Res. Chem. lntermed., Vol. 24, No. 2, pp. 169-181 (1998) 9 VSP 1998

INTEGRATED COMPUTATIONAL CHEMISTRY STUDY FOR ZEOLITE MICROPOROUS MATERIALS A. MIYAMOTO*, A. CHATTERJEE, M. KUBO, H. TAKABA and Y. OUMI Department of Materials Chemistry, Graduate School of Engineering, Tohoku University, Aoba-ku, Aramaki, Sendal 980-77, Japan Received 16 September 1997; accepted 4 November 1997

INTRODUCTION

The knowledge of the structure, dynamics and energies of solid porous catalysts under the reaction conditions are fundamentally important information to develop efficient catalysts. Computational chemistry studies can contribute significantly in achieving an understanding of structure - property relationships by the synthesis of current understanding and data, and by their perspicacity in revealing critical conceptual issues whose resolution demands additional experimentation. In relevance with the importance of the zeolite catalysts in various petroleum, petrochemical and related processes [1 ], as well as due to its growing importance as environmental catalysts [2], as catalysts for the synthesis of fine chemicals [3], as advanced materials in separation technology [4] and as electronic device materials [5], the number of reports of computational studies of zeolites are in proportion. From our earlier experience of applying the molecular dynamics (MD) and computer graphics (CG) techniques to derive valuable insight on the structural aspects of various materials such as supported metal catalysts [6], superconducting materials [7], epitaxial heteroconjunctions [8] etc. we extended the application of these techniques to study zeolite catalysts [9]. The structure and dynamics of the framework atoms, extra framework atoms and guest molecules have been widely studied by various research groups [10-13]. Recently, there have been considerable improvements in the methodology and the cluster models adopted and sophisticated problems such as the adsorption configuration of various molecules and the mechanisms of various reactions have been tackled by QC techniques [14-15]. In this present communication we will give an overview of the computational chemistry methodology used and developed by our recent investigation, followed by the application of the computational chemistry study on different aspects of zeolites such as structure - property correlation, adsorption phenomenon and ion-exchange behavior. This will be followed by the examples of

170

A. Miyamotoet al.

novel application of computational chemistry study in the area of zeozymes, de-NOx processes and in the exciting area of permeation.

COMPUTATIONAL METHODOLOGY

MD Methodology The MD calculations were carried out with the MXDORTO program developed by Kawamura [16]. The Verlet algorithm was used for the calculation of atomic motions, while the Ewald method was applied for the calculation of electrostatic interactions. Temperature and pressure were controlled by means of scaling of atomic velocities and unit cell parameters under 3-dimensional periodic conditions. The two body central force interaction potential was used for all the calculations. Calculations were performed for 200,000 to 500,000 steps, each step corresponds to 1.5 x 10-Is sec. We modified the program time to time in order to keep the temperature values of the solid and gas phases at desired levels during the simulation.

MC Methodology In conventional Grand Canonical MC (GCMC) simulation, the simulations were performed at a certain temperature under the three dimensional periodic boundary condition, e.g. by using Sorption module of B1OSYM/MSI [17]. This kind of simulation is performed in cycles, and each cycle consists of randomly selected attempts either to displace, to create or to destroy a molecule. We developed a new GCMC simulation code where along with the conventional procedure it is possible to displace extra framework cations for the metal-exchanged zeolite [18], which is a very useful tool for analyzing metaltosilicates. Our GCMC has the following features : (i) either GCMC or NVT simulations can be performed with a movable zeolite framework, (ii) calculation and visualization of the adsorbate-zeolite potential energy map in 3-D space is possible and finally (iii) the interacting system can be visualized during the simulation process as an interface.

Semiempirical (AMPAC) and Accurate (DFT) Quantum Chemical (QC) Calculations The semiempirical quantum chemical method using AM1 hamiltonian is adopted to perform the cluster calculations for large tetrameric cluster (four SiO4 terahedra linked by bridging oxygen) models [ 19]. The calculations were carried out using the AMPAC code (QCPE Program No. 539) [20]. Density Functional Theory (DFT) was applied for studying structure property relationship for zeolites along with adsorption phenomenon. The calculations were carried out using the self consistent

Zeolite Microporous Materials

171

Kohn Sham formalism [21] as implemented in the DMOL program of Biosym/MSI based on the work of von Barth and Hedin [22]. The geometry optimization calculations were carried out using Double Numerical with Polarization basis set (DNP) [23]. Local density approximation was performed using JMW (JanakMoruzzi-Witliams) functional as well as using VWN (Vasko-Wiik-Nusair) functional. The non-local gradient corrections were used for getting accurate energy values. We have used non local gradient corrections for exchange Becke [24] and correlation Lee-Young-Parr [25] to calculate self consistently the final energy.

Static Visualization The static visualization by computer graphics (CG) of molecular configurations as snapshots during MD simulations as well as the plots of optimized molecular geometries from DFT calculations were made with the INSIGHTII code of BIOSYM/MSI technologies Inc. on a Silicon Graphics IRIS-4D/25TG engineering workstation. The dynamic visualization for final checking the MD derived equilibrium was made with MOMOVIE and RYUGA [26] programs developed in our laboratory on OMRON-LUNA-88K and HP9000 Model 715/33 workstations, respectively.

Molecular Electrostatic Potential (MEP) Map from the Quantum Chemical Calculations We have devised a methodology [27], where the electrostatic potential can be quantified to understand the spatial volume and the intensity of interaction. The MEP at a certain "r" in space represents a first order approximation to the interaction energy of the molecular charge distribution with the probe of unit charge at that point. The molecular electrostatic potential between the molecular electronic and nuclear charge distribution as well as an external proton at all the points formed by a grid 25 x 25 x 25 points are calculated by a method proposed by Tomasi [28]. The isopotential points are connected to form a contour.

RESULTS AND DISCUSSION A computational chemistry study which has been performed on zeoIites in our laboratory can be divided into two major areas namely: (t) structure and property correlation in zeolites including various phenomenon like adsorption, ion-exchange etc. and (2) novel application areas of computer simulation in the novel fields in zeolite catalysis. We have used MD, MC, Quantum chemical calculations on zeolites to rationalize a better understanding in this area of research.

172

A. Miyamotoet al.

Structure and Property Correlation in Zeolites Zeolite structures. The selective substitution of a silicon by titanium in MFI lattice was investigated by MD methods and CG [29]. It was demonstrated that using a two body interatomic potential of MD is effective to reproduce the framework structure. All eight silicons in the T12 site were replaced by Ti leading to an Si/Ti ratio of 11. The anisotropic lattice expansion due to Ti substitution in various crystallographic sites of MFI lattice was studied. The favorable site for Ti substitution was discussed by comparing the expansions of lattice parameters reported by X-ray diffraction (XRD) studies. It is observed that the T8 site is the most probable site for isomorphous substitution of Ti. A further study on structure and dynamic behavior o f copper, gallium and cerium cations in cation-exchanged ZSM-5 framework was performed using MD and CG [30]. Various oxidation states of these metals were simulated and their relevance in the catalytic activity is also discussed. Depending on the starting position of Cu +, either the migration readily occurred even at 300 K when it is closer to TI 2 site, or a higher temperature, namely 600 K was needed when it is further from T12 site. At lower temperatures, Cu 2+ ion in Cu-ZSM-5 is considered to be a hydrated state as [Cu(OH)] +. Under the reaction conditions, the copper is expected to be in a dehydrated state and hence we have simulated both Cu z~ and [CuO] inside ZSM-5 channels. It was observed that both these species also prefer a location closer to the [A104]-tctrahedron. The variation of the coulomb energy of the lattice with the distance between the positive and negative charges indicates that the coulomb energy is less favorable, when two aluminium atoms are present at a distance 11,~ (at two T12 sites) with a Cu 2+ ion. Thus, Cu 2+ is not stable and it prefers to be Cu + even under reaction conditions. Various Gallium species such as Ga 3~, [GaO] § [Ga(OH)] 2+ and [Ga(OH)2] + were simulated inside ZSM-5. Gallium also migrates similarly towards T12 site where A1 is substituted. Cerium also follows the same trend as that of gallium. The reversible formation of oxidized and reduced states of cations particularly that of copper are the reason for catalytic activity. Whereas the increased dynamic freedom of the oxygen attached to gallium and cerium and the low coordination of these cations in ZSM-5 may be the cause of their improved catalytic activity. MD and QC (DFT) calculations have been performed in a recent study [31] to understand the structure and electronic properties ofmetallosilicates. For the first time, the local structure of the metal substituted zeolite-framework is derived from MD calculations and its electronic structure is determined by DFT calculations for different metal substituted clusters. Our calculations showed that the net electron density on the MO 4 unit, where M is Ti 4§ A13~, Ga 3+ or Fe 3§ is a reliable descriptor for the acidity of metaliosilicates. Furthermore the analysis of MEP maps for the cluster models predict the polarization resultant from various substitutions which

Zeolite Microporous Materials

173

further justifies the acidity trend as observed by experiment.

Ion-exchangeproperties. The applicability of MD and CG to investigate the structure and function of ion-exchanged zeolites was demonstrated [32-33] for (i) reproducing the known structures of various zeolites, (ii) determining the positions of A1 and exchanged cations in the framework and (iii) understanding the dynamic mechanism in the molecular sieving process of O2 and N 2 in A-type zeolites. It is observed that Na ion cannot reach the vicinity of Al cation at 300k, while it can migrate easily at higher temperature. H and Li having smaller ionic radii are located near the vicinity of oxygen anions whereas Na, K or Cs due to their larger ionic radii spread out in the micropores of ZSM-5. The dynamic visualization of the diffusion process of 02 and N 2 at different temperatures suggested that the Na ion at the window plays the crucial role for determining the effective pore radius of NaA ion. The effectiveness and applicability of MD, NMR simulation and CG to the investigation of the sites and distribution of framework aluminums in faujasites, is shown [34]. By using MD calculations, we successfully predicted the adequate aluminum distribution in the NaY zeolite (Si:AI -- 2.43), as it reproduces the sites and occupancies of Na" cations which have been further supported by neutron diffraction techniques. The simulated neutron diffraction patterns show the differences in the distribution of cations among the I" sites in the different patterns at low angle region. The validity of our aluminum distribution model in Na-Y has been tested with simulated 29Si MAS NMR spectra, which shows agreement with both in terms of chemical shifts and intensities.

Adsorption and diffusion behavior. MD and CG were applied to study the dynamic behavior of adsorbed molecules with different diameters and weights in ZSM-5 and mordenite [35]. The medium size monatomic molecules diffused 3dimensionally in the micropores of ZSM-5. The mobility of the adsorbed molecule decreases with increase in its size. The higher diffusivity is observed in the case of ZSM-5 in comparison to mordenite. On further study it is revealed that the flexibility of the framework is present in the case of ZSM-5 or mordenite to allow the diffusion. The MD simulation results also show that the collision between adsorbed molecules greatly decreased the diffusion rate for mordenite, while the dynamics and rapid diffusion were not much retarded by collision for ZSM-5. This is consistent with the experimental results. In another study with a combination of GCMC, DFT and CG the adsorption mechanism of CFCs (chlorofluorocarbons) in zeolite CsNaY are successfully predicted [36]. CFCs are widely used as solvents, refrigerants and propellants. However, the emission of CFCs led to stratospheric ozone depletion and greenhouse

A. Miyamotoet al.

174

effects. The accurate distribution of AP +, Cs § and Na § in the CsNaY (Si/A1 = 2.43) was obtained from MD calculations and NMR simulation, while the potential parameters for CF2C12 were determined by DFT calculations. The experimental adsorption isotherm of the CF2C12 in CsNaY was fully reproduced by GCMC simulation as shown in Figure 1. Furthermore, CG visualization of the calculated results revealed the presence of two different mechanisms of CFEC12 adsorption in CsNaY: (i) at low pressure, the CF2C12 selectively and drastically adsorbed on the Cs cation of CsNaY and (ii) at high pressure, the aggregation of CF2C12 in the supercages of CsNaY occurred. In an earlier review, computational methodology was used to design synthetic sorbents for selective adsorption of molecules [37]. The attempts made to design synthetic sorbents at a molecular level are thoroughly discussed.

Novel Application Areas of Computer Simulation in the Novel Fields in Zeolite Catalysis Zeozymes. Zeozymes are a type of materials which can act as enzymes after encapsulating microporous crystals inside zeolites and can act as selective catalysts. 4O r r/J

r

30 .......,1

0

""

20

O

o

10 ...... 9 ........

0 "t:)

<

0

Calculated data

I

I

I

I

0.01

0.02

0.03

0.04

0.05

Pressure (kPa) F i g u r e 1. Calculated adsorption isotherm ofCF2CI 2 in CsNaY (Si/A1 = 2.43) with the experimental result.

175

Zeolite Microporous Materials

Our aim is to design these catalysts by carrying out electronic structure calculations and to study the encapsulation of enzyme-like active sites into porous mantles. Porphyrins have a significant role in biological actions such as photosynthesis, as oxygen carriers in the decomposition of water. Although porphyrin based catalysts undergo oxidative dimerization as well as degradation, their encapsulation inside microporous materials extends their catalytic life, activity and selectivity. In our recent study [38], we studied the configuration of metal (Mg, Co, Fe) porphyrin complexes inside Y-type zeolites using MD simulations. Figure 2 shows the orientation of the Mg-porphyrin complex in absence of Na cations; the Mgporphyrin is located above the six membered ring composed of zeolite oxygens with the metal center oriented towards the oxygen atoms. Whereas in the presence of Na the metal center (Mg) of the porphyrin is oriented towards the center of the supercage as shown in Figure 3, which is attributed to the electrostatic repulsion with sodium cations.

N

~H Si

Figure 2. Mg-porphyrin complex encapsulated in USY zeolite at 300K. The metal center of the porphyrin interacts with zeolite oxygen anions.

In a recent study we have used QC and CG methods to study the docking of organic molecules pyrrole and acetaldehyde inside faujasite zeolites [39]. The low-energy binding sites for the organics inside the supercage of the faujasite framework were derived from the energy minimization procedure. The exact orientation of these molecules are derived from QC DFT cluster model calculations.

176

A. Miyamoto et al.

Si

,C

Figure 3. Mg-porphyrin complex encapsulated in NaY zeolite at 300 K. The Na cation is located above zeolite oxygens and the Mg atom is oriented toward the supercage center.

The electronic structure of pyrrole and acetaldehyde as well as their adsorption complexes over a suitable cluster model of the faujasite framework are reported. The configuration of tetramethylporphyrin formed by the condensation of pyrrole and acetaldehyde is observed to be inclined inside the supercage. It is difficult experimentally to obtain information about the configuration of the porphyrin inside the zeolite, the effect of exchanged cations, or the size limitations due to the presence of organic substituents on the porphyrin. We therefore performed MD simulation to study the configuration of metal (Mg, Co, Fe) porphyrin complexes encapsulated in Y-type zeolites [40]. It is observed that as the zeolite supercage can hold only one porphyrin the oxidative dimerization which occurs in homogeneous systems can be avoided by the encapsulation. Further, when the zeolite does not contain exchanged cations, the metal center of the porphyrin goes towards the center of the six member ring formed by framework oxygens. In this orientation, the metal interacts with the zeolites; therefore, a low activity of the porphyrin-zeolite catalyst is expected. Now in the situation with exchanged metal ion it is oriented towards the supercage center, so the active site is available to react and a better porphyrin-zeolite catalyst is expected.

De-NOx process.

The utilization of zeolite catalysts as prospective

Zeolite Microporous Materials

177

environmental catalysts has been explored for the past few years. Metal exchanged zeolites have emerged as efficient and economical catalysts for the removal of nitric oxide from combustion exhaust and automobile exhaust gases. Despite the extremely challanging character of the topic, the mechanism of the process is basically unknown; for example, little is known about the mechanism of water poisoning or the specific role of the reductant and the catalyst in the reaction; also the identification of intermediates has not yet been accomplished. We have performed a combination of MD and QC DFT calculations [41] to form an efficient route towards understanding a complex reaction sequence in catalytic reduction of NOx. It is shown on the basis of our result that the large nucleophilic space (Figure 4) around GaO is capable of activating even non-polar hydrocarbons. This nucleophilicity arises from the coordination unsaturation around Ga3+. The proposed mechanism of the interaction of methane with the Ga site in Ga-ZSM-5 agrees well with recent experimental results on efficient catalytic performance of the catalyst.

Figure 4. The MESP map around [H4AIO4][GaO] cluster. Dark contour corresponds to nucleophilic region (-0.05 to -1.00 a.u.) and light contour corresponds to electrophilic region (+0.05 to +1.00 a.u.).

A MD, QC and CG combination methodology has been further applied to rationalize the multistep removal of NOx by methane activated Ga-ZSM-5 [42]. It is obseved that the methane molecule may undergo weak adsorption with polarization of an intramolecular C-H bond prior to the dissociation of the bond. Two types of methane dissociation are proposed, namely, one with the methyl group attached to the exchange Ga ion and the dissociated H atom connected to extraframework oxygen and the other one with the methyl group bonded to the extraframework oxygen and the hydrogen atom attached to the exchanged Ga ion. The former one is the preferred geometry in view of energy stabilization, while both forms highly stabilize the system in comparison to physisorbed methane molecule.

178

A. Miyamoto et al.

It is also observed that the water molecule may be easily dissociated without physisorption on the Ga site, and the dissociated form of water strongly stabilizes the system energetically. This greatly decreases the adsorption energy of the methane molecule on the Ga site and, as a consequence, prevents the activation process of methane. We performed few exhaustive studies in rationalizing the understanding of the adsorption and activation mechanisms of methane and NOx molecules on GaZSM-5, reaction profile and transition state calculations using a combination of MD. QC and CG methodologies [43-45].

Zeolite membranes. Much attention has been given recently to the design and use of inorganic membranes in separation processes. The use of zeolites as inorganic membranes will allow the combination of their catalytic activity and separation ability, which make them act as bifunctional systems. For the first time, we used computational chemistry methodology based on interatomic interactions, such as MD and MC techniques for understanding the dynamic behavior and transport mechanism of butane through the zeolite membrane [46]. The permeation mechanism of n-butane is primarily dominated by surface diffusion or capillary condensation at 373K as shown in Figure 5, since the aggregation of molecules inside the silicalite pore was observed. At higher temperature i.e. 773 K the adsorption becomes smaller and the mechanism can be described as an activated diffusion. In the gas mixture, the permeation of n-butane through the membrane was observed at 373 K, where as iso-butane did not permeate. However diffusion of both isomers individually into the zeolite pores was observed. Our MD results show an enhanced adsorption of iso-butane and a lower permeance of n-butane compared with the behavior of single gases. Since the diffusion of iso-butane is significantly lower than that of n-butane, the few diffused iso-butane molecules reduce the diffusion and permeability of n-butane.

CONCLUSION The content of this review could be summarized to conclude that computational chemistry studies are having a substantial impact on our understanding of structural properties, adsorption, ion-exchange, encapsulation, de-NOx removal and membranes at present and will have enormous predicting power in the near future. The tactics needed to attack the problems lie in a judicious use of different combinations of computational techniques. In this review, it has only been possible to highlight the outline of our studies in the field of zeolite catalysis. All these methodologies are essentially applicable to any other unsolved challenging areas of

179

Zeolite Microporous Materials

~

0 ps

9

~ ps

240 ps

Figure 5. Permeation of n-butane gas through a silicalite membrane at 373 K.

interest in zeolites and other correlated fields and essentially reminds us that a great deal remains to be done. Future explorations should deal with the elimination of the approximations that are often being made, such as QC AMPAC methods. Particularly with the advent of limitless computing power, more accurate QC and MD calculations will become feasible in the foreseeable future. As our future task, we are engaging ourselves in deriving new softwares for applying in this challenging area. So the computational chemistry study has now established its role in exploring the limitaions of experiment to propose plausible explanation for the critical situations.

REFERENCES 1. 2.

N.Y. Chen and T.F. Degnan, Chem. Eng. Prog. Feb., 32 (1988). H. Hamada, Y. Kintaichi, T. Yoshinari, M. Tabata, M. Sasaki, and T.lta, CataL Today 17,

180

3, 4. 5. 6.

9, 10. 11. 12. 13.

14. 15. 16. 17. 18.

19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34.

A. Miyamoto et al. 111 (1993). W. Holderich, M. Hesse, and F. Naumann, Angew. Chem. Int. Ed. Engl. 27, 226 (1988). R.V. Jasra, N.V. Choudary, and S.G.T. Bhat, Sep. Sci. Technol. 26, 885 (1991). G.A. Stucky, Science 247, 669 (1990). A. Miyamoto, T. Hattori, and T. Inui, Appl. Surf Sci. 60161 660 (1992); A. Miyamoto, R. Yamauchi, and M.Kubo, Appl. Surf Sci. 75, 51 81994). A. Miyamoto, Y. Hattori, and T.Inui, Physica C 190, 93 (1991). A. Miyamoto, K. Takeichi, T. Hattori, M.Kubo, and T. Inui, Jpn. J. Appl. Phys. 31, 4463 (1992). A. Miyamoto, K. Matsuba, M.Kubo, K. Kawamura, and T. Inui, Chemistry Lett. 2055 (1991). P. Demontis, G.B. Suffritti, S. Quarteri, E.S. Fois, and A. Gamba, Zeolites 7, 522 (1987). P. Santikary, S. Yashonath, and G. Anathakrishna, a~ Phys. Chem. 96, 10469 (1992). J.O. Titiloye, S.C. Parker, F.S. Stone, and C.R.A. Catlow, ~Phys. Chem. 95, 4038 (1991). J.B. Nicholas, F.R. Trouw, J.E. Mertz, L.E. Iton, and A.J. Hopfinger, J. Phys. Chem. 97, 4149(1993). J. Sauer, Chem. Rev. 89, 199 (1989). S. Beran. In: Theoretical Aspects of Heterogeneous Catalysis, J.B. Moffat (Ed.), van Nostrand Reinhold, New York, (1990), p. 160,. K. Kawamura. In: Introduction to Molecular Simulations, I.Okada and E. Osawa (Eds.), Kaibun-do, Tokyo, 1989, Chapters 6 and 7. Cerius2 Computational and Analytical Instrumentation User's Guide, Release 1.0, Molecular Simulation Inc., Massachusette, USA, 1994. A. Hirotani, K. Mizukami, R. Miura, H. Takaba, T. Miya, A. Fahmi, A. Stifling, M. Kubo, and A. Miyamoto, Appl. Surf Sci., in press. S. Prasad and R. Vetrivel, J. Mol. Catal. 84, 299 (1993). M.J.S. Dewar, E.G. Zoebisch, E.F. Healy, and J.J.P. Stewart, J. Am. Chem. Soc. 107, 3902 (1985). W. Kohn and L.J. Sham, Phys. Rev. A 140, 1133 (1965). U. von Barth and L. Hedin, J. Phys. C 5, 1629 (1972). B. Delley, J. Chem. Phys. 92, 508 (1990). A. Becke, J. Chem. Phys. 88, 2547 (1988). C. Lee, W. Yang, and R.G. Parr, Phys. Rev. B 37, 786 (1988). R. Miura, H. Yamano, R. Yamauchi, M. Katagiri, M. Kubo, R. Vetrivel, and A. Miyamoto, Catal. Today 23, 409 (1995). R. Vetrivel. R.C. Deka, A. Chatterjee, M. Kubo, E. Broclawik, and A. Miyamoto, Theor. Comp. Chem. 3, 509 (1996). J. Tomasi. In: Chemical Application of Atomic and Molecular Electrostaic Potentials, P. Politzer and D.G. Truhler (Eds.), Plenum, New York, 151, 1981. Y. Oumi, K. Matsuba, M. Kubo, T. Inui, and A. Miyamoto, Microporous Mater. 4, 53 (1995). A. Miyamoto, H. Himei, Y. Oka, E. Maruya, M. Katagiri, R. Vetrivel, and M. Kubo, Catal. Today 22, 87 (1994). Y. Oumi, M. Yamadaya, T. Kanougi, M. Kubo, A. Stirling, R. Vetrivel, E. Broclawik, and A. Miyamoto, Catal. Lett. 45, 21 (1997). A. Miyamoto and M. Kubo, Stud. Surf Sci. Catal. 83, 117 (1994). A. Miyamoto, H. ttimei, E. Maruya, M. Katagiri, R. Vetrivel, and M. Kubo, Stud. ,Surf Sci. Catal. 90, 217 (1994). H. Himei, M. Yamadeya, Y. Oumi, M. Kubo, A. Stirling, R. Vetrivel, E. Broclawik, and A. Miyamoto, Microporous Mater. 7, 235 (1996).

Zeolite Microporous Materials 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46.

181

M. Kubo, A. Miyamoto, and T. Inui, Stud Surf Sci. Catal. 80, 333 (1993). K. Mizukami, H. Takaba, Y. Oumi, M. katagiri, M. Kubo, A. Stirling, and A. Miyamoto, Stud Surf Sci. Catal. 105, 1811 (1997). R. Vetrivel, H. Takaba, M. Katagiri, M. Kubo, and A. Miyamoto, Stud Surf Sci. Catal. 99,3 (1996). T. Sato, K. Sugao, Y. Oumi, R. Vetrivel, M. Chatterjee, A. Chatterjee, M. Kubo, A. Stirling, A. Fahmi, and A. Miyamoto, Appl. Surf Sci., in press. A. Chatterjee, R. Vetrivel, M. Kubo, and A. Miyamoto, Stud Surf Sci. Catal. 105, 2339 (1997). T. Sato, K. Sugao, Y. Oumi, R. Vetrivel, M. Chatterjee, A. Chatterjee, M. Kubo, A. Stirling, A. Fahmi, and A. Miyamoto, Appl. Surf Sei., in press. R. Vetrivel, M. Kubo, H. Himei, E. Maruya, M. Katagiri, E. Broclawik, and A. Miyamoto, Stud. Surf Sci. Catal 92,233 81995). H. ltimei, M Yamadaya, M. Kubo, R. Vetrivel, E. Broclawik, and A. Miyamoto, dl Phys. Chem. 99, 12461 (1995). E. Broclawik, tl. Himei, M. Yamadaya, M. Kubo, A. Miyamoto, and R. Vetrivel, J. Chem. Phys. 103, 2102 (1995). M. Yamadaya, H. Himei, T. Kanougi, Y. Oumi, M. Kubo, A. Stirling, R. Vetrivel, E. Broclawik, and A. Miyamoto, Stud. Surf ScL Catal. 105, 1485 (1997). T. Kanougi, K. Furukawa, M. Yamadaya, Y. Oumi, M. Kubo, A. Stirling, A. Fahmi, and A. Miyamoto, Appl. Surf Sci., in press. H. Takaba, R. Koshita, K. Mizukami, Y. Oumi, N. Ito, M. Kubo, A. Fahmi, and A. Miyamoto, J. Membrane Sci., in press.