Environ. Sci. Technol. 2003, 37, 3619-3624
Carbonate Effects on Hexavalent Uranium Adsorption by Iron Oxyhydroxide MAHMOUD WAZNE, GEORGE P. KORFIATIS, AND XIAOGUANG MENG* Center for Environmental Systems, Stevens Institute of Technology, Hoboken, New Jersey 07030
Carbonate dramatically affects the adsorption of uranium (U(VI)) onto iron hydroxides and its mobility in the natural environment. Batch tests, zeta potential measurements, and Fourier transform infrared (FTIR) spectroscopic studies were utilized to characterize the nature of U(VI) adsorption on ferrihydrite. Adsorption isotherms demonstrated that carbonate had a negative effect on U(VI) adsorption on ferrihydrite at pH >6. Zeta potential measurements indicated that U(VI) was adsorbed as a cationic species (SOUO2+) in the absence of carbonate and as anionic U(VI) complexes in the presence of carbonate at neutral pH. FTIR spectroscopic measurement of adsorbed U(VI) suggested that it was retained as uranyl carbonate complexes in the presence of carbonate. An increase in carbonate concentration caused a shift in the antisymmetric stretching vibration of the uranyl (UO22+) U-O bond toward lower wavenumbers, which indicated an increasing carbonate effect in the adsorbed uranyl carbonate complexes. The adsorbed U(VI) species were successfully incorporated into a surface complexation model to describe the adsorption of U(VI) by ferrihydrite from artificial solutions and contaminated water.
Introduction Use of uranium at industrial and military sites has resulted in many cases of uranium contamination. The mobility of uranium is controlled by its interaction with ions and minerals present in nature. Carbonate forms strong complexes with U(VI), thus increasing the solubility of uranium (1). Iron hydroxides, which are ubiquitous in soils, act as natural retardants to uranium migration. The concentration of dissolved carbonate species in acidic aqueous environments is approximately 100-1000 µM, and at pH >7.0, it increases 1 order of magnitude for every one pH unit increase (2). It is of great importance to know uranium interactions with carbonate and iron hydroxides. Although adsorption of U(VI) on iron hydroxides has been studied extensively, many questions remain about the nature of the U(VI) species adsorbed at the iron hydroxide surface. Bargar et al. (3) studied the U(VI) species adsorbed on hematite using zeta potential measurements. They observed that in the absence of U(VI) the hematite surface was positively charged when pH was below 8.0. The addition of U(VI) caused a reversal in zeta potential between pH 6.5 and 8.0. The magnitude of surface charge reversal increased with * Corresponding author telephone: 201-216-8014; fax: 201-2168303; e-mail:
[email protected]. 10.1021/es034166m CCC: $25.00 Published on Web 07/23/2003
2003 American Chemical Society
increasing total U(VI) concentration. The charge reversal was attributed to the formation of anionic ternary uranyl carbonate surface complexes. Ho and Miller (4) assumed that the decreased surface potentials were due to the adsorption of uranyl hemicarbonate at pH 7.6. In an earlier study, Ho and Doren (5) speculated that the shift in the isoelectric point was due to (UO2)3(OH)5+ adsorption. Fourier transform infrared (FTIR) spectroscopy has been used to investigate the U(VI) species adsorbed on hematite (4). The adsorption of U(VI) on hematite surfaces resulted in IR peaks at 910 and 903 cm-1 in the absence and presence of carbonate, respectively. Bargar et al. (6) focused on the antisymmetric stretching vibration ν3 carbonate peak in the 1400-1600-cm-1 region to infer information about U(VI) adsorption on hematite. They concluded that the adsorbed species consisted of bidentate coordination of carbonate anions to U(VI). Bargar et al. (3) used extended X-ray absorption fine structure (EXAFS) to determine how U(VI) was coordinated to the hematite surface. They concluded that at pH 8.0. The EXAFS results suggested the presence of anionic ternary surface U(VI)-bicarbonate complexes that increased in relative concentration with increasing pH. Furthermore, it was determined that U(VI) was attached in a bidentate or tridentate fashion to the hematite surface. Waite et al. (7) used EXAFS to study U(VI) adsorption to the ferrihydrite surface. They conducted their EXAFS experiments at pH 5.0 and 5.5, and they concluded that at pH 5.0 (Figure 4). Aqueous reactions in Table 1 were used to construct the speciation diagrams in Figures 3 and 4. Since the major aqueous species in the absence of carbonate are uranium(VI) hydroxide complexes at pH >5.0, it is frequently assumed these hydroxides complexes are adsorbed on the surface (10, 12). However, the adsorption of all possible uranium(VI) hydroxide complexes will result in the formation of anionic or neutral surface species, which contradicts the increased zeta potential caused by U(VI) adsorption (Figure 5). The only possible cationic surface species is monodentate uranyl, S-O-UO2+, where S-O
FIGURE 5. Possible coordination of U(VI) to ferrihydrite surface. denotes a ferrihydrite surface site. Based on the zeta potential results, the U(VI) adsorption is written as
SOH + UO22+ T SO-UO2+ + H+ In the presence of carbonate, U(VI) may form monodentate, bidentate, and tridentate coordination of uranyl carbonate complexes (Figure 5). All these complexes will cause a decrease in the zeta potential. FTIR Spectroscopic Studies. The surface species of U(VI) on the ferrihydrite were further investigated using FTIR. Uranyl ion UO22+ exhibits three fundamental vibrational modes: symmetric stretching vibration υ1, bending vibration υ2, and antisymmetric stretching vibration υ3. The υ1 UO 22+ VOL. 37, NO. 16, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3621
FIGURE 6. FTIR spectra of U(VI) adsorbed on ferrihydrite surface at pH 6 and [Fe(III)] ) 1.78 × 10-4 M. (A) CT ) 0 M, [U(VI)] ) 0 M. (B) CT ) 1.26 × 10-3 M, [U(VI)] ) 0 M. (C) CT ) 0 M, [U(VI)] ) 4.2 × 10-6 M. (D) CT) 4.2 × 10-5 M, [U(VI)] ) 4.2 × 10-6 M. (E) CT ) 4.2 × 10-4 M, [U(VI)] ) 4.2 × 10-6 M. (F) CT ) 1.26 × 10-3 M, [U(VI)] ) 4.2 × 10-6 M. exhibits in the region 900-750 cm-1 and is Raman active, whereas, UO 22+ υ1 peak appears in the IR spectrum only in case of substantial symmetry lowering. The antisymmetric stretching vibration, υ3 UO22+ (1000-850 cm-1), is active in the infrared and inactive in Raman. υ2 UO 22+ (300-200 cm-1) is IR active (17, 18). Cejka (17) noted that the intensity of the antisymmetric stretching vibration is usually strong or very strong. Ferrihydrite had no absorption peaks in the 900-cm-1 region as shown in Figure 6. Upon loading the ferrihydrite surface with U(VI) in the absence of carbonate, a clear peak appeared at 902 cm-1. Similar U(VI) peaks were observed by Ho and Miller (4) and Duff et al. (19). When both carbonate and uranium were introduced to the system, the U(VI) peak shifted toward lower wavenumbers. The ionic strength and the pH were not changed for FTIR experiments. As the carbonate concentration was increased from 0 to 4.2 × 10-5, 4.2 × 10-4, and 1.26 × 10-3 M, the U(VI) peak shifted from 902 to 89, 888, and 880 cm-1, respectively. Moreover, the uranium peak broadened toward the lower energy side as the carbonate concentration was increased. Many researchers have confirmed the effect of the carbonate ligand on the UO22+ stretching vibration peak shift toward lower wavenumbers (20-25). The shift of the UO22+ stretching vibration peak toward lower wavenumbers increased as the number of carbonate ligands increased (21, 24). Ho and Miller (4) discerned a U(VI) peak at 910 cm-1 in the case of hematite loaded with U(VI) with no carbonate present and a peak at 903 cm-1 when carbonate was introduced to the system. The spectra in Figure 6 indicated that more carbonate ligands were attached to the adsorbed uranyl ion as the total carbonate concentration increased. The ligand effect of carbonate on the UO22+ stretching vibration and the FTIR experimental results indicated that U(VI) adsorbed as uranyl carbonate complexes on the ferrihydrite surface in the presence of carbonate. Even though carbonate has an out-of-plane ν2 C-O bending vibration that exhibits in a region between 880 and 835 cm-1 (17, 26), no clear peak could be assigned to this vibration when ferrihydrite was treated with 1.26 × 10-3 M total carbonate solution (spectrum B in Figure 6). The spectra of blank and carbonate-treated ferrihydrite samples were similar. The absence of carbonate peaks could be attributed to the low intensity of the ν2 C-O bending vibration of the adsorbed carbonate. Surface Complexation Modeling. The diffuse doublelayer surface complexation model (27, 28) was used to describe U(VI) adsorption data in the present work. The equilibrium constants for acid-base surface reactions (eqs 3622
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 16, 2003
18 and 19 in Table 1) are taken from Waite et al. (7). Carbonate adsorption was also incorporated into the model (eqs 20 and 21 in Table 1) since carbonate was proven to adsorb onto iron hydroxides. One or both of eqs 20 and 21 were used by researchers (2, 7, 15, 29) to model carbonate adsorption onto iron hydroxides. The mechanisms involve the exchange of the surface OH functional groups for carbonate or bicarbonate anions where the carbonate or bicarbonate anions are attached to the surface in an inner-sphere fashion. The adsorption constants (eqs 20 and 21 in Table 1) were generated by fitting the double-layer model to carbonate adsorption data reported in the literature (7). The simplest reactions based on our experimental observations of the zeta potential and FTIR experimental results were considered in modeling: monodentate adsorption of uranyl ion, uranyl monocarbonate complex, and uranyl dicarbonate complex in addition to carbonate adsorption (eqs 18-24), and MENTEQA2 (16) extended database aqueous reactions (Table 1) with one kind of adsorption sites. The use of a second uranyl carbonate complex in the model (eq 24) made a significant improvement. Adsorption of the second uranyl carbonate species, uranyl dicarbonate, was justified based on the FTIR data. As the total carbonate concentration increased, the uranyl U-O antisymmetric stretching peak shifted to the right toward lower wavenumbers while simultaneously broadening at the right side, signifying that more carbonate ligands are attached to the adsorbed uranyl ion as the total carbonate concentration increased. The surface sites SO- and the carbonate anions act as ligands in Equations 22, 23 and 24. The surface ligand is attached to the uranyl carbonate moiety equatorial plane in a monodentate fashion whereas the carbonate anion is coordinated to the equatorial plane of the uranyl ion in a bidentate fashion through the two singly bonded oxygens of the carbonate anion. The axial double-bonded oxygens of the uranyl cation are not involved in the coordination to the surface. Burns (30) observed that, in mineral structures, U(VI) is almost always present in an approximately linear uranyl ion UO22+. The uranyl ion is coordinated by four, five, or six equatorial ligands in a planar arrangement perpendicular to axis of the uranyl ion. The equatorial ligands could belong to O2-, OH-, or H2O. The two axial oxygens form the apexes of bipyramids while the four to six equatorial ligands occupy the equatorial positions of the bipyramids. Cejka (17) noted that the equatorial oxygens could belong to various oxyanions including hydroxyl ions, carbonates, sulfates, silicates, phosphates, and arsenates. The number of oxygens in the equatorial plane of the uranyl cation in case of SO- UO2(CO3)23- species is five, which is smaller than the maximum number of ligands observed by Burns (30) and Cejka (17). Moreover, Bargar et al. (3) postulated six equatorial oxygens where two of them were attached to the hematite surface and the rest belonged to carbonate or hydroxyl anions. Similarly, Waite et al. (7) proposed five equatorial oxygens with two oxygens attached to the ferrihydrite surface by edge-sharing bond with an iron octahedron. The set of eqs 22-24 in Table 1 was used to describe U(VI) coordination to ferrihydrite surface, and it constitutes an original contribution in this work. This set is based on direct experimental evidence using zeta potential measurements and FTIR spectroscopic studies. The experimental data in Figure 7 were used to calibrate the model (the actual ionic strength was computed automatically in all model calculations). The adsorption data were generated by mixing an artificial solution consisting of DI water that had uranium concentration of 4.2 × 10-6 M, total carbonate concentration of 0.01 M, and sodium chloride
FIGURE 7. Model fit of U(VI) adsorption data on 1.42 × 10-3 M iron(III) ferrihydrite in closed system. 0.01 M total carbonate, 4.2 × 10-6 M U(VI), and 0.01 M NaCl.
FIGURE 8. Model prediction of U(VI) adsorption on 1.42 × 10-3 M iron(III) ferrihydrite in closed system. 1 × 10-5 M total carbonate, 4.2 × 10-6 M U(VI), and 0.01 M NaCl. concentration of 0.01 M with 1.42 × 10-3 M iron(III) ferrihydrite suspension. U(VI) removal peaked when pH was between 4.5 and 7.0. The removal decreased dramatically when pH was lower than 4.5 and greater than 7.0. The goodness of the model was quantified by the rootmean-square of error (RMSE):
RMSE )
[
1
nd - np
nd
∑ i)1
]
1/2
(% Rem - % Rem∧)
where nd is the number of data points, np is the number of adjustable parameters, i is an index, Rem is the measured U(VI) removal, and Rem∧ is the predicted U(VI) removal by the model. All combinations of eqs 22-24 were considered to optimize fitting the data in Figure 7. The RMSE of different combinations were calculated and compared. The combination with the smallest RMSE was selected. The model (Figure 7) was able to predict both of the adsorption edges at approximately pH 4.0 and 8.0 and the peak adsorbing area between pH 4.5 and 7.0, lagging onethird of a pH unit at the first adsorption edge. The model had a RMSE of 0.133 94. The model was tested on a set of data produced by mixing an artificial solution consisting of DI water that had uranium concentration of 4.2 × 10-6 M, total carbonate concentration of 1 × 10-5 M, and sodium chloride concentration of 0.01 M with 1.42 × 10-3 M iron(III) ferrihydrite suspension. The model was successful in predicting the adsorption curve of the experimental data, as shown in Figure 8, with a RMSE of 0.029 828. A significant difference between the experi-
FIGURE 9. Model prediction of U(VI) adsorption from contaminated water samples collected at Ford Farm site. 1.42 × 10-3 M iron(III) ferrihydrite, [U(VI)] ) 2.94 × 10-6, total alkalinity 11.25 mg/L as CaCO3, and closed system adsorption
FIGURE 10. Model prediction of U(VI) adsorption from contaminated water collected at BTD site. 1.42 × 10-3 M iron(III) ferrihydrite, [U(VI)] ) 4.2 × 10-5, and total alkalinity 177 mg/ L as CaCO3, closed system data. mental data in Figures 7 and 8 was the complete removal of U(VI) at pH >7.0 at low total carbonate concentration. The steep decrease in U(VI) removal at pH >7.0 (Figure 7) was caused by the presence of 0.01 M total carbonate. The similar adsorption edges at approximately pH 4.0 in Figures 7 and 8 indicated that carbonate had no negative effect on U(VI) adsorption at low pH. The adsorption results and modeling predictions were presented in Figures 9 and 10 for the contaminated water from the two DOD sites, Ford Farm and BTD, respectively. While complete U(VI) removal was obtained for the Ford Farm water in a pH range between 5.0 and 9.0 (Figure 9), U(VI) adsorption decreased substantially in the BTD water at pH >7.0 (Figure 10). Since the natural pH of the contaminated water were about 7.0 and 8.7 for Ford Farm and BTD, respectively, it would be more difficult to remove U(VI) from BTD water than from the Ford Farm water by iron hydroxides. The high mobility of U(VI) in the BTD water was attributed to its higher total carbonate content (total alkalinity 177 mg/L as CaCO3) than in the Ford Farm water (total alkalinity 11.2 mg/L as CaCO3). Despite the difference in chemical compositions between the two water samples, the model was able to predict the U(VI) uptake at both of the DOD sites as shown in Figure 9 and Figure 10, respectively. The RMSE of the model prediction of BTD and Ford Farm sites are 0.332 343 and 0.094 082, respectively. The total alkalinity values were converted to total carbonate contents and were used in the model calculations. VOL. 37, NO. 16, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3623
The model prediction of the adsorbed U(VI) species was also plotted versus pH in Figures 9 and 10. The main U(VI) species on the surface was SO-UO2+ in the Ford Farm system (Figure 9). As total carbonate concentration increased, the percentage of the adsorbed uranyl carbonate surface species increased. When pH was between 5.0 and 7.0, SO-UO2(CO3)23- was the predominant surface species in the BTD system due to the high content of total carbonate (Figure 10). There are three pH regions where carbonate exhibits different effects on U(VI) adsorption. The first pH region is at pH 8.0. At pH 8.0, carbonate does not adsorb to the iron hydroxides (2, 15) and thus does not compete for the adsorption sites but rather competes with the surface to complex U(VI) as aqueous uranyl tricarbonate. At pH >8.0 and in the presence of carbonate, uranium(VI) tricarbonate is the predominant aqueous complex (Figure 4) which prevents the adsorption of U(VI). Implications of the Research. The use of zeta potential measurements provided information which led to the conclusion that SO-UO2+ is the adsorbed U(VI) species in the absence of carbonate. In the presence of carbonate, the FTIR measurements were able to trace the increased number of carbonate ligands in the adsorbed uranyl carbonate complexes as total carbonate concentration increased. Zeta potential and FTIR spectral data were used to reach a refined set of inputs for the diffuse double-layer surface complexation model (eqs 22-24). The resulting model was able to predict U(VI) adsorption on iron hydroxides for U(VI) solutions spanning 4 orders of magnitude in total carbonate concentrations over a wide range of pH values. Carbonate was shown to inhibit U(VI) adsorption on ferrihydrite by two different mechanisms, depending on the pH range: at intermediate pH, by forming uranyl carbonate complexes that have lower affinity to adsorb to the iron hydroxide surfaces and by competing for surface sites, while complexing and solubilizing U(VI) in the highest pH range. This research provides new insights into and tools for predicting the migration of uranium in the natural environment, an increasingly important problem in view of the increased use of uranium and depleted uranium in military and industrial applications.
Acknowledgments This work was supported by the DOD. We thank Dr. Mike Dadachov for his help in the FTIR work and the DOD employees at Picatinny Arsenal and Aberdeen Proving Grounds for their help providing facilities and assistance. We also thank Christine Chin Choy, Sunbaek Bang, and Dr. Chuanyong Jing for their thoughtful discussions and help.
3624
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 16, 2003
Literature Cited (1) Grenthe, I.; Fuger, J.; Konings, R. J. M.; Lemire, R. J.; Muller, A. B.; Nguyen-Trung Gregu, C.; Wanner, H. Chemical Thermodynamics of Uranium; Elsevier: New York, 1992. (2) Van Geen, A.; Robertson, A. P.; Leckie, J. O. Geochim. Cosmochim. Acta 1994, 58 (9), 2073-2086. (3) Bargar, J. R.; Reitmeyer, R.; Lenhar, J. J.; Davis, J. A. Geochim. Cosmochim. Acta 2000, 64 (16), 2737-2749. (4) Ho, C. H.; Miller, N. H. J. Colloid Interface Sci. 1986, 110 (1), 165-175. (5) Ho, C. H.; Doren, D. C. Can. J. Chem. 1985, 63, 1100. (6) Bargar, J. R.; Reitmeyer, R.; Davis, J. A. Environ. Sci. Technol. 1999, 33 (14) 2481-2484. (7) Waite, T. D.; Davis, J. A.; Payne, T. E.; Waychunas, G, A.; Xu, N. Geochim. Cosmochim. Acta 1994, 58 (24), 5465-5478. (8) Manceau, A.; Boisset, M. C.; Didier, B.; Spadini, L. Appl. Clay Sci. 1992, 7, 201-230. (9) Dzombak, D. A.; Morel, F. M. M. Surface Complexation Modeling: Hydrous Ferric Oxide; Wiley: New York, 1990. (10) Hsi, C. D.; Langmuir, D. Geochim. Cosmochim. Acta 1985, 49, 1931-1941. (11) Davis, J. A.; Leckie, J. O J. Colloid Interface Sci. 1978, 67, 90-107. (12) Tripathi, V. S. Uranium transport modeling: geochemical data and sub-models. Ph.D. dissertation, Stanford University. 1983. (13) Raven, K. P.; Jain, A.; Loeppert, R. H. Environ. Sci. Technol. 1998, 32, 344-349. (14) Cornell, R. M.; Schwertmann, U. The Iron Oxides: Structure, Properties, Reactions, Occuurrences and Uses; VCH: Weinheim, 1996. (15) Zacahra, J. M.; Girvin, D. C.; Schmidt, R. L.; Resch, C. T. Environ. Sci. Technol. 1987, 22, 589-594. (16) David, S. B.; Allison, J. D. Minteqa2, An Equilibrium Metal Speciation Model: User’s Manual 4.01; Environmental Research laboratory, U.S. Environmental Protection Agency: Athens, GA, 1999. (17) Cejka, J. In Uranium: Mineralogy, Geochemistry and the Environment; Burns, P. C., Finch, R., Eds.; Reviews in Mineralogy 38; Mineralogical Society of America: Washington, DC, 1999. (18) Hoekstra, H. R. Vibrational Spectra. In Gmelin’s Handbook of Inorganic Chemistry. Uranium; Suppl. Vol. A5. Spectra; VerlagChemie: Weinheim, Germany, 1982; Chapter 5, pp 211-240. (19) Duff, M. C.; Coughlin, J. U.; Hunter, D. B. Geochim. Cosmochim. Acta 2002, 66 (20), 3533-3547. (20) McGlynn, S. P.; Smith, J. K.; Neely, W. C. J. Chem. Phys. 1961, 35, 105. (21) Maya, L.; Begun, G. M. Inorg. Nucl. Chem. 1981, 43 (11), 28272832. (22) Vdovenko V. M., Suglobov D. N.; Krasilnikov, V. A. Radiokhimiya 5 1963, 281/8; Sov. Radiochem. 1963, 5, 281/8. (23) Venanzi, L. M.; Day, J. P. J Chem. Soc. A 1966, 1363/7. (24) Koglin, E.; Schnek, H. J.; Schowchau, k. Spectrochim. Acta 1979, 35A. 641-647. (25) Gorbenko-Germanove, D. S.; Zenkova, R. A. Opt. Spectroskopiya 20 1966, 842/7; Opt. Spectry. [USSR] 1966, 20, 467/9. (26) Smith, B. Infrared Spectral Interpretation. A Systematic Approach; CRC Press: Boca Raton, FL,1999. (27) Stumm, W.; Huang, C. P.; Jenkins, S. R. Croat. Chem. Acta 1970, 42, 223-244. (28) Huang, C. P.; Stumm, W. J. Colloid Interface Sci. 1973, 22, 231259. (29) Villalobos, M.; Leckie, O. J. Colloid Interface Sci. 2001, 235, 1532. (30) Burns, P. C. In Uranium: Mineralogy, Geochemistry and the Environment; Burns, P. C., Finch, R., Eds.; Reviews in Mineralogy 38; Mineralogical Society of America: Washington, DC, 1999. (31) Duff, M. C.; Amrhein, C. Soil Sci. Am. J. 1996, 60, 1393-1400. (32) Payne, T. E.; Waite, T. D. Radiochim. Acta 1991, 52/53, 487-493.
Received for review February 24, 2003. Revised manuscript received May 28, 2003. Accepted June 5, 2003. ES034166M
