«THERMODYNAMIC MODELING OF METAL ADSORPTION AND MINERAL SOLUBILITY IN GEOCHEMICAL SYSTEMS A Dissertation Submitted to the Graduate School of the ...»
(4) How does the incorporation of neptunium into the structure of the uranyl silicate soddyite affect the mineral solubility? Soddyite with neptunium in the structure is likely to form under the oxidizing conditions of some nuclear repositories. If the neptunium has a substantial effect on soddyite solubility, the mobility of neptunium and uranium in a geologic repository setting could be much different than currently predicted. To answer this question, I synthesized soddyite powders with varying trace amounts of Np in the structure. Chapter 5 presents the results of solubility experiments performed with these Np-incorporated soddyites.
1.2 Monovalent Cation Adsorption to Bacterial Surfaces Water in the subsurface contains a range of metal ions that compete for adsorption sites on soil and aquifer surfaces. Although the binding of various metals onto bacterial cell wall functional groups has been studied extensively (e.g., Beveridge and Murray, 1976; Beveridge, 1989; Ledin et al., 1997; Yee and Fein, 2001; Fein et al., 2002; Borrok and Fein, 2004; Gorman-Lewis et al., 2005; Covelo et al., 2007; Johnson et al., 2007), the binding constants for monovalent cations, which in many systems are the most concentrated cations present, are not known. Laboratory measurements of divalent metal adsorption onto bacterial surfaces are usually conducted in the presence of a concentrated monovalent salt electrolyte to buffer ionic strength, but monovalent cations are assumed to be inert to adsorption onto bacterial surface functional groups.
The decrease in divalent metal adsorption to surfaces with increasing ionic strength is often attributed to a decrease in the strength and extent of the electric field of the adsorbing surface with increasing ionic strength. However, if monovalent metals adsorb onto surface functional groups and thereby compete with divalent metals for available sites, an alternative approach for modeling the ionic strength effect on divalent metal adsorption may be to account for monovalent adsorption explicitly.
In our study, we test this alternative approach to accounting for the ionic strength effect by using a non-electrostatic surface complexation model (NEM). The NEM approach assumes that the functional groups on surfaces behave the same as dissolved ligands, so that the electrostatic terms can be neglected entirely (Davis and Kent, 1990).
Instead, ions in solution are assumed to directly and simultaneously compete for surface functional groups, the assumption being that monovalent adsorption is onto specific sites rather than only due to electrostatic attraction from the surface electric field in general.
In this regard, NEMs are simpler to apply than electrostatic models. Although electrostatic interactions do occur between ions and surfaces, they are difficult to model due to the difficulty in modeling the nature and extent of the surface electric field, especially in complex realistic systems (Davis et al., 1998). A range of surface electric field models exist, but all are empirical in that they require calibration with a number of adjustable parameters, and typically the models fit experimental adsorption data equally well (Hayes et al., 1991). The NEM approach minimizes the number of required modeling parameters, but without an understanding of site-specific monovalent ion adsorption, the NEM approach cannot account for the effect of ionic strength on cation adsorption. The research presented in Chapter 2 represents an attempt to adapt nonelectrostatic adsorption models to account for these ionic strength effects without having to characterize the nature and extent of the surface electric field.
In the first study of this dissertation (Chapter 2), I conduct experiments to determine the extent of adsorption of monovalent metals to the Gram-positive soil bacteria Bacillus subtilis in order to calibrate a non-electrostatic adsorption model that can account for the effect of ionic strength on metal adsorption. The experiments are performed as a function of ionic strength and pH. Because bacterial cells contain high concentrations of Na+ and K+ that could enter into solution during metal adsorption experiments, the experiments were not conducted with these metals, but instead with Li+ and Rb+. A NEM is used to model the monovalent adsorption data and solve for discrete metal-bacteria binding constants for the monovalent cations. These constants are then used to determine the competitive effect of monovalent cations on the adsorption of Cd2+onto bacterial surface adsorption sites. This approach may serve as an alternative modeling approach to electrostatic models for accounting for ionic strength effects on cation adsorption onto bacterial surfaces.
1.3 Testing the Fumigation-Extraction Method for Soil Microbial Biomass Modeling metal speciation, bioavailability, and mobility in systems such as soils and aquifers using a surface complexation approach requires accurate and precise estimates of the type and quantity of each type of binding site in the system, and the binding constants for the cation of interest onto each binding site type. Soils are complex, having a myriad of possible combinations of constituent parts (Gardiner and Miller, 2004). In the most general sense, natural soils consist of inorganic and organic components and soil moisture. The inorganic fraction consists of minerals such as quartz, iron oxides, feldspars, clays, and micas. The organic portion includes humic substances and microorganisms. In order to develop general models of metal speciation and distribution in soils, not only do we need to know the equilibrium constants for each binding reaction that occurs, but we also need accurate and precise methods to quantify the concentration of each component in a natural soil sample.
Commonly used methods of biomass determination include direct counting of bacteria using a microscope and the use of adsorbent fluorescent dyes for spectrophotometric measurement of cells. These methods yield highly imprecise values for cell concentrations (Poglazova et al., 1996). Jenkinson and Powlson (1976a, b) introduced a biocidal fumigation method for soils to determine the live cell biomasscarbon in soils. This approach is significantly more precise than previous biomass determination methods; it is able to detect microbial biomass differences of 5% to 10% at a 0.05% probability level (Voroney et al., 2008). The work of Jenkinson and Powlson demonstrated that chloroform (CHCI3) fumigation completely lyses live cells in 24 hours, and the evolved gas is collected. Organisms killed during the fumigation process are readily mineralized to CO2, so that the difference in CO2 gas evolution between fumigated and unfumigated samples is a measure of biomass-carbon (Smith et al., 1995).
Vance et al. (1987) and Tate et al. (1988) used a similar method, but extracted organic carbon from fumigated and unfumigated samples using a 0.5 M K2SO4 solution. These solutions were analyzed for total dissolved organic carbon as the measure of biomasscarbon. The Vance et al. (1987) method attributes the enhanced amount of organic carbon extracted from a fumigated sample relative to an unfumigated control exclusively to cell lysis caused by chloroform fumigation. The method would be highly inaccurate if chloroform adsorbed onto any soil components during fumigation and desorbed during the extraction procedure. Under these circumstances, the chloroform would subsequently enhance the amount of organic carbon extracted from a fumigated sample, and therefore the enhancement of organic carbon could not be attributed exclusively to sample biomass.
Although both the Vance et al. (1987) and the Tate et al. (1988) papers are heavily cited and considered current standard procedures for biomass determination in soils, control experiments using individual soil components have never been performed, and the accuracy of the approach remains untested. Haney et al. (1999) questioned the acceptability of the fumigation-extraction method to determine biomass carbon by showing that the amount of carbon extracted using a 0.5 M K2SO4 solution can vary as a function of pH. There is also some evidence that suggests that chloroform can adsorb to soils both from aqueous solution (Dural and Peng, 1995) and from the atmosphere (Chen, 1993). In this study (Chapter 3), I test whether the fumigation-extraction method is valid for different soil types by performing fumigation experiments with individual soil components, including a humic acid, a quartz sand, two types of clays, and a bacterial species. By determining which soil components retain chloroform through the fumigation procedure, this study determines which types of soils are likely to yield inflated biomass carbon readings using the fumigation-extraction method.
1.4 Predicting Metal Adsorption to Mixtures of Geosorbents The ultimate goal of this project is to test the accuracy of surface complexation models that predict metal adsorption and distribution in multi-sorbent systems.
Although metal adsorption in multi-component systems has been studied empirically (e.g., Ledin et al., 1997,1999; Krantz-Rulcker et al., 1996; Fingler et al., 2004), the ability of the SCM approach to predict metal adsorption and distribution in soils has not been tested. Davis et al. (1998) attempted to use a SCM to predict Zn2+ adsorption to mineral assemblages. They used a component additivity (CA) approach to independently predict adsorption onto assemblages of minerals. Specifically, the authors conducted surface titrations and Zn2+ adsorption experiments on individual minerals and calculated proton- and metal-surface complex stability constants for each mineral surface site type. The adsorption behavior of Zn2+ onto mixtures of these minerals was then measured and compared to independent predictions of the adsorption behavior based on the calculated individual stability constants and the known concentration of each site type in the mixtures. Davis et al. (1998) found that the CA approach could not account for the pH-dependent adsorption behavior, likely due to interactions among the minerals that were not accounted for in the CA surface complexation model.
Davis et al. (1998) also tested a generalized composite (GC) approach to modeling adsorption. The GC is a semi-empirical method that attributes metal adsorption onto mixtures of surfaces to generic functional groups, rather than the specific groups of the CA. Unlike the CA approach, the GC approach cannot be extended to other mixtures of the same surfaces because it is only valid for the surface ratios present in the experiment for which it was calibrated. Because of its added flexibility and increased number of adjustable parameters, Davis et al. (1998) found that the GC approach was more successful in accounting for Zn2+ adsorption to mineral assemblages than was the CA approach. However, Pagnanelli et al. (2006) and Fowle and Fein (1999) demonstrated that a non-electrostatic surface complexation model (NEM) approach combined with the CA approach can be successful in predicting metal adsorption to mixtures of pure minerals and bacteria, respectively. It remains unclear if the CA approach is capable of predicting metal distribution in mixtures containing minerals, microbes, and organic acids.
In Chapter 4,1 test whether the CA approach can account for the distribution of Cd, a toxic metal of environmental interest, between an aqueous phase and mixtures of kaolinite, bacteria, iron oxyhydroxide, and dissolved acetate. In order to obtain internally consistent stability constants for the important Cd surface complexes, I first measure adsorption of Cd2+ onto the individual components. From these data, the stability constants between Cd2+ and each individual component are calculated. The distribution of Cd in binary and ternary mixtures of the components is measured in adsorption experiments and the data are compared to independent predictions from the CA approach. The results demonstrate conditions for which the CA approach may be appropriate in predicting Cd2+ distribution, and those for which more complex models that include interactions between the sorbents are necessary.
l.S Solubility of Np-incorporated Soddyite Spent nuclear fuel is likely to alter to uranyl minerals under the moist oxidizing conditions of a geological repository (Finch et al., 1999; Finn et al., 1996; Wronkiewicz et al., 1992,1996). Radionuclides, such as neptunium, may become incorporated into these secondary uranyl mineral structures (Burns et al., 1997), potentially altering the solubility of the phases and hence the mobilities of U and Np in the repository environment. Soddyite ((UC^MSiCUXFbO^) forms as a common alteration product of spent nuclear fuel in laboratory settings (e.g., Finch et al., 1999), and significant concentrations of Np(V), which is likely to be present in an oxizing repository, can be incorporated within soddyite (Klingensmith and Burns, 2007). The solubility and thermodynamic properties of pure soddyite have been studied (Gorman-Lewis et al., 2007). It is unclear, however, what effect Np(V) incorporation into the mineral structure of soddyite has on the mineral solubility or on the extent to which Np is released from the phase.
Gnanapragasam and Lewis (1995) reported that trace levels of radium incorporation exert no effect on the solubility of gypsum and calcite. Curti (1999) reported that the incorporation of rare earth metals and Cd2+ leads to the formation of insoluble carbonates, but the incorporation of Mg2+, K+, Na+ and Li+ forms soluble to very soluble carbonates. The closest analog to this study is that of Rai et al. (2004), who determined the solubility effect of Np(IV) incorporation into uraninite (U02(S)). The authors observed ideal solid solution behavior, or a decrease in the aqueous U(IV) concentration in equilibrium with the solid phase that is equal to the decrease in the mole fraction of U(IV) within the solid phase with increasing extents of Np(IV) substitution.
The ideal substitution behavior found by Rai et al. (2004) is likely due to similarities in size and charge of U(IV) and Np(IV). However, Np(V) forms the neptunyl cation, NpC2+, which has different bonding environment and charge than the uranyl cation, U0 2 2+ (Forbes et al., 2008). In order for Np0 2 + to substitute for U0 2 2+ in soddyite, it is likely that co-substitution of another ion, such as Na+, must occur, creating a more complex, non-ideal substitution mechanism.