Density Functional Model Study of Uranyl Adsorption on the Solvated

Jul 15, 2010 - Benjamı Martorell, Alena Kremleva, Sven Krüger, and Notker Rösch*. Technische UniVersität München, Department Chemie and Catalysis...
7 downloads 0 Views 2MB Size
J. Phys. Chem. C 2010, 114, 13287–13294

13287

Density Functional Model Study of Uranyl Adsorption on the Solvated (001) Surface of Kaolinite Benjamı´ Martorell, Alena Kremleva, Sven Kru¨ger, and Notker Ro¨sch* Technische UniVersita¨t Mu¨nchen, Department Chemie and Catalysis Research Center, 85747 Garching, Germany ReceiVed: February 10, 2010; ReVised Manuscript ReceiVed: June 14, 2010

We studied the adsorption of uranyl on bare and solvated models of the octahedral (001) surface of kaolinite using first-principles density functional calculations. Inner-sphere bidentate complexes adsorbed at partially deprotonated short-bridge sites AlOO(H) and long-bridge sites AlO-AlO(H) were modeled as the most probable adsorption complexes. The uranyl complex at the doubly deprotonated AlO-AlO long-bridge site exhibits a third contact to the surface, not present in the complex at the corresponding short-bridge site. Adsorption at short-bridge sites is energetically favored compared to complexes at long-bridge sites. We were unable to determine stable adsorption complexes of uranyl at singly deprotonated AlO-AlOH longbridge sites. Surface solvation, approximated via an adsorbed monolayer of water molecules, hardly affects the adsorption complexes of uranyl. Contacts U-Oeq in the equatorial plane shorten by 2 pm, U-Al distances elongate by up to 4 pm. In contrast to the bare surface, adsorption complexes at long-bridge and short-bridge sites of the solvated surface exhibit similar stabilities. 1. Introduction Actinide adsorption on clay minerals is a key process in the environmental chemistry of these elements.1-3 In nature adsorption can serve as a possible retardation mechanism of actinide migration. Clay minerals are ubiquitous in the geosphere; thus, adsorption of actinides at clay materials can take place everywhere, in soils, aquifers, or aquatic sediments. Recently scientific interest in actinide adsorption problems has grown in the context of assessing the risk of possible deep geological repositories for radioactive waste.4,5 There clay materials can serve as natural or technical (bentonite) barriers. Clays are typically composed of clay minerals, quartz, calcite, and small amounts of other materials (less than 5%). The clay mineral content of clays is relatively high, ∼50%, and therefore important for the adsorption properties of this material. Several experimental studies addressed the adsorption of radionuclides relevant to nuclear waste on clay rocks and separately on their constituting clay minerals, e.g., montmorillonite, illite, and kaolinite.6-13 Exploring the adsorption on purified clay minerals helps with identifying the minerals pertinent to the sorption of actinides. Experimental investigations on the adsorption of U, Pu, and Np actinides at montmorillonite and kaolinite recently became more frequent.10-13 Some studies explored actinide adsorption on natural clay rocks, Opalinus clay, and CallovoOxfordian clay because these argillaceous rocks are considered as potential host rock formations for geological deep repositories in Switzerland and Germany (Opalinus clay)5,14 and France (Callovo-Oxfordian clay).15 Despite various experimental studies16-23 knowledge on the atomic scale about the speciation and the structures of adsorption complexes on clay materials is scarce. Therefore a computational approach, that is able to provide reliable complementary information, can help to achieve a better understanding and a more detailed interpretation of such adsorption processes. Yet, computational studies of actinides at mineral surfaces are still rare.24-28 * E-mail: [email protected].

For our explorations of actinide adsorption on clay minerals under environmental conditions, we chose uranyl as the most stable form of uranium(VI) in aqueous solution29 and kaolinite as an exemplary clay mineral with a relatively small unit cell. Kaolinite is also a major constituent of the clay mineral content of Opalinus clay.14 The adsorption of uranyl on kaolinite was widely studied experimentally.8,18,20,30,31 Uranyl was found to adsorb to almost 100% on kaolinite at pH above 6, which is close to the pH of the zero-point charge of kaolinite.8,30 In the presence of air, adsorption decreases dramatically at pH >8, most probably due to the formation of uranyl-carbonato complexes.8,30 Time-resolved laser fluorescence spectroscopy (TRLFS) showed uranyl species of two life times adsorbed on kaolinite.18 However, the interpretation of the TRLFS results did not afford an identification of the adsorbed species. Several extended X-ray absorption fine structure (EXAFS) spectroscopic experiments addressed the structure of the adsorbed complexes.20,30,31 Results have been interpreted as inner-sphere complexes of uranyl with two different U-Al/Si distances resolved.30,31 Recent resonant anomalous X-ray reflectivity (RAXR) experiments reveal a more complex adsorption behavior of various cations that indicates the coexistence of inner- and outer-sphere adsorption complexes at mineral surfaces.32 Thus, there are important open questions related to the adsorption of uranyl on kaolinite, e.g., (i) which facets of kaolinite (basal or edge) contribute to which extent to the adsorption, (ii) what are the adsorption sites, and (iii) which adsorption complexes can be formed on these surfaces. Aiming at the clarification of some of these issues, we report in the following a computational approach to model adsorption sites and various species on the octahedral (001) surface of kaolinite which prevails on typical crystallites.33 In a previous study26 of uranyl adsorption at (001) surfaces of kaolinite we addressed octahedral, Al(o), and tetrahedral, Si(t), (001) surfaces of kaolinite, and we found that adsorption at Si(t) is unlikely. In that earlier work, we examined monodentate adsorption of uranyl but explored also some bidentate complexes at sites with O centers that are bound to the same Al center.26 Here we extend

10.1021/jp101300w  2010 American Chemical Society Published on Web 07/15/2010

13288

J. Phys. Chem. C, Vol. 114, No. 31, 2010

Martorell et al.

that study to other sites by examining uranyl adsorption at various pairs of O- and OH surface groups that are bound to neighboring Al centers. Furthermore, in the present work we also approximately account for the solvation of the surface which we model explicitly by means of a monolayer of water molecules. 2. Computational Details We carried out first-principles calculations on supercell models of uranyl adsorbed on the octahedral (001) surface of kaolinite, using a plane-wave method based on density functional theory (DFT) as implemented in the Vienna ab initio simulation package (VASP).34 The structures were optimized with the generalized gradient approximation (GGA) for the exchangecorrelation potential (PW91)35 which is appropriate for the relatively weak interactions present in the models studied. In our previous work, we had optimized structures with the localdensity approximation (LDA).26 Here, we also modeled solvation effects on the adsorption complexes by explicit inclusion of a monolayer of water molecules above the mineral surface. Thus the systems examined include hydrogen bonds, the strength of which is usually overestimated at the LDA level.36 For consistency, all systems without and with explicit solvation of the surface were optimized at the GGA level. Changing from LDA to GGA only slightly affects the bulk parameters of kaolinite.26 The resulting triclinic primitive unit cell, optimized at the GGA level, is characterized by the parameters a ) 521 pm, b ) 905 pm, c ) 745 pm, and R ) 92.2°, β ) 105.1°, γ ) 89.8°. We used a (2 × 2) surface unit cell to study uranyl adsorption. The core electrons were described by the fullpotential projector augmented wave (PAW) method.37,38 Scalar relativistic effects, which are important for heavy metals, were incorporated via effective core potentials37 as well as explicit mass-velocity and Darwin corrections.39 The energy cutoff of the plane-wave expansion was set to 400 eV, which had been found sufficient when optimizing a slab model of a clean kaolinite surface.26 Integration over the Brillouin zone was carried out with a (2 × 2 × 1) k-point grid,40 applying a generalized Gaussian smearing method41 with a smearing width of 0.15 eV. In geometry optimizations, the total energy was converged to 10-4 eV, and forces acting on ions were required to be less than 10-4 eV/pm. 3. Surface Models Kaolinite is a layered two-sheet aluminosilicate42 with the composition Al4Si4O10(OH)8 per unit cell (no permanent charge). Kaolinite layers consist of a “tetrahedral” Si(O,OH)4 sheet Si(t) that is bound to an “octahedral” Al(O,OH)6 sheet Al(o) via inner vertex oxygen centers. Hydrogen bonds connect these two-sheet layers.42 The octahedral basal surface (001) of kaolinite contains four crystallographically different Al centers and six surface OH groups which can be grouped into three pairs of essentially equivalent centers as the space group C1 of the unit cell is only weakly perturbed (Figure 1a).42,43 An earlier DFT study suggested that the unsolvated Al(o) surface of kaolinite exhibits two types of hydroxyl groups:44 two-thirds are “upright” OH groups (uOH), oriented perpendicularly to the surface, and onethird are “lying” OH groups (lOH), essentially oriented parallel to the surface (Figure 1a). The lying OH groups are stabilized by hydrogen bonds to neighboring surface oxygen centers.44 In the present work we will compare the adsorption of uranyl on uOH and lOH groups at various adsorption sites. The basal (001) surface of kaolinite was modeled as a single layer, exposing the (001) surface; this slab model consists of

Figure 1. (a) Top view of the Al(o) basal (001) surface of kaolinite. uOH, upright surface OH group; lOH, lying surface OH group; symmetry inequivalent surface oxygen centers are numbered for convenience, corresponding equivalent centers are labeled by prime. Inspected adsorption sites: solid blue lines, short-bridge sites uu (O1-O2), ul (O2-O3); dashed lines, long-bridge sites uu (O1-O2′) and ul (O1-O3). Arrows indicate the final location of the complexes whose optimization was started at partially deprotonated AlO-AlOH long-bridge sites (see text). Selected O-O distances: O1-O2 ) O2-O3 ) 282 pm, O1-O3 ) 348 pm, O1-O2′ ) 347 pm. (b) Side view of the single-layer model of kaolinite with atomic “sublayers” indicated.

six atomic “sublayers”, H-O-Al-O-Si-O (Figure 1b). Earlier, this single-layer model had been found adequate for describing the kaolinite basal surface.45 The four uppermost sublayers of the Al(o) surface (H-O-Al-O) were optimized together with the adsorption complexes, whereas the two “bottom” sublayers (Si-O) were kept fix at the optimized bulk geometry to ensure bulk boundary conditions. Thus, surface relaxation was taken into account for the four “top” sublayers of the slab model. Experiments showed that uranyl adsorption depends on the pH8,20,30 as direct consequence of the increasing deprotonation of the surface with increasing pH. To simulate this feature, we modeled two types of adsorption sites: sites with (i) two deprotonated OH groups (Os- centers) or (ii) one Os- center and one OH group. Previously we showed26 that only bidentate uranyl adsorption complexes on at least partially deprotonated sites of the Al(o) surface exhibit U-Al distances that are comparable with experiment.20,30,31 Monodentate uranyl adsorption yields rather long U-Al contacts, beyond 360 pm;26 therefore, these complexes may not have been seen in EXAFS experiments.30,31 For the bidentate adsorption mode only neighboring O centers connected to the same Al atom had previously been studied.26 In the following, we will refer to these sites as short-bridge sites (AlOO) because they feature relatively short O-O distances, ∼284 pm (Figure 1a). In addition, here we also considered possible adsorption sites that comprise pairs of O centers attached to neighboring Al centers. These additional adsorption sites are referred to as long-bridge sites (AlO-AlO) as they feature O-O distances up to ∼340 pm (Figure 1a). Adsorption sites are further differentiated according to the contributing surface OH groups: (i) uu sites where both surface OH groups are upright and (ii) ul sites where an upright and a lying OH group are involved. Both experimental46,47 and theoretical48,49 results suggest that uranyl is preferentially 5-fold coordinated in its equatorial plane. Therefore, in the present models, we assumed 5-fold coordination in all complexes

Uranyl Adsorption on Kaolinite studied. Thus, three water molecules of the first coordination sphere of uranyl were explicitly included in the models of bidentate adsorption complexes. The VASP code provides compensating corrections for charged unit cells of cubic lattices.50 We applied this computational strategy to optimize the structures of the molecular species [UO2(H2O)5]2+ and H3O+ in unit cells of 1 × 1 × 1 nm3. For uranyl adsorbed on the basal surfaces of kaolinite, models with neutral unit cells had to be created by invoking surface defects, as done in our previous study;26,51 see the Supporting Information. SiOH defects on the Si(t) side of the slab model were introduced where appropriate to allow neutralization of the unit cell via deprotonation. For consistency with our previous work we here used two SiOH defects per (2 × 2) unit cell, one of which was deprotonated when necessary to compensate the charge of the adsorption complex. The formation energies ∆Eform of the adsorption complexes were estimated as in our previous study; see Supporting Information.26 Neutral SiOH defects hardly affect the geometric parameters of uranyl adsorption complexes: bond distances change at most by 1 pm and the formation energies of complexes by less than 5 kJ mol-1.45 Note, however, that our strategy of charge neutralization is limited by uncertainties in the solvation contribution to the deprotonation energies of the SiOH defects.26 Therefore, formation energies ∆Eform should be compared only for complexes at adsorption sites that exhibit the same (initial) charge q; overall their size represents a rough estimate only.26 4. Results and Discussion 4.1. Uranyl Adsorption on the Bare Al(o) (001) Surface of Kaolinite. Uranyl Adsorption on Doubly Deprotonated Sites AlOO and AlO-AlO. We started by exploring adsorption sites that correspond to an elevated pH value, above the pH of zero-point charge of kaolinite of ∼5.5.52 Two surface OH groups per (2 × 2) unit cell are deprotonated in the corresponding models. Uranyl adsorption at short-bridge sites AlOO had previously been studied at the LDA level where ul and uu sites had been found to yield similar adsorption complexes.26 Here, we consistently used a GGA functional (see section 2). Figures 2 and 3 display these reoptimized structures as well as those of uranyl at long-bridge doubly deprotonated sites. Table 1 collects the formation energies and pertinent structure parameters of these complexes. The adsorption complexes at the short-bridge uu and ul sites are structurally very similar; the largest difference is 2 pm in the bonding distances U-Ow and U-Os (Table 1). Both these complexes at AlOO shortbridge sites are characterized by two U-O bonds to the surface, 213-215 pm and 219-220 pm, respectively. Overall, U-Os distances to the surface are ∼40 pm shorter than those to the aqua ligands, U-Ow (Table 1). There is also a single short U-Al distance, 311 pm. Adsorption complexes at uu and ul long-bridge sites AlO-AlO differ notably more than the corresponding complexes at shortbridge sites AlOO (Table 1). The U-Os bonds in the complexes at AlO-AlO sites are longer than those at AlOO sites (uu AlO-AlO: 218, 229 pm, uu AlOO: 213, 219 pm, ul AlO-AlO: 227, 231 pm, ul AlOO: 215, 220 pm). A third U-Os contact, ∼260 pm, to a surface OH group in complexes at AlO-AlO sites is in the range of U-Ow bonds (Table 1). In view of this third contact, uranyl at long-bridge sites is expected to be closer to the surface. If one estimates the height of the uranium center above the surface with respect to the Si sublayer of the substrate, then uranyl at AlO-AlO sites sits ∼20 pm lower above the surface than in the short-bridge complexes at AlOO sites. The

J. Phys. Chem. C, Vol. 114, No. 31, 2010 13289

Figure 2. Optimized adsorption structures of uranyl adsorbed on doubly deprotonated short-bridge sites of the Al(o) basal (001) surface of kaolinite: (a) uu AlOO and (b) ul AlOO. Hydrogen bonds to the surface are indicated as dashed lines; numbers indicate their lengths (in pm).

Figure 3. Optimized adsorption structures of uranyl adsorbed on doubly deprotonated long-bridge sites of the Al(o) basal (001) surface of kaolinite: (a) uu AlO-AlO and (b) ul AlO-AlO. Hydrogen bonds to the surface are shown as dashed lines (lengths in pm).

complex at the ul AlO-AlO site clearly exhibits coordination number CN ) 6: there are two shorter U-Os bonds, one longer U-Os bond of 261 pm, and three similar U-Ow bonds of 264-265 pm. This results in an average value U-Oeq ) 252 pm, whereas in adsorption complexes at short-bridge AlOO sites U-Oeq ) 245 pm. The adsorption complex at the uu AlO-AlO site exhibits one rather long U-Ow bond, 280 pm. If one views this aqua ligand as part of the first coordination shell of uranyl, then CN ) 6 results with U-Oeq ) 254 pm. On the other hand,

13290

J. Phys. Chem. C, Vol. 114, No. 31, 2010

Martorell et al.

TABLE 1: Structure Parametersa (Distances in pm) and Formation Energies ∆Eform (in kJ mol-1) of Uranyl Adsorption Complexes at Short-Bridge Sites AlOO and AlOOH and Long-Bridge Sites AlO-AlO of the (001) Al(o) Surface of Kaoliniteb model

q

2+

UO2(H2O)5 bare surface uu AlOO ul AlOO uu AlO-AlO, CN ) 6c ul AlO-AlO, CN ) 6c uu AlOOH ul AlOOH solvated surface uu AlOO ul AlOO uu AlO-AlO ul AlO-AlO, CN ) 6c uu AlOOH ul AlOOH

U-Ot

U-Os

178 -2

-1 -2

-1

U-Ow

U-Oeq

245, 245, 248 248, 251

247

U-Al

∆Eform

187 186 187 185 186 185

213, 215, 218, 227, 213, 214,

219 220 229, 260 231, 261 264 266

263, 263, 266, 264, 257, 257,

266, 264, 271, 265, 260, 257,

266 264 280 265 262 265

245 245 254 (249d) 252 251 252

311 311 333, 340 341, 344 332 333

206 195 235 261 -67 -76

187 187 188 185 187 186

220, 216, 222, 229, 220, 217,

222 218 223, 258 232, 267 262 265

247, 256, 252, 249, 243, 250,

253, 259, 259 253, 246, 251,

271 272

243 244 243 249 249 248

315 312 336, 339 341, 350 335 335

218 258 244 247 -67 -35

266 273 258

a

Average uranyl bond length U-Ot, U-Os bond lengths to surface oxygen centers, U-Al distances to the nearest surface Al center, bond lengths U-Ow to aqua ligands, and average equatorial U-O bond length U-Oeq. b q (in e) is the charge of the adsorption site. Data for the penta-aqua complex of the uranyl ion are given for comparison. c The formal coordination number CN is larger than in the other complexes, where CN ) 5. d For CN ) 5, when the longest U-Ow distance is excluded (see text).

Figure 4. Optimized structure of uranyl adsorbed at the long-bridge uu AlO-AlO site solvated by two aqua ligands to preserve the coordination number 5. Coordination bonds from uranium to aqua ligands are shown as solid lines, hydrogen bonds to the surface as dashed lines (lengths in pm).

one may consider the U-Ow contact of 280 pm as too long for the first coordination shell, and thus exclude it from the average equatorial distance; then CN ) 5 and U-Oeq ) 249 pm. To inspect the coordination number of uranyl in this case, we also modeled a long-bridge adsorption complex with only two aqua ligands, coordinated in the equatorial plane of uranyl. The resulting optimized structure (Figure 4) exhibits shorter distances U-Al and U-Os to surface atoms, which indicate a stronger interaction with the surface due to the lower number of aqua ligands that compete for bonding to the substrate. The U-Ow bonds of that complex are not reduced because the two aqua ligands also interact with the surface via hydrogen bonds to neighboring surface OH groups (Figure 4). These features of the ligand shell may be interpreted as consequence of an incomplete first solvation shell of the uranyl ion. Surface solvation changes the orientation of the aqua ligands of the first coordination shell of uranyl and confirms CN ) 5; see section 4.2 and Figure S2c. The U-Al distance, accessible in experiment, is a pertinent structural characteristic for differentiating between adsorption at short-bridge and long-bridge sites. Uranyl complexes at longbridge sites AlO-AlO exhibit two similar U-Al distances in contrast to a single U-Al distance in complexes at short-bridge AlOO sites (Table 1). In the latter complexes, the U center is located approximately above an Al center, and the single U-Al

distance is 311 pm. At AlO-AlO sites uranyl adsorbs between two adjacent Al atoms with two longer U-Al contacts of ∼340 pm. The estimated formation energies of the two adsorption complexes at the short-bridge AlOO sites differ only slightly, by 11 kJ mol-1, in favor of the complex at the ul AlOO site. The preference of this complex likely is due to differences in the hydrogen bonds formed between aqua ligands of uranyl and surface OH groups. In the complex at the ul site, two aqua ligands of the first solvation shell form hydrogen bonds with O · · · H distances of 175 pm, whereas the adsorption complex at the uu site exhibits hydrogen bonds of 174 and 193 pm (Figure 2a,b). Not only the structures, also the formation energies ∆Eform of uranyl adsorption complexes at the uu and ul long-bridge sites vary more strongly. First of all, adsorption of uranyl at long-bridge sites is associated with larger energy changes ∆Eform than for complexes at short-bridges (Table 1). Formation of adsorption complexes at short-bridge sites requires ∼200 kJ mol-1 compared to ∼250 kJ mol-1 at long-bridge sites. We calculated the largest formation energy, ∆Eform ) 261 kJ mol-1, for six-coordinated uranyl at the ul AlO-AlO site, likely because the coordination sphere of uranyl is overcrowded. However, surface solvation, which so far is not taken into account, may affect these trends. The values ∆Eform at analogous short-bridge and long-bridge sites, both of uu or ul type, are too close to each other, within 30-60 kJ mol-1, to make definite conclusions on the relative size of the direct adsorption energies of uranyl. Recall that ∆Eform comprises also the deprotonation energies of the two surface OH groups besides the (direct) adsorption energy of uranyl; see the Supporting Information. The deprotonation of two close-lying OH groups at a short-bridge site will require more energy than the same process at a long-bridge site where the final O- groups are farther apart. Unfortunately, with the present models, these deprotonation energies cannot be estimated in a sufficiently reliable fashion. Uranyl Adsorption on Singly Deprotonated Sites AlOOH and AlO-AlOH. Table 1 also lists results for uranyl adsorption complexes at singly deprotonated short-bridge uu and ul AlOOH sites of the Al(o) (001) surface of kaolinite (at the GGA level). These two complexes exhibit similar structures and formation

Uranyl Adsorption on Kaolinite energies (Table 1), just as those obtained in our previous study at the LDA level.26 Each complex exhibits one short U-Os bond to the deprotonated surface O- center, ∼213 pm, similar to one of the U-Os bonds of the AlOO sites (Table 1). The second U-Os bond to the protonated OH group, ∼265 pm, is in the range of the U-Ow bonds, ∼260 pm. The U-Al distances at the AlOOH sites are ∼20 pm longer than at the AlOO sites, thus comparable to the distances in the complexes at the longbridge AlO-AlO sites. Adsorption complexes prepared at the long-bridge uu and ul AlO-AlOH sites, with uranyl placed slightly above the midpoint between the surface O- center and the OH group of the site, converged to complexes adsorbed at the nearest short-bridge AlOOH sites (see arrows, Figure 1a). Alternative initial structures of complexes at AlO-AlOH sites, constructed by adding a proton to one of the surface O centers of the optimized complex at the AlO-AlO site, also converged to complexes at short-bridge sites; for a description of all initial structures of the adsorption complexes at AlO-AlOH sites, see the Supporting Information. The uranyl moiety binds to the deprotonated surface O- center and rotates to find the closest hydroxyl group to interact with. Hence, adsorption on O-/OH groups of a basal kaolinite surface seems to occur in short-bridge mode only. The most stable structures of uranyl adsorbed at these sites are given in Table 1. Other complexes, at crystallographically nonequivalent sites of the same type, are presented as Supporting Information; these additional results illustrate the accuracy of our optimizations and the scattering of properties for various sites of the same type on the surface, which amounts to up to 2 pm for short bonds, up to 5 pm for long bonds and about 20 kJ mol-1 in the formation energies. The variety of these structures also reflects the complexity of the system under study. Analogous results were observed in a study on uranyl adsorption on the (001) gibbsite surface27 which is a hydroxylated octahedral surface, similar to the Al(o) surface of kaolinite. Therefore, it seems reasonable to compare computational results of uranyl adsorption on these two surfaces. As the basal (001) surface of gibbsite is assumed to be neutral at pH values between 3 and 10,53 where strong uranyl adsorption is observed,54 neutral sites have been explored OH/OH: one long-bridge and two shortbridge sites, all with uranyl adsorbed in bidentate fashion.27 During optimization the complex at the long-bridge AlOH-AlOH site converged to the neighboring short-bridge AlOHOH site.27 Thus, adsorption at protonated long-bridge sites also seems to be excluded. The study on the (001) surface of gibbsite27 did not find any complexes on long-bridge AlO-AlO sites as deprotonation of the sites was assumed to occur after adsorption. Therefore, optimization of uranyl complexes adsorbed at O-/ O- sites was started only from complexes on short-bridge sites.27 In summary, we found one new adsorption mode on the (001) octahedral surface of kaolinite, the doubly deprotonated longbridge AlO-AlO site which was calculated to be slightly less favorable than the corresponding short-bridge AlOO sites. In contrast to the short-bridge adsorption complexes, uranyl at AlO-AlO sites exhibits three contacts to the surface, two shorter and one longer one. This tridentate coordination at the surface may change the coordination number of uranyl at AlO-AlO sites from 5 to 6. Adsorption at singly deprotonated long-bridge AlO-AlOH sites was calculated to be unstable with the structure converging to a short-bridge complex. 4.2. Uranyl Adsorption on the Solvated Al(o) (001) Surface of Kaolinite. Surface SolWation. To approximate the main effects of solvation of the kaolinite surface on uranyl adsorption, we constructed models with an adsorbed monolayer

J. Phys. Chem. C, Vol. 114, No. 31, 2010 13291

Figure 5. Model of the Al(o) basal (001) surface of kaolinite with a (2 × 2) surface unit cell, and its solvation by a monolayer of water molecules. Upper panel shows side view, lower panel represents the top view of the solvated surface.

of water molecules, thus accounting explicitly for short-range solvation effects of the first solvation layer at the surface. Solvation effects had also been studied in a model of gibbsite using an overlayer of water molecules of low density.27 The average O-O distance of that overlayer, >450 pm, was notably longer than the average nearest neighbor O-O distance of bulk water, 285 pm.55 The (2 × 2) supercell model of the Al(o) (001) surface of kaolinite has a surface area of 1.88 nm2 which is almost completely covered by 20 water molecules (Figure 5). The average O-O distance within our model overlayer of water is 280 pm. Optimization yields an ordered structure of the water overlayer, similar to the results obtained in an earlier DFT MD study (PW91)56 where nine adsorbed water molecules were used in a (2 × 1) unit cell. That study also included a calculation on a larger model with 36 water molecules in a (4 × 2) unit cell that yielded the same ordering of the water overlayer.56 In the water monolayer at the Al(o) surface of kaolinite obtained in the present work (Figure 5), each water molecule forms two hydrogen bonds to its neighbors and one to the surface. There are two types of hydrogen bonds to the surface. One features a surface oxygen as electron donor and a hydrogen of a water molecule as electron acceptor, with an Osurf · · · H distance of ∼170 pm. In the second type of hydrogen bonds an oxygen center of a water molecule acts as electron donor; these bonds are longer, 185-190 pm, hence weaker. These two types of hydrogen bonds to the surface were described in an earlier ab initio MD study56 where the most probable Osurf · · · H distance was ∼10 pm shorter than the Ow · · · Hsurf distance, in qualitative agreement with our results. Uranyl Adsorption at the SolWated Surface. To estimate solvation effects, we modeled the adsorption of uranyl complexes, including the three aqua ligands of their first solvation shell, by covering the surface with the same number of water molecules as in the models of the bare surface. We examined adsorption complexes at uu and ul sites of the types AlOO, AlOOH, AlO-AlO, and AlO-AlOH, located on the solvated Al(o) (001) basal surface of kaolinite. By modeling the solvation of the surface with the same number of solvent molecules, for a surface free of adsorbates and with uranyl complexes, the results are expected to be comparable. The initial structures for

13292

J. Phys. Chem. C, Vol. 114, No. 31, 2010

the optimization of uranyl adsorption complexes were constructed as follows: (i) starting with the optimized structure of the solvated surface, the water monolayer was raised sufficiently high above the surface, (ii) the surface was appropriately deprotonated to create a specific surface site to which the adsorbate [UO2(H2O)3]2+ was added, (iii) four to six water molecules of the monolayer were shifted further upward to form a “cavity” for accommodating uranyl, and (iv) the rearranged solvation layer was lowered to its original height to cover the surface together with the adsorbate. The lower part of Table 1 provides pertinent structure parameters of the various adsorption complexes of uranyl as well as their formation energies. As complexes at the uu sites are more stable than complexes at ul sites, we provide the solvated structures of the former as Supporting Information (Figure S2). The complex initially constructed as adsorbed on the longbridge AlO-AlOH site (for the preparation procedure, see section 4.1) again converted to a complex at a short-bridge site during structure optimization. Thus, surface solvation does not seem to stabilize uranyl adsorption at singly deprotonated long-bridge sites. The present extended models confirm that a bidentate adsorption complex of uranyl at O/OH groups is stable only when these two groups are coordinated at the same Al center, i.e., at a short-bridge site. Accordingly, the following discussion is restricted to complexes at short-bridge sites and doubly deprotonated long-bridge sites. The U-Ot bonds of all optimized adsorption complexes are only weakly affected by solvation. The terminal uranyl bonds elongate at most by 1 pm due to solvation as a consequence of additional hydrogen bonds of uranyl oxygen centers with the water molecules of the surrounding. U-Os bonds at AlOO sites are evened out by solvation: the shorter contacts slightly elongate while the longer ones are somewhat reduced. Accordingly, the difference of the two bonds shrinks from ∼7 pm to at most 3 pm (Table 1). The U-Os bonds to surface O- centers of complexes at AlOOH sites increase slightly due to solvation, whereas the U-Os bonds to the neutral OH groups shorten by 1-2 pm. The U-Al distances in complexes at the short-bridge sites are elongated by 1-4 pm whereas they hardly change in complexes at long-bridge AlO-AlO sites. U-Ow bonds to the aqua ligands were shortened by solvation and averages U-Oeq are reduced by 1-3 pm in all complexes studied. Recall that the conductor-like screening model of solvation (COSMO) yields a similar shortening of U-Oeq of 2 pm for complexes of uranyl in solution.57,58 Overall, surface solvation modeled by a monolayer of water molecules does not affect the geometry of the adsorption complexes of uranyl in a substantial way. The pertinent structure parameters preserve all major trends as discussed for the adsorption complexes at the bare Al(o) surface of kaolinite. The complex at the uu AlO-AlO site exhibits only two close contacts of uranium to aqua ligands in the first solvation shell (Figure S2c). Thus, in this complex uranyl forms three bonds to surface O centers and only two bonds to the aqua ligands (Figure S2c). Hence, the model with surface solvation confirms the 5-fold coordination of uranyl at the long-bridge uu site. Interestingly, the adsorption complex at the ul AlO-AlO site preserved its coordination, CN ) 6, even in the model with solvation, as it exhibits three contacts to the surface in addition to the three aqua ligands of its first coordination shell (Table 1). The formation energies of the two complexes at the uu and ul AlO-AlO sites are essentially the same, differing by 3 kJ mol-1 only. The diverse coordination numbers, CN ) 5 at the

Martorell et al. uu AlO-AlO site and CN ) 6 at the ul AlO-AlO site, are reflected in the averages U-Oeq: 243 pm (CN ) 5) and 249 pm (CN ) 6), respectively. This increase of U-Oeq agrees with earlier observations regarding a correlation between the coordination number of uranyl and U-Oeq values for complexes in aqueous solution due to bonding competition of the ligands.57,59 The formation energies of the adsorption complexes in general are larger in the models with solvation. For complexes at the uu sites (AlOO, AlO-AlO, and AlOOH), minor increments are calculated, up to 12 kJ mol-1 (Table 1). Complexes at ul sites are unevenly affected by the addition of the monolayer of water molecules. At the short-bridge ul sites they are destabilized, on average by ∼50 kJ mol-1. Interestingly, the complex at the ul AlO-AlO site (CN ) 6) is stabilized due to surface solvation: its formation energy is reduced by 14 kJ mol-1. Overall uu adsorption sites seem to be more favorable than the corresponding ul sites; hence our selection of structures in Figure S2 of the Supporting Information. However, recall that the formation energies provide only an approximate measure of stability for the various complexes. Including structural variations of the water overlayer modeling surface solvation, these energies vary within 40 kJ mol-1 for complexes at sites with the same charge (Table 1). Therefore, definitive conclusions on energy preferences are not possible without an improved representation of surface solvation. In summary, solvation as modeled in this work does not change the general qualitative trends of properties of innersphere complexes of uranyl at the Al(o) (001) surface of kaolinite: U-Al and U-Oeq distances decrease with the charge q (or the degree of deprotonation) of the adsorption site (Table 1).26 Also, solvation does not stabilize uranyl adsorption at singly deprotonated long-bridge AlO-AlOH sites. Rather, adsorption of uranyl at long bridges is stable only for the doubly deprotonated AlO-AlO sites. On the other hand, the addition of a monolayer of water led to a clarification of the coordination number of uranyl at the uu AlO-AlO site. In contrast to a bare surface, adsorption complexes at long-bridge and short-bridge sites of a solvated surface exhibit similar stabilities. 5. Comparison with Experiment Next, we will compare our results with available experimental data, mainly from EXAFS studies that provide average structure parameters of the adsorption complexes of uranyl at surfaces of kaolinite.20,30,31 In this comparison, we will only consider results from models on the solvated Al(o) surfaces of kaolinite. The basal Al(o) (001) surface can be viewed as result of an ideal cleavage of kaolinite; it dominates for crystallites in solution although a small fraction (∼20%) of kaolinite edge surfaces [oriented in (010) or (110) and related directions]60 may also be available for adsorption. Nevertheless, we do not consider adsorption on edge surfaces in the present work. Recall that all adsorption complexes inspected were optimized at the GGA level where in general bond distances are determined longer than at the LDA level; yet, experience shows that LDA structures as a rule agree well with experiment.54,61,62 Thus, to achieve a more direct comparison with experiment, we will correct the structures obtained at the GGA level for differences to the corresponding LDA results. To estimate these corrections, we compared in detail structural parameters of various adsorption complexes optimized both at the LDA and the GGA levels; see Table S2 of the Supporting Information. Thus, at the GGA level, uranyl U-Ot bonds typically are calculated too long by 2 pm and the average equatorial U-Oeq distance is overestimated by 8 pm. The U-Al distance, although not associated

Uranyl Adsorption on Kaolinite

J. Phys. Chem. C, Vol. 114, No. 31, 2010 13293

TABLE 2: EXAFS Results in Comparison with Calculated Characteristic Distancesa (in pm) and Formation Energies ∆Eform (in kJ mol-1) of Uranyl Adsorption Complexes at Short-Bridge (AlOO and AlOOH) and Long-Bridge (AlO-AlO) Sites of the Solvated (001) Al(o) Kaolinite Surfaceb model

q

U-Ot

U-Oeq

U-Al

∆Eform

uu AlOO ul AlOO uu AlO-AlO ul AlO-AlO, CN ) 6 uu AlOOH ul AlOOH exp. CO2, pH 5c CO2, pH 7d CO2, pH 7e N2, pH 8.5c

-2

185 185 186 183 185 184

235 236 235 241 241 240

309 306 330, 333 335, 344 329 329

218 258 244 247 -67 -35

180 180 180 180

237 234 240 236

310, 330 306, 326 ∼330 309, 329

-1

a Bond lengths corrected for GGA overestimation (see text) by -2 pm for U-Ot, -8 pm for U-Oeq, and -6 pm for U-Al. For details see text and the Supporting Information. b q (in e) is the charge of the adsorption site. c Reference 30. d Reference 31. e Reference 20.

with a bond, is overestimated on average by 6 pm. These approximate corrections will be taken into account in the following. To facilitate the discussion, we compare in Table 2 corrected GGA results for adsorption complexes at the solvated Al(o) surface with available experimental data. Terminal uranyl bonds U-Ot are consistently determined somewhat longer, 183-186 pm, than the experimental value, 180 pm (Table 2). This overestimation can be noted in all available slab-model calculations on uranyl adsorption at mineral surfaces.24,26,27 It was rationalized in part by a large transfer of electron density from the surface, in part by additional hydrogen bonds formed between terminal O centers of uranyl and surface OH groups.27,45 However, if not a model artifact, such hydrogen bonds should also be present in experiment. Shorter experimental values of U-Ot bonds can in part be rationalized by the coexistence of inner- and outer-sphere adsorption complexes which exhibit longer, ∼185 pm, and shorter, ∼180 pm, uranyl bonds, respectively.24,26,45 Moreover, the simultaneous presence of inner- and outer-sphere adsorption complexes was recently observed in experiments on the adsorption of various cations at muscovite surfaces.32 Thus, a fully satisfactory rationalization of the notable overestimation of U-Ot bonds in computational models currently seems to be lacking. The calculated average equatorial U-O distances U-Oeq of 235-241 pm agree well with the experimental results, 234-240 pm (Table 2). The calculated U-Oeq values fall into two groups. Smaller results, 235-236 pm, are obtained for complexes at doubly deprotonated AlOO and AlO-AlO sites. Larger results, 240-241 pm, are associated with singly deprotonated sites AlOOH and complexes at ul AlO-AlO sites with CN ) 6. Thus, in the present models, the calculated U-Oeq values do not only reflect the coordination number of uranyl, but also the charge q of the adsorption site. U-Oeq values increase with the coordination number and decrease with the charge q of the adsorption site. Unfortunately, the experimental U-Oeq values do not seem to show a clear trend with pH. Note that the coordination number is not varied when evaluating the EXAFS data; CN ) 5 is assumed in all cases.20,30,31 However, according to our models, CN ) 6 may occur in adsorption complexes at long-bridge sites where uranyl exhibits a third contact to the surface. Also, experiments thus far are not able to unravel at which sites uranyl adsorbs.

U-Al distances can only be compared to experiment in approximate fashion as they do not reflect bonds and thus will be more sensitive to details of the models or to surface defects. EXAFS resolves two U-Al/Si distances, 306-310 and 326-330 pm.30,31 Similar U-Al/Si distances were estimated for an edgesharing adsorption geometry of uranyl at Al(O,OH)6 octahedra and Si(O,OH)4 tetrahedra.9,30 Thus, bidentate adsorption at edgesharing sites was claimed.9,30 The short-bridge sites studied in the present work are edge sharing sites as uranyl shares the edges of the Al octahedra, while the long-bridge sites are not edge sharing. The calculated U-Al distances in general agree well with the experimental data (Table 2). In our model results the shorter U-Al distance, 306-309 pm, is associated with bidentate complexes at short-bridge AlOO edge-sharing sites (Table 2). For adsorption complexes at long-bridge AlO-AlO sites and at AlOOH sites U-Al distances around 330 pm and higher (up to 344 pm) were calculated. Note that the long-bridge AlO-AlO site is not edge sharing. Thus, we propose to assign the two experimentally determined U-Al distances to different adsorption complexes. Two types of adsorbed species were also found in TRLFS experiments, but a different interpretation was suggested.18 A peculiarity of our models of long-bridge sites is the coordination number of U-Al, which is 2 instead of 1 as in complexes at the AlOOH sites. As mentioned, the coordination numbers were fixed when analyzing the EXAFS data, namely CN ) 1.0 for U-Al.30,31 One also has to account for sites present at the edge surfaces of kaolinite,63 not addressed in the present study. Edge surfaces are considered more reactive,64 and adsorption complexes on them may turn out to contribute notably to experimental averages. Formally sites of different charges, q ) 1 or 2 e, correspond to different pH regimes. At a pH close to the pH of zero-point charge (pHZPC), the surface is neutral and doubly deprotonated sites, q ) 2 e, should be rare. With increasing pH, first the density of singly charged sites, q ) 1 e, will increase. At higher pH, doubly deprotonated sites, q ) 2 e, will appear and eventually will replace singly charged sites. Therefore, one can expect AlOOH sites to dominate at lower pH, slightly above pHZPC, and an increasing fraction of AlOO and AlO-AlO sites at increasing pH. At elevated pH also the number of possible adsorption sites grows, as the long-bridge sites become available. In principle, the resulting structure trends should be reflected in experiment where the pH is varied. Therefore, according to our model calculations, U-Oeq values are predicted to shorten with increasing pH; also the effective coordination number corresponding to the U-Al contacts may increase above 1 due to the presence of complexes at long-bridge sites. On the other hand, even at low pH, the adsorption of uranyl on singly deprotonated AlOOH sites may induce the deprotonation of the nearby surface OH groups and result in an adsorption complex at an AlOO site. In this way, one can rationalize that EXAFS results do not exhibit a trend with pH. In addition, recent RAXR experiments provide evidence that inner- and outer-sphere complexes can coexist on mineral surfaces which makes the adsorption behavior even more complex and structure averages less clear-cut to interpret.32 Finally, one should recall the model character of the present work which did not address adsorption at edge sites. Yet, such sites may play an important role for a quantitative description of uranyl adsorption at kaolinite. Results of a study on uranyl adsorption at kaolinite edge surfaces will be reported elsewhere. Acknowledgment. This work was supported by the German Bundesministerium fu¨r Wirtschaft und Technologie (Grant No. 02E10186) and Fonds der Chemischen Industrie (Germany).

13294

J. Phys. Chem. C, Vol. 114, No. 31, 2010

Supporting Information Available: Information regarding the determination of formation energies ∆Eform. Results for adsorption complexes at various singly deprotonated sites. Comparison of structural parameters of adsorption complexes optimized at LDA and GGA levels of theory. Optimized structures of uranyl adsorption complexes on the solvated Al(o) basal (001) surface of kaolinite. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Thompson, B. M.; Longmire, P. A.; Brookings, D. G. Appl. Geochem. 1986, 1, 335–343. (2) Adsorption of metals by geomedia: Variables, mechanisms and model applications; Jenne, E. A., Ed.; Academic Press: London, 1998. (3) Adsorption of metals by geomedia II: Variables, mechanisms, and model applications; Barnett, M., Kent, D., Eds.; Elsevier: New York, 2008. (4) Kristallin-I, Safety Assessment Report; Nagra Technical Reports NTB 93-22E; National Cooperative for the Disposal of Radioactive Waste: Wettingen, Switzerland, 1994. (5) Hoth, P.; Wirth, H.; Reinhold, K.; Bra¨uer, V.; Krull, P.; Feldrappe, H. Endlagerung radioaktiVer Abfa¨lle in tiefen geologischen Formationen Deutschlands; Bundesanstalt fu¨r Geowissenschaften und Rohstoffe BGR: Hannover, Germany, 2007; p 118. (6) Hartmann, E.; Geckeis, H.; Rabung, Th.; Lu¨tzenkirchen, J.; Fangha¨nel, Th. Radiochim. Acta 2008, 96, 699–707. (7) Sachs, S.; Bernhard, G. Chemosphere 2008, 72, 1441–1447. (8) Krˇepelova´, A.; Sachs, S.; Bernhard, G. Radiochim. Acta 2006, 94, 825–833. (9) Hennig, C.; Reich, T.; Da¨hn, R.; Scheidegger, A. M. Radiochim. Acta 2002, 90, 653–657. (10) Buda, R. A.; Banik, N. L.; Kratz, J. V.; Trautmann, N. Radiochim. Acta 2008, 96, 657–665. (11) Fernandes, M. M.; Baeyens, B.; Bradbury, M. H. Radiochim. Acta 2008, 96, 691–697. (12) Bertetti, F. P.; Pabalan, R. T.; Almendarez, M. G. In Adsorption of metals by geomedia; Jenne, E. A., Ed.; Academic Press: London, 1998; pp 132-150. (13) Migration of actinides in the system clay, humic substances, aquifer; Wissenschaftliche Berichte FZKA 7407; Forschungszentrum Karlsruhe GmbH: Karlsruhe, Germany, 2008. (14) Project Opalinus Clay, Safety Report; Nagra Technical Reports NTB 02-05; National Cooperative for the Disposal of Radioactive Waste: Wettingen, Switzerland, 2002. (15) Safety of Geological Disposal of high-leVel and long-liVed radioactiVe waste in France; An international Peer Review of the “Dossier 2005 Argile” concerning disposal in the callovo-oxfordian formation. NEA (Nuclear Energy Agency) No. 6178, 2006; ISBN 92-64-02299-6. (16) Baumann, N.; Brendler, V.; Arnold, T.; Geipel, G.; Bernhard, G. J. Colloid Interface Sci. 2005, 290, 318–324. (17) Chisholm-Brause, C. J.; Berg, J. M.; Little, K. M.; Matzner, R. A.; Morris, D. E. J. Colloid Interface Sci. 2004, 277, 366–382. (18) Krˇepelova´, A.; Brendler, V.; Sachs, S.; Baumann, N.; Bernhard, G. EnViron. Sci. Technol. 2007, 41, 6142–6147. (19) Sylwester, E. R.; Hudson, E. A.; Allen, P. G. Geochim. Cosmochim. Acta 2000, 64, 2431–2438. (20) Thompson, H. A.; Parks, G. A.; Brown, G. E., Jr. In Adsorption of metals by geomedia; Jenne, E. A., Ed.; Academic Press: London, 1998; pp 350-371. (21) Den Auwer, C.; Simoni, E.; Conradson, S.; Madic, C. Eur. J. Inorg. Chem. 2003, 21, 3843–3859. (22) Payne, T.; Lumpkin, G. R.; Waite, T. D. In Adsorption of metals by geomedia; Jenne, E. A., Eds.; Academic Press: London, 1998; pp 7599. (23) Arai, Y.; McBeath, M.; Bargar, J.; Joye, J.; Davis, J. A. Geochim. Cosmochim. Acta 2006, 70, 2492–2509. (24) Moskaleva, L. V.; Nasluzov, V. A.; Ro¨sch, N. Langmuir 2006, 22, 2141–2145. (25) Greathouse, J. A.; Cygan, R. T. EnViron. Sci. Technol. 2006, 40, 3865–3871. (26) Kremleva, A.; Kru¨ger, S.; Ro¨sch, N. Langmuir 2008, 24, 9515– 9524. (27) Shuller L. C.; Poling J.; Ewing R. C.; Becker U. Goldschmidt Conf. Abst. A864, 2008. (28) Roques, J.; Veilly, E.; Simoni, E. Int. J. Mol. Sci. 2009, 10, 2633– 2661.

Martorell et al. (29) The chemistry of actinide and transactinide elements; Morss, L. R., Edelstein, N. M., Fuger, J., Katz, J. J., Eds.; Springer: Dordrecht, The Netherlands, 2006. (30) Reich, T.; Reich, T. Ye.; Amayri, S.; Drebert, J.; Banik, N. L.; Buda, R. A.; Kratz, J. V.; Trautmann, N. AIP Conf. Proc. 2007, 882, 179– 183. (31) Krˇepelova´, A.; Reich, T.; Sachs, S.; Drebert, J.; Bernhard, G. J. Colloid Interface Sci. 2008, 319, 40–47. (32) Park, C.; Fenter, P. A.; Sturchio, N. C.; Nagy, K. L. Langmuir 2008, 24, 13993–14004. (33) Brady, P. V.; Cygan, R. T.; Nagy, K. L. J. Colloid Interface Sci. 1996, 183, 356–364. (34) (a) Kresse, G.; Furthmu¨ller, J. Phys. ReV. B 1996, 54, 11169–11186. (b) Kresse, G.; Hafner, J. Phys. ReV. B 1994, 49, 14251–14269. (c) Kresse, G.; Hafner, J. Phys. ReV. B 1993, 47, 558–561. (d) Kresse, G.; Hafner, J. Phys. ReV. B 1993, 48, 13115–13118. (e) Kresse, G.; Furthmu¨ller, J. Comput. Mater. Sci. 1996, 6, 15–50. (35) Perdew, J. P.; Wang, Y. Phys. ReV. B 1992, 45, 13244–13249. (36) Sim, F.; St-Amant, A.; Papai, I.; Salahub, D. R. J. Am. Chem. Soc. 1992, 114, 4391–4400. (37) Blo¨chl, P. E. Phys. ReV. B 1994, 50, 17953–17979. (38) Kresse, G.; Joubert, D. Phys. ReV. B 1999, 59, 1758–1775. (39) MacDonald, A. H.; Vosko, S. H. J. Phys. C 1979, 12, 2977–2990. (40) Monkhorst, H. J.; Pack, J. D. Phys. ReV. B 1976, 13, 5188–5192. (41) Methfessel, M.; Paxton, A. T. Phys. ReV. B 1989, 40, 3616–3621. (42) Young, R. A.; Hewat, A. W. Clays Clay Miner. 1988, 36, 225– 232. (43) Bish, D. L. Clays Clay Miner. 1993, 41, 738–744. (44) Benco, L.; Tunega, D.; Hafner, J.; Lischka, H. Am. Mineral. 2001, 86, 1057–1065. (45) Kremleva, A. Environmental chemistry of uranyl: a relativistic density functional study on complexation with humic substances and sorption on kaolinite. Ph.D. Thesis, Technische Universita¨t Mu¨nchen, Munich, 2009. (46) Vallet, V.; Moll, H.; Wahlgren, U.; Szabo´, Z.; Grenthe, I. Inorg. Chem. 2003, 42, 1982–1993. (47) Denecke, M. A.; Reich, T.; Pompe, S.; Bubner, M.; Heise, K. H.; Nitsche, H.; Allen, P. G.; Bucher, J. J.; Edelstein, N. M.; Shuh, D. K. J. Phys. IV 1997, 7, 637–638. (48) Vallet, V.; Macak, P.; Wahlgren, U.; Grenthe, I. Theor. Chem. Acc. 2006, 115, 145–160. (49) Spencer, S.; Gagliardi, L.; Handy, N. C.; Ioannou, A. G.; Skylaris, C. K.; Willets, A.; Simper, A. M. J. Phys. Chem. A 1999, 103, 1831–1837. (50) Vienna Ab initio Simulation Package Website. http://cms.mpi. univie.ac.at/vasp/vasp/node152.html. (51) Lim, K. H.; Shor, A. M.; Zakharieva, O.; Ro¨sch, N. Chem. Phys. Lett. 2007, 444, 280–286. (52) Stumm, W. Chemistry of the solid-water interface. Processes at the mineral-water and particle-water interface in natural systems; Wiley: New York, 1992. (53) Hiemstra, T.; Yong, H.; Van Riemsdijk, W. H. Langmuir 1999, 15, 5942–5955. (54) Hattori, T.; Saito, T.; Ishida, K.; Scheinost, A. C.; Tsuneda, T.; Nagasaki, S.; Tanaka, S. Geochim. Cosmochim. Acta 2009, 73, 5975–5988. (55) (a) Yokoyama, H.; Kannami, M.; Kanno, H. Chem. Phys. Lett. 2008, 463, 99–102. (b) Narten, A. H.; Levy, H. A. J. Chem. Phys. 1971, 55, 2263– 2269. (56) Tunega, D.; Gerzabek, M. H.; Lischka, H. J. Phys. Chem. B 2004, 108, 5930–5936. (57) Schlosser, F.; Kru¨ger, S.; Ro¨sch, N. Inorg. Chem. 2006, 45, 1480– 1490. (58) Vallet, V.; Wahlgren, U.; Schimmelpfennig, B.; Moll, H.; Szabo´, Z.; Grenthe, I. Inorg. Chem. 2001, 40, 3516–3525. (59) Kru¨ger, S.; Schlosser, F.; Ray, R. S.; Ro¨sch, N. In Lecture series on computer and computational sciences; Simos, T., Maroulis, G., Eds; Brill: Netherlands, 2006; Vol. 7, pp 904-907. (60) Kameda, J.; Yamagishi, A.; Kogure, T. Am. Mineral. 2005, 90, 1462–1465. (61) Ziegler, T. Chem. ReV. 1991, 91, 651–667. (62) Go¨rling, A.; Trickey, S. B.; Gisdakis, P.; Ro¨sch, N. In Topics in organometallic chemistry; Brown, J., Hoffmann, P., Eds.; Springer: Heidelberg, Germany, 1999; Vol. 4, pp 109-165. (63) Bickmore, B. R.; Rosso, K. M.; Nagy, K. L.; Cygan, R. T.; Tadainer, C. J. Clays Clay Miner. 2003, 51, 359–371. (64) Bickmore, B. R.; Bosbach, D.; Hochella, M. F.; Charlet, L.; Rufe, E. Am. Mineral. 2001, 86, 411–423.

JP101300W