Removal of Bromate and Perchlorate in Ozone/GAC Systems

AWWA
Research
Foundation
Removal of Bromate
and Perchlorate in
Conventional
Ozone/GAC Systems
Subject Area:
Water Treatment

Removal of Bromate
and Perchlorate in
Conventional____
Ozone/GAC Systems

The mission of the AWWA Research Foundation is to advance the science of water to improve
the quality of life. Funded primarily through annual subscription payments from over 1,000 utili
ties, consulting firms, and manufacturers in North America and abroad, AWWARF sponsors
research on all aspects of drinking water, including supply and resources, treatment, monitoring
and analysis, distribution, management, and health effects.
From its headquarters in Denver, Colorado, the AWWARF staff directs and supports the efforts
of over 500 volunteers, who are the heart of the research program. These volunteers, serving on
various boards and committees, use their expertise to select and monitor research studies to ben
efit the entire drinking water community.
Research findings are disseminated through a number of technology transfer activities, includ
ing research reports, conferences, videotape summaries, and periodicals.

Removal of Bromate
and Perchlonate in
Conventional____
Ozone/GAC Systems
Prepared by:
Mary Jo Kirisits
Jess C. Brown
Vernon L. Snoeyink
Lutgarde M. Raskin
Joanne C. Chee-Sanford
University of Illinois at Urbana-Champaign
3230 Newmark Civil Engineering Laboratory
205 North Mathews Avenue
Urbana, Illinois 61801
and
Sun Liang
Joon Min
The Metropolitan Water District of Southern California
700 North Moreno Avenue
La Verne, California 91750
Sponsored by:
AWWA Research Foundation
6666 West Quincy Avenue
Denver, CO 80235-3098
Published by the
AWWA Research Foundation and the
American Water Works Association

Disclaimer
This study was funded by the AWWA Research Foundation (AWWARF). AWWARF assumes no responsibility for the
content of the research study reported in this publication or for the opinions or statements of fact expressed in the report.
The mention of trade names for commercial products does not represent or imply the approval or endorsement of AWWARF.
This report is presented solely for informational purposes.
Library of Congress Cataloging-in-Publication Data has been applied for.
Copyright 2001
by
AWWA Research Foundation
and
American Water Works Association
Printed in the U.S.A.
ISBN 1-58321-104-7 Printed on recycled paper

CONTENTS
TABLES ................................................................ ix
FIGURES .............................................................. xi
FOREWORD ........................................................... xv
ACKNOWLEDGMENTS .......................;......................... xvii
EXECUTIVE SUMMARY ................................................. xix
CHAPTER 1. INTRODUCTION ............................................. 1
Background - Bromate ............................................... 1
Background - Perchlorate ............................................. 1
Previous Studies - Bromate ............................................ 3
Bromate Reduction with Various Influent Dissolved Oxygen
Concentrations .......................................... 5
Bromate Reduction to Bromide ................................... 8
Bromate Reduction with Various Influent Nitrate Concentrations .......... 9
Bromate Reduction with Various EBCTs ........................... 13
Verification of Biological Bromate Removal ......................... 14
Previous Studies - Perchlorate ......................................... 16
Objectives ........................................................ 17
CHAPTER 2. MATERIALS AND METHODS ................................. 19
Materials Used for Bromate Reduction Experiments ........................ 19
Water ...................................................... 19
Reagents .................................................... 20
Methods Utilized for Bromate Reduction Experiments ....................... 20
Biologically Active Carbon- Small Columns ......................... 20
Biologically Active Carbon - Large Filter ........................... 23
Backwashing the BAG Filters .................................... 25
Heterotrophic Plate Counts ..................................... 25
Denaturing Gradient Gel Electrophoresis ........................... 25
Plating Experiments ........................................... 26
Dilution-to-Extinction Experiments ............................... 28

Serum Bottle Experiments ...................................... 30
Materials Used for Perchlorate Reduction Experiments ....................... 30
Water ...................................................... 30
Reagents and Carbon .......................................... 30
Methods Utilized for Perchlorate Reduction Experiments ..................... 31
Biologically Active Carbon Filters ................................. 31
Abiotic Batch Experiments ...................................... 32
Batch Test Wash Experiments ................................... 32
Metal-Catalyzed GAC Filter Experiments ........................... 33
Metal-Catalyzed GAC Filter Wash Experiments ...................... 33
Ozone/Hydrogen Peroxide/Perchlorate Batch Tests ................... 33
Denaturing Gradient Gel Electrophoresis ........................... 34
Plating Experiments ........................................... 34
Dilution-to-Extinction Experiments ............................... 35
Serum Bottle Experiments ...................................... 35
Analytical Methods ................................................. 37
CHAPTER 3. RESULTS AND DISCUSSION .................................. 39
Bromate.......................................................... 39
Verification of Biological Bromate Removal in the BAG Filters .......... 39
Further Testing the Reproducibility of BAC Filters .................... 42
Backwashing BAC Filters ....................................... 42
Effect of the Initial Bromate Concentration .......................... 44
Effect of the Influent Dissolved Oxygen Concentration ................. 46
Effect of the Influent Nitrate Concentration ......................... 49
Effect of the Influent Sulfate Concentration ......................... 51
Effect ofpH ................................................. 52
Effect of Source Water Type .................................... 53
Perchlorate Reduction in CUW ..........................:........ 60
Microscopic Examination of Microorganisms in the Filters .............. 60
Denaturing Gradient Gel Electrophoresis Experiment .................. 62
Plating Experiments and Microbial Characterization ................... 64
vi

Dilution-to-Extinction Experiments ............................... 69
Serum Bottle Experiments ...................................... 72
Perchlorate - Abiotic Removal ......................................... 73
Calculation of Ion Exchange Capacity for Virgin Norit GAC ............ 73
Abiotic Batch Experiments ...................................... 74
Wash Tests .................................................. 77
Metal-Catalyzed GAC Experiments ............................... 77
Ozone/Hydrogen Peroxide/Perchlorate Batch Tests ................... 80
Perchlorate - Biological Reduction ..................................... 80
BAG Filtration Experiments ..................................... 80
Batch Biological Experiments .................................... 89
Microscopic Examination of Microorganisms in the Filters .............. 96
Denaturing Gradient Gel Electrophoresis Experiment .................. 98
CHAPTER 4. ACTIVITIES AT THE METROPOLITAN WATER DISTRICT
OF SOUTHERN CALIFORNIA ....................................... 99
Introduction ......................................................... 99
Objectives .......................................................... 99
Bench Scale Mini-Column Study ....................................... 100
Materials and Methods ........................................ 100
Results and Discussion ........................................ 104
Pilot-Scale Mobile Pilot Plant Study .................................... Ill
Materials and Methods ........................................ Ill
Results .................................................... 115
CHAPTER 5. SUMMARY AND CONCLUSIONS ............................. 119
Bromate......................................................... 119
Perchlorate....................................................... 122
Metropolitan Water District of Southern California: Ozonated Water
(No Added Electron Donor) for Bromate and Perchlorate Reduction ....... 125
CHAPTER 6. RECOMMENDATIONS TO UTILITIES ......................... 127
REFERENCES ......................................................... 129
ABBREVIATIONS ...................................................... 137
vii

TABLES
1.1 Bromate removal in DFE at pH 6.5 with various DO concentrations .............. 5
1.2 Bromate removal in CUW at pH 7.5 with different influent nitrate
concentrations ............................................... 10
1.3 Batch biological bromate reduction experiments .............................. 15
2.1 Average water quality characteristics of CUW and LMW ..................... 19
2.2 Agar recipe for plating experiments ..................................... 27
2.3 Experimental conditions: ozone/hydrogen peroxide/perchlorate batch tests ......... 34
2.4 Serum bottle experiment 1 using perchlorate-reducing isolate B9
(initial conditions) ............................................. 36
2.5 Serum bottle experiment 2 using perchlorate-reducing isolate B9
(initial conditions) ............................................. 37
3.1 Effect of backwashing on bromate removal in the BAG filter .................. 43
3.2 The effect of influent DO concentration on bromate removal in BAG filters ....... 46
3.3 DOC removal and DO remaining in the BAG experiments using LMW ........... 54
3.4 Testing the isolates for the ability to reduce bromate (12-week incubation;
initial conditions: CUW, 5 mM phosphate buffer, 28.7 ng/L BrO3",
4.82 mg/L NO3-, pH 7.8, 0 mg/L DO) .............................. 65
3.5 Testing the isolates for the ability to reduce bromate (4.5-week incubation;
initial conditions: DDW, 5 mM phosphate buffer, 43.5 ug/L BrO3",
4.92 mg/L NO3', 8.3 mg/L DOC, pH 7.5, 0 mg/L DO) ................. 66
3.6 Challenging isolates with a high nitrate concentration ........................ 67
3.7 Challenging isolates with a high bromate concentration ...................... 68
3.8 Challenging isolates with high concentrations of bromate and nitrate ............ 68
3.9 Summary of experimental evidence for isolates ............................. 70
3.10 Results of dilution-to-extinction experiments .............................. 71
3.11 Abiotic perchlorate batch removal and wash tests ........................... 77
3.12 Perchlorate mass balance: mass perchlorate removed by and recovered from
metal-catalyzed GAG (10 mg/L NO3" rinse) ......................... 79
IX

3.13 Extended perchlorate mass balance: total mass removed by and recovered from
metal-catalyzed GAC (10 + 200 mg/L NO3" rinses) ................... 79
3.14 Ozone/hydrogen peroxide/perchlorate batch tests ........................... 81
3.15 Reduction of perchlorate in aqueous medium using microbial isolates ............ 90
3.16 Dilution-to-extinction experiments: perchlorate ............................ 95
4.1 Historical water quality for Metropolitan source water ...................... 100
4.2 Control study for chemical reduction pathway for perchlorate ................ 109
4.3 Characteristics of Norit carbon used in mobile pilot plant .................... 112
4.4 Mobile pilot plant operating conditions .................................. 115
4.5 Preloading results from the mobile pilot plant ............................. 116

FIGURES
1.1 Effect of influent dissolved oxygen concentration on bromate breakthrough
from the BAG filter ............................................. 6
1.2 Possible depletion of bromate, nitrate, and dissolved oxygen in the biofilm ......... 7
1.3 Reduction of bromate to bromide in a BAG filter ............................ 8
1.4 Effect of influent nitrate concentration on bromate breakthrough from
a BAG filter ................................................. 10
1.5 Nitrogen mass balance in experiments 4 and 5 (25-minute BAG filter) - ammonia,
nitrite, and nitrate ............................................. 11
1.6 Nitrogen mass balance in experiments 6 and 7 (49-50-minute BAG filter) - ammonia,
nitrite, and nitrate ............................................. 12
1.7 Nitrogen speciation in experiment 4 (25-minute BAG filter) .................... 12
1.8 Nitrogen speciation in experiment 6 (49-minute BAG filter) ................... 13
1.9 Percent bromate removal versus EBCT with different influent nitrate
concentrations ............................................... 14
2.1 Schematic of BAG filters using CUW .................................... 22
2.2 Schematic of BAG filters using LMW ................................... 23
2.3 Schematic of multi-port BAG filter using CUW ............................ 24
2.4 Progression of plating and aqueous medium experiments with pure cultures ....... 29
3.1 Repeat of the 10- and 20-minute EBCT filtration experiment (average influent
conditions: CUW, pH 7.5, 2 mg/L DO, 5.0 mg/L NO3', 20 ug/L BrO3") .... 40
3.2 Repeat of the 10- and 20-minute EBCT filtration experiment (average influent
conditions: CUW, pH 7.5, 2 mg/L DO, 0.3 mg/L NO3', 20 |ag/L BrO 3") ... 40
3.3 Changing the influent nitrate concentration from 0.3 to 5.0 mg/L in the
BAG filters .................................................. 41
3.4 Bromate removal in 1- and 2-inch inner-diameter BAG filters ................. 43
3.5 Mass concentration bromate removal with different influent bromate concentrations
(average influent conditions: CUW, pH 7.5, 2 mg/L DO, 0.3 mg/L NO3") ........ 45
3.6 Mass concentration bromate removal with different influent bromate concentrations
(average influent conditions: CUW, pH 7.5, 2 mg/L DO, 5.0 mg/L NO3") ........ 45
xi

3.7 Percent bromate removal with various influent DO concentrations .............. 47
3.8 Effect of influent nitrate concentration on bromate removal in BAG filters
(average influent conditions: CUW, pH 7.5, 20 ug/L BrO3', 2.1 mg/L DO,
26-minute EBCT)............................................. 50
3.9 Effect of influent nitrate concentration on bromate removal in BAG filters
(average influent conditions: CUW, pH 7.5, 20 ug/L BrO3', 2.1 mg/L DO,
51-minute EBCT) ............................................. 50
3.10 Effect of influent sulfate concentration on bromate removal in BAG filters ........ 52
3.11 Effect of influent pH on bromate removal in BAG filters ...................... 53
3.12 Bromate remaining after first BAG filter using LMW ........................ 55
3.13 Nitrate remaining after first BAG filter using LMW ......................... 56
3.14 Bromate remaining after second BAG filter using LMW ...................... 57
3.15 Nitrate remaining after second BAG filter using LMW ....................... 58
3.16 Perchlorate remaining after first BAG filter using LMW ...................... 59
3.17 Perchlorate remaining after second BAG filter using LMW .................... 59
3.18 Rods and cocci from the BAG filter ..................................... 61
3.19 Chain of rods from the BAG filter ...................................... 61
3.20 Aggregated rods from the BAG filter .................................... 61
3.21 Protozoa in the BAC filter ............................................ 62
3.22 DGGE profile of bromate and perchlorate samples .......................... 63
3.23 Bromate reduction in a serum bottle experiment (initial conditions: CUW,
pH 7.7, 19 ug/L BKV, 5.4 mg/L NO3', 25 C) ....................... 72
3.24 Bromate reduction in a serum bottle experiment (initial conditions: CUW,
pH 7.7, 19 ug/L BrO3-, 5.4 mg/L NO3', 4 C) ........................ 73
3.25 Ion exchange of perchlorate to GAC .................................... 74
3.26 Batch perchlorate removal (initial conditions: CUW, pH 7.5, 500 mg/L GAC) ..... 75
3.27 Batch perchlorate removal (initial conditions: DDW, pH 7.5, 500 mg/L GAC) ..... 75
3.28 Batch perchlorate removal (initial conditions: pH 7.5, 500 mg/L GAC) .......... 76
3.29 Batch perchlorate removal (initial conditions: DDW, 500 mg/L GAC) ........... 76
3.30 Removal of perchlorate by GAC mixed with aluminum shot ................... 78
3.31 Removal of perchlorate by copper- and zinc oxide-impregnated carbon .......... 78
xii

3.32 Perchlorate removal vs. nitrate remaining ................................. 82
3.33 Nitrate isolation experiments .......................................... 83
3.34 Comparison of perchlorate removal for nitrate preload and non-nitrate preload
conditions ................................................... 85
3.35 Comparison of perchlorate removal at various EBCTs ....................... 86
3.36 Comparison of perchlorate removal for 25-minute EBCT experiments ........... 87
3.37 BAC filter recovery experiments ....................................... 89
3.38 Isolate B9 serum bottle experiment 2 (initial conditions: DDW, C1O4- 50 ug/L, 0.1
mg/L NCV, DO « 0 mg/L, pH 7.5) ................................ 91
3.39 Isolate B9 serum bottle experiment 2 (initial conditions: DDW, 50 ng/L C1O4-, 0.4
mg/L NCV, DO « 0 mg/L, pH 7.5) ................................ 92
3.40 Isolate B9 serum bottle experiment 2 (initial conditions: DDW, 50 |ig/L C1O4-, 1.6
mg/L NCV, DO « 0 mg/L, pH 7.5) ................................ 92
3.41 Isolate B9 serum bottle experiment 2 (initial conditions: DDW, 50 [ig/L C1O4-, 2.6
mg/L NCV, DO « 0 mg/L, pH 7.5) ................................ 93
3.42 Isolate B9 serum bottle experiment 2: nitrate degradation ..................... 94
3.43 Isolate B9 serum bottle experiment 2: perchlorate degradation ................. 94
3.44 Rods and protozoa from the BAC filter .................................. 97
3.45 Bacterial floe from the BAC filter ....................................... 97
3.46 Large protozoan from the BAC filter .................................... 98
4.1 Schematic of the bench-scale filter setup at Metropolitan (not to scale) ......... 102
4.2 Removal of perchlorate by the mini-column under various conditions ........... 105
4.3 Removal of bromate by the mini-column under various conditions ............. 105
4.4 Mobile pilot plant ....................................................111
4.5 Schematic of the mobile pilot plant setup at Metropolitan (not to scale) ......... 112
4.6 Backwash water supply setup ......................................... 114
4.7 Pre-loading and biological reduction of perchlorate at the pilot-scale ........... 117
4.8 Pre-loading and biological reduction of bromate at the pilot-scale .............. 117
Xlll

FOREWORD
The AWWA Research Foundation is a nonprofit corporation that is dedicated to the
implementation of a research effort to help utilities respond to regulatory requirements and traditional
high-priority concerns of the industry. The research agenda is developed through a process of
consultation with subscribers and drinking water professionals. Under the umbrella of a Strategic
Research Plan, the Research Advisory Council prioritizes the suggested projects based upon current
and future needs, applicability, and past work; the recommendations are forwarded to the Board of
Trustees for final selection. The foundation also sponsors research projects through the unsolicited
proposal process; the Collaborative Research, Research Applications, and Tailored Collaboration
programs; and various joint research efforts with organizations such as the U.S. Environmental
Protection Agency, the U. S. Bureau of Reclamation, and the Association of California Water
Agencies.
This publication is a result of one of these sponsored studies, and it is hoped that its findings
will be applied in communities throughout the world. The following report serves not only as a
means of communicating the results of the water industry's centralized research program but also as
a tool to enlist the further support of the nonmember utilities and individuals.
Projects are managed closely from their inception to the final report by the foundation's staff
and large cadre of volunteers who willingly contribute their time and expertise. The foundation
serves a planning and management function and awards contracts to other institutions such as water
utilities, universities, and engineering firms. The funding for this research effort comes primarily from
the Subscription Program, through which water utilities subscribe to the research program and make
an annual payment proportionate to the volume of water they deliver and consultants and
manufacturers subscribe based on their annual billings. The program offers a cost-effective and fair
method for funding research in the public interest.
A broad spectrum of water supply issues is addressed by the foundation's research agenda:
resources, treatment and operations, distribution and storage, water quality and analysis, toxicology,
economics, and management. The ultimate purpose of the coordinated effort is to assist water
suppliers to provide the highest possible quality of water economically and reliably. The true benefits
are realized when the results are implemented at the utility level. The Foundation's trustees are
pleased to offer this publication as a contribution toward that end.
xv

Perchlorate and chlorate salts are widely used by the chemical, aerospace and defense
industries as oxidizers in propellant, explosives, and pyrotechnics. In April 1997, the limit of
detection for perchlorate decreased from 400 ug/L to 4 |ig/L. Since that time, perchlorate
concentrations ranging from single digits to hundreds of ug/L have been found in drinking water
supplies. Bromate is a regulated disinfection by-product resulting from ozonation of water containing
bromide. It is currently regulated at 10 ng/L. This report describes whether conventional ozone
granular/activated carbon systems can be modified to remove perchlorate and bromate without
sacrificing system performance.
Julius Ciaccia, Jr. James F. Manwaring, P. E.
Chair, Board of Trustees Executive Director
AWWA Research Foundation AWWA Research Foundation
xvi

ACKNOWLEDGMENTS
The authors of this report would like to thank the AWWARF project officer, Kenan Ozekin,
and the PAC (Project Advisory Committee) - Issam Najm (Water Quality and Treatment Solutions,
Inc., Chatsworth, California), Rick Sase (Main San Gabriel Basin Watermaster, Azusa, California)
and Paul Westerhoff (Arizona State University, Tempe, Arizona) for all of their help and advice. The
authors also wish to thank Becky Daugherty, Hatice Inan, and Richard Lin for assistance with data
collection.
xvn

EXECUTIVE SUMMARY
Bromate (BrO3~) and perchlorate (C1O4~) are inorganic ions that have been found in drinking
water. Bromate can be formed during ozonation due to the oxidation of bromide (Haag and Hoigne
1983), and it is regulated by the United States Environmental Protection Agency (USEPA) at a
concentration of 10 ug/L (Waterweek 1998). Based on the work of Kurokawa et al. (1990) and
DeAngelo et al. (1998), it was determined that the ingestion of aqueous bromate caused renal cell
tumors in rats. Perchlorate is a component of munitions and rocket fuel, and the contamination of
drinking water may be related to these sources. No state or federal standard currently exists for
perchlorate, but the USEPA recently set a national action level of 32 |ug/L (USEPA 1999).
Perchlorate is a health concern due to its ability to disrupt the thyroid gland's use of iodine in the
generation of metabolic hormones. Thus, the removal of bromate and perchlorate from drinking
water or their reduction to innocuous end-products is desirable, and requires development of
appropriate technologies. It was believed that the common ionic structure of bromate and
perchlorate would lend itself to a common treatment method for both anions.
RESEARCH OBJECTIVES
The two main objectives of the project were to investigate the biological reduction of
bromate and perchlorate in biologically active carbon (BAG) filters and to investigate the abiotic
reduction of perchlorate in advanced oxidation processes involving granular activated carbon (GAC).
Specific project objectives were as follows:
To show that perchlorate can be removed in BAG adsorbers under the same conditions that
have proven effective for bromate removal, or to show how these conditions must be
changed to obtain biological removal of perchlorate.
+ To investigate the mechanism of the biological reduction process, with
special focus on the role of nitrate-reducing organisms, the types of
microorganisms carrying out the bromate/perchlorate reduction and their
growth requirements, the type and concentration of electron donor, and the
xix

effect of parameters such as bromate and perchlorate concentration, dissolved
oxygen (DO) concentration, nitrate concentration, pH, temperature, and the
role of ozonation preceding BAG.
To investigate advanced oxidation processes to determine whether these reactions can
produce efficient removal of perchlorate, and if so, to develop the process at the bench-scale.
> Emphasis will be placed on the ozone-hydrogen peroxide-GAC process,
virgin GAC, GAC impregnated with metals such as copper and zinc, and
GAC mixed with solids such as iron, zinc, and aluminum.
To optimize the bromate and perchlorate removal processes that show the most encouraging
results and to monitor the product water for potability.
To test the most promising of the bromate and perchlorate removal processes at the pilot-
scale.
APPROACH
The project included bench-scale work conducted at the University of Illinois at Urbana-
Champaign and the Metropolitan Water District of Southern California (MWDSC). In general, one
water quality or operational parameter was changed at a time, in order to observe its effect on
bromate and perchlorate removal. Additionally, pilot-scale work for removal of bromate and
perchlorate using BAG filtration was completed at the La Verne pilot plant at MWDSC; these
experiments were carried out using filter parameters that were tested at the bench-scale.
CONCLUSIONS
1. Increasing the influent DO concentration to the BAG filters from 2.1 mg/L to 13.6
mg/L decreased bromate removal from 40 percent to 11 percent at a 20-minute
empty-bed-contact time (EBCT). The lack of bromate reduction observed in fullxx

scale ozone-BAC plants is likely due to the high concentrations of DO entering the
filters.
1. Filter history is very important to bromate removal in a BAG filter. For instance, the
negative effect of high DO concentrations on the bromate-reducing bacteria was not
immediately reversible when the DO concentration is subsequently reduced.
2. When the influent nitrate concentration to the BAG filters was increased from 0.3 to
42.3 mg/L, bromate removal decreased from 86 to 49 percent at a 26-minute emptybed
contact time.
3. The influent sulfate concentration to the BAG filters did not have a significant effect
on bromate reduction.
4. At a 25-minute EBCT, bromate removal increased from 20 percent to 40 percent
when the influent pH to the BAG filters decreased from 8.2 to near neutral pHs (6.8-
7.5). This is beneficial since pH control may be used to reduce bromate formation
during ozonation and increase biological bromate removal since bromate formation
decreases as the pH decreases.
5. Under similar operating conditions, better bromate removal was observed in BAG
filters operated with Champaign-Urbana water (CUW, a groundwater source) than
using Lake Michigan water (LMW). Thus, composition of the source water can have
a dramatic impact on bromate reduction in a BAG filter.
6. Bromate removal results obtained in a one-inch inner-diameter BAG filter were
reproduced in a 2-inch inner-diameter BAG filter, suggesting that scale-up of the
bench-scale results should not be of concern.
7. Perchlorate and bromate removal were observed in pilot-scale BAG filters.
8. Backwashing the BAG filters using filter effluent with a low DO concentration (2
mg/L) and no chlorine did not adversely affect bromate reduction.
9. For lower bromate concentrations (10-20 [ig/L) and longer EBCTs (25-50 minutes),
bromate removal increased as the influent bromate concentration increased.
However, constant mass concentration bromate removal was observed at higher
influent bromate concentrations (20-50 ug/L) and shorter EBCTs (10-20 minutes),
which may indicate that the electron donor concentration was limiting bromate
removal under these conditions.
xxi

10. Eight bromate-reducing isolates were cultured from the BAG filter, and it is unlikely
that any of them are denitrifiers.
12. Perchlorate was ion exchanged, not reduced, on the surface of GAC. This includes
a GAC bed as well as GAC mixed with aluminum shot and copper- and zincimpregnated
GAC.
13. GAC used in this study had a low ion exchange capacity for perchlorate (0.172 mg
perchlorate per gram GAC).
14. Perchlorate was not chemically reduced in the presence of ozone or ozone/hydrogen
peroxide.
15. BAG filters demonstrated efficient biological reduction of low ng/L concentrations
of perchlorate under anaerobic conditions with the addition of an external electron
donor mixture consisting of acetate, lactate, and pyruvate. The added electron donor
concentration was approximately 1.5-1.8 mg/L as carbon.
16. Biological perchlorate reduction was hindered by the presence of nitrate. When
effluent nitrate concentration increased from 0.05 to 0.35 mg/L, perchlorate removal
decreased from >95 to 65 percent.
17. Perchlorate removal during a zero influent nitrate experiment increased from 60 to
> 95 percent when the filter was preloaded with nitrate prior to the start of the
experiment, as compared to a non-nitrate preloaded condition.
18. The microbial communities present in perchlorate-reducing and bromate-reducing
BAG filters are diverse.
19. Using bench-scale BAG filters following ozonation, complete bromate and
perchlorate removal was observed with 10-50 ug/L of influent bromate and
perchlorate, EBCTs of 25 and 15 minutes, and influent DO concentrations of 0-2
mg/L. No external electron donor was required to achieve this removal.
20. Pilot scale treatment, consisting of ozonation, conventional treatment, and BAG
filtration achieved complete removal of 25 and 50 ng/L of influent bromate and
perchlorate, respectively, without the addition of an external electron donor.
xxn

RECOMMENDATIONS
1. To utilize BAG filtration for bromate and perchlorate reduction, a utility must
determine if reduction of the influent DO concentration to the filter is feasible. Not
only must it be economically feasible, but the utility should also determine that no
unwanted effects are caused by DO reduction (i.e. lack of taste-and-odor removal or
production of hydrogen sulfide in the filter). This could be accomplished through
bench-scale testing with the water and filter medium used by the utility.
2. Bench-scale testing at the utility should also be performed to determine the required
contact time in the BAG filter to obtain the desired removal of bromate and
perchlorate. Contact times between 10 and 40 minutes should be investigated.
3 .Depending on the alkalinity of the water, the utility should consider a combination
of reduction of bromate formation during ozonation and an increase in biological
bromate removal in the BAC filter by pH control.
4. Utilities that treat water contaminated with both nitrate and perchlorate should give
particular consideration to whether or not BAC filtration is a feasible treatment
option. Research has shown that BAC filtration removes nitrate and perchlorate very
efficiently. Additionally, nitrate reduction can enhance perchlorate reduction kinetics,
making BAC filtration particularly attractive for combined nitrate-perchlorate
remediation.
FUTURE RESEARCH
The most important modification to the ozone-BAC process for the application of bromate
removal is the reduction of the influent DO concentration to the filter. Little or no bromate reduction
has been observed at DO concentrations that are typical of the ozone-BAC process. Future work
should address the best way to reduce the influent DO concentration. Additionally, studies should
be conducted to ascertain the effects of a reduced DO concentration on effluent water quality. For
example, a reduced DO concentration to the BAC filter may affect the biodegradation of trace
compounds such as 2-methylisoborneol or atrazine.
In order to truly understand how best to optimize the reduction of bromate or perchlorate in
xxiii

a BAG filter, individual bromate- and perchlorate-reducing cultures must continue to be studied. By
understanding the physiology and nutrient requirements of these organisms, the BAG process may
be designed to select for these organisms. Relatedly, the seeding of particularly efficient bromate-or
perchlorate-reducing organisms to the BAG filters could be tested, and the retention of these
organisms after backwashing should be monitored.
Research to date has indicated that the addition of an external, organic electron donor is
important for the stimulation of efficient, biological perchlorate reduction in a BAG filter. Though
studies done at Metropolitan indicate that ozonated natural organic matter may satisfy electron donor
requirements for biological perchlorate reduction, these results are preliminary. This pre-filtration
ozonation step needs to be further investigated, both at the bench- and full-scales, to ensure that the
electron donor capability of ozonated natural organic matter will suffice to achieve biological
perchlorate reduction under various operating conditions.
Studies must also be undertaken that focus on treatment requirements after water has been
in contact with a BAG filter. The production of any undesirable taste and odor compounds, the
biological stability and chlorine demand of the product water, and reaeration requirements must each
be addressed to ensure the potability of BAC-filtered water.
xxiv

CHAPTER 1. INTRODUCTION
BACKGROUND - BROMATE
Bromate is a disinfection by-product resulting from the ozonation of waters containing
bromide. Bromate formation during ozonation is affected by various factors including pH, and the
concentrations of bromide, ammonia, natural organic matter (NOM), and ozone (von Gunten and
Hoigne 1996, von Gunten and Hoigne 1994, Westerhoff et al. 1993, Haag and Hoigne 1983).
Bromate formation has become an important issue as more utilities are considering ozone to combat
Cryptosporidium and Giardia, color, and taste and odor problems.
Based on the work of Kurokawa et al. (1990) and DeAngelo et al. (1998), it was determined
that the ingestion of aqueous bromate caused renal cell tumors in rats. For a 70-kg adult drinking
2 liters of water per day with a bromate concentration of 5 ng/L, the United States Environmental
Protection Agency (USEP A1998) calculated a lifetime cancer risk of 10"4; a 10"4 cancer risk indicates
that 1 person per 10,000 people would develop cancer. The maximum contaminant level (MCL) for
bromate in the United States has recently been set at 10 ng/L by the Stage 1
Disinfectants/Disinfection By-Products Rule (Waterweek 1998), based on a practical quantification
limit. Therefore, as analytical capabilities advance, a future, lower MCL may be promulgated.
Bromate formation is highly dependent on the bromide concentration. The presence of
bromide results from natural and anthropogenic sources, including seawater intrusion and industrial
discharges. In a survey of 100 U. S. drinking water utilities, Amy et al. (1993) found bromide
concentrations ranging from 2 - 429 ug/L. Krasner et al. (1993) summarized the findings of several
pilot- and full-scale studies that showed bromate concentrations in finished water ranging from 0-60
Hg/L, depending on the conditions and the amount of bromide in the source water. Thus, bromate
formation in excess of the 10 \ig/L standard is a concern for some utilities.
BACKGROUND-PERCHLORATE
In April 1997, the limit of detection (LOD) for perchlorate decreased from 400 ng/L to 4
\igfL (Waterweek 1997). Since that time, perchlorate concentrations ranging from single digits to
hundreds of ng/L have been found in drinking water supplies in California, Nevada, and Arizona.
1

Ammonium perchlorate is an oxidizer that helps rocket fuel ignite and burn, and it is often a
component of munitions. Since ammonium perchlorate has a limited shelf life, it is frequently washed
out of the nation's solid rocket supply. Contamination is likely the result of the large volumes of
perchlorate-containing water generated during this process. Perchlorate contamination could also
be due to Wastes produced by the munitions industry or from areas subjected to substantial use of
explosives, such as mines. The occurrence of perchlorate in drinking water supplies is now being
documented. Extensive perchlorate contamination has been found in drinking water supplies in the
western U.S. For example, perchlorate concentrations of greater than 100 ug/L were detected in the
San Gabriel Basin in Southern California, one of the largest groundwater basins in the United States
(Najm et al. 1999). The state of California has closed 18 drinking water wells due to perchlorate
contamination up to 280 ug/L (Waterweek 1997). Perchlorate has also been detected in Lake Mead,
one portion of the Lower Colorado River system, which supplies drinking water to 10 million people
in Nevada, California, and Arizona. It is thought that most of the Lake Mead contamination is from
the Las Vegas Wash, one of the lake's source waters, in which perchlorate concentrations of more
than 1 mg/L have been measured (Rogers 1997). Ten states have been found to have perchloratecontaminated
drinking water supplies (Betts 1998), and this number is likely to increase, as only six
states do not contain users or manufacturers of perchlorate (Renner 1999). More complete data will
be collected in a survey funded by the American Water Works Association Research Foundation
(AWWARF) that targets both surface and groundwater supplies in the United States.
Perchlorate is of concern due to its ability to disrupt the thyroid gland's use of iodine in the
generation of metabolic hormones. One step in the production of thyroid hormones is the uptake and
transport of iodide (I") within the thyroid gland. The iodide ion must be present for thyroid hormone
production to occur. The thyroid gland can mistake perchlorate for iodide because perchlorate has
a similar molecular volume and charge to iodide. When perchlorate is taken up by the thyroid gland
instead of iodide, the thyroid gland loses some hormone production capacity (Book 1999).
No state or federal standard for perchlorate currently exists in the U.S. since perchlorate
contamination has only recently come to the forefront. However, based on a 1992 health study
conducted by the USEPA, the California Department of Health Services (CDHS) established a
provisional perchlorate action level of 18 ng/L (CDHS 1997). However, the national action level is
32 ug/L, as set by the USEPA (USEPA 1998).

PREVIOUS STUDIES - BROMATE 1
Siddiqui et al. (1994) observed bromate reduction by UV light (180 - 254 nm), and Mills et
al. (1996a) found that the use of a platinised titanium dioxide catalyst enhanced the rate of bromate
reduction by UV light (254 nm). However, UV irradiation for bromate reduction is currently, not
economically viable (Siddiqui et al. 1994).
Bromate reduction by granular activated carbon (GAC) has been demonstrated (Kirisits et al.
1998, Miller 1996, Mills et al. 1996b, Siddiqui et al. 1994, 1996). Bromate removal by GAC is
hindered in natural water by the presence of NOM and anions such as chloride, nitrate, and sulfate.
Using a GAC filter, Kirisits et al. ( 1 998) observed 50 percent bromate breakthrough after 4 1 ,000 bed
volumes for distilled deionized water (DDW) as compared to less than 1,000 bed volumes for natural
water. It was concluded that abiotic reduction of bromate by GAC could not be feasibly sustained
for a significant period of time in natural water.
Using half reactions (Equations 1.1-1.3), reduction potentials were calculated for bromate,
nitrate, and oxygen. The following concentrations, typical for the current study, were used in the
calculations: 20 ug/L BrO/, 20 ug/L Br', 5 mg/L NO/, 2 mg/L O2, pH 7.5, and the partial pressure
of nitrogen (/Vz) was assumed to be 1 since the water was sparged with nitrogen gas to remove
oxygen.
BrO~ + 6H+ + 6e~ -» Br' + 3H2O; E = 0.996 V (1.1)
O2 + 4/T + 4e~ -» 2H2 O; E = 0.767 V (1.2)
2NO3 ~ + 12H+ + 10e- -> N2 + 6H2 O; E = 0.665 V (1.3)
Bromate is a stronger oxidant than both oxygen and nitrate as shown by the calculated
reduction potentials. Although bromate has a higher reduction potential than oxygen and nitrate,
reaction kinetics must also be considered. Enzymes strongly affect the kinetics of electron acceptor
utilization. Bromate reduction can be catalyzed by spinach nitrate reductase, an enzyme that catalyzes
the reduction of nitrate (Barber and Notton 1990). If denitrifying organisms also have the capability
to reduce bromate using nitrate reductase, the nitrate concentration will affect bromate removal.
1 Adapted from Journal AWWA, 91(8), 74-84.
3

The formation and activity of nitrate reductase are also affected by oxygen concentration. For
example, approximately 30 uM oxygen (0.96 mg/L) was found to inhibit the formation of nitrate
reductase in Hyphomicrobium (Meiberg et al. 1980) while oxygen concentrations in the vicinity of
10 uM (0.32 mg/L) inhibited the activity of the enzyme (Tiedje 1988). If the oxygen concentration
is too high, this enzyme may not be available for nitrate or bromate reduction. If conditions permit
nitrate reductase activity, then nitrate and bromate could be competing for use as electron acceptors.
On the other hand, more recent work has demonstrated the occurrence of denitrification under
aerobic conditions (Robertson and Kuenen 1990, Robertson and Kuenen 1984). For example,
aerobically grown Thiosphaerapantotropha, when placed in an anoxic environment, were able to use
nitrate immediately; the lack of a lag time prior to nitrate use indicated that the enzymes involved in
denitrification were already expressed and active (Robertson and Kuenen 1984). Furthermore, these
investigators found that Thiosphaerapantotropha could use oxygen and nitrate simultaneously. Thus,
complete removal of oxygen may not be necessary for bromate reduction. Moreover, the existence
of oxygen and nitrate gradients within the biofilm may permit the use of bromate as an electron
acceptor. DO and nitrate gradients have been found in microbial floes (Vandenabeele et al. 1995)
For example, these investigators found that the DO concentration dropped from 0.25 mM (8 mg/L)
in the bulk to 0.1 mM (3.2 mg/L) in the center of the floe (=650 um = 0.03 in); the nitrate
concentration dropped from 60.1 mg/L in the bulk to 50.8 mg/L approximately 600 um (0.02 in)
into the floe. If the DO concentration reaches an appropriate level for the microorganisms in the
biofilm of the BAG filters to maintain active nitrate reductase, then it may be possible that nitrate and
bromate can be reduced in this region.
Hijnen et al. (1995) demonstrated batch microbial bromate reduction in the absence of oxygen
and nitrate. However, a more recent study by Hijnen et al. (1999) showed bromate reduction in the
presence of nitrate using a denitrifying bioreactor supplemented with ethanol. The authors suggested
that the bromate reduction was taking place in a portion of the biofilm where nitrate had already been
depleted, although they did not discount the possibility of biological bromate reduction below a
threshold nitrate concentration. Hijnen et al. (1999) concluded that the use of a denitrifying
bioreactor for bromate reduction would not be practical due to the extended contact times required
for bromate removal to 0.5 ug/L (proposed Dutch standard) and the post-treatment that would be
required to eliminate the excess ethanol and biomass. It should be noted that the influent water used
for this study had a very high nitrate concentration of 85 mg/L, and it is not surprising that 50 mg/L

ethanol was required to stimulate enough biological growth to remove the nitrate. However, on the
basis of the study by Hijnen et al., one should not conclude that biological bromate reduction is
infeasible for a water that has to meet the U.S. standard for bromate (10 ng/L), especially since the
influent nitrate concentration will likely be significantly less than 85 mg/L.
Experiments concerning the reduction of bromate in biologically active carbon (BAG) filters
have been completed at the University of Illinois at Urbana-Champaign and have shown excellent
bromate removal in the presence of an effluent nitrate concentration between 2 and 8 mg/L. These
results were published by Kirisits and Snoeyink (1999), and are discussed in the following sections.
Bromate Reduction with Various Influent Dissolved Oxygen Concentrations
A set of BAG filtration experiments were run with different influent DO concentrations,
ranging from 2.0 to 8.0 mg/L. Danville filter effluent (DFE), obtained from the Consumer's Illinois
Treatment Plant (Danville, Illinois), was used in conjunction with an empty bed contact time (EBCT)
of approximately 50 minutes in the BAG filter. Since nitrate and bromate may compete for use as
electron acceptors, it is noteworthy that the batch of DFE used for these experiments contained
approximately 25 mg/L nitrate. The pertinent experimental information is shown in Table 1.1.
Table 1.1 Bromate removal in DFE at pH 6.5 with various DO concentrations
Influent
# EBCT (min) DO
(mg/L) NO3-
(mg/L) BrO3-
("g/L)
1 50 4.5 24.9 10
2 49 2.0 24.9 11
3 50 8.0 27.8 11
(Adapted from Journal AWWA, 91(8), 74-84.)
Effluent
DO
N03"
(mg/L) (mg/L)
_2
_2
24.6
24.2
4.9 26.7
% BrO3-
removal
56
61
42
95%
confidence
interval
50-62
54-68
38-46
Data not available

The bromate breakthrough data for these experiments are shown in Figure 1.1. Note that at
the start of the experiment, bromate breakthrough was 70 percent (C/C0 = 0.7) but then gradually
decreased. This is considered to be a period of acclimation for the microorganisms, in which the
microbial communities adjusted to the 50-minute EBCT and the DO concentration of 4.5 mg/L. Prior
to this experiment, the BAC filter had been operated with an 8-minute EBCT and an influent DO
concentration of approximately 8.0 mg/L.
The data in Table 1.1 show 42 percent bromate removal in the presence of DO concentrations
of 4.9 to 8.0 mg/L (Experiment 3). Increased bromate removal was observed as the influent DO
concentration decreased from 8.0 mg/L to 4.5 mg/L (Experiments 3 and 1); the average percent
bromate removals for these two experiments are different at the 95 percent confidence level since
their confidence intervals do not overlap (Table 1.1). The trend of improved bromate removal at
lower DO concentrations was expected since oxygen may inhibit the activity of the enzyme (possibly
nitrate reductase) that catalyzes bromate reduction. However, as the influent DO concentration
decreased from 4.5 to 2.0 mg/L (Experiments 1 and 2), the average percent bromate removal did not
increase in a statistically significant manner. Table 1.1 shows that the 95 percent confidence intervals
overlap for these experiments. Although the effluent DO concentrations from Experiments 1 and 2
GO
c
o
1
U
i.u -
0.8 -
0.6 -
0.4 -
0.2 -
n n -
0
•
• •
4.5 mg/L DO
r •• a .
56% removal
—————— ... i —— _..,„,
2.0 mg/L DO
mm —————•
• • •
• i'. •
61% removal
8.0 mg/L DO
• . ' •
TT. . m . *-
•
, 42% removal
j —.,_. ..... ———————
200 400 600
800 1000
Bed Volumes
Figure 1.1 Effect of influent dissolved oxygen concentration on bromate breakthrough from the
BAC filter (initial conditions: DFE, EBCT «50 minutes, pH 6.5, 10-11 ng/L BrO3")
(Adapted from Journal AWWA, 91(8), 74-84.)

were not measured, they can be presumed to be relatively low since 3.1 mg/L DO was consumed in
Experiment 3. Thus, there may already be a region of considerable size in the BAG filter of
Experiment 1 where the DO is quite low, and bromate removal can take place. Decreasing the
influent DO concentration further in Experiment 2 does not result in additional bromate removal.
The nitrate concentration in these experiments was quite high (approximately 25 mg/L) and probably
was a major factor affecting bromate removal. Such an effect is expected if nitrate reductase is the
enzyme responsible for bromate reduction. Only low concentrations of nitrate (0.3 mg/L -1.1 mg/L)
were removed in the BAG filter. However, even with these relatively high concentrations of nitrate
and a pH of 6.5, which was probably less than optimum for denitrification, bromate removals of 42 -
61 percent were achieved.
It is likely that microorganisms in the early reaches of the filter reduce the DO concentration.
Therefore, organisms deeper within the filter are exposed to lower or negligible DO, and nitrate
reductase synthesis or activity could occur here unimpeded by the presence of DO. Due to
diffusional transport, there will also be a gradient of nitrate and bromate in the biofilm (Figure 1.2).
As suggested by Hijnen et al. (1999), denitrifying microorganisms may possess the ability to reduce
bromate below a threshold nitrate concentration, and the nitrate concentration in a portion of the
biofilm may reach a low enough concentration for bromate to be used as an electron acceptor.
Bulk Solution Containing
Bromate, Nitrate, and DO
Bromate
GAC BIOFILM
Figure 1.2 Possible depletion of bromate, nitrate, and dissolved oxygen in the biofilm
(Adapted from Journal AWWA, 91(8), 74-84.)

Bromate Reduction to Bromide
Since the previous section demonstrated the removal of bromate in the BAG filter, it was then
desirable to determine if bromate was indeed reduced to bromide. To this end, a BAG filter with a
20-minute EBCT was operated with DDW spiked with an electron donor mixture consisting of
lactate, acetate, and pyruvate to maintain the biological activity. This experiment was performed
in DDW so that small changes in bromate and bromide would be easier to detect analytically. More
than half the data in Figure 1.3 show that less bromide was observed than expected on the basis of
bromate removal. In fact, the data points showing bromate removed and bromide produced were up
to 3.5 ng/L apart. In kinetic studies with GAG and bromate, a mass balance on Br showed that the
initial BrO3"-Br in the water was equivalent to the BrO3"-Br and Br" in solution at the end of the test,
plus the Br" displaced by a sulfate wash of the GAG at the end of the experiment (Miller 1996).
However, since the BAG filter in this study had been continuously treating bromate in water for over
one year prior to this experiment, the probability of bromate or bromide adsorbing to the carbon is
3
1
2
CQ
10
--B-Bromate removed
—•-Bromide produced
0
0
—i—
50 100 150
200 250
Bed Volumes
Figure 1.3 Reduction of bromate to bromide in a BAG filter (initial conditions: DDW, EBCT =
20 minutes, pH 7.5, 1.3 mg/LDO, 2.4 mg/LNO3", 18 ug/LBrO 3")
(Adapted from Journal AWWA, 91(8), 74-84.)

one year prior to this experiment, the probability of bromate or bromide adsorbing to the carbon is
quite low. It has been shown that the bromate removal capacity of GAC in the absence of biological
activity is exhausted very quickly in natural water (Kirisits et al. 1998). Therefore, the lack of a
balance on Br in the BAG filter may indicate the presence of other reduced species that are
intermediates in the reduction of bromate to bromide. Alternatively, Br may be associated with or
adsorbed on the microbes, thereby preventing a mass balance on Br.
Bromate Reduction with Various Influent Nitrate Concentrations
Experiments 4-7 (Table 1.2) employed Champaign-Urbana water (CUW) to show the effect
of influent nitrate concentration on bromate removal. Since the results of Experiments 1-3 showed
better bromate removal with lower influent DO concentrations, Experiments 4-7 were operated with
an influent DO concentration of approximately 2 mg/L. The effluent DO concentration from the
filters was typically 0.1- 0.2 mg/L, although some of this oxygen may have been introduced during
the measurement. The bromate removal data from Experiments 4 and 5 are shown in Figure 1.4. The
bromate removal results with the 25-minute EBCT were particularly encouraging because they
indicated that excellent bromate removal can be obtained at EBCTs that could be used in
conventional ozone/GAC plants. Typically, BAG filters are operated with EBCTs in the range of 5
to 20 minutes (Bishop and Bourbigot 1996).
As the influent nitrate concentration increased from 0.2 mg/L to 5.1 mg/L with a 25-minute
EBCT, the bromate removal decreased from 86 percent to 76 percent (Table 1.2, Figure 1.4).
However, when the EBCT was increased to 50 minutes, the bromate removal decreased
insignificantly from 95 percent to 93 percent for the same change in nitrate concentration (Table 1.2).
Less bromate reduction is generally expected as the nitrate concentration is increased since bromate
and nitrate may be competing electron acceptors. However, the increased contact time in Experiment
7 was able to counteract the negative effect of the elevated nitrate concentration, allowing bromate
to be removed to the detection limit. The importance of the effect of nitrate concentration on bromate
removal indicated that the nitrogen relationships in the filters must be more closely examined. The
effluent nitrate concentrations for Experiments 4-7 were always higher than the influent
concentrations, indicating that nitrifying organisms were present. CUW typically contained 0.3-0.8
mg/L NH3-N, and some of this was converted to nitrite and nitrate, thereby reducing the amount of

Table 1.2 Bromate removal in CUW at pH 7.5 with different influent nitrate concentrations
Influent Effluent
# EBCT DO NO3- BrO3" DO NO3' % BrO3'
(min) (mg/L) (mg/L) (ug/L) (mg/L) (mg/L) removal
4 25 2.2 0.2 21 0.1
5 25 2.3 5.1 20 0.1
0.9 86
5.7 76
6 49 2.2 0.2 21 0.2 2.7 95
7 50 2.3 5.1 20 0.2 8.1 93
(Adapted from Journal AWWA, 91(8), 74-84.)
1 0 -|
ao
_c
•a o.s -
1
3 •s uo 0.6 -
-•-5. 1 mg/L influent nitrate (Expt. 5)
— E3-0.2 mg/L influent nitrate (Expt. 4)
1 Q 0.4 -
u
m
1
1
—— m ———— 76% removal
-2 0.2 - • B -» Q_ ^-,
8
LJ 0
u 86% removal
[•y , /•> /\
D D
U.U i i i i i
i i
0 200 400 600 800
Bed Volumes
1000 1200 1400
Figure 1.4 Effect of influent nitrate concentration on bromate breakthrough from a BAG filter
(initial conditions: CUW, EBCT - 25 minutes, pH 7.5, 2.3 mg/L DO, 21 pg/L BrO3")
(Adapted from Journal AWWA, 91(8), 74-84.)
DO in the filters. Figures 1.5 and 1.6 show nitrogen mass balances, including ammonia, nitrite, and
nitrate, for Experiments 4-7. A few of the points show that the effluent nitrogen is greater than the
influent nitrogen by less than 0.1 mg/L as N. This is within the experimental error of the analytical
methods used. In general, the influent nitrogen values agreed quite well with effluent nitrogen values,
except for Experiment 6 (Figure 1.6). Since effluent nitrogen was less than influent nitrogen for this
10

2.0 -
e •
• 8
i 8
a

i.s-
2.0 -
S i- D
sg> i.o -
£
0.5 -
n n -
'.8 . 8 «
-
B
• •
^^ n D
H
H u
n n g
• Influent Nitrogen (Expt. 6) D Effluent Nitrogen (Expt. 6)
• Influent Nitrogen (Expt. 7) O Effluent Nitrogen (Expt. 7)
0 100 200 300
Bed Volumes
400 500
Figure 1.6 Nitrogen mass balance in experiments 6 and 7 (49-50-minute BAG filter) - ammonia,
nitrite, and nitrate
(Adapted from Journal AWWA, 91(8), 74-84.)
00
fi
I.O
1.4 -
1.2 -
1.0 -
0.8 -
0.6 -
0.4 -
0.2 -
nn
• Influent Nitrate
o Effluent Nitrite
X
X
a 7.
X
x.
0
D D
D Effluent Nitrate
- X Influent Ammonia
X
o
8 S
0 100 200 300 400 500
X
o
Bed Volumes
• Influent Nitrite
X Effluent Ammonia
X
X
0 0
° n
o
n
I
x
0
n
i
600 700 800 900
Figure 1.7 Nitrogen speciation in experiment 4 (25-minute BAG filter)
(Adapted from Journal AWWA, 91(8), 74-84.)
12

i S
I
i.o -
1.4 -
1.2 -
1.0 -
0.8 -
0.4 -
0.2 -
no
• Influent Nitrate
• Influent Nitrite
X Influent Ammonia
X X
X
n n n D Q
n D
sk ^t f~\ /^
i i i • Si
n a
0 100 200 300
Bed Volumes
D Effluent Nitrate
o Effluent Nitrite
X Effluent Ammonia
X
i i it i i
400 500
Figure 1.8 Nitrogen speciation in experiment 6 (49-minute BAG filter)
(Adapted from Journal AWWA, 91(8), 74-84.)
Bromate Reduction with Various EBCTs
The EBCT was varied by changing the flowrate through the filters. These data were collected
using CUW at pH 7.5, spiked with 20 ug/L bromate. The average influent DO concentration was
approximately 2 mg/L. As before, the data were collected by using two filters in series so that two
EBCTs could be evaluated simultaneously. Two influent nitrate concentrations were examined -
approximately 0.3 and 5.0 mg/L nitrate. As shown in Figures 1.7 and 1.8, nitrification took place
in the filters and increased the nitrate concentration and decreased the oxygen concentration to near
zero. For example, in the experiment with the 25- and 50-minute EBCTs, even though the influent
nitrate concentration was 0.2 mg/L, the effluent nitrate concentration was between 0.7 and 1.5 mg/L
from the 25-minute column and between 2.1 and 3.5 mg/L from the 50-minute filter. The trend of
bromate removal versus EBCT is plotted in Figure 1.9. The error bars on each data point indicate
the 95 percent confidence interval.
Figure 1.9 shows that bromate removal varied widely with contact time. With a 4-minute
EBCT, bromate removal for both the 0.3 and the 5.0 mg/L influent nitrate experiments was about 10
percent. With a 50-minute EBCT, the effluent bromate concentration for both the 0.3 and the 5.0
13

• 0.3 mg/L influent nitrate
D 5.0 mg/L influent nitrate
0
0
10 20 30 40 50 60
EBCT (min)
Figure 1.9 Percent bromate removal versus EBCT with different influent nitrate concentrations
(initial conditions: CUW, pH 7.5, 2 mg/L DO, 20 ug/L BrO3')
(Adapted from Journal AWWA, 91(8), 74-84.)
mg/L influent nitrate filtration experiments was approximately at the detection limit. Between these
two extremes, however, better bromate removal was obtained when less influent nitrate was present.
The EBCT affects the time available for bromate to diffuse into the biofilm and react. Thus,
for shorter EBCTs, the percent bromate removal is expected to be lower. If EBCT is decreased by
increasing the hydraulic loading to the filter, the thickness of the biofilm should decrease because of
hydraulic shear forces, and bromate removal should accordingly decrease.
Verification of Biological Bromate Removal
In order to verify that the observed bromate removal in the BAC filters was indeed due to
microbial action and not merely a catalytic effect of the carbon, batch biological experiments were
performed. Microorganisms were harvested from the BAC filters, and these microorganisms were
spiked into glass serum bottles containing DDW amended with trace elements or CUW. All bottles
were buffered with 1 mM phosphate buffer, adjusted to pH 7.5, and spiked with 20 ug/L bromate,
14

ut the concentrations of oxygen and nitrate varied in the bottles. The bottles were sampled for
bromate and nitrate on Days 1 and 19 of the experiment (Table 1.3). Table 1.3 shows that bromate
removal did not take place in the absence of microorganisms (Bottle 1). When the same solution
composition was used in Bottle 2, and inoculated with microorganisms from the BAG filters, bromate
reduction did occur. This verifies bromate removal by the microorganisms in the BAG filters. In
spite of the observation of biological bromate removal, this does not rule out the continued reduction
of bromate by the carbon. However, previous work has demonstrated that abiotic bromate reduction
by GAG is not sustainable for long periods of time (Kirisits et al. 1998); for example, 85 percent
bromate breakthrough was observed after a nine-day, non-biological GAG filtration experiment in
natural water with a 2-minute EBCT. While it is still possible that a small amount of abiotic bromate
Table 1.3 Batch biological bromate reduction experiments (initial conditions: pH 7.5, DDW
supplemented with trace salts and metals or CUW)
Bottle #
1
2
3
4
5
6
Water
DDW
DDW
CUW
CUW
CUW
CUW
Day
1
19
1
19
1
19
1
19
1
19
1
19
Bromate
(ug/L)
12.4
11.4
9.4
1.0
11.5
1.0
13.0
2.3
14.1
9.3
13.1
1.0
Nitrate
(mg/L)
0.2
0.2
0.2
0.2
0.3
0.2
6.2
0.2
0.3
4.9
0.3
0.2
DO Inoculum
(mg/L)
0
0 /
0 /
0 /
8.3 /
>0
1.5 /
0
(Adapted from Journal AWWA, 91(8), 74-84.)
15

eduction by GAC is occurring at the extended contact times used in the current BAG filtration
experiments, it is more likely that the preponderance of the bromate reduction observed is biological.
Bottles 3-6 were set up with various combinations of initial DO concentrations (0 - 8.3 mg/L)
and nitrate concentrations (0.2 - 6.2 mg/L), and all bottles exhibited microbial bromate removal. By
Day 19, Bottles 3, 4, and 6 showed bromate removal to approximately the detection limit while the
microorganisms in Bottle 5 had only reduced the bromate concentration from 14.1 ug/L to 9.3 ug/L.
Bottle 5 contained higher final concentrations of DO and nitrate than did the other bottles. On Day
19, Bottle 5 contained 4.88 mg/L nitrate and a non-zero DO concentration as compared to 0.2 mg/L
nitrate and 0 mg/L DO in the other bottles. The elevated concentrations of both DO and nitrate in
Bottle 5 probably caused the slower rate of bromate reduction in this bottle.
PREVIOUS STUDIES - PERCHLORATE
Biological and chemical reduction processes, membrane processes, and ion exchange are being
investigated for perchlorate removal. The latter two are being investigated in separate research
projects under the supervision of AWWARF and are not reviewed here. Reduction processes are
based on the fact that perchlorate is thermodynamically unstable. According to free energy
relationships, perchlorate can oxidize water to form oxygen, resulting in the following overall reaction
(Note: H2O appears in the oxidation half-reaction but not the overall reaction):
C1O4- - Cl- + 2 O2
This reaction has a large energy barrier that keeps it from taking place in the absence of a catalyst.
Similar to free and combined chlorine, chlorite, chlorate, and bromate, the low concentrations of
perchlorate in drinking water are kinetically stable. An objective of this research is to find a
microorganism with the appropriate enzyme, or a catalytic solid, that can enable this reaction to
proceed rapidly.
In a study of biological reduction, Rikken et al. (1996) found that a microorganism in the P-
subgroup of the Proteobacteria was able to use perchlorate as an electron acceptor. Hurley et al.
(1996) described a pilot system that is in place at Tyndall Air Force Base to treat the wash water from
motors containing ammonium perchlorate. In this system, Wolinella succinogenes is the bacterial
16

species that is able to reduce perchlorate to chloride and perchlorate is reduced from 3000 mg/L to
0.5 mg/L. Biological reduction of perchlorate in the ug/L concentration range has only recently been
achieved (Herman and Frankenberger 1999, Logan 1999). Herman and Frankenberger (1999) used
a pure culture of perclace to reduce perchlorate to below the detection limit in sand filters. Logan
(1999) isolated perchlorate-reducing microorganisms from a sample of wastewater and utilized these
microorganisms in various bioreactor configurations to achieve perchlorate reduction. A completely
mixed, autotrophic, hydrogen-oxidizing biofilm reactor has been used by Rittmann et al. to examine
perchlorate, nitrate, and hydrogen utilization mechanisms (Rittman et al. 2000).
In each of the biological reduction processes found in the literature to date, there is a need
to maintain anaerobic conditions. Oxygen is a major inhibitor of perchlorate reduction and exposure
of active perchlorate-reducing microorganisms to air can immediately terminate perchlorate reduction
(Attaway and Smith 1993). Attaway and Smith (1993) reported that in order for perchlorate
reduction to occur, a redox potential of less than -110 mV must be maintained. Other operating
parameters that have remained consistent among biological processes are: 1) neutral pH, 2) moderate
to high temperatures, and 3) a carbon source suitable for respiration (Herman and Frankenberger
1998).
Chemical reduction by GAC, probably with GAC serving as the electron donor, had been
considered to be a promising method of perchlorate reduction. However, studies done by the
California Department of Health Services, found that a virgin bed of GAC removed perchlorate for
only two weeks (Forbes 1997).
OBJECTIVES
The two main objectives of the project were to investigate the biological reduction of bromate
and perchlorate in BAG filters and to investigate the abiotic reduction of perchlorate in advanced
oxidation processes involving GAC. Specific project objectives were as follows:
• To show that perchlorate can be removed in BAC adsorbers under the same conditions that
have proven effective for bromate removal, or to show how these conditions must be changed
to obtain biological removal of perchlorate.
17

• To investigate the mechanism of the biological reduction process, with
special focus on the role of nitrate-reducing organisms, the types of
microorganisms carrying out the bromate/perchlorate reduction and their
growth requirements, the type and concentration of electron donor, and the
effect of parameters such as bromate and perchlorate concentration, dissolved
oxygen (DO) concentration, nitrate concentration, pH, temperature, and the
role of a prefiltration ozonation step.
To investigate advanced oxidation processes to determine whether these reactions can
produce efficient removal of perchlorate, and if so, to develop the process at the bench-scale.
*• Emphasis will be placed on the ozone-hydrogen peroxide-GAC process, virgin
GAC, GAC impregnated with metals such as copper and zinc, and GAC
mixed with solids such as iron, zinc, and aluminum.
To optimize the bromate and perchlorate removal processes that show the most encouraging
results and to monitor the product water for potability.
To test the most promising of the bromate and perchlorate removal processes at the pilot-
scale.
18

CHAPTER 2. MATERIALS AND METHODS
MATERIALS USED FOR BROMATE REDUCTION EXPERIMENTS
Water
Deionized, distilled water (DDW) had a total organic carbon concentration less than 0.3
mg/L. Champaign-Urbana, tap water (CUW) and Lake Michigan water (LMW) were the two natural
waters used in this study. CUW is tapwater (Champaign, Illinois), and it comes from a groundwater
source. Prior to use, CUW was dechlorinated by sodium sulfite addition. Raw LMW was obtained
from the crib of the South Water Production Plant (Chicago, IL); the crib is 3.2 kilometers (2 miles)
from shore at a depth of 9.6 to 10.5 meters (32-35 feet). The main differences between the two
natural waters were the ammonia (NH3) concentrations and biodegradability of the organic matter.
CUW contained less than 1.70 mg/L NH3-N, with an average concentration of 0.80 mg/L, and the
NH3-N concentration in LMW was below the detection limit of 0.05 mg/L. The ammonia in CUW
led to the production of nitrate in the BAG filters; little or no nitrification occurred in LMW, and the
background nitrate concentration was 1.6 mg/L. The organic matter in CUW displayed some degree
of biodegradability in the BAG filter, with an average dissolved organic carbon (DOC) removal of
25 percent. The organic matter in LMW was fairly inert in the BAG filter with an average DOC
removal of 4 percent. Average water characteristics of CUW and LMW are listed in Table 2.1.
Table 2.1 Average water quality characteristics of CUW and LMW
Water
Chloride
Nitrate
Sulfate
DOC
Ammonia
(mg/L)
(mg/L)
(mg/L)
(mg/L)
(mg/LN)
CUW
8
0.3
1
1.4
0.80
LMW
10
1.6
23
2.4

Reagents
Reagent grade sodium bromate (Aldrich Chemical Co., Inc., Milwaukee, WI) and sodium
perchlorate (Sigma Chemical Co., St. Louis, MO) were used to prepare 1000 mg/L stock solutions.
Other chemicals were of reagent grade quality or better. Where appropriate, chemicals were dried
overnight at 105 °C and stored in a desiccator.
METHODS UTILIZED FOR BROMATE REDUCTION EXPERIMENTS
Biologically Active Carbon - Small Columns
BAC filters were constructed in 2.5-cm (1-inch) inner-diameter glass pipes (Ace Glass,
Vineland, NJ). The pipes were capped with Teflon® and steel endcaps. A small layer of 3-mm (0.1-
inch) glass beads was placed below the carbon bed in order to promote uniform flow characteristics
and prevent channeling. Influent was pumped upflow through the columns using peristaltic
Masterflex pumps (Cole-Parmer, Vernon Hills, EL). Each influent container was fitted with a stainless
steel, Teflon®, or Plexiglas® floating cover, to maintain the DO concentration. The influent DO
concentration was adjusted to the desired level by sparging with compressed nitrogen or oxygen.
Four filters were constructed to evaluate bromate removal from CUW using Norit RO 0.8
GAC (Norit Americas Inc., Atlanta, GA). The average GAC pellet diameter and length were 0.8 and
3.4 mm, respectively. Each filter contained one-quarter preloaded carbon and three-quarters fresh
carbon. The preloaded carbon was taken from a bench-scale filter that had been operated with a 2-
minute EBCT for 40,000 bed volumes with dechlorinated surface water (Danville filter effluent,
Consumer's Illinois, Danville, Illinois). At the end of this preloading period, the carbon was used in
experiments not discussed in this report, and approximately 30 percent of the influent DOC and 10
percent of the influent bromate were consistently being removed by the microorganisms during a 45-
day experiment. At this time, the preloaded carbon was considered to be biologically active and at
steady state. Aforementioned, the preloaded carbon was then mixed with fresh carbon in the
construction of four new filters. This was done in order to seed the fresh carbon with
microorganisms, thereby reducing the preloading time required to stimulate biological activity in the
filters.
20

BAG was formed by extensively preloading the carbon filters with CUW that had been
dechlorinated by sulfite addition. The purpose of the preloading period was to establish a biofilm on
the carbon, and then the BAG filter could be used for various biological bromate reduction
experiments. The preloading conditions were determined in accordance with the experiments that
were to immediately follow the preloading time. It was planned to use the BAG filters in experiments
with an influent DO concentration of 2 mg/L and a variety of EBCTs between 4 and 50 minutes.
Therefore, the carbon filters were preloaded with 2,700 bed volumes of dechlorinated CUW with an
EBCT of 10 minutes and an influent DO concentration of approximately 2 mg/L. Two parallel
filtration experiments were set up, each with two columns in series, for a total of four filters. All
filtration experiments with CUW were run with two filters in series, so that two EBCTs could be
evaluated simultaneously. For example, influent water first passed through a 25-minute filter and then
through another 25-minute filter for a total EBCT of 50 minutes. The only difference between the
preloading conditions of the parallel filters was the influent nitrate concentration; one set-up was
operated with an influent nitrate concentration of 0.3 mg/L and the other with an influent nitrate
concentration of 5.0 mg/L. No exogenous electron donor was added to CUW experiments. A
schematic diagram showing the BAG filters using CUW is shown in Figure 2.1.
As the filters were preloaded, the percent bromate removal continued to decrease as the
bromate reduction capacity of the carbon was approached. However, near the end of the preloading
period, the percent bromate removal began to increase. This indicated that while the abiotic reduction
capacity of the carbon was approaching exhaustion, microbiological bromate reduction had
commenced. For the filters being preloaded in the presence of 5.0 mg/L nitrate, the bromate removal
decreased to reach a minimum of 27 percent as the abiotic bromate reduction capacity of the carbon
approached exhaustion but then increased to approximately 53 percent removal by the end of the
preloading time due to the onset of biological bromate reduction. For the filters being preloaded in
the presence of 0.3 mg/L nitrate, the bromate removal decreased to reach a minimum of 37 percent
but then increased to 83 percent removal by the end of the preloading period due to biological
bromate reduction. These data show that the BAG filters were biologically active after the preloading
period. To attain steady state bromate removal, the BAG filters were operated with an EBCT of 25
minutes (each filter) for 30 days. For the last two weeks of this period, the filter being operated with
an influent nitrate concentration of 0.3 mg/L and a 25-minute EBCT removed 83 ± 5 percent
bromate.
21

Influent:
CUW
10-50 ug/L Bromate
pH 6.8-8.2
2-14mg/LDO
0.3 - 42 mg/L Nitrate
t
NOTE: Not to Scale
2.5-cm Inner-Diameter
8.0-cm BAC Bed Height
Support
Material
Figure 2.1 Schematic of BAG filters using CUW
For the filter being operated with an influent nitrate concentration of 5.0 mg/L and a 25-minute
EBCT, the last two weeks of this experiment showed 81 ± 6 percent bromate removal. The
consistent bromate removals observed in these filters demonstrated the attainment of steady state.
New filters were not assembled for each experimental condition (i.e. DO concentration, nitrate
concentration, EBCT). Therefore, zero bed volumes is a term relative to the start of the particular
experiment, not necessarily to the overall life of the filter. Each experimental condition was run for
approximately 2 weeks, to ensure adequate acclimation to the experimental conditions.
Figure 2.2 shows a schematic diagram of another set of filters that was constructed to
evaluate bromate reduction using LMW. Norit 830 GAC (Norit Americas Inc., Atlanta, GA) was
the medium in these filters, and it was taken from the full-scale plant at the Paul M. Neal Water
Treatment Facility (Lake Bluff, IL). This carbon had and average particle diameter of 0.9 mm and
had been used for approximately seven years to treat ozonated LMW. Prior to packing the carbon
into the column, the BAC was rinsed with DDW to remove excessive turbidity. Since this carbon
was already biologically active, no extensive preloading period was required. For some of the LMW
experiments, a syringe pump (74900 series, Cole Farmer Instrument Co., Vernon Hills, IL) was used
to add an electron donor solution containing 1.2 mM
22

Floating Cover
1
Influent:
LMW
20 (ig/L Bromate
pH7.5
2 mg/L DO
1. 6 mg/L Nitrate
Peristaltic Pump
8.0-cm BAG Bed Height
Support
Material
NOTE: Not to Scale
Syringe Pump
Lactate +
Pyruvate
2. 5 -cm Inner-Diameter
Figure 2.2 Schematic of BAG filters using LMW
pyruvic acid (10 electron equivalents/mole3) and 1.0 mM lactic acid (12 electron equivalents/mole3).
Biologically Active Carbon - Large Filter
A BAG filter was constructed in a 5.1-cm (2-inch) inner-diameter glass pipe (Ace Glass,
Vineland, NJ) and capped with rubber stoppers (Figure 2.3). Virgin Norit RO 0.8 GAG (Norit
Americas Inc., Atlanta, GA) was washed with DDW to remove carbon fines and added to the column.
Three evenly-spaced sampling ports were added to the column by a glass-blower, and four EBCTs
could be monitored in every experiment by using the three sampling ports and the effluent line of the
filter. The filter was run upflow using a peristaltic pump (Masterflex, Cole-Parmer Instrument Co.,
Vernon Flills, IL). Compressed nitrogen was used to sparge the influent water to reach the desired
3 The number of electron equivalents per mole was calculated assuming mineralization of
the organic acid to carbon dioxide.
23

DO concentration (2-6 mg/L). The influent was stored in a headspace-free container, with a floating
cover made of stainless steel, in order to maintain the DO concentration.
The GAC was extensively contacted with CUW in order to exhaust the adsorptive capacity
of the carbon. The GAC was preloaded with 7,200 bed volumes of chlorinated CUW with an influent
DO concentration of approximately 4 mg/L and a total EBCT of 5 minutes. CUW was not
dechlorinated during this portion of the preloading due to the operational difficulty that would have
caused because of the high flowrate (-300 mL/min). Since Kirisits et al. (1998) observed 80 percent
bromate breakthrough from a non-biologically active GAC filter after 6,500 bed volumes, it is
unlikely that significant abiotic bromate removal was taking place in the BAG filter after the
preloading period. In order to stimulate biological growth, the filter was subsequently operated for
500 BV with dechlorinated CUW and 2 mg/L DO. During this time, the filter had a total EBCT of
60 minutes, with the sampling ports allowing the evaluation of 15-, 30-, 45-, and 60-minute EBCTs.
. Floating Cover
Influent:
CUW
20 ng/L Bromate
pH7.5
0.3 mg/L Nitrate
2-6 mg/L DO
Sampling Ports
30.6-cmBAC Bed Height
3-mm
Gass Beads
NOTE: Not to Scale
Peristaltic Pump
5.1-cm Inner-Diameter
Figure 2.3 Schematic of multi-port BAC filter using CUW
24

Backwash ing the BAG filters
Two of the small BAG filters using CUW (Figure 2.1) were backwashed to determine its
effect on bromate removal. The backwash water consisted of filter effluent rather than the usual filter
influent so that the microorganisms would not be exposed to higher ammonia and DOC
concentrations than usual. The backwashing procedure consisted of 10 minutes of fluidization with
approximately 35 percent bed expansion. The carbon was mechanically agitated from the top of the
column to promote mixing in the bed.
Heterotrophic Plate Counts
Heterotrophic plate counts (HPCs) were run with water samples collected before and after
backwashing. Bacto® R2A agar (Difco Laboratories, Detroit, MI) was the medium used. 9.1 g of
R2A was dissolved per 500 mL of DDW. This was heated to approximately 70 °C and stirred for 20
minutes to dissolve the medium. The medium was then sterilized by autoclaving (Eagle/Century
Series, Amsco Scientific) for a 20-minute liquid cycle. 5-mL portions of the medium were poured
into sterile petri dishes and allowed to solidify. The Membrane Filter Method was performed
according to Standard Method 9215D (APHA, AWWA, and WEF, 1995). Various volumes of water
were filtered through sterile, gridded filters (0.45 ^im, Millipore, Bedford, MA). Each filter was
placed in a petri dish and incubated in the dark for a minimum of 5 days at 25 °C.
Denaturing Gradient Gel Electrophoresis
Denaturing gradient gel electrophoresis (DGGE) was used to investigate the diversity of the
microbial population in the BAG filters. The DGGE technique consists of extracting ribosomal DNA
(rDNA) from the microbial community, amplifying it by the polymerase chain reaction (PCR), and
then applying it to a polyacrylamide gel that has a gradient of DNA denaturing agents. As the rDNA
migrates through the gel, it melts at a particular location in the denaturing gradient. The position at
which this melting occurs is a function of the base pairs in the DNA fragment. When the fragment
melts, its motility ceases. Thus, one will obtain a DGGE pattern containing different bands that have
25

separated according to their base pair structure. Each band theoretically represents a unique
microorganism. DGGE gels were analyzed by visual inspection.
Plating Experiments
The purpose of the plating experiments was to isolate individual active bromate-reducers from
the BAC filters operated with CUW. One straightforward and common approach to do this is
through the use of solid medium, which facilitates the isolation of a pure culture. Agar medium
(MR2A), containing essential nutrients, bromate, and nitrate, was prepared for the plating
experiments (Table 2.2). MR2A was selected as an appropriate heterotrophic medium for
environmental isolates and is particularly well-suited for denitrifiers, which have been suggested as
an important group of microorganisms in biological bromate reduction. The MR2A was adjusted to
pH 7.0 and autoclaved. The vitamin 2 solution was filter sterilized, stored at 4 °C, and added to the
sterile agar medium before the plates were poured.
The sterile agar plates were degassed in an anaerobic chamber for a period of one week before
an inoculum was applied. Biomass was obtained from the BAC filters operated with CUW and was
diluted in degassed influent medium in the following ratios: 1:10,000, 1:1,000, and 1:100. 0.1 mL
from each dilution was taken and added to a plate using a sterile syringe. The biomass was spread
over the surface of the agar. The plates were sealed with parafilm and placed into anaerobic jars that
had been purged with argon. These jars were then incubated at room temperature. After 3 weeks,
the plates were observed for growth. There were a number of fast-growing microorganisms which
caused colonies to intrude upon each other. For this reason, some biomass was taken from each plate
and restreaked onto a fresh plate. These plates were incubated for only 5 days before they were
checked for growth. Different colony morphologies were observed, and each different morphology
was streaked onto a fresh plate in order to obtain a pure culture. These plates were incubated in an
argon atmosphere for 41 days.
At this time, some of the pure cultures were harvested. Using a sterile syringe, 0.1 mL of
sterile medium was used to resuspend colonies of pure cultures to use as inocula for further testing
in aqueous medium containing bromate.
26

Table 2.2 Agar recipe for plating experiments
MR2A Constituent
KH2PO4
K2HP04
CaCl2 * 2H2O
MgCl2 * 6H2O
FeSO4 * 7H20
NH4C1
Trace metals 1
Vitamin 2
Yeast extract
Peptone
Casamino acids
Sodium pyruvate
Bacto agar
Br03'(ng/L)
NO3- (mM)
DDW
1 Trace metals (500X)
MnCl2 * 4H2O
H3B03
ZnCl2
CoCl2 * 6H2O
NiSO4 * 6H2O
CuCl2 * 2H2O
NaMoO2 * 2H2O
DDW
2 Vitamin solution (1000X)
B 12
Biotin
Calcium pantothenate
Folic acid
Nicotinamide
Pyridoxine HC1
Riboflavin
Thiamine HC1
DDW
(g/L)
0.25
0.40
0.015
0.02
0.007
0.005
0.8
2mL
ImL
0.5
0.5
0.5
0.3
15.0
100
1000
1.0
1000 mL
g/lOOmL
0.25
0.025
0.025
0.025
0.025
0.015
0.0005
100 mL
mg/L
1.0
20.0
50.0
20.0
50.0
100.0
50.0
50.0
1000 mL
27

Three different media were used. The first aqueous medium consisted of CUW, 28.7 ug/L
bromate, 4.8 mg/L nitrate, and 0 mg/L DO at pH 7.8 with 2.5 mM phosphate buffer. The second
aqueous medium was DDW, 43.5 ug/L bromate, 4.9 mg/L nitrate, 8.3 mg/L DOC (0.10 mM acetate,
0.08 mM pyruvate, and 0.07 mM lactate), and 0 mg/L DO at pH 7.5 with 5.0 mM phosphate buffer.
The third aqueous medium consisted of MR2A broth, supplemented with higher-than-usual
concentrations of bromate and/or nitrate. This medium was amended with one of the following: 1
mM nitrate, 45 uM nitrate plus 70 uM bromate, or 70 uM bromate; the cultures in these media were
monitored for visible growth and gas production (N2O). Following incubation of the aqueous
medium experiments, samples were taken for bromate, bromide, nitrite, and nitrate analyses. Figure
2.4 schematically shows the progression of the plating and aqueous medium experiments.
The pure cultures were also tested for facultative growth on MR2 A agar by incubation under
aerobic conditions.
Dilution-to-Extinction Experiments
In addition to the traditional plating technique, another way to isolate microorganisms is
through dilution-to-extinction experiments (progressive dilutions). Two inocula were used for these
experiments. The first inoculum consisted of effluent water collected from the second BAG filter in
series operated with an influent nitrate concentration of 5 mg/L in CUW. The second inoculum
consisted of carbon and water taken from the effluent side of the second BAG filter in series operated
with an influent nitrate concentration of 0.3 mg/L in CUW. Medium for the experiments consisted
of the following: dechlorinated CUW, 5 mg/L nitrate, 30 pgfL bromate, 2.5 mM phosphate buffer.
15 mL of medium were added to each Balch tube. Butyl rubber stoppers were inserted and aluminum
caps were crimped in place. The tubes were autoclaved, cooled, and then inoculated with the
appropriate biomass. Each experimental condition was set up in triplicate. The tubes were then
monitored for bromate, bromide, nitrate, and nitrite. Active dilution samples can be inoculated onto
agar medium for isolation and testing. However, since 8 bromate-reducing isolates have been
cultured using the traditional plate technique, the dilution-to-extinction samples have not yet been
used to isolate additional bromate-reducers.
28

Streaked individual colony morphologies
to get pure isolate (Incubated 41 days)
Pure isolates inoculated to aqueous media
Biomass taken from
BAG filter
A\
3 Dilutions prepared
1:10,000 1:1,000 1:100
J T \ Plates streaked (Incubated 3 weeks)
Confluent growth; plates restreaked
(Incubated 5 days)
Aqueous media: CUW, DDW, M2RA
Bottles sampled to monitor bromate/nitrate reduction
I
Active bromate-reducers will be characterized
by 16S rDNA sequencing and nutritional requirements
Figure 2.4 Progression of plating and aqueous medium experiments with pure cultures

Serum Bottle Experiments
Since the column filtration experiments are labor and time intensive, it is useful to set up batch
biological experiments to test the dependence of bromate reduction on various parameters. 160 mL
serum bottles, equipped with gas-tight butyl rubber stoppers, were used as the batch reactors. Each
bottle contained 140 mL of aqueous medium, which was autoclaved after the bottles were capped.
After the bottles were cooled, biomass from the BAG filters was added.
The medium consisted of the following: CUW, 5 mg/L nitrate, 5 mg/L sulfate, 19 ug/L
bromate, and 1 mM phosphate buffer. Biomass was taken from the BAG filter operated with an
influent nitrate concentration of 5.0 mg/L in CUW, and 1 mL of this inoculum was spiked to each
serum bottle. The serum bottle experiments were used to evaluate the effect of temperature (4 and
25 °C) on bromate reduction.
MATERIALS USED FOR PERCHLORATE REDUCTION EXPERIMENTS
Water
Two types of water were used in this study - DDW and CUW. Typical properties of CUW
are listed in Table 2.1.
Reagents and Carbon
The chemicals used included reagent grade sodium perchlorate (Sigma Chemical Co., St.
Louis, MO) and sodium bromate (Aldrich Chemical Co., Inc., Milwaukee, WI). Other chemicals
were of reagent grade quality or better. Where appropriate, the chemicals were dried overnight at
105 °C and stored in a desiccator. Norit RO 0.8 GAG, lot number 72303-5 (Norit Americas Inc.,
Atlanta, GA), was used in the BAG filters. This carbon is an extruded peat activated by steam. The
30x40 fraction of Calgon F-400 (Calgon Inc., Pittsburgh, PA), lot number 5925-LG, was used for
some kinetic studies. The F-400 is made from bituminous coal.
30

METHODS UTILIZED FOR PERCHLORATE REDUCTION EXPERIMENTS
Biologically Active Carbon Filters
Four filters were constructed to evaluate perchlorate removal. Each filter contained fresh
carbon. BAG was formed by extensively preloading the filters with effluent from the previously
described bromate reducing filters. The influent was spiked with 50 ng/L perchlorate and 20 ug/L
bromate. The columns were operated with a 25-minute EBCT for 600 bed volumes. At the end of
the preloading period, approximately 60 percent of the influent DOC and 90 percent of the influent
bromate were consistently being removed. Therefore, by the end of the preloading period, the filters
were considered biologically active. Two parallel filtration experiments were set up, each with two
columns in series, for a total of four filters. The filtration experiments were often run with two filters
in series, so that two EBCTs could be evaluated simultaneously. Once the filters were rendered
biologically active, the influent base water for one set of filters was switched from bromate filter
effluent to DDW. This allowed for better control over influent composition. The other set of filters
utilized CUW as the base water for approximately one year. However, since perchlorate removal was
not consistent in these filters due to the occurrence of nitrification, the data from these filters are not
reported here. The schematic for the biological bromate filters shown in Figure 2.3 also serves as the
schematic for the biological perchlorate filters.
Syringe pumps (kd Scientific, Boston, Massachusetts) add an electron donor mixture to the
influent at a point shortly before the influent enters the first filter. The electron donor solution was
either a mixture of acetate, lactate, and pyruvate or a mixture of acetate, benzoate, and pyruvate.
Benzoate was sometimes substituted for lactate due to the fact that carboxylic acids are common
ozonation by-products (Camel and Bermond 1998). Since the focus of this project is conventional
ozone/GAC systems, the use of benzoate was very practical. The electron donor concentration was
always 1.5 mg/L of equivalent nitrate demand4 in excess of that required to stoichiometrically remove
the influent concentrations of DO and nitrate. Thus, electron donor concentration was never the
limiting variable.
4The equivalent nitrate demand was determined by calculating the concentration of
electron donor required to stoichiometrically reduce NO3" completely to N2 .
31

Influent was prepared by adding 50 ug/L of perchlorate, the appropriate nitrate concentration,
and 1 mM phosphate buffer to DDW. Flowrates through the BAG filters were adjusted to provide
the desired EBCT. Samples were taken at the influent, at the effluent of the first filter, and at the
effluent of the second filter. Samples were stored at 4 °C until they were analyzed.
Abiotic Batch Experiments
Using ajar test apparatus (B-KER2, Model 7790-400, Phipps and Bird, Richmond, Virginia),
a series of batch tests were run to determine if chemical reduction of perchlorate could be achieved
using GAC under various conditions. Four different GACs were used: 1) virgin Norit, 2) virgin
Calgon F-400,3) acid-washed outgassed (AWOG) Norit, and 4) AWOGCalgon F-400. The AWOG
carbon was prepared by Kirisits (2000). AWOG carbon typically contains lower concentrations of
surface oxygen groups and has a higher pH (Miller 1996). The mean pellet diameter and length for
the Norit carbon were 0.8 and 3.4 mm, respectively. The mean particle diameter for the Calgon F-
400 carbon was approximately 0.7 mm. The carbon to be used in the batch studies was weighed and
stored in DDW overnight to wet it. Two liters of DDW or CUW was placed in each jar for the batch
test studies. The pH of each solution was adjusted, and 1 mM phosphate buffer was added to prevent
pH fluctuation. 50 ug/L of perchlorate was spiked into the batch test jars. A total of 9 different
experiments were run. The behavior of each of the four carbons was examined in both CUW and
DDW at pH 7.5. The ninth experiment used DDW, virgin Norit carbon and pH 2. Once the nine
batch test solutions were appropriately prepared, 500 mg/L of GAC was added to each jar, and the
stirrers were turned on to suspend the GAC. The experiments were run for 5 hours and were
sampled regularly during this time.
Batch Test Wash Experiments
Upon completion of the batch test experiments, the GAC from the four CUW experiments
was removed and placed in separate glass columns equipped with stopcocks and glass wool to
support the carbon. Solutions containing a nitrate concentration of 200 mg/L were used to wash the
GAC. The effluent from the columns were collected in known volumes and analyzed for perchlorate.
The GAC was washed until no perchlorate was detected in the effluent.
32

Metal-Catalyzed GAG Filter Experiments
The two filters constructed for the abiotic, metal-catalyzed experiments were similar to those
used for the BAG filter experiments. One filter was composed of Calgon copper- and zinc oxideimpregnated
carbon. The other filter contained virgin Norit RO 0.8 carbon, to which aluminum shot
was added (10% w/w). The aluminum was mixed thoroughly with the GAC, and the mixture was
added to the column. Influent was made with the following parameters: 1) DDW, 2) DO = ambient
(-8.6 mg/L), 3) pH 2.5, 4) ClO/^, = 3 mg/L, NO3-influent = 0.1 mg/L (ambient). A 3 mg/L influent
perchlorate concentration was used so that reduction products could be monitored. A 25-minute
EBCT was used for each filter, and samples were taken at the influent and effluent of each column.
Metal-Catalyzed GAC Filter Wash Experiments
At the end of the aforementioned experiment, the GAC was rinsed with a 10 mg/L nitrate
solution to wash off any perchlorate or chloride that was ion-exchanged at the carbon surface. This
procedure would help determine if ion exchange or chemical reduction was the mechanism of
perchlorate removal. The effluent from these tests was collected and analyzed for perchlorate.
Rinsing continued until no perchlorate was detected in the effluent. The mass of perchlorate desorbed
was calculated and compared to the mass of perchlorate removed by each filter. Once it was
determined that not all the perchlorate had been recovered after the 10 mg/L nitrate rinse, a 200 mg/L
nitrate rinse was run until no perchlorate was detected in the effluent.
Ozone/Hydrogen Peroxide/Perchlorate Batch Tests
Six batch experiments were set up to examine the effect of ozone and hydrogen peroxide on
perchlorate reduction. The same jar test apparatus that was used in other batch tests was utilized for
these experiments. Table 2.3 describes the experimental conditions for these experiments.
Gaseous ozone was generated by a model GL-1 ozone generator (PCI Ozone Corporation,
West Caldwell, New Jersey) from pure oxygen. Ozone was applied to the batch solutions as gas and
concentrations were determined according to the Indigo Colorimetric Method (4500-O3) described
in Standard Methods for the Examination of Water and Wastewater, 19* Edition (1995).
33

Table 2.3 Experimental conditions: ozone/hydrogen peroxide/perchlorate batch tests
Experiment #
1
2
3
4
5
6
Initial
Perchlorate
46.6
46.5
46.2
43.2
43.5
*
42.8
pH
8.8
8.8
8.8
8.8
8.8
8 .8
Ozone:DOC
(mg/mg)
2
2
2
3
3
3
Ozone (mg/L)
2.94
2.94
2.94
4.41
4.41
4.41
H2O2 : Ozone
(mg/mg)
0.0
0.5
1.0
0.0
0.5
1 .0
Each jar was prepared by adding appropriate concentrations of perchlorate and hydrogen
peroxide to 2 liters of CUW. The appropriate amount of ozone from the batch solution was then
added to the CUW, and samples were taken at time = 0, 5,10, and 15 minutes. Samples were filtered
(0.45 |am) prior to storage at 4 °C and analysis. No solutions or samples were quenched.
Denaturing Gradient Gel Electrophoresis
DGGE was used to investigate the diversity of the microbial population in the BAC filters.
The same DGGE technique used for the bromate samples (described earlier) was used for the
perchlorate samples.
Plating Experiments
The solid medium plating technique used to isolate perchlorate-reducing bacteria was the
same as that used to isolate bromate-reducing bacteria (described earlier). To check for perchloratereducing
ability, microorganisms isolated from the BAC filters were inoculated into an aqueous
medium containing 50 |ig/L perchlorate, 20 |ig/L bromate, 1.0 mg/L nitrate, 5 mM phosphate buffer,
and 0 mg/L DO at pH 7.5. Electron donor was supplied at 1.5 mg/L of equivalent nitrate demand
in excess of that required to stoichiometrically remove 1.0 mg/L nitrate. Figure 2.4 schematically
shows the progression of these plating experiments. Following incubation, samples were taken for
34

perchlorate, nitrite, nitrate, and bromate analyses. To verify the ability of pure cultures to reduce
perchlorate, transfers were made to fresh medium and monitored. Individual isolates will be further
characterized once they are confirmed active for bromate and perchlorate reduction.
Dilution-to-Extinction Experiments
The technique used for the perchlorate dilution-to-extinction experiments was the same as that
used for the bromate dilution-to-extinction experiments (described earlier). The inoculum used for
these experiments was effluent water from the BAG filter. The medium consisted of DDW, 50 ug/L
perchlorate, 20 ng/L bromate, 1.0 mg/L nitrate, and 5 mM phosphate buffer. The electron donor
concentration, comprised of acetate, lactate, and pyruvate, was 1.5 mg/L of equivalent nitrate demand
in excess of that required to stoichiometrically reduce 1.0 mg/L nitrate. Each experimental condition
was run in triplicate.
Serum Bottle Experiments
Serum bottle experiments were run as screening tests for the effect of various parameters on
perchlorate reduction. 160 mL serum bottles, equipped with gas-tight, butyl rubber stoppers, were
used as batch reactors. All serum bottle experiments were performed using a single microbial species
isolated during the plating experiments, described in the previous section. This isolate, B9,
demonstrated the ability to reduce 50 (ig/L perchlorate to below the detection limit after being
inoculated to aqueous medium.
Two different serum bottle experiments were run using isolate B9. Medium for the first
experiment was prepared according to Table 2.4. This experiment was designed to compare several
different operating conditions: 1) nitrate concentration (0.0, 0.3, and 1.5 mg/L), 2) DDW amended
with external electron donor added versus CUW with no external electron donor, 3) bromate
concentration (0, 20 ug/L), 4) lactate versus benzoate as an electron donor. The results of this
experiment led to the design of the second serum bottle experiment, as described in Table 2.5. This
experiment focused on the difference in perchlorate reduction kinetics when the initial nitrate
concentration was varied.
Using 5 mM phosphate buffer, each medium was maintained at pH 7.5. 150 mL of each
35

medium was placed into a serum bottle and sparged with argon gas for 30 minutes to remove the DO.
Butyl stoppers were inserted and aluminum caps were crimped in place. The bottles were autoclaved
and then cooled. To produce a uniform inoculum among serum bottles, a stock solution of biomass
was first prepared. For this solution, 5 mL of DDW was placed into a serum bottle, stoppered,
capped, and then autoclaved. 0.5 mL of sterile DDW was aspirated on the plate containing the B9
isolate to suspend the colonies. This 0.5 mL volume, containing the microbial suspension, was added
to the other 4.5 mL sterile DDW. The bottle was then vortexed for several minutes to create a
uniform microbial suspension. 0.1 mL of this suspension was spiked to each serum bottle. Serum
bottles with no inoculum were also prepared as controls for each medium. Bottles were manually
shaken once per day, and samples were taken once, twice, or three times daily. Each serum bottle
was run in duplicate.
Table 2.4 Serum bottle experiment 1 using perchlorate-reducing isolate B9 (initial conditions)
Serum
bottle
#
1
2
3
4
5
6
7
8
9
10
Water
DDW
DDW
DDW
DDW
DDW
DDW
DDW
DDW
DDW
CUW
(dechlorinated)
50
50
50
50
50
50
50
50
50
50
cio4-
NO3-
(mg/L)
0.0
0.3
1.5
0.0
0.3
1.5
0.0
0.3
1.5
0.0
Calculated concentration
(mg/L as C)
and type of electron donor
2.17, acetate/lactate/pyruvate
2.17, acetate/lactate/pyruvate
2.17, acetate/lactate/pyruvate
2.17, acetate/benzoate/pyruvate
2.17, acetate/benzoate/pyruvate
2.17, acetate/benzoate/pyruvate
2.17, acetate/benzoate/pyruvate
2.17, acetate/benzoate/pyruvate
2.17, acetate/benzoate/pyruvate
No External Donor Added
^
20
20
20
20
20
20
0
0
0
20
11 CUW 50 0.3
(dechlorinated)
12 CUW 50 1.5
(dechlorinated)
No External Donor Added
No External Donor Added
20
20
36

Table 2.5 Serum bottle experiment 2 using perchlorate-reducing isolate B9 (initial conditions)
Serum
bottle #
Water
cio4-
(Hg/L)
N03-
(mg/L)
Calculated concentration (mg/L as C)
and type of electron donor
1
DDW
50
0.1
2.17, acetate/benzoate/pyruvate
2
DDW
50
0.4
2.17, acetate/benzoate/pyruvate
3
DDW
50
1.6
2.17, acetate/benzoate/pyruvate
4
DDW
50
.2.6
2.17, acetate/benzoate/pyruvate
ANALYTICAL METHODS
Bromate, nitrite, bromide, chloride, nitrate, and sulfate were analyzed by ion chromatography.
The ion chromatography system consisted of a Dionex Series 300 ion chromatograph (Sunnyvale,
CA) with an anion self-regenerating suppressor (ASRS-1) and conductivity meter. Three columns
were used in series - Dionex lonpac NG1, AG9-HC and AS9-HC. The purpose of the NGl was to
adsorb organic material that could foul the analytical column. The analytical method included a 9.0
mM sodium carbonate eluant, 1.0 mL/min flowrate, and 250 uL injection loop. Perchlorate was also
analyzed by ion chromatography. Three columns were used in series for perchlorate analysis -Dionex
lonpac NGl, AG11, and AS 11. The analytical method for perchlorate included a 100 mM sodium
hydroxide eluant, 1.0 mL/min flowrate, and a 250 uL injection loop. All anions were detected by
suppressed conductivity.
The LOD for bromate and perchlorate were calculated by the method outlined by Skoog and
Leary (1992) to be approximately 2 ug/L. Any bromate or perchlorate concentration that appeared
to be lower than 2 ug/L was recorded as being at one half of the LOD. Other studies have shown
that this simple method of handling concentrations below the LOD is as good or better than more
complicated mathematical methods (Clarke 1998).
DO was measured using a YSI Model 58 DO meter with a 5905 probe (Yellow Springs, OH).
An Orion 720A pH/Ion Selective Electrode Meter with an Orion ammonia probe 95-12 or pH probe
(Beverly, MA) was used to measure ammonia and pH, respectively. DOC was analyzed with a
Phoenix 8000 TOC Analyzer (Tekmar-Dohrmann, Cincinnati, OH) by UV-persulfate oxidation.
The method detection limits were as follows: bromate = 2 ug/L, bromide

perchlorate = 2 ug/L, chloride

CHAPTER 3. RESULTS AND DISCUSSION
BROMATE
Verification of Biological Bromate Removal in the BAC Filters
Since GAC can abiotically reduce bromate and can also serve as the attachment medium for
microorganisms that use bromate as an electron acceptor, there exists the question of whether both
processes are occurring simultaneously in the BAC filters. Due to the fact that the abiotic bromate
reduction capacity of GAC is quickly exhausted in natural water (Kirisits et al. 1998), it seemed
highly unlikely that the BAC filters were simultaneously supporting abiotic and biotic bromate
reduction. However, to gather further evidence for the hypothesis that most of the observed bromate
reduction in the BAC filters was indeed biological, repeat bromate removal experiments were run to
verify the percent bromate removals obtained with the same BAC filters approximately one year ago.
The 10- and 20-minute EBCT BAC experiments first run in March 1998 were repeated in February
1999 with the same BAC filters; if the percent bromate removals remained the same, this would be
an indication that bromate removal in the BAC filters was biological since it is doubtful that abiotic
bromate reduction was still occurring 1.5 years after the filters were first constructed. The 10- and
20-minute EBCT BAC filtration experiments were repeated for influent nitrate concentrations of 5.0
and 0.3 mg/L.
Figure 3.1 shows the results for the BAC experiment with an influent nitrate concentration
of 5.0 mg/L, and it demonstrates that the percent bromate removal did not change after an additional
year of operation. Thus, only biological removal was occurring in both March 1998 and February
1999. However, when the same comparison was performed for the BAC experiment with an influent
nitrate concentration of 0.3 mg/L, the percent removal in February 1999 did not correspond to the
percent removal in March 1998 (Figure 3.2). This is not interpreted to mean that abiotic bromate
reduction was occurring in the experiments with a 0.3 mg/L influent nitrate concentration in March
1998, which would have artificially inflated the percent bromate removals. Rather, these results are
evidence of a microbial disturbance that occurred in these filters in December 1998. In December
1998, the EBCT in the BAC filters with an influent nitrate concentration of 0.3 mg/L was increased
to approximately 115 minutes so that they could run with a single batch of influent for
39

100
80 -
• March 1998
D February 1999
5 60 H
S
40 -
20 -
0
0 10 15
EBCT (min)
20 25
Figure 3.1 Repeat of the 10- and 20-minute EBCT filtration experiment (average influent
conditions: CUW, pH 7.5, 2 mg/L DO, 5.0 mg/L NO3', 20 ug/L BrO3')
100
80
• March 1998
D February 1999
£ 60

a two-week period. Over the two-week period, the flowrate discharged by the peristaltic pump
decreased so that the EBCT was well over 200 minutes. It is possible that the extended contact time
in the filter resulted in much lower DOC concentrations reaching the lower portions of the filter,
causing that microbial community to die or to change composition. This microbial disturbance had
apparently not been rectified by the time the verification study was run in February 1999. Even
though these filters have been operated for an additional year since the disturbance, they have been
unable to reproduce the bromate removals originally observed in March 1998. This indicates that
filter history has a very important effect on biological bromate removal. The microbial community
that is developed under a particular water quality condition might affect the bromate removal obtained
even after the water quality conditions become more favorable for bromate removal.
To further investigate the performance of the BAG filters that had undergone a disturbance
(Figure 3.2), the influent nitrate concentration to these filters was switched to 5.0 mg/L (Figure 3.3).
In Figure 3.3, the series entitled "Original 5.0 mg/L influent nitrate filters" refers to the BAC filters
that have been operated with an influent nitrate concentration of 5.0 mg/L for two years; the series
entitled "Filters switched from 0.3 to 5.0 mg/L influent nitrate" refers to the BAC filters that were
operated for 2 years with an influent nitrate concentration of 0.3 mg/L but were recently switched

to an influent nitrate concentration of 5.0 mg/L. There is no statistically significant difference 1 in
bromate removal between the two sets of filters (Figure 3.3). These data could indicate a difference
in the microbial communities that were first developed in the two sets of filters. That is, when the
BAG filters operated with an influent nitrate concentration of 0.3 mg/L were first run, a nitratesensitive
microbial community may have developed. This community may have washed out during
the two-week disturbance in December 1998 and has not been able to re-establish itself since that
time.
Further Testing the Reproducibility of BAG Filters
To test the reproducibility of the data obtained with the BAG filter operated with a 0.3 mg/L
influent nitrate concentration, a new BAG filter was developed. This BAG filter was constructed in
a larger column (2-inch inner diameter, Figure 2.3) as compared to the smaller columns used for other
BAG filtration experiments (1-inch inner diameter, Figure 2.1). Due to the multi-port configuration
of this filter (Figure 2.3), EBCTs of 15, 30, 45, and 60 minutes were simultaneously tested. Figure
3.4 compares the bromate removal results from the 1 -inch inner diameter filter (prior to the December
1998 disturbance) to the results from the 2-inch inner diameter filter and demonstrates the
reproducibility of the data.
Since Figure 3.4 demonstrates that a one-inch inner diameter BAG filter produces the same
results as a 2-inch inner diameter BAG filter, this suggests that wall effects are minimal. This is a
useful result from an operational standpoint because the smaller filter requires preparation of smaller
volumes of influent water.
Backwashing BAG Filters
Since the filters utilized for bromate removal did not accumulate appreciable head loss (as
evidenced by a constant flow rate across the filter for a given pumping rate), they were not routinely
backwashed. However, the question arose as to whether bromate removal would decrease due to
backwashing.
'Statistical significance was based on overlap of 95 percent confidence intervals.
42

To establish the effect of backwashing on filter performance, two filters operating with
CUW, a 25-minute EBCT, and an influent nitrate concentration of 0.3 or 5.0 mg/L were chosen for
backwashing. Prior to backwashing, the filters were run to obtain the pre-backwashing bromate
removal percentages. Then the filters were backwashed and run again to obtain the postbackwashing
bromate removal percentages.
As shown in Table 3.1, backwashing did not negatively impact the bromate removal rate.
It should be noted that the backwashing procedure removed a substantial amount of microbial
material from the filters; the backwash water was visually, turbid and contained floes.
100
^ 80
o 60

Samples were collected for heterotrophic plate counts for the following waters: influent
water, pre-backwash filter effluent, post-backwash filter effluent, and the backwash water. Plate
counts were performed according to the membrane filter method. However, the volume of water
filtered through the membranes resulted in extensive microbial growth, and individual colonies were
too numerous to count (TNTC).
Effect of the Initial Bromate Concentration
The mass concentration of bromate removed by the BAG filters was monitored as the influent
bromate concentrations increased from 10 to 40 ug/L with an influent nitrate concentration of 0.3
mg/L (Figure 3.5) and from 10 to 50 ug/L with an influent nitrate concentration of 5.0 mg/L (Figure
3.6). Figure 3.5 shows constant mass concentration bromate removal as the bromate concentration
was increased from 20 to 40 ug/L in the 10- and 20-minute EBCT filters which indicates that the
microorganisms in the BAG filters were not able to reduce additional bromate at the 10- and 20-
minute EBCTs as the influent bromate concentration increased. One may have expected that an
increase in the influent bromate concentration would have increased the driving force for bromate in
the bulk solution to diffuse deeper within the biofilm, thereby increasing the mass of bromate
removed. However, it seems that bromate reduction under these conditions (10- and 20- minute
EBCTs and an influent bromate concentration between 20-40 ug/L) is limited by the reaction rate,
which may indicate that the electron donor concentration is the limiting factor. Figure 3.5 shows that
the mass concentration of bromate removed increased as the influent bromate concentration increased
from 10 to 20 ug/L in the 25- and 50-minute EBCT filters, indicating that the electron donor
concentration was not limiting under these conditions.
Similar experiments were carried out in the BAG filters with an influent nitrate concentration
of 5.0 mg/L. Figure 3.6 shows a general increase in mass concentration bromate removal as the
influent bromate concentration was increased from 10 to 50 ug/L for 10- and 20-minute EBCTs.
However, as the influent bromate concentration was increased from 40 to 50 ug/L, the data points
show that constant mass concentration bromate removal was observed in the 10- and 20-minute
EBCT filters, at the 95 percent confidence level. This indicates that the microorganisms in the BAG
filters were not able to reduce additional bromate at the 10- and 20-minute EBCTs when the influent
44

o
S
25
20 -
15 -
10 -
O 50-min EBCT
• 25-min EBCT
D 20-rrdn EBCT
• 10-minEBCT
I 2
ffl
5 -
0
0 10 20 30 40 50
Influent Bromate Concentration (u,g/L)
Figure 3.5 Mass concentration bromate removal with different influent bromate
concentrations (average influent conditions: CUW, pH 7.5, 2 mg/L DO, 0.3 mg/L NO3")
20 -T
15 -
10 -
o 50-min EBCT
• 25-min EBCT
D 20-min EBCT
• 10-minEBCT
2
CQ
5 -
10 20 30 40
Influent Bromate Concentration (|j,g/L)
50 60
Figure 3.6 Mass concentration bromate removal with different influent bromate concentrations
(average influent conditions: CUW, pH 7.5, 2 mg/L DO, 5.0 mg/L NO3")
bromate concentration was above 40 ug/L. As discussed previously, this may indicate that the
electron donor concentration is limiting bromate removal.
Figure 3.6 also shows that the mass concentration of bromate removed increased as the
influent bromate concentration increased from 10 to 20 ug/L in the 25- and 50-minute EBCT filters,
45

indicating that the electron donor concentration was not limiting under these conditions. Hijnen et
al. (1999) also observed an increase in bromate removal in a denitrifying bioreactor as the influent
bromate concentration increased. Since ethanol was fed to the bioreactor, the concentration of
electron donor did not limit bromate removal in that study.
Effect of the Influent Dissolved Oxygen Concentration
To evaluate the effect of DO concentration on bromate removal, the BAG filters were
•operated with CUW and varying DO concentrations (Table 3.2). The influent nitrate concentration
was relatively constant for all experiments (4.6 -5.1 mg/L), although the nitrate concentration in the
filters increased due to nitrification of the influent ammonia (Table 3.2). The data in Table 3.2 show
that the effluent DO concentration was below the detection limit (0.1 mg/L) when the influent DO
concentration was less than or equal to 5.8 mg/L. When the influent DO concentration was raised
to 8.0 mg/L and higher, the effluent DO concentration was above the detection limit.
Table 3.2 The effect of influent DO concentration on bromate removal in BAC filters
(average influent conditions: CUW, pH 7.5, 20 ug/L BrO3")
Average DO Average nitrate
(mg/L)
Expt. EBCT Influent Effluent
Number (min)
Influent Effluent Average bromate removal
with 95% confidence limits
1 10
20
2.1

In general, the average percent bromate removal decreased as the influent DO concentration
increased (Table 3.2 and Figure 3.7). This was expected since oxygen, a common electron acceptor,
may inhibit the expression or activity of the enzymes required for bromate reduction. Additionally,
the increased DO concentration could cause more competition for electron donor. Note that the
average percent bromate removals cannot be considered statistically different at the 95 percent
confidence level for experiments with influent DO concentrations between 3.8 and 13.6 mg/L.
However, there is a statistically significant decrease in bromate removal, at the 95 percent confidence
level, as the influent DO concentration was increased from 2.1 to 3.8 mg/L.
The average percent bromate removal observed in the experiment with an influent DO
concentration of 13.6 mg/L was only 11 percent after a 21-minute EBCT (Table 3.2). Even though
nitrification and aerobic respiration consumed 8.5 mg/L DO during the 21-minute EBCT, the
remaining DO (> 5.1 mg/L throughout the filter) prevented significant bromate removal. Bromate
removal may also have been prevented by electron donor limitation due to electron donor
consumption by aerobic respiration. Therefore, it is likely that full-scale ozone-B AC plants are not
observing biological bromate reduction in their BAG filters because the DO concentration is too high.
Treatment systems that generate ozone from pure oxygen may have influent DO concentrations
o
50
40 -
30 -
20 -
• 2.0 mg/L influent DO
D 3.9 mg/L influent DO
• 5.9 mg/L influent DO
• 8.2 mg/L influent DO
X 13.8 mg/L influent DO
2 10 H
CQ
0
0 10 15
20 25
EBCT (min)
Figure 3.7 Percent bromate removal with various influent DO concentrations (average
influent conditions: CUW, pH 7.5, 20 ug/L Br03", 5 mg/L NO3')
47

greater than 10 mg/L. For example, in the full-scale, post-ozonation BAG filters at the Paul M. Neal
Water Treatment Facility (Lake Bluff, IL), bromate is not reduced from the influent concentration
(less than 6 ug/L) after a 15-minute EBCT (Soucie 1999); this is likely due to the fact that the
average influent and effluent DO concentrations are 11.5 and 10.7 mg/L, respectively.
Tiedje (1988) noted that the expression and activity of the enzymes required for
denitrification are often affected by the DO concentration. While denitrification is primarily thought
of as an anoxic process, denitrification has also been observed under aerobic conditions (Robertson
and Kuenen 1984). In Thiosphaera pantotropha (now Paracoccus denitrificans [Ludwig et al.
1993]), Bell et al. (1990) found that the nitrate reductases involved under aerobic conditions are
different from those used under anaerobic conditions. A membrane bound nitrate reductase is
primarily expressed under anaerobic conditions; a periplasmic nitrate reductase is primarily
expressed under aerobic conditions but is active under aerobic and anaerobic conditions. Ye et al.
(1994) noted that aerobic denitrifiers, such as Paracoccus denitrificans, are rare organisms.
However Lloyd et al. (1987) demonstrated aerobic denitrification with eight bacterial isolates and
concluded that aerobic denitrification may be common in the environment. Bell et al. (1990)
observed that the aerobic rate of nitrate reduction was approximately one tenth of the anaerobic rate
of nitrate reduction; thus, the slower rate of aerobic denitrification may make aerobic denitrifiers
more difficult to identify. It is possible that aerobic denitrifiers are present in the BAG filters used
in the current study since bromate reduction was observed in the presence of DO (Experiments 4 and
5 in Table 3.2) for 36 days. The low percent bromate removal observed under aerobic conditions
would be consistent with the slow rate of aerobic denitrification observed by Bell et al. (1990).
The effluent nitrate concentration increased as the influent DO concentration increased due
to biological nitrification of ammonia (Table 3.2). However, note that in Experiments 4 and 5, the
average effluent nitrate from the 21-minute EBCT filter was the same as or close to the effluent
concentration from the 10-minute EBCT filter. This indicates that all of the nitrification took place
in the first (10-minute EBCT) filter for these experiments.
In general, the measured DO consumption (Table 3.2) was greater than the calculated
consumption of DO required for the observed nitrate production. This indicates that some of the DO
was also used in aerobic respiration. Also, the ammonia consumption can account for the observed
increase in nitrate concentration. On average, there was only a 6 percent difference between the
48

measured nitrate concentration and the calculated nitrate concentration, based on ammonia
consumption (data not shown).
There is no evidence that significant denitrification took place in the filters, since the influent
nitrogen (ammonia, nitrite, and nitrate) was equal to the effluent nitrogen (data not shown). Based
on the results of Hijnen et al. (1999), one might expect a faster rate of nitrate reduction as compared
to bromate reduction since the nitrate concentration was two orders of magnitude greater than the
bromate concentration in the BAG filters. Due to the low concentration of biodegradable organic
matter in drinking water, it is likely that microbial activity - including denitrification - was limited
by the electron donor concentration. Additionally, a low concentration of electron donor can result
in measurable effluent DO (Experiments 4 and 5, Table 3.2) which can also inhibit the enzymes
required for denitrification. These points are illustrated in a later section (see - Effect of Source
Water Type) where significant nitrate removal was observed upon addition of an external electron
donor to a drinking water source.
Effect of the Influent Nitrate Concentration
Kirisits and Snoeyink (1999) showed a decrease in bromate removal as the influent nitrate
concentration increased over a relatively narrow concentration range (0.3 to 5.0 mg/L). Figures 3.8
and 3,9 expand this result to show how bromate removal decreased as the influent nitrate
concentration increased from 0.3 to 42.3 mg/L. With a 26-minute EBCT, bromate removal dropped
from 86 to 49 percent as the influent nitrate concentration increased from 0.3 to 42.3 mg/L (Figure
3.8). It should be noted that the 86 percent removal performance was sustainable (data point covers
45 days of flow). With a 51-minute EBCT, bromate removal decreased from 95 to 79 percent for
the same increase in nitrate concentration (Figure 3.9). The change in bromate removal was
seemingly smaller for the 51-minute EBCT than for the 26-minute EBCT (16 percent decrease
versus 37 percent decrease) because of the first two points in Figure 3.9. These points show 95
percent bromate removal, which is removal to the detection limit. In these experiments, it is likely
that the BAG filters could have removed more than the 20 |ag/L bromate present in the influent.
Therefore, the 51-minute EBCT seemed to dampen the negative effect of an increased nitrate
concentration due to bromate limitation for the experiments with influent nitrate concentrations of
0.3 and 5.0 mg/L.
49

ffl
10 - Immediately preceding these experiments, the filters were
exposed to an influent DO concentration of 13.6 mg/L DO for 2 weeks
0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0
40.0 45.0
Influent Nitrate Concentration (mg/L)
Figure 3.8 Effect of influent nitrate concentration on bromate removal in BAG filters
(average influent conditions: CUW, pH 7.5, 20 ng/L BrO3', 2.1 mg/L DO,
26-minute EBCT)
£
13
o
s
O
90
80 H
70
60
50 H
40
30 -
20 -
10 -
0
0.0
Immediately preceding these experiments, the filters were
exposed to an influent DO concentration of 13.6 mg/L for 2 weeks
—I————————,———————!———————!————————!————————!————————,————————!——————
5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 45.0
Influent Nitrate Concentration (mg/L)
Figure 3.9 Effect of influent nitrate concentration on bromate removal in BAG filters
(average influent conditions: CUW, pH 7.5, 20 ug/L BrtV, 2.1 mg/L DO,
51 -minute EBCT)
50

Two data points in each of Figures 3.8 and 3.9 are denoted as outliers (influent nitrate
concentrations of 10.9 and 32.1 mg/L). These experiments were run with a different set of BAG
filters than the rest of the experiments shown in Figures 3.8 and 3.9. Prior to the experiments that
generated these data points, the filters were exposed to an influent DO concentration of 13.6 mg/L
for two weeks (Experiment 5, Table 3.2), whereas the other data points were generated from filters
that had always been exposed to an influent DO concentration of 2 mg/L. It is likely that filter history
had an effect on subsequent experiments, since these data points are significantly below the curve
fitted through the other data in Figures 3.8 and 3.9. The negative effect of higher DO concentrations
on the bromate-reducing bacteria was not immediately reversible. These data suggest that the DO
concentration of the backwash water must be carefully considered. Optimally, the backwash water
should have a low DO concentration in order to minimize its deleterious effect on bromate-reducing
bacteria. However, the effects of short-term, high DO concentrations on biological bromate removal
have not yet been assessed; the effect on bromate removal would likely depend on the frequency and
duration of backwash. Simpkin and Boyle (1988) studied the effect of aerobic/anoxic cycles on the
synthesis of nitrate and nitrite reductase in activated sludge and found that these enzymes were
synthesized to at least 50 percent of their maximum levels; they concluded that the repression of
nitrate reductase and nitrite reductase synthesis by oxygen was not as important as the effect of
oxygen on the enzyme activities. Thus, if nitrate reductase is involved in bromate reduction, it is
possible that a short aerated backwash would not be detrimental to bromate removal. However, the
current study showed that the operation of the BAG filters with an influent DO concentration of 13.6
mg/L for two weeks was detrimental in subsequent bromate removal experiments.
Effect of the Influent Sulfate Concentration
In contrast to bromate, nitrate, and DO - which are high potential terminal electron acceptors
(£ 0/gr0 -/Br -=+1.03F; £°'0///0=+0.82F; E°'NO -/NO -=+OA3V)- sulfate is a low potential terminal
electron acceptor E° so ^-/HS -=-0.22V. Thus, sulfate reduction would tend to take place after
bromate, nitrate, and DO have been depleted. Based on this, the sulfate concentration was not
expected to have a considerable impact on bromate reduction in the BAG filter. To test this
hypothesis, a BAG filter was operated with influent sulfate concentrations of 11.1 and 102.7 mg/L.
Sulfate reduction did not take place in any of the experiments, since the influent and effluent sulfate
51

concentrations were equal (Figure 3.10). As the sulfate concentration increased, the average percent
bromate removal declined just slightly in the BAC with the 25- and 50-minute EBCTs (Figure 3.10).
This is similar to the findings of Chen and Hao (1996), who observed that 120 mg/L sulfate slightly
inhibited microbial chromate reduction even though no sulfate reduction was observed.
Effect of pH
It is useful to evaluate the sensitivity of bromate-reducing microorganisms to pH changes, so
that conditions for biological bromate reduction can be optimized. To assess the effect of pH on
bromate reduction, four BAC filtration experiments were run with pH values of 6.8,7.2,7.5, and 8.2.
The system was buffered with 1-2 mM phosphate buffer, and the effluent pH values deviated from
the influent values by less than 0.1 pH units. Bromate removal generally decreased as the influent pH
increased from 6.8 to 8.2 (Figure 3.11). The percent bromate removals observed at pH values of 6.8
and 7.2 are not different at the 95 percent confidence level, which might indicate that the optimum
pH for bromate removal is in this region.
1 o1
4>
1 O
ffl
90 -
80 -
70 -
60 -
50 -
40 -
D 50-min, bromate • 25-rrrin, bromate X influent sulfate
O 50-min, sulfate — 25-rrrin, sulfate
f--------.j----..........^ ;
^ —— ~~~— — --*—— _ *
T~~— —— — — — r
- 180.0 c
r 160.0 -B
- 140.0 |
- 120.0 g u
- 100.0 i
- 80.0 3
X
en
30 -
a
- 60.0 |
20 -
- 40.0 $|
10 -
20.0
*
0 - 11111
- 0.0
0.0 20.0 40.0 60.0 80.0 100.0 120.0
Influent Sulfate Concentration (mg/L)
Figure 3.10 Effect of influent sulfate concentration on bromate removal in BAC filters
(average influent conditions: CUW, pH 7.5, 21 ug/L BrO3", 2.1 mg/L DO, 0.3 mg/L NO/;
52

90 -
^ 80 -
70 -
J 60 - o
11
1
50 -
40 -
30 -
20 -
10 -
n -
• 50-minEBCT
D25-m] mEBCT
4 ' 5
f ' ——
i * i
6.5 7.5
pH
i
8.5
Figure 3.11 Effect of influent pH on bromate removal in BAG filters
(average influent conditions: CUW, 20 \ig/L BrO3', 2.1 mg/L DO, 5.0 mg/L NO3')
Since bromate formation during ozonation is dependent on the pH, bromate formation could
be reduced and biological bromate reduction could be increased by pH control. For example, Song
et al. (1996) observed a 50 percent reduction in bromate formation by reducing the pH from 8.0 to
7.0. Thus, if the pH for ozonation were decreased to the near-neutral range (6.8 to 7.2) to reduce
bromate formation, the data presented in Figure 3.11 show that biological bromate removal would
be improved. Although pH 6.8 and 7.2 yielded the highest percent bromate removal, the optimum
pH can only be found by evaluating a narrower range of pH values since biological reduction rates
can be very sensitive to small changes in pH. For instance, Chen and Hao (1996) observed a doubling
in the initial specific microbial chromate reduction rate as the pH increased from 7.0 to 7.3.
Effect of Source Water Type
To evaluate if the type of source water would have an effect on bromate reduction,
experiments were run with raw water from Lake Michigan using seven-year-old BAG from the Paul
M. Neal Water Treatment Facility (Lake Bluff, IL). Since this plant treats water from Lake Michigan,
it was appropriate to use this BAG for the LMW experiments. The influent DO concentration was
53

approximately 2 mg/L and the pH was 7.5. Table3.3 shows pertinent experimental information. The
LMW experiment was separated into five phases - A through E. Each phase had a different electron
donor concentration and/or EBCT. Table 3.3 shows the concentration of exogenous electron donor
and the total DOC. The total DOC is the sum of the organic matter present in LMW (2.4 mg C/L)
and the concentration of exogenous electron donor.
Two BAG filters were run in series for these experiments so that two EBCTs could be
evaluated simultaneously (Figure 2.2). The results from the lead BAG filter will be discussed first.
Figures 3.12 and 3.13 show the bromate and nitrate remaining in the effluent of the lead BAG filter
for each phase of the experiment. Note that each phase of the experiment in Figure 3.12 is labeled
as "no electron donor added", "low electron donor", or "high electron donor" in order to remind the
reader about the concentration of external electron donor added in each phase.
At the beginning of the experiment (Phase A), no external electron donor was added to the
influent. Table 3.3 shows that little DOC removal was observed in Phase A in the 15-minute EBCT
BAG filter, indicating the lack of biodegradable organic matter in LMW. In the full-scale plant from
which the BAG was obtained, an average of 20 percent DOC removal from LMW is observed in a
15-minute EBCT filter (Soucie 1999). The difference in DOC removal between the
Table 3.3 DOC removal and DO remaining in the BAG experiments using LMW (average
influent conditions: LMW, 2.0 mg/L DO, pH 7.5, 21 ug/L BrO3", 50 ug/L C1O4", 1.6 mg/L NO3')
Phase
EBCT
(min)
Exogenous electron donor
lactate + pyruvate (mg C/L)
Total DOC
(mg/L)
DOC
removal (%)
Effluent DO
(mg/L)
A
15
30
0
2.4
4
2
0.8
0.3
B
15
30
1.0
3.4
34
42
0.2
0.2
C
30
60
3.1
5.5
62
67
0.2
0.2
D
15
30
3.6
6.0
56
73
0.2
0.2
E
15
30
0.6
3.0
36
40
0.2
0.2
54

ench-scale and full-scale B AC filters was likely due to the fact that the bench-scale influent was not
ozonated while the full-scale filter receives ozonated influent (0.1 mg ozone/mg DOC) which can
increase the concentration of biodegradable organic matter. Very little bromate removal was
observed in Phase A (Figure 3.12). Figure 3.13 shows that only a small amount of nitrate was
produced in the BAG (C/C0 > 1) because the influent NH3-N concentration was routinely below the
detection limit of 0.05 mg/L. The effluent DO concentration was 0.8 mg/L (Table 3.3), which
indicates the presence of biological activity in the filter. Kirisits and Snoeyink (1999) showed that
an 18-minute EBCT BAG filter using CUW under similar conditions (pH 7.5, influent DO = 2.0
mg/L, effluent DO = 0.2 mg/L, influent nitrate = 0.3 mg/L, effluent nitrate = 2.1 mg/L, 20 ng/L
bromate) removed 60 percent of the bromate. The reduction of bromate in LMW was likely hindered
by a combination of factors which are explained as follows. Only a small amount of DO was
consumed in the biological nitrification reaction because of the low ammonia concentration. This DO
could then compete with bromate for the biodegradable organic matter. Alternatively, if nitrate
reductase catalyzes bromate reduction, the DO could also have inhibited the synthesis or activity of
the enzyme.
0-minEB
e don
15-min EBCT
Low e" donor
15-min EBCT
High e" donor
15-mmEBCT
No e" donor added
15-min EBCT
Low e" donor
0 1000 2000 3000 4000 5000 6000
Bed Volumes
7000 8000 9000 10000
Figure 3.12 Bromate remaining after first BAG filter using LMW (average influent conditions: Lake
County BAG, LMW, 2.0 mg/L DO, pH 7.5, 21 ug/L BrO3', 50 ng/L C1O4', 1.6 mg/L NO3')
55

fl
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000
Bed Volumes
Figure 3.13 Nitrate remaining after first BAG filter using LMW (average influent conditions: Lake
County BAG, LMW, 2.0 mg/L DO, pH 7.5, 21 ug/L BrO3', 50 ug/L CIO/, 1.6 mg/L NO3')
Lactate and pyruvate (1 mg C/ L) were added to the influent in Phase B, and the EBCT was
maintained at 15 minutes. There was little difference in the amount of bromate removed between
Phases A and B (Figure 3.12), despite the fact that significant DOC removal was observed in Phase
B (Table 3.3). The low effluent DO concentration (0.2 mg/L) suggests that much of the DOC in
Phase B was being used by aerobic microorganisms. Since nitrate removal became significant in
Phase B (Figure 3.13), DOC was also being consumed by nitrate reducers. This may indicate that
a nitrate-reducing population was developing on the carbon but was not able to reduce bromate due
to an insufficient concentration of electron donor.
In Phase C, the EBCT in the BAC was doubled from 15 to 30 minutes, and the external
electron donor concentration was increased to 3.1 mg C/L (Table 3.3) to provide a favorable
environment for bromate reduction. Figure 3.13 shows that the effluent nitrate concentrations were
very low. Figure 3.12 shows that bromate removal increased until the effluent bromate concentration
reached the detection limit.
When the EBCT was reduced from 30 to 15 minutes (Phase D) while the external electron
donor concentration was maintained close to the concentration in Phase C (Table 3.3), complete
56

omate removal was sustained (Figure 3.12). The effluent nitrate concentration also remained very
low (Figure 3.13). This shows that the electron donor concentration, and not the EBCT, was limiting
bromate removal in these experiments. This is different from the results observed in CUW because
bromate removal increased as the EBCT increased, indicating that EBCT was limiting bromate
removal (Kirisits and Snoeyink 1999).
Finally, the external electron donor concentration was decreased to 0.6 nig C/L while
maintaining an EBCT of 15 minutes in Phase E (Table 3.3). Figures 3.12 and 3.13 show that the
effluent bromate and nitrate concentrations began to increase. It is likely that the electron donor
concentration was limiting the growth of the nitrate-reducing and bromate-reducing organisms. With
less electron donor present, there was a decreased demand for nitrate and bromate as electron
acceptors.
Table 3.3 and Figures 3.14 and 3.15 show the results from the two BAG filters in series. The
trend of the data in each phase of Figures 3.14 and 3.15 is the same as that in Figures 3.12 and 3.13.
Comparing phase A between Figures 3.12 and 3.14, when no exogenous electron donor was added
to LMW, bromate removal did not improve even though the contact time was increased from 15 to
30 minutes. This is another indication that the electron donor concentration was limiting bromate
removal from LMW.
1.0
•1 °- 8 H
•a
i
ft 0.6 H

fi
0 1000 2000 3000
4000 5000
Bed Volumes
Figure 3.15 Nitrate remaining after second BAG filter using LMW (average influent conditions:
Lake County BAG, LMW, 2.0 mg/L DO, pH 7.5, 21 ng/L BrCV, 50 ug/L CIO;, 1.6 mg/L NO3')
Based on the results of the LMW experiments, it can be concluded that the biodegradability
of the NOM in the source water can significantly affect the biological bromate removal. Ozonating
the LMW to increase the biodegradable fraction of NOM may improve bromate removal.
Although bromate removal was the focus of the work using Lake County BAG, perchlorate
was also added to the column influent beginning in Phase B (Figures 3.16 and 3.17). At this time,
the concentration of perchlorate in the effluent was increasing, as the ion exchange capacity for
perchlorate on the GAC was approached. Perchlorate reduction increased as the concentration of
exogenous electron donor increased and the effluent nitrate concentration was very low (Phases C
and D). As the effluent nitrate concentration began to increase in Phase E, due to a lower
concentration of external electron donor, perchlorate removal rapidly declined. The dependence of
perchlorate removal on nitrate concentration is further explored later in this chapter.
58

Fraction Perchlorate Remaining, Fraction Perchlorate Remaining,
C/Co

Perchlorate Reduction in CITW
The influent and effluent concentrations of perchlorate to the BAG filters operated with CUW
were also monitored. In all experiments, no evidence of reduction of the influent perchlorate
concentration (50 ng/L) was observed. As discussed later in this chapter, perchlorate reduction is
extremely sensitive to nitrate concentration. The threshold nitrate concentration, at which 95 percent
perchlorate reduction could be obtained, was between 0.35 and 0.05 mg/L. Since CUW contains
ammonia, nitrification took place in the BAG filters. Depending on the experimental conditions,
nitrification caused the nitrate concentration to increase by 1.8-4.6 mg/L. Significant nitrate
reduction did not take place since no external electron donor was added to CUW. Therefore, the
0.35/0.05 mg/L nitrate threshold for perchlorate reduction could not be obtained in these
experiments.
Microscopic Examination of Microorganisms in the Filters
A microscopic examination of biomass taken from the BAG filters operated with CUW was
completed. Four representative pictures are shown in Figures 3.18-3.21, which were obtained using
phase contrast microscopy with 1 OOOx magnification. The dark objects are microorganisms while the
lighter, refractile images are likely debris.
The bacterial morphologies were generally observed to be cocci, rods, chain-forming rods,
and aggregated rods. All of these morphologies were found in both sets of BAC filters operated with
CUW.
Rods and cocci can be observed in Figure 3.18. Large aggregates of bacteria associated with
non-biological particulates were commonly observed. The figure also shows some flocculated
material that has bacteria incorporated into the floe structure. Despite the fact that the carbon filters
looked very "clean" and did not experience significant headless, the microscope images confirm that
there are a lot of microorganisms in the filters.
A chain of rods, illustrated in Figure 3.19, was a common morphology observed in BAC filter
samples. It is interesting to note that Hijnen et al. (1995) isolated some chain-forming rods that were
able to reduce bromate. Figure 3.20 shows a loose aggregation of rods.
60

Figure 3.18 Rods and cocci from the BAG filter
Figure 3.19 Chain of rods from the BAG filter
Figure 3.20 Aggregated rods from the BAG filter
61

Figure 3.21 Protozoa in the BAG filter
Protozoa were also found in the filters. The common morphology seemed to be the shape of
a round-bottom flask. Protozoa can be seen in Figure 3.21. It is not uncommon to find protozoa in
BAG filters, especially filters that have only been backwashed once in two years of operation. Higher
organisms, such as protozoa, graze on the microbial population.
Denaturing Gradient Gel Electrophoresis Experiment
DGGE is a genetic fingerprinting technique used to profile the bacterial diversity present in
a sample. The method is based on the separation of PCR-amplified ribosomal DNA (rDNA)
fragments of the same length but with different sequences. Each band theoretically represents a
unique organism. Figure 3.22 shows a DGGE profile that was obtained with DNA extracted from
the BAC filters operated with CUW. DNA from each sample was extracted and analyzed in
duplicate. The first lane is a standard marker. The next four lanes are from the BAC filters operated
for perchlorate removal and will be discussed in a later section. Lanes 6 and 7 were run with a
microbial sample from the effluent side of the BAC operated with CUW and an influent nitrate
concentration of 5.0 mg/L. Lanes 8 and 9 were run with material isolated from the backwash water
of the BAC filter operated with CUW and an influent nitrate concentration of 0.3 mg/L. Lanes 10
and 11 were run with material isolated from the backwash water of the BAC filter operated with
CUW and an influent nitrate concentration of 5.0 mg/L.
Since the number of bands is a measure of the diversity of the microbial community, it can be
62

seen that the BAG samples contain a diverse array of microorganisms (Figure 3.22). It is possible
that the genetic diversity is even greater in the samples than indicated by the banding profiles. By
analyzing the DNA fragments using a narrow denaturing gradient, it is possible to further resolve
single bands into multiple bands representing distinct sequences. Individual bands, excised from
DGGE gels, can also be further analyzed for DNA sequences, and thus the identity of specific
members of the microbial community can be determined.
There are conserved bands across the samples taken from the BAG filters operated to remove
bromate, and there are some bands that are unique to a particular sample. There are conserved bands
between the samples taken from the BAG filters operated to remove bromate and the BAG filters
operated to remove perchlorate; this was expected since the BAG filters operated to remove
perchlorate were first run with effluent from the BAG filters that were removing bromate.
Lanes 123456789 10 11
Figure 3.22 DGGE profile of bromate and perchlorate samples
63

Future work may use DGGE profiling as a technique for observing significant perturbations in the
population that may relate to the bromate reduction performance of the BAG filter. DGGE could be
run on BAG filter biomass samples after changing water quality parameters or operational conditions
in the BAG filters. For instance, DGGE could be run on biomass samples from the BAG filters that
have been exposed to different pH values, to see if the bromate-reducing isolates are able to persist
under different pH conditions.
Plating Experiments and Microbial Characterization
Biomass was taken from the BAG filters operated with CUW and an influent nitrate
concentration of 5 mg/L, and it was diluted and streaked on agar plates. Fourteen isolates were
obtained from the dilutions of the mixed culture.
Bromate Reduction in CUW
To test some of the isolates for bromate reduction ability, biomass was taken from the plates
and inoculated to Balch tubes containing CUW spiked with bromate and nitrate. The results of the
experiment are summarized in Table 3.4. Little or no bromate removal was demonstrated by the
isolates under the tested conditions. The most bromate reduction was demonstrated by isolate B15,
which reduced bromate from 28 to 2 \igfL after 12 weeks. Nitrite and nitrate data were also obtained
in this experiment. Isolate Bl showed nitrate removal with no nitrite accumulation. Isolates B2,B3,
B4, B9, B11, B12, and B13 showed nitrate removal with nitrite accumulation. Isolates B6, B8, B14,
and B15 did not show substantial nitrate removal. A mass balance on nitrite + nitrate showed that
isolates Bl and Bl 1 may have converted some of the nitrate to unmonitored species (such as NH3
or N2). The nitrogen balance showed that some DO may have been present in the tubes containing
B3 and B13, leading to nitrification. The nitrite + nitrate balance for the rest of the isolates was close
to the control value, indicating that nitrite and nitrate were not converted to unmonitored nitrogen
species.
64

Bromate Reduction in DDW with Exogenous Electron Donor
Based on the poor bromate removal exhibited in the previous experiment, the experiment was
modified and repeated. The aqueous medium was DDW spiked with bromate, nitrate, and exogenous
electron donor. The concentration of electron donor in the medium (0.07 mM lactate, 0.10 mM
acetate, and 0.08 mM pyruvate) was such that 10 mg/L nitrate could be reduced to N2 . Not all of
the same isolates used in the previous experiment (Table 3.4) were used in this experiment. B3, B4,
B12, B13, and B14 could not be sub-cultured after the first culture was grown.
Table 3.4 Testing the isolates for the ability to reduce bromate (12-week incubation; initial
conditions: CUW, 5 mM phosphate buffer, 28.7 \igfL BrO3', 4.82 mg/L NO 3',
pH 7.8, 0 mg/L DO)
Isolate
Control -
no inoculum
Bl
B2
B3
B4
B6
B8
B9
Bll
B12
B13
B14
B15
Bromate
(Hg/L)
28.4
26.6
31.0
26.0
26.0
27.5
28.8
25.7
28.6
26.7
28.9
27.7
24.6
Nitrite
(mg/L)
0.05
0.03
3.51
5.04
3.78
0.20
0.31
3.44
1.50
3.42
4.68
0.04
0.05
Nitrate
(mg/L)
4.82
0.39
0.20
0.21
0.17
4.74
4.48
0.38
0.13
0.13
0.12
4.68
4.84
Nitrite + nitrate
(mg/L as N)
1.10
0.09
1.11
1.58
1.19
1.13
1.11
1.13
0.49
1.07
1.45
1.07
1.11
65

Table 3.5 shows that all of the isolates tested (Bl, B2, B6, B7, B8, B9, BIO, Bl 1, and B15)
were able to reduce bromate to some degree. Isolate B7 showed the best bromate removal after the
4.5-week incubation period, since it was able to reduce bromate to the detection limit. Isolates B2
and B8 showed the least bromate removal. Even with the external electron donor addition, isolates
B6, B8, and B15 were not able to reduce nitrate (Table 3.5) as was observed earlier in the absence
of an external electron donor (Table 3.4). Isolate B2 was unable to remove a substantial amount of
nitrate in the presence of external electron donor (Table 3.5), in contrast to the nitrate removal and
nitrite accumulation observed in the absence of external electron donor (Table 3.4). Isolates Bl, B9,
BIO, and Bl 1 were able to reduce nitrate with nitrite accumulation. Isolate B7 was able to remove
nitrate with no nitrite accumulation.
Table 3.5 Testing the isolates for the ability to reduce bromate (4.5-week incubation; initial
conditions: DDW, 5 mM phosphate buffer, 43.5 ug/L BrO3', 4.92 mg/L NO3', 8.3 mg/L DOC,
pH 7.5, 0 mg/L DO)
Isolate
Control -
no inoculum
Bromate
(Hg/L)
43.5
Nitrite
(mg/L)
0.00
Nitrate
(mg/L)
4.92
Nitrite + nitrate
(mg/L as N)
1.11
Bl
B2
B6
B7
B8
B9
BIO
Bll
B15
36.6
40.6
36.3
1.0
39.5
34.1
32.5
37.6
37.3
0.42
0.16
0.09
0.00
0.14
3.64
3.18
2.49
0.06
2.99
4.35
4.85
0.41
4.85
0.41
0.42
0.42
4.77
0.68
1.03
1.12
0.09
1.14
1.20
1.06
0.85
1.10
66

The mass balance on nitrate + nitrite (Table 3.5) showed that isolates B1, B7, and Bll may
have converted some of the nitrate to unmonitored species (such as NH3 or N2). Isolates B1 and Bll
also demonstrated this behavior in the previous experiment (Table 3.4). The mass balance on nitrate
+ nitrite for the rest of the isolates was close to the control value, indicating that these
microorganisms did not convert nitrate to unmonitored nitrogen species.
Challenging the Isolates
The isolates were also challenged with higher than usual concentrations of bromate and
nitrate. Each isolate was inoculated into aqueous medium containing 14.0 mg/LNO 3 " -N (Table 3.6),
5.6 mg/L BrO 3--Br (Table 3.7), or 0.6 mg/L NO 3'-N / 5.6 mg/L BrO 3'-Br (Table 3.8).
Table 3.6 shows that all of the isolates, except for B15, were able to reduce nitrate. Table
3.7 shows that isolates B2, B6, BIO, and B15 were able to substantially reduce bromate when no
nitrate was added to the medium. Table 3.8 shows that isolates B6, B7, BIO, Bll, and B15 were
able to substantially reduce bromate when nitrate was added to the medium.
Table 3.6 Challenging isolates with a high nitrate concentration (initial conditions: modified
MR2A broth, 14.0 mg/L NO3 -N, pH 7.0, 0 mg/L DO)
Isolate
Bl
B2
B6
B7
B8
B9
BIO
Bll
B15
Nitrate removed
(mg/L as N)
1.1
13.3
13.6
13.6
4.3
13.6
13.6
13.5
-0.6
Nitrite produced
(mg/L as N)
0
12.7
0
0
0
0
0
0
0
67

Table 3.7 Challenging isolates with a high bromate concentration (initial conditions: modified
MR2A broth, 5.6 mg/L BrO 3' -Br, pH 7.0, 0 mg/L DO)
Isolate
Bl
B2
B6
B7
B8
B9
BIO
Bll
B15
Bromate removed
(mg/L as Br)
0.4
1.2
2.4
0.2
0.2
0.2
1.5
0.2
1.2
Bromide produced
(mg/L as Br)
0.4
1.5
1.5
0.4
0.1
0.1
1.0
0.2
1.1
Table 3.8 Challenging isolates with high concentrations of bromate and nitrate (initial conditions:
MR2A broth, 5.6 mg/L BrO 3' -Br, 0.6 mg/L NO3" -N, pH 7.0, 0 mg/L DO)
Isolate
Bromate removed
(mg/L as Br)
Bromide produced
(mg/L as Br)
Nitrate removed
(mg/L as N)
Nitrite produced
(mg/L as N)
Bl
0.2
0.3
0.4
0
B2
0.2
0.2
0.5
0.6
B6
2.1
1.6
0.5
0
B7
0.7
0.5
0.5
0
B8
0.3
-0.2
0.5
0
B9
0.4
0.2
0.5
0
BIO
2.3
2.3
0.6
0
Bll
1.1
0.9
0.6
0
B15
0.8
1.1
0.2
-0.2
68

N2O was not detected and no evidence of trapped gases were present in Durham tubes in any
cultures where nitrate consumption had occurred, indicating that the primary products of nitrate
reduction were not likely to be through denitrification pathways (i.e. N2O or N2 gases). It is likely
that nitrate is being reduced to ammonia, and this will be tested in future experiments.
Additional Isolate Information
All cultures are facultative, since they were able to grow on agar plates that were incubated
anaerobically and those that were incubated aerobically. Isolate B6 was further tested in MR2A
supplemented with yeast extract, casamino acids, acetate, and pyruvate, without additional electron
acceptors present (i.e. no bromate, nitrate, or DO). Isolate B6 grew but growth was significantly less
than when additional electron acceptor was present. This indicates that isolate B6 can grow by
fermentation.
A summary of all the information concerning the isolates is shown in Table 3.9. Isolate B8
is the pure culture that has not shown much promise regarding bromate reduction. Isolates Bl and
B9 have shown promise for bromate reduction at low bromate concentrations (~ ug/L) but not at high
bromate concentrations (- mg/L). Isolates B6, B10, and B15 seem quite robust, since they were able
to reduce bromate under all conditions tested. All of the isolates shown in Table 3.9 will be
sequenced and phylogenetically classified.
Dilution-to-Extinction Experiments
Using a non-traditional enrichment technique, consisting of progressive dilutions, Jackson
et al. (1998) was able to enrich for different microorganisms at different dilutions. For example,
microorganisms that grow very well in the culture medium will be selected for at lower dilutions;
they will be able to outcompete organisms that are originally present in higher numbers but are not
well-suited to the chosen culture medium. On the other hand, microorganisms that were abundant
in number in the original culture will be selected for at higher dilutions since the less abundant
organisms will essentially be diluted out of the culture. In contrast to the plating experiments
described earlier, which targeted pure cultures using an artificial nutrient-rich condition, the dilutionto-extinction
method considers microorganisms which may be "specialists" or whose activity may
69

Table 3.9 Summary of experimental evidence for isolates
High BrO3" concentration Low BrO 3 " concentration
Isolate
Bl
B2
B6
B7
B8
B9
BIO
Bll
B15
Bromate- Bromate- Bromate-reducer, Nitrate- Facultative
reducer, reducer, with nitrate reducer
with nitrate without nitrate
X XX
X X X
X X X XX
X XXX
X X
X XX
XX X XX
X XXX
XXX X
be dependent upon a consortium. This is an alternative strategy that targets the isolation of pertinent
microorganisms under conditions that more closely parallel those under which the BAG filters are
operated (i.e. low nutrient conditions).
The results of this experiment are tabulated in Table 3.10. Using biomass from the BAG
operated with an influent nitrate concentration of 5 mg/L, bromate reduction was observed for all
dilutions and decreased as the dilution increased. Close attention should be paid to the highest
dilution, 1:10,000, since bromate reduction was still observed here. At this high dilution, organisms
not well-suited to the medium were diluted out of the mixture, thereby lessening the complexity of
the microbial population. For all tubes, significant nitrate removal was observed. It is not known
if bromate removal took place before, during or after the nitrate removal.
70

Table 3.10 Results of dilution-to-extinction experiments (3.5-month incubation; initial conditions:
CUW, 28 ug/L BrCV, 7.25 mg/L N(V, pH 7.5, 0 mg/L DO)
Biomass Source - B AC filters operated with an influent nitrate concentration of 5.0 mg/L
Dilution Undiluted
1:10
1:100
1:1,000
1:10,000
Bromate (ng/L) 2.5
9.3
19.1
21.0
23.8
Nitrite (mg/L) 0.07
ND
ND
ND
ND
Nitrate (mg/L)
0.13
0.10
0.10
0.09
0.09
Biomass Source - B AC filters operated with an influent nitrate concentration of 0.3 mg/L
Dilution Undiluted
1:10
1:100
1:1,000
1:10,000
Bromate (ug/L) 1.0
2.0
1.0
1.0
2.6
Nitrite (mg/L) ND
ND
ND
ND
ND
Nitrate (mg/L)
0.09
0.10
0.09
0.09
0.09
Using biomass from the BAG filters operated with an influent nitrate concentration of 0.3
mg/L, bromate removal was observed to levels at or near the detection limit for all dilutions. This
suggests that the bromate-reducing bacteria were numerically dominant in this particular inoculum.
The organisms from the dilution-to-extinction experiments were not plated onto agar medium.
Characterization of the eight bromate-reducing cultures that have been isolated by traditional
techniques (Table 3.9) will be completed before attempting to isolate additional bromate-reducers.
71

Serum Bottle Experiments
To evaluate the effect of temperature on biological bromate reduction, batch experiments
were set up in serum bottles with aqueous medium and inocula from the BAG filters and incubated
at 25 and 4 °C. External electron donor (lactate, acetate, and pyruvate) was added to the bottles after
33 days. As expected, bromate was reduced in the 25 °C experiment (Figure 3.23) while bromate was
not reduced in the 4 °C experiment (Figure 3.24). Figure 3.23 shows that nitrate was also removed
at 25 °C, and bromate reduction took place in the presence of nitrate.
c
_o
1

0
so
o
U
).0 -
I 5.0
U
ffl
Addition of external electron donor
6.0
-- 5.0
-- 4.0
u/ u ' v
3.0 ai o --
2.0
1.0
o •-S
U S
0.0
0 20 40 60 80
Time (days)
100 120
0.0
Figure 3.24 Bromate reduction in a serum bottle experiment (initial conditions: CUW, pH
7.7, 19 ug/L BrO3-, 5.4 mg/L NO3', 4 °C)
PERCHLORATE - ABIOTIC REMOVAL
Calculation of Ion Exchange Capacity for Virgin Norit GAC
CUW spiked with 50 ug/L of perchlorate was filtered through a bed of virgin Norit GAC with
a 25-minute EBCT. Figure 3.25 shows the perchlorate breakthrough curve for the first 5,800 bed
volumes (BV). A mass balance calculation showed that the total mass of perchlorate applied to the
carbon through 5,800 BV equaled the total mass of perchlorate in the effluent through 5,800 BV.
Therefore, any observed perchlorate removal up to 5,800 BV was the result of adsorption, probably
ion exchange. The first 1,600 B V correspond to the phase when perchlorate was undergoing primary
ion exchange. To calculate the ion exchange capacity of the GAC, the area above the curve and
below C/C0= 1 from O^BV< 1,600 (shaded area, Figure 3.25) was integrated. This integration
showed an ion exchange capacity of 0.172 mg perchlorate (1.7 x 10"6 mole) per gram GAC.
73

1000 2000 3000 4000
Bed Volumes
5000 6000
Figure 3.25 Ion exchange of perchlorate to GAC (initial conditions: CUW, 2.5 mg/L DO, 5.0
mg/L NO3-, pH 7.5, 50 ug/L C1O4')
Abiotic Batch Experiments
Figures 3.26 and 3.27 show the batch perchlorate removal results for four GACs in CUW
and DDW, respectively. Perchlorate was abiotically removed by each carbon. In both DDW and
CUW, the virgin Calgon F-400 carbon performed better than the virgin Norit carbon. This is most
likely due to the fact that the Norit carbon particles are much larger than the Calgon F-400 carbon
particles. A larger particle size requires a perchlorate molecule to travel a greater diffusional distance
before reaching an active site on the carbon. The two figures also show that the AWOG treated
carbons perform better than the virgin carbons. This is in agreement with the earlier findings of Miller
(1996) and Kirisits et al. (1998), both of whom attributed the differences in virgin and AWOG carbon
reactivity to changes in surface chemistry generated during the process of acid washing and
outgassing.
Figure 3.28 compares perchlorate removal by virgin Calgon F-400 in CUW and DDW.
Perchlorate removal performance was better in DDW than in CUW. These results were comparable
for each of the 4 carbons. NOM or other inorganic anions in CUW are likely competing with
perchlorate for active sites on the carbon, thus decreasing perchlorate removal.
74

o
I
Io
U
D
o
ou -
50 1
40 -
30 -
20 -
10 -
!•••
"*£ 2 o
A S 5 D D
A A
• Virgin Norit
D Virgin F-400
AAWOGNorit
AAWOGF-400
*
A
n
0 50 100 150 200
Time (min)
250 300 350
Figure 3.26 Batch perchlorate removal (initial conditions: CUW, pH 7.5, 500 mg/L GAC)
••s 0
3 co
o tH
O 5b
J5
o

hd 0 >t> "0
o
— . g «>
1 Perchlorate Concentration f Perchlorate Concentration f f'
S 0*/L) I WV 1 1
« o 5 £ S £ 3 § £ o 5 £ £ £ 3 § £ f
P
1 ° -
w ° 1
—— i —— i —— i —— i —— an —
^ OQ 3r*
S
•n
S cr
0s-
O. 0)
•o
TJ 0
• D
(6
S 1-
O
I °
»
i -
< -
SL 0
2.'
/*N
«-»-
£
§ si § -
B: §
1' i"
a B .
to S tO
^ 0 -
0 °
D
$
o to
§ 3 -
1O u>
> 0 -
h °
u>

Wash Tests
Table 3.11 lists the results of the high nitrate concentration wash tests performed on four of
the carbons used in the batch abiotic perchlorate tests. It can be noted that, within the limits of
detection for the ion chromatographic analysis, the entire mass of perchlorate that was removed by
the GAC was washed off the GAC. Therefore, no chemical reduction of perchlorate took place on
the GAC.
Metal-Catalyzed GAC Experiments
Figures 3.30 and 3.31 show that both the filter containing GAC mixed with aluminum shot
and the filter containing copper- and zinc oxide-impregnated GAC were efficient at removing
perchlorate from influent water.
Both GAC filters were stopped after approximately 500 BV. Up to that point the
GAC/aluminum filter had removed a total of 52.2 mg of perchlorate and the copper- and zinc oxideimpregnated
GAC filter had removed a total of 51.8 mg of perchlorate. At 500 BV, there was no
indication that perchlorate removal efficiencies were decreasing for either filter.
GAC
Virgin Norit
Virgin F-400
AWOG Norit
AWOG F-400
Virgin Norit
Virgin F-400
AWOG Norit
AWOG F-400
Virgin Norit
Table 3.11 Abiotic perchlorate batch removal and wash tests
Water
CUW
CUW
CUW
CUW
DDW
DDW
DDW
DDW
DDW
PH
7.5
7.5
7.5
7.5
7.5
7.5
7.5
7.5
2.0
Mass of perchlorate
removed
G*g)
42.0
71.8
80.5
72.3
60.4
90.7
95.5
95.7
85.3
Mass of perchlorate
washed off carbon
fog)
Not measured
70.2
79.0
72.1
Not measured
Not measured
Not measured
Not measured
83.1
77

00
•1
1
& a o
2 u
0
fc
OH
§
* i-H
-4-*
2
i.u -
0.9 -
0.8 -
0.7 -
0.6 -
0.5 -
0.4 -
0.3 -
0.2 o.i -
0.0 Jr ¥ i • i
• • A A • ***
0 200 400
Bed Volumes
600
Figure 3.30 Removal of perchlorate by GAC mixed with aluminum shot (initial
conditions: DDW, EBCT = 25 minutes, pH 2.0, C1O4" = 2.7 mg/L)
n
c3
'5
'51

The next step was to determine if perchlorate was being removed by adsorption or chemical
reduction. If perchlorate was being reduced completely to chloride, then it should have been possible
to observe a stoichiometric increase in the effluent chloride concentration, based on the amount of
perchlorate being removed in the filters. However, no increase in effluent chloride concentration was
observed. If adsorption was the dominant removal mechanism, it should have been possible to wash
the entire mass of adsorbed perchlorate off the carbon using a concentrated nitrate solution. The
GAC was washed until no perchlorate could be detected in the filter effluent. A 10 mg/L nitrate
solution was used to desorb perchlorate from the metal-catalyzed GAC filters (Table 3.12) followed
by a 200 mg/L nitrate solution (Table 3.13).
When the 10 mg/L nitrate solution was applied, 79 percent of the perchlorate removed by the
GAC/aluminum filter was desorbed, but almost no perchlorate was released from the copper- and zinc
oxide-impregnated GAC filter. When the 200 mg/L nitrate solution was applied, little perchlorate
came off of the GAC/aluminum filter, while essentially all of the perchlorate was released from the
copper- and zinc oxide-impregnated GAC filter.
Table 3.12 Perchlorate mass balance: mass perchlorate removed by and recovered from metal-
catalyzed GAC (10 mg/L NO3' rinse)
Filter
GAC/aluminum
Cu- and ZnOimpregnated
GAC
Mass perchlorate removed
by carbon (mg)
52.10
51.80
Mass perchlorate
recovered (mg)
41.10
0.97
% recovery
78.9
1.9
Table 3.13 Extended perchlorate mass balance: total mass removed by and recovered from metal-
catalyzed GAC (10 + 200 mg/L NO3 " rinses)
Filter Mass perchlorate Mass perchlorate % recovery
removed by carbon washed off carbon
__________________(mg)__________(mg)__________________
GAC/aluminum 52.10 42.73 82.0
Cu- and ZnOimpregnated
GAC_______51.80__________51.39__________99.2_____
79

Thus, perchlorate was removed by the copper- and zinc oxide-impregnated GAC filter by ion
exchange. Most of the perchlorate that was removed by the GAC/aluminum filter was also removed
by ion exchange, although 18 percent (9.37 mg) of the perchlorate removed by the filter was not
recovered. A fraction of the influent perchlorate may have been chemically reduced or may still be
on the carbon, but there is no evidence to support either possibility. Even under the low pH and high
perchlorate concentration conditions in these experiments, ion exchange is the dominant mechanism
of perchlorate removal by GAC.
Ozone/Hydrogen Peroxide/Perchlorate Batch Tests
Table 3.14 shows the results of the ozone/hydrogen peroxide/perchlorate batch tests, and no
perchlorate reduction occurred in the presence of ozone or ozone and hydrogen peroxide.
PERCHLORATE - BIOLOGICAL REDUCTION
BAC Filtration Experiments
Electron Donor Limited Experiments
Figure 3.32 illustrates the results of the BAC filter/syringe pump experiments. The figure
examines the fraction of perchlorate remaining versus the concentration of nitrate in the effluent of
the 25-minute EBCT filter. It is important to note that biological removal of perchlorate has been
achieved. The ion exchange capacity of the GAC had been exhausted prior to the start of this
experiment. There has been no evidence that perchlorate can be abiotically reduced by GAC.
Therefore, any removal demonstrated during this experiment is the result of biological reduction.
The second interesting result illustrated by Figure 3.32 is that with low influent DO
concentrations, the biological reduction of perchlorate is highly sensitive to the concentration of
nitrate present in the water. As nitrate concentration increases, perchlorate removal decreases.
Figure 3.32 also indicates that even though the influent nitrate concentrations were constant,
nitrate removed by the BAC filters was variable. This was caused by the fact that the floating,
80

Jar
cio4-
42.8
40.4
40.5
41.1
6
C/C0
1.00
0.95
0.95
0.96
Table 3.14 Ozone/hydrogen peroxide/perchlorate batch tests
(initial conditions: DDW, pH 8.8)
Time
(min)
Jar
cio4-
1
C/C0
Jar
cio4-
2
C/C0
Jar
cio;
3
C/C0
Jar
cio 4-
4
C/C0
Jar
cio 4-
5
C/C0
oo
0
46.6
1.00
46.5
1.00
46.2
1.00
43.2
1.00
43.5
1.00
5
45.9
0.98
45.6
0.98
43.1
0.93
44.3
1.03
41.8
0.96
10
47.1
1.01
45.9
0.99
44.7
0.97
43.6
1.01
41.0
0.94
15
45.9
0.98
45.3
0.98
43.9
0.95
42.4
0.98
41.4
0.95

2.5mg/L
Influent NOs- =

of nitrate on perchlorate reduction. Three different influent nitrate concentrations were used: 0.1
(concentration of nitrate present in DDW with no external spike), 1.0, and 1.5 mg/L. Figure 3.33
shows the results of the experiments with influent nitrate concentrations of 1.0 and 1.5 mg/L.
Several important results were obtained from these experiments. First, due to the excess
electron donor, the effluent nitrate concentration was constant for a given influent nitrate
concentration. That is, EBCT was controlling nitrate removal in the filters. This is in contrast to the
previous BAG filter experiments which showed fluctuating effluent nitrate concentrations for a given
influent nitrate concentration. Second, for a given nitrate concentration, the degree of perchlorate
removal was essentially constant. Therefore, these experiments have been successful at eliminating
the condition where electron donor concentration limits the removal of perchlorate. The
concentration of nitrate present in the filter has become the variable controlling the removal of
perchlorate. Third, it was observed that the effluent nitrate concentration needed to be maintained
between 0.05 and 0.35 mg/L to obtain >95 percent perchlorate removal. When the influent nitrate
concentration was 1.5 mg/L, the average effluent nitrate concentration was 0.35 mg/L. Under these
1.0
§ 0.9
oB 0.8
1 0-5 H
I 0.4 J
o
f£ c
1 0.2 H
ts
0.0
B B B
v/
/
Influent NO3"= 1.5 mg/L
Effluent NCh" = 0.35 mg/L
B
Influent NO3~= 1.0 mg/L
Effluent NCb" = 0.05 mg/L
0 200 400 600 800 1000 1200
Bed Volumes
Figure 3.33 Nitrate isolation experiments (initial conditions: 2.5 mg/L DO, 50 ng/L C104",
-1.7 mg/L as C added electron donor, pH 7.5)
83

conditions, 65 percent perchlorate removal was observed (Figure 3.33). When the influent nitrate
concentration was 1.0 mg/L, the average effluent nitrate concentration was 0.05 mg/L, and
perchlorate removal was >95 percent (Figure 3.33). Since it has been noted that there is an inverse
relationship between the effluent nitrate concentration and perchlorate removal, the 0.05-0.35 mg/L
range of nitrate concentrations was taken directly from Figure 3.33. This means that these particular
BAG filters can remove perchlorate in the presence of nitrate, but only a low concentration of nitrate
can be present if >95 percent perchlorate removal is expected.
The experiment with no nitrate added was run for 950 bed volumes. For the duration of the
experiment, no perchlorate was measured in the effluent. Thus, in presence of very low nitrate
concentrations (

Effluent Perchlorate, C/C
l.U
0.8 -
0.6-
0.4 -
* * * *
• Non-Nitrate pre-loaded
• Nitrate pre-loaded
0.2 -
B
sa a u • H B B
o.o - ———————,———————,———————————
0 200 400 600 800 1000 1200 1400
Bed Volumes
Figure 3.34 Comparison of perchlorate removal for nitrate preload and non-nitrate
preload conditions (initial conditions: DDW, DO = 2.5 mg/L, C1O4' = 50 ug/L,
NCV = 0 mg/L, 1.5 mg/L as C added electron donor, pH 7.5, EBCT = 15 minutes)
perchlorate-reducers could be reducing nitrate to ammonia for assimilatory purposes. Other than
endogenous respiration, nitrate is the only nitrogen source for the microorganisms in the BAC filter
since there is no ammonia in the influent. If endogenous respiration was not providing sufficient
nitrogen to satisfy microbial nutrient requirements, a portion of the biomass would die, thus
decreasing the quantity of perchlorate-reducing biomass. The more nitrate present, the more nitrogen
available for assimilation and growth. Increased growth means that there would be more cells present
to degrade perchlorate, thus causing an increased rate of perchlorate reduction. This scenario
suggests that ammonia is the key component of the system for the microbial community. Nitrate is
not required per se, but is only used to generate ammonia. A second possible explanation for the
improved perchlorate removal performance when the filter was preloaded with nitrate is that
microorganisms in the filter dissimilatively reduce nitrate to ammonia, nitrous oxide, or dinitrogen
gas to gain energy for growth. The electron transport involved in these reduction pathways directly
leads to the production of energy. Microorganisms can then use this energy to grow, and the
generation of more cells would increase the rate of perchlorate reduction. This scenario suggests that
nitrate is more important for microbial growth than ammonia.
It is also possible that the decrease in perchlorate reduction ability was induced by the
85

suppression of the active perchlorate-reducing enzyme due to a lack of nitrate reduction activity. The
enzymes involved in nitrate reduction may also be responsible for perchlorate reduction, and nitrate
reduction may be required to maintain the activity of these enzymes Without nitrate reduction, these
enzymes may have become inactive.
Perchlorate Removal Versus EBCT with No Influent Nitrate
As an extension of the nitrate preload experiment, perchlorate removal versus EBCT without
an influent nitrate spike was investigated. EBCTs of 25-, 15-, 10-, and 5-minutes were used,
respectively. Each EBCT was first run without any nitrate in the influent. The influent parameters
were as follows: DDW, C1O4" = 50 ug/L, NO3" = 0.1 mg/L (ambient), DO = 2.5 mg/L, phosphate
buffer = 1 mM, and pH 7.5. Electron donor (acetate, benzoate, and pyruvate) was supplied at 1.5
mg/L of equivalent nitrate demand in excess of that required to remove all influent DO. Figure 3.35
illustrates the results of this experiment.
Each EBCT was run for two weeks, which allowed the microbial community time to adjust
to changes in flowrate. Figure 3.35 shows that, on average, perchlorate removal declined as EBCT
1.2
«j uH
|
H a

was decreased. It is interesting to note, however, that during the 10-minute EBCT, perchlorate
removal steadily declined. For the first few days of the 10-minute experiment, approximately 70
percent of the influent perchlorate was being removed. By the end of the two-week period,
perchlorate removal had dropped to about 10 percent. Then, the 5-minute EBCT showed essentially
no perchlorate removal. To determine if this loss in perchlorate reduction capacity was solely a
function of EBCT, the 25-minute EBCT experiment was rerun after the 5-minute experiment was
complete. Figure 3.36 shows the results of this experiment (25-minute EBCT 2) and compares it to
the 25-minute experiment run previously (25-minute EBCT 1). The second 25-minute EBCT was
run approximately six weeks after the end of the first 25-minute EBCT experiment.
The data in Figure 3.36 clearly indicate that the filter had lost most of its ability to reduce
perchlorate. Average perchlorate removal for the first 25-minute experiment was approximately 80
percent. Six weeks later, the second 25-minute experiment showed average perchlorate removal
levels of less than 5 percent. There are three possible explanations for this phenomenon and are the
same as those described for the nitrate preload experiments. The first relates to the nitrogen
requirements of microbial communities. It is possible that the microbial community in the filter relied,
in part, on assimilative nitrate reduction to produce the nitrogen they require for cell growth.
1 .Z
Effluent Perchlorate,
1.0 -
0.8 -
I 0 - 6 -
0.4 -
0.2 -
n n
*— ——— ^— --— -"""
-«— 25 -Minute EBCT 2
-H&- 25 -Minute EBCT 1
^^a — — 0 —— n ——— B ——— B^^
100 200 300 400 500 600
Bed Volumes
700
Figure 3.36 Comparison of perchlorate removal for 25-minute EBCT experiments
(initial conditions: DDW, 2.5 mg/L DO, 50 ug/L C1O4", 0 mg/L NO3", 1.5 mg/L as
C added electron donor, pH 7.5)
87

Without this nitrogen, the community may have died. There is some evidence to suggest, though,
that this may not entirely explain the loss of perchlorate reduction ability. Throughout the
experiment, bromate was added to the influent at a concentration of 20 ug/L. During the first 25-
minute EBCT experiment, influent bromate was reduced to below detection in the filter. During the
second 25-minute experiment, an average of 63 percent of the influent bromate was removed. While
this clearly indicates a decrease in the filter's ability to reduce bromate, significant bromate reduction
ability still existed even after operating without an exogenous source of nitrogen for over two months.
This suggests that endogenous respiration was providing some of the required nutritional nitrogen.
It is possible that the perchlorate-reducing organisms are different than the bromate-reducing
organisms and that the perchlorate-degraders diminished because they were outcompeted for
nitrogen.
A second possible explanation is that there was plenty of nitrogen available to the perchloratedegraders
and that the loss of perchlorate reduction ability was caused by suppression of the
perchlorate-reducing enzyme due to a lack of nitrate reduction. The enzymes responsible for nitrate
reduction may also be involved in perchlorate reduction, and nitrate reduction may be required to
maintain the activity of the perchlorate-reducing enzymes.
A third possible explanation for the loss of perchlorate reduction ability is that the perchloratereducing
microorganisms grow via dissimilative nitrate reduction, and a lack of nitrate reduction
could reduce the population of perchlorate-reducers.
Filter Recovery
In an attempt to recover biological perchlorate reduction in the filter, nitrate was again added
to the influent (Figure 3.37). During the first 1,700 BV, the C/C0 was greater than 1 as perchlorate
desorbed from the GAC. During the previous period when no nitrate was added to the influent, the
GAC ion exchange sites equilibrated with an influent nitrate concentration of 0.1 mg/L, thereby
releasing adsorbed nitrate from the GAC. Perchlorate could then adsorb to these sites. When nitrate
was reintroduced to the influent during the filter recovery experiments, nitrate replaced perchlorate
on the ion exchange sites causing C/C0 for perchlorate to be greater than 1 from 0-1,700 BV.
Figure 3.37 clearly indicates that the filter is regaining its ability to reduce perchlorate. The
actual amount of biological perchlorate reduction that was occurring in the filter throughout the
88

1
1.6
1.4 -
1.2 -
.1.0-
1.5mg/L
Influent NO3" = 1.0 mg/L 0.5 ng/L
0.8 -
0.6 -
w
0.4
0.2 H
0.0
0 1000 2000
Bed Volumes
3000 4000
Figure 3.37 BAG filter recovery experiments (initial conditions: DDW, 2.5 mg/L
DO, 50 ug/L C1CV, 1.5-1.8 mg/L as C added electron donor pH 7.5, EBCT = 25
minutes)
recovery experiments was not detectable due to the fact that biological reduction may have been
masked by desorption. That is, biological reduction may have been occurring throughout the
recovery experiments but could not be measured, as perchlorate was simultaneously desorbing from
the GAC. Experiments will be run to further investigate why the BAG filter's perchlorate
reduction ability deteriorated when nitrate was removed from the influent. These experiments will
isolate whether the perchlorate-reducing microorganisms in the filter simply died due to nitrogen
deprivation, lack of energy for growth, or loss of enzymatic activity caused by a lack of nitrate
reduction.
Batch Biological Experiments
Plating Experiments and Serum Bottle Tests
Twelve isolates which had been growing on agar plates were inoculated into aqueous medium,
as described in Chapter 2. Table 3.15 shows the results of samples taken 3 weeks after the aqueous
medium was inoculated.
89

While some of the isolates reduced small amounts of perchlorate, isolate B9 reduced
perchlorate to below the detection limit. B9 was then used in two sets of serum bottle experiments.
At the inception of the first serum bottle experiment, it was not known how fast isolate B9 would
reduce perchlorate. Therefore, as sample volume was limited, samples were taken on the third, sixth,
and eighth day after inoculation. Even after the eighth day, the bottles containing CUW and no
external electron donor showed no perchlorate removal. The biodegradable fraction of NOM does
not suffice as an electron donor for perchlorate reduction. The DDW serum bottles did demonstrate
perchlorate removal. By the third day, each bottle containing an initial nitrate concentration of 0.3
Table 3.15 Reduction of perchlorate in aqueous medium using microbial isolates
(initial conditions: DDW, 46.3 ug/L CIO/, 1.0 mg/L NCV, 0 mg/L DO, pH 7.5)
Isolate
Control
(no inoculum)
Bl
B2
B3
B4
B5
B6
B7
B8
B9
BIO
Bll
B12
cio4-
(Hg/L)
46.5
45.2
47.3
42.5
46.4
45.3
45.7
42.5
45.1

or 1.5 mg/L nitrate showed no perchlorate remaining. The bottles using benzoate with 0 mg/L nitrate
removed 50 percent of the influent perchlorate by the third day while the bottles using lactate with
0 mg/L nitrate removed 17 percent of the influent perchlorate during the same time period. These
data suggest that pechlorate removal kinetics can be enhanced by the presence of nitrate and that B9
might prefer benzoate slightly over lactate as an electron donor. There were no differences in
perchlorate removal performance between the bottles containing bromate and those without bromate.
Thus, it can be concluded that perchlorate and bromate reduction are independent of each other.
The second serum bottle experiment was designed to further test the differences in perchlorate
reduction kinetics caused by differences in initial nitrate concentration. Figures 3.38-3.41 show
perchlorate and nitrate removal versus incubation time.
0123456
Incubation Time (days)
0.0
Figure 3.38 Isolate B9 serum bottle experiment 2 (initial conditions: DDW, 50
ug/L C1CV, 0.1 mg/L NO3-, DO ^0 mg/L, pH 7.5)
91

~\
I
0
012345
Incubation Time (days)
Figure 3.39 Isolate B9 serum bottle experiment 2 (initial conditions: DDW, 50
Hg/L CIO/, 0.4 mg/L NO3', DO ^0 mg/L, pH 7.5)
2.5
0 1
23456
Incubation Time (days)
Figure 3.40 Isolate B9 serum bottle experiment 2 (initial conditions: DDW, 50
Hg/L C1CV, 1.6 mg/L NO3', DO *0 mg/L, pH 7.5)
0
92

^——s
0 4
0 234
Incubation Time (days)
Figure 3.41 Isolate B9 serum bottle experiment 2 (initial conditions: DDW, 50
\igfL C1O4', 2.6 mg/L NO3', DO ^0 mg/L, pH 7.5)
The first observation that can be made from Figure 3.42 is that isolate B9 does have the ability to
reduce nitrate. Consequently, there is at least one microbial species in the BAG filters that has the
ability to reduce both nitrate and perchlorate. Another interesting observation from these plots is that
significant perchlorate removal did not occur in any of the bottles until the nitrate concentration was
^0.1 mg/L. This falls in the range of nitrate concentrations required for greater than 95 percent
perchlorate removal in the BAG filters, as determined during the nitrate isolation experiments.
Nitrate and perchlorate degradation kinetics are compared in Figures 3.42 and 3.43,
respectively. Figure 3,42 also shows that during the first 1.5 days of incubation, the nitrate was not
reduced by isolate B9. It is possible that there was some residual DO in the bottles that may have
inibited the reduction of nitrate. Once nitrate degradation began, however, B9 required less than one
day to reduce the nitrate concentration to 0.1 mg/L or lower, regardless of the initial nitrate
concentration.
93

j.u -
f-x — x —— x x-x
2.5 -
ll 2.0-
S
u 1.5 -
1
g i.o -
"A — A ——— &-A-A \
0-5 - İ
•• • ^ '*^*, • •.-• H ' i • ——
() 1 2 3 4 5 f
Incubation Time (days)
Figure 3.42 Isolate B9 serum bottle experiment 2: nitrate degradation (initial
conditions: DDW, 50 ng/L CIO/, DO ^0 mg/L, pH 7.5)
0.1mg/LNO3-
0.4 mg/L N03
1.6 mg/L NO3.
2.6 mg/L N03.
0
0 1 23456
Incubation Time (days)
Figure 3.43 Isolate B9 serum bottle experiment 2: perchlorate degradation (initial
conditions: DDW, 50 ng/L C1O4', DO ^0 mg/L, pH 7.5)
94

Figure 3.43 illustrates the differences in perchlorate reduction rates when the initial nitrate
concentration was varied. As the initial nitrate concentration increased, the rate of perchlorate
reduction increased. B9 required approximately 3 days to reduce perchlorate to below the LOD
when the initial nitrate concentration was 1.6 or 2.6 mg/L, while it took over 5 days when the initial
nitrate concentration was 0.1 or 0.4 mg/L. Similar to the BAG filter experiments, there are three
possible explanations for this: 1) increased growth due to dissimilative nitrate reduction, 2) increased
growth due to assimilative nitrate reduction to ammonia (nutrient requirements), and 3) increased
perchlorate-reducing enzymatic activity due to increased nitrate reduction.
Dilution-to-Extinction Experiments
Table 3.16 shows the results of the perchlorate dilution-to-extinction experiments. After six
weeks, the microorganisms reduced perchlorate to below detection in every dilution except 1:100.
Perchlorate reduction may occur at a high dilution and not at a low dilution because at the low
dilution there may be non-perchlorate reducers present that outcompete the numerically dominant
perchlorate-reducers for substrate. This can occur if the non-perchlorate reducers are faster growers
than the perchlorate reducers. At the higher dilutions, these non-pechlorate reducers are diluted out,
leaving the numerically dominant perchlorate-reducers alone in the sample.
Table 3.16 Dilution-to-extinction experiments: perchlorate (initial conditions: DDW, 50 |u.g/L
C1CV, 20 ng/L BrCV, 1.0 mg/L NO3", and 5 mM phosphate buffer
Dilution
C104-(ug/L)
t = 6 weeks
C104- (ng/L)
t = 8 weeks
N03- (mg/L)
t = 8 weeks
N02- (mg/L)
t = 8 weeks
Br0 3-(ug/L)
t = 8 weeks
Control -
no inoculum
53.1
Not
measured
1.05

After eight weeks, the microorganisms reduced all of the bromate and almost all of the nitrate,
and significant nitrite accumulation did not occur except in the zero dilution Balch tubes. The nitrite
degraders may have been outcompeted by other microorganisms under the zero dilution condition.
Microscopic Examination of Microorganisms in the Filters
Biomass from the perchlorate-reducing BAG filters has been examined microscopically.
Three representative pictures are shown in Figures 3.44-3.46. The microscopy was done using phase
contrast with lOOOx magnification. Therefore, the dark objects in the pictures are microorganisms
while the lighter, retractile images are likely debris.
Rods were the most common bacterial morphology observed. The dark specks in Figures
3.44 and 3.45 are these rod-shaped bacteria. They were observed to exist most often as single
organisms (Figure 3.44) but were also found in floe form (Figure 3.45).
Some common morphologies were observed between the biomass from the bromate and
perchlorate BAG filters. Protozoa shaped like round-bottom flasks are one example of this (Figures
3.21 and 3.44). Other morphologies found in the perchlorate samples were clearly distinctive. The
best example of this is the large protozoan that was observed in biomass taken from the the filter of
the DDW apparatus (Figure 3.46). The micrograph shows about one-half of the protozoan. The
small black dots inside this protozoan are bacteria which it has consumed - evidence of microbial
grazing. From this micrograph, it can be noted that this particular protozoan is extremely large -
thousands of times larger than the typical bacteria found in these filters.
96

Figure 3.44 Rods and protozoa from the BAG filter
Figure 3.45 Bacterial floe from the BAG filter
97

Figure 3.46 Large protozoan from the BAG filter
Denaturing Gradient Gel Electrophoresis Experiment
Figure 3.22 shows the DGGE profile obtained with material harvested from various BAC
filters. Each sample was run in duplicate. The first lane is a standard sample. Lanes 2 and 3 were run
with a fresh microbial sample from the effluent side of the CUW perchlorate filter that used CUW for
one year. The microbial sample used in lanes 4 and 5 came from the effluent side of the DDW filter.
Similar to the DGGE profiles generated using samples from the BAC filters operated for bromate
removal, the profiles for the microbial community from the perchlorate-reducing filters reflect a
diverse array of microorganisms. There are conserved bands across the two different samples, while
there are some bands that are unique to each sample.
98

CHAPTER 4. ACTIVITIES AT THE METROPOLITAN WATER DISTRICT
OF SOUTHERN CALIFORNIA
INTRODUCTION
Bench- and pilot-scale studies were conducted at the Metropolitan Water District of Southern
California (MWDSC) to investigate the reduction of perchlorate and bromate in biologically active
carbon (BAG) filters. The foundation for these studies was developed at the University of Illinois
at Urbana-Champaign (UIUC). Work done at UIUC demonstrated that efficient biological
perchlorate and bromate reduction could be achieved and sustained using BAG filtration. Natural
organic matter (NOM) and exogenous organic acids served as the electron donors for bromate and
perchlorate reduction for the bench-scale UIUC experiments, respectively. One unique aspect of the
work done at the MWDSC is the fact that influent water was ozonated to provide the electron donor
necessary for biological reduction. Ozonation of natural water can break large NOM molecules down
into more readily assimilable organic compounds such as aldehydes and carboxylic acids. Other
biological perchlorate treatment processes reported in the literature require the addition of an
exogenous electron donor to support biological reduction (Logan et al. 1999).
Tests were performed at the MWDSC using blends of Colorado River Water (CRW) and
State Project Water (SPW). Results at the bench-scale were used to generate the design of pilotscale
BAG filters, which were utilized to examine the issue of scale-up for biological perchlorate and
bromate reduction.
OBJECTIVES
The objective of the work done at the MWDSC was to investigate the reduction of
perchlorate and bromate in B AC filters after ozonation. Specific goals of the project were as follows:
To demonstrate that low ug/L concentrations of perchlorate and bromate can be
biologically reduced to below detection using the ozonation/activated carbon process
for CRW/SPW blends at the bench-scale.
• To investigate the mechanism of biological reduction, with special focus on the effects
of perchlorate, bromate, nitrate, and dissolved oxygen (DO) concentrations.
• To determine the optimal operating conditions to biologically reduce perchlorate and
bromate using BAG filtration.
• To examine perchlorate and bromate removal using BAG filtration at the pilot-scale.
99

BENCH-SCALE MINI-COLUMN STUDY
Materials and Methods
Water
Currently the La Verne Pilot Plant at the Metropolitan Water District of Southern California
(MWDSC) receives a blend of 25 percent State Project Water (SPW) and 75 percent Colorado River
Water (CRW). Typical water characteristics are given in Table 4.1 based on historical data. More
detailed water quality parameters are presented in a later section.
The raw water blend was ozonated with approximately 2 mg/L of applied ozone using 100-
gallon, 6-inch ozone contactors located at the La Verne Pilot Plant. Compressed air was used as a
feed gas for the ozone generator. The initial DO concentration using compressed air as a feed gas
yielded approximately 8-9 mg/L of DO after the ozonation process. The batch of ozonated water was
spiked with sodium thiosulfate (Certified ACS, Fisher, Pittsburgh, PA) to reduce the DO level, and
the amount of sodium thiosulfate varied depending on the desired DO level (23-160 mg/L sodium
thiosulfate to maintain 0-5 mg/L DO).
Table 4.1 Historical water quality for Metropolitan source water
Parameter Typical Level at Metropolitan
Alkalinity 13 0 mg/L as CaCO3
Bromide between 0.07 and 0.20 mg/L
Bromate ND (< 3 (ig/L) - 55 ug/L (2 mg/L ozone dose)
Perchlorate ND (< 4 ug/L) - 9 ug/L for CRW
ND for SPW
Nitrate 0.25 to 0.40 mg/L as N
Color 3 cu
Iron 0.020 mg/L
Manganese 0.005 mg/L
pH ~8
UVA 0.1 cm- 1
TOC 2.5 mg/L ____
100

Perchlorate is only found in CRW, and the background level concentration ranged from < 4
Mg/L to 9 ug/L (Table 4.1). Thus, perchlorate was spiked (10 to 50 ng/L) to the ozonated water for
the bench-scale study. Flowrates of 1.6 and 2.7 mL/minute were used to achieve 25- and 15-minute
empty-bed contact times (EBCTs), respectively, for a single filter and 50- and 30-minute EBCTs,
respectively, for two filters in series.
Column Setup
The bench-scale columns were constructed in May 1999 at the La Verne Pilot Plant, and have
been in operation since the beginning of June 1999. The columns were modeled after those already
in use at the University of Illinois at Urbana-Champaign in terms of physical setup and flow
characteristics.
Two 60-cm glass columns with 2.5-cm inner diameter (Sigma, St. Louis, MO) were used as
filter columns. The columns have V4-28 PTFE fittings, and Tygon® tubing was used to connect the
columns and the reservoir. The reservoir was constructed from a 15-gallon Low Density
Polyethylene (LDPE) tank with a spigot, and a commercially available plastic device was used as a
floating cover with 1/16" clearance with the reservoir wall. The schematic of the bench-scale setup
is shown in Figure 4.1.
Spent Norit GAC 8x12 (Norit America Inc., Atlanta, GA) was taken from the La Verne Pilot
Plant filters, which had been in service for 18 months, and was used to pack the mini-columns. Each
column was packed with 1 cm of 3-mm glass beads at the top and bottom of an 8-cm bed of GAC.
After spiking the water with 10 to 50 ng/L of perchlorate (sodium perchlorate, Sigma
Chemical Co. St. Louis, MO), a floating cover was placed into the reservoir. The reservoir water was
changed once a week with freshly ozonated water.
101

Effluent
Sample
Bottle
Mini
Column
Floating
Cover
Peristaltic
Pump
Reservoir
Mid Sampling Point
Figure 4.1. Schematic of the bench-scale filter setup at Metropolitan (not to scale)
Sampling
The ozonated water from the bench-scale set-up was analyzed for perchlorate, bromate, and
ultraviolet (UV) absorbance, total organic carbon (TOC), conductivity, and pH. Aldehydes and
carboxylic acids were also measured once bromate and perchlorate reduction were established in the
columns. When appropriate, the concentrations of DO and nitrate were also analyzed. Samples were
collected at the influent, after the first column (midpoint), and after the second column (effluent), as
shown in Figure 4.1.
Analytical Methods
All inorganic and organic analyses were performed at Metropolitan's Water Quality
Laboratory in accordance with the procedures described in Standard Methods (1995), except as
noted in the following sections.
102

Residual ozone. To reduce sample overbleaching and dilution problems during residual
ozone measurements, the gravimetric method was used, which incorporates a slight modification of
the indigo colorimetric method (Standard Methods, 1995). With this method, the sample was
collected in a 125-mL Erlenmeyer flask containing 10 mL of indigo trisulfonate reagent to decolorize
the solution to a light blue. The exact sample volume was gravimetrically determined by comparing
the total weight of the indigo-sample mixture and the flask with the initial weight of the indigo and
flask. This procedure reduced the time needed to collect and analyze the samples. The sample
absorbance was measured by a spectrophotometer (Lambda-SB model; Perkin-Elmer Corp.,
Norwalk, Conn.) with a 1-cm light path length, and the measurement was corrected for the
absorbance of background organics at 800 nm.
Perchlorate. Perchlorate samples were analyzed using an ion chromatograph (model DX300,
Dionex Corp., Sunnyvale, Calif.) modified with a 200-|oL sample loop. An ion-chromatography (1C)
analytical column (model AG11, Dionex Corp.), an anion micromembrane suppressor, and
a conductivity detector were used. The method detection limit was 1.4 ng/L, and the reporting limit
was 4.0 ug/L.
Bromide. Bromide analyses were conducted on an ion chromatograph (model 2010, Dionex
Corp., Sunnyvale, Calif.) with a 20- or 50-uL sample loop. An 1C analytical column (model AS4A/
Dionex), an anion micromembrane suppresser, and a conductivity detector were used. The method
detection limit was 0.01 mg/L, and the reporting limit was 0.02 mg/L.
Bromate. Bromate analyses were performed using a modification of an 1C method described
by Pfaff and Brockhoff (1990). The modified method is described by Kuo et al. (1990). An 1C
analytical column (AS4A, Dionex), an anion micromembrane suppressor, and a conductivity detector
were used. The method detection limit was 1.0 ng/L, and the reporting limit was 3.0 ug/L.
Aldehydes. Formaldehyde and acetaldehyde were analyzed by a derivatization-extraction gaschromatographic
(GC) electron-capture detection method described by Glaze and colleagues (1989).
Other aldehydes, such as glyoxal and methyl glyoxal, were analyzed by a modification (heated
derivatization) of this method, as described by Sclimenti and colleagues (1990).
103

Carboxylic Acids. Carboxylic acids were measured using ion chromatography as described
by Coffey et al. 1997. Organic acids were separated on a high-capacity ion exchange Dionex ASH
column, using a sodium hydroxide solution eluent in the gradient mode, followed by conductivity
detection. The individual method detection limits for each of the three carboxylic acid species
analyzed (acetate, formate, and oxalate) was 15 ug/L.
Dissolved Oxygen. The modified Winkler or iodometric method was used to measure the DO
concentration in water. The samples were analyzed according to the Standard Method 4500-O F
(Sulfamic acid flocculation modification). A DO probe was not used since high backgrounds of
sodium thiosulfate confounded the measurement.
pH. pH was measured using a pH meter with an accuracy of ±0.002 pH units (model 920A;
Orion Research, Inc., Boston, Massachusetts).
Results and Discussion
Perchlorate and Bromate Reduction
The removal of perchlorate and bromate has been observed since the third week
(approximately 660 B Vs) of the mini-column experiment. The columns continued to show the
reduction of perchlorate and bromate after 10 months of operation under various conditions. Figures
4.2 and 4.3 show the removal of perchlorate and bromate by the mini-columns. These figures are
divided up into regions, and different regions are compared in the following sections. During these
tests, there were few instances when the influent perchlorate concentration decreased in the reservoir
(influent sample). This occurred when a freshly ozonated water supply was not available due to pilot
plant shut down. In this case, the influent water was in the reservoir for more than a week (up to 10
days), and the influent concentration is plotted separately (denoted as -7th day) from the regular
influent sample. The influent perchlorate concentrations for these cases were about half the target
value. During the cleaning of the reservoir after these periods when the influent was unchanged for
over a week, a thick, yellowish biofilm was observed at the bottom of the reservoir. This biofilm may
have been responsible for the premature degradation of perchlorate and bromate.
As shown in Figure 4.2, after three months (-2,800 BVs) of mini-column operation, the
influent concentration of perchlorate was reduced from 50 to 10 ng/L (regions 1 and 2). Complete
perchlorate reduction continued to take place. From 3,500 to 5,000 BV, the effect of EBCT was
104

^ influent 0 effluent A mid • inf (~7th day)
Region
1
2
3
4
5
6
7
8
Initial Conditions
C1O4 EBCT DO NO,

studied with an influent perchlorate concentration of 50 ug/L (Figure 4.2, region 3). The influent
flow rate was increased to 2.7 mL/min to achieve a 15-min EBCT after the first column (mid-point).
This condition was tested for approximately one month to confirm the findings, and the condition was
repeated with a lower influent perchlorate concentration of 10 ug/L (region 4 in Figure 4.2). Under
each of these conditions, complete removal of perchlorate was achieved.
Bromate is formed during ozonation at the La Verne pilot plant, and therefore bromate is
present in the influent to the BAG filters. Average bromate formation has been 17 ug/L with a
maximum concentration of 55 ug/L. As found for perchlorate, bromate removal by the mini-columns
has also been observed. Complete bromate removal has been observed for 15- and 25-minute EBCTs
(Figure 4.3).
Effect of Dissolved Oxygen
The effect of DO concentration was also studied in the bench-scale setup using a 15-minute
EBCT. When a low influent DO concentration (0.6-2 mg/L) was used (Figure 4.2, region 5 and
Figure 4.3, region 3), complete perchlorate and bromate removal was observed except for one
perchlorate sample. When the influent DO concentration was increased to 5 mg/L (Figure 4.2, region
6), mid-point samples showed effluent perchlorate concentrations between 8 and 13 ug/L, indicating
that perchlorate removal declined as the influent DO concentration was increased. By contrast, when
the influent DO concentration was increased to 5 mg/L (Figure 4.3, region 4), bromate was removed
to the detection limit.
After the tests at 5 mg/L influent DO were complete, the columns were equilibrated with a
zero influent DO condition to allow for the examination of nitrate effects without the occurrence of
any residual DO effects. Influent perchlorate was also increased from 10 to 50 |jg/L during this
equilibration period. Four days after the new condition was imposed, less than 10 percent perchlorate
reduction was observed in the 15-minute EBCT filter (Figure 4.2, region 7). However, by the second
week of equilibration, complete removal of influent perchlorate was again observed in the 15-minute
EBCT filter.
106

Effect of Nitrate
The background nitrate concentration in the source water at Metropolitan is typically very
low. The influent nitrate concentration was below the detection limit (0.05 mg/L) in the bench-scale
setup except for a few samples during the experiments. There were three instances where nitrate was
detected in the influent, and the values ranged from 0.2 to 1.0 mg/L. Consequently, the filters were
exposed to only low nitrate concentrations for the first 8,000 B Vs. Nitrate was spiked to the influent
at concentrations of 4.0 and 8.3 mg/L for a two-week period with zero influent DO (Figure 4.2,
region 8 and Figure 4.3, region 5). During these experiments, complete removal of nitrate,
perchlorate, and bromate was observed. With complete nitrate removal in the BAG filters,
perchlorate and bromate removal continued unperturbed by the elevated influent nitrate
concentration.
DO Control with Sodium Thiosulfate
In order to facilitate biological utilization of perchlorate and bromate, a zero or low influent
DO concentration was often maintained. Purging of the influent with gaseous nitrogen to remove
DO was tested but not used because of the possibility of also purging ozonated by-products needed
as electron donors. Thus, the DO concentration was chemically controlled with sodium thiosulfate,
but there have been some complications in its use. The fast reaction of thiosulfate with DO yields
tetrathionate (S 4O62~), and this stoichiometry was used to calculate the mass of thiosulfate needed for
DO reduction. Based on this stoichiometry (equation 4.1), the sodium thiosulfate dosage required
to reduce the DO concentration to 0 mg/L was approximately 160 mg/L.
4S2O32" + O2(aq) + 4H+ - 2S 4O62' + 2H2O (4.1)
However, tetrathionate can slowly be oxidized to sulfate as it reacts with DO. Assuming
oxidation of thiosulfate to sulfate, the dose of thiosulfate needed for DO reduction is only one eighth
of the amount calculated by equation 4.1 (20 mg/L). However, reduction of the DO concentration
to 0 mg/L could not be achieved for the latter case within a reasonable time period. After 5 days,
with a 20 mg/L dose of sodium thiosulfate, the DO concentration was 5 mg/L in a study done using
107

the Winkler method (Standard Method 4500-O F) to measure DO. The fast kinetics reaction model
was used and tested to determine the amount of sodium thiosulfate spike to prepare a batch of water.
However, tetrathionate can be oxidized further to sulfate, as it reacts with dissolved oxygen,
and the amount of sodium thiosulfate spike based on this stoichiometry is only 1/8 of the amount
compared with the reaction shown above. However, the reduction to zero DO level in the influent
could not be achieved for the latter case within the reasonable time frame of the experiments. After
5 days, with 20 mg/L sodium thiosulfate dose, the DO concentration was 5 mg/L. Therefore, the
kinetically fast reaction (equation 4.1) was used to determine the sodium thiosulfate spike for DO
control.
One concern with high concentrations of sodium thiosulfate in the reservoir is the possibility
of chemical reduction of perchlorate by sodium thiosulfate. Three control studies were done to verify
that thiosulfate did not reduce perchlorate, and a summary of these results is given in Table 4.3. In
the deionized (Super-Q) water, no perchlorate reduction was found over a two-week period.
Similarly, no perchlorate reduction was observed when this experiment was repeated using an
ozonated blend of CRW/SPW. In the third control study, the same spent GAC used for the minicolumns
was used to determine if thiosulfate can reduce perchlorate in the presence of GAC. The
GAC used for this study was probably not biologically active since it had not been used for drinking
water filtration for about 6 months prior to this experiment. It was found that perchlorate reduction
did not take place in the presence of thiosulfate and GAC.
108

Table 4.2 Control study for chemical reduction pathway for perchlorate (initial values N/A)
Reaction time Water Sodium
(week) thiosulfate used
1
1
Super-Q
Super-Q
no
yes
GAC used
no
no
CKV(ng/L)
173
174
2
Super-Q
yes
no
208
1
Ozonated CRW
no
no
180
1
Ozonated CRW
yes
no
182
2
Ozonated CRW
yes
no
221
1
Super-Q
yes
yes
126
1
Super-Q
no
yes
142
1
Ozonated CRW
yes
yes
100
Ozonation By-products
Carboxylic acids and aldehydes were analyzed from the mini-column experiments since the
third month of filter operation. There has been no clear decrease in carboxylic acids across the filters.
Some of the aldehydes decline in concentration across the filters, which may be indicative of aldehyde
use as an microbial electron donor. However, interpretation of the data was potentially confounded
by 1) difficulty in maintaining a stable concentration of ozone by-products in the water, 2)
interference of sodium thiosulfate with aldehydes during GC analysis, and 3) changes in sodium
thiosulfate concentration as the reaction occurs between the reservoir and the columns. A more
detailed and controlled study is required to isolate and identify the by-products used in the BAC
filters.
TOC and 254
During the first three months of the experiment, the effluent total organic carbon (TOC)
concentration from the BAC filters was approximately 30 percent higher than the influent
concentration. Typical influent and effluent TOC concentrations were 1.8 and 2.4 mg/L, respectively.
109

This increase in TOC concentration across the filters may have been due to biofilm sloughing from
the spent GAC. Since the third month of operation, the average influent and effluent TOC
concentrations have been approximately 2.11 and 2.21 mg/L respectively, which is only a 5 percent
increase.
The UV absorption at 254 nm (filters were not filtered) of the filter effluent, 0.11 cm" 1 , was
approximately 40 percent less than the average influent value (0.18 cm" 1) for the first three months
of the experiment. After this time, the average UV254 values have been 0.13 cm" 1 at the filter influent
and 0.11 cm' 1 at the effluent, which is a decrease of about 15 percent across the filter.
The historical sulfate concentration found in Metropolitan's source water is approximately
250 mg/L, which is also the average value measured in the influent of the mini-column four days after
dosing the reservoir with sodium thiosulfate. This indicates that sodium thiosulfate is not oxidized
to sulfate in the reservoir within four days. However, the sulfate concentration is higher at the mid
point (13 percent higher than the influent concentration) and in the effluent from the second filter (20
percent higher than the influent concentration). The average effluent sulfate concentration from the
second filter is about 300 mg/L. Tetrathionate formed in the reservoir could be oxidized to sulfate
in the presence of GAC, thus elevating the effluent sulfate concentrations.
Sulfur-based reducing chemicals can be effective in reducing the influent DO concentration,
but taste-and-odor problems associated with sulfide may be a concern. For instance, hydrogen sulfide
has been observed effluent samples from the mini-columns. Post-treatment could be employed to
remove hydrogen sulfide, but it may be possible to use other reducing agents that will not cause tasteand-odor
problems. The use of ascorbic acid would facilitate faster reduction of DO; this could
potentially decrease the amount of chemical required for DO reduction and reduce the overall
chemical costs. However, ascorbic acid was not used in this study since ascorbic acid could break
down in the reservoir or filters and provide an additional electron donor source.
110

PILOT-SCALE MOBILE PILOT PLANT STUDY
Materials and Methods
Water
The typical blend of water to the mobile pilot plant is 25 percent SPW and 75 percent CRW.
This is the same blend of water used for the mini-column study. The mobile pilot plant has the
capability to supply 100 percent CRW in case the SPW supply to the mobile pilot plant is shut down
during regular maintenance or for special testing.
Mobile Pilot Plant Setup
The mobile pilot plant, located at the MWDSC's La Verne facility (Figure 4.4), consists of
two parallel trains, which include ozone contactors and conventional treatment. The raw CRW or
CRW/SPW blend is treated by 1) pre-ozonation, 2) rapid mix (high-intensity mixing with addition of
coagulant and cationic polymer), 3) flocculation (slow mixing of water in combination with coagulant
and polymer in order to form floes), 4) sedimentation (settling of floe), and 5) biological filtration (no
Figure 4.4 Mobile pilot plant
111

DO residual across the filters, allowing for biodegradation of perchlorate and bromate).
A schematic of the mobile pilot plant is shown in Figure 4.5, and the arrow indicates
direction of water flow. The flow to the ozone contactor is equal to the flow rate to the head of the
train, which is set at 1.4 gallons per minute (gpm). When small shells or debris clog the pre-filter
outside of the mobile pilot plant, the flow rate decreases; thus the filter is checked and cleaned
regularly.
Mobile Pilot Plant Filter
A 10-ft column with an inner-diameter of 4 inches was filled with 4 feet of virgin Norit
carbon (Hydrodarco GCW 10 X 20). The flow rate to the column is set at 0.1 gpm to attain an
EBCT of 25 minutes. The characteristics of the virgin carbon are given in Table 4.3.
Influent
Alum,
Polymer
Perchlorate
Bromate
Spike
Biological
Filter
Sodium
Tbiosulfate
Spike
CD CD
Air O 3
Compressor Generator O 3 Contactor
Rapid Flocculators Sedimentation
Mix Basin
Effluent
Figure 4.5 Schematic of the mobile pilot plant setup at Metropolitan (not to scale)
Table 4.3 Characteristics of Norit carbon used in mobile pilot plant
______(Norit Americas, Inc., GA)______
Characteristics
Value
BET surface area 1025 m2/g
Average particle diameter
1 mm
Uniformity coefficient
1.2
Total pore volume 0.8 mL/g
Apparent density 0.48 g/mL
112

Mobile Pilot Plant Ozone
The ozonation process consisted of an ozone contactor column and an ozone reactor column.
The raw water was pumped to the top of the ozone contactor at 1.4 gpm, and ozone was applied in
countercurrent fashion. Ozone was bubbled through 0.5-in diameter (1-in long) stainless steel static
diffiisers to generate fine bubbles. The ozone generator used was a tube-type, water-cooled unit
operating on a low-frequency (60 Hz), variable-voltage power supply (model Labo-76,
Praxair/Trailigaz Ozone Co., Cincinnati, Ohio). The ozone contactor column was operated with a
1.2-minute residence time, and the ozone reactor had a 4.7- minute residence time. The water level
was maintained at 10 feet in the 2-inch diameter contact column and in the 4-inch diameter reactor
column.
The ozone generator capacity is 1 Ib/day, with compressed air as the feed gas. The target gasphase
concentration of ozone produced is approximately 0.95 percent by weight. The ozone
concentration in the feed gas was measured by a gaseous ozone monitor (model HC-12, PCI Ozone
Corp., West Caldwell, NJ). Off-gas from the contactor flows through a thermocatalytic ozone
destruction unit prior to discharge to the atmosphere.
Backwash Setup
The backwash water supply setup has been modified to dechlorinate and deoxygenate the
water, so that the BAG filters are protected during the backwash cycle. The backwash water supplied
to the mobile pilot plant is treated Weymouth Plant utility water with approximately 2 mg/L of free
chlorine residual and an ambient DO concentration. The modification included a 55-gallon stainless
steel tank packed with virgin Norit carbon (HYDRODARCO GCW 10 X 20) for dechlorination, an
HOPE tank to spike sodium thiosulfate to deoxygenate the backwash water, and a pump to fluidize
the beds (Figure 4.6). The automatic backwash cycle was disabled to allow a plant operator to
manually backwash the filters at least once every 48 hours.
113

Sodium
Thiosulfate
to reduce
Treated
Backwas
Water
Weymouth
Plant
Utility
Water
GAG tank to
dechlorinate
Figure 4.6 Backwash water supply setup
Operating Conditions
The typical operating conditions for the mobile pilot plant are listed in Table 4.5. A
perchlorate spike of 500 u,g/L was used for the first few weeks to pre-load the virgin carbons. The
spike was subsequently reduced to 50 ug/L. Similarly, 250 (o,g/L of bromate was used for preloading,
and 25 |u,g/L was used during the experimental phase. Chemicals were continuously injected
using individual peristaltic pumps from the 5-gallon carboys. The secondary feed stocks for alum,
polymer (poly-DADMAC, NeoSolutions, Beaver, Pa.), perchlorate, and bromate were prepared
weekly. Sodium thiosulfate injection started after the exchange capacity of the carbon for perchlorate
and bromate was exhausted (~ 6000 BVs). When used, sodium thiosulfate was spiked after the
sediment basin, and three in-line static mixers (six mixing elements per mixer; Cole-Parmer, Vernon
Hills, 111.) were installed to provide adequate mixing so that zero DO was achieved before influent
water entered the filters.
114

Table 4.4 Mobile pilot plant operating conditions
Operating Conditions Current Setting
Influent flow rate ~1.4 gpm
Applied ozone 2 mg/L
Rapid mix influent flow rate 1 gpm
Perchlorate spike 50 jag/L
Bromate spike 0 to 25 (ig/L
Alum dosage 4.5 mg/L
Polymer dosage 1.5 mg/L
Filter influent flow rate 0.1 gpm
Sodium thiosulfate spike -160 mg/L
Influent DO concentration 0 mg/L
Sampling
For the pilot-scale study, perchlorate and bromate concentrations were analyzed at the influent
and effluent of the BAG filter. For selected tests, perchlorate and bromate were also analyzed after
the rapid mix step. Once complete perchlorate and bromate reduction was observed, samples were
also taken for UV, TOC, aldehydes, and carboxylic acids.
Results
Perchlorate and Bromate Preloading
Since the pilot plant filters were comprised of virgin carbon, the filters were subjected to
perchlorate and bromate preloading periods to exhaust the ion exchange capacity of the carbon. This
would ensure that subsequent removal was via biological reduction and not ion exchange. Table 4.5
shows the results of preloading the GAC in the pilot plant experiments. The virgin carbon effectively
removed perchlorate during the initial phase of operation and removed bromate to a lesser degree.
After a few weeks of preloading the carbon with 500 ng/L perchlorate, the effluent perchlorate
concentration from the GAC filter steadily increased.
115

Table 4.5 Preloading results from the mobile pilot plant
Day
1
2
3
4
8
9
Sampling Point
rapid mix effluent
filter influent
filter effluent
rapid mix effluent
filter influent
filter effluent
rapid mix effluent
filter influent
filter effluent
rapid mix effluent
filter influent
filter effluent
rapid mix effluent
filter influent
filter effluent
rapid mix effluent
filter influent
filter effluent
C1CV (ug/L)
632
667
ND
532
577
ND
592
572
ND
640
641
ND
580
580
250
590
570
230
Br03-(ug/L)
257
300
169
248
270
183
288
289
171
238
249
203
262
259
160
265
241
185
Since samples were diluted for ion chromatographic analysis, the ND indicated in Table 4.5 could be
as high as 40 u,g/L. After approximately one week of GAC preloading, perchlorate removal in the
filter decreased to about 60 percent. Bromate was also spiked to the water at 250 |ig/L. Bromate
removal declined steadily for the first few days, as indicated in Table 4.5. Perchlorate and bromate
concentrations were nearly constant from the rapid mix effluent to the filter influent, indicating that
virtually no perchlorate or bromate reduction was taking place during coagulation or sedimentation.
Figures 4.7 and 4.8 show the exhaustion of the GAC's ion exchange capacity for perchlorate
and bromate, respectively. Perchlorate and bromate concentrations of 500 u,g/L and 250 u,g/L were
used initially to minimize the time required to preload the GAC. When the perchlorate spike was
dereased to approximately 60 |ig/L (at 800 BV), the aqueous perchlorate concentration was lower
than the sorbed perchlorate concentration. This negative concentration gradient caused perchlorate
to desorb from the carbon, yielding C/C0 values greater than one. Though the influent bromate spike
was also decreased from its initial concentration, similar desorption behavior was not observed. Both
116

p o
O. i
Fraction Bromate Remaining
C/Co
n>
Fraction Perchlorate Remaining
C/Co
if
-•— ' O
CE. o'
§1
Ks* ^^
CZ • \J
ET o'
3 5T
I o'
2,
cr
3
aO)
O
O "-*>
•o
n>
3
o
09
a>
"2.
o"
r*
CO
O
a
iCO
I
CD

perchlorate and bromate reached equilibrium (C/C0 * 1) before the filters were switched to biological
mode (ozonation and deaeration). No desorption of perchlorate or bromate was observed when the
filters were operating in the biological mode.
Perchlorate and Bromate Reduction in BAC Filters
Approximately 3,000 BVs after the filter was switched to biological mode, greater than 87
percent perchlorate removal was observed (Figure 4.7). Figure 4.8 shows that biological bromate
reduction was also occurring in the filter. During the third week of operation, however, bromate was
not being formed during ozonation, necessitating the addition of a 25 u.g/L bromate spike. Bromate
removal continued to be very efficient.
At about 10,000 BV, the DO control system failed, subjecting the filter to ambient DO
concentrations (8-9 mg/L) for approximately 48 hours. Repairs were made and the thiosulfate spike
reduced the influent DO concentration back down to zero. During the two weeks following the
repair, perchlorate removal was less than 15 percent and bromate removal decreased to 75 percent.
These data are shown as open circles in Figures 4.7 and 4.8, respectively. Subsequently, complete
perchlorate and bromate removal was achieved in the filter. This lagged recovery period was
previously observed for perchlorate when the mini-columns were subjected to 5 mg/L DO during a
controlled study.
Both bench- and pilot-scale experimentation demonstrated efficient biological perchlorate and
bromate reduction in a BAC filter. These results suggest that ozonation followed by deaeration and
biological filtration could comprise an effective strategy for removing low u,g/L concentrations of
perchlorate and bromate from drinking water.
118

CHAPTER 5. SUMMARY AND CONCLUSIONS
BROMATE
A main purpose of this project was to investigate bromate reduction in BAG filters. BAG
filters were constructed at the University of Illinois at Urbana-Champaign to evaluate various
parameters affecting bromate reduction.
To rule out continued abiotic reduction of bromate in the BAG filters, BAG filtration
experiments were repeated one year apart. The same percent bromate removal was observed in
March 1998 as in February 1999 for a set of BAG filters. These results indicated that abiotic bromate
reduction was not significant in these experiments, since abiotic reduction could not likely be
sustained for 11 months between the repeat experiments. A similar experiment was done with
another set of BAG filters, but the percent bromate removal obtained in February 1999 was
significantly less than the percent bromate removal obtained in March 1998. A microbial disturbance,
probably caused by a two-week period in December 1998 with an EBCT greater than 200 minutes,
is likely responsible for the decline in bromate removal in these filters. This indicates that filter history
has a very important effect on biological bromate removal. The microbial community that develops
under a particular water quality condition might affect the bromate removal obtained even after the
water quality conditions become more favorable for bromate removal. A study of the change in the
microbial population resulting from such events will be very useful in understanding and optimizing
BAG filter performance.
The bromate removal results obtained in a one-inch inner diameter BAG filter were able to
be reproduced in a 2-inch inner diameter BAG filter, suggesting that wall-effects are minimal. This
indicates that scale-up of the bench-scale results should not be a concern.
Backwashing the BAG filters using filter effluent with a low DO concentration (2 mg/L) did
not adversely affect bromate reduction. Bromate removal before and after backwashing remained
essentially the same, despite the fact that significant microbial material was removed from the filters
during backwashing.
The mass concentration of bromate removed in the BAG filters was often dependent on the
influent bromate concentration. For lower bromate concentrations (10-20 ng/L) and longer EBCTs
(25-50 minutes), bromate removal increased as the influent bromate concentration increased. This
119

indicated that bromate removal was not limited by the electron donor concentration under these
conditions. However, constant mass concentration bromate removal was observed at higher influent
bromate concentrations (20-50 ng/L) and shorter EBCTs (10-20 minutes) which may indicate that
the electron donor concentration was limiting bromate removal under these conditions. Based on this
information, percent bromate removal would not necessarily remain constant as the influent bromate
concentration changed. The percent bromate removal obtained for a given water will depend on a
combination of factors including the influent bromate concentration, EBCT, and concentration of
electron donor.
Increasing the concentration of influent DO to the BAG filter caused a decrease in bromate
removal. Since it is possible that denitrifying organisms are using nitrate reductase to reduce
bromate, the presence of DO may be inhibiting the synthesis or activity of this enzyme. An increased
DO concentration may also cause the use of more electron donor for aerobic respiration, leaving less
electron donor available for bromate reduction. The lack of bromate reduction observed in full-scale
ozone-BAC plants is likely due to the high concentrations of DO entering the filters. Therefore, it
would be advantageous for treatment plants to reduce the DO concentration entering the BAG filters
as much as possible.
The negative effect of higher concentrations of DO on the bromate-reducing bacteria was not
immediately reversible when the DO concentration was reduced. A BAG filter operated with an
influent DO concentration of 2.0 mg/L, which had previously been exposed to an influent DO
concentration of 13.6 mg/L for two weeks, yielded significantly lower bromate removal efficiency
than expected. This result suggests that the DO concentration of the water used to backwash the
BAG filters should be carefully considered since it may affect subsequent bromate removal.
Increasing the influent nitrate concentration to the BAG filter caused a decrease in bromate
removal. If denitrifiers are reducing bromate in the BAG filter, nitrate and bromate may be competing
for use as terminal electron acceptors by nitrate reductase. Since it has been shown that a number
of non-denitrifying microbial isolates from the BAG can reduce bromate and nitrate, it also is possible
that nitrate and bromate are simply competing for the available electron donor.
Increasing the concentration of sulfate from 11.1 to 102.7 mg/L to the BAG filters only
caused a slight decline in bromate removal for the BAG filters. No significant sulfate reduction was
detected in the filters.
120

Bromate removal increased as the influent pH to the BAG filters decreased from 8.2 to the
near-neutral range of 6.8 to 7.2. It may be possible to reduce bromate formation during ozonation
and increase biological bromate removal through pH control since bromate formation decreases as
the pH decreases.
Under similar operating conditions, better bromate removal was observed in the BAG filters
using CUW (a groundwater source) than using LMW. The reduction of bromate in LMW may have
been hindered directly and/or indirectly by the DO. Only a small amount of DO was consumed in the
biological nitrification reaction because of the low ammonia concentration. This DO could then
compete with bromate for biodegradable organic matter to use in aerobic respiration. Alternatively,
if nitrate reductase catalyzes bromate reduction, the DO could also have inhibited the synthesis or
activity of the enzyme. It is possible that bromate reduction in LMW could have been improved if
the water had been ozonated, since bromate removal did improve when enough exogenous electron
donor was added to remove the DO and nitrate present in LMW. Thus, the biodegradability of the
source water NOM is a key aspect for efficient bromate reduction.
Perchlorate removal was observed in a BAG filter treating LMW spiked with electron donor.
Perchlorate was removed when the nitrate concentration in the filter was very low. No perchlorate
removal was observed in a BAG filter treating CUW because no external electron donor was added
to reduce the nitrate concentration.
Microscopic examination of the biomass present in the BAG filters showed that the
morphology mainly consists of rods and cocci. Some of the rods formed chains and a number of
aggregated rods were observed. Round-bottom protozoa were also seen.
DGGE showed that the microbial population present in the BAG was quite diverse. DGGE
can be run on the 16S rDNA of the bromate-reducing isolates, and these profiles can be compared
to the DGGE profile of the mixed biomass sample from the BAG filter. In this way, the presence of
individual bromate-reducing isolates in the BAG could be confirmed. Future work may use DGGE
profiling as a technique for observing significant perturbations in the population that may relate to
the bromate reduction performance of the BAG filter. DGGE could be run on BAG filter biomass
samples after changing water quality parameters or operational conditions in the BAG filters in order
to observe changes in the bromate-reducing community.
Eight bromate-reducing isolates were cultured. All cultures are facultative, since they were
able to grow on agar plates that were incubated anaerobically and those that were incubated
121

aerobically. All but one of the isolates can reduce nitrate. However, due to the lack of gas
production by the isolates, it is unlikely that any of these bromate-reducing organisms are denitrifiers.
Identification and further characterization of these bromate-reducing microorganisms may aid in
optimizing conditions for bromate reduction in the BAG filters.
Dilution-to-extinction experiments were also performed so that additional bromate-reducing
organisms may be isolated. Since eight bromate-reducing organisms have already been isolated by
traditional plating techniques, the organisms obtained in the dilution-to-extinction experiments will
be retained for future analyses.
Bromate reduction was sensitive to temperature. In batch experiments, inoculated with
biomass from the BAG filters, bromate was reduced at 25 °C but not at 4 °C. Therefore, as the water
temperature changes at the water treatment plant during the year, bromate removal will not likely
remain constant.
The most important modification to the ozone-BAC process for the application of bromate
removal is the reduction of the influent DO concentration to the filter. Little or no bromate reduction
has been observed at DO concentrations that are typical of the ozone-BAC process. Future work
should address the best way to reduce the influent DO concentration. Additionally, studies should
be conducted to ascertain the effects of a reduced DO concentration on effluent water quality. For
example, a reduced DO concentration to the BAG filter may affect the biodegradation of trace
compounds such as 2-methylisoborneol or geosmin.
In conjunction with a reduced DO concentration, other modifications (such as pH adjustment)
may be considered as a means to improve biological bromate removal. However, in order to truly
understand how best to optimize the reduction of bromate in a BAG filter, individual bromatereducing
cultures must continue to be studied. By understanding the physiology and nutrient
requirements of these organisms, the BAG process may be designed to select for these organisms.
PERCHLORATE
Abiotic and biological removal of perchlorate at low ng/L concentrations was investigated.
GAG filter experiments demonstrated that perchlorate is not chemically reduced on the surface of
GAG. Rather, ion exchange is the mechanism by which perchlorate is removed by GAG. An ion
exchange capacity of virgin Norit GAG for perchlorate was calculated. For the influent matrix used,
122

virgin Norit GAC showed an ion exchange capacity of 0.172 mg perchlorate per gram GAC. Batch
experiments using four different carbons, CUW and DDW, and two different pHs confirmed the filter
results that showed ion exchange not chemical reduction as the mechanism of perchlorate removal
by GAC.
Efficient abiotic removal of perchlorate was demonstrated by metal-catalyzed GAC filters.
Norit carbon mixed with aluminum shot and copper- and zinc oxide-impregnated carbon filters were
used for these experiments. Subsequent wash tests demonstrated that no perchlorate was chemically
reduced in the Cu- and ZnO-impregnated GAC filter. Though a small percentage of the perchlorate
removed by the aluminum/GAC filter was not accounted for, the dominant perchlorate removal
mechanism in both metal-catalyzed filters was ion exchange.
Experiments investigating the removal of perchlorate in ozone and ozone plus hydrogen
peroxide indicated that no perchlorate reduction occurred.
The GAC filters were rendered biologically active by extensively contacting them with effluent
from a set of filters that had demonstrated the ability to biologically reduce bromate. Once the GAC
filters were biologically active, the influent matrix for one B AC filter was switched to DDW to allow
for better control over influent conditions. Influent dissolved oxygen concentrations were kept low.
An electron donor solution was added to the influent tank when it was noted that no reduction of
perchlorate was occurring. Due to premature degradation of the electron donor solution, syringe
pumps were included in the treatment scheme to add electron donor at the point of entry to the BAG
bed. Biological reduction of perchlorate was immediately observed. The DDW filter demonstrated
efficient perchlorate removal and results from this filter revealed that perchlorate reduction is highly
sensitive to the concentration of nitrate in the filter. As nitrate concentrations increase, perchlorate
removal decreases.
In order to isolate the effect of nitrate concentration on perchlorate reduction in the BAG
filters, electron donor concentration was increased and EBCT was reduced. This ensured that
electron donor availability was not a constraint. Under these conditions, effluent nitrate
concentrations were constant for a given influent nitrate concentration. A threshold nitrate
concentration between 0.05 mg/L and 0.35 mg/L was established for the attainment of >95 percent
perchlorate removal in the BAG filter.
To investigate the effect of nitrate preloading on the removal of perchlorate, an experiment
was set up that compared perchlorate removal after the filter had been loaded with nitrate for several
123

months to perchlorate removal after the filter did not see nitrate for 2 !/2 weeks prior to the start of
the experiment. The data showed that when compared to the nitrate preload condition, perchlorate
removal for the non-nitrate preload condition decreased approximately 3 5 percent. It is not currently
known if this phenomenon is caused by the deprivation of nutritional nitrogen, a loss in energy
required for growth, or a suppression of perchlorate-reducing enzymes.
Perchlorate removal versus EBCT without influent nitrate was also investigated. EBCTs of
25, 15,10, and 5 minutes were utilized. During the course of this experiment, the BAG filter lost the
ability to reduce perchlorate. When a second 25 minute EBCT experiment was run two months after
the first 25 minute EBCT experiment, perchlorate removal decreased to less than 5 percent.
Perchlorate removals during the first 25 minute EBCT experiment were approximately 80 percent.
Again, a loss in microbial growth, a suppression of perchlorate-reducing enzymes, or the deprivation
of nutritional nitrogen may explain this phenomenon. An experiment is currently being run to recover
the perchlorate-reduction ability in the filter. The filter has been operating for 2 '/2 months with
nitrate in the influent. Desorption of perchlorate from the filter may have masked the degree to which
perchlorate was being biologically reduced during the early portions of the filter recovery experiment.
However, biological perchlorate reduction to below detection is now occurring in the filter.
Batch biological perchlorate experiments have also been performed. Microorganisms that
were isolated from the DDW BAG filter using agar plating techniques were inoculated into aqueous
media. One of the 12 isolates reduced perchlorate to below the detection limit-- isolate B9. Serum
bottle tests using this isolate were run to examine the kinetic effects of different initial nitrate
concentrations on perchlorate removal. Several interesting observations were made. First, isolate
B9 showed the ability to reduce both nitrate and perchlorate. Second, perchlorate removal did not
occur in any serum bottle until the nitrate concentration was ^ 0.1 mg/L. This falls in the range of
nitrate concentrations required for 2t95 percent perchlorate removal in the BAG filters, as determined
during the nitrate isolation experiments. Third, it was found that the greater the initial nitrate
concentration in the serum bottle, the greater the rate of perchlorate reduction. Again, the exact
casue of this has not yet been determined.
Dilution to extinction experiments showed that perchlorate-reducing microorganisms are
numerically dominant in the BAG filter. Bromate was reduced to below the detection limit at every
dilution. Nitrate was reduced from 1.0 mg/L to

Microscopic examination of the biomass present in the BAG filters revealed the most common
morphology to be that of single rod bacteria. These bacteria were found both as isolates and in floes.
Large protozoa were also found in the perchlorate filters. Some morphologies were common
between the bromate and the perchlorate filter samples while others were unique to each sample. A
DGGE analysis run with samples from the perchlorate filters confirmed the microscopic analysis: the
microbial population present in the BAG is diverse, shares some common species with the bromate
filters, and contains some unique microbial species.
METROPOLITAN WATER DISTRICT OF SOUTHERN CALIFORNIA: OZONATED
WATER (NO ADDED ELECTRON DONOR) FOR BROMATE AND PERCHLORATE
REDUCTION
Bench-scale BAG filters following ozonation demonstrated efficient biological removal of
both bromate and perchlorate for approximately 10,000 BV under various operating conditions.
Complete bromate and perchlorate removal was observed with 10-50 ug/L of influent bromate and
perchlorate, EBCTs of 25 and 15 minutes, and DO concentrations of 0-2 mg/L. When the influent
DO was increased to 5mg/L, perchlorate removal decreased by 50 percent while complete reduction
of bromate continued to be observed.
Ozonation by-products are believed to serve as the electron donors needed by the perchlorateand
bromate- reducing microorganisms in the BAG filter. Further study is recommended to isolate
and identify the specific by-products utilized by microorganisms as electron donors.
Given that the filter operates anaerobically, the choice of a sulfur-based reducing chemical
poses a taste-and-odor problem. Post-treatment may be required to meet aesthetic standards if the
current DO reducing strategy is adopted for future studies. However, the use of an alternative
reducing agent should be investigated to prevent any potential taste-and-odor problems.
Pilot-scale experiments were conducted at the Metropolitan's Mobile Pilot Plant facility,
which consists of ozone columns, conventional treatment, and BAG filters. The virgin carbon used
for the pilot-scale work showed saturation of the reduction capacity for bromate and the ion exchange
capacity for perchlorate at approximately 6,000 BV. When the system reached equilibrium (C/C0
si), ozonation and deaeration were started to provide favorable conditions for biological reduction
of bromate and perchlorate. 4,000 BV after switching the filter to this biological mode, complete
125

omate and perchlorate reduction was observed. These results suggest that ozonation followed by
deaeration and biological filtration could comprise an effective strategy for removing low |ig/L
concentrations of bromate and perchlorate from drinking water.
126

CHAPTER 6. RECOMMENDATIONS TO UTILITIES
1. To utilize BAG filtration for bromate and perchlorate reduction, a utility must
determine if reduction of the influent DO concentration to the filter is feasible. Not
only must it be economically feasible, but the utility should also determine that no
unwanted effects are caused by DO reduction (i.e. lack of taste-and-odor removal or
production of hydrogen sulfide in the filter). This could be accomplished through
bench-scale testing with the water and filter medium used by the utility.
2. Bench-scale testing at the utility should also be performed to determine the required
contact time in the BAG filter to obtain the desired removal of bromate and
perchlorate.
3. Depending on the alkalinity of the water, the utility should consider a combination of
reduction of bromate formation during ozonation and an increase in biological
bromate removal in the BAG filter by pH control.
4. Utilities that treat water contaminated with both nitrate and perchlorate should give
particular consideration to whether or not BAG filtration is a feasible treatment
option. Research has shown that BAG filtration removes nitrate and perchlorate very
efficiently. Additionally, nitrate reduction can enhance perchlorate reduction kinetics,
making BAG filtration particularly attractive for combined removal of nitrate and
perchlorate.
127

REFERENCES
Amy, G., M. Siddiqui, W. Zhai, J. DeBroux, and W. Odem. 1993. Nation-Wide Survey of Bromide
Ion Concentrations in Drinking Water Sources. In Proceedings of the American Water Works
Association. Denver, Colorado: AWWA.
APHA, AWWA, and WEF. (American Public Health Association, American Water Works
Association, and Water Environment Federation). 1995. Standard Methods for the Examination
of Water and Wastewater. 19* edition. Washington, D. C.: APHA.
Attaway, H. and M. Smith. 1993. Reduction of Perchlorate by an Anaerobic Enrichment Culture.
Journal of Industrial Microbiology. 12: 408-412.
Barber, M. J. and B. A. Norton. 1990. Spinach Nitrate Reductase - Effects of Ionic Strength and pH
on the Full and Partial Enzyme Activities. Plant Physiology, 93:537-540.
Bell L. C., D. J. Richardson, and S. J. Ferguson. 1990. Periplasmic and Membrane-Bound
Respiratory Nitrate Reductases in Thiosphaera pantotropha. FEBS Letters, 265(l-2):85-87.
Betts, K. S. 1998. Rotating Ion Exchange System Removes Perchlorate. Environmental Science
and Technology. 32(19):454A-455A.
Bishop, S. L., and M. Bourbigot. 1996. Meeting the Proposed Regulations Using Ozone and
Biologically Active Carbon. Journal of the New England Water Works Association, C V( 1): 1 -12.
Book, S. 1999. Health Effects of Perchlorate. Presented at the 1999 Perchlorate Conference,
Ontario, California, March 18-19,1999.
129

Brock, T. D., M. T. Madigan, J. M. Martinko, and J. Parker. 1994. Biology of Microorganisms, 7th
ed. Englewood Cliffs, New Jersey: Prentice Hall.
California Department of Health Services (CDHS) Division of Drinking Water and Environmental
Management Drinking Water Program. 1997. Perchlorate in Drinking Water Basis of DHS=
Provisional Action Level. In Proceedings of the American Water Works Association Research
Foundation Perchlorate Research Issue Group Conference. Ontario, California.
Camel, V. and A. Bermond. 1998. The Use of Ozone and Associated Oxidation Processes in
Drinking Water Treatment. Water Research. 32(11): 3208-3222
Chen, J. M. and O. J. Hao. 1996. Environmental Factors and Modeling in Microbial Chromium
(VI) Reduction. Water Environment Research, 68 (7): 1156-1164.
Clarke, J. U. 1998. Evaluation of Censored Data Methods to Allow Statistical Comparisons Among
Very Small Samples with Below Detection Limit Observations. Environmental Science and
Technology, 32 (1): 177-183.
Coates, J.D. U. Michaelidou, R.A. Bruce, S.M. O' Conner, J.N. Crespi, and L.A. Auchenbach.
1999. Ubiquity and Diversity of Dissimilatory (Per)chlorate-Reducing Bacteria. Applied and
Environmental Microbiology, 65(12): 5234-5241.
DeAngelo, A. B., M. H. George., S. R. Kilburn, T. M. Moore, and D. C. Wolf. 1998.
Carcinogenicity of Potassium Bromate Administered in the Drinking Water to Male B6C3Fj Mice
and F344/N Rats. Toxicologic Pathology, 26 (5): 587-594.
Glaze, W. H., M. Koga, and D. Cancilla. 1989. Ozonation By Products 2: Improvement of Aqueous-
Phase Derivatization Method for the Detection of Formaldehyde and Other Carbonyl Compounds
Formed by the Ozonation of Drinking Water. Environmental Science and Technology, 23: 838.
130

Forbes, C. 1997. Fax sent from Calif. Dept. of Health Services (Drinking Water Field Operations
Branch) to Robert Martin (General Manager of East Valley Water District, CA).
Frankenberger, William T., Jr., et al. 1999. Bacterial Reduction of Perchlorate and Nitrate in
Groundwater by Perclace. Presented at the Perchlorate Conference, Ontario, California, March 18-
19,1999.
H aag, W. R. and J. Hoigne. 1983. Ozonation of Bromide-Containing Waters: Kinetics of Formation
of Hypobromous Acid and Bromate. Environmental Science and Technology, 17: 261-267.
Herman, D. C., and W. T. Frankenberger. 1998. Microbial-Mediated Reduction of Perchlorate in
Groundwater. Journal of Environmental Quality 27:750-754.
Herman, D.C. and W.T. Frankenberger. 1999. Bacterial Reduction of Perchlorate and Nitrate in
Water. Journal of Environmental Quality. 28: 1018-1024.
Hijnen, W. A. M., R. long, and D. van der Kooij. 1999. Bromate Removal in a Denitrifying
Bioreactor used in Water Treatment. Water Research, 33 (4): 1049 - 1053.
Hijnen, W. A. M., R. Voogt, H. R. Veenendaal, H. van der Jagt, and D. van der Kooij. 1995. Bromate
Reduction by Denitrifying Bacteria. Applied and Environmental Microbiology, 61 (1): 239 - 244.
Hurley, J.M., W. Wallace, and E. Coppola. 1996. The Military Engineer. August-September, 56.
Jackson, C. R., E. E. Roden, and P. F. Churchill. 1998. Changes in Bacterial Species Composition
in Enrichment Cultures with Various Dilutions of Inoculum as Monitored by Denaturing Gradient
Gel Electrophoresis. Applied and Environmental Microbiology, 64 (12): 5046-5048.
Kirisits, M. J. and V. L. Snoeyink. 1999. Reduction of Bromate in a BAG Filter. Journal AWWA,
91 (8): 74-84.
131

Kirisits, M. J., V. L. Snoeyink, and J. C. Kruithof. 1998. Reduction of Bromate by Granular
Activated Carbon. In Proceedings of the 1998 National Conference on Environmental Engineering.
Chicago, Illinois.
Krasner, S. W., W. H. Glaze, H. S. Weinberg, and P. A. Daniel. 1993. Bromate Occurrence and
Control: Pilot- and Full-Scale Studies. In Proceedings of the American Water Works Association.
Denver, Colorado: AWWA.
Kuo, C. Y., et al. 1990 Analysis of Inorganic Disinfection By-Products in Ozonated Drinking Water
by Ion Chromatography. In Proceedings of the American Water Works Association. San Diego,
California: AWWA.
Kurokawa, Y., A. Maekawa, M. Takahashi, and Y. Hayashi. 1990. Toxicity and Carcinogenicity
of Potassium Bromate - A New Renal Carcinogen. Environmental Health Perspectives, 87: 309-
335.
Lloyd D., L. Boddy, and K. J. P. Davies. 1987. Persistence of Bacterial Denitrification Capacity
under Aerobic Conditions: The Rule Rather than the Exception. FEMS Microbiology Ecology, 45:
185-190.
Logan, Bruce E. et al. 1999. * Remediation of Contaminated Waters Using Innovative
Bioreactors" Presented at the Perchlorate Conference, Ontario, California, March 18-19.
Ludwig, W., G. Mittenhuber, and C. G. Friedrich. 1993. Transfer of Thiosphaera pantotropha to
Paracoccus denitrificans. Int. J. Syst. Bacteriol., 43: 363-367.
Meiberg, J. B. M., P. M. Bruinenberg, and W. Harder. 1980. Effect of Dissolved Oxygen Tension
on the Metabolism of Methylated Amines in Hyphomicrobium X in the Absence and Presence of
Nitrate: Evidence for x Aerobic' Denitrification. Journal of General Microbiology, 120: 453 - 463.
132

Miller, J. A. 1996. AThe Reduction of Bromate in Drinking Water by Activated Carbon. @ PhD
Thesis. University of Illinois at Urbana-Champaign.
Mills, A., A. Belghazi, and D. Rodman. 1996a. Bromate Removal from Drinking Water by
Semiconductor Photocatalysis. Water Research, 30 (9): 1973-1978.
Mills, A., A. Belghazi, D. Rodman, and P. Hitchins. 1996b. The Removal of Bromate from Potable
Water Using Granular Activated Carbon. Journal of the Institution of Water and Environmental
Management, 10 (3): 215-217.
Najm, L, R. Trussell, L. Boulos, B. Gallagher, R. Bowcock, C. Williams, and D. Clifford. 1999. In
Proceedings of the American Water Works Association. Chicago, Illinois: AWWA.
Pfaff, J. D., and C. A. Brockhoff. 1990. Determining Inorganic Disinfection By-Products by Ion
Chromatography. Journal AWWA, 82:4:192.
Renner, Rebecca. 1999. EPA Draft Almost Doubles Safe Dose of Perchlorate in Water.
Environmental Science and Technology. 33 (5): 110A-111 A.
Rikken, G.B., A.G.M. Kroon, and C.G. van Ginkel. 1996. Appl. Microb. Biotech., 45: 420.
Rittmann, B.E., R. Nurenberg, and I. Najm. 2000. "Autohydrogenotrophic Perchlorate
Reduction." Paper presented at the AWWA/AWWARF Inorganic Contaminants Workshop in
Albuquerque, NM, February 27-29.
Robertson, L.A. and J. G. Kuenen. 1990. Combined Heterotrophic Nitrification and Ae
in Thiosphaera pantotropha and Other Bacteria. Antonie van Leeuwenhoek, 57:139-152.
Robertson L. A. and J. G. Kuenen. 1984. Aerobic Denitrification: A Controversy Revived.
Archives of Microbiology, 139: 351-354.
133

Rogers, K. 1997. Water Contaminated. Las Vegas Review Journal. August 19: 1-3.
Sclimenti, M.J., S.W. Krasner, W.H. Glaze, and H.S. Weinberg, 1990. Ozone Disinfection By-
Products: Optimization of the PFBHA Derivatization Method for the Analysis of Aldehydes. In
Proceedings of the American Water Works Association, San Diego, California: AWWA.
Siddiqui, M., W. Zhai, G. Amy, and C. Mysore. 1996. Bromate Ion Removal by Activated Carbon.
Water Research, 30 (7): 1651-1660.
Siddiqui, M.J., G. Amy, K. Ozekin, W. Zhai, and P. Westerhoff. 1994. Alternative Strategies for
Removing Bromate. Journal AWWA, 86 (10): 81-96.
Simpkin T. J. and W. C. Boyle. 1988. The Lack of Repression by Oxygen of the Denitrifying
Enzymes in Activated Sludge. Water Research, 22 (2): 201-206.
Skoog, D. A. and J. J. Leary. 1992. Principles of Instrumental Analysis. 4*ed. Fort Worth, TX:
Harcourt Brace College Publishers.
Song R., C. Donohoe, R. Minear, P. Westerhoff, K. Ozekin, and G. Amy. 1996. Empirical
Modeling of Bromate Formation during Ozonation of Bromide-Containing Waters. Water Research,
30(5): 1161-1168.
SoucieW. 1999. November 9. Personal communication. Central Lake County Joint Action Water
Agency, Paul M. Neal Water Treatment Facility, 200 Rockland Road, Lake Bluff, Illinois 60044.
Standard Methods for the Examination of Water and Wastewater, 19th Edition, 1995.
Tiedje,J. M. 1988. Ecology of enitrification and Dissimilatory Nitrate Reduction to Ammonium.
In Biology of Anaerobic Microorganisms. Edited by A. J. B. Zehnder. New York: John Wiley and
Sons.
134

USEPA. 1998. Health Risk Assessment/Characterization of the Drinking Water Disinfection
Byproduct Bromate. Office of Science and Technology, Office of Water. March 13,1998. Quoted
in Fed. Reg. 63:61:15673, FR Document 98-8215.
Vandenabeele, J., D. de Beer, R. Germonpre, R. Van de Sande, and W. Verstraete. 1995. Influence
of Nitrate on Manganese Removing Microbial Consortia from Sand Filters. Water Research, 29 (2):
579-587.
von Gunten, U. and J. Hoigne. 1996. Ozonation of Bromide-Containing Waters: Bromate
Formation through Ozone and Hydroxyl Radicals. In Disinfection By-Products in Water Treatment
- The Chemistry of Their Formation and Control. Edited by R. A. Minear and G. L. Amy. Boca
Raton, Florida: Lewis Publishers.
von Gunten, U. and J. Hoigne. 1994. Bromate Formation During Ozonation of Bromide-Containing
Waters: Interaction of Ozone and Hydroxyl Radical Reactions. Environmental Science and
Technology, 28: 1234-1242.
Waterweek. 1997. Perchlorate Detected in Lake Mead Water. AWWA, Denver (Sept. 1, 1997).
Waterweek. 1998. EPA Finalizes M/DBP Rules. AWWA, Denver (December 4,1998).
Westerhoff, P., G. Amy, R. Minear, M. Siddiqui, R. Song, and K. Ozekin. 1993. Bromate Formation
in Laboratory Studies: Influence of Water Quality and Treatment Variables. In Proceedings of the
American Water Works Association. Denver, Colorado: AWWA.
9
Ye R. W., B. A. Averill, and J. M. Tiedje. 1994. Denitrification: Production and Consumption of
Nitric Oxide. Appl. Envir. Microbiol. 60 (4): 1053-1058.
135

ABBREVIATIONS
AWWA
AWWARF
American Water Works Association
American Water Works Research Foundation
BAG
biologically active carbon
CDHS California Department of Health Services
°C degrees Celsius
C
concentration at time, t
C/C0 fraction remaining
cm centimeter
C0
initial concentration
CUW Champaign-Urbana water
DDW
DFE
DO
DOC
distilled deionized water
Danville filter effluent
dissolved oxygen
dissolved organic carbon
EBCT
empty bed contact time
GAC
granular activated carbon
HPC
heterotrophic plate count
L
LOD
liter
limit of detection
M
molar
137

MCL maximum contaminant level
mg milligram
mg/L milligrams per liter
mL/min milliliters per minute
MWDSC Metropolitan Water District of Southern California
pg/L microgram per liter
NOM natural organic matter
PAC Project Advisory Committee
t time
TNTC too numerous to count
U.S. United States
USEPA United States Environmental Protection Agency
138

AWWA
Research
Foundation
Advancing the Sderm of Water*
6666 W. Quincy Avenue, Denver, CO 80235
(303)347-6100
1 P-3.25C-90836-3/01 -CM ISBN 1-58321-104-7