Environ. Sci. Technol. 2008, 42, 4777–4783
Confounding Impacts of Iron Reduction on Arsenic Retention KATHARINE J. TUFANO AND SCOTT FENDORF* School of Earth Sciences, Stanford University, Stanford, California 94305
Received October 16, 2007. Revised manuscript received March 17, 2008. Accepted March 18, 2008.
A transition from oxidizing to reducing conditions has long been implicated to increase aqueous As concentrations, for which reductive dissolution of iron (hydr)oxides is commonly implicated as the primary culprit. Confounding our understanding of processes controlling As retention, however, is that reductive transformation of ferrihydrite has recently been shown to promote As retention rather than release. To resolve the role iron phases have in regulating arsenic concentrations, here we examine As desorption from ferrihydrite-coated sands presorbed with As(III); experiments were performed at circumneutral pH under Fe-reducing conditions with the dissimilatory iron reducing bacterium Shewanella putrefaciens strain CN-32 over extended time periods. We reveal that with the initial phase of iron reduction, ferrihydrite undergoes transformation to secondary phases and increases As(III) retention (relative to abiotic controls). However, with increased reaction time, cessation of the phase transitions and ensuing reductive dissolution result in prolonged release of As(III) to the aqueous phase. Our results suggest that As(III) retention during iron reduction is temporally dependent on secondary precipitation of iron phases; during transformation to secondary phases, particularly magnetite, As(III) retention is enhanced even relative to oxidized systems. However, conditions that retard secondary transformation (more stable iron oxides or limited iron reducing bacterial activity), or prolonged anaerobiosis, will lead to both the dissolution of ferric (hydr)oxides and release of As(III) to the aqueous phase.
Introduction Arsenic desorption from soil and sediment can pose a considerable human health risk through consumption of contaminated water. Arsenic naturally exists in a variety of oxidation states and in numerous organic and inorganic forms. Two valence states of arsenic, As(III) and As(V), predominate in aqueous systems (1), and both form complexes with many solids, inclusive of clay minerals, and iron, manganese, and aluminum oxides, in soils and sediments (2–6). While As(V) binds strongly to solids of most soils and sediments, As(III) adsorption is more specific, binding most appreciably to iron(III) (hydr)oxides (3). Despite being more specific, As(III) binds extensively to Fe(III) (hydr)oxides and, in fact, at circumneutral and higher pH adsorbs more extensively to ferric (hydr)oxides and magnetite than does As(V) under hydrostatic (batch) conditions (2, 7, 8). Furthermore, both As(III) and As(V) form bidentate, binuclear * Corresponding author phone: 650-723-5238; e-mail: fendorf@ stanford.edu. 10.1021/es702625e CCC: $40.75
Published on Web 05/29/2008
2008 American Chemical Society
complexes on iron (hydr)oxides (9, 10). However, multiple complexes are present and a large proportion of As(III) appears to be a labile, outer-sphere complex (6, 11). Adsorption on iron (hydr)oxides can limit As mobility in groundwater provided limited desorption under changing solute conditions. Numerous geochemical conditions can act to promote As desorption from soils and sediments, including increased concentration of competing ligands, shifts in solution pH, and changes in redox condition (1, 12, 13). A shift in redox conditions from oxidizing to reducing conditions appears to be the most ubiquitous pathway responsible for promoting As desorption (14–16). Such a transition can result in microbially induced reductive dissolution (inclusive of transformation) of Fe(III) (hydr)oxides. Microbial reduction of ferrihydrite, a poorly crystalline Fe(III) (hydr)oxide common in soil and sediment, results in dissolution and transformation to lower surface area minerals such as goethite and magnetite (17–19). Consequently, it is reasonable to expect that reductive dissolution of As-bearing ferrihydrite would result in a concomitant release of adsorbed As due to a decrease in surface area, and thus a decrease in reactive surface sites, coupled with an unstable surface environment. In fact, numerous studies have observed As desorption upon reductive dissolution of ferric (hydr)oxides (20, 21). However, recent studies have also observed the contrary; coupled with reductive transformation of ferrihydrite, retention of both As(III) and As(V) increases rather than decreases (7, 22–25). Over the past few decades there has been, in fact, disagreement regarding the dominant mechanism controlling As partitioning between aqueous and solid phases. It has been widely viewed that reduction of As(V) to As(III) promotes As release to the aqueous phase as a consequence of a decrease in adsorption strength of As(III) (3, 26). It has also been proposed that reductive dissolution of Fe(III) (hydr)oxides is the dominant mechanism of As desorption (2, 20, 21). More recently, it has been observed that reductive transformation of ferrihydrite increases retention of both As(III) and As(V) relative to control systems (7, 23, 25). Understanding the processes that lead to As desorption from soils and sediments, and the environmental factors that can enhance or limit these processes, are fundamental to predicting aqueous concentrations of As in surface or groundwater. To help resolve the role of iron reduction on the fate of arsenic, we examined As(III) desorption from packed ferrihydrite columns inoculated with Shewanella putrefaciens strain CN32 under iron reducing conditions at circumneutral pH for periods up to 90 days (270 pore volumes). Furthermore, to investigate whether the rate and magnitude of iron reduction was an additional control on As(III) retention and transport, we exposed identical columns to 3 orders of magnitude variation in electron donor (lactate) concentration. We reveal that as iron reduction proceeds in a ferrihydrite system, temporally sensitive retention and release of As(III) occurs. Initially, a period of As(III) retention commensurate with ferrihydrite transformation to magnetite; a prolonged period of As(III) release coupled with the reductive dissolution of the transformed ferric/ferrous (hydr)oxide substrates then transpires. Additionally, we observe that increasing the rate of iron reduction by providing higher electron donor (lactate in the present case) concentration accelerates As(III) desorption as the Fe(III) (hydr)oxide undergoes more rapid reductive dissolution. VOL. 42, NO. 13, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4777
Materials and Methods Iron (Hydr)oxide Coated Sand Preparation. Ferrihydrite was synthesized according to Schwertmann and Cornell (27) and then centrifuged and washed with distilled, deionized (DDI) water a minimum of four times to remove dissolved salts; it was then mixed with IOTA-6 quartz sand (Unimin Corp., Spruce Pine, NC), air-dried for 2 days, then rinsed repeatedly with distilled, deionized water until the rinsewater appeared free of iron (hydr)oxide particulates. Mineralogy was verified by taking dried subsamples of the ferrihydrite suspension (prior to mixing with quartz sand) and analyzing using X-ray diffraction (XRD). The iron content of the sand was 83.0 mmol Kg-1 as determined by digestion (6 M HCl) and ICP-OES analysis and had a surface area of 247 m2 g-1, as determined by N2-BET analysis. Preparation of Bacterial Cultures. We used the laboratory isolate S. putrefaciens strain CN-32, a common dissimilatory iron reducing bacterium to examine the influence of reductive iron transformations on aqueous As(III) concentrations. Standard methods for culture of anaerobic bacteria and preparation of anoxic media were used throughout. S. putrefaciens was grown aerobically to late log phase in tryptic soy broth (TSB, DIFCO, Detroit, MI) at room temperature and frozen in 20% glycerol at -80 °C. Seed cultures were started from frozen stocks (1 mL in 100 mL TSB) and grown aerobically for 12 h at room temperature (150 rpm). Cell suspensions were prepared by adding 1 mL of the seed culture to 100 mL of TSB and grown to late log phase (12 h; 25 °C; 150 rpm). Cells were harvested by centrifugation (4500g, 8 min, 25 °C), washed twice in 100 mL of 10 mM PIPES buffer (pH 7.1), and resuspended in an artificial groundwater medium. Arsenic(III) Desorption from Iron (Hydr)oxide Coated Sand Columns. Arsenic(III) desorption from ferrihydrite sand was studied under flow conditions using packed sand columns (Kontes Flex-column economy columns, L 1.3 cm; 5 cm long) containing ca. 9.5 g ferrihydrite sand (porosity 0.48). Ferrihydrite sands were mixed with distilled-deionized (DDI) water (250 g in 0.85 L), bubbled with N2 for 1 h, autoclaved, and cooled in an anaerobic glovebox. Once in the anaerobic chamber, DDI water was decanted and replaced with 0.85 L basal salts buffer. Basal salts buffer (BSB) was composed of 10 mM PIPES, 2.7 mM KCl, 0.3 mM MgSO4, 7.9 mM NaCl and 0.4 mM CaCl2.2H2O. Before addition to the sand, the BSB was adjusted to pH 7.1, bubbled with N2 overnight, sealed, and autoclaved. Phosphate is a necessary nutrient for microorganisms; therefore, to minimize competitive desorption, phosphate was presorbed onto the ferrihydrite sand by adding KH2PO4 to the BSB. The phosphate-sand-BSB mixture was then rotated periodically for 3 days, attaining a surface loading of ca. 0.84 µmol Kg-1. The phosphate-BSB was decanted from the sand and replaced by a similar BSB (pH 7.1, 0.85 L) containing As(III) (1.8 mM), again rotating the bottles periodically. After 3 days, the As(III)-containing BSB was decanted, and the remaining sand was rinsed twice with As/P-free BSB. Using a Langmuir isotherm, an adsorption maximum for As(III) on ferrihydrite of 1.75 mol Kg-1 (16.2 mmol Kg-1 ferrihydrite sand) was determined following experimental techniques described previously (7). An As(III) loading of 4.48 mmol Kg1- was achieved on the ferrihydrite-coated sands, which is 28% of the adsorption maximum. Arsenic(III) equilibrated sands were then mixed with washed S. putrefaciens strain CN-32 cells. Prepared As(III)-sand-cell mixtures (107 cells/g ferrihydrite sand) were loaded into columns, then immediately afterward an artificial groundwater media was pumped through the column at ∼3 pore volumes per day (6.8 µL.min-1; ca. 3.2 mL/pore volume); flow occurred from the bottom up to minimize development of preferential flow paths. Influent 4778
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 13, 2008
FIGURE 1. As(III) desorption (A) and Fe elution (B) for three influent lactate concentrations supplied to ferrihydrite-sand columns inoculated with S. putrefaciens CN-32.; initial loading of arsenic (4.48 mmol Kg-1) was 28% of the adsorption maximum at pH 7.1. Hatched lines indicate times when replicate columns were terminated for solid phase analysis media consisted of BSB (composition described previously) amended with 17.8 µM NH4Cl, Mineral solution (28) and 0.08, 0.8, or 7.7 mM lactate. For each experimental condition, three replicate columns were run simultaneously and terminated at interim time points. Columns were terminated after reaction times of 11, 52, and 90 days (33, 157, and 270 pv; indicated on Figures 1 and 2 as hatched lines labeled T1, T2, and T3). For comparison, a similar set of columns were run to measure abiotic As(III) desorption. Here, column sands were prepared in an identical fashion except bacteria were not mixed with the sand and the influent media was not amended with lactate, NH4Cl or Mineral solution; previous experiments indicated that changes in ionic strength caused by addition of these constituents did not effect a change in As(III) desorption. Samples were taken from terminated columns and from the remaining initial sand-cell-mixture for X-ray absorption spectroscopic (XAS) analysis, cell counts, and digestion with 6 M HCl. Throughout the course of the column experiments, solution samples were taken periodically from the influent and effluent for analysis of lactate, acetate and dissolved Fe and As. Analytical Methods. Inductively coupled plasma-optical emission spectroscopy (ICP-OES) was used to measure total dissolved concentrations of iron and arsenic in effluent and in solid phase digests. At lower concentrations, total dissolved As was measured by reacting acidified samples with 6 M HCl and 0.15 M NaBH4 in 0.12 M NaOH to form arsenic hydride (AsH3) prior to ICP-OES measurement. Arsenic detection limit using this technique was 0.04 µM. Organic acids were quantified using a Dionex DX-500 ion chromatograph using an IonPac AS9-HC column with 0.4 mM heptafluorobutyric acid eluent and suppressed ion conductivity detection. Ferrous iron in the aqueous samples and solid phase digests was quantified using the ferrozine method as described previously (18). For cell counts, the sands were mixed (∼1 g wet weight; 10 mL solution) with 0.5 mM sodium pyro-
by pre-edge subtraction followed by spline fitting using the SixPack XAS data analysis package (29). Backgroundsubtracted k3-weighted EXAFS were analyzed using the SixPack interface to IFEFFIT (Newville, 2001) and fitted in the k-range of 3-14 Å-1. Linear combination of model compounds was performed to reconstruct unknown spectra; Fe(II)-phases were constrained to obtain mass balance (within 5%) with extractable Fe(II). The detection limit for linear combination fitting was ∼5%. The set of Fe reference standards included ferrihydrite, goethite, lepidocrocite, magnetite, green rust-chloride, green rust-sulfate, green rustcarbonate, and hematite. Each of these materials were synthesized following the procedures of Schwertmann and Cornell (27).
Results
FIGURE 2. Proportion (relative to initial concentration) of (A) As(III) desorption and (B) Fe loss for three influent lactate concentrations supplied to As(III)-bearing (4.48 mmol Kg-1) ferrihydrite-sand columns inoculated with S. putrefaciens CN-32; As(III) desorption from abiotic columns is also shown in panel A. Difference (C) between proportion (relative to initial concentration) of abiotic As(III) desorption and proportion of As(III) desorption (solid symbols). For reference purposes dashed lines in panel C indicate actual proportion of As(III) desorbed from abiotic and biological columns; and hatched lines indicate times when replicate columns were terminated for solid phase analysis phosphate solution and sonicated for 30 min to remove the cells from the sand. The cells were fixed by adding 50 µL of 37% formaldehyde to 0.95 mL of the sonicated solution. Cells were stained with 4,6-diamidino-2-phenylindole (DAPI) and counted using epifluorescence microscopy. Speciation and distribution of Fe phases were determined using X-ray absorption spectroscopy (XAS). Samples for X-ray absorption spectroscopic analyses were sonicated (
