Environ. Sci. Technol. 2009 43, 8787–8793
Changes in Iron, Sulfur, and Arsenic Speciation Associated with Bacterial Sulfate Reduction in Ferrihydrite-Rich Systems S A M A N T H A L . S A A L F I E L D * ,†,§ A N D B E N J A M I N C . B O S T I C K †,‡ Department of Earth Sciences, Dartmouth College, HB 6105 Fairchild, Hanover, New Hampshire 03755, and Lamont-Doherty Earth Observatory, Columbia University, Palisades, New York 10964
Received June 5, 2009. Revised manuscript received October 1, 2009. Accepted October 14, 2009.
Biologically mediated redox processes have been shown to affect the mobility of iron oxide-bound arsenic in reducing aquifers. This work investigates how dissimilatory sulfate reduction and secondary iron reduction affect sulfur, iron, and arsenic speciation. Incubation experiments were conducted with As(III/ V)-bearing ferrihydrite in carbonate-buffered artificial groundwater enriched with lactate (10 mM) and sulfate (0.08-10 mM) and inoculated with Desulfovibrio vulgaris (ATCC 7757, formerly D. desulfuricans), which reduces sulfate but not iron or arsenic. Sulfidization of ferrihydrite led to formation of magnetite, elemental sulfur, and trace iron sulfides. Observed reaction rates imply that the majority of sulfide is recycled to sulfate, promoting microbial sulfate reduction in low-sulfate systems. Despite dramatic changes in Fe and S speciation, and minimal formation of Fe or As sulfides, most As remained in the solid phase. Arsenic was not solubilized in As(V)loaded incubations, which experienced slow As reduction by sulfide, whereas As(III)-loaded incubations showed limited and transient As release associated with iron remineralization. This suggests that As(III) production is critical to As release under reducing conditions, with sulfate reduction alone unlikely to release As. These data also suggest that bacterial reduction of As(V) is necessary for As sequestration in sulfides, even where sulfate reduction is active.
Introduction Arsenic mobility in aquifers is determined by complex interactions of hydrology, water chemistry, and biological processes. Anaerobic microorganisms play an important role in the release of As in iron-rich aquifers, where As is typically sequestered by iron minerals (1-3). Dissimilatory iron reduction can release As through reductive dissolution of iron oxides, although net As release may be minimal due to readsorption or incorporation in secondary mineral phases (2, 4). Other investigations have demonstrated the ubiquity of As-reducing bacteria (3, 5), which can have complex effects on environmental As mobility. However, further exploration * Corresponding author phone: 501-984-3034; e-mail:
[email protected]. † Dartmouth College. ‡ Columbia University. § Current Address: 15 Loveton Circle, Sparks, MD 21152. 10.1021/es901651k CCC: $40.75
Published on Web 10/28/2009
2009 American Chemical Society
of the effect of diverse microbial metabolisms is necessary to explain As behavior in suboxic and anoxic environments. Bacterial sulfate reduction is recognized as a mechanism for sequestering metals in contaminated environments, primarily through precipitation of metal sulfides (6, 7). The sulfide produced by bacteria can reduce As(V) (8) and precipitate As(III) in sulfide phases (9, 10). However, sulfide may also react with As-bearing iron oxides, producing oxidized S and dissolved Fe2+ (11). Various reduced or mixedvalence Fe minerals (e.g., siderite, magnetite, iron sulfides) may then form, often immobilizing As (12-14). Released As can also be readsorbed onto remaining iron oxides (15), or arsenic-sulfide complexes (e.g., thioarsenites) may form and enhance arsenic solubility (16). Ultimately, the interactions of these products determine how sulfate reduction affects arsenic mobility. Numerous field and modeling-based studies have identified empirical relationships between sulfate reduction and arsenic partitioning (9, 17-21), and considered the various sequestration and mobilization processes. However, these studies do not isolate the effect of sulfate-reducing bacteria (SRBs), but rather consider their impact in combination with iron reduction and other processes. This paper examines the effects of sulfate reduction on As concentrations in batch studies. We assess changes in the speciation of Fe, S, and As in ferrihydrite-rich systems containing a well-characterized strain of Desulfovibrio vulgaris (ATCC strain 7757, formerly D. desulfuricans), which strongly favors sulfate reduction over reduction of iron oxides, especially with lactate as electron acceptor (22, 23). The incubation experiments presented thus isolate the effects of dissimilatory sulfate reduction on solid- and aqueous-phase Fe, S, and As speciation in systems in which As is originally adsorbed to ferrihydrite.
Materials and Methods Incubation Experiments. Ferrihydrite was synthesized by KOH addition to a FeCl3 solution, washed with NaCl (24). The suspension was centrifuged to remove the supernatant, and the ferrihydrite was heat-sterilized at 80 °C for 16 h, to minimize both microbiological content and mineralogical changes (25). Analysis by extended X-ray absorption fine structure (EXAFS) spectroscopy and X-ray diffraction (XRD) before and after this sterilization confirmed that the iron remained ferrihydrite, without formation of crystalline oxides. The surface area of this ferrihydrite was approximately 200 m2/g, as determined using a 3-point BET isotherm with nitrogen gas. The medium used was a sulfate- and lactate-enriched solution of otherwise typical groundwater composition (“artificial groundwater” (26)). This solution consisted of tap water amended with 0.02 mM NH4Cl, 0.07 mM KCl, 10 mM Na-lactate, 0.8 mM MgSO4 or MgCl, 0.5 mM NaHCO3, CaCO3 at 150 mg/L, and a variable amount of NaSO4. The CaCO3 was added in slight excess, to mimic a carbonate-buffered aquifer. Supporting Information (SI) Table S1 lists initial parameters for the various incubation treatments in this study. Total sulfate concentrations in the media were manipulated by adjusting the content of MgSO4 and MgCl2, and by adding additional NaSO4 as needed. The media were autoclaved prior to addition of presterilized Na-lactate and NaHCO3 solutions, and was maintained anoxic in a glovebox (95% N2, 5% H2). Incubations were assembled in serum bottles in the glovebox. Sterilized ferrihydrite (typically 1 g L-1; some with 0.1 g L-1 or 5 g L-1) was first suspended in artificial VOL. 43, NO. 23, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
8787
groundwater medium. Phosphate was added and allowed to adsorb (750 mg/kg on ferrihydrite), followed by As (either 225 mg/kg or 9900 mg/kg on ferrihydrite). The arsenic was loaded in most incubations as arsenate, and occasionally as arsenite. The higher concentration (9900 mg/kg) represents approximately 10% of the sorption capacity of arsenate on ferrihydrite at circumneutral pH (27). Although this is in excess of typical environmental levels, the high concentration was used to maximize changes in aqueous As concentrations during incubation and thus to increase the likelihood of detecting small changes in arsenic behavior. Cultures of D. vulgaris (strain 7757) were rinsed with artificial groundwater and then used to inoculate incubations (at 1 mL culture/100 mL). Additional uninoculated bottles were monitored as negative controls. Once sealed, incubations were shaken at room temperature (22 °C) on a table at 80 rpm. Homogenized samples were extracted by syringe and filtered (0.2 µM), with an initial sampling interval of 1-2 d, increasing over a period of 40-60 d. Aqueous samples were acidified with trace-metal grade concentrated HNO3 for ICP-OES analysis (see below), and filters containing solid samples were coated in glycerol and frozen for X-ray absorption spectroscopy (XAS). Occasional samples (filtered but unacidified) were collected to determine pH, and sulfide and ferrous iron concentrations, which were measured colorimetrically immediately following collection (28, 29). ICP Analysis. Quantification by inductively coupled plasma-optical emission spectroscopy (ICP-OES) (Thermo Electron IRIS Intrepid II XSP) was based on comparison to a three-point standard curve for a multielement standard, which typically afforded 3-4 orders of magnitude linear range in quantitation. Measured concentrations were blank-corrected by subtracting the concentration of a dilute tracemetal grade acid blank. Typical errors were 50% of the solid-phase S (Figure 2), but
