ARTICLE pubs.acs.org/Langmuir
Comparison of Cation Adsorption by Isostructural Rutile and Cassiterite Michael Machesky,*,† David Wesolowski,‡ J€orgen Rosenqvist,b,‡ Milan Predota,§ Lukas Vlcek,‡ Moira Ridley,|| Vaibhav Kohli,4,^ Zhan Zhang,# Paul Fenter,^ Peter Cummings,z Serguei Lvov,0 Mark Fedkin,0 Victor Rodriguez-Santiago,2,0 James Kubicki,9 and Andrei BanduraO †
Illinois State Water Survey, Champaign, Illinois 61820-7495, United States Chemical Sciences Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831-6110, United States § Institute of Physics and Biophysics, Faculty of Science, University of South Bohemia, Branisovska 31, 370 05 Ceske Budejovice, Czech Republic Department of Geosciences, Texas Tech University, Box 41053, Lubbock, Texas 79409-1053, United States ^ Chemical Sciences and Engineering Division # X-ray Science Division Advanced Photon Source, Argonne National Laboratory, Argonne, Illinois 60439, United States z Department of Chemical Engineering, Vanderbilt University, Nashville, Tennessee 37235, United States 0 Department of Energy and Mineral Engineering 9 Department of Geosciences and the Earth & Environmental Systems Institute The Pennsylvania State University, University Park, Pennsylvania 16802, United States O St. Petersburg State University, St. Petersburg, Russia
)
‡
bS Supporting Information ABSTRACT: Macroscopic net proton charging curves for powdered rutile and cassiterite specimens with the (110) crystal face predominant, as a function of pH in RbCl and NaCl solutions, trace SrCl2 in NaCl, and trace ZnCl2 in NaCl and Na Triflate solutions, are compared to corresponding molecular-level information obtained from static DFT optimizations and classical MD simulations, as well as synchrotron X-ray methods. The similarities and differences in the macroscopic charging behavior of rutile and cassiterite largely reflect the cation binding modes observed at the molecular level. Cation adsorption is primarily inner-sphere on both isostructural (110) surfaces, despite predictions that outer-sphere binding should predominate on low bulk dielectric constant oxides such as cassiterite (εbulk ≈ 11). Inner-sphere adsorption is also significant for Rbþ and Naþ on neutral surfaces, whereas Cl- binding is predominately outer-sphere. As negative surface charge increases, relatively more Rbþ, Naþ, and especially Sr2þ are bound in highly desolvated tetradentate fashion on the rutile (110) surface, largely accounting for enhanced negative charge development relative to cassiterite. Charging curves in the presence of Zn2þ are very steep but similar for both oxides, reflective of Zn2þ hydrolysis (and accompanying proton release) during the adsorption process, and the similar binding modes for ZnOHþ on both surfaces. These results suggest that differences in cation adsorption between high and low bulk dielectric constant oxides are more subtly related to the relative degree of cation desolvation accompanying inner-sphere binding (i.e., more tetradentate binding on rutile), rather than distinct inner- and outer-sphere adsorption modes. Cation desolvation may be favored at the rutile (110) surface in part because inner-sphere water molecules are bound further from and less tightly than on the cassiterite (110) surface. Hence, their removal upon inner-sphere cation binding is relatively more favorable.
’ INTRODUCTION Experimental and computational tools that probe mineralwater interfacial structure and dynamics at the atomic level have proven to be absolutely necessary in reducing the ambiguity inherent in rationalizing macroscopic ion adsorption data. Key beneficiaries of this clarity have been ion adsorption frameworks such as the MUSIC and CD models of Hiemstra, van Riemsdijk, r 2011 American Chemical Society
and others,1,2 which have remained relevant in light of growing microscopic scrutiny. The impetus for our own efforts toward clarity was the call by Brown and coauthors3 for the increased use of integrated Received: October 5, 2010 Revised: January 20, 2011 Published: March 21, 2011 4585
dx.doi.org/10.1021/la1040163 | Langmuir 2011, 27, 4585–4593
Langmuir macroscopic and microscopic approaches to decipher mineralwater interface complexity. Rutile, and in particular the predominant (110) crystal face, was the initial focus of our studies, and Zhang et al.4 summarize many of our earlier findings. Subsequently, the isostructural cassiterite (110) surface has received similar attention,5 and neutron scattering has been added to our arsenal of experimental techniques.6 Recent studies have compared rutile and cassiterite with respect to water structure and dynamics via neutron scattering and classical molecular dynamics (CMD) simulations,7 multilayer water adsorption via density functional theory,8 and surface protonation.9,10 A prime motivation for these comparative rutile-cassiterite studies was to test the hypothesis of Sverjensky11,12 that cation adsorption is primarily inner-sphere on metal oxides with high bulk dielectric constants (e.g., rutile ∼120) and outer-sphere on metal oxides with low bulk dielectric constants (e.g., cassiterite ∼11). This hypothesis stems from an earlier study by James and Healy13 who postulated that the work required to remove solvation waters from adsorbing cations is inversely related to the dielectric constant of the solid. Moreover, this theory points to the importance of interfacial water structure in helping to govern ion adsorption, a concept that has been supported by experimental and theoretical studies of various oxide surfaces including hematite,14 goethite,15 corundum,16 and silica.17 Rosenqvist et al.10 briefly compared rutile and cassiterite proton charge curves in RbCl and NaCl media and noted that negative charge development was greater on rutile, particularly in NaCl media, and that this enhancement corresponded with CMD simulations, which revealed more inner-sphere tetradentate binding on rutile (110) than cassiterite (110). Here, we extend this rutilecassiterite comparison to results for two monovalent (Naþ, Rbþ) and two divalent (Sn2þ, Zn2þ) cations for which both macroscopic charging data and molecular modeling and/or X-ray scattering results are available. Specifically, we combine the results of a broad range of experimental/computational information to ascertain the consistency of the macroscopic and microscopic data, and to clearly delineate the similarities and differences in cation binding between rutile and cassiterite at the molecular level.
’ METHODS Most of the results presented and discussed here were generated with experimental and computational methods that are detailed elsewhere. Accordingly, primarily provided below are brief method summaries and relevant literature citations. Net Proton Charge Titrations. These titrations were conducted at 25 C with glass pH electrodes according to procedures detailed in Ridley et al.18 and Machesky et al.9 for rutile and Rosenqvist et al.10 for cassiterite. Approximately 1 g samples of carefully cleaned rutile and cassiterite powders, each dominated by the (110) crystal face, were suspended in ca. 40 mL acidic solutions (pH ∼2.7) of various but precisely known compositions, typically in 0.03 and 0.3 m NaCl, NaTr (Tr = triflate, CF3SO3-), or RbCl, plus trace amounts (10-4 to 10-3 m) of multivalent cations where appropriate. Titrations were then conducted to pH 8.5-11 by adding 15 to 40 aliquots of base titrant (NaOH or RbOH in the respective electrolyte media). From mass and charge balance considerations, the solution excess or deficit of protons is known for each measured pH value, and this excess or deficit can be expressed in terms of net proton charge per unit surface area of solid titrated (C/m2). Blank titrations were routinely conducted to correct for experimental artifacts, such as non-Nernstian electrode behavior. It is important to note that the same sources of rutile and cassiterite powders were used for all titrations, and that these powders were cleaned exhaustively including
ARTICLE
a final hydrothermal pretreatment step (14 days at 200 C in DI water). This helped ensure that repeat titrations were highly reproducible (to within 0.01 C/m2), and that titration curves conducted at several ionic strengths intersected very near zero net proton charge (σH ≈ 0). X-ray Synchrotron Methods. X-ray standing wave (XSW) measurements were conducted on single crystal (110) samples of rutile and cassiterite under micrometer-scale films of the solutions of interest. Procedural details are available for rutile in several previous publications,4,19 and similar procedures were followed in obtaining results from the cassiterite (110) surface. However, high-quality XSW results for the cassiterite (110) surface were much more limited due primarily to difficulty in synthesizing single crystals of sufficient size and quality. Synthesis was necessary, because unlike rutile, an adequate commercial source was not available. Briefly, the cassiterite (110) surface XSW measurements were performed at the XOR Sector 12 (BESSRC) facility at Advanced Photon Source (APS), Argonne National Laboratory (ANL). A symmetric Si (111) high heat load monochromator was used to select the X-ray energy of 10.9 keV. Two sets of double-bounce Si channel cuts were used to further collimate the beam. The typical beam size was ∼0.05 0.5 mm2. Procedural details of these Bragg XSW measurements are presented elsewhere.20 The Zn2þ solutions were 10-5 m ZnTr2 buffered at pH 8.0 with submillimolal concentrations of Tris(hydroxymethyl)aminomethane (Tris). At these solution conditions and room temperature, crystalline ZnO and Zn(OH)2 are undersaturated, and Zn2þaq is the dominant species in solution.21,22 During the X-ray measurements, typical solution thicknesses above the sample surface were estimated to be ∼2-5 μm. The absolute total adsorbed Zn2þ coverage was calibrated by comparing the fluorescence yield to a Zn-ion-implanted standard sample measured in the same experimental geometry. Static DFT Calculations. The planewave pseudopotential implementation of density functional theory (DFT) was used to study the relative energies and structures of possible Zn2þ adsorption sites on the (110) surfaces of rutile and cassiterite. Calculations were performed using the VASP code,23,24 and exchange and correlation were treated within the Perdew-Burke-Ernzerhof (PBE) functional.25 Projectoraugmented-wave (PAW) pseudopotentials26,27 were used in these calculations, allowing us to apply the medium planewave cutoff energy (400 eV). Accurate Ti, Zn, and Sn pseudopotentials, in which the semicore states (3s and 3p (Ti), 3d (Zn), and 4d (Sn)) were treated as valence states, were chosen. Our previous DFT results for the Zn2þrutile system4,19 have primarily employed the CASTEP code.28 Analogous to our previous calculations of Zn2þ adsorption on the rutile surface,4 a three Sn-layer slab model with a 3D-supercell was used to mimic the (110) surface. The resulting orthorhombic cell√consisting of 2 1 surface unit cells had dimensions of a = 2c0 and b = 2a0 in the (001) and (110) directions, respectively, where a0 (4.737 Å) and c0 (3.186 Å), are the experimental29 crystal-lattice parameters for cassiterite. The third translation vector c in z-direction normal to the (110) surface was chosen to be 24 Å, resulting in an approximately 15 Å vacuum gap between the two (110) faces. The positions of all atoms were allowed to relax except the positions of the central layer atoms, which were fixed at the bulk crystal geometry. Additionally, cell parameters a, b, and c were kept constant during optimization. Energy minimization was carried out until all forces on atoms were less than 0.025 eV/Å. The number of H2O molecules was chosen to provide complete filling of the first solvation sphere of a Zn2þ ion. Since at least one coordination place of adsorbed Zn2þ is occupied by surface oxygen atoms, five water molecules were added on each side of the slab to provide the probable sixfold coordination for Zn2þ. To satisfy the neutrality condition, two terminal hydroxyl groups were attached to fivefold tin atoms on each surface. This is the smallest model that can be 4586
dx.doi.org/10.1021/la1040163 |Langmuir 2011, 27, 4585–4593
Langmuir
ARTICLE
used for simulating the adsorption of Zn2þ on the cassiterite (110) surface. Several different initial positions of Zn2þ on this minimal cassiterite (110) surface were investigated to obtain representative information about possible adsorption structures and their energies. The most probable positions of Zn2þ ion on the rutile (110) surface obtained in previous theoretical and X-ray standing wave experiments4,19 were used to constrain these choices. Still, the relative stability of the calculated Zn2þ adsorption structures should be interpreted with caution, because effects from the interaction of periodically repeated structures may not be negligible, especially due to the presence of localized charged centers on Zn. Moreover, the adsorption configurations chosen are certainly not exhaustive. Classical Molecular Dynamics (CMD) Simulations. CMD simulations were carried out according to previously published methods,5,10,30-32 and a summary of those methods including relevant interaction potentials is provided in the Supporting Information section. In short, periodic planar slabs with lateral dimensions equal to 12 6 replicas of the basic unit cell (35.508 38.981 Å2 for rutile and 38.232 40.194 Å2 for cassiterite) were used with a 50 Å separation between opposing surfaces. The space between the surfaces was populated with a total of 2048 SPC/E water molecules, which was sufficient to ensure that bulk water properties (e.g., density, diffusivity) existed in the center of the simulation cell.5,30 Some of the water molecules were replaced by ions and surface hydroxyl groups, depending on the desired system composition. The interactions of water with surface groups were modeled with ab initio derived forcefields derived separately for both rutile33 and cassiterite.34 Negative surface charge was created by removing Hþ from bridging oxygens of the hydroxylated surfaces in a manner which minimized their mutual repulsion. The removal of Hþ was compensated by adding an excess of cations to the solution phase, thereby maintaining electrical neutrality. After an equilibration period exceeding 1 ns, simulations typically proceeded for 1-2 ns with 1 fs time steps.
’ RESULTS Macroscopic. Representative net proton charge data are presented in Figure 1 for rutile (top) and cassiterite (bottom). Results for RbCl, NaCl, and 0.001 m Sr2þ and Zn2þ in background 1:1 electrolyte media are presented because comparable titrations are available for both oxides. The rutile data have been published previously,4,35 as have the cassiterite NaCl and RbCl results.10 The NaCl and RbCl only titration curves also include error bars ((0.01 C/m2) to indicate titration curve reproducibility. For these NaCl and RbCl only titrations, the respective 0.03 and 0.30 m titration curves intersect at pH values very near the zero proton condition, which signifies that the resulting common intersection point pH values (pHcip) can be equated with the pH of zero net proton charge (pHznpc).36 For rutile, this pHznpc value is 5.4 ((0.2) at 25 C in both NaCl and RbCl only solutions,18,35 and a recent independent study has confirmed 5.4 as the pHznpc in NaCl solutions.37 Cassiterite pHcip values at 25 C are 4.3 ((0.1) and 4.1 ((0.2) in NaCl only and RbCl only media, respectively,10 which makes identification of a corresponding single pHznpc value less precise. However, titrations were more reproducible in NaCl media, which suggests that 4.3 (albeit with greater uncertainty, ( 0.2) is better as a unique pHznpc value.10 Net proton charge data for rutile and cassiterite are compared for RbCl (top) and NaCl (bottom) in Figure 2. The net proton charge data are the same as in Figure 1, but they are presented relative to the respective pHznpc values of rutile and cassiterite (pHznpc - pH) to adjust for the 1.1 pH unit difference in pHznpc values, thereby allowing a more direct comparison of proton
Figure 1. Representative net proton charge vs pH data for rutile (top) and cassiterite (bottom). The error bars given for the NaCl and RbCl data ((0.01 C/m2) represent titration curve reproducibility. The dashed lines highlight zero net proton charge and corresponding pHznpc values. Figure legends provide the compositions of the solutions titrated.
charge development under both net positive (pHznpc - pH > 0) and negative (pHznpc - pH < 0) proton charge conditions. Similarly, Figure 3 compares rutile and cassiterite net proton charge data for trace (0.4 to 1 mm) concentrations of Sr2þ (top) and Zn2þ (bottom) in 0.03 and 0.30 m NaCl or NaTr solutions. Microscopic. Table 1 contains CMD simulation results for Rbþ, Naþ, and Sr2þ at three charge states (0, -0.1, and -0.2 C/ m2), and XSW results for Zn2þ at pH 8. The cassiterite CMD results for Rbþ and Naþ are adapted from Rosenqvist et al.,10 while the Sr2þ results have not been published previously. The rutile CMD results are an extension of those previously published31 (Predota et al., in preparation). The Zn2þ results for rutile (110) are from Zhang et al.,19 while those for cassiterite (110) have not been previously reported. The CMD results provided include the total number of cations contained in a particular simulation (#), the number of cations bound as innersphere species per Ti2O4/Sn2O4 surface unit (#IS), their fractional distribution (occupancy) between various inner-sphere binding configurations, the binding distance (above the first Ti or Sn layer at the (110) surface) for the various adsorption geometries, and the weighted average binding distances (the sum of the various fractional occupancies times their binding distances). Figure 4 presents ion distributions (Rbþ, Naþ, and Cl-) above the (110) surface planes of rutile and cassiterite at zero charge as 4587
dx.doi.org/10.1021/la1040163 |Langmuir 2011, 27, 4585–4593
Langmuir
Figure 2. Net proton charge vs pHznpc - pH data for rutile (black symbols) and cassiterite (red symbols) in RbCl (top) and NaCl (bottom) media. Selected proton charge error bars ((0.01 C/m2) are also given. The dashed lines highlight the location of the pHznpc.
obtained from CMD simulations. These distributions are scaled to familiar macroscopic molar concentration units for convenience. Figure 5 presents normalized fluorescence yield and rocking curve XSW data for Zn2þ adsorbed at the (110) surfaces of cassiterite (top) and rutile (bottom) at pH 8. The cassiterite data are the only set of good results obtained for Zn2þ, while the rutile data were obtained during the course of a previous study.20 Figure 6 presents XSW generated 3D iso-density contour surfaces for the 110 surfaces of rutile (top) and cassiterite (bottom) where the top-layer Ti/Sn sites are included to illustrate how the adsorbed Zn ions adopt adsorption sites that are similar to the projected substrate lattice sites. The relative adsorption energies and heights of Zn2þ calculated both for TiO2 and SnO2 (110) surfaces from our static DFT optimizations are compared in Table 2. The minimum energy obtained for six-coordinated Zn2þ bound in monodentate fashion to bridging oxygen (Mono-BO (VI)) is used as the zero level for the energy values given for the four other adsorption structures listed since our X-ray standing wave results have shown that Zn2þ bound in mondentate fashion to the BO is the dominant adsorption configuration on rutile (110).4,19
’ DISCUSSION Zero Surface Charge. The net proton charge curves in 0.03 and 0.30 m NaCl and RbCl for both rutile (Figure 1, top) and cassiterite (Figure 1, bottom) intersect very near the zero net proton charge, which is good evidence that the corresponding
ARTICLE
Figure 3. Net proton charge vs pHznpc - pH data for rutile (black symbols) and cassiterite (red symbols) for Sr2þ (top) and Zn2þ (bottom). The dashed lines highlight the location of the pHznpc.
pH values (5.4 and 4.3) are the respective pHznpc values. Our CMD results given in Table 1 and Figure 4 indicate significant inner-sphere binding of Rbþ and especially Naþ to both rutile and cassiterite at zero surface charge. For rutile, fractional innersphere coverages (per Ti2O4 surface unit) are 0.04 for Rbþ and 0.06 for Naþ, while corresponding coverages for cassiterite are 0.02 for Rbþ and 0.05 for Naþ (Table 1). The corresponding (scaled to molarity) concentrations are represented in Figure 4 by those portions of the Rbþ and Naþ distributions within about 5 Å of the (110) surface planes (marked by vertical dotted lines). The indicated molarities reveal a distinct predominance of Rbþ and Naþ in the inner-sphere region and Cl- in the outer-sphere region. The conventional macroscopic interpretation would then be that cation and anion binding at zero charge is compensatory (rather than insignificant). At the molecular level, however, this compensation has a distinct segregated structure that is similar for both rutile and cassiterite with significant inner-sphere Rbþ and Naþ binding and primarily outer-sphere Cl- binding. The neutral hydroxylated (110) surfaces of rutile and cassiterite (constructed in the CMD simulations assuming all terminal H2O molecules are fully dissociated) are populated by terminal (Ti-OH and Sn-OH) and bridging (Ti2-OH or Sn2-OH) hydroxyl groups, for which the negative charge of the surface oxygen atoms, at the molecular level, remains net attractive for solution cations. That is, the Coulombic interaction of cations and negatively charged surface oxygen atoms with limited flexibility results in well-defined adsorption configurations and density profiles, while the much more freely rotating hydrogen 4588
dx.doi.org/10.1021/la1040163 |Langmuir 2011, 27, 4585–4593
Langmuir
ARTICLE
Table 1. Inner-Sphere Cation Binding Configurations, Occupancies (Fractional), Distances, and Average Distances (above the (110) Surface Me-O Planes) for Rutile (110) (left) and Cassiterite (110) (right)a ion (#b)/ 2
charge (C/m ) TiO2(110) Rbþ(18)/0
site TD TO-TO &BO-TO
Rbþ(30)/-0.1
TD
occupancy
distance (Å)
0.619
3.55
0.381
4.25
0.776
3.55
0.224
4.25
avg.
ion (#b)/
avg.
distance
charge (C/m2)
distance
(Å)/#ISc
SnO2(110) Rbþ(12)/0
3.82/0.04
Rbþ(36)/-0.1
TO-TO &BO-TO
Rbþ(48)/-0.2
TD
0.864
3.55
0.136
4.25
3.71/0.11
Rbþ(48)/-0.2
TO-TO &BO-TO
Naþ(18)/0
Naþ(30)/-0.1
Naþ(51)/-0.2
Sr2þ(16)/0
Sr2þ(22)/-0.1
Sr2þ(31)/-0.2
Zn2þ/pH 8d
TD
0.000
2.9
BO-TO
0.022
3.25
TO-TO
0.978
3.75
TD
0.039
2.9
BO-TO
0.112
3.25
TO-TO
0.850
3.75
TD
0.111
2.9
BO-TO
0.091
3.25
TO-TO
0.798
3.75
TD
0.000
3.25
TO-BO
0.000
3.55
TO-TO
1.000
4.05
TD
0.000
3.25
TO-BO
0.000
3.55
TO-TO TD
1.000 0.417
4.05 3.25
TO-BO
0.261
3.55
TO-TO
0.322
4.05
TO-TO
0.200
2.53
mono-BO
0.590
3.23
3.65/0.23
Naþ(12)/0 3.74/0.06
Naþ(36)/-0.1 3.66/0.14
Naþ(48)/-0.2 3.61/0.28
Sr2þ(16)/0 4.05/0.03 Sr2þ(18)/-0.1 4.05/0.06 Sr2þ(27)/-0.2 3.59/0.15 Zn2þ/pH 8d 3.05/0.50
site
occupancy
distance (Å)
TD
0.391
3.7
TO-2BO 2TO-BO
0.000 0.200
3.88 3.94
TO-BO
0.190
4.16
mono-BO
0.000
4.56
TO-TO
0.121
4.44
mono-TO
0.091
4.96
TD
0.327
3.7
TO-2BO
0.014
3.88
2TO-BO TO-BO
0.117 0.312
3.94 4.16
mono-BO
0.018
4.56
TO-TO
0.092
4.44
mono-TO
0.119
4.96
TD
0.402
3.58
TO-2BO
0.021
3.78
2TO-BO
0.076
3.88
TO-BO mono-TO
0.348 0.024
4.08 4.47
TO-TO
0.043
4.51
mono-TO
0.085
4.99
TD
0.000
3.05
TO-BO
0.104
3.53
mono-BO
0.000
3.56
TO-TO
0.837
3.83
mono-TO TD
0.054 0.015
4.27 3.05
TO-BO
0.201
3.53
mono-BO
0.063
3.56
TO-TO
0.686
3.83
mono-TO
0.024
4.27
TD
0.038
2.96
TO-BO
0.367
3.52
mono-BO TO-TO
0.062 0.497
3.66 3.78
mono-TO
0.020
4.25
TO-BO
0.000
3.59
TO-TO
0.995
4.14
mono-TO
0.002
4.41
TO-BO
0.055
3.59
TO-TO
0.886
4.14
mono-TO TO-BO
0.003 0.042
4.41 3.73
TO-TO
0.957
4.06
mono-TO
0.001
4.41
TO-TO
0.140
3.01
mono-BO
0.330
3.08
(Å)/#ISc
4.01/0.02
4.10/0.09
3.96/0.16
3.80/0.05
3.71/0.18
3.59/0.28
4.14/0.01
4.11/NA
4.05/0.17 3.03/0.07
Rbþ, Naþ, and Sr2þ results (at 0, -0.1, and -0.2 C/m2 equivalent charge), include the total number of ions in the simulation (#) and the number adsorbed inner-spherically (#IS) are from CMD simulations. Zn2þ results are from X-ray scattering experiments (pH 8). b Total number of ions in the simulation cell. c Number of inner-sphere ions per Ti2O4/Sn2O4 surface unit. d X-ray standing wave results where pH was measured. a
4589
dx.doi.org/10.1021/la1040163 |Langmuir 2011, 27, 4585–4593
Langmuir
ARTICLE
Figure 5. XSW measurements for Zn2þ adsorption at the SnO2 (top 2 panels) and TiO2 (bottom 2 panels) aqueous solution interface. The top panels for each surface are the Zn fluorescence yields (right axis), and the bottom panels are the rocking curves for SnO2/TiO2 (110) reflections (left axis). The lines are the best fits to the data. Here, θb is the Bragg angle for the (110) reflection, and f and P refer to the coherent fraction and coherent position, respectively. Solution conditions: pH = 8.04, total [Zn2þ] = 10-5 molal.
Figure 4. Ion concentrations at zero charge as a function of distance above the (110) surface planes of rutile and cassiterite from CMD simulations for RbCl (top) and NaCl (bottom). The vertical dotted line at about 5 Å depicts the approximate inner-sphere and outer-sphere boundary. Note the ion concentration scale break for NaCl.
atoms of surface hydroxyl groups are able to swing away from approaching cations, which minimizes electrostatic repulsion. Conversely, the interaction of Cl- with hydrogen atoms of surface hydroxyl groups is weaker due to the presence of O-ClCoulombic and van der Waals repulsions. As a result, cations bind primarily in inner-sphere fashion and Cl- in outer-sphere fashion, although there is some inner-sphere binding of Cl- on the neutral rutile (110) surface (Figure 4). Additional simulations are required to confirm these results. The equivalent bulk concentrations of our simulations are relatively high, ranging from about 0.7 M before simulation (e.g., scaled from 48 Rbþ in 2000 water molecules) to 0.3 to 0.5 M after equilibration of the simulation (Figure 4), with the decrease due to inner- and outer-sphere binding. Larger-scale simulations could probe lower bulk concentrations. Still, the 0.3 m macroscopic titrations for RbCl and NaCl cross those conducted at 0.03 m at the zero proton condition (Figure 1), which suggests that such larger-scale simulations would yield similar neutral surface
Figure 6. XSW generated 3D iso-density plots (at 60% peak density) for the 110 surfaces of rutile (top) and cassiterite (bottom) overlapping with the respective ball and stick models of those surfaces. The leftmost figure of each pair depicts the location of surface Ti and Sn atoms, and the rightmost figures depict the Zn2þ contours at pH = 8.
results. Moreover, given the prevalence of oxygen-terminated surfaces for metal oxide surfaces in water,38 perhaps the distinct asymmetric cation/anion distribution observed here is the norm rather than exception. Negative Surface Charge. Net proton charge curves become more negative with increasing pH for both oxides for all solution conditions. This negative proton charge development is virtually identical in the presence of Rbþ and Naþ for SnO2 (Figure 1 bottom), while there is a slight enhancement of negative proton charge for Naþ relative to Rbþ for TiO2 (Figure 1, top) at 0.3 m ionic strength. For trace Sr2þ with NaCl as the background 4590
dx.doi.org/10.1021/la1040163 |Langmuir 2011, 27, 4585–4593
Langmuir
ARTICLE
Table 2. Comparison of Znþ2 Adsorption on SnO2 (110) and TiO2 (110) for Several Adsorption Structures Optimized with the VASP23,24 Package relative energy, kJ/mol site type
SnO2
TiO2
height of Zn2þ (Å) SnO2
TiO2
a
Mono-BO (VI )
0
0
3.45
3.49
Mono-BO(IVb)
-68
-80
3.35
3.24
BO-TO(IV)
-84
(-62)c
3.02
3.03
TO-TO(IV)
-90
-121
3.11
3.05
Mono-TO(IV)
-95
-110
4.05
3.93
a
Zn2þ in sixfold coordination. b Zn2þ in fourfold coordination. c May not be completely optimized.
electrolyte, negative proton charge is significantly enhanced for TiO2, but only slightly for SnO2 (relative to the NaCl only titrations). Finally, negative proton charge is greatly enhanced for both oxides in the presence of trace (0.4 to 1 millimolal) Zn2þ. Relative to the respective pHznpc values (pHznpc-pH), negative proton charge development for TiO2 is slightly enhanced relative to SnO2 with Rbþ as the counterion (Figure 2, top), and this relative enhancement is somewhat greater still with Naþ as the counterion (Figure 2, bottom). Trace Sr2þ results in much greater negative proton charge development on TiO2 than SnO2 (Figure 3, top), while negative proton charge development in the presence of trace Zn2þ is similar for both oxides in the pH regime before Zn2þ is adsorbed completely, as signified by the charging curve plateaus (Figure 3, bottom). Most of these macroscopic trends are consistent with the CMD and X-ray results for negatively charged surfaces summarized in Table 1, and the neutral surface static DFT results for Zn2þ in Table 2. For Rbþ, the slightly greater negative charge development on rutile (Figure 2, top) can be rationalized by more Rbþ being adsorbed at the tetradentate (TD) site (0.86 vs 0.40 fractional site occupancy at -0.2 C/m2), which is closest to the surface, and hence more effective at screening negative proton-induced surface charge development. Another measure of this is the weighted average height of Rbþ above the respective (110) surface planes which at -0.2 C/m2 charge is 3.65 Å for rutile and 3.96 Å for cassiterite. For Naþ, the macroscopic and microscopic results are less consistent. The similar average weighted distances at -0.2 C/m2 charge from CMD simulations (3.61 Å rutile, 3.59 Å cassiterite) do not reflect the greater negative surface charge of rutile for that condition (Figure 2, bottom). However, more Naþ is bound in TD fashion on rutile than on cassiterite at -0.2 C/m2 charge (0.11 vs 0.04 fractional site occupancy), which could help account for the enhanced negative surface charge development observed macroscopically. Moreover, the CMD and titration curve results are more consistent at -0.1 C/m2 charge where both the similar average binding distances and predominant bidentate TO-TO binding geometries reflect the nearly identical charging curves (within (0.01 C/m2) at that condition. The rutile CMD results indicate a large shift from exclusive TO-TO binding of Sr2þ at 0 and -0.1 C/m2 to 42% TD binding at -0.2 C/m2. On cassiterite, however, TO-TO binding predominates at all these charge conditions. The corresponding macroscopic manifestation is that negative surface charge develops to a much greater extent on rutile than cassiterite (Figure 3, top). Moreover, negative charge development for Sr2þ on cassiterite is only slightly greater than for Naþ and Rbþ
(Figure 1, bottom). In this instance, the microscopic rationalization is that significant fractions of Naþ and Rbþ are bound closer to the cassiterite surface (at the BO-TO and TD sites, respectively) than is Sr2þ (primarily bound in TO-TO fashion). Consequently, negative surface charge development in the presence of the monovalent Naþ and Rbþ is similar to that of Sr2þ. According to the XSW results (Table 1 and Figure 6), Zn2þ adsorbs in both monodentate (mono-BO) and bidentate (TOTO) fashion above the cassiterite (110) surface at pH 8 with the mono-BO configuration predominant. Zinc adsorption on the rutile (110) surface at pH 8 involves the same sites,19 and the mono-BO configuration again predominates. However, fractional surface coverages, which are an estimate of Stern-layer occupancy,20 are different. Fractional coverage is greater for rutile, with the equivalent monolayer (ML) coverage (1 ML = 1 ion per Ti2O4/Sn2O4 surface unit) being 0.50 ( 0.05 ML for rutile (ideally) and 0.07 ( 0.05 ML for cassiterite at pH 8. This large difference in estimated Stern layer coverage is unexpected given that the surfaces are isostructural, the same inner-sphere binding configurations are involved (mono-BO and TO-TO), and the corresponding macroscopic titration curves (Figure 3, bottom) are more similar. Although the reason for this coverage difference is not known for certain, we suspect that it is due to factors extrinsic to the measured adsorption configurations such as sample surface quality. First, separate XSW measurements of Zn2þ on the rutile (110) surface at pH 8 exhibited rather wide variability in site occupancy (0.12 to 0.50 ML), which was attributed to contaminated or defective surfaces preventing a tighter distribution nearer the observed maximum (0.50 ML) since the observed binding configurations and their relative occupancies (Table 1) remained constant.20 Second, we were only able to obtain a single good-quality XSW measurement of Zn2þ on the cassiterite (110) surface (Figure 5, top), and given the low site occupancy found, we suspect that sample surface quality was not ideal. In any case, more systematic measurements of Zn2þ adsorbed at the cassiterite (110) surface would be needed to obtain a better estimate of the ideal site occupancy. The Zn2þ adsorption distances are also different with the TO-TO binding height being considerably less on rutile (2.53 Å) than cassiterite (3.01 Å), as is clearly evident from the respective iso-density plots in Figure 6. Conversely, the Zn2þ adsorption height is lower on cassiterite for the dominant mono-BO site, although the height difference is less (3.08 vs 3.23 Å) than for the TO-TO site. The static DFT results for Zn2þ (Table 2) suggest that a reduction in coordination number (from 6 to 4) and hydrolysis (Zn2þ to ZnOHþ) accompany adsorption to both rutile and cassiterite. That is, the mono-BO adsorption configuration in fourfold coordination for both rutile and cassiterite (second Table 2 entry) is much lower in energy than the sixfold monoBO configuration (first Table 2 entry). Moreover, the fourfold mono-BO complexes underwent hydrolysis during the course of the relaxations, and the resulting adsorption heights are in better agreement with the corresponding XSW determined heights (Table 1) than are the sixfold mono-BO complexes. The remaining fourfold adsorption complexes given in Table 2 are also hydrolyzed and more stable than the reference sixfold monoBO complex. In fact, for the limited set of adsorption structures tested, the fourfold TO-TO and fourfold mono-TO complexes are most stable for rutile and cassiterite, respectively. This is apparently at odds with our XSW results (mono-BO (IV) predominant), although given the limited scope of our DFT 4591
dx.doi.org/10.1021/la1040163 |Langmuir 2011, 27, 4585–4593
Langmuir calculations, the most certain results are that the mono-BO Zn2þ adsorption complex is more stable in fourfold than sixfold coordination, and that hydrolysis accompanies fourfold but not sixfold coordination. In any case, the generally similar Zn2þ binding modes for rutile and cassiterite are reflected in the macroscopic charging curves (Figure 3, bottom), which are also similar in the pH regime of active Zn2þ adsorption (that is, below the plateaus which represent virtually 100% Zn2þ adsorption). Moreover, even on rutile the Zn2þ charging curves are much steeper than those for Sr2þ (Figure 1) with this increased steepness primarily resulting from Hþ release accompanying Zn2þ hydrolysis. Our comparative CMD and neutron scattering results7 reveal that water structure and dynamics are also different between rutile and cassiterite, with an important distinction being that water molecules that are H-bonded directly to surface oxygen or hydroxyl groups (L2 water molecules) are more strongly bound and complexly structured on cassiterite than rutile. A semiquantitative measure of this difference comes from our recent analysis of CMD derived H-bond configurations for rutile and cassiterite (110) surfaces within the MUSIC model surface protonation framework.9,10 Terminal hydroxyl group oxygen atoms on rutile participate in an average of 1 H-bond (∼1.7 Å long) with L2 water molecule hydrogen atoms, while each terminal hydroxyl group oxygen atom on cassiterite participates in 1.4 H-bonds (∼1.8 Å long) with L2 water molecules. H-bonding between L2 water hydrogen atoms and bridging oxygen atoms is also greater for cassiterite, averaging 1.5 H-bonds (∼1.7 Å long) versus 1.3 H-bonds (∼1.8 Å long) for rutile bridging oxygen atoms. Since it is the L2 water molecules that directly impede inner-sphere cation binding and these waters are H-bonded to cassiterite more strongly, this could account for the greater degree of outersphere binding on cassiterite compared to rutile. In fact, our results demonstrate that the difference in cation adsorption mode between rutile and cassiterite is in the relative degree of cation desolvation upon adsorption, rather than a more distinct inner- vs outer-sphere adsorption mode. Adsorbed Rbþ, Naþ, and Sr2þ are more highly desolvated on rutile than cassiterite, since (on average) a greater fraction of these cations are adsorbed in tetradentate fashion (Table 1), which requires the removal of 4 primary hydration waters in addition to intervening L2 water molecules. However, these same cations also adsorb to cassiterite in inner-sphere fashion, but the more solvated bidentate adsorption configurations are more prominent.
’ CONCLUSIONS Similarities and differences in macroscopic charging curves of rutile and cassiterite in the presence of Rbþ, Naþ, Sr2þ, and Zn2þ correspond closely to actual binding modes observed at the molecular level from combined CMD, DFT, and X-ray reflectivity studies. X-ray results show that Zn2þ adsorption configurations are similar for both oxides, and DFT calculations suggest that adsorbed Zn2þ changes from sixfold to fourfold coordination and also hydrolyses (to ZnOHþ) on both oxides, which together account for the similar and steep charging curves. The CMD results indicate significant inner-sphere Rbþ, Naþ, and Sr2þ binding even on the neutral surfaces of both oxides. As negative charge increases, these cations move to binding sites and configurations closer to the surface, with a consequent increase in their coverage and degree of desolvation. Also, relatively more Rbþ, Naþ, and Sr2þ binds at the tetradentate site on rutile with
ARTICLE
increasing negative charge, which accounts for the enhanced negative charge development on rutile relative to cassiterite. However, because inner-sphere cation binding also predominates on the cassiterite surface, our results are at odds with the theories of James and Healy13 and Sverjensky11,12 which predict that outer-sphere adsorption should dominate on solids with low dielectric constants. Rather, it is the relative degree of cation desolvation upon inner-sphere adsorption (greater for rutile) that distinguishes the charging behavior of rutile and cassiterite in the presence of Rbþ, Naþ, and Sr2þ. Cation desolvation may be easier at the rutile (110) surface in part because inner-sphere or L2 water molecules are bound less tightly than on the cassiterite (110) surface. Hence, their removal to make way for inner-sphere cation binding is relatively more favorable. More such comparative studies, for other oxides and for a wider range of ions and other solution conditions, will help lead to improved ion adsorption frameworks. Larger-scale CMD simulations and DFT calculations would permit a wider range of solution conditions to be mimicked, particularly at lower ionic strengths, which might then help provide a molecular-level interpretation of ill-defined phenomena such as zeta potentials. Similarly, macroscopic charging and adsorption data can be quantified with surface complexation models that can incorporate corresponding molecular-level information such as the CDMUSIC model,2,35 with those modeling frameworks being improved as is necessary to make the greatest use of combined macroscopic-microscopic results.
’ ASSOCIATED CONTENT
bS
Supporting Information. Additonal information as described in the text. This material is available free of charge via the Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION Corresponding Author
*
[email protected]. Present Addresses
b School of Earth and Environment, University of Leeds, Leeds L52 9LT, United Kingdom 4 Intel Corporation, Portland, Oregon 2 U.S. Army Research Laboratory, Coatings, Corrosion, and Engineered Polymers Branch, 4600 Deer Creek Loop, Aberdeen Proving Ground, MD 21005
’ ACKNOWLEDGMENT This work was primarily supported by the U.S. Department of Energy, Office of Basic Energy Sciences, Division of Chemical Sciences, Geosciences, and Biosciences through the project “Nanoscale complexity at the oxide/water interface” (ERKCC41) under contract DE-AC05-00OR22725, Oak Ridge National Laboratory, managed and operated by UT-Battelle, LLC. X-ray measurements were performed at the XOR Sector 12, Advanced Photon Source, Argonne National Laboratory, which is supported by the U.S. Department of Energy grant DE-AC0206CH11357. Some of the rutile titration data were also collected with support from the National Science Foundation (EAR9627784). M. Predota was also supported by the Ministry of Education of the Czech Republic (ME09062). We thank three 4592
dx.doi.org/10.1021/la1040163 |Langmuir 2011, 27, 4585–4593
Langmuir anonymous reviewers for their comments, which resulted in a much improved final manuscript.
’ REFERENCES (1) Hiemstra, T.; Venema, P.; VanRiemsdijk, W. H. J. Colloid Interface Sci. 1996, 184, 680. (2) Hiemstra, T.; VanRiemsdijk, W. H. J. Colloid Interface Sci. 1996, 179, 488. (3) Brown, G. E., Jr.; Henrich, V. E.; Casey, W. H.; Clark, D. L.; Eggleston, C.; Felmy, A.; Goodman, D. W.; Graetzel, M.; Maciel, G.; McCarthy, M. I.; Nealson, K. H.; Sverjensky, D. A.; Toney, M. F. Chem. Rev. 1999, 99, 77. (4) Zhang, Z.; Fenter, P.; Cheng, L.; Sturchio, N. C.; Bedzyk, M. J.; Predota, M.; Bandura, A.; Kubicki, O. J. D.; Lvov, S. N.; Cummings, P. T.; Chialvo, A. A.; Ridley, M. K.; Benezeth, P.; Anovitz, L.; Palmer, D. A.; Machesky, M. L.; Wesolowski, D. J. Langmuir 2004, 20, 4954. (5) Vlcek, L.; Zhang, Z.; Machesky, M. L.; Fenter, P.; Rosenqvist, J.; Wesolowski, D. J.; Anovitz, L. M.; Predota, M.; Cummings, P. T. Langmuir 2007, 23, 4925. (6) Ross, N. L.; Spencer, E. C.; Levchenko, A. A.; Kolesnikov, A. I.; Wesolowski, D. J.; Cole, D. R.; Mamontov, E.; Vlcek, L. Neutron Applications in Earth, Energy and Environmental Sciences 2009, 235. (7) Mamontov, E.; Vlcek, L.; Wesolowski, D. J.; Cummings, P. T.; Wang, W.; Anovitz, L. M.; Rosenqvist, J.; Brown, C. M.; Garcia Sakai, V. J. Phys. Chem. C 2007, 111, 4328. (8) Bandura, A. V.; Kubicki, J. D.; Sofo, J. O. J. Phys. Chem. B 2008, 112, 11616. (9) Machesky, M. L.; Predota, M.; Wesolowski, D. J.; Vlcek, L.; Cummings, P. T.; Rosenqvist, J.; Ridley, M. K.; Kubicki, J. D.; Bandura, A. V.; Kumar, N.; Sofo, J. O. Langmuir 2008, 24, 12331. (10) Rosenqvist, J.; Machesky, M. L.; Vlcek, L.; Cummings, P. T.; Wesolowski, D. J. Langmuir 2009, 25, 10852. (11) Sverjensky, D. A. Geochim. Cosmochim. Acta 2001, 65, 3643. (12) Sverjensky, D. A. Geochim. Cosmochim. Acta 2005, 69, 225. (13) James, R. O.; Healy, T. W. J. Colloid Interface Sci. 1972, 40, 65. (14) Catalano, J. G.; Park, C.; Fenter, P.; Zhang, Z. Geochim. Cosmochim. Acta 2008, 72, 1986. (15) Ghose, S. K.; Waychunas, G. A.; Trainor, T. P.; Eng, P. J. Geochim. Cosmochim. Acta 2010, 74, 1943. (16) Mason, S. E.; Iceman, C. R.; Tanwar, K. S.; Trainor, T. P.; Chaka, A. M. J. Phys. Chem. C 2009, 113, 2159. (17) Yang, Z.; Li, Q.; Chou, K. C. J. Phys. Chem. C 2009, 113, 8201. (18) Ridley, M. K.; Machesky, M. L.; Palmer, D. A.; Wesolowski, D. J. Colloids Surf., A 2002, 204, 295. (19) Zhang, Z.; Fenter, P.; Kelly, S. D.; Catalano, J. G.; Bandura, A. V.; Kubicki, J. D.; Sofo, J. O.; Wesolowski, D. J.; Machesky, M. L.; Sturchio, N. C.; Bedzyk, M. J. Geochim. Cosmochim. Acta 2006, 70, 4039. (20) Zhang, Z.; Fenter, P.; Cheng, L.; Sturchio, N. C.; Bedzyk, M. J.; Machesky, M. L.; Anovitz, L. M.; Wesolowski, D. J. J. Colloid Interface Sci. 2006, 295, 50. (21) Benezeth, P.; Palmer, D. A.; Wesolowski, D. J.; Xiao, C. J. Sol. Chem. 2002, 31, 947. (22) Wesolowski, D. J.; Ziemniak, S. E.; Anovitz, L. M.; Machesky, M. L.; Benezeth, P.; Palmer, D. A. In Aqueous Systems at Elevated Temperatures and Pressures: Physical Chemistry in Steam and Hydrothermal Solutions; Palmer, D. A., Fernandez-Prini, R., Harvey, A. H., Eds.; Elsevier Ltd.: Amsterdam, 2004; p 493. (23) Kresse, G.; Furthm€uller, J. Phys. Rev. B 1996, 54, 11169. (24) Kresse, G.; Furthm€uller, J. Vasp the Guide, Institut f€ur Materialphysik Universit€at Wien, Wein Austria, 2003. (25) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865. (26) Kresse, G.; Joubert, D. Phys. Rev. B 1999, 59, 1758. (27) Bl€ochl, P. E. Phys. Rev. B 1994, 50, 17953. (28) Segall, M. D.; Lindan, P. J. D.; Probert, M. J.; Pickard, C. J.; Hasnip, P. J.; Clark, S. J.; Payne, M. C. J. Phys.: Condens. Matter 2002, 14, 2717.
ARTICLE
(29) Bolzan, A. A.; Fong, C.; Kennedy, B. J.; Howard, C. J. Acta Crystallogr., Sect. B 1997, 53, 373. (30) Predota, M.; Bandura, A. V.; Cummings, P. T.; Kubicki, J. D.; Wesolowski, D. J.; Chialvo, A. A.; Machesky, M. L. J. Phys. Chem. B 2004, 108, 12049. (31) Predota, M.; Zhang, Z.; Fenter, P.; Wesolowski, D. J.; Cummings, P. T. J. Phys. Chem. B 2004, 108, 12061. (32) Predota, M.; Vlcek, L. J. Phys. Chem. B 2007, 111, 1245. (33) Bandura, A. V.; Kubicki, J. D. J. Phys. Chem. B 2003, 107, 11072. (34) Bandura, A. V.; Sofo, J. O.; Kubicki, J. D. J. Phys. Chem. B 2006, 110, 8386. (35) Ridley, M. K.; Hiemstra, T.; van Riemsdijk, W. H.; Machesky, M. L. Geochim. Cosmochim. Acta 2009, 73, 1841. (36) Sposito, G. Environ. Sci. Technol. 1998, 32, 2815. (37) Jonsson, C. M.; Jonsson, C. L.; Sverjensky, D. A.; Cleaves, H. J.; Hazen, R. M. Langmuir 2009, 25, 12127. (38) Brown, G. E., Jr. Science (Washington, DC, U. S.) 2001, 294, 67.
4593
dx.doi.org/10.1021/la1040163 |Langmuir 2011, 27, 4585–4593
