Zinc Adsorption on Clays Inferred from Atomistic Simulations and

Apr 26, 2012 - regulation.1 Excess Zn at polluted sites can be taken up by roots .... Applying the Kröger−Vink notation for point defects,16 these ...
27 downloads 0 Views 381KB Size
Article pubs.acs.org/est

Zinc Adsorption on Clays Inferred from Atomistic Simulations and EXAFS Spectroscopy Sergey V. Churakov* and Rainer Daḧ n Laboratory for Waste Management, Paul Scherrer Institute, CH-5232, Villigen-PSI, Switzerland S Supporting Information *

ABSTRACT: Clay minerals are efficient sinks for heavy metals in the geosphere. Knowing the uptake mechanism of these elements on clays can help to protect the natural environment from industrial pollution. In this study ab initio molecular dynamics (MD) calculations were applied to simulate the uptake of Zn on the edge surfaces of montmorillonite, a dioctahedral clay, and to explain the measured K-edge extended X-ray absorption fine structure (EXAFS) spectra of adsorbed Zn. These experiments were carried out using a high ionic strength Na background electrolyte that enables one to block cation exchange processes and to restrict the Zn uptake to the sorption complexation at the edge sites of clay. The analysis of the experimental data and simulation results suggest that structurally incorporated Zn preferentially substitutes for Al(III) in the trans-symmetric sites of the octahedral layer. At low loading, Zn is incorporated into the outermost trans-octahedra on (010) and (110) edges. At medium loading, Zn forms mono- and bidentate inner-sphere surface complexes attached to the octahedral layer of (010) and (110) edge sites. The maximal site density of innersphere sorption sites inferred from molecular simulations agrees well with site capacities of surface complexation sites derived from macroscopic studies and modeling.



INTRODUCTION At small concentrations Zn is an indispensible component of biochemical processes and cellular activity in the human body. However, long-term consumption of Zn above the recommended dietary allowance may influence the epigenetic gene regulation.1 Excess Zn at polluted sites can be taken up by roots and plants and finally enter the nutrition chain of human beings. The concentrations of total zinc in soils are normally 50 mmol/kg. Thus, the understanding of the mechanisms controlling, for example, the zinc migration in the geosphere is of great importance for the protection of the natural environment. Clay minerals such as illite and montmorillonite are ubiquitous in the environment. Because of their large specific surface area and high structural charge, they control the migration of heavy metals in the geosphere via different uptake mechanisms. One of the major processes for the sequestration of trace concentrations of heavy metals is sorption to clay edge sites and incorporation into clay structures. Whereas sorption is a fast process occurring nearly instantaneously, the incorporation of metal ions into clay minerals occurs over geological time scales. Furthermore, Zn is a divalent transition metal, which shows similar chemical behavior to Ni and Co and can thus also be considered as a natural analogue for radioactive Ni and Co arising from nuclear fuel and radioactive waste from the decommissioning of nuclear power plants. The release of radionuclides from a repository can be considerably retarded © 2012 American Chemical Society

due to interactions with clay minerals. For example, bentonite containing >70 wt % dioctahedral aluminosilicate clays is foreseen as a backfill material in the Swiss concept for a highlevel radioactive waste repository.3 Batch sorption experiments and modeling have suggested that the uptake mechanisms of divalent transition metals on the edge sites of dioctahedral clays is different at low and medium loading.4 A recent extended X-ray absorption fine structure (EXAFS) study has established that surface complexes formed by transition metals have different structures at low and medium loadings.5 Fully resolved structural characteristics of these complexes could not be achieved due to the fact that EXAFS spectroscopy measures a cumulative signal that can potentially arise from different surface sites. An unambiguous discrimination of such complexes is only possible with additional structural information obtained on an atomistic level. Ab initio MD calculations applied here to model the uptake of Zn on edge surfaces of dioctahedral clays and to interpret the K-edge EXAFS measurements of Zn absorbed on montmorillonite offer a molecular-level description of sorption processes of divalent transition metals on clay minerals. To the best of our knowledge this is the first study on the uptake of divalent metal Zn on clay mineral surfaces unifying wet chemistry data, EXAFS measurements, and ab initio calculations. Received: Revised: Accepted: Published: 5713

December 10, 2011 March 26, 2012 April 26, 2012 April 26, 2012 dx.doi.org/10.1021/es204423k | Environ. Sci. Technol. 2012, 46, 5713−5719

Environmental Science & Technology



Article

METHODS Sample Preparation and Sorption Experiments. Two different montmorillonites were chosen for this study: Milos (Island of Milos, Greece; Sü d-Chemie AG, Moosburg, Germany) and STx-1 (Gonzales County, TX; Source Clay Minerals Repository). Milos montmorillonite was taken because it contains 1.8 ± 0.15 mmol/kg of Zn, an amount that can be measured with EXAFS in reasonable time (Table 1). This incorporated Zn was used as an internal standard to

pseudohexagonally ordered layer of octahedra, sandwiched between two tetrahedral siloxane layers. Two thirds of the cation sites in the octahedral layer have cis-symmetry and onethird trans-symmetry with respect to the OH groups. Only twothirds of the octahedra are occupied in montmorillonite, opening a possibility for two distinct structural models. In one model, all trans-octahedra are occupied and half of the cistetrahedra are vacant (cv-model). In the second model, all trans-octahedra are vacant and cis-tetrahedra are occupied (tvmodel). Natural montmorillonites are found in both trans- and cis-vacant forms depending on the nature of isomorphic substitutions (tetrahedral or octahedral) and the concentration of the substituted ions.13−15 Pyrophyllite, the nonswelling clay often used as the simplest structural prototype for smectites, occurs in the tv-structure only. To validate the modeling approach, the Zn complexes incorporated into the montmorillonite sheets were simulated first. Two incorporation mechanisms were considered. Applying the Kröger−Vink notation for point defects,16 these are substitution in Al octahedra, ZnAl ′ , and incorporation in the octahedral vacancies, Zn•• Al . Taking into account the existence of cis and trans sites, five different defect types are possible: cv‑mont •• Zn trans‑Al ′ , cv‑mont Zn cis‑Al ′ , tv‑mont Zn cis‑Al ′ , cv‑mont Zn cis‑v , tv‑mont •• Zntrans‑v. These simulations were performed for a single clay particle with the composition [Zn′Al×Al7Si16O40(OH)16]− 2+ and [Zn•• without water in the interAl ×Al8Si16O40(OH)16] layer. The cell parameters remained fixed at experimental values throughout the simulations, and only atomic coordinates were allowed to relax. The aim of this investigation was (a) to find out whether Zn substitutes for Al (trans or cis sites) or occupies the vacancies (trans or cis sites) and (b) to determine how sensitive molecular modeling calculations combined with ab initio EXAFS calculations are to small changes in the local environment of the Zn atoms in montmorillonite. In a next step, adsorption processes of Zn were modeled taking into account the (010), (110), (130), and (100) edge surfaces for both tv and cv forms of montmorillonite. The composition of the system in these simulations was K2Mg2Al10Si24O60(OH)12×56H2O×Zn2+. For the sake of computational efficiency, the interlayer water between basal planes of clay particles was neglected and only water molecules confined between the edges of clay particles were considered. For the system without interlayer water, we used K as a charge compensation counterion, as it corresponds to the natural clay illite. The cell dimension in the c-direction was fixed to 10 Å, which corresponds to the typical interlayer distance for dehydrated smectites and illites.17 In the setup used, the interlayer K ions do not influence the structure of edge sites and are more then 7 Å away from the Zn ion in the simulation runs. The EXAFS spectroscopy is not capable of measuring the contribution due to K atoms at such a distance. Thus, the modeling setup with K in the interlayers in place of Na is justified for the purpose of this study. The cell parameters in the direction orthogonal to the edge were adjusted by performing classical simulations in isothermal−isobaric ensemble at 1 bar external pressure and temperature equal to 300 K using the CLAYFF force field.18 The optimized dimensions of the supercell for the simulations with the periodic boundary conditions were 10.44 × 28.5 × 10.00 Å and 9.03 × 32.70 × 10.00 Å for (010), (110) and (100), (130) edge surfaces, respectively. These lattice parameters were kept fixed throughout the ab initio simulations.

Table 1. Zn Loadings for the Samples Prepared for the EXAFS Studies in 0.2 NaClO4 Sample

Zn in structure (mmol/kg)

Zn adsorbed (mmol/kg)

Milos-incorporated STx-1-medium STx-1-low Milos-low

1.8 0.4 0.4 1.8

32.0 2.3 2.7

verify the robustness of the data analysis. In contrast, STx-1 was taken because it contains only 0.3 ± 0.1 mmol/kg of Zn (Table 1), an amount that cannot be measured in an Fe-rich clay matrix (Milos and STx-1 contain 1.26 wt % and 0.56 wt % Fe, respectively). Compared to the foreseen added Zn loadings of ∼2 and ∼30 mmol/kg, respectively, the quantity of incorporated Zn in the STx-1 is quite low and thus does not impair the EXAFS signal form the adsorbed species. In a previous study we could demonstrate that under the employed experimental conditions a correction for the incorporated species in the EXAFS signals was not necessary.5 Uptake experiments were carried out at a high ionic strength of Na background electrolyte. Under such a condition the basal planes and the interlayer of clay are saturated with sodium and the Zn adsorbs almost exclusively at the edge sites. The Zn uptake experiments were conducted at room temperature in a glovebox under N2 atmosphere (CO2 and O2 < 5 ppm). After a 14 day reaction time, 40 mL aliquots from the strongly stirred suspensions were slowly filtrated through 47 mm diameter filters (Millipore, 0.4 μm pore size) to prepare highly oriented self-supporting films. The filtrations were performed in a closed vessel under a continuous flow of argon. Excess of solution in the wet films was removed by washing with a few milliliters of deionized water before air-drying at room temperature. Previous studies have shown that highly textured selfsupporting films are obtained with the same protocol.6−11 The filtered solutions were analyzed for Zn, Si, and Al by ICPOES (Table S1, Supporting Information). The obtained Zn loadings are summarized in Table 1. Further details on the conditioning process can be found in the Supporting Information. EXAFS Measurements. The Zn K-edge EXAFS spectra were recorded at the Dutch Belgium Beamline (DUBBLE/ BM26) at the European Synchrotron Radiation Facility (ESRF), Grenoble, France.12 All spectra were recorded at room temperature using a Si(111) monochromator and a ninechannel monolithic Ge solid-state detector. Higher order harmonics were rejected with two mirrors (Si and Pt/Si). The monochromator angle was calibrated by assigning the first inflection point of the K-absorption edge spectrum of Zn metal to 9659 eV. Several scans were averaged to improve the signalto-noise ratio. Modeling Setup. Montmorillonite belongs to the family of 2:1 dioctahedral clays, which can be imagined as a 5714

dx.doi.org/10.1021/es204423k | Environ. Sci. Technol. 2012, 46, 5713−5719

Environmental Science & Technology

Article

Ab Initio Calculations. All ab initio calculations were performed on the basis of the density functional theory using the Gaussian and augmented plane waves (GAPW) method as implemented in the CP2K simulation package.19,20 The technical details of the simulation setup are available in the Supporting Information. The accuracy of the simulation parameters was tested against all electron calculations for small molecular clusters, previous pseudopotential calculations of the pyrophyllite structure,21−24 and available experimental data. The geometries were reproduced within 1.5%. The uptake of aqueous ions on mineral surfaces is an activated process that cannot be easily accessed in a direct ab initio simulation. To promote the ion adsorption and to monitor the structural transformation in the Zn hydration shell during the uptake, the umbrella-sampling and constrained MD techniques were combined. A harmonic restraint (force constant 50 kJ mol−1 Å−2) was applied to the direction of the normal between the Zn ion and the edge. For each value of the Zn-surface distance a 10 ps equilibration run was performed, which was followed by at least a 10 ps sampling run. Then the equilibrium distance of the restrained potential was reduced by 0.5 Å and a new simulation started from the previous atomic coordinates. In total more than 180 ps ab initio trajectories (eight intermediate runs for ion transfer from solution to the surface) were obtained for the (110) edge of the tvmontmorillonite form. This set of simulations enabled the change in the hydration shell of a Zn ion upon sorption to be monitored. A set of unconstrained ab initio MD trajectories (∼10 ps each) were obtained for different inner-sphere complexes of the Zn ion on the (010) and (110) edges of montmorillonite. The starting configurations for these simulations were derived from geometry optimization of inner-sphere Zn complexes suggested by metadynamics calculations (see below). These trajectories were used to obtain thermal averaged EXAFS spectra of relevant inner-sphere complexes of a Zn ion on the edge of montmorillonite (see Modeling of EXAFS Spectra). Metadynamics Calculations. The metadynamics technique25 was used to search for possible conformations of the Zn ion on edge surfaces of montmorillonite. The Zn−Si and Zn− Al coordination numbers were used as reaction coordinates. The parameters of the coordination number function were set as follows: R0 = 3.6 Å and R0 = 3.8 Å for Zn−Si and Zn−Al coordination numbers, respectively (see eq 1 in ref 26), with the exponents of numerator and denominator set to 12 and 24, respectively. Due to slow dynamics of Zn ion, such simulations are not feasible in the ab initio framework. The runs were therefore performed classically with the CLAYFF force field for montmorillonite, the SPC/E water model, and a pairwise interaction potential for the aqueous Zn ion derived by ref 27. The simulations were performed using the FIST module of the CP2K package.20 The structure of the edge sites was taken from the ab initio structure optimization. All atoms belonging to the clay except hydrogen sites and interlayer cations were fixed throughout the simulations. Potential Zn-adsorbing sites of the clay edge were deprotonated. These calculations were used to scan the phase space for potential conformations of Zn ions on the edge sites of montmorillonite. The inner-sphere Zn complexes obtained in classical metadynamic simulations were reoptimized using density functional calculation and further used as initial configuration for ab initio MD simulations. The obtained ab initio MD trajectories were

used to calculate thermal averaged EXAFS spectra of Zn innersphere complexes. Modeling of EXAFS Spectra. Theoretical EXAFS spectra were calculated on the basis of molecular configurations derived in ab initio simulations using the FEFF 8.40 program.28 The scattering potential of the atoms were calculated selfconsistently28,29 for each atomic configurations. Multiscattering paths up to three legs with path lengths up to 8.0 Å were calculated, and the plane wave filtering with an acceptance criterion of 0.5% of the strongest path amplitude was applied. Other parameters were set to the default values. EXAFS spectra of static configurations from geometry optimization runs were modeled using a global Debye−Waller factor that was accounting for both thermal and structural disorder. The EXAFS spectra from MD trajectories were calculated using a different approach in which the disorder was taken into account by averaging individual spectra without Debye−Waller contribution from 400 molecular snapshots separated by a 25 fs time interval. These spectra were used as a basis for the interpretation of the experimental data. The fit to the experimental EXAFS spectra was performed according to eq 1 ⎛ ⎞ k ⎜∑ ai 2 χ (k) i −χ exp (k)⎟⎟ ⎝ i ⎠ 3⎜

2

+ ai 2

2

⇒ min

(1)

using the Euclidian norm defined by the scalar product, where ai2 are fitting parameters and ⟨χ(k)⟩i are the basis ab initio EXAFS spectra. The fitting of experimental data was limited to the interval of k (Å−1) ∈ [3.0:10.0]. The primary indicator for the fit quality was the value of the objective function in eq 1. When the quality of the fit was within the uncertainty of the measured data, the spectra features in the EXAFS signal and the sum of components (closeness to 100%) were considered as additional indicators of fit quality.



RESULTS AND DISCUSSION Structurally Incorporated Zn. Milos and STx-1 are cisvacant.30,31 Traditional EXAFS data analysis of the Zn incorporated species indicated that Zn substitutes for Al and does not fill empty cv-sites from the dioctahedral aluminum sheet.5,32,33 In this work, EXAFS spectra of Zn atoms incorporated into the octahedral sheet of montmorillonite were modeled on the basis of ab initio refined structures to determine whether Zn substitutes for Al in trans- or cis-sites and to estimate the sensitivity of the ab initio XAFS modeling to small changes in the local environment of the Zn atoms in montmorillonite. The geometries of Zn substitutions in tv-form •• (tv‑montZncis‑Al ′ , tv‑montZntrans‑v ) and in cv-form (cv‑montZncis‑Al ′ , cv‑mont cv‑mont •• Zn′trans‑Al, Zncis‑v) (see Methods for details) were optimized, and the calculated EXAFS spectra are shown in Figure 1 along with the measurements. The spectra of tv‑mont cv‑mont cv‑mont Zn•• Zn•• Zncis‑Al ′ are inconsistent trans‑v, cis‑v, and ′ and with the measurements. The spectra of cv‑montZncis‑Al cv‑mont Zn′trans‑Al sites resemble the measured data. However, the strong correlation of these spectra complicates the component analysis of the experimental data (the correlation coefficient is as high as 0.95). Under such circumstances an integral norm alone is an insufficient indicator of the fit quality. It is therefore sensible to analyze features and characteristic shapes of the spectra in addition. The first double-bounced oscillation at k ∼ 3.5/4.0 Å−1 and two shoulders at k ∼ 6.5/8.2 Å−1 are very sensitive to the clay structure and Zn coordination.5 5715

dx.doi.org/10.1021/es204423k | Environ. Sci. Technol. 2012, 46, 5713−5719

Environmental Science & Technology

Article

using integral norm as an indicator of the fit quality suggests that the data can be fitted best as a binary mixture of cv‑mont Zntrans‑Al ′ and cv‑montZncis‑Al ′ sites with 70% and 30% contribution, respectively. The quality of the fit, however, does not improve significantly compared to the fit using a single ′ . Moreover, the characteristic basis component cv‑montZntrans‑Al features of the spectra are better reproduced with a singlecomponent fit. Therefore, the model with the least number of fitting parameters was preferred, implying that Zn substitutes for Al in the trans-octahedra of cv-montmorillonite. This spectrum was further used to account for the fraction of Zn incorporated in bulk montmorillonite in the analysis of EXAFS of the surface complexes. Uptake at Medium Loading. The first concept which was tested to describe the Zn sorption mechanism assumes that at different loadings Zn forms distinct types of inner-sphere complexes on the edges of montmorillonite. The structure of relevant Zn surface complexes was obtained from the metadynamics simulations of the edge−water interface with classical force field, ab initio geometry optimizations, and MD simulations as described in the Methods section. In total, 52 configurations, including mono- and bidentate Zn-complexes attached to oxygen sites in octahedral and tetrahedral layers of the clay edges for (010), (110), (130), and (100) surfaces, were derived and reoptimized using the ab initio calculations. These reoptimized configurations were used for a multidimensional nonlinear regression to experimental data according to eq 1.

Figure 1. Experimental EXAFS spectra of Zn incorporated in Milosmontmorillonite (points) and modeled ab initio EXAFS spectra (lines) of Zn incorporated in the octahedral layer of 2:1 clay. From the ′ species (blue) and cv‑montZncis‑Al ′ top to the bottom: cv‑montZntrans‑Al •• species (dark green), tv‑montZn′cis‑Al (light green), cv‑montZncis‑v species •• species (yellow). (magenta), and tv‑montZntrans‑v

Remarkably, the calculated cv‑montZntrans‑Al ′ spectra reproduce the above-mentioned structural features the best. The fitted value of the global Debye−Waller factor, which accounts for the statistical and thermal disorder of the complexes, was equal to 0.008 Å2. A multiparameter fit with all available basis spectra

Figure 2. (A) Schematic view of adsorption sites on (010) and (110) edge facets of montmorillonite at low and medium loading (red and green octahedra, respectively). Structural incorporation of Zn is illustrated as black polyhedra. trans- and cis-Al octahedra are shown in blue and yellow colors, respectively. Silica tetrahedra are orange. (B) Best fit of the measured EXAFS spectra, from top to bottom: medium loading (STx-1-medium, blue line), low loading (STx-1-low, green line), low loading (Milos-low, red line), and incorporated (Milos, black line). Contribution of sites on different edges of montmorillonite is summarized in Table 2. Sample loading is provided in Table 1. Spectra of individual components for the fit are shown in Figure S1 (Supporting Information). (C−F) Snapshots from atomistic simulations of Zn adsorption on (110) and (010) edge surfaces. Oxygen atoms are red. Aluminum atoms are green. Hydrogen atoms are white. Silica atoms are yellow. Potassium atoms are pink. Zinc atoms are cv‑mont brown. C and D are the dominant sorption complexes at low loading (cv‑mont (010)edgeZn′trans‑Al and (110)edgeZn′trans‑Al, respectively). E and F are the dominant cv cv sorption complexes at medium loading ((110)AlOZn×(H2O)6 and (010)AlOZn×(H2O)6, respectively). 5716

dx.doi.org/10.1021/es204423k | Environ. Sci. Technol. 2012, 46, 5713−5719

Environmental Science & Technology

Article

This preliminary analysis was aimed to filter out the irrelevant structural configurations and to select Zn surface complexes consistent with the measured data for further ab initio MD simulations. A linear combination fit using ab initio basis components has shown that no more than four configurations are necessary to describe the spectra at medium loading. These were two different complexes of Zn on the (010) and (110) edges of a cv-clay. To account for the thermal motion of atoms, four 10 ps long ab initio MD trajectories starting from the atomic configurations suggested by the regression analysis were obtained. The ensemble averaged ab initio EXAFS basis spectra for each of the four configurations were used as a basis to fit the experimental data (Figure S1, Supporting Information). An excellent fit with just two components was obtained (Figure 2, blue curve). These two components resemble the features of bidentate inner-sphere complexes attached to oxygen sites of octahedral layers on the (010) and (110) surfaces. The spectra of Zn bonded to oxygen atoms of the siloxane plane were unnecessary to reproduce the fit. On the basis of these results it was concluded that at medium loading Zn forms bidentate complexes attached to the octahedral layer of (010) and (110) edge sites of montmorillonite (Figure 2, green polyhedra). During the dynamics, these complexes undergoes slow dangling motion out of the octahedral plane. This dangling can be interpreted as dynamic transformations of in-plane bidentate complexes to off-plane monodentate ones and vice verse and would explain the absence of polarization dependence in the previous EXAFS measurements.5 EXAFS spectra of Zn-surface complexes at medium loading do not show distinct features, unlike those of the incorporated species. Somewhat similar featureless spectra are typical for aqueous ions in the solution. It is therefore reasonable to ask if the spectra of Zn adsorbed at medium loading can be explained by the presence of outer-sphere surface complexes. The calculations suggest that the measured spectra cannot be fitted satisfactory with outer-sphere complexes alone. Consideration of outer-sphere complexes together with the inner-sphere does not result in significant improvement of the fit quality compared to the fit without outer-sphere complexes. The modeling suggests that a small amount of outer-sphere complexes cannot be completely ruled out on the basis of the fitting, but at the same time, their consideration is not essential for the successful data interpretation. The sample handling protocol, however, strongly suggests that the significant contribution of the outer-sphere complexes to the EXAFS spectra measured at medium loading can be ruled out. As a part of the sample preparation process, the excess of solution in the clay films is removed by washing the sample with deionized water before air-drying it at room temperature. Outer-sphere complexes are only weakly bound to the clay surface and are easily removed by washing of the samples with water. On the basis of the experimental arguments and the modeling results, we conclude that the sorption of Zn at medium loading is dominated by the inner-sphere complexes at the edge sites of montmorillonite. To investigate the possible mechanism responsible for the formation of the inner-sphere complexes, a sequence of ab initio MD simulation with restrained ion−surface distance was performed. The evolution of the hydration shell as a function of the distance to the edge surface is shown in Figure 3. The absorption of Zn to the (110) edge occurs as follows. A fivecoordinated [Zn×5H2O] complex approaches the surface, acquires a water molecule on the edge, and forms a

Figure 3. Coordination number of Zn ion as function of reaction coordinate. The reaction coordinate is defined as the distance to the second row of the octahedral layers from the surface. OTot designates all oxygen in the Zn solvation shell. OAl are the oxygen atoms bound to Al in the octahedral plane. Ow are oxygen sites of water molecules, and OSi/Al are oxygen sites of the Al−O−Si linkage in clay. At reaction coordinate values below 7.5 Å, Zn forms a bidentate complex to the edge. At values from 7.5 to 10 Å a monodentate complex is formed. At a value above 10 Å Zn forms an outer-sphere complex. Snapshots illustrate the structure of Zn complexes. Only water molecules in the first coordination shell of Zn ion are shown. Oxygen atoms are red. Hydrogen atoms are gray. Silica atoms are light brown. Zn ion is brown. Aluminum and magnesium atoms are green and black, respectively.

monodentate complex in which a water molecule is shared between the coordination shell of the Zn ion and the surface. As the Zn moves ion further toward the surface its hydration shell loses two water molecules and takes up a tetrahedral configuration, whose vertex is attached to the octahedral layer of the clay through the Al−(H2O)−Zn bond. Docked to the surface as a bidentate surface complex the Zn ion accommodates octahedral symmetry for the first coordination shell. The Zn ion then lies within the plane of the octahedral layer and shares a vertex with the Al octahedra. Uptake at Low Loading. The molecular configurations used to fit EXAFS spectra at medium loading were also applied to model the spectra at low absorber concentrations. However, the approach failed, as the fits were unsatisfactorily poor, even when including all available components simultaneously. Thus, another structural model had to be devised. The EXAFS spectra obtained for inner-sphere complexes at medium loadings have a smooth featureless sinusoidal shape, whereas measured EXAFS spectra at low absorber concentration show a characteristic double-bounced oscillation at k ∼ 3.5/4.0 Å−1 and a shoulder at k ∼ 6.5 Å−1 similar to the spectra measured and calculated for the incorporated Zn species. This suggests that sorption at these conditions may correspond to Zn incorporation in the outermost tetrahedra of the montmorillonite edge. To test this hypothesis, a structural optimization of the outermost Zn′Al substitutions on each of the (010), (110), (100), and (130) edges for both tv- and cv-forms of clays was performed. These simulations were performed for a single clay sheet without water molecules between edge surfaces and in the interlayer. Sixteen spectra for cis-vacant montmorillonite and eight spectra for trans-vacant montmorillonite (four and two per each edge type, respectively) were used to fit the measured spectra at low loading. The contribution from the incorporated Zn species was fixed according to the data from the sample characterization (Table 1).5 The fitting indicates that two types of sites are sufficient to explain the spectra, namely, Zn incorporated in the outermost trans-octahedra on (010) edge, cv‑mont ′ , (010)edgeZntrans‑Al 5717

dx.doi.org/10.1021/es204423k | Environ. Sci. Technol. 2012, 46, 5713−5719

Environmental Science & Technology

Article

Table 2. Best Fit of the Experimental EXAFS Data Based on Ab Initio Simulationsa Milos-low component/ sample

Milosincorp

constrainedb

STx-1-low

STx-1-medium

unconstrainedc constrainedb unconstrainedc

cv‑mont ′ (010)edgeZntrans‑Al

25

27

42

44

cv‑mont (110)edgeZn′trans‑Al

27

40

50

59

50d 102

37 104

10d 102

0 103

cv‑mont bulk Zn′trans‑Al

total (%)

102 102

component/sample cv (010)AlOZn×(H2O)6

innersphere cv (010)AlOZn×(H2O)6 innersphere cv‑mont bulk Zn′trans‑Al

constrainedb unconstrainedc 41

23

58

67

0d 99

7 97

a

The contribution of individual components is given in percents. bPreferred data set obtained from constrained optimization. cAn alternative data set obtained from unconstrained optimization. dValues were fixed during the fit according to the results of chemical analysis.

and (110) edge, cv‑mont (110)edgeZn′trans‑Al, of montmorillonite (Figure 2, red polyhedra). Remarkably, both fits for Milos and STx-1 montmorillonite predict very similar contributions from the (010) and (110) edges. Estimation of the Model Uncertainties. To estimate the robustness of the model developed in this work, the data sets for both low and medium loading were reoptimized taking the contribution of the incorporated Zn as a fitting parameter. The results obtained in such unconstrained fits are remarkably similar to the prediction of the reference model, in which the contribution of incorporated species was fixed during the fit according to the wet chemistry data (Table 2). Please note that in both fitting approaches the sum of the components was not considered as a fitting parameter and was not forced to be 100%. Comparing the results of constrained and unconstrained fits, the uncertainty in the calculated contributions of the individual components can be estimated to be ±15%. Within the two sets of results, the ones obtained from “constrained” fitting are preferred, since they represent a model with the least number of fitting parameters and thus are more plausible for such a complex system. It is worthwhile to be reminded here that adsorption at both medium and low loading was successfully reproduced with the two different sets of complexes formed at the (010) and (110) edge surfaces of cis-vacant montmorillonite. Remarkably, the fractional contribution of the complexes formed at (010) and (110) edge surfaces is comparable both at low and medium loading. At the same time, the sorption complexes on the other edge surface types were not consistent with the measured data. This finding indirectly suggests that the (010) and (110) planes are the dominant edge facets in montmorillonite. Implication for the Surface Complexation Modes. Surface complexation models applied to explain sorption experiments consider the site density as a fitting parameter.34,35 To derive site densities from these data the surface area of edge sites need to be known. The BET measurements suggest that the surface area of edges in montmorillonite is ∼8 m2/g.34 The experimentally estimated sorption capacities are ∼40 and ∼2 mmol/kg at medium and low loadings,36 which correspond to site densities of 3.0 and ∼0.15 sites/nm2, respectively. Molecular modeling results of this study suggest that the density of mono- and bidentate inner-sphere surface complexes attached to the octahedral layer of (010) and (110) edge sites at medium loadings do not exceed ∼2 sites/nm2. At low loading Zn is incorporated into the outermost trans-tetrahedra on (010) and (110) edge sites. The surface density of such sites is 2.0 sites/nm2. Typical concentration of divalent cations incorporated into the structure of 2:1 smectites is ∼0.2 substitutions per octahedral site. This would limit the maximal

density of the surface incorporated sites to 0.4 sites/nm2. These values agree well with the estimates provided by empirical surface complexation models. The atomistic modeling of Zn adsorption and EXAFS measurements thus establish a missing link between the wet chemistry data and molecular structure of adsorbed complexes of transition metals on the edges of dioctahedral clays (Figure 2).



ASSOCIATED CONTENT

S Supporting Information *

Additional material as noted in the text. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*Tel: ++41(0)56 310 4113; fax: ++41(0)56 310 2821; e-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The authors thank the staff of the Dutch Belgium Beamline (DUBBLE, ESRF, Grenoble, France) for the support during the XAFS measurements and the European Synchrotron Radiation Facility (ESRF) at Grenoble, France, for the provision of beamtime and acknowledge access to the highperformance computing facilities at the Swiss Center of Scientific Computing (Manno). The authors are grateful to Dr. Bradbury and Dr. Baeyens for a critical discussion of the results. Partial financial support was provided by the National Co-operative for the Disposal of Radioactive Waste (Nagra), Wettingen, Switzerland.



REFERENCES

(1) Mareta, W.; Sandsteada, H. H. Zinc requirements and the risks and benefits of zinc supplementation. J. Trace Elem. Med. Biol. 2006, 20, 3−18. (2) Adriano, D. C. Trace Elements in the Terrestrial Environment; Springer: New York, 1986. (3) NAGRA. Project Opalinus Clay. Safety Report. Demonstration of Disposal Feasibility for Spent Fuel, Vitrified High-Level Waste and LongLived Intermediate-Level Waste (Entsorgungsnachweis); Technical Report 02-05; National Cooperative for the Disposal of Radioactive Waste (NAGRA): Wettingen/Switzerland, 2002; 472 p. (4) Bradbury, M. H.; Baeyens, B. Modelling the sorption of Zn and Ni on Ca-montmorillonite. Geochim. Cosmochim. Acta 1999, 63, 325− 336. (5) Dähn, R.; Bradbury, M.; Baeyens, B. Investigation of the different binding edge sites for Zn on montmorillonite using P-EXAFSThe

5718

dx.doi.org/10.1021/es204423k | Environ. Sci. Technol. 2012, 46, 5713−5719

Environmental Science & Technology

Article

(24) Churakov, S. V. Structure and dynamics of the water films confined between edges of pyrophyllite: A first principle study. Geochim. Cosmochim. Acta 2007, 71 (5), 1130−1144. (25) Laio, A.; Parrinello, M. Escaping free-energy minima. Proc. Natl. Acad. Sci. U. S. A. 2002, 99 (20), 12562−12566. (26) Churakov, S. V.; Iannuzzi, M.; Parrinello, M. Ab Initio study of dehydroxylation−carbonation reaction on brucite surface. J. Phys. Chem. B 2004, 108 (31), 11567−11574. (27) Chillemi, G.; D’Angelo, P.; Pavel, N. V.; Sanna, N.; Barone, V. Development and validation of an integrated computational approach for the study of ionic species in solution by means of effective twobody potentials. The case of Zn2+, Ni2+, and Co2+ in Aqueous Solutions. J. Am. Ceram. Soc. 2002, 124, 1968−1976. (28) Ankudinov, A. L.; Ravel, B.; Rehr, J. J.; Conradson, S. D. Real space multiple scattering calculation of XANES real-space multiplescattering calculation and interpretation of X-ray-absorption near-edge structure. Phys. Rev. B 1998, 58, 7565−7576. (29) Ankudinov, A. L.; Rehr, J. J. Theory of solid-state contributions to the x-ray elastic scattering amplitude. Phys. Rev. B 2000, 62 (4), 2437−2445. (30) Vantelon, D.; Pelletier, M.; Michot, L. J.; Barres, O.; Thomas, F. Fe, Mg and Al distribution in the octahedral layer of montmorillonites. An infrared study in the OH-bending region. Clay Miner. 2001, 36, 369−379. (31) Tsipursky, S. I.; Drits, V. A. The distribution of octahedral cations in the 2:1 layers of dioctahedral smectites studied by obliquetexture electron diffraction. Clay Miner. 1984, 19 (2), 177−193. (32) Manceau, A.; Tommaseo, C.; Rihs, S.; Geoffroy, N.; Chateigner, D.; Schlegel, M.; Tisserand, D.; Marcus, M. A.; Tamura, N.; Chen, Z.S. Natural speciation of Mn, Ni, and Zn at the micrometer scale in a clayey paddy soil using X-ray fluorescence, absorption, and diffraction. Geochim. Cosmochim. Acta 2005, 69 (16), 4007−4034. (33) Vespa, M.; Manceau, A.; Lanson, M. Natural attenuation of zinc pollution in smelter-affected soil. Environ. Sci. Technol. 2010, 44 (20), 7814−7820. (34) Tournassat, C.; Neaman, A.; Villiéras, F.; Bosbach, D.; Charlet, L. Nanomorphology of montmorillonite particles: Estimation of the clay edge sorption site density by low-pressure gas adsorption and AFM observations. Am. Mineral. 2003, 88, 2322. (35) Bradbury, M. H.; Baeyens, B. A mechanistic description of Ni and Zn sorption on Na-montmorillonite. Part II: modelling. J. Contam. Hydrol. 1997, 27 (3−4), 223−248. (36) Bradbury, M. H.; Baeyens, B. Modelling the sorption of Mn(II), Co(II), Ni(II), Zn(II), Cd(II), Eu(III), Am(III), Sn(IV), Th(IV), Np(V) and U(VI) on montmorillonite: Linear free energy relationships and estimates of surface binding constants for some selected heavy metals and actinides. Geochim. Cosmochim. Acta 2005, 69 (4), 875−892.

strong/weak site concept in the 2SPNE SC/CE sorption model. Geochim. Cosmochim. Acta 2011, 75, 5154−5168. (6) Schlegel, M. L.; Manceau, A.; Charlet, L.; Chateigner, D.; Hazemann, J. L. Sorption of metal ions an clay minerals. 3. Nucleation and epitaxial growth of Zn phyllosilicate on the edges of hectorite. Geochim. Cosmochim. Acta 2001, 65, 4155−4170. (7) Schlegel, M. L.; Manceau, A.; Chateigner, D.; Charlet, L. Sorption of metal ions on clay minerals: 1. Polarized EXAFS evidence for the adsorption of Co on the edges of hectorite particles. J. Colloid Interface Sci. 1999, 215, 140−158. (8) Manceau, A.; Chateigner, D.; Gates, W. P. Polarized EXAFS, distance-valence least-squares modeling (DVLS) and quantitive texture analysis approaches to the structural refinement of Garfield nontronite. Phys. Chem. Miner. 1998, 25, 347−365. (9) Dähn, R.; Scheidegger, A. M.; Manceau, A.; Curti, E.; Baeyens, B.; Bradbury, M. H.; Chateigner, D. Th uptake on montmorillonite: A powder and polarized extended X-ray absorption fine structure (EXAFS) study. J. Colloid Interface Sci. 2002, 249, 8−21. (10) Dähn, R.; Scheidegger, A. M.; Manceau, A.; Schlegel, M. L.; Baeyens, B.; Bradbury, M. H.; Chateigner, D. Structural evidence for the sorption of Ni(II) atoms on the edges of montmorillonite clay minerals. A polarized X-ray absorption fine structure study. Geochim. Cosmochim. Acta 2003, 67 (1), 1−15. (11) Dähn, R.; Scheidegger, A. M.; Manceau, A.; Schlegel, M. L.; Baeyens, B.; Bradbury, M. H.; Morales, M. Neoformation of Ni phyllosilicate upon Ni uptake on montmorillonite: A kinetics study by powder and polarized extended X-ray absorption fine structure spectroscopy. Geochim. Cosmochim. Acta 2002, 66 (13), 2335−2347. (12) Nikitenko, S.; Beale, A. M.; Van Der Eerden, A. M. J.; Jacques, S. D. M.; Leynaud, O.; O’Brien, M. G.; Detollenaere, D.; Kaptein, R.; Weckhuysen, B. M.; Bras, W. Implementation of a combined SAXS/ WAXS/QEXAFS set-up for time-resolved in situ experiments. J. Synchrotron Radiat. 2008, 15 (6), 632−640. (13) Tsipursky, S. I.; Drits, V. A. The distribution of octahedral cations in the 2:1 layers of dioctahedral smectites studied by obliquetexture electron diffraction. Clay Miner. 1984, 19, 177−193. (14) Drits, V. A.; McCarty, D. K.; Zviagina, B. B. Crystal-chemical factors responsible for the distribution of octahedral cations over transand cis-sites in dioctahedral 2:1 layer silicates. Clays Clay Miner. 2006, 54, 131−151. (15) Wolters, F.; Lagaly, G.; Kahr, G.; Nueesch, R.; Emmerich, K. A comprehensive characterisation of dioctahedral smectites. Clays Clay Miner. 2009, 57, 115−133. (16) Kröger, F. A. The Chemistry of Imperfect Crystals; NorthHolland: Amsterdam, 1964; p 1039. (17) Kraehenbuehl, F.; Stoeckli, H. F.; Brunner, F.; Kahr, G.; Mueller-Vonmoos, M. Study of the water−bentonite system by vapour adsorption, immersion calorimetry and X-ray techniques: 1. Micropore volumes and internal surface areas, following Dubinin’s theory. Clay Miner. 1987, 22, 1−9. (18) Cygan, R. T.; Liang, J.-J.; Kalinichev, A. G. Molecular models of hydroxide, oxyhydroxide, and clay phases and the development of a general force field. J. Phys. Chem. B 2004, 108, 1255−1266. (19) VandeVondele, J.; Krack, M.; Mohamed, F.; Parrinello, M.; Chassaing, T.; Hutter, J. Quickstep: Fast and accurate density functional calculations using a mixed Gaussian and plane waves approach. Comput. Phys. Commun. 2005, 167 (2), 103−128. (20) http://www.cp2k.org/. (21) Churakov, S. V. Ab initio study of sorption on pyrophyllite: Structure and acidity of the edge sites. J. Phys. Chem. B 2006, 110, 4135−4146. (22) Bickmore, B. R.; Rosso, K. M.; Nagy, K. L.; Cygan, R. T.; Tadanier, C. J. Ab initio determination of edge surface structures for dioctahedral 2:1 phyllosilicates: Implications for acid−base for reactivity. Clays Clay Miner. 2003, 51 (4), 359−371. (23) Larentzos, J.; Greathouse, J.; Cygan, R. An ab initio and classical molecular dynamics investigation of the structural and vibrational properties of talc and pyrophyllite. J. Phys. Chem. C 2007, 111, 12752− 12759. 5719

dx.doi.org/10.1021/es204423k | Environ. Sci. Technol. 2012, 46, 5713−5719