Theoretical Study of Adsorption Sites on the (001) Surfaces of 1:1 Clay

Jeffery A. Greathouse , David B. Hart , and Margaret E. Ochs ... Adélia J. A. Aquino, Daniel Tunega, Martin H. Gerzabek, and Hans Lischka .... Adsorp...
0 downloads 0 Views 331KB Size
Langmuir 2002, 18, 139-147

139

Theoretical Study of Adsorption Sites on the (001) Surfaces of 1:1 Clay Minerals Daniel Tunega,*,†,‡ Georg Haberhauer,‡ Martin H. Gerzabek,‡ and Hans Lischka† Institute for Theoretical Chemistry and Structural Biology, University of Vienna, Wa¨ hringerstrasse 17, A-1090 Vienna, Austria, and Austrian Research Centers Seibersdorf, A-2444 Seibersdorf, Austria Received June 18, 2001. In Final Form: September 7, 2001 Quantum chemical calculations within the framework of a two-layered ONIOM approach using the B3LYP density-functional method were carried out for the study of adsorption sites on regular (001) surfaces of the kaolinite group of clay minerals. A molecular cluster of an appropriate size (about 80 atoms including H atoms saturating dangling bonds) was cut from a structure of a single kaolinite layer. This cluster contains two types of interaction sitessone at the tetrahedral side and one at the octahedral side of the layer. Interactions of water and acetic acid molecules with the tetrahedral side and of water molecule and acetate anion with the octahedral side were investigated. Geometry optimizations were carried out, and interaction energies between the clay surfaces and the molecular species were computed. Hydrogen bridges between the hydroxyl groups of the octahedral surface and the water molecule were observed with total adsorption energy about -8 kcal/mol. The carboxylate group of the acetate anion binds to the octahedral surface hydroxyls via hydrogen bridges also. The respective interaction energy amounts to about -70 kcal/mol. On the other hand, the water molecule and the acetic acid molecule interact with the tetrahedral (001) surface through weaker hydrogen bonds in comparison with those with the octahedral side. Adsorption energies are about -3 to -4 kcal/mol in these cases.

Introduction Clay minerals represent an important inorganic component of soils and affect significantly their physicochemical properties. Furthermore, we want to mention also their importance for industrial applications. Clays mainly exist as very small particles (∼2 µm in diameter) with a high specific surface area and a high chemical surface activity. Adsorption of mobile chemical species from soil solution on mineral surfaces is a very important process. Soil solutions represent chemically very complex systems containing, for example, inorganic ions and organic species originating from natural biochemical processes or from human activities. In many investigations special focus is paid to the behavior of toxically active species, such as radionuclides or pesticides, because of their potential adverse effects on the ecosystem.1 Among organic species, carboxylic acids and their derivatives play an important role.2,3 For example, one group of herbicides represents derivatives of the phenoxyacetic acid. Mainly the carboxyl group is responsible for their relatively high chemical activity. Organic acids affect the pH of soils and interact strongly either with other species in the solution, forming various associates and complexes, or with the surfaces of inorganic/organic solid particles, forming strongly bonded adsorbates.4 Moreover, some of the organic acids also strongly influence the dissolution of soil minerals.5-9 The interaction of water * To whom correspondence should be addressed. E-mail: Daniel. [email protected]. Fax: +43-1-4277-9527. Permanent address: Institute of Inorganic Chemistry, Slovak Academy of Sciences, Du´bravska´ cesta 9, SK-84236 Bratislava, Slovakia. †University of Vienna. ‡Austrian Research Centers Seibersdorf. (1) Schachtschabel, P.; Blume, H. P.; Bru¨mmer, G.; Hartge, K.-H.; Schwertmann, U. Lehrbuch der Bodenkunde; Ferdinand Enke Verlag: Stuttgart, 1998. (2) Huang, W. H.; Keller, W. D. Am. Mineral. 1971, 56, 1082-1095. (3) Welch, S. A.; Ullman, W. J. Geochim. Cosmochim. Acta 1993, 57, 2725-2763. (4) Kubicki, J. D.; Schroeter, L. M.; Itoh, M. J.; Nguyen, B. N.; Apitz, S. E. Geochim. Cosmochim. Acta 1999, 63, 2709-2725.

with clay surfaces plays an important role for soil processes. Therefore, a lot of experimental work has been devoted to the study of clay-water interactions, including the effects of wetting, swelling, intercalation, or ionic exchange.10 For example, X-ray powder diffraction, IR spectroscopy, NMR spectroscopy, differential scanning calorimetry, and thermal gravimetric analysis have been used in the investigation of hydrated kaolinites.11,12 In these investigations two kinds of water molecules in the interlayer space of kaolinite were observed and a dehydration enthalpy of -10.6 kcal/mol was obtained. From the measurement of the desorption isotherm of the waterkaolinite system13 a value of -4.1 kcal/mol was reported for the dehydration enthalpy. Adsorption and calorimetric studies14 of another water-kaolinite system reported a value of -7 kcal/mol for the adsorption enthalpy. Thus, considerable discrepancies in published values of adsorption/desorption processes are observed. The adsorption of small organic acid molecules has also been studied experimentally, for example, by the measurement of adsorption isotherms8,9 or with IR spectroscopy.4 Unfortunately, adsorption enthalpies are not available from these studies. Experimental information on surface complexes at a molecular level is scarce, especially when the clay surface is too complex. It is very difficult to distinguish by experimental means energetically different adsorption (5) Stumm, W.; Furrer, G.; Kunz, B. Croat. Chim. Acta 1983, 46, 593-611. (6) Fox, T. R.; Comerford, N. B.; McFee, W. W. Soil Sci. Soc. Am. J. 1990, 54, 1441-1447. (7) Wieland, E.; Stumm, W. Geochim. Cosmochim. Acta 1992, 56, 3339-3355. (8) Bhatti, J. S.; Comerford, N. B.; Johnston, C. T. Soil Sci. Soc. Am. J. 1998, 62, 152-158. (9) Ward, D. B.; Brady, P. V. Clays Clay Miner. 1998, 46, 453-465. (10) Rae, J.; Parker, A. In Enviromental Interactions of Clays; Rae, J., Parker, A., Eds.; Springer-Verlag: Berlin, 1988. (11) Costanzo, P. M.; Giesse, F. R., Jr.; Lipsicas, M. Clays Clay Miner. 1984, 32, 419-428. (12) Lipsicas, M.; Straley, C.; Constanzo, P. M.; Giesse, F. R., Jr. J. Colloid Interface Sci. 1985, 107, 221-230. (13) Khalfi, A.; Blanchart, P. Ceram. Int. 1999, 25, 409-414. (14) Poliakova, I. G.; Tarasevich, Yu. Kolloid. Zh. 1991, 53, 386-391.

10.1021/la010914e CCC: $22.00 © 2002 American Chemical Society Published on Web 12/08/2001

140

Langmuir, Vol. 18, No. 1, 2002

sites on mineral surfaces (e.g. regular (001) and edge surfaces) or to distinguish between outer surface adsorption and interlayer intercalation. Computer simulation methods are very useful for giving detailed insight into the above adsorption processes. Conventional molecular dynamics or Monte Carlo methods based on empirical potential functions have been used in the study of the interaction of water with clay minerals.15-24 The determination of a balanced set of parameters for these empirical potentials poses a particularly difficult problem here, especially concerning the parameters describing the interaction of aluminum, silicon, and oxygen atoms of the clay with the adsorbed molecules. Therefore, Teppen et al.25 have performed refinements of force field parameters. They have been applied in molecular dynamics calculations to the interactions between clays and some organic species.25-28 Ab initio calculations are much more reliable and do not suffer from parameter-fitting problems. However, they are much more time-consuming. Periodic ab initio pseudopotential calculations were performed on the talc-water and pyrophyllite-water systems.29 First principles molecular dynamics simulations were applied in the study of water adsorption at the surface of muscovite mica.30 In addition to calculations with explicit consideration of periodic boundary conditions, molecular cluster approaches were also used frequently.31 The advantage of cluster calculations is the fact that standard quantum chemical codes for molecular calculations can be used. Both semiempirical32-35 and ab initio investigations35-38 have been published so far. For example, densityfunctional theory (DFT) was used to study the smectitewater interface.35 The interaction of triaminotoulene with a model kaolinite surface was studied by means of the (15) Skipper, N. T.; Refson, K.; McConnell, J. D. C. J. Chem. Phys. 1991, 94, 7434-7445. (16) Delville, A.; Sokolowski, S. J. Phys. Chem. 1993, 97, 62616271. (17) Delville, A. J. Phys. Chem. 1995, 99, 2033-2037. (18) Bridgeman, C. H.; Skipper, N. T. J. Phys.: Condens. Matter 1997, 9, 4081-4087. (19) DeSiqueira, A. V. C.; Skipper, N. T.; Coveney, P. V. Mol. Phys. 1997, 92, 1-6. (20) DeSiqueira, A. V. C.; Skipper, N. T.; Coveney, P. V. Mol. Phys. 1998, 95, 123-123. (21) Skipper, N. T. Mineral. Mag. 1998, 62, 657-667. (22) Greathouse, J.; Sposito, G. J. Phys. Chem. B 1998, 102, 24062414. (23) Smirnov, K. S.; Bougeard, D. J. Phys. Chem. B 1999, 103, 52665273. (24) Shroll, R. M.; Smith, D. E. J. Chem. Phys. 1999, 111, 90259033. (25) Teppen, B. J.; Rasmussen, K.; Bertsch, P. M.; Miller, D. M.; Scha¨fer, L. J. Phys. Chem B 1997, 101, 1579-1587. (26) Teppen, B. J.; Yu, C.-H.; Miller, D. M.; Scha¨fer, L. J. Comput. Chem. 1998, 19, 144-153. (27) Yu, C.-H.; Newton, S. Q.; Norman, M. A.; Miller, D. M.; Scha¨fer, L.; Teppen, B. J. Clays Clay Miner. 2000, 48, 665-681. (28) Yu, C.-H.; Norman, M. A.; Newton, S. Q.; Teppen, B. J.; Miller, D. M.; Scha¨fer, L. J. Mol. Struct. 2000, 556, 95-103. (29) Bridgeman, C. H.; Buckingham, A. D.; Skipper, N. T.; Payne, M. C. Mol. Phys. 1996, 89, 879-888. (30) Odelius, M.; Bernasconi, M.; Parrinello, M. Phys. Rev. Lett. 2000, 78, 2855-2858. (31) Sauer, J. Chem. Rev. 1989, 89, 199-255. (32) Delville, A. Langmuir 1991, 7, 547-555. (33) Delville, A. Langmuir 1992, 8, 1796-1805. (34) Gorb, L. G.; Aksenenko, E. V.; Adams, J. W.; Larson, S. L.; Weiss, C. A.; Leszczynska, D.; Leszczynski, J. J. Mol. Struct. (THEOCHEM) 1998, 425, 129-135. (35) Chatterjee, A.; Iwasaki, T.; Ebina, T.; Hayashi, H. Appl. Surf. Sci. 1997, 121/122, 167-170. (36) Zhanpeisov, N. U.; Adams, J. W.; Larson, S. L.; Weiss, C. A.; Zhanpeisova, B. Z.; Leszczynska, D.; Leszczynski, J. Struct. Chem. 1999, 10, 285-294. (37) Pelmenschikov, A.; Leszczynski, J. J. Phys. Chem B 1999, 103, 6886-6890. (38) Gorb, L. G.; Gu, J.; Leszczynska, D.; Leszczynski, J. Phys. Chem. Chem. Phys. 2000, 2, 5007-5012.

Tunega et al.

Hartree-Fock (HF) approach using a 6-31G basis set;36 the HF approach, DFT, and Møller-Plesset perturbation theory to the second order (MP2) were applied in the study of adsorption of 1,3,5-trinitrobenzene on the siloxane sites of clay minerals;37 and the DFT approach was used in the study of the nitrobenzene-montmorillonite system.38 Even though several investigations of interactions on clay surfaces have been performed, deeper insight into basic adsorption structures even of simple molecules such as water is still missing. It was the purpose of the present work to perform benchmark calculations on the adsorption of selected prototype molecules on well-defined clay surfaces such as the (001) surfaces of kaolinite. Kaolinite minerals are of 1:1 type layered silicates. Thus, they allow the study of two chemically different surfaces, one covered with hydroxyl groups and the other one covered with oxygen atoms. As the first example of adsorbate molecules we selected the water molecule because of its fundamental importance for all solvation processes. As the second set of examples we chose acetic acid (HAc) and acetate anion (Ac-), sincesas has been outlined abovesthe carboxyl group is largely responsible for the activity of organic acids in soil processes. Cluster calculations were carried out on adsorption complexes. The DFT method in combination with flexible basis sets has been chosen for the calculation of the interaction of the adsorbate with the clay surface in order to obtain reliable results. The ONIOM method,39 which has been especially developed for calculations on large molecules or clusters, was applied to allow the use of sufficiently large clusters and still keep the computational effort manageable. In this way we could characterize accurately different adsorption structures on the tetrahedral and octahedral sheets of a kaolinite layer and discuss the formation of hydrogen bonds. Such detailed information is important for the understanding of adsorption processes at a molecular level. These model investigations should lay the basis for future computations on more extended models concerning, for example, solvent effects and/or interactions of more complicated molecules such as pesticides with mineral surfaces. Structural and Computational Details Minerals of the kaolinite group (typical representatives are kaolinite and dickite) have a 1:1 dioctahedral structure.40 They have a common chemical formula Al2Si2O5(OH)4, and they differ only in the layer stacking. The unit cell of dickite consists of two kaolinite layers and is twice as large as the unit cell of kaolinite. An individual layer consists of two connected sheetssa tetrahedral sheet formed from SiO4 tetrahedra sharing corners and an octahedral sheet consisting of AlO6 octahedra sharing edges. Both sheets share a common plane of apical oxygen atoms. One-third of all possible octahedral central positions are empty, resulting in octahedral cavities also causing deformations of the other octahedra occupied by aluminum atoms. Hydroxyl groups participating in hydrogen bonds with basal oxygen atoms of an adjacent layer cover the outer surface of the octahedral sheet. These hydrogen bonds are the main source for the cohesive energy between layers. The tetrahedral side of a layer is characterized by ditrigonal cavities. This surface is built from the plane of basal oxygen atoms each shared by two silicon atoms. These two planes (tetrahedral basal oxygen (39) Dapprich, S.; Komaromi, I.; Byun, K. S.; Morokuma, K.; Frisch, M. J. J. Mol. Struct. (THEOCHEM) 1999, 462, 1-21. (40) Bailey, S. W. Structures of layered silicates. In Crystal structures of clay minerals and their X-ray identification; Brindley, G. W., Brown, G., Eds.; Mineralogical Society: London, 1980.

Sites on the (001) Surfaces of 1:1 Clay Minerals

Langmuir, Vol. 18, No. 1, 2002 141

Figure 1. Relaxed structure of the isolated kaolinite layer: top (upper) and side (lower) views.

atoms and octahedral surface hydroxyl groups) are parallel to the crystallographic (001) surface and can be considered as electronically saturated. The interlayer space in kaolinite and dickite is empty, since the layers are neutral and no compensating cations are required. The typical thickness of a single layer is about 5.4 Å, and the distance between two layers is about 2 Å. Figure 1 shows two different views of the structure of a single kaolinite layer. This visualized structure was obtained from the total structural relaxation of the isolated layer using ab initio periodic calculation (for more details see ref 41). Both surfaces (basal oxygen atoms on the tetrahedral side and surface hydroxyl groups on the octahedral side) are of different chemical nature, and a different behavior is expected in their contact with polar liquids. The cluster model of the single kaolinite layer was derived from the structure of the mineral dickite.42 The model consists of 78 atoms and contains one ditrigonal tetrahedral ring and one octahedral ring. All dangling bonds were saturated with hydrogen atoms. The resulting chemical formula is Si6Al6O36H30. The structure of this cluster is displayed in Figures 2-5. The cluster is denoted in the text as D(O) or D(T) to resolve which surface interacts with studied molecules. The two-level ONIOM method39 as implemented in the Gaussian98 package43 was used for the cluster calculations. The total energy of the cluster C is written as LL HL EONIOM ) ELL C C - EIP + EIP

(1)

where IP designates the inner part of the cluster. LL and HL stand for the low-level and high-level method, respectively. The interaction energy is defined analogously from calculations on the whole cluster and the individual (41) Benco, L.; Tunega, D.; Hafner, J.; Lischka, H. Am. Mineral. 2001, 86, 1057-1065. (42) Joswig, W.; Drits, V. A. N. Jb. Miner. Mh. 1986, 19-22. (43) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A., Jr.; Stratmann, R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A. D.; Kudin, K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi, M.; Cammi, R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.; Ochterski, J.; Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.; Ortiz, J. V.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe, M.; Gill, P. M. W.; Johnson, B. G.; Chen, W.; Wong, M. W.; Andres, J. L.; Head-Gordon, M.; Replogle, E. S.; Pople, J. A. Gaussian 98, revision A.7; Gaussian, Inc.: Pittsburgh, PA, 1998.

Figure 2. Dickite(octahedral)-H2O system. ONIOM(B3LYP/ SVP:PM3) optimized geometry of the model (b) in Table 1. Top and two side views. The stick model represents the inner ONIOM part, and the wire model, the outer ONIOM part.

subsystems as LL HL ∆EONIOM ) ∆ELL C C - ∆EIP + ∆EIP

(2)

142

Langmuir, Vol. 18, No. 1, 2002

Figure 3. Dickite(octahedral)-Ac- system. ONIOM(B3LYP/ SVP:PM3) optimized geometry of the model (b) in Table 1. Top and two side views. The stick model represents the inner ONIOM part, and the wire model, the outer ONIOM part.

Two different ONIOM partitionings of the cluster have been tested. Abbreviations used and characterizations of the inner ONIOM part are given in Table 1. More detailed

Tunega et al.

Figure 4. Dickite(tetrahedral)-H2O system. ONIOM(B3LYP/ SVP:PM3) optimized geometry. Top and two side views. The stick model represents the inner ONIOM part, and the wire model, the outer ONIOM part.

explanation of the ONIOM partitioning will be presented together with the discussion of the results. The geometry optimizations were performed at the ONIOM(B3LYP/SVP:PM3) level. This notation indicates that DFT with the B3LYP functional44 and the polarized

Sites on the (001) Surfaces of 1:1 Clay Minerals

Langmuir, Vol. 18, No. 1, 2002 143 Table 1. List of Adsorption Complexes, Abbreviations Used in the Text, and Chemical Formulas of the Inner ONIOM Layer investigated system

abbreviation

dickite(tetrahedral side)-H2O D(T)-H2O dickite(tetrahedral side)-HAc D(T)-HAc dickite(octahedral side)-H2O D(O)-H2O-a D(O)-H2O-b dickite(octahedral side)-AcD(O)-Ac-a D(O)-Ac-b

composition of the inner ONIOM layera Si6O6H12 Si6O6H12 Al6(OH)8O4H16 Al6(OH)14O10H16 Al6(OH)8O4H16 Al6(OH)14O10H16

a With exclusion of the adsorbed molecule but with inclusion of terminal hydrogen atoms.

In the case of adsorption on the tetrahedral side of the kaolinite layer, the geometry of the clay-layer cluster was kept fixed at the experimental geometry.42 Only the geometry of the adsorbate and its relative position with respect to the clay cluster were optimized. This procedure was chosen in order to avoid artificial edge effects from the geometry optimization of the cluster. Moreover, as will be demonstrated later in the text, the interaction of the adsorbate molecules with the tetrahedral sheet is relatively weak. Therefore, only very small deformations of the surface structure are to be expected. For the interactions on the octahedral side, the positions of the six nearest-neighbor hydroxyl groups of the octahedral sheet were optimized also, since some of them were involved in hydrogen bonds with the adsorbate. For the calculation of interaction energies, the geometries of the subsystems were relaxed under the same conditions as were used for the whole system. This means that the adsorbates were fully optimized and the geometries of the clay layer fragments were partially relaxed in the case of adsorption at the octahedral surface and not relaxed for adsorption at the tetrahedral surface. As follows from the above-mentioned conditions of the geometry relaxation, the position of the inner hydroxyl group was also fixed at the estimated experimental geometry,42 since the distance of the proton of this inner hydroxyl group to any studied adsorbates is more than 3 Å and no strong influence of this hydroxyl group on adsorbates is expected. Of course, additional stabilization of the studied systems will be observed but no significant changes in the formation of hydrogen bonds will occur. Necessary fixing of some part of the clusters and possible artificial edge effects can be circumvented by performing calculations with periodic boundary conditions. Such an approach was used, for example, in the studies41,48-50 of hydroxyl groups in kaolinite and dickite. Investigations on the adsorbed systems studied in this work using first-principle calculations with periodic boundary conditions are in progress. To improve the quality of the SVP basis set with respect to the basis set superposition error (BSSE),51 the basis set for the heavy atoms directly involved in the adsorbentadsorbate binding was augmented with one additional set of s and p functions. The exponents of these functions were obtained by dividing the smallest respective exponent of the SVP basis by a factor of 3. The abbreviation SVP+sp

Figure 5. Dickite(tetrahedral)-HAc system. ONIOM(B3LYP/ SVP:PM3) optimized geometry. Top and two side views. The stick model represents the inner ONIOM part, and the wire model, the outer ONIOM part.

split-valence basis (SVP)45 was used for the inner ONIOM layer and the semiempirical PM3 method46,47 for the outer layer.

(44) Becke, A. D. J. Chem. Phys. 1993, 98, 5648-5652. (45) Scha¨fer, A.; Horn, H.; Ahlrichs, R. J. Chem. Phys. 1992, 97, 2571-2577. (46) Stewart, J. J. P. J. Comput. Chem. 1989, 10, 209-220. (47) Stewart, J. J. P. J. Comput. Chem. 1989, 10, 221-264. (48) Hess, A. C.; Saunders, V. R. J. Phys. Chem. 1992, 96, 43674374. (49) Hobbs, J. D.; Cygan, R. T.; Nagy, K. L.; Schultz, P. A.; Sears, M. P. Am. Mineral. 1977, 82, 657-662. (50) Benco, L.; Tunega, D.; Hafner, J.; Lischka, H. Chem. Phys. Lett. 2001, 333, 479-484. (51) Boys, S. F.; Bernardi, F. Mol. Phys. 1970, 19, 553-566.

144

Langmuir, Vol. 18, No. 1, 2002

Tunega et al.

Figure 6. Contour map of the electrostatic potential in the plane parallel to the (001) plane located 1.5 Å above the oxygen atoms from the octahedral surface hydroxyl groups.

Figure 7. Contour map of the electrostatic potential in the plane parallel to the (001) plane located 1.5 Å above the basal oxygen atoms from the tetrahedral side.

is used for this basis set. Test calculations on the aluminum-hexaaquo complex52 and a set of hydrogenbonded complexes involving acetic acid, acetate anion, acetaldehyde, acetamide, methanol, and phenol53 have shown that the BSSE is substantially decreased in this way. This SVP+sp basis set was used in single-point ONIOM calculations. The accuracy of the ONIOM approach has been tested in two ways. First, for D(O)-H2O and D(O)-Ac- two different sizes of the inner ONIOM layer were chosen. A comparison between the two choices is given in the following section. Second, several computational methods (the semiempirical MNDO,54 PM3,46,47 and AM155 methods and HF/STO-3G56 and HF/SV45) were used for the outer ONIOM layer. To save computer time, the ONIOM calculations with the MNDO, AM1, HF/STO3G, and HF/SV methods were performed as single-point calculations for the ONIOM(B3LYP/SVP:PM3) optimized geometries only. Additionally, single-point B3LYP/SVP and B3LYP/SVP+sp calculations have also been performed for all clusters without using the ONIOM approach. In all calculations interaction energies were corrected for the BSSE using the counterpoise method.51

hydroxyl groups surround a trigonal hole in the octahedral sheet. Figure 7 represents the contour map of the electrostatic potential above the plane of the basal oxygen atoms from the tetrahedral side of the layer at a distance of 1.5 Å. The electrostatic potential at this distance is everywhere negative. The shape of the electrostatic potential reflects the ditrigonal hole in the tetrahedral sheet surrounded by six basal oxygen atoms. Minima are observed above the oxygen and silicon atoms and are not equivalent, since these atoms do not lie in the same plane. One very flat minimum is also observed above the center of the ditrigonal hole. The shape and the symmetry of the electrostatic potential indicate that in this case several, nearly equivalent, local minima will exist for adsorbed molecules. We did not investigate all possibilities but constrained our search for local minima by some reasonable guesses for starting geometries. These starting geometries were constructed in the following way. In the case of the octahedral surface, the water molecule was oriented such that the oxygen atom was placed above the center of the trigonal hole and both water protons were directed out of the layer surface. In this way the water molecule was located in an attractive region of the surface and the three hydrogen atoms of the surface hydroxyl groups have the possibility to form hydrogen bonds. The acetate anion was localized in the similar manner. One oxygen atom of the carboxylate group was placed above the trigonal hole. The anion was rotated in such a way that the C-C bond was perpendicular to the octahedral surface and the second carboxylate oxygen was directed toward the middle hydrogen atom in the top part of Figure 6. In the case of the tetrahedral surface (see Figure 7), we placed the water molecule at the beginning of the geometry optimization with the oxygen atom above the center of the ditrigonal cavity with protons directed toward the plane of the basal oxygen atoms. The acetic acid molecule was placed above the tetrahedral side in such a way that the oxygen atom of the carboxyl group was located above the center of the ditrigonal cavity and the proton of the carboxyl group was directed toward one of the basal oxygen atoms.

Results and Discussion In the first step we investigated the shape of the electrostatic potential above both surfaces of the cluster representing the kaolinite layer. Figure 6 represents the electrostatic potential in the plane parallel to the (001) plane located 1.5 Å above the oxygen atoms of the octahedral surface hydroxyl groups. One can see that positive maxima localized above the hydrogen atoms of the surface hydroxyl groups are dominating the map. Three maxima above three hydroxyl groups are located above the center of the layer fragment. These three (52) Tunega, D.; Haberhauer, G.; Gerzabek, M.; Lischka, H. J. Phys. Chem. A 2000, 104, 6824-6833. (53) Aquino, A. J. A.; Tunega, D.; Gerzabek, M. H.; Haberhauer, G.; Lischka, H. Submitted. (54) Dewar, M. J. S.; Thiel, W. J. Am. Chem. Soc. 1977, 99, 48994907. (55) Dewar, M. J. S.; Zoebisch, E. G.; Healy, E. F.; Stewart, J. J. P. J. Am. Chem. Soc. 1985, 107, 3902-3909. (56) Hehre, W. J.; Stewart, R. F.; Pople, J. A. J. Chem. Phys. 1969, 51, 2657-2664.

Sites on the (001) Surfaces of 1:1 Clay Minerals Table 2. ONIOM(B3LYP/SVP:PM3) Interaction Energies ∆E (BSSE Uncorrected) for the D(O)-H2O and D(O)-AcSystems Computed with Two Different Sizes of the Inner ONIOM Region

a

systema

∆E (kcal/mol)

D(O)-H2O-a D(O)-H2O-b D(O)-Ac-a D(O)-Ac-b

-17.8 -22.1 -60.5 -65.7

For definition see Table 1.

At the beginning of our investigations two choices for the size of the inner ONIOM part (cases a and b in Table 1) were examined in order to evaluate the quality of the ONIOM calculations. The interaction with the octahedral side (adsorption complexes D(O)-Ac- and D(O)-H2O) was tested only. Linking atoms were replaced by hydrogen atoms to saturate the ONIOM inner part. In case a the linking atoms between the inner and outer ONIOM part are apical oxygen atoms (see description in Figure 1). In this model all aluminum atoms are included in the inner ONIOM part. However, the octahedral coordination of aluminum is destroyed by the ONIOM partitioning. Therefore, in case b the silicon atoms are the border atoms and the entire octahedral sheet is included in the inner part. Interaction energies obtained from ONIOM(B3LYP/ SVP:PM3) calculations are given in Table 2. The difference in the adsorption energies computed between models a and b amounts to values between 4 and 5 kcal/mol. This is a relatively small discrepancy in the case of acetate adsorption but quite large (25%) in the case of water adsorption. The positions and geometries of the water and acetate molecules above the hydroxyl surface were much less sensitive to the ONIOM partitioning. In view of this situation we considered it more consistent to use the more extended model b in all later calculations. A similar, extended partitioning was applied for interactions with the tetrahedral side. In all cases the adsorbed molecule is included in the ONIOM inner part. A stick model is used in Figures 2-5 to visualize the inner ONIOM parts, and a wire model, to display the outer ONIOM parts. A. Structures. In Figures 2 and 3 optimized geometries and the most important geometrical parameters for adsorption on the octahedral side of the kaolinite layer are displayed. Figure 2 shows that the water molecule is located directly above the center of the octahedral cavity. Three surface hydroxyl groups, which are surrounding the hole in the octahedral sheet, are involved in hydrogen bonding with the adsorbed water molecule. Two of them act as proton donors to the water oxygen atom at O‚‚‚H distances of about 1.9 Å. They are, therefore, tilted up with respect to the cavity. The oxygen atom of the third hydroxyl group is the proton acceptor for one proton of the water molecule. Therefore, the hydrogen atom of this hydroxyl group is tilted out in contrast to the previous case. The O1‚‚‚H5 distance is 1.628 Å, which is shorter than the two previous ones and thus indicates a significantly stronger hydrogen bond in comparison to the other two hydrogen bonds. The charge transfer through these hydrogen bonds also illustrates this situation. In Table 3 Mulliken net atomic charges57,58 obtained at the B3LYP/ SVP+sp level are displayed for all adsorbed complexes and their subsystems. Quite generally, as expected, the positive charge of the hydrogen atoms involved in hydrogen bonding increases with respect to those for the isolated systems. In the D(O)-H2O case (Table 3a), the largest (57) Mulliken, R. J. Chem. Phys. 1955, 23, 1833-1840. (58) Mulliken, R. J. Chem. Phys. 1955, 23, 1841-1846.

Langmuir, Vol. 18, No. 1, 2002 145

increase (0.214 f 0.336) occurs for proton H5 of the water molecule involved in the shortest hydrogen bond. The negative charge of the oxygen atom O5 of the water molecule is increased because of the charge transfer from the surface hydroxyl groups. In total, the water and surface OH bonds involved in hydrogen bonds are significantly polarized. The other O5H6 bond of the water molecule not involved in any contact with the octahedral surface is polarized too. It is practically free and is prepared to interact with another molecules approaching the octahedral surface. The present analysis shows clearly the bifunctional ability of surface hydroxyl groups, which can act either as proton donor or as proton acceptor. In Figure 3 the optimized structure of the acetate adsorption complex is displayed. The acetate anion is also located above the center of the octahedral hole. The C-C bond is oriented almost perpendicularly to the hydroxyl surface of the layer with the methyl group pointing away from the surface. Both oxygen atoms of the carboxylate group represent proton acceptors and are involved in hydrogen bonds with the four adjacent surface hydroxyl groups. Three O‚‚‚H bond lengths have values of about 1.75 Å and represent the interaction with the same carboxylate oxygen atom while the fourth one has a bond length of about 2.00 Å. The last one is longer than the three previous ones, since it interacts with a more distant hydroxyl group. The polarization of the interacting surface hydroxyl groups is even stronger than that in the case of the dickite-water system, since a charged system interacts with the octahedral surface. This is again documented by the Mulliken net atomic charges (Table 3b). The net atomic charges increase for all atoms of the hydroxyl groups involved in the interaction. In the acetate anion a strong polarization of the C-C and C-H bonds is observed and a large amount of charge (about 0.17e) is transferred to the surface hydroxyl groups. Figures 4 and 5 present views of the adsorption complexes D(T)-H2O and D(T)-HAc on the tetrahedral side of the kaolinite layer. The tetrahedral sheet is electronically fully saturated and only relatively weak interactions are expected. The calculations with water and acetic acid as adsorbates confirmed this assumption. The water molecule is not located directly above the center of the ditrigonal-tetrahedral hole but is shifted to the side of the ditrigonal ring in order to form better contact with two neighboring oxygen atoms in that ring. The water molecule is oriented with hydrogen atoms directed toward two basal oxygen atoms, and the H-H vector is almost parallel to the plane of the basal oxygen atoms. The distances between the protons of the water molecule and the contacting basal oxygen atoms are 2.142 and 2.515 Å, respectively. These large distances (cf. the respective bond lengths of 1.75 Å on the octahedral side of the layer) indicate a relatively weak interaction of the water molecule with the tetrahedral surface. The interaction of the acetic acid molecule with the tetrahedral surface is very similar to that in the case of adsorption of water. The molecule is positioned nearly above the center of the ditrigonal hole and forms a single, weak hydrogen bond (1.940 Å in length) with one basal oxygen atom. All relevant geometrical parameters are displayed in Figure 3. Calculated partial atomic charges (Table 3c,d) show only small charge density redistributions between the adsorbed molecules and the tetrahedral surface. Also, the polarization of adsorbed molecules is smaller in comparison with that for the interactions on the octahedral surface. B. Adsorption Energies. Several low-level methods have been selected for the description of the outer ONIOM layer in order to see how those methods affect the

146

Langmuir, Vol. 18, No. 1, 2002

Tunega et al.

Table 3. Mulliken Atomic Charges of Atoms Involved in the Interactions in All Studied Systemsa (a) D(O)-H2O D(O)-H2O dickite H2O

O1

O2

O3

-1.254 -1.246

-1.215 -0.971

-1.265 -1.246

O4

H1

H2

H3

0.249 0.302

0.314 0.297

0.345 0.302

H4

O5

H5

H6

-0.502

0.336

0.242

-0.428

0.214

0.214

(b) D(O)-AcD(O)-Acdickite Ac-

Hb

O1

O2

O3

O4

H1

H2

H3

H4

O5

O6

C1

C2

-1.383 -1.246

-1.061 -0.971

-1.338 -1.246

-1.091 -1.021

0.409 0.302

0.368 0.297

0.398 0.302

0.403 0.301

-0.593

-0.248

-0.251

0.002

0.085

-0.448

-0.546

-0.019

-0.027

0.085

(c) D(T)-H2O D(T)-H2O dickite H2O

O1

O2

O3

O4

O5

O6

O7

H1

H2

-1.329 -1.045

-1.171 -1.133

-0.968 -0.994

-1.134 -1.133

-1.152 -1.044

-1.446 -1.342

-0.469

0.263

0.256

-0.428

0.214

0.214

(d) D(T)-HAc D(T)-HAc dickite Ac-

O1

O2

O3

O4

O5

O6

H1

O7

O8

C1

C2

Hb

-1.053 -1.045

-1.147 -1.133

-0.989 -0.994

-1.121 -1.133

-1.052 -1.044

-1.479 -1.342

0.336

-0.350

-0.203

-0.120

-0.025

0.080

0.259

-0.347

-0.276

0.167

0.032

0.055

a

The numbering of the atoms corresponds to the numbering of the atoms in Figures 2-5. b Averaged value for protons of the methyl group. Table 4. Dependence of the ONIOM(B3LYP/SVP:LL) Interaction Energy on the Low-Level Method (LL) for All Studied Systems (All Values Are in kcal/mol) outer parta LL method

a

∆ELL C

-

∆ELL IP

HLb (B3LYP) method ∆EHL IP

HL c EIP,BSSE

total d ∆EHL IP,corr

∆EONIOM C

d ∆EONIOM C,corr

PM3 AM1 MNDO HF/STO-3G B3LYP/SV

-1.2 0.8 0.6 1.3 1.3

-20.9

(a) D(O)-H2O -10.3

-10.6

-22.1 -20.1 -20.3 -19.7 -19.6

-11.8 -9.8 -10.0 -9.4 -9.0

PM3 AM1 MNDO HF/STO-3G B3LYP/SV

9.5 -7.4 -11.1 -12.9 -19.5

-75.1

(b) D(O)-Ac-16.1

-59.0

-65.7 -82.6 -86.2 -88.0 -94.6

-49.5 -66.4 -70.1 -71.9 -78.5

PM3 AM1 MNDO HF/STO-3G B3LYP/SV

-2.5 -3.5 -3.1 -2.2 -3.8

-4.9

(c) D(T)-H2O -3.6

-1.4

-7.4 -8.4 -8.1 -7.1 -8.7

-3.9 -4.9 -4.5 -3.6 -5.2

PM3 AM1 MNDO HF/STO-3G B3LYP/SV

-3.3 -3.5 -3.7 -3.9 -4.8

-2.8

(d) D(T)-HAc -3.6

0.7

-6.2 -6.3 -6.5 -6.7 -7.6

-2.6 -2.8 -2.9 -3.1 -4.0

C ) whole cluster; IP ) inner part. b High-level method. c Energy of BSSE.

calculated interaction energies. The ONIOM(B3LYP/SVP: PM3) geometries have been used for that purpose. Results are presented in Table 4, where the total interaction energy is split up into individual components according to eq 2. The second column in Table 4 shows the contribution of the outer ONIOM part to the total interaction energy computed with the low-level method. For neutral systems (Table 4, cases a, c, and d), these contributions are relatively small as compared to those for the charged system D(O)-Ac-. However, they have different relative weights with respect to the total interaction energies. While for the stronger interacting systems D(O)-H2O and D(O)-Ac- this relative weight is less than 15%, it is significantly larger for more weakly interacting systems (about 50% for D(T)-HAc). For the weakly interacting systems D(T)-H2O and D(T)-HAc a relatively small

d

BSSE corrected value.

influence of the choice of the low-level method is found. For the strongly interacting systems D(O)-H2O and D(O)-Ac-, however, the low-level results differ significantly between PM3 on the one side and the remaining methods on the other. The largest difference is observed for the charged D(O)-Ac- system. This fact points to inadequacies coming from the long-range Coulomb interaction between the acetate anion and the outer ONIOM part. Table 4 shows also BSSE corrections calculated for the ONIOM inner part. The BSSE is especially important for the weakly interacting systems. For the strongly interacting ones, the BSSE energy is also quite high but has a smaller relative influence than that in the previous case. The last two columns in Table 4 contain final, uncorrected

Sites on the (001) Surfaces of 1:1 Clay Minerals

Langmuir, Vol. 18, No. 1, 2002 147

ONIOM Table 5. BSSE Corrected Interaction Energies ∆EC,corr Using the SVP and SVP+sp Basis Sets for the Inner ONIOM Part, and BSSE Corrected Interaction Energies ∆EC,corr for the Whole Cluster without ONIOM Approximationa,b

a

system

∆EONIOM C,corr (B3LYP/SVP:MNDO)

∆EC,corr (B3LYP/SVP)

∆EONIOM C,corr (B3LYP/SVP+sp:MNDO)

∆EC,corr (B3LYP/SVP+sp)

D(O)-H2O D(O)-AcD(T)-H2O D(T)-HAc

-10.0 (-10.3) -73.0 (-16.1) -4.5 (-3.6) -2.9 (-3.6)

-10.5 (-6.7) -70.1 (-16.1) -3.8 (-4.5) -2.6 (-5.9)

-7.2 (-9.8) -69.7 (-9.1) -4.7 (-1.5) -4.1 (-2.1)

-8.3 (-5.7) -67.1 (-8.1) -4.1 (-2.7) -2.8 (-2.1)

Energies are in kcal/mol. b BSSE is given in parentheses.

ONIOM ∆EONIOM and BSSE-corrected ∆EC,corr interaction enerC gies. Since the BSSE played an important role for the calculation of adsorption energies, additional single-point calculations using the extended SVP+sp basis set were carried out. SVP and SVP+sp results are compared in Table 5. Moreover, B3LYP single-point calculations have been performed on the entire cluster without ONIOM approximation. Computed adsorption energies are shown in Table 5 as well. The calculations with the SVP+sp basis set show a significant reduction of the BSSE by about 50% except for the D(O)-H2O case, where the effect is much smaller. The B3LYP calculations on the full system serve as reference for the ONIOM approach. Only relatively small differences between ONIOM and full calculation are observed. Extension of the basis set to SVP+sp causes a small shift of the interaction energies. They decrease in absolute value for the strongly interacting systems on the octahedral side and increase for the weakly interacting systems on the tetrahedral side. The adsorption energy of water on the tetrahedral side of the kaolinite layer is about -4.5 kcal/mol after BSSE correction (see Table 5). Since this interaction involves two hydrogen bonds, the energy per hydrogen bond is about -2.3 kcal/mol. Therefore, this hydrogen bond is much weaker than that in the water dimer, for which we computed an interaction energy of -4.7 kcal/mol at the B3LYP/SVP+sp level including BSSE correction. Such weak interaction of the water molecule with the tetrahedral side indicates hydrophobic character of the tetrahedral surface. Unfortunately, direct experimental evidence for the kaolinite group minerals does not exist. However, for example, very similar tetrahedral surfaces occur in minerals of the 2:1 type (two tetrahedral sheets are connected to one octahedral sheet) like talc or pyrophyllite. Contact angle measurements59 on the surfaces of these minerals showed their low-energetic and hydrophobic character, a finding which agrees with our conclusion for the tetrahedral kaolinite surface. With an interaction energy of -2.8 kcal/mol, the interaction of the acetic acid molecule with the tetrahedral surface is even weaker than that in the water case. This confirms the low-energetic character of the tetrahedral surface as well. Interactions with the octahedral side of the layer are much stronger than those with the tetrahedral one. This could already be seen from the discussion of structures in the previous subsection. The adsorption of the water molecule is about a factor of 2 stronger on the octahedral side as compared to the tetrahedral one. This factor of 2 as compared to the three hydrogen bonds formed in the adsorption complex also indicates that not all three bonds could be formed optimally because of steric reasons. The adsorption of the acetate anion is significantly stronger because of the negative charge involved in the interaction and the fact that both oxygen atoms of the carboxylate group are very strong proton acceptors. The high adsorption energy of the water molecule indicates that the

(59) Schrader, M. E.; Shmuel, Y. J. Colloid Interface Sci. 1990, 136, 85-94.

octahedral surface has hydrophilic character and is of the high-energy type and also confirms the very high adsorption energy of the acetate anion. We have shown that the two types of the (001) surfaces of the kaolinite layer are energetically very different and adsorption of the water molecules would occur preferentially on the octahedral surface. Available experimental data of adsorption/desorption energies for water mentioned in the Introduction are in the range -4 to -11 kcal/mol. Our calculated adsorption energies for the water-octahedral surface system (e.g. -8.3 kcal/mol at the B3LYP/SVP+sp level) fall into that interval. However, it is clear that our model calculations do not contain the full complexity of the real systems. Conclusions Systematic quantum chemical calculations have been performed for the study of the adsorption of water, acetate anion, and acetic acid on the tetrahedral and octahedral surfaces of a single kaolinite layer. The water molecule and the acetate anion each form several relatively strong hydrogen bonds with the surface hydroxyls on the octahedral side with interaction energies of -8 and -70 kcal/mol, respectively. These values should be typical for individual interactions with neutral and ionized carboxylic acids. The calculated adsorption energy for the water molecule is in the interval of experimental sorption/ desorption energies.11-14 On the other hand, the water molecule and the acetic acid are only weakly bound to the basal oxygen atoms on the tetrahedral side with interaction energies of about -3 to -4 kcal/mol. From the analysis of computed structures of adsorbed complexes, adsorption energies, and electronic charge distributions, we have shown that the two (001) surfaces (octahedral and tetrahedral) of the kaolinite layer are of different chemical nature. This has been quantified in terms of structural parameters of adsorption complexes and of interaction energies. The octahedral surface is more highly energetic than the tetrahedral one and is more attractive for polar and/or negatively charged species. It offers more possibilities to form hydrogen bonds with adsorbates than the tetrahedral side due to the existence of hydroxyl groups on its surface. These OH groups have a bifunctional character and can act as proton donor and/or proton acceptor, respectively. Hydrogen bonds formed on this octahedral surface can even be stronger because of cooperative effects. Acknowledgment. This work was supported by the Austrian Science Fund, project no. P12969-CHE. We are grateful for technical support and computer time at the DEC Alpha server of the computer center of the University of Vienna. LA010914E