Environ. Sci. Technol. 2004, 38, 2067-2074
Uranium Immobilization by Sulfate-Reducing Biofilms HALUK BEYENAL,† RAJESH K. SANI,‡ BRENT M. PEYTON,‡ ALICE C. DOHNALKOVA,§ JAMES E. AMONETTE,§ AND Z B I G N I E W L E W A N D O W S K I * ,†,+ Center for Biofilm Engineering and Department of Civil Engineering, Montana State University, Bozeman, Montana 59717-3980, Department of Chemical Engineering, Center for Multiphase Environmental Research, Washington State University, Dana Hall Room 118, Pullman, Washington 99164-2710, and Fundamental Science Directorate, Pacific Northwest National Laboratory, P.O. Box 999, K8-96, 3335 Q Avenue, Richland, Washington 99352
Hexavalent uranium [U(VI)] was immobilized using biofilms of the sulfate-reducing bacterium (SRB) Desulfovibrio desulfuricans G20. The biofilms were grown in flat-plate continuous-flow reactors using lactate as the electron donor and sulfate as the electron acceptor. U(VI) was continuously fed into the reactor for 32 weeks at a concentration of 126 µM. During this time, the soluble U(VI) was removed (between 88 and 96% of feed) from solution and immobilized in the biofilms. The dynamics of U immobilization in the sulfate-reducing biofilms were quantified by estimating: (1) microbial activity in the SRB biofilm, defined as the hydrogen sulfide (H2S) production rate and estimated from the H2S concentration profiles measured using microelectrodes across the biofilms; (2) concentration of dissolved U in the solution; and (3) the mass of U precipitated in the biofilm. Results suggest that U was immobilized in the biofilms as a result of two processes: (1) enzymatically and (2) chemically, by reacting with microbially generated H2S. Visual inspection showed that the dissolved sulfide species reacted with U(VI) to produce a black precipitate. Synchrotron-based U L3-edge X-ray absorption near edge structure (XANES) spectroscopy analysis of U precipitated abiotically by sodium sulfide indicated that U(VI) had been reduced to U(IV). Selected-area electron diffraction pattern and crystallographic analysis of transmission electron microscope lattice-fringe images confirmed the structure of precipitated U as being that of uraninite.
Introduction Wastewaters containing uranium (U) and other actinides generated in nuclear reactors in processes related to generating energy and manufacturing nuclear weapons have been discharged to the ground during the past 50 years (1). As a result, U is one of the most common radionuclides in * Corresponding author phone (406)994-5915; fax: (406)994-6098; e-mail:
[email protected]. † Center for Biofilm Engineering, Montana State University. ‡ Washington State University. § Pacific Northwest National Laboratory. + Department of Civil Engineering, Montana State University. 10.1021/es0348703 CCC: $27.50 Published on Web 02/25/2004
2004 American Chemical Society
soils, sediments, and groundwater at Department of Energy (DOE) sites and is therefore of particular environmental concern (2, 3). When selecting remediation processes, it is believed that in situ microbial reduction of U(VI) can be an attractive alternative strategy for remediation of U-contaminated subsurface environments (4, 5). Traditional ex situ remediation processes (e.g., pump and treat methods) are often limited by poor extraction efficiency and production of large volumes of toxic U waste. Also, bringing the contaminants up to the surface can increase health and safety risks for cleanup workers and the public. In situ microbial enzymatic U(VI) reduction to U(IV) provides an attractive alternative remediation strategy since U(IV) precipitates as uraninite (UO2), a highly insoluble U mineral (6, 7). Despite the potential benefits of using dissimilatory metal reducing microorganisms such as sulfate reducers for U precipitation, the U(VI) reduction mechanisms using biofilms in flow systems are poorly understood. It is vital, therefore, to characterize these mechanisms to predict U(VI) natural attenuation or to rationally design active in situ bioremediation strategies for U-contaminated subsurface environments. Sulfate-reducing bacteria (SRB), a major group of subsurface organisms responsible for heavy metal immobilization and groundwater detoxification, are known to reduce U(VI) to uraninite (8-10). The microbial process, when SRB are artificially stimulated in the subsurface using nutrient injection, may offer multiple benefits because other heavy metals and radionuclides contaminating groundwaters, such as chromium Cr(VI), U(VI), and technetium Tc(VII), are also reduced to less soluble forms (11, 12). Reduction of U(VI) to U(IV) using SRB involves at least three processes: (1) U(VI) binding to the cell surface and to extracellular biopolymers (biosorption); (2) chemical reduction of U(VI) by microbially generated hydrogen sulfide (H2S); and (3) bioreduction of U(VI), which is enzymatic dissimilatory metal reduction with U(VI) acting as a terminal electron acceptor. It is not well understood which of these processes contributes to uranium participation, and to what extent, in groundwater U reduction, and the published results are conflicting. Mohagheghi et al. (13) hypothesized that the combined effects of processes 1 and 2 were responsible for U(VI) reduction, while Lovley et al. (5) showed that process 3, bioreduction, was the dominant mechanism for U(VI) reduction in a carbonate buffered SRB culture. Lovley and Phillips (10) showed U(VI) bioreduction in the presence of 30 mM bicarbonate buffer (pH 7), which reinforced their conclusion that in the presence of bicarbonate buffer, sulfide (1 mM) was a poor reductant for U(VI) reduction (5). This effect may seriously decrease the feasibility of using SRB to precipitate uranium in groundwater and has to be carefully evaluated. Published studies have demonstrated that the main factors affecting U(VI) bioreduction are competition between electron acceptors and U speciation in solution. The difficulty of reducing uranium in the presence of carbonate buffer reported by Lovley et al. (5) coincides with the results published by Allison et al. (14), Fredrickson et al. (15), and Brooks et al. (16), who demonstrated that U(VI) was almost entirely complexed in the UO2(CO3)22- and UO2(CO3)34forms. When all these results are combined, they lead to the conclusion that because hexavalent uranium forms complexes with carbonates, it is difficult to reduce by microbially generated hydrogen sulfide and that the prevailing mechanism of uranium reduction under these conditions is bioreduction. This conclusion is not entirely certain, however, VOL. 38, NO. 7, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2067
FIGURE 1. Flat-plate flow cell used to grow the biofilms was constructed of polycarbonate (1/4′′ thick). All fittings were 3/8′′ i.d. with 1/8′′ plastic pipe thread (ACE Hardware Corp., Oak Brook, IL). All fittings were centered and placed close to the edge except the output line, which was positioned above the recycle line. All dimensions are in centimeters, and the figure is sketched without scale. as Spear et al. (17) showed that under nongrowth conditions (10 mM lactate to 20 mM bicarbonate buffer), Desulfovibrio desulfuricans reduced U(VI) faster in the presence of sulfate and suggested that the microbially generated sulfide may have been responsible for the increased rate of U(VI) reduction. Although chemical reduction of U(VI) by sulfide is thermodynamically feasible (13), practical application of such a process may be controlled by the presence of inorganic and organic chelating agents (10, 18). On the basis of all these somewhat conflicting results, we hypothesize that immobilization of U by biogenic sulfide depends on the presence of the U complexing ligands in the solution. To test this hypothesis, U reduction was tested in a microbial growth medium that did not precipitate U. To test U reduction, we used biofilms composed of SRB, D. desulfuricans G20 grown in laboratory scale reactors on quartz surfaces. A metal toxicity medium (MTM) formulated specifically to eliminate formation of metal precipitates and minimize metal complexation by Sani et al. (19) was used in this study. Under these conditions, the results showed that U(VI) was removed both enzymatically and chemically. To better understand the process dynamics, profiles of H2S concentration in biofilms were measured using microelectrodes and used to quantify microbial activity in the biofilms. To isolate the effect of sulfide on U(VI), separate abiotic experiments were performed in batch reactors. To identify the product of the chemical reaction between sulfide and U, synchrotron-based U L3edge X-ray absorption near edge structure (XANES) spectroscopy analysis and transmission electron microscopy (TEM) were used.
Materials and Methods Microorganism, Medium, and Cultivation Conditions. D. desulfuricans G20 used in the study was obtained from J. Wall, University of Missouri-Columbia (Columbia, MO). The G20 strain was derived from D. desulfuricans G100A by Wall et al. (20). D. desulfuricans G20 was maintained in MTM (19), which contained sodium lactate, 45 mM; sodium sulfate, 15 mM; calcium chloride dihydrate, 0.41 mM; ammonium 2068
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 7, 2004
chloride, 19 mM; magnesium sulfate, 4 mM; and 1,4piperazinediethane sulfonic acid disodium salt monohydrate (PIPES), 30 mM. In addition, 0.05 g/L yeast extract and 0.5 g/L tryptone were also added. The pH was adjusted to 7.2 using a solution of 6 M HCl. All components of the medium were of analytical grade and were purchased from Fisher Scientific (Pittsburgh, PA) with the exception of the following: yeast extract and tryptone were obtained from Difco Chemical Company (Detroit, MI), and PIPES was obtained from Aldrich Chemical Co. (Milwaukee, WI). A fresh inoculum for each experiment was prepared by growing a 3-day-old batch culture of D. desulfuricans in 100 mL serum bottles leaving approximately 10 mL of headspace in the bottles. The flat-plate biofilm rectors were inoculated with 60 mL of that inoculum. Biofilm Reactor. The biofilms were grown at room temperature on quartz slides (2.5 × 2.5 cm) placed on the bottom of an open-channel flow reactor made of polycarbonate (3.5 cm deep, 2.5 cm wide, and 34 cm long, with a total working volume of 120 mL) (Figure 1). To maintain pure cultures, the entire system was sterilized with 70% alcohol before each experiment. The reactor was then rinsed with autoclaved water until all the alcohol was removed. Tubing, connectors, air filters, and the growth medium were autoclaved at 121 °C for 15 min. Before addition of the growth medium and during biofilm growth, the system was continuously purged with filtered sterilized nitrogen gas to create positive pressure and prevent oxygen penetration into the system. During the operation, to prevent microbial contamination, the lid of the reactor remained sealed with silicon rubber. The reactor was fed continuously with growth medium at a flow rate of 6.25 mL/h. U(VI) was added to the medium to maintain a 126 µM concentration. The average linear flow velocity in the biofilm reactor was 2.4 cm/s, and the average residence time was 19.2 h. H2S Microelectrodes and Measurement. Microbial activity in SRB biofilms was evaluated as the rate of H2S production. The rate of H2S production was estimated from H2S concentration profiles measured by microelectrodes constructed and used according to Jeroschewski et al. and
FIGURE 2. (A) Schematic view of H2S microelectrode. (B) Schematic view of experimental setup. 1: growth medium; 2: reference electrode*; 3: peristaltic pump; 4: microscope; 5: microelectrode*; 6: micromanipulator*; 7: flat-plate flow cell; 8: computer*; 9: outlet; 10: fresh feed; 11: vent; 12: N2 gas; 13: cell cluster; 14: interstitial void; 15: picoammeter*; 16: fresh feed (used for high feed flow rates). *: These instruments were used only during measurements. Ku ¨hl et al. (21, 22). To prevent damaging the biofilm structure during measurements, the microelectrodes had tip diameters less than 20 µm (Figure 2A). The principle of the hydrogen sulfide microelectrode operation is as follows: H2S diffuses through the silicone membrane covering the tip to the shaft of the microelectrode, which is filled with an alkaline solution of potassium ferricyanide [K3Fe(CN)6]. In this solution, part of the H2S deprotonates to HS-, which is then oxidized to elemental sulfur, S, by the ferricyanide. The reduced form of ferricyanide, ferrocyanide [K4Fe(CN)6], is then reoxidized back to ferricyanide at the internal platinum anode polarized at (+)100 mV against a Pt counter electrode placed approximately 10 µm behind the silicone membrane. The current generated by oxidizing ferro- to ferricyanide is proportional to H2S concentration near the tip of the microelectrode. The electrode was calibrated in solutions of different H2S concentrations, prepared by dissolving Na2S in O2-free phosphate buffer (100 mM, pH 7). The response time of the microelectrodes was about 3 s and was linear over a large H2S concentration range, from 0 to 300 µM H2S. The microelectrodes had the same calibration curve before and after the measurements, within the range of acceptable experimental error. New microelectrodes were used for each H2S measurement. To measure H2S concentration profiles, the microelectrodes were inserted into the reactor through a small opening in the lid while N2(g) continually purged the system, creating
positive pressure in the reactor and preventing ingress of the outside air (Figure 2B). The microelectrodes were mounted on a micromanipulator (Model M3301L, World Precision Instruments, New Haven, CT) equipped with a stepper motor (Model 18503, Oriel, Stratford, CT) and controlled by the Oriel Model 20010 interface. Microelectrodes were introduced from the top of the reactor at an angle perpendicular to the biofilm. During the measurements, the position of the microelectrode’s tip was observed using an Olympus CK2 inverted microscope. The micropositioner was interfaced with a computer, and the microelectrode movement was facilitated by a controller (CTC-283-3, Micro Kinetics) with a positioning precision of 0.1 µm. Custom software was used to control and coordinate the microelectrode movements and the data acquisition (23, 24). The diffusive fluxes of H2S to the bulk liquid were calculated by multiplying the slope of the profiles evaluated at the biofilm surface by the diffusion coefficient of H2S in water (DH2S ) 1.7 × 10-5 cm2/s) (22). Uranium Reduction in Biofilms. To observe U reduction, biofilms were grown in two reactors under identical conditions. After a suitable biofilm had developed, microbial respiration was inhibited in one reactor using sodium azide (NaN3, 5 mM). Then, a known aliquot of U(VI) was introduced into both reactors. Consequently, in the reactor where microbial respiration was inhibited, U(VI) was considered to be reduced only through the action of microbial enzymes since it has been shown that NaN3 does not affect the ability of D. desulfuricans to reduce U(VI) (10). At the same time, in the reactor where sulfate respiration was not inhibited, U(VI) was reduced both enzymatically and with the microbially generated hydrogen sulfide. To quantify the rate of U(VI) removal, microbial activity in the 3-week old biofilm was evaluated from H2S concentration profiles. Then, to inhibit H2S production, the reactor was flushed with a solution of 5 mM NaN3 (pH 7, buffered with 30 mM PIPES) at a flow rate of 150 mL/min for 30 min. At the end of the 30 min period, H2S concentration profiles were measured again to ensure that the NaN3 treatment had been successful, and H2S was not produced. At that time, 18.9 µmol of U(VI) (in 30 mL) was injected to each reactor to give a concentration of 126 µM U(VI). The reactors were continuously fed with 126 µM solution (pH 7, buffered with 30 mM PIPES) at a flow rate of 150 mL/min (from the location 16 shown in Figure 2) for 30 min, to keep the recycle flow rate constant. Finally, samples of the biofilm were removed from the reactors, and the amount of immobilized U was measured in each sample. U(VI) was continuously fed to another reactor operated separately for 32 weeks, and U(VI) removal rates were quantified. Biofilm structure and thickness observed were monitored during the experiments following the procedures given by Beyenal and Lewandowski (24). Uranium(VI) Reduction by Sulfide. To test for U(VI) reduction by sulfide, serum bottles containing PIPES buffer (30 mM, pH 7) or Nanopure water were prepared according to Sani et al. (25). Briefly, serum bottles containing 100 mL of Nanopure water or PIPES buffer (30 mM, pH 7) were autoclaved and put immediately in a glovebox (Model 1025, Forma Scientific Inc., OH) under vacuum (34 kPa) to remove headspace O2. Serum bottles were left overnight in an anaerobic chamber containing a mixture of gases, N2/CO2/ H2 (80:15:5), and then sealed with a butyl rubber septa, capped, and crimped with aluminum seals. Stock solutions of 50 mM U as UO2Cl2‚3H2O (Bodman Industries, Aston, PA) and 1 M sulfide as Na2S‚9H2O (Aldrich Chemical Co., Milwaukee, WI) were prepared in the deaerated water in a glovebox. Serum bottles containing Nanopure water or PIPES buffer were treated with U(VI) (140 µM) and sulfide (1 mM) and were incubated at room temperature (25 °C) at 100 rpm for 48 h. Prior to the sulfide addition, samples for evaluating the initial U(VI) concentrations were taken anaerobically, VOL. 38, NO. 7, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2069
using disposable syringes. In each experiment, sulfide-free controls were used. Each measurement was carried out in triplicate, and the measurements were repeated three times. Uranium Analyses. To determine the extent of U(VI) removal in the batch studies, samples (0.2 mL) of the liquid were withdrawn using a needle and syringe, and the uranium concentration was measured immediately after filtration. We used Gelman Acrodisc filters, pore diameter 0.2 µm, to minimize the reoxidation of the reduced U in solution (26). As described previously by Sani et al. (19), samples were diluted 1000 times with Nanopure water, and then 1 mL of the sample was mixed with 1.5 mL of a proprietary complexing agent, Uraplex (Chemcheck Instruments, Inc., USA). Samples were analyzed with a kinetic phosphorescence analyzer (KPA11, Chemcheck Instruments, Inc.), which uses a pulsed N2pumped dye laser to measure U(VI) concentrations in solution (27). Calibration was performed using UO2Cl2 solutions from 0 to 0.16 µM, yielding a U(VI) detection limit of 0.04 µM with a precision of (5%. Removal of U(VI) from the aqueous phase was evaluated by measuring dissolved U(VI) and total U in the filtered samples. To determine total U, the filtered samples (0.1 mL) were exposed to air for 1 h, and 0.3 mL of concentrated HNO3 was added (10, 18). The samples were vortexed and left for 1 h more exposed to the air and then diluted 100 times using Nanopure water, and total U as U(VI) was measured as described previously. The total concentration of U in the input and output of the biofilm reactors was measured using the method described by Meloan et al. (28). The biofilms containing precipitated U were dissolved in 2 N HNO3 to oxidize U(IV) to U(VI) (29). The samples were centrifuged at 6000 rpm for 15 min, and U(VI) was measured in supernatant as described previously (28). XANES Spectroscopy of U Precipitates. At the end of abiotic experiments, black precipitates, resulting from U(VI) reduction by sulfides, were separated from the aqueous phase by centrifugation at 10 000g for 10 min under anoxic conditions. After decanting the supernatant, the black precipitates were mixed with quartz grains (R-SiO2, 212-300 µm) and dried in a glovebox under an O2-free N2/CO2/H2 (90:5:5) atmosphere. Dried solid samples were packed between Kapton windows in a multi-specimen Al sample holder modified to fit into a Lytle detector chamber. To avoid oxidation, the specimen-mounting process was carried out in a glovebox, and the sample holder was placed inside a BBL Gas-Pak (Becton Dickinson, Franklin Lakes, NJ) container for shipping and storage at the synchrotron before analysis. XANES spectra at the U L3-edge were collected in fluorescence mode at the Pacific Northwest ConsortiumCollaborative Access Team (PNC-CAT) bending-magnet beamline (20-BM-B) located at the Advanced Photon Source (Argonne, IL). Fluorescence data were collected using a scintillation detector with a single-channel analyzer to define the energy window. A strontium filter was placed between the sample and the detector to minimize background from the incident beam. Natural uraninite (nominally UO2) and UO2Cl2‚3H2O samples served as reference standards and were analyzed in transmission mode. During the analysis, the Lytle detector chamber was continuously flushed with He(g) to prevent oxidation of the samples by air. The resulting spectra were normalized for comparative purposes. Transmission Electron Microscopy (TEM). Because of the O2-sensitive nature of the samples, resin embedding of the specimens and thin sectioning of the material embedded in a resin block of the microtome were conducted in an anaerobic glovebox (Ar/H2, 95:5; Coy Laboratory Products, Inc.). The black precipitates resulting from U(VI) reduction by sulfide were washed in deionized water and fixed in 2.5% glutaraldehyde followed by gradual dehydration in ethanol 2070
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 7, 2004
FIGURE 3. H2S concentration profiles in 3-, 6-, and 12-week-old grown biofilms (A) grown without U(VI) and (B) grown with U(VI). Note that both biofilms were grown without U(VI) for first 3 weeks, and then one of the reactor was fed continuously with 126 µM U(VI). series and infiltration in LR White embedding resin. Samples embedded in solid resin blocks were sectioned to 70 nm on an ultramicrotome (Leica Ultracut UCT), and sections were mounted on 200 mesh copper grids coated with Formvar support film sputtered with carbon. Sections were examined using a JEOL 2010 high-resolution transmission electron microscope (HR TEM) equipped with a LaB6 filament operating at 200 kV with resolution of 1.9 Å. Elemental analysis was performed using an Oxford EDS system equipped with a SiLi detector coupled to the TEM and analyzed with ISIS software. Images were digitally collected and analyzed using Gatan’s Digital Micrograph. The d spacings obtained from the selected area electron diffraction (SAED) ring patterns of nanocrystalline minerals were evaluated by Desktop Microscopist (Lacuna Labs) software.
Results and Discussion Activity of Sulfate-Reducing Biofilms in the Presence and Absence of Uranium. The activities of the biofilms were quantified from H2S concentration profiles. The profiles of H2S in 3-, 6-, and 12-week old biofilms grown on quartz with and without U(VI) are shown in Figure 3. For the first 3 weeks, the biofilms in both reactors were grown without U(VI), and then one of the reactors was continuously fed with 126 µM U(VI). The H2S measurements were performed in the middle of a large cell cluster with a diameter of at least 500 µm as described by Beyenal and Lewandowski (24). The H2S concentrations at the bottom of 3-, 6-, and 12-week old biofilms, before addition of U(VI) to the reactor, were 210, 240, and 300 µM, respectively. After addition of U(VI), these concentrations decreased to 100, 160, and 210 µM, respectively. In biofilms grown without U(VI), H2S fluxes were
FIGURE 4. Total U immobilized by the sulfate-reducing biofilm (b), uranium concentration in the feed solution (2), and in the outlet (9). The continuous line shows the hypothetical values expected if 100% of U was immobilized by the biofilm. Note that the time zero refers to the time we started uranium feed. However, at that point, biofilm was 3 weeks old and grown without uranium in the feed. approximately constant, 1.4 × 10-4 µmol/cm2/s, measured at weeks 3, 6, and 12. In biofilms grown with U(VI), H2S fluxes were lower, negligible at week 3, and then 1.2 × 10-4 µmol/cm2/s and 0.82 × 10-4 µmol/cm2/s were measured at weeks 6 and 12, respectively. The lower H2S flux observed for biofilms grown with U(VI) may have been caused by the inhibitory effects of U(VI) or by the reduction of U(VI) by H2S. Figure 3 shows that H2S concentrations in the biofilms increased between 3rd and 12th week of operation. Immobilization of U(VI) by Sulfate-Reducing Biofilms. Figure 4 shows that the total amount of U immobilized in the biofilm increased linearly during the entire time that the biofilm reactor was in operation. Uranium immobilization efficiencies (expressed as the ratio of the amount of U immobilized to the total amount of U introduced to the rector) were 96, 88, and 93% at the end of the 8, 18, and 32 weeks, respectively, consistently showing at least 88% U removal. The maximum U removal efficiency was observed at the eighth week of operation and was hypothetically ascribed to biosorption since these biofilms were initially grown without U. There was a slight increase in U removal between the 18th and 32nd week, possibly caused by either rapid biofilm growth or adaptation of the biofilm microorganisms to U. Combined information from Figures 3 and 4 show that presence of U(VI) affects the following: (1) H2S concentration in the reactor, (2) H2S concentration distribution in the biofilm, and (3) H2S production rate. After adding U(VI), H2S concentration in the bulk phase decreased, which may have resulted from at least one of the following processes: (i) chemical reduction of U(VI) by biogenic H2S; (ii) inhibition of H2S production by U(VI); and (iii) enzymatic reduction of U(VI) rather than sulfate (i.e., bacterial competition for electron acceptors). To separate the possible contributions of these factors, additional investigations were performed. Effect of U(VI) and NaN3 on the Profiles of H2S Concentrations. Figure 5 shows the profiles of H2S concentrations in 3-week-old biofilms before and after adding U(VI) and NaN3. As expected, the highest H2S concentrations were measured near the bottom of the biofilms (175-185 µM), and concentrations decreased toward the surface of the biofilm. After treating the biofilm for 30 min with NaN3, the concentration of H2S across the biofilm decreased about 9-fold, and the H2S flux across the biofilm surface decreased 10-fold. These results were as expected and showed that the addition of NaN3 significantly decreases the activity of SRB biofilms. It was disappointing that NaN3 could not stop H2S production completely in the biofilms. In an attempt to inhibit microbial activity in the biofilm, the biofilm was treated with
FIGURE 5. (A) H2S concentration profiles in a 3-week-old biofilm: (1) U- and NaN3-free biofilm, (2) 30 min after adding NaN3, (3) 30 min after adding U(VI) and NaN3 mixture; (B) H2S concentration profiles in a separate 3-week-old biofilm: (4) U-and NaN3-free biofilm, (5) 30 min after adding U(VI). Note: The data in panels A and B were measured in different reactors, but under the same conditions, to compare the effects of U(VI) on biofilm activity. NaN3 for longer periods, up to 2 h, but the results were almost identical to those shown by profile 2 (data not shown). However, the decrease in H2S concentration when only U(VI) was added (Figure 5B) was six times greater than when NaN3 and U(VI) were added together (Figure 5A). As a result of adding U(VI) and NaN3 (profile 3 in Figure 5A), the H2S concentration at the bottom of the biofilm decreased by approximately 4 µM. This decrease in H2S concentration in the presence of U(VI) may be due either to toxicity of U(VI) on microorganisms in the biofilm or to reducing U(VI) by H2S to form uraninite and elemental sulfur. Similar measurements were repeated at different locations in the same biofilm and repeated in different biofilms. In all measurements, the average H2S concentration was 19 ( 12 µM near the bottom, and H2S concentration decreased after adding uranium to the sodium azide treated samples by 5 ( 3 µM (the difference between line 2 and 3 at the bottom of the biofilm). After the H2S profiles were measured, we removed the cover slides from the reactor and measured total U immobilized by the biofilms. The total amount of U immobilized in the biofilms exposed only to U(VI) was 8.5 ( 1.1 µmol/ cm2, and it was somewhat higher than that measured in biofilms exposed to both U(VI) and NaN3, 6.2 ( 0.6 µmol/ cm2. These results suggest that some of the biogenic sulfide could have reacted with U(VI), which differs from the previously published results for U(VI) reduction by SRB in the presence of a bicarbonate buffer (30 mM, pH 7) (5). U(VI) Reduction by Sulfides. Our results in the previous section clearly show that in continuous flow systems, without VOL. 38, NO. 7, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2071
FIGURE 7. U L3-edge XANES spectra for a U(IV) standard [natural UO2(s)], a U(VI) standard [UO2Cl2(s)], and for the precipitate obtained from a U(VI) solution treated with Na2S under anoxic conditions.
FIGURE 6. (A) Changes in soluble U(VI) concentrations in the presence or absence of 1 mM Na2S (corresponds to approximately 500 µM H2S) in 30 mM PIPES (pH 7) under anoxic conditions. The data points on the graphs represent the average values of triplicate treatments, and the error bars represent the corresponding standard deviations. (B) Changes in H2S concentration in the presence or absence of 1 mM U(VI) in 30 mM PIPES (pH 7) under anoxic conditions. bicarbonate buffer, sulfate-reducing biofilms remove U(VI) enzymatically as well as chemically. Lovley and Phillips (10) established that SRB can enzymatically reduce U(VI) to U(IV) and suggested that the pathway of electron flow to U(VI) in D. vulgaris strain Hildenborough with hydrogen as the electron donor was through hydrogenase to cytochrome c3 to U(VI). Similar enzymatic U(VI) reducing mechanisms using lactate as the electron donor may have occurred in the SRB biofilms in our systems. To separately determine the effect of sulfide on U(VI) reduction, experiments were performed in batch reactors containing U(VI) and sulfide in Nanopure water or in PIPES buffer (30 mM, pH 7). Figure 6 shows the reduction of U(VI) by sulfide in batch reactors under anoxic conditions and demonstrates that both sulfide and U(VI) concentrations decreased in time. In contrast, both sulfide and U(VI) concentrations remained constant in the U(VI)- and sulfide-free control experiment (Figure 6A,B), which corroborated with results from the biofilm reactors. To make sure that U(VI) was reduced by sulfide, and that the PIPES buffer did not contribute to this process, the same measurements were repeated using the Nanopure water instead of PIPES buffer, and the results were nearly identical to those in Figure 6 (data not shown), which suggests that in our systems under sulfate-reducing conditions, U(VI) reduction was likely occurring both enzymatically and chemically. We repeated the experiments shown in Figure 6 using 30 mM bicarbonate buffer instead of PIPES buffer and found that both sulfide and U(VI) concentrations remained constant (results not shown) in time, showing that the reduction of uranium by biogenic sulfides depended on the presence of U-complexing ligands in the solution, which supported our hypothesis that reduction of uranium by biogenic sulfide 2072
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 7, 2004
FIGURE 8. Transmission electron microscopy (TEM) photomicrographs of unstained thin sections from U(VI) incubated with Na2S, illustrating the production of reduced U (A). (B) HR-TEM micrograph of U precipitates. The average particle size of the newly formed biogenic uraninite was 4 nm. depends on the presence of U-complexing ligands in the solution. Characterization of Chemically Reduced Uranium. The U L3-edge XANES spectra for U(VI) samples incubated in the presence of sulfide were similar to that of natural uranite (UO2) (Figure 7), showing that under anaerobic conditions U(VI) was reduced during the sulfide treatment. Figure 8A,B shows transmission electron microscopy (TEM) photomicrographs of unstained thin sections of U precipitates incubated with sodium sulfide. The individual particle diameters were in the range of 2.6-3.6 nm and occurred as discrete and aggregated particles. Our results corroborated with Suzuki et al. (30) who also observed nanometer-sized particles resulting from bacterial U reduction. These samples reveal the presence of uraninite [UO2(s)] as determined by energy-dispersive spectroscopy (EDS, Figure 9) and selected area electron diffraction (SAED, Figure 10). No sulfur peak was observed in the EDS spectra. This suggests that during the U(VI) reduction, sulfide has likely been oxidized to sulfite or sulfate, or that during the sample preparation, elemental sulfur washed away. High-resolution TEM revealed the finegrained nature of the U precipitates that yielded lattice fringe images with d spacings consistent with UO2(s) (Table 1). Interestingly, these d spacings were also consistent with those previously obtained for U reduced by D. desulfuricans and Geobacter metallireducens (Table 1) (4, 10). These results clearly suggest that under anaerobic conditions, sulfide reduced U(VI), and U(IV) was precipitated as UO2(s).
TABLE 1. Comparison of d Spacings of Sulfide-Reduced U to Those from Published Data experimental d spacings (nm)a
JCPDSb 41-1422 d spacings for synthetic UO2(s) (nm)
JCPDSb 05-0550 d spacings for natural UO2(s) (nm)
0.318 0.276 0.194 0.165 0.126
0.3153 0.2733 0.1933 0.1647 0.1254
0.3157 0.2735 0.1934 0.1649 0.1266
a
Present study.
b
d spacings for D. desulfuricans reduced U (nm)c
d spacings for G. metallireducens reduced U (nm)d
0.315 0.274 0.193 0.163-0.166
0.314 0.272 0.193 0.163
Joint Committee on Powder Diffraction-International Centre for Diffraction data. c From ref 10.
FIGURE 9. Energy-dispersive X-ray spectrum for the U particles precipitated by treatment with Na2S. Unlabeled peaks near 8 keV are from Cu grid.
d
From ref 4.
reducing biofilms can reduce U(VI) both enzymatically and chemically. Further work is needed to determine the bicarbonate concentration that significantly inhibits chemical sulfide U(VI) reduction. Median bicarbonate concentrations in natural groundwaters are around 3.3 mM (i.e., 200 mg L-1) (32). Higher concentrations of bicarbonate have been observed in limestone aquifers treated with acidic waste streams, such as at the Natural and Accelerated Bioremediation Research program (NABIR) Field Research Center at Oak Ridge National Laboratory. In summary, our results suggest that in the absence of a bicarbonate-buffered system, sulfate-reducing biofilms will likely remove U(VI) from the contaminated water for significant amounts of time as the result of both enzymatic precipitation and chemical reduction of soluble U(VI) to insoluble uraninite.
Acknowledgments
FIGURE 10. Electron diffraction pattern of the U particles precipitated by Na2S. The electron diffraction pattern of uranium (reduced by sodium sulfide) shows rings characteristic for nanocrystalline material, with d spacing values in a good accordance with the listed nonbiogenic uraninite diffraction data. Our results show that in a flowing system, sulfate-reducing biofilms can immobilize U for more than eight months with high efficiency. In the absence of bicarbonate buffer, sulfatereducing biofilms likely reduced U both enzymatically and chemically using biogenic sulfide. Also under the same conditions, sulfide reduced more than 95% of the U(VI) from solutions containing Nanopure water or PIPES buffer (30 mM, pH 7) during 8 h of incubation. It has also been reported that U(VI) can precipitate as metaschoepite (UO3‚2H2O) with PIPES (15). However, in the present study, under anaerobic conditions, no precipitates were detected in the sulfide-free controls containing PIPES buffer and U(VI) incubated up to 50 h of incubation. In our opinion, microbial and chemical uranium reduction remains a potentially viable method for in situ plume stabilization. It has several possible advantages over other methods of U removal (4, 5). However, after it was shown that in a highly buffered bicarbonate system (30 mM) sulfide was a poor reductant for U(VI) (5), many researchers discontinued their study of sulfide reduction of U(VI) (31). Abandoning microbial reduction as a possible uranium remediation process was, perhaps, not entirely justified because 30 mM is an exceedingly high bicarbonate concentration, and as we have demonstrated, in the absence of bicarbonate buffer and in continuous flow systems, sulfate-
The authors gratefully acknowledge the financial support provided by the Natural and Accelerated Bioremediation Research program (NABIR), Office of Biological and Environmental Research (OBER), and U.S. Department of Energy (DOE), USA (Grants DE-FG03-98ER62630/A001 and DEFG03-01ER63270). The support of the Center for Multiphase Environmental Research and the Department of Chemical Engineering also contributed significantly to this research. The Pacific Northwest National Laboratory (PNNL) is operated for DOE by Battelle Memorial Institute under Contract DE-AC06-76RL0 1830. Part of this research was performed at the Environmental Molecular Sciences Laboratory (EMSL), a national scientific user facility sponsored by the Department of Energy’s Office, located at PNNL.
Literature Cited (1) McCullough, J.; Hazen, T. C.; Benson, S. M.; Metting, F. B.; Palmisano, A. C. Bioremediation of metals and radionuclides...what it is and how it works: A NABIR Primer; Lawrence Berkeley National Laboratory, Berkeley, CA: U.S. Department of Energy, LBNL-42595, 1999. (2) Lloyd, J. R.; Lovley, D. R. Curr. Opin. Biotechnol. 2001, 12, 248253. (3) Riley, R. G.; Zachara, J. M. Nature of chemical contaminants on DOE lands and identification of representative contaminant mixtures for basic subsurface science research; Washington, D.C.: DOE/ER-0547T, 1992. (4) Gorby, Y. A.; Lovley, D. R. Environ. Sci. Technol. 1992, 26, 205207. (5) Lovley, D. R.; Phillips, E. J. P.; Gorby, Y. A.; Landa, E. R. Nature 1991, 350, 413-416. (6) Langmuir, D. Geochim. Cosmochim. Acta 1978, 42, 547-569. (7) Parks, G. A.; Pohl, D. C. Geochim. Cosmochim. Acta 1988, 52, 863-875. (8) Abdelouas, A.; Lutze, W.; Nuttall, H. E. J. Contam. Hydrol. 1999, 36, 353-375. (9) Abdelouas, A.; Lu, Y. M.; Lutze, W.; Nuttall, H. E. J. Contam. Hydrol. 1998, 35, 217-233. (10) Lovley, D. R.; Phillips, E. J. P. Appl. Environ. Microbiol. 1992, 58, 850-856. (11) Lovley, D. R.; Coates, J. D. Curr. Opin. Biotechnol. 1997, 8, 285289. (12) Lloyd, J. R.; Yong, P.; Macaskie, L. E. Appl. Environ. Microbiol. 1998, 64, 4607-4609. VOL. 38, NO. 7, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2073
(13) Mohagheghi, A.; Updegraff, D. M.; Goldhaber, M. B. Geomicrobiol. J. 1985, 4, 153-173. (14) Allison, J. D.; Brown, D. S.; Novo-Gradac, K. J. MINTEQA2/ PRODEFA2, a geochemical assessment model for environmental systems, EPA/600/3-91/021; U.S. Environmental Protection Agency: Cincinnati, 1991. (15) Fredrickson, J. K.; Zachara, J. M.; Kennedy, D. W. et al. Geochim. Cosmochim. Acta 2000, 64, 3085-3098. (16) Brooks, S. C.; Fredrickson, J. K.; Carroll, S. L. et al. Environ. Sci. Technol. 2003, 37, 1850-1858. (17) Spear, J. R.; Figueroa, L. A.; Honeyman, B. D. Environ. Sci. Technol. 1999, 33, 2667-2675. (18) Ganesh, R.; Robinson, K. G.; Reed, G. D.; Sayler, G. S. Appl. Environ. Microbiol. 1997, 63, 4385-4391. (19) Sani, R. K.; Geesey, G.; Peyton, B. M. Adv. Environ. Res. 2001, 5, 269-276. (20) Wall, J. D.; Rappgiles, B. J.; Rousset, M. J. Bacteriol. 1993, 175, 4121-4128. (21) Jeroschewski, P.; Steuckart, C.; Kuhl, M. Anal. Chem. 1996, 68, 4351-4357. (22) Kuhl, M.; Steuckart, C.; Eickert, G.; Jeroschewski, P. Aquat. Microb. Ecol. 1998, 15, 201-209. (23) Beyenal, H.; Lewandowski, Z. Water Res. 2000, 34, 528-538.
2074
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 7, 2004
(24) Beyenal, H.; Lewandowski, Z. Aiche J. 2001, 47, 1689-1697. (25) Sani, R. K.; Peyton, B. M.; Smith, W. A.; Apel, W. A.; Petersen, J. N. Appl. Microbiol. Biotechnol. 2002, 60, 192-199. (26) Anderson, R. F. Nucl. Instrum. Methods Phys. Res., Sect. A 1984, 223, 213-217. (27) Brina, R.; Miller, A. G. Anal. Chem. 1992, 64, 1413-1418. (28) Meloan, L. E.; Holkeboer, P.; Brandt, W. W. Anal. Chem. 1960, 32, 791-793. (29) Ganesh, R.; Robinson, K. G.; Chu, L. L.; Kucsmas, D.; Reed, G. D. Water Res. 1999, 33, 3447-3458. (30) Suzuki, Y.; Kelly, S. D.; Kemner, K. M.; Banfield, J. F. Nature 2002, 419, 134. (31) Finneran, K. T.; Anderson, R. T.; Nevin, K. P.; Lovley, D. R. Soil Sed. Contam. 2002, 11, 339-357. (32) Langmuir, D. Aqueous Environmental Geochemistry; PrenticeHall: Upper Saddle River, NJ, 1997; p 294.
Received for review August 5, 2003. Revised manuscript received January 14, 2004. Accepted January 16, 2004. ES0348703
