Stoichiometry of Removal of NaturalOrganic Matter by Ion ExchangeT R E A V O R H . B O Y E R * A N DP H I L I P C . S I N G E R

Department of Environmental Sciences and Engineering,CB 7431, University of North Carolina at Chapel Hill,Chapel Hill, North Carolina 27599-7431

Received August 3, 2007. Revised manuscript received October17, 2007. Accepted November 1, 2007.

Five anion exchange resins, including a magnetic ion exchange(MIEX) resin, were evaluated for removal of SuwanneeRiver Fulvic Acid (SRFA) in the presence of bicarbonate andchloride. The charge density of SRFA, obtained by potentiometrictitration, was used to perform charge balances for ionexchange reactions involving SRFA, bicarbonate, and chlorideunder different solution conditions. The results clearly showthe equivalence of SRFA uptake and chloride release by ionexchange. Although the structure of the anion exchange resinsdid not affect the stoichiometry of the reaction, the polyacrylicresins did exhibit greater removal of SRFA than the polystyreneresins. The hindered removal of SRFA by the polystyrene resinswas hypothesized to be a result of size exclusion. The MIEXresin, which has a polyacrylic structure, performed similarly tothe other polyacrylic resins. For the MIEX resin, the separationfactor for SRFA over chloride was ∼8 times greater than forbicarbonate over chloride. This work provides an improvedunderstanding of the interactions between natural organic matter(NOM), inorganic anions, and anion exchange resins, andshould result in more effective applications of ion exchangefor the removal of NOM in the treatment of drinking water.

IntroductionIt is well-known that during drinking water treatment, naturalorganic matter (NOM) reacts with chlorine and otherdisinfectants to form halogenated organic disinfectionbyproducts (DBPs), e.g., trihalomethanes (THMs), haloaceticacids (HAAs), haloacetonitriles, haloketones, haloni-tromethanes, and haloaldehydes (1). Both short-term andlong-term health risks have been associated with exposureto tap water that contains these and other DBPs, e.g., a recentstudy by Villanueva and co-workers (2) provided strongevidence that long-term exposure to THMs increases therisk for bladder cancer. Consequently, removal of NOM priorto disinfection is a common strategy for minimizing theformation of DBPs. Treatment processes used to removeNOM include enhanced coagulation, granular activatedcarbon adsorption, nanofiltration, and ion exchange (3–5).To quantify the effectiveness of these treatment processes,dissolved organic carbon (DOC) is commonly used as asurrogate for NOM. Both conventional polymeric anionexchange resins (6–11) and a magnetic ion exchange (MIEX)resin (12–20) have previously been evaluated for removingDOC and controlling the formation of DBPs. The majordifference between these two classes of resins is their mode

of treatment, i.e., conventional polymeric resins are used ina fixed-bed to treat filtered water, whereas MIEX resin isused in a completely mixed flow reactor to treat raw water.Important properties of anion exchange resins includepolymer composition, porosity, and charged functionalgroups. The composition of most resins is either polystyreneor polyacrylic. Polystyrene resins are more hydrophobic thanpolyacrylic resins; as a result, polyacrylic resins tend to havea more open structure and higher water content. The porosityof resins is defined as either macroporous or gel. Macroporousresins are highly porous solids, while gel resins do not containany pores (9). Strong-base resins typically contain either typeI (-N+(CH3)3) or type II (-N+(CH3)2(C2H4OH)) quaternaryammonium functional groups. Due to their ethanolic content,type II resins are more hydrophilic than type I resins (21).Most strong-base anion exchange resins are used in thechloride form.

Among ion exchange resins, polyacrylic, macroporous,strong-base anion exchange resins have been shown to bemost effective for DOC removal (6, 7, 9). In addition, previouswork by the authors, consisting of laboratory-scale batchexperiments and pilot-scale continuous-flow studies, showedthat treatment of raw drinking water with MIEX resin wasmore effective than coagulation at removing DOC andreducing the formation of THMs and HAAs (13, 15). Alimitation of many previous ion exchange studies, however,was that the experiments were conducted using naturalwaters that were not well characterized. As a result, it wasdifficult to quantitatively assess the impacts of raw watercomposition, NOM characteristics, and anion exchange resinproperties on DOC removal. In particular, the mechanismof DOC uptake by anion exchange resins, i.e., ion exchangeversus adsorption, has not been conclusively demonstrated.For example, Fu and Symons (9) concluded that DOC wasremoved via ion exchange, yet others have shown evidenceof both ion exchange and adsorption (8). The mechanism ofDOC removal is important because it influences all aspectsof ion exchange treatment.

Accordingly, the objective of this research was to elucidatethe interactions between NOM, inorganic anions, and anionexchange resins. The ternary ion exchange reaction of interestwas the exchange of NOM and bicarbonate for chloride. Thespecific aims of this work were to quantify the contributionsof ion exchange and adsorption to DOC removal and toexamine the impact of resin structure on DOC uptake. Basedupon previous work, the initial hypothesis was that DOC isremoved by ion exchange.

Experimental SectionModel Waters. NOM-containing model waters were preparedby adding Suwannee River Fulvic Acid (SRFA; referencesample from International Humic Substances Society (22)),sodium bicarbonate (used as a pH buffer), and sodiumchloride to laboratory-grade water (LGW). Control (i.e., NOM-free) model waters were prepared by adding sodium bicar-bonate and sodium chloride to LGW. The pH of the modelwaters was adjusted using either 0.1 N NaOH or 0.1 N HClsolutions. Table 1 shows the composition of the model watersused in this work. Most experiments were conducted usingSRFA I and Control I model waters. The additional modelwaters were used to investigate the impact of higher chlorideconcentrations (SRFA II and Control II) and the absence ofbicarbonate (SRFA III). The DOC concentration in mil-liequivalents (meq) per liter was calculated as the productof the DOC concentration in milligrams carbon per liter andthe charge density of SRFA in milliequivalents per gram

* Corresponding author phone: 919-966-1177; e-mail: tboyer@email.unc.edu.

Environ. Sci. Technol. 2008, 42, 608–613

608 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 2, 2008 10.1021/es071940n CCC: $40.75 2008 American Chemical SocietyPublished on Web 12/07/2007

carbon that was determined following published proceduresfor the direct potentiometric titration of NOM (23, 24). Briefly,a solution containing 300 mg/L NOM and 0.1 M KCl waspurged with nitrogen gas for 30 min to eliminate carbondioxide and then titrated with 0.04 N NaOH solution undera nitrogen atmosphere. The NaOH solution was added in 0.1mL increments; the pH was recorded after each addition oftitrant. All samples were titrated up to approximately pH 11.The charge density of SRFA was calculated based on the pHmeasurements and a charge balance of the solution as follows:

Charge density (meq/g C) )[H+]+ [Na+]- [OH-]

CNOM(1)

where the concentrations of H+, Na+, and OH- are in meq/Land CNOM is the NOM concentration of the solution (g C/L)(24). Potassium and chloride were not included in the chargebalance since they were opposite in charge and equal inconcentration.

Ion Exchange Resins. Five fresh anion exchange resins,from three different manufacturers, were used in this work:Amberlite IRA910 and Amberlite IRA958 from Rohm and HaasCo., Ionac A-641 and Ionac Macro-T from Sybron ChemicalsInc., and MIEX resin from Orica Watercare. All resins had amacroporous structure, strong-base functional groups, andwere used in the chloride-form. IRA910 and A-641 arepolystyrene resins, while MIEX, Macro-T, and IRA958 arepolyacrylic resins. The MIEX resin has magnetic iron oxideincorporated into its polymer structure to aid in separation;the other resins are conventional polymeric ion exchangeresins. The ion exchange resins were cleaned and preparedfollowing published procedures (7, 10); the MIEX resin didnot require any preparation. All ion exchange resins werestored in LGW. To accurately dose small quantities of resin(∼10 mg), the resins were filtered on a Whatman GF/B filterusing a vacuum filter apparatus and then weighed; thisprocedure was similar to one devised by Boening et al. (25).The apparent resin density was defined as the mass of filteredresin per volume of wet settled resin. The volumetric ionexchange capacity (meq/mL wet settled resin) and themoisture content were determined following manufacturers’recommendations and published procedures (26). Illustrativeproperties of the five anion exchange resins used in this workare shown in Table 2; the experimentally determined resinproperties agree well with the manufacturers’ specifications.

Ion Exchange Isotherm Experiments. Bottle-point iso-therms were conducted to investigate ion exchange equi-

librium. It was determined, for both the polyacrylic andpolystyrene resins, that two days of mixing was sufficient toreach equilibrium. All experiments were conducted at anambient laboratory temperature of 22 °C. Briefly, varyingamounts of ion-exchange resin (5–50 mg) were added tosample bottles containing 100 mL of model water and mixedend-over-end at 14 rpm. A majority of the ion exchangeexperiments were conducted using duplicate resin doses.Prior to chemical analysis, all equilibrated samples werefiltered through 0.45 µm Supor-450 membrane filters (pre-rinsed with 500 mL of LGW). As a quality control check, therewas no observed increase in chloride or DOC after equili-brating the resins for two days in LGW. For data analysis,aqueous-phase concentrations were expressed in milliequiv-alents per liter, and resin-phase concentrations were cal-culated as the difference in aqueous-phase concentrationsbefore and after ion exchange equilibrium. The equivalentionic fraction of DOC in solution was calculated by dividingthe aqueous-phase concentration of DOC by the totalaqueous-phase concentration of DOC, bicarbonate, andchloride (i.e., CT); the equivalent ionic fraction of DOC in theresin was calculated by dividing the resin-phase concentra-tion of DOC by the dose of ion-exchange resin. Equivalentionic fractions for bicarbonate and chloride were calculatedin the same manner.

Analytical Methods. An Accumet AB15 pH meter with acalomel combination pH electrode was used to measure pH.Ultraviolet (UV) absorbance was measured on either a HitachiU-2000 or U-3300 spectrophotometer using a 1 cm quartzcell. DOC and dissolved inorganic carbon (DIC) weremeasured on a Shimadzu TOC-VCPH Total Organic CarbonAnalyzer. All DOC and DIC samples were measured induplicate with average values reported; the relative difference(defined as the absolute difference divided by the mean)between DOC and DIC duplicates was <10% and <5%,respectively. The bicarbonate concentration was calculatedby correcting the DIC concentration for temperature andpH (27). Chloride was measured according to U.S. Environ-mental Protection Agency Method 300.0 (28) on a DionexIon Chromatograph model AI-450v.3.32 equipped with anIonPac AG4A-SC guard column and an AS4A-SC analyticalcolumn. The relative difference between chloride duplicateswas <10%.

Results and DiscussionIon Exchange Stoichiometry. The charge density of SRFA asa function of pH was determined by direct potentiometric

TABLE 1. Composition of Model Waters

model water pH DOC (mg C/L) DOC (meq/L) bicarbonate (meq/L) chloride (meq/L) CTa (meq/L)

SRFA I 7.70–7.83 9.2–9.9 0.11–0.12 0.43–0.45 0.009–0.038 0.56–0.61SRFA II 7.49 9.7 0.11 0.43 0.12 0.67SRFA III 6.56 11.0 0.12 0 0.097 0.21Control I 7.88–8.04 0 0 0.47–0.48 0.008–0.016 0.48–0.49Control II 7.50 0 0 0.45 0.092 0.54

a CT ) DOC + bicarbonate + chloride.

TABLE 2. Properties of Ion Exchange Resins

resin structurea capacityb (meq/mL) water contentb (% (m/m))

IRA910 polystyrene, type II 1.0 (g1.0) 43 ( 6 (54–61)A-641 polystyrene, type I 0.93 ( 0.02 (1.0) 58 ( 2 (56–64)MIEX polyacrylic, type I 0.52 ( 0.02 65 ( 1Macro-T polyacrylic, type I 0.65 ( 0.01 (0.85) 68 ( 1 (66–72)IRA958 polyacrylic, type I 0.86 ( 0.02 (g0.8) 69 ( 1 (66–72)

a Macroporous, chloride-form. b Average ( one standard deviation (manufacturer’s data).

VOL. 42, NO. 2, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 609

titration in a carbon-dioxide-free solution. The titration ofSRFA was done in duplicate; the agreement between the twotitration curves was very good, as shown in Figure 1. Theduplicate titration curves were averaged using an interpola-tion routine in MATLAB. The carboxyl acidity (defined as thecharge density at pH 8) of SRFA was 11.8 ( 0.1 meq/g C. Thisvalue is in good agreement with published values for SRFA(e.g., Ritchie and Perdue (23) reported a carboxyl acidity of12.2 meq/g C for SRFA). The charge density of SRFA wasused to calculate the DOC concentration on an equivalentcharge basis and to quantitatively examine the hypothesisthat DOC was removed via ion exchange.

The ion exchange reaction of interest is the equilibrationof an anion exchange resin in the chloride form with a solutioncontaining DOC, bicarbonate, and chloride. The chargebalance equation for this reaction is written as follows:

Cl-+DOC+HCO3-+Cl-)

DOC+HCO3-+Cl-+DOC+HCO3

-+Cl- (2)

where overbars represent resin-phase species in meq/g offiltered resin and all aqueous-phase concentrations are inmeq/L. At equilibrium, for a purely ion exchange reaction,the decrease in the aqueous-phase concentrations of DOCand bicarbonate must equal the increase in the aqueous-phase concentration of chloride. Figure 2 shows the resultsof multiple batch equilibrium experiments using MIEX resinand the model waters shown in Table 1. The y-axis shows thenet increase in the aqueous concentration of chloride (i.e.,equilibrium chloride concentration minus initial chlorideconcentration), while the x-axis shows the net decrease inthe aqueous concentrations of DOC and bicarbonate to-gether. For duplicate samples, both the mean increase inchloride and the mean decrease in DOC and bicarbonatetogether had a coefficient of variation (CV; defined as thestandard deviation divided by the mean) of <10%. Due tothe substantial removal of DOC and bicarbonate, the pH ofthe equilibrated samples decreased somewhat as the resindose increased. The observed decreases in pH were as follows:0.02–0.44 (SRFA I and II), 0.2–1 (SRFA III), and 0.04–0.5(Control I and II). To account for the change in pH of theequilibrated samples, the DOC concentration (meq/L) wasadjusted using the measured DOC concentration (mg C/L)and the charge density of SRFA at the appropriate pH; thebicarbonate concentration was adjusted based on themeasured DIC concentration and pH. The variations in pH

did not affect the capacity of the ion exchange resins since allresins had strong-base functional groups. The legend in Figure2 indicates the chemical species taking part in the ion exchangereaction, e.g., HCO3

-/Cl- indicates bicarbonate-chloride ex-change. The y ) x line in Figure 2 represents the stoichio-metric exchange of chloride on the resin for bicarbonate andDOC in the solution. The bicarbonate-chloride exchangeexperiments in the absence of DOC were conducted as acontrol since it was expected that bicarbonate would beremoved exclusively via ion exchange. Figure 2 shows thatall of the data points for HCO3

-/Cl- are clustered about they ) x line and verifies that bicarbonate was indeed removedvia ion exchange; the relative difference between chloriderelease and bicarbonate uptake was <18%. Of particularinterest are the data points for the DOC/HCO3

-/Cl- experi-ments and the DOC/Cl- experiments. For both systems, thepoints are also tightly clustered about the y ) x line; therelative difference between chloride release and DOC/bicarbonate uptake was <15%. The results of the DOCexchange experiments mirrored the results of the controlexperiments and clearly show that, at approximately pH 8,SRFA was removed entirely by ion exchange. The range ofDOC removal observed in these experiments was ∼5–9 mgC/L (0.06–0.1 meq/L); bicarbonate removal was ∼0.01–0.2meq/L, indicating that the changes in DOC were comparableto those of bicarbonate.

To probe the relative contributions of ion exchange andadsorption to DOC removal, an ion exchange experimentwas conducted using SRFA I model water and a weak-acid,magnetic ion exchange (WA-MIEX) resin provided by OricaWatercare. The WA-MIEX resin was similar in compositionto the MIEX resin described in Table 2, but it had no anionexchange properties because of its weak-acid cation exchangefunctional groups. As a result, any removal of DOC couldonly occur by adsorption. After two days of mixing, no DOCor bicarbonate removal by WA-MIEX was observed (see TableS1 in the Supporting Information (SI)). These results confirmthe results illustrated in Figure 2 that removal of SRFAoccurred exclusively via ion exchange.

Additional ion exchange experiments were conducted withthe other resins using SRFA I and Control I model waters.Figure 3 shows the results of these experiments in a fashionanalogous to that of Figure 2. As discussed above, the DOC

FIGURE 1. Charge density of Suwannee River Fulvic Acidresulting from direct potentiometric titration.

FIGURE 2. Illustrative ion exchange stoichiometry for MIEXresin. Corresponding model waters are as follows: Control Iand II (HCO3

-/Cl-), SRFA I and II (DOC/HCO3-/Cl-), and SRFA III

(DOC/Cl-).

610 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 2, 2008

and bicarbonate concentrations were adjusted based on theequilibrium pH. For duplicate samples, both the meanincrease in chloride and the mean decrease in bicarbonateand DOC had a CV of <14%. The ion exchange resins areidentified in the legend; solid symbols represent bicarbon-ate-chloride exchange in the NOM-free (control) solution,while open symbols represent DOC-bicarbonate-chlorideexchange. As expected, the bicarbonate-chloride exchangedata for IRA910 and A-641 illustrate equivalent exchange ofchloride for bicarbonate; the relative difference betweenchloride release and bicarbonate uptake was <6%. All of thedata for the DOC-bicarbonate-chloride system in Figure 3are tightly clustered about the y)x line (the relative differencebetween chloride release and DOC/bicarbonate uptake was<11%), and illustrate the stoichiometric exchange of DOCand bicarbonate for chloride. Both the polystyrene resins(IRA910 and A-641) and the polyacrylic resins (MIEX, Macro-T,andIRA958)exhibitedthesameionexchangestoichiometry.

To ensure that the results shown in Figures 2 and 3 werenot dictated solely by bicarbonate removal, the ratio of DOCremoval to total DOC/bicarbonate removal was calculated(i.e., ∆DOC/(∆DOC + ∆HCO3

-)). Table S2 in the SI showsthese fractional DOC removals for the data in Figure 3. Therange of fractional DOC uptakes for the polystyrene andpolyacrylic resins were 0.14–0.29 and 0.33–0.84, respectively.(The finding that the polyacrylic resins have a higher fractionaluptake of DOC than the polystyrene resins will be discussedin the next section.) The data in Table S2 indicate that DOCuptake accounted for a substantial fraction of the materialbalance, and that the ion exchange stoichiometry shown inFigures 2 and 3 required DOC removal for closure. The resultsin this section quantitatively confirm the stoichiometricremoval of SRFA by ion exchange.

Impact of Resin Structure. In previous ion exchangestudies, researchers often compared ion exchange resins onan equal volume or mass basis (7, 8, 10). Using this approach,however, it is difficult to separate the influences of resinstructure (i.e., polymer composition, water content, andporosity) and ion exchange capacity on solute removal.Therefore, to compare the resins on an equivalent basis andto isolate the impact of resin structure, resin doses wereconverted to milliequivalents of exchange capacity per liter

of solution using the apparent density and volumetric ionexchange capacity of each resin. Figure 4a and b show therelative concentrations of DOC and bicarbonate, respectively,as a function of resin dose. Removal of DOC was evaluatedusing the SRFA I model water, while removal of bicarbonatewas evaluated using the Control I model water. As will bediscussed in the subsequent section, the presence ofbicarbonate did not adversely affect removal of DOC. Forduplicate samples, resin doses had a CV of 3–14% (thevariation in the resin dose was primarily due to experimentaluncertainty in the apparent density), DOC removal had a CVof <12%, and bicarbonate removal had a CV of <3%.

Figure 4a illustrates two important features: (1) the threepolyacrylic resins (MIEX, Macro-T, and IRA958) all exhibitedsimilar removal of DOC, and (2) the polyacrylic resinsexhibited much greater removal of DOC than the polystyreneresins (IRA910 and A-641). It is noteworthy that the mag-netically enhanced polyacrylic resin (MIEX) performedsimilarly to the two conventional polyacrylic resins (IRA958and Macro-T), since there have been limited comparisonsof these two types of resins. When considered on anequivalent capacity basis, data from Humbert et al. (10)showed that MIEX resin performed similarly to a macroporouspolystyrene resin with a water content of 73%. Figure 4a andTable 2 show that the polyacrylic resins exhibited greater

FIGURE 3. Illustrative ion exchange stoichiometry for fiveanion exchange resins. Solid symbols represent bicar-bonate-chloride exchange (Control I model water); opensymbols represent DOC-bicarbonate-chloride exchange (SRFAI model water).

FIGURE 4. Impact of resin structure on (a) DOC removal (SRFAI model water) and (b) bicarbonate removal (Control I modelwater). All concentrations are normalized by their respectiveinitial model water concentration (C0).

VOL. 42, NO. 2, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 611

removal of DOC and had higher water contents (65–69%versus 43–58%) than the polystyrene resins. It is often difficult,however, to distinguish between the effects of polymercomposition, water content, and porosity. For example, Boltoet al. (7) observed a direct relationship between DOC removaland water content, without reporting the pore size of theresins, and concluded that water content, in general, was agood predictor of resin performance. Although IRA910 (typeII resin) had a lower water content than A-641 (type I resin),it is somewhat surprising that IRA910 removed less DOCthan A-641, since strong-base anion exchange resins withtype II functional groups have been previously shown toremove more NOM than would be expected based on theirwater content (7). In addition, other researchers haveattributed poor removal of DOC by polystyrene resins to sizeexclusion (8, 9, 29). The specific UV absorbance (SUVA) wascalculated for the data in Figure 4a to assess the possibilityof size exclusion. SUVA (L/mg ·m) is defined as the ratio ofthe UV absorbance at 254 nm (UV254 (1/cm)) to the DOCconcentration (mg C/L), and is reported to be directlycorrelated with the aromatic carbon content and molecularweight of NOM (30, 31). The analysis showed that the SUVAof the water decreased following treatment with the poly-acrylic resins, whereas the SUVA increased following treat-ment with the polystyrene resins (see Figure S1 in the SI).These findings demonstrate that the polyacrylic resinspreferentially removed UV254-absorbing organic compounds,and suggest that NOM with a higher molecular weight and/or aromatic carbon content was excluded by the polystyreneresins.

Bicarbonate removal in the DOC-free solution was usedto further investigate the influences of polymer composition,water content, and porosity on performance of the resins. Itwas assumed that size exclusion could be neglected whenconsidering bicarbonate-chloride exchange. Figure 4b showsthe relative bicarbonate concentration as a function of theequivalent resin dose, and illustrates that both MIEX resinand the polystyrene resins (IRA910 and A-641) removedbicarbonate to a similar degree. This demonstrates that fora small inorganic anion such as bicarbonate, the driving forcefor ion exchange is the capacity of the anion exchange resin,and that resin structure has a negligible effect. Table S3 inthe SI shows a comparison of the percent removal of DOCand bicarbonate by IRA910, A-641, and MIEX. It is interestingthat IRA910 exhibits 11% lower removal of DOC thanbicarbonate, whereas MIEX resin shows 51% greater removalof DOC than bicarbonate. Since polystyrene resins have beenshown to have a higher selectivity for hydrophobic organicanions than inorganic anions (32), the lower removal of DOCthan bicarbonate, combined with other results presented inthis section, suggest that the poor removal of DOC by thepolystyrene resins is a consequence of specific interactionsbetween the resin structure and NOM, e.g., size exclusion.It is expected that, as the ionic strength of a solution increases,size exclusion of DOC by polystyrene resins would becomeless important due to contraction of NOM molecules (8).

Isotherms and Separation Factors. Ion exchange iso-therms for SRFA and bicarbonate on MIEX resin werecompared to elucidate the SRFA-bicarbonate-chloridesystem. Figure 5 shows the equivalent ionic fraction of anionA in the resin (yA) as a function of its equivalent ionic fractionin solution (xA). The solid circles and triangles represent binaryisotherms for SRFA-chloride and bicarbonate-chlorideexchange, respectively; the open circles show the isothermfor SRFA in the presence of chloride and bicarbonate.Although CT varied by a factor of ∼2.5 between the binaryand ternary systems, the DOC concentrations were ap-proximately equal for both systems. The diagonal lineillustrates the hypothetical scenario in which MIEX resin hasequal preference for both generic anion A and chloride. In

addition, isotherms that fall above the diagonal illustratefavorable exchange, while isotherms that fall below thediagonal illustrate unfavorable exchange. Considering onlythe binary isotherms (i.e., solid symbols), Figure 5 showsthat MIEX resin favors SRFA over chloride while, at best, theresin has approximately equal preference for bicarbonateand chloride.

Separation factors were calculated for the binary isothermsto quantify the affinity of MIEX resin for SRFA and bicarbon-ate. The separation factor for anion A over chloride is definedas RA/Cl ) (yAxCl)/(xAyCl), where A can represent either SRFAor bicarbonate. If RA/Cl is >1, then anion A is preferred by theresin. At yA ) 0.5, the separation factors for SRFA-chlorideand bicarbonate-chloride were 5.7 and 0.76, respectively.In addition, Figure S2 in the SI illustrates that the separationfactors for SRFA and bicarbonate are relatively constant withrespect to resin loading. To the authors’ knowledge, theseare the first published separation factors for MIEX resin.Figure 5 shows that when the total ionic strength of the modelwater is increased by the addition of sodium bicarbonate,the affinity of MIEX resin for SRFA increases (see solid andopen circles). Semmens and Gregory (33) reported a similarrelationship between selectivity and ionic strength forcarboxylate uptake by a strong-base anion exchange resin.Thus, the removal of SRFA is not adversely affected by thepresence of bicarbonate; rather, it appears to be slightlypromoted.

This work illustrates the stoichiometric removal of SRFAvia ion exchange by five different anion exchange resins.While resin structure did not affect the reaction stoichiometry,the polyacrylic resins exhibited greater removal of SRFA thanthe polystyrene resins. This is hypothesized to be a result ofsize exclusion by the polystyrene resins. In addition, MIEXresin had a greater affinity for SRFA over bicarbonate andchloride. Although not investigated as part of this research,it would be of interest to know how changes in NOMcharacteristics, e.g., charge density, aromaticity, and mo-lecular weight, would affect NOM uptake via ion exchange,and how inorganic anions, such as sulfate, would competewith NOM for ion exchange sites. (This is the subject of afollow-up study.) A selectivity sequence for well-characterizedNOM extracts and strongly competing inorganic anions

FIGURE 5. Illustrative ion exchange isotherms for MIEX resin.Open circles show DOC isotherm in the presence ofbicarbonate and chloride (SRFA I); solid circles show DOCisotherm in the presence of only chloride (SRFA III); and solidtriangles show bicarbonate isotherm in the presence ofchloride (no DOC, Control I).

612 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 2, 2008

potentially could be developed. Such a study would furtherelucidate the behavior of NOM with respect to its removalby ion exchange and ultimately improve the efficacy of ionexchange treatment for removal of NOM.

AcknowledgmentsThe support of T.H.B. by the U.S. Environmental ProtectionAgency (EPA) under the Science to Achieve Results (STAR)Graduate Fellowship Program (FP-91678201-1), and by theAmerican Water Works Association (AWWA) through an AbelWolman Doctoral Fellowship is gratefully acknowledged. EPAand AWWA have not officially endorsed this publication andthe views expressed herein may not reflect the views of EPAor AWWA.

Supporting Information AvailableTabulated data for WA-MIEX ion exchange experiments;tabulated data for the fractional removal of DOC; tabulateddata for the impact of resin structure on removal of DOC andbicarbonate; analysis of the impact of resin structure ontreated water SUVA; and analysis of separation factors as afunction of resin loading. This information is available freeof charge via the Internet at http://pubs.acs.org.

Literature Cited(1) Krasner, S. W.; Weinberg, H. S.; Richardson, S. D.; Pastor, S. J.;

Chinn, R.; Sclimenti, M. J.; Onstad, G. D.; Thruston, A. D.Occurrence of a new generation of disinfection byproducts.Environ. Sci. Technol. 2006, 40 (23), 7175–7185.

(2) Villanueva, C. M.; Cantor, K. P.; Grimalt, J. O.; Malats, N.;Silverman, D.; Tardon, A.; Garcia-Closas, R.; Serra, C.; Carrato,A.; Castano-Vinyals, G.; Marcos, R.; Rothman, N.; Real, F. X.;Dosemeci, M.; Kogevinas, M. Bladder cancer and exposure towater disinfection by-products through ingestion, bathing,showering, and swimming in pools. Am. J. Epidemiol. 2007, 165(2), 148–156.

(3) Bolto, B.; Dixon, D.; Eldridge, R.; King, S. Removal of THMprecursors by coagulation or ion exchange. Water Res. 2002, 36(20), 5066–5073.

(4) Siddiqui, M.; Amy, G.; Ryan, J.; Odem, W. Membranes for thecontrol of natural organic matter from surface waters. WaterRes. 2000, 34 (13), 3355–3370.

(5) Jacangelo, J. G.; Demarco, J.; Owen, D. M.; Randtke, S. J. Selectedprocesses for removing NOM–an overview. J. Am. Water WorksAssoc. 1995, 87 (1), 64–77.

(6) Anderson, C. T.; Maier, W. J. Trace organics removal by anionexchange resins. J. Am. Water Works Assoc. 1979, 71 (5), 278–283.

(7) Bolto, B.; Dixon, D.; Eldridge, R.; King, S.; Linge, K. Removal ofnatural organic matter by ion exchange. Water Res. 2002, 36(20), 5057–5065.

(8) Croue, J. P.; Violleau, D.; Bodaire, C.; Legube, B. Removal ofhydrophobic and hydrophilic constituents by anion exchangeresin. Water Sci. Technol. 1999, 40 (9), 207–214.

(9) Fu, P. L. K.; Symons, J. M. Removing aquatic organic substancesby anion exchange resins. J. Am. Water Works Assoc. 1990, 82(10), 70–77.

(10) Humbert, H.; Gallard, H.; Suty, H.; Croue, J. P. Performance ofselected anion exchange resins for the treatment of a high DOCcontent surface water. Water Res. 2005, 39 (9), 1699–1708.

(11) Kim, P. H. S.; Symons, J. M. Using anion exchange resins toremove THM precursors. J. Am. Water Works Assoc. 1991, 83(12), 61–68.

(12) Singer, P. C.; Bilyk, K. Enhanced coagulation using a magneticion exchange resin. Water Res. 2002, 36, 4009–4022.

(13) Boyer, T. H.; Singer, P. C. Bench-scale testing of a magnetic ionexchange resin for removal of disinfection by-product precur-sors. Water Res. 2005, 39 (7), 1265–1276.

(14) Allpike, B. P.; Heitz, A.; Joll, C. A.; Kagi, R. I.; Abbt-Braun, G.;Frimmel, F. H.; Brinkmann, T.; Her, N.; Amy, G. Size exclusionchromatography to characterize DOC removal in drinking watertreatment. Environ. Sci. Technol. 2005, 39 (7), 2334–2342.

(15) Boyer, T. H.; Singer, P. C. A pilot-scale evaluation of magneticion exchange treatment for removal of natural organic materialand inorganic anions. Water Res. 2006, 40 (15), 2865–2876.

(16) Drikas, M.; Chow, C. W. K.; Cook, D. The impact of recalcitrantorganic character on disinfection stability, trihalomethaneformation and bacterial regrowth: An evaluation of magneticion exchange resin (MIEX ) and alum coagulation. J. WaterSupply Res. Technol.-AQUA 2003, 52 (7), 475–487.

(17) Fearing, D. A.; Banks, J.; Guyetand, S.; Eroles, C. M.; Jefferson,B.; Wilson, D.; Hillis, P.; Campbell, A. T.; Parsons, S. A.Combination of ferric and MIEX for the treatment of a humicrich water. Water Res. 2004, 38 (10), 2551–2558.

(18) Johnson, C. J.; Singer, P. C. Impact of a magnetic ion exchangeresin on ozone demand and bromate formation during drinkingwater treatment. Water Res. 2004, 38 (17), 3738–3750.

(19) Morran, J. Y.; Drikas, M.; Cook, D.; Bursill, D. B. Comparisonof MIEX treatment and coagulation on NOM character. WaterSci. Technol.: Water Supply 2004, 4 (4), 129–137.

(20) Singer, P. C.; Schneider, M.; Edwards-Brandt, J.; Budd, G. C.Magnetic ion exchange for the removal of disinfection by-product precursors: pilot plant findings. J. Am. Water WorksAssoc. 2007, 99 (4), 128–139.

(21) Gregory, J.; Dhond, R. V. Anion exchange equilibria involvingphosphate, sulphate, and chloride. Water Res. 1972, 6, 695–702.

(22) Ravichandran, M.; Aiken, G. R.; Reddy, M. M.; Ryan, J. N.Enhanced dissolution of cinnabar (mercuric sulfide) by dissolvedorganic matter from the FL Everglades. Environ. Sci. Technol.1998, 32, 3305–3311.

(23) Ritchie, J. D.; Perdue, E. M. Proton-binding study of standardand reference fulvic acids, humic acids, and natural organicmatter. Geochim. Cosmochim. Acta 2003, 67 (1), 85–96.

(24) Lin, Y. P.; Singer, P. C.; Aiken, G. R. Inhibition of calciteprecipitation by natural organic material: Kinetics, mechanism,and thermodynamics. Environ. Sci. Technol. 2005, 39 (17), 6420–6428.

(25) Boening, P. H.; Beckmann, D. D.; Snoeyink, V. L. Activated carbonversus resin adsorption of humic substances. J. Am. Water WorksAssoc. 1980, 72 (1), 54–59.

(26) Helfferich, F. Ion Exchange; Dover Publications, Inc.: New York,1995.

(27) Stumm, W.; Morgan, J. J. Aquatic Chemistry; John Wiley andSons, Inc.: New York, 1996.

(28) U.S. Environmental Protection Agency. Method 300.0, Revision2.2: Determination of Inorganic Anions by Ion Chromatography;EPA: Cincinnati, OH, 1999.

(29) Aiken, G. R.; Thurman, E. M.; Malcolm, R. L.; Walton, H. F.Comparison of XAD macroporous resins for the concentrationof fulvic-acid from aqueous-solution. Anal. Chem. 1979, 51 (11),1799–1803.

(30) Chin, Y. P.; Aiken, G.; Oloughlin, E. Molecular-weight, poly-dispersity, and spectroscopic properties of aquatic humicsubstances. Environ. Sci. Technol. 1994, 28 (11), 1853–1858.

(31) Weishaar, J. L.; Aiken, G. R.; Bergamaschi, B. A.; Fram, M. S.;Fujii, R.; Mopper, K. Evaluation of specific ultraviolet absorbanceas an indicator of the chemical composition and reactivity ofdissolved organic carbon. Environ. Sci. Technol. 2003, 37 (20),4702–4708.

(32) Li, P.; SenGupta, A. K. Genesis of selectivity and reversibility forsorption of synthetic aromatic anions onto polymeric sorbents.Environ. Sci. Technol. 1998, 32, 3756–3766.

(33) Semmens, M.; Gregory, J. Selectivity of strongly basic anionexchange resins for organic anions. Environ. Sci. Technol. 1974,8 (9), 834–839.

ES071940N

VOL. 42, NO. 2, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 613