«THERMODYNAMIC MODELING OF METAL ADSORPTION AND MINERAL SOLUBILITY IN GEOCHEMICAL SYSTEMS A Dissertation Submitted to the Graduate School of the ...»
(1998) also applied the generalized composite (GC) approach, a semi-empirical method that assigns generic functional groups to the mineral mixture. The GC approach was more successful in modeling Zn adsorption behavior onto the sediment, but this approach is not predictive, and can only be applied to systems for which laboratory calibration exist. Kulczycki et al. (2005) found that the Pb and Cd adsorption onto twocomponent systems comprised of ferrihydrite and either Bacillus subtilis or Escherichia coli bacterial cells was less than what would be expected by summing the metal adsorption to each component determined in one-component experiments. The authors speculated that adhesion between the ferrihydrite and the bacteria masked some of the surface sites on these sorbents, decreasing the adsorption capacity of the mixture.
Several authors have successfully applied SCM models to mixtures of minerals or bacteria, in some cases with dissolved organic ligands present in the systems. For example, Pagnanelli et al. (2006) measured the adsorption of protons and Pb to quartz, muscovite, clinochlore, goethite, and hematite, individually, using a non-electrostatic SCM. They found that a component additivity SCM, using the acidity and equilibrium constants determined for each pure mineral phase, could successfully predict Pb distribution in mixtures of these solids. Fowle and Fein (1999) demonstrated that the CA approach can be successful in predicting metal adsorption to mixtures of Bacillus licheniformis and Bacillus subtilis bacterial cells. Additionally, Yee and Fein (2003) showed that Cd, Co, Sr, and Zn adsorption onto complex mixtures of 10 species of Gram-negative and Gram-positive bacteria can be predicted with the CA approach. Lund et al. (2008) used a component additivity approach, with a diffuse layer model (DLM) to account for electric field effects, to predict the adsorption of Cu onto mixtures of hydrous ferric oxide (HFO) and kaolinite. The goodness-of-fits of the models to the experimental data were dependent on the model fits to the HFO and kaolinite individually. For mixtures of HFO and kaolinite, the authors postulated that interactions between these solids did not significantly impact Pb adsorption.
Although some tests of the CA approach have been conducted, few have tested its ability to account for metal distributions in systems containing both bacteria and minerals. Applying the CA approach to mixtures of minerals and bacteria may be problematic because bacteria can adhere to mineral surfaces, blocking reactive sites on the bacteria and the mineral (Lower et al., 2001; Ams et al., 2004). In this study, we test the ability of the CA approach to predict metal adsorption behavior in systems that contain mixtures of common soil components: kaolinite, hydrous ferric oxide (HFO), Bacillus subtilis bacterial cells, and acetate as a representative simple dissolved organic acid. We use literature values to describe the proton reactivity and site concentrations for each sorbent, and we calibrate the model by measuring Cd adsorption onto each sorbent separately, using the results to calculate stability constants for the Cd-surface complexes. Our objective in this study is not to determine stability constants that can describe Cd adsorption over a wide range of Cd and sorbent concentrations. Rather, the emphasis is on testing if the CA approach can account for sorbent competition for Cd at a few specific experimental conditions. For this reason, we use the Cd-sorbent stability constants that we calculate for a given Cd: sorbent concentration ratio in single-sorbent systems in order to construct CA models that predict the distribution of Cd in multisorbent systems that have the same Cd:total sorbent concentration ratio. We vary the ratios of the sorbents, keeping the total sorbent concentrations constant, and we test systems that both include and exclude the effects of acetate complexation. We compare the CA predictions to the observed extents of adsorption for these systems to test the validity of the CA approach and to determine its strengths and limitations.
4.2 Methods 4.2.1 Preparation of bacterial cells Bacillus subtilis, a Gram-positive soil bacterium, was selected as the biosorbent.
The cell wall of this species is well-characterized (e.g., Beveridge, 1989) and the acidity constants and site concentrations of surface functional groups are well-constrained (Fein.
et al., 1997; 2005). The bacteria were grown and harvested in a manner similar to that described by Borrok et al. (2004). The bacteria were initially grown on agar slants made of 0.5% yeast extract and trypticase soy agar. Cells from the slant were transferred to a 3 ml test tube containing trypticase soy broth (TSB) and 0.5% yeast extract, and allowed to grow for 24 hours at 32°C. After the growth period, bacteria were transferred to 11 solutions of the same composition, and allowed to grow for another 24 hours at 32°C.
Bacteria were harvested in the stationary phase by centrifuging the broth at 9,000 g for 10 minutes to pellet the bacteria. After decanting the broth, the bacteria were washed four times in a 0.1 M NaClC«4 solutions. Between each wash, the bacteria were centrifuged at 8,100 g for 5 minutes to pellet the bacteria, the supernatant was decanted, and the cells were suspended in a fresh 0.1 M NaC104 electrolyte. After the washing cycles, the bacteria were transferred to a weighed centrifuge tube after the final wash, and centrifuged one time for 4 min and two times for 30 min at 8100 g, decanting the remaining supernatant each time. The resulting wet weight of the B. subtilis bacterial pellet is approximately 5 times the dry weight (Borrok et al., 2004). The method of preparation described here removes excess growth media and adsorbed cations from the bacterial surface, and renders the bacteria alive, but metabolically inactive (Wightman and Fein, 2005).
4.2.2 Preparation of mineral powders High-defect kaolinite, KGa-2, from the Source Clay Repository, was used as the clay sorbent in our systems. Following a procedure similar to that used by Schroth and Sposito (1997), the kaolinite was washed repeatedly in a 1 M NaC104 solution that was previously adjusted to pH 3.0 with concentrated HC1. After each wash, the supernatant pH was measured, the clay suspension was centrifuged at 8100 g for 5 minutes to pellet the clay, and the supernatant was discarded. This process was repeated until the supernatant pH stabilized at 3.0. The clay was then washed in non-acidified solutions of NaCKX gradually decreasing in ionic strength from 1 M to 0.1 M. Threefinalwashes in 0.1 M NaClC4, the electrolyte and ionic strength of the adsorption experiments, were conducted, and the pH stabilized at 4.5. The clay was then dried at 25°C, ground to a fine powder, and stored in a sealed centrifuge tube.
HFO was produced by titrating a 1.01 solution of 0.05 M Fe(N03)3»9H20 with small volumes of concentrated NaOH to increase the solution pH to 6.0. The solution was stirred for 24 h to allow the HFO to precipitate fully, after which the solution was decanted and the HFO powder was washed three times with 18 MQ ultrapure water. The powder was dried and stored in a sealed polycarbonate test tube.
4.2.3 Cd adsorption experiments Batch Cd adsorption experiments were performed at initial Cd concentrations of
8.9 x 10"5 M and 8.9 x 10"6 M, referred to hereafter as ' 10 ppm' and ' 1 ppm' experiments for simplicity. To determine the stability constants for each important Cdsurface complex, we measured the adsorption of Cd onto each sorbent separately as a function of pH. 1 g l"1 of kaolinite, B. subtilis, or HFO was suspended in a 0.1 M NaC104 solution. A small volume of a 1000 mg l'1 Cd stock solution was added to each bacterial suspension, with the amount determined gravimetrically, to achieve the desired Cd concentration. The Cd stock solution was prepared from a Cd(C104)2 salt. While stirring, the bulk suspension then was divided into 8 ml aliquots in polycarbonate test tubes, and small volumes of concentrated HNO3 or NaOH were used to adjust the pH of each experiment so that a set covered a pH range between approximately 2 and 8. This pH range was selected to avoid the precipitation of Cd-hydroxides that occurs at the experimental Cd concentrations under higher pH conditions. The test tubes were then placed on a rotary shaker for 2 h, after which the final pH of each solution was measured. The experimental systems were then centrifuged at 8,100 g for 10 min to pellet the solid sorbent, and the supernatant was decanted and filtered through 0.45 um Nylon membranes. The resultingfilteredsupernatants were acidified with 15 yd of 15.8 N HN03.
Allfilteredexperimental supernatants were analyzed for Cd concentrations on the same day that they were collected, using inductively coupled plasma - optical emission spectroscopy (ICP-OES). The amount of Cd adsorbed in each experimental system was determined by difference between the initial known Cd concentration in each experiment and the measured Cd concentration remaining in solution after equilibration with each sorbent. Aqueous Cd standards for ICP-OES calibration were prepared gravimetrically from a 1000 mg l"1 Cd stock solution made from a Cd(NOa)2 salt, diluted to desired concentrations using the same 0.1 M NaClCU matrix as was used in the experimental systems. The Cd signal strength reported by the ICP-OES did not vary significantly with solution ionic strength, and analytical uncertainty as determined by repeat analysis of standards was ±3%.
Parent suspensions, consisting of 1 g l"1 of each sorbent in 0.1 M NaC104, were mixed to generate systems containing two or three sorbents. In this way, the total sorbent concentration in one, two, and three component systems was a constant 1 g l"1, while the ratio of one sorbent to another could be varied. After mixing, a small aliquot of the Cd stock solution described above was added to the mixture. The solutions were divided into polycarbonate test tubes and the pH adjusted with small volumes of concentrated NaOH or HNO3. The test tubes were then placed on a rotary shaker for 2 h, after which the equilibrium pH was measured. Centrifugation,filtration,and acidification sampling procedures were the same as those described above for the onecomponent systems.
In the experiments that involved acetate, 1 g l'1 of sorbent was initially suspended in a 0.1 M NaC104 solution that contained 0.3 M acetate as sodium acetate. The suspension was then spiked with Cd to the desired concentration, and the experiments were conducted using the same procedure as is described above. Experiments in the presence of dissolved acetate were conducted with one-sorbent kaolinite, B. subtilis, or HFO systems, and in the presence of all three sorbents simultaneously.
4.3 Results and Discussion 4.3.1 Cd adsorption onto individual sorbents The extents of Cd adsorption onto 1 g T1 HFO, 1 g l'1 Bacillus subtilis cells, and 1 g l"1 kaolinite are depicted as a function of pH for the 10 ppm (Figure 4.1 A) and 1 ppm Cd (Figure 4. IB) experiments. For each of the sorbents, Cd adsorption generally increases with increasing pH from 2 to 8. At both Cd concentrations, HFO adsorbs less than 20% of the total Cd in the systems below pH 5, and the extent of adsorption increases most significantly between pH 6 and 7 to nearly 100% in the 1 ppm Cd experiments and to nearly 80% in the 10 ppm Cd experiments. Bacillus subtilis cells exhibit a more shallow pH adsorption edge than does the HFO. The cells adsorb Cd to a similar extent to that of HFO below pH 6, but at higher pH, the extent of adsorption does not increase as much for the bacterial systems as it does for HFO, with the maximum extent of adsorption being 75% at pH 7.9 for the 1 ppm Cd experiments, and 40% at pH
7.7 for the 10 ppm Cd experiments. The kaolinite clay adsorbs Cd more weakly on a per
gram basis than the HFO or the B. subtilis cells, never exceeding 20% of the total Cd concentration in the experiments.
4.3.2 Modeling approach We employ a non-electrostatic model (NEM) surface complexation approach to model the Cd adsorption behavior to Bacillus subtilis cells and to kaolinite, and we use a diffuse double layer model (DLM) to describe Cd adsorption to the HFO surface. Fein et al. (2005) performed potentiometric titrations of Bacillus subtilis cells and demonstrated that both electrostatic and non-electrostatic models can fit potentiometric titration data for bacteria equally well, but that surface electric field effects are small for bacteria. For our modeling, we employ the Fein et al. (2005) four-site NEM to describe the surface protonation and site concentrations on the bacterial cell wall. The NEM approach may be preferable for complex geologic applications because it requires fewer fitting parameters than do electrostatic models of surface electric field effects. Schroth and Sposito (1997,1998) conducted potentiometric titrations and metal adsorption experiments on the KGa-2 kaolinite clay used here, and employed two amphoteric, proton-active surface sites and a permanent, pH independent structural charge site to describe protonation and metal adsorption onto the clay surface. The permanent structural charge sites account for the adsorption effects of isomorphic substitutions in the mineral lattice; the amphoteric sites likely represent surface silica and alumina functional groups. The surface reactivity of HFO is described by the DLM of Dzombak and Morel (1990) that employs strong and weak amphoteric surface sites to explain surface protonation and metal adsorption at =FeOH sites. The protonation and cadmium adsorption reactions, and related stability constants for each sorbent are listed in Table 4.1.
4.3,2.1 Calculation of Cd-Bacillus subtilis stability constants The basis of the Cd adsorption model is the non-electrostatic protonation model of Fein et al. (2005). In this model, deprotonation of cell wall organic acid functional
groups is described by the generic reaction:
where At represents one of the four organic acid functional group types needed to account for the protonation behavior of the cell wall (/ = 1 - 4), and R represents the cell wall macromolecule to which the functional group At is attached. The mass action
equation for the deprotonation reaction is: