Environ. Sci. Technol. 2004, 38, 1399-1407
Use of Spectroscopic Techniques for Uranium(VI)/Montmorillonite Interaction Modeling A . K O W A L - F O U C H A R D , † R . D R O T , * ,† E. SIMONI,† AND J. J. EHRHARDT‡ Institut de Physique Nucle´aire, Groupe de Radiochimie, Universite´ Paris XI, 91406 Orsay, France, and Laboratoire de Chimie Physique et Microbiologie pour l’Environnement, UMR-7564 CNRS-Universite´ Henri Poincare´ Nancy I, 405 Rue de Vandoeuvre, 54600 Villers-le`s-Nancy, France
To experimentally identify both clay sorption sites and sorption equilibria and to understand the retention mechanisms at a molecular level, we have characterized the structure of hexavalent uranium surface complexes resulting from the interaction between the uranyl ions and the surface retention groups of a montmorillonite clay. We have performed laser-induced fluorescence spectroscopy (LIFS) and X-ray photoelectron spectroscopy (XPS) on uranyl ion loaded montmorillonite. These structural results were then compared to those obtained from the study of uranyl ions sorbed onto an alumina and also from U(VI) sorbed on an amorphous silica. This experimental approach allowed for a clear determination of the reactive surface sites of montmorillonite for U(VI) sorption. The lifetime values and the U4f XPS spectra of uranium(VI) sorbed on montmorillonite have shown that this ion is sorbed on both exchange and edge sites. The comparison of U(VI)/ clay and U(VI)/oxide systems has determined that the interaction between uranyl ions and montmorillonite edge sites occurs via both ≡AlOH and ≡SiOH surface groups and involves three distinct surface complexes. The surface complexation modeling of the U(VI)/montmorillonite sorption edges was determined using the constant capacitance model and the above experimental constraints. The following equilibria were found to account for the uranyl sorption mechanisms onto montmorillonite for metal concentrations ranged from 10-6 to 10-3 M and two ionic strengths (0.1 and 0.5 M): 2≡XNa + UO22+ S (≡X)2UO2 + 2Na+, log K0exch ) 3.0; ≡Al(OH)2 + UO22+ S ≡Al(OH)2UO22+, log K0Al ) 14.9; ≡Si(OH)2 + UO22+ S ≡SiO2UO2 + 2H+ , log K0Si1 ) -3.8; and ≡Si(OH)2 + 3UO22+ + 5H2O S ≡SiO2(UO2)3(OH)5- + 7H+, log K0Si2 ) -20.0.
Introduction Long-term storage of nuclear fuels in deep geological sites is one of the major issues of the nuclear fuel cycle. One of the main concerns regarding the safe storage of nuclear waste in underground repository is the migration of radiotoxic elements through the geosphere (1, 2). Potentially released radionuclides such as U, Np, Pu, Am, and Cm may adsorb * Corresponding author phone: +33 1 69 15 73 42; fax: +33 1 69 15 71 50; e-mail:
[email protected]. † Universite ´ Paris XI. ‡ Universite ´ Henri Poincare´ Nancy I. 10.1021/es0348344 CCC: $27.50 Published on Web 01/16/2004
2004 American Chemical Society
onto mineral surfaces thereby enhancing retardation. Under environmental conditions, uranium typically occurs in the hexavalent form as the mobile aqueous uranyl ion (UO22+). The sorption of this ion onto solid surfaces has been widely studied because this process has a significant effect on transport properties (3-9). Several investigations regarding ion sorption processes on geological materials were mainly focused on the macroscopic aspects of the ionic species interaction with the mineral surface (4-7). This approach, however, did not yield any direct structural information on the chemical environment of the sorbed ion. We present here the use of two structural techniques that allow one to investigate the sorption mechanisms at a molecular scale and thus identify the species involved in the retention processes: laser-induced fluorescence spectroscopy (LIFS) and X-ray photoelectron spectroscopy (XPS). Both techniques are suitable as LIFS is highly sensitive for the uranyl species (10, 11) and XPS is a method well-adapted to probe the surfaces (12, 13) and is also sensitive for uranium. Characteristics of clay materials such as low permeability, high sorption capacity, and plasticity make them an effective barrier against radionuclides migration (14). Few studies were devoted to the microscopic study of uranium sorption on clay minerals, especially in acidic pH ranges (3, 8, 9, 15-17). In this study, we have investigated the sorption of the uranyl ion onto montmorillonite. Montmorillonite is a clay mineral composed of two-dimensional sheets of tetrahedral SiO4 alternating with sheets of octahedral Al(O,OH)6 (18) whose retention properties for fission products or actinides are acknowledged. For uranium sorption onto montmorillonite in suspension with low pH and low ionic strength values, cation exchange at negatively charged surface sites located on the basal planes was identified as the main retention mechanism (3, 16). Above pH 5, the surface hydoxyl groups provided additional sorption sites, located on the edges of the clay sheets, which then could interact with the uranyl species. The existence of both exchange and edge sites surface complexes on montmorillonite were already experimentally demonstrated using molecular spectroscopic probes (16, 17). Because montmorillonite presents several types of sorption sites (exchange sites, amphoretic edge sites), uranyl sorption mechanisms on this mineral are expected to be complex. The accurate determination of the species involved in the sorption process requires reference solids to simplify the studied system. Based on its aluminosilicate structure, montmorillonite can be viewed as a sophisticated assemblage of “basic structural units”, which could be represented as “aluminol” and “silanol” sites. The investigation of the U(VI) sorption mechanisms onto both alumina and silica solids and the comparison of the obtained results with those obtained for the U(VI)/montmorillonite system will allow one to clearly identify the sorbed uranyl species on montmorillonite versus pH and ionic strength. Then, the surface complexation modeling (constant capacitance model) of the U(VI) retention data onto montmorillonite can be completed on the basis of these structural constraints. The determined sorption constants appear more reliable because they take into account several constraints obtained from independent spectroscopic techniques. Finally, the validity of the sorption constants values can be tested by considering sorption edges obtained for a wide range of experimental conditions and especially for low cation concentrations where spectroscopic techniques are not sensitive enough. VOL. 38, NO. 5, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1399
Experimental Section Materials. The Na-montmorillonite was obtained from a Wyoming bentonite, marketed under the name Volclay. Prior to the sorption experiments, this clay was treated as follows: after dispersion in a 0.1 M HCl solution for 24 h, the obtained suspension was neutralized (pH 7) by adding concentrated NaOH solution. Following the neutralization process, the solution was centrifuged (2860g for 30 min). In a second step, sodium saturation was completed by adding a 0.5 M NaCl solution (contact time of 3 d). Next, the rinsing operation was completed by dialysis with a suspension volume over deionized water volume ratio equal to 0.5. The water used in the rinsing process was changed two times per day and then tested by measuring its conductivity value. This purification step allowed for the removal of different soluble phases (such as calcite), thereby resulting in a homoionic montmorillonite clay. The recovered solid was mainly montmorillonite, but a low percentage of silicated phases was still present (lower than 10% in weight: quartz, albite/ orthose) as observed by electron probe microanalysis (EPMA). The specific surface area of the clay was found to be 35 m2/g determined by N2-BET analysis, and the cation exchange capacity (CEC) was determined as 63 mequiv/100 g by using the ammonium acetate method (19). A silica gel and γ-alumina were chosen as model surfaces. These reference oxide compounds were purchased from Merck and used without further purification. The N2-BET specific surface areas of alumina and silica were 140 and 380 m2/g, respectively (20, 21). Uranium(VI) stock solutions were prepared by dissolving a known amount of UO2(NO3)2‚6H2O (Merck) in a previously acidified NaClO4 solution (pH 1) to prevent cation hydrolysis. The stock solution concentrations were about 5 × 10-3 or 10-2 M. The exact uranyl concentrations ((2%) were determined by R-liquid scintillation using a Tri-Carb spectrometer supplied by Packard (Camberra Co., Meriden, CT) and the Alphaex scintillation cocktail according to the protocol described in the literature (22). Sorption Experiments. The sorption of uranyl ion onto the three solids was studied at room temperature with pH values ranging from 2 to 8. Sorption experiments were carried out in a batch mode with crystal polystyrene or high-density polyethylene tubes in order to avoid the sorption of the uranyl ion to the tube walls. These uranyl solutions were prepared by adding an appropriate volume of the stock solution to the background salt solution (NaClO4). The experiments were conducted by mixing 0.2 g of solid with 20 mL of the uranyl solution present at various concentrations (from 10-3 to 10-6 M). The initial pH value was varied between 2 and 8 by adding to the suspension negligible volumes of concentrated HClO4 or NaOH solutions of analytical grade. The ionic strength was maintained at 0.5 or 0.1 M for the U(VI)/montmorillonite system, but in the case of alumina and silica, it was always maintained at I ) 0.1 M because the ionic strength has no influence on uranyl ion sorption on these oxides since innersphere complexes are involved (16, 21, 23). Moreover, for the spectroscopic investigation, additionnal experiments were also conducted with a 0.05 M ionic strength solution, but no sorption edge was realized in such conditions. The suspensions were stirred for 7 d, and the final pH value for each tube was measured directly in the suspension with a combined AgCl/NaCl electrode and a Tacussel pHmeter. The solution and solid were separated by centrifugation (3 h at 2860g). The supernatant was used to quantify the amount of uranium(VI) sorbed on the substrate. The uranium sorption rate was calculated according to the following equation: U(VI)sorbed (%) ) (1 - A/A0) × 100 where A0 represented the initial activity (Bq) in the solution and A represented the final activity after the sorption step (R-liquid 1400
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 5, 2004
scintillation counting measurement). The resulting solids were dried at room temperature and stored in a drier before being analyzed using LIFS and XPS. Spectroscopic Measurements. The uranyl emission spectra (characteristic strong green fluorescence centered at 520 nm) were collected on the previously dried samples, at room temperature (298 K) to compare the optical results with those obtained from XPS investigation. It was previously shown using EXAFS measurements (24) that the drying step does not affect the structure of the surface complexe. The fluorescence measurements were performed using a Continuum pulsed laser Nd:YAG (7-ns pulse duration) coupled with a Panther OPO, a SPEX 270M Jobin Yvon monochromator, a Hamamatsu photonics photomultiplier, and a Princeton Instruments Inc. model PG 200 pulse generator. The samples were excited at 355 or 430 nm to provide the best signal-to-noise ratio. The experimental setup was controlled by a PC with the Princeton Instruments Inc. WINSPEC program. The uncertainties on the position of the uranyl emission bands were approximately 1 nm, and the error bars associated to the corresponding lifetimes were lower than 10% (variation of the decay time value for several samples prepared in same conditions). All XPS spectra were collected on an ESCA apparatus with a multi-detection electron analyzer (VSW HA150, fixed analyzer transmission mode). A Mg KR source of photons (1253.6 eV and half-width of 0.9 eV) was used for both survey (FAT ) 90 eV) and narrow scans (FAT ) 22 eV). The powdered samples were fixed on a copper plate before being placed in the analytical chamber (10-9 mbar vacuum). As the studied compounds were electric insulators, the obtained spectra were corrected from the charge effect using as internal reference the C 1s line from adventitious aliphatic carbon (284.6 eV). The recorded lines, U 4f and C 1s were fitted using the XPSPEAK 3.0 program (25) after subtraction of the background (Shirley baseline). The uranium 4f spin-orbit splitting was held at 10.8 eV, and the component ratio (4f5/2: 4f7/2) was constrained at 0.8 (24, 26, 27). Moreover, a satellite peak (well-known for U(VI)) with an intensity around 10% of the main line was present on each spectrum at 3.5-3.7 eV toward higher binding energy (28, 29). Considering our device, the typical error bar associated to the binding energies was (0.3 eV.
Results and Discussion U(VI)/Na-Montmorillonite. As shown in Figure 1, the sorption edges of the uranyl ion on Na-montmorillonite present three distinct parts: (i) Below pH 4, the U(VI) sorption rate slowly increases. Moreover, the amount of sorbed uranium depends on both the ionic strength and the cation concentration. The greater the ionic strength (or metal concentration), the lower the sorption rate. Such behavior is commonly attributed to an interaction between aqueous ions and permanently negative charged sites, called exchange sites (4, 30), which leads to the formation of outer-sphere complexes. (ii) For pH values ranging from 4 to 6, the slope of the sorption edge curve is greater than for the previous region and the uranyl sorption rate increases up to 100%. (iii) For pH values greater than 6, all of the uranyl ions are sorbed on the clay; the amount sorbed remains near 100%. However, when 3 mg of calcite was added to the system corresponding to roughly 1.5% in weight (clay composition before the purification step), we observe a decrease of the sorption rate from pH 6 (around 90%) to pH 8 (around 30%) followed by an increase to 100% for pH values greater than 8 (Figure 2). This observation clearly shows that carbonate ions strongly affect the uranyl retention on Na-montmorillonite by complexing the uranyl ions in solution as aqueous carbonato complexes, which are not sorbed on the solid as
FIGURE 1. Sorption edges of U(VI)/montmorillonite system and corresponding surface species repartition diagrams: (a) [U(VI)] ) 10-4 M, I ) 0.5 M; (b) [U(VI)] ) 10-4 M, I ) 0.1 M; (c) [U(VI)] ) 10-3 M, I ) 0.5 M; (d) [U(VI)] ) 10-6 M, I ) 0.1 M. Experimental points (dots), calculated curve (straight), (≡X)2UO2 (dotted line), ≡Al(OH)2UO22 + (triangles), ≡SiO2UO2 (open squares), and ≡SiO2(UO2)3(OH)5- (crosses).
FIGURE 2. Comparison of U(VI) sorption edge on Na-montmorillonite with (open triangles) and without (squares) 3 mg of calcite added ([U(VI)] ) 10-4 M, I ) 0.5 M). it was already mentioned in the literature (31-35). This result also confirms that calcite was removed during the purification of the clay. According to the above macroscopic observations, the spectroscopic study of the sorbed uranyl species on clay was performed on U(VI)/Na-montmorillonite system without additional calcite. Moreover, because 10-4 M is an intermediate uranyl concentration in this study, the spectroscopic investigation was mainly focused on samples prepared with such a concentration. Nevertheless, as a comparison, some results were also obtained with samples corresponding to higher cation concentrations (10-3 M). We have recorded the emission spectra and the corresponding lifetimes of the uranyl ion sorbed on the Namontmorillonite prepared in NaClO4 solution (ionic strength from 0.05 to 0.5 M) at different pH values. The shape of the emission spectra clearly depends on the ionic strength and on the pH values of the suspension (Figure 3): (i) When the pH values are lower than 5, the main emission band is located at 522 nm (Figure 3a). The associated lifetimes
FIGURE 3. Fluorescence spectra of U(VI) sorbed onto Namontmorillonite vs pH and ionic strength: (a) I ) 0.5 M, [U(VI)] ) 10-4 M; pH 4.2 (I), pH 5.1 (II), and pH 8.0 (III). (b) [U(VI)] ) 10-4 M, pH 5.1; I ) 0.5 M (I) and I ) 0.05 M (II). VOL. 38, NO. 5, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1401
FIGURE 4. U4f XPS spectra (experimental and fit) of U(VI) sorbed onto Na-montmorillonite: [U(VI)] ) 10-4 M, I ) 0.1 M. (a) pH 4.2 and (b) pH 5.9.
TABLE 1. Spectroscopic Characteristics vs pH of U(VI) Species Sorbed onto Na-Montmontrillonite system
exptl conditions
decay times (µs)
U4f7/2 binding energies (eV)
pH < 3 10 383.7 U(VI)/montmorillonite 4 < pH < 5 10 + 55 + 190 383.7 + 382.6 pH > 5 55 + 190 + 400 382.6 + 381.9
were measured at 10 ( 2 µs (m - τ0), 55 ( 6 µs (m - τ1), and 190 ( 20 µs (m - τ2). Moreover, some additional experiments were performed at low pH values (less than 3) and low ionic strength (approximately 0.05 M) in order to facilitate the ion exchange process. Under such conditions only the decay time m - τ0 was measured (Table 1). (ii) When the pH values increase (from 5 to 8), the main fluorescence band shifts toward greater wavelengths (Figure 3a): 526 nm. This evolution is correlated with the appearance of another species having a longer lifetime (400 ( 30 µs named m - τ3 in the following text). Moreover, the decay time m - τ0 disappears when the pH increases. Thus, three decay times are required to fit the samples corresponding to the top of the sorption edge: m - τ1, m - τ2, and m - τ3 (Table 1). (iii) For each pH value, the emission spectra become more unresolved when the ionic strength is increased (Figure 3b). As expected according to the structure of montmorillonite, this first study shows that different environments of the sorbed uranyl ion exist on the clay, which could correspond to different sorption sites and/or different uranyl species. It is sometimes quite difficult to obtain a reasonable signalto-noise ratio when sorption processes are spectroscopically investigated, which can lead to questionnable interpretation. Then, XPS analysis has also been carried out on the samples. The obtained results appear to be in very good agreement with the optical investigation (Table 1). These results depend on the equilibrium sorption pH value as well (Figure 4): (i) When the ionic strength of the suspension is maintained at 0.05 mol/L and the pH is fixed at approximately 4, the corresponding U4f7/2 XPS spectrum (Figure 4a) was fitted considering two components: 383.7 ( 0.3 eV and 382.6 ( 0.3 eV. The full width at half maximum (fwhm) is equal to 2.3 eV, which agrees with previously published works dealing with uranyl sorbed onto phosphate and silicate materials (24, 26, 27). (ii) For the samples prepared at pH higher than 5 (Figure 4b), two components were used to fit the U4f7/2 peak: 382.6 ( 0.3 eV and 381.9 ( 0.3 eV (fwmh ) 2.3 eV). Thus, the 1402
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 5, 2004
FIGURE 5. Sorption edges of U(VI) sorbed onto oxide solids: I ) 0.1 M, [U(VI)] ) 10-4 M. Silica (triangles) and alumina (squares). highest binding energy (383.7 eV) observed for low pH values disappears when the pH increases, and a low binding energy (381.9 eV) appears when the pH is greater than 5. In the same time, the third binding energy (382.6 eV) remains the same for both low (4-5) and high pH values (above 5). The use of both optical and XPS techniques has clearly shown that the interaction between uranyl aqueous species and montmorillonite leads to several sorbed species that depend on the pH value (Table 1). The next step of the study is to identify these species. The investigation of U(VI) sorption onto both alumina and silica can allow one to reach that goal if these solid surfaces are seen as the “basic structural units” of the clay. U(VI)/Alumina. The sorption edges of U(VI) sorbed onto oxide solids are shown in Figure 5. For the U(VI)/alumina system, the sorption edge is observed for pH values ranging from 3.5 to 5.0. A 100% sorption rate is observed around pH 5, and it remains constant up to pH 8. All the sorption curves observed in this study do not show any decrease of the sorption rate for pH values greater than 5 as expected (35) because experiments were carried out in closed tubes where carbonate concentrations remain very low (around 10-5 M). This estimation considers a pCO2 ) 10-3.5 atm and the dead volume in the batch tubes (15 mL). An example of the emission spectra obtained for U(VI) sorbed on γ-alumina is shown in Figure 6a. In comparison with the U(VI)/clay system, no clear difference in the shape and the position of the spectra was observed for all analyzed samples prepared at different pH values: the maxima positions are located at 497, 518, and 540 nm. However, some differences were found in the fluorescence lifetimes analysis: for the samples prepared using pH values lower than 5, only one species (characterized by a - τ1 ) 45 ( 5 µs) was observed, while for pH values higher than 6, two uranyl lifetimes were observed: a - τ1 ) 45 ( 5 µs and a - τ2 ) 120 ( 10 µs. Hence, a new uranyl surface complex appears on the solid when the pH increases (greater than 6), which could arise from the U(VI) speciation (sorption of hydrolyzed and/or polynuclear species) as shown in Figure 8 (calculations were performed using Chess2.5; 36). The uranyl sorbed species observed at pH values lower than 5 (a - τ1) was also identified by a single U4f7/2 binding energy located at 381.8 ( 0.3 eV, as shown in Figure 7, whereas the U4f binding energy at 380.2 ( 0.3 eV was associated to the second species (characterized by the a - τ2) that appears above pH 5.5. The presence of two types of U(VI)/alumina complexes depending on the pH value is in agreement with the literature (16). Comparison of the U(VI) speciation (Figure 8a) with the above results suggests that the first species (a - τ1 and 381.8
TABLE 2. Spectroscopic Characteristics of U(VI) Species Sorbed onto Silica and Alumina Surfaces and Corresponding Attribution τ (µs)
EB (eV) 380.2 381.8 382.2 383.3
45
65
120 (UO2)n(OH)m(2n-m)+ on
UO22+ on Al2O3
180 Al2O3
400 (UO2)n(OH)m(2n-m)+ on SiO2
UO2H3SiO4+ on SiO2 UO22+ on SiO2
FIGURE 7. U4f XPS spectra of U(VI) sorbed onto alumina (experimental and fit): [U(VI)] ) 10-4 M, I ) 0.1 M. (a) pH 4 and (b) pH 6.
FIGURE 6. Fluorescence spectra of uranium(VI) sorbed ([U(VI)] ) 10-4 M, I ) 0.1 M) onto alumina (a) or onto silica (b) vs pH. (a) pH 4.3 (I) and pH 6.3 (II). (b) pH 7.0 (I), pH 5.9 (II), pH 4.5 (III), and pH 3.7 (IV). eV) arises from the interaction between free UO22+ ion and aluminol sites (Table 2). The second species (a - τ2 and 380.2 eV) results from the interaction of aluminol surface sites and the (UO2)3(OH)5+ complexes for instance (Table 2) because this species is one of major aqueous complexes for pH values greater than 5. These findings agree with previous work where the sorption of a polynuclear species onto alumina occurs for pH values greater than 6 (16). U(VI)/Silica. The uranium(VI) sorption edge (Figure 5) begins at pH 2 and reaches the 100% adsorption at pH 5. A sorption edge spread over 3 pH units is often relevant of the presence of many surface complexes (37). The LIFS study was completed for samples prepared between pH 3.5 and pH 8.0. Fluorescence spectra of uranyl ion sorbed on silica versus pH are shown in Figure 6b. The emission bands located at 498, 518, and 541 nm for pH values around 4 are redshifted when the sorption pH value increases (506, 528, and 552 nm for pH ranging from 6 to 8). This behavior, which is identical to the one observed on the U(VI)/Na-montmorillonite system, indicates the presence of more than one uranyl complex onto the silica surface. In addition, the analysis of the fluorescence decays yields more details about the
FIGURE 8. Solution speciation of uranyl ([U(VI)] ) 10-4 M, [H2CO3] ) 1.2 × 10-5 M, I ) 0.1 M) and [H4SiO4] ) 1.5 × 10-4 M (a) or [H4SiO4] ) 2 × 10-3 M (b). Stability constants from Grenthe et al. (51, 52) and Moll et al. (53). Uranyl species are noted as (p, q, r) where p refers to UO22+, q refers to OH-, and r refers to CO32-. identification of the surface species. When the sorption pH value is lower than 5, two fluorescence lifetimes were identified: s - τ1 ) 65 ( 6 µs and s - τ2 ) 180 ( 20 µs. Above pH 6, two lifetimes were also necessary to explain the fluorescence decays: the lifetime value s - τ2 at 180 µs was still observed and a new value, s - τ3 ) 400 ( 30 µs, was observed while s - τ1 disappeared. Between pH 5 and pH 6, the three fluorescence lifetime values were necessary to explain the fluorescence decays. These results agree with the findings of Gabriel (38): decay times values of 170 and 360 µs were identified for uranyl ions sorption (10-6 M) on another similar silica (Aerosil 200 by Degussa). The analysis of the XPS spectra yielded similar results (Figure 9). For samples prepared around pH 4 (Figure 9a), the XPS spectrum was fitted considering two components (U4f7/2): 382.2 and 383.3 eV with a fwhm equal to 2.3 eV. The VOL. 38, NO. 5, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1403
FIGURE 9. U4f XPS spectra of U(VI) sorbed onto silica (experimental and fit): [U(VI)] ) 10-4 M, I ) 0.1 M. (a) pH 4 and (b) pH 6. XPS spectrum of a sample prepared at pH 6.4 was also fitted with two components (Figure 9b): the first one is the same as the previous value located at 383.3 eV, the second one, 381.8 eV, is significantly different from the binding energy at 382.2 eV observed around pH 4, evidence of the formation of a new surface species. Therefore, the series of LIFS and XPS results (three sorbed uranyl species on silica, depending on the pH value) are in very good agreement. For the uranium(VI)/alumina system, the presence of these three surface complexes could also be explained by solution speciation taking into account the amount of dissolved silicates released by the silica ([Si]solution ∼ 2 × 10-3 M) (38, 39). For the lower pH values, the main aqueous species are UO22+ and UO2H3SiO4+ (Figure 8b); thus, the two first surface complexes (s - τ1 ) 65 µs; 382.2 eV) and (s - τ2 ) 180 µs; 383.3 eV) are probably relevant to the sorption of these species onto silanol sites (Table 2). It is quite difficult to unambiguously address the spectroscopic values to one of these species. However, because 170 µs was allocated to the ≡SiO2UO2 surface species (38), we can reasonably address the second surface species (s - τ2 ) 180 µs; 383.3 eV) to the sorption of UO22+ onto silica. Then, s - τ1 corresponds to the sorption of UO2H3SiO4+ onto silica. The third lifetime (s - τ3 ) 400 µs) only observed for the higher pH values accounts for the sorption of hydrolyzed uranyl species such as (UO2)3(OH)5+ for instance (17, 40). This assumption is supported by the work of Sylwester et al. (16), who have shown using XAS the presence of U-U backscattering in the case of U(VI) sorbed onto silica at pH around 6. Mechanism of U(VI) Sorption onto Montmorillonite. For all systems under study, we have first verified that no precipitate was formed at high pH values during the sorption process. Indeed, XPS measurements did not give any evidence for the presence of neither uranium(VI) carbonate solids (C1s line) nor uranium(VI) hydroxyde such as schoepite, for instance. Comparison of the spectroscopic results obtained for the U(VI)/Na-montmorillonite system (Table 1) with both uranium(VI) oxide systems (Table 2) shows that it is possible to propose a sorption mechanism of uranyl ion onto this clay: (i) The surface species characterized by m - τ0 ) 10 µs (only observed at low pH values) is correlated to the presence of an unusually high binding energy (383.7 eV) for uranium. These spectroscopic parameters (10 µs; 383.7 eV) are not observed for the reference systems. On one hand, several studies of uranium(VI) sorption onto smectite clays have shown that the interaction of uranyl species and exchange sites occurs via an outer-sphere complex (3, 16, 17). This result has also been reported for other cations (41). The decay 1404
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 5, 2004
time of such a species is expected to be quite similar to the one observed in aqueous solution (2 µs for aqueous uranyl ion). On the other hand, the unusually high binding energy is only observed for low pH values and montmorillonite system, which suggests that this energy could correspond to uranyl sorbed in the interlayer space of the clay as supported by other studies performed with Eu(III) (42), which have shown that much greater binding energies are obtained when the cation is retained by exchange sites as compared to edge sites. We concluded that the uranyl ion sorbed in the interlayer space is characterized by the following spectroscopic parameters: 10 µs and 383.7 eV. (ii) The longer decay time observed for U(VI) sorbed onto montmorillonite (m - τ3 ) 400 µs) at high pH values is also observed with silica for pH > 6 (s - τ3); moreover, the binding energy around 381.8 eV is obtained for both systems (381.8 eV for silica and 381.9 eV for montmorillonite). It appears that the nature of this uranyl species is probably the same for both solids, and we propose that this U(VI) species (probably a polynuclear one) is sorbed onto “silanol” edge sites of montmorillonite with the following spectroscopic characteristics: 400 µs and 381.9 eV. (iii) The other surface species characterized by m - τ1 ) 55 µs and m - τ2 ) 190 µs are observed whatever the pH values investigated. The longer decay time (m - τ2 ) 190 µs) is very close to the one attributed to the sorption of free UO22+ onto silica (180 µs). We can therefore conclude that m - τ2 is relevant to the sorption of free uranyl ion onto “silanol” edge sites of the clay. Moreover, m - τ1 ) 55 µs is very close (within the uncertainties) to the value obtained for the sorption of UO22+ onto alumina (45 µs) but also the one of UO2H3SiO4+ onto silica (65 µs). However, the amount of dissolved silicates arising from the dissolution of montmorillonite was determined to be approximately 1.5 × 10-4 M in our experimental conditions (ICP-AES measurements). This value is in good agreement with the work of Wanner et al. (43). For such a silicates concentration, the amount of UO2H3SiO4+ aqueous complex is less than 10% (Figure 8a). Thus, under such conditions, the complexation of U(VI) by silicates is not favored, and an uranium(VI) silicate complex is not expected to be sorbed in our experimental conditions, which led us to address m - τ1 to the sorption of free uranyl ion onto aluminol edge sites. This conclusion is supported by the work of Hennig et al. (44) and Da¨hn et al. (45), who have recently shown, using X-ray absorption spectroscopy, that aluminol sites are active toward cations retention. Finally, with regards to XPS results, we can underline that only two binding energies are observed for pH values higher than 6 when U(VI) is sorbed onto montmorillonite (381.9 and 382.6 eV). The lower binding energy (which is correlated to the lifetime equals 400 µs) was attributed to an interaction between U(VI) species and silanol edge sites. It appears that the binding energy equal to 382.6 eV is probably correlated to both species characterized by the decay times m - τ1 and m - τ2, which is supported by the fact that at low pH values, 382.6 eV is also obtained while m - τ1 and m - τ2 are measured. Thus, 382.6 eV appears to be an average binding energy value. Concluding the above discussion, the spectroscopic characteristics of the U(VI) sorbed species onto montmorillonite are 10 µs and 383.7 eV for the ion exchange process; 55 µs for the interaction between UO22+ and aluminol edge sites and 190 µs for the sorption of UO22+ onto silanol edge sites, both characterized by the average binding energy at 382.6 eV; 400 µs and 381.9 eV for the sorption of a polynuclear U(VI) species ((UO2)3(OH)5+) onto silanol edge sites. This investigation coupling LIFS and XPS measurements has demonstrated that uranyl aqueous species interact with both edge sites and exchange sites of montmorillonite. The results have shown that three sorbed species arise from the
TABLE 3. Surface Reactions and Corresponding Constant Values Used for Montmorillonitea
a
montmorillonite surface reactions
log K0
≡Al(OH) S ≡AlO- + H+ ≡Al(OH) + H+ S ≡AlOH2+ ≡Si(OH) S ≡SiO- + H+ ≡XNa + H+ S ≡XH + Na+
-9.4 7.9 -7.8 1.0
From ref 46.
interaction between uranyl species and aluminol and silanol sites. Moreover, ion exchange is only observed for pH values lower than 5. As expected, the U(VI)/montmorillonite system is complex because four different sorbed species are identified. Indeed, not only the substrate presents several reactive surface sites but also uranyl speciation is rather complex (with the appearance of several hydrolyzed and polynuclear aqueous species) for pH values greater than 4-5, which makes the system extremely complex to study. The coupling of LIFS and XPS investigations has allowed one to identify all these sorbed species. Then, these results can be used as experimental constraints for the thermodynamical modeling of the U(VI)/montmorillonite system sorption edges. Surface Complexation Modeling. A multisite surface complexation model was considered in this study. This model included cation exchange sites together with surface complexation sites. The sorption data modeling was completed according to the above structural investigation. The characteristics of the amphoteric behavior of the montmorillonite were previously described (Table 3) as well as the concentrations of each type of sites (19, 42): 5.75 × 10-4 mol/g for exchange sites and for the edge sites, 1.75 × 10-5 and 2.5 × 10-4 mol/g for aluminol and silanol sites, respectively. The surface acidity constants (Ka) relative to silanol and aluminol sites were previously reported (46). These constants refer to protonation/deprotonation processes of the hydroxylated surface groups according to
≡SOH + H+ S ≡SOH2+
(K1)
≡SOH S ≡SO- + H+ (K1) Using the surface electrostatic potential, the expression of K1 and K2 are
K1 ) K2 )
[≡SOH2+]
exp(FΨ0/RT)
[≡SOH][H+]
[≡SO-][H+] exp(-FΨ0/RT) [≡SOH]
where F is the Faraday constant, Ψ0 is the surface electrostatic potential, R is the ideal gas constant, and T is the temperature expressed in Kelvin. No potentiometric titrations were completed on the montmorillonite used in this study. Surface acidity constants were previously determined for alumina and silica solids (20, 21). In a first approximation, it was assumed that these constants are still valid to account for the amphoteric behavior of the montmorillonite (46). Thus, the used aluminol and silanol acidity constants were those determined for alumina and silica reference solids, respectively. We have chosen the constants reported by Hurel et al. (46) because they were obtained considering the same surface complexation model as used in this study. The Constant Capacitance Model (CCM) included in the FITEQL3.2 code (iterative nonlinear least squares optimization program) was
used to complete the sorption data modeling (47). This model allows one to consider ionic strengths as high as 0.5 M and also implies a rather low number of adjustable parameters (inner-layer capacitance, acidity constants, and sites densities). The inner-layer capacitance value considered in these simulations (1 F/m2) is the empirical value commonly used for the oxide minerals (48). Note that this value was kept constant, which seems to be a reasonable assumption since the variation of the ionic strength for the sorption edges was only from 0.1 to 0.5 M. Therefore, because the surface site density was already determined, only the uranium sorption constants need to be calculated. For the fitting procedure, the errors were estimated to (0.1 pH unit, and the absolute uncertainty on the sorbed uranyl concentration was considered to be 5% of the initial metal concentration. An estimation of 0.5 logarithm unit was postulated for the calculated sorption constants logarithms. This value corresponds to an estimation regarding to the goodness of the fit for the range of experimental conditions considered in this work. Because surface complexation modeling requires the use of equilibria, the reversibility of the sorption processes under study was checked as follows: after the sorption step, negligible volumes of concentrated HClO4 were added to the system and the aqueous uranyl concentration as well as the corresponding pH value was then measured after 7 d of shaking. For all the studied conditions, the sorption processes were found to be fully reversible. For example, in the solution where [U(VI)] ) 10-4 M and Ι ) 0.5 M, the addition of various volumes of concentrated perchloric acid to suspensions after sorption at pH 6 (sorption rate near 100%) led to pH 4.4 (sorption rate 53%) and pH 3 (sorption rate 26%). These results agree with the forward sorption edge (within the error bars) indicating the reversibility of the processes. The sorption data modeling was completed according to our above structural investigation. Several spectroscopic investigations using XAS experiments have shown that uranyl is sorbed onto several mineral substrates as bidentate surface complexes (24, 27, 49, 50) as well as for montmorillonite (16, 44). Therefore, the formation of monodentate surface complexes was not considered in the fitting procedure. The hydrolysis and aqueous complexation constants used in this work to account for uranyl chemistry were those reported by Grenthe et al. (51, 52), and the constant reported by Moll et al. (53) was considered for uranyl complexation with silicic acid. We had no information regarding the number of proton released during the sorption process and the choice of the sorption equilibria was based on the goodness of the modeling for this point. The best fits were obtained considering the following sorption equilibria:
2≡XNa + UO22+ S (≡X)2UO2 + 2Na+ ≡Al(OH)2 + UO22+ S ≡Al(OH)2UO22+ ≡Si(OH)2 + UO22+ S ≡SiO2UO2 + 2H+
(Kexch) (KAl) (KSi1)
≡Si(OH)2 + 3UO22+ + 5H2O S
≡SiO2(UO2)3(OH)5- + 7H+
(KSi2)
The ionic strength corrections for surface acidity constants as well as aqueous complexation of uranyl were completed using the Davies equation. The same set of surface complexation constants was able to successfully account for all the sorption conditions (Figure 1). These constants, corrected to zero ionic strength, are as follows: log K0exch ) 3.0 ( 0.5, VOL. 38, NO. 5, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1405
log K0Al ) 14.9 ( 0.2, log K0Si1 ) -3.8 ( 0.2, and log K0Si2 ) -20.0 ( 0.1. The surface species repartition diagrams (Figure 1) are in quite good agreement with the structural results, with mainly three species for pH values ranging from 4 to 5 and three for pH greater than 6. It can be concluded that, whatever the uranyl concentration, the same species with the same sorption constants are able to account for all the experimental conditions considered in this work (U(VI) concentrations and ionic strengths). This is a fundamental point in the field of radionuclides migration where the concentrations of the released cations are expected to be very low and then do not allow a structural investigation with regards to the sensitivity of the spectroscopic techniques. Thus, because the uranyl sorption mechanisms are the same for high (10-4-10-3 M) and low (10-6 M) uranyl concentrations, this study clearly shows that spectroscopic techniques are very powerful tools for the retention processes study. However, even if the sorbed species are the same, some differences can be underlined in the surface species repartition depending mainly on the total uranyl concentration. At low uranyl concentrations (
