WATER RESEARCH

Editor-in-Chief
Mogens Henze
Institute of Environment & Resources
Technical University of Denmark
Bygningstorvet
DK-2800 KGS Lyngby
Denmark
Tel: 45 4525 1477
Fax: 45 4593 2850
E-mail: moh@er.dtu.dk
Editors
J. Block
Université H. Poincaré, Nancy I
France
David Dixon
University of Melbourne
Australia
Hiroaki Furumai
The University of Tokyo
Japan
Gregory Korshin
University of Washington
USA
Anna Ledin
Technical University of Denmark
Denmark
Eberhard Morgenroth
University of Illinois Urbana-Champaign
USA
W. Rauch
University Innsbruck
Austria
Maria Reis
Universidade Nova de Lisboa/FCT
Portugal
Hang-Shik Shin
Korea Advanced Institute of Science
and Technology
Korea
Mark van Loosdrecht
Delft University of Technology
The Netherlands
Thomas Ternes
Bundesanstalt für Gewässerkunde
Germany
Stefan Wuertz
Univ. of California, Davis
USA
Hanqing Yu
University of Science & Technology of China
China
Associate Editors
Andrew Baker
The University of Birmingham
UK
WATER RESEARCH
A Journal of the International Water Association
Damien Batstone
The University of Queensland
Australia
G-H. Chen
The Hong Kong University of Science
& Technology
Hong Kong
China
Tom Curtis
Univ. of Newcastle upon Tyne
UK
Ana Deletic
Monash University
USA
Francis de los Reyes III
North Carolina State University
USA
Rob Eldridge
The University of Melbourne
Australia
Rosina Girones
University of Barcelona
Spain
Stephen Gray
Victoria University
Australia
Kate Grudpan
Chiang Mai University
Thailand
Xiaodi Hao
Beijing University of Civil Engineering
and Architecture
China
E.E. Herricks
University of Illinois - Urbana
USA
H-Y. Hu
Tsinghua University
China
P.M. Huck
University of Waterloo
Canada
Bruce Jefferson
Cranfi eld University
UK
Sergey Kalyuzhnyi
Moscow State University
Russian Federation
Jaehong Kim
Georgia Institute of Technology
USA
Jes La Cour Jansen
Lund Institute of Technology
Sweden
G. Langergraber
BOKU/Univ. of Natural Res. and
Applied Life Scs.
Austria
S-L. Lo
National Taiwan University
Taiwan
Y. Matsui
Hokkaido University
Japan
Max Maurer
EAWAG
Switzerland
How Yong Ng
National University of Singapore
Singapore
Satoshi Okabe
Hokkaido University
Japan
S.L. Ong
National University of Singapore
Singapore
Jong M. Park
Pohang University of Science & Technology
Korea
Miguel Salgot
University of Barcelona
Spain
David Sedlak
University of California, Berkeley
USA
Jean-Philippe Steyer
LBE-INRA
France
M. Takahashi
Hokkaido University
Japan
Kai Udert
EAWAG
Switzerland
V.P. Venugopalan
BARC Facilities
India
E. von Sperling
Federal University of Minas Gerais
Brazil
A.I. Zouboulis
Aristotle University
Greece
Editorial Offi ce
Mailing address
Water Research
Radarweg 29
1043 NX Amsterdam
The Netherlands
Tel: 31 20 485 2701
Fax: 31 20 485 2521; 31 20 485 3264
E-mail: wr-eo@elsevier.com
Publication information: Water Research (ISSN 0043-1354). For 2010, volume 44 is scheduled for publication. Subscription prices are available upon request from the
publisher or from the Elsevier Customer Service Department nearest you or from this journal’s website (http://www.elsevier.com/locate/watres). Further information
is available on this journal and other Elsevier products through Elsevier’s website (http://www.elsevier.com). Subscriptions are accepted on a prepaid basis
only and are entered on a calendar year basis. Issues are sent by standard mail (surface within Europe, air delivery outside Europe). Priority rates are available upon
request. Claims for missing issues should be made within six months of the date of dispatch.
Author Enquiries: Please visit this journal’s homepage at http://www.elsevier.com/locate/watres. You can track accepted articles at http://www.elsevier.com/
trackarticle and set up e-mail alerts to inform you of when an article’s status has changed. Also accessible from here is information on copyright, frequently asked
questions and more. Contact details for questions arising after acceptance of an article, especially those relating to proofs, will be provided by the publisher.
Orders, claims, and journal enquiries: Please contact the Elsevier Customer Service Department nearest you:
St. Louis: Elsevier Customer Service Department, 11830 Westline Industrial Drive, St. Louis, MO 63146, USA; phone: (877) 8397126 [toll free within the USA];
(+1) (314) 4537076 [outside the USA]; fax: (+1) (314) 5235153; e-mail: JournalCustomerService-usa@elsevier.com
Oxford: Elsevier Customer Service Department, The Boulevard, Langford Lane, Kidlington OX5 1GB, UK; phone: (+44) (1865) 843434; fax: (+44) (1865)
843970; e-mail: JournalsCustomerServiceEMEA@elsevier.com
Tokyo: Elsevier Customer Service Department, 4F Higashi-Azabu, 1-Chome Bldg, 1-9-15 Higashi-Azabu, Minato-ku, Tokyo 106-0044, Japan; phone: (+81) (3)
5561 5037; fax: (+81) (3) 5561 5047; e-mail: JournalsCustomerServiceJapan@elsevier.com
Singapore: Elsevier Customer Service Department, 3 Killiney Road, #08-01 Winsland House I, Singapore 239519; phone: (+65) 63490222; fax: (+65) 67331510;
e-mail: JournalsCustomerServiceAPAC@elsevier.com
Application for membership of International Water Association should be made to: Executive Director, IWA, Alliance House, 12 Caxton Street,
London SW1H 0QS, U.K. (Tel.: +44 207 654 5500; Fax: +44 207 654 5555; e-mail: members@IWAhq.org.uk; website: http://www.IWAhq.org.uk).
Registered Charity (England) No. 289269. Individual membership is available from £30 upwards.
For details contact IWA.

Editorial
Editorial to special issue in Water Research
Emerging contaminants in water
The chemical pollution of natural waters is one of the big
challenges of the 21st century. Based on the rapid evolution in
analytical chemistry, with the possibility to detect more polar
compounds a whole new suite of ‘‘emerging contaminants’’
such as pharmaceuticals, hormones and perfluorinated
compounds has been identified in various compartments of
the water cycle including both natural and technical aquatic
systems. The discovery of these micropollutants in the aquatic
environment has triggered research efforts to investigate
sources and mitigation strategies with conventional and novel
treatment processes and assess their importance with regard
to ecotoxicology and human health. In this special issue on
emerging contaminants, we have compiled 20 research articles
and 2 reviews, which deal with these timely issues.
As mentioned above, quantification of micropollutants in
aquatic systems is a key requirement to assess their fate.
Several studies in this issue address the determination of
pharmaceuticals in archived biosolids (Halden et al.), in
wastewater treatment (Lindberg et al.) and lagoon treatment
(Wong et al.). Another study uses analytical data to evaluate
the fraction of pharmaceutical residues in wastewater originating
from hospitals (Ort et al.). Finally, a review covers the
occurrence and fate of phytoestrogens in the environment
(Liu et al.).
To further elucidate the relevance of the micropollutants
detected in various aquatic compartments, their (eco)toxicological
potential has to be assessed. Several papers address
related issues for individual compounds (lipid regulators,
Fernandez-Pinas et al.) or classes of compounds (ionic liquids,
Yun et al.) and for pharmaceuticals in advanced wastewater
treatment systems such as powdered activated carbon and
ozonation with in vivo and in vitro tests (Escher et al., Stalter
et al.). Furthermore, one study is focused on toxicity nanotube
suspensions (Tarabara et al.), which have been on the radar of
emerging contaminants recently. Finally, one paper focuses
on the toxicological relevance of emerging contaminants for
drinking water (Schriks et al.).
In recent years, municipal wastewater has been recognized
as an important source of micropollutants to the
water research 44 (2010) 351
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
receiving water bodies. Therefore, mitigation strategies for
the minimization of the discharge of these compounds play
an increasingly important role in the urban water management.
In this special issue, the removal of benzotriazoles
(Reemtsma et al.) and pharmaceuticals, caffeine and DEET (Yu
et al.) and other emerging contaminants (Rosal et al.) has been
investigated for conventional wastewater treatment.
Numerous papers address the oxidative removal of micropollutants
from wastewater with chlorine, chlorine dioxide,
ferrate, ozone, advanced oxidation processes, such as UV/
H2O2, (solar) photo-Fenton and non-thermal plasma (Lee
et al., Malato et al., Reungoat et al., Gerrity et al., Mendez-
Arriaga). Other options for removal of micropollutants
include membrane processes such as reverse osmosis (Hu
et al.) and nanofiltration (Yangali-Quintanilla et al.) as well as
sorption on sludge (Carrere et al.). The challenges caused by
harmful algae producing toxins for desalination operations
were reviewed by David Caron and co-workers. Finally, mitigation
of micropollutants may also occur during managed
aquifer recharge (Drewes et al.) or in biological Fenton-like
processes (Vicent et al.).
In short, we are very happy to provide you this Theme Issue
on Emerging Contaminants. We appreciate the contributions
by the authors, reviewers and editorial staff of Water Research
to this project.
Thomas Ternes*
Federal Institute of Hydrology (BFG),
Am Mainzer Tor 1, 56068 Koblenz, Germany
*Corresponding author. Tel.: þ49 261 1306 5560;
fax: þ49 261 1306 5363.
Urs von Gunten
EAWAG, Ueberlandstrasse 133, Duebendorf CH-8600,
Switzerland
0043-1354/$ – see front matter
ª 2010 Published by Elsevier Ltd.
doi:10.1016/j.watres.2010.01.015

Review
Environmental fate and toxicity of ionic liquids: A review
Thi Phuong Thuy Pham a , Chul-Woong Cho a , Yeoung-Sang Yun a,b, *
a Department of Bioprocess Engineering, Chonbuk National University, Jeonju, Chonbuk 561-756, Republic of Korea
b Division of Semiconductor and Chemical Engineering and Research Institute of Industrial Technology, Chonbuk National University,
Chonbuk 561-756, Republic of Korea
article info
Article history:
Received 31 May 2009
Received in revised form
27 August 2009
Accepted 12 September 2009
Available online 24 September 2009
Keywords:
Ionic liquids
Toxicity
Degradation
Biodegradation
Environmental fate
Sorption
Contents
water research 44 (2010) 352–372
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
abstract
Ionic liquids (ILs) are organic salts with low melting point that are being considered as
green replacements for industrial volatile organic compounds. The reputation of these
solvents as ‘‘environmental friendly’’ chemicals is based primarily on their negligible vapor
pressure. Nonetheless, the solubility of ILs in water and a number of literature
documenting toxicity of ILs to aquatic organisms highlight a real cause for concern. The
knowledge of ILs behavior in the terrestrial environment, which includes microbial
degradation, sorption and desorption, is equally important since both soil and aquatic
milieu are possible recipients of IL contamination. This article reviews the achievements
and current status of environmental risk assessment of ILs, and hopefully provides
insights into this research frontier.
ª 2009 Elsevier Ltd. All rights reserved.
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
2. Toxicological aspect of ILs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354
2.1. Effects of ILs in an enzyme level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354
2.2. Antibacterial activity of ILs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 356
2.3. Toxicity of ILs to algae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
2.4. Cytotoxicity of ILs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
2.5. Phytotoxicity of ILs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363
2.6. Toxicity of ILs to invertebrates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363
2.7. Inhibitory effects of ILs on vertebrates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 364
3. Environmental fate of ILs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 364
3.1. Chemical degradation of ILs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 364
3.2. Biodegradability of ILs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 365
3.3. Sorption of ILs in environmental systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 367
* Corresponding author. Division of Semiconductor and Chemical Engineering, Chonbuk National University, Jeonju, Chonbuk 561-756,
Republic of Korea. Tel.: þ82 63 270 2308; fax: þ82 63 270 2306.
E-mail address: ysyun@chonbuk.ac.kr (Y.-S. Yun).
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.watres.2009.09.030

4. Concluding remarks and future directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 368
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369
1. Introduction
Most of volatile organic compounds (VOCs) commonly used in
industrial applications cause a major concern in the current
chemical processing industry. The main problems are the
toxicity of the organic solvents to both the process operators
and the environment as well as the volatile and flammable
nature of these solvents which make them a potential explosion
hazard (Schmid et al., 1998). Recently, the deleterious
effects of many solvents combined with serious
environmental issues, such as atmospheric emissions and
contamination of aqueous effluents are making their use
prohibitive. Thus, many researchers have focused on the
development of ‘‘green engineering’’ which represents
research aimed at finding environmentally benign alternatives
to harmful chemicals. Among the neoteric solvents applicable
in ‘‘green technologies’’ ionic liquids (ILs) have garnered
increasing attention over the others such as supercritical CO 2
(Blanchard et al., 1999; Blanchard and Brennecke, 2001;
Kazarian et al., 2000) and aqueous biphasic systems (Myasoedov
et al., 1995; Rogers et al., 1995; Willauer et al., 1999).
Ionic liquids, formerly known as molten salts, constitute
one of the hottest areas in chemistry these days. Basically,
they have melting points below 100 C, which can be achieved
by incorporating a bulky asymmetric cation into the structure
together with a weakly-coordinating anion (Ranke et al., 2004).
The unique, highly solvating, yet non-coordinating environment
of ILs provides an attractive medium for various types of
chemical processes. Also, the physical properties of ILs can be
tailored by a judicious variation in the length and branching of
the alkyl chain and the anionic precursor (Fuler et al., 1997;
Huddleston et al., 2001). In this way, ILs can be made taskspecific
for a certain application. The almost limitless structural
possibilities of ILs, as opposed to limited structural
variations within molecular solvents, make them ‘‘designer
water research 44 (2010) 352–372 353
Fig. 1 – Applications of ionic liquids.
solvents’’ (Marsh et al., 2004; McFarlane et al., 2005; Sheldon,
2005). Some independent reports (Hagiwara and Ito, 2000;
Olivier, 1999; Welton, 1999) and many reviews (Earle and
Seddon, 2000; Rooney and Seddon, 2001) have highlighted ILs
as representing a state-of-the-art, innovative approach to
sustainable chemistry, with the argument that their vapor
pressure is immeasurably low and they are not flammable.
Recently, the application of these liquids as reaction media for
organic synthesis, catalysis, or biocatalysis has been well
documented (Earle and Seddon, 2000; Wasserscheid and
Keim, 2000; Welton, 1999) (Fig. 1). Gordon (2001) pointed out
that there is an obvious advantage in performing many reactions
in ILs due to the improvement in process economics,
reaction activity, selectivity and yield.
Although ILs can lessen the risk of air pollution due to their
insignificant vapor pressure, they do have significant solubility
in water (Anthony et al., 2001; McFarlane et al., 2005;
Wong et al., 2002). As a result, this is the most likely medium
through which ILs will be released into the environment. Ionic
liquids currently are not widely used in industrial applications;
nonetheless, continued development and further use of
these solvents may lead to accidental discharge and
contamination. The properties that make them be the target
of industrial interest (i.e. high chemical, thermal stability and
non-volatility) suggest potential problems with degradation or
persistence in the environment. In general, the deficiency of
information and uncertainty surrounding the environmental
impact of ILs is a major barrier to the utilization of these
compounds by industry. Initial efforts have been made to
overcome this drawback and offer a preliminary insight into
the behavior of ILs in the aqueous environments. These
studies provided extensive data sets, e.g. on (eco)toxicity,
biodegradability, bioaccumulation and distribution of ILs in
different environmental compartments. Therefore, it is
necessary to consolidate all the available data in a single

354
review to lay the groundwork for more comprehensive
community and ecosystem investigations. The overall objective
of this review is to systematically gather and interpret
existing information about the fate, removal options and
(eco)toxicological assessment strategies of ILs.
2. Toxicological aspect of ILs
The current literature represents a number of studies addressing
the biological effects of ILs evaluated on the basis of
toxicological test systems. The ILs toxicities towards these
systems of different levels of biological complexity as well as
several environmental compartments (Fig. 2) aresuccessively
discussed in the following subsections. All the structures of
IL compounds discussed in this review were listed in Table 1.
The acronyms used for these substances were adapted
from Ranke et al. (2007a). In this way, the cation head groups
were abbreviated as ‘‘IM’’ for imidazolium, ‘‘Py’’ for pyridinium,
‘‘Pyr’’ for pyrrolidinium, ‘‘Mor’’ for morpholinium, ‘‘Pip’’ for
piperidinium, ‘‘Quin’’ for quinolinium, ‘‘N’’ for quaternary
ammonium and ‘‘P’’ for quaternary phosphonium. The alkyl
chains attached to the head group were given as numbers
corresponding to the number of carbon in the alkyl residues. For
example, the 1-butyl-3-methylimidazolium moiety was denoted
as IM14. In case the carbon chain length equals or exceeds 10,
the numbers were separated by a hyphen (e.g. IM1-10 indicated
1-decyl-3-methylimidazolium). Particularly, for pyridinium
water research 44 (2010) 352–372
entities, the carbon-bound alkyl chains were appended to the
head group at different positions and the abbreviation was
made by noting the position of attachment and a symbol for the
attached group (e.g. Py4-2Me for 1-butyl-2-methylpyridinium).
The anionic components were shortened as they are in the
periodic table for the halides. For tetrafluoroborate, hexafluorophosphate,
bis(trifluoromethylsulfonyl)imide, dicyanamide
and hydrogen sulfate the abbreviations were BF4, PF6,
(CF 3SO 2) 2N, CN(N) 2 and HSO 4 in respective to their structural
formula.
2.1. Effects of ILs in an enzyme level
Enzyme inhibition data by ILs include those of the acetylcholinesterase
from electric eel (Electrophorus electricus)
(Arning et al., 2008; Jastorff et al., 2005; Matzke et al., 2007;
Ranke et al., 2007b; Stasiewicz et al., 2008; Stock et al., 2004;
Torrecilla et al., 2009; Zhang and Malhotra, 2005), the AMP
deaminase (Sk1adanowski et al., 2005) and the antioxidant
enzyme system of mouse liver (Yu et al., 2009a). The enzyme
acetylcholinesterase plays the most important role in nerve
response and function. Also, acetylcholinesterase catalyzes
the hydrolysis of acetylcholinesters with a relative specificity
for acetylcholine, which is a neurotransmitter common to
many synapses throughout mammalian nervous systems
(Fulton and Key, 2001; Massoulié et al., 1993). Thus, an inhibition
of acetylcholinesterase leads to various adverse effects
in neuronal processes, such as heart diseases or myasthenia
Fig. 2 – The flexible (eco)toxicological test battery considering aquatic and terrestrial compartments as well as different
trophic levels including enzymes, luminescent marine bacteria, freshwater green algae, mammalian cells, duckweed,
freshwater crustacean and zebrafish (Adapted from Matzke et al. (2007) by permission of the Royal Society of Chemistry).

Table 1 – Selection of cationic and anionic structures of commonly used ionic liquids.
Cation
water research 44 (2010) 352–372 355
CH 3
CH 3
R 2
Head group Side chain
N N +
R 1
Imidazolium (IM)
CH 3
N +
Pyridinium (Py)
N +
CH 3
R 1
Pyrrolidinium (Pyr)
O
N +
CH 3
R 1
Morpholinium (Mor)
N +
CH3 Piperidinium (Pip)
N +
R 1
R 1
Quinolinium (Quin)
N +
R1 R2
R 4
R3
Quaternary ammonium (N)
P +
R1 R2
R 4
R3
Quaternary phosphonium (P (
R 1
R1 ¼ -C2H5, -C3H7, -C4H9,
-C 5H 11, -C 6H 13, -C 7H 15,
-C 8H 17, -C 9H 19, -C 10H 21,
-C14H29, -C16H33, -C18H37,
-C19H39
R 2 ¼ -CH 3,-C 2H 5
R1 ¼ -C2H5, -C3H7, -C4H9,
-C 5H 11, -C 6H 13, -C 8H 17
R1 ¼ -C4H9, -C6H13, -C8H17
R 1 ¼ -C 4H 9
R1 ¼ -C4H9
R1 ¼ -C4H9, -C6H13, -C8H17
R 1-4 ¼ -CH 3,-C 2H 5,-C 3H 7,
-C4H9, -C6H13
R 1-4 ¼ -C 4H 9,-C 6H 13, -C 14H 29
(continued on next page)

356
Table 1 (continued)
in humans (Chemnitius et al., 1999; Pope et al., 2005). Ranke
et al. (2007b) published a comprehensive collection of acetylcholinesterase
inhibition values for 292 compounds covering
a large variety of ILs and closely related salts. Among these
data, only those of the commonly tested ILs are summarized
in Table 2 for the ease of comparison of ILs toxicity from
molecular up to organism levels of biological complexity.
It was found that all observed inhibitory effects on the enzyme
could be exclusively accounted for the cationic moiety (Arning
et al., 2008). In particular, the IL with pyridinium as cationic
core structure inhibited the enzyme slightly stronger than the
imidazolium analogue whereas the compounds based on
phosphonium was less inhibitory. All anion species exerted
no effect on the enzyme activity with only exception of the
fluoride anion and the fluoride containing [SbF6] and [PF6]
species. Both species are known to readily undergo hydrolysis
in contact with moisture and thus the fluoride seems to be the
active compound. The non-inhibiting effects of anion might
be explained by their limited interactions with the active site
of this enzyme (Matzke et al., 2007). In addition, a correlation
between an increasing chain length of the side chains
connected to the cationic head groups and an enhanced
inhibitory potential of the ILs was found. It is believed that the
mechanism involves the similarity of the positively charged
imidazolium or pyridinium to the choline part that binds to
the anionic site of the enzyme, such that the longer alkyl
chain results in an improved fit (Stock et al., 2004).
Sk1adanowski et al. (2005) discussed the usefulness of in
vitro AMP deaminase inhibition assay as a potential molecular
method in prospective risk analysis of imidazolium-based ILs.
The results revealed that IM14 salts associated with [PF 6] ,
[BF 4] , p-tosylate and [Cl] demonstrated a dose-dependent
inhibition of AMP deaminase activity. The IC50 values
(concentration of ILs inhibiting 50% of enzyme activity) for
those containing a fluorine compartment [PF6] and [BF4] are
lower (5 mM) than those for [Cl] and p-tosylate (10 mM), which
indicated the adverse effect of these fluoride-containing
anions. The other study on enzyme inhibition assay dealt with
Head group Side chain
Anion Chloride Cl
Bromide Br
Tetrafluoroborate [BF4]
Hexafluorophosphate [PF6]
Bis(trifluoromethylsulfonyl)imide [(CF 3SO 2) 2N]
Dicyanamide [(CN)2N]
water research 44 (2010) 352–372
the effects of acute exposure of intraperitoneal injection of
aqueous IM18 Br on the antioxidant enzymes of the treated
mouse liver (Yu et al., 2009a). The antioxidant enzymes tested
included superoxide dismutase, catalase, glutathione peroxidase
and glutathione-S-transferase. The results showed that
administration of IM18 Br modified activities of these defense
enzymes in mouse liver, and caused damage to livers of
treated mice at median lethal dose (LD 50) of 35.7 mg/kg.
Though data published by these authors did not cover
a large variety of ILs, the enzyme inhibition assays suggest the
trend in which cationic moiety is the dominating factor
influencing the toxicity of ILs, especially when substituted
with a long alkyl side chain. Regarding the anion types,
perfluoronated ions are of toxicological interest due to
hydrolysis resulting in HF formation, while the others cause
less prominent effect.
2.2. Antibacterial activity of ILs
Bacteria serve as an ideal starting point for ILs toxicity
estimations as they have short generation times. Preliminary
toxicological investigations have shown quaternary ammonium
and pyridinium compounds have critical inhibitory
effects on a variety of bacteria and fungi (Babalola, 1998;
Kelman et al., 2001; Li et al., 1998). In the studies of Pernak’s
group (Cieniecka-Ros1onkiewicz et al., 2005; Pernak et al.,
2001a; Pernak et al., 2001b; Pernak and Chwa1a, 2003; Pernak
et al., 2003; Pernak et al., 2004a), they observed a trend of
increasing toxicity with an increase in the alkyl chain length
substituent in the pyridinium, imidazolium and quaternary
ammonium salts to various bacteria including rods, cocci
and fungi. As a measure of microbial activity of imidazolium
and pyridinium ILs with varying alkyl chain lengths, Docherty
and Kulpa (2005) also used a group of microorganisms
possessing a variety of physiological and respiratory activities.
It was found that imidazolium and pyridinium bromides
incorporated hexyl- and octyl-chain had considerable antimicrobial
effect to pure cultures of Escherichia coli,
F
F
F
B -
F
F
P -
F
F
F
F
F
N -
O O
F3C CF3 S S
O
O
N -
N N

Table 2 – Toxicity of ILs to different levels of biological complexity including enzyme, bacteria, algae, rat cell line, human cell lines, duckweed and invertebrate.
Compound Log10EC50 (mM) a
IM12 Cl 2.06 14
IM12 BF4 IM12 PF6
2.05 14
IM12 (CF3SO2)2N 2.03 14
IM13 Cl 2.27 14
IM13 BF4 2.28 0.03 18
IM13 PF6
2.22 14
IM14 Cl 1.91 0.04 11
Acetylcholin esterase Vibrio
fischeri
2.05 14
IM14 Br 1.90 0.02 18
IM14 BF 4 1.98 0.018 11
Escherichia
coli
Pseudo kirchneriella
subcapitata
Scenedesmus
vacuolatus
IPC-81 HeLa MCF7 b
Lemna
minor
4.55 10
4.33 0.11 19
N.A. N.A. 2.78 0.06 19
N.A. N.A. N.A. N.A. N.A.
N.A. 5.25 0.06 9
N.A. N.A. 3.44 14
4.00 0.04 20
N.A. N.A. N.A.
N.A. N.A. N.A. N.A. 3.92 14
N.A. N.A. N.A. N.A.
N.A. N.A. N.A. N.A. N.A. 3.26 0.04 20
N.A. N.A. N.A.
N.A. N.A. N.A. N.A. >4.30 14
N.A. N.A. N.A. N.A.
3.94 0.06 13
N.A. N.A. N.A. 3.47 14
N.A. N.A. N.A. N.A.
N.A. N.A. N.A. N.A. >3.00 14
N.A. N.A. N.A. N.A.
3.71 0.14 4
2.95 5
3.34 0.13 6
3.47 0.04 19
4.01 0.05 4
3.07 0.03 13
3.35 5
3.27 0.09 6
3.55 0.04 13
3.10 0.17 6
3.12 0.35 16
3.07 0.29 6
N.A. 2.34 0.01 21
N.A. 3.46 0.062 3
4.60 0.02 9
N.A. 2.11 11
2.26 0.08 11
3.55 14
N.A. 3.43 14
3.12 14
N.A. N.A. 2.82 11
3.44 0.11 20
3.72 0.05 17
3.66 0.08 20
Daphnia
magna c
1.93 4
1.93 0.06 1
N.A. N.A. 1.57 4
1.56 0.07 1
1.85 0.06 22
N.A. 2.49 7
1.68 4
1.67 0.11 1
IM14 PF6 2.15 0.05 18
4.15 0.06 9
2.20 0.04 21
N.A. 3.10 14
4.14 0.22 17
N.A. N.A. 1.85 4
1.85 0.10 1
IM14 (CF3SO2)2N 1.96 0.021 11
3.39 0.08 4
2.55 0.15 9
1.80 0.07 12
1.81 0.15 11
2.68 14
3.07 0.08 20
N.A. 2.45 0.08 11
N.A.
IM14 (CN)2N 1.95 0.07 18
3.67 0.10 4
2.99 5
N.A. N.A. N.A. 3.15 14
N.A. N.A. N.A. N.A.
IM15 Cl 1.96 14
N.A. N.A. N.A. N.A. >3.00 14
N.A. N.A. N.A. N.A.
IM15 BF4
1.86 14
3.14 0.02 13
N.A. N.A. N.A. >3.00 14
N.A. N.A. N.A. N.A.
IM15 PF6
1.85 14
N.A. N.A. N.A. N.A. >3.00 14
N.A. N.A. N.A. N.A.
IM16 Cl 1.92 14
1.94 10
2.32 0.16 6
2.91 0.09 13
N.A. 1.92 0.09 21
0.08 19
2.85 14
N.A. N.A. N.A. N.A.
IM16 Br N.A. 1.42 0.12 4
0.81 5
N.A. 2.57 0.15 2
N.A. N.A. N.A. N.A. N.A. 0.78 4
1.06 0.04 22
IM16 BF4 1.88 14
3.18 0.03 13
N.A. N.A. N.A. 2.98 14
N.A. N.A. N.A. N.A.
IM16 PF6
1.88 14
2.17 0.06 6
3.25 0.67 9
N.A. N.A. 2.91 14
N.A. N.A. N.A. N.A.
IM16 (CF3SO2) 2N 2.15 14
N.A. 2.53 0.15 9
N.A. N.A. 2.24 14
N.A. 2.81 8
N.A. N.A.
IM17 Cl 2.07 14
N.A. N.A. N.A. N.A. 2.53 14
N.A. N.A. N.A. N.A.
IM17 BF4
2.12 14
2.44 0.06 13
N.A. N.A. N.A. 2.58 14
N.A. N.A. N.A. N.A.
IM17 PF6
1.91 14
N.A. N.A. N.A. N.A. 2.30 14
N.A. N.A. N.A. N.A.
IM18 Cl 1.60 14
1.19 0.11 6
1.01 0.06 19
N.A. 1.46 21
2.67 0.37 19
2.01 14
N.A. N.A. N.A. N.A.
IM18 Br N.A. 0.63 0.07 4
0.07 5
N.A. 1.65 0.25 2
N.A. N.A. 2.48 0.04 20
N.A. N.A. 1.33 4
0.54 0.12 22
(continued on next page)
water research 44 (2010) 352–372 357

Table 2 (continued)
Compound Log 10EC 50 (mM) a
Acetylcholin esterase Vibrio
fischeri
Escherichia
coli
Pseudo kirchneriella
subcapitata
Scenedesmus
vacuolatus
IPC-81 HeLa MCF7 b
IM18 BF4 1.53 0.025 11
1.41 0.07 13
N.A. N.A. 2.30 11
1.59 14
2.48 0.02 20
2.84 8
0.90 7
N.A.
IM18 PF6 2.03 14
0.95 0.12 6
2.64 0.15 9
N.A. N.A. 1.96 14
N.A. N.A. N.A. N.A.
IM18 (CF3SO2)2N 2.03 14
N.A. N.A. N.A. N.A. 1.64 14
2.28 0.02 20
N.A. N.A. N.A.
IM19 BF4 N.A. 0.72 0.04 13
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
IM1-10 Cl 1.09 14
0.50 0.07 13
0.23 0.06 19
N.A. N.A. 3.57 0.06 19
1.34 14
N.A. N.A. N.A. N.A.
IM1-10 BF4 1.10 0.04 18
0.18 0.06 13
N.A. N.A. N.A. 0.77 14
N.A. N.A. N.A. N.A.
IM1-10 PF6 1.68 14
N.A. N.A. N.A. N.A. 1.50 14
N.A. N.A. N.A. N.A.
IM1-14 Cl 0.54 14
0.15 0.07 19
N.A. N.A. 2.48 0.2 19
0.42 14
N.A. N.A. N.A. N.A.
IM1-16 Cl 0.68 14
0.23 0.08 19
N.A. N.A. > 2.00 19
0.19 14
N.A. N.A. N.A. N.A.
IM1-18 Cl 0.96 14
1.45 0.05 19
N.A. N.A. > 2.00 19
0.01 14
N.A. N.A. N.A. N.A.
IM1-19 Cl 1.36 14
N.A. N.A. N.A. N.A. 1.40 14
N.A. N.A. N.A. N.A.
IM1-19 BF4 1.43 14
N.A. N.A. N.A. N.A. 1.65 14
N.A. N.A. N.A. N.A.
IM1-19 PF6
1.62 14
N.A. N.A. N.A. N.A. 1.85 14
N.A. N.A. N.A. N.A.
IM22 Br 2.08 14
N.A. N.A. N.A. N.A. >3.00 14
N.A. N.A. N.A. N.A.
IM23 Br 2.21 14
N.A. N.A. N.A. N.A. >3.30 14
N.A. N.A. N.A. N.A.
IM24 BF4 2.03 0.01 18
2.8 0.04 13
N.A. N.A. N.A. 3.26 14
4.36 0.09 17
N.A. N.A. N.A.
IM25 BF4 N.A. 3.14 13
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
IM26 Br 1.77 14
N.A. N.A. N.A. N.A. 2.01 14
N.A. N.A. N.A. N.A.
IM26 BF4 1.84 14
2.15 0.05 13
N.A. N.A. N.A. 2.26 14
N.A. N.A. N.A. N.A.
IM2-10 Br 0.92 14
N.A. N.A. N.A. N.A. 0.53 14
N.A. N.A. N.A. N.A.
Py Cl >3.00 14
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
Py2 Cl 2.10 14
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
Py3 Br 2.22 14
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
Py3 (CF3SO2)2N 2.21 14
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
Py4 Cl 1.70 14
3.41 0.08 4
2.64 5
3.18 0.06 19
N.A. 2.57 0.06 21
2.59 0.11 19
N.A. N.A. N.A. 2.32 0.18 19
N.A.
Py4 Br 1.77 14
3.40 0.01 4
2.73 5
N.A. N.A. N.A. 3.90 14
3.50 0.07 20
N.A. N.A. N.A.
Py4 BF4 1.80 14
N.A. N.A. N.A. N.A. 3.18 14
N.A. N.A. N.A. N.A.
Py4 PF6
1.84 14
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
Py4 (CN)2N N.A. 3.31 0.10 4
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
2.61 5
Py5 Br 1.52 14
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
Py5 (CF3SO2)2N 1.55 14
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
Py6 Cl 1.72 14
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
Py6 Br N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. 1.07 4
Py6 PF6 1.76 14
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
Py6 (CF3SO2)2N 1.85 14
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
Py8 Cl 1.60 14
N.A. N.A. N.A. N.A. 1.27 14
N.A. N.A. N.A. N.A.
Py8 (CF3SO2) 2N 1.40 14
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
Py4-2Me Cl 0.70 14
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
Py4-2Me BF4
0.82 14
N.A. N.A. N.A. N.A. 3.25 14
N.A. N.A. N.A. N.A.
Lemna
minor
Daphnia
magna c
358
water research 44 (2010) 352–372

Py4-3Me Cl 1.15 14
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
Py4-3Me Br N.A. 2.75 0.13 4
N.A. 3.46 0.062 3
N.A. N.A. N.A. N.A. N.A. 1.76 4
2.12 5
Py4-3Me BF4 1.53 0.02 18
N.A. N.A. N.A. N.A. 3.30 14
N.A. N.A. N.A. N.A.
Py4-3Me PF6 1.45 0.02 18
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
Py4-3Me (CN) 2N 1.22 14
2.66 0.05 4
1.99 5
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
Py6-3Me Cl 1.06 14
1.44 10
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
Py6-3Me Br N.A. 2.06 0.16 4
N.A. N.A. N.A. N.A. N.A. N.A. N.A. 0.59 4
1.48 5
Py6-4Me Cl 1.44 14
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
Py6-4Me BF4
1.48 14
N.A. N.A. N.A. N.A. 2.17 14
N.A. N.A. N.A. N.A.
Py8-3Me Cl 0.64 14
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
Py8-3Me Br N.A. 0.79 0.05 4
N.A. N.A. N.A. N.A. N.A. 1.00 8
N.A. 0.40 4
Py8-4Me Cl 1.11 14
Py8-4Me BF4 1.22 14
Pyr14 Cl 1.92 14
Pyr14 Br 1.93 14
0.25 5
N.A. N.A. N.A. N.A. 1.63 14
N.A. N.A. N.A. N.A. 1.49 14
>4.30 19
N.A. N.A. 3.37 0.10 19
>4.30 14
N.A. N.A. 3.67 0.28 3
N.A. 3.77 14
N.A. N.A. N.A. N.A.
N.A. N.A. N.A. N.A.
N.A. N.A. 2.16 0.25 19
N.A.
N.A. 4.64 0.02 15
N.A. N.A.
Pyr14 BF4 1.91 14
N.A. N.A. N.A. N.A. 2.90 14
N.A. N.A. N.A. N.A.
Pyr14 (CF3SO2)2N 2.13 14
N.A. N.A. >2.38 12
2.53 19
3.01 14
N.A. 3.14 8
2.98 0.32 19
N.A.
Pyr14 (CN)2N 1.98 14
N.A. N.A. N.A. N.A. 4.23 14
N.A. N.A. N.A. N.A.
Pyr16 Cl 2.48 14
2.99 10
N.A. N.A. N.A. 2.91 14
N.A. N.A. N.A. N.A.
Pyr16 (CF3SO2) 2N 2.60 14
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
Pyr18 Cl 2.36 14
N.A. N.A. N.A. N.A. 2.59 14
N.A. N.A. N.A. N.A.
Pyr18 BF4
2.02 14
N.A. N.A. N.A. N.A. 1.82 14
N.A. N.A. N.A. N.A.
Pyr66 2.08 14
N.A. N.A. N.A. N.A. 1.23 14
N.A. N.A. N.A. N.A.
Mor14 Cl N.A. >4.30 19
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
Mor14 Br 2.71 14
N.A. N.A. N.A. >4.00 19
>4.30 14
N.A. N.A. 3.11 0.13 19
N.A.
Mor14 (CF3SO2) 2N 2.78 14
N.A. N.A. N.A. 2.00 19
3.43 14
N.A. N.A. 3.15 0.13 19
N.A.
Pip14 Br 1.83 14
4.27 0.09 19
N.A. N.A. 3.27 0.12 19
4.03 14
N.A. 4.15 0.04 15
0.47 19
N.A.
Pip14 (CF3SO2)2N 1.78 14
N.A. N.A. N.A. 2.08 19
3.41 14
N.A. 2.93 8
2.85 0.07 19
N.A.
Quin4 Br 0.79 14
N.A. N.A. N.A. N.A. 2.32 14
N.A. N.A. N.A. N.A.
Quin4 BF4 0.62 14
N.A. N.A. N.A. N.A. 2.16 14
N.A. N.A. N.A. N.A.
Quin6 BF4
0.48 14
N.A. N.A. N.A. N.A. 1.07 14
N.A. N.A. N.A. N.A.
Quin8 Br N.A N.A. N.A. N.A. N.A. 0.03 14
N.A. N.A. N.A. N.A.
Quin8 BF4 0.30 14
N.A. N.A. N.A. N.A. 0.17 14
N.A. N.A. N.A. N.A.
N1111 Br N.A. >5.00 4
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
N1114 (CF3SO2)2N 2.60 14
N.A. N.A. N.A. N.A. 3.61 14
N.A. N.A. N.A. N.A.
N1123 (CF3SO2) 2N 2.34 14
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
N1124 Cl 2.06 14
N.A. N.A. N.A. >4.00 19
>4.30 14
N.A. N.A. 0.83 0.67 19
N.A.
N1124 (CF3SO2)2N 2.03 14
N.A. N.A. N.A. 1.78 0.17 19
3.43 13
N.A. N.A. N.A. N.A.
N2222 Cl 2.80 14
N.A. N.A. N.A. N.A. >3.48 14
N.A. N.A. N.A. N.A.
N2222 Br N.A. >5.00 4
N.A. N.A. N.A. N.A. 4.26 0.04 20
N.A. N.A. N.A.
N2226 Br N.A. 2.46 0.16 4
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
N4444 Br 2.30 14
3.27 0.07 4
N.A. N.A. N.A. 2.25 14
N.A. N.A. N.A. 1.47 4
P4444 Br 2.61 14
2.71 4
N.A. N.A. N.A. 1.66 14
N.A. N.A. N.A. 0.95 4
P666-14 Br 2.85 14
3.41 0.02 4
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
4.25 8
4.15 8
(continued on next page)
water research 44 (2010) 352–372 359

360
Table 2 (continued)
Compound Log10EC50 (mM) a
Daphnia
magna c
Lemna
minor
IPC-81 HeLa MCF7 b
Scenedesmus
vacuolatus
Pseudo kirchneriella
subcapitata
Escherichia
coli
Acetylcholin esterase Vibrio
fischeri
N.A. N.A. N.A. N.A. 0.48 14
N.A. N.A. N.A. N.A.
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
N.A. N.A. N.A. N.A. 0.24 14
1.90 0.05 20
N.A. N.A. N.A.
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
P666-14 BF4 3.47 0.08 18
P666-14 PF6
>3.30 17
P666-14 (CF3SO2)2N >3.48 14
P666-14 (CN) 2N 3.40 0.2 18
References: 1 Bernot et al. (2005a); 2 Cho et al. (2007); 3 Cho et al. (2008a,b); 4 Couling et al. (2006); 5 Docherty and Kulpa (2005); 6 Garcia et al. (2005); 7 Jastorff et al. (2005); 8 Kumar et al. (2009); 9 Lee et al. (2005);
10 11 12 13 14 15 16 17 18
Luis et al. (2007); Matzke et al. (2007); Pretti et al. (2008); Ranke et al. (2004); Ranke et al. (2007b); Salminen et al. (2007); Samorì et al. (2007); Stepnowski et al. (2004); Stock et al. (2004);
19 Stolte et al. (2007a); 20 Wang et al. (2007); 21 Wells and Coombe (2006); 22 Yu et al. (2009).
a N.A. means not available (not determined).
b Toxicity of ILs is expressed as log10IC50 (mM) in case of MCF7 cell line.
c Toxicity of ILs is expressed as log10LC50 (mM) in case of D. magna.
water research 44 (2010) 352–372
Staphylococcus aureus, Bacillus subtilis, Pseudomonas fluorescens
and Saccharomyces cerevisiae. The anion performed nearly no
effect on antimicrobial activity in the case of imidazolium
analogues (Docherty and Kulpa, 2005; Garcia et al., 2005; Lee
et al., 2005; Pernak et al., 2003; Pernak et al., 2004a) whereas
this was not the case for phosphonium salts. Within the group
of alkyltrihexylphosphonium ILs in the study of Cieniecka-
Ros1onkiewicz et al. (2005), both cation structure and the type
of anion had effects on the biological activity.
The antibacterial activity of ILs not only involves in
hampering the growth rate of microbes but also interferes
with their productivity. Matsumoto et al. (2004a) tested the
toxicity of imidazolium-based ILs to lactic acid producing
bacterium Lactobacillus rhamnosus to examine whether these
compounds can replace conventional organic solvents in the
extractive fermentation of lactate. The results showed that
the bacterium L. rhamnosus grew, consumed glucose, and
produced lactate in the presence of imidazolium-based ILs.
A change of alkyl length in the imidazolium cation had little
difference on the survival of the cells. In a similar study
(Matsumoto et al., 2004b), they focused on hiochii bacteria,
Lactobacillus homochiochii and Lactobacillus fructivorans and also
found that the bacteria could produce lactic acid in the presence
of ILs. Nonetheless, the lactic acid producing activities of
these bacteria generally decreased with the extension of alkyl
chain length in the imidazolium cation moiety.
Water miscible ILs had various effects on the physiology of
Clostridium sporogenes when tested as additives in culture
media or reaction media for reduction of nitrobenzene
(Dipeolu et al., 2008). In their study, 2-hydroxyethyltrimethylammonium
dimethylphosphate and N,N-dimethylethanolammonium
acetate increased the growth rate of C. sporogenes;
by contrast, IM14 BF4 and IM12 EtSO 4 inhibited growth.
Although IM12 EtSO 4 inhibited growth, it was sufficiently
non-toxic to allow efficient reduction of nitrobenzene using
harvested cells. Thus, it is recommended that both noninhibitory
and partially inhibitory ILs should be screened for
use in biotransformation. Nonetheless, Ganske and Bornscheuer
(2006) referred that ILs could have substantial inhibitory
effects on the growth of microorganisms when they
explored the effects of the two most commonly used ILs IM14
BF 4 and IM14 PF 6 on the growth of E. coli, Pichia pastoris and
Bacillus cereus.
Regarding inhibition assays used in assessment of environmental
potential risk of a compound in aquatic milieu, the
bioluminescence assay using Vibrio fischeri (formerly known as
Photobacterium phosphoreum) is one of the most applied (Kaiser
and Palabrica, 1991; Steinberg et al., 1995). This is a rapid, costeffective,
and well-established method for toxicity determination
focusing on environmental issues, and also a standard
ecotoxicological bioassay in Europe (DIN EN ISO 11348). The
published data on ILs toxicity towards V. fischeri were listed on
Table 2 and were comprehensively interpreted in the study of
Peraccini et al. (2007). Although it has been claimed that
modifications of the anion lead to changes in chemical and
physical properties of ILs (Sheldon, 2001), no clear increase in
toxicity caused by the anion could be observed, and toxicity
seemed to be determined mainly by the cationic component
(Ranke et al., 2004). This is likely explained by the fact that
lipophilic part of the molecules can be intercalated into the

membrane, whereas their ionic head group is at least partially
solvated in the aqueous solution, as suggested by Austin et al.
(1998). The ILs toxicity was also observed to correlate directly
with the length of the n-alkyl residues in the methylimidazolium
cation (Romero et al., 2008). Interestingly, Ranke
et al. (2004) noted a slight hormetic effect at concentrations
below inhibitory concentrations. Concerning the anionic
influence, compounds with [PF6] were found to be slightly
more toxic than compounds with other anions in their study
(Ranke et al., 2004). The anion [(CF 3SO 2) 2N] showed no
intrinsic toxicity to V. fischeri in the report of Matzke et al.
(2007); in contrast, an increased in toxicity was found for all
tested compounds combined with [(CF3SO2)2N] for V. fischeri
(Stolte et al., 2007a). Couling et al. (2006) extended the bioluminescence
inhibition assay to pyridinium derivatives and it
was noted that the quaternary ammonium compounds
seemed to be less toxic to V. fischeri than the pyridinium
and imidazolium analogues. Also, the quantitative structureproperty
relationship (QSPR) modeling suggested that imidazolium
cations, with two nitrogen atoms, are predicted to
be more toxic than pyridinium moieties, which only have
one nitrogen atom in the structure. In addition, the QSPR
correlation predicted that quaternary ammonium cations are
less toxic than those with cations containing nitrogen-bearing
rings, which was in agreement with the experimental results
(Couling et al., 2006). However, in contrast to the cases of
aromatic ILs and ammonium compounds, the authors were
unsuccessful in modeling the behavior of phosphonium salts
using the developed correlation.
2.3. Toxicity of ILs to algae
As algae are primary producers, either directly or indirectly, of
organic matter required by animals in freshwater food chains,
their ecology is crucial in providing the energy for sustaining
other higher trophic levels. The ubiquity of algae makes these
organisms ideal for toxicological studies and, because they
have a short life cycle they can respond quickly to environmental
change (Blaise, 1993; Lewis, 1995). To date, several
groups have focused their attention on the use of algal
primary producers to assess the effects of ILs to aquatic
environments (Cho et al., 2007; Cho et al., 2008a,b,c;
Grabinska-Sota and Kalka, 2006; Kulacki and Lamberti, 2008;
Lata1a et al., 2005; Matzke et al., 2007; Matzke et al., 2008; Pham
et al., 2008a,b; Pretti et al., 2009; Stolte et al., 2007a; Wells and
Coombe, 2006). Cho and co-workers used Pseudokirchneriella
subcapitata (formerly known as Selenastrum capricornutum) to
study the effect of different head groups, side chains and
anions of ILs on algal growth rate and photosynthetic activity.
The data revealed that the toxic influence of ILs on growth
rates were more significant than those of photosynthetic
performance (Pham et al., 2008b). Once again, the trend of
increasing toxicity with increasing alkyl chain length was
observed in their reports (Cho et al., 2007; Pham et al., 2008b).
Regarding the anionic effects, P. subcapitata was sensitive
to the anion moieties in the order: [SbF6] > [PF6] >
[BF4] > [CF3SO3] > [C8H17OSO3] > [Br] z [Cl] . In particularly,
it was found that with respect to IL incorporating perfluorinated
anion (i.e. IM14 BF4), EC50 values (concentrations
which lead to a 50% reduction of the exposed organisms
water research 44 (2010) 352–372 361
relative to control) of the previously prepared stock solution
(6 months prior to experiment) were significantly lower
compared to those of the freshly made one (Pham et al.,
2008a). This might be due to hydrolytic effects of IM14 BF4
leading to fluoride formation, as confirmed by ion chromatography
analysis. This implies that after ILs are released into
the aqueous system; they can become more hazardous than
expected by laboratory data with fresh ILs. In a detailed study
on hydrolysis of fluoride-containing anions, Cho et al. (2008a)
showed that IM14 SbF 6 generated a greater amount of fluoride
compared to IM14 BF 4, but no fluoride formation occurred
with the hexafluorophosphate. When only small amounts of
fluoride ions were formed from IM14 SbF6 and IM14 BF4 within
96 h, the formed fluoride ion did not affect the algal growth
rate. Nevertheless, the fluoride ion formation from IM14 BF4
increased with incubating time of the stock solution; thus, the
toxicity might significantly increase according to the further
formed fluoride ions. In view of cationic effect, Pyr14 Br was
found to be the least toxic of all the ILs tested to P. subcapitata
(Cho et al., 2008b). For the limnic green alga Scenedesmus
vacuolatus, a severe toxicity was found for 1-butyl-4-(dimethylamino)pyridinium,
whereas the quaternary ammonium
and morpholinium compounds exhibited no toxicity (Stolte
et al., 2007a). Despite the extensive studies on the toxicological
impact of ILs towards freshwater phytoplankton, inhibition
mechanism of both the growth rate and photosynthetic
activity by ILs has not been described by the authors.
Lata1a et al. (2005), who selected two marine algae Oocystis
submarina (green algae) and Cyclotella meneghiniana (diatom) as
testing organisms, found that the two species differed
dramatically in their ability to recover from IL exposure.
Additionally, it was discovered that IL toxicity declined with
increasing salinity. The lower toxicity of IL in this case is
probably due to the reduced permeability of IL cations through
the algal cell walls. High amounts of chloride provide a good
ion-pairing environment for imidazolium cations, which
consequently compete with hydroxyl or silanol functional
groups in the cell-wall structure of green alga and diatom,
respectively. Though no information on EC50 values was
described, the facts emerged from this work provide useful
information in the further fate assessment of ILs in marine
environments.
2.4. Cytotoxicity of ILs
As a cellular test system, promyelotic leukemia rat cell line
IPC-81 has been frequently used in cytotoxicity assays of ILs,
with the reduction of the WST-1 dye as an indicator of cell
viability (Matzke et al., 2007; Ranke et al., 2004; Ranke et al.,
2007a; Stasiewicz et al., 2008; Stolte et al., 2006; Stolte et al.,
2007b; Torrecilla et al., 2009). It was observed that ILs with
polar ether, hydroxyl and nitrile functional groups within the
side chains exhibited low cytotoxicity compared to those
incorporated with ‘‘simple’’ alkyl side chains (Kumar et al.,
2009; Stasiewicz et al., 2008; Stolte et al., 2007b). Those functional
groups were thought to impede cellular uptake by
membrane diffusion and reduce lipophilicity based interactions
with the cell membrane (Stolte et al., 2007b). Taking
a closer look at the effects of sub-structural elements of ILs,
[(CF3SO2)2N] anion and 4-(dimethylamino)pyridinium cation

362
were described to have intrinsic effects of anion and head
group on cytotoxicity, respectively (Stolte et al., 2007b). The
well known side chain length effect (decrease in EC50 values
with elongation of the alkyl side chain) could also be
confirmed in these studies.
To date many studies have analyzed the toxicity of ILs on
human cell lines (Frade et al., 2007; García-Lorenzo et al.,
2008; Hassoun et al., 2002; Kumar et al., 2009; Salminen et al.,
2007; Stepnowski et al., 2004; Wang et al., 2007). These in vitro
systems have been extremely beneficial in studying the
molecular basis of chemical’s biological activity, including its
toxic mode of action (Blaauboer et al., 1998) and could facilitate
extrapolation of in vitro data with regard to possible
effects on humans (Malich et al., 1997). Most of studies dealt
with HeLa cells exemplifying prototypical cells of the human
epithelium which is normally the site of first contact of an
organism with toxicants. According to Stepnowski et al.
(2004), the cytotoxicity data implied that effects of IM14
cation coupled with chloride, tetrafluoroborate or hexafluorophosphate
were probably dependent on the anionic
moieties. The lowest effect concentrations for tetrafluoroborate
species were found to be 0.63 mM, whereas
hexafluorophosphate and chloride inhibited HeLa cell growth
at comparably high concentrations of >10 mM. Surprisingly,
when the anion effect was compared, the strongest inhibition
was found for [PF6] . This might be due to hydrolysis
affecting fluoride formation, thus causing serious toxicological
consequences through the decomposition product.
A similar phenomenon was observed by Ranke et al. (2004) in
IPC-81 leukemia cells, where the lower toxicity of 1-n-butyl-3methylimidazolium
hexafluorophosphate in comparison to
the hexafluorophosphate anion alone was explained by
reduced anion uptake due to the formation of an ion pair.
The anion in this ion pair can, however, also be partially
decomposed. This was shown in recent work by Swatloski
et al. (2003), who identified traces of 1-n-butyl-3-methylimidazolium
fluoride hydrate as a decomposition product
formed during the purification of the 1-n-butyl-3-methylimidazolium
hexafluorophosphate.
As shown by Wang et al. (2007) the phosphonium
bis(trifluoromethylsulfonyl)imide salts performed the highest
inhibitory to HeLa cells, followed by alkylimidazolium,
alkylpyridinium, alkyltriethylammonium and N-alkyl-N,Ndimethyl-N-(2-hydroxylethyl)ammonium
salts, in decreasing
order. For each cation class the toxicity increased with
increasing chain length of the alkyl substituent for a given anion:
1-ethyl-3-methylimidazolium bromide yielded an EC50 of
8.4 mM, substituting the ethyl moiety for a butyl group led to an
EC50 of 2.8 mM, and for an octyl moiety an EC50 of 0.3 mM. This
result was consistent with what has been observed in other
studies. Salts containing the tetrafluoroborate anion showed the
highest EC50, followed closely by bromide and chloride. Bis(trifluoromethylsulfonyl)imide
salts were significantly more toxic
than their halide counterparts. However, the effect of changing
the anion was smaller than that of changing the alkyl substituent,
e.g. while 1-ethyl-3-methylimidazolium tetrafluoroborate
was observed to have an EC 50 of 9.9 mM, the corresponding
bromide and bis(trifluoromethylsulfonyl)imide salts had EC50 of
8.4 and 1.8 mM, respectively – these all considerably less
toxic than 1-octyl-3-methylimidazolium bromide.
water research 44 (2010) 352–372
The CaCo-2 cells were used in the study of García-Lorenzo
et al. (2008) with the aim of a convenient screening method
for obtaining first rough estimates for the toxic potential of
ILs. The obtained data showed that in general, ILs with longer
alkyl chains were more lipophilic than those with shorter
alkyl chains. The former can be presumed to have a tendency
to be incorporated into the phospholipid bilayers of biological
membranes. In this respect, some authors have indicated
that the increased toxicity of longer ILs can be accounted for
enhanced membrane permeability altering the physical
properties of the lipid bilayer (Lata1a et al., 2005; Ranke et al.,
2004; Stepnowski et al., 2004). Additionally, it has been
proposed that the mode of toxic action for ILs takes place
through membrane disruption because of the structural
similarity of imidazolium-based ILs to detergent, pesticides
and antibiotics able to cause membrane-bound protein
disturbance (Docherty and Kulpa, 2005). Recently, Ranke et al.
(2007a,b) have demonstrated that lipophilicity of ILs dominates
their in vitro cytotoxicity over a wide range of structural
variations. The contribution of the anionic part of the ILs to
the observed biological effect was evaluated by comparing
the EC 50 values obtained for the cations IM16 and IM18,
combined with two different anions [Cl] and [PF 6] . For both
cations, a stronger toxic effect was found for chloride derivatives,
but not for fluoride containing hexafluorophosphate.
A similar result was reported by Stock et al. (2004) where the
inhibitory effects of IM14 Cl and IM14 PF6 on the acetylcholinesterase
activity were compared. In addition, slightly
higher cytotoxicity for the chloride derivative has also been
observed when the cytotoxicity of IM14 Cl and IM14 PF6 on
HeLa cells was tested (Stepnowski et al., 2004). This implies
the effect of perfluorinated ions is not drastic to all but vary
according to species of organisms tested. Several authors
have pointed out that altering the anion has only minimal
effects on the toxicity of several imidazolium compounds
(Bernot et al., 2005a; Garcia et al., 2005; Ranke et al., 2004).
This indicates that ILs toxicity seems to be related to the alkyl
chain branching and to the hydrophobicity of the imidazolium
cation but not to the various anions. In this respect,
a recent study using the IPC-81 rat leukemia cell line with
a large pool of anions demonstrated that most of the
commercially available anions showed no or only marginal
cytotoxic effects. However, anionic compartments with
lipophilic and hydrolysable structural elements are likely to
be of considerable relevance with respect to the toxicity of ILs
(Stolte et al., 2006).
In a recent study (Frade et al., 2007), the human cell lines
such as HT-29 and CaCo-2 cells were utilized to estimate the
inhibitory effect of ILs with several types of cations and
anions. In both cells, IM14, IM12OH (1-(2-hydroxyethyl)-3methylimidazolium),
IM12O2O1 (1-(2-(2-methoxyethoxy)ethyl)-3-methylimidazolium)
and cholines were the least
toxic cations independently of the anion. Within the studied
combinations, it can be noted that IM14 PF 6, IM14 acesulfame,
IM12OH BF 4/PF 6, IM12O2O1 BF 4/PF 6, IM12OH acesulfame and
IM12OH saccharine are not toxic and present good alternatives
to organic solvents. Meanwhile, increasing the length of
the substituent chain may contribute to a significant
increasing of imidazolium toxicity. It was also noted that
[(CF3SO2)2N] anion decreased the toxicity to a large extent,

independently of the cation and for both cell types, which was
in accordance with Salminen et al. (2007).
2.5. Phytotoxicity of ILs
The studies on phytotoxic activity of ILs were conducted
mostly on the duckweed, Lemna minor, a common aquatic
vascular plant (Jastorff et al., 2005; Larson et al., 2008; Matzke
et al., 2007; Stolte et al., 2007a). In general, 1-alkyl-3methylimidazolium
compounds with longer alkyl chains were
more toxic to L. minor than those with short alkyl chain
lengths. Imidazolium and pyridinium cations with butyl
groups had similar EC50s (the concentrations that produced
a 50% reduction in root growth) (39.07 and 32.54 mM, respectively);
while the equivalent ammonium cation had a much
higher EC50 (101.48 mM; i.e., less toxic) (Larson et al., 2008). In
consideration of anionic effect, [(CF 3SO 2) 2N] was found to
cause moderate toxicity to this duckweed (EC 50 ¼ 6300 mM)
(Matzke et al., 2007). On the other hand, this anion had no or
even a positive influence on the observed effects on L. minor
(Stolte et al., 2007a).
Focusing on the terrestrial environment, Matzke et al.
(2009a) investigated the influence of differently composed
soils, with varying contents of the clay minerals smectite and
kaolinite, on the toxicity of different anion species of imidazolium-based
ILs towards the wheat Triticum aestivum. The
data showed that IM14 (CF3SO2)2N appeared the most toxic,
independently of the type and concentration of added clay.
This is totally in contrast to the findings of Stolte et al. (2007a),
who reported that [(CF 3SO 2) 2N] caused no harm to L. minor,
indicating the toxic effect of this anion is different between
certain plants. The toxicity of 1-butyl-3-methylimidazolium
incorporated chloride, tetrafluoroborate and hydrogen sulfate
was mainly controlled by the cationic moiety. The observed
effects varied according to the added clay type and clay
concentration. An increase of clay content resulted in less
inhibitory effects of these substances. On the contrary, for
IM14 combined with bis(trifluoromethylsulfonyl)imide the
addition of clay minerals led to higher toxicity compared to
the reference soil. Since results are contradictious further
study is necessary to unravel the underlying mechanism.
Moreover, a detailed study on the effect of IM14 BF 4 on the
wheat T. aestivum seedlings (Wang et al., 2009) showed that
IM14 BF4 was hazardous to the early development of wheat
and had varying effects on different organs. At low concentrations,
IM14 BF4 did not inhibit, and even promoted, wheat
seedling growth. Nonetheless, at high concentrations, this IL
inhibited wheat seedling growth significantly and decreased
chlorophyll content, thereby reducing photosynthesis and
plant growth. Therefore, the authors suggested that dilution
could decrease the toxicity of IM14 BF 4 to plants and would be
a good method for remediating IL-polluted environments.
In another research, the phytotoxicity tests of chiral ILs
containing (-)-nopyl derivatives were carried out in a plant
house using spring barley (Hordeum vulgare) which is a monocotyledonous
plant, and a common radish (Raphanus sativus L.
subvar. radicula Pers.) which is a dicotyledonous plant (Ba1czewski
et al., 2007). According to the data obtained, increasing
the concentration of ILs resulted in a systematic decrease in
the crop fresh weight of total sprouts and the crop fresh
water research 44 (2010) 352–372 363
weight per plant, both for spring barley and for common
radish. It could also be noted that common barley was a more
resistant plant which fairly well tolerates test IL concentrations
up to 200 mg kg 1 of soil; whereas, for radish, the growth
and development inhibiting concentration is 100 mg kg 1 of
soil. Using the same target plant (H. vulgare), Pernak et al.
(2004b) reported that the 1,3-dialkoxymethylimidazolium
tetrafluoroborate salts introduced to the soil at concentration
of 1,000 mg kg 1 , or 100 mg kg 1 dry mass of soil, were found
to exert a phytotoxic effect on monocotyledonous plants.
On the other hand, at a concentration of 10 mg kg 1 no such
effect on the growth of the roots was notified. Concerning
phytotoxicity of ILs to garden cress (Lepidium sativum L.) in soil
environment, Studzińska and Buszewski (2009) have proved
that hazardous effects of imidazolium ILs are closely
connected with organic matter content in soil. Soil with more
organic carbon was observed to sorb IL cations more extensively
than soil with little or no organic matter; hence, the
more fertile in soil, the lower probability of hazardous effect of
ILs to plants. On the other hand, the hazardous character of
analyzed ILs was strongly connected with their hydrophobicity,
indicating that the more hydrophobic IL, the higher
decrease of seed germination.
Although intensive work has not been conducted on
phytotoxic influence of ILs, the available data offer initial
hints for environmental scientists dealing with the potential
impact of ILs towards aqueous and terrestrial plants.
2.6. Toxicity of ILs to invertebrates
Ecotoxicological literature of ILs to invertebrates mainly focus
on the use of Daphnia magna as a test organism (Bernot et al.,
2005a; Couling et al., 2006; Garcia et al., 2005; Grabinska-Sota
and Kalka, 2006; Luo et al., 2008; Nockemann et al., 2007; Pretti
et al., 2009; Samorì et al., 2007; Wells and Coombe, 2006; Yu
et al., 2009b). Daphnia is an important link between microbial
and higher trophic levels (McQueen et al., 1986), and has been
the subject of hundreds of intensive ecological studies. The
results of all studies again observed the well-established link
between toxicity and alkyl chain length of the tested ILs
containing imidazolium, pyridinium or quaternary ammonium
as counter cations. The most toxic compound towards
D. magna was found to be IM18 Br whereas the least toxic one
was IM14 Cl with log10EC50 values of 1.33 and 1.93, respectively
(Table 2). Also, the nature of the anion was suggested to
have smaller effects compared to those of the cation. In a
recent study, Luo et al. (2008) investigated the developmental
toxicity of IM18 Br on D. magna. It was found that this
compound exhibited toxicity on the development of three
generation of D. magna with the decrease of number of
offspring and average brood size correlated to increasing IM18
Br concentrations. This indicated that IM18 Br could cause
deleterious effect to the population of Daphnia and indirectly
disturb freshwater food webs. Couling et al. (2006) used
experimental data to determine which part of the IL molecule
is responsible for the observed toxic effects through a quantitative
structure-property relationship (QSPR) modeling. In this
respect, correlative and predictive equations were generated
and proved that there was a distinct influence of the length of
alkyl residues attached to the aromatic nitrogen atoms

364
towards D. magna. Moreover, the models predicted that the
toxicity increased slightly with increasing number of aromatic
nitrogen atoms in cation ring. This implies that ammonium
salts are less toxic than pyridinium salts, which in turn are
less toxic than imidazolium moieties. Interestingly, it was
noted that methylating the aromatic carbons could be
effective in reducing toxicity to D. magna, indicating that 1-nbutylpyridinium
bromide can be more toxic than 1-n-butyl-3methylpyridinium
bromide, which is more toxic than 1-nbutyl-3,5-dimethylpyridinium
analogue. The QSPR, though is
still in its infancy, has contributed initial guidelines for
a rationale design of a new category of ILs with an acceptable
environmental profile.
Other studies include data on the snail Physa acuta (Bernot
et al., 2005b), the spring tail Folsomia candida (a soil invertebrate)
(Matzke et al., 2007), Caenorhabditis elegans (a soil
roundworm) (Swatloski et al., 2004) and Dreissena polymorpha
(zebra mussel) (Costello et al., 2009). It was also demonstrated
a positive relationship between alkyl chain length and toxicity
in these reports. In the research of Bernot et al. (2005b), the
estimated LC 50 (median lethal concentration) ranged from
3.50 to 1799.8 mM (0.54 to 3.26 in the logarithmic form), which
implied that P. acuta are less sensitive to ILs than are D. magna
(log10LC50 (mM) ranging from 1.33 to 1.93 (Table 2)). Also, the
authors observed that at low concentrations, the IL may
suppress snail movement, but concentrations above this
threshold level trigger an escape response, causing the
organism to move faster. Grazing patterns, nonetheless,
showed that snails grazed less at higher IL concentrations.
Physa spp. are key components of freshwater food webs,
because they graze algae and are themselves important prey
for fish and invertebrate predators (Bernot and Turner, 2001;
Osenberg and Mittelbach, 1989). Thus, nonlethal IL concentrations
affected P. acuta behaviors, potentially influencing
individual fitness and good web interactions.
2.7. Inhibitory effects of ILs on vertebrates
Zebrafish (Danio rerio) plays an important role in ecotoxicology
as a prominent model vertebrate. Concerning toxicity of ILs to
the zebrafish, Pretti et al. (2006) revealed that ILs may cause
a completely different effect on fish according to their chemical
structures. As imidazolium, pyridinium and pyrrolidinium
showed a LC50 (lethal effect) >100 mg L 1 , they could be
regarded as non-highly lethal towards zebrafish. On the other
hand, the ammonium salts showed LC50 remarkably lower
than that reported for organic solvents and tertiary amines.
In general, these data referred that fish are less sensitive to ILs
toxicity compared to other species belonging to lower trophic
levels.
In a recent report, Li et al. (2009) used the frog Rana nigromaculata
as an amphibian model for toxicity testing.
Amphibians are often the main vertebrate group prone to
contaminant exposure in aquatic systems mostly because
their larvae live in water (Lahr, 1997; Mann and Bidwell, 2000).
In their study, they evaluated the toxic effects of IM18 Br on
the early embryonic development of the frog R. nigromaculata.
The results demonstrated that the highest embryonic
mortality occurred in the neural plate stage, followed by the
early gastrula and early cleavage stages with the LC50 values
water research 44 (2010) 352–372
being 42.4, 43.4 and 85.1 mg/L, respectively, indicating that the
developmental toxicity of IM18 Br in the frog was stagesensitive.
The number of dead embryos was also found to
increase with the increasing concentrations of the IL IM18 Br.
The developmental impact of IM18 Br was claimed not only in
this finding but also in the work of Luo et al. (2008), who
investigated on D. magna.
Other work in the literature has focused on the acute
toxicity of ILs on rats and mice (Bailey et al., 2008; Cheng et al.,
2009; Landry et al., 2005; Pernak and Czepukowicz, 2001; Sipes
et al., 2008). The values of acute toxicity of 3-hexyloxymethyl-
1-methylimidazolium tetrafluoroborate were found to be
LC50 ¼ 1400 and 1370 mg kg 1 for female and male Wistar rats,
respectively (Pernak and Czepukowicz, 2001). Bailey et al.
(2008) studied the effects of prenatal exposure of mice to IM14
Cl due to the potential for human exposure as a result of water
or soil contamination from industrial effluent or accidental
spills. As shown in the experimental data, after being contacted
to the IL, fetal weight was considerably reduced at the
two highest concentrations (169 and 225 mg kg 1 d 1 ).
Malformations were also somewhat more numerous at the
highest dosage, suggesting that IM14 Cl may be teratogenic.
Maternal toxicity was also present, indicating that IM14 Cl
appeared to be developmentally toxic at maternally toxic
dosages. Also, IM14 Cl has been shown to cause thermal irritation
when applied topically to rats, but produced only
minimal contact sensitization when evaluated in the mouse
local lymph node assay (Landry et al., 2005). Additionally, in
this report, it is worth noting that the transdermal toxicity of
IM14 Cl was influenced by the vehicle of administration. Use
of the organic solvent, dimethylformamide, accentuated the
acute toxicity. Very high concentrations of IM14 Cl (up to 95%
IM14 Cl in water) applied to the rat skin were markedly less
acutely toxic. This result may have a practical guideline that to
reduce the acute toxicity, ILs can be handled in pure form with
water as a co-solvent.
3. Environmental fate of ILs
3.1. Chemical degradation of ILs
Ionic liquids possess excellent chemical and thermal stability,
which gives, unfortunately, a negative aspect for their treatment
after usage prior to disposal. To assess the persistence of
ILs in the environment as well as verify possibilities of their
cleanup by chemical methods, several groups have focused
their attention on oxidative and thermal degradation of ILs in
aqueous media (Awad et al., 2004; Baranyai et al., 2004;
Berthon et al., 2006; Itakura et al., 2008; Li et al., 2007; Morawski
et al., 2005; Siedlecka and Stepnowski, 2009; Siedlecka
et al., 2008a,b; Stepnowski and Zaleska, 2005). Pioneering work
in the field of oxidative degradation was done by Stepnowski
and Zaleska (2005) and Morawski et al. (2005) who showed that
the greatest degradation efficiency for imidazolium ILs was
achieved with a combination of UV light and a catalytic
oxidant such as hydrogen peroxide or titanium dioxide.
Subsequently, Li et al. (2007) studied the oxidative degradation
of 1,3-dialkylimidazolium ILs in hydrogen peroxide/acetic acid
medium assisted by ultrasonic chemical irradiation. It was

observed that 99% of tested compounds was degraded after
72 h. In addition, advanced oxidative degradation in the
presence of reactive peroxides generated by Fenton reagent
has been applied for the removal of ILs from water (Siedlecka
and Stepnowski, 2009; Siedlecka et al., 2008a,b). According to
the results, in a Fenton system with 1 mM of Fe(III) and
100 mM of H2O2, more than 97% of IM14 Cl was observed to
degrade after 90 min. For Pyr14 Cl, IM16 Cl and IM18 Cl, the
levels of degradation were 92%, 88% and 68%, respectively.
Investigations of the degradation mechanisms indicated IM18
Cl was more resistant to oxidation by OH radicals cleaved
from H 2O 2, suggesting that the oxidation rates of imidazolium
ILs by OH are structure-dependent (Siedlecka and Stepnowski,
2009). The level of degradation was dependent on the alkyl
chain length, consistent with Stepnowski and Zaleska (2005),
who indicated that lengthening the alkyl chain lowered the
rate of IL degradation. On contrast, the different length of the
side chains and the type of anions did not affect the degradation
process (Li et al., 2007). Regarding the thermal degradation
studies of alkylimidazolium salts (Awad et al., 2004),
extension of the alkyl chain enhanced the thermo-oxidative
degradation of imidazolium salts. Interestingly, methyl
substitution in the 2-position (i.e. between the two N atoms)
was observed to decrease the oxidative decomposition of
imidazolium ILs. The longer alkyl chain was also observed to
induce an enhancement in photocatalytic decomposition of
ILs (Morawski et al., 2005), which was not in the case of
oxidative degradation (Stepnowski and Zaleska, 2005; Siedlecka
and Stepnowski, 2009). Nonetheless, detailed account on
the degradation of ILs by photocatalysis is required to verify
this phenomenon.
3.2. Biodegradability of ILs
In contrast to chemical degradation, which requires the
assistance of a certain oxidant for catalysis, biodegradation is
the microbial breakdown of chemical compounds. Biodegradation
seems to be more environmentally friendly compared
to chemical decomposition process. The initial attempt to
examine the degradation potential of different IM14 cations
combined with [Br] , [BF 4] , [PF 6] , [N(CN) 2] , [(CF 3SO 2) 2N]
and octylsulfate as the counter ion was done using the Sturm
and Closed-Bottle test protocols by the group of Scammells
(Garcia et al., 2005; Gathergood and Scammells, 2002; Gathergood
et al., 2004; Gathergood et al., 2006). Nonetheless, no
compound showed significant degree of biodegradation with
the exception of the octylsulfate-containing IL. The next step
study on the biodegradation of ILs involved the design of ILs
containing biodegradable side chains (Gathergood and
Scammells, 2002). The design was done according to the
principles of Boethling (Boethling, 1994, 1996; Howard et al.,
1991) who identified three important parameters including
the potential sites of enzymatic hydrolysis (for example,
esters and amides) and oxygen in form of hydroxyl, aldehyde
or carboxylic acid groups as well as unsubstituted linear alkyl
chains (especially 4 carbons) and phenyl rings, which
represent possible sites for attack by oxygenases. However,
for a balance between chemical properties and biodegradability,
not all of these factors were suitable for ILs. The
water research 44 (2010) 352–372 365
addition of oxygen containing functional groups such as
alcohols, aldehyde and carboxylic acids was reported to
restrict the ILs performance as reaction media whereas the
incorporation of phenyl rings was known to increase the
melting points of IL solvents (McGuinness and Cavell, 2000).
Therefore, ester or amide group was selected to be coupled in
alkyl side chain of ILs. The introduction of ester groups
derived from a C2 acid and C4 or higher alcohol in the
3-N-substitutent was demonstrated to increase the biodegradation
of imidazolium-based ILs (Gathergood et al., 2004). This
can be explained by the fact that introduction of ester moiety
probably provides a site susceptible to enzymatic attack
(Gathergood et al., 2004; Gathergood et al., 2006) and hence,
improves the biodegradation level. Though the addition of
amide group is informed to improve the biodegradation of
organic compounds (Boethling, 1994, 1996; Howard et al.,
1991), no critical enhancement of biological degradation was
noted when this group was appended into the imidazoliumbased
ILs (Gathergood et al., 2004). However, no compound
could be classified as ‘‘readily biodegradable’’ corresponding
to Organization for Economic Cooperation and Development
(OECD) standards (U.S. EPA, 1998), for which 60–70% or greater
biodegradation by activated sludge microbial inoculate is
required within a 10-day window in a 28-day period. Finally,
the combination of the octylsulfate anion and imidazolium
cation containing ester side chains resulted in readily biodegradable
IL (Gathergood et al., 2006). Recently, Stolte et al.
(2008) also paid their attention on investigation of functional
groups incorporating alkyl chain ILs. Nonetheless, the introductions
of terminal hydroxyl, carboxyl, ether and nitrile
groups did not improve the biological degradation as
expected.
Kumar et al. (2006) investigated the fate of IM14 BF 4 when
in contact with soil-microorganisms, wastewater microorganisms,
Pseudomonas putida and E. coli. Although IM14 BF4
was indicated to be recalcitrant in Sturm and Closed-Bottle
test assays as mentioned above, it was observed in this study
that P. putida was able to break down IM14 BF4 after 15 days of
incubation. The breakdown products were monitored
using GC-MS and identified to be 1-H-methylimidazole and
1-H-butylimidazole, which were in consistent with the
theoretical metabolism scheme proposed by Jastorff et al.
(2003). In case of bacteria from soil and wastewater, the
metabolic intermediates appeared on the 12th day. It was
also noted that different intermediate peaks were observed at
different retention time with different microbes, indicating
that the degradation mechanism of IM14 BF4 may vary in
correspondence to certain microbes and metabolic pathways.
In another study, the biodegradation pathway of IM18 moiety
(Fig. 3) was proposed based on intermediate products via
HPLC-MS analysis after 24-day period of incubation with
activated sludge (Stolte et al., 2008). The metabolism of IM18
cation appeared to undergo oxidation reactions catalyzed
probably by mono-oxygenases, e.g. the cytochrome P 450
system on the terminal methyl group (u-oxidation). The
alcohol formed was subsequently oxidized and converted
into aldehydes, and then into carboxylic acids by dehydrogenases.
The resulting carboxylic acids then might undergo
b-oxidation and finally generated two carbon fragments that
can enter the tricarboxylic acid cycle as acetyl Co-A (Fig. 3).

366
retention time
in min.
8.9
13.5 / 14.4
12.2 / 12.7
10.0 – 12.5
19.5
16.2
26.5
24.2
26.7
m/z +
195
211
209
225
183
197
155
169
141
intensity
4*10 5
3*10 5 / 2*10 5
1*10 6 / 2*10 6
2*10 4 – 6*10 4
1*10 5
3*10 5
0.5*10 5
4*10 6
1*10 5
N N +
N N +
N N +
N N +
N N +
N N +
N N +
N N +
O
OH
The proposed pathways provide basic information to both
environmental scientists and chemical engineers; however,
no studies have sought to examine the toxicity of metabolic
products after degradation of ILs. This issue is of paramount
importance since metabolism might not always end in less
toxic products.
Subsequently, Wells and Coombe (2006) extended the
microbial degradation study with ammonium, imidazolium,
phosphonium and pyridinium compounds by measuring the
biological oxygen demand. The authors observed no biodegradability
of cations incorporated short chains (C 4) within
this test series, which was in agreement with Docherty et al.
(2007) and Stolte et al. (2008). For longer alkyl chains (C 12, C 16
and C 18) containing ILs, a strong inhibitory effect of these
compounds on the inoculum used was found, indicating the
active microbial consortium was significantly impacted by ILs
toxicity. In recent studies (Docherty et al., 2007; Grabinska-
Sota and Kalka, 2004; Harjani et al., 2008; Stasiewicz et al.,
2008), pyridinium-based ILs were reported to be fully catabolized
by microbial community in activated sludge. This can be
inferred from the fact that degradation pathways for pyridine –
water research 44 (2010) 352–372
OH
OH
O
OH
O
OH
H
N N +
O
O
OH
OH
N N +
N N +
N N +
the precursor of pyridinium-based compounds – under
aerobic and anaerobic conditions were intensively investigated
in the work of Kaiser et al. (1996). With respect to the
common 1,3-dialkylpyridinium ILs, Pham et al. (2009) reported
that after 21 days of incubation, microorganisms from activated
sludge were able to break down Py4-3Me Br. Analyses of
HPLC and MS/MS demonstrated that this biodegradation
led to the formation of 1-hydroxybutyl-3-methylpyridinium,
1-(2-hydroxybutal)-3-methylpyridinium, 1-(2-hydroxyethyl)-
3-methylpyridinium and methylpyridine. Based on these
intermediate products, biodegradation pathways were also
suggested (Fig. 4), thereby providing the basic information
which might be useful for assessing the factors related to the
environmental fate and behavior of this commonly used
pyridinium IL. Although this is the first report on biodegradation
intermediates and pathway of pyridinium ILs, the
authors have failed to systematically screen a single microorganism
or a microbial consortium responsible for biodegradation
of ILs. Therefore, it is needed to further investigate
which type of microorganism is adaptable to ILs and which is
responsible for degradation process. Also, the broken
OH
O
OH O
+
Fig. 3 – Biodegradation pathways of 1-octyl-3-methylimidazolium by activated sludge microbial community (Reproduced
from Stolte et al. (2008) by permission of the Royal Society of Chemistry).
m/z +
211
209
225

C
H 3
C
H 3
N +
N +
C
H 3
structures much further than methylpyridinium were not
measured in the study. In particular, the possibility of
cleavage of heterocyclic ring in ILs molecular structure (both
pyridinium in this work and imidazolium in the study of Stolte
et al. (2008)) has not been indentified. This issue should be
clarified through further studies. Interestingly, Py4-3Me Br
was not found to be metabolized by the activated sludge
community (Docherty et al., 2007). This was attributed likely
to the high IL concentration used in the study of Docherty’s
group, which consequently inhibited the microbial consortium.
Nonetheless, it was demonstrated that the structural
manipulation of the pyridinium skeleton may lead to ILs with
greater biodegradable extent compared to imidazolium-based
compouds (Harjani et al., 2008).
Concerning the anionic effect, ILs with halide counter ions
were postulated to be more stable to degradation than
perfluoronated ions (Awad et al., 2004; Gathergood and
Scammells, 2002). In a preliminary study, Gathergood and
Scammells (2002) confirmed this assumption and showed that
the biodegradation efficiency decreased in the order
[PF 6] > [BF 4] > [Br] with 60%, 59% and 48% of CO 2 evolution
values, respectively. In a later study (Gathergood et al., 2006),
it was found that the octylsulfate anion is considerably more
biodegradable than the other commonly used anions.
The alkyl chain with C4, C6 or C8 was found to increase the
rate of degradation (Docherty et al., 2007; Stolte et al., 2008);
nonetheless, further increasing the chain length to C12, C16 or
C18 was noted to cause toxic effect towards inoculum (Wells
and Coombe, 2006). However, it was stated that the long octyl
side chain was not a compulsory factor for biodegradation, but
more important is a certain overall lipophilicity of the
compound (Stolte et al., 2008).
I
C
H 3
C
H 3
OH
H
+
+
N +
N +
N +
water research 44 (2010) 352–372 367
OH
C
H 3
C
H 3
CH 3
OH
O
O
OH
II
C
H 3
N + C H3 CH3 OH
N +
OH
+ C
H 3
Fig. 4 – Biodegradation pathways of 1-butyl-3-methylpyridinium entity by microorganisms from activated sludge (Reprinted
with permission from Pham et al. (2009). Copyright 2009 American Chemical Society).
C
H 3
N +
H
CH 3
3.3. Sorption of ILs in environmental systems
+
C
H 3
Because the aquatic and terrestrial environments are possible
recipients for contaminants, the distribution and behavior of
ILs in soil are also extremely important. The retention and
mobility of ILs in soils and sediments are strongly influenced by
its tendency to be sorbed onto the various components of the
soil matrix (Stepnowski, 2005). Since imidazolium-based ILs
possess high electron acceptor potential of delocalized
aromatic systems and hydrophobic components (e.g. the alkyl
chain) (Stepnowski, 2005), they could be sorbed onto soils and
sediments via several mechanisms. In a preliminary study,
Stepnowski (2005) proved that electrostatic interactions
contributed to the sorption of the imidazolium cations. Moreover,
totally contrast trends were also observed demonstrating
an extremely strong and practically irreversible sorption onto
fine-textured marine sediments and a relatively weakly and
reversibly binding to peaty soil (with the highest organic
carbon content) (Stepnowski, 2005). This indicates the importance
of the mineral component of the soil (sediment) in the
sorption mechanism of ILs. Also, in this work, the author
pointed out that compounds with longer alkyl chains were
irreversibly bound to the soil component, which was in
agreement with Stepnowski et al. (2007) and Matzke et al.
(2009b). Interestingly, Beaulieu et al. (2008) did not find a positive
effect of alkyl chain length on the sorption of alkylimidazolium-based
ILs to aquatic sediments and suggested that
hydrophobic interactions were not the most important sorption
mechanism. The contrast results between these groups
imply that sorption mechanisms of ILs may vary according to
properties and composition of the environmental systems. The
studies suggest that ILs may be retained by aquatic sediments;
OH

368
nonetheless, the toxic action of these sorbed materials towards
aquatic organisms has not been addressed. Also, further
efforts should be continued to elucidate the reversibility of IL
sorption on these sediments to give a better understanding of
the ILs fate after being released into aqueous environment.
Matzke et al. (2009b) investigated the influences of the two
different clay minerals kaolinite and smectite as well as of
organic matter on the cation sorption and desorption behaviors
of three imidazolium based ILs including IM14 BF 4, IM18
BF 4 and IM14 (CF 3SO 2) 2N in soil. The addition of organic matter
and clays was observed to increase the sorption/decrease the
desorption of all ILs tested, and in particular smectite had
striking effects on the sorption efficiency of all substances.
It is worth noting that not only the cationic moiety with
different alkyl side chain lengths but the anionic compartments
can also play an importance role in the sorption/
desorption processes. Imidazolium compounds with BF4 as
a counter anion showed higher sorption capacity compared to
that of [(CF3SO2)2N] , indicating the high potential of this type
of IL to form ionic pairs in the soil matrix (Matzke et al., 2009b).
However, further work with a variety of ILs incorporated
different cationic and anionic moieties should be carried out
to verify these phenomena.
Gorman-Lewis and Fein (2004) examined the sorption
behavior of IM14 Cl onto a range of surfaces which are
commonly found in the near-surface environment. The results
suggested that IM14 Cl could be minimally retarded by noninterlayer
clay system and might lead to unimpeded transport
through subsurface groundwater. Also, the adsorption
capacity of this IL onto bacterial surfaces was not high, which
might be due to the low hydrophobicity of IM14 Cl. Additionally,
investigations of the adsorption of IM14 Cl towards
different media carried out by our group (Vijayaraghavan et al.,
2009) have shown that retardation of this compound was
possible only by an ion-exchange resin and activated carbon,
which was in consistency with the work of Anthony et al.
(2001). However, no significant adsorption of IM14 Cl in the
media of a fermentation waste (Corynebacterium glutamicum)
and dried activated sludge was observed in our study.
Conclusively, the data currently available demonstrated
that ILs incorporated imidazolium cation can be sorbed to
organic matter, whether found in aquatic sediments or
terrestrial soils, and the presence of clays significantly
enhanced the sorption capacity.
4. Concluding remarks and future directions
Ionic liquids, of which the most often cited attribute is their
negligible vapor pressure, have been suggested as a green
alternative to traditional organic solvents with the desire to
minimize diffusion to the atmosphere. Low volatility, however,
does not completely eliminate potential environmental
hazards and might pose serious threats to aquatic and terrestrial
ecosystems. The studies of environmental fate and
toxicity of ILs have shown that the ILs commonly used to date
are toxic in nature and their toxicities vary considerably across
organisms and trophic levels. In general, the effect of anionic
moieties is not drastic as the alkyl length effect except for the
case of [(CF3SO2)2N] , which shows a clear (eco)toxicological
water research 44 (2010) 352–372
hazard potential. The other perfluorinated anions have been
also proved to be hazardous due to hydrolytically unstable
properties. In addition, the introduction of functional polar
groups to the alkyl chain has been shown to reduce the toxicity
of ILs and increase the biodegradation efficiency to some
extent. This indicates the possibility of tailoring ILs by coupling
suitable functional groups to their structure, which in turn
leads to a more environmental friendly compound. The side
chain length effect has been found to be consistent in all levels
of biological complexity as well as different environmental
compartments. Also, an increase in alkyl-chain length, or lipophilicity,
was observed to be related to an increase in the rate
of degradation as well as an increase in toxicity. This indicates
a conflict of aims between minimizing the toxicity and maximizing
the biodegradability of these neoteric solvents.
Regarding the cationic compartment, pyridinium has been
found to be more environmental friendly than imidazolium
from both viewpoints of toxicology and microbial degradation.
It can therefore be suggested that the structural manipulation
of the pyridinium skeleton should be considered in design of
a sustainable IL. From the currently available data, it is clear
that some commonly used ILs are very far away from the image
of green chemicals that are often cited in the literature. The
uncertainties in their sustainable development hinder the
applications of ILs under real conditions. Although some
attempts have been made to give important hints in the
prospective design and synthesis of inherently safer ILs,
comprehensive studies dealing with the behaviors of ILs in
aqueous media still await to be conducted. The important
features required for the thorough insight into environmental
fate of ILs include, but are not limited to:
- Providing more fundamental understanding into the
mechanism for IL-induced toxicity to different levels of
biological complexity. The underlying mechanisms of IL
toxicity have rarely been studied.
- Assessing the biodegradability of cationic and anionic
compartments and toxicity of their degradation intermediates.
This may provide useful information in consciously
designing safer chemicals.
- Investigating the aerobic and anaerobic biodegradation of
ILs, which would suggest initial guidelines for the treatment
of ILs waste by using the existing aerobic and anaerobic
wastewater treatment facilities. Especially, anaerobic
degradation awaits to be investigated.
- Defining which organisms or enzymes may promote
degradation pathways and determining specific microbial
consortium or cultivatable communities capable of
biotransformation of ILs.
- Performing the ecotoxicity and biodegradation tests in real
environmental conditions instead of controlled conditions
of laboratory experiments, which would be advantageous in
understanding the fate and behavior of ILs under real
conditions. For this, the potential toxicological effects at
population level and community level should be addressed.
It must be encouraged to use tools such as experimental
mesocosms to study the effects of ILs at higher levels of
organization.
- Creating database of environmentally benign structure
moieties of ILs based upon their toxicological and

iodegradation information, which would be practically
useful as a reference for manufacturers and regulators to
properly develop and regulate the use of ILs.
Acknowledgements
This work was supported by NRF Grant funded by the Korean
Government (KRF-2007-521-D00106, NRL 2009-0083194, and in
part WCU R31-2008-000-20029-0).
references
Anthony, J.L., Maginn, E.J., Brennecke, J.F., 2001. Solution
thermodynamics of imidazolium-based ionic liquids and
water. J. Phys. Chem. B 105, 10942–10949.
Arning, J., Stolte, S., Böschen, A., Stock, F., Pitner, W.-R., Welz-
Biermann, U., Jastorff, B., Ranke, J., 2008. Qualitative and
quantitative structure activity relationships for the inhibitory
effects of cationic head groups, functionalized side chains and
anions of ionic liquids on acetylcholinesterase. Green Chem.
10, 47–58.
Austin, R.P., Barton, P., Davis, A.M., Manners, C.N., Stansfield, M.C.,
1998. The effect of ionic strength on liposome-buffer and
1-octanol-buffer distribution coefficients. J. Pharm. Sci. 87,
599–607.
Awad, W.H., Gilman, J.W., Nyden, M.,Harris,R.H.,Sutto,T.E.,
Callahan, J., Trulove, P.C., DeLong, H.C., Fox, D.M., 2004.
Thermal degradation studies of alkyl-imidazolium salts and
theirapplicationinnanocomposites. Thermochim. Acta
409, 3–11.
Babalola, G.O., 1998. Anti-bacterial activity of synthetic
N-heterocyclic oxidizing compounds. Lett. Appl. Microbiol. 26,
43–46.
Bailey, M.M., Townsend, M.B., Jernigan, P.L., Sturdivant, J., Hough-
Troutman, W.L., Rasco, J.F., Swatloski, R.P., Rogers, R.D.,
Hood, R.D., 2008. Developmental toxicity assessment of the
ionic liquid 1-butyl-3-methylimidazolium chloride in CD-1
mice. Green Chem. 10, 1213–1217.
Ba1czewski, P., Bachowska, B., Bia1as, T., Biczak, R., Wieczorek, W.
M., Balińska, A., 2007. Synthesis and phytotoxicity of new
ionic liquids incorporating chiral cations and/or chiral anions.
J. Agric. Food Chem. 55, 1881–1892.
Baranyai, K.J., Deacon, G.B., MacFarlane, D.R., Pringle, J.M., Scott, J.L.
, 2004. Thermal degradation of ionic liquids at elevated temperatures.
Aust. J. Chem. 57, 145–147.
Beaulieu, J.J., Tank, J.L., Kopacz, M., 2008. Sorption of imidazoliumbased
ionic liquids to aquatic sediments. Chemosphere 70,
1320–1328.
Bernot, R.J., Turner, A.M., 2001. Predator identity and traitmediated
indirect effects in a littoral food web. Oecologia 129,
139–146.
Bernot, R.J., Brueseke, M.A., Evans-White, M.A., Lamberti, G.A.,
2005a. Acute and chronic toxicity of imidazolium-based
ionic liquids on Daphnia magna. Environ. Toxicol. Chem. 24,
87–92.
Bernot, R.J., Kennedy, E.E., Lamberti, G.A., 2005b. Effects of ionic
liquids on the survival, movement, and feeding behavior of
the freshwater snail, Physa acuta. Environ. Toxicol. Chem. 24,
1759–1765.
Berthon, L., Nikitenko, S.I., Bisel, I., Berthon, C., Faucon, M.,
Saucerotte, B., Zorz, N., Moisy, Ph, 2006. Influence of gamma
water research 44 (2010) 352–372 369
irradiation on hydrophobic room-temperature ionic liquids
[BuMeIm] PF 6 and [BuMeIm] (CF 3SO 2) 2N. Dalton Trans, 2526–2534.
Blaauboer, B.J., Balls, M., Barratt, M., Casati, S., Coecke, S.,
Mohamed, M.K., Moore, J., Rall, D., Smith, K.R., Tennant, R.,
Schwetz, B.A., Stokes, W.S., Younes, M., 1998. Alternative
testing methodologies and conceptual issues. Environ. Health
Perspect 106, 413–418.
Blaise, C.R., 1993. In: Richardson, M.L. (Ed.), Ecotoxicology
Monitoring. VCH, Weinheim, pp. 83–108.
Blanchard, L.A., Hancu, D., Beckman, E.J., Brennecke, J.F., 1999.
Green processing using ionic liquids and CO 2. Nature 399, 28–29.
Blanchard, L.A., Brennecke, J.F., 2001. Recovery of organic
products from ionic liquids using supercritical carbon dioxide.
Ind. Eng. Chem. Res. 40, 287–292.
Boethling, R.S., 1994. Cationic Surfactants, Surfactant Science
Series, vol. 53. Marcel Dekker, New York, pp. 95–135.
Boethling, R.S., 1996. Designing safer chemicals. ACS Symp. Ser.
640, 156.
Chemnitius, J.-M., Sadowski, R., Winkel, H., Zech, R., 1999.
Organophosphate inhibition of human heart muscle
cholinesterase isoenzymes. Chem. Biol. Interact 120, 183–192.
Cheng, Y., Wright, S.H., Hooth, M.J., Sipes, I.G., 2009.
Characterization of the disposition and toxicokinetics of
N-butylpyridinium chloride in male F-344 rats and female
B6C3F1 mice and its transport by organic cation transporter 2.
Drug Metab. Dispos. 37, 909–916.
Cho, C.-W., Pham, T.P.T., Jeon, Y.-C., Vijayaraghavan, K., Choe, W.-S.,
Yun, Y.-S., 2007. Toxicity of imidazolium salt with anion bromide
to a phytoplankton Selenastrum capricornutum:effectofalkylchain
length. Chemosphere 69, 1003–1007.
Cho, C.-W., Pham, T.P.T., Jeon, Y.-C., Yun, Y.-S., 2008a. Influence of
anions on the toxic effects of ionic liquids to a phytoplankton
Selenastrum capricornutum. Green Chem. 10, 67–72.
Cho, C.-W., Jeon, Y.-C., Pham, T.P.T., Vijayaraghavan, K., Yun, Y.-S.,
2008b. The ecotoxicity of ionic liquids and traditional organic
solvents on microalga Selenastrum capricornutum. Ecotoxicol.
Environ. Saf. 71, 166–171.
Cho, C.-W., Pham, T.P.T., Jeon, Y.-C., Min, J., Jung, H.Y., Lee, D.S.,
Yun, Y.-S., 2008c. Microalgal photosynthetic activity
measurement system for rapid toxicity assessment.
Ecotoxicology 17, 455–463.
Cieniecka-Ros1onkiewicz, A., Pernak, J., Kubis-Feder, J., Ramani, A.,
Robertson, A.J., Seddon, K.R., 2005. Synthesis, anti-microbial
activities and anti-electrostatic properties of phosphoniumbased
ionic liquids. Green Chem. 7, 855–862.
Costello, D.M., Brown, L.M., Lamberti, G.A., 2009. Acute toxic effects
of ionic liquids on zebra mussel (Dreissena polymorpha) surviving
and feeding. Green Chem. 11, 548–553.
Couling, D.J., Bernot, R.J., Docherty, K.M., Dixon, J.K., Maginn, E.J.,
2006. Assessing the factors responsible for ionic liquid toxicity
to aquatic organisms via quantitative structure-property
relationship modeling. Green Chem. 8, 82–90.
Dipeolu, O., Green, E., Stephens, G., 2008. Effects of watermiscible
ionic liquids on cell growth and nitro reduction using
Clostridium sporogenes. Green Chem. 11, 397–401.
Docherty, K.M., Kulpa, C.F., 2005. Toxicity and antimicrobial
activity of imidazolium and pyridinium ionic liquids. Green
Chem. 7, 185–189.
Docherty, K.M., Dixon, J.K., Kulpa, C.F., 2007. Biodegradability of
imidazolium and pyridinium ionic liquids by an activated
sludge microbial community. Biodegradation 18, 481–493.
Earle, M.J., Seddon, K.R., 2000. Ionic liquids. Green solvents for the
future. Pure Appl. Chem. 72, 1391–1398.
Frade, R.F.M., Matias, A., Branco, L.C., Afonso, C.A.M., Duarte, C.
M.M., 2007. Effect of ionic liquids on human colon carcinoma
HT-29 and CaCo-2 cell lines. Green Chem. 9, 873–877.
Fuler, J., Carlin, R.T., Osteryoung, R.A., 1997. The room temperature
ionic liquids 1-ethyl-3-methylimidazolium tetrafluoroborate:

370
electrochemical couples and physical properties. J.
Electrochem. Soc. 144, 3881–3886.
Fulton, M.H., Key, P.B., 2001. Acetylcholinesterase inhibition in
estuarine fish and invertebrates as an indicator of
organophosphorus insecticide exposure and effects. Environ.
Toxicol. Chem. 20, 37–45.
Ganske, F., Bornscheuer, U.T., 2006. Growth of Escherichia coli,
Pichia pastoris and Bacillus cereus in the presence of the ionic
liquids [BMIM] [BF 4] and [BMIM] [PF 6] and organic solvents.
Biotechnol. Lett. 28, 465–469.
Garcia, M.T., Gathergood, N., Scammells, P.J., 2005. Biodegradable
ionic liquids. part II: effect of the anion and toxicology. Green
Chem. 7, 9–14.
García-Lorenzo, A., Tojo, E., Tojo, J., Teijeira, M., Rodríguez-
Berrocal, F.J., González, M.P., Martínez-Zorzano, V.S., 2008.
Cytotoxicity of selected imidazolium-derived ionic liquids in
the human Caco-2 cell line. Sub-structural toxicological
interpretation through a QSAR study. Green Chem. 10, 508–516.
Gathergood, N., Scammells, P.J., 2002. Design and preparation of
room-temperature ionic liquids containing biodegradable side
chains. Aust. J. Chem. 55, 557–560.
Gathergood, N., Garcia, M.T., Scammells, P.J., 2004. Biodegradable
ionic liquids: part I. Concept, preliminary targets and
evaluation. Green Chem. 6, 166–175.
Gathergood, N., Scammells, P.J., Garcia, M.T., 2006. Biodegradable
ionic liquids. Part III: the first readily biodegradable ionic
liquids. Green Chem. 8, 156–160.
Gordon, C.M., 2001. New developments in catalysis using ionic
liquids. Appl. Catal. A 222, 101–117.
Gorman-Lewis, D.J., Fein, J.B., 2004. Experimental study of the
adsorption of an ionic liquid onto bacterial and mineral
surfaces. Environ. Sci. Technol 38, 2491–2495.
Grabinska-Sota, E., Kalka, J., 2004. An assessment of the toxicity
of pyridinium chlorides and their biodegradation
intermediates. Environ. Int 28, 687–690.
Grabinska-Sota, E., Kalka, J., 2006. Toxicity of imidazolium chlorides
to aquatic organisms. Polish J. Environ. Stud. 15, 405–409.
Hagiwara, R., Ito, Y., 2000. Room temperature ionic liquids of
alkylimidazolium cations and fluoroanions. J. Fluorine Chem.
105, 221–227.
Harjani, J.R., Singer, R.D., Garcia, M.T., Scammells, P.J., 2008. The
design and synthesis of biodegradable pyridinium ionic
liquids. Green Chem. 10, 436–438.
Hassoun, E.A., Abraham, M., Kini, V., Al-Ghafri, M., Abushaban, A.,
2002. Cytotoxicity of the ionic liquid, 1- N-butyl-3methylimidazolium
chloride. Res. Commun, Pharmacol.
Toxicol 7, 23–31.
Howard, P.H., Boethling, R.S., Stiteler, W., Meylan, W., Beauman, J.,
1991. Development of a predictive model for biodegradability
based on BIODEG, the evaluated biodegradation data base. Sci.
Total Environ. 110, 635–641.
Huddleston, J.G., Visser, A.E., Reichert, W.M., Willauer, H.D.,
Broker, G.A., Rogers, R.D., 2001. Characterization and
comparison of hydrophilic and hydrophobic room
temperature ionic liquids incorporating the imidazolium
cation. Green Chem. 3, 156–164.
Itakura, T., Hirata, K., Aoki, M., Sasai, R., Yoshida, H., Itoh, H.,
2008. Decompostition and removal of ionic liquid in aqueous
solution by hydrothermal and photocatalytic treatment.
Environ. Chem. Lett. doi:10.1007/s10311-008-0177-7.
Jastorff, B., Störmann, R., Ranke, J., Mölter, K., Stock, F.,
Oberheitmann, B., Hoffmann, W., Hoffmann, J., Nüchter, M.,
Ondruschka, B., Filser, J., 2003. How hazardous are ionic
liquids? Structure-activity relationships and biological testing
as important elements for sustainability evaluation. Green
Chem. 5, 136–142.
Jastorff, B., Mölter, K., Behrend, P., Bottin-Weber, U., Filser, J.,
Heimers, A., Ondruschka, B., Ranke, J., Schaefer, M.,
water research 44 (2010) 352–372
Schröder, H., Stark, A., Stepnowski, P., Stock, F., Störmann, R.,
Stolte, S., Welz-Biermann, U., Ziegert, S., Thöming, J., 2005.
Progress in evaluation of risk potential of ionic liquids-basis
for an eco-design of sustainable products. Green Chem. 7,
362–372.
Kaiser, K.L.E., Palabrica, V.S., 1991. Photobacterium phosphoreum,
toxicity data index. Water Poll. Res. J. Can. 26, 361–431.
Kaiser, J.P., Feng, Y.C., Bollag, J.M., 1996. Microbial metabolism of
pyridine, quinoline, acridine, and their derivatives under
aerobic and anaerobic conditions. Microbiol. Rev. 60, 483–498.
Kazarian, S.G., Briscoe, B.J., Welton, T., 2000. Combining ionic
liquids and supercritical fluids: in situ ATR-IR study of CO 2
dissolved in two ionic liquids at high pressures. Chem.
Commun., 2047–2048.
Kelman, D., Kashman, Y., Rosenberg, E., Ilan, M., Ifrach, I.,
Loya, Y., 2001. Antimicrobial activity of the reef sponge
Amphimedon viridis from the Red Sea: evidence for selective
toxicity. Aquat. Microb. Ecol 24, 9–16.
Kulacki, K.J., Lamberti, G.A., 2008. Toxicity of imidazolium ionic
liquids to freshwater algae. Green Chem. 10, 104–110.
Kumar, R.A., Papaïconomou, N., Lee, J.-M., Salminen, J., Clark, D.S.,
Prausnitz, J.M., 2009. In vitro cytotoxicities of ionic liquids: effect
of cation rings, functional groups, and anions. Environ. Toxicol.
24, 388–395.
Kumar, S., Ruth, W., Sprenger, B., Kragl, U., 2006. On the
biodegradation of ionic liquid 1-butyl-3-methylimidazolium
tetrafluoroborate. Chim. Oggi 24, 24–26.
Lahr, J., 1997. Ecotoxicology of organisms adapted to life in
temporary freshwater ponds in arid and semi-arid regions.
Arch. Environ. Contam. Toxicol 32, 50–57.
Landry, T.D., Brooks, K., Poche, D., Woolhiser, M., 2005. Acute
toxicity profile of 1-butyl-3-methylimidazolium chloride. Bull.
Environ. Contam. Toxicol 74, 559–565.
Larson, J.H., Frost, P.C., Lamberti, G.A., 2008. Variable toxicity of
ionic liquid-forming chemicals to Lemna minor and the
influence of dissolved organic matter. Environ. Toxicol. Chem.
27, 676–681.
Lata1a, A., Stepnowski, P., N˛edzi, M., Mrozik, W., 2005. Marine
toxicity assessment of imidazolium ionic liquids: acute effects
on the Baltic algae Oocystis submarina and Cyclotella
meneghiniana. Aqua. Toxicol 73, 91–98.
Lee, S.-M., Chang, W.-J., Choi, A.-R., Koo, Y.-M., 2005. Influence of
ionic liquids on the growth of Escherichia coli. Korean. J. Chem.
Eng 22, 687–690.
Lewis, M.A., 1995. In: Rand, G.M. (Ed.), Fundamentals of
Aquatic Toxicology: Effects, Environment Fate, and Risk
Assessment, second ed. Taylor and Francis, Washington,
DC, USA, pp. 135–170.
Li, G., Shen, J., Zhu, Y., 1998. Study of pyridinium-type functional
polymers. II. Antibacterial activity of soluble pyridinium-type
polymers. J. Appl. Polym. Sci. 67, 1761–1768.
Li, X., Zhao, J., Li, Q., Wang, L., Tsang, S.C., 2007. Ultrasonic
chemical oxidative degradations of 1,3-dialkylimidazolium
ionic liquids and their mechanistic elucidations. Dalton Trans,
1875–1880.
Li, X.-Y., Zhou, J., Yu, M., Wang, J.-J., Pei, Y.C., 2009. Toxic effects of
1-methyl-3-octylimidazolium bromide on the early embryonic
development of the frog Rana nigromaculata. Ecotox. Environ.
Saf. 72, 552–556.
Luis, P., Ortiz, I., Alcado, R., Irabien, A., 2007. A novel group
contribution method in the development of a QSAR for
predicting the toxicity (Vibrio fischeri EC 50) of ionic liquids.
Ecotoxicol. Environ. Saf 67, 423–429.
Luo, Y.-R., Li, X.-Y., Chen, X.-X., Zhang, B.-J., Sun, Z.-J., Wang, J.-J.,
2008. The developmental toxicity of 1-butyl-3-octylimidazolium
bromide on Daphnia magna. Environ. Toxicol 23, 736–744.
Malich, G., Markovic, B., Winder, C., 1997. The sensitivity and
specificity of the MTS tetrazolium assay for detecting the in

vitro cytotoxicity of 20 chemicals using human cell lines.
Toxicology 124, 179–192.
Mann, R.M., Bidwell, J.R., 2000. Application of the FETAX protocol
to assess the development toxicity of nonylphenol ethoxylate
to Xenopus laevis and two Australian frogs. Aquat. Toxicol 51,
19–29.
Marsh, K.N., Boxall, J.A., Lichtenthaler, R., 2004. Room
temperature ionic liquids and their mixtures–a review. Fluid
Phase Equilibr 219, 93–98.
Massoulié, J., Pezzementi, L., Bon, S., Krejci, E., Vallette, F.-M.,
1993. Molecular and cellular biology of cholinesterases. Prog.
Neurobiol. 41, 31–91.
Matsumoto, M., Mochiduki, K., Fukunishi, K., Kondo, K., 2004a.
Extraction of organic acids using imidazolium-based ionic
liquids and their toxicity to Lactobacillus rhamnosus. Sep.Purif.
Technol. 40, 97–101.
Matsumoto, M., Mochiduki, K., Kondo, K., 2004b. Toxicity of ionic
liquids and organic solvents to lactic acid-producing bacteria.
J. Biosci. Bioeng 98, 344–347.
Matzke, M., Stolte, S., Thiele, K., Juffernholz, T., Arning, J., Ranke, J.,
Welz-Biermann, U., Jastorff, B., 2007. The influence of anion
species on the toxicity of 1-alkyl-3-methylimidazolium ionic
liquids observed in an (eco)toxicological test battery. Green
Chem. 9, 1198–1207.
Matzke, M., Stolte, S., Böschen, A., Filser, J., 2008. Mixture effects
and predictability of combination effects of imidazolium
based ionic liquids as well as imidazolium based ionic liquids
and cadmium on terrestrial plants (Triticum aestivum) and
limnic green algae (Scenedesmus vacuolatus). Green Chem. 10,
784–792.
Matzke, M., Stolte, S., Arning, J., Uebers, U., Filser, J., 2009a. Ionic
liquids in soils: effects of different anion species of
imidazolium based ionic liquids on wheat (Triticum aestivum)
as affected by different clay minerals and clay concentrations.
Ecotoxicology 18, 197–203.
Matzke, M., Thiele, K., Müller, A., Filser, J., 2009b. Sorption and
desorption of imidazolium based ILs in different soil types.
Chemosphere 74, 568–574.
McFarlane, J., Ridenour, W.B., Luo, H., Hunt, R.D., Depaoli, D.W.,
Ren, R.X., 2005. Room temperature ionic liquids for
separating organics from produced water. Sep. Sci. Technol
40, 1245–1265.
McGuinness, D.S., Cavell, K.J., 2000. Donor-functionalized
heterocyclic carbine complexes of palladium(II): efficient
catalysts for C-C coupling reactions. Organometallics 19,
741–748.
McQueen, D.J., Post, J.R., Mills, E.L., 1986. Trophic relationships in
freshwater pelagic ecosystems. Can. J. Fish. Aqua. Sci. 43,
1571–1581.
Morawski, A.W., Janus, M., Goc-Maciejewska, I., Syguda, A.,
Pernak, J., 2005. Decomposition of ionic liquids by
photocatalysis. Pol. J. Chem. 79, 1929–1935.
Myasoedov, B.F., Molochnikova, N.P., Shkinev, V.M., Spivakov, B.Y.,
1995. The Behavior of Actinides in Two-Phase Aqueous Systems
Based on Polyethylene Glycol. Plenum, New York, pp. 91.
Nockemann, P., Thijs, B., Driesen, K., Janssen, C.R., Hecke, K.V.,
Meervelt, L.V., Kossmann, S., Kirchner, B., Binnemans, K.,
2007. Choline saccharinate and choline acesulfamate: ionic
liquids with low toxicities. J. Phys. Chem. B 111, 5254–5263.
Olivier, H., 1999. Recent developments in the use of non-aqueous
ionic liquids for two-phase catalysis. J. Mol. Cat. A: Chem. 146,
285–289.
Osenberg, C.W., Mittelbach, G.G., 1989. Effects of body size on the
predator–prey interaction between pumpkinseed sunfish and
gastropods. Ecol. Monogr. 59, 405–432.
Pernak, J., Czepukowicz, A., 2001. New ionic liquids and their
antielectrostatic properties. Ind. Eng. Chem. Res. 40, 2379–
2383.
water research 44 (2010) 352–372 371
Pernak, J., Rogo _za, J., Mirska, I., 2001a. Synthesis and
antimicrobial activities of new pyridinium and
benzimidazolium chlorides. Eur. J. Med. Chem. 36, 313–320.
Pernak, J., Kalewska, J., Ksycińska, H., Cybulski, J., 2001b.
Synthesis and anti-microbial activities of some pyridinium
salts with alkoxymethyl hydrophobic group. Eur. J. Med.
Chem. 36, 899–907.
Pernak, J., Chwa1a, P., 2003. Synthesis and anti-microbial activities of
choline-like quaternary ammonium chlorides. Eur. J. Med. Chem.
38, 1035–1042.
Pernak, J., Sobaszkiewicz, K., Mirska, I., 2003. Anti-microbial
activities of ionic liquids. Green Chem. 5, 52–56.
Pernak, J., Goc, I., Mirska, I., 2004a. Anti-microbial activities of
protic ionic liquids with lactate anion. Green Chem. 6, 323–329.
Pernak, J., Sobaszkiewicz, K., Foksowicz-Flaczyk, J., 2004b. Ionic
liquids with symmetrical dialkoxymethyl-substituted
imidazolium cations. Chem. Eur. J 10, 3479–3485.
Pham, T.P.T., Cho, C.-W., Vijayaraghavan, K., Min, J., Yun, Y.-S.,
2008a. Effect of imidazolium-based ionic liquids on the
photosynthetic activity and growth rate of Selenastrum
capricornutum. Environ. Toxicol. Chem. 27, 1583–1589.
Pham, T.P.T., Cho, C.-W., Min, J., Yun, Y.-S., 2008b. Alkyl-chain
length effects of imidazolium and pyridinium ionic liquids on
photosynthetic response of Pseudokirchneriella subcapitata.
J. Biosci. Bioeng 105, 425–428.
Pham, T.P.T., Cho, C.-W., Jeon, C.-O., Chung, Y.-J., Lee, M.-W.,
Yun, Y.-S., 2009. Identification of metabolites involved in the
biodegradation of the ionic liquid 1-butyl-3-methylpyridinium
bromide by activated sludge microorganisms. Environ. Sci.
Technol 43, 516–521.
Peraccini, D., Chiappe, C., Intorre, L., Pretti, C., 2007. In: Letcher, T.
M. (Ed.), Thermodynamics, Solubility and Environmental
Issues. Elsevier, UK, pp. 259–278.
Pope, C., Karanth, S., Liu, J., 2005. Pharmacology and toxicology of
cholinesterase inhibitors: uses and misuses of a common
mechanism of action. Environ. Toxicol. Pharmacol 19, 433–446.
Pretti, C., Chiappe, C., Pieraccini, D., Gregori, M., Abramo, F.,
Monni, G., Intorre, L., 2006. Acute toxicity of ionic liquids to
the zebrafish (Danio rerio). Green Chem. 8, 238–240.
Pretti, C., Chiappe, C., Baldetti, I., Brunini, S., Monni, G., Intorre, L.,
2009. Acute toxicity of ionic liquids for three freshwater
organisms: Pseudokirchneriella subcapitata, Daphnia magna and
Danio rerio. Ecotoxicol. Environ. Saf. 72, 1170–1176.
Ranke, J., Mölter, K., Stock, F., Bottin-Weber, U., Poczobutt, J.,
Hoffmann, J., Ondruschka, B., Filser, J., Jastorff, B., 2004.
Biological effects of imidazolium ionic liquids with varying
chain length in acute Vibrio fischeri and WST-1 cell viability
assays. Ecotox. Environ. Saf 58, 396–404.
Ranke, J., Müller, A., Bottin-Weber, U., Stock, F., Stolte, S., Arning, J.,
Störmann, R., Jastorff, B., 2007a. Lipophilicity parameters for
ionic liquid cations and their correlation to in vitro cytotoxicity.
Ecotox. Environ. Saf 67, 430–438.
Ranke, J., Stolte, S., Störmann, R., Arning, J., Jastorff, B., 2007b.
Design of sustainable chemical products – the example of
ionic liquids. Chem. Rev. 107, 2183–2206.
Rogers, R.D., Bond, A.H., Bauer, C.B., Zhang, J., Rein, S.D., Chomko, R.R.,
Roden, D.M., 1995. Partitioning behavior of 99 Tc and, 129 I from
simulated Handford tank wastes using polyethylene-glycol based
aqueous biphasic systems. Solvent Extr. Ion Exch. 13, 689–713.
Romero, A., Santos, A., Tojo, J., Rodríguez, A., 2008. Toxicity and
biodegradability of imidazolium ionic liquids. J. Hazard. Mater
151, 268–273.
Rooney, D.W., Seddon, K.R., 2001. In: Wypych, G. (Ed.), Handbook
of Solvents. ChemTech Publishing, Toronto, pp. 1459–1484.
Salminen, J., Papaiconomou, N., Kumar, R.A., Lee, J.-M., Kerr, J.,
Newman, J., Prausnitz, J.M., 2007. Physicochemical properties
and toxicities of hydrophobic piperidinium and pyrrolidinium
ionic liquids. Fluid Phase Equilibr 261, 421–426.

372
Samorì, C., Pasteris, A., Galletti, P., Tagliavini, E., 2007. Acute
toxicity of oxygenated and nonoxygenated imidazoliumbased
ionic liquids to Daphnia magna and Vibrio fischeri.
Environ. Toxicol. Chem. 26, 2379–2382.
Schmid, A., Kollmer, A., Mathys, R.G., Withot, B., 1998. Development
toward large-scale bacterial bioprocesses in the presence of bulk
amounts of organic solvents. Extremophiles 2, 249–256.
Sheldon, R., 2001. Catalytic reactions in ionic liquids. Chem.
Commun, 2399–2407.
Sheldon, R.A., 2005. Green solvents for sustainable organic
synthesis: state of the art. Green Chem. 7, 267–278.
Siedlecka, E.M., Stepnowski, P., 2009. The effect of alkyl chain
length on the degradation of alkylimidazolium- and
pyridinium-type ionic liquids in a Fenton-like system.
Environ. Sci. Pollut. Res. 16, 453–458.
Siedlecka, E.M., Mrozik, W., Kaczyński, Z., Stepnowski, P.,
2008a. Degradation of 1-butyl-3-methylimidazolium chloride
ionic liquid in a Fenton-like system. J. Hazard. Mater 154,
893–900.
Siedlecka, E.M., Go1˛ebiowski, M., Kumirska, J., Stepnowski, P.,
2008b. Identification of 1-butyl-3-methylimidazolium chloride
degradation products formed in Fe(III)/H2O2 oxidation system.
Chem. Anal. (Warsaw) 53, 943–951.
Sipes, I.G., Knudsen, G.A., Kuester, R.K., 2008. The effects of dose
and route on the toxicokinetics and disposition of 1-butyl-
3-methylimidazolium chloride in male F-344 rats and female
B6C3F1 mice. Drug Metab. Dispos. 36, 284–293.
Sk1adanowski, A.C., Stepnowski, P., Kleszczyński, K.,
Dmochowska, B., 2005. AMP deaminase in vitro inhibition by
xenobiotics. A potential molecular method for risk assessment of
synthetic nitro- and polycyclic musks, imidazolium ionic liquids
and N-glucopyranosylammoniumsalts.Environ.Toxicol.Phar.
19, 291–296.
Stasiewicz, M., Mulkiewicz, E., Tomczak-Wandzel, R.,
Kurmirska, J., Siedlecka, E.M., Go1ebiowski, M., Gajdus, J.,
Czerwicka, M., Stepnowski, P., 2008. Assessing toxicity and
biodegradation of novel, environmentally benign ionic liquids
(1-alkoxymethyl-3-hydroxypyridinium chloride, saccharine
and acesulfamates) on cellular and molecular level. Ecotox.
Environ. Saf 71, 157–165.
Steinberg, S.M., Poziomek, E.J., Engelmann, W.H., Rogers, K.R., 1995.
A review of environmental applications of bioluminescence
measurements. Chemosphere 30, 2155–2197.
Stepnowski, P., Sk1adanowski, A.C., Ludwiczak, A., qaczyńska, E.,
2004. Evaluating the cytotoxicity of ionic liquids using human
cell line HeLa. Hum. Exp. Toxicol 23, 513–517.
Stepnowski, P., 2005. Preliminary assessment of the sorption of
some alkyl imidazolium cations as used in ionic liquids to
soils and sediments. Aust. J. Chem. 58, 170–173.
Stepnowski, P., Zaleska, A., 2005. Comparison of different
advanced oxidation processes for the degradation of room
temperatureionicliquids.J.Photoch.Photobio.A170,45–50.
Stepnowski, P., Mrozik, W., Nichthauser, J., 2007. Adsorption of
alkylimidazolium and alkylpyridinium ionic liquids onto
natural soils. Environ. Sci. Technol 41, 511–516.
Stock, F., Hoffmann, J., Ranke, J., Störmann, R., Ondruschka, B.,
Jastorff, B., 2004. Effects of ionic liquids on the
acetycholinesterase–a structure-activity relationship
consideration. Green Chem. 6, 286–290.
Stolte, S., Arning, J., Bottin-Weber, U.,Matzke,M.,Stock,F.,Thiele,K.,
Uerdingen, M., Welz-Biermann, U., Jastorff, B., Ranke, J., 2006.
Anion effects on the cytotoxicity of ionic liquids. Green Chem. 8,
621–629.
Stolte, S., Matzke, M., Arning, J., Böschen, A., Pitner, W.-R.,
Welz-Biermann, U., Jastorff, B., Ranke, J., 2007a. Effects of
water research 44 (2010) 352–372
different head groups and functionalized side chains on
the aquatic toxicity of ionic liquids. Green Chem. 9, 1170–1179.
Stolte, S., Arning, J., Bottin-Weber, U., Müller, A., Pitner, W.-R.,
Welz-Biermann, U., Jastorff, B., Ranke, J., 2007b. Effects of
different head groups and functionalized side chains on the
cytotoxicity of ionic liquids. Green Chem. 9, 760–767.
Stolte, S., Abdulkarim, S., Arning, J., Blomeyer-Nienstedt, A.-K.,
Bottin-Weber,U.,Matzke,M.,Ranke,J.,Jastorff,B.,Thöming, J.,
2008. Primary biodegradation of ionic liquid cations, identification
of degradation products of 1-methyl-3-octylimidazolium chloride
and electrochemical wastewater treatment of poorly
biodegradable compounds. Green Chem. 10, 214–224.
Studzińska, S., Buszewski, B., 2009. Study of toxicity of imidazolium
ionic liquids to watercress (Lepidium sativum L.). Anal. Bioanal.
Chem. 393, 983–990.
Swatloski, R.P., Holbrey, J.D., Rogers, R.D., 2003. Ionic liquids are
not always green: hydrolysis of 1-butyl-3-methylimidazolium
hexafluorophosphate. Green Chem. 5, 361–363.
Swatloski, R.P., Holbrey, J.D., Memon, S.B., Caldwell, G.A.,
Caldwell, K.A., Rogers, R.D., 2004. Using Caenorhabditis elegans
to probe toxicity of 1-alkyl-3-methylimidazolium chloride
based ionic liquids. Chem. Commun, 668–669.
Torrecilla, J.S., García, J., Rojo, E., Rodríguez, R., 2009. Estimation
of toxicity of ionic liquids in leukemia rat cell line and
acetylcholinesterase enzyme by principal component
analysis, neural networks and multiple linear regressions. J.
Hazard. Mater 164, 182–194.
U.S. Environmental Protection Agency 1998. Fate, transport, and
transformation test guidelines. OPPTS 835.3110. Readily
Biodegradability.
Vijayaraghavan, K., Pham, T.P.T., Cho, C.-W., Won, S.W., Choi, S.
B., Mao, J., Kim, S., Kim, Y.-R., Chung, B.W., Yun, Y.-S., 2009.
An assessment on the interaction of a hydrophilic ionic liquid
with different sorbents. Ind. Eng. Chem. Res. 48, 7283–7288.
Wang, L.-S., Wang, L., Wang, L., Wang, G., Li, Z.-H., Wang, J.-J.,
2009. Effect of 1-butyl-3-methylimidazolium tetrafluoroborate
on the wheat (Triticum aestivum L.) seedlings. Environ. Toxicol
24, 296–303.
Wang, X., Ohlin, C.A., Lu, Q., Fei, Z., Hu, J., Dyson, P.J., 2007.
Cytotoxicity of ionic liquids and precursor compounds
towards human cell line HeLa. Green Chem. 9, 1191–1197.
Wasserscheid, P., Keim, W., 2000. Ionic liquids–New ‘‘solutions’’ for
transition metal catalysis. Angew. Chem. Int. Ed. 39, 3772–3789.
Wells, A.S., Coombe, V.T., 2006. On the freshwater ecotoxicity and
biodegradation properties of some common ionic liquids. Org.
Pro. Res. Dev. 10, 794–798.
Welton, T., 1999. Room-temperature ionic liquids, solvents for
synthesis and catalysis. Chem. Rev. 99, 2071–2083.
Willauer, H.D., Huddleston, J.G., Griffin, S.T., Rogers, R.D., 1999.
Partitioning of aromatic molecules in aqueous biphasic
systems. Sep. Sci. Technol 34, 1069–1090.
Wong, D.S.H., Chen, J.P., Chang, J.M., Chou, C.H., 2002. Phase
equilibria of water and ionic liquids [emim][PF 6] and [bmim][PF 6].
Fluid Phase Equilibr 194–197, 1089–1095.
Yu, M., Li, S.-M., Li, X.-Y., Zhang, B.-J., Wang, J.-J., 2009a. Acute
effects of 1-octyl-3-methylimidazolium bromide ionic liquid
on the antioxidant enzyme system of mouse liver. Ecotox.
Environ. Saf. 71, 903–906.
Yu, M., Wang, S.-H., Luo, Y.-R., Han, Y.-W., Li, X.-Y., Zhang, B.-J.,
Wang, J.-J., 2009b. Effects of the 1-alkyl-3-methylimidazolium
bromide ionic liquids on the antioxidant defense system of
Daphnia magna. Ecotox. Environ. Saf. 72, 1798–1804.
Zhang, C., Malhotra, S.V., 2005. Increased paraoxon detection by
acetylcholinesterase inactivation with ionic liquid additives.
Tatanta 67, 560–563.

A QSAR model for predicting rejection of emerging
contaminants (pharmaceuticals, endocrine disruptors)
by nanofiltration membranes
Victor Yangali-Quintanilla a,b, *, Anwar Sadmani a , Megan McConville a,b ,
Maria Kennedy a , Gary Amy a,b
a
UNESCO-IHE, Institute for Water Education, Westvest 7, 2611AX Delft, The Netherlands
b
Delft University of Technology, Stevinweg 1, 2628CN Delft, The Netherlands
article info
Article history:
Received 25 February 2009
Received in revised form
4 June 2009
Accepted 26 June 2009
Available online 3 July 2009
Keywords:
Pharmaceuticals
Endocrine disruptors
Nanofiltration
Modelling
QSAR
water research 44 (2010) 373–384
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
abstract
A quantitative structure activity relationship (QSAR) model has been produced for predicting
rejection of emerging contaminants (pharmaceuticals, endocrine disruptors,
pesticides and other organic compounds) by polyamide nanofiltration (NF) membranes.
Principal component analysis, partial least square regression and multiple linear regressions
were used to find a general QSAR equation that combines interactions between
membrane characteristics, filtration operating conditions and compound properties for
predicting rejection. Membrane characteristics related to hydrophobicity (contact angle),
salt rejection, and surface charge (zeta potential); compound properties describing
hydrophobicity (log Kow, log D), polarity (dipole moment), and size (molar volume, molecular
length, molecular depth, equivalent width, molecular weight); and operating conditions
namely flux, pressure, cross flow velocity, back diffusion mass transfer coefficient,
hydrodynamic ratio (Jo/k), and recovery were identified as candidate variables for rejection
prediction. An experimental database produced by the authors that accounts for 106
rejection cases of emerging contaminants by NF membranes as result of eight experiments
with clean and fouled membranes (NF-90, NF-200) was used to produce the QSAR model.
Subsequently, using the QSAR model, rejection predictions were made for external
experimental databases. Actual rejections were compared against predicted rejections and
acceptable R 2 correlation coefficients were found (0.75 and 0.84) for the best models.
Additionally, leave-one-out cross-validation of the models achieved a Q 2 of 0.72 for internal
validation. In conclusion, a unified general QSAR equation was able to predict rejections of
emerging contaminants during nanofiltration; moreover the present approach is a basis to
continue investigation using multivariate analysis techniques for understanding
membrane rejection of organic compounds.
ª 2009 Elsevier Ltd. All rights reserved.
* Corresponding author: UNESCO-IHE, Institute for Water Education, Westvest 7, 2611AX Delft, The Netherlands. Tel.: þ31 15 215 1745;
fax: þ31 15 215 2921.
E-mail addresses: v.a.yangaliquintanilla@tudelft.nl, v.yangaliquintanilla@unesco-ihe.org (V. Yangali-Quintanilla).
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.watres.2009.06.054

374
1. Introduction
Nanofiltration (NF) and reverse osmosis (RO) are technologies
that provide medium to high rejections of organic
compounds present as emerging contaminants (micropollutants)
in water, namely endocrine disrupting
compounds (EDCs), pharmaceutically active compounds
(PhACs), and pesticides (Kiso et al., 2001; Schäfer et al., 2003;
Kimura et al., 2003, 2004; Nghiem et al., 2004). The presence
of micropollutants has been identified in surface water
bodies, sewage treatment plant effluents, and stages of
drinking water treatment plants and even at trace-levels in
finished drinking water (Kolpin et al., 2002; Heberer, 2002;
Castiglioni et al., 2006). The possible effects on aquatic
organisms and human health, associated with the
consumption of water containing low concentrations of
single compounds, have been presented in toxicology studies
(Pomati et al., 2006; Escher et al., 2005; Vosges et al., 2008).
The studies demonstrate that researchers do not yet
understand the exact risks from decades of persistent
exposure to random combinations of low levels of pharmaceuticals,
EDCs, and other organic compounds; hence, the
long-term effects of consumption of water containing low
concentrations of micropollutants will remain as an unanswered
question for the foreseeable future. Meanwhile,
water treatment facilities are implementing monitoring
programs, research organizations dealing with water reuse
have published reports, and studies have addressed the topic
(Drewes et al., 2006; Verliefde et al., 2007). An important
aspect to deal with the problem has been the identification
of compound physicochemical properties and membrane
characteristics to explain transport, adsorption and removal
of micropollutants by different mechanisms, explicitly size/
steric exclusion, hydrophobic adsorption and partitioning,
and electrostatic repulsion (Kiso et al., 2001; Schäfer et al.,
2003; Kimura et al., 2003; Nghiem et al., 2004, Ozaki and Li,
2002; Van der Bruggen and Vandecasteele, 2002; Bellona and
Drewes, 2005; Xu et al., 2005). A number of articles have
proposed a mechanistic understanding of the interaction
between membranes and organic compounds; others have
tried to apply fitting parameter models to model rejection
(Cornelissen et al., 2005; Kim et al., 2007; Verliefde et al.,
2008). However, there have been few models to ‘‘predict’’ the
rejection of compounds. To overcome that status, our
objective was to create a general QSAR model to predict
rejections based on an integral approach that considers
membrane characteristics, filtration operating conditions
and physicochemical compound properties. A quantitative
structure activity relationship (QSAR) is a method that
relates an activity of a set of compounds quantitatively to
chemicals descriptors (structure or property) of those
compounds (Sawyer et al., 2003). QSAR has the objective of
prediction but maintaining a relationship to mechanistic
interpretation. Applications of QSAR for the development of
models to find relationships between membranes and
organic compounds have been presented in journals related
to drug discovery and medicinal chemistry for analysis of
permeability of membranes to organic compounds (Ren
et al., 1996; Fujikawa et al., 2007). The study of reverse
water research 44 (2010) 373–384
osmosis membranes has also experienced the application of
QSAR principles. Campbell et al. (1999) performed a QSAR
analysis of surfactants influencing attachment of a mycobacterium
to cellulose and aromatic polyamide reverse
osmosis membranes; their objective was to understand the
relationship between surfactant molecular properties and
activity on the membrane surface to inhibit bacterial
attachment to the membrane in order to reduce biofilm
formation and to increase permeate production. More
recently, Libotean et al. (2008) developed an artificial neural
network model based on quantitative structure-property
relations, the model claim to predict organic solute passage
through reverse osmosis membranes considering simultaneous
correlation of organic solutes (molecular descriptors)
and membrane properties. Alike, our study uses the concept
of QSAR analysis to quantify an activity, compound rejection
by a membrane, in terms of organic compound physicochemical
properties, membrane characteristics (salt rejection,
pure water permeability, molecular weight cut-off,
charge, hydrophobicity) and operating conditions (pressure,
flux, cross flow velocity, back diffusion mass transfer coefficient,
recovery). In this work a QSAR model was constructed
with internal experimental data used for training.
The model was internally validated using measures of
goodness of fit and prediction. Subsequently, after identification
of a relationship in form of an equation, estimation of
rejections for an external dataset for different compounds
and membranes were used to externally validate the model.
Similarly, rejections of more emerging organic contaminants
can be predicted in advance, before nanofiltration or reverse
osmosis applications. Nevertheless, the QSAR model is
applicable in the range of boundary experimental conditions
that will be defined in the experimental section of this
publication (Section 3).
2. Theory
2.1. Principal component analysis
Principal component analysis (PCA) is a method that allows
simplification of many variables into a group of a few variables
that might be measuring the same principles of a system. It
may occur that a system considers an abundance of variables
to explain a process; in this case principal component analysis
reduces the redundancy of information. The general objectives
of PCA are data reduction and interpretation. Although p
variables (components) are required to reproduce the total
system variability, often much of this variability can be
accounted for by a small number k of principal components.
The k is the number of components (reduced) that represent
the initial p variables. In general, PCA is concerned with
whether the covariances or correlations between a set of
observed variables x 1, x 2, ., x p can be explained in terms of
a smaller number of unobservable components, c1, c2, ., ck,
where k < p. Thus, there is as much information in the k
components as there is in the original p variables. Comprehensive
details about the theory of PCA can be found elsewhere
(Jolliffe, 2002; Johnson and Wichern, 2007; Everitt and
Dunn, 2001).

2.2. Multiple linear regression
Multiple linear regression is a method of analysis for assessing
the strength of the relationship between a set of explanatory
variables known as independent variables, and a single
response or dependent variable. Applying multiple regression
analysis to a set of data results in what are known as regression
coefficients, one for each explanatory variable. The
multiple regression model for a response variable, y, with
observed values, y1, y2, ., yn (where n is the sample size) and q
explanatory variables, x1, x2, ., xq with observed values, x1i,
x2i, ., xqi for i ¼ 1, ., n, is
yi ¼ b 0 þ b 1x1i þ b 2x2i þ / þ b qxqi þ 3i
The regression coefficients, b0, b1, ., bq, are generally estimated
by least squares. The term 3i is the residual or error for
individual i and represents the deviation of the observed value
of the response for this individual from that expected by the
model. These error terms are assumed to have a normal
distribution with variance s 2 . The fit of a multiple regression
model can be judged with calculation of the multiple correlation
coefficient, R, defined as the correlation between the
observed values of the response variable and the values predicted
by the model. The squared value of R (R 2 ) gives the
proportion of the variability of the response variable accounted
for by the explanatory variables. Analysis of variance
(ANOVA) will provide an F-test of the null hypothesis that
each of b0, b1, ., bq, is equal to zero, or in other words that R 2 is
zero (Landau and Everitt, 2004).
2.3. Principal component regression and partial least
squares regression
Principal component regression (PCR) is a method in which
the components from the principal component method are
used for regression. Hence, the principal components of
the matrix X are used as regressors of a dependent Y. The
orthogonality of the principal components eliminates the
multicollinearity problem. But, the problem of choosing an
optimum subset of predictors remains. A possible strategy is
to keep only a few of the first components. But they are chosen
to explain X rather than Y, and therefore, nothing guarantees
that the principal components, which ‘‘explain’’ X, are relevant
for Y. Problems may arise, however, if there is a lot of
variation in X. PCR finds, somewhat uncritically, those latent
variables that describe as much as possible of the variation in
X. But sometimes the variable itself gives rise to only small
variations in X, and if the interferences vary a lot, then the
latent variables found by PCR may not be particularly good at
describing Y. In the worst case important information may be
hidden in directions in the X-space that PCR interprets as
disturbance, and therefore leaves out. Partial least squares
regression (PLS) is able to cope better with this problem, by
forming variables that are relevant for describing Y. By
contrast, PLS regression finds components from X that are
also relevant for Y. Specifically, PLS regression searches for
a set of components (called latent vectors) that performs
a simultaneous decomposition of X and Y with the constraint
that these components explain as much as possible of the
water research 44 (2010) 373–384 375
(1)
covariance between X and Y. This step generalizes PCA. The
goal of PLS regression is to predict Y from X and to describe
their common structure (Abdi, 2003; Jørgensen and Goegebeur,
2006).
3. Experimental
3.1. Chemicals, membranes, materials and
experimental conditions
The list of organic compounds representing emerging
contaminants is presented in Table 1. The compounds were
selected considering: their occurrence in surface water and
drinking water, their identification as priority emerging
contaminants, the availability of analytical methods, their
quantitative and qualitative representation of physicochemical
properties. The pharmaceutical compounds (caffeine,
sulfamethoxazole, acetaminophen, phenacetin, phenazone,
carbamazepine, naproxen, ibuprofen, metronidazole), endocrine
disrupting compounds (17b-estradiol, estrone, bisphenol
A, nonylphenol, atrazine) and sodium alginate were
purchased from Sigma–Aldrich (Sigma–Aldrich, Schnelldorf,
Germany). Potassium chloride, sodium hydroxide, hydrochloric
acid and magnesium sulphate anhydrous were
purchased from J.T.Baker (J.T.Baker, Deventer, Netherlands).
Two thin film composite NF membranes were selected for
this study (NF-200 and NF-90, Dow-Filmtec, Dow Chemical
Co., Midland, MI). The experimental setup consisted of two
filtration SEPA CF II (GE Osmonics, Minnetonka, MN) cells and
cell holders in parallel, in order to increase permeate
production and achieve hydrodynamic conditions, two
hydraulic pumps (Power Team, Bega Int. BV, Netherlands),
a 60 litres stainless steel tank (Tummers, Netherlands),
a positive displacement pump (Hydra-Cell pump, Wanner
Eng. Inc., Minneapolis, MN), a frequency converter (VLT
microdrive, Danfoss, SA), a chiller/heater (Julabo, Germany),
control needle valves, pressure gauges, flow meters,
a proportional pressure relief valve and stainless steel
tubings (Swagelok BV, Netherlands), a digital balance
(Sartorius, Germany) and, a computer for flow rate data
acquisition. A piece of membrane was compacted with
deionised water for 6 h at a pressure of 276 kPa before performing
an experiment with the membrane. The experiments
were conducted in a recycle mode in which permeate
and concentrate were recirculated into the feed tank for the
first 72 h (a pre-equilibration period); then, permeate was
collected within the next 24 h. The feed solution of all the
experiments contained a cocktail of 14 compounds (concentration
ranging from 6.5 to 65 mg/L). The main reason for
conducting experiments at concentrations of mg/L was to
accelerate steady state (after membrane adsorption) conditions
in a limited time (3 days). At very low concentrations
more time would be needed to achieve steady state conditions
and at short time tests low concentrations may lead to
over-estimation of rejections. A specific flux decline of 15% of
initial flux was targeted to foul the membranes using a feed
solution containing w10 mg/L DOC of sodium alginate. The
study of Lee et al. (2004) concluded that polysaccharides were
important membrane foulants. Sodium alginate was used as

376
Table 1 – List of compounds and physicochemical properties.
Name Molecular Acid
weight
(g/mol)
pKa
20 C a
b
Log Kow a
Log D Dipole
(pH 7) moment
(debye) c
surrogate of polysaccharides. Alginate is frequently used as
a model for organic matter of algae origin (Henderson et al.,
2008). Experiments were carried out for both clean and fouled
NF-90 and NF-200 membranes. The selection of membranes
was based on a qualitative rejection assessment of emerging
contaminants with molecular weight of more than 150 Da by
membranes with a MWCO between 200 and 300 Da. A total of
eight experiments were performed; at hydrodynamic ratio of
pure water permeation flux to back diffusion mass transfer
coefficient (J0/k) ofw1 and recovery 3% (NF-90 clean, NF-200
clean), at J0/k of w1 and recovery 8% (NF-200 clean), at J0/k of
w2 and recovery 3 (NF-90 clean), and at J 0/k of w2 and
recovery 8 (NF-90 clean and fouled, NF-200 clean and fouled).
The calculation method of k (back diffusion mass transfer
coefficient) was presented in a previous publication (Yangali-
Quintanilla et al., 2009). All the experiments were carried out
at a controlled temperature of 20 C, an ionic strength of
10 mM as KCl and a pH of 7. The transmembrane pressures
were in the range of 276–483 kPa. The fluxes were between
4.3 and 30.2 L/m 2 per day; cross flow velocities were between
0.5 and 4.5 cm/s. The experiments produced a total internal
dataset of 106 rejection cases; the dataset can be accessed as
supplementary data. The boundary experimental conditions
of the internal dataset are presented in Table 2. The internal
dataset was used to develop the model. An external dataset
that gathered three different datasets was used for validation
of the model. The external dataset is presented as supplementary
data. Experimental conditions for the first part of
the external dataset can be obtained from Kim et al. (2007)
and Yangali-Quintanilla et al. (2008). Experimental conditions
for the second and third parts can be found in Verliefde et al.
(2008); the data correspond to filtration experiments using
synthetic water solutions.
Molar
volume d
(cm 3 /
mol)
Molec.
length
(nm) e
3.2. Analytical equipment, analyses of compounds
and membranes
The pharmaceuticals and endocrine disruptors (with the
exception of atrazine) were analyzed by Technologiezentrum
Wasser (TZW, Karlsruhe, Germany). The detection limit was
10 ng/L per compound. The uncertainty of estimates was of
15% according to a validation method of the analysis
protocol and due to high concentrations of the samples (mg/L).
The analyses of pharmaceuticals were performed according to
Table 2 – Data range of membrane characteristics,
operating conditions and rejections.
Variable Unit Min. value Max. value
Molecular weight cut-off
(MWCO)
Pure water permeability
(PWP)
Molec.
width
(nm) e
Da 200 300
L/m 2 per
day/kPa
Molec.
depth
(nm) e
Equiv.
width
(nm) e
Classification f
Acetaminophen 151 10.2 0.46 0.23 4.55 120.90 1.14 0.68 0.42 0.53 HL-neutral
Phenacetine 179 N/A 1.58 1.68 4.05 163.00 1.35 0.69 0.42 0.54 HL-neutral
Caffeine 194 N/A 0.07 0.45 3.71 133.30 0.98 0.87 0.56 0.70 HL-neutral
Metronidazole 171 N/A 0.02 0.27 6.30 117.80 0.93 0.90 0.48 0.66 HL-neutral
Phenazone 188 N/A 0.38 0.54 4.44 162.70 1.17 0.78 0.56 0.66 HL-neutral
Sulfamethoxazole 253 5.7 0.89 0.45 7.34 173.10 1.33 0.71 0.58 0.64 HL-ionic
Naproxen 230 4.3 3.18 0.34 2.55 192.20 1.37 0.78 0.75 0.76 HB-ionic
Ibuprofen 206 4.3 3.97 0.77 4.95 200.30 1.39 0.73 0.55 0.64 HB-ionic
Carbamazepine 236 N/A 2.45 2.58 3.66 186.50 1.20 0.92 0.58 0.73 HB-neutral
Atrazine 216 N/A 2.61 2.52 3.43 169.80 1.26 1.00 0.55 0.74 HB-neutral
17b-estradiol 272 10.3 4.01 3.94 1.56 232.60 1.39 0.85 0.65 0.74 HB-neutral
Estrone 270 10.3 3.13 3.46 3.45 232.10 1.39 0.85 0.67 0.76 HB-neutral
Nonylphenol 220 10.3 5.71 5.88 1.02 236.20 1.79 0.75 0.59 0.66 HB-neutral
Bisphenol A 228 10.3 3.32 3.86 2.13 199.50 1.25 0.83 0.75 0.79 HB-neutral
a ADME/Tox Web Software.
b Experimental database: SRC PhysProp Database.
c Chem3D Ultra 7.0.
d ACD/ChemSketch Properties Batch.
e Molecular Modeling Pro.
f HL ¼ hydrophilic, HB ¼ hydrophobic.
water research 44 (2010) 373–384
0.86 2.23
Salt rejection (SR) a
– 0.96 0.98
Zeta potential (ZP) MV 48.04 10.78
Contact angle (CA) 39.3 58.0
Pressure (P) kPa 276 483
Cross flow velocity (v) cm/s 0.5 4.5
Back diffusion mass
transfer coefficient (k)
cm/s 2.70E-04 5.99E-04
Flux (J ) L/m 2 per day 4.3 30.2
Hydrodynamic ratio
(J0/k)
– 1 2
Recovery (recov) % 3 8
Rejection (rejection) % 17.7 99.0
a 2000 mg/L MgSO4, 25 C, recovery 15%, pressure 1034 kPa, pH 8.

the protocols described by Sacher et al. (2001, 2008). Analyses
of 17b-estradiol, estrone, nonylphenol and bisphenol A were
done by gas chromatography/mass spectrometry (GC/MS)
after automated solid-phase extraction onto a polymeric
material and subsequent silylation of the analytes. First, 10 ml
of a 50 ng/ml solution of 4-n-nonylphenol in acetone which
was used as internal standard for the overall procedure were
added to an aliquot of the water sample (1000 ml). Automated
solid-phase extraction (Tekmar AutoTrace, Germany) was
done on plastic cartridges filled with 200 mg of bondelut
material (Fa. Varian, Darmstadt, Germany). After the enrichment
step the solid-phase material was dried in a gentle
stream of nitrogen. Elution was done with 4 ml of acetone. The
acetone was evaporated to 100 ml in a stream of nitrogen and
to dryness in a drying oven at 80 C. The dry residue was
reconstituted with 100 ml of a silylation reagent mixture
(N-methyl-N-trimethylsilyltrifluoro acetamide (MSTFA)/2%
trimethyliodo silane). After a reaction time of 20 min at 80 C
(drying oven), determination of the derivatives was done by
GC/MS using a 6890 GC/MS system from Agilent Technologies
(Waldbronn, Germany).
Concentrations of atrazine were determined using microplate
enzyme-linked immunoabsorbent assay (ELISA) kits
(Abraxis LLC, Warminster, PA). Atrazine was determined with
a detection limit of 0.04 mg/L, and uncertainty of 15%. To
determine the hydrophobicity of membranes, contact angles
of clean and fouled membrane surfaces were measured with
CAM200 optical contact angle and surface tension meter (KSV
Instruments, Finland) at Delft University of Technology; to
measure contact angle, the sessile drop method was used.
Surface charge, in terms of zeta potential, of clean and fouled
membranes was quantified using ELS-8000 zeta potential
analyzer (Otsuka Electronics, Japan). The zeta potential analyses
were determined using a Milli-Q water solution at pH 7
and ionic strength of 10 mM KCl. The zeta potential was
determined using the electrophoresis method using a cell
consisting of membrane and quartz cells. The zeta potential
was calculated from the electrophoretic mobility using the
Smoluchowski formula, a detailed explanation of calculation
was provided in a previous publication (Shim et al., 2002). The
pH of the solutions was measured using a calibrated Metrohm
691 pH-meter (Metrohm AG, Herisau, Switzerland); the electrical
conductivity and temperature were measured with
a WTW Cond 330i (WTW GmbH, Weilheim, Germany) portable
conductivity meter. Clean and fouled membranes were characterized
to determine magnesium sulphate salt rejection at
standard conditions specified by manufacturers, a pure water
solution containing 2000 mg/L of magnesium sulphate at 25 C
and pH 8 was filtrated at pressure of 1034 kPa and recovery of
15%.
3.3. Characterization and classification of compounds
The acid dissociation constant as log K a (pK a) was used to
determine the speciation of the organic compound in ionic
species at pH 7. For hydrophobicity determination, log Kow and
log D were used; log Kow is the octanol–water partition coefficient
and log D is the ratio of the equilibrium concentrations
of all species (unionized and ionized) of a molecule in octanol
to the same species in the water phase. Values of pKa and log D
water research 44 (2010) 373–384 377
were calculated by ADME/Tox web software. Solubility and
log Kow values were obtained from SRC Physprop experimental
database. The value of the molecular dipole moment
was equal to the vector sum of the individual bond dipole
moments. Dipole moment was calculated by Chem3D Ultra
7.0, Cambridgesoft. Size descriptors included molar volume
(MV), molecular length, molecular width, molecular depth and
equivalent molecular width. The molecular length is defined
as the distance between the two most distant atoms. The
molecular width and molecular depth (width > depth) are
measured by projecting the molecule on the plane perpendicular
to the length axis and the equivalent molecular width
is defined as the geometric mean of width and depth (Santos
et al., 2006). Molar volumes were calculated using the program
ACD/ChemSketch Properties Batch, ACD/Labs; and Molecular
Modeling Pro, ChemSW, was used to compute size descriptors
after optimization geometry of a molecule from the interaction
of conformational analysis and energy minimization with
a semi-empiric method MOPAC-PM3. Based on pKa and log -
Kow values, the compounds were classified as hydrophilic
neutral, hydrophilic ionic, hydrophobic ionic and hydrophobic
neutral (see Table 1). Compounds with log Kow 2 were
referred to as hydrophobic; therefore those with log Kow < 2
were hydrophilic. The classification was based on an early
reference (Connell, 1990). Although the value of 2 may seem
low to consider hydrophobicity, the classification was not
used in constructing the models and therefore the magnitude
of log Kow or log D became more important. Table 1 shows the
calculated values of molecular weight, pKa, log Kow, log D,
dipole moment, molar volume, molecular length, molecular
width, molecular depth and equivalent width.
4. Results and discussion
4.1. QSAR methodology
The procedure to find a general QSAR equation to describe
rejection was performed in four phases. The first phase was
the organization of data from the experimental part. The data
comprised of 106 rejection cases. The database showing the
rejection cases is presented as supplementary data. A total of
21 initial variables were used. The variables considered as
compound descriptors were molecular weight (MW), solubility,
log Kow, log D, dipole moment, molar volume, molecular
length, molecular width, molecular depth and equivalent
width; variables describing membrane characteristics were
molecular weight cut-off (MWCO), pure water permeability
(PWP), magnesium sulphate salt rejection (SR), charge of the
membrane as zeta potential (ZP), and hydrophobicity as
contact angle (CA); variables describing operating conditions
were operating pressure (P), cross flow velocity (v), back
diffusion mass transfer coefficient (k), flux (J ), ratio of pure
water permeation flux J 0 and back diffusion mass transfer
coefficient (J0/k) and recovery. The range of values for
membrane characteristics, operating conditions and rejections
is presented in Table 2. The second phase was dedicated
to the process of variables reduction using a correlation
matrix and factor analysis with principal component analysis.
The third phase corresponded to the regression analysis. In

378
the third phase three methodologies were implemented; the
first was principal component analysis (PCA), with sequential
application of multiple linear regressions (MLR). The second
method was the use of partial least squares (PLS) regression
and MLR; and the third method was the use of MLR only. The
last phase was the validation process. The model was internally
validated using measures of goodness of fit (regression
coefficients) and prediction (leave-one-out cross-validation);
Section 4.5 gives details about the validation process. External
validation of the general QSAR model was implemented by
predicting rejections for an external dataset of experiments
performed with different compounds and membranes, and
with comparable operating conditions. PCA and PLS were
performed using the research and statistical package SPSS
Statistics 16.0. Leave-one-out cross-validations of the models
were performed with MobyDigs (Talete, Milano, Italy).
4.2. Variables reduction with principal component
analysis and QSAR model
The correlation matrix of the initial 21 variables was scrutinized
in order to obtain a not positive definite matrix as
a requisite of PCA; the matrix is accompanied as supplementary
data. A matrix is called not positive definite when there
are both positive and negative eigenvalues. In the case of
symmetric matrices, such as a correlation matrix, positive
definiteness will only hold if the matrix and every principal
submatrix have a positive determinant. A non-positive definite
input matrix may signal a perfect linear dependency of
one variable on another, known as collinearity. This was the
case for MWCO and salt rejection (SR) that were perfectly
linearly correlated. Therefore application of PCA considering
independently one variable or the other will give the same
results of variables reduction and number of components. In
other words, MWCO will not be excluded with PCA, the
variable will be separated in advance and the results obtained
for SR may be replaced by the variable MWCO, or vice versa.
Once an appropriate matrix was defined, the variables were
analyzed in terms of how significant their correlations with
rejection were; those correlations are also shown as an additional
row and column of the 21 21 variables matrix. Rejection
is only a reference variable to evaluate correlation with
the rest of variables. After a sequential implementation of
PCA, three components were extracted; they defined the
initial database of 21 variables with 11 variables describing
three relations namely membrane/operating-conditions
(comp. 1: flux, pure water permeability, salt rejection, zeta
potential, mass transfer coefficient, cross flow velocity),
hydrophobicity/size (comp. 2: length, log Kow, log D) and size
(comp. 3: equivalent width, depth). The final three components
accounted for 89.3% explanation of total variance. It is
important to mention that these results were produced for the
experimental dataset.
The next step was the implementation of multiple linear
regression (MLR) using the new set of variables. The use of
MLR after PCA presents the advantage of a more simplified
modelling approach. Moreover, the analysis of data before
MLR may help to identify variables that are similar in
response, which was the case of SR and MWCO. The dependent
variable for all regression analyses was rejection. Two
water research 44 (2010) 373–384
methods of linear regression were used, the first method is
called enter (forced) method; which performs a regression
with the contribution of all variables entered to model the
dependent variable. The second method is stepwise regression;
which is a more sophisticated method. Each variable is
entered in sequence and its contribution is assessed according
to an F-test. In the present study an F-test with a statistical
significance >0.10 implied removal of the variable, and F-test
with a significance

Elimelech, 2003). Thus, SR is ultimately serving as a comparison
parameter between membranes of the same type
(aromatic polyamide) but with narrow differences in pore size,
and possibly with differences in charge. The QSAR equation
merges information about interaction of membrane characteristics,
filtration operating conditions and organic
compounds properties to predict rejections during nanofiltration.
According to Eq. (2), contact angle and zeta potential
as measurements of hydrophilicity and membrane surface
charge, respectively, were not part of the equation and
therefore did not contribute quantitatively to model rejection.
However, size/steric hindrance effects related to salt rejection,
and hydrophobicity of the solutes were part of the model
equation. In conclusion, rejection increased by size/steric
hindrance effects; but hydrophobicity decrease rejection due
to adsorption and partitioning mechanisms.
The results of PCA and the model cannot be generalized to
experimental conditions outside of the boundary experimental
conditions defined in Section 3. For instance, excessive
changes of pH affect the ionic speciation of charged
compounds, obviously pK a values of solutes and pH of feed
waters will determine boundary conditions for applicability of
the model. Changes of membrane properties such as charge
and pore size due to swelling will also influence the model.
Other considerations for application of the model are the type
of membrane used (aromatic polyamide), fluxes, pressures
and cross flow velocities. Cellulose acetate or even different
membrane chemistry than NF-90 or NF-200 will influence the
PCA and the model. Nonetheless, the approach is valid and
can be generalized under certain conditions in upscale NF
applications.
4.3. QSAR model after partial least squares
regression and MLR
The following variables: MW, solubility, MV, MWCO, ZP, CA, P,
v, J, Jo/k, recovery and width, were removed after partial least
squares (PLS) regression. Therefore, the final PLS model is
defined by variables named log Kow, log D, dipole, length,
depth, eqwidth, PWP, SR and k. The main advantage of PLS is
the ability to handle collinearity among the independent
variables. In contrast, principal component analysis can not
control collinearity. Another advantage of PLS was that the
calculation process was simpler. However, PLS is used more as
a predictive technique and not as an interpretive technique
such as MLR. Therefore, in this exploratory analysis, PLS
regression serves as a variable selection process and as
prelude for implementation of MLR. Once again as occurred
with PCA, a reduced number of variables simplified the
implementation of MLR. After applying stepwise regression to
the reduced number of variables, the model result obtained
was also Eq. (2).
4.4. QSAR model after multiple linear regression
To finalize the model development under various statistical
application scenarios, the implementation of direct multiple
linear regression was performed. The R 2 is calculated for all
possible subset models. Using this technique, the model with
the largest R 2 is declared the best linear model. However, this
water research 44 (2010) 373–384 379
technique has some disadvantages. First, the R 2 increases
with each variable included in the model. Therefore, this
approach encourages including all variables in the best model
although some variables may not significantly contribute to
the model. This approach also contradicts the principal of
parsimony that encourages as few parameters in a model as
possible. Thus, the application of MLR without previous data
analysis is a possibility when the number of variables is
limited. Another disadvantage during the present study was
that MLR was not able to distinguish collinearity between
variables. This was the case for variables MWCO and SR;
MWCO can replace the role of SR in Eq. (2). The new equation
presented an R 2 of 0.75; the F-test was 52.5 with a significance
of w0%. The equation was the following
265:150 eqwidth 117:356 depth þ 81:662 length
rejection ¼
5:229 log D 0:272 MWCO 62:565
(3)
It is important to mention that Eq. (3) may have been equally
defined in Section 4.2 if the variable selected before implementing
PCA was MWCO and not SR as explained before;
therefore it is not surprising that Eqs. (2) and (3) only show
differences in two coefficients. However, considering practical
or operational facts, it may be more difficult to determine
changes of MWCO when fouling occurs; on the other hand,
salt rejections tests are part of monitoring practices. In addition,
the variable of MWCO may be difficult to define when
a range of MWCO exists for a membrane rather than an
approximate single MWCO value if compared to salt rejection,
although that fact is not desirable for a membrane because
will greatly influence rejection of solutes with sizes close to or
in the range of MWCO.
It may be argued that a simple stepwise MLR will produce
the same results of a PCA or PLS followed by MLR, however,
the application of MLR without PCA or PLS has the disadvantage
of removing or adding appropriate variables during the
iterative process of stepwise MLR. The stepwise MLR is only
dependent on the fulfillment of a statistical condition; it even
may happen that different combination of variables defines
a good equation. However, the advantage of PCA or PLS is that
only important variables are considered, and only those will
be part of the final MLR implementation.
4.5. Internal and external validation of the QSAR model
Actual (measured) rejection values (106 rejection cases) versus
modelled (fitted) rejections of the data used to generate the
model are shown in Fig. 1, the dataset is provided as supplementary
data; a 95% confidence interval shows that very few
modelled rejections were out of that interval. Besides of a good
fit of a model, it is necessary an assessment of the predictive
power of the model, i.e. appropriate robustness (Eriksson
et al., 2003). The R 2 is the most widely used measure of the
ability of a QSAR model to reproduce the internal data in the
training (goodness of fit), but does not explain its robustness
and prediction power. One technique to evaluate prediction is
the leave-one-out cross-validation technique, in which one
case at a time is iteratively held-out from the training set and
the rest is used for model development and the excluded case
is predicted by the developed model (Gramatica, 2007).

380
According to Gramatica, the predictive power of a model may
be estimated by the goodness of prediction parameter Q 2
leave-one-out (1-PRESS/TSS, where PRESS is the predictive
error sum of squares and TSS is the total sum of squares). In
general, a Q 2 > 0.5 is regarded as good and Q 2 > 0.9 as excellent
(Eriksson et al., 2003). For the developed QSAR models, the
model with SR (Eq. (2)) presented a Q 2 leave-one-out of 0.72,
and the model with MWCO (Eq. (3)) presented a Q 2 leave-oneout
of 0.72. After internal cross-validation it was demonstrated
that Eqs. (2) and (3) were valid to model rejection,
however, an adjustment must be made to the equation before
using it to compare measured vs. predicted rejections for
external databases. This adjustment was necessary to overcome
the mathematical structure of the equation. Using
a physical interpretation, it was evident that size parameters
referring to variables length and equivalent width may be
large enough to cause rejection predictions over 100%, which
can be explained after observing positive coefficients for
equivalent width and length. This situation may also be
detrimental for rejection predictions of ionic compounds of
medium to large size (0.6–1.2 nm as equivalent width) that
mostly are rejected due to electrostatic repulsion and less
steric hindrance. Therefore Eqs. (2) and (3) can be transformed
to the following conditional equation
rejection ¼
a b
measured rejection (%)
HB-ion
HB-neu
HL-ion
HL-neu
95%
confidence
interval
100 if QSAR model 100
QSAR model
modeled rejection (%)
length, eqwidth,
depth, log D, SR
R² = 0.75
An external dataset (that gathered three different datasets)
was selected for external validation of the QSAR model. The
external dataset is presented as supplementary data. The first
part of the external dataset corresponds to membrane Filmtec
NF-90. The second part corresponds to NF membrane Trisep
TS-80 and the third part corresponds to NF membrane Desal
HL. Desal HL membrane has a main difference with NF-90 and
Trisep TS-80 membranes, viz. the MWCO of Desal HL is in the
range of 150–300 Da, while NF-90 and Trisep TS-80 were
reported to have a MWCO of 200 Da. Therefore an average
MWCO of 225 Da was assumed for Desal HL during the
application of Eq. (3). It is worthwhile to mention that the
water research 44 (2010) 373–384
(4)
measured rejection (%)
HB-ion
HB-neu
HL-ion
HL-neu
95%
confidence
interval
modeled rejection (%)
Fig. 1 – QSAR model of experimental database: (a) Eq. (2); and (b) Eq. (3).
length, eqwidth,
depth, log D,
MWCO
R² = 0.75
second and third parts of the external dataset were generated
using spiral wound membrane elements instead of flat sheet
membranes. Fig. 2a show plotted results of measured rejections
vs. predicted rejections after calculations with Eq. (4), for
QSAR model with Eq. (2) (SR). According to Fig. 2aanR 2 of 0.75
was obtained after considering all compounds of the external
dataset. However, after observing the rejection cases by NF-90
for bromoform (BF) and trichloroethene (TCE), it appeared that
BF and TCE may be considered atypical results because their
rejections did not correspond to their size and hydrophobicity
when compared to other compounds with comparable
molecular descriptors. According to Table 3 BF has approximately
the same hydrophobicity and polarity as CF, but BF is
bigger than CF; therefore, measured rejection for BF was
expected to be higher than rejection of CF (0%) due to size
exclusion. The measured rejection of TCE (3%) was not
comparable to the rejection of perchloroethene (39%)
although they have the same length but a very small difference
in equivalent width. In an experiment conducted by Kim
et al. (2007) rejections of BF and TCE were of 50 and 33%,
respectively, for a low pressure reverse osmosis membrane. It
was also observed that the measured rejection (53%) of
linuron (LNU) may be considered as atypical observation
because a higher rejection was expected. According to Table 3,
monolinuron is close to LNU in size and polarity; thus
measured rejection of monolinuron was of 77%, higher than
53%. Also according to Table 3, carbamazepine is close to LNU
in size and hydrophobicity; thus measured rejection of
carbamazepine was of 94%, higher than 53%. A similar
explanation was given for the low rejection (70%) of N-acetyl-
L-tyrosine (NAT) by Desal HL compared to 94% rejection by
TS-80 as can be seen in Table 3. Other non-expected rejection
was that of 2-(1H)-quinoline (QNL), with a rejection of 22%
when compared to 2-methoxyethanol and perchloroethene
with rejections of 32 and 39%, respectively; even though the
length of QNL is greater than the length of 2-methoxyethanol
and perchloroethene, a lower rejection of QNL was observed.
However, rejections of methacetin (MTC) and NDPA may be
influenced by their small equivalent width as can be seen in

a b
measured rejection (%)
100
80
60
40
ETH
MTBE
R 2 20
QNL
NDPA
0
BF TCE
MTC
= 0.75
0 20 40 60 80 100
predicted rejection (%)
Table 3. The response of the model for rejection of those
particular compounds (with 2*equivalent width < or w length)
was of over prediction, but the model presented better
response for rejections of metribuzin, atrazin and N-acetyl-Ltyrosine
(all with length >2*equivalent width) as observed in
Table 3. Ethanol (ETH) and MTBE with rejections of 38 and 60%,
respectively, were expected to be lower for Desal HL
membrane because it was observed that rejection of ETH was
of 9% for TS-80; and MTBE was expected to have a rejection
compared to that of 2-methoxyethanol (32%) or perchloroethene
(39%) due to proximities in size. After selection
water research 44 (2010) 373–384 381
NDPA
MTC
NAT
LNU
0
0 20 40 60 80 100
predicted rejection (%)
R 2 = 0.84
and justified separation of the mentioned rejection cases,
Fig. 2b presents the predictions of the external dataset with an
R 2 of 0.84. In a similar explanatory scenario, Figs. 3a,b show
measured rejections vs. predicted rejections after calculations
with Eq. (4), for QSAR model with equation 3 (MWCO). The
main difference between Figs. 2b and 3b was that the model
with MWCO (Fig. 3b) showed a lower R 2 (0.80) than the model
with SR (R 2 ¼ 0.84), meaning that the latter had a better
goodness of fit for external prediction response. Moreover, the
characterization of magnesium sulphate salt rejection for
a membrane may be preferred instead of MWCO, particularly
Table 3 – Partial list of rejections and compound properties of external dataset.
Compound Abb. Length Eqwidth Depth Log D Dipole Measured rejection Predicted rejection Membrane
Chloroform 0.53 0.42 0.35 1.97 1.12 0 0 NF-90
Ethanol 0.64 0.52 0.51 0.31 1.55 9 14 TS-80
Ethanol ETH 0.64 0.52 0.51 0.31 1.55 38 0 Desal HL
Carbontetrachloride 0.64 0.6 0.57 2.83 0.30 35 26 NF-90
Bromoform BF 0.69 0.56 0.48 2.40 1.00 0 33 NF-90
MTBE MTBE 0.77 0.63 0.59 0.94 1.37 60 25 Desal HL
Trichloroethene TCE 0.78 0.49 0.36 2.29 0.95 3 36 NF-90
Perchloroethene 0.78 0.59 0.45 3.40 0.11 39 46 NF-90
2-methoxyethanol 0.87 0.52 0.51 0.77 0.25 32 34 TS-80
2-(1H)-quinoline QNL 1.00 0.52 0.36 1.26 3.38 22 53 TS-80
NDPA NDPA 1.16 0.60 0.53 1.36 3.40 45 69 TS-80
NDPA NDPA 1.16 0.60 0.53 1.36 3.40 19 55 Desal HL
Metribuzin 1.17 0.74 0.64 0.47 0.52 97 99 TS-80
Carbamazepine 1.20 0.73 0.58 2.45 3.66 88 94 TS-80
Linuron LNU 1.21 0.69 0.53 3.20 2.11 53 84 TS-80
Monolinuron 1.22 0.69 0.65 2.30 2.02 79 77 TS-80
Atrazin 1.26 0.74 0.55 2.61 3.43 91 100 TS-80
Methacetin MTC 1.28 0.52 0.42 1.03 2.20 38 72 TS-80
Methacetin MTC 1.28 0.52 0.42 1.03 2.20 5 58 Desal HL
N-acetyl-L-tyrosine NAT 1.33 0.71 0.60 2.18 3.45 94 100 TS-80
N-acetyl-L-tyrosine NAT 1.33 0.71 0.60 2.18 3.45 70 100 Desal HL
measured rejection (%)
NF-90 TS-80 HL Linear ( ) NF-90 TS-80 HL Linear ( )
Fig. 2 – Predicted rejections for external dataset using magnesium sulphate SR: (a) all external data; and (b) selected external
data.
100
80
60
40
20

382
measured rejection (%)
100
80
60
40
20
for nanofiltration and low pressure reverse osmosis
membranes; besides, the effect of fouling in membranes can
also be quantified by salt rejection experiments. We can state
that the QSAR model with SR demonstrated to be acceptable
for the external dataset of NF-90, Trisep TS-80 and Desal HL
with an R 2 of 0.75 and 0.84 for the total external dataset and
justified selected external dataset, respectively. Although the
model can be valid with limitations related to boundary
experimental conditions mentioned in Section 3, its applicability
and approach can be of value for the construction of
a model with combined datasets organized in training and
testing groups.
5. Conclusions
0
ETH
BF
MTBE
TCE
QNL
NDPA
MTC
R 2 = 0.74
0 20 40 60 80 100
predicted rejection (%)
– A general QSAR model equation was developed to merge
information about interaction of membrane characteristics,
filtration operating conditions and solute properties
to predict rejections of emerging contaminants during
nanofiltration.
– The QSAR model identified that the most important
variables that influence rejection of organic solutes were
log D, salt rejection, equivalent width, depth and length.
– Rejection increased by size/steric hindrance effects,
solute hydrophobicity decreased rejection due to adsorption
and partitioning mechanisms.
– Salt rejection incorporated steric hindrance and electrostatic
repulsion effects that were related to the membrane
structure and operating conditions.
– The use of MWCO was acceptable for modelling purposes;
however NF membranes with a broad range of MWCO
(pore size and distribution) may difficult estimation of
rejection of contaminants, thus magnesium sulphate salt
rejection may be more appropriate.
water research 44 (2010) 373–384
a b
NDPA
MTC
NAT
LNU
measured rejection (%)
100
80
60
40
20
0
0 20 40 60 80 100
predicted rejection (%)
Acknowledgements
The authors acknowledge Delft Cluster and EU Techneau
Project for funding this project. The authors also acknowledge
Tae-Uk Kim and Arne Verliefde for providing data and details
of their research. The authors thank Filmtec (Dow Chemical
Co.) for donating the membranes, Dr. Jaeweon Cho of GIST
(Korea), Dr. Frank Sacher of TZW (Germany) and Steven
Mookhoek of TU Delft (Netherlands) for contributing with
analytical results and facilities.
Appendix.
Supplementary data
Supplementary data associated with this article can be found,
in the online version, at doi:10.1016/j.watres.2009.06.054
references
R 2 = 0.80
NF-90 TS-80 HL Linear ( ) NF-90 TS-80 HL Linear ( )
Fig. 3 – Predicted rejections for external dataset using MWCO: (a) all external data; and (b) selected external data.
Abdi, H., 2003. Partial least squares (PLS) regression. In: Lewis-
Beck, M., Bryman, A., Futing, T. (Eds.), Encyclopaedia of Social
Sciences Research Methods. Sage, Thousand Oaks, CA.
Bellona, C., Drewes, J.E., 2005. The role of membrane surface
charge and solute physico-chemical properties in the rejection
of organic acids by NF. Journal of Membrane Science 249 (1–2),
227–234.
Campbell, P., Srinivasan, R., Knoell, T., Phipps, D., Ishida, K.,
Safarik, J., Cormack, T., Ridgway, H., 1999. Quantitative
structure-activity relationship (QSAR) analysis of surfactants
influencing attachment of a Mycobacterium sp. to cellulose

acetate and aromatic polyamide reverse osmosis membranes.
Biotechnology and Bioengineering 64 (5), 527–544.
Castiglioni, S., Bagnati, R., Fanelli, R., Pomati, F., Calamari, D.,
Zuccato, E., 2006. Removal of pharmaceuticals in sewage
treatment plants in Italy. Environmental Science &
Technology 40 (1), 357–363.
Connell, D.W., 1990. Bioaccumulation of Xenobiotic Compounds.
CRC Press, Boca Raton, FL, pp. 75-110.
Cornelissen, E.R., Verdouw, J., Gijsbertsen-Abrahamse, A.J.,
Hofman, J.A.M.H., 2005. A nanofiltration retention model for
trace contaminants in drinking water sources. Desalination
178 (1–3), 179–192.
Drewes, J.E., Amy, G., Kim, T.-U., Xu, P., Bellona, C.,
Oedekoven, M., Macalady, D., 2006. Rejection of Wastewater-
Derived Micropollutants in High-Pressure Membrane
Applications Leading to Indirect Potable Reuse Effect of
Membranes and Micropollutant Properties. WateReuse,
Alexandria, VA.
Eriksson, L., Jaworska, J., Worth, A.P., Cronin, M.T.D.,
McDowell, R.M., Gramatica, P., 2003. Methods for reliability
and uncertainty assessment and for applicability evaluations
of classification and regression-based QSARs. Environmental
Health Perspectives 111 (10), 1361–1375.
Escher, B.I., Bramaz, N., Eggen, R.I.L., Richter, M., 2005. In vitro
assessment of modes of toxic action of pharmaceuticals in
aquatic life. Environmental Science & Technology 39 (9),
3090–3100.
Everitt, B.S., Dunn, G., 2001. Applied Multivariate Data Analysis.
Arnold, London.
Fujikawa, M., Nakao, K., Shimizub, R., Akamatsu, M., 2007. QSAR
study on permeability of hydrophobic compounds with
artificial membranes. Bioorganic & Medicinal Chemistry 15
(11), 3753–3767.
Gramatica, P., 2007. Principles of QSAR models validation:
internal and external. QSAR & Combinatorial Science 26 (5),
694–701.
Heberer, T., 2002. Occurrence, fate, and removal of
pharmaceutical residues in the aquatic environment: a review
of recent research data. Toxicology Letters 131 (1–2), 5–17.
Henderson, R.K., Baker, A., Parsons, S.A., Jefferson, B., 2008.
Characterisation of algogenic organic matter extracted from
cyanobacteria, green algae and diatoms. Water Research 42
(13), 3435–3445.
Hoek, E.M.V., Elimelech, M., 2003. Cake-enhanced concentration
polarization: a new fouling mechanism for salt-rejecting
membranes. Environmental Science & Technology 37 (24),
5581–5588.
Johnson, R.A., Wichern, D.W., 2007. Applied Multivariate
Statistical Analysis. Pearson Prentice Hall, Upper Saddle
River, NJ.
Jolliffe, I.T., 2002. Principal Component Analysis. Springer-Verlag,
New York.
Jørgensen, B., Goegebeur, Y., 2006. Multivariate Data Analysis and
Chemometrics. Department of Statistics, University of
Southern Denmark.
Kim, T.-U., Drewes, J.E., Summer, R.S., Amy, G., 2007. Solute
transport model for trace organic neutral and charged
compounds through nanofiltration and reverse osmosis
membranes. Water Research 41 (17), 3977–3988.
Kimura, K., Amy, G., Drewes, J., Heberer, T., Kim, T., Watanabe, Y.,
2003. Rejection of organic micropollutants (disinfection byproducts,
endocrine disrupting compounds, and
pharmaceutically active compounds) by NF/RO membranes.
Journal of Membrane Science 227 (1–2), 113–121.
Kimura, K., Toshima, S., Amy, G., Watanabe, Y., 2004. Rejection of
neutral endocrine disrupting compounds (EDCs) and
pharmaceutical active compounds (PhACs) by RO membranes.
Journal of Membrane Science 245 (1–2), 71–78.
water research 44 (2010) 373–384 383
Kiso, Y., Sugiura, T., Kitao, T., Nishimura, K., 2001. Effects of
hydrophobicity and molecular size on rejection of aromatic
pesticides with nanofiltration membranes. Journal of
Membrane Science 192 (1–2), 1–10.
Kolpin, D.W., Furlong, E.T., Meyer, M.T., Thurman, E.M., Zaugg, S.D.,
Barber, L.B., Buxton, H.T., 2002. Pharmaceuticals, hormones,
and other organic wastewater contaminants in U.S. Streams,
1999–2000: a National Reconnaissance. Environmental Science
& Technology 36 (6), 1202–1211.
Landau, S., Everitt, B., 2004. A Handbook of Statistical Analysis
using SPSS. Chapman & Hall/CRC, Boca Raton, FL.
Lee, N., Amy, G., Croue, J.-P., Buisson, H., 2004. Identification and
understanding of fouling in low-pressure membrane (MF/UF)
filtration by natural organic matter (NOM). Water Research 38
(20), 4511–4523.
Libotean, D., Giralt, J., Rallo, R., Cohen, Y., Giralt, F., Ridgway, H.,
Rodriguez, G., Phipps, D., 2008. Organic compounds passage
through RO membranes. Journal of Membrane Science 313
(1-2), 23–43.
Nghiem, L.D., Schäfer, A.I., Elimelech, M., 2004. Removal of
natural hormones by nanofiltration membranes:
measurement, modelling, and mechanisms. Environmental
Science & Technology 38 (6), 1888–1896.
Nghiem, L.D., Schäfer, A.I., Elimelech, M., 2005. Pharmaceutical
retention mechanisms by nanofiltration membranes.
Environmental Science & Technology 39 (19), 7698–7705.
Ozaki, H., Li, H., 2002. Rejection of organic compounds by ultralow
pressure reverse osmosis membrane. Water Research 36
(1), 123–130.
Pomati, F., Castiglioni, S., Zuccato, E., Fanelli, R., Vigetti, D.,
Rossetti, C., Calamari, C., 2006. Effects of a complex mixture of
therapeutic drugs at environmental levels on human
embryonic cells. Environmental Science & Technology 40 (7),
2442–2447.
Ren, S., Das, A., Lien, E.J., 1996. QSAR analysis of membrane
permeability to organic compounds. Journal of Drug Targeting
4 (2), 103–107.
Sacher, F., Lange, F.T., Brauch, H.-J., Blankenhorn, I., 2001.
Pharmaceuticals in groundwaters: analytical methods and
results of a monitoring program in Baden-Württemberg,
Germany. Journal of Chromatography A 938 (1–2), 199–210.
Sacher, F., Ehmann, M., Gabriel, S., Graf, C., Brauch, H.-J., 2008.
Pharmaceutical residues in the river Rhinedresults of a onedecade
monitoring programme. Journal of Environmental
Monitoring 10 (5), 664–670.
Santos, J.L.C., de Beukelaar, P., Vankelecom, I.F.J., Velizarov, S.,
Crespo, J.G., 2006. Effect of solute geometry and orientation on
the rejection of uncharged compounds by nanofiltration.
Separation and Purification Technology 50 (1), 122–131.
Sawyer, C.N., McCarty, P.L., Parkin, G.F., 2003. Chemistry for
Environmental Engineering and Science. McGraw-Hill, New
York, pp. 479-520.
Schäfer, A.I., Nghiem, L.D., Waite, T.D., 2003. Removal of the
natural hormone estrone from aqueous solutions using
nanofiltration and reverse osmosis. Environmental Science &
Technology 37 (1), 182–188.
Shim, Y., Lee, H.-J., Lee, S., Moon, S.-H., Cho, J., 2002. Effects of
natural organic matter and ionic species on membrane surface
charge. Environmental Science & Technology 36 (17), 3864–3871.
Van der Bruggen, B., Vandecasteele, C., 2002. Modeling of the
retention of uncharged molecules with nanofiltration. Water
Research 36 (5), 1360–1368.
Van der Bruggen, B., Schaep, J., Wilms, D., Vandecasteele, C.,
1999. Influence of molecular size, polarity and charge on the
retention of organic molecules by nanofiltration. Journal of
Membrane Science 156 (1), 29–41.
Verliefde, A., Cornelissen, E., Amy, G., Van der Bruggen, B., van
Dijk, H., 2007. Priority organic micropollutants in water

384
sources in Flanders and the Netherlands and assessment of
removal possibilities with nanofiltration. Environmental
Pollution 146 (1), 281–289.
Verliefde, A.R.D., Cornelissen, E.R., Heijman, S.G.J., Verberk, J.Q.J.C.,
Amy, G.L., Van der Bruggen, B., van Dijk, J.C., 2008. The role of
electrostatic interactions on the rejection of organic solutes in
aqueous solutions with nanofiltration. Journal of Membrane
Science 322 (1), 52–66.
Vosges, M., Braguer, J.-C., Combarnous, Y., 2008. Long-term
exposure of male rats to low-dose ethinylestradiol (EE2)
in drinking water: effects on ponderal growth and on
litter size of their progeny. Reproductive Toxicology 25
(2), 161–168.
water research 44 (2010) 373–384
Xu, P., Drewes, J.E., Bellona, C., Amy, G., Kim, T.-U., Adam, M.,
Heberer, T., 2005. Rejection of emerging organic
micropollutants in nanofiltration-reverse osmosis membrane
applications. Water Environment Research 77 (1), 40–48.
Yangali-Quintanilla, V., Kim, T.-U., Kennedy, M., Amy, G., 2008.
Modeling of RO/NF membrane rejections of PhACs and organic
compounds: a statistical analysis. Drinking Water Engineering
and Science 1 (1), 7–15.
Yangali-Quintanilla, V., Sadmani, A., McConville, M.,
Kennedy, M., Amy, G., 2009. Rejection of pharmaceutically
active compounds and endocrine disrupting compounds by
clean and fouled nanofiltration membranes. Water Research
43 (9), 2349–2362.

Harmful algae and their potential impacts on desalination
operations off southern California
David A. Caron a, *, Marie-Ève Garneau a , Erica Seubert a , Meredith D.A. Howard a,b ,
Lindsay Darjany a , Astrid Schnetzer a , Ivona Cetinić a , Gerry Filteau c , Phil Lauri d ,
Burton Jones a , Shane Trussell e
a
Department of Biological Sciences, University of Southern California, 3616 Trousdale Parkway, Los Angeles, CA 90089-0371, USA
b
Southern California Coastal Water Research Project, 3535 Harbor Blvd., Suite 110, Costa Mesa, CA 92626, USA
c
Separation Processes, Inc., 3156 Lionshead Avenue, Suite 2, Carlsbad, CA 92010, USA
d
West Basin Municipal Water District, 17140 Avalon Blvd., Suite 210, Carson, CA 90746, USA
e
Trussell Technologies, Inc., 6540 Lusk Boulevard, Suite C175, San Diego, CA 92121, USA
article info
Article history:
Received 2 April 2009
Received in revised form
12 June 2009
Accepted 23 June 2009
Available online 30 June 2009
Keywords:
Harmful algal blooms
Desalination
Red tides
Phytoplankton
Phytotoxins
water research 44 (2010) 385–416
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
abstract
* Corresponding author. Tel.: þ1 213 740 0203; fax: þ1 213 740 8123.
E-mail address: dcaron@usc.edu (D.A. Caron).
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.watres.2009.06.051
Seawater desalination by reverse osmosis (RO) is a reliable method for augmenting
drinking water supplies. In recent years, the number and size of these water projects have
increased dramatically. As freshwater resources become limited due to global climate
change, rising demand, and exhausted local water supplies, seawater desalination will play
an important role in the world’s future water supply, reaching far beyond its deep roots in
the Middle East. Emerging contaminants have been widely discussed with respect to
wastewater and freshwater sources, but also must be considered for seawater desalination
facilities to ensure the long-term safety and suitability of this emerging water supply.
Harmful algal blooms, frequently referred to as ‘red tides’ due to their vibrant colors, are
a concern for desalination plants due to the high biomass of microalgae present in ocean
waters during these events, and a variety of substances that some of these algae produce.
These compounds range from noxious substances to powerful neurotoxins that constitute
significant public health risks if they are not effectively and completely removed by the RO
membranes. Algal blooms can cause significant operational issues that result in increased
chemical consumption, increased membrane fouling rates, and in extreme cases, a plant to
be taken off-line. Early algal bloom detection by desalination facilities is essential so that
operational adjustments can be made to ensure that production capacity remains unaffected.
This review identifies the toxic substances, their known producers, and our present
state of knowledge regarding the causes of toxic episodes, with a special focus on the
Southern California Bight.
ª 2009 Elsevier Ltd. All rights reserved.

386
1. Introduction
1.1. General overview of harmful algal blooms:
a growing global concern
Microscopic algae constitute an essential component of all
aquatic food webs. Photosynthetic production of organic
material by this diverse group of species comprises the
primary source of nutrition for all heterotrophic forms of life
in much of the world’s ocean and freshwater ecosystems.
Microalgae can reach high abundances in the plankton during
periods of optimal growth and reduced grazing pressure by
herbivores. Such localized mass proliferations are known as
algal (or phytoplankton) blooms. In addition, a small proportion
of microalgal species are capable of producing a number
of noxious or toxic compounds that cause a variety of adverse
effects on ecosystem structure and function. These
substances pose the potential for ecosystem damage, food
web disruption and marine animal mortality, and present
a significant human health risk through the consumption of
contaminated seafood and, in at least one case, direct exposure
to water or aerosols containing these toxic compounds.
Additionally, the algal biomass and the associated organic
load cause significant desalination operational issues,
impacting the pretreatment system and possibly forcing the
treatment plant to be taken off-line (Petry et al., 2007).
Countless human deaths resulting from the consumption
of seafood contaminated with algal toxins have been avoided
through rigorous monitoring programs, but sea life has not
been so fortunate. Approximately one half of all unusual
marine mammal mortality incidents are now attributable to
the ingestion of food or prey contaminated by harmful algal
blooms (Ramsdell et al., 2005). Losses in revenue due to the
direct contamination of seafood products and indirect effects
on tourism and other uses of coastal areas have been estimated
in the tens of millions of dollars annually in the U.S.
states along the Pacific coast (Trainer et al., 2002).
There is now convincing evidence that harmful algal
bloom (HAB) events are increasing at local, regional and global
scales worldwide (Smayda, 1990; Hallegraeff, 1993, 2003;
Anderson et al., 2002; Glibert et al., 2005a) and along the North
American west coast in particular (Horner et al., 1997; Trainer
et al., 2003). This increased occurrence may be due in part to
better detection of HAB episodes in recent years or the global
dispersal of toxic algal species via the transport of resting
spores in ships’ ballast waters (Hallegraeff and Bolch, 1992;
Burkholder et al., 2007), but another very likely cause is the
increasing impact of anthropogenic activities on coastal
ecosystems (Smayda, 1990; Anderson et al., 2002; Glibert et al.,
2005b, 2006; Howard et al., 2007; Cochlan et al., 2008; Kudela
et al., 2008a). Recent reports reveal extensive and, in some
cases, newly emerging occurrences of HABs along the coasts
of the U.S. (Fig. 1). These incidents engender a variety of
noxious impacts on ecosystems and public health, including
direct effects on organisms due to the production of acutely
toxic substances, and indirect effects such as reduced availability
of dissolved oxygen in the water column resulting from
the decomposition of the extensive amounts of organic
substances usually produced during such blooms. The
water research 44 (2010) 385–416
Fig. 1 – Distribution of some well-known regional HAB
issues along U.S. shores, including (a) Alaska and (b)
Hawaii. Causes and impacts of these poisoning events are
defined in Tables 1–3. Summarized from information
presented on the Harmful Algae webpage (http://www.
whoi.edu/redtide/).
dramatic increases in biomass and organic load that accompany
these events pose a significant threat to seawater desalination
facilities (Gaid and Treal, 2007).
1.2. Regional HAB issues along U.S. coastlines
Harmful algae are present throughout U.S. coastal waters, but
not all species are of equal concern in all regions (Fig. 1). For
example, toxic species of the dinoflagellate genus Alexandrium
are common over vast stretches of the U.S. coastline, but
coastal regions of the northeastern and northwestern U.S.
appear to experience particularly high rates of occurrence of
toxic ‘red tides’ caused by these species. The neurotoxins
produced by Alexandrium, called saxitoxins, cause paralytic
shellfish poisoning (PSP) in humans when ingested through
contaminated seafood (particularly filter-feeding shellfish).
Similarly, several toxic species of the diatom genus Pseudonitzschia
occur along the entire U.S. coastline but significant
concentrations of the neurotoxin, domoic acid, produced by
these species have historically constituted a health threat
primarily in the northeastern and northwestern U.S. (Bates
et al., 1989) where it has been documented as the cause of
amnesic shellfish poisoning (ASP) in humans. However, high
concentrations of domoic acid in the plankton and in diverse
planktivorous organisms have been recently documented
along the entire Pacific coast of the U.S. (Scholin et al., 2000;
Trainer et al., 2002; Schnetzer et al., 2007), as well as in the
Gulf of Mexico (Pan et al., 1998). Domoic acid has been
attributed to numerous marine animal mortalities along the
U.S. west coast. In the Gulf of Mexico, primarily along the
west coast of Florida, extensive and recurrent blooms of the
dinoflagellate Karenia brevis produce a suite of toxins, known
as brevetoxins, that can be aerosolized by breaking waves and

induce neurotoxic shellfish poisoning (NSP) in people inhaling
the aerosols (see review of Kirkpatrick et al., 2004). The
Tampa Bay seawater desalination facility is the only operating
seawater desalination treatment plant of significant
size in the United States. It is located along the west coast of
Florida and is likely to encounter algal blooms that contain
brevetoxin.
Less toxic blooms also take place with regional specificity.
The pelagophyte Aureococcus anophagefferens causes ‘brown
tides’ in coastal waters of Rhode Island, near Long Island (NY)
and southward along the mid-Atlantic coast of the U.S. since
1985. No specific toxins have been identified from A. anophagefferens,
and no human fatalities have been directly attributed
to these blooms. Nevertheless, this species appears to be
unpalatable or inhibitory to many filter-feeding mollusks and
has caused substantial mortality among these populations,
including commercially valuable species (Bricelj and Lonsdale,
1997).
Other microalgal species can disrupt food webs or cause
reductions in water quality without producing acutely toxic
conditions. Among these are the ‘colorful’ red tides of the
dinoflagellate Lingulodinium polyedrum, a yessotoxin producer,
that have occurred periodically throughout several decades
along the south and central Californian coasts (Horner et al.,
1997; Gregorio and Pieper, 2000). These blooms have so far
been found to be relatively innocuous in these waters but
massive accumulations of these cells could have significant
impact on desalination plants because of increased turbidity,
high suspended solids and organic loading of influent water.
Furthermore, accumulations of cells in protected harbors can
cause fish mortality by depleting oxygen dissolved in the
water, further challenging influent screening and pretreatment
systems at desalination plants. Other taxa, such as
species of the prymnesiophyte genus Phaeocystis, produce
substances that can lead to enormous buildups of sea foam
along coasts (Armonies, 1989).
1.3. Desalination, plankton and water quality issues
Large research programs have developed within different
geographic areas throughout the U.S. to address regional HAB
issues. These programs are designed to study the species,
toxins and environmental causes of HAB outbreaks. These
efforts, as well as local, county, state and federal monitoring
programs provide basic information for marine resource use
and have focused almost exclusively on threats to human
health via the consumption of contaminated seafood. Unfortunately,
few if any of these programs provide sufficient
information on appropriate temporal and spatial resolution
for thoroughly assessing the potential impact of HAB events
on reverse osmosis desalination operations. Moreover, toxin
analyses have primarily examined the presence of these
substances in particulate material (plankton or animal tissue,
particularly shellfish and finfish), and therefore may be poor
predictors for the amount of toxins that might occur in
seawater in the dissolved state during algal blooms, which
would be most likely to be loaded onto reverse osmosis
membranes during desalination.
There are two potential impacts that HABs may have on
seawater desalination facilities: (1) algal toxins in ocean water
water research 44 (2010) 385–416 387
pose a significant treatment challenge for the reverse osmosis
system to ensure that these molecules are effectively removed
and (2) increased turbidity, total suspended solids and total
organic content resulting from algal biomass and growth
challenge the entire desalination facility’s treatment train.
The significance of these issues will depend on the specific
algae forming a bloom and the toxin(s) or other substances
that they produce, the magnitude and duration of the bloom,
and the specific desalination process conducted. For example,
multi-effect distillation and multi-flash distillation might be
susceptible to (2) but would be much less affected by toxins in
the water (1). Desalination using reverse osmosis presumably
would be vulnerable to both issues. Therefore, for the latter
desalination approach, a thorough understanding of HAB
episodes in terms of incidence and seasonality, vertical and
horizontal spatial distribution, as well as biological aspects
such as algal composition within a geographical region could
help optimize the design and operational efficiency of desalination
plants employing reverse osmosis.
This paper provides an overview of HABs occurring along
the continental U.S. coastline with special emphasis on the
southwestern U.S., and provides some insight on the potential
impacts that these events may have on the seawater desalination
process. In recent years, this geographical area has
become a focal point of discussions regarding desalination
(Cooley et al., 2006) because of its sizable population and the
particularly tenuous nature of the water supply to this region.
Although numerous issues involving the desalination process
are now being examined (Separation Processes Inc., 2005; Gaid
and Treal, 2007; Pankratz, 2008, 2009), very limited information
exists on the risks that algal blooms pose to seawater
desalination facilities. A review of the major species
producing harmful blooms, the substances they produce, and
information on the spatial and temporal distributions of
blooms are presented along with some conclusions on their
potential impacts. This paper also provides some general
guidelines on how early detection may help prevent or minimize
the impact of HABs on a desalination facility’s production
capacity or its water quality.
2. Toxin producers and toxin concentrations
of the west coast
A variety of toxins including several powerful neurotoxins are
produced by microalgae, and a number of these toxins and
potentially toxic algal species have been detected on the U.S.
west coast (Table 1). The ability to rapidly detect and quantify
toxic algae in natural water samples is problematic at this
time. Many of these species are difficult to identify using light
microscopy. For this reason, new genetic and immunological
methods for species identification and enumeration have
been appearing rapidly in the literature (Miller and Scholin,
1998; Bowers et al., 2000, 2006; Coyne et al., 2001; Caron et al.,
2003; Galluzzi et al., 2004; Anderson et al., 2005; Mikulski et al.,
2005, 2008; Ahn et al., 2006; Handy et al., 2006; Moorthi et al.,
2006; Iwataki et al., 2007, 2008; Demir et al., 2008; Matsuoka
et al., 2008). Moreover, many toxin-producing algal species
exhibit variable toxin production in response to environmental
conditions, and among different strains of the same species

388
Table 1 – Planktonic species occurring along the west coast of the U.S. that are potential concerns for reverse osmosis
operations.
Microalgae Toxin(s) Poisoning Event References
Diatoms
Pseudo-nitzschia spp.
P. australis b
P. cuspidata b
P. delicatissima b
P. fraudulenta b
P. multiseries b
P. pungens b
P. pseudodelicatissima b
P. seriata a
Dinoflagellates
Alexandrium spp.
A. acatenella a
A. catenella b
A. fundyense a
A. hiranoi a
A. ostenfeldii a
A. tamarense a
Dinoflagellates
Lingulodinium polyedrum b
Gonyaulax spinifera a
Protoceratium reticulatum a,c
Dinoflagellates
Dinophysis spp.
D. acuminata a
D. acuta a
D. caudate
D. fortii a
D. norvegica a
D. rotundata a
D. tripos a
Prorocentrum spp.
P. micans
P. minimum a,d
Raphidophytes
Chattonella marina a
Fibrocapsa japonica a
Heterosigma akashiwo a
Domoic acid (DA) Amnesic Shellfish Poisoning (ASP)
even when isolated from the same geographic region (Smith
et al., 2001; Trainer et al., 2001; Kudela et al., 2004).
Laboratory experiments have revealed a wide range of
physico-chemical factors that increase or decrease toxin
Human effects
Gastro-intestinal symptoms
Neurologic symptoms
Death
Ecosystem effects
Marine mammal mortalities
Bird mortalities
Saxitoxins (STXs) Paralytic Shellfish Poisoning (PSP)
Human effects
Gastro-intestinal symptoms
Paralysis
Death
Ecosystem effects
Marine mammal mortalities
Yessotoxins (YTXs) Human and ecosystem
effects
None reported
Okadaic acid (OA)
Dinophysistoxins (DTXs)
Pectenotoxins (PTXs)
water research 44 (2010) 385–416
Diarrhetic Shellfish
Poisoning (DSP)
Human effects
Gastro-intestinal symptoms
Ecosystem effects
None reported
Brevetoxins (PbTxs) Neurotoxic Shellfish
Poisonining (NSP)
Human effects
Gastroenteritis
Neurologic symptoms
Respiratory irritation
and/or failure
Ecosystem effects
Marine mammal
mortalities
Fish mortality events
a Reported to produce toxin.
b Reported to produce toxin on the west coast of the United States.
c Conflicting reports on toxicity of P. reticulatum cultures isolated from California, Washington and Florida.
d Reported to be present on the west coast of Mexico.
Subba Rao et al. (1988), Bates et al. (1989),
Martin et al. (1990), Buck et al. (1992),
Garrison et al. (1992), Rhodes et al. (1996),
Horner et al. (1997), Lundholm et al. (1997),
Rhodes et al. (1998), Trainer et al. (2000, 2001),
Baugh et al. (2006)
Sommer and Meyer (1937),
Gaines and Taylor (1985),
Steidinger (1993),
Scholin et al. (1994),
Taylor and Horner (1994),
Jester (2008)
Holmes et al. (1967),
Satake et al. (1997, 1999),
Draisci et al. (1999a),
Paz et al. (2004, 2007),
Armstrong and Kudela (2006),
Rhodes et al. (2006), Howard et al. (2007)
Holmes et al. (1967),
Yasumoto et al. (1980),
Murata et al. (1982),
Yasumoto et al., (1985), Cembella (1989),
Lee et al. (1989), Horner et al. (1997),
Cembella (2003), Miles et al. (2004),
Shipe et al. (2008), Sutherland (2008)
Loeblich and Fine (1977),
Hershberger et al. (1997),
Gregorio and Connell (2000),
Hard et al. (2000),
Tyrell et al. (2002), O’Halloran et al. (2006)
production by harmful species of algae, and which appear to be
species-specific (see review of Granéli and Flynn, 2006). Reports
of factors inducing toxin production have sometimes been
conflicting, presumably indicating that multiple factors, or

perhaps generally stressful conditions, may stimulate toxin
production. Factors affecting toxin production include: (1)
temperature (Ono et al., 2000); (2) light intensity (Ono et al., 2000);
(3) salinity (Haque and Onoue, 2002a,b); (4) trace metal availability,
especially iron (Ladizinsky and Smith, 2000; Rue and
Bruland, 2001; Maldonado et al., 2002; Wells et al., 2005; Sunda,
2006) but also copper (Maldonado et al., 2002) andselenium
(Mitrovic et al., 2004, 2005); (5) macronutrient availability
including silicate (Pan et al., 1996b; Fehling et al., 2004; Kudela
et al., 2004), phosphate (Pan et al., 1996a, 1998; Fehling et al.,
2004), nitrogen (Bates et al., 1991; Pan et al., 1998; Kudela
et al., 2004) and combinations of nutrient limitation (Anderson
et al., 1990; Flynn et al., 1994; John and Flynn, 2000); (6) cellular
elemental ratios of nutrients and physiological stress (Granéli
and Flynn, 2006; Schnetzer et al., 2007); (7) growth phase
(Anderson et al., 1990; Bates et al., 1991; Flynn et al., 1994;
Johansson et al., 1996; Maldonado et al., 2002; Mitrovic et al.,
2004). The precise combination(s) of environmental factors that
select for population growth of particular algal species within
diverse natural assemblages, and the specific conditions that
induce toxin production, are poorly understood for most
harmful algae. This present state of knowledge makes it difficult
to predict the timing, duration or spatial extent of the vast
majority of HAB events and the toxic events resulting from
them.
Our ability to thoroughly characterize HABs is also
complicated by the complex array of toxins produced by algae.
Marine algal species produce a suite of toxic components
(Yasumoto and Murata, 1993), and unidentified toxins
undoubtedly remain to be described. Additionally, most toxins
are actually composed of families of closely related
compounds. Slightly different forms of a toxin can exhibit very
different levels of toxicity, or may be characterized differently
by some detection methods and analytical approaches
(Garthwaite et al., 2001; Lefebvre et al., 2008). Such complexity
and variability can sometimes yield vague or contradictory
conclusions regarding the exact source of toxicity in a natural
sample (Bates et al., 1978; Paz et al., 2004, 2007).
Finally, characterization of HAB events is complicated by
inherent difficulties associated with linking specific toxins
measured in natural water samples to a specific algal species
in a complex, natural phytoplankton assemblage and, as
noted above, the presence of toxic species in a water sample
does not necessarily indicate the presence of toxins. Despite
these shortcomings, there is considerable knowledge of many
of the major algal toxins and their producers in U.S. coastal
waters that constitute the most important potential concerns
for desalination activities because they are the most likely to
be encountered in ocean water intakes.
2.1. Domoic acid
2.1.1. Toxin description and activity
Domoic acid (Fig. 2; Table 3) is an amino acid derivative
belonging to the kainoid class of compounds containing three
carboxyl groups and one secondary amino group (Wright
et al., 1990; Jeffery et al., 2004). All four groups are charged at
neutral pH, and the carboxyl groups become successively
protonated as pH decreases, yielding five possible protonated
forms of domoic acid (Quilliam, 2003; Jeffery et al., 2004). There
water research 44 (2010) 385–416 389
are currently ten known isomers of domoic acid, including the
isodomoic acids A through H and the domoic acid 5’ diestereomer
(Jeffery et al., 2004).
Domoic acid and other members of the kainoid class are
glutamate analogues that interfere with neurochemical
pathways by binding to glutamate receptors of brain neurons
(Wright et al., 1990; Quilliam, 2003). The resulting effect of
these neuroexcitants, or excitotoxins, is a continuous stimulation
of the neurons, which can lead to rupture and/or
eventual formation of lesions (Wright et al., 1990). Depolarized
neurons result in short-term memory loss (Clayden et al.,
2005), which has led to the common name for the illness
related to the consumption of seafood contaminated with
domoic acid: amnesic shellfish poisoning (ASP). Symptoms of
ASP include gastroenteritis (vomiting, diarrhea, abdominal
cramps) that can be experienced in humans within 24 h after
ingestion, and neurological symptoms of confusion, memory
loss, disorientation, seizures, coma and/or cranial nerve
palsies that are typically experienced within 48 h (Perl et al.,
1990; Wright et al., 1990). The number of human illnesses
resulting from domoic acid poisoning has been few (Horner
et al., 1997), likely due to active monitoring of fisheries.
However, cultured blue mussels (Mytilus edilus) contaminated
with domoic acid poisoned 107 people and killed three during
the first major documented ASP outbreak in 1987 on Prince
Edward Island, Canada (Perl et al., 1990).
ASP poses a serious threat to marine wildlife, and the
deaths of thousands of marine mammals and sea birds have
been attributed to domoic acid intoxication (Bates et al., 1989;
Scholin et al., 2000; Gulland et al., 2002; Caron et al., unpublished
data). The first documented poisoning episode of
marine animals related to domoic acid on the U.S. west coast
was attributed to Pseudo-nitzschia australis and occurred in
September 1991 off central California (Table 2; Buck et al.,
1992; Fritz et al., 1992). High concentrations of domoic acid
were also detected in Washington and Oregon in the 1990s
(Wekell et al., 1994; Adams et al., 2000; Trainer et al., 2002), and
a decade later in coastal waters off southern California
(Schnetzer et al., 2007). The frequency and severity of these
toxic events appears to be increasing (Trainer et al., 2007).
2.1.2. Producers
The production of domoic acid and its isomers is confined to
approximately a dozen chain-forming marine pennate diatom
species within the genus Pseudo-nitzschia (Bates and Trainer,
2006), a genus containing species that form long chains of
cells attached at their ends (Fig. 3a and b). The main toxin
producing species that have been documented on the U.S.
west coast include: P. australis, P. delicatissima, P. fraudulenta, P.
multiseries, P. pungens, P. pseudodelicatissima, P. seriata and
P. cuspidata (Tables 1 and 2). These species are distinguished
based on fine morphological features of their silica frustules
(Fig. 3a and b). These distinctions are subtle and require
careful electron microscopical analysis and elaborate taxonomic
training. As a consequence, historical misidentifications
are not unusual and debates regarding some species
descriptions are still unresolved.
It is surprising that the first reports of ASP on the west coast
of the U.S. were not recorded until the 1990s, even though
Pseudo-nitzschia species have been recorded in surveys of

390
phytoplankton species in the Southern California Bight since
1917 (Allen, 1922, 1924, 1928, 1936, 1940, 1941; Reid et al., 1970,
1985; Lange et al., 1994; Fryxell et al., 1997; Thomas et al., 2001).
Given that these species generally comprise a significant
portion of the total diatom assemblage in these waters, it can
be surmised that either toxin production has increased in these
west coast species, or that poisoning events prior to the 1990s
have occurred but have not been attributed to these diatoms.
Historical accounts of ‘unusual animal mortality events’ along
the U.S. west coast tend to support the latter hypothesis.
There have been increasing numbers of toxic events
recorded along the U.S. west coast (Table 2), notably in Puget
Sound (Trainer et al., 2003, 2007), Monterey Bay (Vigilant and
Silver, 2007; R. Kudela, unpubl. data), Santa Barbara Channel
(Trainer et al., 2000; Anderson et al., 2006; Mengelt, 2006), San
Pedro Channel (Busse et al., 2006; Schnetzer et al., 2007),
Newport Beach (Busse et al., 2006) and San Diego (Lange et al.,
1994; Busse et al., 2006). Most recently, toxic blooms of Pseudonitzschia
in the Long Beach-Los Angeles Harbor and San Pedro
Channel have been particularly toxic, with some of the highest
domoic acid concentrations recorded for the U.S. west
coast (Caron et al., unpublished data). The increased incidence
and severity of these toxic episodes off the western U.S. coast
parallels the increase in frequency and intensity of harmful
water research 44 (2010) 385–416
Fig. 2 – Chemical structures of commonly encountered toxins produced by microalgae in U.S. coastal waters.
algal blooms observed globally (Smayda, 1990; Hallegraeff,
1993, 2003; Anderson et al., 2002; Glibert et al., 2005b).
2.2. Saxitoxins
2.2.1. Toxin description and activity
Saxitoxin is a complex guanidine-based alkaloid that exists as
more than 30 identified analogues in nature (Llewellyn, 2006).
It is the most powerful marine toxin currently known and
among the most dangerous poisons on Earth, except for some
venoms and bacterial toxins (Schantz et al., 1957). Due to its
acute toxicity, saxitoxin is currently listed as a chemical
weapon in Schedule 1 of the Chemical Weapons Convention
(Llewellyn, 2006). Saxitoxins display a rigid tricyclic core
(Fig. 2; Table 3) and are very stable in biological and physiological
solutions (Rogers and Rapoport, 1980). This nitrogenrich
molecule and its chemical relatives are polar and have
a positive charge at pH 7.7 (Shimizu et al., 1981). Consequently,
they are soluble in water and alcohols, and insoluble
in organic solvents (Schantz et al., 1957).
Saxitoxins are known to disrupt the flow of ions through
voltage gate sodium channels (Catterall, 1992; Cestele and
Catterall, 2000). It has also been recently discovered that they
have the ability to bind to calcium (Su et al., 2004) and

Table 2 – Distribution and concentrations of marine toxins in plankton of confirmed toxin producers in U.S. west coast waters.
Toxin(s) Location and year Causative species Particulate mgL 1
(nmol L 1 Cellular
) pg cell 1
Dissolved pg mL 1
(nmol L 1 )
Domoic
acid
Washington coast and Juan de
Fuca Eddy, WA (1997, 1998)
P. pseudodelicatissima b.d.–2.7
Pseudo-nitzschia spp.
3.6–8.7
References
b.d.–4.6 Adams et al. (2000),
Trainer
et al. (2001, 2002)
Penn Cove, WA (1997) P. pungens b.d.–0.8 Trainer et al. (1998)
P. multiseries
P. australis
P. pseudodelicatissima
Washington coast, WA (2001) P. australis b.d.–0.03 Marchetti et al. (2004)
Washington coast, WA (2003) Pseudo-nitzschia spp. (0.4–15) 2 10 4 –0.3 a
(b.d.–4.3) a
0.1–94.4 (1–5)
Baugh et al. (2006)
Puget Sound, WA (2005) P. pseudodelicatissima b.d.–14 Trainer et al. (2007)
Pseudo-nitzschia spp.
Central Oregon coast, OR (1998) P. australis 0.5 35 Trainer et al. (2001)
Pt. Año Nuevo,
San Francisco, CA (1998)
Bolinas Bay,
San Francisco, CA (2003)
P. pungens 0.1–0.7 0.3–6.3 Trainer et al. (2000)
P. multiseries
P. australis 0.15–9.4 Howard et al. (2007)
Monterey Bay, CA (1991, 1998) P. australis b.d.–12.3 3–37 Buck et al. (1992),
0.1–6.7 7.2–75
Garrison
et al. (1992), Walz
et al. (1994), Scholin
et al. (2000)
Monterey Bay, CA (1998) P. pseudodelicatissima 0.1–0.4 0.8–1.2 Trainer et al. (2000,
P. multiseries 0.67 6
2001)
Monterey Bay, CA (2000) Pseudo-nitzschia spp. b.d.–24 b.d.–8491 Bargu et al. (2002, 2008)
P. australis
Monterey Bay, CA (2002–2003) Pseudo-nitzschia spp. 24 Vigilant and Silver
(2007)
Morro Bay, CA (1998) P. australis 1.3–7.4 37–78 Trainer et al. (2000,
2001)
San Luis Obispo, CA (2003–2005) P. australis 1.5–7.6 9–38 Mengelt (2006)
P. multiseries
Point Conception, CA (1998) P. australis 2.2–6.3 15–22 Trainer et al. (2000)
Santa Barbara, CA (1998) P. australis 0.5–1.2 0.1–0.9 Trainer et al. (2000)
P. pungens
P. pseudodelicatissima
(continued on next page)
water research 44 (2010) 385–416 391

Table 2 (continued)
Toxin(s) Location and year Causative species Particulate mgL 1
(nmol L 1 )
Cellular
pg cell 1
Dissolved pg mL 1
(nmol L 1 )
References
Santa Barbara Channel, CA (2003) P. australis 0.03–1.7 0.14–2.1 Anderson et al. (2006)
Santa Barbara (Santa Rosa Island
and north San Miguel) (2004)
Southern California Bight,
CA (2003, 2004)
San Diego and Orange
counties, CA (2004)
P. australis 6–12 b.d.–80 Mengelt (2006)
P. multiseries
Pseudo-nitzschia spp. 5.6–12.7 b.d.–117 Schnetzer et al. (2007)
P. australis
P. cuspidata
P. australis b.d.–2.33 Busse et al. (2006)
P. multiseries
Saxitoxins Sequim Bay, WA (2004–2007) Alexandrium spp. 0.02–0.5 150–800 Lefebvre et al. (2008)
Oregon coast, OR (2004) Alexandrium spp. 0.004–0.028 Jester et al.
(unpublished data)
Humboldt Bay, CA (2004) A. catenella 1.6–19 a
San Mateo County coast, CA (2004) A. catenella 2.1–62.6 a
Monterey Bay, CA (2004) A. catenella 0.6–31.3 a
Jester (2008)
Jester (2008)
Jester (2008)
Monterey Bay, CA (2003–2005) A. catenella b.d.–0.962 Jester et al. (2009b)
Morro Bay, CA (2004) A. catenella 1.4–16.6 a
Yessotoxin La Jolla, CA (1993) Lingulodinium polyedrum 0.002–0.02 a
Brevetoxins Indian Inlet, Bald Eagle Creek
and Torque Canal, DE (2000)
b.d.: Below detection limit.
a Toxin concentration from cells in culture.
0–0.05 a
Jester (2008)
Armstrong and Kudela
(2006)
Howard et al. (2008)
Chattonella cf. verruculosa 0.008–

potassium channels (Wang et al., 2003) and to be a weak
inhibitor of neuronal nitric oxide synthase (reviewed in Llewellyn,
2006). These activities directly affect the nervous
system, and the consumption of seafood containing saxitoxin
can result in serious human illness and death, commonly
referred to as paralytic shellfish poisoning (PSP).
Minor symptoms of PSP, such as burning or tingling sensation
of the lips and face, dizziness, headache, salivation, intense
thirst and perspiration, vomiting, diarrhea and stomach
cramps, can be experienced within 30 min after the consumption
of contaminated seafood (Llewellyn, 2006). The consumption
of a lethal dose can result in death within hours due to
muscular paralysis and respiratory difficulty followed by
complete respiratory arrest. PSP outbreaks result in more than
2000 illnesses worldwide each year, with a 5–10% mortality rate
(Hallegraeff, 2003). PSP toxins also have adverse effects on
marine wildlife that can cause mortalities among fish, marine
mammal and seabird populations (Geraci et al., 1989; Montoya
et al., 1996; Shumway et al., 2003). There are presently no records
of unusual animal mortality events along the Californian coast
that are attributable to saxitoxin poisoning (Jester et al., 2009b),
but occurrence of the toxins in species consumed by humans is
sufficient to warrant year-round monitoring.
2.2.2. Producers
Saxitoxins are biosynthesized by dinoflagellates in marine
ecosystems, most notably species within the genus Alexandrium
(Fig. 3c), as well as Gymnodinium catenatum, Pyrodinium
bahamense var. copressum, and by some cyanobacteria in
freshwater ecosystems (Hallegraeff, 2003). Blooms of toxic and
noxious dinoflagellates are often referred to as ‘red tides’
because of the red discoloration of water created by the
accessory pigments of the cells. However, toxic levels of saxitoxins
can be attained at dinoflagellate abundances that do not
significantly discolor the water because of the exceptionally
high potency of saxitoxins (Burkholder et al., 2006). This situation
exists for Alexandrium in that it does not typically reach
‘bloom’ abundances on the U.S. west coast, and constant toxin
monitoring is necessary to identify toxic conditions (Langlois,
2007; IOC HAB Programme, 2008; Jester et al., 2009a).
Alexandrium species and measurable saxitoxin concentrations
are common along the U.S. west coast (Table 1), although
concentrations reported for this toxin typically have not been
as high as noted along the U.S. east coast (Table 2). Thus, few
west coast studies have contributed to our understanding the
dynamics of Alexandrium abundances and saxitoxin production
while ongoing research in the Gulf of Maine represents
the most comprehensive regional study of this dinoflagellate
(Anderson et al., 2005; McGillicuddy et al., 2005). Combined
field observations, laboratory studies and modeling efforts
have led to a scenario for toxic events along the northeastern
coast of the U.S. that involve an interplay between river
runoff, resuspension of dinoflagellate cysts from coastal
sediments, favorable offshore growth conditions, and winds
that generate onshore flow into coastal shellfish areas. A
monitoring study of PSP in Puget Sound (WA) from 1993 to
2007 underscores that the timing and location of PSP
outbreaks and high Alexandrium abundances are highly variable
and not easily predicted from local or large-scale climate
data (Moore et al., 2009). However, the study points out that
water research 44 (2010) 385–416 393
periods of warm air and low stream flow may favor saxitoxin
accumulation in sentinel mussels (Moore et al., 2009).
2.3. Brevetoxins
2.3.1. Toxin description and activity
Brevetoxins are polyether, non-polar compounds that depolarize
cell membranes by opening voltage gate sodium ion
channels and induce enhanced inward flux of ions into cells
(Lin et al., 1981; Baden, 1983, 1989; Purkerson et al., 1999).
Brevetoxins exist as two structural types and multiple analogs
possessing various levels of potency (Baden, 1989; Cembella,
2003; Kirkpatrick et al., 2004). The types differ in their ladderframe
polycyclic ether structural backbones and are designated
type A and type B (Fig. 2). The brevetoxin derivatives
found in the marine environment (PbTx-2, PbTx-3 and PbTx-9;
Table 3) are produced most commonly by dinoflagellate and
raphidophyte algae and are of the structural type B (Baden,
1989; Baden et al., 2005).
Brevetoxins bind to site 5 of the voltage-sensitive sodium
channel in neurons, causing these channels to remain open
and fire repeatedly (Catterall, 1992; Cestele and Catterall,
2000). Brevetoxin poisoning in humans is referred to as
neurotoxic shellfish poisoning (NSP), and includes gastrointestinal
symptoms of nausea, diarrhea and abdominal pain,
neurologic symptoms of paresthesia, and respiratory irritation
and/or failure (Kirkpatrick et al., 2004).
The effects of brevetoxins on human health are well
documented along the western coast of Florida where severe,
nearly annual red tides caused by the dinoflagellate Karenia
brevis release large amounts of brevetoxins into the air when
the fragile cells are broken in breaking waves at the water’s
edge (Kusek et al., 1999; Kirkpatrick et al., 2004). The aerosolized
toxins constitute a significant health risk when they
are inhaled and, as a result, K. brevis blooms are one of the
most intensively studied and best-understood regional HABs.
Red tides caused by K. brevis have been implicated in marine
mammal fatalities, fish kills and human illnesses. Brevetoxins
have not been reported from the U.S. west coast, and therefore
no known human fatalities or health issues have yet been
attributed to brevetoxins from that region.
2.3.2. Producers
Several dinoflagellate species and a few raphidophyte species
produce a suite of brevetoxins (Baden, 1989). K. brevis, the
most notorious brevetoxin producer within the Gulf of
Mexico, has not been observed on the west coast of the U.S.,
but several species of raphidophytes that are potential brevetoxin
producers have been documented (Tables 1 and 2).
Heterosigma akashiwo (Fig. 3e), Chattonella marina (Fig. 3f) and
Fibrocapsa japonica have been isolated and cultured from
coastal waters off southern California. In general there are few
reports of significant blooms of these species on the west
coast, although blooms of raphidophytes in San Francisco Bay
and Delaware Inland Bays have been observed with cell
abundances in excess of 10 8 cells L 1 (Herndon et al., 2003;
Coyne et al., 2005). In part, this lack of information is
a consequence of the fact that raphidophyte species are
notoriously difficult to identify using traditional microscopical
techniques because they preserve poorly (Hallegraeff and

394
water research 44 (2010) 385–416

Hara, 1995; Throndsen, 1997). Recently developed genetic
approaches for the identification and quantification of some
raphidophytes are beginning to provide much-needed tools
for studying the distributions and ecology of these HAB
species (Handy et al., 2006; Demir et al., 2008). Despite the
difficulties of characterizing these blooms, fish kills have been
attributed to raphidophyte blooms on the west coast of the
U.S. although these studies have not quantified brevetoxins
(Hershberger et al., 1997; Hard et al., 2000).
2.4. Diarrhetic shellfish toxins
2.4.1. Toxin description and activity
Toxins that cause diarrhetic shellfish poisoning (DSP) include
okadaic acid, dinophysistoxins and pectenotoxins (Ramsdell
et al., 2005). Okadaic acid is a monocarboxylic acid named for
the marine sponge Halichondria okadai from which it was first
isolated (Tachibana et al., 1981). Okadaic acid can also be
found in natural water samples in polar and non-polar esteric
forms (Prassopoulou et al., 2009). The first dinophysistoxin
described was isolated from the mussel M. edilus and was
found to be a methyl form of okadaic acid (Murata et al., 1982).
Okadaic acid and dinophysistoxins are linear polyethers
(Fig. 2) and the mode of action is the inhibition of protein
phosphatases (Takai et al., 1987; Bialojan and Takai, 1988;
Haystead et al., 1989), enzymes that play a key role in
dephosphorylation in many biological processes including cell
cycle regulation. The pectenotoxins are lipid soluble and differ
structurally from other diarrhetic toxins in that they possess
a lactone ring (Fig. 2), and not considered to be protein phosphatase
inhibitors, but have a high actin-depolarizing action
(Hori et al., 1999). There is speculation that pectenotoxins may
not produce diarrhetic effects (Cembella, 2003).
DSP toxins (Table 3) were named for the human symptoms
resulting from the ingestion of contaminated shellfish,
including inflammation of the intestinal tract, diarrhea,
abdominal cramps, vomiting and nausea beginning 30 min to
a few hours after ingestion (Hallegraeff, 2003). In addition to the
symptoms listed above, okadaic acid is known to be a strong
tumor promoter (Suganuma et al., 1988), although the potential
health implications of this activity due to the ingestion of
contaminated seafood is unknown. There are presently no
documented cases of DSP resulting from okadaic acid, dinophysistoxins
or pectenotoxins on the U.S. west coast, and thus
these toxins are not regularly monitored in the marine environment.
DSP toxins have been detected in mussels and water
samples from California (Sutherland, 2008), so it is possible
that DSP has occurred on the U.S. west coast but has been
attributed to other sources of contamination.
water research 44 (2010) 385–416 395
2.4.2. Producers
Okadaic acid and dinophysistoxins are produced by a few
species of the dinoflagellate genus Prorocentrum (Cembella,
2003) and most commonly by species of the genus Dinophysis.
Dinophysis species present on the western U.S coast include
D. acuminata, D. acuta, D. caudata, D. fortii, D. norvegica, D.
rotundata, and D. tripos (Table 2). D. acuminata and D. fortii have
been documented in Californian waters for many years
(Bigelow and Leslie, 1930). D. acuminata produces okadaic acid
(Yasumoto et al., 1985), D. fortii produces okadaic acid, dinophysistoxins
and pectenotoxins (Yasumoto et al., 1980; Murata
et al., 1982) while D. rotundata and D. tripos produce
dinophysistoxin-1 (Lee et al., 1989).
Dinophysis species are technically not phytoplankton, but
heterotrophic protists that retain chloroplasts acquired from
their prey. Dinophysis acquires its chloroplasts by preying on
ciliates, which in turn prey on cryptophyte algae. This
complex trophic relationship has made the culture of these
species unsuccessful until recently (Park et al., 2006), and
therefore no information on the environmental conditions
that influence toxin production by these species presently
exists. However, genus members are easily distinguished by
light microscopy because they possess a pronounced ‘keel’
and other unique morphological aspects (Fig. 3d). These
species are often encountered in plankton samples. Abundances
of 10 3 cells L 1 are commonly encountered (Nishitani
et al., 2005), and occasionally they may reach abundances in
excess of 10 5 cells L 1 (Carpenter et al., 1995).
2.5. Yessotoxin
2.5.1. Toxin Description
Yessotoxin (Fig. 2; Table 3) is a chiral molecule of high polarity
due to the presence of two sulfate groups. The molecule
consists of fused polyether rings organized into a ladder
shaped skeleton (Yasumoto and Murata, 1993; Wright and
Cembella, 1998), a structure similar to other ladder-like polyether
toxins such as the ciguatera toxin complex (ciguatoxins
and maitotoxin), gambieric acids and brevetoxins (Yasumoto
and Murata, 1993; Wright and Cembella, 1998). There are
nearly 100 analogs of yessotoxin that have been identified to
date (Satake et al., 1997, 1999; Ciminiello et al., 1998, 2000,
2001; Daiguji et al., 1998; Miles et al., 2004, 2005a,b, 2006; Paz
et al., 2006).
The yessotoxin class was named for the species of
scallop, Patinopecten yessoensis, in which it was initially
detected (Murata et al., 1987). Yessotoxin was originally
classified in the DSP-toxin class because it was detected with
other DSP toxins, but it appears that it does not induce
Fig. 3 – Major algal toxin producers occurring along the U.S. west coast. (a) Pseudo-nitzschia australis, a producer of domoic
acid; (b) Scanning electron micrograph of Pseudo-nitzschia australis; (c) Alexandrium catenella, a producer of saxitoxin; (d)
Dinophysis sp., a producer of okadaic acid; (e) Heterosigma akashiwo, a producer of brevetoxins; (f) Chattonella marina,
a producer of brevetoxins; (g) Cochlodinium sp.; (h) Lingulodinium polyedrum, a producer of yessotoxin; (i) Phaeocystis globosa
colony; (j) foam produced by the prymnesiophyte, Phaeocystis accumulating along the shore; (k) Unconcentrated seawater
from King Harbor, City of Redondo Beach, with significant discoloration due to an algal bloom; (l) Prorocentrum sp., the
dominant organism in (k); and (m) higher magnification of Prorocentrum sp. from (l). Scale bars [ 10 mm. Photo (b) courtesy of
Peter Miller, (c) courtesy of Carmelo Tomas, (j) courtesy of Cindi Heslin.

Table 3 – Summary of toxins that can be present in Southern California waters. MW: molecular weight.
Toxin Properties Formula MW Mode of action References
Domoic acid (DA) Hydrosoluble
At pH 7: DA 3
Saxitoxins (STXs) Hydrosoluble
pH 7: Stable
C 15H 21NO 6 311.14 Binds to glutamate receptors in the brain
disrupting normal neurochemical
transmission
C 10H 17N 7O 4 299.3 Bind to site 1 of voltage-sensitive sodium
channels and block sodium conductance;
bind to calcium and potassium channels
Brevetoxins (PbTxs) Liposoluble Bind to site 5 of voltage-sensitive sodium
Brevetoxin 2 (PbTx 2) C 50H 70O 14 895.1
Brevetoxin 3 (PbTx 3) C50H72O15 897.1
Brevetoxin 9 (PbTx 9) C 50H 74O 14 899.1
Diarrhetic shellfish toxins
channels, shifting activation to more
negative membrane potentials and block
channel activation
Okadaic acid (OA)
Dinophysistoxins (DTXs)
C44H68O13 805 Inhibits protein phosphatases, inhibits
dephosphorylation of proteins
Pectenotoxins (PTXs) Liposoluble High actin-depolarizaing action
Yessotoxins (YTXs) Hydrosoluble C55H80O21S2Na2 1187.3 Activation of phosphodiesterase in
the presence of external Ca 2þ ;
Disruption of the E-cadherin–catenin
system in epithelial cells and potentially
disrupting its tumour suppressive functions
Wright et al. (1990), Quilliam (2003)
Wong et al. (1971), Wang et al. (2003), Su et al. (2004),
Catterall (1992), Cestele and Catterall (2000)
Lin et al. (1981), Baden (1983, 1989),
Purkerson et al. (1999)
Tachibana et al. (1981), Murata et al. (1982),
Yasumoto et al. (1985), Takai et al. (1987),
Bialojan and Takai (1988), Haystead et al. (1989),
Hori et al. (1999)
Murata et al. (1987), Takahashi et al. (1996),
Alfonso et al. (2003), Ronzitti et al. (2004)
396
water research 44 (2010) 385–416

diarrhetic effects (Ogino et al., 1997). Accordingly, the regulation
of the European Commission on marine biotoxins now
considers yessotoxins separately from DSP toxins (European
Commission, 2002). The great number of yessotoxin analogs
complicates toxicity studies, and may explain the sometimes
contradictory reports of its mode of action. Studies have
shown that lysosomes, the immune system and the thymus
(with tumorigenic implications) are the biological targets of
yessotoxin (Franchini et al., 2004; Malagoli et al., 2006), while
other reports have indicated cardiotoxic effects (Terao et al.,
1990; Ogino et al., 1997; Aune et al., 2002). The cardiotoxicity
of yessotoxin might be attributed to phosphodiesterase
activation in the presence of external calcium ions (Alfonso
et al., 2003). Unlike the other marine toxins mentioned
above, there have been no reported human health issues or
marine mammal deaths associated with yessotoxins.
2.5.2. Producers
There are three known yessotoxin-producing dinoflagellates,
Protoceratium reticulatum, Lingulodinium polyedrum, and
Gonyaulax spinifera, and they have all been observed in coastal
waters off the western U.S. (Table 1). According to phylogenetic
analyses of available rRNA gene sequences, the capacity
for yessotoxin production appears to be restricted to the order
Gonyaulacales (Howard et al., 2009). However, toxin production
among strains within each species appears to be highly
variable.
Yessotoxin has been detected in L. polyedrum isolates
cultured from around the globe (Tubaro et al., 1998; Draisci
et al., 1999a; Strom et al., 2003; Paz et al., 2004), including
isolates from Californian coastal waters (Armstrong and
Kudela, 2006). The reported cellular concentrations in the
latter cells ranged from below detection to 1.5 pg cell 1 , indicating
that L. polyedrum is significantly less toxic than
P. reticulatum or G. spinifera. Yessotoxin has been recorded in
blue mussels at low concentrations along the U.S. west coast
(Table 2) during red tides caused by L. polyedrum, as well as
during non-bloom conditions, but yessotoxin production by
this dinoflagellate appears to be less than toxin levels
produced by isolates from other geographical regions (see
Table 2 in Howard et al., 2008). Expansive and dense blooms of
L. polyedrum (Fig. 3 h) have been reported in California since
1901, but there have only been anecdotal reports of health
problems associated with the red tides caused by L. polyedrum
(Kudela and Cochlan, 2000).
Yessotoxin production by isolates of P. reticulatum has
been confirmed (Satake et al., 1997; Boni et al., 2002; Miles
et al., 2002; Riobo et al., 2002; Stobo et al., 2003; Samdal et al.,
2004; Eiki et al., 2005), but concentrations ranging from
below detection to 79 pg cell 1 have been reported for
isolates from Washington, California and Florida (Paz et al.,
2004, 2007). Isolates of G. spinifera appear to be the most
prolific yessotoxin producers. Concentrations in New Zealand
isolates ranged from below detection to 200 pg cell 1
(Rhodes et al., 2006), more than 20-fold higher than the
per-cell toxicity of P. reticulatum and 600-fold higher than
L. polyedrum. G. spinifera does not generally bloom in high
densities on the U.S. west coast, but it has been frequently
observed at low abundances (Howard et al., 2008; M. Silver,
pers. comm.) and has reached bloom concentrations in
water research 44 (2010) 385–416 397
Tomales Bay, north of San Francisco (G. Langlois pers.
comm.). Yessotoxins are monitored in New Zealand, Europe
and Japan but they are not routinely measured on the U.S.
west coast. Howard et al. (2008) was the first study to confirm
yessotoxins in California and Washington coastal waters,
albeit at very low concentrations.
2.6. Toxin detection and quantification
A wide variety of methodologies and technologies exist to
detect, characterize and quantify the major toxins produced
by microalgae. These approaches can be broadly divided into
those used to characterize biological activity (toxicity assays)
and those used to identify specific chemical structure(s)
(immunological, various analytical techniques). Because of
the highly variable approaches employed, and in most
cases the highly diverse set of compounds comprising a toxin
class, the methods provide somewhat different estimates of
absolute toxin concentrations or presumed toxicity. It is also
noteworthy that the majority of the protocols used to measure
algal toxins have focused on the analysis of tissue samples
that may be a source of human contamination (e.g. shellfish or
finfish tissue) or plankton material filtered from seawater
samples. Relatively few studies have examined the concentrations
of toxins dissolved in seawater (Table 2). For this
reason, limits of detection for toxins in seawater are poorly
known for most approaches. However, based on analytical
approaches presently available, a practical limit of detection
for a few of the major concerns (domoic acid and saxitoxins) is
in the range of 0.1 mg per liter of seawater for immunological
approaches, but a detection limit of approximately 0.01 mg per
liter for dissolved domoic acid in seawater using highperformance
liquid chromatography has been reported
(Pocklington et al., 1990). Detection limits for toxins in
particulate material in the water can be significantly lower
because particles can be concentrated by filtration prior to
extraction. Knowledge of the concentrations of dissolved
toxins would be preferable from the perspective of reverse
osmosis desalination operations because dissolved toxins
(rather than cell-bound toxins) would most likely impact
these membranes.
Domoic acid has been detected and quantified in seawater,
plankton, shellfish extract and homogenate, as well as sea bird
and mammalian body fluid (e.g. blood, urine, amniotic
fluid, cerebral spinal fluid). Analytical approaches for these
measurements include commercially available immunological
techniques (enzyme-linked immunosorbent assay; ELISA)
(Garthwaite et al., 1998, 2001), high pressure liquid chromatography
(HPLC) with ultraviolet (UV) diode array detection
(DAD) (Quilliam et al., 1989; Quilliam, 2003), receptor binding
assay (RBA) (Van Dolah et al., 1994), mouse bioassay (MBA) or
liquid chromatography–mass spectrometry (LC–MS). The
results obtained by these various approaches are not yet
completely compatible or comparable, and therefore comparisons
across studies using different analytical methods can
be problematic. In general, the choice of an approach is
a compromise between cost of analysis (or access to costly
equipment), sample throughput, sensitivity and analytical goal
(e.g. thorough chemical characterization versus overall

398
toxicity, or human health risk versus scientific understanding
of bloom dynamics).
Saxitoxins can be rapidly detected and quantified with
commercial ELISA kits, but the high specificity of these tests
precludes the recognition of certain members of the saxitoxin
family, especially the neo-saxitoxin (Garthwaite et al., 2001).
Saxitoxin detection and quantification is often accomplished
by HPLC, RBA, LC–MS, and MBA. Regulatory programs for
seafood consumption are still based on ‘lethal mouse dosage’.
Similarly, detection and quantification of brevetoxin and its
derivatives in seawater, shellfish, and mammalian body fluids
can be accomplished using commercially available ELISA kits
(Naar et al., 2002), by HPLC, RBA (Van Dolah et al., 1994), LC–MS
and MBA. A comparative study also quantified brevetoxins by
radioimmunoassay (RIA) and a neuroblastoma (N2A) cytotoxicity
assay (Twiner et al., 2007).
Quantification of the DSP toxin suite can be underestimated
by ELISA because commercial ELISA assays are
usually optimized to detect okadaic acid and not the dinophysistoxins
(Garthwaite et al., 2001). HPLC with fluorimetric
detection (HPLC-FLD) has been commonly used to detect and
quantify okadaic acid, its polar and non-polar esters, as well
as dinophysistoxins (Lee et al., 1987). The MBA method has
also been routinely used for the detection of DSP toxins.
The detection and quantification of yessotoxin is problematic
because of the extensive suite of derivatives that may
exist. HPLC-FLD analysis (Yasumoto and Takizawa, 1997),
MBA, LC-MS (Draisci et al., 1999b; Paz et al., 2006, 2008) and
ELISA (Samdal et al., 2004, 2005) have been employed.
2.7. Other potentially toxic, noxious
and nuisance organisms
Reports of newly occurring HAB species, or recognition of
extant issues that have gone previously undocumented, are
water research 44 (2010) 385–416
increasing our awareness of other potentially harmful bloomforming
algae along the U.S. west coast. For example, blooms
of an emerging potentially toxic organism, Cochlodinium sp.
(Fig. 3g) off central California have recently been reported
(Curtiss et al., 2008; Iwataki et al., 2008; Kudela et al., 2008b).
This species is difficult to identify using light microscopy, and
therefore researchers have begun to use gene sequences to
provide accurate identification (Iwataki et al., 2007, 2008;
Matsuoka et al., 2008). While this organism has only recently
reached sufficient abundances to discolor Californian waters,
there has already been one reported abalone loss in central
California that appears to be associated with a bloom of
Cochlodinium (R. Kudela pers. comm).
Species of the dinoflagellate Prorocentrum (Fig. 3k–m) occasionally
bloom in Californian coastal waters where they can
attain very high abundances periodically, causing discolorations
of the water and nuisance accumulations of algae
(Holmes et al., 1967; Shipe et al., 2008). The Prorocentrum
species known to occur in Californian waters have not
yet demonstrated DSP toxin production, but species from
elsewhere in the world are known to produce theses toxins
(Table 1). Similarly, massive blooms of the dinoflagellate
Akashiwo sanguinea are common in coastal waters of southern
and central California and have recently been the cause of
seabird mortality due to surfactant-like proteins (Jessup et al.,
2009). Although they may not be overtly toxic, these blooms
can cause animal mortalities, deplete oxygen, and result in an
increased organic and biomass loading to a seawater desalination
facility.
The prymnesiophyte Phaeocystis globosa infrequently
attains high abundances off the Californian coast (Armonies,
1989). This species produces single cells that are

Phaeocystis are consumed by many zooplankton species but
the colonies are typically a poor food source. Selective feeding
on single cells appears to favor colony formation and the
accumulation of colonies in the water column (Netjstgaard
et al., 2007). When released via colony destruction or algal life
cycle events, the colony matrix material is easily worked into
a ‘sea foam’ that can form layers many centimeters (even
meters) thick at the ocean surface or along the coastline over
fairly extensive regions (Fig. 3j).
3. Spatial and temporal patterns
of harmful algae
A fundamental aspect of the biology of harmful algal blooms,
and of vital importance for desalination operations, is the
tendency for rapid and dramatic changes in the spatial and
temporal distributions of these species. These changes occur
rapidly across a wide range of scales, and pose challenges for
documenting and predicting these distributions. Numerous
approaches and instruments have been developed to characterize
the dynamics of phytoplankton communities. These
approaches presently have significant limitations on their
abilities to identify species composition of a bloom, but they
provide crucial information on the emergence and longevity
of bloom events as well as their vertical and geographical
extent. This information can help seawater desalination
facilities adjust operations to ensure reliable production for
the duration of the bloom.
3.1. Temporal variability
Significant temporal variations in the abundances of phytoplankton
take place on time scales ranging from hours to
decades. Short-term temporal variability (hours to a few weeks)
can be a consequence of rapid population growth or consumption
by herbivores, sinking of senescent populations, diel
vertical migration, tidal movement, and aggregation or
dispersal by physical processes such as water mass convergences
or divergences.
Diel vertical migration of several dinoflagellates has been
attributed to geotaxis, phototaxis and nutrient status (Eppley
et al., 1968; Blasco, 1978; Cullen and Horrigan, 1981; Levandowsky
and Kaneta, 1987). Classic responses involve nighttime
sinking out of surface waters to deeper water where
nutrient concentrations are greater, and rising into surface
waters for photosynthesis during daytime. Shifts in nutrient
cell quotas that accompany these migrations may have
significant implications for toxin production because cell
toxicity can be related to nutrient status of the cells (Anderson
et al., 1990; Flynn et al., 1994; John and Flynn, 2000; Flynn,
2002).
Diel-to-weekly variations in phytoplankton abundance can
be characterized using self-contained or wirelessly networked
sensor packages. Data collected in King Harbor of the City of
Redondo Beach, CA (Fig. 4) demonstrate the efficacy of these
instruments for providing high temporal resolution of chlorophyll
a fluorescence (which approximates phytoplankton
biomass) and pertinent environmental factors (e.g. dissolved
oxygen and temperature) that provide insight into the factors
water research 44 (2010) 385–416 399
controlling the pattern. These data reveal a 4-fold variation in
phytoplankton standing stock over a two-day period. In
addition, changes in water quality criteria were easily and
rapidly identified (e.g. decrease in dissolved oxygen concentration
near noontime on September 13). Daily variations in
the latter parameter can be extreme at night during algal
blooms. High resolution, short-term monitoring approaches
allow rapid detection of sentinel parameters, and in turn
provide information for the development of predictive models
of HAB events.
Chlorophyll a fluorescence sensors provide valuable
information on the short-term temporal patterns of total algal
biomass, but these instruments cannot identify noxious algal
species within a phytoplankton assemblage. More sophisticated
instruments now coming online offer that possibility.
For example, the Environmental Sample Processor (http://
www.mbari.org/ESP) isanin situ instrument capable of performing
real-time identifications of HAB and other microbial
species, as well as toxin analyses such as domoic acid using
on-board molecular analyses (Greenfield et al., 2006). Similarly,
handheld devices now exist for the detection of some
HAB species and toxins in the field (Casper et al., 2007). These
rapid and highly specific analyses are becoming valuable tools
for quick determinations of toxin presence resulting from
algal blooms. These more costly instruments can be used to
improve the information available on bloom composition
once the cheaper and more readily available sensors identify
an emerging bloom event.
Seasonal variability of HAB species and their toxins along
the Californian coast can be gleaned from the records of the
Marine Biotoxin Monitoring Program (MBMP) of the California
Department of Public Health (CDPH), a program started after
a major domoic acid outbreak in the fall of 1991. At present,
the annual effort involves the analysis of approximately 300
shellfish samples for domoic acid and >1000 samples for PSP
toxins from all fifteen Californian coastal counties (CDPH,
2007). Shellfish toxin information provides a reasonable
representation of toxins in the upper water column over
seasonal and annual scales such as demonstrated in studies
on toxic dinoflagellates (Montojo et al., 2006; Jester et al.,
2009a; Moore et al., 2009). Detailed information on the
sampling effort is provided in the MBMP annual reports
(Langlois, 2007).
Distributions of domoic acid and PSP toxins along the
Californian coast during the period 2002–2007 based on the
MBMP data reveal both seasonal and geographical trends (Figs.
5 and 6). Monthly averages for the 6-year period (histograms)
and maximal concentrations (triangles and lines) showed
detectable concentrations of PSP toxins and domoic acid in
shellfish in nearly all months for all Californian coastal
counties (Fig. 5). Domoic acid concentrations showed
pronounced seasonality, with very high peaks during spring
and much smaller peaks during fall. This temporal trend is
concordant with previous observations along the southern
Californian coast (Lange et al., 1994; Walz et al., 1994; Schnetzer
et al., 2007; Shipe et al., 2008). Fall domoic acid peaks correspond
to minor blooms of toxic Pseudo-nitzschia occasionally
noted on the west coast of the U.S. (Bolin and Abbott, 1960;
Buck et al., 1992; Walz et al., 1994; Trainer et al., 2002;
Schnetzer et al., 2007).

400
Domoic acid
(µg/100g)
PSP toxins
(µg/100g)
70000
60000
50000
40000
30000
20000
15000
10000
5000
Seasonal peaks in domoic acid along the U.S. west coast
differ along a latitudinal gradient. Highest domoic acid
concentrations in Washington have been observed during the
fall (Trainer, 2002; Office of Shellfish and Water Protection, 2008;
Moore et al., 2009) compared to spring blooms that are common
in southern California (Schnetzer et al., 2007). This north-south
trend in seasonality is also evident on a smaller latitudinal scale.
Blooms along the Californian coast have been more frequently
observed later in the year in northern counties (Humboldt
versus Santa Barbara counties in Fig. 6). The time lag between
bloom periods observed from south to north may be related to
the timing of the California Current System (CCS) upwelling
maximum, which brings nutrients into surface waters and
promotes phytoplankton growth. The CCS upwelling occurs in
early spring in southern California, in June off Washington,
0
1600
1400
1200
1000
800
600
400
300
200
100
0
water research 44 (2010) 385–416
Mean of monthly averages
Absolute maximum
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Fig. 5 – Seasonal variability of domoic acid (upper panel) and PSP toxin (lower panel) concentrations along the Californian
coast. Monthly averages of data collected from 2002 and 2007 summarized for fifteen Californian coastal counties are shown
as histograms. Also shown are the maximal values recorded during each month over the entire study period (triangles and
solid lines). Toxin concentrations were derived from shellfish tissue. Data from CDPH (Langlois, 2007).
and throughout summer in northern California and Oregon
(Reid et al., 1956; Landry, 1989).
Monthly averages as well as maximal PSP toxin concentrations
showed less pronounced seasonal variability than
domoic acid during the period 2002–2007, although the highest
concentrations were recorded from July to September (Fig. 5).
This seasonal pattern of maximal PSP is in agreement with the
last 25 years of monitoring results (Langlois, 2007). It is also
consistent with the 1927–1989 observations on the Californian
coast indicating that most significant concentrations of the
toxin take place between May and October (Price et al., 1991).
The highest PSP toxin concentrations in shellfish have also
been observed in the summer and the fall periods off the
Washington and Oregon coasts (Trainer et al., 2002; Determan,
2003).

North
South
Domoic acid
(µg/100 g)
Domoic acid
(µg/100 g)
14000
12000
10000
8000
6000
4000
2000
0
14000
12000
10000
8000
6000
4000
2000
0
Domoic acid
Santa Barbara County
Diel to seasonal temporal patterns in ASP and PSP
outbreaks off the U.S. west coast are augmented by large
interannual variability in the intensity and frequency of
HABs. Interannual variations in HABs presumably are
related, at least in part, to changes in atmospheric and
hydrographic features modulated by the 3–7 year cycles of El
Niño-Southern Oscillations (ENSOs) (Price et al., 1991; Horner
et al., 1997), but the exact relationship between HABs and
these climatic events is not clear because there are still
relatively few observations spanning these temporal
regimes. Warming off California during El Niño episodes
reduces seasonal upwelling, enhances physical stratification
in the CCS and lowers the nutricline in the water column
(the depth at which nutrient concentrations increase
rapidly). It is clear that these changes in water stability and
nutrient availability have significant impacts on plankton
productivity and community structure, but the specific
responses of the phytoplankton communities vis-à-vis HAB
events are not yet predictable (Barber and Chavez, 1983).
ASP outbreaks occurred during El Niño episodes of 1991 and
1997–98 along the coasts of Oregon, Washington and California
(Table 2; Moore et al., 2008), but the specific factors contributing
to toxic Pseudo-nitzschia blooms during these events
could not be clearly identified (Horner and Postel, 1993; Trainer
et al., 2000). Interannual variability in ASP and PSP concentrations
in coastal waters along the Californian coast was
evident and substantial in the MBMP dataset during the period
2002–2007 (Fig. 6). ASP and PSP outbreaks were frequent, but
water research 44 (2010) 385–416 401
Humboldt County
**** * ** ** ** **
April May June
** * * **
April May June
PSP toxins
(µg/100 g)
PSP toxins
(µg/100 g)
140
120
100
80
60
40
20
0
140
120
100
80
60
40
20
0
PSP toxins
Marin County
July August September
San Luis Obispo County
2002
2003
2004
2005
2006
2007
* not detected
** * * * * * * ** *
July August September
Fig. 6 – Interannual variation in toxin concentration in four Californian coastal counties from 2002 to 2007 during three
months exhibiting high concentrations of domoic acid (April, May, June) and saxitoxin (July, August, September). Humboldt
and Marin counties are located north of Santa Barbara and San Luis Obispo counties. Toxin concentrations were measured
from shellfish tissues. Data from CDPH (Langlois, 2007).
they varied in intensity and the timing of peak concentrations
between the years within a single geographical location, and
between southern and northern Californian counties in the
same year and season. Notably, the magnitude of this variability
is on the same order of magnitude as the variability
observed on short-term (daily) or seasonal time scales.
Little is known regarding the longer time scale fluctuations
in HABs along the U.S. west coast. Multi-decadal fluctuations
in ocean temperature are known to provoke shifts in the
biological regime (Chavez et al., 2003), and it is anticipated
that climatic shifts would affect the timing, intensity or
frequency of phytoplankton blooms. Long-term regime shifts
may affect the occurrence of blooms of L. polyedrum,
a producer of yessotoxin, along the coast of southern
California (Tables 1 and 2). The Pacific Ocean was cooler in the
years preceding 1976, and red tides dominated by L. polyedrum
commonly developed along the Southern California Bight
during fall (Gregorio and Pieper, 2000). During a recent warm
regime (1976-mid 1990s), red tides occurred during winter and
spring and persisted until summer in the region of the Los
Angeles River mouth (Gregorio and Pieper, 2000). Recent
massive blooms of L. polyedrum during fall may indicate
a return to the pre-1976 conditions (Moorthi et al., 2006).
The generality surmised from data depicting short- to longterm
temporal variability in phytotoxin dynamics is that
variability can be high at all scales. Given our present state of
understanding regarding the specific combination of forcing
factors that give rise to this high variability, it is difficult to

402
Fig. 7 – Two-dimensional presentations of chemical and physical data collected by an autonomous vehicle (Webb Slocum
glider, Teledyne Webb Research, East Falmouth, MA) along a nearshore–offshore transect at Newport Beach, CA shown in (f)
(indicated on map by the blue line). Contour plots are shown for (a) temperature, (b) chlorophyll a fluorescence, (c) salinity,
(d) backscattered light and (e) water density. The Webb Slocum glider is an autonomous vehicle commonly employed in
coastal ecosystems. This buoyancy-driven underwater vehicle generates horizontal motion by ascending and descending
with pitched wings (Schofield et al., 2007). A rudder directs heading while buoyancy is controlled by pumping seawater into
and out of the nose of the vehicle. This long-lived, low-power glider achieves horizontal velocities of approximately 25–
30 cm s L1 with vertical velocities of 10–15 cm s L1 .
accurately predict the timing and magnitude of toxic blooms.
For these reasons, monitoring at multiple temporal scales is
necessary to adequately characterize plankton dynamics.
3.2. Spatial variability
Spatial variability of HABs is considerable at multiple scales,
analogous (and strongly related) to the temporal variability
described above. Blooms can be highly localized (10s of
meters) or expansive (100s of kilometers), and distributions
vertically within the water column are heterogeneous over
scales of centimeters to meters. The geographical extent and
heterogeneous nature of the U.S. west coast results from
differences in local hydrography that are manifested in smalland
large-scale differences in spatial patterns of toxic blooms.
Regional-scale variations in HAB distributions are illustrated
by MBMP data during 2002–2007 for ASP and PSP
concentrations observed in counties along northern and
central California (Fig. 6). ASP events (frequency and level of
water research 44 (2010) 385–416
toxicity) were generally lower in northerly Humboldt County
during this period relative to Santa Barbara County nearly
1000 km to the south. More recently, high domoic acid
concentrations have been observed within the Southern California
Bight, presumably indicating a continuing southward
movement of the Pseudo-nitzschia spring blooms (Langlois,
2007; Schnetzer et al., 2007). On the other hand, Marin County
(north of San Francisco) exhibited higher monthly PSP values
during most months than San Luis Obispo County located
nearly 500 km to the south. This general latitudinal trend in
PSP events is consistent with findings that the three southernmost
counties (Los Angeles, Orange and San Diego)
generally experience low concentrations of PSP relative to
northern California (Price et al., 1991; Langlois, 2007).
Regional-scale and geographical differences in ASP and PSP
events have also been reported along the coastline of Oregon
(Trainer et al., 2002).
Small-scale spatial variability (horizontally and vertically)
can also be dramatic. HAB events can be highly restricted

geographically (e.g. relegated to a protected embayment).
Even within spatially extensive blooms, phytoplankton
biomass is often highly discontinuous over very small spatial
scales because of differences in local circulation, and wind or
wave forcing. Considerable spatial variability in the abundance
of P. australis within Monterey Bay has been observed,
a phenomenon that has been attributed to advective forces
(Buck et al., 1992). Discontinuities within a water column, such
as thermoclines, haloclines, nutriclines and light absorption
often leads to the establishment of subsurface microlayers
where phytoplankton biomass can be many-fold elevated
relative to algal standing stocks only centimeters above or
below (Dekshenieks et al., 2001; Rines et al., 2002).
Characterization of phytoplankton spatial distributions
include approaches ranging from shore-based sampling, to
the use of oceanographic ships, to remote sensing of largescale
patterns using satellite imagery. Vertical profiling of
phytoplankton assemblages can be accomplished using
over-the-side, ship-based instrument packages and more
recently autonomous vehicles equipped with a variety of
sensor packages. The use of autonomous vehicles to provide
synoptic measurements of phytoplankton biomass from
chlorophyll a fluorescence and pertinent chemical/physical
parameters is state-of-the-art for obtaining two-dimensional
cross-sections or three-dimensional patterns in the water
column (Fig. 7).
Autonomous vehicles provide time- and depth-stamped
measurements of a variety of parameters that can be optimized
for a specific mission. Such an instrument deployed off Newport
Beach, CA (blue line in Fig. 7f) during May–June 2008 yielded
detailed patterns of chemical/biological parameters that
provided information on the extent and vertical distribution of
phytoplankton biomass (Fig. 7a–e). Evidence of upwelling in this
118°10’W
Particulate domoic acid
(µg L-1)
118°00’W
0.01-0.50
0.51-1.00
1.01-1.50
1.51-2.00
2.01-2.50
2.51-3.00
3.01-3.50
3.51-4.00
4.01-4.50
4.51-5.00
33°45’N
33°40’N
33°35’N
Fig. 8 – Spatial variability in domoic acid concentrations
contained in the total particulate material (algal cells, other
microbes and detritus) in surface waters within the San
Pedro Bay area including the Long Beach-Los Angeles
harbor area. Data were collected at 20 sampling stations
(locations indicated by filled circles).
water research 44 (2010) 385–416 403
spring deployment was apparent from the upward-pointing
isotherms (temperature) and isopycnals (density) on the
shoreward end of the transect (left sides of Fig. 7a, e). This was
particularly evident in the temperature plot 5–10 km from shore
between the surface and 20 m (Fig. 7a, red circle). Nutrients (not
shown) were generally depleted in surface waters, and
increased with decreasing temperature. This physico-chemical
structure is concordant with a significant subsurface maximum
in chlorophyll a concentration, indicative of a response of the
phytoplankton assemblage to elevated nutrient concentrations
at this depth (red circle on left in Fig. 7b), as previously observed
(Jones et al., 2002). The cross-sectional picture provided by the
autonomous vehicle indicated that the phytoplankton
community had a patchy structure on the scale of meters
(vertically) and kilometers (horizontally). Two major horizontal
patches were observed at 6–12 km and at 15–21 km from shore
(red circles in Fig. 7b).
Measurements in addition to chlorophyll a fluorescence can
add information on the observed general patterns such as
particle size distributions derived from the optical backscatter
spectrum. The backscatter profile obtained at a wavelength of
532 nm indicated a patch of particles that were not of algal
nature (Fig. 7d, red circle). The wavelength-dependent slope of
the backscatter, which is dependent on the particle size distribution,
can also be mapped to indicate the size-class of phytoplankton
particles that dominate the chlorophyll a maxima.
This information is particularly important for a seawater
desalination facility, where the incoming particle size distribution
is known to impact the source water filterability.
Autonomous vehicles allow nearly synoptic measurements
of the spatial distribution of phytoplankton and ancillary
parameters. Many of these instruments operate for
significant periods of time (weeks) and thereby supply
temporal as well as spatial coverage. Knowledge of the
temporal evolution and spatial organization of coastal marine
systems enables a better understanding of the linkages
between physical processes and the biological responses that
contribute to the formation of algal blooms. Moreover, data
from these instruments can be telemetered to the laboratory
in near-real time and used to direct costly efforts such as
shipboard sampling, or plan operations for land-based activities
and measurements.
Broad-scale, horizontal distributions can also be acquired
via shipboard sampling programs (Fig. 8). Shipboard work is
time and labor intensive, but onboard sample processing
enables more sophisticated analyses than autonomous vehicles
are presently capable of providing. Moreover, ships and
other manned platforms permit time series studies at a single
study site. Shipboard sampling conducted in April 2008 in the
Long Beach-Los Angeles harbor area and the adjacent San
Pedro Channel demonstrated considerable spatial variability
in the distribution of HAB species and their toxins within this
relatively small area (approximately 500 km 2 ). Results indicated
a patchy distribution of domoic acid in particulate
material (i.e. within phytoplankton cells) collected near the
surface with highest concentrations in the vicinity of the
harbor breakwater and at several offshore locations (Fig. 8).
Intermediate regions exhibited toxin concentrations that
were more than an order of magnitude less than these
maxima.

404
3.3. Environmental driving factors
Characterizing the factors that lead to the stimulation of
harmful algae and the production of toxins by these algae has
been an area of very active research for decades. These studies
involve field observations to document the spatiotemporal
extent of blooms and toxin concentrations in plankton and
marine life, and laboratory experiments aimed at understanding
the key environmental factors leading to HAB events
and toxin production. The overall results gleaned from many
years of work group into three basic categories: (1) factors and
conditions leading to phytoplankton blooms in general, (2)
factors leading specifically to the growth of HAB species, and
(3) factors leading to toxin production.
Numerous factors have been implicated as contributors to
the observed global expansion of HABs (Smayda, 1990; Hallegraeff,
1993, 2003; Anderson et al., 2002; Glibert et al., 2005b).
Phytoplankton blooms occur naturally as a consequence of
the vertical mixing of deep, nutrient-rich waters into lighted
surface waters. This process occurs seasonally in temperate
environments due to winter storm events, and due to coastal
upwelling events caused by appropriate regional wind conditions.
There is no a priori reason why these ‘natural’ sources
of nutrients cannot lead to HAB events, but the global increase
in frequency and severity of HABs implies that human activities
may be an underlying reason for this escalation.
Eutrophication of coastal ecosystems is a growing global
concern that has clear consequences for blooms of nearshore
algal populations (Anderson et al., 2008; Heisler et al., 2008;
Howarth, 2008). Nutrient enrichment has been implicated in
harmful blooms occurring in some protected bays, but the
linkage between nutrient discharges mediated by human
activities and many HAB events is still unconfirmed. For
example, field studies have shown that coastal upwelling of
nitrate-rich waters can be a driving factor leading to toxigenic
Pseudo-nitzschia blooms along the U.S. west coast (Horner
et al., 2000; Scholin et al., 2000; Anderson et al., 2006) but the
specific role of river/coastal runoff in domoic acid production
is unclear (Scholin et al., 2000; Schnetzer et al., 2007). The
importance of nutrient discharge into coastal waters is, of
course, dependent on the amount of nutrients available for
phytoplankton growth from natural sources but the latter
term is poorly defined in most situations. Constructing
nutrient budgets for coastal ecosystems is an area ripe for
future work. In the meantime, it has been speculated that
anthropogenic nutrient sources, such as elevated nutrient
concentrations in river discharge, coastal runoff from agricultural
land, and sewage discharge may significantly
increase the total amounts of nutrients available for the
growth of coastal phytoplankton (Scholin et al., 2000; Glibert
et al., 2005a,b, 2006; Howard et al., 2007; Kudela et al., 2008a).
There now exists a basic understanding of the general
conditions that favor the growth of phytoplankton per se
(Allen et al., 2008). Despite this basic understanding, there is
still only limited information on the specific conditions that
selectively stimulate the growth of harmful algal bloomforming
species of phytoplankton. As a result, mathematical
models that attempt to predict HAB events tend to be more
correlative than deterministic (i.e., they identify the
water research 44 (2010) 385–416
conditions that may promote a HAB, rather than the conditions
that will promote a bloom). One generality is that rarely
can one identify a ‘silver bullet’, a single parameter or set of
circumstances that provide an accurate prediction of the
occurrence of a particular HAB species. The environmental
circumstances leading to the dominance of a HAB population
over all other species of algae in a given locale are composed of
a complex set of physical, chemical and biological conditions
with poorly known variances, and these conditions appear to
be species-specific.
Biological factors contributing to the success or demise of
individual HAB taxa include allelopathy among competing
phytoplankton, mixotrophy by HAB species, and the deterrence
of potential consumers via the production of noxious or
toxic compounds (Strom et al., 2003; Burkholder et al., 2008;
Buskey, 2008; Flynn, 2008; Smayda, 2008). These biological
interactions presuppose a period of stable environmental
conditions in order that the scenarios of allelopathy, grazer
deterrence or phagotrophic activity of HAB species can play
themselves out. This requirement may explain why the
formation of a stable water mass appears to play a role in the
development of some HAB events (Scholin et al., 2000).
Models predicting the population growth of potentially
toxic algae are necessary for understanding bloom dynamics,
but these models must also integrate information on toxin
induction. Many toxins do not appear to be constitutively
produced by algae, but are induced by a variety of specific
environmental conditions that are not completely understood.
Silica, phosphorus, nitrogen and trace metal limitations,
and nutrient or elemental ratios (in addition to or
instead of absolute concentrations) have all been implicated
in toxin induction (Pan et al., 1996a, 1998; Rue and Bruland,
2001; Fehling et al., 2004; Wells et al., 2005; Granéli and Flynn,
2006; Schnetzer et al., 2007). Again, species-specific differences
(and perhaps strain-specific differences) may exist in
the factors promoting toxin production. Physical aspects such
as temperature and light intensity may stimulate toxin
production by some harmful algae (Ono et al., 2000).
Relative fluorescence
7
45
6
Relative fluorescence
TMP
40
35
5
30
4
25
3
20
2
15
10
1
5
0
0
0 5 10 15 20 25 30 35 40 45
Days
Transmembrane pressure
(TMP, psi)
Fig. 9 – Correlation between chlorophyll a fluorescence
(open squares, in relative fluorescence units) and
transmembrane pressure (filled circles) in a pilot-scale
reverse osmosis desalination system. Increased loading of
phytoplankton biomass resulted in greater
transmembrane pressure.

Pseudo-nitzschia spp.
(cells L-1 )
Particulate domoic acid
(µg L-1 )
Dissolved domoic acid
(µg L-1 )
2.0x10 5
1.5x10 5
1.0x10 5
5.0x10 4
4. Desalination operations and HAB events
A thorough understanding of the specific factors and conditions
giving rise to harmful algal blooms and toxin production
in coastal waters will require a great deal of additional
research before accurate models for predicting these toxic
events will be readily available. Until then, appropriate
monitoring strategies to detect imminent bloom events and
the ability to track the evolution of an active bloom, coupled
with an understanding of the potential toxins being produced,
0
4.0
3.0
2.0
1.0
0.5
0.0
3.0
2.5
2.0
1.5
1.0
0.5
0.0
01-Mar-05
01-May-05
water research 44 (2010) 385–416 405
01-Jul-05
01-Sep-05
01-Nov-05
Fig. 10 – Time series of abundances of Pseudo-nitzschia spp. cells (top), domoic acid concentrations in particulate material
(middle) and dissolved in seawater (bottom) at a coastal monitoring site in El Segundo, CA.
01-Jan-06
01-Mar-06
01-May-06
01-Jul-06
01-Sep-06
01-Nov-06
the toxin chemistry, and their rejection by seawater reverse
osmosis membranes, provide a seawater desalination facility
with the best strategy for making operational adjustments to
ensure that the treatment plant capacity or product water
quality remains unaffected.
4.1. Concern with harmful algal blooms and their
toxin production
As seawater desalination has continued to become more costeffective
and less energy intensive (Al-Sahlawi, 1999), many

406
communities are planning or implementing seawater desalination
facilities (Al-Sahlawi, 1999; Burbano et al., 2007). The
selected pretreatment procedures and the process engineering
that determines the ultimate facility design is entirely
dependent upon the source water quality (e.g. suspended
solids, turbidity, organic material content, algal cell content,
etc.) and its variations, particularly for facilities incorporating
open intakes (Bonnelye et al., 2004b; Gaid and Treal, 2007).
When the seawater desalination process is performed by
reverse osmosis membranes, the selection and proper operation
of a pretreatment system is paramount to the success of
the downstream desalination process (Tenzer et al., 1999;
Bonnelye et al., 2004b; Separation Processes Inc., 2005; Gaid
and Treal, 2007; Petry et al., 2007).
Algal blooms are known to have significant negative
impacts on reverse osmosis desalination facilities. A variety of
pretreatment trains have been considered to address the
difficult source water quality associated with algal blooms,
where the organic and biomass load increase dramatically
(Adin and Klein-Banay, 1986; Al Arrayedhy, 1987; Hasan Al-
Sheikh, 1997; Watson, 1997; Abdul Azis et al., 2000; Bonnelye
et al., 2004a,b; Burbano et al., 2007; Gaid and Treal, 2007; Petry
et al., 2007; Peleka and Matis, 2008). Recently, microfiltration
and ultrafiltration membrane pretreatment has been identified
as a component of a preferred pretreatment train due to the
consistent, high quality water produced by membrane filtration,
especially when compared to conventional processes
(Wilf and Schierach, 2001). However, one significant drawback
to implementation of these modern pretreatment technologies
is that they are as susceptible, or possibly more so, to significant
algal blooms (Bonnelye et al., 2004a).
An early warning system can provide information to
a seawater desalination facility so that functional changes can
be made to efficiently maintain operations even as source
water quality deteriorates. Turbidity sensors offer a rapid
measurement of the total amount of suspended particles in
the intake water for making these decisions. On in situ
instruments such as the Slocum glider, backscatter
measurements yield this type of information (Fig. 7d).
Measurements of chlorophyll a fluorescence can augment this
surveillance approach by estimating the degree to which algal
bimoass contributes to the total load of suspended particles.
Responses to deteriorating water quality may include chemical
additions, use of additional pretreatment equipment, or
additional staff preparations (e.g. maintenance activities,
guaranteeing all membranes and filters are clean in preparation
for an event) to continuously deliver a high quality
feedwater to the reverse osmosis system that will produce the
desalinated drinking water.
4.2. Addressing spatiotemporal variability in HAB
abundance and early detection
As detailed above, it is essential that a desalination facility
incorporate a means of rapid algal bloom detection so that,
when necessary, proper process changes can be made to
maintain the production capacity. Sensors for detecting an
eminent algal bloom can be located at the desalination facility
to inform personnel regarding changes in water quality that
are directly observed on the source water. Fig. 9 presents the
water research 44 (2010) 385–416
transmembrane pressure (TMP) of a microfiltration system
that serves as pretreatment to a pilot-scale reverse osmosis
desalination system along with the levels of chlorophyll
a fluorescence observed in the feedwater. It is clear from this
figure that increased membrane fouling rates (e.g. faster daily
rise in the TMP) were associated with increasing chlorophyll
a fluorescence (i.e. increased algal biomass) in the source
water.
It is well known that higher concentrations of algae cause
increased membrane fouling rates in microfiltration systems
that are frequently incorporated, or considered, in today’s
desalination facilities (Gijsbertsen-Abrahamse et al., 2006; Lee
and Walker, 2006; Reiss et al., 2006). A more complete approach
might include a monitoring system located offshore that
measures some of the primary factors influencing algal
blooms, such as nutrient monitoring in near-real time using
new in situ sensor technology (Glibert et al., 2008). Such information
would be useful to both the desalination facility and
HAB researchers who are continually improving their understanding
of the causative factors that produce HABs and their
associated toxins. Using the information provided by offshore
sensors, the desalination facility personnel could note trends
and shifts in driving factors that generate algal blooms and
make any chemical orders or perform maintenance procedures
that have significant lead times. The same offshore sensor
might also incorporate real-time monitors of sentinel parameters
for changes in algal biomass, such as turbidity and chlorophyll
a, allowing the facility to prepare for changes in
chemical additions and redundant equipment service.
Monitoring of basic chemical parameters of seawater (e.g.
chlorophyll a concentrations) will provide valuable information
for facility operations, but this activity is not sufficient to
fully assess potentially toxic conditions that might arise from
algae that do not require high standing stocks to constitute
a significant toxic threat, such as Alexandrium species.
Species-specific approaches, such as automated in situ
instruments or laboratory-based methods, as well as chemical/immunological
analyses that identify and quantify
specific algal toxins are necessary to more thoroughly characterize
the potential hazards posed by HAB species. The
consistent removal of these potentially toxic substances
through the reverse osmosis process is both a function of
size (e.g. molecular weight) and charge (e.g. zeta potential)
(Amy et al., 2005). Depending on the size and charge of
the contaminant of concern, the rejection, or removal, by the
reverse osmosis process will differ. It is important that we
continue to broaden our knowledge on potentially toxic
substances excreted by algal stock and their associated
blooms.
The approach for obtaining this information would be best
complemented with knowledge of the species that are present
regionally, the potential problems they pose (e.g. specific toxins
and the amounts of soluble microbial products and extracellular
polymeric substances excreted), the spatial extent of HAB
episodes, and their seasonality. The seasonality of Pseudo-nitzschia
spp. and domoic acid near the intake of a pilot desalination
plant in El Segundo, CA, exemplifies the usefulness of routine
monitoring for identifying potentially toxic conditions in coastal
waters adjacent to a plant (Fig. 10). Abundances of Pseudo-nitzschia
spp. and concentrations of domoic acid contained in the

algal cells or dissolved in the intake water exhibited a springtime
peak. Knowledge of the seasonality of this toxic bloom-forming
species allows intensive sampling of coastal waters during
spring when toxic events are more common, improving the
overall effectiveness of the monitoring effort and making it
more cost effective.
Historical and real-time information on the spatial distribution
of HABs can provide information vital for optimizing
design and performance of desalination operations. Local/
regional hydrography, and resultant algal blooms can differ
dramatically. When constructing a new intake pipeline, the
selection of its location (e.g. depth and distance from shore) can
be greatly enhanced through the use of offshore monitoring
devices and efforts to take into account the presence of any
local accumulations of algal biomass due to currents, water
mass convergences/divergences or internal waves, and also
subsurface maxima in algal abundance. Properly locating
offshore monitoring can provide significant information that
will allow optimal location of a new intake pipeline or identification
of issues that might affect an existing one, thereby
significantly reducing the organic and suspended solid loads
present in the feedwater during algal bloom events. These
considerations will ease pretreatment operations, reduce the
cost of water production, and help improve the facility’s
longevity.
5. Conclusions
5.1. Potential impacts, unresolved issues
and research prospectus
The presence of harmful algae in coastal waters that might be
employed in reverse osmosis desalination pose potential
problems for these operations that have been known to even
cause desalination facilities to temporarily cease production
(Tenzer et al., 1999; Pankratz, 2008). As the number of seawater
desalination facilities continues to grow with lower costs and
increasing demand, it is essential that these operating facilities
develop the tools necessary to allow process changes and
ensure capacity objectives continue to be met. Regardless of
the pretreatment configuration, changes in source water
quality require adjustments and these changes need to carefully
coordinate to ensure that the reverse osmosis membranes
are not irreversibly fouled or damaged in the process.
Benchmark work is required to establish the effectiveness
of the seawater reverse osmosis process in dealing with HAB
toxins and other phytoplankton-derived substances. Even if
advanced pretreatment technologies such as microfiltration
are implemented upstream of the reverse osmosis process,
passage of transparent extracellular material produced by the
algal bloom (Alldredge et al., 1993; Hong et al., 1997) may affect
reverse osmosis membrane performance. Additionally, the
physical durability of phytoplankton varies greatly and the
pretreatment process might disrupt cells and create significantly
higher concentrations of dissolved organic substances,
including toxins, than were originally present in the source
water. For example, dissolved domoic acid has been observed
in the seawater passing through the prefiltration process
(Fig. 10, bottom panel) but it is unclear if these values are
water research 44 (2010) 385–416 407
higher due to cell breakage as a result of the prefiltration
process. Therefore, it is important that the international
desalination community carefully characterize these potential
contaminants and their removal to improve treatment
approaches in seawater desalination.
To our knowledge, there are no published reports on the
effectiveness of reverse osmosis for removing dissolved algal
toxins from seawater. Some of these toxin molecules (e.g.
domoic acid) are near the theoretical molecular size of molecules
rejected by reverse osmosis membranes, but experimental
studies are required to validate the effective of this
process on toxin removal. In addition, more information will
be needed to understand the potential impact of discharged
brine and pretreatment backwash water resulting from the
reverse osmosis desalination process on the ecology of coastal
ecosystems. The use of ferric sulfate or ferric chloride as
a pretreatment coagulant would concentrate toxic algae and
their associated toxins if they are present in the intake water.
Similarly, the discharge of brine resulting from the reverse
osmosis process would contain elevated concentrations of
dissolved algal toxins relative to unfiltered seawater. The
degree of concentration of these toxins would not be expected
to be large, but the significance of these processes will depend
on the starting concentrations in the raw intake or prefiltered
water and the degree of concentration due to treatment. There
is presently no information on algal toxins in these
discharges.
HABs on the U.S. west coast exhibit significant generalities
across geographical and temporal scales (e.g. many of the
same species occur throughout the region), but the details of
bloom dynamics differ with geographic location, depth and
season (and perhaps on interannual and decadal scales). The
high degree of variability associated with these events makes
constant monitoring of HABs in intake water for desalination
a vital issue. Regional HAB programs and regulatory agencies
along the U.S. west coast presently provide useful information
for some known potential problems (e.g. ASP and PSP toxins)
for end users that need information on coastal water quality.
Awareness (and augmentation) of this information could
improve planning and safe operation of desalination facilities.
Monitoring of newly emerging HAB concerns (e.g. Cochlodinium
spp.), or HABs and toxins that are presently poorly characterized
(e.g. NSP, DSP, and yessotoxin poisoning) should also
be implemented in the future to allow evaluation of their
potential impacts on desalination processes.
New technologies for toxin detection and quantification,
and in situ monitoring of biological and chemical parameters
are rapidly improving our ability to monitor coastal ecosystems
and identify potentially problematic situations involving
HABs. Advances in in situ observing technologies (sensor
networks, autonomous sensor-equipped vehicles) provide the
capability for obtaining unprecedented resolution in the
spatial and temporal distributions of chemical and physical
parameters, and some biologically important features
(Sukhatme et al., 2007). New approaches and instruments for
toxin detection help identify contaminated seafood products,
and constitute sentinels for the threats of HABs to marine
animal populations. Future uses of coastal waters for desalination
will also benefit from, and contribute to, these
activities.

408
Acknowlegements
The authors are grateful to Mr. Mark Donovan at Separation
Processes, Inc. for providing the data for Fig. 9. The preparation
of this manuscript was supported in part by funding from
a contract between the West Basin Municipal Water District,
Department of Water Resources and the University of Southern
California, National Oceanic and Atmospheric Administration
grants NA05NOS4781228 and NA07OAR4170008, Sea Grant
NA07OAR4170008, National Science Foundation grants CCR-
0120778 (Center for Embedded Networked Sensing; CENS),
DDDAS-0540420, MCB-0703159, and a NASA Earth and Space
Science Fellowship Grant NNX06AF88H. M.-È. Garneau was
supported by a fellowship from the Fonds québécois de
recherche sur la nature et les technologies (FQRNT).
references
Abdul Azis, P.K., Al-Tisan, I., Al-Daili, M., Green, T.N., Dalvi, A.G.I.,
Javeed, M.A., 2000. Effects of environment on source water for
desalination plants on the eastern coast of Saudi Arabia.
Desalination 132 (1–3), 29–40.
Adams, N.A., Lesoing, M., Trainer, L., 2000. Environmental
influences on domoic acid accumulation in razor clams on the
Washington coast. Journal of Shellfish Research 19 (2),
1007–1015.
Adin, A., Klein-Banay, C., 1986. Pretreatment of seawater by
flocculation and settling for particulates removal.
Desalination 58 (3), 227–242.
Ahn, S., Kulis, D.M., Erdner, D.L., Anderson, D.M., Walt, D.R., 2006.
Fiber-optic microarray for simultaneous detection of multiple
harmful algal bloom species. Applied and Environmental
Microbiology 72 (9), 5742–5749.
Al Arrayedhy, M., 1987. Pre- and post-treatment at the RO plant at
RA’s Abu Jarjur, Bahrain. Desalination 63 (C), 81–94.
Al-Sahlawi, M.A., 1999. Seawater desalination in Saudi Arabia:
economic review and demand projections. Desalination 123
(2–3), 143–147.
Alfonso, A., de la Rosa, L., Vieytes, M.R., Yasumoto, T., Botana, L.M.,
2003. Yessotoxin, a novel phycotoxin, activates
phosphodiesterase activity: effect of yessotoxin on cAMP
levels in human lymphocytes. Biochemical Pharmacology 65
(2), 193–208.
Alldredge, A.L., Passow, U., Logan, B.E., 1993. The abundance and
significance of a class of large, transparent organic particles in
the ocean. Deep-Sea Research Part I Oceanographic Research
Papers 40 (6), 1131–1140.
Allen, W.E., 1922. Observations on surface distribution of marine
diatoms between San Diego and Seattle. Ecology 3 (2), 140–145.
Allen, W.E., 1924. Surface catches of marine diatoms and
dinoflagellates made by U.S.S. Pioneer between San Diego and
Seattle in 1923. In: University of California Publications in
Zoology, vol. 26, pp. 243–248.
Allen, W.E., 1928. Review of five years of studies on phytoplankton at
Southern California piers, 1920–1924, inclusive. Bulletin of the
Scripps Institution of Oceanography Technical Services 1,
357–401.
Allen, W.E., 1936. Occurrence of marine plankton diatoms in a tenyear
series of daily catches in Southern California. American
Journal of Botany 23 (2), 60–63.
Allen, W.E., 1940. Summary of results of twenty years of
researches on marine phytoplankton. Proceedings of the 6th
Pacific Scientific Congress 3, 576–583.
water research 44 (2010) 385–416
Allen, W.E., 1941. Twenty years’ statistical studies of marine
plankton dinoflagellates of Southern California. American
Midland Naturalist 26, 603–635.
Allen, J.I., Smyth, T.J., Siddom, J.R., Holt, M., 2008. How well can
we forecast high biomass algal biomass events in a eutrophic
coastal sea? Harmful Algae 8 (1), 70–76.
Amy, G., Kim, T.U., Yoon, J., Bellona, C., Drewes, J., Pellegrino, J.,
Heberer, T., 2005. Removal of micropollutants by NF/RO
membranes. Water Science and Technology: Water Supply 5 (5),
25–33.
Anderson, C.R., Brzezinski, M.A., Washburn, L., Kudela, R., 2006.
Circulation and environmental conditions during a toxigenic
Pseudo-nitzschia australis bloom in the Santa Barbara Channel,
California. Marine Ecology Progress Series 327, 119–133.
Anderson, D.M., Kulis, D.M., Sullivan, J.J., Hall, S., Lee, C., 1990.
Dynamics and physiology of saxitoxin production by the
dinoflagellates Alexandrium spp. Marine Biology 104 (3), 511–524.
Anderson, D.M., Glibert, P.M., Burkholder, J.M., 2002. Harmful algal
blooms and eutrophication: nutrient sources, composition, and
consequences. Estuaries 25 (4B), 704–726.
Anderson, D.M., Kulis, D.M., Keafer, B.A., Gribble, K.E., Marin, R.,
Scholin, C.A., 2005. Identification and enumeration of
Alexandrium spp. from the Gulf of Maine using molecular
probes. Deep-Sea Research Part II 52 (19–21), 2467–2490.
Anderson, D.M., Burkholder, J.M., Cochlan, W.P., Glibert, P.M.,
Gobler, C.J., Heil, C.A., Kudela, R.M., Parsons, M.L., Rensel, J.E.J.,
Townsend, D.W., Trainer, V.L., Vargo, G.A., 2008. Harmful algal
blooms and eutrophication: examining linkages from selected
coastal regions of the United States. Harmful Algae 8 (1), 39–53.
Armonies, W., 1989. Occurrence of meiofauna in Phaeocystis
seafoam. Marine Ecology Progress Series 53, 305–309.
Armstrong, M., Kudela, R., 2006. Evaluation of California isolates
of Lingulodinium polyedrum for the production of yessotoxin.
African Journal of Marine Science 28 (2), 399–401.
Aune, T., Sørby, R., Yasumoto, T., Ramstad, H., Landsverk, T., 2002.
Comparison of oral and intraperitoneal toxicity of yessotoxin
towards mice. Toxicon 40 (1), 77–82.
Baden, D.G., 1983. Marine food-borne dinoflagellate toxins.
International Review of Cytology 83, 99–150.
Baden, D.G., 1989. Brevetoxins: unique polyether dinoflagellate
toxins. The FASEB Journal 3 (7), 1807–1817.
Baden, D.G., Bourdelais, A.J., Jacocks, H., Michelliza, S., Naar, J.,
2005. Natural and derivative brevetoxins: historical
background, multiplicity and effects. Environmental Health
Perspectives 113 (5), 621–625.
Barber, R.T., Chavez, F.P., 1983. Biological consequences of El
Niño. Science 222 (4629), 1203–1210.
Bargu, S., Powell, C.L., Coale, S.L., Busman, M., Doucette, G.J.,
Silver, M.W., 2002. Krill: a potential vector for domoic acid in
marine food webs. Marine Ecology Progres Series 237, 209–216.
Bargu, S., Powell, C.L., Wang, Z., Doucette, G.J., Silver, M.W., 2008.
Note on the occurrence of Pseudo-nitzschia australis and
domoic acid in squid from Monterey Bay, CA (USA). Harmful
Algae 7 (1), 45–51.
Bates, H.A., Kostriken, R., Rapoport, A., 1978. The occurrence of
saxitoxin and other toxins in various dinoflagellates. Toxicon
16 (6), 595–601.
Bates, S.S., Trainer, V.L., 2006. The ecology of harmful diatoms. In:
Granéli, E., Turner, J.T. (Eds.), Ecology of Harmful Algae.
Springer, Berlin, pp. 81–93.
Bates, S.S., Bird, C.J., Defreitas, A.S.W., Foxall, R., Gilgan, M.,
Hanic, L.A., Johnson, G.R., McCulloch, A.W., Dodense, P.,
Pocklington, R., Quilliam, M.A., Sim, P.G., Smith, J.C., Rao, D.V.
S., Todd, C.D., Walter, J.A., Wrigh, J.L.C., 1989. Pennate diatom
Nitzschia pungens as the primary source of domoic acid, a toxin
in shellfish from eastern Prince Edwards Island, Canada.
Canadian Journal of Fisheries and Aquatic Sciences 46 (7),
1203–1215.

Bates, S.S., DeFreitas, A.S.W., Milley, J.E., Pocklington, R.,
Quilliam, M.A., Smith, J.C., Worms, J., 1991. Controls on
domoic acid production by the diatom Nitzschia pungens f.
multiseries in cultures: nutrients and irradiance. Canadian
Journal of Fisheries and Aquatic Sciences 48 (7), 1136–1144.
Baugh, K.A., Bush, J.M., Bill, B.D., Lefebvre, K.A., Trainer, V.L.,
2006. Estimates of specific toxicity in several Pseudo-nitzschia
species from the Washington coast, based on culture and field
studies. African Journal of Marine Science 28 (2), 403–407.
Bialojan, C., Takai, A., 1988. Inhibitory effect of a marine-sponge
toxin, okadaic acid, on protein phosphatases. Specificity and
kinetics. Biochemical Journal 256 (1), 283–290.
Bigelow, H.B., Leslie, M., 1930. Reconnaissance of the waters and
plankton of Monterey Bay July 1928. Bulletin of the Museum of
Comparative Zoology 70 (5), 430–581.
Blasco, D., 1978. Observations on the diel migration of marine
dinoflagellates off the Baja Californian coast. Marine Biology
46 (1), 41–47.
Bolin, R.L., Abbott, D.P., 1960. Studies on the marine climate and
phytoplankton of the central coastal area of California,
1954–1960. California Cooperative Oceanic Fisheries
Investigations 9, 23–45.
Boni, L., Ceredi, A., Guerrini, F., Milandri, A., Pistocchi, R., Poletti,
R., Pompei, M., 2002. Toxic Gonyaulax Grindley Reinecke in the
North-Western Adriatic Sea (Italy). In: Tenth International
Conference on Harmful Algal Blooms, St. Petersburg, FL
(abstract).
Bonnelye, V., Snaz, M.A., Comte, C., Colas, F., Plasse, L., Gueguen, F.,
2004a. Ultrafiltration versus conventional pre-treatment
upstream RO: advantages and limits, Seoul, Korea.
Bonnelye, V., Sanz, M.A., Durand, J.P., Plasse, L., Gueguen, F.,
Mazounie, P., 2004b. Reverse osmosis on open intake
seawater: pre-treatment strategy. Desalination 167 (1–3),
191–200.
Bourdelais, A.J., Tomas, C.R., Naar, J., Kubanek, J., Baden, D.G., 2002.
New fish-killing alga in coastal Delaware produces neurotoxins.
Environmental Health Perspectives 110 (5), 465–470.
Bowers, H.A., Tengs, T., Glasgow Jr., H.B., Burkholder, J.M.,
Rublee, P.A., Oldach, D.W., 2000. Development of real-time
PCR assays for rapid detection of Pfiesteria piscicida and related
dinoflagellates. Applied and Environmental Microbiology 66
(11), 4641–4648.
Bowers, H.A., Tomas, C., Tengs, T., Kempton, J.W., Lewitus, A.J.,
Oldach, D.W., 2006. Raphidophyceae (Chadefaud ex Silva)
systematics and rapid identification: sequence analyses and
real-time PCR assays. Journal of Phycology 42 (6), 1333–1348.
Bricelj, V.M., Lonsdale, D.J., 1997. Aureococcus anophagefferens:
causes and ecological consequences of brown tides in U.S.
mid-Atlantic coastal waters. Limnology and Oceanography 42
(5), 1023–1038.
Buck, K.R., Uttal-Cooke, L., Pilskaln, C.H., Roelke, D.L., Villac, M.C.,
Fryxell, G.A., Cifuentes, L., Chavez, F.P., 1992. Autecology of
the diatom Pseudo-nitzschia australis, a domoic acid producer,
from Monterey Bay, California. Marine Ecology Progress Series
84 (3), 293–302.
Burbano, A.A., Adham, S.S., Pearce, W.R., 2007. The state of fullscale
RO/NF desalination - Results from a worldwide survey.
Journal of American Water Works Association 99 (4), 116–127.
Burkholder, J.M., Azanza, R.V., Sako, Y., 2006. The ecology of
harmful dinoflagellates. In: Granéli, E., Turner, J.T. (Eds.),
Ecology of Harmful Algae. Springer, Berlin, pp. 53–66.
Burkholder, J.M., Hallegraeff, G.M., Melia, G., Cohen, A.,
Bowers, H.A., Oldach, D.W., Parrow, M.W., Sullivan, M.J.,
Zimba, P.V., Allen, E.H., Kinder, C.A., Mallin, M.A., 2007.
Phytoplankton and bacterial assemblages in ballast water of
U.S. military ships as a function of port of origin, voyage time,
and ocean exchange practices. Harmful Algae 6 (4), 486–518.
water research 44 (2010) 385–416 409
Burkholder, J.M., Glibert, P.M., Skelton, H.M., 2008. Mixotrophy,
a major mode of nutrition for harmful algal species in eutrophic
waters. Harmful Algae 8 (1), 77–93.
Buskey, E.J., 2008. How does eutrophication affect the role of
grazers in harmful algal bloom dynamics? Harmful Algae 8 (1),
152–157.
Busse, L.B., Venrick, E.L., Antrobus, R., Miller, P.E., Vigilant, V.,
Silver, M.W., Mengelt, C., Mydlarz, L., Prezelin, B.B., 2006.
Domoic acid in phytoplankton and fish in San Diego, CA, USA.
Harmful Algae 5 (1), 91–101.
Caron, D.A., Dennett, M.R., Moran, D.M., Schaffner, R.A., Lonsdale, D.
J., Gobler, C.J., Nuzzi, R., McLean, T.I., 2003. Development and
application of a monoclonal-antibody technique for counting
Aureococcus anophagefferens, an alga causing recurrent brown
tides in the Mid-Atlantic United States. Applied and
Environmental Microbiology 69 (9), 5492–5502.
Carpenter, E.J., Janson, S., Boje, R., Pollehne, F., Chang, J., 1995.
The dinoflagellate Dinophysis norvegica: biological and
ecological observations in the Baltic Sea. European Journal of
Phycology 109.
Casper, E.T., Patterson, S.S., Bhanushali, P., Farmer, A., Smith, M.,
Fries, D.P., Paul, J.H., 2007. A handheld NASBA analyzer for the
field detection and quantification of Karenia brevis. Harmful
Algae 6 (1), 112–118.
Catterall, W.A., 1992. Cellular and molecular biology of voltagegated
sodium channels. Physiological Reviews 72 (Suppl. 4.),
S15–S48.
CDPH, 2007. Preharvest Shellfish Protection and Marine Biotoxin
Monitoring Program. California Department of Public Health.
Cembella, A.D., 1989. Occurrence of okadaic acid, a major
diarrheic shellfish toxin, in natural populations of Dinophysis
spp. from the eastern coast of North America. Journal of
Applied Phycology 1 (4), 307–310.
Cembella, A.D., 2003. Chemical ecology of eukaryotic microalgae
in marine ecosystems. Phycologica 42 (4), 420–447.
Cestele, S., Catterall, W.A., 2000. Molecular mechanisms of
neurotoxin action on voltage-gated sodium channels.
Biochimie 82 (9–10), 883–892.
Chavez, F.P., Ryan, J., Lluch-Cota, S.E., Niquen, M., 2003. From
anchovies to sardines and back: multidecadal change in the
Pacific ocean. Science 299 (5604), 217–221.
Ciminiello, P., Fattorusso, E., Forino, M., Poletti, R., Viviani, R., 2000.
Structure determination of carboxyhomoyessotoxin, a new
yessotoxin analogue isolated from Adriatic mussels. Chemistry
Research in Toxicology 13 (8), 770–774.
Ciminiello, P., Fattorusso, E., Forino, M., Magno, S., Poletti, R., 2001.
42,43,44,45,46,47,55-Heptanor-41-oxohomoyessotoxin, a new
biotoxin from mussels of the Northern Adriatic Sea. Chemical
Research in Toxicology 14 (5), 596–599.
Ciminiello, P., Fattorusso, E., Forino, M., Magno, S., Poletti, R.,
Viviani, R., 1998. Isolation of adriatoxin a new analogue of
yessotoxin from mussels of the Adriatic Sea. Tetrahedron Letters
39 (48), 8897–8900.
Clayden, J., Read, B., Hebditch, K., 2005. Chemistry of domoic acid,
isodomoic acids, and their analogues. Tetrahedron 61 (24),
5713–5724.
Cochlan, W.P., Herndon, J., Kudela, R.M., 2008. Inorganic and organic
nitrogen uptake by the toxigenic diatom Pseudo-nitzschia australis
(Bacillariophyceae). Harmful Algae 8 (1), 111–118.
Cooley, H., Gleick, P.H., Wolff, G., 2006. Desalination, with a grain
of salt: a California perspective. In: Hart, I. (Ed.). Pacific
Institute for Studies in Development, Environment, and
Security, Hayward, CA, p. 88.
Coyne, K.J., Hutchins, D.A., Hare, C.E., Cary, S.C., 2001. Assessing
temporal and spatial variability in Pfiesteria piscicida
distributions using molecular probing techniques. Aquatic
Microbial Ecology 24 (3), 275–285.

410
Coyne, K.J., Handy, S.M., Demir, E., Whereat, E.B., Hutchins, D.A.,
Portune, K.J., Doblin, M.A., Cary, S.C., 2005. Improved
quantitative real-time PCR assays for enumeration of harmful
algal species in field samples using an exogenous DNA reference
standard. Limnology and Oceanography Methods 3, 381–391.
Cullen, J.J., Horrigan, S.G., 1981. Effects of nitrate on the diurnal
vertical migration, carbon to nitrogen ratio, and the
photosynthetic capacity of the dinoflagellate Gymnodinium
splendens. Marine Biology 62 (2–3), 81–89.
Curtiss, C.C., Langlois, G.W., Busse, L.B., Mazzillo, F., Silve, M.W.,
2008. The emergence of Cochlodinium along the California
Coast (USA). Harmful Algae 7 (3), 337–346.
Daiguji, M., Satake, M., Ramstad, H., Aune, T., Naoki, H.,
Yasumoto, T., 1998. Structure and fluorometric HPLC
determination of 1-desulfoyessotoxin, a new yessotoxin
analog isolated from mussels from Norway. Natural Toxins 6
(6), 235–239.
Dekshenieks, M.M., Donaghay, P.L., Sullivan, J.M., Rines, J.E.B.,
Osborn, T.R., Twardowski, M.S., 2001. Temporal and spatial
occurrence of thin phytoplankton layers in relation to physical
processes. Marine Ecology Progress Series 223, 61–71.
Demir, E., Coyne, K.J., Doblin, M.A., Handy, S.M., Hutchins, D.A.,
2008. Assessment of microzooplankton grazing on Heterosigma
akashiwo using a species-specific approach combining
quantitative real-time PCR (QPCR) and dilution methods.
Microbial Ecology 55 (4), 583–594.
Determan, T., 2003. Paralytic shellfish poisoning (PSP) patterns in
Puget Sound shellfish in 2001-A report for the Puget Sound
Ambient Monitoring Program Office of Food Safety and
Shellfish Programs. Washington State Department of Health,
Olympia.
Draisci, R., Ferretti, E., Palleschi, L., Marchiafava, C., Poletti, R.,
Milandri, A., Ceredi, A., Pompei, M., 1999a. High levels of
yessotoxin in mussels and presence of yessotoxin and
homoyessotoxin in dinoflagellates of the Adriatic Sea. Toxicon
37, 1187–1193.
Draisci, R., Palleschi, L., Giannetti, L., Lucentini, L., James, K.J.,
Bishop, A.G., Satake, M., Yasumoto, T., 1999b. New approach to
the direct detection of known and new diarrhetic shellfish toxins
in mussels and phytoplankton by liquid chromatography-mass
spectrometry. Journal of Chromatrography 847 (1–2), 213–221.
Eiki, K., Satake, M., Koike, K., Ogata, T., Mitsuya, T., Oshima, Y.,
2005. Confirmation of yessotoxin production by the
dinoflagellate Protoceratium reticulatum in Mutsu Bay. Fisheries
Science 71 (3), 633–638.
Eppley, R.W., Holm-Harisen, O., Strickland, J.D.H., 1968. Some
observations on the vertical migration of dinoflagellates.
Journal of Phycology 4 (4), 333–340.
European Commission, 2002. Directive Commission Decision, No.
C2002 1001. Official Journal of the European Communities
March 15, 2002, 1.75/63, 16.3.
Fehling, J., Davidson, K., Bolch, C.J., Bates, S.S., 2004. Growth and
domoic acid production by Pseudo-nitzschia seriata
(Bacillariophyceae) under phosphate and silicate limitation.
Journal of Phycology 40 (4), 674–683.
Flynn, K.J., 2002. Toxin production in migrating dinoflagellates:
a modelling study of PSP producing Alexandrium. Harmful
Algae 1 (2), 147–155.
Flynn, K.J., 2008. Attack is not the best form of defense: lessons from
harmful algal bloom dynamics. Harmful Algae 8 (1), 129–139.
Flynn, K., Franco, J.M., Fernandez, P., Reguera, B., Zapata, M.,
Wood, G., Flynn, K.J., 1994. Changes in toxin content, biomass
and pigments of the dinoflagellate Alexandrium minutum during
nitrogen refeeding and growth into nitrogen or phosphorus
stress. Marine Ecology Progres Series 111, 99–109.
Franchini, A., Marchesini, E., Poletti, R., Ottaviani, E., 2004. Lethal
and sub-lethal yessotoxin dose-induced morpho-functional
water research 44 (2010) 385–416
alterations in intraperitoneal injected Swiss CD1 mice. Toxicon
44 (1), 83–90.
Fritz, L., Quilliam, M.A., Wright, J.L.C., Beale, A.M., Work, T.M.,
1992. An outbreak of domoic acid poisoning attributed to the
pennate diatom Pseudo-nitzschia australis. Journal of Phycology
28 (4), 439–442.
Fryxell, G.A., Villac, M.C., Shapiro, L.P., 1997. The occurrence of
the toxic diatom genus Pseudo-nitzschia (Bacillariophyceae) on
the West Coast of the USA, 1920–1996: a review. Phycologia 36
(6), 419–437.
Gaid, K., Treal, Y., 2007. Le dessalement des eaux par osmose
inverse: l’expÈrience de VÈolia Water. Desalination 203 (1–3),
1–14.
Gaines, G., Taylor, F.J.R., 1985. An exploratory analysis of PSP
patterns in British Columbia: 1942–1984. In: Anderson, D.M.,
White, A.W., Baden, D.G. (Eds.), Toxic dinoflagellates. Elsevier
Science, New York, pp. 439–444.
Galluzzi, L., Penna, A., Bertozzini, E., Vila, M., Garces, E.,
Magnani, M., 2004. Development of a real-time PCR assay for
rapid detection and quantification of Alexandrium minutum
(a dinoflagellate). Applied and Environmental Microbiology 70
(2), 1199–1206.
Garrison, D.L., Conrad, S.M., Eilers, P.P., Waldron, E.M., 1992.
Confirmation of domoic acid production by Pseudo-nitzschia
australis (Bacillariophyceae) cultures. Journal of Phycology 28
(5), 604–607.
Garthwaite, I., Ross, K.M., Miles, C.O., Hansen, R.P., Foster, D.,
Wilkins, A.L., Towers, N.R., 1998. Polyclonal antibodies to
domoic acid, and their use in immunoassays for domoic acid
in sea water and shellfish. Natural Toxins 6 (3–4), 93–104.
Garthwaite, I., Ross, K.M., Miles, C.O., Briggs, L.R., Towers, N.R.,
Borrell, T., Busby, P., 2001. Integrated Enzyme-Linked
Immunosorbent Assay screening system for amnesic,
neurotoxic, diarrhetic, and paralytic shellfish poisoning toxins
found in New Zealand. Journal of AOAC International 84 (5),
1643–1648.
Geraci, J.R., Anderson, D.M., Timperi, R.J., St. Aubin, D.J., Early, G.A.,
Prescott, J.H., Mayo, C., 1989. Humpback whales (Megaptera
novaeangliae) fatally poisoned by dinoflagellate toxin. Canadian
Journal of Fisheries and Aquatic Sciences 46 (11), 1895–1898.
Gijsbertsen-Abrahamse, A.J., Schmidt, W., Chorus, I., Heijman, S.G.J.,
2006. Removal of cyanotoxins by ultrafiltration and
nanofiltration. Journal of Membrane Science 276 (1–2), 252–259.
Glibert, P.M., Anderson, D.M., Gentien, P., Granéli, E., Sellner, K.G.,
2005a. The global, complex phenomena of harmful algal
blooms. Oceanography 18 (2), 137–147.
Glibert, P.M., Seitzinger, S., Heil, C.A., Burkholder, J.M., Parrow, M.W.,
Codispoti, L.A., Kelly, V., 2005b. The role of eutrophication in
the global proliferation of harmful algal blooms. Oceanography
18 (2), 198–209.
Glibert, P.M., Harrison, J., Heil, C., Seitzinger, S., 2006. Escalating
worldwide use of urea – a global change contributing to
coastal eutrophication. Biogeochemistry 77 (3), 441–463.
Glibert, P.M., Kelly, V., Alexander, J., Cosdispoti, L.A., Boicourt, W.C.,
Trice, T.M., Michael, B., 2008. In situ nutrient monitoring:
a tool for capturing nutrient variability and the antecedent
conditions that support algal blooms. Harmful Algae 8 (1),
175–181.
Granéli, E., Flynn, K., 2006. Chemical and physical factors
influencing toxin content. In: Granéli, E., Turner, J.T. (Eds.),
Ecology of Harmful Algae. Springer, Berlin, pp. 229–241.
Greenfield, D.I., Marin, R., Jensen, S., Massion, E., Roman, B.,
Feldman, J., Scholin, C., 2006. Application of the Environmental
Sample Processor (ESP) methodology for quantifying Pseudonitzschia
australis using ribosomal RNA-targeted probes in
sandwich and fluorescent in situ hybridization. Limnology and
Oceanography Methods 4, 426–435.

Gregorio, D.E., Connell, L., 2000. Range of Heterosigma akashiwo
(Raphidophyceae) expanded to include California, USA. In:
Ninth International Conference on Harmful Algal Blooms,
Tasmania, Australia (abstract).
Gregorio, D.E., Pieper, R.E., 2000. Investigations of red tides along
the southern California. Bulletin of Southern California
Academy of Sciences 99 (8), 147–160.
Gulland, F.M.D., Haulena, M., Fauquier, D., Langlois, G., Lander, M.E.,
Zabka, T., Duerr, R., 2002. Domoic acid toxicity in California sea
lions (Zalophus californianus): clinical signs, treatment and
survival. Veterinary Record 150 (15), 475–480.
Hallegraeff, G., Hara, Y., 1995. Taxonomy of harmful marine
raphidophytes. In: Hallegraef, G., Anderson, D.M., Cembella, A.D.
(Eds.), Manual on Harmful Marine Microalgae. IOC UNCESCO,
Paris, pp. 365–371.
Hallegraeff, G.M., 1993. A review of harmful algal blooms and
their apparent global increase. Phycologia 32 (2), 79–99.
Hallegraeff, G.M., 2003. Harmful algal blooms: a global overview.
In: Hallegraeff, G.M., Anderson, D.M., Cembella, A.D. (Eds.),
Manual on Harmful Marine Microalgae. Monographs on
Oceanographic Methodology. UNESCO Publishing, Paris
pp. 25–49.
Hallegraeff, G.M., Bolch, C.J., 1992. Transport of diatom and
dinoflagellate resting spores in ships’ ballast water:
implications for plankton biogeography and aquaculture.
Journal of Plankton Research 14 (8), 1067–1084.
Handy, S.M., Hutchins, D.A., Cary, S.C., Coyne, K.J., 2006.
Simultaneous enumeration of multiple raphidophyte species
by multiprobing and mutiplexing: capabilities and limitations.
Limnology and Oceanography: Methods 4, 193–204.
Haque, S.M., Onoue, Y., 2002a. Effects of salinity on growth and
toxin production of a noxious phytoflagellate, Heterosigma
akashiwo (Raphidophyceae). Botanica Marina 45 (4), 356–363.
Haque, S.M., Onoue, Y., 2002b. Variation in toxin compositions of
two harmful raphidophytes, Chattonella antiqua and Chattonella
marina, at different salinities. Environmental Toxicology 17 (2),
113–118.
Hard, J.J., Connell, L., Hershberger, W.K., Harrell, L.W., 2000.
Genetic variation in mortality of chinook salmon during
a bloom of the marine alga Heterosigma akashiwo. Journal of
Fish Biology 56 (6), 1387–1397.
Hasan Al-Sheikh, A.H., 1997. Seawater reverse osmosis
pretreatment with an emphasis on the Jeddah Plant operation
experience. Desalination 110 (1–2), 183–192.
Haystead, T.A.J., Sim, A.T.R., Carling, D., Honnor, R.C., Tsukitani, Y.,
Cohen, P., Hardie, D.G., 1989. Effects of the tumour promoter
okadaic acid on intracellular protein phosphorylation and
metabolism. Nature 337, 78–81.
Heisler, J., Glibert, P.M., Burkholder, J.M., Anderson, D.M.,
Cochlan, W., Dennison, W.C., Dortch, Q., Gobler, C.J., Heil, C.A.,
Humphries, E., Lewitus, A., Magnien, R., Marshall, H.G.,
Sellner, K., Stockwell, D.A., Stoecker, D.K., Suddleson, M., 2008.
Eutrophication and harmful algal blooms: a scientific
consensus. Harmful Algae 8 (1), 3–13.
Herndon, J., Cochlan, W.P., Horner, R., 2003. Heterosigma akashiwo
blooms in San Francisco Bay. Interagency Ecological Program
for the San Francisco Estuary Newsletter 16, 46–48.
Hershberger, P.K., Rensel, J.E., Postel, J.R., Taub, F.B., 1997.
Heterosigma bloom and associated fish kill. Harmful Algae
News 16 (1), 4.
Holmes, R.W., Williams, P.M., Eppley, R.W., 1967. Red water in
La Jolla, 1964–1966. Limnology and Oceanography 12 (3),
503–512.
Hong, Y., Smith Jr., W.O., White, A.M., 1997. Studies on
transparent exopolymer particles (TEP) produced in the Ross
Sea (Antarctica) and by Phaeocystis antarctica
(Prymnesiophyceae). Journal Of Phycology 33 (3), 368–376.
water research 44 (2010) 385–416 411
Hori, M., Matsuura, Y., Yoshimoto, R., Ozaki, H., Yasumoto, T.,
Karaki, H., 1999. Actin depolymerizing action by marine toxin,
pectenotoxin-2. Folia Pharmacologica Japonica 114 (Suppl. 1),
225–229.
Horner, R.A., Postel, J.R., 1993. Toxic diatoms in western
Washington waters (U.S. west coast). Hydrobiologia 269/270
(1), 197–205.
Horner, R.A., Garrison, D.L., Plumley, F.G., 1997. Harmful algal
blooms and red tide problems on the U.S. west coast.
Limnology and Oceanography 42 (2), 1076–1088.
Horner, R.A., Hickey, B.M., Postel, J.R., 2000. Pseudo-nitzschia
blooms and physical oceanography off Washington state.
South African Journal of Marine Science 22 (1), 299–308.
Howard, M.D.A., Cochlan, W.P., Ladizinsky, N., Kudela, R.M., 2007.
Nitrogenous preference of toxigenic Pseudo-nitzschia australis
(Bacillariophyceae) from field and laboratory experiments.
Harmful Algae 6 (2), 206–217.
Howard, M.D.A., Silvera, M., Kudela, R.M., 2008. Yessotoxin
detected in mussel (Mytilus californicus) and phytoplankton
samples from the U.S. west coast. Harmful Algae 7 (5),
646–652.
Howard, M.D., Smith, G.J., Kudela, R.M., 2009. Phylogenetic
relationships of yessotoxin-producing dinoflagellates, based
on the large subunit and internal transcribed spacer ribosomal
DNA domains. Applied and Environmental Microbiology 75 (1),
54–63.
Howarth, R.W., 2008. Coastal nitrogen pollution: a review of
sources and trends globally and regionally. Harmful Algae 8
(1), 14–20.
IOC HAB Programme, 2008. IOC Harmful Algal Bloom Website.
Intergovernmental Oceanic Commission of UNESCO.
Iwataki, M., Kawami, H., Matsuoka, K., 2007. Cochlodinium
fulvescens sp nov (Gymnodiniales, Dinophyceae), a new chainforming
unarmored dinoflagellate from Asian coasts.
Phycological Research 55 (3), 231–239.
Iwataki, M., Kawami, H., Mizushima, K., Mikulski, C.M.,
Doucette, G.J., Relox, J.R., Anton, A., Fukuyo, Y., Matsuoka, K.,
2008. Phylogenetic relationships in the harmful dinoflagellate
Cochlodinium polykrikoides (Gymnodiniales, Dinophyceae)
inferred from LSU rDNA sequences. Harmful Algae 7 (3),
271–277.
Jeffery, B., Barlow, T., Moizer, K., Paul, S., Boyle, C., 2004. Amnesic
shellfish poison. Food and Chemical Toxicology 42 (4),
545–557.
Jessup, D.A., Miller, M.A., Ryan, J.P., Nevins, H.M., Kerkering, H.A.,
Mekebri, A., Crane, D.B., Johnson, T.A., Kudela, R.M., 2009.
Mass stranding of marine birds caused by a surfactantproducing
red tide. PLOS One 4, E4550.
Jester, R., 2008. An investigation into the prevalence of
Alexandrium derived toxins in marine food webs. Ph.D.,
University of California, Santa Cruz, Santa Cruz, pp. 116.
Jester, R., Lefebvre, K.A., Langlois, G., Vigilant, V., Baugh, K.A.,
Silver, M.W., 2009a. A shift in the dominant toxin-producing
algal species in central California alters phycotoxins in food
webs. Harmful Algae 8 (2), 291–298.
Jester, R.J., Baugh, K.A., Lefebvre, K.A., 2009b. Presence of
Alexandrium catenella and paralytic shellfish toxins in finfish,
shellfish and rock crabs in Monterey Bay, California, USA.
Marine Biology 156 (3), 493–504.
Johansson, N., Granéli, E., Yasumoto, T., Carlsson, P., Legrand, C.,
1996. Toxin production by Dinophysis acuminata and D. acuta
cells grown under nutrient sufficient and deficient conditions.
In: Yasumoto, T., Oshima, Y., Fukuyo, Y. (Eds.), Harmful and
Toxic Algal Blooms. Intergovernmental Oceanographic
Commission of UNESCO, Paris, pp. 277–279.
John, E.H., Flynn, K.J., 2000. Growth dynamics and toxicity of
Alexandrium fundyense (Dinophyceae): the effect of changing N:

412
P supply ratios on internal toxin and nutrient levels. European
Journal of Phycology 35 (1), 11–23.
Jones, B.H., Noble, M.A., Dickey, T.D., 2002. Hydrographic and
particle distributions over the Palos Verdes continental shelf:
spatial, seasonal and daily variability. Continental Shelf
Research 22 (6–7), 945–965.
Kirkpatrick, B., Fleming, L.E., Squicciarini, D., Backer, L.C.,
Clark, R., Abraham, W., Benson, J., Cheng, Y.S., Johnson, D.,
Pierce, R., Zaias, J., Bossart, G.D., Baden, D.G., 2004. Literature
review of Florida red tide: implications for human health
effects. Harmful Algae 3 (2), 99–115.
Kudela, R.M., Cochlan, W.P., 2000. Nitrogen and carbon uptake
kinetics and the influence of irradiance for a red tide bloom off
southern California. Aquatic Microbial Ecology 21 (1), 31–47.
Kudela, R.M., Roberts, A., Armstrong, M., 2004. Laboratory
analyses of nutrient stress and toxin production in Pseudonitzschia
spp. from Monterey Bay, California. In: Steidinger, K.,
Landsberg, J., Tomas, C., Vargo, G. (Eds.). Florida
Environmental Research Institute and UNESCO, St. Petersberg,
pp. 136–138.
Kudela, R.M., Lane, J.Q., Cochlan, W.P., 2008a. The potential role
of anthropogenically derived nitrogen in the growth
of harmful algae in California, USA. Harmful Algae 8 (1),
103–110.
Kudela, R.M., Ryan, J.P., Blakely, M.D., Lane, J.Q., Peterson, T.D.,
2008b. Linking the physiology and ecology of Cochlodinium to
better understand harmful algal bloom events: a comparative
approach. Harmful Algae 7 (3), 278–292.
Kusek, K.M., Vargo, G., Steidinger, K., 1999. Gymnodinium breve in
the field in the lab and in the newspaper - scientific and
journalistic analysis of Florida red tides. Contributions in
Marine Science 34, 1–229.
Ladizinsky, N., Smith, G.J., 2000. Accumulation of domoic acid by
the coastal diatom Pseudo-nitzschia multiseries: a possible
copper complexation strategy. Journal of Phycology 36, 41.
Landry, M.R., 1989. Broad-scale distributional patterns of
hydrographic variables on the Washington/Oregon shelf. In:
Landry, M.R., Hickey, B.M. (Eds.), Coastal Oceanography of
Washington and Oregon. Elsevier, New York, pp. 1–40.
Lange, C.B., Reid, F.M.H., Vernet, M., 1994. Temporal distribution
of the potentially toxic diatom Pseudo-nitzschia australis at
a coastal site in Southern California. Marine Ecology Progress
Series 104 (3), 309–312.
Langlois, G., 2007. Marine Biotoxin Monitoring Program Annual
Report. California Department of Public Health for California
Department of Fish and Game, Sacramento.
Lee, J., Walker, H.W., 2006. Effect of process variables and natural
organic matter on removal of microcystin-LR by PAC-UF.
Environmental Science and Technology 40 (23), 7336–7342.
Lee, J.S., Yanagi, T., Kenma, R., Yasumoto, T., 1987. Fluorometric
determination of diarrhetic shellfish toxins by highperformance
liquid chromatography. Agricultural and
Biological Chemistry 51 (3), 877–881.
Lee, J.S., Igarashi, T., Fraga, S., Dah, E., Hovgaard, P., Yasumoto, T.,
1989. Determination of diarrhetic toxins in various
dinoflagellate species. Journal of Applied Phycology 1 (2),
147–152.
Lefebvre, K.A., Bill, B.D., Erickson, A., Baugh, K.A., O’Rourke, L.,
Costa, P.R., Nance, S., Trainer, V.L., 2008. Characterization of
intracellular and extracellular saxitoxin levels in both field
and cultured Alexandruim spp. samples from Sequim Bay,
Washington. Marine Drugs 6, 103–116.
Levandowsky, M., Kaneta, P.J., 1987. Behaviour in dinoflagellates.
In: Taylor, F.J.R. (Ed.), The Biology of Dinoflagellates. Blackwell,
Oxford, pp. 360–397.
Lin, Y.Y., Risk, M., Ray, S.M., Van Engen, D., Clardy, J., Golik, J.,
James, J.C., Nakanishi, K., 1981. Isolation and structure of
brevetoxin B from the ‘red tide’ dinoflagellate Ptychodiscus
water research 44 (2010) 385–416
brevis (Gymnodinium breve). Journal of American Chemical
Society 103 (22), 6773–6775.
Llewellyn, L.E., 2006. Saxitoxin, a toxic marine natural product
that targets a multitude of receptors. Natural Product Reports
23 (2), 200–222.
Loeblich III, A.R., Fine, K.E., 1977. Marine chloromonads: More
widely distributed in neritic environments than previously
thought. Proceedings of the Biological Society of Washington
90 (2), 388–399.
Lundholm, N., Skov, J., Pocklington, R., Moestrup, Ã., 1997. Studies
on the marine planktonic diatom Pseudo-nitzschia. 2.
Autecology of P. pseudodelicatissima based on isolates from
Danish coastal waters. Phycologia 36, 381–388.
Malagoli, D., Marchesini, E., Ottaviani, E., 2006. Lysosomes as the
target of yessotoxin invertebrate and vertebrate cell lines.
Toxicology Letters 167 (1), 75–83.
Maldonado, M.T., Hughes, M.P., Rue, E.L., Wells, M.L., 2002. The
effect of Fe and Cu on growth and domoic acid production by
Pseudo-nitzschia multiseries and Pseudo-nitzschia australis.
Limnology and Oceanography 47 (2), 515–526.
Marchetti, A., Trainer, V.L., Harrison, P.J., 2004. Environmental
conditions and phytoplankton dynamics associated
with Pseudo-nitzschia abundance and domoic acid in the
Juan de Fuca eddy. Marine Ecology Progress Series 281,
1–12.
Martin, J.L., Haya, K., Burridge, L.E., Wildish, D.J., 1990. Nitzschia
pseudodelicatissima - a source of domoic acid in the Bay of
Fundy, eastern Canada. Marine Ecology Progress Series 67 (2),
177–182.
Matsuoka, K., Iwataki, M., Kawami, H., 2008. Morphology and
taxonomy of chain-forming species of the genus Cochlodinium
(Dinophyceae). Harmful Algae 7 (3), 261–270.
McGillicuddy Jr., D.J., Signell, R.P., Stock, C.A., Keafer, B.A.,
Keller, M.D., Ehtland, R.D., Anderson, D.M., 2005. A
mechanisms for offshore initiations of harmful algal blooms
in the coastal Gulf of Maine. Journal of Plankton Research 25,
1131–1138.
Mengelt, C., 2006. Ultraviolet Photoecology, Dark Survival, and
Seasonal Abundance of Pseudo-nitzschia australis and P.
multiseries in Coastal Waters of Central California.
Dissertation, University of California, Santa Barbara.
Mikulski, C.M., Morton, S.L., Doucette, G.J., 2005. Development
and application of LSU rRNA probes for Karenia brevis in the
Gulf of Mexico, USA. Harmful Algae 4 (1), 49–60.
Mikulski, C.M., Park, Y.T., Jones, K.L., Lee, C.K., Lim, W.A., Lee, Y.,
Scholin, C.A., Doucette, G.J., 2008. Development and field
application of rRNA-targeted probes for the detection of
Cochlodinium polykrikoides Margalef in Korean coastal waters
using whole cell and sandwich hybridization formats.
Harmful Algae 7 (3), 347–359.
Miles, C.O., Aasen, J., Dahl, E., Samdal, I., Aune, T., Briggs, L.R.,
Olseng, C.D., Edvardsen, B., Naustvoll, L.J., 2002. Yessotoxin in
Norwegian Blue Mussels caused by Protoceratium reticulatum.
In: Tenth International Conference on Harmful Algal Blooms,
St. Petersburg, FL (abstract).
Miles, C.O., Wilkins, A.L., Samdal, I.A., Sandvik, M., Petersen, D.,
Quilliam, M.A., Nausvoll, L.J., Rundberget, T., Torgersen, T.,
Hovgaard, P., Jensen, D.J., Cooney, J.M., 2004. A novel
pectenotoxin, PTX-12, in Dinophysis spp. and shellfish from
Norway. Chemical Research and Toxicology 17 (11),
1423–1433.
Miles, C.O., Wilkins, A.L., Hawkes, A.D., Selwood, A.I., Jensen, D.J.,
Munday, R., Cooney, J.M., Beuzenberg, V., 2005a.
Polyhydroxylated amide analogs of yessotoxin from
Protoceratium reticulatum. Toxicon 45 (1), 61–71.
Miles, C.O., Samdal, I.A., Aasen, J.A.G., Jensen, D.J., Quilliam, M.A.,
Petersen, D., Briggs, L.M., Wilkins, A.L., Rise, F., Cooney, J.M.,
MacKenzie, A.L., 2005b. Evidence for numerous analogs of

yessotoxin in Protoceratium reticulatum. Harmful Algae 4 (6),
1075–1091.
Miles, C.O., Wilkins, A.L., Hawkes, A.D., Selwood, A.I., Jensen, D.J.,
Cooney, J.M., Beuzenberg, V., MacKenzie, L., 2006.
Identification of 45-hydroxy-46,47-dinoryessotoxin, 44-oxo-
45,46,47-trinoryessotoxin, and 9-methyl-42,43,44,45,46,47,55hepta-nor-38-en-41-oxoyessotoxin,
and partial
characterization of some minor yessotoxins, from
Protoceratium reticulatum. Toxicon 47 (2), 229–240.
Miller, P.E., Scholin, C.A., 1998. Identification and enumeration of
cultured and wild Pseudo-nitzschia (Bacillariophyceae) using
species-specific LSU rRNA-targeted fluorescent probes and
filter-based whole cell hybridization. Journal of Phycology 34
(2), 371–382.
Mitrovic, S.M., Fernández Amandi, M., MacKenzie, L., Furey, A.,
James, K.J., 2004. Effects of selenium, iron and cobalt
addition to growth and yessotoxin production of the toxic
marine dinoflagellate Protoceratium reticulatum in culture.
Journal of Experimental Marine Biology and Ecology 313 (2),
337–351.
Mitrovic, S.M., Hamilton, B., MacKenzie, L., Furey, A., James, K.J.,
2005. Persistence of yessotoxin under light and dark
conditions. Marine Environmental Research 60 (3), 397–401.
Montojo, U.M., Sakamoto, S., Cayme, M.F., Gatdula, N.C., Furio, E.
F., Relox, J.R.J., Sato, S., Fukuyo, Y., Kodam, M., 2006.
Remarkable difference in accumulation of paralytic shellfish
poisoning toxins among bivalve species exposed to Pyrodinium
bahamense var. compressum bloom in Masinloc bay. Philippines
Toxicon 48 (1), 85–92.
Montoya, N.G., Akselman, R., Franco, J.M., Carreto, J.I., 1996.
Paralytic shellfish toxins and mackerel (Scomber japonicus)
mortality in the Argentine sea. In: Yasumoto, T., Oshima, Y.,
Fukuyo, Y. (Eds.), Harmful and Toxic Algal Blooms.
Intergovernmental Oceanographic Community, UNESCO,
Paris, pp. 417–420.
Moore, S.K., Mantua, N.J., Kellogg, J.P., Newton, J.A., 2008. Local
and large-scale climate forcing of Puget Sound oceanographic
properties on seasonal to interdecadal timescales. Limnology
and Oceanography 53 (5), 1746–1758.
Moore, S.K., Mantua, N.J., Hickey, B.M., Trainer, V.L., 2009. Recent
trends in paralytic shellfish toxins in Puget Sound,
relationships to climate, and capacity for prediction of toxic
events. Harmful Algae 8 (3), 463–477.
Moorthi, S.D., Countway, P.D., Stauffer, B.A., Caron, D.A., 2006.
Use of quantitative real-time PCR to investigate the dynamics
of the red tide dinoflagellate Lingulodinium polyedrum.
Microbial Ecology 52 (1), 136–150.
Murata, M., Shimatani, M., Sugitani, Y., Oshima, Y.,
Yasumoto, T., 1982. Isolation and structural elucidation of
the causative toxin of the diarrhetic shellfish poisoning.
Bulletin of the Japanese Society of Scientific Fisheries 48 (4),
549–552.
Murata, M., Kumagai, M., Lee, J.S., Yasumoto, T., 1987. Isolation
and structure of Yessotoxin, a novel polyether compound
implicated in diarrhetic shellfish poisoning. Tetrahedron
Letters 28 (47), 5869–5872.
Naar, J., Bourdelais, A., Tomas, C., Kubanek, J., Whitney, P.L.,
Flewelling, L., Steidinger, K., Lancaster, J., Baden, D.G., 2002.
A competitive ELISA to detect brevetoxins from Karenia brevis
(formerly Gymnodinium breve) in seawater, shellfish, and
mammalian body fluid. Environmental Health Perspectives
110 (2), 179–185.
Netjstgaard, J.C., Tang, K.W., Steinke, M., Dutz, J., Koski, M.,
Antajan, E., Long, J.D., 2007. Zooplankton grazing on
Phaeocystis: a quantitative review and future challenges.
Biogeochemistry 83 (1–3), 147–172.
Nishitani, G., Yamaguchi, M., Ishikawa, A., Yanagiya, S.,
Mitsuya, T., Imai, M., 2005. Relationships between occurrences
water research 44 (2010) 385–416 413
of toxic Dinophysis species (Dinophyceae) and small
phytoplanktons in Japanese coastal waters. Harmful Algae 4
(4), 755–762.
O’Halloran, C., Silver, M., Holman, T., Scholin, C., 2006.
Heterosigma akashiwo in central California waters. Harmful
Algae 5 (2), 124–132.
Office of Shellfish and Water Protection, 2008. 2007 Annual
Report: Commercial and Recreational Shellfish Areas in
Washington State Olympia. Office of Shellfish and Water
Protection.
Ogino, H., Kumagai, M., Yasumoto, T., 1997. Toxicologic
evaluation of Yessotoxin. Natural Toxins 5 (6), 255–259.
Ono, K., Khan, S., Onoue, Y., 2000. Effects of temperature and light
intensity on the growth and toxicity of Heterosigma akashiwo
(Raphidophyceae). Aquaculture Research 31 (5), 427–433.
Pan, Y., Subba Rao, D.V., Mann, K.H., 1996a. Changes in domoic
acid production and cellular chemical composition of the
toxigenic diatom Pseudo-nitzschia muliseries under
phosphate limitation. Journal of Phycology 32 (3), 371–381.
Pan, Y., Subba Rao, D.V., Mann, K.H., Li, K.W., Harrison, W.G.,
1996b. Effects of silicate limitation on production of domoic
acid, a neurotoxin, by the diatom Pseudo-nitzschia multiseries. I.
Batch culture studies. Marine Ecology Progress Series 131 (1–3),
225–233.
Pan, Y., Bates, S.S., Cembella, A.D., 1998. Environmental stress
and domoic acid production by Pseudo-nitzschia:
a physiological perspective. Natural Toxins 6 (3–4), 127–135.
Pankratz, T., 2008. Red tides close desal plants. Water
Desalination Report 44, 1.
Pankratz, T., 2009. Speaking of red tides. Water Desalination
Report 45, 2.
Park, M.G., Kim, S., Kim, H.S., Myung, G., Kang, Y.G., Yih, W., 2006.
First successful culture of the marine dinoflagellate Dinophysis.
Aquatic Microbial Ecology 45 (2), 101–106.
Paz, B., Riobo, P., Fernandez, A.L., Fraga, S., Franco, J.M., 2004.
Production and release of yessotoxins by the dinoflagellates
Protoceratium reticulatum and Lingulodinium polyedrum in
culture. Toxicon 44 (3), 251–258.
Paz, B., Riobo, P., Souto, M.L., Gil, L.V., Norte, M., Fernandez, J.J.,
Franco, J.M., 2006. Detection and identification of
glycoyessotoxin A in a culture of the dinoflagellate
Protoceratium reticulatum. Toxicon 48 (6), 611–619.
Paz, B., Daranas, A.H., Cruz, P.G., Franco, J.M., Pizarro, G.,
Souto, M.L., Norte, M., Fernández, J.J., 2007. Characterization of
okadaic acid related toxins by liquid chromatography coupled
with mass spectrometry. Toxicon 50 (2), 225–235.
Paz, B., Daranas, A.H., Norte, M., Riobo, P., Franco, J.M.,
Fernández, J.J., 2008. Yessotoxins, a group of marine polyether
toxins: an overview. Marine Drugs 6 (6), 73–102.
Peleka, E.N., Matis, K.A., 2008. Application of flotation as
a pretreatment process during desalination. Desalination 222
(1–3), 1–8.
Perl, T.M., Bedard, L., Kosatsky, T., Hockin, J.C., Todd, E.C.D.,
McNutt, L.A., Remis, R.S., 1990. Amnesic shellfish poisoning:
a new clinical syndrome due to domoic acid. Canadian
Diseases Weekly Report 16 (Suppl 1E), 7–8.
Petry, M., Sanz, M.A., Langlais, C., Bonnelye, V., Durand, J.P.,
Guevara, D., Nardes, W.M., Saemi, C.H., 2007. The El Coloso
(Chile) reverse osmosis plant. Desalination 203 (1–3),
141–152.
Pocklington, R., Milley, J.E., Bates, S.S., Bird, C.J., deFreitas, A.S.W.,
Quilliam, M.A., 1990. Trace determination of domoic acid and
phytoplankton by high-performance liquid chromatography
of the fluorenylmethoxycarbonyl (FMOC) derivative.
International Journal of Environmental Analytical Chemistry
38, 351–358.
Prassopoulou, E., Katikou, P., Georgantelis, D., Kyritsakis, A.,
2009. Detection of okadaic acid and related esters in mussels

414
during diarrhetic shellfish poisoning (DSP) episodes in
Greece using the mouse bioassay, the PP2A inhibition assay
and HPLC with fluorimetric detection. Toxicon 53 (2),
214–227.
Price, D.W., Kizer, K.W., Hansgen, K.H., 1991. California’s paralytic
shellfish poisoning prevention program, 1927–89. Journal of
Shellfish Research 10 (11), 119–145.
Purkerson, S.L., Baden, D.G., Fieber, L.A., 1999. Brevetoxin
modulates neuronal sodium channels in two cell lines derived
from rat brain. Neurotoxicology 20 (6), 909–920.
Quilliam, M., 2003. Chemical methods for domoic acid, the
amnesic shellfish poisoning (ASP) toxin. In: Hallegraeff, G.,
Anderson, D., Cembella, A. (Eds.), Manual on Harmful Marine
Microalgae, Monographs on Oceanographic Methodology.
Intergovernmental Oceanographic Commission (UNESCO),
Paris, pp. 247–266.
Quilliam, M.A., Sim, P.G., McCulloch, A.W., McInnes, A.G., 1989.
High-performance liquid chromatography of domoic acid,
a marine neurotoxin, with application to shellfish and
plankton. International Journal of Environmental Analytical
Chemistry 36 (3), 139–154.
Ramsdell, J.S., Anderson, D.M., Glibert, P.M., 2005. HARRNESS,
Harmful Algal Research and Response: a National
Environmental Science Strategy 2005–2015. Ecological Society
of America, Washington, DC, 96 pp.
Reid, J.L., Roden, G.I., Wyllie, J.G., 1958. Studies of the California
Current system. California Cooperative Oceanic Fisheries
Investigations Progress Report, 1 July 1956–1 January 1958,
pp. 27–56.
Reid, F.M.H., Fuglister, E., Jordan, J.B., 1970. The ecology of the
plankton off La Jolla, California, in the period April through
September, 1967. Part 5: phytoplankton taxonomy and
standing crop. Bulletin of the Scripps Institute of
Oceanography Techical Series 17, 51–66.
Reid, F.M.H., Lange, C.B., White, M.M., 1985. Microplankton
species assemblages at Scripps Pier from March to November
1983 during the 1982–1983 El Niño Event. Botanica Marina 28
(10), 443–452.
Reiss, C.R., Robert, C., Owen, C., Taylor, J.S., 2006. Control of MIB,
geosmin and TON by membrane systems. Journal of Water
Supply: Research and Technology – AQUA 55 (2), 95–108.
Rhodes, L., White, D., Syhre, M., Atkinson, M., 1996. Pseudonitzschia
species isolated from New Zealand coastal waters:
domoic acid production in vitro and links with shellfish
toxicity. In: Yasumoto, T., Oshima, Y., Fukuyo, Y. (Eds.),
Harmful and Toxic Algal Blooms. IOC of UNESCO, Paris
pp. 155–158.
Rhodes, L., Scholin, C., Garthwaite, I., Haywood, A., Thomas, A.,
1998. Domoic acid producing Pseudo-nitzschia species deduced
with whole cell DNA probe-based and immunochemical
assays. In: Reguera, B., Blanco, J., Fernández, M.L., Wyatt, T.
(Eds.), Harmful Algae. Xunta de Galicia and
Intergovernmental Oceanographie Commission of UNESCO,
pp. 274–277.
Rhodes, L., McNabb, P., de Salas, M., Briggs, L., Beuzenberg, V.,
Gladstone, M., 2006. Yessotoxin production by Gonyaulax
spinifera. Harmful Algae 5 (2), 148–155.
Rines, J.E.B., Donaghay, P.L., Dekshenieks, M.M., Sullivan, J.M.,
Twardowski, M.S., 2002. Thin layers and camouflage: hidden
Pseudo-nitzschia spp. (Bacillariophyceae) populations in a fjord
in the San Juan Island, Washington, USA. Marine Ecology
Progress Series 225, 123–137.
Riobo, P., Paz, B., Fernandez, M., Fraga, S., Franco, J.M., 2002.
Lipophylic toxins of different strains of Ostreopsidaceae and
Gonyaulacaceae. In: Tenth International Conference on
Harmful Algal Blooms, St. Petersburg, FL (abstract).
Rogers, R.S., Rapoport, H., 1980. The pKas of saxitoxin. Journal of
American Chemical Society 102 (24), 7335–7339.
water research 44 (2010) 385–416
Ronzitti, G., Callegari, F., Malaguti, C., Rossini, G.P., 2004. Selective
disruption of the E-cadherin–catenin system by an algal toxin.
British Journal of Cancer 90 (5), 1100–1107.
Rue, E.L., Bruland, K.W., 2001. Domoic acid binds iron and copper:
a possible role for the toxin produced by the marine diatom
Pseudo-nitzschia. Marine Chemistry 76 (1–2), 127–134.
Samdal, I.A., Naustvoll, L.J., Olseng, C.D., Briggs, L.R., Miles, C.O.,
2004. Use of ELISA to identify Protoceratium reticulatum as
a source of yessotoxin in Norway. Toxicon 44 (1),
75–82.
Samdal, I.A., Aesen, J.A.G., Briggs, L.R., Dahl, E., Miles, C.O., 2005.
Comparison of ELISA and LC–MS analyses for yessotoxins in
blue mussels (Mytilus edulis). Toxicon 46, 7–15.
Satake, M., Mackenzie, L., Yasumoto, T., 1997. Identification of
Protoceratium reticulatum as the biogenetic origin of yessotoxin.
Natural Toxins 5 (4), 164–167.
Satake, M., Ichimura, T., Sekiguchi, K., Yoshimatsu, S., Oshima, Y.,
1999. Confirmation of yessotoxin and 45, 46, 47trinoryessotoxin
production by Protoceratium reticulatum
collected in Japan. Natural Toxins 7 (4), 147–150.
Schantz, E.J., Mold, J.D., Stanger, D.W., Shavel, J., Riel, F.J.,
Bowden, J.P., Lynch, J.M., Wyler, R.S., Riegel, B., Sommer, H.,
1957. Paralytic shellfish poison. VI. A Procedure for the
isolation and purification of the poison from toxic clam and
mussel tissues. Journal of American Chemical Society 79 (19),
5230–5235.
Schnetzer, A., Miller, P.E., Schaffner, R.A., Stauffer, B.A., Jones, B.H.,
Weisberg, S.B., DiGiacomo, P.M., Berelson, W.M., Caron, D.A.,
2007. Blooms of Pseudo-nitzschia and domoic acid in the San
Pedro Channel and Los Angeles harbor areas of the Southern
California Bight, 2003-2004. Harmful Algae 6 (3), 372–387.
Schofield, O., Kohut, J., Aragon, D., Creed, L., Graver, J.,
Haldeman, C., Kerfoot, J., Roarty, H., Jones, C., Webb, D.,
Glenn, S., 2007. Slocum gliders: robust and ready. Journal of
Field Robotics 24 (6), 473–485.
Scholin, C.A., Herzog, M., Sogin, M., Anderson, D.M., 1994.
Identification of group-and strain-specific genetic markers for
globally distributed Alexandrium (Dinophyceae). Part II.
Sequence analysis of a fragment of the LSU rRNA gene. Journal
of Phycology 30 (6), 999–1011.
Scholin, C.A., Gulland, F., Doucette, G.J., Benson, S., Busman, M.,
Chavez, F.P., Cordaro, J., DeLong, R., De Vogelaere, A.,
Harvey, J., Haulena, M., Lefebvre, K.A., Lipscomb, T.,
Loscutoff, S., Lowenstine, L.J., Marin 3rd, R., Miller, P.E.,
McLellan, W.A., Moeller, P.D., Powell, C.L., Rowles, T.,
Silvagni, P., Silver, M., Spraker, T., Trainer, V.L., Van
Dolah, F.M., 2000. Mortality of sea lions along the central
California coast linked to a toxic diatom bloom. Nature 403
(6765), 80–84.
Separation Processes Inc., 2005. Investigation of Microfiltration
and Reverse Osmosis Seawater Desalination. National Water
Research Institute.
Shimizu, Y., Hsu, C.P., Genenah, A.A., 1981. Structure of saxitoxin
in solutions and stereochemistry of dihydrosaxitoxins. Journal
of American Chemical Society 103, 605–609.
Shipe, R.F., Leinweber, A., Gruber, N., 2008. Abiotic controls of
potentially harmful algal blooms in Santa Monica Bay,
California. Continental Shelf Research 28 (18), 2584–2593.
Shumway, S.E., Allen, S.M., Boersma, P.D., 2003. Marine birds and
harmful algal blooms: sporadic victims or under-reported
events? Harmful Algae 2 (1), 1–17.
Smayda, T.J., 1990. Novel and nuisance phytoplankton blooms in
the sea: evidence for a global epidemic. In: Granéli, E.,
Gundström, B., Edler, L., Anderson, D.M. (Eds.), Toxic Marine
Phytoplankton. Elsevier, New York, pp. 29–40.
Smayda, T.J., 2008. Complexity in the eutrophication–harmful
algal bloom relationship, with comment on the importance of
grazing. Harmful Algae 8 (1), 140–151.

Smith, G., Ladizinsky, N., Miller, P., 2001. Amino acid profiles in
species and strains of Pseudo-nitzschia from Monterey Bay,
California: insights into the metabolic role(s) of domoic acid.
In: Hallegraef, G., Blackburn, S., Bolch, C., Lewis, R. (Eds.). IOC,
UNESCO, Hobart, Tasmania, pp. 324–327.
Sommer, H., Meyer, K.F., 1937. Paralytic shellfish poisoning.
Archives of Pathology 24, 560–598.
Steidinger, K.A., 1993. Some taxonomic and biologic aspects of
toxic dinoflagellates. In: Falconer, I. (Ed.), Algal Toxins in
Seafood and Drinking Water. Academic Press, London, pp. 1–28.
Stobo, L.A., Lewis, J., Quilliam, M.A., Hardstaff, W.R., Gallacher, S.,
Webster, L., Smith, E.A., McKenzie, M., 2003. Detection of
yessotoxin in UK and Canadian isolates of phytoplankton and
optimization and validation of LC–MS methods. Canadian
Technical Report of Fisheries and Aquatic Sciences 2498, 8–14.
Strom, S., Wolfe, G.V., Slajer, A., Lambert, S., Clough, J., 2003.
Chemical defenses in the microplankton II: inhibition of
protist feeding by B-dimethylsulfoniopropionate (DMSP).
Limnology and Oceanography 48 (1), 230–237.
Su, Z., Sheets, M., Ishida, H., Li, F., Barry, W.H., 2004. Saxitoxin
blocks L-type ICa. Journal of Pharmacology and Experimental
Therapeutics 308 (1), 324–329.
Subba Rao, D.V., Quilliam, M.A., Pocklington, R., 1988. Domoic
acid – a neurotoxic amino acid produced by the marine diatom
Nitzschia pungens in culture. Canadian Journal of Fisheries and
Aquatic Sciences 45 (12), 2076–2079.
Suganuma, M., Fujiki, H., Suguri, H., Yoshizawa, S., Hirota, M.,
Nakayasu, M., Ojika, M., Wakamatsu, K., Yamada, K.,
Sugimura, T., 1988. Okadaic acid: an additional non-phorbol-
12-tetradecanoate-13-acetate-type tumor promoter.
Proceedings of the National Academy of Sciences, United
States of America 85 (6), 1768–1771.
Sukhatme, G.A.D., Zhang, B., Oberg, C., Stauffer, B.A., Caron, D.A.,
2007. The design and development of a wireless robotic
networked aquatic microbial observing system.
Environmental Engineering Science 24, 205–215.
Sunda, W., 2006. Trace metals and harmful algal blooms. In:
Granéli, E., Turner, J.T. (Eds.), Ecology of Harmful Algae.
Springer, Berlin, pp. 203–214.
Sutherland, C.M., 2008. Diarrhetic Shellfish Toxins Linked to Local
Dinophysis Populations in the California Coastal Waters of
Monterey Bay. Dissertation, University of California, Santa
Cruz, Santa Cruz.
Tachibana, K., Scheuer, P.J., Tsukitani, Y., Kikuchi, H., Engen, D.V.,
Clardy, J., Gopichand, Y., Schmitz, F.J., 1981. Okadaic acid,
a cytotoxic polyether from two marine sponges of the genus
Halichondria. Journal of American Chemical Society 103 (9),
2469–2471.
Takahashi, H., Kusumi, T., Kan, Y., Satake, M., Yasumoto, T., 1996.
Determination of the absolute configuration of yessotoxin,
a polyether compound implicated in diarrhetic shellfish
poisoning, by NMR spectroscopic method using a chiral
anisotropic reagent, methoxy-(2-naphthyl) acetic acid.
Tetrahedron Letters 37 (39), 7087–7090.
Takai, A., Bialojan, C., Troschka, M., Rüegg, J.C., 1987. Smooth
muscle myosin phosphatase inhibition and force
enhancement by black sponge toxin. FEBS Letters 217 (1),
81–84.
Taylor, F.J.R., Horner, R.A., 1994. Red tides and other problems
with harmful algal blooms in Pacific Northwest coastal waters.
In: Review of the Marine Environment and Biota of Strait of
Georgia, Puget Sound and Juan de Fuca Strait. Canadian
Fisheries and Aquatic Science Technical Report 1948,
pp. 175–186.
Tenzer, B., Adin, A., Priel, M., 1999. Seawater filtration for fouling
prevention under stormy conditions. Desalination 125 (1–3),
77–88.
water research 44 (2010) 385–416 415
Terao, K., Ito, E., Oarada, M., Murata, M., Yasumoto, T., 1990.
Histopathological studies on experimental marine toxin
poisoning: the effects in mice of yessotoxin isolated from
Patinopecten yessoensis and of a desulfated derivative. Toxicon
28 (9), 1095–1104.
Thomas, W.H., Zavoico, T., Hewes, C., 2001. Historical
Phytoplankton Species Time-Series Data (1917–1939) from the
North American Pacific Coast. Scripps Institution of
Oceanography. Ref. no. 01-12.
Throndsen, J., 1997. The plankton marine flagellates. In:
Tomas, C.R. (Ed.), Identifying marine phytoplankton.
Academic Press, San Diego, pp. 591–729.
Trainer, V.L., 2002. Harmful algal blooms on the U.S. west coast.
In: Taylor, F.J., Trainer, V.L. (Eds.), Harmful Algal Blooms in the
PICES Region of the North Pacific, pp. 89–118. PICES Scientific
Report 23.
Trainer, V.L., Adams, N.G., Bill, B.D., Anulacion, F., Wekell, J.C.,
1998. Concentration and dispersal of a Pseudo-nitzschia bloom
in Penn Cove, Washington, USA. Natural Toxins 6 (3–4),
113–126.
Trainer, V.L., Adams, N.G., Bill, B.D., Stehr, C.M., Wekell, J.C.,
Moeller, P., Busman, M., Woodruff, D., 2000. Domoic acid
production near California coastal upwelling zones, June 1998.
Limnology and Oceanography 45 (8), 1818–1833.
Trainer, V.L., Adams, N.G., Wekell, J.C., 2001. Domoic acid
producing Pseudo-nitzschia species off the U.S. west coast
associated with toxification events. In: Hallegraeff, G.M.,
Blackburn, S.I., Bolch, C.J., Lewis, R.J. (Eds.), Harmful Algal
Blooms 2000. Intergovernmental Oceanographic Commission
of UNESCO, Paris, pp. 46–49.
Trainer, V.L., Hickey, B.M., Horner, R.A., 2002. Biological and
physical dynamics of domoic acid production off the
Washington Coast. Limnology and Oceanography 47 (5),
1438–1446.
Trainer, V.L., Eberhart, B.T.L., Wekell, J.C., Adams, N.G.,
Hanson, L., Cox, F., Dowell, J., 2003. Paralytic shellfish toxins in
Puget Sound, Washington State. Journal of Shellfish Research
22 (1), 213–223.
Trainer, V.L., Cochlan, W.P., Erickson, A., Bill, B.D., Cox, F.H.,
Borchert, J.A., Lefebvre, K., 2007. Recent domoic acid closures
of shellfish harvest areas in Washington State inland
waterways. Harmful Algae 6 (3), 449–459.
Tubaro, A., Sidari, L., Della Loggia, R., Yasumoto, T., 1998.
Occurrence of yessotoxin in phytoplankton and mussels from
Northern Adriatic Sea. In: Proceedings of the VIII International
Conference on Harmful Algae, Vigo, Spain (abstract).
Twiner, M.J., Dechraoui, M.Y.B., Wang, Z., Mikulski, C.M., Henry, M.
S., Pierce, R.H., Doucette, G.J., 2007. Extraction and analysis of
lipophilic brevetoxins from the red tide dinoflagellate Karenia
brevis. Analytical Biochemistry 369 (1), 128–135.
Tyrell, J.V., Connell, L.B., Scholin, C.A., 2002. Monitoring for
Heterosigma akashiwo using a sandwich hybridization assay.
Harmful Algae 1 (2), 205–214.
Van Dolah, F.M., Finley, E.L., Haynes, B.L., Doucette, G.J.,
Moeller, P.D., Ramsdell, J.S., 1994. Development of rapid and
sensitive high throughput pharmacologic assays for marine
phycotoxins. Natural Toxins 2 (4), 189–196.
Vigilant, V.L., Silver, M., 2007. Domoic acid in benthic flatfish on
the continental shelf of Monterey Bay, California, USA. Marine
Biology 151 (6), 2053–2062.
Walz, P.M., Garrison, D.L., Graham, W.M., Cattey, M.A.,
Tjeerdema, R.S., Silver, M.W., 1994. Domoic acid-producing
diatom blooms in Monterey Bay, California: 1991–1993.
Natural Toxins 2 (5), 271–279.
Wang, J., Salata, J.J., Bennett, P.B., 2003. Saxitoxin is a gating
modifier of hERG Kþ channels. Journal of General Physiology
121 (6), 583–598.

416
Watson, M., 1997. Control of Membrane Fouling Due to Algae.
Applied Research Department. Technology Transfer Note.
No.12.
Wekell, J.C., Gauglitz, E.J.J., Barnett, H.J., Hatfield, C.L., Simons, D.,
Ayres, D., 1994. Occurrence of domoic acid in Washington
state razor clams (Siliqua patula) during 1991–1993. Natural
Toxins 2 (4), 197–205.
Wells, M.L., Trick, C.G., Cochlan, W.P., Hughes, M.P., Trainer, V.L.,
2005. Domoic acid: the synergy of iron, copper, and the
toxicity of diatoms. Limnology and Oceanography 50 (6),
1908–1917.
Wilf, M., Schierach, M.K., 2001. Improved performance and cost
reduction of RO seawater systems using UF pretreatment.
Desalination 135 (1–3), 61–68.
Wong, J.L., Oesterlin, R., Rapoport, H., 1971. Struture of
saxitoxin. Journal of American Chemical Society 93 (26),
7344–7345.
Wright, J.L.C., Cembella, A.D., 1998. Ecophysiology and
biosynthesis of polyether marine biotoxins. In: Anderson, D.M.,
Cembella, A.D., Hallegraeff, G.M. (Eds.), The Physiological
water research 44 (2010) 385–416
Ecology of Harmful Algal Blooms. Springer-Verlag, Heidelberg,
pp. 427–451.
Wright, J.L.C., Bird, C., de Freitas, A., Hampson, D., McDonald, J.,
Quilliam, M., 1990. Chemistry, biology, and toxicology of
domoic acid and its isomers. Canadian Diseases Weekly
Report 16 (Suppl. 1E), 21–26.
Yasumoto, T., Murata, M., 1993. Marine toxins. Chemical Reviews
93, 1867–1909.
Yasumoto, T., Takizawa, A., 1997. Fluorometric measurement of
yessotoxins in shellfish by high-pressure liquid
chromatography. Bioscience, Biotechnology and Biochemistry
61 (10), 1775–1777.
Yasumoto, T., Oshima, Y., Sugawara, W., Fukuyo, Y., Oguri, H.,
Igarashi, T., Fujita, N., 1980. Identification of Dinophysis fortii as
the causative organism of diarrhetic shellfish poisoning.
Bulletin of the Japanese Society of Scientific Fisheries 46 (11),
1405–1411.
Yasumoto, T., Murata, M., Oshima, Y., Sano, M., Matsumoto, G.K.,
Clardy, J., 1985. Diarrhetic shellfish toxins. Tetrahedron 41 (6),
1019–1025.

Occurrence and removal of pharmaceuticals, caffeine and
DEET in wastewater treatment plants of Beijing, China
Qian Sui, Jun Huang, Shubo Deng, Gang Yu*, Qing Fan
POPs Research Centre, Department of Environmental Science & Engineering, Tsinghua University, Beijing 10084, China
article info
Article history:
Received 14 January 2009
Received in revised form
5 July 2009
Accepted 8 July 2009
Available online 15 July 2009
Keywords:
Pharmaceuticals
Wastewater
Removal efficiency
Advanced treatment
Risk assessment
China
1. Introduction
abstract
With the progress of sensitive analytical techniques, the
frequent detection of various pharmaceuticals in the aquatic
environment has received global concerns of both the
academic community and the public (Daughton and Ternes,
1999; Jones et al., 2005). After intake by humans or animals,
the pharmaceuticals will be partially converted to metabolites,
however, partially excreted unchanged or as conjugates,
and finally delivered to the wastewater treatment plants
(WWTPs). As there is no unit specifically designed to remove
these compounds, the elimination by most WWTPs seems to
be inefficient (Ternes, 1998; Castiglioni et al., 2006; Lishman
et al., 2006; Nakada et al., 2006; Santos et al., 2007; Vieno et al.,
2007b; Xu et al., 2007; Gulkowska et al., 2008; Paxeus, 2004).
Together with treated wastewater, these compounds are
water research 44 (2010) 417–426
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
* Corresponding author. Tel.: þ86 10 62787137; fax: þ86 10 62794006.
E-mail address: yg-den@tsinghua.edu.cn (G. Yu).
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.watres.2009.07.010
The occurrence and removal of 13 pharmaceuticals and 2 consumer products, including
antibiotic, antilipidemic, anti-inflammatory, anti-hypertensive, anticonvulsant, stimulant,
insect repellent and antipsychotic, were investigated in four wastewater treatment plants
(WWTPs) of Beijing, China. The compounds were extracted from wastewater samples by
solid-phase extraction (SPE) and analyzed by ultra-performance liquid chromatography
coupled with tandem mass spectrometry (UPLC–MS/MS). Most of the target compounds were
detected, with the concentrations of 4.4 ng L 1 –6.6 mgL 1 and 2.2–320 ng L 1 in the influents
and secondary effluents, respectively. These concentrations were consistent with their
consumptions in China, and much lower than those reported in the USA and Europe. Most
compounds were hardly removed in the primary treatment, while their removal rates ranging
from 12% to 100% were achieved during the secondary treatment. In the tertiary treatment,
different processes showed discrepant performances. The target compounds could not be
eliminated by sand filtration, but the ozonation and microfiltration/reverse osmosis (MF/RO)
processes employed in two WWTPs were very effective to remove them, showing their main
contributions to the removal of such micro-pollutants in wastewater treatment.
ª 2009 Elsevier Ltd. All rights reserved.
released to the aquatic environment, and consequently found
to contaminate the receiving water bodies (Lindqvist et al.,
2005; Kasprzyk-Hordern et al., 2009), or even raw water sources
of drinking water treatment plant (Ternes et al., 2002;
Vieno et al., 2007a; Radjenovic et al., 2008). Meanwhile, results
of toxicology studies have revealed that some pharmaceuticals
are suspected to have direct toxicity to certain aquatic
organisms (Ferrari et al., 2003; Jjemba, 2006; Grung et al., 2008;
Quinn et al., 2008). Besides, their continual but undetectable
effects could accumulate slowly, and finally lead to irreversible
change on wildlife and human beings (Daughton and
Ternes, 1999). Therefore, the occurrence and behavior of
pharmaceuticals in the WWTPs, which are both the sink and
source of the compounds, should be focused on. So far,
concentrations of pharmaceuticals from various therapeutic
classes in the WWTPs have been well documented in the

418
North America (Thomas and Foster, 2005; Lishman et al.,
2006), Japan (Nakada et al., 2006) and some European countries
(Ternes, 1998; Castiglioni et al., 2006; Santos et al., 2007; Vieno
et al., 2007b; Jones et al., 2007; Paxeus, 2004). Reported species
and concentrations of pharmaceuticals varied from country to
country, and plant to plant, owing to the different usage
patterns. Meanwhile, the removal efficiencies of pharmaceuticals
also varied much (Nakada et al., 2006; Gulkowska et al.,
2008), indicating that the removal could be affected by both
the compound-specific properties, and the factors concerning
specific WWTPs, such as types of treatment processes, solids
retention time (SRT), hydraulic retention time (HRT),
temperature, etc. In recent years, very few studies about the
situation in China have been reported. Only one specific
therapeutic class, antibiotics, has been investigated by limited
previous studies (Xu et al., 2007; Gulkowska et al., 2008; Chen
et al., 2008). Therefore, it is necessary and important to
investigate the occurrence and removal of pharmaceuticals
from different therapeutic classes in the WWTPs of China.
Due to the low efficiency of conventional wastewater
treatment processes, some advanced treatment technologies
have been evaluated. Ozonation was found to be effective to
remove pharmaceuticals in real municipal WWTPs of Japan
(Nakada et al., 2007; Okuda et al., 2008) and Germany (Ternes
et al., 2003). Nanofiltration (NF) and reverse osmosis (RO)
membrane filtration, the well-proven technologies to remove
pharmaceuticals from different kinds of waters, have also been
applied at bench, pilot and full scale (Khan et al., 2004; Nghiem
et al., 2005; Drewes et al., 2005; Al-Rifai et al., 2007; Watkinson
et al., 2007; Comerton et al., 2008; Radjenovic et al., 2008).
Retention behavior of pharmaceuticals during the processes
associated with physicochemical properties of pharmaceuticals,
membranes as well as the solution chemistry, and
mechanisms of pharmaceutical rejection have been discussed
in Kimura et al. (2004), Nghiem et al. (2005), Nghiem and
Coleman (2008) and Comerton et al. (2008). Recently, considering
the requirement of reclaimed water, several advanced
treatment facilities have been installed in the WWTPs of
Beijing. However, the removal efficiency of micro-pollutants,
such as pharmaceuticals, has not been evaluated yet.
In the present study, we investigated the contamination
levels of 13 pharmaceuticals and 2 consumer products from 8
classes (i.e. antibiotic, antilipidemic, anti-inflammatory, antihypertensive,
anticonvulsant, stimulant, insect repellent and
antipsychotic) in four WWTPs of Beijing, China, which have
different advanced treatment units, and evaluated the elimination
efficiencies of the target pharmaceuticals. To the best
of our knowledge, this is the first report on the occurrence and
removal of pharmaceuticals and consumer products from
multiple classes in the WWTPs of China, especially for the
situation during the advanced treatment processes.
2. Materials and methods
2.1. Chemicals
All the standards including chloramphenicol (CP), nalidixic
acid (NA), trimethoprim (TP), bezafibrate (BF), clofibric acid
(CA), gemfibrozil (GF), diclofenac (DF), indometacin (IM),
water research 44 (2010) 417–426
ketoprofen (KP), mefenamic acid (MA), metoprolol (MTP),
carbamazepine (CBZ), caffeine (CF), N,N-diethyl-meta-toluamide
(DEET) and sulpiride (SP) (Appendix) were of analytical
grade (>90%), and purchased from Sigma–Aldrich (Steinheim,
Germany). Isotopically labeled compounds, used as internal
standards, were 13 C-phenacetin obtained from Sigma–
Aldrich, and 3 D-mecoprop from Dr. Ehrenstorfer (Augsburg,
Germany). HPLC grade methanol, acetone, dichloromethane,
hexane, as well as formic acid were provided by Dikma (USA),
and ultra-pure water was produced by a Milli-Q unit (Millipore,
USA). Stock solutions of individual compound were prepared in
methanol and mixture standards with different concentrations
were prepared by diluting the stock solutions before each
analytical run. All the solutions were stored at 4 C in the dark.
2.2. Sample collection
Four full-scale municipal WWTPs, referred as A, B, C and D,
were selected in our study. These WWTPs employ similar
conventional treatment processes: primary treatment to
remove particles coupled with secondary biological treatment.
For the secondary biological treatment processes,
WWTPs A and D employ anaerobic/anoxic/oxic (A 2 /O) activated
sludge process, anoxic/oxic (A/O) activated sludge
process is adopted in WWTP B, and WWTP C employs oxidation
ditch (OD). Other detailed information on each WWTP,
such as inhabitants served, daily flow, HRT and SRT are shown
in Table 1. Part of the secondary effluents was further treated
in WWTPs A, B and D, by the processes of ultrafiltration
(UF)/ozone, sand filtration (SF) and microfiltration/reverse
osmosis (MF/RO), respectively. In WWTP A, a dead-end
ultrafiltration system (Zenon GE) is used. The whole system
has 6 trains of Zee-Weed 1000 membrane. Each train contains
9 cassettes of 57–60 modules per cassette. The membrane,
with the pore size of 0.02 mm, is made by PVDF. The module is
operated in an outside/in configuration at a constant flow of
23 L (m 2 h) 1 and the total treatment capacity reaches
80,000 m 3 d 1 . The membrane is hydraulically backwashed at
a constant flow rate of 34 (m 2 h) 1 , and 29 times per day. The
backwash phase lasts for 1 min. Maintenance cleaning is
conducted once per day. Membranes are soaked in the sodium
hypochlorite solution (50 mg L 1 ) for 25 min. For the ozonation
process, gaseous ozone is generated from an ozone generator
(Mitsubishi Electric). The ozone dosage and contact time in the
reaction tank is 5 mg L 1 and 15 min, respectively. The pH of
the wastewater before ozonation ranges 6.5–8.0 and shows no
significant change after ozonation. As the heart of the
advanced treatment in WWTP D, a spiral-wound crossflow
module is employed for the reverse osmosis (RO) membrane
filtration. The RO membrane (Filmtec, DOW) is made from
a thin-film composite polyamide material. Each module is
designed to operate at a water flux of 1.3 m 3 h 1 , and a product
water recovery of 75–80%. The trans-membrane pressure is
between 0.04 and 0.06 MPa, and the salt rejection remained at
the level of 99%. Every 3–6 months, normally when the transmembrane
pressure reaches above 0.06 MPa, the membrane is
cleaned with 0.1%(w) sodium hydroxide solution (for organic
foulants), 2%(w) citric acid (for inorganic foulants) and 0.5%(w)
formaldehyde (as biocide). Schematic diagram of treatment
processes in the four WWTPs is shown in Fig. 1.

The samples were collected once from the four WWTPs
during June and July 2008, with no compensation for HRT. All
of them were collected as grab samples in duplicate (500 mL
for influents and 1000 mL for the others) in prewashed amber
glass bottles, kept in the cooler and transported to the laboratory.
Immediately after delivery to the laboratory, they were
filtered through prebaked (400 C, >4 h) glass microfiber filters
(GF/F, Whatman) to remove particles and stored at 4 C before
extraction.
2.3. Sample extraction and analysis
The method for the extraction and analysis of pharmaceuticals
and consumer products is presented elsewhere (Sui et al.,
in press) and briefly described here. After the solid-phase
extraction (SPE) cartridges (Oasis, HLB, 200 mg, 6 mL) were
a WWTP A
Influent
Screen
b WWTP B
Influent
Screen
c WWTP C
Influent
Screen
d WWTP D
Influent
Screen
Grit
Removal
Grit
Removal
Grit
Removal
Grit
Removal
water research 44 (2010) 417–426 419
Table 1 – Information of the WWTPs investigated.
WWTP Inhabitants
served 10 3
Daily flow
( 10 3 m 3 HRT (h) SRT (d) Secondary treatment Tertiary treatment
)
A 814 400 11 12–15 A 2 /O UF/ozone
B 2400 1000 11 20 A/O SF
C 480 200 15 12–16 OD –
D 2415 600 10.67 15 A 2 /O MF/RO
A/O
treatment
Primary
Clarifier
Primary
Clarifier
Oxidation
Ditch
conditioned, wastewater samples, added with internal standards
and adjusted to pH ¼ 7, were introduced to the cartridge
via a PTFE tube, at a flow rate of 5–10 mL min 1 . After washing
by 5 mL of 5.0% methanol solution, the cartridge was dried
under vacuum for 2 h and eluted with 5 mL of methanol. The
extract was then concentrated to 0.4 mL under a gentle
nitrogen stream and stored at 4 C for analysis. Concentrations
of the target compounds were analyzed using ultraperformance
liquid chromatography coupled with tandem
mass spectrometry (UPLC–MS/MS). Analytes were separated
using Waters Acquity UPLC system (Waters Corporation, USA)
equipped with Acquity UPLC BEH C18 column (50 2.1 mm,
particle size of 1.7 mm), and detected by Quattro Premier XE
tandem quadrupole mass spectrometry (Waters Corp., USA)
equipped with an electrospray ionization source. The analysis
was carried out in multiple reaction monitoring (MRM) mode,
Secondary
Clarifier
A 2 /O
treatment
A 2 /O
treatment
Secondary
Clarifier
Secondary
Clarifier
Ultrafiltration
Secondary Effluent
Secondary
Clarifier
Secondary
Effluent
Ozonation
Sand Filtration
Secondary Effluent
MF
`
RO
Secondary Effluent
Tertiary
Effluent
Tertiary
Effluent
Tertiary
Effluent
Fig. 1 – Schematic diagram of the treatment processes in the four WWTPs and sampling site location (C).

420
and in general, two precursor ion/product ion transitions were
monitored for one compound with the purpose of quantification
and confirmation.
2.4. Quality control
For each sampling, 500 mL Milli-Q water in an amber glass
bottle as a field blank was brought to the WWTPs, exposed to
the environment where the samples were taken from, and
then delivered back to the laboratory with samples. For each
set of samples (normally 10 samples), at least one procedural
blank was prepared from ultra-pure water in the laboratory.
Both the field blanks and procedural blanks were run identically
to the wastewater samples, and the concentrations of
target compounds were below the limit of quantification
(LOQ). The absolute recoveries, calculated by comparing the
concentrations of target compounds in spiked and unspiked
wastewaters, were proved to be 73–102% and 50–95% in the
effluent and influent for most compounds, respectively. While
for several compounds (i.e. sulpiride, gemfibrozil, mefenamic
acid), the absolute recoveries were not satisfactory. However,
13 C-phenacetin and 3 D-mecoprop, the surrogate standards
used for positive and negative ion mode respectively were able
to compensate for the loss of most analytes, and relative
recoveries were 67–130% for all the analytes in the effluent
and 79–140% in the influents except mefenamic acid (251%)
and nalidixic acid (178%). Therefore, the concentrations of
these two compounds in the wastewater influents were not
quantitatively determined and reported. The LOQs were
0.3–5.5 ng L 1 and 0.7–20 ng L 1 in the effluent and influent,
respectively. Detailed information about the calibration,
recoveries, LOQ, matrix effects, etc. were described in Sui et al.
(in press), and briefly listed in Table 2. As duplicate samples
were collected at each sampling site, mean concentrations
were adopted. In most cases, deviations of duplicate samples
were less than 20%. For some tertiary effluent samples, low
concentrations of some target compounds (i.e. caffeine, DEET,
carbamazepine) resulted in slightly higher deviations.
3. Result and discussion
3.1. Influents
As shown in Fig. 2, 12 target compounds were detected in all
the influent samples from the four WWTPs, while ketoprofen
was below LOQ in all wastewater samples. The most
abundant compounds detected were the consumer products,
caffeine (3.4–6.6 mgL 1 ) and N,N-diethyl-meta-toluamide
(0.6–1.2 mgL 1 ), probably due to the large consumption of
drinks containing caffeine (i.e. coffee, tea, etc.) and wide
application of insect repellent during the summer time when
we sampled. Diclofenac, trimethoprim, sulpiride, carbamazepine,
indometacin and metoprolol showed relatively high
concentrations (Fig. 2). A similar composition distribution
was observed among all the influents of the four WWTPs.
The concentrations of target pharmaceuticals except
diclofenac and trimethoprim, were much lower than those
reported in the European and North American countries
(Thomas and Foster, 2005; Lishman et al., 2006; Vanderford
and Snyder, 2006; Santos et al., 2007; Gomez et al., 2007; Vieno
et al., 2007b; Huerta-Fontela et al., 2008). For instance, the
concentrations of ketoprofen in the wastewater influents
were recorded to be 2.0 0.6 mgL 1 in Finland (Lindqvist et al.,
2005), 200 ng L 1 in Australia (Al-Rifai et al., 2007), and
300–1360 ng L 1 in Spain (Santos et al., 2007), while in the
influents of four WWTPs in Beijing, it could not be detected.
Concerning gemfibrozil, which is used to lower cholesterol
and triglyceride levels in the blood, the contamination level
found in the present study was 24–140 ng L 1 , even 1 or 2 order
of magnitude lower than those in the USA (4770 ng L 1 ,
Vanderford and Snyder, 2006) and Canada (418 ng L 1 ,
Table 2 – Instrumental quantification limit (IQL), limit of quantification (LOQ), absolute recovery (AR), relative recovery (RR)
and matrix suppression of target compounds.
Compounds IQL (pg) LOQ (ng L 1 ) AR (n¼6, %) RR (n ¼ 6, %) Matrix effect (%)
Effluent Influent Effluent a
Influent a
Effluent a
Influent a
BF 0.5 0.3 0.7 74 (3) 58 (5) 94 (4) 94 (10) 27
CA 10 5.4 16 74 (4) 50 (4) 94 (6) 82 (8) 30
CBZ 2.5 1.0 2.8 100 (4) 71 (12) 130 (6) 133 (23) 6
CF 2.5 1.7 3.3 60 (3) 61 (37) 77 (4) 114 (70) 49
CP 2.5 1.0 2.3 98 (7) 86 (9) 124 (10) 140 (17) 39
DEET 2.5 1.3 3.2 76 (3) 63 (6) 99 (5) 118 (14) 24
DF 10 4.7 9.4 86 (12) 85 (19) 109 (16) 139 (32) 4
GF 10 4.2 20 95 (12) 40 (11) 120 (16) 65 (18) 28
IM 2.5 1.3 2.8 80 (12) 73 (13) 101 (16) 119 (22) 1
KP 10 5.5 18 69 (9) 44 (3) 87 (12) 72 (7) 46
MA 10 5.5 5.2 73 (11) 154 (24) 92 (15) 251 (41) 13
MTP 2.5 1.1 3.3 88 (3) 60 (3) 115 (5) 113 (9) 15
NA 2.5 1.1 2.1 91 (9) 95 (9) 118 (12) 178 (20) 3
SP 0.5 0.4 1.0 51 (5) 42 (3) 67 (7) 79 (7) 63
TP 2.5 1.0 2.7 102 (7) 74 (16) 132 (9) 139 (31) 5
a Value in the brackets refers to the deviation of the recovery.
water research 44 (2010) 417–426

a
Concentration (ng L -1 )
b
Concentration (ng L -1 )
10000
1000
100
10
1
10000
1000
100
10
1
DEET
DEET
CF
CF
CP
TP
BF
CA
Lishman et al., 2006). The low levels of target pharmaceuticals
were probably due to the lower per capita consumption in
China than in the countries with higher socioeconomic
statuses, where medical care is more prevalent (Thomas and
Foster, 2005). The per capita consumption rate of gemfibrozil
in China is estimated to be 0.036 mg person 1 d 1 (Table 3),
lower than those in Germany (0.2 mg person 1 d 1 , Ternes,
1998) and Canada (0.2 mg person 1 d 1 , Lishman et al., 2006).
Since the levels of target pharmaceuticals were somewhat
different from those of European and North American
countries, we theoretically calculate the concentration of
pharmaceuticals in the wastewater influent by the following
equation (Lindqvist et al., 2005; Nakada et al., 2006)
T e% I 1012
Cpred ¼
365 P Q
where C pred is the predicted concentration of the pharmaceutical
in wastewater influent (ng L 1 ); T is the total production
of a pharmaceutical both for human and animal use in
China per year (ton year 1 ), P is the population of China, e% is
the amount of the pharmaceutical excreted unchanged, I is
the number of inhabitants served and Q is the influent flow
(m 3 d 1 ). The predicted concentrations of gemfibrozil, diclofenac,
indometacin, ketoprofen, carbamazepine, and sulpiride
were comparable to those measured in the influents
(Table 3). Much lower measured concentration than predicted
concentration of chloramphenicol was probably because it
had been forbidden for use in food and aquaculture in China
since 2005, and the available data about the production of
pharmaceuticals were based on the year of 2004. It should be
noticed that since there is no available data on the total
consumption of any pharmaceutical, we used figures for total
production of individual pharmaceutical instead. Therefore,
the differences between the amounts actually produced and
applied as well as the amount used in human and veterinary
medicine could not be distinguished, which might result in
overestimation of the theoretical concentration. Nevertheless,
the comparability between the predicted concentrations
and measured concentrations illustrates the overall reasonability
of the approach.
Table 3 – Outputs, per capita consumption, predicted concentrations (PECs) and measured concentrations (MECs) of some
pharmaceuticals in the wastewater influents of WWTPs investigated.
Compound Output a
(tons year 1 )
GF
DF
IM
MTP
Pharmaceuticals & Consumer Products
CP
TP
BF
CA
GF
DF
IM
MA
MTP
Influents
CBZ
CBZ
SP
Secondary effluents
Pharmaceuticals & Consumer Products
Fig. 2 – Concentrations of target pharmaceuticals in
wastewater influents (a) and secondary effluents (b) of four
WWTPs in Beijing.
water research 44 (2010) 417–426 421
SP
Per capita
consumption
(mg person 1 d 1 )
Excreted
unchanged b (%)
PEC
(ng L 1 )
(1)
MEC
(mean, ng L 1 )
CP 1929 4.051 5–10 844 31
TP 2352 4.939 45 6173 400
GF 17 0.036 76 76 60
DF 328 0.689 15 287 318
IM 277 0.582 10–20 242 129
KP 92 0.193 2.7 15 n.d. c
CBZ 395 0.830 2–3 58 113
SP 98 0.206 15–70 243 157
a From CMEIN (2005).
b From Bolton and Null (1981), Ternes (1998), Khan and Ongerth (2004), Niwa et al. (2005), Nakada et al. (2006), Jjemba (2006).
c n.d. ¼ Not detected.

422
3.2. Secondary effluent
Similar to the influent samples, ketoprofen was below the LOQ
in all the secondary effluent samples. Nalidixic acid and
chloramphenicol were detected only in one WWTP, with the
concentration of 8.1 and 19 ng L 1 , respectively. The mean
concentrations of the other 12 compounds ranged from 5 to
200 ng L 1 (Fig. 2). Diclofenac, N,N-diethyl-meta-toluamide,
trimethoprim, sulpiride and indometacin showed high
concentrations in the secondary effluents. Carbamazepine
and metoprolol followed, with the concentrations ranging
from 69 to 120 ng L 1 , and 60 to 108 ng L 1 , respectively. Other
compounds, such as caffeine, gemfibrozil and mefenamic
acid, occurred at the lowest levels. Despite of a wide variation
of trimethoprim from different WWTPs, the composition
profiles of target pharmaceuticals in secondary effluents from
the four WWTPs were quite similar (Fig. 3).
The concentration levels of most pharmaceuticals and
consumer products detected in the secondary effluent were
also lower than those reported in the Europe. They were over
100 ng L 1 , in some cases even up to 500 ng L 1 in the wastewater
effluents of the European countries (Santos et al., 2007;
Gomez et al., 2007; Ternes, 1998; Vieno et al., 2007b). While in
the present study, 10 out of 15 compounds were less than
100 ng L 1 , and none of them exceeded 400 ng L 1 in any
effluent samples (Fig. 2). Our results were in agreement with
those in Japan (Nakada et al., 2006), Korea (Kim et al., 2007) and
some other cities of China (Xu et al., 2007; Gulkowska et al.,
2008; Chen et al., 2008). For instance, the concentrations of
chloramphenicol in the effluents of 4 WWTPs in Guangzhou
were

et al., 2008; Gomez et al., 2007). The removal rate of bezafibrate
was found to be 87% in six Italian WWTPs in summer time
(Castiglioni et al., 2006), very similar to that observed in our
study. The concentrations of DEET were decreased by more
than 80% during the biological treatment in the WWTP of
Japan (Okuda et al., 2008), slightly better than our results. A
second group of pharmaceuticals, including three antiinflammatory
drugs, clofibric acid, gembrozil, metoprolol and
sulpiride, had lower removal rates with large variation in
different WWTPs studied. For instance, 28–53% of diclofenac,
a representative of the anti-inflammatory drugs, was removed
by secondary treatment in the WWTPs, which was between
26% in Finland (Lindqvist et al., 2005) and 69% in Germany
(Ternes, 1998). The elimination of these compounds may be
highly dependent on the configurations and operation conditions
of individual WWTP as well as wastewater characteristics,
and thus no definitive conclusion could be reached.
Higher load of carbamazepine was found in the secondary
effluent than in the primary effluent, indicating negative
removal efficiency during the secondary treatment. Some
carbamazepine was found to be excreted as the form of
conjugates (Vieno et al., 2007b), which was biodegraded to
carbamazepine by enzymatic processes during the secondary
treatment, resulting in additional amounts of carbamazepine
in the secondary effluent. However, as the calculations of all
the removal efficiencies were based on grab samples that were
not sampled with a hydraulic lag in the present study, some
error might be brought in due to diurnal variation of the
concentration. Therefore, the present study only provided
a snapshot of the removal of pharmaceuticals and consumer
products in the WWTPs of Beijing. To better illustrate that,
24-h composite samples that are lagged by HRT should be
collected and analyzed in further studies.
It has been reported that high HRT (>12 h) and SRT (>10 d)
may contribute to an increased removal rate of pharmaceuticals
(Jones et al., 2007; Vieno et al., 2007b). In the present
study, the WWTP C, in which the HRT was higher than the
others, was the best in removing these compounds, due to
increased contact time of target compounds and the microorganisms.
On the other hand, the different SRTs did not have
significant effects on the removal efficiency, probably because
the SRTs in all the four WWTPs were relatively high (>10 d),
and without large differences. In addition, it is noteworthy
that the WWTP C employed oxidation ditches, which showed
better removal of natural estrogens and estrogenic activity
than A/O (Hashimoto et al., 2007). It also could be the reason
for the higher removal efficiencies in the WWTP C. Further
investigation for different types of WWTPs is necessary to
confirm the results mentioned above.
3.4. Removal efficiency in advanced treatment processes
The removal efficiencies of the pharmaceuticals during the SF,
UF/ozonation, as well as MF/RO treatment in three corresponding
WWTPs are listed in Table 5.
Generally, sand filtration was not effective for these
compounds. Only trimethoprim, DEET and gemfibrozil were
removed slightly during this treatment process. It should be
noticed that these compounds were efficiently removed in the
secondary treatment, indicating that the biodegradation on
water research 44 (2010) 417–426 423
Table 5 – Removal efficiencies (%) of target
pharmaceuticals and consumer products by advanced
treatment processes in studied WWTPs.
Compound WWTP A WWTP B WWTP D
UF Ozone SF MF/RO
DEET 0–50 50–80 0–50 >90
CF 90
BF 0–50 0–50 0–50 >90
CA 90 90
IM 0–50 >90 90
MA 0–50 80–90 90
SP 0–50 >90 0–50 >90
the biofilm present on the sand particle, rather than the
removal with particles, may be the main reason for their
elimination (Gobel et al., 2007).
The results showed that ozonation is effective in removing
most of the target compounds, probably due to the operation
conditions employed in WWTP A (ozone dosage: 5 mg L 1 ,
contact time: 15 min). Carbamazepine, diclofenac, indomethacin,
sulpiride and trimethoprim were significantly
eliminated, with the removal rates of above 95%. The double
bond in the azepine ring of carbamazepine and pyrrole ring of
indomethacin, and the non-protonated amine of diclofenac
and trimethoprim were susceptible to ozone attack (Vieno
et al., 2007a; Nakada et al., 2007; Westerhoff et al., 2005). The
removal efficiencies of DEET and metoprolol were modest.
The amide group, which is not reactive with ozone, could be
the reason for the modest removal of DEET (Nakada et al., 2007).
Low removal efficiencies were found for bezafibrate, clofibric
acid, as well as caffeine. Only 14% of bezafibrate disappeared
in the ozone process, consistent with its low rate constants
with ozone (590 50 M 1 S 1 , Huber et al., 2003). The reaction
site of bezafibrate is the R-oxysubstituent (–O–C(CH 3) 2COOH)
on one of the aromatic rings. However, as the pK a of bezafibrate
is 3.6, the R-oxysubstituent cannot be deprotonated and
consequently the overall rate constant at pH > 4 is much lower
(Huber et al., 2003). It should be noticed that during the
ozonation, most of the pharmaceuticals were not mineralized
but transformed to the oxidation products. For instance, three
oxidation products containing quinazoline-based functional
groups were identified during the ozonation of CBZ (Mcdowell
et al., 2005).
The good performance of ozonation in the present study
was consistent with Ternes et al. (2003), Huber et al. (2005)
and Okuda et al. (2008). When 5 mg L 1 ozone was applied to
the effluent of a municipal WWTP in Germany (contact time:
18 min), target compounds, such as trimethoprim, carbamazepine,
indomethacin, clofibric acid, were removed by
more than 50% (Ternes et al., 2003). Huber et al. (2005)
conducted a pilot study on the oxidation of pharmaceuticals
during ozonation of conventional activated sludge (CAS) and
membrane bio-reactor (MBR) effluents with various ozone
dosages, and found that macrolide and sulfonamide antibiotics,
estrogens, and acidic pharmaceuticals diclofenac,

424
naproxen and indomethacin were oxidized by more than
90–99% for ozone doses 2mgL 1 in all effluents.
The elimination by ultrafiltration in the WWTP A was low
for all the investigated compounds. The molecular weight cutoff
(MWCO) of UF membranes was much higher than 1000 Da,
thus UF membranes showed poor retention of all the investigated
pharmaceuticals, of which the molecular weight are
less than 400 Da. The removal of individual target compound
was less than 50%, and might be due to the adsorption onto
the membrane. It has been also demonstrated that UF
membrane typically had less than 40% retention of 27 PPCPs,
and the mass balances calculated based on the concentration
of each compound in feed, permeate and retentate showed
the observed retention was significantly governed by adsorption
(Yoon et al., 2006).
In contrast, MF/RO employed in WWTP D was very effective.
In the effluent of MF/RO, all the target compounds except
caffeine were not detected. Generally, one or combination of
three basic mechanisms could be involved during the rejection
of solute by NF/RO membrane: steric effect, charge
exclusion and adsorption (Radjenovic et al., 2008). For most
pharmaceuticals, the rejections were considered to be dominated
by steric interaction in ‘‘tight’’ NF or RO membrane
filtration (Nghiem et al., 2005; Radjenovic et al., 2008). As most
investigated compounds have molecular weights about
200–400 Da, smaller than MWCO of RO membrane applied,
excellent rejection of most pharmaceuticals by RO membrane
was observed in this study as well as in previous studies
(Kimura et al., 2004; Al-Rifai et al., 2007; Radjenovic et al.,
2008). Besides, membrane fouling and the presence of organic
matter in the wastewater effluents likely contributed to higher
rejections of pharmaceuticals, especially for some hydrophobic
ionogenic compound (Nghiem and Coleman, 2008;
Comerton et al., 2008).
Nevertheless, the rejections of two compounds, caffeine
and mefenamic acid were slightly lower (i.e. 50–80% and
0–50%, respectively). The concentration of mefenamic acid in
feed wastewaters of MF/RO membrane process was very low,
only a bit higher than its LOQ in the wastewater effluent,
which could be the reason for the low rejection rate. The low
retention of caffeine in the present study was inaccordance
with Drewes et al. (2005). They found that in two full-scale RO
facilities, target EDCs and PPCPs were efficiently rejected to
below detection limit except for caffeine, still detected in the
permeates. The physiochemical properties might explain the
low rejection rate of caffeine. As a representative of hydrophilic
and non-ionic compounds, the rejection driven by
charge exclusion and adsorption is negligible, and steric
exclusion is solely responsible for the retention of caffeine
(Nghiem et al., 2005). However, the molecular weight of
caffeine is 195 Da, smaller than other target compounds, and
might result in the decreased removal efficiency during the RO
membrane filtration process.
Compared to the other two, the WWTPs employing ozone
and RO membrane filtration as advanced treatment were
more efficient in removing pharmaceuticals. For these
WWTPs, the advanced treatment made a significant contribution
to the total elimination of most pharmaceuticals
(Fig. 5). Therefore, the utility of efficient advanced treatment
could be considered as a tool to reduce pharmaceuticals in the
water research 44 (2010) 417–426
a
Removal Contribution (%)
b
Contribution to removal efficiency (%)
120
100
80
60
40
20
0
-20
180
160
140
120
100
80
60
40
20
0
-20
-40
-60
DEET
DEET
CF
CP
municipal wastewater treatment plants. However, the problems
of membrane fouling and further treatment or disposal
of retentate challenge the application of RO membrane
filtration (Van der Bruggen et al., 2008). For ozonation, as most
of the pharmaceuticals could not be mineralized, and oxidation
products are formed from parent pharmaceutical
compounds (Mcdowell et al., 2005), more research is required
to identify the oxidation products and their potential toxicity
during the partial oxidation process (Nakada et al., 2007).
Besides, economic feasibility should be evaluated by estimating
the energy consumption and investment and
operation costs for both advanced treatment processes (Joss
et al., 2008).
4. Conclusion
Tertiary treatment
Conventional treatment
CF
TP
TP
BF
BF
13 out of 15 pharmaceuticals and consumer products from
eight classes were detected at four WWTPs in Beijing, China.
The concentrations of most compounds in the influent and
secondary effluent were lower than those reported in the USA
and Europe, but consistent with the production profile of the
CA
GF
DF
Pharmaceutical
CA
GF
DF
Pharmaceutical
IM
IM
MTP
MTP
CBZ
CBZ
SP
SP
Tertiary treatment
Secondary treatment
Primary treatment
Fig. 5 – Contributions of primary, secondary (or
conventional treatment) and tertiary treatment to the total
elimination of selected pharmaceuticals in WWTP A (a)
and WWTP D (b).

pharmaceuticals in China. According to the result of risk
assessment for the secondary effluent, only diclofenac might
pose a risk to the aquatic environment. The removal efficiencies
by the conventional treatment varied for different
compounds, depending on their chemical structures, physiochemical
properties, as well as the specific treatment
processes utilized at each WWTP. Further removal could be
achieved by adopting some advanced treatment processes,
such as ozonation and MF/RO. However, others, such as
sand filtration, showed low efficiency in removing these
compounds from secondary effluent.
Acknowledgement
This study was supported by the National Science Fund for
Distinguished Young Scholars (No. 50625823).
Appendix.
Supplementary data
Supplementary information related to this article can be
found at doi:10.1016/j.watres.2009.07.010.
references
Al-Rifai, J.H., Gabelish, C.L., Schafer, A.I., 2007. Occurrence of
pharmaceutically active and non-steroidal estrogenic
compounds in three different wastewater recycling schemes
in Australia. Chemosphere 69 (5), 803–815.
Bolton, S., Null, G., 1981. Caffeine: psychological effects, use and
abuse. Orthomolecular Psychiatry 10 (3), 202–211.
Castiglioni, S., Bagnati, R., Fanelli, R., Pomati, F., Calamari, D.,
Zuccato, E., 2006. Removal of pharmaceuticals in sewage
treatment plants in Italy. Environmental Science and
Technology 40 (1), 357–363.
Chen, H.C., Wang, P.L., Ding, W.H., 2008. Using liquid
chromatography–ion trap mass spectrometry to determine
pharmaceutical residues in Taiwanese rivers and
wastewaters. Chemosphere 72 (6), 863–869.
China Medicine Economic Information Net (CMEIN), 2005.
Chinese Medical Statistical Yearbook. Products section. China
Medicine Economic Information Net (CMEIN), Beijing, China.
Comerton, A.M., Andrews, R.C., Bagley, D.M., Hao, C.Y., 2008. The
rejection of endocrine disrupting and pharmaceutically active
compounds by NF and RO membranes as a function of
compound and water matrix properties. Journal of Membrane
Science 313 (1–2), 323–335.
Daughton, C.G., Ternes, T.A., 1999. Pharmaceuticals and personal
care products in the environment: agents of subtle change?
Environmental Health Perspectives 107 (S6), 907–938.
Drewes, J.E., Bellona, C., Oekekoven, M., Xu, P., Kim, T.U., Amy, G.,
2005. Rejection of wastewater-derived micropollutants in
high-pressure membrane applications leading to indirect
potable reuse. Environmental Progress 24 (4), 400–409.
European Environment Agency, 1998. Environmental Risk
Assessment – Approaches, Experiences and Information
Source. European Environment Agency, London, pp. 68–86.
Ferrari, B., Paxeus, N., Giudice, R.L., Pollio, A., Grric, J., 2003.
Ecotoxicological impact of pharmaceuticals found in treated
water research 44 (2010) 417–426 425
wastewaters: study of carbamazepine, clofibric acid, and
diclofenac. Ecotoxicology and Environmental Safety 55 (3),
359–370.
Gobel, A., Mcardell, C.S., Joss, A., Siegrist, H., Giger, W., 2007. Fate
of sulfonamides, macrolides, and trimethoprim in different
wastewater treatment technologies. Science of the Total
Environment 372 (2–3), 361–371.
Gomez, M.J., Bueno, M.J.M., Lacorte, S., Fernandez-Alba, A.R.,
Aguera, A., 2007. Pilot survey monitoring pharmaceuticals and
related compounds in a sewage treatment plant located on the
Mediterranean coast. Chemosphere 66 (6), 993–1002.
Grung, M., Kallqvist, T., Sakshaug, S., Skurtveit, S., Thomas, K.V.,
2008. Environmental assessment of Norwegian priority
pharmaceuticals based on the EMEA guideline. Ecotoxicology
and Environmental Safety 71 (2), 328–340.
Gulkowska, A., Leung, H.W., So, M.K., Taniyasu, S., Yamashita, N.,
Yeung, L.W.Y., Richardson, B.J., Lei, A.P., Giesy, J.P., Lam, P.K.S.,
2008. Removal of antibiotics from wastewater by sewage
treatment facilities in Hongkong and Shenzhen, China. Water
Research 42 (1–2), 395–403.
Hashimoto, T., Onda, K., Nakamura, Y., Tada, K., Miya, A.,
Murakami, T., 2007. Comparison of natural estrogen removal
efficiency in the conventional activated sludge process and
the oxidation ditch process. Water Research 41 (10),
2117–2126.
Huber, M.M., Canonica, S., Park, G.Y., Gunten, U.V., 2003.
Oxidation of pharmaceuticals during ozonation and advanced
oxidation processes. Environmental Science and Technology
37 (5), 1016–1024.
Huber, M.M., Gobel, A., Joss, A., Hermann, N., Loffler, D.,
Mcardell, C.S., Ried, A., Siegrist, H., Ternes, T.A., Gunten, U.V.,
2005. Oxidation of pharmaceuticals during ozonation of
municipal wastewater effluents: a pilot study. Environmental
Science and Technology 39 (11), 4290–4299.
Huerta-Fontela, M., Galceran, M.T., Martin-Alonso, J., Ventura, F.,
2008. Occurrence of psychoactive stimulatory drugs in
wastewaters in north-eastern Spain. Science of the Total
Environment 397 (1–3), 31–40.
Huschek, G., Hansen, P.D., Maurer, H.H., Krengel, D., Kayser, A.,
2004. Environmental risk assessment of medicinal products
for human use according to European Commission
recommendations. Environmental Toxicology 19 (3), 226–240.
Jjemba, P.K., 2006. Excretion and ecotoxicity of pharmaceutical
and personal care products in the environment. Ecotoxicology
and Environmental Safety 63 (1), 113–130.
Jones, O.A.H., Voulvoulis, N., Lester, J.N., 2005. Human
pharmaceuticals in wastewater treatment processes. Critical
Reviews in Environmental Science and Technology 35 (4),
401–427.
Jones, O.A.H., Voulvoulis, N., Lester, J.N., 2007. The occurrence
and removal of selected pharmaceutical compounds in
a sewage treatment works utilising activated sludge
treatment. Environmental Pollution 145 (3), 738–744.
Joss, A., Siegrist, H., Ternes, T.A., 2008. Are we about to upgrade
wastewater treatment for removing organic micropollutants.
Water Science and Technology 57 (2), 251–255.
Kasprzyk-Hordern, B., Dinsdale, R.M., Guwy, A.J., 2009. The
removal of pharmaceuticals, personal care products,
endocrine disruptors and illicit drugs during wastewater
treatment and its impact on the quality of receiving waters.
Water Research 43 (2), 363–380.
Khan, S.J., Wintgens, T., Sherman, P., Zaricky, J., Schafer, A.I.,
2004. Removal of hormones and pharmaceuticals in the
advanced water recycling demonstration plant in Queensland,
Australia. Water Science and Technology 50 (5), 15–22.
Khan, S.J., Ongerth, J.E., 2004. Modelling of pharmaceutical
residues in Australian sewage by quantities of use and
fugacity calculations. Chemosphere 54 (3), 355–367.

426
Kim, S.D., Cho, J., Kim, I.S., Vanderford, B.J., Snyder, S.A., 2007.
Occurrence and removal of pharmaceuticals and endocrine
disruptors in South Korean surface, drinking, and waste
waters. Water Research 41 (5), 1013–1021.
Kimura, K., Toshima, S., Amy, G., Watanabe, Y., 2004. Rejection of
neutral endocrine disrupting compounds (EDCs) and
pharmaceutical active compounds (PhACs) by RO membranes.
Journal of Membrane Science 245 (1–2), 71–78.
Lishman, L., Smyth, S.A., Sarafin, K., Kleywegt, S., Toito, J.,
Peart, T., Lee, B., Servos, M., Beland, M., Seto, P., 2006.
Occurrence and reductions of pharmaceuticals and personal
care products and estrogens by municipal wastewater
treatment plants in Ontario, Canada. Science of the Total
Environment 367 (2–3), 544–558.
Lindqvist, N., Tuhkanen, T., Kronberg, L., 2005. Occurrence of
acidic pharmaceuticals in raw and treated sewages and in
receiving waters. Water Research 39 (11), 2219–2228.
Mcdowell, D.C., Huber, M.M., Wagner, M., Gunten, U.V.,
Ternes, T.A., 2005. Ozonation of carbamazepine in drinking
water: identification and kinetic study of major oxidation
products. Environmental Science and Technology 39 (20),
8014–8022.
Nakada, N., Tanishima, T., Shinohara, H., Kiri, K., Takada, H.,
2006. Pharmaceutical chemicals and endocrine disrupters in
municipal wastewater in Tokyo and their removal during
activated sludge treatment. Water Research 40 (17), 3297–3303.
Nakada, N., Shinohara, H., Murata, A., Kiri, K., Managaki, S.,
Sato, N., Takada, H., 2007. Removal of selected
pharmaceuticals and personal care products (PPCPs) and
endocrine-disrupting chemicals (EDCs) during sand filtration
and ozonation at a municipal sewage treatment plant. Water
Research 41 (19), 4372–4382.
Nghiem, L.D., Schafer, A.I., Elimelech, M., 2005. Pharmaceutical
retention mechanisms by nanofiltration membrane.
Environmental Science and Technology 39 (19), 7698–7705.
Nghiem, L.D., Coleman, P.J., 2008. NF/RO filtration of the
hydrophobic ionogenic compound triclosan: transport
mechanisms and the influence of membrane fouling.
Separation and Purification Technology 62 (3), 709–716.
Niwa, T., Inoue, S., Shiraga, T., Takagi, A., 2005. No inhibition of
cytochrome P450 activities in human liver microsomes by
sulpiride, an antipsychotic drug. Biological and
Pharmaceutical Bulletin 28 (1), 188–191.
Okuda, T., Kobayashi, Y., Nagao, R., Yamashita, N., Tanaka, H.,
Tanaka, S., Fujii, S., Konishi, C., Houwa, I., 2008. Removal
efficiency of 66 pharmaceuticals during wastewater treatment
process in Japan. Water Science and Technology 57 (1), 65–71.
Paxeus, N., 2004. Removal of selected non-steroidal antiinflammatory
drugs (NSAIDs), gemfibrozil, carbamazepine,
beta-blockers, trimethoprim and triclosan in conventional
wastewater treatment plants in five EU countries and their
discharge to the aquatic environment. Water Science and
Technology 50 (5), 253–260.
Quinn, B., Gagne, F., Blaise, C., 2008. An investigation into the acute
and chronic toxicity of eleven pharmaceuticals (and their
solvents) found in wastewater effluent on the cnidarian, Hydra
attenuata. Science of the Total Environment 389 (2–3), 306–314.
Radjenovic, J., Petrovic, M., Ventura, F., Barcelo, D., 2008. Rejection
of pharmaceuticals in nanofiltration and reverse osmosis
water research 44 (2010) 417–426
membrane drinking water treatment. Water Research 42 (14),
3601–3610.
Santos, J.L., Aparicio, I., Alonso, E., 2007. Occurrence and risk
assessment of pharmaceutically active compounds in
wastewater treatment plants. A case study: Seville city
(Spain). Environment International 33 (4), 596–601.
Sui, Q., Huang, J., Deng, S. B., Yu, G. Rapid determination of
pharmaceuticals from multiple therapeutic classes in
wastewater by solid-phase extraction and ultra-performance
liquid chromatography tandem mass spectrometry. Chinese
Science Bulletin, in press.
Ternes, T.A., 1998. Occurrence of drugs in German sewage
treatment plants and rivers. Water Research 32 (11),
3245–3260.
Ternes, T.A., Meisenheimer, M., Mcdowell, D., Sacher, F.,
Brauch, H.J., Haist-gulde, B., Preuss, G., Wilme, U.,
Zulei-seibert, N., 2002. Removal of pharmaceuticals during
drinking water treatment. Environmental Science and
Technology 36 (17), 3855–3863.
Ternes, T.A., Stuber, J., Herrmann, N., Mcdowell, D., Ried, A.,
Kampmann, M., Teiser, B., 2003. Ozonation: a tool for removal
of pharmaceuticals, contrast media and musk fragrances from
wastewater? Water Research 37 (8), 1976–1982.
Thomas, P.M., Foster, G.D., 2005. Tracking acidic
pharmaceuticals, caffeine, and triclosan through the
wastewater treatment process. Environmental Toxicology and
Chemistry 24 (1), 25–30.
Vanderford, B.J., Snyder, S.A., 2006. Analysis of pharmaceuticals
in water by isotope dilution liquid chromatography/tandem
mass spectrometry. Environmental Science and Technology
40 (23), 7312–7320.
Van der Bruggen, B., Manttari, M., Nystrom, M., 2008.
Drawbacks of applying nanofiltration and how to avoid them:
a review. Separation and Purification Technology 63 (2),
251–263.
Vieno, N.M., Harkki, H., Tuhkanen, T., Kronberg, L., 2007a.
Occurrence of pharmaceuticals in river water and their
elimination in a pilot-scale drinking water treatment plant.
Environmental Science and Technology 41 (14), 5077–5084.
Vieno, N., Tuhkanen, T., Kronberg, L., 2007b. Elimination of
pharmaceuticals in sewage treatment plants in Finland. Water
Research 41 (5), 1001–1012.
Watkinson, A.J., Murby, E.J., Costanzo, S.D., 2007. Removal of
antibiotics in conventional and advanced wastewater
treatment: implications for environmental discharge and
wastewater recycling. Water Research 41 (18), 4164–4176.
Westerhoff, P., Yoon, Y., Snyder, S., Wert, E., 2005. Fate of
endocrine-disruptor, pharmaceutical, and personal care
product chemicals during simulated drinking water treatment
processes. Environmental Science and Technology 39 (17),
6649–6663.
Xu, W.H., Zhang, G., Li, X.D., Zou, S.C., Li, P., Hu, Z.H., Li, J., 2007.
Occurrence and elimination of antibiotics at four sewage
treatment plants in the Pearl River Delta, South China. Water
Research 41 (19), 4526–4534.
Yoon, Y., Westerhoff, P., Snyder, S.A., Wert, E.C., 2006.
Nanofiltration and ultrafiltration of endocrine disrupting
compounds, pharmaceuticals and personal care products.
Journal of Membrane Science 270 (1–2), 88–100.

Application of the combination index (CI)-isobologram
equation to study the toxicological interactions of lipid
regulators in two aquatic bioluminescent organisms
Ismael Rodea-Palomares a,1 , Alice L. Petre b,1 , Karina Boltes b , Francisco Leganés a ,
José Antonio Perdigón-Melón b , Roberto Rosal b , Francisca Fernández-Piñas a, *
a
Departamento de Biología, Facultad de Ciencias, Universidad Autónoma de Madrid, 2 Darwin Street, Cantoblanco, 28049 Madrid, Spain
b
Departamento de Ingeniería Química, Universidad de Alcalá, Alcalá de Henares, E-28871 Madrid, Spain
article info
Article history:
Received 26 March 2009
Received in revised form
13 July 2009
Accepted 18 July 2009
Available online 25 July 2009
Keywords:
Antagonism
Combination index-isobologram
equation
Cyanobacterium
Fibrates
Synergism
Vibrio fischeri
1. Introduction
abstract
Fibrates and statins (HMG-CoA reductase inhibitors) are the
main lipid-lowering drugs prescribed either alone or in
combination therapy in order to decrease plasma cholesterol
levels and reduce the incidence of coronary heart disease.
Although partially displaced by statins, the total number of
fibrate prescriptions is in constant increase in the United
States (Holoshitz et al., 2008). Fibric acids are the active forms
water research 44 (2010) 427–438
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
* Corresponding author. Tel.: þ34 9 1497 8176; fax: þ34 9 1497 8344.
E-mail address: francisca.pina@uam.es (F. Fernández-Piñas).
1
Both authors contributed equally to this work.
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.watres.2009.07.026
Pharmaceuticals in the aquatic environment do not appear singly and usually occur as
complex mixtures, whose combined effect may exhibit toxicity to the aquatic biota. We
report an environmental application of the combination index (CI)-isobologram equation,
a method widely used in pharmacology to study drug interactions, to determine the nature
of toxicological interactions of three fibrates toward two aquatic bioluminescent organisms,
Vibrio fischeri and the self-luminescent cyanobacterial recombinant strain Anabaena
CPB4337. The combination index-isobologram equation method allows computerized
quantitation of synergism, additive effect and antagonism. In the Vibrio test, the fibrate
combinations showed antagonism at low effect levels that turned into an additive effect or
synergism at higher effect levels; by contrast, in the Anabaena test, the fibrate combinations
showed a strong synergism at the lowest effect levels and a very strong antagonism at high
effect levels. We also evaluated the nature of the interactions of the three fibrates with a real
wastewater sample in the cyanobacterial test. We propose that the combination indexisobologram
equation method can serve as a useful tool in ecotoxicological assessment.
ª 2009 Elsevier Ltd. All rights reserved.
of fibrates and belong to the nuclear receptor superfamily of
ligand-activated transcription factors. Gemfibrozil and fenofibrate
are the fibrates currently marketed in the US, whereas
bezafibrate is also available in Europe and other developed
countries (Lambropoulou et al., 2008). Fenofibric acid, 2-[4-(4chlorobenzoyl)phenoxy]-2-methylpropanoic
acid, is the active
metabolite of fenofibrate, the inactive prodrug marketed and
dispensed. Gemfibrozil, 5-(2,5-dimethylphenoxy)-2,2-dimethylpentanoic
acid and bezafibrate, p-[4-[chlorobenzoylamino-

428
ethyl]-phenoxy]-b-methylpropionic acid, are also fibric acid
derivatives with similar pharmacokinetic behaviour (Miller and
Spence, 1998).
The occurrence of lipid regulators in the discharge of
treated urban and municipal wastewater has been relatively
well documented. Bezafibrate has been detected in effluents
of two British STP with averages up to 230 ng/L (Kasprzyk-
Hordern et al., 2009). Metcalfe et al. (2003) found around 1 mg/L
of gemfibrozil in effluents of Canadian STP, whereas fenofibrate
has been reported in concentrations up to 0.5 mg/L in the
influent of several Brazilian STP (Stumpf et al., 1999).
Andreozzi et al. (2003) found lipid regulators in the effluent of
several European STP at concentrations up to 4.76 mg/L (gemfibrozil),
1.07 mg/L (bezafibrate) and 0.16 mg/L (fenofibrate).
Rosal et al. (2008), reported the occurrence of bezafibrate and
gemfibrozil at levels of 139 and 608 ng/L respectively in the
effluent of a Spanish STP. In the same plant Rodríguez et al.
(2008) found 165 ng/L of fenofibric acid, 61 ng/L of bezafibrate
and 143 ng/L of gemfibrozil.
It is also significant that removal efficiencies observed in
current STP are not always high. Fent et al. (2006) reported
maximum removal rates of 50–75% for fenofibric acid and
gemfibrozil and somewhat higher for bezafibrate, although for
the later, efficiencies below 15% have also been reported.
Stumpf et al. (1999) reported a 45% removal of fenofibric acid
by an activated sludge conventional treatment. Kasprzyk-
Hordern et al. (2009) encountered an average degradation of
bezafibrate not higher than 67%. On the other hand, Castiglioni
et al. (2006) reported that the removal efficiency of
bezafibrate during an activated sludge treatment greatly
varied from 15% in winter to 87% in summer.
At measured environmental concentrations as those
reported above (mostly in the ng/L and mg/L range), many
studies have shown that the risk of acute toxicity is unlikely
(Fent et al., 2006; Han et al., 2006; Borgmann et al., 2007);
however, there is a lack of data on chronic toxicity effects.
Moreover, pharmaceuticals in the aquatic environments occur
as complex mixtures from different classes, not as single
contaminants (Gros et al., 2007); thus, although the concentration
of individual pharmaceuticals is low, their mixture
could prove ecotoxicologically significant (Brain et al., 2004).
Current methods of risk assessment usually focus on the
assessment of single chemicals, which may underestimate
the risk associated with toxic action of mixtures; probably for
this reason, in the last years there is an increasing number of
studies dealing with complex mixtures of pharmaceuticals
(Cleuvers, 2003, 2004; Crane et al., 2006; Han et al., 2006;
Borgmann et al., 2007; Christensen et al., 2007; Pomati et al.,
2008; Quinn et al., 2009). However, assessment of combined
toxicities is not an easy issue. Basically, two different models
are in use for the prediction of mixture toxicity, i.e., concentration
addition, when pharmaceuticals have a similar mode
of toxic action, and response addition or independent action,
when pharmaceuticals have different modes of toxic action
(Cleuvers, 2003; Teuschler, 2007). However, toxicological
interactions, synergisms or antagonisms, between the pharmaceuticals
and their effects can occur independently of
mode of action; moreover, in most cases, the pharmacological
mechanisms of action is known but the toxic mode of action
may remain unknown (Cleuvers, 2003; Chou, 2006). In an effort
water research 44 (2010) 427–438
to overcome this limitation, we report an environmental
application of a method widely used in pharmacology to
interpret drug interactions; this method, termed as the
median-effect/combination index (CI)-isobologram equation
(Chou, 2006) allows quantitative determinations of chemical
interactions where CI 1 indicate synergism,
additive effect and antagonism, respectively. One important
property of the method is that previous knowledge of the
mechanisms of action of each chemical is not required.
Besides, the method takes into account both the potency and
the shapes of the dose-effect curve of each chemical. The
method has been computerized allowing an automated
simulation of synergism and antagonism at different
concentrations and at different effect levels of the chemicals
in a mixture.
The aim of our study was to assess the nature of the
toxicological interactions of three fibrates, gemfibrozil, bezafibrate
and fenofibric acid, by the method of combination
index (CI)-isobologram equation. The three pharmaceuticals
were used singly or in two- and three-drug combinations. As
toxicity endpoint we have chosen the bioluminescent
response of two prokaryotes, the naturally luminescent Vibrio
fischeri and the recombinant bioluminescent cyanobacterium
Anabaena sp. PCC 7120 CPB4337 (hereinafter, Anabaena
CPB4337), both bioluminescent organisms have proved very
useful in evaluating the toxicity of individual fibrates in
a previous study (Rosal et al., 2009). For Anabaena CPB4337, we
also evaluated the nature of the interactions of the three
fibrates with a wastewater sample from a local STP, which
already proved very toxic to the cyanobacterium (Rosal et al.,
2009).
2. Materials and methods
2.1. Materials
Gemfibrozil (þ99%) and bezafibrate (þ98%) were purchased
from Sigma–Aldrich. Fenofibric acid was produced from
fenofibrate (Sigma–Aldrich, þ99% purity) by hydrolysis.
A suspension of fenofibrate in isopropanol (30 wt.%, 400 mL)
was refluxed during 4 h with an aqueous sodium hydroxide
solution (2.0 M, 200 mL). After cooling to less than 70 C,
a solution of hydrochloric acid (1.0 M, 325 mL) was slowly
added while keeping the temperature over 60 C. The product
crystallized after cooling and keeping at room temperature
during 4 or more h. The product was filtered and rinsed with
water and dried overnight at 60 C under nitrogen. The purity
of the product was over 97% checked by HPLC. Solubility of
acidic drugs in water is strongly pH dependent with few data
considering this variable. Comerton et al. (2007) reported
a solubility of 10.9 mg/L of gemfibrozil in water, but we could
solve over 125 mg/L in 2 mM MOPS (3-[N-morpholino] propanesulfonic
acid) at pH 6 and higher quantities for the pH at
which V. fischeri bioassays were performed. In all cases, we
avoided the use of solvents and the upper limit for the
concentrations of the studied compounds was their solubility
in pure water or wastewater at the pH of the bioassay.
Wastewater samples were collected from the secondary
clarifier of a STP located in Alcalá de Henares (Madrid) that

eceives domestic wastewater with a minor contribution of
industrial effluents from facilities located near the city. This
STP used a conventional activated sludge treatment and has
been designed for a total capacity of 375,000 equivalent
inhabitants with a maximum flow rate of 3000 m 3 /h. In
a recent previous study (Rosal et al., 2009), we found that this
wastewater was very toxic to Anabaena cells with a wastewater
dilution as low as 0.11 causing 50% luminescence
inhibition (wastewater EC 50).
2.2. Toxicity tests
Bioassays with the photo-luminescent bacteria Vibrio fischeri
were carried out according to ISO 11348-3 standard protocol
(ISO, 2007). This bioassay measures, during the prescribed
incubation period, the decrease in bioluminescence induced
in the cell metabolism due to the presence of a toxic
substance. The bacterial assay used the commercially available
Biofix Lumi test (Macherey-Nagel, Germany). The bacterial
reagent is supplied freeze-dried (Vibrio fischeri NRRL-B
11177) and was reconstituted and incubated at 3 C for 5 min
before use. The desired pH was set by using NaOH or HCl. The
analysis media was 0.34 M NaCl (2% w/v) and tests were
performed at 15 C and the measurements of light were made
using a luminometer (Optocomp I). The effect of toxicants or
toxicant mixtures (i.e., fibrates or fibrate combinations) was
measured as percent inhibition with respect to the light
emitted under test conditions in the absence of any toxic
influence. Toxicity values were routinely obtained after
30 min exposure. Phenol and ZnSO 4 7H 2O have been used as
toxicity standards and all tests have been replicated to ensure
reproducibility.
The bioassays using the recombinant bioluminescent
cyanobacterium Anabaena CPB4337 were based on the inhibition
of constitutive luminescence caused by the presence of
any toxic substance (Rodea-Palomares et al., 2009; Rosal et al.,
2009). Anabaena CPB4337 was routinely grown at 28 C in the
light, ca. 65 mmol photons m 2 s 1 on a rotary shaker in 50 mL
AA/8 (Allen and Arnon, 1955) supplemented with nitrate
(5 mM) in 125 ml Erlenmeyer flasks and 10 mg/mL of neomycin
sulphate (Nm). Luminescence inhibition-based toxicity assays
were performed as follows: 160 mL from five to seven serial
dilutions of each tested toxicant or toxicant mixture (i.e.;
fibrates or fibrate combinations) plus a control (ddH 2O buffered
with MOPS at pH 5.8) were disposed in an opaque white
96-well microtiter plates. 40 mL cells, grown as described, were
washed twice and resuspended in ddH2O buffered with MOPS
at pH 5.8 and were added to the microtiter plate wells to reach
a final cell density at OD750 nm of 0.5. The luminescence of
each sample was recorded every 5 min for up to 1 h in the
Centro LB 960 luminometer. Three independent experiments
with duplicate samples were carried out for all Anabaena
toxicity assays. CuSO 4 has been used as toxicity standard and
all tests have been replicated to ensure reproducibility.
2.3. Experimental design of fibrate combinations
Solutions of gemfibrozil (Gm), bezafibrate (Bz) and fenofibric
acid (Fn) prepared as described above were used singly and in
two (Bz þ Gm; Fn þ Gm; Fn þ Bz) and three (Fn þ Gm þ Bz)
water research 44 (2010) 427–438 429
combinations. Anabaena and Vibrio fischeri cells were treated
with serial dilutions of each fibrate individually and with
a fixed constant ratio (1:1), based on the individual EC50 values,
in their binary and ternary combinations. Five dilutions (serial
dilution factor ¼ 2) of each fibrate and combination plus
a control were tested in three independent experiments with
replicate samples.
For evaluating the nature of the interaction of fibrates with
wastewater, binary combinations of each fibrate plus wastewater
(Fn þ WW; Gm þ WW; Bz þ WW) and a quaternary
combination of the three fibrates plus wastewater (Fn þ Gm þ
Bz þ WW) were also prepared and tested for Anabaena
CPB4337. Anabaena cells were treated with serial dilutions of
each fibrate and wastewater individually and with a fixed
constant ratio (1:1), based on the individual EC50 values, in
their binary and quaternary combinations. Five dilutions
(serial dilution factor ¼ 2) of each fibrate and wastewater and
their combinations plus a control were tested in three independent
experiments with replicate samples. The experimental
design is shown in Table 1.
All individual fibrate, wastewater and their combination
assays were carried out at the same time as recommended by
Chou (2006) to maximize computational analysis of data.
2.4. Median-effect and combination index (CI)isobologram
equations for determining combined
fibrate interactions
The results were analyzed using the median-effect/combination
index (CI)-isobologram equation by Chou (2006) and Chou
and Talalay (1984) which is based on the median-effect principle
(mass-action law) (Chou, 1976) that demonstrates that
there is an univocal relationship between dose and effect
independently of the number of substrates or products and of
the mechanism of action or inhibition. This method involved
plotting the dose-effect curves for each compound and their
combinations in multiple diluted concentrations by using the
median-effect equation:
fa D
¼
fu Dm
m
Where D is the dose, Dm is the dose for 50% effect (e.g., 50%
inhibition of bioluminescence or EC 50), fa is the fraction
affected by dose D (e.g., 0.75 if cell bioluminescence is inhibited
by 75%), fu is the unaffected fraction (therefore, fa ¼ 1 fu),
and m is the coefficient of the sigmoidicity of the dose-effect
curve: m ¼ 1, m > 1, and m < 1 indicate hyperbolic, sigmoidal,
and negative sigmoidal dose-effect curve, respectively.
Therefore, the method takes into account both the potency
(Dm) and shape (m) parameters. If Eq. (1) is rearranged, then:
D ¼ Dm½fa=ð1 faÞŠ 1=m
The Dm and m values for each fibrate are easily determined by
the median-effect plot: x ¼ log (D) versus y ¼ log ( fa/fu) which
is based on the logarithmic form of Eq. (1). In the medianeffect
plot, m is the slope and log (Dm) is the x-intercept. The
conformity of the data to the median-effect principle can be
ready manifested by the linear correlation coefficient (r) of the
data to the logarithmic form of Eq. (1) (Chou, 2006).
(1)
(2)

Table 1 – Experimental design for determining toxicological interactions of fenofibric acid [Fn (D)1], gemfibrozil [Gm (D)2], bezafibrate [Bz (D)3] and their binary and ternary
combinations for Vibrio fischeri and Anabaena CPB4337 bioluminescence tests.
Pure fibrate experiments Fibrates plus wastewater experiments
Vibrio fischeri Anabaena CPB4337 Anabaena CPB4337
Dilutions Single toxicant Single toxicant Single toxicant
Fn Gm Bz Fn Gm Bz Fn Gm Bz WW**
(D) 1 (D) 2 (D) 3 (D) 1 (D) 2 (D) 3 (D) 1 (D) 2 (D) 3 (D) 4
1
⁄4 (EC50) 0.4 8.75 37.5 2.5 2.5 12.5
1
⁄4 (EC50) 2.5 2.5 12.5 0.025
1
⁄2 (EC50) 0.8 17.5 75 5 5 25
1
⁄2 (EC50) 5 5 25 0.05
1 (EC50) 1.6 35 150 10 10 50 1 (EC50) 10 10 50 0.1
2 (EC50) 3.2 70 300 20 20 100 2 (EC50) 20 20 100 0.2
4 (EC50) 6.4 140 600 40 40 200 4 (EC50) 40 40 200 0.4
Two toxicant combo Two toxicant combo Two toxicant combo
(D) 1 þ (D) 2 (1.6:35) (D) 1 þ (D) 2 (1:1) (D) 1 þ (D) 4 (1:0.01)
1
⁄4 (EC50) 0.4 8.75 2.5 2.5
1
⁄4 (EC50) 2.5 0.025
1
⁄2 (EC50) 0.8 17.5 5 5
1
⁄2 (EC50) 5 0.05
1 (EC50) 1.6 35 10 10 1 (EC50) 10 0.1
2 (EC50) 3.2 70 20 20 2 (EC50) 20 0.2
4 (EC50) 6.4 140 25* 25* 4 (EC50) 40 0.4
(D) 1 þ (D) 3 (1.6:150) (D) 1 þ (D) 3 (1:5) (D) 2 þ (D) 4 (1:0.01)
1
⁄4 (EC50) 0.4 37.5 2.5 12.5
1
⁄4 (EC50) 2.5 0.025
1
⁄2 (EC50) 0.8 75 5 25
1
⁄2 (EC50) 5 0.05
1 (EC50) 1.6 150 10 50 1 (EC50) 10 0.1
2 (EC50) 3.2 300 20 100 2 (EC50) 20 0.2
4 (EC50) 6.4 600 30* 150* 4 (EC50) 40 0.4
(D)2 þ (D)3 (35:150) (D)2 þ (D)3 (1:5) (D)3 þ (D)4 (1:0.002)
1
⁄4 (EC50) 8.75 37.5 2.5 12.5
1
⁄4 (EC50) 12.5 0.025
1
⁄2 (EC50) 17.5 75 5 25
1
⁄2 (EC50) 25 0.05
1 (EC50) 35 150 10 50 1 (EC50) 50 0.1
2 (EC50) 70 300 20 100 2 (EC50) 100 0.2
4 (EC50) 140 600 40 200 4 (EC50) 200 0.4
Three toxicant combo Three toxicant combo Four toxicant combo
(D)1 þ (D)2 þ (D)3 (1.6:35:150) (D)1 þ (D)2 þ (D)3 (1:1:5) (D)1 þ (D)2 þ (D)3 þ (D)4 (1:1:5:0.01)
1
⁄4 (EC50) 0.4 8.75 37.5 2.5 2.5 12.5 1/8 (EC50) 1.25 1.25 6.25 0.0125
1
⁄2 (EC50) 0.8 17.5 75 5 5 25
1
⁄4 (EC50) 2.5 2.5 12.5 0.025
1 (EC50) 1.6 35 150 10 10 50
1
⁄2 (EC50) 5 5 25 0.05
2 (EC50) 3.2 70 300 20 20 100 1 (EC50) 10 10 50 0.1
4 (EC50) 6.4 140 600 40 40 200 2 (EC50) 20 20 100 0.2
For the Anabaena test, the design for the experiment with the wastewater [WW (D4)] sample is also included. The experimental design is based on EC50 ratios as proposed by Chou and Talalay (1984).
EC 50 is the effective concentration of a toxicant which caused a 50% bioluminescence inhibition. The combination ratio was approximately equal to the EC 50 ratio of the combination components (i.e.,
close to their equipotency ratio). *Upper maximal possible dose due to the solubility limit of fibrates in pure water. **EC 50 for wastewater is the dilution which caused 50% luminescence inhibition. (D) 1,
(D)2 and (D)3 in mg/L, (D)4 is the dilution of wastewater in ddH2O.
430
water research 44 (2010) 427–438

These parameters were then used to calculate doses of the
fibrates and their combinations required to produce various
effect levels according to Eq. (1); for each effect level, combination
index (CI) values were then calculated according to the
general combination index equation for n chemical combination
at x% inhibition (Chou, 2006):
n ðCIÞx ¼ Xn
j¼1
ðDÞj ðDxÞj ¼ Xn
j¼1
ðDmÞ j
ðDxÞ1 n ½DŠj = Pn 1 ½DŠ
n h
fax = 1 j
fax j
io 1=mj
where n (CI) x is the combination index for n chemicals (e.g.,
fibrates) at x% inhibition (e.g., bioluminescence inhibition);
(Dx) 1 n is the sum of the dose of n chemicals that exerts x%
inhibition in combination, {[Dj]/ Pn 1 ½DŠ} is the proportionality
of the dose of each of n chemicals that exerts x% inhibition in
combination; and (Dm)j {( fax)j/[1 ( fax)j]} 1/mj is the dose of each
drug alone that exerts x% inhibition. From Eq. (3), CI1 indicates synergism, additive effect and antagonism,
respectively.
2.5. Analysis of results
Computer program CompuSyn (Chou and Martin, 2005,
Compusyn Inc, USA) was used for calculation of dose-effect
curve parameters, CI values, fa-CI plot (plot representing CI
(3)
versus fa, the fraction affected by a particular dose; see Eq. (1))
and polygonograms (a polygonal graphic representation
depicting synergism, additive effect and antagonism for three
or more drug combinations). Linear regression analyses were
computed using MINITAB Release 14 for Windows (Minitab
Inc; USA).
3. Results
3.1. Toxicological interactions of fibrate combinations in
Vibrio fischeri and Anabaena CPB4337 bioluminescence
tests
Applying the combination index-isobologram method, we
evaluated the nature of gemfibrozil (Gm), fenofibric acid (Fn)
and bezafibrate (Bz) interactions both in Vibrio fischeri and
Anabaena CPB4337 bioluminescence tests. Table 2 shows the
dose-effect curve parameters (Dm, m and r) of the three
fibrates singly and their binary and ternary combinations, as
well as mean combination index (CI) values of fibrate combinations.
Dm was the dose required to produce the medianeffect
(analogous to the EC 50); Dm values for Fn were the
lowest both, in Vibrio and Anabaena tests, Dm values for Gm
were in the same range for both Vibrio and Anabaena while Bz
Table 2 – Dose-effect relationship parameters and mean combination index (CI) values (as a function of fractional inhibition
of luminescence) of gemfibrozil (Gm), fenofibric acid (Fn), and bezafibrate (Bz) individually and of their binary and ternary
combinations on Vibrio fischeri and Anabaena CPB4337 bioluminescence tests.
Drug combo Vibrio fischeri
Dose-effect parameters CI values
Dm m r EC10 EC50 EC90
mg/L (mM)
Fn 1.45 (4.01) 0.78 0.989 – – –
Gm 20.58 (82.11) 1.53 0.966 – – –
Bz 252.07 (696.46) 1.15 0.975 – – –
Gm þ Bz 78.20 (234.20) 1.54 0.991 1.13 0.13 Add 0.97 0.04 Add 0.86 0.05 Syn
Fn þ Bz 153.79 (424.93) 1.09 0.981 2.98 0.15 Ant 1.71 0.03 Ant 1.17 0.06 Ant
Fn þ Gm 9.84 (38.74) 1.15 0.973 0.99 0.17 Add 0.75 0.05 Syn 0.86 0.08 Syn
Fn þ Gm þ Bz 55.69 (166.69) 1.23 0.993 1.46 0.06 Ant 1.01 0.02 Add 0.99 0.03 Add
Anabaena CPB4337
Dose-effect parameters CI values
Dm m r EC 10 EC 50 EC 90
mg/L (mM)
water research 44 (2010) 427–438 431
Fn 8.53 (23.62) 0.96 0.971 – – –
Gm 10.69 (42.67) 0.81 0.959 – – –
Bz 12.56 (34.70) 1.08 0.990 – – –
Gm þ Bz 19.17 (56.88) 0.84 0.972 1.06 0.15 Add 1.57 0.06 Ant 2.5 0.22 Ant
Fn þ Bz 13.92 (38.49) 0.76 0.965 0.55 0.06 Syn 1.19 0.04 Ant 2.59 0.14 Ant
Fn þ Gm 12.26 (41.45) 0.46 0.955 0.13 0.02 Syn 1.29 0.05 Ant 12.9 2.33 Ant
Fn þ Gm þ Bz 6.62 (19.45) 0.53 0.960 0.09 0.01 Syn 0.57 0.02 Syn 3.92 0.19 Ant
The parameters m, Dm and r are the antilog of x-intercept, the slope and the linear correlation coefficient of the median-effect plot, which
signifies the shape of the dose-effect curve, the potency (EC50), and conformity of the data to the mass-action law, respectively (Chou, 1976; Chou
and Talalay, 1984; Chou, 2006). Dm and m values are used for calculating the CI values (Eq. (3)); CI 1 indicate synergism (Syn),
additive effect (Add), and antagonism (Ant), respectively. EC 10,EC 50 and EC 90, are the doses required to inhibit bioluminescence 10, 50 and 90%,
respectively. Computer software CompuSyn was used for automated calculation and simulation.

432
Dm values were an order of magnitude higher in the Vibrio test
(Rosal et al., 2009); m was the Hill coefficient used to determine
the shape of the dose-response curve, hyperbolic (m ¼ 1),
sigmoidal (m > 1) or negative sigmoidal (m < 1); also shown in
the table, linear regression correlation coefficients (r-values)
of the median-effect plots were >0.95 in all cases, indicating
the conformity of the data to the median-effect principle
which qualifies for further studies using this method.
The Dm and m values for single fibrates and for their
combination mixtures were used for calculating synergism or
antagonism based on the CI Eq. (3) (Chou, 2006). Fig. 1 shows
the fa-CI plot of fibrate interactions both for Vibrio (Fig. 1a) and
Anabaena tests (Fig. 1b); the fa-CI plot depicts the CI value
a
Combination index, CI
b
Combination index, CI
3.0
2.5
2.0
1.5
1.0
0.5
0.0
0.0 0.2 0.4 0.6 0.8 1.0
3.0
2.5
2.0
1.5
1.0
0.5
Fraction affected, fa
0.0
0.0 0.2 0.4 0.6 0.8 1.0
Fraction affected, fa
Fig. 1 – Combination index plot ( fa-CI plot) for a set of three
fibrate combinations: Fn D Bz (–6–), Bz D Gm (–B–),
Fn D Gm (–,–) and Fn D Gm D Bz (–;–) for Vibrio fischeri
test (a) and Anabaena CPB4337 test (b). CI values are plotted
as a function of the fractional inhibition of
bioluminescence ( fa) by computer simulation (CompuSyn)
from fa [ 0.10 to 0.95. CI 1 indicates
synergism, additive effect and antagonism, respectively.
At least three independent experiments with two
replicates were used. The vertical bars indicate 95%
confidence intervals for CI values based on sequential
deletion analysis (SDA) (Chou and Martin, 2005).
Fn [ fenofibric acid, Bz [ bezafibrate and
Gm [ gemfibrozil.
water research 44 (2010) 427–438
versus fa (effect level or fraction of luminescence inhibited by
a fibrate singly or in combination with respect to the control)
for two (Fn þ Bz; Fn þ Gm and Bz þ Gm) and three fibrate
(Fn þ Gm þ Bz) combinations. The fa-CI plot is an effectoriented
plot that shows the evolution of the kind of interaction
(synergism, antagonism, additive effect) as a function of
the level of the effect ( fa) of a particular toxicant on the
reference organism ( fa, where ECa ¼ fa 100; i.e.,
EC 10 ¼ f10 100). In the Vibrio test (Fig. 1a), the Bz þ Gm and
Fn þ Gm binary combination showed a slight antagonism at
very low fa values and slight synergism (Fn þ Gm) or nearly
additive effects (Bz þ Gm) at the highest fa values, the Fn þ Bz
combination showed a strong antagonism at low effect levels
but the antagonism decreased and approached an additive
kind of interaction at the highest fa levels; the ternary
combination (Fn þ Gm þ Bz) showed a moderate antagonism
at low fa values that also turned into a nearly additive effect at
fa values above 0.4. Correlation analyses were made between
CI values of the fibrate ternary combination and CI values of
each of the fibrate binary combinations to determine which
binary combination interaction was predominant in the
ternary mixture (Table 3); the highest correlation coefficient
was found for the Fn þ Bz combination (r ¼ 0.91), suggesting
that this combination interaction predominated in the three
fibrate mixture. The fa-CI plot of the Anabaena test (Fig. 1b)
showed the opposite pattern of interactions as the three
binary and the ternary combinations showed from slight to
strong synergism at the lowest fa values that turned into a very
strong antagonism at fa values over 0.5; the ternary combination
(Fn þ Gm þ Bz) closely followed the interaction pattern
of the binary Fn þ Gm combination, this is confirmed by the
highest correlation coefficient found between the CI values of
the ternary combination and the CI values of the Fn þ Gm
combination (r ¼ 0.996) which suggests that in the Anabaena
test, this particular combination seemed to be the predominant
in the ternary toxicological interaction. Selected average
CI values for both Vibrio fischeri and Anabaena CPB4337 tests at
three representative dose levels (EC10, EC50 and EC90) and the
combined effects are summarized in Table 2.
3.2. Toxicological interactions of wastewater and
fibrate combinations in the Anabaena CPB4337
bioluminescence test
In a recent previous study (Rosal et al., 2009), we found that
a wastewater sample collected from a local STP was very toxic
to Anabaena cells with a wastewater dilution of 0.11 causing
50% luminescence inhibition (wastewater EC50). The observed
toxicity was attributed to the combined toxicities of over thirty
micropollutants, which included fibrates as well as other
pharmaceuticals (Rosal et al., 2008). We sought to investigate
the nature of the interaction between the wastewater (WW)
and the three fibrates in binary (Fn þ WW; Bz þ WW and
Gm þ WW) and quaternary (Fn þ Gm þ Bz þ WW) combinations;
for these experiments, the wastewater itself was
regarded as a toxicant; the experimental design was analogous
to the one for the three fibrate interactions and is also shown
in Table 1. The r-values of the median-effect plots were >0.95
in all cases, indicating that the data conformed to the medianeffect
principle (not shown). Fig. 2 shows the fa-CI plot for each

Table 3 – Correlation analyses between CI values of fibrate ternary and fibrate D wastewater quaternary combinations ( y)
and their binary combinations (x) for Vibrio fischeri and Anabaena CPB4337 tests.
Test organism Combinations Regression parameters
of the binary fibrate-wastewater combination and the
quaternary combination; as can be observed, in a broad range
of fa values, the binary combinations showed a strong synergism;
however, at fa values above 0.8, the binary Fn þ WW and
Bz þ WW combinations approached an additive effect and at
fa values above 0.95, these two combinations yielded antagonism;
by contrast, the Gm þ WW combination became even
more synergistic. The quaternary combination interaction
showed a strong synergism through a broad range of fa values
but also turned into slight antagonism at fa values above 0.95,
Combination index, CI
closely resembling the pattern of the Fn þ WW and Bz þ WW
interactions which is confirmed by the highest r value
(r ¼ 0.999) in the correlation analyses (Table 3), which suggests
a predominant effect of Fn and Bz in the quaternary
interaction.
The computer software CompuSyn (Chou and Martin, 2005)
displays a type of graphic termed polygonogram, which is
a semiquantitative method of representing interactions
between three or more compounds at a determined fa value.
This graphic allows a simplified visual presentation of the
overall results. Fig. 3 shows the polygonogram for the three
fibrates and the wastewater at four fa values; synergism is
indicated by solid lines and antagonism by broken ones; the
thickness of the lines indicates the strength of the interaction.
The polygonogram clearly shows the synergistic interaction of
wastewater in combination with each of the three fibrates at
low fa values and the antagonistic interaction that appeared at
the highest fa value, 0.99, for the Fn þ WW and the Bz þ WW
combinations.
The same wastewater sample collected from a local STP
was proved as responsible of stimulation of the bioluminescence
activity of Vibrio fischeri to 110–120% of that of the
control. Moreover, the EC50 values for the fibrates in the
wastewater were higher than those for fibrates in pure water
(Rosal et al., 2009). The same trend was observed comparing
the dose-effect curve parameters (Dm, m and r) for the ternary
combination (Fn þ Gm þ Bz) of fibrates in ddH 2O and wastewater.
The dose required to produce the median-effect (Dm)in
Vibrio fischeri test when (Fn þ Gm þ Bz) were solved in wastewater
was 131.936 compared to 55.6951 mg/L required when
ddH2O was employed. CI values could not be calculated for
Vibrio fischeri due to the fact that the wastewater itself was not
toxic to this bacterium; synergism or antagonism could not be
properly estimated (Chou, 2006).
4. Discussion
xo m r
V. fischeri Fn þ Gm þ Bz versus Gm þ Bz 0.614 1.77 0.83
Fn þ Bz 0.594 0.281 0.91
Fn þ Gm 0.067 1.40 0.81
Anabaena CPB4337 Fn þ Gm þ Bz versus Gm þ Bz 5.876 4.39 0.91
Fn þ Bz 2.716 3.00 0.941
Fn þ Gm 0.282 0.247 0.996
WW þ Fn þ Gm þ Bz versus Gm þ Bz 0.079 0.372 0.999
Fn þ Bz 0.199 0.246 0.998
Fn þ Gm 0.464 0.017 0.897
Fn þ WW 0.253 1.31 0.999
Gm þ WW 2.131 2.41 0.89
Bz þ WW 0.003 0.865 0.999
Fn ¼ fenofibric acid, Bz ¼ bezafibrate, Gm ¼ gemfibrozil, WW ¼ wastewater. The parameters of linear regression equations: x0 (value of y when
x ¼ 0); m (slope) and r (correlation coefficient) with all p-values of 0.001. Analyses were computed using MINITAB Release 14 for Windows.
2.0
1.5
1.0
0.5
0.0
0.0 0.2 0.4 0.6 0.8 1.0
Fractions affected, fa
Fig. 2 – Combination index plot ( fa-CI plot) for a set of three
fibrates and toxic wastewater sample in their binary and
quaternary combinations: Gn D WW (–6–), Fn D WW (–B–),
Bz D WW (–,–) and Fn D Gm D Bz D WW (–;–) for the
Anabaena CPB4337 test. CI values are plotted as a function of
the fractional inhibition of bioluminescence ( fa) by
computer simulation (CompuSyn) from fa [ 0.10 to 0.95.
CI 1 indicates synergism, additive effect and
antagonism, respectively. At least three independent
experiments with two replicates were used. The vertical
bars indicate 95% confidence intervals for CI values based on
sequential deletion analysis (SDA) (Chou and Martin, 2005).
Fn [ fenofibric acid, Bz [ bezafibrate, Gm [ gemfibrozil,
WW [ wastewater.
water research 44 (2010) 427–438 433
The three fibrates that we have used in our study are lipid
modifying agents that are effective in lowering elevated

434
Fig. 3 – Polygonograms showing the toxicological interactions of three fibrates and a toxic wastewater sample in their
binary combinations (Fn D Bz, Bz D Gm, Fn D Gm, Gm D WW, Fn D WW, Bz D WW) as calculated by CompuSyn (Chou and
Martin, 2005) for Anabaena CPB4337 test at four effect levels: fa [ 0.1 (a), fa [ 0.5 (b), fa [ 0.9 (c) and fa [ 0.99 (d). Solid lines
indicate synergism, broken lines indicate antagonism. The thickness of the line represents the strength of synergism or
antagonism. Figure generated by CompuSyn (Chou and Martin, 2005).
plasma triglycerides and cholesterol in humans (Staels et al.,
1998). These pharmaceuticals are highly used, ubiquitous and
persistent (Daughton and Ternes, 1999), they are found at ng/L
to mg/L levels in many STP effluents, surface waters, estuaries
of rivers and even in sea water (for a review, see Hernando
et al., 2007). Although non-target organisms; the continuous
release of these substances into the environment may cause
acute or chronic toxicity to the aquatic biota. Regarding
fibrates, in the recent literature there are many reports dealing
with individual toxicity of different fibrates in a range of
aquatic organisms from primary producers to consumers;
a great variability has been found in the sensitivity of the
different test organisms toward these pharmaceuticals
(Hernando et al., 2007). However, pharmaceuticals such as
fibrates do not occur singly in a polluted environment and are
usually found as mixtures, therefore, for risk assessment
strategies it is important to know the combined effects of
pharmaceuticals in non-target organisms (Teuschler, 2007).
There are two concepts widely used for the prediction of
mixture toxicity: concentration addition (CA) and independent
action (IA) (Backhaus et al., 2003; Vighi et al., 2003;
Backhaus et al., 2004; Junghans et al., 2006). CA is used for
mixtures whose components act in a similar mode of action
while IA is based on the idea of dissimilar action, meaning
that the compounds have different mechanisms of action;
however, as discussed by Cleuvers (2003) the terms similar/
water research 44 (2010) 427–438
dissimilar action may be misleading. Pharmaceuticals such as
fibrates may have the same pharmacological mechanism of
action [i.e., interaction with the binding peroxisome proliferator-activated
receptor a (PPARa)] in their target organism,
humans; however, if fibrates released in the aquatic environments
prove toxic to different non-target organisms, the
exact mechanism of toxicity (probably different to the pharmacological
mode of action) should be investigated in depth
before choosing which approach, CA or IA, to use. In fact, only
if toxicity is regarded as non-specific at all, the concept of CA
may be used although it may also have limitations. Cleuvers
(2003) found that two totally different pharmaceuticals,
a fibrate and an anti-epileptic drug, followed the concept of CA
in the Daphnia toxicity test and the concept of IA in an algal
test; both pharmaceuticals apparently shared the same nonspecific
toxic mode of action for both organisms; so it
appeared that the concept of CA or IA did not depend on
a similar/dissimilar mode of action but on the tested
organism. The author also discussed that by definition, when
using CA, substances applied below their individual noneffect
concentration (NOEC) will contribute to the total effect
of the mixture while when using IA, substances applied below
their NOEC will not contribute to the total effect of the
mixture, meaning that any combination effect will probably
be higher if the substances follow the concept of CA and this
may be misleading when considering the terms synergism or

antagonism because as also discussed by Chou (2006), synergism
or antagonism may occur independently of a similar or
dissimilar mode of action. In this context, Fent et al. (2006)
tested mixtures of different kinds of pharmaceuticals
(including fibrates) that might have estrogenic activity in
a yeast reporter system; they applied the CA model and found
that it had severe limitations when the dose-response curves
of the individual pharmaceuticals were not identical or at low
effect concentrations. As pharmaceuticals released in the
environment may have such diverse dose-effect relationships,
the lack of appropriate prediction suggests limitation of
the CA mixtures concept.
To study the nature of the combined fibrate interactions
(synergism, additive effect, antagonism) for the Vibrio fischeri
and Anabaena CPB4337 bioluminescence tests, we have followed
the combination index (CI)-isobologram equation
method of Chou (2006) and Chou and Talalay (1984); a method
widely used to study drug interactions in pharmacology. This
method may be considered a fractional analysis technique for
drug interactions (Berenbaum, 1981; Bovill, 1998) that is
independent of the mode of action and considers both the
potency (EC 50, Dm) and the shape (m) of the dose-effect curve
for each drug. The method allows prediction of synergism/
antagonism at all effect levels ( fa) for a combination of n
drugs; in contrast with the classical graphical isobologram
method (Berenbaum, 1981; Bovill, 1998) that cannot be used
for more than three compounds and have also graphical
limitations to show all effect levels. By using this method, we
have been able to determine the nature of interactions for
a wide range of effect levels of three fibrates in binary and
ternary combinations in two different bioluminescent organisms.
However, the nature of these interactions was not
uniform along the fa levels range in any of the two organisms.
In Vibrio fischeri, antagonism predominated at low and intermediate
fa levels but at the highest effect levels, interactions
became additive or slightly synergistic. In Anabaena, a dual
synergistic/antagonistic behaviour was observed with synergism
predominating at fa levels below 0.4–0.5 and strong
antagonism above these fa values. It is difficult to give an
explanation to this phenomenon because the combination
index method only allows quantitative determination of
synergism or antagonism and the elucidation of the mechanism
by which synergism or antagonism occurs is a separate
issue that needs a different kind of approach. However,
tentatively, antagonism, which could be considered the
predominant interaction in Vibrio fischeri and Anabaena, might
be explained by the structural similarity of fibrates which are
related pharmaceuticals that share a common structural
motif, a cyclic head and a hydrophobic tail (Rosal et al., 2009);
at the fa levels where antagonism is found in both organisms,
fibrates may compete with one another for the same target/
receptor sites. The slight synergism found at very high levels
in Vibrio fischeri could perhaps be explained by the fact that at
very high concentrations, fibrates may somehow combine to
increase toxicity by an unspecific way of action that is probably
not related to their pharmacological mechanism.
Perhaps, the most puzzling interaction is the observed high
synergism at very low fa levels in Anabaena; the mechanism of
such synergistic interaction is not readily apparent. One could
speculate that these fibrates at very low concentrations could
water research 44 (2010) 427–438 435
involve what Jia et al. (2009) in their extensive review of
mechanisms of drug combinations call ‘‘facilitating actions’’
that means that secondary actions of one drug enhances the
activity or level of another drug in the mixture or alternatively
‘‘complementary actions’’ when drugs act at the same target
at different sites, at overlapping sites or at different targets of
the same pathway. However, in the literature there are very
few reports on possible targets of fibrates on the prokaryotic
cell; English et al. (1994) reported that peroxisome proliferators
such as fibrates have been shown to induce cytochrome
P450BM-3 which catalyzes the hydroxylation of fatty
acids, in Bacillus megaterium. Garbe (2004) reported that
fibrates induced methyltransferase Rv0560c with a function in
the biosynthesis of isoprenoid compounds in Mycobacterium
tuberculosis; Garbe (2004) suggested that both effects may act
on the plasma membrane, modulating its properties. In
mitochondria, which have significant features that resemble
those of prokaryotes, fibrates have been found to inhibit
respiratory complex I (NDH-1 complex) and to interfere with
mitochondrial fatty acid oxidation (Scatena et al., 2007).
Whether fibrates may exert similar effects in Vibrio fischeri and
Anabaena to those observed in Bacillus or mitochondria needs
further research. In this context, we have found that, as the fa-
CI plots show, fibrate interactions do not follow the same
pattern in both bacteria, this may be due to the different origin
and position in the food web of Vibrio fischeri, a heterotrophic
marine prokaryote and Anabaena CPB4337, a recombinant
strain of an obligate phototrophic freshwater prokaryote; in
fact, Anabaena presents intracellular photosynthetic
membranes called thylakoids where several functionally
distinct NDH-1 complexes have been found with roles both in
respiration and photosynthesis (Battchikova and Aro, 2007). If
fibrates are also affecting NDH-1 complexes in Anabaena, their
effects might be very different to those in Vibrio fischeri; so,
although we have measured the same toxicity endpoint in
both bacteria, i.e., luminescence inhibition, the combined
effects of fibrates seem to depend on the test organism.
Ince et al. (1999) assessed toxic interactions of heavy
metals in binary mixtures on Vibrio fischeri and the freshwater
aquatic plant Lemna minor and found that most binary metal
mixtures exhibited only antagonistic interactions in the plant
opposed to fewer antagonistic and some synergistic interactions
in the heterotrophic bacterium. These authors also
found that in the bacterium, the nature of the interaction
(synergism or antagonism) also changed with the effect level
of the binary metal combinations, although the authors did
not provide a mechanistic explanation for this variability.
Cheng and Lu (2002) made a comparison of joint interactions
of organic toxicants in binary mixtures in Escherichia coli and
Vibrio fischeri and found that toxicants with the same mechanisms
of toxicity displayed mostly additive or antagonistic
interactions in E. coli and Vibrio fischeri; however a synergistic
interaction was found between glutardialdehyde and butyraldehyde
in Vibrio. Synergistic effects in both bacteria were
mostly associated with toxicants with different mechanisms
of toxicity, although antagonism clearly predominated. They
also found that for a total of 44 organic binary mixtures, only
six mixtures resulted in identical type of interaction in both
bacteria. From our results and those of other authors’ (Ince
et al., 1999; Cheng and Lu, 2002; Cleuvers, 2003) one may

436
conclude that previous knowledge of the mechanism of toxic
action of a compound is not useful enough to predict which
kind of interactions it will display when combined with other
toxicants with the same or different toxic mechanism; also, as
we have shown, the nature of the interaction may depend on
the effect level of the mixture. In addition, different types of
organisms will show completely different responses to
mixtures of potential toxicants.
We previously found that a local wastewater was very toxic
for the Anabaena CPB4337 test but non-toxic at all for the Vibrio
fischeri or Daphnia magna tests. This wastewater is a mixture of
over thirty micropollutants, mostly pharmaceuticals of
different therapeutics groups that, besides the fibrates used in
this study, included antibiotics, analgesics/anti-inflammatories,
b-blockers, antidepressants, anti-epileptics/psychiatrics,
ulcer healing compounds, diuretics and bronchodilators;
personal care products and some priority organic pollutants
are also present (Rosal et al., 2008). The method of Chou allows
to combine one drug mixture with another drug mixture and
determine their interactions; therefore, we studied the nature
of the interaction of fibrates and wastewater in the Anabaena
bioluminescence test; interestingly, we found that in a wide
range of effect levels, the interaction of wastewater and the
three fibrate combination was synergistic; particularly, at very
low fa values which means that fibrates are at low concentrations
and the wastewater is diluted several-fold, the method
predicted a strong synergism; this may be due, as discussed
above, to the observed synergistic interactions of fibrates with
one another as well as interactions with some of the detected
micropollutants when present at very low concentrations. This
observed synergism may be environmentally relevant since
most pharmaceuticals such as fibrates do not usually show
acute toxicity on non-target organisms when tested at real
environmental concentrations (Hernando et al., 2007) but in
a mixture, if they act synergistically, they could prove toxic for
a test organism even at low concentrations; these results agree
with those found by Hernando et al. (2004) who reported
synergistic toxic effects for Daphnia magna test when wastewater
was spiked with environmental concentrations of
several pharmaceuticals including fibrates. By contrast, our
results show that at high fa values ( fa > 0.8), the combined
interaction of the quaternary fibrates þ wastewater combination,
the binary Fn þ WW and Bz þ WW combinations
approached an additive effect and eventually became antagonistic;
in our previous study, the wastewater itself decreased
Anabaena bioluminescence by 84% with a lower confidence
limit of 76% and an upper confidence limit of 91%; when the
wastewater was spiked with increasing concentrations of each
fibrate we found that, with the exception of gemfibrozil, the
EC50 values for the fibrates in the wastewater were higher than
those for fibrates in pure water; this was attributed either to
reduced bioavailability or to antagonistic effects of fibrates
with other chemicals present in the wastewater; although we
did not use the method of Chou, we obtained similar results to
the ones we report in this study; that is, at high effect levels
(>84% luminescence inhibition) the interaction of fibrates with
wastewater, except the Gm þ WW combination, showed
antagonism.
Based on our results, we propose that the combination
index (CI)-isobologram equation, a method widely used in
water research 44 (2010) 427–438
pharmacology both for in vitro and in vivo bioassays, may also
be applied in environmental toxicology as a general method to
define interactions of potential toxicants in mixtures in any
test organism and/or toxicological endpoint of interest and
could be especially useful for risk assessment strategies that
take into account the toxicological interactions of substances
in a mixture.
5. Conclusions
We report an environmental application of the combination
index (CI)-isobologram equation to study the nature of the
interactions of fibrate combinations in two bioluminescent
aquatic organisms. The method allowed calculating synergism
or antagonism of binary and ternary fibrate combinations at all
effect levels simultaneously; we could also test the method
with a real wastewater sample in binary and quaternary
combination with the fibrates, finding that at very low effect
levels, the fibrates acted synergistically with the wastewater in
the Anabaena test. The proposed method may be used with
other test organisms and/or toxicological endpoints and could
be particularly useful for risk assessment approaches to
toxicity of complex mixtures.
Acknowledgements
The research was funded by the Spanish Ministry of Education
through grants CTM2005-03080/TECNO and CSD2006-00044
and the Comunidad de Madrid, grants 0505/AMB-0395 and
0505/MB/0321.
references
Allen, M.B., Arnon, D.I., 1955. Studies on nitrogen-fixing blue
green algae. I Growth and nitrogen fixation by Anabaena
cylindrica Lemm. Plant Physiol 30 (4), 366–372.
Andreozzi, R., Raffaele, M., Nicklas, P., 2003. Pharmaceuticals in
STP effluents and their solar photodegradation in aquatic
environment. Chemosphere 50 (10), 1319–1330.
Backhaus, T., Altenburger, R., Arrhenius, A˚ ., Blanck, H., Faust, M.,
Finizio, A., Gramatica, P., Grote, M., Junghans, M., Meyer, W.,
Pavan, M., Porsbring, T., Scholze, M., Todeschini, R., Vighi, M.,
Walter, H., Horst Grimme, L., 2003. The BEAM-project:
prediction and assessment of mixture toxicities in the aquatic
environment. Continental Shelf Res. 23 (17–19), 1757–1769.
Backhaus, T., Arrhenius, A., Blanck, H., 2004. Toxicity of a mixture
of dissimilarly acting substances to natural algal
communities: predictive power and limitations of
independent action and concentration addition. Environ. Sci.
Technol 38 (23), 6363–6370.
Battchikova, N., Aro, E.-M., 2007. Cyanobacterial NDH-1
complexes: multiplicity in function and subunit composition.
Physiol. Plant 131 (1), 22–32.
Berenbaum, M.C., 1981. Criteria for analyzing interactions
between biologically active agents. Adv. Cancer Res. 35,
269–335.
Borgmann, U., Bennie, D.T., Ball, A.L., Palabrica, V., 2007. Effect of
a mixture of seven pharmaceuticals on Hyalella azteca over
multiple generations. Chemosphere 66 (7), 1278–1283.

Bovill, J.G., 1998. Analysis of drug interactions. Baillière’s Clin.
Anaesthesiol 12 (2), 135–168.
Brain, R.A., Johnson, D.J., Richards, S.M., Hanson, M.L.,
Sanderson, H., Lam, M.W., Young, C., Mabury, S.A., Sibley, P.K.,
Solomon, K.R., 2004. Microcosm evaluation of the effects of an
eight pharmaceutical mixture to the aquatic macrophytes Lemna
gibba and Myriophyllum sibiricum. Aquat. Toxicol 70 (1), 23–40.
Castiglioni, S., Bagnati, R., Fanelli, R., Pomati, F., Calamari, D.,
Zuccato, E., 2006. Removal of pharmaceuticals in sewage
treatment plants in Italy. Environ. Sci. Technol 40 (1),
357–363.
Chen, C.-Y., Lu, C.-L., 2002. An analysis of the combined effects of
organic toxicants. Sci. Total Environ 289 (1–3), 123–132.
Chou, T.C., 1976. Derivation and properties of Michaelis–Menten
type and Hill type equations for reference ligands. J. Theor.
Biol. 59 (2), 253–276.
Chou, T.C., 2006. Theoretical basis, experimental design, and
computerized simulation of synergism and antagonism in
drug combination studies. Pharmacol. Rev. 58 (3), 621–681.
Chou, T.C., Martin, N., 2005. CompuSyn for Drug Combinations:
PC Software and User’s Guide: A Computer Program for
Quantification of Synergism and Antagonism in Drug
Combinations and the Determination of IC 50 and ED 50 and
LD 50 Values. ComboSyn, Inc., Paramus, NJ.
Chou, T.C., Talalay, P., 1984. Quantitative analysis of dose-effect
relationships: the combined effects of multiple drugs or
enzyme inhibitors. Adv. Enzyme Regul 22, 27–55.
Christensen, A.M., Faaborg-Andersen, S., Ingerslev, F., Baun, A.,
2007. Mixture and single-substance toxicity of selective
serotonin reuptake inhibitors toward algae and crustaceans.
Environ. Toxicol. Chem. 26 (1), 85–91.
Cleuvers, M., 2003. Aquatic ecotoxicity of pharmaceuticals
including the assessment of combination effects. Toxicol. Lett.
142 (3), 185–194.
Cleuvers, M., 2004. Mixture toxicity of the anti-inflammatory
drugs diclofenac, ibuprofen, naproxen, and acetylsalicylic
acid. Ecotoxicol. Environ. Saf 59 (3), 309–315.
Comerton, A.M., Andrews, R.C., Bagley, D.M., Yang, P., 2007.
Membrane adsorption of endocrine disrupting compounds
and pharmaceutically active compounds. J. Memb. Sci. 303
(1–2), 267–277.
Crane, M., Watts, C., Boucard, T., 2006. Chronic aquatic
environmental risks from exposure to human
pharmaceuticals. Sci. Total Environ 367 (1), 23–41.
Daughton, C.G., Ternes, T.A., 1999. Pharmaceuticals and personal
care products in the environment: agents of subtle change?
Environ. Health Perspect 107 (6), 907–938.
English, N., Hughes, V., Wolf, C.R., 1994. Common pathways of
cytochrome P450 gene regulation by peroxisome proliferators
and barbiturates in Bacillus megaterium ATCC14581. J. Biol.
Chem. 269 (43), 26836–26841.
Fent, K., Escher, C., Caminada, D., 2006. Estrogenic activity of
pharmaceuticals and pharmaceutical mixtures in a yeast
reporter gene system. Reprod. Toxicol 22 (2), 175–185.
Garbe, T.R., 2004. Co-induction of methyltransferase Rv0560c by
naphthoquinones and fibric acids suggests attenuation of
isoprenoid quinone action in Mycobacterium tuberculosis. Can. J.
Microbiol. 50 (10), 771–778.
Gros, M., Petrovic, M., Barcelo, D., 2007. Wastewater treatment
plants as a pathway for aquatic contamination by
pharmaceuticals in the Ebro river basin (northeast Spain).
Environ. Toxicol. Chem. 26 (8), 1553–1562.
Han, G.H., Hur, H.G., Kim, S.D., 2006. Ecotoxicological risk of
pharmaceuticals from wastewater treatment plants in Korea:
occurrence and toxicity to Daphnia magna. Environ. Toxicol.
Chem. 25 (1), 265–271.
Hernando, M.D., Petrovic, M., Fernández-Alba, A.R., Barceló, D.,
2004. Analysis by liquid chromatography-electrospray
water research 44 (2010) 427–438 437
ionization tandem mass spectrometry and acute toxicity
evaluation for beta-blockers and lipid-regulating agents in
wastewater samples. J. Chromatogr. A 1046 (1–2), 133–140.
Hernando, M.D., Agüera, A., Fernández-Alba, A.R., 2007. LC-MS
analysis and environmental risk of lipid regulators. Anal.
Bioanal. Chem. 387 (4), 1269–1285.
Holoshitz, N., Alsheikh-Ali, A.A., Karas, R.H., 2008. Relative safety
of gemfibrozil and fenofibrate in the absence of concomitant
cerivastatin use. Am. J. Cardiol 101 (1), 95–97.
Ince, N.H., Dirilgen, N., Apikyan, I.G., Tezcanli, G., Üstün, B., 1999.
Assessment of toxic interactions of heavy metals in binary
mixtures: a statistical approach. Arch. Environ. Contam.
Toxicol 36, 365–372.
International Organization for Standardization, 2007. Water
quality - Determination of the inhibitory effect of water
samples on the light emission of Vibrio fischeri (Luminescent
bacteria test), ISO 11348-3 Revised version, Geneva,
Switzerland.
Jia, J., Zhu, F., Ma, X., Cao, Z.W., Li, Y.X., Chen, Y.Z., 2009.
Mechanisms of drug combinations: interaction and network
perspectives. Nat. Rev. Drug Discov 8 (2), 111–128.
Junghans, M., Backhaus, T., Faust, M., Scholze, M., Grimme, L.H.,
2006. Application and validation of approaches for the
predictive hazard assessment of realistic pesticide mixtures.
Aquat. Toxicol 76 (2), 93–110.
Kasprzyk-Hordern, B., Dinsdale, R.M., Guwy, A.J., 2009. The
removal of pharmaceuticals, personal care products,
endocrine disruptors and illicit drugs during wastewater
treatment and its impact on the quality of receiving waters.
Water Res. 43 (2), 363–380.
Lambropoulou, D.A., Hernando, M.D., Konstantinou, I.K.,
Thurman, E.M., Ferrer, I., Albanis, T.A., Fernandez-Alba, A.R.,
2008. Identification of photocatalytic degradation products of
bezafibrate in TiO 2 aqueous suspensions by liquid and gas
chromatography. J. Chromatogr. A 1183 (1–2), 38–48.
Metcalfe, C.D., Miao, X.S., Koenig, B.G., Struger, J., 2003.
Distribution of acidic and neutral drugs in surface waters near
sewage treatment plants in the lower Great Lakes, Canada.
Environ. Toxicol. Chem. 22 (12), 2881–2889.
Miller, D.B., Spence, J.D., 1998. Clinical pharmacokinetics of fibric
acid derivatives (fibrates). Clin. Pharmacokinet 34 (2), 155–162.
Pomati, F., Orlandi, C., Clerici, M., Luciani, F., Zuccato, E., 2008.
Effects and interactions in an environmentally relevant
mixture of pharmaceuticals. Toxicol. Sci. 102 (1), 129–137.
Quinn, B., Gagne, F., Blaise, C., 2009. Evaluation of the acute,
chronic and teratogenic effects of a mixture of eleven
pharmaceuticals on the cnidarian, Hydra attenuata. Sci. Total
Environ 407 (3), 1072–1079.
Rodea-Palomares, I., González-García, C., Leganés, F., Fernández-
Piñas, F., 2009. Effect of pH, EDTA, and anions on heavy
metal toxicity toward a bioluminescent cyanobacterial
bioreporter. Arch. Environ. Contam. Toxicol. doi:10.1007/
s00244-008-9280-9.
Rodríguez, A., Rosal, R., Perdigón, J.A., Mezcua, M., Agüera, A.,
Hernando, M.D., Letón, P., Fernández-Alba, A.R., García-
Calvo, E., 2008. In: Barceló, D., Petrovic, M. (Eds.), Emerging
Contaminants from Industrial and Municipal Waste. Springer-
Verlag, Berlin, pp. 127–175.
Rosal, R., Rodríguez, A., Perdigón-Melón, J.A., Mezcua, M.,
Hernando, M.D., Letón, P., García-Calvo, E., Agüera, A.,
Fernández-Alba, A.R., 2008. Removal of pharmaceuticals and
kinetics of mineralization by O 3/H 2O 2 in a biotreated
municipal wastewater. Water Res. 42 (14), 3719–3728.
Rosal, R., Rodea-Palomares, I., Boltes, K., Fernández-Piñas, F.,
Leganés, F., Gonzalo, S., Petre, A., 2009. Ecotoxicity assessment
of lipid regulators in water and biologically treated
wastewater using three aquatic organism. Environ. Sci. Pollut.
Res. doi:10.1007/s11356-009-0137-1.

438
Scatena, R., Bottoni, P., Botta, G., Martorana, G.E., Giardina, B.,
2007. The role of mitochondria in pharmacotoxicology:
a reevaluation of an old, newly emerging topic. Am. J. Physiol.
Cell Physiol 293 (1), C12–C21.
Staels, B., Dallongeville, J., Auwerx, J., Schoonjans, K., Leitersdorf, E.,
Fruchart, J.C., 1998. Mechanism of action of fibrates on lipid and
lipoprotein metabolism. Circulation 98 (19), 2088–2093.
Stumpf, M., Ternes, T.A., Wilken, R.D., Rodrigues, S.V., Baumann, W.,
1999. Polar drug residues in sewage and natural waters in the
state of Rio de Janeiro. Brazil. Sci. Total Environ 225 (1–2), 135–141.
water research 44 (2010) 427–438
Teuschler, L.K., 2007. Deciding which chemical mixtures risk
assessment methods work best for what mixtures. Toxicol.
Appl. Pharmacol 223 (2), 139–147.
Vighi, M., Altenburger, R., Arrhenius, A., Backhaus, T.,
Bodeker,W.,Blanck,H.,Consolaro,F.,Faust,M.,
Finizio, A., Froehner, K., Gramatica, P., Grimme, L.H.,
Gronvall, F., Hamer, V., Scholze, M., Walter, H., 2003.
Water quality objectives for mixtures of toxic chemicals:
problems and perspectives. Ecotoxicol. Environ. Saf 54 (2),
139–150.

Toxication or detoxication? In vivo toxicity assessment of
ozonation as advanced wastewater treatment with the
rainbow trout
Daniel Stalter a, *, Axel Magdeburg a , Mirco Weil b , Thomas Knacker b ,Jörg Oehlmann a
a
Goethe University Frankfurt am Main, Biological Sciences Division, Department of Aquatic Ecotoxicology, Siesmayerstrasse 70, 60054
Frankfurt, Hessen, Germany
b
ECT Oekotoxikologie GmbH, Böttgerstrasse 2–4, 65439 Flörsheim, Germany
article info
Article history:
Received 23 March 2009
Received in revised form
16 July 2009
Accepted 18 July 2009
Available online 25 July 2009
Keywords:
Sewage treatment
Pharmaceuticals
Oxidation byproducts
Vitellogenin
Emerging contaminants
Advanced oxidation process
Rainbow trout
Fish early life stage toxicity test
1. Introduction
abstract
Wastewater (WW) is one of the major sources of micropollutants
introduced into the aquatic environment
(Schwarzenbach et al., 2006). The large spectrum of pharmaceuticals
and personal care products (PPCPs) occurring in
water research 44 (2010) 439–448
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
* Corresponding author. Tel.: þ49 69 7982 4882; fax: þ49 69 7982 4748.
E-mail address: stalter@bio.uni-frankfurt.de (D. Stalter).
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.watres.2009.07.025
Ozonation as advanced wastewater treatment method is an effective technique for
micropollutant removal. However, the application of this method carries the inherent
danger to produce toxic oxidation byproducts. For an ecotoxicological assessment
conventionally treated wastewater, wastewater after ozonation and ozonated wastewater
after sand filtration were evaluated in parallel at an operating treatment plant via the fish
early life stage toxicity test (FELST) using rainbow trout (Oncorhynchus mykiss).
The FELST revealed a considerable developmental retardation of test organisms
exposed to ozonated WW. This was accompanied by a significant decrease in body weight
and length compared to reference water, to the conventionally treated WW and to the
ozonated water after sand filtration. Hence sand filtration obviously prevents from adverse
ecotoxicological effects of ozonation.
An additional test with yolk-sac larvae resulted in a significant reduction of vitellogenin
levels in fish exposed to ozonated wastewater compared to fish reared in conventionally
treated wastewater. This demonstrates the effective removal of estrogenic activity by
ozonation.
Adverse ozonation effects may have been a result of the conversion of chemicals into
more toxic metabolites. However, sand filtration reduced toxication effects indicating that
these oxidation byproducts are readily degradable or adsorbable. The results indicate that
in any case ozonation should not be applied without subsequent post treatment appropriate
for oxidation byproducts removal (e.g. sand filtration).
ª 2009 Elsevier Ltd. All rights reserved.
surface waters (Daughton and Ternes, 1999), particularly with
regard to mixture toxicity, may pose a potential threat to
aquatic wildlife. But also single substances may have the
capability to endanger the ecosystem as for example diclofenac
can lead to an impairment of the general health condition
of rainbow trout at environmentally relevant concentrations

440
Nomenclature
AOC assimilable organic carbon
C control
COD chemical oxygen demand
DOC dissolved organic carbon
EE2 17a-ethinylestradiol
ELISA enzyme linked immunosorbent assay
EQS environmental quality standard
FELST fish early life stage toxicity test
FS wastewater after final sedimentation
GJIC gap junctional intercellular communication
MS222 tricaine methanesulfonate
NH4–N ammonium nitrogen
(Schwaiger et al., 2004; Triebskorn et al., 2004). 17a-ethinylestradiol
(EE2) is considered to be the most potent estrogenic
active compound for fish with the lowest observed effect
concentration of 0.1 ng/L for vitellogenesis in rainbow trout
(Purdom et al., 1994). Moreover it leads to a reduced egg fertilisation
success and skewed sex ratio toward females of fat
head minnows at a LOEC of 0.32 ng/L (Parrott and Blunt, 2005).
Furthermore, industrial surfactants like alkylphenolic ethoxylates
are likely to be responsible for feminization effects in
rainbow trout (Jobling et al., 1996) and complex mixtures of
estrogenic chemicals present in the environment have been
shown to act additively (Brian et al., 2005). The concentrations
of such compounds in WW often exceed effect concentrations
(Ternes et al., 1999; Ying et al., 2002; Tixier et al., 2003) and
thus even treated WW contains high amounts of estrogenic
substances that impair the endocrine system of exposed fish
(Larsson et al., 1999; Rodgers-Gray et al., 2001) resulting in
feminization effects of males and consequently reduced
fertility of field populations (Jobling et al., 2002).
Besides, in the European Union the reduction of the
contamination of surface waters with hazardous substances
is defined by the Water Framework Directive requiring a ‘‘good
status’’ for all coastal and inland waters until 2015 (European
Commission, 2000). One tool is the implementation of environmental
quality standards (EQSs) for mono substance
pollutants exhibiting a significant risk to the aquatic
ecosystem. The discharge of such compounds, classified as
priority substances, is envisaged to be progressively reduced
by 2025 or 5 years after inclusion in the list for priority
substances, respectively.
However, till now so called ‘‘emerging contaminants’’ are
only marginally addressed and PPCPs are not included in this
list. But the latter has to be reviewed at least every 4 years and
provisions have been made to add several hazardous
emerging contaminants. Amongst others the inclusion of
diclofenac, EE2 and carbamazepine is discussed and additionally
recommended by the European Parliament since 2007
(Jensen, 2007). Moreover EQSs meanwhile have been proposed
for several PPCPs based on reliable ecotoxicity data (e.g.
diclofenac: 0.1 mg/L, EE2: 0.03 ng/L, carbamazepine: 0.5 mg/L;
Jahnel et al., 2006; Moltmann et al., 2007). Unfortunately, some
PPCPs exceed the EQSs in surface waters of some urbanised
water research 44 (2010) 439–448
NO 3–N nitrate nitrogen
O wastewater after ozonation
OECD Organisation for Economic Cooperation and
Development
OS wastewater after ozonation and sand filtration
P phosphate
PAH polycyclic aromatic hydrocarbons
PAC powdered activated carbon
PPCPs pharmaceuticals and personal care products
rcf relative centrifugal force
SPM suspended particulate matter
VTG vitellogenin
WW wastewater
WWTP wastewater treatment plant
regions. That is, inter alia, a result of the fact that conventional
WW treatment plants are not designed for appropriate
removal of PPCPs (Daughton and Ternes, 1999). Nevertheless
in many countries surface waters serve as raw water resources
for drinking water supply. Due to the public’s demand for safe
drinking water and ecosystem health advanced WW treatment
methods are required to allow for sufficient removal of
micropollutants from sewage treatment effluents. Therefore,
end of pipe techniques like activated carbon filtration and
ozonation of WW might be crucial to achieve regulative
requirements in a medium-term perspective.
Huber et al. (2005) demonstrated that ozonation of
secondary effluent is an effective tool for the removal of
a wide range of pharmaceuticals among them diclofenac,
carbamazepine and the estrogens estradiol, estrone and EE2,
with degradation rates of more than 90% for ozone doses of
2 mg/L. Merely iodinated X-ray contrast media and a few
acidic pharmaceuticals were oxidized only partially. Nakada
et al. (2007) confirmed the effective elimination of 22 pharmaceuticals
during ozonation in an operating treatment
plant. A further advantage of the ozonation is the sanitizing
property of the method (Tyrrell et al., 1995). The effective
removal of micropollutants and pathogens indicates that
ozonation is a suitable technique for advanced WW treatment
to reduce the contamination of the aquatic environment.
Actually a reduced toxicity of a mixture of six pharmaceuticals
treated by ozonation towards algae and rotifer was
demonstrated by Andreozzi et al. (2004). Reduced estrogenic
activity of an EE2-containing WW after ozonation was shown
by Huber et al. (2004) and ozone treatment of WW significantly
decreased the toxic potency of urban pollutants to the freshwater
mussel Elliptio complanata (Gagne et al., 2007). However,
ozone treatment typically transforms chemical compounds
but does not mineralize them entirely. Consequently, not only
the concentrations of mother compounds but also the inclusion
of transformation products and their mixtures should be
assessed for toxicity, too.
For a risk-benefit analysis an extensive ecotoxicological
evaluation of ozonated WW is essential. In vitro bioassays
covering different modes of toxic action (e.g. estrogenicity,
acetylcholine esterase activity or non-specific baseline
toxicity) and performed with enriched water samples are

Table 1 – Wastewater quality summary (mg/L).
SPM COD P NH4–N NO3–N pH
Final sedimentation
Median 4.8 17 0.19 0.1 8.3
10% percentile 3.6 15 0.17 0.04 5.4 8.12
90% percentile 8.96 23.8 0.22 0.27 10.1 8.4
n 37 23 37 37 2 23
Ozonation and sand filtration
Median 2 15 0.17 0.04 9.8 8.3
10% percentile 1.6 12.2 0.14 0.04 7.28 8.02
90% percentile 2.4 19.8 0.19 0.15 12.48 8.4
n 37 23 37 37 37 23
suitable tools for toxicity characterization of WW as shown by
Escher et al. (2008). These methods were shown to be highly
sensitive even for identification and characterization of low
toxic water samples. Nevertheless, as typical screening tools
they are not designed to replace chronic in vivo tests of whole
effluent samples. To allow for a more comprehensive and
integrative assessment of the potentially hazardous impact of
WW ozonation we applied the fish early life stage toxicity test
(FELST) with rainbow trout (Oncorhynchus mykiss).
2. Materials and methods
Ozonated and conventionally treated WW were tested in
parallel on site at the wastewater treatment plant (WWTP)
Wüeri (Regensdorf, Switzerland) with the fish early life stage
toxicity test (FELST) in a flow-through system.
2.1. Characterization of the wastewater treatment plant
The municipal WWTP Wüeri operates experimentally with
a full scale ozonation reactor after final sedimentation and
with a sand filtration step after the ozone reactor. Table 1
shows the WW quality parameters. The water parameters
after the final sedimentation and after the ozone reactor are
on the same level and therefore exemplarily shown for final
sedimentation water. Low ammonium and phosphate
concentrations indicate that the treatment plant is already
working well. The dissolved organic carbon (DOC) ranged from
5.4 to 5.9 mg/L and the pH was nearly constant in all test
waters. The WWTP serves for a population equivalent of
25,000. The median discharge in the experimental period was
6190 m 3 /day (10th percentile: 4430 m 3 /day, 90th percentile:
10,500, n ¼ 109) and the applied ozone concentration was in
a range between 0.4 and 1 mg O 3/mg DOC.
water research 44 (2010) 439–448 441
2.2. Experimental setup
WW from three different sampling points of serial treatment
steps was tested (Fig. 1): after the final sedimentation (FS),
after the ozone reactor (O) and after additional sand filtration
(OS).
Test waters were passively transported through high-grade
steel pipes to aerated high-grade steel reservoir tanks. The
retention time in the pipes was adjusted to be at least 45 min
to avoid ozone residuals reaching the exposure vessels.
Indeed no ozone was detected by indigo blue method (Bader
and Hoigne, 1981) in the reservoir tank during maximum
required flow-through and during maximum applied ozone
concentration (1 mg O3/mg DOC). From reservoir tanks test
waters were transported through polytetrafluoroethylene
tubes via a peristaltic pump (IPC24, Ismatec, Wertheim–
Mondfeld, Germany) to the exposure vessels each equipped
with a passive discharge device and tempered using
a temperature-controlled water bath. Constant temperature
conditions in the water bath were achieved with a flowthrough
cooling system (Van der Heijden, Dörentrup,
Germany). The flow-through rates in the exposure vessels
ranged from 11 mL per minute up to 44 mL per minute (2–8 fold
water exchange per day in the exposure vessels) depending on
the fish size to match the loading rate criteria (OECD, 1992b).
Reconstituted water according to OECD guideline 203 (1992a)
was used as control water (C).
The FELST was performed with the rainbow trout (Oncorhynchus
mykiss) according to OECD guideline 210 (1992b) with
a constant water temperature of 10 2 C as well as darkness
for embryo development and 12 2 C with 12/12 h light/dark
photoperiod post hatch. Sixty newly fertilized eggs per replicate
were exposed to the test waters for 65 days in high-grade
steel 10 L tanks. One test series was performed with unfiltered
WW and a second with membrane filtered WW (pore size:
0.4 mm, Kubota Corp., Osaka, Japan) to minimize microbial
impacts. Macromolecules and organic compounds were not
retained. However, it has to be considered that membrane
filtration additionally removes suspended particulate matter
and consequently all particle bound pollutants. The filter
membranes were placed in the reservoir vessels.
A third test was performed with yolk-sac fry (5 days post
hatch, 30 larvae per exposure vessel) and non-filtered test
waters because we postulated a reduced sensitivity to pathogen
contamination of the larvae compared to the egg stage.
The test duration was 64 days. All tests were performed with
undiluted WW to increase the probability to detect differences
between the treatment groups. With the beginning of swim up
(the swim up process marks a developmental transition from
larval stage to juvenile fish stage and is characterized with the
Fig. 1 – Sampling points at the wastewater treatment plant. Abbreviations: FS, after final sedimentation; O, after the ozone
reactor; OS, after sand filtration.

442
beginning of exogenous ingestion) the fish were fed four times
per day (trout starter, 4% body weight per day).
The tests were run with four (C) and three (FS, O, OS)
replicates in the first FELST with unfiltered WW and two (C,
OS) and three replicates (FS, O) per test water in the remaining
two tests. The latter were performed in parallel and therefore
it was not possible to use consistently three replicates per
WW-group due to limited space capacities. In each test replicates
were placed randomized in the water bath. Observations
on egg coagulation, hatching, mortality, swim up, malformation
and abnormal behaviour were recorded daily. Fish were
humanely killed by MS222 (tricaine methanesulfonate, Sigma–
Aldrich, St. Louis, USA) overdose. Individual fish were blotted
dry, weight and body length were determined. Afterwards fish
were frozen in liquid nitrogen and stored at 80 C until
vitellogenin detection in whole body homogenates.
2.3. Vitellogenin detection
Whole body homogenates of 11 fish (from the third test with
yolk-sac fry) per replicate were prepared as described by
Holbech et al. (2006) with slight modifications. Aliquots of 0.3 g
frozen fish, excised between head and pectoral fin, were
mixed with 10 fold of the body weight of homogenisation
buffer (50 mM Tris–HCl pH 7.4; 1% protease inhibitor cocktail
(P 8340, Sigma–Aldrich, St. Louis, USA)) and homogenated
with a dispersing apparatus (T18 basic Ultra–Turrax Ò , IKA,
Staufen, Germany). The homogenate was centrifuged for
30 min at 20,000 rcf and the supernatant was used for vitellogenin
analysis. Vitellogenin (VTG) was detected with
a rainbow trout vitellogenin ELISA test kit (Biosense, Bergen,
Norway) using a 1:20 dilution.
2.4. Statistical analysis
Complete statistical analyses (Kruskal–Wallis with Dunn’s
post test, Fisher’s exact test, non-linear regression with variable
slope) were performed using GraphPad Prism version 5.0
for windows (GraphPad software, San Diego, CA, USA). All
given error values indicate the standard error (SE) and in all
figures error bars display the SE. Kruskal–Wallis with Dunn’s
post test was chosen to test on significant differences because
data were not normally distributed in all cases. For quantal
data Fisher’s exact test was applied (mortality, swim up,
coagulation). When toxicity endpoints could be analysed
individually for each test animal (biomass, body length,
vitellogenin), the collected data are presented and statistically
evaluated on a per specimen basis.
3. Results
3.1. FELST with unfiltered wastewater
Unfiltered WW caused an increased coagulation rate of the
exposed eggs in the FELST (Fig. 2). The eggs exposed to WW
after final sedimentation (FS) were completely coagulated
after 18 days with a 50% coagulation time of 12.1 days. The
coagulation of the eggs exposed to WW after the ozone reactor
(O) was considerably delayed compared to FS. However the
water research 44 (2010) 439–448
Fig. 2 – Cumulative coagulation of Oncorhynchus mykiss
eggs (mean values ± SE) exposed to differently treated
wastewaters. Abbreviations: C, control water; FS, final
sedimentation; O, ozonation; OS, ozonation and sand
filtration; SE, standard error. Time scale is logarithmized.
After 40 days all treatment groups differ significantly from
each other (Fisher’s exact test: p < 0.001; n [ 240 (C), 180
(FS, O, OS)).
coagulation after 40 days achieves still an average rate of
87.2 4.8% with a 50% coagulation time of 17.6 days. The
lowest coagulation rate in the WW treatments occurred after
sand filtration (OS; 64.4 4.0% after 40 days; 50% coagulation
time: 25.8 days). After 10–15 days exposure, fungus mycelia
(first appearing in the FS vessels) were observed in all WW
exposure vessels. Mycelia were found on and between eggs as
well as vorticellas on the eggs whereas the reference water (C)
remained observably free from mycelia and vorticellas.
Hatching success in the control group (53.4 5.2%) did not
meet validity criteria (>66%; OECD guideline 210).
3.2. FELST with membrane filtered wastewater
To exclude microbial impairment the second FELST was performed
with membrane filtered WW and eggs were obtained
from another fish hatchery.
All test vessels remained free from microbial contamination
throughout the test duration. The egg coagulation rates in
the WW treatment groups of this experimental series were
reduced compared to the first FELST with a maximum of 25%
(O, OS) and only 20.1 1.1% after FS (Fig. 3A). Egg coagulation
was significantly increased and hatching success significantly
decreased in the WW treatment groups compared to the
control ( p < 0.05, Fisher’s exact). The coagulation rates were
slightly but not significantly increased in O and OS compared
to FS (Fig. 3A). The hatching progress was slightly delayed in O
compared to FS and OS but hatching success achieved at least
75% in all treatments (Fig. 3B). The control group met the
validity criteria according to OECD guideline 210 (egg coagulation:
10.0 1.7%, hatching success: 90.0 1.7%).

Fig. 3 – Cumulative coagulation of eggs (A) and cumulative hatching of larvae (B) from Oncorhynchus mykiss (mean
values ± SE) exposed to differently treated and membrane filtered wastewaters. Abbreviations: C, control water; FS, final
sedimentation; O, ozonation; OS, ozonation and sand filtration; SE, standard error. Time scale is logarithmized. After 40 and
36 days, respectively all wastewater treatment groups are significantly different from the control (Fisher’s exact test:
p < 0.01–0.05; n [ 120 (C, OS), 180 (FS, O)).
Fig. 4 shows the cumulative swim up of hatched fish
beginning after 45 days of exposure. The swim up is considerably
delayed in all WW treatment groups compared to the
control. This effect is most notable in O, even if compared to
Fig. 4 – Cumulative swim up of Oncorhynchus mykiss larvae
(mean values ± SE) exposed to differently treated and
membrane filtered wastewaters. Abbreviations: C, control
water; FS, final sedimentation; O, ozonation; OS, ozonation
and sand filtration; SE, standard error. Time scale is
logarithmized. After 64 days the O group differs
significantly from the other treatment groups (Fisher’s
exact test: p < 0.001–0.01; n [ 95 (C), 116 (FS), 103 (O), 75
(OS)).
water research 44 (2010) 439–448 443
FS and OS. At the end of the experiment only 83.3 1.6% of the
fish swam up in O while 100% swam up in C and FS and
97.6 2.4% in OS. Hereby the swim up success after 64 days is
significantly decreased in O compared to the other treatment
groups ( p < 0.01, Fisher’s exact). The 50% swim up time in the
control is 47.4 1.0 days, which is only slightly increased in FS
and OS (50.2 1.0; 49.3 1.0) but obviously increased in O
(57.5 1.04).
The biomass as well as the body length of fish is significantly
decreased ( p < 0.001, Kruskal–Wallis with Dunn’s post
test) in all WW treatments compared to the control (Fig. 5).
Both endpoints in fish exposed to O are furthermore significantly
decreased compared to the FS group ( p < 0.001) and
significantly decreased compared to OS ( p < 0.05).
Generally the mortality is comparatively low in all WW
treatment groups (Fig. 6), as they are fulfilling the validity
criteria for controls according to OECD guideline 210 (survival
after hatch 70%). Nevertheless the mortality in the WW
treatment groups is slightly increased compared to C. The
highest and significantly increased mortality rate was detected
in the ozonated water (24.0 4.0%, p < 0.05, Fisher’s exact).
3.3. Fish test starting with yolk-sac fry & vitellogenesis
The fish test starting with the yolk-sac stage revealed no
statistically significant differences in development between
WW treatments and the control. The biomass exhibited no
major deviations between exposure groups (

444
Fig. 5 – Biomass (A) and body length (B) in percentage relative to the control (0.28 ± 0.01 g and 34.1 ± 0.2 mm, respectively) of
Oncorhynchus mykiss (mean values ± SE) exposed to differently treated and membrane filtered wastewaters. Abbreviations:
C, control water; FS, final sedimentation; O, ozonation; OS, ozonation and sand filtration; SE, standard error. Significant
differences to the control are indicated with white asterisks, between treatments with black asterisks (Kruskal–Wallis with
Dunn’s post test: +, p < 0.05; +++, p < 0.001; n [ 73–114).
Kruskal–Wallis with Dunn’s post test) compared to fish maintained
in reference water (11.0 6.6 ng/mL) whereas it was
significantly decreased in the O and OS groups (4.6 1.9 and
4.8 1.9 ng/mL, respectively; p 0.01) compared to FS (Fig. 7).
4. Discussion
4.1. FELST with unfiltered wastewater
Unfiltered WW led to an increased coagulation rate of the
exposed eggs in the FELST (Fig. 2). High coagulation rates
Fig. 6 – Mortality of Oncorhynchus mykiss post hatch (mean
values ± SE) exposed to differently treated and membrane
filtered wastewaters. Abbreviations: C, control water; FS,
final sedimentation; O, ozonation; OS, ozonation and sand
filtration; SE, standard error. Significant differences to
control are indicated with white asterisks (Fisher’s exact
test: +, p < 0.05; n [ 108 (C), 143 (FS), 135 (O), 90 (OS)).
water research 44 (2010) 439–448
presumably are a result of microbial contamination as the
reference water (C) remained free from mycelia and vorticellas
whereas in the WW treatment groups fungi mycelia and
vorticellas were found after 10–15 days exposure. However,
the hatching success in the control (53.4 5.2%) did not
comply with the validity criteria (>66%) according to OECD
guideline (1992b) indicating that egg quality was not sufficient
to run a valid test. The disinfectant effect of ozonation (Tyrrell
et al., 1995) is likely to be responsible for the delayed egg
Fig. 7 – Whole body vitellogenin concentration of
Oncorhynchus mykiss (mean values ± SE) after 60 days
exposure to differently treated wastewaters starting with
the yolk-sac stage. Abbreviations: FS, final sedimentation;
O, ozonation; OS, ozonation and sand filtration; C, control
water; SE, standard error. Significant differences to control
are indicated with white asterisks, between treatments
with black asterisks (Kruskal–Wallis with Dunn’s post test:
+, p < 0.05; ++, p < 0.01; n [ 22 (C, OS) – 33 (FS, O)).

coagulation in the O group compared to FS. Ozonation is also
known to produce high amounts of assimilable organic
carbons (AOC) which may have allowed a fast regrowth of
microorganisms (Huang et al., 2005), resulting in still high
coagulation rates in the O group. Sand filtration reduces the
amount of suspended particulate matter (SPM) and of AOC
(Wang and Summers, 1996). This may have reduced microbial
development and resulting coagulation rates.
4.2. FELST with membrane filtered wastewater
Exposure to membrane filtered WW revealed a distinct delay
of the swim up process in the O group compared to FS and
OS (Fig. 4) accompanied by a significant decrease in body
weight and body length (Fig. 5). The reduced biomass and
body length after ozonation is most likely a result of the
delayed development because fish start exogenous ingestion
at the onset of the swim up process. Consequently trout that
swim up later may suffer from developmental disadvantages.
Oxidative byproducts in ozonated WW may have impeded
embryonic and/or larval development of fish. Possibly
because sand filtration is able to remove ozonation metabolites
(e.g. aldehydes, glyoxal, AOC; Wang and Summers, 1996)
the effect in OS was reduced. However, no single compound
could be clearly identified for the retardation effect. Possibly
the sum of aldehydes (e.g. formaldehyde, glyoxal, acetaldehyde),
carboxylic acids (e.g. formate), ketones and brominated
organic compounds formed due to ozonation (Huang
et al., 2005; Wert et al., 2007) caused these effects. Unfortunately,
no chronic toxicity data for juvenile rainbow trout are
available in literature for these compounds. Nevertheless,
Wang and Summers (1996) were able to demonstrate that
sand filtration efficiently removes aldehydes produced
during ozone application, supporting this assumption.
Besides, the formation of more complex organic metabolites
evoked by ozonation may result in an increased toxicity
compared to chemical precursors as it has already been
documented for polycyclic aromatic hydrocarbons (PAHs). In
the experiments of Luster-Teasley et al. (2002, 2005) chrysene
and pyrene and byproducts as a result of ozonation were
examined for their ability to disrupt gap junctional intercellular
communication (GJIC), an indicator for tumor
promoting properties. Among the transformation products,
aldehydic compounds exhibited an increased toxicity
compared to the precursor substance while their carboxylic
structure analogue showed no GJIC disrupting activity. For
the antiepileptic drug carbamazepine McDowell et al. (2005)
identified three new oxidation products of unknown toxicity
after ozone treatment. Furthermore, a possible increase in
mutagenicity after ozonation of WW, as observed by Monarca
et al. (2000) with the Ames test, verifies the potential of
ozonation to produce toxic oxidation byproducts. In this
context conversion of the widely used fungicide tolylfluanide
into the carcinogen N-nitrosodimethylamine during ozonation
was discovered by Schmidt and Brauch (2008). Based on
these results it is likely that ozonation of WW leads to an
increased number of unknown and potentially toxic metabolites
depending on the composition of WW at the outset and
the post treatment (e.g. sand filtration). Petala et al. (2006)
demonstrated with the Vibrio fischeri bioluminescence test,
water research 44 (2010) 439–448 445
that toxicity originating from ozonated WW decreases with
increasing storage time, indicating a rapid decomposition of
toxic metabolites.
In spite of the developmental retardation and the reduced
biomass no significant differences in mortality rates were
observed between the WW treatment groups (Fig. 6). Mortality
was highest directly after ozonation (O, 24.0 4.0%). Considering
the developmental retardation and the decreased
biomass of fish exposed to ozonated water the increased
mortality of O group specimens compared to FS and OS is
noticeable and, although not significant, potentially a result of
a reduced fitness of the fish in the ozonated water.
It might be assumed that the retarded development is
a consequence of an unspecific and general impairment of the
fish’s health condition. This may increase their sensitivity
towards environmental and anthropogenic stressors, thus
leading to an increased mortality. Furthermore, under field
conditions the delayed development possibly increases the
risk for the fish to fall prey to predators since before swim up
the larvae rest on the bottom, not capable to abscond effectively.
Based on the assumption that the compounds causing
adverse effects after ozonation are readily degradable it is
possible that the detoxication after the ozone reactor will
occur in the river as well. However, the discharge of ozonated
WW without sand filtration would bear the risk to endanger
fish populations in a considerable range of the receiving river,
depending on WW load and flow velocity. Therefore, ozonation
should always be followed by a sand filtration and further
studies should focus on whether sand filtration is capable to
remove oxidation byproducts sufficiently.
4.3. Fish test starting with yolk-sac fry & vitellogenesis
The fish test with non-filtered WW starting with the yolk-sac
stage resulted in negligible and insignificant deviations of
development and biomass between treatment groups. Ozonation
had no effect on these toxicity endpoints. These results
indicate that yolk-sac fry 5 days post hatch are less sensitive
to ozonation metabolites, compared to fish embryos and
newly hatched fish, respectively. Nevertheless, the amount of
suspended particulate matter in the exposure vessels and in
the Teflon tubes was accompanied by an increased biofilm
formation. This may have contributed to an increased detoxication
of the ozonated WW due to (bio-) degradation of
oxidation byproducts.
The significant increase of vitellogenin content in fish
exposed to FS compared to the reference water indicates an
environmentally relevant contamination of the WW with
estrogenic active compounds (Fig. 7). After ozonation the VTG
content decreased even below the control level. The formation
of antiestrogenic compounds is not likely because the VTG
decrease compared to control level is not significant and
phenols, as an important functional group interacting with
the estrogen receptor (Nishihara et al., 2000), are particularly
susceptible to ozone attack (von Gunten, 2003). These results
confirm the high efficiency of ozonation to eliminate estrogenic
contamination in WW as already shown by Huber et al.
(2004) with an in vitro test system. Based on these results it
can be assumed that ozonation is well suited to reduce the
estrogenic burden of WW below environmental relevance.

446
Regarding ecosystem health it should be considered that
estrogen and xenoestrogen concentrations detected in the
environment have been shown to impair the sustainability of
wild fish populations (Kidd et al., 2007), arguing for the
establishment of an end of pipe technique for estrogen
removal to protect fish populations in surface waters receiving
high amounts of WW.
4.4. Ozonation vs. alternatives
Currently there is a controversy regarding the appropriate
advanced WW treatment techniques to reduce contamination
of the aquatic ecosystem with micropollutants. Alternative
processes other than ozonation also deliver promising results
with regard to the removal of contaminants. Wastewater
treatment with powdered activated carbon (PAC) achieves
reduction rates of 75–90% for pharmaceuticals (including
X-ray contrast media) with PAC dosages of 10–20 mg/L
(Püttmann et al., 2008). The main advantage of PAC treatment
is that a reduced chemical concentration is equivalent to
removal of chemicals whereas ozonation leads basically to
a transformation of the compounds. According to cost estimations
PAC treatment is expected to be about 30% more
expensive than ozonation (Joss et al., 2008). However, the
main disadvantage of PAC treatment might be the need for the
disposal of the used and contaminated carbon. Membrane
filtration is suitable for micropollutants retention but as
a result of considerably higher requirement for energy and
technical equipment economically not competitive with
ozonation or activated carbon treatment (Joss et al., 2008).
Overall, end of pipe techniques are presumably a suitable
solution to reduce toxicity of hazardous WW in a mediumterm
perspective. Nevertheless in the long term source control
strategies such as wastewater separation (e.g. urine separation),
ecologically correct disposal of drugs by end users, reuse
or recycling by the pharmaceutical industry or alternative
medical treatments to drug therapies (Daughton, 2003; Joss
et al., 2006) could offer environmentally friendly and affordable
options.
5. Conclusions
The sanitizing effect of ozonation was confirmed. Coagulation
rates of eggs exposed to ozonated wastewater were on
a lower level compared to those exposed to conventionally
treated wastewater. Disinfection is presumably only efficient
when linked to sand filtration because of rapid
recovery of microorganisms due to increased assimilable
organic carbon formation as a result of ozonation.
Membrane filtered wastewater (for removal of microorganisms)
reveals developmental retardation directly after
ozonation. After sand filtration this adverse effect disappears.
The reduced biomass and body length in fish exposed
to ozonated wastewater is most probably a result of the
formation of toxic oxidation byproducts.
The mortality in the ozonated wastewater was significantly
increased as a result of retarded development. Impairment of
the fish’s health condition may increase the sensitivity
towards environmental and anthropogenic stressors.
water research 44 (2010) 439–448
Developmental retardation might increase the risk for the fish
to fall prey to predators because swim up stage is delayed.
Ozonation should not be applied without appropriate
barrier for oxidation byproducts. Effectiveness of sand
filtration for removal of oxidation byproducts should be
further evaluated.
Reduction of vitellogenin content in fish exposed to ozonated
wastewater on control level confirms the suitability of
this technique to reduce estrogenic activity, possibly below
environmental relevance.
Acknowledgements
The authors should like to express their gratitude to the staff
from WWTP Wüeri for their technical cooperation and
Adriano Joss from the EAWAG for helpful suggestions and
technical support. Ulrike Schulte-Oehlmann is acknowledged
for critical comments and helpful suggestions on the manuscript.
This study was part of the EU project Neptune (contract
no 036845, SUSTDEV-2005-3.II.3.2) within the Energy, Global
Change and Ecosystems Programme of the Sixth Framework
(FP6-2005-Global-4) and co-funded by Bundesamt für Umwelt
(BAFU), Bern (CH) within the Strategy MicroPoll Programme
(contract no 05.0013.PJ/F471-0916).
references
Andreozzi, R., Campanella, L., Fraysse, B., Garric, J., Gonnella, A.,
Lo Giudice, R., Marotta, R., Pinto, G., Pollio, A., 2004. Effects of
advanced oxidation processes (AOPs) on the toxicity of
a mixture of pharmaceuticals. Water Science and Technology
50 (5), 23–28.
Bader, H., Hoigne, J., 1981. Determination of ozone in water by the
indigo method. Water Research 15 (4), 449–456.
Brian,J.V.,Harris,C.A.,Scholze,M.,Backhaus,T.,Booy,P.,
Lamoree,M.,Pojana,G.,Jonkers,N.,Runnalls,T.,Bonfa,A.,
Marcomini, A., Sumpter, J.P., 2005. Accurate prediction of
the response of freshwater fish to a mixture of estrogenic
chemicals. Environmental Health Perspectives 113 (6),
721–728.
Daughton, C.G., 2003. Cradle-to-cradle stewardship of drugs for
minimizing their environmental disposition while
promoting human health. II. Drug disposal, waste reduction,
and future directions. Environmental Health Perspectives
111 (5), 775–785.
Daughton, C.G., Ternes, T.A., 1999. Pharmaceuticals and personal
care products in the environment: agents of subtle change?
Environmental Health Perspectives 107, 907–938.
Escher, B.I., Bramaz, N., Quayle, P., Rutishauser, S.,
Vermeirssen, E.L.M., 2008. Monitoring of the ecotoxicological
hazard potential by polar organic micropollutants in sewage
treatment plants and surface waters using a mode-of-action
based test battery. Journal of Environmental Monitoring 10 (5),
622–631.
European Commission, 2000. Directive 2000/60/EC of the
European Parliament and of the Council of 23 October 2000
establishing a framework for Community action in the field
of water policy. Official Journal of the European
Communities L327. Retrieved from: http://eur-lex.europa.eu/

LexUriServ/LexUriServ.do?uri¼OJ: L:2000:327:0001:0072:EN:
PDF [12.02.2009].
Gagne, F., Andre, C., Cejka, P., Gagnon, C., Blaise, C., 2007.
Toxicological effects of primary-treated urban wastewaters,
before and after ozone treatment, on freshwater mussels
(Elliptio complanata). Comparative Biochemistry and Physiology
C – Toxicology & Pharmacology 145 (4), 542–552.
Holbech, H., Kinnberg, K., Petersen, G.I., Jackson, P., Hylland, K.,
Norrgren, L., Bjerregaard, P., 2006. Detection of endocrine
disrupters: evaluation of a fish sexual development
test (FSDT). Comparative Biochemistry and Physiology
C – Toxicology & Pharmacology 144 (1), 57–66.
Huang, W.J., Fang, G.C., Wang, C.C., 2005. The determination
and fate of disinfection by-products from ozonation of
polluted raw water. Science of the Total Environment 345
(1-3), 261–272.
Huber, M.M., Ternes, T.A., von Gunten, U., 2004. Removal of
estrogenic activity and formation of oxidation products during
ozonation of 17a-ethinylestradiol. Environmental Science &
Technology 38 (19), 5177–5186.
Huber, M.M., Gobel, A., Joss, A., Hermann, N., Loffler, D.,
McArdell, C.S., Ried, A., Siegrist, H., Ternes, T.A., von
Gunten, U., 2005. Oxidation of pharmaceuticals during
ozonation of municipal wastewater effluents: a pilot study.
Environmental Science & Technology 39 (11), 4290–4299.
Jahnel, J., Neamtu, M., Schudoma, D., Frimmel, F.H., 2006.
Scientific risk assessment of considered water relevant
substances (Bestimmung von Umweltqualitaetsnormen fuer
potenziell gewaesserrelevante Stoffe). Acta Hydrochimica et
Hydrobiologica 34 (4), 389–397.
Jensen, G.K., 2007. European Parliament increases number of
chemicals earmarked for priority action by European
Commission. The Lancet Oncology 8 (5), 382–383.
Jobling, S., Sheahan, D., Osborne, J.A., Matthiessen, P.,
Sumpter, J.P., 1996. Inhibition of testicular growth in rainbow
trout (Oncorhynchus mykiss) exposed to estrogenic
alkylphenolic chemicals. Environmental Toxicology and
Chemistry 15 (2), 194–202.
Jobling, S., Coey, S., Whitmore, J.G., Kime, D.E., Van Look, K.J.W.,
McAllister, B.G., Beresford, N., Henshaw, A.C., Brighty, G.,
Tyler, C.R., Sumpter, J.P., 2002. Wild intersex roach (Rutilus
rutilus) have reduced fertility. Biology of Reproduction 67 (2),
515–524.
Joss, A., Klaschka, U., Knacker, T., Liebig, M., Lienert, J., Ternes, T.A.,
Wennmalm, A., 2006. Source control, source separation. In:
Ternes, T.A., Joss, A. (Eds.), Human Pharmaceuticals, Hormones
and Fragrances. The Challenge of Micropollutants in Urban
Water Management. ISBN 1843390930, pp. 353–384.
Joss, A., Siegrist, H., Ternes, T.A., 2008. Are we about to upgrade
wastewater treatment for removing organic micropollutants?
Water Science and Technology 57 (2), 251–255.
Kidd, K.A., Blanchfield, P.J., Mills, K.H., Palace, V.P., Evans, R.E.,
Lazorchak, J.M., Flick, R.W. (2007). Collapse of a fish population
after exposure to a synthetic estrogen. Proceedings of the
National Academy of Sciences of the United States of America
104(21), 8897–8901.
Larsson, D.G.J., Adolfsson-Erici, M., Parkkonen, J.,
Pettersson, M., Berg, A.H., Olsson, P.E., Forlin, L., 1999.
Ethinylestradiol – an undesired fish contraceptive? Aquatic
Toxicology 45 (2-3), 91–97.
Luster-Teasley, S.L., Yao, J.J., Herner, H.H., Trosko, J.E., Masten, S.J.,
2002. Ozonation of chrysene: evaluation of byproduct mixtures
and identification of toxic constituent. Environmental Science
& Technology 36 (5), 869–876.
Luster-Teasley, S.L., Ganey, P.E., DiOrio, M., Ward, J.S.,
Maleczka, R.E., Trosko, J.E., Masten, S.J., 2005. Effect of
byproducts from the ozonation of pyrene: biphenyl-2,2 0 ,6,6 0 -
tetracarbaldehyde and biphenyl-2,2 0 ,6,6 0 -tetracarboxylic acid
water research 44 (2010) 439–448 447
on gap junction intercellular communication and neutrophil
function. Environmental Toxicology and Chemistry 24 (3),
733–740.
McDowell, D.C., Huber, M.M., Wagner, M., Von Gunten, U.,
Ternes, T.A., 2005. Ozonation of carbamazepine in drinking
water: identification and kinetic study of major oxidation
products. Environmental Science & Technology 39 (20),
8014–8022.
Moltmann, J.F., Liebig, M., Knacker, T., Keller, M., Scheurer, M.,
Ternes, T., 2007. Relevance of Endocrine Disrupting
Substances and Pharmaceuticals in Surface Waters
(Gewässerelevanz endokriner Stoffe und Arzneimittel). Final
report. Federal Environment Agency. Funding code 20524205.
Monarca, S., Feretti, D., Collivignarelli, C., Guzzella, L., Zerbini, I.,
Bertanza, G., Pedrazzani, R., 2000. The influence of different
disinfectants on mutagenicity and toxicity of urban
wastewater. Water Research 34 (17), 4261–4269.
Nakada, N., Shinohara, H., Murata, A., Kiri, K., Managaki, S.,
Sato, N., Takada, H., 2007. Removal of selected
pharmaceuticals and personal care products (PPCPs) and
endocrine-disrupting chemicals (EDCs) during sand filtration
and ozonation at a municipal sewage treatment plant. Water
Research 41 (19), 4373–4382.
Nishihara, T., Nishikawa, J., Kanayama, T., Dakeyama, F., Saito, K.,
Imagawa, M., Takatori, S., Kitagawa, Y., Hori, S., Utsumi, H.,
2000. Estrogenic activities of 517 chemicals by yeast two-hybrid
assay. Journal of Health Science 46 (4), 282–298.
OECD, 1992a. Fish Acute Toxicity Test. OECD Guidelines for the
Testing of Chemicals. No. 203. Organisation for Economic
Cooperation and Development, Paris.
OECD, 1992b. Fish Early Life Stage Toxicity Test. OECD Guidelines
for the Testing of Chemicals. No. 210. Organisation for
Economic Cooperation and Development, Paris.
Parrott, J.L., Blunt, B.R., 2005. Life-cycle exposure of fathead
minnows (Pimephales promelas) to an ethinylestradiol
concentration below 1 ng/L reduces egg fertilization success
and demasculinizes males. Environmental Toxicology 20 (2),
131–141.
Petala, M., Samaras, P., Zouboulis, A., Kungolos, A.,
Salkellaropoulos, G., 2006. Ecotoxicological properties of
wastewater treated using tertiary methods. Environmental
Toxicology 21 (4), 417–424.
Püttmann, W., Keil, F., Oehlmann, J., Schulte-Oehlmann, U., 2008.
Strategy to reduce pharmaceuticals in drinking water –
technical approach (Wassertechnische Strategien zur
Reduzierung der Trinkwasserbelastung durch
Arzneimittelwirkstoffe). Umweltwissenschaften und
Schadstoff-Forschung 20 (3), 209–226.
Purdom, C.E., Hardiman, P.A., Bye, V.V.J., Eno, N.C., Tyler, C.R.,
Sumpter, J.P., 1994. Estrogenic effects of effluents from sewage
treatment works. Chemical Ecology 8 (4), 275–285.
Rodgers-Gray, T.P., Jobling, S., Kelly, C., Morris, S., Brighty, G.,
Waldock, M.J., Sumpter, J.P., Tyler, C.R., 2001. Exposure of
juvenile roach (Rutilus rutilus) to treated sewage effluent
induces dose-dependent and persistent disruption in gonadal
duct development. Environmental Science & Technology 35
(3), 462–470.
Schmidt, C.K., Brauch, H.J., 2008. N, N-dimethosulfamide as
precursor for N-nitrosodimethylamine (NDMA) formation
upon ozonation and its fate during drinking water treatment.
Environmental Science & Technology 42 (17), 6340–6346.
Schwaiger, J., Ferling, H., Mallow, U., Wintermayr, H.,
Negele, R.D., 2004. Toxic effects of the non-steroidal antiinflammatory
drug diclofenac Part 1: histopathological
alterations and bioaccumulation in rainbow trout. Aquatic
Toxicology 68 (2), 141–150.
Schwarzenbach, R.P., Escher, B.I., Fenner, K., Hofstetter, T.B.,
Johnson, C.A., von Gunten, U., Wehrli, B., 2006. The challenge

448
of micropollutants in aquatic systems. Science 313 (5790),
1072–1077.
Ternes, T.A., Stumpf, M., Müller, J., Haberer, K., Wilken, R.D.,
Servos, M., 1999. Behavior and occurrence of estrogens in
municipal sewage treatment plants – I. Investigations in
Germany, Canada and Brazil. Science of the Total
Environment 225 (1-2), 81–90.
Tixier, C., Singer, H.P., Oellers, S., Müller, S.R., 2003. Occurrence
and fate of carbamazepine, clofibric acid, diclofenac,
ibuprofen, ketoprofen, and naproxen in surface waters.
Environmental Science & Technology 37 (6), 1061–1068.
Triebskorn, R., Casper, H., Heyd, A., Eikemper, R., Köhler, H.R.,
Schwaiger, J., 2004. Toxic effects of the non-steroidal antiinflammatory
drug diclofenac Part II. Cytological effects in
liver, kidney, gills and intestine of rainbow trout (Oncorhynchus
mykiss). Aquatic Toxicology 68 (2), 151–166.
water research 44 (2010) 439–448
Tyrrell, S.A., Rippey, S.R., Watkins, W.D., 1995. Inactivation of
bacterial and viral indicators in secondary sewage
effluents, using chlorine and ozone. Water Research 29
(11), 2483–2490.
von Gunten, U., 2003. Ozonation of drinking water: Part I.
Oxidation kinetics and product formation. Water Research 37
(7), 1443–1467.
Wang, J.Z., Summers, R.S., 1996. Biodegradation behavior of
ozonated natural organic matter in sand filters. Revue des
Sciences de l’Eau 9 (1), 3–16.
Wert, E.C., Rosario-Ortiz, F.L., Drury, D.D., Snyder, S.A., 2007.
Formation of oxidation byproducts from ozonation of
wastewater. Water Research 41 (7), 1481–1490.
Ying, G.G., Williams, B., Kookana, R., 2002. Environmental fate of
alkylphenols and alkylphenol ethoxylates – a review.
Environment International 28 (3), 215–226.

The role of organic matter in the removal of emerging
trace organic chemicals during managed aquifer recharge
T. Rauch-Williams, C. Hoppe-Jones, J.E. Drewes*
Advanced Water Technology Center (AQWATEC), Colorado School of Mines, Environmental Science and Engineering Division,
Golden, CO 80401-1887, United States
article info
Article history:
Received 20 April 2009
Received in revised form
11 August 2009
Accepted 20 August 2009
Available online 27 August 2009
Keywords:
Trace organic chemicals (TOrC)
Groundwater recharge
Effluent organic matter
Managed aquifer recharge
Riverbank filtration
Biotransformation
Co-metabolism
Primary substrate
1. Introduction
abstract
Managed aquifer recharge (MAR) systems, such as riverbank
filtration (RBF) and soil aquifer treatment (SAT), are widely
used natural processes for drinking water augmentation
projects using source water that might be impaired by
wastewater discharge. Previous studies have demonstrated
that MAR systems are effective in dampening and reducing
water research 44 (2010) 449–460
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
* Corresponding author.
E-mail address: jdrewes@mines.edu (J.E. Drewes).
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.watres.2009.08.027
This study explored the effect of different bulk organic carbon matrices on the fate of trace
organic chemicals (TOrC) during managed aquifer recharge (MAR). Infiltration through
porous media was simulated in biologically active column experiments under aerobic and
anoxic recharge conditions. Wastewater effluent derived organic carbon types, differing in
hydrophobicity and biodegradability (i. e., hydrophobic acids, hydrophilic carbon, organic
colloids), were used as feed substrates in the column experiments. These carbon substrates
while fed at the same concentration differed in their ability to support soil biomass growth
during porous media infiltration. Removal of degradable TOrC (with the exception of
diclofenac and propyphenazone) was equal or better under aerobic versus anoxic porous
media infiltration conditions. During the initial phase of infiltration, the presence of
biodegradable organic carbon (BDOC) enhanced the decay of degradable TOrC by
promoting soil biomass growth, suggesting that BDOC served as a co-substrate in a cometabolic
transformation of these contaminants. However, unexpected high removal
efficiencies were observed for all degradable TOrC in the presence of low BDOC concentrations
under well adopted oligotrophic conditions. It is hypothesized that removal under
these conditions is caused by a specialized microbial community growing on refractory
carbon substrates such as hydrophobic acids. Findings of this study reveal that the
concentration and character of bulk organic carbon present in effluents affect the degradation
efficiency for TOrC during recharge operation. Specifically aerobic, oligotrophic
microbiological soil environments present favorable conditions for the transformation of
TOrC, including rather recalcitrant compounds such as chlorinated flame retardants.
ª 2009 Elsevier Ltd. All rights reserved.
the concentrations of dissolved organic carbon (DOC) as well
as various trace organic contaminants (TOrC) that might be
present in impaired source waters (Drewes and Fox, 1999;
Brauch et al., 2000; Grünheid et al., 2005). The presence of
TOrC has become a key concern for drinking water augmentation
projects during the past decade (Kolpin et al., 2002;
Heberer, 2002; Focazio et al., 2008). Although adverse human
health effects caused by these compounds at concentrations

450
commonly observed in impaired water resources are very
unlikely (Schwab et al., 2005), minimizing exposure of
wastewater derived contaminants in these projects is desired.
Previous research on the fate of TOrC during MAR has
primarily focused on collecting anecdotal and site specific
information on their occurrence and removal (Drewes et al.,
2002; Montgomery-Brown et al., 2003; Grünheid et al., 2005;
Dillon et al., 2008). Studies delineating the mechanisms and
boundary conditions for the transformation of wastewater
derived TOrC during MAR are lacking.
Previous studies demonstrated that the type and bioavailability
of effluent organic matter (EfOM) controls the extent of
soil biomass growth in MAR systems (Rauch and Drewes, 2004;
Rauch and Drewes, 2005). EfOM may consequently effect the
degradation of TOrC by serving as a co-substrate in microbiologically
facilitated transformations (Stratton et al., 1983).
The diversity and expression of the soil microbial community
also depends on composition and concentration of the organic
carbon substrate controlling trophic cycles in the subsurface
(Preuß and Nehrkorn, 1996; Szewzyk et al., 1998). The
composition of EfOM (i.e. in terms of its bioavailability) is
primarily determined by the degree of wastewater treatment
employed (Drewes and Fox, 1999), which can vary widely from
primary to conventional to advanced wastewater treatment.
As a result, effluent qualities fed to MAR systems can vary in
biodegradable dissolved organic carbon (BDOC) concentrations
from less than 1 up to 15 mg/L or more. As a consequence,
soil microbial communities growing on different
levels of BDOC can differ widely in total biomass and diversity.
The objectives of this research were to investigate the role
of 1) abiotic vs. biotic conditions, 2) BDOC and 3) the type of
organic carbon matrices on the removal of select TOrC, such
as pharmaceutical residues, personal care products, and
household chemicals, during MAR.
2. Methodology
2.1. Target organic contaminants
Compounds selected for this study represent small molecular
weight organic chemicals (180 to 360 Dalton) that are hydrophilic
at neutral pH regimes as indicated by an octanol/water
partition coefficient at pH 7 (log DpH¼7 of less than 2.6). TOrC
with these properties have a high potential to migrate into
groundwater and are not expected to be adsorbed onto porous
media. The molecular structures and physicochemical properties
of the target compounds chosen for this study are presented
in Table 1. These compounds cover a wide range of
biodegradability as previously reported for soil/water systems.
The anticonvulsants carbamazepine and primidone have
been classified as recalcitrant during wastewater treatment
and MAR in earlier studies (Heberer, 2002; Drewes et al., 2003;
Clara et al., 2004). The chlorinated phosphate esters tris
(1-chloroisopropyl)-phosphate (TCPP) and tris(2-chloroethyl)phosphate
(TCEP) are two widely used flame retardants and
are persistent in the aquatic environment (Heberer et al., 2001;
Fries and Püttmann, 2003). During bank filtration in Germany,
however, TCPP and TCEP exhibited a significant reduction,
which was attributed to biotransformation in the aquifer
water research 44 (2010) 449–460
(Heberer et al., 2003). Amy and Drewes (2006) also reported
removal of TCEP to concentrations below the detection limit
after two years of subsurface travel in an MAR facility supporting
that chlorinated flame retardants can be biotransformed
under anoxic conditions. Propyphenazone is
a poorly biodegradable analgesic that persists during RBF
(Heberer et al., 2003) and SAT (Drewes et al., 2003). For diclofenac,
a popular analgesic drug, low removal due to biodegradation
or adsorption was reported (Buser et al., 1998; Möhle
et al., 1999) unless soils contain a high organic carbon content
(Drillia et al., 2003). Several studies report a faster degradation
of diclofenac under anoxic conditions as compared to aerobic
conditions (Zwiener and Frimmel, 2003; Hua et al., 2003).
Opposing results were reported by Schmidt et al. (2004) in that
diclofenac was almost completely removed during aerobic
bank filtration but recalcitrant during anaerobic recharge.
Ibuprofen, ketoprofen, and naproxen are common analgesics
that are well degradable during wastewater treatment (Buser
et al., 1999; Zwiener and Frimmel, 2003; Carballa et al., 2004)
and during soil infiltration (Sedlak and Pinkston, 2001; Drewes
et al., 2003). Gemfibrozil is a commonly prescribed blood lipid
regulator found in unconfined shallow aquifers impacted by
wastewater infiltration in the low to moderate ng/L-concentration
range (Heberer and Stan, 1997; Drewes and Shore,
2001; Heberer, 2002). Gemfibrozil was removed below the limit
of detection within a few weeks during groundwater recharge
using SAT (Drewes et al., 2003).
2.2. Analytical methods
2.2.1. GC/MS analysis
The following TOrC were analyzed by gas chromatography
coupled with mass spectrometry (GC/MS) using a HP 6890
gas chromatograph and a HP 5973 quadrupole mass spectrometer
from Agilent Technologies (Waldbronn, Germany)
adopting a method published by Reddersen and Heberer
(2003): gemfibrozil, primidone, diclofenac sodium salt,
carbamazepine, ketoprofen, naproxen, phenacetine, tris
(2-chloroethyl)phosphate (TCEP) (Sigma Aldrich Chemicals),
tris(chloropropyl)phosphate (TCPP), and propyphenazone
(Pfaltz & Bauer, Inc.). Stock solutions were prepared by dissolving
the compounds in milli-Q water adjusted to a pH of
10, during sonification and in the dark. A volume of 500 mL
of sample was collected and filtered (0.45 mm, Whatman)
prior to solid phase extraction. 10,11-dihydrocarbamazepine
(Sigma Aldrich Chemicals) and 2-(m-chlorophenoxy) propionic
acid (Sigma Aldrich Chemicals) were used as surrogate
standards. The limits of detection (LofD) and limit of
quantification (LoQ) for the target compounds ranged from 5
to 10 ng/L and from 10 to 50 ng/L, respectively. Only parent
target compounds were investigated in this study, metabolites
and conjugates were not analyzed.
2.2.2. HPLC analysis
A high performance liquid chromatography (HPLC) HP 1100
(Agilent Technologies) combined with a UV diode array detector
(DAD) was used to quantify concentrations of primidone, phenacetine,
carbamazepine, naproxen, diclofenac, and ibuprofen
in the lower mg/L range during adsorption breakthrough tests.
The samples were directly injected without filtration or other

Table 1 – Physical–chemical properties of target organic compounds.
Compound Structure pKa log P
(log DpH7)
Carbamazepine
(anticonvulsant)
Diclofenac sodium
(analgesic/antiinflammatory)
Gemfibrozil (blood
lipid regulator)
Ibuprofen
(analgesic/antiinflammatory)
Ketoprofen
(analgesic/antiinflammatory)
Naproxen
(analgesic/antiinflammatory)
Phenacetine
(antipyretic)
Primidone
(anticonvulsant)
Propyphenazone
(analgesic)
water research 44 (2010) 449–460 451
Expected
degradability
13.9 2.67 Persistent
4.0
4.8
4.4
4.2
4.0
n/a
12.3
n/a
4.06
(1.06)
4.39
(2.19)
3.72
(1.12)
2.81
(0.01)
3.00
(0.00)
1.63
(1.63)
0.4
(0.4)
1.74
(1.74)
Degradable
(anoxic)
Relatively
persistent
Degradable
(better
aerobic)
Degradable
Degradable
Relatively
persistent
Persistent
Poor/
persistent
References
Mersmann et al.,
2002; Schmidt
et al., 2004
Buser et al., 1998;
Möhle et al., 1999;
Drillia et al., 2003;
Zwiener and Frimmel,
2003, Hua et al.,
2003; Schmidt et al., 2004
Heberer and Stan, 1997;
Drewes and Shore, 2001;
Heberer, 2002; Drewes et al., 2003
Buser et al., 1999;
Zwiener and Frimmel,
2003; Carballa et al., 2004;
Sedlak and Pinkston, 2001;
Drewes et al., 2003
Buser et al., 1999;
Zwiener and Frimmel,
2003; Carballa et al., 2004;
Sedlak and Pinkston, 2001;
Drewes et al., 2003
Buser et al., 1999;
Zwiener and Frimmel,
2003; Carballa et al., 2004;
Sedlak and Pinkston, 2001;
Drewes et al., 2003
Buser et al., 1999;
Zwiener and Frimmel,
2003; Carballa et al., 2004;
Sedlak and Pinkston, 2001;
Drewes et al., 2003
Buser et al., 1999;
Zwiener and Frimmel,
2003; Carballa et al., 2004;
Sedlak and Pinkston, 2001;
Drewes et al., 2003
Heberer et al., 2003;
Drewes et al., 2003
(continued on next page)

452
Table 1 (continued)
Compound Structure pKa log P
(log DpH7)
Tris(2chloroethyl)phosphate
(TCEP) (flame
retardant)
sample preparation. The mobile phase consisted of two
solvents: 1) type 1 water (adjusted to pH 2.5 with o-phosphoric
acid (HPLC grade, Fisher Scientific) and buffered with 25 mmol
potassium phosphate monobasic (Fisher Scientific)), and 2)
acetonitrile (HPLC grade, Mallinckrodt ChromAR HPLC) with the
same concentration of o-phosphoric acid as solvent 1). Sample
injection (100 mL) occurred in triplicates for each sample at a flow
rate of 2 mL/min. A mobile phase gradient was applied during
the run for solvent 2 (acetonitrile) of 30% at t ¼ 0 min., 80% at
t ¼ 12 min., and 30% at t ¼ 15 min. with a post-run time of 7 min.
for column cleaning (30% of solvent 2). The detection wavelengths
(band width 10 nm) were set to 205 nm at t ¼ 0min.for
quantification of primidone, phenacetine, and carbamazepine,
230 nm at t ¼ 6.8 min. (naproxen), and 205 nm at 7.8 min. for
detection of diclofenac and ibuprofen with a reference wavelength
of 330 nm (band width 60 nm) for all compounds. Standards
were run in the range of 5–500 mg/L. The detection limits
for each compound were 5 mg/L.
2.2.3. DOC/UV absorbance/nitrate/ammonium
A Sievers 800 Total Organic Carbon Analyzer (GE, Boulder)
was used for DOC quantification after microfiltration
(0.45 mm, Whatman) (Standard Method 5310C). UV absorbance
(UVA) measurements were conducted at 254 nm using
a Beckman Coulter DU 800 Spectrophotometer after 0.45 mm
filtration (Standard method 5910B). The specific UV absorbance
(SUVA) was calculated as the ratio of UVA and DOC.
Ammonium and nitrate concentrations were measured
using the Nessler and the chromotropic acid method (HACH)
with a detection range of 0.02–2.0 mg/l and 0.2–30 mg/L,
respectively.
2.2.4. Biomass
Soil biomass was determined as total viable biomass (i.e.
viable, not necessarily active bacteria) using phospholipid
extraction (PLE) as described in Rauch and Drewes (2005).
Analyses were conducted in triplicates from the top-soil
(0–2 cm, infiltration zone) of the columns.
2.2.5. Wastewater effluent
Secondary treated wastewater effluent served as the feed to
column systems and as source for subsequent organic carbon
fractionation. The secondary treated effluent was collected
from a local wastewater treatment plant employing nitrification
and partial denitrification. The average DOC concentration
of the effluent was 8.74 1.44 mg/L.
water research 44 (2010) 449–460
n/a
0.48
(0.48)
n/a not applicable.
pKa and log P values calculated by software ACD.
log D pH¼7 values calculated using equations proposed by Scherrer and Howard, (1977).
Expected
degradability
Relatively
persistent
2.2.6. Organic carbon fractionation
Organic matter (less than 1 mm in size as defined for this study)
of the secondary effluent sample was isolated into three
organic fractions: colloidal organic carbon, hydrophobic acids
(HPO-A), and hydrophilic carbon (HPI) following the procedure
described in Rauch and Drewes (2005) with some modifications.
The sample was concentrated by vacuum rotary evaporation
at 45 C (concentration factor: 40). The concentrate
was separated by dialysis (Spectra/Por, Spectrum, 6000–
8000 Dalton) at pH 4–5 into organic colloids and DOC. The
permeate of the dialysis (containing DOC) was collected and
further separated into HPI and HPO-A by XAD-8 fractionation
according to Leenheer et al. (2000) using a capacity factor of
k 0 ¼ 4. A carbon mass balance was performed for each carbon
fractionation based on UVA and DOC measurements for
quality control. In average, the secondary treated effluent
contained 16 percent colloidal organic carbon, 38 percent
HPO-A, 37 percent HPI, and 8 percent hydrophobic neutrals
(HPO-N) (Rauch and Drewes, 2004). Organic carbon isolates
were diluted to 3 mg/L DOC using milli-Q water, adjusted for
ion strength, macro nutrient (nitrogen, phosphate) and micro
nutrient concentrations (Rauch and Drewes, 2004) and
promptly utilized as column feed waters.
2.3. Experimental set-up
References
Heberer et al., 2001;
Heberer et al., 2003,
Amy and Drewes, 2006;
Fries and Püttmann, 2003
2.3.1. Anoxic column system
The columns (PC system) were operated to study TOrC removal
during simulated MAR under anoxic conditions and up to 3–4
weeks of retention time in the subsurface. The anoxic columns
consisted of four 1-m plexiglass columns (15 cm i.d.) in series
filled with aquifer material (d50 ¼ 0.8 mm, foc ¼ 0.003%) (Fig. SI-1,
Supplemental Information). The column system was operated
in flow-through mode at a loading rate of 0.065 m/d under
saturated, anoxic (denitrifying) flow conditions (Table 2). The
hydraulic retention time of the system using four columns in
series was previously determined as 25 days. The columns had
been continuously fed with secondary or tertiary treated effluents
for over 6 years and for 5 months with the secondary
treated effluent quality employed in this study prior to spiking of
TOrC. The column influent was regularly purged with nitrogen
gas to keep dissolved oxygen concentrations below 0.5 mg/L.
Samples were collected once to twice a week from column
influent and the four column effluents, respectively, and
analyzed for DOC, UV absorbance, pH, conductivity and TOrC
concentrations.

Table 2 – Operational parameters and water quality changes of columns used in this study. a
Label Feed water
matrix
PC Anoxic columns
(EfOM)
C1 Hydrophobic acids
(HPO-A)
C2 Hydrophilic carbon
(HPI)
C3 Effluent organic
matter (EfOM)
C4 Organic colloids
(Org. Colloids)
Abiotic Abiotic column,
Type I water
Length
(m)
4 1m¼
4m
2.3.2. Organic carbon fraction columns
Four columns (C1–C4) were operated to examine the effect of
different organic carbon matrices on soil biomass growth and
TOrC removal during aerobic conditions and short retention
times (less than 1 day) in the subsurface. This system consisted
of four parallel plexiglass columns (L ¼ 30 cm, i.d. ¼ 5 cm) filled
with silica sand (d50 ¼ 0.65 mm, foc ¼ 0.004%). The columns
were operated under saturated, aerobic flow conditions at
a loading rate of 0.9 m/d representing a retention time of about
19 h in the porous media. The system had been acclimated for
over two years to the infiltration of the four organic carbon
fractions derived from secondary effluent (HPO-A, HPI, organic
colloids, EfOM), respectively. Approximately six weeks prior to
spiking TOrC to the column systems, the feed water concentrations
of all four fractions was adjusted to 3 mg/L as DOC.
Samples for DOC, UVA and pH were taken once to twice a week
from the four column influents and effluents. Three consecutive
TOrC spiking events were conducted. For each event,
hydraulically corresponding samples of column influents and
effluents were collected, respectively, after a complete breakthrough
of the spike solution was observed using conductivity
as a conservative tracer (typically after 24 h). After the 1st
spiking event (day 0), the 2nd and 3rd spikes occurred on day
11 and 22, respectively. Set-up and operation conditions of the
different columns employed in this study are summarized in
Table 2. All small column systems received feed waters that
were adjusted to a pH of 8.0 0.2 to minimize adsorption
effects of acidic TOrC onto porous media.
2.3.3. Adsorption test
An additional column was operated under abiotic conditions
to assess the role of physical adsorption of selected TOrC onto
porous media. The column set-up and operation was the same
as described for the biologically active columns C1–C4. The
adsorption column was filled with aquifer material from an
RBF site (d50 ¼ 0.55 mm, bulk density ¼ 1.81 g/cm 3 , effective
porosity ¼ 0.36, fOC ¼ 0.066%). Six TOrC were spiked into
secondary effluent in a concentration range of 175–226 mg/L
adjusted to a pH of 7.0 (feed water). To minimize biotransformation
during this experiment, the column was kept
Organic Carbon Predominant
Feed water Effluent BOC
Redox Condition
(mg/L) (mg/L) (mg/L)
6.9 See Fig. 1 for
DOC column profile
abiotic by adding sodium azide at a concentration of 2 mM.
Sodium bromide was added as a conservative tracer and
bromide was measured by ion chromatography (Dionex,
Sunnyvale, CA) in the column effluent. The column was
operated in single flow-through mode under saturated flow
conditions at a hydraulic loading rate of 0.33 m/d. Effluent
samples were collected using a fraction collector and analyzed
by HPLC analysis. Retardation factors (Rd) were fitted to the
step input breakthrough curves for each compound using the
software program CXTFIT (Toride et al., 1995). Adsorption
coefficients (K d) were calculated according to the following
relationship
Rd ¼ rKd
þ 1 (1)
3
where Rd is the retardation factor, r is the sand bulk density,
and 3 is the sand porosity.
3. Results and discussion
3.1. Adsorption behavior of target TOrC
In order to assess whether physical retardation would be
a contributing factor in the removal of select TOrC in the
column experiments, an abiotic column experiment was
conducted to study sorption of the compounds onto the virgin
porous media used in the columns. None of the six TOrC
included in this experiment exhibited significant retardation
during flow through porous media as compared to the
conservative tracer bromide (indicated by R d values very close
to 1, Table 3). These findings reveal that sorption did not play
an important role in the attenuation of these compounds onto
the porous media used in the columns.
3.2. Organic carbon removal
Total Viable
Soil Biomass b
(nmol PO4 3 /g d.w soil)
Anoxic 17.3 1.3
0.3 3.1 0.9 2.7 0.02 0.4 0.9 Aerobic 3.3 0.4
0.3 2.8 0.4 2.6 0.1 0.3 0.4 Aerobic 10.5 5.3
0.3 3.0 0.4 2.7 0.4 0.4 0.4 Aerobic 20.6 2.5
0.3 3.1 0.5 2.2 0.3 0.8 0.5 Aerobic 27.2 1.8
0.3

454
Table 3 – Retardation and sorption coefficients of select
TOrC derived during abiotic column study.
TOrC Rd Kd (mL/g) Mass recovery (%)
Naproxen 1.30 0.08 94.2
Ibuprofen 1.10 0.03 92.7
Phenacetine 1.02 n.a. n.a.
Diclofenac 1.41 0.01 93.6
Primidone 1.00

Table 4 – Summary of TOrC concentrations observed in column studies.
TOrC (ng/L) Feed
water
1-m
Column
effluent
Anoxic Column HPO-A (C1) HPI (C2) EfOM (C3) Org. colloids (C4)
2-m
Column
effluent
3-m
Column
effluent
4-m
column
effluent
Column
influent
Column
effluent
Column
influent
Column
effluent
Column
influent
Column
effluent
Column
influent
Carbamazepine 279–296 (n ¼ 3) n.a. n.a. n.a. 318 n.a. n.a. n.a. n.a.
Diclofenac sodium 307–416 (n ¼ 2) 75 134 110 80–120 (n ¼ 2) 604 538 n.a. n.a. 710 682
538 190 70

456
Fig. 2 – Removal of TOrC in anoxic column system.
first-order degradation kinetics are indicative of organic
carbon degradation at concentrations below the half-velocity
coefficient Ks, or under mass transport limitations that keep
biodegradation rates below the biological capacity. Zero-order
reaction kinetics are indicative of degradation below the biological
degradation potential, where either substrate concentrations
of trace contaminants are significantly higher than
the affinity for the specific substrate, or the limiting substrate
is delivered at a constant rate (Simoni et al., 2001; Alexander,
1999). It should be noted that, while the column results
ng/L removed in column
900
800
700
600
500
400
300
200
100
0
Gemfibrozil Ketoprofen Naproxen
3.3 10.4 27.2
HPO-A HPI Org. Colloids
3- -1
nmol org-PO4 g
Fig. 3 – Relationship between TOrC removal, soil biomass,
and organic carbon substrate in columns C1, C2, and C4
(1st spike event).
water research 44 (2010) 449–460
indicate a linear removal profile for ketoprofen, naproxen,
gemfibrozil, and ibuprofen, the removal profile might also
have been affected by a change in soil microbial community
composition with column depth caused by changing organic
carbon substrate composition and concentrations along the
flow path. Thus, definite conclusions about the kinetic order
are not possible without additional experiments.
Due to the low ambient concentrations of trace organic
contaminants (occurring in the ng/L range) it is unlikely that
energy from TOrC metabolism is sufficient for biomass
maintenance and growth. It was hypothesized that at trace
levels, TOrC were transformed by co-metabolism (co-utilization).
In this context the term co-metabolism describes
a transformation of a trace compound, which is by itself
unable to support cell replication, and requires the presence of
another transformable organic compound (primary carbon
source) that allows microorganisms to obtain energy for their
metabolism and growth (Brandt et al., 2003; Stratton et al.,
1983; Clara et al., 2005). Because the primary substrate
(biodegradable bulk organic carbon) was present in the
column experiment in concentrations two orders of magnitude
above common concentrations of TOrC in the environment,
co-metabolism offers a possible explanation for the
observed removal of the target TOrC, that were fed at trace
concentrations. This hypothesis was further tested by examining
the effect of organic carbon subgroups present in EfOM
and differing in biodegradability on soil biomass growth and
TOrC removal.
3.6. Effect of aqueous organic carbon matrices
on TOrC removal
In order to further study the effect of soil biomass and aqueous
organic carbon matrices on the removal of TOrC, aerobic
column experiments were conducted using different organic
carbon bulk fractions (EfOM, HPO-A, HPI, and organic colloids)
isolated from secondary treated effluent. The different organic
bulk substrates were fed at similar DOC concentrations to
different column systems and were able to support different
amounts of viable soil biomass (Table 2). During three consecutive
spiking tests, hydraulically corresponding influent and
effluent samples were collected from each column (Fig. 3)with
the exception of the column fed with organic colloids for which
only two spiking tests were conducted.
The results of the first spiking test for the HPI and organic
colloid columns are presented in Fig. 3. Findings suggest that
the more viable soil biomass was present on the column
media, the more complete was the removal for gemfibrozil,
ketoprofen, and naproxen. In other terms the concentration of
biodegradable effluent derived organic carbon substrate
limited the transformation of TOrC. (Phenacetine was
removed below the quantification limit in all columns and
removal could, thus, not be related to biomass activity in the
columns.) Organic colloids (BDOC content 0.86 0.48 mg/L)
appeared to have promoted the degradation of TOrC by
serving as the primary substrate and establishing a relatively
higher microbial population that was able to co-metabolize
TOrC better as compared to HPI organic carbon. TOrC removal
in the HPI column was possibly lower because HPI substrate
(BDOC content 0.25 0.41 mg/L) supported less biomass

emoval %
100
90
80
70
60
50
40
30
20
10
0
growth to metabolize TOrCs. During longer exposure times to
the contaminants the performance of the HPI column for
TOrC removal improved. In the 2nd and 3rd spiking event,
ketoprofen, phenacetine, gemfibrozil, and naproxen exhibited
increasingly better removal exceeding 95 percent 22 days after
the first spike (Fig. 4) suggesting that the microbial community
had adapted to metabolizing TOrC.
In accordance with the performance of the anoxic column
system (Fig. 2), ibuprofen was equally efficient removed (>85%)
in the presence of all organic carbon fractions and in all three
spiking events (Table 4 and Fig. 4) under both aerobic and anoxic
conditions. This confirms previously reported findings that
ibuprofen is very well degradable under various field conditions
(Sedlak and Pinkston, 2001; Drewes et al., 2003). In contrast to the
anoxic column system, propyphenazone was not eliminated in
any of the four short columns(Table 4). Consistent with previous
studies, propyphenazone was found to require longer retention
times (>1 week) to exhibit partial removal under MAR conditions
(Drewes et al., 2003; Heberer et al., 2003). The anticonvulsant
drugs were not eliminated in any of the columns
investigated (Table 4). Based on these study results, effluentderived
BDOC does not seem to be a suitable substrate to stimulate
co-metabolic decay of the selected anticonvulsant drugs
(i.e. primidone, carbamazepine) under environmentally relevant
bulk and TOrC concentrations. Whereas TCPP was not
removed in the anoxic columns, this flame retardant seems to
be removed in the aerobic HPO-A, HPI, and EfOM column after
a lag time of two to three weeks after starting to feed this
compound to the column, possibly indicating biological acclimatization
leading to some degree of attenuation of this
compound (Table 4). However, results are inconsistent when
comparing influent and effluent concentrations in the different
columns, therefore further studies are recommended to solidify
this observation.
Given that the first column of the anoxic system was
operated at a longer residence time (about 6 days) compared to
the aerobic columns (less than 1 day), it is noticeable that most
degradable TOrC (i.e. gemfibrozil, ibuprofen, ketoprofen,
naproxen) were significantly better transformed under
aerobic conditions as compared to anoxic recharge
87
1st spike 2nd spike (day 11 after 1st spike)
3rd spike (day 22 after 1st spike)
561
925
water research 44 (2010) 449–460 457
206
544
700
550
506
Ketoprofen Naproxen Phenacetine Gemfibrozil Ibuprofen
508
113
200
243
490 495 499
Fig. 4 – Microbial adaptation to TOrC in HPI column (values on top of bars represent removal in ng/L).
conditions. Diclofenac exhibited consistent removal during
anoxic conditions but not during aerobic recharge, which is
consistent with findings reported by Zwiener and Frimmel
(2003) and Hua et al. (2003).
3.7. Removal of TOrC in oligotrophic environments
Although the HPO-fraction provided little soil biomass growth
(

458
Because oligotrophic organisms are more likely to be prevalent
in native environments characterized by low-carbon
fluxes than in wastewater treatment plants, the degradation
of TOrC may be more efficient during soil percolation than in
wastewater treatment plants (Daughton and Ternes, 1999).
It is anticipated that oligotrophic organisms reside in MAR
systems in soil depths below the immediate infiltration zone
(which in contrast is likely dominated by microbial organisms
growing primarily on easily degradable EfOM). Further investigations
directed at the interrelation between effluent derived
organic carbon and the succession of biocommunities in soil
based on trophic carbon sequences seem key to the understanding
of mechanisms and limitations to TOrC removal in
MAR systems.
4. Conclusions
Hydrophilic, small-molecular size TOrC can survive conventional
and advanced wastewater treatment processes and are
one of the key concerns in drinking water augmentation
projects using impaired source waters. More research is
needed to better understand the trophodynamic role of organic
carbon on the removal of TOrC in MAR applications. In this
study we conducted column experiments to illuminate key
factors for the metabolic removal of twelve TOrC during soil
infiltration. Experiments were designed to resemble natural
environments (flow through systems, long biological acclimatization
times, complex aqueous carbon compositions, environmentally
relevant bulk and TOrC concentrations) in order
to reveal interactions between bulk organic carbon and TOrC
removal that are relevant for full-scale MAR conditions.
Removal efficiencies varied widely among the target
compounds but were generally higher during aerobic soil
infiltration with the exception of diclofenac that was faster
degraded during anoxic recharge. Propyphenazone required
a longer residence time for removal than was provided in the
aerobic columns employed in this study. The biological reactions
for all other degradable TOrC were fast at ambient
concentrations (ng/L level) and completed within the first 2 m
of porous media infiltration. Threshold concentrations below
which further degradation was negligible were observed for
acidic TOrC under anoxic conditions in the lower ng/L-range.
Intensive research has been conducted on stimulating the
degradation of certain pollutants by artificially adding carbon
substrates (Horvath, 1973, Semprini, 1997). Findings of this
study demonstrated that naturally available organic matter in
aqueous environments can promote the degradation of
certain TOrC by serving as a secondary carbon substrate for
their removal. The composition of effluent-derived bulk
organic carbon had a strong effect on the biological removal of
the target TOrC and is based on several mechanisms. BDOC,
prevalent in form of colloidal and hydrophilic carbon, stimulates
soil biomass growth and induces secondary substrate
utilization of TOrC. Soil biomass in aerobic systems was able
to adapt to most acidic drugs showing an improved removal
over time. Soil biomass growth occurred even in response to
the infiltration of recalcitrant hydrophobic acids. This organic
carbon substrate induced an oligotrophic microbial community
that was more effective in removing TOrC than during
water research 44 (2010) 449–460
copiotrophic metabolism in the presence of higher concentrations
of BDOC. Consistent with previous observations, the
anticonvulsant drugs carbamazepine and primidone exhibited
no removal under any conditions examined in this study.
Acknowledgments
Partial funding for this study was provided by the Gwangju
Institute of Technology in Korea. We are thankful for analytical
support during this study provided by Dr. Thomas Heberer
at the Technical University in Berlin, Germany, Matt
Oedekoven, and Stephan Wagner.
Appendix. Supplementary data
Supplementary data associated with this article can be found
in the online version at doi:10.1016/j.watres.2009.08.027.
references
Amy, G., Drewes, J.E., 2006. Soil-aquifer treatment (SAT) as
a natural and sustainable wastewater reclamation/reuse
technology: fate of wastewater effluent organic matter (EfOM)
and trace organic compounds. Environmental Monitoring and
Assessment 129 (1–3), 19–26.
Alexander, M., 1999. Biodegradation and Bioremediation.
Academic Press, San Diego, California.
Brandt, B.W., Van Leeuwen, I.M.M., Kooijman, S.A.L.M., 2003.
A general model for multiple substrate biodegradation.
Application to co-metabolism of structurally non-analogous
compounds. Water Research 37, 4843–4854.
Brauch, H.-J., Sacher, F., Denecke, E., Tacke, T., 2000. Effectiveness
of bank filtration for the removal of polar organic trace
contaminants. GWF Wasser Abwasser 141 (4) 2260–234 (in
German).
Buser, H.R., Poiger, T., Müller, M.D., 1998. Occurrence and fate of
the pharmaceutical drug diclofenac in surface waters: rapid
photodegradation in a lake. Environmental Science and
Technology 32 (22), 3449–3456.
Buser, H.R., Poiger, T., Müller, M.D., 1999. Occurrence and
environmental behavior of the chiral pharmaceutical drug
ibuprofen in surface waters and in wastewater.
Environmental Science and Technology 33 (15), 2529–2535.
Carballa, M., Omil, F., Lema, J.M., Lllompart, M., Jares, C.G.,
Rodriguez, I., Gomez, M., Ternes, T., 2004. Behaviour of
pharmaceuticals, cosmetics and hormones in a sewage
treatment plant. Water Research 38, 2918–2926.
Clara, M., Strenn, B., Kreuzinger, N., 2004. Carbamazepine as
a possible anthropogenic marker in the aquatic environment:
investigations on the behaviour of carbamazepine in
wastewater treatment and during groundwater infiltration.
Water Research 38, 947–954.
Clara, M., Kreuzinger, N., Strenn, B., Gans, O., Kroiss, H., 2005. The
solids retention time – a suitable design parameter to evaluate
the capacity of wastewater treatment plants to remove
micropollutants. Water Research 39, 97–106.
Daughton, C.G., Ternes, T.A., 1999. Pharmaceuticals and personal
care products in the environment: agents of subtle change?
Environmental Health Perspectives 107 (Suppl. 6) December
1999.

Dillon, P., Page, D., Vanderzalm, J., Pavelic, P., Toze, S., Bekele, E.,
Sidhu, J., Prommer, H., Higginson, S., Regel, R., Rinck-
Pfeiffer, S., Purdie, M., Pitman, C., Wintgens, T., 2008. A critical
evaluation of combined engineered and aquifer treatment
systems in water recycling. Water Science and Technology 57
(5), 753–762.
Drewes, J.E., Fox, P., 1999. Fate of natural organic matter (NOM)
during groundwater recharge using reclaimed water. Water
Science and Technology 40 (9), 241–248.
Drewes, J.E., Shore, L.S., 2001. Concerns about pharmaceuticals in
water reuse, groundwater recharge, and animal waste. In:
American Chemical Society (Ed.), Pharmaceuticals and
personal care products in the environment. Symp. Ser, vol.
791, pp. 206–228.
Drewes, J.E., Heberer, T., Reddersen, K., 2002. Fate of
pharmaceuticals during indirect potable reuse. Water Science
& Technology 46 (3), 73–80.
Drewes, E.J., Heberer, T., Rauch, T., Reddersen, K., 2003. Fate of
pharmaceuticals during groundwater recharge. Ground Water
Monitoring and Remediation 23 (3), 64–72.
Drillia, P., Stamatelatou, K., Lyberatos, G., 2003. Mobility of
Pharmaceuticals in Soil, 8th Conference on Environmental
Science and Technology, Lemnos, Greece.
Focazio, M.J., Kolpin, D.W., Barnes, K.K., Furlong, E.T., Meyer, M.T.,
Zaugg, S.D., Barber, L.B., Thurman, M.E., 2008. A national
reconnaissance for pharmaceuticals and other organic
wastewater contaminants in the United States–II untreated
drinking water sources. Science of the Total Environment 402
(2–3), 201–216.
Fries, E., Püttmann, W., 2003. Monitoring of the three
organophosphate esters TBP, TCEP and TBEP in river water
and ground water (Oder, Germany). Journal of Environmental
Monitoring 5 (2), 346–352.
Grünheid, S., Amy, G., Jekel, M., 2005. Removal of bulk dissolved
organic carbon (DOC) and trace organic compounds by bank
filtration and artificial recharge. Water Research 39, 3219–3228.
Heberer, T., Stan, H.-J., 1997. Determination of clofibric acid and
N-(phenylsulphonyl)-sarcosine in sewage, river and drinking
water. International Journal of Environmental Analytical
Chemistry 67, 113–124.
Heberer, T., Verstraeten, I.M., Meyer, M.T., Mechlinski, A.,
Reddersen, K., 2001. Occurrence and fate of pharmaceuticals
during bank filtration–preliminary results from investigations in
Germany and the United States. Water Resources Update 20, 4–17.
Heberer, T., 2002. Occurrence, fate and removal of
pharmaceutical residues in the aquatic environment: a review
of recent research data. Toxicology Letters 131, 5–17.
Heberer, T., Mechlinski, A., Fanck, B., 2003. NASRI – Occurrence
and Fate of Pharmaceuticals During Bank Filtration,
Conference Wasser Berlin, KompetenzZentrum Wasser Berlin,
pp. 41–49.
Horvath, R.S., 1973. Enhancement of co-metabolism of
chlorobenzoates by the co-substrate enrichment technique.
Applied Microbiology 25 (6), 961–963.
Hua, J., An, P., Winter, J., Gallert, C., 2003. Elimination of COD,
microorganisms and pharmaceuticals from sewage by
trickling through sandy soil below leaking sewers. Water
Research 37, 4395–4404.
Kolpin, D.W., Furlong, E.T., Meyer, M.T., Thurman, E.M., Zaugg, S.
D., Barber, L.B., Buxton, H.T., 2002. Pharmaceuticals,
hormones, and other organic wastewater contaminants in U.
S. streams, 1999–2000: a national reconnaissance.
Environmental Science & Technology 36, 1202–1211.
Leenheer, J.A., Croue, J.P., Benjamin, M., Korshin, G.V., Hwang, C..,
Bruchet, A., Aiken, G.R., 2000. Comprehensive isolation of
natural organic matter from water for spectral characterizations
and reactivity testing. In: Barett, S.E., Krasner, S.W., Amy, G.L.
water research 44 (2010) 449–460 459
(Eds.), Natural Organic Matter and Disinfection By-Products, vol.
761. ACS Symposium Series, pp. 68–83.
Margesin, R., Schinner, F., 2001. Bioremediation (natural
attenuation and biostimulation) of diesel-oil-contaminated
soil in an alpine glacier skiing area. Applied and
Environmental Microbiology 67 (7), 2127–3133.
McCarty, P.L., Reinhard, M., Rittmann, B.E., 1981. Trace organics in
groundwater. Environmental Science and Technology 15 (1), 40–51.
Mersmann, P., Scheytt, T., Heberer, T., 2002. Column experiments
on the transport of pharmaceutically active compounds in the
saturated zone. Acta Hydrochimica et Hydrobiologica 30 (5–6),
275–284 (in German).
Möhle, E., Kempter, C., Kern, A., Metzger, J.W., 1999.
Examination of the degradation of drugs in municipal sewage
plants using liquid chromatography-electrospray mass
spectrometry. Acta Hydrochimica et Hydrobiologica 27 (6),
430–436 (in German).
Montgomery-Brown, J., Drewes, J.E., Fox, P., Reinhard, M., 2003.
Behavior of alkylphenol polyethoxlate metabolites during soil
aquifer treatment. Water Research 37, 3672–3681.
Preuß, G., Nehrkorn, A., 1996. Succession of microbial
communities during bank filtration and artificial groundwater
recharge Artificial Recharge of Groundwater. Proceedings of
an International Symposium, A.-L. Kivimäki and T. Suokko,
eds., Helsinki, Finland, June 3–5, 1996, Nordic Hydrological
Programme, pp. 215–221.
Rauch, T., Drewes, J.E., 2004. Assessing the removal potential of
soil-aquifer treatment systems for bulk organic matter. Water
Science & Technology 50 (2), 245–253.
Rauch, T., Drewes, J.E., 2005. A novel approach for quantifying
bulk organic carbon removal in groundwater recharge
systems. Journal of Environmental Engineering 131 (6),
909–923.
Reddersen, K., Heberer, T., 2003. Multi-compound methods for
the detection of pharmaceutical residues in various waters
applying solid phase extraction (SPE) and gas chromatography
with mass spectrometric (GC-MS) detection. Journal of
Separation Sciences 26, 1443–1450.
Rittmann, B.E., McCarty, P.L., 2001. Environmental Biotechnology.
Principles and Applications. McGraw-Hill, Boston.
Scherrer, R.A., Howard, S.M., 1977. Use of distribution coefficients
in quantitative structure-activity relationships. Journal of
Medical Chemistry 20 (1), 53–58.
Schmidt, C.K., Lange, F.T., Brauch, H.-J., 2004. Assessing the
Impact of Different Redox Conditions and Residence Times on
the Fate of Organic Micropollutants during Riverbank
Filtration, Fourth International Conference on
Pharmaceuticals and Endocrine Disrupting Chemicals in
Water, October 13–15, Minneapolis, Minnesota.
Schwab, B.W., Hayes, E.P., Fiori, J.M., Mastrocco, F.J., Roden, N.M.,
Cragin, D., Meyerhoff, R.D., D’aco, V.J., Anderson, P.D., 2005.
Human pharmaceuticals in US surface waters: a human
health risk assessment. Regulatory Toxicology and
Pharmacology 42 (3), 296–312.
Sedlak, D.L., Pinkston, K.E., 2001. Factors affecting the
concentrations of pharmaceuticals released to the aquatic
environment. Journal of Contemporary Water Research and
Education 120, 56–64. September.
Semprini, L., 1997. Strategies for the aerobic co-metabolism of
chlorinated solvents. Current Opinion in Biotechnology 8 (3),
296–308.
Simoni, S.F., Schäfer, A., Harms, H., Zehnder, A.J.B., 2001. Factors
affecting mass transfer limited biodegradation in saturated
porous media. Journal of Contaminant Hydrology 50, 99–120.
Stratton, R.G., Namkung, E., Rittmann, B.E., 1983. Secondary
utilization of trace organic by biofilms on porous media.
Journal American Water Works Association 75, 463–469.

460
Szewzyk, U., Kalmbach, S., Manz, W., La˚ ngmark, J., and
Stenström, T.-A., 1998. Artificial groundwater as the future
water supply of greater Stockholm IV: Phylogenetic
diversity of the unsaturated zones of experimental sand
filter columns. Artificial recharge of groundwater.
Proceedings of the third international symposium on
artificial recharge of groundwater–Tisar 98, J.H. Peters, ed.,
Amsterdam, Netherlands, 21–25 September 1998,
Rotterdam, pp. 391–393.
Toride, N., Leij, F.J., van Genuchten, M.T., 1995. The CXTFIT Code
for Estimating Transport Parameters from Laboratory or Field
water research 44 (2010) 449–460
Tracer Experiments. Version 2.0. U.S. Salinity Laboratory Res.
Rep. 137, U.S. Salinity, Riverside, CA.
Van der Meer, J.R., Roelofsen, W., Schraa, G., Zehnder, A.J.B., 1987.
Degradation of low concentrations of dichlorobezenes and 1,2,4thrichlorobenzene
by Pseudomonas sp. strain P51 in nonsterile
soil columns. FEMS Microbiology Letters 45 (6), 333–341.
Zwiener, C., Frimmel, F.H., 2003. Short-term tests with a pilot
sewage plant and biofilm reactors for the biological
degradation of the pharmaceutical compounds clofibric acid,
ibuprofen, and diclofenac. Science of the Total Environment
309, 201–211.

Toxicological relevance of emerging contaminants for
drinking water quality
Merijn Schriks a, *, Minne B. Heringa a , Margaretha M.E. van der Kooi a , Pim de Voogt a,b ,
Annemarie P. van Wezel a
a KWR Watercycle Research Institute, P.O. Box 1072, 3430 BB Nieuwegein, The Netherlands
b Institute for Biodiversity and Ecosystem Dynamics, University of Amsterdam, Nieuwe Achtergracht 166, 1018 WV,
Amsterdam, The Netherlands
article info
Article history:
Received 17 May 2009
Received in revised form
14 August 2009
Accepted 21 August 2009
Available online 26 August 2009
Keywords:
Risk assessment
Drinking water
(Provisional) drinking water
guideline value
Threshold of Toxicological
Concern (TTC)
1. Introduction
abstract
Due to anthropogenic activities, freshwater systems worldwide
are confronted with thousands of compounds. In the
European Union, for example, there are more than 100 000
registered chemicals (EINECS), of which 30 000–70 000 are in
daily use. About 300 million tons of synthetic compounds
annually used in industrial and consumer products, partially
find their way to natural waters (Schwarzenbach et al., 2006).
A major contribution to chemical contamination originates
from wastewater discharges that impact surface water quality
water research 44 (2010) 461–476
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
* Corresponding author. Tel.: þ31 30 6069564.
E-mail address: merijn.schriks@kwrwater.nl (M. Schriks).
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.watres.2009.08.023
The detection of many new compounds in surface water, groundwater and drinking water
raises considerable public concern, especially when human health based guideline values
are not available it is questioned if detected concentrations affect human health. In an
attempt to address this question, we derived provisional drinking water guideline values
for a selection of 50 emerging contaminants relevant for drinking water and the water
cycle. For only 10 contaminants, statutory guideline values were available. Provisional
drinking water guideline values were based upon toxicological literature data. The
maximum concentration levels reported in surface waters, groundwater and/or drinking
water were compared to the (provisional) guideline values of the contaminants thus
obtained, and expressed as Benchmark Quotient (BQ) values. We focused on occurrence
data in the downstream parts of the Rhine and Meuse river basins. The results show that
for the majority of compounds a substantial margin of safety exists between the maximum
concentration in surface water, groundwater and/or drinking water and the (provisional)
guideline value. The present assessment therefore supports the conclusion that the
majority of the compounds evaluated pose individually no appreciable concern to human
health.
ª 2009 Elsevier Ltd. All rights reserved.
with incompletely removed organic contaminants (Kolpin
et al., 2004; Snyder et al., 2001). Additional contamination
comes from diffuse agricultural activities, in which over 140
million tons of fertilizers and several million tons of pesticides
are applied each year, and from atmospheric deposition. Such
contamination can become an increasing problem for
drinking water supplies, especially since the European REACH
legislation may drive producers to develop newly designed
less lipophilic/bioaccumulative chemicals that will be inherently
more difficult to remove by traditional drinking water
treatment techniques.

462
Recently, Loos et al. (2009) presented an EU-wide monitoring
study on 35 organic compounds in European river
waters in concentrations up to 40 mg/L. In addition, we have
shown the occurrence of emerging polar contaminants,
such as benzotriazoles and metabolites of illicit drugs (e.g.
benzoylecgonine, desalkylflurazepam and 9-carbonic acid-d-
9-THC) in groundwaters and surface waters in the
Netherlands (Hogenboom et al., 2009; Van Leerdam
et al., 2009).
Many of these emerging contaminants raise considerable
toxicological and public concern, especially when human
health based guideline values are unavailable. At present,
the World Health Organization (WHO) and the US Environmental
Protection Agency (US EPA) have derived approximately
125 statutory guideline values for contaminants in
drinking water (Cotruvo, 1988; US EPA, 2006; WHO, 2006).
However, the potential health effects of many emerging
contaminants present in the water cycle and the potential
human health concern associated with direct water ingestion
have not been evaluated and statutory standards are
not available. Therefore, we proposed earlier to assess the
potential human health concern of unknown non-genotoxic
compounds (lacking structural alerts that raise concern for
potential genotoxicity) by comparing the environmental or
drinking water concentration to a TTC (Threshold of Toxicological
Concern) derived target value (Mons et al., 2008).
The TTC concept was developed in the context of food
safety to obtain a first clue on risks of unregulated chemicals
present at low levels. The TTC thus derived is based on the
molecular structure of the chemical involved and its related
mode of toxic action (Munro et al., 1996). Assuming a daily
intake of 2 l/day of drinking water, and a maximum contribution
of 10% from drinking water to the total exposure –
both of which are standard assumptions for deriving
drinking water-quality guidelines (WHO, 2006) –TTCbased
target values proposed for drinking water are 0.1 mg/L for
non-genotoxic compounds and 0.01 mg/L for genotoxicants
(Mons et al., 2008). This value is based on a TTC level of
1.5 mg/person/d (0.15 mg/person/d for substances containing
structural alerts that raise concern for potential genotoxicity
with an acceptable lifetime cancer risk of 10 6 ) for
compounds in food (Kroes et al., 2004). Since the TTC based
value is rather conservative, an over-estimation of the
actual risk may be the result. For emerging contaminants,
a more profound human health based assessment may
therefore be very valuable. The objectives of the present
study were twofold. The first objective was to collect existing
drinking water guideline values for a selection of 50
emerging contaminants relevant for the water cycle. If
existing guideline values were not available, provisional
guideline values were derived with the aid of relevant
toxicological literature data. The second aim was to
compare the maximum concentration levels reported in
surface water, groundwater and/or drinking water to the
(provisional) guideline values of the contaminants thus
obtained, and express this as a Benchmark Quotient (BQ)
value (further abbreviated as ‘‘BQ value’’). The present study
does not attempt to quantify mixture interactions, since for
compounds with an unknown mode of action there is no
accepted methodology for such an assessment.
water research 44 (2010) 461–476
2. Materials and methods
The toxicological assessment of the compounds presented in
this paper comprises of a tiered approach in five consecutive
steps (Fig. 1). First, the compounds to be assessed were
selected. Second, n-octanol–water partition coefficients
(log K ow) were obtained and compounds with a log K ow > 3
were excluded from further assessment. This log K ow cut off
value is applied as a default threshold; for compounds with
a log K ow above 3 it is less likely that they pass drinking water
treatment plants (Westerhoff et al., 2005). Third, if available,
statutory drinking water guideline values were obtained from
websites of competent authorities; else provisional guideline
values were derived with the aid of toxicological data relevant
for humans as reported in literature. Fourth, measured
maximum surface water, groundwater and/or drinking water
concentrations were obtained from various sources and
compared to (provisional) guideline values. Finally, a BQ value
was calculated from the maximum concentrations reported
and the (provisional) drinking water guideline values
obtained. These steps are described in more detail below.
2.1. Selection of compounds for assessment
A priority list representing a broad range of chemical classes
was formulated with more than 100 compounds of interest.
The arguments for inclusion were (i) questions related to
toxicity posed by Dutch drinking water companies, (ii) potential
low removal efficiency during drinking water production,
(iii) appearance in recent literature and (iv) occurrence in
surface waters, groundwaters and drinking water as determined
by ourselves and others in various screening studies.
2.2. Collection of compound-specific data
2.2.1. n-Octanol–water partition coefficients (log K ow)
All log Kow values were obtained with the aid of the estimation
program KOWWIN (US EPA, v1.67). An exception was
made for perfluoroctane sulfonate (PFOS) and perfluoroctanoic
acid (PFOA), for which accurate log K ow values
cannot be calculated with estimation software. For these
compounds the log Kow values were obtained from a database
(Krop and de Voogt, 2008).
2.2.2. Toxicological data
As illustrated in Fig. 1 (step 3), the first step was to obtain existing
statutory drinking water guideline values from e.g. the US EPA
(URL1) and the WHO (URL2). If not available, the second step was
to obtain an established (by an (inter)national organization)
Tolerable Daily Intake (TDI), Acceptable Daily Intake (ADI) or
Reference Dose (RfD) and subsequently a provisional drinking
water guideline value was derived as further described in
Section 2.3. If not available, in a third step toxicity data collection
focused primarily on established (by an (inter)national organization)
lowest/no observed (adverse) effect levels (LO/NO(A)ELs)
and subsequently a TDI was calculated as further described in
Section 2.3. Finally, in a fourth step, miscellaneous toxicological
information was collected and a TDI was calculated accordingly.
In the case of insufficient human relevant toxicological data the

compound of interest was not further evaluated and removed
from the list. To facilitate the interpretation for which
compounds the toxicity database is strong and less strong, all
compounds were categorized (Table 2); (A) representing
compounds with a statutory drinking water guideline value, (B)
representing compounds with an established TDI, ADI or RfD,
(C) representing compounds for which the TDI was calculated
with an established LO(A)EL or NO(A)EL and (D) representing
compounds for which the TDI was calculated with miscellaneous
toxicological information.
TDIs, ADIs, RfDs and/or other chronic toxicity data were
sourced from peer-reviewed scientific papers and from other
sources such as ‘‘grey literature’’. In addition, a literature
search was performed on the internet and/or toxicological
relevant data were obtained from the US EPA IRIS database,
the European Chemicals Bureau (ECB), the Organization for
Economic Cooperation and Development (OECD), the Dutch
National Institute for Health and the Environment (RIVM), the
US Food and Drug Administration (US FDA), the Joint Meeting
FAO/WHO Meetings on Pesticide Residues (JMPR), the Dutch
Expert Committee for Occupational Standards (DECOS), the
Dutch board for authorization of plant protection products
and biocides (CTGB), the Scientific Committee on Occupational
Exposure Limits (SCOEL), the US National Toxicology
Program (NTP), the Joint FAO/WHO Committee on Food
Additives (JECFA), the Food and Agriculture Organization
(FAO), the Danish veterinary and food administration, the
European Union (EU), the US National Research Council (NRC),
the EFSA scientific panel on contaminants in the food chain
(CONTAM) and the Hazardous Substance Data Bank (HSDB).
water research 44 (2010) 461–476 463
Fig. 1 – Flow diagram of the assessment conducted in the present study. Abbreviations: log K ow, n-octanol–water partition
coefficient; GLV, drinking water guideline value; ADI, acceptable daily intake; RfD, reference dose; TDI, tolerable daily intake;
LO/NO(A)EL, lowest/no observed (adverse) effect level; numbers correspond to consecutive steps as described Section 2.
Compound categories: see Section 2.2.2.
2.2.3. Occurrence data
Collection of occurrence data focused primarily on maximum
concentrations of compounds measured in the downstream
parts of the Rhine and Meuse river basins during the past
decade. If not available, maximum concentrations in other
surface waters and/or groundwaters were sought. The
primary source of occurrence data of compounds in surface
waters were the annual reports of the Dutch Association of
River Water Companies (RIWA) and the German Association
of River Water Companies (ARW). Occurrence data of
compounds in drinking water were obtained from the REWAB
data set (restricted water-quality data from the Dutch water
companies). If not available, alternative resources were
searched such as ‘‘grey literature’’, unpublished data, or
publicly available sources such as RIVM, the Dutch ministry of
transport, public works and water management (Rijkswaterstaat)
and the WHO. Additional data on the occurrence of
compounds was obtained from peer-reviewed scientific
papers and an internet based literature search.
2.3. Derivation of provisional drinking water guideline
values
A drinking water guideline value represents the concentration
of a constituent that does not exceed tolerable risk to the
health of the consumer over a lifetime (WHO, 2006). In some
cases, an odour-threshold value may be much lower than the
health based guideline value. To calculate a provisional health
based guideline value, the general methodology was applied

464
Table 1 – List of compounds assessed in the present study.
No. Compound CAS log K ow No. Compound CAS Log K ow
1 1,4-Dioxane 2
2 2,6-Dichlorobenzamide (BAM) 3
3 4-Methylbenzenesulfonamide
(p-toluenesulfonamide,
4-tolylsulfonamide) 2
4 Acetylsalicylate
(aspirin, acetyl salicylic acid) 6
5 Alpha-amino-3-hydroxy-
5-methyl-4-isoxazole propionic
acid (AMPA) 3
6 Amidotrizoic acid (diatrizoic acid) 1
7 Bentazone 3
8 Benzene 2
9 Benzothiazole 2
10 Benzotriazole (1H-benzotriazole) 2
11 Bis(2-chloroisopropyl)ether (BCIPE) 2
12 Carbamazepine 6
13 Carbendazim 3
14 Chloridazon (pyrazon) 3
15 Clofibric acid 6
16 Dichlorophenoxyacetic acid (2,4-D) 3
17 Diethyl phthalate 2
18 Diethyl toluamide (DEET) 3
19 Diethylamine (DEA) 2
20 Diethylene glycol dimethyl ether
(diglyme, bis(2-methoxy ethyl)ester)) 2
21 Diethylene triamine penta acetic
acid (DTPA) 2
22 Dimethenamid 3
23 Dimethylamine (DMA) 2
24 Diuron 3
25 Ethyl tert-butyl ether (ETBE) 4
123-91-1 0.3 a
2008-58-4 0.8 a
70-55-3 0.9 b
50-78-2 1.2 a
77521-29-0 2.2 b
117-96-4 1.4 b
25057-89-0 2.3 a
71-43-2 2.1 a
95-16-9 2.0 a
95-14-7 1.4 a
108-60-1 2.5 a
298-46-4 2.5 a
10605-21-7 1.5 a
1698-60-8 1.1 a
882-09-7 2.6 a
94-75-7 2.8 a
84-66-2 2.4 a
134-62-3 2.2 a
109-89-7 0.6 a
111-96-6 0.4 a
as described by Van Leeuwen (2000) and the WHO (2006). For
compounds without a statutory drinking water guideline
value, first the Tolerable Daily Intake (TDI) was determined.
The point of departure (POD) for calculating the TDI was
mostly a chronic LO(A)EL, NO(A)EL, benchmark dose level,
maximum tolerated dose (MTD) or lowest effective safe dose.
In case only inhalatory toxicity data could be found, a routeto-route
extrapolation was carried out according to toxicological
methods as described by Stokinger and Woodward
(1958). An appropriate safety factor to extrapolate between
species (inter-species differences), inter-individual differences
(intraspecies differences), exposure route/duration and
quality of the data was utilized as part of the TDI calculation
(Van Leeuwen and Vermeire, 2007). Secondly, a drinking water
equivalent level (DWEL) was calculated by multiplying the TDI
water research 44 (2010) 461–476
67-43-6 4.2 b
87674-68-8 2.2 a
124-40-3 0.4 a
330-54-1 2.7 a
637-92-3 1.9 b
26 Ethylenediamine tetra
acetic acid (EDTA) 2
27 Glyphosate 3
28 Imidacloprid 3
29 Iohexol 1
30 Iomeprol (iomeron) 1
31 Iopamidol 1
32 Iopromide 1
33 Isoproturon 3
34 Methyl tert-butyl ether (MTBE) 4
35 Metoprolol 6
36 n-Butylbenzenesulphonamide 2
37 Nicosulfuron 3
38 n-Nitrosodimethylamine
(NDMA) 2
39 p,p 0 -Sulfonyldiphenol 2
40 Perfluoroctanesulfonate
(PFOS) (potassium-salt) 5
41 Perfluorooctanoic acid
(PFOA) 5
42 Phenazone 6
43 Simazine 3
44 Sulfamethoxazole 6
45 Tolyltriazole 2
46 Trichloroethene 2
47 Triethylphosphate
(ethylphosphate) (TEP) 2
48 Triphenylphosphine
oxide (TPPO) 2
49 Tris(2-chloroethyl)
phosphate (TCEP) 2
50 Urotropine (methenamine,
hexamine) 2
60-00-4 3.9 b
1071-83-6 4.0 a
138261-41-3 0.6 a
66108-95-0 3.1 a
78649-41-9 1.4 b
62883-00-5 2.4 a
73334-07-3 2.1 a
34123-59-6 2.9 a
1634-04-4 0.9 a
37350-58-6 1.9 a
3622-84-2 2.3 b
111991-09-4 1.2 b
62-75-9 0.6 a
80-09-1 1.7 b
2795-39-3 1.08 c
335-67-1 2.8 c
60-80-0 0.4 a
122-34-9 2.2 a
723-46-6 0.9 a
29385-43-1 NA
79-01-6 2.4 a
78-40-0 0.8 a
791-28-6 2.8 a
115-96-8 1.4 a
100-97-0 4.2 b
NA not available. Chemical categories: 1 Iodinated contrast media; 2 Miscellaneous organic compounds; 3 Miscellaneous pesticides; 4 Oxygenated
gasoline additives; 5 Perfluorinated organic compounds; 6 Pharmaceuticals.
a log K ow values were derived from US EPA’s KOWWIN experimental database.
b log Kow values were calculated with the aid of US EPA’s KOWWIN.
c log Kow values were obtained from Krop and de Voogt (2008).
by a typical average body weight of 70 kg and division by
a daily water consumption of 2 l. Finally, to account for the
fraction of the TDI allocated to drinking water, the DWEL was
multiplied by an allocation factor to give the provisional
guideline value. In most cases, when there was insufficient
exposure information to derive chemical-specific allocation
factors, a default allocation factor of 10% was used.
2.4. Evaluation of water-quality data in the context of
human health
Of each compound the maximum concentration level reported
in surface waters, groundwater and/or drinking water was
compared to its (provisional) guideline value and was
expressed as a BQ value (concentration in water divided by

Table 2 – Parameters used for derivation of (provisional) drinking water guideline values.
Compound Point of departure (POD) Category a SUF b TDI, ADI or RfD
(mg/kg bw/d)
1,4-Dioxane POD is an oral slope factor as derived by US EPA (1988a) of 0.011 per mg/kg bw/d from a 110-week study
in rats (NCI, 1978).
2,6-Dichlorobenzamide (BAM) POD is a NOAEL of 4.5 mg/kg bw/d for decreased body weight in both sexes and increased liver weight
in males as derived by the Danish EPA (2004) from a study in which dogs were fed BAM in the diet for 2
years (Wilson and Thorpe, 1971), with an uncertainty factor of 300 (100 for inter- and intraspecies
variation and 3 for uncertainties in the dataset).
4-Methylbenzenesulfonamide
(p-toluenesulfonamide,
4-tolylsulfonamide)
Acetylsalicylate (aspirin, acetyl
salicylic acid)
Alpha-amino-3-hydroxy-5-methyl-
4-isoxazole propionic acid (AMPA)
POD is a NOAEL of 300 mg/kg bw/d for reproductive effects (decrease lactation index and litter weight at
birth) as derived from a GLP compliant study in which rats were administered 4methylbenzenesulfonamide
via oral gavage for 42 days (OECD/SIDS, 1994), with an uncertainty factor
of 400 (100 for inter- and intraspecies variation and 4 for extrapolation to chronic exposure).
POD is a LOEL of 10 mg/person from a human study as derived by the Dutch National Institute for
Health and the Environment (RIVM) (Versteegh et al., 2007).
POD is a NOAEL of 32 mg/kg bw/d as derived by the WHO (2005a) from a 26-month toxicity study in rats
(study reference unknown).
Amidotrizoic acid (diatrizoic acid) POD is the highest therapeutic dose of 50 mg/person/d (0.71 mg/kg bw/d) as derived by the Dutch
National Institute for Health and the Environment (RIVM) (Versteegh et al., 2007).
Bentazone POD is a NOAEL of 9 mg/kg bw/d as derived by the WHO (1998) from a 2-year dietary toxicity study in
rats (study reference unknown).
Benzene POD is a risk estimate as derived by the WHO (2003a) from a 2-year gavage study in rats and mice (NTP,
1986).
Benzothiazole POD is a NOEL of 5.1 mg/kg bw/d as derived by the WHO/JECFA (2003) from a study in which rats were
administered benzothiazole in the diet for 90 days (Morgareidge, 1971). Daily observations revealed no
treatment related effects in histopathological/haematological parameters, body weight, food
consumption and liver/kidney weights. An uncertainty factor of 200 was applied (100 for inter- and
intraspecies variation and 2 for extrapolation to chronic exposure).
Benzotriazole (1H-benzotriazole) POD is a LOAEL of 295 mg/kg bw/d for histological changes in the liver, decreased body weight gain and
inflammation of the prostate/uterus as derived by the Dutch Expert Committee for Occupational
Standards (DECOS, 2000) from a study in which rats were administered benzotriazole in the diet for 78
weeks (BIBRA Toxicology International, 1995), with an uncertainty factor of 1000 (100 for inter- and
intraspecies variation and 10 for extrapolation of a LOAEL to a NOAEL).
Bis(chloroisopropyl)ether (BCIPE) POD is a NOAEL of 35.8 mg/kg bw/d as derived by the US EPA (1989) from a 24-month chronic toxicity
study in mice (Mitsumori et al., 1979).
Carbamazepine POD is a maximum tolerated dose (MTD) of 250 mg/kg bw/d as derived by Snyder et al. (2008) from a 2year
study in rats showing evidence of carcinogenicity (Singh et al., 2005).
Carbendazim POD is a NOAEL of 2.5 mg/kg bw/d as derived by the WHO/JECFA (1995) from a 2-year study in dogs
(Sherman, 1972).
Chloridazon (pyrazon) POD is a NOAEL of 5.4 mg/kg bw/d for adverse histopathological/haematological changes, decreased
food intake, lower body weight gain and higher organ weights (liver, kidney, thyroid gland) as derived
from a study in which rats were orally administered chloridazon (method of administration
unspecified) for 7 weeks (ECB, 2000b), with an uncertainty factor of 100 (inter- and intraspecies
variation).
Clofibric acid POD is a LOEL of 1 mg/kg bw/d as derived by the Dutch National Institute for Health and the
Environment (RIVM) (Versteegh et al., 2007) from an 8-week oral study in humans (Larsen et al., 1994).
(Provisional)
guideline
value (mg/L)
B NA NA 30 d
C 300 0.015 52.5
D 400 0.75 2600
B 20 0.007 25
A 100 0.3 900 c
B 10 NA 250 000
A 100 0.1 300 c
A NA NA 10 c,d
C 200 0.026 90
C 1000 0.295 1000
B 1000 0.04 140
B NA 0.00034 1
B 100 0.03 105
D 100 0.054 189
B 100 0.01 30
(continued on next page)
water research 44 (2010) 461–476 465

Table 2 (continued)
Compound Point of departure (POD) Category a SUF b TDI, ADI or RfD
(mg/kg bw/d)
Dichlorophenoxyacetic acid (2,4-D) POD is a NOAEL of 1 mg/kg bw/d as derived by the WHO (2003b) from a 1-year study of toxicity in dogs
and a 2-year study of toxicity and carcinogenicity in rats (study reference unknown).
Diethyl phthalate POD is a NOAEL of 750 mg/kg bw/d as derived by US EPA (1988b) from a 16-week toxicity study in rats
(Brown et al., 1978).
Diethyl toluamide (DEET) POD is a NOEL of 100 mg/kg/bw/d based on clinical signs, reduced haemoglobin/haematocrit levels and
histological changes in liver, lymph nodes and uterus as derived by the California Environmental
Protection Agency (California EPA, 2000) from a study in which beagle dogs were orally administered
DEET (gelatin capsules) for 1 year (Goldenthal, 1994), with an uncertainty factor of 56 (14 for inter- and
intraspecies variation and 4 for extrapolation to chronic exposure).
Diethylamine (DEA) POD is a LOAEL of 75 mg/m 3 for reduced mean body weights and adverse histopathological effects
(lesions of the nasal mucosa) as derived by the scientific committee on occupational exposure limits
(SCOEL, 2002) from a study in which rats were exposed to DEA via the inhalatory route for 24 weeks
(6.5 h/d, 5d/wk) (Lynch et al., 1986), with an uncertainty factor of 50 (5 for the absence of human data
and a NOAEL and 10 for route-to-route extrapolation uncertainties).
Diethylene glycol dimethyl ether
(diglyme, bis(2-methoxy ethyl)ester))
Diethylene triamine penta acetic acid
(DTPA)
POD is a NOAEL 25 mg/kg bw/d for developmental effects (adversely affected implants per liter and
decreased weight gain) as derived by the WHO (2002) from a study in which rabbits were administered
diglyme via oral gavage for 13 days (NTP, 1987), with an uncertainty factor of 500 (100 for inter- and
intraspecies variation and 5 for uncertainties in the dataset).
POD is a NOAEL of 100 mg/kg bw/d for developmental effects (increased fetal deformations) as derived
from a study according to OECD guideline 414 in which rats were administered DTPA (in its sodium
form) via oral gavage during day 6–15 of pregnancy (ECB, 2000c), with an uncertainty factor of 1000 (100
for inter- and intraspecies variation and 10 for extrapolation to chronic exposure).
Dimethenamid POD is a NOAEL of 7 mg/kg bw/d as derived by the WHO/JMPR (2005) from a 24-month study in rats
given diets containing racemic dimethenamid (study reference unknown).
Dimethylamine (DMA) POD is a LOAEL of 19 mg/m 3 for concentration-related lesions in the respiratory/olfactory mucosa as
derived by the scientific committee on occupational exposure limits (SCOEL, 1991) from a study in
which rats and mice were exposed to DMA via the inhalatory route for 2 years (6 h/d, 5d/wk) (CIIT,
1990), with an uncertainty factor of 50 (5 for the absence of human data and a NOAEL and 10 for routeto-route
extrapolation uncertainties).
Diuron POD is a NOEL of 0.625 mg/kg bw/d as derived by US EPA (1988c) from a 2-year feeding study in dogs
(DuPont, 1964).
Ethyl tert-butyl ether (ETBE) POD is a NOAEL of 500 ppm (29.1 mg/kg bw/d h ) for testes degeneration as derived from a study in which
rats were exposed to ETBE via the inhalatory route for 13 weeks (Medinsky et al., 1999), with an
uncertainty factor of 200 (100 for inter- and intraspecies variation and 2 for extrapolation to chronic
exposure).
Ethylenediamine tetra acetic Acid
(EDTA)
POD is a NOAEL of 250 mg/kg bw/d (190 mg/kg bw/d as the free acid) as derived by the WHO/JECFA
(1973) from a 2-year toxicity study in rats (study reference unknown).
Glyphosate POD is a NOAEL of 32 mg/kg bw/d as derived by the WHO (2005a) from a 26-month study of toxicity in
rats fed technical-grade glyphosate (study reference unknown).
Imidacloprid POD is a NOAEL of 5.7 mg/kg bw/d as derived by the WHO/JMPR (2001) from a 2-year study of toxicity
and carcinogenicity in rats (study reference unknown).
Iohexol POD is a safe dose of 75 g/person/d (1.07 g/kg bw/d) as derived by the Dutch National Institute for
Health and the Environment (RIVM) (Versteegh et al., 2007).
(Provisional)
guideline
value (mg/L)
A 100 0.01 30 c
B 1000 0.8 2800
C 56 1.8 6250
C 50 2.14 f
750
C 500 0.05 175
D 1000 0.1 350
B 100 0.07 245
C 50 0.54 g
190
B 300 0.002 7
D 200 0.15 525 e
A 100 1.9 600 b,c
A 100 0.3 900 c
B 100 0.06 210
B 10 NA 375 000
466
water research 44 (2010) 461–476

Iomeprol (iomeron) POD is a NOEL of 2 g Iodine/kg bw/d (equal to 4 g iomeprol/kg bw/d) for adverse effects on liver and
kidney (non-lipid cytoplasmic vacuolization of hepatocytes and renal tubular epithelium cells) as
derived from a study in which dogs were intravenously exposed to iomeprol for 28 days (Morisetti et al.,
1994), with an uncertainty factor of 2100 (35 for inter- and intraspecies, 10 for route-to-route
extrapolation and 6 for extrapolation to chronic exposure).
D 2100 1.9 6700
Iopamidol POD is a safe dose of 83 g/person/d (1.19 g/kg bw/d) as derived by the Dutch National Institute for
Health and the Environment (RIVM) (Versteegh et al., 2007).
B 10 NA 415 000
Iopromide POD is a safe dose of 50 g/person/d (0.71 g/kg bw/d) as derived by the Dutch National Institute for
Health and the Environment (RIVM) (Versteegh et al., 2007).
B 10 NA 250 000
Isoproturon POD is a NOAEL of 3 mg/kg bw/d as derived by the WHO (2003c) from a 90-day study in dogs and a 2-year
feeding study in rats (study reference unknown).
A 1000 0.003 9 c
Methyl tert-butyl ether (MTBE) POD is a NOAEL of 300 mg/kg bw/d as derived by the Dutch National Institute for Health and the
Environment (RIVM) (Swartjes et al., 2004) from a 90-day oral toxicity study in rats (Robinson et al.,
1990).
B 1000 0.3 9400 e
Metoprolol POD is a LOEL of 100 mg/person/d (1.42 mg/kg bw/d) as derived by the Dutch National Institute for
Health and the Environment (RIVM) (Versteegh et al., 2007).
B 100 0.014 50
n-Butylbenzenesulphonamide POD is a NOAEL of 50 mg/kg bw/d for adverse treatment related macro- or microscopic effects (liver
enlargement, hepatocyte hypertrophy, thymic atrophy and lymphocytolysis) as derived from a study
according to OECD guideline 407 (GLP compliant) in which rats were administered nbutylbenzenesulphonamide
via oral gavage for 28 days (Proviron Fine Chemicals, 2003), with an
uncertainty factor of 600 (100 for inter- and intraspecies differences, 6 for extrapolation to chronic
exposure).
D 600 0.083 292
Nicosulfuron POD is a NOAEL of 199 mg/kg bw/d as derived from a 2-year study in rats (URL3). B 1000 0.2 700
n-Nitrosodimethylamine (NDMA) POD is a tumor dose (TD05, dose level that causes 5% in increase in tumour incidence over the
background) of 18 mg/kg bw/d as derived by the WHO (2008) from a detailed 2-year cancer dose–
response study in rats (Peto et al., 1991a,b).
A NA NA 0.1 c,d
p,p 0 -Sulfonyldiphenol POD is a NOEL of 10 mg/kg bw/d for histopathological effects (hyperplasia of the mucosal epithelium,
hypertrophy of hepatocytes), decreased food consumption/body weight gain and increased liver
weights as derived from a study according to OECD guideline 421 in which male and female rats were
administered p,p 0 -sulfonyldiphenol via oral gavage for respectively 45 days and from 14 days before
mating to day 3 of lactation (Mitsubishi Chemical Safety Institute Ltd, date of study unknown), with an
uncertainty factor of 600 (100 for inter- and intraspecies differences, 6 for extrapolation to chronic
exposure).
Perfluoroctane sulfonate (PFOS) POD is a NOAEL of 0.03 mg/kg bw/d as derived by the Scientific Panel on Contaminants in the Food
Chain (CONTAM) (EFSA, 2008) from a 182-day study in Cynomolgus monkeys (Seacat et al., 2002).
Perfluorooctanoic acid (PFOA) POD is a benchmark dose level (BMDL10) of 0.3 mg/kg bw/d as derived by the Scientific Panel on
Contaminants in the Food Chain (CONTAM) (EFSA, 2008) from a number of studies in mice and male
rats (study references unknown).
Phenazone POD is a LOEL of 250 mg/person/day (3.57 mg/kg bw/d) as derived by the Dutch National Institute for
Health and the Environment (RIVM) (Versteegh et al., 2007).
Simazine POD is a NOAEL of 0.52 mg/kg bw/d as derived by the WHO (1996) from a 2-year combined chronic
toxicity/oncogenicity study in rats (Ciba-Geigy, 1988; unpublished study submitted to WHO).
Sulfamethoxazole POD is a NOAEL of 25 mg/kg/d as derived by Schwab et al. (2005) from a 60-week study in rats (Swarm
et al., 1973).
Tolyltriazole POD is a NOAEL of 150 mg/kg bw/d for observed mild apathy as derived from a study in which rats were
administered tolyltriazole via oral gavage for 29 days (Benzotriazoles Coalition, 2001; ECB, 2000a), with
an uncertainty factor of 600 (100 for inter- and intraspecies variation and 6 for extrapolation to chronic
exposure).
D 600 0.017 60
B 200 0.00015 0.5
B 200 0.0015 5.3
B 100 0.036 125
A 1000 0.52 2 c
B 200 0.13 440
D 600 0.25 875
(continued on next page)
water research 44 (2010) 461–476 467

Table 2 (continued)
Compound Point of departure (POD) Category a SUF b TDI, ADI or RfD
(mg/kg bw/d)
Trichloroethene POD is a benchmark dose level (BMDL10) of 0.146 mg/kg bw/d as derived by the WHO (2005b)/Health
Canada (2003) from a developmental toxicity study in rats (Dawson et al., 1993).
Triethylphosphate (ethylphosphate)
(TEP)
POD is a NOEL of 335 mg/kg bw/d for fertility effects (effects on litter size) as derived from a study in
which rats were administered TEP via the food for a unknown period (OECD/SIDS, 1998), with an
uncertainty factor of 600 (100 for inter- and intraspecies variation and 6 for extrapolation to chronic
exposure).
Triphenylphosphine oxide (TPPO) POD is a NOAEL of 8 mg/kg bw/d for salivation, vomiting, diarrhea, histopathological (liver damage and
skeletal muscle atrophy)/haematological (elevated GPT, GOT and alkaline phosphatase activities,
reduced haemoglobin/haematocrit levels) parameters as derived from a study in which dogs were
administered TPPO via the food for 3 months (ECB, 2000d), with an uncertainty factor of 1000 (100 for
inter- and intraspecies variation and 10 for extrapolation to chronic exposure).
Tris(2-chloroethyl)phosphate (TCEP) POD is a NOAEL of 22 mg/kg bw/d for increased relative liver and kidney weights as derived from
a study in which rats were administered TCEP via oral gavage for 16 weeks (NTP, 1991), with an
uncertainty factor of 1000 (10 for inter- and intraspecies variation and 10 for extrapolation to chronic
exposure and uncertainty in genotoxic potential).
Urotropine (methenamine, hexamine) POD is a NOAEL of 15 mg/kg bw/d as derived by the WHO/JECFA (1974) from a teratogenicity study
(exposure from the fourth to fifty-sixth day after mating) in dogs (Hurni and Ohder, 1973).
(Provisional)
guideline
value (mg/L)
A 100 0.0015 20 c
D 600 0.56 1950
D 1000 0.008 28
D 1000 0.022 77
B 100 0.15 500
NA not available.
a Categories: A) statutory drinking water guideline value available; B) established TDI, ADI or RfD available; C) TDI calculated with a established LO(A)EL or NO(A)EL; D) TDI calculated with
miscellaneous toxicological information.
b Uncertainty factors.
c WHO drinking water guidelines (WHO, 2006).
d Based on a specific cancer risk level of 10 5 .
e The odour-threshold value for drinking water preparation is 15 mg/L for MTBE and w1 mg/L for ETBE (Swartjes et al., 2004; Van Wezel et al., 2009).
f Using the Stokinger–Woodward approach (Stokinger and Woodward, 1958), a TDI of 150 mg/person/d (2.14 mg/kg bw/d) can be calculated from the 8-h threshold limit value (15 mg/m 3 ) assuming
100% oral/inhalatory absorption and a 8-h total workshift ventilation of 10 m 3 .
g Using the Stokinger–Woodward approach (Stokinger and Woodward, 1958), a TDI of 40 mg/kg/person/d (0.54 mg/kg bw/d) can be calculated from the 8-hour threshold limit value (3.8 mg/m 3 )
assuming 100% oral/inhalatory absorption and a 8-h total workshift ventilation of 10 m 3 .
h A TDI of 0.15 mg/kg bw/day can be calculated from the NOAEL (500 ppm ¼ 29.1 mg/kg/bw/d) assuming 100% oral/inhalatory absorption, a specific lung retention of 25% and a minute ventilation
volume for rat of 45 mL.
468
water research 44 (2010) 461–476

guideline value) in order (i) to provide perspective on what the
occurrence of emerging contaminants might signify to human
health and (ii) to help prioritize further investigations. A BQ
value of 1 represents a (drinking) water concentration equal to
the (provisional) guideline value. Compounds with a BQ value
of 1 in drinking water may be of potential human health
concern if the water were to be consumed over a lifetime
period. Compounds with a BQ value 0.1 in drinking water
were identified as those that may warrant further investigation;
this is consistent with various US State and Federal
practices (Toccalino, 2007). For compounds found in surface
waters and groundwater the BQ value threshold to carry out
an additional assessment was set at an arbitrary value of 0.2,
since these source waters are purified in drinking water
treatment plants which provides extra safety. Compounds in
surface waters/groundwater or drinking water with a BQ value
of 0.2 or 0.1 respectively, are presumed to present no
appreciable concern to human health.
3. Results
3.1. Selection of compounds
For only 50 compounds out of the original list, statutory
guideline values or useful toxicity and occurrence data could
be found. These compounds constitute the final list, which
includes compounds from various groups such as iodinated
contrast media, pharmaceuticals, oxygenated gasoline additives,
perfluorinated organic compounds, miscellaneous
organic compounds and pesticides (Table 1). Natural and
synthetic steroid hormones such as 17b-estradiol, 17a-ethynylestradiol
and estrone were not included in this assessment,
as they are removed relatively easily in drinking water
purification processes (Nghiem et al., 2004).
3.2. (Provisional) drinking water guideline values
For 10 compounds WHO statutory drinking water guideline
values were available and these compounds were classified
as category A. For the remaining 40 compounds a provisional
guideline value was established with the aid of toxicological
data. An established TDI, ADI or RfD was available for 22
compounds (category B). In 7 cases when there was no TDI,
ADI or RfD available, an established NO(A)EL or LO(A)EL was
used to calculate a TDI and subsequently a provisional
drinking water guideline value (category C). For the remaining
11 compounds, miscellaneous toxicological data was
used to calculate a TDI and subsequently a provisional
drinking water guideline value (category D). As tabulated in
Table 2, (provisional) guideline values ranged from
0.0001 mg/L for NDMA to 415 mg/L for the iodinated contrast
medium iopamidol. All iodinated contrast media had relatively
high provisional guideline values, ranging from 6.7 mg/
L (iomeprol) to 415 mg/L (iopamidol). In the cases of MTBE
and ETBE, the human health based guideline values were at
least one order of magnitude higher than the corresponding
odour-threshold based guideline values of respectively 15 mg/
L and w1 mg/L (Swartjes et al., 2004; Van Wezel et al., 2009).
For two compounds (DMA and DEA) the provisional guideline
water research 44 (2010) 461–476 469
values were established by route-to-route extrapolation of
inhalatory LOAELs to oral LOAELs. Since benzene was evaluated
to be genotoxic/carcinogenic (IARC group I) and NDMA
(IARC group 2A) and 1,4-dioxane (IARC group 2B) are respectively
suspected non-genotoxic and genotoxic carcinogens,
their corresponding (provisional) guideline values are
provided as an upper bound lifetime cancer risk to an individual
of 10 5 , or the odds that one case of cancer would
result for every 100 000 persons subjected to continuous
exposure over a 70-year lifetime.
3.3. Concentration of compounds in surface waters,
groundwaters and drinking water
The maximum concentrations of compounds reported in
surface waters and/or groundwaters are summarized in
Table 3. Measured maximum surface water concentrations
were available for 37 of the 50 compounds in the annual
reports of RIWA and ARW. For two compounds (MTBE and
clofibric acid) maximum concentrations in Dutch groundwater
are reported. For the remaining compounds, the
maximum concentration reported in surface waters was
taken from other sources (see Section 2.2.3).
The six compounds with the highest reported maximum
concentrations in surface waters were EDTA (29 mg/L), DTPA
(12.2 mg/L), p,p 0 -sulfonyldiphenol (10 mg/L), urotropine (10 mg/
L), 1,4-dioxane (10 mg/L) and AMPA (5 mg/L), whereas in
groundwater a relatively high concentration was found for
MTBE (27.3 mg/L) showing the environmental relevance of this
compound. The highest maximum concentration of iodinated
contrast media in surface waters was reported for
iomeprol (0.97 mg/L).
Table 3 also summarizes the maximum concentrations of
compounds reported in drinking water. Data on the occurrence
of compounds in drinking water were relatively scarce,
and limited to 35 compounds. For 18 compounds, drinking
water concentrations were obtained from the Dutch REWAB
database and for 17 compounds drinking water concentrations
were taken from reports by others. Drinking water
concentrations for the remaining compounds could not be
found. The highest maximum concentration reported was for
EDTA (13.6 mg/L), followed by DTPA (9 mg/L), metoprolol
(2.1 mg/L) and BCIPE (1.9 mg/L).
3.4. Comparison of compound concentrations to
(provisional) guideline values (BQ value)
For all compounds found in surface waters, groundwaters
and drinking water the calculated BQ value was

Table 3 – Reported concentrations in surface waters, groundwaters and drinking water and comparison to (provisional) drinking water guideline values expressed as
Benchmark Quotient (BQ) values.
Compound Surface waters and groundwaters Drinking water
Max conc (mg/L) (number
of measurements, year)
Source Ref BQ
value a
Max conc (mg/L) (number
of measurements, year)
Source Ref BQ
value a
1,4-Dioxane 10 (NA, 1997) SW, NL (4) 0.3 0.5 (NA) UDW, NL (4) 0.02
2,6-Dichlorobenzamide (BAM) 0.05 (40, 2002-2006) SW, NL (11) 0.001 0.23 (14, 2002) FDW, USA (10) 0.004
4-Methylbenzenesulfonamide
(p-toluenesulfonamide,
4-tolylsulfonamide)
0.06 (20, 2005) SW, NL (13) 0.00002 NA
Acetylsalicylate (aspirin, acetyl
salicylic acid)
0.065 (NA, 2007) SW, NL (15) 0.003 0.12 (12, 2007) FDW, NL (15) 0.005
Alpha-amino-3-hydroxy-5-methyl-
4-isoxazole propionic acid (AMPA)
5 (499, 2005) SW, NL (11) 0.006 1.1 (6, 2001) FDW, NL (10) 0.001
Amidotrizoic acid (diatrizoic acid) 0.63 (189, 2006) SW, GER (2) 0.000003 0.25 (6, 2006) FDW, NL (10) 0.000001
Bentazone 0.1 (126, 2007) SW, NL (11) 0.0003 0.28 (11, 2006) FDW, NL (10) 0.0009
Benzene 0.74 (116, 2001) SW, NL (11) 0.07 0.96 (12, 2005) TW, NL (10) 0.1
Benzothiazole 0.03 (3, 2008) SW, NL (6) 0.0003 0.01 (10, 2007) FDW, NL (14) 0.0001
Benzotriazole (1H-benzotriazole) 0.54 (11, 2007) SW, NL (11) 0.0005 0.2 (10, 2007) FDW, NL (14) 0.0002
Bis(chloroisopropyl)ether (BCIPE) 2.9 (15, 1984-1985) GW, NL (6) 0.02 1.9 (9, 1982-1984) FDW, NL (6) 0.01
Carbamazepine 0.227 (263, 2003) SW, NL (11) 0.2 0.03 (2, 2007) FDW, NL (15) 0.03
Carbendazim 1.5 (111, 2006) SW, BE (11) 0.01 NA
Chloridazon (pyrazon) 0.3 (68, 2002) SW, BE (11) 0.002 NA
Clofibric acid 0.091 (NA, 2007) BFGW, NL (15) 0.003 0.14 (2, 2007) FDW, NL (15) 0.005
Dichlorophenoxyacetic acid (2,4-D) 0.2 (34, 2006) SW, NL (11) 0.007 0.11 (5, 2002) TW, NL (10) 0.004
Diethyl phthalate 0.9 (8, 2005) SW, NL (11) 0.0003 NA
Diethyl toluamide (DEET) 0.06 (36, 2005) SW, NL (11) 0.00001 0.03 (1, 2005) TW, NL (10) 0.000005
Diethylamine (DEA) 0.29 (38, 2007) SW, NL (11) 0.0004 NA
Diethylene glycol dimethyl
ether (diglyme, bis(2-methoxy ethyl)ester))
3.64 (11, 2007) SW, NL (11) 0.02 0.15 (13, 2007) UDW, NL (1) 0.0009
Diethylene triamine penta acetic acid (DTPA) 12.2 (53, 2005) SW, NL (11) 0.03 9 (2, 2001) FDW, NL (10) 0.03
Dimethenamid 0.12 (4, 2005) SW, NL (9) 0.0005 NA
Dimethylamine (DMA) 0.34 (42, 2005) SW, NL (11) 0.002 NA
Diuron 0.68 (386, 2002) SW, BE (11) 0.1 0.08 (2, 2005) TW, NL (10) 0.01
Ethyl tert-butyl ether (ETBE) 1.2 (97, 2006) SW, GER (2) 0.002 (1.2 b ) NA
Ethylenediamine tetra acetic Acid (EDTA) 29 (192, 2005) SW, NL (11) 0.05 13.6 (7, 2001) FDW, NL (10) 0.02
Glyphosate 1.2 (291, 2006) SW, NL (11) 0.001 0.46 (3, 2006) TW, NL (10) 0.0005
Imidacloprid 0.06 (1, 2007) SW, NL (11) 0.0003 NA
Iohexol 0.5 (180, 2005) SW, NL (11) 0.000001 0.06 (1, 2007) FDW, NL (15) 0.0000002
Iomeprol (iomeron) 0.97 (172, 2007) SW, NL (11) 0.0001 0.01 (1, 2006) FDW, NL (10) 0.000001
Iopamidol 0.714 (188, 2006) SW, NL (11) 0.000002 0.1 (6, 2006) FDW, NL (10) 0.0000002
Iopromide 0.56 (186, 2004) SW, NL (11) 0.000002 0.04 (2, 2007) FDW, NL (15) 0.0000002
Isoproturon 0.31 (256, 2002) SW, NL (11) 0.03 0.02 (1, 2004) TW, NL (10) 0.002
Methyl tert-butyl ether (MTBE) 27.3 (14, 2003-2005) GW, NL (7) 0.003 (1.8 b ) 1.25 (27, 2006) FDW, NL (10) 0.0001 (0.08 b )
Metoprolol 0.2 (114, 2006) SW, NL (11) 0.004 2.1 (2, 2005) FDW, NL (10) 0.04
n-Butylbenzenesulphonamide 0.78 (300-400, 2004-2006) SW, NL (6) 0.003 0.05 (2, 2004) TW, NL (13) 0.0002
Nicosulfuron 0.17 (5, 2007) SW, NL (11) 0.0002 NA
n-Nitrosodimethylamine (NDMA) 0.0071 (38, 2006) SW, NL (11) 0.07 0.002 (21, 2007) UDW, NL (5) 0.02
470
water research 44 (2010) 461–476

p,p0-Sulfonyldiphenol 10 (300, 2005-2006) SW, BE (6) 0.1 NA
Perfluoroctane sulfonate (PFOS) 0.11 (NA, 2008) SW, NL (8) 0.2 0.02 (NA, 2005/2006) TW, GER (12) 0.04
Perfluorooctanoic acid (PFOA) 0.647 (NA, 2005/2006) SW, GER (13) 0.1 0.52 (NA, 2005/2006) TW, GER (12) 0.1
Phenazone 0.11 (26, 2005) SW, NL (12) 0.0009 0.03 (8, 2005) FDW, NL (15) 0.0002
Simazine 0.13 (124, 2006) SW, BE (11) 0.07 0.06 (4, 2004) FDW, NL (10) 0.03
Sulfamethoxazole 0.11 (170, 2005) SW, NL (11) 0.0003 0.03 (4, 2007) FDW, NL (15) 0.00007
Tolyltriazole 0.29 (11, 2007) SW, NL (11) 0.0003 NA
Trichloroethene 1.35 (206, 2006) SW, NL (11) 0.07 1.75 (8, 2005) FDW, NL (10) 0.09
Triethylphosphate (ethylphosphate) (TEP) 0.189 (25, 2007) SW, NL (11) 0.0001 NA
Triphenylphosphine oxide (TPPO) 0.344 (88, 2003) SW, NL (11) 0.01 0.13 (6, 2007) FDW, NL (6) 0.005
Tris(2-chloroethyl)phosphate (TCEP) 0.29 (8, 2006) SW, NL (11) 0.004 NA
Urotropine (methenamine, hexamine) 10 (NA, 1997-1998) SW, GER (3) 0.02 NA
Sources: bank-filtrated groundwater (BFGW); groundwater (GW); surface water (SW); finished drinking water (FDW); tap water (TW); unspecified drinking water (UDW). Locations: Belgium (BE);
Germany (GER); the Netherlands (NL); United States of America (USA). References: (1) Anonymous data Dutch drinking water companies; (2) ARW database; URL4; (3) Brauch et al., 2000; (4) ECB, 2002; (5)
Kleinneijenhuis and Puijker, 2008; (6) KWR internal data (KWR, 2009); (7) De Voogt et al., 2008; (8) Loos et al., 2009; (9) Mout et al., 2007; (10) REWAB database, 2009; (11) RIWA database; URL5 (12)
Skutlarek et al., 2006; (13) van Beelen, E.S.E. (HWL, the Netherlands), personal communication; (14) Van Leerdam et al., 2009; (15) Versteegh et al., 2007. NA: Not available; an absence of measured
concentrations above the detection limit.
a BQ value: ratio of reported maximum concentration to the (provisional) guideline value (cf. Table 2).
b Based on an odour-threshold value (cf. Table 2).
water research 44 (2010) 461–476 471
there is a concern for drinking water production, because
consumers will not accept odourous drinking water. The BQ
values of the remaining compounds calculated for surface
waters and groundwater ranged between 0.1 (p,p 0 -sulphonylphenol,
diuron) and 0.000001 (iohexol). For all iodinated
contrast media the concentration in surface waters was at
least three orders of magnitude less than the (provisional)
guideline value. For drinking water, the BQ values of these
compounds were even lower (Fig. 2B). For two compounds
(benzene and PFOA) found in drinking water, the BQ value
was equal to 0.1 indicating that additional assessments such
as establishing trends may be warranted. However, for 15
compounds occurrence data in drinking water were not
available and therefore the human health concern associated
with drinking water consumption due to presence of any of
these compounds remains unknown.
4. Discussion
Rapid new developments in analytical chemistry lead to the
detection and quantification of many emerging contaminants
in drinking water and its environmental sources (surface
water and groundwater). Since toxicological information is
often absent, such compounds are a growing concern for
drinking water companies and their customers.
The present study attempts to address potential human
health concern associated with water containing emerging
contaminants. The 50 compounds included in this study
represent a broad range of chemical classes for which
maximum concentrations in surface waters, groundwater
and/or drinking water were obtained in the downstream parts
of the Rhine and Meuse basins. The results as presented in
Fig. 2 indicate that a substantial margin exists between the
(provisional) guideline value and the maximum concentrations
of most compounds reported in surface waters,
groundwaters and/or drinking water.
The compounds evaluated with a relatively high BQ value
(i.e. a high potential human health concern) and a known
carcinogenic action are 1,4-dioxane, benzene and NDMA. The
(provisional) guideline values for 1.4-dioxane (30 mg/L), benzene
(10 mg/L) and NDMA (0.1 mg/L) used in the present study are
based on a specific cancer risk level of 10 5 . However, when
applying a specific risk level of 10 6 ,asiscommonpracticeinthe
Netherlands, the provisional guidelines value would be 3 mg/L,
1 mg/L and 0.01 mg/L, respectively. This would result in BQ values
much higher than the arbitrary thresholds for surface waters
and drinking water employed in the present study. This indicates
that very low concentrations of these compounds in
drinking water could lead to a potential carcinogenic effect, and
we conclude that for these compounds it is important to
monitor trends in their (environmental) occurrence.
Furthermore, Fig. 2 illustrates that the (provisional) guideline
values of the majority of non-genotoxic compounds are at least
two orders of magnitude above the Threshold of Toxicological
Concern (TTC)-based drinking water target value for non-genotoxic
compounds (0.1 mg/L). In addition, the (provisional)
guideline values (expressed as a specific risk level of 10 6 )ofthe
three compounds with known carcinogenic action (1,4-dioxane,
benzene and NDMA) are equal (NDMA) or much higher (1,4

472
Fig. 2 – Comparison of compound concentrations in (A) surface/groundwaters and (B) drinking water to (provisional)
guideline values. Benchmark Quotient (BQ) thresholds are indicated with dashed lines. Threshold of Toxicological Concern
(TTC) based target value for non-genotoxic compounds (0.1 mg/L) is indicated with a dotted line. Numbers correspond to
compounds as tabulated in Table 1.
dioxane and benzene) than the TTC derived target value of
0.01 mg/L for genotoxic compounds. This illustrates that the TTC
based drinking water target value may be a conservative value
ideally suited for exposure based waiving of compounds for
which there is no sufficient toxicological information, which
can be followed up by a more data-intensive evaluation.
Two perfluorinated organic compounds were evaluated in
the present study. For PFOA in drinking water a BQ value equal
to the arbitrary threshold of 0.1 was calculated, whereas for
PFOS in surface waters a BQ value of 0.2 was calculated. These
persistent compounds are becoming a global problem, and
PFOA and PFOS have already been detected in the ng/L range
in, e.g. European and Japanese tapwaters (Ericson et al., 2007;
Loos et al., 2007; Norimitsu et al., 2004). Recently, Skutlarek
et al. (2006) observed at sampling site Neheim (river Ruhr
catchment, a tributary of the river Rhine, Germany)
water research 44 (2010) 461–476
concentrations of PFOA of 0.65 mg/L in Lake Moehne, and
0.53 mg/L in corresponding drinking water, respectively. The
authors concluded that water treatment steps may not
effectively eliminate perfluorinated compounds to a sufficient
extent, although approximately 50% of the waterworks at the
Ruhr river are equipped with activated carbon filters. Hence,
more research should be devoted to the behavior of perfluorinated
organic compounds in drinking water treatment
processes.
Several structurally related iodinated contrast media
(iopamidol, iohexol, iomeprol and iopromide) were evaluated
in the present research. However, their calculated BQ values
are much lower than the BQ threshold above which further
investigations would be warranted. Iopromide, for example, is
a relatively non-toxic compound with a reported safe dose
(intravenous) of 50 g/person/d (Versteegh et al., 2007). Despite

the absence of human health effects, these compounds may
deserve further attention from an environmental impact
point of view. Since environmental sublethal effects of
iodinated contrast media to organisms are largely unknown,
taken together with high persistence and environmental
presence at relatively high concentrations, additional environmental
assessments may be necessary.
For compounds with a low (provisional) guideline value as
identified in the present study (e.g. carbamazapine), additional
environmental monitoring may be warranted to characterize
concentrations and to establish trends in their occurrence. As
shown by Walraven and Laane (2009), river flow rates may
influence contaminant concentrations seasonally, thus
resulting in substantially varying BQ values. For example, it can
be observed that the riverine concentration of the fuel
oxygenate MTBE is highly dependent on the flow of the river
Meuse. Similar patterns may occur for other compounds,
resulting in (temporarily) exceedance of the BQ threshold.
The evaluation as presented here supports the conclusion
that the majority of the selected compounds as found in
surface waters, groundwater and drinking water do not pose
an appreciable concern to human health. This finding of no
adverse effect to human health from exposure to trace
quantities of compounds (e.g. pharmaceuticals) in surface
waters and/or drinking water is supported by other results
reported in the literature. Kingsbury et al. (2008) recently
evaluated the potential health effects of 148 organic
compounds in source water and finished water. The authors
showed that the annual mean concentration of all compounds
detected in finished water was less than the established
human health benchmarks. Furthermore, Snyder et al. (2008)
arrived at the same findings after evaluating human health
effects associated with potential drinking water exposure of
a suite of 62 indicator pharmaceuticals and potential endocrine
disrupting compounds.
Despite the absence of any concern to human health,
drinking water remains a major point of consumer concern
and some residual uncertainties need further exploration. For
example, drinking water guideline values are developed using
toxicity information for single compounds. Hence, the longterm
cumulative dose-additive or synergistic effects of low
concentrations of contaminants co-occurring as mixtures on
human health and potentially sensitive sub-populations
remain currently unknown. Understanding and implementing
of such information is important for the development of
future (enforceable) guideline values. Finally, the relatively
large data gap on occurrence of compounds in drinking water
should compel further research and assessment, especially
for those compounds with a low (provisional) guideline value.
5. Major conclusions
For most compounds evaluated in the present assessment,
a substantial margin exists between the (provisional)
guideline value and the maximum concentrations in
surface waters, groundwaters and/or drinking water.
The TTC based drinking water target values (0.1 mg/L and
0.01 mg/L for non-carcinogenic compounds, respectively) as
proposed earlier are the conservative values they are meant
water research 44 (2010) 461–476 473
to be. They are optimally suited to provide exposure based
waiving.
The concentrations in drinking water of compounds such as
MTBE, ETBE, 1,4-dioxane, NDMA and benzene should be
monitored closely, since their guideline values are easily
exceeded.
Alkylated perfluorinated compounds such as PFOA and
PFOS are environmentally persistent compounds and their
increasing occurrence in (the sources of) drinking water
should be monitored closely.
For compounds with a very low (provisional) guideline value
(e.g. mutagenic and carcinogenic compounds) it is important
to better establish trends in their environmental occurrence.
From a toxicological point of view iodinated contrast media
as present in drinking water, such as amidotrizoic acid
iopamidol, iohexol and iopromide, are not a direct concern
for human health. However, further environmental assessment
may be necessary, especially since the sublethal
(ecological) effects of these compounds are largely unknown.
Better understanding of the potential mixture effects of
emerging compounds present in drinking water is important
for the development of future guideline values.
Acknowledgements
The authors wish to acknowledge the significant contribution
to this work by Leo Puijker (KWR) and Ruud Jansen. Thanks
also to Dr. Corine Houtman (HWL, the Netherlands) for her
valuable comments on the manuscript. The research
described was funded by the Joint Research Programme of the
Dutch Water utilities (BTO).
references
Benzotriazoles Coalition, 2001. Benzotriazoles category
justification and testing rationale. Synthetic Organic Chemical
Manufacturers Organization. http://www.epa.gov/oppt/
chemrtk/pubs/summaries/benzo/c13456.pdf (accessed June
2009).
BIBRA Toxicology International, 1995. Toxicity Profile of
Benzotriazole. BIBRA Toxicology International, Carshalton
(Surrey), UK.
Brauch, H.J., Sacher, F., Denecke, E., Tacke, T., 2000. Efficiency of
river-bank filtration for the removal of polar organic
compounds (Werksamkeit der Uferfiltration fuer die
Enfernung von polaren organischen Spurenstoffen). Wasser-
Abwasser 141 (4), 226–234.
Brown, D., Butterworth, K.R., Gaunt, I.F., Grasso, P., Gangolli, S.D.,
1978. Short-term oral toxicity study of diethyl phthalate in the
rat. Food Cosmet. Toxicol. 16, 415–422.
California EPA (California Environmental Protection Agency),
2000. N,N-diethyl-meta-toluamide (DEET) risk
characterization document. Department of pesticide
regulation. Document number RCD 00-01, Sacramento,
California, USA.
CIIT (Chemical Industry Institute of Toxicology), 1990. Twenty
four month final report inhalation toxicity of dimethylamine
in F344 rats and B6C3F1 mice. CIIT Archives, Docket #11957.
Ciba-Geigy, 1988. Combined chronic toxicity/oncogenicity study
in rats. Basle, Ciba-Geigy, 1988 (unpublished study submitted
to WHO).

474
Cotruvo, J.A., 1988. Drinking water standards and risk
assessment. Regul. Toxicol. Pharmacol. 8 (3), 288–299.
Danish EPA (Danish Environmental Protection Agency), 2004.
Evaluation of Health Hazards by Exposure to BAM (2,6-
Dichlorobenzamide) and Risk Characterisation of Drinking
Water Exposure. Environmental Project Nr. 943 2004.
Dawson, B.V., Johnson, P.D., Goldberg, S.J., Ulreich, J.B., 1993.
Cardiac teratogenesis of halogenated
hydrocarbon-contaminated drinking water. J. Am. Coll.
Cardiol. 21 (6), 1466–1472.
De Voogt, P., Vink, C., Puijker, L.M., 2008. MTBE and ETBE in Dutch
groundwater (MTBE en ETBE in Nederlands grondwater), KWR
08.043, KWR, Nieuwegein, The Netherlands.
DuPont de Nemours and Company, Inc., 1964. Report MRID No.
00017763, 00080899.
DECOS (Dutch Expert Committee for Occupational Standards),
2000. 1,2,3-benzotriazole. Health-based recommended
occupational exposure limit. No. 2000/14OSH, The Hague, The
Netherlands.
ECB (European Chemicals Bureau), 2000a. IUCLID dataset on
benzotriazoles.
ECB (European Chemicals Bureau), 2000b. IUCLID dataset on
Chloridazon.
ECB (European Chemicals Bureau), 2000c. IUCLID dataset on
Diethylene triamine penta acetic acid.
ECB (European Chemicals Bureau), 2000d. IUCLID dataset on
triphenylphosphine oxide.
ECB (European Chemicals Bureau), 2002. European Risk
Assessment Report 1,4-dioxane.
EFSA (European Food Safety Authority), 2008. Opinion of the
scientific panel on contaminants in the food chain on
perfluorooctane sulfonate (PFOS), perfluorooctanoic acid
(PFOA) and their salts. EFSA J. 53, 1–131.
Ericson, I., Nadal, M., van Bavel, B., Lindström, G., Domingo, J.L.,
2007. Levels of perfluorochemicals in water samples from
Catalonia, Spain: is drinking water a significant contribution
to human exposure? Environ. Sci. Pollut. Res. 15, 614–619.
Goldenthal, E.I. (International Research and Development Corp.),
1994. Evaluation of DEET in a one-year chronic oral toxicity
study in dogs. DEET Joint Venture/Chemical Specialties
Manufacturers Assoc. DPR vol. 50191-164, Rec. No. 131265.
Health Canada, 2003. Benchmark Dose for TCE in Drinking Water.
Health Canada, Healthy Environments and Consumer Safety
Branch, Biostatistics Unit, April, Ottawa, Ontario.
Hogenboom, A.C., Van Leerdam, J.A., De Voogt, P., 2009. Accurate
mass screening and identification of emerging contaminants
in environmental samples by liquid chromatography–hybrid
linear ion trap Orbitrap mass spectrometry. J. Chromatogr. A
1216 (3), 510–519.
Hurni, H., Ohder, H., 1973. Reproduction study with formaldehyde
and hexamethylenetetramine in beagle dogs. Food Cosmet.
Toxicol. 11 (3), 459–462.
Kingsbury, J.A., Delzer, C.D., Hopple, J.A., 2008. Anthropogenic
organic compounds in source water of nine community water
systems that withdraw from streams, 2002–2005, USGS,
Reston, Virginia, USA, Report 2008–5208.
Kleinneijenhuis, A.J., Puijker, L.M., 2008. Monitoring program for
DMSA, DMST, N,N-dimethylsulfamide and its ozone reaction
product NDMA in untreated and ozone-treated water. TNO
report 7686, Delft, The Netherlands.
Kolpin, D.W., Skopec, M., Meyer, M.T., Furlong, E.T., Zaugg, S.D.,
2004. Urban contribution of pharmaceuticals and other
organic wastewater contaminants to streams during differing
flow conditions. Sci. Total Environ. 328 (1–3), 119–130.
Kroes, R., Renwick, A.G., Cheeseman, M., Kleiner, J.,
Mangelsdorf, I., Piersma, A., Schilter, B., Schlatter, J.,
Van Schothorst, F., Vos, J.G., Wurtzen, G., 2004. Structurebased
thresholds of toxicological concern (TTC): guidance for
water research 44 (2010) 461–476
application to substances present at low levels in the diet.
Food Chem. Toxicol. 42, 65–83.
Krop, H. de Voogt, P., 2008. Physicochemical parameters and
source markers of PFAS. Report IVAM ISO doc O/0818,
Amsterdam, The Netherlands, pp. 1–147.
KWR Watercycle Research Institute, 2009. KWR internal data,
Nieuwegein, the Netherlands.
Larsen, M.L., Illingworth, D.R., O’Malley, J.P., 1994. Comparative
effects of gemfibrozil and clofibrate in type III
hyperlipoproteinemia. Atherosclerosis 106, 235–240.
Loos, R., Wollgast, J., Huber, T., Hanke, G., 2007. Polar herbicides,
pharmaceutical products, PFOS, PFOA, and nonylpheniol and
its carboxylates and ethoxylates in surface and tap waters
around Lake Maggiore in Northern Italy. Anal. Bioanal. Chem.
387, 1469–1478.
Loos, R., Gawlik, B.M., Locoro, G., Rimaviciute, E., Contini, S.,
Bidoglio, G., 2009. EU-wide survey of polar organic persistent
pollutants in European river waters. Environ. Pollut. 157 (2),
561–568.
Lynch, D.W., Moorman, W.J., Stober, P., Lewis, T.R., Iverson, W.O.,
1986. Subchronic inhalation of diethylamine vapour in Fisher-
344 rats: organ toxicity. Fund. Appl. Toxicol. 6 (3), 559–565.
Medinsky, M.A., Wolf, D.C., Cattley, R.C., Wong, B., Janszen, D.B.,
Farris, G.M., Wright, G.A., Bond, J.A., 1999. Effects of
a thirteen-week inhalation exposure to ethyl tertiary
butyl ether on Fischer-344 rats and CD-1 mice. Toxicol. Sci.
51, 108–118.
Mitsubishi Chemical Safety Institute Ltd. (date of study unknown).
4,4’-sulfonyldiphenol. Summary of repeated dose and
developmental studies. Kashima Laboratory, Ibaraki, Japan.
Mitsumori, K., Usui, T., Takahashi, K., Shirasu, Y., 1979. Twentyfour
month chronic toxicity studies of dichlorodiisopropyl
ether in mice. J. Pestic. Sci. 4 (3), 323–335.
Mons, M., Van der Hoek, J.P., Stoks, P., Van der Kooij, D., 2008.
Vision on and guideline values for xenobiotic substances in
drinking water (Visie op en streefwaarden voor
milieuvreemde stoffen in drinkwater). H 2O 4, 1–4 (in Dutch).
Morgareidge, K., 1971. 90-Day feeding study with benzothiazole in
rats. Unpublished report from Food and Drug Research
Laboratories, Inc., Maspeth, New York, USA. Submitted to
WHO by Flavor and Extract Manufacturers’ Association of the
United States.
Morisetti, A., Tirone, P., Luzzani, F., de Haen, C., 1994.
Toxicological safety assessment of iomeprol, a new X-ray
contrast agent. Eur. J. Radiol. 1 (18 Suppl), S21–31.
Mout, L., Jansen, L., van der Pijll, S., Nierop, S., 2007. Implementation
program dimethenamid-p. Clean sources, now and in the future
(uitvoeringsprogramma dimethenamid-p. Schone bronnen, nu
en in de toekomst: tweede reeks knelpunten (http://www.
schonebronnen.nl/). The Hague, The Netherlands.
Munro, I.C., Ford, R.A., Kennepohl, E., Sprenger, J.G., 1996.
Correlation of structural class with no-observed effect levels:
a proposal for establishing a threshold of concern. Food Chem.
Toxicol 34, 829–867.
NCI (National Cancer Institute), 1978. Bioassay of 1,4-dioxane for
possible carcinogenicity, CAS No. 123-91-1. NCI
Carcinogenesis Technical Report Series No. 80. DHEW
Publication No. (NIH) PB-285-711, Bethesda, Maryland, USA.
Nghiem, L.D., Schafer, A.I., Elimelech, M., 2004. Removal of
natural hormones by nanofiltration membranes:
measurement, modeling, and mechanisms. Environ. Sci.
Technol. 38 (6), 1888–1896.
Norimitsu, S., Kouji, H., Kayoko, I., Kazuaki, S., Takeo, Y., Akio, K.,
2004. Perfluorooctanoate and perfluorooctane sulfonate
concentrations in surface water in Japan. J. Occup. Health 46,
49–59.
NTP (National Toxicology Program), 1986. Toxicology and
carcinogenesis studies of benzene in F344/N rats and B6C3F1

mice (gavage studies). Research Triangle Park, NC, USA, US
Department of Health and Human Services. (Technical report
series No. 289).
NTP (National Toxicology Program), 1987. Teratologic evaluation
of diethylene glycol dimethyl ether (CAS No. 111-96-6)
administered to New Zealand White rabbits on gestation days
6 through 19. Research Triangle Park, NC, USA, National
Institute of Environmental Health Sciences, National
Toxicology Program (NTP-87-108; PB 87–209532).
NTP (National Toxicology Progam), 1991. Toxicology and
carcinogenesis studies of tris(2-chloroethyl)phosphate in
F344/N rats and B6C3F1 mice. Technical report series No. 391.
NTP TR 391, Research Triangle Park, NC, USA.
OECD/SIDS (Organization for Economic Development and
Cooperation/Screening Information DataSet), 1994. p-
Toluenesulfonamide. UNEP publications, Nairobi, Kenya.
OECD/SIDS (Organization for Economic Development and
Cooperation/Screening Information DataSet), 1998.
Triethylphosphate. UNEP publications, Nairobi, Kenya.
Peto, R., Gray, R., Brantom, P., Grasso, P., 1991a. Effects on 4080
rats of chronic ingestion of N-nitrosodiethylamine or Nnitrosodimethylamine:
a detailed dose–response study.
Cancer Res. 51, 6415–6451.
Peto, R., Gray, R., Brantom, P., Grasso, P., 1991b. Dose and time
relationships for tumor induction in the liver and esophagus
of 4080 inbred rats by chronic ingestion of Nnitrosodiethylamine
or N-nitrosodimethylamine. Cancer Res.
51, 6452–6469.
Proviron Fine Chemicals, 2003. IUCLID dataset on Nbutylbenzenesulphonamide.
REWAB Database, 2009. KWR Watercycle Research Institute,
Nieuwegein, The Netherlands.
Robinson, M., Bruner, R.H., Olson, G.R., 1990. Fourteen- and
ninety-day oral toxicity studies of methyl-t-butyl ether in
Sprague-Dawley rats. J. Am. Coll. Toxicol. 9, 525–540.
Schwarzenbach, R.P., Escher, B.I., Fenner, K., Hofstetter, T.B.,
Johnson, C.A., Von Gunten, U., Wehrli, B., 2006. The challenge
of micropollutants in aquatic systems. Science 313 (5790),
1072–1077.
SCOEL (Scientific Committee on Exposure Limits), 1991. No 11b.
Recommendations of the SCOEL for DMA.
SCOEL (Scientific Committee on Exposure Limits), 2002. No 91.
Recommendations of the SCOEL for DEA.
Schwab, B.W., Hayes, E.P., Fiori, J.M., Mastrocco, F.J., Roden, N.M.,
Cragin, D., Meyerhoff, R.D., D’Aco, V.J., Anderson, P.D., 2005.
Human pharmaceuticals in US surface waters: a human
health risk assessment. Regul. Toxicol. Pharmacol. 42 (3),
296–312.
Seacat, A.M., Thomford, P.J., Hansen, K.J., Olsen, G.W., Case, M.T.,
Butenhoff, J.L., 2002. Subchronic toxicity studies on
perfluorooctanesulfonate potassium salt in cynomolgus
monkeys. Toxicol. Sci. 68 (1), 249–264.
Sherman, H., 1972. Long-term feeding studies in rats and dogs
with 2-benzimadazole carbamic acid, methyl ester (INE-965)
(50% and 70% MBC wettable powder formulations).
Unpublished report from E.I. DuPont de Nemours and Co., Inc.,
Haskell Laboratory, Newark, Delaware, USA.
Singh, G., Driever, P.H., Sander, J.W., 2005. Cancer risk in people
with epilepsy: the role of antiepileptic drugs. Brain 128 (Pt 1),
7–17.
Skutlarek, D., Exner, M., Farber, H., 2006. Perfluorinated
surfactants in surface and drinking waters. Environ. Sci.
Pollut. Res. 13 (5), 299–307.
Snyder, S.A., Villeneuve, D.L., Snyder, E.M., Giesy, J.P., 2001.
Identification and quantification of estrogen receptor agonists
in wastewater effluents. Environ. Sci. Technol. 35 (18),
3620–3625.
Snyder, S.A., Trenholm, R.A., Snyder, E.M., Bruce, G.M.P.,
Hemming, J.D.C., 2008. Toxicological Relevance of EDCs and
water research 44 (2010) 461–476 475
Pharmaceuticals in Drinking Water. Awwa Research
Foundation, Denver, USA, p. 484.
Stokinger, H.E., Woodward, R.L., 1958. Toxicologic methods for
establishing drinking water standards. J. Am. Water Works
Ass. 50, 515–529.
Swarm, R.L., Roberts, G.K., Levy, A.C., Hines, L.R., 1973.
Observations on the thyroid gland in rats following the
administration of sulfamethoxazole and trimethoprim.
Toxicol. Appl. Pharm. 24 (3), 351–363.
Swartjes, F.A., Baars, A.J., Fleuren, R.H.L.J., Otte, P.F., 2004. Risk
limits for MTBE (methyl tertiair-butyl ether) in soil, sediment,
groundwater, surface water, drinking water and for drinking
water production (risicogrenzen voor MTBE (Methyl tertiair-
Butyl Ether) in bodem, sediment, grondwater,
oppervlaktewater, drinkwater en voor drinkwaterbereiding)
RIVM report 711701039/2004, Bilthoven, The Netherlands.
Toccalino, P.L., 2007. Development and application of healthbased
screening levels for use in water-quality assessments.
US Geological Survey Scientific Investigations Report 2007-
5106, 12 p, Reston, Virginia, USA.
URL1: http://www.epa.gov/waterscience/criteria/drinking/
dwstandards.pdf, accessed June 2009.
URL2: http://www.who.int/water_sanitation_health/dwq/
guidelines/en/, accessed June 2009.
URL3: http://www.ctb.agro.nl/ctb_files/11995_01.html, accessed
May 2009 (In Dutch).
URL4: http://www.iawr.org/index.php, publications, accessed
January 2009.
URL5: http://www.riwa-rijn.org/riwa_en.html, publications,
accessed January 2009.
US EPA (United States Environmental Protection Agency) IRIS
database (1988a). 1,4-dioxane.
US EPA (United States Environmental Protection Agency) IRIS
database, 1988b. Diethyl phthalate.
US EPA (United States Environmental Protection Agency) IRIS
database, 1988c. Diuron.
US EPA (United States Environmental Protection Agency) IRIS
database, 1989. Bis(chloroisopropyl)ether.
US EPA (United States Environmental Protection Agency), 2006.
National Primary Drinking Water Standards. http://www.epa.
gov/safewater/consumer/pdf/mcl.pdf (accessed August 2009).
Van Leerdam, J.A., Hogenboom, A.C., Kooi van der, M.E., de
Voogt, P., 2009. Determination of polar 1-H-benzotriazoles and
benzothialzoles in water by solid-phase extraction and liquid
chromatography LTQ Orbitrap mass spectrometry. Int. J. Mass
Spectrom. 282 (3), 99–107.
Van Leeuwen, F.X., 2000. Safe drinking water: the toxicologist’s
approach. Food Chem. Toxicol. 38 (1 Suppl), S51–S58.
Van Leeuwen, C.J., Vermeire, T., 2007. Risk Assessment of
Chemicals, second ed., Springer, ISBN 978-4020-6101-1.
Van Wezel, A.P., Puijker, L., Vink, C., Versteegh, A., de Voogt, P.,
2009. Odour and flavour thresholds of gasoline additives
(MTBE, ETBE and TAME) and their occurrence in Dutch
drinking water collection areas. Chemosphere 76, 672–676.
Versteegh, J., Van Der Aa, N., Dijkman, E., 2007. Pharmaceuticals
in drinking water and sources of drinking water - results of
monitoring campaign 2005/2006 (geneesmiddelen in
drinkwater en drinkwaterbronnen - resultaten van het
meetprogramma 2005/2006), 53, RIVM, Bilthoven, The
Netherlands. Number: R(T)-1-67.
Walraven, N., Laane, R.W., 2009. Assessing the discharge of
pharmaceuticals along the Dutch coast of the North sea. Rev.
Environ. Contam. Toxicol. 199, 1–18.
Westerhoff, P., Yoon, Y., Snyder, S., Wert, E., 2005. Fate of
endocrine-disruptor, pharmaceutical, and personal care
product chemicals during simulated drinking water treatment
processes. Environ. Sci. Technol. 37 (17), 6649–6663.
WHO (World Health Organization), 1996. Simazine in drinkingwater.
Background document for development of WHO

476
Guidelines for drinking-water quality. WHO/SDE/WSH/03.04/
42, Geneva, Switzerland.
WHO (World Health Organization), 1998. Guidelines for
drinking water quality. Second edition, volume 1
recommendations: addendum, ISBN 92 4 154514 3, Geneva,
Switzerland.
WHO (World Health Organization), 2002. Diethylene glycol
dimethyl ether. Concise International Chemical Assessment
Document 41, Geneva, Switzerland.
WHO (World Health Organization), 2003a. Benzene in drinking
water. WHO report WHO/SDE/WSH/03.04/24, Geneva,
Switzerland.
WHO (World Health Organization), 2003b. 2,4-D in drinking water.
WHO report WHO/SDE/WSH/03.04/70, Geneva, Switzerland.
WHO (World Health Organization), 2003c. Isoproturon in
Drinking-water. Background document for development of
WHO Guidelines for Drinking-water Quality WHO/SDE/WSH/
03.04/37, Geneva, Switzerland.
WHO (World Health Organization), 2005a. Glyphosate and AMPA
in drinking-water. Background document for development of
WHO Guidelines for Drinking-water Quality WHO/SDE/WSH/
03.04/97, Geneva, Switzerland.
WHO (World Health Organization), 2005b. Trichloroethene in
Drinking-water, Background document for development of
WHO Guidelines for Drinking-water Quality. WHO/SDE/WSH/
05.08/22, Geneva, Switzerland.
WHO (World Health Organization), 2006. Guidelines for drinking
water quality, First Adendum to third ed., Recommendations,
vol. 1, 2006, Geneva, Switzerland.
WHO (World Health Organization), 2008. N-
Nitrosodimethylamine in drinking-water. Background
water research 44 (2010) 461–476
document for preparation of WHO Guidelines for drinkingwater
quality. Geneva, World Health Organization (WHO/HSE/
AMR/08.03/8), Geneva, Switzerland.
WHO/JECFA (World Health Organization/joint FAO/WHO expert
committee on food additives), 1973. Food Additives Series 32;
Toxicological Evaluation of Certain Food additives and
Contaminants. Prepared by the forty-first meeting of the joint
FAO/WHO expert committee.
WHO/JECFA (World Health Organization/joint FAO/WHO expert
committee on food additives), 1974. Food Additives Series No.
5; Seventeenth Report of the Joint FAO/WHO Expert
Committee on Food Additives, World Health Organization
technical report series, 1974, No. 539; FAO Nutrition Meetings
Report Series, 1974, No. 53.
WHO/JECFA (World Health Organization/joint FAO/WHO expert
committee on food additives), 1995. 892. Carbendazim
(Pesticide residues in food: 1995 evaluations Part II
Toxicological & environmental).
WHO/JECFA (World Health Organization/joint FAO/WHO expert
committee on food additives), 2003. WHO Food Additives
Series No. 50; sulfur-containing heterocyclic compounds.
WHO/JMPR (World Health Organization/Joint Meeting on
Pesticide Residues), 2001. Toxicological evaluations
imidacloprid. Pesticide residues in food.
WHO/JMPR (World Health Organization/Joint Meeting on
Pesticide Residues), 2005. Dimethenamid-p. pp. 189–239.
Wilson, A.B., Thorpe, E., 1971. Toxicity studies on the ‘‘Prefix’’
residue 2,6-dichlorobenzamide: two year oral experiment with
dogs. Tunstall Laboratory. Report number: Lab Project
Number: T507531; Group research report: TLGR.0028.71, pp.
1–28.

Monitoring the biological activity of micropollutants during
advanced wastewater treatment with ozonation and activated
carbon filtration
M. Macova a , B.I. Escher a, *, J. Reungoat b , S. Carswell a,c , K. Lee Chue a ,
J. Keller b , J.F. Mueller a
a
University of Queensland, National Research Centre for Environmental Toxicology (EnTox), 39 Kessels Road, Brisbane, QLD 4108, Australia
b
University of Queensland, Advanced Water Management Centre (AWMC), Brisbane, QLD 4072, Australia
c
Queensland Health Forensic and Scientific Services, Organics Laboratory, 39 Kessels Road, Brisbane, QLD 4108, Australia
article info
Article history:
Received 13 May 2009
Received in revised form
28 August 2009
Accepted 7 September 2009
Available online 16 September 2009
Keywords:
Bioassays
Baseline toxicity
Phytotoxicity
Estrogenicity
Genotoxicity
Toxic equivalency concept
water research 44 (2010) 477–492
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
abstract
* Corresponding author. Tel.: þ61 (0) 7 3274 9180; fax: þ61 (0) 7 3274 9003.
E-mail address: b.escher@uq.edu.au (B.I. Escher).
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.watres.2009.09.025
A bioanalytical test battery was used to monitor the removal efficiency of organic micropollutants
during advanced wastewater treatment in the South Caboolture Water Reclamation
Plant, Queensland, Australia. This plant treats effluent from a conventional sewage treatment
plant for industrial water reuse. The aqueous samples were enriched using solid-phase
extraction to separate some organic micropollutants of interest from metals, nutrients and
matrix components. The bioassays were chosen to provide information on groups of chemicals
with a common mode of toxic action. Therefore they can be considered as sum indicators
to detect certain relevant groups of chemicals, not as the most ecologically or human health
relevant endpoints. The baseline toxicity was quantified with the bioluminescence inhibition
test using the marine bacterium Vibrio fischeri. The specific modes of toxic action that were
targeted with five additional bioassays included aspects of estrogenicity, dioxin-like activity,
genotoxicity, neurotoxicity, and phytotoxicity. While the accompanying publication discusses
the treatment steps in more detail by drawing from the results of chemical analysis as well as
the bioanalytical results, here we focus on the applicability and limitations of using bioassays
for the purpose of determining the treatment efficacy of advanced water treatment and for
water quality assessment in general. Results are reported in toxic equivalent concentrations
(TEQ), that is, the concentration of a reference compound required to elicit the same response
as the unknown and unidentified mixture of micropollutants actually present. TEQ proved to
be useful and easily communicable despite some limitations and uncertainties in their derivation
based on the mixture toxicity theory. The results obtained were reproducible, robust
and sensitive. The TEQ in the influent ranged in the same order of magnitude as typically seen
in effluents of conventional sewage treatment plants. In the initial steps of the treatment
chain, no significant degradation of micropollutants was observed, and the high levels of
dissolved organic carbon probably affected the outcome of the bioassays. The steps of coagulation/flocculation/dissolved
air flotation/sand filtration and ozonation decreased the effectbased
micropollutant burden significantly.
ª 2009 Elsevier Ltd. All rights reserved.

478
1. Introduction
Water quantity and water quality issues have led many
countries to explore water reuse and water reclamation as
alternatives to provide water for direct use in industrial
applications as well as agricultural and gardening use, and
for indirect use to supplement water reservoirs. Australia is
amongst the countries most affected by uneven distribution
of water resources and droughts. Therefore, various water
treatment and desalination plants have been put into
operation, among them the Water Reclamation Plant in
South Caboolture, Queensland, Australia (van Leeuwen
et al., 2003).
Besides hygiene issues, organic micropollutants are of
major concern because classical wastewater treatment with
biological activated sludge only partially removes hazardous
chemicals. Furthermore there is the potential that persistent
and toxic micropollutants and transformation products could
build up during the recycling process.
Urban and agricultural wastewaters contain a high
number of organic micropollutants, including pharmaceuticals
and personal care products, hormones and industrial
chemicals, biocides and pesticides (Schwarzenbach et al.,
2006). A large number of studies have investigated the fate
and removal of various individual and specific groups of
micropollutants by treatment processes like biological activated
sludge treatment (Ternes et al., 2004; Joss et al., 2005),
ozonation and advanced oxidation processes (von Gunten,
2003; Huber et al., 2004; Huber et al., 2005; Lee et al., 2008) or
activated carbon treatment (Sanchez-Polo et al., 2006; Snyder
et al., 2007) but little is known how these treatment processes
change the composition and biological activity of the
micropollutants.
Toxicity testing may provide complementary information
to chemical analysis on the sum of micropollutants present in
treated water. However, whole-effluent toxicity testing, which
is a valuable tool for hazard assessment of industrial and
municipal effluents (Chapman, 2000), is inappropriate to
evaluate the fate of micropollutants during advanced water
treatment because acute toxicity tests are not sensitive
enough for purified water and long-term chronic tests are too
expensive and time-consuming for routine monitoring. In
addition, the physical treatment of wastewater alters the
matrix and electrolyte composition and it becomes virtually
impossible to differentiate the effect caused by micropollutants
from that induced by the matrix, such as organic
matter content, pH, ionic strength or nutrient content.
Therefore, over the last years, we have developed
a framework for bioanalytical quantification of organic
micropollutants, which has three main cornerstones: Firstly,
the aqueous samples are enriched using solid-phase extraction
(SPE) to separate the organic micropollutants of interest
from metals and matrix components. This SPE method was
previously validated for recovery and extraction yield for
selected chemicals including pesticides and pharmaceuticals
(Bengtson Nash et al., 2005a; Escher et al., 2005; Escher et al.,
2008a). Secondly, a battery of bioanalytical methods were
selected that cover a non-specific bioassay, the bioluminescence
inhibition of the marine bacterium Vibrio fischeri and
water research 44 (2010) 477–492
five bioassays for specific modes of toxic action that are
indicative for groups of chemicals of particular relevance for
human and environmental health, including aspects of
estrogenicity, dioxin-like activity, genotoxicity, neurotoxicity,
and phytotoxicity (Muller et al., 2007; Muller et al., 2008;
Escher et al., 2008a; Escher et al., in press). Table 1 summarizes
the key features of the bioassays and more details are
given in the Supporting Information.
Finally,we developed a coherent data evaluation method that
is based on the toxic equivalency concept (Villeneuve et al., 2000;
Escher et al., 2008b). Thus, the effects at various steps of the
treatment chain can be easily compared and treatment efficiency
can be compared between the different effect endpoints.
Bioassays in combination with SPE have been successfully
applied by several groups to evaluate various water treatment
options (Escher et al., 2008a; Cao et al., 2009). In particular, the
bioluminescence inhibition test with Vibrio fischeri and the
Salmonella based test for genotoxicity have been widely used
despite their recognized limitations (Johnson, 2005; Petala
et al., 2006; Petala et al., 2008). The present work is more
extensive covering several additional toxicity endpoints,
allowing high-throughput monitoring applications while
remaining cost-efficient.
Here we present this framework for the first time to evaluate
all steps of enhanced water treatment in the South
Caboolture Water Reclamation Plant (van Leeuwen et al.,
2003). This paper focuses on the advantages and limitations in
the application of the bioanalytical test battery, and the
interpretation and significance of the results. In the accompanying
paper, the results obtained from the bioanalytical
tools are compared with chemical analytical data for a variety
of pharmaceuticals and pesticides and conclusions for
advanced wastewater treatment are drawn (Reungoat et al.,
2010).
2. Materials and methods
2.1. Samples and sites
Four consecutive sets of 24-hour composite samples were
collected on 11-07-08, 22-07-08, 27-07-08 and 06-08-08 using
refrigerated autosamplers at 10 sites of the South Caboolture
Water Reclamation Plant (Table 2).
2.2. Solid phase extraction
Immediately after sampling, concentrated HCl (36%) was
added to a final concentration of 5 mM for preservation. It was
demonstrated in earlier work that a pharmaceutical cocktail
in a wastewater matrix had highest recoveries for HLB at pH 3
(Escher et al., 2005).
Samples were extracted using 1 g OASIS Ò HLB solid phase
material in 20 mL cartridges (Waters, Australia) following
centrifugation (only S1 sample) and filtration with a glass fibre
filter. After conditioning the cartridges with 10 mL methanol
and 20 mL of 5 mM HCl in MilliQ water, a known volume of
sample was percolated under vacuum. The cartridges were
dried overnight under vacuum and were eluted with 10 mL

Table 1 – Bioanalytical test battery (adapted from Muller (2008)).
Assay Measured by Target mode of
toxic action
Bioluminescence
inhibition test
Acetylcholinesterase
(AChE)
Inhibition Assay
Imaging-PAM
Assay
methanol and 10 mL hexane:acetone (1:1). All eluates were
evaporated to dryness and reconstituted with 0.5 mL of ethanol.
2.3. Bioanalytical tools
The bioassays were performed as previously reported (Escher
et al., 2005; Muller et al., 2007; Muller et al., 2008; Escher et al.,
2008a; Escher et al., in press) using the methods cited in Table
1 and described in detail in the Supporting Information.
2.4. Relative enrichment factor of the samples REF
For each sample, the enrichment factor of the solid phase
extraction was calculated using (Eq. 1) which represents the
enrichment of the extract compared to the source water.
enrichment factorSPE ¼ Vwater
Reduction in luminescence of
the naturally bioluminescent
marine bacteria Vibrio fischeri
Product of the enzymatic
reaction (colorimetric)
Increased Chl a fluorescence,
inversely proportional to PS II
photosynthetic yield
E-SCREEN Cellular proliferation in
estrogenic dependent cell line
AhR-CAFLUX Induction of green
fluorescent protein under the
control of Ah receptor;
(fluorescence)
umuC assay Induction of ß-galactosidase
enzyme as an indicator of
DNA damage;
Product of the enzymatic
reaction (colorimetric)
Vextract
Sample volume varied from 2.0 to 2.5 L. The final volume of
each extract was 0.5 mL therefore the enrichment factor SPE
was between 3500 and 5500 depending on the sample.
An aliquot of the enriched sample extracts were then
added to the microtiter plate of the respective bioassay and
serially diluted by a test medium to obtain a concentrationeffect
curve. A dilution factor of each bioassay was calculated
using (Eq. 2). The dilution factors ranged from 0.05 to 0.007 for
the highest dose depending on the bioassay.
water research 44 (2010) 477–492 479
Effect on energy
status through
cytotoxicity and/or
baseline toxicity
Inhibition of the
enzyme, which
hydrolyses the
neurotransmitter
acetylcholine
PS II derived
photosynthesis
inhibition
(1)
Chemicals that
give response
Literature reference
All chemicals International Standard
Organisation, 1998;
Johnson, 2005;
Farré et al., 2006; Escher
et al., 2008a
Organophosphates
and carbamate
insecticides
Triazine and phenylurea
herbicides
Estrogenicity Estrogens,
estrogenic industrial
chemicals
Dioxin-like activity;
Aryl hydrocarbon
(Ah) receptor
activation
Genotoxicity;
DNA damage
(SOS-response)
Polychlorinated
dibenzodioxins/
furans (PCDD/F)
and biphenyls (PCB)
or polycyclic
aromatic
hydrocarbons (PAH)
Chlorinated
byproducts,
aromatic amines,
polycyclic aromatic
hydrocarbons (PAH)
Ellman et al., 1961;
Deutsche Norm, 1995;
Hamers et al., 2000
Schreiber et al., 2002;
Schreiber et al., 2007
Soto et al., 1995; Körner
et al., 1999
Nagy et al., 2002; Zhao
and Denison, 2004
Oda et al., 1985;
Reifferscheid et al., 1991;
International Standard
Organisation, 2000
volume of extract added to bioassay
dilution factorbioassay ¼
total volume of bioassay
(2)
The final relative enrichment factor REF is the combination
of the enrichment of the extraction and the dilution in the
bioassay (Eq. 3) (Escher et al., 2006; Muller et al., 2007) and
represents the enrichment or dilution of the original sample
in each bioassay (Table 3). The REF is equivalent to concentration
and is expressed in the units [Lwater sample/Lbioassay].
REF ¼ dilution factorbioassay enrichment factorSPE (3)
2.5. Evaluation of the concentration-effect curves
Due to the nature of the endpoints, the six bioanalytical tests
of the battery could not be evaluated with a single data evaluation
model (Table 3). In general, all concentration effect
curves of the reference compounds and sample extracts followed
a log-logistic function (Eq. 4), which was fitted using
Prism 5.0 software (GraphPad, San Diego, CA, USA). Adjustable
parameters are the minimum and maximum of the effect,
min and max, the slope s, and the effect concentration
inducing 50% of the maximum effect, EC50. If the concentration-effect
curve dropped sharply at the range of higher doses
indicating that the sample has a cytotoxic effect, the cytotoxic
concentrations were excluded from the concentration-effect
assessment.

480
Table 2 – Sample details. Samples were each taken after the described process. More details are given in Reungoat et al.
(2010). TOC/DOC data represent the average of two samplings (22-07-08 and 06-08-08).
Sample ID Site description TOC (mg/L) DOC (mg/L)
effect¼min
þ
max min
1þ10 s ðlog EC 50 log ðconcentrationof reference compoundor REF of sampleÞÞ
Three bioassays, the inhibition of bioluminescence of Vibrio
fischeri, inhibition of AChE and inhibition of PSII photosynthetic
yield, all produced effects between 0% and 100%, therefore the
maximum effect in (Eq. 4) could be set to 100% (Table 3).
Two bioassays, the AhR-CAFLUX and the E-SCREEN, yield
information on the induction of a receptor or proliferation of
cells and there is no absolute maximum effect. Dependent on
the mode of action, the maximum effect can be higher or
lower than the maximum effect of the reference compounds.
The extent of difference of maximally attainable effect yields
some additional information on the mode of action and the
type of compounds that induce the effect. Details of each
assay are discussed below. The curve fit is then performed in
such a way that the maximum effect in equation 4 is an
additional fitting parameter.
Finally the umuC assay exhibited a linear concentrationeffect
curve. At low effect levels the log-logistic concentration
effect curve (Eq. 4) is congruent with a linear concentrationeffect
curve. Unlike the other assays, where effects were
compared to a reference compound, genotoxicity is expressed
as the concentration that induces the threshold of induction
defined by the EN ISO guideline (International Standard
Organisation, 2000). This way of data evaluation needs further
refinement in the future because it is not directly comparable
to the results in the other bioassays and the effect concentration
cannot be translated into TEQ.
The detection limit of the bioassays was defined as three
times the standard deviation of the response using the lowest
concentration of the standard that induced an effect significantly
different from the control.
2.6. Estimating equivalent concentrations
in the samples
For an easy-to-follow reporting of the effect, in all but one
bioassay, the effects were reported in terms of toxic equivalent
(4)
concentrations (Villeneuve et al., 2000; Escher et al., 2008b). The
equivalent concentration represents the concentration of
the reference compound required to produce the same effect as
the mixture of different compounds in the sample.
Equivalent concentrations were calculated from concentration-effect
curves of the reference compound and the
samples. Both reference compounds and samples generally
followed a sigmoidal log–concentration-effect curve (Eq. 4),
which is equivalent to a linear function with respect to nonlogarithmic
concentration at small effect levels. Sample
extracts are normally composed of a mixture of unknown
substances at unknown concentrations, so the concentration
unit of the sample in the concentration effect curve is based
on the Relative Enrichment Factor (REF) in units of
[Lwater sample/Lbioassay].
Toxic equivalent concentrations (TEQ) were calculated as
the ratio of EC50 values of the reference compound to the EC50
of the sample (Eq. 5).
TEQ ¼ EC50ðreference compoundÞ
EC50ðsampleÞ
2 g
3
Lbioassay 4 ¼
g
5 ð5Þ
L water sample
L bioassay
Lwater sample
Since the EC50(reference compound) is expressed in concentration
units such as [g/Lbioassay] and the EC50(sample) in
concentration units [Lwater sample/Lbioassay], representing the
enrichment or dilution (REF) of the sample required to elicit the
50% effect, the toxic equivalent concentration is expressed in
[g/Lwater sample].
In principle, any effect level can be used to derive TEQ,
provided that the concentration-effect curves are parallel i.e.
have the same slope, which is in fact one of the prerequisites for
the validity of the toxic equivalency concept (Villeneuve et al.,
2000). Therefore, if the effect of the sample in the bioassay does
not reach 50%, toxic equivalency can be estimated using EC20 of
the sample and EC20 of the corresponding reference compound.
2.7. QA/QC
avg sd avg sd
Blank MilliQ water n.a. n.a. n.a. n.a.
S1 Influent (Effluent from WWTP) 20 4 17 4
S2 Denitrification 24 n.a. 21 n.a.
S3 Pre-Ozonation 21 7 21 4
S4 Coagulation/flocculation/DAFF 11 1 11 0.5
S5 Main ozonation 9.9 0.9 10 0.9
S6 Activated carbon filtration 6.6 0.6 6.6 0.9
S7 Effluent 7.1 0.5 6.8 0.7
C3 Biological sand filter fed with S4 water 8.5 1 8.8 0.8
C4 Biological activated carbon filter fed with S4 water 4.0 0.4 4.2 0.04
C5 Biological activated carbon filter fed with S5 water 3.6 0.07 4.1 0.08
avg - average; sd - standard deviation; n.a. - not applicable.
water research 44 (2010) 477–492
To determine background levels of contamination associated
with the extraction process as well as assessing any effect of

Table 3 – Evaluation of the concentration effect curves and expression of the bioanalytical results.
Assay Concentration-effect
assessment
Baseline Toxicity –
Bioluminescence
inhibition in
Vibrio
fischeri
Neurotoxicity –
AChE
Phytotoxicity –
Max-I-PAM
Estrogenicity –
E-SCREEN
Ah-Receptor –
AhR-CAFLUX
Genotoxicity –
Umu
Top: 100
Bottom: 0
Slope: 1
Top: w
Bottom: w
Slope: 1
Top: at least
100 and then w
Bottom: w
Slope: 1
the solvent, 2.0 L of MilliQ water were extracted identically to
the samples and analysed in all bioassays as a procedural
blank.
For quality control and assurance purposes two replicates
of randomly selected samples (2 sites per each sampling; for
details see Supporting information Table S1–S6) were
collected and analysed in all bioassays on a different day than
the first replicate to evaluate repeatability of the SPE and the
bioassays. Each sample extract was tested in the bioassays in
duplicates or triplicates per run, depending on the assay.
All individual data sets are summarised in the Supporting
information Table S1–S6. Since the variability among the four
sampling events was not higher than the variability between
the sample replicates collected at the same time, we report in
the main manuscript the average sd of four sample extracts
at four different consecutive sampling events (n ¼ 4, whith
exception of S2, where n ¼ 3). Analysis of variance (ANOVA)
was used to test the differences among the average TEQ of
samples S1-C5, including the blank, in comparison to S1.
3. Results and discussion
REF
range
3.1. Baseline toxicity – bioluminescence inhibition
in vibrio fischeri
Reference compound;
Positive control
0.4–38 Virtual baseline
toxicant;
Phenol a
The EC50 of phenol was 130 37 mg/L over 8 plates in 3 different
repetitions of the assay, confirming the reproducibility and
repeatability of the assay. While in all other assays, the reference
compound served two purposes, as quality control and to
define the reference for the TEQ, in the bioluminescence inhibition
test, phenol was only used for quality control.
EC 50 of the
reference
compound
12 mg/L;
130 mg/L
Result
expression
Baseline toxicity
equivalent
concentrations
(baseline-TEQ)
0.04–115 Parathion 120 mg/l Parathion equivalent
concentrations (PTEQ)
1.5–258 Diuron 16 mg/L Diuron equivalent
concentrations
(DEQ)
0.0006–55 17ß-Estradiol (E2) 2 ng/L Estradiol equivalent
concentrations (EEQ);
Relative proliferative
effect (RPE)
0.004–55 2,3,7,8-
Tetrachlorodibenzodioxin
(TCDD)
Linear regression 2.2–110 (-S9) 2-amino
anthracene a
(þS9) 4-nitroquinoline-
N-oxide a
a Only used as positive control, not as reference compound.
b Quantification limit; w variable; n.a. - not applicable.
water research 44 (2010) 477–492 481
8 ng/L TCDD equivalent
concentrations
(TCDDEQ)
n.a. REF that elicits genotoxic
effect with threshold
Induction Ratio of
1.5
Detection limit
20% inhibition
of bioluminescence
0.3 mg/L
0.01 mg/L
0.02 ng/L
0.04 ng/L
Maximum REF
achieved in the
assay b
Toxicity of the sample was expressed in baseline toxicity
equivalents (baseline-TEQ) derived from a baseline toxicity
QSAR (quantitative structure-activity relationship) (Fig. 1A)
using a virtual compound with log K ow of 3 and molecular
weight of 300 g/mol as a reference, which equates to an EC50 of
12 mg/L (Escher et al., 2008b). The rationale behind this choice
is the fact that there is not a singular positive control, as every
chemical exhibits baseline toxicity. For specifically acting
compounds the baseline effect is marginal and cannot be
resolved if the single chemical is tested. However, in a mixture
with a large number of chemicals with a variety of specific
modes of action, as is the case in a wastewater sample, the
baseline toxicity might actually dominate the overall mixture
effect (Warne and Hawker, 1995). Thus, baseline toxicity will
provide an integrative measure of the combination of chemicals
that act together in the mixture and each chemical’s
contribution is essentially weighted by its hydrophobicity only
(Escher and Schwarzenbach, 2002). Fig. 1A depicts the relationship
between hydrophobicity and effect concentration on
a log-log plot and how the EC50 of a virtual baseline toxicant is
derived (note that the hydrophobicity scale is based on liposome-water
partitioning not octanol-water partitioning),
therefore the virtual baseline toxicant with log K ow of 3 is
situated slightly higher than 3 on the log-hydrophobicity
scale.
Samples were tested for baseline toxicity at 8 different
concentrations after serial dilution 1:2, with the REF ranging
from 0.4 to 38. All previously tested baseline toxicants
exhibited a common slope of 1 (Escher et al., 2008b), and
therefore the slope of the samples was fixed to 1, too. Effect
concentrations (EC50) of the samples used to calculate the
toxic equivalent concentrations were expressed in REFs,

482
Fig. 1 – Concentration-effect curves of the reference compounds and selected samples tested in bioluminescence inhibition
assay (A, B), AChE inhibition assay (C, D) and I-PAM (E, F) assay. Bottom and top of each curve was fixed to 0 and 100,
respectively. Quality of the fit expressed as R 2 was >0.90, except for S7-effluent sample tested in AChE assay, where the R 2
was 0.74. (A) Baseline toxicity QSAR for the bioluminescence inhibition test with Vibrio fischeri and derivation of the EC50 of
the virtual baseline toxicant that is used as reference compound in this assay (data and QSAR from Escher et al. (2008b)).
Error bars indicate SD.
representing the enrichment or dilution of the sample
required to elicit the 50% effect, and are summarised in Table 4.
The EC 50 of the blank was 59, which means that if MilliQ
water were enriched 59 times, the blank extract would elicit
50% bioluminescence inhibition. This EC 50 was extrapolated
from experimental data, where the highest measured REF was
38, which caused a 25% effect. It is not reasonable to enrich the
samples to a higher REF because the enrichment of impurities
of solvents and materials or contamination during the
enrichment procedure would mask the measurable effect.
The influent, denitrification and ozonation samples all
exhibited EC50 in the range of 4–5.6, i.e. the samples had to be
water research 44 (2010) 477–492
enriched by approximately five times to see an effect. The
concentration window for a visible effect is relatively narrow.
A sample that is not enriched (REF ¼ 1) or even diluted would
not give any visible response. The necessary enrichment is not
the only reason for using SPE for sample preparation:
a majority of matrix compounds are removed by SPE, such as
salts and particulates. All organisms are very sensitive to
variable electrolyte concentration. In this assay too little salt
would decrease the bioluminescence of the marine bacterium
(Escher et al., 2008a), and salting up is difficult because the salt
content of water samples is unknown. In contrast to salts and
particulate matter, we assume that dissolved organic carbon

Table 4 – Sample enrichment required to produce a 50% effect in each assay expressed as the average ± sd of four 24 h
sampling events.
ID Site description Bioluminescence inhibition AChE I-PAM
(DOC), in particular the low molecular weight fraction, is
partially retained by SPE.
Baseline-TEQ increased slightly from influent (2.3 mg/L) to
denitrification (2.8 mg/L) and after pre–ozonation (3.2 mg/L).
However the stepwise increase was not statistically significant
(ANOVA, p ¼ 0.05). Significant decrease by more than
a factor of two in TEQ was observed after coagulation/flocculation/DAFF
(1.4 mg/L). Baseline-TEQ after the main ozonation,
activated carbon filtration and in the effluent ranged
water research 44 (2010) 477–492 483
EC50 (REF) EC50 (REF) EC50 (REF)
avg sd avg sd avg sd
Blank 59 0.81 BDL (

484
Fig. 2 – Relationship between DOC concentration and
baseline-TEQ. The black diamonds are the data from this
study (from Tables 1 and 5). The empty squares refer to
a previous study performed in Switzerland (all data refer to
secondary effluent) (Escher et al., in press). Error bars
indicate SD.
matter that is co-extracted with SPE. Assimilable organic carbon
may be taken up by microorganisms and contribute to bioluminescence
inhibition, while higher molecular-weight fractions
of organic matter may bind micropollutants, potentially
causing a decrease in their bioavailability and thus decrease the
response. Note also that organic micropollutants contribute to
DOC in DOC measurements. Therefore it is virtually impossible
to differentiate between assimilable organic carbon and organic
micropollutants.
The TEQ of the influent of the reclamation plant, which is the
effluent of a conventional wastewater treatment plant (WWTP),
was higher than of the secondary effluent of a WWTP in
Switzerland (Escher et al., 2008a; Escher et al., in press). It must
also be noted that the DOC levels were significantly lower in the
Swiss WWTP effluent than in the present study. In fact,
the baseline-TEQ of the Swiss study (Escher et al., in press)fallon
the lower left corner of Fig. 2, but they fit the correlation. This
observation supports the conclusion that DOC contributed to
the effects measured in this bioassay.
3.2. Neurotoxicity – acetylcholinesterase
inhibition assay
The concentration-effect curve of the reference compound,
parathion, followed a log-logistic function with the slope 1 and
the EC50 of 120 32 mg/L (Fig. 1C), using a commercially available
acetylcholinesterase enzyme from Electrophorus electricus.
The EC50 of parathion in our work was higher than data
reported in literature: 27 mg/L (Escher et al., in press) and 26 mg/L
(Escher et al., 2008a) with the use of bovine acetylcholinesterase.
The difference in the sensitivity of the assay might be
due to a different origin of acetylcholinesterase enzyme.
The samples were tested at 12 different concentrations
after serial dilution 1:2 with the REF range from 0.04 to 115.
The best-fit slope of the samples’ concentration-effect curve
was 1 (Fig. 1D) as it was for the reference compound. Effective
concentrations required to elicit 50% inhibition are summarised
in Table 4. Influent samples had to be enriched 52 times to
elicit 50% inhibition, in contrast to effluent, which would
water research 44 (2010) 477–492
require an enrichment of 1100 times to elicit the same effect.
As discussed above, we did not apply such high enrichment
factors but had to extrapolate the EC50. The effect of the blank
and samples treated with activated carbon was less than 20%,
defined as the detection limit of the assay. Therefore the
effective concentrations were not extrapolated from the
concentration-effect curves.
Results of the acetylcholinesterase inhibition assay were
expressed as parathion equivalent concentrations (PTEQ) and
are summarised in Table 5. PTEQ of the samples showed firstly
a slight increase after pre-ozonation to 4.2 mg/L in comparison
with the influent PTEQ of 3.1 mg/L, and then a significant
decrease to 1.9 mg/L after the main ozonation. Activated
carbon filtration further reduced PTEQ below the detection
limit (

measurement is somewhat compromised, when several nonspecifically
acting chemicals in the mixture mask the phytotoxic
effect (Escher et al., 2008a). In fact, when single baseline
toxicants (Escher et al., 2008b) were tested for the inhibition of
photosynthesis yield they showed a response when
measuring Y after 2 hours incubation, albeit at higher
concentrations than for the corresponding growth inhibition
endpoint, and the curve was steeper with a slope of 1.9. Since
it is impossible to quantitatively differentiate between the
contributions from phytotoxicity and cytotoxicity (unless
cytotoxicity is so overwhelming that it makes the photosynthesis
yield measurements impossible), we used the slope of
the reference compound diuron to fit the samples even
though the quality of fit was sub-optimal.
The phytotoxic response of the samples after 2 hours of
incubation expressed as diuron equivalent concentration DEQ
is summarised in Table 5. Denitrification significantly
increased the DEQ (0.29 mg/L) in comparison with the influent
DEQ of 0.11 mg/L. The pre-ozonation step itself had no significant
effect (0.34 mg/L). After coagulation/flocculation/DAFF,
the DEQ decreased by 85% to 0.05 mg/L. DEQ decreased in small
steps after the main ozonation and activated carbon filtration
stages and in the effluent samples. No significant additional
decrease in DEQ was observed after sand filter or biological
activated carbon filters. All DEQ after coagulation/flocculation/DAFF
are close to the detection limit and are based on
water research 44 (2010) 477–492 485
Fig. 4 – Concentration-effect curves of selected samples and corresponding standards tested in E–SCREEN (A, B) and AhR-
CAFLUX (C, D) assay. The slope of the curves was fixed to 1. Bottom of the curves was an adjustable parameter in both
assays. The top of the curve was adjustable in the E–SCREEN assay, while in AhR-CAFLUX, the top was set to the level of at
least 100% of the TCDD maximal effect. Quality of the fit expressed as R 2 ranged for selected samples and reference
compounds from 0.96 to 0.99, except for Blank and S7-effluent samples tested in E-SCREEN assay, where the R 2 was 0.3 and
0.4, respectively due to a very low response of both samples. Error bars indicate SD.
a single data point, therefore should be interpreted with some
caution.
The initial slight but not statistically significant increase in
TEQ from sample S1 to S3 is unexpected at first sight. It is
interesting to note, however, that this small increase was
common for baseline toxicity and slightly less clear also for
neurotoxicity (Figs. 2 and 3). The increase in baseline toxicity
was explained by an increase of DOC, which is presumably
partially co-extracted at low pH. The small increase in DEQ
could also be caused by baseline toxicants interfering with the
measurement of the photosynthesis yield. We have not
observed such an effect in our previous work (Escher et al.,
2008a; Escher et al., in press). However, we have never
encountered DOC concentrations exceeding 10 mg/L in any of
our previous work. It is unclear how much of the DOC is coextracted
with the micropollutants by SPE. The effect of DOC
needs to be investigated in future work but is beyond the
scope of the present study. Possibly, extraction at a higher pH
might reduce the fraction of co-extracted DOC due to the
higher negative charge of DOC under these conditions.
Previous work on comparing herbicides concentrations
with DEQ often showed a relatively good correlation and
a substantial fraction (> 50% of the response) could be
explained by the measured herbicides (Bengtson Nash et al.,
2005b; Bengtson Nash et al., 2006; Escher et al., 2006; Muller
et al., 2008). There were some cases, such as samples collected

486
Table 6 – Sample enrichment required to produce a defined effect in each assay: 50% effect in E–SCREEN assay and 20% of
maximal TCDD effect in AhR-CAFLUX assay. Results represent the average ± sd of 4 samplings.
ID Site description E-SCREEN AhR-CAFLUX
from the Thames River, where the biological DEQ were much
higher than the chemically analysed DEQ (Bengtson Nash
et al., 2006). The authors suggested this difference was due to
the presence of additional herbicides that were not analysed
with the analytical chemical methodology, but given the new
evidence it is also possible that additional cytotoxicity by nonspecifically
acting micropollutants and DOC could have
contributed to the discrepancy.
3.4. Estrogenic activity: E-SCREEN assay
In the E-SCREEN, bottom and top of the concentration-effect
curve (Eq. 4), both adjustable parameters, refer to the
minimum or maximum stimulatory response, i.e., proliferation
of the estrogen-dependent cells in E–SCREEN. The slopes
of the concentration-effect curves of the reference compound,
17ß-estradiol, and the samples were fixed to 1 (Figs. 4A and B).
The maximum of the curve yields information on the mode of
action, thus being able to distinguish among full agonist,
water research 44 (2010) 477–492
Before Clean-up After Clean-up
RPE Mode of Action EC50 (REF) EC20 TCDD (REF) EC20 TCDD (REF)
avg sd avg sd avg sd avg sd
Blank 0.05 – Not estrogenic BDL (

Table 7 – Sample enrichment required to produce
a genotoxic effect [ Induction Ratio of 1.5 (IR1.5) in umuC
assay without and with metabolic activation. Results
represent the average ± sd of 4 samplings.
ID Site description umuC -S9 umuC þ S9
EC IR1.5 (REF) EC IR1.5 (REF)
avg sd avg sd
Blank >78 >78
S1 Influent (Effluent from WWTP) 5.2 2 17 3
S2 Denitrification 6.9 2 22 3
S3 Pre-ozonation 6.3 2 18 3
S4 Coagulation/flocculation/DAFF 15 7 28 8
S5 Main ozonation 78 20 >87
S6 Activated carbon filtration >84 >84
S7 Effluent >69 >73
C3 Biological sand
filter fed with S4 water
24 13 34 16
C4 Biological activated
carbon filter
fed with S4 water
>80 >80
C5 Biological activated
carbon filter
fed with S5 water
>79 >79
(1667 times dilution) to 55 (55 times enrichment of the original
water sample). Sample REF required to produce a 50% effect
(EC50) in the E–SCREEN assay ranged from 0.27 (equivalent to 4
times dilution of the original water sample) to 6.7 (equivalent to
6.7 times enrichment of the original water sample) (Table 3). The
wide range of REF windows allows one to monitor a wide range
of samples collected throughout the treatment chain.
The estrogenicity of the samples was quantified with two
parameters: the relative proliferative effect (RPE) and the
estradiol equivalent concentration (EEQ). RPE of the samples,
representing the ratio of the maximum proliferation induced
by test samples compared to 17b-estradiol (also referred to as
relative efficacy), are summarised in Table 6. The samples from
the influent up to ozonation were fully or partially agonistic,
while the remainder of the samples were not classified estrogenic
and no EEQ could be derived for these samples.
When a sample was classified as either a partial or full
agonist, an estradiol equivalent concentration (EEQ) was
calculated from the EC 50. Influent to the reclamation plant
(representing the effluent of the WWTP) showed full agonistic
activity (RPE > 0.8) with the estradiol equivalent concentration
EEQ of 6 ng/L (Table 5). This was comparable with the EEQs
reported for WWTP effluent in the literature: 2 ng/L) and the resulting effluent of the ozonation contained
0.6 ng/L EEQ (Escher et al., in press).
Estrogenicity was further markedly altered after activated
carbon filtration. Estrogenic effect of the sample treated with
activated carbon (S6) was less than 20% of the estradiol effect
(RPE < 0.2), which corresponds to the detection limit of the
assay, therefore the EEQ was less than 0.02 ng/L.
Estrogenicity (RPE) of the effluent sample was slightly
increased, but did not reach 50% effect of estradiol, therefore
the sample was classified as not estrogenic and the EEQ was
not quantified. The estrogenic effect was below quantification
limit. Taking into account the REF of the sample in this assay,
the EEQ of the effluent was less than 0.06 ng/L.
Besides the main treatment chain, additional treatment
processes were assessed in this study: sand filtration of the
coagulation/flocculation/DAFF effluent (C3) and biological
activated carbon filtration of the coagulation/flocculation/
DAFF effluent (C4) and main ozonation effluent (C5). Sand
filtration significantly decreased the EEQ of the coagulation/
flocculation/DAFF effluent from 9.8 ng/L to 0.63 ng/L. Biological
activated carbon filtration decreased the estrogenic effect
of both effluents to below detection limit (

488
reproducibility and repeatability of the assay and is in good
agreement with the previous levels from the literature of
5.9 ng/L (Nagy et al., 2002) and 3.9 ng/L (Zhao and Denison,
2004).
TCDD was chosen as the reference compound as it is one of
the most potent agonists for this assay. However, high affinity
ligands for the AhR also include dibenzofurans and biphenyls as
well as a variety of polycyclic aromatic hydrocarbons, benzoflavones
and other chemicals (Denison and Nagy, 2003). To
assess the contribution of chemicals other than dioxins, furans
and dioxin-like PCBs to the potency estimates, sulfuric acid silica
gel clean-up was performed in this study, which removed most
organic chemicals (e.g. PAHs) except persistent chemicals such
as dioxins, furans and PCBs. Original sample extracts, as well as
the extracts cleaned with sulfuric acid silica gel were tested with
the AhR-CAFLUX assay at 5 different concentrations after serial
dilution 1:10, with the REF ranging from 55 to 0.004. The response
of many samples, especially after clean-up, did not reach 50%
TCDD response. Therefore effective concentrations of the
samples required to elicit 20% effect of TCDD (EC 20 TCDD) were
interpolated from concentration-effect curves (Table 6). The EC 20
TCDD of the samples ranged from 1.4 to 8.2. The effective
concentration of the same samples was significantly altered
after clean up and ranged from 14 to 22, i.e. samples after acid
silica gel clean up had to be enriched approximately 3–10 times
more to elicit the same effect.
The response in the AhR-CAFLUX assay was expressed as the
2,3,7,8-TCDD-equivalent concentration (TCDDEQ). TCDDEQ was
significantly decreased by the main ozonation to 0.44 ng/L in
comparison with the TCDDEQ of 0.83 ng/L in the influent. Sand
filtration did not change the TCDDEQ of the coagulation/flocculation/DAFF
effluent, while treatment with the biological activated
carbon filters reduced the TCDD equivalent concentration
to 0.33 ng/L a level not significantly different from the blank.
The TCDDEQ of the samples after the sulfuric acid silica gel
clean-up were reduced to levels not significantly different
from the blank (Table 5). These results indicate that the
observed effect in the AhR-CAFLUX assay prior to the sulfuric
acid silica gel clean-up step may be attributed to chemicals
other than dioxins, furans and PCBs.
3.6. Genotoxicity-UmuC assay
The umuC genotoxicity assay can detect both cytotoxic and
genotoxic effects. The extracts were tested both with (þS9) and
without ( S9) exogenous metabolic activation. Response on
this assay is determined as an induction ratio (IR) (Fig. 5). An
IR 1.5 is considered genotoxic, providing the sample is not
cytotoxic (growth < 0.5). For samples that demonstrate significant
genotoxic response, the effect concentrations EC IR 1.5
were derived from the linear concentration-effect curve as
depicted in Fig. 5. Effect concentrations are in dimensionless
units of REF and represent how many times the samples must
be concentrated or diluted to elicit a threshold IR of 1.5 in the
assay (Table 7). Results are expressed as 1/ECIR 1.5 therefore
a higher number represents a higher genotoxic effect (Table 5).
Unfortunately no TEQ could be derived for this endpoint but in
theory this should be possible and in the future we will work on
implementing the TEQ concept for this assay. This is somehow
water research 44 (2010) 477–492
problematic because of the frequently observed cytotoxicity of
the samples, which can only be partially accounted for.
Overall, the umuC test without metabolic activation gave
higher responses than the test with metabolic activation
indicating that some of the micropollutants that are responsible
for genotoxicity are detoxified by the liver enzyme fraction
S9. As in all other bioassays, the denitrification and preozonation
did not affect the activity in the umuC test but
coagulation/flocculation/DAFF decreased the genotoxicity
without S9 by 58% and that with S9 by 35%. Subsequent main
ozonation completely removed the effect for the test with prior
S9 treatment and drastically reduced it in the absence of S9.
4. Conclusion
A similar test battery has been previously used for various
applications in wastewater treatment efficiency and water
quality assessment. To apply the test battery to a treatment
train with very subtle steps and for relatively pure water was
a challenge. This challenge was met by the following measures:
A comprehensive framework was developed that centres
around the bioassays but links to sample preparation with
SPE prior to testing and to the data evaluation scheme.
Quality control was implemented and all effects were
related to reference compounds, thereby correcting for
slight day-to-day variability of the test results.
The use of SPE allows a relatively non-specific extraction of
water samples to concentrate a wide range of relevant
pollutants to levels where they can exhibit a measurable
response, while matrix components such as salts and
metals are removed from the sample during SPE.
The detection window of the bioassays is adjustable. For the
application in the present study the concentration range of
the sample in the bioassay ranged between 1700 times dilution
and 250 times enrichment of the sample. The detection
window depends on the constitution of the sample and the
type of bioassay and is only limited by the effect of the blank
at very high enrichment and nonspecific cytotoxicity of the
sample for each of the assays. By dynamically changing
the relative enrichment window of the sample throughout
the study one can adapt the various bioassays to a wide
variety of samples and results are robust and reproducible.
Results are reported as toxic equivalent concentrations,
a format that allows for comparison to results from chemical
analysis. This is also a key feature of the bioanalytical toolsthey
are no direct indicators for environmental health but
they are indicators of the presence of mixtures of chemicals,
accounting for the mixture toxicity of the unidentified
chemicals in an environmental sample. To report bioanalytical
results as toxic equivalent concentrations is better
suited for hazard assessment and extrapolation than the
commonly reported % effect of the water sample in a given
bioassay because it is too difficult to extrapolate effects while
it is possible to compare effect concentrations of chemicals
across different level of biological organisation. So for
example, if a sample elicits 0.5 mg/L DEQ one can compare
this value with the environmental quality standard for

diuron to estimate what potential long-term effects could be
expected from such an exposure.
The study also highlighted some difficulties that remain and
some lessons are to be learned. One of the weak points of the
approach is the sample preparation with SPE. However, this is
a common problem for chemical analysis and can only be circumvented
if there are recovery standards for every single
compound to be analysed. Such an approach is not possible for
bioassays because spiked chemicals would add to the mixture
toxicity. A shortcoming of the SPE method that was noticed here
for the first time is the co-extraction of an unknown fraction of
dissolved organic matter, an issue that requires further attention
in the future.
In addition, volatile compounds will evaporate during SPE
but they would also cross contaminate in a 96 well plate, so it
is a procedural advantage to have removed the volatile
chemicals prior to toxicity testing. It must also be noted that
oxidation by-products formed during ozonation are often
more hydrophilic than their parent compound and might
therefore not be extracted efficiently by SPE.
In addition, one must be aware that mixture effects are
assessed when applying bioanalytical tools to complex samples
and there is always the problem that the subtle effect of newly
formed by-products might be shielded by other micropollutants
present in the mixture. Some of the oxidation byproducts
stemming from ozonation are very reactive, for example aldehydes
and other electrophiles formed from ß-blockers during
ozonation (Benner and Ternes, 2009), others are very hydrophilic
organics such as NDMA (N-nitrosodimethylamine) or
inorganics such as bromate (Hollender et al., in press) noneof
these would be caught with the bioanalytical tools used here but
require targeted chemical analysis for quantification. However,
ozonation often only mildly oxidizes organic micropollutants
and many resulting transformation products are likely to retain
some of their toxic potential (Escher et al., 2008c) or lose their
specific receptor-binding effect but retain non-specific baseline
toxicity (Lee et al., 2008) and will thus contribute to the mixture
toxicity in the bioanalytical tools applied in the present study.
While the results are discussed in more detail and
compared to chemical analysis in the accompanying paper
(Reungoat et al., 2010), some general conclusions can be
drawn:
The influent of the water reclamation plant had similar
levels of effects for the different endpoints as we have
encountered in previous studies for wastewater after
secondary treatment (Escher et al., 2008a; Escher et al., in
press).
The effluent of the water reclamation plant had strongly
reduced effect levels compared to the influent, reducing the
baseline-TEQ by 79%, the DEQ by 76%, the estrogenicity to
below the detection limit of 0.02 ng/L EEQ, the PTEQ by 88%
and the genotoxicity to below the detection limit. Only the
AhR-CAFLUX assay did not yield any information on the
removal efficiency of the treatment because the level of
persistent inducers of the arylhydrocarbon receptor
remained close to the detection limit. This was not too
unexpected given that these are typically very hydrophobic
compounds, which are usually adsorbed and not present in
water research 44 (2010) 477–492 489
the water column. In contrast, the non-persistent inducers
of the AhR were clearly reduced by a factor of two during
ozonation.
The most efficient steps with respect to removal of the toxic
responses in all selected assays proved to be the coagulation/flocculation/DAFF,
the main ozonation and (biological)
activated carbon filtration steps.
This paper demonstrates the first comprehensive application
of the bioanalytical tool framework on an advanced water
treatment plant. One future goal would be to automate the
assays to a degree that they can be used routinely by
commercial laboratories or water companies themselves. At
this stage, there remain some scientific questions to be
resolved, including (1) the sample treatment and dosing, (2)
extension of the battery and (3) data interpretation. (1) SPE and
dosing by solvent spiking will always change the composition
of an undefined mixture. A combination of passive sampling to
a biomimetic polymer and passive dosing directly from the
polymer into the bioassay might be a way to overcome this
limitation. (2) A battery of bioassays will never be comprehensive
because of the myriads of receptors and regulatory
pathways in an organism. Nevertheless the battery would
benefit from inclusion of additional endpoints, such as chemically
induced immunosuppression or even other examples of
mode-of-action categories already included, e.g. other endocrine
endpoints. (3) At this stage we can say how much of
a toxic equivalent concentrations is removed by a certain
treatment step and we can give comparative results but we will
never be able to deduce ecological or health consequences
from that. This discussion must be lead independently. One
solution would be to compare the TEQ of the mixture with the
defined Environmental Water Quality Criteria (EQS) for single
substances of the Water Framework Directive (European
Commission, 2006) or guideline values for drinking water (e.g.
NHMRC–NRMMC, 2004) but not for all reference compounds
are such EQS and guideline values available.
Acknowledgements
This work was co-funded by the Urban Water Security Research
Alliance under the Enhanced Treatment Project, the CRC Water
Quality and Treatment Project No. 2.0.2.4.1.1. Dissolved Organic
Carbon Removal by Biological Treatment and by EnTox. The
National Research Centre for Environmental Toxicology (EnTox)
is a joint venture of the University of Queensland and Queensland
Health Forensic and Scientific Services (QHFSS).
The authors acknowledge the Moreton Bay Water for access to
the South Caboolture Water Reclamation Plant; Ray McSweeny
and Paul McDonnell (Moreton Bay Water) for their help during
sampling and Chris Pipe-Martin (Ecowise) for providing information
on the South Caboolture Water Reclamation Plant.
The authors are grateful to Renee Muller (Gold Coast
Water), Fred Leusch (Griffith University/EnTox), Karen
Kennedy (EnTox) and Pam Quayle (EnTox) for constructive
discussions and for help with the data analysis method, Marita
Goodwin (EnTox) for running various assays and for helpful
comments Christina Carswell (QHFSS) for extracting the

490
samples and Chris Paxman (EnTox) for technical assistance.
We would like to thank also Michael Denison (University of
California Davis, USA) for providing the H4G1.1c2 cells, Georg
Reifferscheid (German Federal Institute of Hydrology,
Germany) for providing the bacteria Salmonella typhimurium
TA1535/pSK1002, and Ana Soto (Tufts University, USA) for
providing the MCF7-BOS cells.
Appendix. Supporting information
Supporting information related to this article can be found at
doi:10.1016/j.watres.2009.09.025.
references
Bengtson Nash, S.M., Schreiber, U., Ralph, P.J., Mueller, J.F., 2005a.
The combined SPE: ToxY-PAM phytotoxicity assay;
application and appraisal of a novel biomonitoring tool for the
aquatic environment. Biosensors and Bioelectronics 20 (7),
1443–1451.
Bengtson Nash, S.M., McMahon, K., Eaglesham, G., Mueller, J.F.,
2005b. Application of a novel phytotoxicity assay for the
detection of herbicides in Hervey Bay and the Great Sandy
Straits. Marine Pollution Bulletin 51, 351–360.
Bengtson Nash, S.M., Goddard, J., Muller, J.F., 2006. Phytotoxicity
of surface waters of the Thames and Brisbane river estuaries:
a combined chemical analysis and bioassay approach for the
comparison of two systems. Biosensors and Bioelectronics 21,
2086–2093.
Benner, J., Ternes, T.A., 2009. Ozonation of propranolol: formation
of oxidation products. Environmental Science & Technology
43, 5086–5093.
Cao, N., Yang, M., Zhang, Y., Hu, J., Ike, M., Hirotsuji, J., Matsui, H.,
Inoue, D., Sei, K., 2009. Evaluation of wastewater reclamation
technologies based on in vitro and in vivo bioassays.
Ecotoxicology and Environmental Safety 407 (5), 1588–1597.
Chapman, P.M., 2000. Whole effluent toxicity testing – usefulness,
level of protection, and risk assessment. Environmental
Toxicology and Chemistry 19 (1), 3–13.
Denison, M.S., Nagy, S.R., 2003. Activation of the aryl hydrocarbon
receptor by structurally diverse exogenous and endogenous
chemicals. Annual Reviews in Pharmacology and Toxicology
43, 309–334.
Deutsche Norm, 1995. In: DIN, N.W.N.i. (Ed.), Suborganismische
Verfahren: Teil 1: Bestimmung von Cholinesterase-
Hemmenden Organophosphat- und Carbamat-Pestiziden
(Cholinesterase-Hemmtest) DIN 38415-1: 1995-02, Berlin,
Germany.
Ellman, G.L., Courtney, K.D., Andres, V., Featherstone, R.M., 1961. A
new and rapid colorimetric determination of acetylcholinesterase
activity. Biochemical Pharmacology 7, 88–95.
Escher, B., Schwarzenbach, R.P., 2002. Mechanistic studies on
baseline toxicity and uncoupling as a basis for modeling
internal lethal concentrations in aquatic organisms. Aquatic
Sciences 64, 20–35.
Escher, B.I., Bramaz, N., Maurer, M., Richter, M., Sutter, D., von
Känel, C., Zschokke, M., 2005. Screening test battery for
pharmaceuticals in urine and wastewater. Environmental
Toxicology and Chemistry 24 (3), 750–758.
Escher, B.I., Quayle, P., Muller, R., Schreiber, U., Mueller, J., 2006.
Passive sampling of herbicides combined with effect analysis
in algae using a novel high-throughput phytotoxicity assay
water research 44 (2010) 477–492
(Maxi-Imaging- PAM). Journal Environmental Monitoring 8,
456–464.
Escher, B.I., Bramaz, N., Quayle, P., Rutishauser, S., Vermeirssen, E.,
2008a. Monitoring of the ecotoxicological hazard potential by
polar organic micropollutants in sewage treatment plants and
surface waters using a mode-of-action based test battery.
Journal Environmental Monitoring 10, 622–631.
Escher, B.I., Bramaz, N., Mueller, J.F., Quayle, P., Rutishauser, S.,
Vermeirssen, E., 2008b. Toxic equivalent concentrations
(TEQs) for baseline toxicity and specific modes of action as
a tool to improve evaluation of ecotoxicity tests on
environmental samples. Journal Environmental Monitoring
10, 612–621.
Escher, B.I., Baumgartner, R., Lienert, J., Fenner, K., 2008c.
Predicting the ecotoxicological effects of transformation
products. SpringerLink online. In: Boxall, A. (Ed.), Handbook of
Environmental Chemistry – Degradation of Synthetic
Chemicals in the Environment. Springer, Heidelberg,
Germany. doi:10.1007/6998_2_015. http://www.springerlink.
com/content/48r2t364q183q654/?p ¼ ec460152bfd
14906adc89378463e4997&pi ¼ 5 published Wednesday,
April 30, 2008.
Escher, B.I., Bramaz, N. and Ort, C., 2009. Monitoring the
treatment efficiency of a full scale ozonation on a
sewage treatment plant with a mode-of-action based test
battery. Journal Environmental Monitoring, doi: 10.1039/
b907093a.
European Commission, 2006. Proposal for a Directive of the
European Parliament and the Council on Environmental
Quality Standards in the Field of Water Policy and Amending
Directive 2000/60/EC COM(2006) 398 Final. COM(2006) 398
Final. European Parliament, Committee on the Environment,
Public Health and Food Safety, Brussels.
Farré, M., Martinez, E., Hernando, M., Fernandez-Alba, A., Fritz, J.I.,
Unruh, E., Mihail, O., Sakkas, V., Morbey, A., Albanis, T.A.,
Brito, F., Hansen, P.D., Barcelo, D., 2006. European ring
exercise on water toxicity using different bioluminescence
inhibition tests based on Vibrio fischeri, in support to the
implementation of the water framework directive. Talanta,
323–333.
GWRC, 2008. In: Leusch, F.D.L. (Ed.), Tools to Detect Estrogenic
Activity in Environmental Waters. Global Water Research
Coalition, London, UK, p. 84.
Hamers, T., Molin, K.R.J., Koeman, J.H., Murk, A.J., 2000. A smallvolume
bioassay for quantification of the esterase inhibiting
potency of mixtures of organophosphates and carbamate
insceticides in rainwater: development and optimization.
Toxicological Sciences 58, 60–67.
Hollender, J., Zimmermann, S.G., Koepke, S., Krauss, M.,
McArdell, C.S., Ort, C., Singer, H., von Gunten, U., Siegrist, H.,
2009. Elimination of organic micropollutants in a municipal
wastewater treatment plant upgraded with a full-scale postozonation
followed by sand filtration. Environmental Science
& Technology, Articles ASAP (Article), in press. doi: 10.1021/
es9014629.
Huber, M.M., Ternes, T.A., von Gunten, U., 2004. Removal of
estrogenic activity and formation of oxidation products during
ozonation of 17 alpha-ethinylestradiol. Environmental
Science & Technology 38, 5177–5186.
Huber, M.M., Göbel, A., Joss, A., Hermann, N., Löffler, D.,
McArdell, C.S., Ried, A., Siegrist, H., Ternes, T.A., von
Gunten, U., 2005. Oxidation of pharmaceuticals during
ozonation of municipal wastewater effluents: a pilot study.
Environmental Science & Technology 39, 4290–4299.
International Standard Organisation 1998 Water Quality–
determination of the inhibitory effect of water samples on the
light emission of Vibrio fischeri (luminescent bacteria test), EN
ISO 11348-3, Geneva, Switzerland.

International Standard Organisation 2000 Water quality –
Determination of the genotoxicity of water and waste water
using the umu-test, EN ISO 13829(2000) and 38415-3 (1996),
Geneva, Switzerland.
Johnson, B.T., 2005. Microtox acute toxicity test. In: Blaise, C.,
Ferard, J.-F. (Eds.), Small-Scale Freshwater Toxicity
Investigations. Toxicity Test Methods, vol. 1. Springer.
Joss, A., Keller, E., Alder, A.C., Gobel, A., McArdell, C.S., Ternes, T.,
Siegrist, H., 2005. Removal of pharmaceuticals and fragrances
in biological wastewater treatment. Water Research 39 (14),
3139–3152.
Körner, W., Hanf, V., Schuller, W., Kempter, C., Metzger, J.,
Hagenmaier, H., 1999. Development of a sensitive E-screen
assay for quantitative analysis of estrogenic activity in
municipal sewage plant effluents. Ecotoxicology and
Environmental Safety 225 (1–2), 33–48.
Körner, W., Bolz, U., Sussmuth, W., Hiller, G., Schuller, W.,
Hanf, V., Hagenmaier, H., 2000. Input/output balance of
estrogenic active compounds in a major municipal sewage
plant in Germany. Chemosphere 40 (9–11), 1131–1142.
Körner, W., Spengler, P., Bolz, U., Schuller, W., Hanf, V.,
Metzger, J.W., 2001. Substances with estrogenic activity in
effluents of sewage treatment plants in southwestern
Germany. 2. Biological analysis. Environmental Toxicology
and Chemistry 20 (10), 2142–2151.
Lee, Y., Escher, B.I., von Gunten, U., 2008. Efficient removal of
estrogenic activity during oxidative treatment of waters
containing steroid estrogens. Environmental Science &
Technology 42 (17), 6333–6339.
Legler, J., Zeinstra, L.M., Schuitemaker, F., Lanser, P.H., Bogerd, J.,
Brouwer, A., Vethaak, A.D., De Voogt, P., Murk, A.J., Van der
Burg, B., 2002. Comparison of in vivo and in vitro reporter gene
assays for short-term screening of estrogenic activity.
Environmental Science & Technology 36 (20), 4410–4415.
Leusch, F.D., Chapman, H.F., van den Heuvel, M.R., Tan, B.L.,
Gooneratne, S.R., Tremblay, L.A., 2006. Bioassay-derived
androgenic and estrogenic activity in municipal sewage in
Australia and New Zealand. Ecotoxicology and Environmental
Safety 65 (3), 403–411.
Muller, R., Tang, J., Thier, R., Mueller, J., 2007. Combining passive
sampling and toxicity testing for evaluation of mixtures of
polar organic chemicals in sewage treatment plant effluents.
Journal Environmental Monitoring 9, 104–109.
Muller, R. 2008 Passive sampling and bioanalysis of organic
chemicals in surface water. Ph.D. thesis. University of
Queensland, St. Lucia, Qld, Australia, 201pp. Accessible at
http://library.uq.edu.au/search~S7?/XRenee þ
Muller&searchscope ¼ 7&SORT ¼ AZ/XRenee þ
Muller&searchscope ¼ 7&SORT ¼ AZ&extended ¼ 0&SUBKEY
¼ Renee%20Muller/21%22C23%22C23%22CB/frameset&FF ¼
XRenee þ Muller&searchscope ¼ 27&SORT ¼
AZ&21%22C21%22C
Muller, R., Schreiber, U., Escher, B.I., Quayle, P., Bengtson Nash, S.,
Mueller, J., 2008. Rapid exposure assessment of PSII herbicides
in surface water using a novel chlorophyll a fluorescence
imaging assay. Science Total Environment 401, 51–59.
Murk, A.J., Legler, J., van Lipzig, M.M.H., Meerman, J.H.N.,
Belfroid, A.C., Spenkelink, A., van der Burg, B., Rijs, G.B.J.,
Vethaak, D., 2002. Detection of estrogenic potency in
wastewater and surface water with three in vitro bioassays.
Environmental Toxicology and Chemistry 21 (1), 16–23.
Nagy, S., Sanborn, J., Hammock, B., Denison, M., 2002.
Development of a green fluorescent protein-based cell
bioassay for the rapid and inexpensive detection and
characterization of Ah receptor agonists. Toxicological
Sciences 65, 200–210.
NHMRC–NRMMC, 2004. Australian Drinking Water Guidelines
(ADWG). National Health and Medical Research Council
water research 44 (2010) 477–492 491
(NHMRC) in Collaboration With the Natural Resource
Management Ministerial Council (NRMMC), Commonwealth
of Australia. www.nhmrc.gov.au/publications/synopses/_
files/adwg_11_06.pdf.
Oda, J., Nakamura, S.I., Oki, I., Kato, T., Shinagawa, H., 1985.
Evaluation of the new system (umu-test) for the detection of
environmental mutagens and carcinogens. Mutatation
Research 147, 219–229.
Petala, M., Samaras, P., Zouboulis, A., Kungolos, A.,
Salkellaropoulos, G., 2006. Ecotoxicological properties of
wastewater treated using tertiary methods. Environmental
Toxicology 21 (4), 417–424.
Petala, M., Samaras, P., Zouboulis, A., Kungolos, A.,
Sakellaropoulos, G.P., 2008. Influence of ozonation on the in
vitro mutagenic and toxic potential of secondary effluents.
Water Research 42 (20), 4929–4940.
Reifferscheid, G., Heil, J., Oda, Y., Zahn, R.K., 1991. A microplate
version of the SOS/umu-test for rapid detection of genotoxins
and genotoxic potentials of environmental samples. Mutation
Research 253 (3), 215–222.
Reungoat, J., Macova, M., Escher, B.I., Carswell, S., Mueller, J.F.,
Keller, J., 2010. Removal of micropollutants and reduction of
biological adverse effects in a full scale reclamation plant
using ozonation and activated carbon filtration. Water
Research 44 (2), 625–637.
Rutishauser, B.V., Pesonen, M., Escher, B.I., Ackermann, G.E.,
Aerni, H.-R., Suter, M.J.-F., Eggen, R.I.L., 2004. Comparative
analysis of estrogenic activity in sewage treatment plant
effluents involving three in vitro assays and chemical analysis of
steroids. Environmental Toxicology and Chemistry 23, 857–864.
Sanchez-Polo, M., Salhi, E., Rivera-Utrilla, J., von Gunten, U., 2006.
Combination of ozone with activated carbon as an alternative
to conventional advanced oxidation processes. Ozone-Science
& Engineering 28 (4), 237–245.
Schreiber, U., Mueller, J.F., Haugg, A., Gademann, R., 2002. New
type of dual-channel PAM chlorophyll fluorometer for highly
sensitive water toxicity biotests. Photosynthesis Research 74,
317–330.
Schreiber, U., Quayle, P., Schmidt, S., Escher, B.I., Mueller, J., 2007.
Methodology and evaluation of a highly sensitive algae
toxicity test based on multiwell chlorophyll fluorescence
imaging. Biosensors and Bioelectronics 22, 2554–2563.
Schwarzenbach, R.P., Escher, B.I., Fenner, K., Hofstetter, T.B.,
Johnson, C.A., von Gunten, U., Wehrli, B., 2006. The challenge
of micropollutants in aquatic systems. Science 313 (5790),
1072–1077.
Snyder, S.A., Adham, S., Redding, A.M., Cannon, F.S., DeCarolis, J.,
Oppenheimer, J., Wert, E.C., Yoon, Y., 2007. Role of
membranes and activated carbon in the removal of endocrine
disruptors and pharmaceuticals. Desalination 202, 156–181.
Soto, A.M., Sonnenschein, C., Chung, K.L., Fernandez, M.F., Olea, N.
, Serrano, F.O., 1995. The E-SCREEN assay as a tool to identify
estrogens: an update on estrogenic environmental pollutants.
Environmental Health Perspectives 103 (Suppl. 7), 113–122.
Tan, B.L.L., Hawker, D.W., Muller, J.F., Leusch, F.D.L., Tremblay, L.A.,
Chapman, H.F., 2007. Comprehensive study of endocrine
disrupting compounds using grab and passive sampling at
selected wastewater treatment plants in South East Queensland,
Australia. Environment International 33 (5), 654–669.
Ternes, T.A., Joss, A., Siegrist, H., 2004. Scrutinizing pharmaceuticals
and personal care products in wastewater treatment.
Environmental Science & Technology 38 (17), 393A–399A.
van Leeuwen, J., Pipe-Martin, C., Lehmann, R.M., 2003. Water
reclamation at South Caboolture, Queensland, Australia.
Ozone: Science & Engineering 25, 107–120.
Villeneuve, D.L., Blankenship, A.L., Giesy, J.P., 2000. Derivation and
application of relative potency estimates based on in-vitro
bioassays. Environmental Toxicology Chemistry 19, 2835–2843.

492
von Gunten, U., 2003. Ozonation of drinking water: part I.
Oxidation kinetics and product formation. Water Research 37
(7), 1443–1467.
Warne, M.S.J., Hawker, D.W., 1995. The number of components
in a mixture determines whether synergistic and
antagonistic or additive toxicity predominate – the funnel
water research 44 (2010) 477–492
hypothesis. Ecotoxicology and Environmental Safety 31 (1),
23–28.
Zhao, B., Denison, M., 2004. Dvelopment and characterization of
a green fluorescence protein-based rat cell bioassay system
for detection of Ah receptor ligands. Organohalogen
Compounds 66, 3332–3337.

An evaluation of a pilot-scale nonthermal plasma advanced
oxidation process for trace organic compound degradation
Daniel Gerrity*, Benjamin D. Stanford, Rebecca A. Trenholm, Shane A. Snyder
Applied Research and Development Center, Southern Nevada Water Authority, River Mountain Water Treatment Facility, P.O. Box 99954,
Las Vegas, NV 89193-9954, USA
article info
Article history:
Received 22 May 2009
Received in revised form
3 September 2009
Accepted 10 September 2009
Available online 17 September 2009
Keywords:
Advanced oxidation process (AOP)
Nonthermal plasma (NTP)
Trace organic compound
Pharmaceutical
Endocrine disrupting compound
(EDC)
1. Introduction
abstract
Wastewater-derived contaminants, including pharmaceuticals
and personal care products (PPCPs), endocrine disrupting
compounds (EDCs), and other trace organic compounds, have
been a significant aspect of environmental research efforts for
decades (Snyder et al., 2003). However, adverse impacts on
aquatic ecosystems and increased public awareness of trace
organic compounds in water supplies have stimulated recent
interest in this issue (Snyder et al., 2003). Due to rapid population
growth and increasingly stressed water supplies,
particularly in semi-arid regions, many cities are turning to
alternative sources of drinking water such as indirect potable
reuse. In some systems, wastewater effluent is having
a greater impact on reservoir water quality due to drought and
overdraft (Benotti et al., 2009a). Over time, these trends may
water research 44 (2010) 493–504
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
* Corresponding author. Tel.: þ1 (702) 856 3666; fax: þ1 (702) 856 3647.
E-mail address: Daniel.Gerrity@snwa.com (D. Gerrity).
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.watres.2009.09.029
This study evaluated a pilot-scale nonthermal plasma (NTP) advanced oxidation process
(AOP) for the degradation of trace organic compounds such as pharmaceuticals and
potential endocrine disrupting compounds (EDCs). The degradation of seven indicator
compounds was monitored in tertiary-treated wastewater and spiked surface water to
evaluate the effects of differing water qualities on process efficiency. The tests were also
conducted in batch and single-pass modes to examine contaminant degradation rates and
the remediation capabilities of the technology, respectively. Values for electrical energy per
order (EEO) of magnitude degradation ranged from

494
ozone, granular activated carbon (GAC), nanofiltration/reverse
osmosis, and advanced oxidation) can achieve significant
removal or transformation (Ternes et al., 2002; Westerhoff
et al., 2005; Snyder et al., 2007). Advanced oxidation processes
(AOPs) are often used to achieve further reductions in trace
organics, target recalcitrant compounds, and comply with
certain water reuse regulations (e.g., California Department of
Public Health Title 22 requirements for NDMA and 1,4dioxane).
The most common AOPs consist of UV/peroxide and
ozone/peroxide, but emerging technologies such as
nonthermal plasma (NTP) provide viable alternatives.
This study focuses on a novel pilot-scale NTP reactor
developed by Aquapure Technologies, Ltd. (Upper Galilee,
Israel). The NTP uses high-voltage electrical pulses across
fiber-like electrodes to form a corona discharge, which in turn
generates UV light, ozone, and hydroxyl radicals (Locke et al.,
2006). The light generated by the corona discharge spans
wavelengths of approximately 250 nm to 1,000 nm with
multiple peaks at wavelengths associated with radical
formation (Locke et al., 2006). Although the light is lowintensity
(Locke et al., 2006), it has been credited with
a capacity for NDMA destruction (Even-Ezra et al., 2009),
which is generally inefficient in ozone-based oxidation
processes (Mitch et al., 2003).
NTP is considered to be highly efficient because little energy
is lost in heating the surrounding fluid, which allows the
energy to be focused on the excitation of electrons (Pekarek,
2003). In contrast to the electron beam technology, which
introduces electrons from an external source, the NTP used in
this study employs a corona discharge that excites electrons in
the ambient air directly above the target water matrix
(Pekarek, 2003). Other NTP configurations are also available,
including those that create a corona discharge in aerosols or
within the target water matrix (Pekarek, 2003; Locke et al.,
2006). The level of treatment can be adjusted by changing the
frequency (e.g., from 500 to 1,000 Hz) or voltage (up to 40 kV) of
the electrical pulses. The ionizing capability of the electrical
pulses generates singlet oxygen atoms, which in turn form
ozone and hydroxyl radicals. One of the most significant
benefits of NTP is the fact that oxidants can be generated
without the addition of costly chemicals or UV lamps, which
require cleaning and are hindered by high turbidity and matrix
absorbance. However, the NTP technology has limited
commercial availability and has not been tested in full-scale
drinking water and wastewater treatment applications so its
long-term reliability, efficiency, and effectiveness are still
unknown. In addition, the NTP technology is currently limited
to low-flow applications due to its thin-film configuration.
Although plasma-based technologies have been studied
extensively for the degradation of volatile organic compounds
(VOCs) (Masuda et al., 1995; Hwang and Jo, 2005; Ighigeanu
et al., 2005; Thevenet et al., 2007; Kim et al., 2008), many of
these studies have focused on air quality and/or bench-scale
configurations. Due to limitations associated with conventional
AOPsdprimarily the need for chemical additiond
studies of novel AOPs for water and wastewater treatment
applications are gaining popularity. A recent pilot-scale study
utilizing the Aquapure technology evaluated the efficacy of
NTP in degrading trichloroethylene (TCE), N-nitrosodimethylamine
(NDMA), 1,4-dioxane, and methyl tert-butyl
water research 44 (2010) 493–504
ether (MTBE) in batch experiments and single-pass remediation
(Even-Ezra et al., 2009). The current study uses the same
pilot-scale system to evaluate the degradation of PPCPs and
EDCs in water and wastewater. Despite the lack of PPCP/EDC
data for plasma-based technologies, other AOPs, particularly
UV/peroxide (Huber et al., 2003; Pereira et al., 2007; Snyder
et al., 2007; Benotti et al., 2009b), ozone/peroxide (Acero and
von Gunten, 2001; Snyder et al., 2007), and photocatalysis
(Benotti et al., 2009b; Westerhoff et al., 2009), have been tested
extensively with respect to trace organic degradation. In their
review of NTP applications, Locke et al. (2006) discussed the
need for comparisons of NTP with more common AOPs to
determine whether there is a net benefit associated with
employing NTP technologies. Among other research needs,
Locke et al. (2006) specifically identified energy consumption
and applicability to different waste matrices as useful points
of comparison. To this end, the objective of this study is to
expand the NTP knowledge base by evaluating its efficacy and
energy efficiency in degrading seven PPCPs and EDCs in
tertiary-treated wastewater and spiked surface water.
2. Experimental section
2.1. Selection of wastewater-derived contaminants
In order to narrow the scope of this research, a subset of the
numerous compounds detected in previous occurrence
studies was selected for evaluation. The indicator compounds
were selected based on their magnitude and frequency of
occurrence in water and wastewater (Snyder et al., 2007),
varying physical/chemical characteristics and resulting
susceptibility to treatment (Ternes et al., 2002; Westerhoff
et al., 2005; Snyder et al., 2007), and ease of analytical methods
(Trenholm et al., 2009). The seven compounds targeted in this
study include meprobamate (anti-anxiety), dilantin (also
known as phenytoin; anticonvulsant), primidone (anticonvulsant),
carbamazepine (anticonvulsant), atenolol (betablocker),
trimethoprim (antibiotic), and atrazine (herbicide
and suspected EDC). Their structures and characteristics are
described in Snyder et al. (2007). Although these compounds
have generated considerable interest in the research, treatment,
and regulatory arenas, only atrazine is currently regulated
by the United States (U.S.) Environmental Protection
Agency (EPA) at a maximum contaminant level (MCL) of 3 mg/L.
2.2. Nonthermal plasma pilot unit
The pilot-scale unit depicted in Fig. 1 is an ‘‘electrode-to-plate’’
NTP prototype (Locke et al., 2006) developed by Aquapure
Technologies, Ltd.; the pilot skid contains two reactors connected
in series. Water flows in a thin film (z5 mm) along the
stainless steel ground electrode (anode) where it is exposed to
high-voltage electrical pulses. The electrical pulses from the
generator have frequencies ranging from 500 to 1,000 Hz,
maximum voltage of 8.0 kV, maximum current of 100 A,
maximum energy of 1 J, and rise time of approximately 18 ns
(Locke et al., 2006; Even-Ezra et al., 2009). Depending on the
applied voltage and frequency, each reactor draws between 0.4
and 1.0 kW, and the external pumping and controls draw

approximately 2.8 kW. The electrical pulses generate a corona
discharge that stretches approximately 1 cm from the carbon
fiber-like cathode to the water surface. The distance between
the carbon electrode and water surface can be adjusted at
multiple points along the reactor to level the system and
optimize the process for a particular water matrix. The 1-cm
distance was selected to achieve a consistent corona discharge
and minimize sparking within the reactor, which would be
destructive to the carbon fibers. Despite the harsh conditions
experienced by the gas-phase electrode, no visible wear was
detected on the carbon fibers over the duration of the experiments.
With continuous, long-term operation of the NTP
reactor, it is likely that the carbon fibers would eventually
require replacement, but this life span is currently unknown.
After traveling through the first reactor, the water is
collected in a storage tank before being pumped into the
ozone injector system or the second reactor. The ozone
injector system combines a portion of the treated water
from the first reactor with ozone-rich air (z2 g/m 3 ) from
the reactor headspace using a Venturi inductor. For the
single-pass configuration, this mixture is collected in
another storage tank for additional ozone contact time
before being mixed and discharged with effluent from the
first reactor. Excess air from the ozone contactor is then
drawn out of the system by vacuum and quenched in an
ozone destruct unit. The hydraulic residence times in the
reactor, storage tank, and ozone contactor are approximately
0.5 min, 4–7.5 min, and 1.5 min, respectively. The
hydraulic residence times in the reactor and storage tank
depend on the process flow rate, whereas the residence
time in the ozone contactor remains relatively constant
because the injector flow rate (z40 L/min) is controlled by
water research 44 (2010) 493–504 495
Fig. 1 – Schematic of NTP pilot unit. The bottom figure illustrates a profile view of the electrodes within the reactor.
a separate pump. The process is then repeated in the
second reactor before being pumped out of the pilot skid.
For the batch configuration, the effluent pump is replaced
by a recirculation pump that allows the water to cycle
continuously through a single reactor and ozone contactor.
Additional details related to the NTP reactor used in this
study are provided in Even-Ezra et al. (2009).
2.3. Experimental design
The study was divided into three experiments: (1) a batch
experiment with 150 L of tertiary-treated wastewater (i.e.,
grit removal, primary settling, activated sludge, secondary
settling, and dual media filtration) at a recirculation rate of
8.0 L/min, (2) a single-pass experiment with tertiary-treated
wastewater at a flow rate of 15.5 L/min, and (3) a single-pass
experiment with spiked surface water at a flow rate of
11.4 L/min. For experiment (3), spiked surface water (Colorado
River water from Lake Mead, NV) was pumped into the
reactor from a 3,000-gallon polyethylene water tank
(American Tank Company, Windsor, CA). The spiking solution
was prepared by dissolving neat standards in water in
order to avoid the introduction of radical scavenging
solvents such as methanol. The spiking solution was added
to the 3,000-gallon tank followed by recirculation for 24 h to
facilitate homogenization. The two water types (wastewater
and surface water) were selected to provide an assessment
of NTP treatment efficiency for matrices with different
organic loadings. The initial pH, total alkalinity, total organic
carbon (TOC), UV254 absorbance, and trace organic
compound concentrations for the three experiments are
provided in Table 1. In all tables, ‘‘Batch’’ refers to the batch

496
Table 1 – Initial water quality conditions.
Parameter Batch (1) WW (2) a,c
SSW (3) b,c
pH 7.0 7.0 8.0
Total Alkalinity 126 mg/L as 126 mg/L as 137 mg/L as
CaCO3 CaCO3 CaCO3 TOC 5.9 mg-C/L 5.6 mg-C/L 2.6 mg-C/L
UV254 Abs. 0.119 cm 1
N/A 0.036 cm 1
Meprobamate 275 ng/L 256 25 e ng/L 933 15 ng/L
Dilantin 119 ng/L 165 10 ng/L 1,005 74 ng/L
Primidone 124 ng/L 146 6 ng/L 1,375 96 ng/L
Carbamazepine 176 ng/L 219 9 ng/L 1,600 115 ng/L
Atenolol 378 ng/L 413 8 ng/L 1,018 57 ng/L
Trimethoprim 36 ng/L 39 2 ng/L 1,200 82 ng/L
Atrazine d
N/A N/A 1,118 159 ng/L
a WW (2) ¼ Single-pass wastewater experiment.
b SSW (3) ¼ Single-pass spiked surface water experiment.
c Four influent samples were taken throughout the experiment to
calculate a representative ambient concentration and evaluate
temporal variability in target compound concentrations.
d Atrazine was spiked in the SSW experiment but was not detected
in the wastewater.
e Errors represent 1 standard deviation.
wastewater experiment, ‘‘WW’’ refers to the single-pass
wastewater experiment, and ‘‘SSW’’ refers to the single-pass
spiked surface water experiment.
The generator settings for the NTP pilot differed for each of
the experiments, as described in Table 2. For the batch
experiment (1), a single reactor/generator was operated at
a frequency of 500 Hz and a voltage of 8.0 kV. To assess the
degradation rate, ten samples were collected at generator
(AOP-specific) energy consumption values ranging from 0 to
7.3 kWh/m 3 . For the single-pass wastewater experiment (2),
four different generator scenarios provided treatment levels
ranging from 0.7 to 1.8 kWh/m 3 for the generators and 3.8 to
4.8 kWh/m 3 for the overall pilot skid. Samples were collected
from the effluent line exiting the pilot skid. Residual oxidants
in each sample were quenched with calcium thiosulfate.
During the wastewater experiments, the dissolved ozone
concentration in the NTP reactor effluent ranged from
approximately 0.05 mg/L to 0.10 mg/L depending on the
generator settings. The single-pass surface water experiment
(3) incorporated nine different generator scenarios ranging
from 0.6 to 2.6 kWh/m 3 for the generators and 4.8 to 6.8 kWh/
m 3 for the pilot skid. For the surface water, samples were
taken from Reactor 1, Ozone Tank 1, Reactor 2, and the pilot
skid effluent to evaluate the extent of contaminant degradation
in each system component. During the surface water
experiment, the dissolved ozone concentrations in the reactor
effluent ranged from 0.25 to 0.50 mg/L depending on the
generator settings.
2.4. Analytical methods
Ozone gas concentrations in the reactor headspace were
determined by online measurements with an OMC20 ozone
monitor and IN-2000-L2-LC ozone analyzer (IN USA, Norwood,
MA). Dissolved ozone residuals were measured with the
indigo trisulfonate method (Yates and Stenstrom, 2000). TOC
and UV 254 absorbance were analyzed according to standard
methods 5310B and 5910, respectively. Excitation-emission
matrices (EEM) were also developed for the single-pass
wastewater samples using a QuantaMaster UV-Vis QM4
Steady State Spectrofluorometer (Photon Technology International,
Inc., Birmingham, NJ).
Trace organic compounds were extracted and analyzed
using on-line solid phase extraction and liquid chromatography
with tandem mass spectrometry (SPE-LC-MS/MS)
according to Trenholm et al. (2009). All standards and reagents
were of the highest purity commercially available. All
Table 2 – Experimental test conditions.
Experiment Generator 1 Generator 2 Total generator
energy
(1) Batch a
Frequency
(Hz)
Voltage
(kV)
water research 44 (2010) 493–504
Energy
(kWh/m 3 )
Frequency
(Hz)
Voltage
(kV)
Energy
(kWh/m 3 )
Total skid
energy
(kWh/m 3 ) (kWh/m 3 )
Off Off 0.00 500 8.0 N/A N/A N/A
(2) WW a – 1 Off Off 0.00 500 8.0 0.71 0.71 3.75
WW – 2 800 8.0 1.04 Off Off 0.00 1.04 4.08
WW – 3 600 8.0 0.87 500 8.0 0.71 1.58 4.62
WW – 4 800 8.0 1.04 500 8.0 0.71 1.75 4.79
(3) SSW b – 1 Off Off 0.00 600 5.9 0.63 0.63 4.78
SSW – 2 Off Off 0.00 800 5.9 0.85 0.85 5.01
SSW – 3 500 8.0 0.97 Off Off 0.00 0.97 5.12
SSW – 4 Off Off 0.00 1000 5.9 1.07 1.07 5.23
SSW – 5 1000 8.0 1.53 Off Off 0.00 1.53 5.68
SSW – 6 500 8.0 0.97 600 5.9 0.63 1.60 5.75
SSW – 7 500 8.0 0.97 1000 5.9 1.07 2.04 6.19
SSW – 8 1000 8.0 1.53 600 5.9 0.63 2.16 6.31
SSW – 9 1000 8.0 1.53 1000 5.9 1.07 2.60 6.75
a The batch and WWTP experiments were conducted with ambient concentrations of trace organics in tertiary-treated wastewater.
b The SSW experiments were conducted with raw surface water (Lake Mead) spiked with a suite of trace organics.

pharmaceuticals were obtained from Sigma-Aldrich (St. Louis,
MO, USA). Meprobamate-d3 and trimethoprim-d9 were
obtained from Toronto Research Chemicals (Ontario, Canada).
Phenytoin-d10 and atenolol-d7 were obtained from C/D/N
Isotopes (Pointe-Claire, Canada). Carbamazepine-d10 and primidone-d5
were obtained from Cambridge Isotope Laboratories
(Andover, MA, USA). All solvents were trace analysis grade
from Burdick and Jackson (Muskegon, MI). Reagent water was
obtained using a Milli-Q Ultrapure Water Purification System
(Millipore, Bedford, MA, USA). All concentrated stocks were
prepared in methanol and stored at 20 C, while mixed
spiking solutions were prepared in reagent water and stored
at 4 C.
On-line SPE was performed using a SymbiosisÔ Pharma
(Spark Holland, Emmen, the Netherlands) automated SPE in
XLC mode. Briefly, a 10-mL volume of sample was measured
in a volumetric flask at which time isotopically-labeled
standards were spiked at 100 ng/L. This provided sufficient
sample volume for duplicates, matrix spikes, and dilutions, if
necessary. A 1.5-mL aliquot of each sample was then
transferred to a 2-mL autosampler vial, although only 1.0 mL
was used for extractions. Extractions were performed using
Waters Oasis HLB Prospekt cartridges (30 mm, 2.5 mg,
10 1 mm, 96 tray) (Milford, MA). Prior to sample loading,
each cartridge was sequentially conditioned with 1 mL of
dichloromethane (DCM), MTBE, methanol, and reagent water
(Milli-Q). Samples were loaded onto the SPE cartridge at
1 mL/min after which the cartridge was washed with 1 mL of
reagent water. After sample loading, the analytes were
eluted from the SPE cartridge to the LC column with 200 mL
methanol, using the LC peak focusing mode. A 5-mM
ammonium acetate solution and methanol gradient was
used for LC mobile phases with a flow rate of 800 mL/min.
Analytes were separated using a 150 4.6 mm Luna C18(2)
with a 5-mm particle size (Phenomenex, Torrance, CA).
Method reporting limits (MRLs) were chosen at 3 to 5 times
the method detection limit (MDL). MRLs were 10 ng/L for all
compounds, except for atenolol, which was 25 ng/L. All
analytes were quantified using isotope dilution. Stringent
QA/QC protocols (i.e., matrix spikes, duplicate samples, field
blanks, and laboratory blanks) were followed throughout the
duration of the experiment. Using the on-line SPE method,
the concentrations of duplicate samples rarely varied by
greater than 5%.
3. Results and discussion
3.1. Experiment 1 – batch experiment with tertiarytreated
wastewater
Contaminant degradation profiles during the batch wastewater
experiment are provided in Fig. 2. The initial ambient
concentrations for the target contaminants ranged from
36 ng/L for trimethoprim to as high as 378 ng/L for atenolol
(Table 1). Although included in the analytical method, atrazine
was below the method reporting limit (MRL) for both
wastewater experiments. As indicated in Table 2, the generator
was operated at a frequency of 500 Hz, a voltage of 8.0 kV,
and a recirculation rate of 8.0 L/min.
water research 44 (2010) 493–504 497
C/C 0
1.0
0.8
0.6
0.4
0.2
0.0
0 1 2 3 4 5 6 7 8
Generator Energy Consumption (kWh/m 3 )
Meprobamate Dilantin Primidone Carbamazepine Atenolol Trimethoprim
Fig. 2 – Contaminant degradation profiles from the batch
experiment (1) with ambient concentrations of trace
organic compounds in tertiary-treated wastewater. This
graph illustrates the relationship between contaminant
degradation and generator (AOP-specific) energy
consumption. The dashed lines represent samples that
were below the MRL for those compounds.
The degradation of each detected contaminant was
assumed to be pseudo first order based on a plot of ln(C/C 0)
versus generator energy consumption. Degradation rates, 95%
confidence intervals, and R 2 values are provided in Table 3.
Due to the low initial trimethoprim concentration, the extent
of degradation was limited, as indicated by the dashed line
representing the MRL in Fig. 2. Despite the limited degradation,
it is apparent that trimethoprim and carbamazepine are
highly susceptible to NTP treatment. This result is consistent
with previous studies assessing oxidation strategies for trace
organic compounds. Westerhoff et al. (2005) reported nearly
100% degradation of trimethoprim and carbamazepine with
standard chlorination and ozonation practices. Dilantin and
atenolol degraded more slowly during the NTP treatment,
which is also consistent with oxidation by chlorine or ozone
(Westerhoff et al., 2005). Finally, meprobamate has been
shown to be highly resistant to chlorination and ozonation in
comparison to other trace organic compounds (Westerhoff
et al., 2005). This conclusion is supported by the NTP results in
that meprobamate required nearly double the energy to
Table 3 – Pseudo first order rate constants for batch
experiment (1) in tertiary-treated wastewater.
Compound k (m 3 /kWh) 95% CI b
R 2
Meprobamate 0.36 0.03 0.98 9
Dilantin 0.52 0.15 0.92 5
Primidone 0.40 0.15 0.88 5
Carbamazepine 1.03 0.35 0.93 4
Atenolol 0.61 0.22 0.88 5
Trimethoprim 0.71 N/A 0.92 2
Atrazine a
N/A N/A N/A N/A
a Atrazine concentrations were below MRL in all samples.
b 95% CI ¼ 95% confidence interval.
N

498
achieve a specified level of degradation. Primidone was not
quite as resistant as meprobamate, but its degradation was
slower and more variable than the other compounds.
With respect to bulk organic parameters, there was no
significant change in TOC over the duration of the experiment.
The consistent TOC level was expected considering that
mineralization is highly energy intensive in hydroxyl radicaldominated
processes (Gerrity et al., 2009), and high CT values
are required in ozone-dominated processes (Rosal et al., 2009).
The limited energy consumption, low dissolved ozone
concentrations, and limited ozone contact time were likely
insufficient to induce measurable organic mineralization in
the NTP reactor.
Although there was no reduction in TOC, UV254 absorbance
was reduced by more than 65%; this change was correlated to
the degradation of target compounds in Fig. 3 as a potential
surrogate measure of process efficacy. Quantifying trace
organic compounds requires specialized equipment and
training, and the analytical methods are very expensive and
time-consuming. For this reason, it is generally impractical for
water and wastewater utilities to employ trace contaminant
monitoring programs for their systems. Implementation of
this type of surrogate framework (i.e., reduction in UV 254
absorbance in lieu of trace contaminant degradation) would
allow utilities to assess the performance of certain treatment
processes for emerging contaminants. For oxidation
processes, UV254 absorbance is an easily measured parameter
that appears to show consistent correlations with the degradation
of trace organic compounds (Wert et al., 2009). Fig. 3
provides support for this surrogate framework because it
demonstrates strong correlations between the degradation of
the target compounds and reduction in UV 254 absorbance. The
slopes of the regression lines indicate that relative contaminant
degradation (i.e., percent degradation) is actually more
rapiddmore than twice as fast for nearly all compoundsdthan
the reduction in UV254 absorbance. The slopes
also show a general agreement with the degradation profiles
% Degradation of Trace Organic Compounds
100%
80%
60%
40%
20%
water research 44 (2010) 493–504
in Fig. 2. Although trimethoprim was expected to degrade
rapidly, its slope of 6.25 is likely an outlier and artifact of its
small sample size. With the exception of trimethoprim, the
magnitude of the slopes and the negative intercepts are
similar to those reported in the Wert et al. (2009) ozonation
study. The negative intercepts indicate that there are initial
reductions in UV254 absorbance that precede degradation of
the target compounds. This may be attributable to initial
oxidant/radical scavenging by the bulk organic matter.
Values for electrical energy per order (EEO) of magnitude
degradation (Bolton et al., 1996), which is often measured in
kWh/m 3 -log, are also provided in Table 4. Since the batch
experiment involved a continuous treatment, the EEO values
were calculated based on first order degradation rates and
linear regression. In other words, a linear regression was
generated for ln(C/C0) against generator energy consumption,
and the EEO was calculated as ln(0.1) divided by the slope of
the linear regression. The EEO values ranged from 2.2 kWh/m 3
for carbamazepine to 6.4 kWh/m 3 for meprobamate. It is
important to note that the batch EEO values may be overestimates
because the batch system requires approximately
10–15 min to achieve its maximum ozone concentration after
starting the generator.
3.2. Experiment 2 – single-pass experiment with
tertiary-treated wastewater
Contaminant degradation profiles for the single-pass wastewater
experiment are illustrated in Fig. 4. The initial ambient
concentrations for the target contaminants were similar to
the batch experiment and ranged from 39 ng/L for trimethoprim
to as high as 413 ng/L for atenolol (Table 1). As indicated
in Table 2, this experiment involved a single-reactor
configuration and both reactors operating in series at a flow
rate of 15.5 L/min. All scenarios for this experiment involved
voltages of 8.0 kV. The reactor consumed 0.7 kWh/m 3 with
Generator 2 operating at a frequency of 500 Hz during the
Trimethoprim 6.25 -90% 1.00 2
0%
0% 20% 40% 60% 80% 100%
% Reduction in UV 254 Absorbance
Linear Regression Models:
Contaminant Slope Intercept R 2 N
Meprobamate 1.56 -14% 0.97 8
Dilantin 2.09 -22% 0.99 4
Primidone 2.08 -28% 0.98 4
Carbamazepine 2.75 -30% 1.00 3
Atenolol 2.68 -44% 0.99 4
Fig. 3 – Correlation between the degradation of target compounds and UV 254 absorbance during the batch experiment (1).
This analysis excludes samples that were below the MRL and the final five carbamazepine samples (>3 kWh/m 3 ; see Fig. 2)
due to high variability near the MRL. The slopes were all significant at the 0.03 level or better, and intercepts were all
significant at the 0.07 level or better.

Table 4 – AOP-specific EEO values (kWh/m 3 -log).
Contaminant (1) (2) (3) UV c
Batch
SSW
a
Min.
WW b
Med.
WW b
Max.
WW b
Min.
SSW b
Med.
SSW b
Max.
SSW b
WW-1 scenario, and the reactor consumed 1.0 kWh/m 3 with
Generator 1 operating at a frequency of 800 Hz during the
WW-2 scenario. WW-3 and WW-4 involved both reactors
operating in series at 600/500 Hz (1.6 kWh/m 3 ) and 800/500 Hz
(1.8 kWh/m 3 ), respectively. The energy consumption values
for the entire skid, which include energy related to the
generators, pumps, and controls, are also provided for each
scenario.
The relative degradation rates were identical to those of
the batch experiment. For example, carbamazepine and
trimethoprim experienced the most rapid degradation, and
primidone and meprobamate were the most recalcitrant
compounds. At the highest level of energy consumption
(1.8 kWh/m 3 ), every compound except meprobamate experienced
at least 70% degradation. The evaluation of
C/C 0
UV/H 2O 2 c,d
SSW
UV/TiO 2 c,e
SSW
Meprobamate 6.4 4.6 10 14 2.1 3.5 5.3 6.6 1.0 6.8
Dilantin 4.4 2.7 3.1 3.5 1.1 2.0 3.1 2.1 1.0 2.2
Primidone 5.8 2.8 4.3 4.8 1.1 2.2 3.3 3.7 0.6 3.9
Carbamazepine 2.2 1.3 1.8 2.1

500
Fig. 5 – Excitation-emission matrices as a function of generator energy consumption for the single-pass experiment (2) with
tertiary-treated wastewater. The regions are described as follows: (I) Aromatic protein, (II) aromatic protein II, (III) fulvic
acid-like, (IV) soluble microbial byproduct-like, and (V) humic acid-like (Chen et al., 2003).
scope of this study, but future studies should address this
issue in greater detail.
3.3. Experiment 3 – single-pass experiment with spiked
surface water
Contaminant degradation profiles, including that of atrazine,
for the single-pass spiked surface water experiment are
illustrated in Fig. 6. The initial spiked concentrations were
much higher than the ambient wastewater concentrations as
all of the compounds were spiked at greater than 900 ng/L
(Table 1). Similar to the single-pass wastewater experiment,
this experiment involved a single-reactor configuration and
both reactors operating in series at a flow rate of 11.4 L/min
(Table 2). For this experiment, Generator 1 was operated at
a voltage of 8.0 kV, and Generator 2 was operated at a voltage
of 5.9 kV. The generator and overall skid energy consumption
values ranged from 0.6 to 2.6 kWh/m 3 and 4.8 to 6.8 kWh/m 3 ,
respectively.
The relative degradation rates were similar to those of the
wastewater experiments. For example, meprobamate was the
most difficult to degrade, while trimethoprim and carbamazepine
rapidly dropped below the MRL in all treated samples.
The lower organic loading of the surface water (i.e., lower
initial TOC concentration and UV254 absorbance), which
water research 44 (2010) 493–504
resulted in a higher residual ozone concentration (0.25 to
0.50 mg/L), likely contributed to the improved degradation. In
contrast to the wastewater experiments, atenolol degradation
was more similar to trimethoprim and carbamazepine, while
primidone degradation was more similar to that of dilantin.
As expected, the degradation of atrazine, which is considered
a recalcitrant compound (Westerhoff et al., 2005), was almost
identical to that of meprobamate. For all of the compounds
with concentrations above the MRL, the degradation peaked at
about 70% for the most recalcitrant compounds and 85% for
the moderately recalcitrant compounds after reaching an
energy consumption of 1.6 kWh/m 3 . At the highest energy
level, the effluent concentrations of the residual compounds
ranged from 140 ng/L for dilantin to 390 ng/L for atrazine.
During Experiment 3, there was also greater fluctuation for
several time points. Since the samples represent discrete,
rather than continuous (i.e., batch), treatment levels, one
would anticipate greater variability, but there was still a relatively
consistent decrease in contaminant concentration over
the range of energy values tested.
The EEO values for all of the compounds were dramatically
lower in the surface water experiment than the wastewater
experiments, presumably due to the lower scavenging
capacity of the surface water. Meprobamate and atrazine had
the highest EEOs with maximum values of 5.3 and 6.3 kWh/

C/C 0
1
0.8
0.6
0.4
0.2
0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
m 3 -log, respectively. The maximum values generally occurred
during the experiments when one or both of the generators
were operating at 1,000 Hz. Although operating both generators
at 1,000 Hz achieved the highest level of degradation,
increasing the pulse frequency generally yielded diminishing
rates of return with respect to efficiency. Operating Generator
2 at a frequency of 600 Hz provided the most efficient treatment
conditions for all of the target compounds. However,
Generator 1 had been operated immediately prior to that
scenario, so the residual ozone from Reactor 1 likely contributed
to this increased efficiency. Although this overestimates
the anticipated degree of degradation, it indicates that it may
be beneficial to incorporate ozone contact tanks before and
after each reactor to maximize degradation. Based on this
Generator Energy Consumption (kWh/m 3 )
Meprobamate Dilantin Primidone Carbamazepine Atenolol Trimethoprim Atrazine
Fig. 6 – Contaminant degradation profiles from the single-pass experiment (3) with spiked surface water. This graph
illustrates the relationship between contaminant degradation and generator (AOP-specific) energy consumption.
Concentration (ng/L)
1600
1400
1200
1000
800
600
400
200
0
water research 44 (2010) 493–504 501
observation, the median values are likely the most
appropriate EEO estimates since they always corresponded
to both reactors operating simultaneously. For the
compounds with reportable concentrations, the median
EEO values ranged from 1.0 to 3.7 kWh/m 3 for atenolol and
atrazine, respectively.
A subset of samples was also analyzed to assess the
contribution of the system components (i.e., reactor vs. ozone
contactor) to contaminant degradation. The results from one
set of samples (SSW-8) are provided in Fig. 7; the other sample
sets yielded similar trends and therefore are not shown. Based
on these results, it appears that the ozone contactors provided
the majority of the degradation for the more recalcitrant
compounds, including meprobamate, dilantin, primidone,
Influent Reactor 1 Ozone 1 Reactor 2 Effluent
Sample Location
Meprobamate Dilantin Primidone Carbamazepine Atenolol Trimethoprim Atrazine
Fig. 7 – Comparison of the pilot system components in degrading the target compounds during the single-pass experiment
(3) with spiked surface water. This data refers to samples from SSW-8 with generator and skid energy consumption values
of 2.2 and 6.3 kWh/m 3 , respectively. The small increase in concentrations from Ozone 1 to Reactor 2 is explained by mixing
of Reactor 1 and Ozone 1 effluent prior to passage into Reactor 2 (see Fig. 1).

502
% Degradation of Trace Organic Compounds
100%
80%
60%
40%
20%
Linear Regression Models:
Contaminant Slope Intercept R 2
N
Meprobamate 2.40 0 0.72 9
Dilantin 1.35 45% 0.84 9
Primidone 1.52 40% 0.81 9
Atrazine 2.27 0 0.74 9
0%
0% 5% 10% 15% 20% 25% 30% 35%
atenolol, and atrazine. However, carbamazepine and
trimethoprim experienced rapid degradation in the reactor
itself with the additional oxidation pathways (i.e., UV light,
hydroxyl radicals, and ozone).
The spiked surface water experiment was also evaluated
based on general water quality parameters. For example, the
initial bromide concentration in the surface water was
approximately 62 mg/L, and additional analyses indicated that
bromate increased only slightly (maximum of 2.4 mg/L) above
the ambient surface water concentration of 1.7 mg/L. Similar
to the wastewater experiments, there was no significant
change in TOC over the duration of the experiment, but the
reduction in UV254 absorbance was correlated to the degradation
of the target contaminants in Fig. 8. In a manner
similar to Fig. 3, Fig. 8 illustrates strong positive correlations
between contaminant degradation and reduction in UV 254
absorbance. Furthermore, the degradation of target contaminants
was still more rapid than the reduction in UV 254
absorbance. In contrast to the batch wastewater experiment
and data provided in Wert et al. (2009), the regression lines
for dilantin and primidone have positive intercepts, while the
intercepts for meprobamate and atrazine were found to be
insignificant ( p > 0.09). Wert et al. (2009) proposed that the
correlations between contaminant degradation and UV254
absorbance were independent of wastewater quality, but it
appears that the correlations differ when the matrices
themselves are drastically different (i.e., wastewater vs.
surface water). Specifically, the positive intercepts for
dilantin and primidone in the surface water experiment
indicate that these target compounds started to degrade
(z40–45%) prior to any reduction in UV 254 absorbance. In
addition to contributing to their smaller slopes, this may
indicate that there was less oxidant/radical scavenging and
that the bulk organic matter was more recalcitrant in the
surface water compared to the wastewater. It is also interesting
to note that meprobamate has the smallest absolute
intercept for the wastewater experiment and an insignificant
water research 44 (2010) 493–504
% Reduction in UV254 Absorbance
Meprobamate Dilantin Primidone Atrazine
Fig. 8 – Correlation between the degradation of detected target compounds and UV254 absorbance during the single-pass
experiment (3) with spiked surface water. The slopes were all significant at the 0.001 level, the intercepts for primidone and
dilantin were significant at the 0.001 level, and the intercepts for meprobamate and atrazine were considered insignificant
( p > 0.09; regressions for meprobamate and atrazine were forced through the origin).
intercept for the surface water experiment, and atrazine also
has an insignificant intercept for the surface water experiment.
Therefore, the initial degradation of these highly
resistant compounds is more similar to the initial degradation
of the bulk organic matter. For the surface water
experiments, this supports the theory that the aromatic
organic matter in the surface water was more recalcitrant
than that of the wastewater. Based on these observations, it
appears that the UV 254 surrogate framework is still valid, but
there may be differences in the correlations for dissimilar
water matrices.
Table 4 provides EEO values for the three experiments in
this study, and it also provides EEO values from Benotti et al.
(2009b), which evaluated the efficacy of photolysis, UV/H2O2,
and photocatalysis in degrading PPCPs and EDCs. Benotti et al.
(2009b) provides an interesting point of comparison since the
water matrix used for that study (spiked surface water from
Lake Mead), target compounds, and analytical methods were
identical. As indicated by the data, NTP was generally more
efficient than photolysis and photocatalysis for each of the
compounds, but was either comparable or slightly less efficient
than UV/H2O2 at a peroxide dose of 10 mg/L. It is
important to note that the NTP technology does not require
UV lamps or costly peroxide feeds to generate the oxidative
species. Therefore, NTP may be a viable alternative to the
more common UV/H2O2 process despite the slightly higher
energy requirements for some compounds. Although the NTP
achieved comparable efficiencies to those of UV/H 2O 2 in these
experiments, identification of the best alternative for
a particular application also depends on capital costs and
reliability. Since the NTP technology has not been fully
commercialized, it is difficult to compare its capital costs and
reliability with more established technologies. Future evaluations
of large-scale NTP reactors, including long-term
studies, will provide valuable information in determining how
NTP compares to more established technologies on a life-cycle
basis.

4. Conclusions
With respect to treatment efficacy, the pilot-scale NTP system
demonstrated rapid degradation of the target compounds.
The relative degradation rates were consistent between the
three phases of the study with carbamazepine and trimethoprim
as the most susceptible compounds and meprobamate,
primidone, and atrazine as the most recalcitrant compounds.
Atenolol and dilantin were generally more resistant to
oxidation than carbamazepine and trimethoprim. These
similar relative degradation rates justify the use of indicator
compounds in evaluations of PPCP and EDC treatment. The
EEO values for the optimal scenarios were generally less than
5 kWh/m 3 -log for all of the target compounds, particularly in
the spiked surface water with a lower organic loading. With
process optimization (i.e., optimization of electrode height for
each water matrix) and limited modifications to the system,
including more effective use of ambient ozone, the EEO values
could likely be reduced further.
More common AOPs, including UV/H 2O 2 and ozone/H 2O 2,
require significant chemical addition and residual H 2O 2
quenching, which represent a significant portion of their
operational costs. The primary benefit of the NTP AOP is the
ability to generate UV light, ozone, and hydroxyl radicals
without chemical addition or the use of UV lamps. However,
the true capital costs and reliability of large-scale NTP reactors
for water and wastewater treatment are still relatively unclear.
Also, since oxidation by molecular ozone may be the dominant
mechanism for many compounds, future studies should
provide direct comparisons of NTP with conventional ozone
generators to compare system performance. These expanded
studies are necessary before one can fully determine the
applicability of the novel NTP technology for water and
wastewater treatment. However, these preliminary results
indicate that NTP may be a viable alternative to more common
AOPs due to its chemical-free operation and comparable
energy requirements for trace contaminant degradation.
Acknowledgments
We would like to thank the SNWA Applied Research and
Development Center staff, particularly Janie Zeigler, Shannon
Ferguson, Christy Meza, and Elaine Go, for assisting with the
various analyses. We would also like to thank Dvir Solnik, Itay
Even-Ezra, Gal Bitan, and Shahar Nuriel of Aquapure Technologies
for use of the nonthermal plasma pilot unit and for
providing crucial training and guidance during the study.
references
Acero, J.L., von Gunten, U., 2001. Characterization of oxidation
processes: ozonation and the AOP O 3/H 2O 2. J. AWWA 93 (10),
90–100.
Benotti, M.J., Stanford, B.D., Snyder, S.A., 2009a. Increased loading
of pharmaceuticals and endocrine disrupting compounds to
source drinking water due to climate change. J. Environ. Qual.
Accepted.
water research 44 (2010) 493–504 503
Benotti, M.J., Stanford, B.D., Wert, E.C., Snyder, S.A., 2009b.
Evaluation of a photocatalytic reactor membrane pilot system
for the removal of pharmaceuticals and endocrine disrupting
compounds from water. Water Res. 43 (6), 1513–1522.
Benotti, M.J., Trenholm, R.A., Vanderford, B.J., Holaday, J.C.,
Stanford, B.D., Snyder, S.A., 2009c. Pharmaceuticals and
endocrine disrupting compounds in U.S. drinking water.
Environ. Sci. Technol. 43 (3), 597–603.
Bolton, J.R., Bircher, K.G., Tumas, W., Tolman, C.A., 1996. Figuresof-merit
for the technical development and application of
advanced oxidation processes. J. Adv. Oxid. Technol. 1 (1),
13–17.
Chen, W., Westerhoff, P., Leenheer, J.A., Booksh, K., 2003.
Fluorescence excitation-emission matrix regional integration
to quantify spectra for dissolved organic matter. Environ. Sci.
Technol. 37 (24), 5701–5710.
Even-Ezra, I., Mizrahi, A., Gerrity, D., Snyder, S., Salveson, A.,
Lahav, O., 2009. Application of a novel plasma-based advanced
oxidation process for efficient and cost effective destruction of
refractory organics in tertiary effluents and contaminated
groundwater. Desalin. Water Treat. Accepted.
Gerrity, D., Mayer, B., Ryu, H., Crittenden, J., Abbaszadegan, M.,
2009. A comparison of pilot-scale photocatalysis and
enhanced coagulation for disinfection byproduct mitigation.
Water Res. 43, 1597–1610.
Huber, M.M., Canonica, S., Park, G., von Gunten, U., 2003.
Oxidation of pharmaceuticals during ozonation and advanced
oxidation processes. Environ. Sci. Technol. 37, 1016–1024.
Hwang, Y.H., Jo, Y.M., 2005. Decomposition of odorous gases in
a pilot-scale nonthermal plasma reactor. J. KOSAE 21 (2),
57–65.
Ighigeanu, D., Martin, D., Zissulescu, E., Macarie, R., Oproiu, C.,
Cirstea, E., Iovu, H., Calinescu, I., Iacob, N., 2005. SO 2 and NO x
removal by electron beam and electrical discharge induced
non-thermal plasmas. Vacuum 77, 493–500.
Kim, S.D., Cho, J., Kim, I.S., Vanderford, B.J., Snyder, S.A., 2007.
Occurrence and removal of pharmaceuticals and endocrine
disruptors in South Korean surface, drinking, and waste
waters. Water Res. 41, 1013–1021.
Kim, H., Han, J., Sakaguchi, Y., Minami, W., 2008. Simultaneous
oxidation of NO x and SO 2 by a new non-thermal plasma
reactor enhanced by catalyst and additive. Plasma Sci.
Technol. 10 (1), 53–56.
Kolpin, D.W., Furlong, E.T., Meyer, M.T., Thurman, E.M., Zaugg, S.
D., Barber, L.B., Buxton, H.T., 2002. Pharmaceuticals,
hormones, and other organic wastewater contaminants in U.
S. streams, 1999–2000: A national reconnaissance. Environ.
Sci. Technol. 36, 1202–1211.
Locke, B.R., Sato, M., Sunka, P., Hoffmann, M.R., Chang, J.S., 2006.
Electrohydraulic discharge and nonthermal plasma for water
treatment. Ind. Eng. Chem. Res. 45, 882–905.
Masuda, S., Hosokawa, S., Tu, X., Wang, Z., 1995. Novel plasma
chemical technologiesdPPCP and SPCP for control of gaseous
pollutants and air toxics. J. Electrostatics 34, 415–438.
Mitch, W.A., Sharp, J.O., Trussell, R.R., Valentine, R.L., Alvarez-
Cohen, L., Sedlak, D.L., 2003. N-nitrosodimethylamine (NDMA)
as a drinking water contaminant: a review. Environ. Eng. Sci.
20 (5), 389–404.
Pekarek, S., 2003. Non-thermal plasma ozone generation. Acta
Polytech. 43 (6), 47–51.
Pereira, V.J., Linden, K.G., Weinberg, H.S., 2007. Evaluation of UV
irradiation for photolytic and oxidative degradation of
pharmaceutical compounds in water. Water Res. 41 (19),
4413–4423.
Rosal, R., Rodriguez, A., Perdigon-Melon, J.A., Petre, A., Garcia-
Calvo, E., 2009. Oxidation of dissolved organic matter in the
effluent of a sewage treatment plant using ozone combined
with hydrogen peroxide (O 3/H 2O 2). Chem. Eng. J. 149, 311–318.

504
Snyder, S.A., Wert, E.C., Lei, H., Westerhoff, P., Yoon, Y., 2007.
Removal of EDCs and Pharmaceuticals in Drinking and
Reuse Treatment Processes. Water Research Foundation,
Denver, CO.
Snyder, S.A., Westerhoff, P., Yoon, Y., Sedlak, D.L., 2003.
Pharmaceuticals, personal care products, and endocrine
disruptors in water: implications for the water industry.
Environ. Eng. Sci. 20 (5), 449–469.
Ternes, T.A., 1998. Occurrence of drugs in German sewage
treatment plants and rivers. Water Res. 32 (11), 3245–3260.
Ternes, T.A., Meisenheimer, M., McDowell, D., Sacher, F.,
Brauch, H., Haist-Gulde, B., Preuss, G., Wilme, U., Zulei-
Seibert, N., 2002. Removal of pharmaceuticals during drinking
water treatment. Environ. Sci. Technol. 36, 3855–3863.
Thevenet, F., Guaitella, O., Puzenat, E., Herrmann, J.M.,
Rousseau, A., Guillard, C., 2007. Oxidation of acetylene by
photocatalysis coupled with dielectric barrier discharge. Catal.
Today 122, 186–194.
water research 44 (2010) 493–504
Trenholm, R.A., Vanderford, B.J., Snyder, S.A., 2009. On-line SPE
LC-MS/MS analysis of pharmaceutical indicators in water:
a green alternative to conventional methods. Talanta 79,
1425–1432.
Wert, E.C., Rosario-Ortiz, F.L., Snyder, S.A., 2009. Using ultraviolet
absorbance and color to assess pharmaceutical oxidation
during ozonation of wastewater. Environ. Sci. Technol. 43,
4858–4863.
Westerhoff, P., Yoon, Y., Snyder, S., Wert, E., 2005. Fate of
endocrine-disruptor, pharmaceutical, and personal care
product chemicals during simulated drinking water treatment
processes. Environ. Sci. Technol. 39, 6649–6663.
Westerhoff, P., Moon, H., Minakata, D., Crittenden, J., 2009.
Oxidation of organics in retentates from reverse osmosis
wastewater reuse facilities. Water Res. 43 (16), 3992–3998.
Yates, R.S., Stenstrom, M.K., 2000. Gravimetric sampling
procedure for aqueous ozone concentrations. Water Res. 34
(4), 1413–1416.

Single-walled carbon nanotubes dispersed in aqueous media
via non-covalent functionalization: Effect of dispersant
on the stability, cytotoxicity, and epigenetic toxicity
of nanotube suspensions
Alla L. Alpatova a , Wenqian Shan a , Pavel Babica b , Brad L. Upham b,c ,
Adam R. Rogensues a , Susan J. Masten a , Edward Drown d,1 , Amar K. Mohanty d,2 ,
Evangelyn C. Alocilja e , Volodymyr V. Tarabara a, *
a
Department of Civil and Environmental Engineering, Michigan State University, East Lansing, MI 48824, USA
b
Center for Integrative Toxicology, Michigan State University, East Lansing, MI 48824, USA
c
Department of Pediatrics and Human Development, Michigan State University, East Lansing, MI 48824, USA
d
School of Packaging, Michigan State University, East Lansing, MI 48824, USA
e
Department of Biosystems and Agricultural Engineering, Michigan State University, East Lansing, MI 48824, USA
article info
Article history:
Received 2 June 2009
Received in revised form
14 September 2009
Accepted 17 September 2009
Available online 30 September 2009
Keywords:
Single-walled carbon nanotubes
Dispersion
Non-covalent functionalization
Cytotoxicity
Epigenetic toxicity
water research 44 (2010) 505–520
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
abstract
As the range of applications for carbon nanotubes (CNTs) rapidly expands, understanding
the effect of CNTs on prokaryotic and eukaryotic cell systems has become an important
research priority, especially in light of recent reports of the facile dispersion of CNTs in
a variety of aqueous systems including natural water. In this study, single-walled carbon
nanotubes (SWCNTs) were dispersed in water using a range of natural (gum arabic,
amylose, Suwannee River natural organic matter) and synthetic (polyvinyl pyrrolidone,
Triton X-100) dispersing agents (dispersants) that attach to the CNT surface non-covalently
via different physiosorption mechanisms. The charge and the average effective hydrodynamic
diameter of suspended SWCNTs as well as the concentration of exfoliated SWCNTs
in the dispersion were found to remain relatively stable over a period of 4 weeks. The
cytotoxicity of suspended SWCNTs was assessed as a function of dispersant type and
exposure time (up to 48 h) using general viability bioassay with Escherichia coli and using
neutral red dye uptake (NDU) bioassay with WB-F344 rat liver epithelia cells. In the E. coli
viability bioassays, three types of growth media with different organic loadings and salt
contents were evaluated. When the dispersant itself was non-toxic, no losses of E. coli and
WB-F344 viability were observed. The cell viability was affected only by SWCNTs dispersed
using Triton X-100, which was cytotoxic in SWCNT-free (control) solution. The epigenetic
toxicity of dispersed CNTs was evaluated using gap junction intercellular communication
(GJIC) bioassay applied to WB-F344 rat liver epithelial cells. With all SWCNT suspensions
except those where SWCNTs were dispersed using Triton X-100 (wherein GJIC could not be
measured because the sample was cytotoxic), no inhibition of GJIC in the presence of
SWCNTs was observed. These results suggest a strong dependence of the toxicity of
* Corresponding author. Tel.: þ517 432 1755; fax: þ517 355 0250.
E-mail address: tarabara@msu.edu (V.V. Tarabara).
1
Present address: Department of Chemical Engineering and Materials Science, Michigan State University, East Lansing, MI 48824, USA.
2
Present address: Department of Plant Agriculture, University of Guelph, 50 Stone Rd. E., Guelph, Ontario, N1 G 2W1 Canada.
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.watres.2009.09.042

506
1. Introduction
The discovery (Iijima, 1991) and subsequent extensive characterization
of carbon nanotubes (CNTs) have revealed a class
of materials with extraordinary electrical, mechanical, and
thermal properties (Tasis et al., 2006). The wider application of
CNTs in electronic, optical, sensing, and biomedical fields has
been impeded by the low solubility of as-produced CNTs in
polar liquids and by the strong tendency of CNTs to aggregate
due to hydrophobic–hydrophobic interactions (Lin et al., 2004).
Dispersing CNTs and ensuring long-term stability of CNTs
suspended in a liquid medium have proved especially challenging
for aqueous systems.
1.1. Dispersing CNTs in aqueous media
Recent efforts on the development of efficient and facile
methods of dispersing CNTs in aqueous media have been
focused on the hydrophilization of CNT with molecules that
bind to the CNT surface non-covalently (O’Connell et al., 2001;
Bandyopadhyaya et al., 2002; Star et al., 2002; Islam et al., 2003;
Moore et al., 2003; Didenko et al., 2005; Wang et al., 2005;
Bonnet et al., 2007; Grossiord et al., 2007; Hyung et al., 2007; Liu
et al., 2007). Such non-covalent functionalization has great
promise as the modification-induced changes in the electronic
and mechanical properties of CNTs are minimized
(Yang et al., 2007). Various surfactants (Islam et al., 2003;
Moore et al., 2003; Grossiord et al., 2007), synthetic polymers
(e.g., polyvinyl pyrrolidone (O’Connell et al., 2001; Didenko
et al., 2005), poly(ethylene glycol) (Vaisman et al., 2006), polyphosphazene
(Park et al., 2006)), natural organic matter
(NOM) (Hyung et al., 2007; Liu et al., 2007; Saleh et al., 2009),
biomolecules (e.g., proteins (Karajanagi et al., 2006; Zong et al.,
2007), aminoacids (Georgakilas et al., 2002), DNA (Enyashin
et al., 2007)), and carbohydrates (e.g., cyclodextrines (Dodziuk
et al., 2003), amylose (Bonnet et al., 2007), starch (Star et al.,
2002), and GA (Bandyopadhyaya et al., 2002)) have been evaluated
as dispersants for CNTs. Dispersion via non-covalent
functionalization is based on the direct contact between
a CNT and a dispersant molecule (Liu et al., 1998; Grossiord
et al., 2007). Such modification of the CNT surface facilitates
the disaggregation (i.e. debundling) of CNT bundles into
smaller diameter bundles (Liu et al., 1998) or even individual
CNTs (O’Connell et al., 2002; Hyung et al., 2007) and leads to
the stabilization of suspended CNTs via steric or electrostatic
repulsion mechanisms or both. (See Supporting Documentation
(SD), Section S.1.1, for a brief review of mechanisms of
non-covalent functionalization of CNTs.)
The dispersion of CNTs in water has been enhanced by
mixing (Didenko et al., 2005; Hyung et al., 2007), sonication
(Bandyopadhyaya et al., 2002; Liu et al., 2007; Salzmann et al.,
2007), or mixing followed by sonication (O’Connell et al., 2001,
water research 44 (2010) 505–520
SWCNT suspensions on the toxicity of the dispersant and point to the potential of noncovalent
functionalization with non-toxic dispersants as a method for the preparation of
stable aqueous suspensions of biocompatible CNTs.
ª 2009 Elsevier Ltd. All rights reserved.
2002; Star et al., 2002; Islam et al., 2003; Moore et al., 2003;
McDonald et al., 2006; Grossiord et al., 2007). These treatments
were applied both in the absence (Salzmann et al., 2007) and in
the presence of solubilizing agents: NOM (Hyung et al., 2007),
Triton X-100 (Islam et al., 2003), Triton X-405 (Chappell et al.,
2009), PVP-1300 (Didenko et al., 2005), GA (Bandyopadhyaya
et al., 2002), and starch (Star et al., 2002). A three-step
approach to solubilizing single-walled carbon nanotubes
(SWCNTs) with amylose was suggested by Kim et al. (Kim
et al., 2004): dispersion of SWCNT in water by sonication followed
by treatment with amylose in dimethylsulfoxide
(DMSO)–H2O mixture, followed by sonication allowing for
molecularly controlled encapsulation of CNTs.
1.2. Stability of CNT suspensions in water
Previous studies of the long-term changes in suspensions of
dispersed CNTs have focused on monitoring changes in the
concentration of suspended CNTs (Jiang and Gao, 2003; Tseng
et al., 2006; Lee et al., 2007; Marsh et al., 2007). By measuring
UV–vis absorption at certain wavelength: 253 nm (Jiang and
Gao, 2003), 300 nm (Sinani et al., 2005), 500 nm (Bahr et al.,
2001; Lee et al., 2007), 530 nm (Marsh et al., 2007), and 800 nm
(Hyung et al., 2007) the change in the concentration of suspended
CNTs with time was determined.
Aqueous suspensions of non-functionalized CNTs are
known to be unstable. There are considerable quantitative
differences, however, in the reported stability data for nonfunctionalized
CNTs. The concentration of unmodified multiwalled
carbon nanotubes (MWCNTs) suspended in deionized
water was reported to decline 86 % over 2 h in one study
(Marsh et al., 2007) and only 50% over 500 h in another study
(Jiang and Gao, 2003). The suspensions of unmodified SWCNTs
in deionized water were found to completely precipitate after
only 4 h (Tseng et al., 2006).
It was found that non-covalent modification of CNT
surface drastically improved the stability of CNT suspensions
(Jiang and Gao, 2003; Tseng et al., 2006; Hyung et al., 2007; Lee
et al., 2007; Marsh et al., 2007). The stability of the suspension
of non-covalently functionalized CNTs was found to be
a function of the type (Hyung et al., 2007; Lee et al., 2007) and
concentration (Chappell et al., 2009) of the dispersant, CNT
length (Marsh et al., 2007) and the presence of inorganic salts
in the solution (Saleh et al., 2009). SWCNTs stabilized with
oligothiophene-terminated poly(ethylene glycol) produced
suspensions that were significantly more stable than CNTs
dispersed in water with the aid of sodium dodecyl sulfate
(SDS) and Pluronic F127 (Lee et al., 2007). CNT suspensions in
model Suwannee River NOM (SRNOM) solutions and in
Suwannee River water were found to be considerably more
stable than suspensions of CNTs dispersed in 1% aqueous
solution of SDS (Hyung et al., 2007). Chappell et al. showed

that the stability of MWCNTs dispersed using two types of
humic acids, Triton X-405, Brij 35, and SDS was enhanced in
dose-dependent manner (Chappell et al., 2009). In a study of
the aggregation kinetics of MWCNTs in aquatic systems,
Suwannee River humic acid was shown to significantly
enhance stability of MWCNTs suspensions in the presence of
monovalent and divalent salts (Saleh et al., 2009). While
studying the stability of MWCNTs in water, Marsh and coworkers
(Marsh et al., 2007) observed that wrapping of the
annealed CNTs with 1% SDS or charge doping increased their
stability in the suspension; the authors demonstrated that
shorter CNTs form more stable dispersions than longer CNTs
of the same diameter. In summary, there is growing evidence
that by the appropriate choice of a dispersant that modifies
CNT surface non-covalently, highly stable CNT suspensions
can be produced. However, to date there have been no systematic
investigations that correlated long-term changes in size and charge
of CNTs dispersed via non-covalent functionalization with the
dispersion stability.
1.3. Toxicity of CNTs
1.3.1. Toxicity of CNTs towards eukaryotic cells: cytotoxicity
Most nanomaterial toxicity studies have been performed with
mammalian cells, in particular with lung and skin cell
cultures reflecting the understanding that the most likely
routes of an organism’s exposure to nanomaterials are
respiratory and dermal contact. With the development of
methods of CNT dispersion in aqueous media, the assessment
of the toxicity of such dispersed CNTs becomes very important
in view of their increased mobility and potential to enter
water supplies. While the toxicity of CNTs was studied with
respect to the type of the CNT (MWCNT versus SWCNT) (Ding
et al., 2005; Jia et al., 2005; Kang et al., 2008a), surface functionalization
(Sayes et al., 2006; Yu et al., 2007; Meng et al.,
2009; Yun et al., 2009), CNT length (Muller et al., 2005; Kang
et al., 2008b; Simon-Deckers et al., 2008), exposure dose
(Cherukuri et al., 2006; Flahaut et al., 2006; Sayes et al., 2006;
Kang et al., 2007; Pulskamp et al., 2007; Yang et al., 2008, 2009;
Ye et al., 2009), degree of purification (Fiorito et al., 2006;
Flahaut et al., 2006; Elias et al., 2007; Simon-Deckers et al.,
2008), and degree of dispersion (Wick et al., 2007), only very
limited information exists on the biocompatibility of CNTs as
a function of surface coating.
Several hypotheses have been put forth to explain the likely
pathways of CNT toxicity: (i) oxidative stress induced by the
formation of reactive oxygen species (ROS) generated at the
surface of CNTs (Manna et al., 2005; Yang et al., 2008; Ye et al.,
2009); (ii) the presence of residual catalyst, which is used during
the CNT manufacturing process (Kang et al., 2008a);
(iii) physical contact between a CNT and a cell (Kang et al., 2007);
or (iv) a combination of these factors (Kang et al., 2008a). The
dispersant used for the stabilization of CNTs in suspension was
suggested as another possible cause of the observed nanotube
toxicity (Sayes et al., 2006) (SeeSD, Section S.1.2 for a brief
review of possible mechanisms of CNT toxicity).
In most studies of the toxicity of non-covalently functionalized
CNTs the dispersants were surfactants. In vivo
assessment of the toxicity of SWCNTs modified with Pluronic
F108 administered intravenously to rabbits showed the
water research 44 (2010) 505–520 507
absence of the acute toxicity (Cherukuri et al., 2006). Pluronic
F-127-coated MWCNTs did not affect cell viability, apoptosis,
and ROS formation in mouse and human neuroblastoma cells
(Vittorio et al., 2009), and mouse cerebral cortex (Bardi et al.,
2009). In contrast, MWCNTs dispersed using Pluronic F68
caused cell death, changes in cell size and complexity, ROS
production, interleukin-8 (IL-8) gene expression and nuclear
factor (NF)-jB activation (Ye et al., 2009). CNTs dispersed using
Tween 80 were toxic to mesothelioma cells (Wick et al., 2007),
led to an inflammation of murine allergic airway with
augmented humoral immunity (Inoue et al., 2009), and
induced inflammatory and fibrotic responses when intracheally
administrated to rats (Muller et al., 2005). More ROS
were observed upon exposure of human lung epithelial or
primary bronchial epithelial cells to SWCNTs dispersed using
dipalmitoylphosphatidylcholine, a major component of lung
surfactant (Herzog et al., 2009), as compared to dipalmitoylphosphatidylcholine-free
samples. Low toxicity was observed
in in vivo experiments when SWCNTs were dispersed using
Tween 80 and intravenously injected into mice (Yang et al.,
2008). In another study, SWCNTs dispersed in 1% SDS aqueous
solution showed no cytotoxicity with respect to the human
alveolar epithelial cells (Wörle-Knirsch et al., 2006). CNTs
dispersed using Pluronic PF-127 solution did not affect
viability, apoptosis and ROS generation in the human neuroblastoma
cells after 3 days of incubation; however, cell
proliferation decreased as incubation time increased to 2
weeks (Vittorio et al., 2009). In the only paper that mentioned
the potential toxicity effect of the dispersant, the authors
suggested that excess surfactant was responsible for the
observed increase in toxicity; in this work, the controlled
exposure of cells to 1% Pluronic F108 produced a 10% decrease
in the cells viability (Sayes et al., 2006).
Very little is known about the effect of non-covalent wrapping
with dispersants other than surfactants on the biocompatibility of
CNTs. In vitro cytotoxicity of CNTs wrapped with poly(methyl
vinylketone) decorated with a-N-acetylgalactosamine (Chen
et al., 2006), nano-1 peptide (Chin et al., 2007) and cholesterolend-capped
poly(2-methacryloyloxyethyl phosphorylcholine)
(Xu et al., 2008) were examined after their contact with human
lung epithelial-like cells, normal skin fibroblasts and human
umbilical vein endothelial cell line. No impact on cell growth
and proliferation was demonstrated in all three cases. Cytotoxicity
of GA-stabilized MWCNTs upon exposure to A549 cells
was observed by LDH, XTT and MTT bioassays in (Simon-
Deckers et al., 2008). Two possible reasons were hypothesized
to be responsible for this effect: (i) increased availability of GAstabilized
MWCNTs and (ii) different intercellular accumulation
pathway of GA-stabilized MWCNTs as compared to when
carbon nanotubes were exposed in bundles.
1.3.2. Toxicity of CNTs towards eukaryotic cells: genotoxicity
and epigenetic toxicity
Most toxicological studies of CNTs focused on the evaluation of
cytotoxicity (Sayes et al., 2006; Elias et al., 2007; Gutierrez et al.,
2007; Kang et al., 2007; Wick et al., 2007; Kang et al., 2008a,b;
Yang et al., 2008; Bardi et al., 2009; Herzog et al., 2009; Inoue et al.,
2009; Kang et al., 2009; Vittorio et al., 2009; Yang et al., 2009).
However, genotoxicity (Kang et al., 2008a; Di Sotto et al., 2009;
Lindberg et al., 2009; Wirnitzer et al., 2009; Yang et al., 2009)and

508
epigenetic toxicity (see SD, Section S.1.3) (Upham et al., 1994);
(Trosko et al., 1998) are other possible causes of cell damage or
other adverse effects. The genotoxicity of SWCNTs dispersed
using fetal bovine serum at (5–10) mg/mL concentration, with
respect to primary mouse embryo fibroblasts was demonstrated
using Comet assay (Yang et al., 2009). In another recent
study, CNTs (dispersed in BEGM cell culture medium and subjected
to ultrasonication) induced a dose-dependent increase in
DNA damage as indicated by Comet assay and caused a significant
increase in micronucleated cells (micronucleus assay) in
human bronchial epithelial cells (Lindberg et al., 2009). Kang
et al. observed high levels of stress-related gene products in
Escherichia coli upon its exposure to CNTs, with the quantity and
magnitude of expression being much higher in the presence of
SWCNTs (Kang et al., 2008a); in this study CNTs were either
deposited on a PVDF membrane surface or dispersed in saline
solution. No mutagenic effect was observed in Salmonella
microcosme test with baytubes Ò (high purity MWCNTs
agglomerates, sonicated 10 min in deionized water) (Wirnitzer
et al., 2009) and in bacterial reverse mutation assay (Ames test)
with Salmonella typhimurium TA 98 and TA 100 strains and
with E. coli WP2uvrA strain exposed to MWCNTs dispersed in
DMSO (Di Sotto et al., 2009). There have been no reports to date on the
epigenetic toxicity of carbon nanomaterials.
1.3.3. Toxicity of CNTs in prokaryotic cells
Most toxicity studies have focused on the effect of CNTs on
mammalian cell lines. Only limited information exists on the
cytotoxic effects of CNTs towards bacterial cells. Recently,
antimicrobial activity of SWCNTs (Kang et al., 2007, 2008a,
2009) and MWCNTs (Kang et al., 2008a,b, 2009) either suspended
in aqueous solution (Kang et al., 2007, 2008a) or
deposited on the surface of a PVDF microfilter (Kang et al.,
2007, 2008a,b, 2009) towards gram-negative bacteria (E. coli
and Pseudomonas aeruginosa (Kang et al., 2007, 2008a,b, 2009)
and gram-positive (Staphylococcus epidermidis and Bacillus subtilis
(Kang et al., 2009)) bacteria was reported. It was suggested
that membrane damage to the cells was caused by the direct
physical contact between the CNT and cell (Kang et al., 2007)
or by a combination of direct physical contact and oxidative
stress (Kang et al., 2008a). No effect on the percentage of E. coli
inactivation was observed upon exposure of SRNOM-stabilized
SWCNTs as compared of SWCNTs dispersed in the
absence of SRNOM (Kang et al., 2009). Significant antimicrobial
activity of CNTs composite films against Staphylococcus aureus
and Staphylococcus warneri was reported in a separated study
(Narayan et al., 2005). In contrast to the findings of above
studies, no inhibition of E. coli growth and proliferation was
reported in a study where microchannel-structured MWCNTs
scaffolds were immersed into the culture medium with the
cells (Gutierrez et al., 2007).
1.4. Effect of growth media on the toxicity of
nanoparticles
The existing literature highlights the importance of the
growth media in cytotoxicity testing and links the apparent
cytotoxic effect of the nanomaterial to the salt and organic
content of the culture and related physicochemical characteristics
of the nanomaterial (Lyon et al., 2005; Tong et al.,
water research 44 (2010) 505–520
2007). Lyon et al. showed that in media with high salt
content, the size of nC60 aggregates tends to increase as
compared to that observed in low salt media (Lyon et al.,
2005). In another study, MWCNTs were found to stimulate
growth of unicellular protozoan Tetrahymena pyriformis in
growth medium, which contained proteose peptone, yeast
extract, and glucose; the increase in growth was attributed to
the formation of peptone-MWCNTs conjugates, which were
taken up by the microorganism (Zhu et al., 2006). While the
aforementioned studies indicate that salt and organic
composition of the medium, in which exposure studies are
performed, may influence the interaction of CNTs with
bacteria, there have been no studies that comparatively evaluated
CNT toxicity in growth media with different organic loadings and
salt contents.
1.5. Objectives of this study
This study addressed some of the knowledge gaps identified
above. Aqueous suspensions of SWCNTs functionalized by
a range of non-covalently bound dispersants of natural (NOM,
GA, amylose) and synthetic (PVP, Triton X-100) origin, were
prepared and evaluated in terms of their physicochemical and
toxicity properties. The study pursued the following
objectives:
(i) To evaluate the long-term stability and its physicochemical
determinants for non-covalently functionalized SWCNTs as
a function of dispersant type. The evolution of the concentration,
size, and charge of suspended SWCNTs was
studied over the period of 28 days.
(ii) To assess time-dependent cytotoxicity of non-covalently functionalized
SWCNTs with respect to bacteria and mammalian
cells as a function of dispersant type and growth media. The
effect of SWCNTs on E. coli was studied by measuring cell
viability after 3 h, 24 h, and 48 h of exposure in three
types of growth media with different organic loadings
and salt contents. The cytotoxicity of dispersed SWCNTs
towards mammalian cells was assessed in NDU bioassay
with rat liver epithelial cells after 30 min and 24 h of
incubation.
(iii) To assess time-dependent epigenetic toxicity of non-covalently
functionalized SWCNTs as a function of dispersant type. In this
study, we evaluated epigenetic toxicity of the SWCNTs to
rat liver epithelial cells as a function of dispersant type in
GJIC bioassay after 30 min and 24 h of incubation.
2. Approach
The set of chemically diverse dispersants was chosen based
on their demonstrated effectiveness in solubilizing CNTs
(Bandyopadhyaya et al., 2002; O’Connell et al., 2002; Star et al.,
2002; Islam et al., 2003; Moore et al., 2003; Hyung et al., 2007;
Liu et al., 2007) and potential adverse effects (Burnette, 1960;
Chourasia and Jain, 2004; Dayeh et al., 2004; Schmitt et al.,
2008). NOM, GA and PVP have LD50 doses of (54.8–58.5) mg
(intravenous administration in mice) (EMEA, 1999), 2,000 mg
(oral administration in rats) (Schmitt et al., 2008), and
100,000 mg (oral administration in rats) (Burnette, 1960) per kg

of body weight, respectively. Amylose has been reported to be
used in a colon-specific drug delivery due to its low toxicity
and high biodegradability (Chourasia and Jain, 2004). Triton
X-100 was reported to be toxic to protozoa, fish, and
mammalian cells (Dayeh et al., 2004).
Three batches of dispersed SWCNTs were prepared. The
first batch was used to comprehensively evaluate the longterm
stability of SWCNT suspensions over a period of 4 weeks
in terms of concentration, effective hydrodynamic diameter,
and z-potential of stabilized SWCNTs.
The second batch was used to evaluate the viability of E.
coli cells after their contact with dispersed SWCNTs by the
quantification of colony forming units. To elucidate the
effects of ionic and organic composition of the growth
medium, the cytotoxicity of the SWCNTs suspended in three
growth media of different organic and salt compositions
was evaluated. First, in order to assess the acute cytotoxicity
of dispersed SWCNTs, we conducted experiments in
0.1 M NaCl. Then, in order to investigate the effect of
SWCNTs on the ability of E. coli form colonies over time, we
employed three types of growth media: (i) LB medium with
higher organic chemical and salt content, (ii) MD medium
with low salt content and organic load, and (iii) 0.1 M NaCl
solution.
The third batch was used to evaluate cyto- and epigenetic
toxicity of the dispersed SWCNTs against rat liver epithelial
cells by the (i) NDU and (ii) GJIC bioassays. The NDU assesses
cell viability by measuring the accumulation of neutral red
dye in lysosomes, which depends on the cell’s capacity to
maintain pH gradients through the maintaining membrane
integrity and production of adenosine triphosphate. The
amount of dye incorporated by the cell is quantified spectrophotometrically
(Borenfreund and Puerner, 1985). The NDU
bioassay has been used to assess cytotoxicity of CNTs (Flahaut
et al., 2006). The GJIC bioassay (Borenfreund and Puerner,
1985; Weis et al., 1998; Herner et al., 2001; Satoh et al., 2003)
utilizes the ability of epigenetic tumor promoters to alter level
of GJIC (Yamasaki, 1990; Trosko et al., 1991). The degree of GJIC
was quantified by measuring the distance (area) the fluorescent
dye ( 1200 Da) travels (occupies) between the cells after
a given time. Both NDU and GJIC bioassays were carried out
with WB-F344 rat liver epithelial cells exposed to dispersed
SWCNTs for different periods of time. The WB-F344 cell line
was chosen because these normal diploid rat liver epithelial
cells have already used in numerous studies of cytotoxic or
epigenetic effects (e.g., Herner et al., 2001) thus allowing us to
have a comparative basis.
3. Materials and methods
3.1. CNTs
SWCNTs (purity > 90%), produced by catalytic chemical vapor
deposition, were obtained from Cheap Tubes, Inc (Brattleboro,
VT) and used as received. The SWCNTs were used as obtained,
allowing us to mimic what might occur in the environment.
As indicated by the manufacturer, the SWCNTs had an inside
diameter in the range of 0.8 nm to 1.6 nm, an outer diameter in
the range of 1 nm to 2 nm and were 5 mm to30mm in length.
water research 44 (2010) 505–520 509
3.2. Non-covalent functionalization: dispersants and
dispersion procedures
3.2.1. Dispersants
Gum arabic (approx. 250 kDa; a complex mixture of saccharides
and glycoproteins obtained from the acacia tree), PVP
(approx. 29 kDa), Triton X-100 (approx. 625 Da; polyethylene
glycol p-(1,1,3,3-tetramethylbutyl)-phenyl ether) and amylose
(molecular weight not specified; a polymeric form of glucose
and a constituent of potato starch) were purchased from
Sigma–Aldrich (Milwaukee, WI). SRNOM reverse osmosis
isolation was obtained from International Humic Substances
Society (St Paul, MN). The molecular structure of dispersants
used in this study is presented in Fig. 1.
3.2.2. Dispersion procedure
Aqueous solutions of PVP, Triton X-100 were prepared by
dissolving PVP (40 mg) and Triton X-100 (0.4 mL) in 40 mL of
water and adjusting the pH of both solutions to 7 with 0.1 M
HCl. The aqueous solution of GA was prepared by mixing 1 g of
GA in 50 mL of water and adjusting the pH of the mixture to 7
with 0.1 M HCl; this mixture was left to settle for 24 h and then
40 mL of supernatant was collected for use in the stability or
toxicity studies. To prepare NOM solutions, two flasks of
40 mL of water, each containing 8 mg of NOM, were stirred for
24 h. Each solution was then filtered through a 0.22 mm filter
under vacuum. The pH of one solution was kept at its original
value (approx. 3.5) while the pH in another solution was
adjusted to 7 with 0.1 M NaOH. SWCNTs were added to the
above solutions to result in a 1 mg (SWCNT)/mL loading and
were sonicated in Aquasonic 50 T water bath (VWR Scientific
Products Corp, West Chester, PA) for 20 min.
SWCNTs were dispersed with amylose based on modified
version of the three-step approach reported by Kim et al. (Kim
et al., 2004). SWCNTs (40 mg) were sonicated in 25 mL of water
(pH 7) at approximately 75 W for 5 min using Sonicator 3000
(Misonix, Inc., Farmingdale, NY) equipped with a microprobe.
100 mg of amylose in 6.28 mL of DMSO were prepared and
added to the sonicated suspension of SWCNT in water so that
the DMSO/water ratio was 20% by volume. The resulting
mixture was sonicated for another 5 min in order to remove
the excess amylose and DMSO. The suspension was sonicated,
centrifuged and refilled with water four times (See SD,
Section S.2.3).
Following the sonication, SWCNTs suspensions were
divided into 12 mL aliquots and each aliquot was transferred
into 15 mL centrifuge tube. After 24 h standing time, the top
4 mL of each suspension was withdrawn in order to be used in
the stability or toxicity studies. Overall, three batches of
dispersed SWCNT suspensions were prepared and were used,
correspondingly, for 1) the study of the long-term stability of
SWCNTs suspensions, 2) the general viability bioassay with
E. coli, and 3) NDU and GJIC bioassays.
3.3. Physicochemical characterization of SWCNT
suspensions
In order to quantitatively assess the long-term stability
SWCNT suspensions, a combination of characterization
methods was employed.

510
a
GALP
1
3
ARAF
1
3
3
ARAF
ARAF
1
1
3 6
1
GALP
3 1
GALP
3 6 1
GALP
3 1
GALP
6
6
3 1
3 1
RHAP GALP
RHAP GALP
6 6
1
1
GA
GA
4 4
1
ARAF
GALP
1
3
ARAF
1
1
ARAF
GALP = D-GALACTOPYRANOSE ARAF = L-ARABOFURANOSE
GA = D-GLUCURONIC ACID RHAP = L-RHAMNOPYRANOSE
b
d
C 8 H 17
3.3.1. UV–vis spectrophotometry. Quantification of SWCNTs
concentration in suspension
The absorbance of SWCNT suspensions over the (200–800) nm
wavelength range was measured over a period of 4 weeks
(Multi-spec 1501, Shimadzu, Kyoto, Japan). For each type of
SWCNT suspension prepared, the absorbance of SWCNT-free
solution of the corresponding dispersant was used as
a baseline.
SWCNT suspensions used in our toxicity studies likely
contained both individual and bundled SWCNTs. In fact, DLS
measurements indirectly confirmed (see Section 4.2) the
presence of dispersed SWCNT bundles and direct TEM
imaging (see SD, Section S.3.3, Fig. S5) also showed the presence
of SWCNTs bundles (note that TEM results need to be
interpreted with caution as evaporation-induced aggregation
could have contributed to bundling during TEM sample
preparation). Incomplete exfoliation is the most likely and
most environmentally relevant scenario; unfortunately, it also
N
O
n
O
water research 44 (2010) 505–520
OH
n
OH
OH
H 2 COH
HO
O
N
H 2
(a-1)
CH 2 OH
OH
O
(a-3)
O
OH
O OH
HOOC
HO
OH
HO
COOH
COH
OH
(a-2)
CH 3
OH OH
HOOC
(a-4)
entails difficulties with quantifying the total concentration of
suspended nanotubes as larger CNT bundles tend to separate
from the suspension.
In this study, we used UV–vis spectrophotometry to estimate
the concentration of exfoliated SWCNTs in suspensions.
It is known that only fully exfoliated SWCNTs absorb in the
(200–1200) nm wavelength range. Bundled NTs do not absorb
significantly in this range due to the tunnel coupling between
metallic and semiconductive SWNTs (Lauret et al., 2004;
Grossiord et al., 2007). Therefore, the UV absorbance by
a SWCNT suspension can be used to selectively measure the
concentration of individually dispersed (i.e. fully exfoliated or
debundled) SWCNTs.
To determine the concentration of suspended exfoliated
SWCNTs, we first prepared a separate set of suspensions with
ten-fold reduced SWCNT loading with respect to the CNT
loading used in long-term stability and toxicity studies. Four
suspensions (with PVP, NOM pH 3.5, NOM pH 7, and GA used
O
O
OH
OH
OH OH
O O
OH
O
HO
OH
O O
Fig. 1 – Molecular structure of solubilizers: (a) Gum arabic (a-1) galactose, (a-2) rhamnose, (a-3) arabinose, (a-4) glucuonic
acid; (b) Poly(vinyl pyrrolidone); (c) A building block of humic acids, which has a compositional similarity to SRNOM; Dots
represent chiral centers; (d) Triton X-100; (e) Amylose.
c
e
n
OH
COOH
OH
OH

as dispersants) with 10-fold reduced SWCNT content (0.1 g/L)
were prepared using the same procedures as described in
Section 3.2 except that they were subjected to rigorous, prolonged
sonication for 1 h at power of (70–80) W using horn
sonicator (Sonicator 3000, Misonix, Inc., Farmingdale, NY). No
settling of SWCNTs was observed over the short term
following the application of this treatment indicating
complete dispersion of SWCNTs. UV–vis absorbance spectra
were recorded at different dilutions and a calibration curve for
absorption at 500 nm (Bahr et al., 2001; Huang et al., 2002;
Sinani et al., 2005; Lee et al., 2007; Salzmann et al., 2007) as
a function of concentration was constructed, and coefficient
of molecular extinction, 3, was determined. This coefficient
was used to estimate the concentration of exfoliated SWCNTs
in suspensions used in long-term stability and toxicity
experiments. Note that by subjecting suspended SWCNTs to
a very intense sonication treatment we aimed at maximizing
the extent of exfoliation; however, it was not possible to
ascertain that the exfoliation was complete. Thus coefficients
of molecular extinction and exfoliated SWCNT concentrations
determined using the recorded calibration curves were estimated
values. SWCNT characteristics measured upon the
preparation of CNT suspensions are given in Table S1 in SD.
3.3.2. Size and charge of dispersed SWCNTs
The effective hydrodynamic diameter and z-potential of suspended
SWCNTs were determined after 1, 4, 7, 14, 21, and 28
days of settling. The size and charge were measured by
dynamic light scattering (DLS) and phase analysis light scattering
techniques, respectively (ZetaPALS, BI_MAS Option,
Brookhaven Instrument Corp., Holtsville, NY). The Smoluchowski
equation was applied to convert the measured
electrophoretic mobility of dispersed SWCNT to z-potential.
Transmission electron microscopy (TEM) imaging was used as
an auxiliary method to aid in the interpretation of dynamic
light scattering data (see SD, Sections S.2.5 and S.3.4).
3.4. Toxicity assessment: E. coli viability assay
3.4.1. Media preparation
Luria–Bertani (LB) growth medium was prepared according to
the standard procedure (Atlas, 1993). Minimal Davis medium
with 90% reduced potassium phosphate concentration
(MD medium) was prepared (Fortner et al., 2005). 0.1 M NaCl
solution was prepared by dissolving 9 g of NaCl in 1 L of water
(pH 7) and autoclaving it for 15 min at 1 bar and 121 C. For the
toxicity assessments, each component of LB, MD and NaCl
media was prepared in 4-fold higher concentration as
compared to original protocol and the aliquot part of the corresponding
medium was added to SWCNTs suspension
(as further described in ‘‘Quantification of cell viability’’
Section). Luria–Bertani Petri plates were prepared according to
the published method (Atlas, 1993).
3.4.2. Preparation of E. coli culture
E. coli K12 stock was prepared in glycerol and stored at 80 C.
Prior to use, the stock was defrosted, and 30 mL of LB medium
were inoculated with 5 mL of the stock. After overnight growth
at 37 C, 5 mL of this suspension was spread onto LB agar plate
and cultured at 37 C. Once distinct colonies were formed, the
water research 44 (2010) 505–520 511
agar plate was transferred to the refrigerator and kept at 4 C
for up to one month. E. coli suspensions to be used in SWCNT
cytotoxicity studies were prepared by scraping one colony
from the surface of a Petri plate by aseptic loop and immersing
the loop into 10 mL of LB or MD media in a 50 mL centrifuge
tube. Tubes were placed on a shaker in an incubator 37 C for
12 h. When 0.1 M NaCl was used as an exposure medium in
colony forming units bioassay, E. coli were first grown in the LB
medium, centrifuged for 5 min at 2250 rpm and washed with
0.1 M NaCl as follows: the supernatant was decanted and
replaced with an equal volume of 0.1 M NaCl, vortexed and
resuspended by centrifugation. The washing procedure was
repeated twice, presuming that after this treatment most of
the remaining organic constituents of LB medium were
removed from the E. coli suspension.
3.4.3. Quantification of cell viability
The SWCNTs suspension (1.425 mL), growth medium (475 mL)
and 100 mL of the stock E. coli suspension were transferred into
a 15 mL centrifuge tube and incubated under gentle shaking at
37 C. Samples were taken after 3, 24, and 48 h and a series of
dilutions (10 4 –10 6 ) was prepared for each sample. Five samples
of 10 mL and 20 mL from each dilution were placed onto an agar
plate and incubated at 37 C until distinct colonies developed.
Colony forming units (CFU)/1 mL were calculated for each
sample. Each experiment was run in triplicates with negative
(bacterial suspension with the corresponding amount of
ultrapure water) and vehicle (solution of the corresponding
dispersant) controls, herein called vehicle control I and
vehicle control II, respectively. The results are reported as
a fraction of control (FOC) standard deviation (STD), calculated
as the ratio of the average number of colonies grown
after E. coli exposure to SWCNTs suspensions or vehicle
control II to the average number of colonies grown in vehicle
control I plates.
3.5. Toxicity assessment: neutral red dye uptake
bioassay
For the NDU bioassay we adapted a published procedure
(Borenfreund and Puerner, 1985; Weis et al., 1998; Satoh et al.,
2003). A solution (0.015 w/v) of neutral red dye (3-amino-7-
(dimethylamino)-2-methylphenazine hydrochloride) in
D-medium was incubated at 37 C for 2 h and filtered through
a 0.22 mm syringe filter (Millipore Corp., New Bedford, MA) to
remove undissolved dye and ensure sterile conditions.
Confluent WB-F344 cells (see SD, Section S.2.6 for details on the
preparation of the cells) were exposed to 500 mL of each of
SWCNTs suspension and incubated at 37 C for 30 min and 24 h
under gentle shaking in a humidified atmosphere containing
5% CO 2. After cells were exposed to SWCNTs, the exposure
medium was removed by aspiration and the cells were washed
with 1 mL of phosphate saline buffer (PBS). Following washing,
2 mL of the neutral red dye solution per plate was added and
the cells were incubated for 1 h at 37 C in the humidified
atmosphere containing 5% CO2. Upon incubation, the cells
were rinsed three times with PBS, and 2 mL of aqueous solution
containing 1% acetic acid and 50% ethanol was added to each
plate to lyse the cells. 1.5 mL of the lysate was transported into
2 mL microcentrifuge tube and the optical density was

512
recorded at 540 nm using a Beckman DU 7400 diode array
detector (Beckman Instruments, Inc., Schaumburg, IL). The
background absorbance was measured at 690 nm and then
subtracted from the original absorbance. Each experiment was
conducted in triplicates. The neutral red dye uptake was
reported as the FOC (absorption of neutral red in the chemically
treated sample divided by the absorption of neutral red in the
nontreated control I sample). FOC values of 1.0 indicates noncytotoxic
response while FOC values

a b
Concentration (mg/L)
c
-potential (mV)
200
160
120
80
40
0
0
0 7 14 21 28 0 7 14 21 28
Time (d)
Time (d)
0
-15
-30
-45
Triton X-100 stabilizes nanotubes by the formation of hemimicelles
that cover nanotube surface with benzene rings
providing p–p stacking between the surfactant molecule and
nanotube core (Islam et al., 2003). The exfoliation (debundling)
of SWCNTs in the presence of SRNOM was suggested to occur
through the interaction of the aromatic moieties of natural
organic matter and nanotube surface (Hyung et al., 2007; Liu
et al., 2007; Hyung and Kim, 2008). Despite the differences in
the physiosorption mechanisms responsible for the stabilization
of CNTs in aqueous suspensions, all dispersants were
effective, albeit to somewhat different extents.
For all SWCNT suspensions, the calculated values of the
concentration of exfoliated SWCNTs correlated well to the
visual observations described above. As seen from Fig. 2a, the
dispersion efficiency in terms of the concentration of
dispersed SWCNTs was a function of the type of the dispersant,
with GA producing suspensions having highest
concentration of SWCNTs. The data were consistent among
the three prepared batches of SWCNTs (see SD, Table S1).
Didenko et al. (Didenko et al., 2005) suggested that after
covering one single carbon nanotube with a long polymeric
molecule, the remaining strands would react with other
uncovered or partially covered SWCNTs thus bundling several
nanotubes together. We speculate that this mechanism where
water research 44 (2010) 505–520 513
Effective diameter (nm)
-60
-45
0 7 14 21 28 0 7 14 21 28
Time (d)
Time (d)
Fig. 2 – The time-dependent characterization of (a) estimated concentration of the exfoliated SWCNTs in the dispersion,
(b) effective diameter and (c) z-potential of dispersed SWCNTs, and (d) z-potential of dispersants alone (-B- for GA/SWCNTs,
-,- for PVP/SWCNTs, -6- for NOM3.5/SWCNTs, -:- for NOM7/SWCNTs, --- for Triton X-100/SWCNTs, -C- for Amylose/
SWCNTs). Notes: (1) The z-potential measurement was not conducted for amylose solution, because amylose is not watersoluble
under room temperature. (2) Reported are results of only those measurements that were statistically significant, i.e.
when sufficiently high photon count rates were recorded in dynamic light scattering measurements. (3) The absorption by
suspensions of amylose-stabilized SWCNTs was not measured because the nanotube content in these suspensions could
not be precisely determined; this was due to the incomplete separation at the centrifugation step of the suspension
preparation process.
d
-potential (mV)
1000
800
600
400
200
0
-15
-30
one polymer molecule links two or more nanotubes may be
responsible for the higher dispersing efficiency of GA (by far
the largest molecule among the target dispersants) towards
SWCNTs compared to other dispersants. This hypothesis is
supported by the measurements of the effective hydrodynamic
diameter (Fig. 2b), and the estimation of the length of
suspended particles with the size and the length of GA/
SWCNTs being higher than those of PVP, Triton X-100 and
SRNOM (both pHs). As the measured diameter of GA-stabilized
SWCNTs aggregates (which could also be expected to be
rather porous) was still relatively small, gravity settling did
not lead to a significant removal of GA-stabilized SWCNTs
from the dispersion.
When comparing the solubilizing ability of SRNOM at
different pH, one could see that SRNOM is a more efficient
dispersant at pH 3.5 than at pH 7. This is consistent with the
findings by Huyng et al. (Hyung and Kim, 2008) who reported
that adsorption of SRNOM to MWCNTs increased when pH
decreased due to more compact and coiled conformation of
NOM at acidic pHs. When pH increases, carboxylic and
phenolic groups of SRNOM deprotonate resulting in higher
electrostatic repulsion between a CNT and a SRNOM molecule
and, as a consequence, in a lower amount of organic matter
adsorbed on the CNT surface. The better dispersion of

514
SWCNTs at pH 3.5 could also be attributed, in part, to the steric
hindrance imposed by SRNOM when a higher surface density
of NOM on the surface of CNT bundles results in stronger
repulsion between CNTs. This observation is supported by the
long-term z-potential measurements (Fig. 2c), where the
surface charge of stabilized SRNOM/SWCNTs at pH 3.5
became less negative up to day 7, and then stabilized with
increasing settling time. At the same time the surface charge
of SWCNTs dispersed in SRNOM solution at pH 7 gradually
increased after day 5. Indeed, had the electrostatic repulsion
been the sole mechanism of SWCNTs stabilization in SRNOM
solutions, the surface charge on the SWCNTs would have
more rapidly become less negative at pH 3.5 than at pH 7. In
summary, in terms of the effectiveness of SWCNT stabilization,
the dispersants were ranked as follows: GA > Triton
X-100 > PVP > NOM (pH 3.5) > NOM (pH 7).
4.2. Hydrodynamic size of dispersed SWCNTs
The hydrodynamic size of CNTs in a suspension is an
important characteristic that affects the stability and,
possibly, toxicity of dispersed CNTs. It should be noted that
the effective hydrodynamic diameter of suspended SWCNTs
measured using DLS can only be used as a very rough estimate.
While the DLS data are interpreted with the assumption
that the primary scatterers are spherical with an aspect ratio
of 1, dispersed SWCNTs are long and tubular, with a very high
aspect ratio (see SD, Fig. S3). To obtain a better approximation
of the size distribution of dispersed SWCNT, the multimodal
size distribution model was used. TEM imaging was employed
to obtain auxiliary information on the size and morphology of
stabilized SWCNT. Even though the measurement of effective
hydrodynamic diameter cannot be relied on to compare the
sizes of suspended SWCNTs in different dispersion media,
these measurements can be used to compare how the average
particle size for a given dispersant changes with settling time.
The values of effective diameter as a function of time for
different SWCNT suspensions are given in Fig. 2b.
Generally, the effective size of SWCNTs dispersed using
Triton X-100, GA and NOM did not change significantly with
increasing settling time (Fig. 2b). In the case of amylosedispersed
SWCNTs, a slight decrease in the effective size was
observed due to settling of larger and unstable SWCNT
aggregates, which left behind more uniformly sized smaller
amylose-stabilized CNT clusters. The concentration of
SWCNTs in the suspension was negatively correlated with the
effective size of SWCNT. The notable apparent exception was
the suspension of SWCNTs dispersed using GA. However, it
should be noted that the aqueous suspension of GA
contained GA colloids of the size that was 1) comparable to the
size of dispersed SWCNTs and 2) decreased during the 4 weeks
of settling (see SD, Fig S5). Thus, the effective size measurements
for SWCNTs dispersed using GA should be interpreted
with caution.
4.3. Charge of dispersed SWCNTs
For all suspensions, the z-potential of dispersed SWCNTs was
negative and nearly constant over the entire duration of the
experiment (Fig. 2c and SD, Table S1). Each z-potential
water research 44 (2010) 505–520
a
Fraction of control (FOC)
b
Fraction of control (FOC)
c
Fraction of control (FOC)
1.2
1
0.8
0.6
0.4
0.2
0
1.2
1
0.8
0.6
0.4
0.2
0
1.2
1
0.8
0.6
0.4
0.2
0
NOM 3.5 NOM 7 GA Amylose PVP Triton
NOM 3.5 NOM 7 GA Amylose PVP Triton
NOM 3.5
NOM 7 GA Amylose PVP Triton
Fig. 3 – Viability of E. coli in (a) 0.1 M NaCl, (b) LB medium,
and (c) MD medium. Exposure time: (a) 3 h, (b) 48 h, (c) 48 h.
Open, hatched and cross-hatched bars correspond to
control I, control II and dispersed SWCNTs. In the case of
amylose ultrapure water was used as control.
measurement for SWCNT suspensions was accompanied by
a measurement of the z-potential of the dispersants in
a SWCNT-free aqueous solution (Fig. 2d). After 1 d, only the GA
solution scattered light sufficiently to enable a statistically
meaningful z-potential measurement. After 4 d, solutions of
NOM (pH 7), PVP, and Triton X-100 also scattered light

a 1.2
b
Fraction of control (FOC)
1
0.8
0.6
0.4
0.2
0
NOM3.5 NOM7 GA Amylose PVP Triton
sufficiently. After 7 d, all the dispersants except for NOM (pH
3.5) achieved acceptable count rates during z-potential
measurements. The gradual change in the scattering ability of
the dispersant solutions may correspond to the aggregation of
the dispersant molecules. The dispersants were ranked in
order from the most to the least negative z-potential of
dispersed SWCNTs: NOM (pH 7) > NOM (pH 3.5) > GA z Triton
X-100 > PVP z amylose. The highly negative charge of NOMstabilized
SWCNTs was likely due to the charge of NOM. The
low stability of amylose-dispersed SWCNTs could be attributed
to the combination of low surface charge and steric
repulsion between stabilized SWCNTs in the dispersion.
4.4. Assessment of cell toxicity in prokaryotic and
eukaryotic systems
4.4.1. General viability bioassay with E. coli (prokaryotic)
Immediately after E. coli cells were exposed to SWCNTs suspended
in growth medium as well as after 3 h of exposure, no
SWCNT aggregation was visually observed (see SD, Table S2)
in experiments with all three types of the media – 0.1 M NaCl,
MD (medium with lower salt organics content) and LB
(medium with higher salt and organics content). After 24 h of
incubation, limited precipitation was observed for all SWCNT
water research 44 (2010) 505–520 515
Fraction of control (FOC)
1.2
1
0.8
0.6
0.4
0.2
0
NOM3.5 NOM7 GA Amylose PVP Triton
Fig. 4 – Neutral red dye uptake by WB-F344 cells as a function of dispersant type (a – 30 min, b – 24 h of exposure). Open,
hatched and cross-hatched bars correspond to control I, control II and dispersed SWCNTs. In the case of amylose control I
and control II were ultrapure water.
suspensions in LB and MD media. After 48 h of incubation
more precipitation occurred in each type of media.
No inhibition of E. coli colony forming ability was observed
after 3 h of incubation with amylose-, NOM-, GA- and PVPstabilized
SWCNTs in 0.1 M NaCl (Fig. 3a). The FOC values did
not decline in any of these samples and no significant difference
in the influence on CFU counts was found between
dispersed SWCNTs and solutions of corresponding dispersants.
When the bacteria suspension was brought in contact
with Triton X-100-stabilized SWCNTs, approx. 25% loss of the
cell viability was measured as compared to vehicle control I
samples. However, there was no statistical difference between
FOC of Triton-stabilized SWCNTs and the control (solution of
Triton X-100 only). It remains unclear if the observed cytotoxicity
of the these suspensions was due to i) SWCNTs
stabilized by Triton X-100 or ii) the residual ‘‘free’’ (i.e. not
associated with suspended SWCNTs) Triton X-100 potentially
present in the solution or iii) the combined effect of both
Triton X-100/SWCNT and dissolved Triton X-100. The
complete separation of the Triton X-100/SWCNT and
dissolved Triton X-100 could not be accomplished using
centrifugation. Even for very long centrifugation times, the
supernatant had grayish color indicating that some fraction of
SWCNTs was not removed.
Fig. 5 – Representative phase contrast images of WB-F344 cells incubated with 500 mLofH 2O (control I), 500 mL of Triton X-100
(control II) and 500 mL of Triton X-100-stabilized SWCNTs. Black dots observed in Control II and Triton X-100-solubilized
SWCNTs samples correspond to dead WB-F344 cells. All images were taken at 2003 magnification. Scale bar is 50 mm.

516
Fig. 6 – Representative phase contrast (upper row) and UV epifluorescent (bottom row) images of WB-F344 cells incubated
with 500 mLofH2O (control I), 500 mL of NOM pH 3.5 (control II) and 500 mL of NOM-stabilized SWCNTs (pH 4). All images were
taken at 2003 magnification. Bright dots correspond to cells that absorb the dye; the absorption is indicative of cellular
health. All images were taken at 2003 magnification. Scale bar is 50 mm.
The ability of E. coli to grow and to form colonies in the
presence of amylose-, GA-, PVP-, and NOM-stabilized SWCNTs
in LB medium mimicked both control samples regardless of
the contact time (see SD, Fig. 3b and Fig. S6). On the contrary,
21 and 18 % mortality rates after 3 h of incubation were
observed for E. coli in Triton X-100/SWCNTs suspension and
Triton X-100 solution, respectively. After 24 h of contact, the
number of colonies grown on the Petri plates decreased by
30 % when bacteria were in contact with Triton X-100-stabilized
SWCNTs and by 27 % when bacteria were in Triton X-100
only solution (see SD, Fig. S6) as compared to vehicle control I
a 1.2
b
Fraction of control (FOC)
1
0.8
0.6
0.4
0.2
0
NOM 3.5 NOM 7 GA Amylose PVP
water research 44 (2010) 505–520
plates. As the exposure time increased to 48 h, E. coli resumed
its growth; with FOC of both Triton X-100 suspensions
approaching values for vehicle control I sample (Fig. 3b). When
dispersed SWCNTs were tested in MD medium, no losses in
cell viability for amylose-, GA-, PVP- and NOM-stabilized
SWCNTs were measured (Fig. 3c; also see SD, Fig. S7). In the
case of Triton X-100 suspensions, the reduction in E. coli
survival was observed after 3 h of exposure; comparable
losses of the E. coli viability were also observed for both Triton
X-100-stabilized SWCNTs and the Triton X-100 only solution.
However, after 24 h of exposure, fewer E. coli colonies were
Fraction of control (FOC)
1.2
1
0.8
0.6
0.4
0.2
0
NOM 3.5 NOM 7 GA Amylose PVP
Fig. 7 – GJIC in WB-F344 cells as a function of dispersant type (a – 30 min, b – 24 h of exposure). Open, hatched and
cross-hatched bars correspond to control I, control II and dispersed SWCNTs. In the case of amylose control I and control II
were ultrapure water.

observed on Petri plates from suspensions of Triton X-100stabilized
SWCNTs and Triton X-100 solution only (vehicle
control II) (33 and 44 %, respectively) (see SD, Fig. S7). As the
incubation time increased to 48 h, bacterial CFU counts in the
presence of both Triton X-100 samples increased and
approached that of vehicle control I samples (Fig. 3c).
As was previously mentioned, the exposure of Triton
X-100-stabilized SWCNTs to MD medium for 24 h resulted in
formation of large flake-like aggregates, which settled out of
the suspension. It is possible that this low salt medium
induced specific interactions between bacterial cells and
Triton X-100-coated SWCNTs, which were suppressed in the
medium having a higher ionic strength. This led to a larger
decrease in E. coli viability in these samples in comparison
with the losses in Triton X-100 solution only. Damaged or dead
cell attached to the SWCNTs aggregates and formed debris at
the bottom of the testing tubes. The simultaneous effect of
bacterial and nanoparticles settling out of the solution has
been previously described (Tang et al., 2007) where the
response of Shewanella oneidensis to C 60–NH 2 nanoparticles
was studied.
The fact that we observed inhibition of E. coli viability after
24 h of exposure in MD medium and did not observed this
effect in LB medium highlights the importance of growth
medium in cytotoxicity testing. Similarly, E. coli exposure to
SWCNTs, stabilized with amylose, GA, PVP and NOM (both pH)
did not affect CFU counts in any of the three media pointing to
the importance of the dispersant in the assessments of
biocompatibility of CNTs.
4.4.2. NDU cytotoxicity bioassay on eukaryotic cells
The assessment of cytotoxicity was determined by the
viable uptake of neutral red into F344-WB rat liver epithelial
cells. Nonviable cells lack the ability to absorb this dye.
Results (Fig. 4) indicated that SWCNTs dispersed using NOM
(pH 3.5), NOM (pH 7), GA, amylose and PVP were not
cytotoxic. However, Triton X-100 detergent induced
a significant cytotoxic effect independent of SWCNTs. This
latter observation can also be seen under phase contrast
microscopy of cells treated with Triton X-100 for 30 min
(Fig. 5). The morphology of the cells drastically changed
from that of the normal control cells, in which the cells
became clearer due to loss of most cytosolic contents,
except for the cell nuclei. After a 24 h treatment with Triton
X-100, the cells completely detached or solubilized from the
culture plates (data not shown).
4.5. Assessment of epigenetic toxicity on eukaryotic cells:
GJIC bioassay
A representative fluorescent micrograph of the GJIC assay
after treatment with (NOM þ SWCNTs) is shown in Fig. 6.
The migration of the fluorescent dye through several cell
layers is an indicator that the gap junction channels are
open. After WB-F344 cells were incubated with dispersed
SWCNTs suspensions for (1) 30 min and (2) 24 h, there was
no significant affect on GJIC for all suspensions tested
(Fig. 7). Due to Triton X-100-stabilized SWCNTs being
cytotoxic, GJIC was not measured in cells exposed to this
mixture. Also, when cells were examined under phase
water research 44 (2010) 505–520 517
contrast, no changes in size and shape of the individual
cells were detected (Fig. 6). Hence, it can be concluded that
under these experimental conditions WB-F344 cells exposed
to dispersed SWCNTs retain normal intercellular communication
irrespective of the applied treatment.
5. Conclusions
The stability of aqueous suspensions of SWCNTs non-covalently
functionalized by a range of natural (gum arabic,
amylose, Suwannee River natural organic matter) and
synthetic (polyvinyl pyrrolidone, Triton X-100) dispersants
that bind to SWCNT surface via different physiosorption
mechanisms was evaluated. Despite the differences, all
SWCNT suspensions remained relatively stable over a period
of 4 weeks as indicated by the evolution of the concentration
of suspended exfoliated SWCNT and by the fact that both the
average effective hydrodynamic diameter and z-potential of
suspended SWCNTs were relatively constant. The epigenetic
toxicity of dispersed SWCNTs was evaluated for the first time
using a gap junctional intercellular communication assay
with WB-F344 rat liver epithelial cells resulting in no inhibition
by SWCNTs non-covalently functionalized using GA, PVP,
amylose, and SRNOM. There were no cytotoxic effects of these
SWCNTs suspensions on either the prokaryotic, bacterial (E.
coli), or eukaryotic (WB-F344 rat liver epithelial) cell types.
Only when the dispersant itself was toxic, were losses of cell
viability observed. Bacterial CFU counts or neutral red
uptake was affected only by SWCNTs dispersed using Triton
X-100, which showed cytotoxicity in the SWCNT-free
solution (vehicle control II). The results suggest a strong
dependence of the toxicity of SWCNT suspensions on the
toxicity of the solubilizing agent and point to the potential of
non-covalent functionalization with non-toxic dispersants as
a method for the preparation of aqueous suspensions of
biocompatible CNTs.
Acknowledgements
This work was supported by the National Science Foundation
research grants CBET-0608320, CBET-056828 and OISE-
0530174, by the National Institute of Environmental Health
Sciences grants R01 ES013268-01A2 and Grant-in-Aid for
Science Research. We thank Michael Housinger for his assistance
with the preparation of SWCNT suspensions. ARR
acknowledges the support by Michigan AWWA Fellowship for
Water Quality and Treatment Study. The contents of this work
are solely the responsibility of the authors and do not necessarily
represent the official views of the National Institute of
Environmental Health Sciences.
Appendix.
Supplementary data
Supplementary data associated with this article can be found
in the online version, at doi: doi:10.1016/j.watres.2009.09.042.

518
references
Atlas, R.M., 1993. Handbook of Microbiological Media. CRC Press,
Boka Raton.
Bahr, J.L., Mickelson, E.D., Bronikovski, M.J., Smalley, R.E., Tour, J.M.,
2001. Dissolution of small diameter single-wall carbon
nanotubes in organic solvents. Chemical Communications,
193–194.
Bandyopadhyaya, R., Nativ-Roth, E., Regev, O., Yerushalmi-
Rozen, R., 2002. Stabilization of individual carbon nanotubes
in aqueous solutions. Nano Letters 2 (1), 25–28.
Bardi, G., Tognini, P., Ciofani, G., Raffa, V., Costa, M.,
Pizzorusso, T., 2009. Pluronic-coated carbon nanotubes do
not induce degeneration of cortical neurons in vivo and in
vitro. Nanomedicine: Nanotechnology, Biology, and Medicine
5, 96–104.
Bonnet, P., Albertini, D., Bizot, H., Bernard, A., Chauvet, O., 2007.
Amylose/SWNT composites: from solution to film – synthesis,
characterization and properties. Composite Science and
Technology 67, 817–821.
Borenfreund, E., Puerner, J.A., 1985. Toxicology determined in
vitro by morphological alterations and neutral red absorption.
Toxicology Letters 24, 119–124.
Burnette, L.W., 1960. A review of the physiological properties of
PVP. Proceedings of the Scientific Section of the Toilet Goods
Association 38, 1–4.
Chappell, M.A., George, A.J., Dontsova, K.M., Porter, B.E., Price, C.L.,
Zhou, P., Morikawa, E., Kennedy, A.J., Steevens, J.A., 2009.
Surfactive stabilization of multi-walled carbon nanotube
dispersions with dissolved humic substances. Environmental
Pollution 157, 1081–1087.
Chen, X., Tam, U.C., Czlapinski, J.L., Lee, G.S., Rabuka, D., Zettl, A.,
Bertozzi, C.R., 2006. Interfacing carbon nanotubes with living
cells. Journal of American Chemical Society 128, 6292–6293.
Cherukuri, P., Gannon, C.J., Leeuw, T.K., Schmidt, H.K.,
Smalley, R.E., Curley, S.A., Weisman, R.B., 2006. Mammalian
pharmacokinetics of carbon nanotubes using intrinsic nearinfrared
fluorescence. Proceedings of the National Academy of
Sciences 103, 18882–18886.
Chin, S.F., Baughman, R.H., Dalton, A.B., Diekmann, G.R.,
Draper, R.K., Mikoryak, C., Musselman, I.H., Poenitzsch, V.Z.,
Xie, H., Pantano, P., 2007. Amphiphilic helical
peptide enhances the uptake of single-walled carbon
nanotubes by living cells. Experimental Biological Medicine
232 (9), 1236–1244.
Chourasia, M.K., Jain, S.K., 2004. Polysaccharides for colon
targeted drug delivery. Drug Delivery 11 (2), 129–148.
Dayeh, V.R., Chow, S.L., Schirmer, K., Lynn, D.H., Bols, N.C.,
2004. Evaluating the toxicity of Triton X-100 to protozoan,
fish, and mammalian cells using fluorescent dyes as
indicators of cell viability. Ecotoxicology and Environmental
Safety 57 (3), 375–382.
Di Sotto, A., Chiaretti, M., Carru, G.A., Bellucci, S., Mazzantia, G.,
2009. Multi-walled carbon nanotubes: lack of mutagenic
activity in the bacterial reverse mutation assay. Toxicology
Letters 184, 192–197.
Didenko, V.V., Moore, V.C., Baskin, D.S., Smalley, R.E., 2005.
Visualization of individual single-walled carbon nanotubes by
fluorescent polymer wrapping. Nano Letters 5 (8), 1563–1567.
Ding, L.H., Stilwell, J., Zhang, T.T., Elboudware, J.O., Jiang, H.J.,
Selegue, J.P., Cooke, P.A., Gray, J.W., Chen, F.F., 2005. Molecular
characterization of the cytotoxic mechanism of multiwall
carbon nanotubes and nanoonions on human skin fibroblast.
Nano Letters 5, 2448–2464.
Dodziuk, H., Ejchart, A., Anczewski, W., Ueda, H., Krinichnaya, E.,
Dolgonosa, G., Kutner, W., 2003. Water solubilization,
water research 44 (2010) 505–520
determination of the number of different types of single-wall
carbon nanotubes and their partial separation with respect to
diameters by complexation with h-cyclodextrin. Chemical
Communications, 986–987.
El-Fouly, M.H., Trosko, J.E., Chang, C.C., 1987. Scrape-loading and
dye transfer: a rapid and simple technique to study gap
junctional intercellular communications. Experimental Cell
Research 168, 422–430.
Elias, A.L., Carrero-Sanchez, J.C., Terrones, H., Endo, M.,
Laclette, H.P., Mauricio, M., 2007. Viability studies of pure
carbon- and nitrogen-doped nanotubes with Entamoeba
Histolytica: from amoebicidal to biocompatible structures.
Small 3 (10), 1723–1729.
Enyashin, A.N., Gemming, S., Seifert, G., 2007. DNA-wrapped
carbon nanotubes. Nanotechnology 18, 245702.
Fiorito, S., Serafino, A., Andreola, F., Bernier, P., 2006. Effects of
fullerenes and single-wall carbone nanotubes on murine and
human macrophages. Carbon 44, 1100–1105.
Flahaut, E., Durrieu, M.C., Remy-Zolghadri, M., Bareille, R.,
Baquey, C., 2006. Investigation of the cytotoxicity of CCVD
carbon nanotubes towards human umbilical vein endothelial
cells. Carbon 44 (6), 1093–1099.
Fortner, J.D., Lyon, D.Y., Sayes, C.M., Boyd, A.M., Falkner, J.C.,
Hotze, E.M., Alemany, L.B., Tao, Y.J., Guo, W., Ausman, K.D.,
Colvin, V.L., Hughes, J.B., 2005. C 60 in water: nanocrystal
formation and microbial response. Environmental Science
and Technology 39 (11), 4307–4316.
Georgakilas, V., Tagmatarchis, N., Pantarotto, D., Bianco, A.,
Briand, J.-P., Prato, M., 2002. Amino acid functionalisation of
water soluble carbon nanotubes. Chemical Communications,
3050–3051.
Grossiord, N., Schhoo, P., Meuldijk, J., Koning, C.E., 2007.
Determination of the surface coverage of exfoliated carbon
nanotubes by surfactant molecules in aqueous solution.
Langmuir 23, 3646–3653.
Gutierrez, M.C., Garcia-Carvajal, Z.Y., Hortiguela, M.J., Yuste, L.,
Rojo, F., Ferrer, M.L., Monte, F., 2007. Biocompatible MWCNT
scaffolds for immobilization and proliferation of E. coli. Journal
of Materials Chemistry 17, 2992–2995.
Herner, H.A., Trosko, J.E., Masten, S.J., 2001. The epigenetic
toxicity of pyrene and related ozonation byproducts
containing an aldehyde functional group. Environmental
Science and Technology 35 (17), 3576–3583.
Herzog, E., Byrne, H.J., Davoren, M., Casey, A., Duschl, A.,
Oostingh, G.J., 2009. Dispersion medium modulates oxidative
stress response of human lung epithelial cells upon exposure
to carbon nanomaterial samples. Toxicology and Applied
Pharmacology 236, 276–281.
Huang, W.J., Taylor, S., Fu, K.F., Lin, Y., Zhang, D.H., Hanks, T.W.,
Rao, A.M., Sun, Y.P., 2002. Attaching proteins to carbon
nanotubes via diimide-activated amidation. Nano Letters 2 (4),
311–314.
Humic acids and their sodium salts. EMEA-The European agency
for the evaluation of medicinal products. Available via. http://
www.emea.europa.eu/pdfs/vet/mrls/055499en.pdf, 1999.
Hyung, H., Fortner, J.D., Hughes, J.B., Kim, J.H., 2007. Natural organic
matter stabilizes carbon nanotubes in the aqueous phase.
Environmental Science and Technology 41 (1), 179–184.
Hyung, H., Kim, J.-H., 2008. Natural organic matter (NOM)
adsorption to multi-walled carbon nanotubes: effect of NOM
characteristics and water quality parameters. Environmental
Science and Technology 42 (12), 4416–4421.
Iijima, S., 1991. Helical microtubules of graphitic carbon. Nature
354 (6348), 56–58.
Inoue, K.-i., Koike, E., Yanagisawa, R., Hirano, S., Nishikawa, M.,
Takano, H., 2009. Effects of multi-walled carbon nanotubes on
a murine allergic airway inflammation model. Toxicology and
Applied Pharmacology 237 (3), 306–316.

Islam, M.F., Rojas, E., Bergey, D.M., Johnson, A.T., Yodh, A.G.,
2003. High weight fraction surfactant solubilization of
single-wall carbon nanotubes in water. Nano Letters 3 (2),
269–273.
Jia, G., Wang, H., Yan, L., Wang, X., Pei, R., Yan, T., Zhao, Y.,
Guo, X., 2005. Cytotoxicity of carbon nanomaterials: singlewall
nanotube, multi-wall nanotube, and fullerene.
Environmental Science and Technology 39, 1378–1383.
Jiang, J., Gao, L., 2003. Production of aqueous colloidal dispersions
of carbon nanotubes. Journal of Colloid and Interface Science
260, 89–94.
Kang, S., Herzberg, M., Rodrigues, D.F., Elimelech, M., 2008a.
Antibacterial effects of carbon nanotubes: size does matter!
Langmuir 24 (13), 6409–6413.
Kang, S., Mauter, M.S., Elimelech, M., 2008b. Physicochemical
determinants of multiwalled carbon nanotube bacterial
cytotoxicity. Environmental Science and Technology 42 (19),
7528–7534.
Kang, S., Mauter, M.S., Elimelech, M., 2009. Microbial cytotoxicity
of carbon-based nanomaterials: implications for river water
and wastewater effluent. Environmental Science and
Technology 43 (7), 2648–2653.
Kang, S., Pinault, M., Pfefferle, L.D., Elimelech, M., 2007. Singlewalled
carbon nanotubes exhibit strong antimicrobial activity.
Langmuir 23 (17), 8670–8673.
Karajanagi, S.S., Yang, H.C., Asuri, P., Sellitto, E., Dordick, J.S.,
Kane, R.S., 2006. Protein-assisted solubilization of singlewalled
carbon nanotubes. Langmuir 22 (4), 1392–1395.
Kim, O.-K., Je, J., Baldwin, J.W., Kooi, S., Pehrsson, P.E., Buckley, L.J.,
2004. Solubilization of single-wall carbon nanotubes by
supramolecular encapsulation of helical amylose. Journal of
American Chemical Society 125, 4426–4427.
Lauret, J.-S., Voisin, C., Cassabois, G., Roussignol, P., Delalande, C.,
Capes, L.E., Valentin, E., Jost., O., 2004. Photocreated carrier
dynamics in isolated carbon nanotubes. Semiconductor
Science and Technology 19, S486–S488.
Lee, J.U., Huh, J., Kim, K.H., Park, C., Jo, W.H., 2007. Aqueous
suspension of carbon nanotubes via non-covalent
functionalization with oligothiophene-terminated
poly(ethylene glycol). Carbon 45, 1051–1057.
Lin, Y., Taylor, S., Li, H.P., Fernando, K.A.S., Qu, L.W., Wang, W.,
Gu, L.R., Zhou, B., Sun, Y.P., 2004. Advances toward
bioapplications of carbon nanotubes. Journal of Materials
Chemistry 14 (4), 527–541.
Lindberg, H.K., Falck, G.C.-M., Suhonen, S., Vippola, M.,
Vanhalab, E., Catalána, J., Savolainena, K., Norppa, H., 2009.
Genotoxicity of nanomaterials: DNA damage and micronuclei
induced by carbon nanotubes and graphite nanofibres in
human bronchial epithelial cells in vitro. Toxicology Letters
186, 166–173.
Liu, J., Rinzler, A.G., Dai, H.J., Hafner, J.H., Bradley, R.K., Boul, P.J.,
Lu, A., Iverson, T., Shelimov, K., Huffman, C.B., Rodriguez-
Macias, F., Shon, Y.S., Lee, T.R., Colbert, D.T., Smalley, R.E.,
1998. Fullerene pipes. Science 280 (5367), 1253–1256.
Liu, Y.Q., Gao, L., Zheng, S., 2007. Debundling of single-walled
carbon nanotubes by using natural polyelectrolytes.
Nanotechnology 18 (36), 365702.
Lyon, D.Y., Fortner, J.D., Sayes, C.M., Colvin, V.L., Hughes, J.B.,
2005. Bacterial cell association and antimicrobial activity of
a C60 water suspension. Environmental Toxicology and
Chemistry 24, 2757–2762.
Manna, S.K., Sarkar, S., Barr, J., Wise, K., Barrera, E.V., Jejelowo, O.,
Rice-Ficht, A., Ramesh, G.T., 2005. Single-walled carbon
nanotube induces oxidative stress and activates nuclear
transcription factor-kB in human keratinocytes. Nano Letters
5, 1676–1684.
Marsh, D.H., Rance, G.A., Zaka, M.H., Whitby, R.J., Khlobystov, A.N.,
2007. Comparison of the stability of multiwalled carbon
water research 44 (2010) 505–520 519
nanotube dispersions in water. Physical Chemistry Chemical
Physics 9 (40), 5490–5496.
McDonald, T.J., Engtrakul, C., Jones, M., Rumbles, G., Heben, M.J.,
2006. Kinetics of PL quenching during single-walled
carbon nanotube rebundling and diameter-dependent
surfactant interactions. Journal of Physical Chemistry B 110,
25339–25346.
Meng, J., Yanga, M., Song, L., Kong, H., Wang, C.Y., Wang, R.,
Wang, C., Xie, S.S., Xua, H.Y., 2009. Concentration control of
carbon nanotubes in aqueous solution and its influence on the
growth behavior of fibroblasts. Colloids and Surfaces B:
Biointerfaces 71, 148–153.
Moore, V.C., Strano, M.S., Haroz, E.H., Hauge, R.H., Smalley, R.E.,
Schmidt, J., Talmon, Y., 2003. Individually suspended singlewalled
carbon nanotubes in various surfactants. Nano Letters
3 (10), 1379–1382.
Muller, J., Huaux, F., Moreau, N., Misson, P., Heilier, J.-F., Delos, M.,
Arras, M., Fonseca, A., Nagy, J.B., Lison, D., 2005. Respiratory
toxicity of multi-wall carbon nanotubes. Toxicology and
Applied Pharmacology 207 (31), 221–231.
Narayan, R.J., Berry, C.J., Brigmon, R.L., 2005. Structural and
biological properties of carbon nanotube composite films.
Materials Science and Engineering B 123, 123–129.
O’Connell, M.J., Bachilo, S.M., Huffman, C.B., Moore, V.C.,
Strano, M.S., Haroz, E.H., Rialon, K.L., Boul, P.J., Noon, W.H.,
Kittrell, C., Ma, J., Hauge, R.H., Weisman, R.B., Smalley, R.E.,
2002. Band gap fluorescence from individual single-walled
carbon nanotubes. Science 297, 593–596.
O’Connell, M.J., Boul, P.J., Ericson, L.M., Huffman, C., Wang, Y.,
Haroz, E., 2001. Reversible water-solubilization of singlewalled
carbon nanotubes by polymer wrapping. Chemical
Physics Letters 342, 265–271.
Park, H.J., Heo, H.Y., Lee, S.C., Park, M., Lee, S.-S., Kim, J., Chang, J.Y.,
2006. Dispersion of single-walled carbon nanotubes in water
with polyphosphazene polyelectrolyte. Journal of Inorganic and
Organometallic Polymers and Materials 16 (4) 359–364.
Pulskamp, K., Worle-Knirsch, J.M., Hennrich, F., Kern, K., Krug, H.F.,
2007. Human lung epithelial cells show biphasic oxidative burst
after single-walled carbon nanotube contact. Carbon 45, 2241–
2249.
Saleh, N.D., Pfefferle, L.D., Elimelech, M., 2009. Aggregation
kinetics of multiwalled carbon nanotubes in aquatic systems:
measurements and environmental implications.
Environmental Science and Technology 42 (21), 7963–7969.
Salzmann, C.G., Chu, B.T.T., Tobias, G., Llewellyn, S.A.,
Green, M.L.H., 2007. Quantitative assessment of carbon
nanotube dispersions by Raman spectroscopy. Carbon 45 (5),
907–912.
Satoh, A.Y., Trosco, J., Masten, S.J., 2003. Epigenetic toxicity of
hydroxylated biphenyls and hydroxylated polychlorinated
biphenyls on normal rat liver epithelial cells. Environmental
Science and Technology 37, 2727–2733.
Sayes, C.M., Liang, F., Hudsona, J.L., Mendeza, J., Guob, W.,
Beach, J.M., Moore, V.C., Doyle, C.D., West, J.L., Billups, W.E.,
Ausmanb, K.D., Colvin, V.L., 2006. Functionalization density
dependence of single-walled carbon nanotubes cytotoxicity in
vitro. Toxicology Letters 161, 135–142.
Schmitt, D., Tran, N., Riefler, S., Jacoby, J., Merkel, D., Marone, P.,
Naouli, N., 2008. Toxicologic evaluation of modified gum
acacia: mutagenicity, acute and subchronic toxicity. Food and
Chemical Toxicology 46, 1048–1054.
Simon-Deckers, A., Gouget, B., Mayne-L’Hermiteb, M., Herlin-
Boimeb, N., Reynaudb, C., Carrièrea, M., 2008. In vitro
investigation of oxide nanoparticle and carbon nanotube
toxicity and intracellular accumulation in A549 human
pneumocytes. Toxicology 253, 137–146.
Sinani, V.A., Gheith, M.K., Yaroslavov, A.A., Rakhnyanskaya, A.A.,
Sun, K., Mamedov, A.A., Wicksted, J.P., Kotov, N.A., 2005.

520
Aqueous dispersions of single-wall and multiwall carbon
nanotubes with designed amphiphilic polycations. Journal of
American Chemical Society 127 (10), 3463–3472.
Star, A., Steuerman, D.W., Heath, J.R., Stoddart, J.F., 2002.
Starched carbon nanotubes. Angewandte Chemie
International Edition 41 (14), 2508–2512.
Tang, Y.J., Ashcroft, J.M., Chen, D., Min, G., Kim, C.-H.,
Murkhejee, B., Larabell, C., Keasling, J.D., Chen, F.F., 2007.
Charge-associated effects of fullerene derivatives on microbial
structural integrity and central metabolism. Nano Letters 7 (3),
754–760.
Tasis, D., Tagmatarchis, N., Bianco, A. and Prato, M., 2006.
Chemistry of carbon nanotubes. 106 (3), 1105–1136.
Tong, Z., Bischoff, M., Nies, L., Applegate, B., Turco, R., 2007.
Impact of fullerene (C60) on a soil microbial community.
Environmental Science and Technology 41, 2985–2991.
Trosko, J.E., Chang, B.V., Madhucar, J.E., Klauning, J.E., 1991.
Chemical, oncogene, and growth factor inhibition of gap
junctional intercellular communication: an alternative
hypothesis to carcinogenesis. Pathobiology 58, 265–278.
Trosko, J.E., Chang, C.-C., Upham, B., Wilson, M., 1998. Epigenetic
toxicology as toxicant-induced changes in intracellular
signalling leading to altered gap junctional intercellular
communication. Toxicology Letters 102–103, 71–78.
Tseng, C.-H., Wang, C.-C., Chen, C.-Y., 2006. Modification of
multi-walled carbon nanotubes by plasma treatment and
further use as templates for growth of CdS nanocrystals.
Nanotechnology 17, 5602–5612.
Upham, B.L., Masten, S.J., Lochwood, B.R., Trosko, J.E., 1994.
Nongenotoxic effects of polycyclic aromatic hydrocarbons and
their ozonation by-products on the intercellular
communication of rat liver epithelial cells. Fundamental and
Applied Toxicology 23, 470–475.
Vaisman, L., Marom, G., Wagner, H.D., 2006. Dispersions of
surface-modified carbon nanotubes in water-soluble and
water-insoluble polymers. Advanced Functional Materials 16
(3), 357–363.
Vittorio, O., Raffa, V. and Cuschieri, A., Influence of purity and surface
oxidation on cytotoxicity of multi-wall carbon nanotubes with
human neuroblastoma cells. Nanomedicine: Nanotechnology,
Biology, and Medicine, in press, doi:10.1016/j.nano.2009.02.006.
Wang, Y., Iqbal, Z., Malhotra, S.V., 2005. Functionalization of
carbon nanotubes with amines and enzymes. Chemical
Physics Letters 402, 96–101.
Weis, L.M., Rummel, A.M., Masten, S.J., Trosko, J.E., Upham, B.L.,
1998. Bay or baylike regions of polycyclic aromatic
hydrocarbons were potent inhibitors of gap junctional
water research 44 (2010) 505–520
intercellular communication. Environmental Health
Perspectives 106 (1), 17–22.
Wick, P., Manser, P., Limbach, L.K., Dettlaff-Weglikowskab, U.,
Krumeich, F., Roth, S., Stark, W.J., Bruinink, A., 2007. The
degree and kind of agglomeration affect carbon nanotube
cytotoxicity. Toxicology Letters 168, 121–131.
Wirnitzer, U., Herbold, B., Voetz, M., Ragot, J., 2009. Studies on the
in vitro genotoxicity of baytubes Ò , agglomerates of engineered
multi-walled carbon-nanotubes (MWCNT). Toxicology Letters
186, 160–165.
Wörle-Knirsch, J.M., Pulskamp, K., Krug, H.F., 2006. Oops they did
it again! Carbon nanotubes hoax scientists in viability assays.
Nano Letters 6 (6), 1261–1268.
Xu, F.-M., Xu, J.-P., Ji, J., Shen, J.-C., 2008. A novel biomimetic
polymer as amphiphilic surfactant for soluble and
biocompatible carbon nanotubes (CNTs). Colloids and
Surfaces B: Biointerfaces 67, 67–72.
Yamasaki, Y., 1990. Gap junctional intercellular communication
and carcinogenesis. Carcinogenesis 11, 1051–1058.
Yang, H., Liu, C., Yang, D., Zhanga, H., Xia, Z., 2009. Comparative
study of cytotoxicity, oxidative stress and genotoxicity induced
by four typical nanomaterials: the role of particle size, shape
and composition. Journal of Applied Toxicology 29, 69–78.
Yang, S.-T., Wang, X., Jia, G., Guc, Y., Wang, T., Nie, H., Ge, C.,
Wang, H., Liu, Y., 2008. Long-term accumulation and low
toxicity of single-walled carbon nanotubes in intravenously
exposed mice. Toxicology Letters 181, 182–189.
Yang, Z., Chen, X.H., Chen, C.S., 2007. Noncovalent-wrapped
sidewall multi-walled carbon nanotubes functionalization
with polyimide. Polymer Composites 28 (1), 36–41.
Ye, S.-F., Wu, Y.-H., Hou, Z.-Q., Zhang, Q.-Q., 2009. ROS and NF-jB
are involved in upregulation of IL-8 in A549 cells exposed to
multi-walled carbon nanotubes. Biochemical and Biophysical
Research Communications 379, 643–648.
Yu, B.Z., Yang, J.S., Li, W.X., 2007. In vitro capability of multi-walled
carbon nanotubes modified with gonadotrophin releasing
hormone on killing cancer cells. Carbon 45 (10), 1921–1927.
Yun, Y., Dong, Z., Tan, Z., Schulz, M.J., Shanov, V., 2009. Fibroblast
cell behavior on chemically functionalized carbon
nanomaterials. Materials Science and Engineering C 29, 719–725.
Zhu, Y., Ran, T., Li, Y., Guo, J., Li, W., 2006. Dependence of the
cytotoxicity of multi-walled carbon nanotubes on the culture
medium. Nanotechnology 16, 4668–4674.
Zong, S., Cao, Y., Jua, H., 2007. Direct electron transfer of hemoglobin
immobilized in multiwalled carbon nanotubes enhanced
grafted collagen matrix for electrocatalytic detection of
hydrogen peroxide. Electroanalysis 19 (7–8), 841–846.

Oxidation of atenolol, propranolol, carbamazepine and
clofibric acid by a biological Fenton-like system mediated by
the white-rot fungus Trametes versicolor
Ernest Marco-Urrea a , Jelena Radjenović b , Gloria Caminal c , Mira Petrović b,d ,
Teresa Vicent a, *, Damià Barceló b,e
a
Departament d’Enginyeria Química and Institut de Ciència i Tecnologia Ambiental, Universitat Autònoma de Barcelona (UAB),
08193 Bellaterra, Spain
b
Department of Environmental Chemistry, IDAEA-CSIC, c/Jordi Girona 18–26, 08034 Barcelona, Spain
c
Unitat de Biocatàlisis Aplicada associada al IQAC (CSIC-UAB), Escola Tècnica Superior d’Enginyeria, UAB, 08193 Bellaterra, Spain
d
Institució Catalana de Recerca i Estudis Avançats (ICREA), Barcelona, Spain
e
Institut Català de Recerca de l’Aigua (ICRA), Parc Científic i Tecnológic de la Universitat de Girona, Pic de Peguera, 15, 17003 Girona, Spain
article info
Article history:
Received 31 March 2009
Received in revised form
16 September 2009
Accepted 21 September 2009
Available online 22 September 2009
Keywords:
Trametes versicolor
Pharmaceuticals
Hydroxyl radical
Carbamazepine
Beta-blockers
Clofibric acid
1. Introduction
abstract
The presence of pharmaceuticals and personal care products
(PPCPs) and their metabolites in waters has become an
emerging environmental issue. To date, most attention has
been focused on identification, fate and distribution of PPCPs
water research 44 (2010) 521–532
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Biological advanced oxidation of the pharmaceuticals clofibric acid (CA), carbamazepine
(CBZP), atenolol (ATL) and propranolol (PPL) is reported for the first time. Extracellular
oxidizing species were produced through a quinone redox cycling mechanism catalyzed by
an intracellular quinone reductase and any of the ligninolytic enzymes of Trametes versicolor
after addition of the lignin-derived quinone 2,6-dimethoxy-1,4-benzoquinone (DBQ)
and Fe 3þ -oxalate in the medium. Time-course experiments with approximately 10 mg L 1
of initial pharmaceutical concentration resulted in percent degradations above 80% after
6 h of incubation. Oxidation of pharmaceuticals was only observed under DBQ redox
cycling conditions. A similar degradation pattern was observed when CBZP was added at
the environmentally relevant concentration of 50 mgL 1 . Depletion of DBQ due to the attack
of oxidizing agents was assumed to be the main limiting factor of pharmaceutical degradation.
The main degradation products, that resulted to be pharmaceutical hydroxylated
derivatives, were structurally elucidated. The detected 4- and 7-hydroxycarbamazepine
intermediates of CBZP degradation were not reported to date. Total disappearance of
intermediates was observed in all the experiments at the end of the incubation period.
ª 2009 Elsevier Ltd. All rights reserved.
in municipal wastewater treatment plants (WWTPs), which
are commonly found at very low concentrations (low ppb
levels) (Radjenovic et al., 2007, 2009, Reemtsma et al., 2006). To
avoid their potential adverse health effects on both humans
and environment, research efforts are underway to develop
efficient techniques for their removal. Among the alternatives
* Corresponding author. Departament d’Enginyeria Química and Institut de Ciència i Tecnologia Ambiental. Universitat Autònoma de
Barcelona (UAB). 08193 Bellaterra, Spain. Tel.: þ34 935812142; fax: þ34 935812013.
E-mail address: Teresa.Vicent@uab.cat (T. Vicent).
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.watres.2009.09.049

522
to traditional water treatment processes, advanced oxidation
processes (AOPs) have received great attention to degrade
PPCPs (Klavarioti et al., 2009; Song et al., 2008). AOPs can be
broadly described as aqueous phase oxidation methods based
on the intemediacy of highly reactive species such as hydroxyl
radicals ( , OH). This radical acts as a non-selective, very strong
oxidant agent with the ability to react with chemicals giving
dehydrogenated or hydroxylated derivates, until achieving
their total mineralization. To date, the production of , OH
radicals was based on chemical, photochemical, photocatalytical
or electrochemical techniques, but little is known
about the possibility to use biological systems to produce , OH
radicals for pollutant degradation (Klavarioti et al., 2009). This
biological approach was proposed recently for several whiterot
fungi (WRF) including Trametes versicolor and Pleurotus
eringyi through a simple quinone redox cycling mechanism
that is referred to as advanced biooxidation (Gómez-Toribio
et al., 2009a).
WRF are basidiomycetes that are capable of extensive
aerobic lignin depolymeration and mineralization due to the
low substrate specificity and high reactivity of the enzymes
they produce (principally laccase and peroxidases). Such
combinations of enzymes with auxiliary oxidoreductive
enzymes such as cytochrome P450 system are commonly
involved in the degradation of several PPCPs as for example
carbamazepine, clofibric acid, ibuprofen and several endocrine
disrupting chemicals (Cabana et al., 2007; Marco-Urrea
et al., 2009a).
In this work, the anti-epileptic drug carbamazepine (CBZP),
lipid regulator clofibric acid (CA), and b-blockers propranolol
(PPL) and atenolol (ATL) were the selected pharmaceuticals due
to their long-term use in Europe and North America and their
subsequent occurrence in the aquatic environment (Alder
et al., 2006). The metabolism of these pharmaceuticals in
mammals has been well established. It is known that CBZP
undergoes extensive hepatic metabolism by the cytochrome
P450 system, with oxidation to 10,11-dihydro-10,11-epoxy
carbamazepine, its further hydration to 11-dihydro-10,11epoxy
carbamazepine and conjugation of the latter one with
glucuronide being the main metabolic pathway of CBZP (Kerr
et al., 1994). CA is the pharmacologically active derivative of
clofibrate and several other fibrates, and it is metabolized to
reactive acylating derivatives that have been shown to transacylate
glutathione forming the corresponding glutathione
conjugate (Grillo and Benet, 2001). The primary metabolic
pathways of PPL are glucuronidation, side-chain oxidation and
ring oxidation, with the main metabolites arising from Ndealkylation
of the isopropanolamine side-chain, naphthalene
ring hydroxylation, and side-chain O-glucuronidation (Luan
et al., 2005). ATL is excreted eagerly via urine as an unchanged
compound (90%) (Dollery, 1991), with a small percentage of
atenolol-glucuronide (0.8–4.4%) and hydroxyatenolol (1.1–
4.4%, hydroxylation of the benzilic position).
Limited removal of ATL and CA and no removal of CBZP
have been frequently observed in municipal WWTPs and
subsequently they deserve more attention due to the high risk
of passing through later barriers in partly closed water cycles
(Radjenovic et al., 2007; Reemtsma et al., 2006; Ternes, 1998).
Joss et al. (2003) found that CBZP was not expected to produce
acute toxic effects in the aquatic biota, but chronic and
water research 44 (2010) 521–532
synergistic effects with other chemicals could not be
excluded. According to their results and regarding the present
European legislation on the classification and labelling of
chemicals (92/32/EEC), they classified CBZP as ‘‘R52/53 Harmful
to aquatic organisms and may cause long-term adverse
effects in the aquatic environment’’. On the other side, CA has
been reported to be a non-hazardous, yet environmentally
persistent compound (estimated persistence of 21 years in the
environment) (Buser et al., 1998; Ferrari et al., 2003). Nevertheless,
the possibility of endocrine disruption activity of CA
through interference with cholesterol synthesis was reported
(Pfluger and Dietrich, 2001), which is particularly important
regarding its long-term effects. Although specific environmental
effects of b-blockers are low, in a mixture with other bblockers
the effect of concentration addition can occur, as
shown in tests with Daphnia magna and phototoxicity assays
with green algae (Escher et al., 2006). Considering these facts,
removal techniques need to be developed to effectively
degrade the selected pharmaceuticals.
The objective of this study was to demonstrate for the first
time the degradation of the aforementioned pharmaceuticals
by biologically induced oxidizing species, produced by the
WRF T. versicolor. The advanced biooxidation strategy consists
of the incubation of fungi with a lignin-derived quinone (2,6,dimethoxy-1,4-benzoquione,
DBQ) and chelated ferric ion
(Fe 3þ -oxalate). Under these conditions, fungi catalyze the
conversion of the quinone into hydroquinone (DBQH2) byan
intracellular quinone reductase, and subsequent oxidation of
DBQH2 to semiquinone radicals (DBQ, ) is performed in the
extracellular medium by any of the lignin modifying enzymes
of the WRF (laccase and peroxidases). Then, Fenton’s reagent
is formed by DBQ , autoxidation catalyzed by Fe 3þ , in which
Fe 2þ ,
and superoxide anion radical (O2 ) are generated
(DBQ , þ Fe 3þ -oxalate / DBQ þ Fe 2þ -oxalate; and Fe 2þ -
oxalate þ O2 # Fe 3þ -oxalate þ O2 , ), followed by O2 ,
dismutation
(O2 , þ HO2 , þ H þ / O2 þ H2O2). Production of , OH
via quinone redox cycling described above was previously
proposed by estimating production of 2-thiobarbituric acid
reactive substances (TBARS) from 2-deoxyribose and hydroxylation
of 4-hydroxybenzoic acid producing 3,4-dimethoxybenzoic
acid, showing a high correlation between formation
of Fenton’s reagent and TBARS production (Gómez-Toribio
et al., 2009a). The inhibitory effect of , OH scavengers and
catalase on TBARS production rate further strengthened , OH
generation by this mechanism (Gómez-Toribio et al., 2009a;
Guillén et al., 2000). However, besides production of , OH
radicals, other oxidizing species such as ferryl ion might also
be produced in typical Fenton reactions under certain conditions
of pH and concentration of organic and inorganic ligands
(Hug and Leupin, 2003).Therefore, more research is needed to
ascertain the relative contribution of , OH produced under
quinone redox cycling conditions.
This quinone redox cycle mechanism applied to WRF
increases the range of environmental pollutants susceptible to
be degraded by these microorganisms due to the high oxidation
power and low substrate specificity of the oxidizing species
generated in comparison with their ligninolytic enzymes
(Marco-Urrea et al., 2009b). These oxidizing species are
produced rapidly in the extracellular medium and almost total
destruction of pollutants can be achieved during the first hours

of incubations without the need of adaptation of fungi to
pollutants (Gómez-Toribio et al., 2009a, b). Furthermore,
degradation metabolites were elucidated, which were further
oxidized either by the induced oxidizing agents or by the ligninolytic
enzyme system of fungus. Studies concerning degradation
of pharmaceuticals by fungi are very scarce, with 1- and
2-hydroxy ibuprofen, and 1,2-dihydroxy ibuprofen being the
only fungal metabolites of pharmaceutical reported up to date
(Marco-Urrea et al., 2009a). Nevertheless, the knowledge of
these metabolites is necessary for safe application of fungi
biocatalyst as bioremediation technology, as well as for the
elucidation of the key steps of the degradation process.
2. Materials and methods
2.1. Chemicals
CA (CAS No. 882-09-7), CBZP (CAS No. 298-46-4), ATL (CAS No.
29122-68-7), PPL (CAS No. 3506-09-0) and DBQ (CAS No. 530-55-2)
were obtained from Sigma–Aldrich Co. All other chemicals used
were of analytical grade.
2.2. Fungus and culture conditions
T. versicolor (ATCC#42530) was maintained by subculturing on
2% malt extract agar slants (pH 4.5) at room temperature.
Subcultures were routinely made every 30 days.
Pellets of T. versicolor were produced by inoculating 1 mL of
a mycelial suspension, prepared as described previously
(Marco-Urrea et al., 2008), in 1 L Erlenmeyer flask containing
250 mL of malt extract medium. This was shaken (135 rpm,
r ¼ 25 mm) at 25 C for 5 days. Subsequent pellets formed by
this process were transferred to another 1 L Erlenmeyer flask
containing 250 mL of a defined medium described elsewhere
(Marco-Urrea et al., 2008) and were also incubated for 2 days in
shaking conditions.
2.3. Degradation experiments
In time-course experiments, induction of oxidizing agents in
T. versicolor via quinone redox cycling was routinely performed
as follows. Mycelial pellets from each sample flask
were collected by filtration and washed three times with
MilliQ water. Appropriate amounts of 2-day old washed
mycelium pellets were incubated in the reaction mixture (see
figure legends). Reaction mixture contained 500 mM DBQ, 100–
300 mM Fe 3þ -oxalate, and 100 mM Mn 2þ in 25 mL 20 mM phosphate
buffer, pH 5, based on previous optimization of , OH
production in quinone redox cycling (Gómez-Toribio et al.,
2009b). Twenty mL of a solution containing the corresponding
pharmaceutical in acetonitrile was added into the flasks to
give the desired final pharmaceutical concentration (approximately
10 mg L 1 ). The flasks were incubated at 25 oC and
130 rpm and samples were taken at each point for analysis.
The samples were filtered through a Millex-GV (Millipore)
0.22 mm filter and subsequently analyzed by HPLC.
In order to investigate the degradation of pharmaceuticals
present at environmentally relevant concentrations, CBZP
was selected as a model compound and tests flasks were
water research 44 (2010) 521–532 523
amended with CBZP to a 50 mg L 1 concentration. Erlenmeyer
flasks containing 50 mL of the corresponding reaction mixture
were sacrificed at each experiment time and were filtered
through 0.45 mm glass fiber filter from Whatman. The target
compound was extracted in one step by solid phase extraction
with Oasis HLB cartridges (60 mg adsorbent, Waters, Barcelona,
Spain) as is described elsewhere (Radjenovic et al., 2007).
Briefly, the cartridges were preconditioned sequentially with
5 mL of methanol and 5 mL of deionized water at neutral pH.
The cartridge was dried under vacuum and was eluted with
two 2-mL portions of methanol and subsequently concentrated
to dryness under a gentle nitrogen stream. The extracts
were reconstituted with 0.5 mL 25:75 (v/v) acetonitrile-water.
To obviate the possible influence of light on pharmaceuticals
stability, all the experiments were carried out in the dark.
Each experiment included control flasks (without Fe 3þ -
oxalate). All the results included in the text and shown in
figures are the mean and standard deviation of duplicate
experiments.
2.4. Analytical procedures
2.4.1. Analysis of pharmaceuticals and DBQ
Analysis of the pharmaceuticals and DBQ were performed
using a Dionex 3000 Ultimate HPLC equipped with a UV
detector at 230 nm. The column temperature was 30 C and
a sample volume of 20 mL was injected from a Dionex autosampler.
Chromatographic separation of CBZP, PPL, CA and
DBQ was achieved on a GraceSmart RP 18 column
(250 mm 4 mm, particle size 5 mm). The mobile phase consisted
of 6.9 mmol L 1 acetic acid adjusted to pH 4 (by NaOH)
with 35% v/v acetonitrile. It was delivered isocratically at 1 mL
min 1 as was described elsewhere (Stafiej et al., 2007). ATL
analysis was performed with a Ascentis C18 column
(150 mm 4.6 mm, particle size 5 mm). Mobile phase of 0.01 M
ammonium acetate (pH 7) and mobile phase B (acetonitrile)
were delivered at flow rate of 1.2 mL min 1 and used for
gradient elution of ATL (t ¼ 0 min A ¼ 95%, t ¼ 20 min A ¼ 80%).
The detection limit was calculated to be

524
Fig. 1 – Time-course of CBZP (a) and CA (b) degradation in T. versicolor cultures under quinone redox cycle conditions.
Symbols: (B) CBZP and CA in the reaction mixture (25 mL 20 mM phosphate buffer containing 500 mM DBQ, 100–300 mM
Fe 3D -oxalate, 100 mM Mn 2D , and 10 mg L L1 of the corresponding pharmaceutical); (C) CBZP and CA in the control treatment
(non-containing DBQ and Fe 3D -oxalate); (-) DBQ(H2) in the reaction mixture; (:) metabolite formation expressed as relative
area (A/A0) where A is the corresponding metabolite and A0 is the parent drug in the control treatment at time zero. In the
case of CBZP where two hydroxylated isomers were found, black and white triangles refer to P254A and B, respectively.
Incubations were carried out with 1.5 ± 0.1 and 1.7 ± 0.2 mg dry weight mL L1 for CBZ and CA, respectively.
increased to 60% at 11 min, further increased to 90% of A in the
next 3 min and held isocratically for 2 min. Together with
returning to initial conditions, the total run time was 20 min.
CA was analyzed in the negative ion (NI) mode with mobile
phases consisting of (A) methanol; and (B) water. The elution
started at 5% of A for the first min, which was raised to 70% in
the next 7 min, to 90% in the next 2 min, and then returned to
initial conditions. The total run time was 13 min. The injection
volume of the sample was 10 mL.
The mass spectrometry analysis on the QqToF instrument
was performed in wide pass quadrupole mode, for MS
experiments, with the ToF data being collected between m/z
50–800. The capillary and cone voltages were set to 3000 and
25–30 V, respectively. Data were collected in the centroid
mode, with a scan accumulation time of 1 s. The instrument
was operated at a resolution of 5000 (FWHM). The nebulisation
gas was set to 500 L h 1 at a temperature of 350 C,
water research 44 (2010) 521–532
the cone gas was set to 50 L h 1 , and the source temperature
to 120 C. All analyses were acquired using an independent
reference spray via the LockSpray interference to ensure
accuracy and reproducibility. Sulfaguanidine (C7H10N4O2S)
was used as the internal lock mass in both PI mode
([M þ H] þ ¼ m/z 215.0602) and NI mode ([M H] ¼ m/z
213.0446). The LockSpray frequency was set at 11 s. Fragmentation
of precursor ions was done by applying collision
energies in the range of 10–35 eV, using argon as a collision
gas at a pressure of ~20 psi.
Elemental compositions of the molecular ions and their
fragments were determined and exact masses were calculated
with the help of MassLynx V4.1 software incorporated in the
instrument. Since the software calculation of the accurate
mass of cation is performed by adding a hydrogen atom
instead of proton, mass of one electron (i.e., 0.0005) was subtracted
from the calculated mass.

2.4.3. Mycelial dry weight
Mycelial dry weights were determined by vacuum filtering the
cultures through reweighed glass filters (Whatman GF/C,
Maidstone, England). The filters containing the mycelial mass
were placed in glass dishes and dried at 100 C to constant
weight.
3. Results and discussions
3.1. Oxidation of CA, CBZP, PPL and ATL by T.
versicolor under quinone redox cycling conditions
The strategy proposed here to degrade the selected pharmaceuticals
couples both AOP and biological strategies by
inducing the production of oxidizing agents such as , OH in the
white-rot fungus T. versicolor. In the experiments conducted,
water research 44 (2010) 521–532 525
Fig. 2 – Time-course of PPL (a) and ATL (b) degradation in T. versicolor cultures under quinone redox cycle conditions. Culture
conditions were as described in the legend for Fig. 1. Symbols: (B) PPL and ATL in the reaction mixture; (C) PPL and ATL in
the control treatment; (-) DBQ(H2) in the reaction mixture; (:) metabolite formation expressed as relative area (A/A0) where
A is the corresponding metabolite and A0 is the parent drug in the control treatment at time zero. Incubations were carried
out with 1.9 ± 0.1 mg dry weight mL L1 for both PPL and ATL.
oxalate was used as chelant agent of Fe 3þ instead of EDTA and
Mn 2þ was added into the incubation mixture, since the
oxidizing power under these conditions was substantially
improved as discussed previously (Gómez-Toribio et al.,
2009b). Figs. 1 and 2 show the time-courses degradation of the
selected pharmaceuticals in test flasks amended with
approximately 10 mg L 1 of the target pharmaceuticals, when
incubated under DBQ redox cycling conditions. Control
treatments in the absence of DBQ and Fe 3þ but containing
fungus were performed to demonstrate both the positive
involvement of oxidizing agents formed from the quinone
redox cycling mechanism and to discard the role of fungal
sorption in the removal of pharmaceuticals. Furthermore,
control treatments permit us to rule out the intracellular and
extracellular enzymatic system of T. versicolor in the degradation
of the pharmaceuticals. Previously CA and CBZP were
reported to be degraded by the cyt P450 system of T. versicolor
grown in a chemically define medium (Marco-Urrea et al.,

526
Fig. 3 – Time-course of CBZP degradation added at low
concentration of 50 mg L L1 in T. versicolor cultures under
quinone redox cycle conditions. Symbols: (;) CBZP in the
reaction mixture; (B) CBZP in the control treatment; (C)
CBZP in uninoculated bottles. Incubations were carried out
with 3.8 ± 0.1 mg dry weight mL L1 . Values plotted are
means ± standard deviations for triplicate cultures.
2009a) but as can be observed in Figs. 1 and 2 degradation did
not occur in control treatments for the incubation period
studied.
Results in Figs. 1 and 2 show a steep decrease of the
concentration of pharmaceuticals in the first hours of incubation
when the fungus was incubated under quinone redox
cycling conditions. Approximately 50% of CBZP, CA and ATL
were degraded the first hour and a plateau was reached above
80% after the sixth hour of incubation. However, the incubation
was continued to 24 h to investigate the further degradation
of possible metabolites formed during the degradation
process (see Section 3.3).
The consumption of DBQ was also investigated in the same
samples used for the determination of pharmaceutical
degradation. This analysis included the determination of DBQ
and its counterpart (DBQH 2) levels and is shown in Figs. 1 and
2 as the sum of both: DBQ(H 2). A high correlation between
DBQ(H2) and pharmaceutical degradation was found, suggesting
that DBQ(H2) consumption was due to the attack of the
induced oxidizing agents. Furthermore, since quinone redox
cycling and therefore the production of oxidizing species
cannot be sustained without the presence of DBQ(H2), the
plateau observed in the pharmaceutical degradation rate can
be ascribed to DBQ(H 2) depletion.
In an effort to discuss the relevance of this strategy in
regards to the normal concentrations of pharmaceuticals
found in aqueous environments (that are frequently detected
at levels from ng L 1 to low mg L 1 ), one experiment was
carried out at 50 mg L 1 . For this purpose CBZP was chosen as
model compound since it is found to be recalcitrant in sewer
treatment facilities and in the environment (Zhang et al.,
2008). As is shown in Fig. 3 the concentration of CBZP was
decreased to approximately 10 mg L 1 , due to the mechanisms
of fungal sorption and oxidation-induced degradation. In this
water research 44 (2010) 521–532
experiment, adsorption and transformation of pharmaceutical
by fungus cannot be clearly distinguished, since live
cultures can also degrade adsorbed pollutants by intracellular
mechanisms (Blánquez et al., 2004). In this case, adsorption
was higher than that observed at 10 mg L 1 probably due to
the highest concentration of fungal mycelium in the experimental
flasks (see figure legend). Similarly to that described
for higher concentrations, the plateau in CBZP degradation
under quinone redox cycling conditions was observed at the
fifth hour, when DBQ(H 2) was expected to be depleted under
the tested conditions.
3.2. QqToF-MS 2 fragmentation pathways of
pharmaceuticals and their metabolites
Table 1 summarizes the exact masses of molecular and fragment
ions, together with recalculated mass errors and double
bond equivalents (DBEs) given by the software (mass
measurements accuracy threshold of 5 mg L 1 ). Since all
fragment ions had an even number of electrons (i.e., DBE was
half integer number in all cases), all neutral losses had likewise
even electron configurations. These data were obtained under
optimized conditions of collision energy and cone voltage in
ESI(þ) and ESI( )MS 2 experiments on the QqToF instrument.
In ESI(þ) full-scan experiments performed at UPLC-QqToF
instrument, ATL (molecular ion [M þ H] þ 267) eluted at
3.4 min. Another peak was observed at 3.2 min, with the
molecular ion [M þ H] þ 283, denominated as metabolite P282.
Considering that the mass of P282 was shifted 16 Da upwards
relative to the parent compound, hydroxylation by , OH radical
attack was assumed. In Table 1 are presented calculated and
measured masses of fragment ions of ATL and its metabolite
P282, determined in ESI(þ)-MS 2 experiments at the QqToF-MS
instrument. The most abundant fragment ions in the MS 2
spectrum of ATL were detected m/z 190 (loss of 77 Da, isopropylamine
and water) and m/z 145 (further cleavage of CO
and intermolecular cyclization and rearrangement) (see
Fig. 4c). In the fragmentation pattern of the molecular ion
[M þ H] þ 283, characteristic losses of isopropyl group (42 Da)
and ammonia (17 Da) afforded the fragment ions at m/z 241
and 224 (see Fig. 4d). The subsequent losses of water and CO
from the fragment ion m/z 224 led to the formation of m/z 206
and 178 fragment ions. The fragment ion m/z 161 was formed
through cleavage of isopropylamine, water, CO and ammonia
and intramolecular cyclization, as suggested in the insert in
Fig. 4d. The fragment ion m/z 116 was also present in the MS 2
spectrum of P282, indicating that the hydroxylation had to
occur at the p-hydroxyphenylacetamide fraction of ATL. From
the comparison of the two fragmentation patterns, it was
deduced that the –OH group was attached to the alkyl sidechain,
at the C-atom next to the ether oxygen.
In the case of PPL, MS 2 fragmentation of the molecular ion
[M þ H] þ 260 rendered intense signals at m/z 183 and 157,
corresponding to the subsequent cleavages of aminoisopropyl
moiety and water, and C2H2, respectively (see Fig. 4a). The less
intense fragment ions at m/z 242 and 218 were formed through
cleavages of water and isopropylamine, respectively. The
fungal metabolite of PPL eluted 0.6 min earlier, with the
molecular ion [M þ H] þ 276, and was denominated as P275.
Besides the molecular ion with mass 16 Da greater than

Table 1 – Accurate mass measurement of product ions of carbamazepine (CBZP), clofibric acid (CA), propranolol (PPL), and atenolol (ATL) and their degradation products
P254, P230, P275 and P282, respectively, as determined by UPLC-QqToF in ESI(D) MS 2 mode for CBZP, PPL and ATL, and ESI(L) MS 2 mode for CA.
Comp. Precursor ion/product ion Elemental formula Mass (m/z) Error DBE a
Exp. Theor. mDa ppm
CBZP [M þ H] þ
C15H13N2O 237.1019 237.1022 0.3 1.3 10.5 8.93
[M þ Na] þ
C15H12N2ONa 259.0844 259.0842 0.2 0.8 10.5
[M þ H NH3] þ
C15H10NO 220.0761 220.0757 0.4 1.8 11.5
[M þ H CONH] þ
C14H12N 194.0960 194.0964 0.4 2.1 9.5
P254 [M þ H] þ
C15H13N2O2 253.0976 253.0972 0.4 1.6 10.5 7.24; 7.81
[M þ Na] þ
C15H12N2O2Na 275.0799 275.0791 0.8 2.9 10.5
[M þ H NH3] þ
C15H10NO2 236.0710 236.0706 0.4 1.7 11.5
[M þ H CONH] þ
C14H12NO 210.0906 210.0913 0.7 3.3 9.5
[M þ H CONH3] þ
C14H10NO 208.0762 208.0762 0 0 10.5
[M þ H CONH CO] þ
C13H12N 182.0965 182.0964 0.1 0.5 8.5
CA [M H] C10H10ClO3 213.0327 213.0324 0.3 1.4 5.5 5.90
[M H C4H6O2] C6H4ClO 126.9944 126.9956 1.2 9.4 4.5
[M H C6H4ClO] C4H6O2 85.0294 85.0290 0.4 4.7 2.5
P230 [M H] C10H10ClO4 229.0254 229.0268 1.4 6.1 5.5 6.40
[M H C4H6O2] C6H4ClO2 142.9909 142.9905 0.4 2.8 4.5
PPL [M þ H] þ
C16H22NO2 260.1653 260.1645 0.8 3.1 6.5 6.7
[M þ H H2O] þ
C16H20NO 242.1547 242.1539 0.8 3.3 7.5
[M þ H C(CH3)2] þ
C13H16NO2 218.1170 218.1176 0.6 2.7 6.5
[M þ H H2O NH2CH(CH3) 2] þ
C13H11O 183.0814 183.0804 1.0 5.5 8.5
[M þ H H2O NH2CH(CH3) 2 C2H2] þ
C11H9O 157.0648 157.0648 0 0 7.5
[M þ H C10H8O] þ
C6H14NO 116.1066 116.1070 0.4 3.4 0.5
P275 [M þ H] þ
C16H22NO3 276.1583 276.1594 1.1 4.0 6.5 6.1
[M þ H H2O] þ
C16H20NO2 258.1492 258.1489 0.3 1.2 7.5
[M þ H H2O O NH2CH(CH3) 2] þ
C13H11O 183.0795 183.0804 0.9 4.9 8.5
[M þ H H2O O NH2CH(CH3)2 C2H2] þ
C11H9O 157.0664 157.0648 1.6 10.1 7.5
[M þ H C10H8O2] þ
C6H14NO 116.1079 116.1070 0.9 7.7 0.5
ATL [M þ H] þ
C14H23N2O3 267.1696 267.1704 0.8 3.0 4.5 3.4
[M þ H C(CH3)2] þ
C11H17N2O3 225.1222 225.1234 5.3 1.2 4.5
[M þ H C(CH3)2 NH3] þ
C11H14NO3 208.0975 208.0969 0.6 2.9 5.5
[M þ H C(CH3) 2 H2O] þ
C11H12NO2 190.0874 190.0863 1.1 5.8 6.5
[M þ H C(CH3) 2 NH3 H2O CO NH3] þ
C10H9O 145.0650 145.0648 0.2 1.4 6.5
[M þ H CHCONH2 H2O C(CH3)2 NH3] þ
C9H9O 133.0658 133.0648 1.0 7.5 5.5
[M þ H C8H9NO2] þ
C6H14NO 116.1070 116.1071 0.1 0.9 0.5
P282 [M þ H] þ
C14H23N2O4 283.1663 283.1652 1.1 3.9 4.5 3.2
[M þ H C(CH3) 2] þ
C11H17N2O4 241.1186 241.1183 0.3 1.2 4.5
[M þ H C(CH3)2 NH3] þ
C11H14NO4 224.0901 224.0917 1.6 7.1 5.5
[M þ H C(CH3)2 NH3 H2O] þ
C11H12NO3 206.0807 206.0812 0.5 2.4 6.5
[M þ H C(CH3) 2 NH3 H2O CO] þ
C10H12NO2 178.0862 178.0863 0.1 0.6 5.5
[M þ H C(CH3)2 NH3 CO H2O NH3] þ
C10H9O2 161.0617 161.0597 2.0 12.4 6.5
[M þ H C8H9NO3] þ
C6H14NO 116.1055 116.1070 1.5 12.9 0.5
Only product ions with abundances higher than 10% are taken into account.
a DBE, double bond equivalent.
b tr-UPLC retention time.
tr b ,min
water research 44 (2010) 521–532 527

528
Fig. 4 – Spectra obtained in ESI(D)-MS/MS experiments at QqToF instrument (cone voltages 20–25 V, collision energies 15–
25 eV) for: (a) standard mixture of PPL in methanol/water (v/v, 25/75) at 10 mg L L1 , (b) degradation product of PPL, P275, in
the sample taken after 2 h, (c) standard mixture of ATL in methanol/water (v/v, 25/75) at 10 mg L L1 , and (d) degradation
product of ATL, P282, in the sample taken after 4 h.
[M þ H] þ 260, fragment ion at m/z 258 was observed, equivalent
to the m/z 242 in the MS 2 spectrum of PPL (Fig. 4b). The rest of
the observed fragment ions of the molecular ion [M þ H] þ 276
were the same (i.e., m/z 116, 157 and 183), indicating that the
naphthalene moiety was the site of , OH radical attack.
In full-scan experiments CBZP eluted at 8.94 min, whereas
two new peaks emerged at 7.24 and 7.81 min. The identical
fragmentation patterns of these two peaks and their molecular
weight (MW) 16 Da higher than the parent compound
(see Figs. 5c,d) suggested that the formed products were two
hydroxylated isomers of CBZP, denominated as P254A and B.
In Table 1 exact and measured masses were presented for
a more intense M254B. The collision-induced-dissociation
experiments with CBZP (molecular ion [M þ H] þ 237) revealed
the formation of only two fragment ions at m/z 220 and 194, as
a consequence of the loss of ammonia and CONH group,
respectively. Equivalent fragments were observed for P254A,B,
at m/z 236 and 210. The strong signal observed at m/z 208
suggested that the –OH group was probably attached at the C4
and C7-atom for the two observed isomers, since the loss of 2
H-atoms could be favoured by intramolecular cyclization as
presented in the insert of Fig. 5d. Further cleavage of CO in the
water research 44 (2010) 521–532
fragment ion m/z 208 afforded the signal at m/z 182. Nevertheless,
based on the retention times only it could not be
deduced which peak belongs to which product. Probably the
formation of intramolecular hydrogen bond between the
attached oxygen and hydrogen atoms of the amide group
caused one of the isomers to elute ~0.6 min later than the
other.
Finally, ESI( ) full-scan screenings of samples from the
experiments with CA (molecular ion [M þ H] þ 213) revealed
a new signal appearing 0.5 min after the CA peak, at 6.40 min.
Similar to the other metabolites identified, its MW was
deferring for 16 Da relative to the original drug, with the
molecular ion [M þ H] þ 229 (i.e., P230). Very poor MS 2 fragmentation
was observed for both parent compound and its
fungal metabolite. The collision of [M þ H] þ 213 rendered two
fragment ions at m/z 126 and 85, representing the cleavage of
the ether bond (see insert in Fig. 5a), whereas in the spectrum
of [M þ H] þ 229 only one fragment ion was detected at m/z 142,
equivalent to the m/z 126 but with mass shifted upwards for
16 Da. Nevertheless, this was enough evidence to assume
hydroxylation of the benzene ring, although the exact position
of the attached –OH group could not be deduced.

The degradation intermediates of each pharmaceutical
detected in this section are found in Fig. 6.
3.3. Discussion of the results on the fungal metabolites
of pharmaceuticals
The formation of metabolites was depicted in Figs. 1 and 2 and
was expressed as relative area (A/A0) measured by integration
of LC-MS peaks of the corresponding metabolite (A) and the
parent drug in the control treatments at time zero (A0), since
due to the lack of authentic analytical standards for the newly
identified products their quantitative determination was not
possible. Examination of the profiles show qualitatively
similar formation pattern of metabolites across the incubation
time, obtaining a maximum value of A/A 0 between the
first and fourth hour and disappearing completely in all the
cases after 24 h. The metabolites appeared to be formed at low
concentration levels in comparison with the parent drug, with
a maximum A/A0 values for each case in the range of 0.012
(PPL, Fig. 2a) to 0.27 (CA, Fig. 1b).
The same compound that was identified in our experiments
as the fungal metabolite of ATL, P282, was previously
reported as one of the intermediate products of solar photo-
water research 44 (2010) 521–532 529
Fig. 5 – Spectra obtained in MS/MS experiments at QqToF instrument (cone voltages 15–30 V, collision energies 15–25 eV)
for: (a) standard mixture of CA in methanol/water (v/v, 25/75) at 10 mg L L1 , (b) degradation product of CA, P230, in the
sample taken after 2 h, (c) standard mixture of CBZP in methanol/water (v/v, 25/75) at 10 mg L L1 , and (d) degradation product
of CBZP, P254, in the sample taken after 2 h.
Fenton and TiO2 photocatalytic treatment of ATL in pilotscale
compound parabolic collectors (Radjenovic et al., 2009).
In the case of these photocatalytic treatments, P282 was
observed together with its keto tautomer (P280), and was
formed through , OH-mediated reactions. Furthermore, Song
et al. (2008) reported a product of ATL with MW 282, formed in
the reaction of ATL with , OH radicals. This product was
identified by 60 Co g-irradiation and LC-MS, although the , OH
attacks was assumed to occur at the benzene ring. In the
same study hydroxylated derivative of PPL was also detected
with MW 275 and –OH group attached at the naphthalene
ring, which coincides with the structure of P275 identified in
the present study. Hydroxylation of naphthalene moiety is
phase I reaction in the human metabolism of PPL, in which 4hydroxy
PPL is formed (Luan et al., 2005). Cytochrome P450
isozymes were found to be responsible for oxidative metabolic
pathways of PPL in humans proceeding through naphthalene
ring-hydroxylations at the 4-, 5-, and 7-positions and
side-chain N-desisopropylation (Masubuchi et al., 1994).
However, in our study no metabolites were found in fungal
control treatments indicating that P275 was generated via
induction of oxidizing agents instead of fungal cyt P450
mechanism.

530
Fig. 6 – Selected pharmaceuticals and proposed structures of their degradation products.
Sirés et al. (2007) proposed the formation of a hydroxylated
intermediate as the observed in our study after the
hydroxylation of CA on its C 2-position by electro-Fenton and
photoelectro-Fenton. This proposed metabolite was not
identified by pure standards in the mentioned study but it
was assigned to a dehydrated species of 2-(4-chloro-2hydroxyphenoxy)-2-methylpropionic
acid on the basis of its
mass fragmentation spectra. On the other hand, by using an
HPLC-DAD and FLD, Doll and Frimmel (2004) proposed TiO2
photocatalytic degradation pathway of CA, which did not
include the hydroxylated product as the one observed in our
study. During photocatalysis CA underwent substitution of
–Cl by –OH group in the para-position, whereas products such
as 2-(4-hydroxyphenoxy)-isobutyric acid and hydroquinone
were formed. On the other side, cleavage of isobutyric acid
from the side-chain afforded 4-chlorophenol and 4-chlorocatechol.
Interestingly, in the abovementioned reports,
substrates of laccase such as phenol and 4-chlorophenol
have been described as typical metabolites of CA degradation
by AOP and once produced they could therefore be mineralized
by the ligninolytic enzymatic system of WRF. But
perhaps even more interesting is the production of different
water research 44 (2010) 521–532
quinone metabolites (also described for CA degradation by
, OH attack) that could theoretically be incorporated into the
quinone redox cycling enhancing the degradation rates of
the target pollutant.
As far as CBZP is concerned, there are few studies
regarding identification of products in , OH-induced degradation
pathway. Vogna et al. (2004) reported the formation of
toxic acridine intermediates in UV/H2O2 treatment of CBZP,
formed in , OH and , HO2-mediated reactions. Chiron et al.
(2006) also found acridine as major photodegradation intermediate
of CBZP after direct photolysis, but in the presence of
Fe þ3
they identified a hydroxycarbamazepine structure
derived from a , OH attack on the C 6 aromatic ring. The
formation of 2- and 3-hydroxylated derivatives of CBZP has
been reported in the metabolism of CBZP by the cytochrome
P450 in mammals (Mandrioli et al., 2001; Pearce et al., 2002) as
well as in the two model fungi Cunninghamella elegans and
Umbelopsis ramanniana used to study mammalian metabolism
for many drugs (Kang et al., 2008). As far as we know, the two
hydroxylated metabolites of CBZP in the 4- and 7-position
(P254A and B) detected in this study are reported for the first
time.

4. Conclusions
Biological advanced oxidation of ATL, CA, CBZP and PPL was
demonstrated for the first time by inducing extracellular
oxidizing species in T. versicolor. Under quinone redox cycling
conditions, pharmaceuticals degradation (added at 10 mg L 1 )
reached a plateau after 6 h of incubation achieving degradation
levels up to 80%. The plateau observed was due to DBQ
depletion, which was shown to be concomitant to pharmaceutical
degradation. The feasibility of this novel technology
to degrade pharmaceuticals at low levels commonly found in
the environment (50 mg L 1 ) was also demonstrated. Hydroxylated
intermediates of all pharmaceuticals were identified
and they disappeared after the incubation period, suggesting
the total mineralization of the target compound.
Acknowledgements
This work was supported by the Spanish MICINN (project
CTM2007-60971/TECNO) and MMAMRM (project 010/PC08/3-04).
The Department of Chemical Engineering of the UAB is the
Unit of Biochemical Engineering of the XRB de la Generalitat de
Catalunya. E. Marco-Urrea, T. Vicent and G. Caminal are
members of a Consolidated Research Group of Catalonia
(2005SGR 00220). Jelena Radjenović gratefully acknowledges
the JAE Program (CSIC-European Social Funds).
references
Alder, A.C., Bruchet, A., Carballa, M., Clara, M., Joss, A., Löffler, D.,
McArdell, C.S., Miksch, K., Omil, F., Tuhkanen, T., Ternes, T.A.,
2006. Consumption and occurrence. In: Ternes, T.A., Joss, A.
(Eds.), Human Pharmaceuticals, Hormones and Fragrances.
IWA Publishing, London, pp. 15–54.
Blánquez, P., Casas, N., Font, X., Gabarrell, X., Sarrà, M.,
Caminal, G., Vicent, T., 2004. Mechanism of textile metal dye
biotransformation by Trametes versicolor. Water Research 38
(8), 2166–2172.
Buser, H.-R., Müller, M.D., Theobald, N., 1998. Occurrence of the
pharmaceutical drug clofibric acid and the herbicide
mecoprop in various Swiss Lakes and in the North Sea.
Environmental Science and Technology 32 (1), 188–192.
Cabana, H., Jones, J.P., Agathos, S.N., 2007. Elimination of
endocrine disrupting chemicals using white-rot fungi and
their lignin modifying enzymes: a review. Engineering in Life
Sciences 7 (5), 429–456.
Chiron, S., Minero, C., Vione, D., 2006. Photodegradation
processes of the antiepileptic drug carbamazepine, relevant to
estuarine waters. Environmental Science and Technology 40
(19), 5977–5983.
Doll, T.E., Frimmel, F.H., 2004. Kinetic study of photocatalytic
degradation of carbamazepine, clofibric acid, iomeprol and
iopromide assisted by different TiO 2 materials – determination
of intermediates and reaction pathways. Water Research 38 (4),
955–964.
Dollery, C., 1991. Therapeutic Drugs. Churchill Livingstone,
Edinburgh.
Escher, B.I., Bramaz, N., Richter, M., Lienert, J., 2006. Comparative
ecotoxicological hazard assessment of beta-blockers and their
water research 44 (2010) 521–532 531
human metabolites using a mode-of-action-based test battery
and a QSAR approach. Environmental Science and Technology
40 (23), 7402–7408.
Ferrari, B., Paxéus, N., Giudice, R.L., Pollio, A., Garric, J., 2003.
Ecotoxicological impact of pharmaceuticals found in treated
wastewaters: study of carbamazepine, clofibric acid, and
diclofenac. Ecotoxicology and Environmental Safety 55 (3),
359–370.
Gómez-Toribio, V., García-Martín, A.B., Martínez, M.J.,
Martínez, A.T., Guillén, F., 2009a. Induction of extracellular
hydroxyl radical production by white-rot fungi through
quinone redox cycling. Applied and Environmental
Microbiology 75 (12), 3944–3953.
Gómez-Toribio, V., García-Martín, A.B., Martínez, M.J.,
Martínez, A.T., Guillén, F., 2009b. Enhancing the production of
hydroxyl radicals by Pleurotus eryngii via quinone redox cycling
for pollutant degradation. Applied and Environmental
Microbiology 75 (12), 3954–3962.
Grillo, M.P., Benet, L.Z., 2001. Interaction of gglutamyltranspeptidase
with clofibryl- S-acyl-glutathione in
vitro and in vivo in rat. Chemical Research in Toxicology 14 (8),
1033–1040.
Guillén, F., Gómez-Toribio, V., Martínez, M.J., Martínez, A.T., 2000.
Production of hydroxyl radical by the synergistic action of
fungal laccase and aryl alcohol oxidase. Archives of
Biochemistry and Biophysics 383 (1), 142–147.
Hug, S.J., Leupin, O., 2003. Iron-catalyzed oxidation of arsenic(III)
by oxygen and by hydrogen peroxide: pH-dependent
formation of oxidants in the Fenton reaction. Environmental
Science and Technology 37 (12), 2734–2742.
Joss, A., Repetto, G., Rios, J.C., Hazen, M.J., Molero, M.L., del
Peso, A., Salguero, M., Fernandez-Freire, P., Perez-Martin, J.M.,
Camean, A., 2003. Ecotoxicological evaluation of
carbamazepine using six different model systems with
eighteen endpoints. Toxicology in Vitro 17 (5-6), 525–532.
Kang, S.-I., Kang, S.-Y., Hur, H.-G., 2008. Identification of fungal
metabolites of anticonvulsant drug carbamazepine. Applied
Microbiology and Biotechnology 79 (4), 663–669.
Kerr, B.M., Thummel, K.E., Wurden, C.J., Klein, S.M., Kroetz, D.L.,
Gonzalez, F.J., Levy, R.H., 1994. Human liver carbamazepine
metabolism: role of CYP3A4 and CYP2C8 in 10,11-epoxide
formation. Biochemical Pharmacology 47 (11), 1969–1979.
Klavarioti, M., Mantzavinos, D., Casinos, D., 2009. Removal of
residual pharmaceuticals from aqueous systems by advanced
oxidation processes. Environment International 35 (2),
402–417.
Luan, L.-J., Shao, Q., Ma, J.-Y., Zeng, S., 2005. Stereoselective
urinary excretion of S-( )- and R-(þ)-propranolol glucuronide
following oral administration of RS-propranolol in Chinese
Han subjects. World Journal of Gastroenterology 11 (12),
1822–1824.
Mandrioli, R., Albani, F., Casamenti, G., Sabbioni, C., Raggi, M.A.,
2001. Simultaneous high-performance liquid chromatography
determination of carbamazepine and five of its metabolites in
plasma of epileptic patients. Journal of Chromatography B 762
(2), 109–116.
Marco-Urrea, E., Parella, T., Gabarrell, X., Caminal, G., Vicent, T.,
Reddy, C.A., 2008. Mechanistics of trichloroethylene
mineralization by the white-rot fungus Trametes versicolor.
Chemosphere 70 (3), 404–410.
Marco-Urrea, E., Pérez-Trujillo, M., Vicent, T., Caminal, G., 2009a.
Ability of white-rot fungi to remove selected pharmaceuticals
and identification of degradation products of ibuprofen by
Trametes versicolor. Chemosphere 74 (6), 765–772.
Marco-Urrea, E., Aranda, E., Caminal, G., Guillén, F., 2009b.
Induction of hydroxyl radical production in Trametes versicolor
to degrade recalcitrant chlorinated hydrocarbons. Bioresource
Technology 100 (23), 5757–5762.

532
Masubuchi, Y., Hosokawa, S., Horie, T., Suzuki, T., Ohmori, S.,
Kitada, M., Narimatsu, S., 1994. Cytochrome P450 isozymes
involved in propranolol metabolism in human liver
microsomes. The role of CYP2D6 as ring-hydroxylase and
CYP1A2 as N-desisopropylase. Drug Metabolism and
Disposition 22 (6), 909–915.
Pearce, R.E., Vakkalagadda, G.R., Steven Leeder, J., 2002. Pathways
of carbamazepine bioactivation in vitro I. Characterization of
human cytochromes P450 responsible for the formation of
2- and 3-hydroxylated metabolites. Drug Metabolism and
Disposition 30 (11), 1170–1179.
Pfluger, P., Dietrich, D.R., 2001. Effects on pharmaceuticals in the
environmentdan overview and principle considerations. In:
Kümmerer, K. (Ed.), Pharmaceuticals in the
EnvironmentdSources, Fate, Effects and Risks. Springer, Berlin,
pp. 11–17.
Radjenovic, J., Petrovic, M., Barceló, D., 2007. Analysis of
pharmaceuticals in wastewater and removal using
a membrane bioreactor. Analytical and Bioanalytical
Chemistry 387 (4), 1365–1377.
Radjenovic, J., Sirtori, C., Petrovic, M., Barceló, D., Sixto, M., 2009.
Solar photocatalytic degradation of persistent
pharmaceuticals at pilot-scale: Kinetics and characterization
of major intermediate products, Applied catalysis B:
Environmental 89 (1–2), 255–264.
water research 44 (2010) 521–532
Reemtsma, T., Weiss, S., Mueller, J., Petrovic, M., González, S.,
Barceló, D., Ventura, F., Knepper, T.P., 2006. Polar pollutants
entry into the water cycle by municipal wastewater:
a European perspective. Environmental Science and
Technology 40 (17), 5451–5458.
Sirés, I., Arias, C., Cabot, P.Ll., Centellas, F., Garrido, J.A.,
Rodríguez, R.M., Brillas, E., 2007. Degradation of clofibric acid
in acidic aqueous medium by electro-Fenton and
photoelectro-Fenton. Chemosphere 66 (9), 1660–1669.
Song, W., Cooper, W.J., Mezyk, S.P., Greaves, J., Peake, B.M., 2008.
Free radical destruction of b-blockers in aqueous solution.
Environmental Science and Technology 42 (4), 1256–1261.
Stafiej, A., Pyrzynska, K., Regan, F., 2007. Determination of antiinflammatory
drugs and estrogens in water by HPLC with UV
detection. Journal of Separation Science 30 (7), 985–991.
Ternes, T.A., 1998. Occurrence of drugs in German sewage
treatment plants and rivers. Water Research 32 (11), 3245–
3260.
Vogna, D., Marotta, R., Andreozzi, R., Napolitano, A., D’Ischia, M.,
2004. Kinetic and chemical assessment of the UV/H 2O 2
treatment of antiepileptic drug carbamazepine. Chemosphere
54 (4), 497–505.
Zhang, Y., Geissen, S.U., Gal, C., 2008. Carbamazepine and
diclofenac: removal in wastewater treatment plants and
occurrence in water bodies. Chemosphere 73 (8), 1151–1161.

Loadings, trends, comparisons, and fate of achiral and
chiral pharmaceuticals in wastewaters from urban tertiary
and rural aerated lagoon treatments
Sherri L. MacLeod a,1 , Charles S. Wong a,b, *
a
Department of Chemistry, University of Alberta, Edmonton, Alberta, T6G 2G2, Canada
b
Environmental Studies Program and Department of Chemistry, Richardson College for the Environment, University of Winnipeg,
Winnipeg, Manitoba, R3B 2E9, Canada
article info
Article history:
Received 1 June 2009
Received in revised form
14 September 2009
Accepted 23 September 2009
Available online 27 September 2009
Keywords:
Pharmaceuticals
Passive samplers
Wastewater
Loadings
Trends
Chiral drugs
1. Introduction
abstract
Pharmaceutical occurrence, fate, and toxicity in aquatic
ecosystems has been studied for more than 10 years (Daughton
and Ternes, 1999; Halling-Sorensen et al., 1998; Kolpin
water research 44 (2010) 533–544
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
A comparison of time-weighted average pharmaceutical concentrations, loadings and
enantiomer fractions (EFs) was made among treated wastewater from one rural aerated
lagoon and from two urban tertiary wastewater treatment plants (WWTPs) in Alberta,
Canada. Passive samplers were deployed directly in treated effluent for nearly continuous
monitoring of temporal trends between July 2007 and April 2008. In aerated lagoon effluent,
concentrations of some drugs changed over time, with some higher concentrations in
winter likely due to reduced attenuation from lower temperatures (e.g., less microbially
mediated biotransformation) and reduced photolysis from ice cover over lagoons; however,
concentrations of some drugs (e.g. antibiotics) may also be influenced by changing use
patterns over the year. Winter loadings to receiving waters for the sum of all drugs were
700 and 400 g/day from the two urban plants, compared with 4 g/day from the rural plant.
Per capita loadings were similar amongst all plants. This result indicates that measured
loadings, weighted by population served by WWTPs, are a good predictor of other effluent
concentrations, even among different treatment types. Temporal changes in chiral drug
EFs were observed in the effluent of aerated lagoons, and some differences in EF were
found among WWTPs. This result suggests that there may be some variation of microbial
biotransformation of drugs in WWTPs among plants and treatment types, and that the
latter may be a good predictor of EF for some, but not all drugs.
ª 2009 Elsevier Ltd. All rights reserved.
et al., 2002; Stamm et al., 2008). While drugs have been
consistently detected in surface water, the effects of these
complex mixtures of bioactive pollutants on non-target
aquatic organisms are not fully understood (Fent et al., 2006).
Accurate exposure scenarios are essential to assess toxicity
Abbreviations: CR, Capital Region; DL, method detection limit; EF, enantiomer fraction; GB, Gold Bar; LLB, Lac La Biche; PEC, predicted
environmental concentration; POCIS, polar organic chemical integrative sampler; SD, standard deviation; TWA, time-weighted average;
WWTP, wastewater treatment plant.
* Corresponding author. Richardson College for the Environment, University of Winnipeg, Winnipeg, Manitoba, R3B 2E9, Canada.
E-mail addresses: smacleod@ualberta.ca (S.L. MacLeod), wong.charles.shiu@alum.mit.edu (C.S. Wong).
1 Department of Chemistry, Université de Montréal, Montréal, Quebec, H3C 3J7, Canada.
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.watres.2009.09.056

534
and assign risk. Exposure, in turn, can be determined by either
prediction or by measurement of wastewater or surface water
concentrations. Prescription and sales information are not
readily available, and may not be ideal for predicting environmental
concentrations of some drugs (Coetsier et al., 2009)
because of the uncertainty associated with use and metabolism
(Letzel et al., 2009). In addition, removal of drugs from
wastewater may range from 0 to 100% depending on the drug
and the treatment process (Batt et al., 2007; Letzel et al., 2009,
Lishman et al., 2006), even for the same compound such as
diclofenac (Letzel et al., 2009), further complicating prediction
of environmental concentrations. In Europe, where a tiered
approach is taken for requirements of fate and effects studies,
predicted environmental concentrations (PECs) are calculated
as a worst-case scenario, without accounting for any metabolism,
biodegradation, or retention by treatment (EMEA, 2006).
For PEC values above 10 ng/L, higher tiered assessment is
required (EMEA, 2006), and while in a French study, PECs were
accurate for some drugs (carbamazepine, diclofenac,
propranolol), local measured environmental concentrations
of others were found to be much lower than those predicted,
even very close (10 m) to effluent outfalls (Coetsier et al., 2009).
Wastewater and surface water concentrations may be
heavily affected by abiotic transformation processes such as
photolysis, which depends on light intensity over the seasons
(Vieno et al., 2005); as well as biotransformation, which would
depend on microbial activity and on temperature. The latter is
particularly significant for the numerous chiral drugs. As with
other chiral pollutants (Wong, 2006), the enantiomers of
pharmaceuticals are affected identically by abiotic processes,
but may be affected differently by biologically mediated
processes in wastewater treatment (Fono and Sedlak, 2005;
MacLeod, 2009; MacLeod et al., 2007b; Nikolai et al., 2006). Drug
enantiomers may also exhibit differential toxicity to aquatic
life (Stanley et al., 2006). Thus, a robust means by which to
predict environmental loadings and fate of drugs from wastewater
effluents is valuable for exposure and risk assessment.
Direct measurement of drug loadings from wastewater
treatment plants (WWTPs) has proven accurate for prediction
of drug concentrations in receiving waters (Letzel et al., 2009;
MacLeod, 2009). Such measurements can be accomplished via
repeated grab sampling (Sacher et al., 2008), which is tedious
and time-consuming, or continuous flow-proportional effluent
collection (Letzel et al., 2009). However, passive sampling
devices such as the Polar Organic Chemical Integrative Sampler
(POCIS) (Alvarez et al., 2004) can provide a simple means of
obtaining time-weighted average (TWA) loadings of polar
contaminants, such as drugs, over longer time periods with
lower detection limits (Jones-Lepp et al., 2004; MacLeod et al.,
2007a; Zhang et al., 2008). While a high-frequency, discrete
sampling approach may capture the real-time heterogeneity of
contaminant loadings, passive samplers have the advantage of
providing integrated, continuous sample collection, and are
thus useful as an alternative or complementary sampling
approach for dissolved phase chemicals.
Most research on drugs in the environment has, thus far,
focused on measurement of drugs in the waste or receiving
waters from larger (10,000 to millions) population centers
(Carballa et al., 2007; Chen et al., 2006; Hua et al., 2006; Kimura
et al., 2007; Lajeunesse et al., 2008; Letzel et al., 2009; Lissemore
water research 44 (2010) 533–544
et al., 2006; Loraine and Pettigrove, 2006; Managaki et al., 2007;
Sacher et al., 2008; Vieno et al., 2005), with little attention paid
to the drug output from the wastewater of smaller (

communities. Treated effluent from LLB is discharged
continuously to Field Lake which discharges into Lac La Biche
(the lake) via Red Deer Brook, while GB and CR both discharge
treated effluent to the North Saskatchewan River and we
have previously measured drugs in both of these receiving
waters (MacLeod, 2009; MacLeod et al., 2007a). To our
knowledge, this is the first study to use POCIS to undertake
long-term, nearly continuous, monitoring of drugs, particularly
chiral drugs, in municipal effluents from a small population
center.
2. Materials and methods
2.1. Materials
We used POCIS (Environmental Sampling Technologies,
St. Joseph, MO) in the ‘pharmaceutical’ configuration: 200 mg
Oasis HLB sorbent (Waters, Milford, MA) between two polyethersulfone
membranes (45.8 cm 2 total surface area) held
together by stainless steel rings and attached to a spindle
holder. This configuration was used as POCIS sampling rates
are available for the target analytes (Tables 1 and S2), all
commonly used drugs. Chemical analytes (>98% purity) and
used as received (sources in Table S2). Solvents and reagents
for chromatography (Fisher Scientific, Ottawa, Canada) were
of HPLC grade; nanopure water (18 MU cm) was supplied by
a Nanopure Ultrapure system (Barnsteam/Thermolyne,
Dubuque, IA). Details are published elsewhere (MacLeod, 2009;
MacLeod et al., 2007a, b).
Table 1 – Pooled winter time-weighted average
concentrations for drugs detected at all three WWTPs in
each of the three deployment periods.
Analyte Gold
Bar
(ng/L) SD
Capital
Region
(ng/L) SD
Lac la
Biche
(ng/L) SD
Atenolol 72 270 120 440 36 160
Carbamazepine 200 130 290 190 63 39
Celecoxib 17 8 28 13 22 10
Citalopram 23 5 67 8 5.7 1.7
Clarithromycin 220 140 420 270 120 90
Codeine 450 380 920 710 990 710
Diclofenac 1200 680 2,000 1300 460 280
Erythromycin 13 10 19 15 3.1 2.8
Gemfibrozil 150 59 190 72 72 34
Metoprolol 58 47 65 52 29 23
Naproxen 66 58 530 510 280 250
Paroxetine 0.7 0.9 1.4 1.8 0.2 0.2
Propranolol 4.7 3.0 14 8.5 1.4 1.1
Sotalol 27 17 66 22 16 11
Temazepam 130 71 160 76 120 60
Triclosan 1.5 2.3 53 69 6.9 8.7
Trimethoprim 38 39 65 70 10 12
Sum of drugs
(mg/L)
2.2 5.0 2.2
Pooled values calculated from n ¼ 6 for Gold Bar and Capital Region
WWTPs, n ¼ 9 for Lac la Biche WWTP.
water research 44 (2010) 533–544 535
2.2. Passive sampling in WWTP effluent
Sampling of WWTP effluent took place between July 2007 and
April 2008 (Table S1).
During each sampling period at the CR and LLB WWTPs,
two and three POCIS, respectively, were protected in perforated
stainless steel cages secured with nylon rope, and
deployed in final effluent. During each sampling period at GB
WWTP, two caged samplers were deployed in a tank into
which treated effluent was continuously pumped. Field blanks
were brought to all sites at deployment and retrieval, at which
time POCIS were individually put on ice until return to the
laboratory, where they were stored at 20 C until chemical
extraction.
2.3. Chemical extraction and instrumental analysis
Details on chemical extraction and instrumental analysis are
provided elsewhere (MacLeod, 2009; MacLeod et al., 2007a, b)
and in Supplemental Information. In brief, POCIS sorbent was
washed into a glass column, and methanol used to elute target
analytes. The extract was filtered, reduced to 5–10 mL via
rotary evaporation, filtered through 0.22 mm Acrodisc polytetrafluoroethylene
syringe filters (Pall Life Sciences, Ann
Arbor, MI, USA), and nitrogen-evaporated to dryness. Internal
standards (Table S2) were added, and extracts were reconstituted
to 1 mL with methanol. Achiral chromatography for
determining concentrations (MacLeod et al., 2007a) was
performed with an Ultra C18 reversed-phase column
(150 mm 4.6 mm internal diameter (i.d.) 5 mm particle size
(dp), Restek, Bellefonte, PA, USA), while enantioselective
chromatography for determining enantiomer compositions
was performed with a Chirobiotic V column (250 mm 4.6 mm
i.d. 5 mm d p, Advanced Separation Technologies, Whippany,
NJ, USA) for all chiral analytes except temazepam, for which
a Chiralpak AD-RH column (MacLeod, 2009) was used
(150 mm 4.6 mm i.d. 5 mm d p, Daicel Chemical Industries,
West Chester, PA). Analytes were detected by electrospray
ionization-tandem mass spectrometry using an Applied Biosystems
QTrap triple-quadruple instrument (Foster City, CA) in
multiple reaction monitoring mode.
2.4. Quality assurance/quality control and
data handling
Detection limits (DLs, Table S2) were determined as the
analyte concentration in 1 mL POCIS extract providing
a signal-to-noise ratio of 3, and ranged from 0.001 to 1.6 ng/
POCIS. Drugs were not detected in any blanks. Calculated
time-weighted average concentrations and loadings (hereafter
called ‘‘concentrations’’ and ‘‘loadings’’, respectively) are
presented as mean standard deviation (SD). Data analysis
was carried out as previously described elsewhere (MacLeod,
2009; MacLeod et al., 2007a, b). Statistical comparisons
between two groups were carried out by t-test, while three or
more groups were compared by ANOVA, with either Tukey’s
test for multiple comparisons or with Dunnett’s test for
comparison with a control, as indicated. Differences noted are
statistically significant (all p < 0.05). As expected, analytes for
which the POCIS sampling rate was known with higher

536
precision (Table S2) were generally more conducive to statistical
significance testing.
For chiral drugs atenolol, citalopram, fluoxetine, metoprolol,
nadolol, pindolol, propranolol, salbutamol, sotalol, and
temazepam, enantiomer fractions (EFs) described enantiomer
composition (Harner et al., 2000):
EF ¼
E1
E1 þ E2
ðþÞ
¼
ðþÞþð Þ
where E 1 and E 2 are the first and second eluted enantiomers,
respectively, when the enantiomer optical rotation was
unknown (i.e., all analytes except atenolol, propranolol, and
fluoxetine). Further information on data quality and handling
is available in Supplementary Data.
3. Results and discussion
3.1. Time-weighted average pharmaceutical
concentrations in treated effluents
3.1.1. Aerated lagoons: Lac la Biche
Most studies to date have relied on grab or short-term
composite sampling to assess temporal trends in pharmaceutical
concentrations in effluents (Batt et al., 2007; Chen et al,
2006; Hua et al., 2006; Loraine and Pettigrove 2006, Stülten et al.,
2008; Vieno et al., 2005). However, temporal trends assumed
from grab sampling may not be representative of longer time
periods. Because POCIS can provide longer and continuous
temporal integration, it may provide more accurate assessment
of temporal trends. However, passive samplers cannot
provide reliable information on concentration fluctuations
over time scales shorter than sample integrative periods,
which can result in very different conclusions about temporal
changes in concentration over time (Shaw and Mueller, 2009).
Drug concentrations at LLB (Fig. 2) measured over six
sampling periods of 35–51 days (Table S1) ranged from 0.07 ng/L
(paroxetine, propranolol) to 1100 ng/L (codeine). Fenoprofen,
fluoxetine, ketoprofen, omeprazole, roxithromycin, sildenafil,
sulfadimethoxine, sulfamethazine, sulfapyridine, sulfisoxazole,
tadalafil and vardenafil were not detected in LLB treated
effluent above their respective DLs in any sampling period
(Table S2). Concentrations of atenolol and propranolol were
similar to effluent grab sampled from LLB in September 2005
(Nikolai et al., 2006); however, metoprolol was higher (200 ng/L).
Diclofenac and naproxen concentrations at LLB were also
similar to that in effluents of the activated sludge WWTP of
Aura, Finland, which has a similar population (4000) to LLB
(Vieno et al., 2005). Concentrations were also generally similar
to those in an aerated lagoon WWTP in Louisiana, USA (Conkle
et al., 2008) for drugs common to both studies: atenolol, carbamazepine,
gemfibrozil, metoprolol, naproxen, and sotalol.
Thus, our TWA concentrations are likely representative of
those expected in small community WWTP effluents.
There were statistically significant temporal concentration
changes (Fig. 2) for celecoxib, citalopram, clarithromycin,
codeine, diclofenac, erythromycin, gemfibrozil, naproxen,
propranolol, sotalol, and temazepam. Temporal concentration
trends were similar for clarithromycin, codeine, diclofenac,
water research 44 (2010) 533–544
(1)
erythromycin, naproxen and propranolol, with higher
concentrations during winter sampling periods (December to
April) than at other times (July to December, Fig. 2). For atenolol,
carbamazepine, metoprolol, triclosan and trimethoprim,
the effluent concentrations did not change significantly
over the sampling periods (Fig. 2). For drugs with temporal
differences in concentration, the changes in concentration
ranged from 2 (propranolol) to 960 ng/L (codeine), whereas
drugs without statistically significant temporal differences
had smaller concentration changes over time, from 8 (triclosan)
to 85 ng/L (atenolol).
While the average daily volume of treated effluent
discharged decreased over the six sampling periods (Table S1),
this decrease was not statistically significant. Thus, TWA
loadings are reflective of concentrations. Of the drugs studied,
photodegradation likely plays a major role in the dissipation
of diclofenac (Andreozzi et al., 2003), naproxen (Lin et al., 2006;
Lin and Reinhard, 2005), and propranolol (Andreozzi et al.,
2003) in aquatic systems. The higher effluent concentrations
of those drugs in winter (Fig. 2) may be at least partially the
result of decreased photolysis, due to ice cover on lagoons
from November to March (Siebold, 2008) blocking sunlight
from the water column (Vieno et al., 2005), and fewer hours of
sunlight during those months (MacLeod, 2009; Vieno et al.,
2005). It is impossible to ascertain the amount of photodegradation
in the lagoon during ice-free periods with our
data. While the LLB lagoon wastewater was not clear in color,
it is likely that photodegradation of drugs took place at least at
the lagoon surface during ice-free periods, and continuous
aeration of lagoon waters for oxygenation purposes would
bring water at depth to the surface to provide some photodegradation
treatment. Lower winter temperatures also likely
influenced elimination of pharmaceuticals, through lower
reaction rate constants for abiotic transformation, and
reduced microbial activity for biotransformation (Kim and
Carlson, 2007). Effluent pH was 8 0.5 over the six sampling
periods and effluent temperatures ( C) were 21 3, 16 2,
4 3, 1 0, 1.4 0.5, 4 2, chronologically for the six
sampling periods (Siebold, 2008). The concentrations of those
three drugs decreased again in March/April, as temperatures
and number of daylight hours increased, and ice cover was
reduced or absent on the lagoons (Fig. 2). Since naproxen can
be readily biodegraded in sewage (Yu et al., 2006), decreased
biodegradation may also contribute to the higher effluent
concentrations for naproxen in winter.
It is possible that changes in temperature may affect
calculated TWA concentrations, as sampling rates used in
this study were determined at room temperature (22–28 C,
MacLeod et al., 2007a), and WWTP temperatures during
colder months were considerably less. However, we cannot
presently quantify the compound-dependent effect of such
changes on our observed trends. Lower temperatures would
result in lower analyte aqueous diffusion coefficients and
hence lower sampling rates, estimated to be roughly 75%
over a 20 C ambient temperature change (MacLeod et al.,
2007a). When applied to water colder than under the calibration
condition, sampling rates would then provide an
underestimate of the calculated TWA concentrations, the
effect of which would be compound-dependent. Such an
effect may be pronounced in the case of drugs with higher

concentrations in winter than in other times (e.g., antibiotics
such as clarithromycin and erythromycin, Fig. 2). Temporal
effects on concentrations were less for other analytes, and
may either be insignificant (e.g., triclosan and gemfibrozil) or
of little practical significance even if statistically significant,
given the relatively narrow range (e.g., from a factor of two,
to an order of magnitude) in concentrations over the seasons
studied (Fig. 2).
water research 44 (2010) 533–544 537
Fig. 2 – Time-weighted average concentrations (ng/L ± SD; ) and enantiomer fractions (EF) of chiral drugs (,) in LLB treated
wastewater during six sampling periods from July 2007 to August 2008 (Table S1). Hash marks on x-axis indicate break in
continuous sampling between September and October 2008. Significant concentration differences are indicated by lowercase
letters while EF differences are indicated by upper-case letters. Plots without letters do not have concentrations or EFs,
as appropriate, statistically different from one another over sampling periods. Non-racemic EFs are indicated by an asterisk,
and EFs that were corrected for enantiomer-specific matrix effects are indicated by ^. RS, racemic standards. Error bars (SD)
represent propagated error from replicate POCIS, analytical processing and instrumental uncertainty, and error associated
with POCIS sampling rates.
Without measurements of influent concentration, we
cannot discern whether temporal concentrations in effluent
are more influenced by changes in removal efficiency, as
suggested as likely by many studies (Hua et al., 2006; Kim and
Carlson, 2007; Loraine and Pettigrove, 2006; Sacher et al., 2008;
Vieno et al., 2005) or in use patterns over time, which may be
the case for antibiotics. In a Swiss study (McArdell et al., 2003),
higher macrolide concentrations were found in winter than in

538
other periods, consistent with our observations for the macrolides
clarithromycin and erythromycin (Fig. 2). In the Swiss
study, as with this study, raw wastewater concentrations
were not measured so removal efficiency could not be
assessed. However, the Swiss authors cited a personal
communication which indicated that macrolide sales were
higher in winter, indicating that increased use may contribute
to higher environmental concentrations.
The winter concentrations in LLB treated effluent, which
were sampled mostly at the same time as those at GB and CR
(see Section 3.1.3), changed over time for some but not all of
our target analytes (Fig. 2). Concentration changes were less
prevalent among the three winter sampling months,
compared with other periods. Among the sampling periods
between December and April (Tukey’s test), there were
statistically significant changes in effluent concentration for
only three drugs. The celecoxib concentration in December/
January was three times that in January/March and twice that
in March/April with magnitudes of difference of 20 and 15 ng/L,
respectively. Sotalol had December/January concentrations
were eight times that in both January/March and March/April
and magnitudes of 33 ng/L for both comparisons. While the
concentration of citalopram in January/March was 50% higher
than that in March/April, the magnitude of this difference was
only 2 ng/L. The general lack of effluent concentration change
over the winter indicates that there is no change in drug use
patterns and/or in the ability of aerated lagoons to remove
drugs effectively over this season.
It is important to note that concentration changes in
WWTPs serving small communities may well be sensitive to
changes in use patterns of the population served. If a drug was
completely excreted from the body in unchanged form and
did not undergo any removal during wastewater treatment,
then the addition of a single additional individual’s twice-aday,
100 mg active ingredient dose at LLB would result in an
effluent concentration change of 100 ng/L, which would rival
existing concentrations at LLB (Fig. 2)! In actuality, only
a fraction of most drugs are excreted as parent compounds,
and WWTP treatment provides some elimination. For
example, 3% of carbamazepine is excreted unmetabolized
from humans, with 30% removal based on a survey of Canadian
WWTPs (Miao et al., 2005). Assuming these conditions
hold for LLB, then a 2 ng/L change in effluent concentration, or
a 3–10% increase (Fig. 2), would result from the same dosage.
However, without specific usage information, excretion data,
and WWTP-specific removal efficiencies, we cannot make
more specific predictions of the sensitivity of drug concentrations
in wastewater on usage in small communities.
3.1.2. Tertiary treatment: Gold Bar and Capital Region
The concentration of detectable drugs (Table 1, Table S3) in GB
effluent ranged from 0.03 (omeprazole) to 1350 ng/L (diclofenac)
while that in CR ranged from 0.09 (roxithromycin) to
2520 ng/L (diclofenac). These concentrations are similar to
grab sample measurements for our analytes in other Canadian
WWTP effluents (Hua et al., 2006; Lishman et al., 2006;
Miao et al., 2005), indicating that our passive sampler derived
TWA concentrations were comparable with previous work.
However, concentrations of chiral drugs in this study were
generally lower by a factor of at least 10 compared with a grab
water research 44 (2010) 533–544
sample from GB in April 2007 (Macleod et al., 2007b), suggesting
possible annual or other temporal changes in
concentrations as discussed below.
Since only two samplers were deployed in each of GB and
CR during each sampling period (Table S1, Table S3), temporal
trends could not be statistically assessed. However, few
temporal trends were evident. In GB effluent, the temazepam
concentration in March was double (range >100 ng/L) that in
December/January, and in CR effluent, the naproxen concentration
increased seven-fold over the winter (range >700 ng/L).
Also in CR effluent, the roxithromycin concentration in
December/January was nine times higher than that in both
February/March and March (range 0.8 ng/L), whereas the
opposite trend was seen for sulfamethazine (increased three
times over time, range 3 ng/L). Fenoprofen, ketoprofen, sulfadimethoxine,
sulfisoxazole, and tadalafil were not detected
in either urban WWTP above their respective DLs.
When the TWA concentrations for each drug were pooled
for each WWTP (n ¼ 6 for each WWTP), statistical comparisons
revealed no significant differences between urban
effluent concentrations for any drug (Table 1). As with the
aerated lagoon effluent, there was little evidence of changes in
drug concentration over the monthly integrated sampling
periods. Seasonality in removal efficiency and/or or drug use
patterns could not be assessed for the tertiary plants as only
winter data was collected. The concentrations of each drug
were similar between the two tertiary plants (Table 1), likely
a reflection of similar per capita drug and water use between
the populations since treatment processes are nearly identical
in the two plants (Letzel et al., 2009). We do not expect
concentrations in these large-community WWTPs to be
influenced significantly by use changes in a small number of
individuals; using the same assumptions as in the previous
section, the same individual’s carbamazepine dose change
would result in a 0.02 ng/L increase in GB concentrations, an
insignificant amount.
3.1.3. Comparison of concentrations in aerated lagoon and
tertiary treated effluents in winter
Drug concentrations in effluents depend on a number of
factors. Removal efficiency during treatment is known to vary
by drug and by WWTP (Batt et al., 2007; Letzel et al., 2009).
Differences in sewage retention time (Batt et al., 2007; Göbel
et al., 2007; Kimura et al., 2007), and/or the co-metabolic
activity of microbes (Joss et al., 2006; Kim and Carlson, 2007;
Kimura et al., 2007; Yu et al., 2006) can produce differences in
the efficacy of wastewater treatment for removal of some
drugs. For other chemicals, sorption capacity, rather than
retention time or biodegradation, can be most important for
removal during treatment (Carballa et al., 2007; Joss et al.,
2006). Differences in sunlight attenuation while wastewater is
in outdoor tanks (tertiary) and lagoons may also play a role for
some drugs, as the turbidity of the wastewater may vary, thus
varying the amount of sunlight attenuation and thus the
amount available for photolysis. Drugs in GB and CR were
exposed to sunlight for less than 24 h, and received at least 8 s
of low power UV exposure (City of Edmonton, 2009) for
disinfection of effluent before release. The drugs in LLB are
exposed to sunlight for as long as the daylight hours of 90 days
(Duigou, 2006), providing more time for photolysis.

Temperature can also play a role, as previously noted. In
addition, both GB and CR use secondary activated sludge
treatment. These tanks are warmed in the winter to maintain
efficiency, especially in cold climates such as that in Alberta.
Thus, the water temperatures in those tanks may be higher
than that in the LLB lagoon pond under ice (w4 C). As a result,
biological activity capable of biodegrading drugs may be more
active in GB and CR compared with LLB.
Notwithstanding these potential differences in incidental
removal during treatment, the concentrations of several drugs
detected at all three plants were generally comparable, with
CR having the highest total concentration (Table 1). Most
drugs (17 out of 24 detectable) were found in all three WWTPs
in winter (Table 1). No differences in winter pooled TWA
concentrations (three winter sampling periods with n ¼ 9 for
LLB, and n ¼ 6 for GB and for CR) among the three effluents for
atenolol, celecoxib, codeine, metoprolol, naproxen, paroxetine,
temazepam, triclosan and trimethoprim. For eight other
drugs, the concentrations in urban effluents were from 2 to 12
times higher than that in rural effluent, with concentration
differences that ranged from 9 to 1600 ng/L (Table 1). For citalopram
and gemfibrozil, the concentrations at GB were
greater than those at LLB. For carbamazepine, citalopram,
clarithromycin, diclofenac, erythromycin, gemfibrozil,
propranolol, and sotalol, the concentrations in CR effluent
were greater than those in LLB effluent. Between the two
urban WWTPs, the concentrations of citalopram, propranolol
and sotalol at CR were greater than those at GB. Several drugs
(fluoxetine, paroxetine, roxithromycin, sildenafil, sulfamethazine,
sulfapyridine, vardenafil), infrequently or not detected
in the rural WWTP effluent, were consistently detected in
the urban WWTP effluents (Tables S2–S3). These drugs were
generally present at low concentrations (100 ng/L). The discrepancy with sulfapyridine may be
explained by differences in drug use among the communities,
but is more likely caused by treatment differences. Sulfapyridine
is generally administered via sulfasalazine, which is
cleaved in the colon to release sulfapyridine (10–35%), while
another 30–50% is released as sulfasalazine and a metabolite
(Neumann, 1989), both of which may be cleaved in biological
wastewater treatment to release sulfapyridine (Göbel et al.,
2005). It is possible that such cleavage does not occur in the
aerated lagoons, leading to a lower concentration of sulfapyridine
in LLB effluent. For the other drugs, differences
reflect differences between the communities in terms of water
and/or drug use, and/or a difference in the fate of drugs in the
WWTPs (Hua et al., 2006; Kim and Carlson, 2007; Loraine and
Pettigrove, 2006; Sacher et al., 2008; Vieno et al., 2005).
3.2. Loadings of pharmaceuticals to surface waters from
aerated lagoon and tertiary treated effluents
Average daily loadings (Table S4) in winter were calculated for
each plant by multiplying the average daily water discharge by
the TWA concentration of each drug during each sampling
period. Total winter loadings for GB, CR, and LLB were 700, 400
and 4 g/day, respectively. For the winter sampling periods,
average daily loadings from GB were generally twice that
from CR. Average daily loadings from LLB were on the order
water research 44 (2010) 533–544 539
of mg/day. This was at least an order of magnitude lower than
loadings from GB and CR, as expected based on the populations
served (Letzel et al., 2009). The above result suggests
that population-weighted per capita TWA daily loadings
would be similar among the three WWTPs. Indeed, this was
the case for most drugs, for which there were no statistically
significant differences among population-weighted average
daily loadings (Fig. 3), e.g., atenolol, celecoxib, clarithromycin,
codeine, gemfibrozil, metoprolol, naproxen, temazepam,
triclosan, and trimethoprim. Urban per capita average daily
loadings were greater than those from the rural WWTP in
a few cases: CR was greater than LLB (carbamazepine,
citalopram, diclofenac, propranolol, erythromycin, sotalol),
GB was greater than LLB (citalopram), while the per capita
loading from CR was greater than GB for three drugs
(citalopram, propranolol, sotalol). These per capita loading
differences, although statistically significant, may not be
practically significant, as all were all less than an order of
magnitude (two to eight times) and the absolute differences
ranged from 2 (propranolol, CR > GB) to 390 mg/person per day
(diclofenac, CR > LLB). All of these differences in loadings
between WWTPs were also noted as differences in concentration.
Loading differences were not found for clarithromycin
and gemfibrozil, for which differences in concentration were
found among WWTPs. Thus, we conclude that population is
a good predictor of human-use drug release into receiving
waters, at least for over-the-counter or long-term prescription
drugs such as those in this study.
Such predictions are likely to be more accurate when based
on direct measurements, particularly from demographically
similar regions (Stamm et al., 2008), and most accurate when
also based on comparable wastewater treatment. In a ten-year
study of drugs in the Rhine, concentrations of diclofenac and
carbamazepine were found to be relatively constant and given
consistency in use patterns, and although raw wastewater
concentrations were not measured, the authors attributed any
temporal variations to differences in WWTP removal
efficiency (Sacher et al., 2008). Their conclusion was that any
measures that had been implemented to reduce loadings were
thus far ineffective. Because our estimates are based on
continuous measurements of pharmaceutical concentrations
(Letzel et al., 2009; Miao et al., 2005; Stamm et al., 2008), rather
than instantaneous grab sampling, they are likely to be more
robust as TWA concentrations from passive samplers are less
susceptible to short-term fluctuations in concentration over
periods shorter than the sampler deployment periods. Efforts
have been undertaken to find a suitable mechanism for
removal of drugs from the waste stream (Ikehata et al., 2008),
and such efforts should continue, because as our results
further highlight, even sophisticated WWTPs like GB and CR
produce wastewater with drug concentrations that are
comparable with water treated by aerated lagoons.
As previously noted, loadings at WWTPs serving small
communities are potentially more sensitive to small changes
in use patterns than larger communities. The twice-a-day,
100 mg/dose from one individual at LLB would result in
a loading increase of 50 mg/person per day at 100% excretion
and no WWTP removal, or 1 mg/person per day (4% increase,
Fig. 3) using the carbamazepine assumptions previously
given. At GB, the equivalent increase in loadings would

540
water research 44 (2010) 533–544
Fig. 3 – Time-weighted average per capita daily loadings (:, mg/person per day) and enantiomer fractions (,, EF), both ± SD,
of drugs at three WWTPs: GB, Gold Bar; CR, Capital Region (n [ 6), LLB, Lac La Biche. Differences in daily loadings are
indicated by lower-case letters while EF Statistically significant differences are indicated by upper-case letters. Plots without
letters do not have loadings or EFs, statistically different from one another over sampling periods. Asterisks indicate EFs
that are significantly non-racemic compared with racemic standards. Error bars (SD) represent propagated error from
replicate POCIS, analytical processing and instrumental uncertainty, and error associated with POCIS sampling rates.

generally be insignificant (Fig. 3), at 0.27 and 0.006 mg/person
per day, respectively.
3.3. Enantiomer fractions of chiral drugs in
treated effluents
3.3.1. Temporal trends in enantiomer fractions of chiral drugs
in aerated lagoon effluent
We measured EFs in LLB to discern whether temporal changes
existed in enantiomer composition of chiral drugs released by
WWTPs. Changes in EF may arise from differential interaction
of chiral drug enantiomers with other chiral chemicals (e.g.,
enzymes). These may result from temporal differences in
enantiomer-specific metabolism of patients (Mehvar and
Brocks, 2001; Schmekel et al., 1999), and/or from enantiomerspecific,
microbial enzyme mediated biotransformations
during wastewater treatment, as previously noted for atenolol
(MacLeod et al., 2007b; Nikolai et al., 2006), citalopram
(MacLeod et al., 2007b) and propranolol (Fono and Sedlak,
2005; MacLeod et al., 2007b; Nikolai et al., 2006), but not
metoprolol (Fono and Sedlak, 2005; MacLeod et al., 2007b;
Nikolai et al., 2006), salbutamol (MacLeod et al., 2007b), or
sotalol (MacLeod et al., 2007b). Because of enantiomer-specific
metabolism by the target organism, chiral pharmaceuticals
may not necessarily enter the environment with the same
enantiomeric composition as the dose. This is in contrast to
chiral persistent organic pollutants, which generally enter the
environment as a racemate, or sometimes as a single enantiomer.
Thus, changes in enantiomeric composition from the
excreted EF, not just from a racemic value of 0.5, would be
environmentally significant as such a change would indicate
post-release biochemical weathering. However, given that
few environmental measurements of drug EFs exist to date,
our discussion here focuses on changes in EF from the racemate,
the most likely enantiomeric composition of formulated
chiral drugs not sold as single enantiomers (e.g., naproxen).
With the exception of tempazepam, chiral drugs were
generally non-racemic in LLB effluent (Fig. 2) compared with
racemic standards (Dunnett’s test). Atenolol was enriched in
( )-atenolol during some sampling periods, with changes over
time, consistent with previous observations in 2005 at LLB
(Nikolai et al., 2006). Citalopram was mainly enriched in the
first-eluted enantiomer except during August/September.
Metoprolol was enriched in the E2-enantiomer when nonracemic,
which was not previously observed at LLB in individual
grab samples (Nikolai et al., 2006). Salbutamol was
always enriched in the second eluted enantiomer, except
during January/March wherein it was E1-salbutamol that was
enriched. Sotalol was always non-racemic, with no change in
EF over time. Temazepam was racemic with one exception,
wherein E 2-temazepam was enriched. Propranolol EFs could
not be measured (supplemental data).
Temporal changes in EF were observed (Tukey’s test) for all
drugs except sotalol in LLB effluent (Fig. 2, Table S5). Without
direct measurement of the EF in influent and during the
treatment process, we cannot conclusively determine
whether EFs in effluent arose from metabolism prior to bodily
excretion or from processes during treatment. The latter may
be plausible, given changes in activity and composition of
microbial consortia during biological sewage treatment from
water research 44 (2010) 533–544 541
temperature changes over the seasons. For drugs such as
citalopram and salbutamol, changes in EF may also reflect
differences in availability or use of single-enantiomer
formulations (Maier et al., 2001). Because of the temporal
changes in EF observed, our results show that an EF measured
at one time point may not necessarily be characteristic of the
enantiomer composition of chiral drugs discharged from
a WWTP, and hence insufficient to characterize exposures in
receiving waters.
3.3.2. Comparison of chiral drug EFs from aerated lagoons
and tertiary treated wastewater
Winter EFs from LLB and from the urban tertiary WWTPs were
compared to gain insight on whether EFs could be predicted by
treatment type, and may possibly be caused by similar human
metabolism and release and/or microbial degradation during
WWTP treatment. Due to low sample sizes, statistical tests and
trend analyses were not carried out on the temporal EF data
from the urban tertiary WWTPs. Instead, winter EFs were pooled
for each drug from GB and CR and compared with EF standard
(Dunnett’s test), to each other (t-test, with Welch’s correction for
unequal variances as needed), and to the pooled EFs from the
three winter sampling periods at LLB (Tukey’s test). Standard
deviations in pooled data reflect propagated variability from
both temporal changes in EFsample and from replicates.
The two tertiary WWTP effluents had similar EFs during
winter for all chiral drugs except sotalol (Fig. 3), suggesting
that both treatment plants were processing those drugs in the
same stereoselective manner during microbial treatment.
At all three WWTPs (Fig. 3), metoprolol EFs were racemic in
effluents, while sotalol was enriched in the E 2-enantiomer and
differed in EF between the tertiary effluents. Racemic metoprolol
was previously observed in both GB and LLB effluents
(MacLeod, 2009; MacLeod et al., 2007b; Nikolai et al., 2006) and
in the effluent of a WWTP in Dallas TX (Fono et al., 2006), with
no change in EF from raw to treated wastewater (MacLeod
et al., 2007b). After correction for enantiomer-specific matrix
effects (MacLeod et al., 2007b), sotalol was previously found to
be racemic in GB effluent (MacLeod et al., 2007b) via grab
samples. This observation suggests that changes in sotalol EFs
might have occurred over time in GB effluent; however, we
cannot unequivocally conclude that this must be the case
given our earlier caveats that a single EF cannot necessarily
characterize the enantiomer composition. No statistically
significant differences were observed for salbutamol EFs in
the three effluents, despite enrichment of E 2-salbutamol in
both tertiary effluents. This observation is largely a result of
the high variability in salbutamol EFs in LLB and CR. In turn,
the variability may be a function of the drug’s input to
WWTPs, given that salbutamol is available as an R-salbutamol-only
formulation, while S-salbutamol is metabolized
slower than its antipode (Schmekel et al., 1999). Citalopram
(Fig. 3) was enriched in the E 1-enantiomer in all effluents, with
differences in EF between tertiary and aerated lagoon output.
As with salbutamol, this difference could be influenced
by differences in prescription patterns, given the availability
of escitalopram, the S-enantiomer formulation. Citalopram
was previously found with similar EFs (0.55–0.65) in GB
effluent (MacLeod et al., 2007b). Temazepam was enriched in
the E2-enantiomer at both GB and LLB, but racemic in CR,

542
wherein its EF was different from that in LLB, but not from that
in GB. No information regarding temazepam EFs in treated
effluent is available, except our previous results (MacLeod,
2009). For atenolol, all effluents had EFs that were different
from one another, with GB containing racemic atenolol, CR
containing more (þ)-atenolol and LLB containing more
( )-atenolol. The EF of atenolol from healthy renal excretion is
0.476 (Mehvar and Brocks, 2001), and 90% of atenolol is
excreted unchanged in urine (Bourne, 1981). Thus, a process
other than healthy renal excretion, and potentially specific to
microbial communities of WWTPs, was influencing atenolol
EFs in effluent. Likewise, the differences in EFs observed
among WWTP effluents is likely due, at least in part, to variations
in enantioselective microbial biotransformation of
drugs during wastewater treatment.
As we have noted, it is not clear what the enantiomeric
composition of drugs may be upon excretion from humans,
but non-racemic release is likely. As such, we must delineate
between environmental biotransformation and in changes in
enantiomeric composition of the source material in assessing
the environmental fate of drug enantiomers, as has been done
for chiral pesticides (e.g., metolachlor) also released in nonracemic
amounts (Kurt-Karakus et al., 2008).
4. Conclusions
Although other studies have used POCIS to monitor drugs in
treated wastewater, none have done so for more than 30 days
(Jones-Lepp et al., 2004; Zhang et al., 2008). Herein, we monitored
LLB effluent for 271 days and found that the concentration
of drugs in effluent varied over seasonal time scales. Other
studies have mainly focused on larger urban centers. By
comparing drugs in both rural and urban wastewater effluents,
we find that pharmaceutical loadings to aquatic ecosystems
scale with population. However, we also found that the
concentrations of drugs in effluent from WWTPs with biological
activated sludge treatment and UV disinfection were
comparable with concentrations in effluent from simpler
treatment methods like aerated lagoons. Although absolute
loadings were generally lower, drugs were still detectable in
effluent from small communities. Thus, if pharmaceutical
contamination of surface water presents a risk to aquatic life,
that risk is also present near smaller centers, especially those
which are landlocked and not discharging to a large surface
water body to provide effluent dilution.
Overall, temporal changes in EF were found for all chiral
drugs except sotalol, with temporal changes in concentration
for all chiral drugs except atenolol and metoprolol. The
concentration of atenolol and metoprolol in LLB effluent did
not change over time, but the EFs of these drugs did change.
Conversely, although the concentration of sotalol in LLB
effluent changed over time, the EF remained constant.
Although temporal changes in concentration may not be
occurring, the relative proportions of enantiomers being
released to the environment may be changing over time. Also,
for some drugs the EF may remain constant, but the concentration
of the drug may change.
Similar conclusions can be drawn from the comparison
between EFs and per capita daily loadings among the three
water research 44 (2010) 533–544
WWTPs. Loading and EF differences were found among
WWTPs for citalopram and sotalol, while for atenolol and
temazepam, there were no differences in loading but there
were changes in EF, and for metoprolol there were no differences
in loading or EF among the three WWTPs. Simply put,
although concentrations and loadings may be accurately
predicted based on population-weighted measured effluent
concentrations in other treatment plants, the EFs of chiral
drugs released by one WWTP may not necessarily reflect that
released by another, probably from site-specific biochemical
activity. Thus, enantiomer-specific monitoring of chiral drugs
may be useful, particularly for drugs with enantiomer-specific
toxicity to aquatic life (Stanley et al., 2006).
Acknowledgements
We thank D. Bleackley, V. Cooper, M. Ross, B. Asher and
E. McClure (University of Alberta), R. Litwinow and D. Seehagel
(GB WWTP), D. Cikaluk (CR WWTP) and G. Siebold (LLB WWTP)
for sampling assistance, and H. Li (Trent University) for
providing unpublished POCIS sampling rates for citalopram and
sotalol. Funding was provided to C.S.W. from the Canada
Research Chairs Program, Canada’s Natural Sciences and Engineering
Research Council (NSERC), and the Society of Environmental
Toxicology and Chemistry (SETAC) Early Career Award
for Applied Ecological Research, cosponsored by the American
Chemistry Council; and to S.L.M. as fellowships from NSERC,
Alberta Ingenuity Fund, and the American Chemical Society
Division of Analytical Chemistry, sponsored by DuPont.
Appendix.
Supplementary data
Information about sampling locations and times, analytical
and data treatment procedures, concentrations, loadings, and
EFs are available in the online supplementary data to
Supplementary data associated with this article can be found
in the online version at doi:10.1016/j.watres.2009.09.056.
references
Alberta Capital Region Wastewater Commission, 2009. http://
www.acrwc.ab.ca accessed February 25.
Alvarez, D.A., Petty, J.D., Huckins, J.N., Jones-Lepp, T.L.,
Getting, D.T., Goddard, J.P., Manahan, S.E., 2004. Development
of a passive, in situ, integrative sampler for hydrophilic organic
contaminants in aquatic environments. Environ. Toxicol.
Chem. 23, 1640–1648.
Andreozzi, R., Raffaele, M., Nicklas, P., 2003. Pharmaceuticals in
STP effluents and their solar photodegradation in the aquatic
environment. Chemosphere 50, 1319–1330.
Batt, A.L., Kim, S., Aga, D.S., 2007. Comparison of the occurrence of
antibiotics in four full-scale wastewater treatment plants with
varying designs and operations. Chemosphere 68, 428–435.
Bourne, G.R., 1981. The metabolism of b-adrenoreceptor blocking
drugs. In: Bridges, J.W., Chasseaud, L.F. (Eds.), Progress in Drug
Metabolism 6. John Wiley & Sons Ltd, pp. 77–110.

Carballa, M., Omil, F., Ternes, T., Lema, J.M., 2007. Fate of
pharmaceutical and personal care products (PPCPs) during
anaerobic digestion of sewage sludge. Water Res. 41, 2139–2150.
Chen, M., Ohman, K., Metcalfe, C., Ikonomou, M.G., Amatya, P.L.,
Wilson, J., 2006. Pharmaceuticals and endocrine disruptors in
wastewater treatment effluents and in the water supply
system of Calgary, Alberta, Canada. Water Qual. Res. J. Can.
41, 351–364.
City of Edmonton, 2009. http://www.edmonton.ca. accessed
February 25.
Coetsier, C.M., Spinelli, S., Lin, L., Roig, B., Touraud, E., 2009.
Discharge of pharmaceutical products (PPs) through
a conventional biological sewage treatment plant: MECs vs
PECs? Environ. Int 35, 787–792.
Conkle, J.L., White, J.R., Metcalfe, C.D., 2008. Reduction of
pharmaceutically active compounds by a lagoon wetland
wastewater treatment system in Southeast Louisiana.
Chemosphere 73, 1741–1748.
Daughton, C.G., Ternes, T.A., 1999. Pharmaceuticals and personal
care products in the environment: agents of subtle change?
Environ. Health. Perspect 107, 907–938.
Duigou, E., 2006. Town of Lac La Biche, Personal communication.
EMEA, 2006. Note for guidance on environmental risk assessment of
medicinal products for human use. The European Agency for the
Evaluation of Medicinal Products (EMEA) CMPC/SWP/4447/00.
Environment Canada, 2009. http://www.ec.gc.ca/soer-ree/
English/Indicator_series/new_issues.cfm?issue_id¼7&tech_
id¼28#bio_pic. accessed February 27.
Fent, K., Weston, A.A., Caminada, D., 2006. Ecotoxicology of
human pharmaceuticals. Aquat. Toxicol 76, 122–159.
Fono, L.J., Sedlak, D.L., 2005. Use of the chiral pharmaceutical
propranolol to identify sewage discharges into surface waters.
Environ. Sci. Technol 39, 9244–9252.
Fono, L.J., Kolodziej, E.P., Sedlak, D.L., 2006. Attenuation of
wastewater-derived contaminants in a effluent-dominated
river. Environ. Sci. Technol 40, 7257–7262.
Göbel, A., McArdell, C.S., Joss, A., Siegrist, H., Giger, W., 2007. Fate
of sulfonamides, macrolides, and trimethoprim in different
wastewater treatment methodologies. Sci. Total Environ 372,
361–371.
Göbel, A., Thomsen, A., McArdell, C.S., Joss, A., Giger, W., 2005.
Occurrence and sorption behavior of sulfonamides,
macrolides, and trimethoprim in activated sludge treatment.
Environ. Sci. Technol 39, 3981–3989.
Halling-SBrensen, B., Nielsen, S.N., Lanzky, P.F., Ingerslev, F.,
Holten-LützhBft, H.C., JBrgensen, S.E., 1998. Occurrence, fate
and effects of pharmaceutical substances in the environment
– a review. Chemosphere 36, 357–393.
Harner, T., Wiberg, K., Norstrom, R., 2000. Enantiomer fractions
are preferred to enantiomer ratios for describing chiral
signatures in environmental analysis. Environ. Sci. Technol
34, 218–220.
Hua, W.Y., Bennett, E.R., Miao, X.-S., Metcalfe, C.D., Letcher, R.J.,
2006. Seasonality effects on pharmaceuticals and s-triazine
herbicides in wastewater effluent and surface water from the
Canadian side of the Upper Detroit River. Environ. Toxicol.
Chem. 25, 2356–2365.
Ikehata, K., El-Din, M.G., Snyder, S.A., 2008. Ozonation and
advanced oxidation treatment of emerging organic pollutants
in water and wastewater. Ozone Sci. Eng 30, 21–26.
Jones-Lepp, T.L., Alvarez, D.A., Petty, J.D., Huckins, J.N., 2004.
Polar organic chemical integrative sampling and liquid
chromatography-electrospray/ion-trap mass spectrometry for
assessing selected prescription and illicit drugs in treated
sewage effluents. Arch. Environ. Contam. Toxicol 47, 427–439.
Joss, A., Zabcynski, S., Göbel, A., Hoffmann, B., Löffler, D.,
McArdell, C.S., Ternes, T.A., Thomsen, A., Siegrist, H., 2006.
Biological degradation of pharmaceuticals in municipal
water research 44 (2010) 533–544 543
wastewater treatment: proposing a classification scheme.
Water Res. 40, 1686–1696.
Kim, S.-C., Carlson, K., 2007. Temporal and spatial trends in
the occurrence of human and veterinary antibiotics in
aqueous and river sediment matrices. Environ. Sci. Technol
41, 50–57.
Kimura, K., Hara, H., Watanabe, Y., 2007. Elimination of selected
acidic pharmaceuticals from municipal wastewater by an
activated sludge system and membrane bioreactors. Environ.
Sci. Technol 41, 3708–3714.
Kolpin, D.W., Furlong, E.T., Meyer, M.T., Thurman, E.M., Zaugg, S.D.,
Barber, L.B., Buxton, H.T., 2002. Pharmaceuticals, hormones,
and other organic wastewater contaminants in U.S. streams,
1999–2000: a national reconnaissance. Environ. Sci. Technol 36,
1202–1211.
Kurt-Karakus, P.B., Bidleman, T.F., Muir, D.C.G., Cagampan, S.J.,
Struger, J., Sverko, E., Small, J.M., Jantunen, L.M., 2008. Chiral
current-use herbicides in Ontario streams. Environ. Sci.
Technol 42, 8452–8458.
Lajeunesse, A., Gagnon, C., Sauvé, S., 2008. Determination of basic
antidepressants and their N-desmethyl metabolites in raw
sewage and wastewater using solid-phase extraction and
liquid chromatography-tandem mass spectrometry. Anal.
Chem. 80, 5325–5333.
Letzel, M., Metzner, G., Letzel, T., 2009. Exposure assessment of
the pharmaceutical diclofenac based on long-term
measurements of the aquatic input. Environ. Int 35, 363–368.
Lin, A.Y.-C., Reinhard, M., 2005. Photodegradation of common
environmental pharmaceuticals and estrogens in river water.
Environ. Toxicol. Chem. 24, 1303–1309.
Lin, A.Y.-C., Plumlee, M.H., Reinhard, M., 2006. Natural
attenuation of pharmaceuticals and alkylphenol
polyethoxylate metabolites during river transport:
photochemical and biological transformation. Environ.
Toxicol. Chem. 25, 1458–1464.
Lishman, L., Smyth, S.A., Sarafin, K., Kleywegt, S., Toito, J.,
Peart, T., Lee, B., Servos, M., Beland, M., Seto, P., 2006.
Occurrence and reductions of pharmaceuticals and personal
care products and estrogens by municipal wastewater
treatment. Sci. Total Environ 367, 544–588.
Lissemore, L., Hao, C., Yang, P., Sibley, P.K., Mabury, S.,
Solomon, K.R., 2006. An exposure assessment for selected
pharmaceuticals within a watershed in Southern Ontario.
Chemosphere 64, 717–729.
Loraine, G.A., Pettigrove, M.E., 2006. Seasonal variations in
concentrations of pharmaceuticals and personal care
products in drinking water and reclaimed wastewater in
Southern California. Environ. Sci. Technol 40, 687–695.
MacLeod, S.L., 2009. Pharmaceutical occurrence and fate in
wastewater and receiving surface waters in two Alberta
watersheds. PhD dissertation. University of Alberta,
Edmonton, Alberta.
MacLeod, S.L., McClure, E.L., Wong, C.S., 2007a. Laboratory
calibration and field deployment of the Polar Organic
Chemical Integrative Sampler for pharmaceuticals and
personal care products in wastewater and surface water.
Environ. Toxicol. Chem. 26, 2517–2529.
MacLeod, S.L., Sudhir, P., Wong, C.S., 2007b. Stereoisomer
analysis of wastewater-derived beta-blockers, selective
serotonin re-uptake inhibitors, and salbutamol by highperformance
liquid chromatography-tandem mass
spectrometry. J. Chromatogr. A 1170, 23–33.
Maier, N.M., Franco, P., Lindner, W., 2001. Separation of
enantiomers: needs, challenges, perspectives. J. Chromatogr.
A 906, 3–33.
Managaki, S., Murata, A., Takada, H., Tuyen, B.C., Chiem, N.H.,
2007. Distribution of macrolides, sulfonamides, and
trimethoprim in tropical waters: ubiquitous occurrence of

544
veterinary antibiotics in the Mekong Delta. Environ. Sci.
Technol 41, 8004–8010.
McArdell, C.S., Molnar, E., Suter, M.J.-F., Giger, W., 2003.
Occurrence and fate of macrolide antibiotics in wastewater
treatment plants and in the Glatt Valley Watershed.
Switzerland. Environ. Sci. Technol 37, 5479–5486.
Mehvar, R., Brocks, D.R., 2001. Stereospecific pharmacokinetics
and pharmacodynamics of beta-adrenergic blockers in
humans. J. Pharm. Pharm. Sci. 4, 185–200.
Miao, X.-S., Yang, J.-J., Metcalfe, C.D., 2005. Carbamazepine and its
metabolites in wastewater and in biosolids in a municipal
wastewater treatment plant. Environ. Sci. Technol 39, 7469–7475.
Neumann, J., 1989. Untersuchungen zur Bioverfügbarkeit und
Pharmakokinetik von Sulfasalazin und seinen Metaboliten.
PhD thesis. Free University of Berlin.
Nikolai, L.N., McClure, E.L., MacLeod, S.L., Wong, C.S., 2006.
Stereoisomer quantification of the b-blocker drugs atenolol,
metoprolol, and propranolol in wastewaters by chiral highperformance
liquid chromatography-tandem mass
spectrometry. J. Chromatogr. A 1131, 103–109.
Sacher, F., Ehmann, M., Gabriel, S., Graf, C., Brauch, H.-J., 2008.
Pharmaceutical residues in the river Rhine-results of a onedecade
monitoring programme. J. Environ. Monit. 10, 664–670.
Schmekel, B., Rydberg, I., Norlander, B., Sjöswärd, K.N., Ahlner, J.,
Anderson, R.G.G., 1999. Stereoselective pharmacokinetics of
S-salbutamol after administration of the racemate in healthy
volunteers. Eur. Respir. J 13, 1230–1235.
Shaw, M., Mueller, J.F., 2009. Time integrative passive sampling:
how well do Chemcatchers integrate fluctuating pollutant
concentrations? Environ. Sci. Technol 43, 1443–1448.
Siebold, G., 2008. Lac La Biche County, Personal communication.
water research 44 (2010) 533–544
Stamm, C., Alder, A.C., Fenner, K., Hollender, J., Kraus, M.,
McArdell, C.S., Ort, C., Schneider, M.K., 2008. Spatial and
temporal patterns of pharmaceuticals in the aquatic
environment. Geog. Compass 2/3, 920–955.
Stanley, J.K., Ramirez, A.J., Mottaleb, M., Chambliss, C.K.,
Brooks, B.W., 2006. Enantiospecific toxicity of the b-blocker
propranolol to Daphnia magna and Pimephales promelas.
Environ. Toxicol. Chem. 25, 1780–1786.
Statistics Canada, 2009. http://www12.statcan.ca. accessed
February 25.
Stülten, D., Zühlke, S., Lamshöft, M., Spiteller, M., 2008.
Occurrence of diclofenac and selected metabolites in sewage
effluents. Sci. Total Environ 405, 310–316.
Vieno, N.M., Tuhkanen, T., Kronberg, L., 2005. Seasonal variation
in the occurrence of pharmaceuticals in effluents from
a sewage treatment plant and in the recipient water. Environ.
Sci. Technol 39, 8220–8226.
Vasskog, T., Anderssen, T., Pedersen-Bjergaard, S., Kallenborn, R.,
Jensen, E., 2008. Occurrence of selective serotonin reuptake
inhibitors in sewage and receiving waters at Spitsbergen and
in Norway. J. Chromatogr. A 1185, 194–205.
Wong, C.S., 2006. Environmental fate processes and biochemical
transformations of chiral emerging organic pollutants. Anal.
Bioanal. Chem. 386, 544–558.
Yu, J.T., Bouwer, E.J., Coelhan, M., 2006. Occurrence and
biodegradability studies of selected pharmaceuticals and personal
care products in sewage effluent. Agric. Water Mgmt 86, 72–80.
Zhang, Z., Hibberd, A., Zhou, J.L., 2008. Analysis of emerging
contaminants in sewage effluent and river water:
comparison between spot and passive sampling. Anal. Chim.
Acta 607, 37–44.

Degradation of fifteen emerging contaminants at mgL L1 initial
concentrations by mild solar photo-Fenton in MWTP effluents
N. Klamerth a,b , L. Rizzo c , S. Malato a, *, Manuel I. Maldonado a ,
A. Agüera b , A.R. Fernández-Alba b
a
Plataforma Solar de Almería-CIEMAT. Carretera Senés km 4, 04200 Tabernas (Almería), Spain
b
Pesticide Residue Research Group, University of Almería, 04120 Almería, Spain
c
Department of Civil Engineering, University of Salerno, via Ponte don Melillo, 84084 Fisciano (SA), Italy
article info
Article history:
Received 9 March 2009
Received in revised form
23 September 2009
Accepted 28 September 2009
Available online 4 October 2009
Keywords:
Emerging contaminants
Pharmaceuticals treatment
Photo-Fenton
Solar photocatalysis
1. Introduction
abstract
Due to their growing use, pharmaceuticals, like the antiinflammatory
ibuprofen, the antibiotic flumequine and the
antiepileptic carbamazepine, endocrine disruptors like
bisphenol A and atrazine, personal care products like oxybenzone
and parabens (PHBA), synthetic musks and
fragrances like musk xylene and galaxolide, pesticides like
isoproturon and endosulphan and illicit drugs like THC aznd
cocaine, to name just a few, and other xenobiotic substances,
are found in increasing quantities in wastewater, surface
water, and even in drinking water (Kasprzyk-Hordern et al.,
2009; Kim et al., 2007; Mitch et al., 2003; Esplugas et al., 2007;
Ternes, 1998). Since the use of these substances cannot be
controlled or eliminated as they are ever present in our daily
water research 44 (2010) 545–554
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
* Corresponding author. Tel.: þ34 950387940; fax: þ34 950365015.
E-mail address: sixto.malato@psa.es (S. Malato).
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.watres.2009.09.059
The degradation of 15 emerging contaminants (ECs) at low concentrations in simulated and
real effluent of municipal wastewater treatment plant with photo-Fenton at unchanged pH
and Fe ¼ 5mgL 1 in a pilot-scale solar CPC reactor was studied. The degradation of those
15 compounds (Acetaminophen, Antipyrine, Atrazine, Caffeine, Carbamazepine, Diclofenac,
Flumequine, Hydroxybiphenyl, Ibuprofen, Isoproturon, Ketorolac, Ofloxacin, Progesterone,
Sulfamethoxazole and Triclosan), each with an initial concentration of 100 mgL 1 ,
was found to depend on the presence of CO3 2 and HCO3 (hydroxyl radicals scavengers) and
on the type of water (simulated water, simulated effluent wastewater and real effluent
wastewater), but is relatively independent of pH, the type of acid used for release of
hydroxyl radicals scavengers and the initial H2O2 concentration used. Toxicity tests with
Vibrio fisheri showed that degradation of the compounds in real effluent wastewater led to
toxicity increase.
ª 2009 Elsevier Ltd. All rights reserved.
lives, their release into the environment has to be optimized
and restricted, as they pose risks to the environment, public
health and aquatic systems and they are responsible for
building up microbiological resistance, feminisation of higher
organisms and ecotoxicological issues (Laville et al., 2004).
Particularly relevant examples of such emerging contaminants
(ECs), such as those mentioned above, which are ubiquitously
present in influents and effluents of MWTPs in the
high ng L 1 to low mgL 1 range, do not need to be persistent to
be hazardous, because they are introduced continuously into
the environment (Fono et al., 2006; Jackson and Sutton, 2008;
Nakada et al., 2008; Petrovic et al., 2003).
Conventional MWTPs, typically based on biological
processes, are capable of removing some substances, but nonbiodegradable
compounds may escape the treatment and be

546
released into the environment (Göbel et al., 2007; Carballa
et al., 2004; Ternes et al., 2007). ECs have been found in the
MWTP effluents at mean concentrations ranging from 0.1 to
20 mgL 1 (Castiglioni et al., 2006; Martínez Bueno et al., 2007;
Radjenović et al., 2007; Richardson, 2007; Zhao et al., 2009).
Concern about the growing problem of the continuously rising
concentrations of these compounds must be emphasized, and
therefore, the application of more thorough wastewater
treatment protocols, including the use of new and improved
technologies, is a necessary task.
Reusable water should be free of these persistent, toxic,
endocrine-disrupting or non-biodegradable substances,
(Radjenović et al., 2007; Teske and Arnold, 2008), and therefore,
an effective tertiary treatment is required to remove
these substances completely.
Among the advanced technologies that may be used to
remove these pollutants, (Gogate and Pandit, 2004; Saritha
et al., 2007; Huber et al., 2005) advanced oxidation processes
(AOPs), through the generation of hydroxyl radicals which are
able to mineralise most organic molecules yielding CO 2 and
inorganic ions as final products, are a particularly attractive
option. (Farré et al., 2005; Gebhardt and Schröder, 2007; Gültekin
and Ince, 2007; Ning et al., 2007; Rosenfeldt and Linden,
2004; Rosenfeldt et al., 2007; Ternes et al., 2003) The generation
of the OH radicals can be achieved: electrochemically (Cañizares
et al., 1999; Pelegrini et al., 2001; Zhou et al., 2005),
sonochemically (Mantzavinos et al., 2004; Papadaki et al.,
2004; Lesko et al., 2006), photochemically (Esplugas et al., 2005;
Bali et al., 2003; Bremner et al., 2006), and by homogeneous or
heterogenous catalysis (Zepp et al., 1992; Martinez et al., 2005)
in acid or basic media (Glaze et al., 1987; Hislop and Bolton,
1999; Neyens and Baeyens, 2003). Most of the AOPs make use
of a combination of either oxidants and irradiation (O 3/H 2O 2/
UV), or a catalyst and irradiation (Fe 2þ /H 2O 2; UV/TiO 2). The
drawbacks which make them economically disadvantageous
depend on the specific AOP: (i) High electricity demand (e.g.
ozone and UV-based AOPs), (ii) the relatively large amounts of
oxidants and/or catalysts (e.g. ozone, hydrogen peroxide and
iron-based AOPs), and (iii) the pH operating conditions (e.g.
Fenton and photo-Fenton). This is why, although AOPs are
well known for their capacity for oxidising and mineralising
almost any organic contaminant, commercial applications are
still scarce. Processes like photo-Fenton may be applied to
commercial applications by using solar energy as a light
source, optimizing the pH range and the amounts of oxidant/
catalyst required.
AOP efficiency in the removal of ECs has typically been
studied in demineralised water and bench scale, at initial
concentrations in the milligram-to-gram range, which is not
realistic compared to the concentrations detected in real
water and wastewater (Farré et al., 2005; Lapertot et al., 2007;
Kassinos et al., 2009; Malato et al., 2007). This work focused on
solar photo-Fenton degradation of the ECs typically found in
the effluents of MWTPs, leaving the treated wastewater suitable
for reuse. Moreover, to make the process of interest for
practical applications high iron concentrations (mM range),
excessive amounts of H 2O 2 and a pH under 3 must be avoided
(Pignatello et al., 2006). A new approach aimed at finding
a very mild photo-Fenton treatment (low iron concentration
and H2O2 dose at neutral pH), has been proposed (Moncayo-
water research 44 (2010) 545–554
Lasso et al., 2008; Klamerth et al., 2009). In this paper, a pilotscale
solar photo-Fenton treatment was run at a pH between 6
and 7, with starting concentrations of 5 mg L 1 Fe, and
50 mg L 1 H2O2. Synthetic water (SW), simulated effluent
wastewater (SE) and real effluent wastewater (RE) were tested
in this study to which a mixture of 15 ECs, consisting of
pharmaceuticals, pesticides and personal care products,
selected from a list of 80 compounds found in MWTP effluents
in previous studies (Martínez Bueno et al., 2007), were added at
low concentrations (100 mgL 1 each).
2. Material and methods
2.1. Reagents
All reagents used for chromatographic analyses, acetonitrile,
methanol, and ultrapure (MilliQ) water, were HPLC grade.
Analytical standards for chromatography analyses were
purchased from Sigma-Aldrich. Table 1 and Scheme 1 list the
15 compounds (pharmaceuticals, pesticides and personal care
products) used. Photo-Fenton experiments were performed
using iron sulphate (FeSO4$7H2O), reagent-grade hydrogen
peroxide (30% w/v), sulphuric acid and hydrochloric acid for
carbonate stripping, all provided by Panreac. The filters used
were syringe driven 0.2 mm Millex nylon membrane filters
from Millipore. The reagents used in the preparation of the
synthetic water (NaHCO 3, CaSO 4$2H 2O, MgSO 4, and KCl), and
simulated effluent wastewater (Peptone, Meat extract, Urea,
K 2HPO 4, CaCl 2$2H 2O, MgSO 4$7H 2O and NaCl) were provided by
Panreac.
2.2. Polluted waters
Taking into consideration that typical EC concentrations in the
effluent are in the 0.1–20.0 mgL 1 range, it was decided to work
at 100 mgL 1 , which is a compromise between (i) a high enough
concentration to characterise kinetics using conventional
Table 1 – LOD, LOQ and absorption l of the selected
compounds.
Name LOD
(mgL 1 )
LOQ
(mgL 1 )
Absorption
l [nm]
Acetaminophen 1.4 2.8 245
Antipyrine 2.1 4.2 205
Atrazine 0.6 1.2 223
Caffeine 0.7 1.5 205
Carbamazepine 0.8 1.6 211
Diclofenac 2.7 5.5 277
Flumequine 2.1 4.3 248
Hydroxybiphenyl 1.6 3.2 243
Ibuprofen 2.5 5.0 222
Isoproturon 1.5 3.0 205
Ketorolac 2.1 4.2 321
Ofloxacin 1.3 2.6 295
Progesterone 2.0 4.0 248
Sulfamethoxazole 1.4 2.7 267
Triclosan 5.0 10.0 280

Name Structure Name Structure
Acetaminophen
Ibuprofen
analgesic /
antipyretic
Antipyrine
analgesic
Atrazine
herbicide
Caffeine
stimulant
Carbamazepine
anticonvulsant
Diclofenac
antiinflammatory
Flumequine
broadspectrum
antibiotic
Hydroxybiphenyl
biocide
HO NH
O
N
Cl
N
O
N
N N
O
N
NH 2
O
nonsteroidal
antiinflammatory
Isoproturon
phenylurea
herbicide
H
H
N N N Ketorolac
O
N
H
N
Cl
Cl
N
N
O
OH
antiinflammatory
Ofloxacin
gramnegative
antibiotic
Progesterone
steroid
hormone
Sulfamethoxaz
ole
bacteriostatic
antibiotic
F COOH Triclosan
OH
water research 44 (2010) 545–554 547
O
N
anti-bacterial/
fungal agent
N
N
O
O
F
N
O
H 2 N S
Scheme 1 – Name and structure of the 15 selected compounds.
Cl
O
Cl
O
O
N
H
N
H
N
O
O
O
OH
OH
N
N
O
O
O
O
OH
Cl
OH

548
C/C 0
1.0
0.5
0.0
1.0
0.5
0.0
1.0
0.5
0.0
Fenton
-40 -20 0 20 40 60 80 100
t EXP , min
illumination
t 30W , min
Acetaminophen
Caffeine
Ofloxacine
Antipyrine
Sulfamethoxazole
Carbamazepine
Flumequine
Ketorolac
Atrazine
Isoproturon
Hydroxybenzone
Diclofenac
Ibuprofen
Progesterone
Triclosan
Fig. 1 – Degradation of the 15 ECs (0.1 mg L L1 each) by
photo-Fenton with 5 mg L L1 Fe without pH adjustment in
SW acidified for release of carbonates.
analytical techniques, and (ii) low enough to simulate real
conditions.
Synthetic water (SW), simulated effluent wastewater (SE) and
real effluent wastewater (RE) to which a mixture of the 15 ECs
previously listed had been added at low concentrations
(100 mgL 1 ) were tested. The experimental protocol was designed
to study solar photo-Fenton with a relatively uncomplicated
aqueous matrix (SW) first, before gradually increasing
complexity by studying SE and finally RE, in order to acquire
information on process operation.
Demineralised water used for preparing SW and SE in the
pilot plant was supplied by the Plataforma Solar de Almería
C/C 0
1.0
0.5
0.0
1.0
0.5
0.0
1.0
0.5
0.0
Fenton
t EXP , min
illumination
water research 44 (2010) 545–554
(PSA) distillation plant (conductivity < 10 mScm 1 ,Cl ¼ 0.7–
0.8 mg L 1 , NO3 ¼ 0.5 mg L 1 , organic carbon < 0.5 mg L 1 ).
The preparation of standard moderately-hard freshwater
(SW) was carried out taking into account the characteristics of
ground water in the province of Almería (southern Spain). SW
was prepared by mixing 96 mg L 1 of NaHCO3, 60mgL 1 of
CaSO4$2H2O, 60 mg L 1 of MgSO4, and 4 mg L 1 of KCl (Standard
Methods, 1998). The simulated effluent wastewater (SE)
consists of SW and Peptone (32 mg L 1 ), Meat extract
(22 mg L 1 ), Urea (6 mg L 1 ), K 2HPO 4 (28 mg L 1 ), CaCl 2$2H 2O
(4 mg L 1 ), MgSO 4$7H 2O(2mgL 1 ) and NaCl (7 mg L 1 ) which
derives to an initial DOC (dissolved organic carbon) of
25 mg L 1 (OECD, 1999). RE was taken downstream of the
Almería MWTP secondary biological treatment and used as
received within the next 2 days. Initial COD and DOC were at
60 mg L 1 and 25 mg L 1 , respectively.
2.3. Solar photo-Fenton pilot plant
Photo-Fenton experiments were performed at the Plataforma
Solar de Almería in a pilot compound parabolic collector (CPC)
solar plant designed for solar photocatalytic applications. This
reactor is composed of two 11dL modules with twelve Pyrex
glass tubes (30 mm O.D.) mounted on a fixed platform tilted 37
(local latitude). The water flows (20 L min 1 ) directly from one
module to the other and finally to a 10dL reservoir. The piping
and valves (3 L) between the reactor and the tank are black
HDPE, which is highly resistant to chemicals, weatherproof
and opaque, preventing any photochemical effect from outside
the collectors. The total illuminated area is 3 m 2 , the total
volume (two modules þ reservoir tank þ piping and valves) is
35 L (VT) and the irradiated volume is 22 L (Vi). Solar ultraviolet
radiation (UV) was measured by a global UV radiometer
(KIPP&ZONEN, model CUV 3) mounted on a platform tilted 37
consumed H 2O 2
0 60 120 180 240 300
t 30W , min
Acetaminophen
Caffeine
Ofloxacine
Antipyrine
Sulfamethoxazole
Carbamazepine
Flumequine
Ketorolac
Atrazine
Isoproturon
Hydroxybenzone
Diclofenac
Ibuprofen
Progesterone
Triclosan
Fig. 2 – Degradation of the 15 ECs (0.1 mg L L1 each) by photo-Fenton (5 mg L L1 Fe, no pH adjustment) and hydrogen peroxide
consumption in SE acidified for carbonate release.
120
90
60
30
0
consumed H 2 O 2 , mg L -1

Table 2 – Final concentration of ECs in two different experiments with SE after similar irradiation times, with
[Fe]0 [ 5mgL L1 : 300 min with [H2O2]0 [ 50 mg L L1 and addition of 50 mg L L1 when the initial dose was almost consumed
and so on; 336 min with [H2O2]0 [ 5mgL L1 and addition of H2O2 every 30–60 min until the end of the experiment. Fenton
degradation (%) in the dark and first order kinetic constants (k) for photo-Fenton are also included.
[H2O2]0 ¼ 50 mg L 1
[H2O2]0 ¼ 5mgL 1
(the same as the CPCs). The temperature inside the reactor was
continuously recorded by a temperature probe (Crioterm
PT-100 3H) inserted in the piping. A plant diagram has been
published elsewhere (Kositzi et al., 2004). With Eq. (1), combination
of the data from several days’ experiments and their
comparison with other photocatalytic experiments is possible,
UV Vi
t30W;n ¼ t30W;n 1 þ Dtn ; Dtn ¼ tn tn 1; t0 ¼ 0ðn ¼ 1Þ (1)
30 VT
where tn is the experimental time for each sample, UV is the
average solar ultraviolet radiation (l < 400 nm) measured
between tn 1 and tn, and t30W is a ‘‘normalized illumination
time’’. In this case, time refers to a constant solar UV power of
30 Wm 2 (typical solar UV power on a perfectly sunny day
around noon).
2.4. Experimental setup
t 30W C/C 0 Fenton
deg. X [%]
The mixture of the 15 compounds (100 mgL 1 each) dissolved in
Methanol at 2.5 g L 1 (mother solution) was added directly
(1.4 mL) into the pilot plant and well homogenized by turbulent
recirculation for 15 min. Methanol was used to dissolve the
contaminants (TOC from methanol was 12 mg L 1 ). The pH in
SW was 7.6, in SE 7.8 and in RE 8.0, which favoured bicarbonate
ions, so at the beginning of the process, when the collectors
were still covered, enough acid to remove CO 3 2 and HCO3 ,
known OH radical scavengers, was added. Recirculation time
for this process was between 15 and 60 min. When the total
inorganic carbon (TIC) concentration was below the desired
level (10 mg L 1 ), 50 mg L 1 of peroxide was added and homogenised
by recirculation for 15 min. Finally, Fe 2þ as FeSO4$7H20
was added (5 mg L 1 ). After 15 min of recirculation, in which
the Fenton reaction started, the collectors were uncovered and
photo-Fenton began. Hydrogen peroxide and iron were
measured in every sample taken. The experiments normally
water research 44 (2010) 545–554 549
k [min 1 ] t 30W C/C 0 Fenton
deg. X [%]
lasted as long as there was H 2O 2 to be consumed (usually 3–4 h),
in the case of experiments which lasted longer than that, the
experiment was covered at the end of the day and the next day,
after recirculation for 15 min, a sample was taken, peroxide
was added (50 mg L 1 ) and the covers again removed.
2.5. Analytical measurements
k [min 1 ]
Acetaminophen 261

550
Table 3 – Final EC concentration in RE with
[Fe]0 [ 5mgL L1 ,[H2O2]0 [ 50 mg L L1 after an irradiation
time of t30W [ 276 min. Fenton degradation (%) in the
dark and first order kinetic constant (k) for photo-Fenton
are also included.
t30W C/C0 Fenton k [min
deg. X [%]
1 ]
Acetaminophen 109

eactor (t ExP ¼ 15 min). Removal of all compounds but atrazine
(18%) was in the range between 30% (caffeine) and 60%
(ibuprofen). Fenton degradation was very quick after adding just
5mgL 1 of Fe 2þ but it did not proceed further until the reactor
was uncovered, as Fe 3þ reduction to Fe 2þ was inefficient without
illumination. All the pollutants but triclosan (87%) and atrazine
(77%) were efficiently degraded (removal over 94%) under photo-
Fenton conditions, at t30W ¼ 90 min. Moreover, some pollutants
(ofloxacin and diclofenac) were degraded to below the LOD as
early as t 30W ¼ 32 min. The total amount of H 2O 2 consumed was
37 mg L 1 and the iron concentration decreased due to precipitation
from 5 mg L 1 to 3.1 mg L 1 , probably due to colloidal
aggregation, while the pH varied from an original 7.6–5.3 (after
adding the FeSO4$7H2O) to a final pH ¼ 3.8 at t30W ¼ 90 min.
Similar results were found by adding H2SO4 (56 mg L 1 )
instead of HCl. H2SO4 has certain advantages (handling,
volatility, etc.) over HCl, and Cl ions may act as radical
scavengers. Experiments did not show any significant difference
between using HCl or H2SO4. Indeed, the total concentration
of chloride and sulphate measured in SW was
43 mg L 1 Cl when HCl was added and 145 mg L 1 SO 4 2 when
H 2SO 4 was added respectively. It is well known that inorganic
species (chloride, sulphate, phosphates, etc) are usually
detrimental to the photo-Fenton reaction rate (Pignatello
et al., 2006) due to complexation of these ions with Fe 2þ or
Fe 3þ , or to the scavenging of hydroxyl radicals and formation
of less reactive inorganic ions (De Laat et al., 2004), but it has
also been observed that they significantly reduce the photo-
Fenton reaction rate only if they occur at concentrations over
500 mg L 1 , usually very far from the concentration found in
natural water (Bacardit et al., 2007; Zapata et al., 2009).
3.2. Photo-Fenton tests using SE
The following experiment dealt with the photo-Fenton treatment
of SE, which was characterised by original TOC and COD
of 37 mg L 1 and 80 mg L 1 , respectively. Two different
approaches to the H2O2 dose were studied: (i) 50 mg L 1 were
added at the beginning and again when the initial dose was
almost consumed and so on; (ii) 5 mg L 1 were added every
30–60 min (also depending on consumption) until the end of
the experiment. In this case, 56 mg L 1 H 2SO 4 acid were added.
A recirculation time of 15 min was necessary for carbonates to
be released (TIC < 5mgL 1 ). As can be seen in Fig. 2, degradation
is much slower with both Fenton (t < 0) and photo-
Fenton (t30W > 0) than in the experiment with SW shown in
Fig. 1. The residual concentration of the contaminants at the
end of the experiments (t30W ¼ 300 min and 336 min, respectively)
can be seen in Table 2. The pH varied during the
experiment with high H 2O 2 concentrations from the original
7.6–4.9 at the end, while the iron concentration varied from
5mgL 1 to 3.2 mg L 1 . During the experiment with low H 2O 2
concentrations the pH varied from 8.1 to 4.9, while iron
concentration went from 4.8 to 3.1 mg L 1 . The EC degradation
behaviour did not change significantly, no matter which
hydrogen peroxide dose approach was used, as can be seen in
Table 2, although the overall amount of peroxide consumed
was much lower (110 mg in 50-mg-L 1 additions compared to
52 mg with the 5-mg-L 1 additions). It should be mentioned,
although not relevant to the purpose of the experiments,
water research 44 (2010) 545–554 551
because the main purpose was to degrade ECs, that TOC
mineralisation was rather low in both cases (around 25%).
3.3. Photo – Fenton treatment of RE
The initial DOC, TIC and COD were 36 mg L 1 , 106 mg L 1 and
60 mg L 1 , respectively. In this case, 406 mg L 1 H2SO4 were
added to reach < 20 mg L 1 TIC and recirculation time in the pilot
plant necessary to remove carbonate species was 30 min.
According to previous tests, photo-Fenton treatment at a TIC over
20 mg L 1 leads to very slow degradation. Interestingly, although
the degradation rate of all ECs was faster than in the previous
experiments in SE, as observed in Table 3, DOC degradation was
similar to SE (around 20%). This can be explained by the presence
of humic acids in RE which produce solvated electrons and
hydroxyl radicals upon irradiation (Fukushima and Tatsumi,
2001; Prosen and Zupančič-Kralj, 2005; Auger et al., 2002).
3.4. The effect of photo-Fenton treatment on
V. fisheri toxicity
In Fig. 3,theresultsofV. fisheri toxicity tests with SE and RE after
photo-Fenton experiments are compared. Three different steps
in toxicity behaviour (marked 1, 2 and 3 in Fig. 3) can be
described. Before photo-Fenton starts, SE toxicity (20–30% inhibition)
may be due to the mixture of pollutants as well as to the
formation of intermediates during Fenton in the dark, particularly
from the fast oxidation of some of the ECs. On the contrary,
simultaneous activation (negative inhibition) is also observable
in the RE experiment. This completely different behaviour may
be explained by the RE – nutrients and salt contents, which in
someway favoured the growth of V. fisheri until the toxic effect of
pollutants and their oxidation intermediates was neutralized.
In both SE and RE, when photo-Fenton starts up, from an of
H2O2 dose ¼ 0mgL 1 up to H2O2 dose ¼ 50 mg L 1 , the degradation
of parent pollutants begins forming more toxic intermediates,
which drastically increase the inhibition rate in SE
(Fig. 3, Step 2). The previously favourable conditions of RE
probably can no longer efficiently balance the increasingly
toxic effect, so inhibition start to gradually increase in RE too.
In the last step (3) inhibition with RE is quite constant, probably
because toxic organic intermediates take longer to
mineralise; on the other hand, toxicity in SE almost totally
disappeared. If we compare these results with residual
concentration of the four parent ECs selected for comparison
(selected for their representativeness of the behaviour of the
15 different compounds) it may be observed that photo-Fenton
treatment was more efficient in the RE experiment than in
the SE. So in terms of toxicity, the higher removal rate in RE
experiment may result in a correspondingly larger amount of
more toxic intermediates than in the SE experiment or the
generation of other intermediates from RE organics. Therefore,
toxicity assessment be should taken into account as well
as EC degradation during AOP wastewater treatment.
4. Conclusions
- The experiments showed that ECs at low concentrations
(mgL 1 range) can be successfully degraded to negligible

552
concentrations with solar photo-Fenton at low iron
concentrations (5 mg L 1 ) and low initial H2O2 (50 mg L 1 )
concentrations without adjusting the pH.
- One limiting factor for the degradation rate is the presence of
CO3 2 and HCO3 - which are very efficient OH radical scavengers
and which have to be removed prior to the photo-
Fenton reaction. This can be done either with HCl or H2SO4,
as the range of the final concentration of Cl or SO4 2 does not
influence the reaction. Therefore, either one may be used as
economic/operating reasons recommend, but not for any
kinetic effect. It should be taken into account that around
400 mg L 1 of H 2SO 4 (0.05 V kg 1 ) were necessary for CO 2
stripping in RE, which means around 0.025 V m 3 . However,
these costs should be evaluated on a case by case basis as it
strongly depends on the particular RE.
- Changing the operating conditions by lowering H2O2
concentration (5 mg L 1 ) and adjusted by properly
controlled dosing could lead to comparable degradation of
the compounds with lower peroxide consumption.
- Degradation is faster in RE due to humic acids producing
solvated electrons and hydroxyl radicals upon irradiation,
and these contribute to the radicals produced by photo-
Fenton.
Toxicity from the formation of oxidation intermediates
should be taken into account in the treatment of effluents
from wastewater treatment plants by AOPs, because they
could increase final toxicity.
Acknowledgements
Funding for this work was provided by the Spanish Ministry of
Science and Innovation under the Consolider-Ingenio 2010
programme (Project CSD2006–00044 TRAGUA; http://www.
consolider-tragua.com) and by the Andalusia Regional Government
(Project no. P06-TEP-02329). Nick Klamerth would to thank
the University of Almería and CIEMAT for his Ph. D. research
grant. Luigi Rizzo wishes to thank the University of Salerno
(Italy) and the Province of Salerno for the research grant which
allowed him to work in the above mentioned project at the
Plataforma Solar de Almería. The authors also wish to thank
Mrs. Deborah Fuldauer for English language correction.
references
Auger, J.P., Richard, C., Trubetskaya, O., Trubetskoj, O., Lévèque, J.,
Andreux, F., 2002. Photoinductive efficiency of soil extracted
humic and fulvic acids. Chemosphere 49, 259–262.
Bacardit, J., Stötzner, J., Chamarro, E., Esplugas, S., 2007. Effects of
salinity on the photo-Fenton process. Industrial and
Engineering Chemistry Research 46, 7615–7619.
Bali, U., Catalkaya, E.C., Sengul, F., 2003. Photochemical
degradation and mineralisation of phenol: a comparative
study. Journal of Environmental Science and Health – Part A
Toxic/Hazardous Substances and Environmental Engineering
38, 2259–2275.
Bremner, D.H., Burgess, A.E., Houllemare, D., Namkung, K., 2006.
Phenol degradation using hydroxyl radicals generated from
zero-valent iron and hydrogen peroxide. Applied Catalysis B:
Environmental 63, 15–19.
water research 44 (2010) 545–554
Cañizares, P., Domínguez, J.A., Rodrigo, M.A., Villaseñor, J.,
Rodríguez, F., 1999. Effect of the current intensity in the
electrochemical oxidation of aqueous phenol wastes at an
activated carbon and steel anode. Industrial and Engineering
Chemistry Research 38, 3779–3785.
Carballa, M., Omil, F., Lema, J.M., Llompart, M., Garcia-Jares, C.,
Rodriguez, I., Gomez, M., Ternes, T., 2004. Behaviour of
pharmaceuticals, cosmetics and hormones in a sewage
treatment plant. Water Research 38, 2918–2926.
Castiglioni, S., Bagnati, R., Fanelli, R., Pomati, F., Calamari, D.,
Zuccato, E., 2006. Removal of pharmaceuticals in sewage
treatment plants in Italy. Environmental Science and
Technology 40, 357–363.
De Laat, J., Truong Le, G., Legube, B., 2004. A comparative study of
the effects of chloride, sulfate and nitrate ions on the rates of
decomposition of H 2O 2 and organic compounds by Fe(II)/H 2O 2
and Fe(III)/H 2O 2. Chemosphere 55, 715–723.
Esplugas, S., Rodríguez, M., Malato, S., Pulgarin, C., Contreras, S.,
Curcó, D., Giménez, J., 2005. Optimizing the solar photo Fenton
process in the treatment of contaminated water.
Determination of intrinsic kinetic constants for scale-up. Solar
Energy 79, 360–368.
Esplugas, S., Bila, D., Krause, L., Dezotti, M., 2007. Ozonation and
advanced oxidation technologies to remove endocrine
disrupting chemicals (EDCs) and pharmaceuticals and
personal care products (PPCPs) in water effluents. A review.
Journal of Hazardous Materials 149, 631–642.
Farré, M.J., Franch, M.I., Malato, S., Ayllon, J., Peral, J.,
Doménech, X., 2005. Degradation of some biorecalcitrant
pesticides by homogeneous and heterogeneous photocatalytic
ozonation. Chemosphere 58, 1127–1133.
Fono, L.J., Kolodziej, E.P., Sedlak, D.L., 2006. Attenuation of
wastewater-derived contaminants in an effluent
dominated river. Environmental Science and Technology 40,
7257–7262.
Fukushima, M., Tatsumi, K., 2001. Degradation pathways of
pentachlorophenol by photo-Fenton systems in the presence
of iron(III), humic acids and hydrogen peroxide.
Environmental Science and Technology 35, 1771–1778.
Gebhardt, W., Schröder, H., 2007. Liquid chromatography-
(tandem) mass spectrometry for the follow-up of the
elimination of persistent pharmaceuticals during wastewater
treatment applying biological wastewater treatment and
advanced oxidation. Journal of Chromatography A 1160, 34–43.
Glaze, W., Kang, W., Chapin, H., 1987. Chemistry of water
treatment processes involving ozone, hydrogen peroxide
and ultraviolet radiation. Ozone Science & Engineering 9,
335–352.
Göbel, A., McArdell, C., Joss, A., Siegrist, H., Giger, W., 2007. Fate of
sulfonamides, macrolides and trimethoprim in different
wastewater treatment technologies. Science of the Total
Environment 372, 361–371.
Gogate, P.R., Pandit, A.B., 2004. A review of imperative
technologies for wastewater treatment. I: oxidation
technologies at ambient conditions. Advances in
Environmental Research 8, 501–551.
Gültekin, I., Ince, N.H., 2007. Synthetic endocrine disruptors in
the environment and water remediation by advanced
oxidation processes. Journal of Environmental Management
85, 816–832.
Hislop, K.A., Bolton, J.R., 1999. The photochemical generation of
OH radicals in the UV–Vis/ferrioxalate/H2O2 system.
Environmental Science and Technology 33, 3119–3126.
Huber, M., Coronen, S., Ternes, T., von Gunten, U., 2005.
Oxidation of pharmaceuticals during water treatment with
chlorine dioxide. Water Research 39, 3607–3617.
Jackson, J., Sutton, R., 2008. Sources of endocrine disrupting
chemicals in urban wastewater, Oakland, CA. Science of the
Total Environment 405, 153–160.

Kassinos, D., Varanva, N., Michael, C., Piera, P., 2009.
Homogenous oxidation of aqueous solutions of atrazine and
fenitrothion through dark and photo-Fenton reactions.
Chemosphere 74, 866–872.
Kasprzyk-Hordern, B., Dinsdale, R.M., Guwy, A.J., 2009. The
removal of pharmaceuticals, personal care products,
endocrine disruptors and illicit drugs during wastewater
treatment and its impact on the quality of receiving waters.
Water Research 43, 363–380.
Kim, S.D., Cho, J., Kim, I.S., Vanderford, B.J., Snyder, S.A., 2007.
Occurrence and removal of pharmaceuticals and endocrine
disruptors in South Korean surface, drinking, and waste
waters. Water Research 41, 1013–1021.
Klamerth, N., Miranda, N., Malato, S., Agüera, A., Fernández-
Alba, A.R., 2009. Degradation of emerging contaminants at low
concentrations in MWTPs effluents with mild solar photo-
Fenton and TiO 2. Catalysis Today 144, 124–130.
Kositzi, M., Poulios, I., Malato, S., Cáceres, J., Campos, A., 2004.
Solar photocatalytic treatment of synthetic municipal
wastewater. Water Research 38, 1147–1154.
Lapertot, M., Ebrahimi, S., Dazio, S., Rubinelli, A., Pulgarin, C.,
2007. Photo-Fenton and biological integrated process for
degradation of a mixture of pesticides. Journal of
Photochemistry and Photobiology A: Chemistry 186, 34–40.
Laville, N., Ait-Aissa, S., Gomez, E., Casellas, C., Porcher, J.M.,
2004. Effects of human pharmaceuticals on cytotoxicity, EROD
activity and ROS production in fish hepatocytes. Toxicology
196, 41–55.
Lesko, T., Colussi, A., Hoffmann, M., 2006. Sonochemical
decomposition of phenol: evidence for a synergistic effect of
ozone and ultrasound for the elimination of total organic
carbon from water. Environmental Science and Technology
40, 6818–6823.
Malato, S., Blanco, J., Alarcón, D.C., Maldonado, M.I., Fernández-
Ibáñez, P., Gernjak, W., 2007. Photocatalytic decontamination
and disinfection of water with solar collectors. Catalysis
Today 122, 137–149.
Mantzavinos, D., Vassilakis, C., Pantidou, A., Psillakis, E.,
Kalogerakis, N., 2004. Sonolysis of natural phenolic
compounds in aqueous solutions: degradation pathways and
biodegradability. Water Research 38, 3110–3118.
Martinez, F., Calleja, G., Melero, J.A., Molina, R., 2005.
Heterogeneous photo-Fenton degradation of phenolic
aqueous solutions over iron containing SBA-15 catalyst.
Applied Catalysis B: Environmental 60, 181–190.
Martínez Bueno, M.J., Agüera, A., Gómez, M.J., Hernando, M.D.,
García-Reyes, J.F., Fernández-Alba, A.R., 2007. Application of
liquid chromatography/quadrupole-linear ion trap mass
spectrometry and time-of-flight mass spectrometry to the
determination of pharmaceuticals and related
contaminants in wastewater. Analytical Chemistry 79,
9372–9384.
Mitch, W.A., Sharp, J.O., Trussel, R.R., Valentine, R.L., Alvarez-
Cohen, L., Seldak, D.L., 2003. N-Nitrosodimethylamine (NDMA)
as a drinking water contaminant: A review. Environmental
Engineering and Science 20, 389–404.
Moncayo-Lasso, A., Pulgarin, C., Benítez, N., 2008. Degradation of
DBPs’ precursors in river water before and after slow sand
filtration by photo-Fenton process at pH 5 in a solar CPC
reactor. Water Research 42, 4125–4132.
Nakada, N., Kiri, K., Shinohara, H., Harada, A., Kuroda, K.,
Takizawa, S., Takada, H., 2008. Evaluation of pharmaceuticals
and personal care products as water-soluble molecular
markers of sewage. Environmental Science and Technology
42, 6347–6353.
Ning, B., Graham, N., Zhang, Y., Nakonechny, M., El-Din, M.G.,
2007. Degradation of endocrine disrupting chemicals by
ozone/AOPs. Ozone Science & Engineering 29, 153–176.
water research 44 (2010) 545–554 553
Neyens, E., Baeyens, J., 2003. A review of classic fenton’s
peroxidation as an advanced oxidation technique. Journal of
Hazardous Materials 98, 33–50.
Nogueira, R.F.P., Oliveira, M.C., Paterlini, W.C., 2005. Simple and
fast spectrophotometric determination of H 2O 2 in photo-
Fenton reactions using metavanadate. Talanta 66, 86–91.
OECD Guidelines for testing of Chemicals, Simulation Test-
Aerobic Sewage Treatment 303A, 1999.
Papadaki, M., Emery, R.J., Abu-Hassan, M.A., Diaz-Bustos, A.,
Metcalfe, I.S., Mantzavinos, D., 2004. Sonocatalytic oxidation
processes for the removal of contaminants containing
aromatic rings from aqueous effluents. Separation and
Purification Technology 34, 35–42.
Pelegrini, R.T., Freire, R.S., Duran, N., Bertazzoli, R., 2001.
Photoassisted electrochemical degradation of organic
pollutants on a DSA type oxide electrode: process test for
a phenol synthetic solution and its application for the E1
bleach kraft mill effluent. Environmental Science and
Technology 35, 2849–2853.
Petrovic, M., Gonzalez, S., Barceló, D., 2003. Analysis and removal
of emerging contaminants in wastewater and drinking water.
Trends in Analytical Chemistry 22, 685.
Pignatello, E., Oliveros, A., McKay, 2006. Advanced oxidation
processes for organic contaminant destruction based on the
Fenton reaction and related chemistry. Critical Reviews in
Environmental Science and Technology 36, 1–84.
Prosen, H., Zupančič-Kralj, L., 2005. Evaluation of photolysis and
hydrolysis of atrazine and its first degradation product in the
presence of humic acids. Environmental Pollution 133, 517–529.
Radjenović, J., Petrović, M., Barceloć, D., Petrović, M., 2007.
Advanced mass spectrometric methods applied to the study of
fate and removal of pharmaceuticals in wastewater
treatment. Trends in Analytical Chemistry 26, 1132–1144.
Richardson, S., 2007. Water analysis: emerging contaminants and
current issues. Reviews in Analytical Chemistry 79, 4295–4324.
Rosenfeldt, E.J., Linden, K.G., 2004. Degradation of endocrine
disrupting chemicals bisphenol A, ethinyl estradiol and estradiol
during UV photolysis and advanced oxidation processes.
Environmental Science and Technology 38, 5476–5483.
Rosenfeldt, E.J., Chen, P.J., Kullman, S., Linden, K.G., 2007.
Destruction of estrogenic activity in water using UV advanced
oxidation. Science of the Total Environment 377, 105–113.
Saritha, P., Aparna, C., Himabindu, V., Anjaneyulu, Y., 2007.
Comparison of various advanced oxidation processes for the
degradation of 4-chloro-2-nitrophenol. Journal of Hazardous
Materials 149, 609–614.
United Book Press Inc.. In: Clesceri, L.S., Greenberg, A.E., Eaton, A.
D. (Eds.), Standard Methods for the Examination of Water and
Wastewater, 20th ed. American Public Health Association,
American Water Works Association, Water Environment
Federation, Baltimore, Maryland.
Ternes, T., 1998. Occurrence of drugs in German sewage
treatment plants and rivers. Water Research 32, 3245–3260.
Ternes, T., Struber, J., Herrmann, N., McDowell, D., Ried, A.,
Kampmann, M., Teiser, B., 2003. Ozonation: a tool for removal
of pharmaceuticals contrast media and musk fragrances from
wastewater. Water Research 37, 1976–1982.
Ternes, T., Bonerz, M., Herrmann, N., Teiser, B., Andersen, R.,
2007. Irrigation of treated wastewaters in Braunschweig,
Germany: an option to remove pharmaceuticals and musk
fragrances. Chemosphere 66, 894–904.
Teske, S.S., Arnold, R.G., 2008. Removal of natural and xenoestrogens
during conventional wastewater treatment. Reviews
in Environmental Science and Biotechnology 7, 107–124.
Zapata, A., Oller, I., Bizani, E., Sánchez-Pérez, J.A., Maldonado, I.,
Malato, S., 2009. Evaluation of operational parameters
involved in solar photo-Fenton degradation of a commercial
pesticide mixture. Catalysis Today 144, 94–99.

554
Zepp, R.G., Faust, B.C., Hoigné, J., 1992. Hydroxyl radical
formation in aqueous reactions pH(3–8) of iron(II) with
hydrogen peroxide: the photo-fenton reaction. Environmental
Science and Technology 26, 313–319.
Zhao, J.-L., Ying, G.-G., Wang, L., Yang, J.-F., Yang, X.-B., Yang, L.-H.,
Li, X., 2009. Determination of phenolic endocrine disrupting
chemicals and acidic pharmaceuticals in surface water of the
water research 44 (2010) 545–554
Pearl River in South China by GC-MS. Science of the Total
Environment 407, 962–974.
Zhou, M., Dai, Q., Lei, L., Ma, C., Wang, D., 2005. Long life
modified lead dioxide anode for organic wastewater
treatment: electrochemical characteristics and degradation
mechanism. Environmental Science and Technology 39,
363–370.

Oxidative transformation of micropollutants during municipal
wastewater treatment: Comparison of kinetic aspects of
selective (chlorine, chlorine dioxide, ferrate VI , and ozone)
and non-selective oxidants (hydroxyl radical)
Yunho Lee a , Urs von Gunten a,b, *
a
Eawag, Swiss Federal Institute of Aquatic Science and Technology, Ueberlandstrasse 133, P.O. Box 611, 8600 Duebendorf, Switzerland
b
Institute of Biogeochemistry and Pollutant Dynamics, ETH Zurich, 8092 Zurich, Switzerland
article info
Article history:
Received 1 October 2009
Received in revised form
13 November 2009
Accepted 23 November 2009
Available online 27 November 2009
Keywords:
Oxidation processes
Ozone
Chlorine
Chlorine dioxide
Ferrate VI
Hydroxyl radical
water research 44 (2010) 555–566
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
abstract
Chemical oxidation processes have been widely applied to water treatment and may serve
as a tool to minimize the release of micropollutants (e.g. pharmaceuticals and endocrine
disruptors) from municipal wastewater effluents into the aquatic environment. The
potential of several oxidants for the transformation of selected micropollutants such as
atenolol, carbamazepine, 17a-ethinylestradiol (EE2), ibuprofen, and sulfamethoxazole was
assessed and compared. The oxidants include chlorine, chlorine dioxide, ferrate VI , and
ozone as selective oxidants versus hydroxyl radicals as non-selective oxidant. Secondorder
rate constants (k) for the reaction of each oxidant show that the selective oxidants
react only with some electron-rich organic moieties (ERMs), such as phenols, anilines,
olefins, and deprotonated-amines. In contrast, hydroxyl radicals show a nearly diffusioncontrolled
reactivity with almost all organic moieties (k 10 9 M 1 s 1 ). Due to a competition
for oxidants between a target micropollutant and wastewater matrix (i.e. effluent
organic matter, EfOM), a higher reaction rate with a target micropollutant does not
necessarily translate into more efficient transformation. For example, transformation
efficiencies of EE2, a phenolic micropollutant, in a selected wastewater effluent at pH 8
varied only within a factor of 7 among the selective oxidants, even though the corresponding
k for the reaction of each selective oxidant with EE2 varied over four orders of
magnitude. In addition, for the selective oxidants, the competition disappears rapidly after
the ERMs present in EfOM are consumed. In contrast, for hydroxyl radicals, the competition
remains practically the same during the entire oxidation. Therefore, for a given oxidant
dose, the selective oxidants were more efficient than hydroxyl radicals for transforming
ERMs-containing micropollutants, while hydroxyl radicals are capable of transforming
micropollutants even without ERMs. Besides EfOM, ammonia, nitrite, and bromide were
found to affect the micropollutant transformation efficiency during chlorine or ozone
treatment.
ª 2009 Elsevier Ltd. All rights reserved.
* Corresponding author at: Eawag, Swiss Federal Institute of Aquatic Science and Technology, Ueberlandstrasse 133, P.O. Box 611,
8600 Duebendorf, Switzerland. Tel.: þ41 44 823 5270; fax: þ41 44 823 5028.
E-mail address: vongunten@eawag.ch (U. von Gunten).
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.watres.2009.11.045

556
1. Introduction
Effluents of municipal wastewater treatment plants (WWTPs)
have been identified as a major source of micropollutants,
such as hormones, pharmaceuticals, and personal care
products (Ternes and Joss, 2006). To minimize a discharge of
micropollutants from WWTPs to the receiving waters and
thus to prevent adverse ecological effects, there have been
discussions on upgrading WWTPs with additional treatment
steps such as activated carbon and ozonation (Nowotny et al.,
2007; Hollender et al., 2009). This is also desirable for the
quality of drinking water if the wastewater-impacted
receiving waters are used as a drinking water source via
indirect potable reuse.
The focus of the current study is on the chemical oxidation.
Chemical oxidants, such as chlorine, chlorine dioxide, ozone,
and hydroxyl radicals, have been widely applied for disinfection
of drinking waters and wastewaters and often for the
transformation/elimination of undesired micropollutants
from drinking water (Letterman, 1999). Chlorine is worldwide
one of the most commonly used disinfectant for drinking
water and wastewater treatment (Letterman, 1999). Chlorine
dioxide has been used as an alternative disinfectant for
chlorine, often to minimize the formation of chlorine-based
disinfection by-products (e.g. trihalomethane) (Gates, 1998;
Letterman, 1999). The application of ozone in drinking water
treatment is also widespread and the main reasons for its use
are disinfection and oxidation (e.g. micropollutants transformation)
or a combination of both (Hoigné, 1998; von
Gunten, 2003a). Even though chlorine, chlorine dioxide, and
ozone have been widely used for disinfection (Letterman,
1999), their potentials for transforming micropollutants in
wastewaters have received attention only in recent years
(Ternes and Joss, 2006). In addition, ferrate VI is an emerging
water treatment chemical due to its dual functions as an
oxidant and a subsequent coagulant/precipitant as ferric
hydroxide (Lee et al., 2009). Finally, the hydroxyl radical is
a primary oxidant in advanced oxidation processes (AOPs)
such as UV/H2O2 and O3/H2O2 and their application are
considered for transforming various micropollutants in
drinking waters as well as wastewaters for water reuse
(Letterman, 1999; Asano et al., 2007). The hydroxyl radical is
quite different from other oxidants with respect to its high
reactivity and thus low selectivity (Buxton et al., 1988). In this
paper, chlorine, chlorine dioxide, ferrate VI and ozone will be
referred as selective oxidants as opposed to non-selective
hydroxyl radicals.
The oxidants mentioned above can achieve a transformation
of various micropollutants during wastewater
treatment. The transformation efficiency mainly depends on:
(i) the reactivity of the oxidant to target micropollutants; and
(ii) towards matrix components present in water (e.g. dissolved
organic matter, DOM) which determines the stability of
the oxidant. In general, the reactions of the oxidants follow
second-order kinetics, i.e. the reaction rate is proportional to
the concentrations of both reactants. The corresponding
second-order reaction rate constants (k) range from 1to
10 10 M 1 s 1 in aqueous solution (Buxton et al., 1988; Deborde
and von Gunten, 2008; Lee et al., 2009; Neta et al., 1988; von
water research 44 (2010) 555–566
Gunten, 2003a). Currently, k values are available for a wide
range of organic compounds including many micropollutants
(Fig. 1 and Table 1 as examples). For ozone reactions more
than 500 k values have been measured (Neta et al., 1988 von
Gunten, 2003a), and a few thousand k values are available for
hydroxyl radical reactions (Buxton et al., 1988). There are also
a few hundred k values available for chlorine, chlorine
dioxide, and ferrate VI (Deborde and von Gunten, 2008; Lee
et al., 2009; Neta et al., 1988). These second order rate
constants show that the selective oxidants react preferentially
with electron-rich organic moieties (ERMs), such as activated
aromatic compounds (i.e. phenol, aniline, and polycyclic
aromatics), organosulfur compounds, and deprotonated
amines. Additionally, ozone and ferrate VI react with olefins.
Contrary to the selective oxidants, hydroxyl radicals react
with almost all organic moieties (e.g. aliphatic CeH bond) with
nearly diffusion-controlled rates (i.e. k 10 9 M 1 s 1 ).
Therefore, micropollutants containing the ERMs
mentioned above can be potentially transformed by the
selective oxidants. On the contrary, hydroxyl radicals can be
effective for transforming any type of micropollutant.
However, a high reactivity of the oxidants to the ERMs is also
responsible for a rapid oxidant consumption by the ERMs
included in the matrix components, such as DOM. In addition
to DOM, some inorganic species such as ammonia, nitrite,
sulfide, iron(II), and manganese(II) can be potential
consumers of the selective oxidants. In the case of hydroxyl
radicals, their rapid consumption by almost all types of
matrix components can lead to a significantly lower transformation
efficiency of micropollutants than the selective
oxidants. Therefore, the kinetic information (i.e. k values) of
the oxidants with the ERMs and some other moieties is
crucial to assess the transformation efficiency of structurallydiverse
micropollutants by a given oxidant in a real wastewater
matrix.
The aim of this study was to compare each of the selective
oxidants and non-selective hydroxyl radicals with respect to
their efficiency for transforming micropollutants during the
treatment of the wastewater effluents. This comparison is
important to provide criteria for choosing an appropriate
oxidation process for a specific class of micropollutants based
on the chemical structure (i.e. presence of ERM). For this
purpose, the following issues were addressed:
The k of each oxidant with selected organic model
compounds and micropollutants (containing various ERMs)
available in literature were summarized and compared.
The oxidant consumption kinetics in a wastewater effluent
were determined.
The oxidative transformations of selected micropollutants
containing various ERMs were conducted in a wastewater
effluent for varying oxidant doses (0–150 mM). The transformation
efficiency was then compared for the oxidants
based on the required oxidant dose to achieve a certain
degree of a micropollutant transformation.
The effect of ammonia, nitrite, and bromide was investigated
in respect to the transformation efficiency of selected
micropollutants during the treatment of a wastewater
effluent by the selected oxidants and hydroxyl radicals.

-1 -1
s
a b c
k,
M
10
10
10 8
10 6
10 4
10 2
10 0
HO.
d e f
-1 -1
s
k,
M
10 10
10 8
10 6
10 4
10 2
10 0
HO.
HFeO 4 –
2. Materials and methods
2.1. Reagents and wastewater
Atenolol (ATL), carbamazepine (CBZ), 17a-ethinylestradiol
(EE2), ibuprofen (IBP), and sulfamethoxazole (SMX) were
obtained from Sigma–Aldrich with a purity higher than 98%.
Stock solutions of these pharmaceuticals were prepared with
Milli-Q purified water (Barenstad). Other chemicals and
solvents were purchased from various commercial suppliers
and used as received.
A 10-L grab sample of secondary wastewater effluent
(RDWW) was obtained from the Regensdorf wastewater
treatment plant near Zurich, Switzerland. The sample was
transported to the laboratory within several hours of sampling
and vacuum-filtered with a 0.45 mm cellulose nitrate
membrane prior to storage at 4 C. RDWW has a DOC of 5.0 mg
CL 1 , 1.2 mM ammonia (17 mg NL 1 ), 6 mM nitrite (84 mg NL 1 ),
5.3 mM carbonate alkalinity, 0.35 mM (28 mg L 1 )ofBr and
apHofw8. The concentrations of selected micropollutants
(ATL, CBZ, EE2, IBP, and SMX) originally present in RDWW
O 3
phenol aniline butenol (olefin)
ClO 2
HOCl
glycine
o (1 amine)
HOCl
O 3
HFeO 4 –
water research 44 (2010) 555–566 557
HO. HO.
HO.
HOCl
HFeO 4 –
HFeO 4 –
O 3
ClO 2
HOCl
dimethylamine
o (2 amine)
5 6 7 8 9 10 5 6 7 8 9 10 5 6 7 8 9 10
pH pH pH
O 3
ClO 2
were negligible relative to the concentrations spiked for
oxidation experiments, which were 0.2–1 mM for each
substrate.
2.2. Oxidants
HO.
HFeO 4 –
Stock solutions of chlorine (5–20 mM) were prepared by
diluting a commercial solution of sodium hypochlorite (10%
active chlorine, Riedel-deHaen, Germany). Chlorine dioxide
(w5 mM) was produced by mixing potassium peroxodisulfate
(K 2S 2O 8) with sodium chlorite (NaClO 2) according to a method
described by Gates (1998). Ozone was produced with a Fischer
502 ozone generator and its stock solutions (w1.5 mM) were
produced by sparging ozone-containing oxygen through Milli-
Q water that was cooled in an ice bath (Bader and Hoigné,
1981). Potassium ferrate VI of high purity (w90%) was prepared
by the method of Thompson et al. (1951). Stock solutions of
ferrate VI (2 mM) were freshly prepared by dissolving solid
samples of potassium ferrate VI (K2FeO4) in Milli-Q water and
subsequent filtration (pH z 9.2). Hydroxyl radicals were in
situ generated by photolysis of H2O2 (0.2–5 mM). Irradiations
O 3
trimethylamine
o (3 amine)
Fig. 1 – pH dependent second-order rate constants (k) for the reaction of the oxidants, chlorine (HOCl), chlorine dioxide (ClO2),
ferrate VI (HFeO4 L ), hydroxyl radicals (HO ), and ozone (O3) with (a) phenol, (b) aniline, (c) butenol, (d) glycine, (e)
dimethylamine, and (f) trimethylamine. k values for chlorine: Deborde and von Gunten, (2008); for chlorine dioxide and
ozone: Neta et al., (1988); for ferrate VI : Lee et al. (2005a, 2008, 2009); and for hydroxyl radicals: Buxton et al., (1988). Speciesspecific
second-order rate constants and the pKa values of the compounds used to generate the data are provided in
Supplementary Information, SI-excel.
O 3
ClO 2
HOCl

558
Table 1 – Second-order rate constants for the reaction of oxidants with selected pharmaceuticals and inorganic species
(ammonia, nitrite, and bromide). a
17a-Ethinylestradiol (EE2),
synthetic steroid estrogen
Sulfamethoxazole (SMX),
sulfonamide antibiotic
Carbamazepine (CBZ),
antiepileptic drug
Atenolol (ATL), b-blocker
Ibuprofen (IBP), antiphlogistic
water research 44 (2010) 555–566
Compounds pK a Oxidant k (species
of compound),
M 1 s 1
k at pH 8,
M 1 s 1
10.4 HO 9.8 10 9
9.8 10 9
O3 1.8 10 5 (neutral),
3.7 10 9 1.5 10
(anion)
7
ClO2 4.6 10 8 (anion) 1.8 10 6
HFeO4 9.4 10 2 (neutral),
5.4 10 5 4.2 10
(anion)
2
HOCl 3.5 10 5 (anion) 3.4 10 2
Ref.
Huber et al. (2003)
Deborde et al. (2005)
Huber et al. (2005)
Lee et al. (2005a)
Deborde et al. (2004)
5.7 HO 5.5 10 9
5.5 10 9
Huber et al. (2003)
O3 4.7 10 4 (neutral),
5.7 10 5 5.7 10
(anion)
5
Dodd et al. (2006)
ClO2 7.9 10 3 (anion) 7.9 10 3
Huber et al. (2005)
HFeO4 9.2 10 3b
95 Lee et al. (2009)
HOCl 1.1 10 3 (neutral),
2.4 10 3 5.7 10
(anion)
2
Dodd and Huang
(2004)
– HO 8.8 10 9
O3 3.0 10 5
ClO 2

Table 1 (continued)
Compounds pKa Oxidant k (species
of compound),
M 1 s 1
were performed in quartz tubes using a merry-go-round
photoreactor (Hans Mangels GmbH, Bornheim-Roisdor,
Germany), equipped with a low pressure mercury lamp (Heraues
Noblelight model TNN 15/32, nominal power 15 W)
emitting essentially monochromatic light at l ¼ 254 nm. In the
case of SMX, a medium pressure mercury lamp (Heraues
Noblelight model TQ 718, nominal power 500 W) with a 0.25 M
sodium nitrate filter solution producing light at l > 320 nm
was used to minimize the direct photo-transformation of SMX
(less than 10%). Fluence and fluence rate were determined by
chemical actinometry at low optical density using 5 mM
aqueous atrazine as an actinometer described by Meunier
et al. (2006). Stock solutions of each oxidant were standardized
spectrophotometrically based on their molar absorption
coefficient: 3 ¼ 350 M 1 cm 1 at 292 nm for OCl (Johnson and
Margerum, 1991), 3 ¼ 1230 M 1 cm 1 at 359 nm for ClO2 (Gates,
1998), 3 ¼ 3000 M 1 cm 1 at 260 nm for O3 (Liu et al., 2001),
3 ¼ 1150 M 1 cm 1 at 510 nm for FeO4 2
(Lee et al., 2005b), and
3 ¼ 40 M 1 cm 1 at 240 nm for H2O2 (Bader et al., 1988).
2.3. Oxidant consumption kinetics and transformation
of selected micropollutants in wastewater
Oxidant consumption experiments for the selective oxidants
were performed in a secondary wastewater effluent (RDWW)
at pH 8 (20 mM borate buffer) in 50 mL reaction volume. Forty
to 45 mM of each oxidant was applied to RDWW and 0.2–1 mL
of the reaction solution was sampled for the measurements of
residual oxidants over a 1 h reaction time.
Transformation of the selected micropollutants was tested
in RDWW spiked with each micropollutant and treated by
each oxidant at various oxidant doses (0–150 mM) at 24 1 C
and pH 8. The micropollutants, ATL (amine), CBZ (olefin), EE2
(phenol), and SMX (aniline), were selected as representative
substances with different ERMs that belong to different usage
classes and have an environmental relevance (Ternes and
Joss, 2006). IBP was chosen as a model micropollutant without
the ERMs. Each micropollutant was spiked separately at
concentrations of 0.2–1 mM into 20 mL of the wastewater
buffered at pH 8. For the selective oxidants, stock solutions of
water research 44 (2010) 555–566 559
k at pH 8,
M 1 s 1
Br bromide – HO w1.1 10 9
w1.1 10 9
von Gunten (2003b)
O3 2.6 10 2
2.6 10 2
Liu et al. (2001)
ClO2

560
irradiation time. For details of the kinetic method, see
Supplementary Information, SI-word.
2.5. Analytical methods
Oxidant concentrations in wastewater effluents were
measured by several colorimetric methods. Chlorine and
chlorine dioxide were determined by the ABTS method
developed by Pinkernell et al. (2000); ozone by the indigo
method (Bader and Hoigné, 1981); ferrate VI by the ABTS
method described by Lee et al. (2005b); H2O2 by the DPD/
peroxidase method (Bader et al., 1988). Micropollutant (or
pCBA) concentrations at sub to low mM range were determined
with an Agilent 1100 series HPLC system equipped with
a Nucleosil 100-5 C18 column (125 4 or 250 4 mm) and a UV
(diode array) and/or fluorescence detector. Separation was
achieved by different gradient programs with the aqueous
phase consisting either of 10 mM phosphoric acid or 10 mM
citric acid (pH 5) and the solvent phase either of acetonitrile or
methanol. The injection volume was 100 mL and the flow rate
was set to 1 mL min 1 . The repeatability was determined to be
4% while limit of quantifications (LOQ) were inferred from the
lowest standard concentrations used. The limits of quantification
were in all cases sufficient to determine a transformation
of micropollutants of more than 96% ( 1.4 logtransformation).
3. Results and discussion
3.1. Rate constants of oxidants with selected organic
compounds
Fig. 1 summarizes the pH-dependent k for the reaction of the five
oxidants (chlorine, chlorine dioxide, ferrate VI , ozone, and
hydroxyl radical) with the selected ERM-containing model
compounds, such as phenol and aniline (activated aromatic
compounds), butenol (olefin), glycine (primary (1 ) amine),
dimethylamine (secondary (2 ) amine), and trimethylamine
(tertiary (3 ) amine). If the k value is lower than 1 M 1 s 1 ,itisnot
shown in Fig. 1. The species-specific k values which were used to
generate Fig. 1 are provided in Supplementary Information, SIexcel.
For more details on the kinetics of each oxidant, refer the
corresponding references provided in SI-excel.
The k values for phenol and aniline, as two aromatic
compounds, decrease in the order of hydroxyl radical -
> ozone > chlorine dioxide > chlorine z ferrate VI (Figs. 1a,b).
The k values for the other compounds show a slightly different
reactivity order. Chlorine and chlorine dioxide are not reactive
to butenol, therefore the reactivity trend for butenol is
hydroxyl radical > ozone > ferrate VI (Fig. 1c). For three amine
compounds, chlorine shows a particular high reactivity
whereas chlorine dioxide has a low reactivity. For glycine as 1
amine, the k values decrease in the order of hydroxyl radical
> chlorine > ozone > ferrate VI > chlorine dioxide (Fig. 1d).
For dimethylamine as 2 amine, the reactivity order is
hydroxyl radical > ozone z chlorine > ferrate VI > chlorine
dioxide (Fig. 1e). Finally, for trimethylamine as 3 amine, the
reactivity order is hydroxyl radical > ozone > chlorine dioxide
chlorine > ferrate VI (Fig. 1f).
water research 44 (2010) 555–566
In the pH range 5–10, hydroxyl radicals react with phenol,
aniline, and butenol with nearly diffusion-controlled rates.
However, the k values for the reaction with three amine
compounds are one or two orders of magnitude lower (10 7 –
10 8 M 1 s 1 at pH 6–8). This is due to the relative low reactivity of
hydroxyl radicals with the protonated-amine moieties. Nevertheless,
the k of hydroxyl radicals with the amine compounds
with longer hydrocarbon-chains can be higher than 10 9 M 1 s 1 .
This is due to the high reactivity of hydroxyl radical with CeH
bonds (k [ 10 8 M 1 s 1 ) (Buxton et al., 1988). In addition,
hydroxyl radicals react with benzenes (i.e. non-activated
aromatics) with k > 10 9 M 1 s 1 (Buxton et al., 1988). Ozone
reacts fast with all ERMs shown in Fig. 1:thek values decrease in
the order of aniline phenol > butenol dimethylamine z
trimethylamine > glycine in the pH range 6–8. Chlorine dioxide
shows a high reactivity only with phenol, aniline, and trimethylamine.
Chlorine is reactive to all ERMs except butenol. The k
values decrease in the order of glycine > dimethylamine
aniline > phenol z trimethylamine in the pH
range 6–8. Finally, ferrate VI reacts with all ERMs except trimethylamine.
The k values decrease in the order of aniline > phenol
z glycine > butenol dimethylamine in the pH range 6–8.
For the selective oxidants, ozone shows the widest range of k
values (i.e. w7 orders of magnitude) while ferrate VI shows the
narrowest range of k values (i.e. w4 orders of magnitude) toward
the selected ERMs in the pH range 6–8.
3.2. Consumption kinetics of the selective oxidants in
a wastewater effluent
Fig. 2 shows the consumption kinetics of the selective
oxidants (ozone, ferrate VI , chlorine, and chlorine dioxide) in
a secondary wastewater effluent from Regensdorf,
Switzerland (RDWW) at pH 8 for an oxidant dose of 40–45 mM.
The selected RDWW has a low concentration of inorganic
nitrogen species ([ammonia] ¼ 1.2 mM and [nitrite] ¼ 6 mM),
therefore the consumption of each oxidant is mainly caused
by EfOM ([DOC] ¼ 5mg C L 1 ). Except ferrate VI , all oxidants
showed fast consumptions within 2 min after the oxidant
addition. The stability of the oxidants follows the order of
ferrate VI z chlorine dioxide chlorine [ ozone. Ozone was
depleted in less than 2 min (Fig. 2a, inset), whereas the other
oxidants were slowly consumed within 60 min. The
observed low stability of ozone is consistent with its high
reactivity to all ERMs shown in Fig. 1. Chlorine and chlorine
dioxide showed a similar decay pattern: a rapid decrease
within 2 min after the oxidant addition (termed ‘initial
phase’), followed by a slow decrease over 60 min of reaction
time (termed ‘second phase’). Based on the k values shown in
Fig. 1, the chlorine consumption in the initial phase is
expected to be mainly caused by aniline- and 1 &2 aminemoieties
while in the second phase phenolic- and 3 aminemoieties
become important. In the case of chlorine dioxide,
the consumption in the initial phase might be mainly caused
by phenolic- and aniline-moieties while in the second phase
by 2 &3 amine-moieties. Contrary to chlorine and chlorine
dioxide, ferrate VI decreased smoothly along the reaction time,
which is consistent with its narrow distribution of k values
with the ERMs shown in Fig. 1. The achieved oxidant exposures
for the conditions shown in Fig. 2 were 4.5 10 2 M 1

a
C oncentration,
µM
c
C oncentration,
µM
50
40
30
20
10
0
50
40
30
20
10
0
0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
Time, min
0 10 20 30 40 50 60
0 10 20 30 40 50 60
Time, min
s 1 for ferrate VI and chlorine dioxide, 3.6 10 2 M 1 s 1 for
chlorine, and 6.0 10 4 M 1 s 1 for ozone.
3.3. Transformation of selected micropollutants in
a wastewater effluent: comparison between selective
oxidants and non-selective hydroxyl radicals
C oncentration,
µM
A comparison was made for the transformation efficiency of
selected micropollutants in a secondary wastewater effluent
(RDWW) by the selective oxidants (chlorine, chlorine dioxide,
ferrate VI , and ozone) and the non-selective hydroxyl radicals.
The selected micropollutants include compounds with ERMs,
such as 17a-ethinylestradiol (EE2, phenol), sulfamethoxazole
(SMX, aniline), carbamazepine (CBZ, olefin), and atenolol (ATL,
2 amine), and ibuprofen (IBP, no ERM). For the selective
oxidants, the reaction time of 1 h was given to simulate realistic
treatment conditions. Ozone was completely consumed
after 1 h whereas for chlorine, chlorine dioxide, and ferrate VI ,
more than 10% of the oxidant concentration remained after
1 h if the oxidant doses were higher than 40 mM (Fig. 2). For
hydroxyl radicals, the oxidant doses were quantified by the
kinetic method described in the Supplementary Information,
SI-word. Fig. 3 shows the logarithm of the relative residual
concentration of each selected micropollutant during treatment
of RDWW at pH 8 as a function of the oxidant doses. The
transformation efficiency of micropollutants varied significantly
depending on the types of the selective oxidants and
50
40
30
20
10
water research 44 (2010) 555–566 561
b
ozone ferrateVI 0 10 20 30 40 50 60
d
chlorine chlorine dioxide
0 10 20 30 40 50 60
Time, min
Fig. 2 – Consumption kinetics of the selective oxidants, (a) ozone, (b) ferrate VI , (c) chlorine, and (d) chlorine dioxide, in
a secondary wastewater effluent (RDWW) at pH 8. Symbols represent measured data and lines connect each data point to
show the trend.
micropollutants, and will be discussed in the following
sections.
3.3.1. EE2 and SMX
For these two micropollutants containing activated aromatic
moieties, all four selective oxidants were more efficient than
hydroxyl radicals in terms of the oxidant doses required to
achieve more than one-log (90%) transformation (Figs. 3a,b).
For the selective oxidants, especially in the case of chlorine
and ferrate VI , there was a lag-phase at lower oxidant doses
where the micropollutant concentration decreased little with
increasing oxidant doses. The low transformation efficiency
in the lag-phase is attributed to the high competition for the
selective oxidants between EE2/SMX and the ERMs of EfOM.
With an increase of the selective oxidant doses higher than
the initial consumption by ERMs, the residual concentration of
EE2 and SMX started to decrease significantly. This can be
understood by the depletion of the most reactive ERMs of
EfOM and the corresponding smaller competition for EE2 and
SMX toward the selective oxidants. For hydroxyl radicals, on
the contrary, the log-relative residual concentration of EE2
and SMX was linearly proportional to the applied hydroxyl
radical dose. This indicates that the magnitude of the
competition for hydroxyl radicals between micropollutants
and wastewater effluent matrix remained constant during the
entire oxidation. Therefore, it can be concluded that the initial
oxidation products of EfOM by hydroxyl radicals have similar
k values for their reaction with hydroxyl radicals compared to

562
l 0
og
C/
C
log C/
C0
a
0
-1
17a-ethinylestradiol (EE2)
0
Sulfamethoxazole (SMX)
0
Carbamazepine (CBZ)
ClO2 HOCl
O3 -
HFeO4 ClO2 -2
HOCl
ClO2 -2
O3 HOCl
-2
O3 -
HFeO4 0 20 40 60 80 100 0 30 60 90 120 150 0 30 60 90 120 150
Oxidant dose, µM Oxidant dose, µM Oxidant dose, µM
0
-1
O 3
HO.
Atenolol (ATL) Ibuprofen (IBP)
0
ClO2 HOCl
ClO2HOCl O3 + t-BuOH
HO.
-
HFeO4 -2
0 30 60 90 120 150
-2
0 30 60 90 120 150
-2
0 30 60 90 120 150
Oxidant dose, µM Oxidant dose, µM Oxidant dose, µM
the parent EfOM. This leads to a constant consumption rate of
hydroxyl radicals. Hence, selective oxidants can be much
more efficient than hydroxyl radicals especially when a higher
degree of transformation is required (e.g. >90%) for EE2 or
SMX. The characteristics of transformations of EE2 and SMX
by hydroxyl radicals (i.e. a linear log-transformations vs.
hydroxyl radical doses) were also observed for CBZ, ATL, and
IBP (Figs. 3c–e, respectively).
For EE2 (Fig. 3a), the transformation efficiency to achieve an
one-log transformation for the oxidants was in the order of
chlorine dioxide (3 mM) > ozone z ferrate VI (10 mM) > chlorine
(20 mM) > hydroxyl radical (50 mM) where the value in the
parenthesis shows the required oxidant dose. The very high
efficiency of chlorine dioxide for EE2 transformation can be
understood by its high reactivity to phenolic moieties (Fig. 1a),
but relative low reactivity to amine moieties (Figs. 1d–f). Since
phenolic and amine moieties are expected to be the two main
ERMs present in EfOM, chlorine dioxide can be selective for
the transformation of the micropollutants containing
phenolic moieties in presence of EfOM. In contrast, the relatively
low efficiency of chlorine for EE2 transformation can be
understood by its high reactivity to amine moieties (Figs. 1d–f),
but low reactivity to phenolic moieties (Fig. 1a).
The one-log transformation efficiency in the case of SMX
(Fig. 3b) was in the order of chlorine dioxide (15 mM) z ozone
(20 mM) > ferrate VI
(45 mM) z chlorine (48 mM) > hydroxyl
water research 44 (2010) 555–566
b c
-1
-1
HO.
HO. HO.
-1
-
HFeO4 d e f
O 3
-
HFeO4 0
-1
p-chlorobenzoate (pCBA)
Fig. 3 – Logarithm of the residual concentrations (log(c/c 0)) of selected micropollutants as a function of oxidant doses in
a secondary wastewater effluent (RDWW) at pH 8: (a) EE2, (b) SMX, (c) CBZ, (d) ATL, and (e) IBP. Symbols represent measured
data and lines connect each data point to show the trend. The lines for hydroxyl radicals represent the linear regression of
data. For the selective oxidants, the reaction time of 1 h was given to simulate realistic treatment conditions. Hydroxyl
radicals were in situ produced by UV254 nm/H2O2 photolysis and their doses were quantified by the kinetic method described
in the Supplementary Information, SI-word. For (e) IBP, ‘O3 D tBuOH’ indicates that the ozonation was conducted in
presence of 60 mM tBuOH to exclude the contribution of hydroxyl radicals to the IBP transformation.
HO.
radical (80 mM). The relative lower transformation of SMX
compared to EE2 at the same oxidant dose (Figs. 3a vs. b) are
consistent with the lower k values of SMX than EE2 for the
reaction with the oxidants at pH 8 except the case of chlorine
(Table 1). The sulfonyl-moiety deactivates the aniline-moiety
of SMX toward oxidation as an electron-withdrawing group.
Therefore, the k values for the reaction of SMX with the
selective oxidants are lower than those for aniline (SI-excel).
In contrast, the cyclic ring-moiety activates the phenolicmoiety
of EE2 toward oxidation as an electron-donating group.
Therefore, the k values for the reaction of EE2 with the
selective oxidants are higher than those for phenol (SI-excel).
The smaller transformation of SMX than EE2 at the same
chlorine doses (Figs. 3a vs. b) is not consistent with the 1.7-fold
higher k for the reaction of chlorine with SMX at pH 8 (Table 1).
Dodd et al. found that the reaction of chlorine with SMX
produced the N-chlorinated SMX (i.e. chlorine-adduct to the
aniline-moiety of SMX) as an initial product (Dodd and Huang,
2004). The k values reported in Table 1 correspond to the rate
constant of this initial reaction step. The N-chlorinated SMX
can be reduced back to the parent, SMX, upon its reaction with
reductants such as thiosulfate, which was employed as
a quenching reagent in this study. Therefore, the actual extent
of SMX transformation during chlorination could be smaller
than the predictions based on the k values corresponding to
the initial chlorine addition to SMX. Finally, even though the k

values of ozone or chlorine dioxide for its reaction with EE2
and SMX are several orders of magnitude higher than those of
chlorine or ferrate VI (Table 1), the transformation efficiency in
RDWW among the selective oxidants differed only within
a factor of 20. This is caused by their competitive consumption
by micropollutants and EfOM.
3.3.2. CBZ
The one-log transformation efficiency in the case of CBZ
(Fig. 3c) was in the order of ozone (30 mM) > ferrate VI
(43 mM) > hydroxyl radical (100 mM) [ chlorine z chlorine
dioxide. Chlorine and chlorine dioxide achieved little transformations
of CBZ (

564
3.5.2. Ammonia and nitrite
In addition to EfOMs, inorganic nitrogen compounds, such as
ammonia and nitrite, commonly present in poorly-nitrified/
denitrified wastewater effluents can rapidly consume some
oxidants and affect the transformation efficiency of micropollutants.
The effect of ammonia and nitrite was investigated
using EE2 as an example. From preliminary experiments,
oxidant doses to achieve w80% transformation of EE2 in
RDWW were determined. As a next step, the RDWW spiked
with additional ammonia or nitrite (up to 100 mM) was treated
with the pre-determined oxidant dose. In the case of the
selective oxidants, the residual EE2 concentration was determined
after 1 h of reaction time.
Fig. 4a shows that in the case of chlorine, ammonia
significantly decreased the degree of transformation of EE2. At
an ammonia concentration higher than 20 mM ([NH4 þ ]0/
[HOCl]0 1), the EE2 transformation was less than 10%. This is
because chlorine reacts rapidly with ammonia, forming
chloramines (Deborde and von Gunten, 2008). When the
þ
[NH4 ]0/[HOCl] 0 1, monochloramine (NH2Cl) is the predominant
chloramine species which has a low reactivity to EE2
(k ¼ 0.24 M 1 s 1 ) (Lee et al., 2008). For the other selective
oxidants and hydroxyl radicals, the EE2 transformation was
not affected by the presence of ammonia within the tested
concentration. This is consistent with the low reactivity of
these oxidants with ammonia (Table 1).
Fig. 4b shows that nitrite significantly reduced the transformation
of EE2 in the case of chlorine and ozone, which is
consistent with the high k values of these two oxidants with
nitrite (Table 1). For chlorine, the EE2 transformation decreased
from 80 to 40% with an increase of the nitrite concentration
from 6 to 30 mM ([NO2 ] 0/[HOCl] 0 ¼ from 0.3 to 1.5) and remained
rather constant at w40% with a further increase of the nitrite
concentration even up to 100 mM ([NO2 ]0/[HOCl]0 ¼ 5). The
oxidation of nitrite by chlorine to nitrate proceeds through
some reactive intermediates such as NO2Cl and NO2 þ (Johnson
% transformation
+
[NH4 ]0 /[HOCl] 0
0 1 2 3 4 5
100
80
60
40
20
HOCl
O 3 ClO 2 Fe(VI) HO.
0
0 20 40 60 80 100
+
[NH4 ]0 , µM
water research 44 (2010) 555–566
a b
100
80
60
40
20
and Margerum, 1991). Reactions of these intermediates with
EE2 might be responsible for the incomplete inhibition of EE2
transformation by chlorine at the high nitrite concentration.
For ozone, the EE2 transformation decreased gradually from 80
to 20% with an increase of the nitrite concentration from 6 to
100 mM ([NO2 ]0/[O3]0 ¼ from 0.8 to 12.5). Nitrite had little effect
on the EE2 transformation in the case of chlorine dioxide and
ferrate VI , which is consistent with the relatively low k values of
these two oxidants with nitrite (Table 1). For hydroxyl radical,
nitrite has a slightly positive effect on EE2 transformation even
though it reacts rapidly with nitrite ( OH þ NO 2 / OH þ NO 2 ,
k ¼ 10 10 M 1 s 1 , Table 1). In RDWW spiked with nitrite at the
concentration of 100 mM, w80% of the produced hydroxyl
radicals is scavenged by nitrite during the initial phase of the
UV/H2O2 treatment. This is based on the relative reaction rate
of hydroxyl radicals with nitrite vs. the remaining matrix
components (i.e. EfOM, H2O2, and carbonate). Therefore, the
slightly enhanced transformation of EE2 in presence of 100 mM
nitrite might be explained by the reaction of EE2 with some
intermediates from the reaction of hydroxyl radical with
nitrite such as NO 2 (i.e. EE2 þ NO 2 / EE2 oxid þ NO 2 ). The k
value for the reaction of NO 2 with EE2 is estimated to be
w10 5 M 1 s 1 at pH 8 from the k value for the reaction of NO 2
with 4-methylphenol (Neta et al., 1988).
3.5.3. Bromide
Bromide is present in waters and wastewaters at concentrations
of 10 to several hundred mgL 1 (von Gunten, 2003b). Since
the oxidation of bromide typically produces bromine, which is
highly reactive to phenol- and amine-moieties, the presence of
bromide during oxidative wastewater treatment can affect the
transformation efficiency of micropollutants. Among
the selective oxidants, only chlorine and ozone react with
bromide (Table 1), generating bromine. Since bromine is about
three orders of magnitude more reactive toward phenols than
chlorine (Lee and von Gunten, 2009), transformation of
−
[NO2 ]0 /[O3 ] 0
0 2 4 6 8 10 12
−
[NO2 ]0 /[HOCl] 0
0 1 2 3 4 5
0
0 20 40 60 80 100
−
[NO2 ]0 , µM
Fig. 4 – Effect of (a) ammonia (NH 4 D ) and (b) nitrite (NO2 L ) on the transformations of EE2 during treatment of a secondary
wastewater effluent (RDWW) by different oxidants at pH 8. Preliminary experiments were conducted to determine the
oxidant dose for each oxidant to achieve a 80% transformation of EE2 in RDWW without additionally spiked ammonia and
nitrite. They were 20 mM for chlorine, 3 mM for chlorine dioxide, 8 mM for ozone, 8 mM for ferrate VI , and 37 mM for hydroxyl
radicals. Symbols represent measured data and lines connect each data point to show the trend.
HO.
ClO 2
Fe(VI)
HOCl
O 3

phenolic micropollutants can be significantly enhanced during
chlorination of bromide-containing waters. A recent study
showed that bromine produced from bromide was mainly
responsible for 17a-ethinylestradiol transformation during
chlorination of a wastewater effluent (Lee and von Gunten,
2009). In the case of ozone, most phenolic- and amine-moieties
are already oxidized by ozone before the significant formation
of bromine from bromide. Hence, the presence of bromide
affects little the transformation efficiency of micropollutants
during ozonation. However, an enhanced transformation of 1
amine-containing micropollutants is expected due to the
relatively low reactivity of ozone versus high reactivity of
bromine to 1 amines. For example, the k values at pH 7 for the
reaction with glycine are 1.6 10 2 and 4.7 10 5 M 1 s 1 for
ozone and bromine, respectively (SI-excel). In addition, like to
the enhanced ammonia oxidation by ozone in the presence of
bromide as a catalyst (Haag et al., 1984), N-brominated 1
amines can be oxidized by ozone with a faster rate than the
parent 1 amines and the corresponding reaction regenerates
bromide. Finally, bromide has only a small effect on the
transformation efficiency of micropollutants during advanced
oxidation processes. This is due to (i) the low reactivity of
hydroxyl radicals with bromide ( OH þ Br / OH þ Br ,
k ¼ 1.1 10 9 M 1 s 1 , Table 1) which results in the formation of
HOBr and in the case of H2O2-based AOPs (ii) the fast reaction of
HOBr with H2O2 leading to bromide (von Gunten, 2003b).
Formation of potentially carcinogenic bromate during
ozonation of bromide-containing waters is one of drawbacks
of ozonation. In recent two studies, only low concentration of
bromate formation (i.e.

566
Buxton, G.V., Greenstock, C.L., Helman, W.P., Ross, W.P., 1988.
Critical review of rate constants for reactions of hydrated
electrons, hydrogen atoms and hydroxyl radicals ( OH/ O )in
aqueous solution. J. Phys. Chem. Ref. Data 17, 513–886.
Deborde, M., von Gunten, U., 2008. Reactions of chlorine with
inorganic and organic compounds during water
treatmentdkinetics and mechanisms: a critical review. Water
Res. 42, 13–51.
Deborde, M., Rabouan, S., Gallard, H., Legube, B., 2004. Aqueous
chlorination kinetics of some endocrine disruptors. Environ.
Sci. Technol 38, 5577–5583.
Deborde, M., Rabouan, S., Duguet, J.P., Legube, B., 2005. Kinetics of
aqueous ozone-induced oxidation of some endocrine
disruptors. Environ. Sci. Technol 39, 6086–6092.
Dodd, M.C., Huang, C.H., 2004. Transformation of the
antibacterial agent sulfamethoxazole in reactions with
chlorine: kinetics, mechanisms, and pathways. Environ. Sci.
Technol 38, 5607–5615.
Dodd, M.C., Buffle, M.O., von Gunten, U., 2006. Oxidation of
antibacterial molecules by aqueous ozone: moiety-specific
reaction kinetics and application to ozone-based wastewater
treatment. Environ. Sci. Technol 40, 1969–1977.
Gates, D., 1998. The Chlorine Dioxide Handbook. American Water
Works Association, Denver.
Haag, W.R., Hoigné, J., Bader, H., 1984. Improved ammonia
oxidation by ozone in the presence of bromide ion during
water treatment. Water Res. 18, 1125–1128.
Hoigné, J., 1998. Chemistry of aqueous ozone and transformation
of pollutants by ozonation and advanced oxidation processes.
In: Hrubec, J. (Ed.), The Handbook of Environmental
Chemistry. Springer-Verlag, Berlin Heidelberg, pp. 83–141.
Hollender, J., Zimmermann, S.G., Koepke, S., Krauss, M.,
McArdell, C.S., Ort, C., Singer, H., von Gunten, U., Siegrist, H.,
2009. Elimination of organic micropollutants in a municipal
wastewater treatment plant upgraded with a full-scale postozonation
followed by sand filtration. Environ. Sci. Technol 43,
7862–7869.
Huber, M.M., Canonica, S., Park, G.Y., von Gunten, U., 2003.
Oxidation of pharmaceuticals during ozonation and advanced
oxidation processes. Environ. Sci. Technol 37, 1016–1024.
Huber, M.M., Korhonen, S., Ternes, T.A., von Gunten, U., 2005.
Oxidation of pharmaceuticals during water treatment with
chlorine dioxide. Water Res. 39, 3607–3617.
Johnson, D.W., Margerum, D.W., 1991. Non-metal redox kinetics:
a reexamination of the mechanism of the reaction between
hypochlorite and nitrite ions. Inorg. Chem. 30, 4845–4851.
Kumar, K., Margerum, D.W., 1987. Kinetics and mechanisms of
general-acid-assisted oxidation of bromide by hypochlorite
and hypochlorous acid. Inorg. Chem. 26, 2706–2711.
Lee, Y., Yoon, J., von Gunten, U., 2005a. Kinetics of the oxidation
of phenols and phenolic endocrine disruptors during water
treatment with ferrate (Fe(VI)). Environ. Sci. Technol 39,
8978–8984.
Lee, Y., Yoon, J., von Gunten, U., 2005b. Spectrophotometric
determination of ferrate (Fe(VI)) in water by ABTS. Water Res.
39, 1946–1953.
Lee, Y., Escher, B.I., von Gunten, U., 2008. Efficient removal of
estrogenic activity during oxidative treatment of waters
containing steroid estrogens. Environ. Sci. Technol 42, 6333–6339.
water research 44 (2010) 555–566
Lee, Y., von Gunten, U., 2009. Transformation of 17aethinylestradiol
during water chlorination: effects of bromide
on kinetics, products, and transformation pathways. Environ.
Sci. Technol 43, 480–487.
Lee, Y., von Gunten, U. Kinetics of ferrate VI reactions with
inorganic compounds in water. in preparation.
Lee, Y., Zimmermann, S.G., Kieu, A.T., von Gunten, U., 2009. Ferrate
(Fe(VI)) application for municipal wastewater treatment:
a novel process. Environ. Sci. Technol 43, 3831–3838.
Letterman, R.D., 1999. Water Quality and Treatment, fifth ed.
McGraw-Hill, New York.
Liu, Q., Schurter, L.M., Muller, C.E., Aloisio, S., Francisco, J.S.,
Margerum, D.W., 2001. Kinetics and mechanisms of aqueous
ozone reactions with bromide, sulfite, hydrogen sulfite, iodide,
and nitrite ions. Inorg. Chem. 40, 4436–4442.
Meunier, L., Canonica, S., von Gunten, U., 2006. Implications of
sequential use of UV and ozone for drinking water quality.
Water Res. 40, 1864–1876.
Neta, P., Huie, E., Ross, A.B., 1988. Rate constants for reactions of
inorganic radicals in aqueous solution. J. Phys. Chem. Ref.
Data 17, 1027–1264.
Nowotny, N., Epp, B., von Sonntag, C., Fahlenkamp, H., 2007.
Quantification and modelling of the elimination behavior of
ecologically problematic wastewater micropollutants by
adsorption on powdered and granulated activated carbon.
Environ. Sci. Technol 41, 2050–2055.
Pinkernell, U., Nowack, B., Gallard, H., von Gunten, U., 2000. Methods
for the photometric determination of reactive bromine and
chlorine species with ABTS. Water Res. 34, 4343–4350.
Pinkston, K.E., Sedlak, D.L., 2004. Transformation of aromatic
ether- and amine-containing pharmaceuticals during chlorine
disinfection. Environ. Sci. Technol 38, 4019–4025.
Rosario-Ortiz, F.L., Mezyk, S.P., Doud, D.F.R., Snyder, S.A., 2008.
Quantitative correlation of absolute hydroxyl radical rate
constants with non-isolated effluent organic matter bulk
properties in water. Environ. Sci. Technol 42, 5924–5930.
Sharma, V.K., Bloom, J.T., Joshi, V.N., 1998. Oxidation of ammonia
by ferrate(VI). J. Environ. Sci. Health A33, 635–650.
Ternes, T.A., Joss, A., 2006. Human Pharmaceuticals, Hormones
and Fragrances. The Challenge of Micropollutants in Urban
Water Management. IWA Publishing, London, New York.
Thompson, G.W., Ockerman, G.W., Schreyer, J.M., 1951.
Preparation and purification of potassium ferrate. VI. J. Am.
Chem. Soc. 73, 1379–1381.
von Gunten, U., 2003a. Ozonation of drinking water: Part I. Oxidation
kinetics and product formation. Water Res. 37, 1443–1467.
von Gunten, U., 2003b. Ozonation of drinking water: Part II.
Disinfection and by-products formation in presence of
bromide, iodide or chlorine. Water Res. 37, 1469–1487.
Wert, E.C., Rosario-Ortiz, F.L., Drury, D.D., Snyder, S.A., 2007.
Formation of oxidation byproducts from ozonation of
wastewater. Water Res. 41, 1481–1490.
Westerhoff, P., Mezyk, S.P., Cooper, W.J., Minakata, D., 2007.
Electron pulse radiolysis determination of hydroxyl radical rate
constants with Suwannee river fulvic acid and other dissolved
organic matter isolates. Environ. Sci. Technol 41, 4640–4646.
Zepp, R.G., Hoigné, J., Bader, H., 1987. Nitrate-induced
photooxidation of trace organic chemicals in water. Environ.
Sci. Technol 21, 443–450.

A review of phytoestrogens: Their occurrence
and fate in the environment
Ze-hua Liu a,b, *, Yoshinori Kanjo a , Satoshi Mizutani a
a
Department of Urban Engineering, Graduate School of Engineering, Osaka City University,
3-3-138 Sugimoto, Sumiyoshi-ku, Osaka 558-8585, Japan
b
Department of Chemistry and Chemical Engineering, Shanghai University of Engineering Science,
333 Longteng Road, Songjiang, Shanghai 201620, PR China
article info
Article history:
Received 11 December 2008
Received in revised form
12 March 2009
Accepted 16 March 2009
Available online 24 March 2009
Keywords:
Phytoestrogens
Endocrine disrupting compounds
Wastewater
Urinary excretion rate
Monitoring
1. Introduction
abstract
Endocrine disrupting compounds (EDCs) are chemicals with
the potential to elicit negative effects on the endocrine
systems of humans and wildlife. They include a broad class of
chemicals such as natural estrogens/androgens, synthetic
estrogens/androgens, industrial chemicals as well as phytoestrogens
(Liu et al., 2009). As EDCs exist in extremely low
concentrations (mg/L or ng/L) in the environment, one of the
key points for the research is sensitive and accurate
measurement of EDCs. Nowadays, two categories of methods
are available; chemical analyses such as LC–MS/MS, GC–MS as
well as HPLC and bioassays (in vitro and in vivo). Chemical
water research 44 (2010) 567–577
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Phytoestrogens are plant compounds with estrogenic activities. Many edible plants, some
of which are common in the human diet, are rich in phytoestrogens. Almost all phytoestrogens
eaten daily by people were reported partly recovered in urine or feces, which can
be regarded as one of the main sources of their occurrence in municipal wastewaters. As
they may act as one part of the endocrine disrupting compounds (EDCs) in water systems,
some phytoestrogens have been monitored and detected in wastewater and other various
environments. It is very difficult to monitor numerous unknown EDCs in complex wastewater
samples, and it is helpful if some estimation of target EDCs can be done before
monitoring. With this in mind, this review will: (1) summarize estrogenic activities or
estrogenic potencies of phytoestrogens by different bioassays; (2) summarize daily urinary
excretion rates of phytoestrogens by humans, and compare their urinary excretion rates to
that of estrone, which suggests that most phytoestrogens may occur in municipal wastewaters;
(3) collect and summarize published data on the occurrence and fate of phytoestrogens
in various environments.
ª 2009 Elsevier Ltd. All rights reserved.
analyses can measure precise concentrations of known target
EDCs in environmental samples, while bioassays provide
direct information of the estrogenic activity of complex
mixtures of EDCs that are likely to occur in water irrespective
of the causative compounds. For large-scale monitoring
research, bioassays (in vitro) seem most appropriate due to
their low cost and high throughput capability. However, for
the complexity of environmental samples (usually wastewater
samples), values measured by different bioassays on
the same samples differed greatly, which have lead to questions
over the validity and effectiveness of bioassays for
environmental samples. To find widely acceptable standard
bioassays for environmental samples, the chemically derived
* Corresponding author: Department of Urban Engineering, Graduate School of Engineering, Osaka City University, 3-3-138 Sugimoto,
Sumiyoshi-ku, Osaka 558-8585, Japan. Tel./fax: þ81 6 6605 3048.
E-mail address: liuzehua78@yahoo.co.jp (Z.-h. Liu).
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.watres.2009.03.025

568
values are often compared to those of bioassays, which
accelerates monitoring a multitude of endocrine-like chemicals
in environmental samples (Furuichi et al., 2004; Nakada
et al., 2004; Tan, 2006; Nelson et al., 2007; Salste et al., 2007). In
Fernandez et al. (2007), as many as 30 EDCs have been monitored
in 9 wastewater treatment plants in Canada and the
results of chemically derived estrogen equivalents (EEQ) were
also compared with those measured by recombinant yeast
assay (RYA). The monitoring of numerous EDCs in wastewater
is difficult, and would be preferable if some guidelines were
applicable to wastewater in which EDCs may exist in high
values, while other EDCs may not.
Phytoestrogens are defined as any plant compounds
structurally and/or functionally similar to ovarian and
placental estrogens and their active metabolites (Whitten and
Patisaul, 2001). Phytoestrogens with rather high concentrations
widely exist in numerous plants such as soybeans,
fruits, cabbages and so on (Liggin et al., 2000; US Department
of Agriculture, USDA, 2003). Some of these are widelyconsumed
foods, especially in Asia, where high amounts of
soy are consumed in direct or indirect ways (Adlercreutz et al.,
1991; Nagata et al., 2007). Although most research on phytoestrogens
has suggested that they yield positive effects on
human health (Setchell and Cassidy, 1999; Lissin and Cook,
2000; Kris-Etherton et al., 2002), some research denoted that
most phytoestrogens also acted as endocrine disruptors,
which could cause intersex in aquatic organisms (Kiparissis
et al., 2003; Tzchori et al., 2004). In addition, genotoxicity of
phytoestrogens was reviewed by Stopper et al. (2005), who
pointed out some possible adverse effects of phytoestrogens
on humans.
Due to the possible negative effects mentioned above,
much attention has been paid to the occurrence and fate of
phytoestrogens by environmental researchers in municipal/
industrial wastewater or surface waters, in which they were
detected with high frequency (Kiparissis et al., 2001; Lagana
et al., 2004; Bacaloni et al., 2005; Erbs et al., 2007). Liu et al. (in
press) detected 15 EDCs at two municipal wastewater treatment
plants, in which the concentrations of daidzein in
influents were from 3900 to 12,000 ng/L and its chemically
derived estrogenic equivalents (EEQ) were 0.9–5.3% of those
measured by the ER-binding assay (bisphenol A (BPA) and
nonylphenol (NP) totally contributed 0.7–4.1% in the same
samples), indicating a relatively high proportion in wastewater
which was not negligible. Like natural estrogens or
androgens, the main source of phytoestrogens in municipal
wastewaters can also be attributed to daily excretions of urine
or feces by humans. Therefore, a thorough understanding of
phytoestrogens from daily excretions by humans is helpful for
estimating their possible concentrations in municipal wastewater.
This estimation has been applied to natural estrogens
(Johnson et al., 2000; Johnson and Williams, 2004), and it is
also helpful for suggesting possible priority phytoestrogens
when monitoring for numerous EDCs in wastewater samples
is carried out. Although there is much data on the excretion
rates of phytoestrogens from human urine, few conclusions
have been drawn for the purpose above. In addition, in terms
of convenience to the comparison of chemical analyses with
bioassays, relative binding affinities (RBA) or estrogenic/
androgenic potencies are also important.
water research 44 (2010) 567–577
The objectives of this review are: (1) to summarize reported
estrogenic/androgenic activities of phytoestrogens by
different bioassays; (2) to summarize the daily urinary excretion
rates of phytoestrogens by humans; and (3) to summarize
reported data on the occurrence and fate of phytoestrogens in
the various environments.
2. Estrogenic/androgenic activities of
phytoestrogens
Phytoestrogens are a large family with two major classes,
flavonoids and lignans. The number of flavonoids is greater
than the number of lignans. Flavonoids also possess higher
relative estrogenic activities than those in lignans (Whitten
and Patisaul, 2001). One interest for researchers of environmental
engineering relating to EDCs is the estrogenic/androgenic
activities of phytoestrogens. There is much available
data on their estrogenic/androgenic activities by different
bioassays. To give an overview of their estrogenic/androgenic
activities, some published data are outlined in Table 1, in
which the estrogen standard substance is 17b-estradiol (E2)
and the androgen standard substance is testosterone (Te) or
methyltrienolone (R1881). From Table 1, it is easily seen that
almost all phytoestrogens are far weaker estrogenic
substances, except in the transient gene expression assay
(TGEA), in which the estrogenic potencies of some phytoestrogens
are at the same level, or even stronger than that of E2.
Compared to the estrogenic activities of industrial chemicals
BPA or NP (Blair et al., 2000; Gutendorf and Westendorf, 2001;
Liu et al., 2009), most phytoestrogens are in the same order of
magnitude, which denote that the estrogenic activities
derived from phytoestrogens can not be neglected when they
remain in high concentrations in water samples. In addition,
some phytoestrogens also possess androgenic activities
(Table 1). With the RBA or estrogenic potency in Table 1, we
can calculate the contribution ratio of phytoestrogens (Ratio)
to values measured by bioassays using Eq. (1) and the contribution
ratio of phytoestrogens (Ratioc) to the chemically
derived EEQ using Eq. (2):
Ratio ¼ EEQ P
c RBAiðEPiÞ Ci
¼
(1)
EEQb EEQb where EEQc and EEQb are the chemically calculated EEQ and
the measured EEQ by bioassay, respectively; Ci represents the
concentration of an individual EDC in wastewater sample or
the urinary excretion rate of an individual phytoestrogen from
human and RBAi(EP)i is its corresponding estrogenic activity.
Ratioc ¼ EEQi ¼
EEQc RBAiðEPiÞ Ci
P
RBAiðEPiÞ Ci
where EEQ i is the chemically derived EEQ of an individual
phytoestrogen in urine or an individual EDC in wastewater.
To monitor phytoestrogens in complex wastewater or
other water samples, it is very important to estimate their
concentrations possibly existing in environmental samples.
For phytoestrogens in municipal wastewater, one of the main
sources can be regarded as human urine or feces. Therefore,
their daily urinary excretion rates were concluded as shown
below.
(2)

Table 1 – Estrogenic/androgenic activities of phytoestrogens by different bioassays * .
Class Subclass Phytoestrogen CAS RBA (ER) RBA (AR) Estrogenic potency (EP)
hER(a) a;b
hER(b) b
hAR(a) a
rAR(a) c
RGA a (ER(a)) YES d
(ER(a))
TGEA b
ER(a) ER(b)
Flavonoids Isoflavone Daidzein 486-66-8 1.8e-3; 1e-3 5e-3 N.B – 6.67e-5 3e-6 0.97 0.8
Dihydrodaidzein 17238-05-0 –; – – – – – – – –
Formononetin 485-72-3 2.65e-4;

570
Table 2 – Daily excretion rates of phytoestrogens from human urine.
Phytoestrogens Gender condition
(number)
water research 44 (2010) 567–577
Diet Daily excretion rates
(average)
mg/d mgE2/d c
Mean EEQ Reference
Genistein Men (267) Normal a
n.d.–7579.4 (132.4) 0.159 Low et al., 2005
Women (250) Normal – b (2172.7) 2.607 Dai et al., 2002 d
Women (19) Soy protein diets 1810.6–2837.5 (2357.8) 2.829 Lampe et al., 2001
Women (268) Normal 31.5–1251.3 (219.8) 0.264 Tonkelaar et al., 2001 d
Women (18) Normal 183.5–280.8 (227) 0.272 Xu et al., 2000
7.1 mg total isoflavones/d 362.4–492.9 (422.7) 0.507
65 mg total isoflavones/d 1521.1–2096.2 (1785.7) 2.143
132 mg isoflavones/d 3045.8–4144.6 (3553.1) 4.264
Men (49) Normal – (78.4) 0.094 Lampe et al., 1999
Women (49) Normal – (45.9) 0.055
Men (2) Normal 368.6–587.2 (477.9) 0.573 Adlercreutz et al., 1995 e
Women (3) Normal (2 vegetarians) 482.6–3653.3 (1749.9) 2.100 Adlercreutz et al., 1995
Women (18) Normal 10.8–281.0 (48.6) 0.058 Lampe et al., 1994
Daidzein Men (267) Normal n.d.–4309.8 (227.0) 0.409 Low et al., 2005
Women (250) Normal – (4390.7) 7.903 Dai et al., 2002
Women (19) Soy protein diets 2517.0–5008.5 (3686.4) 6.636 Lampe et al., 2001
Women (18) Normal 200.8–323.4 (254.7) 0.458 Xu et al., 2000
7.1 mg total isoflavones/d 303.8–523.5 (398.9) 0.718
65 mg total isoflavones/d 1281.6–2255.8 (1700.3) 3.061
132 mg isoflavones/d 2722.6–4692.5 (3574.3) 6.434
Men (49) Normal – (211.0) 0.380 Lampe et al., 1999
Women (49) Normal – (94.1) 0.169
Men (2) Normal 1074.2–1113.6 (1093.9) 1.969 Adlercreutz et al., 1995
Women (3) Normal (2 vegetarians) 1240.4–5865.3 (3624.1) 6.523 Adlercreutz et al., 1995
Women (18) Normal 28.0–2333.9 (180.5) 0.325 Lampe et al., 1994
Women (41) Normal 56.1–322.9 (124.2) 0.224 Adlercreutz et al., 1986
Dihydrodaidzein Women (250) Normal – (1340.2) – Dai et al., 2002
Women (18) Normal 151.2–222.7 (183.5) – Xu et al., 2000
7.1 mg total isoflavones/d 221.7–339.5 (274.4) –
65 mg total isoflavones/d 853.1–1334.1 (1066.8) –
132 mg isoflavones/d 1528.8–2341.4 (1891.9) –
ODMA Men (267) Normal n.d.–381.9 (48.1) – Low et al., 2005
Women (250) Normal – (1299.1) – Dai et al., 2002
Women (19) Soy protein diets 206.6–1859.5 (707.0) – Lampe et al., 2001
women (18) Normal 142.6–214.4 (174.8) – Xu et al., 2000
7.1 mg total isoflavones/d 206.4–297.8 (247.9) –
65 mg total isoflavones/d 1045.2–1344.5 (1265.5) –
132 mg isoflavones/d 2236.9–3227.8 (2687.3) –
Men (2) Normal 22.8–335.2 (179.0) – Adlercreutz et al., 1995
Women (3) Normal (2 vegetarians) 171.0–627.1 (379.6) – Adlercreutz et al., 1995
Women (41) Normal 6.5–27.4 (12.7) – Adlercreutz et al., 1986
Equol Men (267) Normal n.d.–1280.5 (18.2) 0.157 Low et al., 2005
Women (19) Soy protein diet n.d.–581.4 (271.0) 2.331 Lampe et al., 2001
Women (18) Normal 45.1–64.4 (53.8) 0.447 Xu et al., 2000
7.1 mg total isoflavones/d 61.3–120.4 (85.8) 0.738
65 mg total isoflavones/d 126.9–249.5 (178.1) 1.532
132 mg isoflavones/d 189.5–372.4 (265.8) 2.286
Men (49) Normal – (21.8) 0.187 Lampe et al., 1999
Women (49) Normal – (17.0) 0.146
Men (2) Normal 26.6–117.5 (72.1) 0.620 Adlercreutz et al., 1995
Women (3) Normal (2 vegetarians) 16.9–60.8 (37.7) 0.324 Adlercreutz et al., 1995
Women (18) Normal 14.5–43.6 (26.6) 0.229 Lampe et al., 1994
Women (41) Normal 15.5–24.7 (20.4) 0.175 Adlercreutz et al., 1986

Table 2 (continued)
Phytoestrogens Gender condition
(number)
3. Occurrence and fate of phytoestrogens in
the environment
3.1. Phytoestrogens in human urine
Phytoestrogens in human urine are from the diet or dietary
metabolites, and their concentrations in urine have been
found dose dependent on their consumption (Karr et al.,
1997; Zhang et al., 1999; Xu et al., 2000; Peeters et al., 2003).
For disease related items, there has been much research on
the daily excretion rates of urinary phytoestrogens, in which
the excretion rates varied greatly with race and dietary
customs (Horn-Ross et al., 1997; Seow et al., 1998). To give an
overview, published data on urinary excretion rates are
outlined in Table 2. In order to compare the estrogenic
potencies of phytoestrogens from human urine, the authors
water research 44 (2010) 567–577 571
Diet Daily excretion rates
(average)
mg/d mgE2/d c
Mean EEQ Reference
Glycitein Men (267) Normal n.d.–556.2 (51.3) 0.014 Low et al., 2005
Women (250) Normal – (690.8) 0.193 Dai et al., 2002
Women (18) Normal 73.9–94.1 (83.3) 0.023 Xu et al., 2000
7.1 mg total isoflavones/d 109.4–141.3 (120.8) 0.034
65 mg total isoflavones/d 149.0–292.8 (209.9) 0.059
132 mg isoflavones/d 222.3–436.9 (311.8) 0.087
Coumestrol Women (18) Normal 8.3–9.5 (8.9) 0.023 Xu et al., 2000
7.1 mg total isoflavones/d 7.9–9.2 (8.6) 0.023
65 mg total isoflavones/d 8.4–9.7 (9.0) 0.024
132 mg isoflavones/d 8.8–10.2 (9.5) 0.025
Enterolactone Men (267) Normal n.d.–13218.5 (1020.4) 0.033 Low et al., 2005
Women (250) Normal – (1873.5) 0.060 Dai et al., 2002
Women (19) Soy protein diets 477.3–2386.7 (1555.1) 0.050 Lampe et al., 2001
Women (268) Normal 19.0–6157.9 (1276.2) 0.041 Tonkelaar et al., 2001
Women (18) Normal 364.3–417.1 (389.6) 0.012 Xu et al., 2000
7.1 mg total isoflavones/d 347–489.6 (412.3) 0.013
65 mg total isoflavones/d 599.6–858.3 (717.2) 0.023
132 mg isoflavones/d 960.6–1352.0 (1139.6) 0.036
Men (49) Normal – (683.2) 0.022 Lampe et al., 1999
Women (49) Normal – (584.7) 0.019
Men (2) Normal 426.9–1592.8 (1009.9) 0.032 Adlercreutz et al., 1995
Women (3) Normal (2 vegetarians) 1305.2–2488.7 (1851.2) 0.059 Adlercreutz et al., 1995
Women (41) Normal 611.6–1244.0 (914.3) 0.029 Adlercreutz et al., 1986
Naringenin Women (250) Normal – (1217.0) 0.065 Dai et al., 2002
Men and women (77) Normal – (653.8) 0.035 Nielsen et al., 2002
Quercetin Men and women (77) Normal – (23.4) – Nielsen et al., 2002
Phloretin Men and women (77) Normal – (84.8) 0.013 Nielsen et al., 2002
Kaempferol Men and women (77) Normal – (51.3) 0.015 Nielsen et al., 2002
a Normal diet.
b Not available.
c Calculated as EEQi in Eq. (2), values of RBA based on ER-binding assay (NITE, 2008).
d Transferred from nmol/mg creatinine or mmol/mol creatinine to mg/d, and creatinine excretion rate of 1 g/d for women was chosen (Jackson,
1966).
e Calculated from their free phytoestrogens and their conjugates.
converted the urinary excretion rate of the individual phytoestrogens
into a corresponding value, calculated as EEQi in
the numerators of Eq. (2); we label it ‘‘excretion rate of EEQ’’.
As is shown in Table 2, out of 12 phytoestrogens, genistein
and daidzein are the two most important phytoestrogens
evaluated by both urinary excretion rate and excretion rate
of EEQ; followed by equol, for which its urinary excretion
rates by normal diets are low but with a relatively high
corresponding excretion rate of EEQ. Enterolactone and
naringenin are the phytoestrogens with high urinary excretion
rates, but very low estrogenic potencies (low RBA, Table
1). Therefore, the corresponding excretion rates of EEQ by
enterolactone and naringenin seem relatively low. Data of
urinary excretion rates of some phytoestrogens, especially
for quercetin, phloretin as well as kaempferol, is very
limited. For better comparison, it would be helpful if more
data were available.

572
In addition, biochanin A and formononetin have been
found existing in high concentrations in red clovers of
Canada, and have also been detected in many fruits, beans
and vegetables (Mazur et al., 1996; Sivesind and Seguin, 2005;
Kuhnle et al., 2007, 2008). Few data on urinary excretion rates
are available. As a reference, one published paper showed the
average fecal excretion rate was about 1.5 mg/d for biochannin
A and 11.9 mg/d for formononetin (Jenner et al.,
2005). A phytosterol b-sitosterol with estrogenic activity
(Table 1) was found rich in a wide range of fruits, vegetables,
nuts, and many kinds of seed oils (Grosso et al., 2000; Beveridge
et al., 2002, 2005; Iwatsuki et al., 2003; Holser et al.,
2004; Phillips et al., 2005; Jimenez-Escrig et al., 2006; Zhang
et al., 2006). Although there is no available data on the
urinary excretion rate of b-sitosterol, very high human fecal
excretion rates of 52–322 mg/d and a high human fecal
concentration of 121 mg/g in dry feces (average) were reported
(Eneroth et al., 1964; Leeming et al., 1996), which suggested
that b-sitosterol might also exist in wastewater with a very
high concentration.
From the view of statistics, a rough estimation on the
arithmetic average of urinary excretion rates was done on
the data of normal diets in Table 2, which was also applied
on human urinary excretion rates of natural estrogens and
androgens (Liu et al., submitted for publication). If the loss of
phytoestrogens during the course of flow to wastewater
treatment plants is low, their concentrations in wastewater
can be compared to those of natural estrogens. Although it is
well known that 17b-estradiol (E2) is the strongest natural
estrogen, estrone (E1) is the one which was detected
frequently with high concentrations in effluent of WWTPs
(Baronti et al., 2000; Joss et al., 2004; Johnson et al., 2007; Liu
et al., 2009). Therefore, the average urinary excretion rates of
phytoestrogens were compared to that of E1 as shown in
Fig. 1, and the urinary excretion rate of E1 was modified from
Liu et al. (submitted for publication), for which the urinary
excretion rates of two pregnant women out of 177 persons
(11.3&) were also included. From Fig. 1 it is known that the
urinary excretion rates of phytoestrogens are 1–86 times that
of E1, except for coumestrol (0.45 times); Among phytoestrogens
in Table 2, genistein and daidzein, frequently detected
in wastewaters, are 35 and 86 times that of E1,
urinary excretion rates(µg/d)
10000
1000
100
10
1
E1
Genistein
Daidzein
Dihydrodaidzein
ODMA
Glycitein
Equol
Coumestrol
Enterlactone
Naringerin
Quercetin
Phloretin
Fig. 1 – Comparison of urinary excretion rates between
phytoestrogens and E1.
water research 44 (2010) 567–577
Kaemperol
respectively. According to published results, the average
concentrations of genistein and daidzein in influent of one
WWTP in Italy were about 3 and 9 times that of the detected
E1 (Lagana et al., 2004), while the concentrations of daidzein
in 12 influent samples of two municipal WWTPs in Japan
showed about 8–46 times (22 times on average) that of the
detected E1 (Liu et al., in press). Compared to detected
concentrations for genistein and daidzein, their estimated
concentrations in Fig. 1 are about 10 times and 2–10 times
higher. The large differences may result from the different
loss rates before they reach WWTPs, different excretion
ratios of urine and feces for phytoestrogens and E1, as well as
regional dietary customs. Although coumestrol is the leastpresent
urinary excretion phytoestrogen in Fig 1, it was also
reported existing in municipal WWTPs (Bacaloni et al., 2005).
Based on this fact, the other 11 phytoestrogens in Fig. 1 are
very likely to be present in municipal wastewater. Based on
EEQ, phytoestrogens such as daidzein, genistein, equol and
glycitein should not be neglected, for which the excretion
rates of EEQ are about 35%, 10%, 2% and 1% that of human
urinary E1. As enterlactone and naringerin are much weaker
estrogens than daidzein (Table 1), although their urinary
excretion rates are on the same level as daidzein, the corresponding
excretion rates of EEQ are just about 0.5% and
0.7% that of human urinary E1. However, the urinary excretion
rates of enterlactone in some regions may be much
higher than that in Fig. 1 (Lampe et al., 1994; Wang, 2002;
Milder et al., 2005; Valentin-Blasini et al., 2005; Penalvo et al.,
2008), which may yield a higher contribution ratio in wastewater.
In addition, the urinary excretion rates of dihydrodaidzein
and ODMA are almost the same as that of
daidzein, but the data on their estrogenic activities is not
available; therefore, their corresponding excretion rates of
EEQ can not be applied here.
3.2. Phytoestrogens in wastewaters and the other
water bodies
As predicted above that most phytoestrogens may exist in
wastewater, some phytoestrogens have been proven to not
only exist in wastewater but also in surface waters (Kawanishi
et al., 2004; Lagana et al., 2004; Bacaloni et al., 2005; Erbs et al.,
2007). Mycoestrogens (Table 1) are zearalenone and its derivatives.
They are macrocyclic toxins produced by several
Fusarium species, colonizing maize, sorghum, wheat, barley
oats and other cereal grains, which were also reported
occurring in wastewaters and surface waters (Lagana et al.,
2004). Published data on the occurrence and fate of phytoestrogens
in the course of WWTPs are included in Table 3.
As stated in Section 1, to give a whole profile of EDCs in the
wastewater of WWTPs, a combination of chemical analyses
and bioassays has been used, and estrogenic contribution
ratios of detected EDCs to those of chemical derived EEQ have
also been compared. As shown in Table 3, it is known that
some phytoestrogens have been detected in both influent and
effluent, and their removal efficiencies varied greatly as
shown for natural estrogens and androgens (Liu et al., 2009).
Data in Table 3 also suggested that the estrogenic contribution
ratios of phytoestrogens could not be ignored in wastewaters
when present in relatively high concentrations. For example,

Table 3 – Occurrence and fate of phytoestrogens in wastewater.
Phytoestrogens Country WWTP
type (n)
the authors calculated that six phytoestrogens in influents of
a municipal WWTP contributed over 2.4% of total chemically
calculated EEQ including E1, E2, E3 and EE2 (RBA values of
EDCs are chosen from NITE, 2008), and increased to over 4.5%
for the corresponding effluents (Lagana et al., 2004); In 12
influent samples of two municipal WWTPs, the contribution
ratios of daidzein accounted for 3.0–14.8% of total chemically
calculated EEQ (Liu et al., in press). In the influents of a swine
wastewater treatment plant, the contribution ratios of phytoestrogen
equol accounted for as high as about 64% of total
chemically derived EEQ (Furuichi et al., 2006). In addition, for
the extremely high concentrations of b-sitosterol in wastewaters
(Table 3), its estrogenic contribution ratio in wastewater
seems relatively high and it also can not be neglected.
Similar to natural estrogens and androgens, about 97–
100% of phytoestrogens excreted from human urine are
phytoestrogen conjugates, in which 62–97% of phytoestrogens
are glucuronide conjugates (Adlercreutz et al., 1995).
Although there is no information on the occurrence and fate
of phytoestrogen conjugates in WWTPs, the glycoside
conjugates of daidzein and genistein (daidzin and genistin),
which have structures similar to those of their glucuronide
conjugates, have been proven hydrolyzable to daidzein and
water research 44 (2010) 567–577 573
Methods LOD Influent Effluent Average
removal (%)
Reference
Daidzein Germany AS (2) – 10 – n.d.–10 – Pawlowski et al., 2003
Italy AS (1) LC–MS/MS 2.5–3.5 75–120 7–16 88 Lagana et al., 2004
Italy AS (2) LC–MS/MS 3.5 97–1685 5–81 88 or 98 Bacaloni et al., 2005
Japan AS (2) LC–MS/MS 5–60 b
3900–12000 5.7–23 99.8 Liu et al., in press
Daidzin Italy AS (2) LC–MS/MS 9–15 18–74 n.d.–22 69 or 100 Bacaloni et al., 2005
Genistein Canada – (1) LC–MS/MS – 13100 10100 23 Kiparissis et al., 2001
Italy As (1) LC–MS/MS 1.3–4.3 195–384 15–83 81 Lagana et al., 2004
Italy AS (2) LC–MS/MS 5–6 160–954 7–22 97 Bacaloni et al., 2005
Genistin Italy AS (2) LC–MS/MS 9–13 11–62 n.d.–19 66 or 100 Bacaloni et al., 2005
Glycitein Italy AS (2) LC–MS/MS 6–8 6–25 n.d.–16 62 or 88 Bacaloni et al., 2005
Formononetin Italy AS (2) LC–MS/MS 5–8 n.d.–10 n.d. – Bacaloni et al., 2005
Coumestrol Italy AS (2) LC–MS/MS 2–5 9–23 n.d.–19 66 or 100 Bacaloni et al., 2005
Biochanin-A Italy As (1) LC–MS/MS 0.9–5.4 8–18 3–5 73 Lagana et al., 2004
Italy AS (2) LC–MS/MS 5–9 14–76 n.d.–21 67 or 90 Bacaloni et al., 2005
Equol Japan – (1) LC–MS/MS – 9.4e5–1.1e6 170–410 99.99 Furuichi et al., 2006
Zearlenone Italy AS (1) LC–MS/MS 0.6–1.7 3–18 3–10 53 Lagana et al., 2004
a-Zearalanol Italy AS (1) LC–MS/MS 5.4–8.3 n.d.–10 n.d.–7 0 Lagana et al., 2004
b-Zearalanol Italy AS (1) LC–MS/MS 3.7–8.4 n.d.–8 n.d.–5 33 Lagana et al., 2004
a-Zearalenol Germany AS (2) – 10 – 20 or 30 – Pawlowski et al., 2003
b-Sitosterol Switzerland – (1) GC–MS 125 – 400–1300 – Baig et al., 2008
Canada AS (3);
TF/SC (1);
L (1)
GC–HRMS 12–39 7303–35033 350–42110 – Fernandez et al., 2007
USA – (2) GC–MS – 32000 a
– – Cain et al., 2008
Germany – (2) GC–MS – 14000or 16000 – – Radke, 2005
Germany AS (2) – 50 – 60 or 90 – Pawlowski et al., 2003
Germany – (7) GC–MS/MS – 15900 1900 88 Heberer, 2002
France AS (2) GC–MS – 4194–49920 1100–50165 – Quemeneur and Marty,
1994
Enterolactone Australian – (1) LC–MS 1 600 – – Kang et al., 2006
Finland AS (1) LC–MS/MS – 1223 659 46 Smeds et al., 2007
AS, activated sludge; TF/SC, trickling filter/solid contact; L, lagoon; –, not available; n.d., not detected; LOD, limit of detection.
a Maximum concentration.
b Limit of quantification.
genistein by the glucosidase of human intestinal bacteria
(Hur et al., 2000). The detected daidzin and genistin in some
effluents of WWTPs suggested their incomplete hydrolysis
during the whole activated sludge process (Table 3). In addition,
based on the fact that natural estrogen conjugates
occurred both the influent and effluent of WWTPs (D’Ascenzo
et al., 2003; Isobe et al., 2003; Komori et al., 2004; Reddy et al.,
2005; Nakada et al., 2006), phytoestrogen conjugates may also
be present in influent and effluent of WWTPs with high
frequency.
Published data on the occurrence of phytoestrogens in
other various environments are collected in Table 4. From
Table 4, it is known that most phytoestrogens including
mycoestrogens were found in rivers, and in some occasions
the concentrations of phytoestrogens in drainage water were
much higher than those in river water, which indicated that
the drainage water was also a source of phytoestrogens in
river waters. In Table 4 b-sitosterol showed very high
concentrations in rivers and it was also detected in the
atmosphere. However, it has not been detected in drinking
water (Quemeneur and Marty, 1994; Heberer, 2002; Martins
et al., 2007; Stackelberg et al., 2007). In addition, enterlactone
was seldom monitored in wastewater as a micro-pollutant,

574
Table 4 – Phytoestrogens in the other various environments.
Phytoestrogens River water
(ng/L)
whereas the recent investigation denoted that it not only
existed in wastewater, but also in river water, lake water, sea
water and drinking water (Table 4).
4. Conclusions
Drainage water
(ng/L)
Urinary excretion rates of phytoestrogens from humans and
their occurrence and fate in the environment were summarized
here. It was concluded that most phytoestrogens may be
present in municipal wastewaters as can be estimated from
their urinary excretion rates. Although phytoestrogens are
very weak estrogens when compared to natural estrogens,
their possible high concentrations in wastewater may lead
them to contribute a non-negligible part of EEQ. Compared to
natural estrogens or industrial chemicals such as BPA, NP and
Octylphenol (OP), the research of phytoestrogens in wastewaters
is just beginning. However, with the progress of
analytic methods, knowledge of phytoestrogens in wastewater,
such as removal mechanisms of phytoestrogens at
water research 44 (2010) 567–577
River sediments
(ng/g)
WWTPs, will grow. On some occasions, chemical advanced
oxidation should also be adopted to degrade phytoestrogens
when purer water is required.
references
Sea water
(ng/L)
Drinking
water
Air
(pg/m 3 )
Daidzein 2–3 [1]; 5–30 [3] – a
2–4 [2];
42900 [4];

Bacaloni, A., Cavaliere, C., Faberi, A., Foglia, P., Samperi, R.,
Lagana, A., 2005. Determination of isoflavones and coumestrol
in river water and domestic wastewater sewage treatment
plants. Analytica. Chimica. Acta 531, 229–237.
Baig, S., Hansmann, G., Paolini, B., 2008. Ozone oxidation of
estrogenic active substances in wastewater and drinking
water. Wat. Sci. Technol 58, 451–458.
Baronti, C., Curini, R., D’Ascenzo, G., Corcia, A.D., Gentili, A.,
Samperi, R., 2000. Monitoring natural and synthetic estrogens
at activated sludge sewage treatment plants and in a receiving
river water. Environ. Sci. Technol 34, 5059–5066.
Beck, M., Radke, M., 2006. Determination of sterols, estrogens and
inorganic ions in waste water and size-segregated aerosol
particles emitted from waste water treatment. Chemosphere
64, 1134–1140.
Beck, I.C., Bruhn, R., Gandrass, J., Ruck, W., 2005. Liquid
chromatography-tandem mass spectrometry analysis of
estrogenic compounds in coastal surface water of the Baltic
Sea. J. Chromatogr. A 1090, 98–106.
Beveridge, T.H.J., Li, T.S.C., Drover, J.C.G., 2002. Phytosterol content
in American ginseng seed oil. J. Agric. Food Chem. 50, 744–750.
Beveridge, T.H.J., Girard, B., Kopp, T., Drover, J.C.G., 2005. Yield
and composition of grape seed oils extracted by supercritical
carbon dioxide and petroleum ether: varietal effects. J. Agric.
Food Chem. 53, 1799–1804.
Blair, R.M., Fang, H., Branham, W.S., Hass, B.S., Dial, S.L.,
Moland, C.L., Tong, W., 2000. The estrogen receptor relative
binding affinities of 188 natural and xenochemicals: structural
diversity of ligands. Toxicol. Sci. 54, 138–153.
Bovee, T.F.H., Helsdingen, R.J.R., Rietjens, I.M.C.M., Keijer, J.,
Hoogenboom, R.L.A.P., 2004. Rapid yeast estrogen bioassays
stably expressing human estrogen receptor a and b, and green
fluorescent protein: a comparison of different compounds with
both receptor types. J. Steroid Biochem. Mol. Biol. 91, 99–109.
Cain, T.G., Kolpin, D.W., Vargo, J.D., Wichman, M.D., 2008.
Occurrence of antibiotics, pharmaceuticals and sterols at
select surface and wastewater sites in Iowa. http://info.ngwa.
org/gwol/pdf/042379985.pdf. 2008.11.24.
Chou, C.C., Liu, Y.P., 2004. Determination of fecal sterols in the
sediments of different wastewater outputs by GC-MS. Intern. J.
Environ. Anal. Chem. 84, 379–388.
D’Ascenzo, G., Corcia, A.D., Mancini, A.G.R., Mastropasqua, R.,
Nazzari, M., Samperi, R., 2003. Fate of natural estrogen
conjugates in municipal sewage transport and treatment
facilities. Sci. Total Environ 302, 199–209.
Dai, Q., Franke, A.A., Jin, F., Shu, X.O., Hebert, J.R., Custer, L.J.,
Cheng, J.R., Gao, Y.T., Zheng, W., 2002. Urinary excretion of
phytoestrogens and risk of breast cancer among Chinese
women in Shanghai. Cancer Epidem. Biomarkers 11, 815–821.
Devane, M., Saunders, D., Gilpin, B., 2006. Faecal sterols and
fluorescene whiteners as indicators of the source of faecal
contamination. Chem. New Zealand 70, 74–77.
Eneroth, P., Hellstrom, K., Ryhage, R., 1964. Identification and
quantification of neutral fecal steroids by gas-liquid
chromatography and mass spectrometry: studies of human
excretion during two dietary regimens. J. Lipid Res. 5, 245–262.
Erbs, M., Hoerger, C.C., Hartmann, N., Bucheli, T.D., 2007.
Quantification of six phytoestrogens at the nanogram per liter
level in aqueous environmental samples using 13 C 3-labeled
internal standards. J. Agric. Food Chem. 55, 8339–8345.
Fang, H., Tong, W., Branham, W.S., Moland, C.L., Dial, S.L.D.,
Hong, H., Xie, Q., Perkins, R., Owens, W., Sheehan, D.M., 2003.
Study of 202 natural, synthetic, and environmental chemicals
for binding to the androgen receptor. Chem. Res. Toxicol 16,
1338–1358.
Fattore, E., Benfenati, E., Marelli, R., Cools, E., Fanelli, R., 1996.
Sterols in sediment samples from Venice lagoon. Italy.
Chemosphere 33, 2383–2393.
water research 44 (2010) 567–577 575
Fernandez, M.P., Ikonomou, M.G., Buchanan, I., 2007. An
assessment of estrogenic organic contaminants in Canadian
wastewaters. Sci. Total Environ 373, 250–269.
Furuichi, T., Kannan, K., Giesy, J.P., Masunaga, S., 2004.
Contribution of known endocrine disrupting substances to the
estrogenic activity in Tama River water samples from Japan
using instrumental analysis and in vitro reporter gene assay.
Wat. Res. 38, 4491–4501.
Furuichi, T., Kannan, K., Suzuki, K., Tanaka, S., Giesy, J.P.,
Masunaga, S., 2006. Occurrence of estrogenic compounds in
and removal by a swine farm waste treatment plant. Environ.
Sci. Technol 40, 7896–7902.
Grosso, N.R., Nepote, V., Guzman, C.A., 2000. Chemical
composition of some wild peanut species (Arachis L.) seeds. J.
Agric. Food Chem. 48, 806–809.
Gutendorf, B., Westendorf, J., 2001. Comparison of an array of in
vitro assays for the assessment of the estrogenic potential of
natural and synthetic estrogens, phytoestrogens and
xenoestrogens. Toxicology 166, 79–89.
Hartmann, N., Erbs, M., Wettstein, F.E., Schwarzenbach, R.P.,
Bucheli, T.D., 2007. Quantification of estrogenic mycotoxins at
the ng/L level in aqueous environmental samples using
deuterated internal standards. J. Chromatogr. A 1138, 132–140.
Heberer, T., 2002. Tracking persistent pharmaceutical residues
from municipal sewage to drinking water. J. Hydrol 266,
175–189.
Holser, R.A., Bost, G., Boven, M.V., 2004. Phytosterol composition
of hybrid Hibiscus seed oils. J. Agric. Food Chem. 52, 2546–2548.
Horn-Ross, P.L., Barnes, S., Kirk, M., Coward, L., Parsonnet, J.,
Hiatt, R.A., 1997. Urinary phytoestrogen levels in young
women from a multiethnic population. Cancer Epidem.
Biomar 6, 339–345.
Hur, H.G., Lay Jr., J.O., Beger, R.D., Freeman, J.P., Rafii, F., 2000.
Isolation of human intestinal bacteria metabolizing the
natural isoflavone glycosides daidzin and genistin. Arch.
Microbiol. 174, 422–428.
Isobe, T., Shiraishi, H., Yasuda, M., Shinoda, A., Suzuki, H.,
Morita, M., 2003. Determination of estrogens and their
conjugates in water using solid-phase extraction followed by
liquid chromatography-tandem mass spectrometry. J.
Chromatogr. A 984, 195–202.
Iwatsuki, K., Akihisa, T., Tokuda, H., Ukiya, M., Higashihara, H.,
Mukainaka, T., Izuka, M., Hayashi, Y., Kimura, Y., Nishino, H.,
2003. Sterol ferulates, sterols, and 5-alk(en)yl resorcinols from
wheat, rye and corn bran oils and their inhibitory effects on
Epstein-Barr virus activation. J. Agric. Food Chem. 51, 6683–
6688.
Jackson, S., 1966. Creatinine in urine as an index of urinary
excretion rate. Health Phys. 12, 843–850.
Jenner, A.M., Rafter, J., Halliwell, B., 2005. Human fecal water
content of phenolics: the extent of colonic exposure to
aromatic compounds. Free Radic. Biol. Med 38, 763–772.
Jimenez-Escrig, A., Santos-Hidalgo, A.B., Saura-Calixto, F., 2006.
Common sources and estimated intake of plant sterols in the
Spanish diet. J. Agric. Food Chem 54, 3462–3471.
Johnson, A.C., Williams, R.J., 2004. A model to estimate influent
and effluent concentrations of estradiol. estrone, and
ethinylestradiol at sewage treatment works. Environ. Sci.
Technol 38, 3649–3658.
Johnson, A.C., Belfroid, A., Di Corcia, A., 2000. Estimating steroid
oestrogen inputs into activated sludge treatment works and
observations on their removal from the effluent. Sci. Total
Environ 256, 163–173.
Johnson, A.C., Williams, R.J., Simpson, P., Kanda, R., 2007. What
difference might sewage treatment performance make to
endocrine disruption in rivers. Environ. Pollut 147, 194–202.
Joss, A., Andersen, H., Ternes, T., 2004. Removal of estrogens in
municipal wastewater treatment under aerobic and anaerobic

576
conditions: consequences for plant optimization. Environ. Sci.
Technol 38, 3047–3055.
Kang, J.G., Price, W.E., Hick, L.A., 2006. Simultaneous
determination of isoflavones and lignans at trace levels in
natural waters and wastewater samples using liquid
chromatography/electrospray ionization ion trap mass
spectrometry. Rapid Commun. Mass Spectrom 20, 2411–2418.
Karr, S.C., Lampe, J.W., Hutchins, A.M., Slavin, J.L., 1997. Urinary
isoflavonoid excretion in humans is dose dependent at low to
moderate levels of soy-protein comsumption. Am. J. Clin. Nutr
66, 46–51.
Kawanishi, M., Takamura-Enya, T., Ermawati, R., Shimohara, C.,
Sakamoto, M., Matsukawa, K., Matsuda, T., Murahashi, T.,
Matsushi, T., Wakabayashi, K., Watanabe, T., Tashiro, Y.,
Yagi, T., 2004. Detection of genistein as an estrogenic
contaminant of river water in Osaka. Environ. Sci. Technol 38,
6424–6429.
Kiparissis, Y., Hughes, R., Metcalfe, C., Ternes, T., 2001.
Identification of the isoflavonoid genistein in bleached kraft
mill effluent. Environ. Sci. Technol 35, 2423–2427.
Kiparissis, Y., Balch, G.C., Metcalfe, T.L., Metcalfe, C.D., 2003.
Effects of the isoflavones genistein and equol on the gonadal
development of Japanese medaka (Oryzias latipes). Environ.
Health Persp 111, 1158–1163.
Komori, K., Tanaka, H., Okayasu, Y., Yasojima, M., Sato, C., 2004.
Analysis and occurrence of estrogen in wastewater in Japan.
Wat. Sci. Technol 50, 93–100.
Kostelac, D., Rechkemmer, G., Briviba, K., 2003. Phytoestrogens
modulate binding response of estrogen receptor a and b to the
estrogen response element. J. Agric. Food Chem. 51, 7632–
7635.
Kris-Etherton, P.M., Hecker, K.D., Bonanome, A., Coval, S.M.,
Binkoski, A.E., Hilpert, K.F., Griel, A.E., Etherton, T.D., 2002.
Bioactive compounds in foods: their role in the prevention of
cardiovascular disease and cancer. Am.. J. Med 113 (Suppl), 71–88.
Kuhnle, G.G.C., Dell’Aquila, C., Kussmaul, M., Bingham, S.A., 2007.
Extraction and quantification of phytoestrogens in foods using
automated solid-phase extraction and LC/MS/MS. Anal. Chem.
79, 9234–9239.
Kuhnle, G.G.C., Dell’aquila, C., Aspinall, S.M., Runswick, S.A.,
Mulligan, A.A., Bingham, S.A., 2008. Phytoestrogen content of
beverages, nuts, seeds, and oils. J. Agric. Food Chem. 56, 7311–
7315.
Kuiper, G., Lemmen, J.G., Carlsson, B., Corton, J.C., Safe, S.H., van
der Saag, P.T., van der Burg, B., Gustafsson, J.A., 1998.
Interaction of estrogenic chemicals and phytoestrogens with
estrogen receptor b. Endocrinology 139, 4252–4263.
Lagana, A., Bacaloni, A., De Leva, I., Faberi, A., Fago, G., Marino, A.,
2004. Analytical methodologies for determining the
occurrence of endocrine disrupting chemicals in sewage
treatment plants and natural waters. Analytica. Chimica. Acta
501, 79–88.
Lampe, J.W., Wartini, M.C., Kurzer, M.S., Adlercreutz, H., Slavin, J.
L., 1994. Urinary lignan and isoflavonoid excretion in
premenopausal women consuming flaxseed powder. J. Clin.
Nutr 60, 122–128.
Lampe, J.W., Gustafson, D.R., Hutchins, A.M., Martini, M.L., Li, S.,
Wahala, K., Grandits, G.A., Potter, J.D., Slavin, J.L., 1999.
Urinary isoflavonoid and lignan excretion on a western diet:
relation to soy, vegetable, and fruit intake. Cancer Epidem.
Biomarkers 8, 699–707.
Lampe, J.W., Skor, H.E., Li, S., Wahala, K., Howald, W.N., Chen, C.,
2001. Wheat bran and soy protein feeding do not alter urinary
excretion of the isoflavan equol in premenopausal women. J.
Nutr 131, 740–744.
Leeming, R., Ball, A., Ashbolt, N., Nichols, P., 1996. Using faecal
sterols from humans and animals to distinguish faecal
pollution in receiving waters. Wat. Res. 30, 2893–2900.
water research 44 (2010) 567–577
Liggin, J., Bluck, L.J.C., Runswick, S., Atkinson, C., Coward, W.A.,
Bingham, S.A., 2000. Daidzein and genistein content of fruits
and nuts. J. Nutr. Biochem 11, 326–331.
Lissin, L.W., Cook, J.P., 2000. Phytoestrogens and cardiovascular
health. J. Am. Coll. Cardiol 35, 1403–1410.
Liu, Z.H., Kanjo, Y., Mizutani, S., 2009. Removal mechanisms for
endocrine disrupting compounds (EDcs) in wastewater
treatment-physical means, biodegradation, and chemical
advanced oxidation: A review. Sci. Total Environ 407, 731–748.
Liu, Z.H., Kanjo, Y., Mizutani, S. Excretions of natural estrogens
and androgens from humans, and their occurrence and fate in
the environments. Sci. Total Environ, submitted for
publication.
Liu, Z.H., Mamoru, I., Kanjo, Y., Yamamoto, A. Profile and removal
of endocrine disrupting chemicals (EDCs) in two municipal
wastewater treatment plants by using an ER (AR) competitive
ligand binding assay and chemical analyses. J. Environ. Sci. in
press. doi:10.1016/S1001-0742(08)62356-6.
Low, Y.L., Taylor, J.I., Grace, P.B., Dowsett, M., Folkerd, E.,
Doody, D., Dunning, A.M., Scollen, S., Mulligan, A.A., Welch, A.
A., Luben, R.N., Khaw, K.T., Day, N.E., Wareham, N.J.,
Bingham, S.A., 2005. Polymorphisms in the CYP19 gene may
affect the positive correlations between serum and urine
phytoestrogen metabolites and plasma androgen
concentrations in men. J. Nutr 135, 2680–2686.
Markiewicz, L., Garey, J., Adlercreutz, H., Gurpide, E., 1993. In vitro
bioassays of non-steroidal phytoestrogens. J. Steroid Biochem.
Mol. Biol. 45, 399–405.
Martins, C., Fillmann, G., Montone, R.C., 2007. Natural and
anthropogenic sterols inputs in surface sediments of Patos
Lagoon. Brazil. J. Braz. Chem. Soc. 18, 106–115.
Matthews, J., Celius, T., Halgren, R., Zacharewski, T., 2000.
Differential estrogen receptor binding of estrogenic
substances: a species comparison. J. Steroid Biochem. Mol.
Biol. 74, 223–234.
Mazur, W., Fotsis, T., Wahala, K., Ojalas, S., Salakka, A.,
Adlercreutz, H., 1996. Isotope dilution gas chromatographicmass
spectrometric method for the determination of
isoflavonoids, coumestrol, and lignans in food samples. Anal.
Biochem 233, 169–180.
Milder, I.E.J., Feskens, E.J.M., Arts, I.C.W., de Mesquita, H.B.B.,
Hollman, P.C.H., Kromhout, D., 2005. Intake of the plant
lignans secoisolariciresinol, matairesinol, lericiresinol, and
pinoresinol in Dutch men and women. J. Nutr 135, 1202–1207.
Mueller, S., Simon, S., Chae, K., Metzler, M., Korach, K., 2004.
Phytoestrogens and their human metabolites show distinct
agonistic and antagonistic properties on estrogen
receptora(ERa) and ER(b) in human cells. Toxicol. Sci. 80,
14–25.
Nagata, Y., Miyanaga, N., Okumura, K., Goto, K., Naito, S.,
Fujimoto, K., Hirao, Y., Takahashi, A., Tsukamoto, T.,
Akaza, H., 2007. Dietary isoflavones may protect against
prostate cancer in Japanese men. J. Nutr 137, 1974–1979.
Nakada, N., Nyunoya, H., Nakamura, M., Hara, A., Iguchi, T.,
Takada, H., 2004. Identification of estrogenic compounds in
wastewater effluent. Environ. Toxicol. Chem. 23, 2807–2815.
Nakada, N., Higashitani, T., Miyajima, K., Komori, K., Suzuki, Y.,
2006. Development of method for identification of major
substances inducing estrogenic activity contained in sewage
and river waters. J. Environ. Chem. 16, 389–401 (in Japanese
with English abstract).
Nelson, J., Bishay, F., van Roodselaar, A., Ikonomou, M., Law, F.C.
P., 2007. The use of in vitro bioassays to quantify endocrine
disrupting chemicals in municipal wastewater treatment
plants effluents. Sci. Total Environ 374, 80–90.
Nielsen, S.E., Freese, R., Kleemola, P., Mutanen, M., 2002.
Flavonoids in human urine as biomarkers for intake of fruits
and vegetables. Cancer Epidem. Biomarkers 11, 459–466.

Nishihara, T., Nishikawa, J.I., Kanayama, T., Dakeyama, F.,
Saito, K., Imagawa, M., Takatori, S., Kitagawa, Y., Hori, S.,
Utsumi, S., 2000. Estrogenic activities of 517 chemicals by
yeast two-hybrid assay. J. Health Sci. 46, 282–298.
NITE, 2008. http://www.safe.nite.go.jp/english/index.html.
2008.11.
Pawlowski, S., Ternes, T., Bonerz, M., Kluczka, T., van der Burg, B.,
Nau, H., Erdinger, L., Braunbeck, T., 2003. Combined in situ and
in vitro assessment of the estrogenic activity of sewage and
surface water samples. Toxicol. Sci. 75, 57–65.
Peeters, P.H.M., Keinan-Boker, L., van der Schouw, Y.T.,
Grobbee, D.E., 2003. Phytoestrogens and breast cancer risk.
Review of the epidemiological evidence. Breast Cancer Res.
Treat 77, 177–183.
Penalvo, J.L., Adlercreutz, H., Uehara, M., Ristimaki, A.,
Watanabe, S., 2008. Lignan content of selected foods from
Japan. J. Agric. Food Chem. 56, 401–407.
Phillips, K.M., Ruggio, D.M., Ashraf-Khorassan, M., 2005.
Phytosterol composition of nuts and seeds commonly
consumed in the United States. J. Agric. Food Chem. 53, 9436–
9445.
Quemeneur, M., Marty, Y., 1994. Fatty acids and sterols in
domestic wastewaters. Wat. Res. 28, 1217–1226.
Radke, M., 2005. Sterols and anionic surfactants in urban aerosol:
emissions from wastewater treatment plants in relation to
background concentrations. Environ. Sci. Technol 39, 4391–
4397.
Reddy, S., Iden, C.R., Brownawell, B., 2005. Analysis of steroid
conjugates in sewage influent and effluent by liquid
chromatography-tandem mass spectrometry. Anal. Chem. 77,
7032–7038.
Salste, L., Leskinen, P., Virta, M., Kronberg, L., 2007.
Determination of estrogens and estrogenic activity in
wastewater effluent by chemical analysis and bioluminescent
yeast assay. Sci. Total Environ 378, 343–351.
Seow, A., Shi, C.Y., Franke, A.A., Hankin, J.H., Lee, H.P., Yu, M.C.,
1998. Isoflavonoid levels in spot urine are associated with
frequency dietary soy intake in a population-based sample of
middle-aged and older Chinese in Singapore. Cancer Epidem.
Biomar 7, 135–140.
Setchell, K.D.R., Cassidy, A., 1999. Dietary isoflavones: biological
effects and relevance to human health. J. Nutr 129 (Suppl),
758–767.
Sivesind, E., Seguin, P., 2005. Effects of the environment, cultivar,
maturity, and preservation method on red clover isoflavone
concentration. J. Agric. Food Chem. 53, 6397–6402.
Smeds, A.I., Willfor, S.M., Pietarinen, S.P., Peltonen-Sainio, P.,
Reunanen, M.H.T., 2007. Occurrence of ‘‘mammalian’’ lignans
in plant and water sources. Planta 226, 639–646.
water research 44 (2010) 567–577 577
Stackelberg, P.E., Furlong, E.T., Meyer, M.T., Zaugg, S.D.,
Hendersen, A.K., Reissman, D.B., 2004. Persistence of
pharmaceutical compounds and other organic wastewater
contaminants in a conventional drinking-water-treatment
plant. Sci. Total Environ 329, 99–113.
Stackelberg, P.E., Gibs, J., Furlong, E.T., Meyer, M.T., Zaugg, S.D.,
Lippincott, R.L., 2007. Efficiency of conventional drinkingwater-treatment
processes in removal of pharmaceuticals and
other organic compounds. Sci. Total Environ 377, 255–272.
Song, T.T., Hendrich, S., Murphy, P.A., 1999. Estrogenic activity of
glycitein, a soy isoflavone. J. Agric. Food Chem. 47, 1607–1610.
Stopper, H., Schmitt, E., Kobras, K., 2005. Genotoxicity of
phytoestrogens. Mutat. Res. 574, 139–155.
Tan, B.L.L., 2006. Chemical and biological analyses of selected
endocrine disruptors in wastewater treatment plants in South
East Queensland, Australia. Doctorate dissertation. Griffith
University, Australia.
Tonkelaar, I.D., Keinan-Boker, L., Veer, P.V., Arts, C.J.M.,
Adlercreutz, H., Thijssen, J.H.H., Peeters, H.M., 2001. Urinary
phytoestrogens and postmenopausal breast cancer risk.
Cancer Epidem. Biomarkers 10, 223–228.
Tzchori, I., Degani, G., Elisha, R., Eliyahu, R., Hurvitz, A., Vaya, J.,
Moav, B., 2004. The influence of phytoestrogens and
oestradiol-17b on growth and sex determination in the
European eel (Anguilla anguilla). Aquat. Res. 35, 1213–1219.
USDA (US Department of Agriculture), 2003. USDA database for
the flavonoid content of selected foods. http: www.nal.usda.
gov/fnic/foodcomp.
Valentin-Blasini, L., Sadowski, M.A., Walden, D., Caltabiano, L.,
Needham, L.L., Barr, D.B., 2005. Urinary phytoestrogen
concentrations in the U.S. population (1999–2000). J. Expo.
Anal. Environ. Epidemiol 15, 509–523.
Wang, L.Q., 2002. Mammalian phytoestrogens: enterodiol and
enterolactone. J. Chromatogr. B 777, 289–309.
Whitten, P.L., Patisaul, H.B., 2001. Cross-species and interassay
comparisons of phytoestrogen action. Environ. Health Persp
109 (Suppl), 5–20.
Xu, X., Duncan, A.M., Wangen, K.E., Kurzer, M.S., 2000. Soy
consumption alters endogenous estrogen metabolism in
postmenopausal women. Cancer Epidem. Biomarkers 9, 781–786.
Zhang, Y., Wang, G.J., Song, T.T., Murphy, P.A., Hendrich, S., 1999.
Urinary disposition of the soybean isoflavones daidzein,
genistein and glycitein differs among humans with moderate
fecal isoflavone degradation activity. J. Nutr 129, 957–962.
Zhang, X., Cambrai, A., Miesch, M., Roussi, S., Raul, F., Aoude-
Werner, D., Marchioni, E., 2006. Separation of D5-and D7phytoesterols
by adsorption chromatography liquid
chromatography for quantitative analysis of phytosterols in
foods. J. Agric. Food Chem. 54, 1196–1202.

Occurrence of emerging pollutants in urban wastewater
and their removal through biological treatment followed
by ozonation
Roberto Rosal a, *, Antonio Rodríguez a , José Antonio Perdigón-Melón a , Alice Petre a ,
Eloy García-Calvo a , María José Gómez b , Ana Agüera b , Amadeo R. Fernández-Alba b
a
Department of Chemical Engineering, University of Alcalá, 28771 Alcalá de Henares, Spain
b
Department of Analytical Chemistry, University of Almería, 04010 Almería, Spain
article info
Article history:
Received 24 February 2009
Received in revised form
26 June 2009
Accepted 3 July 2009
Available online 9 July 2009
Keywords:
Emerging pollutants
Pharmaceuticals
Personal care products
Wastewater treatment
Ozonation
1. Introduction
abstract
The presence of a wide variety of pharmaceutical and
personal care products (PPCP) in water and wastewater has
been frequently reported after the findings of Ternes (1998)
and Daughton and Ternes (1999). These compounds are
a source of concern because they are used and released in
large quantities and their physical and chemical properties
water research 44 (2010) 578–588
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
This work reports a systematic survey of over seventy individual pollutants in a Sewage
Treatment Plant (STP) receiving urban wastewater. The compounds include mainly pharmaceuticals
and personal care products, as well as some metabolites. The quantification in
the ng/L range was performed by Liquid Chromatography–QTRAP–Mass Spectrometry and
Gas Chromatography coupled to Mass Spectrometry. The results showed that paraxanthine,
caffeine and acetaminophen were the main individual pollutants usually found
in concentrations over 20 ppb. N-formyl-4-amino-antipiryne and galaxolide were also
detected in the ppb level. A group of compounds including the beta-blockers atenolol,
metoprolol and propanolol; the lipid regulators bezafibrate and fenofibric acid; the antibiotics
erythromycin, sulfamethoxazole and trimethoprim, the antiinflammatories diclofenac,
indomethacin, ketoprofen and mefenamic acid, the antiepileptic carbamazepine
and the antiacid omeprazole exhibited removal efficiencies below 20% in the STP treatment.
Ozonation with doses lower than 90 mM allowed the removal of many individual
pollutants including some of those more refractory to biological treatment. A kinetic model
allowed the determination of second order kinetic constants for the ozonation of bezafibrate,
cotinine, diuron and metronidazole. The results show that the hydroxyl radical
reaction was the major pathway for the oxidative transformation of these compounds.
ª 2009 Elsevier Ltd. All rights reserved.
contribute to their widespread distribution into the environment.
The presence of small concentration of PPCP has been
associated to chronic toxicity, endocrine disruption and the
development of pathogen resistance. The consequences are
particularly worrying in aquatic organisms as they are subjected
to multigenerational exposure (Halling-Sørensen et al.,
1998). The presence of micropollutants also endangers the
reuse of treated wastewater, a generally proposed solution to
* Corresponding author. Department of Chemical Engineering, Facultad de Quamica, University of Alcalá, 28771 Alcalá de Henares,
Madrid, Spain. Tel.: þ34 91 885 5099; fax: þ34 91 885 5088.
E-mail address: roberto.rosal@uah.es (R. Rosal).
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.watres.2009.07.004

Notation
Ci concentration of a given organic compound (M)
CHO$ concentration of hydroxyl radical (M)
CO3 concentration of dissolved ozone (M)
CO3 equilibrium concentration of dissolved
ozone (M)
DO3 diffusivity of ozone in water, m 2 s 1
Dow apparent octanol–water partition coefficient
(dimensionless)
fOH fraction of a given compound degraded by
hydroxyl radicals
Ha Hatta number, defined in Eq. (5) (dimensionless)
achieve a sustainable water cycle management (Muñoz et al.,
2009). PPCP represent a rising part of the trace organic
micropollutants found in urban and domestic wastewaters
that reach sewage treatment plants (STP), either metabolised
or not (Castiglioni et al., 2006). Many of these substances
escape to conventional activated sludge wastewater treatments
allowing them to reach surface water streams and
distribute in the environment (Tauxe-Wuersch et al., 2005).
The need for treatment technologies that can provide safe
treated effluents led to the proposals for upgrading STP and to
implement new competing technologies for biological degradation
of organic matter like membrane bioreactor (Radjenovic
et al., 2009). In addition to these strategies, effective
tertiary treatment technologies are also needed in order to
ensure a safe use for reclaimed wastewater. The available
technologies include oxidation processes alone or combined
with nanofiltration or reverse osmosis (Ernst and Jekel, 1999).
Many oxidation processes have been described for the
removal of organic compounds in wastewater. Ozone-based
and Advanced Oxidation Processes (AOP) using hydrogen
peroxide or radiation have been repeatedly proposed for this
task (Gogate and Pandit, 2004a; Ikehata et al., 2006). Proper
combinations of AOP can also be considered in order to treat
the more refractory pollutants. Fenton and Fenton-based
systems, heterogeneous photocatalysis, and ultraviolet or
ozone-based oxidation processes have been described (Gogate
and Pandit, 2004b; Comninellis et al., 2008). The choice of the
most suitable technology or combination lies on the quality
required for the reclaimed water.
Ozone has been largely used as oxidant in drinking water
treatment and repeatedly proposed to remove organics in
wastewater treatment (Raknes, 2005; Beltrán, 2004). The ozone
molecule can react with many organic compounds, particularly
those unsaturated or containing aromatic rings or
heteroatoms also being able to decompose in water to form
hydroxyl radicals. In a previous work (Rosal et al., 2008a) we
studied the ozonation of wastewater from the secondary
clarifier of urban and domestic STP by using ozone (pH w 8) and
ozone–hydrogen peroxide (O3/H2O2). The presence of hydrogen
peroxide improved the mineralization of dissolved carbon
from 15% to over 90% after one hour, from which most part took
place during the first five minutes. The disappearance of
a selected group of over thirty pharmaceuticals revealed
removal efficiencies were over >99% for most compounds after
water research 44 (2010) 578–588 579
kL liquid-phase individual mass transfer
coefficient, m s –1
kO3 second order ozone-based rate constant for
a direct ozonation reaction (M 1 s 1 )
kHO$ second order rate constant for a reaction with
hydroxyl radical (M 1 s 1 )
kR second order rate constant for the ozonation of
a given compound (M 1 s 1 )
kLa volumetric mass transfer coefficient (s 1 )
Kow octanol–water partition coefficient for neutral
species (dimensionless)
TOD dose of ozone transferred to the liquid (M)
five minutes on stream, with slightly better results in the
absence of hydrogen peroxide. This result is consistent with
the fact that many PPCP directly react with ozone with large
second order kinetic constants (Huber et al., 2003).
The objective of this research was to verify the occurrence
and fate of 84 pollutants of different classes, mainly PPCP
traced during wastewater treatment in a conventional urban
STP. These include: pharmaceuticals (analgesics, antidepressants,
antiinflammatories, antibiotics, antiepileptics, betablockers
and lipid regulators among others), personal care
products (sunscreen agents, synthetic musks), stimulants
(caffeine, nicotine) and some metabolites (clofibric acid, cotinine,
several metabolites of dipyrone). The effectiveness of
the STP process for the removal of these compounds has been
assessed in a monitoring program undertaken in Alcalá de
Henares (Madrid) during a one-year period with samples
taken before and after a biological activated sludge with
nutrient removal process. The research was also trying to
identify the impact of ozone exposure on individual pollutants
encountered in the secondary effluent. The doses of ozone
required for a given degree of removal of the most representative
individual pollutants has been assessed. For certain
pollutants, the determination of second order kinetic
constants for the ozonation reaction in a real wastewater
matrix was also performed.
2. Material and methods
2.1. Materials and plant description
Wastewater samples were taken every month over a one-year
period from the input and output of the secondary clarifier of
a STP located in Alcalá de Henares (Madrid). This plant treats
a mixture of domestic and industrial wastewater from some
facilities located near the city with a nominal capacity of
3000 m 3 /h of raw wastewater. The Council Directive 91/271/
EEC concerning urban wastewater treatment established
strict limits for the wastewater discharged to sensitive areas
particularly by plants serving an equivalent population of
10 000 or more. In the period prior to the sampling campaign,
the need to adapt STP to the new conditions forced, the
implementation of a biological nutrient removal process
aimed to the elimination of phosphorous and nitrogen. The

580
biological treatment worked under a traditional A2O multistage
configuration with nitrification–denitrification and
enhanced phosphorus removal by phosphorus-accumulating
microorganisms. The treatment takes place in three zones:
anaerobic, anoxic and oxic. The nitrate produced in the oxic
zone is recycled, with mixed liquor to the anoxic zone where
denitrification takes place. The return sludge from the settler
is recycled to the anaerobic zone where the influent and
sludge are mixed under anaerobic conditions. All samples
were immediately processed or stored in a refrigerator (

2.5. Ozonation
The ozonation runs were carried out in a 5-L glass jacketed
reactor operating in semi-batch mode. A temperature of 25 C,
chosen to be close to average ambient temperature, was kept
using a Huber Polystat cc2 and monitored throughout the runs
bymeansofaPt100ResistanceTemperatureDetector(RTD).
Ozone was produced by a corona discharge ozonator (Ozomatic,
119 SWO100) fed by an AirSep AS-12 PSA oxygen
generation unit. The gas containing about 9.7 g/Nm 3 ozone
was bubbled by means of a porous glass disk with a gas flow of
0.36 Nm 3 /h. The reaction vessel was agitated with a Teflon
four-blade impeller at 1000 rpm. The mass transfer coefficient
was determined in transient runs with pure water with
a value of kLa ¼ 0.010 0.005 s 1 . Additional details on the
experimental set-up and procedure are given elsewhere
(Rosal et al., 2008a,b). Throughout the runs, certain samples
were withdrawn for analysis at prescribed intervals. Residual
ozone was removed by bubbling nitrogen in order to prevent
oxidation reactions to continue. For the desorption conditions
used, residual ozone fell down below 10% in less than 20 s.
During the ozonation reactions a moderate increase of pH
took place that can be attributed to the stripping of CO2 from
solution. In conditions that favour the generation of hydroxyl
radicals, this effect is not observed and pH tend to decrease
during ozonation due to the accumulation of carboxylic acids.
The choice of a pH higher than that of the raw wastewater not
only favours hydroxyl radical mediated reactions, but allowed
to keep an almost constant pH value of 8.5 0.1 by using the
feed-back control procedure described above.
3. Results and discussion
3.1. Occurrence
The pollutants analyzed were mainly PPCP and their metabolites
together with some agrochemicals. A list of the
72 anthropogenic emerging pollutants detected in at least one
wastewater sample from the influent to the biological treatment
is detailed in the Appendix together with the analytical
method applied, the molecular formula and the octanol–water
partition coefficient, when available. The limit of quantification
for the biological effluent (LOQ), for most compounds in
the tens of ng/L, is also shown in the Appendix that also lists
the compounds checked but not detected with their limits of
detection (LOD). Several pesticides like chlorfenvinphos and
the herbicide isoproturon were never detected, as expected
considering the urban origin of wastewater. Atrazine,
however, was found in all samples with an average concentration
at the inlet of the biological treatment of 109 ng/L.
Diuron was encountered in two samples and simazine in
three with maximum concentrations of 196 and 32 ng/L,
respectively. Some drugs like paroxetine, the antibiotic cefotaxime
and the antiinflammatory fenoprofen were not
detected in any of the analyzed samples. A similar negative
result was reported for paroxetine by Terzić et al. (2008),
whereas the occurrence of fenoprofen was reported in
concentrations as high as 0.759 mg/L in Canadian wastewater
facilities (Metcalfe et al., 2004).
water research 44 (2010) 578–588 581
The detailed data for the concentrations of the most significant
individual pollutants are shown in Table 2. Itincludes
maximum and minimum values of those compounds encountered
over their quantification limit in at least 4 samples in the
influent of the biological treatment. The compounds excluded
for not complying with this criterion comprise, among other,
carbamazepine-10,11-epoxide that was detected in 2 samples
with a maximum concentration of 63 ng/L, sotalol, detected
three times in the influent with concentrations in the 25–30 ng/L
range or celestolide with a maximum influent concentration of
30 ng/L. In addition, fenofibrate was encountered in one sample
with 101 ng/L, whereas its human metabolite, fenofibric acid
has been detected in all samples at concentrations as high as
117 ng/L. Table 2 also excludes some compounds not systematically
analyzed during the sampling campaign due to
improvements in the analytical procedure. Certain compounds
such as terbutaline or simazine have been encountered in
several samples, but these data have not been considered
statistically significant because they were found at concentrations
so close to the LOQ that removal efficiency in STP could not
be properly assessed and, therefore, they were not included in
Table 2. Among them, diazepam was detected in half of the
samples with a maximum concentration of 8 ng/L in the
influent and 5 ng/L in the effluent, the average in both cases
being near 3 ng/L (that equals LOQ). Also, mepivicaine was
found in 7 samples with an average concentration of 8 ng/L in
the influent and 7 ng/L in the effluent and maximum concentration
in the influent of 14 ng/L.
Paraxanthine, caffeine and acetaminophen were the individual
pollutants usually found in higher concentration, with
averages near 20 ppb in the influent. N-formyl-4-amino-antipiryne
(4-FAA) and galaxolide also exhibited averages in the
mg/L level. The fact that caffeine is the dominant micropollutant
in STP is not new. Caffeine has been detected in
many surface streams and STP effluents in concentration as
high as 230 mg/L (Ternes et al., 2001; Heberer et al., 2002).
Recently, Wilcox et al. (2009) also report caffeine, paraxanthine
and acetaminophen as the most frequently detected
compounds, in the influent of conventional septic systems.
Muñoz et al. (2009) found several substances at mg/L level, with
the highest concentrations corresponding to caffeine, acetaminophen,
atenolol, and paraxanthine, that exceeded 40 mg/L
each. In this work, atenolol was part of the group of
compounds found with averages over the ppb limit that
included ciprofloxacin, hydrochlorthiazide, ibuprofen,
N-acetyl-4-amino-antipiryne (4-AAA), naproxen, nicotine,
and ofloxacin. The occurrence of dipyrone (metamizol) residues
in STP effluents has been less frequently assessed, but
Feldmann et al. (2008) have recently reported concentrations
up to 7 mgL 1 in influents and effluents of STP in Berlin. These
dipyrone residues have been attributed to effluents originated
in hospitals more than to private households. Nicotine was
found in concentrations as high as 12 mg/L in the influent
and 148 ng/L in the effluent. Cotinine, its major urinary
metabolite, was detected in the biological effluent with
a concentration of 100 ng/L. These values agree with those
published by Buerge et al. (2008) who found concentrations of
cotinine of approximately 1–10 mg/L in untreated wastewater,
and 0.01–0.6 mg/L in the treated effluent, corresponding to
elimination efficiencies of over 90%.

582
Table 2 – Concentrations of pollutants in the influent and effluent of the studied STP calculated for compounds detected
over LOQ in at least four influent samples along the monitoring programme. Averages and removal efficiencies have been
calculated excluding concentrations below LOQ.
Compound pKa a
Influent (ng/L) Effluent (ng/L) Removal
Maximum Minimum Average Maximum Minimum Average
efficiency (%)
4-aminoantipyrine (4-AA) 4.3 3325 262 1517 2253 127 676 55.4
4-methylaminoantipyrine (4-MAA) 4.3 1894 314 880 1098 34 291 66.9
Acetaminophen 9.4 37 458 1571 23 202

elow LOQ. It is well known that during wastewater treatment
in STP, PPCP as well as their metabolites partition into the
solid phase or remain dissolved depending on their hydrophobicity.
The hydrophobicity of a neutral compound can be
expressed as its octanol–water partition coefficient, Kow, but in
the case of compounds that can exist in ionized form, the
acid–base equilibrium must also be taken into account. The
pH-dependent or apparent octanol–water distribution coefficient,
D ow, that considers both the dissociation constant of
acidic of basic solutes, pK a, and the current pH of wastewater
can be derived from the Herderson–Hasselbalch equations
(Scheytt et al., 2005). For acidic compounds, like indomethacin
or naproxen, that are dissociated at the pH of wastewater, the
equation yields:
Kow
Dow ¼
1 þ 10pH pKa
In the case of basic drugs, the apparent partition coefficient
can be expressed using the pKa for the corresponding conjugate
acid:
Kow
Dow ¼
1 þ 10pKa pH (2)
Codeine or fluoxethine are among compounds that are
substantially dissociated at ambient pH, whose average was
7.61 0.39, with boundaries representing the 95% confidence
interval. For neutral substances, D ow ¼ K ow.
There is experimental evidence that the removal of organic
pollutants in STP is largely controlled by sorption process with
solid–water distribution coefficients being a function of the
octanol–water distribution coefficient Dow (Carballa et al.,
Removal efficiency (%)
100
90
80
70
60
50
40
30
water research 44 (2010) 578–588 583
6
(1)
1 2
2008). The removal efficiency obtained in this work for the
compounds indicated in Table 2 has been related to Dow and
the results represented in Fig. 1. For a significant group of
compounds, ranging from ketorolac (44%) to galaxolide (88%),
a clear relationship is observed between removal efficiency
and Dow. A group of five compounds were almost completely
removed in the STP even they present relatively low Dow
values. They are the metabolite of caffeine paraxanthine,
caffeine itself, acetaminophen (paracetamol), ibuprofen and
nicotine, compounds otherwise usually found in high
concentrations in raw urban wastewater (Table 2). A similar
result was reported by Muñoz et al. (2008) who reported
concentrations of caffeine, acetaminophen, atenolol, and
paraxanthine that exceeded 40 mg/L each but a substantial
removal in an activated sludge STP (90%, >99%, 43%, and 67%,
respectively). Joss et al. (2005) reported removal efficiencies for
ibuprofen beyond its quantification limit (>90%) and also
coincident are data corresponding to naproxen (50–80%).
This work also identified 14 compounds with removal
efficiencies fell below 20%, but exhibiting intermediate D ow
values ( 2 < D ow < 2.5). They have been marked by the lower
square in Fig. 1 and are the beta-blockers atenolol, metoprolol
and propanolol; the lipid regulator bezafibrate and the
metabolite of fenofibrate fenofibric acid; the antibiotics
erythromycin, sulfamethoxazole and trimethoprim; the antiinflammatories
diclofenac, indomethacin, ketoprofen and
mefenamic acid; the antiepileptic carbamazepine and the
antiacid omeprazole. Joss et al. (2005) reported no significant
removal of sulfamethoxazole and carbamazepine and partial
elimination of diclofenac, results generally coincident with
this work. For musk fragrances, galaxolide and tonalide, Joss
3 4
5
22
17
11
18
19
8
7
10
12
9 14
21
16 15
20
28
10
0
-8.00 -6.00 -4.00
27 32
35
31
39
30 34
29
40
36 37
-2.00 0.00 2.00 4.00 6.00 8.00
log D ow
Fig. 1 – Removal efficiency during conventional activated sludge treatment: (1) paraxanthine, (2) caffeine,
(3) acetaminophen,(4) nicotine, (5) ibuprofen,(6) ketorolac,(7) clofibric acid,(8) furosemide, (9) ciprofloxacin,(10) fluoxethine,
(11) ofloxacin,(12) naproxen,(13) hydrochlorothiazide, (14) 4-aminoantipyrine, (15) metronidazole, (16) N-acetyl-4-aminoantipiryne,
(17) codeine, (18) N-formyl-4-amino-antipiryne, (19) 4-methylaminoantipyrine, (20) ranitidine, (21) antipyrine,
(22) gemfibrozil, (23) benzophenone-3,(24) triclosan,(25) tonalide, (26) galaxolide, (27) atenolol, (28) sulfamethoxazole,
(29) fenofibric acid, (30) metoprolol, (31) bezafibrate, (32) ketoprofen, (33) trimethoprim, (34) diclofenac, (35) indomethacine,
(36) propanolol, (37) mefenamic acid, (38) omeprazole, (39) carbamazepine, (40) erythromycin.
23
24
26
25

584
et al. (2005) reported, however, removal efficiencies somewhat
lower (>50%) than those found in this work (w85%). Carballa
et al. (2004) also reported overall removal efficiencies of the
ranging between 70% and 90% for fragrances in coincidence
with this work and generally higher for antiinflammatories
(40–60%) and sulfamethoxazole. In a further work, Carballa
et al. (2007) indicated different removal efficiencies for several
PPCP during the anaerobic digestion of sewage sludge with
high efficiencies for naproxen or sulfamethoxazole. In
general, a high variability can be expected as a consequence of
different pollutant concentrations, changes in the processes
of the STP as well as operational conditions (Liu et al., 2008).
3.3. Removal by ozonation
The ozonation of a dissolved organic compound may take
place by direct reaction with ozone molecule and also through
the action of secondary oxidants produced from ozone in
aqueous medium. Among them, the strongest oxidant is
hydroxyl radical, generally associated with the oxidation of
the more refractory substances. A mass balance to a given
compound in solution yields the following expression:
dCi
dt ¼ kO 3 CiCO 3 þ kHO$CiCHO$
Elovitz and von Gunten (1999) introduced a parameter R ct
defined as the relationship between the concentrations of
ozone and hydroxyl radical at any moment. Based on the
observation that R ct can be considered constant at least
though certain periods of the ozonation runs, the concentration
of the two main oxidants involved in ozonation reactions
can be related so that Eq. (3) can be solved without experimental
determination of CHO$:
ln Ci
Z t
Z t
¼ kO CO dt þ kHO$
3
3
Cio
0
0
Z t
¼ kR
0
CO dt 3
CHO$dt ¼ kO 3 þ RctkHO$
Z t
CO dt 3
0
The kinetic parameter kR would behave as second order
kinetic constant provided Rct is constant throughout the
sampling period. As shown below, this work shows that Rct is
constant during ozonation at least for the samples taken after
ozone appeared in solution. The preceding findings also lie on
the assumption of slow kinetic regime. The kinetics of
a heterogeneous gas–liquid semicontinuous process is determined
by the relative rates of absorption and chemical reaction
and is characterized by Hatta number. It represents the
maximum rate of chemical reaction relative to the maximum
rate of mass transfer, yielding, for a second order reaction, the
following expression:
Ha ¼
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
kO Ci;oDO 3 3
kL
In the case of wastewater treatment, ozone reacts with
many compounds in a complex parallel reaction system so
that kO Ci;o can be substituted by 3 P
kO Ci;o. As raw wastewater
3
i
contains a number of compounds whose second order direct
reaction constants with ozone are very large (Huber et al.,
water research 44 (2010) 578–588
(3)
(4)
(5)
2003) mass transfer is likely to limit the ozonation rate during
the first reaction minutes. Once ozone appears in solution, the
following inequality holds:
kLa C O3
CO 3
> X
i
kO 3 Ci;oCO 3
Therefore an upper limit can be obtained for Ha as follows:
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Ha <
kLa C O 3
C O3
kL
1 DO 3
where DO 3 is the diffusivity of ozone in water
(1.77 10 9 m 2 s 1 ). The equilibrium concentration of ozone
C O3 was calculated from Henry’s law using the correlation of
Rischbieter et al. (2000) from the concentration in the gas phase
that was 9.4 g/Nm 3 . The value of the mass transfer coefficient,
kL ¼ 5.5 10 5 ms 1 , was calculated by using the correlation of
Calderbank and Moo-Young (1961). At the beginning of the run,
there was no ozone in solution and Ha is supposed to reach
high values. After about three minutes, the concentration of
ozone that increased during runs approaching equilibrium was
over 2 10 3 mM. This, considering that C O3 ¼ 0:05 mM,
ensures Ha < 0.3 and, therefore, the kinetic regime was slow and
for the sample taken at 4 min and those taken thereafter.
Second order kinetic constants could be obtained as indicated
before for bezafibrate, cotinine, diuron, ketoprofen and
metronidazole. The logarithmic concentration decay is represented
against the integral ozone dose in Fig. 2 in which, the
experimental points represent data from samples taken at 4, 6,
10 and 15 min, the first being considered the initial concentration,
Cio. The comparison with literature data shows that the
transformation of ozone-resistant micropollutants takes place
primarily via indirect radical oxidation inhibited by the wastewater
matrix. The values reported in the literature for
the second order direct and indirect rate constants are:
bezafibrate, kO3 ¼ 590 M 1 s 1 and kHO$ ¼ 7:40 10 9 M 1 s 1
in Huber et al. (2005); ketoprofen, kO3 ¼ 0:4 M 1 s 1 and
kHO$ ¼ 8:40 10 9 M 1 s 1 in Real et al. (2009), diuron,
ln(C io/C i)
4
3
2
1
0
0 0.2 0.4 0.6 0.8 1
Fig. 2 – Logarithmic decay of the concentration of diuron
(B), metronidazole (C), ketoprofen (D) bezafibrate (,) and
cotinine (-) as a function of the integral ozone exposure.
(Two y-axes have been used for clarity.)
6
5
4
3
2
1
0
(6)
(7)
ln(C io/C i)

kO3 ¼ 16:5 M 1 s 1 and kHO$ ¼ 4:6 109 M 1 s 1 in Benitez et al.
(2007) and kO3 ¼ 14:7 M 1 s 1 and kHO$ ¼ 6:6 109 M 1 s 1 in
De Laat et al. (1996) ; metronidazole, kO3 ¼ < 350 M 1 s 1
in Sánchez-Polo et al. (2008) and kHO$ ¼ 1:98 109 M 1 s 1 in
Johnson and Mehrvar (2008). No data have been published
for cotinine. The ozonation of bezafibrate was also studied
by Dantas et al. (2007) who reported a direct kinetic constant
kO3 ¼ 4:24 103 0:66 103 M 1 s 1 for pH 7 and kR ¼ 1.0
10 4
1.07 10 3 M 1 s 1 for pH 8. The reported direct rate
constant for ketoprofen is particularly low and allows to calculate
Rct by dividing the value of kR obtained in this work by kHO$
reportedby Real et al. (2009). The value found, Rct ¼ 3.62 10 7 ,is
very close to those that could be derived using Eq. (4) for bezafibrate
and diuron. The partial contribution of the direct ozone
and radical pathways can be calculated for a given organic
compound as follows:
kHO$CHO$Ci
fOH ¼
kHO$CHO$Ci þ kO Ci 3
¼
kHO$Rct
kHO$Rct þ kO 3
Using literature constants, the fraction degraded by hydroxyl
radicals were 0.819 for bezafibrate, 0.996 for diuron, 1.000 for
ketoprofen and >0.887 for metronidazole. The contribution of
direct ozone reaction is obviously low because the five
compounds studied are relatively refractory to ozonation. By
means of Eq. (8) and taking into account that kR ¼ kO þ kHO$$Rct,
3
the second order rate constants, kR, can be derived from literature
data with the following results: metronidazole
1.07 10 3 M 1 s 1 , bezafibrate 3.27 10 3 M 1 s 1 and diuron
2.41 10 3 M 1 s 1 (Benitez et al., 2007). The two last are in good
agreement with experimental values. In the case of metronidazole,
the low value obtained for Rct suggests that kHO could
have been underestimated.
The efficiency of ozonation for the removal of the main
micropollutants whose concentration in biologically treated
wastewater was >10 ng/L is indicated in Table 3. The table
shows the evolution of the concentration for samples taken
during ozonation up to a reaction time of 15 min and the
amount of ozone required for a given degree of removal. Only
concentrations above LOQ are shown. These LOQ apply for
ozonated samples and are lower than those reported in the
Appendix for untreated samples. They could be reached,
because of the higher preconcentration factor applied to the
ozonated samples (see experimental section) and the reduction
in the matrix effects observed in these cleaner extracts.
The amount of ozone transferred to the liquid at a certain
reaction time, TOD, was determined, from the integration of
the ozone absorption rate equation:
0
1
Z t
TODðtÞ ¼ kLa@C
t CO dtA
(9)
O3 3
0
For the experimental conditions used in this work CO 3 was
always less then 10% of C O3 and, therefore, TOD(t) was
essentially linear with time. The last right column of Table 3
shows the dose of ozone required for the complete removal
(no detection) of a given compound or to achieve certain
removal efficiency in cases where ozonation was unable to get
complete oxidation in less than 15 min corresponding to
a dose of ozone of 0.34 mmol/L of wastewater. It is to be noted
that a change in the pH of wastewater modifies the ozone
water research 44 (2010) 578–588 585
(8)
doses required for a given effect. A pH increase causes
a higher hydroxyl radical exposure increasing Rct and leading,
for a similar integral radical exposure, to lower ozone doses.
Also, it is important to point out that a change of pH may
affect the direct ozonation rates of dissociating compounds. In
our case, raising pH from that of raw wastewater to 8.5, at
which the ozonation took place, could affect the ozonation
rate of some proton-accepting compounds whose pKa w 8,
particularly benzophenone-3, hydrochlorthiazide, nicotine,
ofloxacin and triclosan.
A group of 15 compounds rapidly disappear during the first
120 s on stream, with ozone doses

586
Table 3 – Removal of pollutants contained in wastewater during ozonation. The ozone doses are those required to reach
concentrations below the limit of quantification (LOQ a ) in treated samples.
Ozonation time (min) LOQ a
0 2 4 6 10 15 Ozone doses
for remotion
3-(4-methylbenzylidene) camphor 39 55 50 65 39 72 54 Not removed
4-aminoantipyrine 19 58 – – – – –

4. Conclusions
This research showed the regular presence of over seventy
anthropogenic individual pollutants, some of which are
encountered in relatively high amounts. In raw sewage,
25 compounds were detected in the mg/L range, 15 of which
exceeded this level in yearly averages. Acetaminophen (paracetamol)
and caffeine were persistently detected over 1 ppb
in untreated wastewater. Galaxolide was also encountered in
high concentration, but its occurrence showed a significant
variability. Two metabolites, paraxanthine from caffeine and
4-FAA from metamizol were also found in high amounts in
almost all samples. Other metabolites detected were 4-AA and
4-MAA from dipyrone, fenofibric acid from fenofibrate and
4-AAA also from metamizol. This findings stress the need for
exploring not only the pattern PPCP, but also their metabolic
or photodegradation intermediates.
The efficiency of removal of PPCP in STP was roughly
dependent on its hydrophobicity expressed as apparent
octanol–water distribution coefficient, D ow, a parameter that
takes into account the octanol–water partition coefficient,
Kow, as well as the dissociation constant of acidic or basic
compounds. For most compounds, the removal efficiency
during biological treatment increased with hydrophobicity as
expected considering the higher sorption of non-polar
compounds on sludge. Certain compounds showed important
deviations with a group of relatively polar substances formed
by paraxanthine, caffeine, acetaminophen, ibuprofen and
nicotine that were almost completely removed during biological
treatment. Another group formed by 14 compounds
exhibited removal efficiencies below 20%, even their octanol–
water distribution coefficient was not particularly low. They
are all important pharmaceuticals prescribed and delivered to
sewage in high amounts and include the beta-blockers atenolol,
metoprolol and propanolol; the lipid regulator bezafibrate,
fenofibric acid; the antibiotics erythromycin,
sulfamethoxazole and trimethoprim; the antiinflammatories
diclofenac, indomethacin, ketoprofen and mefenamic acid;
the antiepileptic carbamazepine and the antiacid omeprazole.
An ozonation treatment yielded high removal efficiencies
of most individual pollutants detected in treated wastewater.
The kinetic analysis of the part of the run that takes place in
the slow kinetic regime, allowed the determination of second
order kinetic constants for the ozonation of bezafibrate, cotinine,
diuron, ketoprofen and metronidazole in its wastewater
matrix. This work shows that Rct, a concept developed for
drinking water, can be applied for the ozonation of wastewater
as it was constant at least for the samples taken after
ozone appeared in solution. The ratio of the concentrations of
ozone and hydroxyl radical, Rct, during this period was
3.62 10 7 and the fraction degraded by hydroxyl radicals was
in the 0.819–1.000 range for the aforementioned compounds.
Even when most compounds disappear for doses lower than
340 mM, and most for less than 90 mM, some pollutants,
essentially personal care products were not significantly
removed during ozonation. These were the UV filters
3-(4-methylbenzylidene) camphor and Ethylkexyl methoxycinnamate,
the sunscreen agent Benzophenone-3 and the
aromatic nitro musk xylene. On the other hand, certain polar
water research 44 (2010) 578–588 587
compounds like diclofenac, indomethacin or the betablockers
atenolol, metoprolol and propanolol, which are
poorly removed in the activated sludge conventional treatment,
exhibit large ozonation rates and can be removed from
treated wastewater using moderate ozone doses.
Acknowledgements
The authors wish to express their gratitude to the Ministry of
Education of Spain (Contract CONSOLIDER-INGENIO 2010
CSD2006-00044), the Dirección General de Universidades e
Investigación de la Comunidad de Madrid, Research network
0505/AMB-0395.
Appendix.
Supplementary information
Supplementary data associated with this article can be found
in the doi: 10.1016/j.watres.2009.07.004.
references
Beltrán, F.J., 2004. Ozone Reaction Kinetics for Water and
Wastewater Systems. CRC, Boca Raton, pp. 113–149.
Benitez, F.J., Real, F.J., Acero, J.L., Garcia, C., 2007. Kinetics of the
transformation of phenyl-urea herbicides during ozonation of
natural waters: rate constants. Water Res. 41, 4073–4084.
Bueno, M.J.M., Agüera, A., Gómez, M.J., Hernando, M.D.,
García, J.F., Fernández-Alba, A.R., 2007. Application of liquid
chromatography/quadrupole-linear ion trap mass
spectrometry and time-of-flight mass spectrometry to the
determination of pharmaceuticals and related contaminants
in wastewater. Anal. Chem. 79, 9372–9384.
Buerge, I.J., Kahle, M., Buser, H.R., Müller, M.D., Poiger, T., 2008.
Nicotine derivatives in wastewater and surface waters:
application as chemical markers for domestic wastewater.
Environ. Sci. Technol. 42, 6354–6360.
Calderbank, P.H., Moo-Young, M.B., 1961. The continuous phase
heat and mass transfer properties of dispersions. Chem. Eng.
Sci. 16, 39–54.
Carballa, M., Fink, G., Omil, F., Lema, J.M., Ternes, T., 2008.
Determination of the solid–water distribution coefficient (K d)
for pharmaceuticals, estrogens and musk fragrances in
digested sludge. Water Res. 42, 287–295.
Carballa, M., Omil, F., Lema, J.M., Llompart, M., García-Jares, C.,
Rodríguez, I., Gómez, M., Ternes, T., 2004. Behavior of
pharmaceuticals, cosmetics and hormones in a sewage
treatment plant. Water Res. 38, 2918–2926.
Carballa, M., Omil, F., Ternes, T., Lema, J.M., 2007. Fate of
pharmaceutical and personal care products (PPCPs) during
anaerobic digestion of sewage sludge. Water Res. 41,
2139–2150.
Castiglioni, S., Bagnati, R., Fanelli, R., Pomati, F., Calamari, D.,
Zuccato, E., 2006. Removal of pharmaceuticals in sewage
treatment plants in Italy. Environ. Sci. Technol. 40, 357–363.
Comninellis, C., Kapalka, A., Malato, S., Parsons, S.A., Poulios, I.,
Mantzavinos, D., 2008. Advanced oxidation processes for
water treatment: advances and trends for R&D. J. Chem.
Technol. Biotechnol. 83, 769–776.

588
Dantas, R.F., Centerino, M., Marotta, R., Sans, C., Esplugas, S.,
Andreozzi, A., 2007. Bezafibrate removal by means of
ozonation: primary intermediates, kinetics, and toxicity
assessment. Water Res. 41, 2525–2532.
Daughton, C.G., Ternes, T.A., 1999. Pharmaceuticals and personal
care products in the environment: agents of subtle change?
Environ. Health Perspect. 107, 907–938.
De Laat, J., Maouala-Makata, P., Dore, M., 1996. Rate constants for
reactions of ozone and hydroxyl radicals with several phenylureas
and acetamides. Environ. Technol. 17, 707–716.
Elovitz, M.S., von Gunten, U., 1999. Hydroxyl radical/ozone ratios
during ozonation processes I. The R ct concept. Ozone Sci. Eng.
21, 239–260.
Ernst, M., Jekel, M., 1999. Advanced treatment combination for
groundwater recharge of municipal wastewater by
nanofiltration and ozonation. Water Sci. Technol. 40, 277–284.
Feldmann, D.F., Zuehlke, S., Heberer, T., 2008. Occurrence, fate and
assessment of polar metamizole (dipyrone) residues in hospital
and municipal wastewater. Chemosphere 71, 1754–1764.
Gogate, P.R., Pandit, A.B., 2004a. A review of imperative technologies
for wastewater treatment I: oxidation technologies at ambient
conditions. Adv. Environ. Res. 8, 501–551.
Gogate, P.R., Pandit, A.B., 2004b. A review of imperative
technologies for wastewater treatment II: hybrid methods.
Adv. Environ. Res. 8, 553–597.
Gómez, M.J., Gómez-Ramos, M.M., Agüera, A., Mezcua, M.,
Herrera, S., Fernández-Alba, A.R., 2009. A new gas
chromatography/mass spectrometry method for the
simultaneous analysis of target and non-target organic
contaminants in waters. J. Chromatogr. A, 1216, 4071–4082.
Halling-Sørensen, B., Nielsen, B.N., Lanzky, P.F., Ingerslev, F.,
Lützhoft, H.C.H., Jorgensen, S.E., 1998. Occurrence, fate and
effects of pharmaceutical substances in the environment,
a review. Chemosphere 36, 357–394.
Heberer, T., Reddersen, K., Mechlinski, A., 2002. From municipal
sewage to drinking water: fate and removal of pharmaceutical
residues in the aquatic environment in urban areas. Water Sci.
Technol. 46, 81–88.
Huber, M.M., Canonica, S., Park, G.Y., Gunten, U., 2003. Oxidation
of pharmaceuticals during ozonation and advanced oxidation
processes. Environ. Sci. Technol. 37, 1016–1024.
Huber, M.M., Göbel, A., Joss, A., Hermann, N., Löffler, D.,
Mcardell, C.S., Ried, A., Siegrist, H., Ternes, T.A., von
Gunten, U., 2005. Oxidation of pharmaceuticals during
ozonation of municipal wastewater effluents: a pilot study.
Environ. Sci. Technol. 39, 4290–4299.
Ikehata, K., Naghashkar, N.J., El-Din, M.G., 2006. Degradation of
aqueous pharmaceuticals by ozonation and advanced
ozonation processes: a review. Ozone Sci. Eng. 28, 353–414.
Johnson, M.B., Mehrvar, M., 2008. Aqueous metronidazole
degradation by UV/H 2O 2 process in single- and multi-lamp
tubular photoreactors: kinetics and reactor design. Ind. Eng.
Chem. Res. 47, 6525–6537.
Joss, A., Keller, E., Alder, A.C., Göbel, A., McArdell, C.S., Ternes, T.,
Siegrist, H., 2005. Removal of pharmaceuticals and fragrances
in biological wastewater treatment. Water Res. 39, 3139–3152.
Kupper, T., Plagellat, C., Brändli, R.C., de Alencastro, L.F.,
Grandjean, D., Tarradellas, J., 2006. Fate and removal of
polycyclic musks, UV filters and biocides during wastewater
treatment. Water Res. 40, 2603–2612.
Liu, Z., Kanjo, Y., Mizutani, S., 2008. Removal mechanisms for
endocrine disrupting compounds (EDCs) in wastewater
treatment d physical means, biodegradation, and chemical
advanced oxidation: a review. Sci. Total Environ.. doi:10.1016/
j.scitotenv.2008.08.039.
water research 44 (2010) 578–588
Metcalfe, C., Miao, X.S., Hua, W., Letcher, R., Servos, M., 2004.
Pharmaceuticals in the Canadian environment. In:
Kümmerer, K. (Ed.), Pharmaceuticals in the Environment,
Sources, Fate Effects and Risks. Springer, pp. 67–90.
Muñoz, I., Gómez, M.J., Molina-Díaz, A., Huijbregts, M.A.J.,
Fernández-Alba, A.R., García-Calvo, E., 2008. Ranking potential
impacts of priority and emerging pollutants in urban
wastewater through life cycle impact assessment.
Chemosphere 74, 37–44.
Muñoz, I., Rodríguez, A., Rosal, R., Fernández-Alba, A.R., 2009. Life
cycle assessment of urban wastewater reuse with ozonation
as tertiary treatment. A focus on toxicity-related impacts. Sci.
Total Environ. 407, 1245–1256.
Radjenovic, J., Petrovic, M., Barceló, D., 2009. Fate and distribution
of pharmaceuticals in wastewater and sewage sludge of the
conventional activated sludge (CAS) and advanced membrane
bioreactor (MBR) treatment. Water Res. 43, 831–841.
Raknes, K.L., 2005. Ozone in Drinking Water Treatment: Process
Design, Operation, and Optimization. American Water Works
Association, Denver.
Real, F.J., Benitez, F.J., Acero, J.L., Sagasti, J.J.P., Casas, F., 2009.
Kinetics of the chemical oxidation of the pharmaceuticals
primidone, ketoprofen, and diatrizoate in ultrapure and
natural waters. Ind. Eng. Chem. Res. 48, 3380–3388.
Rischbieter, E., Stein, H., Shumpe, A., 2000. Ozone solubilities in
water and aqueous salt solutions. J. Chem. Eng. Data 45,
338–340.
Rosal, R., Rodríguez, A., Perdigón-Melón, J.A., Mezcua, M.,
Hernando, M.D., Letón, P., García-Calvo, E., Agüera, A.,
Fernández-Alba, A.R., 2008a. Removal of pharmaceuticals and
kinetics of mineralization by O 3/H 2O 2 in a biotreated
municipal wastewater. Water Res. 42, 3719–3728.
Rosal, R., Rodríguez, A., Perdigón-Melón, J.A., Petre, A., García-
Calvo, E., 2008b. Oxidation of dissolved organic matter in the
effluent of a sewage treatment plant by ozone combined with
hydrogen peroxide (O 3/H 2O 2). Chem. Eng. J. 149, 311–318.
Sánchez-Polo, M., Rivera-Utrilla, J., Prados-Joya, G., Ferro-
Garcıa, M.A., Bautista-Toledo, I., 2008. Removal of
pharmaceutical compounds, nitroimidazoles, from waters by
using the ozone/carbon system. Water Res. 42, 4163–4171.
Scheytt, T., Mersmann, P., Lindstädt, R., Heberer, T., 2005.
1-Octanol/water partition coefficients of 5 pharmaceuticals
from human medical care: carbamezepine, clofibric acid,
diclofenac, ibuprofen and propyphenazone. Water Air Soil
Pollut. 165, 3–11.
Tauxe-Wuersch, A., De Alencastro, L.F., Grandjean, D.,
Tarradellas, J., 2005. Occurrence of several acidic drugs in
sewage treatment plants in Switzerland and risk assessment.
Water Res. 39, 1761–1772.
Ternes, T.A., 1998. Occurrence of drugs in German sewage
treatment plants and rivers. Water Res. 32, 3245–3260.
Ternes, T., Bonerz, M., Schmidt, T., 2001. Determination of
neutral pharmaceuticals in wastewater and rivers by liquid
chromatography–electrospray tandem mass spectrometry.
J. Chromatogr. A 938, 175–185.
Terzić, S., Senta, I., Ahel, M., Gros, M., Petrović, M., Barcelo, D.,
Müller, J., Knepper, T., Martí, I., Ventura, F., Jovančić, P.,
Jabučar, D., 2008. Occurrence and fate of emerging wastewater
contaminants in Western Balkan Region. Sci. Total Environ.
399, 66–77.
Wilcox, J.D., Bahr, J.M., Hedman, C.J., Hemming, J.D.C.,
Barman, M.A.E., Bradbury, K.R., 2009. Removal of organic
wastewater contaminants in septic systems using
advanced treatment technologies. J. Environ. Qual. 38,
149–156.

Degradation of the emerging contaminant ibuprofen
in water by photo-Fenton
Fabiola Méndez-Arriaga*, Santiago Esplugas, Jaime Giménez
Departamento de Ingeniería Química, Facultad de Química, Universidad de Barcelona, C/ Martí i Franquès 1, 08028 Barcelona, Spain
article info
Article history:
Received 18 April 2009
Received in revised form
3 July 2009
Accepted 8 July 2009
Available online 15 July 2009
Keywords:
AOP
Photo-Fenton
Non-steroidal anti-inflammatory
drug ibuprofen
1. Introduction
abstract
The recent public interest regarding the presence of pharmaceutical
pollutants in water has raised important concerns due
to the unknown environmental impact and possible damages to
the botany and fauna present in aquatic systems (Halling-Sorensen
et al., 1998; Ternes et al., 2004). One of the most consumed
medications corresponds to the classification of the Non-
Steroidal Anti-Inflammatory Drugs (NSAIDs) with more than 70
million annual prescriptions in the world (Takagi et al., 2006).
Moreover, several recent reports have confirmed the presence of
water research 44 (2010) 589–595
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
* Corresponding author. Tel.: þ34 93 403 9789; fax: þ34 93 402 1291.
E-mail address: fmendeza@ub.edu (F. Méndez-Arriaga).
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.watres.2009.07.009
In this study the degradation of the worldwide Non-Steroidal Anti-Inflammatory Drug
(NSAID) ibuprofen (IBP) by photo-Fenton reaction by use of solar artificial irradiation was
carried out. Non-photocatalytic experiments (complex formation, photolysis and UV/Vis-
H2O2 oxidation) were executed to evaluate the isolated effects and additional differentiated
degradation pathways of IBP. The solar photolysis cleavage of H2O2 generates hydroxylated-IBP
byproducts without mineralization. Fenton reaction, however promotes hydroxylation
with a 10% contamination in form of a mineralization. In contrast photo-Fenton in
addition promotes the decarboxylation of IBP and its total depletion is observed. In absence
of H2O2 a decrease of IBP was observed in the Fe(II)/UV–Vis process due to the complex
formation between iron and the IBP-carboxylic moiety. The degradation pathway can be
described as an interconnected and successive principal decarboxylation and hydroxylation
steps. TOC depletion of 40% was observed in photo-Fenton degradation. The iron-IBP
binding was the key-point of the decarboxylation pathway. Both decarboxylation and
hydroxylation mechanisms, as individual or parallel process are responsible for IBP
removal in Fenton and photo-Fenton systems. An increase in the biodegradability of the
final effluent after photo-Fenton treatment was observed. Final BOD5 of 25 mg L 1 was
reached in contrast to the initial BOD5 shown by the untreated IBP solution (BOD5 < 1mgL 1 ).
The increase in the biodegradability of the photo-Fenton degradation byproducts
opens the possibility for a complete remediation with a final post-biological treatment.
ª 2009 Elsevier Ltd. All rights reserved.
the NSAID ibuprofen (IBP), in effluents of wastewater treatment
plants (WTPs) (Alonso et al., 2006; Gómez et al., 2008). Concentrations
of IBP in the environment are reported between
10 ng L 1 and 169 mgL 1 (Santos et al., 2007). A significant
removal rate of this drug is reached by biological oxidation, in
some cases more than 70% (Carballa et al., 2006). In spite of, its
apparently high biodegradation rate, the ecological risk remains
still high, due to the main byproducts generated during the
biological oxidation. These byproducts, hydroxyl-IBP and carboxy-IBP,
have shown quite similar toxicological consequences
in aquatic environment (Roberts and Thomas, 2006).

590
On the other hand, recent research has shown promising
results in the removal of pharmaceutical pollutants through
the application of Advanced Oxidation Processes (AOPs). The
AOPs are oxidative processes applied for treatment of
contaminants in water, soils and air, based on the presence
and reactivity of the hydroxyl radicals ( OH) which are
generated in atmospheric or subcritical conditions of
temperature and pressure with or without catalyst and/or
reactive energy (electrochemical, UV–Vis or ultrasounds)
(Méndez-Arriaga et al., 2009). The final common byproducts
after AOPs have been reported to be easily degraded by biological
oxidation (Torres et al., 2003). Although AOPs use
different reacting systems (homogeneous or heterogeneous
phases, irradiated or dark conditions, etc.), they are mainly
characterized by the production of hydroxyl radicals with
consecutive unselective attack on the organic material.
Several AOPs have been employed to remove NSAIDs (as
diclofenac, naproxen, IBP, ketoprofen, etc.) but special focus
was placed on the heterogeneous photocatalyst with TiO2,
photo-Fenton solar process, ozonation process, electrochemistry
(Sirés Sadornil Ignacio, 2006) and more recently on
ultrasound energy (Hartmann et al., 2008; Méndez-Arriaga
et al., 2008b). Photo-Fenton reaction is one of the most widely
used AOPs due to its environment friendly application of solar
energy. The extremely complex chemistry of the oxidation
system based on the reaction between iron and hydrogen
peroxide, the called Fenton reaction, has become of great
interest for environmental applications.
Well documented information, in a critical compilation
about the basis of Fenton chemistry as well as its applications
to wastewater treatment, can be founded in Venkatadri and
Peters (1993), Bigda (1995), Chen and Pignatello (1999), Tarr
(2003), Burkitt (2003), Neyens and Baeyens (2003), Safarzadeh-
Amiri et al. (1996) and more recently in a widely recommended
critical review, written by Pignatello et al. (2006). Photo-Fenton,
an attractive alternative for degradation of emerging
contaminants, is based on the redox reaction between Fe(II)
and H2O2 with the concomitant generation of OH and
consecutive catalytic role of iron (Fe(II)/Fe(III)) and it is
improved under UV–Vis irradiation.
Even the Fenton and photo-Fenton reactions have been
applied for wide types of organic contaminants, their use for
the degradation of pharmaceutical compounds has been
applied to a smaller extent than other AOPs like photocatalysis
or O 3.
The Fenton and photo-Fenton processes have been
employed in the treatment of wastewater containing pharmaceutical
pollutants such as azathriopine, asparginase, DCF,
penicillin, a-methyl-phenylglycine, metronidazole, lincomicyn
metabolites, sulfamethazine, sulfamethoxazole, amoxicillin,
bezafibrate, paracetamol, biguanides, guanidine,
triamines, pharmaceutical industry and hospital wastewaters
among others (Klaviaroti et al., 2008; González et al., 2007;
Pérez-Moya et al., 2007; Pérez-Estrada et al., 2005).
Degradation of IBP by photo-Fenton reaction has not been
reported yet. Therefore, the goals of this work were to evaluate
the photo-Fenton ability to degrade IBP in an aqueous
solution with a simulated solar irradiation source as well as to
assess the biodegradability of the identified byproducts in the
effluent after the treatment.
water research 44 (2010) 589–595
2. Reagents, equipments and
analytical procedures
IBP (Sigma–Aldrich), FeSO4$7H2O (Panreac) and H2O2 (30% w/v)
(Panreac) were used without pre-treatment. The experimental
device has been already described in earlier communications
(Méndez-Arriaga et al., 2008a). Initial IBP aqueous solution
(0.87 mM, pH 6.25 0.25) was pumped from the feeding stirred
tank to the Duran-photoreactor (0.08 L) placed inside the
Solarbox (Co.Fo.Me.Gra, Italy). The Xe lamp (Phillips OP 1 kW,
6.9 mEinsteins s 1 in 290–400 nm wavelength range), placed
top-inside of the Solarbox, was switched on, and samples at
different irradiation times were withdrawn. In all experiments,
fresh H2O2 and Fe(II)-acid dilutions were daily
prepared. For photo-Fenton experiments a peristaltic pump
was used to dose solutions of H2O2 and H2SO4 in the entrance
of the photoreactor. Several concentrations of H2O2 were
chosen, from 0 to 0.32 mM (0.04, 0.08, 0.16 and 0.32 mM), with
a dose rate of 0.032 mM min 1 . pH adjust was carried out with
H 2SO 4 diluted solution reaching a final pH 3. IBP concentration
was monitored by HPLC by using an RP-C 18 Trace Extrasil
OD52-5 Micromet 25 0.46 Teknockroma column in isocratic
mode with acetonitrile (99.8% Panreac) and acetic acid (0.25 M)
in 75–25 proportion. Aliquots were also used to measure the
residual concentration of H2O2 by spectrophotometric method
(Nogueira et al., 2005) with a Perkin Elmer UV/VIS Lambda 20.
Total Organic Carbon (TOC) measurements were obtained by
a Shimadzu TOC-V CNS auto sampler instrument. BOD5
determinations were obtained according to Standard Methods
(5120) by respirometric single measuring system OxiTop
procedure. A Liquid Chromatography–Electrospray Ionization
Jasco AS-2050 plus IS mass spectrometer (LC/ESI-MS, TOF
Mariner) was employed to identify several byproducts.
3. Results and discussion
3.1. Photolysis, H 2O 2 and UV–Vis/H 2O 2
photolysis effects
Non-photocatalytic assays were undertaken to evaluate their
isolated influence on the degradation of IBP. For the photolysis
experiment, the IBP solution (initial concentration of 0.87 mM)
was irradiated throughout the photoreactor during 2 h. On the
other hand, several ratios of IBP-H2O2 (ranging from 1:0.001 to
1:10) were mixed in 30 mL vials at free pH conditions (pH
6.25 0.25), constant controlled temperature (30 C) and
continuous dark stirring during 24 h. Furthermore, H2O2
photolysis process was carried out irradiating a solution of
0.87 mM of IBP with 0.32 mM of H 2O 2 during 2 h.
Photolysis let IBP and TOC concentrations unchanged. The
negligible degradation attained by photolysis was not unexpected,
due to the low IBP molar absorption coefficient above
300 nm. Similar results were observed for all ratios IBP-H2O2
after 24 h darkness. Any difference in the original concentration
of IBP was observed and TOC conversion showed an
insignificant change, thus addition of peroxide in the dark
gave a quite stable IBP solution. In contrast, UV–Vis/H2O2
process showed an evident decrease in the concentration of

IBP. When solution was irradiated in presence of H 2O 2, almost
40% of degradation took place after 2 h. The degradation of IBP
can be attributed to the photochemical cleavage of hydrogen
peroxide to yield hydroxyl radicals by light absorption (Tuhkanen,
2004):
H 2O 2 þ hv / 2 OH (1)
In the configuration of our system (Xe lamp and Duranglass
photoreactor) there is an overlap between the emitted
light and the irradiation yet available to be absorbed for H 2O 2
approximately from 290 to 350 nm.
Thus, the elimination of IBP has the unique contribution of
the hydroxyl radical attack produced by H2O2 photolysis but
not due to their separate elements (irradiation or H2O2). An
important remark is the fact that TOC did not decrease. Thus,
the composition of the remained organic solution after 2 h of
UV–Vis/H 2O 2 process could be addressed either to the
hydroxylated byproducts of IBP with weak acidic properties
according to the pH evolution observed from 6.5 to 5.5. The
hydroxylation process of IBP is a consequence of its scavenger
ability in favor of OH. Several indications of the OH scavenger
property in NSAID have been widely reported (Hamburger and
McCay, 1990; Bilodeau et al., 1995; Aruoma and Halliwell, 1988;
Halliwell et al., 1995; Patricò et al., 1999).
3.2. Fenton dark and photo-Fenton reaction
For Fenton dark experiments, several Fe(II) concentrations
(0.15 mM, 0.29 mM; 0.60 mM and 1.2 mM) were selected. Iron
acid solutions were added to 1 L IBP solution (0.87 mM, pH
6.25 0.25) under stirring and controlled temperature (30 C).
H 2O 2 (0.32 mM) was immediately added in the tank and
samples were periodically withdrawn. On the other hand,
photo-Fenton experiments were carried out with similar
conditions in 1.5 L IBP solution as described in Section 2. Iron
acid solution (1.2 mM) was added in the continuously flowed
solution. Several concentrations of H2O2 were chosen, from
0 to 0.32 mM, with a dose rate of 0.032 mM min 1 . pH adjust
was carried out with H2SO4 diluted solution reaching a final
pH 3.
Fig. 1 depicts the IBP conversion after 2 h of Fenton dark
reaction. As higher Fe(II) concentration, higher IBP degradation
is reached. IBP depletion by Fenton involves the chemical
reaction with H 2O 2 as free OH generator (Haber and Weiss,
1934):
Fe(II) þ H2O2 / Fe(III) þ OH þ OH (2)
Fe(III) þ H2O2 / Fe(II) þ O2H þ H þ
Iron acts as catalyst by the simultaneous species Fe(II)/
Fe(III) present during the reaction meanwhile H2O2 is continually
consumed forming OH. The increase of Fe(II) promotes
IBP decrease, in the maximum case ca. of 60% corresponding
to 10% of TOC reduction. Early report (Kennedy et al., 1990)
suggests that for ratios equal or higher to 0.13 of iron/IBP all
water research 44 (2010) 589–595 591
(3)
IBP/ IBPo
1.00
0.75
0.50
0.25
0.00
0 30 60 90 120
Time (min)
Fig. 1 – Effect of Fe(II) concentration on removal of IBP
(0.87 mM) by Fenton dark reaction.(B) 0.15 mM
Fe(II),(:)0.29 mM Fe(II); (,)0.60 mM Fe(II) and (A)1.2 mM
Fe(II). In all cases 0.32 mM of H2O2.
the coordination sites of the metal are occupied by IBP and the
OH production is hampered. However, in this work under the
ratios tested – between 0.12 and 0.33 – the catalyst has shown
to be reactive with H2O2 promoting the degradation of IBP.
On the other hand, in photo-Fenton experiments more
than 80% removal of IBP and almost 40% decrease of the initial
TOC were achieved by using 0.32 mM H2O2 and 1.2 mM Fe(II).
These values are four folds higher than the observed under
dark conditions. Fig. 2 shows the IBP/IBP o for photo-Fenton
process together with photolysis, H 2O 2/UV–Vis, Fenton and
Fe(II)/UV–Vis process, and Fig. 3 shows the concomitant TOC
removal for all the above described processes.
Fig. 2 shows an evident change in the kinetic degradation
of IBP by means of photo-Fenton reaction in comparison with
the dark Fenton series above described. The increase in the
reactivity of the photo-Fenton reaction is attributed to the
simultaneous OH generation from the H2O2/UV–Vis and
photo-Fenton systems. The photo-assisted reaction (photo-
Fenton) gives typically faster rates of degradation and
mineralization than the thermal (‘‘dark’’) reaction. In photo-
Fenton reaction the production of OH is significantly
increased under illuminated conditions due to the stimulating
photoreduction of Fe(III) to Fe(II):
IBP/ IBPo
1.00
0.75
Photolysis
H 2 O 2 /UV-Vis
0.50
Fenton
Fe(II)/UV-Vis
0.25
Photo-Fenton (1.2mM Fe(II))
0.00
0 30 60 90
0.04mM H2O2 0.08mM H2O2 0.16mM H2O2 0.32mM H
120
2O2 Irradiation Time (min)
Fig. 2 – Removal of IBP (0.87 mM) by (-) Photolysis; (,)
H 2O 2/UV–Vis (0.32 mM H 2O 2); (A) Fenton (1.2 mM Fe(II),
0.32 mM H 2O 2); (>)Fe(II)/UV–Vis (1.2 mM Fe(II)); Photo-
Fenton in all cases 1.2 mM Fe(II) and (:) 0.04 mM H 2O 2;(6)
0.08 mM H2O2; (C) 0.16 mM H2O2; (B) 0.32 mM H2O2.

592
% TOC removal
50
40
30
20
10
0
Photo-Fenton 1.2mM Fe(II)
Fig. 3 – TOC removal for H 2O 2/UV–Vis (0.32 mM H 2O 2);
Fenton (1.2 mM Fe(II), 0.32 mM H 2O 2); Fe(II)/UV–Vis (1.2 mM
Fe(II)); Photo-Fenton with 1.2 mM Fe(II) and 0.04 mM H 2O 2,
0.08 mM H 2O 2, 0.16 mM H 2O 2, 0.32 mM H 2O 2. In all cases
0.87 mM initial IBP concentration.
Fe(III) þ H2O þ hv / Fe(II) þ H þ þ OH (4)
Ferrous ion is continuously recycled by irradiation and
therefore it is not depleted during the course of the oxidation,
as stated by Zepp et al. (1992). Additionally Pignatello et al.
(1999) have shown the formation of iron complexes, as
(FeOH) 2þ – predominant in acidic conditions – which have
adsorption bands in the UV–Vis range (between 180 and
410 nm) (Sirés Sadornil Ignacio, 2006), and act as additional
hydroxyl radical source in photo-Fenton reaction.
Fe(II) þ H2O2 / Fe(OH) 2þ þ OH (5)
Degradation of IBP was directly proportional to the amount
of H2O2 employed. However, H2O2 was not in all cases full
consumed. For example, almost 60% of 0.32 mM of H2O2
remained unconsumed. The same proportion is observed by
using 0.16 mM however a clear decrease tendency was
observed between 90 and 120 min. In contrast 0.08 and
0.04 mM of H 2O 2 were almost full consumed after 120 min.
Fenton reaction shows a rapid degradation phase resulting
from a burst of $OH by Eq. (2). However when H2O2 is in large
excess the extent of this burst phase will depend on the Fe/
contaminate molar ratio because the ratio determines the
OH/contaminant ratio in the burst phase (Pignatello et al.,
2006).
Moreover due to the pH was adjusted into the first 30 min,
an early parallel formation of Fe(III) could promote in strong
manner the Fenton-like reaction (Eq. (3)) rather than Fenton
reaction. Since reaction in Eq. (3) is so much slower than
reaction in Eq. (2), the parallel formation of Fe(III) also results
in a slower initial rate of IBP and similar slow consume of
H2O2.
In the dark series more than 90% of the H2O2 remained
after 2 h of reaction. TOC depletion was enhanced under
irradiated conditions in the photo-Fenton reaction. As noted
in Fig. 3 the maximum TOC removal is observed around 40%.
The elimination of organic carbon is also due to the formation
water research 44 (2010) 589–595
of volatile compounds (see also Section 3.5). Thus IBP oxidation
and TOC depletion depend on experimental conditions
and ratios between reactants, in a rather complex way due to
the large number of reactions involved.
3.3. Iron effect: IBP as organic ligand of iron
and its photolysis
An additional important query in Fenton process is the evaluation
of possible complex formation between the organic
compound and the metallic catalyst. Normally in Fenton
applications the iron-complex formation with organic ligands
is employed in order to widen the pH range, to avoid precipitation
of Fe(III)-oxyhydroxide and/or to increase the ironphotolysis
process (Andreozzi et al., 2006). However, undesirable
effects can be also observed if the active coordination
sites in the metal are not able to generate hydroxyl radicals in
the Fenton reaction.
The experiment carried out with 1.2 mM Fe(II) without
H 2O 2, under irradiated condition (> doted in Fig. 2), gave an
unexpected 60% of IBP removal as shown in Fig. 2. The
removal of IBP using Fe(II)/UV–Vis configuration (without
H 2O 2) was in deep studied by additional experiments varying
the iron concentration under irradiated and free pH 5 conditions.
Fig. 4 shows the IBP evolution, for 0.012, 1.2 and 5 mM
FeSO4.7H2O concentrations, during 2 h of irradiation. Elimination
of IBP is directly proportional to the amount of the
irradiated-iron employed. IBP binds iron and under UV–Vis
irradiated conditions this iron–IBP complex is photoactive. At
pH 5, a large fraction of iron is in the form of Fe(III). Coordination
of IBP with Fe(II) is probably negligible and would not be
photolabile in the wavelength range used. Therefore, the
reaction was due to the Fe(III)-complex photolysis.
Iron forms complexes with organic compounds, especially
those acting as polydentate ligands like Fe(III)–carboxylate or
Fe(III)–polycarboxylate complexes (Pignatello et al., 2006). In
the case of NSAID – which are characterized for their
carboxylic acid moiety – some authors have shown their
capacity to bind iron to form coordination complex. For
instance, Kennedy et al. (1990) suggest that the complex
formation between IBP and iron can be reached through the
carboxylic moiety. Under our experimental conditions the
possible production of decarboxylated byproducts and acidic
IBP (mM)
1.00
0.75
0.50
0.25
0.00
0 30 60
Time (min)
90
Fig. 4 – Effect of Fe(III)/IBP coordinated complex photolysis
on IBP degradation during 120 min of UV–Vis irradiation.
Initial Fe(II) acid ion added (C) 0.012 mM; (:) 1.2 mM; (-)
5 mM and 0.87 mM IBP initial concentration.
120

Table 1 – Isolated and combined effects of Fenton and
photo-Fenton reagents on the IBP conversion and TOC
removal observed at 2 h.
Process A þ B / C XIBP TOC
removal
Fe(III) IBP þ hv Decarboxylated byproducts 0.70 20%
H2O2 þ IBP Stable solution without reaction 0.00 0%
H2O2 þ IBP þ hv Hydoxylated bycompounds 0.30 2%
Fe(II) þ H2O2 þ IBP Fe(III)-IBP coordinated complex
þ hydroxylated byproducts
0.50 10%
Fe(II) þ H2O2 þ hv
þ IBP
Decarboxylated þ hydroxylated
þ biodegradable compounds
1.00 40%
aliphatic compounds are related with the change of the pH
during the reaction from initial 5 to final 3.5. Higher iron–IBP
complex amount, higher depletion of IBP through the
carboxylic moiety, in agreement with reports of oxidation in
iron-complex with organic functional group RCOO (Pignatello
et al., 2006). The ligand-to-metal-charge-transfer reduction of
the metal center promotes decarboxylation of IBP, at least
partially, and further degradation of decarboxylated bycompounds
is also driven under whole photo-Fenton system.
3.4. Comparison between photo-Fenton processes and
its isolated or combined reagents
The photo-Fenton degradation of IBP is strongly marked for
the hydroxyl attack and decarboxylation process. In dark
water research 44 (2010) 589–595 593
H 3 C
CH 3
CH3 CH3 CH3 OH
CH3 CH3 CH3 H3C 4
O
HO
5
CH3 H3C 6 OH
H3C 7
O
conditions, the catalytic reaction is less favorable and low TOC
removal is observed. In Table 1 all defined stages of photo-
Fenton process are described together with their simultaneous
contributions of isolated and combined reagents. The
degradation of IBP by iron irradiated process seems to be the
principal way as initial mechanism of degradation even if
H2O2 is not present (0.7 and 20% of IBP conversion and TOC
removal respectively). In photo-Fenton process, the OH yield
improves and simultaneous scavenges the isobutyl moiety of
IBP. Thus the carboxylic moiety of IBP binds the iron while the
isobutyl moiety scavenges the OH radical generated.
3.5. Byproducts from photo-Fenton process
IBP byproducts generated by photo-Fenton treatment (1.2 mM
Fe(II), 0.32 mM H 2O 2, 2 h of irradiation) were identified by an
LC/ESI-TOF-MS, into the m/z range of 65–1000 in negative
ionization mode.
Fig. 5 depicts the proposed reaction pathway with the
corresponding intermediates identified. The remaining TOC
after treatment of IBP (m/z ¼ 206, TR 28) consisted mostly of
two principal types of byproducts: decarboxylated and
hydroxylated metabolites of IBP. Hydroxy-IBP byproducts
identified were the three structures 1 (m/z ¼ 222, TR 16), 2 (m/
z ¼ 222, TR 23) and 3 (m/z ¼ 222, TR 18). The dihydroxylated
compound 10 (m/z ¼ 238, TR 19) was founded at low intensity.
Compound 1 suffers either the cleavage of C 1–C 2 bond of the
isobutyl moiety, forming 4 (m/z ¼ 164, TR 13) and isopropanol
or ketone acetone; or the decarboxylation forming 5
CH 3
CH3 CH3 CH3 OH
CH3 OH
CH3 HO
1 CH3 O
H3C 2 OH
O
H3C 3
CH 3
IBP
O
OH
H3C H3C 8 OH
9
O
CH3 CH3 CH 3
HO
H3C 10
CH 3
CH 3
O
OH
CH3 HO
O
O
OH
Fig. 5 – Byproducts identified and possible degradation pathway by photo-Fenton reaction (1.2 mM Fe(II), 0.32 mM H2O2, 2h
of irradiation, 0.87 mM IBP initial concentration). ID Name, Molecular Mass [Fragments m/z] IBP Ibuprofen, 206 [205 L ; 161 L ].
1 2-HydroxyIBP, 222 [221 L ; 177 L ]. 2 1-HydroxyIBP-, 222 [221 L ; 177 L ]. 3 2-hydroxy-2-[4-(2methylpropyl)phenyl]propanoic
acid, 222 [221 L ; 191 L ]. 4 (2RS)-2-(4-Methylphenyl)propanoic acid, 164 [163 L ; 133 L ]. 5 1-ethyl-4-(2-hydroxy)isobutylbenzene,
178 [177 L , 149 L ]. 6 1-ethyl-4-(1-Hydroxy)isobutylbenzene, 176 [175 L ; 133 L ]. 7 1-[4-(2-Methylpropyl)phenyl]ethanone, 176
[175 L ; 133 L ]. 8 4-(1-hydroxy-2-methylpropyl)acetophenone, 192 [191 L ; 177 L ]. 9 2-Methyl-1-phenylpropane, 134[133 L ;
119 L ]. 10 2-hydroxy-2-[4-(2methylpropyl)phenyl]peroxic acid, 238[237 L ; 221 L ].

594
(m/z ¼ 178, TR 15.5) and formic acid. In the case of 2, the
evidence of its decarboxylation could be represented by the
byproduct 6 (m/z ¼ 178, TR 15.5). However undifferentiated
retention times by MS analysis with its analogous compound
5 were obtained.
In the case of 3, the degradation process follows two
possible pathways by either decarboxylation and deprotonation
to reach 7 (m/z ¼ 176, TR 21) and formic acid, and by
dihydroxylation to obtain the byproduct 10. However
byproduct 10 can be also formed by a non-free radical
pathway involving a nucleophilic exchange of OH of the
carboxylic acid with H 2O 2. 7 can be again hydroxylated in
position 1 of the isobutyl to reach 8 (m/z ¼ 191, TR 25) and also
can be decarboxylated to form 9 (m/z ¼ 133, TR 15) and
acetaldehyde.
An approximation of the main byproducts remained after
photo-Fenton treatment could be assigned to compounds
7 and 9 with 23 and 36% of the total intensities. Thus the
degradation of IBP by photo-Fenton goes through the simultaneous
cleavage of the carboxyl acid moiety to produce formic
acid or acetaldehyde and scavenger of hydroxyl radicals. As
described before the formation of high soluble bycompounds
isopropanol or formic acid and acetone promotes stable
dissolution. The presence of those soluble compounds after 2 h
of reaction improves the biodegradability in the effluent
treated. Initial BOD5 for 0.87 mM of untreated-IBP was

and effects of pharmaceutical substances in the environment -
a review. Chemosphere 36, 357–393.
Hartmann, J., Bartels, P., Mau, U., Witter, M., Tumpling, W.V.,
Hofmann, J., Nietzschmann, E., 2008. Degradation of the drug
diclofenac in water by sonolysis in presence of catalysts.
Chemosphere 70, 453–461.
Kennedy, T.P., Rao, N.V., Noah, W., Michael, J.R., Jafri, M.H.,
Gurtner, G.H., Hoidal, J.R., 1990. Ibuprofen prevents oxidant
lung injury and in vitro lipid-peroxidation by chelating iron.
J. Clin. Invest. 86, 1565–1573.
Klavarioti, M., Mantzavinos, D., Kassinos, D., 2008. Removal of
residual pharmaceuticals from aqueous systems by advanced
oxydation processes. Review article. Environment
International 35 (2), pp. 402–417.
Méndez-Arriaga, F., Esplugas, S., Gimenez, J., 2008a.
Photocatalytic degradation of non-steroidal antiinflammatory
drugs with TiO 2 and simulated solar irradiation.
Water Res. 42, 585–594.
Méndez-Arriaga, F., Torres Palma, R., Pétrier, C., Esplugas, S.,
Gimenez, J., Pulgarin, C., 2008b. Ultrasonic treatment of water
contaminated with ibuprofen. Water Res. 42, 4243–4248.
Méndez-Arriaga F., March 2009. Advanced Oxidation Processes
(Photocatalysis, Photo-Fenton and Sonolysis) for Removal of
Pharmaceutical Pollutants in Water. Doctoral thesis,
Barcelona University.
Neyens, E., Baeyens, J., 2003. A review of the classic Fenton’s
peroxidation as an advanced oxidation technique. J. Hazard.
Mater. 98, 33–50.
Nogueira, R.F.P., Oliveira, M.C., Paterlini, W.C., 2005. Simple and
fast spectrophotometric determination of H2O2 in photo-
Fenton reactions using metavanadate. Talanta 66, 86–91.
Patricò, D., Pasin, M., Barry, O.P., Ghiselli, A., Sabatino, G.A.,
Iuliano, L., FiztGerald, G., Violi, F., 1999. Iron-dependent
human platelet activation and hydroxyl radical formation:
involvement of protein kinase C. Circulation 99, 3118–3124.
Pérez-Estrada, L., Malato, S., Gernjak, W., Agüera, A.,
Thurman, M., Ferrer, I., Fernandez-Alba, A., 2005. Photo
Fenton degradation of diclofenac: identification of main
intermediates and degradation pathway. Environ. Sci.
Technol. 39 (21), 8300–8306.
Pérez-Moya, M., Graells, M., del Valle, L., Centelles, E., Mansilla, H.
, 2007. Fenton and photo-Fenton degradation of
2-chlorophenol: Multivariate analysis and toxicity monitoring.
Catalysis Today, 124, 3-4, 30 2007, 163–171.
Pignatello, J.J., Liu, D., Huston, P., 1999. Evidence for an additional
oxidant in the photoassisted Fenton reaction. Environ. Sci.
Technol. 33, 1832–1839.
water research 44 (2010) 589–595 595
Pignatello, J.J., Oliveros, E., MacKay, A., 2006. Advanced oxidation
processes for organic contaminant destruction based on the
Fenton reaction and related chemistry. Crit. Rev. Environ. Sci.
Technol. 36, 1–84.
Roberts, P.H., Thomas, K.V., 2006. The occurrence of selected
pharmaceuticals in wastewater effluent and surface
waters of the lower Tyne catchment. Sci. Total Environ. 356,
143–153.
Safarzadeh-Amiri, A., Bolton, J.R., Cater, S.R., 1996. The use of iron
in advanced oxidation technologies. J. Adv. Oxid. Technol. 1,
18–26.
Santos, J.L., Aparicio, I., Alonso, E., 2007. Occurrence and risk
assessment of pharmaceutically active compounds in
wastewater treatment plants. A case study: Seville city
(Spain). Environ. Int. 33, 596–601.
Sirés Sadornil Ignacio, 2006. Electrochemical Advanced Oxidation
Processes for the Removal of the Drugs Paracetamol, Clofibric
Acid and Chlorophene from Waters. Doctoral thesis,
Universidad de Barcelona.
Takagi, T., Ramachandran, C., Bermejo, M., Yamashita, S., Yu, L.X.,
Amidon, G.L., 2006. A provisional biopharmaceutical
classification of the top 200 oral drug products in the United
States, Great Britain, Spain, and Japan. Mol. Pharmacol. 3,
631–643.
Tarr, M.A., 2003. Chemical degradation methods for wastes and
pollutants. Chemical Degradation Methods for Wastes and
Pollutants. In: Environmental Science and Pollution Control
Series, vol. 26 165–200.
Ternes, T.A., Herrmann, N., Bonerz, M., Knacker, T., Siegrist, H.,
Joss, A., 2004. A rapid method to measure the solid-water
distribution coefficient (K–d) for pharmaceuticals and musk
fragrances in sewage sludge. Water Res. 38, 4075–4084.
Torres, R., Sarria, V., Torres, W., Peringer, P., Pulgarin, C., 2003.
Electrochemical treatment of industrial wastewater
containing 5-amino-6-methyl-2-benzimidazolone: toward and
electrochemical–biological coupling. Water Res. 37, 7–13.
Tuhkanen, T., 2004. UV/H 2O 2 processes. In: Parsons, S. (Ed.),
Advanced Oxidation Processes for Water and Wastewater
Treatment. IWA, London UK, pp. 86–109.
Venkatadri, R., Peters, R., 1993. Chemical oxidation technologies:
ultraviolet light/hydrogen peroxide, Fenton’s reagent and
titanium dioxide-assisted photocatalysis, hazard. Waste
Hazard. Mater. 10, 107–149.
Zepp, R.G., Faust, B.C., Hoigne, J., 1992. Hydroxyl radical
formation in aqueous reactions (pH 3–8) of iron(II) with
hydrogen-peroxide - the photo-Fenton reaction. Environ. Sci.
Technol. 26, 313–319.

Polar pollutants in municipal wastewater and the water cycle:
Occurrence and removal of benzotriazoles
Thorsten Reemtsma a, *, Ulf Miehe b , Uwe Duennbier c , Martin Jekel b
a
Federal Institute for Risk Assessment, Chemicals Safety, Thielallee 88–92, 14195 Berlin, Germany
b
Technical University of Berlin, Department of Water Quality Control, Sekr KF 4, Strasse des 17 Juni 135, 10623 Berlin, Germany
c
Berliner Wasserbetriebe, Laboratory, Motardstrasse 35, 13629 Berlin, Germany
article info
Article history:
Received 3 April 2009
Received in revised form
9 July 2009
Accepted 11 July 2009
Available online 18 July 2009
Keywords:
Municipal wastewater
Wastewater
Treatment
Benzotriazole
Bank filtration
Ozonation
Activated carbon
Groundwater
Corrosion inhibitor
Rivers
Rhine
Elbe
1. Introduction
abstract
Benzotriazoles, in which a 1,2,3-triazole ring is condensed to
a benzene ring, are a class of high production volume
chemicals (HPVC) used in a wide variety of applications. The
basic benzotriazole components are 1H-benzotriazole (BTri,
also abbreviated as BTZ, BTA or BTAH) and the 1H-methylbenzotriazoles
(tolyltriazoles; TTri, available as technical
water research 44 (2010) 596–604
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
1H-benzo-1,2,3-triazole (BTri) and its methylated analogues (tolyltriazole, TTri) are corrosion
inhibitors used in many industrial applications, but also in households in dishwashing
agents and in deicing fluids at airports and elsewhere. BTri and one of the TTri-isomers
(4-TTri) are typical examples of polar and poorly degradable trace pollutants. Benzotriazole
elimination in four wastewater treatment plants (WWTP) in Berlin ranged from 20 to 70%
for 5-TTRi over 30 to 55% for BTri to insignificant for 4-TTri. WWTP effluent concentrations
were in the range of 7–18 mg/L of BTri, 1–5 mg/L of 4-TTri and 0.8–1.2 mg/L of 5-TTri. BTri and
4-TTri proved to be omnipresent in surface waters of the rivers Rhine and Elbe with
concentrations increasing from

enzotriazoles may dissociate and occur as anions at slightly
basic pH or in contact with metal cations. At neutral pH in
aqueous solution benzotriazoles are expected to be non-ionic
rather than cationic.
Owing to their NH acidity, benzotriazoles show metal
complexing properties. BTri and TTri prevent corrosion of
metals and steels, especially of copper and brass (Hart et al.,
2004), by forming a thin complexing film on the metal surface
(Chadwick and Hashemi, 1978). For this reason BTri or TTri are
used in the metal finishing industry, in semiconductor
industry (Hollingsworth et al., 2005), in milk processing (Verheyen
et al., 2009), and added to various fluids that come in
contact with metals, such as aircraft deicing and antiicing
fluids (Gruden et al., 2001), cooling liquids, brake fluids, metalworking
fluids, etc. BTri is also used for silver protection in
dishwashing machines (Ort et al., 2005). The annual production
of benzotriazoles was reported to be in the range of
9000 tons/year worldwide (Reemtsma et al., 2006). Benzotriazole
can also be employed in the developing process in the
photographic industry (Bergthaller, 2002) and is a biocide
included in Annex I of Council Directive 98/8/EC (European
Commission, 2007).
Structurally more complex hydroxiphenyl-benzotriazoles
are found in personal care products, where they act as stabilizers
for UVA filter components (De Orsi et al., 2006). In the
European Union some benzotriazole derivatives are included
in Annex VII, part I of Council Directive 76/768/EEC as registered
cosmetic ingredients (European Commission, 2000).
Structurally related and partly chlorinated benzotriazole
derivatives are used as stabilizers in plastic material such as
polyethylene terephthalate (PET) (Bentayeb et al., 2007).
First reports on the occurrence of benzotriazoles in the
environment were related to their use in open systems as
aircraft deicing fluids at airports (Cancilla et al., 1998, 2003).
A few years later it was recognized that BTri and TTri are
much more widely distributed in surface waters (Reemtsma
et al., 2006; Voutsa et al., 2006; Giger et al., 2006) in 0.1–1 mg/L
concentrations and are regularly discharged with municipal
wastewater even after state-of-the art biological treatment
in mg/L concentrations (Weiss and Reemtsma, 2005; Voutsa
et al., 2006; Weiss et al., 2006). Only one of these studies
differentiated between the two TTri-isomers (Weiss et al.,
2006) by liquid-chromatography tandem mass spectrometry
(LC-MS/MS) (Weiss and Reemtsma, 2005). With this method
it could be shown that the two isomers behave quite differently
in the environment. While 5-TTri was mineralized
within 15 days in an aerobic biodegradation test with activated
sludge, the 4-TTri isomer remained stable (Weiss et al.,
2006). The much higher stability of 4-TTri as compared to
5-TTri has also been observed in wastewater treatment
and in surface waters (Weiss and Reemtsma, 2005; Weiss
et al., 2006).
Unlike the N-substituted 1,2,4-triazoles, which are used as
herbicides, the benzo-1,2,3-triazoles have only limited biological
activity. EC50 values for the acute toxicity to Vibrio
fisheri were in the range of 7–21 mg/L and NOAEC levels for
invertebrates and vertebrates in the range of 10–100 mg/L
(Pillard et al., 2001). BTri has been shown to exhibit antiestrogenic
activity in vitro in a yeast assay but not in vivo to
fish (Harris et al., 2007).
water research 44 (2010) 596–604 597
Despite these favourable ecotoxicity data the widespread
occurrence of benzotriazoles in the environment should be
a reason to improve our knowledge on the environmental
behaviour of these compounds. Inter alia, the fate of benzotriazoles
in surface waters is not well studied, especially for
the TTri-isomers. It could also be worthwhile to become aware
of those processes that are suitable for the removal of benzotriazoles
from wastewater as well as from raw waters used
for drinking water production.
2. Materials and methods
2.1. Wastewater treatment plants
Samples are taken from the four largest wastewater treatment
plants (WWTPs) in Berlin with treated water flows between
40,000 and 200,000 m 3 /day. All plants are equipped with
primary sedimentation, conventional activated sludge (CAS)
treatment with nutrient removal (N/P removal) and secondary
clarification. The plants differ in the P removal process:
WWTP-R and W use biological P-elimination (Bio-P), WWTP-S
with Bio-P and chemical precipitation and WWTP-M with
chemical precipitation only. Hydraulic retention time (HRT) in
the plants ranged from 11 to 24 h for the biological stage and
the sludge retention time (SRT) from 8 to 25 days. The influent
to WWTP-R had an organic load of 300–900 mg/L chemical
oxygen demand (COD), while it ranged from 700 to 1100 mg/L
COD for the other three plants.
Samples of the influent (raw sewage) and the effluent of the
secondary clarifier were taken as 24 h mixed samples between
June and December 2006. Sampling was performed only at dry
weather flow, and the sampling of effluent was corrected for
hydraulic retention time in each of the plants.
2.1.1. Flocculation-rapid filtration (F-RF)
A polishing of the effluent of WWTP-R was investigated by
rapid filtration on two experimental dual-media filters (0.8 m
sand/1 m anthracite layer) that treated 100–180 L/h. Particle
sizes were 0.7–1.25 mm for sand and 1.4–2.5 mm for anthracite
in Filter 1 and 1.1–1.6 mm for sand and 2.5–4.0 mm for
anthracite in Filter 2. Filter velocities ranged from 6 to 10 m/h.
Filter backwash was performed every 1–2 days.
Before filtration the effluent was treated by flocculation
with either Al 3þ (1 mg/L) or Fe 3þ (2 mg/L). During some phases
10–20 mg/L of powdered activated carbon (PAC; Norit Super
SAE) were added during flocculation (flocculation, PAC, rapid
filtration: F-PAC-RF).
2.2. Field sites and field samples
2.2.1. Rivers Rhine and Elbe
Single grab samples were taken with a bucket in January 2007
at sites used for sampling for the environmental specimen
bank of the Federal Environment Agency (UBA).
2.2.2. River Havel
Samples of the river Havel were taken at the surface water
treatment plants Spandau (upstream, n ¼ 5) and Beelitzhof
(downstream; n ¼ 8) from July to November 2006.

598
2.2.3. Bank filtration
Bank filtration was studied at two sites in Berlin (Germany), at
Lake Tegel and Lake Wannsee/Havel. Details of the bank
filtration, the hydrology of the sites and the transects installed
have been published before (Gruenheid et al., 2005; Massmann
et al., 2008). Samples were withdrawn from several observation
wells and from the production well at a distance of about
100 m from the lakeshore. Samples, taken in April 2008 from
the Tegel site and in May 2008, from both sites, were analyzed
for benzotriazoles.
2.2.4. Surface water treatment plants
Two surface water treatment plants (Tegel and Beelitzhof)
were investigated. These plants are designed to reduce the
total phosphorus load of the surface water (median of 265 mg/L
for Tegel and 129 mg/L for Beelitzhof) to values between 5 and
30 mg/L by coagulation/flocculation, sedimentation and dualmedia
filtration. The influent of the Tegel plant consists of
25–80% effluent of WWTP-S, diluted with surface water. The
Tegel plant treats 130,000–300,000 m 3 /day, the final filters
have a height of 1.8 m and an average filter velocity of 6.6 m/h.
The Beelitzhof plant treats surface water of Lake Wannsee
(part of the River Havel system) downstream of the City of
Berlin. A volume of 12,000 m 3 /day is treated in this plant and
filtration occurs on filters of 1.6 m height and at a filter velocity
of 5 m/h.
2.3. Lab experiment
2.3.1. PAC sorption studies
Batch experiments for isotherm calculation were performed
in either ultrapure water or effluent of WWTP-R with start
concentrations of 100 mg/L of BTri or TTri (technical mix consisting
of 40% 4-TTri and 60% 5-TTri (Weiss and Reemtsma,
2005). Between 1 and 25 mg/L of PAC (Norit Super SAE) were
added from suspensions (0.1–1 g/L) stirred at 600 rpm. The
bottles were continuously mixed on a horizontal shaker
(120 rpm) for 24 h. All experiments were performed in duplicate.
Stability of the compounds was recorded in a control
experiment without PAC addition. Dissolved concentrations
of BTri and TTri (sum of both isomers) were determined after
membrane filtration (0.45 mm cellulose-nitrate membrane
filters) by LC-MS/MS analysis. A Freundlich model was fitted to
the data.
water research 44 (2010) 596–604
2.4. Analysis of benzotriazoles
All field samples were stored in glass bottles, transported to
the lab in ice chests, filtered over 0.45 mm membrane filters
and stored in a refrigerator at 4 C pending analysis, usually
within the next 24 h.
Benzotriazoles were analyzed by LC-MS/MS with an isocratic
elution by methanolic eluents on a diphenyl column
(Pursuit Diphenyl, 3 mm, 150 2 mm) that allowed separation
of the two TTri-isomers. Details have been published before
(Weiss and Reemtsma, 2005). Samples of surface water and
WTTP influents and effluents were analyzed by direct injection
(20 mL for surface water and WWTP effluent; 10 mL for
WWTP influent) into the LC-MS/MS system, while groundwater
samples were analyzed after enrichment of a volume of
50 mL by solid phase extraction (SPE) on a Oasis HLB 200 mg
cartridge. With this procedure an LOQ of 0.15 mg/L (surface
water) and 0.25 mg/L (wastewater) was obtained for direct
injection and 0.05 mg/L after SPE (groundwater) for each of the
three analytes.
3. Results and discussion
3.1. Wastewater treatment plants
Four WWTP in Berlin, treating municipal wastewater as a mix
of household and industrial wastewater were investigated for
the occurrence and removal of benzotriazoles over a 6 month
period (Fig. 1, Table S1). Mean influent concentration for the
four plants ranged from 17 to 44 mg/L for BTri and from 1.1 to
4.9 mg/L for each of the TTri-isomers. BTri and TTri concentrations
were both higher in WWTP-M as compared to the
other three plants. This was not due to a generally more
concentrated wastewater in this plant, because the BTri/DOC
ratio of 0.6& was also twice as high as the value for the other
three plants (0.2–0.3&). The source of this elevated input of
benzotriazoles to this WWTP is unknown.
These influent concentrations are comparable to those
reported in previous studies: mean concentrations of 6.6 mg/L
BTri and 0.7 mg/L for TTri were found in a monitoring of eight
WWTP in Western Europe (Reemtsma et al., 2006). In a study
carried out in Switzerland on 10 WWTP even higher influent
concentrations of 15–73 mg/L for BTri and 1.1–5.6 mg/L for
Fig. 1 – Concentrations of (a) BTri, (b) 5-TTri and (c) 4-TTri in the influent and effluent of the four largest WWTP in Berlin (R
(n [ 5), S (n [ 8), W (n [ 9): 05/2006–04/2007; M (n [ 8): 10/2005–04/2007).

TTri-isomers were recorded (Voutsa et al., 2006). In a previous
study on WWTP-R in Berlin 12.0 mg/L for BTri and 1.3 mg/L for
5-TTri and 2.1 mg/L for 4-TTri were reported as the 2-year
average (Weiss et al., 2006), which is at the lower end of
concentrations determined in this study.
Benzotriazoles in municipal wastewater may originate
from household due to their use as corrosion inhibitors in
dishwashing agents. This could be proven for a residential
area by modeling the loads in its sewer system (Ort et al.,
2005). When wastewater of a larger urban area is collected in
one sewer system, industrial effluents, e.g. from metal finishing
or semiconductor industry, may also be of relevance.
Milligram per litre concentrations of TTri have only recently
been reported in wastewater from milk processing (Verheyen
et al., 2009). In mixed sewer systems runoff from streets
(traffic) may also contribute. Airports, which are a well studied
point source of benzotriazole emission (Cancilla et al., 1998,
2003), may also contribute to the load in municipal wastewater
if their surface runoff is collected and directed to the
municipal sewer system (Weiss et al., 2006).
Of the three benzotriazoles, none was eliminated
completely during wastewater treatment. Generally relative
removal was best for 5-TTri, but highly variable between the
plants; it ranged from 19 to 69%. BTri showed moderate
removal (29–58%), whereas for 4-TTri only one plant showed
a significant but still limited removal (34%) (Fig. 1, Table S1).
Only moderate removal of BTri and TTri (sum of both isomers)
has been reported before in a monitoring of eight WWTP in
Western Europe (Reemtsma et al., 2006) and for a variety of
WWTP in Switzerland (Voutsa et al., 2006).
This different removal of the two TTri-isomers observed
in this study is in agreement with lab experiments, which
showed that 5-TTri can be mineralized by activated sludge
while 4-TTri is persistent (Weiss et al., 2006). These differences
are attributed to the position of the methyl substituent
at the ring system. Previous studies reported that the aerobic
biodegradability of positional isomers of bicyclic aromatic
compounds depends upon the substitution pattern and that
isomers with the ortho-substituent (as in 4-TTri) were not
biodegradable, while the isomer with the meta-substituent (as
in 5-TTri) were degraded (Weiss and Reemtsma, 2005).
The different removal behaviour of the three benzotriazoles
can be highlighted by calculating the concentration
ratios of either BTri or 5-TTri to the most persistent 4-TTri.
Both these ratios tend to decrease during biological treatment
from 6–7 to 4–5 for BTri/4-TTri and from 0.7–0.9 to 0.3–0.8 for
5-TTri/4-TTri (Table S1). These ratios are especially useful to
differentiate between dilution (decrease in concentrations but
no change in the concentration ratios) and biodegradation
(decrease in concentrations and in concentration ratios). This
underlines that an appropriate analytical method that separates
the isomers is essential to adequately study the environmental
fate of TTri (Weiss and Reemtsma, 2005).
Owing to the incomplete removal BTri median values of
7–18 mg/L were found for the effluents of the four WWTP.
Single values up to 38 mg/L were recorded in the effluent of
WWTP-M. In a Swiss study BTri effluent concentrations
ranging from 10 to 100 mg/L have been found (Voutsa et al.,
2006). These are remarkably high concentrations for single
compounds in WWTP effluents. In a study of eight European
water research 44 (2010) 596–604 599
WWTP, in which 28 trace pollutants have been monitored,
BTri ranked second in effluent concentration; it was outcompeted
only by ethylenediamine tetraacetic acid (EDTA)
with 100 mg/L (Reemtsma et al., 2006).
As shown previously the elimination of 5-TTri and BTri in
biological wastewater treatment can be improved to about
60% by using membrane bioreactors. But even with this
upgraded biological treatment system the 4-TTri concentration
was not reduced (Weiss et al., 2006).
As a measure to prioritize trace pollutants in WWTP effluents
with regard to their putative concentration in further
compartments of the water cycle a so-called water cycle
spreading index (WCSI) has been proposed (Reemtsma et al.,
2006). The WCSI is the ratio of the effluent concentration and
the degree of removal in a WWTP. In previous studies the order
of the three benzotriazoles in terms of increasing WCSI was
5-TTri (16 mg/L), BTri (22 mg/L) and 4-TTri (46 mg/L) (Reemtsma
et al., 2006; Weiss et al., 2006). From the influent and effluent
data of these four WWTP the order would be 5-TTri (1–13 mg/L),
4-TTri (4–27 mg/L, 74 mg/L) and BTri (10–54 mg/L). Accordingly,
BTri would be expected to be the benzotriazole occurring in
highest concentration in surface waters and also in later
compartment of a water cycle, closely followed by 4-TTri. From
its WCSI (lower effluent concentration and better removal)
5-TTri would be the least critical components.
3.2. Polishing of WWTP effluents
If the biological processes in WWTP are not suited to remove
polar pollutants from the wastewater, physico-chemical posttreatment
may be used to avoid discharge of these pollutants
into surface waters and their spreading along partially closed
water cycles.
For benzotriazoles, ozonation has been shown to be
a suitable post-treatment process. An ozone dose of 1 mg/mg
DOC in wastewater was sufficient for complete removal of
BTri and TTri (Weiss et al., 2006). No difference in the reactivity
of the three benzotriazoles towards ozone could be
determined.
3.2.1. Flocculation, sand filtration and powdered
activated carbon
In the present study we investigated whether flocculation
followed by rapid sand filtration, with and without addition of
powdered activated carbon (PAC) can also be used for such
a polishing step.
A flocculation and rapid filtration (F-RF) unit (100–180 L/h)
was operated for more than 2 years on site of the WWTP-R
with different flocculants (Al 3þ 1 mg/L, Fe 3þ 2 mg/L) and
different filter velocities (6–7.5 m/h). Based on experiences in
the deironing step of water treatment plants it was hypothesized
that the biomass establishing in the subsequent (bio-)
filters may remove some of the trace pollutants. The flocculation
with either Al 3þ or Fe 3þ had no effect on the concentration
of any of the benzotriazoles. However, 5-TTri showed
an average removal of 50% in the subsequent rapid filtration
(RF) step (Table S1). During some periods the removal was
almost complete (Fig. 2). Thus, biodegradation of 5-TTri on the
filter column must have been comparatively fast, as the residence
time in the filter was about 20 min, only. These data

600
Fig. 2 – Mean concentration of BTri, 4-TTri and 5-TTri in the
feed water and the effluent of a rapid filtration column
(filled with 0.8 m sand and 1 m anthracite) after flocculation
with 1–2 mg/L Fe 3D (F-RF) with and without addition of PAC
(10–20 mg/L) The error bars indicate standard deviation of
three samplings at independent days.
confirm the much better biodegradability of 5-TTri as
compared to BTri and 4-TTri.
The whole process was much more effective when PAC
was added during flocculation (F-PAC-RF) (Fig. 2, Table S1).
With the addition of 20 mg/L of PAC BTri could be reduced by
81% to filter effluent concentrations around 3 mg/L and 5-TTri
as well as 4-TTri were removed to levels below the LOQ of
0.25 mg/L. A PAC concentration of 10 mg/L proved insufficient
in effluent polishing (Fig. 2). In batch experiments with WWTP
effluent, however, 10 mg/L of PAC were able to reduce the BTri
and TTri concentration by 40 and 80%, respectively, in 24 h.
This indicates that the contact time in the rapid filtration
process was too short to make full use of the sorption capacity
of the PAC.
3.3. Occurrence in surface water
3.3.1. Rhine and Elbe
The two major rivers of Germany and also important European
rivers, Rhine (catchment area around 200,000 km 2 ) and Elbe
(catchment area around 150,000 km 2 ), were analyzed for their
benzotriazole concentrations. BTri was quantified (LOQ ¼
0.05 mg/L) in all but one sample with concentrations ranging
from 0.09 to 0.68 mg/L and 4-TTri in six of the nine samples (0.05–
0.46 mg/L; Table 1). Clear trends are seen along both rivers, such
as the increasing concentration of BTri (from 0.13 to 0.35 mg/L
over the 700 km in the Rhine and from

discharged directly into trenches and creeks. However, the
concentration of benzotriazoles in such surface runoffs has
not been analyzed, yet.
3.3.2. River Havel
Also, water of the river Havel that passes the City of Berlin
from the north to the south was analyzed for a period of
5 months, because it is an important component of the
partially closed water cycle of the city (Gruenheid et al., 2005).
While the upstream samples did not show measurable
concentrations of the triazoles, the downstream samples
contained an average 1.6 mg/L of BTri, 0.34 mg/L of 5-TTri and
2.1 mg/L of 4-TTri (Table 1). Here 4-TTri is the most prominent
benzotriazole, which is consistent with our view on the relative
stability of these three compounds.
These surface water concentrations in river Havel downstream
of the city of Berlin are about a factor of 5–10 higher
than those found in Rhine and Elbe (Table 1). This is due to the
high portion of WWTP effluent in the rivers of Berlin with their
low water flow. Increased surface water concentration of BTri
with increasing wastewater portion has been found also for
different Swiss rivers (Giger et al., 2006).
The widespread occurrence of BTri and 4-TTri in surface
waters may also have consequences for the quality of raw
waters used for drinking water production. Along all three
rivers, Rhine, Elbe and Havel, water is abstracted via bank
filtration for drinking water production.
3.4. Treatment of surface water
3.4.1. Flocculation and rapid filtration
At several locations in Berlin (Fig. S1) surface water is treated
by flocculation with Fe 3þ , sedimentation and rapid filtration to
remove dissolved phosphorous and to avoid eutrophication in
critical areas. Because removal of some trace pollutants, such
as para-toluenesulfonamide, has been recorded in these
surface water treatment plants (Richter et al., 2008) we
investigated whether benzotriazoles could also be removed.
However, no significant removal was obtained for any of the
three compounds in each plant. For BTri and 4-TTri, these
results agreed to those obtained for effluent polishing by F-RF
(see above; Table S1). For 5-TTri, a 50% removal had been
obtained in the RF step of the effluent, presumably by
biodegradation. The lack of removal in the corresponding step
with surface water, at the same contact time of 20 min, may
indicate, that either microorganisms from the activated
sludge may be required for this degradation or that the
concentration of 5-TTri in the surface water (0.3–0.5 mg/L) was
too low to induce this biodegradation process.
3.4.2. (River) bank filtration
River bank filtration is an important source of raw water for
drinking water production and the passage of the littoral and
the underground is an important process to improve and
ensure water quality. This is true on a global scale, for
Germany (16% of bank filtration; Hiscock and Grischek, 2002)
but especially for the City of Berlin that gains more than 50% of
the drinking water for its 3.5 Mio inhabitants from bank
filtrate (Gruenheid et al., 2005).
water research 44 (2010) 596–604 601
The concentrations of the three benzotriazoles were
determined along two bank filtration transects towards the
production wells at two different sites (Lake Tegel and Lake
Wannsee) in Berlin (Fig. S1) At both sites the distance of the
production well from the lakeshore is about 100 m, corresponding
to travel times of the water in the range of
4–5 months (Gruenheid et al., 2005; Massmann et al., 2008).
Surface water in Lake Tegel consists of 15–30% effluent of
WWTP-S (Gruenheid et al., 2005). The water abstracted in the
production well consists of approx. 75% lake water and 25%
land-sided groundwater (inland aquifer), corresponding to
a volumetric contribution of previous wastewater of 11–22%
(Gruenheid et al., 2005). Redox conditions are not stable but
fluctuate between anoxic and aerobic depending on water
temperature and the mode of operation of production wells
(Gruenheid et al., 2005). At the second site the hydrological
situation is largely similar.
For two reasons the concentration data (Fig. 3) are a bit
scattered. (i) Not all observation wells sampled at the same
depth and, thus, redox conditions are not identical. (ii) All
samples were taken within 2 days while the water needs
months to travel along the transects from the lakeshores to
the production wells. Thus, not the same body of water was
sampled along one transect. Long-term investigations at these
two sites have shown that seasonal fluctuations can be
recorded in the groundwater body (Massmann et al., 2008).
Despite these difficulties both transects show an ongoing
continuous decrease in BTri concentration from the lakeshore
(distance 0) to the respective production well (Figs. 3a,d). At
Lake Tegel the lakewater concentration of around 2 mg/L is
diminished to 0.2 mg/L in the production well, which corresponds
to a 90% concentration decrease due to dilution and
removal (Fig. 3a). If one considers the dilution by land-sided
groundwater with no BTri contamination, the removal of BTri
during this underground passage would be 86%. At the bank
filtration site at Lake Wannsee the BTri concentration was
reduced by 75% to 0.3 mg/L in the production well (Fig. 3d). The
concentration gradient of BTri at the Lake Tegel transect is
quite parallel to the DOC gradient (Fig. 3a), which may suggest
a cometabolic degradation of BTri.
Removal of the least stable 5-TTri from surface water was
obviously much faster and, likely, complete (Figs. 3c,f). This is
well visible for the Wannsee site, where the first observation
well was placed right in the colmation zone at the lakeshore.
Here the 5-TTri concentration was already diminished to
0.1 mg/L (Fig. 3f), while the concentrations of 4-TTri was only
slightly reduced as compared to the lake water (Fig. 3e).
However, also 4-TTri was removed significantly along both
transects, but the concentration gradient was less steep
(Figs. 3b,e). At Lake Wannsee, where the 4-TTri concentration
in surface water tends to be higher than in Lake Tegel the
concentration in the production well in 100 m distance
remained at 0.13 mg/L (Fig. 3e).
It is unclear by which mechanism 4-TTri was removed.
For sandy aquifer material a retardation factor of 1.1 has
been experimentally determined (Jia et al., 2007). As sandy
material is predominant also at these bank filtration sites,
a removal by sorption appears unlikely. Rather a very slow
microbial degradation may occur. Slow biodegradation of
trace concentrations of organic contaminants taking place in

602
Fig. 3 – Concentration of (a, d) BTri, (b, e) 4-TTri and (c, f) 5-TTri in bank filtration transects at two sites in Berlin. (left) at Lake
Tegel (04/2008 and 05/2008); DOC concentration also shown in (a); (right) at Lake Wannsee (04/2008). PW: production well.
Shaded area below LOQ.
time periods of months are, however, difficult to verify
experimentally.
Whatever process may be responsible, bank filtration was
the most effective natural process for removal of benzotriazoles
from waters. This may also be true for other poorly
degradable polar substances. However, one has to consider
the duration of this treatment process of several months. The
concentration profiles of the transects also show that these
long residence times in the subsurface are essential to remove
poorly degradable polar pollutants such as BTri and 4-TTri,
while for 5-TTri much shorter residence times would have
been sufficient.
These findings consistently show that neither BTri nor
4-TTri are completely eliminated during bank filtration, even
after a travel time of 4 months (120 days).
3.5. Drinking water treatment
Benzotriazoles are widespread in surface waters and
‘‘natural’’ processes such as biological wastewater treatment,
water research 44 (2010) 596–604
biologically active filters and bank filtration are not sufficient
to remove them completely. Thus, BTri and 4-TTri may occur
in raw waters used for drinking water production and technical
treatment processes should be available to remove such
polar pollutants during drinking water treatment. While the
suitability of ozonation for the removal of BTri and TTri with
moderate dosages of ozone has been shown before (Weiss
et al., 2006; Legube et al., 1987), it was investigated here
whether activated carbon treatment could also be used to
remove trace concentrations of benzotriazoles.
Batch experiments with deionized water were performed
to record isotherms for BTri and TTri (as the sum of both
isomers; Fig. 4). The Freundlich parameters K F ¼ 84 (mg/g)/
(mg/L) n for BTri and 187 (mg/g)/(mg/L) n for TTri and n ¼ 0.44
for BTri and 0.52 for TTri indicate a moderate tendency to
adsorb onto activated carbon. It is stronger for the methylated
benzotriazoles (TTri) than for the non-substituted analogue
(BTri).
Based on these isotherm data one can assume that trace
concentrations of TTri that may occur in raw waters are well

Fig. 4 – Isotherms for the adsorption of BTri and TTri (sum
of both isomers) onto PAC (Norit Super SAE) from deionized
water with a start concentration C 0 [ 100 mg/L (mean of
two experiments).
removable by GAC filtrations, as it has been found for the
pharmaceutical diclofenac with a comparable K F values of 140
(mg/g)/(mg/L) n in a previous study (Ternes et al., 2002). For BTri
with its lower KF value, however, the suitability of GAC filtration
is questionable. Rather breakthrough in GAC filtration
may occur, as found for clofibric acid with a comparable KF
value of 71 (mg/g)/(mg/L) n (Ternes et al., 2002). A final evaluation
of the suitability of GAC for the removal of the triazoles
would have to be performed site-specific, as the sorption
tendency and the sorption kinetics in a certain water may be
influenced by its matrix constituents and its pH.
4. Conclusion
Benzotriazoles are typical examples of polar and poorly
degradable pollutants and they behave characteristically in
partially closed water cycles. Biological treatment is insufficient
for their complete removal from water, as shown for
wastewater treatment, surface waters, bank filtration and
biologically active filters. Owing to its wide use and biological
resistance BTri appears to be one of the most prominent
single trace pollutants in WWTP effluents (mg/L concentrations),
and it is omnipresent also in surface waters. Concentrations
of 4-TTri tend to increase along the large European
river systems Rhine and Elbe, indicating the continuous
discharge of this compound into the rivers and its high
stability also in surface waters. Bank filtration with residence
times of several months in the underground proved the most
effective natural process for removing these polar and poorly
degradable trace pollutants.
But neither BTri nor 4-TTri were removed completely and
both compounds may occur at reduced concentration in raw
waters used for drinking water production. Fortunately, these
aromatic compounds of high electron density can be removed
by ozonation and, at least, TTri also by activated carbon
filtration, either in wastewater effluent polishing or in
drinking water treatment. Thus, an intrusion of BTri and TTri
from wastewater into drinking water in partially closed water
water research 44 (2010) 596–604 603
cycles with indirect potable reuse can be avoided by these
technical treatment steps.
Data presented here clearly show an increasing resistance
to biodegradation from 5-TTri over BTri to 4-TTri. Thus, the
widespread contamination of the water cycle with benzotriazoles
could be avoided to a large extent if solely 5-TTri
would be used instead of BTri and of the technical mixture of
5-TTri and 4-TTri.
Acknowledgements
Funding of this research by Berliner Wasserbetriebe (project
‘‘Barrieren’’) is gratefully acknowledged. We are grateful to
Jutta Jakobs for assistance in the laboratory and to Ina Engel
for performing the activated carbon study. The river water
samples were received from the Department of Hydrogeology
of the Free University of Berlin.
references
Bentayeb, K., Batlle, R., Romero, J., Nerin, C., 2007. UPLC-MS as
a powerful technique for screening the nonvolatile
contaminants in recycled PET. Anal. Bioanal. Chem. 388,
1031–1038.
Bergthaller, P., 2002. Couplers in color photography – chemistry
and function – Part 3. Imaging Sci. J. 50, 233–276.
Cancilla, D.A., Martinez, J., Van Aggelen, G.C., 1998. Detection of
aircraft deicing/antiicing fluids in perched water monitoring well
at an international airport. Environ. Sci. Technol. 32, 3834–3835.
Cancilla, D.A., Baird, J.C., Rosa, R., 2003. Detection of aircraft
deicing additives in groundwater and soil samples from
Fairchild Air Force base, a small to moderate user of deicing
fluids. Bull. Environ. Contam. Toxicol. 70, 868–875.
Chadwick, D., Hashemi, T., 1978. Adsorbed corrosion inhibitors
studied by electron spectroscopy: benzotriazoles on copper
and copper alloys. Corros. Sci. 18, 39–51.
Corsi, S.R., Zitomer, D.H., Field, J.A., Cancilla, D.A., 2003.
Nonylphenol ethoxylates and other additives in aircraft
deicers, antiicers and waters receiving airport runoff. Environ.
Sci. Technol. 37, 4031–4037.
De Orsi, D., Gianni, G., Gogliardi, L., Porra, R., Berri, S., Bolasco, A.,
Carpani, I., Tonelli, D., 2006. Simple extraction and HPLC
determination of UV-A and UV-B filters in sunscreen products.
Chromatographia 64, 509–515.
European Commission, 2000. Twenty-Fourth Commission
Directive 2000/6/EC Adapting to Technical Progress Annexes
II, III, VI and VII to Council Directive 76/768/EEC on the
Approximation of the Laws of the Member States Relating to
Cosmetic Products.
European Commission, 2007. Commission Regulation (EC) No.
1451/2007 of Directive 98/8/EC Concerning the Placing of
Biocidal Products on the Market.
Giger, W., Schaffner, C., Kohler, H.-P.E., 2006. Benzotriazole and
tolyltriazole as aquatic contaminants. 1. Input and occurrence
in rivers and lakes. Environ. Sci. Technol. 40, 7186–7192.
Gruden, C.L., Dow, S.M., Hernandez, M.T., 2001. Fate and toxicity
of aircraft deicing fluids additives through anaerobic
digestion. Water Environ. Res. 73, 72–79.
Gruenheid, S., Amy, G., Jekel, M., 2005. Removal of bulk dissolved
organic carbon (DOC) and trace organic compounds by bank
filtration and artificial recharge. Water Res. 39, 3219–3228.

604
Harris, C.A., Routledge, E.J., Schaffner, C., Brian, J.V., Giger, W.,
Sumpter, J.P., 2007. Benzotriazole is antiestrogenic in vitro but
not in vivo. Environ. Toxicol. Chem. 26, 2367–2372.
Hart, D.S., Davis, L.C., Erickson, L.E., Callender, T.M., 2004.
Sorption and partitioning of benzotriazole compounds.
Microchem. J. 77, 9–17.
Hiscock, K.M., Grischek, T., 2002. Attenuation of groundwater
pollution by bank filtration. J. Hydrol. 266, 139–144.
Hollingsworth, J., Sierra-Alvarez, R., Zhou, M., Ogden, K.L., Field, J.A.,
2005. Anaerobic biodegradability and methanogenic toxicity of
key constituents in copper chemical mechanical planarization
effluent of the semiconductor industry. Chemosphere 59,
1219–1228.
Jia, Y., Breedveld, G.D., Aagaard, P., 2007. Column studies on
transport of deicing additive benzotriazole in a sandy aquifer
and a zerovalent iron barrier. Chemosphere 69, 1409–1418.
Legube, B., Guyon, S., Dore, M., 1987. Ozonation of aqueous
solutions of nitrogen heterocyclic compounds: benzotriazoles,
atrazine and amitrole. Ozone Sci. Eng. 9, 233–246.
Loos, R., Gawlik, B.M., Locoro, G., Rimaviciute, E., Contini, S.,
Bidoglio, G., 2009. EU-wide survey of polar organic persistent
pollutants in European river waters. Environ. Pollut. 157,
561–568.
Massmann, G., Sueltenfuss, J., Duennbier, U., Knappe, A.,
Taute, T., Pekdeger, A., 2008. Investigation of groundwater
residence times during bank filtration in Berlin: a multi-tracer
approach. Hydrol. Process 22, 788–801.
Ort, C., Schaffner, C., Giger, W., Gujer, W., 2005. Modeling
stochastic load variations in a sewer system. Water Sci.
Technol. 52, 113–122.
water research 44 (2010) 596–604
Pillard, D.A., Cornell, J.S., Dufresne, D.L., Hernandez, M.T., 2001.
Toxicity of benzotriazole and benzotriazole derivatives to
three aquatic species. Water Res. 35, 557–560.
Reemtsma, T., Weiss, S., Mueller, J., Petrovic, M., Gonzalez, S.,
Barcelo, D., Ventura, F., Knepper, T.P., 2006. Polar pollutant
entry into the water cycle by municipal wastewater:
a European perspective. Environ. Sci. Technol. 40, 5451–5458.
Richter, D., Massmann, G., Duennbier, U., 2008. Identification and
significance of sulphonamides (p-TSA, o-TSA, BSA) in an
urban water cycle (Berlin, Germany). Water Res. 42, 1368–1378.
Ternes, T.A., Meisenheimer, M., McDowell, D., Sacher, F.,
Brauch, H.-J., Haist-Gulde, B., Preuss, G., Wilme, U., Zulei-
Seibert, N., 2002. Removal of pharmaceuticals during drinking
water treatment. Environ. Sci. Technol. 36, 3855–3863.
Verheyen, V., Cruickshank, A., Wild, K., Heaven, M.W., McGee, R.,
Watkins, M., Nash, D., 2009. Soluble, semivolatile phenol and
nitrogen compounds in milk-processing wastewaters. J. Dairy
Sci. 92, 3484–3493.
Voutsa, D., Hartmann, P., Schaffner, C., Giger, W., 2006.
Benzotriazoles, alkylphenols and bisphenol A in municipal
wastewaters and in the Glatt River, Switzerland. Environ. Sci.
Pollut. Res. 13, 333–341.
Weiss, S., Reemtsma, T., 2005. Determination of benzotriazoles
corrosion inhibitors from aqueous environmental samples by
liquid chromatography-electrospray ionization-tandem mass
spectrometry. Anal. Chem. 77, 7415–7420.
Weiss, S., Jakobs, J., Reemtsma, T., 2006. Discharge of three
benzotriazole corrosion inhibitors with municipal wastewater
and improvements by membrane bioreactor treatment and
ozonation. Environ. Sci. Technol. 40, 7193–7199.

Determining the fraction of pharmaceutical residues
in wastewater originating from a hospital
Christoph Ort a, *, Michael G. Lawrence a , Julien Reungoat a , Geoff Eaglesham b ,
Steve Carter b , Jurg Keller a
a
The University of Queensland, Advanced Water Management Centre (AWMC), Brisbane, QLD 4072, Australia
b
Queensland Health Forensic and Scientific Services, Organics Laboratory, QLD 4108, Australia
article info
Article history:
Received 14 May 2009
Received in revised form
21 July 2009
Accepted 1 August 2009
Available online 5 August 2009
Keywords:
Experimental design
Audit data
Quantification
Prediction
Loads
Consumption
1. Introduction
1.1. Brief overview
abstract
Hospital wastewater (HWW) is normally discharged directly,
without pre-treatment, to sewers. Despite mostly being only
water research 44 (2010) 605–615
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
* Corresponding author. Tel.: þ61 7 3365 4730; fax: þ61 7 3365 4726.
E-mail address: c.ort@awmc.uq.edu.au (C. Ort).
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.watres.2009.08.002
Pharmaceutical residues in water are frequently analysed and discussed in connection with
sewage treatment, ecotoxicity and, natural and drinking water quality. Among different
localities hospitals are suspected, or implied, to be a major and highly variable source of
pharmaceuticals that substantially contribute to the total wastewater load. In this study, the
contribution of pharmaceuticals from a hospital to a sewage treatment plant (STP) serving
around 45,000 inhabitants was evaluated. Approximately 200 hospital beds result in a hospital
bed density of 4.4 beds per 1000 inhabitants, which is a typical value for developed world
countries. Prior to sampling, a sound systems analysis was performed, and a sophisticated
continuous flow-proportional sampling regime was applied. Hence, overall experimental
uncertainty was reduced to a minimum, and measurements provide clear evidence that, for 28
of 59 investigated substances, over 85% of the pharmaceutical residue loads do not originate
from the hospital when applying a conservative error estimation. Only for 2 substances,
trimethoprim (18%) and roxithromycin (56%), was the maximum observed contribution of the
hospital >15%. On average, the contribution of the hospital for the compounds detected in
both, hospital effluent and sewage treatment plant influent was small and fairly constant. Five
compounds were only detected in hospital wastewater, and 24 neither in the hospital wastewater
nor in the total wastewater at the influent of the STP. For these compounds no experimental
contribution could be calculated. For the compounds where audit data for both the
national consumption and the specific hospital under investigation were available, a prediction
of the fraction of pharmaceuticals originating from the hospital was performed. Three
quarters of the compounds, classified with the existing audit data, were in the same ‘‘hospital
contribution category’’ as determined by measurements. For most of the other compounds,
plausible reasons could be identified to explain the observed deviations.
ª 2009 Elsevier Ltd. All rights reserved.
a small fraction of the total wastewater volume in the influent
of a sewage treatment plant (STP), HWW has gained increasing
scientific and public attention in the last decade. This is, in part
due to the observation and expectation that HWW is a source
for undesirable constituents, such as (multi-)antibiotic-resistant
bacteria (Baquero et al., 2008; Kummerer, 2004). In other

606
publications, the emission from hospitals was estimated for
antibiotics, anaesthetics, disinfectants, heavy metals, AOX
(Adsorbable Organic Halogens), iodised X-ray contrast media
and cytostatic agents (e.g. Kummerer, 2001). The latter were
also investigated in detail by Lenz et al. (2007). Furthermore,
a number of toxicity assays were performed (Boillot et al., 2008;
Ferk et al., 2009; Hartmann et al., 1998). As a result, it has been
suggested in some studies that pre-treatment of HWW prior to
discharge into the sewers provides a reasonable solution
(Gautam et al., 2007; Lenz et al., 2007; Pauwels and Verstraete,
2006). However, this view is not unanimously supported. The
separate treatment of HWW to reduce the development of
resistant bacteria was questioned (Kummerer, 2009): the
substantial amount of antibiotics used outside of hospitals (in
Germany more than 75%) seems to be a plausible reason that
resistant bacteria are also abundant in wastewater not
receiving any HWW. Additionally, Boillot et al. (2008) found
quantitatively far fewer microorganisms in the effluents of
hospitals than in urban wastewaters which is consistent with
other studies. With regard to pharmaceuticals, Lenz et al.
(2007) report that 1) for some pharmaceuticals merely a small
fraction of the amounts administered in the hospital were
actually found in its effluent (i.e. 0.1–0.2% for doxorubicin, 0.5–
4.5% for 5-fluorouracil and 27–34% for total platinum) and 2)
a complete onsite wastewater treatment process is needed to
significantly remove targeted pharmaceuticals. This includes
full physical and biological treatment steps, not only advanced
processes. Capturing all sources within a hospital (wards,
laboratories) may be further complicated by the fact that
different facilities discharge through different pipes to the
common sewer. This particularly holds true for large existing
hospital complexes.
Therefore, local circumstances need to be considered and
the contribution of an individual hospital needs to be assessed
in relation to the total load in a STP catchment. To our
knowledge, only a few publications explicitly quantify pharmaceutical
residues (subsequently referred to as ‘pharmaceuticals’)
excreted within hospitals compared to the total
pharmaceutical load in the corresponding STP influents
(Feldmann et al., 2008; Heberer and Feldmann, 2005; Thomas
et al., 2007). However, these studies are limited to a small
number of pharmaceuticals, or make an assumption on the
water flow instead of measuring the wastewater flow onsite to
determine actual loads.
In view of the local situation in South East Queensland
where it is proposed to recycle wastewater for indirect potable
reuse, it is sensible to consider whether pre-treatment of
HWW will provide a significant benefit. From two previous
research papers relevant for the region of interest also dealing
with pharmaceuticals the contribution of hospitals cannot be
derived (Khan and Ongerth, 2004; Watkinson et al., 2009).
Therefore, the goal of our study is to determine accurately
the contribution of a hospital to the total pharmaceutical load
found at the inlet of the corresponding STP by means of
measurements. Additionally, this experimentally data
obtained from a limited time period is then compared with
readily available audit data. It shall be assessed whether the
contribution of a hospital can be predicted reliably without
any additional administrative effort, i.e. without extra surveys
on the hospital wards for day-specific consumptions. If
water research 44 (2010) 605–615
measurements matched with the prediction, the same kind
(comprehensiveness and quality) of information can be used
at other locations to make a prediction, a priori without
laborious measurements.
The focus of this research is on dissolved pollutants which
cannot be eliminated in conventional wastewater treatment.
Pollutants showing poor to moderate biological removal need
to be transformed by chemical reactions (e.g. oxidation) or
separated by physical processes (e.g. adsorption onto activated
carbon).
1.2. Systems analysis
The prediction and experimental quantification of pharmaceutical
mass fluxes in the wastewater of a specific STP catchment
are laborious. A sound understanding of the whole system is
required prior to setting up a predictive model, and performing
a confirmative sampling campaign. This particularly holds true
when attempting to attribute different fractions to a multitude
of individual sources, for example if there are several hospitals
and multiple smaller health care facilities in a catchment. Due to
the lack of generally accessible consumption data at sufficiently
high spatial and temporal resolution, models often provide only
a prediction of an average load. Additionally, the latter is prone
to uncertainty due to varying transformations of pharmaceuticals
during human metabolism.
While it would be ideal to have a list of all health care
facilities with size, services provided and precise pharmaceutical
consumption, just obtaining generally available
consumption data is a tedious task in itself. The ‘‘institutional
resolution’’ is often not sufficient without additional administrative
effort, i.e. temporary surveys of the wards in the
hospital(s) under investigation (Feldmann et al., 2008; Kummerer,
2001). Furthermore, the (average) household pharmaceutical
consumption needs to be estimated from national or
state-wide sales and/or prescription data if regional data is not
available.
Moreover, collecting representative samples requires
a thorough knowledge of the sewer layout and awareness of
potentially highly variable concentrations and loads in the
course of a day. Clearly, accurate chemical analysis of a nonrepresentative
sample is not adequate to characterise a real
full-scale system.
1.3. Sampling issues
Accurately quantifying pharmaceutical loads in hospital effluents
or sewers close to any source (sub-catchments, households
or industry) is a demanding undertaking. It requires a substantial
experimental effort and is still prone to uncertainties. The
latter are extremely hard to quantify if sampling is carried out
with conventional (unsophisticated) devices, i.e. auto-samplers
operated in a discrete sampling mode with (too) long time
intervals, or grab samples. Rarely are fluctuations of concentrations
and loads assessed in separate experiments at high
temporal resolution prior to the ‘‘real’’ measuring campaigns.
These pre-experiments are very expensive and may not provide
the data to answer the actual research question. However, if the
applied sampling protocol does not result in the collection of
a representative sample, then the care taken in the following

processes of transport, storage, preparation and chemical
analyses with a sophisticated method cannot make up for this
deficiency (de Gruijter et al., 2006). Subsequent (even sophisticated)
statistical analyses of non-representative samples are
unreliable and the resulting conclusions will therefore be of
limited value. In some cases, the large variation observed in
previous studies may not be ‘‘true natural variation’’ but
instead, may simply be an artefact caused by inadequate
sampling (Ort et al., in preparation).
Therefore, strong emphasis has been put on obtaining
representative samples for this study. In Ort and Gujer (2006)
a method was presented to estimate the required sampling
frequency in order to not exceed a certain sampling error. In
gravity sewers this results in fairly short time intervals if the
substance of interest is contained in a small number of
‘‘wastewater pulses’’ per day (e.g. toilet flushes containing
a specific excreted pharmaceutically active compound).
Sampling frequencies that are too low result in large sampling
uncertainties, especially in the case of only a few patients per
day (Weissbrodt et al., 2009). The often claimed problem of
‘‘limited storage capacity in an auto-sampler’’ can be easily
solved by replacing the glass bottles more than once per day.
This may be more laborious, but it is a much better solution than
using a time-proportional sampling mode, which does not take
samples weightedaccording to the flow inthe sewer. In contrast,
physical boundary conditions such as deep sewers resulting in
long dead times for purging the sampling hose or limited access
to pressurised sewers are more difficult to overcome.
2. Material and methods
2.1. Sewage treatment plant and catchment
characteristics
A total of approximately 45,000 inhabitants in two geographically
separated sub-catchments, Morayfield and Caboolture,
are connected to the South Caboolture STP (subsequently only
referred to as STP) which is operated with two sequencing
batch reactors (SBRs). It treats a daily dry weather flow of
approximately 10,000 m 3 . During long dry periods with high
level water restrictions, this value can drop to 7500 m 3 day 1 .
Morayfield is drained by gravity sewers and contributes
two thirds of the total wastewater. It is only pumped once, at
the STP itself. Caboolture makes up for one third of the total
influent and is a largely pressurised sewer system with
numerous pumping stations. At specific times of the day the
flow is diverted at the influent of the STP and stored in two
large buffer tanks (800 m 3 each) before being pumped to the
SBRs. This combination of sewers and the complex influent
layout of the STP results in very high hydraulic fluctuations
(see Fig. 1). Hours with almost zero flow contrast with hours
around 250–300 L s 1 and in between, the flow varies rapidly
and significantly. During wet weather the relative flow variations
are less significant due to higher base flow.
2.2. Hospital characteristics
Caboolture Public Hospital has 190 beds and offers all services
of a modern regional hospital (listed in Table SI 1, see
water research 44 (2010) 605–615 607
supporting information). A small private hospital providing
mainly day surgery (only around 10 beds) and a small dental
surgery also drain into the same sewer. The wastewater from
the private hospital cannot be accessed separately. Other
small health care facilities within this sewer catchment make
consultations to out-patients, and therefore, the wastewater
from these facilities are not expected to significantly add to
the pharmaceutical load of the STP. The hospital bed density
for the whole STP catchment is 4.4 beds per 1000 inhabitants.
All HWW is collected in a sewage pumping station (SPS CT-51,
subsequently referred to as SPS) before being pumped to the
primary rising main. There is no residential wastewater
contributing to this SPS and the hydraulic residence time in
the main sewer to the STP is approximately 30 min to 1 h
(hydraulic calculations provided by the Regional Council for
the decisive time in the morning when samples at the SPS and
the STP needed to be coordinated). The average daily volume
during dry periods pumped at the SPS is approximately 75 m 3
which is 1% of the total wastewater volume discharged to the
STP. The occupancy of hospital beds in Caboolture during the
sampling period was close to 100% which is representative for
the year to date average.
Unfortunately no comprehensive database exists with
regard to other health care facilities in the catchment of the
STP. Hence, an internet search was performed. Four aged care
facilities with a total capacity of 443 beds were found (297 high
care and 146 low care) with an unknown occupancy rate.
Furthermore, a total of 14 addresses for doctors plus 12
dentists were found. If mass fluxes at the influent of the STP
were significantly higher than expected from average national
consumption and hospital usage, further investigations of
these facilities would be warranted.
2.3. Sampling
Continuous flow-proportional sampling modes were applied
in this study to minimise sampling error. Continuously
diverting a small flow-proportional side stream is conceptually
the best solution to obtain representative samples for
dissolved compounds. However, low velocities in the side
stream prevent proper sampling of solids and long-term
operation may lead to biofilm growth. Due to the limited time
of sampling biofilm growth is not considered problematic in
this instance.
Sampling over consecutive days was preferred to the
alternative option of collecting samples on single days
distributed over a longer period. This drastically reduces the
effect of unknown system behaviour: missing a ‘‘decisive’’
HWW packet at the STP is then limited to the first hour of the
first day and the last hour of the last day. All other water
packets are captured, although they might be attributed to the
STP sample a day later. However, this would merely lead to
higher variability of the hospital’s contribution and not to
a non-quantifiable effect.
2.3.1. Sampling protocol for Caboolture Hospital (SPS CT-51)
The HWW is not easily accessible before it enters the SPS.
Furthermore it would have been very difficult to set up an
accurate flow meter to measure flow in a small open channel
with intermittent, partially very low flows and to use this data

608
Flow [L s -1 ]
500
400
300
200
100
0
Dry weather flow (cv = 0.72)
Wet weather flow (cv = 0.44)
8am 10am 12pm 2pm 4pm 6pm 8pm 10pm 12am 2am 4am 6am 8am
to control the speed of the sampling pump. Instead, plumbers
from the Regional Council fitted a tap in the rising main of the
SPS (see Fig. 2). The tap is upstream of the non-return valve
before the HWW enters the primary rising main leading to the
STP. Electricians from the Regional Council installed an
actuator after the tap which only opens when the pump of the
SPS empties the wet well. Water runs without a sampling
pump due to the pressure in the rising main. Under normal
operating conditions there are about 24 pumping cycles per
day, triggered automatically based on to the water level in the
SPS. While it was found that the flow during one cycle is fairly
constant, it can vary significantly among cycles due to variable
hydraulic conditions in the primary rising main. Therefore,
a manual operating mode was adopted, disabling the auto
level control. This allowed for using the full storage capacity of
the wet well. Starting at 7 AM it was emptied again at 12 PM, 6
PM and 7 AM the following day which required personnel to be
present three times per day (confined space). The pump
Time [h]
Fig. 1 – Two examples for typical flow patterns at the influent of the sewage treatment plant; cv [ coefficient of variation
(standard deviation/mean).
wet well
SPS CT-51
water research 44 (2010) 605–615
pump
control
unit
operates at about 2500 L min 1 and the sampling side stream
was adjusted with the tap to approximately 1 L min 1 ,
resulting in a sampling volume of about 10 L per pump cycle.
In comparison, the dead volume of the tap installation
including hose was 0.5 L (ca. 5% of the sampling volume).
The three samples were collected in separate glass bottles,
and analysed separately. The concentrations of the individual
samples were multiplied with the flow for the corresponding
pump cycle, and summed to obtain a 24-h load. Rough diurnal
variations could also be determined with this sampling
procedure, but they are not relevant for the system and time
scales under investigation, and hence they are not further
discussed in this paper.
2.3.2. Sampling protocol at the sewage treatment plant
To sample for the same ‘‘water packets’’ as at the SPS, sampling
started at 7:45 AM in the influent of the STP. The storage tanks
start filling at 8 AM and are emptied completely during night
dry pit
hose
actuator
tap
non-return
valve
effluent Caboolture Hospital
sample
bottle
to STP
Fig. 2 – Schematic drawing of the sampling point at the sewage pumping station (SPS) CT-51 (not to scale): All hospital
wastewater is discharged to the wet well of the SPS and intermittently pumped to the primary rising main leading to the
sewage treatment plant (STP). Upstream of the non-return valve a stand pipe with a tap and an actuator was fitted. This
allows for taking flow-proportional samples during individual pump cycles.
primary rising main

time, and in the early morning hours. This guarantees that
wastewater is not stored and dragged on over different 24-h
sampling periods. Flow rates in the influent are routinely
measured at high temporal resolution. A wire connected to an
analogue digital converter provides a 4–20 mA signal from the
PLC (programmable logic controller) linear to the flow in the
sewer to control the speed of the sampling pump. The peristaltic
pump (Watson Marlow 520UN, programmable interface, water
proof casing, equipped with a 520R2 pump head and 3.2 mm
tube bore) was tested in the lab to ensure its linear behaviour
over the full speed range under similar physical boundary
conditions (suction height approximately 2 m, pressure height
negligible). The pump speed was set to 0 rpm (revolutions per
minute) for 0 L s 1 in the sewer (pumping 0 mL min 1 )andto
34 rpm for 1000 L s 1 (pumping 69.4 mL min 1 ). The finest
increment of the pump is 0.1 rpm equivalent to 2.9 L s 1
wastewater flow in the influent of the STP. With this setup
approximately 15 L of wastewater were collected in a 20 L glass
bottle which was located in a refrigerated container. Two field
blanks were collected: to this end 0.5 L of MilliQ water was used
to rinse the sampling tube and subsequently 0.5 L MilliQ water
was pumped through the tube to be analysed in the laboratory.
No substances were detected above the limit of quantification.
2.4. Chemical analyses
After collection, the continuously refrigerated samples were
transported to the laboratory where they were filtered the same
day and preserved before analysis. All samples were analysed
for 59 substances by Queensland Health Forensic and Scientific
Services (QHFSS). A detailed description of the method consisting
of solid phase extraction followed by concentration prior
to quantification by LC–MS/MS (liquid chromatography coupled
with tandem mass spectrometry) is given in the supplementary
information SI 2, accompanied with an alphabetical list of all
compounds (Tables SI 2.2 and SI 2.3).
As the method does not allow for correction of absolute
analytical extraction recoveries in raw wastewater samples,
we report relative loads. In order to compare hospital effluent
samples with samples from the influent of the STP, it is
necessary to assume that matrix effects between these
sample types are similar. Any systematic error in recovery is
therefore cancelled out when calculating ratios of loads, i.e.
contribution of HWW to the total influent of the STP.
2.5. Uncertainty assessment
Flows in completely filled pressurised pipes can be measured
more accurately than flows in open water channels (gravity
flow). For this study a maximum error of 10% was assumed,
which equals to 6% (¼10/3 0.5 ) as single standard deviation of
a normal distribution. For chemical analysis a random uncertainty
(reproducibility) of 20% for all compounds was chosen
(see Tables SI 2.1–2.3). The two errors are independent, and
Gaussian error propagation results in an overall uncertainty
estimate for calculated loads of 21% (¼[6 2 þ 20 2 ] 0.5 ).
The flow-proportional continuous sampling procedure
covers all fluctuations in the wastewater over time. Since it is
a reasonable assumption that dissolved compounds are
completely mixed over the whole pipe cross section in the
water research 44 (2010) 605–615 609
influent works, no additional errors need to be taken into
account due to sampling.
2.6. Pharmaceutical audit data
2.6.1. National consumption
An extract from the DUSC database (Drug Utilisation Sub-
Committee) for the year 2008 is listed for the compounds
investigated in this study (see supporting information, Table SI
3). It comprises the sum of subsidised drugs (subsidised under
the Pharmaceutical Benefits Scheme (PBS) and the Repatriation
Pharmaceutical Benefits Scheme (RPBS) and processed by
Medicare Australia) and non-subsidised drugs (under PBS copayment
and private prescriptions). The amounts of non-subsidised
drugs were estimated from continuous data on all
prescriptions dispensed from a validated sample of 370
community based pharmacies. The available data do not
include drugs dispensed to public hospital in-patients, pharmacy
over-the-counter drugs (i.e. non-prescription) and drugs
supplied by supermarkets.
2.6.2. Amounts administered to in-patients in Caboolture
Public Hospital
No specific survey was carried out during the sampling period on
the wards. Routinely stored audit data for a current 12-month
period (2007–2008) was made available by the pharmacy of the
Caboolture Public Hospital. For each pharmaceutical, a specific
database query was performed to derive the amounts exclusively
used for hospitalised in-patients; pharmaceuticals given
to out-patients (in consultations and pharmacy) were not
considered, since they will be taken and excreted at home. The
total annual hospital load was determined after summing the
contributions of all medications containing the pharmaceutically
active compound of interest.
It has to be noted that the amounts derived from this database
are amounts supplied by the pharmacy to the individual
wards and not the amounts effectively administered. However,
it is generally not the hospital’s policy to discard drugs to the
(solid or liquid) waste system, both from a financial and environmental
point of view. Nevertheless, some unused drugs for
in-patients may be collected on the wards and returned to the
pharmacy for reuse or proper disposal. Hence, these drugs do
not contribute to the load in the HWW. However, in discussion
with relevant hospital staff these amounts are considered to be
very limited and are not assessed within this study.
3. Results and discussion
3.1. Evaluation of wastewater volumes
The four consecutive weekdays, mid-February 2009 when
sampling took place, were during a wet period, with flows 1.5–
2 times higher than normal dry weather flow (i.e. surface
runoff in catchments and infiltration to sewage pumping
stations). In Table 1 the flows at the two sampling locations
over the corresponding 24-h periods are summarised. During
the sampling period, the hospital contributed less than 1% of
the total wastewater flow to the STP.

610
Table 1 – Wastewater volumes over 24 h at the SPS CT-51 (hospital wastewater) and the influent to the STP.
Influent STP [m 3 ] Hospital wastewater
(flow at SPS)
7:45 AM – 7:45 AM of the following day 7 AM–7 AM of the following
day [m 3 ]
3.2. Evaluation of relative pharmaceutical loads
To obtain relative pharmaceutical loads, measured concentrations
were multiplied with the corresponding 24-h flow at
each sampling location and normalised by the highest STP
influent load. Four examples representing four different
groups of pharmaceuticals are charted in Fig. 3. Absolute
concentration values are not reported because they are difficult
to compare among different studies; they highly depend
on the sewer system (separate or combined) and on the
hydraulic conditions (dry or wet weather flow). The key figures
chosen for statistical evaluation are presented in Table 2, and
discussed subsequently in detail for one example (atenolol,
a beta-blocker, see also Fig. 3A).
The numbers in black circles ( ) refer to the corresponding
column in Table 2:
Concentration values for atenolol in the influent of the
STP were, on average, 10 times higher than the limit of
quantification (LOQ).
Fraction of influent STP [%]
Day 1 16/2/09 14,064 109 0.8
Day 2 17/2/09 16,921 129 0.8
Day 3 18/2/09 19,059 138 0.7
Day 4 19/2/09 14,347 127 0.9
A
Loads (normalised to
max. STP influent load)
Atenolol
1-3%
1-2%
2-4%
1-2%
fraction of STP influent
C Paracetamol
D
Loads (normalised to
max. STP influent load)
1.2
1.0
0.8
0.6
0.4
0.2
0.0
1.2
1.0
0.8
0.6
0.4
0.2
0.0
day 1
day 1
day 2
day 3
STP influent
day 2
day 3
STP influent
day 4
day 4
day 1
day 1
water research 44 (2010) 605–615
Hospital effluent
day 2
day 3
day 4
Hospital effluent
day 2
day 3
day 4
3-7%
3-7%
3-8%
4-10%
fraction of STP influent
B
The concentrations in the hospital effluent were on
average 2 times higher than in the STP influent.
The STP influent loads show only little day-to-day variation
(cv ¼ 0.06, cv ¼ coefficient of variation ¼ standard
deviation/mean). Day-to-day variation is smaller than
the estimated overall uncertainty.
The loads in the hospital effluent varied more (cv ¼ 0.27).
On average the hospital contributed only 1.8% to the
total atenolol load in the influent of the STP.
For a conservative error estimation, a maximum
contribution of the hospital was calculated by dividing
the upper uncertainty value of the hospital effluent by
the lower uncertainty value of the STP influent for each
day (see Fig. 3). Over all four days, the highest maximum
contribution for atenolol was 3.5%.
Over all four days, the smallest minimum contribution
for atenolol was 0.9% (analogue procedure as in ).
The prediction for an average contribution of the
hospital based on audit data is 0.6% (see more details in
Section 3.4).
0.8
0.6
0.4
0.2
0.0
Gabapentin
1.20 1.2
1.20
Hospital effluent
0.03
1.0
0.04
0.02
0.01
0.00
Trimethoprim
1-2%
1-2%
1-3%
3-7%
fraction of STP influent
1.20 1.2
1.20
Hospital effluent
0.06
0.15
1.0
0.04
0.02
0.00
Loads (normalised to
max. STP influent load)
Loads (normalised to
max. STP influent load)
0.8
0.6
0.4
0.2
0.0
day 1
day 1
day 2
day 2
day 3
STP influent
day 3
STP influent
day 4
day 4
day 1
day 1
day 2
day 2
day 3
day 3
day 4
day 4
8-18%
6-15%
6-13%
7-16%
fraction of STP influent
Fig. 3 – Measured, relative pharmaceutical loads over 24-h periods in the influent of the STP and effluent of the hospital for
four consecutive weekdays. Error bars include uncertainty of flow measurements (±6%) and chemical analysis (±20%),
resulting in an overall uncertainty of ±21% (single standard deviation). Note the different scales for the y-axis of STP influent
and hospital effluent.
0.03
0.02
0.01
0.00
0.10
0.05
0.00

Table 2 – Classification of substances according to the contribution of the hospital to the total load in the influent of the STP
(see Section 3.3 for more explanations of key figures marked with black circles ). LOQs for all compounds are between 0.1
and 2 mgL L1 .
Classification Substance Therapeutic
group f
CSTP
LOQ
C Hospital
CSTP
Coefficient of variation
for loads
Influent
STP
Contribution of hospital
wastewater [% of total STP influent]
Measured average
predicted
with audit
data d
Hospital Min Mean Max
Max 5% Atenolol BB 10.7 2.0 0.06 0.27 0.9 1.8 3.5 0.6
Atorvastatin HL

612
Table 3 – Comparison with other hospital wastewater studies.
Number of hospitals Number of beds
per 1000 inhabitants
Classification of all substances according to maximum
observed contribution from the hospital ( ).
The consistent results for atenolol are reflected across
most of the 30 detected substances. Representatives of other
pharmaceutical groups show also fairly constant loads over
the four-day period: gabapentin (an anticonvulsant), paracetamol
(an analgesic) and trimethoprim (an antibiotic, see
Fig. 3B–D).
From the 59 substances, 5 were detected only in the HWW
but not in the influent of the STP and 24 substances were not
detected above the LOQ in any of the samples. The 30
substances detected at both locations were classified for the
hospital’s contribution to the total influent of the STP. To this
end, the maximum observed contribution including uncertainty
as a conservative estimate was used (see description
before in ). The hospital’s contribution for 17 substances was
at all times ‘‘smaller than 5%’’, 11 additional substances fall in
the category ‘‘smaller than 15%’’ and only 2 substances were
‘‘above 15%’’ (trimethoprim and roxithromycin with a worst
case estimate of 18% and 56% respectively). For most
substances quantified in both STP influent and hospital
effluent, the variations of the loads in the HWW were on
average 2.4 times higher than in the influent to the STP. The
small number of hospital patients compared to the potentially
large number of individuals taking these pharmaceuticals at
home is a valid explanation for this observed difference in
variation.
water research 44 (2010) 605–615
Investigated substances
(% in influent of the corresponding
STP originating from hospitals)
2 4.4 diclofenac (1.4) a
ibuprofen (0.7) a
metoprolol (1.5) a
paracetamol (12) a
tetracycline (0.5) a
trimethoprim (14) a
1 3.6 5 X-ray contrast
media (50) b
cytostatics (max 7.5) b
More than 5 12.1 carbamazepine (15) c
diclofenac (10) c
More than 5 12.1 metamizol (50) c
Four out of the 5 substances only detected in the HWW
were just above the LOQ. With the 100 fold dilution in the
influent of the STP the LOQ would have to be at least three
orders of magnitude lower to reliably quantify the hospital’s
(high) contribution. When assuming that the concentrations
in the influent of the STP were equivalent to the corresponding
LOQ, only a one-sided estimation with regard to the hospital’s
contributions from >5% up to >50% can be made.
However, in some cases this deviates from the prediction
based on audit data (see chapter 3.3).
3.3. Comparison with audit data
Location of study
Oslo, Norway (Thomas et al., 2007)
Winterthur, Switzerland (Weissbrodt et al., 2009)
Berlin, Germany (Heberer and Feldmann, 2005)
Berlin, Germany (Feldmann et al., 2008)
a Concentrations measured over 12 weeks, loads estimated with water consumption, sum of the two major hospitals (an unknown number of
other smaller hospitals/health care facilities are located in the catchment).
b Influent STP was not measured in this study, percentage refers to loads of pharmaceuticals quantified in the hospital’s effluent compared to
day-specific administered amounts.
c Only measured in the effluent of one hospital and then extrapolated for the whole catchment based on audit data of the other hospitals.
If the consumption of pharmaceuticals in a STP catchment
can be estimated from existing national sales or prescription
data, and audit data for the hospital are available, the
contribution of the hospital can be calculated with the
following equation
ConsCab:Hosp:$excretion ratio
contributionðhospitalÞ ¼
ConsCab:Pop:$excretion ratio þ ConsCab:Hosp:$excretion ratio
measured loadðhospitalÞ$recovery$accuracy
y
measured loadðSTP catchmentÞ$recovery$accuracy with ConsCab:Pop: ¼ ConsAUS
$45; 000 ð1Þ
20; 000; 000
where Cons is the consumption, Cab stands for Caboolture,
Pop. for population, AUS for Australia and Hosp. for hospital. It
becomes evident that the transformation due to human
metabolism (excretion ratio) cancels out of the equation when
assumed to be similar for patients in the hospital and for
people at home. The consumption of pharmaceuticals in the
STP catchment is estimated by calculating an average per
capita consumption from the national consumption data
multiplied with the number of inhabitants in the catchment.
The consumption of in-patients in the hospital is added to the

domestic consumption to obtain an estimate for the total STP
influent load (see also Table SI 3).
The prediction for 27 compounds where both national and
hospital audit data were available, in some cases deviated
significantly from the experimentally determined values.
However, only 8 substances would have been classified
differently based on audit data when applying strict boundaries
for the classification which does not change the overall
picture substantially.
Possible reasons for three examples are briefly discussed: 1)
The overestimation in the case of ibuprofen may be reasonably
explained by the fact that the national consumption is likely to
be substantially underestimated because ibuprofen can also be
obtained over the counter and in supermarkets without
prescription. 2) A patient who regularly takes histamine
blockers (at home) is likely to take them with him if he is being
hospitalised (for any treatment not related or interfering with
histamine blockers). This is one of the cases where patients
may bring their own medication to the hospital and is also
assumed to be valid for beta-blockers and diuretics. 3) In some
countries trimethoprim is often applied together with sulphamethoxazole
(combination item) and hence would be
expected in a similar ratio. In Australia, the consumption
pattern is different: 70% of trimethoprim is sold as single item
(general public) and in the hospital under investigation even
90% are administered as individual compound.
In other cases the explanation may be sought in a higher or
lower than average number of patients being treated during the
sampling period in the hospital. However, if the number of
treated patients shall be estimated from measurements, the
excretion ratio and absolute recoveries for chemical analyses
need to be taken in to account (see Eq. (1)). This makes it difficult
to compare measured influent loads from a STP with audit data
from an individual health care facility to reliably calculate the
health care facility’s contribution to the total influent of the STP.
3.4. Comparison with other studies
The Caboolture catchment, with 4.4 beds per 1000 inhabitants,
is comparable with two other studies (3.6 and 4.4 beds per 1000
inhabitants, see Table 3). Without audit data for the hospitals
and general public, the load estimations based on measured
concentrations and an estimate for wastewater based on
average water consumption in the study by Thomas et al. (2007)
make a direct comparison difficult. However, higher contributions
were also found for paracetamol and trimethoprim. In
the study by Weissbrodt et al. (2009) the loads at the influent of
the STP were not measured. The percentage determined in this
study is the amounts measured in the sewer divided by the
amounts administered on the corresponding days. The
compounds investigated in the Swiss study are iodinated X-ray
contrast media and cytostatics, both compounds almost
exclusively administered in hospitals. Only 50% of the X-ray
contrast media and a maximum of 7.5% of the cytostatics were
quantified in the hospital’s effluent, implying that the
remaining part is most likely ‘‘carried home’’ by patients and
excreted in household toilets. In the studies by Heberer and
Feldmann (2005) and Feldmann et al. (2008) the hospital bed
density is significantly higher (12.1 beds per 1000 inhabitants)
with a sub-catchment bed density of 24. Pharmaceutical loads
water research 44 (2010) 605–615 613
were measured in the influent of the STP and in selected
hospital effluents. With day-specific hospital consumption
data the contribution of the other hospitals was estimated,
resulting in a total hospital contribution of 15% (carbamazepine),
10% (diclofenac) and 50% (metamizole, not measured in
our study). Although the results seem to be in good agreement
with our study, the limited number of compounds, the various
approaches used, and the different catchment characteristics
preclude a comprehensive comparison.
3.5. Hospital wastewater treatment and catchments in
South East Queensland
Over 800 pharmaceuticals, disinfectants and other substances
are recorded in the DUSC and the hospital database. Whilst
the 59 substances analysed for in this study presents one of
the more comprehensive studies of the relative contribution
of a hospital to total load in wastewater, we do not claim that
these results can be extrapolated for each of these 800
substances, at all hospitals, or medical research activities in
general. As is often the case, the selection of these 59
substances was based upon the availability of a validated
analytical method. Despite this ‘‘limitation’’, even if there
were substances that originate almost exclusively from
hospital wastewater, or if measures were taken to prevent
pharmaceutical residues entering hospital wastewater
(source control, separate collection of urine and faeces
(Heinzmann et al., 2008)) or if hospital wastewater was treated
on site, over 85% of the total load for the majority of the
pharmaceuticals investigated in this study would still reach to
the STP because they are excreted by the public at home in
their households. Even for very specific compounds, almost
exclusively administered in hospitals, the trends in many
health care systems are moving towards shorter hospitalisations
or even treatment of out-patients (particularly diagnostics).
Two examples are the iodinated X-ray media and
cytostatics: although administered in high amounts in
hospitals, they cannot be recovered to 100% and hence solely
attributed to hospital effluent (Weissbrodt et al., 2009).
Relevance to other catchments in South East Queensland
(SEQ): The three catchments of main interest within SEQ, the
ones with advanced water treatment plants for providing
purified recycled water to the region (for planned indirect
potable reuse scheme), have approximately 8 hospital beds
per 1000 inhabitants (Luggage Point, eleven hospitals), 0.4
(Gibson Island, one hospital) and 1.7 (Bundamba, five hospitals).
While the hospital in Caboolture (this study) contributes
4.2 beds per 1000 inhabitants (total in the catchment 4.4), the
biggest individual hospital in the catchment of Luggage Point
accounts for only 1.5. A desktop exercise analysing audit data
from the sum of all hospitals in these catchments is proposed
to evaluate if further steps are required. This includes the
planning of future sampling campaigns and the potential
benefit of treating some hospitals’ wastewater at the source.
4. Conclusions
Measurements: For several, widely applied pharmaceuticals,
an individual hospital seems to be a small additional point

614
source in the catchment of a sewage treatment plant. In
this study a hospital with 4.4 hospital beds per 1000
inhabitants contributed less than 15% to the total load in
the influent of the sewage treatment plant for 28
substances, detected in both hospital effluent and STP
influent, which is in good agreement with estimates from
other studies. Considering a conservative worst case
uncertainty estimation, the hospital contribution only
exceeded 15% for two substances, roxithromycin (max.
56%) and trimethoprim (max. 18%).
Audit data: The contribution of the hospital calculated with
audit data and the chosen classification reveals good
agreement with actual measurements for three quarters of
the substances. National audit data to calculate the
consumption by the general public in a catchment and
hospital data for in-patients appear to be good predictors.
This approach can be used with some confidence for
substances where no analytical method exists to experimentally
determine concentrations and loads or where the
LOQ is not low enough. This needs to be tested for other
countries (dependant upon the comprehensiveness and
quality of national and hospital audit data).
Sampling in general: Sampling campaigns in hospital wastewater
are prone to high uncertainty due to a highly dynamic
system (flow and concentrations). All effort should be
undertaken to understand the system (behaviour) prior to
setting up a sound sampling protocol to ensure that representative
samples can be obtained.
Other catchments in South East Queensland: The preliminary
analysis based on hospital bed densities suggests focusing
on the catchment of the STP at Luggage Point (approximately
8 hospital beds per 1000 inhabitants). However it has
to be noted that this hospital bed density consists of 3
major public hospitals and a series of private hospitals.
Since measurements will be very expensive to assess all
hospitals’ contributions. A detailed desktop analysis of all
audit data is planned to identify if there are major sources
and if measurements at selected locations may be
appropriate.
Hospital wastewater treatment: If, for whatever motivation,
hospital wastewater shall be treated separately onsite, it
must be noted, that for many substances no major overall
reduction can be achieved since many pharmaceuticals are
taken on a regular basis at home. With the current trend to
shorter hospitalisations and treatments (diagnostics) of outpatients,
this also holds true for compounds mainly
administered in hospitals.
Acknowledgments
We would like to acknowledge the following institutions and
individuals who contributed to this study: the South East
Queensland Urban Water Security Research Alliance (financial
support); David Fillmore, Rick Jones and Wayne Batchler
from Moreton Bay Water (assistance in setting up proper
sampling points and support during the campaign);
Benjamin Tan and Mary Hodge from Queensland Health
water research 44 (2010) 605–615
Forensic and Scientific Services (logistics and processing
samples); John Doonan, Greg Jackson and Daniel Field from
Queensland Health (making contacts and providing hospital
data); Maxine Robinson and Chris Raymond from The Drug
Utilisation Sub-Committee of the Pharmaceutical Benefits
Advisory Committee, Commonwealth Department of Health
and Aged Care (providing Australian drug-use statistics);
Christa McArdell from Eawag (valuable discussion and
review of manuscript); and the Swiss National Science
Foundation (Grant PBEZP2-122958 awarded to the first
author).
Appendix.
Supplementary information
Information about the supplementary material can be found
at doi:10.1016/j.watres.2009.08.002.
references
Baquero, F., Martínez, J.-L., Cantón, R., 2008. Antibiotics and
antibiotic resistance in water environments. Current Opinion
in Biotechnology 19, 260–265.
Boillot, C., Bazin, C., Tissot-Guerraz, F., Droguet, J., Perraud, M.,
Cetre, J.C., Trepo, D., Perrodin, Y., 2008. Daily physicochemical,
microbiological and ecotoxicological fluctuations of a hospital
effluent according to technical and care activities. Science of
the Total Environment 403 (1–3), 113–129.
de Gruijter, J., Brus, D., Bierkens, M., Knotters, M., 2006. Sampling
for Natural Resource Monitoring. Springer.
Feldmann, D.F., Zuehlke, S., Heberer, T., 2008. Occurrence, fate
and assessment of polar metamizole (dipyrone) residues in
hospital and municipal wastewater. Chemosphere 71 (9),
1754–1764.
Ferk, F., Misik, M., Grummt, T., Majer, B., Fuerhacker, M.,
Buchmann, C., Vital, M., Uhl, M., Lenz, K., Grillitsch, B.,
Parzefall, W., Nersesyan, A., Knasmuller, S., 2009. Genotoxic
effects of wastewater from an oncological ward. Mutation
Research – Genetic Toxicology and Environmental
Mutagenesis 672 (2), 69–75.
Gautam, A.K., Kumar, S., Sabumon, P.C., 2007. Preliminary study
of physico-chemical treatment options for hospital
wastewater. Journal of Environmental Management 83 (3),
298–306.
Hartmann, A., Alder, A.C., Koller, T., Widmer, R.M., 1998.
Identification of fluoroquinolone antibiotics as the main
source of umuC genotoxicity in native hospital wastewater.
Environmental Toxicology and Chemistry 17 (3), 377–382.
Heberer, T., Feldmann, D., 2005. Contribution of effluents from
hospitals and private households to the total loads of
diclofenac and carbamazepine in municipal sewage effluents –
modeling versus measurements. Journal of Hazardous
Materials 122 (3), 211–218.
Heinzmann, B., Schwarz, R.-J., Schuster, P., Pineau, C., 2008.
Decentralized collection of iodinated X-ray contrast media in
hospitals – results of the feasibility study and the practice test
phase. Water Science and Technology 57 (2).
Khan, S.J., Ongerth, J.E., 2004. Modelling of pharmaceutical
residues in Australian sewage by quantities of use and
fugacity calculations. Chemosphere 54 (3), 355–367.

Kummerer, K., 2001. Drugs in the environment: emission of
drugs, diagnostic aids and disinfectants into wastewater by
hospitals in relation to other sources – a review. Chemosphere
45 (6–7), 957–969.
Kummerer, K., 2004. Resistance in the environment. Journal of
Antimicrobial Chemotherapy 54 (2), 311–320.
Kummerer, K., 2009. Antibiotics in the aquatic environment –
a review – part II. Chemosphere 75, 435–441.
Lenz, K., Mahnik, S.N., Weissenbacher, N., Mader, R.M., Krenn, P.,
Hann, S., Koellensperger, G., Uhl, M., Knasmuller, S., Ferk, F.,
Bursch, W., Fuerhacker, M., 2007. Monitoring, removal and
risk assessment of cytostatic drugs in hospital wastewater.
Water Science and Technology 56 (12), 141–149.
Ort, C., Gujer, W., 2006. Sampling for representative
micropollutant loads in sewer systems. Water Science and
Technology 54 (6–7), 169–176.
Ort, C., Lawrence, M.G., Joss, A. Sampling for pharmaceuticals
and personal care products to assess representative
water research 44 (2010) 605–615 615
environmental loads: sewers and wastewater treatment
plants, in preparation.
Pauwels, B., Verstraete, W., 2006. The treatment of hospital
wastewater: an appraisal. Journal of Water and Health 4 (4),
405–416.
Thomas, K.V., Dye, C., Schlabach, M., Langford, K.H., 2007.
Source to sink tracking of selected human pharmaceuticals
from two Oslo city hospitals and a wastewater treatment
works. Journal of Environmental Monitoring 9 (12), 1410–
1418.
Watkinson, A.J., Murby, E.J., Kolpin, D.W., Costanzo, S.D., 2009.
The occurrence of antibiotics in an urban watershed: from
wastewater to drinking water. Science of the Total
Environment 407, 2711–2723.
Weissbrodt, D., Kovalova, L., Ort, C., Pazhepurackel, V., Moser, R.,
Hollender, J., Siegrist, H., McArdell, C., 2009. Mass flows of Xray
contrast media and cytostatics in hospital wastewater.
Environmental Science & Technology 43 (13), 4810–4817.

A three-compartment model for micropollutants sorption
in sludge: Methodological approach and insights
Maialen Barret, Dominique Patureau, Eric Latrille, Hélène Carrère*
INRA, UR050, Laboratoire de Biotechnologie de l’Environnement, Avenue des Etangs, 11100 Narbonne, France
article info
Article history:
Received 11 June 2009
Received in revised form
17 August 2009
Accepted 20 August 2009
Available online 27 August 2009
Keywords:
Activated sludge
Bioavailability
Isotherm
Sorption model
Xenobiotic
1. Introduction
abstract
Organic micropollutants such as Polycyclic Aromatic Hydrocarbons
(PAHs) present very low water solubility and are
highly hydrophobic. Due to these properties, they tend to sorb
to organic matter. In the context of wastewater treatment,
sorption phenomena firstly impact on physical fate of these
compounds, determining the fluxes distribution throughout
the physical separation steps (eg. settling, centrifugation). But
these physical mechanisms only determine micropollutants
dispersion, whereas biological mechanisms are mainly
responsible for their removal. Biological mechanisms may
also be influenced by sorption, throughout bioavailability
limitation towards biodegradation. Bioavailability might be
the main limiting factor of PAHs biodegradation in sludge.
water research 44 (2010) 616–624
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
* Corresponding author. Tel.: þ33 468 425 168; fax: þ33 468 425 160.
E-mail address: carrere@supagro.inra.fr (H. Carrère).
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.watres.2009.08.029
In sludge resulting from wastewater treatment, organic micropollutants sorb to particles
and to dissolved/colloidal matter (DCM). Both interactions may influence their physical and
biological fate throughout the wastewater treatment processes. To our knowledge, sludge
has never been considered as a three-compartment matrix, in which micropollutants
coexist in three states: freely dissolved, sorbed-to-particles and sorbed-to-DCM. A methodology
is proposed to concomitantly determine equilibrium constants of sorption to
particles (Kpart) and to DCM (KDCM). Polycyclic Aromatic Hydrocarbons (PAHs) were chosen
as model compounds for the experiments. The logarithm of estimated equilibrium
constants ranged from 3.1 to 4.3 and their usual correlation to PAH hydrophobicity was
verified. Moreover, PAH affinities for particles and for DCM could be compared. Affinity for
particles was found to be stronger, probably due to their physical and chemical characteristics.
This work provided a useful tool to assess the freely dissolved, sorbed-to-particles
and sorbed-to-DCM concentrations of contaminants, which are necessary to accurately
predict their fate. Besides, guidelines to investigate the link between sorption and the
fundamental concept of bioavailability were proposed.
ª 2009 Elsevier Ltd. All rights reserved.
Micropollutants bioavailability has been widely investigated
in soils and many tools were subsequently designed to quantify
it (Reid et al., 2000; Tang et al., 2002; van Straalen et al., 2005;
Oleszczuk, 2007; Hickman et al., 2008; Oleszczuk, 2008).
However, these studies referred to different definitions of
bioavailability. Moreover, bioavailability and bioaccessibility
were sometimes confused. Semple et al. (2004, 2007) gathered
most of the encountered concepts into two definitions:
a bioavailable compound is ‘‘freely available to cross an
organism’s cellular membrane from the medium the organism
inhabits at a given time’’; a bioaccessible compound is ‘‘available
to cross an organism’s cellular membrane from the environment,
if the organism has access to the chemical’’. These
definitions imply that a bioaccessible compound can become
bioavailable after some time.

Nomenclature
DCM Dissolved and Colloidal Matter
DM Dry Matter of the sludge, which includes particles
and DCM
PAH Polycyclic Aromatic Hydrocarbon
Cfree Concentration of freely dissolved PAHs (mg/mL)
CDCM Concentration of sorbed-to-DCM PAHs (mg/gDCM) Cpart Concentration of sorbed-to-particles PAHs
(mg/gPART) On the contrary, micropollutants bioavailability in sludge
has been little studied. Related concepts and mechanisms are
less precisely described than in soil. Some of them may be
similar between soil and sludge, although matrix characteristics
and micropollutants/matrix contact time completely differ.
From a general point of view, micropollutants are usually
assumed to be bioavailable when they are located in the
aqueous phase and bioaccessible when they are initially sorbed-to-particles
and can be transferred to the aqueous phase
during the process (Byrns, 2001; Artola-Garicano et al., 2003;
Langford et al., 2005; Urase and Kikuta, 2005; Dionisi et al., 2006).
Otherwise, the assumption that a fraction of sorbed-to-particles
compounds would be directly bioavailable can also be
encountered (Fountoulakis et al., 2006). Anyway, even if the
nature of the link between sorption phenomena and bioavailability
was never demonstrated and is still subject to controversy
when studying sludge, the existence of a link is generally
admitted. Thus, to study sorption phenomena appears to be the
first step towards bioavailability investigation.
In published studies dealing with micropollutants sorption
in sludge, sludge have been considered as a two-compartment
system, corresponding to two states for micropollutants: sorbed-to-particles
and aqueous (Byrns, 2001; Artola-Garicano
et al., 2003; Ternes et al., 2004; Dionisi et al., 2006; Carballa
et al., 2008). However, sorption phenomena also occur inside
aqueous phase. Indeed, micropollutants have been shown to
interact with dissolved and colloidal matter (DCM) of various
origins (Chiou et al., 1986; Perminova et al., 2001) including
sludge from wastewater treatment plant (Holbrook et al., 2004).
The sorption to DCM is likely to influence micropollutants
bioavailability in sludge as well as sorption to particles
(Mackay and Fraser, 2000; Vinken et al., 2004). Thus, the freely
dissolved and sorbed-to-DCM micropollutants states should be
differentiated. Sludge should then be considered as a threecompartment
matrix, with three states of micropollutants:
freely dissolved, sorbed-to-DCM and sorbed-to-particles. To
our knowledge, such a model has been used to study sorption
phenomena in aquatic environmental systems such as sediments
(Schrap and Opperhuizen, 1992; Lee and Kuo, 1999;
Amiri et al., 2005) but it has never been applied to sludge.
Micropollutants sorption equilibria in environmental
samples including sludge are often modelled by Freundlich
isotherms with a Freundlich coefficient close to 1, which is
equivalent to a linear isotherm (Hung et al., 2004; Ivashechkin
et al., 2004; Arias-Estevez et al., 2007). As a consequence, the
equilibrium of sorption to particles is usually assumed to fit
linear equations (Ternes et al., 2004; Carballa et al., 2006;
Dionisi et al., 2006; Katsoyiannis et al., 2006) as well as the
water research 44 (2010) 616–624 617
C aqu
K DCM
Kpart
Kglobal
Concentration of apparently dissolved
micropollutant (mg/mL)
Equilibrium constant of PAHs sorption to DCM
(mL/gDCM) Equilibrium constant of PAHs sorption to particles
(mL/gPART)
Coefficient of PAHs partition between aqueous
phase and particles (mL/gPART)
equilibrium of sorption to DCM (Yamamoto et al., 2003; Zhou
et al., 2007).
The objective of this study was to propose and to validate
a methodological approach to study micropollutants sorption
equilibria in sludge, considering sludge as a three-compartment
matrix. Even though this investigation focused on PAHs,
the methodology and its implications may be transposed to
a wider range of micropollutants.
2. Materials and methods
2.1. Sludge source
All experiments were performed using activated sludge from
a urban wastewater treatment plant. The influent was entirely
domestic. The plant consisted in preliminary treatments,
primary settling, aerated tank, secondary settling and thermophilic
anaerobic digestion of sewage sludge. Hydraulic
retention time in the aerated tank was 0.36 day. After collection,
sludge was divided into several samples and stored at
20 C in order to guaranty that all following experiments
were performed with exactly the same matrix. DCM fraction
represented the dissolved and colloidal matters which were
not differentiated, defined as the centrifugation supernatant
(12 000 g, 20 min, 35 C) filtered at 1.2 mm (Whatman GF/C
glass microfiber filter). As some particles were unwillingly
suspended back after centrifugation because of little pellet
cohesiveness, the filtration step was performed to remove
these particles. The DCM mass retained by the filter was
negligible, as the DCM fraction beyond the size of 1.2 mm was
insignificant (data not shown). On total sludge and on DCM
fraction, the dry matter (DM) was measured by weighing the
samples after heating at 105 C during 24 h and the organic
matter after heating at 550 C during 2 h. The defrosted activated
sludge contained 65.4 1.2 g DM/L, constituted of 7 1%
of DCM and 93 2% of particles. Organic matter respectively
accounted for 72 and 80% of particles and DCM dry matter.
2.2. Chemicals
All solvents were purchased from J.T.Baker. Mixtures are
indicated in volume percentage. Fluorene, phenanthrene,
anthracene, fluoranthene, pyrene, benzo(a)anthracene,
chrysene, benzo(b)fluoranthene, benzo(k)fluoranthene,
benzo(a)pyrene, dibenzo(a,h)anthracene, benzo(g,h,i)perylene
and indeno(1,2,3,c,d)pyrene powders were obtained from
Dr Ehrenstorfer GmbH. Each PAH was dissolved in

618
dichloromethane at 1 g/L. The spiking mix was prepared from
these individual concentrated solutions, adding 5 mL of each,
evaporating solvent under gentle nitrogen flow and dissolving
in 50 mL of acetonitrile. Final concentrations were 100 mg/L
for each PAH. This mixture was used in equilibrium experiments,
as competitive phenomena are often assumed not to
occur at low concentration (Krauss and Wilcke, 2005; Dionisi
et al., 2006; Zhou et al., 2007).
The 10 mg/L standard solution of PAHs in acetonitrile was
provided by Dr Ehrenstorfer GmbH. Dilution factors from 10 to
1000 were applied to obtain 6 calibration levels. Standards
were stored at –20 C.
2.3. PAHs quantification in sludge
400 mL of sludge were centrifuged during 20 min at
12 000 g, at 35 C. The supernatant was filtered at 1.2 mm
with glass microfiber filters (GF/C Whatman) to obtain the
aqueous phase containing the free and sorbed-to-DCM
micropollutants, to which 1% of formaldehyde at 37% was
added for conservation. The particulate phase (pellet) was
frozen at 20 C and freeze-dried.
Extraction from aqueous phase was performed using SPE
cartridges with 1 g C18 sorbent from Macherey-Nagel.
Cartridges were washed and conditioned with 2 5mL of
methanol:toluene:acetonitrile (33:33:33), 2 5 mL of methanol
and 2 5 mL of ultra pure water. 100 mL of sample were
percolated through the extraction columns at a flow rate of
2 mL/min. After the sample percolated, the sorbent was dried
in a stream of atmospheric air during 30 min. Adsorbates were
eluted with 2 5 mL of methanol:toluene (50:50). Extracts
were dried under nitrogen flow and dissolved into 1 mL of
acetonitrile. Each sample was extracted in duplicate and corrected
by the recovery performance determined on the same
aqueous sample (same DCM concentration) spiked at
approximately the same PAHs concentration.
Extraction from solid phase was carried out as described by
Trably et al. (2004).
PAHs quantification in extracts from aqueous and solid
phases was performed according to the method optimised by
the same authors (Trably et al., 2004). The sum of PAHs
concentration in aqueous and in solid phase was compared to
the spiking concentration to ensure that no significant losses
occurred during the experiment.
2.4. The three-compartment model
A micropollutant present in sludge can be located in one
among three physical compartments (Fig. 1): the freely dissolved
one (concentration Cfree, mg/mL), the sorbed-to-DCM
one (concentration CDCM, mg/gDCM) and the sorbed-to-particles
one (concentration Cpart, mg/gPART).
At equilibrium, the three-compartment system can be
described by the two following equations:
Kpart ¼ Cpart
and
Cfree
water research 44 (2010) 616–624
(1)
Micropollutant
KDCM ¼ CDCM
Cfree
where K part is the equilibrium constant of micropollutant
sorption to particles (mL/g PART) andK DCM is the equilibrium
constant of sorption to DCM (mL/g DCM). C free is very difficult to
obtain experimentally whereas Caqu (concentration of apparently
dissolved micropollutant, sum of the freely dissolved and
sorbed-to-DCM states, mg/mL) and Cpart are easily measurable.
From Caqu and Cpart measurement, Kglobal (mL/gPART) can be
estimated:
Kglobal ¼ Cpart
Caqu
KpartCfree
Kpart
¼
¼
KDCMCfree½DCMŠþCfree KDCM½DCMŠþ1
where [DCM] represents the DCM concentration (g/mL). K global
is not a thermodynamical equilibrium constant but a partition
coefficient, which is system-dependant. Equation (3) was
previously used in several studies dealing with the effect of
DCM on micropollutants sorption in environmental aquatic
systems (Schrap and Opperhuizen, 1992; Amiri et al., 2005).
2.5. Equilibration time
The kinetics of PAHs sorption was monitored in order to
determine the equilibration time. For this, eight sludge
samples were firstly mixed at 100 rpm at 35 C in centrifugation
flasks in order to stabilize the matrix. Then, each sample
was spiked at 5 mg PAH/g DM to start with the sorption kinetics.
The aqueous PAHs concentration was measured after 1, 3, 4, 6,
8, 10, 14 and 24 h of incubation (extractions were carried out in
duplicate).
2.6. Linearity verification
Particle
K part KDCM
Dissolved and
colloidal matter
(DCM)
Fig. 1 – Representation of the three-compartment model of
a micropollutant in sludge.
At constant DCM concentration, K global is constant. As
a consequence, the linearity of C part ¼ f(C aqu) is a linearity
indicator of equilibria (1) and (2). Five samples at 0.7 g DCM/L
and 4.9 g PART/L were therefore directly put in centrifugation
flasks and equilibrated at 35 C on a mixing table (100 rpm).
After 30 min of equilibration, they were spiked with the
spiking mix at different concentrations ranging from 1 to 5
mgHAP/gDM and incubated again at 35 C for 4 h. The samples
were then centrifuged at 35 C and both Cpart and Caqu were
measured according to quantification section.
(2)
(3)

2.7. Determination of sorption equilibrium constants
Kpart and KDCM can be extracted from equation (3) by assessing
Kglobal with various DCM concentrations for each PAH. Variations
of DCM concentrations from 0.6 to 4.5 gDCM/L were
obtained by diluting sludge with its own supernatant previously
separated and with water in different proportions. To
ensure that this dilution method did not modify pH and that
no particulate matter was released into DCM, the pH and the
chemical oxygen demand (COD) were measured in 10 samples
diluted by 1.2 (3.7 gDCM/L) to 200 (0.02 gDCM/L) fold factor. The
COD was chosen here because it is a fast and little volume
consuming method.
The same procedure previously detailed was applied to
equilibrate, spike, incubate and quantify Cpart and Caqu in the
samples from 0.6 to 4.5 gDCM/L. Hence, Kglobal dependence to
DCM concentration was measured. A non linear regression
algorithm of Levenberg-Marquardt type (Marquardt, 1963) was
used to minimize the sum of square errors and to estimate the
two parameters of the K global model: K part and K DCM. In some
cases, the algorithm did not converge, due to the measurement
noise and to high K DCM[DCM] values compared to 1.
Indeed, when the ‘‘þ1’’ term becomes negligible in the
denominator of equation (3), this leads to Equation (4):
Kglobalz
Kpart
KDCM½DCMŠ
¼ a
½DCMŠ
An infinity of couples (Kpart;KDCM) can be solution of this
equation, defined by the ratio a¼Kpart/KDCM. a could be
precisely determined by the algorithm. Afterwards, introduction
of a into equation (3) allowed us to estimate Kpart for
each Kglobal measurement and its corresponding [DCM]:
Kglobal
Kpart ¼
1 Kglobal ½DCMŠ
a
The average of calculated Kpart values and the corresponding
KDCM were used with the Levenberg-Marquardt algorithm
in order to evaluate standard deviations and correlation
coefficient. This procedure was applied to dibenzo(a,h)anthracene,
benzo(g,h,i)perylene and indeno(1,2,3,c,d)pyrene.
3. Results and discussion
3.1. Hypotheses verification
From the initial DCM concentration in sludge (12 700 mgO2/L),
the DCM concentration in diluted samples (from 1.2 to 200 fold
factor) could be predicted, assuming that no matter was
transferred from particles into aqueous phase. These predicted
concentrations were compared to measured values. The ratio
between them was very close to 1 (Fig. 2). Only for dilution
factors higher than 20, a little fraction of particulate matter
(15%) was found to be released, since the ratio reached 1.15.
The highest dilution factor applied in sorption experiments
was 7.5, which ensures that no particulate matter release
occurred during these experiments. As a consequence, the
DCM concentrations estimated thanks to the dilution factor
were reliable. Moreover the pH was not modify by the dilution
water research 44 (2010) 616–624 619
(4)
(5)
Measured COD (mgO2/L) /
theoretical COD (mgO2/L)
1,3
1,2
1,1
1
0,9
0,8
1 10 100
5,4
1000
Dilution factor
Fig. 2 – Ratio between measured and theoretical aqueous
COD (mg/L) (black circles) and pH (grey triangle) as
a function of sludge dilution factor. The dotted frame
indicates the range of dilution factors used for sorption
experiments.
(Fig. 2), implying that the physicochemical properties of DCM
can be assumed to be unchanged.
Besides, the absence of particulate matter release in
diluted samples indicates that sludge DCM does not behave
like sorbable/desorbable matter. Indeed, matter is likely to be
either dissolved/colloidal or particulate according to intrinsic
parameters. This implies that PAHs sorption to particles via
the sorption of DCM on which PAHs are sorbed does not occur,
which validates the ‘‘three-compartment model / two equilibria’’
description (Fig. 1).
The kinetics of fluorene and benzo(a)pyrene sorption in
sludge are presented in Fig. 3. The sorption equilibrium state
for this two PAHs was achieved within 1 h shaking, and it was
the same for all other PAHs (data not shown). This value of 1 h
for equilibrium achievement is consistent with literature
(Dionisi et al., 2006) and suggests that during wastewater
treatment processes, bioavailability is not limited by dynamic
phenomena, as contact between xenobiotics and sludge lasts
for several hours or days. Furthermore, a 4 h incubation time
was selected for equilibrium experiments.
The linear regressions performed to extract K global from
C part ¼ f(C aqu) produced regression coefficients R 2 ranging
from 0.85 to 0.95 (Table 1). This is consistent with the
hypothesis of linearity assumed in the model. Moreover,
Kglobal was plotted as a function of DCM concentration (Fig. 4)
for each PAH. The obtained models fitted very well the
experimental data with R 2 ranging from 0.68 to 0.95, which
again reinforced the model.
3.2. K global dependence to DCM concentration
According to the considered PAH, the measured partition
coefficients varied of a ten fold factor in the set of tested
conditions (Fig. 4). Indeed, a little shift of DCM concentration
was shown to imply an important variation of Kglobal, above all
at low concentrations. Moreover, when sludge is not considered
as a three-compartment system and measured Kglobal is
equated to an equilibrium constant, this so-called constant
can be largely underestimated. For example, measured Kglobal
6,4
6,2
6
5,8
5,6
pH

620
Aqueous fraction (%)
100
80
60
40
20
Fluorene
for benzo(b)fluoranthene at 0.6 g DCM/L was about 2000 mL/
g PART whereas the real equilibrium constant K part was estimated
at 6920 mL/gPART (Fig. 4). In this case, assimilating
Kglobal to the equilibrium constant would lead to an underestimation
of 71%. Overall, this underestimation varied from 57
to 94% within the PAHs family and the more hydrophobic the
PAH, the higher the error. Such an underestimation prevents
from accurately predicting micropollutants sorption in sludge.
All these results highlighted the necessity to consider the
third phase to precisely assess the micropollutants fate.
The obtained partition coefficients could not be strictly
compared to literature data. Indeed, the studies published
until now about PAHs sorption in sludge are very few, and
authors neither take into account the three compartments nor
measure DCM concentration. Moreover, our results underlined
the strong Kglobal dependence to this concentration,
which obliges to compare Kglobal values at equivalent DCM
concentrations.
3.3. Comparison between K part and K DCM
0
0 10 20 30
0
0 10 20 30
t (h)
t (h)
The logarithm of thermodynamical equilibrium constants
Kpart and KDCM extracted from Kglobal dependence to DCM
concentration are presented in Table 1. Standard deviation
water research 44 (2010) 616–624
Aqueous fraction (%)
100
80
60
40
20
Benzo(a)pyrene
Fig. 3 – Percentage of initially spiked PAHs that remain in aqueous phase as a function of equilibration time. Since PAHs are
spiked in liquid form, at time 0, the aqueous fraction is assumed to be 100%.
was higher for highly hydrophobic compounds than for
weakly hydrophobic ones, due to mathematical difficulties
mentioned in the paragraph 2.7. The first way to reduce
standard deviations would be to multiply measurements in
order to decrease experimental uncertainty due to (i) sludge
heterogeneity and (ii) quantification procedure. The second
way would be to optimize measurements distribution on the
DCM concentration scale. However, this option is limited for
practical reasons. Indeed, the upper DCM concentration limit
comes from the DCM concentration in raw sludge that cannot
be passed over. Besides, to minimize DCM concentration, no
supernatant volume should be added and a small sludge
volume should be diluted with water. Nevertheless, particles
concentration would concomitantly decrease and particles
quantity needed for PAH quantification would constitute the
lower DCM concentration limit. Nonetheless, standard deviation
values for most PAHs were acceptable in comparison
with the literature data on adsorption topic, since standard
deviations, although very seldom estimated, are in this range
(Kordel et al., 1997; Dionisi et al., 2006; Carballa et al., 2008).
Again, K part comparison with previously published data
was hardly feasible. However, at low particulate matter
concentration (1 g/L), which indicates a probable very low
DCM concentration, Dionisi et al. (2006) measured a K global of
Table 1 – log Kow, partition coefficient Kglobal and coefficient R 2 obtained from linear regression of Cpart [ f(Caqu) at 0.7 gDCM/L,
and equilibrium constants extracted from Kglobal dependence to DCM concentration in the range 0.6–4.5 gDCM/L.
Compound log Kow Partition coefficient Equilibrium constants
K global (mL/ g part) R 2
log K part
log K DCM
Fluorene 4.18 638 0.89 3.21 0.23 3.24 0.37
Phenanthrene 4.46 2142 0.92 3.56 0.24 3.35 0.36
Anthracene 4.5 1493 0.92 3.45 0.17 3.11 0.29
Fluoranthene 4.9 2357 0.85 4.01 0.44 3.46 0.61
Pyrene 4.88 3196 0.95 3.71 0.37 3.20 0.62
Benzo(a)anthracene 5.63 2712 0.92 3.78 0.21 3.33 0.32
Chrysene 5.63 1833 0.92 3.86 0.40 3.47 0.56
Benzo(b)fluoranthene 6.04 2381 0.89 3.84 0.38 3.49 0.52
Benzo(k)fluoranthene 6.21 1857 0.90 4.27 0.76 4.02 0.85
Benzo(a)pyrene 6.06 566 0.89 4.01 0.55 3.83 0.65
Dibenzo(a,h)anthracene 6.86 1701 0.90 4.34 1.57 4.12 1.71
Benzo(g,h,i)perylene 6.78 1542 0.89 4.11 1.10 4.00 1.20
Indeno(1,2,3,c,d)pyrene 6.58 1035 0.87 4.03 0.83 3.81 0.97

)
( mL/
gP
ART
mL/
gP
AR
) Kglobal ( T
mL/
gP
AR
) Kglobal ( T
P ) Kglobal Kglobal ( mL/
g ART
)
Kglobal ( mL/
gP
ART
1200
800
400
Fluorene
K part = 1 606
K DCM = 1 717
R 2 = 0.857
0
0 1 2 3 4 5
8000
Fluoranthene
Kpart = 10 225
6000
4000
KDCM = 2 894
R2 = 0.823
2000
0
0 1 2 3 4 5
Chrysene
5000
4000
3000
Kpart = 7 251
KDCM = 2 940
R2 = 0.854
2000
1000
3000
2000
1000
9700 5000 mL/g PART for pyrene in an activated sludge. This
value might be compared to our K part, asK part corresponds to
K global limit when DCM concentration tends to zero. We
calculated a K part in the range 2200–12 000 mL/g PART for pyrene
(Table 1), which is in good agreement with Dionisi et al.’s
(2006) values.
KDCM is easier to compare with literature data than Kpart,
since several studies were performed in absence of particles. As
an illustration, Holbrook et al. (2004) studied pyrene adsorption
onto sludge DCM. Depending on sample origin and size
Phenanthrene
K part = 3 659
K DCM = 2 244
R 2 = 0.902
0
0 1 2 3 4 5
4000
Pyrene
Kpart = 5 089
3000
2000
KDCM = 1 594
R2 = 0.680
1000
0
0 1 2 3 4 5
Benzo(b)fluoranthene
5000
4000
3000
Kpart = 6 920
KDCM = 3 073
R2 = 0.877
2000
1000
3000
2000
1000
Anthracene
K part = 3 636
K DCM= 2 117
R 2 = 0.833
0
0 1 2 3 4 5
Benzo(a)anthracene
5000
4000
3000
Kpart = 7 941
KDCM = 3 337
R2 = 0.876
2000
1000
0
0 1 2 3 4 5
8000
Benzo(k)fluoranthene
Kpart = 18 604
6000
4000
KDCM = 10 429
R2 = 0.950
0
0 1 2 3 4 5
0
0 1 2 3 4 5
0
0 1 2 3 4 5
5000
4000
3000
Kpart = 10 283
KDCM = 6 745
R
2000
1000
2 8000
= 0.940
6000
4000
Kpart = 22 076
KDCM = 13 202
R
2000
2 5000
= 0.863
4000
3000
Kpart = 12 770
KDCM = 9 992
R
2000
1000
2 Benzo(a)pyrene Dibenzo(a,h)anthracene Benzo(g,h,i)perylene
α = 1.672
α = 1.278
= 0.895
5000
4000
3000
2000
1000
0
0 1 2 3 4 5
Indeno(1,2,3,c,d)pyrene
Kpart = 10 599
KDCM = 6 465
R2 α = 1.639
= 0.823
0
0 1 2 3 4 5
DCM (g/L)
water research 44 (2010) 616–624 621
0
0 1 2 3 4 5
DCM (g/L)
2000
0
0 1 2 3 4 5
DCM (g/L)
Fig. 4 – Measured (mark) and modelled (line) K global (mL/g PART) as a function of DCM concentration (g/L). The a coefficient (for
dibenzo(a,h)anthracene, benzo(g,h,i)perylene and indeno(1,2,3,c,d)pyrene), Kpart,KDCM and regression coefficients estimated
thanks to the algorithm are presented on each graph. The corresponding logarithmic values and standard deviation for Kpart
and KDCM are presented in Table 1.
fractionation, the authors measured some K DCM values ranging
from below 1000–80 000 mL/g OC (g of organic carbon). Since
organic carbon usually represents approximately 25% of the
total matter in sludge (Dignac et al., 2000), these values are
widespread from below 250 to 20 000 mL/gDCM. This range is in
accordance with our estimation of 380–6607 mL/gDCM (Table 1).
The common and general correlation between the logarithm
of equilibrium constants and hydrophobicity of PAHs
(Karickhoff et al., 1979; Poerschmann and Kopinke, 2001) was
observed for both Kpart and KDCM (Fig. 5). This demonstrates

622
log K part , log K DCM
7
6
5
4
3
2
y = 0.308x + 2.13
y = 0.348x + 1.61
1
4 4.5 5 5.5 6 6.5 7
log K ow
Fig. 5 – log Kpart (black rounds) and log KDCM (grey rounds)
as a function of PAHs log K ow.
that PAHs sorption to particles and to DCM may be driven by
hydrophobic interactions. Moreover, the linear regressions
revealed very similar slopes (0.31 and 0.35) but different
origins (2.13 and 1.61), resulting in different values (Fig. 5).
Indeed, the PAHs affinity for particles was found to be stronger
than for DCM. This finding can be related to the compartments
characteristics since equilibrium constants reflect the
affinity of PAHs for the corresponding compartment. In this
work, the compartments were differentiated according to
their physical criteria, as defined in the 2.1. Section: the
density criteria imposed by centrifugation, followed by the
size criteria of filtration. Other physical properties such as
hydrophobicity, surface charge and specific surface area
differentiate particles and DCM. In addition of physical properties,
Bougrier et al. (2008) demonstrated that biochemical
composition (chemical oxygen demand, carbohydrates and
proteins content) is different in particles and DCM. These
properties are influenced by sludge history in terms of
substrate (wastewater), involved microbiological population,
and process operating parameters. For example, high sludge
retention time in activated sludge process reduces particles
specific surface area (Liss et al., 2002) and leads to more
hydrophobic particles (Liao et al., 2001) with higher lipid
content (Liss et al., 2002). Quantifying the influence of physical
and biochemical sludge characteristics on the sorption equilibrium
constants of PAHs would be very interesting.
3.4. Assessment of PAHs compartment distribution
Kpart and KDCM determination and dependence to hydrophobicity
provided the parameters needed to simulate PAHs
distribution between the three sludge compartments.
Aqueous and freely dissolved percentages were simulated in
two theoretical and extreme cases: for a weakly hydrophobic
PAH and for a strongly hydrophobic one (Fig. 6).
The simulation demonstrated that freely dissolved fraction
does not exceed 0.4% of the total concentration for a very
hydrophobic compound and 2.5% for a weakly hydrophobic
one. This fraction is slightly influenced by DCM concentration
in both cases.
On the contrary, aqueous fraction of both PAHs appeared
to strongly depend on DCM concentration. This result is of
water research 44 (2010) 616–624
%
35
30
25
20
15
10
5
0
0 2 4
6 8
DCM (g/L)
Fig. 6 – Percentage of free (plots) and aqueous (linked plots)
concentrations modelled for a theoretical weakly
hydrophobic PAH (log K ow [ 4.2, square) and for
a theoretical very hydrophobic PAH (log Kow [ 6.8, triangle)
as a function of dissolved and colloidal matter
concentration, particulate concentration being fixed at
10 g/L. For this simulation, Kpart and KDCM of both PAHs
were estimated thanks to the relationship presented in
Fig. 5. The reversion point is located at 6.01 gDCM/L, which
corresponds to 25.5% of aqueous PAHs. Over this value,
aqueous fraction is slightly higher for the very
hydrophobic PAH than for the weakly hydrophobic one.
great concern since aqueous fraction is usually equated to
bioavailable fraction, as previously mentioned. It means that
low DCM concentrations favour higher percentage in aqueous
phase for weakly hydrophobic PAHs than for very hydrophobic
ones, whereas reasonable high concentrations
discriminate less PAHs. The extrapolation of the model to
extremely concentrated conditions ([DCM] > 6 g/L) shows that
the aqueous concentration of a very hydrophobic PAH could
equal and even exceed a weakly hydrophobic one (Fig. 6). In
the usual bioavailability hypothesis ‘‘aqueous is bioavailable’’,
this would mean that a very hydrophobic PAH could be more
bioavailable than a weakly hydrophobic one. This case is
really atypical, since micropollutants are usually assumed to
be less bioavailable when they are more hydrophobic (Mackay
and Fraser, 2000). It is an important insight derived from the
three-compartment approach. The reversion point would be
located at [DCM] ¼ 6.0 g/L for the studied activated sludge, as
particulate matter concentration was fixed at 10 g/L for the
simulation. However, it has to be reminded that the model
was established for DCM concentrations up to 4.5 g/L. Moreover,
DCM at the reversion point would represent 37.5% of the
total matter, which is much higher than common values
(Bougrier et al., 2008). As a conclusion, the latter case is little
probable in real systems, but it could be interesting to create it
in laboratory investigations.
Apart from the descriptive issue previously discussed, such
simulations constitute a new tool for micropollutants physical
fate assessment. They will allow to predict the fate of PAHs
throughout the separation steps encountered in wastewater

treatment plants. In the example of a settler, the flux of sorbedto-particles
PAHs in solid effluent (for agricultural disposal)
can be quantified as well as the flux of sorbed-to-DCM and
dissolved PAHs carried by clarified water (discharged into
surface water). Indeed, Katsoyiannis and Samara (2007)
observed an influence of DCM concentration on micropollutants
fate, without quantifying this effect. The threecompartment
approach could provide the missing data to
quantitatively take this interaction between micropollutants
and DCM into account. As a result, the methodology presented
here constitutes a real improvement in the assessment of
micropollutants fate throughout treatment plants and of their
environmental discharges.
3.5. Towards bioavailability
Quantifying PAHs distribution in sludge samples has an
important outcome concerning its consequences on
bioavailability topic. Indeed, it may be useful to verify the
usual hypothesis of identity between aqueous and bioavailable
fractions (Artola-Garicano et al., 2003; Dionisi et al., 2006).
This assumption may be verified by measuring biodegradation
at different aqueous concentrations. Initial biodegradation
rate in batch experiments could be an appropriate
indicator for biodegradation. The assumption would be
confirmed if biodegradation rate correlates with aqueous
concentration. Guidelines for such an experiment are
provided by the present work: it suggests that different
aqueous concentrations of one PAH might be achieved thanks
to different DCM concentrations.
On the other hand, investigation on the link between freely
dissolved and bioavailable fractions of one PAH appears more
difficult, as freely dissolved concentration varied little in the
presented conditions. It means that this link might be investigated
through either the comparison between different
compounds in one matrix or the comparison of one
compound in various matrixes for which this compound has
different affinities.
These guidelines are of great importance for the modeling
of micropollutants biological fate in sludge, since until now,
models have been built on hypotheses which have never been
clearly demonstrated.
4. Conclusions
The three-compartment methodology proposed to investigate
micropollutants sorption in sludge was validated. The results
underlined the fact that a reliable tool for micropollutants
behaviour prediction in sludge should integrate a threecompartment
model since DCM concentration influences
micropollutants distribution in a significant way. Indeed, the
usual biphasic approach was shown to lead to an important
underestimation of equilibrium constants. Despite the lack of
published data about PAH sorption in sludge, the results could
be confronted to a few studies. The confrontation showed the
consistency of equilibrium constant values and of dependency
to PAH hydrophobicity.
Besides, PAH affinities for particles and for DCM could be
compared. Affinity for particles was found to be stronger. This
water research 44 (2010) 616–624 623
may be due to the differences between particles and DCM in
term of physical and chemical characteristics. Further investigation
is necessary to include the influence of both sludge
and micropollutant characteristics in the three-compartment
model. This research, being presently undergone by the
authors, will enable to predict PAHs behaviour in any sludge,
as a function of its characteristics. The future predictive
model would allow to estimate the PAH distribution in any
sludge in order (i) to determine fluxes throughout physical
separation steps and (ii) to confront PAHs distribution
between compartments to their biodegradation, which might
help to elucidate bioavailability issue.
Acknowledgements
This study was funded by the Institut National de Recherche
Agronomique and the Ministère de l’Enseignement Supérieur
et de la Recherche du Gouvernement Français, France, and the
authors gratefully acknowledge them.
references
Amiri, F., Bornick, H., Worch, E., 2005. Sorption of phenols onto
sandy aquifer material: the effect of dissolved organic matter
(DOM). Water Research 39, 933–941.
Arias-Estevez, M., Fernandez-Gandara, D., Garcia-Falcon, M.S.,
Garcia-Rio, L., Mejuto, J.C., Simal-Gandara, J., 2007. Sorption of
PAHs to colloid dispersions of humic substances in water.
Bulletin of Environmental Contamination and Toxicology 79,
251–254.
Artola-Garicano, E., Borkent, I., Damen, K., Jager, T., Vaes, W.H.J.,
2003. Sorption kinetics and microbial biodegradation activity
of hydrophobic chemicals in sewage sludge: model and
measurements based on free concentrations. Environmental
Science and Technology 37, 116–122.
Bougrier, C., Delgenès, J.P., Carrère, H., 2008. Effects of thermal
treatments on five different waste activated sludge samples
solubilisation, physical properties and anaerobic digestion.
Chemical Engineering Journal 139, 236–244.
Byrns, G., 2001. The fate of xenobiotic organic compounds in
wastewater treatment plants. Water Research 35, 2523–2533.
Carballa, M., Fink, G., Omil, F., Lema, J.M., Ternes, T., 2008.
Determination of the solid-water distribution coefficient (K-d)
for pharmaceuticals, estrogens and musk fragrances in
digested sludge. Water Research 42, 287–295.
Carballa, M., Omil, F., Alder, A.C., Lema, J.M., 2006. Comparison
between the conventional anaerobic digestion of sewage
sludge and its combination with a chemical or thermal pretreatment
concerning the removal of pharmaceuticals and
personal care products. Water Science and Technology 53,
109–117.
Chiou, C.T., Malcolm, R.L., Brinton, T.I., Kile, D.E., 1986. Water
solubility enhancement of some organic pollutants and
pesticides by dissolved humic and fulvic acids. Environmental
Science and Technology 20, 502–508.
Dignac, M.F., Ginestet, P., Rybacki, D., Bruchet, A., Urbain, V.,
Scribe, P., 2000. Fate of wastewater organic pollution during
activated sludge treatment: nature of residual organic matter.
Water Research 34, 4185–4194.
Dionisi, D., Bertin, L., Bornoroni, L., Capodicasa, S., Papini, M.P.,
Fava, F., 2006. Removal of organic xenobiotics in activated
sludges under aerobic conditions and anaerobic digestion of

624
the adsorbed species. Journal of Chemical Technology and
Biotechnology 81, 1496–1505.
Fountoulakis, M.S., Stamatelatou, K., Batstone, D.J., Lyberatos, G.,
2006. Simulation of DEHP biodegradation and sorption during
the anaerobic digestion of secondary sludge. Water Science
and Technology 54, 119–128.
Hickman, Z.A., Swindell, A.L., Allan, I.J., Rhodes, A.H., Hare, R.,
Semple, K.T., Reid, B.J., 2008. Assessing biodegradation
potential of PAHs in complex multi-contaminant matrices.
Environmental Pollution 156, 1041–1045.
Holbrook, R.D., Love, N.G., Novak, J.T., 2004. Investigation of
sorption behavior between pyrene and colloidal organic
carbon from activated sludge processes. Environmental
Science and Technology 38, 4987–4994.
Hung, N.V., Tateda, M., Ike, M., Fujita, M., Tsunoi, S., Tanaka, M.,
2004. Sorption of biodegradation end products of nonylphenol
polyethoxylates onto activated sludge. Journal of
Environmental Sciences-China 16, 564–569.
Ivashechkin, P., Corvini, P.F.X., Dohmann, M., 2004. Behaviour of
endocrine disrupting chemicals during the treatment of
municipal sewage sludge. Water Science and Technology 50,
133–140.
Karickhoff, S.W., Brown, D.S., Scott, T.A., 1979. Sorption of
hydrophobic pollutants on natural sediments. Water Research
13, 241–248.
Katsoyiannis, A., Samara, C., 2007. The fate of dissolved
organic carbon (DOC) in the wastewater treatment process
and its importance in the removal of wastewater
contaminants. Environmental Science and Pollution
Research 14, 284–292.
Katsoyiannis, A., Zouboulis, A., Samara, C., 2006. Persistent
organic pollutants (POPs) in the conventional activated sludge
treatment process: model predictions against experimental
values. Chemosphere 65, 1634–1641.
Kordel, W., Hennecke, D., Franke, C., 1997. Determination of the
adsorption-coefficients of organic substances on sewage
sludges. Chemosphere 35, 107–119.
Krauss, M., Wilcke, W., 2005. Persistent organic pollutants in soil
density fractions: distribution and sorption strength.
Chemosphere 59, 1507–1515.
Langford, K.H., Scrimshaw, M.D., Birkett, J.W., Lester, J.N., 2005.
The partitioning of alkylphenolic surfactants and
polybrominated diphenyl ether flame retardants in activated
sludge batch tests. Chemosphere 61, 1221–1230.
Lee, C., Kuo, L., 1999. Quantification of the dissolved organic
matter effect on the sorption of hydrophobic organic
pollutant: application of an overall mechanistic sorption
model. Chemosphere 38, 807–821.
Liao, B.Q., Allen, D.G., Droppo, I.G., Leppard, G.G., Liss, S.N., 2001.
Surface properties of sludge and their role in bioflocculation
and settleability. Water Research 35, 339–350.
Liss, S.N., Liao, B.Q., Droppo, I.G., Allen, D.G., Leppard, G.G., 2002.
Effect of solids retention time on floc structure. Water Science
and Technology 46, 431–438.
Mackay, D., Fraser, A., 2000. Bioaccumulation of persistent
organic chemicals: mechanisms and models. Environmental
Pollution 110, 375–391.
Marquardt, D., 1963. An algorithm for least-squares estimation of
nonlinear parameters. SIAM Journal of Applied Mathematics
11, 431–441.
Oleszczuk, P., 2007. Investigation of potentially bioavailable
and sequestrated forms of polycyclic aromatic
water research 44 (2010) 616–624
hydrocarbons during sewagesludgecomposting.
Chemosphere 70, 288–297.
Oleszczuk, P., 2008. Tenax-TA extraction as predictor for free
available content of polycyclic aromatic hydrocarbons (PAHs)
in composted sewage sludges. Journal of Environmental
Monitoring 10, 883–888.
Perminova, I.V., Grechishcheva, N.Y., Kovalevskii, D.V.,
Kudryavtsev, A.V., Petrosyan, V.S., Matorin, D.N., 2001.
Quantification and prediction of the detoxifying properties of
humic substances related to their chemical binding to
polycyclic aromatic hydrocarbons. Environmental Science and
Technology 35, 486–490.
Poerschmann, J., Kopinke, F.D., 2001. Sorption of very
hydrophobic organic compounds (VHOCs) on dissolved humic
organic matter (DOM). 2. Measurement of sorption and
application of a Flory–Huggins concept to interpret the data.
Environmental Science and Technology 35, 1142–1148.
Reid, B.J., Stokes, J.D., Jones, K.C., Semple, K.T., 2000.
Nonexhaustive cyclodextrin-based extraction technique for
the evaluation of PAH bioavailability. Environmental Science
and Technology 34, 3174–3179.
Schrap, S.M., Opperhuizen, A., 1992. On the contradictions
between experimental sorption data and the sorption
partitioning model. Chemosphere 24, 1259–1282.
Semple, K.T., Doick, K.J., Jones, K.C., Burauel, P., Craven, A.,
Harms, H., 2004. Defining bioavailability and bioaccessibility
of contaminated soil and sediment is complicated.
Environmental Science and Technology 38, 228A–231A.
Semple, K.T., Doick, K.J., Wick, L.Y., Harms, H., 2007. Microbial
interactions with organic contaminants in soil: definitions,
processes and measurement. Environmental Pollution 150,
166–176.
Tang, J.X., Liste, H.H., Alexander, M., 2002. Chemical assays of
availability to earthworms of polycyclic aromatic
hydrocarbons in soil. Chemosphere 48, 35–42.
Ternes, T.A., Herrmann, N., Bonerz, M., Knacker, T., Siegrist, H.,
Joss, A., 2004. A rapid method to measure the solid-water
distribution coefficient (K-d) for pharmaceuticals and musk
fragrances in sewage sludge. Water Research 38, 4075–4084.
Trably, E., Delgenes, N., Patureau, D., Delgenes, J.P., 2004.
Statistical tools for the optimization of a highly reproducible
method for the analysis of polycyclic aromatic hydrocarbons
in sludge samples. International Journal of Environmental
Analytical Chemistry 84, 995–1008.
Urase, T., Kikuta, T., 2005. Separate estimation of adsorption and
degradation of pharmaceutical substances and estrogens in
the activated sludge process. Water Research 39, 1289–1300.
van Straalen, N.M., Donker, M.H., Vijver, M.G., van Gestel, C.A.M.,
2005. Bioavailability of contaminants estimated from uptake rates
into soil invertebrates. Environmental Pollution 136, 409–417.
Vinken, R., Hollrigl-Rosta, A., Schmidt, B., Schaffer, A., Corvini, P.F.X.,
2004. Bioavailability of a nonylphenol in dependence on the
association to dissolved humic substances. Water Science and
Technology 50, 277–283.
Yamamoto,H.,Liljestrand,H.M.,Shimizu,Y.,Morita,M.,2003.Effects
of physical–chemical characteristics on the sorption of selected
endocrine disruptors by dissolved organic matter surrogates.
Environmental Science and Technology 37, 2646–2657.
Zhou,J.L.,Liu,R.,Wilding,A.,Hibberd,A.,2007.Sorptionof
selected endocrine disrupting chemicals to different aquatic
colloids. Environmental Science and Technology 41,
206–213.

Removal of micropollutants and reduction of biological
activity in a full scale reclamation plant using ozonation
and activated carbon filtration
J. Reungoat a, *, M. Macova b , B.I. Escher b , S. Carswell b,c , J.F. Mueller b , J. Keller a
a
The University of Queensland, Advanced Water Management Centre (AWMC), Qld 4072, Australia
b
The University of Queensland, National Research Centre for Environmental Toxicology (EnTox), QLD 4108, Australia
c
Queensland Health Forensic and Scientific Services, Organics Laboratory, QLD 4108, Australia
article info
Article history:
Received 12 May 2009
Received in revised form
14 September 2009
Accepted 21 September 2009
Available online 1 October 2009
Keywords:
Water reuse
Oxidation
Adsorption
Pesticides
Pharmaceuticals
Bioassays
water research 44 (2010) 625–637
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
abstract
* Corresponding author. Tel.: þ61 734466251; fax: þ61 733654726.
E-mail address: j.reungoat@awmc.uq.edu.au (J. Reungoat).
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.watres.2009.09.048
Pharmaceutical compounds are found in secondary treated effluents up to mgL -1 levels and
therefore discharged into surface waters. Since the long term effects of these compounds
on the environment and human health are, to date, largely unknown, implementation of
advanced treatment of wastewaters is envisaged to reduce their discharge. This is of
particular relevance where surface waters are used as drinking water sources and when
considering indirect potable reuse. This study aimed at assessing the removal of organic
micropollutants and the concurrent reduction of their biological activity in a full scale
reclamation plant treating secondary effluent. The treatment consists of 6 stages: denitrification,
pre-ozonation, coagulation/flocculation/dissolved air flotation and filtration
(DAFF), main ozonation, activated carbon filtration and final ozonation for disinfection. For
that purpose, representative 24-hour composite samples were collected after each stage.
The occurrence of 85 compounds was monitored by LC/MS-MS. A battery of 6 bioassays
was also used as a complementary tool to evaluate non-specific toxicity and 5 specific toxic
modes of action. Results show that, among the 54 micropollutants quantified in the
influent water, 50 were removed to below their limit of quantification representing more
than 90% of concentration reduction. Biological activity was reduced, depending on the
specific response that was assessed, from a minimum of 62% (AhR response) to more than
99% (estrogenicity). The key processes responsible for the plant’s performances were the
coagulation/flocculation/DAFF, main ozonation and activated carbon filtration. The effect
of these 3 processes varied from one compound or bioassay to another but their combination
was almost totally responsible for the overall observed reduction. Bioassays yielded
complementary information, e.g. estrogenic compounds were not detected in the
secondary effluent by chemical analysis, but the samples had an estrogenic effect. The
main ozonation formed oxidation by-products of the organic micropollutants but
decreased the level of non-specific toxicity and other specific toxic modes of action,
demonstrating that the mixture of oxidation by-products was less potent than the mixture
of the parent compounds for the considered effects.
ª 2009 Elsevier Ltd. All rights reserved.

626
1. Introduction
In the last decade, numerous studies around the world have
demonstrated the presence of pharmaceutical compounds in
domestic wastewaters. These compounds are removed to
different degrees by the commonly used biological activated
sludge processes. While some compounds (e.g. ibuprofen,
paracetamol) are effectively removed, others (e.g. carbamazepine,
diclofenac) are barely affected by the biological treatment
(Onesios et al., 2009). As a result, some pharmaceutical
compounds are released into the environment, contaminating
surface waters. This situation is of concern as these
compounds have been designed to affect biological systems
and long term exposure effects on the biological elements in
the environment are still largely unknown. Human health is
also at risk when treated wastewater is discharged to water
bodies that are used as drinking water sources or considered
for indirect potable reuse. Therefore, advanced treatment of
wastewaters has to be envisaged to reduce the load of these
compounds in discharged effluents.
Several technologies have proven to be effective in
removing pharmaceutical compounds from water: activated
carbon adsorption (Nowotny et al., 2007; Snyder et al., 2007;
Ternes et al., 2002; Westerhoff et al., 2005; Yu et al., 2008),
ozonation and advanced oxidation processes (Esplugas et al.,
2007; Huber et al., 2003; Huber et al., 2005; Kim et al., 2008;
Nakada et al., 2007; Ternes et al., 2003; Zwiener and Frimmel,
2000) and membrane filtration (Kimura et al., 2004; Snyder
et al., 2007; Yoon et al., 2007). Activated carbon adsorption and
ozonation are considered to be economically feasible for
tertiary treatment of wastewaters by Joss et al. (2008). Nevertheless,
none of these processes can remove all the
compounds of concern and to date, no extensive study has
been carried out on the efficiency of the sequential combination
of these two processes for the removal of micropollutants
from secondary treated wastewater.
Chemical analysis of micropollutants in water has gone
through major developments in recent years and today’s
analytical techniques can measure some compounds at
concentrations as low as a few ng L 1 . Chemical analysis is
powerful to measure the concentration of targeted
compounds but does not cover the whole range of chemicals
that might be present in the effluent of a wastewater treatment
plant. Analytical techniques usually focus on pharmaceutically
active parent compounds whereas a variable
fraction is excreted from the human body as metabolites.
Moreover, biological treatment of wastewater may form yet
unidentified compounds through biodegradation reactions;
e.g. 17b-estradiol is degraded to estrone (Ternes et al., 1999).
Subsequent chemical treatments such as ozonation can also
form by-products. Chemical analysis of a limited suite of
compounds does not allow assessment of the potential biological
adverse effects of the wastewater as the cumulative
effect of the mixture of chemicals that may be present cannot
be easily integrated. Bioassays have been utilised as complementary
monitoring tools for the assessment of possible biological
effects of chemicals that are present in a particular
water sample (Escher et al., 2008b; Muller et al., 2007). These
bioanalytical tools are designed to quantify non-specific
water research 44 (2010) 625–637
toxicity or particular toxic modes of action (e.g. estrogenicity,
genotoxicity, phytotoxicity) induced by a sample on a biological
organism or a biological process. To date, bioassays have
not been widely used in the evaluation of water treatment
processes and have rarely been accompanied with chemical
analysis (Macova et al., 2010).
The present study assessed the removal of selected
micropollutants (pharmaceuticals and pesticides) and the
decrease of biological activity along a full scale water reclamation
plant using ozonation and activated carbon adsorption
to treat a secondary effluent. This aimed at identifying the
key steps in the treatment process as well as the key operating
parameters. Results of chemical analysis and bioassays were
compared to determine whether the use of both techniques
was redundant or allowed a deeper understanding of the
performances of individual treatment processes. The performance
and relevance of the treatment train is further discussed
in the context of indirect potable reuse.
2. Materials and methods
2.1. South Caboolture Water Reclamation Plant
The South Caboolture Water Reclamation Plant was designed
to reduce riverine pollution from the 40,000 population
equivalent wastewater treatment plant and to provide recycled
water to industry and community consumers. Whilst the
plant provides water for non-potable applications, it has been
designed to meet drinking water standards. The treatment
process detailed in Fig. 1 incorporates biological denitrification,
pre-ozonation, coagulation/flocculation/dissolved air
flotation-sand filtration (DAFF), main ozonation, biological
activated carbon filtration and final ozonation for disinfection.
The activated carbon was renewed in March 2008 after 9 years
of service, its adsorption capacity was assumed not to be
exhausted at the time sampling took place, 4 months later.
van Leeuwen et al. (2003) published more details on the
process and its performances.
2.2. Sample collection
Four sets of samples were collected over winter 2008 under
dry weather conditions including three during week days and
one during the weekend (11-07-08, 22-07-08, 27-07-08 and 06-
08-08). Water temperature across the plant was 22 2 C and
pH was 7.0 0.5. Samples were collected at 7 sampling points
along the treatment train, labelled S1 to S7 on Fig. 1, in order to
evaluate the performance of individual treatment steps. As
the flow rate in the reclamation plant is constant (8 megalitres
per day), representative samples were collected as time
proportional 24-h composites. For S1 and S7, refrigerated
auto-samplers collected 200 mL every 15 minutes; for the
remaining sampling locations, a continuous flow pump
collected samples at a flow rate of 7 mL min -1 . At each point,
samples were collected into a glass bottle pre-washed with
MilliQ water and HPLC grade acetone. The samples were
protected from light and refrigerated during collection. For
micropollutant analysis, 1 L was transferred into solvent

washed amber glass bottles and preserved with sodium thiosulfate.
For the bioassays, 2 L were transferred into solvent
washed amber glass bottles and hydrochloric acid (36%) was
added to a final concentration of 5 mM for preservation. For
dissolved organic carbon (DOC) measurements, 100 mL were
filtered through a 0.45 mm PTFE membrane and collected in
plastic (HDPE) bottles rinsed with the sample beforehand. All
samples were transported on ice and stored frozen or at 4 C.
2.3. Chemical analyses
2.3.1. Dissolved organic carbon
The dissolved organic carbon (DOC) was measured as nonpurgable
organic carbon (NPOC) with an Analytik Jena multi
N/C 3100 instrument at the Advanced Water Management
Centre (AWMC). For each sample, 2d3 replicates were
measured, giving a relative standard deviation of less than 3%.
2.3.2. Micropollutants
Micropollutant analysis was carried out by Queensland Health
Forensic and Scientific Services (QHFSS, accredited by the
National Association of Testing Authorities, Australia, and ISO
9001 certified). The method consisted of solid phase extraction
(SPE), concentration and quantification by liquid chromatography
coupled with tandem mass spectrometry (LC/MSdMS).
This method, described in details in the supporting information
SI 1, allowed the quantification of 85 compounds (listed in
Table SI 3 in the supporting information SI 2) selected on the
water research 44 (2010) 625–637 627
Fig. 1 – South Caboolture Water Reclamation Plant process (adapted from van Leeuwen et al. (2003)). S1 to S7: sampling
locations. HRT: hydraulic retention time. SRT: sludge retention time.
basis of quantity of usage of the particular compounds, their
potential toxicity and their resistance to degradation. Their
limit of quantification (LOQ) was 0.01 mgL -1 in most cases.
Concentrations were calculated using an internal calibration
method. Results were not corrected for recovery efficiency of
the extraction method. Recoveries of single compounds are
assumed to be consistent along the treatment as the matrix is
similar therefore; calculation of the removal efficiency is not
affected. Indeed, Gros et al. (2009) determined the recoveries
of 73 pharmaceutical compounds by SPE in raw wastewater
and treated effluent and their results show that the recovery
for a given compound generally vary by less than 20% from
one matrix to the other despite the fact that these matrices are
very different. The standard deviation of the method is 20%
(Table SI 1 in the supporting information SI 1). When the
reported outlet concentration was below the LOQ of the
compound, removal efficiency was calculated as a minimum
value using the LOQ as outlet concentration.
2.3.3. Bioassays
Six bioassays, discussed in details in Macova et al. (2010), were
applied to the samples collected in the study. Baseline toxicity
was determined with the non-specific Vibrio fischeri bioluminescence
inhibition test. Five additional specific toxic modes
of action were also evaluated: estrogenic activity (E-SCREEN
assay), arylhydrocarbon receptor response (CAFLUX assay),
neurotoxicity (acetylcholinesterase inhibition assay), phytotoxicity
(PSII inhibition I-PAM assay) and genotoxicity (umuC

628
assay). Water samples were extracted by SPE using Oasis HLB
cartridges. Full dose response curves were determined for
a serial dilution of the extract for each bioassay. Results were
expressed as toxic equivalent concentrations (TEQ) except for
the umuC assay. The TEQ represents the concentration of
a given reference compound that would be required to
produce the same effect as the mixture of the various
compounds in the sample. When the outlet TEQ was below
the LOQ of the bioassay, removal efficiency was calculated as
a minimum value using the LOQ as outlet TEQ. In the umuC
assay, the response is determined as an induction ration (IR),
an IR 1.5 is considered genotoxic. For genotoxic samples,
ECIR1.5 corresponds to how many times the sample must be
concentrated or diluted to elicit an IR of 1.5. Results are
expressed as 1/ECIR1.5 therefore a higher number represents
a higher genotoxic effect.
3. Results and discussion
3.1. Micropollutants
In the reclamation plant’s influent, 54 of the 85 targeted
compounds had a median concentration above their LOQ
confirming that conventional activated sludge treatment does
not completely remove these micropollutants from wastewater
(Fig. 2). The concentrations ranged from 0.01 to
2.10 mgL -1 with the exception of gabapentin, which was
consistently found at higher concentrations ranging from
5.60–6.50 mgL -1 (data provided in the supporting information
in Table SI 3). The factor between the minimum and the
maximum concentrations measured for each individual
compound was generally close to or lower than 2, with
a maximum of 3.6 observed for iopromide. No clear pattern
Fig. 2 – Number of compounds quantified and dissolved
organic carbon (DOC) after indicated stage along the
treatment train. Bars represent the number of compounds
(left y-axis) with a median concentration above the LOQ
(4 samples). Dots represent DOC on two different sampling
days (right y-axis). Denit: denitrification; Pre-O3: preozonation;
DAFF: dissolved air flotation-sand filtration;
Main O3: main ozonation; AC: activated carbon; Fin O3:
final ozonation.
water research 44 (2010) 625–637
could be distinguished between the different sampling days.
The increase or decrease of single compound concentrations
from one day to another appeared to be random, even when
comparing the sample collected during the weekend to
samples collected during week days. Twenty-five compounds
had an influent median concentration above 0.10 mgL –1 , their
fate is further detailed hereafter. The removal efficiency of
these compounds was determined in each treatment step
except when the concentration before treatment was lower
than ten times the LOQ and below LOQ after treatment. This
criterion was used to allow the determination of removals up
to 90% in any case. The DOC was measured for two sets of
samples (22-07-08 and 06-08-08), and varied from 14.2 to
19.7 mg L -1 in the influent water. The following paragraphs
discuss the performance of each individual stage of the
treatment by comparing concentrations of the selected
compounds in the inlet and outlet water for the considered
process.
3.1.1. Denitrification
The number of compounds above LOQ was unchanged after
denitrification and the concentrations of the 25 selected
compounds decreased by less than 20% (with the exception of
atenolol, 38%). To our knowledge, the removal of micropollutants
by denitrifying bacteria has not been specifically
studied to date. In the present case, methanol is added to the
water prior to the denitrification reactor and the bacteria use it
as the primary electron donor which may prevent or reduce
the degradation of micropollutants. The observed increase in
DOC after the denitrification was probably due to residual
methanol.
3.1.2. Pre-ozonation
The number of compounds quantified was stable through the
pre-ozonation process (Fig. 2) and the concentration of the 25
selected compounds decreased by less than 30%. The ozone
dose applied in the pre-ozonation step is approximately
2mgL -1 and the DOC was approximately 20 mg L -1 corre-
-1
sponding to an ozone dose of 0.1 mgO3 mgDOC.
In such conditions,
ozone is rapidly consumed by the organic matter (Buffle
et al., 2006a,b) therefore micropollutants oxidation is limited.
Ozonation has been proved to be very effective for oxidising
various micropollutants in secondary treated wastewaters
(Hollender et al., 2009; Huber et al., 2005; Snyder et al., 2006;
Ternes et al., 2003; Wert et al., 2009) but with higher ozone
-1
doses of at least 0.25 to 0.50 mgO3 mgDOC. Snyder et al. (2006)
also investigated a full scale wastewater treatment plant
using pre-ozonation with a low ozone dose (not specified)
prior to biological activated carbon filtration, and similarly,
observed removal of micropollutants inferior to 30%.
3.1.3. Coagulation/flocculation/DAFF
In the coagulation/flocculation/DAFF process, the negatively
charged colloids are neutralised by the addition of a chemical
agent (aluminium sulphate) which causes them to aggregate
in floccs (flocculation) which are subsequently removed by
dissolved air flotation and sand filtration. The main purpose of
this stage is to decrease the DOC, which was achieved effectively
as demonstrated by the 40–50% DOC reduction (Fig. 2).
This reduction of the DOC is important to achieve maximal

performance of the main ozonation process as discussed
above. After the DAFF, 53 compounds were quantified and the
concentrations of the 25 selected compounds decreased by
less than 20% except for atenolol (42%), caffeine (29%), gabapentin
(44%), gemfibrozil (32%) and roxithromycin (37%). It is
also interesting to note that paracetamol concentration
increased by 44% in the process, which could not be explained.
No literature reporting investigations of the removal of
micropollutants from treated wastewaters by coagulationflocculation
could be identified. However, several studies on
lab-scale and full scale drinking water treatment systems
have reported removals lower than 50% for several pharmaceuticals
and pesticides by such solids removal processes
(Adams et al., 2002; Ternes et al., 2002; Thuy et al., 2008; Vieno
et al., 2006; Westerhoff et al., 2005). Suarez et al. (2009) also
showed similar limited removal in hospital wastewaters.
Nevertheless, Suarez et al. (2009) and Westerhoff et al. (2005),
showed that removals of up to 80% can be achieved for highly
hydrophobic compounds (i.e logKow > 6) suggesting that the
micropollutants removal during coagulation-flocculation
occurs via hydrophobic interactions with neutral particles.
Most of the pharmaceuticals and pesticides measured in this
study are hydrophilic or of moderate hydrophobicity with
logKow < 4(Table SI 3 in supporting information SI 2); therefore
little removal can be expected in the coagulation/flocculation/
DAFF stage. Among the exceptions observed, the removal of
gemfibrozil can be explained by its more hydrophobic nature
(logKow ¼ 4.77, EPI SUITE 4.0) but other compounds are very
hydrophilic and no explanation for their removal could be
advanced.
3.1.4. Main ozonation
-1
The main ozonation stage (approximately 0.5 mgO3 mgDOC) decreased the concentration of 26 compounds below LOQ
(Fig. 2). Among the 25 selected compounds, 9 showed
water research 44 (2010) 625–637 629
a reduction of more than 90% and 13 others were reduced by
more than 70% while iopromide and gabapentin were reduced
by 55% (Fig. 3). Removal of naproxen was not determined
because its concentration was lower than ten times its LOQ
before treatment and below LOQ after. The higher removal
efficiency obtained by the main ozonation stage compared
with the pre-ozonation is due to the higher ozone dosage
relative to the DOC content. These results are in agreement
with previous findings (Buffle et al., 2006b; Hollender et al.,
2009; Huber et al., 2005; Snyder et al., 2006; Ternes et al., 2003;
Wert et al., 2009). The reaction of organic compounds with
molecular ozone is selective and only certain groups of
compounds react rapidly, e.g. aliphatic molecules with double
bonds, deprotonated amines and aromatics with an activating
group. Most of the pharmaceuticals and pesticides possess
one of these moieties and would be expected to react with
molecular ozone with high second order rate constants (kO3).
The results show that the removal of micropollutants
increased with increasing value of kO3 but even compounds
weakly reactive with ozone (i.e. iopromide and MCPA; Table 1)
were oxidised. Oxidation in ozonation processes can also
occur via the attack of hydroxyl radicals formed by ozone
decomposition in water. Hydroxyl radicals are highly reactive
with almost any organic molecule; the second order rate
constants of oxidation by hydroxyl radicals are typically in the
10 9 M -1 s -1 range (Table 1). Therefore, the observed removals of
iopromide and MCPA suggest a substantial exposure to
hydroxyl radicals. Indeed, Buffle et al. (2006a,b) showed that
the ozonation of wastewaters leads to high exposure to
hydroxyl radicals; in the first instants of the reaction, the
hydroxyl radical production is equivalent to, or greater than
what can be obtained with advanced oxidation processes. To
our knowledge, gabapentin oxidation by ozone has not been
reported in the literature and no reaction rate data could be
found. Gabapentin possesses both carboxylic and amine
Fig. 3 – Median removal of selected compounds (median of influent concentration > 0.10 mgL -1 ) by the main ozonation stage
and the combination of the main ozonation and the activated carbon (AC) filtration stages. Error bars represent minimum
and maximum removal, no error bar means that the compound was below LOQ after treatment; therefore removal was
calculated as a minimum using the LOQ. C S4: concentration before main ozonation; C S5: concentration after main ozonation;
CS6: concentration after activated carbon filtration.

630
Table 1 – Removal of selected pharmaceuticals in the main ozonation stage and their first order reaction rate constants with
ozone (kO3) and hydroxyl radicals (kHO ). The temperature (T ) and pH given are the experimental conditions used for the
determination of the reaction rates.
Compound Removal kO3 (M 1 s 1 ) pH T ( C) kHO (109.M 1 s 1 ) pH T ( C) Ref.
Iopromide 55% 91% 3.0 10 4 –1.0 10 6
2 20
Trimethoprim >93% 2.7 10 5
7 20 6.5 0.2 7 25
Sulphamethoxazole >93% ~2.5 10 6
7 20 5.5 0.7
5.5 10 5
7 20 7 25
Diclofenac >94% ~10 6
7 20 7.5 1.5
(1.84 0.15) 10 4
6 25 7 25
Paracetamol >96% 4.11
6 k
10 7 25 2.2 0.017 5.5 25
Carbamazepine >98% ~3 10 5
7 20 8.8 1.2 7 25
(7.81 1.31) 10 4
25
2.05 0.14 5 room
a Huber et al. (2003).
b Benitez et al. (2004).
c Benner et al. (2008).
d Song et al. (2008).
e Dodd et al. (2006).
f Rivas et al. (2009).
g Vogna et al. (2004b).
h Andreozzi et al. (2003).
i Andreozzi et al. (2002).
j Vogna et al. (2004a).
k Calculated from Andreozzi et al (2003).
groups, their pKa are 3.89 and 9.56 respectively (Zour et al.,
1992). Therefore gabapentin is present as a zwitterion under
usual wastewater pH conditions. Protonated amines and
deprotonated carboxylic acids are typically not highly reactive
with ozone (Hoigné and Bader, 1983) and gabapentin oxidation
can be expected to be controlled by hydroxyl radical reactions.
As a primary amine, gabapentin has an expected second order
rate constant for the reaction with ozone of about 50 M -1 s -1
which is significantly higher than for iopromide. However, the
similar behaviour of the two substances shows that their
decrease is controlled by hydroxyl radical reactions. The DOC
decreased by less than 10% in the process, showing that
ozonation does not lead to substantial mineralization of dissolved
organic matter.
3.1.5. Activated carbon filtration
Activated carbon filtration further removed the micropollutants
and only 2 compounds were quantified above their
LOQ downstream: roxithromycin (0.01 mgL -1 ) and gabapentin
(0.70 mgL -1 ). The removal efficiency of the compounds having
a median concentration of at least ten times their LOQ prior to
activated carbon filtration was calculated: oxazepam, tramadol
and venlafaxine were removed by more than 90% and
gabapentin was removed by 53%. Adsorption onto activated
carbon also removed 20–30% of the DOC showing a less
effective adsorption of organic matter than of micropollutants.
Previous studies (Ormad et al., 2008; Snyder et al.,
2007; Ternes et al., 2002; Westerhoff et al., 2005) demonstrated
water research 44 (2010) 625–637
that powdered and granular activated carbon can efficiently
remove micropollutants from natural water sources used for
drinking water. To date, the removal of micropollutants from
wastewaters by activated carbon has not been extensively
studied; Snyder et al. (2007) found that a water reuse facility
using granular activated carbon without regular replacement/
regeneration provided little removal. In the present work, the
activated carbon had been replaced about 4 months before the
collection of the samples so it is assumed that its adsorption
capacity was not yet exhausted. Adsorption onto activated
carbon is driven mainly by hydrophobic interactions and
Westerhoff et al. (2005) observed that the trend in the removal
efficiency of micropollutants by activated carbon can be predicted
from logKow values. Gabapentin is zwitterionic at
neutral pH as shown above and much more hydrophilic
(logKow ¼ -1.37) compared to oxazepam, tramadol and venlafaxine
(logKow ¼ 2.32, 3.01 and 3.28 respectively). This
explains the difference observed in the removal efficiencies of
these compounds. It can be expected that, with time, the
adsorption capacity of the activated carbon filter will decrease
and eventually be exhausted while biological activity will
develop in the filter and contribute to biodegradation of the
by-products formed by ozonation (Simpson, 2008).
3.1.6. Final ozonation
In the final effluent, after final ozonation, 4 compounds had
a median concentration above the LOQ: gabapentin (0.45 mgL -1 ),
roxithromycin (0.01 mgL -1 ), DEET (0.03 mgL -1 ) and caffeine
a
b
c
d
a
e
c
d
f
e
a
e
a
g
h
a
i
j

(0.02 mgL -1 ). The median concentrations of caffeine and DEET
are very close to the LOQ and are below the LOQ prior to the
final ozonation stage; therefore one can assume that their
quantification is due to analytical variations rather than an
actual concentration variation. It is difficult to evaluate the
efficiency of the final ozonation in the removal of micropollutants
because they all had concentrations equal or below
the LOQ before the treatment except gabapentin. The median
removal of gabapentin in the final ozonation was only 20%,
which is low compared to the main ozonation, and is due to
-1
the lower ozone dose employed (approximately 0.3 mgO3 mgDOC). The DOC was not removed in this treatment step.
3.1.7. Overall treatment
The full treatment decreased the concentration of 50 of the 54
compounds quantified in the influent water to levels below
LOQ (Fig. 2). Concomitantly, DOC was also reduced by 55–60%
in the effluent compared to the influent water. Overall, among
the 25 selected compounds, 11 were removed by more than
95% and 11 by more than 89%. The median removal of gabapentin
was 86% and the removals of naproxen and iopromide
were not calculated because their concentration was lower
than 10 times their LOQ in the influent and below their LOQ in
the effluent. The 4 remaining compounds were gabapentin
(0.45 mgL -1 ), roxithromycin (0.01 mgL -1 ), DEET (0.03 mgL -1 ) and
caffeine (0.02 mgL -1 ).
The first 3 stages of the treatment train (i.e. denitrification,
pre-ozonation and coagulation/flocculation/DAFF) did not
decrease the number of micropollutants quantified in the
water (Fig. 2). After these 3 stages, the concentrations of the 25
compounds that had an influent median concentration of at
least 0.10 mgL -1 were generally still greater than 50% compared
to the influent concentration (Fig. 4). The main ozonation and
water research 44 (2010) 625–637 631
the activated carbon adsorption played a key role in the
treatment bringing the concentration of 26 and 25 compounds
below LOQ respectively (Fig. 2). The main ozonation generally
decreased the micropollutants to less than 20% of their
influent concentration and activated carbon filtration further
removed the compounds to levels below LOQ except for
gabapentin and roxythromycin (Fig. 4). The combined effects
of the main ozonation and the activated carbon filtration
processes decreased the concentration of 10 of the 25 selected
micropollutants by more than 95% and by more than 89% for
12 of the 15 remaining compounds compared to their
concentration prior to the main ozonation (Fig. 3). Gabapentin
concentration was reduced by 79%. These results show that
ozonation followed by activated carbon filtration is a very
effective combination of processes to remove micropollutants
from secondary treated wastewater. The key steps in the
removal of the DOC were the DAFF and the activated carbon
adsorption. Although the DAFF reduced the concentration of
micropollutants by less than 30%, it also played a key role
indirectly by reducing the DOC which enhanced the performances
of the main ozonation.
3.2. Bioanalytical results
The influent biological activity was higher than the blank
(MilliQ water) in all the bioassays (Table 2). The effect of the
treatment train on each bioassay is discussed individually
hereafter.
3.2.1. Baseline toxicity
The Vibrio fischeri bioluminescence inhibition test is a nonspecific
bacterial toxicity test widely recognised in the field of
ecotoxicology as the standard assay for acute cytotoxicity. The
Fig. 4 – Median relative concentrations of selected compounds (median of influent concentration > 0.10 g L -1 ) after the
dissolved air flotation-sand filtration (DAFF, S4), the main ozonation (S5) and the activated carbon (S6) stages (error bars
represent maximum and minimum values). C is the concentration after the specified treatment step and the reference
concentration, C0, is the concentration in the influent water of the reclamation plant.

632
Table 2 – Maximum, median and minimum biological activity in the influent and effluent of the reclamation plant and
overall maximum, median and minimum decrease observed.
Bioassay Result expression Blank Influent Effluent Decrease (%)
assay reflects the general ‘‘energy status’’ of the bacteria and
can indicate the toxic potency of a broad spectrum of
compounds with different modes of action. Denitrification and
pre-ozonation did have a slight stimulatory effect on the
baseline toxicity whereas the following treatment steps were
able to substantially decrease baseline toxicity (Fig. 5). The TEQ
was reduced to 62, 37 and 21% of the influent water level after
the coagulation/flocculation/DAFF, the main ozonation and
the activated carbon filtration respectively. As is discussed in
more details in Macova et al. (2010), an almost linear correlation
exists between DOC level and TEQ. The slight increase in
TEQ after denitrification cannot be related to the residual
methanol added before this process, which is still partially
present after denitrification as is discussed above, but is likely
to be related to some non-volatile organic chemicals. The 52%
decrease of TEQ in the DAFF stage is accompanied by a 40–50%
reduction in DOC, whereas micropollutants concentrations
were generally reduced by less than 30% in the meantime.
Although the SPE that is performed prior to toxicity testing
should be able to remove a substantial fraction of the DOC and
will remove all of the residual methanol, some DOC, most
likely smaller breakdown products that have similar
Max Med Min Max Med Min Max Med Min
Baseline toxicity Baseline toxicity EqC a (TEQ, mg L 1 ) 0.21 2.9 2.1 2.0 0.72 0.52 0.31 84 78 67
Estrogenicity Estradiol EqC (EEQ, ng L 1 )

Activated carbon filtration reduced the baseline toxicity by
50% and the DOC by 30 to 35%. Activated carbon effectively
adsorbs the more hydrophobic compounds, which is again
consistent with the general trend discussed above that the
more hydrophobic compounds have a higher toxic activity
than the more hydrophilic ones. Based on this fact, identification
of the compounds exhibiting a high toxic activity could
start with the identification of the more hydrophobic
compounds.
The final ozonation did not further reduce the baseline
toxicity compared to after activated carbon filtration. The
effluent TEQ was approximately 80% lower than the
influent TEQ (Fig. 5) and only 2.5 times higher than
the blank (Table 2). This , indicates that the residual toxicity
is of no concern, unless the residual organic chemicals and
organic matter inducing this effect were of very specific
potency. This latter question was tested with a series of
specific endpoints that respond to environmentally relevant
modes of toxic action.
3.2.2. Estrogenic activity
The E-SCREEN assay specifically responds to natural
hormones and other compounds that can mimic the activity
of the female sex hormone estradiol. The estrogenic activity of
the samples is expressed as an estradiol equivalent concentration
(EEQ). The median influent EEQ was 5.8 ng L -1 ; higher
than levels previously reported in South East Queensland.
Most of the effluents from 12 activated sludge wastewater
treatment plants tested by Leusch et al. (2006) had EEQs below
4ngL -1 and sometimes below 1 ng L -1 . Hormones and endocrine
disrupting compounds (EDCs) were not investigated in
this study but an earlier sampling campaign of the secondary
treated wastewater showed that the concentrations of
measured estrogenic compounds (17b-estradiol, 17a-ethynylestradiol,
estrone, estriol, bisphenol A, nonylphenol) were all
below the LOQ of 1 ng L -1 (Table SI 4 in supporting information
SI 3). This demonstrates the relevance of using bioassays as
complementary tools to chemical analysis for the assessment
of water quality and process performances.
Denitrification did not affect the estrogenicity (Fig. 5). Preozonation
with an ozone dose of approximately 0.10 mgO3
-1
mgDOC reduced the EEQ by 34% compared to the influent. This
is higher than the removal previously observed by Snyder
et al. (2006) who measured the EEQ reduction induced by
various ozone doses in treated wastewater with a DOC of
6.38 mg L -1 . They found that an ozone dose of 2.1 mg L -1
1
(0.33 mgO3 mgDOC)
only removed 18% of the EEQ but, with
ozone doses of 3.6 mg L -1 1
(0.56 mgO3 mgDOC)
and above, 90%
or more removal could be achieved. In a recent study on a full
scale ozonation in a Swiss sewage treatment plant (STP), the
dose-dependency of removal of micropollutants yielded
similar results (Escher et al., 2009). While most endpoints
showed a clear dose-dependency of reduction of effects, the
reduction of estrogenicity was already large at low ozone
doses and depended more on the EEQ than on the ozone
dose. When estrogenicity was already below a certain level,
which was very close to the detection limit, the quantification
of further reduction became difficult and prone to large
uncertainty. For the remaining samples, ozone doses of 1.6–
5.3 mg L -1 in the presence of 4.2–6.0 mg L -1 DOC lead to more
water research 44 (2010) 625–637 633
than 90% reduction of estrogenicity. This is consistent with
laboratory experiments that demonstrated that almost all
first-generation transformation products of estrogenic
chemicals had severely decreased estrogenic potency (Lee
et al., 2008). Thus ozonation can be considered as a fairly
selective oxidation, where even low doses selectively target
one of the most environmentally relevant modes of toxic
action, namely estrogenicity.
After the coagulation/flocculation/DAFF stage the EEQ
increased drastically by a median factor of 3.3 compared to the
level prior to treatment. At this treatment step, the concentration
of DOC is greatly reduced (by 40–50%), and there is
a likelihood that the estrogenic chemicals that were bound to
DOC were released during this treatment step. We have
previously observed with another estrogenicity assay that
DOC appears to reduce the bioavailability of estrogens
(B. Escher, unpublished results). Estrogenic chemicals are
typically relatively hydrophobic and bind well to DOC (Neale
et al., 2008). In general DOC is not bioavailable in bioassays
(the discussion on the small breakdown products above is an
exception of this general paradigm) and micropollutants sorbed
to DOC would not be bioavailable either. A large fraction of
the matrix and also the DOC is supposed to be removed by SPE
but, given the color of the extracts, it is possible that
a substantial fraction of larger DOC is co-extracted. In addition,
for the E-SCREEN test, it was demonstrated that the
presence of serum proteins modulates the free and bioavailable
concentration of estrogenic chemicals (Heringa et al.,
2004). This effect was also hydrophobicity dependent and was
much more pronounced for the more hydrophobic octylphenol
than for the less hydrophobic estradiol. Protein binding is
generally less important than binding to DOC or lipids,
therefore while the effect on bioavailability was not very large
for estradiol in the study of Heringa et al. (2004); it might well
be relevant under the conditions of the present study. This
hypothesis needs to be evaluated in the future by exploring
the correlation between size distribution of naturally occurring
DOC and effect on bioavailability, estrogenicity and
toxicity.
The main ozonation reduced the EEQ by a median value of
92 and 95% compared to the level of the reclamation plant’s
influent and to the level before treatment respectively
whereas DOC was not affected. It can be concluded that the
mixture of by-products formed by the oxidation of the estrogenic
compounds by ozone and hydroxyl radicals have
a much lower estrogenic activity than the mixture of parent
compounds, which is consistent with expectations as discussed
above and in Lee et al. (2008).
Activated carbon filtration was able to efficiently adsorb
residual estrogenic compounds and further reduced the EEQ
by another 95% to levels below the detection limit of 0.02 ng L -1
and the final effluent concentration was below the quantification
limit of 0.06 ng L -1 . The overall treatment efficiency for
the removal of estrogenic activity was greater than 99%. This
is in good agreement with observations on a full scale ozonation
in a Swiss STP (Escher et al., 2009). As discussed above
the analytically determined concentrations of (xeno)estrogens
were below the quantification limit, therefore for this
endpoint the very sensitive bioassay poses a great advantage
despite the observed limitations due to matrix effects.

634
3.2.3. Ah-receptor response
The CAFLUX assay targets dioxins and dioxin-like compounds
such as polychlorinated biphenyls (PCBs) but can also respond
to other chemicals such as polycyclic aromatic hydrocarbons
(PAHs) (Macova et al., 2010). The results of the test are
expressed as 2,3,7,8-tetrachlorodibenzo-p-dioxin equivalent
concentration (TCDDEQ). The median TCDDEQ of the influent
water was 0.82 ng L -1 and there was no significant variation
along the first three steps of the treatment process; i.e denitrification,
pre-ozonation and coagulation/flocculation/DAFF
(Fig. 5). The main ozonation removed about 50% of the TCDDEQ
but subsequent activated carbon filtration and final ozonation
did not show further important removal and the median
TCDDEQ of the final effluent was approximately 3.9 times
higher than the blank (Table 2). Two sets of samples were
submitted to a sulphuric acid silica gel clean up procedure that
aims at removing organic chemicals except those that are not
oxidised such as polychlorinated dibenzodioxins, -furans and
PCBs. The samples were then tested again with the CAFLUX
assay to evaluate the contribution of these very persistent
chemicals (i.e dioxins, furans and dioxin-like PCBs). Results
showed that after clean up the TCDDEQ was not significantly
different from the blank (values ranged from 0.09 to 0.11 ng L -1 ).
This shows that the effect induced by the samples without
sulphuric acid silica gel clean up is not due to the presence of
dioxins, furans or dioxin-like PCBs but was caused by other
chemicals. Since none of these groups of chemicals were
quantified by chemical analysis in this study, no comparison
between chemical and biological analysis is possible.
3.2.4. Genotoxicity
The umuC assay responds specifically to genotoxic compounds
that cause DNA damages. To detect genotoxic effects caused by
metabolites, the test is also performed in presence of a rat liver
extract that can transform indirect genotoxicants to metabolites
that are DNA damaging compounds. The median influent
1/ECIR1.5 were 0.19 and 0.060 in the absence and presence of the
rat liver extract respectively, showing that the sample was less
genotoxic after metabolisation. This is what one would
commonly expect, an exception would be PAHs that are activated
by metabolism. Denitrification and pre-ozonation did not
have a substantial influence on genotoxicity (Fig. 5). The coagulation/flocculation/DAFF
stage decreased 1/EC IR1.5 by 59%
compared to the influent. The main ozonation drastically
reduced the genotoxicity, 1/EC IR1.5 was reduced by 80 and 93%
compared to the DAFF effluent and to the influent of the plant
respectively. After activated carbon filtration as well as in the
final effluent, 1/ECIR1.5 was below the LOQ of the bioassay (Table 2).
In every case, the genotoxicity of the metabolised sample was
lower than the non-metabolised sample, indicating that the
type of chemicals inducing the genotoxic effect did not change
over the treatment.
3.2.5. Neurotoxicity
Neurotoxicity is measured by the inhibition of the enzyme
acetylcholinesterase (AChE). Organophosphate and carbamate
pesticides specifically bind to this enzyme and the
results are expressed as parathion equivalent concentration
(PTEQ). The median PTEQ in the secondary treated wastewater
was 3.1 mgL -1 ; denitrification and pre-ozonation did not
water research 44 (2010) 625–637
reduce the PTEQ whereas DAFF decreased it by 31% compared
to influent (Fig. 5). Unlike the other bioassays, the effect of the
main ozonation on PTEQ was not significant but activated
carbon filtration reduced it drastically to level below the
quantification limit of the bioassay (0.30 mgL -1 ) which represents
more than an 80% and 90% decrease compared to the
main ozonation effluent and the plant influent water respectively.
This observation is consistent with theoretical expectation,
as it is known that compounds like diazinon and
chlorpyrifos, which often constitute a large fraction of the
acetylcholinesterase inhibitors, are not well oxidised by
ozone. In contrast, these compounds are fairly hydrophobic
(logKow ¼ 3.96 and 4.66 respectively), therefore sorption to
activated carbon can be expected. A similar removal pattern
has been observed for acetylcholinesterase inhibitors in the
above-mentioned Swiss STP: none of the single removal steps
(biological treatment, ozonation, sand filtration) had a high
removal efficiency but all steps taken together produced
a satisfactory overall removal (Escher et al., 2009).
3.2.6. Phytotoxicity
The I-PAM assay is sensitive to herbicides that directly inhibit
photosynthesis; the results are reported as a diuron equivalent
concentration (DEQ). The DEQ of the influent water ranged from
0.05 to 0.22 mgL -1 with a median value of 0.10 mgL -1 (Table 2).
Diuron concentrations were measured by chemical analysis; it
was reported in every sample of the influent water from 0.02 to
0.04 mgL -1 , suggesting that its contribution to the effect observed
was limited. Among the other herbicides quantified, only
simazine is also a photosystem II inhibitor with a relative
potency of 0.15 (Muller et al., 2008). Simazine concentrations in
the influent ranged from 0.05 to 0.19 mgL -1 . These two
compounds considered together accounted for 17–93% of the
measured DEQ demonstrating the interest of this bioassay to
take into account non-measured compounds. The DEQ
increased by factors of 2.2 and 3.5 after denitrification and preozonation
respectively but variation from one day to another
was large therefore it is difficult to draw a conclusion (Fig. 5). This
increase was accompanied with a slight increase in baseline
toxicity and could therefore be caused by baseline toxicants
interfering with the measurement of the photosynthesis yield
(Macova et al., 2010). The coagulation/filtration/DAFF stage
reduced DEQ by 67% and 88% compared to the plant’s influent
water and to the pre-ozonated water respectively. After the
main ozonation the DEQ was not further significantly reduced;
Diuron’s concentrations were equal to or below the LOQ of
0.01 mgL -1 and simazine was removed by approximately 50%,
their contribution accounting for 16–38% of the observed DEQ.
Subsequent activated carbon filtration and final ozonation did
not further affect the DEQ but diuron and simazine were
removed below their LOQ. The overall treatment achieved 75%
median decrease of DEQ, the effluent median DEQ was 0.03 mgL -1
(Table 2).
3.2.7. Overall treatment
The biological activity of the effluent was lower compared to
the influent and close or equal to the blank level showing that
the treatment process could effectively decrease the biological
adverse effects observed with the bioassays; from 62% for the
AhR response to more than 99% for estrogenicity (Table 2). The

key treatment steps responsible for the decrease of biological
activity are the DAFF stage, the main ozonation and the activated
carbon adsorption (Fig. 5).
3.3. Indirect potable reuse considerations
Micropollutants reported concentrations were compared to the
guidelines values for indirect potable reuse given in the
Australian Guidelines for Water Recycling: Augmentation of
Drinking Water Supplies (Table SI 3 in supporting information
SI 2). The reported concentrations of the measured compounds
were found to be below the guideline values in the influent
water of the reclamation plant before any treatment. After
removal by the advanced treatment process, concentrations
were several orders of magnitude below the guideline values.
For information purpose, median equivalent concentrations
obtained with the bioassays were compared to the corresponding
reference compound’s guideline value when
available. Note however, that the effect caused by a mixture
cannot be compared directly to a guideline value of a single
compound. Moreover, the bioassays used here are acute tests
and no conclusions can be drawn about chronic effects.
Nevertheless such a comparison gives an impression on the
expected hazard of the mixture but must be communicated
with caution to a lay audience. For estrogenicity, neurotoxicity
and phytotoxicity the reference compounds were estradiol,
parathion and diuron and the guidelines values were 175 ng L -1 ,
10 mgL -1 and 30 mgL -1 respectively. Similarly to individual
compounds concentrations, the bioassays equivalent
concentrations were already below the guidelines values in
the water entering the reclamation plant. Final effluent
median equivalent concentrations were also several orders of
magnitude below the corresponding guideline values, i.e.
more than 2900, 33 and 428 fold for estrogenicity, neurotoxicity
and phytotoxicity respectively.
For the parameters considered, the water quality complies
with the requirements of the Australian Guidelines for Water
Recycling: Augmentation of Drinking Water Supplies. This
suggests that such a treatment train could be considered as an
alternative to the combination of microfiltration and reverse
osmosis for indirect potable reuse schemes. It has the advantage
of not producing a waste stream and would be certainly
less energy intensive. Nevertheless, before this process can be
recommended for indirect potable reuse, additional consideration
needs to be given to the overall risk management
strategies of the treatment train. Moreover, the removal of
pathogens such as viruses and bacteria has to be assessed as
well as the potential to form disinfection by-products due to
the remaining DOC levels.
4. Conclusions
The assessment of a tertiary treatment train regarding the
removal of micropollutants and decrease of biological activity
leads to the following conclusions:
1) The treatment train reduced the concentration of the 54
micropollutants quantified in the secondary treated
wastewater as well as the biological activity observed in
water research 44 (2010) 625–637 635
bioassays. Overall concentration reductions were typically
higher than 90% and most of the compounds were removed
to levels lower than 0.01 mgL 1 . The observed effects
decreased by 62% to more than 90% depending on the assay
and levels in the final effluent were close to or equivalent to
the blank’s. The effect of individual processes varied from
one compound and bioassay to another but the combination
of the coagulation/flocculation/DAFF stage, the main
ozonation and the activated carbon filtration was responsible
for the major part of the observed reduction.
2) The use of a battery of bioassays as complementary tools to
chemical analysis yielded valuable additional information on
the water quality and the process efficiency. The results
showed the limitations of chemical analysis to assess potential
biological adverse effects and the ability of bioassays to
take into account the presence of non-measured compounds,
formed transformation products and/or mixture effects.
3) Concurrent measurement of DOC, micropollutants and
biological activity showed that the DOC level can influence
the effect observed in the bioassays. Baseline toxicity was
shown to correlate well with the DOC levels. It is supposed
that non-targeted organic matter is co-extracted with the
targeted organic micropollutants in the SPE purification
step. More understanding of the nature and quantity of
these co-extracted compounds is needed.
4) Ozone dose relative to DOC content is a key parameter for
the ozonation performance. A low ozone dose of 0.1 mgO3
-1
mgDOC did not affect the concentration of micropollutants
-1
or biological activity whereas a dose of 0.5 mgO3 mgDOC was
able to remove most of the micropollutants by more than
70% and substantially decrease the biological activity.
5) Degradation products are a major concern in ozonation
processes. Here, the results from the battery of bioassays
show that the main ozonation leads to lower baseline and
specific toxic effects. It can be concluded that the mixture of
degradation products formed have an overall less harmful
potential than the mixture of parent compounds.
Acknowledgements
This work was co-funded by the Urban Water Security
Research Alliance under the Enhanced Treatment Project, the
CRC Water Quality and Treatment Project No. 2.0.2.4.1.1 –
Dissolved Organic Carbon Removal by Biological Treatment
and by EnTox. The National Research Centre for Environmental
Toxicology (EnTox) is a joint venture of the University
of Queensland and Queensland Health Forensic and Scientific
Services. The authors acknowledge the following institution
and individuals who contributed to this study: Moreton Bay
Water for access to the South Caboolture Water Reclamation
Plant; Ray McSweeny and Paul McDonnell (Moretom Bay
Water) for their help during sampling; Chris Pipe-Martin
(Ecowise) for providing information on the South Caboolture
Water Reclamation Plant; Geoff Eaglesham, Steve Carter and
Mary Hodge (Queensland Health Forensic and Scientific
Services) for the analysis of micropollutants and discussions
on the analytical method; Wolfgang Gernjak, Michael G.
Lawrence, Christoph Ort and Maria Jose Farre Olalla

636
(Advanced Water Management Centre) for fruitful discussions
and proof reading of the manuscript.
Appendix.
Supplementary data
Supplementary data associated with this article can be found
in the online version, at doi:10.1016/j.watres.2009.09.048.
references
Adams, C., Wang, Y., Loftin, K., Meyer, M., 2002. Removal of
antibiotics from surface and distilled water in conventional
water treatment processes. Journal of Environmental
Engineering–ASCE 128 (3), 253–260.
Andreozzi, R., Caprio, V., Marotta, R., Vogna, D., 2003.
Paracetamol oxidation from aqueous solutions by means of
ozonation and H2O2/UV system. Water Research 37 (5),
993–1004.
Andreozzi, R., Marotta, R., Pinto, G., Pollio, A., 2002.
Carbamazepine in water: persistence in the environment,
ozonation treatment and preliminary assessment on algal
toxicity. Water Research 36 (11), 2869–2877.
Benitez, F.J., Acero, J.L., Real, F.J., Roman, S., 2004. Oxidation of
MCPA and 2,4-D by UV radiation, ozone, and the combinations
UV/H2O2 and O-3/H2O2. Journal of Environmental Science
and Health Part B-Pesticides Food Contaminants and
Agricultural Wastes 39 (3), 393–409.
Benner, J., Salhi, E., Ternes, T., von Gunten, U., 2008. Ozonation of
reverse osmosis concentrate: Kinetics and efficiency of beta
blocker oxidation. Water Research 42 (12), 3003–3012.
Buffle, M.O., Schumacher, J., Meylan, S., Jekel, M., von Gunten, U.,
2006a. Ozonation and advanced oxidation of wastewater:
effect of O-3 dose, pH, DOM and HO center dot-scavengers on
ozone decomposition and HO center dot generation. Ozone:
Science and Engineering 28 (4), 247–259.
Buffle, M.O., Schumacher, J., Salhi, E., Jekel, M., von Gunten, U.,
2006b. Measurement of the initial phase of ozone decomposition
in water and wastewater by means of a continuous quench-flow
system: application to disinfection and pharmaceutical
oxidation. Water Research 40 (9), 1884–1894.
Dodd, M.C., Buffle, M.O., Von Gunten, U., 2006. Oxidation of
antibacterial molecules by aqueous ozone: Moiety-specific
reaction kinetics and application to ozone-based wastewater
treatment. Environmental Science & Technology 40 (6),
1969–1977.
Escher, B.I., Bramaz, N., Mueller, J.F., Quayle, P., Rutishauser, S.,
Vermeirssen, E.L.M., 2008a. Toxic equivalent concentrations
(TEQs) for baseline toxicity and specific modes of action as
a tool to improve interpretation of ecotoxicity testing of
environmental samples. Journal of Environmental Monitoring
10 (5), 612–621.
Escher, B.I., Bramaz, N., Ort, C., 2009. Monitoring the treatment
efficiency of a full scale ozonation on a sewage treatment
plant with a mode-of-action based test battery. Journal of
Environmental Monitoring 11 (10), 1836–1846.
Escher, B.I., Bramaz, N., Quayle, P., Rutishauser, S.,
Vermeirssen, E.L.M., 2008b. Monitoring of the ecotoxicological
hazard potential by polar organic micropollutants in sewage
treatment plants and surface waters using a mode-of-action
based test battery. Journal of Environmental Monitoring 10 (5),
622–631.
Esplugas, S., Bila, D.M., Krause, L.G.T., Dezotti, M., 2007.
Ozonation and advanced oxidation technologies to remove
water research 44 (2010) 625–637
endocrine disrupting chemicals (EDCs) and pharmaceuticals
and personal care products (PPCPs) in water effluents. Journal
of Hazardous Materials 149 (3), 631–642.
Gros, M., Petrovic, M., Barcelo, D., 2009. Tracing pharmaceutical
residues of different therapeutic classes in environmental
waters by using liquid chromatography/quadrupole-linear ion
trap mass spectrometry and automated library searching.
Analytical Chemistry 81 (3), 898–912.
Heringa, M.B., Schreurs, R., Busser, F., Van Der Saag, P.T., Van Der
Burg, B., Hermens, J.L.M., 2004. Toward more useful in vitro
toxicity data with measured free concentrations.
Environmental Science and Technology 38 (23), 6263–6270.
Hoigné, J., Bader, H., 1983. Rate constants of reactions of ozone with
organic and inorganic compounds in water–II: dissociating
organic compounds. Water Research 17 (2), 185–194.
Hollender, J., Zimmermann, S.G., Koepke, S., Krauss, M.,
McArdell, C.S., Ort, C., Singer, H., von Gunten, U., Siegrist, H.,
2009. Elimination of organic micropollutants in a municipal
wastewater treatment plant upgraded with a full-scale post-
Ozonation followed by sand filtration. Environmental Science
and Technology 43 (20), 7862–7869.
Huber, M.M., Canonica, S., Park, G.Y., von Gunten, U., 2003.
Oxidation of pharmaceuticals during ozonation and advanced
oxidation processes. Environmental Science and Technology
37 (5), 1016–1024.
Huber, M.M., Gobel, A., Joss, A., Hermann, N., Loffler, D.,
McArdell, C.S., Ried, A., Siegrist, H., Ternes, T.A., von
Gunten, U., 2005. Oxidation of pharmaceuticals during
ozonation of municipal wastewater effluents: a pilot study.
Environmental Science and Technology 39 (11), 4290–4299.
Joss,A.,Siegrist,H.,Ternes,T.A.,2008.Areweaboutto
upgrade wastewater treatment for removing organic
micropollutants? Environmental Science and Technology
57 (2), 251–255.
Kim, I.H., Tanaka, H., Iwasaki, T., Takubo, T., Morioka, T., Kato, Y.,
2008. Classification of the degradability of 30 pharmaceuticals
in water with ozone, UV and H2O2. Water Science and
Technology 57 (2), 195–200.
Kimura, K., Toshima, S., Amy, G., Watanabe, Y., 2004. Rejection of
neutral endocrine disrupting compounds (EDCs) and
pharmaceutical active compounds (PhACs) by RO membranes.
Journal of Membrane Science 245 (1–2), 71–78.
Lee, Y., Escher, B.I., Von Gunten, U., 2008. Efficient removal of
estrogenic activity during oxidative treatment of waters
containing steroid estrogens. Environmental Science and
Technology 42 (17), 6333–6339.
Leusch, F.D.L., Chapman, H.F., van den Heuvel, M.R., Tan, B.L.L.,
Gooneratne, S.R., Tremblay, L.A., 2006. Bioassay-derived
androgenic and estrogenic activity in municipal sewage in
Australia and New Zealand. Ecotoxicology and Environmental
Safety 65 (3), 403–411.
Macova, M., Escher, B.I., Reungoat, J., Carswell, S., Lee Chue, K.,
Keller, J., Mueller, J.F., 2010. Monitoring the biological activity
of micropollutants during enhanced wastewater treatment
with ozonation and activated carbon filtration. Water
Research 44 (2), 477–492.
Muller, R., Schreiber, U., Escher, B.I., Quayle, P., Nash, S.M.B.,
Mueller, J.F., 2008. Rapid exposure assessment of PSII
herbicides in surface water using a novel chlorophyll
a fluorescence imaging assay. Science of the Total
Environment 401 (1–3), 51–59.
Muller, R., Tang, J.Y.M., Thierb, R., Mueller, J.F., 2007. Combining
passive sampling and toxicity testing for evaluation of mixtures
of polar organic chemicals in sewage treatment plant effluent.
Journal of Environmental Monitoring 9 (1), 104–109.
Nakada, N., Shinohara, H., Murata, A., Kiri, K., Managaki, S.,
Sato, N., Takada, H., 2007. Removal of selected
pharmaceuticals and personal care products (PPCPs) and

endocrine-disrupting chemicals (EDCs) during sand filtration
and ozonation at a municipal sewage treatment plant. Water
Research 41 (19), 4373–4382.
Neale, P.A., Escher, B.I., Schafer, A.I., 2008. Quantification of
solute–solute interactions using negligible-depletion solidphase
microextraction: measuring the affinity of estradiol to
bulk organic matter. Environmental Science and Technology
42 (8), 2886–2892.
Nowotny, N., Epp, B., vonSonntag, C., Fahlenkamp, H., 2007.
Quantification and modeling of the elimination behavior of
ecologically problematic wastewater micropollutants by
adsorption on powdered and granulated activated carbon.
Environmental Science and Technology 41 (6), 2050–2055.
Onesios, K.M., Yu, J.T., Bouwer, E.J., 2009. Biodegradation and
removal of pharmaceuticals and personal care products in
treatment systems: a review. Biodegradation 20 (4), 441–466.
Ormad, M.P., Miguel, N., Claver, A., Matesanz, J.M., Ovelleiro, J.L.,
2008. Pesticides removal in the process of drinking water
production. Chemosphere 71 (1), 97–106.
Rivas, J., Gimeno, O., Encinas, A., Beltrán, F., 2009. Ozonation of
the pharmaceutical compound ranitidine: Reactivity and
kinetic aspects. Chemosphere 76 (5), 651–656.
Simpson, D.R., 2008. Biofilm processes in biologically active
carbon water purification. Water Research 42 (12), 2839–2848.
Snyder, S.A., Adham, S., Redding, A.M., Cannon, F.S., DeCarolis, J.,
Oppenheimer, J., Wert, E.C., Yoon, Y., 2007. Role of membranes
and activated carbon in the removal of endocrine disruptors
and pharmaceuticals. Desalination 202 (1–3), 156–181.
Snyder, S.A., Wert, E.C., Rexing, D.J., Zegers, R.E., Drury, D.D.,
2006. Ozone Oxidation of Endocrine Disruptors and
Pharmaceuticals in Surface Water and Wastewater. Ozone:
Science and Engineering 28 (6), 445–460.
Song, W.H., Cooper, W.J., Mezyk, S.P., Greaves, J., Peake, B.M.,
2008. Free radical destruction of beta-blockers in aqueous
solution. Environmental Science & Technology 42 (4),
1256–1261.
Suarez, S., Lerna, J.M., Omil, F., 2009. Pre-treatment of hospital
wastewater by coagulation-flocculation and flotation.
Bioresource Technology 100 (7), 2138–2146.
Ternes, T.A., Kreckel, P., Mueller, J., 1999. Behaviour and
occurrence of estrogens in municipal sewage treatment plants –
II. Aerobic batch experiments with activated sludge. Science of
the Total Environment 225 (1–2), 91–99.
Ternes,T.A.,Meisenheimer,M.,McDowell,D.,Sacher,F.,Brauch,H.J.,
Gulde, B.H., Preuss, G., Wilme, U., Seibert, N.Z., 2002. Removal of
water research 44 (2010) 625–637 637
pharmaceuticals during drinking water treatment.
Environmental Science and Technology 36 (17), 3855–3863.
Ternes, T.A., Stuber, J., Herrmann, N., McDowell, D., Ried, A.,
Kampmann, M., Teiser, B., 2003. Ozonation: a tool for removal
of pharmaceuticals, contrast media and musk fragrances from
wastewater? Water Research 37 (8), 1976–1982.
Thuy, P.T., Moons, K., van Dijk, J.C., Anh, N.V., Van der Bruggen, B.,
2008. To what extent are pesticides removed from surface
water during coagulation-flocculation? Water and Environment
Journal 22 (3), 217–223.
van Leeuwen, J., Pipe-Martin, C., Lehmann, R.M., 2003. Water
reclamation at South Caboolture, Queensland, Australia.
Ozone: Science and Engineering 25 (2), 107–120.
Vieno, N., Tuhkanen, T., Kronberg, L., 2006. Removal of
pharmaceuticals in drinking water treatment: effect of chemical
coagulation. Environmental Technology 27 (2), 183–192.
Vogna, D., Marotta, R., Andreozzi, R., Napolitano, A., d’Ischia, M.,
2004a. Kinetic and chemical assessment of the UV/H2O2
treatment of antiepileptic drug carbamazepine. Chemosphere
54 (4), 497–505.
Vogna, D., Marotta, R., Napolitano, A., Andreozzi, R., d’Ischia, M.,
2004b. Advanced oxidation of the pharmaceutical drug
diclofenac with UV/H2O2 and ozone. Water Research 38 (2),
414–422.
Wert, E.C., Rosario-Ortiz, F.L., Snyder, S.A., 2009. Effect of ozone
exposure on the oxidation of trace organic contaminants in
wastewater. Water Research 43 (4), 1005–1014.
Westerhoff, P., Yoon, Y., Snyder, S., Wert, E., 2005. Fate of
endocrine-disruptor, pharmaceutical, and personal care
product chemicals during simulated drinking water treatment
processes. Environmental Science and Technology 39 (17),
6649–6663.
Yoon, Y., Westerhoff, P., Snyder, S.A., Wert, E.C., Yoon, J., 2007.
Removal of endocrine disrupting compounds and
pharmaceuticals by nanofiltration and ultrafiltration
membranes. Desalination 202 (1–3), 16–23.
Yu, Z.R., Peldszus, S., Huck, P.M., 2008. Adsorption characteristics
of selected pharmaceuticals and an endocrine disrupting
compound – naproxen, carbamazepine and nonylphenol – on
activated carbon. Water Research 42 (12), 2873–2882.
Zour, E., Lodhi, S.A., Nesbitt, R.U., Silbering, S.B., Chaturvedi, P.R.,
1992. Stability studies of gabapentin in aqueous-solutions.
Pharmaceutical Research 9 (5), 595–600.
Zwiener, C., Frimmel, F.H., 2000. Oxidative treatment of
pharmaceuticals in water. Water Research 34 (6), 1881–1885.

Removal of natural hormone estrone from secondary effluents
using nanofiltration and reverse osmosis
Xue Jin 1 , Jiangyong Hu*, Say Leong Ong
Division of Environmental Science and Engineering, National University of Singapore, 10 Kent Ridge Crescent, 119260, Singapore
article info
Article history:
Received 19 December 2008
Received in revised form
1 June 2009
Accepted 23 September 2009
Available online 13 October 2009
Keywords:
Estrone
Nanofiltration
Reverse osmosis
Rejection
Treated effluent
Binding
1. Introduction
abstract
Recently, the use of treated municipal wastewater for
groundwater recharge and indirect drinking water reuse has
become promising worldwide. Nanofiltration (NF) and reverse
osmosis (RO) membranes are widely used in this kind of
environmental application (Schäfer et al., 2005). However, the
presence of endocrine disrupting chemicals (EDCs) in the
aquatic environment has raised great consumer concern due
to their potential health risk (Tim, 1997; Adams, 1998; Croley
water research 44 (2010) 638–648
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
The rejection of steroid hormone estrone by nanofiltration (NF) and reverse osmosis (RO)
membranes in treated sewage effluent was investigated. Four NF/RO membranes with
different materials and interfacial characteristics were utilized. To better understand
hormone removal mechanisms in treated effluent, effluent organic matters (EfOM) were
fractionated using column chromatographic method with resins XAD-8, AG MP-50 and IRA-
96. The results indicate that the presence of EfOM in feed solution could enhance estrone
rejection significantly. Hydrophobic acid (HpoA) organic fraction made a crucial contribution
to this ‘‘enhancement effect’’. Hydrophobic base (HpoB) could also improve estrone
rejection while hydrophobic neutral (HpoN) and hydrophilic acid (HpiA) with low aromaticity
had little effects. The increment in estrone rejection was predominantly attributed to
the binding of estrone by EfOM in feed solutions, which led to an increase in molecular
weight and appearance of negative charge (for the HpoA case) and thus an increased level
of estrone rejection. However, the improvement of estrone rejection by HpoA decreased
with increasing calcium ion concentration. The important conclusion of this study is, first,
hydrophobic acid macromolecules are recommended to be added into feed water to
improve the rejection of trace hormone during NF/RO membrane process, and, second,
removal of calcium ions via pretreatment and application of membrane with more negative
charge at its interface can greatly intensify this ‘‘enhancement effect’’.
ª 2009 Elsevier Ltd. All rights reserved.
et al., 2000; Baker, 2001; Iguchi et al., 2001). Amongst many
kinds of EDCs, the impact of steroid hormones such as
estrone, estradiol and ethinylestradiol is prominent because
their potency can be several thousand times higher than that
of others (Desbrow et al., 1998; Ternes et al., 1999; Johnson and
Sumpter, 2001). It is noteworthy to emphasize that concentrations
of some steroid hormones in secondary effluent are
often high enough to harm wildlife although it is within ng/L
range (Baronti et al., 2000; Williams et al., 2003; Soliman et al.,
2004). In the light of the magnitude of this problem, NF/RO
* Corresponding author. Tel.: þ65 65164540; fax: þ65 67744202.
E-mail addresses: jinxuesky@ucla.edu (X. Jin), esehujy@nus.edu.sg (J. Hu), cveongsl@nus.edu.sg (S.L. Ong).
1
Current address: Civil & Environmental Engineering Department, University of California, 5732 Boelter Hall, Los Angeles, CA 90095-
1593, USA
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.watres.2009.09.057

membranes and different post-treatment technologies are
essential to remove EDCs from treated effluent before water
reuse and ensure the safety of drinking water.
Removal of EDCs using NF/RO membranes in single-organic
solution (the experiments were conducted with target EDCs
spiked in synthetic model solutions in the absence of other
organics) has been reported in several recent studies (Kimura
et al., 2003, 2004; Schäfer et al., 2003; Nghiem et al., 2004a, b).
Many factors may influence the interaction between EDCs and
membrane and thus their rejection including physicochemical
properties of EDCs, such as molecular weight (MW), molecular
size, charge (as a function of dissociation constant), hydrophobicity
and polarity; membrane properties, such as material,
surface charge (zeta potential), hydrophobicity, molecular
weight cut-off (MWCO), porosity and desalting degree; and
solution chemistry, such as pH, ionic strength and presence of
divalent cations (Bellona et al., 2004, Schäfer et al., 2005).
Another important factor influencing the rejection efficiency is
initial concentration of target EDCs in feed solution. Kimura
et al. (2003) reported that experiments conducted at a ng/l
concentration range resulted in a lower rejection efficiency as
compared to experiments conducted at a mg/l range. This
observation substantiates that experiments conducted within
the ‘‘realistic’’ concentration range are necessary to provide an
accurate value of rejection efficiency. It should, however, be
pointed out that the results achieved from a single-organic
solution cannot be extrapolated to the mixtures typically
encountered in a real aquatic environment.
In previous works, we have investigated the influence of
other dissolved organic matter (DOM) in feed solution on the
removal of steroid hormone estrone during NF processes (Jin
et al., 2007). The results indicated that estrone rejection depended
on the type of DOM co-present. Neutral and highly hydrophilic
dextran had little effect on estrone rejection. The presence
of humic acid without phenolic groups but great aromaticity
improved estrone rejection slightly. In contrast, estrone rejection
was obviously increased by the presence of hydrophobic
acid fraction (HpoA) derived from secondary effluent. Therefore,
it is expected that the rejection mechanism of EDCs in real water
matrix, such as secondary effluent, is very complicated.
Organic matter in secondary effluent is referred to as
effluent organic matter (EfOM) and can be categorized into
fractions based on the difference in hydrophobicity and
charge properties using a column chromatographic method
(Leenheer, 1981; Thurman and Malcolm, 1981; Imai et al., 2002;
Hu et al., 2003). Adsorption of some fractions in EfOM onto
membranes would be expected to alter the membrane surface
properties, which subsequently results in changes in the
rejection of EDCs (Ng and Elimelech, 2004; Yoon et al., 2004; Xu
et al., 2006; Steinle-Darling et al., 2007). Alternatively, EDCs
can bind to some fractions in EfOM and be retained together,
which can also affect their rejection characteristics (Agbekodo
et al., 1996; Devitt et al., 1998; Majewska-Nowak et al., 2002;
Nghiem et al., 2004a, b; Hu et al., 2007; Jin et al., 2007). Both
interactions mentioned above (EfOM-membrane and EfOM-
EDCs) are markedly influenced by the physicochemical characteristics
of EfOM. Therefore, different fractions of EfOM
would demonstrate diverse impacts on EDCs rejection by
membranes. To our knowledge, however, very little information
is available on the rejection of trace EDCs by NF/RO
water research 44 (2010) 638–648 639
membranes with the presence of different fractions of EfOM.
In order to better understand the removal mechanism of
EDCs in secondary effluent by NF/RO membranes and optimize
membrane performance, it is important to clarify this
issue.
It is known that divalent cations (especially calcium) affect
the characteristics of EfOM, membrane charges and thus the
two above-mentioned interactions (EfOM–membrane and
EfOM–EDCs). For example, Devitt et al. (1998) reported that
atrazine–NOM association decreased in the presence of
calcium. Therefore, calcium ion is predicted to affect the
rejection of EDCs with the presence of other EfOM. However,
to date, the intricate relationship between physicochemical
characteristics of EfOM, calcium ion concentration and
membrane performance on EDC rejection remains poorly
understood. In order to gain insight into the removal mechanism
of EDCs in treated effluent, more research effort is
needed to better understand the above relationship.
In this study, we investigated the removal of steroid hormone
estrone from secondary effluent using a bench-scale cross-flow
NF/RO system. We examined the effect of various organic fractions
isolated from secondary effluent and membrane characteristics
on the removal of estrone by this process. After
determining the key organic fraction which makes a crucial
contribution to the enhanced estrone rejection by EfOM, the
influence of calcium ion concentration on estrone rejection with
the presence of the key fraction was investigated.
2. Materials and methods
2.1. NF/RO membranes and characterization
Four NF/RO membranes, denoted as DL, CK, AK and CG (Sepa
membranes, OSMONICS, USA), were used. DL and AK are polyamide
(PA) composite membranes. CK and CG are made of
cellulose acetate (CA) polymer. New membranes were soaked in
ultrapure water for 24 h with the water being replaced regularly
prior to use. Membrane properties were characterized as
described elsewhere (Jin et al., 2007) and are shown in Table 1.
Briefly, contact angle was determined with ultrapure water at
pH 5.8 0.2 using a goniometer (Ramé-hart contact angle
goniometer, Imaging System, Mountain Lakes, NJ, USA). Zeta
potential of the membrane surface was measured in the electrolyte
background solution (1 mM NaHCO3 and 8 mM NaCl, pH
7) using an Electro Kinetic Analyzer (EKA) from Anton Paar
(Austria). To characterize the membrane pore size, MWCO was
estimated by filtering the electrolyte background solution containing
polyethylene glycol (PEG) with MW of 200, 400, 600 and
1000 Da, purchased from Fluka (Switzerland). Filtration tests
were performed using the same cross-flow membrane test unit
as used in the estrone filtration tests with clean membrane.
A PEG concentration of approximately 6 ppm as dissolved
organic carbon (DOC) was used for each measurement.
2.2. Membrane filtration unit
A laboratory-scale cross-flow membrane filtration unit was
used (Fig. S1). Membrane coupon (13 cm by 13 cm) was housed
in a stainless steel membrane cell. Feed solution was fed from

640
Table 1 – Membrane properties.
Membrane Membrane type Contact angle ( ) Zeta potential a
a reservoir by a pump, capable of providing a maximum
pressure of 4137 kPa and a maximum flow of 20 l/min. The
temperature inside the feed reservoir was maintained
constant (22 0.1 C) by a recirculating chiller. The pressure
(1379 kPa with DL, CK and AK, and 2758 kPa with CG) inside
the membrane cell was set with a back-pressure regulator and
a bypass valve.
Each experiment was conducted in two steps: (1) the
membrane was compacted with ultrapure water until the
water flux remained constant; (2) the reservoir was emptied
and filled with test solution. The solution was then filtered at
constant transmembrane pressure. All experiments were
conducted for 24 h. During filtration, both permeate and
concentrate were recirculated back to the feed tank. At specified
intervals, permeate and feed samples were collected for
analysis. Rejection (R) is defined as R ¼ 100 ð1 Cp=C fÞ,
where C p and C f are the permeate and feed concentrations,
respectively. The effect of EfOM on estrone rejection was
determined based on the rejection measured at the end of
filtration where stable rejection was achieved.
2.3. EfOM fractionation method
Secondary effluent was collected from a local municipal
wastewater treatment plant. The water was first filtered through
amicrofiltration(MF)unit(0.05mm, Desal/Osmonics, USA) to
simulate pretreatment ahead of NF/RO processes in the real
plant, which is commonly used to remove suspended solids and
colloids. The total organic carbon (TOC) and total dissolved solid
(TDS) of the water samples ranged from 9.2 to 10.8 ppm and 455
to 618 ppm, respectively. Other characteristics of the MF treated
secondary effluent are summarized in Table S1.
The filtrates were subsequently fractionated into six
subcomponents: HpoA, hydrophobic bases (HpoB), hydrophobic
neutrals (HpoN), hydrophilic acids (HpiA), hydrophilic
bases (HpiB) and hydrophilic neutrals (HpiN) using a method
developed by Leenheer (1981). SupeliteÔ XAD-8 resin
(SUPELCO, USA.), AG-MP-50 cation exchange resin (BioRad,
USA) and Amberlite IRA-96 anion exchange resin (Rohm and
Haas, France) were the resins used. Details on the instruments
and fractionation methods have been published previously
(Hu et al., 2003). The DOC fractionation results for the
secondary effluent are presented in Table 2. HpoA predominated
in secondary effluent and accounted for 36.6–40.3% of
the total amount of dissolved organic matter. These results do
not agree with an earlier study by Imai et al. (2002) who found
different results on the fractions with HpiA as the most
water research 44 (2010) 638–648
MWCO (Da) b
abundant fraction for the samples taken from Japan. Thus,
DOM composition in secondary effluent varies with the origin
of wastewater and the type of treatment process used. In this
study, the experimental results obtained in MF treated
secondary effluent should be analyzed based on its specific
DOM-fraction distribution as reported here.
2.4. Chemical and solution chemistry
Typical flux/pressure
(m/s per Pa) c
DL Polyamide NF 30.7 2.7 16.3 2.7 490 1:46 10 5 =6:89 10 5
CK Cellulose acetate NF 54.2 1.4 6.0 0.3 560 1:32 10 5 =1:52 10 6
AK Polyamide RO 41.7 1.3 19.9 0.2

In all feed waters, initial estrone concentration was maintained
at 100 ng/L. Solution with different calcium ion
concentrations (0, 0.3 and 0.6 mM) were accomplished by
adding CaCl2 stock solution, while the total ionic strength was
kept constant by adjusting NaCl concentration.
2.5. Analytical methods
2.5.1. Estrone detection
Estrone was extracted from aqueous sample by OASIS HLB
cartridge (500 mg) from Waters, Singapore. Before the analytes
were extracted, 80 ng/l of E1-d4 was added to the sample
water to increase accuracy of the analytical data. The
cartridge was subsequently conditioned by 6 ml of diethylether,
5 ml of methanol and 10 ml of ultrapure water. The
sample water (400 ml) was loaded through the cartridge at
flow rate of 3 ml/min with the aid of a vacuum pump. After
sample loading, the cartridge was eluted with 10 ml of
a methanol–diethylether solution (10:90, v/v). The eluant was
collected in a brown glass vial and was allowed to dry under
a gentle stream of nitrogen. After drying, the residuals were
re-dissolved in 0.4 ml of methanol and then analyzed by liquid
chromatography–tandem mass spectrometry (MDS Sciex API
2000 tandem triple-quadropole mass spectrometer) from
Applied Biosystems, USA. Analytes were chromatographed on
a 2.1 150 mm column filled with 3.5 mm C18 reversed phase
packing (Agilent, Singapore). The multiple reaction monitoring
(MRM) mode was chosen for quantification. After
observing collision-induced dissociation (CID) spectra
obtained by full-scan production experiments, the following
MRM pairs were chosen: E1: 269.2/145.0; 269.2/143.0; E1-d4:
273.0/146.9. Estrone recovery was not matrix-dependent due
to using internal standard. The mean recovery and relative
standard deviation (RSD) was 99 7.4% (n ¼ 16). The limit of
detection was 0.5 ng/l, calculated using a signal-to-noise ratio
of 3. Details with respect to basic information on this method
are provided separately (Jin, 2007).
2.5.2. TOC and UV 254 analysis
In all cases, the concentrations of EfOM were detected by
a TOC analyzer (Shimadzu TOC-Vcsh, Japan). TOC and DOC
are used interchangeably in this study because secondary
effluent was filtered through 0.05 mm membrane before using.
The specific ultraviolet absorption of selected organic fraction
at 254 nm (SUVA254) was calculated by dividing the absorbance
at 254 nm measured using the Shimadzu UV-1700
UV-Visible Spectrophotometer (Japan) by the DOC level of the
solution.
2.5.3. Molecular Weight Measurement
High performance size exclusion chromatography (HPSEC)
was performed to determine both weight- and number-averaged
molecular weights. Instrumentation consisted of a SHI-
MADZU (Japan) HPLC instrument LC-10AT VP with a Polymer
Laboratories aquagel-OH column and a SHIMADZU SPD-M10A
diode array detector operating at 260 cm. The HPSEC system
was calibrated using polystyrene sulfonates standards of
known average MW between 210 and 13000, and acetone.
Mobile phase was composed of ultrapure water buffered with
phosphate to pH of 6.8.
water research 44 (2010) 638–648 641
3. Results and discussion
3.1. Rejection of estrone by NF/RO membranes in
electrolyte background solution
Filtration of estrone in electrolyte background solution (1 mM
NaHCO 3 and 8 mM NaCl, pH 7) provides the baseline from
which the effect of other organic matter on estrone rejection
can be found. The open symbols in Fig. 1 present the observed
rejections of estrone in electrolyte background solution by
four kinds of NF/RO membranes. It appeared that estrone
rejection was initially higher than 90% and decreased
dramatically and then stabilized at later filtration stage.
The excellent removal performances of all membranes at
the initial filtration stage can be attributed to their adsorption
capabilities and steric hindrance. However, the adsorption
effect can only contribute to the short-term removal of
estrone. As the feed solution is continuously filtered through
the membrane, more and more available sites on the
membrane are occupied by adsorbed estrone. When the
partition of estrone between feed solution and membrane
reaches equilibrium, there is no further net adsorption effect
taking place and thus the contribution from adsorption would
be negligible. Under this condition, size exclusion would
become the overriding removal mechanism at the later
filtration stage. Consistent with the MWCO values of the
membranes, the ultimate rejection of estrone followed the
order: AK > CG > DL > CK. AK membrane with MWCO value
below the MW of estrone can remove estrone efficiently.
However, for the other three kinds of membranes with MWCO
values larger than the MW of estrone, ultimate rejection was
extremely low. To avoid overestimation of estrone rejection,
in this paper, all of the determined rejections were based on
23 h of filtration when stable rejection was achieved.
3.2. Rejection of estrone by NF/RO membranes in
secondary effluent matrix
Estrone rejection in MF treated secondary effluent, compared
with that in electrolyte background solution, using four kinds
of NF/RO membranes was monitored and the results are presented
in Fig. 1, whereas the corresponding DOC rejection in
secondary effluent is presented in Fig. 2. Estrone removal
efficiencies in MF treated secondary effluent matrix for all
NF/RO membranes used were consistently higher than the
results obtained in electrolyte background solution. The
improvements in ultimate estrone rejection were 6.5%, 32.5%,
21.6% and 11.6% for AK, DL, CK and CG membranes, respectively.
The results obtained here are consistent with an earlier
study by Comerton et al. (2008), who investigated the influence
of water matrix on EDC rejection by RO and NF membranes.
They reported a significant increase in EDCs rejection by tight
NF membranes in membrane bioreactor effluent and natural
waters when compared to the ultrapure water.
The improvement in estrone rejection could be due to the
following reasons. Firstly, permeate flux decline was observed in
MF treated secondary effluent matrix during separation
processes (in electrolyte background solution, reduction in
permeate flux was not observed for any membrane tested,

642
a b
Estrone Rejection (%)
Estrone Rejection (%)
100
80
60
40
20
0
100
80
60
40
20
0
Electrolyte background
Secondary effluent
0 5 10 15 20 25
Filtration Time (h)
which can be attributed to the very low concentration of estrone
in the feed solution). After 23 h of filtration, the flux decreased by
about 13.0%, 8.9%, 33.8% and 17.5% for DL, CK, AK and CG
membrane, respectively. This suggests that the membranes
were fouled by EfOM. The entrapment of EfOM onto the
membrane active layer restricted the passage of contaminants,
resulting in an increase in the observed rejection of both DOC
(Fig. 2) and estrone at the ng/l level (Fig. 1). Similar increases in
trace organic rejection as a result of organic fouling have been
reported in previous studies (Agenson and Urase, 2007; Bellona
and Drewes, 2007; Nghiem and Hawkes, 2007). However, simple
EfOM fouling may not be the main reason, because the sequence
Estrone Rejection (%)
c d
Electrolyte background
Secondary effluent
0 5 10 15 20 25
Filtration Time (h)
Estrone Rejection (%)
100
80
60
40
20
0
100
80
60
40
20
Electrolyte background
Secondary effluent
0 5 10 15 20 25
Filtration Time (h)
Electrolyte background
Secondary effluent
0
0 5 10 15 20 25
Filtration Time (h)
Fig. 1 – Estrone rejection in electrolyte background solution and MF treated secondary effluent by (a) DL, (b) CK, (c) AK and (d) CG.
DOC Rejection (%)
100
95
90
85
80
75
70
65
60
DL
CK
AK
CG
0 5 10 15 20 25
Filtration Time (h)
Fig. 2 – DOC rejection in MF filtered secondary effluent by
four kinds of NF/RO membranes.
water research 44 (2010) 638–648
of the improvement in estrone rejection (DL > CK > CG > AK)
was mismatched with the sequence based on permeate flux
decline in secondary effluent matrix (AK > CG > DL > CK).
Therefore the enhanced estrone rejection can be mainly attributed
to the other reason, namely that, secondly, estrone molecules
may bind to some fractions of EfOM in bulk solution and be
retained together, resulting in a reduction in the passage of
estrone molecules through membrane. The remainder of this
paper is focused on proving this hypothesis.
3.3. Rejection of estrone in the presence of specific
organic fractions at same TOC level
In order to optimize membrane performance, it is essential to
identify which organic fraction in secondary effluent would
make a crucial contribution to the ‘‘enhancement effect’’ on
estrone removal. Amongst the six fractions derived from
EfOM, HpiB was present in insignificant quantities and thus
was expected to have a negligible influence on estrone
removal. HpiN, as the last fraction derived from the resin
fractionation system, inevitably contains some impurities of
other fractions (Hu et al., 2003). Moreover, as shown in our
previous work (Jin et al., 2007), dextran representing hydrophilic
neutral organics had little effect on estrone rejection.
Therefore, in this study, the research was focused on the
influences of HpoA, HpoB, HpoN and HpiA on estrone rejection
by NF/RO membranes.
As shown earlier, DL and CK membranes demonstrated
a greater improvement in estrone rejection in MF treated
secondary effluent compared to AK and CG membranes.
Investigations were firstly conducted to compare the capability
of each isolated fraction to improve estrone removal

a
Estrone Rejection (%)
b
Estrone Rejection (%)
100
80
60
40
20
0
100
80
60
40
20
0
Electrolyte background
With HpoA
With HpoB
With HpoN
With HpiA
0 5 10 15 20 25
Filtration Time (h)
Electrolyte background
With HpoA
With HpoB
With HpoN
With HpiA
0 5 10 15 20 25
Filtration Time (h)
Fig. 3 – Observed rejection of estrone with the presence of
different organic fractions at same DOC level (1 ppm) by (a)
DL and (b) CK.
using DL and CK membranes. To avoid the influence of DOC
values, the initial concentration of each organic fraction was
adjusted to a DOC value of approximately 1 ppm, which was
close to the original concentrations of HpoB, HpoN and HpiA
present in secondary effluent (Table 2).
Estrone rejection in the presence of each fraction over
a 24 h filtration period is presented in Fig. 3. It is evident that
estrone rejection was significantly improved by the presence
of HpoA, rising from 15.3% to 41.9% with DL membrane and
from 8.2% to 15.3% with CK membrane. The presence of HpoB
also resulted in an improvement in estrone rejection of as
high as 3.6% and 4.1% with DL and CK, respectively. In
contrast, HpoN and HpiA did not exhibit any obvious increase
in estrone rejection for any of the membranes tested,
although HpoN and HpiA were removed more efficiently than
HpoB (Fig. 4). The differences in the ‘‘enhancement effect’’
between HpoA and HpiA indicate that hydrophobic organic
fractions have a greater contribution to the enhanced estrone
rejection in secondary effluent than hydrophilic organic
fractions.
In all experiments, no obvious permeate flux decline (

644
a
DOC Rejection (%)
b
DOC Rejection (%)
100
90
80
70
60
50
40
30
20
10
0
100
90
80
70
60
50
40
30
20
10
0
0 5 10 15 20 25
Filtration Time (h)
HpoA
HpoB
HpoN
HpiA
HpoA
HpoB
HpoN
HpiA
0 5 10 15 20 25
Filtration Time (h)
Fig. 4 – Observed rejection of bulk organics (as DOC) with
the presence of different organic fractions at same DOC
level (1 ppm) by (a) DL and (b) CK.
molecules can bind to bulk DOM in feed matrix and subsequently
be retained together. In this case, the ‘‘enhancement
effect’’ on estrone removal was controlled not only by the
binding between estrone and DOM in feed solution, but also by
the rejection efficiency of the DOM with which estrone
molecules are associated. Therefore, the weak effect of HpoB
on estrone rejection is due to the inefficient removal of HpoB
by CK membrane (Fig. 7b). It is apparent that the HpoA fraction,
for all membranes used, had a significantly greater
‘‘enhancement effect’’ than the other two hydrophobic fractions
on ultimate estrone removal. After 23 h of filtration, the
presence of HpoA increased estrone rejection by 29.8%, 13.3%
and 20.5% for DL, CK and CG membranes, respectively. This
could be attributed to a combination of the following reasons.
Firstly, the structural characteristics of HpoA (phenolic group)
water research 44 (2010) 638–648
Fig. 5 – Schematic of possible hydrogen bonding between
phenolic groups of estrone molecule and HpoA and
between phenolic group in estrone molecule and aromatic
amine group in HpoB (A–E stand for any possible functional
groups within HpoB fraction).
allowed estrone molecules to bind with HpoA in feed solution
through hydrogen bonding, thus estrone could be removed
together with HpoA. Secondly, HpoA with the highest
concentration (Table 2) in feed solution would have the
highest chance to associate with estrone molecules. Thirdly,
HpoA with the largest MW of more than 4000 Da and a negative
charge could be removed effectively (Fig. 7). The formation
of estrone-HpoA complex in bulk solution led to
a significant increase in MW (from 270 Da to more than
4000 Da) and the appearance of negative charge and thus an
increased level of estrone rejection by negative charged
membranes based on the mechanisms of size exclusion and
charge repulsion. Therefore, it could be concluded that the
HpoA fraction was the greatest contributor to the ‘‘enhancement
effect’’ on estrone rejection in treated effluent.
It is worthwhile to note that the ‘‘enhancement effect’’ of
HpoA followed the order of DL > CG > CK. This sequence
agrees with the sequence based on the negative charge of the
membrane surface (Table 1). According to the above discussion,
the binding of estrone with HpoA in feed solution was
critically important for the ‘‘enhancement effect’’. Less negative
charge of membrane surface would result in more HpoA
adsorption onto membrane due to the weaker electrostatic
repulsion between HpoA and membrane. Consequently, less
HpoA molecules would be available in feed solution for binding
with estrone, resulting in a weaker improvement in estrone
rejection. The results suggest that a more negatively-charged
Table 3 – Characteristics of specific organic fractions derived from secondary effluent.
Parameter HpoA HpoB HpoN HpiA
MW (Da) 4192–4442 762–1751 1406–2658 3628–3825
r 1.29–1.43 1.01–1.84 1.01–1.27 1.18–1.32
SUVA254 (m 1 L/mg C) 2.58–4.25 2.06–3.81 1.19–1.57 0.75–0.81

a
Estrone Rejection (%)
b
Estrone Rejection (%)
c
Estrone Rejection (%)
100
80
60
40
20
0
100
80
60
40
20
0
100
80
60
40
20
0
Electrolyte background
With HpoA
With HpoB
With HpoN
0 5 10 15 20 25
Filtration Time (h)
Electrolyte background
With HpoA
With HpoB
With HpoN
0 5 10 15 20 25
Filtration Time (h)
Electrolyte background
With HpoA
With HpoB
With HpoN
0 5 10 15 20 25
Filtration Time (h)
Fig. 6 – Observed rejection of estrone with the presence of
different hydrophobic organic fractions (at their original
concentration in secondary effluent) by (a) DL, (b) CK and (c) CG.
membrane would benefit the interaction between HpoA and
estrone and thus estrone rejection.
3.5. Effect of calcium ion concentration on estrone
rejection in HpoA-containing solution
Based on the discussion above, the estrone-binding capacity
of HpoA is the major contributor to its significant ‘‘enhancement
effect’’ on estrone rejection by NF/RO membranes. It is
known that divalent cations, such as calcium, can influence
the binding of trace contaminants by humic substances
water research 44 (2010) 638–648 645
a
DOC Rejection (%)
b
DOC Rejection (%)
c
DOC Rejection (%)
100
90
80
70
60
50
40
30
20
10
0
100
90
80
70
60
50
40
30
20
10
0
100
90
80
70
60
50
40
30
20
10
0
HpoA
HpoB
HpoN
0 5 10 15 20 25
Filtration Time (h)
HpoA
HpoB
HpoN
0 5 10 15 20 25
Filtration Time (h)
HpoA
HpoB
HpoN
0 5 10 15 20 25
Filtration Time (h)
Fig. 7 – Observed rejection of bulk organics (as DOC) with the
presence of different hydrophobic organic fractions (at their
original concentration in SE) by (a) DL, (b) CK and (c) CG.
(Schlautman and Morgan, 1993). Therefore, calcium ion
concentration in feed solution is predicted to affect the
estrone rejection in HpoA-containing solution.
Fig. 8 presents the influence of HpoA on estrone rejection
by DL membrane at different calcium concentrations. It is
evident that while the presence of HpoA could increase the
rejection of estrone, a higher calcium ion concentration tended
to reverse this effect. In particular, the addition of 0.3 mM
calcium ion in feed solution resulted in the ‘‘enhancement

646
a
Estrone Rejection (%)
b
Estrone Rejection (%)
c
Estrone Rejection (%)
100
80
60
40
20
0
100
80
60
40
20
0
100
80
60
40
w/o HpoA
with HpoA
0 5 10 15 20 25
Filtration Time (h)
w/o HpoA
with HpoA
0 5 10 15 20 25
Filtration Time (h)
w/o HpoA
with HpoA
0 5 10 15 20 25
Filtration Time (h)
Fig. 8 – Influence of HpoA on estrone rejection by DL
membrane at different calcium ion concentration (a) 0 mM
Ca 2D , (b) 0.3 mM Ca 2D and (c) 0.6 mM Ca 2D .
effect’’ of HpoA on ultimate estrone rejection by DL
membrane decreasing to 18.0% as compared to the improvement
of 29.8% with the absence of calcium ion. When calcium
ion concentration was increased to 0.6 mM, HpoA did not
exhibit observable improvement in estrone rejection.
In addition, negligible differences in permeate flux decline
and DOC rejection were observed with increasing calcium
concentration (data are not shown, but are available on
request). This observation is unexpected and also differs
water research 44 (2010) 638–648
from other published investigations (Hong and Elimelech,
1997; Tang et al., 2007), in which a more serious permeate
flux decline with an increased calcium concentration was
reported. This phenomenon was possibly due to the relatively
low calcium ion concentration added in this study.
Such low calcium ion concentration might not be enough to
overcome the electrostatic repulsive energy barrier between
HpoA and membrane and thus cannot result in a severe
permeate flux decline. In this case, the diminished effect of
HpoA on estrone rejection by membrane with increasing
calcium concentration can most probably be attributed to the
lower estrone–HpoA complex formation. Calcium ions can
interact specifically with acidic functional groups (predominantly
carboxylic groups) of humic substances (Hong and
Elimelech, 1997). Such interactions reduce the charge repulsion
between negative functional groups on HpoA molecules
due to charge shielding and neutralization. As a result, the
structure of HpoA tends to change from a stretched and
linear configuration to a coiled and compact configuration
with increased calcium concentration (Clark and Lucas,
1998). This would result in the interactive sites within HpoA
molecules becoming less easily reachable by estrone molecules
by restricting access to interior sorption sites. In this
case, the increase in calcium concentration would decrease
the ability of HpoA to bind estrone molecules in feed water.
This observation is similar to the results of Schlautman and
Morgan (1993) who reported that the presence of calcium at
neutral pH decreased the ability of humic acid to bind perylene.
They attributed the finding to the fact that the
compression of the HA structure decreased the size of voids
in humic acid into which target solute could partition. Since
the presence of calcium ions was not conducive to the
binding of estrone by HpoA, the effect of increasing calcium
concentration, even at relatively low levels, was to push
estrone rejection towards the ‘‘free’’ estrone limit of rejection
obtained in the absence of HpoA.
4. Conclusions
The results reported in this paper demonstrate that NF and
loose RO membranes cannot remove trace steroid hormone
efficiently in single-organic solution. However, the removal
efficiency can be enhanced significantly in the presence of
EfOM in feed solution. The HpoA fraction played a paramount
role in the ‘‘enhancement effect’’. This enhancement is
mainly attributed to the binding of estrone molecules to HpoA
in feed solutions and being removed with the negatively
charged HpoA macromolecules based on the mechanisms of
size exclusion and charge repulsion. However, the presence
of calcium ion tended to diminish this effect. Application of
membrane with more negative-charge at its interface can
greatly intensify this ‘‘enhancement effect’’.
Supplementary data
Supplementary data associated with this article can be found
in online version at doi:10.1016/j.watres.2009.09.057.

eferences
Adams, N.R., 1998. Clover phyto-oestrogens in sheep in Western
Australia. Pure and Applied Chemistry 70 (9), 1855–1862.
Agbekodo, K.M., Legube, B., Dard, S., 1996. Atrazine and simazine
removal mechanisms by nanofiltration: influence of natural
organic matter concentration. Water Research 30 (11), 2535–2542.
Agenson, K.O., Urase, T., 2007. Change in membrane performance
due to organic fouling in nanofiltration (NF)/reverse osmosis
(RO) applications. Separation and Purification Technology 55
(2), 147–156.
Baker, V.A., 2001. Endocrine disrupters – testing strategies to
assess human hazard. Toxicology in Vitro 15 (4-5), 413–419.
Baronti, C., Curini, R., D’Ascenzo, G., Di Corcia, A., Gentili, A.,
Samperi, R., 2000. Monitoring natural and synthetic estrogens
at activated sludge sewage treatment plants and in a receiving
river water. Environmental Science & Technology 34 (24),
5059–5066.
Bellona, C., Drewes, J.E., 2007. Viability of a low-pressure nanofilter
in treating recycled water for water reuse applications: a pilotscale
study. Water Research 41 (17), 3948–3958.
Bellona, C., Drewes, J.E., Xu, P., Amy, G., 2004. Factors affecting
the rejection of organic solutes during NF/RO treatment –
a literature review. Water Research 38 (12), 2795–2809.
Clark, M.M., Lucas, P., 1998. Diffusion and partition of humic acid
in a porous ultrafiltration membrane. Journal of Membrane
Science 143 (1–2), 13–25.
Comerton, M.A., Andrews, C.R., Bagley, M.D., Hao, C., 2008. The
rejection of endocrine disrupting and pharmaceutically active
compounds by NF and RO membranes as a function of
compounds and water matrix properties. Journal of
Membrane Science 313 (1-2), 323–335.
Croley, T.R., Hughes, R.J., Koenig, B.G., Metcalfe, C.D., March, R.E.,
2000. Mass spectrometry applied to the analysis of estrogens
in the environment. Rapid Communications in Mass
Spectrometry 14 (13), 1087–1093.
Desbrow, C., Routledge, E.J., Brighty, G.C., Sumpter, J.P., Waldock, M.,
1998. Identification of estrogenic chemicals in STW effluent. 1.
Chemical fractionation and in vitro biological screening.
Environmental Science & Technology 32 (11), 1549–1558.
Devitt, E.C., Ducellier, F., Cote, P., Wiesner, M.R., 1998. Effects of
natural organic matter and the raw water matrix on the
rejection of atrazine by pressure-driven membranes. Water
Research 32 (9), 2563–2568.
Hong, S., Elimelech, M., 1997. Chemical and physical aspects of
natural organic matter (NOM) fouling of nanofiltration
membranes. Journal of Membrane Science 132 (2), 159–181.
Hu, J.Y., Ong, S.L., Shan, J.H., Kang, J.B., Ng, W.J., 2003. Treatability of
organic fractions derived from secondary effluent by reverse
osmosis membrane. Water Research 37 (19), 4801–4809.
Hu, J.Y., Jin, X., Ong, S.L., 2007. Rejection of estrone by
nanofiltration: influence of solution chemistry. Journal of
Membrane Science 302 (1-2), 188–196.
Iguchi, T., Watanabe, H., Katsu, Y., 2001. Developmental effects of
estrogenic agents on mice, fish, and frogs: a mini-review.
Hormones and Behavior 40 (2), 248–251.
Imai, A., Fukushima, T., Matsushige, K., Kim, Y.H., Choi, K.,
2002. Characterization of dissolved organic matter in
effluents from wastewater treatment plants. Water Research
36 (4), 859–870.
Jin, X., 2007. Rejection of steroid hormone estrone by NF/RO
membranes. PhD thesis. Department of Civil Engineering,
National University of Singapore, Singapore.
Jin, X., Hu, J.Y., Ong, S.L., 2007. Influence of dissolved organic
matter on estrone removal by NF membranes and the role of
their structures. Water Research 41 (14), 3077–3088.
water research 44 (2010) 638–648 647
Johnson, A.C., Sumpter, J.P., 2001. Removal of endocrinedisrupting
chemicals in activated sludge treatment works.
Environmental Science & Technology 35 (24), 4697–4703.
Kimura, K., Amy, G., Drewes, J.E., Heberer, T., Kim, T.U.,
Watanabe, Y., 2003. Rejection of organic micropollutants
(disinfection by-products, endocrine disrupting compounds,
and pharmaceutically active compounds) by NF/RO
membranes. Journal of Membrane Science 227 (1–2), 113–121.
Kimura, K., Toshima, S., Amy, G., Watanabe, Y., 2004. Rejection of
neutral endocrine disrupting compounds (EDCs) and
pharmaceutical active compounds (PhACs) by RO membranes.
Journal of Membrane Science 245 (1–2), 71–78.
Leenheer, J.A., 1981. Comprehensive approach to preparative
isolation and fractionation of dissolved organic carbon from
natural waters and wastewaters. Environmental Science &
Technology 15 (5), 578–587.
Majewska-Nowak, K., Kabsch-Korbutowicz, M., Dod, M.,
Winnicki, T., 2002. The influence of organic carbon
concentration on atrazine removal by UF membranes.
Desalination 147 (1–3), 117–122.
Ng, H.Y., Elimelech, M., 2004. Influence of colloidal fouling on
rejection of trace organic contaminants by reverse osmosis.
Journal of Membrane Science 244 (1–2), 215–226.
Nghiem, L.D., Hawkes, S., 2007. Effects of membrane fouling
on the nanofiltration of pharmaceutically active
compounds (PhACs): Mechanisms and role of membrane
pore size. Separation and Purification Technology 57 (1),
176–184.
Nghiem, L.D., Manis, A., Soldenhoff, K., Schäfer, A.I., 2004a.
Estrogenic hormone removal from wastewater using NF/RO
membranes. Journal of Membrane Science 242 (1–2),
37–45.
Nghiem, L.D., Schäfer, A.I., Elimelech, M., 2004b. Removal of
natural hormones by nanofiltration membranes:
measurement, modeling, and mechanisms. Environmental
Science & Technology 38 (6), 1888–1896.
Schäfer, A.I., Fane, A.G., Waite, T.D., 2005. Nanofiltration:
Principles & Applications. Elsevier Advanced Technology,
Oxford, New York.
Schäfer, A.I., Nghiem, L.D., Waite, T.D., 2003. Removal of the
natural hormone estrone from aqueous solutions using
nanofiltration and reverse osmosis. Environmental Science &
Technology 37 (1), 182–188.
Schlautman, M.A., Morgan, J.J., 1993. Effects of aqueous chemistry
on the binding of polycyclic aromatic hydrocarbons by
dissolved humic materials. Environmental Science &
Technology 27 (5), 961–969.
Soliman, M.A., Pedersen, J.A., Suffet, I.H., 2004. Rapid gas
chromatography–mass spectrometry screening method for
human pharmaceuticals, hormones, antioxidants and
plasticizers in water. Journal of Chromatography A 1029 (1-2),
223–237.
Steinle-Darling, E., Zedda, M., Plumlee, H.M., Ridgway, F.H.,
Reinhard, M., 2007. Evaluating the impacts of membrane
type, coating, fouling, chemical properties and water
chemistry on reverse osmosis rejection of seven
nitrosoalklyamines, including NDMA. Water Research 41 (17),
3959–3967.
Tang, Y.C., Kwon, Y.N., Leckie, O.J., 2007. Fouling of reverse
osmosis and nanofiltration membranes by humic acid –
effects of solution composition and hydrodynamic
conditions. Journal of Membrane Science 290 (1–2),
86–94.
Ternes, T.A., Stumpf, M., Mueller, J., Haberer, K., Wilken, R.D.,
Servos, M., 1999. Behavior and occurrence of estrogens in
municipal sewage treatment plants - I. Investigations in
Germany, Canada and Brazil. The Science of the Total
Environment 225 (1–2), 81–90.

648
Thurman, E.M., Malcolm, R.L., 1981. Preparative isolation of
aquatic humic substances. Environmental Science &
Technology 15 (4), 463–466.
Tim, Z., 1997. In vitro bioassays for assessing estrogenic substances.
Environmental Science & Technology 31 (3), 613–623.
Williams, R.J., Johnson, A.C., Smith, J.J., Kanda, R., 2003. Steroid
estrogens profiles along river stretches arising from sewage
treatment works discharges. Environmental Science &
Technology 37 (9), 1744–1750.
Xu, P., Drewes, J.E., Kim, T.U., Bellona, C., Amy, G., 2006. Effect of
membrane fouling on transport of organic contaminants in
water research 44 (2010) 638–648
NF/RO membrane applications. Journal of Membrane Science
279 (1–2), 165–175.
Yamamoto, H., Liljestrand, H.M., Shimizu, Y., Morita, M., 2003.
Effects of physical-chemical characteristics on the sorption of
selected endocrine disruptors by dissolved organic matter
surrogates. Environmental Science and Technology 37 (12),
2646–2657.
Yoon, Y., Westerhoff, P., Yoon, J., Snyder, S.A., 2004. Removal of
17 beta estradiol and fluoranthene by nanofiltration and
ultrafiltration. Journal of Environmental Engineering 130 (12),
1460–1467.

Screening of antimycotics in Swedish sewage treatment
plants – Waters and sludge
Richard H. Lindberg*, Jerker Fick, Mats Tysklind
Department of Chemistry, Umea˚ University, SE-901 87 Umea˚, Sweden
article info
Article history:
Received 25 March 2009
Received in revised form
19 October 2009
Accepted 23 October 2009
Available online 31 October 2009
Keywords:
Antimycotics
Sewage water
Sludge
Mass flow
1. Introduction
abstract
There are approximately 40 antimycotics, i.e. pharmaceuticals
that can be used to treat fungal infections in humans,
many of which are imidazoles, triazoles or allylamines. Antimycotics
are usually administered topically, although a few,
such as ketoconazole and fluconazole, are also used perorally.
All antimycotics currently used are general cytochrome P450
inhibitors that reportedly affect CYP families 1–3. For instance,
the imidazoles and triazoles strongly inhibit CYP3A4 and 2C9,
while the allylamines are known to affect CYP2D6 (Sweetman,
2007). This has potentially significant implications for their
use as pharmaceuticals since the P450s are responsible for the
initial metabolism of ca. 75% of xenobiotics in humans, hence
water research 44 (2010) 649–657
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
* Corresponding author. Tel.: þ46 90 786 5464; fax: þ46 90 12 7655.
E-mail address: richard.lindberg@chem.umu.se (R.H. Lindberg).
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.watres.2009.10.034
Concentrations of six pharmaceutical antimycotics were determined in the sewage water,
final effluent and sludge of five Swedish sewage treatment plants (STPs) by solid phase
extraction, liquid/solid extraction, and liquid chromatography–electrospray tandem mass
spectrometry. The antimycotics were quantified by internal standard calibration. The
results were used to estimate national flows that were compared to predictions based on
sales figures. Fluconazole was the only one of the six investigated antimycotics that was
detected (at concentrations ranging from 90 to 140 ng L 1 ) in both raw sewage water and
final effluent. Negligible amounts of this substance were removed from the aqueous phase,
and its levels were below the limit of quantification in all of the analyzed sludge samples.
In contrast, clotrimazole, ketoconazole and econazole were present in all of the sludge
samples, at concentrations ranging between 200 and 1000 mgkg 1 , dry weight. There were
close correlations between the national measured and predicted antimycotic mass flows.
Antimycotic fate analysis, based on sales figures, indicated that 53% of the total amount of
fluconazole sold appeared in the final effluents of the STPs, while 1, 155, 35, 209 and 41% of
the terbinafine, clotrimazole, ketoconazole, econazole and miconazole sold appeared in the
digested dewatered sludge.
ª 2009 Elsevier Ltd. All rights reserved.
they could have undesirable effects on patients’ ability to
eliminate other substances (Sweetman, 2007; Guengerich,
2008). More specifically, the imidazole and triazole antimycotics
inhibit the essential activity of CYP51 (also known as
ERG11 in fungi) in ergosterol synthesis in fungal pathogens,
thereby increasing the permeability of their membranes,
inhibiting their growth and/or killing them (Graybill and
Drutz, 1980; Yoshida et al., 2000; Lepesheva and Waterman,
2007). The antimycotic activity of allylamine antimycotics is
also based on interference with ergosterol biosynthesis, but by
inhibiting squalene epoxidase (Ryder, 1992).
Gunnarson et al. have recently shown that six species
commonly used in aquatic ecotoxicology tests possess CYP51
orthologs that have more than 70% homology to the drug

650
targets (Gunnarsson et al., 2008). Further, imidazole and triazole
antimycotics have been shown to inhibit CYP19, which
participates in the conversion of androgens to estrogens
(Trösken et al., 2004; Gyllenhammar et al., 2009), with serious
potential consequences for sex differentiation in exposed
vertebrates, including testicular development in female
salmon (Piferrer et al., 1994). Gyllenhammar et al. (2009) have
also shown that exposure to the imidazole clotrimazole can
have complex effects on aromatase activity in the gonads and
brain of Xenopus tropicalis larvae; decreasing brain aromatase
activity during their gonadal differentiation and increasing
gonadal aromatase activity during metamorphosis.
Concentrations of antimycotics have been analyzed
(generally by solid phase extraction, SPE, followed by liquid (or
gas) chromatography mass spectrometry, LC(GC)–MS/MS) in
various aquatic environments and sewage flows (Kolpin et al.,
2004; Thomas and Hilton, 2004; Roberts and Bersuder, 2006;
Roberts and Thomas, 2006; Peschka et al., 2007; Kahle et al.,
2008; Van De Steene and Lambert, 2008). Clotrimazole levels in
the Main and Rhine rivers in Germany, and both the raw
sewage water and final effluent from a German sewage
treatment plant (STP), have been reported to be in the low
ng L 1 range (Peschka et al., 2007). Comparable concentrations
of clotrimazole have also been reported in two UK studies,
which assessed sewage water from the Howton STP and
samples collected along the Tyne estuary (Thomas and Hilton,
2004; Roberts and Thomas, 2006). Somewhat higher levels of
clotrimazole and fluconazole, between 10 and 110 ng L 1 , have
been found in Swiss STP waters, along with fluconazole at
levels close to its limit of quantification (LOQ), 1 ng L 1 ,in
Swiss lakes (Kahle et al., 2008). In addition, quantifiable levels
of miconazole have been found in both surface and sewage
waters in the UK (Roberts and Bersuder, 2006), although it was
not found at levels exceeding its reporting limit (17.5 ng L 1 )in
an analysis of the contribution of wastewater contaminants to
US streams (Kolpin et al., 2004). However, no published studies
to our knowledge have analyzed the concentrations of antimycotics
in digested sewage sludge.
The six substances examined here were chosen to represent
a range, in terms of applications and physicochemical
properties, of antimycotics commonly used in Sweden.
Information on these six substances and the internal standard
sulconazole (which is not used in this country) is presented in
Table 1. Little is known about the fate and ecotoxic potential of
these substances, thus the aim of the presented study was to
increase information on their occurrence, levels and fate
within STPs. In addition, the collected data were cross-validated
against predictive equivalents. To our knowledge, this is
the first published study to present national screening data on
antimycotics in both sewage water and sludge.
2. Experimental
2.1. Chemicals
All of the antimycotics were bought from Sigma Aldrich
(Stockholm, Sweden) except for terbinafine, which was
obtained from LGC Standards (Bora˚s, Sweden). Formic acid
and methanol (HPLC-grade) were purchased from JT Baker
water research 44 (2010) 649–657
(Deventer, the Netherlands), acetonitrile (HPLC-grade) from
Fischer chemicals (Zurich, Switzerland), and sulfuric acid
from Merck (Darmstadt, Germany). Water was purified (to
18.2 MU cm resistivity) using a MAXIMA HPLC ultra pure
water system (ELGA, High Wycombe Bucks, England). Stock
and working (including calibration standards) solutions of
individual antimycotics were prepared in methanol and
diluted in water, respectively.
2.2. Sample sites, collection and pre-extraction
procedure
Raw sewage water, final effluent and sludge samples were
taken from five Swedish STPs during October and November
2007. The STPs surveyed were Stockholm (Henriksdal), Gothenburg
(Ryaverken), Umea˚ , Alingsa˚s (Nolhaga), and Bollebygd,
which were selected to represent a wide range of STPs in
terms of size and catchment area. Sewage water from the
Umea˚ county hospital (UCH) was also collected during the
same time period. In addition, raw sewage water, raw sewage
water particles, raw sludge and digested dewatered sludge
were collected from the Umea˚ STP in April 2008 to assess the
fate of antimycotics during the treatment process. The locations
of the STPs and data on their waste flows and catchment
areas are provided in Fig. 1. All of the STPs treat sewage water
mechanically, chemically and biologically. Nitrogen is also
actively removed at the Henriksdal, Ryaverken and Nolhaga
STPs, and the final solid product of all the STPs is anaerobically
digested sludge, except at Bollebygd, where only clarifiers
are used.
Flow-proportional samples of raw sewage water and final
effluents were collected from each of the five STPs in each
case over a full single day, while grab samples of the UCH
sewage water and digested dewatered sludge were collected.
After collection, the sewage water (2 L in brown glass bottles)
and sludge samples (500 g in brown glass beakers) were
immediately stored at 18 C prior to extraction. Grab
sampling was also used in the antimycotic fate study at the
Umea˚ STP, in which raw sewage water (1 þ 10 L), final effluent
(2 L), primary sludge (2 L) and 500 g of digested dewatered
sludge samples were collected between 8:30 and 9:00 on each
of two consecutive days in April 2008 (except for the digested
dewatered sludge, which was only collected on the first of the
sampling days due to technical problems). The raw sewage
water (1 L) and final effluent samples for SPE were filtered
(through 0.45 mm filters) immediately after collection, while
the remaining raw sewage water (10 L) and primary sludge
samples were centrifuged at 4800 rpm for 20 min, to concentrate
the particles sufficiently for analysis (Lindberg et al.,
2006), then air-dried for 24 h. All of the samples were then
stored at 18 C awaiting extraction.
2.3. Sample preparation – sewage waters
The aqueous samples were thawed, filtered through 0.45 mm
MFÔ membrane filters (Millipore, Sundbyberg, Sweden),
acidified to pH 3 with sulfuric acid and 500 ng of the internal
standard (IS) sulconazole was added to 500 mL portions of the
resulting samples, which were then subjected to SPE using
Evolute ABN (6 mL, 200 mg) solid sorbent columns (Biotage,

Table 1 – Information on the antimycotics investigated in this study.
Terbinafine
(allylamine)
Clotrimazole
(imidazole)
Ketoconazole
(imidazole)
Fluconazole
(triazole)
Econazole
(imidazole)
Miconazole
(imidazole)
IS sulconazole
(imidazole)
O
N
N
N
Cl
N N O
Cl
Cl
Cl
N
Cl
N
HO
N
N
N
N
N
H
O
N
O
N
Cl
F F
N
O
O
N
S
Cl
water research 44 (2010) 649–657 651
N
Cl
N
Cl
N
Cl
Cl
Cl
Cl
Application Use a
Topical 43
Topical 11
(kg year 1 )
EAS b Log K ow c
g
g
PEC d
(ng L 1 )
UD e BCF f
5.81 6 m 5370
6.26 2 m 13,183
Topical/peroral 399/22 –/0% 4.45 (4.35) 60 r 447
Peroral 121 80% u 0.25 20 r 3
Topical 67
Topical 134
Not used
in Sweden
g
g
5.61 10 r 4169
6.25 20 r 12,882
a Estimated from Apoteket AB’s Swedish sales during 2007.
b Excretion of active substance through urine and/or feces (LIF, 2008).
c Log Kow values estimated by KowWin, EPI suiteÔ (EPI SuiteÔ) (experimental values in parentheses).
d Predicted environmental concentrations in surface waters, calculated according to EMEA guidelines Phase 1, except that Swedish sales values
were used rather than the supposed penetration factor (Fpen), in combination with the daily defined dose (European Medicines Agency, 2006).
e Ultimate degradation (m ¼ months, r ¼ recalcitrant) estimated by BioWin, EPI suiteÔ (EPI SuiteÔ).
f Estimated bioconcentration factor as estimated by BcfWin, EPI suiteÔ (EPI SuiteÔ).
g No excretion due to topical application.
6.46

652
Uppsala, Sweden) that had been conditioned with 6 mL of
methanol and 6 mL water adjusted to pH 3 with sulfuric acid.
Each sample was applied at a flow rate of 5 mL min 1 , cleaned
using 5 mL of aqueous methanol (5%), and eluted by two
washes of 3 mL methanol with 0.5% formic acid. The eluates
were evaporated almost to dryness and finally reconstituted
with 0.7 mL aqueous acetonitrile (10%).
2.4. Sample preparation – sludge
The sludge samples were thawed, then 2.0 g portions from
each sample were placed in separate glass centrifuge tubes.
The IS was added, 500 ng, and the samples were subjected to
liquid/solid extraction step by adding 10.0 mL of methanol to
each tube and ultrasonically agitating them for 10 min. They
were subsequently centrifuged at 4800 rpm for 10 min, and
the supernatants were placed in separate 30 mL glass vials.
The residues were then re-extracted in the same manner and
the two supernatants were combined. The combined extracts
(approximately 19 mL) were made up to 200 mL with purified
water research 44 (2010) 649–657
Fig. 1 – Mass flows of the measured antimycotics (g day L1 ) in the raw sewage water, final effluent and digested dewatered
sludge of five Swedish sewage treatment plants.
water, acidified to pH 3, and finally subjected to SPE using the
protocol described above for the aqueous samples.
The dry weights of the samples were determined by
weighing portions after drying them at 105 C for 24 h.
2.5. Liquid chromatography–mass spectrometry
Sample extracts and calibration solutions (10 mL) were injected
using an AS 3000 autoinjector (Thermo Finnigan, San Jose, CA,
US) into a Hydrosphere C18 guard column (10 4 mm i.d.,
5 mm particle size; YMC Inc., Wilmington, NC, US) followed by
an analytical column of the same type (150 4.6 mm i.d., 5 mm
particle size). The antimycotics were then eluted by a linear
gradient of 90:10 to 70:30 A:B (v/v) over 17 min followed by
a further 4-min linear gradient to 30:70 (v/v), where A and B are
water and acetonitrile, respectively (both acidified with 0.1%
formic acid), at a 0.8 mL min 1 flow rate, generated using
a P4000 HPLC pump (Spectra system, Thermo Finnigan) at
a constant temperature of 25 C.
The eluting antimycotics were detected by an LCQ Duo ion
trap mass spectrometer (Thermo Finnigan) with an

electrospray ion source operating in positive ion mode. The
source voltage was maintained at a constant 6.0 kV and the
heated capillary temperature was set to 250 C. The MS/MS
parameters and the collision energies were optimized semiautomatically
and manually, respectively.
2.6. Extraction yield assessment
Absolute extraction yields (including matrix suppression of
precursor/product ions during instrumental analysis) were
investigated by fortifying 500 ng of each analyte and the IS to
raw sewage water (500 mL, n ¼ 4) and digested sludge samples
(2.0 g, n ¼ 3). The SPE of the water samples began immediately
but the sludge samples were left for an hour prior to the liquid/
solid extraction. The extraction yield was evaluated by
a comparison of the analyte and IS peak areas of the samples
to the corresponding peak areas in a calibration solution. The
same experimental set up was used in order to assess corrected
extraction yields with the exception that the IS was
added post extraction. The corrected extraction yields were
based on peak area ratios (analyte peak area/IS peak area) of
the samples to the corresponding peak areas ratios in a calibration
solution.
2.7. Quantification
The antimycotics were identified by retention time, precursor
and product ion m/z. The most abundant transitions were
monitored and the IS calibration method (peak area ratios of
analytes and IS) was used for all substances except fluconazole
(external calibration without recovery correction). The
ten-point calibration curve, based on 500 ng of sulconazole
and ranging from 1 ng mL 1 to 1000 ng mL 1 of the analytes,
was prepared in aqueous acetonitrile (10%).
2.8. Mass flow calculations
Mass flows (g day 1 ) of each antimycotic substance at each
STP were calculated using the obtained concentrations and
the aqueous and solid flows during the sample collection
periods, and the daily production of digested dewatered
sludge (dw) obtained from monthly production figures
water research 44 (2010) 649–657 653
supplied by staff of the STPs. The predicted mass flows
(g day 1 ) were based on total sales of the active substances to
both institutional and non-institutional care during October
and November 2007 (Svensson, 2008), the number of people
connected to each STP and the total population of Sweden
(Statistics Sweden).
3. Results and discussion
3.1. Quantification
At least two transitions were monitored for each of the antimycotics
(see Table 2), except ketoconazole, for which only
one detectable product ion was formed, high CEs only resulted
in additional low m/z product ions with weak signal intensities.
An approach that could have been used to improve the
level of confirmation is the use of the other Cl-isotope of
ketoconazole, unfortunately it was not considered in this
work. In addition, although two product ions were recorded
for fluconazole, signals from one of them were often severely
suppressed by matrix components, so it was not used for
quantification. In general, matrix suppression of the analyte
and IS precursor/product ion signals was immense and in this
context it should be noted that the Evolute ABN sorbent
column used in the SPE process was designed specifically for
quantifying analytes in dirty extracts, since it reduces interference
by matrix components such as macromolecules, but
the sewage water and sludge extracts were brownish in color,
indicating that it failed to remove substantial proportions of
low molecular weight substances from these samples. It
should also be noted that IS calibration was not used to
quantify fluconazole, since its extraction yields were much
higher than those of the other antimycotics (including the IS),
and it correlated most poorly with the IS in terms of retention
time and Log Kow. In addition, the IS was not natively present
in the samples, thus an underestimation of the antimycotics
was avoided. Terconazole was also considered as an IS,
however, traces of it were found in the sewage waters and
sludge from the southern part of Sweden, most likely due to
its use as a pesticide in the agricultural areas, and it was
therefore not used.
Table 2 – Monitored ions, retention times, method detection limits and extraction yields of the antimycotics in aqueous and
solid matrices.
MS/MS transitions CE a (AU) t r (min) MLOQs b
Extraction
yields (%)
IS corrected
extraction yields (%)
Waters (ng L 1 ) Sludge (mgkg 1 ) Waters Solids Waters Solids
Terbinafine 292.8 > 141.4, 205.4, 170.3 31 17.3 5 3 27 9 40 7 133 48 109 26
Clotrimazole 277.2 > 165.5, 242.5 35 17.2 5 3 23 5 34 8 117 31 79 6
Ketoconazole 532.9 > 490.7 33 14.9 20 10 9 5 48 15 104 21 108 9
Fluconazole 307.3 > 238.3 23 13.3 50 40 104 12 64 6 – –
Econazole 383.0 > 125.3, 193.1 34 18.6 80 40 10 6 53 16 115 21 115 10
Miconazole 417.7 > 161.4, 229.4 38 19.8 100 50 5 3 41 12 69 28 95 17
Sulconazole (IS) 399.3 > 330.8, 183.2 22 19.1 – 13 10 43 10 – –
a Collision energy in LC–MS/MS (arbitrary units).
b Method limits of quantification based on the second calibration curve point in terms of mass per liter (liquid samples) and dry weight (sludge
samples).

654
3.2. Concentrations and mass flows of the antimycotics
in sewage water and sludge
The results from the Umea˚ STP antimycotic fate and the
national screening study are presented in Table 3. The
concentrations of the antimycotics in both the aqueous and
solid matrices were generally similar to, or slightly lower than,
previously reported levels of antimycotics (Kahle et al., 2008)
and other pharmaceuticals, such as antibiotics (Golet et al.,
2003; Carballa et al., 2004; Göbel et al., 2005a,b; Lindberg et al.,
2005, 2006). However, the only substances found at higher
levels than their respective LOQs in the aqueous samples were
fluconazole (in both the Umea˚ STP and hospital samples) and
ketoconazole (in the hospital samples). The levels found in the
final effluent were in close correlation with the raw sewage
water (just below LOQ in raw sewage water of Stockholm but
the chromatographic peak was identified and above three
times the signal to noise ratio) hence elimination from the
aqueous phase is minimal. Fluconazole was not present in
any of the sludge samples thus it is the only one of the
monitored antimycotics that will be directly released into
recipient waters of the STPs. Kahle et al. came to the same
conclusion regarding the fate of fluconazole in their study
made at Swiss STPs (Kahle et al., 2008). Three of the other
antimycotics (clotrimazole, ketoconazole and econazole) were
detected in sludge samples collected from all of the studied
STPs. However, miconazole was not detected in any samples
Table 3 – Sewage water and sludge antimycotic concentrations.
National screening study a,b
STP Um STP St STP Al STP Go STP Bo UCH
RSW, ng L 1
from the Alingsa˚s or Bollebygd STPs, and terbinafine was not
detected in samples from the Umea˚ and Alingsa˚s STPs.
According to EMEA guidelines (see Table 1), predicted environmental
surface water concentrations can only be considered
acceptable for fluconazole, and at best misleading for the
other substances, since sludge is not included in the equation.
It should be noted here that the distribution of the antimycotics
between the aqueous and solid phases seems to be
consistent with their respective Log Kow values. However,
models based on Log K ow are not suitable to predict the environmental
fate of pharmaceuticals (Tolls, 2001), and
substances with similar Log K ow values have been shown to
sorb to sludge to widely differing degrees (Lindberg et al., 2005,
2006).
Antimycotic mass flows (g day 1 ) were established to: (a)
investigate the absolute distribution of these substances
between the aqueous and solid phases, and (b) cross-validate
them with mass flows based on sales statistics. The
recorded mass flows from the STP screening study are
presented together with the predicted values in Fig. 1. The
recorded antimycotic mass flows were similar (within an
order of magnitude) to their respective predicted equivalents
at each STP with two exceptions: a) the measured
mass flows in Bollebygd STP were ca. 100-fold lower than
predicted, possibly because it is the smallest of the investigated
STPs, and that Bollebygd is a small town lacking
both a hospital and large industries; and b) the much lower
RSW, ng L 1
Terbinafine d d d d d d d
Clotrimazole d d d d d d d
Ketoconazole d d d d d 30 d
Fluconazole 90 d d d d 570 120
Econazole d d d d d d d
Miconazole d d d d d d d
FE, ng L 1
water research 44 (2010) 649–657
FE, ng L 1
Terbinafine d d d d d d
Clotrimazole d d d d d d
Ketoconazole d d d d d d
Fluconazole 140 100 d d d 100
Econazole d d d d d d
Miconazole d d d d d d
Fate in Umea˚ STP study a,c
DDS, mgkg 1 (dw) RSP, mgkg 1 (dw) RS, mgkg 1 (dw) DDS, mgkg 1 (dw)
Terbinafine d 4 d 7 30 d 40 d
Clotrimazole 30 120 120 110 50 10 310 60
Ketoconazole 910 910 480 670 280 980 1300 1800
Fluconazole d d d d d d d d
Econazole 590 1000 660 680 210 470 240 550
Miconazole 160 970 d 190 d d d 370
a RSW ¼ raw sewage water, FE ¼ Final effluent, DDS ¼ digested dewatered sludge, RSP ¼ raw sewage particles, and RS ¼ raw sludge.
b STP Um ¼ Umea˚ ; STP St ¼ Stockholm, Henriksdal; STP Al ¼ Alingsa˚ s; STP Go ¼ Gothenburg, Ryaverken; STP Bo ¼ Bollebygd; UCH ¼ Umea˚
county hospital (Norrlands Universitetssjukhus).
c Mean concentrations from two days of sampling.
d Below LOQ.

observed mass flow of terbinafine, which suggests that this
topically used substance (of the ones included) is only to
a small extent present in the sewage waters post application.
In Umea˚, the use of ketoconazole and fluconazole in
the institutional care accounts for ca. 23% and 56% of the
total amounts sold, respectively. However, while the mass
flow of fluconazole in the UCH sewage water represented
31% of the total mass flow in the STP raw sewage water,
a substantially smaller proportion of the ketoconazole of
UCH (1%) was found in the solid flows of the STP (digested
dewatered sludge). In addition, mean mass flow values
normalized to the number of people connected to each STP
indicated that 53% of the purchased fluconazole eventually
appeared in the final effluents, while 1, 35, 41, 155 and
209% of terbinafine, ketoconazole, miconazole, clotrimazole
and econazole, respectively, appeared in digested dewatered
sludge. In addition, the relative distribution between
the antimycotics based on sales statistics (total sales nonand
institutional care) in comparison to the observed
results in dewatered digested sludge, with the exception of
final effluents for fluconazole, can be seen in Fig. 2.
Although the comparison would be more accurate by using
raw sewage water and raw sewage particles and sales
statistics of each catchment area of the STPs, the posttreatment
findings at the STPs show similar distribution of
the antimycotics as to the predicted equivalent.
water research 44 (2010) 649–657 655
Fig. 2 – The relative distribution of antimycotics based on
sales records compared to observed results. Abbreviations:
STP Um [ Umea˚; STP St [ Stockholm, Henriksdal; STP
Al [ Alingsa˚s; STP Go [ Gothenburg, Ryaverken; STP
Bo [ Bollebygd; STP mean [ mean value of the
distribution of the antimycotics.
3.3. Fate analysis within Umea˚ sewage treatment plant
The mass flow results from the fate analysis in Umea˚ STP are
given in Fig. 3. Again, fluconazole was not detected in any of
the solid matrices, and its mass flows in the raw sewage water
Fig. 3 – Mass flows of antimycotics (g day L1 ) within the Umea˚ sewage treatment plant.

656
and the final effluent were very similar (removal efficiency ca.
13%). Thus, these results corroborate the indications from the
screening study that fluconazole is at best marginally affected
by current sewage water treatments. In contrast, clotrimazole,
ketoconazole and econazole were all present in the solid flow
of the raw sewage water, and they were also, together with
terbinafine, detected in the raw sludge. The results for clotrimazole
are consistent with data presented by Kahle et al.
(2008), who found extractable amounts of this substance in
the suspended solids of raw sewage water of Swiss STPs.
Clotrimazole, ketoconazole, econazole and terbinafine were
not strongly affected by sludge digestion either, and all of
them (apart from terbinafine) were found at levels exceeding
their respective LOQs in the resultant sludge. Traces of
miconazole were also observed post-digestion.
Our results show that fluconazole is released at significant
levels into the aquatic environment, and although this
substance is reportedly the least effective CYP19 aromatase
inhibitor of the azole antimycotics (Trösken et al., 2004), it is
a moderate Cyp3A4 and a potent Cyp2C9 inhibitor, and thus
may have significant direct ecotoxicological effects. The
possibility that it may also have adverse indirect environmental
effects cannot be excluded, since (for example) ketoconazole
has been shown to increase the sensitivity of
rainbow trout to ethynylestradiol (Hasselberg et al., 2008).
However, sludge is the major route for antimycotics into the
environment, and thus their exposure levels and potential
negative effects will depend on the fate of the sludge. Sludge is
commonly used as a fertilizing agent (e.g. for golf courses and
forests) or as landfill material, so we recommend that future
environmental risk assessments of this group of pharmaceuticals,
not only focus on the aqueous environment, but also
consider their potential impact on terrestrial organisms.
4. Conclusions
The levels of fluconazole in the raw sewage water were
similar to those of the final effluent and in addition it was
not observed in digested dewatered sludge.
Terbinafine, clotrimazole, ketoconazole, econazole and
miconazole were only present in sludge samples, in turn,
not strongly affected by sludge digestion.
Observed antimycotic mass flows were in general similar to
their respective predicted equivalents at each STP except for
terbinafine and for the smallest STP included in this study.
Future environmental risk assessments of antimycotics
should consider both the aqueous and the terrestrial
environment.
Acknowledgements
We gratefully acknowledge financial support from the
Swedish Research Council for Environment, Agricultural
Sciences and Spatial Planning (FORMAS). We also thank the
personnel at the sewage treatment plants for their assistance.
water research 44 (2010) 649–657
references
Carballa, M., Omil, F., Lema, J.M., Llompart, M., Garcia-Jares, C.,
Rodriguez, I., Gomez, M., Ternes, T., 2004. Behaviour of
pharmaceuticals, cosmetics and hormones in a sewage
treatment plant. Water Res. 38, 2918–2926.
EPI SuiteÔ was Developed by the EPA’s Office of Pollution
Prevention Toxics and Syracuse Research Corporation (SRC),
2000. U.S. Environmental Protection Agency.
European Medicines Agency, 2006. Pre-authorisation Evaluation
of Medicines for Human Use. http://www.emea.europa.int/
pdfs/human/swp/444700en.pdf (accessed 11.12.2008).
Göbel, A., Thomsen, A., McArdell, C.S., Joss, A., Giger, W., 2005a.
Occurrence and sorption behaviour of sulfonamides,
macrolides, and trimethoprim in activated sludge treatment.
Environ. Sci. Technol. 39, 3981–3989.
Göbel, A., McArdell, C.S., Suter, M.J.-F., Giger, W., 2005b. Trace
determination of macrolide and sulfonamide antimicrobials,
a human sulfonamide metabolite, and trimethoprim in
wastewater using liquid chromatography coupled to
electrospray tandem mass spectrometry. Anal. Chem. 76,
4756–4764.
Golet, E.M., Xifra, I., Siegrist, H., Alder, A.C., Giger, W., 2003.
Environmental exposure assessment of fluoroquinolone
antibacterial agents from sewage to soil. Environ. Sci. Technol.
37, 3243–3249.
Graybill, J.R., Drutz, D.J., 1980. Ketoconazole – a major innovation
for treatment of fungal disease. Ann. Int. Med. 93, 921–923.
Guengerich, P.F., 2008. Cytochrome P450 and chemical toxicology.
Chem. Res. Toxicol. 21, 70–83.
Gunnarsson, L., Jauhiainen, A., Kristiansson, E., Nerman, O.,
Larsson, D.G.J., 2008. Evolutionary conservation of human
drug targets in organisms used for environmental risk
assessments. Environ. Sci. Technol. 42, 5807–5813.
Gyllenhammar, I., Eriksson, H., Söderqvist, A., Lindberg, R.H.,
Fick, J., Berg, C., 2009. Clotrimazole exposure modulates
aromatase activity in gonads and brain during gonadal
differentiation in Xenopus tropicalis frogs. Aquat. Toxicol. 91,
102–109.
Hasselberg, L., Westerberg, S., Wassmur, B., Celander, M.C., 2008.
Ketoconazole, an antifungal imidazole, increases the
sensitivity of rainbow trout to 17a-ethynylestradiol exposure.
Aquat. Toxicol. 86, 256–264.
Kahle, M., Buerge, I.J., Hauser, A., Müller, M.D., Poiger, T., 2008.
Azole fungicides: occurrence and fate in wastewaters and
surface waters. Environ. Sci. Technol. 42, 7193–7200.
Kolpin, D.W., Skopec, M., Meyer, M.T., Furlong, E.T., Zaugg, S.D.,
2004. Urban contribution of pharmaceuticals and other
organic wastewater contaminants to streams during differing
flow conditions. Sci. Total Environ. 328, 119–130.
Lepesheva, G.I., Waterman, M.R., 2007. Sterol 14
alpha-demethylase cytochrome P450 (CYP51), a P450 in all
biological kingdoms. Biochim. Biophys. Acta-General Subjects
1770, 467–477.
LIF, 2008. Läkemedelsindustriföreningen (the Swedish
association of the pharmaceutical industry). FASS. Elanders,
Kungsbacka.
Lindberg, R.H., Wennberg, P., Johansson, M.I., Tysklind, M.,
Andersson, B.A.V., 2005. Screening of human antibiotic
substances and determination of weekly mass flows in five
sewage treatment plants in Sweden. Environ. Sci. Technol. 39,
3421–3429.
Lindberg, R.H., Olofsson, U., Rendahl, P., Johansson, M.I.,
Tysklind, M., Andersson, B.A.V., 2006. Behavior of
fluoroquinolones and trimethoprim during mechanical,
chemical and active sludge treatment of sewage water and
digestion of sludge. Environ. Sci. Technol. 40, 1042–1048.

Peschka, M., Roberts, P.H., Knepper, T.P., 2007. Analysis, fate
studies and monitoring of the antifungal agent clotrimazole in
the aquatic environment. Anal. Bioanal. Chem. 389, 959–968.
Piferrer, F., Zanuy, S., Carillo, M., Solar, I.I., Devlin, R.H.,
Donaldson, E.M., 1994. Brief treatment with an aromatase
inhibitor during sex differentiation causes chromosomally
female salmon to develop as normal, functional males. J. Exp.
Zool. 270, 255–262.
Roberts, P.H., Bersuder, P., 2006. Analysis of OSPAR priority
pharmaceuticals using high-performance liquid
chromatography–electrospray ionisation tandem mass
spectrometry. J. Chromatogr. A 1134, 143–150.
Roberts, P.H., Thomas, K.V., 2006. The occurrence of selected
pharmaceuticals in wastewater effluent and surface waters of
the lower Tyne catchment. Sci. Total Environ. 356, 143–153.
Ryder, N.S., 1992. Terbinafine: mode of action and properties of the
squalene epoxidase inhibition. Br. J. Dermatol. 126 (Suppl. 39), 2–7.
Statistics Sweden. http://www.scb.se/templates/
tableOrChart____78315.asp (accessed 02.10.08).
Svensson, Elin, 2008. Apoteket AB, Sweden. Personal
communication.
water research 44 (2010) 649–657 657
Sweetman, S.C. (Ed.), 2007. Martindale: The Complete Drug
Reference. Pharmaceutical Press, London Electronic version,
Edition 070214.
Thomas, K.V., Hilton, M.J., 2004. The occurrence of selected
human pharmaceutical compounds in UK estuaries. Mar.
Pollut. Bull. 49, 436–444.
Tolls, J., 2001. Sorption of veterinary pharmaceuticals in soils:
a review. Environ. Sci. Technol. 35, 3397–3406.
Trösken, E.R., Scholz, K., Lutz, R.W., Volkel, W., Zarn, J.A., Lutz, W.
K., 2004. Comparative assessment of the inhibition of
recombinant human CYP19 (aromatase) by azoles used in
agriculture and as drugs for humans. Endocr. Res. 30, 387–394.
Van De Steene, J.C., Lambert, W.E., 2008. Validation of a solidphase
extraction and liquid chromatography–electrospray
tandem mass spectrometric method for the determination of
nine basic pharmaceuticals in wastewater and surface water
samples. J. Chromatogr. A 1182, 153–160.
Yoshida, Y., Aoyama, Y., Noshiro, M., Gotoh, O., 2000. Sterol 14demethylase
P450 (CYP51) provides a breakthrough for the
discussion on the evolution of cytochrome P450 gene
superfamily. Biochem. Biophys. Res. Commun. 273 (3), 799–804.

Pharmaceuticals and personal care products in archived U.S.
biosolids from the 2001 EPA national sewage sludge survey
Kristin McClellan, Rolf U. Halden*
Center for Environmental Biotechnology, The Biodesign Institute at Arizona State University, 1001 S. McAllister Avenue, Tempe,
AZ 85287-5701, USA
article info
Article history:
Received 1 July 2009
Received in revised form
15 December 2009
Accepted 21 December 2009
Available online 4 January 2010
Keywords:
Municipal sludge
Organic wastewater contaminants
Risk assessment
Land application
1. Introduction
abstract
Pharmaceuticals and personal care products (PPCPs) are
common contaminants of the environment, and have been
detected in surface water (Kolpin et al., 2002; Cahill et al.,
2004; Moldovan, 2006; Roberts and Thomas, 2006; Tamtam
et al., 2008), groundwater (Heberer et al., 2000; Lindsey et al.,
water research 44 (2010) 658–668
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
In response to the U.S. National Academies’ call for a better assessment of chemical
pollutants contained in the approximately 7 million dry tons of digested municipal sludge
produced annually in the United States, the mean concentration of 72 pharmaceuticals and
personal care products (PPCP) were determined in 110 biosolids samples collected by the
U.S. Environmental Protection Agency (EPA) in its 2001 National Sewage Sludge Survey.
Composite samples of archived biosolids, collected at 94 U.S. wastewater treatment plants
from 32 states and the District of Columbia, were analyzed by liquid chromatography
tandem mass spectrometry using EPA Method 1694. Thirty-eight (54%) of the 72 analytes
were detected in at least one composite sample at concentrations ranging from 0.002 to
48 mg kg 1 dry weight. Triclocarban and triclosan were the most abundant analytes with
mean concentrations of 36 8 and 12.6 3.8 mg kg 1 (n ¼ 5), respectively, accounting for
65% of the total PPCP mass found. The loading to U.S. soils from nationwide biosolids
recycling was estimated at 210–250 metric tons per year for the sum of the 72 PPCPs
investigated. The results of this nationwide reconnaissance of PPCPs in archived U.S.
biosolids mirror in contaminant occurrences, frequencies and concentrations, those
reported by the U.S. EPA for samples collected in 2006/2007. This demonstrates that PPCP
releases in U.S. biosolids have been ongoing for many years and the most abundant PPCPs
appear to show limited fluctuations in mass over time when assessed on a nationwide
basis. The here demonstrated use of five mega composite samples holds promise for
conducting cost-effective, routine monitoring on a regional and national basis.
ª 2009 Elsevier Ltd. All rights reserved.
2001; Fick et al., 2009), and drinking water (Stackelberg et al.,
2004; Loraine and Pettigrove, 2006; Loos et al., 2007; Focazio
et al., 2008), as well as in agricultural soils subject to land
application of digested municipal sludge (Kinney et al., 2008;
Kupper et al., 2004; Wu et al., 2009), also known as biosolids.
Wastewater treatment plants were identified as one possible
source for surface water contamination. Over-the-counter
Abbreviations: DL, Detection limit; EPA, Environmental protection agency; NEBRA, North east biosolids and residuals association;
NSSS, National sewage sludge survey; PPCP, Pharmaceuticals and personal care products; RPD, Relative percent difference;
TNSSS, Targeted national sewage sludge survey.
* Corresponding author. Tel.: þ1 480 727 0893; fax: þ1 480 727 0889.
E-mail addresses: Kristin.McClellan@asu.edu (K. McClellan), Halden@asu.edu (R.U. Halden).
0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.watres.2009.12.032

and prescription drugs enter the wastewater via excretion of
urine and feces containing parental drugs and their conjugates
as well as other metabolites, or from disposal of
unwanted or expired medications (Halling-Sorensen et al.,
1998; Fent et al., 2006). Similarly, chemical constituents of
personal care products may be directly disposed of into
domestic wastewater. Removal of PPCPs during municipal
wastewater treatment is rarely complete, thereby creating
a pathway for entry of these compounds into aquatic environments
via wastewater reclamation (Halling-Sorensen
et al., 1998; Ternes, 1998; Daughton and Ternes, 1999; Hirsch
et al., 1999) and into terrestrial environments via land
application of biosolids (Ternes et al., 2004a).
Of the more than 7 million tons of sewage sludge produced
in the United States in 2004, about 50% was applied to land as
fertilizer or soil amendment, and 45% was disposed of in
landfills or as landfill cover (NEBRA, 2007). Terrestrial environments
can offer effective biological, physical, and chemical
attenuation mechanisms for manmade pollutants. However,
they also can act as a source term for chemical migration into
surface and groundwater from biosolids runoff and leachate.
Pharmaceutical compounds are designed to be biologically
active and therefore may have effects on non-target organisms
even at trace concentrations extant in terrestrial and
aquatic environments. While acute toxic effects of pharmaceuticals
on non-target organisms have been investigated for
some compounds, chronic toxicity and potential subtle environmental
effects are only scarcely known (Fent et al., 2006).
Also insufficiently investigated is the effect of mixtures of
pharmaceuticals on aquatic organisms, although biochemical
interactions of drugs in humans are well known. Of additional
concern is the possible uptake of contaminants into food
crops grown on agricultural fields that were fertilized with
biosolids (Kumar et al., 2005; Dolliver et al., 2007). Currently,
no regulation exists in the United States for PPCPs contained
in biosolids, and a need for more information on the occurrence
of and risk from these compounds has been noted by
the National Research Council of the National Academies of
the United States (National Research Council, 2002). In the
past, analytical methods were limited, especially for trace
analyses of complex environmental samples. In 2007, the
release of U.S. EPA method 1694 (USEPA, 2007a) for the analysis
of PPCPs in various matrices afforded the opportunity to
analyze biosolids samples using a standardized protocol.
The U.S. EPA has performed national sewage sludge surveys
(NSSS) in 1989, 2001, and 2007. The survey conducted in 2001
served to evaluate the potential need for regulations of trace
levels of dioxins and polychlorinated biphenyls (USEPA, 2007b).
After the 2001 survey was completed, unused samples were
released to a nationwide repository of biosolids samples now
maintained at the Biodesign Institute at Arizona State University.
This investigation evaluated occurrences and concentrations
of PPCPs in biosolids from the year 2001 to enable risk
assessments and to establish a national baseline for evaluating
temporal trends of PPCPs in U.S. biosolids. The analysis
of composite samples by EPA Method 1694 was employed to
determine average concentrations of 72 PPCPs in archived
biosolids collected by the EPA, as a representative sample of
the more than 16,000 treatment plants located in the contiguous
United States.
water research 44 (2010) 658–668 659
2. Materials and methods
2.1. Sampling procedure
Biosolids samples with solid contents between 1% and 30%
were obtained from 94 wastewater treatment plants in 32
states and the District of Columbia for the 2001 National
Sewage Sludge Survey (USEPA, 2007b). They were selected by
the U.S. EPA to obtain a representative estimate of the
occurrence of chemical contaminants in sewage sludge that is
disposed of primarily by land application. Information on the
exact sampling locations is available in Table S1 of the
supplementary material (SM). This survey aimed to estimate
levels of dioxins, dibenzofurans, and coplanar polychlorinated
biphenyls in biosolids. The sampling was conducted
by the U.S. EPA between February and March 2001, and
samples were collected according to sampling procedures
developed by the U.S. EPA (USEPA, 2001). Samples were only
taken from fully processed sewage sludges intended for
disposal. Eighty-nine of the 94 WWTPs had one single system
for sludge treatment, therefore one sample was collected. Five
facilities had two systems for treating their sludges, therefore
two samples were taken from each of these plants. In addition,
duplicate samples were collected from 15% of facilities
(14 samples) for precision analysis. This amounted to 113
samples overall. After completion of the 2001 NSSS, the
samples were acquired by the Halden laboratory for further
studies. For the 7-year period between acquisition and analysis
in the spring of 2008, samples were stored at 20 C.
2.2. Composite sample preparation
From the 113 biosolids samples acquired from the EPA, three
were excluded from analysis because the sample containers
were broken or compromised; the remaining 110 samples were
randomly grouped into five groups. Composite samples were
prepared by weighing out approximately 1 g of dry weight from
each sample and pooling it to obtain 5 composites each containing
solids from between 21 and 24 individual samples.
A duplicate of composite sample #3 was prepared to serve as
a blind duplicate.
2.3. Sample analysis
The samples were analyzed by AXYS Analytical Services (2045
Mills Road West, Sydney, British Columbia, Canada V8L 3S8)
according to EPA method 1694 (USEPA, 2007a). For the purpose
of compound detection, the 72 analytes were divided into four
groups. All analytes were separated by liquid chromatography
and detected by tandem mass spectrometry. For compounds
with a respective labeled analog, the concentration was
determined using the isotope dilution technique. The corresponding
concentrations are deemed to be of high quality. For
compounds where a labeled analog was not available, the
concentration was determined using an external calibration.
Quantitative data for analytes not determined by the isotope
dilution method are judged to be less robust. More detailed
information on the analysis method and accuracy and precision
criteria for acceptance of analytical data is available in

660
supplementary material (SM). Further information on study
limitations can be found in the discussion section.
2.4. Quality assurance
To ensure system and laboratory performance, several tests
were performed before sample analysis. Calibration accuracy
was verified using a calibration standard solution with labeled
and native analytes. Retention times of native and labeled
compounds had to be within 15 seconds of the respective
retention time established during the previous calibration. In
addition, ongoing precision and recovery were ensured. Lab
blanks were analyzed before each sample analysis. A duplicate
sample analysis was performed by the lab for each batch
between 7 and 20 samples. In addition to these standard
procedures, a blind duplicate was included in the sample set to
evaluate analysis precision. Precision is expressed as relative
percent difference (RPD) between each pair of measured
concentrations. It was calculated using the following equation,
RPD½%Š ¼ jCsample Cduplicatej 100
Csample þ Cduplicate
2
where C sample and C duplicate are, respectively, the concentration
detected in the original sample and in its duplicate
sample, and RPD is the relative percent difference.
2.5. Modeling of soil and porewater concentration
For assessment of their potential environmental impact, the
concentrations of PPCPs after mixing with agricultural soil
were calculated. The mixing ratio with soil was assumed
based on the EPA-recommended rate of biosolids application
of up to 4.5 dry kg per m 2 and by further assuming incorporation
into soil to a depth of 10 cm (USEPA, 1994). The soil bulk
density was assumed to equal 1.3 g cm 3 and the biosolids
bulk density 1.6 g cm 3 . Though soil moisture content is
known to vary widely depending on soil properties, for the
purpose of this calculation, a soil moisture content of 22% (v/v)
was used, as reported by others for agricultural soil (De Lannoy
et al., 2006). A typical soil organic carbon fraction was
assumed (Causarano et al., 2008), as was a biosolids organic
carbon fraction of 0.4 (USEPA, 2007b). The concentration in soil
and porewater at equilibrium was calculated based on the
organic carbon fraction of the soil-biosolids mixture and the
compound specific organic-carbon distribution coefficient
(K OC) of each analyte (Chalew and Halden, 2008). Calculations
took into consideration the soil moisture content, as well as
the volume of biosolids added to the soil, and were conducted
for an environmentally relevant pH range of 7–9. The
concentrations of PPCPs in porewater of soil/biosolid mixtures
were calculated using the equations below,
mbiosolids
Cbiosolids
msoilþbiosolids
Cporewater ¼
fporewater rporewater þ KOC fOC
msoilþbiosolids
Csoil ¼ mbiosolids
Cbiosolids
msoilþbiosolids
fporewater rporewater Cporewater
msoilþbiosolids
where m is the dry mass in kg m 3 of the solid matrix, C the
concentration in mgkg 1 , r the density in kg m 3 , and fporewater
water research 44 (2010) 658–668
(1)
(2)
(3)
and f OC the dimensionless fractions of, respectively, porewater
and organic carbon in the soil/biosolids mixture.
2.6. Drug usage and ecotoxicity data
Information of prescription drug consumption was obtained
from Internet sources (www.drugtopics.com). Exotoxicity
data were taken from EPA’s Ecotox database (www.epa.gov/
ecotox).
2.7. Modeling of annual loading to agricultural soil
The annual loading of PPCPs contained in biosolids was based
on a production of 5.6–7 million dry tons of sewage sludge in
the U.S., of which 50–60% is applied to land (National Research
Council, 2002; Jones-Lepp and Stevens, 2007; NEBRA, 2007).
3. Results
3.1. Data quality assurance
Lab blanks showed no detections above the detection limit for
any of the analytes except for ciprofloxacin [61 mg kg 1 dry
weight (dw); detection limit (DL) 32 mg kg 1 dw] and erythromycin-H2O
(2.6 mg kg 1 dw; DL 1.5 mg kg 1 dw)]. However,
concentrations detected in biosolids were 100- and 40-times
greater than the background level detected in lab blanks.
Therefore, measured concentrations for both analytes were
accepted.
Recovery for all analytes typically was good with an average
of 112% but some notable outliers were observed leading to
a range of 12–493% (See Table 1 and Table S2, SM). A total of
10 analytes exceeded the methods’ lower and upper control
limits. For 5 analytes (anhydrotetracycline, azithromycin,
cimetidine, 4-epianhydrochlortetracycline, and 4-epianhydrotetracycline)
recovery rates were below the method’s lower
control limit of 40%. The concentrations reported for these
analytes in biosolids may represent underestimates. The
recovery rates of 5 analytes (clinafloxacin, enrofloxacin,
lomefloxacin, ofloxacin, and sarafloxacin) were above the
method’s upper control limits. The respective concentrations
reported for these analytes may represent overestimations. All
of the above mentioned analytes were quantified using labeled
surrogate standards. In the case of clinafloxacin, enrofloxacin,
lomefloxacin, ofloxacin, and sarafloxacin, 13 C3- 15 N-ciprofloxacin
was used as an isotope-labeled surrogate standard.
The recoveries were determined from spiked quality control
samples, where recovery of 13 C3- 15 N-ciprofloxacin was below
the method’s lower control limit. This led to extremely high
recoveries of the above mentioned analytes. In the biosolids
samples, recovery of 13 C 3- 15 N-ciprofloxacin was always within
the method’s control limits. However, the lack of adequate
recovery of 13 C 3- 15 N-ciprofloxacin in quality control samples
suggests that concentrations reported in biosolids for clinafloxacin,
enrofloxacin, lomefloxacin, ofloxacin, and sarafloxacin
should be interpreted with caution.
The duplicate analysis revealed 18% relative percent
difference (RPD) for all analytes. The blind duplicate analysis
of a subset of 10 analytes revealed 28% RPD. Both RPD values

Table 1 – Analytical results and summary statistics for pharmaceuticals and personal care products detected in U.S. biosolids collected in 2001 and reported on a dry weight basis.
Substance name CAS RN Detection
limit
[mg kg 1 ]
Anhydrotetracycline g
Azithromycin g
Standard
deviation
of detection
limit [mgkg 1 ]
n ¼ 5
Frequency
measured
[%]
Maximum
concentration
[mg kg 1 ]
Mean detected
concentration
[mg kg 1 ]
Standard
deviation
of mean
concentration
[mg kg 1 ] n ¼ 5
Recovery
[%]
Isotope
dilution
quantification
Use Lowest effect
concentration
for aquatic
biota [mg L 1 ]
Projected
annual
land application
[kg yr 1 ] f
4496-85-9 74.1 7.4 100 880 392 249 51 Antibiotic 1300–1600
83905-01-5 5.6 1.6 100 1220 838 224 12 Antibiotic 2800–3500
Caffeine 58-08-2 59.0 16.4 100 643 248 200 78 Stimulant
Carbamazepine 298-46-4 5.6 1.6 100 238 163 56.4 139 Anticonvulsant 100 b (Jos
Chlortetracycline 57-62-5 22.9 6.2 60 43.5 23.4 16.9 159 Antibiotic 36 c (Brain
Cimetidine g
0.05 a (Bantle
et al., 1994)
et al., 2003)
et al., 2004)
51481-61-9 6.4 2.0 100 893 504 208 41 Antacid 1700–2100
Ciprofloxacin 85721-33-1 20.9 6.4 100 10800 6858 2348 98 Antibiotic
5 d (Halling-
Sorensen, 2001)
Clarithromycin 81103-11-9 5.6 1.6 100 94.6 66.2 25.5 114 Antibiotic 220–270
Codeine 76-57-3 11.2 3.2 20 29.7 n.d. 114 Analgesic
Cotinine 486-56-6 9.9 3.3 100 38.6 28.1 8.3 103
Nicotine
metabolite
Diltiazem 42399-41-7 1.3 0.1 100 109 45.2 34.2 110 Antianginal 150–190
Diphenhydramine 58-73-1 2.3 0.6 100 1740 1166 516 101 Antihistamine 3900–4800
Doxycycline 564-25-0 23.9 6.3 100 1780 966 436 85 Antibiotic 3200–4000
Enrofloxacin h
93106-60-6 19.5 6.5 60 28.6 n.d. 190 Antibiotic 49 d (Robinson
et al., 2005)
4-Epianhydrotetracycline
g
4465-65-0 77.1 20.2 100 399 261 71.7 39 Antibiotic 900–1100
4-Epichlortetracycline
14297-93-9 56.3 16.0 40 93.0 n.d. 161 Antibiotic
4-Epitetracycline 23313-80-6 74.4 18.8 100 3040 2376 517 82 Antibiotic 8000–9800
Erythromycin-H 2O 114-07-8 1.6 0.2 100 183 81.5 52.3 97 Antibiotic
Fluoxetine 54910-89-3 8.2 1.0 100 258 171 46.6 89 Antidepressant
Gemfibrozil 25812-30-0 12.2 12.3 100 159 152 13.2 107
Antihyperlipidemic
22 700 c (Williams
et al., 1992)
36 h, c (Flaherty
and Dodson, 2005)
30,040 c (Zurita
et al., 2007)
Ibuprofen 15687-27-1 122 123 80 359 246 121 109 Anti-inflammatory 830–1000
Isochlortetracycline 514-53-4 22.5 6.4 60 36.0 n.d. 70 Antibiotic
Lomefloxacin h
98079-51-7 13.2 2.1 40 16.1 n.d. 388 Antibiotic 106 e (Robinson
et al., 2005)
Metformin 657-24-9 119 32.5 80 456 305 152 116 Antidiabetic 1000–1300
Miconazole 22916-47-8 5.8 1.3 100 1100 777 266 86 Antifungal 2600–3200
Minocycline 10118-90-8 1880 524 80 2630 1884 939 54 Antibiotic 6300–7800
830–1000
550–680
80–100
23,000–28,000
100–120
270–340
580–710
510–630
(continued on next page)

Table 1 (continued)
Substance name CAS RN Detection
limit
[mg kg 1 ]
Standard
deviation
of detection
limit [mgkg 1 ]
n ¼ 5
Frequency
measured
[%]
Maximum
concentration
[mg kg 1 ]
Mean detected
concentration
[mg kg 1 ]
Standard
deviation
of mean
concentration
[mg kg 1 ] n ¼ 5
Naproxen 22204-53-1 24.3 24.6 100 273 119 79 106
Recovery
[%]
Isotope
dilution
quantification
Antiinflammatory
Use Lowest effect
concentration
for aquatic
biota [mg L 1 ]
Projected
annual
land application
[kg yr 1 ] f
Norfloxacin 70458-96-7 63.5 6.3 100 418 289 74.0 136 Antibiotic 18 000 b (Yang
et al., 2008)
970–1200
Ofloxacin h
82419-36-1 7.7 4.2 100 8140 5446 1941 493 Antibiotic 21 d (Robinson
et al., 2005)
18000–23,000
Oxytetracyclin 79-57-2 22.8 5.8 100 114 87.5 22.2 90 Antibiotic 50 d (Hanson
et al., 2006)
300–360
Ranitidine 66357-35-5 6.5 1.7 100 30.1 21.0 8.0 51 Antacid 70–90
Sulfamethoxazole 723-46-6 2.9 0.7 20 3.3 n.d. 98 Antibiotic
9 e (Brain
et al., 2008)
Sulfanilamide 63-74-1 56.2 16.0 40 87.3 n.d. 101 Antibiotic
Tetracycline 60-54-8 57.5 15.5 100 2790 1914 691 124 Antibiotic 47 e (Brain
et al., 2004)
Thiabendazole 148-79-8 5.9 1.2 100 370 110 131 103 Fungicide 310 c (EPA, 2000) 370–460
Triclocarban 101-20-2 183 67.0 100 48100 36060 8049 96 Disinfectant
Triclosan 3380-34-5 487 493 100 19700 12640 3816 105 Disinfectant
Trimethoprim 738-70-5 10.6 3.8 60 60.5 26.0 21.5 97 Antibiotic
0.101 c
(EPA/OTS, 1992)
0.12 b (Wilson
et al., 2003)
16 000 b (Luetzhoft
et al., 1999)
400–500
6400–7900
120,000–150,000
42,000–52,000
a Amphibium.
b Green algae.
c Crustacean.
d Cyanobacteria.
e Macrophyte.
f Projected deposition rates of PPCPs on land based on 5.6–6.9 million dry tons of annual sewage sludge production and a land application rate of 60% (National Research Council, 2002; USEPA, 2003; USGS, 2006;
Jones-Lepp and Stevens, 2007).
g Concentration may be underestimated due to low recovery in quality control samples.
h Concentration may be overestimated due to high recovery in quality control samples.
90–110

were in control with respect to the target RPD of 30% or less.
The RPD value improved to 11% for 9 analytes, when excluding
results for metformin, whose lack of detection in the blind
duplicate unfavorably increased the summary statistic for
measurement precision. A non-blinded duplicate analysis,
performed by the contract laboratory for these 10 analytes,
showed an RPD value of 11%.
3.2. Study representativeness and sample integrity
The prolonged storage of samples between sampling event
and analysis may have allowed for the chemical degradation
of labile analytes to occur. Therefore, any results of this study
are conservative with respect to the detection frequency and
concentration of compounds found. In other words, due to
pooling of a large number of samples, analytes occurring
infrequently and at low concentrations may have been diluted
out to below the detection limit. A comparison of the presented
data with the EPA data (USEPA, 2009), reveals that the
mean concentrations of all analytes show no statistically
significant difference within the 95th percentile confidence
interval. Therefore, the prolonged storage did not impair the
detection of multiple analytes at elevated concentrations in
archived samples.
3.3. Occurrence of PPCPs in biosolids
All composites tested positive for at least 26 analytes. Of the 72
PPCPs targeted, 38 (54%) were detected at concentrations
ranging from the low parts-per-billion (ppb) to the parts-permillion
(ppm) range. These 38 PPCPs include 8 that have not
previously been reported in biosolids in the peer-reviewed
literature but that also were observed in the U.S. EPA’s TNSSS
published online (USEPA, 2009). The remaining 34 PPCPs were
Concentration in biosolids [µg kg -1 dry weight]
100
000
10
000
1000
100
10
1
triclocarban
triclosan
‡
ciprofloxacin
ofloxacin
*
4-epitetracycline
tetracycline
minocycline
water research 44 (2010) 658–668 663
‡
diphenhydramine
doxycycline
azithromycin
miconazole
cimetidine
‡
‡
*
anhydrotetracycline
metformin
norfloxacin
‡
*
not detected in any of the composite biosolids samples. The
mean total concentration of all targeted PPCPs combined in
the five composite samples was 74.4 mg kg 1 dw of sewage
sludge 21.4 mg kg 1 standard deviation (n ¼ 5).
The two most abundant contaminants were the disinfectants
triclocarban (48% of total detected PPCP mass) and triclosan
(17%) (Fig. 1). Their mean concentrations were 36 8
and 12.6 3.8 mg kg 1 dw (n ¼ 5), respectively (Table 1). The
second most abundant class of PPCPs found was antibiotics. In
order of decreasing concentration, ciprofloxacin, ofloxacin, 4epitetracycline,
tetracycline, minocycline, doxycycline and
azithromycin were found at concentrations between 6.8 2.3
and 0.8 0.2 mg kg 1 dw (n ¼ 5) (Table 1). The combined mass
of all antibiotics constituted about 29% of the total mass of
PPCPs per sample.
In addition to the 3 tetracyclines reported above, three
tetracycline antibiotics were found for which peer-reviewed
occurrence data in sewage sludge thus far were lacking.
Anhydrotetracycline, 4-epianhydrotetracycline, and chlortetracycline
(in order of decreasing concentration) together
accounted for 6.9 2.5 mg kg 1 dw (n ¼ 5) (Table 1). Also not
previously reported in biosolids were two prescription drugs,
metformin and ranitidine. They together accounted for 0.4%
of the total mass of PPCPs found in sewage sludge composites.
4. Discussion
4.1. Study limitations
For this study, a relatively large number of individual samples
were combined to form 5 composite pools or mega composites.
This approach served to reduce the number of samples to be
analyzed in order to obtain a defensible estimate of mean
*
4-epianhydrotetracycline
caffeine
ibuprofen
fluoxetine
carbamazepine
gemfibrozil
naproxen
thiabendazole
oxytetracycline
erythromycin-H2O clarithromycin
diltiazem
sulfanilamide
*
*
4-epichlortetracycline
cotinine
trimethoprim
chlortetracycline
ranitidine
*
‡*
ppm
ppb
*
‡*
isochlortetracycline
enrofloxacin
codeine
lomefloxacin
Fig. 1 – Rank order of mean concentrations for 38 PPCPs detected in composites of a total of 110 U S biosolids samples from
94 treatment plants in 32 states and the District of Columbia. Newly detected compounds are shown in darker hue. Error
bars depict ± one standard deviation (n [ 5). Some concentrations represent estimates only (z) and some analytes were
detected inconsistently (*).
sulfamethoxazole *

664
analyte concentrations across the various treatment facilities
represented. While being efficient and economical for the
intended purpose, this approach was not well suited to capture
the full spectrum of concentrations of individual PPCPs as
a function of plant type, treatment processes employed, populations
served, as well as of geographical locations and
climate zones represented. As a comparison with the EPA
TNSSS data reveals (USEPA, 2009), the detection frequency of
less abundant analytes was significantly reduced in composite
samples compared to individual sample analysis. Therefore,
analytes that were not detected will represent a conservative
estimate, and may still occur at detectable concentrations in
individual samples from specific plants. While the mega
composite approach cannot serve to determine variability
between the large numbers of WWTPs studied, it was found to
be suitable for identifying major contaminants of concern as
well as their average concentration in a large sample set.
4.2. Sanitizing agents
Triclocarban, which in previous U.S. studies had been found in
concentrations ranging from 5.97 to 51 mg kg 1 dry weight
(dw) (Heidler et al., 2006; Chu and Metcalfe, 2007; Sapkota et al.,
2007), was detected in every composite sample assayed. Also
found in significant amounts was triclosan, which has been
observed in a number of U.S. studies of sewage sludge with
reported concentrations ranging from 0.53 to 30 mg kg 1 dw
(Chu and Metcalfe, 2007; Heidler and Halden, 2007; Kinney
et al., 2008; McAvoy et al., 2002; USEPA, 2003). The EPA TNSSS
(USEPA, 2009) found triclocarban and triclosan at concentrations
of up to 441 and 133 mg kg 1 dw, respectively, with mean
concentrations at 38.7 59.7 and 12 18 mg kg 1 dw (n ¼ 74),
respectively. The relatively high mean concentrations of triclocarban
and triclosan reported here and by the U.S. EPA
(USEPA, 2009) in U.S. biosolids are in line with the intense
usage of these antimicrobials and their high octanol/water
partitioning coefficient (log KOW) of 4.9 and 4.8, respectively
(both at neutral pH), which indicates significant potential of
both compounds for sorption to biosolids (Halden and Paull,
2005). In addition, triclosan was found to be persistent in
biosolids after aeration (Ying et al., 2007). Both triclosan and
triclocarban concentrations found in the present study fall
within the range of concentrations previously reported.
Information on previously reported concentrations of PPCPs in
biosolids is available in Table S2, SM.
4.3. Antibiotics
The most abundant antibiotic was ciprofloxacin (6.8 2.3 mg
kg 1 dw; n ¼ 5), which is among the 30 most prescribed drugs
in the United States, according to Internet sources. Ciprofloxacin,
a metabolite of enrofloxacin, is polar and therefore
prone to electrostatic interactions with the negatively charged
surfaces of microbes that are found in high concentrations
especially in secondary sludge (Ternes et al., 2004b). Ciprofloxacin
has been detected in sewage sludge by several studies
conducted in Sweden and Switzerland. Detected concentrations
ranged from 6 10 5 –11 mg kg 1 dw (Golet et al., 2002;
Lindberg et al., 2005, 2006, 2007). Therefore, the mean
concentration found in the present study (6.8 2.3 mg kg 1
water research 44 (2010) 658–668
dw; n ¼ 5) falls in the mid range of concentrations reported
from outside of the U.S. The EPA TNSSS also identified ciprofloxacin
as an abundant microcontaminant in sewage sludge,
which was detected in every sample analyzed, yielding
a mean concentration of 8.7 8.5 mg kg 1 dw.
At a mean concentration of 5.4 1.9 mg kg 1 dw, ofloxacin
was the fourth most abundant contaminant found in biosolids.
It is fairly hydrophilic (log KOW of 0.2) and does not appear on
the list of the top 200 prescription drugs. Its detection at
elevated levels in every composite sample came as a surprise.
However, ofloxacin had been found in concentrations of up to
2mgkg 1 dw in biosolids from Sweden (Lindberg et al., 2005). It
can be speculated that the carboxyl group contained in ofloxacin
may form an ion complex with exchangeable cations
associated with negatively charged surfaces.
Among the 10 most abundant PPCPs of the present study
were three tetracycline antibiotics, 4-epitetracycline, tetracycline
and minocycline, which had not been reported in
biosolids before in the peer-reviewed literature. In addition,
doxycycline was found as the ninth most abundant PPCP at
a mean concentration of 1 0.4 mg kg 1 dw. It had not been
detected in biosolids from the U.S. before and only once in
a Swedish study that found concentrations similar to those
reported here (Lindberg et al., 2005). All tetracycline antibiotics
are fairly hydrophilic (log KOW of 1.33 for tetracycline/4-epitetracycline,
0.42 for minocycline, and 1.36 for doxycycline).
Despite their hydrophilic character, the mean concentrations
reported here are around 1–2 mg kg 1 dw. Tetracycline antibiotics
are known to precipitate with ions of magnesium, calcium
and ferric iron, and therefore accumulate in the solid fraction
during wastewater treatment. Tetracycline, doxycycline and
minocycline also rank among the 200 prescription drugs most
widely used in the U.S. Other tetracycline antibiotics analyzed
in this study were only found at mean concentrations below
0.4 mg kg 1 dw, but the concentrations of anhydrotetracycline,
4-epianhydrotetracycline and 4-epianhydrochlortetracycline
were possibly underestimated due to low recoveries. The sum
of tetracycline antibiotics found in sewage sludge constitute
about 8 1.3mgkg 1 dw, which is similar to the findings of the
EPA TNSSS that found about 5 mg kg 1 dw (USEPA, 2009).
Azithromycin was found at 0.8 0.2 mg kg 1 dw. It ranked
as the 6th most frequently prescribed drug in 2007 and is also
fairly hydrophobic. Due to both these properties and the fact
that biodegradation of azithromycin was found to be insignificant
(Ericson, 2007), one would expect high concentrations
in biosolids. Yet, there are 7 drugs (not counting triclocarban
and triclosan) that were found at higher concentrations.
However, a low recovery of azithromycin of only 12% may
indicate that actual azithromycin concentrations in biosolids
are much higher than the detected concentration. Azithromycin
has been detected in previous studies at concentrations
of up to 6.5 mg kg 1 dw in the U.S. (Jones-Lepp and
Stevens, 2007) and at up to 0.16 mg kg 1 dw in sludge from
Germany and Switzerland (Gobel et al., 2005a, 2005b).
4.4. Bioavailability and soil/porewater equilibria
To explore the importance of bioavailability of PPCPs
sequestered in biosolids, the concentrations of individual
PPCPs that are anticipated to occur in the solid and liquid

phases upon mixing of land applied biosolids into soil were
calculated (Fig. 2). In fully equilibrated soil-biosolids mixtures
(assuming an EPA-recommended mixing ratio of approximate
25:1), concentrations of PPCPs on soil particles are expected to
fall into the ppb range for most analytes, except for triclocarban,
which was projected to be present at around 1 ppm(w/w)
dw. Since most of the PPCPs found in biosolids are fairly
hydrophobic, their calculated concentrations in porewater at
equilibrium typically were quite low (Fig. 2 B) (Kinney et al.,
2008). A comparison of these estimated dissolved PPCP levels
in porewater with the lowest effect concentrations for aquatic
organisms (red circles in Fig. 2 B) suggests that the leaching of
dissolved PPCPs into surface waters probably does not present
an important mechanism for exposure of aquatic biota for the
majority of analytes detected. Concentrations calculated for
soil porewater typically were several orders of magnitude
below the lowest effect concentration reported for aquatic
organisms. Notable exceptions were 6 analytes (Fig. 2) that are
expected to yield potentially problematic concentrations in
porewater after land application and partitioning of biosolidsderived
PPCPs. These include the antibiotics ciprofloxacin,
ofloxacin and tetracycline, as well as the stimulant caffeine,
Conc. in dry weight soil at
pH 7-9 [µg kg-1 ]
Conc. in porewater
at pH 7-9 [µg L-1 ]
1000
100
10
1
0.
1
0.
01
100
000
10
000
1000
100
10
1
0.
1
0.
01
0.
001
0.
0001
‡
*
‡
‡
*
‡
*
‡*
and the two sanitizing agents triclosan and triclocarban
(Fig. 2).
4.5. Risk assessment data gaps
These results suggest that aside from the 7 notable exceptions
discussed above, the majority of the PPCPs detected in
biosolids in this study likely exert no acute effects on aquatic
organisms, assuming that biosolids are applied as regulated
by the EPA (USEPA, 1994) and that the migration of solids from
agricultural land to surface water via soil erosion and runoff
can be completely prevented. While the former assumption is
plausible, the latter may not always apply, as soil erosion is
a common phenomenon.
However, chronic toxicity as well as effects from
mixtures of PPCPs on non-target organisms cannot be
assessed due to lack of appropriate toxicity data. It has been
speculated that the presence of sub-therapeutic concentrations
of antibiotics may adversely affect soil microbial
community structures as well as induce spreading of resistance
among bacterial pathogens. In addition to influencing
microbial populations, it has been shown that some
*
*
‡*
*
4-epianhydrotetracycline
4-epichlortetracycline
4-epitetracycline
anhydrotetracycliine
azithromycin
caffeine
carbamazepine
chlortetracycline
cimetidine
ciprofloxacin
clarithromycin
codeine
cotinine
diltiazem
diphenhydramine
doxycycline
enrofloxacin
erythromycin-H2O fluoxetine
gemfibrozil
ibuprofen
isochlortetracycline
lomefloxacin
metformin
miconazole
minocycline
naproxen
norfloxacin
ofloxacin
oxytetracycline
ranitidine
sulfamethoxazole
sulfanilamide
tetracycline
thiabendazole
triclocarban
triclosan
trimethoprim
‡
*
‡
‡
*
‡
water research 44 (2010) 658–668 665
*
‡*
*
*
‡*
*
4-epianhydrotetracycline
4-epichlortetracycline
4-epitetracycline
anhydrotetracycliine
azithromycin
caffeine
carbamazepine
chlortetracycline
cimetidine
ciprofloxacin
clarithromycin
codeine
cotinine
diltiazem
diphenhydramine
doxycycline
enrofloxacin
erythromycin-H2O fluoxetine
gemfibrozil
ibuprofen
isochlortetracycline
lomefloxacin
metformin
miconazole
minocycline
naproxen
norfloxacin
ofloxacin
oxytetracycline
ranitidine
sulfamethoxazole
sulfanilamide
tetracycline
thiabendazole
triclocarban
triclosan
trimethoprim
*
*
‡
‡
*
*
*
*
*
*
100
000
10
000
1000
100
10
1
0.
1
0.
01
0.
001
0.
0001
Fig. 2 – Predicted equilibrium concentrations of PPCPs associated with particulates of soil-biosolids mixtures (top) and
dissolved in porewater (bottom) after land application of biosolids on agricultural soil. Data depict the environmentally
relevant range between pH 7 and 9. Circles represent the lowest ecological effect concentrations contained in the EPA Ecotox
database. Some concentrations were calculated based on estimates only (z) and some analytes were detected inconsistently (*).
Effect concentration [µg L -1 ]

666
antibiotics, specifically drugs belonging to the tetracyclines,
fluoroquinolones and sulfonamides, may be taken up by
crop plants (Migliore et al., 2003; Kumar et al., 2005; Dolliver
et al., 2007). This presents a potential exposure pathway to
humans through ingestion of contaminated food and may
result in the promotion of resistant bacteria in humans
(Shoemaker et al., 2001).
Furthermore, information is lacking to determine risks to
the health of agricultural soils and soil-dwelling organisms.
Few studies have examined the half-lives in soils of PPCPs
sequestered in biosolids. The effect of ppm levels of sanitizing
agents on soil microbial communities has rarely been investigated
to date (Liu et al., 2009). Similarly, the half-life of PPCPs
in biosolids-amended soils will require additional research to
inform risk assessment analyses. Passage through municipal
digesters and chemical aging may reduce the bioavailability of
PPCPs and with it the risk of chemical uptake of and exposure
to soil-dwelling organisms. However, a reduced bioavailability
also may imply a prolonged half-life of these compounds in
the environment, with possible delayed release in sensitive
compartments. Furthermore, studies are lacking on the
potential of biosolids-derived antimicrobials and antibiotics to
exert selective pressure for the enrichment of drug-resistant
microorganisms, a scenario demonstrated in vitro (Braoudaki
and Hilton, 2004). Also unavailable are threshold concentrations
for toxic effects in terrestrial organisms of many PPCPs,
including some that have been detected in biosolids at ppm
levels. The EPA’s Ecotox database (www.epa.gov/ecotox)
provides only some toxicity values in aquatic organisms for
the PPCPs investigated here. Bioaccumulation and biomagnification
are other aspects that will need to be considered
for risk assessment purposes. Although PPCPs are not
typically thought of as representing persistent hydrophobic
pollutants, some may be subject to bioaccumulation and
possibly biomagnification thereafter in both terrestrial and
aquatic environments. Bioaccumulation was demonstrated
for triclosan and triclocarban in lab and field studies that
examined uptake of these compounds from soil, sediment
and water (Coogan and La Point, 2008; Higgins et al., 2009;
Kinney et al., 2008).
5. Conclusions
Overall, this study reemphasizes the significance of biosolids
recycling as a mechanism for the release of PPCPs into the
environment. Based on the mean concentrations of all analytes
detected, it is estimated that the total loading to U.S. soils
from nationwide biosolids recycling is on the order of
210–250 metric tons per year for the 72 PPCPs investigated
here.
It is concluded that, despite large variations found by the
U.S. EPA between different treatment plants, mean concentrations
of PPCPs in U.S. biosolids on a nationwide basis have
remained fairly constant between 2001 and 2007 and possibly
longer. Good agreement between this and the U.S. EPA study
further suggests that the here demonstrated use of mega
composite samples represents a cost-effective approach for
collecting regional and nationwide information on average
concentrations of contaminants in biosolids.
water research 44 (2010) 658–668
Acknowledgements
We thank Rick Stevens and Harry B. McCarty from the U.S. EPA
for providing samples from the 2001 National Sewage Sludge
Survey. We also thank Randhir Deo and Jochen Heidler for
their valuable input. This work was supported in part by the
National Institute of Environmental Health Sciences by NIEHS
grant 1R01ES015445 and by the Johns Hopkins University
Center for a Livable Future.
Appendix.
Supplementary material
Additional details regarding experimental procedures and
results can be found in the Supplementary Material accompanying
this article. This information is available free of
charge via the Internet at www.iwaponline.com.
Supplementary data associated with this article can be
found in the online version at doi:10.1016/j.watres.2009.12.
032.
references
Bantle, J.A., Burton, D.T., Dawson, D.A., Dumont, J.N., Finch, R.A.,
Fort, D.J., Linder, G., Rayburn, J.R., Buchwalter, D.,
Gaudethull, A.M., Maurice, M.A., Turley, S.D., 1994. Fetax
interlaboratory validation study - Phase II testing.
Environmental Toxicology and Chemistry 13 (10), 1629–1637.
Brain, R.A., Johnson, D.J., Richards, S.M., Sanderson, H., Sibley, P.K.,
Solomon, K.R., 2004. Effects of 25 pharmaceutical compounds
to Lemna gibba using a seven-day static-renewal test.
Environmental Toxicology and Chemistry 23 (2), 371–382.
Brain, R.A., Ramirez, A.J., Fulton, B.A., Chambliss, C.K., Brooks, B.
W., 2008. Herbicidal Effects of Sulfamethoxazole in Lemna
gibba: Using p-Aminobenzoic Acid As a Biomarker of Effect.
Environmental Science & Technology 42 (23), 8965–8970.
Braoudaki, M., Hilton, A.C., 2004. Low level of cross-resistance
between triclosan and antibiotics in Escherichia coli K-12 and
E. coli O55 compared to E. coli O157. Fems Microbiology
Letters 235, 305–309.
Cahill, J.D., Furlong, E.T., Burkhardt, M.R., Kolpin, D., Anderson, L.G.,
2004. Determination of pharmaceutical compounds in surfaceand
ground-water samples by solid-phase extraction and highperformance
liquid chromatography-electrospray ionization
mass spectrometry. Journal of Chromatography A 1041, 171–180.
Causarano, H.J., Franzluebbers, A.J., Shaw, J.N., Reeves, D.W.,
Raper, R.L., Wood, C.W., 2008. Soil organic carbon fractions
and aggregation in the Southern Piedmont and Coastal Plain.
Soil Science Society of America Journal 72, 221–230.
Chalew, T., Halden, R., 2008. Environmental exposure of aquatic
and terrestrial biota to triclosan and triclocarban. Journal of
the American Water Resources Association 45 (1), 4–13.
Chu, S.G., Metcalfe, C.D., 2007. Simultaneous determination of
triclocarban and triclosan in municipal biosolids by liquid
chromatography tandem mass spectrometry. Journal of
Chromatography A 1164, 212–218.
Coogan, M.A., La Point, T.W., 2008. Snail bioaccumulation of
triclocarban, triclosan, and methyltriclosan in a North Texas,
USA, stream affected by wastewater treatment plant runoff.
Environmental Toxicology and Chemistry 27, 1788–1793.

Daughton, C.G., Ternes, T.A., 1999. Pharmaceuticals and personal
care products in the environment: agents of subtle change?
Environmental Health Perspectives 107, 907–938.
De Lannoy, G.J.M., Verhoest, N.E.C., Houser, P.R., Gish, T.J., Van
Meirvenne, M., 2006. Spatial and temporal characteristics of
soil moisture in an intensively monitored agricultural field
(OPE3). Journal of Hydrology 331, 719–730.
Dolliver, H., Kumar, K., Gupta, S., 2007. Sulfamethazine uptake by
plants from manure-amended soil. Journal of Environmental
Quality 36, 1224–1230.
Ericson, J.F., 2007. An evaluation of the OECD 308 water/sediment
systems for investigating the biodegradation of
pharmaceuticals. Environmental Science & Technology 41,
5803–5811.
Fent, K., Weston, A.A., Caminada, D., 2006. Ecotoxicology of
human pharmaceuticals. Aquatic Toxicology 76, 122–159.
Fick, J., Soderstrom, H., Lindberg, R.H., Phan, C., Tysklind, M.,
Larsson, D.G.J., 2009. Contamination of surface,
ground, and drinking water from pharmaceutical
production. Environmental Toxicology and Chemistry 28,
2522–2527.
Focazio, M.J., Kolpin, D.W., Barnes, K.K., Furlong, E.T., Meyer, M.T.,
Zaugg, S.D., Barber, L.B., Thurman, M.E., 2008. A national
reconnaissance for pharmaceuticals and other organic
wastewater contaminants in the United States – II untreated
drinking water sources. Science of the Total Environment 402,
201–216.
Gobel, A., Thomsen, A., McArdell, C.S., Alder, A.C., Giger, W.,
Theiss, N., Loffler, D., Ternes, T.A., 2005a. Extraction and
determination of sulfonamides, macrolides, and
trimethoprim in sewage sludge. Journal of Chromatography A
1085, 179–189.
Gobel, A., Thomsen, A., McArdell, C.S., Joss, A., Giger, W., 2005b.
Occurrence and sorption behavior of sulfonamides,
macrolides, and trimethoprim in activated sludge treatment.
Environmental Science & Technology 39, 3981–3989.
Golet, E.M., Strehler, A., Alder, A.C., Giger, W., 2002.
Determination of fluoroquinolone antibacterial agents in
sewage sludge and sludge-treated soil using accelerated
solvent extraction followed by solid-phase extraction.
Analytical Chemistry 74, 5455–5462.
Halden, R.U., Paull, D.H., 2005. Co-occurrence of triclocarban and
triclosan in US water resources. Environmental Science &
Technology 39, 1420–1426.
Halling-Sorensen, B., Nielsen, S.N., Lanzky, P.F., Ingerslev, F.,
Lutzhoft, H.C.H., Jorgensen, S.E., 1998. Occurrence, fate and
effects of pharmaceutical substances in the environment -
a review. Chemosphere 36, 357–394.
Halling-Sorensen, B., 2001. Inhibition of aerobic growth and
nitrification of bacteria in sewage sludge by antibacterial
agents. Archives of Environmental Contamination and
Toxicology 40 (4), 451–460.
Hanson, M.L., Knapp, C.W., Graham, D.W., 2006. Field assessment
of oxytetracycline exposure to the freshwater macrophytes
Egeria densa Planch and Ceratophyllum demersum L.
Environmental Pollution 141 (3), 434–442.
Heberer, T., Fuhrmann, B., Schmidt-Baumler, K., Tsipi, D.,
Koutsouba, V., Hiskia, A., 2000. Occurrence of pharmaceutical
residues in sewage, river, ground-, and drinking water in
Greece and Germany. Abstracts of Papers of the American
Chemical Society 219 U623.
Heidler, J., Halden, R.U., 2007. Mass balance assessment of
triclosan removal during conventional sewage treatment.
Chemosphere 66, 362–369.
Heidler, J., Sapkota, A., Halden, R.U., 2006. Partitioning,
persistence, and accumulation in digested sludge of the
topical antiseptic triclocarban during wastewater treatment.
Environmental Science & Technology 40, 3634–3639.
water research 44 (2010) 658–668 667
Higgins, C.P., Paesani, Z.J., Chalew, T.E.A., Halden, R.U., 2009.
Bioaccumulation of triclocarban in Lumbriculus variegates.
Environmental Toxicology and Chemistry 65, 141–148.
Hirsch, R., Ternes, T., Haberer, K., Kratz, K.L., 1999. Occurrence of
antibiotics in the aquatic environment. Science of the Total
Environment 225, 109–118.
Jones-Lepp, T.L., Stevens, R., 2007. Pharmaceuticals and personal
care products in biosolids/sewage sludge: the interface
between analytical chemistry and regulation. Analytical and
Bioanalytical Chemistry 387, 1173–1183.
Jos, A., Repetto, G., Rios, J.C., Hazen, N., Molero, M.L., del Peso, A.,
Salguero, M., Fernandez-Freire, P., Perez-Martin, J.M.,
Camen, A., 2003. Ecotoxicological evaluation of
carbamazepine using six different model systems with
eighteen endpoints. Toxicology in Vitro 17 (5–6), 525–532.
Kinney, C.A., Furlong, E.T., Kolpin, D.W., Burkhardt, M.R.,
Zaugg, S.D., Werner, S.L., Bossio, J.P., Benotti, M.J., 2008.
Bioaccumulation of pharmaceuticals and other anthropogenic
waste indicators in earthworms from agricultural soil
amended with biosolid or swine manure. Environmental
Science & Technology 42, 1863–1870.
Kolpin, D.W., Furlong, E.T., Meyer, M.T., Thurman, E.M., Zaugg, S.D.,
Barber, L.B., Buxton, H.T., 2002. Pharmaceuticals, hormones, and
other organic wastewater contaminants in US streams,
1999–2000: a national reconnaissance. Environmental Science &
Technology 36, 1202–1211.
Kumar, K., Gupta, S.C., Baidoo, S.K., Chander, Y., Rosen, C.J., 2005.
Antibiotic uptake by plants from soil fertilized with animal
manure. Journal of Environmental Quality 34, 2082–2085.
Kupper, T., Berset, J.D., Etter-Holzer, R., Furrer, R., Tarradellas, J.,
2004. Concentraions and specific loads of polycyclic musks in
sewage sludge originating from a monitoring network in
Switzerland. Chemosphere 54, 1111–1120.
Lindberg, R.H., Wennberg, P., Johansson, M.I., Tysklind, M.,
Andersson, B.A.V., 2005. Screening of human antibiotic
substances and determination of weekly mass flows in five
sewage treatment plants in Sweden. Environmental Science &
Technology 39, 3421–3429.
Lindberg, R.H., Olofsson, U., Rendahl, P., Johansson, M.I., Tysklind, M.
, Andersson, B.A.V., 2006. Behavior of fluoroquinolones and
trimethoprim during mechanical, chemical, and active sludge
treatment of sewage water and digestion of sludge.
Environmental Science & Technology 40, 1042–1048.
Lindberg, R.H., Bjorklund, K., Rendahl, P., Johansson, M.I.,
Tysklind, M., Andersson, B.A.V., 2007. Environmental risk
assessment of antibiotics in the Swedish environment with
emphasis on sewage treatment plants. Water Research 41,
613–619.
Lindsey, M.E., Meyer, M., Thurman, E.M., 2001. Analysis of trace
levels of sulfonamide and tetracycline antimicrobials, in
groundwater and surface water using solid-phase extraction
and liquid chromatography/mass spectrometry. Analytical
Chemistry 73, 4640–4646.
Liu, F., Ying, G.-G., Yang, L.-H., Zhou, Q.-X., 2009. Terrestrial
ecotoxicological effects of the antimicrobial agent triclosan.
Ecotoxicology and Environmental Safety 72, 86–92.
Loos, R., Wollgast, J., Huber, T., Hanke, G., 2007. Polar herbicides,
pharmaceutical products, perfluorooctanesulfonate (PFOS),
perfluorooctanoate (PFOA), and nonylphenol and its
carboxylates and ethoxylates in surface and tap waters
around Lake Maggiore in Northern Italy. Analytical and
Bioanalytical Chemistry 387, 1469–1478.
Loraine, G.A., Pettigrove, M.E., 2006. Seasonal variations in
concentrations of pharmaceuticals and personal care products
in drinking water and reclaimed wastewater in Southern
California. Environmental Science & Technology 40, 687–695.
Luetzhoft, H.C.H., Halling-Sorensen, B., Jorgensen, S.E., 1999.
Algal toxicity of antibacterial agents applied in Danish fish

668
farming. Archives of Environmental Contamination and
Toxicology 36 (1), 1–6.
McAvoy, D.C., Schatowitz, B., Jacob, M., Hauk, A., Eckhoff, W.S.,
2002. Measurement of triclosan in wastewater treatment
systems. Environmental Toxicology and Chemistry 21,
1323–1329.
Migliore, L., Cozzolino, S., Fiori, M., 2003. Phytotoxicity to and
uptake of enrofloxacin in crop plants. Chemosphere 52,
1233–1244.
Moldovan, Z., 2006. Occurrences of pharmaceutical and personal
care products as micropollutants in rivers from Romania.
Chemosphere 64, 1808–1817.
National Research Council, 2002. Biosolids Applied to Land:
Advancing Standards and Practices Washington, D.C.
North East Biosolids and Residuals Association, 2007. A National
Biosolids Regulation, Quality, End Use & Disposal Survey
Tamworth, NH.
Roberts, P.H., Thomas, K.V., 2006. The occurrence of selected
pharmaceuticals in wastewater effluent and surface waters of
the lower Tyne catchment. Science of the Total Environment
356, 143–153.
Robinson, A.A., Belden, J.B., Lydy, M.J., 2005. Toxicity of
fluoroquinolone antibiotics to aquatic organisms.
Environmental Toxicology and Chemistry 24 (2), 423–430.
Sapkota, A., Heidler, J., Halden, R.U., 2007. Detection of
triclocarban and two co-contaminating chlorocarbanilides in
US aquatic environments using isotope dilution liquid
chromatography tandem mass spectrometry. Environmental
Research 103, 21–29.
Shoemaker, N.B., Vlamakis, H., Hayes, K., Salyers, A.A., 2001.
Evidence for extensive resistance gene transfer among
Bacteroides spp. and among Bacteroides and other genera in
the human colon. Applied and Environmental Microbiology
67, 561–568.
Stackelberg, P.E., Furlong, E.T., Meyer, M.T., Zaugg, S.D.,
Henderson, A.K., Reissman, D.B., 2004. Persistence of
pharmaceutical compounds and other organic wastewater
contaminants in a conventional drinking-watertreatment
plant. Science of the Total Environment 329, 99–113.
Tamtam, F., Mercier, F., Le Bot, B., Eurin, J., Dinh, Q.T.,
Clement, M., Chevreuil, M., 2008. Occurrence and fate of
antibiotics in the Seine river in various hydrological
conditions. Science of the Total Environment 393, 84–95.
Ternes, T.A., 1998. Occurrence of drugs in German sewage
treatment plants and rivers. Water Research 32, 3245–3260.
Ternes, T.A., Herrmann, N., Bonerz, M., Knacker, T., Siegrist, H.,
Joss, A., 2004a. A rapid method to measure the solid-water
water research 44 (2010) 658–668
distribution coefficient (K-d) for pharmaceuticals and musk
fragrances in sewage sludge. Water Research 38, 4075–4084.
Ternes, T.A., Joss, A., Siegrist, H., 2004b. Scrutinizing
pharmaceuticals and personal care products in wastewater
treatment. Environmental Science & Technology 38,
392A–399A.
USEPA, 1994. Biosolids Recycling: Beneficial Technologies for
a Better Environment. Office of Water. U.S. Environmental
Protection Agency, Washington, D.C. EPA 832-R-94–009.
USEPA, 2001. Sampling Procedures for the 2001 National Sewage
Sludge Survey. Office of Science and Technology, Washington,
D.C.
USEPA, 2003. Workshop on Emerging Pollutants. Region/ORD
Workshop on Emerging Pollutants. U.S. EPA, Chicago, IL.
USEPA, 2007a. 2001 National Sewage Sludge Survey Report
Washington, D.C. EPA-822-R-07-006.
USEPA, 2007b. Method 1694: Pharmaceuticals and Personal Care
Products in Water, Soil, Sediment, and Biosolids by HPLC/MS/
MS Washington, D.C. EPA-821-R-08–002.
USEPA, 2009. Targeted National Sewage Sludge Survey
Washington, D.C. EPA-822-R-08–014.
Williams, R.R., Bell, T.A., Lightner, D.V., 1992. Shrimp
antimicrobial testing: II. Toxicity testing and safety
determination for twelve antimicrobials with penaeid shrimp
larvae. Journal of Aquatic Animal Health 4 (4), 262–270.
Wilson, B.A., Smith, V.H., Denoyelles, F., Larive, C.K., 2003. Effects
of three pharmaceutical and personal care products on
natural freshwater algal assemblages. Environmental Science
& Technology 37 (9), 1713–1719.
Wu, C.X., Witter, J.D., Spongberg, A.L., Czajkowski, K.P., 2009.
Occurrence of selected pharmaceuticals in an agricultural
landscape, western Lake Erie basin. Water Research 43,
3407–3416.
Yang, L.H., Ying, G.G., Su, H.C., Stauber, J.L., Adams, M.S.,
Binet, M.T., 2008. Growth-inhibiting effects of 12 antibacterial
agents and their mixtures on the freshwater microalga
Pseudokirchneriella subcapitata. Environmental Toxicology and
Chemistry 27 (5), 1201–1208.
Ying, G.G., Yu, X.Y., Kookana, R.S., 2007. Biological
degradationoftriclocarbanandtriclosaninasoilunder
aerobic and anaerobic conditions and comparison with
environmental fate modelling. Environmental Pollution
150, 300–305.
Zurita, J.L., Repetto, G., Jos, A., Salguero, M., Lopez-Artiguez, M.,
Camean, A.M., 2007. Toxicological effects of the lipid regulator
gemfibrozil in four aquatic systems. Aquatic Toxicology 81 (1),
106–115.