Article pubs.acs.org/est
Dual Mechanism Conceptual Model for Cr Isotope Fractionation during Reduction by Zerovalent Iron under Saturated Flow Conditions Julia H. Jamieson-Hanes,† Richard T. Amos,‡ David W. Blowes,*,† and Carol J. Ptacek† †
Department of Earth and Environmental Sciences, University of Waterloo, 200 University Avenue West, Waterloo, ON N2L 3G1, Canada ‡ Department of Earth Sciences, Carleton University, 1125 Colonel By Drive, Ottawa, ON K1S 5B6, Canada S Supporting Information *
ABSTRACT: Chromium isotope analysis is rapidly becoming a valuable complementary tool for tracking Cr(VI) treatment in groundwater. Evaluation of various treatment materials has demonstrated that the degree of isotope fractionation is a function of the reaction mechanism, where reduction of Cr(VI) to Cr(III) induces the largest fractionation. However, it has also been observed that uniform flow conditions can contribute complexity to isotope measurements. Here, laboratory batch and column experiments were conducted to assess Cr isotope fractionation during Cr(VI) reduction by zerovalent iron under both static and saturated flow conditions. Isotope measurements were accompanied by traditional aqueous geochemical measurements (pH, Eh, concentrations) and solid-phase analysis by scanning electron microscopy and X-ray absorption spectroscopy. Increasing δ53Cr values were associated with decreasing Cr(VI) concentrations, which indicates reduction; solid-phase analysis showed an accumulation of Cr(III) on the iron. Reactive transport modeling implemented a dual mechanism approach to simulate the fractionation observed in the experiments. The faster heterogeneous reaction pathway was associated with minimal fractionation (ε = −0.2‰), while the slower homogeneous pathway exhibited a greater degree of fractionation (ε = −0.9‰ for the batch experiment, and ε = −1.5‰ for the column experiment).
■
sulfides, green rust, and siderite.12 The degree of fractionation varies with the type of reactive material, which may indicate that the change in Cr isotope signature is determined by the mechanism of Cr(VI) removal. The majority of the published Cr isotope literature has focused on characterizing Cr isotope fractionation under static conditions.4,5,8−10,12−15 A number of studies have also observed changes in δ53Cr values at various field sites;4,14,16−22 however, very little work has been done to understand the potential influence of transport on Cr isotope trends. Previous work by the authors of this study has employed column and flowthrough cell experiments to explore Cr isotope behavior under uniform flow conditions and to compare the results with static batch experiments.6,23,24 These flow experiments are designed to provide insight into both the temporal and spatial evolution of Cr isotope ratios while also allowing enough mass to be recovered for spectroscopic analyses.
INTRODUCTION Chromium (Cr) contamination of groundwater is a widespread concern and often results from industrial activities such as leather tanning and electroplating.1 In its oxidized form, Cr(VI) is highly toxic and mobile, occurring as HCrO4−, CrO42−, and Cr2O72− oxyanions.2 Remediation techniques traditionally focus on reduction to Cr(III), as this form has lower toxicity and is sparingly soluble, which thus effectively removes the Cr from solution.3 Recently, Cr isotope ratio measurements have emerged as an additional tool to track Cr(VI) migration in groundwater. Chromium has four stable isotopes, 50Cr, 52Cr, 53Cr, and 54Cr, with natural abundances of 4.35%, 83.8%, 9.5%, and 2.37%. Redox processes have been shown to induce significant isotope fractionation, where the lighter isotopes are preferentially reacted and the remaining Cr(VI) pool becomes enriched in 53 Cr relative to 52Cr.4 This enrichment is expressed as a positive shift in δ53Cr in units of per mil (‰). Many of the common electron donors used for Cr(VI) treatment have been evaluated for Cr isotope fractionation, including organic carbon,5−7 aqueous Fe(II),8 certain bacteria,9−11 and permeable reactive barrier (PRB) materials such as Fe(II)-doped goethite, iron © XXXX American Chemical Society
Received: December 21, 2014 Revised: April 2, 2015 Accepted: April 3, 2015
A
DOI: 10.1021/es506223a Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
disassembled within the glovebox to provide reactive material for the batch experiment. A 50 mg L−1 Cr(VI) input solution was prepared by dissolving K2Cr2O7 (99+% Baker Analyzed ACS Reagent) in CaCO3-saturated high-purity water and then purging overnight with Ar. This Cr(VI) concentration was selected to be relevant to documented incidents of groundwater contamination, where concentrations as high as 14 600 mg L−1 have been reported.3 The batch experiment was initialized by combining 6 g of aged ZVI and 150 mL of the Cr(VI) input solution in a series of 250 mL amber glass bottles in the anaerobic glovebox. The bottles were hand-tumbled twice per day to ensure good contact between the solids and the solution. Input solution composed of 50 mg L−1 Cr(VI) in CaCO3saturated water was pumped in an upward direction through the 22 cm long column containing aged ZVI. A flow rate of approximately 4.2 pore volumes (PVs) per day was employed. Input and effluent samples were collected twice per week over the duration of the experiment. Profile sampling was conducted using the six sampling ports located along the length of the column, beginning at the top of the column to avoid disturbing the flow field. Four full profiles were collected after 33 days, 41 days, 55 days, and 76 days. Aqueous Sampling. All sampling of the batch experiment was conducted within the anaerobic glovebox. Bottles were sacrificially sampled at each time point; at seven of the 14 sampling times, two bottles were sampled simultaneously to provide information regarding experimental reproducibility. An unfiltered water sample was collected for immediate pH and redox measurement using an Orion Ross 815700 pH electrode (Thermo Scientific, Waltham, MA) and an Orion 9678 Eh electrode (Thermo Scientific). The performance of the electrodes was checked using pH 4, 7, and 10 buffers and ZoBell’s35 and Light’s36 solutions prior to sampling. The remaining solution was vacuum-filtered through 0.45 μm cellulose acetate filters (Whatman, UK) to halt the reaction. Alkalinity measurements were performed on a filtered sample (0.2 μm Supor membrane, Acrodisc, Pall, UK) using phenolthalein and bromocresol green-methyl red indicators and titrating to the end point with 0.16 N H2SO4. Remaining samples were filtered at 0.2 μm and collected for anions, cations, Cr(VI), and Cr isotope measurements. All aqueous sampling of the column experiment was carried out in the same manner as the batch experiment. Cr(VI) concentrations were measured immediately following sampling on a Hach DR/2010 spectrophotometer at 540 nm using the 1,5-diphenylcarbohydrazide method. Inorganic anion concentrations were determined by ion chromatography (Dionex DX 600). Samples acidified to pH < 2 with tracemetal grade HNO3 were used for the determination of cation concentrations by inductively coupled plasma optical emission spectrometry (ICP-OES; Thermo Scientific iCAP 6500) and inductively coupled plasma mass spectrometry (ICP-MS; Thermo Scientific XSeries 2). Cr Isotope Measurements. Sample preparation and Cr isotope analyses were conducted following the techniques described in Jamieson-Hanes et al.6 Briefly, the acidified samples were combined with a 50Cr−54Cr double spike prior to oxidation with ammonium persulfate. The oxidized spikesample mixtures were retained on Bio Rad AG 1-X8 anion exchange resin that had been conditioned with trace metal grade HNO3, while the impurities were flushed through.
Reactive transport modeling is often used to provide insight into geochemical interactions in experimental systems. Isotope fractionation has already been implemented into several geochemical models to simulate a variety of isotope systems including carbon,25−30 chlorine, and sulfur.31 More recently, reactive transport modeling of Cr isotope fractionation has been performed on samples collected from field sites21,32,33 and from laboratory experiments.23,24,34 Jamieson-Hanes et al.6 conducted complementary batch and column experiments using organic carbon (OC) from leaf mulch to reduce Cr(VI). This study found that although the isotope data from the batch experiment followed the expected Rayleigh-type curve, the behavior of the Cr isotope fractionation differed under continuous flow conditions. The reactive transport model MIN3P was employed to simulate Cr isotope fractionation in these OC batch and column experiments.23 The model-based analysis provided an alternative hypothesis to the original conclusion derived from interpretation of the experimental results regarding the mechanism of Cr(VI) removal and the influence of continuous flow conditions on Cr isotope fractionation. Jamieson-Hanes et al.24 examined isotope fractionation during Cr(VI) reduction by zerovalent iron (ZVI) in a flow-through cell experiment. Similar to the OC column simulations, a dual mechanism approach was used to model the complex isotope behavior observed in these flow-through cell experiments. However, no comparison was made with a ZVI system under static conditions. In this study, a batch experiment was conducted using granular ZVI to characterize Cr isotope fractionation under static conditions. A complementary column experiment was conducted simultaneously with the batch experiment. The objective of the present study was to further investigate the complex isotope behavior observed in Jamieson-Hanes et al.24 while providing a direct comparison between static and continuous flow conditions during Cr(VI) reduction by ZVI. Aqueous samples were collected for analysis of water chemistry and Cr isotope ratios, while solid-phase analyses were performed on the reactive material at the conclusion of the experiments. MIN3P was used to simulate Cr isotope fractionation in both the batch and column experiments by following the dual mechanism approach in Jamieson-Hanes et al.24 The analytical and numerical modeling techniques were combined to evaluate the isotope fractionation effects associated with static and flow conditions and to further elucidate the cause of this behavior.
■
MATERIALS AND METHODS Experimental Setup. Granular ZVI (Connelly GMP, Chicago, IL) was used as the reactive material for both the batch and column experiments. The ZVI was first sieved at 60− 16 mesh (0.4−1.6 mm) and then soaked in 1 M HCl for several hours to remove the majority of the oxidized coatings. The material was rinsed with Ar-purged high-purity water and vacuum filtered in an anaerobic glovebox (Coy Laboratory Products, Inc., Grass Lake, MI) under a 5% H2/balance N2 atmosphere. Two Plexiglas columns, each measuring 22 × 5 cm2, were packed with the acid-cleaned ZVI. Argon-purged deionized water was pumped upward through both columns for 2 weeks, followed by Ar-purged CaCO3-saturated water for 4 weeks, to simulate the aging of a PRB. After reducing conditions had developed in the columns, as confirmed by redox measurements of the effluent, one column was B
DOI: 10.1021/es506223a Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
transport model MIN3P38 with modification to simulate isotope fractionation.23,31 Jamieson-Hanes et al.24 simulated Cr(VI) reduction by ZVI in flow-through cell experiments consistent with Gheju39 who reported both heterogeneous (eqs 2 and 3) and homogeneous (eqs 8 and 9) reduction of Cr(VI). In this study, the same conceptual model is used to be consistent with the previous work. Specifically, the reduction of the 52Cr(VI) and 53Cr(VI) isotopologues can occur by the Fe0:
Following reduction by 2 N HNO3, the purified samples were eluted and prepared for analysis. Cr isotope measurements were performed on a multicollector ICP-MS (Thermo Scientific Neptune), which allowed the simultaneous measurement of all four Cr isotopes, 50Cr, 52 Cr, 53Cr, and 54Cr, along with 49Ti, 50V, and 56Fe to correct for isobaric interferences. Polyatomic interferences of 40Ar14N on 54Cr and 40Ar16O on 56Fe were minimized by off-center peak analysis in medium resolution mode. Instrumental mass bias was corrected using a double-nested iterative routine, and the true isotopic compositions of the samples were expressed as δ53Cr in per mil (‰) relative to the NIST SRM 979 Cr isotope standard, where
k52
0.667 52CrO4 2 − + Fe0 + 4H+ ⎯→ ⎯ Fe2 + + 0.667 52Cr(OH)2+ + H 2O
(2) k53
0.667 53CrO4 2 − + Fe0 + 4H+ ⎯→ ⎯ Fe2 + + 0.667 53Cr(OH)2+ + H 2O
⎡ ( Cr/ Cr) ⎤ sample − ( Cr/ Cr)standard ⎥ × 1000 δ 53Cr = ⎢ ⎢⎣ ⎥⎦ (53Cr/ 52Cr)standard 53
52
53
52
(3)
where the rate of each reaction is a function of the Cr(VI) concentrations, the volume fraction of reactive ZVI, the surface area of ZVI, and the passivation by secondary minerals aragonite and ferrihydrite as described by Jeen et al.:40
(1)
Two aliquots of SRM 979 were prepared with each set of samples to assess the success of the purification and the MCICP-MS analysis. Uncertainty of the isotope measurements was calculated to be ±0.08‰ based on two times the root−mean− square difference of six unknown samples prepared in duplicate (one duplicate pair per batch of prepared samples) following the approach of Basu and Johnson.12 Solid-Phase Cr Characterization. Solid material from the batch experiment was vacuum-filtered in the glovebox, sealed in polyethylene (PE) bottles in a portable anaerobic chamber, frozen, and freeze-dried prior to analysis. Samples of column solids were collected from six locations along the length of the column under anaerobic conditions and were frozen and freezedried in the same manner as the batch samples. Secondary precipitates were examined on several samples using field emission-scanning electron microscopy (FE-SEM; Leo1530, Carl Zeiss SMT GmbH, Germany) with energy dispersive spectroscopy (EDS; EDAX Pegasus 1200, AMETEX Inc.). The samples were mounted on Al stubs with C tape and coated with a 10−12 nm thick Au layer to ensure conductance. An accelerating potential of 15 kV was used for backscatter electron (BSE) imaging and collection of semiquantitative EDS spectra. Synchrotron-based X-ray absorption near edge structure (XANES) spectroscopy was performed at the GSECARS beamline 13-BM-D at the Advanced Photon Source, Argonne National Laboratory (Argonne, IL). Samples from each location along the length of the column and from various times through the batch experiment were mounted separately in 1 mm thick Al sample holders between two layers of Kapton tape. Finely ground reference materials were spread on polyethylene terephthalate (PET) tape (Scotch Magic Tape, 3M, St. Paul, MN) and layered to achieve a thickness of 300− 500 μm. All samples and standards were analyzed at the Cr Kedge (5989 eV) in bulk mode with an unfocused incident beam (∼0.5 × 3 mm2). Spectra for unknown samples were collected in fluorescence mode using a four-element Si drift detector (Vortex ME-4, SII NanoTechnology USA Inc., Northridge, CA), whereas spectra for reference materials were collected in transmission mode. Smoothing and linear combination fitting of XANES data were carried out using the program ATHENA, which is a component of the IFEFFIT software package.37 Reactive Transport Modeling. Aqueous geochemistry and Cr isotope results from both the batch and column experiments were simulated using the multicomponent reactive
RFe0 − 52Cr(VI) = k52S0 exp(αferriφferri(x , t )αaragφarag (x , t )) {52CrO4 2 −}{φFe0 − Cr(VI)(x , t )}
(4)
where S0 is the initial reactive surface area of the iron (m2 iron L−1 bulk), αi is the proportionality constant for mineral phase i, and φi is the volume fraction of mineral phase i at a specific location and time (m3 mineral m−3 bulk). An analogous rate expression is derived for the 53Cr(VI) isotopologue (see Supporting Information, Table S1), and the rate parameters k52 and k53 are a function of the overall reaction rate constant (keff): keff = k52 + k53
(5)
and the isotope fractionation factor (α) k53 = α53R 53 k52
(6)
where R53 is the ratio of 53Cr(VI) to 52Cr(VI) in solution. For convenience, the extent of fractionation is expressed in terms of isotope fractionation using the epsilon notation (ε) according to the following equation: ε (‰) = (α − 1) × 1000
(7)
Similarly, Cr(VI) reduction can occur by aqueous ferrous iron CrO4 2 − + 3Fe2 + + 6H+ →
52
Cr(OH)2+ + 3Fe3 + + 2H 2O
(8)
CrO4 2 − + 3Fe2 + + 6H+ → 53Cr(OH)2+ + 3Fe3 + + 2H 2O
(9)
52
53
where the ferrous iron is produced through eq 2 or 3, or following the reduction of water by Fe0: Fe0(s) + 2H+ → Fe2 + + H 2(aq)
(10)
The rate of Cr(VI) reduction by ferrous iron is simulated to be dependent on the aqueous concentrations of Cr(VI) and ferrous iron, with eqs 5 and 6 describing the relationship between the overall reaction rate, the ratio of the Cr(VI) isotopologues, and the fractionation factor (see Table S1 for reaction stoichiometry, reaction rate expressions, and model parameters). The rate of ZVI oxidation by water is dependent on the surface area of ZVI and the passivation by secondary minerals aragonite and ferrihydrite (Table S1, Supporting Information). C
DOI: 10.1021/es506223a Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology The rate of Cr(VI) removal has been reported to depend on Cr(VI) concentrations, reactive surface area, and also pH (e.g., Gould41). However, the rate has been often reported as first order with no dependence on pH (Gheju39 and references therein). Here we simulate the rate to be independent of pH to simplify the calibration of the model given the complexity and coupled nature of the geochemical system. In the simulations, the reaction of Cr(VI) with ZVI is considered to be dependent on the reactive ZVI volume fraction, which is a fraction of the total ZVI volume. In contrast, reaction of water with ZVI is considered to be dependent on the total ZVI volume fraction. Conceptually, these assumptions suggest that Cr(VI) reduction by ZVI is dependent on specific surface chemistry on the ZVI, although the modeling cannot distinguish the exact nature of this mechanism. In addition to Cr(VI) reduction, the simulated reaction network includes the precipitation of Cr hydroxide and the precipitation or dissolution of aragonite, ferrihydrite, ferrous hydroxide, and ferrous hydroxycarbonate. Rate expressions and relevant parameters are summarized in Table S1 of the Supporting Information. Model input parameters are summarized in Table 1.
Figure 1. Batch experiment Cr(VI) concentrations and corresponding δ53Cr values for the first 2 days, after which time Cr(VI) concentrations were too low for isotope analysis. Error bars of δ53Cr values represent external reproducibility of 0.08‰. Simulation results are shown for both the Cr(VI) concentrations (solid line) and the δ53Cr values (dashed line).
Information). This trend is consistent with the corrosion of ZVI coupled with the reduction of H2O and Cr(VI), reactions which both consume H+.42,43 Alkalinity decreased from an initial value of 128 mg L−1 as CaCO3 to 58 mg L−1, which suggests precipitation of carbonate minerals such as calcite or ferrous hydroxycarbonate (Fe2(OH)2CO3). Formation of secondary carbonate minerals contributed to the decrease in the reactive surface area of the ZVI and thus influenced the rate of Cr(VI) removal.44 Finally, redox measurements indicated a relatively weak trend toward reducing conditions. Column Experiment. Input Cr(VI) concentrations averaged 44.7 ± 2.5 mg L−1 (2σ) over the course of the column experiment. Complete Cr treatment was observed in the effluent during the first 55 days of flow (230 PVs; Figure 2). Breakthrough began between 55 and 62 days (230 and 260 PVs). After the initial appearance of Cr, the effluent Cr concentration increased rapidly at first, followed by a slower increase. At the end of the experiment (126 days, or 530 PVs), the effluent Cr concentration was 73% of the input concentration. Profile samples collected at 33 days, 41 days, 55 days, and 76 days captured the progression of the Cr front along the length of the column, which demonstrates the loss of reactivity of the ZVI over time. Again, good agreement was found between the Cr(VI) and total Cr measurements, which suggests rapid removal of Cr(III) from solution (linear regression; slope = 1.00, R2 = 0.99). Effluent pH values exhibited first an increasing trend to a maximum value of 10.64, then a decreasing trend during Cr breakthrough; however, the effluent pH was consistently higher than the average input value of 8.16 ± 0.22 (Figure S2, Supporting Information). Despite variability in the input alkalinity, effluent alkalinity values remained fairly stable over the course of the experiment, averaging 51 ± 10 mg L−1 as CaCO3. Strongly reducing conditions were observed in the column effluent at the beginning of the experiment, with Eh values as low as −470 mV. As the reactivity of the ZVI became exhausted and the Cr concentration increased in the effluent, the Eh also increased up to a final value of 250 mV. Profile samples taken throughout the experiment also documented the overall geochemical evolution of the experimental system (Figure S3, Supporting Information).
Table 1. Reactive Transport Model Parameters physical parameters
batch experiment
column experiment
Darcy velocity (m s−1) porosity dispersivity (m s−1) 1-D model length (cm) discretization interval (cm) component concentrations
− 0.51 − − −
5.45 × 10−6 0.51 0.016 22 1 inflow boundary
initial condition batch
pH CrO42− (mol L−1) 53 CrO42− (mol L−1) 52 Cr(OH)2+ (mol L−1) 53 Cr(OH)2+ (mol L−1) Fe2+ (mol L−1) Fe3+ (mol L−1) K1+ (mol L−1) H2(aq) (mol L−1) CO32− (mol L−1) Ca2+ (mol L−1) 52
8.77 8.04 9.12 1.00 1.00 1.00 1.00 8.49 1.0 2.56 1.32
× × × × × × × × × ×
10−4 10−5 10−10 10−10 10−10 10−10 10−4 10−10 10−3 10−3
column 8.86 8.98 1.02 1.00 1.00 1.00 1.00 1.00 1.00 1.00 8.20
× × × × × × × × × ×
10−19 10−19 10−10 10−10 10−10 10−10 10−10 10−10 10−5 10−4
column 8.16 8.05 9.13 1.00 1.00 1.00 1.00 8.72 1.00 2.33 1.18
× × × × × × × × × ×
10−4 10−5 10−10 10−10 10−10 10−10 10−4 10−10 10−3a 10−3b
a Average value: 1.78 × 10−3−2.88 × 10−3. bAverage value: range 1.05 × 10−3−1.37 × 10−3.
■
RESULTS AND DISCUSSION Aqueous Results. Batch Experiment. Samples from the batch experiment exhibited a rapid decrease in Cr(VI) concentrations, with approximately 40% removal after the first hour (Figure 1). Slower rates were observed for the remainder of the experiment, and complete removal (
