Minimization of short-term low-pressure membrane fouling using a magnetic ion exchange (MIEX®) resin

Accepted Manuscript Minimization of Short-Term Low-Pressure Membrane Fouling Using a Magnetic Ion ® Exchange (MIEX ) Resin Panitan Jutaporn, Philip C. Singer, Rose M. Cory, Orlando Coronell PII:

S0043-1354(16)30204-4

DOI:

10.1016/j.watres.2016.04.007

Reference:

WR 11964

To appear in:

Water Research

Received Date: 23 January 2016 Revised Date:

1 April 2016

Accepted Date: 5 April 2016

Please cite this article as: Jutaporn, P., Singer, P.C., Cory, R.M., Coronell, O., Minimization of Short® Term Low-Pressure Membrane Fouling Using a Magnetic Ion Exchange (MIEX ) Resin, Water Research (2016), doi: 10.1016/j.watres.2016.04.007. 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.

Terrestrial DOM

Microbial DOM

Terrestrial DOM

TE D

M AN U

SC

Microbial DOM

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

Removal by MIEX

Contribution to fouling

ACCEPTED MANUSCRIPT

1 2 3 4

Minimization of Short-Term Low-Pressure Membrane Fouling Using a Magnetic Ion Exchange (MIEX®) Resin Panitan Jutaporna, Philip C. Singera, Rose M. Coryb, Orlando Coronella,* a

Department of Environmental Sciences and Engineering, Gillings School of Global Public Health, The University of North Carolina at Chapel Hill, Chapel Hill, NC 27599, USA b Department of Earth and Environmental Sciences, University of Michigan, Ann Arbor, MI 48109, USA

13

dissolved organic matter (DOM) and membrane fouling by DOM. The magnetic ion exchange

14

resin MIEX® (Ixom Watercare Inc.) has been demonstrated to remove substantial amounts of

15

DOM from many source waters, suggesting that MIEX can both reduce DOM content in

16

membrane feed waters and minimize LPM fouling. We tested the effect of MIEX pretreatment

17

on the reduction of short-term LPM fouling potential using feed waters varying in DOM

18

concentration and composition. Four natural and two synthetic waters were studied and a

19

polyvinylidene fluoride (PVDF) hollow-fiber ultrafiltration membrane was used in membrane

20

fouling tests. To evaluate whether MIEX removes the fractions of DOM that cause LPM fouling,

21

the DOM in raw, MIEX-treated, and membrane feed and backwash waters was characterized in

22

terms of DOM concentration and composition. Results showed that: (i) the efficacy of MIEX to

23

reduce LPM fouling varies broadly with source water; (ii) MIEX preferentially removes

24

terrestrial DOM over microbial DOM; (iii) DOM is a more important contributor to LPM fouling

25

than terrestrial DOM, relative to their respective concentrations in source waters; and (iv) the

26

fluorescence intensity of microbial DOM in source waters can be used as a quantitative indicator

27

of the ability of MIEX to reduce their membrane fouling potential. Thus, when ion exchange

28

resin processes are used for DOM removal towards membrane fouling reduction, it is advisable

29

to use a resin that has been designed to effectively remove microbial DOM.

RI PT

5 6 7 8 9 10 11 12

*Corresponding author [tel: 1-919-966-9010; fax: +1-919-966-7911; e-mail: [email protected]]

AC C

EP

TE D

M AN U

SC

Abstract: Two challenges to low-pressure membrane (LPM) filtration are limited rejection of

1

ACCEPTED MANUSCRIPT

Keywords: ultrafiltration, MIEX, magnetic ion exchange resin, dissolved organic matter,

31

fluorescence, fouling

32

1. Introduction

33

Low-pressure membrane (LPM) filtration (i.e., microfiltration, ultrafiltration) has been widely

34

employed in drinking water treatment and wastewater reuse because it can effectively remove

35

microbial and particulate contaminants with relatively low energy consumption (Zularisam et al.,

36

2006). One important challenge to LPM filtration is low rejection of dissolved organic matter

37

(DOM) as a result of the relatively large size of the membrane pores compared to the average

38

molecular size of DOM (Fan et al., 2001). Reduction of DOM content in drinking water

39

treatment is desirable because DOM is a source of undesirable color, taste and odor (Carroll et al.,

40

2000), coagulant and oxidant demand, and disinfection byproducts (DBPs) (Croué et al., 1999a).

41

A second important challenge to LMP filtration is membrane fouling by DOM (Fan et al., 2001;

42

Huang et al., 2009a), which consists of the deposition of organic matter on the membrane surface

43

and/or membrane pores. LPM fouling by DOM leads to decreased water productivity, and

44

increased cleaning frequency and operational costs (Cheryan, 1998).

45

The DOM content in source waters, and therefore LPM fouling by DOM, can be successfully

46

reduced by coagulation pretreatment prior to membrane filtration, with higher coagulant and

47

associated operational costs needed for waters with higher DOM content (Carroll et al., 2000).

48

As an alternative, anion exchange resins can also efficiently remove DOM from source waters

49

(Bolto et al., 2002; Fearing et al., 2004). One such resin, the strong base anion exchange resin

50

MIEX® DOC (Ixom Watercare Inc.), was developed to remove dissolved organic carbon (DOC)

51

in water treatment applications (Neale & Schäfer, 2009). MIEX resin beads contain magnetized

AC C

EP

TE D

M AN U

SC

RI PT

30

2

ACCEPTED MANUSCRIPT

cores (Slunjski et al., 2000) that enhance particle agglomeration and settling during the resin

53

separation process (Boyer & Singer, 2006). The magnetized core is surrounded by a macro-

54

porous polymer shell with quaternary amine functional groups that allow for DOC removal by

55

ion exchange (Wert et al., 2005). The average size of the resin beads is 180 µm (Singer & Bilyk,

56

2002), and this relatively small size results in a high surface area that enhances DOC removal

57

kinetics and regeneration efficiency (Johnson & Singer, 2004).

58

In recent years, MIEX has been proven effective in removing DOC (including DBP precursors)

59

from a broad range of source waters (Boyer & Singer, 2005; Fearing et al., 2004; Humbert et al.,

60

2005), and in reducing the coagulant dose needed for particulate removal (Huang et al., 2012a;

61

Humbert et al., 2007; Singer & Bilyk, 2002). MIEX has also been reported to remove a broad

62

range of organics within the DOM pool, including hydrophilic and hydrophobic organic acids

63

(Singer & Bilyk, 2002; Zhang et al., 2006) varying from low to high molecular weight (Humbert

64

et al., 2005). Given the demonstrated ability of MIEX to remove organics from water, it is

65

reasonable to expect that MIEX can potentially minimize LPM fouling by DOM.

66

Previous studies have reported that MIEX can reduce LPM fouling either by direct pretreatment

67

of LPM influent waters (Dixon et al., 2010; Zhang et al., 2006, 2007, 2008), or by integrated use

68

of MIEX and other processes such as coagulation/flocculation (Fearing et al., 2004; Huang et al.,

69

2012a; Humbert et al., 2007). However, the effectiveness of MIEX to reduce LPM fouling

70

without additional pretreatment steps varies across different studies. For example, some

71

researchers have reported that MIEX reduced LPM fouling potential (Dixon et al., 2010; Huang

72

et al., 2012b; Zhang et al., 2006, 2007), while others have reported no change in fouling potential

73

after MIEX pretreatment (Fabris et al., 2007; Fan et al, 2008; Humbert et al., 2007). Contrasting

AC C

EP

TE D

M AN U

SC

RI PT

52

3

ACCEPTED MANUSCRIPT

findings on the effectiveness of MIEX to reduce LPM fouling could be due to differences in

75

experimental procedures, or in water quality among the source waters tested.

76

DOM is a complex mixture of organic molecules derived from the decay of terrestrial (plant and

77

soil) and microbial matter (Leenheer & Croué, 2003). Depending on the contribution of

78

terrestrial and microbial sources to the DOM pool, the organic compounds comprising DOM can

79

range from more hydrophobic to more hydrophilic, and from high to low molecular weight,

80

respectively (Leenheer & Croué, 2003). These DOM properties are expected to influence its

81

LPM fouling potential because hydrophobic and steric interactions are influential factors in

82

membrane fouling (Cho et al., 2000; Howe & Clark, 2002), with reports indicating that higher

83

DOM aromaticity (Fan et al., 2001) and higher molecular-weight DOM (Fabris et al., 2007; Lee

84

at al., 2005) lead to greater water flux decline.

85

Despite the expected importance of DOM composition on LPM fouling, only a few studies have

86

investigated how MIEX may preferentially remove (or not) the fractions of DOM that cause

87

fouling (Dixon et al., 2010; Fabris et al., 2007; Humbert et al., 2007). For example, Dixon et al.

88

(2010) concluded that a portion of LPM fouling was caused by organic acids which were

89

removed by MIEX. Humbert et al. (2007) found that high molecular-weight DOM of microbial

90

origin contributed substantially to LPM fouling, and this DOM fraction was not well removed by

91

MIEX due to size exclusion from resin pores. Fabris et al. (2007) found that MIEX was not

92

effective at removing very high molecular-weight (>50,000 Da) colloidal DOM, a fraction

93

strongly associated with LPM fouling. These results suggest that DOM source, in addition to

94

hydrophobicity and size, influence fouling, and that all fractions of DOM are not removed to the

95

same extent by MIEX.

AC C

EP

TE D

M AN U

SC

RI PT

74

4

ACCEPTED MANUSCRIPT

DOM source and composition can be readily analyzed by absorbance and fluorescence

97

spectroscopy (Coble et al., 1990; Cory & McKnight, 2005). For example, percent aromaticity of

98

DOM is strongly correlated with specific ultraviolet absorbance (SUVA) (Weishaar et al., 2003),

99

which is defined as ultraviolet absorbance at 254 nm (UVA254) normalized to DOC concentration.

RI PT

96

Also, quantitative identification of DOM origins and structural properties can be obtained from

101

excitation-emission matrices (EEMs) (Coble et al., 1990; Coble, 1996; McKnight et al., 2001),

102

which are 3-dimensional fluorescence intensity landscapes resulting from collections of

103

fluorescence emission scans for a range of excitation wavelengths. The relative concentration of

104

different types of fluorophores in a water sample can be examined on the basis of specific peak

105

intensities from the EEMs. Specifically, peak A (240-270 nm excitation and 380-470 nm

106

emission) and peak C (300-340 nm excitation and 400-450 nm emission) represent terrestrial

107

humic materials, while peak T (260-290 nm excitation and 326-350 emission nm) represents

108

microbial protein-like fluorophores (Cory & McKnight, 2005; Yamashita & Tanoue, 2003).

109

The objective of this study was to investigate the relationship between DOM removal by MIEX

110

(extent and type of DOM removed) and the associated LPM fouling reduction. We used a range

111

of water sources, one common procedure for all fouling experiments, and a comprehensive

112

methodology for characterizing the nature of DOM in source waters. We present MIEX

113

pretreatment and LPM fouling results along with DOM characterization via DOC, UVA254,

114

SUVA, and fluorescence EEMs analysis. In the discussion and conclusion sections, we interpret

115

our findings to (i) explain why MIEX pretreatment reduces LPM fouling only for some water

116

sources, and (ii) suggest which types of source waters may benefit most from MIEX

117

pretreatment to reduce LPM fouling.

118

2. Materials and Methods

AC C

EP

TE D

M AN U

SC

100

5

ACCEPTED MANUSCRIPT

2.1. Water samples

120

We used four natural waters and two synthetic waters spanning broad ranges of DOC (1.8-11

121

mg/L), UVA254 (0.056-0.208 cm-1), and SUVA (2.0-5.4 L/mg.m) levels. The various water

122

quality parameters of interest for the six “raw” waters are presented in Table 1. “Raw” refers to

123

the state of the waters after laboratory filtration, as described below. Natural water samples were

124

collected from the intake of drinking water treatment plants using, as source waters, the

125

University Lake (UL) and Cane Creek (CC) reservoirs in Orange County, North Carolina, the

126

Tennessee River in Muscle Shoals (MS), Alabama, and Clear Lake in West Palm Beach (PB),

127

Florida, all located in the United States of America. All natural water samples were transported

128

to our laboratories the same day they were collected, either directly from nearby water treatment

129

plants or via overnight shipping refrigerated with cold-packs. Immediately after arrival, all water

130

samples were filtered with Whatman (Piscataway, NJ) GF/C 1.2-µm glass fiber filters to remove

131

suspended solids, and then stored at 4oC in the dark until needed. Synthetic waters were prepared

132

using Suwannee River humic acid standard II (SR) and Pony Lake fulvic acid reference (PL)

133

from the International Humic Substances Society (IHSS). The SR and PL isolates are IHSS

134

references for terrestrial humic acids and microbial fulvic acids, respectively (Averett et al.,

135

1994; Brown et al., 2004), and have properties at the ends of the chemical spectrum found for

136

DOM in natural waters (Cory et al., 2010). The SR isolate possesses properties associated with

137

terrestrial DOM, such as high SUVA and high aromaticity (Averett et al., 1994; Boyer et al.,

138

2008; McKnight & Aiken, 1998), and has been widely used as a DOM model in LPM fouling

139

studies (Aoustin et al., 2001; Jones & Melia, 2000; Schäfer et al., 2000; Yuan & Zydney, 1999,

140

2000). The PL isolate has properties associated with microbial DOM, such as low SUVA and

141

low aromaticity (Boyer et al., 2008; Brown et al., 2004; McKnight et al., 1994). Synthetic waters

AC C

EP

TE D

M AN U

SC

RI PT

119

6

ACCEPTED MANUSCRIPT

were prepared by dissolving the SR and PL isolates in ultrapure water (>18 MΩ⋅cm) adjusted to

143

pH 10.5 using 0.1 N sodium hydroxide (NaOH), stirring the solution overnight for complete

144

dissolution of the isolate, and adjusting the pH to 6.5-7.5 using 0.1 N hydrochloric acid (HCl)

145

before use.

146

2.2. MIEX pretreatment

147

MIEX DOC resin, provided by Ixom Watercare Inc., was stored wet in a 5% NaCl solution. The

148

resin concentration is reported as milliliters of settled resin per liter of water. For dosing

149

purposes, the resin was allowed to settle in a glass graduated cylinder for 30 minutes before

150

measuring the settled resin volume. Batch experiments were used to evaluate the removal of

151

DOC and UVA254 as a function of MIEX dose (1-10 mL/L). The experiments were performed

152

using a jar test apparatus with 2-L square jars (Phipps and Bird Inc., Richmond, VA). The

153

mixing protocol entailed mixing at 100 rpm for 1 hour and settling for 15 min. After settling, the

154

supernatant was decanted and vacuum-filtered through Whatman GF/C 1.2-µm glass fiber filters.

155

For pretreatment of waters used in fouling tests, we compared a 2 mL/L MIEX resin dose (i.e., a

156

typical dose in full-scale treatment plants) to a higher MIEX dose of 10 mL/L. Each condition

157

for MIEX pretreatment experiments was tested in triplicate and the average of the triplicate

158

results is reported.

159

2.3. Bench-scale membrane fouling tests

160

For the six water sources listed in Table 1, we investigated the LPM fouling potential of raw

161

waters (Raw), raw waters pretreated with MIEX doses of 2 mL/L (MIEX2) and 10 mL/L

162

(MIEX10), and raw waters diluted (Dilute) with ultrapure water to the same DOC concentration

163

of waters pretreated with a MIEX dose of 2 mL/L. The membranes used consisted of hydrophilic

AC C

EP

TE D

M AN U

SC

RI PT

142

7

ACCEPTED MANUSCRIPT

polyvinylidene fluoride (PVDF), hollow-fiber ultrafiltration membranes (GE Water & Process

165

Technologies, Oakville, Ontario, Canada) with nominal pore size of 0.04 µm and outside

166

diameter of 1.95 mm. The PVDF membrane had a contact angle in the range of 60-65o (provided

167

by the manufacturer), which is similar to that of other hydrophilic PVDF membranes, e.g. Myat

168

et al., (2013). The membranes were tested using the setup depicted in Figure 1 which included a

169

hollow-fiber module consisting of four fibers, each having a length of 16 cm, providing a total

170

membrane surface area of 39.2 cm2. The module was operated under a constant vacuum pressure

171

of 5.8 psi (40 kPa) provided by a peristaltic pump (Masterflex, Cole-Parmer, Vernon Hills, IL).

172

Permeate flow rate was monitored continuously using a digital balance (Ohaus, Parsippany, NJ).

173

Each fouling run lasted at least six hours, and each experimental condition was tested in

174

duplicate. Feed, concentrate and permeate water samples were collected for organic matter

175

analyses (see Section 2.4) at 1, 3 and 5 hr during the filtration run. All tests were conducted at

176

room temperature (20±2oC). The initial permeate flux varied in the range of 8.43-10.8 cm/hr. For

177

each source water tested, the relative standard deviation of the average initial permeate water

178

fluxes for Raw, MIEX2 and MIEX10, and Dilute waters was always less than 5%.

179

The LPM fouling potential of each water was quantified using the unified membrane fouling

180

index (UMFI) as defined by (Huang et al., 2008)

181

J0 = 1 + UMFI ⋅VS , Jt

182

where J 0 (m/s) and J t (m/s) correspond to the water fluxes at time zero and time t, respectively,

183

and VS (L/m2) corresponds to the cumulative volume of water filtered at time t per unit area of

184

membrane. Greater UMFI values correspond to greater LPM fouling potential (Huang et al.,

185

2009b). The UMFI values calculated in this study describe total fouling (including both

AC C

EP

TE D

M AN U

SC

RI PT

164

(1)

8

ACCEPTED MANUSCRIPT

reversible and irreversible fouling) and were obtained by fitting the experimental J 0 / J t and VS

187

data from the 6-hour filtration runs to Equation 1.

188

At the end of each fouling experiment, hydraulic and chemical cleaning were performed. For

189

hydraulic cleaning, the membrane module was first backwashed at 100 kPa using 70 mL of

190

ultrapure water and then rinsed by spraying 30 mL of ultrapure water on the membrane surface.

191

For each membrane fouling test with raw and MIEX -treated waters, the backwash and rinse

192

water (a total of 100 mL) were collected in clean beakers for organic matter analyses (see

193

Section 2.4). Chemical cleaning was performed after hydraulic cleaning, and consisted of

194

submerging and aerating the module in 1% sodium hypochlorite (NaOCl) solution for 15

195

minutes (Nghiem et al., 2009). To assess the water flux recovery due to membrane cleaning, we

196

measured the clean water permeate flux (40 kPa, average of 15 min) after each hydraulic

197

cleaning and chemical cleaning procedures. Given that the water permeate flux recovered by

198

hydraulic cleaning was always higher than 94.2% (average of 98.8%, except for one test where it

199

was 90.3%), the corresponding organic matter collected represented well the organic matter

200

contributing to fouling. For chemical cleaning, the water permeate flux recovery was always

201

higher than 97.7% (average of 101%), indicating effective membrane cleaning by NaOCl.

202

2.4. Characterization of organic matter

203

The DOC concentration (mg/L) in water was measured using a TOC-V organic carbon analyzer

204

(Shimadzu, Atlanta, GA). UVA254 measurements were made using a 1-cm quartz cell and a U-

205

2000 spectrophotometer (Hitachi Instruments Inc., Danbury, CT). SUVA values were calculated

206

as SUVA=100(UVA254/DOC). Fluorescence measurements were conducted in a 1-cm quartz cell

207

using Aqualog and Fluorolog-321 spectro-fluorometers (Horiba JobinYvon, Edison, NJ, USA),

AC C

EP

TE D

M AN U

SC

RI PT

186

9

ACCEPTED MANUSCRIPT

and a diode array UV-Vis spectrophotometer (Hewlett Packard, Palo Alto, CA, USA). EEMs

209

fluorescence spectroscopy was employed to identify the relative abundance of organic fractions

210

from different origins in the samples analyzed, namely terrestrial humics and microbial protein-

211

like organics, as identified by their emission intensities at excitation/emission coordinates of 250

212

nm/450 nm (component A, terrestrial humics stimulated by UV excitation), 350 nm/450 nm

213

(component C, terrestrial humics stimulated by visible excitation), and 275 nm/340 nm

214

(component T, microbial protein-like organics) (Coble, 1996; Stedmon et al., 2003). EEMs

215

obtained with ultrapure water blanks were collected daily and subtracted from the EEM of each

216

sample. Fluorescence intensities are reported in Raman units (RU) by normalization of the

217

intensities to the area under the water-Raman peak at an excitation of 350 nm (Cory et al., 2010;

218

Stedmon et al., 2003). Additional details about fluorescence spectroscopy analyses are provided

219

in the Supporting Materials.

220

3. Results and Discussion

221

3.1. Removal of organic matter by MIEX pretreatment

222

Figure 2 presents results for the removal of DOC and UVA254 from water by MIEX treatment.

223

The results show that MIEX was effective at removing significant levels of DOC (38-80% with a

224

2 mL/L MIEX dose, Figure 2a) and UVA254 (36-87% with a 2 mL/L MIEX dose, Figure 2b)

225

from all waters. Statistical analyses of the results in Figure 2 showed that DOC removal was not

226

strongly correlated with the initial DOC concentration of the raw water (R2=0.37 for DOC

227

removal with a 2 mL/L MIEX dose). Similarly, UVA254 removal did not correlate well with the

228

initial UVA254 (R2=0.18 for UVA254 removal with a 2 mL/L MIEX dose). These findings are

229

consistent with previous studies, which reported significant removal of DOC by MIEX but no

AC C

EP

TE D

M AN U

SC

RI PT

208

10

ACCEPTED MANUSCRIPT

significant relationship between raw water DOC concentration and DOC removal by MIEX

231

(Boyer & Singer, 2006; Mergen et al., 2008; Singer et al., 2009; Singer et al., 2007). Together

232

with these previous studies, our results indicate that DOC and UVA254 content in raw waters are

233

not good predictors of the effectiveness of MIEX treatment at removing organic content.

234

For nearly all waters and MIEX doses, UVA254 removal was higher than DOC removal (except

235

for SR water at a 2 mL/L MIEX dose, Figure 2). These results demonstrate that MIEX

236

preferentially removes the UV-absorbing fraction of organic matter over the non-UV absorbing

237

fraction, thus resulting in lower SUVA values in treated waters than in raw waters. The affinity

238

of MIEX for the UV-absorbing fraction of organics has been previously reported in the literature

239

(Fearing et al., 2004; Johnson & Singer, 2004; Kitis et al., 2007; Wert et al., 2005).

240

For all natural waters and the synthetic PL water (representative of source waters enriched in

241

microbial DOM (Brown et al., 2004)), removal of organic material increased appreciably with

242

MIEX doses up to approximately 2 mL/L at which point removal started to plateau (Figure 2).

243

This observation is consistent with the results obtained by Johnson and Singer (2004) for

244

groundwater from California (USA). By contrast, for the SR water (representative of source

245

waters enriched in terrestrial DOM (Averett et al., 1994)), removal of organics increased

246

gradually across the entire MIEX dose range and was lower than that for the other waters at

247

MIEX doses of up to 5 mL/L. One likely contributor to the behavior observed in Figure 2 for

248

DOM removal as a function of MIEX dose (i.e., removal starting to plateau at a resin dose of 2

249

mL/L for some waters, and lack of correlation between DOM removal and MIEX dose) is that

250

there is excess ion exchange capacity in virtue of the MIEX dose compared to the equivalents of

251

charged organic matter present in solution. For example, using an ion exchange capacity of 0.5

252

meq/mL for MIEX resin and a charge density of 10 meq/g-C for DOM (Boyer & Singer, 2008),

AC C

EP

TE D

M AN U

SC

RI PT

230

11

ACCEPTED MANUSCRIPT

a MIEX dose of 2 mL/L is equivalent to 1 meq/L and a DOM content of 10 mg-C/L is equivalent

254

to 0.1 meq/L, which would indicate 10 times excess ion exchange capacity. Mass transfer

255

limitations are also likely contributors to the behavior observed in Figure 2 for DOM removal.

256

For example, the larger average size of the organics in SR water than organics in the other waters

257

may result in size exclusion from the resin pores; size exclusion is thought to be the primary

258

mechanism responsible for reduced organic matter removal by anion exchange resins (Bolto et

259

al., 2002; Croué et al., 1999b). The lower removals for SR water may also be due to kinetics, as

260

the higher molecular weight of the SR organics would make them absorb into the resin more

261

slowly. SR organics have a larger average molecular weight (2,748 amu) (Aoustin et al., 2001)

262

compared with PL organics (1,260-1,470 amu) (Brown et al., 2004).

263

3.2. Reduction in membrane fouling potential by MIEX pretreatment

264

Given that the extent of DOC and UVA254 removal started to plateau at approximately 2 mL/L

265

for all but one water, we used a 2 mL/L MIEX dose for pretreatment of waters used in LPM

266

fouling tests; as stated in Section 2.2, this dose is representative of those used in full-scale

267

treatment plants. We also used a 10 mL/L MIEX dose to determine whether higher doses could

268

further reduce membrane fouling. The results for MS water are presented in Figure 3a and show

269

that MIEX pretreatment led to a reduction in LPM fouling, as indicated by a lower flux decline

270

over time for MIEX-pretreated water than for raw water. Similar results (not shown) were

271

obtained for all natural and synthetic waters, though the extent of fouling reduction varied with

272

source water. We fitted experimental results for J 0 / J t and VS to Equation 1 to obtain UMFI

273

values (see Figure 3b for illustrative example using MS water). The UMFI results are presented

274

in Figure 3c for all waters and show that, for a 2 mL/L MIEX dose, MIEX pretreatment led to

AC C

EP

TE D

M AN U

SC

RI PT

253

12

ACCEPTED MANUSCRIPT

UMFI reductions as low as 3.5% (UL water) and up to 56% (SR water). Increasing the MIEX

276

dose by a factor of 5, from 2 mL/L to 10 mL/L, resulted in only marginal increases in percentage

277

UMFI reduction (Figure 3c). This observation is consistent with the marginal increase in

278

organics removal observed when the MIEX dose was increased from 2 mL/L to 10 mL/L (Figure

279

2). The marginal increase in UMFI reduction observed with a 5-fold increase in MIEX dose

280

occurred even for the SR water for which the removal of organic matter increased continuously

281

as a function of MIEX dose.

282

3.3. Relationship between the removal of organics by MIEX pretreatment and reduction in

283

membrane fouling potential

284

We evaluated the association between fouling potential reduction (quantified by the percent

285

UMFI reduction) and DOM removal by MIEX. The results (Figure 4a) show that there was not a

286

strong correlation between fouling potential reduction and DOC removal, and that higher DOC

287

removal did not result in higher fouling potential reduction. Similarly, there was not a strong

288

correlation between fouling potential reduction and UVA254 or SUVA reduction (Figures A1a-b

289

in Supplementary Material). A strong association was not observed either between fouling

290

potential reduction and initial DOC, UVA254, or SUVA values in raw waters (Figures A2a-c in

291

Supplementary Material). Therefore, neither the extent of removal of DOC, UVA254, or SUVA

292

by MIEX, nor the initial DOC, UVA254, or SUVA values in raw waters, are good indicators of

293

the reduction in fouling potential by MIEX. These results are consistent with previous studies

294

(Henderson et al., 2011; Lee, et al., 2004) that reported no significant correlation between DOC

295

and fouling when more than one source water was investigated, and our own observation that

296

there was not a strong correlation between the fouling potential of raw waters (quantified by the

297

UMFI) and their DOC concentration (Figure 4b). Similarly, there was not a strong correlation

AC C

EP

TE D

M AN U

SC

RI PT

275

13

ACCEPTED MANUSCRIPT

between fouling potential and DOC, UVA254, or SUVA levels for all waters tested (i.e., raw,

299

MIEX-pretreated, and diluted, Figures A3a-c in Supplementary Material).

300

3.4. Relationship between the organic fractions removed by MIEX pretreatment and those

301

that cause membrane fouling

302

Given that neither DOM removal by MIEX, nor initial DOM concentration, explain the

303

variability in fouling potential reduction by MIEX, we investigated the relationship between the

304

organic matter fractions removed by MIEX and those that fouled the membranes. Specifically,

305

for each source water, we compared the fouling potential of waters pretreated with a 2mL/L

306

MIEX dose (MIEX2) and waters diluted with ultrapure water (Diluted) to the same DOC

307

concentration of the MIEX2 waters (see Figure 3c). In this case, differences in fouling potential

308

could only be due to different characteristics of the organic matter because the waters had

309

approximately the same DOC concentration. Any differences in organic matter composition

310

between MIEX2 and diluted waters could only have been produced by MIEX pretreatment.

311

For the natural waters, the fouling potential of MIEX2 waters was 2-4 times as high as that of

312

their respective diluted waters (Figure 3c). For example, although the DOC concentrations of

313

MIEX2 (3.3 mg/L) and diluted (3.5 mg/L) PB waters were similar, the UMFI of the MIEX2 PB

314

water (0.0066 m2/L) was approximately three times as high as that of the diluted PB water

315

(0.0021 m2/L). The different fouling potentials consistently observed for MIEX2 waters and

316

diluted waters with the same DOC concentration indicate that the two waters contained different

317

fractions of organic matter which resulted in different fouling behavior. Moreover, given that the

318

MIEX2 waters had a higher fouling potential than the diluted waters, we conclude that MIEX

319

preferentially removed DOM fractions that were not dominant contributors to membrane fouling.

AC C

EP

TE D

M AN U

SC

RI PT

298

14

ACCEPTED MANUSCRIPT

The synthetic PL water (representative of water enriched in microbial DOM) showed a similar

321

trend to that observed for the natural waters (i.e., higher UMFI in MIEX2 water than in Diluted

322

water, Figure 3c). In contrast, the synthetic SR water (representative of water enriched in

323

terrestrial DOM) showed the opposite trend (i.e., lower UMFI in MIEX2 water than in Diluted

324

water, Figure 3c). Given that natural waters showed the same trend as the synthetic PL water in

325

terms of fouling potential reduction upon MIEX pretreatment and dilution of the raw water, we

326

conclude that microbial DOM was likely a main source of fouling in the natural waters

327

investigated.

328

3.5. Fate of terrestrial and microbial DOM during MIEX treatment of raw waters and

329

membrane fouling tests

330

We characterized the relative abundance of terrestrial and microbial DOM in raw and MIEX-

331

pretreated waters, as well as in foulant layers, using EEMs. A representative EEM obtained for

332

raw UL water is shown in Figure 5a. The EEM analysis indicates that the main components of

333

organic matter present in raw UL water were terrestrial DOM (components A and C) and

334

microbial DOM (component T) (Leenheer & Croué, 2003; McKnight et al., 2001; Stedmon et al.,

335

2003). Components A, C and T were also the main fluorescent organic components identified in

336

the EEMs of all other raw waters investigated (not shown).

337

The fluorescence intensity of components A, C, and T for all raw waters is presented in Figure

338

5b, and the percentage removal of the fluorescence intensity of each component by MIEX 2

339

mL/L pretreatment is presented in Figure 5c. Three observations can be made from the results in

340

Figure 5 regarding the relative abundance of the different types of organic fractions in the raw

341

waters and their respective removal by MIEX pretreatment. First, in every raw water, the

AC C

EP

TE D

M AN U

SC

RI PT

320

15

ACCEPTED MANUSCRIPT

terrestrial DOM (components A and C) had a greater fluorescence intensity than microbial DOM

343

(component T), thus indicating greater abundance of terrestrial DOM over microbial DOM

344

(Figures 5a and 5b). This is consistent with literature findings for most natural waters (Coble,

345

2014). Second, there was no significant difference between the removal of the two terrestrial

346

DOM components (components A and C) by MIEX pretreatment for any of the waters (Figure

347

5c). Lastly, the removal of terrestrial DOM (75-90% fluorescence loss of components A and C,

348

Figure 5c) by MIEX was always higher than the removal of the corresponding microbial DOM

349

(44-80% fluorescence loss of component T, Figure 5c). The preferential removal by MIEX of

350

humic substances over biopolymers, such as proteins and polysaccharides, was previously

351

reported for secondary effluent (Myat et al., 2013), and was explained by size

352

exclusion/blockage of the resin exchange sites by the high molecular weight biopolymers

353

compared to the smaller humic molecules. Overall, compared to microbial DOM, terrestrial

354

DOM was both more abundant in all of the natural waters tested and was removed to a greater

355

extent by MIEX pretreatment.

356

To probe the fraction of DOM that fouled the membranes, we analyzed the EEMs of the

357

backwash waters of membranes fouled with raw waters and MIEX2 waters. Figures 6a and 6b

358

show illustrative EEMs obtained for the backwash waters of raw UL and MIEX2 UL waters,

359

respectively. Both EEMs are qualitatively similar, indicating the presence of both microbial

360

DOM (component T) and terrestrial DOM (components A and C), and relatively high and low

361

fluorescence intensities associated with microbial DOM and terrestrial DOM, respectively.

362

EEMs with similar characteristics were obtained for all of the other backwash waters (not

363

shown). Given that the membranes were backwashed with ultrapure water, the presence of

AC C

EP

TE D

M AN U

SC

RI PT

342

16

ACCEPTED MANUSCRIPT

fluorescence signals associated with microbial and terrestrial DOM in the backwash waters

365

indicates that both microbial and terrestrial DOM were present in the foulant layers.

366

The EEMs obtained for all backwash waters were analyzed, and the corresponding results for the

367

fluorescence intensities of components A, C, and T are presented in Figures 6c-d. A comparison

368

between the fluorescence intensities for raw waters (Figure 5b) and the corresponding

369

fluorescence intensities for backwash waters of membranes fouled with the raw waters (Figure

370

6c) indicates that the relative abundance of microbial DOM (component T) with respect to that

371

of terrestrial DOM (components A and C) was larger in all backwash waters than in their

372

respective raw waters. Specifically, while the ratio between the fluorescence intensity of

373

component T and the fluorescence intensity of component A for the raw SR, PL, CC, UL, MS

374

and PB waters were 0.06, 0.16, 0.20, 0.21, 0.22 and 0.13, respectively, the corresponding ratios

375

for the backwash waters were 0.13, 0.45, 0.78, 1.13, 1.78 and 1.02. The same trend is observed

376

from a comparison of the fluorescence intensities for MIEX2 waters and the corresponding

377

fluorescence intensities for backwash waters of membranes fouled with MIEX2 waters. These

378

results indicate that microbial DOM caused preferential fouling over terrestrial DOM, relative to

379

their respective abundance in feed waters, despite the fact that all feed waters had a significantly

380

higher concentration of terrestrial DOM than of microbial DOM. This finding is consistent with

381

previous studies showing that protein-like materials were major contributors to LPM membrane

382

fouling during filtration of secondary effluent (Fan et al. 2008; Nguyen et al., 2010) and natural

383

surface water (Lee et al., 2006; Shao et al. 2014), and that fouling correlates with biopolymers in

384

the filtration of surface water (Yamamura et al., 2007).

385

A comparison between the fluorescence intensities for backwash waters of membranes fouled

386

with raw waters (Figure 6c) and membranes fouled with corresponding MIEX2 waters (Figure

AC C

EP

TE D

M AN U

SC

RI PT

364

17

ACCEPTED MANUSCRIPT

6d) shows that the relative abundance of microbial DOM with respect to that of terrestrial DOM

388

was greater in the foulant layers of membranes fouled with the MIEX-pretreated waters. This

389

observation is consistent with our previous finding that MIEX preferentially removes terrestrial

390

DOM over microbial DOM (Figures 5b-c) and therefore that the feed water to the membranes

391

after MIEX treatment is enriched with respect to microbial DOM compared to terrestrial DOM.

392

Clean water permeability tests of the membrane after hydraulic cleaning showed that the foulant

393

layers from our fouling tests were effectively removed by hydraulic cleaning (i.e., average of

394

98.8% permeate water flux recovery). Lee et al., (2004) reported that fouling of UF membranes

395

was caused mainly by cake/gel layer formation and this fouling could be recovered effectively by

396

backwashing. Similarly, Myat et al. (2014) reported that cake layer formation was the dominant

397

fouling mechanism in filtration studies in which the filter cake was comprised mainly of

398

biopolymers that could be extensively removed by hydraulic backwashing. The similarity

399

between our results and those from the literature discussed above in terms of foulant layer

400

composition and its removal by hydraulic cleaning suggests that cake layer formation was the

401

primary fouling mechanisms in our study.

402

Overall, the results show that while microbial DOM (component T) is present in lowest

403

abundance in all of the raw waters tested, it is disproportionately more important to membrane

404

fouling than terrestrial DOM, relative to their respective abundance in the feed water. In addition,

405

microbial DOM is also the fraction that is least effectively removed by MIEX pretreatment.

406

Given that microbial contributions to the DOM pool increase with increasing water residence

407

time (Cory et al., 2007), MIEX is likely more effective in fouling reduction in stream waters (e.g.,

408

MS) than in source waters with larger water residence times such as lakes and impoundments

409

(e.g., PB).

AC C

EP

TE D

M AN U

SC

RI PT

387

18

ACCEPTED MANUSCRIPT

3.6. Component T as a quantitative indicator of membrane fouling potential reduction by

411

MIEX

412

We used least-square linear regression analysis to evaluate the correlation between LPM fouling

413

potential reduction by MIEX pretreatment and the fluorescence intensity of component T in raw

414

waters (Figure 7). The results show that the percent UMFI reduction by MIEX and the

415

fluorescence intensity of component T were strongly correlated (R2=0.95, p=0.001 for MIEX2,

416

and R2=0.79, p=0.04 for MIEX10) via an inverse linear correlation. Similar analyses performed

417

for components A and C (Figures A4a-b in Supplementary Material) indicated that the UMFI

418

reduction by MIEX 2 and 10 mL/L pretreatment was not strongly correlated to their respective

419

fluorescence intensities (R2=0.37-0.58, p=0.08-0.28 for component A, and R2=0.42-57, p=0.08-

420

0.24 for component C). These results indicate that, at a significance level of 0.05, membrane

421

fouling potential reduction by MIEX is strongly correlated only to the fluorescence intensity of

422

microbial DOM (not terrestrial DOM), where the effectiveness in membrane fouling potential

423

reduction by MIEX decreases as the fluorescence intensity of microbial DOM increases in raw

424

waters. Therefore, the results indicate that the fluorescence intensity of component T in a source

425

water may serve as a quantitative indicator of the effectiveness of MIEX treatment to reduce

426

membrane fouling potential for that water.

427

Overall, the results indicate that given that microbial DOM causes preferential fouling over

428

terrestrial DOM, but MIEX does not remove microbial DOM as well as it removes terrestrial

429

DOM, the efficacy of MIEX to reduce LPM fouling is lower in waters enriched with microbial

430

DOM. Therefore, for membrane pre-treatment applications targeting fouling potential reduction,

431

further development of the MIEX resin should target increasing the removal of microbial DOM.

432

4. Conclusions

AC C

EP

TE D

M AN U

SC

RI PT

410

19

ACCEPTED MANUSCRIPT

We studied the minimization of low-pressure PVDF membrane fouling by pretreatment of feed

434

waters using MIEX resin. We investigated the relationship between DOM removal by MIEX

435

(extent and type of DOM removed) and the associated LPM fouling reduction. Our results and

436

discussion support the following main conclusions:

437

•

RI PT

433

MIEX pretreatment reduced the membrane fouling potential (quantified by the percent UMFI reduction) of all waters. However, the reduction in fouling potential varied

439

significantly among the different waters and did not correlate strongly with the extent of

440

removal of DOC, UVA254, or SUVA by MIEX, nor the initial DOC, UVA254, or SUVA

441

values in raw waters. •

M AN U

442

SC

438

Organic matter characterization using fluorescence EEMs showed that MIEX preferentially removed terrestrial DOM over microbial DOM. However, relative to their

444

respective abundance in feed waters, microbial DOM was disproportionately more

445

important to membrane fouling than terrestrial DOM.

446

•

TE D

443

Membrane fouling potential reduction was strongly correlated to the fluorescence intensity of microbial DOM at excitation/emission coordinates of 275 nm/340 nm via an

448

inverse linear function. Given that microbial contributions to the DOM pool increase with

449

increasing water residence time, MIEX is likely more effective to reduce the fouling

451

AC C

450

EP

447

potential of stream waters than of source waters with larger water residence times (e.g., lakes).

452

Based on the above conclusions, when ion exchange resin processes are used for DOM removal

453

towards membrane fouling reduction, it is advisable to use a resin that has been designed to

454

effectively remove microbial DOM.

20

ACCEPTED MANUSCRIPT

Acknowledgement

456

This work was supported by Ixom Watercare Inc. The Royal Thai Office of the Civil Service

457

Commission supported Panitan Jutaporn through a Royal Thai Government Scholarship under

458

the Ministry of Science and Technology, Thailand. The authors thank Katherine Harrold and

459

Alex Gorzalski for their assistance in fluorescence EEM spectroscopy analyses and data

460

processing, and Dr. Treavor Boyer and Dr. Miguel Arias for helpful discussions. We also thank

461

our collaborators at the Orange Water and Sewer Authority (Carrboro, NC), Muscle Shoals (AL)

462

Water Treatment Plant, and West Palm Beach (FL) Water Treatment Plant who facilitated and

463

helped with water sample collection.

464

Supplementary Material

465

Fluorescence spectroscopy analyses; Correlation between percent reduction of UMFI and percent

466

removal of UVA254, and SUVA in raw waters; Correlation between percent reduction of UMFI

467

and initial DOC concentration,UVA254, and SUVA levels in raw waters; Correlation between

468

UMFI and initial DOC concentration, UVA254, and SUVA levels in raw, MIEX2, MIEX10 and

469

diluted waters; Correlation between UMFI reduction by MIEX 2 and 10 mL/L and the

470

fluorescence intensity of components A and C in all waters tested.

471

References

472 473

Aoustin, E., Schafer, A. I., Fane, A. G., & Waite, T. D. (2001). Ultrafiltration of natural organic matter. Separation and Purification Technology, 22-23, 63–78.

474 475 476

Averett, R. C., Leenheer, J. A., McKnight, D. M., & Thorn, K. A. (1994). Humic Substances in the Suwannee River, Georgia; Interactions, Properties, and Proposed Structures. U.S. Geological Survey Water-Supply Paper, 2373.

477 478

Bolto, B., Dixon, D., Eldridge, R., King, S., & Linge, K. (2002). Removal of natural organic matter by ion exchange. Water Research, 36(20), 5057–65.

AC C

EP

TE D

M AN U

SC

RI PT

455

21

ACCEPTED MANUSCRIPT

Boyer, T. H., & Singer, P. C. (2005). Bench-scale testing of a magnetic ion exchange resin for removal of disinfection by-product precursors. Water Research, 39(7), 1265–76.

481 482 483

Boyer, T. H., & Singer, P. C. (2006). A pilot-scale evaluation of magnetic ion exchange treatment for removal of natural organic material and inorganic anions. Water Research, 40(15), 2865–76.

484 485

Boyer, T. H., & Singer, P. C. (2008). Stoichiometry of removal of natural organic matter by ion exchange. Environmental Science & Technology, 42(2), 608–13.

486 487 488

Boyer, T. H., Singer, P. C., & Aiken, G. R. (2008). Removal of dissolved organic matter by anion exchange: effect of dissolved organic matter properties. Environmental Science & Technology, 42(19), 7431–7.

489 490 491

Brown, A., McKnight, D. M., Chin, Y.-P., Roberts, E. C., & Uhle, M. (2004). Chemical characterization of dissolved organic material in Pony Lake, a saline coastal pond in Antarctica. Marine Chemistry, 89(1-4), 327–337.

492 493 494

Carroll, T., King, S., Gray, S. R., Bolto, B. A., & Booker, N. A. (2000). The fouling of microfiltration membranes by NOM after coagulation treatment. Water Research, 34(11), 2861–2868.

495

Cheryan, M. (1998). Ultrafiltration and Microfiltration Handbook. Lancaster, PA: CRC Press.

496 497 498

Cho, J., Amy, G., & Pellegrino, J. (2000). Membrane filtration of natural organic matter : factors and mechanisms affecting rejection and flux decline with charged ultrafiltration ( UF ) membrane. Journal of Membrane Science, 164, 89–110.

499 500

Coble, P. G. (1996). Characterization of marine and terrestrial DOM in seawater using excitation-emission matrix spectroscopy. Marine Chemistry, 51(4), 325–346.

501 502

Coble, P. G. (2014). Aquatic Organic Matter Fluorescence. New York: Cambridge University Press.

503 504 505

Coble, P. G. ., Green, S. A. ., Blough;, N. V., & Gagosian, R. B. (1990). Characterization of dissolved organic matter in the Black Sea by fluorescence spectroscopy. Nature, 348, 432 – 435.

506 507 508

Cory, R. M., & McKnight, D. M. (2005). Fluorescence spectroscopy reveals ubiquitous presence of oxidized and reduced quinones in dissolved organic matter. Environmental Science & Technology, 39(21), 8142–9.

509 510 511

Cory, R. M., McKnight, D. M., Chin, Y.-P., Miller, P., & Jaros, C. L. (2007). Chemical characteristics of fulvic acids from Arctic surface waters: Microbial contributions and photochemical transformations. Journal of Geophysical Research, 112, 1–14.

512 513 514

Cory, R. M., Miller, M. P., Mcknight, D. M., Guerard, J. J., & Miller, P. L. (2010). Effect of instrument-specific response on the analysis of fulvic acid fluorescence spectra. Limnology and Oceanography: Methods, 8, 67–78.

515 516 517

Croué, J.-P., Debroux, J. F., Gary, A., Aiken, G. R., & Leenheer, J. A. (1999). Natural organic matter: structural characteristics and reactive properties. In P. C. Singer (Ed.), Formation and Control of Disinfection By-Products in Drinking Water (pp. 65–93). Denver, CO:

AC C

EP

TE D

M AN U

SC

RI PT

479 480

22

ACCEPTED MANUSCRIPT

518

American Water Works Association. Dixon, M. B., Morran, J. Y., & Drikas, M. (2010). Extending membrane longevity by using MIEX as a pre-treatment. Journal of Water Supply: Research and Technology—AQUA, 59(2-3), 92–99.

522 523 524

Fabris, R., Lee, E. K., Chow, C. W. K., Chen, V., & Drikas, M. (2007). Pre-treatments to reduce fouling of low pressure micro-filtration (MF) membranes. Journal of Membrane Science, 289, 231–240.

525 526 527

Fan, L., Harris, J. L., Roddick, F. A., & Booker, N. A. (2001). Influence of the characteristics of natural organic matter on the fouling of microfiltration membranes. Water Research, 35(18), 4455–63.

528 529 530

Fan, L., Nguyen, T., Roddick, F. a., & Harris, J. L. (2008). Low-pressure membrane filtration of secondary effluent in water reuse: Pre-treatment for fouling reduction. Journal of Membrane Science, 320, 135–142.

531 532 533

Fearing, D. a, Banks, J., Guyetand, S., Monfort Eroles, C., Jefferson, B., Wilson, D., … Parsons, S. a S. a. (2004). Combination of ferric and MIEX for the treatment of a humic rich water. Water Research, 38, 2551–2558.

534 535 536

Henderson, R. K., Subhi, N., Antony, A., Khan, S. J., Murphy, K. R., Leslie, G. L., … Le-Clech, P. (2011). Evaluation of effluent organic matter fouling in ultrafiltration treatment using advanced organic characterisation techniques. Journal of Membrane Science, 382, 50–59.

537 538

Howe, K. J., & Clark, M. M. (2002). Fouling of microfiltration and ultrafiltration membranes by natural waters. Environmental Science & Technology, 36(16), 3571–6.

539 540 541

Huang, H., Cho, H.-H., Jacangelo, J. G., & Schwab, K. J. (2012). Mechanisms of Membrane Fouling Control by Integrated Magnetic Ion Exchange and Coagulation. Environmental Science & Technology, 46(19), 10711–10717.

542 543 544

Huang, H., Cho, H.-H., Schwab, K. J., & Jacangelo, J. G. (2012). Effects of magnetic ion exchange pretreatment on low pressure membrane filtration of natural surface water. Water Research, 46(17), 5483–90.

545 546

Huang, H., Schwab, K., & Jacangelo, J. G. (2009). Pretreatment for Low Pressure Membranes in Water Treatment : A Review. Environmental Science & Technology, 43(9), 3011–3019.

547 548 549

Huang, H., Young, T. A., & Jacangelo, J. G. (2008). Unified Membrane Fouling Index for Low Pressure Membrane Filtration of Natural Waters: Principles and Methodology. Environmental Science & Technology, 42(3), 714–720.

550 551 552

Huang, H., Young, T., & Jacangelo, J. G. (2009). Novel approach for the analysis of bench-scale, low pressure membrane fouling in water treatment. Journal of Membrane Science, 334(1-2), 1–8.

553 554 555

Humbert, H., Gallard, H., Jacquemet, V., & Croué, J.-P. (2007). Combination of coagulation and ion exchange for the reduction of UF fouling properties of a high DOC content surface water. Water Research, 41(17), 3803–11.

556

Humbert, H., Gallard, H., Suty, H., & Croué, J.-P. (2005). Performance of selected anion

AC C

EP

TE D

M AN U

SC

RI PT

519 520 521

23

ACCEPTED MANUSCRIPT

557 558

exchange resins for the treatment of a high DOC content surface water. Water Research, 39(9), 1699–708. Johnson, C. J., & Singer, P. C. (2004). Impact of a magnetic ion exchange resin on ozone demand and bromate formation during drinking water treatment. Water Research, 38(17), 3738–50.

562 563 564

Jones, K. L., & Melia, C. R. O. (2000). Protein and humic acid adsorption onto hydrophilic membrane surfaces : effects of pH and ionic strength. Journal of Membrane Science, 165, 31–46.

565 566 567

Kitis, M., İlker Harman, B., Yigit, N. O., Beyhan, M., Nguyen, H., & Adams, B. (2007). The removal of natural organic matter from selected Turkish source waters using magnetic ion exchange resin (MIEX®). Reactive and Functional Polymers, 67(12), 1495–1504.

568 569 570

Lee, N., Amy, G., & Croue, J.-P. (2006). Low-pressure membrane (MF/UF) fouling associated with allochthonous versus autochthonous natural organic matter. Water Research, 40(12), 2357–68.

571 572 573

Lee, N., Amy, G., Croue, J.-P., & Buisson, H. (2004). Identification and understanding of fouling in low-pressure membrane (MF/UF) filtration by natural organic matter (NOM). Water Research, 38(20), 4511–23.

574 575

Lee, N., Amy, G., & Lozier, J. (2005). Understanding natural organic matter fouling in lowpressure membrane filtration. Desalination, 178(1-3), 85–93.

576 577

Leenheer, J. A., & Croué, J.-P. (2003). Peer reviewed: Characterizing aquatic dissolved organic matter. Environmental Science & Technology, 1, 18A–26A.

578 579 580

McKnight, D. M., & Aiken, G. R. (1998). Sources and Age of Aquatic Humus. In D. Hessen & L. Tranvik (Eds.), Aquatic Humic Substances: Ecology and Biogeochemistry (pp. 9 – 40). New York: Springer.

581 582

McKnight, D. M., Andrews, E. D., Spaulding, S. A., & Aiken, G. R. (1994). Aquatic fulvic acids in algal-rich antarctic ponds. Limnol. Oceanogr., 39(8), 1972–1979.

583 584 585

McKnight, D. M., Boyer, E. W., Westerhoff, P. K., Doran, P. T., Kulbe, T., & Andersen, D. T. (2001). Spectrofluorometric characterization of dissolved organic matter for indication of precursor organic material and aromaticity. Limnology and Oceanography, 46(1), 38–48.

586 587

Mergen, M. R. D., Jefferson, B., Parsons, S. a, & Jarvis, P. (2008). Magnetic ion-exchange resin treatment: impact of water type and resin use. Water Research, 42, 1977–1988.

588 589 590

Myat, D. T., Mergen, M., Zhao, O., Stewart, M. B., Orbell, J. D., Merle, T., … Gray, S. (2013). Effect of IX dosing on polypropylene and PVDF membrane fouling control. Water Research, 47, 3827–3834.

591 592 593

Myat, D. T., Mergen, M., Zhao, O., Stewart, M. B., Orbell, J. D., Merle, T., … Gray, S. R. (2014). Membrane fouling mechanism transition in relation to feed water composition. Journal of Membrane Science, 471, 265–273.

594 595

Neale, P. a., & Schäfer, A. I. (2009). Magnetic ion exchange: Is there potential for international development? Desalination, 248, 160–168.

AC C

EP

TE D

M AN U

SC

RI PT

559 560 561

24

ACCEPTED MANUSCRIPT

Nghiem, L., Tadkaew, N., & Sivakumar, M. (2009). Removal of trace organic contaminants by submerged membrane bioreactors. Desalination, 236(1-3), 127–134.

598 599 600 601

Nguyen, S. T., Roddick, F. a, & Harris, J. L. (2010). Membrane foulants and fouling mechanisms in microfiltration and ultrafiltration of an activated sludge effluent. Water Science and Technology : A Journal of the International Association on Water Pollution Research, 62(9), 1975–83.

602 603

Schäfer, A. I., Fane, A. G., & Waite, T. D. (2000). Fouling effects on rejection in the membrane filtration of natural waters. Desalination, 131, 215–224.

604 605 606 607

Shao, S., Liang, H., Qu, F., Yu, H., Li, K., & Li, G. (2014). Fluorescent natural organic matter fractions responsible for ultrafiltration membrane fouling: Identification by adsorption pretreatment coupled with parallel factor analysis of excitation-emission matrices. Journal of Membrane Science, 464, 33–42.

608 609

Singer, P., Boyer, T., Holmquist, A., & Morran, J. (2009). Integrated analysis of NOM removal by magnetic ion exchange. American Water Works Association, 101:1(January), 65–73.

610 611

Singer, P. C., & Bilyk, K. (2002). Enhanced coagulation using a magnetic ion exchange resin. Water Research, 36, 4009–22.

612 613 614

Singer, P. C., Schneider, M., Edwards-Brandt, J., & Budd, G. C. (2007). MIEX for removal of DBF precursors: Pilot-plant findings. Journal / American Water Works Association, 99(4), 128–139.

615 616

Slunjski, M., Cadee, K., & Tattersall, J. (2000). MIEX ® Resin Water Treatment Process. In Aquatech Amsterdam. Amsterdam, Netherlands.

617 618 619

Stedmon, C., Markager, S., & Bro, R. (2003). Tracing dissolved organic matter in aquatic environments using a new approach to fluorescence spectroscopy. Marine Chemistry, 82(34), 239–254.

620 621 622

Weishaar, J. L., Aiken, G. R., Bergamashi, B. A., Fram, M. S., Fujii, R., & Mopper, K. (2003). Evaluation of specific ultra-violet absorbance as an indicator of the chemical content of dissolved organic carbon. Environmental Chemistry, 37, 4702–4708.

623 624 625

Wert, E. C., Edwards-Brandt, J. C., Singer, P. C., & Budd, G. C. (2005). Evaluating Magnetic Ion Exchange Resin (MIEX)® Pretreatment to Increase Ozone Disinfection and Reduce Bromate Formation. Ozone: Science & Engineering, 27, 371–379.

626 627 628

Yamamura, H., Kimura, K., & Watanabe, Y. (2007). Mechanism involved in the evolution of physically irreversible fouling in microfiltration and ultrafiltration membranes used for drinking water treatment. Environmental Science and Technology, 41(19), 6789–6794.

629 630

Yamashita, Y., & Tanoue, E. (2003). Chemical characterization of protein-like fluorophores in DOM in relation to aromatic amino acids. Marine Chemistry, 82, 255–271.

631 632

Yuan, W., & Zydney, A. L. (1999). Humic acid fouling during microfiltration. Journal of Membrane Science, 157(July 1998), 1–12.

633 634

Yuan, W., & Zydney, A. L. (2000). Humic Acid Fouling during Ultrafiltration. Environmental Science & Technology, 34(23), 5043–5050.

AC C

EP

TE D

M AN U

SC

RI PT

596 597

25

ACCEPTED MANUSCRIPT

Zhang, R., Vigneswaran, S., Ngo, H. H., & Nguyen, H. (2006). Magnetic ion exchange (MIEX®) resin as a pre-treatment to a submerged membrane system in the treatment of biologically treated wastewater. Desalination, 192, 296–302.

638 639 640

Zhang, R., Vigneswaran, S., Ngo, H., & Nguyen, H. (2007). A submerged membrane hybrid system coupled with magnetic ion exchange (MIEX®) and flocculation in wastewater treatment. Desalination, 216(1-3), 325–333.

641 642 643

Zhang, R., Vigneswaran, S., Ngo, H., & Nguyen, H. (2008). Fluidized bed magnetic ion exchange (MIEX®) as pre-treatment process for a submerged membrane reactor in wastewater treatment and reuse. Desalination, 227(1-3), 85–93.

644 645

Zularisam, A. W., Ismail, A. F., & Salim, R. (2006). Behaviours of natural organic matter in membrane filtration for surface water treatment — a review. Desalination, 194, 211–231.

SC

RI PT

635 636 637

AC C

EP

TE D

M AN U

646

26

ACCEPTED MANUSCRIPT

RI PT

Table 1. Water quality of raw waters.

UVA254 (cm-1)

DOC (mg/L)

SUVA (L/mg.m)

SR: Suwannee River (GA, USA) humic acid standard II

7.0

0.193

3.4

5.6

PL: Pony Lake (Antarctica) fulvic acid reference

6.8

0.167

CC: Cane Creek Reservoir, OWASA WTP* intake, Chapel Hill, NC, USA

6.3

6.3

(Dec 2013) 6.9

MS: Muscle Shoals WTP* intake, AL, USA

2.2

6.1

2.2

0.201

7.5

2.7

0.146

6.2

2.4

7.5

0.056

1.8

3.0

8.2

0.208

11

2.0

EP

PB: West Palm Beach WTP* intake, FL, USA

M AN U

(Nov 2012)

7.7

0.139

TE D

UL: University Lake, OWASA WTP* intake, Chapel Hill, NC, USA

SC

pH

Water source

AC C

*WTP: water treatment plant.

1

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

Figure 1. Schematic of ultrafiltration (UF) membrane system used for low-pressure membrane (LPM) fouling tests.

ACCEPTED MANUSCRIPT

100

(a)

RI PT

60

SR, DOC0 = 4.36 mg/L

40

PL, DOC0 = 7.73 mg/L

UL, DOC0 = 6.16 mg/L

SC

DOC removal (%)

80

20

MS, DOC0 = 1.84 mg/L

0 0 100

2 (b)

60

4

6

TE D

UVA254 removal (%)

80

M AN U

PB, DOC0 = 10.5 mg/L

40

10

12

-1

SR, UVA254,0 = 0.225 cm

-1

PL, UVA254,0 = 0.167 cm

-1

UL, UVA254,0 = 0.146 cm

-1

MS, UVA254,0 = 0.056 cm

EP

20

8

-1

PB, UVA254,0 = 0.208 cm

AC C

0

0

2

4

6

8

10

12

MIEX dose (mL/L)

Figure 2. (a) DOC and (b) UVA254 removal as a function of MIEX dose in batch tests. The initial DOC concentration (DOC0) and UVA254 (UVA254,0) in the raw waters are presented in the legends in panels (a) and (b), respectively. The water sources and water qualities of the raw waters are described in Table 1.

ACCEPTED MANUSCRIPT

1.0

(a)

0.8

J/J0

0.6

Raw, DOC = 1.8 mg/L MIEX2, DOC = 0.8 mg/L MIEX10, DOC = 0.7 mg/L

0.2 0.0 00:00

02:00

04:00

06:00

SC

2.4 2.2 2.0 1.8 1.6 1.4 1.2 1.0 0.8

J0/Jt = 0.0033VS +1 (b) Raw MIEX2 MIEX10 J0/Jt = 0.002VS + 1

M AN U

J0/Jt

Time (hh:mm)

J0/Jt = 0.0012VS +1

0

100

200 300 2 VS (L/m )

TE D

0.010 0.008

2

UMFI (m /L)

RI PT

0.4

400

Raw MIEX2 MIEX10 Diluted

500

(c)

0.006 0.004

EP

0.002 0.000

AC C

SR

PL

CC

UL

MS

PB

Water sources

Figure 3. (a) Illustrative normalized flux decline as a function of time for MS raw water (Raw), and MS raw water treated with MIEX doses of 2 mL/L (MIEX2) and 10 mL/L (MIEX10). (b) Illustrative fit of experimental results for J 0 / J t as a function of VS to Equation 1 for MS Raw (R2=0.97), MIEX2 (R2=0.99), and MIEX10 (R2=0.95) waters, where the slope of the linear regression corresponds to the unified membrane fouling index (UMFI). (c) UMFI for all source waters in fouling tests performed with Raw, MIEX2, MIEX10, and raw waters diluted with ultrapure water (Diluted) to approximately the same DOC concentration of MIEX2 waters. A MIEX dose of 10 mL/L was not tested for the CC water. Error bars correspond to the difference between the average of results obtained for duplicate samples and the result for either of the two samples. The water sources and water qualities of raw waters are described in Table 1.

1

ACCEPTED MANUSCRIPT

(a)

SR

40

MS

30 20

SC

CC PB

10

PL

UL

0 0 0.010

20 (b)

40 60 DOC removal (%)

80

CC

TE D

0.008 0.006

100

PB

UL

MS

0.004

EP

UMFI (m2/L)

RI PT

50

M AN U

UMFI reduction (%)

60

0.002

SR PL

AC C

0.000

0

2

4

6 8 DOC (mg/L)

10

12

Figure 4. (a) Percent UMFI reduction by 2 mL/L MIEX pretreatment as a function of percent DOC removal, and (b) UMFI as a function of DOC concentration in raw waters.

ACCEPTED MANUSCRIPT

600

(a)

1.0

400

C 300

T

2.5

0.5

0.0

A

400 500 Emission (nm)

300 (b)

A C T

2.0 1.5 1.0 0.5 0.0 PL

CC UL Water sources

TE D

SR

120

600

M AN U

Fluorescent intensity (RU)

RI PT

1.5

MS

PB A C T

(c)

100 80

EP

Removal (%)

Raman unit

500

SC

Excitation (nm)

2.0

60

AC C

40 20 0

SR

PL

CC

UL

MS

PB

Water sources

Figure 5. (a) Fluorescence excitation-emission matrix (EEM) of raw UL water. Components A and C correspond to terrestrial DOM, and component T corresponds to microbial DOM. (b) Fluorescence intensity (Raman units) of components A, C and T for all raw waters. (c) Removal of components A, C, and T by MIEX 2 mL/L pretreatment from all raw waters. Error bars correspond to the standard deviation of results obtained in triplicate tests.

(b)

AC C

2

0.0

MI EX

2

PB

PB

Water sources

MS

2

UL

MS MI EX

CC

MI EX

PL

0.1

2

SR

EP

0.0

0.2

2

0.1

600

0.3

CC MI EX

0.2

0.0

A

A C T

0.4

2

TE D

0.3

(d)

MI EX

0.4

0.2

400 500 Emission (nm)

300

0.5

A C T

(c)

600

MI EX

400 500 Emission (nm)

PL

A

0.6

0.4

C

T

0.0

SR

0.5

300

T 300

400

M AN U

C 300

Fluorescent intensity (RU)

0.4

500

SC

400

Raman unit

0.8

Excitation (nm)

500

Fluorescent intensity (RU)

Excitation (nm)

1.2

Raman unit

600

(a)

UL

600

RI PT

ACCEPTED MANUSCRIPT

Water sources

Figure 6. (a, b) Fluorescence excitation-emission matrices (EEMs) of the backwash waters of membranes fouled with (a) raw UL water and (b) UL water pretreated with a MIEX dose of 2 mL/L (MIEX2). Components A and C correspond to terrestrial DOM, and component T corresponds to microbial DOM. (c, d) Fluorescence intensity of components A, C, and T for the backwash waters of membranes fouled with (c) raw and (d) MIEX2 waters for all source waters. Error bars correspond to the difference between the average of samples from duplicate filtration tests and the value of either of the samples.

RI PT

ACCEPTED MANUSCRIPT

MIEX2 MIEX10

80 60

R2 =0.79

SR MS

40

MS

R2 = 0.95

20

SC

SR

PL

M AN U

% UMFI Reduction

100

PB

CC

PL

PB

UL UL

EP

TE D

0 0.0 0.1 0.2 0.3 0.4 0.5 Fluorescence intensity of component T (RU)

AC C

Figure 7. Low-pressure membrane fouling potential reduction by MIEX 2 and 10 mL/L pretreatment as a function of the fluorescence intensity of component T (microbial DOM). The figure contains data for all six water sources treated with 2 mL/L MIEX dose (MIEX2) and 10 mL/L MIEX dose (MIEX10), except CC+MIEX10 experiment due to limited water sample availability.

ACCEPTED MANUSCRIPT

Highlights

EP

TE D

M AN U

SC

RI PT

PVDF membrane fouling reduction by MIEX varied significantly among different waters Both terrestrial and microbial DOM fouled PVDF ultrafiltration membranes Microbial DOM was more important to membrane fouling than terrestrial DOM MIEX preferentially removed terrestrial DOM over microbial DOM Fouling reduction correlated well with microbial DOM via an inverse linear function

AC C

• • • • •