Accepted Manuscript

Long-term performance of biological ion exchange for the removal of natural organic
matter and ammonia from surface waters

Nargess Amini, Isabelle Papineau, Veronika Storck, Pierre R. Bérubé, Madjid

Mohseni, Benoit Barbeau

PII: S0043-1354(18)30600-6
DOI: 10.1016/j.watres.2018.07.057
Reference: WR 13956

To appear in: Water Research

Received Date: 6 June 2018

Revised Date: 20 July 2018
Accepted Date: 24 July 2018

Please cite this article as: Amini, N., Papineau, I., Storck, V., Bérubé, P.R., Mohseni, M., Barbeau,
B., Long-term performance of biological ion exchange for the removal of natural organic matter and
ammonia from surface waters, Water Research (2018), doi: 10.1016/j.watres.2018.07.057.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.
 ACCEPTED MANUSCRIPT

Stable IEX Capacity using Weekly Regeneration

PT
• High & Stable NOM Removal
Features
• No Nitrification
Weekly
IEX IEX IEX

RI
Regeneration Resin Resin Resin
(IEX) Removal
Mechanism • Anion Exchange

SC
IEX
Resin
Decrease of IEX Capacity to Complete Resin Exhaustion

U
• High to Moderate NOM Removal
Features
• Nitrification in Warm Waters

AN
(BIEX) IEX IEX
Without Resin Resin • Anion Exchange
IEX +
Regeneration + • Biodegradation
Resin
Biofilm Biofilm Removal

M
• Bioregeneration
Growth Mechanism • Probable Secondary Anion Exchange

D
TE
Stable Performance With Weekly Backwash
• Low NOM Removal
Features
• Nitrification
EP
BAC
BAC BAC BAC
Removal • Biodegradation
Mechanism
C
AC
 ACCEPTED MANUSCRIPT

1 Long-Term Performance of Biological Ion Exchange for the Removal of Natural Organic Matter
2 and Ammonia from Surface Waters
3 Nargess Amini, Isabelle Papineau, Veronika Storck, Pierre R. Bérubé, Madjid Mohseni and
4 Benoit Barbeau.
5 Nargess Amini : nargess.amini@polymtl.ca

PT
6 Corresponding author, M.Sc. student, Department of Civil, Geological & Mining Engineering,
7 Ecole Polytechnique de Montréal, 2900 boulevard Édouard-Montpetit, Montréal, Québec,

RI
8 Canada H3T 1J4
9 Isabelle Papineau : i.papineau@polymtl.ca

SC
10 Research Associate, Department of Civil, Geological & Mining Engineering, Ecole Polytechnique
11 de Montréal, 2900 boulevard Édouard-Montpetit, Montréal, Québec, Canada H3T 1J4
12 Veronika Storck : veronika.storck@polymtl.ca
13
U
Post-doctoral fellow, Department of Civil, Geological & Mining Engineering, Ecole Polytechnique
AN
14 de Montréal, 2900 boulevard Édouard-Montpetit, Montréal, Québec, Canada H3T 1J4
15 Pierre R. Bérubé : berube@mail.ubc.ca
M

16 Ph.D., P.Eng., Professor, Department of Civil Engineering, The University of British Columbia,
17 6250 Applied Science Lane, Vancouver, BC, V6T 1Z4
D

18 Madjid Mohseni : madjid.mohseni@ubc.ca

19 Ph.D., P.Eng., Professor, Department. of Chemical & Biological Engineering, University of British
TE

20 Columbia. Vancouver, BC, V6T 1Z4

21 Benoit Barbeau : benoit.barbeau@polymtl.ca
EP

22 Ph.D., P.Eng., Professor, Department of Civil, Geological & Mining Engineering, Ecole
23 Polytechnique de Montréal, 2900 boulevard Édouard-Montpetit, Montréal, Québec, Canada
24 H3T 1J4
C

25
AC

26 Abstract
27 Anionic exchange is an effective treatment option for the removal of natural organic matter from

28 surface waters. However, the management of the spent brine regenerant often limits the adoption of

29 this process. The current study reports one year of operation of ion exchange resins under biological

30 mode (BIEX, i.e. without regeneration to promote biofilm growth on the media) compared to the
 ACCEPTED MANUSCRIPT

31 performance of (i) ion exchange with weekly regeneration (IEX), (ii) granular activated carbon under

32 biological mode (BAC) and (ii) granular activated carbon under adsorption mode (GAC). Four parallel

33 pilot filters (GAC, BAC, IEX and BIEX) were fed with a colored and turbid river water without

34 pretreatment. Although IEX provided the best performance (80 % DOC removal) throughout the study,

PT
35 BIEX achieved a similar performance to IEX prior to DOC breakthrough (92 days) and subsequently

36 achieved a mean DOC removal of 62 % in warm water conditions. The GAC filter was rapidly exhausted

RI
37 (2 weeks) while the BAC filter only provided a 5 % DOC reduction. Full nitrification was observed on both

SC
38 the BIEX and BAC filters under warm water conditions (> 15oC). After one year of operation, BIEX was

39 successfully regenerated with brine. According to a mass balance, 69% of DOC removal in BIEX was due

40

U
to ion exchange while we assume the remainder was biodegraded. Operation of ion exchange in
AN
41 biological mode is a promising option to reduce spent brine production while still achieving high DOC

42 removal.
M

43 Keywords: Ion Exchange, Activated Carbon, Biological Mode, Natural Organic Matter, Ammonia,
D

44 Drinking Water
TE

45 Highlights
46 • In the biological IEX filter, DOC breakthrough occurred after 60 days.
EP

47 • After DOC breakthrough, BIEX reduced DOC from 7 mg C/L to 2-3 mg C/L in warm water.

•
C

48 Nitrification in warm water was as efficient in BIEX filters as in BAC filters

•
AC

49 BIEX media was successfully regenerated after 331 days of operation.

50 • NOM removal in BIEX was mostly (69%) due to ion exchange.

51 1. Introduction
52 Natural Organic Matter (NOM) is ubiquitous in surface waters (Matilainen and Sillanpaa 2010, Pelekani

53 and Snoeyink 1999). Adequate NOM removal during drinking water treatment is of importance as the
 ACCEPTED MANUSCRIPT

54 presence of NOM deteriorates water quality and disrupts several water treatment processes. These

55 deleterious NOM impacts are numerous: poor aesthetic water quality such as taste and odours

56 (Christman and Ghassemi 1966), formation of chlorinated disinfection by-products (DBPs) (Kleiser and

57 Frimmel 2000, Xie 2003), potential bacterial regrowth and biofilm formation in distribution systems (van

PT
58 der Kooij 1992), reduction of micropollutants sorption on activated carbon (Smith and Weber 1985),

59 membrane fouling (Amy and Cho 1999, Nilson and DiGiano 1996), impact on UV and UV-based advanced

RI
60 oxidation processes (Sarathy et al. 2011), decrease in the rate of oxidation of iron and manganese

SC
61 (Graveland and Heertjes 1975), etc. Given this long list of negative impacts, it is not surprising that NOM

62 removal has received much attention in the scientific literature.

63
U
Available water treatment processes for NOM removal are numerous and include coagulation, high
AN
64 pressure membranes (e.g. NF), sorption-based processes (ion exchange and activated carbon) and

65 biological treatment. For small water systems in remote areas, selecting an appropriate process is
M

66 challenging as the issues of cost and complexity of operation are important design constraints. Passive
D

67 and robust systems with low production of residuals are desirable. Biological filtration with activated
TE

68 carbon (BAC) has been considered as an economical and passive option for the removal of dissolved

69 organic carbon (DOC) and DBP precursors. However, DOC removal under steady-state BAC is typically in
EP

70 the low range of 5-20 % (Terry and Summers 2017). On the other hand, ion exchange (IEX) is a promising

71 robust and simple treatment alternative to remove color, DBP precursors and chlorine demand as
C

72 typically 85-95 % of NOM is negatively charged (Boyer et al. 2008) and, therefore, potentially removable
AC

73 by IEX. However, spent brine management is an important drawback of this option as severe discharge

74 limits for sodium and/or chloride exist under many environmental regulations in order to protect

75 ecosystems. In addition, regenerant (NaCl) transport to the water treatment facility can be an important

76 constraint for remote communities such as the ones found in Canada. The development of a robust

77 NOM removal process with low chemical usage, low effluent waste discharge and high performance
 ACCEPTED MANUSCRIPT

78 would clearly represent an important breakthrough for the design of small water systems in remote

79 communities.

80 As part of RES’EAU-WATERNET, a strategic network dedicated to small water systems, we recently

PT
81 reported the possibility to operate IEX contactors in an extended operation cycle without regeneration

82 (months rather than days), a concept referred to hereafter as biological ion exchange (BIEX) (Schulz et

RI
83 al. 2017, Winter et al. 2018). The lab-scale study of Schulz et al. (2017) compared three abiotic vs. three

84 biotic columns (≈ 2 BV/h, EBCT = 30 min) of Purolite A860 resin. The columns were fed for 2 months

SC
85 (2800 BV) with 0.45 µm pre-filtered surface waters with neutral pH and constant DOC (≈ 5 mg/L) at

86 room temperature (22oC). Three different in-situ regeneration strategies were tested on each biotic and

87
U
abiotic columns (brine, caustic plus brine or peracetic acid prior to regeneration with caustic and brine).
AN
88 All regeneration strategies were able to fully recover IEX capacity. The presence of biofilm did not

89 impact regeneration efficacy. Biotic columns could remove approximately 60% DOC in contrast to
M

90 abiotic columns which removed approximately 40% DOC. As a follow-up, Winter et al. (2018) compared
D

91 the performance of a 3-month acclimatized BIEX with a 5-yr BAC filter using identical test conditions
TE

92 (temperature, EBCT & source waters) to the study of Schulz et al. (2017). Over 11 months of operation

93 (16,000 BV), approximately 56±7% DOC removal was maintained in the BIEX while BAC filtration
EP

94 operating in parallel only provided a 15±5% DOC removal.

95 Considering the overall cost of salt and brine disposal associated to IEX, it is of interest to increase our
C

96 understanding of conditions that favor BIEX performance to reduce salt consumption while allowing
AC

97 sufficient NOM removal to meet DBP regulation. Thus, the objective of this study was to investigate the

98 impact of long-term operation of IEX resins without regeneration in order to promote biological activity

99 on the media. For this purpose, we operated pilot columns for a period of 60 weeks. The pilot was

100 directly fed by the Des Prairies River, a source water with high DOC (≈ 7 mg C/L), low alkalinity (≈ 30 mg

101 CaCO3/L) and variable turbidity (5-58 NTU). This long-term operation was performed to confirm the
 ACCEPTED MANUSCRIPT

102 potential viability of BIEX operation mode and its potential superiority against BAC or GAC filtration,

103 even over a long period of operation.

104 2. Materials
Materials and methods

PT
105 2.1. Source water characteristics
106 The pilot plant was located at the Pont-Viau water treatment plant (Laval, Canada) which is fed by the

RI
107 Des Prairies River, a colored and low mineralized surface water (Table 1) currently treated at full-scale

108 with a ballasted flocculation, inter-ozonation, biological activated carbon filtration and post-

SC
109 chlorination. During the current study (February 2017 to April 2018), the source water exhibited a

110 significant turbidity (14 NTU), high dissolved organic carbon (DOC) concentration (7.1 mg C/L) and the

111 temperature fluctuated from 1.4 to 23.6oC.

U
AN
112 2.2. Experimental set-
set-up and operating conditions
M

113 The pilot plant (Figure 1) consisted of four parallel filtration columns (PVC, 10 cm diameter and 2 m

114 height) containing 1 m (= 8.1 L) of either (i) ion exchange resins (IEX), (ii) biologically active ion exchange
D

115 resins (BIEX), (iii) granular activated carbon (GAC) or (iv) biological activated carbon (BAC). The columns
TE

116 were equipped with several sampling taps to allow sample collection at various empty bed contact times

117 (EBCT). The resin type used for IEX and BIEX columns consists of Purolite A860, an anionic strong base
EP

118 resin. The only difference between the IEX and BIEX filter was the frequency of regeneration. Both

119 media were new at the onset of the project. A weekly regeneration of the IEX filter was done by filtering
C

120 2 bed volumes (BV) of 120 g NaCl/L whereas the BIEX filter was never regenerated (except for a single
AC

121 regeneration assay performed after about one year of operation). The coal-based activated carbon

122 (AquaCarb® 816, Evoqua, USA) was either fresh (GAC) or exhausted (BAC). The BAC media was collected

123 from the full-scale filter of the Pont-Viau water treatment plant (WTP) after two years of operation and

124 was therefore considered exhausted.

 ACCEPTED MANUSCRIPT

125 The columns were continuously operated at a filtration rate of 2 m/h (2 BV/h or 270 mL/min) which

126 corresponds to an empty bed contact time (EBCT) of 30 min. This filtration rate is lower than commonly

127 used for IEX (5-15 m/h) contactors or BAC filters (10 m/h). This lower velocity was motivated by the fact

128 that (i) the columns were fed by turbid source waters and (ii) the goal was to develop a process which

PT
129 minimizes maintenance. All four filtration columns were backwashed weekly, first using air (2 min at

130 55 m/h), then water to achieve a media expansion of 50 %. Backwash was continued until the backwash

RI
131 effluent turbidity was < 10 NTU or until 40 L (4 BV) of backwash effluent was collected. The IEX filter was

SC
132 regenerated after performing the backwash.

133 2.3. Monitoring

Monitoring filter performance

134
U
Sampling procedures: Influent and effluent streams were monitored weekly for DOC, turbidity,
AN
135 temperature, chloride (Cl-), trihalomethane (THM), haloacetic acid (HAA) precursors, and ammonia

136 (NH3). IEX and BIEX anion exchange capacities (AEC) were monitored weekly by sampling filter media at
M

137 depths of 5, 15, 50 and 90 cm. Liquid and solid media samples were collected at various depths after 7,
D

138 19 and 35 weeks of pilot plant operation (in April, July and November 2017) to assess DOC, NH3, nitrite

(NO2-), nitrate (NO3-), adenosine triphosphate (ATP), biological nitrifying activity, chloride (Cl-) and
TE

139

140 sulfate (SO42-). In addition, pilot plant performance was compared to the full-scale WTP performance
EP

141 after 19 weeks (July 2017) and 35 weeks of operation (November 2017).

142 Liquid sample characterization: DOC was quantified according to Standard Method 5310C using a
C

143 UV/persulfate TOC-meter (Sievers 5310C, GE Water, USA) after filtration through 0.45 µm pore-size
AC

144 filters (Supor® 450 PES, PALL). Turbidity was analyzed using Standard Method 2130B (Standard Methods

145 2012) using a Hach 2100 turbidy meter. NO2-, NO3-, Cl- and SO42- were quantified in filtered samples (0.45

146 µm) by an ion chromatograph (ICS 5000 AS-DP DIONEX) equipped with an AS18 column according to

147 method MA 300 Ions 1.3(CEAEQ 2014). Bicarbonate was estimated from titration-based alkalinity

148 measurements. THM and HAA precursors were quantified under Uniform Formation Conditions (UFC)
 ACCEPTED MANUSCRIPT

149 (Summers et al. 1996), i.e. by maintaining a free chlorine residual of (1.0 ± 0.5) mg Cl2/L after a contact

150 time of 24 hours at pH 8.0 and 20°C. THM and HAA were analyzed by gas chromatography (7890B GC

151 system from Agilent Technologies) according to methods 524.2 (THM) and 552.3 (HAA) (USEPA 2003).

152 NH3 was quantified with the indophenol colorimetric method NF T90-015 (AFNOR 2000).

PT
153 Media sample characterization: Media samples were characterized for nitrifying biomass, total biomass

RI
154 and AEC. Nitrifying bacterial activity was analyzed on filter media according to the method of Kihn et al.

155 (2000). Briefly, media samples (2 cm3) were collected at various depths with pre-cut sterile plastic

SC
156 syringes. Media samples were washed and then resuspended in a nitrifier medium spiked with 10 mg

157 N/L of NH4Cl before incubation at 30oC for 30 minutes while maintaining a constant organic-free air

158
U
sparging. After incubation, formation of NO2- and NO3- concentrations were measured by colorimetry
AN
159 (Jones 1984). Potential nitrifying activity is reported as µg N nitrified/cm3/h.
M

160 Total biomass was estimated using ATP measurements performed on detached biofilm from the solid

161 media using 6 cycles of sonication at 20 W on 5 g of media, resuspended after each cycle in 50 mL of
D

162 sterile phosphate buffer. After each sonication cycle, the supernatant (≈ 50 mL) was recovered and
TE

163 mixed to produce a composite sample which was filtered (mesh size 0.2 micron, Quench-Gone Syringe

164 Filters, (DIS-SFQG-25), LuminUltra, USA) to retain the bacterial and exclude any extracellular ATP.
EP

165 UltraLyse (LuminUltra, USA) was then filtered to lyse the bacterial ATP retained onto the filter and

166 recover intracellular ATP of detached biomass. After luminase injection in the filtrate, luminescence was
C

167 read on a TriStar2 Multimode Reader LB 942 (BERTHOLD Technologies). Total biomass is reported as ng
AC

168 ATP/cm3 of media.

169 Anion exchange capacity (AEC) of IEX and BIEX resins was monitored by titration. IEX beads (10 mL) were

170 added into 170 mL NaNO3 (25.6 g/L) which was agitated at 190 rpm for another 30 min to displace Cl- by

171 NO3-. The beads were then removed and an aliquot of the filtrate (15-30 mL) was spiked with 1 mL of
 ACCEPTED MANUSCRIPT

172 K2CrO4 (20 g/L) and then titrated with AgNO3 (0.04 N) until the solution changed to an orange color due

173 to AgCl precipitation. AEC is expressed as mEq/mL of resin.

174 2.4. Statistical analysis

PT
175 Significance tests were performed using analyses of variance (ANOVA) with the usual significance level

176 set at p = 0.05. Statistical analyses were conducted using Statistica 13.0 (TIBCO Software, USA).

RI
177 3. Results

SC
178 3.1. Evolution of source water temperat
temperature

179 The pilot was started on February 28th, 2017 and was operated for a period of 420 days. The water

U
180 temperature was less than 2oC until April and progressively increased up to 20oC in June. The water
AN
181 temperature was above 20oC from June to September, before slowly declining to less than 4oC in

182 November. It remained at 4°C for the rest of the project which ended in April 2018. The BIEX column
M

183 was regenerated in January 31st, 2018 (after 331 days of operation). Following the regeneration of BIEX,

184 pilot performance was monitored for an additional 90 days.

D
TE

185 3.2. Natural organic matter removal

186 Source water and effluent DOC concentrations as well as UV absorbance of the four columns operated
EP

187 in parallel were assessed through time (Figure 2 and Figure 3), while the distribution of DOC

188 measurements prior to BIEX regeneration (331 days) was used to summarize the overall performance of
C

189 each media type (Figure 4). The source water DOC was fairly stable with an average of 7.1 mg C/L during
AC

190 the study. DOC removal of the BAC filter was marginal (≈ 0.48 mg C/L or 7 %), while the GAC only

191 provided significant removals during the first two weeks of operation. After approximately 200 days of

192 operation, the GAC was fully exhausted and provided the same performance as the BAC (p > 0.05). The

193 IEX column offered the highest performance as the effluent DOC in was sustained in the low range of 1-2

194 mg C/L (i.e. an average 80 % DOC removal). The BIEX gave an equivalent performance to IEX for the first
 ACCEPTED MANUSCRIPT

195 50 days. After this period, the BIEX effluent DOC concentration progressively rose to a maximum of 4.0

196 mg C/L after 90 days. Interestingly, as the water temperature rose above 15oC, the BIEX effluent DOC

197 concentration progressively decreased, an indication that biodegradation most likely became a

198 significant DOC removal mechanism. As water temperature declined below 15oC in fall (after 250 days),

PT
199 the BIEX effluent DOC concentration started to rise again suggesting that a large part of DOC removal

200 during summer may have been due to biological activity. After 331 days of operation (January 23rd,

RI
201 2018), the BIEX was regenerated and its performance proved to be equivalent to the performance of the

SC
202 IEX column (i.e. 1.8-2.0 mg C/L) for the 30 days following BIEX regeneration. Throughout the study, DOC

203 profiles through the filter bed performed after 7, 19 and 35 weeks of operation (Figure S1) indicated

204

U
that most of the DOC removal by BIEX and IEX was achieved in less than 10-15 min of empty bed contact
AN
205 time. Therefore, we suggest that an operation with an EBCT of 15 min should be considered for a BIEX

206 filter design in order to reduce capital costs.

M

207 The impact of temperature on DOC removal was assessed by calculating the activation energies (Ea) for
D

208 the BAC, the BIEX and the IEX columns. Activation energies can be calculated by plotting a linear
TE

209 regression between ln k vs. 1/T where T is the water temperature (in Kelvins) and k is the apparent

210 kinetic constant calculated for each sampling campaign (k was approximated as ∆DOC/EBCT). For the
EP

211 BIEX, only removal data after 100 days of operation were used in order to retain performance under the

212 suspected biological mode. The results (Error! Reference source not found.) indicate that the activation
C

213 energies are respectively (20 ± 5), (30 ± 4) and (30 ± 8) kJ/mole for the IEX, BIEX and BAC filters. The BAC
AC

214 data should be interpreted with caution due to the very low DOC removals measured for this media

215 (which explains the poorer fit). The BIEX filter was more sensitive to variations in water temperature as

216 opposed to the IEX filter. The activation energies of BAC filters have been reported as 54 kJ/mole in the

217 CHABROL model (calculated using data from Laurent et al. (1999)) or 45 (ozonated waters) and 55

218 kJ/mole (non-ozonated waters) (calculated using data from Terry and Summers 2017).
 ACCEPTED MANUSCRIPT

219 3.3. Exhaustion of ion exchange capacity

220 In order to better distinguish the mechanisms responsible for NOM removal, the chloride release in the

221 BIEX and IEX effluents were assessed along with the DOC removal (Error! Reference source not found.).

222 The chloride release from IEX varied from 21 to 34 mg/L from week to week (Error! Reference source

PT
223 not found.a). The chloride release from the IEX filter was constant and not related to the DOC removal

RI
224 performance (DOC/DOC0) which was high (DOC/DOC0 < 0.25) throughout the year. In contrast, chloride

225 release from BIEX progressively declined from > 15 mg/L to zero after 90 days (Error! Reference source

SC
226 not found.b) which coincided exactly with the DOC breakthrough observed in the BIEX effluent. This

227 result suggests that the primary IEX capacity (i.e. due to chloride) was exhausted at that time, which

228
U
corresponds to 4320 bed volumes (BV). However, it is possible that NOM displaced other anions on the
AN
229 media after this period (e.g. sulfate or bicarbonate). Finally, BIEX regeneration at 331 days of operation

230 resumed chloride release showing a value similar to IEX at the same period.
M

231 Ion exchange capacity within BIEX and IEX was also quantified by recovering media from different
D

232 column depths and measuring the residual IEX capacity (Error! Reference source not found. and Figure
TE

233 S3 for profiles). The IEX capacity for fresh resin was measured in the lab as equal to 0.68 mEq/ml of

234 resin. Results for the IEX filter (Figure 7a) indicates that the weekly regenerations were efficient as the
EP

235 IEX capacities in the middle (50 cm) and bottom (90 cm) of the filter were stable and very high. In

236 contrast, the IEX capacity inside the BIEX declined below 0.1 mEq/mL of resin after 10 weeks of
C

237 operation. The regeneration after 48 weeks increased the IEX capacity to values similar to that of the IEX
AC

238 at the same period.

239 To understand the mechanism of DOC removal, a carbon mass balance was calculated for BIEX and IEX

240 using the influent/effluent DOC concentrations as well as the DOC measured in the brine recovered after

241 BIEX and IEX regenerations (Table 2). Out of the 543 g of carbon removed by BIEX in 331 days, 68.5%
 ACCEPTED MANUSCRIPT

242 was due to ion exchange and the remainder was probably due to biodegradation. As expected, DOC

243 removal in the IEX filter is essentially due to ion exchange (99.4%).

244 3.4. Biomass measurement on colonized media

PT
245 Biomass measurements were performed on the BAC, BIEX and IEX filters after 7 and 19 weeks of

246 operation (April and July 2017). Figures 8a and 8b present the ATP profiles through the depth of the

RI
247 columns for these two sampling campaigns. In general, ATP amounts were higher in the BAC filter,

248 except for the sample at the top of the BIEX filter recovered in April. The biomass density was not a

SC
249 significant predictor of BIEX performance compared to the BAC filter. The average biomasses (based on

250 the profiles) were calculated as (7.0 ± 6.8), (18 ± 18) and (27 ± 11) ng/cm3 of IEX, BIEX and BAC media,

251
U
respectively. However, we suspect that ATP measurements were not providing an accurate evaluation of
AN
252 the biomass. We observed that a schmutzdecke was developing in the upper portion of the BIEX filter as

253 opposed to the other filters. Breaking down this layer with air injection during the first step of the
M

254 backwash (BW) was important to properly clean the media during backwash. It was also observed that
D

255 the BIEX filter required a longer ripening time for turbidity compared to IEX and BAC when the filters
TE

256 were put back in service (see Figure 7c). This excess turbidity most likely results from the biomass

257 sloughing during backwash, as such behavior (i.e. long ripening after BW) was not observed on the IEX
EP

258 filter (which was not observed to develop a schmutzdecke). Therefore, ATP measurements may have

259 been inadequate to characterize the biofilm density found in the upper layer of the BIEX filter due to the
C

260 difficulty to correctly sample the schmutzdecke. The average effluent turbidities during the study were
AC

261 4.0, 5.5, 5.8 and 5.9 NTU for the BIEX, BAC, GAC and IEX filters, respectively. The BIEX filter was

262 therefore providing slightly higher retention of particulate matter, although these filters are not

263 designed for this specific water treatment objective. The BIEX filter also exhibited higher headloss (data

264 not shown: ≈12% higher than IEX) than the other columns.
 ACCEPTED MANUSCRIPT

265 3.5. Nitrification

266 Microbial activity was assessed by monitoring ammonia removal by the columns. Ammonium, a cation,

267 is not expected to be efficiently removed by anion exchange resins. However, nitrate, an anion, is

268 expected to be efficiently removed. For BIEX, it is expected that nitrifying microorganisms will convert

PT
269 ammonia to nitrate which may or may not be removed depending on the degree of exhaustion of the

RI
270 BIEX media. Figures 9a and 9b present the performance of the four media types to remove ammonia

271 during the first 140 days of operation. One last sampling campaign was performed after 340 days at cold

SC
272 temperatures (< 4oC). During the first 50 days of operation, the BAC was the only filter to reduce

273 ammonia below 10 µg N/L as the media was already biologically active at the onset of the study

274

U
(sampled from the WTP after two years of use). The GAC and BIEX eventually achieved performance
AN
275 equivalent to the BAC filter following an acclimation period of 50 and 78 days, respectively. Throughout

276 the study, the IEX column offered the lowest ammonia removal, an observation suggesting that the
M

277 weekly brine regeneration was negatively impacting the population of nitrifiers. Nitrate and nitrite
D

278 profiles were measured across the media after 7 and 35 weeks of operation (Figures 9c and 9d). After 7

279 weeks of operation, ion exchange was still prevalent within the BIEX as can be noted by the increase of
TE

280 nitrate in the upper portion of the column. This phenomenon was also observed within the IEX column
EP

281 which evidenced that anion displacement was occurring in these media (Figure 9c). After 35 weeks of

282 operation, ammonia removal was low in the IEX filter while it was close to 100 % in the BAC and BIEX
C

283 filters. For these two filters, the ammonia was converted into nitrate (Figure 9d), which was not
AC

284 removed by neither BAC nor BIEX, whose IEX capacity had vanished. Only IEX efficiently removed nitrate

285 by ion exchange. At that time, potential nitrifying activity was only detected in the upper media layer (5

286 cm) of the BAC (1.82 µg N/cm3/h) and the BIEX filter (4.51 µg N/cm3/h).
 ACCEPTED MANUSCRIPT

287 3.6. Removal of THM and HAA precursors

288 The removal of THM and HAA5 precursors was monitored for the first 120 days of operation (Figure 10).

289 As expected, the removals were consistent with the effluent DOC concentrations: the IEX filter provided

290 the lowest average THM-UFC (45 µg/L) and HAA5-UFC (41 µg/L) concentrations. The THM and HAA

PT
291 precursors in the BIEX effluent reached a peak at 92 days of operation, which corresponded to the DOC

RI
292 breakthrough. After this event, the concentrations of THM-UFC and HAA5-UFC declined during the

293 summer to averages of 88 and 51 µg/L, respectively.

SC
294 After 19 weeks of operation (July 2017), a sampling campaign was performed to compare the removal of

U
295 THM precursors in the IEX, BIEX and BAC effluents as opposed to the unit treatment processes in place
AN
296 at the full-scale WTP (Figure S2). Interestingly, the IEX column provided a lower concentration of THM

297 precursors (29 µg/L) than the full-scale plant (35 µg/L) which included ballasted flocculation, inter-
M

298 ozonation and dual media sand/BAC filtration. In contrast, the performance of the BIEX (72 µg/L) was

299 similar to the effluent from the clarifier (77 µg/L). This was an impressive performance considering that
D

300 the BIEX column had been in operation for 135 days (6480 BV) without regeneration.
TE

301 4. Disc
Discussion
EP

302 BAC filtration can typically remove 5-20 % of NOM depending on the characteristics of the source water,

303 EBCT and water temperature (Terry and Summers, 2017). The performance of our BAC filter fell within
C

304 this range. On the contrary, the BIEX filter provided a largely superior performance. After ion exchange
AC

305 exhaustion of BIEX (based on chloride release), DOC removals as high as 64 % were observed during

306 summer, a performance comparable with the lab-scale pilots that we reported in 2017-2018, fed with a

307 different surface water (5 mg DOC/L) and a 30 min EBCT BIEX filter (Schulz et al. 2017, Winter et al.

308 2018). The mechanisms responsible for the superior performance of BIEX compared to BAC have not yet

309 been fully elucidated. Theoretical IEX breakthrough was estimated to occur after59 days based on (i) the
 ACCEPTED MANUSCRIPT

310 average sulfate (0.19 mEq) and DOC concentrations (5 mg C/L removed = 0.05 mEq assuming a charge

311 density of 10 mEq/g C at pH 7), (ii) the fresh IEX capacity in the column (5.4 Eq), (iii) the condition of

312 operation (2 BV/h), and (iv) neglecting other anions (bicarbonate & nitrate). Differences in performance

313 of the BIEX and IEX were first observed after 51 days of operation which is consistent with the estimated

PT
314 capacity of the bed. Therefore, we conclude that IEX was not the sole mechanism at play to explain the

315 long-term performance of the BIEX column.

RI
316 In the studied source water, humic substances compose approximately 75 % of the DOC. To achieve high

SC
317 DOC removal, this NOM fraction, reputed to be very difficult to biodegrade (Catalán et al. 2017), must

318 therefore be partially removed. Using LC-OCD analysis of various surface waters, it has been shown by

319
U
Catalan et al. (2017) that 20-50 % of aquatic NOM can biodegraded. Our organic carbon mass balance
AN
320 indicates that 31 % of the DOC was probably biodegraded. As anion exchange resins have a superior
M

321 capacity to sorb NOM compared to GAC, it is speculated that IEX resins offer a more favorable

322 environment for microbial growth than activated carbon. The higher NOM loading present on the
D

323 media, the high macroporosity of resins and the weak electrostatic bonding of NOM to its surface may
TE

324 favour biodegradation activity, which, in turn, would liberate new sorption sites for additional reaction,

325 as suggested in the concept of bioregeneration (El Gamal et al. 2018). Under such scenario, the BIEX
EP

326 filter would be best described as a process with simultaneous ion exchange and bioregeneration. The

327 organic carbon mass balance indicates that most of the DOC in the BIEX filter (69%) is removed by ion
C

328 exchange while this value is over 99% for the IEX filter. The activation energy that we calculated for BIEX
AC

329 was also more consistent to an IEX filter than a BAC filter.

330 Important NOM removal on BIEX was observed after DOC breakthrough even though chloride release

331 from the media was negligible. Apart from biodegradation, secondary ion exchange mechanisms (e.g.

332 displacement of sulfate by NOM) may also be responsible for the long-term performance of BIEX.
 ACCEPTED MANUSCRIPT

333 Additional investigations will be needed to assess the role of secondary ion exchange on NOM removal.

334 It is anticipated that such phenomenon would be highly source-water specific (i.e. mineralization).

335 Biomass development on anionic resins is currently considered as a nuisance by resins suppliers. Many

PT
336 suppliers will recommend not to extend regeneration intervals beyond 48-72 h in order to control

337 biomass growth. During this project, we were interested in promoting microbial growth. Heterotrophic

RI
338 and nitrifying biomasses were measured on the BIEX media at amounts that were not largely different

339 from the BAC filter. The measured biomass densities on both media fall in the low range of reported

SC
340 values in the literature for BAC filters (Gibert et al. 2013, Velten et al. 2011, Velten et al. 2007, Zhang et

341 al. 2017). However, the observed formation of a dense and difficult-to-break-up biological layer

342
U
(schmutzdecke) in the upper portion of the BIEX is clearly an issue that will require more attention, given
AN
343 that the low density of anion exchange resins makes the backwashing process more challenging than for
M

344 other denser granular media.

345 Biomass is thought to potentially reduce regeneration efficacy. However, in an earlier study, a BIEX filter
D

346 operated for 60 days (2600 BV) was shown to be effectively regenerated with any of the three different
TE

347 tested strategies of regeneration (brine, caustic plus brine or peracetic acid prior to regeneration with

348 caustic and brine) (Winter et al. 2018). During the present study, the BIEX filter also recovered its ion
EP

349 exchange performance after a regular regeneration following one year (47 weeks) of sustained
C

350 operation (≈ 15,840 BV). Nevertheless, the resins morphology (color and size) were impacted by this
AC

351 extreme scenario of operation (pictures are provided in Figure S4) which also failed to provide the

352 desired effluent water quality (DOC ≤ 2 mg/L) during the entire period. Therefore, the recommended

353 strategy would be to perform the regeneration of the (BIEX) column when breakthrough of DOC is noted

354 in the effluent as opposed to the common strategy based on a very low fixed number of bed volumes. In

355 addition, systems in Nordic climates like Canada would benefit from a regeneration before the cold-
 ACCEPTED MANUSCRIPT

356 water season due to the adverse impact of lower temperature on biodegradation and, to a lower extent,

357 ion exchange. These recommendations would lead to an important reduction in salt usage which will

358 make IEX a more sustainable treatment alternative.

PT
359 5. Conclusion
360 Four parallel pilot filtration columns were fed with surface water with a high DOC and low mineralization

RI
361 for a one-year period. The IEX column, regenerated weekly, presented the best performance for DOC

362 removal but the lowest for ammonia removal. The BIEX filter provided a largely superior performance to

SC
363 the BAC filter with respect to NOM removal but similar nitrification capacities. The following BIEX

U
364 performance was noted:
AN
365 • Effluent DOC concentrations of BIEX were below the treatment goal of ≤ 2 mg C/L for 64 days (≈

366 3072 BV), rose to 4.0 mg C/L after breakthrough, stabilized at 2.5 mg C/L in warm water conditions
M

367 (3072-6768 BV) and finally rose again up to 5.4 mg C/L under winter conditions (< 4oC). After 141

368 days of operation (6768 BV), the BIEX was still able to lower TOC from 7.1 to 2.5 mg C/L (October
D

369 2017), although its residual anion exchange capacity was below detection.
TE

370 • The performance for DOC removal of BIEX in warm waters (62 %) was similar to the one observed

371 at lab scale (56 %) in a previous study using a different surface water (Winter et al. 2018). In both
EP

372 cases, BAC performance for DOC removal was significantly lower (5-15 %). THM and HAA precursors
C

373 were also significantly lower in the BIEX than in the BAC effluent.
AC

374 • Most DOC removal in the BIEX occurred in the first 15 minutes EBCT. Most of the DOC removal in

375 the BIEX filter (69%) was due to ion exchange.

376 • After one year of operation, the BIEX column was successfully regenerated.

377 After ion exchange resin exhaustion, the BIEX column was observed to support heterotrophic biomass.

378 Considering the difficulty to biodegrade humic substances, we speculate that the media is being
 ACCEPTED MANUSCRIPT

379 regenerated biologically. Further studies are ongoing to discriminate the role of biodegradation and ion

380 exchange in the removal of NOM from exhausted ion exchange resins.

381

PT
382 ACKNOWLEDGEMENTS
383 This research was supported by RES’EAU-WATERNET, an NSERC Strategic Network dedicated to the

RI
384 provision of safe water in small and rural communities. The authors wish to acknowledge the support of
385 Jérôme Leroy, Julie Philibert, Jacinthe Mailly and Claire Wauquiez for their assistance with the chemical
386 analysis, Marie Wendy Andriantsarafara for her assistance with biological analysis and Gabriel St-Jean

SC
387 and Mireille Blais for the support in the pilot plant operation. We would also like to thank Joerg Winter
388 for the numerous useful discussions of the results and Kim Lompe for her final review of the manuscript.
389 This research was conducted in the CREDEAU facilities, a Canadian Foundation for Innovation research
390 infrastructure. Finally, we wish to thank the City of Laval for allowing us to setup the pilot plant at their

U
391 facility.
AN
392

393 REFERENCES
M

394 Amy, G. and Cho, J. (1999) Interactions between natural organic matter (NOM) and membranes:
395 rejection and fouling. Water Science and Technology 40(9), 131-139.
396 Boyer, T.H., Singer, P.C. and Aiken, G.R. (2008) Removal of dissolved organic matter by anion exchange:
397 effect of dissolved organic matter properties. Environmental Science & Technology 42(19), 7431-7437.
D

398 Catalán, N., Casas-Ruiz, J.P., von Schiller, D., Proia, L., Obrador, B., Zwirnmann, E. and Marcé, R. (2017)
399 Biodegradation kinetics of dissolved organic matterchromatographic fractions in an intermittent river.
TE

400 Journal of Geophysical Research: Biogeosciences 122(1), 131-144.

401 CEAEQ (2014) MA. 300 Ions 1.3 - Détermination des anions: méthode par chromatographie ionique,
402 Centre d'expertise en analyse environnementale du Québec.
403 Christman, R.F. and Ghassemi, M. (1966) Chemical nature of organic color in water. American Water
EP

404 Works Association.

405 El Gamal, M., A. Mousa, H., H.El-Naas, M., Zacharia, R. and Judd, S. (2018) Bio-regeneration of activated
406 carbon: A comprehensive review. Separation and Purification Technology 197, 345-359.
C

407 Gibert, O., Lefevre, B., Fernandez, M., Bernat, X., Paraira, M., Calderer, M. and Martınez-Llado, X. (2013)
408 Characterising biofilm development on granular activated carbon used for drinking water production.
AC

409 Water Research.

410 Graveland, A. and Heertjes, P.M. (1975) Removal of manganese from ground water by heterogeneous
411 autocatalytic oxidation. Chemical Engineering Research and Design.
412 Jones, M. (1984) Nitrate reduction by shaking with cadmium: Alternative to cadmium columns. Water
413 Res 18(5), 643-646.
414 Kihn, A., Laurent, P. and Servais, P. (2000) Measurement of potential activity of fixed nitrifying bacteria
415 in biological filters used in drinking water production. Journal of Industrial Microbiology and
416 Biotechnology 24(3), 161-166.
 ACCEPTED MANUSCRIPT

417 Kleiser, G. and Frimmel, F.H. (2000) Removal of precursors for disinfection by-products (DBPs) -
418 differences between ozone- and OH-radical-induced oxidation. Science of the Total Environment 256(1),
419 1-9.
420 Laurent, P., Prévost, M., Cigana, J., Niquette, P. and Servais, P. (1999) Biodegradable organic matter
421 removal in biological filters: evaluation of the chabrol model. Water Research 33(6), 1387-1398.
422 Matilainen, A. and Sillanpaa, M. (2010) Removal of natural organic matter from drinking water by
423 advanced oxidation processes. Chemosphere 80(4), 351-365.

PT
424 Nilson, J.A. and DiGiano, F.A. (1996) Influence of NOM composition on nanofiltration. Journal of the
425 American Water Works Association 88(5), 53-66.
426 Pelekani, C. and Snoeyink, V.L. (1999) Competitive adsorption in natural water: role of activated carbon
427 pore size. Water Research 33(5), 1209-1219.

RI
428 Sarathy, S.R., Bazri, M.M. and Mohseni, M. (2011) Modeling the Transformation of Chromophoric
429 Natural Organic Matter during UV=H2O2 Advanced Oxidation. American Society of Civil Engineers.
430 Schulz, M., Winter, J., Wray, H., Barbeau, B. and Bérubé, P. (2017) Biologically-active ion exchange (BIEX)

SC
431 for NOM-removal and membrane fouling prevention. Water Science and Technology: Water Supply.
432 Smith, E.H. and Weber, W.J. (1985) The Effect of Dissolved Organic Matter on Adsorption Capacity of
433 Organic Compounds on Activated Carbon. American Water Works Association.

U
434 Summers, R.S., Hooper, S.M., Shukairy, H.M., Solarik, G. and Owen, D. (1996) Assessing DBP yield:
435 uniform formation conditions. Journal of the American Water Works Association 88(6), 80-93.
AN
436 Terry, L.G. and Summers, R.S. (2017) Biodegradable organic matter and rapid-rate biofilter performance:
437 A review. Water Research.
438 van der Kooij, D. (1992) Assimilable organic carbon as an indicator of bacterial regrowth. Journal of the
439 American Water Works Association 84(2), 57-65.
M

440 Velten, S., Boller, M., Köster, O., Helbing, J., Weilenmann, H.U. and Hammes, F. (2011) Development of
441 biomass in a drinking water granular active carbon (GAC) filter. Water Research 45(19), 6347-6354.
442 Velten, S., Hammes, F., Boller, M. and Egli, T. (2007) Rapid and direct estimation of active biomass on
D

443 granular activated carbon through adenosine tri-phosphate (ATP) determination. Water Research 41(9),
444 1973-1983.
445 Winter, J., Wray, H.E., Schulz, M., Vortisch, R., Barbeau, B. and Bérubé, P.R. (2018) The impact of loading
TE

446 approach and biological activity on NOM removal by ion exchange resins. Water Research 134, 301-310.
447 Xie, Y. (2003) Disinfection Byproducts in Drinking Water: Formation, Analysis, and Control, Lewis
448 Publishers, New York.
EP

449 Zhang, S., Gitungo, S.W., Axe, L., Raczko, R.F. and Dyksen, J.E. (2017) Biologically active filters e An
450 advanced water treatment process for contaminants of emerging concern. Water Research, 31-41.

451
C
AC
 ACCEPTED MANUSCRIPT

Table 1. Source water characteristics of the Des Prairies River (February 2017 to April 2018).

Parameter Temp. Turbidity DOC2 TOC3 pH Chloride Alkalinity

1
(˚C) (NTU ) (mg/L) (mg/L) (mg/L) (mg CaCO3/L)

Mean ± standard 12.3 13.9 7.09 7.44 7.2 5.8 34

± 7.9 ± 12.1 ± 0.34 ± 0.41 ± 0.2 ± 2.6

PT
Deviation ±7

Number of
39 39 36 24 39 17 42
samples

RI
1 2 3
Nephelometric Turbidity Unit, Dissolved Organic Carbon, Total Organic Carbon. Values are arithmetic averages
with standard deviations. Sulfate concentration is typically in the range of 6-10 mg/L.

U SC
AN
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT

Table 2. Organic carbon mass balances in the BIEX and IEX filters

Influent Effluent Removal (g C)

Type of Filter
(g C) (g C) Total Mechanisms

372 g
Ion Exchange1

PT
BIEX 897 g 354 g 543 g (68.5%)
(After 331 d) (100%) (39.5%) (60.5%) 171 g Assumed

RI
(31.5%) biodegradation
16.8 g
Ion Exchange1
23.1 g 6.3 g 16.9 g (99.4%)

SC
IEX (7 d)
(100%) (27%) (73%) 0.1 g Assumed
(0.6%) biodegradation

U
1: Based on the DOC measured in the spent brine (18.6 g/L for BIEX and 0.84. g/L for IEX).
AN
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
Figure 1. Pilot-plant schematic consisting of four downflow filtration columns filled with GAC, BAC, BIEX
or IEX filter medium. V = 2 m/h. EBCT = 30 min.
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT

600 25 600 25

Temperature of Source Water (ᵒC)

Temperature of Source Water (°C)
aTHM-UFC bHAA5-UFC
500 500

HAA-UFC (µg/L)
20 20
THM-UFC (µg/L)

PT
400 400
15 15
300 300

RI
10 10
200 200
5 5

SC
100 100

0 0 0 0
20
40
60
80
0

100
120

20

40

60

80
0

100

120
U
Days of Operation Days of Operation
Figure 10. THM (a) and HAA (b) precursors concentrations measured under uniform formation
AN
conditions (UFC) in source water and in BIEX, IEX, GAC and BAC effluents. Source water temperature:
dotted line. EBCT = 30 min.
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
Figure 2. Weekly Dissolved Organic Carbon (DOC) monitoring in the source water (SW) and GAC, BAC, BIEX
and IEX effluents over a period of 390 days of operation. EBCT = 30 min, V = 2 m/h, 48 BV/d.
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
Figure 3 - Weekly UV absorbance at 254 nm (UVA254) monitoring in the source water (SW) and GAC, BAC,
M
BIEX and IEX effluents over a period of 390 days of operation. EBCT = 30 min, V = 2 m/h, 48 BV/d.
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
Figure 4. Summary of Dissolved Organic Carbon (DOC) concentrations in the source water and the GAC,
BAC, BIEX and IEX effluents (n = 36 samples over 338 days, i.e. until the first BIEX regeneration). The
groups A, B, C and D were statistically different one from another. Group AB (the BAC filter) was not
statistically different from the source water (A) and the GAC filter (B).
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT

1 / T (K-1) 1 / T (K-1) 1 / T (K-1)

3.3E-03

3.4E-03

3.5E-03

3.6E-03

3.3E-03

3.4E-03

3.5E-03

3.6E-03
3.3E-03

3.4E-03

3.5E-03

3.6E-03
-8.4 -8.5 -11.0
y = -3540x + 0.69
y = -2430x - 0.36 y = -3618x + 3.41
-8.5 R² = 0.32
R² = 0.60 R² = 0.85 -11.2
-8.7

PT
-8.6
-11.4
ln k

ln k
ln k
-8.7 -8.9
-11.6

RI
-8.8
-9.1
-11.8
-8.9
aIEX bBIEX cBAC

SC
-9.0 -9.3 -12.0

Figure 5. Estimation of the energies of activation (temperature effect) for the a. IEX, b. BIEX and c. BAC
columns. The slope of the regression line is equal to Ea/R (J/mole). For example, Ea for IEX is given by

U
2430 x 8.31 = 20 193 J/mole = 20.2 kJ/mole. Ea of IEX and BIEX were calculated with data obtained after
100 days of operation.
AN
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT

PT
RI
SC
Figure 6. Monitoring of ion exchange capacity exhaustion (through chloride release) in parallel with DOC
breakthrough in the (a) IEX column and (b) BIEX column. BIEX regeneration occurred at t = 331 days =

U
15,888 BV.

AN
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT

PT
RI
SC
Figure 7. Monitoring of ion exchange capacity exhaustion (through chloride release) in parallel with DOC

U
breakthrough in the (a) IEX column and (b) BIEX column. BIEX regeneration occurred at t = 331 days =
15,888 BV.
AN
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT

EBCT (min) EBCT (min)

0 7.5 15 22.5 30 0 7.5 15 22.5 30
60 60 20

Effluent Turbidity (NTU)

50 a April 2017 50 b July 2017 C Oct. 2017
15

ATP (ng/cm3)
ATP (ng/cm3)

40 40

PT
30 30 10

20 20
5

RI
10 10
0 0 0
0 25 50 75 100 0 25 50 75 100 0 10 20 30 40 50

SC
Media Depth (cm) Media Depth (cm) Time (min) After Backwash

U
Figure 8. Biomass density (ATP) profiles through depth of the BAC, BIEX and IEX after (a) 7 weeks of
operation (T= 10oC) and (b) 19 weeks of operation (T=23oC). (c) Typical effluent turbidity ripening after
performing a backwash. AN
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
Figure 9. Ammonia removal with respect to (a) ammonia through time and (b) impact of temperature on
ammonia removal. Nitrate and nitrite formation through depth of the BAC, BIEX and IEX columns after
(c) 7 weeks of operation and (d) 35 weeks of operation.
D
TE
C EP
AC