Subsurface Iron and Arsenic Removal
qj78kp8
qj78kp8
- No tags were found...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
Doris van Halem<br />
<strong>Subsurface</strong> <strong>Iron</strong> <strong>and</strong><br />
<strong>Arsenic</strong> <strong>Removal</strong><br />
for drinking water treatment in Bangladesh
<strong>Subsurface</strong> <strong>Iron</strong> <strong>and</strong> <strong>Arsenic</strong> <strong>Removal</strong><br />
for drinking water treatment in Bangladesh<br />
Proefschrift<br />
ter verkrijging van de graad van doctor<br />
aan de Technische Universiteit Delft,<br />
op gezag van de Rector Magnificus prof. ir. K.C.A.M. Luyben,<br />
voorzitter van het College van Promoties<br />
in het openbaar te verdedigen op ma<strong>and</strong>ag 3 oktober 2011 om 15:00 uur<br />
door<br />
Doris VAN HALEM<br />
civiel ingenieur<br />
geboren te Eindhoven
Dit proefschrift is goedgekeurd door de promotoren:<br />
Prof. ir. J.C. van Dijk<br />
Prof. dr. G.L. Amy<br />
Samenstelling promotiecommissie:<br />
Rector Magnificus<br />
voorzitter<br />
Prof. ir. J.C. van Dijk<br />
Technische Universiteit Delft, promotor<br />
Prof. dr. G.L. Amy<br />
UNESCO-IHE/Technische Universiteit Delft, promotor<br />
Dr. ir. J.Q.J.C. Verberk<br />
Technische Universiteit Delft<br />
Prof. dr. K.M.U. Ahmed University of Dhaka<br />
Prof. dr. ing- U. Rott<br />
Technische Universität Stuttgart<br />
Prof. dr. ir. H.H.G. Savenije Technische Universiteit Delft<br />
Prof. dr. P.J. Stuyfz<strong>and</strong><br />
Vrije Universiteit Amsterdam<br />
Prof. dr. ir. W.G.J. van der Meer Technische Universiteit Delft, reservelid<br />
ISBN: 978-90-8957-022-2<br />
Published by Water Management Academic Press<br />
Copyright © 2011 by D. van Halem<br />
All rights reserved. No part of the material protected by the copyright may be reproduced or utilized in any form or by any means,<br />
electronic or mechanical, including photocopying, recording or by any information storage <strong>and</strong> retrieval system, without written<br />
permission from the copywright owner.<br />
Cover design: S. Willems (www. sjorswillems.nl)<br />
Printed in the Netherl<strong>and</strong>s
Acknowledgements<br />
Acknowledgements<br />
The desire to make a difference has been the driving<br />
force to start my PhD research, but along the road I<br />
have also increasingly enjoyed the scientific aspects of<br />
it. This dissertation could not have been written without<br />
the people around me, so thank you all for making my<br />
PhD period anything but lonely.<br />
First of all I thank my two promotors, Prof. Hans<br />
van Dijk <strong>and</strong> Prof. Gary Amy, who have given me the<br />
opportunity to take this new research direction. Hans,<br />
your enthusiasm <strong>and</strong> confidence have encouraged me<br />
to approach this research in my own way. I enjoyed<br />
working with you, especially because you know how to<br />
transfer your knowledge (<strong>and</strong> gezond boerenverst<strong>and</strong>)<br />
without dominating the discussion. Gary, thank you<br />
for sharing your broad expertise with me, <strong>and</strong> I am<br />
glad that I also got to know you as a kind <strong>and</strong> caring<br />
person. With two promotors operating from a distance<br />
in the last year(s), either retired or working in Jeddah,<br />
it was essential to have a daily supervisor. Jasper, thank<br />
you for all the guidance <strong>and</strong> laughs, you are a valuable<br />
colleague.<br />
My time in Bangladesh was a success thanks<br />
to the great support of Rick Johnston (Unicef), his<br />
know-how <strong>and</strong> generosity helped me get started in this<br />
beautiful country. At the Department of Public Health<br />
Engineering I want to thank Ihtishamul Huq, Sudhir<br />
Ghosh <strong>and</strong> Shah Alam for tackling loads of political <strong>and</strong><br />
practical hinders, <strong>and</strong> for assisting in the monitoring<br />
program. Dhonnobad.<br />
The field research in the Netherl<strong>and</strong>s would not<br />
have been possible without the support <strong>and</strong> hospitality<br />
of Oasen Drinking Water Company <strong>and</strong> Vitens<br />
Drinking Water Supply at WTPs Lekkerkerk, De Put<br />
<strong>and</strong> Loosdrecht. I am also thankful for the funding of<br />
iii
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
AgentschapNL in their InnoWator program.<br />
During the past four years I was lucky to<br />
supervise many (international) students, whose results<br />
have contributed to several of my published papers. I<br />
especially want to thank Samuele, David <strong>and</strong> Harmen<br />
who have battled the O 2<br />
<strong>and</strong> challenged me with<br />
surprising questions.<br />
At TU Delft, I want to thank my colleagues<br />
for accompanying me on this PhD journey, which<br />
took us from coffee breaks, past tropical beaches <strong>and</strong><br />
inspiring discussions (on water bugs <strong>and</strong> hydrophobia),<br />
to running the 10km. With special thanks to Mieke,<br />
Weren, David, Bas, Luuk, Arne, Sheng, Anke, Petra,<br />
Tonny, Martine <strong>and</strong> Stefan. I wish S<strong>and</strong>ra, Moshiur,<br />
Zahid <strong>and</strong> Debasish all the best in pursuing their PhD<br />
within the newly started NWO WOTRO project on<br />
<strong>Subsurface</strong> <strong>Arsenic</strong> <strong>Removal</strong>. I also want to thank my<br />
colleagues at UNESCO-IHE, who have made the start of<br />
my PhD more enjoyable. Addionally, I want to mention<br />
the TU Delft <strong>and</strong> UNESCO-IHE laboratory staffs who<br />
have provided great assistance from analysis with GF-<br />
AAS to designing (oxygen-free) column installations.<br />
Apart from my direct colleagues, it has been<br />
very inspiring to discuss my findings with so many<br />
(international) scientists at conferences <strong>and</strong> meetings,<br />
<strong>and</strong> during peer-review processes. A PhD path is not<br />
something you walk by yourself in solitude, because<br />
it is the interaction with the world that makes you<br />
reconsider your route.<br />
Naturally it is impossible to enjoy four years of<br />
research without the occasional distraction of friends<br />
<strong>and</strong> family. I want to thank you all for asking me about<br />
my research, <strong>and</strong> maybe even more for not asking.<br />
In the final months of my PhD period it was great<br />
to have the support of my two paranimphs, Geerte <strong>and</strong><br />
Harmen, who share my enthusiasm for research <strong>and</strong><br />
helped me with the last hurdles.<br />
I realize that it is because of my loving parents<br />
that I approach life with an open <strong>and</strong> positive mind.<br />
Pap en mam, thanks for supporting me, your believe in<br />
me gives me wings.<br />
Most of all I want to thank Wieger, for always<br />
keeping me close whether I am in Delft or on the other<br />
side of the world. I do not know why we are so lucky,<br />
but life is amazing with you. And Ferre, you are my<br />
little explorer, curious <strong>and</strong> brave, the best example a<br />
scientist can get.<br />
Een rivier dankt haar naam<br />
aan de oevers, want het water<br />
stroomt anoniem voorbij.<br />
T. van Lieshout,1987<br />
iv
Table of contents<br />
Table of contents<br />
Acknowledgements iii<br />
1 Introduction 1<br />
2 <strong>Subsurface</strong> iron removal in the Netherl<strong>and</strong>s 23<br />
3 Small-scale subsurface iron <strong>and</strong> arsenic removal in Bangladesh 37<br />
4 Simulation of adsorptive-catalytic oxidation in s<strong>and</strong> columns 51<br />
5 Cation exchange during subsurface iron removal 67<br />
6 Characterization of accumulated deposits 81<br />
7 Catalysis by accumulated deposits 99<br />
8 Influence of groundwater composition 111<br />
9 Concluding remarks 125<br />
Summary 135<br />
Samenvatting 139<br />
List of publications 145<br />
Biography 148<br />
v
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
vi
1<br />
Introduction<br />
Parts of this chapter are based on:<br />
van Halem et al. (2009) Drinking Water Engineering <strong>and</strong> Science 2: 29-34<br />
van Halem et al. (2010) Desalination 248: 241-248
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
1 An introduction to the arsenic<br />
problem<br />
A worldwide problem<br />
The World Health Organization estimated in 2001 that<br />
about 130 million people worldwide are exposed to<br />
arsenic concentrations above 50 μg.L -1 (WHO, 2001).<br />
Affected countries include Bangladesh (>30 million<br />
exposed people), India (40 million), China (1.5 million)<br />
<strong>and</strong> the United States (2.5 million). The problem of<br />
arsenic-contaminated source waters is, however, not<br />
confined to these countries, as illustrated in Figure<br />
1.1. According to the United Nations Synthesis report,<br />
arsenic poisoning is the second most important health<br />
hazard related to drinking water (Johnston et al., 2001).<br />
Only contamination by pathogenic microorganisms has<br />
a bigger impact worldwide. <strong>Arsenic</strong> contamination of<br />
groundwater (Box 1.1) has been found to occur due to<br />
(Smedley <strong>and</strong> Kinniburgh, 2002):<br />
• geothermally- influenced groundwater;<br />
• mineral dissolution (e.g., pyrite oxidation);<br />
• desorption in the oxidising environment, <strong>and</strong>;<br />
• reductive desorption <strong>and</strong> dissolution.<br />
Reductive dissolution of young arsenic-bearing<br />
sediments is the cause of the large-scale arsenic<br />
contamination of the strongly reducing aquifers in the<br />
Bengal Delta. Also in China the reducing conditions<br />
in the subsurface are the cause of arsenic mobilization.<br />
Figure 1.1 <strong>Arsenic</strong>-affected countries of the world (Smedley <strong>and</strong> Kinniburgh, 2002; Appleyard et al., 2006;<br />
Petrusevski et al., 2007; Smedley et al., 2007; Gunduz et al., 2009; Jovanovic et al., 2011)<br />
2
1 Introduction<br />
Box 1.1<br />
<strong>Arsenic</strong> speciation, including an Eh-pH diagram for aqueous arsenic species in the system As-O 2<br />
-H 2<br />
O at 25°C <strong>and</strong> 1 bar total<br />
pressure (Smedley <strong>and</strong> Kinniburgh, 2002)<br />
1<br />
<strong>Arsenic</strong> is stable in four oxidation states (+5, +3, 0, -3)<br />
under redox conditions occurring in aquatic systems<br />
(Furguson <strong>and</strong> Gavis, 1972). However, in the aqueous<br />
environment, arsenic occurs mainly as two species; trivalent<br />
arsenic, arsenite or As(III) <strong>and</strong> pentavalent arsenic, arsenate<br />
or As(V). Under oxidizing <strong>and</strong> aerated conditions, the<br />
predominant form of arsenic in water is As(V) (as HAsO 4<br />
2-<br />
or H 2<br />
AsO 4<br />
-<br />
around neutral pH), whereas if reducing<br />
conditions prevail, As(III) (around neutral pH as H 3<br />
AsO 3<br />
) is<br />
the predominant arsenic compound (WHO, 2001). As(III)<br />
is thermodynamically unstable in oxic environments <strong>and</strong><br />
oxidizes to As(V); however studies have shown that this<br />
reaction proceeds slowly when oxygen is the only oxidant<br />
(AWWARF, 2000; Kim et al., 2000; Stollenwerk, 2003).<br />
Therefore both As species can be found in natural systems.<br />
The speciation of arsenic is controlled by pH within a<br />
particular oxidation state, while the redox potential controls<br />
the distribution of arsenic species between the two oxidation<br />
states (Masscheleyn et al., 1999; Bose <strong>and</strong> Sharma, 2002).<br />
Concentrations up to 1,800 μg.L -1 have been measured in<br />
Inner Mongolia, a northern province of China (Smedley<br />
et al., 2003). In Vietnam <strong>and</strong> Cambodia, arsenic<br />
concentrations were also observed to be high (up to 1,340<br />
μg.L -1 ) due to reductive dissolution of young sediments<br />
(Buschmann et al., 2007; Buschmann et al., 2008).<br />
<strong>Arsenic</strong> mobilization caused by mineral dissolution<br />
has been found in active volcanic areas of Italy (Aiuppa<br />
et al., 2003) <strong>and</strong> inactive volcanic regions in Mexico<br />
(Armienta <strong>and</strong> Segovia, 2008). Volcanism in the Andes<br />
has lead to arsenic contamination of groundwater in<br />
Chile <strong>and</strong> Argentina (Smedley <strong>and</strong> Kinniburgh, 2002).<br />
Also mining activities have been found to contribute to<br />
arsenic contamination in Latin American groundwater<br />
(Smedley <strong>and</strong> Kinniburgh, 2002). Mining activities may<br />
cause the oxidation of arsenic-bearing sulphide minerals<br />
resulting in the release of arsenic into groundwater.<br />
Smedley <strong>and</strong> Kinniburgh (2002) listed cases of arsenic<br />
contamination caused by mining activities in Canada,<br />
Germany, Ghana, Greece, Mexico, South Africa,<br />
Thail<strong>and</strong>, UK, USA <strong>and</strong> Zimbabwe. In the past years,<br />
more <strong>and</strong> more countries have found their waters to<br />
3
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
1 be affected by arsenic contamination due to mining<br />
wastes, e.g., Pol<strong>and</strong>, Korea <strong>and</strong> Brazil (Marszałek <strong>and</strong><br />
Wasik, 2000; Woo <strong>and</strong> Choi, 2001; Borba et al., 2003).<br />
More recently, groundwater in Burkino Faso was found<br />
to be contaminated by arsenic with concentrations up<br />
to 1,630 μg.L -1 , caused by mining activities (Smedley et<br />
al., 2007). Furthermore, Gunduz et al. (2009) reported<br />
elevated arsenic levels ( As(V) > monomethylarsonate<br />
(MMA) > dimethylarsinate (DMA) (Jain <strong>and</strong> Ali,<br />
2000). In areas with elevated arsenic concentrations in<br />
the environment, the exposure is not solely confined<br />
to drinking water. <strong>Arsenic</strong> (organic <strong>and</strong> inorganic) is<br />
also found in a wide range of food products, like fish,<br />
4
1 Introduction<br />
meat <strong>and</strong> rice (WHO, 2001b; Williams et al., 2006;<br />
Sambu <strong>and</strong> Wilson, 2008). Intake through air may also<br />
be significant, especially close to industrial sources<br />
(WHO, 2001b).<br />
The cancer risk at low-to-moderate exposure<br />
concentrations in drinking water is still under debate<br />
(Smith et al., 2002; Celik et al., 2008). Most risk<br />
estimations use data from Taiwanese studies (Tseng et al.,<br />
1968; Chen et al., 1992), since limited epidemiological<br />
information is available from elsewhere in the world.<br />
In Bangladesh, results indicated at least a doubling of<br />
lifetime mortality risk from liver, bladder, <strong>and</strong> lung<br />
cancers (229.6 vs 103.5 per 100,000 population) owing<br />
to arsenic in drinking water (Chen <strong>and</strong> Ahsan, 2004).<br />
This has an enormous impact since it is estimated that<br />
of the 140 million inhabitants of Bangladesh, over 37<br />
million are at risk of arsenic poisoning (WHO, 2001;<br />
Chowdhury et al., 2006). The cancer risks from arsenic<br />
in drinking water were assessed by Smith <strong>and</strong> coworkers<br />
(1992) for the USA situation. At that time the<br />
national guideline in the USA was 50 μg.L -1 <strong>and</strong> they<br />
concluded that the lifetime risks of dying from cancer<br />
due to arsenic in drinking water was 21 in 1,000 adults.<br />
For a concentration of 2.5 μg.L -1 the risk would still be<br />
1 in 1,000 adults, which they found comparable to the<br />
lifetime cancer risk of passive smoking. More recently,<br />
based on male bladder cancer with an excess risk of<br />
1 in 10,000 for 75-year lifetime exposure, the arsenic<br />
guideline is recommended to be 3.4 μg.L -1 (Liao et al.,<br />
2009).<br />
The World Health Organization has published an<br />
overview document on the toxicology <strong>and</strong> legislation<br />
for arsenic in drinking water (WHO, 2003). They<br />
conclude that the maximum likelihood “for bladder<br />
<strong>and</strong> lung cancer for US populations exposed to 10 μg<br />
of arsenic per litre in drinking water are, respectively,<br />
12 <strong>and</strong> 18 per 10,000 population for females <strong>and</strong> 23<br />
<strong>and</strong> 14 per 10,000 population for males”. The WHO<br />
has a general rule that no substance may have a higher<br />
lifetime risk of more than 1 in 100,000. Purely based on<br />
health effects, the WHO guideline of 10 μg.L -1 would, in<br />
that respect, not suffice. The main reason to maintain<br />
this guideline is, therefore, merely practical from<br />
economic <strong>and</strong> engineering perspective <strong>and</strong> not health<br />
related. The US Environmental Agency (EPA, 1998) <strong>and</strong><br />
the US Natural Resources Defense Council (NRDC,<br />
2000) even recommend arsenic guidelines below 1<br />
μg.L -1 to attain an acceptable lifetime cancer risk. It is<br />
noteworthy that EPA considered life-time skin cancer<br />
risk only <strong>and</strong> did not include arsenic intake through<br />
food due to a lack of reliable data. The consumption<br />
of arsenic through food could overestimate the<br />
current risk calculations <strong>and</strong> EPA indicates a possible<br />
uncertainty of one order of magnitude. The WHO <strong>and</strong><br />
EPA do not provide information on whether inorganic<br />
arsenic is genotoxic or non-genotoxic, because current<br />
epidemiologic studies are inadequate for that. In their<br />
background documents, these organizations describe<br />
the cancer risks based on both approaches. Although<br />
the uncertainties concerning the health risks due to<br />
5<br />
1
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
1 arsenic in drinking water are undeniable, it is clear that<br />
extremely low concentrations, or its complete absence,<br />
are desirable to avoid these potential risks.<br />
Household water treatment <strong>and</strong><br />
safe storage<br />
Household water treatment <strong>and</strong> storage (HWTS) has<br />
been recognized to be an effective measure in reducing<br />
diarrheal diseases (Sobsey, 2002; Fewtrell et al., 2005;<br />
Clasen, 2008). HWTS treats the water in homes to<br />
remove microbial <strong>and</strong>/or chemical contaminants. The<br />
targeted microbial contaminants generally include<br />
helminthes, protozoa, bacteria <strong>and</strong> viruses, in order to<br />
reduce diarrhoea cases <strong>and</strong> diseases such as bilharzias<br />
<strong>and</strong> cholera. HWTS is a so-called point-of-use water<br />
treatment, greatly reduceing the risk of recontamination<br />
of the treated water during transport (Sobsey, 2002).<br />
HWTS systems to remove microorganisms include<br />
technologies based on chlorination, coagulationfiltration,<br />
solar disinfection, ceramic <strong>and</strong> (bio)s<strong>and</strong><br />
filtration (Sommer et al., 1997; Stauber et al., 2006;<br />
van Halem et al., 2007). Although HWTS is widely<br />
promoted as an appropriate intervention (UNICEF,<br />
2009; WHO, 2011), studies have shown that not<br />
all systems are as effective in reducing pathogenic<br />
microorganisms. Hunter (2009) predicted that over 12<br />
months, ceramic filters were likely to be still effective<br />
at reducing disease, whereas solar disinfection,<br />
chlorination, <strong>and</strong> coagulation-chlorination had little if<br />
any long-term public health benefit. Additionally, the<br />
Bios<strong>and</strong> filter, alongside the ceramic filter, was found to<br />
have the greatest potential to become widely used <strong>and</strong><br />
to be most effective in reducing waterborne disease <strong>and</strong><br />
death (Sobsey et al., 2008). Such findings point out that<br />
not all HWTS interventions are sustainable, illustrating<br />
the need for knowledge on technology development,<br />
adoptation <strong>and</strong> socio-economic dynamics.<br />
In the past decade, HWTS solutions have<br />
also been developed for the removal of arsenic; most<br />
of these technologies have translated conventional<br />
treatment methods to a household solution. When<br />
considering arsenic contamination in (anoxic or<br />
anaerobic) groundwater, the dominant arsenic species is<br />
generally As(III). Around neutral pH As(III) (H 3<br />
AsO 3<br />
)<br />
is uncharged <strong>and</strong> therefore difficult to remove with<br />
processes that rely on surface charge (ion exchange, iron<br />
hydroxide adsorption). As(V) (HAsO 4<br />
2-<br />
or H 2<br />
AsO 4-<br />
)<br />
can be more easily removed because it is negatively<br />
charged <strong>and</strong> is, therefore, relatively easily incorporated<br />
into the iron oxide matrix during iron removal. In<br />
conventional groundwater treatment usually preoxidation<br />
to As(V) is required to remove As(III) from<br />
the water. Hypochlorite <strong>and</strong> permanganate are common<br />
oxidants to catalyze the reaction of As(III) oxidation. A<br />
lot of attention has been given to the development of<br />
technologies based on the co-precipitation of arsenic<br />
in flocs during coagulation (e.g., by dosing of ferric<br />
6
1 Introduction<br />
sulfate or ferric chloride) <strong>and</strong> arsenic adsorption<br />
to media, like activated alumina <strong>and</strong> granular iron<br />
oxide/oxide (GFH; Banerjee et al., 2008). In all cases<br />
the arsenic binds to the positively-charged surface<br />
of the (iron hydroxide) matrix. Especially at low to<br />
moderate arsenic concentrations, the technology<br />
of arsenic adsorption is relatively effective. Other<br />
arsenic removal technologies include ion exchange,<br />
coagulation/filtration <strong>and</strong> reverse osmosis (AWWARF,<br />
2000; EPA, 2007; Johnston et al., 2001). Most processes<br />
are, however, not designed for household application<br />
in a developing country context. The sustainability<br />
of a HWTS can be addressed based on five criteria<br />
(van Halem et al., 2009): accessibility, water quality,<br />
water quantity, functionality, <strong>and</strong> environmental<br />
footprint. Accessibility entails both the affordability<br />
<strong>and</strong> availability of the technology <strong>and</strong> should be<br />
expressed in money, time or distance. A product can be<br />
accessible through a local shop or organization, but in a<br />
globalizing world, also through an ordering system on<br />
the internet. The criteria of water quality <strong>and</strong> quantity<br />
are formulated by the World Health Organization<br />
(WHO, 2006; WHO, 2007), being 2L per person per day. Functionality depends on<br />
the operation <strong>and</strong> maintenance activities for the users,<br />
which strongly determines the social acceptance of a<br />
technology. The environmental footprint is generally<br />
not the main concern of most users, but long-term<br />
1<br />
brick<br />
chips<br />
coarse s<strong>and</strong><br />
cast iron<br />
coarse s<strong>and</strong><br />
coagulation <strong>and</strong><br />
flocculation bucket<br />
cloth filter<br />
scooping net<br />
coarse s<strong>and</strong><br />
charcoal<br />
fine s<strong>and</strong><br />
brick<br />
chips<br />
media<br />
media<br />
perforated plate<br />
covered with<br />
cotton filter<br />
s<strong>and</strong><br />
cloth filter<br />
s<strong>and</strong><br />
EVOH resin<br />
geo textile<br />
synthetic cloth<br />
scooping net<br />
scooping net<br />
charcoal<br />
Figure 1.2 <strong>Arsenic</strong> household water treatment <strong>and</strong> storage (from left to right): Sono Filter, Alcan Filter, Shawdesh Filter <strong>and</strong> Read-F<br />
(BETV-SAM, 2011)<br />
7
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
1 issues regarding chemical/electricity consumption <strong>and</strong><br />
production of a waste stream may seriously affect a<br />
rural community.<br />
The majority of the adsorption-based HWTS<br />
technologies targeting arsenic removal, rely on iron<br />
oxides, such as cast iron (Sono Filter, Figure 1.2a),<br />
iron coated bricks/s<strong>and</strong> (Shapla Filter/IHE Filter)<br />
<strong>and</strong> iron nails (Kanchan Filter), or activated alumina<br />
(Alcan Filter, Figure 1.2b). In order to remove iron<br />
from the groundwater these filters use a bucket-style<br />
s<strong>and</strong> filter. Also frequently combined with a s<strong>and</strong><br />
filter are oxidation <strong>and</strong> coagulation-flocculation<br />
processes by dosing chemicals such as permanganate,<br />
sodium hypochlorite, iron or aluminum sulphate<br />
(DPHE-Danida Bucket Treatment Unit; 2-Kolshi<br />
Filter; Shapla Filter; Shawdesh Filter, Figure 1.2c) or<br />
atmospheric oxygen (Asia <strong>Arsenic</strong> Network Filter). Ion<br />
exchange resins have also been utilized for household<br />
arsenic removal (Read-F, Figure 1.2d; Tetrahedron),<br />
in combination with a cloth <strong>and</strong>/or s<strong>and</strong> filter. Most<br />
arsenic HWTS solutions consist of buckets on a tripod<br />
<strong>and</strong> appear to be similar, but reported efficacies vary<br />
widely between 90% (Sutherl<strong>and</strong>, 2002;<br />
IGRAC, 2007; BETV-SAM, 2011).<br />
Many factors influence removal efficacies (e.g.,<br />
water quality, consumer operation <strong>and</strong> maintenance),<br />
but before the implementation of HWTS one must<br />
have solid <strong>and</strong> statistically valid results showing<br />
that it is a safe technology. The post-deployment<br />
monitoring is generally poor or absent, even further<br />
dem<strong>and</strong>ing for a safe <strong>and</strong> robust verification of the<br />
newly developed technologies. The Bangladesh<br />
Environmental Technology Verification – Support to<br />
<strong>Arsenic</strong> Mitigation (BETV-SAM) program has tested<br />
15 arsenic removal technologies <strong>and</strong> has approved only<br />
6 of them for sale. The majority of arsenic removal units<br />
failed to perform according to the set guidelines (www.<br />
verification-unit.org). Nevertheless, there are arsenic<br />
HWTS solutions that show potential, although further<br />
research <strong>and</strong> documentation is needed to determine<br />
the long-term effectiveness. Additionally, the treatment<br />
of water in the homes requires capacity building in<br />
these rural communities. The users need to be aware<br />
of the operation <strong>and</strong> maintenance obligations. because<br />
all filters will at one stage show breakthrough of arsenic<br />
<strong>and</strong> many leave a waste stream that should be disposed<br />
of as hazardous waste.<br />
<strong>Subsurface</strong> <strong>Iron</strong> <strong>and</strong> <strong>Arsenic</strong><br />
<strong>Removal</strong><br />
A new HWTS approach to remove arsenic from<br />
groundwater is by retention in the subsurface,<br />
preventing the generation of a waste stream (Sarkar<br />
<strong>and</strong> Rahman, 2001; Rott et al., 2002). Immobilizing the<br />
arsenic in the aquifer provides the additional benefit<br />
that existing infrastructure, namely the (shallow) tube<br />
well h<strong>and</strong> pump, can be utilized for treatment. In that<br />
8
1 Introduction<br />
case there is no longer the need for above-ground water<br />
treatment. This novel technology, <strong>Subsurface</strong> <strong>Arsenic</strong><br />
<strong>Removal</strong> (SAR), is based on the existing technology of<br />
<strong>Subsurface</strong> <strong>Iron</strong> <strong>Removal</strong> (SIR), or in-situ iron removal.<br />
<strong>Subsurface</strong> <strong>Iron</strong> <strong>Removal</strong> (SIR)<br />
With a patent originating from 1900 (von Oesten,<br />
1900), subsurface iron removal was first introduced<br />
in Sweden in 1971 (Mettler, 2002), but soon other<br />
European countries followed with subsurface iron<br />
removal plants, including Switzerl<strong>and</strong>, Germany,<br />
Denmark, France <strong>and</strong> the Netherl<strong>and</strong>s (van Beek,<br />
1985; Mettler, 2002). The principle of subsurface iron<br />
removal is that aerated water is periodically injected<br />
into an anoxic or anaerobic aquifer through a tube<br />
well, partially displacing the original Fe 2+ -containing<br />
groundwater. The O 2<br />
-rich injection water oxidizes the<br />
Fe 2+ in the subsurface environment around the tube<br />
well. When the flow is reversed, groundwater with<br />
low Fe concentrations is abstracted. More water with<br />
reduced iron concentrations can be abstracted (volume<br />
V) than was injected (volume Vi), i.e., this volumetric<br />
ratio (V/Vi) determines the efficiency of the system.<br />
When O 2<br />
-rich water is injected into the aquifer, the<br />
adsorbed Fe 2+ will oxidize heterogeneously:<br />
Equation 1.1<br />
S − OFe( II) + 0.25O + 1.5 H O →S − OFe( III)( OH ) + H<br />
+ 0 +<br />
2 2 2<br />
With S being the solid iron oxide surface. The freshly<br />
formed Fe 3+ iron hydroxides provide new adsorption<br />
sites, for soluble Fe 2+ during the abstraction phase. The<br />
adsorption of Fe 2+ occurs in the absence of oxygen <strong>and</strong><br />
can be schematically formulated as:<br />
Equation 1.2<br />
o 2+ + +<br />
S − OH + Fe →S − OFe( II ) + H<br />
It should be noted that the use of S-OH 0 in the equation<br />
is simplified, as in reality adsorbed Fe 2+ may transfer<br />
an electron to the solid (Hiemstra <strong>and</strong> van Riemsdijk,<br />
2007). The adsorptive capacity of the soil depends on<br />
the type of iron oxides present in the soil, as adsorption<br />
capacities for the amorphous ferrihydrite are larger<br />
than for crystalline mineral structures with lower<br />
surface areas (goethite, lepidocrocite). The presence<br />
of Fe 3+ oxides are known to catalyze the Fe 2+ oxidation<br />
reaction (Tamura et al., 1980; Stumm <strong>and</strong> Morgan,<br />
1996), making this mode of iron immobilization very<br />
attractive. Besides this theory of adsorptive-catalytic<br />
oxidation (Rott, 1985; van Beek, 1985), it has also been<br />
proposed that the injection of O 2<br />
-rich water onsets the<br />
exchange of adsorbed Fe 2+ with other cations, such as<br />
calcium (Appelo et al., 1999; Appelo <strong>and</strong> Postma, 2005),<br />
that Fe 2+ is effectively incorporated into the mineral<br />
structure through interfacial electron transfer (Mettler,<br />
2002), or that iron oxidizing bacteria (IOB) enhance<br />
Fe 2+ removal (Hallberg <strong>and</strong> Martinell, 1976).<br />
It has been widely reported that a major benefit<br />
of subsurface iron removal is that the technology<br />
increases in efficacy with every injection-abstraction<br />
cycle (Boochs <strong>and</strong> Barovic, 1981; Braester <strong>and</strong> Martinell,<br />
9<br />
1
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
1<br />
Figure 1.3 Light microscope image (top left) of a region<br />
containing roots with large As/Fe ratio, along with<br />
the distribution of As, Fe <strong>and</strong> Mn (μ-XRF). Obtained<br />
with permission from Frommer et al. (2011)<br />
1988; Grombach, 1985; Hallberg <strong>and</strong> Martinell, 1976;<br />
Mettler, 2002; Rott, 1985; van Beek, 1985), providing<br />
it with a great advantage over the above-ground<br />
alternative, which is sensitive to clogging. Clogging of<br />
the aquifer by iron sludge has not been reported as a<br />
limitation of the subsurface iron removal technology.<br />
The deposited iron precipitates have been characterized<br />
as compact, crystalline mineral structures (Mettler at<br />
al, 2001). Additionally it has been proposed that iron<br />
precipitates at various distances from the well, resulting<br />
in an even spreading of the precipitates around the tube<br />
well (Boochs <strong>and</strong> Barovic, 1981; Appelo et al., 1999).<br />
In the past decades several well designs have<br />
been implemented, including the single-well pushpull<br />
design, the dual well design <strong>and</strong> the Vyredox TM<br />
method (Hallberg <strong>and</strong> Martinell, 1976). The Vyredox TM<br />
method consists of a circle of injection wells around<br />
one abstraction well, providing a subsurface oxidation<br />
screen for Fe 2+ adsorption. In the dual well design, one<br />
well is used for production while the other is being<br />
used for injection <strong>and</strong> vice versa. The single-well design<br />
can be found most attractive for the shallow tube well<br />
HWTS, as a normal production well can be relatively<br />
easily transitioned into a subsurface iron removal well.<br />
Apart from iron removal, the SIR technology has also<br />
been applied for the removal of manganese <strong>and</strong> arsenic,<br />
with varying results (van Beek, 1985; Rott et al., 2002).<br />
Manganese precipitates are generally found closer to<br />
the well, as oxidation kinetics are slower than for iron<br />
(Hallberg <strong>and</strong> Martinell, 1976; van Beek, 1985), whereas<br />
arsenic is mostly bound to the iron oxides (Dixit <strong>and</strong><br />
Hering, 2003). The spatial separation of Fe, Mn <strong>and</strong> As<br />
deposits is beautifully visualized by a study of the O 2<br />
<strong>and</strong> groundwater interaction around a rice plant root<br />
in a Bangladeshi paddy field (Figure 1.3; Frommer et<br />
al., 2011). In the Netherl<strong>and</strong>s subsurface iron removal<br />
is even operated at some locations to enhance the<br />
nitrification process in subsequent bios<strong>and</strong> filters (de<br />
Vet et al., 2009), illustrating the beneficiary side effects<br />
of SIR.<br />
<strong>Subsurface</strong> <strong>Arsenic</strong> <strong>Removal</strong> (SAR)<br />
Although the removal of arsenic during subsurface<br />
iron removal can be considered a side effect, it may<br />
10
1 Introduction<br />
also be approached as the main objective at sites highly<br />
contaminated with arsenic. In that case we consider<br />
iron removal to be the beneficiary side effect <strong>and</strong><br />
<strong>Subsurface</strong> <strong>Arsenic</strong> <strong>Removal</strong> (SAR) the technology.<br />
Either way, arsenic removal depends greatly on the iron<br />
removal processes, as precipitated iron oxides form the<br />
adsorptive surfaces for the arsenic (Dixit <strong>and</strong> Hering,<br />
2003):<br />
Equation 1.3<br />
Equations 1.4a <strong>and</strong> 1.4b<br />
As(III) adsorption onto iron oxides is relatively<br />
insensitive to pH as found in the aquatic environment,<br />
though slightly favouring the pH range of 5 to 8 (Dixit<br />
<strong>and</strong> Hering, 2003). In natural waters, the adsorption<br />
(a) injection phase<br />
S − OH + H3AsO3 →S − H<br />
2AsO3 + H<br />
2O<br />
−<br />
S − OH + H<br />
2<br />
AsO4<br />
→ S − HAsO4<br />
+ H<br />
2O<br />
2−<br />
S − OH + HAsO4 → S − AsO4<br />
+ H<br />
2O<br />
−<br />
2−<br />
of the negatively charged As(V) is better at higher<br />
pH, because the point of zero charge of iron oxides is<br />
reportedly between 7 <strong>and</strong> 9 (Sposito, 1989; van Geen<br />
et al., 1994; Peacock <strong>and</strong> Sherman 2004). The type<br />
of iron oxide mineral structure also determines the<br />
arsenic adsorption, mainly due to the large difference<br />
in specific surface area <strong>and</strong> accompanying number of<br />
adsorptive surface sites. Apart from arsenic adsorption,<br />
there may also be arsenic immobilization through coprecipitation<br />
with Fe 2+ to ferrihydrite, or, depending on<br />
pH <strong>and</strong> Fe:As ratio, precipitation of other complexes<br />
(such as simplesite; Johnston, 2008).<br />
In Bangladesh, elevated iron <strong>and</strong> arsenic<br />
concentrations have been found to often co-occur in<br />
groundwater (Nickson et al., 2000). Presence of elevated<br />
iron levels is considered a key criterion for the technical<br />
feasibility of SAR. In the small-scale setting, subsurface<br />
arsenic removal relies on the existing infrastructure<br />
of a h<strong>and</strong>-pump/shallow tube-well <strong>and</strong> retains iron<br />
(b) abstraction phase<br />
1<br />
Ground water level<br />
containing Fe 2+ <strong>and</strong> As(III)<br />
Ground water level<br />
containing Fe 2+ <strong>and</strong> As(III)<br />
O2 front<br />
iron oxide<br />
adsorbed iron oxideFe 2+<br />
<strong>and</strong> As(III)<br />
with adsorbed<br />
Fe 2+ <strong>and</strong> As(III)<br />
Injected water front<br />
Figure 1.4 Principle of subsurface iron removal (a) injection <strong>and</strong> (b) abstraction<br />
11
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
1 <strong>and</strong> arsenic in the subsurface (Figure 1.4). As such,<br />
it has potential advantages over other household <strong>and</strong><br />
community arsenic removal systems, such as SONO<br />
<strong>and</strong> Alcan (Sutherl<strong>and</strong> et al., 2002; IGRAC 2007;<br />
BETV-SAM, 2011):<br />
• no costly filter media <strong>and</strong> maintenance is needed;<br />
• the tube well is the 1st preferred option for drinking<br />
water in rural Bangladesh (WSP/Worldbank, 2003);<br />
<strong>and</strong> available to a majority of the rural poor in their<br />
household;<br />
• (minimal) additional hardware beyond the existing h<strong>and</strong><br />
pump is affordable <strong>and</strong> locally available/repairable;<br />
• iron is also removed which improves colour <strong>and</strong> taste<br />
of the water; greatly enhancing potential for social<br />
acceptance;<br />
• iron could be a visible indicator for arsenic presence (<strong>and</strong><br />
aid in post-deployment monitoring of water quality);<br />
• groundwater-irrigation leading to arsenic accumulation<br />
in crops (rice) <strong>and</strong> harvest reduction may also be<br />
mitigated.<br />
Clearly, the SIR/SAR technology scores high for the<br />
sustainability criterion of accessibility. And the absence<br />
of a waste stream <strong>and</strong> electricity/chemicals shows<br />
the great potential of this technology regarding the<br />
environmental footprint. The social acceptance <strong>and</strong><br />
proper operation <strong>and</strong> maintenance (functionality) is<br />
yet to be determined, but the co-removal of iron may<br />
well aid in that. The criteria of sufficient water quantity<br />
<strong>and</strong> water quality includes, for an arsenic removal<br />
system, the production of >2L with
1 Introduction<br />
to be investigated. In order to do so, knowledge needs<br />
to be gained regarding the subsurface processes that<br />
determine the site-specific efficacy of SIR/SAR.<br />
This thesis<br />
<strong>Arsenic</strong> contamination of shallow tube well drinking<br />
water in Bangladesh is an urgent developmental <strong>and</strong><br />
health problem (British Geological Survey/DPHE,<br />
2001; World Health Organization, 2001; Smith et al.,<br />
2002), disproportionately affecting the rural poor,<br />
i.e., those most reliant on this source of drinking<br />
water. Current arsenic mitigation solutions, including<br />
(household) arsenic removal options, do not always<br />
provide a sustainable alternative for safe drinking<br />
water in rural Bangladesh. A novel HWTS combines<br />
the removal of iron <strong>and</strong> arsenic by utilizing the existing<br />
h<strong>and</strong> pump infrastructure. This new technology shows<br />
great potential, but for safe implementation there is the<br />
need to address urgent research questions regarding<br />
the production of sufficient water quality <strong>and</strong> water<br />
quantity. The problem description of this thesis has<br />
been formulated as follows:<br />
<strong>Subsurface</strong> <strong>Iron</strong> <strong>and</strong> <strong>Arsenic</strong> <strong>Removal</strong> is a<br />
promising safe water solution, however, there is the<br />
need for better underst<strong>and</strong>ing of the (subsurface)<br />
processes determining the sustainable operation<br />
in diverse geochemical settings.<br />
This problem has been addressed by identifying research<br />
questions within three main knowledge gaps: (1) Fe 2+ /<br />
As(III) immobilization processes, (2) site-specific<br />
effectiveness, <strong>and</strong> (3) safe <strong>and</strong> sustainable application.<br />
The knowledge gaps include research topics that are<br />
within <strong>and</strong> outside the scope of this thesis, as has been<br />
summarized below.<br />
Fe 2+ /As(III) immobilization processes<br />
Although much work has already been done on the<br />
oxidation <strong>and</strong> adsorption kinetics of iron <strong>and</strong> arsenic,<br />
the rapidly changing redox conditions during injectionabstraction<br />
cycles make SIR/SAR a unique application.<br />
A combination of Fe 2+ oxidation, precipitation,<br />
adsorption <strong>and</strong> exchange may occur during injectionabstraction<br />
cycles, resulting in the immobilization<br />
of iron. Adsorptive-catalytic oxidation <strong>and</strong> cation<br />
exchange have been proposed to dominate the<br />
subsurface iron removal process (van Beek, 1985; Appelo<br />
et al., 1999), with arsenic removal being a beneficiary<br />
side effect. However, to target effective arsenic removal<br />
it is desirable to underst<strong>and</strong> the behaviour of this<br />
constituent during the adsorptive-catalytic oxidation<br />
mechanism, including the speciation of arsenic. In this<br />
thesis the competition of iron oxidizing bacteria for the<br />
same dissolved oxygen as the chemical Fe 2+ oxidation<br />
reaction will not be addressed. This does, however, not<br />
mean that it is assumed that IOB communities play an<br />
insignificant role during SIR, as the observed presence<br />
of Gallionella spp. in subsurface treated groundwater<br />
13<br />
1
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
1 could point towards vivid microbial activity (van<br />
Halem et al., 2008). The focus of this thesis lies on the<br />
supply of oxygen to adsorbed <strong>and</strong> exchangeable Fe 2+<br />
<strong>and</strong> to unravel the complex combination of adsorption,<br />
(heterogeneous) oxidation <strong>and</strong> cation exchange<br />
reactions that determine the removal efficacy of iron<br />
<strong>and</strong> arsenic.<br />
Site-specific effectiveness in diverse (geochemical)<br />
conditions<br />
The decentralized character of a household water<br />
treatment <strong>and</strong> storage solution requires that a smallscale<br />
SIR/SAR performs in a diversity of geochemical<br />
conditions. Fe 2+ <strong>and</strong> As(III) oxidation <strong>and</strong> adsorption<br />
reactions are influenced by groundwater quality<br />
parameters such as pH, temperature, Fe:As ratio,<br />
competing cations or anions, <strong>and</strong> more. Apart from<br />
the water composition, it is also reported in the<br />
literature that certain mineral structures may catalyze<br />
the oxidation <strong>and</strong>/or adsorption process (e.g., Fe 3+<br />
oxides, calcite). On the other h<strong>and</strong>, the presence of<br />
other mineral structures may seriously threaten safe<br />
SIR/SAR operation, such as (arseno)pyrite. Both<br />
groundwater <strong>and</strong> soil composition are given parameters<br />
at a certain site, <strong>and</strong> may vary significantly from well<br />
to well. It is therefore evident that more knowledge<br />
needs to be gained on the performance of SIR/SAR in<br />
diverse geochemical settings. Clearly, local variations<br />
in socio-economic <strong>and</strong> geohydrological conditions<br />
could also greatly influence the effective small-scale<br />
Fe 2+ /As immoblization<br />
site-specific efficacy<br />
II: experience in the Netherl<strong>and</strong>s<br />
safe <strong>and</strong> sustainable<br />
III: small-scale application<br />
IV: adsorptive-catalytic oxidation<br />
V: Fe 2+ exchange<br />
VI: clogging<br />
by deposits<br />
VII: catalysis by deposits<br />
VIII: influence groundwater composition<br />
Figure 1.5 Synthesis between chapters <strong>and</strong> the formulated knowledge gaps<br />
14
1 Introduction<br />
implementation of SIR/SAR. Although indisputable<br />
important, considerations regarding social acceptance,<br />
seasonal fluctuations <strong>and</strong> spatial solute transport were<br />
beyond the scope of this thesis.<br />
Safe <strong>and</strong> sustainable application<br />
H<strong>and</strong> pump tube wells are the prime source of drinking<br />
water in rural Bangladesh <strong>and</strong> may therefore never be<br />
threatened in their ability to produce sufficient <strong>and</strong> safe<br />
water. The application of SIR/SAR as a HWTS requires a<br />
thorough investigation of the long-term environmental<br />
<strong>and</strong> health effects of this technology. The subsurface<br />
surroundings of the shallow tube well will be affected<br />
by the accumulation of Fe/As deposits once SIR/SAR<br />
is in operation. Knowledge needs to be gained on<br />
whether these deposits could clog the aquifer <strong>and</strong>/or<br />
well <strong>and</strong> whether these deposits may release arsenic<br />
during irregular injection-abstraction modes. Clogging<br />
of the aquifer has not been reported in large-scale SIR<br />
wells in Europe, but injection volumes at family owned<br />
h<strong>and</strong> pumps are one-thous<strong>and</strong>th of the 1,000-2,000 m 3<br />
in full scale plants. SIR/SAR at household level requires<br />
new boundary conditions for sustainable application,<br />
including operational modes <strong>and</strong> post-deployment<br />
monitoring.<br />
The defined knowledge gaps form the basis for the<br />
main research questions of the individual chapters<br />
of this thesis. The research questions overlap one or<br />
more knowledge gaps, as illustrated by the schematic<br />
synthesis overview in Figure 1.5. The research questions<br />
per chapter are:<br />
I. What lessons can be drawn from the past experiences<br />
with SIR in the Netherl<strong>and</strong>s (Chapter 2)?<br />
II. Can SIR/SAR be applied at a h<strong>and</strong> pump/tube-well level<br />
in Bangladesh (Chapter 3)?<br />
III. What is the contribution of the adsorptive-catalytic oxidation<br />
mechanism on the retention of Fe 2+ <strong>and</strong> As(III)<br />
during SIR/SAR (Chapter 4)?<br />
IV. Does Fe 2+ exchange play a role during an injection-abstraction<br />
cycle (Chapter 5)?<br />
V. What kinds of deposits accumulate around a SIR well<br />
<strong>and</strong> do they clog the aquifer (Chapter 6)?<br />
VI. Does the presence of accumulated deposits catalyze the<br />
subsurface removal of Fe <strong>and</strong> As (Chapter 7)?<br />
VII. What is the influence of groundwater composition on<br />
the efficacy of SIR/SAR (Chapter 8)?<br />
In Chapter 2 an overview is given of three decades<br />
of experience with subsurface iron removal in the<br />
Netherl<strong>and</strong>s. This overview focuses on current/past<br />
practices, site-specific efficacy <strong>and</strong> long-term operation<br />
<strong>and</strong> is the starting point of the research presented in<br />
this thesis.<br />
Chapter 3 focuses primarily on the technical feasibility<br />
of small-scale application of SIR/SAR at family level in<br />
Manikganj (Bangladesh), with injection volumes below<br />
1m 3 .<br />
Chapter 4 translates the field observations from<br />
Bangladesh <strong>and</strong> the Netherl<strong>and</strong>s, to injection-<br />
15<br />
1
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
1 abstraction s<strong>and</strong> columns in order to investigate the<br />
adsorptive-catalytic oxidation mechanism during<br />
successive cycles.<br />
Additional s<strong>and</strong> column experiments were conducted<br />
to investigate the proposed occurrence of cation<br />
exchange during injection <strong>and</strong> abstraction, as presented<br />
in Chapter 5.<br />
Clogging of the aquifer during long-term operation of<br />
SIR is a concern that is addressed in Chapter 6, where<br />
results from characterized deposits near a 12-year-old<br />
subsurface iron removal well are presented.<br />
The same sediments were used for column studies to<br />
investigate the effect of the accumulated deposits on<br />
adsorptive-catalytic Fe 2+ oxidation, Fe 2+ exchange <strong>and</strong><br />
As(III) removal during SIR/SAR (Chapter 7).<br />
Chapter 8 elaborates on the influence of groundwater<br />
composition on the efficacy of SIR/SAR. The limiting<br />
<strong>and</strong>/or catalyzing effect on SIR/SAR of commonly cooccurring<br />
ions, such as phosphate, nitrate <strong>and</strong> calcium,<br />
are investigated with s<strong>and</strong> columns in laboratory <strong>and</strong><br />
natural groundwater.<br />
The final chapter, Chapter 9, provides the reader with<br />
the concluding remarks of this thesis. These conclusions<br />
provide a broader perspective than the individual<br />
chapters <strong>and</strong> recommendations are given for future<br />
research.<br />
References<br />
Aiuppa A., W. D’Aless<strong>and</strong>ro, C. Federico, B. Palumbo <strong>and</strong><br />
M. Valenza (2003) The aquatic geochemistry of arsenic in<br />
volcanic groundwaters from southern Italy, Appl Geochem<br />
18(9): 1283-1296.<br />
Appelo C. A. J., B. Drijver, R. Hekkenberg <strong>and</strong> M. de Jonge<br />
(1999) Modeling in situ iron removal from ground water,<br />
Ground Water 37(6): 811-817.<br />
Appelo C. A. J. <strong>and</strong> W. W. J. M. de Vet (2003) Modeling in situ<br />
iron removal from groundwater with trace elements such<br />
as As. In <strong>Arsenic</strong> in groundwater. A. H. Welch <strong>and</strong> K.G.<br />
Stollenwerk. Kluwer Academic, Boston.<br />
Appelo C.A.J. <strong>and</strong> D. Postma (2005) Geochemistry, groundwater<br />
<strong>and</strong> pollution. Balkema, Rotterdam, 2nd edition.<br />
Appleyard S. J., J. Angeloni <strong>and</strong> R. Watkins (2006) <strong>Arsenic</strong>rich<br />
groundwater in an urban area experiencing drought<br />
<strong>and</strong> increasing population density, Perth, Australia, Applied<br />
Geochemistry 21(1): 83-97.<br />
Armienta M. A. <strong>and</strong> N. Segovia (2008) <strong>Arsenic</strong> <strong>and</strong> fluoride<br />
in the groundwater of Mexico, Environ Geochem <strong>and</strong> Hlth<br />
30(4): 345-353.<br />
AWWA Research Foundation (2000) <strong>Arsenic</strong> treatability options<br />
<strong>and</strong> evaluation of residuals management issues, American<br />
Water Works Association, Denver.<br />
Banerjee K., G.L. Amy, M. Prevost, S. Nourc, M. Jekel, P.M.<br />
Gallagher <strong>and</strong> C.D. Blumenscheine (2008) Kinetic <strong>and</strong><br />
thermodynamic aspects of adsorption of arsenic onto<br />
granular ferric hydroxide (GFH), Water Research 42: 3371 –<br />
3378.<br />
16
1 Introduction<br />
BETV-SAM: Bangladesh Environmental Technology Verification<br />
– Support to <strong>Arsenic</strong> Mitigation, www.verification-unit.org.<br />
Boochs P. W. <strong>and</strong> G. Barovic (1981) Numerical-Model Describing<br />
Groundwater Treatment by Recharge of Oxygenated Water,<br />
Water Resources Research 17(1): 49-56.<br />
Borba R. P., B. R. Figueiredo <strong>and</strong> J. Matschullat (20030<br />
Geochemical distribution of arsenic in waters, sediments <strong>and</strong><br />
weathered gold mineralized rocks from <strong>Iron</strong> Quadrangle,<br />
Brazil, Environ Geol 44(1): 39-52.<br />
Bose P., A. Sharma (2002) Role of iron in controlling speciation<br />
<strong>and</strong> mobilization of arsenic in subsurface environment,<br />
Water Research 36: 4916-4926.<br />
Braester C. <strong>and</strong> R. Martinell (1988) The Vyredox <strong>and</strong> Nitredox<br />
methods of in situ treatment of groundwater, Water Science<br />
<strong>and</strong> Technology 20(3): 149-163.<br />
British Geological Survey/DPHE: <strong>Arsenic</strong> contamination of<br />
groundwater in Bangladesh (2001) Volume 2: Final report,<br />
BGS Technical Report WC/00/19.<br />
Buschmann J., M. Berg, C. Stengel <strong>and</strong> M.L. Sampson (2007)<br />
<strong>Arsenic</strong> <strong>and</strong> manganese contamination of drinking water<br />
resources in Cambodia: Coincidence of risk areas with low<br />
relief topography, Environ Sci <strong>and</strong> Technol 41(7): 2146-2152.<br />
Buschmann J., M. Berg, C. Stengel, L. Winkel, M.L. Sampson,<br />
P.T.K. Trang <strong>and</strong> P.H. Viet (2008) Contamination of drinking<br />
water resources in the Mekong delta floodplains: <strong>Arsenic</strong> <strong>and</strong><br />
other trace metals pose serious health risks to population,<br />
Environ Int 34(6): 756-764.<br />
Celik I., L. Gallicchio, K. Boyd, T.K. Lam, G. Matanoski, X.<br />
Tao, M. Shiels, E. Hammond, L. Chen, K. A. Robinson, L.E.<br />
Caulfield, J.G. Herman, E. Guallar <strong>and</strong> A.J. Alberg (2008)<br />
<strong>Arsenic</strong> in drinking water <strong>and</strong> lung cancer: a systematic<br />
review, Environ Res 108: 48-55.<br />
Chen C.J., C.W. Chen, M.M. Wu <strong>and</strong> T.L. Kuo (1992) Cancer<br />
potential in liver, lunch bladder <strong>and</strong> kidney due to injected<br />
inorganic arsenic in drinking water, Br J Cancer 66: 888-892.<br />
Chen Y. <strong>and</strong> H. Ahsan (2004) Cancer burden from arsenic in<br />
drinking water in Bangladesh, Am J Public Health 94(5): 741-<br />
744.<br />
Chowdhury M.A.I., M.T. Uddin, M.F. Ahmed, M.A. Ali <strong>and</strong><br />
S.M. Uddin (2006) How does arsenic contamination of<br />
groundwater cause severity <strong>and</strong> health hazard in Bangladesh,<br />
J Appl Sci 6(6): 1275-1286.<br />
Clasen T. (2008) Scaling Up Household Water Treatment:<br />
Looking Back, Seeing Forward. Geneva: World Health<br />
Organization.<br />
de Vet W.W.J.M., I.J.T. Dinkla, G. Muyzer, L.C. Rietveld <strong>and</strong><br />
M.C.M. van Loosdrecht (2009) Molecular characterization<br />
of microbial populations in groundwater sources <strong>and</strong> s<strong>and</strong><br />
filters for drinking water production, Water Research 43(1):<br />
182-194.<br />
Dixit S. <strong>and</strong> J. G. Hering (2003) Comparison of arsenic(V) <strong>and</strong><br />
arsenic(III) sorption onto iron oxide minerals: implications<br />
for arsenic mobility. Environ. Sci. Technol. 37: 4182-4189.<br />
Dixit S. <strong>and</strong> J. G. Hering (2006) Sorption of Fe(II) <strong>and</strong> As(III)<br />
on goethite in single- <strong>and</strong> dual-sorbate systems, Chemical<br />
Geology 228(1-3): 6-15.<br />
Environmental Protection Agency: <strong>Arsenic</strong> Treatment<br />
Technologies, Environmental Technology Verification<br />
Program, EPA/600/S-07/007 (2007) www.epa.gov/etv/vtdws.html#traic.<br />
17<br />
1
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
1<br />
Environmental Protection Agency: Integrated Risk Information<br />
System, <strong>Arsenic</strong>, inorganic; CASRN 7440-38-2 (1998) www.<br />
epa.gov/NCEA/iris/subst/0278.htm.<br />
Ferguson J.F. <strong>and</strong> J. Gavis (1972) A review of the arsenic cycle in<br />
natural waters, Water Research 6: 1259-1274.<br />
Fewtrell L., R. B. Kaufman, D. Kay, W. Enanoria, L. Haller<br />
<strong>and</strong> J. M. Colford Jr (2005) Water, sanitation, <strong>and</strong> hygiene<br />
interventions to reduce diarrhoea in less developed countries:<br />
a systematic review <strong>and</strong> meta-analysis, Lancet Infect Dis. 5:<br />
42–52.<br />
Frommer J., A. Voegelin, J. Dittmar, M.A. Marcus <strong>and</strong> R.<br />
Kretzschmar (2011) Biogeochemical processes <strong>and</strong> arsenic<br />
enrichment around rice roots in paddy soil: results from<br />
micro-focused X-ray spectroscopy, European Journal of Soil<br />
Science 62: 305-317.<br />
Grombach P. (1985) Groundwater treatment in situ in the<br />
aquifer, Water Supply 3(1): 13-18.<br />
Gunduz O., C. Simsek <strong>and</strong> A. Hasozbek (2009) <strong>Arsenic</strong> pollution<br />
in the groundwater of Simav Plain, Turkey: its impact on<br />
water quality <strong>and</strong> human health, Water Air Soil Pollut 205<br />
(1-4): 43-62.<br />
Hallberg R. O. <strong>and</strong> R. Martinell (1976) Vyredox - in situ<br />
purification of groundwater, Ground Water, 14(2): 88-93.<br />
Hiemstra T. <strong>and</strong> W.H. van Riemsdijk (2007) Adsorption <strong>and</strong><br />
surface oxidation of Fe(II) on metal (hydr)oxides, Geochim.<br />
Cosmochim. Acta 71: 5913-5933.<br />
Hunter P.R. (2009) Household Water Treatment in Developing<br />
Countries: Comparing Different Intervention Types Using<br />
Meta-Regression. Environmental Science <strong>and</strong> Technology 43<br />
(23): 8991–8997.<br />
IGRAC (2009) <strong>Arsenic</strong> in groundwater: Overview <strong>and</strong> evaluation<br />
of removal methods. International Groundwater Resources<br />
Assessment Centre, Report nr. SP 2007-2.<br />
International Agency for Research on Cancer: Some drinkingwater<br />
disinfectants <strong>and</strong> contaminants, including arsenic.<br />
(2004) IARC Monographs on the evaluation of carcinogenic<br />
risks to humans, Volume 84.<br />
Jain C.K. <strong>and</strong> I. Ali (2000) <strong>Arsenic</strong>: occurrence, toxicity <strong>and</strong><br />
speciation techniques, Water Research 43(17):4304-4313.<br />
Johnston R., H. Heijnen <strong>and</strong> P. Wurzel (2001) Chapter 6. Safe<br />
Water Technology, in United Nations Synthesis Report on<br />
<strong>Arsenic</strong> in Drinking Water, World Health Organisation.<br />
Johnston R. (2008) Chemical interactions between iron <strong>and</strong><br />
arsenic in water, PhD dissertation, University of North<br />
Carolina at Chapel Hill.<br />
Jovanovic, D., B. Jakovljevic, Z. Rasic-Milutinovic, K. Paunovic,<br />
G.Pekovic <strong>and</strong> T. Knezevic (2011) <strong>Arsenic</strong> occurrence in<br />
drinking water supply systems in ten municipalities in<br />
Vojvodina Region, Serbia, Environmental Research 111(2):<br />
315-318.<br />
Kim M.J. <strong>and</strong> J. Nriagu (2000) Oxidation of arsenite in<br />
groundwater using ozone <strong>and</strong> oxygen, The Science of the<br />
Total Environment 247: 71-79.<br />
Liao, C.M., H.H. Shen, C.L. Chen, L.I.Hsu, T.L. Lin, S.C. Chen<br />
<strong>and</strong> C.J. Chen (2009) Risk assessment of arsenic-induced<br />
internal cancer at long-term low dose exposure, Journal of<br />
Hazardous Materials 165: 652-663.<br />
Marszalek, H. <strong>and</strong> M. Wasik (2000) Influence of arsenic-bearing<br />
gold deposits on water quality in Zloty Stok mining area (SW<br />
Pol<strong>and</strong>), Environ Geol 39(8): 888-892.<br />
18
1 Introduction<br />
Masscheleyn P.H., R.D. Delaune, W.H. Patrick (1999) Effect of<br />
redox potential <strong>and</strong> pH on arsenic speciation <strong>and</strong> solubility<br />
in a contaminated soil, Environmental Science & Technology<br />
25: 1414-1419.<br />
Mettler S., M. Abdelmoula, E. Hoehn, R. Schoenenberger, P.<br />
Weidler <strong>and</strong> U. von Gunten (2001) Characterization of iron<br />
<strong>and</strong> manganese precipitates from an in situ groundwater<br />
treatment plant, Ground Water 6: 921-930.<br />
Mettler S. (2002) In-situ removal of iron from groundwater:<br />
Fe(II) oxygenation, <strong>and</strong> precipitation products in a<br />
calcareous aquifer, PhD dissertation, Swiss Federal Institute<br />
of Technology, Zurich.<br />
Natural Recourses Defense Council: <strong>Arsenic</strong> <strong>and</strong> old laws: A<br />
scientific <strong>and</strong> public health analysis of arsenic occurrence<br />
in drinking water, its health effects, <strong>and</strong> EPA’s outdated<br />
arsenic tap water st<strong>and</strong>ard, 2000: http://www.nrdc.org/water/<br />
drinking/arsenic/aolinx.asp.<br />
Nickson R., J. McArthur, P. Ravencroft, W.G. Burgess <strong>and</strong> K.M.<br />
Ahmed (2000) Mechanism of arsenic release to groundwater,<br />
Bangladesh <strong>and</strong> West Bengal, Applied Geochemistry 15(4):<br />
403-413.<br />
Petrusevski B., W. van der Meer, J. Baker, F. Kruis, S.K. Sharma<br />
<strong>and</strong> J.C. Schippers (2007) Innovative approach for treatment<br />
of arsenic contaminated groundwater in Central Europe,<br />
Water Sci Technol: Water Supply 7(3): 131–138.<br />
Peacock C. L. <strong>and</strong> D. M. Sherman (2004) Copper(II) sorption<br />
onto goethite, hematite <strong>and</strong> lepidocrocite: A surface<br />
complexation model based on ab initio molecular geometries<br />
<strong>and</strong> EXAFS spectroscopy, Geochimica Et Cosmochimica<br />
Acta 68(12): 2623-2637.<br />
Rott U. (1985) Physical, chemical <strong>and</strong> biological aspects of the<br />
removal of iron <strong>and</strong> manganese underground. Water Supply<br />
3(2), 143-150.<br />
Rott U., C. Meyer <strong>and</strong> M. Friedle (2002) Residue-free removal<br />
of arsenic, iron, mangenese <strong>and</strong> ammonia from groundwater.<br />
Water Science <strong>and</strong> Technology: Water Supply 2(1), 17-24.<br />
Sambu S. <strong>and</strong> R. Wilson (2008) <strong>Arsenic</strong> in food <strong>and</strong> water – a<br />
brief history, Toxicol Ind Health 24: 217-226.<br />
SAR Technology (2011) In-situ arsenic treatment. www.<br />
insituarsenic.org.<br />
Sarkar A. R. <strong>and</strong> O.T. Rahman (2001) In-situ removal of arsenic<br />
- experiences of DPHE-Danida pilot project. In Technologies<br />
for arsenic removal from drinking water, Bangladesh<br />
University of Engineering <strong>and</strong> Technology <strong>and</strong> The United<br />
Nations University, Bangladesh.<br />
Sen Gupta B., S. Chatterjee, U. Rott, H. Kauffman, A.<br />
B<strong>and</strong>opadhyay, W. de Groot, N.K. Nag, A.A. Borbonell-<br />
Barrachina <strong>and</strong> S. Mukherjee (2009) A simple chemical free<br />
arsenic removal method for community, Environmental<br />
Pollution 157: 3351–3353.<br />
Smedley P.L <strong>and</strong> D.G. Kinniburgh (2002) A review of the source,<br />
behaviour <strong>and</strong> distribution of arsenic in natural waters, Appl<br />
Geochem 17: 517–568.<br />
Smedley P. L., M. Zhan, G. Zhang <strong>and</strong> Z. Luo (2003) Mobilisation<br />
of arsenic <strong>and</strong> other trace elements in fluviolacustrine aquifers<br />
of the Huhhot Basin, Inner Mongolia, Appl Geochem 18(9):<br />
1453-1477.<br />
Smedley P. L., J. Knudsen <strong>and</strong> D. Maiga (2007) <strong>Arsenic</strong> in<br />
groundwater from mineralised Proterozoic basement rocks<br />
of Burkina Faso, Appl Geochem 22(5): 1074-1092.<br />
19<br />
1
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
1<br />
Smith A.H., C. Hopenhayn-Rich, M.N. Bates, H.M. Goeden, I.<br />
Hertz-Picciotto, H.M. Duggan, R. Wood, M.J. Kosnett <strong>and</strong><br />
M.T. Smith (1992) cancer risks from arsenic in drinking<br />
water, Environmental Health Perspectives 97: 259-267.<br />
Smith A.H., E.O. Lingas <strong>and</strong> M. Rahman (2000) Contamination<br />
of drinking-water by arsenic in Bangladesh: a public health<br />
emergency, Bulletin of the World Health Organization, 78<br />
(9): 1093-1103.<br />
Smith A.H., P.A. Lopipero, M.N. Bates <strong>and</strong> C.M. Steinmaus<br />
(2002) <strong>Arsenic</strong> epidemiology <strong>and</strong> drinking water st<strong>and</strong>ards,<br />
Science 296: 2145-2146.<br />
Sobsey M.D. (2002) Managing water in the home: accelerated<br />
health gains from improved water supply. Geneva: The World<br />
Health Organization (WHO/SDE/WSH/02.07)<br />
Sobsey, M.D., C.E. Stauber, L.M. Casanova, J.M. Brwon <strong>and</strong><br />
M.A. Elliott (2008) Point of Use Household Drinking Water<br />
Filtration: A Practical, Effective Solution for Providing<br />
Sustained Access to Safe Drinking Water in the Developing<br />
World, Environ, Sci. Technol. 42: 4261–4267.<br />
Sommer B., A. Mariño, Y. Solarte, M.L. Salas, C. Dierolf, C.<br />
Valiente, D. Mora, R. Rechsteiner, P. Setter, W. Wirojanagud,<br />
H. Ajarmeh, A. Al-Hassan <strong>and</strong> M. Wegelin (1997) SODIS<br />
- An emerging water treatment process. Journal of Water<br />
Supply: Research <strong>and</strong> Technology - AQUA 46(3): 127-137.<br />
Sposito G. (1989) The Chemistry of Soils, New York, Oxford<br />
University Press.<br />
Stauber C.E., M.A. Elliot, F. Koksal, G.M. Ortiz, F.A. DiGiano<br />
<strong>and</strong> M.D. Sobsey (2006) Characterisation of the bios<strong>and</strong> filter<br />
for E. coli reductions from household drinking water under<br />
controlled laboratory <strong>and</strong> field use conditions, Water Science<br />
<strong>and</strong> Technology 54 (3): 1-7<br />
Stollenwerk K.G. Geochemical processes controlling transport<br />
of arsenic in groundwater: a review of adsorption (2003) In<br />
<strong>Arsenic</strong> in groundwater. A. H. Welch <strong>and</strong> K.G. Stollenwerk.<br />
Kluwer Academic, Boston.<br />
Stumm W. <strong>and</strong> J.J. Morgan (1996) Aquatic chemistry: chemical<br />
equilibrium <strong>and</strong> rates in natural waters. 3rd edition, Wiley<br />
Interscience, New York.<br />
Stuyfz<strong>and</strong>, P.J., P. van Rossem <strong>and</strong> I. Mendizabal (2006) Does<br />
arsenic, in groundwaters of the compound Rhine-Meuse-<br />
Scheldt-Ems delta, menace drinking water supply in the<br />
Netherl<strong>and</strong>s? Proceedings of IAH meeting, Utrecht.<br />
Sutherl<strong>and</strong> D., P.M. Swash, A.C. MacQueen, L.E. McWilliam,<br />
D.J. Ross <strong>and</strong> S.C. Wood (2002) A field based evaluation of<br />
household arsenic removal technologies for the treatment<br />
of drinking water, Environmental Technology 23(12): 1385-<br />
1404.<br />
Tamura H., S. Kawamura <strong>and</strong> M. Hagayama (1980) Acceleration<br />
of the oxidation of Fe 2+ ions by Fe(III)-oxyhhydroxides,<br />
Corrosion Science 20, 963-971.<br />
Tseng W.P., H.M. Chu, S.W. How, J.M. Fong, C.S. Lin <strong>and</strong> S.<br />
Yeh (1968) Prevalence of skin cancer in an endemic area of<br />
chronic arsenicism in Taiwan, J Nat Cancer Inst 40: 453-463.<br />
UNICEF (2009) Promotion of household water treatment<br />
<strong>and</strong> safe storage in UNICEF WASH programmes, Water,<br />
Environment <strong>and</strong> Sanitation Section, New York.<br />
van Beek C.G.E.M. (1985) Experiences with underground water<br />
treatment in the Netherl<strong>and</strong>s, Water Supply 3: 1-11.<br />
van Geen A., A. P. Robertson <strong>and</strong> J.O. Leckie (1994) Complexation<br />
of Carbonate Species at the Goethite Surface - Implications<br />
20
1 Introduction<br />
for Adsorption of Metal-Ions in Natural-Waters, Geochimica<br />
Et Cosmochimica Acta 58(9): 2073-2086.<br />
van Halem, D., S.G.J. Heijman, A.I.A. Soppe, J.C. van Dijk <strong>and</strong><br />
G.L. Amy (2007) Ceramic silver-impregnated pot filters for<br />
household drinking water treatment in developing countries:<br />
material characterization <strong>and</strong> performance study. Water<br />
Science <strong>and</strong> Technology, Water Supply 7(5-6), 9-17.<br />
van Halem, D., W. W. J. M. de Vet, G.L. Amy <strong>and</strong> J.C. van<br />
Dijk (2008) <strong>Subsurface</strong> iron removal for drinking water<br />
production: underst<strong>and</strong>ing the process <strong>and</strong> exploiting<br />
beneficial side effects, Water Quality Technology Conference,<br />
Cincinnati, American Water Works Association.<br />
van Halem, D., H. van der Laan, S.G.J. Heijman, J.C. van Dijk<br />
<strong>and</strong> G.L. Amy (2009) Assessing the sustainability of the<br />
silver-impregnated ceramic pot filter for low-cost household<br />
drinking water treatment, Physics <strong>and</strong> Chemistry of the<br />
Earth 34: 36-42.<br />
von Oesten, G. (1900) Verfahren zur Enteisenung von<br />
Grundwasser im Untergrund selbst, Patentschrift Nr. 114709,<br />
Kaiserliches Patentamt Berlin.<br />
Williams P.N., M.R. Islam, E.E. Adomako, A. Raab, S.A. Hossain,<br />
Y.G. Zhu, J. Feldmann <strong>and</strong> A.A. Meharg (2006) Increase in<br />
rice grain arsenic for regions of Bangladesh irrigating paddies<br />
with elevated arsenic in groundwater, Environ. Sci. Technol.,<br />
40(16): 4903-4908.<br />
Woo, N. C. <strong>and</strong> M. C. Choi (2001) <strong>Arsenic</strong> <strong>and</strong> metal<br />
contamination of water resources from mining wastes in<br />
Korea, Environ Geol 40(3): 305-311.<br />
World Health Organization (2001) United Nations synthesis<br />
report on arsenic in drinking water, Geneva.<br />
World Health Organization (2001b) <strong>Arsenic</strong> <strong>and</strong> arsenic<br />
compounds, Environmental Health Criteria 224, Second<br />
edition, Geneva.<br />
World Health Organization (2003) <strong>Arsenic</strong> in drinking-water,<br />
Background document for development of WHO guidelines<br />
for drinking-water quality, Geneva.<br />
World Health Organization (2006) Guidelines for drinkingwater<br />
quality, First Addendum to Third Edition, Volume 1<br />
Recommendations, Geneva.<br />
World Health Organization (2007) Combating waterborne<br />
diseases at the household level, Geneva.<br />
World Health Organization (2011) Household water treatment<br />
<strong>and</strong> safe storage, www.who.int/household_water/en/<br />
WSP/Worldbank (2003) Fighting <strong>Arsenic</strong>: Listening to Rural<br />
Communities, Willingness to Pay for <strong>Arsenic</strong>-Free Safe<br />
Drinking Water in Bangladesh. Geneva.<br />
21<br />
1
2<br />
<strong>Subsurface</strong> iron removal in the<br />
Netherl<strong>and</strong>s<br />
This chapter is based on:<br />
van Halem et al. (2011) Journal of American Water Works Association: to be submitted
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
2<br />
An introduction to subsurface<br />
iron removal<br />
<strong>Iron</strong> removal in the Netherl<strong>and</strong>s<br />
In the Netherl<strong>and</strong>s there has been a long history of<br />
groundwater treatment, either above-ground or below<br />
the surface. Since the start of drinking water supply in<br />
the Netherl<strong>and</strong>s, the preferred source has always been<br />
microbiologically safe groundwater (in the Netherl<strong>and</strong>s<br />
this source can be found in confined s<strong>and</strong>y aquifers<br />
in most parts of the country; Smeets et al., 2009).<br />
Conventional above-ground treatment consists in<br />
general of the combination of aeration, e.g., cascade<br />
or plate aeration, <strong>and</strong> subsequent s<strong>and</strong> filtration. Prechlorination<br />
of the groundwater is not applied, as<br />
iron removal can also be achieved with the (slower)<br />
oxidation kinetics as obtained by aeration. <strong>Iron</strong> removal<br />
processes can be approached from a chemical <strong>and</strong><br />
biological perspective, with the latter being considered<br />
only since several decades (Mouchet, 1992; Czekalla et<br />
al., 1985). Chemical iron removal can be subdivided<br />
into oxidation-flocculation <strong>and</strong> adsorptive-oxidation.<br />
The formation of flocs results in faster clogging of the<br />
filter bed, whereas adsorptive iron removal produces a<br />
neatly ordered coating onto the s<strong>and</strong> grains (Sharma<br />
et al., 2001). Biological iron removal relies on the<br />
presence of iron oxidizing bacteria, such as Gallionella<br />
spp., to immobilize iron in the filter bed. An additional<br />
advantage of the bios<strong>and</strong> filters is that nitrification of<br />
ammonium-containing anaerobic groundwater can<br />
be obtained as well. Apart from aeration-filtration,<br />
also nanofiltration, reverse osmosis, activated carbon<br />
filtration <strong>and</strong> pellet softening reactors are sometimes<br />
practised to treat groundwater in the Netherl<strong>and</strong>s.<br />
After treatment, the groundwater is usually biologically<br />
stable <strong>and</strong>, therefore, post-chlorination of the water to<br />
prevent bacterial growth is not needed (Smeets et al.,<br />
2009).<br />
The principle of SIR<br />
For groundwater treatment a new iron removal<br />
technology was introduced in the 1970s, based on<br />
similar oxidation-adsorption principles as s<strong>and</strong><br />
filtration, but utilizing the natural treatment capacity of<br />
the aquifer. The principle of <strong>Subsurface</strong> <strong>Iron</strong> <strong>Removal</strong><br />
(SIR), or in-situ iron removal, is that aerated water is<br />
A<br />
B<br />
1.0<br />
C<br />
C 0<br />
Injection Oxygen front<br />
C<br />
C 0<br />
0.5<br />
0<br />
1.0<br />
0.5<br />
0<br />
Injection<br />
front<br />
distance from the well<br />
Fe 2+ front Fe 2+ front<br />
Abstraction<br />
t 2<br />
t 1<br />
distance from the well<br />
Figure 2.1 Behaviour of injection water, oxygen <strong>and</strong> iron front<br />
at a distance from the well (obtained from van Beek,<br />
1983)<br />
24
2 <strong>Subsurface</strong> iron removal in the Netherl<strong>and</strong>s<br />
periodically injected into an anoxic or anaerobic aquifer<br />
through a tube well (Figure 2.1A), partially displacing<br />
the original Fe 2+ -containing groundwater. The O 2<br />
-<br />
rich injection water oxidizes Fe 2+ in the subsurface<br />
environment around the tube well. When the flow is<br />
reversed, groundwater with low Fe concentrations is<br />
abstracted (Figure 2.1B). More water with reduced iron<br />
concentrations can be abstracted (volume V) than was<br />
injected (volume Vi), i.e., this volumetric ratio (V/Vi)<br />
determines the efficiency of the system.<br />
When O 2<br />
-rich water is injected into the Fe 2+<br />
saturated subsurface environment the adsorbed Fe 2+<br />
will oxidize:<br />
Equation 2.1<br />
S − OFe( II) + 0.25O + 1.5 H O →S − OFe( III)( OH ) + H<br />
+ 0 +<br />
2 2 2<br />
for the amorphous ferrihydrite is larger than crystalline<br />
mineral structures with lower surface areas (goethite,<br />
lepidocrocite). Besides the adsorptive-catalytic<br />
oxidation theory (Rott, 1985; van Beek, 1985), it has<br />
also been proposed that the injection of O 2<br />
-rich water<br />
promotes the exchange of adsorbed Fe 2+ with other<br />
cations, such as calcium (Appelo et al., 1999; Appelo<br />
<strong>and</strong> Postma, 2005), that Fe 2+ is effectively incorporated<br />
into the mineral structure through interfacial electron<br />
transfer (Mettler, 2002) or that iron oxidizing bacteria<br />
(IOB) enhance Fe 2+ removal (Hallberg <strong>and</strong> Martinell,<br />
1976).<br />
2<br />
With S being the solid iron oxide surface. The freshly<br />
formed Fe 3+ iron hydroxides provide new adsorption<br />
sites, for soluble Fe 2+ during the abstraction phase. The<br />
adsorption of Fe 2+ occurs in the absence of oxygen <strong>and</strong><br />
can be schematically formulated as:<br />
Equation 2.2<br />
o 2+ + +<br />
S − OH + Fe →S − OFe( II ) + H<br />
It should be noted that the use of S-OH 0 in the equation<br />
is simplified, as in reality adsorbed Fe 2+ may transfer an<br />
electron to the solid, creating an equivalent to Fe(OH) 2<br />
(Hiemstra <strong>and</strong> van Riemsdijk, 2007). The adsorptive<br />
capacity of the aquifer sediments depends on the type of<br />
iron oxides present in the soil, as adsorption capacities<br />
Figure 2.2 <strong>Subsurface</strong> iron removal patent from 1900 (obtained<br />
from Olthoff, 1986)<br />
25
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
2<br />
26<br />
SIR history<br />
With a patent originating from 1900 (von Oesten;<br />
Figure 2.2) subsurface iron removal by injection of<br />
oxygen-rich water into a tube well was first proposed in<br />
Berlin. In 1902 Piesch had a similar idea <strong>and</strong> patented<br />
a subsurface iron filter, where air was allowed to pass<br />
through the ground just before abstraction (Olthoff,<br />
1986; Figure 2.3). With only little experimental work<br />
in the following decades, it was not until 1971 that<br />
the Finns applied subsurface iron removal (Olthoff,<br />
1986), with a complete new multiple-well design<br />
(Vyredox TM ). Soon other European countries followed<br />
Figure 2.3 <strong>Subsurface</strong> iron removal patent from 1902 (obtained<br />
from Olthoff, 1986)<br />
with subsurface iron removal plants, including Sweden,<br />
Switzerl<strong>and</strong>, Germany, Denmark, France <strong>and</strong> the<br />
Netherl<strong>and</strong>s (van Beek, 1985; Mettler, 2002).<br />
It has been widely reported that a major<br />
benefit of subsurface iron removal is that the technology<br />
increases in efficacy with every injection-abstraction<br />
cycle (Boochs <strong>and</strong> Barovic, 1981; Braester <strong>and</strong> Martinell,<br />
1988; Grombach, 1985; Hallberg <strong>and</strong> Martinell, 1976;<br />
Mettler, 2002; Rott, 1985; van Beek, 1985), providing<br />
it with a great advantage over the above-ground<br />
alternative, which is sensitive to clogging. Clogging of<br />
the aquifer by iron sludge has not been reported as a<br />
limitation of the subsurface iron removal technology,<br />
<strong>and</strong> deposited iron precipitates have been characterized<br />
as compact, crystalline mineral structures (Mettler at al,<br />
2001; van Halem et al., 2011). Additionally it has been<br />
proposed that iron precipitates at various distances<br />
from the well, resulting in an even spreading of the<br />
precipitates around the tube well (Boochs <strong>and</strong> Barovic,<br />
1981; Appelo et al., 1999). The microbial water quality<br />
of the abstracted groundwater has not been found to<br />
be negatively affected by the injection of aerated water<br />
(van Beek, 1983).<br />
In the past decades several well designs have<br />
been implemented, including the single-well pushpull<br />
design, the dual well design <strong>and</strong> the Vyredox TM<br />
method (Hallberg <strong>and</strong> Martinell, 1976). The Vyredox TM<br />
method consists of a circle of injection wells around<br />
one abstraction well (also referred to as “star” or<br />
“satellite”), providing a subsurface oxidation screen for
2 <strong>Subsurface</strong> iron removal in the Netherl<strong>and</strong>s<br />
Fe 2+ adsorption. In the dual well design, two wells are<br />
intermittently used for injection or abstraction. The<br />
single-well design can be found most in Dutch drinking<br />
water production practices, as a normal production well<br />
can be relatively easily transitioned into a subsurface<br />
iron removal well. Apart from iron removal, the SIR<br />
technology has also been applied for the removal of<br />
manganese <strong>and</strong> arsenic, with varying results (van Beek,<br />
1985; Rott et al., 2002; van Halem et al., 2010). In the<br />
Netherl<strong>and</strong>s it is even operated at some locations to<br />
enhance the nitrification process in subsequent bios<strong>and</strong><br />
filters (de Vet et al., 2009), illustrating the beneficial<br />
side effects of SIR.<br />
Although subsurface iron removal has been<br />
applied for several decades in Europe it has not found<br />
widespread attention elsewhere. The objective of this<br />
study was to present an overview of the experience at<br />
two water treatment plants (WTPs) in the Netherl<strong>and</strong>s<br />
that have operated subsurface iron removal for several<br />
decades: WTP Corle <strong>and</strong> WTP Lekkerkerk. The<br />
overview includes information on the site-specific<br />
conditions <strong>and</strong> Fe removal efficacies, resulting in an<br />
analysis of the current knowledge on the influence of<br />
groundwater matrix <strong>and</strong> operational mode on the longterm<br />
<strong>and</strong> sustainable application of SIR.<br />
<strong>Subsurface</strong> iron removal sites<br />
Vitens Water Supply Company<br />
The majority of the subsurface iron removal<br />
treatment plants are located in the eastern provinces<br />
of the Netherl<strong>and</strong>s (Figure 2.4). Vitens Water Supply<br />
Company is currently responsible for the operation of<br />
10 SIR plants. SIR wells have been operated since 1977<br />
at WTP Vorden <strong>and</strong> WTP ‘t Klooster, whereas WTP<br />
Corle, WTP Eerbeek <strong>and</strong> WTP de Pol started subsurface<br />
treatment in 1980. A decade later, WTP Aalten was<br />
constructed in 1990 <strong>and</strong> has been applying SIR ever<br />
since. The number of SIR wells varies per location with<br />
4 wells at WTP Aalten up to 14 wells at WTP De Pol; in<br />
total 77 SIR wells are currently in operation by Vitens.<br />
In general, Vitens operates their SIR wells based on the<br />
volumetric V/Vi ratio, where iron concentration may<br />
never exceed 1.8 µmol.L -1 . The V/Vi ratio is determined<br />
for every well individually <strong>and</strong> then operated for several<br />
years before this ratio is retested, unless of course,<br />
iron concentrations exceed the limit of 1.8 µmol.L -1 .<br />
At a number of the water treatment plants Vitens has<br />
started to enrich the injection water with pure oxygen<br />
to increase the oxygen content from 0.28 mmol.L -1 to<br />
0.55 mmol.L -1 in order to operate at higher V/Vi ratios.<br />
Currently, the treatment plants of Vitens operate their<br />
SIR wells with injection volumes of 2,000 or 3,000 m 3<br />
<strong>and</strong> V/Vi ratios varying between 6 <strong>and</strong> 20.<br />
2<br />
27
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
2<br />
Oasen Drinking Water Company<br />
The three locations in the west of the Netherl<strong>and</strong>s,<br />
WTP Lekkerkerk (since 1998), WTP De Put (1981-<br />
1988, <strong>and</strong> 1996-current) <strong>and</strong> WTP ‘t Kromme Gat<br />
(1983-2010), are operated by Oasen Drinking Water<br />
Company. Oasen operates subsurface aeration, a less<br />
intensive form of subsurface iron removal, since 1980.<br />
The operation of the injection-abstraction wells are not<br />
optimized to target iron removal, but the purpose of<br />
1. Aalten<br />
2. Corle<br />
3. De Pol<br />
4. Dinxperlo<br />
5. Gorssel<br />
6. Noordijk<br />
7. ‘Klooster<br />
8. Vorden<br />
9. Zutphen<br />
10. Eerbeek<br />
11. Lekkerkerk<br />
12. ’t Kromme Gat<br />
13. De Put<br />
subsurface aeration is the enhancement of nitrification<br />
in the dry biofilters (de Vet, 2011). After subsurface<br />
aeration the water is mixed with other sources before<br />
treatment. The total amount of injection water is only<br />
1% of the total abstracted raw water. The retention of<br />
iron, arsenic <strong>and</strong> phosphate is, in this case, a side effect<br />
observed during subsurface treatment for enhanced<br />
nitrification. As iron <strong>and</strong> other constituents are allowed<br />
to arrive at elevated concentrations in the well, the<br />
measurements at these locations provide information<br />
on the constituent behaviour during an injectionabstraction<br />
cycle.<br />
Influence of pH<br />
The pH of the groundwater influences the efficiency<br />
of subsurface iron removal both during injection<br />
(oxidation) <strong>and</strong> abstraction (adsorption). During<br />
injection the heterogeneous oxidation reaction is<br />
catalyzed at higher pH ranges (Tamura et al., 1980;<br />
Sung <strong>and</strong> Morgan, 1980):<br />
d<br />
[ Fe ]<br />
dt<br />
Equation 2.3<br />
[ S − OH ][ Fe ][ O ]<br />
[ H ]<br />
0 2+<br />
2+<br />
2<br />
hetero<br />
= − k<br />
2<br />
+<br />
Figure 2.4 Geographical locations of subsurface iron removal<br />
WTPs in the Netherl<strong>and</strong>s (Gorssel has been shut<br />
down)<br />
The value of k 2<br />
depends on the type of mineral<br />
structure, being lowest for the more crystalline iron<br />
(oxy)hydroxides (goethite, lepidocrocite). When<br />
28
2 <strong>Subsurface</strong> iron removal in the Netherl<strong>and</strong>s<br />
pH<br />
8.0<br />
7.5<br />
7.0<br />
6.5<br />
6.0<br />
0 5 10 15<br />
V/Vi<br />
WTP Corle<br />
van Beek, 1983<br />
Figure 2.5 Relation between pH <strong>and</strong> V/Vi ratio at the 12 SIR<br />
wells at WTP Corle <strong>and</strong> data from van Beek (1983);<br />
[O 2<br />
] of injection water was 0.28mM.<br />
pH<br />
7.8<br />
7.6<br />
7.4<br />
7.2<br />
7.0<br />
6.8<br />
5 10 15 20 25<br />
V/Vi<br />
Figure 2.6 Relation between pH <strong>and</strong> V/Vi ratio at 12 SIR wells<br />
at WTP Corle with [O 2<br />
] of injection water of 0.55mM<br />
the abstraction phase starts, the Fe 2+ -containing<br />
groundwater travels passed oxidized Fe 3+ oxide surfaces<br />
for adsorption. The adsorptive capacity of the soil<br />
depends on the type of Fe 3+ oxides present in the soil,<br />
as adsorption capacities for amorphous Fe 3+ oxides are<br />
larger than more crystalline mineral structures with<br />
lower surface areas. The adsorption of cations onto a<br />
surface is a result of two major forces: the chemical<br />
<strong>and</strong> the coulombic forces (Dzombak <strong>and</strong> Morel,<br />
1990; Stumm <strong>and</strong> Morgan, 1996). In case of an Fe 3+<br />
oxide surface the chemical forces are most dominant,<br />
resulting in higher adsorption capacities at higher pH<br />
(Sharma, 2001; Dixit <strong>and</strong> Hering 2006).<br />
Figure 2.5 shows the relation between the V/Vi<br />
ratio <strong>and</strong> the groundwater pH in 12 subsurface iron<br />
removal wells at WTP Corle. At the normal oxygen<br />
concentration in the injection water of 0.28 mmol.L -1<br />
at WTP Corle, the V/Vi rises from 3 to 9 between pH<br />
7.0 <strong>and</strong> 7.4. The field experiments by van Beek (1983)<br />
show a similar correlation for different subsurface iron<br />
removal locations, with a V/Vi of 2-3 at pH of 6.3. With<br />
injecting higher oxygen concentrations, namely 0.55<br />
mmol.L -1 , the difference between the V/Vi at pH 7.0<br />
<strong>and</strong> 7.4 is even more pronounced as the V/Vi rose from<br />
9 to 24 in this pH range (Figure 2.6). It may therefore be<br />
concluded that the pH has, as expected, an enormous<br />
effect on subsurface iron removal. These results<br />
illustrate that without the proper pH (approximately<br />
>7.0) the Fe 2+ oxidation <strong>and</strong>/or adsorption cannot be<br />
achieved sufficiently during SIR.<br />
29<br />
2
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
2<br />
Successive cycles<br />
As reported before (Boochs <strong>and</strong> Barovic, 1981; Braester<br />
<strong>and</strong> Martinell, 1988; Grombach, 1985; Hallberg <strong>and</strong><br />
Martinell, 1976; Mettler, 2002; Rott, 1985; van Beek,<br />
1985) the iron removal efficacy is known to improve<br />
with every injection-abstraction cycle. Figure 2.7<br />
illustrates that such an efficacy increase is also observed<br />
during the first cycles at WTP Lekkerkerk. Several<br />
mechanisms have been proposed to be responsible for<br />
this observation:<br />
• Incomplete Fe breakthrough enhances removal in future<br />
cycles (Appelo et al., 1999);<br />
• Pyrite oxidation during initial cycles;<br />
• Growth of iron oxidizing bacterial communities in the<br />
oxidation zone (Hallberg <strong>and</strong> Martinell, 1976);<br />
• Increased adsorptive surface area on the soil material<br />
[Fe] mmol.L -1<br />
through accumulation of precipitated iron oxides (Rott,<br />
1985; van Beek, 1985).<br />
0.12<br />
0.08<br />
0.04<br />
0.00<br />
0 2 4 6 8 10 12 14 16<br />
V/Vi<br />
Cycle 1<br />
Cycle 7<br />
Cycle 22<br />
Figure 2.7 <strong>Iron</strong> measurements during cycle 1, 7 <strong>and</strong> 22 at WTP<br />
Lekkerkerk<br />
A combination of above mechanisms may contribute<br />
to the improved efficacy of SIR <strong>and</strong> it is difficult<br />
to differentiate between them with field data.<br />
The first hypothesis, regarding the incomplete Fe<br />
breakthrough, is based on the assumption that lower<br />
Fe 2+ concentrations are present around a tube well at<br />
the end of the abstraction phase. These lowered Fe 2+<br />
concentrations are subsequently pushed into the aquifer<br />
during injection, displacing the original groundwater<br />
even further. During abstraction these lowered Fe 2+<br />
concentrations will load the oxidation zone to a lesser<br />
extent than the original groundwater, resulting in<br />
better retardation of Fe. As can be seen Figure 2.7, the<br />
subsurface iron removal well at WTP Lekkerkerk has<br />
been operated with complete breakthrough of Fe from<br />
the start. Nevertheless, the Fe removal efficacy increases<br />
with every cycle, right from the beginning.<br />
The second hypothesis may be responsible for<br />
that. The occurrence of pyrite oxidation (FeS 2<br />
), can be<br />
determined based on field data as sulphate is released in<br />
the reaction (Appelo <strong>and</strong> Postma, 2005):<br />
Equation 2.4<br />
15 7<br />
2<br />
FeS 4<br />
4 2<br />
− +<br />
2<br />
+ O2<br />
+ H<br />
2O<br />
→ Fe(<br />
OH )<br />
3<br />
+ 2SO4<br />
+ H<br />
This equation summarizes the oxidation of both<br />
disulfide <strong>and</strong> ferrous iron, <strong>and</strong> subsequent hydrolysis<br />
of Fe 3+ to ferrihydrite. The oxidation of pyrite results in<br />
the mobilization of sulphate <strong>and</strong> a drop in pH. Sulphate<br />
concentrations have been monitored during the first 18<br />
cycles at WTP Lekkerkerk <strong>and</strong> after SIR was temporary<br />
30
2 <strong>Subsurface</strong> iron removal in the Netherl<strong>and</strong>s<br />
[SO4] in mmol.L -1<br />
0.7<br />
0.6<br />
0.5<br />
0.4<br />
stop injection<br />
0 20 40 60 80<br />
abstracted volume (x10,000m 3 )<br />
Figure 2.8 Sulphate measurements during <strong>and</strong> after the<br />
operation of subsurface iron removal<br />
stopped (Figure 2.8). The measurements show a slight<br />
increase in sulphate during SIR, from 0.5 mmol.L -1<br />
at the end of a cycle to 0.6 mmol.L -1 , indicating the<br />
occurrence of sulphate releasing processes, such as<br />
pyrite oxidation. Once SIR had stopped, sulphate<br />
concentrations slowly reduced to their original value.<br />
When combining this information with the observed<br />
mobilization of arsenic, it may be assumed that also<br />
some pyrite-bound arsenic or arsenopyrite (FeAs S<br />
) was<br />
present in the targeted aquifer. It should be noted that<br />
arsenic concentrations did not exceed 0.2 µmol.L -1 <strong>and</strong><br />
after dry bios<strong>and</strong> filtration the concentrations were<br />
always below the provisional guideline of the World<br />
Health Organization (10 µg.L -1 ; 2006). Although iron<br />
is released from the pyrite as well, it is immediately<br />
oxidized <strong>and</strong> immobilized <strong>and</strong> will therefore not be<br />
found in elevated concentrations in the abstracted<br />
water. The pyrite oxidation process, however, limits the<br />
iron removal efficacy, as part of the oxygen is consumed<br />
for oxidation of disulfide (Equation 2.4).<br />
Co-removal of manganese,<br />
phosphate <strong>and</strong> arsenic<br />
Apart from subsurface iron removal subsurface<br />
manganese removal has also been operated in the<br />
Netherl<strong>and</strong>s <strong>and</strong> other European countries (Braester<br />
<strong>and</strong> Martinell, 1988; van Beek 1983) <strong>and</strong>, more recently,<br />
in Egypt (Olsthoorn, 2000). Van Beek (1983) reported<br />
V/Vi efficiencies for manganese removal of 3.5 <strong>and</strong> 12,<br />
<strong>and</strong> indicated that operation for manganese removal<br />
is more sensitive to operational variations than iron.<br />
Abiotic Mn 2+ oxidation is a slower reaction than Fe 2+<br />
oxidation <strong>and</strong> will start once Fe 2+ oxidation is more<br />
[Mn] mmol.L -1<br />
0.025<br />
0.020<br />
0.015<br />
0.010<br />
0.005<br />
0.000<br />
0 5 10 15<br />
V/Vi<br />
Cycle 1<br />
Cycle 7<br />
Cycle 22<br />
Figure 2.9 Manganese measurements during cycle 1, 7 <strong>and</strong> 22 at<br />
WTP Lekkerkerk<br />
31<br />
2
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
2<br />
[PO4 3- ] mmol.L -1<br />
0.05<br />
0.04<br />
0.03<br />
0.02<br />
0.01<br />
0.00<br />
Cycle 1<br />
Cycle 7<br />
Cycle 22<br />
0 5 10 15<br />
V/Vi<br />
Figure 2.10 Phosphate measurements during cycle 1, 7 <strong>and</strong> 22 at<br />
WTP Lekkerkerk<br />
[As] µmol.L -1<br />
0.20<br />
0.15<br />
0.10<br />
0.05<br />
0.00<br />
0 5 10 15<br />
V/Vi<br />
Cycle 1<br />
Cycle 7<br />
Cycle 22<br />
Figure 2.11 <strong>Arsenic</strong> measurements during cycle 1, 7 <strong>and</strong> 22 at<br />
WTP Lekkerkerk<br />
or less complete. Mn 4+ precipitates are found closer<br />
to the subsurface treatment well than Fe 3+ precipitates<br />
(Hallberg <strong>and</strong> Martinell, 1976; Rott, 1985; van Beek<br />
1985). The separation between iron <strong>and</strong> manganese<br />
deposits has also been observed in the dry filters at<br />
WTP Lekkerkerk (de Vet, 2011). In a completely<br />
different setting, namely in oxygen-diffusion around<br />
rice plants, the same Fe-Mn separation was found<br />
(Frommer et al., 2011). The manganese measurements<br />
at WTP Lekkerkerk during cycle 1, 7 <strong>and</strong> 22 are shown<br />
in Figure 2.9. <strong>Removal</strong> efficiencies increased slightly<br />
after multiple cycles, but not as distinctive as for iron.<br />
Anion co-removal during SIR has been reported<br />
in the literature before (Rott et al., 2002, Appelo <strong>and</strong><br />
de Vet, 2003); the measurements for phosphate <strong>and</strong><br />
arsenic are depicted in Figure 2.10 <strong>and</strong> Figure 2.11,<br />
respectively. Phosphate behaviour correlates well to<br />
the removal efficacy of iron, showing significant lower<br />
concentrations during cycle 22 than cycle 1 <strong>and</strong> 7. For<br />
arsenic the breakthrough curves are not as uniform,<br />
<strong>and</strong> during the first cycles arsenic concentrations do<br />
not seem to be very stable. It even seems as if arsenic<br />
is mobilized during cycle 1 <strong>and</strong> 7, whereas that process<br />
does not seem relevant anymore during cycle 22. An<br />
explanation for the release of arsenic during early cycles<br />
may be that the oxidized pyrite was arsenic-bearing,<br />
resulting in elevated arsenic concentrations in solution.<br />
This has also been observed during an experiment with<br />
injection of aerated water into a deep tube well at the<br />
nearby located Langerak (Wallis et al., 2010).<br />
Contaminant remobilization<br />
Fe 3+ oxides will accumulate in the aquifer around a<br />
tube well when subsurface iron removal is operated for<br />
several years (Mettler et al., 2001; van Halem et al., 2011).<br />
Although clogging of the aquifer has not been reported,<br />
32
2 <strong>Subsurface</strong> iron removal in the Netherl<strong>and</strong>s<br />
there is still a concern about what happens to the<br />
accumulated deposits when operation is stopped. The<br />
aquifer will return to its anoxic or anaerobic state <strong>and</strong><br />
reducing conditions will prevail. Reductive dissolution<br />
of the deposits may promote the mobilization of iron,<br />
manganese or even trace compounds, such as arsenic<br />
(Appelo <strong>and</strong> Postma, 2005):<br />
Equation 2.5<br />
[Fe], [Mn] in mmol.L -1<br />
0.1<br />
0.08<br />
0.06<br />
0.04<br />
0.02<br />
0<br />
0 100,000 200,000 300,000 400,000<br />
2<br />
CH O<br />
4<br />
2<br />
+ −<br />
+ H<br />
2O<br />
→ CO2<br />
+ 4H<br />
+ e<br />
Equation 2.6<br />
+ − 2+<br />
Fe ( OH ) As + 3H<br />
+ e → Fe + H O + As<br />
3<br />
3<br />
Although this is a valid concern, especially at sites<br />
with high organic carbon content in the soil <strong>and</strong>/or<br />
groundwater, the mobilization of iron has not been<br />
observed to exceed its background value after stopping<br />
with subsurface iron removal (Figure 2.12). At WTP<br />
Lekkerkerk, the iron <strong>and</strong> manganese concentrations<br />
have been measured at the SIR well before <strong>and</strong> after<br />
the well was exposed to periodic injection. Dissolved<br />
organic carbon (DOC) concentrations in the abstracted<br />
groundwater fluctuated around 3 mg.L -1 (±0.5). This<br />
particular SIR well had been in operation for 18<br />
injection-abstraction cycles, <strong>and</strong> Figure 2.12 illustrates<br />
the results from the start of cycle 18. The measurements<br />
show an initial increase in Fe concentration, but after<br />
140,000 m 3 (approx. 10 months) the concentration<br />
stabilized at 0.08 mmol.L -1 . This is lower than the iron<br />
concentrations observed during cycle 1 <strong>and</strong> 7 (Figure<br />
2<br />
Pumped volume (m 3 )<br />
<strong>Iron</strong><br />
Manganese<br />
Figure 2.12 <strong>Iron</strong> <strong>and</strong> manganese measurements once subsurface<br />
iron removal had stopped after 18 injectionabstraction<br />
cycles (WTP Lekkerkerk)<br />
2.7), which supports the theory that pyrite oxidation<br />
may have occurred during the initial cycles. Manganese<br />
concentrations also increased initially, but stabilized<br />
at 0.015 mmol.L -1 , which corresponds closely to the<br />
background concentration in Figure 2.9. Based on<br />
these findings it can be concluded that accumulated<br />
iron <strong>and</strong> manganese oxides during subsurface iron<br />
removal were not remobilized to result in elevated<br />
concentrations to the groundwater at WTP Lekkerkerk<br />
once SIR operation had stopped.<br />
<strong>Iron</strong> or manganese remobilization has not been<br />
observed in the >2 years period that subsurface iron<br />
removal was stopped at WTP Lekkerkerk. Nevertheless,<br />
trace (heavy) metals incorporated or adsorbed to<br />
the Fe/Mn oxide matrix could be released through<br />
desorption, exchange or re-crystalization (Jessen et al.,<br />
2005). Desorption may occur if groundwater conditions<br />
33
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
2<br />
[PO 4<br />
-3 ] in mmol.L-1<br />
0.05<br />
0.04<br />
0.03<br />
0.02<br />
0.01<br />
0<br />
0<br />
0 100,000 200,000 300,000 400,000<br />
0.1<br />
0.08<br />
0.06<br />
0.04<br />
0.02<br />
[As] in µmol.L -1<br />
did not exceed 0.01 mmol.L -1 , which is far below the<br />
concentrations measured during cycle 1 (Figure 2.10).<br />
<strong>Arsenic</strong> concentrations remained below 0.1 µmol.L -1<br />
<strong>and</strong> were thus also not found to remobilize after SIR<br />
was stopped at WTP Lekkerkerk.<br />
Pumped volume (m 3 )<br />
Conclusions<br />
Phosphate <strong>Arsenic</strong><br />
Figure 2.13 Phosphate <strong>and</strong> arsenic measurements once subsurface<br />
iron removal had stopped after 18 injectionabstraction<br />
cycles (WTP Lekkerkerk)<br />
suddenly change (e.g., pH), which is not likely in a<br />
normal well field. Exchange of the adsorbed anion with<br />
other passing anions with higher affinity could promote<br />
the release of retained arsenic or phosphate from the soil<br />
grain surface. However, once groundwater conditions<br />
stabilize around the stopped SIR well no large water<br />
quality variations are expected. Re-crystallization<br />
of (amorphous) iron or manganese oxides to oxides<br />
of higher crystallinity is a slow process, but could<br />
release small amounts of trace elements. Although<br />
Figure 2.13 shows some variations in the arsenic <strong>and</strong><br />
phosphate concentrations, it is obvious that high peak<br />
concentrations are not observed. It may be concluded<br />
that the retained phosphate was not mobilized to<br />
elevated solute concentrations once injection was no<br />
longer applied. Phosphate concentrations generally<br />
During subsurface iron removal at WTP Lekkerkerk,<br />
iron, manganese, phosphate <strong>and</strong> arsenic were measured<br />
to be retained in the aquifer at varying efficacies. SIR<br />
technology was found to improve in efficacy with<br />
every successive cycle, which correlates well with<br />
findings in other literature. No peak concentrations<br />
of the studied constituents were observed in the<br />
abstracted groundwater once SIR was stopped <strong>and</strong><br />
therefore not threatening the sustainability of the<br />
groundwater production well. pH was shown to be the<br />
most pronounced water quality parameter determining<br />
the efficacy of SIR, the operational V/Vi ratio more<br />
than doubled between pH 7.0 <strong>and</strong> 7.4 at WTP Corle.<br />
The experiences in the Netherl<strong>and</strong>s have shown<br />
that subsurface iron removal is an effective, robust<br />
<strong>and</strong> sustainable iron removal technology with great<br />
potential for worldwide application.<br />
34
2 <strong>Subsurface</strong> iron removal in the Netherl<strong>and</strong>s<br />
References<br />
Appelo C. A. J., B. Drijver, R. Hekkenberg <strong>and</strong> M. de Jonge<br />
(1999) Modeling in situ iron removal from ground water,<br />
Ground Water 37(6): 811-817.<br />
Appelo C. A. J. <strong>and</strong> W. W. J. M. de Vet (2003) Modeling in situ<br />
iron removal from groundwater with trace elements such<br />
as As. In <strong>Arsenic</strong> in groundwater. A. H. Welch <strong>and</strong> K.G.<br />
Stollenwerk. Kluwer Academic, Boston.<br />
Appelo C.A.J. <strong>and</strong> D. Postma (2005) Geochemistry, groundwater<br />
<strong>and</strong> pollution. Balkema, Rotterdam, 2nd edition.<br />
Boochs P. W. <strong>and</strong> G. Barovic (1981) Numerical-Model Describing<br />
Groundwater Treatment by Recharge of Oxygenated Water,<br />
Water Resources Research 17(1): 49-56.<br />
Braester C, <strong>and</strong> R. Martinell (1988) The Vyretox <strong>and</strong> Nitredox<br />
methods of in situ treatment of groundwater, Water Science<br />
<strong>and</strong> Technology 20: 149-163.<br />
Czekalla, C., W. Mevius <strong>and</strong> H. Hanert (1985) Quantative<br />
removal of iron <strong>and</strong> manganese by microorganisms in rapid<br />
s<strong>and</strong>filters (in situ investigations). Water Supply 3(1): 111-<br />
123.<br />
de Vet W.W.J.M., I.J.T. Dinkla, G. Muyzer, L.C. Rietveld <strong>and</strong><br />
M.C.M. van Loosdrecht (2009) Molecular characterization<br />
of microbial populations in groundwater sources <strong>and</strong> s<strong>and</strong><br />
filters for drinking water production, Water Research 43(1):<br />
182-194.<br />
de Vet, W.W.J.M. (2011) Biological drinking water treatment<br />
of anaerobic groundwater trickling filters, PhD dissertation,<br />
Delft University of Technology.<br />
Dixit S. <strong>and</strong> J. G. Hering (2006) Sorption of Fe(II) <strong>and</strong> As(III)<br />
on goethite in single- <strong>and</strong> dual-sorbate systems, Chemical<br />
Geology 228(1-3): 6-15.<br />
Dzombak D. A. <strong>and</strong> F.M.M. Morel (1990) Surface complexation<br />
modeling: hydrous ferric oxide, Wiley.<br />
Frommer J., A. Voegelin, J. Dittmar, M.A. Marcus <strong>and</strong> R.<br />
Kretzschmar (2011) Biogeochemical processes <strong>and</strong> arsenic<br />
enrichment around rice roots in paddy soil: results from<br />
micro-focused X-ray spectroscopy, European Journal of Soil<br />
Science 62: 305-317.<br />
Grombach, P. (1985) Groundwater treatment in situ in the<br />
aquifer. Water Supply 3(1): 13-18.<br />
Hallberg R. O. <strong>and</strong> R. Martinell (1976) Vyredox - in situ<br />
purification of groundwater, Ground Water 14(2): 88-93.<br />
Hiemstra T. <strong>and</strong> W.H. van Riemsdijk (2007) Adsorption <strong>and</strong><br />
surface oxidation of Fe(II) on metal (hydr)oxides. Geochim.<br />
Cosmochim. Acta 71: 5913-5933.<br />
Jessen S., F. Larsen, C.B. Koch <strong>and</strong> E. Arvin (2005) Sorption<br />
<strong>and</strong> Desorption of <strong>Arsenic</strong> to Ferrihydrite in a S<strong>and</strong> Filter,<br />
Environ. Sci. Technol 39: 8045-8051.<br />
Mettler S., M. Abdelmoula, E. Hoehn, R. Schoenenberger, P.<br />
Weidler <strong>and</strong> U. von Gunten (2001) Characterization of iron<br />
<strong>and</strong> manganese precipitates from an in situ groundwater<br />
treatment plant, Ground Water 6: 921-930.<br />
Mettler, S. (2002) In-situ removal of iron from groundwater:<br />
Fe(II) oxygenation, <strong>and</strong> precipitation products in a<br />
calcareous aquifer, PhD dissertation, Swiss Federal Institute<br />
of Technology, Zurich.<br />
Mouchet P. (1992) From conventional to biological removal of<br />
iron <strong>and</strong> manganese in France, Journal AWWA 84(4): 158-<br />
167.<br />
35<br />
2
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
2<br />
Olsthoorn T.N. (2000) Background of subsurface iron <strong>and</strong><br />
manganese removal, Amsterdam Water Supply Research <strong>and</strong><br />
Development, Hydrology Department.<br />
Olthoff R. (1986) Die Enteisenung und Entmanganung von<br />
Grundwasser im Aquifer, PhD thesis Hannover University.<br />
Rott U. (1985) Physical, chemical <strong>and</strong> biological aspects of the<br />
removal of iron <strong>and</strong> manganese underground, Water Supply<br />
3: 143-150.<br />
Rott U., C. Meyer <strong>and</strong> M. Friedle (2002) Residue-free removal<br />
of arsenic, iron, manganese <strong>and</strong> ammonia from groundwater,<br />
Water Sci Technol: Water Supply 2(1): 17-24.<br />
Sharma, S.K. (2001) Adsorptive iron removal from groundwater.<br />
PhD dissertation, IHE Delft.<br />
Sharma S.K., J. Kappelhof, M. Groenendijk <strong>and</strong> J.C. Schippers<br />
(2001) Comparison of physiochemical iron removal<br />
mechanisms in filters, J. water Supply Res. Techn. Aqua 50(4),<br />
187-198.<br />
Smeets, P.W.M.H., G.J. Medema <strong>and</strong> J.C. van Dijk (2009) The<br />
Dutch secret: how to provide safe drinking water without<br />
chlorine in the Netherl<strong>and</strong>s, Drink. Water Eng. Sci., 2: 1–14.<br />
Stumm, W. <strong>and</strong> J.J. Morgan (1996) Aquatic chemistry: chemical<br />
equilibrium <strong>and</strong> rates in natural waters. 3rd edition, Wiley<br />
Interscience, New York.<br />
Sung W. <strong>and</strong> J.J. Morgan (1980) Kinetics <strong>and</strong> products of ferrous<br />
iron oxygenation in aqueous systems, Environmental Science<br />
& Technology 14: 561-568.<br />
Tamura, H., S. Kawamura <strong>and</strong> M. Hagayama (1980) Acceleration<br />
of the oxidation of Fe 2+ ions by Fe(III)-oxyhhydroxides,<br />
Corrosion Science 20: 963-971.<br />
van Beek C.G.E.M. (1983) Ondergrondse ontijzering, een<br />
evaluatie van uitgevoerd onderzoek (in Dutch). KIWA<br />
mededeling 78.<br />
van Beek C.G.E.M. (1985) Experiences with underground water<br />
treatment in the Netherl<strong>and</strong>s. Water Supply 3: 1-11.<br />
van Halem D., S. Olivero, W.W.J.M de Vet, J.Q.J.C Verbek, G.L.<br />
Amy <strong>and</strong> J.C. van Dijk (2010) <strong>Subsurface</strong> iron <strong>and</strong> arsenic<br />
removal for shallow tube well drinking water supply in rural<br />
Bangladesh. Water Research 44: 5761-5769.<br />
van Halem, D., W.W.J.M. de Vet, , J.Q.J.C Verbek, G.L. Amy<br />
<strong>and</strong> J.C. van Dijk (2011) Characterization of accumulated<br />
precipitates during subsurface iron removal, Applied<br />
Geochemistry 26: 116–124.<br />
von Oesten, G. (1900) Verfahren zur Enteisenung von<br />
Grundwasser im Untergrund selbst, Patentschrift Nr. 114709,<br />
Kaiserliches Patentamt Berlin.<br />
Wallis I., H. Prommer, C.T. Simmons, V. Post, <strong>and</strong> P.J. Stuyfz<strong>and</strong><br />
(2010) Evaluation of Conceptual <strong>and</strong> Numerical Models<br />
for <strong>Arsenic</strong> Mobilization <strong>and</strong> Attenuation during Managed<br />
Aquifer Recharge, Environ. Sci. Technol. 44: 5035–5041.<br />
36
3<br />
Small-scale subsurface iron <strong>and</strong><br />
arsenic removal in Bangladesh<br />
This chapter is based on:<br />
van Halem et al. (2010) Water Science <strong>and</strong> Technology (62): 2702-2709
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
3<br />
Introduction<br />
The most well-known <strong>and</strong> severe case of arsenic<br />
poisoning through drinking water is currently ongoing<br />
in Bangladesh. Two-thirds of the tube wells installed<br />
over the last three decades, roughly three million<br />
in total, have been reported to produce arsenic in<br />
concentrations above the permissible level of 10<br />
µg.L -1 set by the World Health Organization (BGS/<br />
DPHE, 2001). It is estimated that 35 to 100 million<br />
people are at risk of drinking arsenic-contaminated<br />
water in Bangladesh above the WHO guideline (BGS/<br />
DPHE, 2001; WHO, 2001; Chowdhury et al., 2006).<br />
In a bulletin (Smith et al., 2000), the World Health<br />
Organization reports that long-term exposure to<br />
arsenic in groundwater, at concentrations over 500<br />
µg.L -1 , causes death in 1 in 10 adults (including lung,<br />
bladder <strong>and</strong> skin cancers). The tube wells were installed<br />
with the firm conviction that they would contribute to<br />
a secure <strong>and</strong> reliable drinking water supply, in order<br />
to put an end to various contagious diseases caused<br />
by the use of (microbial unsafe) surface water. It is a<br />
bitter observation that it is this very approach that<br />
has led to widespread arsenic poisoning through<br />
drinking water. The large well-to-well variability in<br />
arsenic concentrations bears the consequence that in<br />
the villages all wells need to be tested. Currently an<br />
estimated 5 million wells have been tested <strong>and</strong> in 1.4<br />
million wells the arsenic concentration was measured<br />
above the national st<strong>and</strong>ard of 50 µg.L -1 . In Bangladesh,<br />
the source of arsenic in groundwater is deposited<br />
sediments from the Himalayas <strong>and</strong> strongly reducing<br />
conditions cause reductive dissolution of the arsenicrich<br />
iron hydroxides (Smedley <strong>and</strong> Kinniburgh, 2002).<br />
In the reducing environment arsenic predominantly<br />
occurs as arsenite or As(III), when oxidising conditions<br />
prevail the dominant species is generally arsenate or<br />
As(V).<br />
In industrialized countries arsenic contamination<br />
is also a recognized problem (van Halem et al, 2009).<br />
The removal of arsenic is achieved through commercial<br />
adsorption media, e.g. GFH (granular ferric hydroxide),<br />
or by means of coagulation/filtration, ion exchange or<br />
membrane filtration. These methods are expensive<br />
<strong>and</strong> therefore unavailable to the poor living in rural<br />
areas. Household water treatment systems with noncommercially<br />
available adsorption media, such as<br />
Sono <strong>and</strong> Alcan, show potential (Sutherl<strong>and</strong> et al.,<br />
2002). However, further research <strong>and</strong> documentation is<br />
needed on the long-term efficacy <strong>and</strong> potential risk of<br />
microbial contamination of the treated water. All filters<br />
will at one stage either clog <strong>and</strong>/or show breakthrough<br />
of arsenic. Some filters can be regenerated; however,<br />
many leave an arsenic-rich waste stream. The treatment<br />
option presented in this study can be operated without a<br />
waste stream, because the iron <strong>and</strong> arsenic are retained<br />
in the subsurface. The adsorption sites for arsenic are<br />
generated in the aquifer, providing a low-cost <strong>and</strong><br />
robust subsurface filter.<br />
38
3 Small-scale subsurface iron <strong>and</strong> arsenic removal in Bangladesh<br />
The principle of subsurface iron <strong>and</strong> arsenic removal is<br />
that aerated water is periodically injected into an anoxic<br />
aquifer through a tube well (Figure 3.1, left), partially<br />
displacing the groundwater. The injected water oxidises<br />
adsorbed Fe 2+ on the soil grains, resulting in adsorptive<br />
surface area of iron hydroxides. When the flow is<br />
reversed, soluble Fe 2+ in the abstracted groundwater<br />
is adsorbed onto the Fe 3+ oxide coated soil grains <strong>and</strong><br />
water with reduced iron concentrations is abstracted<br />
(Figure 3.1, right). Injection is started again once<br />
elevated iron levels arrive at the well. More water with<br />
reduced iron concentrations can be abstracted (volume<br />
V) than was injected (volume Vi), i.e., this volumetric<br />
ratio (V/Vi) determines the efficiency of the system.<br />
The affected area in the subsurface around the tube well<br />
is referred to as the oxidation zone.<br />
<strong>Subsurface</strong> or in-situ iron removal has been used<br />
in central Europe for many decades (Hallberg <strong>and</strong><br />
Martinell, 1976; Rott, 1985; van Beek, 1985; Appelo<br />
et al., 1999; Mettler, 2002), but the application of<br />
subsurface treatment for the removal of arsenic from<br />
groundwater is a relatively new approach (Sarkar <strong>and</strong><br />
Rahman, 2001; Rott et al., 2002, van Halem et al., 2008).<br />
<strong>Arsenic</strong> is known to adsorb to Fe 3+ oxides (Dzombak<br />
<strong>and</strong> Morel, 1990), <strong>and</strong> since these are formed during<br />
subsurface treatment, arsenic can be retained in the<br />
subsurface. This technology has the potential to be a<br />
cost-effective way to provide safe drinking water in<br />
rural areas in decentralized applications. Existing<br />
shallow tube wells can be modified to be operated<br />
under infiltration <strong>and</strong> abstraction conditions. Such a<br />
small-scale set-up has been the focus of an earlier study<br />
in Bangladesh (Maijdee area), where 0.5m 3 of aerated<br />
3<br />
(a) injection phase<br />
(b) abstraction phase<br />
Ground water level<br />
containing Fe 2+ <strong>and</strong> As(III)<br />
Ground water level<br />
containing Fe 2+ <strong>and</strong> As(III)<br />
O2 front<br />
iron oxide<br />
adsorbed iron oxideFe 2+<br />
<strong>and</strong> As(III)<br />
with adsorbed<br />
Fe 2+ <strong>and</strong> As(III)<br />
Injected water front<br />
Figure 3.1 Small-scale subsurface iron <strong>and</strong> arsenic removal in Bangladesh<br />
39
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
3<br />
water was injected into the shallow aquifer (Sarkar<br />
<strong>and</strong> Rahman, 2001). This study was executed at three<br />
different sites, with As concentrations ranging from 1.5<br />
to 17 μmol.L -1 <strong>and</strong> molar Fe:As ratios between 1 <strong>and</strong> 12.<br />
The design (Figure 3.2) consisted of a regular shallow<br />
tube well (screen depth 10-12m) <strong>and</strong> a storage tank with<br />
a plate aerator (up to 0.16 mmol O 2<br />
.L -1 ). After injection,<br />
the tube well was left undisturbed for 12 hours before<br />
abstraction was started. Though As concentrations were<br />
significantly lowered, concentrations below the WHO<br />
guideline were not reported. C/C 0<br />
=0.5 was reached<br />
at V/Vi=3 for all studied sites. A similar study was<br />
executed on the other side of the Indian border with 6<br />
small-scale SAR plants (Sen Gupta et al., 2009). In this<br />
study injection volumes were higher at 2m 3 <strong>and</strong> aeration<br />
was achieved with sprinklers/shower heads (Figure<br />
3.3). <strong>Arsenic</strong> concentrations varied between 1.2 <strong>and</strong><br />
3.8 μmol.L -1 <strong>and</strong> molar Fe:As ratios range from 12 to 50<br />
(SAR Technology, 2011). The authors reported that As<br />
concentrations to below 0.027 μmol.L -1 were achieved<br />
at all sites, but it should be noted that volumetric ratios<br />
did not exceed V/Vi=2. Other researchers have studied<br />
the co-removal of As during SIR in Europe <strong>and</strong> found<br />
a reduction of As from < 0.5 μmol.L -1 to below the<br />
WHO guideline (Rott et al. 2002; Appelo <strong>and</strong> De Vet,<br />
2003). Miller (2006) combined the pulsed injection of<br />
aerated water <strong>and</strong> ferrous iron to remove As(V).Welch<br />
et al. (2000) investigated subsurface or in-situ arsenic<br />
removal with a combination of aerated water <strong>and</strong> the<br />
injection of ferric chloride. Preliminary results showed<br />
reduction of 1.3 μmol.L -1 As(V) to below the WHO<br />
guideline of 0.13 μmol.L -1 (10 μg.L -1 ; WHO, 2006).<br />
Previous studies have shown the potential<br />
of small-scale subsurface iron <strong>and</strong> arsenic removal,<br />
from tube well<br />
Aeration tray<br />
from tube well<br />
Shower aeration<br />
V i = 2m 3 V i = 2m 3<br />
to tube well<br />
V i = 0.5m 3<br />
iron sludge<br />
to tube well<br />
Injection tank<br />
Storage tank<br />
Figure 3.2 Injection facilities of Sarkar <strong>and</strong> Rahman (2001)<br />
Figure 3.3 Injection facilities of Sen Gupta et al.(2009)<br />
40
3 Small-scale subsurface iron <strong>and</strong> arsenic removal in Bangladesh<br />
but have also indicated that the success of this<br />
technology may be site-specific. For this technology<br />
to be considered a safe arsenic mitigation solution it<br />
is inevitable that more knowledge needs to be gained<br />
about the operational conditions that control the<br />
efficacy of SIR/SAR. The objective of this study was to<br />
investigate the performance of small-scale SIR/SAR in<br />
rural Bangladesh.<br />
Materials <strong>and</strong> methods<br />
Household shallow tube wells with suction h<strong>and</strong> pumps<br />
are widely distributed in Bangladesh <strong>and</strong> the aim of the<br />
subsurface iron/arsenic removal technology is to use<br />
the existing infrastructure optimally. The Manikganj<br />
district, just west of Dhaka, was selected for this study,<br />
since this area is known to have high iron, arsenic <strong>and</strong><br />
manganese concentrations in the groundwater. After<br />
testing numerous wells, two sites (A <strong>and</strong> B) were selected<br />
with the same arsenic concentrations (1.94 μg.L -1 ). The<br />
groundwater pH at the two test locations was 6.85 <strong>and</strong><br />
the groundwater composition is illustrated in Table<br />
3.1. The initial iron concentrations at location A <strong>and</strong><br />
B was 0.018 mmol.L -1 <strong>and</strong> 0.27 mmol.L -1 , respectively,<br />
resulting in molar Fe:As ratios of 9 <strong>and</strong> 140. Manganese<br />
concentrations also differed at the two locations,<br />
namely, (A) 0.045 mmol.L -1 <strong>and</strong> (B) 0.006 mmol.L -1 .<br />
The experimental set-up (Figure 3.4) was<br />
connected to an existing h<strong>and</strong> pump with tube well in<br />
the upper aquifer. The 1.5-inch tube-well had a depth<br />
of 31 meters <strong>and</strong> a perforated well length of 3 meter.<br />
As an added precaution, the set-up was placed with<br />
families who already had arsenic treatment since 2001<br />
(SIDKO system; BETV-SAM, 2011). For the purpose<br />
of subsurface treatment, the existing situation was<br />
modified with a pipe <strong>and</strong> valve for injection. After<br />
subsurface treatment, the groundwater was pumped<br />
(electrical suction pump) into the SIDKO system for<br />
aeration, s<strong>and</strong> filtration <strong>and</strong> Granular Ferric Hydroxide<br />
filtration (AdsorpAs, Harbauer GmbH). The treated<br />
water, low in arsenic (
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
3<br />
water meter<br />
pump<br />
valve<br />
inline<br />
monitoring<br />
sampling<br />
tap<br />
iron<br />
filter<br />
aeration<br />
tank<br />
storage<br />
tank<br />
SIDKO<br />
system<br />
arsenic<br />
filter<br />
oxygen concentrations were reached by aerating the<br />
water with a combination of a shower head (installed<br />
in the SIDKO system) <strong>and</strong> subsequent bubble aeration<br />
in the storage tank. Analysis of the water samples<br />
was done with field test kits (Wagtech International:<br />
Palintest <strong>and</strong> Arsenator) <strong>and</strong> confirmed in the<br />
laboratory (Perkin-Elmer Flame AAS 3110; Perkin-<br />
Elmer GF-AAS 5100PC). Duplicates or triplicates were<br />
taken to check the method of sampling <strong>and</strong> accuracy<br />
of analysis. <strong>Arsenic</strong> speciation was done with a field<br />
method (Clifford et al., 2004) using anion exchange<br />
resin columns (Amberlite IRA400). Multimeters<br />
(HACH 340i) were fixed inline to the experimental setup<br />
to monitor pH (WTW SenTix 41), dissolved oxygen<br />
(WTW Cellox 325), ORP (WTW SenTix ORP) <strong>and</strong><br />
electric conductivity (TetraCon 325). Measurements<br />
were registered on a computer with Multilab Pilot<br />
v5.06 software. The injection <strong>and</strong> abstraction volumes<br />
were monitored using water meters. Operation started<br />
in October 2008, just after the monsoon season, <strong>and</strong><br />
continued until May 2009. At both locations, the<br />
families shared their arsenic treatment facility with<br />
the community <strong>and</strong> the weekly water consumption<br />
was 5.7-8.7 m 3 <strong>and</strong> 2.4-2.9 m 3 for location A <strong>and</strong> B,<br />
respectively. Operational conditions, such as injection<br />
frequency <strong>and</strong> production discharge, varied due to<br />
irregular operation. Normally the set-ups were used<br />
for the families’ water production, however, during<br />
research periods the operation was intensified. In this<br />
paper, all results are presented using the ratio between<br />
abstracted volume (V) <strong>and</strong> injected volume (Vi), V/<br />
Vi. This volumetric ratio determines the efficiency of<br />
the system. Every period of injection-abstraction starts<br />
at V/Vi=0 <strong>and</strong> is referred to as a cycle, with the first<br />
injection-abstraction period being cycle 1. Injection<br />
was done the night before, so abstraction was started at<br />
12-16 hours after injection.<br />
Figure 3.4 Experimental set-up for subsurface iron <strong>and</strong> arsenic<br />
removal at existing SIDKO system<br />
42
3 Small-scale subsurface iron <strong>and</strong> arsenic removal in Bangladesh<br />
Results <strong>and</strong> discussion<br />
Typical breakthrough curve<br />
The graph in Figure 3.5 presents a typical breakthrough<br />
curve after injection. The electrical conductivity of<br />
the injection water was slightly lower than of the<br />
groundwater, <strong>and</strong> gives a good indication when the<br />
groundwater water arrives in the well <strong>and</strong> can thus be<br />
considered a conservative tracer. This was, as expected,<br />
at the moment when the volume of produced water was<br />
equal to the injected volume for this particular cycle,<br />
i.e., C/C 0<br />
=0.5 at around V/Vi=1. Although the tracer<br />
arrived at the well at this point, it is noteworthy that it<br />
took multiple V/Vi units before complete breakthrough<br />
C/C0<br />
1<br />
0.5<br />
0<br />
0 2 4 6 8<br />
V/Vi<br />
iron<br />
tracer<br />
oxygen<br />
arsenic<br />
Figure 3.5 Typical breakthrough curve for a tracer (electrical<br />
conductivity), dissolved oxygen, iron <strong>and</strong> arsenic.<br />
Cycle 6 at location B with an injection volume of Vi =<br />
744L<br />
was observed. The tailing of this curve suggests that<br />
injection water remains in the subsurface after V/Vi=1<br />
<strong>and</strong> a mixture of groundwater <strong>and</strong> injection water is<br />
withdrawn. It has been suggested before that stagnant<br />
zones, or “dead-end pores” in the oxidation zone<br />
influence the subsurface removal of iron (Rott et al.,<br />
2002). Diffusion between these stagnant zones <strong>and</strong> the<br />
passing groundwater delays the complete breakthrough<br />
of a constituent. This would mean that although the<br />
majority of the injected water volume arrived at the well<br />
before V/Vi=1.5, there was still a small percentage in<br />
the stagnant zones at that moment <strong>and</strong> equilibrium was<br />
only found after V/Vi=5. The dissolved oxygen curve did<br />
not show similar tailing, since oxygen concentrations<br />
reduced immediately after the start of abstraction to
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
3<br />
this a mass balance can be calculated <strong>and</strong>, for this cycle,<br />
was calaculted to be an approximate removal of 0.66<br />
mmol <strong>and</strong> 0.77 µmol for iron <strong>and</strong> arsenic, respectively.<br />
One isolated cycle does however not describe the<br />
potential of subsurface iron <strong>and</strong> arsenic removal, since<br />
this technology is known for its increasing efficacy with<br />
every successive cycle.<br />
Influence of Fe:As ratio in the groundwater<br />
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal is applied directly<br />
at the shallow tube well <strong>and</strong> therefore depends greatly<br />
on the occurring groundwater geochemistry. One<br />
of the parameters that may enhance the effective coremoval<br />
of arsenic with the iron is the presence of<br />
significant iron levels in the groundwater. Figures 3.6<br />
<strong>and</strong> 3.7 show the breakthrough of arsenic <strong>and</strong> iron for<br />
cycle 6 at locations A <strong>and</strong> B. At location A, the initial<br />
iron concentration was on average 0.018 mmol.L -1<br />
(molar Fe:As = 9) <strong>and</strong> the dominant arsenic species<br />
in the groundwater was As(III), measured with anion<br />
exchange resin to be 80-83% of total arsenic. Under<br />
anoxic conditions, with a pH of 6.8 <strong>and</strong> temperature<br />
of 25°C, arsenic predominantly occurs as H 3<br />
AsO 3<br />
in<br />
the aquatic environment (Ferguson <strong>and</strong> Gavis, 1972).<br />
The injection water (657L for cycle 6) contained >90%<br />
arsenic(V) in concentrations below 0.1 μmol.L. -1 in all<br />
cycles. <strong>Subsurface</strong> iron removal at location A shows for<br />
cycle 6 that after V/Vi=6 the measured concentration<br />
was still 65% lower than the initial iron level. Although<br />
iron levels are significantly reduced in the subsurface,<br />
(C/C0)Fe<br />
1<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0<br />
location A<br />
location B<br />
0 1 2 3 4 5 6 7 8<br />
V/Vi<br />
Figure 3.6 Measurements of iron for cycle 6 at locations A<br />
(molar Fe:As = 9, Vi= 657L) <strong>and</strong> B (molar Fe: As =<br />
140, Vi=744L)<br />
(C/C0)As<br />
1<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0<br />
0 1 2 3 4 5 6 7 8<br />
V/Vi<br />
location A<br />
location B<br />
Figure 3.7 Measurements of arsenic for cycle 6 at locations A<br />
(molar Fe:As = 9, Vi= 657L) <strong>and</strong> B (molar Fe: As =<br />
140, Vi=744L)<br />
44
3 Small-scale subsurface iron <strong>and</strong> arsenic removal in Bangladesh<br />
only small amounts of arsenic were removed from the<br />
groundwater. At V/Vi=2.5, an arsenic concentration<br />
of C/C 0<br />
=0.8 had been reached, which exceeds more<br />
than 10 times the WHO guideline of 10 µg.L -1 . It is<br />
noteworthy that the original arsenic concentration<br />
was not reached within the research period of V/Vi=6,<br />
potentially due to the long lasting effects of diffusion in<br />
the stagnant zones.<br />
At location B, the iron concentration was<br />
significantly higher at 0.27 mmol.L -1 (molar Fe:As =<br />
140), which could potentially result in better arsenic<br />
removal. Figure 3.6 shows that iron breakthrough was<br />
faster at this location, which was expected since the<br />
same amount of oxygen was injected at both locations.<br />
The breakthrough of arsenic is comparable with location<br />
A (Figure 3.7), <strong>and</strong> although the curve shows some<br />
tailing, it is clear that the arsenic concentration rose<br />
more or less simultaneously with the groundwater front<br />
(C/C0)Mn<br />
1<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0<br />
0 2 4 6 8<br />
V/Vi<br />
Figure 3.8 Manganese measurements during cycle 6 at location<br />
A (Vi= 657L)<br />
at V/Vi=1. This strongly indicates that the occurrence<br />
of high iron concentrations in the groundwater does<br />
not promote the effective co-removal of arsenic during<br />
subsurface treatment. The formed adsorptive iron<br />
hydroxide surface area is apparently either insufficiently<br />
available for arsenic adsorption, or other constituents,<br />
such as competing anions, are occupying the available<br />
adsorption sites. The removal of manganese during<br />
subsurface treatment was not an objective of this study,<br />
but during cycle 6 this inorganic contaminant was also<br />
measured at location A. Figure 3.8 illustrates that Mn<br />
breakthrough occurs immediately upon abstraction<br />
<strong>and</strong> that after V/Vi=6 the background concentration of<br />
0.046 mmol.L -1 was reached.<br />
Oxygen level of injection water<br />
Theoretically, 1 mole of oxygen can oxidise 4 moles<br />
of Fe 2+ . It might seem logical to assume that the more<br />
oxygen is injected into the subsurface, the higher<br />
the removal of iron will be. However, the absence of<br />
available Fe 2+ or the presence of other oxidizeable<br />
compounds may interfere with this assumption during<br />
subsurface treatment. During the injection phase,<br />
the oxygen front lags behind the injected water front,<br />
as illustrated in Figure 3.1. This results in a limited<br />
mixing of iron-rich groundwater <strong>and</strong> oxygen-rich<br />
injection water. Instead, the dissolved oxygen reaches<br />
the adsorbed Fe 2+ on the soil grains. Heterogeneous<br />
oxidation is extremely fast, so it may be assumed that<br />
consumption by other reducing compounds will only<br />
45<br />
3
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
3<br />
start once Fe 2+ oxidation is complete. This would mean<br />
that if more oxygen is injected into the subsurface than<br />
adsorbed Fe 2+ is available for oxidation, this oxygen<br />
will be consumed by other (a)biotic reactions. In other<br />
words, the injection of high oxygen levels may not only<br />
contribute to more oxidation of iron <strong>and</strong> is instead<br />
potentially consumed in other reactions. For the design<br />
of a community-scale subsurface treatment plant it<br />
is useful to know what oxygen concentrations are at<br />
least required in the injection water. Simple aeration<br />
tanks, as used in this study, have a limited capacity <strong>and</strong><br />
aeration of anoxic/anaerobic (iron-rich) groundwater<br />
to saturation levels is difficult to reach. Figure 3.9<br />
shows the breakthrough curves for iron after injecting<br />
different oxygen concentrations into the subsurface.<br />
Increased iron retardation is visible for higher oxygen<br />
concentrations. The difference between
3 Small-scale subsurface iron <strong>and</strong> arsenic removal in Bangladesh<br />
oxygen concentrations does enhance the subsurface<br />
removal of iron. At small injection volumes (500 m 3 , it is known that<br />
iron is retained in the subsurface (van Beek, 1985;<br />
van Halem, 2008). The objective of this study is,<br />
however, to apply this technology on a small-scale with<br />
injection volumes no more than 1 m 3 . As mentioned<br />
before, subsurface iron removal is known to increase<br />
in efficiency with every successive cycle. Figure 3.11<br />
shows the iron breakthrough curves at location B for<br />
cycle 1, 3, 6, 12, 16, <strong>and</strong> 20. It clearly shows that iron<br />
removal improved significantly in this period. During<br />
cycle 1, C/C 0<br />
=0.5 for iron was measured just after V/<br />
Vi = 2, while during cycle 20 this was at V/Vi >> 5.<br />
When extrapolating these results linearly, C/C 0<br />
=0.5 was<br />
[Fe] mmol.L -1<br />
0.3<br />
0.2<br />
0.1<br />
0<br />
Cycle 1 Cycle 3<br />
Cycle 6 Cycle 12<br />
Cycle 16 Cycle 20<br />
0 1 2 3 4 5<br />
V/Vi<br />
Figure 3.11 <strong>Iron</strong> breakthrough curves during successive cycles<br />
(Location B). Injection volumes (Vi) were 864L, 980L,<br />
744L, 840L, 786L, <strong>and</strong> 827L for cycles 1, 3, 6, 12, 16,<br />
<strong>and</strong> 20, respectively<br />
potentially even as late as V/Vi > 15, but the duration<br />
of the experiment was not long enough to confirm this<br />
with measurements. Based on these findings it can<br />
therefore be concluded that subsurface iron removal<br />
can be applied at decentralized small-scale treatment<br />
plants with injection volumes as low as 1 m 3 .<br />
The breakthrough curves for arsenic were also<br />
monitored <strong>and</strong> did not show the same improvement as<br />
for iron (Figure 3.12). The curves for all cycles show<br />
more or less the same trend, with C/C 0<br />
=0.5 around V/<br />
Vi = 1.5. The WHO drinking water guideline of 0.13<br />
μmol.L -1 was already exceeded before V/Vi = 1, which<br />
is clearly insufficient retardation for this system to be<br />
called an arsenic removal technique. An improved<br />
As removal efficiency had been reported in other<br />
publications (Rott et al., 2002; Appelo <strong>and</strong> de Vet, 2003),<br />
but does not seem to account for the sites in this study.<br />
47<br />
3
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
3<br />
The limited arsenic adsorption to the freshly formed Fe 3+<br />
oxide surfaces could be caused by insufficient contact<br />
time in the subsurface during abstraction to reach<br />
equilibrium in the oxidation zone. Rough calculations<br />
show that the oxidation zone, with an injection volume<br />
of 0.9 m 3 is approximately 0.5 m around the well. At an<br />
average pumping rate of 1.2 m 3 .h -1 this would result in<br />
a contact time in the oxidation zone of approximately<br />
20 minutes. In other words, once the groundwater<br />
arrives at the oxidation zone, 0.5 m from the well, it<br />
takes the water a travel time of 20 minutes to reach<br />
the well. It is noteworthy that in reality the oxidation<br />
front lags behind the injected water front, so the actual<br />
oxidised zone may be even smaller. Though arsenic<br />
adsorption may not be complete within 20 minutes, it<br />
would theoretically suffice to adsorb part of the arsenic.<br />
A design consideration to increase the contact time<br />
[As] µmol.L -1<br />
2.0<br />
1.5<br />
1.0<br />
0.5<br />
0.0<br />
Cycle 1 Cycle 3<br />
Cycle 6 Cycle 12<br />
Cycle 16 Cycle 20<br />
0 1 2 3 4 5<br />
V/Vi<br />
Figure 3.12 <strong>Arsenic</strong> breakthrough curves during successive cycles<br />
(Location B). Injection volumes (Vi) were 864L, 980L,<br />
744L, 840L, 786L, <strong>and</strong> 827L for cycles 1, 3, 6, 12, 16,<br />
<strong>and</strong> 20, respectively<br />
C/C0<br />
1<br />
0.5<br />
0<br />
0 2 4 6 8<br />
V/Vi<br />
Phosphate<br />
<strong>Arsenic</strong><br />
<strong>Iron</strong><br />
Figure 3.13 Phosphate, arsenic <strong>and</strong> iron breakthrough during<br />
cycle 8 at location B (Vi=643L)<br />
is to reduce the length of the well filter. Adsorption<br />
of zero-valent As(III) is known to be less effective to<br />
iron oxides than the negatively charged As(V). Since<br />
arsenic occurs at these sites predominantly as As(III),<br />
adsorption may be limited. Another explanation for<br />
the absence of arsenic adsorption may be sought in<br />
the competition with other anions (Stachowicz, 2007).<br />
Especially phosphate is present in concentrations high<br />
enough to interfere with arsenic adsorption (0.053<br />
mmol.L -1 ). Phosphate was measured to be delayed<br />
during abstraction, also when arsenic was already<br />
breaking through (Figure 3.13). The competitive role of<br />
phosphate during subsurface arsenic removal has been<br />
reported before by Appelo <strong>and</strong> de Vet (2003).<br />
48
3 Small-scale subsurface iron <strong>and</strong> arsenic removal in Bangladesh<br />
Conclusions<br />
<strong>Subsurface</strong> treatment was successfully operated at<br />
small-scale (injection volumes below 1 m 3 ) for the<br />
removal of iron. For arsenic, however, the system did<br />
not prove to be very effective. <strong>Arsenic</strong> retardation was<br />
limited <strong>and</strong> breakthrough of 10 µg.L -1 (WHO guideline)<br />
was observed before V/Vi=1, which corresponds to the<br />
moment of groundwater arrival at the well. <strong>Iron</strong> removal<br />
was more effective after multiple injection-abstraction<br />
cycles <strong>and</strong> after injection of higher dissolved oxygen<br />
concentrations. <strong>Subsurface</strong> arsenic removal was also<br />
slightly more effective at higher oxygen concentrations,<br />
but did not show the same improvement after<br />
successive cycles. Two explanations are proposed to be<br />
responsible for this result, either (i) contact times with<br />
the freshly formed iron oxides were insufficient, or (ii)<br />
competing anions, such as phosphate, interfered with<br />
the adsorption process.<br />
References<br />
Appelo C. A. J., B. Drijver, R. Hekkenberg <strong>and</strong> M. de Jonge<br />
(1999) Modeling in situ iron removal from ground water,<br />
Ground Water 37(6): 811-817.<br />
Appelo C. A. J. <strong>and</strong> W. W. J. M. de Vet (2003) Modeling in situ<br />
iron removal from groundwater with trace elements such<br />
as As. In <strong>Arsenic</strong> in groundwater. A. H. Welch <strong>and</strong> K.G.<br />
Stollenwerk. Kluwer Academic, Boston.<br />
BETV-SAM: Bangladesh Environmental Technology Verification<br />
– Support to <strong>Arsenic</strong> Mitigation, www.verification-unit.org.<br />
British Geological Survey/DPHE (2001). <strong>Arsenic</strong> contamination<br />
of groundwater in Bangladesh, Volume 2: Final report, BGS<br />
Technical Report WC/00/19.<br />
Chowdhury M.A.I., M.T. Uddin, M.F. Ahmed, M.A. Ali <strong>and</strong><br />
S.M. Uddin (2006) How does arsenic contamination of<br />
groundwater cause severity <strong>and</strong> health hazard in Bangladesh,<br />
J Appl Sci 6(6): 1275-1286.<br />
Clifford D.A., S. Karori, G. Ghurye <strong>and</strong> S. Samanta (2004) Field<br />
speciation method for arsenic inorganic species, AWWA<br />
Research Foundation, Denver.<br />
Dzombak D. A. <strong>and</strong> F. M. M. Morel (1990) Surface complexation<br />
modeling: hydrous ferric oxide, Wiley.<br />
Ferguson J.F. <strong>and</strong> J. Gavis (1972) A review of the arsenic cycle in<br />
natural waters, Water Research 6: 1259-1274.<br />
Hallberg R. O. <strong>and</strong> R. Martinell (1976) Vyredox - in situ<br />
purification of groundwater, Ground Water, 14(2): 88-93.<br />
Mettler S. (2002) In-situ removal of iron from groundwater:<br />
Fe(II) oxygenation, <strong>and</strong> precipitation products in a<br />
calcareous aquifer, PhD dissertation, Swiss Federal Institute<br />
of Technology, Zurich.<br />
Miller G. P. (2006) <strong>Subsurface</strong> treatment for arsenic removal.<br />
Denver, American Water Works Association Research<br />
Foundation: 59.<br />
Rott U. (1985) Physical, chemical <strong>and</strong> biological aspects of the<br />
removal of iron <strong>and</strong> manganese underground, Water Supply<br />
3: 143-150.<br />
Rott U., C. Meyer <strong>and</strong> M. Friedle (2002) Residue-free removal<br />
49<br />
3
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
3<br />
of arsenic, iron, manganese <strong>and</strong> ammonia from groundwater,<br />
Water Sci Technol: Water Supply 2(1): 17-24.<br />
Sarkar A. R. <strong>and</strong> O.T. Rahman (2001) In-situ removal of arsenic<br />
- experiences of DPHE-Danida pilot project. In Technologies<br />
for arsenic removal from drinking water, Bangladesh<br />
University of Engineering <strong>and</strong> Technology <strong>and</strong> The United<br />
Nations University, Bangladesh.<br />
SAR Technology (2011) In-situ arsenic treatment. www.<br />
insituarsenic.org.<br />
Sen Gupta B., S. Chatterjee, U. Rott, H. Kauffman, A.<br />
B<strong>and</strong>opadhyay, W. de Groot, N.K. Nag, A.A. Borbonell-<br />
Barrachina <strong>and</strong> S. Mukherjee (2009) A simple chemical free<br />
arsenic removal method for community, Environmental<br />
Pollution 157: 3351–3353.<br />
Smedley P.L <strong>and</strong> D.G. Kinniburgh (2002) A review of the source,<br />
behaviour <strong>and</strong> distribution of arsenic in natural waters, Appl<br />
Geochem 17: 517–568.<br />
Smith A.H., E.O. Lingas <strong>and</strong> M. Rahman (2000) Contamination<br />
of drinking-water by arsenic in Bangladesh: a public health<br />
emergency, Bulletin of the World Health Organization, 78<br />
(9): 1093-1103.<br />
Stachowicz M. (2007) Solubility of arsenic in multi-component<br />
systems: from the microscopic to the macroscopic scale, PhD<br />
thesis, Wageningen University, the Netherl<strong>and</strong>s.<br />
Sutherl<strong>and</strong> D., P.M. Swash, A.C. MacQueen, L.E. McWilliam,<br />
D.J. Ross <strong>and</strong> S.C. Wood (2002) A field based evaluation of<br />
household arsenic removal technologies for the treatment<br />
of drinking water, Environmental Technology 23(12): 1385-<br />
1404.<br />
van Beek C.G.E.M. (1983) Ondergrondse ontijzering, een<br />
evaluatie van uitgevoerd onderzoek (in Dutch). KIWA<br />
mededeling 78.<br />
van Beek C. G. E. M. (1985). Experiences with underground<br />
water treatment in the Netherl<strong>and</strong>s, Water Supply, 3(2), 1-11.<br />
van Halem, D., W. W. J. M. de Vet, G.L. Amy <strong>and</strong> J.C. van<br />
Dijk (2008) <strong>Subsurface</strong> iron removal for drinking water<br />
production: underst<strong>and</strong>ing the process <strong>and</strong> exploiting<br />
beneficial side effects, Water Quality Technology Conference,<br />
Cincinnati, American Water Works Association.<br />
van Halem D., S.A. Bakker, G.L. Amy <strong>and</strong> J.C. van Dijk (2009)<br />
<strong>Arsenic</strong> in drinking water: a worldwide water quality concern<br />
for water supply companies, Drinking Water Engineering <strong>and</strong><br />
Science 2 (1): 29-34.<br />
Welch A.H., K.G. Stollenwerk, L. Feinson <strong>and</strong> D.K. Maurer<br />
(2000) Preliminary evaluation of the potential for insitu<br />
arsenic removal from ground water. In: <strong>Arsenic</strong> in<br />
Groundwater of Sedimentary Aquifers, 31st International<br />
Geological Congress, Rio de Janeiro, Brazil.<br />
World Health Organization (2001) United Nations synthesis<br />
report on arsenic in drinking water, Geneva<br />
World Health Organization (2006) Guidelines for Drinkingwater<br />
Quality, First addendum to third edition, Volume1,<br />
Recommendations, Geneva<br />
50
Simulation of adsorptive-catalytic<br />
oxidation in s<strong>and</strong> columns<br />
4<br />
This chapter is based on:<br />
van Halem et al. (2010) Water Research 44: 5761-5769
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
4<br />
Introduction<br />
<strong>Arsenic</strong> contamination of shallow tube well drinking<br />
water in Bangladesh is an urgent developmental <strong>and</strong><br />
health problem (British Geological Survey/DPHE,<br />
2001; World Health Organization, 2001; Smith et al.,<br />
2002), disproportionately affecting the rural poor, i.e.,<br />
those most reliant on this source of drinking water.<br />
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal, relies on the<br />
existing infrastructure of a h<strong>and</strong>-pump/shallow tubewell<br />
<strong>and</strong> retains iron <strong>and</strong> arsenic in the subsurface. As<br />
such, it has potential advantages over other household<br />
<strong>and</strong> community arsenic removal systems, such as<br />
SONO <strong>and</strong> Alcan (Sutherl<strong>and</strong> et al., 2002):<br />
• no costly filter media <strong>and</strong> maintenance is needed;<br />
• the tube well is the 1st preferred option for drinking<br />
water in rural Bangladesh (WSP/Worldbank, 2003);<br />
<strong>and</strong> available to a majority of the rural poor in their<br />
household;<br />
• (minimal) additional hardware beyond the existing h<strong>and</strong><br />
pump is affordable <strong>and</strong> locally available/repairable;<br />
• iron is also removed which improves colour <strong>and</strong> taste<br />
of the water; greatly enhancing potential for social<br />
acceptance;<br />
• iron could be a visible indicator for arsenic presence (<strong>and</strong><br />
aid in post-deployment monitoring of water quality);<br />
• groundwater-irrigation leading to arsenic accumulation<br />
in crops (rice) may also be mitigated.<br />
By injecting oxygen-rich water into an anoxic aquifer,<br />
both homogenous <strong>and</strong> heterogeneous oxidation of<br />
ferrous iron will occur in the aquifer. Homogeneous<br />
oxidation of ferrous iron takes place in solution, <strong>and</strong><br />
predominantly occurs at the interface of injected water<br />
<strong>and</strong> original, anoxic groundwater. Based on the large<br />
surface area of iron hydroxides on the soil grains in<br />
the subsurface, it is thought that the heterogeneous<br />
reaction of ferrous iron oxidation on the surface of<br />
ferric iron oxides is dominant during subsurface iron<br />
removal. In literature, the system’s efficacy is explained<br />
by adsorptive-catalytic oxidation (van Beek 1985; Rott,<br />
1985), where adsorbed ferrous iron is oxidized to form<br />
new adsorption sites. On its way into the aquifer, the<br />
injected water oxidizes adsorbed ferrous iron <strong>and</strong> thus<br />
“regenerates” the subsurface for adsorption during<br />
abstraction:<br />
Equation 4.1<br />
S − OFe( II) + 0.25O + 1.5 H O →S − OFe( III)( OH ) + H<br />
+ 0 +<br />
2 2 2<br />
Due to the rapid consumption of oxygen during<br />
injection of aerated water, the oxygen front will lag<br />
behind the injected water front. When heterogeneous<br />
ferrous iron oxidation is complete, the iron hydroxide<br />
surface is available for adsorption of ferrous iron <strong>and</strong><br />
oxyanions, such as arsenic(III), during groundwater<br />
abstraction:<br />
Equation 4.2 <strong>and</strong> 4.3<br />
o 2+ + +<br />
S − OH + Fe →S − OFe( II ) + H<br />
S − OH + H3AsO3 →S − H<br />
2AsO3 + H<br />
2O<br />
52
4 Simulation of adsorptive-catalytic oxidation in s<strong>and</strong> columns<br />
Once the Fe 3+ oxide surface is exhausted, no more Fe 2+<br />
or arsenic is adsorbed <strong>and</strong> iron/arsenic breakthrough<br />
is observed in the produced water. Hence, during<br />
abstraction the iron/arsenic front is delayed <strong>and</strong> more<br />
iron-free water can be produced than was injected.<br />
Every period of injection-abstraction is referred to<br />
as a cycle, with the first injection-abstraction period<br />
being cycle 1. More water with reduced iron/arsenic<br />
concentrations can be abstracted (volume V) than was<br />
injected (volume Vi), i.e., this volumetric ratio (V/Vi)<br />
determines the efficiency of the system.<br />
<strong>Subsurface</strong> or in-situ iron removal has been<br />
used in central Europe for many decades (Hallberg <strong>and</strong><br />
Martinell, 1976; Boochs <strong>and</strong> Barovic, 1981; Jechlinger<br />
et al., 1985; Rott, 1985; van Beek, 1985; Appelo et al.,<br />
1999; Mettler, 2002), but the application of subsurface<br />
treatment for the removal of arsenic from groundwater<br />
is a relatively new approach (Rott et al. 2002, van<br />
Halem et al. 2010). This technology has the potential<br />
to be a cost-effective way to provide safe drinking<br />
water in rural areas in decentralized applications.<br />
With minimal investments in additional equipment,<br />
the existing infrastructure (h<strong>and</strong> pumps/shallow tube<br />
wells) can be modified to be operated under injection<br />
<strong>and</strong> abstraction conditions. In literature, a reduction<br />
of arsenic concentrations from maximum 40 μg.L -1 to<br />
below the WHO guideline (10 μg.L -1 , WHO, 2006) has<br />
been reported with the injection of aerated water into<br />
the aquifer (Rott et al., 2002; Appelo <strong>and</strong> De Vet, 2003).<br />
In Bangladesh the subsurface removal of higher arsenic<br />
levels was investigated by Sarkar <strong>and</strong> Rahman (2001),<br />
namely, 500 - 1300 μg.L -1 . In that study concentrations<br />
as low as 10 μg.L -1 were not reached, nevertheless,<br />
more than 50% removal was observed. In the absence<br />
of naturally occurring soluble ferrous iron, other<br />
researchers have studied the simultaneous injection<br />
of aerated water with ferric or ferrous iron (Welch et<br />
al., 2000; Miller, 2006). Preliminary results showed<br />
reduction of 100 μg.L -1 arsenic(V) to below the WHO<br />
guideline. Although these results are promising, only<br />
little is known about the limitations of this technology<br />
in the diverse geochemical settings of Bangladesh.<br />
The focus of this study was to identify the dominant<br />
processes in subsurface iron <strong>and</strong> arsenic removal in<br />
order to assess the applicability for rural Bangladesh.<br />
The methodology included (1) a field study with a<br />
community-scale facility in Manikganj, Bangladesh to<br />
assess the potential of decentralized subsurface iron<br />
<strong>and</strong> arsenic removal, <strong>and</strong> (2) column experiments with<br />
natural groundwater to simulate the shifting redox<br />
conditions in the oxidation zone during subsurface<br />
iron <strong>and</strong> arsenic removal. The column experiments<br />
provided controlled conditions for the investigation of<br />
the adsorptive-catalytic oxidation mechanism, whereas<br />
the test facility enabled to study subsurface treatment<br />
in the complex subterranean environment.<br />
53<br />
4
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
4<br />
Materials <strong>and</strong> methods<br />
Test facility, Bangladesh<br />
Household shallow tube wells with suction h<strong>and</strong> pumps<br />
are widely distributed in Bangladesh <strong>and</strong> the objective<br />
of subsurface iron <strong>and</strong> arsenic removal is to use this<br />
existing infrastructure. The Manikganj district, 40 km<br />
west of Dhaka, was selected for this study, since the area<br />
is known to have high iron <strong>and</strong> arsenic concentrations<br />
in the groundwater. A site was selected with elevated<br />
iron concentrations (0.27 mmol.L -1 ) <strong>and</strong> arsenic<br />
concentrations (1.94 μmol.L -1 ). Unlike other parts of<br />
the Manikganj district, manganese concentrations were<br />
not found to be high (5.46 μmol.L -1 ) at this particular<br />
location. For phosphate, however, the groundwater did<br />
show elevated levels (52.6 μmol.L -1 ). The OR potential<br />
of the groundwater was measured to be on average<br />
-170mV <strong>and</strong> the pH of the groundwater was 6.85.<br />
The experimental set-up (Figure 4.1) was<br />
connected to an existing h<strong>and</strong> pump with tube well in<br />
the upper aquifer. The 1.5-inch tube-well had a depth of<br />
31 meters <strong>and</strong> a perforated well length of 3 meter. As an<br />
added precaution, the set-up was placed with a family<br />
who already had arsenic treatment since 2001 (SIDKO<br />
system, BCSIR 2003). For the purpose of subsurface<br />
treatment, the existing situation was modified with a<br />
pipe <strong>and</strong> valve for injection. After subsurface treatment,<br />
the groundwater was pumped (electrical suction pump)<br />
into the SIDKO system for aeration, s<strong>and</strong> filtration<br />
<strong>and</strong> Granular Ferric Hydroxide filtration (AdsorpAs,<br />
Harbauer GmbH). The treated water, low in arsenic<br />
<strong>and</strong> iron, was collected in a 1 m 3 storage tank <strong>and</strong> used<br />
for injection into the aquifer. The maximum injection<br />
volume was therefore limited to 1 m 3 . Analysis of the<br />
water samples was done with field test kits (Wagtech<br />
International: Palintest <strong>and</strong> Arsenator) <strong>and</strong> confirmed<br />
in the laboratory (Perkin-Elmer Flame AAS 3110;<br />
Perkin-Elmer GF-AAS 5100PC). Duplicates or<br />
triplicates were taken to check the method of sampling<br />
<strong>and</strong> accuracy of analysis. <strong>Arsenic</strong> speciation was<br />
done with a field method (Clifford et al., 2004) using<br />
anion exchange resin columns (Amberlite IRA400).<br />
Multimeters (HACH 340i) were fixed inline to the<br />
water meter<br />
pump<br />
valve<br />
inline<br />
monitoring<br />
sampling<br />
tap<br />
iron<br />
filter<br />
aeration<br />
tank<br />
storage<br />
tank<br />
SIDKO<br />
system<br />
arsenic<br />
filter<br />
Figure 4.1 Experimental set-up for subsurface iron <strong>and</strong> arsenic<br />
removal at existing SIDKO system<br />
54
4 Simulation of adsorptive-catalytic oxidation in s<strong>and</strong> columns<br />
experimental set-up to monitor pH (WTW SenTix 41), <strong>and</strong> phosphate concentration of 33 μmol.L -1 . The<br />
dissolved oxygen (WTW Cellox 325), OR potential groundwater was pumped onto the columns during<br />
(WTW SenTix ORP) <strong>and</strong> electric conductivity the abstraction phase of a cycle, <strong>and</strong> the injection phase<br />
(TetraCon 325). Measurements were registered on consisted of drinking water. The drinking water had<br />
a computer with Multilab Pilot v5.06 software. The an oxygen concentration of 0.28 mmol.L -1 , a slightly<br />
injection <strong>and</strong> abstraction volumes were monitored higher pH of 7.4, <strong>and</strong> iron, manganese <strong>and</strong> arsenic were<br />
using water meters. Operation started in October 2008, below detection limits.<br />
just after the monsoon season, <strong>and</strong> continued until<br />
The experimental set-up (Figure 4.2) consisted<br />
May 2009. The family shared their arsenic treatment of duplicate transparent PVC columns with a length of<br />
facility with their community <strong>and</strong> the weekly water 80cm <strong>and</strong> an inner diameter of 36mm (wall thickness<br />
consumption was 2.4-2.9 m 3 . Operational conditions, 2mm). The columns were filled with washed (24h<br />
such as injection frequency <strong>and</strong> production discharge, with 3% HCl) filter s<strong>and</strong> (grain size = 0.6-1.2mm,<br />
varied due to irregular operation. Normally the set-ups D 10<br />
=0.75mm) that contained 48.4 μmol Fe.g -1 .ds<br />
were used for the families’ water production, however, after total iron extraction with 5M HCl. The pushpull<br />
operational mode of injection-abstraction at<br />
during research periods the operation was intensified.<br />
Injection was done the night before, <strong>and</strong> abstraction the test facility was simulated in the 1D plug-flow<br />
was started at least 12 hours after injection.<br />
environment of the columns with down flow (1.0 L.h -1<br />
±0.05) for both injection <strong>and</strong> abstraction. At the start<br />
Anoxic column experiments<br />
of the experiments the columns were conditioned<br />
The raw groundwater of Oasen Drinking Water with groundwater, until complete breakthrough of<br />
Company drinking water treatment plant Lekkerkerk iron occurred. Anoxic conditions were maintained<br />
in the Netherl<strong>and</strong>s was used as influent for the column in the columns by using an airtight FESTO system (6<br />
experiments. In addition, arsenic(III) (NaAsO 2<br />
, Fisher) x 1 PUN, I.D. 4mm) with matching connectors <strong>and</strong><br />
was added to simulate high arsenic conditions as found valves. The flow rate in the columns (2.16 m.h -1 ±0.11)<br />
in Bangladesh. In order to evaluate the effect of different was controlled with a multi-channel pump <strong>and</strong> PVC<br />
Fe:As ratios in the groundwater, several lower arsenic tubing with low gas permeability. The set-up remained<br />
concentrations were dosed. During the research period under constant positive hydrostatic pressure to prevent<br />
the groundwater had an average pH of 7.1, a nearly oxygen. An injection-abstraction cycle started with 1.5<br />
constant temperature of 12 °C, iron concentration of pore volume of (oxic) injection water <strong>and</strong> subsequently<br />
95 μmol.L -1 , manganese concentration of 11 μmol.L -1 the influent was switched to (anoxic) groundwater for<br />
4<br />
55
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
4<br />
multiple pore volumes. Electrical conductivity was used<br />
as a conservative tracer from which the pore volume<br />
could be calculated to be on average 0.37L (±0.005).<br />
For the columns, the V/Vi was calculated by dividing<br />
the produced water (V) by one pore volume (Vi), since<br />
the latter corresponds to the actual oxidized volume of<br />
s<strong>and</strong> in the column. The water quality parameters were<br />
monitored until at least C/C 0<br />
= 0.8 was reached for<br />
iron <strong>and</strong> arsenic (C= measured concentration, <strong>and</strong> C 0<br />
=<br />
original concentration), <strong>and</strong> runtimes of the columns<br />
per cycle varied between 9.2 <strong>and</strong> 16.1 pore volumes.<br />
During the experiments samples were taken for iron<br />
analysis (Perkin-Elmer Flame AAS 3110) <strong>and</strong> arsenic<br />
analysis (GF-AAS; Perkin-Elmer 5100PC). <strong>Arsenic</strong><br />
speciation was done with a field method (Clifford<br />
et al., 2004) using anion exchange resin columns<br />
(Amberlite IRA400). On-line measurements were<br />
done for dissolved oxygen (WTW Cellox 325), E h<br />
potential (WTW SenTix ORP), pH (WTW SenTix<br />
41), <strong>and</strong> electrical conductivity WTW (TetraCon 325).<br />
Measurements were registered on a computer with<br />
Multilab Pilot v5.06 software.<br />
pH ORP EC DO<br />
Waste/Sample<br />
Point<br />
H 2<br />
O<br />
+<br />
O 2<br />
Discharge Pump<br />
Groundwater<br />
Dosing Pump<br />
N 2<br />
Column 1<br />
Column 2<br />
NaAsO 2<br />
Figure 4.2 Experimental column set-up at WTP Lekkerkerk, the Netherl<strong>and</strong>s<br />
56
4 Simulation of adsorptive-catalytic oxidation in s<strong>and</strong> columns<br />
Results <strong>and</strong> discussion<br />
2<br />
1.6<br />
0.2<br />
0.16<br />
Breakthrough curves at test facility<br />
A dimensionless retardation factor (R) has been defined<br />
to represent the delayed arrival of iron or arsenic in<br />
the well compared to the original groundwater. R+1<br />
is equal to the V/Vi when the C/C 0<br />
(C= measured<br />
concentration, <strong>and</strong> C 0<br />
= original concentration) for<br />
iron or arsenic equals 0.5 divided by the V/Vi for a<br />
conservative tracer, e.g., electrical conductivity, at C/C 0<br />
= 0.5:<br />
C/C0<br />
R<br />
Well<br />
1<br />
0.5<br />
0<br />
TracerC/ C0<br />
0 1 2 3 4 5 6 7 8<br />
V/Vi<br />
⎛ V ⎞<br />
⎜ ⎟<br />
⎝Vi<br />
⎠Fe<br />
C / C0=<br />
0.5 4.5<br />
RFe<br />
= − 1 = = 3.5<br />
⎛ V ⎞<br />
1<br />
⎜ ⎟<br />
⎝Vi<br />
⎠<br />
TracerC / C0=<br />
0.5<br />
Equation 4.4<br />
⎛V<br />
⎞<br />
⎜ ⎟<br />
⎝Vi ⎠<br />
( PV )<br />
FeC/ C<br />
Fe<br />
0 C/<br />
C0<br />
= −1⇒ RColumn<br />
= −1<br />
⎛V<br />
⎞<br />
( PV )<br />
TracerC/ C0<br />
⎜ ⎟<br />
⎝Vi<br />
⎠<br />
iron<br />
tracer<br />
oxygen<br />
Figure 4.3 Typical breakthrough curve for electrical conductivity,<br />
dissolved oxygen, <strong>and</strong> iron, including determination<br />
of the retardation factor<br />
[As] µmol.L -1<br />
1.2<br />
0.8<br />
0.4<br />
0<br />
0 1 2 3 4 5 6 7 8<br />
V/Vi<br />
total As As(III) total Fe<br />
Figure 4.4 Development of arsenic <strong>and</strong> iron at the test facility in<br />
Manikganj, Bangladesh (cycle 20)<br />
The determination of the retardation factor is<br />
illustrated in Figure 4.3 for cycle 6 at the communityscale<br />
test facility in Manikganj, in this case the R Fe<br />
for iron is 4.5. Figure 4.4 depicts the breakthrough<br />
of total arsenic, arsenic(III) <strong>and</strong> iron during cycle 20<br />
at the test facility in Bangladesh. The graph clearly<br />
shows that iron breakthrough was delayed significantly,<br />
since the background concentration of 0.27 mmol.L -1<br />
was not reached at V/Vi = 7.5. The retardation factor<br />
(R Fe<br />
) for iron has an estimated value of 12. It can be<br />
calculated that the total amount of removed iron would<br />
in that case be approximately 2.6 moles. The volume<br />
of injected water for this particular cycle was 827L<br />
<strong>and</strong> had an oxygen concentration of 0.17 mmol.L -1 ,<br />
which adds up to a total amount of injected oxygen of<br />
±0.14 moles. In the case that all injected oxygen was<br />
consumed by subterranean adsorbed ferrous iron, <strong>and</strong><br />
0.12<br />
0.08<br />
0.04<br />
0<br />
[Fe] mmol.L -1<br />
4<br />
57
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
4<br />
thus used for the formation of new iron hydroxide<br />
surfaces, the measured iron removal does not even<br />
closely correspond to the equation that 1 mol of oxygen<br />
can oxidize 4 moles of ferrous iron (Equations 4.1 <strong>and</strong><br />
4.2). In other words, iron removal at this particular site<br />
was much more effective than can be explained by the<br />
theory of adsorptive-catalytic oxidation.<br />
<strong>Arsenic</strong> breakthrough started immediately at<br />
V/Vi=0 <strong>and</strong> reached complete breakthrough before<br />
V/Vi=5. During this cycle, the retardation factor for<br />
arsenic did not even reach 1. In the initial stage of the<br />
cycle all arsenic that breaks through is arsenic(III),<br />
but after V/Vi=4 arsenic(V) also arrived at the well. In<br />
total, 2.6 mmol of arsenic is removed during this cycle,<br />
of which 1.1 mmol is arsenic(V). This gave an arsenic<br />
adsorption ratio of 1.0 mmol As/mol of removed iron.<br />
It may be concluded that the efficient iron removal does<br />
not promote the equivalent co-removal of arsenic. Also,<br />
iron does not provide a visible indicator for arsenic<br />
presence at this site – which could have been an aid in<br />
post-deployment monitoring of the water quality.<br />
Breakthrough curves for anoxic column<br />
experiments<br />
In the columns, oxygen-rich drinking water was dosed<br />
to the columns for 1.5 pore volumes <strong>and</strong> remained in<br />
the columns overnight (16 hours). In the morning,<br />
columns were re-started with natural groundwater<br />
<strong>and</strong> monitored for the retardation of arsenic <strong>and</strong> iron.<br />
Since the columns were operated under strict plug<br />
EC (C/C0)<br />
pH<br />
O2 (mmol.L -1 )<br />
Eh (mV)<br />
1<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0<br />
7.2<br />
7.1<br />
7<br />
0.25<br />
0.2<br />
0.15<br />
0.1<br />
0.05<br />
0<br />
0<br />
-50<br />
-100<br />
-150<br />
0.0 0.5 1.0 1.5<br />
Pore volume<br />
0.0 0.5 1.0 1.5<br />
Pore volume<br />
0.0 0.5 1.0 1.5<br />
Pore volume<br />
0.0 0.5 1.0 1.5<br />
Pore volume<br />
Figure 4.5 Development of the tracer (electrical conductivity),<br />
pH, dissolved oxygen <strong>and</strong> Eh potential at the start of<br />
an injection cycle in the columns<br />
58
4 Simulation of adsorptive-catalytic oxidation in s<strong>and</strong> columns<br />
flow conditions, it can be assumed that homogeneous<br />
oxidation <strong>and</strong> precipitation were very limited <strong>and</strong> that<br />
heterogeneous oxidation <strong>and</strong> adsorption processes,<br />
<strong>and</strong> thus adsorptive-catalytic oxidation, dominated.<br />
The injection of aerated water into the Fe 2+ saturated<br />
columns onsets the oxidation of adsorbed Fe 2+ ,<br />
resulting in a pH drop. Figure 4.5 illustrates the<br />
development of the tracer (electrical conductivity),<br />
pH, oxygen concentration <strong>and</strong> E h<br />
potential during the<br />
[As] µmol.L -1<br />
[Fe] mmol.L -1<br />
4<br />
3<br />
2<br />
1<br />
0<br />
0.1<br />
0.08<br />
0.06<br />
0.04<br />
0.02<br />
0<br />
total As<br />
As(III)<br />
Initial - total As<br />
Initial - As(III)<br />
0 2 4 6 8 10 12 14<br />
0 2 4 6 8 10 12 14<br />
V/Vi<br />
total Fe<br />
Initial Fe<br />
Figure 4.6 Breakthrough of arsenic <strong>and</strong> iron in the columns<br />
(cycle 14)<br />
start of an injection cycle. The amount of consumed<br />
oxygen can be registered by the retardation of the<br />
oxygen curve compared to the conservative tracer,<br />
electrical conductivity. The total oxygen consumption<br />
during injection was 0.05 mmol (results not shown),<br />
which corresponds to approximately ¼ of the amount<br />
of removed iron (Equation 4.1). Based on this mass<br />
balance, it can be concluded that the oxygen retardation<br />
was indeed caused by heterogeneous oxidation of<br />
ferrous iron in the column.<br />
The typical breakthrough curves of arsenic <strong>and</strong><br />
iron are depicted in Figure 4.6 for one of the columns<br />
(cycle 14). The arsenic concentration spiked to the<br />
influent consisted of 3.7 μmol.L -1 , of which 2.8 μmol.L -1<br />
was arsenic(III). The graph shows that the original<br />
arsenic concentration was reached just before V/Vi =<br />
7, with a retardation factor (R As<br />
) of 1. Like in the test<br />
facility, arsenic(V) was initially completely removed,<br />
but passed the columns around V/Vi = 4. The iron<br />
content of the natural groundwater was 95 μmol.L -1 <strong>and</strong><br />
in the columns this concentration was reduced with a<br />
retardation factor of 6. The total amount of removed<br />
iron was 0.21 mmol, which yields an arsenic adsorption<br />
ratio of 24.8 mmol As per mol of removed iron.<br />
Retardation factors at community-scale test<br />
facility<br />
<strong>Subsurface</strong> iron removal has been frequently reported<br />
to increase in efficacy with every successive cycle<br />
(Hallberg <strong>and</strong> Martinell, 1976; Rott, 1985; van Beek,<br />
59<br />
4
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
4<br />
1985; Jechlinger et al., 1985; Boochs <strong>and</strong> Barovic, 1981;<br />
Mettler, 2002; Braester <strong>and</strong> Martinell, 1988). Figure<br />
4.7 shows that this was also the case for the test facility<br />
in Bangladesh, hence R Fe<br />
increased from 1 to 14 after<br />
multiple cycles. It is noteworthy that the injection<br />
volume at this facility was limited to only 1 m 3 , which<br />
can be considered very small-scale compared to<br />
existing treatment plants in Europe where injection<br />
volumes typically vary between 500 <strong>and</strong> 1,000 m 3 (van<br />
Beek, 1985; Appelo <strong>and</strong> de Vet, 2003). It is therefore an<br />
important finding that, even at small-scale, subsurface<br />
iron removal is effective <strong>and</strong> could provide iron-free<br />
water in decentralized facilities in rural areas. R As<br />
did<br />
not increase with every successive cycles, it remained<br />
stable at around 1, illustrating that the process which is<br />
responsible for the increasingly effective iron removal<br />
Retardation factor<br />
20<br />
15<br />
10<br />
5<br />
0<br />
arsenic<br />
iron<br />
0 2 4 6 8 10 12 14 16 18 20<br />
Cycle #<br />
Figure 4.7 Retardation factors for iron <strong>and</strong> arsenic during<br />
successive cycles at the test facility in Manikganj,<br />
Bangladesh<br />
during subsurface treatment did not promote an<br />
equally effective co-removal of arsenic.<br />
There is a general consensus in the literature<br />
that adsorptive-catalytic oxidation is the dominant<br />
mechanism in subsurface iron removal, but the<br />
increasing efficacy of subsurface iron removal has<br />
also been attributed to bacterial activity (Hallberg<br />
<strong>and</strong> Martinell, 1976; Jechlinger et al., 1985; Rott, 1985;<br />
Grombach, 1985), occurrence of dead-end pores or<br />
stagnant zones (Boochs <strong>and</strong> Barovic, 1981), growth of<br />
the oxidation zone with every cycle (Appelo et al., 1999),<br />
<strong>and</strong> oxidation of reductants other than Fe 2+ during<br />
initial cycles (van Beek, 1985). Heterogeneous<br />
oxidation of Fe 2+ is extremely fast, so oxidation of<br />
other reductants is unlikely to be favored. Nevertheless,<br />
oxidation of, e.g., pyrite could be a secondary reaction,<br />
resulting in elevated iron (<strong>and</strong> sulphate) concentrations<br />
in the produced water. Such mobilization of iron<br />
could underestimate the actual iron removal efficiency<br />
through adsorptive-catalytic oxidation during initial<br />
cycles. In the columns, only the adsorptive-catalytic<br />
oxidation mechanism within the oxidation zone was<br />
simulated. The clean filter s<strong>and</strong> did not contain other<br />
reductants that consume oxygen during initial cycles<br />
<strong>and</strong> changing transport mechanisms are unlikely to be<br />
relevant in the columns, since the conservative tracer<br />
tests confirmed stable plug conditions after multiple<br />
cycles. Bacterial activity was checked by Phase Contrast<br />
Microscopy of the produced water during cycle 21 <strong>and</strong><br />
no iron oxidizers like Gallionella spp. were found. In<br />
60
4 Simulation of adsorptive-catalytic oxidation in s<strong>and</strong> columns<br />
other words, the adsorptive-catalytic oxidation process<br />
during subsurface iron <strong>and</strong> arsenic removal is expected<br />
to be the dominant process in the columns.<br />
Retardation factors for anoxic column<br />
experiments<br />
The retardation factors for iron <strong>and</strong> arsenic for the<br />
different injection-abstraction cycles in the columns are<br />
depicted in Figure 4.8. It should be noted that dosing of<br />
arsenic to the natural groundwater was started after 7<br />
cycles; therefore the arsenic retardation factors for the<br />
initial cycles are not included in the graph. Successive<br />
cycles in the columns show similar retardation factors<br />
for arsenic as the test facility. R As<br />
remained around<br />
Retardation factor<br />
10<br />
8<br />
6<br />
4<br />
2<br />
0<br />
0 2 4 6 8 10 12 14 16 18 20 22<br />
Cycle #<br />
arsenic iron arsenic-C2 iron-C2<br />
Figure 4.8 Retardation factors for iron <strong>and</strong> arsenic during<br />
successive cycles in the duplicate columns (C1 <strong>and</strong><br />
C2)<br />
1 <strong>and</strong> did not increase after multiple cycles. In the<br />
columns, the removal of arsenic is expected to be<br />
purely adsorptive, <strong>and</strong> the strong correlation with the<br />
results from the test facility suggests that those results<br />
were also achieved with adsorptive arsenic removal.<br />
<strong>Iron</strong> retardation in the columns did increase<br />
initially, but remained more or less stable at R Fe<br />
=7<br />
after the first 6 cycles. The efficacy did not improve<br />
significantly with every cycle as was found in the test<br />
facility, <strong>and</strong> since the dominant mechanism in the<br />
columns was adsorptive-catalytic oxidation it can be<br />
concluded that this mechanism was not responsible for<br />
increasing efficacies at the test facility. Apparently this<br />
mechanism does not provide sufficient new adsorptive<br />
surface area through freshly formed iron hydroxides to<br />
improve the system’s efficacy with every cycle. Hence,<br />
the mechanism which was responsible for improved<br />
iron removal with every successive cycle in the field<br />
situation does not prevail in the columns. Unlike in<br />
the columns, at the test facility bacterial activity <strong>and</strong>/or<br />
occurrence of stagnant zones may control the improved<br />
iron removal efficacy with every cycle.<br />
Fe:As ratio of the groundwater<br />
The total amount of arsenic removed per cycle in the<br />
duplicate columns varied between 1.6 <strong>and</strong> 3.6 μmol.<br />
cycle -1 <strong>and</strong> did not increase with every successive cycle.<br />
It appears that the sites available for arsenic adsorption<br />
are regenerated during every injection phase, but the<br />
number of sorption sites does not seem to increase<br />
61<br />
4
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
4<br />
due to the freshly retained iron in the columns. On<br />
average, the amount of removed arsenic per mol of<br />
removed iron is 8.4 <strong>and</strong> 10.0 mmol As per mol Fe<br />
for the duplicate columns. This is valid for an arsenic<br />
concentration of 3.1 μmol.L -1 <strong>and</strong> an iron concentration<br />
of 95 μmol.L -1 . The amount of available sorption sites<br />
appears stable with every cycle, so the breakthrough of<br />
arsenic would, in theory, be delayed in case of lower<br />
arsenic concentrations (Equation 4.3). To study the<br />
effect of the molar Fe:As ratio in groundwater on the<br />
adsorptive removal efficiency of arsenic, the column<br />
experiments were repeated with different arsenic(III)<br />
concentrations: 3.1, 1.5, <strong>and</strong> 0.9 μmol.L -1 .<br />
<strong>Iron</strong> concentrations in the groundwater<br />
remained constant at 95 μmol.L -1 , thus the Fe:As<br />
ratios were 28, 63, <strong>and</strong> 103. The results for arsenic<br />
[As] µmol.L -1<br />
4<br />
3<br />
2<br />
Fe:As = 28<br />
Fe:As = 63<br />
breakthrough in the duplicate columns are depicted in<br />
Figure 4.9. It clearly shows that, independent of arsenic<br />
concentration, the breakthrough trend <strong>and</strong> retardation<br />
factor (R As<br />
=~1) were the same at the different Fe:As<br />
ratios. In other words, the percent breakthrough<br />
curves for arsenic at different Fe:As ratios matched<br />
exactly. This implies that although the amount of<br />
removed iron was equal per cycle, the total amount of<br />
removed arsenic was not. At a Fe:As ratio of 28, 63, <strong>and</strong><br />
103, the total amount of removed arsenic was 3.9, 1.6,<br />
<strong>and</strong> 1.1 μmol, respectively. The sorption sites on the<br />
freshly formed iron hydroxide surfaces were apparently<br />
only available for arsenic adsorption up to a V/Vi of<br />
7-9, independent of the Fe:As ratio. This indicates that<br />
the available sorption sites may have been occupied by<br />
other sorbates in the groundwater, such as competing<br />
anions (e.g., phosphate), which limited the adsorption<br />
of arsenic. It is noteworthy that the Fe:As ratio at the test<br />
facility was even higher than in the columns (Fe:As =<br />
140) <strong>and</strong> also complete arsenic breakthrough occurred<br />
around V/Vi of 5-6.<br />
1<br />
Fe:As = 103<br />
General discussion<br />
0<br />
0 2 4 6 8 10 12<br />
V/Vi<br />
Figure 4.9 <strong>Arsenic</strong> breakthrough curves for duplicate columns<br />
with Fe:As ratios of 28, 63 <strong>and</strong> 103<br />
The total amount of iron that was removed per cycle<br />
was stable over time at an average of 0.30 <strong>and</strong> 0.29<br />
mmol for the duplicate columns. The retardation curve<br />
for oxygen showed that only a portion of the injected<br />
62
4 Simulation of adsorptive-catalytic oxidation in s<strong>and</strong> columns<br />
oxygen was consumed for oxidation of adsorbed<br />
ferrous iron. After one pore volume, with a travel time<br />
of approximately 22 minutes, the oxygen concentration<br />
had reached 82% of its original concentration again.<br />
Of the total amount of injected oxygen (155 μmol)<br />
around 75-105 μmol passed the column. Only 50-<br />
80 μmol O 2<br />
/cycle was consumed during the injection<br />
phase in the columns, corresponding to the molar<br />
Fe:O 2<br />
removal ratio of 4 <strong>and</strong> thus consistent with the<br />
theory of adsorptive-catalytic oxidation (Equation 4.1<br />
<strong>and</strong> 4.2). In other words, the oxidation of adsorbed<br />
ferrous iron was complete during the injection phase,<br />
illustrating that the iron oxidation reaction was fast <strong>and</strong><br />
the adsorbed ferrous iron reacts with only a portion of<br />
the total amount of injected oxygen.<br />
For operational purposes this is an important<br />
finding, since, above a certain threshold value, the<br />
oxygen concentration does not control the adsorptioncatalytic<br />
oxidation, but rather the injection volume.<br />
The key is to oxidize as much soil grain surface area<br />
as possible, since this will provide new sorption sites<br />
for iron <strong>and</strong> arsenic. It is thus unlikely that injection of<br />
chemical oxidants (such as permanganate) will improve<br />
subsurface iron removal efficiencies. Such chemicals<br />
may even inhibit any bacterial activity that could be<br />
responsible for the improved iron removal efficiency<br />
with every successive cycle. Although only a portion<br />
of injected oxygen is used for rapid heterogeneous<br />
iron oxidation, the surplus oxygen in the oxidation<br />
zone is available for consumption by other adsorbed<br />
components – such as iron-oxidizing bacteria.<br />
Freshly formed iron oxides usually have high<br />
(adsorptive) surface areas (Cornell <strong>and</strong> Schwertmann,<br />
1996) <strong>and</strong> enhance the removal of ferrous iron.<br />
Interfacial Electron Transfer (IET, Jeon et al., 2001; Jeon<br />
et al., 2003) has been proposed to describe the “loss” of<br />
ferrous iron in an ferrous/ferric iron system. IET entails<br />
the transport of an electron to adsorbed ferrous iron<br />
from the incorporated iron hydroxide, creating new<br />
sorption sites at the surface. The theory of IET could<br />
provide an explanation for the improved subsurface<br />
iron removal efficiency at the test facility; however,<br />
these results were not reproduced by the adsorptivecatalytic<br />
oxidation in the columns. It is therefore more<br />
likely that other processes, such as bacterial activity or<br />
transport phenomena, are responsible for the enhanced<br />
iron removal in full-scale facilities.<br />
The community-scale test facility in Manikganj<br />
has shown that iron removal increases after multiple<br />
cycles, but arsenic removal remains stable at a<br />
retardation factor of 1. Hence, the amount of arsenic<br />
removed per mol of removed iron reduces with every<br />
successive cycle. This indicates that arsenic adsorption<br />
during subsurface treatment is controlled by the<br />
amount of oxidized iron, <strong>and</strong> not by the amount of<br />
removed iron. In the columns, the arsenic adsorption<br />
is also stable at R As<br />
=1 but, unlike at the test facility, iron<br />
removal does not improve with every successive cycle.<br />
The mechanism of adsorptive-catalytic oxidation is<br />
isolated in the columns from other potential removal<br />
63<br />
4
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
4<br />
processes, showing that subsurface arsenic removal is,<br />
indeed, controlled by the amount of oxidized iron per<br />
cycle. In the field, enhancement of subsurface arsenic<br />
removal can therefore only be achieved by increasing<br />
the oxidation zone, i.e. the volume of injected water. A<br />
high molar Fe:As ratio of the groundwater has not been<br />
shown to promote improved co-removal of arsenic<br />
with the iron. A proposed explanation for this finding is<br />
that arsenic removal is limited by the presence of other<br />
constituents in the natural groundwater, competing<br />
for the same adsorption sites. The process of arsenic<br />
adsorption is not limited by ferrous iron sorption<br />
(Dixit <strong>and</strong> Hering, 2006), but may be limited by the<br />
presence of competing anions in the multi-component<br />
environment, such as phosphate (Stachowicz, 2008).<br />
In the columns, phosphate concentrations were 10<br />
to 37 times higher than the arsenic concentrations,<br />
<strong>and</strong> phosphate adsorption may thus dominate over<br />
arsenic adsorption. In practice, competing anions will<br />
frequently co-occur in the groundwater with arsenic <strong>and</strong><br />
could therefore locally threat the efficacy of subsurface<br />
arsenic removal. One could overcome this limitation by<br />
improving the current design to reach larger injection<br />
volumes through utilizing rainwater for injection. In<br />
areas with heavy rainfall during the monsoon season(s),<br />
it could even be considered to combine subsurface<br />
arsenic removal with artificial rainwater recharge <strong>and</strong><br />
recovery. With such a design, the oxidation zone is<br />
scaled-up <strong>and</strong> arsenic/iron adsorption would occur in<br />
a much larger area around the well.<br />
Conclusions<br />
At the community-scale test facility in Bangladesh,<br />
subsurface iron removal showed great potential<br />
for decentralized application in rural areas. The<br />
efficacies were much higher than could be explained<br />
by the adsorptive-catalytic oxidation in the column<br />
experiments <strong>and</strong> therefore other ((a)biotic or transport)<br />
processes must contribute to the system’s efficacy.<br />
Unlike iron removal, subsurface arsenic removal did<br />
not increase after multiple cycles, illustrating that<br />
the process which is responsible for the effective iron<br />
removal did not promote an equally effective coremoval<br />
of arsenic.<br />
The strong correlation between field <strong>and</strong><br />
column results indicates that arsenic adsorption during<br />
subsurface treatment is controlled by the amount of<br />
adsorbed iron that is oxidized, <strong>and</strong> not by the amount<br />
of removed iron. For operational purposes this is<br />
an important finding, since apparently the oxygen<br />
concentration of the injection water does not control<br />
the arsenic removal, like it does for iron. On the other<br />
h<strong>and</strong>, increasing the injection volume may be a better<br />
approach to enhance arsenic removal. No relation has<br />
been observed in this study between the amount of<br />
removed arsenic <strong>and</strong> the Fe:As ratio of the groundwater.<br />
It is proposed that the removal of arsenic is limited<br />
by the presence of other anions, such as phosphate,<br />
competing for the same adsorption sites.<br />
64
4 Simulation of adsorptive-catalytic oxidation in s<strong>and</strong> columns<br />
References<br />
Appelo C. A. J., B. Drijver, R. Hekkenberg <strong>and</strong> M. de Jonge<br />
(1999) Modeling in situ iron removal from ground water,<br />
Ground Water 37(6): 811-817.<br />
Appelo C. A. J. <strong>and</strong> W. W. J. M. de Vet (2003) Modeling in situ<br />
iron removal from groundwater with trace elements such<br />
as As. In <strong>Arsenic</strong> in groundwater. A. H. Welch <strong>and</strong> K.G.<br />
Stollenwerk. Kluwer Academic, Boston.<br />
BCSIR (2003) Performance evaluation <strong>and</strong> verification of five<br />
arsenic removal technologies, ETV-AM Field Testing <strong>and</strong><br />
Technology Verification Program, Dhaka.<br />
Boochs P. W. <strong>and</strong> G. Barovic (1981) Numerical-model describing<br />
groundwater treatment by recharge of oxygenated water.<br />
Water Resources Research 17(1): 49-56.<br />
Braester C. <strong>and</strong> R. Martinell (1988) The Vyredox <strong>and</strong> Nitredox<br />
methods of in situ treatment of groundwater. Water Science<br />
<strong>and</strong> Technology 20(3): 149-163.<br />
British Geological Survey/DPHE (2001) <strong>Arsenic</strong> contamination<br />
of groundwater in Bangladesh, Volume 2: Final report, BGS<br />
Technical Report WC/00/19.<br />
Clifford D.A., S. Karori, G. Ghurye <strong>and</strong> S. Samanta (2004) Field<br />
speciation method for arsenic inorganic species, American<br />
Water Works Association Research Foundation, Denver.<br />
Cornell R. M. <strong>and</strong> U. Schwertmann (1996) The iron oxides -<br />
structure, properties, reaction, occurrence <strong>and</strong> uses, VCH-<br />
Germany <strong>and</strong> USA.<br />
Dixit S. <strong>and</strong> J. G. Hering (2006) Sorption of Fe(II) <strong>and</strong> As(III)<br />
on goethite in single- <strong>and</strong> dual-sorbate systems, Chemical<br />
Geology 228(1-3): 6-15.<br />
Grombach P. (1985) Groundwater treatment in situ in the<br />
aquifer, Water Supply 3(1): 13-18.<br />
Hallberg R. O. <strong>and</strong> R. Martinell (1976) Vyredox - in situ<br />
purification of groundwater, Ground Water 14(2): 88-93.<br />
Jechlinger G., W. Kasper, F. Scholler <strong>and</strong> F. Seidelberger (1985)<br />
The removal of iron <strong>and</strong> manganese in groundwaters through<br />
aeration underground, Water Supply 3(1): 19-25.<br />
Jeon B. H., B.A. Dempsey, W.D. Burgos <strong>and</strong> R.A. Royer (2001)<br />
Reactions of ferrous iron with hematite, Colloids <strong>and</strong><br />
Surfaces A: Physicochemical <strong>and</strong> Engineering Aspects 191(1-<br />
2): 41-55.<br />
Jeon B. H., B.A. Dempsey <strong>and</strong> W.D. Burgos (2003) Kinetics <strong>and</strong><br />
mechanisms for reactions of Fe(II) with iron(III) oxides.<br />
Environmental Science <strong>and</strong> Technology 37(15): 3309-3315.<br />
Mettler S. (2002) In-situ removal of iron from groundwater:<br />
Fe(II) oxygenation, <strong>and</strong> precipitation products in a<br />
calcareous aquifer, PhD dissertation, Swiss Federal Institute<br />
of Technology, Zurich.<br />
Miller G. P. (2006) <strong>Subsurface</strong> treatment for arsenic removal,<br />
American Water Works Association Research Foundation:<br />
59, Denver.<br />
Rott U. (1985) Physical, chemical <strong>and</strong> biological aspects of the<br />
removal of iron <strong>and</strong> manganese underground. Water Supply<br />
3(2): 143-150.<br />
Rott U., C. Meyer <strong>and</strong> M. Friedle (2002) Residue-free removal<br />
of arsenic, iron, mangenese <strong>and</strong> ammonia from groundwater.<br />
Water Science <strong>and</strong> Technology: Water Supply 2(1), 17-24.<br />
Sarkar A. R. <strong>and</strong> O.T. Rahman (2001) In-situ removal of arsenic<br />
- experiences of DPHE-Danida pilot project. In Technologies<br />
for arsenic removal from drinking water, Bangladesh<br />
65<br />
4
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
4<br />
University of Engineering <strong>and</strong> Technology <strong>and</strong> The United<br />
Nations University, Bangladesh.<br />
Smith A.H., P.A. Lopipero, M.N. Bates <strong>and</strong> C.M. Steinmaus<br />
(2002) <strong>Arsenic</strong> epidemiology <strong>and</strong> drinking water st<strong>and</strong>ards,<br />
Science 296, 2145-2146.<br />
Stachowicz M., T. Hiemstra <strong>and</strong> W. H. Van Riemsdijk (2008)<br />
Multi-competitive interaction of As(III) <strong>and</strong> As(V) oxyanions<br />
with Ca 2+ , Mg 2+ 3− 2−<br />
, PO 4<br />
4 , <strong>and</strong> CO 3<br />
ions on goethite, Journal<br />
of Colloid <strong>and</strong> Interface Science 320, 400–414.<br />
Sutherl<strong>and</strong> D., P.M. Swash, A.C. MacQueen, L.E. McWilliam,<br />
D.J. Ross <strong>and</strong> S.C. Wood (2002) A field based evaluation of<br />
household arsenic removal technologies for the treatment<br />
of drinking water. Environmental Technology 23(12): 1385-<br />
1404.<br />
van Beek C. G. E. M. (1985) Experiences with underground<br />
water treatment in the Netherl<strong>and</strong>s, Water Supply 3(2): 1-11.<br />
van Halem D., S. G. J. Heijman, G. L. Amy <strong>and</strong> J. C. van Dijk<br />
(2010) <strong>Subsurface</strong> arsenic removal for small-scale application<br />
in developing countries. Desalination 251: 241-248.<br />
Welch A.H., K.G. Stollenwerk, L. Feinson <strong>and</strong> D.K. Maurer<br />
(2000) Preliminary evaluation of the potential for insitu<br />
arsenic removal from ground water. In <strong>Arsenic</strong> in<br />
Groundwater of Sedimentary Aquifers, 31st International<br />
Geological Congress, Rio de Janeiro, Brazil.<br />
World Health Organization (2001) United Nations synthesis<br />
report on arsenic in drinking water. Geneva.<br />
World Health Organization (2006) Guidelines for Drinkingwater<br />
Quality, First addendum to third edition, Volume1,<br />
Recommendations, Geneva.<br />
WSP/Worldbank (2003) Fighting <strong>Arsenic</strong>: Listening to Rural<br />
Communities, Willingness to Pay for <strong>Arsenic</strong>-Free Safe<br />
Drinking Water in Bangladesh. Geneva.<br />
66
5<br />
Cation exchange during subsurface<br />
iron removal<br />
This chapter is based on:<br />
van Halem et al. (2010) Water Research: under review
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
5<br />
Introduction<br />
Based on the premise that heterogeneous oxidation<br />
prevails during <strong>Subsurface</strong> <strong>Iron</strong> <strong>Removal</strong> (SIR), the<br />
adsorptive-catalytic oxidation theory summarizes the<br />
processes during an injection-abstraction cycle (Rott,<br />
1985; van Beek, 1985). Besides the adsorptive-catalytic<br />
oxidation theory, it has also been proposed that the<br />
injection of O 2<br />
-rich water onsets the exchange of<br />
adsorbed Fe 2+ with other cations, such as Ca 2+ <strong>and</strong> Na +<br />
(Appelo et al., 1999; Appelo <strong>and</strong> Postma, 2005):<br />
Equations 5.1 <strong>and</strong> 5.2<br />
XCa + Fe ⇔ XFe + Ca<br />
2+ 2+<br />
2+ +<br />
2XNa + Fe ⇔ X<br />
2Fe + 2Na<br />
With X being the exchange sites <strong>and</strong> exchange<br />
coefficients, following the Gaines-Thomas convention,<br />
of K Ca\Fe<br />
= 0.7 <strong>and</strong> K Na\Fe<br />
=0.6 (Appelo <strong>and</strong> Postma, 2005).<br />
Once Fe 2+ has exchanged/desorbed, the dissolved O 2<br />
in the injection water oxidizes the soluble Fe 2+ . In the<br />
presence of high Fe 3+ oxide concentrations, as generally<br />
found in the aquifer, this oxidation reaction can also<br />
occur heterogeneously. Subsequent hydrolysis results in<br />
the formation of immobile iron oxides or even mobile<br />
colloidal material (Wolthoorn, 2003). During the<br />
abstraction phase, these iron hydroxides provide new<br />
surface sites for Fe 2+ adsorption <strong>and</strong>/or cation exchange.<br />
The process of cation exchange (CIEX) during SIR has<br />
been schematized in Figure 5.1 for a simplified system<br />
containing Fe 2+ during abstraction <strong>and</strong> Ca 2+ , Na + <strong>and</strong><br />
O 2<br />
during injection. Before the start of injection, both<br />
soluble <strong>and</strong> adsorbed Fe 2+ are present (Figure 5.1A).<br />
During injection the cations in the injection water, Ca 2+<br />
<strong>and</strong> Na + , exchange with the adsorbed Fe 2+ on the soil<br />
grains (Figure 5.1B). The desorbed Fe 2+ is then flushed<br />
deeper into the aquifer, partially mixing with O 2<br />
in the<br />
injection water, resulting in hydrolyzed Fe 3+ precipitates<br />
(Fe(OH) 3<br />
; Figure 5.1C). When abstraction is started,<br />
the flow is reversed <strong>and</strong> the Fe 2+ in the groundwater<br />
is retained on the mineral surface <strong>and</strong> on the Fe 3+<br />
hydroxides, either through CIEX or adsorption (Figure<br />
5.1D).<br />
The contribution of cation exchange to the<br />
system’s efficiency depends on the water composition<br />
of the injection water <strong>and</strong> groundwater, but also on<br />
the exchangeable Fe 2+ on the aquifer material. The<br />
cation exchange capacity (CEC) of the soil depends,<br />
in order of importance, on the clay, organic carbon<br />
<strong>and</strong> iron hydroxide content (Appelo <strong>and</strong> Postma,<br />
2005). The occurrence of CIEX is difficult to extract<br />
from field data, as iron oxidizes after it has exchanged,<br />
<strong>and</strong> will not reach the well in its ferrous, soluble<br />
state. However, for the purpose of optimizing the<br />
subsurface iron removal process or the application<br />
for other inorganic constituents, such as arsenic<br />
(Rott et al., 2002; van Halem et al., 2010) it is vital to<br />
underst<strong>and</strong> the underlying mechanisms. The objective<br />
of this study was therefore to simulate the process of<br />
oxidation, adsorption <strong>and</strong> cation exchange in injectionabstraction<br />
column studies, <strong>and</strong> to investigate to what<br />
68
5 Cation exchange during subsurface iron removal<br />
A<br />
Fe 2+<br />
Fe 2+ Fe 2+<br />
Fe Fe 2+<br />
2+<br />
Fe 2+ Fe 2+<br />
Fe 2+<br />
Fe 2+ 2+<br />
Fe 2+ Fe 2+<br />
Fe 2+<br />
Fe 2+<br />
Fe 2+<br />
Fe 2+<br />
Mineral Surface<br />
Fe 2+ Fe<br />
Fe 2+ Fe 2+ Fe 2+<br />
Fe 2+ Fe 2+<br />
Fe 2+<br />
Fe 2+<br />
Fe 2+<br />
Na + Ca 2+<br />
O Na +<br />
2<br />
O 2<br />
O<br />
Ca Na Ca 2+<br />
+<br />
2+<br />
2<br />
Na +<br />
Na Ca +<br />
2+<br />
Na + Ca 2+ Ca 2+ Ca 2+<br />
Na +<br />
Ca 2+<br />
Na + Na +<br />
Na +<br />
Ca 2+ Ca 2+<br />
Na +<br />
Fe 2+<br />
Fe 2+<br />
Mineral Surface<br />
Fe Fe Fe 2+<br />
2+<br />
2+<br />
Fe 2+<br />
Fe 2+<br />
Fe 2+ Fe 2+ Fe 2+<br />
B<br />
5<br />
O 2<br />
Ca 2+<br />
Na +<br />
Ca 2+<br />
Na O +<br />
2<br />
Ca 2+ O Na +<br />
2<br />
O 2<br />
O<br />
Na +<br />
Ca 2+ 2<br />
Ca 2+ Na<br />
Na +<br />
+<br />
Ca<br />
Ca 2+<br />
2+<br />
Na<br />
Na + +<br />
Ca Ca Fe(OH) 2+ 2+ Fe 2+<br />
Ca 2+<br />
3<br />
Fe 2+ Fe 2+<br />
Na +<br />
Fe 2+<br />
Ca 2+ Mineral Surface Ca 2+<br />
Na<br />
Fe 2+ Mineral Surface<br />
Fe 2+<br />
+<br />
Na +<br />
Fe 2+<br />
Na +<br />
Fe 2+ Fe 2+<br />
Ca 2+<br />
O Na Fe + 2+<br />
2 Fe(OH)<br />
Fe 3+<br />
Fe 3+ 3 Fe(OH)<br />
3<br />
Fe 2+<br />
Fe 2+<br />
Fe 2+ Fe2+<br />
Na<br />
Fe 2+<br />
+<br />
Na Fe Fe 2+<br />
2+<br />
+<br />
O 2<br />
O 2<br />
O 2<br />
Fe(OH)<br />
Ca 2+ 3<br />
Fe Fe 2+ 2+<br />
Fe Fe(OH) Fe 3+ 2+<br />
3<br />
O 2<br />
C<br />
Fe 3+<br />
Fe 2+<br />
Fe 2+<br />
Fe Fe 2+ 2+<br />
Figure 5.1 Schematic presentation of cation exchange during subsurface iron removal on the s<strong>and</strong> grain surface<br />
D<br />
69
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
5<br />
extent the occurrence of cation exchange influences the<br />
subsurface iron removal process. The column studies<br />
have been performed with a Na + exchanger during<br />
injection, either in the presence or absence of O 2<br />
,<br />
with synthetic groundwater <strong>and</strong> in the more complex<br />
environment of natural groundwater.<br />
Materials <strong>and</strong> methods<br />
The experimental set-up (Figure 5.2) consisted of<br />
duplicate transparent PVC columns with a length of<br />
30cm <strong>and</strong> an inner diameter of 36mm (wall thickness<br />
2mm). During all experiments, the columns were<br />
wrapped in aluminium foil to exclude light. The columns<br />
were filled with washed (24h with 1M HCl) filter s<strong>and</strong><br />
(500g; grain size = 0.5-0.8mm; D 10<br />
=0.58). The absence<br />
of other mineral structures than quartz on the s<strong>and</strong><br />
material was checked with X-ray Powder Diffraction<br />
(Bruker D5005; Brain PSD). The push-pull operational<br />
mode of injection-abstraction was simulated in the 1D<br />
plug-flow environment of the columns with down flow<br />
for injection <strong>and</strong> up flow abstraction (1.1 L.h -1 ±0.05).<br />
An injection-abstraction cycle started with 14 (±0.5)<br />
pore volumes of injection water, to allow for complete<br />
breakthrough of dissolved O 2<br />
. Subsequently the influent<br />
was switched to groundwater to allow retention of Fe 2+ .<br />
Electrical conductivity was used as a conservative tracer<br />
from which the pore volume could be calculated to be<br />
on average 0.12L (±0.002). The flow rate in the columns<br />
(2.7 m.h -1 ) was controlled with a multi-channel pump<br />
<strong>and</strong> PVC tubing with low gas permeability. Anoxic<br />
conditions were maintained in the columns by using<br />
an airtight FESTO system (6 x 1 PUN, I.D. 4mm) with<br />
matching connectors <strong>and</strong> valves.<br />
The experiments were executed (i) in the<br />
laboratory with synthetic groundwater <strong>and</strong> (ii)<br />
at a drinking water treatment plant with natural<br />
groundwater. At the start of each experiment the<br />
columns were conditioned with synthetic or natural<br />
groundwater, until complete breakthrough of iron<br />
occurred, <strong>and</strong> the E h<br />
potential stabilized. A normal<br />
injection mode consisted of demineralized water<br />
containing a pH buffer (5mM NaHCO 3<br />
) <strong>and</strong> 0.28mM<br />
O 2<br />
. However, to study the role of cation exchange,<br />
injection cycles have also been performed in the<br />
absence of O 2<br />
<strong>and</strong>/or Na + . The water quality during the<br />
abstraction phase was constant during the experiments,<br />
being either synthetic or natural groundwater.<br />
The synthetic groundwater was produced by<br />
sprinkling demineralized water on a 6m gas stripping<br />
column containing stainless steel Pall Rings. From the<br />
bottom pure N 2<br />
was blown into the degassing column<br />
to sparge out all O 2<br />
. Before entering the s<strong>and</strong> columns,<br />
the water was checked for O 2<br />
with the Orbiphere<br />
(HACH Lange; M1100 Sensor; 410 Analyser) to<br />
ensure concentrations below 1.5 µmol.L -1 . Addition of<br />
stock solutions for FeSO 4<br />
, NaHCO 3<br />
<strong>and</strong>/or NaCl was<br />
done with a dosing pump followed by a static mixer.<br />
70
5 Cation exchange during subsurface iron removal<br />
pH correction was achieved by addition of HCl or<br />
NaOH <strong>and</strong> all stock solutions were sparged with N 2<br />
in order to ensure the absence of O 2<br />
. The experiments<br />
were performed with synthetic groundwater of pH 6.9<br />
(±0.02), a temperature of 20°C (±0.1), Fe concentration<br />
of 0.1 mmol.L -1 (±0.01), pH buffer of 5mM NaHCO 3<br />
<strong>and</strong> ionic strength buffer of 1.6mM NaCl. To study the<br />
occurrence of cation exchange in the multicomponent<br />
groundwater matrix, the column set-up was transported<br />
to a groundwater treatment plant (WTP Loosdrecht,<br />
Vitens Water Supply Company). The groundwater,<br />
naturally containing Fe 2+ , was used as feed water for<br />
the column experiments. During the research period<br />
the groundwater had an average pH of 7.2 (±0.01), a<br />
constant temperature of 11 °C, 0.1 mmol Fe.L -1 (±0.01),<br />
3.3 μmol Mn.L -1 , 1.07 mmol Ca.L -1 , 0.52 mmol Na.L -1 ,<br />
0.28 mmol Si.L -1 , <strong>and</strong> 0.14 mmol SO 4<br />
.L -1 .<br />
Fe analysis of the water samples was done with<br />
an Atomic Absorption Spectrometer (Perkin-Elmer<br />
Flame AAS 3110). In-line measurements were done for<br />
dissolved oxygen (Orbisphere <strong>and</strong> WTW Cellox 325),<br />
Eh potential (WTW SenTix ORP), pH (WTW SenTix<br />
H 2<br />
O<br />
5<br />
pH ORP EC DO<br />
Waste/Sample<br />
Point<br />
N 2<br />
Stainless steel Pall Rings<br />
H 2<br />
O<br />
+<br />
O 2<br />
Discharge Pump<br />
Dosing Pump<br />
Column 2<br />
Column 1<br />
Fe/As<br />
NaHCO 3<br />
Figure 5.2 Schematic overview of experimental column set-up<br />
71
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
5<br />
41), <strong>and</strong> electrical conductivity (WTW TetraCon 325).<br />
Measurements were registered on a computer with<br />
Multilab Pilot v5.06 software.<br />
Results<br />
Injection-abstraction cycles<br />
A regular injection phase during subsurface iron<br />
removal consists of the injection of aerated water, in<br />
most cases drinking water from the clean water reservoir.<br />
In the experiments, demineralized water was used<br />
for injection containing a pH buffer (5mM NaHCO 3<br />
)<br />
<strong>and</strong> a 0.28mM O 2<br />
concentration. During injection<br />
the tracer passed the column at 1 pore volume (PV),<br />
but O 2<br />
concentrations were delayed in the columns,<br />
corresponding to an approximate O 2<br />
consumption<br />
of 0.15mmol in the synthetic groundwater columns<br />
(results not shown). Based on the stoichiometric ratio<br />
for Fe 2+ oxidation of 4, this could result in a total Fe<br />
removal of 0.6mmol for one cycle.<br />
To what extent iron is delayed compared to the<br />
groundwater determines the efficacy of the subsurface<br />
Fe removal technology. In general, groundwater<br />
treatment plants operate a V/Vi ratio based on the<br />
moment Fe starts to arrive at the well, thus (C/C 0<br />
) Fe<br />
>0.<br />
In the experiments, [Fe 2+ ] is allowed to breakthrough, in<br />
order to calculate the dimensionless retardation factor<br />
R. R Fe<br />
+1 is calculated from a Vi corresponding to the<br />
pore volume when the tracer (electrical conductivity)<br />
is C/C 0<br />
= 0.5, <strong>and</strong> V is the number of pore volumes that<br />
can be abstracted with iron concentrations below (C/<br />
C 0<br />
) Fe<br />
=0.5. In the case of a column study, the oxidation<br />
zone is limited to the size of the column, i.e., 1 pore<br />
volume, making the calculation of V/Vi redundant:<br />
R<br />
Well<br />
TracerC/ C0<br />
Equation 5.3<br />
⎛V<br />
⎞<br />
⎜ ⎟<br />
⎝Vi ⎠<br />
( PV )<br />
FeC/ C<br />
Fe<br />
0 C/<br />
C0<br />
= −1⇒ RColumn<br />
= −1<br />
⎛V<br />
⎞<br />
( PV )<br />
TracerC/ C0<br />
⎜ ⎟<br />
⎝Vi<br />
⎠<br />
The Fe retardation during abstraction for the regular<br />
injection-abstraction cycle is shown for both the<br />
synthetic <strong>and</strong> natural groundwater columns in Figure<br />
5.3. The tracer passed the columns at PV=1, but<br />
elevated Fe concentrations were not observed until<br />
30 <strong>and</strong> 8 pore volumes for the synthetic <strong>and</strong> natural<br />
columns, respectively. The Fe retardation in the natural<br />
groundwater columns (R Fe<br />
= 9) was measured to be<br />
lower than for synthetic water (R Fe<br />
= 42). The reduction<br />
in removal efficiency can be explained by the presence<br />
of competing cations in the natural groundwater,<br />
binding to the same sites as Fe 2+ . For instance, Ca 2+ is<br />
known to inhibit the Fe 2+ adsorption onto virgin <strong>and</strong><br />
iron hydroxide coated s<strong>and</strong> (Sharma, 2001).<br />
72
5 Cation exchange during subsurface iron removal<br />
(C/C0)Fe<br />
1<br />
0.5<br />
0<br />
R=9 R=42<br />
0 20 40 60 80<br />
pore volumes<br />
synthetic<br />
natural<br />
synthetic<br />
natural<br />
Figure 5.3 Fe breakthrough in duplicate columns after injection<br />
of water containing 0.28mM O 2<br />
<strong>and</strong> 5mM NaHCO 3<br />
for columns loaded with synthetic <strong>and</strong> natural<br />
groundwater (C 0<br />
=0.1mM Fe 2+ )<br />
When considering the synthetic groundwater column<br />
results it was calculated from the retardation factor<br />
(R Fe<br />
= 42) that the Fe retention was approximately 0.6<br />
mmol/cycle. This value correlates well to the 0.6 mmol<br />
Fe calculated from the O 2<br />
consumption in the columns.<br />
Based on this finding it may be concluded that all O 2<br />
consumption was used for Fe 2+ oxidation, however,<br />
there is still the question of whether cation exchange<br />
plays a (catalyzing) role. Adsorbed Fe 2+ could either<br />
directly (heterogeneously) oxidize on the surface,<br />
like proposed with the catalytic adsorptive-oxidation<br />
mechanism, or it could exchange with other cations<br />
before oxidizing to Fe 3+ hydroxides. During a regular<br />
injection-abstraction cycle, the Fe:O 2<br />
mass balance<br />
does not differentiate between these two mechanisms,<br />
as Fe 2+ will always oxidize in the presence of O 2<br />
<strong>and</strong> be<br />
retained in the columns.<br />
Fe 2+ -Na + exchange<br />
After reaching steady state with the synthetic<br />
groundwater the columns were loaded for an injection<br />
phase without O 2<br />
, but with elevated Na + concentrations.<br />
Na + was added as 0.01M NaHCO 3<br />
or 0.5M NaCl.<br />
Additionally a control cycle was tested without the<br />
addition of any Na + . It is noteworthy that the pH during<br />
injection with water containing low buffer capacities<br />
(absence of HCO 3-<br />
) remained above pH 7.5 <strong>and</strong> the pH<br />
drop lasted for a maximum of 3 pore volumes. Figure<br />
5.4 shows the Fe concentrations that were measured<br />
during the injection phases. The dotted lines represent<br />
the results from the geochemical surface complexation<br />
model PHREEQC (v2.15; Parkhurst <strong>and</strong> Appelo, 1999).<br />
Fe desorption was measured from the column material<br />
at concentrations 1.5 to 25 times the C 0<br />
of 0.1mM. The<br />
PHREEQC model results even show a higher desorbed<br />
concentration, indicating that during sampling the<br />
high peak concentration was missed. The amount<br />
of exchanged Fe was higher in the case of 0.5M Na + ,<br />
resulting in subsequent higher removal efficiencies<br />
during abstraction (Figure 5.4). R Fe<br />
was measured to<br />
be 11 <strong>and</strong> 30, for 0.01M <strong>and</strong> 0.5M Na + , respectively.<br />
The same experiment was conducted with 0.1M NaCl,<br />
showing the same retardation as for 0.5M NaCl (not<br />
shown, confirmed with PHREEQC), which means<br />
5<br />
73
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
5<br />
that maximum Na+ regeneration has occurred at 0.1M<br />
NaCl.<br />
The retardation of Fe during this particular cycle<br />
provides information on the exchangeable Fe 2+ capacity<br />
of the s<strong>and</strong>, as retention of approximate 0.4 mM Fe can<br />
be recalculated to a Cation Exchange Capacity (CEC)<br />
of 1.55 meq.kg -1 (0.78 mmol Fe.kg -1 ) for the column<br />
A<br />
mM Fe<br />
B<br />
(C/C0)Fe<br />
0.01M NaHCO3 0.01M NaHCO3 Phreeqc<br />
no Na no Na Phreeqc<br />
0.5M NaCl 0.5M NaCl Phreeqc<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0<br />
1<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0<br />
Injection<br />
0 2 4 6 8 10<br />
pore volumes<br />
Abstraction<br />
0 20 40 60<br />
pore volumes<br />
Figure 5.4 Fe measurements during injection of 0.01M NaHCO 3<br />
,<br />
Na-free or 0.5M NaCl without O 2<br />
<strong>and</strong> abstraction of<br />
synthetic groundwater (C 0<br />
=0.1mM Fe 2+ ), dotted lines<br />
represent results from PHREEQC (v2.15)<br />
8<br />
6<br />
4<br />
2<br />
0<br />
mM Fe (0.5M NaCl)<br />
s<strong>and</strong>. This CEC value is lower than for average aquifer<br />
s<strong>and</strong> (10 meq.kg -1 ; Appelo <strong>and</strong> Postma, 2005), as can be<br />
expected based on the absence of clay, organic carbon<br />
<strong>and</strong> iron oxides on this clean (silica) filter s<strong>and</strong>. The CEC<br />
value correlates well with the Fe 2+ adsorption capacity<br />
of virgin s<strong>and</strong> of 0.74 mmol Fe.kg -1 observed by Sharma<br />
et al. (1999). In that study the adsorption of Fe 2+ onto<br />
clean s<strong>and</strong> (D=0.7-1.25mm) at pH 7.0 in the absence<br />
of oxygen was investigated. When using the measured<br />
exchange value of 1.55 meq.kg -1 in PHREEQC, the<br />
measured results were simulated well. These results<br />
show that also in the absence of O 2<br />
, but in the presence<br />
of abundant Na + , a retardation of Fe can be achieved to<br />
a certain level. The desorption of Fe during injection<br />
shows the exchange of attached Fe 2+ with soluble Na + ,<br />
<strong>and</strong> subsequent retardation of Fe points towards the<br />
vise versa exchange of Fe 2+ <strong>and</strong> Na + . The occurrence<br />
of Fe 2+ -Na + exchange, even on clean filter s<strong>and</strong>, can<br />
potentially play an important role during subsurface<br />
iron removal. However, this depends strongly on cation<br />
concentrations in the injection water <strong>and</strong> on the CEC<br />
of the aquifer material. The control cycle with Na + -free<br />
water confirmed that the Fe 2+ exchange was very low<br />
during injection (Figure 5.4). Fe was subsequently not<br />
significantly delayed in the columns, with an R Fe<br />
of 1.<br />
CIEX in natural groundwater<br />
The role of CIEX in the multi-component environment<br />
of natural groundwater was studied in experiments<br />
using groundwater from a water treatment plant<br />
74
5 Cation exchange during subsurface iron removal<br />
(WTP Loosdrecht) instead of synthetic groundwater.<br />
Under these natural conditions the injection cycles<br />
in the presence <strong>and</strong> absence of O 2<br />
<strong>and</strong>/or Na + were<br />
investigated. Figure 5.5A depicts the Fe desorption<br />
during the injection in the absence of oxygen. A peak<br />
concentration almost reaching the initial concentration<br />
of 0.1mM was detected. At 0.1M NaCl the Fe leaching<br />
A<br />
mM Fe<br />
0.1<br />
0.08<br />
0.06<br />
0.04<br />
Injection<br />
5mM NaHCO3<br />
5mM NaHCO3<br />
0.1M NaCl<br />
0.1M NaCl<br />
no Na<br />
no Na<br />
from the columns was higher than for 5mM Na + ,<br />
resulting in a slightly better Fe retardation during the<br />
abstraction phase (Figure 5.5). The Fe desorption after<br />
injection in the absence of Na + was most likely caused<br />
by the cations in the natural groundwater present in the<br />
columns just before injection. It is noteworthy that the<br />
leaching in these natural groundwater columns was up<br />
to 20 times lower than in the synthetic groundwater<br />
columns, as can be seen from Figure 5.4 <strong>and</strong> Figure<br />
5.5, not exceeding the background Fe concentration of<br />
0.1mM.<br />
The Fe retardation factors that were measured<br />
in all experiments for both synthetic <strong>and</strong> natural<br />
5<br />
B<br />
(C/C0)Fe<br />
0.02<br />
0<br />
1<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0<br />
0 2 4 6 8 10 12<br />
pore volumes<br />
Abstraction<br />
0 5 10 15<br />
pore volumes<br />
Retardation factor (R Fe)<br />
(A) 5mM NaHCO3/0.28mM O2<br />
50<br />
synthetic natural<br />
40<br />
30<br />
20<br />
10<br />
0<br />
(B) 5mM NaHCO3* (no O2)<br />
(C) 0.1M NaCl (no O2)<br />
(D) no Na/O2<br />
(E) 0.28mM O2 (no Na)<br />
Injection water quality<br />
* for synthetic 10mM NaHCO3 was used<br />
Figure 5.5 Fe measurements during (A) injection of 0.01M<br />
NaHCO 3<br />
, 0.5M NaCl or Na-free water without O 2<br />
<strong>and</strong> (B) abstraction with natural groundwater<br />
Figure 5.6 The measured Fe retardation factors for cycles with<br />
different injection modes with abstraction of either<br />
synthetic or natural groundwater<br />
75
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
5<br />
groundwater are summarized in Figure 5.6, showing<br />
the decrease in efficiency in the natural groundwater<br />
columns compared to the synthetic groundwater<br />
columns. The R Fe<br />
after injection with or without<br />
oxygen show a reduced value in the multi-component<br />
environment of natural groundwater (injection modes<br />
A, B, C, E). The Fe 2+ exchange/adsorption is clearly<br />
limited by the presence of other cations, such as Ca 2+ ,<br />
resulting in retardation values below 12. In the absence<br />
of oxygen, the R Fe<br />
was somewhat similar for both<br />
5mM <strong>and</strong> 0.1M Na + , i.e., CIEX between Na + <strong>and</strong> Fe 2+<br />
is not significantly enhanced at higher Na + . Also in the<br />
absence of Na + during injection, the Fe 2+ retardation was<br />
limited in natural groundwater compared to synthetic<br />
water (injection mode E), as the R Fe<br />
dropped from 33 to<br />
8 in the multi-component environment of the natural<br />
groundwater columns. It should be noted that although<br />
R Fe<br />
between injection modes A, C <strong>and</strong> E correlate well,<br />
the Fe breakthrough trend looks different (not shown).<br />
After injection of Na-free water, the Fe concentrations<br />
never reached below the detection limit, i.e., some<br />
Fe always passed the columns. This was observed<br />
in the synthetic <strong>and</strong> natural groundwater columns,<br />
indicating that some CIEX is needed to reach ultra low<br />
Fe concentrations at the beginning of an abstraction<br />
phase.<br />
In Figure 5.7 the measurements for Na <strong>and</strong><br />
Ca are depicted, showing some desorption of Ca<br />
during injection (Figure 5.7A). For the duration of 9<br />
pore volumes, the Na concentration increased during<br />
injection from the concentration in the groundwater<br />
(C/C 0<br />
=0.12) to the concentration in the injection water<br />
(C/C 0<br />
=1). The tailing of both curves indicates that<br />
CIEX proceeds between these cations. The abstraction<br />
period shows the reversed trend (Figure 5.7B), as the Na<br />
concentration of the groundwater (C 0<br />
) is reached after<br />
approximately 8 pore volumes. The Ca concentration<br />
is already restored after 6 pore volumes. It should be<br />
A<br />
B<br />
(C/C0)Na,Ca<br />
(C/C0)Na,Ca<br />
1<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0<br />
1<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0<br />
Na<br />
Na<br />
Ca<br />
Ca<br />
0 2 4 6 8 10<br />
pore volumes<br />
0 5 10 15<br />
pore volumes<br />
Figure 5.7 Na <strong>and</strong> Ca measurements during an (A) injection<br />
<strong>and</strong> (B) abstraction cycle with natural groundwater<br />
76
5 Cation exchange during subsurface iron removal<br />
noted that Ca 2+ complexation with natural organic<br />
matter (NOM) may have occurred as well, but because<br />
of the low binding contact time can be considered<br />
insignificant. It may be concluded that CIEX occurs,<br />
in the multi-component environment of natural<br />
groundwater, between a wide range of cations, resulting<br />
in retardation during the abstraction phase. The effect<br />
of CIEX is most predominant on Fe retardation, as Fe<br />
concentrations in the groundwater are relatively low<br />
compared to Ca 2+ <strong>and</strong> Na + .<br />
Fe 2+ desorption theory<br />
The column studies in the absence of O 2<br />
<strong>and</strong>/or Na +<br />
confirmed that, in addition to the adsorptive-oxidation<br />
process (van Beek, 1985; Rott, 1985), CIEX occurs<br />
during subsurface iron removal. The Na + concentrations<br />
of actual injection water will always be lower than the<br />
concentrations in the experiments. The experiments<br />
therefore overestimate the actual occurrence of CIEX<br />
during subsurface iron removal. However, the CEC of<br />
clean filter s<strong>and</strong> in the columns is much lower than that<br />
of actual aquifer s<strong>and</strong> (on average 10 times) resulting in<br />
lower CIEX. At the beginning of a full-scale injection<br />
cycle, the CIEX will happen in the presence of O 2<br />
,<br />
resulting in immediate oxidation. While injection<br />
proceeds, the injection water will penetrate further into<br />
the aquifer, but the O 2<br />
front will lag behind, caused by<br />
the consumption of O 2<br />
during Fe 2+ oxidation. In other<br />
words, the injected water front does not contain any O 2<br />
,<br />
but contains cations for exchange with Fe 2+ . The further<br />
the injection water flows into the aquifer, the larger the<br />
distance between the injected water front <strong>and</strong> the O 2<br />
front becomes. In this moving zone where O 2<br />
is absent,<br />
the process of CIEX prevails <strong>and</strong> Fe 2+ desorbs from the<br />
soil material <strong>and</strong> travels deeper into the aquifer (Figure<br />
5.4A <strong>and</strong> Figure 5.5A).<br />
In theory, these elevated Fe 2+ concentrations<br />
never come in contact with the O 2<br />
in the injection water<br />
<strong>and</strong> will not oxidize in this cycle (Figure 5.8). When<br />
abstraction starts <strong>and</strong> the flow direction is reversed,<br />
the desorbed Fe 2+ passes the available adsorption sites<br />
on the soil grains closer to the production well <strong>and</strong> is<br />
removed through either adsorption or CIEX. In other<br />
words, the injection phase of subsurface iron removal<br />
mobilizes a part of the adsorbed Fe 2+ through CIEX<br />
but does not subsequently oxidize the Fe 2+ . One may<br />
state that this proportion of mobilized Fe 2+ limits the<br />
system’s efficacy during the abstraction phase, as part<br />
of the adsorption sites will be occupied by the desorbed<br />
C<br />
C 0<br />
1<br />
0.5<br />
Separation of Fe 2+ <strong>and</strong> O2<br />
O2 front<br />
Injection<br />
water<br />
Fe 2+<br />
desorption<br />
distance from the well<br />
Figure 5.8 Schematic representation of the separation between<br />
O 2<br />
front <strong>and</strong> Fe 2+ desorption peak<br />
77<br />
5
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
5<br />
Fe 2+ <strong>and</strong> not by the Fe 2+ present in the groundwater.<br />
On the other h<strong>and</strong>, the exchange may also enhance<br />
SIR, as the desorbed Fe 2+ is pushed deeper into the<br />
aquifer <strong>and</strong> concentrations may be lowered through the<br />
buffering capacity of the aquifer material. Whether the<br />
Fe 2+ truly enhances or limits the efficacy of SIR is yet<br />
to be determined, because it all depends on the actual<br />
separation between the Fe 2+ desorption front <strong>and</strong> the<br />
O 2<br />
front. The investigation of this front separation<br />
was not the focus of these column experiments <strong>and</strong> is<br />
recommended for future research.<br />
Injection elevated O 2<br />
concentrations<br />
CIEX may play a role during subsurface iron removal,<br />
but it is the supply of O 2<br />
to the (im)mobile Fe 2+ that<br />
determines the efficiency of the system. Appelo et<br />
al. (1999) concluded that increasing the oxidant<br />
concentration of the injected water would be useless<br />
as long as the efficiency is limited by the amount of<br />
exchangeable Fe 2+ capable of consuming the oxidant.<br />
The observation that injected oxygen will not be used<br />
in the absence of available oxidizeable Fe 2+ is valid,<br />
but there is also a reason why injection of higher O 2<br />
concentrations can increase the system’s efficacy.<br />
Namely, if the injection water contains higher O 2<br />
concentrations, the O 2<br />
front will not lag far behind the<br />
injected water front. The desorbed Fe 2+ will then come<br />
in contact with O 2<br />
for oxidation <strong>and</strong> not leach out of the<br />
oxidation zone into the aquifer. Thus by injecting higher<br />
oxidant concentrations, the exchangeable Fe 2+ fraction<br />
V/Vi<br />
25<br />
20<br />
15<br />
10<br />
5<br />
0<br />
1 2 3 4 5 6 7 8 9 10 11 12<br />
well #<br />
0.28mM O2<br />
0.55mM O2<br />
Figure 5.9 The operational V/Vi ratio of 12 wells at WTP Corle<br />
with (0.55mM) <strong>and</strong> without (0.28mM) the injection<br />
of 47mol.m -3 pure oxygen<br />
on the soil material can be utilized for oxidation <strong>and</strong><br />
will contribute as hydrolyzed Fe 3+ hydroxides with<br />
their exchangeable/adsorptive surface area during the<br />
following abstraction phase.<br />
At water treatment plant Corle (Vitens Water<br />
Supply Company) there has been extensive experience<br />
with the injection of elevated oxygen concentrations<br />
into subsurface iron removal wells. In the past they<br />
injected 3,000m 3 of drinking water containing 0.28mM<br />
O 2<br />
, but they have changed the operational mode to<br />
the injection of 2,000m 3 with an O 2<br />
concentration of<br />
0.55mM. Although the O 2<br />
concentration increased with<br />
a factor of 1.9, the total O 2<br />
injection increased only with<br />
a factor 1.3, from 0.84 to 1.1 *10 3 mol. As a result of this<br />
operational change, the volumetric ratio for abstraction<br />
(V) <strong>and</strong> injection (Vi) has increased from an average<br />
7 to 16. This immense <strong>and</strong> sudden efficiency increase<br />
78
5 Cation exchange during subsurface iron removal<br />
was clearly caused by the operational change, as an<br />
increasing efficacy caused by successive cycles (Appelo<br />
et al., 1999) may not be expected at stablized subsurface<br />
iron removal wells. WTP Corle calculates V/Vi based<br />
on the moment when Fe breakthrough starts, so the<br />
moment [Fe]>2µM is registered.<br />
Figure 5.9 depicts the relation of the operational<br />
parameter V/Vi for 12 subsurface iron removal<br />
wells after injection of 0.28mM O 2<br />
<strong>and</strong> 0.55mM O 2<br />
.<br />
Although there was some variation between the results,<br />
considering these are operational data from 12 different<br />
wells, it can be concluded that on average the V/Vi<br />
increases by an approximate factor 2. This indicates that<br />
not the available (adsorbed <strong>and</strong>/or exchangeable) Fe 2+<br />
is limiting during injection, but the supply of O 2<br />
to the<br />
(in)soluble Fe 2+ . The field measurements support the<br />
theory that increasing the oxidant concentration has a<br />
positive effect on the subsurface removal of Fe, but they<br />
do not prove that it is actually the Fe 2+ desorption front<br />
that is targeted by the higher O 2<br />
dose. The field results<br />
point towards the assumption that Fe 2+ is abundantly<br />
available on the soil material. In conclusion, CIEX may<br />
occur in the aquifer during injection, but does not seem<br />
to limit the system’s efficiency when injecting elevated<br />
oxygen concentrations.<br />
Conclusions<br />
In s<strong>and</strong> column experiments with synthetic <strong>and</strong> natural<br />
groundwater it was found that cation exchange (Na + -<br />
Fe 2+ ) occurs during the injection-abstraction cycles of<br />
subsurface iron removal. The Fe 2+ exchange increased<br />
at higher Na + concentration in the injection water,<br />
but decreased in the presence of other cations in the<br />
groundwater. Field results with injection of high O 2<br />
concentrations indicated that not the exchangeable<br />
Fe 2+ on the soil material is the limiting factor during<br />
injection, but it is the supply of O 2<br />
to the available Fe 2+ .<br />
References<br />
Appelo C. A. J., B. Drijver, R. Hekkenberg <strong>and</strong> M. de Jonge<br />
(1999) Modeling in situ iron removal from ground water.<br />
Ground Water 37(6): 811-817.<br />
Appelo C.A.J. <strong>and</strong> D. Postma (2005) Geochemistry, groundwater<br />
<strong>and</strong> pollution. Balkema, Rotterdam, 2nd edition.<br />
Parkhurst D.L. <strong>and</strong> C.A.J. Appelo (1999) User’s guide to phreeqc<br />
(version 2) – a computer program for speciation, batchreaction,<br />
one-dimentional transport, <strong>and</strong> inverse geochemical<br />
calculations. Water-Resources Inverstigation Report 99-4259,<br />
US Geological Survey.<br />
Rott U., (1985) Physical, chemical <strong>and</strong> biological aspects of the<br />
removal of iron <strong>and</strong> manganese underground. Water Supply<br />
3(2): 143-150.<br />
79<br />
5
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
5<br />
Rott U., C. Meyer <strong>and</strong> M. Friedle (2002) Residue-free removal<br />
of arsenic, iron, mangenese <strong>and</strong> ammonia from groundwater.<br />
Water Science <strong>and</strong> Technology: Water Supply 2(1): 17-24.<br />
Sharma S.K., M.R. Greetham, J.C. Schippers (1999) Adsorption<br />
of iron(II) onto filter media. J Water SRT-Aqua 48(3): 84-91.<br />
Sharma S.K. (2001) Adsorptive iron removal from groundwater.<br />
PhD dissertation, Wageningen University.<br />
van Beek C. G. E. M. (1985) Experiences with underground<br />
water treatment in the Netherl<strong>and</strong>s. Water Supply 3(2): 1-11.<br />
van Halem D., S. Olivero, W.W.J.M. de Vet, J.Q.J.C. Verberk,<br />
G.L. Amy <strong>and</strong> J.C. van Dijk (2010) <strong>Subsurface</strong> iron <strong>and</strong><br />
arsenic removal for shallow tube well drinking water in rural<br />
Bangladesh. Water Research 44: 5761-5769.<br />
Wolthoorn A. (2003) <strong>Subsurface</strong> aeration of anaerobic<br />
groundwater; iron colloid formation <strong>and</strong> the nitrification<br />
process. Ph.D. dissertation, Wageningen University.<br />
80
6<br />
Characterization of accumulated<br />
deposits<br />
This chapter is based on:<br />
van Halem et al. (2011) Applied Geochemistry 26: 116-124
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
6<br />
Introduction<br />
<strong>Subsurface</strong> iron removal has been implemented since<br />
the 1970s in Europe (Hallberg <strong>and</strong> Martinell, 1976;<br />
Mettler, 2002), nevertheless the technology has not<br />
yet found widespread application elsewhere. The iron<br />
removal occurs in the aquifer <strong>and</strong> is therefore not<br />
as visible as conventional above-ground rapid s<strong>and</strong><br />
filtration, resulting in concerns regarding the longterm<br />
sustainability. Clogging of the aquifer is an issue<br />
that is raised frequently, however, in literature there is<br />
an agreement that clogging of the aquifer does not pose<br />
a serious threat to subsurface iron removal (Appelo<br />
et al., 1999; Braester <strong>and</strong> Martinell, 1988; Grombach,<br />
1985; van Beek, 1985). The absence of aquifer clogging<br />
has been proposed to be caused by (Appelo et al., 1999;<br />
Rott et al., 2002; van Beek 1985):<br />
• decreasing porosity with time, leading to further<br />
infiltration of injection water into the aquifer;<br />
• iron precipitation at different infiltration distances, i.e.,<br />
variable size of the oxidation zone;<br />
• iron deposits predominantly formed in dead-end pores<br />
or stagnant zones;<br />
• transformation of voluminous amorphous iron<br />
hydroxides to less voluminous, crystalline hydroxides.<br />
Although previous publications have not reported<br />
aquifer clogging around subsurface iron removal wells,<br />
doubts on the sustainability of this technology still limit<br />
the widespread application. However, this inexpensive<br />
technology holds great promise for the developed<br />
<strong>and</strong> developing country context, with the potential<br />
application for arsenic (van Halem et al., 2010) <strong>and</strong><br />
manganese removal.<br />
<strong>Iron</strong> hydroxide deposits near subsurface<br />
treatment wells have been previously investigated<br />
in literature. Braester <strong>and</strong> Martinell (1988) have<br />
been unable to microscopically identify any severe<br />
precipitation at the oldest Swedish plant built in 1971.<br />
In some samples precipitates were interpreted as<br />
bacterial iron stalks, but also mineral transformation<br />
into haematite has been observed. Oxalate/dithionite<br />
extractions of aquifer material near subsurface treatment<br />
wells have shown that half of the iron hydroxides were<br />
poorly crystalline <strong>and</strong> X-ray diffraction showed traces<br />
of lepidocrocite (Olthoff, 1986). Mettler et al. (2001)<br />
characterized the iron precipitates close to a subsurface<br />
iron removal well in Switzerl<strong>and</strong>. With chemical<br />
extraction <strong>and</strong> 57 Mössbauer spectroscopy they found<br />
that iron was mainly deposited as ferric hydroxides <strong>and</strong><br />
consisted for 50-100% of goethite.<br />
The mechanism of adsorptive-catalytic oxidation<br />
would suggest the formation of neatly ordered Fe 3+<br />
hydroxides, rather than voluminous amorphous iron<br />
sludge. Also, cation exchange (Appelo, et al., 1999),<br />
Interfactial Electron Transfer (Jeon et al., 2001; Mettler<br />
2002) <strong>and</strong> recrystallization (Pederson et al., 2005) have<br />
been proposed to occur during subsurface iron removal<br />
resulting in different hypotheses on iron precipitate<br />
accumulation with time. The main objective of this study<br />
was to identify <strong>and</strong> characterize iron accumulation<br />
82
6 Characterization of accumulated deposits<br />
deposits in the aquifer near subsurface iron removal<br />
wells to assess the sustainability regarding clogging of<br />
the aquifer. An additional objective was to investigate<br />
the potential co-accumulation of other groundwater<br />
constituents, such as arsenic <strong>and</strong> manganese.<br />
Materials <strong>and</strong> methods<br />
Soil sample collection<br />
Oasen Drinking Water Supply Company, in<br />
the Netherl<strong>and</strong>s, has extensive experience with<br />
groundwater-related operational modes, such as<br />
subsurface treatment <strong>and</strong> (river) bank filtration (de Vet<br />
et al., 2009). The production wells 03 <strong>and</strong> 04 at Oasen<br />
Water Treatment Plant (WTP) “De Put” have been<br />
intermittently used for subsurface iron removal since<br />
1996. Injection of 1,500-2,200 m 3 drinking water was<br />
done every other month, resulting in cycles of 2 months.<br />
Both wells were in operation for one month <strong>and</strong> stood<br />
still for the other month, while the other well was in<br />
operation. During operation, water is abstracted at a<br />
rate of ±1,250-1,600 m 3 .day -1 , for subsequent treatment<br />
with bios<strong>and</strong> filters. The average groundwater quality of<br />
the reference wells is summarized in Table 6.1.<br />
The accumulation of iron <strong>and</strong> other constituents<br />
onto aquifer soil material during subsurface iron<br />
removal was assessed by analysis of drilled samples<br />
from aquifer material. Bore holes (SonicMast with<br />
SonicDrill 2x7) were drilled at 5m from injection/<br />
abstraction wells 03 <strong>and</strong> 04, <strong>and</strong> between the two wells<br />
(at ±50m from both wells, Figure 6.1). The borehole<br />
between well 03 <strong>and</strong> 04 can be used as a reference,<br />
since, at 50m, it was outside the affected oxidation zone.<br />
800mL samples were taken every meter at the depths<br />
from 10 to 28 meters below ground level. Classification<br />
of the drilled material showed that at approximately -9<br />
<strong>and</strong> -12m for well 03 <strong>and</strong> 04, respectively, the clay layer<br />
ended <strong>and</strong> coarse s<strong>and</strong> material (brownish) started. The<br />
drillings showed similar variations, with greyish coarse<br />
s<strong>and</strong> from a depth of 20-25 <strong>and</strong> 22-28 m for wells 03 <strong>and</strong><br />
04, respectively. The lower clay layer started at 26-27<br />
m, showing that the wells were drilled into a confined<br />
aquifer. The depth of the perforated well filters was 12-<br />
6<br />
Table 6.1 Average water quality parameters of groundwater at Oasen WTP “De Put” (reference wells 1992-2008)<br />
pH Fe Mn Ca Cl Mg Na HCO 3<br />
SO 4<br />
NH 4<br />
Si As Pb Ni<br />
µmol.L -1 mmol.L -1 nmol.L -1<br />
7.3 47 11 2.1 2.7 0.46 2.0 3.9 0.5 0.2 0.2 44 1.3 26<br />
83
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
6<br />
5m<br />
-12m<br />
-27m<br />
injection/abstraction wells<br />
50m<br />
bore holes<br />
50m<br />
Well 03 Well 04<br />
-15m<br />
-26m<br />
Figure 6.1 Schematic overview of injection/abstraction wells <strong>and</strong><br />
drilled bore holes<br />
27 m <strong>and</strong> 15-26 m for wells 03 <strong>and</strong> 04, respectively. The<br />
soil samples were taken in September 2008, subsurface<br />
treatment started in 1996 <strong>and</strong> had been operated for 12<br />
years at the moment of sampling.<br />
5m<br />
Chemical extraction<br />
Depending on the subsequent analyses, the samples<br />
were oven dried at 40°C or 105°C. The organic matter<br />
was measured after ignition at 450°C (Craft, 1988), with<br />
an average of all samples of 0.21%(±0.15), 0.25%(±0.13),<br />
<strong>and</strong> 0.23%(±0.14), for the reference well, well 03 <strong>and</strong><br />
well 04, respectively. Samples from every meter depth<br />
were analysed in a certified laboratory with ICP-MS<br />
scan after chemical extraction with: Acid-Oxalate for<br />
amorphous iron <strong>and</strong> manganese; or, Nitric Acid for<br />
total iron <strong>and</strong> a wide range of elements (including,<br />
Na, Mg, Al, Si, K, Ca, Mn, Sr, Ni, Sc, Ti, As). Acid-<br />
Oxalate extraction consisted of reductive dissolution<br />
of amorphous iron/manganese with an oxalate buffer<br />
solution which consisted of 0.2M ammonium oxalate<br />
<strong>and</strong> 0.2M oxalic acid (at pH=3). During digestion the<br />
sample was agitated <strong>and</strong> light was excluded.<br />
The samples at 19-20, 20-21 <strong>and</strong> 21-22m depth<br />
in the reference well <strong>and</strong> near well 04 were oven dried at<br />
40°C <strong>and</strong> sieved (mesh sizes 00.63, 0.106, 0.125, 0.150,<br />
0.250, 0.50, 1.0, <strong>and</strong> 2.0mm). Duplicate samples between<br />
0.02 to 0.7 mg were taken from every sieve fraction for<br />
chemical extraction (1) with 5M Hydrochloric Acid for<br />
24 hours to extract total iron (Heron et al., 1994), <strong>and</strong><br />
(2) with 1M Hydrochloric Acid addition to 50mL of<br />
demineralised water to dissolve carbonates at pH 3.5-<br />
4.0 (Kroetsch <strong>and</strong> Wang, 2006). Once the samples were<br />
diluted they were filtered with 0.2μm filters (cellulose<br />
acetate) for total iron analysis with Flame Atomic<br />
Absorption Spectrometer (Perkin-Elmer AAS 3110).<br />
84
6 Characterization of accumulated deposits<br />
ESEM-EDX<br />
Sieved soil samples (fraction
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
6<br />
production. The stoichiometric ratio of injected oxygen<br />
versus removed iron is theoretically 1:4, at WTP “De<br />
Put” this ratio was between 1:3.3 <strong>and</strong> 1:3.8 in the year<br />
2006-2007, meaning that 81 to 94% of the injected<br />
oxygen was consumed for iron oxidation. The amount<br />
of iron that has been retained in the subsurface during<br />
12 years of operation can be estimated based on the<br />
historical water quality data. The average iron removal<br />
per cycle is an estimated 1.4•10 3 mol/cycle <strong>and</strong> 1.6•10 3<br />
mol/cycle for well 03 <strong>and</strong> 04, respectively. Although<br />
there is no difference in operation between the wells,<br />
well 04 performs better than well 03. Nevertheless, it<br />
can be roughly calculated that in a period of 12 years<br />
approximately 1•10 5 mol of iron has been retained in<br />
the subsurface for each well.<br />
Figure 6.3 shows the results for total iron<br />
extraction with Nitric Acid on the soil samples from<br />
[Fe] mmol Fe.L -1<br />
0.06<br />
0.04<br />
0.02<br />
0.00<br />
21.8.06 29.11.06 9.3.07 17.6.07 25.9.07 3.1.08 12.4.08 21.7.08<br />
Well 03 Well 04 Reference<br />
Figure 6.2 Fluctuating iron concentration pattern in the<br />
produced groundwater in wells 03 <strong>and</strong> 04 (year:<br />
2006-2008)<br />
a depth of 12 to 28m at a distance of 5m from wells<br />
03 <strong>and</strong> 04 <strong>and</strong> in the reference bore hole. The graph<br />
indicates that iron accumulation was found in distinct<br />
soil layers near both wells. Near well 03, elevated iron<br />
levels were found at a depth of 12 to 15m <strong>and</strong> from 23<br />
to 26m. The difference between the reference value <strong>and</strong><br />
the sample at 5m from this well varied between 11.5<br />
<strong>and</strong> 271.7 mmol Fe.(kg.ds) -1 . Compared to well 04, less<br />
iron seems to have accumulated near well 04, which<br />
corresponds to the water quality data in Figure 6.2.<br />
The iron accumulation near well 04 was mainly found<br />
at a depth from 19 to 21m, where iron concentrations<br />
were between 12.1 <strong>and</strong> 390.8 mmol.(kg.ds) -1 higher<br />
than the reference bore hole. The accumulation of iron<br />
at specific depths could indicate that preferred flow<br />
lines were followed by the injected water, e.g., due to<br />
higher permeability. Certain soil layers may have a<br />
lower resistance both during injection <strong>and</strong> abstraction,<br />
resulting into a greater contribution to the subsurface<br />
treatment system than layers with other properties.<br />
However, based on the classification of the aquifer<br />
material (coarser s<strong>and</strong> or more silt/clay; data not<br />
shown) no relation was found between soil layers <strong>and</strong><br />
iron accumulation.<br />
The Acid-Oxalate extraction method was used<br />
to dissolve only the amorphous or poorly crystalline<br />
iron hydroxides in the drilled soil samples. The results<br />
in Figure 6.3 show increased iron concentrations<br />
compared to the reference level near both wells. The<br />
soil layers where iron accumulated were more or less<br />
86
6 Characterization of accumulated deposits<br />
similar to Figure 6.3, indicating that indeed the injected<br />
water affected mainly certain soil layers. Compared to<br />
the reference samples, amorphous iron has accumulated<br />
up to 35.0 <strong>and</strong> 62.2 mmol.(kg.ds) -1 for well 03 <strong>and</strong> 04,<br />
respectively.<br />
Based on the assumption that all iron accumulates<br />
as iron (oxy)hydroxides, it can be calculated what<br />
portion of accumulated iron is amorphous or crystalline:<br />
Fe crys<br />
=Fe total<br />
-Fe amorph<br />
. The percentage of amorphous <strong>and</strong><br />
crystalline iron that has accumulated over 12 years in<br />
12<br />
mmol Fe.(kg.ds) -1<br />
0 200 400 600 800<br />
the affected layers was found to be 56-100% crystalline<br />
near well 03 <strong>and</strong>, 67-100% near well 04. In order to<br />
ascertain the limited contribution of iron in carbonates<br />
on the soil grains, the carbonates were dissolved at pH<br />
3.5-4.0 with Hydrochloric Acid at the different mesh<br />
sizes (Kroetsch <strong>and</strong> Wang, 2006). All samples taken<br />
in the reference well <strong>and</strong> near well 04 between 19 <strong>and</strong><br />
22m confirmed that the iron release during carbonate<br />
removal was very small at
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
of crystalline deposits excludes the accumulation of<br />
massive iron sludge volumes in the subsurface. The<br />
presence of amorphous deposits indicates that iron<br />
initially (partially) precipitates as amorphous iron, but<br />
transforms to more crystalline iron hydroxides with<br />
time.<br />
Grain size vs. iron accumulation<br />
The samples at depths 19-20, 20-21 <strong>and</strong> 21-22m near<br />
well 04 <strong>and</strong> in the reference well were selected for<br />
sieve analyses <strong>and</strong> subsequent extraction with 5M<br />
Hydrochloric Acid, as these soil layers were clearly<br />
affected by subsurface iron removal. All samples were<br />
classified as s<strong>and</strong>y aquifer, with brownish s<strong>and</strong>. The<br />
samples near well 04 were more orange coloured, than<br />
the samples from the reference well. Sieve analyses<br />
showed a very similar size distribution in all six<br />
samples with 95% of the total weight between mesh<br />
sizes 0.125 <strong>and</strong> 0.250mm. Additional analyses with<br />
the Hydrometer Method showed that the clay <strong>and</strong> silt<br />
500<br />
400<br />
S =<br />
φ<br />
6<br />
s s ρ<br />
i<br />
j<br />
25<br />
20<br />
6<br />
mmol Fe.(kg.ds) -1<br />
300<br />
200<br />
Well 04<br />
Reference<br />
15<br />
10<br />
mmol Fe.m -2<br />
100<br />
5<br />
0<br />
0<br />
0.063-0.106 0.106-0.125 0.125-0.150 0.150-0.250 0.250-0.50 0.50-1.0 1.0-2.0<br />
Grain size (mm)<br />
Reference Well 04<br />
Figure 6.4 Relation between grain size <strong>and</strong> iron per geometric surface area (lines) <strong>and</strong> per weight dry solids<br />
(bars) for samples taken at 19-22m in reference well <strong>and</strong> near well 04<br />
88
6 Characterization of accumulated deposits<br />
content of the samples was no more than 3%w/w <strong>and</strong><br />
the clay content was below 0.5-0.6 %w/w.<br />
Figure 6.4 depicts the average iron concentrations<br />
for 19-22m depth, distributed over the different<br />
s<strong>and</strong> grain sizes. The smallest fractions (
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
[C04]-[CREF]<br />
5<br />
4<br />
3<br />
2<br />
1<br />
mmol/kg.ds<br />
*10e-1 mmol/kg.ds<br />
mmol Ca.(kg.ds) -1<br />
0 100 200 300 400<br />
12<br />
14<br />
16<br />
6<br />
0<br />
Na Mn Sc Ti V As Sr<br />
Figure 6.5 Accumulated trace compounds near well 04 at depths<br />
between 19 <strong>and</strong> 22m<br />
was not sufficiently regular or accurate to monitor a<br />
reduction in the abstracted water due to subsurface<br />
iron removal. For arsenic, however, the concentrations<br />
were very low (
6 Characterization of accumulated deposits<br />
the majority of the Cation Exchange Capacity (CEC)<br />
in the groundwater, with 4.2 meq.L -1 of the total 7.2<br />
meq.L -1 (Ca+Mg+Na+Fe+Mn). The organic carbon<br />
<strong>and</strong> clay content can be used to estimate the CEC of the<br />
soil with the following empirical formula (Appelo <strong>and</strong><br />
Postma, 2005):<br />
Equation 6.1<br />
CEC( meq / kg)<br />
= 7 ⋅ (% clay)<br />
+ 35⋅<br />
(% organicC)<br />
The clay content has been determined with the<br />
Hydrometer Method <strong>and</strong> the organic carbon can be<br />
calculated from the organic matter concentration.<br />
When calculating the organic carbon from the organic<br />
matter content, the bound water needs to be taken into<br />
account when significant proportions of clay minerals<br />
are present. The clay content in the studied sediments<br />
was Fe > Na > Si > Al > Ca,<br />
with an average iron content of 7.6 mol% <strong>and</strong> an O:Fe<br />
mol% ratio of 5.3 (±2.25). The O:Fe mol% ratio in the<br />
reference coatings at 19-22m depth were much lower<br />
with an average of 1.3 (±0.57), the iron content was<br />
slightly higher at 9.9 mol%. The reduction of carbon<br />
near well 04 <strong>and</strong> the increase in oxygen indicates a<br />
change in coating composition to iron phases high<br />
in oxygen, like iron (oxy)hydroxides (Fe(OH) 3<br />
or<br />
α-FeOOH).<br />
Although based on the acid digestion results the<br />
iron content increased significantly near well 04, the<br />
ESEM-EDX results do not reflect this. This indicates<br />
that the coatings themselves have not gained more<br />
percent iron, but the overall amount of coating(s) must<br />
have increased. As the results from chemical extraction<br />
had already indicated, both sodium <strong>and</strong> calcium were<br />
found to be present in the iron coatings. The Fe:Ca<br />
91<br />
6
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
(a)<br />
(b)<br />
6<br />
(c)<br />
(d)<br />
1 Figure 6.7 ESEM pictures of iron coatings (a) iron coated grain at 19-20m; (b) iron precipitate on s<strong>and</strong> grain at 21-22m; (c) calcium<br />
sulphate coating at 23-34m; <strong>and</strong> (d) pyrite at 23-24m<br />
92
6 Characterization of accumulated deposits<br />
%mol ratio was on average 4.8 (±1.65) <strong>and</strong> 5.4 (±1.26)<br />
near well 04 <strong>and</strong> the reference well, respectively. The<br />
precipitates in Figure 6.7(c) <strong>and</strong> (d) have been found<br />
in the deeper layer at 23-24m. ESEM-EDX showed<br />
that the Ca:S ratio in coating (c) is 0.85, which roughly<br />
corresponds to CaSO 4<br />
(gypsum). Calcium sulfate has<br />
not been observed in any of the samples taken between<br />
19 <strong>and</strong> 22m depth, neither near well 04 nor in the<br />
reference well. The same was the case for the occurrence<br />
of the framboidal pyrite (FeS 2<br />
, (d)). In the presence of<br />
oxygen, pyrite is known to oxidize <strong>and</strong> release sulphate<br />
<strong>and</strong> iron into the groundwater. In addition, pyrite<br />
oxidation is an acidifying reaction <strong>and</strong> therefore not<br />
promoting the (rapid) oxidation of Fe 2+ . The presence<br />
of sulphide minerals is a vital difference between the<br />
depths with iron accumulation (19-22m) <strong>and</strong> without<br />
iron accumulation (deeper than 22m).<br />
X-Ray Powder Diffraction <strong>and</strong> 57 Fe<br />
Mössbauer Spectroscopy<br />
The same ten soil samples (
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
6<br />
hyperfine values of haematite. The Fe 2+ contribution,<br />
doublet with a Quadrupole Splitting (Q.S.) of ~ 2.39<br />
mm.s -1 , can be assigned to iron in phyllosilicates/ clay<br />
minerals like kaolinite <strong>and</strong> illite (Murad <strong>and</strong> Cashion,<br />
2004) or smectites (Heller-Kallai <strong>and</strong> Rozenson, 1981).<br />
Since the Fe 3+ doublets of most iron hydroxides can<br />
overlap irresolvable, it is recommended to measure<br />
Mössbauer spectra also at 77 <strong>and</strong>/or 4.2 K to allow a<br />
definitive identification of the iron structures <strong>and</strong> the<br />
quantitative determination of their proportions.<br />
The XRPD peaks for calcite allowed for the<br />
calculation of molar weight percentages by quantitative<br />
analyses with the Reference Intensity Ratio method.<br />
Calcite concentrations in the pre-treated samples<br />
(ground coating at
6 Characterization of accumulated deposits<br />
specific type of well clogging, intermittent abstraction<br />
to prevent well clogging has also been proposed by van<br />
Beek et al. (2010). In case of subsurface iron removal<br />
wells, the well is stopped <strong>and</strong> the flow direction is<br />
reversed, resulting in an even more effective reduction<br />
of the particle accumulation on the well screen. The<br />
contribution of particles in clogging is apparently more<br />
important than the possible effect of subsurface iron<br />
removal, indicating that clogging of the aquifer <strong>and</strong>/<br />
or well is no threat to the sustainable operation of this<br />
technology.<br />
Clogging of the well <strong>and</strong> aquifer is thus not<br />
noticeably occurring, even after removal of more than<br />
100,000 moles of iron. If this amount of iron would<br />
have been removed by conventional rapid s<strong>and</strong> filters,<br />
it is most likely that clogging by the voluminous iron<br />
sludge would have occurred. Water treatment sludge<br />
retains excessive amounts of water: e.g., 85% (w/w)<br />
after 5 months of sedimentation (Georgaki et al., 2004).<br />
Assuming formation of ferrihydrite (approximate<br />
composition, 5Fe 2<br />
O 3<br />
•9H 2<br />
O), the total amount of iron<br />
sludge would add up to 1.28•10 6 kg in 12 years of<br />
subsurface iron removal. Pure ferrihydrite has a density<br />
of 4,000 kg.m -3 (Cornell <strong>and</strong> Schwertmann, 1996), <strong>and</strong><br />
thus a very conservative estimation of the sludge density<br />
is 1,450 kg.m -3 . After 12 years of subsurface iron removal<br />
800<br />
600<br />
rehabilitation<br />
rehabilitation<br />
6<br />
Drawdown (cm)<br />
400<br />
normal production<br />
wells<br />
subsurface iron<br />
removal well<br />
200<br />
0<br />
2.7.06 10.10.06 18.1.07 28.4.07 6.8.07 14.11.07 22.2.08 1.6.08 9.9.08<br />
Figure 6.9 Drawdown development in normal production wells <strong>and</strong> a subsurface iron removal well at WTP<br />
De Put (year 2006-2008)<br />
95
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
6<br />
a sludge volume of >880 m 3 would have been produced.<br />
This is more than half of the injection (pore)water<br />
volume, <strong>and</strong>, even with a large (pore) storage capacity<br />
it is unthinkable that such a sludge amount remains<br />
unnoticed by the operator, i.e., clogging of the aquifer<br />
should have been observed. The finding that 56-100%<br />
of the iron has accumulated as compact, crystalline<br />
iron (oxy)hydroxides indicates that no voluminous<br />
water-containing iron sludge is produced subsurface.<br />
The presence of amorphous iron hydroxides, however,<br />
indicates that iron initially precipitates as ferrihydrite<br />
<strong>and</strong> subsequently transforms to iron crystal phases.<br />
This process in itself reduces the volume of the iron<br />
precipitates significantly, but could occur together with<br />
one or more of the previously proposed processes: iron<br />
precipitation in dead-end pores or deeper infiltration<br />
into the aquifer with time.<br />
More interesting is the existence of actively<br />
contributing soil layers where iron has accumulated<br />
versus soil layers where subsurface iron removal seems<br />
not to have occurred. Whether it was due to preferred<br />
flow paths or preferred geochemical mineralogy<br />
conditions (e.g., calcite); subsurface iron removal clearly<br />
favoured certain soil layers. The occurrence of such soil<br />
layers <strong>and</strong> the overall contribution of these layers in<br />
the production capacity of the well is site-specific <strong>and</strong><br />
greatly influences the well’s performance. It is in this<br />
observation that the exact physical distribution of the<br />
iron precipitates, e.g., in dead-end pores or distance<br />
from the well, is not relevant for the clogging of the<br />
aquifer. The assumed dominant process of adsorptivecatalytic<br />
oxidation during subsurface iron removal,<br />
with or without additional cation exchange (Appelo<br />
et al., 1999) or interfactial electron transfer (Mettler,<br />
2002), cannot be further supported based the presented<br />
data. Nevertheless, there seems to be a close correlation<br />
between the occurrence of iron <strong>and</strong> calcium (as calcite)<br />
in the analysed samples. This finding points out the<br />
proposed catalyzing role of calcite in the oxidation of<br />
Fe 2+ during subsurface iron removal (Mettler et al.,<br />
2009), which needs further investigation.<br />
The co-occurrence of iron in the subsurface<br />
in three phases, soluble Fe 2+ in the groundwater,<br />
poorly ordered amorphous iron hydroxides <strong>and</strong><br />
crystalline iron hydroxides, supports existing theory<br />
on recrystallisation of amorphous iron hydroxides. In<br />
the presence of soluble Fe 2+ , freshly formed ferrihydrite<br />
was found to transform to the more crystalline goethite<br />
after 5 days (Jeon et al., 2003). Pederson et al. (2005)<br />
observed complete ferrihydrite transformation into<br />
new <strong>and</strong> more stable phases such as lepidocrocite <strong>and</strong><br />
goethite within 5 days, <strong>and</strong> it was hypothesized that this<br />
was induced by the catalytic action of soluble Fe 2+ . In<br />
the subsurface iron removal system with soluble Fe 2+ ,<br />
amorphous iron hydroxides <strong>and</strong> stable (oxy)hydroxides<br />
it is likely that recrystallization of the freshly formed<br />
precipitation occurs, resulting in a more crystalline<br />
accumulation of iron near injection/abstraction wells.<br />
96
6 Characterization of accumulated deposits<br />
Conclusions<br />
After 12 years of operation, iron has accumulated<br />
at specific depths near the subsurface iron removal<br />
wells. Whether it was due to preferred flow paths<br />
or geochemical mineralogy conditions; subsurface<br />
iron removal clearly favoured certain soil layers. The<br />
majority of accumulated iron was characterized as<br />
crystalline, suggesting that precipitated amorphous<br />
iron hydroxides have transformed to iron hydroxides<br />
of higher crystallinity. These crystalline, compact<br />
iron hydroxides have not noticeably clogged the<br />
investigated well <strong>and</strong>/or aquifer between 1996 <strong>and</strong><br />
2008. The subsurface iron removal wells even need less<br />
frequent rehabilitation, as drawdown increases slower<br />
than in normal production wells. Other groundwater<br />
constituents, such as manganese, arsenic <strong>and</strong> strontium<br />
were found to co-accumulate with iron. Acid extraction<br />
<strong>and</strong> ESEM-EDX showed that calcium occurred together<br />
with iron <strong>and</strong> with X-Ray Powder Diffraction it was<br />
identified as calcite.<br />
References<br />
Appelo C. A. J., B. Drijver, R. Hekkenberg <strong>and</strong> M. de Jonge<br />
(1999) Modeling in situ iron removal from ground water,<br />
Ground Water 37(6): 811-817.<br />
Appelo C.A.J. <strong>and</strong> D. Postma (2005) Geochemistry, groundwater<br />
<strong>and</strong> pollution. Balkema, Rotterdam, 2nd edition.<br />
Braester C. <strong>and</strong> R. Martinell (1988) The Vyredox <strong>and</strong> Nitredox<br />
methods of in situ treatment of groundwater. Water Science<br />
<strong>and</strong> Technology 20(3): 149-163.<br />
Cornell R. M. <strong>and</strong> U. Schwertmann (1996) The iron oxides -<br />
structure, properties, reaction, occurrence <strong>and</strong> uses. VCH-<br />
Germany <strong>and</strong> USA.<br />
Craft C.B., S.W. Broome <strong>and</strong> E.D. Seneca (1988) Nitrogen,<br />
phosphorus <strong>and</strong> organic carbon pools in natural <strong>and</strong><br />
transplanted marsh soils. Estuaries 11(4): 272-280.<br />
de Vet W. W. J. M., C.C.A. van Genuchten, M.C.M. van<br />
Loosdrecht <strong>and</strong> J.C. van Dijk (2009) Water quality <strong>and</strong><br />
treatment of river bank filtrate. Drinking Water Engineering<br />
<strong>and</strong> Science Discussions 2: 127–159.<br />
Georgaki I., A.W.L. Dudeney <strong>and</strong> A.J. Monhemius (2004)<br />
Monhemius characterisation of iron-rich sludge: correlations<br />
between reactivity, density <strong>and</strong> structure. Minerals<br />
Engineering 17: 305–316.<br />
Grombach P. (1985) Groundwater treatment in situ in the<br />
aquifer. Water Supply 3: 13-18.<br />
Hallberg R. O. <strong>and</strong> R. Martinell (1976) Vyredox - in situ<br />
purification of groundwater. Ground Water 14(2): 88-93.<br />
Heron G., C. Crouzet, C.M. Bourg <strong>and</strong> T.H. Christensen<br />
(1994) Speciation of Fe(II) <strong>and</strong> Fe(III) in contaminated<br />
aquifer sediments using chemical extraction techniques.<br />
Environmental Science <strong>and</strong> Technology 28(9): 1698-1705.<br />
Jeon B. H., B.A. Dempsey, W.D. Burgos <strong>and</strong> R.A. Royer (2001)<br />
Reactions of ferrous iron with hematite. Colloids <strong>and</strong><br />
Surfaces A: Physicochemical <strong>and</strong> Engineering Aspects 191(1-<br />
2): 41-55.<br />
97<br />
6
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
6<br />
Jeon B. H., B.A. Dempsey <strong>and</strong> W.D. Burgos (2003) Kinetics <strong>and</strong><br />
mechanisms for reactions of Fe(II) with iron(III) oxides.<br />
Environmental Science <strong>and</strong> Technology 37(15): 3309-3315.<br />
Kroetsch D. <strong>and</strong> C. Wang (2006) Chapter 55: Particle size<br />
distribution. In:Soil sampling <strong>and</strong> methods of analysis.<br />
Canadian Soil Society, Taylor & Francis Group, LLC.<br />
Mettler S., M. Abdelmoula, E. Hoehn, R. Schoenenberger, P.G.<br />
Weidler <strong>and</strong> U. von Gunten (2001) Characterization of iron<br />
<strong>and</strong> manganese precipitates from an in situ groundwater<br />
treatment plant. Groundwater 39(6): 921-930.<br />
Mettler S. (2002) In-situ removal of iron from groundwater:<br />
Fe(II) oxygenation, <strong>and</strong> precipitation products in a<br />
calcareous aquifer. PhD dissertation, Swiss Federal Institute<br />
of Technology, Zurich.<br />
Mettler S., M. Wolthers, L. Charlet <strong>and</strong> U. von Gunten (2009)<br />
Sorption <strong>and</strong> catalytic oxidation of Fe(II) at the surface of<br />
calcite. Geochimica et Cosmochimica Acta 73: 1826-1840.<br />
Murad E. <strong>and</strong> J.H. Johnston (1987) <strong>Iron</strong> oxides <strong>and</strong> oxyhydroxides,<br />
in: Mössbauer Spectroscopy Applied to Inorganic Chemistry,<br />
G.J. Long, Ed. Plenum, New York, 507-582.<br />
Murad E. <strong>and</strong> U. Wagner (1994) The Mössbauer Spectrum of<br />
illites. Clay Minerals 29, 1-10.<br />
Murad E. <strong>and</strong> J.D. Cashion (2004) Mössbauer Spectroscopy of<br />
environmental materials <strong>and</strong> their industrial utilization,<br />
Kluwer, Boston.<br />
Olthoff R. (1986) <strong>Removal</strong> of iron <strong>and</strong> manganese in the aquifer.<br />
Inst. Siedlungswasserwirtschaft Abfalltechn. 63. University of<br />
Hannover, Germany.<br />
Pedersen H. D., D. Postma, R. Jakobsen <strong>and</strong> O. Larsen (2005)<br />
Fast transformation of iron oxyhydroxides by the catalytic<br />
action of aqueous Fe(II). Geochimica et Cosmochimica Acta<br />
69(16): 3967-3977.<br />
Rott U. (1985) Physical, chemical <strong>and</strong> biological aspects of the<br />
removal of iron <strong>and</strong> manganese underground. Water Supply<br />
3(2): 143-150.<br />
Rott U., C. Meyer <strong>and</strong> M. Friedle (2002) Residue-free removal<br />
of arsenic, iron, mangenese <strong>and</strong> ammonia from groundwater.<br />
Water Science <strong>and</strong> Technology: Water Supply 2(1): 17-24.<br />
Timmer H., J.D. Verdel <strong>and</strong> A.G. Jongmans (2003) Well clogging<br />
by particles in Dutch well fields. Journal AWWA 95(8): 112-<br />
118.<br />
van Beek C. G. E. M. (1985) Experiences with underground<br />
water treatment in the Netherl<strong>and</strong>s. Water Supply 3(2): 1-11.<br />
van Beek C. G. E. M., R. Breedveld, M. Tas, <strong>and</strong> R. Kollen<br />
(2010) Prevention of wellbore clogging by intermittent<br />
abstraction. Groundwater Monitoring & Remediation, DOI:<br />
10.1111/j.1745-6592.2010.01307.x.<br />
van Halem D., S. Olivero, W.W.J.M. de Vet, J.Q.J.C. Verbek., G.L.<br />
Amy <strong>and</strong> J.C. van Dijk (2010) <strong>Subsurface</strong> iron <strong>and</strong> arsenic<br />
removal for shallow tube well drinking water supply in rural<br />
Bangladesh. Water Research 44: 5761-5769.<br />
98
7<br />
Catalysis by accumulated deposits<br />
This chapter is based on:<br />
van Halem et al. (2011) Water Science <strong>and</strong> Technology: submitted
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
7<br />
Introduction<br />
<strong>Subsurface</strong> iron removal relies on the presence of<br />
(adsorbed) Fe 2+ available for oxidation during the<br />
injection phase (van Halem et al., 2011). The amount<br />
of adsorbed <strong>and</strong>/or exchangeable Fe 2+ depends,<br />
apart from the groundwater matrix, on the aquifer<br />
material composition. In general, the more adsorptive<br />
surface sites on the material, the more adsorbed Fe 2+<br />
is available for oxidation. Adsorption/exchange surface<br />
sites are mostly provided by organic material, clay<br />
<strong>and</strong> Fe 3+ oxides. Apart form adsorptive surface sites,<br />
the Fe 3+ mineral surfaces are also known to catalyze<br />
the heterogeneous oxidation of Fe 2+ (Tamura et al.,<br />
1976), which is relevant to subsurface iron removal.<br />
In addion to Fe 3+ oxides, also the presence of calcite<br />
has been proposed to enhance Fe 2+ immobilization<br />
(Mettler, 2002). Both mineral surfaces may be found in<br />
the aquifers <strong>and</strong> could locally enhance subsurface iron<br />
removal.<br />
Once subsurface iron removal has been in<br />
operation for multiple cycles, the Fe 3+ oxide mineral<br />
surface will increase. In the affected area around the<br />
tube well, the so-called oxidation zone, these Fe 3+<br />
oxides have been found to accumulate (Mettler et al.,<br />
2001; van Halem et al., 2011). Such accumulated iron<br />
deposits have not been reported to clog the aquifer, but<br />
they may have a positive effect on the immobilization<br />
of both Fe 2+ <strong>and</strong> As(III). An improved efficacy over<br />
successive cycles has already been observed at many<br />
full scale wells, mainly for the removal of iron.<br />
<strong>Subsurface</strong> arsenic removal has not shown the same<br />
improving trend after multiple cycles (van Halem et al.,<br />
2010), but experiments have never lasted longer than<br />
a year (or 20 cycles). Investigation of arsenic retention<br />
during injection-abstraction cycles by accumulated<br />
deposits could provide information on the long-term<br />
performance of subsurface arsenic removal.<br />
Accumulated Fe 3+ oxides act as a sorbent for Fe 2+<br />
during the abstraction phase, while during injection<br />
these mineral surfaces enhance heterogeneous<br />
oxidation. The amount of Fe 2+ adsorbed during<br />
abstraction, theoretically determines the amount of<br />
O 2<br />
consumption during injection. Thus, the more<br />
accumulated Fe 3+ , the more O 2<br />
consumption, Fe 2+<br />
oxidation <strong>and</strong> subsequent Fe 2+ removal. This is only<br />
the case if the amount of O 2<br />
is not limiting <strong>and</strong> if the<br />
adsorbed Fe 2+ is reached by the passing O 2<br />
. The latter<br />
could be limited by the occurrence of cation exchange<br />
during injection, as exchangeable Fe 2+ may be released<br />
from the mineral surface (e.g., Ca 2+ -Fe 2+ ; Appelo et al.,<br />
1999; van Halem et al., 2011). These complex mineralwater<br />
interaction reactions occur simulteously during<br />
surbsurface iron <strong>and</strong> arsenic removal. The objective<br />
of this study was to isolate these reactions in order<br />
to investigate the influence of the accumulated (Fe 3+ )<br />
deposits on the behaviour of Fe 2+ <strong>and</strong> As(III). Column<br />
experiments were executed with sediments obtained<br />
100
7 Catalysis by accumulated deposits<br />
from within <strong>and</strong> outside the oxidation zone nearby a<br />
12-year-old iron removal well of Oasen Drinking Water<br />
Company (van Halem et al., 2011). <strong>Subsurface</strong> injectionabstraction<br />
cycles were simulated in the columns with<br />
natural <strong>and</strong> synthetic groundwater to investigate the<br />
contribution of Fe 2+ oxidation, adsorption <strong>and</strong> exchange<br />
on these sediments. In addition, the experiments with<br />
these sediments are compared to clean filter s<strong>and</strong> for<br />
the co-removal of As(III) during injection-abstraction<br />
cycles.<br />
Materials <strong>and</strong> methods<br />
Composition of sediments<br />
Bore holes (SonicMast with SonicDrill 2x7) were drilled<br />
at WTP De Put at 5m from the injection/abstraction<br />
wells <strong>and</strong> 50m from the wells (van Halem et al., 2011).<br />
The borehole at 50m was considered to be outside<br />
the affected oxidation zone. 800mL sediment samples<br />
were taken every meter at the depths from 10 to 28<br />
meters below ground level. Classification of the drilled<br />
material showed that at approximately -12m the clay<br />
layer ended <strong>and</strong> coarse s<strong>and</strong> layer (brownish) started.<br />
Greyish coarse s<strong>and</strong> was found at a depth of 22-26 m<br />
<strong>and</strong> the lower clay layer started at 26-27 m, showing<br />
that the well was into a confined aquifer. The depth of<br />
the perforated well filters was from 15 to 26 m. The soil<br />
samples were taken in September 2008, subsurface iron<br />
removal started in 1996 <strong>and</strong> had been operated for 12<br />
years at the moment of sampling.<br />
Samples from every meter depth were analysed<br />
in a certified laboratory with ICP-MS scan after<br />
chemical extraction with: Acid-Oxalate for amorphous<br />
iron <strong>and</strong> manganese; or, Nitric Acid for total iron <strong>and</strong><br />
a wide range of compounds (including, Na, Mg, Al, Si,<br />
K, Ca, Mn, Sr, Ni, Sc, Ti, As). Acid-Oxalate extraction<br />
consisted of reductive dissolution of amorphous iron/<br />
manganese with an oxalate buffer solution which<br />
consisted of 0.2M ammonium oxalate <strong>and</strong> 0.2M oxalic<br />
7<br />
Table 7.1<br />
Chemical composition of the sediments (A), (B) <strong>and</strong> (C)<br />
Sediment depth distance from well Fe am-Fe Mn Ca Na<br />
m m mmol.kg -1<br />
A 21-22 5 419 65 3.1 265 1.8<br />
B 23-24 5 210 24 1.8 49 1.8<br />
C 21.22 50 241 19 1.9 178 1.6<br />
101
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
acid (at pH=3). During digestion the sample was<br />
agitated <strong>and</strong> light was excluded.<br />
For the column studies 3 different samples were<br />
selected, namely (A) one from within the oxidation zone<br />
at 21m depth; (B) one from within the oxidation zone<br />
at 23m depth; <strong>and</strong> (C) one from outside the oxidation<br />
zone at 21m depth. The chemical composition of these<br />
sediments is summarized in Table 7.1. Additionally,<br />
clean silica filter s<strong>and</strong> (washed for 24h with 1M HCl)<br />
was used in one of the columns as a reference.<br />
Column studies<br />
For the experiments two different column set-ups<br />
were used: (i) 30cm columns <strong>and</strong> (ii) 9cm columns<br />
(Figure 7.1). The small columns were used for the<br />
experiments with specific sediment samples, as the<br />
sample volume was too small to fill the larger columns.<br />
As a reference filter s<strong>and</strong> was also used in one of these<br />
columns. All columns had an inner diameter of 38mm<br />
<strong>and</strong> were wrapped in aluminium foil to exclude light.<br />
The experiments were executed at a drinking water<br />
treatment plant, WTP Loosdrecht of Vitens Drinking<br />
Water Supply, with natural groundwater. During the<br />
pH ORP EC DO<br />
Waste/Sample<br />
Point<br />
Discharge Pump<br />
7<br />
H 2<br />
O<br />
+<br />
O 2<br />
Groundwater<br />
Dosing Pump<br />
N 2<br />
Column FS<br />
Column A<br />
Column B<br />
Column C<br />
NaAsO 2<br />
Figure 7.1 Experimental column set-up with filter s<strong>and</strong> column (FS) <strong>and</strong> columns with sediments A, B <strong>and</strong> C<br />
102
7 Catalysis by accumulated deposits<br />
research period the groundwater had an average pH of<br />
7.2 (±0.01), a constant temperature of 11 °C, 0.1 mmol<br />
Fe.L -1 , 3.3 μmol Mn.L -1 , 1.07 mmol Ca.L -1 , 0.52 mmol<br />
Na.L -1 , 0.28 mmol Si.L -1 , 0.5 μmol PO 4<br />
3-<br />
.L -1 , <strong>and</strong> 0.14<br />
mmol SO 4<br />
.L -1 .<br />
At the start of each experiment the columns<br />
were conditioned with the groundwater, until complete<br />
breakthrough of iron occurred, <strong>and</strong> the redox potential<br />
stabilized. In order to simulate the injection-abstraction<br />
cycles of subsurface iron removal, the flow was reversed<br />
from up flow to down flow to start an injection phase.<br />
A normal injection mode consisted of injection with<br />
drinking water originating from the same water<br />
treatment plant. However, to measure the exchangeable<br />
Fe 2+ , injection cycles have also been performed in the<br />
absence of O 2<br />
<strong>and</strong>/or Na + . In addition to the experiments<br />
with natural groundwater, some experiments were<br />
also executed with synthetic groundwater, namely<br />
experiments to study the absence/presence of Fe 2+ <strong>and</strong><br />
experiments to study the behaviour of As(III). These<br />
experiments were conducted in 30cm columns with<br />
filter s<strong>and</strong> <strong>and</strong> a mixture of sediment samples. The<br />
synthetic groundwater was produced by distributing<br />
demineralized water through a 6m gas stripping column<br />
containing stainless steel Pall Rings. From the bottom<br />
pure N 2<br />
was blown into the degassing column to sparge<br />
out all O 2<br />
. Before entering the s<strong>and</strong> columns, the water<br />
was checked for O 2<br />
(Orbisphere; HACH Lange; M1100<br />
Sensor; 410 Analyser) to ensure concentrations below<br />
1.5µmol.L -1 . Addition of stock solutions for FeSO 4<br />
,<br />
NaHCO 3<br />
<strong>and</strong>/or NaCl was done with a dosing pump<br />
followed by a static mixer. pH correction was achieved<br />
by addition of HCl or NaOH <strong>and</strong> all stock solutions<br />
were sparged with N 2<br />
in order to ensure the absence<br />
of O 2<br />
. The experiments were performed with synthetic<br />
groundwater of pH 6.9 (±0.02), a temperature of 20°C<br />
(±0.1), Fe concentration of 0.1 mmol.L -1 (±0.01), pH<br />
buffer of 5mM NaHCO 3<br />
<strong>and</strong> ionic strength buffer of<br />
1.6mM NaCl.<br />
The push-pull operational mode of injectionabstraction<br />
was simulated in the plug-flow environment<br />
of the columns with down flow for injection <strong>and</strong> up<br />
flow for abstraction (1.2 L.h -1 ±0.05). An injectionabstraction<br />
cycle started with 14 (±0.5) pore volumes<br />
of injection water to achieve >80% breakthrough of O 2<br />
.<br />
Subsequently the influent was switched to groundwater<br />
for multiple pore volumes to allow breakthrough of Fe<br />
<strong>and</strong> As. Electrical conductivity was used as a conservative<br />
tracer from which the pore volume could be calculated<br />
to be 0.04-0.05L, depending on the sediment type.<br />
The weight of the sediment material per column was<br />
0.16kg <strong>and</strong> 0.17kg for filter s<strong>and</strong> <strong>and</strong> natural sediment,<br />
respectively. The flow rate in the columns (2.0-2.5 m.h -<br />
1<br />
) was controlled with a multi-channel pump <strong>and</strong> PVC<br />
tubing with low gas permeability. Anoxic conditions<br />
were maintained in the columns by using an airtight<br />
FESTO system (6 x 1 PUN, I.D. 4mm) with matching<br />
connectors <strong>and</strong> valves.<br />
Fe <strong>and</strong> As analyses of the water samples were<br />
done with an Atomic Absorption Spectrometer<br />
103<br />
7
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
(Perkin-Elmer Flame AAS 3110; Perkin-Elmer GF-<br />
AAS 5100PC). In-line measurements were done for<br />
dissolved oxygen (Orbisphere <strong>and</strong> WTW Cellox 325),<br />
ORP(WTW SenTix ORP), pH (WTW SenTix 41),<br />
<strong>and</strong> electrical conductivity (WTW TetraCon 325).<br />
Measurements were registered on a computer with<br />
Multilab Pilot v5.06 software.<br />
(C/C0)Fe<br />
1<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0<br />
filter s<strong>and</strong><br />
sediment A<br />
sediment B<br />
sediment C<br />
0 10 20 30 40<br />
7<br />
104<br />
Results<br />
Regular injection-abstraction cycle<br />
Figure 7.2 shows the breakthrough curves for Fe in the<br />
groundwater after injection with tap water (0.28mM<br />
O 2<br />
) in the columns with the different sediments. The<br />
Fe breakthrough is the fastest for the clean filter s<strong>and</strong><br />
column, with a retardation to C/C 0<br />
=0.5 of 6.5 pore<br />
volumes (R Fe<br />
=6.5). The Fe retardation factor for the<br />
reference sample from outside the oxidation zone,<br />
sediment (C), is slightly higher at 9. The presence of<br />
iron <strong>and</strong> manganese oxides, natural organic matter <strong>and</strong><br />
clay content will enhance the Fe removal compared to<br />
clean filter s<strong>and</strong>. The columns filled with s<strong>and</strong> from<br />
inside the oxidation zone (A, B) showed better Fe<br />
removal efficacies with retardation factors 20 <strong>and</strong> 14,<br />
respectively. The column with the highest Fe content in<br />
the sediment (A) was most effective in the retention of<br />
Fe 2+ during a regular injection-abstraction cycle. The<br />
pore volumes<br />
Figure 7.2 Fe behaviour during injection-abstraction cycle in<br />
columns containing filter s<strong>and</strong>, sediments A, B or C<br />
R Fe<br />
of this column was 3 times higher than of the clean<br />
filter s<strong>and</strong> column, illustrating that the presence of<br />
(accumulated) iron hydroxides enhanced the removal<br />
of Fe. Although the Fe content in the sediment can<br />
account for the enhanced performance of column A,<br />
it does not differentiate between columns B <strong>and</strong> C;<br />
both columns have more or less the same Fe content.<br />
Between these two columns a major difference can be<br />
found in the amount <strong>and</strong> type of calcium mineral. The<br />
calcium in the sediment of column C has been identified<br />
as calcite (XRPD; van Halem et al., 2011), whereas the<br />
calcium in column B was linked to the presence of<br />
calcium sulphates (SEM-EDX; van Halem et al., 2011).<br />
Column C did not perform much better than the filter<br />
s<strong>and</strong> column, indicating that the presence of high<br />
calcite concentrations did not enhance the subsurface<br />
iron removal process.
7 Catalysis by accumulated deposits<br />
O 2<br />
consumption<br />
The laminar plug flow in the columns permits only very<br />
little mixing of the injection water with the groundwater,<br />
meaning that heterogeneous oxidation will prevail, i.e.,<br />
adsorptive-catalytic oxidation reaction. The columns<br />
do not simulate processes that occur on the boundary<br />
of the oxidation zone, but merely a fragment of soil that<br />
is flushed by injection water within the zone. The O 2<br />
in the injection water, or tap water, passes the small<br />
columns after a few pore volumes to allow complete<br />
oxidation of all Fe 2+ on the soil material. The retardation<br />
of O 2<br />
during injection shows the O 2<br />
consumption for<br />
the Fe 2+ oxidation reaction:<br />
Equation 7.1<br />
S − OFe( II) + 0.25O + 1.5 H O →S − OFe( III)( OH ) + H<br />
[O2] mmol.L -1<br />
0.4<br />
0.3<br />
0.2<br />
0.1<br />
0<br />
+ 0 +<br />
2 2 2<br />
0 2 4 6 8 10<br />
pore volumes<br />
filter s<strong>and</strong><br />
sediment A<br />
Figure 7.3 O 2<br />
concentrations during an injection phase at WTP<br />
Loosdrecht for the filter s<strong>and</strong> <strong>and</strong> sediment A<br />
Figure 7.3 shows the O 2<br />
consumption for the column<br />
with filter s<strong>and</strong> <strong>and</strong> the column containing sediment<br />
A. Clearly, the sediment from within the oxidation<br />
zone consumes more O 2<br />
, illustrating the abundance<br />
of adsorbed Fe 2+ on the sediment. The more adsorbed<br />
Fe 2+ is oxidized during injection, the more adsorption<br />
sites will be available during the abstraction phase. The<br />
enhanced performance of sediment A has already been<br />
demonstrated in Figure 7.2.<br />
The amount of adsorbed Fe 2+ that is available<br />
during injection depends on both the number of<br />
available adsorption sites (= type of sediment; Figure<br />
7.2) <strong>and</strong> the saturation of this surface with Fe 2+ from<br />
solution. The saturation of the soil material in the<br />
columns with Fe 2+ has been achieved by loading the<br />
columns with groundwater from WTP Loosdrecht<br />
until complete breakthrough of Fe, <strong>and</strong> stabilization<br />
of pH <strong>and</strong> ORP. Only once this steady state situation<br />
had been reached, the columns were injected with<br />
tap water. In order to investigate the difference in O 2<br />
consumption between columns saturated with Fe 2+ <strong>and</strong><br />
columns unsaturated with Fe 2+ , the filter s<strong>and</strong> columns<br />
were loaded with synthetic groundwater either with or<br />
without 0.1mM Fe 2+ . Subsequently in both columns a<br />
regular injection-abstraction cycle was performed with<br />
0.28mM O 2<br />
. Figure 7.4 depicts the O 2<br />
concentration<br />
during injection <strong>and</strong> abstraction in both experiments.<br />
O 2<br />
breakthrough was significantly more delayed during<br />
injection in the columns saturated with Fe 2+ than the<br />
columns without Fe 2+ , i.e., more O 2<br />
was consumed for<br />
105<br />
7
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
7<br />
A<br />
[O2] mmol.L -1<br />
0.4<br />
0.3<br />
0.2<br />
0.8<br />
0.1<br />
no Fe2+<br />
0.6<br />
Fe2+ saturated<br />
0.4<br />
0<br />
R Fe =7<br />
0.2<br />
0 2 4 6 8 10<br />
0<br />
in columns saturated or unsaturated with Fe 2+<br />
B<br />
0.4<br />
pore volumes<br />
-10 -5 0 5 10 15 20 25<br />
no Fe2+<br />
pore volumes<br />
0.3<br />
Fe2+ saturated<br />
Figure 7.5 Fe behaviour during injection of 0.1 M NaCl <strong>and</strong><br />
subsequent abstraction of natural groundwater<br />
0.2<br />
(sediment C)<br />
0.1<br />
0<br />
0 2 4 6 8 10<br />
pore volumes<br />
Figure 7.4 O 2<br />
behaviour during (A) injection <strong>and</strong> (B) abstraction<br />
[O2] mmol.L -1<br />
adsorbed Fe 2+ oxidation. During abstraction (Figure<br />
7.4B) more O 2<br />
was observed to leave the column with<br />
no Fe 2+ , which corresponds to the finding that, in the<br />
absence of adsorbed Fe 2+ , less O 2<br />
will be consumed. It<br />
is noteworthy that O 2<br />
can be found in the abstracted<br />
water for nearly 10 pore volumes.<br />
injection phase<br />
C/C0<br />
1.6<br />
1.4<br />
1.2<br />
1<br />
abstraction phase<br />
Exchangeable Fe 2+<br />
The Fe 2+ exchange on the material in the columns can<br />
be calculated from the Fe retention measurements after<br />
injection with a solution of 0.1M NaCl, in the absence of<br />
O 2<br />
. During such an injection phase, abundant Na + will<br />
be available to exchange with the exchangeable Fe 2+ on<br />
the s<strong>and</strong> material. The amount of exchangeable Fe 2+ on<br />
the s<strong>and</strong> material can be estimated either from the Fe 2+<br />
leaching from the columns during injection, or from<br />
the Fe 2+ retention during abstraction. It is, however,<br />
practically almost impossible to measure the relatively<br />
short Fe 2+ peak passing after injection of 1 pore volume.<br />
The sample size (30mL) dilutes the initial peak <strong>and</strong> the<br />
Fe concentration in the first sample will therefore be<br />
lower than the actual Fe peak. During abstraction the<br />
breakthrough process is much slower, making it feasible<br />
106
7 Catalysis by accumulated deposits<br />
Table 7.2<br />
Retardation factor, Fe retention <strong>and</strong> exchangeable Fe 2+ for filter s<strong>and</strong> <strong>and</strong> sediments A, B <strong>and</strong> C<br />
Column Retardation factor Fe retained Exchangeable Fe 2+<br />
0.1M NaCl tap water mM Fe mM Fe.kg -1 meq.kg -1<br />
filter s<strong>and</strong> 7 6 0.04 0.20 0.40<br />
A 14 14 0.07 0.41 0.83<br />
B 10 11 0.05 0.30 0.60<br />
C 7 7 0.04 0.22 0.44<br />
to accurately measure the total amount of Fe retained in<br />
the columns. Figure 7.5 depicts the Fe measurements<br />
during injection (negative pore volumes) <strong>and</strong> during<br />
abstraction (positive pore volumes) for column C. The<br />
retardation factor for this cycle was 7 <strong>and</strong> the total<br />
amount of retained Fe was calculated to be 0.04mM,<br />
corresponding to 0.22mM.kg -1 of sediment. The sites<br />
available for exchangeable Fe 2+ on this particular<br />
sediment with this particular groundwater type can be<br />
calculated to be 0.44 meq.kg -1 .<br />
It should be noted that the experiments<br />
were executed with natural groundwater, thus the<br />
exchangeable Fe 2+ on the sediments are limited by the<br />
presence of other cations, such as Ca 2+ . An overview of<br />
the retained Fe in the other columns is given in Table<br />
7.2, both for injection of O 2<br />
-free tap water <strong>and</strong> 0.1M<br />
NaCl demineralized water. The table shows that Fe 2+<br />
exchange was higher in sediment (A) from within<br />
the oxidation zone, reaching up to 0.83 meq.kg -1 . The<br />
presence of high iron oxides is known to enhance Fe 2+<br />
exchange (Appelo <strong>and</strong> Postma, 2005). The presence of<br />
calcite did not have the same effect on the exchangeable<br />
Fe 2+ fraction, as both clean s<strong>and</strong> <strong>and</strong> calcite-containing<br />
sediment (C) have the same Fe 2+ retention in the<br />
columns.<br />
Fe 2+ oxidation<br />
In order to differentiate between a catalytic oxidation<br />
reaction <strong>and</strong> the occurrence of cation exchange, the<br />
columns have been loaded with O 2<br />
-containing water,<br />
either with or without 0.1M NaCl. The injectionabstraction<br />
cycle without 0.1M NaCl simulates the<br />
oxidation-adsorption process during subsurface iron<br />
removal without the interference of cation exchange.<br />
The results in Figure 7.6 show Fe retardation factors of<br />
6, 13, 16 <strong>and</strong> 17 for the filter s<strong>and</strong>, <strong>and</strong> sediments C,<br />
B, A, respectively. R Fe<br />
for the filter s<strong>and</strong>, <strong>and</strong> sediments<br />
C, B, A, were measured to be 10, 13, 22 <strong>and</strong> 31 after<br />
injection with a solution containing 0.1M NaCl <strong>and</strong><br />
0.28mM O 2<br />
. These retardation factors are higher<br />
than the regular injection-abstraction cycle with O 2<br />
-<br />
containing tap water for all sediments, except sediment<br />
107<br />
7
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
7<br />
108<br />
RFe<br />
35<br />
30<br />
25<br />
20<br />
15<br />
10<br />
5<br />
0<br />
filter s<strong>and</strong><br />
0.28mM O2 - no NaCl<br />
0.28mM O2 - 0.1M NaCl<br />
Sediment A Sediment B Sediment C<br />
Figure 7.6 Fe retardation factors after injection with<br />
demineralised water <strong>and</strong> 0.1M NaCl solution, both<br />
containing 0.28mM O 2<br />
C. This sediment originates from outside the oxidation<br />
zone <strong>and</strong> shows less affinity for cation exchange during<br />
subsurface iron removal than the other sediments.<br />
However, the exchangeable Fe 2+ of sediment C was the<br />
same as for filter s<strong>and</strong>, whereas the column with filter<br />
s<strong>and</strong> performed better at higher Na concentrations. An<br />
explanation for this deviation may be the fact that the<br />
clean filter s<strong>and</strong> did not contain any catalyzing material,<br />
such as calcite. Sediment C contained significant calcite<br />
content which may have caused rapid heterogeneous<br />
Fe 2+ oxidation/incorporation, <strong>and</strong> therefore limiting<br />
the potential role of cation exchange. Sediments A <strong>and</strong><br />
B have a higher exchangeable Fe 2+ fraction, resulting<br />
in better performance after injection of high Na<br />
concentrations.<br />
(C/C0)As<br />
1.0<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0.0<br />
0 10 20 30 40 50 60 70 80<br />
pore volumes<br />
filter s<strong>and</strong> 1<br />
filter s<strong>and</strong> 2<br />
sediments 1<br />
sediments 2<br />
Figure 7.7 As(III) behaviour during an injection-abstraction<br />
cycle in duplicate columns with filter s<strong>and</strong> <strong>and</strong> a<br />
sediment mixture from within the oxidation zone<br />
Co-removal of As(III)<br />
A mixture of the sediments A <strong>and</strong> B was used to<br />
study the behaviour of As during regular injectionabstraction<br />
cycles. Before injection with 0.28mM O 2<br />
solution, the columns were saturated with synthetic<br />
groundwater, containing 0.1mM Fe 2+ <strong>and</strong> 2µM As(III).<br />
Figure 7.7 depicts the results for arsenic measurements<br />
during the abstraction phase. The behaviour of arsenic<br />
is very similar for the columns containing clean filter<br />
s<strong>and</strong> or a mixture of sediments A <strong>and</strong> B. Clearly, the<br />
breakthrough of arsenic is not delayed by sediments<br />
from within the oxidation zone compared to the clean<br />
filter s<strong>and</strong>. <strong>Arsenic</strong> breakthrough starts immediately<br />
upon abstraction <strong>and</strong> C/C 0<br />
=0.5 is reached at 5 to<br />
10 pore volumes. Based on these results subsurface<br />
arsenic removal does not seem to be enhanced by iron<br />
deposits nearby subsurface removal wells. Also, the
7 Catalysis by accumulated deposits<br />
enhanced O 2<br />
consumption (Figure 7.3) in the columns<br />
does not translate in better arsenic retention. The<br />
adsorptive capacity of the sediments was insufficient<br />
to significantly enhance the As removal, which shows<br />
that while subsurface iron removal may benefit from<br />
accumulated deposits (Figure 7.5), the same does not<br />
apply to As(III).<br />
Conclusions<br />
Sediments obtained inside the oxidation zone of a<br />
12-year-old subsurface iron removal well showed<br />
better Fe 2+ retention during injection-abstraction<br />
cycles in the column environment than clean filter<br />
s<strong>and</strong> or sediments from outside the oxidation zone.<br />
This finding was supported by the fact that more<br />
O 2<br />
was consumed during injection in columns with<br />
sediments from within the oxidation zone. In addition,<br />
the exchangeable Fe 2+ fraction for sediments was twice<br />
as high as for sediments that had been unaffected by<br />
subsurface iron removal, resulting in a doubling of the<br />
Fe retardation factor. The same cannot be concluded<br />
for As(III) retardation, as sediments from within the<br />
oxidation zone did not show better removal than clean<br />
filter s<strong>and</strong>.<br />
References<br />
Appelo C. A. J., B. Drijver, R. Hekkenberg <strong>and</strong> M. de Jonge<br />
(1999) Modeling in situ iron removal from ground water,<br />
Ground Water 37(6): 811-817.<br />
Appelo C.A.J. <strong>and</strong> D. Postma (2005) Geochemistry, groundwater<br />
<strong>and</strong> pollution. Balkema, Rotterdam, 2nd edition.<br />
Braester C. <strong>and</strong> R. Martinell (1988) The Vyredox <strong>and</strong> Nitredox<br />
methods of in situ treatment of groundwater, Water Science<br />
<strong>and</strong> Technology 20(3): 149-163.<br />
Grombach P. (1985) Groundwater treatment in situ in the<br />
aquifer, Water Supply, 3(1): 13-18.<br />
Hallberg R. O. <strong>and</strong> R. Martinell (1976) Vyredox - in situ<br />
purification of groundwater, Ground Water, 14(2): 88-93.<br />
Mettler S., M. Abdelmoula, E. Hoehn, R. Schoenenberger, P.<br />
Weidler <strong>and</strong> U. von Gunten (2001) Characterization of iron<br />
<strong>and</strong> manganese precipitates from an in situ groundwater<br />
treatment plant, Ground Water 6: 921-930.<br />
Mettler S. (2002) In-situ removal of iron from groundwater:<br />
Fe(II) oxygenation, <strong>and</strong> precipitation products in a<br />
calcareous aquifer, PhD dissertation, Swiss Federal Institute<br />
of Technology, Zurich.<br />
Mettler S., M. Wolthers, L. Charlet <strong>and</strong> U. von Gunten (2009)<br />
Sorption <strong>and</strong> catalytic oxidation of Fe(II) at the surface of<br />
calcite. Geochimica et Cosmochimica Acta 73: 1826-1840.<br />
Rott U. (1985) Physical, chemical <strong>and</strong> biological aspects of the<br />
removal of iron <strong>and</strong> manganese underground. Water Supply<br />
3(2): 143-150.<br />
Rott U., C. Meyer <strong>and</strong> M. Friedle (2002) Residue-free removal<br />
of arsenic, iron, mangenese <strong>and</strong> ammonia from groundwater.<br />
109<br />
7
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
Water Science <strong>and</strong> Technology: Water Supply 2(1): 17-24.<br />
van Beek C.G.E.M. (1985) Experiences with underground water<br />
treatment in the Netherl<strong>and</strong>s, Water Supply 3: 1-11.<br />
van Halem D., S. Olivero, W.W.J.M. de Vet, J.Q.J.C. Verbek, G.L.<br />
Amy <strong>and</strong> J.C. van Dijk (2010) <strong>Subsurface</strong> iron <strong>and</strong> arsenic<br />
removal for shallow tube well drinking water supply in rural<br />
Bangladesh. Water Research (44): 5761-5769.<br />
van Halem D., W. de Vet, J. Verberk, G. Amy <strong>and</strong> H. van Dijk<br />
(2011) Characterization of accumulated precipitates during<br />
subsurface iron removal. Applied Geochemistry 26: 116–124.<br />
van Halem, D., D.H. Moed, J. Q.J.C. Verberk, G.L. Amy <strong>and</strong><br />
J.C. van Dijk (2011) Cation exchange during subsurface iron<br />
removal. Water Research: under review.<br />
7<br />
110
8<br />
Influence of groundwater<br />
composition<br />
This chapter is based on:<br />
van Halem et al. (2011) Science of the Total Environment: submitted
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
8<br />
112<br />
Introduction<br />
In Bangladesh, elevated iron <strong>and</strong> arsenic concentrations<br />
have been found to often co-occur in anoxic <strong>and</strong><br />
anaerobic groundwater (Nickson et al., 2000). The<br />
presence of both iron <strong>and</strong> arsenic are a prerequisite for<br />
the operation of <strong>Subsurface</strong> <strong>Iron</strong> <strong>and</strong> <strong>Arsenic</strong> <strong>Removal</strong><br />
(SIR/SAR; van Halem et al., 2010)). The injection of<br />
aerated water promotes the oxidation of Fe 2+ , resulting<br />
in the formation of iron oxides nearby subsurface<br />
treatment wells. More water with reduced iron <strong>and</strong><br />
arsenic concentrations can be abstracted (volume V)<br />
than was injected (volume Vi), i.e., this volumetric<br />
ratio (V/Vi) determines the efficiency of the system.<br />
The deposited iron oxides provide adsorptive surface<br />
area for both Fe 2+ <strong>and</strong> arsenic during abstraction (Table<br />
8.1). The adsorptive capacity of iron oxides has been<br />
found to be limited by the presence of other inorganic<br />
ions, such as phosphate, silicate <strong>and</strong> bicarbonate<br />
(goethite; Stachowicz, 2008). Also Fe 2+ adsorption<br />
<strong>and</strong> precipitation is affected by the presence of other<br />
inorganic ions, including calcium (Sharma, 2001;<br />
Ciardelli et al., 2006). The majority of these compounds<br />
are commonly found in Bangladeshi ground- <strong>and</strong><br />
drinking water (BGS/DPHE, 2001; Ciardelli et al., 2006).<br />
It is hypothesized that the groundwater composition<br />
will greatly influence the site-specific effectiveness of<br />
SIR/SAR under diverse geochemical conditions.<br />
There have been several pilot studies to<br />
investigate the potential of SAR, with various results. In<br />
Bangladesh three small-scale injection facilities (0.5m 3 )<br />
were constructed with a plate aerator, allowing injection<br />
of oxygen-containing water with 0.16 mmol.L -1 (Sarkar<br />
<strong>and</strong> Rahman, 2001). <strong>Arsenic</strong> concentrations were<br />
found to be reduced from 1.5 µmol.L -1 to below 0.67<br />
µmol.L -1 (national guideline; 50 µg.L -1 ) until V/Vi<br />
reached 4. At higher arsenic concentrations, namely 6.7<br />
µmol.L -1 <strong>and</strong> 17.0 µmol.L -1 , the reduction of arsenic did<br />
not reach below 2.7 µg.L -1 <strong>and</strong> 6.7 µg.L -1 , respectively.<br />
Across the Indian border in the same Bengal Delta,<br />
Sen Gupta <strong>and</strong> co-workers (2009) reported that at<br />
6 different community SAR plants a V/Vi ratio of 4<br />
to 6 could be operated providing arsenic-free water<br />
to the people. Background arsenic levels were not<br />
reported in this publication, but the surrounding wells<br />
Table 8.1<br />
As(III), arsenite<br />
As(V), arsenate<br />
Fe 2+ , ferrous iron<br />
Adsorption reactions <strong>and</strong> constants for As(V), As(III)<br />
<strong>and</strong> Fe 2+ onto the iron oxide goethite (Dixit <strong>and</strong><br />
Hering, 2003; Dixit <strong>and</strong> Hering, 2006)<br />
Reactions<br />
S − OH + AsO + 3H →S − H AsO + H O<br />
S − OH + AsO + H →S − HAsO + H O<br />
S − OH + AsO<br />
S − OH + AsO<br />
3− +<br />
3 2 3 2<br />
3− + −<br />
3<br />
2<br />
3 2<br />
3−<br />
4<br />
2<br />
3−<br />
4<br />
− + → − +<br />
+<br />
+ H → S − HAsO4<br />
+ H<br />
2O<br />
+<br />
2−<br />
+ H → S − AsO + H O<br />
2+ + +<br />
S OH Fe S OFe( II ) H<br />
2+ +<br />
S − OH + Fe + H<br />
2O →S − OFe( II ) OH + 2H<br />
4<br />
−<br />
2
8 Influence of groundwater composition<br />
had concentrations exceeding 6.7 µmol.L -1 . <strong>Arsenic</strong><br />
removal during subsurface treatment has also been<br />
observed by Rott <strong>and</strong> co-workers (2002), reducing<br />
arsenic concentrations from 0.5 µmol.L -1 to below 0.13<br />
µg.L -1 after 20 cycles. Van Halem et al. (2010) found<br />
less encouraging results at sites in Bangladesh with 1.9<br />
µg As(III).L -1 , with immediate arsenic breakthrough<br />
upon abstraction (Vi=1m 3 ). These pilot studies have<br />
indicated that there is potential for the technology<br />
of SAR, nevertheless at some sites the efficacy seems<br />
to be much better than at others. Unfortunately, only<br />
very little is reported on the soil <strong>and</strong> groundwater<br />
composition at these field sites, making it difficult to<br />
extract information on the influence of limiting <strong>and</strong>/<br />
or catalyzing compounds. Appelo <strong>and</strong> de Vet (2003)<br />
reported the behaviour of a wide range of groundwater<br />
compounds during subsurface treatment, proposing<br />
that phosphate was responsible for sudden arsenic peaks<br />
in the abstracted water. The arsenic concentrations<br />
at the studied site were, however, many times lower<br />
(80% breakthrough of dissolved O 2<br />
. Subsequently the<br />
influent was switched to groundwater to allow retention<br />
of Fe 2+ <strong>and</strong> As(III). Electrical conductivity was used<br />
as a conservative tracer from which the pore volume<br />
could be calculated to be, on average, 0.12 L (±0.002).<br />
The flow rate in the columns (2.7 m.h -1 ) was controlled<br />
with a multi-channel pump <strong>and</strong> PVC tubing with low<br />
gas permeability. Anoxic conditions were maintained<br />
113<br />
8
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
H 2<br />
O<br />
pH ORP EC DO<br />
Waste/Sample<br />
Point<br />
N 2<br />
Stainless steel Pall Rings<br />
H 2<br />
O<br />
+<br />
O 2<br />
Discharge Pump<br />
Dosing Pump<br />
Column 2<br />
Column 1<br />
Fe/As<br />
NaHCO 3<br />
Figure 8.1 Schematic overview of experimental column set-up<br />
8<br />
114<br />
in the columns by using an airtight FESTO system (6<br />
x 1 PUN, I.D. 4 mm) with matching connectors <strong>and</strong><br />
valves. All injection-abstraction experiments were<br />
performed twice in the duplicate columns, with freshly<br />
filled columns before each experiment. At the start of<br />
each experiment the columns were conditioned with<br />
synthetic groundwater (containing the compounds<br />
of interest), until complete breakthrough of iron<br />
occurred, <strong>and</strong> the Eh potential stabilized. An injection<br />
mode consisted of demineralized water containing<br />
a pH buffer (5mM NaHCO 3<br />
) <strong>and</strong> 0.28mM O 2<br />
. The<br />
abstraction phase consisted of synthetic groundwater<br />
with pH 6.9, a temperature of 20°C, 0.07mM Fe 2+ , 2.7µM<br />
As(III), pH buffer (5mM NaHCO 3<br />
) , ionic strength<br />
buffer (1.6mM NaCl) <strong>and</strong> additionally the inorganic<br />
compound of interest (phosphate, silicate, nitrate,<br />
calcium or manganese). The chemicals (reagent grade,<br />
Sigma-Alldrich) were dosed as FeSO 4<br />
.7H 2<br />
O, FeCl 2<br />
,<br />
NaAsO 2<br />
, NaHCO 3<br />
, NaCl, Na 2<br />
HPO 4<br />
, Na 2<br />
SO 3<br />
.5H 2<br />
O,<br />
NaNO 3<br />
, Ca(NO 3<br />
) 2<br />
.4H 2<br />
O <strong>and</strong> MnSO 4<br />
.H 2<br />
O.
8 Influence of groundwater composition<br />
The synthetic groundwater was produced by<br />
distributing demineralized water through a 6m gas<br />
stripping column containing stainless steel Pall Rings.<br />
From the bottom pure N 2<br />
was blown into the degassing<br />
column to sparge out all O 2<br />
. Before entering the s<strong>and</strong><br />
columns, the water was checked for O 2<br />
(Orbisphere;<br />
HACH Lange; M1100 Sensor; 410 Analyser) to ensure<br />
concentrations below 1.5 µmol.L -1 . Addition of stock<br />
solutions was done with a dosing pump followed by a<br />
static mixer. pH correction was achieved by addition of<br />
HCl or NaOH <strong>and</strong> all stock solutions were sparged with<br />
N 2<br />
in order to ensure the absence of O 2<br />
.<br />
<strong>Iron</strong> analysis of the water samples was done<br />
with an Atomic Absorption Spectrometer (Perkin-<br />
Elmer Flame AAS 3110). <strong>Arsenic</strong> analysis with<br />
Graphite Furnace Atomic Absorption Spectrometer<br />
(Perkin-Elmer 5100PC) with an electrode discharge<br />
lamp (EDL) <strong>and</strong> a Ni(NO 3<br />
) 2<br />
.6H 2<br />
O matrix modifier.<br />
C/C0<br />
1<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0<br />
R Fe = 42<br />
0 20 40 60 80<br />
pore volumes<br />
Fe Column 1<br />
Fe Column 2<br />
oxygen<br />
tracer<br />
Figure 8.2 Behaviour of O 2<br />
, tracer <strong>and</strong> Fe 2+ during the reference<br />
injection-abstraction cycles<br />
In-line measurements were done for dissolved oxygen<br />
(Orbisphere <strong>and</strong> WTW Cellox 325), E h<br />
potential (WTW<br />
SenTix ORP), pH (WTW SenTix 41), <strong>and</strong> electrical<br />
conductivity (WTW TetraCon 325). Measurements<br />
were registered on a computer with Multilab Pilot v5.06<br />
software.<br />
Results<br />
Fe 2+ -O 2<br />
system<br />
A regular injection phase during full scale subsurface<br />
iron or arsenic removal consists of the injection of<br />
aerated water, in most cases drinking water from the<br />
clean water reservoir. In the experiments described in<br />
the paper, demineralized water was used for injection<br />
containing a pH buffer (5mM NaHCO 3<br />
) <strong>and</strong> a 0.28mM<br />
O 2<br />
concentration. After injection, the influent was<br />
switched to synthetic groundwater for the abstraction<br />
phase, containing 0.07mM Fe 2+ , 2.7µM As(III), pH<br />
buffer (5mM NaHCO 3<br />
) <strong>and</strong> ionic strength buffer (1.6<br />
mM NaCl). The measurements in Figure 8.2 illustrate<br />
that once abstraction has started first the tracer passed<br />
the column (electrical conductivity), with a C/C 0<br />
at<br />
1 pore volume. The initial oxygen concentration of<br />
0.28mM was pushed out of the column <strong>and</strong> reached<br />
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
8<br />
116<br />
the delay of Fe front arrival at the well compared to<br />
the injection water front. The more water with low Fe<br />
concentrations can be abstracted from the well/column,<br />
the better the efficacy of SIR.<br />
In general, groundwater treatment plants<br />
operate a V/Vi ratio based on the moment Fe starts to<br />
arrive at the well, thus C/C 0<br />
>0. In the experiments, Fe 2+<br />
is allowed to breakthrough, in order to calculate the<br />
dimensionless retardation factor R. R Fe<br />
-1 is calculated<br />
from a Vi corresponding to the pore volume when the<br />
tracer (electrical conductivity) is C/C 0<br />
= 0.5, <strong>and</strong> V is<br />
the number of pore volumes that can be abstracted with<br />
iron concentrations below (C/C 0<br />
) Fe<br />
=0.5. The reference<br />
injection-abstraction cycles in Figure 8.2 show a R Fe<br />
of<br />
42, meaning that 43 times more water with [Fe]
8 Influence of groundwater composition<br />
Clearly, the presence of Fe 2+ in the changing redox<br />
conditions in the column improved As(III) retention. It<br />
should be noted that, unlike for Fe, the As breakthrough<br />
starts almost immediately upon abstraction. The As<br />
breakthrough showed extensive tailing, resulting in<br />
a high R As<br />
. The tailing during As(III) breakthrough<br />
may indicate incomplete adsorption kinetics in the<br />
columns, which is not surprising as residence times<br />
of the synthetic groundwater in the column was no<br />
more than 10 minutes (empty bed contact time). Such<br />
residence times correlate well with actual retention<br />
times in small-scale applications of SIR/SAR (Sarkar<br />
<strong>and</strong> Rahman, 2001; van Halem et al., 2010), but reaching<br />
As(III) adsorption equilibrium on iron oxides may take<br />
longer than that (Raven et al., 1998). Irregular operation<br />
of h<strong>and</strong> pump wells may increase the residence time<br />
of groundwater in the oxidation zone, potentially<br />
increasing removal efficacies. Additionally it should<br />
be taken into account that although As breakthrough<br />
is observed immediately, the arsenic concentration<br />
passed the WHO <strong>and</strong> Bangladeshi guideline at 4 <strong>and</strong><br />
8 pore volumes, respectively. An SIR/SAR efficiency<br />
of V/Vi=4 without the need of any operational costs<br />
(chemicals, electricity) may well be very effective for<br />
household application.<br />
Phosphate<br />
The results in Figure 8.2 <strong>and</strong> Figure 8.3 of the reference<br />
injection-abstraction cycles in the Fe 2+ -As(III)-O 2<br />
system show promising results for the household<br />
application of SIR/SAR. Nevertheless, the presence<br />
of other inorganic groundwater constituents may<br />
interfere, reducing Fe <strong>and</strong> As removal efficacies. In<br />
Bangladesh, PO 4<br />
3-<br />
concentrations are measured in<br />
the shallow aquifer between 0.01mM PO 4<br />
3-<br />
at pH>7 inhibits the precipitation of Fe 3+ (Ciardelli et<br />
al., 2008; Guan et al., 2009). The adsorbed H 2<br />
PO 4<br />
-<br />
<strong>and</strong><br />
HPO 4<br />
2-<br />
maintains the iron oxide surfaces as negatively<br />
charged preventing the particles from hydrolyzing<br />
<strong>and</strong> subsequently precipitating. Sharma et al. (2001)<br />
observed reduced adsorptive capacity for Fe 2+ of virgin<br />
s<strong>and</strong> in the presence of 0.01mM PO 4<br />
3-<br />
, 23% at pH<br />
6.8. Nevertheless, the results in Figure 8.4 illustrate<br />
that subsurface iron removal in the s<strong>and</strong> columns is<br />
not affected by the presence of PO 4<br />
3-<br />
, Fe retardation<br />
C/C0<br />
1<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0<br />
R As = 4<br />
R Fe = 44<br />
0 20 40 60 80<br />
pore volumes<br />
Fe Column 1<br />
As Column 1<br />
Fe Column 2<br />
As Column 2<br />
Figure 8.4 Fe <strong>and</strong> As retardation in the presence of 0.01 mM<br />
PO 4<br />
3-<br />
117<br />
8
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
8<br />
118<br />
remains almost the same at 44. This may be explained<br />
by the fact that the s<strong>and</strong> grains (partially) retained the<br />
non-hydrolysed iron oxide particles through filtration,<br />
adsorption or exchange, whereas the batch experiments<br />
from the literature did not include filtration.<br />
The effect of phosphate on the removal of As(III)<br />
during subsurface arsenic removal is two-fold, namely<br />
through (i) inhibition of Fe 2+ oxidation-precipitation<br />
<strong>and</strong> (ii) competitive adsorption. Competitive<br />
adsorption between As(III) <strong>and</strong> PO 4<br />
3-<br />
onto iron oxides<br />
has been studied in detail (Jain <strong>and</strong> Loeppert, 2000;<br />
Ciardelli et al., 2007; Stachowicz et al., 2008), making<br />
PO 4<br />
3-<br />
the most important competing anion for arsenic<br />
(Su <strong>and</strong> Puls, 2001; Guan et al., 2009). As(III) behaviour<br />
during oxidation <strong>and</strong> precipitation of soluble Fe 2+ was<br />
studied by Ciardelli et al. (2008), reporting an As(III)<br />
removal reduction from 73% to 63% in the presence<br />
of 0.07mM PO 4<br />
3-<br />
-P at pH 7.2. The combination of its<br />
ability to occupy the same adsorption sites as arsenic<br />
<strong>and</strong> its elevated concentrations compared to trace<br />
amount of arsenic makes PO 4<br />
3-<br />
a strong competitor in<br />
natural waters. This observation correlates well with<br />
the results from the column experiments (Figure 8.4),<br />
showing that As retardation was reduced from 30 to 4<br />
in the presence of 0.01mM PO 4<br />
3-<br />
.<br />
Silicate<br />
Silicon concentrations of groundwater in Bangladesh<br />
are commonly high, with median concentrations in the<br />
shallow aquifer of 0.71 mmol.L -1 (BGS/DPHE, 2001).<br />
C/C0<br />
1<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0<br />
R As = 5<br />
R Fe = 57<br />
0 20 40 60 80 100<br />
pore volumes<br />
Fe Column 1<br />
As Column 1<br />
Fe Column 2<br />
As Column 2<br />
Figure 8.5 Fe <strong>and</strong> As retardation in the presence of 0.2mM SiO 2<br />
Silicon, or in groundwater found as silicate, is known<br />
to interact with Fe 3+ to form soluble polymers <strong>and</strong> higly<br />
dispersed colloids (Iler, 1979; Meng et al., 2000) <strong>and</strong><br />
hinder the Fe 3+ iron oxide precipitation <strong>and</strong> aggregation<br />
(Lui et al., 2007a; Lui et al., 2007b). The negative effect<br />
of silicate on the precipitation of iron oxides is greater<br />
at higher pH (7-9). This dispersed colloidal material<br />
cannot be removed by filtration (Davis et al., 2001),<br />
so it may be found to pass the injection-abstraction<br />
columns. Fe 2+ adsorption onto virgin s<strong>and</strong> has also<br />
been reported to be inhibited by the presence of silicate,<br />
reducing by 14 <strong>and</strong> 27% in the presence of 10 <strong>and</strong> 1.4<br />
mmol.L -1 Si, respectively (Sharma, 2001). The results in<br />
Figure 8.5 do not show a clear inhibition of iron removal<br />
after the addition of 0.2mM SiO 2<br />
, the retardation factor<br />
R Fe<br />
is even higher at 57. It is, however, noteworthy<br />
that the iron concentrations rise immediately after<br />
one pore volume to C/C 0<br />
=0.06, which correlates to<br />
an approximate 5.4 µM Fe.L -1 . This instant increase
8 Influence of groundwater composition<br />
in iron concentration has not been observed in the<br />
absence of silicate <strong>and</strong> continues until 40 pore volumes.<br />
It is proposed that the formation of mobile colloidal<br />
material prevents the adsorption of this iron fraction<br />
onto the column s<strong>and</strong>, resulting in measurements of<br />
total iron after the columns.<br />
Below pH 7 silicate is found to occur as H 2<br />
SiO 3<br />
,<br />
whereas at higher pH it may be found as HSiO 3-<br />
. At<br />
higher pH the affinaty of the negatively charged silicate<br />
for adsorption onto iron oxides increases, resulting in<br />
more competition with arsenic (Meng et al., 2000; Guan<br />
et al., 2009). Ciardelli et al. (2008) found a reduction<br />
in As(III) removal from 73% to 67% during oxidationprecipiation<br />
of Fe 2+ after addition of 1.1 mmol.L -1 .<br />
Figure 8.5 illustrates that subsurface arsenic removal<br />
is negatively affected by the presence of 0.2mM SiO 2<br />
.<br />
The retardation factor R As<br />
was reduced from 30 to 65<br />
which is very much comparable to the impact of 0.01<br />
mM PO 4<br />
3-<br />
.<br />
Nitrate<br />
In anaerobic groundwater, an NO 3-<br />
-reducing<br />
environment, one expects to find dissolved NH 4<br />
+<br />
rather<br />
than NO 3-<br />
. Nevertheless, the aeration <strong>and</strong> storage of<br />
anaerobic groundwater for the purpose of injection<br />
may lead to nitrification of NH 4<br />
, resulting in elevated<br />
NO 3<br />
-<br />
concentrations. Additionally, in Bangladesh,<br />
shallow aquifers have been measured to contain nitrate<br />
concentrations up to 1.2 mM (Majumder et al., 2008).<br />
The effect of 1mM NO 3<br />
-<br />
in the synthetic groundwater<br />
C/C0<br />
1<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0<br />
R As = 4<br />
R Fe = 47<br />
0 20 40 60 80 100<br />
pore volumes<br />
Fe Column 1<br />
As Column 1<br />
Fe Column 2<br />
As Column 2<br />
Figure 8.6 Fe <strong>and</strong> As retardation in the presence of 1mM NO 3<br />
-<br />
on the iron <strong>and</strong> arsenic retardation was studied during<br />
the injection-abstraction cycles in the columns. Figure<br />
8.6 shows that Fe removal did not differentiate much<br />
compared to the reference cycle, with a slightly higher<br />
R Fe<br />
of 47. Previous studies showed that addition of<br />
NaNO 3<br />
had no effect on Fe 2+ adsorption onto virgin<br />
s<strong>and</strong> <strong>and</strong> iron oxides (Hayes <strong>and</strong> Leckie, 1987; Dzombak<br />
<strong>and</strong> Morel, 1990; Sharma, 2001).<br />
Su et al. (2001) investigated the influence of<br />
nitrate on As(III) removal <strong>and</strong> found that when adding<br />
1 mM NaNO 3<br />
to a solution containing As(III) <strong>and</strong><br />
zerovalent iron oxides, As(III) removal was reduced<br />
by 55%. Push-pull column studies by Johnston (2008),<br />
where a 0.1M NaNO 3<br />
ionic strength buffer was used,<br />
showed As(III) retardation of R As<br />
=6-10 at pH 7 with<br />
goethite coated s<strong>and</strong> columns. The R As<br />
in that study was<br />
significantly lower than the reference cycles presented<br />
in Figure 8.3, but corresponds more or less to the R As<br />
in Figure 8.6. In the presence of 1mM NO 3<br />
-<br />
the arsenic<br />
119<br />
8
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
8<br />
120<br />
retardation factor was reduced from 30 to 4, illustrating<br />
the significant effect of this constituent on subsurface<br />
arsenic removal.<br />
Calcium<br />
One of the major cations in groundwater is calcium,<br />
especially in the calcium-carbonate groundwaters of<br />
Bangladesh (Ciardelli et al., 2007). Ca 2+ concentrations<br />
are generally much higher than Fe 2+ concentrations,<br />
with a median <strong>and</strong> maximum in Bangladesh of<br />
0.88mM <strong>and</strong> 9.1mM, respectively (BGS/DPHE, 2001).<br />
Competition/exchange between these divalent cations<br />
during SIR may therefore be expected, with Ca 2+ being<br />
in favour because of its concentration advantage. A<br />
reduction in Fe 2+ adsorption onto virgin <strong>and</strong> iron oxide<br />
coated s<strong>and</strong> of 42% <strong>and</strong> 12%, respectively, was observed<br />
in the presence of 1.2mM Ca 2+ at pH 6.8 (Sharma, 2001).<br />
Calcium was added in this column study as Ca(NO 3<br />
) 2<br />
,<br />
resulting in 1.2mM Ca 2+ , but also 2.4mM NO 3-<br />
. The<br />
C/C0<br />
1<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0<br />
R As = 4<br />
R Fe = 15-24<br />
0 10 20 30 40<br />
pore volumes<br />
Fe Column 1<br />
As Column 1<br />
Fe Column 2<br />
As Column 2<br />
Figure 8.7 Fe <strong>and</strong> As retardation in the presence of 1.2mM Ca 2+<br />
previous experiment illustrated that NO 3<br />
-<br />
had almost<br />
no effect on Fe removal during SIR, but there was a<br />
strong effect on As removal. Therefore conclusions can<br />
be drawn based on the results presented in Figure 8.7<br />
regarding the influence of Ca 2+ on Fe removal, but not<br />
on the removal of As. Stachowicz et al. (2008) did not<br />
observe an effect of Ca 2+ on As(III) adsorption onto<br />
goethite. It is shown in Figure 8.7 that the R As<br />
is almost<br />
the same after addition of Ca(NO 3<br />
) 2<br />
as after addition<br />
of NaNO 3<br />
, thus not providing interesting information<br />
on the effect on Ca 2+ . The R Fe<br />
was, however, clearly<br />
lower than in earlier injection-abstraction cycles,<br />
with a variation between 15 <strong>and</strong> 24 for the duplicate<br />
experiments. The variation between the duplicate cycles<br />
cannot be explained at this stage, but the reduced R Fe<br />
confirms that the presence of Ca 2+ inhibited the removal<br />
of Fe 2+ during SIR. Fe concentrations start to rise in the<br />
water after approximately 5 to 8 pore volumes, resulting<br />
in an operational V/Vi ratio of 5 to 8, which is much<br />
lower than the V/Vi=25 of the reference cycle.<br />
Manganese<br />
<strong>Subsurface</strong> iron removal has frequently been associated<br />
with co-removal of manganese (Hallberg <strong>and</strong> Martinell,<br />
1976; van Beek 1985; Rott, 1985). Mn 2+ oxidation is<br />
slower than abiotic Fe oxidation <strong>and</strong> it is proposed that<br />
manganese removal does not start in the subsurface<br />
until iron removal is complete. This results in a spatial<br />
separation between the formation of manganese <strong>and</strong><br />
iron oxides, with the latter further away from the
8 Influence of groundwater composition<br />
C/C0<br />
1<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0<br />
Fe Column 1<br />
As Column 1<br />
Fe Column 2<br />
As Column 2<br />
R As = 28<br />
0 10 20 30 40 50<br />
pore volumes<br />
R Fe = 34<br />
Figure 8.8 Fe <strong>and</strong> As retardation in the presence of 0.06 mM<br />
Mn 2+<br />
well. Mn 2+ oxidation is known to be catalyzed in the<br />
presence of Mn 4+ oxides, <strong>and</strong> also As(III) oxidation<br />
<strong>and</strong> adsorption is suggested to be enhanced by Mn 4+<br />
oxides. Nevertheless, Mn 2+ is not expected to compete<br />
with Fe 2+ for the injected O 2<br />
in the columns, as the Mn 2+<br />
oxidation kinetics are much slower <strong>and</strong> the contact time<br />
in the columns is relatively short. Mn 2+ may, however,<br />
inhibit the Fe 2+ adsorption. Sharma (2001) observed a<br />
reduction in Fe 2+ adsorption in the presence of 0.03mM<br />
Mn onto virgin <strong>and</strong> iron coated s<strong>and</strong> with 44% <strong>and</strong> 6%,<br />
respectively.<br />
During injection-abstraction cycles with<br />
0.06mM Mn 2+ , the Fe removal was slightly lower<br />
compared to the reference cycles, with a R Fe<br />
of 34<br />
(Figure 8.8). The moment that Fe 2+ was detected did<br />
not differ from the reference cycles, showing that the<br />
operational V/Vi ratio is not affected by the presence of<br />
0.06mM Mn. As(III) behaviour remained the same in<br />
the presence of soluble Mn 2+ , resulting in R As<br />
=28.<br />
Conclusions<br />
It was found that during injection-abstraction cycles<br />
in the Fe 2+ -As(III)-O 2<br />
column system it was possible<br />
to reach very effective iron <strong>and</strong> arsenic retardation,<br />
though arsenic breakthrough started immediately<br />
upon abstraction. <strong>Arsenic</strong> removal was found to be<br />
significantly inhibited by the presence of 0.01mM<br />
phosphate, 0.2mM silicate, <strong>and</strong> 1mM nitrate, illustrating<br />
the vulnerability of this <strong>Subsurface</strong> <strong>Arsenic</strong> <strong>Removal</strong><br />
technology in diverse geochemical settings. <strong>Subsurface</strong><br />
<strong>Iron</strong> <strong>Removal</strong> was found not be as sensitive to other<br />
inorganic groundwater compounds, though iron<br />
retardation was found to be limited by 1.2mM calcium<br />
<strong>and</strong> 0.06mM manganese. Total iron removal was<br />
enhanced by the presence of 0.2mM silicate, but low<br />
iron concentrations (5µM) were found to breakthrough<br />
immediately upon abstraction.<br />
References<br />
Appelo C. A. J. <strong>and</strong> W. W. J. M. de Vet (2003) Modeling in situ<br />
iron removal from groundwater with trace elements such<br />
as As. In <strong>Arsenic</strong> in groundwater. A. H. Welch <strong>and</strong> K.G.<br />
Stollenwerk. Kluwer Academic, Boston.<br />
Ciardelli M.C., H. Xu <strong>and</strong> N. Sahai (2008) Role of Fe(II),<br />
phosphate, silicate, sulfate, <strong>and</strong> carbonate in arsenic uptake<br />
by coprecipitation in synthetic <strong>and</strong> natural groundwater<br />
121<br />
8
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
8<br />
Water Research 42(3): 615-624.<br />
Davis C.C., W.R. Knocke <strong>and</strong> M. Edwards (2001) Implications<br />
of aqueous silica sorption to iron hydroxide: mobilization of<br />
iron colloids <strong>and</strong> interference with sorption of arsenate <strong>and</strong><br />
humic substances. Environ. Sci. Technol. 35: 3158-3162.<br />
Dixit S. <strong>and</strong> J. G. Hering (2003) Comparison of arsenic(V) <strong>and</strong><br />
arsenic(III) sorption onto iron oxide minerals: implications<br />
for arsenic mobility. Environ. Sci. Technol. 37: 4182-4189.<br />
Dixit S. <strong>and</strong> J. G. Hering (2006) Sorption of Fe(II) <strong>and</strong> As(III)<br />
on goethite in single- <strong>and</strong> dual-sorbate systems, Chemical<br />
Geology, 228(1-3): 6-15.<br />
Dzombak D.A. <strong>and</strong> F.M.M. Morel (1990) Surface complexation<br />
modelling – hydrous ferric oxides. John Wiley & Sons.<br />
Guan X., H. Dong, J. Ma <strong>and</strong> L. Jiang (2009) <strong>Removal</strong> of arsenic<br />
from water: effects of competing anions on As(III) removal in<br />
KMnO 4<br />
-Fe(II) process. Water Research 43: 3891-3899.<br />
Hallberg R. O. <strong>and</strong> R. Martinell (1976) Vyredox - in situ<br />
purification of groundwater. Ground Water 14(2): 88-93.<br />
Hayes K.F. <strong>and</strong> J.O. Leckie (1987) Modelling ionic strength effects<br />
on cation adsorption ad hydrous oxide/solution interfaces.<br />
Journal of Colloid <strong>and</strong> Interface Science (115): 564-572.<br />
Iler R. K. (1979) The Chemistry of Silica. Wiley-Interscience,<br />
New York.<br />
Johnston R. (2008) Chemical interactions between iron <strong>and</strong><br />
arsenic in water, PhD dissertation, University of North<br />
Carolina at Chapel Hill.<br />
Liu R.P., J.H. Qu, S.J. Zia, G.S. Zhang <strong>and</strong> G.B. Li (2007a) Silicate<br />
hindering in situ formed ferric hydroxide precipiation:<br />
inhibing arsenic removal from water. Environ. Eng. Sci. 24:<br />
707-715.<br />
122<br />
Liu R.P., X. Li, S.J. Zia, Y.L. Yang, R.C. Wu <strong>and</strong> G.B. Li (2007b)<br />
Calcium-enhanced ferric hydroxide co-precipitation of<br />
arsenic in the presence of silicate. Water Environ. Res. 79:<br />
2260-2264.<br />
Majumder R.K., M.A. Hasnat, S. Hossain, K. Ikeue, M.<br />
Machida (2008) An exploration of nitrate concentrations in<br />
groundwater aquifers of central-west region of Bangladesh.<br />
Journal of Hazardous Materials 159(2-3): 536-543.<br />
Meng X., S. Bang <strong>and</strong> G.P. Korfiatis (2000) Effects of silicate,<br />
sulfate, <strong>and</strong> carbonate on arsenic removal by ferric chloride.<br />
Water Research 34(4): 1255-1261.<br />
Nickson R., J. McArthur, P. Ravencroft, W.G. Burgess <strong>and</strong> K.M.<br />
Ahmed (2000) Mechanism of arsenic release to groundwater,<br />
Bangladesh <strong>and</strong> West Bengal, Applied Geochemistry 15(4):<br />
403-413.<br />
Raven K.P., A. Jain <strong>and</strong> R.H. Loeppert (1998) Arsenite <strong>and</strong><br />
arsenate adsorption on ferrihydrite: kinetics, equilibrium,<br />
<strong>and</strong> adsorption envelopes. Environ. Sci. Technol. 32: 344-349.<br />
Rott U. (1985) Physical, chemical <strong>and</strong> biological aspects of the<br />
removal of iron <strong>and</strong> manganese underground. Water Supply<br />
3(2): 143-150.<br />
Sarkar A. R. <strong>and</strong> O.T. Rahman (2001) In-situ removal of arsenic<br />
- experiences of DPHE-Danida pilot project. In Technologies<br />
for arsenic removal from drinking water, Bangladesh<br />
University of Engineering <strong>and</strong> Technology <strong>and</strong> The United<br />
Nations University, Bangladesh.<br />
Sen Gupta B., S. Chatterjee, U. Rott, H. Kauffman, A.<br />
B<strong>and</strong>opadhyay, W. de Groot, N.K. Nag, A.A. Borbonell-<br />
Barrachina <strong>and</strong> S. Mukherjee (2009) A simple chemical free<br />
arsenic removal method for community, Environmental
8 Influence of groundwater composition<br />
Pollution 157: 3351–3353.<br />
Sharma, S.K. (2001) Adsorptive iron removal from groundwater.<br />
PhD dissertation, Wageningen University.<br />
Stachowicz M., T. Hiemstra <strong>and</strong> W. H. Van Riemsdijk (2008)<br />
Multi-competitive interaction of As(III) <strong>and</strong> As(V) oxyanions<br />
with Ca 2+ , Mg 2+ 3− 2−<br />
, PO 4<br />
, <strong>and</strong> CO 3<br />
ions on goethite, Journal of<br />
Colloid <strong>and</strong> Interface Science 320, 400–414.<br />
Su C., <strong>and</strong> R.W. Puls (2001) Arsenate <strong>and</strong> arsenite removal by<br />
zerovalent iron: effects of phosphate, silicate, carbonate,<br />
borate, sulphate, chromate, molybdate, <strong>and</strong> nitrate, relative<br />
to chloride. Environmental Science <strong>and</strong> Technology<br />
35(22):4562-4568.<br />
van Beek C. G. E. M. (1985) Experiences with underground<br />
water treatment in the Netherl<strong>and</strong>s. Water Supply 3(2): 1-11.<br />
van Halem, D., S. Olivero, W.W.J.M. de Vet, J.Q.J.C. Verberk,<br />
G.L. Amy <strong>and</strong> J.C. van Dijk (2010) <strong>Subsurface</strong> iron <strong>and</strong><br />
arsenic removal for shallow tube well drinking water in rural<br />
Bangladesh. Water Research 44: 5761-5769.<br />
8<br />
123
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
8<br />
124
Concluding remarks9
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
9<br />
126<br />
Main conclusion<br />
In this concluding chapter the results from this thesis<br />
are combined to address the main problem definition<br />
as stated in Chapter 1. The combination of s<strong>and</strong> column<br />
studies, field experiments, sediment characterization<br />
<strong>and</strong> past experiences in the Netherl<strong>and</strong>s have gained<br />
knowledge on the applicability of SIR/SAR for smallscale<br />
drinking water treatment. The considerations per<br />
knowledge gap have contributed to the formulation of<br />
the main conclusion of this thesis:<br />
The results presented in this thesis have shown<br />
that <strong>Subsurface</strong> <strong>Iron</strong> <strong>Removal</strong> is a safe <strong>and</strong><br />
sustainable small-scale drinking water treatment<br />
solution. <strong>Subsurface</strong> <strong>Arsenic</strong> <strong>Removal</strong> has been<br />
found to be less suitable for h<strong>and</strong> pump application<br />
<strong>and</strong> vulnerable to diverse geochemical conditions.<br />
This conclusion underlines the great potential of SIR<br />
for household water treatment <strong>and</strong> storage (HWTS),<br />
but also points out the variable performance of SAR.<br />
This two-fold conclusion does not mean that the<br />
application of SAR should no longer be considered a<br />
safe option, but indicates that this novel technology is<br />
no silver bullet for arsenic mitigation. The efficacy of<br />
SAR is sensitive to geochemical variations, which is<br />
in conflict with the decentralized character of HWTS.<br />
This vital observation, however, applies not just for SAR<br />
but for the majority of (arsenic) HWTS solutions. In<br />
general, no household water treatment option has been<br />
designed to suit all conditions, illustrating the urgent<br />
need to investigate the (technical) boundary conditions<br />
in which a HWTS solution can succeed. In this thesis<br />
the efficacy of SIR/SAR has been described as the ratio<br />
between the injected volume (Vi) <strong>and</strong> the abstracted<br />
volume (V). The more water that can be abstracted<br />
with reduced iron <strong>and</strong> arsenic, the lower the frequency<br />
of injection <strong>and</strong> the higher the efficacy. This thesis has<br />
shown that removal of iron was more encouraging<br />
than for arsenic, meaning that for SAR operation users<br />
need to inject more frequently (depending on the<br />
groundwater composition). It therefore depends on<br />
the users’ willingness to inject whether SIR/SAR can be<br />
considered an appropriate <strong>and</strong> safe alternative to current<br />
arsenic mitigation solutions. The SIR/SAR technology<br />
has many socio-economic advantages (Chapter 1),<br />
including the removal of iron. The odour, taste <strong>and</strong><br />
colour of the drinking water are significantly improved<br />
by SIR/SAR, which could potentially increase the social<br />
acceptance <strong>and</strong> willingness to operate the technology<br />
at low V/Vi ratios. A study on the socio-economic<br />
feasibility of this technology, within the formulated<br />
technological boundary conditions, is therefore highly<br />
recommended.<br />
The three identified knowledge gaps in Chapter 1<br />
are addressed in this concluding chapter: (1) Fe 2+ /As(III)<br />
immobilization processes; (2) site-specific effectiveness<br />
in diverse (geochemical) conditions; <strong>and</strong> (3) safe <strong>and</strong><br />
sustainable application. The findings of every individual<br />
chapter are placed in a broader perspective to support<br />
the main conclusion of this thesis. Additionally an
9 Concluding remarks<br />
outlook is given on the recommended research <strong>and</strong> the<br />
future of small-scale SIR/SAR.<br />
Fe 2+ /As(III) immobilization processes<br />
A complex combination of (heterogeneous) oxidation,<br />
precipitation, adsorption <strong>and</strong> cation exchange reactions<br />
simultaneously occur in the subsurface, making it<br />
difficult to differentiate between these processes in<br />
field experiments. In s<strong>and</strong> column experiments the<br />
processes of adsorptive-catalytic oxidation <strong>and</strong> Fe 2+<br />
exchange were isolated. During adsorptive-catalytic<br />
oxidation, the consumption of 1 mole of O 2<br />
resulted in<br />
the stoichiometric removal of 4 moles of Fe 2+ (Chapter<br />
4 <strong>and</strong> 7). Columns containing sediments from within<br />
a 12-year-old oxidation zone consumed more oxygen<br />
during injection than clean filter s<strong>and</strong> <strong>and</strong> subsequently<br />
retained more Fe 2+ during abstraction (Chapter 7).<br />
Although the adsorptive-catalytic oxidation process<br />
could describe the regeneration of the subsurface<br />
by injection of dissolved O 2<br />
, it could not account for<br />
the improved efficiency after multiple cycles. The<br />
retardation of Fe was found to increase significantly per<br />
cycle at the small-scale SIR/SAR set-ups in Bangladesh.<br />
After multiple cycles, the Fe 2+ removal efficacies at these<br />
sites exceeded the stoichiometric Fe 2+ :O 2<br />
ratio for Fe 2+<br />
oxidation (Chapter 3). This efficiency increase can be<br />
described based on the memory of the SIR/SAR system,<br />
i.e., previous injection-abstraction cycles have not used<br />
the freshly formed adsorptive surface sites completely<br />
because the injection phase was started before complete<br />
breakthrough of Fe 2+ . Hence, these adsorption sites<br />
are available for future cycles, resulting in increasing<br />
efficacy (modelled by Appelo et al., 1999). The size<br />
of the oxidation zone should therefore be considered<br />
dynamic in size <strong>and</strong> water quality.<br />
This is also illustrated by the behaviour of<br />
exchangeable Fe 2+ during injection. In the absence of O 2<br />
<strong>and</strong> in the presence of high cation concentrations the<br />
adsorbed Fe 2+ was found be released from the s<strong>and</strong> grain<br />
surface <strong>and</strong> pushed deeper into the aquifer (Chapter<br />
5). The role of this cation exchange process in the<br />
efficacy of SIR/SAR may nevertheless be insignificant,<br />
as exchanged Fe 2+ will either be oxidized or adsorbed<br />
in- or outside the oxidation zone. It was hypothesized<br />
by Appelo et al. (1999) that injection of higher oxidant<br />
concentrations would not enhance SIR performance, as<br />
the available exchangeable Fe 2+ determines the efficacy.<br />
Field results with injection of high O 2<br />
concentrations,<br />
0.55mM instead of 0.28mM, increased the operational<br />
V/Vi ratio from an average 7 to 16 (WTP Corle, Chapter<br />
5). This indicates that the exchangeable Fe 2+ on the soil<br />
material is not necessarily the limiting factor during<br />
injection, but rather the supply of O 2<br />
to the subsurface<br />
Fe 2+ . Apart from the O 2<br />
concentration in the injection<br />
water, also the pH of the groundwater strongly affects<br />
the Fe 2+ immobilization. At higher pH (>6.5) Fe 2+<br />
removal was found to be enhanced (Chapter 2), which<br />
is clearly based on the catalysed Fe 2+ oxidation <strong>and</strong><br />
adsorption reaction in higher pH ranges (Tamura et al.,<br />
1980; Sharma, 2001; Dixit <strong>and</strong> Hering, 2006). Of the<br />
127<br />
9
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
9<br />
128<br />
investigated inorganic groundwater constituents, only<br />
Ca 2+ was found to significantly inhibit the removal of<br />
Fe 2+ during injection-abstraction cycles (Chapter 9).<br />
As(III) behaviour during adsorptive-catalytic<br />
oxidation in the columns proved to be independent of<br />
the sediment source, Fe:As ratio <strong>and</strong> successive cycles<br />
(Chapter 3, 4 <strong>and</strong> 7). The latter was confirmed at the<br />
small-scale SIR/SAR set-up in Bangladesh, where<br />
arsenic was found in the abstracted water immediately<br />
upon abstraction (Chapter 3). Unlike iron removal,<br />
arsenic removal did not benefit from enhanced O 2<br />
consumption in the columns (Chapter 7), so co-removal<br />
of As(III) during SIR was not evident. As(III) retention<br />
is a secondary process during SIR/SAR, depending<br />
completely on the (chemical) affinity of As(III) for the<br />
freshly formed Fe 3+ oxides (amorphous; Chapter 6).<br />
Speciation of As(III) <strong>and</strong> As(V) in the abstracted water<br />
revealed that As(V) was slightly more delayed than<br />
As(III), illustrating the charge advantage of As(V) over<br />
As(III) at near neutral pH (Chapter 3 <strong>and</strong> 4). In general<br />
it can be concluded that the adsorptive process of As(III)<br />
removal depends strongly on the presence of other (oxy)<br />
anions in the groundwater (PO 4<br />
3-<br />
, NO 3-<br />
, SiO 2<br />
; Chapter<br />
9). Investigations with natural groundwater, in column<br />
<strong>and</strong> full-scale experiments, have shown that although<br />
SIR could perform under these different geochemical<br />
conditions, As(III) removal was negatively influenced<br />
by these complex water matrices. Independent of the<br />
available surface sites, either in the aquifer sediment or<br />
freshly formed by Fe 2+ oxidation, the presence of other<br />
anions -usually in much higher concentrations than<br />
As(III)- limited the As(III) adsorption.<br />
Site-specific effectiveness in diverse<br />
geochemical conditions<br />
The presence of elevated arsenic concentrations in<br />
the groundwater shows a very strong spatial (wellto-well)<br />
variability, even if Fe 2+ concentrations are<br />
constant (Smedley <strong>and</strong> Kinniburgh, 2002). These local<br />
geochemical variations require that an arsenic mitigation<br />
solution performs in a wide range of conditions. In<br />
the absence of limiting (inorganic) constituents, the<br />
removal of Fe 2+ <strong>and</strong> As(III) was extremely effective in<br />
columns with synthetic groundwater (R Fe<br />
=42; R As<br />
=30)<br />
at pH 6.9 (Chapter 8). At the small-scale SIR/SAR setup<br />
in Bangladesh, with the same pH, the retardation<br />
of Fe <strong>and</strong> As was much lower (R Fe<br />
=14; R As<br />
=1; Chapter<br />
3). In order to reduce the risk of chronic arsenic<br />
poisoning, concentrations below the WHO guideline<br />
(10 µg.L -1 ) are desirable, but those were only reached<br />
during abstraction of the As-free injection water. The<br />
effect of groundwater composition on SIR/SAR was<br />
investigated with s<strong>and</strong> columns <strong>and</strong> it was found that<br />
subsurface arsenic removal was seriously inhibited by<br />
the presence of PO 4<br />
3-<br />
, NO 3-<br />
, <strong>and</strong> SiO 2<br />
, illustrating the<br />
vulnerability of this technology in diverse geochemical<br />
settings (Chapter 8). <strong>Subsurface</strong> iron removal was<br />
observed not to be as sensitive to other inorganic<br />
groundwater constituents, although iron retardation<br />
was limited by Ca 2+ . From three decades of experience
9 Concluding remarks<br />
with SIR in the Netherl<strong>and</strong>s it can be concluded that pH<br />
is a vital parameter. The operational V/Vi ratio more<br />
than doubled between pH 7.0 <strong>and</strong> 7.4 at WTP Corle<br />
(Chapter 2), showing that a low pH can inhibit effective<br />
SIR operation. A pH buffer (as HCO 3-<br />
) to maintain<br />
the pH during the acidifying Fe 2+ oxidation reaction is<br />
desired for operation of SIR, though this may entail the<br />
presence of elevated Ca 2+ concentrations.<br />
At the investigated sites in Bangladesh a higher<br />
Fe:As(III) ratio did not significantly enhance As removal<br />
(Chapter 3), which was also confirmed in column<br />
experiments (Chapter 4). From the perspective of<br />
adsorptive-catalytic oxidation this makes sense, because<br />
the formation of Fe 3+ oxides (for subsequent adsorption<br />
of As) depends on the supply of oxygen to the adsorbed<br />
Fe 2+ . As long as the soil grains are saturated with Fe 2+ ,<br />
the presence of higher Fe 2+ concentrations will not<br />
necessarily lead to more adsorptive surface area for As.<br />
<strong>Arsenic</strong> removal efficiencies from natural groundwater<br />
were limited by the presence of competing anions. The<br />
retention of Fe 2+ during injection-abstraction cycles<br />
was found to be successful in all experiments, both in<br />
the laboratory <strong>and</strong> field. The presence of accumulated<br />
deposits, mainly as Fe 3+ oxides, was found to enhance<br />
the adsorptive-catalytic oxidation process. For As(III)<br />
these deposits did not stimulate the removal process, as<br />
sediments from within the oxidation zone performed the<br />
same as clean filter (silica) s<strong>and</strong>. As(III) removal seems<br />
to rely primarily on the oxidation of Fe 2+ during a cycle,<br />
independent of the sediment composition. Though<br />
geochemical composition of sediments was not found<br />
to significantly influence As(III) removal, the presence<br />
of arsenic-bearing sulphide minerals may threaten<br />
the SAR technology during initial cycles. At WTP<br />
Lekkerkerk, arsenic peaks (
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
9<br />
130<br />
great potential of SIR, as the removal of iron improves<br />
colour <strong>and</strong> taste of the water; greatly enhancing social<br />
acceptance. From a technological perspective, the<br />
effective removal of Fe 2+ has an additional advantage,<br />
as the injection procedure start before arrival of Fe 2+ at<br />
the well. Aeration is more effective in the absence of<br />
Fe 2+ , making it easier to achieve high O 2<br />
concentrations<br />
without the production of iron sludge.<br />
HWTS may never threaten the long-term<br />
availability of a water source. For SIR/SAR this means<br />
that a shallow tube well may not produce water of lesser<br />
quality or quantity before, during or after use of this<br />
technology. Clogging of the aquifer around a subsurface<br />
iron removal well was addressed in Chapter 6, showing<br />
that after 12 years of operation, iron had accumulated<br />
at specific depths near the subsurface iron removal<br />
wells. Whether it was due to preferred flow paths<br />
or geochemical mineralogy conditions, subsurface<br />
iron removal clearly favoured certain soil layers. The<br />
majority of accumulated iron was characterized as<br />
crystalline, suggesting that precipitated amorphous<br />
iron oxides had transformed to iron oxides of higher<br />
crystallinity. These crystalline, compact iron oxides<br />
had not noticeably clogged the investigated well <strong>and</strong>/or<br />
aquifer. The subsurface iron removal wells even needed<br />
less frequent rehabilitation, as drawdown increase<br />
was slower than in normal production wells. Other<br />
groundwater constituents, such as manganese, arsenic<br />
<strong>and</strong> strontium were found to co-accumulate with iron.<br />
The accumulation of groundwater constituents<br />
around a SIR/SAR well is a direct consequence of this<br />
technology. And although it may seem undesirable<br />
from a sustainability perspective, these accumulated<br />
deposits do not harm human health or environment,<br />
as long as they remain immobile. Both Fe <strong>and</strong> As are<br />
abundantly available in sediments throughout the world<br />
(Smedley <strong>and</strong> Kinniburgh, 2002), but it merely depends<br />
on the local groundwater conditions whether these<br />
constituents will dissolve. The remobilization of iron,<br />
manganese, arsenic <strong>and</strong> phosphate was investigated<br />
when a SIR well was stopped after 18 injectionabstraction<br />
cycles (Chapter 2). At this particular<br />
site, none of the studied constituents was found to<br />
be released in the groundwater to concentrations<br />
exceeding the background value for the operational<br />
period of 2 years. The application of SIR at that site had<br />
therefore not threatened the long-term operation of<br />
this groundwater production well.<br />
The future of small-scale SIR/SAR<br />
This thesis has shown the potential of SIR as a household<br />
drinking water treatment solution, nevertheless a<br />
HWTS solution will only prove smart once it sustains<br />
in the users’ environment. This elementary observation<br />
comprises the real challenge for researchers who<br />
design technologies for developing countries. Some<br />
technologies might not yet be ready for implementation<br />
in the near future, but can form a base for upcoming<br />
trends. (Drinking water) engineers <strong>and</strong> researchers
9 Concluding remarks<br />
have the tendency to focus on the technology, rather<br />
than on the social boundary conditions. In addition,<br />
enthusiasm about a new treatment system can be the<br />
driving force to implement the technology fast. In the<br />
process implementers can then forget the need for<br />
objective scientific research. Only when the advantages<br />
<strong>and</strong>, perhaps more importantly, the limitations of<br />
a technology are identified is it safe to scale-up the<br />
implementation. The research presented in this<br />
thesis can be seen as an example where a promising<br />
technology, according to the involved universities <strong>and</strong><br />
national partners in Bangladesh, has been critically<br />
evaluated on its technical feasibility. The role of<br />
drinking water researchers can be vital in this, because<br />
during development of new technologies researchers<br />
must (i) remain critical of their own work, despite<br />
enthusiasm for the technology under investigation, <strong>and</strong><br />
(ii) show flexibility <strong>and</strong> patience to let their findings<br />
be slowly adopted <strong>and</strong>, ultimately, be improved by<br />
the users. Currently rural Bangladesh seems to be<br />
a playground for researchers to test their ‘products’,<br />
<strong>and</strong> communication with local authorities is minimal.<br />
Even though this seems to be the fast way forward<br />
scientifically, working together will eventually give the<br />
best long-term solutions. Only by joining modern <strong>and</strong><br />
endogenous knowledge true sustainable solutions will<br />
develop.<br />
In this perspective the main conclusion of<br />
this thesis should be re-formulated, as only longterm<br />
application of small-scale SIR will prove its<br />
sustainability in Bangladesh. Nevertheless, the results<br />
in this thesis have shown that SIR is technically feasible<br />
in a wide range of (geochemical) settings <strong>and</strong> holds<br />
therefore great promise for decentralized application.<br />
<strong>Subsurface</strong> arsenic removal did not show the same<br />
promise at household level. The performance of SAR at<br />
small-scale may be enhanced by enriching the injection<br />
water with higher O 2<br />
concentrations, e.g., with pure O 2<br />
,<br />
though this may be practically infeasible, for logistic,<br />
safety <strong>and</strong> economic reasons. It may also be, however,<br />
all a matter of scale. At larger injection volumes the<br />
oxidation zone is exp<strong>and</strong>ed <strong>and</strong> higher retention times<br />
of arsenic-contaminated groundwater in this zone can<br />
be reached. In favourable groundwater composition<br />
conditions (low competing anions concentrations, high<br />
pH), the subsurface removal of arsenic may still be<br />
feasible at community <strong>and</strong>/or municipality level. Such<br />
medium- or large-scale applications do not benefit<br />
from the existing h<strong>and</strong> pump infrastructure, but may<br />
be a safe arsenic mitigation solution in more densely<br />
populated areas, as frequently found in Bangladesh.<br />
In rural settings the scale issue may be overcome by<br />
combining SAR with <strong>Subsurface</strong> Rainwater Storage,<br />
where monsoon rain is utilized as injection water.<br />
Rainwater harvesting has been promoted as a safe<br />
drinking water alternative, but above-ground storage<br />
adds the risk of microbial contamination. <strong>Subsurface</strong><br />
storage could overcome this problem, but research is<br />
recommended to investigate the effect of rain water<br />
(low in pH <strong>and</strong> alkalinity) on the mobility of arsenic.<br />
131<br />
9
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
9<br />
132<br />
It is also recommended to investigate the relationship<br />
between the size of the oxidation zone (contact time)<br />
<strong>and</strong> the removal of arsenic, which could be achieved<br />
by monitoring larger injection-abstraction columns,<br />
in combination with solute transport modelling. In<br />
addition, the threshold concentrations of competing<br />
anions for which SAR could still be effectively<br />
operated are yet to be determined, which can be done<br />
by exp<strong>and</strong>ing the column studies as presented in this<br />
thesis.<br />
<strong>Iron</strong> contamination of groundwater is not a<br />
health hazard like arsenic, but the iron removal benefits<br />
the users as well. Not just for aesthetic improvement<br />
of drinking water quality, but for all water-consuming<br />
activities (cooking, washing, laundry, etc.). <strong>Arsenic</strong><br />
removal is primarily needed for drinking (<strong>and</strong> cooking)<br />
water, making the application of SIR in combination<br />
with above-ground arsenic removal very viable. SIR may<br />
thus be very useful for families in rural areas, just not as<br />
an arsenic mitigation solution. Subsequent (household)<br />
water treatment is needed in order to remove arsenic<br />
from the groundwater, which may be even more<br />
effective with subsurface treated water because: (i) iron<br />
is removed, reducing the clogging of the filters; <strong>and</strong><br />
(ii) phosphate concentrations are reduced, leaving the<br />
available adsorption sites on the media available for<br />
arsenic adsorption. Some HWTS, however, rely on the<br />
presence of Fe 2+ , either for coagulation or regeneration<br />
of the adsorptive surfaces, making those systems less<br />
promising for combination with SIR.<br />
<strong>Subsurface</strong> manganese removal was not within the<br />
scope of this thesis, but over the past years elevated<br />
manganese concentrations have been reported<br />
in Bangladeshi groundwater. Chronic exposure<br />
to manganese through drinking water may have<br />
neurological effects above the WHO guideline of 0.4<br />
mg.L -1 (WHO, 2006). <strong>Subsurface</strong> manganese removal<br />
has been successfully operated in several European<br />
countries (Hallberg <strong>and</strong> Martinell, 1976; van Beek 1983)<br />
<strong>and</strong>, more recently, in Egypt (Olsthoorn, 2000). It may<br />
also be feasible in decentralized, rural settings, though<br />
previous experiments have indicated its sensitivity to<br />
operational fluctuations (van Beek, 1983), making it<br />
less attractive for h<strong>and</strong> pump application.<br />
References<br />
Appelo C. A. J., B. Drijver, R. Hekkenberg <strong>and</strong> M. de Jonge<br />
(1999) Modeling in situ iron removal from ground water,<br />
Ground Water 37(6): 811-817.<br />
Dixit S. <strong>and</strong> J. G. Hering (2006) Sorption of Fe(II) <strong>and</strong> As(III)<br />
on goethite in single- <strong>and</strong> dual-sorbate systems, Chemical<br />
Geology, 228(1-3): 6-15.<br />
Hallberg R. O. <strong>and</strong> R. Martinell (1976) Vyredox - in situ<br />
purification of groundwater, Ground Water, 14(2): 88-93.<br />
Olsthoorn, T.N. (2000) Background of subsurface iron <strong>and</strong><br />
manganese removal. Amsterdam Water Supply Research <strong>and</strong><br />
Development, Hydrology Department.
9 Concluding remarks<br />
Sharma S.K. (2001) Adsorptive iron removal from groundwater.<br />
PhD dissertation, IHE Delft.<br />
Smedley P.L <strong>and</strong> D.G. Kinniburgh (2002) A review of the source,<br />
behaviour <strong>and</strong> distribution of arsenic in natural waters, Appl<br />
Geochem 17: 517–568.<br />
Tamura H., S. Kawamura <strong>and</strong> M. Hagayama (1980) Acceleration<br />
of the oxidation of Fe 2+ ions by Fe(III)-oxyhydroxides,<br />
Corrosion Science 20, 963-971.<br />
van Beek C.G.E.M. (1983) Ondergrondse ontijzering, een<br />
evaluatie van uitgevoerd onderzoek (in Dutch). KIWA<br />
mededeling 78.<br />
World Health Organization (2006) Guidelines for drinkingwater<br />
quality, First Addendum to Third Edition, Volume 1<br />
Recommendations, Geneva.<br />
9<br />
133
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
9<br />
134
Summary<br />
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
- Doris van Halem<br />
Summary<br />
<strong>Arsenic</strong> contamination of shallow tube well drinking<br />
water in Bangladesh is an urgent developmental <strong>and</strong><br />
health problem. This problem disproportionately<br />
affects the rural poor, i.e., those most reliant on this<br />
source of drinking water. Current arsenic mitigation<br />
solutions, including (household) arsenic removal<br />
options, do not always provide a sustainable alternative<br />
for safe drinking water in rural Bangladesh. A novel<br />
household water treatment <strong>and</strong> storage (HWTS)<br />
technology, <strong>Subsurface</strong> <strong>Arsenic</strong> <strong>Removal</strong> (SAR),<br />
relies on the existing technology of <strong>Subsurface</strong> <strong>Iron</strong><br />
<strong>Removal</strong> (SIR), or in-situ iron removal. The principle<br />
of SIR is that aerated water is periodically injected<br />
into an anoxic or anaerobic aquifer through a tube<br />
well, partially displacing the original Fe 2+ -containing<br />
groundwater. The O 2<br />
-rich injection water oxidizes the<br />
adsorbed Fe 2+ in the subsurface environment around<br />
the tube well. The flow direction is reversed during<br />
groundwater abstraction, resulting in adsorption<br />
of dissolved Fe 2+ (<strong>and</strong> arsenic) on the oxidized Fe 3+<br />
surface. Subsequently groundwater with reduced iron<br />
<strong>and</strong> arsenic concentrations can be abstracted. However,<br />
for SIR/SAR to be considered a viable arsenic mitigation<br />
tool, the following problem needs to be addressed:<br />
Problem description<br />
<strong>Subsurface</strong> <strong>Iron</strong> <strong>and</strong> <strong>Arsenic</strong> <strong>Removal</strong> is a<br />
promising safe water solution, however, there is the<br />
need for better underst<strong>and</strong>ing of the (subsurface)<br />
processes determining the sustainable operation<br />
in diverse geochemical settings.<br />
The research methodology consisted of four<br />
components: (1) data analysis of existing SIR plants<br />
in the Netherl<strong>and</strong>s, (2) monitoring of a test facility in<br />
135<br />
S
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
S<br />
136<br />
Bangladesh, (3) s<strong>and</strong> column studies in laboratory <strong>and</strong><br />
field, <strong>and</strong> (4) sediment characterization nearby SIR<br />
wells. The methodology allowed to address the main<br />
problem definition by focusing on the following:<br />
• past experiences in the Netherl<strong>and</strong>s (chapter 2);<br />
• small-scale applicability of SIR/SAR (chapter 3);<br />
• adsorptive-catalytic oxidation of Fe 2+ (chapters 4 <strong>and</strong> 7);<br />
• co-removal of As(III) (chapters 3,4, <strong>and</strong> 7);<br />
• cation exchange during injection (chapter 5);<br />
• clogging of the aquifer (chapter 6);<br />
• influence of groundwater composition (Chapter 8).<br />
The field <strong>and</strong> laboratory data have gained insight in the<br />
occurring (geo)chemical processes under the shifting<br />
redox conditions during injection-abstraction cycles,<br />
resulting in two main conclusions.<br />
Conclusion #1<br />
The results in this thesis have shown that<br />
<strong>Subsurface</strong> <strong>Iron</strong> <strong>Removal</strong> is technically feasible<br />
in a wide range of (geochemical) settings <strong>and</strong><br />
has therefore great potential for h<strong>and</strong> pump<br />
application.<br />
The experiences in the Netherl<strong>and</strong>s have shown that<br />
subsurface iron removal is an effective, robust <strong>and</strong><br />
sustainable iron removal technology with great potential<br />
for worldwide application. pH showed to be the most<br />
pronounced water quality parameter determining the<br />
efficacy of SIR, as the operational V/Vi ratio more<br />
than doubled between pH 7.0 <strong>and</strong> 7.4 at WTP Corle.<br />
Apart from iron, it was measured that also manganese,<br />
phosphate <strong>and</strong> arsenic were retained in the aquifer. In<br />
general, successive cycles resulted in improved removal<br />
efficacies, illustrating the technology’s potential for safe<br />
<strong>and</strong> long-term operation. Additionally, when SIR was<br />
stopped for several years, no peak concentrations of the<br />
studied constituents were observed in the abstracted<br />
groundwater.<br />
For h<strong>and</strong> pump application of SIR, the injection<br />
volumes, as operated in the Netherl<strong>and</strong>s were reduced<br />
from over 1,000m 3 to below 1m 3 . The field study<br />
in Manikganj, Bangladesh, showed that SIR could<br />
be successfully operated at small-scale. The system<br />
gained in efficiency after multiple injection-abstraction<br />
cycles, <strong>and</strong> after injection of water with higher O 2<br />
concentrations.<br />
The injection-abstraction cycles were simulated<br />
in s<strong>and</strong> column experiments, where the observed<br />
(stoichiometric) 1:4 ratio for O 2<br />
consumption <strong>and</strong> Fe 2+<br />
removal confirmed the occurrence of the adsorptivecatalytic<br />
oxidation mechanism. Also, O 2<br />
consumption<br />
<strong>and</strong> subsequent Fe 2+ removal were enhanced in the<br />
columns by accumulated Fe 3+ deposits. Nevertheless,<br />
the adsorptive-catalytic oxidation mechanism could<br />
not account for the efficacy increase after multiple<br />
cycles. The O 2<br />
consumption by secondary processes<br />
during initial cycles (e.g., pyrite oxidation) could<br />
provide an explanation, but it is more likely that the<br />
spatial (volume) increase of the subsurface oxidation<br />
zone enhances the Fe 2+ immobilization per cycle.
Summary<br />
Apart from adsorptive-catalytic oxidation, also cation<br />
exchange has been proposed to occur during injectionabstraction<br />
cycles. In s<strong>and</strong> column experiments with<br />
synthetic <strong>and</strong> natural groundwater it was found that<br />
cation exchange (Na + -Fe 2+ ) occurs during subsurface<br />
iron removal. The Fe 2+ exchange increased at higher<br />
Na + concentration in the injection water, but decreased<br />
in the presence of other cations in the groundwater.<br />
Field results at WTP Corle with injection of high O 2<br />
(0.55 mmol.L -1 ) concentrations showed better removal<br />
than the normal 0.28 mmol.L -1 O 2<br />
, indicating that not<br />
the exchangeable Fe 2+ on the soil material is the limiting<br />
factor during injection, but it is the supply of O 2<br />
to the<br />
available Fe 2+ . From a groundwater quality perspective,<br />
it was found in the column studies that SIR was not<br />
significantly limited or enhanced by other inorganic<br />
groundwater constituents, though iron retardation was<br />
lower in the presence of 1.2 mM calcium or 0.06 mM<br />
manganese.<br />
One of the concerns for sustainable application<br />
of SIR is the risk of clogging of the aquifer by longterm<br />
operation of SIR. Therefore sediments nearby<br />
12-year-old SIR wells were characterized for their<br />
chemical composition. At the investigated site (WTP<br />
De Put), iron has accumulated at specific depths near<br />
the subsurface iron removal wells. Whether it was due<br />
to preferred flow paths or geochemical mineralogy<br />
conditions; subsurface iron removal clearly favoured<br />
certain soil layers. The majority of accumulated iron<br />
was characterized as crystalline, suggesting that<br />
precipitated amorphous iron oxides have transformed<br />
to iron oxides of higher crystallinity. These crystalline,<br />
compact iron oxides had not noticeably clogged the<br />
investigated well <strong>and</strong>/or aquifer between 1996 <strong>and</strong> 2008.<br />
The subsurface iron removal wells even needed less<br />
frequent rehabilitation, as drawdown increases slower<br />
than in normal production wells. Other groundwater<br />
constituents, such as manganese, arsenic <strong>and</strong> strontium<br />
were found to co-accumulate with iron.<br />
Conclusion #2<br />
<strong>Subsurface</strong> <strong>Arsenic</strong> <strong>Removal</strong> has been found to<br />
be less suitable for h<strong>and</strong> pump application <strong>and</strong><br />
vulnerable to diverse geochemical conditions.<br />
The h<strong>and</strong> pump field study in Bangladesh did not prove<br />
to be very effective for the removal of arsenic. <strong>Arsenic</strong><br />
(as As(III)) retardation was limited <strong>and</strong> breakthrough<br />
of 10 µg.L -1 (provisional guideline of the World<br />
Health Organization) was observed before V/Vi=1,<br />
which corresponds to the moment of groundwater<br />
arrival at the well. Unlike iron removal, subsurface<br />
arsenic removal did not increase after multiple cycles,<br />
illustrating that the process which is responsible for<br />
the effective iron removal did not promote an equally<br />
effective co-removal of As(III). No relation has been<br />
observed in this study between the amount of removed<br />
As(III) <strong>and</strong> the Fe 2+ :As(III) ratio of the groundwater.<br />
And because arsenic is observed in the abstracted<br />
groundwater amply before iron, the latter cannot be<br />
used as a visible indicator for arsenic presence, i.e., aid<br />
137<br />
S
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
S<br />
138<br />
in post-deployment monitoring.<br />
During injection-abstraction cycles in s<strong>and</strong> columns<br />
with a synthetic Fe 2+ -As(III)-O 2<br />
system it was possible<br />
to reach significantly better arsenic removal than at<br />
the field study in Bangladesh <strong>and</strong> in the columns with<br />
groundwater (WTP Lekkerkerk <strong>and</strong> WTP Loosdrecht).<br />
This can be explained by the competition with other<br />
anions in the natural groundwater. In the presence of<br />
0.01mM phosphate, 0.2mM silicate, or 1mM nitrate,<br />
arsenic removal was seriously inhibited, illustrating<br />
the vulnerability of this SAR technology in diverse<br />
geochemical settings.<br />
This vital observation, however, applies not<br />
just for SAR but for the majority of (arsenic) HWTS<br />
solutions. In general, no household water treatment<br />
option has been designed to suit all conditions,<br />
illustrating the urgent need to investigate the (technical)<br />
boundary conditions in which a HWTS solution can<br />
succeed. In favourable groundwater composition<br />
conditions (low competing anions concentrations,<br />
high pH), the subsurface removal of arsenic may still<br />
be feasible. However, the decentralized character of<br />
HWTS does not promote extensive knowledge on the<br />
groundwater composition at h<strong>and</strong> pump level, making<br />
community- <strong>and</strong>/or municipality scale facilities more<br />
attractive.<br />
A newly recognized groundwater quality<br />
concern in Bangladesh is contamination of groundwater<br />
by elevated manganese concentrations. Chronic<br />
exposure to manganese through drinking water may<br />
have neurological effects above the WHO guideline of<br />
0.4 mg.L -1 . The (small-scale) application of subsurface<br />
manganese removal would be an interesting approach<br />
to investigate. <strong>Iron</strong> contamination of drinking water<br />
is not a health hazard, but the iron removal benefits<br />
the users as well. Not just for aesthetic improvement<br />
of drinking water quality, but for all water-consuming<br />
activities (cooking, washing, laundry, etc.). <strong>Arsenic</strong><br />
removal is primarily needed for drinking (<strong>and</strong> cooking)<br />
water, making the application of SIR in combination<br />
with subsequent (household) arsenic removal very<br />
viable. <strong>Arsenic</strong> filters may even perform better as iron<br />
<strong>and</strong> phosphate levels are reduced by SIR, reducing both<br />
clogging <strong>and</strong> adsorptive competition.<br />
A HWTS solution will only prove smart<br />
once it sustains in the users’ environment. This<br />
elementary observation comprises the real<br />
challenge for researchers who study technologies<br />
for developing countries. Some technologies might<br />
not yet be ready for implementation in the near<br />
future, but can form a base for upcoming trends.<br />
(Drinking water) engineers <strong>and</strong> researchers have<br />
the tendency to focus on the technology, rather than<br />
on the social boundary conditions. In addition,<br />
enthusiasm about a new treatment system can be<br />
the driving force to implement the technology fast.<br />
In the process, implementers can then forget the<br />
need for objective scientific research. Only when<br />
the advantages <strong>and</strong>, perhaps more importantly,<br />
the limitations of a technology are identified is it<br />
safe to scale-up the implementation.
Samenvatting<br />
Ondergrondse ijzer-en arseenverwijdering voor drinkwaterzuivering in Bangladesh<br />
- Doris van Halem<br />
Samenvatting<br />
Arseenverontreiniging van ondiep grondwater is<br />
een groot ontwikkelings- en gezondheidsprobleem<br />
in Bangladesh. Dit probleem raakt de armen op<br />
het plattel<strong>and</strong> onevenredig, omdat zij het meest<br />
afhankelijk zijn van deze bron van drinkwater. Huidige<br />
oplossingen voor het arseenprobleem, waaronder<br />
arseenverwijdering op huishoudschaal, vormen niet<br />
altijd een duurzaam alternatief voor veilig drinkwater<br />
op het plattel<strong>and</strong> van Bangladesh. Een nieuwe household<br />
water treatment <strong>and</strong> storage (HWTS) technologie<br />
combineert ondergrondse arseenverwijdering (SAR)<br />
met de besta<strong>and</strong>e technologie van ondergrondse<br />
ijzerverwijdering (SIR). Het principe van SIR is dat<br />
belucht water periodiek wordt geïnjecteerd in een<br />
zuurstofloze of anaërobe watervoerende laag, waardoor<br />
het originele Fe 2+ -bevattende grondwater gedeeltelijk<br />
verplaatst wordt. Het O 2<br />
-rijke injectie water oxideert<br />
het geadsorbeerde Fe 2+ in de ondergrond rondom de<br />
put. Tijdens grondwaterontrekking keert de stroming<br />
om waardoor opgelost Fe 2+ (en arseen) kan adsorberen<br />
aan het geoxideerde Fe 3+ oppervlak. Hierdoor kan<br />
vervolgens grondwater met verlaagde ijzer en arseen<br />
concentraties onttrokken worden. Voordat SIR/SAR<br />
beschouwd kan worden als een haalbaar hulpmiddel ter<br />
bestrijding van het arseenprobleem, moet het volgende<br />
probleem opgelost worden:<br />
Probleembeschrijving<br />
Ondergrondse ijzer- en arseenverwijdering is een<br />
veelbelovende oplossing voor veilig drinkwater,<br />
maar er is de noodzaak voor beter inzicht in<br />
de (ondergrondse) processen die de duurzame<br />
werking onder verschillende geochemische<br />
condities bepalen.<br />
139<br />
S
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
S<br />
140<br />
De onderzoeksmethodiek bestond uit vier<br />
componenten: (1) data analyse van besta<strong>and</strong>e SIR<br />
zuiveringen in Nederl<strong>and</strong>, (2) het monitoren van<br />
een proefopstelling in Bangladesh, (3) z<strong>and</strong>kolom<br />
studies in het veld en laboratorium, en (4) sediment<br />
karakterisering nabij SIR putten. Met hulp van deze<br />
methodes is er gericht onderzoek gedaan naar:<br />
• ervaringen uit het verleden in Nederl<strong>and</strong> (hoofdstuk 2);<br />
• kleinschalige toepassing van SIR/SAR (hoofdstuk 3);<br />
• adsorptieve-katalytische oxidatie van Fe 2+ (hoofdstukken<br />
4 en 7);<br />
• co-verwijdering van As(III) (hoofdstukken 3,4 en 7);<br />
• kationuitwisseling tijdens injectie (hoofdstuk 5);<br />
• verstopping van het watervoerend pakket (hoofdstuk 6);<br />
• invloed van de grondwatersamenstelling (hoofdstuk 8).<br />
De veld- en laboratoriumresultaten hebben inzicht<br />
gegeven in de (geo)chemische processen die<br />
plaatsvinden onder de ver<strong>and</strong>erende redox condities<br />
tijdens de injectie-onttrekkingscycli, resulterend in<br />
twee hoofdconclusies.<br />
Conclusie #1<br />
De resultaten in dit proefschrift hebben<br />
aangetoond dat ondergrondse ijzerverwijdering<br />
technisch haalbaar is in een breed scala van<br />
(geochemische) condities en heeft daarom veel<br />
potentie voor de toepassing op h<strong>and</strong>pomp schaal.<br />
De ervaringen in Nederl<strong>and</strong> hebben aangetoond<br />
dat ondergrondse ijzerverwijdering een effectieve,<br />
robuuste en duurzame technologie is met een groot<br />
potentieel voor wereldwijde toepassing. pH bleek de<br />
meest uitgesproken waterkwaliteitsparameter voor het<br />
bepalen van de effectiviteit van SIR. De operationele V/<br />
Vi-verhouding verdubbelde ruimschoots tussen pH 7,0<br />
en 7,4 op PS Corle. Naast ijzer, werden ook mangaan,<br />
fosfaat en arseen vastgelegd in het watervoerende<br />
pakket. Opeenvolgende cycli resulteerden in een<br />
verbeterde verwijderingsefficiëntie, wat het potentieel<br />
van deze technologie voor veilig en langdurig gebruik<br />
illustreert. Nadat SIR werd gestopt voor een aantal jaren,<br />
zijn er bovendien in die periode geen piekconcentraties<br />
van de onderzochte best<strong>and</strong>delen waargenomen in het<br />
onttrokken grondwater.<br />
Voor h<strong>and</strong>pomp toepassing van SIR werden<br />
de injectievolumes teruggebracht van meer dan 1.000<br />
m 3 , zoals gebruikelijk in Nederl<strong>and</strong>, tot onder de 1m 3 .<br />
Het veldonderzoek in Manikganj, Bangladesh, toonde<br />
aan dat SIR met succes zou kunnen worden toegepast<br />
op kleine schaal. De systeemefficiëntie had profijt van<br />
opeenvolgende injectie-ontrekkingscycli en van injectie<br />
met hogere O 2<br />
concentraties.<br />
De injectie-ontrekkingscycli zijn gesimuleerd<br />
in z<strong>and</strong>kolom experimenten, waarbij de waargenomen<br />
(stoichiometrische) 1:4 verhouding voor O 2<br />
verbruik en Fe 2+ verwijdering het optreden van<br />
het adsorptieve-katalytische oxidatie mechanisme<br />
bevestigde. Aanvullend werden het O 2<br />
verbruik en<br />
de daaropvolgende Fe 2+ verwijdering versterkt in de<br />
kolommen door aanwezigheid van geaccumuleerde
Samenvatting<br />
Fe 3+ oxiden. Niettemin is het adsorptieve-katalytische<br />
oxidatie mechanisme niet verantwoordelijk voor de<br />
verbeterde efficiëntie na meerdere cycli. Het O 2<br />
verbruik<br />
van secundaire processen tijdens de eerste cycli<br />
(bijvoorbeeld pyriet oxidatie), kan wel een verklaring<br />
zijn, maar het is waarschijnlijker dat de ruimtelijke<br />
(volume) vergroting van de ondergrondse oxidatie<br />
zone de Fe 2+ immobilisatie per cyclus verhoogt.<br />
Naast adsorptieve-katalytische oxidatie, is de<br />
hypothese dat ook kationenuitwisseling een rol speelt<br />
tijdens injectie-ontrekkingscycli. In de z<strong>and</strong>kolom<br />
experimenten met synthetisch en natuurlijk<br />
grondwater bleek dat kationuitwisseling (Na + -Fe 2+ )<br />
optreedt tijdens ondergrondse ontijzering. De Fe 2+<br />
uitwisseling verhoogde bij hogere Na + -concentraties<br />
in het geïnjecteerde water, maar daalde in de<br />
aanwezigheid van <strong>and</strong>ere kationen in het grondwater.<br />
Veldresultaten op PS Corle met het injecteren van<br />
water van hoge O 2<br />
(0,55 mmol/l) concentraties toonde<br />
een betere verwijdering dan de normale 0,28 mmol/l<br />
O 2,<br />
wat aangeeft dat niet de uitwisselbare Fe 2+ op het<br />
bodemmateriaal de beperkende factor is tijdens injectie,<br />
maar de aanvoer van O 2<br />
aan de beschikbare Fe 2+ . Met<br />
betrekking tot het effect van de grondwaterkwaliteit,<br />
bleek in de kolomstudies dat SIR niet significant<br />
beperkt of versterkt wordt door de aanwezigheid van<br />
<strong>and</strong>ere anorganische grondwater best<strong>and</strong>delen, hoewel<br />
de ijzer retardatie lager was in de aanwezigheid van 1,2<br />
mmol/l calcium of 0,06 mmol/l mangaan.<br />
Een a<strong>and</strong>achtspunt voor de duurzame toepassing<br />
van SIR is het risico van verstopping van het<br />
watervoerend pakket door langdurig gebruik van SIR.<br />
Sedimenten nabij 12-jarige SIR putten zijn daarom<br />
geanalyseerd op hun chemische samenstelling.<br />
Op de onderzochte locatie (PS De Put), is ijzer<br />
geaccumuleerd op specifieke dieptes in de buurt van<br />
de ondergrondse ontijzeringsputten. Veroorzaakt<br />
door voorkeursstromingen of geochemische condities;<br />
ondergrondse ontijzering had duidelijk een voorkeur<br />
voor bepaald bodemlagen. De meerderheid van het<br />
geaccumuleerde ijzer is gekarakteriseerd als kristallijn<br />
dat suggereert dat neergeslagen amorfe ijzeroxiden<br />
zijn getransformeerd naar ijzeroxiden van hogere<br />
kristalliniteit. Deze kristallijne en compacte ijzeroxiden<br />
hebben de onderzochte put en/of het watervoerend<br />
pakket niet merkbaar verstopt tussen 1996 en 2008.<br />
De ondergrondse ontijzeringsputten hoefden zelfs<br />
minder vaak gereinigd te worden, vanwege het verschil<br />
in peilhoogte dat langzamer toenam dan in de normale<br />
onttrekkingsputten. Uit metingen is ook gebleken<br />
dat <strong>and</strong>ere grondwaterbest<strong>and</strong>delen, zoals mangaan,<br />
arseen en strontium accumuleerden samen met het<br />
ijzer.<br />
Conclusie #2<br />
Ondergrondse arseenverwijdering is minder<br />
geschikt bevonden voor h<strong>and</strong>pomp toepassing en<br />
is kwetsbaar in diverse geochemische condities.<br />
S<br />
141
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
S<br />
142<br />
Uit het h<strong>and</strong>pomp veldonderzoek in Bangladesh is<br />
SAR niet effectief gebleken voor de verwijdering van<br />
arseen. Arseen (als As(III)) retardatie was beperkt en<br />
de doorbraak van 10 μg/l (voorlopige richtlijn van de<br />
Wereldgezondheidsorganisatie) werd waargenomen<br />
nog voor V/Vi =1, wat overeenkomt met het moment<br />
van grondwater aankomst in de put. In tegenstelling<br />
tot de verwijdering van ijzer, nam ondergrondse<br />
arseenverwijdering niet toe na meerdere cycli. Dit<br />
illustreert dat het proces dat verantwoordelijk is voor<br />
de effectieve verwijdering van ijzer niet een evenredige<br />
efficiëntie gar<strong>and</strong>eert voor As(III). In deze studie<br />
is geen relatie waargenomen tussen de hoeveelheid<br />
verwijderde As(III) en de Fe 2+ :As(III) verhouding van<br />
het grondwater. Bovendien kan ijzer niet gebruikt<br />
worden als zichtbare indicator voor de aanwezigheid<br />
van arseen, aangezien arseen waargenomen wordt<br />
in het grondwater voordat opgelost ijzer bij de put<br />
arriveert.<br />
Tijdens de injectie-ontrekkingscycli in de<br />
z<strong>and</strong>kolommen met een synthetisch Fe 2+ -As(III)-<br />
O 2<br />
-systeem was het mogelijk om significant<br />
betere arseenverwijdering te bereiken dan bij het<br />
veldonderzoek in Bangladesh en bij de kolommen<br />
met grondwater (PS Lekkerkerk en PS Loosdrecht).<br />
Dit kan verklaard worden door de competitie met<br />
<strong>and</strong>ere anionen in het natuurlijke grondwater. In de<br />
aanwezigheid van 0,01mmol/l fosfaat, 0,2 mmol/l<br />
silicaat of 1 mmol/l nitraat, was de arseenverwijdering<br />
erg beperkt, welke de kwetsbaarheid van de SAR<br />
technologie in verschillende geochemische condities<br />
aantoont.<br />
Deze essentiële waarneming geldt echter<br />
niet alleen voor SAR, maar voor de meerderheid<br />
van (arseen) HWTS oplossingen. Geen enkele<br />
huishoudelijke waterbeh<strong>and</strong>elingsoptie is ontworpen<br />
om aan alle mogelijke voorwaarden te voldoen en er<br />
is dus een dringende noodzaak om de (technische)<br />
r<strong>and</strong>voorwaarden waarin een HWTS oplossing<br />
kan slagen te onderzoeken. Met een gunstige<br />
grondwatersamenstelling (lage concentraties<br />
van concurrerende anionen, hoge pH), kan de<br />
ondergrondse arseenverwijdering nog steeds haalbaar<br />
zijn. Echter, het decentrale karakter van HWTS is niet<br />
bevorderlijk voor de aanwezigheid van voldoende<br />
kennis van de grondwatersamenstelling ter plaatse,<br />
waardoor zuivering op community- of gemeentelijke<br />
schaal aantrekkelijker is.<br />
Een nieuw waterkwaliteitsa<strong>and</strong>achtspunt in<br />
Bangladesh is de verontreiniging van grondwater<br />
met verhoogde mangaan concentraties. Chronische<br />
blootstelling aan mangaan via drinkwater kan<br />
neurologische effecten hebben boven de richtlijn<br />
van 0,4 mg/l (Wereldgezondheidsorganisatie).<br />
De (kleinschalige) toepassing van ondergrondse<br />
mangaanverwijdering zou daarom een interestante<br />
onderzoeksrichting zijn. IJzerverontreiniging van het<br />
drinkwater vormt geen gevaar voor de gezondheid, maar<br />
de gebruikers hebben wel profijt van ijzerverwijdering.<br />
Niet alleen voor esthetische verbetering van de
Samenvatting<br />
drinkwaterkwaliteit, maar voor alle water verbruikende<br />
activiteiten (koken, wassen, etc.). Verwijdering van<br />
arseen is alleen nodig voor drink- en kookwater, maar<br />
huidige (huishoud) arseenfilters worden beperkt in<br />
hun efficiëntie door de aanwezigheid van ijzer en<br />
fosfaat in het grondwater (verstopping en adsorptieve<br />
competitie). De combinatie van SIR gevolgd door een<br />
arseenfilter kan deze beperking wegnemen, aangezien<br />
tijdens SIR zowel ijzer als fosfaat concentraties verlaagd<br />
worden.<br />
Een HWTS oplossing is pas een ‘smart<br />
solution’ als het duurzaam st<strong>and</strong>houdt in<br />
de gebruikersomgeving. Deze elementaire<br />
observatie omvat de echte uitdaging voor<br />
wetenschappers die onderzoek doen naar<br />
technologieën voor ontwikkelingsl<strong>and</strong>en.<br />
Sommige technologieën zijn misschien nog niet<br />
klaar voor implementatie in de nabije toekomst,<br />
maar kunnen een basis vormen voor toekomstige<br />
trends. (Drinkwater) ingenieurs en onderzoekers<br />
hebben de neiging om zich te concentreren op<br />
de technologische mogelijkheden, in plaats van<br />
op de maatschappelijke r<strong>and</strong>voorwaarden.<br />
Daarnaast kan enthousiasme over een nieuw<br />
zuiveringssysteem de drijvende kracht worden<br />
om de technologie te snel te implementeren. In<br />
het proces van realisatie kunnen betrokkenen<br />
de noodzaak van objectief wetenschappelijk<br />
onderzoek vergeten. Alleen wanneer de voordelen<br />
en, misschien nog belangrijker, de beperkingen<br />
van een technologie zijn geïdentificeerd is het<br />
veilig om de implementatie op te schalen.<br />
S<br />
143
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
S<br />
144
List of publications<br />
List of publications<br />
International refereed journals<br />
van Halem D., W. de Vet, J. Verberk, G. Amy, H. van Dijk<br />
(2011) Characterization of accumulated precipitates during<br />
subsurface iron removal, Applied Geochemistry 26: 116-124.<br />
van Halem D., S. Olivero, W.W.J.M. de Vet, J.Q.J.C. Verberk, G.L.<br />
Amy <strong>and</strong> J.C. van Dijk (2010) <strong>Subsurface</strong> iron <strong>and</strong> arsenic<br />
removal for shallow tube well drinking water supply in rural<br />
Bangladesh, Water Research 44: 5761-5769.<br />
van Halem D., R. Johnston, I.M. Huq, S.K. Ghosh, J.Q.J.C.<br />
Verberk, J.C. van Dijk <strong>and</strong> G.L. Amy (2010) <strong>Subsurface</strong> iron<br />
<strong>and</strong> arsenic removal: low-cost technology for communitybased<br />
drinking water supply in Bangladesh. Water Science<br />
<strong>and</strong> Technology: Water Supply 62(11): 2702-2709.<br />
van Halem D., H. van der Laan, S.G.J. Heijman, J.C. van Dijk<br />
<strong>and</strong> G.L. Amy (2009) Assessing the sustainability of the<br />
silver-impregnated ceramic pot filter for low-cost household<br />
drinking water treatment, Physics <strong>and</strong> Chemistry of the<br />
Earth 34: 36-42.<br />
van Halem D., S.G.J. Heijman, G.L. Amy, <strong>and</strong> J.C. van Dijk<br />
(2009) <strong>Subsurface</strong> arsenic removal for small-scale application<br />
in developing countries, Desalination 248(1-2): 241-248.<br />
van Halem D., S.A. Bakker, G.L. Amy, <strong>and</strong> J.C. van Dijk (2009)<br />
<strong>Arsenic</strong> in drinking water: a worldwide water quality concern<br />
for water supply companies, Drinking Water Engineering <strong>and</strong><br />
Science 2: 29-34.<br />
van Halem D., S.G.J. Heijman, J.C. van Dijk <strong>and</strong> G.L. Amy<br />
(2007) Ceramic silver-impregnated pot filters for household<br />
drinking water treatment in developing countries: material<br />
characterization <strong>and</strong> performance study, Water Science &<br />
Technology: Water Supply 7(5-6): 9-17.<br />
Publications in preparation<br />
van Halem D., D.H. Moed, J.Q.J.C. Verberk, G.L. Amy <strong>and</strong> J.C.<br />
van Dijk (2011) Cation exchange during subsurface iron<br />
145<br />
P
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
P<br />
removal, Water Research: under review.<br />
van Halem D. D.H. Moed, J.Q.J.C. Verberk, G.L. Amy <strong>and</strong> J.C.<br />
van Dijk (2011) The influence of groundwater matrix on<br />
subsurface arsenic <strong>and</strong> iron removal, Science of the Total<br />
Environment: submitted.<br />
van Halem J.Q.J.C. Verberk, G.L. Amy <strong>and</strong> J.C. van Dijk (2011)<br />
Accumulated deposits near subsurface iron removal wells:<br />
do they catalyze the Fe 2+ removal process? Water Science <strong>and</strong><br />
Technology: submitted.<br />
van Halem D., W.W.J.M. de Vet, M. de Jonge, G.L. Amy <strong>and</strong> J.C.<br />
van Dijk (2011) <strong>Subsurface</strong> iron removal in the Netherl<strong>and</strong>s,<br />
Journal of American Water Works Association: to be<br />
submitted.<br />
Editor of proceedings<br />
van Halem D. <strong>and</strong> W. M. Savenije (2009). International Ceramic<br />
Pot Filter Conference: Proceedings, International Pot Filter<br />
Conference, Atlanta, WEF Disinfection.<br />
Conference proceedings<br />
van Halem D.H. Moed, J.Q.J.C. Verberk, G.L. Amy <strong>and</strong> J.C. van<br />
Dijk (2011) <strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking<br />
water production: influence of the multi-component<br />
groundwater matrix, IWA Groundwater, Belgrade.<br />
van Halem J.Q.J.C. Verberk, G.L. Amy <strong>and</strong> J.C. van Dijk (2011)<br />
Accumulated deposits near subsurface iron removal wells: do<br />
they catalyze the Fe 2+ removal process? IWA Groundwater,<br />
Belgrade.<br />
van Halem D., J.Q.J.C. Verberk <strong>and</strong> J.C. van Dijk (2011)<br />
Decentralised subsurface iron <strong>and</strong> arsenic removal for<br />
146<br />
drinking water production: field <strong>and</strong> laboratory investigations,<br />
Conference on Integrated Water Management, Perth.<br />
van Halem D., P. Smeets, et al. (2010) Natural Engineered<br />
Systems: a sustainable alternative for drinking water<br />
treatment. AEESP session: Water Sustainability, WEFTEC<br />
2010, New Orleans: 1-13.<br />
Li Z., D. van Halem, et al. (2010) Review of high arsenic<br />
groundwater in China, 4th International Conference on<br />
Bioinformatics <strong>and</strong> Biomedical Engineering, Chengdu,<br />
China.<br />
van Halem D., R. Johnston, et al. (2009) <strong>Subsurface</strong> iron <strong>and</strong><br />
arsenic removal: low-cost technology for community-based<br />
drinking water supply in Bangladesh, 1 st IWA Development<br />
Congress, Mexico City.<br />
van Halem D., H. van der Laan, et al. (2009) <strong>Subsurface</strong><br />
iron removal: modelling the process, BeNeLux Young<br />
Water Professionals, Eindhoven, International Water<br />
Association.<br />
Bloem S. C., D. van Halem, et al. (2009) Silver impregnated<br />
ceramic pot filter: Flow rate versus the removal efficiency<br />
of pathogens, WEF Disinfection: International Ceramic Pot<br />
Filter Conference, Atlanta, WEF.<br />
van Halem D., W. W. J. M. de Vet, et al. (2008) <strong>Subsurface</strong> iron<br />
removal for drinking water production: underst<strong>and</strong>ing the<br />
process <strong>and</strong> exploiting beneficial side effects, Water Quality<br />
Technology Conference, Cincinnati, American Water Works<br />
Association.<br />
van Halem D., S. G. J. Heijman, et al. (2008) <strong>Subsurface</strong> arsenic<br />
removal for small-scale application in developing countries,<br />
Water <strong>and</strong> Sanitation in International Development <strong>and</strong>
List of publications<br />
Disaster Relief, Edinburgh.<br />
van Halem D., S. G. J. Heijman, et al. (2008) Design considerations<br />
for small-scale subsurface arsenic removal in developing<br />
countries, <strong>Arsenic</strong> in the environment: <strong>Arsenic</strong> from nature<br />
to humans, Valencia.<br />
van Halem D., S. G. J. Heijman, et al. (2007) Low-cost ceramic<br />
membrane filtration for application in developing countries:<br />
the ceramic silver-impregnated pot filter, Water Quality<br />
Technology Conference, Charlotte, American Water Works<br />
Association.<br />
van Halem D., S. G. J. Heijman, et al. (2007). Ceramic silverimpregnated<br />
pot filters for household drinking water<br />
treatment in developing countries: material characterization<br />
<strong>and</strong> performance study, International Conference on Water<br />
Management <strong>and</strong> Technology Applications in Developing<br />
Countries, Kuala Lumpur, IWA Specialist Conference.<br />
National publications<br />
Bakker S.A., D. van Halem, J.C. van Dijk, <strong>and</strong> G.L. Amy (2008)<br />
Arseen in drinkwater: niet alleen een probleem in Bangladesh,<br />
H 2<br />
O: Tijdschrift voor watervoorziening en waterbeheer.<br />
van Halem D. (2008) Drinking water research for all,<br />
Vakantiecursus, Delft University of Technology, J.C. van Dijk<br />
<strong>and</strong> N. Krikke (eds), Delft.<br />
van Halem D., A.I.A. Soppe, J. Kroesbergen <strong>and</strong> H. van der<br />
Jagt (2006) Effectiviteit van de met zilver geïmpregneerde<br />
keramische potfilter, H 2<br />
O: Tijdschrift voor watervoorziening<br />
en waterbeheer.<br />
P<br />
147
<strong>Subsurface</strong> iron <strong>and</strong> arsenic removal for drinking water treatment in Bangladesh<br />
Biography<br />
Doris van Halem (Eindhoven, 1981) graduated in 2000<br />
from secondary school (Eindhoven, VWO/IGCSE).<br />
Her interest in water <strong>and</strong> development started during<br />
her studies in Civil Engineering at Delft University of<br />
Technology, where she was involved in water projects<br />
in Benin <strong>and</strong> Sri Lanka. For her MSc thesis she<br />
investigated the efficacy of low-cost ceramic filters,<br />
locally produced around the world. As a follow-up she<br />
organized, on behalf of Aqua for All, the 1 st international<br />
CPF conference at WEF Disinfection in Atlanta.<br />
After graduating cum laude in 2006 she started her<br />
PhD research under the guidance of Prof. Hans van<br />
Dijk (TU Delft) <strong>and</strong> Prof. Gary Amy (UNESCO-IHE).<br />
The research entitled “<strong>Subsurface</strong> <strong>Iron</strong> <strong>and</strong> <strong>Arsenic</strong><br />
<strong>Removal</strong> for drinking water treatment in Bangladesh”<br />
was presented at many (international) IWA, AWWA<br />
<strong>and</strong> WEF conferences. Her articles have been published<br />
in international refereed journals <strong>and</strong> she is active as a<br />
reviewer for several journals, including Water Science<br />
<strong>and</strong> Technology <strong>and</strong> Water Research.<br />
B<br />
148