THERMODYNAMIC MODELING OF U(VI) ADSORPTION ONTO BACTERIA: IMPLICATIONS FOR QUANTIFYING BIOAVAILABILITY RELATIONSHIPS.

Transcription

1 THERMODYNAMIC MODELING OF U(VI) ADSORPTION ONTO BACTERIA: IMPLICATIONS FOR QUANTIFYING BIOAVAILABILITY RELATIONSHIPS A Dissertation Submitted to the Graduate School of the University of Notre Dame in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy by Ling Sheng Jeremy B. Fein, Director Graduate Program in Civil Engineering and Geological Sciences Notre Dame, Indiana April 2013

2 Copyright 2013 Ling Sheng

3 THERMODYNAMIC MODELING OF U(VI) ADSORPTION ONTO BACTERIA: IMPLICATIONS FOR QUANTIFYING BIOAVAILABILITY RELATIONSHIPS Abstract by Ling Sheng Metal adsorption onto bacteria not only affects the mobility and transport of metals in geologic systems, but also may control the bioavailability of metal ions to bacteria. Surface complexation modeling (SCM) represents a flexible approach for quantifying the extent of metal adsorption onto bacterial cell walls, and hence may be used to predict the fate and transport of metals in water-rock systems and even the bacterial bioavailability of metals. The goal of this dissertation research is to test if metal adsorption onto bacteria controls the bioavailability of metals to bacteria and whether it is possible to use a SCM approach to predict metal bioavailability. Hence, the studies described in this dissertation examine possible correlations between U(VI) adsorption onto bacteria and the kinetics of enzymatic reduction of U(VI) by bacteria, and also put rigorous constraints on the uranyl bacterial surface complexation reactions in order to better understand the bioavailability of uranyl to bacteria, and the fate of uranium in natural and engineered environments.

4 Ling Sheng Three studies are presented in this dissertation. The first study (Chapter 2) examines U(VI) reduction by bacteria in the presence of variable concentrations of EDTA or dissolved Ca. The results of this study demonstrate that EDTA increases the U(VI) reduction rate by forming U 4+ -EDTA aqueous complexes which remove U(IV) from the cell surface after reduction, and prevent UO 2 precipitation on the cell wall, thereby preventing blockage of U(VI) binding sites. Furthermore, the data demonstrate that dissolved Ca decreases the U(VI) bioreduction rate by forming Ca-uranyl-surface complexes, and that the U(VI) in these surface complexes is not easily reducible. Therefore, the concentration of Ca-free uranyl surface complexes controls the U(VI) bioreduction rate in the presence of dissolved Ca. In summary, the results of this study indicate that U speciation, both of U(VI) before reduction and of U(IV) after reduction, affects the reduction kinetics, and that thermodynamic modeling of the U speciation may be useful in the prediction of reduction kinetics in realistic geologic settings. In the second study (Chapter 3), I measured U(VI) adsorption onto bacteria as a function of dissolved NaHCO 3 concentration. The data provide unequivocal evidence for the formation of a series of uranyl-, uranyl-hydroxide-, uranyl-carbonate-hydroxide-, and uranyl-carbonate-bacterial surface complexes. The calculated stability constants for the uranyl-bacterial complexes from this study provide a framework for estimating the adsorption and speciation of U(VI) on bacterial cell walls in complex environments. Based on the results of the first two studies, the third study (Chapter 4) focused on the enzymatic reduction of U(VI) by bacteria in systems with elevated concentrations of NaHCO 3, revealing a strong positive correlation between the U(VI) reduction rate and

5 Ling Sheng the total concentration of adsorbed U(VI). This positive correlation indicates that the speciation and adsorption of U(VI) on the bacterial cell wall control the kinetics of enzymatic reduction of U(VI) by bacteria. The successful use of surface complexation modeling to relate U speciation to enzymatic reduction rates in this study may enable predictions of enzymatic U(VI) reduction kinetics or bioavailability of uranium to bacteria in complex geologic settings.

6 CONTENTS Figures... iv Tables... vii Acknowledgments... viii Chapter 1: Introduction Overview Effects of U(VI) Speciation on the Rate of U(VI) Reduction by Bacteria Controls on U(VI) Adsorption by Bacteria in the presence of Dissolved Inorganic Carbonate Further Test of Correlations between the U(VI) Speciation and the Rate of U(VI) Reduction by Bacteria... 7 Chapter 2: The Effects of Uranium Speciation on the Rate of U(VI) Reduction by Shewanella oneidensis MR Introduction Experimental Methods U(VI) Reduction Experiments U(VI) Adsorption Experiments Analytical Methods Results Results of U(VI) Reduction Experiments with EDTA Results of U(VI) Reduction Experiments with Ca Results of U(VI) Adsorption Experiments Discussion Effect of EDTA on the Rate of U(VI) Reduction Effect of Ca on the Rate of U(VI) Reduction Conclusions Chapter 3: Uranium Adsorption by Shewanella oneidensis MR-1 as a function of NaHCO 3 Concentration Introduction Experimental Methods Bacteria Preparation U(VI) Adsorption Experiments ii

7 3.3 Results and Discussion U(VI) Adsorption Experiments Thermodynamic Modeling Conclusions Chapter 4: Uranium Reduction by Shewanella oneidensis MR-1 as a function of NaHCO 3 Concentration: Surface Complexation Control of Reduction Kinetics Introduction Experimental Methods Bacteria Preparation U(VI) Reduction Experiments Results and Discusstion U(VI) Reduction Experiments Thermodynamic Modeling Conclusions Chapter 5: Conclusions Appendix A: Aqueous and Surface Speciation Reactions used in Thermodynamic Modeling Appendix B: U(VI) Speciation Diagrams Appendix C: U(VI) Reduction Data Appendix D: Uranyl Surface Complexation Reactions for S. oneidensis MR Bibliography iii

8 FIGURES Figure 2.1: The concentrations of (a) total dissolved U and (b) dissolved U(VI) remaining in solution in the presence of different concentrations of EDTA (0.0, 0.5, 1.5, 3.0 & 5.0 mm) as a function of time. Symbols represent the average of triplicate experiments, and the error bars represent the associated standard deviation (1σ) of each value Figure 2.2: The concentrations of (a) total dissolved U and (b) dissolved U(VI) remaining in solution in the presence of different concentrations of dissolved Ca (0.0, 0.5, 1.5, 2.5 & 5.0 mm) as a function of time. Symbols represent the average of triplicate experiments, and the error bars represent the associated standard deviation (1σ) of each value Figure 2.3: Percentage of U(VI) adsorbed onto S. oneidensis MR-1 cells as a function of ph. Total concentration of U(VI) was M; ionic strength was 0.1 M NaClO 4. The open squares represent the experimental adsorption data for experiments with g (wet mass)/l cells. The solid squares represent experimental adsorption data for experiments with 0.25 g (wet mass)/l cells. The dashed and solid curves represent predicted extents of U(VI) adsorption, calculated using the set of averaged K values listed in Table Figure 2.4: Measured U(VI) reduction rates plotted as a function of the calculated concentration of the aqueous U 4+ -EDTA complex at the 2.3 h time point of the U(VI) reduction experiments with EDTA Figure 2.5: (a) Measured U(VI) reduction rates plotted as a function of the concentration of the Ca-uranyl-carbonate bacterial surface complex ([R-L2-Ca 2 UO 2 (CO 3 ) 3 ] - ). (b) Measured U(VI) reduction rates plotted as a function of the total concentration of the Ca-free uranyl surface complexes ([R-L2-UO 2 ] +, [R-L2-UO 2 CO 3 ] - and [R-L3- UO 2 (CO 3 ) 3 ] 5- ) Figure 3.1: The percentage of U(VI) adsorbed onto 1 g(wet mass)/l of S. oneidensis MR-1 in the presence of 0.1 M NaClO 4 and varying concentrations of NaHCO 3 as a function of ph. Total concentration of uranium was 0.25 mm in each experiment; concentrations of NaHCO 3 were 0.0, 0.2, 2.4, 11.9 and 30.0 mm iv

9 Figure 3.2: The open diamonds represent experimental adsorption data for experiments with 0.0 mm NaHCO 3. The dashed and solid curves represent predicted extents of U(VI) adsorption Figure 3.3: The open diamonds represent experimental adsorption data for experiments with 0.2 mm NaHCO 3. The dashed curve represents predicted extents of U(VI) adsorption using the set of K values calculated from the 0.0 mm NaHCO 3 dataset (Table 3.1). The solid curve represents predicted extents of U(VI) adsorption using the set of K values listed in Table 3.1 for the 0.2 mm NaHCO 3 dataset Figure 3.4: The open diamonds represent experimental adsorption data for experiments with (a) 2.4 mm (b) 11.9 mm and (c) 30.0 mm NaHCO 3. The dashed curves represent predicted extents of U(VI) adsorption using averaged K values calculated from datasets of 0.0 and 0.2 mm NaHCO 3 listed in Table 3.1. The solid curves represent predicted extents of U(VI) adsorption using the set of K values listed in Table 3.1 for each dataset of 2.4 mm, 11.9 mm and 30.0 mm NaHCO Figure 3.5: The open diamonds represent experimental adsorption data for experiments with (a) 0.0 mm (b) 0.2 mm (c) 2.4 mm (d) 11.9 mm and (e) 30.0 mm NaHCO 3. The solid curves represent predicted extents of U(VI) adsorption using the set of average K values listed in Table Figure 4.1: The concentrations of dissolved U(VI) remaining in solution in the presence of (a) 2.4 mm; (b) 5.0 mm; (c) 7.2 mm; (d) 11.9 mm; (e) 21.0 mm; (f) 30.0 mm NaHCO 3 as a function of time. In each figure, open triangles represent the nonacidified dataset in which the sample was immediately analyzed for concentration of remaining U(VI) at each sampling time; solid diamonds represent the acidified dataset in which the sample was acidified to approximately ph 1.5 for ~40 min before analysis of U(VI) concentration; open squares represent the data points used to calculate the initial rate of U(VI) reduction by S. oneidensis Figure 4.2: The average initial rate of U(VI) reduction by S. oneidensis as a function of NaHCO 3 concentration in the experiments. The error bars represent the associated standard deviation (1σ) of each value Figure 4.3: The total concentration of U(VI) adsorbed onto the cell wall of S. oneidensis as a function of NaHCO 3 concentration in the experiments Figure 4.4: Average initial U(VI) reduction rates as a function of the total concentration of U(VI) adsorbed onto the cell wall of S. oneidensis. The error bars represent the associated standard deviation (1σ) of each value Figure B.1: Speciation diagrams for systems with a total U(VI) concentration of 0.25 mm, and NaHCO3 concentrations of: (a) 0.0, (b) 0.2, (c) 2.4, (d) 11.9 and (e)30.0 mm v

10 in 0.1 M NaClO4 solution. Only species with concentrations greater than 5% of the total U(VI) concentration are shown Figure C.1: The concentrations of dissolved U(VI) remaining in solution in the presence of (a, b) 2.4 mm; (c, d) 5.0 mm; (e, f) 7.2 mm; (g, h) 11.9 mm; (i, j) 21.0 mm; (k, l) 30.0 mm NaHCO 3 as a function of time. In each figure, open triangles represent the non-acidified dataset; solid diamonds represent the acidified dataset; open squares represent the data points used to calculate the initial rate of U(VI) reduction by S. oneidensis vi

11 TABLES Table 2.1 Uranyl Surface Complexation Reactions for both S. oneidensis and B. subtilis 33 Table 2.2 Calculated uranyl surface complexes formed in the Ca experiments Table 3.1 Calculated Log K values for uranyl surface complexes formed on the cell wall of S. oneidensis MR Table 4.1 Calculated uranyl surface complexes and average reduction rate for each NaHCO 3 concentration Table A.1 Aqueous and Surface Speciation Reactions used in Thermodynamic Modeling.93 Table A.1 Aqueous and Surface Speciation Reactions used in Thermodynamic Modeling.94 Table A.1 Aqueous and Surface Speciation Reactions used in Thermodynamic Modeling.95 Table D.1 Uranyl Surface Complexation Reactions for S. oneidensis MR vii

12 ACKNOWLEDGMENTS I would like to thank my advisor, Dr. Jeremy Fein, for his guidance, continuous encouragement and support during my graduate studies. I also would like to thank the members of my dissertation committee, Dr. Patricia Maurice, Dr. Joshua Shrout, Dr. Peter Burns, and the member of my proposal committee, Dr. Chongzheng Na, for their time and helpful advice. I thank Jennifer Szymanowski for technical help with the experiments. Part of the analyses for this research were conducted using equipment in the Center for Environmental Science and Technology (CEST) at University of Notre Dame, technical support from the staff at CEST, Dennis Birdsell, Jon Loftus, and Suzyanne Guzicki, is appreciated. I am also grateful to the members of Fein group for their support and friendship. A special thanks goes to my family, especially my husband, Dr. Yanghai Yu. This research was supported by a NSF Environmental Molecular Science Institute grant to University of Notre Dame (EAR ), and a CEST Bayer Pre-doctoral Fellowship to L.S., as well as a University of Notre Dame Sustainable Energy Initiative grant to J.F. viii

13 CHAPTER 1: INTRODUCTION 1.1 Overview The biotic ligand model (BLM) has been developed to estimate the acute toxicity of metals to organisms (e.g., Di Toro et al., 2001; Santore et al., 2001; Paquin et al., 2002). Basically, the BLM relates the toxicity or bioavailability of a metal to an organism to the concentration of that metal that is bound to a generic biotic ligand on the organism in question. The foundation of the BLM is the assumption that bioavailability of a metal to an organism is controlled by adsorption of the metal onto some binding site on the organism. The BLM accounts for inhibition effects on toxicity from the aqueous organic and inorganic metal-ligand complexation as well as the effects of competing cations (e.g., Mg 2+, Na +, H +, etc.) on metal toxicity. However, as originally formulated, the generic biotic ligand of the BLM is not defined stoichiometrically and hence the equilibrium constant for the metal adsorption reaction involving this generic, unspecified biotic ligand is not a true constant, but varies as a function of system parameters (e.g., ph, ionic strength, solution composition, etc.). As a result, predictions of metal toxicity or bioavailability under more complex geologic settings by the BLM may be highly inaccurate. In my dissertation research, I expand on the BLM approach and apply it to bacteria by replacing the generic ligand binding reactions with adsorption reactions that 1

14 accurately portray cell wall binding using a surface complexation modeling approach. Surface complexation models (SCM) have been developed to model metal adsorption onto bacterial cell walls (e.g., Plette et al., 1996; Fein et al., 1997; Cox et al., 1999; Daughney and Fein, 1998; Fein, 2000; 2006). Instead of treating metal binding onto an organism using one generic unspecified ligand as in traditional BLMs, my proposed approach involves mechanistically accurate metal binding reactions involving each different functional group type that is exposed on the bacterial cell wall. The advantage of the SCM approach is that the binding reactions are balanced chemical reactions that occur on a molecular scale, and hence the binding constants that describe the adsorption are true thermodynamic equilibrium constants that do not vary with system conditions other than pressure and temperature, and hence can be applied to predict bioavailability in systems that are more complex than have been studied directly in the laboratory. The stoichiometry of the important cell wall adsorption reactions can be determined by several approaches (Hennig et al., 2001; Kelly et al., 2002; Boyanov et al., 2003; Burnett et al., 2006; Guine et al., 2006; Mishra et al. 2007). The approach that I use represents an advanced BLM in that the metal binding is characterized in a more realistic and sophisticated manner than in the traditional BLM. I use this approach to test whether the adsorption of metal ions onto bacterial cell walls represents the first step of bioavailability of these metal ions. The quantitative prediction of the concentrations and speciation of metal ions on the cell surface based on SCM enables the examination of those possible correlations. If those correlations can be confirmed, 2

15 the SCM may be applied to predict the bioavailability of metal ions under a wide range of geologic settings. Previous studies have suggested that metal bioavailability may be linked to the extent and speciation of metal adsorption onto bacterial cell walls (e.g., Pawlik et al., 1993; Franklin et al., 2000; Borrok et al., 2005a; VanEngelen et al., 2010). For example, Borrok et al. (2005a) found that the chemotactic response of Escherichia coli away from aqueous Ni is strongly correlated to the extent of Ni adsorbed onto cell wall functional groups. Franklin et al. (2000) also reported that the toxicity of U and Cu to bacteria was positively related to the concentration of metal that binds to bacterial cells. However, our understanding of the possible correlations between the speciation of metals on the bacterial cell surface and the relevant metabolic processes is still poor. Metal ions bind to bacterial cell wall through complexation with deprotonated functional groups on the cell wall (e.g., Sherbert 1978; Beveridge and Murray, 1980; Fein et al., 1997). Numerous studies have successfully determined the molecular-scale mechanisms responsible for metal binding onto bacteria (e.g., Plette et al., 1996; Yee and Fein, 2001; Gorman-Lewis et al., 2005). Hence, we can take the next step of using this type of information to improve BLMs by incorporating SCM to describe metal binding. In addition, these advanced BLMs must be tested to determine if a correlation exists between metal adsorption and bacterial bioavailability of metal ions. My dissertation research projects test the hypothesis that the concentration and speciation of metals on the bacterial cell wall control some metabolic activities. The research addresses the following scientific questions: (1) Does uranyl adsorption and speciation on bacterial cell walls control the 3

16 rate of enzymatic U(VI) reduction by bacteria? (2) What are the important uranylbacterial surface complexes that control the U(VI) adsorption behavior under conditions with high concentrations of dissolved inorganic carbon? and (3) If uranyl speciation on the bacterial cell wall controls the rate of U(VI) reduction by bacteria, is it possible to derive a kinetic model based on that correlation? 1.2 Effects of U(VI) Speciation on the Rate of U(VI) Reduction by Bacteria Under anaerobic conditions, the mobility of U in the subsurface can be retarded by the reduction of soluble U(VI) species into insoluble U(IV) solid phases (Bonatti et al., 1971; Langmuir, 1978; Senko et al., 2002). Bacteria can reduce U(VI) to U(IV) through metabolic activity (Lovley et al., 1991; Lovley and Phillips, 1992a; Lovley et al., 1993). Metal- or sulfate-reducing bacterial species such as Geobacter metallireducens, Shewanella oneidensis and Desulfovibrio desulfuricans can couple the reduction of U(VI) to the oxidation of electron donors such as H 2 or lactate, and under laboratory conditions, the reduction of U(VI) to U(IV) can be both rapid and complete. Although considerable research has focused on the mechanisms of U(VI) reduction by bacteria (e.g., Brooks et al., 2003; Haas and Northup, 2004; Stewart et al., 2007; Suzuki et al., 2010), the controls on the kinetics of the reduction process are poorly defined. Borrok et al. (2005a) found that the chemotactic response of Escherichia coli to Ni 2+ in solution is directly related to the concentration of Ni 2+ adsorbed on the cell surface. It is likely that the reduction of U(VI) by bacteria is similarly controlled by the speciation and concentration of U on the bacterial cell surface. 4

17 In this study, I test if adsorption reactions control bacterial reductive precipitation of U(VI) to U(IV) and hence it is also crucial to quantify the tendency of U to bind to cell wall components. There have been no studies of U(IV) adsorption onto bacteria, but U(VI) adsorption onto bacteria has been examined and the thermodynamic stabilities of U(VI)-bacterial surface complexes have been quantified using a SCM approach (Fowle et al., 2000; Haas et al., 2001; Gorman-Lewis et al., 2005). In the study described in Chapter 2, I measured the U(VI) reduction rate by Shewanella oneidensis MR-1 in the presence of variable concentrations of EDTA or dissolved Ca, measuring both total dissolved U and aqueous U(VI) remaining in solution in order to discern U(VI) adsorption controls from U(IV) desorption controls on the reduction kinetics. I also measured the extent of U(VI) adsorption onto S. oneidensis in order to quantify the thermodynamic stabilities of important U(VI)-bacterial surface complexes, and I attempted to relate those stabilities to the observed reduction kinetics. 1.3 Controls on U(VI) Adsorption by Bacteria in the presence of Dissolved Inorganic Carbonate Metal adsorption onto bacterial surfaces not only can affect the mobility and transport of metals in geologic systems (e.g., Beveridge and Murray, 1976; Harvey and Leckie, 1985; Konhauser et al., 1993; Fein, 2000), but also may control the bioavailability of metal ions by bacteria (e.g., Borrok et al., 2005a; Sheng et al., 2011). Hence, a better understanding of uranyl-bacterial surface complexation reactions is needed not only to better constrain U partitioning in bacteria-bearing aqueous systems, but also to develop 5

18 quantitative models of the bacterial bioavailability of U. In addition, we need to better understand U(VI) binding to bacteria in the presence of high concentrations of dissolved inorganic carbonate (DIC) in order to improve carbonate leaching of U from the subsurface both for remediation and for primary U mining (Mason et al., 1997; Francis et al., 1999; IAEA, 1993). Surface complexation modeling (SCM) represents a flexible approach for quantifying the extent of metal adsorption onto bacterial cell walls (e.g., Plette et al., 1996; Fein et al., 1997; Ngwenya et al., 2003; Guiné et al., 2006), and hence can be used to predict the fate and transport of metals in water-rock systems (Fein, 2000). Several studies have examined the adsorption of U(VI) onto bacteria in the presence of DIC (Haas et al., 2001; Gorman-Lewis et al., 2005; Sheng et al., 2011), however, these previous experiments were conducted at only one DIC concentration and the measurements do not provide rigorous constraints on the stoichiometries of the important adsorbed uranyl complexes. In the study described in Chapter 3, I measured the adsorption of aqueous U(VI) onto Shewanella oneidensis strain MR-1 as a function of NaHCO 3 concentration in solution. The molal ratio of DIC : U concentration in this study ranged from 0 to 120 : 1. A thermodynamic surface complexation modeling approach was applied to interpret the experimental adsorption data in order to determine the dominant adsorption reactions and the stability constants for the important uranyl-bacterial surface complexes. The wide range of DIC concentrations that was studied is needed in order to more rigorously constrain the identities and stability constants of the important uranyl-carbonatebacterial surface species. It is particularly important to determine these parameters in 6

19 order to model the effects of bacterial adsorption of U(VI) and the bioavailability of U(VI) in a range of natural and engineered DIC-bearing aqueous systems. 1.4 Further Test of Correlations between the U(VI) Speciation and the Rate of U(VI) Reduction by Bacteria In order to understand and model the fate and mobility of U in groundwater systems and also to enhance the effectiveness of U bioremediation strategies, it is crucial to determine the controls on the enzymatic U(VI) bioreduction rate. A considerable amount of research has investigated the mechanisms of U(VI) reduction by bacteria (e.g., Brooks et. al, 2003; Haas and Northup, 2004; Neiss et al., 2007; Suzuki et al., 2010). However, the controls on the kinetics of the reduction process and the connection between aqueous complexation and its effect on U(VI) reduction rates are poorly defined, especially in complex realistic geologic systems. Previous studies suggest that U(VI) speciation and distribution in geologic systems affects U(VI) reduction rates (Ulrich et al., 2011; Stewart et al., 2011), but a better understanding of the relationship between U speciation and U bioavailability is needed in order to derive speciation-based kinetic rate laws. Sheng et al. (2011) proposed that the speciation of both U(VI) and U(IV) on bacterial cell walls controls the kinetics of U(VI) reduction by bacteria. Sheng et al. (2011) measured the rate of U(VI) reduction by Shewanella oneidensis in the presence of dissolved Ca and EDTA, and found strong correlations between the speciation and concentration of U on the cell surface and the U(VI) reduction rate. Other studies also suggest that metal bioavailability is linked to the extent and speciation of 7

20 metal adsorption onto bacterial cell walls (e.g., Pawlik et al., 1993; Franklin et al., 2000; Borrok et al., 2005a; VanEngelen et al., 2010). Dissolved bicarbonate strongly affects the aqueous speciation of U(VI), as well as the adsorption and speciation of U(VI) on the bacterial cell wall (Sheng and Fein, 2012). Hence, measuring the rate of U(VI) reduction by bacteria as a function of NaHCO 3 concentration in solution would rigorously test the hypothesis that the extent and speciation of U(VI) adsorbed onto bacterial cell walls controls the kinetics of enzymatic U(VI) reduction by bacteria. In this study, I used the stability constants for the important uranyl-bacterial complexes derived in Chapter 3 (Sheng and Fein, 2012) to calculate the concentration and speciation of U(VI) on the bacterial cell wall under each of the experimental conditions, and I test if a relationship exists between the calculated extent of U(VI) adsorption onto the bacterial cell walls and the observed U(VI) reduction rate. Determining if such a relationship exists provides critical insights into the controls on U(VI) bioavailability to bacteria, and is crucial in order to design effective bioremediation strategies based on enzymatic reduction of U(VI) in realistic geologic systems. 8

21 CHAPTER 2: THE EFFECTS OF URANIUM SPECIATION ON THE RATE OF U(VI) REDUCTION BY SHEWANELLA ONEIDENSIS MR Introduction The oxidation state of U is one of several factors that control the mobility and fate of U in geologic settings. U(VI), in the form of the uranyl cation UO 2+ 2, is the thermodynamically stable form of U under oxic conditions. Uranyl minerals exhibit markedly higher solubilities than U(IV) phases (Langmuir, 1997), so U mobility in aerobic environments can be high. However, in anaerobic settings, the mobility of U in the subsurface can be retarded by the reduction of soluble U(VI) species into insoluble U(IV) solid phases (Bonatti et al., 1971; Langmuir, 1997). Bacteria can reduce U(VI) to U(IV) through metabolic activity (Lovley et al., 1991; Lovley and Phillips, 1992a; Lovley et al., 1993). Metal- or sulfate-reducing bacterial species such as Geobacter metallireducens (Lovley et al., 1991), Shewanella oneidensis (Lovley et al., 1991) and Desulfovibrio desulfuricans (Lovley and Phillips, 1992a) can couple the reduction of U(VI) to the oxidation of electron donors such as H 2 and lactate, and under laboratory conditions, the reduction of U(VI) to U(IV) can be both rapid and complete. As a result, bioreduction of U(VI) may represent a viable remediation approach for anaerobic 9

22 groundwater systems contaminated with U(VI) (Lovley and Phillips, 1992b; Finneran et al., 2002; Anderson et al., 2003; Wu et al., 2007). Although considerable research has focused on the mechanisms of U(VI) reduction by bacteria, the controls on the kinetics of the reduction process are poorly defined. Brooks et al. (2003) found that the presence of dissolved Ca could decrease the rate and extent of U(VI) reduction by Shewanella oneidensis, Desulfovibrio desulfuricans 0 and Geobacter sulfurreducens, likely due to the formation of the aqueous Ca 2 UO 2 (CO 3 ) 3 complex. Inhibition of U(VI) reduction rates by bacteria in the presence of dissolved Ca was also observed by Stewart et al. (2007) and by Neiss et al. (2007). Haas and Northup (2004) measured the removal rate of U(VI) by Shewanella oneidensis in the presence of a range of multi-dentate organic acids, and found that the initial rates of U removal from solution decreased with increasing stability constant values for the 1:1 aqueous U(VI):ligand and U(IV):ligand complexes. Because Haas and Northup (2004) only measured total U remaining in solution and not U(VI) specifically, it remains unclear whether the observed effects were caused by ligand-retarded U(VI) reduction due to aqueous U(VI)-ligand complexation, by ligand-promoted dissolution of the solid phase UO 2 that formed on the bacteria, by ligand-promoted desorption of U 4+ from the bacterial cell wall before UO 2 could form, or by a combination of these processes. Furthermore, Suzuki et al. (2010) demonstrated that strong complexing ligands like citrate, NTA and EDTA retard UO 2 precipitation in U(VI) bioreduciton experiments by forming aqueous complexes with U(IV). Behrends and Van Cappellen (2005) found that U(VI) speciation in systems containing both bacteria and hematite nanoparticles control 10

23 the pathway and kinetics of U(VI) reduction. The results of these previous studies suggest that the cell wall speciation of U may control the kinetics and extent of reductive precipitation of U(VI) by bacteria. Metal adsorption represents an important mechanism in a number of other bacterial metabolic processes as well. For example, Borrok et al. (2005a) demonstrated that the chemotactic response of Escherichia coli away from aqueous Ni is directly related to the concentration of Ni adsorbed onto the bacterial cell wall. We propose that the cell wall speciation of U similarly controls U(VI) reduction by bacteria. If adsorption/desorption reactions control bacterial reductive precipitation of U(VI) to U(IV), then it is crucial to quantify the tendency of U to bind cell wall components. There have been no studies of U(IV) adsorption onto bacteria, but U(VI) adsorption onto bacteria has been examined and the thermodynamic stabilities of U(VI)- bacterial surface complexes have been quantified (Fowle et al., 2000; Haas et al., 2001; Gorman-Lewis et al., 2005). Adsorption of U(VI) onto bacterial cell walls is controlled by complexation of aqueous uranyl ions and aqueous uranyl complexes with phosphoryl and carboxyl functional groups within the bacterial cell wall (Kelly et al. 2002). Fowle et al. (2000) observed extensive adsorption of UO 2+ 2 onto the Gram-positive species Bacillus subtilis under low ph conditions, with increasing U(VI) adsorption with increasing ph up to approximately ph 5. Gorman-Lewis et al. (2005) extended these observations to higher ph conditions and demonstrated that U(VI) could adsorb to the bacterial cell wall extensively, even under conditions where aqueous uranyl-carbonate, uranyl-hydroxide, and Ca-uranyl-carbonate complexes dominate the aqueous U(VI) 11

24 budget. These studies use a surface complexation approach to determine stability constants for the important U(VI)-bacterial surface complexes, and these constants can serve as a basis for understanding the role of U(VI) adsorption in controlling bioreduction of U(VI). In this study, we measured the kinetics and extent of U(VI) reduction by Shewanella oneidensis MR-1 in the presence of variable concentrations of EDTA or dissolved Ca, measuring both total dissolved U and aqueous U(VI) remaining in solution in order to discern U(VI) adsorption controls from U(IV) desorption controls on the reduction kinetics. We also measured the extent of U(VI) adsorption onto S. oneidensis in order to quantify the thermodynamic stabilities of important U(VI)-bacterial surface complexes, and we relate those stabilities to the observed reduction kinetics. 2.2 Experimental Methods U(VI) Reduction Experiments Bacteria Preparation Shewanella oneidensis strain MR-1 was grown aerobically following procedures described previously (Fein et al., 1997; Fowle and Fein, 2000). Cells were maintained on agar plates made of trypticase soy agar with 0.5% yeast extract. Cells were first transferred from the agar plate to a tube containing 3 ml of sterile trypticase soy broth (TSB) with 0.5% yeast extract. After being incubated at 32 o C for 24 h, the cell 12

25 suspension was transferred to 2 L of sterile TSB with 0.5% yeast extract and incubated at 32 o C for another 24 h. Cells of S. oneidensis were harvested by centrifugation at 10,970 g for 5 min. Cells were washed twice by resuspending them in 20 ml of sterile anoxic 0.1 M NaCl before the reduction experiments conducted with EDTA, or in 20 ml of sterile anoxic 30 mm NaHCO3 before the reduction experiments with Ca. Between each wash, cells were pelleted by centrifugation at 8,100 g for 5 min. After the washes, the cells were resuspended in 10 ml of sterile anoxic 0.1 M NaCl (for the EDTA experiments) or 30 mm NaHCO 3 (for the Ca experiments) in order to create a concentrated parent cell suspension that was used in the U(VI) reduction experiments U(VI) Reduction Experiments with EDTA The anaerobic procedures and the general composition of the experimental medium that we used for the EDTA experiments were similar to those described previously (Balch and Wolfe, 1976; Lovley et. al, 1991; Haas and Northup, 2004). The experimental medium was consisted of 5 mm NH 4 Cl, 1 mm KCl, 30 mm NaHCO 3, 40 mm Na-lactate, and vitamins solution and trace elements as per Lovely and Phillips (1988). Different amounts of EDTA disodium salt were added to the experimental medium in order to achieve final EDTA concentrations of 0.0, 0.5, 1.5, 3.0 and 5.0 mm. The ph of each experimental medium was adjusted to 7.0 using small aliquots of concentrated NaOH and/or HCl. After being heated on a hotplate, and bubbled with an 85% N 2 /10% CO 2 /5% H 2 gas mixture for ~15 min, the experimental medium was transferred to an 13

26 anaerobic glovebox chamber, containing an atmosphere with the same 85% N 2 /10% CO 2 /5% H 2 gas composition. Sterile serum bottles were filled with 50 ml of experimental medium each, sealed inside the glove box, and then autoclaved outside the glovebox at 120 o C for 20 min. The ph of each medium with different EDTA concentrations was checked to be around 7.0 after being autoclaved. A uranyl acetate stock was prepared with UO 2 (CH 3 COO) 2 2H 2 O and ultrapure 18 MΩ water and its concentration was determined by inductively coupled plasma optical emission spectroscopy (ICP-OES). The U(VI) stock solution was bubbled with the 85% N 2 /10% CO 2 /5% H 2 gas mixture for ~30 min, then filter-sterilized (0.2 µm) and injected into a sterile evacuated serum bottle inside the anaerobic chamber. The filter-sterilized (0.2 µm) uranyl acetate stock was added to each experimental serum bottle to achieve an initial dissolved U(VI) concentration of 0.5 mm. Two ml of solution was transferred from each experimental bottle to measure the initial U(VI) and total dissolved U concentrations by fluorescence spectrometry and ICP-OES, respectively. For each concentration (0.0, 0.5, 1.5, 3.0 and 5.0 mm) of EDTA in the reduction experiments, there were three replicate sample bottles injected with 200 µl of the parent cell suspension and two cell-free control bottles with the same aqueous composition but without bacteria. The cell-free control experiments were to test if any U(VI) reduction or any loss of total U could be attributed to abiotic factors under the experimental conditions. During the experimentals, all of the sample bottles and the control bottles were agitated gently and continuously at room temperature (~25 o C). To determine the initial cell density, 0.5 ml of well-mixed cell suspension was removed 14

27 from three sample bottles right after injection of bacteria, 50 µl of formalin was added, and cell density was determined through direct cell counting using a Petroff-Hausser cell counting chamber. The average initial cell density in the EDTA experiments was 3.0(±0.7) 10 7 cells/ml. At selected sampling times, approximately 3.5 ml of sample solution was removed under anaerobic and sterile conditions from each bottle during the initial 2.3 h of the experiment. 0.5 ml of each 3.5 ml sample was filtered through a 0.2 µm Millipore Millex PTFE filter and analyzed for dissolved U(VI) concentration by fluorescence spectrometry. A 2.5 ml aliquot of each 3.5 ml sample was filtered through a 0.2 µm Millipore Millex PTFE filter and acidified with 4.5 µl of concentrated HNO 3 (15.8 N) for total dissolved U analysis by ICP-OES. The concentration of U(IV) in the experimental systems was calculated by difference between the measured total U and dissolved U(VI) concentrations U(VI) Reduction Experiments with Ca The experimental medium for the U(VI) reduction experiments with dissolved Ca consisted of 30 mm NaHCO 3 and 50 mm Na-lactate (Brooks et al., 2003). The ph of the experimental medium was adjusted to 7.0 using small aliquots of concentrated NaOH and/or HCl. The ph adjustment did not significantly change the ionic strength of the experimental solutions. The experimental serum bottles containing 50 ml of sterile anoxic experimental medium were prepared the same way as described in the EDTA experiments part. A Ca stock solution was prepared with CaCl 2 2H 2 O and ultrapure 18 15

28 MΩ water and its concentration was determined by ICP-OES. The Ca stock solution was stirred and placed inside the anaerobic chamber overnight, then filter-sterilized (0.2 µm) and injected into a sterile evacuated serum bottle inside the anaerobic chamber. Aliquots of the Ca stock solution were injected into the experimental serum bottles to achieve experimental Ca concentrations of 0.0, 0.5, 1.5, 2.5 and 5.0 mm. The ph of the medium of each Ca concentration was measured to be around ph 7.0 after addition of Ca. A initial 0.5 mm of U(VI) in each experimental bottle was achieved by following the same procedures used for the EDTA experiments. Five bottles were prepared the same way as the EDTA experiments for each concentration of Ca: three replicate sample bottles were injected with parent cell suspension; the other two bottles were cell-free controls. The determination of the initial aqueous U(VI) and total dissolved U concentrations and the average initial cell density [3.2(±0.6) 10 7 cells/ml], and also the sampling procedure for the initial 5 h after cell injection were the same as the EDTA experiments. During the experimental course, all the sample bottles and the control bottles were agitated gently and continuously at room temperature (~25 o C) U(VI) Adsorption Experiments Bacteria Preparation S. oneidensis MR-1 was grown following the same procedures described above. The bacteria were washed following the procedure outlined in previous work (Fein et al., 1997; Fowle et al., 2000). After 48 h growth, cells of S. oneidensis were harvested by centrifugation at 10,970 g for 5 min. Then the cells were rinsed 5 times in 0.1 M NaClO 4 16

29 with a centrifugation step of 8,100 g for 5 min between each rinse. The cells were then centrifuged twice at 8,100 g for 30 min, pouring off the supernatant after each centrifugation, in order to determine the wet mass of S. oneidensis to be used in the experiments. The wet mass is approximately 8 times the dry mass of the biomass (Borrok et al., 2005b). We assume that no U(VI) reduction occurred during these aerobic adsorption experiments, an assumption consistent with the results of Kelly et al. (2002) who conducted an EXAFS study of U(VI) adsorption onto Bacillus subtilis bacterial cells U(VI) Adsorption Experiments For the U(VI) adsorption experiments, an aqueous filter-sterilized uranyl acetate stock solution was prepared as described above. Aliquots of the U stock solution were added to a 0.1 M NaClO 4 electrolyte solution to form a U(VI)-electrolyte solution containing ~10 ppm (~ M) of U(VI). Before adding bacteria, this solution was sampled to determine the initial total dissolved U concentration by ICP-OES. Then in each of two separate Teflon bottles, a weighed volume of a parent suspension that contained washed S. oneidensis cells in 0.1 M NaClO 4 was added to the 10 ppm U(VI)- electrolyte solution to form two bacteria-u(vi)-electrolyte suspensions: one with g/l (wet mass) of S. oneidensis and another with 0.25 g/l of S. oneidensis. The ph values of these two bacteria-u(vi)-electrolyte suspensions were adjusted to between 6~8 using aliquots of concentrated HNO 3 and NaOH. A number of 8 ml aliquots were taken from the parent suspensions and placed in Teflon tubes. The ph of each individual suspension 17

30 was adjusted to cover the ph range of 3~9 using small volumes of HNO 3 and NaOH. After the ph adjustment, all the reaction tubes were placed on a rotating rack and gently agitated for 3 h. The final ph of each suspension was measured, and each suspension was then filtered through a 0.2-µm Millipore Millex PTFE filter and acidified with concentrated HNO 3 for ICP-OES analysis. The concentration of the adsorbed U was calculated by difference between the measured initial and final aqueous U(VI) concentrations Analytical Methods Total Dissolved U Analysis by ICP-OES A Perkin-Elmer Optima 2000DV ICP-OES system was used to determine total dissolved U in solution. Matrix-matched blanks and standards covering the probable range of U in solution were prepared. The standards including the blank were reanalyzed after running every 30~40 samples in order to check machine drift. Analytical uncertainty was approximately ± 2%, as determined by repeat analyses of an aqueous U standard, and the operational detection limit for U on the ICP-OES was determined to be approximately 60 ppb Dissolved U(VI) Analysis by Fluorescence Spectrometry A PTI Quantamaster QM-4 spectrofluorometer was used to measure the phosphorescence decay of U(VI) in order to determine the concentration of U(VI) in solution, following the general approach and principles described in previous studies 18

31 (Gorby and Lovley, 1992; Brina and Miller, 1992). The spectrofluorometer system uses a xenon flash lamp as an excitation source and exhibits a linear dynamic range for aqueous U(VI) concentrations from 0.05 mm to 1.5 mm. The spectrofluorometer measures phosphorescence decay by recording the change in intensity of the phosphorescence signal emitted by excited U(VI) atoms in the sample as a function of 2+ time. UO 2 complexing agents are added to aqueous samples to prevent the uranyl ion from quenching after excitation (Sill and Peterson, 1947). Each sample (0.5 ml) was filtered, acidified with 0.25 ml of 12.1 N HCl, and diluted 200 times with ultrapure 18 MΩ water. 1.5 ml of Uraplex (the complexing agent) was then added to 1 ml of diluted sample and the solution was analyzed immediately on the spectrofluorometer, using an excitation wavelength of 420 nm, an emission wavelength of 515 nm and slit width of 17 nm. Matrix-matched blanks and standards covering the probable range of U(VI) in solution were measured to construct a calibration curve and to quantify the U(VI) concentrations in the samples. Analytical uncertainty was approximately ± 4.5%, as determined by repeat analyses of an aqueous U(VI) standard, and the detection limit for U(VI) under these analytical conditions is approximately 0.05 mm. 2.3 Results Results of U(VI) Reduction Experiments with EDTA Figure 2.1 illustrates the results of U(VI) reduction by S. oneidensis MR-1 in the presence and absence of EDTA, depicting the measured concentrations of total dissolved U and aqueous U(VI) as a function of time in Figure 2.1a & 2.1b, respectively. 19

32 There was no measurable loss of total dissolved U or aqueous U(VI) in the cell-free control experiments (data not shown), indicating no significant U(VI) reduction or adsorption onto the experimental apparatus under the experimental conditions. In the experiments without EDTA, the concentration of total dissolved U decreased steadily from an initial concentration of 0.52 mm to 0.45 mm after 2.3 h (Figure 2.1a). The addition of EDTA to the system caused the total dissolved U concentrations to remain unchanged over the course of the experiment for each of the EDTA concentrations studied (Figure 2.1a). Additionally, all of the experiments that contained EDTA exhibited respectively clear solutions with no visible black U(IV)O 2 precipitate. Conversely, a black U(IV)O 2 precipitate formed in the system that did not contain EDTA. These results are consistent with previous studies such as those by Haas and Northup (2004) and Suzuki et al. (2010). 20

33 Figure 2.1: The concentrations of (a) total dissolved U and (b) dissolved U(VI) remaining in solution in the presence of different concentrations of EDTA (0.0, 0.5, 1.5, 3.0 & 5.0 mm) as a function of time. Symbols represent the average of triplicate experiments, and the error bars represent the associated standard deviation (1σ) of each value. 21

34 In contrast to the total dissolved U concentrations, the concentration of U(VI) in solution decreased steadily over the course of the 2.3 h experiments for all EDTA concentrations studied (Figure 2.1b). In the system without EDTA, the U(VI) concentrations decreased from 0.52 mm to 0.24 mm during the experiment. The total dissolved U concentrations were significantly higher than the U(VI) concentrations at each sampling time, suggesting that some of the U(IV) that was produced during the experiment passed through the filtration membrane and contributed to the total dissolved U concentration. The U(IV) that passed through the filter was likely in the form of UO 2 particles in the EDTA-free experiments and in the form of aqueous EDTA-U(IV) complexes in the EDTA-bearing systems. The U(VI) concentration profiles are direct measurements of the extent of U(VI) reduction and are unaffected by the presence of U(IV) in the samples, and these values alone were used to determine U(VI) reduction rates. Despite the presence of U(IV) species in the experimental samples, the total dissolved U measurements can still be used qualitatively to constrain reduction mechanisms. In general, the presence of EDTA yielded a faster rate of U(VI) reduction, with decreasing U(VI) concentrations at a given sampling time with increasing EDTA concentration. For example, after 2.3 h, the 0.0, 3.0, and 5.0 mm EDTA experiments contained 0.24, 0.13, and 0.04 mm of U(VI) in solution, respectively. The magnitude of the error bars in Figure 2.1b demonstrates that the experiment with the lowest concentration of EDTA (0.5 mm EDTA) does not exhibit a significantly different reduction rate compared to the EDTA-free controls. Additionally, the 0.5 mm EDTA 22

35 dataset in Figure 2.1b appears to exhibit a significantly longer lag phase prior to significant reduction of U(VI), however the precise timing of this lag phase is difficult to constrain due to the experimental uncertainties coupled with the slow reduction rate of these experiments Results of U(VI) Reduction Experiments with Ca The presence of dissolved Ca exerted an opposite effect to that of EDTA on the rate of U(VI) reduction (Figure 2.2a & 2.2b). The experiment without dissolved Ca exhibited a steady decrease in total dissolved U concentration over the 5 h experiment, from an initial concentration of 0.49 mm to 0.24 mm after 5 h (Figure 2.2a). In general, with increasing dissolved Ca concentration in the experiment, the rate of decrease in the concentration of total dissolved U slowed (Figure 2.2a). The U(VI) concentrations in the Ca experiments exhibited a slower rate of reduction with increasing dissolved Ca concentrations (Figure 2.2b). The experiment without dissolved Ca exhibited a steady decrease in U(VI) concentrations with time, and as in the EDTA experiments, the U(VI) concentrations in solution at any given time during the experiment were lower than the measured total dissolved U concentrations, likely due to the presence of UO 2 particles in the aqueous samples. The presence of Ca significantly slowed the rate of U(VI) reduction. For example, the concentrations of U(VI) in solution after 5 h in the 0.0, 0.5, and 5.0 mm Ca experiments were 0.12, 0.23, and 0.42 mm. Though our initial U(VI) concentration was much higher than that used by Brooks et al. (2003), our Ca experiments are consistent with their results in that the presence of dissolved Ca significantly slows the observed U(VI) bioreduction rate. 23

36 Figure 2.2: The concentrations of (a) total dissolved U and (b) dissolved U(VI) remaining in solution in the presence of different concentrations of dissolved Ca (0.0, 0.5, 1.5, 2.5 & 5.0 mm) as a function of time. Symbols represent the average of triplicate experiments, and the error bars represent the associated standard deviation (1σ) of each value. 24

37 2.3.3 Results of U(VI) Adsorption Experiments The results of our U(VI) adsorption experiment (Figure 2.3) indicate that S. oneidensis exhibits similar U(VI) adsorption behavior as a function of ph to Bacillus subtilis (Fowle et al., 2000; Gorman-Lewis et al., 2005). We observed extensive U(VI) adsorption onto S. oneidensis over the entire ph range studied (ph 3~9). The extent of U(VI) adsorption increased from ph 3 to 5, remained relatively constant from ph 5 to 6.5, and decreased slightly above ph 6.5. The percent of total U adsorbed onto the bacteria varied only between 29% and 75% in the presence of 0.25 g/l bacteria and between 62% and 98 % in the presence of g/l bacteria over the entire ph range studied. Increasing the bacterial concentration in the adsorption experiments from 0.25 to g/l bacteria increased the extent of U adsorbed onto the bacterial cells by approximately 30% across the ph range studied. 25

38 Figure 2.3: Percentage of U(VI) adsorbed onto S. oneidensis MR-1 cells as a function of ph. Total concentration of U(VI) was M; ionic strength was 0.1 M NaClO 4. The open squares represent the experimental adsorption data for experiments with g (wet mass)/l cells. The solid squares represent experimental adsorption data for experiments with 0.25 g (wet mass)/l cells. The dashed and solid curves represent predicted extents of U(VI) adsorption, calculated using the set of averaged K values listed in Table

39 2.4 Discussion Effect of EDTA on the Rate of U(VI) Reduction The addition of EDTA to the solutions in the U(VI) reduction experiments could cause two possible competing effects: 1) aqueous UO EDTA complexation could sequester U(VI) away from the bacteria in the system prior to reduction, causing a decrease in the rate of U(VI) reduction during the experiments; or 2) after U(VI) reduction to U(IV) on the cell wall, aqueous U 4+ -EDTA complexation could draw U(IV) away from bacterial cell wall sites of reduction, speeding the reduction rate by freeing sites for more U(VI) to adsorb and become reduced. As pointed out by Haas and Northup (2004), the total dissolved U measurements do not distinguish between these two controls on the reduction rate as either mechanism would result in enhanced total dissolved U with increasing EDTA concentration. In the first case, the total dissolved U would be present as U(VI), and in the latter case, the aqueous U would be present as an aqueous U 4+ -EDTA complex. However, the U(VI) measurements clearly demonstrate that EDTA increases the rate of U(VI) reduction in these systems, and that the reduction rate increases with increasing EDTA concentration. Our observations are consistent only with the formation of aqueous U 4+ -EDTA complexes, which prevent the precipitation of UO 2, and maintain virtually all of the U in solution. The results of this investigation also suggest that in the EDTA experiments the rate controlling step in the enzymatic reduction of U(VI) is not the adsorption or desorption rate of U(VI). If the U(VI) speciation controlled the reduction rate, then the reduction rate would decrease with increasing EDTA concentration due to the sequestration of U(VI) in solution by aqueous 27

40 EDTA complexes. We observed the opposite effect, which is best explained by aqueous U(IV)-EDTA complexation. Our Ca experiments demonstrate an opposite effect to that of EDTA in that increasing concentrations of Ca in the experimental solutions decrease the rate of disappearance of both total dissolved U and aqueous U(VI). The most likely explanation for this observation is that the presence of Ca promotes the formation of aqueous Cauranyl-carbonate complexes and sequester U(VI) away from the bacterial cells. Although these complexes can adsorb onto the bacterial cell wall, the presence of Ca may decrease the net amount of adsorbed U, sequestering U(VI) away from the cell wall, and thereby slowing U(VI) reduction. The effect of Ca on U(VI) reduction is opposite to that of EDTA: EDTA forms more stable complexes with U 4+ than it does with UO 2+ 2, so the presence of EDTA in the experiments affects the speciation of the reduction product more than it does the speciation of U(VI) in solution, and hence speeds reduction. Conversely, Ca does not affect U(IV) speciation or the solubility of UO 2. Therefore, the addition of Ca to the experimental systems does not affect the removal rate of U(IV) from the cell wall or make reduction sites more available after the reduction of U(VI) to U(IV). However, increasing Ca concentration does affect the aqueous speciation of U(VI), promoting the formation of Ca-uranyl-carbonate aqueous complexes which sequester more U(VI) away from the bacterial cells. The decrease in U(VI) concentration on the cells with increasing Ca concentration appears to slow U(VI) reduction. Our results suggest that the bacterial cell wall complexation of both U(VI) before reduction and U(IV) after reduction controls U(VI) reduction rates in our experimental 28

41 systems. There have been no measurements of U 4+ adsorption onto bacteria, so we can not quantify the effect of EDTA on the U(IV) speciation on the cell wall. However, as a proxy for models of U 4+ speciation on the cell wall, we relate the U(VI) reduction rate in the EDTA experiments to the concentration of the aqueous U 4+ -EDTA complex (Figure 2.4). By assuming a pseudo-first-order reaction rate, we calculated the initial reduction rate (mm/h) of U(VI) as the decrease in aqueous U(VI) concentration versus time, defined by the slope of the best-fitting trendline of the datasets for each EDTA concentration in Figure 2.1b. The concentration of the aqueous U 4+ -EDTA complex was assumed to be the difference between the measured concentrations of total dissolved U and aqueous U(VI) for each EDTA concentration, and the data shown in Figure 2.4 represent the concentrations of the U 4+ -EDTA complex for each EDTA concentration at 2.3 h. The data from the experiments without EDTA were excluded from this treatment because most of the U(IV) that remained in the samples after filtration was likely present as UO 2 particles, whereas virtually all of the U(IV) in the EDTA-bearing systems was present as the aqueous U 4+ -EDTA complex. Figure 2.4 depicts a strong correlation between the U(VI) reduction rate and the concentration of the aqueous U 4+ -EDTA complex at 2.3 h. Similar strong correlations exist between the reduction rate and the concentration of the U 4+ -EDTA complex at the other sampling times (correlation not shown here). The strong relationship between the U(VI) reduction rate and the concentration of the aqueous U 4+ -EDTA complex supports our conclusion that EDTA controls the reduction rate by removing U 4+ from reduction sites on the bacterial cell wall. Similarly, Suter et al. (1991) argued that the mineral surface release rate of Fe is 29

42 the rate-controlling step in the reductive dissolution of Fe(III) (hydr)oxides. Our results also suggest that thermodynamic models of U(IV) speciation in realistic systems could yield reasonable predictions of the U(VI) reduction rates. U(VI) bioreduction rates are likely enhanced by ligands such as EDTA that preferentially bind with U(IV) relative to U(VI). For example, the presence of humic acid, which forms highly stable aqueous complexes with U(IV), also enhances the rate of U(VI) reduction by S. oneidensis (Gu et al., 2005) perhaps in part due to complexation effects. Figure 2.4: Measured U(VI) reduction rates plotted as a function of the calculated concentration of the aqueous U 4+ -EDTA complex at the 2.3 h time point of the U(VI) reduction experiments with EDTA. 30

43 2.4.2 Effect of Ca on the Rate of U(VI) Reduction Quantifying U(VI) adsorption onto S. oneidensis We use our U(VI) adsorption measurements to determine the speciation of U(VI) on S. oneidensis in order to determine if a relationship exists between cell wall uranyl speciation and U(VI) reduction rates in the U(VI) reduction experiments with Ca. We follow the modeling approach described by Gorman-Lewis et al. (2005) who measured U(VI) adsorption onto Bacillus subtilis and used the results to determine the stability constants for the important U(VI)-bacterial surface complexes. Metal adsorption measurements conducted as a function of ph constrain the number of sites involved in metal binding, the ph range of influence, and the stability constants for the important metal-bacterial surface complexes. We used the program FITEQL (Herbelin and Westall, 1994) for the equilibrium thermodynamic modeling of the U adsorption data, accounting for aqueous speciation using reactions 14 to 33 listed in Appendix A. Activity coefficients for ions were calculated within FITEQL using the Davies equation. A discrete pk a 4-site non-electrostatic model was used to model the protonation behavior of the S. oneidensis cell wall functional groups (Mishra et al., 2010). We refer to Sites 1-4 as the sites with pk a values of 3.3 ± 0.2, 4.8 ± 0.2, 6.7 ± 0.4, and 9.4 ± 0.5, respectively. The bacterial site density for each site was calculated according to the site densities of S. oneidensis described by Mishra et al. (2010), which are 8.9(±2.6) 10-5, 1.3(±0.2) 10-4, 5.9(±3.3) 10-5 and 1.1(±0.6) 10-4 mol per gram of wet mass for sites 1-4, respectively. 31

44 The adsorption experiments were conducted with the systems open to the atmosphere, so CO 2 in the aqueous systems was assumed to be in equilibrium with atmospheric CO 2. We deconvolved the adsorption reactions that control U(VI) adsorption across the ph range studied by first modeling only the low ph (from ph 3 to 5) adsorption. Below ph 5, U is present in solution dominantly as the UO 2+ 2 cation, so the range of likely adsorption reactions is restricted. A model with UO 2+ 2 binding onto deprotonated Site 2 best fits the low ph data, with a reaction stoichiometry and calculated K value given in Table 2.1. Because of the diminished importance of UO 2+ 2 in the aqueous U budget above ph 5 and the increased importance of uranyl-hydroxide and -carbonate complexes, Reaction 1 from Table 2.1 cannot account for the observed extent of U(VI) adsorption from ph 5 to 7. Therefore, we fix the K value for Reaction 1 to our calculated value, and test a range of adsorption reactions involving binding of the important uranyl-hydroxide and -carbonate complexes onto deprotonated forms of Sites 2 and 3 to account for the mid-ph adsorption behavior, and using the V(Y) output of FITEQL to quantify model fits. Reaction 2 in Table 2.1 yields the best-fit to data in the mid-ph region; the inclusion of Reaction 3 in Table 2.1 is required in order to account for the observed extents of adsorption between ph 7 and 9. We modeled the 0.25 g/l and the g/l U(VI) adsorption datasets separately, and both yielded best-fit models that include Reactions 1-3 in Table 2.1 (Figure 2.3). The values listed for the stability constants for these reactions in Table 2.1 are the averages of the two datasets of stability constant values, which were calculated based on 0.25 g/l and g/l adsorption data, respectively. The uncertainties listed in Table2.1 were calculated by 32

45 determining the range of K values that account for the observed range of experimental values for the extent of adsorption. The average K values for Reactions 1-3 were used to generate the model curves that are plotted in Figure 2.3. The figure demonstrates that the set of average K values can account for the data well both as a function of ph and as a function of bacterial concentration. TABLE 2.1 URANYL SURFACE COMPLEXATION REACTIONS FOR BOTH S. ONEIDENSIS AND B. SUBTILIS Log K (S. oneidensis) Log K (b) (B. subtilis) 1 UO R-L2 - (a) = [R-L2-UO 2 ] ± ± UO 2 CO R-L2 - (a) = [R-L2-UO 2 CO 3 ] ± ± UO 2 (CO 3 ) R-L3 - (a) = [R-L3-UO 2 (CO 3 ) 3 ] ± ± 0.6 (a) R-L#- represents S. oneidensis functional groups, Sites 1-4, with pk a values of 3.3 ± 0.2, 4.8 ± 0.2, 6.7 ± 0.4, and 9.4 ± 0.5, respectively (Mishra et al., 2010). (b) Gorman-Lewis et al. (2005). 33

46 The adsorption behavior that we observed for S. oneidensis is similar to that of B. subtilis. The log K values that we calculate for Reactions 1-3 for S. oneidensis are listed in Table 2.1, along with the corresponding log K values for B. subtilis reported by Gorman- Lewis et al. (2005). The calculated K values for Reactions 1 and 3 for S. oneidensis are in excellent agreement with corresponding K values for B. subtilis. The K values for Reaction 2 for these two bacterial species do not agree within uncertainties and may indicate enhanced adsorption by S. oneidensis under circumneutral ph conditions. The similarities in stability constants for Reactions 1 and 3 are consistent with the observations of similar Cd adsorption behavior and stability constants for a range of bacteria and bacterial consortia (Yee and Fein, 2001). Based on these similarities, we assumed that the stability constants for the important Ca-bacterial surface complexes for S. oneidensis (Reactions 1 and 2 in Appendix A) are the same as those for B. subtilis, which were determined previously by Gorman-Lewis et al. (2005) Relationship between U Cell Wall Speciation and U(VI) Reduction Rate The calculated stability constants for the uranyl-bacterial surface complexes enable calculation of the extent of U(VI) adsorption and the speciation of adsorbed U(VI) under the conditions of the U(VI) reduction experiments with Ca. These calculations enable tests of whether relationships exist between the observed reduction rates and the speciation of U(VI) on the bacterial cell wall in the Ca-bearing U(VI) reduction experiments. The initial reduction rates of U(VI) in the presence of different Ca concentrations were determined by calculating the slopes of the best-fitting trendlines 34

47 of the data for each Ca concentration in Figure 2.2b. The calculations of the U(VI) speciation in the systems of the U(VI) reduction experiments with Ca account for aqueous uranyl-hydroxide, -carbonate, -lactate, acetate, and bacterial surface complexation using the reactions and stability constants listed in Table 2.1 and Appendix A. The system was also constrained with mass balance constraints on dissolved carbonate, lactate, acetate, bacterial sites and U(VI) concentrations. Bacterial concentrations in the experiments were determined from the experimental suspensions, yielding an average experimental cell density of 3.2(±0.6) 10 7 cells/ml. This cell density was transformed into the wet mass density by dividing by a conversion factor of 1.9(±0.6) cells/g, which was determined based on our cell mass-cell counts transformation experiments, in which cells of specific wet mass were counted by direct cell counting method after suspended in specific volume of solution. Bacterial site concentrations were calculated using S. oneidensis site densities reported by Mishra et al. (2010). The calculated total binding site concentration was 0.61 mm for the Ca experiments. The U(VI) speciation in the system without Ca was calculated in a similar way to the procedures described above, but excluding all Ca-bearing reactions in Appendix A. The modeling results of the U(VI) reduction experiments with Ca indicate the presence and the concentrations of the important uranyl surface complexes, which are shown in Table

48 TABLE 2.2 CALCULATED URANYL SURFACE COMPLEXES FORMED IN THE CA EXPERIMENTS [Ca] (mm) [R-L2-UO 2 ] + (a) [R-L2- UO 2 CO 3 ] - (a) Uranyl Surface Complexes (mm) [R-L3- [R-L2-5- (a) UO 2 (CO 3 ) 3 ] Ca 2 UO 2 (CO 3 ) 3 ] - (a) Total Adsorbed U(VI) NONE (a) R-L#- represents S. oneidensis functional groups, Sites 1-4, with pk a values of 3.3 ± 0.2, 4.8 ± 0.2, 6.7 ± 0.4, and 9.4 ± 0.5, respectively (Mishra et al., 2010). 36

49 We use our modeling results to test whether the observed reduction rates correlate with either the total concentration of adsorbed U(VI), the concentration of adsorbed Ca-uranyl-carbonate species, or the total concentration of adsorbed uranyl complexes that do not involve Ca. There is no consistent relationship between the total adsorbed U(VI) concentration and the U(VI) reduction rates. Figure 2.5a depicts a strong negative correlation between the reduction rate and the concentration of the Ca-uranylcarbonate surface complex ([R-L2-Ca 2 UO 2 (CO 3 ) ] 3- ). The strong negative correlation suggests that the U(VI) that is present as this species on the bacterial cell wall is not available for reduction, and the higher the concentration of U(VI) bound as R-L2- Ca 2 UO 2 (CO 3 ) 3-, the slower the reduction rate. Conversely, Figure 2.5b illustrates a strong positive correlation between the observed rate of U(VI) reduction in the Ca experiments and the sum of the concentrations of the non-ca-uranyl-carbonate surface complexes ([R-L2-UO] 2+, [R-L2-UO 2 CO] 3- and [R-L3-UO 2 (CO 3 ) 3 ] 5- ). The positive correlation shown in Figure 2.5b strongly suggests that the only U(VI) that is available for reduction is the adsorbed U(VI) that exists on the cell wall as Ca-free uranyl surface complexes. Therefore, increasing the concentration of these Ca-free uranyl surface complexes also increases the rate of U(VI) reduction by S. oneidensis under the experimental conditions, and the concentration of these species on the bacterial cell wall is the primary control on the reduction rate. 37

50 Figure 2.5: (a) Measured U(VI) reduction rates plotted as a function of the concentration of the Ca-uranyl-carbonate bacterial surface complex ([R-L2-Ca 2 UO 2 (CO 3 ) 3 ] - ). (b) Measured U(VI) reduction rates plotted as a function of the total concentration of the Ca-free uranyl surface complexes ([R-L2-UO 2 ] +, [R-L2-UO 2 CO 3 ] - and [R-L3-UO 2 (CO 3 ) 3 ] 5- ). 38

51 We also considered the possibility that the Ca inhibition effect was due to competition for binding sites on the cell surface between U and Ca. Although the concentration of adsorbed Ca increases with increasing Ca in the experimental systems (modeling results not shown here), the surface sites are significantly undersaturated with bound Ca or U, and our modeling results of the bacterial surface speciation (Table 2.2) indicate that the concentration of total U(VI) adsorbed onto the bacteria does not change significantly as a function of Ca concentration in the experiments. That is, the Ca concentration affects the speciation of surface-bound uranium, but does not affect the total concentration of bound uranium through competitive adsorption. 2.5 Conclusions Our experiments demonstrate that EDTA increases the rate of U(VI) reduction by S. oneidensis and that dissolved Ca decreases the reduction rate. In the EDTA-presenting U(VI) reduction systems, EDTA maintains all of the uranium in solution by forming U 4+ - EDTA complex after reduction of U(VI) to U(IV). Also there is a strong relationship between the U(VI) reduction rate and the concentration of the aqueous U 4+ -EDTA complex in the system. So we conclude that EDTA speeds U(VI) bioreduction by removing U 4+ from reduction sites on the bacterial cell wall. We measured uranyl adsorption onto S. oneidensis and use the results to search for relationships between the reduction rate and the speciation of U(VI) on the bacterial cell wall in the systems of U(VI) reduction experiments with Ca. We find that the reduction rate is negatively correlated with the concentration of [R-L2-Ca 2 UO 2 (CO 3 ) ] 3- suggesting that U(VI) cannot 39

52 be reduced when present as the Ca-uranyl-carbonate complex on the bacterial cell wall. However, a strong positive correlation exists between the reduction rate and the concentration of uranyl-bacterial surface complexes that do not involve Ca. This finding suggests that U must be present on the cell wall as one of these complexes for reduction to occur, and the reduction rate is directly proportional to the concentration of these complexes on the cell wall. The findings of this study could enable quantitative predictions of U(VI) reduction rates in realistic geologic systems. The presence of aqueous ligands such as EDTA which form significantly more stable aqueous complexes with U than with UO 2 will increase the reduction rate of U(VI) by bacteria, and these ligands will also prevent the precipitation of UO 2 and help to maintain U dissolved in solution. Therefore, although the U(VI) reduction rate is increased, the mobility of uranium would be unaffected when strong complexing agents such as EDTA are present. A similar ligand effect may contribute to the enhancement of enzymatic Pu(IV) reduction by bacteria in the presence of NTA (Rusin et al., 1994). The results from our Ca experiments suggest that the rate of enzymatic U(VI) reduction is directly proportional to the concentration of Ca-free uranyl bacterial surface complexes, and that system components that prevent the formation of these surface complexes would slow the reduction rate. For example, the presence of mineral surfaces that adsorb UO 2+ 2 and thereby compete with the bacterial cell wall sites in binding UO 2+ 2 may significantly slow bacterial U(VI) reduction rate in field settings. Surface complexation modeling offers a means for accounting for these competitive effects, and the relationships derived here may apply to these more 40

53 complex settings. The modeling approach outlined here can be used to calculate the speciation of U(VI) on the bacterial cell wall and thereby to predict the rate of enzymatic reduction of U(VI) to U(IV) by bacteria. 41

54 CHAPTER 3: URANIUM ADSORPTION BY SHEWANELLA ONEIDENSIS MR-1 AS A FUNCTION OF NAHCO 3 CONCENTRATION 3.1 Introduction Metal adsorption onto bacterial surfaces can affect the mobility and transport of metals in geologic systems (e.g., Beveridge and Murray, 1976; Harvey and Leckie, 1985; Konhauser et al., 1993; Fein, 2000). Several studies have examined U(VI) adsorption onto bacterial cell walls. U(VI) adsorbs strongly onto bacteria, exhibiting ph-dependent adsorption behavior that is likely caused by a range of uranyl surface complexes on the cell wall (Fowle et al., 2000; Haas et al., 2001; Sar and F D'Souza, 2001; Gorman-Lewis et al., 2005; Sheng et al., 2011). In addition to the direct effect of adsorption on the partitioning of mass in geologic systems, metal adsorption onto bacteria may control the bioavailability of metal ions by bacteria (e.g., Pawlik et al., 1993; Franklin et al., 2000; Borrok et al., 2005a; Sheng et al., 2011). For example, the extent and speciation of U(VI) adsorbed onto bacterial cell walls may control the initial rate of U(VI) enzymatic reduction by bacteria under anaerobic conditions (Sheng et al., 2011). Hence, a better understanding of uranyl-bacterial surface complexation reactions is needed not only to better constrain uranium partitioning in bacteria-bearing aqueous systems, but also to develop quantitative models of the bacterial bioavailability of uranium. 42

55 U(VI) adsorption onto bacteria increases with increasing ph under the low ph conditions (less than approximately ph 5) where the uranyl cation dominates the aqueous U(VI) speciation. In systems containing dissolved inorganic carbonate (DIC), as ph increases above 5, a series of aqueous uranyl-hydroxide, uranyl-hydroxidecarbonate, and uranyl-carbonate complexes become important. The presence of these aqueous complexes competes to some extent with bacterial cell wall binding sites for uranyl, causing a decrease in the extent of U(VI) that adsorbs onto bacteria (Haas et al., 2001; Gorman-Lewis et al., 2005; Sheng et al., 2011). Surface complexation modeling represents a flexible approach for quantifying the extent of metal adsorption onto bacterial cell walls (e.g., Plette et al., 1996; Fein et al., 1997; Ngwenya et al., 2003; Guiné et al., 2006), and hence can be used to predict the fate and transport of metals in water-rock systems (Fein, 2000). The increase in adsorption with increasing ph up to ph 5 can be modeled as uranyl cation adsorption onto deprotonated cell wall functional groups according to the following reaction (Fowle et al., 2000; Haas et al., 2001; Gorman-Lewis et al., 2005): UO >Ai 1- >Ai-UO 2 + (1) where >Ai 1- represents a deprotonated functional group bound to the bacterial cell wall. Because the uranyl cation activity becomes negligible with increasing ph above ph 5, models based on Reaction (1), including those that involve UO 2+ 2 adsorption onto multiple sites, predict dramatically decreasing U(VI) adsorption with increasing ph above ph 5 (Haas et al., 2001; Gorman-Lewis et al., 2005). However, between approximately ph 5 and 8, observed extents of U(VI) adsorption onto bacteria are 43

56 extensive both in DIC-free systems and systems open to atmospheric CO 2, and only decrease slightly with increasing ph (Gorman-Lewis et al., 2005). The elevated extent of U(VI) adsorption that is observed relative to predicted extents of adsorption based only on Reaction (1) represents strong evidence that other aqueous uranyl species adsorb to a significant extent onto bacterial cell walls. Although the available data are unequivocal regarding the existence of these additional uranyl-bacterial surface complexes, there is considerable uncertainty regarding the identities and thermodynamic stabilities of these complexes, because only a limited number of conditions have been probed. X-ray absorption spectroscopy (XAS) studies indicate that adsorption of U(VI) onto bacterial cell walls under low ph conditions (ph < 5) is controlled by uranyl complexation with phosphoryl and carboxyl functional groups within the bacterial cell wall (Hennig et al., 2001; Kelly et al., 2002; Merroun et al., 2005). However, XAS studies of uranyl adsorption onto bacteria are limited to ph values less than 5, and do not provide constraints on the identities of the uranyl-bacterial surface complexes that control U(VI) adsorption onto bacteria above ph 5 in the presence of DIC. It is imperative to place rigorous constrains on the identities and stability constants of the important uranyl-carbonate-bacterial surface species under conditions with ph greater than 5 and high concentrations of DIC in order to model uranium remediation and mining strategies that involve carbonate leaching (Mason et al., 1997; Francis et al., 1999; IAEA, 1993) and to enhance the effectiveness of uranium remediation at sites with high concentrations of DIC in the groundwater. Haas et al. 44

57 (2001) modeled U(VI) adsorption onto Shewanella putrefaciens up to ph 10, invoking binding of UO 2 (OH) 0 2 onto cell wall functional groups above ph 5; Gorman-Lewis et al. (2005) modeled the enhanced U(VI) adsorption onto Bacillus subtilis above ph 5 in the presence of DIC concentrations in equilibrium with the atmosphere with reactions involving the adsorption of UO 2 CO 0 3, (UO 2 ) 2 CO 3 (OH) - 3 and UO 2 (CO 3 ) 4-3 ; and Sheng et al. (2011) modeled U(VI) adsorption onto Shewanella oneidensis invoking adsorption of UO 2 CO 0 3 and UO 2 (CO 3 ) 4-3 onto the bacterial cell wall above ph 5. Each of these studies consisted of bulk U(VI) adsorption measurements with solutions open to the atmosphere. Because these previous experiments were conducted with only one DIC concentration (open to the atmospheric CO 2 ), the measurements do not provide rigorous constraints on the stoichiometries of the important adsorbed uranyl complexes. In this study, we measured the adsorption of aqueous U(VI) onto Shewanella oneidensis strain MR-1 as a function of NaHCO 3 concentration in solution. The molal ratio of DIC : U(VI) in this study ranged from 0 to 120 : 1. A thermodynamic surface complexation modeling approach was used to interpret the experimental adsorption data in order to determine the dominant adsorption reactions and the stability constants for the important uranyl-bacterial surface complexes. The wide range of DIC concentrations that we studied is needed in order to rigorously constrain the identities and stability constants of the important uranyl-carbonate-bacterial surface species. It is particularly important to determine these parameters in order to model the effects of 45

58 bacterial adsorption of uranium and the bioavailability of uranium in a range of natural and engineered DIC-bearing aqueous systems. 3.2 Experimental Methods Bacteria Preparation U(VI) adsorption was measured onto Shewanella oneidensis strain MR-1, a facultative gram-negative bacterial species. All of the growth media and solutions used in this study were made with ultrapure 18 MΩ water. Cells of S. oneidensis were cultured aerobically and prepared following similar procedures to those described in Sheng et al. (2011). The bacterial cells were first transferred from an agar plate to a test tube containing 7 ml of autoclaved trypticase soy broth (TSB) with 0.5% yeast extract, and incubated at 32 o C for 24 h. The cell suspension in the tube was then transferred to 2 L of autoclaved TSB with 0.5% yeast extract, and incubated at 32 o C for another 24 h. Cells were collected by centrifugation at 10,970 g for 5 min after the 48 h growth. Collected cells were then rinsed five times in 0.1 M NaClO4. The cells were recollected by centrifugation at 8,100 g for 5 min after each rinse. After the fifth rinse, the supernatant was discarded, and the cells were then centrifuged twice at 8,100 g for 30 min without any further addition of 0.1 M NaClO 4. The supernatant was poured off after each centrifugation in order to determine the wet mass of cells that was to be used in the following U(VI) adsorption experiments. The wet mass is approximately 8 times the dry mass of the biomass (Borrok et al., 2005b). A known wet mass of cells was added to a weighed volume of 0.1 M NaClO 4 electrolyte solution to form a concentrated parent 46

59 cell suspension, and this parent suspension was diluted gravimetrically for the U(VI) adsorption experiments U(VI) Adsorption Experiments An aqueous filter-sterilized (using a nylon membrane with pore size of 0.2 µm) uranyl acetate stock solution was prepared according to procedures described previously (Sheng et al., 2011). The concentration of uranium in the uranyl acetate stock solution was approximately 160 mm, which was determined by inductively coupled plasma optical emission spectroscopy (ICP-OES). A weighed amount of solid NaHCO3 was dissolved in 0.1 M NaClO 4 electrolyte solution to achieve NaHCO 3 concentrations of 0.0, 0.2, 2.4, 11.9 or 30.0 mm. Experiments with 0.0 mm NaHCO 3 were conducted inside a glovebox chamber with a gas composition of 95% N 2 /5% H 2, and the 0.1 M NaClO 4 solution was bubbled with 95% N 2 /5% H 2 for ~30 min before being transferred into the glovebox. All the other experiments with dissolved NaHCO 3 were conducted outside the glovebox open to the atmosphere. For each concentration of NaHCO 3, an aliquot of the uranyl acetate stock solution was added to attain an initial dissolved U(VI) concentration of 0.25 mm in each experiment. Before the addition of bacterial cells to the experiments, samples were taken of the experimental solution in order to determine the initial total dissolved uranium concentration by ICP-OES. A weighed volume of the parent cell suspension as prepared above was added to the 0.25 mm U(VI)-electrolyte solution to form a parent bacteria-u(vi)-electrolyte suspension with 1.0 g/l (wet mass) of S. oneidensis. The ph of the parent suspension was then adjusted to be between 6 and 8 using small aliquots of concentrated HNO 3 and/or NaOH. The parent suspension 47

60 was then separated into Teflon tubes with approximately 8 ml in each. The ph of the suspension in each Teflon tube was adjusted to cover the ph range of approximately 3 to 9 using small aliquots of concentrated HNO 3 and/or NaOH. After the ph adjustment, all of the Teflon tubes were placed on a rotating rack (~30 rpm) for 3 h. After the 3 h equilibration period, the final ph of the suspension was measured, and then each suspension was centrifuged at 6,600 g for 2 min and finally filtered through a 0.2-µm Millipore Millex PTFE filter. The filtered solution was then acidified with concentrated HNO 3 in preparation for ICP-OES analysis of the final concentration of dissolved uranium. The concentration of adsorbed U(VI) was determined by difference between the measured initial and final aqueous uranium concentrations. Cell-free control experiments for each concentration of NaHCO 3 were also conducted, and did not indicate any significant uranium loss due to adsorption onto the experimental apparatus (data not shown). A Perkin-Elmer Optima 2000DV ICP-OES system was used to analyze the concentration of total dissolved uranium in solution. Matrix-matched blanks and standards covering the probable range of uranium concentrations in solution were prepared. The standards and the blank were re-analyzed after running every 30~40 samples in order to check machine drift. Analytical uncertainty was approximately ± 2%, as determined by repeat analyses of an aqueous uranium standard, and the operational detection limit for uranium on the ICP-OES was determined to be approximately 60 ppb. 48

61 3.3 Results and Discussion U(VI) Adsorption Experiments The observed extent of U(VI) adsorption onto S. oneidensis is independent of NaHCO 3 concentration below ph 5 (Figure 3.1). The extent of adsorption for each of the NaHCO 3 concentrations increases from approximately 18% at ph 3 to 38% at ph 5. The extents of adsorption in the 0.0 mm and 0.2 mm NaHCO 3 experiments are similar across the ph range of these experiments, with the concentration of adsorbed U(VI) continuing to increase above ph 5 to approximately 90% at ph 8.7 and 79% at ph 8.4 for the 0.0 and 0.2 mm NaHCO 3 experiments, respectively. However, above ph 5, the extent of U(VI) adsorption that we observed varies significantly as a function of NaHCO 3 concentration, with U(VI) adsorption decreasing with increasing NaHCO 3 concentration in solution. At ph 6.7, the measured extent of U(VI) adsorption in the experiments with 0.2 mm NaHCO 3 is approximately 63%, and decreases to approximately 46%, 12%, and 1% in the 2.4, 11.9, and 30.0 mm NaHCO 3 experiments, respectively (Figure 3.1). Because of the effect of dissolved NaHCO 3 on the adsorption behavior under the higher ph conditions studied, the ph at which the observed maximum extent of U(VI) adsorption occurs also changes as a function of NaHCO 3 concentration, with the maximum adsorption in the 2.4, 11.9 and 30.0 mm NaHCO 3 experiments occurring at approximately ph 5.9, 5.9 and 5.0, respectively. 49

62 Figure 3.1: The percentage of U(VI) adsorbed onto 1 g(wet mass)/l of S. oneidensis MR-1 in the presence of 0.1 M NaClO 4 and varying concentrations of NaHCO 3 as a function of ph. Total concentration of uranium was 0.25 mm in each experiment; concentrations of NaHCO 3 were 0.0, 0.2, 2.4, 11.9 and 30.0 mm. 50

63 3.3.2 Thermodynamic Modeling We use a thermodynamic surface complexation modeling approach to model the adsorption of U(VI) onto the cell surface of S. oneidensis. In this approach, U(VI) adsorption onto bacteria is modeled as interactions between a range of adsorbing aqueous uranyl species and specific sites on the bacterial cell wall. The adsorption reactions are modeled to involve discrete, negatively charged deprotonated sites, with the deprotonation reactions of the bacterial cell wall functional groups written as: R L n H 0 R L n H (2) where R is the bacterial cell wall macromolecule to which the functional groups are attached, and Ln represents a particular functional group type that is present on the bacterial cell wall. We use a non-electrostatic model (Fein et al., 2005), so the mass action equation for reaction (2) is expressed as: K a [ R L [ R L n n ] a H 0 H ] (3) where [R-L - n ] and [R-L n -H 0 ] represent the concentration of deprotonated and protonated functional groups of each bacterial cell wall species in moles per liter of solution, respectively, and represents the activity of H + in solution. We use the 4-site model of Mishra et al. (2010) to characterize the proton-active binding sites on S. oneidensis, with Sites 1-4 exhibiting pk a values of 3.3 ± 0.2, 4.8 ± 0.2, 6.7 ± 0.4, and 9.4 ± 0.5, respectively. The corresponding site densities for Sites 1-4 are 8.9(±2.6) 10-5, 1.3(±0.2) 10-4, 5.9(±3.3) 10-5 and 1.1(±0.6) 10-4 moles per gram of wet mass, respectively. 51

64 The adsorption of an aqueous metal cation (M m+ ) onto bacteria has been represented as the following reaction (e.g., Fein et al., 1997): R L n M m R L n M (m 1) (4) with a mass action equation for this reaction expressed as: K ads [ R L [ R L M n n ] a ( m 1) m M ] (5) where K ads represents the equilibrium constant for Reaction (4). This approach is suitable for conditions at which the dissolved metal is present exclusively as the aqueous cation. Below approximately ph 5, dissolved U(VI) is present in solution dominantly as the aqueous uranyl cation, UO 2+ 2, and we use this approach to account for the observed adsorption behavior under conditions with ph < 5. However, with increasing ph above ph 5, the aqueous speciation of U(VI) becomes much more complex, with a range of aqueous uranyl-hydroxide, uranyl-carbonate, and mixed uranyl-hydroxide-carbonate complexes forming. In general, the speciation of U(VI) in solution can be represented by the following generic aqueous complexation reaction: xuo 2 2 2x 2 y z 2 yco 3 zoh ( UO2) x( CO3) y( OH) z (6) Because both the speciation of aqueous U(VI) and the speciation of the bacterial cell wall change as a function of ph, it is impossible to unequivocally determine both the adsorbing U(VI) species and the binding sites responsible for the observed adsorption behavior based on bulk adsorption measurements only. However, we can place constraints on these parameters, and we do so by modeling U(VI) adsorption with the following generic reaction: 52

65 R L n xuo 2 2 2x 2 y z 1 2 yco 3 zoh R Ln ( UO2) x( CO3) y( OH) z (7) The goal of this study is to construct a model that could quantitatively account for the observed U(VI) adsorption behavior as a function of both ph and DIC concentration. Without spectroscopic constraints on the binding sites involved in adsorption, we take the approach of modeling the adsorption starting with the low ph data, collected under conditions at which the aqueous U(VI) speciation is most simple and where only Site 1 sites are deprotonated to a significant extent. With increasing ph, we ascribe the observed adsorption to the dominant aqueous U(VI) species present at each experimental ph, as calculated from the reactions given in Appendix A, binding onto deprotonated Site 1 sites. If the observed extent of U(VI) adsorption exceeds the capacity of Site 1 to account for it, then we include binding onto deprotonated Site 2 sites, and so on. The general modeling approach is similar to the approach described by Gorman-Lewis et al. (2005) and Sheng et al. (2011), except that here we assume that U(VI) preferentially adsorbs onto low pk a sites. Because the speciation of the aqueous U(VI) is most simple under low ph conditions and in the DIC-free 0.0 mm NaHCO 3 experiments, and because only Site 1 is deprotonated under these conditions, we begin the construction of our adsorption model by modeling only the 0.0 mm NaHCO 3 data below ph 5. Increasing ph causes more sites to deprotonate and also complicates the aqueous U(VI) speciation, so starting with the low ph data provides a firm foundation for the modeling. Using the ph < 5 data from the 0.0 mm NaHCO 3 experiments, we attributed the observed adsorption 53

66 to the formation of R-L 1 -UO + 2, using x = 1, y = z = 0 in Reaction (7). Using FITEQL (Herbelin and Westall, 1994), the data were used, in conjunction with the aqueous complexation reactions listed in Appendix A including reactions and 31, in which carbonate is not involved in the aqueous speciation, to calculate the stability constant for [R- L 1 -UO 2 ] +. If we use this calculated stability constant to predict the extent of U(VI) adsorption as a function of ph across the entire ph range studied, the predictions fit the data well below ph 4.5, but the model predicts a decrease in adsorption above ph 4.5 to negligible adsorption at approximately ph 6 due to the dramatic decrease in the concentration of aqueous UO 2+ 2 above ph 4.5 (dashed curve labeled [R-L 1 -UO 2 ] + in Figure 3. 2). Clearly, our data, which show a continued increase in adsorption above ph 4.5, are strong evidence that additional adsorption reactions become important with increasing ph. Although our data unequivocally indicate the presence of one or more different adsorbing uranyl species than UO 2+ 2 above ph 4.5, the data do not uniquely define the identities of those species. We modeled the enhanced adsorption progressively with increasing ph, choosing the aqueous uranyl species that predominates over a particular ph range as the adsorbing species for that ph range. As described previously, the aqueous complexation reactions listed in Appendix A, including reactions and 31, were used to calculate the aqueous speciation of U(VI) from ph 3 to 9 under the 0.0 mm NaHCO 3 experimental conditions (Figure B.1 in Appendix B). The approach that we used next to model the 0.0 mm NaHCO 3 data was to ascribe the enhanced adsorption between ph 4.5 and 7.5 to an adsorption reaction involving the dominant aqueous U(VI) 54

67 species in this ph range, (UO 2 ) 3 (OH) + 5. We modeled all of the ph < 7.5 data simultaneously by invoking the formation of both [R-L 1 -UO 2 ] + and [R-L 1 -(UO 2 ) 3 (OH) 5 ] 0. It is possible that Site 2 is at least partially responsible for the adsorption that we observed above ph 5, but without direct spectroscopic evidence, it is impossible to determine which sites are responsible for adsorption under the experimental conditions. Therefore, we ascribed adsorption only onto Site 1 until that site saturates, at which point we included adsorption onto Site 2 and so on. We followed a similar approach for the rest of the studied ph range for the 0.0 mm NaHCO 3 data. For these data, all adsorption can be modeled by invoking adsorption onto Site 1 only, and the data require the formation of 4 uranyl-bacterial surface complexes: [R-L 1 -UO 2 ] +, [R-L 1 - (UO 2 ) 3 (OH) 5 ] 0, [R-L 1 -(UO 2 ) 4 (OH) 7 ] 0 and [R-L 1 -(UO 2 ) 3 (OH) 7 ] 2-. The calculated stability constants for these complexes are tabulated in Table 3.1, and the model fit to the data is shown in Figure 3.2. The three dashed curves in Figure 3.2 represent predicted extents of adsorption based on models that include [R-L 1 -UO 2 ] + only, [R-L 1 -UO 2 ] + and [R- L 1 -(UO 2 ) 3 (OH) 5 ] 0 only, and [R-L 1 -UO 2 ] +, [R-L 1 -(UO 2 ) 3 (OH) 5 ] 0 and [R-L 1 -(UO 2 ) 4 (OH) 7 ] 0 only, and these curves are shown as evidence for the need for additional uranyl-bacterial complexes in order to account for the observed adsorption behavior across the entire ph range studied. The final model involving all 4 uranyl-bacterial surface complexes yields an excellent fit to the experimental data over the entire ph range studied. 55

68 Figure 3.2: The open diamonds represent experimental adsorption data for experiments with 0.0 mm NaHCO 3. The dashed and solid curves represent predicted extents of U(VI) adsorption. 56

69 TABLE 3.1 CALCULATED LOG K VALUES FOR URANYL SURFACE COMPLEXES FORMED ON THE CELL WALL OF S. ONEIDENSIS MR-1. Uranyl Surface Complexes [NaHCO 3 ] (mm) [R-L 1 - UO 2 ] + (a) [R-L 1 - (UO 2 ) 3 (OH) 5 ] 0 (a) [R-L 1 - (UO 2 ) 4 (OH) 7 ] 0 (a) [R-L 1 - (UO 2 ) 3 (OH) 7 ] 2- (a) [R-L 2 - (UO 2 ) 2 CO 3 (OH) 3 ] 2- (a) [R-L 1 - UO 2 CO 3 ] - (a) * * * * * * * * * * * * * AVG σ * Calculation of Log K value was not conducted for this dataset (see text). (a) R-L # - represents S. oneidensis MR-1 functional groups, Sites 1-4, with pk a values of 3.3 ± 0.2, 4.8 ± 0.2, 6.7 ± 0.4, and 9.4 ± 0.5, respectively (Mishra et al., 2010). 57

70 In the next stage of modeling, we used the stability constants that we calculated from the 0.0 mm NaHCO 3 dataset to predict the extent of U(VI) adsorption that we would expect in the 0.2 mm NaHCO 3 experiments (Figure 3.3). The prediction model included reactions in Appendix A, including those which involve aqueous uranylcarbonate complexation reactions (Reactions in Appendix A), but did not include any adsorption reactions that form uranyl-carbonate-bacterial surface complexes. The prediction curve fits the data reasonably well up to approximately ph 6. However, this model prediction dramatically underestimates the extent of adsorbed U(VI) above ph 6, where the aqueous U(VI) speciation becomes dominated by uranyl-carbonate species (see the corresponding aqueous U(VI) speciation diagram for this NaHCO 3 concentration in Figure B.1b in Appendix B). The observed enhanced extent of U(VI) adsorption relative to that predicted from the model derived from the 0.0 mm NaHCO 3 data represents strong evidence that at least one aqueous uranyl-carbonate species must adsorb onto the bacteria above ph 6, causing the adsorption that we observed under those conditions. The dominant aqueous U(VI) species above ph 6 in the system with 0.2 mm NaHCO 3 is (UO 2 ) 2 CO 3 (OH) - 3 (Figure B. 1b), so to account for the observed enhanced adsorption, we added the adsorption reaction that creates [R-L 2 - (UO 2 ) 2 CO 3 (OH) 3 ] 2- to the reactions that we used to model the 0.0 mm NaHCO 3 data and solved for the stability constant of this new uranyl-bacterial surface complex. Invoking adsorption onto Site 2 was required because the extent of adsorption in the 0.2 mm NaHCO 3 experiments above ph 6 exceeded the capacity of Site 1. Again, it is possible 58

71 that other sites were involved in the adsorption, but without spectroscopic constraints it is impossible to determine the sites involved. The stability constants for the surface complexes that control the U(VI) adsorption in the 0.2 mm NaHCO 3 system (listed in Table 3.1) were calculated based on the 0.2 mm NaHCO3 dataset only, independent of the 0.0 mm NaHCO 3 data. However, we fixed the stability constant for [R-L1- (UO 2 ) 4 (OH) 7 ] 0 at its value that was calculated from the 0.0 mm NaHCO 3 data, because (UO 2 ) 4 (OH) 0 7 is only a minor species when 0.2 mm NaHCO 3 is present in the system, and an independent calculation of the stability constant with the 0.2 mm NaHCO 3 data is not possible. The model curve calculated using the set of K values for the 0.2 mm NaHCO 3 dataset in Table 3.1 fits the data very well across the studied ph range (Figure 3.3). 59

72 Figure 3.3: The open diamonds represent experimental adsorption data for experiments with 0.2 mm NaHCO 3. The dashed curve represents predicted extents of U(VI) adsorption using the set of K values calculated from the 0.0 mm NaHCO 3 dataset (Table 3.1). The solid curve represents predicted extents of U(VI) adsorption using the set of K values listed in Table 3.1 for the 0.2 mm NaHCO 3 dataset. 60

73 For the three highest NaHCO 3 concentrations, we used a similar modeling approach as we did for the 0.0 and 0.2 mm NaHCO 3 datasets. First, we used the averaged stability constants that were calculated from the 0.0 and 0.2mM NaHCO 3 data to predict the extent of adsorption that we would expect under those NaHCO 3 concentrations (dashed curves in Figure 3.4a, b, and c). For each of these conditions, the prediction curve significantly underestimates the observed extent of U(VI) adsorption in the middle of the ph range studied, where the aqueous U(VI) speciation is dominated by UO 2 CO 0 3 for each NaHCO 3 concentration (Figure B.1 c, d and e in Appendix B). Therefore, in each case we assumed that the additional adsorption reaction involves the binding of UO 2 CO 0 3 onto a cell wall site, and we included the reaction that forms [R-L 1 - UO 2 CO 3 ] - along with the reactions that were used to model the 0.0 and 0.2 mm NaHCO 3 systems. In these cases, because aqueous uranyl-hydroxide complexes are not major species under any of the studied conditions, we could not use the 2.4, 11.9, or 30.0 mm NaHCO 3 data to independently solve for the stability constants for the uranyl-hydroxidebacterial surface complexes, and we fixed these values at the averaged values shown in Table 3.1, calculated from the 0.0 and 0.2 mm NaHCO 3 data. Taking each of the three highest NaHCO 3 concentration datasets separately, the stability constants for [R-L 1 - UO 2 ] +, [R-L 2 -(UO 2 ) 2 CO 3 (OH) 3 ] 2- and [R-L 1 -UO 2 CO 3 ] - were calculated, and the resulting set of calculated stability constants for each condition (tabulated in Table 3.1) yields an excellent fit to each dataset as shown in Figure

74 Figure 3.4: The open diamonds represent experimental adsorption data for experiments with (a) 2.4 mm (b) 11.9 mm and (c) 30.0 mm NaHCO 3. The dashed curves represent predicted extents of U(VI) adsorption using averaged K values calculated from datasets of 0.0 and 0.2 mm NaHCO 3 listed in Table 3.1. The solid curves represent predicted extents of U(VI) adsorption using the set of K values listed in Table 3.1 for each dataset of 2.4 mm, 11.9 mm and 30.0 mm NaHCO 3. 62

75 In summary, our U(VI) adsorption data provide unequivocal evidence for the formation of a series of uranyl-, uranyl-hydroxide-, uranyl-hydroxide-carbonate-, and uranyl-carbonate-bacterial surface complexes. The average values of the calculated stability constants are listed in Table 3.1. The uncertainty associated with each stability constant value was calculated directly from the range of K values calculated for each complex, or, for stability constants that were calculated from fewer than three sets of data (e.g., for [R-L 1 -(UO 2 ) 3 (OH) 5 ] 0, [R-L 1 -(UO 2 ) 4 (OH) 7 ] 0 and [R-L 1 -(UO 2 ) 3 (OH) 7 ] 2- ), we determined the uncertainty by varying the K value while fixing all others in order to determine the K value range needed to encompass 95% of the experimental data under the ph range where the associated complex dominates the adsorption. In general, the calculated K values from the different datasets in this study are consistent with each other within their calculated uncertainties, especially the K values for [R-L 1 -UO 2 ] +, [R-L 2 - (UO 2 ) 2 CO 3 (OH) 3 ] 2- and [R-L 1 -UO 2 CO 3 ] -. However, the K value for [R-L 1 -(UO 2 ) 4 (OH) 7 ] 0 is not well constrained by our data because only the 0.0 mm NaHCO 3 dataset can be used to solve for it independently. In order to test the accuracy of the overall set of calculated K values to account for the range of adsorption behaviors that we observed as functions of ph and NaHCO 3 concentration, we used the average stability constant values to predict the extent of adsorption expected under each experimental condition studied. These predicted adsorption curves are shown in Figure 3.5. The set of average stability constants from this study provides excellent fits to all of the datasets (Figure 3.5), accounting for the effects of both ph and dissolved NaHCO 3 on the adsorption behavior of U(VI) by S. oneidensis. 63

76 Figure 3.5: The open diamonds represent experimental adsorption data for experiments with (a) 0.0 mm (b) 0.2 mm (c) 2.4 mm (d) 11.9 mm and (e) 30.0 mm NaHCO 3. The solid curves represent predicted extents of U(VI) adsorption using the set of average K values listed in Table

77 3.4 Conclusions The extent of U(VI) adsorption onto S. oneidensis below ph 5 is independent of NaHCO 3 concentration in solution, likely because the adsorption is caused by complexation of the uranyl cation with cell wall functional groups and uranyl-carbonate complexes are not involved in the binding. Above ph 5, increasing NaHCO 3 concentration causes the extent of U(VI) adsorption to decrease, but the observed extent of U(VI) adsorption is higher than that predicted by accounting for aqueous uranyl-carbonate complexation only and neglecting adsorption of uranyl-carbonate species onto the bacteria. Therefore, our adsorption data provide unequivocal evidence for the formation of a series of uranyl-, uranyl-hydroxide-, uranyl-hydroxide-carbonate-, and uranyl-carbonate-bacterial surface complexes. Contrary to the results of Gorman- Lewis et al. (2005), our data do not require the adsorption of UO 2 (CO 3 ) 4-3 onto the - bacterial cell wall. However, adsorption reactions involving binding of (UO 2 ) 2 CO 3 (OH) 3 and UO 2 CO 0 3 onto bacterial cell wall functional groups control the U(VI) adsorption onto bacteria under conditions with elevated NaHCO 3 concentrations. The thermodynamic model that we construct in order to account for the observed U(VI) adsorption behavior is not unique. We have assumed that the dominant aqueous uranyl species at each ph studied is the adsorbing species, and we have ascribed all binding onto Sites 1 and 2 on the bacterial cell wall. Clearly, detailed spectroscopic data are required as a function of ph and NaHCO 3 concentration in order to place better constraints on the identities and compositions of the important uranyl-bacterial surface complexes. However, the model that we have developed is a reasonable one and provides an excellent overall fit to the 65

78 data, capturing the ph and NaHCO 3 dependencies well. Our results demonstrate that bacteria can compete with DIC to control the distribution of U(VI) in aqueous systems, and the calculated stability constants for the uranyl-bacterial surface complexes from this study provide a framework for estimating the adsorption and speciation of U(VI) on bacterial cell walls in complex environments. These modeling results may further improve our ability to understand bacterial effects on U(VI) speciation, bioavailability, and remediation in geologic systems. 66

79 CHAPTER 4: URANIUM REDUCTION BY SHEWANELLA ONEIDENSIS MR-1 AS A FUNCTION OF NAHCO 3 CONCENTRATION: SURFACE COMPLEXATION CONTROL OF REDUCTION KINETICS 4.1 Introduction A number of groundwater systems have been contaminated by uranium as a result of mining activities or from improper disposal of waste from nuclear materials processing. The mobility of U in the subsurface can be dramatically reduced when soluble U(VI) is reduced to insoluble U(IV) solid phases under anaerobic conditions (e.g., Langmuir, 1978; Gorby and Lovley, 1992; Senko et al., 2002). Enzymatic reduction of U(VI) by bacteria can be both rapid and complete under controlled conditions (Lovley et al., 1991; Francis et al., 1994; Sani et al., 2002), and hence represents a potentially efficient and inexpensive U remediation approach (Finneran et. al, 2002; Anderson et al., 2003; Wu et al., 2007). In order to understand and model the fate and mobility of U in groundwater systems and also to enhance the effectiveness of U bioremediation strategies, it is crucial to determine the controls on the enzymatic U(VI) bioreduction rate. A considerable amount of research has investigated the mechanisms of U(VI) reduction by bacteria (e.g., Brooks et. al, 2003; Ulrich et al., 2011; Stewart et al. 2011). For example, Ulrich et al. (2011) examined U(VI) reduction by Shewanella oneidensis as a function of 67

80 dissolved bicarbonate and Ca concentrations, and related the change in reduction kinetics that they observed to changes in the dominant U(VI) species in solution that accompanied changing fluid compositions. Similarly, Stewart et al. (2011) observed that dissolved Ca and the presence of iron oxides both decrease the U(VI) reduction rate by affecting the aqueous U(VI) speciation. However, the controls on the kinetics of the reduction process and the connection between aqueous complexation and its effect on U(VI) reduction rates are still poorly defined, especially in complex realistic geologic systems. Previous studies suggest that U(VI) speciation and distribution in geologic systems may affect U(VI) reduction rates, but a better understanding of the relationship between U speciation and U bioavailability is needed in order to derive speciation-based kinetic rate laws. Sheng et al. (2011) proposed that the speciation of both U(VI) and U(IV) on bacterial cell walls controls the kinetics of U(VI) reduction by bacteria. Sheng et al. (2011) measured the rate of U(VI) reduction by S. oneidensis in the presence of dissolved Ca and EDTA, and found strong correlations between the speciation and concentration of U(VI) on the cell surface and the U(VI) reduction rate. Other studies also suggest that metal bioavailability can be linked to the extent and speciation of metal adsorption onto bacterial cell walls (e.g., Franklin et al., 2000; Borrok et al., 2005; VanEngelen et al., 2010). For example, the chemotactic response of Escherichia coli away from aqueous Ni is strongly correlated to the extent of Ni adsorbed onto cell wall functional groups (Borrok et al., 2005a), and the toxicity of U and Cu to bacteria is positively correlated to the concentration of the metals that bind to bacterial cells 68

81 (Franklin et al., 2000). In addition, VanEngelen et al. (2010) demonstrated that the presence of high concentrations of bicarbonate in solution significantly inhibits the toxicity of U(VI) to bacteria due to the formation of negatively charged uranyl-carbonate aqueous complexes which are less bioavailable to the bacteria than carbonate-free U(VI) aqueous species. The presence of bicarbonate under circumneutral to basic ph conditions decreases the concentration of U(VI) that adsorbs onto bacteria (Sheng and Fein, 2012). Hence, it is likely that U(VI) bioavailability is controlled by cell wall adsorption of U(VI), and that the decreased toxicity observed by VanEngelen et al. (2010) was caused by decreased adsorption and/or a change in cell wall speciation of U(VI) that accompanied the introduction of bicarbonate to solution. In order to test the hypothesis that the extent and speciation of U(VI) on the bacterial cell wall controls the kinetics of enzymatic U(VI) reduction by bacteria, in this study we measured the rate of U(VI) reduction by bacteria as a function of NaHCO 3 concentration in solution. Bicarbonate not only is ubiquitous in natural geologic systems, but also is commonly used as a flushing agent to mobilize subsurface U in remediation and mining approaches (Zhou and Gu, 2005; Santos and Ladeira, 2011; IAEA, 1993). We use stability constants for the important uranyl-bacterial surface complexes (Sheng and Fein, 2012) to calculate the concentration and speciation of U(VI) on the bacterial cell wall under each of the experimental conditions, and we test if a relationship exists between the calculated extent of U(VI) adsorption onto the bacterial cell walls and the observed U(VI) reduction rate. Determining if such a relationship exists will provide insights into the controls on U(VI) bioavailability to bacteria, and is 69

82 crucial in order to design effective bioremediation strategies based on enzymatic reduction of U(VI) in realistic geologic systems. 4.2 Experimental Methods Bacteria Preparation The facultative Gram-negative bacterium Shewanella oneidensis strain MR-1 was used in this study. Cells of S. oneidensis were cultured aerobically and prepared following similar procedures to those described previously (Sheng et al., 2011). Cells were collected by centrifugation at 10,970 g for 5 min after 48 h aerobic growth. Cells were washed twice by resuspending them in 20 ml of sterile anoxic NaHCO 3 solution, and then were resuspended in 10 ml of the same solution to make a parent bacterial suspension. The concentration of NaHCO 3 used in the wash solution was the same as the concentration of NaHCO 3 that was to be used in the specific U(VI) reduction experiments for which each batch of bacteria was being prepared. A small portion of the parent bacterial suspension was sampled in order to measure the cell concentration in the suspension following the procedures described in Fein et al. (2005). The rest of the 10 ml parent bacterial suspension was transferred into an anaerobic glovebox for use in the U(VI) reduction experiments. All the growth media and solutions used in this study were made with ultrapure 18 MΩ water. 70

83 4.2.2 U(VI) Reduction Experiments The reduction experiments were conducted inside a glovebox chamber with an anaerobic atmosphere of 95% N 2 /5% H 2, and with a palladium catalyst heater unit to remove any remnant oxygen. We used a gas analyzer within the glovebox to monitor oxygen concentrations, insuring that oxygen was present below the 1 ppm detection limit of the analyzer at all times. The general procedures and analytical methods used were similar to those described by Sheng et al. (2011). The experimental medium consisted of 50 mm Na-lactate, and NaHCO 3 at either 2.4, 5.0, 7.2, 11.9, 21.0 or 30.0 mm. First, a 50 mm Na-lactate solution was bubbled with a 95% N 2 /5% H 2 gas mixture for ~30 min outside the glovebox to remove dissolved oxygen. The Na-lactate solution was then transferred into the glovebox and separated into serum bottles, which were sealed, removed from the glovebox, and then autoclaved at 120 o C for 20 min. Inside the glovebox, a weighed mass of solid NaHCO 3 was added to a weighed amount of the anoxic sterile 50 mm Na-lactate solution to achieve a NaHCO 3 concentration of 2.4, 5.0, 7.2, 11.9, 21.0 or 30.0 mm. The ph of this lactate-bicarbonate solution was adjusted to be approximately 7.0 using small aliquots of concentrated NaOH and/or HCl. A filtersterilized (using 0.2 µm nylon membrane filters) uranyl acetate stock solution was prepared as described in Sheng et al. (2011), the concentration of which was determined to be approximately 200 mm by inductively coupled plasma optical emission spectroscopy (ICP-OES). A volume of the uranyl acetate stock solution was added to a weighed volume of the lactate-bicarbonate solution inside a Teflon bottle to achieve an initial aqueous U(VI) concentration of 0.25 mm. The exact total dissolved U 71

84 concentration of a sample of this solution was determined by ICP-OES, and was taken to be the initial U(VI) concentration in the reduction experiments. A weighed volume of the parent bacterial suspension was added to a weighed volume of the U(VI)-lactatebicarbonate solution to achieve an experimental cell concentration of 1.0 g (weight mass)/l. After addition of the cells, the experimental suspension was stirred continuously and sampled at selected times. The ph was periodically measured and maintained to be approximately 7.0 by using small aliquots of concentrated NaOH and/or HCl. U(VI) can adsorb to a significant extent onto S. oneidensis at ph 7.0 even in the presence of NaHCO3 (Sheng and Fein, 2012), so it is likely that a portion of the U(VI) that was removed from solution was present as adsorbed U(VI) on the bacterial cells. If we directly measured the U(VI) left in solution only, we would overestimate the reduction rate by failing to account for the U(VI) that was adsorbed onto the bacterial cells. Therefore, in order to measure all of the U(VI) remaining in the system at each sampling time, we conducted a U(VI) desorption step on each sample that promoted desorption of U(VI) from the bacteria and caused all remaining U(VI) in the system to be present in solution and available for analysis. When sampling, 8 ml was extracted from the reaction vessel and transferred into a Teflon tube, sealed inside the glovebox, and then centrifuged at 6,600 g for 2 min and filtered through a 0.2-µm Millipore Millex PTFE filter outside the glovebox. The sample was analyzed immediately for dissolved U(VI) concentration by fluorescence spectrometry. Analysis of those samples yields a before desorption concentration of U(VI) in the experiment. At each sampling time, a second 8 ml sample was also extracted and transferred to another Teflon tube and 72

85 acidified with 60 µl of 12.1 N HCl, to decrease the ph of the sample to approximately 1.5. After 40 min of agitation, the acidified sample was sealed inside the glovebox and then transferred to the outside for centrifugation, filtration, and immediate analysis for dissolved U(VI) concentration. Because the desorption step caused all remaining U(VI) in the experimental systems to partition into solution, we used these after desorption U(VI) concentrations to define the U(VI) reduction rate. The reduction experiments were conducted in triplicate for each NaHCO 3 concentration. A set of desorption control experiments were carried out to determine if the acidification step in the reduction experiments could release adsorbed U(VI) into solution. In the control experiments, 1.0 g(wet mass)/l of cells prepared as described above were added to a solution which consisted of 50 mm Na-lactate solution, different concentrations of NaHCO 3 (2.4, 5.0, 7.2, 11.9, 21.0 or 30.0 mm), and 0.25 mm total U (added as uranyl acetate). The ph was maintained at approximately 7.0 for ~2.5h. The sample was then acidified with 12N HCl to approximately ph 1.5 for 40 min. The cells were removed by centrifugation and filtration as described previously, and the concentrations of initial (before addition of cells) and final (after removal of cells) total dissolved uranium in the solution were analyzed by ICP-OES. The results of these desorption control experiments indicate that the acidification step promoted desorption of approximately 92% of the total uranium in the system, and only 8% of the uranium was present as adsorbed U on the cells at ph 1.5. Cell-free control experiments, which followed the same experimental procedures as the reduction experiments but without the addition of cells, were also conducted in order to determine if any U(VI) was lost due 73

86 to adsorption onto the experimental apparatus or for any possible reason other than reduction by bacteria. No measurable loss of U(VI) was observed in these abiotic control experiments. A PTI Quantamaster QM-4 spectrofluorometer was used to measure the phosphorescence decay of U(VI) in order to determine the concentration of U(VI) in solution, following the general approach and principles described in previous studies (Gorby and Lovley, 1992; Brina and Miller, 1992). Those previous studies used a kinetic phosphorescence analyzer (KPA) which uses a pulsed nitrogen laser as an excitation source and has an extremely low detection limit (~1ng/L) (Brina and Miller, 1992). Similar to KPA, the spectrofluorometer measures phosphorescence decay by recording the change in intensity of the phosphorescence signal emitted by excited U(VI) atoms in the sample as a function of time. The spectrofluorometer that was used in this study uses a xenon flash lamp, and has a higher detection limit (~2 ppm) than KPA. However, because the initial concentration of U(VI) in our experiments was 0.25 mm (~60 ppm), the spectrofluorometer provides adequate U(VI) analytical resolution and precision under our experimental conditions. The setup parameters of the spectrofluorometer and sample preparation were similar to those reported by Sheng et al. (2011). Each sample (0.5 ml) was acidified with 0.25 ml of 12.1 N HCl, and diluted 150 times with ultrapure 18 MΩ water. 1.5 ml of Uraplex (complexing agent) was then added to 1 ml of diluted sample and the solution was analyzed immediately on the spectrofluorometer, using an excitation wavelength of 420 nm, an emission wavelength of 515 nm and slit width of 17 nm. Matrix-matched blanks and standards covering the 74

87 probable range of U(VI) in solution were measured to construct a calibration curve and to quantify the U(VI) concentrations in the samples. The spectrofluorometer exhibits a linear dynamic range for aqueous U(VI) concentrations from 2 ppm to 90 ppm under our experimental conditions. Analytical uncertainty was approximately ± 4.5%, as determined by repeat analyses of an aqueous U(VI) standard. 4.3 Results and Discusstion U(VI) Reduction Experiments Acidified samples from experiments with 2.4 and 5.0 mm NaHCO 3 concentrations (Figure 4.1a and b) yielded higher measured U(VI) concentrations than did the non-acidified samples, due to significant adsorption of U(VI) species onto bacterial cell walls. However, experiments with higher NaHCO 3 concentrations, depicted in Figure 4.1c-f, showed no significant difference between the acidified and nonacidified samples, likely because U(VI) species do not adsorb onto bacterial cell wall sites to a measurable extent at neutral ph in the presence of these higher NaHCO 3 concentrations (Sheng and Fein, 2012). Each of the experimental systems exhibited an initial linear decrease in U(VI) concentration as a function of time. For the systems with lower NaHCO 3 concentrations, the observed reduction rate decreased markedly when approximately 60-70% of the original U(VI) had been reduced. The experiments with 21.0 and 30.0 mm NaHCO 3 did not exhibit this change in reduction rate. However, the reduction rates that were observed in these experiments were slower than those observed in the experiments with lower NaHCO 3 concentrations. At the end of the

88 and 30.0 mm NaHCO 3 experiments, the U(VI) concentration had only decreased by approximately 30 % from its original value, so perhaps did not reach a low enough U(VI) concentration to cause the change in reduction rate. In our exercise of relating reduction rates to the speciation and concentration of U(VI) on the bacteria, we only considered the data that define the initial linear reduction rates, with these initial reduction rates depicted in Figure 4.1 as solid lines. The initial reduction rate (mm/h) for each NaHCO 3 concentration was determined from a linear fit to the U(VI) measurements from the acidified samples only, and was calculated using the data points that defined the initial linear relationship. For example, the initial rate of reduction for the 2.4 mm NaHCO 3 experiment shown in Figure 4.1a was determined from the first four data points collected during the first 1 h of the experiment, while the reduction rate for the 30.0 mm NaHCO 3 experiment (Figure 4.1f) was determined from the entire dataset. The data shown in Figure 1 are representative ones for each NaHCO 3 concentration, and the other experimental results are included in Appendix C. The average initial reduction rate and the associated uncertainty (1σ) for each of the experimental conditions are shown in Table 4.1, and were calculated from the three experiments conducted at each NaHCO 3 concentration. 76

89 Figure 4.1: The concentrations of dissolved U(VI) remaining in solution in the presence of (a) 2.4 mm; (b) 5.0 mm; (c) 7.2 mm; (d) 11.9 mm; (e) 21.0 mm; (f) 30.0 mm NaHCO 3 as a function of time. In each figure, open triangles represent the non-acidified dataset in which the sample was immediately analyzed for concentration of remaining U(VI) at each sampling time; solid diamonds represent the acidified dataset in which the sample was acidified to approximately ph 1.5 for ~40 min before analysis of U(VI) concentration; open squares represent the data points used to calculate the initial rate of U(VI) reduction by S. oneidensis. 77

90 TABLE 4.1 CALCULATED URANYL SURFACE COMPLEXES AND AVERAGE REDUCTION RATE FOR EACH NAHCO 3 CONCENTRATION. [NaHCO 3 ] (mm) [R-L 2 - (UO 2 ) 2 CO 3 (OH) 3 ] (mm) 2- (a) [R-L 1 - UO 2 CO 3 ] - (a) (mm) Total Adsorbed U(VI) (b) (mm) Average Initial Reduction Rate (mm/h) ± 1σ ± ± ± ± ± ± (a) R-L#- represents S. oneidensis functional groups, Sites 1-4, with pk a values of 3.3 ± 0.2, 4.8 ± 0.2, 6.7 ± 0.4, and 9.4 ± 0.5, respectively (Mishra et al., 2010). (b) Total adsorbed U(VI) includes all the uranyl surface complexes formed on the cell wall: [R-L 1 -UO 2 ] +, [R-L 1 -(UO 2 ) 3 (OH) 5 ] 0, [R-L 1 -(UO 2 ) 4 (OH) 7 ] 0, [R-L 1 -(UO 2 ) 3 (OH) 7 ] 2-, [R- L 2 -(UO 2 ) 2 CO 3 (OH) 3 ] 2-, [R-L 1 -UO 2 CO 3 ] - (Sheng and Fein, 2012). 78

91 The reduction rate data clearly indicate that increasing the concentration of NaHCO 3 in solution under the experimental conditions significantly decreases the rate of U(VI) reduction by S. oneidensis (Figure 4.2). This observation is consistent with the results of previous studies (e.g., Luo et al., 2007; Ulrich et al., 2011), which note a similar dependence of the U(VI) reduction rate on NaHCO 3 concentration in solution. As the concentration of NaHCO 3 increases from 2.4 mm to 30.0 mm in the reduction experiments, the initial U(VI) reduction rate decreases from to mm/h (Figure 4.2). The reduction rates that we observed in the 21.0 and 30.0 mm NaHCO 3 experiments are the same within experimental uncertainty, with a value of ± mm/h for the 21.0 mm NaHCO 3 system and a value of ± mm/h for the 30.0 mm NaHCO 3 system. The lack of an observed effect of increasing NaHCO 3 concentration on the reduction rate under the two highest NaHCO 3 conditions is likely due to the increased difficulty in obtaining precise rate determinations for these conditions where the U(VI) reduction rates are so slow. 79

92 Figure 4.2: The average initial rate of U(VI) reduction by S. oneidensis as a function of NaHCO 3 concentration in the experiments. The error bars represent the associated standard deviation (1σ) of each value. 80

93 4.3.2 Thermodynamic Modeling A non-electrostatic surface complexation modeling approach (Fein et al., 2005; Sheng et al., 2011) and the program FITEQL (Herbelin and Westall, 1994) were used to calculate the extent and speciation of U(VI) that was adsorbed onto the cell surface of S. oneidensis under each experimental condition. The calculations accounted for aqueous U(VI) complexation with hydroxide and carbonate, and also the important uranyl surface complexation reactions determined by Sheng and Fein (2012) for S. oneidensis (refer to reactions 8-33 Appendix A and reactions 1-6 in Appendix D for the reactions considered in the calculations along with corresponding equilibrium constant values and their sources). The speciation calculations indicate that the uranyl surface complexes that formed at ph 7.0 under our experimental conditions are dominantly [R-L 2 - (UO 2 ) 2 CO 3 (OH) 3 ] 2- and [R-L 1 -UO 2 CO 3 ] - (Table 4.1). The extent of total adsorbed U(VI) at ph 7.0 decreases as the concentration of NaHCO 3 increases in the experiments (Figure 4.3). A strong positive correlation exists between the observed U(VI) reduction rate and the calculated total concentration of U(VI) that adsorbed onto the bacterial cell walls (Figure 4.4). The relationship depicted in Figure 4.4 strongly suggests that the bioavailability of U(VI) under the experimental conditions is controlled by the extent of U(VI) adsorption onto the bacterial cell walls. Increasing the concentration of NaHCO 3 in the experimental systems causes a decrease in the extent of U(VI) adsorption onto the cells, and hence a concomitant decrease in the U(VI) reduction rate by the bacteria. These results are consistent with those of Sheng et al. (2011) who measured the effect 81

94 of dissolved Ca on the reduction rate of U(VI) by S. oneidensis, and found a strong negative correlation between the reduction rate and the concentration of Ca-uranylbacterial complexes, but a strong positive correlation between the U(VI) reduction rate and the total concentration of Ca-free uranyl-bacterial complexes. 82

95 Figure 4.3: The total concentration of U(VI) adsorbed onto the cell wall of S. oneidensis as a function of NaHCO 3 concentration in the experiments. 83

96 Figure 4.4: Average initial U(VI) reduction rates as a function of the total concentration of U(VI) adsorbed onto the cell wall of S. oneidensis. The error bars represent the associated standard deviation (1σ) of each value. 84

97 Previous studies suggest that U speciation in solution affects the rate of U(VI) reduction by bacteria (e.g., Stewart et al., 2011; Ulrich et al., 2011). However, under the experimental conditions, direct contact between U(VI) and the enzymatic electron pathways within the bacterial cell wall, specifically multiple sites located in the outer membrane and periplasm (Beliaev et al., 2001; Wall and Krumholz, 2006), is required in order for U(VI) reduction to occur. Therefore, although the extent of adsorption and hence reduction rates are clearly related to aqueous U(VI) speciation, this relationship is indirect only. Our approach of relating the reduction rate to the speciation and concentration of U(VI) on the cell wall represents a more mechanistic model that reflects the fact that U(VI) adsorption is the rate-controlling step during U(VI) reduction. The results of this study provide the framework for using a surface complexation modeling approach for deriving quantitative rates laws for U(VI) reduction by bacteria. This modeling approach not only reflects the mechanisms that are responsible for U(VI) reduction, but it also enables flexible predictions of the initial rate of U(VI) reduction by bacteria in complex geologic settings through calculation of the speciation and concentration of U(VI) that adsorbs onto the cell wall. 4.4 Conclusions The observed rate of U(VI) reduction by S. oneidensis in this study decreases with increasing NaHCO 3 concentration in the systems. We calculated the species and concentration of U(VI) that is adsorbed onto the bacteria under the experimental conditions using the thermodynamic surface complexation model and stability constants 85

98 for the important uranyl-bacterial complexes developed by Sheng and Fein (2012). There is a strong positive correlation between the calculated concentration of adsorbed U on the bacterial cell wall and the observed U(VI) reduction rate under those conditions. The correlation between the U(VI) reduction rate and the total concentration of uranyl surface complexes indicates that the speciation and concentration of U(VI) on the bacterial cell wall control the kinetics of enzymatic reduction of U(VI) by bacteria, and the modeling approach outlined here represents a means for predicting enzymatic U(VI) reduction kinetics in complex geologic settings. In a broader sense, our results are consistent with a growing body of evidence that suggests that adsorption controls a wide range of metabolic functions that involve metal cations, and that surface complexation modeling represents a potent approach for quantitatively modeling those processes in complex multi-component systems. 86

99 CHAPTER 5: CONCLUSIONS This dissertation consists of three studies whose overall objective is to create more sophisticated biotic ligand models of metal bioavailability by using site-specific surface complexation models (SCM) to account for the bioavailability. In my research, I put more rigorous constraints on the uranyl bacterial surface complexation reactions that control U adsorption in systems containing high concentrations of dissolved inorganic carbon (DIC), and I used these results to test for possible correlations between U(VI) adsorption and speciation on bacterial cell walls and the kinetics of enzymatic reduction of U(VI) by bacteria. In Chapter 2, my study found that adsorption/desorption reactions of U onto bacterial cell walls can control the kinetics of enzymatic reduction of U(VI) by bacteria. I determined this by measuring the rate of U(VI) reduction by S. oneidensis in the presence of different concentrations of EDTA and dissolved Ca under anaerobic conditions. In the U(VI) reduction experiments, I measured both total dissolved U and aqueous U(VI) remaining in solution as a function of time in order to discern U(VI) adsorption controls from U(IV) desorption controls on the reduction kinetics. In separate experiments, the extent of U(VI) adsorption onto S. oneidensis was also measured in order to quantify the thermodynamic stabilities of the important U(VI)- 87

100 bacterial surface complexes. In the EDTA experiments, the rate of U(IV) production increased with increasing EDTA concentration. However, the total dissolved U concentrations remained constant and identical to the initial U concentrations during the course of the experiments for all EDTA-bearing systems. Additionally, the U(VI) reduction rates in the EDTA experiments exhibited a strong correlation to the concentration of the aqueous U 4+ -EDTA complex. I concluded from these observations that the U(VI) reduction rate increases with increasing EDTA concentration, likely due to U 4+ -EDTA aqueous complexation which removes U(IV) from the cell surface and prevents UO 2 precipitation. In the Ca experiments, the U(VI) reduction rate decreased as Ca concentration increased. The thermodynamic modeling results based on the U(VI) adsorption data demonstrate that U(VI) was adsorbed onto the bacterial surface in the form of a Ca-uranyl-carbonate complex in addition to a number of other Ca-free uranylbacterial surface complexes. The observed U(VI) reduction rates in the presence of Ca exhibit a strong negative correlation to the concentration of the Ca-uranyl-carbonate bacterial surface complex, but a strong positive correlation to the total concentration of all the other Ca-free uranyl-bacterial surface complexes. Thus, the concentration of these Ca-free uranyl surface complexes appears to control the rate of U(VI) reduction by S. oneidensis in the presence of dissolved Ca. The results of this study demonstrate that U speciation, both of U(VI) before reduction and of U(IV) after reduction, affects the reduction kinetics, and that thermodynamic modeling of the U speciation may be useful in the prediction of reduction kinetics in realistic geologic settings. 88

101 As discussed in Chapter 2, U(VI) adsorption and speciation on the bacterial cell wall can control the kinetics of U(VI) bioreduction, so it is important to better constrain the important uranyl bacterial surface complexation reactions which control U adsorption onto bacteria. To rigorously constrain the important uranyl bacterial surface complexation reactions under conditions with high dissolved organic carbon (DIC) is vital in order to better understand and predict the fate and mobility of U in various natural or engineered systems with high levels of DIC. The identities and thermodynamic stabilities of uranyl-hydroxide-carbonate surface complexes are poorly constrained by previous studies, and the research that I present in Chapter 3 placed more rigorous constraints on the stoichiometries of these surface complexes by measuring adsorption of aqueous U(VI) onto S. oneidensis as a function of dissolved NaHCO 3 concentration. A thermodynamic SCM approach was used to interpret the adsorption data. The experimental results indicate that the extent of U(VI) adsorption onto S. oneidensis below ph 5 is independent of NaHCO 3 concentration in solution, likely because the adsorption is caused by complexation of the uranyl cation with cell wall functional groups and uranyl carbonate complexes are not involved in the binding. Above ph 5, increasing NaHCO 3 concentration causes the extent of U(VI) adsorption to decrease, but the observed extent of U adsorption is higher than that predicted by accounting for aqueous uranyl-carbonate complexation only and neglecting adsorption of uranylcarbonate species onto the bacteria. Therefore, a series of uranyl-, uranyl-hydroxide-, uranyl-carbonate-hydroxide-, and uranyl-carbonate-bacterial surface complexes are required in order to account for the observed adsorption behavior, and I used the 89

102 adsorption measurements to constrain values for the stability constants of these complexes. The modeling results suggest that uranyl-carbonate-bacterial surface complexes form and that competition between these complexes and uranyl-carbonate aqueous complexes controls the U(VI) adsorption behavior under high ph conditions in systems with high dissolved inorganic carbonate concentrations. The calculated stability constants for the uranyl-bacterial complexes from this study provide a framework for estimating the adsorption and speciation of U(VI) on bacterial cell walls in complex environments. These modeling results may also further improve our ability to understand bacterial effects on U speciation, bioavailability, and remediation in geologic systems. Based on the results of studies described in Chapters 2 and 3, in Chapter 4 I further test the hypothesis that the speciation of U(VI) on the bacterial cell wall controls the kinectics of U(VI) reduction by bacteria and hence the bacterial bioavailability of U. I measured the U(VI) reduction rate by S. oneidensis in the presence of elevated concentrations of NaHCO 3, and my experimental results indicate that the rate of U(VI) reduction decreases with increasing NaHCO 3 concentration in the experiments. I also calculated the speciation of U(VI) on the bacterial cell wall using the adsorption model developed in Chapter 3. I found a strong correlation between the U(VI) reduction rate and the total concentration of U(VI) adsorbed on the bacterial cell wall. This positive correlation indicates that the speciation and adsorption of U(VI) on the bacterial cell wall controls the kinetics of enzymatic reduction of U(VI) by bacteria. The successful use of thermodynamic modeling to relate U(VI) speciation to enzymatic reduction rates in 90

103 this study may also enable predictions of enzymatic U(VI) reduction kinetics in complex geologic settings. Overall, this dissertation provides new insights into mechanisms that control the kinetics of U(VI) reduction by bacteria. The results suggest that advanced biotic ligand models can be constructing by using surface complexation modeling to account for the bioavailability of metals in complex settings. The thermodynamic model that I constructed in Chapter 3 to account for the observed U(VI) adsorption behavior is not unique. I assumed that the dominant aqueous uranyl species at each ph studied is the adsorbing species, and I ascribed all binding onto Sites 1 and 2 on the bacterial cell wall. However, the model that I develop is a reasonable one and provides an excellent overall fit to the data, capturing the ph and NaHCO 3 dependencies well. Clearly, further detailed spectroscopic studies are required as a function of ph and NaHCO 3 concentration in order to better constrain the identities and compositions of the important uranyl-bacterial surface complexes. Studies described in Chapters 2 and 4 indicate that the U(VI) adsorption/desorption reactions can control the rate of U(VI) reduction by bacteria. To further test those correlations in the future, it would be valuable to study the reduction of U(VI) by bacteria in the presence of more components that could affect the U speciation and/or adsorption onto the cell wall such as minerals that can adsorb U extensively. Those further studies may enable the application of surface complexation modeling to predict U(VI) reduction rates, and even U bioavailability, in realistic natural and engineered systems. Additionally, future studies may also extend this correlation to other metals to test the hypothesis that metal 91

104 adsorption/speciation on the bacterial cell wall controls the metal bioavailability or metal-related metabolic activities of bacteria, such as reduction of Fe(III) by bacteria. 92

105 APPENDIX A: AQUEOUS AND SURFACE SPECIATION REACTIONS USED IN THERMODYNAMIC MODELING. TABLE A.1 AQUEOUS AND SURFACE SPECIATION REACTIONS USED IN THERMODYNAMIC MODELING. Log K (I=0) Reference Ca Surface Complexation Reactions 1 Ca 2 UO 2 (CO 3 ) R-L2 - (a) = [R-L2-Ca 2 UO 2 (CO 3 ) 3 ] (b) 2 Ca 2+ + R-L2 - (a) = [R-L2-Ca] (b) Ca Aqueous Complexaion Reactions 3 Ca CO UO = Ca(UO 2 )(CO 3 ) 3 4 2Ca CO UO = Ca 2 (UO 2 )(CO 3 ) 3 5 Ca CO 3 = CaCO 3 6 Ca 2+ + H CO 3 = CaHCO (c) (c) 3.22 (d) (d) 7 Ca 2+ + OH - = CaOH (d) 93

106 TABLE A.1 AQUEOUS AND SURFACE SPECIATION REACTIONS USED IN THERMODYNAMIC MODELING. Log K (I=0) Reference Lactate & Acetate Protonation and Complexaion Reactions 8 H + + C 3 H 5 O 3 (Lactate) - = C 3 H 6 O 3 (Lactate) (e) 9 Ca 2+ + C 3 H 5 O 3 (Lactate) - = Ca-C 3 H 5 O 3 (Lactate) (e) UO 2 + C 3 H 5 O 3 (Lactate) - = UO C 3 H 5 O 3 (Lactate) (f) 11 H + + C 2 H 3 O 2 (Acetate) - = C 2 H 4 O 2 (Acetate) (e) 12 Ca 2+ + C 2 H 3 O 2 (Acetate) - = Ca-C 2 H 3 O 2 (Acetate) (e) UO 2 + C 2 H 3 O 2 (Acetate) - = UO C 2 H 3 O 2 (Acetate) (f) Uranyl Hydroxide Aqueous Complexation Reactions 14 UO H 2 O = UO 2 OH + + H (g) 15 UO H 2 O = UO 2 (OH) 2(aq) + 2H (g) 16 UO H 2 O = UO 2 (OH) H (g) 17 UO H 2 O = UO 2 (OH) H (g) 18 2UO H 2 O = (UO 2 ) 2 OH 3+ + H (g) 19 2UO H 2 O = (UO 2 ) 2 (OH) H (g) 20 3UO H 2 O = (UO 2 ) 3 (OH) H (g) 21 3UO H 2 O = (UO 2 ) 3 (OH) H (g) 22 3UO H 2 O = (UO 2 ) 3 (OH) H (g) 23 4UO H 2 O = (UO 2 ) 4 (OH) H (g) 94

107 TABLE A.1 AQUEOUS AND SURFACE SPECIATION REACTIONS USED IN THERMODYNAMIC MODELING. Log K (I=0) Reference Uranyl Carbonate Aqueous Complexation Reactions 24 UO CO 3 = UO 2 CO 3(aq) 9.94 (g) 25 UO CO UO CO 3 2- = UO 2 (CO 3 ) 2 2- = UO 2 (CO 3 ) (g) (g) 27 3UO CO 3 2- = (UO 2 ) 3 (CO 3 ) (g) 28 2UO H 2 O + CO UO H 2 O + CO 3 2- = (UO 2 ) 2 CO 3 (OH) H (g) = (UO 2 ) 3 CO 3 (OH) H (g) 30 11UO H 2 O + 6CO 3 2- = (UO 2 ) 11 (CO 3 ) 6 (OH) H (g) Other Reactions 31 H 2 O = H + + OH (d) 32 H 2 CO 3 = H + + HCO 3-33 H 2 CO 3 = 2H + + HCO (d) (d) (a) R-L#- represents different S. oneidensis funtional groups, Sites 1-4, with pka values of 3.3±0.2, 4.8±0.2, 6.7±0.4, and 9.4±0.5, respectively (Mishra et al., 2010). (b) Gorman-Lewis et al., (2005). (c) Dong and Brooks, (2006). (d) Martell and Smith, (2001). (e) Martell and Smith, (1977). (f) The K values of these reactions have only been determined at ionic strength (I) = 1 (Martell and Smith, 1977). The K values were extrapolated to infinite dilution (I=0) here using the Davies equation. (g) Guillaumont et al., (2003). 95

108 APPENDIX B: U(VI) SPECIATION DIAGRAMS Figure B.1: Speciation diagrams for systems with a total U(VI) concentration of 0.25 mm, and NaHCO3 concentrations of: (a) 0.0, (b) 0.2, (c) 2.4, (d) 11.9 and (e)30.0 mm in 0.1 M NaClO4 solution. Only species with concentrations greater than 5% of the total U(VI) concentration are shown. 96

109 97

110 98

111 99

112 APPENDIX C: U(VI) REDUCTION DATA Figure C.1: The concentrations of dissolved U(VI) remaining in solution in the presence of (a, b) 2.4 mm; (c, d) 5.0 mm; (e, f) 7.2 mm; (g, h) 11.9 mm; (i, j) 21.0 mm; (k, l) 30.0 mm NaHCO 3 as a function of time. In each figure, open triangles represent the nonacidified dataset; solid diamonds represent the acidified dataset; open squares represent the data points used to calculate the initial rate of U(VI) reduction by S. oneidensis. 100

113 101

114 102