Influence of Anionic Cosolutes and pH on Nanoscale Zerovalent Iron

Jul 11, 2012 - Technol. , 2012, 46 (15), pp 8365–8373 ... SO42–, and ClO4– suspensions decreased by 95% over 1 month but were generally equivale...
0 downloads 0 Views 3MB Size
Download PDF
Article pubs.acs.org/est

Influence of Anionic Cosolutes and pH on Nanoscale Zerovalent Iron Longevity: Time Scales and Mechanisms of Reactivity Loss toward 1,1,1,2-Tetrachloroethane and Cr(VI) Yang Xie† and David M. Cwiertny‡,* †

Department of Chemical and Environmental Engineering, University of California, Riverside A242 Bourns Hall Riverside, California 92521, United States ‡ Department of Civil and Environmental Engineering, University of Iowa, 4105 Seamans Center Iowa City, Iowa 52242, United States S Supporting Information *

ABSTRACT: Nanoscale zerovalent iron (NZVI) was aged over 30 days in suspension (2 g/L) with different anions (chloride, perchlorate, sulfate, carbonate, nitrate), anion concentrations (5, 25, 100 mN), and pH (7, 8). During aging, suspension samples were reacted periodically with 1,1,1,2tetrachloroethane (1,1,1,2-TeCA) and Cr(VI) to determine the time scales and primary mode of NZVI reactivity loss. Rate constants for 1,1,1,2TeCA reduction in Cl−, SO42−, and ClO4− suspensions decreased by 95% over 1 month but were generally equivalent to one another, invariant of concentration and independent of pH. In contrast, longevity toward 1,1,1,2-TeCA depended upon NO3− and HCO3− concentration, with complete reactivity loss over 1 and 14 days, respectively, in 25 mN suspensions. X-ray diffraction suggests that reactivity loss toward 1,1,1,2TeCA in most systems results from Fe(0) conversion into magnetite, whereas iron carbonate hydroxide formation limits reactivity in HCO3− suspensions. Markedly different trends in Cr(VI) removal capacity (mg Cr/g NZVI) were observed during aging, typically exhibiting greater longevity and a pronounced pH-dependence. Notably, a strong linear correlation exists between Cr(VI) removal capacities and rates of Fe(II) production measured in the absence of Cr(VI). While Fe(0) availability dictates longevity toward 1,1,1,2-TeCA, this correlation suggests surface-associated Fe(II) species are primarily responsible for Cr(VI) reduction.

■

tially relative to passive FeCO3 (siderite), allowing free CO32− to promote Fe(0) corrosion. Klausen et al.27 observed reactivity toward organohalides and nitroaromatics that was initially enhanced (over 90 days) at a high bicarbonate concentration (20 mM), but ultimately inhibited due to formation of passive iron carbonate precipitates.31 Unlike these other anions, nitrate is reducible by zerovalent iron,14,32,33 which generates passive ferric iron (Fe(III)) (oxyhydr)oxides that dampen reactivity and shorten longevity.13,17,18 While the roles of anions as inhibitors or promoters should not change in NZVI systems, NZVI performance is likely far more sensitive to their presence due to its greater corrosion rate. Reardon et al. observed NZVI corrosion rates (∼2 mol kg−1d−1)34 orders of magnitude greater than those of granular iron samples (1.4−31.3 × 10−4 mol kg−1d−1) and electrolytic iron powder (7.9 × 10−2 mol kg−1d−1).35,36 With such markedly higher corrosion, solubility limits for passive phases (e.g.,

INTRODUCTION Although nanoscale zerovalent iron (NZVI) treatment zones are becoming a popular alternative to permeable reactive barriers (PRBs),1−4 questions surround their reactive lifetime. To date, field-scale monitoring data5−8 are limited, and laboratory aging studies9−11 have yet to examine reactivity loss in systems with geochemical complexity representative of the subsurface. While generally accepted that NZVI longevity will be shorter than that of PRBs (operative several years to decades12), establishing time scales of NZVI reactivity loss in response to geochemical conditions will help identify optimal sites for application. From established behavior of granular iron,13−18 the identity and concentration of anionic cosolutes will dictate NZVI longevity. Chloride13,15,19 and sulfate13,20 are corrosion promoters capable of destabilizing passive surface films,21 and thus have been reported to enhance reactivity. (Bi)carbonate has been extensively investigated,13,22−29 and its influence typically depends on concentration and exposure time. Recently, Bi et al.30 observed enhanced reactivity toward 4chloronitrobenzene at low carbonate concentrations, where it was hypothesized that amorphous Fe(OH)2 forms preferen© 2012 American Chemical Society

Received: Revised: Accepted: Published: 8365

May 2, 2012 July 9, 2012 July 11, 2012 July 11, 2012 dx.doi.org/10.1021/es301753u | Environ. Sci. Technol. 2012, 46, 8365−8373

Environmental Science & Technology

Article

siderite in HCO3− systems) will be achieved more readily, in turn accelerating NZVI reactivity loss. For promoters (e.g., Cl− and SO42−), enhanced corrosion rates may also shorten longevity by increasing electron transfer to nontargets (e.g., oxygen and water). In this case, the anions’ influence on surface oxide formation will be critical because this layer regulates the rate of electron transfer from the metallic particle core to species at the particle−water interface.37,38 In a noteworthy series of studies, Liu and Lowry39 reported the following reactivity trend (from most to least reactive) toward TCE in NZVI suspensions with 5 mN concentration of different anions: Cl− > SO42− > HCO3− > HPO42− >NO3−. However, this trend reflects pseudo-first-order rate constants for TCE reduction measured over relatively short periods of time (6 days). As such, it only provides the immediate impact of these anions on NZVI reactivity while providing little indication as to how long the enhanced or diminished reactivity is sustained in each solution. A follow-up study40 characterized NZVI corrosion products generated over 6 months in anaerobic, unbuffered suspensions with either 10 mN Cl−, NO3−, SO42−, HPO4−, or HCO3−. After one month, the following trend in NZVI oxidation was observed based upon the Fe(0) remaining in the particle as determined by EXAFS (from most to least oxidized): NO3− > Cl− > SO42− > HPO4− > HCO3−. Notably, the reactivity of these aged NZVI suspensions toward a model pollutant target was not simultaneously measured. Because these same authors previously showed TCE reduction rates to be independent of Fe(0) content for aged NZVI,9 it is not possible to translate their characterization results into an effective performance lifetime. Here, we simultaneously monitor changes in reactivity and material properties for NZVI suspensions aged up to one month in solutions of Cl−, HCO3−, NO3−, SO42−, and ClO4− over a range of concentrations (5, 25, and 100 mN) and buffered at environmentally relevant pH values (pH 7 and 8). Over time, NZVI reactivity loss was quantified via multiple metrics including rates of aqueous Fe(II) production, pseudofirst-order rate constants (kobs values) for 1,1,1,2-tetrachloroethane (1,1,1,2-TeCA) reduction, hexavalent chromium (Cr(VI)) removal capacities (in mg of Cr/g NZVI), and oxidation−reduction potential (ORP), which is often used to monitor NZVI field-scale performance.5,8 Inclusion of 1,1,1,2TeCA and Cr(VI) as model pollutants allows the total reducing capacity available in aged NZVI suspensions to be evaluated; from our prior work,11 1,1,1,2-TeCA is degraded over short time scales by Fe(0), whereas Cr(VI) can be reduced both by Fe(0) and forms of aqueous and solid phase Fe(II). Thus, differences in reactivity trends toward 1,1,1,2-TeCA and Cr(VI) provide an indirect measure of how the relative amounts of reducing equivalents available as Fe(0) and Fe(II) evolve during aging. NZVI particles were also characterized over time using transmission electron microscopy (TEM) and X-ray diffraction (XRD) to relate measured reactivity loss to corrosion product development.

prepared by weighing 400 mg NZVI powder into a 250 mL Kimax bottle and combining it with 200 mL of 50 mM HEPEs buffer with an appropriate ionic composition. The anion concentrations were 5, 25, and 100 mN for Cl−, SO42− and ClO4− and 5 and 25 mN for HCO3− and NO3−. Sodium salts of all anions were used. ClO4− was included because it has previously been assumed to interact minimally with granular iron surfaces,13 yet it has also been shown to be a weak oxidant for NZVI41 and granular iron.42 Thus, its influence on NZVI longevity remains largely unknown. HEPES buffer (pKa value of 7.55) provided reasonable buffering capacity at both pH 7 and 8. Our rationale for using organic buffers was discussed previously.11 It was necessitated because suspension pH increased considerably in the absence of buffer due to water reduction. After assembly, the bottles were briefly sonicated (∼10 s) to disperse the NZVI powder and then aged for 30 day while sealed. During aging, suspensions did not contain a model pollutant, with water reduction (i.e., corrosion) representing the only aging process. The suspensions were vigorously mixed by hand daily, and when necessary, small amounts of 5 N HCl were added to maintain a pH of 7.0 (±0.2) or 8.0 (±0.2). Even with 50 mM HEPES, pH slightly increased over time from water reduction. Because higher HEPES concentrations might influence reactivity, small volumes of HCl were used to maintain the pH of all suspensions, even those with anions other that Cl−. This process resulted in the addition of less than 1 mM of Cl− to each suspension. Evaluation of Aged NZVI Reactivity. Suspension reactivity was evaluated via measurement of kobs values for 1,1,1,2-TeCA reduction and removal capacities of chromate (CrO42−) in mg of Cr/g NZVI according to procedures in the SI and detailed previously.11 Reactivity studies were conducted with suspension samples periodically withdrawn during aging. To remove any contribution from aqueous Fe(II) that accumulated in suspension over time, samples were first centrifuged (8000 rpm for 5 min), and the supernatant was decanted inside the anaerobic chamber. The aged particles were then washed twice with deoxygenated, deionized water prior to being resuspended to the desired solid loading (typically 1 or 2 g/L) in the appropriate buffer. Also measured (see SI) was the rate of aqueous Fe(II) production in aged NZVI suspensions, which we present as a metric of NZVI corrosion. Briefly, washed suspensions of aged NZVI (1 g/L) were prepared in an appropriate ionic solution. Samples were withdrawn over time, passed through a 0.2 μm filter and subsequently analyzed for aqueous Fe(II). Oxidation−reduction potential (ORP) was also measured over time using a single-junction ORP electrode (Oakton Instruments). Analytical Methods. Dissolved Fe(II) and Cr(VI) concentrations were quantified colorimetrically using 1,10phenanthroline43,44 and diphenylcarbazide,45 respectively. The concentration of 1,1,1,2-TeCA and its sole reduction product 1,1-dichloroethylene (1,1-DCE) were determined via GC/ECD according to established methods.11 ORP measurements were made by placing the electrode into a well-mixed suspension of 2 g/L NZVI with at least 25 mL of total volume. Stable readings were typically achieved after 1 min. The probe was calibrated with saturated solutions of quinhydrone at pH 4 and 7 according to manufacturer specifications. Further analytical details are in the SI. Materials Characterization. Changes in NZVI mineralogy during aging were investigated with powder XRD, and particle

■

MATERIALS AND METHODS Reagents. NZVI was acquired from Nanostructured and Amorphous Materials, Inc. (Houston, TX) and characterized in detail previously.11 Unless noted, aging experiments were conducted with the same lot of NZVI powder. A complete reagent list is provided in the Supporting Information (SI). Aging of NZVI Suspensions. Within an anaerobic chamber (97% N2, 3% H2), 2 g/L suspensions of NZVI were 8366

dx.doi.org/10.1021/es301753u | Environ. Sci. Technol. 2012, 46, 8365−8373

Environmental Science & Technology

Article

Figure 1. Change in kobs(1,1,1,2-TeCA) values over aging time in the presence of (a) 5 mN and (b) 25 mN Cl−, SO42−, ClO4−, HCO3−, and NO3− at pH 8. The dashed line represents the loss coefficient for 1,1,1,2-TeCA in NZVI-free controls, as described in the text. Uncertainties represent 95% confidence intervals from regression analyses used to determine kobs values. Reactors contained a reductant loading equivalent to 2 g/L of fresh NZVI, 50 mM HEPES buffer to stabilize pH, and the appropriate anion composition.

morphology was examined with TEM. As in our prior work,11 samples for XRD analysis were deposited on a microscope slide, dried within the anaerobic chamber, and analyzed immediately after drying. The typical analysis time was 30 min. Evidence of oxidation during analysis was not observed, with select samples prepared in up to 50% glycerol (by volume), which slows oxidation, yielding identical diffraction patterns to the same samples prepared without glycerol. TEM samples were dried on a standard Cu grid within the anaerobic chamber prior to imaging. Additional details regarding sample preparation and instrumentation are in the SI and our previous work.11

equivalent kobs(1,1,1,2-TeCA) values, whereas SO42− exhibited slightly depressed reactivity over intermediate time scales (5− 14 d). In these suspensions, reactivity increased roughly 3-fold over the first two days of aging, which previous studies10,39 have attributed to autoreduction of a passive oxide layer initially present on the particle surface. In contrast, 5 mN NO3− immediately dampened NZVI reactivity by an order of magnitude relative to other systems. Values of kobs(1,1,1,2TeCA) slowly rebounded over 5 days in NO3− systems, also increasing 3-fold due in part to autoreduction. Because NO3− reduction occurred in parallel (SI Figure S1), this rebound may also reflect consumption of NO3− and, in turn, a greater number of reducing equivalents available for 1,1,1,2-TeCA. After autoreduction, the decrease in kobs(1,1,1,2-TeCA) values over time roughly followed exponential decay. Plots of ln[kobs(1,1,1,2-TeCA)] versus time (SI Figure S2), therefore, provide an empirical coefficient for reactivity loss (or kRL value) in aged NZVI suspensions. The near-equivalence in kobs(1,1,1,2-TeCA) values for 5 mN Cl−, ClO4−, HCO3−, and SO42− suspensions corresponds to equal rates of reactivity loss, with kRL values of ∼0.1 d−1 in these systems. Thus, although these suspensions maintained measurable reactivity over 30 days, the extent of reactivity loss was substantial (∼95%). In contrast, exposure to 5 mN NO3− shortened NZVI longevity; loss of 1,1,1,2-TeCA was equivalent to controls after 30 days. Nevertheless, the rate of reactivity loss in 5 mN NO3− suspensions (kRL ∼ 0.1 d−1) was on par with other suspensions. For 25 mN suspensions at pH 8 (Figure 1b), kobs(1,1,1,2TeCA) values for Cl−, SO42− and ClO4− were approximately equal to those measured at 5 mN. These systems also exhibited a similar influence from autoreduction and comparable rates of reactivity loss (i.e., roughly 95% reactivity loss over 30 days; see SI Figure S2). In fact, NZVI reactivity and longevity toward 1,1,1,2-TeCA at pH 8 were unaffected up to 100 mN for Cl−, SO42−, and ClO4− (SI Figure S3).

■

RESULTS AND DISCUSSION Trends in Reactivity Loss toward 1,1,1,2-TeCA. Figure 1 illustrates the change in kobs values for 1,1,1,2-TeCA reduction [kobs(1,1,1,2-TeCA)] during aging in the presence of 5 mN (Figure 1a) and 25 mN (Figure 1b) anionic cosolutes at pH 8. Data are presented from at least duplicate suspensions. In all systems, 1,1,1,2-TeCA concentrations exhibited exponential decay, thus kobs(1,1,1,2-TeCA) values were obtained from semilog plots of 1,1,1,2-TeCA concentration versus time. Transformation of 1,1,1,2-TeCA occurred via reductive dechlorination; 1,1-dichloroethylene was the only detectable reaction product and sorption-corrected carbon mass balances were >90%. Minor 1,1,1,2-TeCA losses occurred in NZVI-free controls, attributable to sorption on the septa used to seal reactors. The loss coefficient measured in these controls is represented by the dashed horizontal line in Figure 1. Aged NZVI was assumed to possess no further reducing capacity toward 1,1,1,2-TeCA when kobs(1,1,1,2-TeCA) values were equivalent to the loss measured in controls (>99% reactivity loss relative to maximum kobs values). A notable feature in Figure 1a is the relatively modest reactivity differences observed among most 5 mN suspensions. Suspensions of Cl−, ClO4−, and HCO3− yielded essentially 8367

dx.doi.org/10.1021/es301753u | Environ. Sci. Technol. 2012, 46, 8365−8373

Environmental Science & Technology

Article

Figure 2. Change in Cr(VI) removal capacity (in mg of Cr/g NZVI) as a function of aging time in 5 and 25 mN suspensions of Cl−, SO42−, ClO4−, HCO3−, and NO3− at pH 7 and 8 (see key for specific details). Data are shown on the same y-axis scale for the purpose of comparison. Reactors contained a reductant loading equivalent to 1 g/L of fresh NZVI, 50 mM HEPES buffer to stabilize pH, and the appropriate anion composition.

these conditions, although verification via measurement of reduction products (e.g., Cl−) was not performed. Notably, Cao et al.41 observed 60% reduction of a 200 mg/L ClO4− solution by NZVI (20 g/L) over 28 days at room temperature and pH 8.0, yielding Cl−. While our experiments were conducted at a lower solid loading (2 g/L), NZVI was aged herein at a considerably higher ClO4− concentration (25 mN ∼ 2500 mg/ L). Trends in Reactivity Loss toward Cr(VI). Cr(VI) removal capacities during aging are shown in Figure 2 for 5 and 25 mN anionic suspensions at pH 7 and 8. In Figure 2, y-axes share the same range to facilitate intersystem comparisons. From earlier work,11 reduction was the primary mechanism of Cr(IV) loss in our systems. Unlike kobs(1,1,1,2-TeCA) values, Cr(VI) removal capacities exhibited a much clearer pH dependence. Generally, suspensions aged at pH 7 exhibited, at least initially, nearly twice the removal capacity of those aged at pH 8, behavior most easily observed for SO42− and HCO3−. At a specific pH value, trends in Cr(VI) removal were also largely insensitive to anion concentration, which again is most apparent for SO42− and HCO3− systems. As exceptions, rates of reactivity loss were greatest in 25 mN suspension of Cl− and ClO4−, but only at pH 7. Recall, that 25 mN suspensions of these anions also exhibited depressed reactivity toward 1,1,1,2-TeCA at pH 7 (SI Figure S2). Overall, however, trends in reactivity loss toward Cr(VI) were rather different from those observed for 1,1,1,2-TeCA. No influence of autoreduction was apparent (i.e., removal capacities were either constant or decreased slightly during initial periods

NZVI longevity was shortened considerably in 25 mN HCO3− and NO3− systems at pH 8. After initially increasing over 2 days via autoreduction, kobs(1,1,1,2-TeCA) values decreased more abruptly in 25 mN HCO3− suspensions, and 1,1,1,2-TeCA reduction was not observed beyond 21 days. Reactivity loss occurred most rapidly in 25 mN NO3− systems, in which 1,1,1,2-TeCA reduction ceased after 1 day. Only ∼20% of available NO3− was reduced by NZVI prior to complete reactivity loss (SI Figure S1), suggesting consumption of reducing equivalents by NO3− inhibited reactivity toward 1,1,1,2-TeCA. Values of kobs(1,1,1,2-TeCA) during aging at pH 7 are shown in SI Figure S4 for 5 and 25 mN suspensions and are mostly equivalent to those observed at pH 8 with a few notable exceptions. In certain instances, peak kobs(1,1,1,2-TeCA) values after autoreduction were greater in suspensions at pH 7, behavior most noticeable for 5 mN NO3− systems but also observed for 5 mN Cl− and ClO4−. Thus, the rate and extent of autoreduction appears to be pH dependent. At 25 mN, kobs(1,1,1,2-TeCA) values in Cl− and ClO4− suspensions exhibited a modest decrease at pH 7 relative to pH 8 (most easily seen in SI Figure S2), which is counter to the pHdependence commonly observed for NZVI11 and granular iron.46 Despite the smaller kobs(1,1,1,2-TeCA) values in 25 mN Cl− at pH 7, the rate of reactivity loss remained invariant of pH (kRL ∼ 0.1 d−1 at both pH 7 and 8; SI Figure S2). In contrast, the depressed kobs(1,1,1,2-TeCA) values in 25 mN ClO4− at pH 7 were accompanied by a nearly 2-fold greater rate of reactivity loss (SI Figure S2). We speculate that this enhanced rate of reactivity loss arises from ClO4− reduction by NZVI under 8368

dx.doi.org/10.1021/es301753u | Environ. Sci. Technol. 2012, 46, 8365−8373

Environmental Science & Technology

Article

Figure 3. Development of corrosion products for NZVI aged at pH 8 in 5 and 25 mN HCO3− suspensions. For each respective concentration, TEM images (a and c, respectively) and XRD patterns (b and d, respectively) are shown as a function of aging time. In XRD patterns diffraction lines associated with specific phases are identified, and select phases are indicated by arrows in TEM images. These include α-Fe for Fe(0), M for magnetite, GR for carbonate green rust, and ICH for iron carbonate hydroxide. Diffraction lines were identified based upon comparison to d-spacings and 2θ values provided by Kohn et al.31 and references therein.

indicates loss of Fe(0) to levels below detection and predominance of magnetite over time scales that coincide with ∼90% decrease in kobs(1,1,1,2-TeCA) values from their maxima. TEM images (SI Figure S7) also revealed secondary phases with hexagonal plate morphology consistent with green rust47 after 30−40 days aging in Cl−, SO42− and ClO4− suspensions. Notably, an Fe(0) signal was apparent in XRD characterization of Cl− suspensions aged for 30 days at pH 8 but was below the limit of detection by XRD at pH 7 (SI Figure S6), likely due to the higher rate of Fe(0) corrosion at lower pH. This may explain the greater reactivity toward 1,1,1,2TeCA measured at pH 8.0 in 25 mN Cl− systems. We have previously shown that air-oxidized NZVI suspensions retain a significant portion of Fe(0) but are unreactive toward 1,1,1,2-TeCA because of a passive ferric oxide coating formed during oxidation.11 In complement with results herein, we conclude that 1,1,1,2-TeCA reduction requires Fe(0) not only be present at appreciable levels in the particle core, but its reducing equivalents must also be readily accessible to solution via a conductive oxide surface layer such as magnetite or the green rust phases observed in Cl−, SO42− and ClO4− suspensions. For HCO3− suspensions, the formation rate and nature of corrosion products were essentially identical at pH 7 and 8 (SI Figure S8) but were strongly influenced by bicarbonate concentration. At 5 mN, TEM images (Figure 3a) collected over the first week revealed small amounts of a secondary mineral phase associated with NZVI aggregates with the characteristic hexagonal plate morphology of carbonate green rust.47 This phase was not present at sufficient quantities to be observed via XRD, with magnetite and residual Fe(0) representing the only detectable phases after 7 days (Figure

of aging), and removal of Cr(VI) was measurable in nearly all suspensions over the entire duration of aging. As a notable example, NZVI aged in 25 mN HCO3−, which was passive toward 1,1,1,2-TeCA after 14 days, maintained nearly constant reactivity toward Cr(VI) over 30 days at both pH values. Similarly, while Cr(VI) removal was suppressed in NO3− suspensions, the only instance of complete reactivity loss occurred in 25 mN suspensions at pH 8 after 5 days. The general disparity in reactivity trends exhibited toward Cr(VI) and 1,1,1,2-TeCA supports unique reductants for each species in aged NZVI suspensions. Role of Corrosion Product Formation in Reactivity Loss toward 1,1,1,2-TeCA. Trends in corrosion product formation most readily explain reactivity loss toward 1,1,1,2TeCA. For example, characterization of NZVI aged in 25 mN NO3− at pH 7 is shown in SI Figure S5. Consistent with other reports26 TEM images (SI Figure S5a) revealed no significant morphological changes over 30 days, yet diffraction patterns (SI Figure S5b) indicate rapid loss of Fe(0) accompanied by magnetite formation over 2 h. The time scale of Fe(0) consumption closely parallels reactivity loss toward 1,1,1,2TeCA, consistent with 1,1,1,2-TeCA reduction being closely linked to the availability of Fe(0) in suspension. Thickening of the magnetite surface layer, promoted by nitrate as an oxidant, suppresses reactivity until the near-complete consumption of Fe(0) stops 1,1,1,2-TeCA reduction entirely. Oxidation of Fe(0) to magnetite also appears to be the primary mechanism for reactivity loss toward 1,1,1,2-TeCA in Cl−, SO42−, and ClO4− suspensions. Characterization results for 25 mN Cl− suspensions at pH 7 and 8 are shown in SI Figure S6, and aging products in pH 7 suspensions of Cl−, SO42−, and ClO4− are compared in SI Figure S7. In all instances, XRD 8369

dx.doi.org/10.1021/es301753u | Environ. Sci. Technol. 2012, 46, 8365−8373

Environmental Science & Technology

Article

Figure 4. Cr(VI) removal capacity plotted as a function of Fe(II) production rate. Data are presented as a function of (a) anion identity and concentration and (b) suspension pH. In (a), open symbols represent 5 mN suspensions and solid symbols represent 25 mN suspensions. In (b), the result of linear regression analysis for all data with nonzero Fe(II) production rates is shown, along with the equation for the resulting best fit line.

SI Figure S8), Cr(VI) removal capacity was essentially constant over time and invariant with HCO3− concentration (Figure 2). Thus, although ICH hinders electron transfer between Fe(0) and 1,1,1,2-TeCA, we propose it is sufficiently redox active so as to sustain Cr(VI) removal. The similarity among corrosion products in Cl−, SO42− and ClO4− systems also is inconsistent with differences in removal capacity, rates of reactivity loss, and the pH dependence for Cr(VI) removal in these systems. Characterization results further support a scenario in which the source of reactivity and pathways responsible for reactivity loss differ substantially between Cr(VI) and 1,1,1,2-TeCA. Correlation of NZVI Reactivity with ORP and Fe(II) Production Rate of Aged Suspensions. As discussed further in the SI, ORP was a poor indicator of suspension reactivity toward 1,1,1,2-TeCA and Cr(VI) (SI Figure S9). For Cr(VI) removal, a much stronger predictor was the rate of Fe(II) production in aged suspensions. Plots of Cr(VI) removal capacity as a function of measured Fe(II) production rate yielded a positive, linear relationship (Figure 4). Fe(II) production rates were measured in the absence of 1,1,1,2TeCA and Cr(VI) using aged NZVI particles that were initially water-washed to remove soluble Fe(II) that accumulated during aging. Rates of Fe(II) production were then quantified from the slope of the initially linear portion (t < 15 min) of aqueous Fe(II) concentration versus time plots (SI Figure S10). Notably, for all anions, soluble Fe(II) concentrations were below limits of detection (∼1 μM) during removal capacity studies. This observation is not entirely unexpected because ferrous iron rapidly reduces Cr(VI) at circumneutral pH values,49,50 and thus would not likely accumulate in bulk solution during the repeated addition of Cr(VI) to the reactor. In Figure 4 data from NO3− systems are plotted along the yaxis because rates of Fe(II) production were not quantifiable in these suspensions. Whereas most systems exhibited an initially linear increase in Fe(II) concentration over time (SI Figure S10a), soluble Fe(II) concentrations in 5 mN NO3− were roughly constant or slightly decreased over time (SI Figure

3b). After 30 days, TEM images revealed another secondary phase with a sheet-like morphology. The XRD pattern for this new phase is consistent with iron carbonate hydroxide, which possesses a distinct diffraction pattern relative to other iron− carbonates including green rust and siderite.31 Iron carbonate hydroxide (ICH; [Fe3(OH)2.2CO3]) has previously been observed in field-scale PRBs48 and in laboratory column studies with granular iron.31 In the latter work, SEM imaging of particle cross sections revealed a surface-associated phase with thin platelet morphology consistent with the phase identified in Figure 3. At 25 mM HCO3−, TEM (Figure 3c) and diffraction patterns (Figure 3d) indicate that ICH formed more extensively and over much shorter time scales (∼1 day). TEM images show that ICH tends to generate in close proximity to NZVI aggregates, with most NZVI aggregates being enveloped by its sheet-like morphology (Figure 3 and SI Figure S8). Collectively, evidence suggests that ICH formation is primarily responsible for NZVI reactivity loss toward 1,1,1,2TeCA in HCO3− suspensions. Suspensions of 5 mN HCO3− were largely free of detectable ICH over 30 days and exhibited comparable reactivity to ClO4−, SO42−, and Cl− systems. In contrast, extensive ICH formation over two weeks in 25 mN suspensions coincides with reactivity loss toward 1,1,1,2-TeCA. Given the manner in which ICH grows on and around NZVI particle aggregates, a probable mechanism for reactivity loss involves the relatively large ICH crystals blocking access to reactive sites on the NZVI surface. Influence of Corrosion Product Formation on Longevity toward Cr(VI). Corrosion product formation cannot sufficiently explain the trends in Cr(VI) removal capacity observed for aged NZVI. In HCO3− systems, for example, the time scales and nature of product formation were independent of pH (SI Figure S8), yet Cr(VI) removal capacity was almost 2-fold greater at pH 7 than pH 8 (Figure 2). Also, despite obvious differences in the rate and extent of ICH formation in 5 and 25 mN HCO3− suspensions (Figure 3 and 8370

dx.doi.org/10.1021/es301753u | Environ. Sci. Technol. 2012, 46, 8365−8373

Environmental Science & Technology

Article

S10b). In 25 mN NO3− systems, aqueous Fe(II) was below detection in all samples. As discussed in the SI, we attribute this behavior to the oxidation of Fe(II) by either NO3− or its reduction products (e.g., nitrite).51 Ignoring nitrate-containing suspensions, the linear relationship in Figure 4 appears independent of anion identity and concentration (Figure 4a), as data from all other anion systems are distributed over the range of Cr(VI) removal capacities and Fe(II) production rates measured. This is also the case when data are categorized based upon extent of aging (SI Figure S11), which also yields a random distribution. In contrast, pH is the key suspension variable governing Cr(VI) removal; categorizing Cr(VI) removal capacities by suspension pH (Figure 4b) produces clear clusters with respect to Fe(II) production. Low pH values correspond to both the highest rates of Fe(II) production and the greatest Cr(VI) removal capacities, and vice versa. Additional Cr(VI) removal studies spanning a broader range of pH values (suspensions of Cl− and SO42− aged over 7 days at pH 6 and 9) also adhered to this relationship. Thus, regardless of anion identity, anion concentration, and the duration of aging, the extent of Cr(VI) removal by NZVI can be predicted entirely from its Fe(II) production rate, a metric of Fe(0) corrosion. Using all Cr(VI) removal capacity data corresponding to nonzero Fe(II) production rates (i.e., excluding NO3 − suspensions and additional experiments conducted at pH 9.0, for which soluble Fe(II) was below detection), an empirical relationship between these variables can be developed via linear regression (Figure 4b). This best-fit regression line has a nonzero y-axis intercept (27 ± 3 mg Cr(VI)/g NZVI), which corresponds to the Cr(VI) removal capacity when Fe(II) production is not appreciable. The y-axis intercept from the linear regression agrees well with the average Cr(VI) removal (22 ± 7 mg Cr(VI)/g NZVI) experimentally measured in suspensions for which Fe(II) production rates were not quantifiable (n = 33). In an early investigation of Cr(VI) reaction with granular iron, Powell et al.52 proposed that chromate (CrO42−) reduction occurred primarily via Fe(II) generated at anodic sites on the granular iron surface, while also noting the possibility for direct electron transfer to CrO42− from cathodic Fe(0) sites. We contend that the correlation in Figure 4 supports such a scenario. Specifically, we hypothesize that the yaxis intercept value reflects the extent of Cr(VI) reduction via electron transfer from cathodic Fe(0) sites initially exposed at or near the NZVI particle-solution interface. When exposed to solution at corrosion pits, cracks or defects in the surface oxide layer, Fe(0) can either reduce water (protons) or transfer electrons directly to CrO42−. At higher pH values (pH > 8.0) where water reduction (and Fe(II) dissolution) slows, CrO42− would be expected to receive the majority of reducing equivalents, but subsequent precipitation of Cr(III)−Fe(III) hydroxides53 would ultimately block these sites and prevent further Cr(VI) removal. In contrast, at lower pH values (pH < 8.0), higher corrosions rates would produce larger quantities of ferrous iron species at anodic sites on the NZVI surface. Because soluble Fe(II) was never measurable in the presence of Cr(VI), these reactive forms of ferrous iron must exist entirely in the near-surface region of the NZVI particle, as was originally proposed by Powell et al.52 Although the exact nature of these ferrous iron species is not known (e.g., Fe(II)-containing solids, adsorbed Fe(II) or dissolved Fe(II) that reacts before it can diffuse into bulk solution), they appear to be the dominant

entity responsible for Cr(VI) reduction in NZVI suspensions below pH 8. Finally, kobs(1,1,1,2-TeCA) values exhibit no dependence on Fe(II) production rate (SI Figure S12). Thus, we propose that Cr(VI) reduction is largely dominated by near-surface Fe(II) generated via NZVI corrosion, whereas 1,1,1,2-TeCA reduction is controlled by the availability of Fe(0) on the NZVI particle surface. This scenario would explain why Cl−, SO42−, ClO4−, and low concentrations of HCO3− all exhibited essentially equivalent reactivity toward 1,1,1,2-TeCA but divergent reactivity toward Cr(VI). Cr(VI) removal will be sensitive to the quantity and nature of Fe(II) species produced on the NZVI surface during aging, while 1,1,1,2-TeCA reduction will only be affected by the ability of these phases to block Fe(0) surface sites. The latter scenario is consistent with increases in kobs(1,1,1,2-TeCA) values observed in response to autoreduction, during which thinning of the overlying surface oxide would make reducing equivalents from Fe(0) more accessible to solution, whereas Cr(VI) removal was not affected. Environmental Implications. Evidence herein suggests that 1,1,1,2-TeCA and Cr(VI) are reduced by different reactive entities in NZVI suspensions (i.e., solution accessible Fe(0) and surface-associated Fe(II), respectively), and these reactive entities display different time scales and mechanisms of reactivity loss in response to pH and anionic cosolutes during aging. Consequently, it is difficult, if not impossible, to develop broadly generalizable trends in NZVI longevity applicable across contaminant classes and over a range of geochemical conditions. We also show that the influence of specific corrosion products on NZVI longevity is equally difficult to predict; formation of ICH completely inhibits NZVI reactivity toward 1,1,1,2-TeCA but exerts little influence on Cr(VI) removal capacity. Perhaps the only generalization safely made is that cosolutes capable of accepting reducing equivalents from Fe(0) (e.g., NO3−) will dampen reactivity and shorten longevity more so than species that simply alter the rate and products of NZVI corrosion. Due to concerns over material cost and potentially adverse consequences of engineered nanomaterials in the environment, NZVI should only be used when it holds an obvious advantage over more traditional Fe(0) technologies. Given its considerably higher rate of corrosion, NZVI is by far the best treatment strategy for pollutants susceptible to reduction not only by Fe(0) but also by forms of ferrous iron (e.g., Cr(VI) and nitroaromatics). We caution that we have largely ignored the role that insoluble Cr(III)−Fe(III) phases generated via Cr(VI) reduction play in limiting NZVI longevity. Our aging protocol, which simulates reactivity loss arising from water reduction, is not meant to imply that NZVI treatment zones will stay active toward Cr(VI) over several months. Rather, treatment zone lifetime will be heavily influenced, if not entirely dictated, by the formation of insoluble products of Cr(VI) reduction and Fe(II) oxidation, the extent of which will scale with available Cr(VI) concentration. A significant outcome of this work is the correlation linking Fe(II) production rate in aged suspensions with Cr(VI) removal (Figure 4), which merits further exploration with other commercially available forms of nanoscale and granular Fe(0). If observed for other Fe(0) reductants, it should prove practically useful for identifying materials best-suited for application at Cr(VI) contaminated sites. The ability to predict treatment efficiency from a relatively easily measurable property such as Fe(II) production rate may also help guide the 8371

dx.doi.org/10.1021/es301753u | Environ. Sci. Technol. 2012, 46, 8365−8373

Environmental Science & Technology

Article

(15) Gotpagar, J.; Lyuksyutov, S.; Cohn, R.; Grulke, E.; Bhattacharyya, D. Reductive Dehalogenation of trichloroethylene with zero-valent iron: Surface profiling microscopy and rate enhancement studies. Langmuir 1999, 15, 8412−8420. (16) Hao, Z.-W.; Xu, X.-H.; Jin, J.; He, P.; Liu, Y.; Wang, D.-H. Simultaneous removal of nitrate and heavy metals by iron metal. J. Zhejiang Univ. Sci. 2005, 6, 307−310. (17) Ritter, K.; Odziemkowski, M. S.; Simpgraga, R.; Gillham, R. W.; Irish, D. E. An in situ study of the effect of nitrate on the reduction of trichloroethylene by granular iron. J. Contam. Hydrol. 2003, 65, 121− 136. (18) Schlicker, O.; Ebert, M.; Fruth, M.; Weidner, M.; Wilst, W.; Dahmke, A. Degradation of TCE with iron: The role of competing chromate and nitrate reduction. Ground Water 2000, 38 (3), 403−409. (19) Klausen, J.; Ranke, J.; Schwarzenbach, R. P. Influence of solution composition and column aging on the reduction of nitroaromatic compounds by zerovalent iron. Chemosphere 2001, 44, 511−517. (20) Lipczynskakochany, E.; Harms, S.; Milburn, R.; Sprah, G.; Nadarajah, N. Degradation of carbon tetrachloride in the presence of iron- and sulfur-containing compounds. Chemosphere 1994, 29, 1477− 1489. (21) MacDougall, B.; Graham, M. J. In Corrosion Mechanisms in Theory and Practice; Marcus, P., Oudar, J., Eds.; Marcel Dekker: New York, 1995; pp 143−173. (22) Agrawal, A.; Tratnyek, P. G. Reduction of nitro aromatic compounds by zero-valent iron metal. Environ. Sci. Technol. 1995, 30, 153−160. (23) Mackenzie, P. D.; Horney, D. P.; Sivavec, T. M. Mineral precipitation and porosity losses in granular iron columns. J. Hazard. Mater. 1999, 68, 1−17. (24) Roh, Y.; Lee, S. Y.; Elless, M. P. Characterization of corrosion products in the permeable reactive barriers. Environ. Geol. 2000, 40, 184−194. (25) Phillips, D. H.; Gu, B.; Watson, D. B.; Roh, Y.; Liang, L.; Lee, S. Y. Performance evaluation of a zerovalent iron reactive barrier: Mineralogical characteristics. Environ. Sci. Technol. 2000, 34, 4169− 4176. (26) Agrawal, A.; Ferguson, W. J.; Gardner, B. O.; Christ, J. A.; Bandstra, J. Z.; Tratnyek, P. G. Effects of carbonate species on the kinetics of dechlorination of 1,1,1-trichloroethane by zero-valent iron. Environ. Sci. Technol. 2002, 36 (20), 4326−4333. (27) Klausen, J.; Vikesland, P. J.; Kohn, T.; Burris, D. R.; Ball, W. P.; Roberts, A. L. Longevity of granular iron in groundwater treatment processes: Solution composition effects on reduction of organohalides and nitroaromatic compounds. Environ. Sci. Technol. 2003, 37, 1208− 1218. (28) Jeen, S. W.; Gillham, R. W.; Blowes, D. W. Effects of carbonate precipitates on long-term performance of granular iron for reductive dechlorination of TCE. Environ. Sci. Technol. 2006, 40, 6432−6437. (29) Jeen, S.-W.; Jambor, J. L.; Blowes, D. W.; Gillham, R. W. Precipitates on granular iron in solutions containing calcium carbonate with trichloroethene and hexavalent chromium. Environ. Sci. Technol. 2007, 41, 1989−1994. (30) Bi, E.; Bowen, I.; Devlin, J. F. Effect of mixed anions (HCO3−, SO42−, ClO4−) on granular iron (Fe0) reactivity. Environ. Sci. Technol. 2009, 43, 5975−5981. (31) Kohn, T.; Livi, K. J. T.; Roberts, A. L.; Vikesland, P. J. Longevity of granular iron in groundwater treatment processes: Corrosion product development. Environ. Sci. Technol. 2005, 39, 2867−2879. (32) Alowitz, M. J.; Scherer, M. M. Kinetics of nitrate, nitrite and Cr(VI) reduction by iron metal. Environ. Sci. Technol. 2002, 36, 299− 306. (33) Westerhoff, P. Reduction of nitrate, bromate, and chlorate by zerovalent iron (Fe-0). J. Environ. Eng 2003, 129, 10−16. (34) Reardon, E. J.; Fagan, R.; Vogan, J. L.; Przepiora, A. Anaerobic corrosion reaction kinetics of nanosized iron. Environ. Sci. Technol. 2008, 42, 2420−2424.

development of more reactive iron-based reductants than current commercially available formulations.

■

ASSOCIATED CONTENT

S Supporting Information *

Method and reagents details, and additional reactivity and characterization data as a function of aging time. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*Phone: 319.335.1401; fax: 319.335.5660; e-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS We acknowledge the Central Facility for Advanced Microscopy and Microanalysis (CFAMM) at UC Riverside for assistance with TEM and Dr. Pingyung Feng for assistance with XRD. We also thank the three anonymous reviewers whose recommendations greatly improved the quality of this work.

■

REFERENCES

(1) Li, L.; Fan, M.; Brown, R. C.; Van Leeuwn, J.; Wang, J.; Wang, W.; Song, Y.; Zhang, P. Synthesis, properties, and environmental applications of nanoscale iron-based materials: A review. Crit. Rev. Env. Sci. Tec. 2006, 36, 405−431. (2) Li, X.; Elliot, D. W.; Zhang, W.-X. Zero-valent iron nanoparticles for abatement of environmental pollutants: Materials and engineering aspects. Crit. Rev. Solid State 2006, 31, 111−122. (3) Tratnyek, P. G.; Johnson, R. L. Nanotechnologies for environmental cleanup. Nano Today 2006, 1, 44−48. (4) Zhang, W.-X. Nanoscale iron particles for environmental remediation: An overview. J. Nanopart. Res. 2003, 5, 323−332. (5) Elliot, D. W.; Zhang, W.-X. Field assessment of nanoscale bimetallic particles for groundwater treatment. Environ. Sci. Technol. 2001, 35, 4922−4926. (6) Henn, K. W.; Waddill, D. W. Utilization of nanoscale zero-valent iron for source remediationA case study. Rem. J. 2006, 16, 57−77. (7) Wei, Y.-T.; Wu, S.-C.; Chou, C.-M.; Che, C.-H.; Tsai, S.-M.; Lien, H.-L. Influence of nanoscale zero-valent iron on geochemical properties of groundwater and vinyl chloride degradation: A field case study. Water Res. 2010, 44, 131−140. (8) He, F.; Zhao, D.; Paul, C. Field assessement of carboxymethyl cellulose stabilized iron nanoparticles for in situ destruction of chlorinated solvents in source zones. Water Res. 2010, 44, 2360−2370. (9) Liu, Y.; Lowry, G. V. Effect of particle age (Fe0 content) and solution pH on NZVI reactivity: H2 evolution and TCE dechlorination. Environ. Sci. Technol. 2006, 40, 6085−6090. (10) Sarathy, V.; Tratnyek, P. G.; Nurmi, J. T.; Baer, D. R.; Amonette, J. E.; Chun, C. L.; Penn, R. L.; Reardon, E. J. Aging of iron nanoparticles in aqueous solution: Effects on structure and reactivity. J. Phys. Chem. C 2008, 112, 2286−2293. (11) Xie, Y.; Cwiertny, D. M. Use of dithionite to extend the reactive lifetime of nanoscale zero-valent iron treatment systems. Environ. Sci. Technol. 2010, 44, 8649−8655. (12) Henderson, A. D.; Demond, A. H. Long-term performance of zero-valent iron permeable reactive barriers: A critical review. Environ. Eng. Sci. 2007, 24 (4), 401−423. (13) Devlin, J. F.; Allin, K. O. Major anion effects on the kinetics and reactivity of granular iron in glass-encased magnet batch reactor experiments. Environ. Sci. Technol. 2005, 39, 1868−1874. (14) Cheng, I. F.; Muftikian, R.; Fernando, Q.; Korte, N. Reduction of nitrate to ammonia by zero-valent iron. Chemosphere 1997, 35, 2689−2695. 8372

dx.doi.org/10.1021/es301753u | Environ. Sci. Technol. 2012, 46, 8365−8373

Environmental Science & Technology

Article

(35) Reardon, E. J. Zerovalent irons: Styles of corrosion and inorganic control on hydrogen pressure buildup. Environ. Sci. Technol. 2005, 39, 7311−7317. (36) Reardon, E. J. Anaerobic corrosion of granular iron: Measurement and interpretation of hydrogen evolution rates. Environ. Sci. Technol. 1995, 29, 2936−2945. (37) Nurmi, J. T.; Tratnyek, P. G.; Sarathy, V.; Baer, D. R.; Amonette, J. E.; Pecher, K.; Wang, C.; Linehan, J. C.; Matson, D. W.; Penn, R. L.; Driessen, M. D. Characterization and properties of metallic iron nanoparticles: Spectroscopy, electrochemistry, and kinetics. Environ. Sci. Technol. 2005, 39, 1221−1230. (38) Yan, W.; Herzing, A. A.; Kiely, C. J.; Zhang, W.-X. Nanoscale zero-valent iron (nZVI): Aspects of the core-shell structure and reactions with inorganic species in water. J. Contam. Hydrol. 2010, 118, 96−104. (39) Liu, Y.; Phenrat, T.; Lowry, G. V. Effect of TCE concentration and dissolved groundwater solutes on NZVI-promoted TCE dechlorination and H2 evolution. Environ. Sci. Technol. 2007, 41, 7881−7887. (40) Reinsch, B. C.; Forsberg, B.; Penn, R. L.; Kim, C. S.; Lowry, G. V. Chemical transformations during aging of zerovalent iron nanoparticles in the presence of common groundwater dissolved constituents. Environ. Sci. Technol. 2010, 44, 3455−3461. (41) Cao, J.; Elliott, D. W.; Zhang, W.-X. Perchlorate reduction by nanoscale iron particles. J. Nanopart. Res. 2005, 7, 499−506. (42) Moore, A. M.; De Leon, C. H.; Young, T. M. Rate and extent of aqueous perchlorate removal by iron surfaces. Environ. Sci. Technol. 2003, 37, 3189−3198. (43) Stucki, J. W. The quantitative assay of minerals for Fe2+ and Fe3+ using 1,10-phenanthroline. II. A photochemical method. Soil Sci. Soc. Am. J. 1981, 45, 638−641. (44) Stucki, J. W.; Anderson, W. L. The quantitative assay of minerals for Fe2+ and Fe3+ using 1,10-phenanthroline. I. Sources of variability. Soil Sci. Soc. Am. J. 1981, 45, 633−637. (45) Bose, M. Mechanism of the reaction between dichromate and diphenylcarbazide. Nature 1952, 170, 213. (46) Cwiertny, D. M.; Roberts, A. L. On the nonlinear relationship between kobs and reductant mass loading in iron batch systems. Environ. Sci. Technol. 2005, 39, 8948−8957. (47) Cornell, R. M.; Schwertmann, U. The Iron Oxides: Structure, Properties, Reactions, Occurrences and Uses, 2nd ed.; Wiley-VCH: Weinheim, Germany, 2003; p 664. (48) Wilkin, R. T.; Puls, R. W. Capstone Report on the Application, Monitoring and Performacne of Permeable Reactive Barriers for Ground Water Remediation; U.S. Environmental Protection Agency, Office of Research and Development: Cincinnati, OH, 2003. (49) Fendorf, S. E.; Li, G. Kinetics of chromate reduction by ferrous iron. Environ. Sci. Technol. 1996, 30, 1614−1617. (50) Sedlak, D. L.; Chan, P. G. Reduction of hexavalent chromium by ferrous iron. Geochim. Cosmochim. Acta 1997, 61, 2185−2192. (51) Fanning, J. C. The chemical reduction of nitrate in aqueous solution. Coord. Chem. Rev. 2000, 199, 159−179. (52) Powell, R. M.; Puls, R. W.; Hightower, S. K.; Sabatini, D. A. Coupled iron corrosion and chromate reduction: Mecahsnism for subsurface remediation. Environ. Sci. Technol. 1995, 29, 1913−1922. (53) Li, X.-Q.; Cao, J.; Zhang, W.-X. Stoichiometry of Cr(VI) immobilization using nanoscale zerovalent iron (nZVI): A study with high-resolution X-ray photoelectron spectroscopy (HR-XPS). Ind. Eng. Chem. Res. 2008, 47, 2131−2139.

8373

dx.doi.org/10.1021/es301753u | Environ. Sci. Technol. 2012, 46, 8365−8373