Introduction

The application of biotechnology for remediation of metals and radionuclides has been well documented (Banaszak et al. 1998, 1999a, b; Caccavo et al. 1992; Farrell et al. 1999; Gorby and Lovley 1992; Haas and Dichristina 2002; Liu et al. 2002; Lovley et al. 1993; Reed et al. 2010; Rittmann et al. 2002a; Songkasiri et al. 2002; Truex et al. 1997). Unlike for organic contaminants, a bioremediation strategy for radionuclides and metals transforms them into phases that should render them immobile and recalcitrant (NRC 2000). This often proceeds via reduction of a radionuclide to an oxidation state that is less soluble.

The occurrence of plutonium (Pu) in the subsurface environment is a concern due to its long half-life (2.4 × 104 years) and toxicity (Neu et al. 2005). Under oxidizing conditions, the predominant forms of plutonium are Pu(V)O2 + and Pu(VI)O2 2+, which form distinct inorganic/organic complexes when present as a contaminant in groundwater (Choppin 2003; Cleveland and Rees 1981). While PuO2 + (i.e., Pu(V)) does not form hydroxyl complexes until pH > 8 and should be very mobile under most subsurface conditions, PuO2 2+ (i.e., Pu(VI)) forms strong hydroxyl complexes and sorbs strongly to aquifer surfaces. However, Pu(VI) is easily reduced and not likely to persist in biologically active systems unless it undergoes irreversible aggregation and polymerization, which may stabilize it against reduction, leading to enhanced subsurface mobility and persistence in the oxidized form (Francis 2007; Reed et al. 2010).

The key to plutonium immobilization in the subsurface is its reduction to Pu(IV) and Pu(III) species. Of these two reduced oxidation states, Pu(IV) exhibits a lower solubility and a much higher tendency towards aggregation, polymer formation, and sorption. From this perspective, Pu(IV) is by far the preferred target oxidation state (Francis 2007; Reed et al. 2010).

Only a few papers have been published on bioreduction of Pu under anaerobic conditions: by Bacillus polymyxa and circulans (Rusin et al. 1994), Geobacter metallireducens GS15 and Shewanella oneidensis MR1 (Boukhalfa et al. 2007; Icopini et al. 2009), and Clostridium sp. (Francis et al. 2008). Furthermore, evidence of Pu reduction by abiotic mechanisms (Rai et al. 2002; Reed et al. 2006; von Gunten and Benes 1995) suggests that bioremediation strategies in subsurface environment require an understanding of coupled abiotic and biotic processes for accurate prediction of plutonium fate in complex subsurface environment (Banaszak et al. 1999a; Lovley 1993; Neu et al. 2002; Reed et al. 2010).

Our group published work on bioreduction of higher-valent uranium and plutonium (Reed et al. 2007), where part of that work investigated Pu(V) reduction in the absence and presence of Fe3+ and NTA. Specifically, we investigated the bioreduction of Pu(V) under anaerobic conditions with Shewanella alga BrY (a facultative metal-reducing bacterium (Caccavo et al. 1992, 1996) in the presence and absence of Fe3+–NTA, an aqueous form of ferric iron. Although Pu(V) reduction occurred in both cases, reduction was significantly slower in the presence of iron. The suggested explanation was that Fe3+ was the preferred electron acceptor (over Pu(V)), but Fe2+, once generated by bioreduction, caused concurrent abiotic reduction of Pu(V).

In this study, we use modeling analyses to understand the mechanisms governing the previously observed slow reduction of Pu(V) in the presence of iron (Reed et al. 2007). We also discuss new spectroscopic results that expand our ability to interpret the experimental findings.

Materials and methods

All methods to synthesize PuO2 +; assay for Pu, Fe, lactate, and other organics; determine oxidation states of PuO2 + and Fe (Fe3+ and Fe2+); grow S. alga; and carry out anaerobic experiments are the same as reported in Reed et al. (2007). For this work, we added spectroscopic analyses of the reduced Pu solid formed. The spectroscopic analysis was done using X-ray absorption near edge spectroscopy (XANES) with MR-CAT beamline at the Advanced Photon Source (APS) following the methods of Songkasiri et al. (2002).

Modeling

Background information on the biogeochemical model CCBATCH

Modeling was performed using the biogeochemical model CCBATCH (Rittmann et al. 2002b; VanBriesen and Rittmann 2000a, b), which was developed by our team to quantitatively link all the different types of reactions that control the fate of radionuclides and a range of other metals and organic co-contaminants. The basic structure of CCBATCH is described here, and then special features needed to use the model to represent the experiments reported here are described. Certain details, such as parameter values, are reported as needed in “Results and discussion” section.

CCBATCH explicitly couples biological electron-donor and -acceptor consumptions to simultaneous chemical reactions in order to determine the effect of biological reactions on the fate of various components in the system. The original CCBATCH model (VanBriesen and Rittmann 1999, 2000a) couples microbially catalyzed reactions, which are kinetically controlled, with aqueous phase acid/base and complexation reactions, which are at thermodynamic equilibrium. Rittmann et al. (2002b) added a sub-model that links kinetically or equilibrium-controlled precipitation/dissolution to the microbial and aqueous-phase reactions. We used the equilibrium sub-model. CCBATCH was designed to describe batch reactions, such as those performed in this study. The microbial sub-model includes oxidation of an electron-donor substrate (e.g., lactate in our studies), synthesis and endogenous decay of biomass, stoichiometric utilization of an electron acceptor, and stoichiometric consumption or generation of inorganic carbon, ammonium–nitrogen, and acidic hydrogen. The equilibrium feature of the sub-model precipitates or dissolves just the amount of solid phase so that the aqueous phase speciation of the cation and anion match the solubility product (Ksp) for every time step. In general, precipitation consumes basic species, and the sub-model represents this through stoichiometric production of acidic hydrogen. Finally, CCBATCH computes pH based on changes in acidic hydrogen and solving a proton condition. To solve for equilibrium speciation, CCBATCH uses a Newton–Raphson technique that combines the aqueous-phase mass balances with mass action equilibrium expressions for all relevant acid/base and complexation reactions, and it can compute the equilibrium pH from the proton condition when the pH is not fixed.

Upgrading CCBATCH for anaerobic growth

To use CCBATCH to represent the experiments in which we saw slow reduction of Pu in the presence of Fe3+–NTA, we first needed to upgrade the model to include anaerobic growth of S. alga using Fe3+, not O2, as the primary electron acceptor. One challenge is that Fe3+ complexes with many anionic species. We added complexes of Fe3+ with the common components of the medium, along with their respective equilibrium constants and stoichiometric mass balances; the complexes and formation constants are listed in Table 1a.

Table 1 (a) Complexes and formation constants for Fe3+ complexes; (b) Kinetic parameters for Fe3+ and PuO2 + reduction
Full size table

A second challenge is that the bioavailabilities of the various Fe3+ species are not known (Haas and Dichristina 2002). Since the pH of our experiments was buffered at neutral, pH has minimal effect on the relative concentrations of all Fe3+ species, and we could use total Fe3+ as the bioavailable form of Fe3+. Table 1b summarizes the parameters used to describe Fe3+ bioreduction and biomass growth from it: qlactate = maximum specific rate of lactate utilization, Ka,Fe(III) = electron acceptor concentration that gives half of the maximum growth rate, and b = first-order endogenous-decay rate. Should only one (or more than one) Fe3+ species be bioavailable, its concentration remains in a constant ratio with total Fe3+; thus, the model would represent the experimental results equally well, but with the value of qlactate increased in proportion to the concentration ratio of total Fe3+ to the bioavailable species. The bioreduction of Fe3+ was not limited by lactate, since it was in excess.

The yield and stoichiometry for all components involved in Fe3+ bioreduction were based on the following overall reaction for lactate utilization coupled to biomass synthesis (Rittmann and McCarty 2001):

$$ 0.25{\text{CH}}_{3} {\text{CHOHCOO}}^{ - } + 0.77{\text{Fe}}^{3 + } + 0.0115{\text{NH}}_{4}^{\, + } \to 0.0115{\text{C}}_{5} {\text{H}}_{7} {\text{O}}_{2} {\text{N}} + 0.25{\text{CH}}_{3} {\text{COO}}^{ - } + 0.1925{\text{H}}_{2} {\text{CO}}_{3} + 0.7815{\text{H}}^{ + } + 0.77{\text{Fe}}^{2 + } $$
(1)

We also included in the model loss of S. alga cells due to cell decay (Rittmann and McCarty 2001; Hacherl and Kosson 2003):

$$ 0.05{\text{C}}_{5} {\text{H}}_{7} {\text{O}}_{2} {\text{N}} + {\text{Fe}}^{3 + } \to 0.05{\text{NH}}_{4}^{\, + } + 0.25{\text{H}}_{2} {\text{CO}}_{3} + 0.95{\text{H}}^{ + } + {\text{Fe}}^{2 + } $$
(2)

Biotic PuO2 + reduction

We added plutonium reactions into the model to represent reduction of PuO2 + in the presence and absence of Fe3+–NTA for the experiments in Reed et al. (2007). In both the cases, we chose Pu3+ as the ultimate reduced form of plutonium based on X-ray absorption near edge spectroscopy (XANES) analysis of the solids formed at the end of the experiments. The potential for Pu(V)O2 + reduction to Pu3+ has been reported before under anaerobic conditions by metal-reducing bacteria: Geobacter metallireducens GS15 and Shewanella oneidensis MR1 (Boukhalfa et al. 2007), Geobacter sulfurreducens and Shewanella oneidensis (Renshaw et al. 2009), and Clostridium sp. (Francis et al. 2008).

Table 2 lists the formation constants of the major aqueous-phase species of Pu(V)O2 +, Pu(IV)4+, and Pu(III)3+ included in the modeling. The last entry in Table 2 is the solubility product for PuPO4(am), the assumed solid phase for Pu3+, which was modeled with the equilibrium feature of the precipitation/dissolution sub-model in CCBATCH (Rittmann et al. 2002b). The parameters that describe PuO2 + bioreduction—k 1 = rate constant of biotic reduction without supporting growth, k 2 = abiotic reduction rate of PuO2 + to Pu4+, and k 3 = abiotic reduction rate of Pu4+ to Pu3+—are given in Table 1b. Based on our observations and literature findings (Boukhalfa et al. 2007; Icopini et al. 2009), we assumed that biotic reduction of Pu(V) does not support growth of S. alga cells. The yield and stoichiometry for all components involved in PuO2 + bioreduction were based on the following overall reaction that coupled to lactate utilization:

$$ 0.25{\text{CH}}_{3} {\text{CHOHCOO}}^{ - } + 0.5{\text{PuO}}_{2}^{\, + } + {\text{H}}^{ + } \to 0.25{\text{CH}}_{3} {\text{COO}}^{ - } + 0.25{\text{H}}_{2} {\text{CO}}_{3} + 0.5{\text{Pu}}^{3 + } + 0.5{\text{H}}_{2} {\text{O}} $$
(3)
Table 2 Formation constants for major Pu(V, IV, III) aqueous complexes at ionic strength = 0.1
Full size table

Adding abiotic PuO2 + reduction

We represented the abiotic reductions of PuO2 + to Pu4+(Eq. 4) and Pu4+ to Pu3+(Eq. 5) by biologically produced Fe2+ with the following overall reaction:

$$ {\text{PuO}}_{2}^{\, + } + {\text{Fe}}^{2 + } + 4{\text{H}}^{ + } \to {\text{Fe}}^{3 + } + {\text{Pu}}^{4 + } + 2{\text{H}}_{2} {\text{O}} $$
(4)
$$ {\text{Pu}}^{4 + } + {\text{Fe}}^{2 + } \to {\text{Fe}}^{3 + } + {\text{Pu}}^{3 + } $$
(5)

For these reactions, we assumed first-order kinetics with respect to PuO2 +, Pu4+ and Fe2+ concentrations:

$$ {\text{Rate}} = k_{2} \left[ {{\text{PuO}}_{2}^{\, + } } \right]\;\left[ {{\text{Fe}}^{2 + } } \right] $$
(6)
$$ {\text{Rate}} = k_{3} \left[ {{\text{Pu}}^{4 + } } \right]\;\left[ {{\text{Fe}}^{2 + } } \right] $$
(7)

The values of k 2 and k 3 are listed in Table 1b.

Results and discussion

Figure 1 compares the experimental (Reed et al. 2007) and (new) modeling results for bioreduction of Fe3+–NTA by S. alga. The modeling results for the transformations of all major components involved in bioreduction of Fe3+ accurately track the experimental results until ~12 h. This period included the parallel and stoichiometric utilization of lactate and reduction of Fe3+ to Fe2+. S. alga showed a small amount of growth, which is clearly shown in the inset. After ~14 h, the model shows plateaus for all the components due to complete consumption of the limiting reactant, Fe3+. The experiments do not show the total reduction of Fe3+ between 12 and 14 h. This can be attributed to the sudden loss of cell viability, as clearly shown in the inset. Thus, the modeling results up to 12 h represent well the transformations of all major components when S. alga was active. Furthermore, given the slow decay rate [i.e., 0.024 day−1 (Table 1)], we do not anticipate a significant loss of S. alga cells due to decay in the timeframe of the experiments conducted in this study.

Fig. 1
figure 1

Bioreduction of Fe3+–NTA by Shewanella alga: comparison of modeling and experimental results (Reed et al. 2007). Model simulations are represented by lines, and experimental results are represented by data points

Full size image

Figure 2A compares model-calculated to experimental results (from Reed et al. 2007) for the bioreduction of PuO2 + to Pu3+ by S. alga. Although the experimental results for PuO2 + are sparse, modeling simulations show a good match to the experimental results past ~17 h. Also shown are model simulations for the stoichiometric oxidation of lactate to acetate and the formation of Pu3+. Modeling indicates that full bioreduction of 10 μM of PuO2 + required oxidation of ~0.18% of the initial lactate, which is practically not measureable.

Fig. 2
figure 2

A Comparison of model-calculated to experimental results (Reed et al. 2007) for the biotic reduction of PuO2+ by Shewanella alga. The solid line is the model simulation, and open data points are experimental results. Also shown are model simulations for consumption of lactate, and productions of acetate and Pu3+. B Speciation of Pu3+ in the absence (solid lines) and presence (dashed line) of Pu3+ precipitate formation as PuPO4(am), but with no NTA

Full size image

Figure 2B shows the model-simulated speciation of Pu3+ in the absence (solid lines) or presence (dashed lines) of Pu3+ precipitate formation, i.e., PuPO4(am). NTA is not present in the medium. When the precipitation subroutine was “turned off” (solid lines for Pu species), the major aqueous Pu3+ species formed and their respective percentages are 26.4% for Pu(OH)2+ (aq) and 73.6% for Pu3+ (aq). The high percentage of free Pu3+ (aq) illustrates the lack of concentration and/or strength of anions in the medium to complex with all the Pu3+. “Turning on” the precipitation subroutine (dashed line) precipitated all the Pu3+ as PuPO4(am). This means that Pu3+ complexation with anions present in the medium was not strong enough to keep Pu3+ in a soluble form when NTA was not present.

Given that the normal anions in the medium could not complex Pu3+ strongly enough to keep it in solution, we explored the effects of having the strong complexing ligand NTA present in the medium. Figure 3 shows modeling simulations for the effect of different concentrations of NTA on Pu3+ precipitation. At 10 μM NTA (equivalent to the initial PuO2 + concentration in the medium), all the Pu3+ precipitated, and the fate of Pu(III) was the same as Fig. 2B with precipitation turned on. In the presence of 1 mM NTA (100 fold more than the initial PuO2 + concentration), Pu3+ did not precipitate at all, because the strength of Pu3+–NTA complex was great enough to completely out-compete Pu3+ precipitation. Decreasing the NTA concentration from 1 to 0.1 mM and 0.05 mM freed Pu3+ (aq), which combined with PO4 3− to form PuPO4(am) to a greater extent with less NTA.

Fig. 3
figure 3

Model simulations for the effect of different concentrations of NTA on Pu3+ precipitate formation as PuPO4(am)

Full size image

Next, we tested our hypothesis for the slowed reduction of plutonium in the presence of Fe3+–NTA: i.e., preferential reduction of Fe3+ followed by abiotic reduction of PuO2 + by the biogenic Fe2+. We previously showed experimental evidence to support abiotic reduction of PuO2 + by biogenically produced Fe2+ (Reed et al. 2007). Figure 4A compares the model-calculated and experimental results for the bioreduction of PuO2 + by S. alga in the presence of Fe3+–NTA. It also shows the buildup of Fe2+ upon biotic reduction of Fe3+. The modeling simulation provide a good fit to the experimental results of the previous experiments (Reed et al. 2007), supporting our hypothesis that the reduction of PuO2 + in the presence of Fe3+ was slowed (~40 h, compared to <17 h for direct biotic reduction of PuO2 +—Fig. 2A) by preferential reduction of Fe3+, but that the produced Fe2+ (Figs. 1, 4A—inset) abiotically reduced PuO2 + to Pu4+ and then to Pu3+. Although abiotic reduction of Pu by Fe2+ was reported for Pu(IV)O2(am) (Rai et al. 2002) and Pu(VI) (Reed et al. 2006), we are the first to describe stepwise reductions of PuO2 + to Pu4+ and to Pu3+ and to quantify how it is coupled with the biotic reactions that produce the Fe2+.

Fig. 4
figure 4

A Comparison of model-calculated and experimental results (Reed et al. 2007) for abiotic (in the presence of Fe3+–NTA) reductions of PuO2+ to Pu4+, followed by Pu4+ to Pu3+, by produced Fe2+. Model simulations are represented by lines, and experimental results are represented by data points. Also shown is the buildup of produced Fe2+ upon preferred biotic reduction of Fe3+ (inset). B Speciation of Pu3+ in the absence (i) and presence (ii) of Pu3+ precipitate formation as PuPO4(am)

Full size image

Figure 4B shows the speciation of Pu3+ in the absence (i) or presence (ii) of PuPO4(am) precipitation. In both cases, almost all (99.8%) of Pu3+ was Pu3+–NTA(aq), which can be attributed to the strength of Pu3+–NTA complex (e.g., Fig. 3). The remaining ~0.2% of Pu3+ was speciated as free Pu3+ (aq) and Pu(OH)2+ (aq), with an additional ~0.07% of PuPO4(am) formed only when the precipitation subroutine was turned on (ii). In contrast to the no-NTA results of Fig. 2B, the presence of NTA in the medium kept Pu3+ in a soluble form in Fig. 4B.

While the desired end product is a bio-precipitated plutonium phase, modeling interpretation of experimental results shows that strong complexing ligands form soluble plutonium phases that are mobile, thus defeating an immobilization strategy. Such undesired reduced products were also observed upon reductions of uranium—forming U(IV)–citric acid complex (Francis and Dodge 2008)—and plutonium—forming Pu3+–EDTA (Boukhalfa et al. 2007) and Pu3+–NTA (Rusin et al. 1994).

Conclusions

We used biogeochemical modeling of experimental results in Reed et al. (2007) to advance our understanding of the fate of Pu in the absence and presence of Fe3+–NTA under anaerobic conditions. In all experiments, S. alga reduced PuO2 + to Pu3+, and evidence indicates that the Pu3+ could be precipitated as PuPO4(am). Modeling simulations support that reduction in the absence of Fe3+–NTA was from direct biotic PuO2 + reduction, but that PuO2 + reduction in the presence of Fe3+–NTA was due to an abiotic reduction reaction by biogenically produced Fe2+. These results explain that PuO2 + reduction was slowed in the presence of Fe3+–NTA because the bacteria preferentially reduced Fe3+ to Fe2+, which then abiotically reduced Pu(V) stepwise to Pu(IV) and then to Pu(III). Modeling results also show that the degree of PuPO4(am) precipitation depended on the concentration of the strong complexing ligand NTA. While precipitation out-competed complexation when NTA had an equimolar or smaller concentration compared to Pu, an excess of NTA could completely prevent precipitation of PuPO4(am).