Environ. Sci. Technol. 2008, 42, 8007–8013
Geochemical Modeling of Reactions and Partitioning of Trace Metals and Radionuclides during Titration of Contaminated Acidic Sediments F A N Z H A N G , * ,† W E N S U I L U O , † JACK C. PARKER,‡ BRIAN P. SPALDING,‡ SCOTT C. BROOKS,† DAVID B. WATSON,† PHILIP M. JARDINE,† AND BAOHUA GU† Environmental Sciences Division, Oak Ridge National Laboratory, P.O. Box 2008, MS 6038, Oak Ridge, Tennessee 37831, and Department of Civil and Environmental Engineering, University of Tennessee, 62 Perkins Hall, Knoxville, Tennessee 37996
Received January 31, 2008. Revised manuscript received August 08, 2008. Accepted August 18, 2008.
Many geochemical reactions that control aqueous metal concentrations are directly affected by solution pH. However, changes in solution pH are strongly buffered by various aqueous phase and solid phase precipitation/dissolution and adsorption/desorption reactions. The ability to predict acid-base behavior of the soil-solution system is thus critical to predict metal transport under variable pH conditions. This study was undertaken to develop a practical generic geochemical modeling approach to predict aqueous and solid phase concentrations of metals and anions during conditions of acid or base additions. The method of Spalding and Spalding was utilized to model soil buffer capacity and pH-dependent cation exchange capacity by treating aquifer solids as a polyprotic acid. To simulate the dynamic and pH-dependent anion exchange capacity, the aquifer solids were simultaneously treated as a polyprotic base controlled by mineral precipitation/ dissolution reactions. An equilibrium reaction model that describes aqueous complexation, precipitation, sorption and soil buffering with pH-dependent ion exchange was developed using HydroGeoChem v5.0 (HGC5). Comparison of model results with experimental titration data of pH, Al, Ca, Mg, Sr, Mn, Ni, Co, and SO42- for contaminated sediments indicated close agreement, suggesting that the model could potentially be used to predict the acid-base behavior of the sediment-solution system under variable pH conditions.
Introduction Historical disposal of wastes at U.S. Department of Energy facilities associated with nuclear material production and processing have created extensive subsurface contamination problems (1). For example, at the Oak Ridge Reservation (ORR) in east Tennessee, wastes containing strong acid, inorganic, organic, and radioactive materials were released into unlined trenches, pits, ponds, and streams and have resulted in approximately 6 km2 (1500 acres) of contaminated groundwater with low pH and high concentrations of * Corresponding author email:
[email protected]. † Oak Ridge National Laboratory ‡ University of Tennessee 10.1021/es800311m CCC: $40.75
Published on Web 10/01/2008
2008 American Chemical Society
aluminum, iron, calcium, magnesium, manganese, nitrate, and various trace metals such as nickel and cobalt and radionuclides such as uranium and technetium. Aquifer pH exerts strong effects on precipitation, redox, complexation, and sorption reactions, and hence, on contaminant mobility. Ability to model pH buffering reactions and pH-dependent surface charge coupled with pH-dependent reactions affecting contaminant mobility is critical to understanding migration risks and designing effective remediation strategies. Hydrolysis and precipitation of pure mineral species involving individual metals or radionuclides are well documented (2-4). Adsorption of ions onto relatively homogeneous materials, such as amorphous aluminosilicates (5), amorphous silica (6), montmorillonite (7, 8), manganese oxide (9), has been extensively investigated. Individual minerals show a wide range in ion exchange capacity depending upon pH, mineral structure, structural substitutions, and the specific surface of the mineral accessible to water (5, 6, 10). For natural materials, multisite exchange models have been used to simulate distributions of distinct exchange sites (11-13). Because natural soils are composed of a complex mixture of individual minerals, which themselves exhibit heterogeneous surface characteristics subject to alteration through coatings, it is difficult if not impracticable to comprehensively model all geochemical reactions that may occur in response to acid or base additions given great uncertainties and variability in mass fractions of various species and values of relevant equilibrium and kinetic parameters. Most ion-exchange models of natural soils use pHdependent exchange capacity based on laboratory measurements. Such models provide satisfactory predictions of sorption under individual pH values with the corresponding fixed exchange capacities (14-17), but they may result in deviations between modeled and experimental pH values (14). Spalding and Spalding have proposed to simulate soil buffering and pH-dependent cation exchange capacity by treating soil solids as an insoluble quadraprotic weak acid with four independent ionization constants, allowing soil acid buffering to be fit to empirical titration data (18). The formulation was successfully applied to simulate Sr adsorption competing with Na, K, Ca, Mg cations under alkaline conditions. This study investigates the modeling of geochemical reactions (including various aqueous phase, precipitation/ dissolution, and adsorption/desorption reactions) that control attenuation of trace metals during base additions to acidic-contaminated sediments. The objective is to describe a practical yet generic approach for modeling acid-base behavior of soil systems that can be utilized to predict pH buffering and geochemical behavior of competing cations (including K, Na, Ca, Mg, Sr, Mn, Co, Ni, Al, U) and anions (including sulfate and U) over a wide range in pH and to demonstrate the model by application to soil titration experiments on ORR soil materials. In this study, the aquifer solids were simultaneously treated as a polyprotic acid and a polyprotic base to model pH-dependent cation exchange capacity (CEC) and anion exchange capacity (AEC), respectively. The model also describes pH-dependent AEC due to precipitation/dissolution of Al and Fe minerals to dynamically simulate changes in surface site density.
Materials and Methods Sediment Materials and Characterization. Sediment samples used for titration experiments were obtained from cores recovered during the installation of borehole FWB-103 located VOL. 42, NO. 21, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
8007
adjacent to the S-3 Ponds site (http://www.esd.ornl.gov/ orifrc/; accessed 10 Jan 2008). These materials are composed largely of saprolite derived from interbedded shale, limestone, and sandstone sequences. Two samples were used in these experiments: a less contaminated shallow sediment Sample A from a depth of 8.53-10.67 m with a pH of 3.92, CEC and AEC of 73.0 and 5.3 mmolc/kg, and extractable Fe, Al, and Mn oxides of 135.5, 42.4, and 20.8 mmol/kg, respectively, and a deeper more contaminated Sample D from a depth of 11.58-13.11 m with a pH of 3.82, CEC and AEC of 135.3 and 12.6 mmolc/kg, and extractable Fe, Al, Mn oxides of 170.2, 70.9, and 28.8 mmol/kg, respectively. Differences in CEC and AEC between the two samples can be reasonably attributed to differences in surface area, carbonate content, and metal oxide contents (10). Titration Experiments. Batch titration experiments of contaminated sediments were performed to evaluate the geochemical reaction dynamics under varying pH conditions. The experiments utilized 5 g homogenized soil samples placed in a series of polyethylene vials with 25 mL 0.03 M KCl, to which various amounts (0-1.2 mL) of 2 M NaOH were added to produce a final pH ranging from 3.5 to 10. The total solution volume was made up with deionized water to 30 mL for all samples. The sample vials were placed on an end-over-end shaker for 2 days. An aliquot of suspension was filtered with a 0.2 µm polytetrafluoroethylene (PTFE) syringe filter, and the clear supernatant solution was analyzed for final pH, common anions (e.g., SO42- and NO3-) by ion chromatograph, soluble metals (e.g., Al, Fe, Ca, Mg, Mn, Co, Ni, Sr, and U) and Si preserved in 0.1 M HNO3 by inductively coupled plasma mass spectrometry (19). Based on duplicate analyses on selected samples, measurement errors were estimated to be about 5% for pH and 10% for other measurements, considering possible interferences and dilutions necessary for the analysis of the constituents in a wide concentration range. Modeling Tools. The computer code HydroGeoChem v5.0 (HGC5) (20) is a comprehensive model for fluid flow, thermal and reactive transport. It can handle heterogeneous, fully anisotropic media within three-dimensional domains. The biogeochemical reactive transport program of HGC5 was used as our primary modeling tool. The program is designed for generic biogeochemical reaction networks, which may include both equilibrium and kinetic reactions with user specified formulations. HGC5 was coupled with the nonlinear inversion code PEST (21) to enable calibration of specified model coefficients using measured data. The fits were carried out using the whole set of reactions to calibrate to measured pH, Ca, Mg, Sr, Mn, Ni, Co, sulfate, and U concentrations. Logarithms of concentrations were employed to give equal relative weight to all data points (11). Geochemical Model Description. The geochemical processes governing changes in the solution composition during the soil titration were simulated by an equilibrium reaction path model for both homogeneous and heterogeneous reactions. The homogeneous reactions considered were aqueous complexation reactions (Details are provided in Supporting Information Table S1). The heterogeneous reactions included were precipitation/dissolution reactions (Details are provided in Supporting Information Table S2), pH buffering reactions, cation exchange reactions and anion exchange reactions. The HGC5 program was operated in batch mode to calculate equilibrium distributions of elements between aqueous and solid phases. Iterative calculations were performed to identify controlling precipitation and dissolution reactions and solve for individual species concentrations. First, ion activity products (ICPs) of all simulated minerals were calculated and compared to corresponding solubility values. The most oversaturated phase was then allowed to precipitate and the 8008
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 21, 2008
governing equation (i.e., ICP ) solubility) was included to solve for the concentration of the precipitate together with other species. The computational process was repeated until all saturated phase precipitation limitations are satisfied. To model titration in the presence of mineralogically complex sediments, an approach proposed by Spalding and Spalding (18) was utilized, which treats aquifer solids as a quadraprotic acid designated H4X, where H4X represents aquifer solids and calibrates logK values for ORR soil materials studied (Table 1). Molar concentrations of the H4X were calculated according to measured CEC values and estimated logKX1, logKX2, logKX3, and logKX4 of the buffering reactions (Table 1). Cation exchange reactions involving cationic K, Na, Al, Ca, Mg, Mn, Ni, Co, Sr, and U aqueous species with negatively charged surfaces were considered in the model (Table 1). The total CEC associated with aquifer solids represented by ionized H4X species is computed as CEC ) H3X- + 2H2X2- + 3HX3- + 4X4-. Studies have shown that the ability of soils to adsorb SO42and U is predominantly influenced by pH and the amount of Al and Fe oxyhydroxides (19, 22-29). Similar to the CEC modeling, pH-dependent AEC associated with aquifer solids was represented by ionization of a hypothetical polyprotic base Y(OH)2 (Table 1) and computed as AEC ) YOH+ + 2Y2+. Molar concentrations of Y(OH)2 are calculated according to the measured AEC and estimated logKY1 and logKY2 values (Table 1). A fraction (f) of Al and Fe precipitates generated during titration were assumed to produce surfaces with neutral or positive charge (Table 1). Anion exchange reactions involving SO42-, and anionic U aqueous species were considered in the model (Table 1). For ion-exchange reactions, the Gaines-Thomas (10) selectivity coefficient formulation was used, for example, KCa )
[Ca2+]{Nax ⁄ CEC}2 [Na+]2{2Cax2 ⁄ CEC}
(1)
where [i] denotes an activity (dimensionless) of ion i and exchangeable ions (curly brackets) are expressed as their equivalent fractions of the total ion exchange capacity. While it is known that cation-exchange selectivity coefficients vary significantly among various cations and different type of soils (10, 30), the variations among same groups of monovalent ions such as Na and K or divalent ions such as Ca and Mg are typically small (18). Therefore, cation-exchange selectivity coefficients for the aquifer material in this study were determined by calibrating coefficients to measurements made on the study material, assuming that ion-exchange selectivity coefficients for monovalent alkali metal cations Na+ and K+ are the same (i.e., logKK ) 0 in Table 1), those for divalent alkaline earth metal cations Ca2+, Mg2+, and Sr2+ are the same (i.e., logKCa ) logKMg ) logKSr in Table 1), and those for divalent transition metal cations Mn2+, Ni2+, and Co2+ are the same (i.e., logKMn ) logKNi ) logKCo in Table 1) in order to minimize estimation error due to overparameterization. Although K can be selectively adsorbed by the clay mineral Illite, selective interlayer sites are often Kdominated under natural conditions so that the assumption of nonselectivity should not substantially affect the results. In future studies, direct measurement of Na and K sorption can be undertaken to further assess this issue. We also assumed the initial solution to be metastable with respect to solid minerals. An initial speciation calculation was performed to obtain concentrations of individual aqueous species given the pH and total nitrate, sulfate, Si, Al, Ca, Mg, Mn, Fe, Ni, Co, Sr, and U concentrations of the initial solution. Concentrations of exchangeable ions were then calculated by assuming equilibrium with the solution phase. The model assumed the titration system to be closed to the atmosphere, i.e., we did not assume equilibrium with
TABLE 1. Soil pH Buffering and Ion Exchange Reactions in the Model reactions X-
H+
H4X ) H3 + H3X- ) H2X2- + H+ H2X2- ) HX3- + H+ HX3- ) X4- + H+
parameters
estimate ( std dev.
logKX1 logKX2 logKX3 logKX4
0.00 a -3.27 ( 0.02 -5.59 ( 0.06 -8.12 ( 0.08
cation exchange reactions
parameters
estimate(stddev.
K+
logKK logKCa logKMg logKSr logKMn logKNi logKCo logKAl logKU1
0.0 (18) -1.24 ( 0.08 ) logKCa ) logKCa 1.66 ( 0.09 ) logKMn ) logKMn -0.49 (10) 1.12 ( 0.12
Na+
+ Nax ) Kx + 2Na+ + Cax2 ) 2Nax + Ca2+ 2Na+ + Mgx2 ) 2Nax + Mg2+ 2Na+ + Srx2 ) 2Nax + Sr2+ Mn2+ + Cax2 ) Mnx2 + Ca2+ Ni2+ + Cax2 ) Nix2 + Ca2+ Co2+ + Cax2 ) Cox2 + Ca2+ 2/3 · Al3+ + Cax2 ) 2/3 · Alx3 + Ca2+ UO22+ + Cax2 ) UO2x2 + Ca2+
Al and Fe precipitates related AEC reactions
parameters
estimate(stddev.
YOH+ + H2O ) Y(OH)2 + H+ Y2+ + H2O ) YOH+ + H+ f Ν (Al and Fe precipitates)) Y(OH)2 + YOH+ + Y2+
logKY1 logKY2 F
-14.0 a -3.61 ( 0.12 0.27 ( 0.02
ation exchange reactions y2SO4 + UO2(CO3)2
2-
a
) y2UO2(CO3)2 + SO4
2-
parameters
estimate(stddev.
logKU2
7.94 ( 0.28
Parameter with given value has no effect on simulation and confidence internal cannot be determined.
FIGURE 1. Observed (symbols) and simulated (lines) pH during titration. atmospheric CO2. However, the NaOH solution used for titration was assumed to contain 0.07 mol of total carbonate per mole NaOH (19).
Results and Discussion Titration Curves. Titration of soil samples A and D with NaOH showed broad regions of buffering (Figure 1), reflecting the complex nature of the solution plus aquifer solid system. In order to satisfactorily model the titration curves, logKX1, logKX2, logKX3, and logKX4 of the buffering reactions were adjusted together with other reaction parameters (Table 1). The reaction path model did an adequate job of describing
the titration curves (Figure 1) with r (2) between observation and simulation ) 0.985. When logKX1, logKX2, logKX3, and logKX4 were increased by 0.1 units one at a time with other parameters fixed, pH simulation changed by a maximum of 0.00, 0.40, 0.09, and 0.10 pH units, respectively. The relative concentration of individual species H4X compared to other buffer species (H3X-, H2X2-, HX3-, and X4-) was determined by logKX1. At the given parameter values, the concentration of H4X was small enough to be neglected over the entire pH range indicating that simulated pH is not sensitive to logKX1. Simulation of Solid Phase Minerals. Precipitation of specific mineral phases was computed over the duration of the titration by the geochemical model (Figure 2). In addition to the buffer capacity conferred by the generic soil polyprotic acid, solution pH was also buffered by Al hydrolysis resulting in the formation of aqueous hydrolysis species as well as solid-phase Al(OH)3. The onset of precipitation occurred at progressively higher pH levels for Al2Si2O5(OH)4/Al(OH)3 or Al2Si2O5(OH)4/AlSO4 · 5H2O, Al(OH)3, and FeCo0.1(OH)3.2. For sample A, which reached a final pH slightly greater than 9, MnCO3, Ni(OH)2 and Co(OH)2 were also predicted to precipitate. Although CaCO3, MgCO3 and CoCO3 were included in the model, none of them was predicted to precipitate during titration with NaOH, since only a small amount of carbonate was assumed to be present. Al, Si, and Fe Concentrations. The observed and simulated concentrations of total aqueous Al, Si and Fe during titration with NaOH are presented in Figure 3a. We observe and simulate generally decreasing aqueous Al, Si and Fe concentrations as pH increases. The model predicted that most Al precipitated as Al(OH)3 by pH ∼5.0 (with NaOH addition of ∼7 mmol/L to sample A and ∼12 mmol/L to sample D); more than 70% of Si coprecipitated with Al as Al2Si2O5(OH)4 by pH ∼4 (with the addition of ∼3 mmol/L NaOH); and majority of Fe coprecipitated with Co forming FeCo0.1(OH)3.2 by pH ∼6.0 (with NaOH addition of ∼10 mmol/L to sample A and ∼20 mmol/L to sample D). VOL. 42, NO. 21, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
8009
FIGURE 2. Predicted amounts of mineral precipitation with various quantities of NaOH added to the initially acidic sediment solution. Little Al was left in solution and insignificant amounts of sorbed Al occurred on charged mineral surfaces at higher pH values. The model simulated Al concentration in solution satisfactorily for sample A and sample D. Since precipitation is predicted to be the dominant process controlling Al concentration, the Al ion-exchange reaction parameter KAl (Table 1) could not be accurately estimated from the titration experiment. Therefore, logKAl was specified based on Appelo and Postma (2005) (10). Aqueous Fe concentrations are under-predicted by the model at a few high pH values for Sample A and at a low pH for Sample D. This could be attributed to omitting formation of Fe(III)-organic matter complexes in the model and reflect variability in natural organic matter among the sediment samples. It does not take much soil organic matter to chelate low concentrations of Fe(III) and elevate iron solubility (31). Ca, Mg, and Sr Concentrations. Aqueous Ca, Mg, and Sr concentrations during titration of soil solutions with NaOH are presented in Figure 3b. As solution pH increases with NaOH addition, the soil quadraprotic weak acid H4X is predicted to change from H4X f H3X- f H2X2- f HX3- f X4-, which increases the CEC. Consequently, observed and predicted aqueous concentrations of Ca, Mg, and Sr gradually decrease during the titration process. Similarity of titration curves for Ca, Mg, and Sr reflects the same charge and similar hydration behavior of the alkali earth metals. Ion-exchange affinities of Ca2+, Mg2+, and Sr2+ were thus assumed to be the same (logKCa ) logKMg ) logKSr, Table 1). The r (2) values between observed and simulated Ca, Mg, and Sr concentrations are 0.954, 0.970, and 0.934, respectively. The small deviations are probably due to the use of invariant selectivity coefficients of these ions over the entire pH ranges. When logKCa was increased by 0.1 units, simulated Ca, Mg, and Sr concentrations exhibited maximum changes equal to 2.50, 2.50, and 2.44% of their initial concentrations, respectively. No precipitation of Ca, Mg, and Sr minerals were predicted to occur during titration for either soil sample, thus ion exchange was the dominant process affecting aqueous Ca2+, Mg2+, and Sr2+ concentrations. The model did a good job of 8010
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 21, 2008
describing the cation exchange behavior via the exchange reactions and the pH-dependent CEC associated with the soil quadraprotic acid H4X. Mn, Ni, and Co Concentrations. Aqueous concentrations of Mn, Co, and Ni also decreased with increasing pH as CEC increased (Figure 3c). The titration curves for Mn, Co, and Ni behaved in a similar manner, and ion-exchange affinities for these transition metals were assumed to be the same (logKMn ) logKNi ) logKCo, Table 1). The positive value of logKMn (1.66 ( 0.09, Table 1) for the exchange reaction of Mn2+ + Cax2 ) Mnx2 + Ca2+ indicates that Mn2+, Ni2+, and Co2+ have a greater exchange affinity than Ca2+, Mg2+, and Sr2+. Therefore, we observe earlier decreases of aqueous Mn, Ni, and Co concentrations (Figure 3c) than for Ca, Mg, and Sr (Figure 3b). Titration curves for aqueous Mn2+, Ni2+, and Co2+ were described well by the model. The r (2) values between observed and simulated Mn, Ni, and Co concentrations were 0.953, 0.941, and 0.981, respectively. When logKMn was increased by 0.1 units, simulated Mn, Ni, and Co concentrations exhibited maximum changes of 3.16, 3.37, and 3.09% of their initial concentrations, respectively. Our model indicates that sorption is the primary mechanism up to pH ∼6. At higher pH, Co2+ was predicted to coprecipitate with Fe3+ forming FeCo0.1(OH)3.2. For sample A, when pH exceeded ∼9 (with the addition of >18 mmol/L NaOH), more carbonate could be introduced since the titration was not performed in a closed system, resulting in the precipitation of MnCO3, Ni(OH)2, and Co(OH)2 minerals. Thus, the loss of Mn2+, Ni2+, and Co2+ from solution was first caused by cation exchange only, followed by a combination of cation exchange and precipitation as pH increased. Sulfate and U Concentrations. Concentrations of total aqueous sulfate and U during titration with NaOH are presented in Figure 3d. Model predictions agreed well with observed SO42- concentrations with r2 ) 0.956. Simulated sulfate concentrations strongly depend on pH and AEC due to aluminum precipitation reactions that form additional anion exchange sites. Fe precipitation had little effect on AEC because of its relatively low concentration as compared with that of Al. As solution pH increased with the addition of NaOH, the hypothetical base Y(OH)2 is predicted to change from Y2+ f Y(OH)+ f Y(OH)2, resulting in decreased AEC and desorption of previously sorbed sulfate. However, the general trend of decreasing AEC as pH increases may be interrupted by increasing AEC associated with newly precipitated Al(OH)3. Because uranium (U) is one of the major contaminants of concern at the site, its behavior in the experiments and accompanying simulations was of particular interest. The oxidized forms of U(VI) are predominant in the soil and groundwater at the site (32). Since all experiments were performed under oxic conditions, all U present was expected to be in the hexavalent oxidation state. Most U(VI) was removed from solution by pH ∼5 with the addition of ∼7 mmol/L NaOH to sample A and ∼12 mmol/L to sample D (Figure 3d). These observations were described well by the model (Figure 4). In the low pH region, U(VI) loss from solution was predicted due to cation exchange involving the cationic U species represented by UO22+ forming UO2x2 (Table 1). As pH increases, the formation of anionic uranyl-carbonate species increases. Sorption of uranyl-carbonate was represented by UO2(CO3)22- forming y2UO2(CO3)2 (Table 1). The simulation agreed well with the observed aqueous U concentrations with r (2) ) 0.988 (Figure 3d). When logKU1 and logKU2 were each increased by 0.1 units, the resulting maximum changes in simulated U concentration were 2.16 and 2.14% of the initial U concentration, respectively. Dissolved carbonate was not measured in experiments during this study. Simulations assumed that the NaOH
FIGURE 3. Observed (symbols) and simulated (lines) aqueous (a) Al, Si and Fe; (b) Ca, Mg and Sr; (c) Mn, Ni and Co; (d) Sulfate and U concentrations during titration. titration solution contains 0.07 mol of total carbonate per moleNaOH(19).Sincecarbonateaffectsanionicuranyl-carbonate sorption, uncertainty in the amount of carbonate may introduce uncertainty in the estimation of U sorption parameters. It is therefore suggested that future experiments include measurements of carbonates or be performed under controlled CO2 conditions. Concluding Remarks. This study demonstrates the importance of solid phase protonation-deprotonation reactions in buffering pH, ion exchange, and precipitation or coprecipitation of metals in sediment materials. For complex mineral assemblages, treating the heterogeneous solid phase as insoluble polyprotic acids and bases facilitates simulta-
neously modeling solid-phase buffer capacity and pHdependent CEC and AEC. Incorporating effects of Al and Fe mineral precipitation and dissolution in the proposed model allows dynamic changes in the quantity of charged surfaces not accounted for by the acid-base formulations. Combining the solid phase polyprotic acid and base model with a geochemical reaction network that includes precipitation/ dissolution reactions, ion exchange reactions, and solution complexation reactions has been proven a practical and accurate means to simulate the titration behavior as well as distributions of various metal species between and within aqueous and solid phases. This proposed modeling approach could potentially provide an effective means to simulate VOL. 42, NO. 21, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
8011
FIGURE 4. Predicted sorbed and precipitated U concentrations during titration.
effects of soil pH manipulations on the mobility of inorganic contaminants in subsurface systems. Work is currently in progress to incorporate the geochemical model in dynamic transport simulations to design and analyze field experiments and ultimately to assess field strategies to control and attenuate groundwater contaminants.
Acknowledgments This research was funded by the U.S. Department of Energy, Office of Science, Biological and Environmental Research Programs. Oak Ridge National Laboratory is managed by UT- Battelle, LLC, for the U.S. Department of Energy under Contract DE-AC05-00OR22725.
Supporting Information Available Supporting Information is free of charge via the Internet at http://pubs.acs.org.
Literature Cited (1) Whicker, F. W.; Hinton, T. G.; MacDonell, M. M.; Pinder, J. E.; Habegger, L. J. Avoiding destructive remediation at DOE sites. Science 2004, 303, 1615–1616. (2) Baes, C. F.; Mesmer, R. E. The Hydrolysis of Cations; John Wiley and Sons: 1976, 489 pp. (3) Bibak, A.; Moberg, J. P.; Borggaard, O. K. Cobalt retention by Danish spodosol samples in relation to contents of organicmatter and aluminum, iron and manganese oxides. Acta Agric. Scand., Sect. B 1995, 45, 0153-0158. (4) Towle, S. N.; Bargar, J. R.; Brown, G. E.; Parks, G. A. Surface precipitation of Co(II) (aq) on Al2O3. J. Colloid Interface Sci. 1997, 187, 62–82. (5) Jara, A. A.; Goldberg, S.; Moraa, M. L. Studies of the surface charge of amorphous aluminosilicates using surface complexation models. J. Colloid Interface Sci. 2005, 292, 160–170. (6) Pivovarov, S. Adsorption of ions onto amorphous silica: Ion exchange model. J. Colloid Interface Sci. 2008, 319, 374–376. (7) Fletcher, P.; Sposito, G. The Chemical Modeling of Clay/ Electrolyte interactions for Montmorillonite. Clay Miner. 1989, 24, 375–391. (8) Duc, M.; Thomas, F.; Gaboriaud, F. Coupled chemical processes at clay/electrolyte interface: A batch titration study of Na8012
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 21, 2008
montmorillonites. J. Colloid Interface Sci. 2006, 300, 616–625. (9) Tonkin, J. W.; Balistrieri, L. S.; Murray, J. W. Modeling sorption of divalent metal cations on hydrous manganese oxide using the diffuse double layer model. Appl. Geochem. 2004, 19, 29–53. (10) Appelo, C. A. J.; Postma, D. Geochemistry, Groundwater and Pollution, 2nd ed.; A.A. Balkema Publishers: Leiden, The Netherlands: 2005. (11) Steefel, C. I.; Carroll, S.; Zhao, P.; Roberts, S. Cesium migration in Hanford sediment: a multisite cation exchange model based on laboratory transport experiments. J. Contam. Hydrol. 2003, 67, 219–246. (12) Jacquier, P.; Ly, J.; Beaucaire, C; The. ion-exchange properties of the Tournemire argillite I. Study of the H, Na, K, Cs, Ca and Mg behaviour. Appl. Clay Sci. 2004, 26, 163–170. (13) Vulava, V. M.; Kretzschmar, R.; Barmettler, K.; Voegelin, A.; Grolimund, D.; Borkovec, M. Cation competition in a natural subsurface material: Prediction of transport behavior. Water Resour. Res 2002, 38, 5. (14) Ludwig, B.; Khanna, P. K.; Anurugsa, B.; Folster, H. Assessment of cation and anion exchange and pH buffering in an Amazonian Ultisol. Geoderma 2001, 102, 27–40. (15) Cances, B.; Ponthieu, M.; M; Castrec-Rouelle, Aubry, E.; Benedetti, M. F. Metal ions speciation in a soil and its solution: experimental data and model results. Geoderma 2003, 113, 341– 355. (16) Ferrell, R. E.; Aagaard, P.; Forsman, J.; Greenwood, L.; Zheng, Z. Application of a geochemical transport model to predict heavy metal retention (Pb) by clay liners. Appl. Clay Sci. 2002, 21, 59–66. (17) Liu, C.; Zachara, J. M.; Smith, S. C. A cation exchange model to describe Cs+ sorption at high ionic strength in subsurface sediments at Hanford site, USA. J. Contam. Hydrol. 2004, 68, 217–238. (18) Spalding, B.; Spalding, I. R. Chemical equilibria model of Strontium-90 adsorption and transport in soil in response to dynamic alkaline conditions. Environ. Sci. Technol. 2001, 35, 365–373. (19) Gu, B.; Brooks, S. C.; Roh, Y.; Jardine, P. M. Geochemical reactions and dynamics during titration of a contaminated groundwater with high uranium, aluminum, and calcium. Geochi. Cosmochim. Acta 2003, 67, 2749–2761. (20) Yeh, G.-T.; Sun, J. T.; Jardine, P. M.; Burgos, W. D.; Fang, Y. L.; Li, M. H.; Siegel, M. D. HydroGeoChem 5.0: a Three-Dimensional Model of Coupled Fluid Flow, Thermal Transport, And Hydrogeochemical Transport through Variable Saturated Conditions, Version 5.0, 2004. (21) Doherty, J. PEST: Model Independent Parameter Estimation; Watermark Numerical Computing: Brisbane, Australia, 2004. (22) Bazer-Bachi, F.; Descostes, M.; Tevissen, E.; Meier, P.; Grenut, B.; Simonnot, M.-O.; Sardin, M. Characterization of sulphate sorption on Callovo-Oxfordian argillites by batch, column and through-diffusion experiments. Phys. Chem. Earth 2007, 32, 552– 558. (23) Arbestain, M. C.; Barreal, M. E.; Macias, F. Relating sulfate sorption in forest soils to lithological classes, as defined to calculate critical loads of acidity. Sci. Total Environ. 1999, 241, 181–195. (24) Arbestain, M. C.; Barreal, M. E.; Macias, F. Sulfate sorption in nonvolcanic andisols and andic soils from Galicia, NW Spain. Geoderma 2001, 104, 75–93. (25) King, H. B.; Wang, M. K.; Zhuang, S. Y.; Hwong, J.-L.; Liu, C. P.; Kang, M. J. Sorption of sulfate and retention of cations in forest soils of Lien-Hua-Chi watershed in central Taiwan. Geoderma 2006, 131, 143–153. (26) Martinson, L.; Alveteg, M.; Warfvinge, P. Parameterization and evaluation of sulfate adsorption in a dynamic soil chemistry model. Environ. Pollut. 2003, 124, 119–125. (27) GOBRAN, G. R.; SELIM, H. M.; HULTBERG, H.; ANDERSSON, I. Sulfate adsorption-desorption in a Swedish forest soil. Water, Air, Soil Pollut. 1998, 108, 411–424. (28) DAVIS, J. A.; MEECE, D. E.; KOHLER, M.; CURTIS, G. P. Approaches to surface complexation modeling of Uranium(VI) adsorption on aquifer sediments. Geochim. Cosmochim. Acta 2004, 68 (18), 3621–3641. (29) Tao, Z. Y.; Chu, T. W.; Du, J. Z.; Dai, X. X.; Gu, Y. J. Effect of fulvic acids on sorption of U(VI), Zn, Yb, I, and Se(IV) onto oxides of aluminum, iron, and silicon. Appl. Geochem. 2000, 15, 145–151. (30) Neal, C.; Robsona, A. J. Integrating variations of soil-adsorbed cations into a cation exchange model. Science. Total Environ. 1997, 199, 277–292. (31) Green-Pedersen, H.; Pind, N. Preparation, characterization, and sorption properties for Ni(II) of iron oxyhydroxide-montmo-
rillonite. Colloids Surf., A 2000, 168, 133–145. (32) Gu, B.; Watson, D. B.; Phillips, D. H.; Liang, L., Biogeochemical, Mineralogical, And Hydrological Characteristics of an Iron Reactive Barrier Used for Treatment of Uranium and Nitrate. In Groundwater Remediation of Trace Metals, Radionuclides,
and Nutrients, With Permeable Reactive Barriers; Naftz, D. L., Morrison, S. J., Davis, J. A., Fuller, C. C., Eds.; Academic Press: New York2002; pp 305-342.
ES800311M
VOL. 42, NO. 21, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
8013
