Chromium(VI) Reduction Kinetics byZero-Valent Iron in Moderately HardWater with Humic Acid: IronDissolution and Humic AcidAdsorptionT O N G Z H O U L I U , †

D A N I E L C . W . T S A N G , ‡ , § A N DI R E N E M . C . L O * , †

Department of Civil Engineering, The Hong Kong Universityof Science and Technology, Hong Kong, China, and Institutefor the Environment, The Hong Kong University of Scienceand Technology, Hong Kong, China

Received August 17, 2007. Revised manuscript receivedDecember 1, 2007. Accepted December 3, 2007.

In zerovalent iron treatment systems, the presence of multiplesolution components may impose combined effects thatdiffer from corresponding individual effects. The copresenceof humic acid and hardness (Ca2+/Mg2+) was found to influenceCr(VI) reduction by Fe0 and iron dissolution in a way differentfrom their respective presence in batch kinetics experiments withsynthetic groundwater at initial pH 6 and 9.5. Cr(VI) reductionrate constants (kobs) were slightly inhibited by humic acidadsorption on iron filings (decreases of 7–9% and 10–12% inthe presence of humic acid alone and together with hardness,respectively). The total amount of dissolved Fe steadilyincreased to 25 mg L-1 in the presence of humic acid alonebecausetheformationofsolubleFe-humatecomplexesappearedto suppress iron precipitation. Substantial amounts of solubleand colloidal Fe-humate complexes in groundwater may arouseaesthetic and safety concerns in groundwater use. Incontrast, the coexistence of humic acid and Ca2+/Mg2+

significantly promoted aggregation of humic acid and metalhydrolyzed species, as indicated by XPS and TEM analyses,which remained nondissolved (>0.45 µm) in solution. Thesemetal-humate aggregates may impose long-term impacts onPRBs in subsurface settings.

IntroductionPermeable reactive barriers (PRBs) using zerovalent iron (Fe0)as a reactive medium have been proven to be a viable andcost-effective technology in a number of laboratory-, pilot-,and full-scale studies for removing inorganics (e.g., chromate,nitrate, bromate, and arsenate) (1–4) as well as chlorinatedhydrocarbons and nitroaromatic compounds from ground-water (4–7). The majority of early studies investigatedchemical reduction of contaminants under simplified solu-tion conditions with a single contaminant (1, 5, 6). Although

recent studies started to take into account the effects ofsolution composition on contaminant reduction (7–10), mostof the effects were singly studied. Two or more solutioncomponents that coexist together may interact with oneanother and exert combined effects that are different fromtheir respective individual effects. To develop a betterunderstanding of the performance of Fe0 PRBs under morecomplicated geochemical conditions, the present studyinvestigates the Cr(VI) reduction by Fe0 in natural organicmatter (NOM)-rich groundwater in the absence and presenceof hardness.

In Fe0 treatment systems, the removal mechanisms ofCr(VI) are believed to involve instantaneous adsorption ofCr(VI) on Fe0 surface where electron transfer takes placeand Cr(VI) is reduced to Cr3+ with oxidation of Fe0 to Fe3+,and subsequently, Cr3+ precipitates as Cr3+ hydroxides and/or mixed Fe3+/Cr3+ (oxy)hydroxides (11–14). The reactivityof Fe0 is highly dependent on its surface characteristics andaqueous chemistry of groundwater. NOM is ubiquitous inshallow aquifers where Fe0 PRBs are most applicable, andhas a high tendency to be adsorbed on different mineralsurfaces such as Fe oxides (15–18). Adsorption of NOM, evenat low surface coverage, could significantly modify theproperties of mineral surfaces. It has been found that NOMadsorption could adjust, or even reverse, the electrostaticcharges of mineral surfaces (18) that would alter theinteractions between the surface and ions in solution. Theadsorbed NOM may also block the surface sites of Fe0 wherethe chemical reduction of contaminants takes place. Severalstudies have shown that the presence of NOM inhibited Fe0

reactivity for chlorinated hydrocarbon degradation, whichwas attributed to strong competition between organics andNOM for reactive sites (7–9, 19) and alteration of reductionpotential of surface sites (7).

Although Cr(VI) reduction by NOM was possible, it wasfound to occur at a very slow rate (20, 21). On the other hand,Dries et al. (10) suggested that the formation of metal-humatecomplexes in solution may be responsible for the delay ofmetal removal (Zn2+ and Ni2+) by Fe0, whereas the removalof chromate (CrO4

2-) was not significantly affected due largelyto its little interaction with negatively charged humic acid.Nevertheless, a significant amount of dissolved Fe wasreleased from Fe0 columns fed with humic acid, which wasnot further investigated. NOM in solution is known to havestrong binding affinity for Fe3+ (22, 23) to form stable andsoluble complexes. The formation of Fe-humate complexesmay influence the precipitation of corrosion products oniron filings, which would lead to Fe0 surface passivationreducing PRB reactivity (7, 24).

On the other hand, it has been found that hardness (e.g.,Ca2+ or Mg2+) and carbonate in solution would influence thereactivity of Fe0 for Cr(VI) removal by the formation ofpassivated precipitates, such as CaCO3 and Mg(OH)2 (25, 26).With the presence of Ca2+ and Mg2+ (equivalent to 100–200mg L-1 as CaCO3) in solution, Fe0 removal capacity towardCr(VI) was reduced by 10–17% (26) to 45% (27) in differentcolumn experiments. However, the removal rate of trichlo-roethylene (TCE) by Fe0 was slightly lower in hard ground-water with high alkalinity than soft groundwater with lowalkalinity (28).

Previous studies have demonstrated that the presence ofNOM or hardness in groundwater could, respectively,produce negative impacts on the removal of contaminantsby Fe0; but, there is a lack of study to assess the effects ofcoexisting NOM and hardness, which is particularly importantin shallow aquifers where water is often moderately hard to

* Corresponding author phone: 852-23587157; fax: 852-23581534;e-mail: cemclo@ust.hk.

† Department of Civil Engineering.‡ Institute for the Environment.§ Current address: Environmental and Water Resources Engineer-

ing, Department of Civil and Environmental Engineering, ImperialCollege London, South Kensington Campus, London SW7 2AZ.

Environ. Sci. Technol. 2008, 42, 2092–2098

2092 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 6, 2008 10.1021/es072059c CCC: $40.75 2008 American Chemical SocietyPublished on Web 02/09/2008

hard. This may be a challenging environment for theapplication of Fe0 PRB, because Ca2+ and Mg2+ are knownto promote NOM aggregation in solution due to chargeneutralization and conformation compaction of humicsubstances as a result of electrostatic binding of divalentcations (23). Additionally, Ca2+ and NOM may have synergisticeffects that alter the charge and potential of the head end ofdiffuse double layer (d-plane) of mineral surfaces (29),enhancing the adsorption of Ca2+ and NOM on mineralsurfaces. All these suggest that the effects of NOM or hardnessalone on Cr(VI) reduction may not be similar to those of Fe0

treatment systems containing both NOM and hardness;therefore, the objectives of this study were to evaluate thekinetics of Cr(VI) reduction, Fe dissolution, and NOMadsorption in a batch setting of Fe0 treatment system in thepresence of NOM (as humic acid) and/or hardness.

Experimental SectionZero-Valent Iron and Surface Charaterization. Iron filings(ETI-CC-1004) were obtained from Connelly GPM Inc. Theirelemental iron content, grain size, specific surface area, andparticle density were 96.28%, 0.25–2.0 mm, 1.8 m2 g-1, and6.43 g cm-3, respectively (26). Sieved fractions of 18–35 meshwere used without chemical pretreatment. Chemical oxida-tion states and compositions analyses of iron filings surfaceusing X-ray photoelectron spectroscopy (XPS, Perkin-ElmerPHI 5600) revealed that the iron filings were covered by apassive film of Fe2O3, which was consistent with the findingsof Ritter et al. (30). The depth profile scans of time-of-flightsecondary ion mass spectroscopy (Tof-SIMS, Perkin-ElmerPHI 7200) showed that the thickness of this Fe2O3 film wasat least 5 nm. The point of zero charge (pHpzc) of the ironfilings was determined to be pH 7.6 using a zeta potentialanalyzer (Zetaplus).

Chemicals and Chemical Analyses. Solutions containingCr(VI) and various compositions were prepared by dissolvingchemicals (K2Cr2O7, NaCl, CaCl2 ·2H2O, MgCl2.6H2O, Na2SO4,FeCl3 ·6H2O, NaOH, and HCl) into ultrapure water. Thechemicals are reagent grade and obtained from Riedel-deHaën and BDH. Ultrapure water (Barnstead D11911) wasdeoxygenated by purging with nitrogen gas for an hour priorto usage. Commercially available sodium salt of humic acid(Aldrich) was used to represent NOM. Humic acid stocksolution was prepared by dissolving certain amount of humicacid sodium salts into ultrapure water followed by filteringthrough 0.45-µm acetate cellulose membranes (ADVANTEC).Concentrations of humic acid were expressed as dissolvedorganic carbon (mg L-1 as DOC). Background electrolyte was5 mM NaCl unless specified. Initial solution pH was adjustedto 6 and 9.5 by dropwise addition of 0.01N NaOH and 0.01NHCl, to cover the groundwater pH range from slightly acidicto alkaline conditions that have been widely used in manystudies on PRBs (4). No buffer was used. Solution pH wasmeasured using pH electrode (Orion model 420A). Concen-trations of Cr(VI) and Fe2+ were measured by 1,5-diphenyl-carbazide (with detection limit and reproducibility of 10 µg/Land (2%, respectively) (31) and 1,10-phenanthroline colo-rimetric methods (with detection limit of 10 µg/L) (32),respectively, using UV/visible spectrophotometer (Ultrospec4300 Pro) at wavelengths of 540 and 510 nm. Concentrationsof humic acid were determined by a TOC analyzer (ShimadzuTOC-5000A). Total concentrations of dissolved metals (Ca,Cr, Fe, and Mg) were determined using atomic absorbancespectrometer (AAS, Varian 220FS).

Batch Experiments. Batch experiments were conductedusing 40 mL glass vials containing 0.4 g of Fe0 and 38 mL of10 mg L-1 Cr(VI) solutions with or without humic acid and/or hardness. To investigate the effects of humic acid on Fe0

reactivity, concentrations of humic acid were varied from 0to 20 mg L-1 as DOC, a typical range in groundwater (23).

To investigate the effects of copresent humic acid andhardness, solutions contained 0 or 20 mg L-1 (as DOC) humicacid and 0.8 mM Ca2+ or Mg2+, equivalent to 80 mg L-1 asCaCO3 hardness representing a moderately hard water. Vialswere sealed with screw caps containing Telfon-lined rubbersepta and shaken end-over-end at 26 rpm under roomtemperature (23 ( 1 °C). Solutions were sampled at regulartime intervals up to 30 min and filtered through 0.45 µmmembranes, followed by immediate measurement of pH andsubsequent chemical analyses. All batch experiments wererun in duplicate. Cr(VI) reduction by Fe0 was described bya pseudo-first-order kinetics model, of which the observedfirst-order rate constant, kobs (min-1) was the slope of theplot of ln (C/C0) versus time (where C0 and C are initial Cr(VI)concentration and its concentration at any time during thereduction reaction, respectively). All reported kobs values wereobtained with regression coefficient (R2) greater than 0.96.

In addition, two sets of supplementary experiments wereconducted to examine the competitive adsorption betweenan oxyanion (SO4

2-, which undergoes adsorption only andwas applied in place of CrO4

2-) and humic acid on iron filings;and the aggregation of humic acid in the presence of Ca2+

and Fe3+ in solution (without iron filings). The concentrationsof humic acid, SO4

2-, Ca2+, and Fe3+ were 20 mg L-1 as DOC,18.46 mg L-1 (equivalent to 10 mg L-1 CrO4

2- in molarity),0.8 mM, and 20 mg L-1, respectively. Other experimentalprocedures were the same as before.

XPS, TEM, and SEM Analyses. The aggregates/precipitatesthat were observed to form in suspensions during the batchexperiments were collected, and then freeze-dried forsubsequent XPS and scanning/transmission electron mi-croscopic (SEM, model JEOL-6300F; TEM, model JEOL-2010F) analyses. For experiments conducted at initial pH 9.5in the absence/presence of humic acid, samples were filteredand centrifuged at 20 000 rpm (HITACHI himac CR 21GII).The solids separated by centrifugation were collected andfreeze-dried for subsequent TEM analyses. All freeze-driedsamples were stored in N2-filled glass bottles.

Results and DiscussionsCr(VI) Reduction Kinetics. Table 1 summarizes the observedpseudo-first-order kinetics rate constants (kobs) of Cr(VI)reduction. The kobs values decreased as initial pH increasedwhether humic acid was present or not. The dependence ofkobs on pH has been widely reported (11, 33, 34). Since bothCr(VI) reduction and iron corrosion consume H+, low initialpH would favor the overall reaction (CrO4

2- + Fe0 + 8H+SCr3+ + Fe3+ + 4H2O) (11, 34) that solution pH was raisedsteadily to 9–10 (Figure S2 in the Supporting Information).Another reason is that at initial pH 9.5, the iron filings (withpHpzc 7.6) would develop negative surface charges andelectrostatic repulsion with CrO4

2-, thereby decreasing Cr(VI)reduction rates.

Addition of humic acid did not show any significantinhibitory effect on Fe0 reactivity toward Cr(VI) reduction atboth initial pH 6 and 9.5: humic acid in solution reduced kobs

values by about 7–9% (Table 1). In addition, under the sameinitial pH, the difference in kobs values was indistinguishablewith humic acid concentration increasing from 5 to 20 mgL-1 as DOC. This is in agreement with the findings of Drieset al. (10), which concluded that chromate removal was notaffected by humic acid. A comparison between the first-order rate constants (Table 2 in ref 10) also revealed a decreaseof about 14% of Cr(VI) reduction in the presence of humicacid despite relatively large standard deviation. By contrast,the inhibitory effect of humic acid was more crucial fortrichloroethylene and carbon tetrachloride degradation, ofwhich the reactivity was decreased by about 40% in batchexperiments (8) or by 2- to 5-fold in column studies (7, 9).As mentioned in the introduction, this was attributed to

VOL. 42, NO. 6, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 2093

humic acid adsorption that out-competes the contaminantsfor the reactive sites on iron surfaces (7–9, 19) and alters thereduction potential of neighboring sites (7). In view of themarginal suppression of Cr(VI) reduction in this study, itappears that the extent to which humic acid was adsorbed(Supporting Information Table S1) did not effectively blockthe surface or out-compete Cr(VI), or the majority of humicacid and Cr(VI) were adsorbed on different types of reactivesites.

The sole presence of hardness (0.8 mM Ca2+ or Mg2+)slightly enhanced Cr(VI) reduction with a 6% increase of kobs

values at initial pH 9.5, whereas it had little effect at initialpH 6 (Table 1). The effect of ionic strength was found to benegligible that using 7.4 mM NaCl as background electrolytedid not affect Cr(VI) reduction; therefore, the enhancementwas ascribed to the presence of Ca2+ or Mg2+. Decreases ofCa2+ and Mg2+ concentrations in solutions were observedduring the course of kinetics experiments (SupportingInformation Figure S1), which might reflect Ca2+ and Mg2+

adsorption or precipitation. Cation adsorption on iron surfacethrough electrostatic attraction increases with increasing pH(35, 36). At high initial pH, adsorbed Ca2+ and Mg2+ couldreduce the electrostatic repulsion between the negativelycharged iron surface (pHpzc 7.6) and chromate, thus enhanc-ing Cr(VI) adsorption and reduction.

When humic acid was present together with hardness, aminor decrease of Cr(VI) reduction rates (10–12%) wasobserved at initial pH 6 (Table 1), probably because Ca2+ andMg2+ could promote NOM adsorption on iron oxides due tosynergistic effects (29) that resulted in greater blockage ofreactive sites. But at initial pH 9.5, adsorbed Ca2+ and Mg2+

probably enhanced Cr(VI) reduction that obscured theinhibitory effect of NOM adsorption; consequently, notice-able difference in Cr(VI) reduction was not observed. A greaterdecrease of Ca2+ and Mg2+ concentrations (20–25%) in thepresence of humic acid than in its absence (4–14%) (Sup-porting Information Figure S1) suggested the correlation ofCa2+ and Mg2+ with humic acid adsorption/aggregation.

Humic Acid Adsorption/Aggregation. The fate of humicacid is of particular concern. Humic acid that is adsorbedcould significantly modify the characteristics of iron surfaceand, in turn, influence the performance of Fe0 PRB; whereashumic acid that remains in aqueous phase could complexor aggregate with cations (e.g., Fe3+) and subsequentlyinfluence the metal precipitation on iron filings that leadsto Fe0 surface passivation.

Figure 1 shows the changes of the amounts of humic acidand dissolved iron in solutions along with time under differentsolution conditions. The decrease in humic acid concentra-tion has been generally regarded as a reflection of humicacid adsorption on iron surfaces (8–10, 37). In the absence

of Cr(VI), steady humic acid adsorption (up to 25%) on ironfilings was observed (curves 1 and 2, Figure 1a) regardlessof pH values (Supporting Information Figure S2a), whichsuggests the specific adsorption of humic acid (e.g., ligandexchange) should have occurred despite electrostatic repul-sion between the iron filings surface and humic acid at highpH (15, 16). Recent FTIR study also confirmed the formationof inner-sphere surface complex between nanoscale Fe0 andhumic acid (37).

The amounts of adsorbed humic acid showed dispro-portional increase with increasing initial humic acid con-centrations (Supporting Information Table S1), indicatingthat humic acid adsorption sites on iron filings of smallspecific surface area (1.8 m2 g-1) were limited and easilysaturated. For this reason, the majority of humic acidremained in solution in this study. It should be noted thathumic acid obtained from different sources may exhibitdifferent extents of adsorption (7, 8), which was speculatedto reflect the differences in the characters of humic acid. Inchromate/humic acid binary solutions, apparently less humicacid adsorption (curves 3 and 4, Figure 1a) signified thatchromate could compete with humic acid for adsorptionsites to some degree despite specific interactions betweenhumic acid and iron filings surface. Supplementary experi-ments with solutions containing sulfate and humic acid(Supporting Information Figure S3) corroborated the com-petition between oxyanion and humic acid for a fraction ofadsorption sites on the iron surface.

However, distinct kinetics of humic acid adsorption wasrevealed when Ca2+ or Mg2+ was also present in solution(curves 5–8, Figure 1c). Binding with multivalent cations (suchas Fe3+, Ca2+, and Mg2+) can lead to a more compactconformation of NOM molecules by reducing intramolecularelectrostatic repulsions and by forming multidentate com-plexes with neighboring functional groups on NOM mol-ecules (23); or can reduce the electrostatic repulsion betweenNOM molecules and negatively charged surfaces (17, 29).BotheffectsarefavorableforNOMadsorptionandaggregation.

To evaluate whether the decreases in humic acid con-centration derived from aggregation regardless the presenceof Fe0 (because aggregates with size >0.45 µm in solutionwere filtered out), supplementary experiments were con-ducted using 20 mg L-1 (as DOC) humic acid, 0.8 mM Ca2+,and/or 20 mg L-1 Fe3+ (prepared by dissolving FeCl3 ·6H2Ointo ultrapure water) without the addition of iron filings(Supporting Information Figure S4). In the solution contain-ing Ca2+ only, humic acid aggregation was negligible on thistime scale. Nevertheless, the copresence of Ca2+ and Fe3+ insolution resulted in spontaneous humic acid aggregation,that is, a 27% decrease of humic acid concentration in solutionshortly after Fe3+ was added. The concentrations of humic

TABLE 1. Observed Pseudo-First Order Cr(VI) Reduction Rate Constantsa

initial pH 6 initial pH 9.5

electrolyte components

humic acidconcentraion

(mg L-1 as DOC) kobs, min-1 t1/2 min R 2 kobs, min-1 t1/2 min R 2

5 mM NaCl 0 0.0499 ( 0.0011 13.89 ( 0.30 0.9977 0.0427 ( 0.0008 16.23 ( 0.30 0.99635 0.0456 ( 0.0007 15.20 ( 0.23 0.9940 0.0398 ( 0.0010 17.42 ( 0.43 0.9903

10 0.0460 ( 0.0008 15.07 ( 0.26 0.9971 0.0391 ( 0.0011 17.73 ( 0.49 0.990115 0.0452 ( 0.0005 15.34 ( 0.17 0.9967 0.0396 ( 0.0007 17.50 ( 0.30 0.996420 0.0465 ( 0.0007 14.91 ( 0.22 0.9986 0.0389 ( 0.0009 17.82 ( 0.40 0.9959

5 mM NaCl and 0.8 mM CaCl2 0 0.0494 ( 0.0005 14.03 ( 0.14 0.9996 0.0452 ( 0.0007 15.34 ( 0.23 0.996620 0.0450 ( 0.0013 15.40 ( 0.43 0.9990 0.0433 ( 0.0011 16.01 ( 0.40 0.9992

5 mM NaCl and 0.8 mM MgCl2 0 0.0491 ( 0.0007 14.12 ( 0.20 0.9933 0.0455 ( 0.0009 15.23 ( 0.30 0.997620 0.0439 ( 0.0005 15.79 ( 0.18 0.9711 0.0413 ( 0.0006 16.78 ( 0.24 0.9914

7.4 mM NaCl 0 0.0504 ( 0.0013 13.75 ( 0.35 0.9785 0.0428 ( 0.0004 16.20 ( 0.15 0.9856a The ionic strength of 7.4 mM NaCl is equivalent to 5 mM NaCl plus 0.8 mM CaCl2 or MgCl2, and was used as reference

for comparison. Errors in table indicate standard deviation of duplicated batch experiments.

2094 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 6, 2008

acid and Fe3+ in solution remained unchanged thereafter.Dissolved Fe3+ was also shown to form settleable aggregateswith NOM via charge neutralization or bridging of stretchedhumic macromolecules (38).

Therefore, it is possible that the substantial decrease(about 85%) of dissolved humic acid in Ca2+ or Mg2+ solutions(Figure 1c) resulted from both adsorption on iron filings andaggregation in solution. The latter was particularly importantin the copresence of dissolved iron and Ca/Mg, because thepresence of these divalent cations can enhance aggregationthrough double layer compression or chemical association(39). Nevertheless, the molecular size and composition ofhumic acid fractions that are adsorbed on the iron surfaceor that remain as aggregates in solution are unclear andrequire detailed characterization in future.

Comparisons between curves 5–8 in Figure 1c revealedthat significant humic acid adsorption/aggregation took placeearlier in solutions with Ca2+ than Mg2+, and at low initialpH than high initial pH, respectively. It has been shown that,compared with Mg2+, Ca2+ exhibits higher binding affinitywith humic acid (40) and leads to greater humic acidaggregation due to its smaller size of hydrated ion (41). Thus,the enhancement of humic acid adsorption on Fe0 by Ca2+

to a greater degree was in agreement with Giasuddin et al.(37). In addition, since humic acid adsorption/aggregationincreases substantially with decreasing pH due to greaterproton neutralization, electrostatic attraction, and specificinteractions (17, 18, 38), the decrease of humic acid con-centration was observed earlier at low initial pH than highinitial pH. It should be noted that, despite different startingtime, humic acid adsorption/aggregation in solutions withCa2+ or Mg2+, at low initial pH or high initial pH almostreached the same extent at the end of experiments(Figure 1c).

Iron Dissolution. Dissolved iron was released as a resultof iron corrosion. The predominant form should be Fe3+

over the course of Cr(VI) reduction in the absence of humicacid because Fe2+, which may result from anaerobic ironcorrosion of water and autoreduction of Fe2O3 film, couldbe completely oxidized by Cr(VI) instantaneously (14).Colorimetric tests indicated the absence of Fe2+ in solutionswithout humic acid. However, it should be noted that onlyFetotal in solution was measured in the presence of humicacid due to the interference of humic acid with thecolorimetric measurement of Fe2+. For the solutionscontaining Cr(VI) alone, the amount of dissolved iron wasfar less at initial pH 6 than 9.5 (curves 9 and 10, Figure 1b)despite greater Cr(VI) reduction at lower pH. It is likelythat most dissolved Fe3+ expeditiously formed precipitatessuch as Fe3+ hydroxides and/or mixed Fe3+/Cr3+ (oxy)hy-droxides (11–14), which would coagulate fast near pHpzc

(pH 7–8.5 for Fe (oxy)hydroxides) but remain stable athigher pH (42). It is, therefore, speculated that at highsolution pH there was substantial amount of colloidalprecipitates (<0.45 µm) that could remain stable in solutionresulting in an increase of dissolved Fe3+ concentration(23). Additional experiments that applied centrifugation(at 20 000 rpm) to the 0.45 µm filtered samples showed adecrease of about 90% in dissolved iron concentration,suggesting that most of the dissolved Fe3+ should exist asfine colloids in solution. It was further confirmed by theobservation of fine particles (∼0.1 µm) in TEM images(Figure 2a) that the formation of colloidal iron precipitatesin solution.

In solutions containing humic acid alone, less significantiron dissolution was expected because there was no ironcorrosion induced by Cr(VI) reduction. However, the resultsdisplayed unexpectedly higher dissolved iron concentra-tion (comparing curves 1, 2 and 3, 4, Figure 1b). Anautoreduction reaction (Fe2O3 + Fe0 + 6H+ ⇒ 3Fe2+ +3H2O) (30) that took place on iron filing surface mayaccount for iron dissolution without Cr(VI) reduction.

FIGURE 1. Amounts of humic acid (a) and (c), and dissolved iron (b) and (d) in solutions along with time in batch kineticexperiments of Cr(VI) reduction by Fe0 under various conditions. Solutions contained: (b) humic acid alone at initial pH 6 (line 1); (O)humic acid alone at initial pH 9.5 (line 2); (9) both humic acid and Cr(VI) at initial pH 6 (line 3); (0) both humic acid and Cr(VI) atinitial pH 9.5 (line 4); (1) Cr(VI) alone at initial pH 6 (line 9); (3) Cr(VI) alone at initial pH 9.5 (line 10). (2)humic acid and Cr(VI) withCa2+ at initial pH 6 (line 5); (4) humic acid and Cr(VI) with Ca2+ at initial pH 9.5 (line 6); (() humic acid and Cr(VI) with Mg2+ atinitial pH 6 (line 7); ()) humic acid and Cr(VI) with Mg2+ at initial pH 9.5 (line 8). Initial concentrations of Cr(VI), humic acid, Ca2+/Mg2+ were 10 mg l-1, 20 mg l-1 as DOC, and 0.8 mM respectively. Background electrolyte was 5 mM NaCl. Lines are for guidanceonly. Error bars indicate standard deviation of duplicate batch experiments.

VOL. 42, NO. 6, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 2095

Comparing curves 3, 4 and 9, 10 in Figure 1b, the amountof dissolved iron was apparently higher in the presence ofhumic acid. Taking into account the high affinity of Fe2+

and Fe3+ for binding with humic acid (22, 23), the majorityof humic acid that remain in solution (curves 1–4, Figure1a) probably formed soluble complexes with dissolved iron,which would otherwise form precipitates. Centrifugationof the filtered samples resulted in a 50% decrease of ironconcentration (in contrast to a 90% decrease when withouthumic acid), indicating that about 40% of the dissolvediron was soluble Fe-humate complexes and the rest 50%was colloids. In addition, TEM observations (Figure 2b)indicate the copresence of Fe-humate colloids and smallaggregates in solution.

In the presence of Ca2+ or Mg2+, the change of dissolvediron concentration was similar to that of humic acid overthe course of kinetics experiments (curves 5–8, Figure 1cand d). Considerable amounts of visible aggregates wereformed in solutions after certain reaction times and filteredout of solution (>0.45 µm), resulting in significantdecreases of humic acid and dissolved Fe concentrations.The presence of large aggregates was clearly evidenced byTEM and SEM images (Figure 2c and Supporting Informa-tion Figure S5). Additionally, Fe-, Cr- and Ca-hydrolyzedspecies and carboxylic groups in humic acid were incor-porated in these aggregates. XPS spectra (Figure 3 andSupporting Information Figure S6) showed that the bindingenergy of O(1s) centered at 531.6 eV and a small peak ofC(1s) appeared at 288.5 eV, indicating hydroxyl oxygenand carboxyl carbon, respectively (Supporting InformationTable S2). Thus, these aggregates might comprise Fe-, Ca-,and Cr-hydrolyzed species and adsorbed/aggregated hu-mic acid. Charge neutralization, coprecipitation, and/oradsorption of humic acid and hydrolyzed metal speciesare probable mechanisms (38, 39). The less substantial pHincrease compared with solutions without hardness (curves5–8, Supporting Information Figure S2b) also signified theconsumption of OH- by aggregation. It appears that largeaggregates were formed by dissolved iron and humic acidin the presence of Ca/Mg, whereas soluble complexes andcolloids of Fe-humate were formed in the absence of Ca/Mg (Supporting Information Figure S7), indicating thatFe-humate aggregation was significantly enhanced by thecopresence of hardness ions.

Implications for Fe0 PRB Application. The inhibitoryeffects of humic acid on Fe0 reactivity toward Cr(VI) reductionwere not significant at both low and high initial pH; kobs

values were reduced by 7–9% and 10–12% in the absenceand presence of hardness, respectively. However, in contrastto previous studies that mainly focused on the reduction

efficiency, this study raised important concerns about thebehaviors of humic acid and dissolved iron during Cr(VI)reduction, which were shown to be different when humicacid was present alone and together with hardness. In theabsence of hardness, a large fraction of humic acid remainingin solutions appears to be an issue since it significantlyincreased the amount of dissolved iron in the form of solubleand colloidal Fe-humate complexes. Although less ironprecipitates formed on the surface of iron filings may prolongthe longevity of Fe0 PRBs (7, 24), elevated dissolved ironconcentration (up to 25 mg L-1) in groundwater may arouseaesthetic and safety concerns (43, 44). On the contrary, thecoexistence of humic acid and Ca2+/Mg2+ significantlypromoted the formation of large aggregates that comprisedmetal-hydrolyzed species and humic acid. The deposition ofaggregates may result in precipitation on the iron surfaceand thereby block the pores, decrease the permeability, and

FIGURE 2. TEM images of freeze-dried solids: (a) separated by centrifugation from filtered samples with Cr(VI) alone at initial pH 9.5;(b) collected from samples with Cr(VI) and humic acid at initial pH 6; and (c) collected from samples with Cr(VI), humic acid andCa2+ at initial pH 6.

FIGURE 3. XPS spectra of freeze-dried aggregates formed in thesolution with Cr(VI), humic acid, and Ca2+ at initial pH 6: (a)O1s and (b) C1s.

2096 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 6, 2008

develop preferential flow path of PRB. The long-term impactsof the aggregates on Fe0 PRB remain uncertain. Therefore,the major concern of Fe0 applications would shift dependingon whether or not NOM is present in hard water.

AcknowledgmentsThis work was supported by the Hong Kong Research GrantsCouncil under grant HKUST RGC 617006.

Supporting Information AvailableAdditional information on adsorbed amounts of humic acid(Table S1), identification of XPS spectral lines (Table S2);amounts of Ca2+ and Mg2+ in solutions (Figure S1); pHchanges along with Cr(VI) reduction kinetics experiments(Figure S2); results of two sets of supplementary experiments(Figure S3 and S4); SEM images, and XPS spectra of freeze-dried aggregates (Figure S5 and S6); and particle sizedistribution of precipitates/aggregates (Figure S7). Thismaterial is available free of charge via the Internet at http://pubs.acs.org.

Literature Cited(1) Cantrell, K. J.; Kaplan, D. I.; Wietsma, T. W. Zero-valent iron for

the in situ remediation of selected metals in groundwater. J.Hazard. Mater. 1995, 42, 201–212.

(2) Blowes, D. W.; Ptacek, C. J.; Benner, S. G.; McRae, C. W. T.;Bennett, T. A.; Puls, R. W. Treatment of inorganic contaminantsusing permeable reactive barriers. J. Contam. Hydrol. 2000, 45,123–137.

(3) Xie, L.; Shang, C. Role of humic acid and quinone modelcompounds in bromate reduction by zerovalent iron. Environ.Sci. Technol. 2005, 39, 1092–1100.

(4) Lo, I. M. C.; Surampalli, R. Y.; Lai, K. C. K., Eds. Zero-Valent IronReactive Materials for Hazardous Waste and Inorganics Removal;American Society of Civil Engineers: Reston, VA, 2006.

(5) Matheson, L. J.; Tratnyek, P. G. Reductive dehalogenation ofchlorinated methanes by iron metal. Environ. Sci. Technol. 1994,28, 2045–2053.

(6) Orth, W. S.; Gillham, R. W. Dechlorination of trichloroethenein aqueous solution using Fe0. Environ. Sci. Technol. 1996, 30,66–71.

(7) Klausen, J.; Vikesland, P. J.; Kohn, T.; Burris, D. R.; Ball, W. P.;Roberts, A. L. Longevity of granular iron in groundwatertreatment processes: solution composition effects on reductionof organohalides and nitroaromatic compounds. Environ. Sci.Technol. 2003, 37, 1208–1218.

(8) Tratnyek, P. G.; Scherer, M. M.; Deng, B.; Hu, S. Effects of naturalorganic matter, anthropogenic surfactants, and model quinoneson the reduction of contaminants by zero-valent iron. WaterRes. 2001, 35, 4435–4443.

(9) Marconetto, S.; Gui, L.; Gillham, R. Adsorption of natural organicmatter and its effects on TCE degradation by iron PRBs. IAHSPubl. 2005, 297, 389–397. Bringing Groundwater Quality Researhto the Watershed Scale, proceedings of GQ 2004, the 4thInternational Groundwater Quality Conference. Waterloo,Canada, July, 2004.

(10) Dries, J.; Bastiaens, L.; Springael, D.; Kuypers, S.; Agathos, S. N.;Diels, L. Effect of humic acids on heavy metal removal by zero-valent iron in batch and continuous flow column systems. WaterRes. 2005, 39, 3531–3540.

(11) Powell, R. M.; Puls, R. W.; Hightower, S. K.; Sabatini, D. A. Couplediron corrosion and chromate reduction: mechanisms forsubsurface remediation. Environ. Sci. Technol. 1995, 29, 1913–1922.

(12) Pratt, A. R.; Blowes, D. W.; Ptacek, C. J. Products of chromatereduction on proposed subsurface remediation material. En-viron. Sci. Technol. 1997, 31, 2492–2498.

(13) Blowes, D. W.; Ptacek, C. J.; Jambor, J. L. In-situ remediationof Cr(VI)-contaminated groundwater using permeable reactivewalls: laboratory studies. Environ. Sci. Technol. 1997, 31, 3348–3357.

(14) Eary, L. E.; Rai, D. Chromate removal from aqueous wastes byreduction with ferrous ion. Environ. Sci. Technol. 1988, 22, 972–977.

(15) Tipping, E. The adsorption of aquatic humic substances by ironoxides. Geochim. Cosmochim. Acta 1981, 45, 191–199.

(16) Gu, B.; Schmitt, J.; Chen, Z.; Liang, L.; MaCarthy, J. F.Adsorption and desorption of natural organic matter on ironoxide: mechanisms and models. Environ. Sci. Technol. 1994,28, 38–46.

(17) Vermeer, A. W. P.; van Riemsdijk, W. H.; Koopal, L. K. Adsorptionof humic acid to mineral particles. 1. Specific and electrostaticinteraction. Langmuir 1998, 14, 2810–2819.

(18) Saito, T.; Koopal, L. K.; van Riemsdijk, W. H.; Nagasaki, S.; Tanaka,S. Adsorption of humic acid on goethite: isotherms, chargeadjustments, and potential profiles. Langmuir 2004, 20, 689–700.

(19) Doong, R.; Lai, Y. Effect of metal ions and humic acid on thedechlorination of tetrachloroethylene by zerovalent iron. Chemo-sphere 2006, 64, 371–378.

(20) Wittbrodt, P. R.; Palmer, C. D. Reduction of Cr(VI) in the presenceof excess soil fulvic acid. Environ. Sci. Technol. 1995, 29, 255–263.

(21) Jardine, P. M.; Fendorf, S. E.; Mayes, M. A.; Larsen, I. L.; Brooks,S. C.; Bailey, W. B. Fate and transport of hexavalent chromiumin undisturbed heterogeneous soil. Environ. Sci. Technol. 1999,33, 2939–2944.

(22) Senesi, N.; Griffith, S. M.; Schnitzer, M.; Townsend, M. G. Bindingof Fe3+ by humic materials. Geochim. Cosmochim. Acta 1977,41, 969–976.

(23) Tipping, E. Cation Binding by Humic Substances; CambridgeUniversity Press: Cambridge, 2002.

(24) Kohn, T.; Livi, K. J. T.; Roberts, A. L.; Vikesland, P. J. Longevityof granular iron in groundwater treatment processes: corrsionproduct development. Environ. Sci. Technol. 2005, 39, 2867–2879.

(25) Lo, I. M. C.; Lam, C. S. C.; Lai, K. C. K. Competitive effects oftrichloroethylene on Cr(VI) removal by zero-valent iron. J.Environ. Eng. 2005, 131, 1598–1606.

(26) Lo, I. M. C.; Lam, C. S. C.; Lai, K. C. K. Hardness and carbonateeffects on the reactivity of zero-valent iron for Cr(VI) removal.Water Res. 2006, 40, 595–605.

(27) Karvonen, A. Cation effects on chromium removal in permeablereactive walls. J. Environ. Eng. 2004, 130, 863–866.

(28) D’Andrea, P.; Lai, K. C. K.; Kjeldsen, P.; Lo, I. M. C. Effectsof groundwater inorganics on the reductive dechlorinationof TCE by zero-valent iron. Water, Air, Soil Pollut. 2005, 162,401–420.

(29) Weng, L. P.; Koopal, L. K.; Hiemstra, T.; Meeussen, J. C. L.; vanRiemsdijk, W. H. Interactions of calcium and fulvic acid at thegoethite-water interface. Geochim. Cosmochim. Acta 2005, 69,325–339.

(30) Ritter, K.; Odziemkowski, M. S.; Gillham, R. W. An in situ studyof the role of surface filims on granular iron in the permeableiron wall technology. J. Contam. Hydrol. 2002, 55, 87–111.

(31) Milacic, R.; Štupar, J.; Kožuh, N.; Korošin, J. Critical evaluationof three analytical techniques for the determination ofchromium(VI) in soil extracts. Analyst 1992, 117, 125–130.

(32) Eaton, A. D.; Clesceri, L. S.; Greenberg, A. E. Standard Methodsfor the Examination of Water and Wastewater, 19 ed.; AmericanPublic Health Association: Washington, DC, 1995.

(33) Gould, J. P. The kinetics of hexavalent chromium reduction bymetallic iron. Water Res. 1982, 16, 871–877.

(34) Alowitz, M. J.; Scherer, M. M. Kinetics of nitrate, nitrite, andCr(VI) reduction by iron metal. Environ. Sci. Technol. 2002, 36,299–306.

(35) Ali, M. A.; Dzombak, D. A. Effects of simple organic acids onsorption of Cu2+ and Ca2+ on goethite. Geochim. Cosmochim.Acta 1996, 60, 291–304.

(36) Rietra, R. P. J. J.; Hiemstra, T.; van Riemsdijk, W. H. Interactionbetween calcium and phosphate adsorption on goethite.Environ. Sci. Technol. 2001, 35, 3369–3374.

(37) Giasuddin, A. B. M.; Kanel, S. R.; Choi, H. Adsorption of humicacid onto nanoscale zerovalent iron and its effect on arsenicremoval. Environ. Sci. Technol. 2007, 41, 2022–2027.

(38) Jung, A. V.; Chanudet, V.; Ghanbaja, J.; Lartiges, B. S.; Bersillon,J. L. Coagulation of humic substances and dissolved organicwith a ferric salt: an electron energy loss spectroscopy inves-tigation. Water Res. 2005, 3849–3862.

(39) Duan, J.; Wang, J.; Graham, N.; Wilson, F. Coagulation of humicacid by aluminium sulphate in saline water conditions. De-salination 2002, 150, 1–14.

VOL. 42, NO. 6, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 2097

(40) Nakahara, O.; Wada, S. I. Ca2+ and Mg2+ adsorption by anallophanic and a humic Andisol. Geodema 1994, 61, 203–212.

(41) Wall, N. A.; Choppin, G. R. Humic acids coagulation: influenceof divalent cations. Appl. Geochem. 2003, 1573–1582.

(42) Cornell, R. M.; Schwertmann, U. Cation The Iron Oxides:Structure, Properties, Reactions, Occurrences and Uses. Bindingby Humic Substances; Wiley-VCH: Weinheim, 2002.

(43) Munter, R.; Ojaste, H.; Sutt, J. Complexed iron removal fromgroundwater. J. Environ. Eng. 2005, 131, 1014–1020.

(44) Pacini, V. A.; Ingallinella, A. M.; Sanguinetti, G. Removal of ironand manganese using biological roughing up flow filtrationtechnology. Water Res. 2005, 39, 4463–4475.

ES072059C

2098 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 6, 2008