Zeolites: A Refined Tool for Designing Catalytic Sites

Studies in Surface Science and Catalysis 97 ZEOLITES: A REFINED TOOL FOR DESIGNING CATALYTIC SITES This page intentio...

Author: Laurent Bonneviot | Serge Kaliaguine

21 downloads 751 Views 20MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

Studies in Surface Science and Catalysis 97 ZEOLITES: A REFINED TOOL FOR DESIGNING CATALYTIC SITES

This page intentionally left blank

Studies in Surface Science and Catalysis Advisory Editors: B. Delmon and J.T. Yates Vol. 97

ZEOLITES: A REFINED TOOL FOR DESIGNING CATALYTIC SITES Proceedings of t h e International Zeolite S y m p o s i u m , Qu(~bec, O c t o b e r 15-20,1995

Canada,

Edited by

Laurent Bonneviot Departement de Chimie, CERPIC, Facu/te des Sciences et de Genie, Universite Lava/, Quebec, Canada G 1K 7P4

Serge Kaliaguine Departement de Genie Chimique, CERPIC, Faculte des Sciences et de Genie, Universite Lava/, Quebec, Canada G 1K 7P4

1995 ELSEVIER A m s t e r d a m - - L a u s a n n e - - N e w Y o r k - - O x f o r d --- S h a n n o n --- T o k y o

ELSEVIER SCIENCE B.V. Sara Burgerhartstraat 25 P.O. Box 521, 1000 AM Amsterdam, The Netherlands

ISBN 0-444-82130-9 91995 Elsevier Science B.V. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior written permission of the publisher, Elsevier Science B.V., Copyright & Permissions Department, P.O. Box 521, 1000 AM Amsterdam, The Netherlands. Special regulations for readers in the U.S.A.- This publication has been registered with the Copyright Clearance Center Inc. (CCC), 222 Rosewood Drive, Danvers, MA 01923. Information can be obtained from the CCC about conditions under which photocopies of parts of this publication may be made in the U.S.A. All other copyright questions, including photocopying outside of the U.S.A., should be referred to the copyright owner, Elsevier Science B.V., unless otherwise specified. No responsibility is assumed by the publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein. This book is printed on acid-free paper. Printed in The Netherlands

CONTENTS Foreword . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . PL-1

M-l-1

One and Two-Dimensional Solid-State NMR Investigations of the ThreeDimensional Structures of Zeolite-Organic Sorbate Complexes C.A. Fyfe, H. Grondey, A.C. Diaz, G.T. Kokotailo, Y. Feng, Y. Huang, K.C. Wong-Moon, K.T. Mueller, H. Strobl and A.R. Lewis . . . . . . . . . . . . .

XI XII

1

The Use of Small, Weakly Basic Probe Molecules for the Investigation of BrCnsted Acid Sites in Zeolites by NMR Spectroscopy E. Brunner . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

11

Determination of the Environment of Titanium Atoms in TS-1 Silicalite by Ti K-edge X-ray Absorption Spectroscopy, 29Si and 1H Nuclear Magnetic Resonance- L. Le Noc, C. Cartier dit Moulin, S. Solomykina, D. Trong On, C. Lortie, S. Lessard and L. Bonneviot . . . . . . . . . . . . . . . .

19

An in situ 13C MAS NMR Study of Toluene Alkylation with Methanol over H-ZSM-11 - I.I. Ivanova and A. Corma . . . . . . . . . . . . . . . . . . . . . .

27

KL-1

Strategies for Zeolite Synthesis by Design- M.E. Davis . . . . . . . . . . . . . .

35

M-2-1

Use of Modified Zeolites as Reagents Influencing Nucleation in Zeolite Synthesis - S.I. Zones and Y. Nakagawa . . . . . . . . . . . . . . . . . . . . . . . . .

45

Templating Studies Using 3,7-Diazabicyclo[3.3.1]nonane Derivatives: Discovery of New Large-Pore Zeolite SSZ-35 - Y. Nakagawa . . . . . . . . . .

53

Synthesis of ZSM-48 Type Zeolite in Presence of Li, Na, K, Rb and Cs Cations - G. Giordano, A. Katovic, A. Fonseca and J.B. Nagy

61

Out-of-Plane Bending Vibrations of Bridging OH Groups in Zeolites: A New Characteristic of the Geometry and Acidity of a Broensted Site - L.M. Kustov, E. Loeffler, V.L. Zholobenko and V.B. Kazansky

63

In situ FTIR Microscopic Investigations on Acid Sites in Cloverite G. MUller, G. Eder-Mirth and J.A. Lercher . . . . . . . . . . . . . . . . . . . . . . .

71

Broensted Sites of Enhanced Acidity in Zeolites: Experimental Modelling - M.A. Makarova, S.P. Bates and J. Dwyer . . . . . . . . . . . . . . . .

79

Modelling of Structure and Reactivity in Zeolites - C.R.A. Catlow, R.G. Bell, J.D. Gale and D.W. Lewis . . . . . . . . . . . . . . . . . . . . . . . . . . .

87

M-l-3

M-l-4

M-2-2

M-2-3

M-2-4

M-2-5

M-2-6

PL-2

VI

Tu-l-1

Tu-l-2

Tu-l-3

Tu-l-4

KL-2

Tu-2-1

Tu-2-2

Tu-2-3

Tu-2-4

Tu-2-5

Computational Studies of Water Adsorption in Zeolites - S.A. Zygmunt, L.A. Curtiss and L.E. Iton . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

101

Modeling of Adsorption Properties of Zeolites - A. Goursot, I. Papai, V. Vasilyev and F. Fajula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

109

Loading and Location of Water Molecules in the Zeolite Clinoptilolite Y.M. Channon, C.R.A. Catlow, R.A. Jackson and S.L. Owens . . . . . . . . . .

117

Withdrawal of Electron Density by Cations from Framework Aluminum in Y-Zeolite Determined by A1 XAFS Spectroscopy D.C. Koningsberger and J.T. Miller . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

125

Geometry of the Active Sites in Zeolites under Working Conditions F. Fajula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

133

Characterization of Hexagonal and Lamellar Mesoporous Silicas, Alumino- and Gallosilicates by Small-Angle X-Ray Scattering (SAXS) and Multinuclear Solid State NMR - Z. Gabelica, J.-M. Clacens, R. Sobry and G. Van den Bossche . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

143

Characterization of a Cubic Mesoporous MCM-48 Compared to a Hexagonal MCM-41 - R. Schmidt, M. St6cker and O.H. Ellestad . . . . . . . .

149

Synthesis and Characterization of Transition Metal Containing Mesoporous Silicas - S. Gontier and A. Tuel . . . . . . . . . . . . . . . . . . . . . .

157

Bimodal Porous Materials with Superior Adsorption Properties C.J. Guo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

165

MCM-41 Type Silicas as Supports for Immobilized Catalysts - D. Brunel, A. Cauvel, F. Fajula and F. DiRenzo . . . . . . . . . . . . . . . . . . . . . . . . . . . .

173

Tu-2-6

Synthesis of Mesop0rous Manganosilicates Mn-MCM-41, Mn-MCM-48 and Mn-MCM-L at a Low Surfactant/Si Ratio - D. Zhao and D. G o l d f a r b . . 181

W-l-1

Alkane Oxidation Catalyzed by Zeolite Encapsulated Ruthenium Perfluorophthalocyanines - K.J. Balkus Jr., A. Khanmamedova and M. Eissa . . . . . .

189

W-l-2

MOCVD in Zeolites Using Mo(CO) 6 and W(CO)6 as Precursors - S. Djajanti and R.F. Howe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197

W-l-3

Tailored Synthesis, Characterization and Properties of ZnO, CdO and SnO 2 Nano Particles in Zeolitic Hosts - M. Wark, H.-J. Schwenn, M. Warnken, N.I. Jaeger and B. Boddenberg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

205

VII

Host-Guest Interactions in Zeolite Cavities - A. Zecchina, R. Buzzoni, S. Bordiga, F. Geobaldo, D. Scarano, G. Ricchiardi and G. Spoto . . . . . . . .

213

PL-4

Diffusion in Zeolites - D.M. Ruthven

223

Th-l-1

Frequency Response Study of Mixture Diffusion of Benzene and Xylene Isomers in Silicalite-1 - D. Shen and L.V. Rees . . . . . . . . . . . . . . . . . . . .

235

2D EXSY 129Xe NMR: New Possibilities for the Study of Structure and Diffusion in Microporous Solids - I.L. Moudrakovski, C.I. Ratcliffe and J. Ripmeester . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

243

Diffusion of C10-C24n-paraffins and perfluorotributylamine in clay catalysts- B. Liao, M. Eic, D.M. Ruthven and M.L. Occelli . . . . . . . . . . .

251

Sorption Properties of Dealuminated Large Crystals of ZSM-5: A New Approach to the Description of Isotherms- J. Kornatowski, M. Rozwadowski, W. Lutz and W.H. Baur . . . . . . . . . . . . . . . . . . . . . . .

259

Convective Methods to Investigate Multi-component Sorption Kinetics on Microporous Solids- A. Micke and M. Btilow . . . . . . . . . . . . . . . . . . .

269

KL-4

Zeolites as the Key Matrix for Superior deNOx Catalysts - T. Inui . . . . . . .

277

Th-2-1

Catalytic Properties of Palladium Exchanged Zeolites in the Reduction of Nitrogen Oxide by Methane in the Presence of Oxygen: Influence of Hydrothermal Ageing - C. Descorme, P. G61in, M. Primet, C. L6cuyer and J. Saint-Just . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

287

The Effect of Preparation and Steaming on the Catalytic Properties of Cu- and Co-ZSM-5 in Lean NO x Reduction - P. Ciambelli, P. Corbo, M. Gambino, F. Migliardini, G. Minelli, G. Moretti and P. Porta . . . . . . . .

295

Adsorption Sites for Benzene in the 12R Window Zeolites: A Molecular Recognition Effect- B.L. Su . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

303

Lewis Basic and Lewis Acidic Sites in Zeolites - M. Huang, S. Kaliaguine and A. Auroux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

311

Effect of the Framework Composition on the Nature and the Basicity of Intrazeolitic Cesium Oxides. Correlation Activity/Basicity - M. Lasp6ras, I. Rodriguez, D. Brunel, H. Cambon and P. Geneste . . . . . . . . . . . . . . . . .

319

KL-3

Th-l-2

Th-l-3

Th-l-4

Th-l-5

Th-2-2

Th-2-3

Th-2-4

Th-2-5

...........................

VIII Th-2-6

F-l-1

F-l-2

F-l-3

F-l-4

KL-5

F-2-1

F-2-2

F-2-3

F-2-4

P-2

P-6

Hydrothermal and Alkaline Stability of High-Silica Y Zeolites Generated by Combining Substitution and Steaming - W. Lutz, E. LOftier and B. Zibrowius . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

327

Electron Spin Resonance and Electron Spin Echo Modulation Spectroscopy of Ni(I) in SAPO-5 and SAPO-11 - M. Hartmann, N. Azuma and L. Kevan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

335

Selective Acidic, Oxidative and Reductive Reactions over ALPO-11 and Si or Metal Substituted ALPO-11 - P.S. Singh, R. Bandyopadhyay, R.A. Shaikh and B.S. Rao . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

343

Characterization and Catalytic Performance of PdSAPO-5 Molecular Sieves - T.-C. Xiao, L.-D. An and H.-L. Wang . . . . . . . . . . . . . . . . . . . . .

351

FTIR Study of the Acidic Properties of Substituted Aluminophosphates V. Zholobenko, A. Garforth, L. Clark and J. Dwyer . . . . . . . . . . . . . . . . .

359

Selective Oxidation with Redox Metallosilicates in the Production of Fine Chemicals- P. Ratnasamy and R. Kumar . . . . . . . . . . . . . . . . . . . . .

367

Novel Model Catalysts Containing Supported MFI-type Zeolites N. van der Puil, E.C. Rodenburg, H. van Bekkum and J.C. Jansen . . . . . . .

377

Oligomerization of Butenes with Partially Alkaline-Earth Exchanged NiNaY Zeolites - B. Nkosi, F.T.T. Ng and G.L. Rempel . . . . . . . . . . . . . .

385

Isomerization of C 8 Aromatic Cut. Improvement of the Selectivity of MOR- and MFI- Catalysts by Treatments with Aqueous Solutions of (NH4)2SiF6 - E. Benazzi, J.M. Silva, M.F. Ribeiro, F.R. Ribeiro and M. Guisnet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

393

Contribution of Framework and Extra-framework A1 and Fe Cations in ZSM-5 to Disproportionation and C 3 Alkylation of Toluene - J. (2ejka, N. Zilkov~i, Z. Tvart)~kov~i and B. Wichterlov~i . . . . . . . . . . . . . . . . . . . . .

401

NO x Adsorption Complexes on Zeolites Containing Metal Cations and Strong Lewis Acid Sites and their Reactivity in CO and CH 4 Oxidation: a Spectroscopic Study - L.M. Kustov, E.V. Smekalina, E.B. Uvarova and V.B. Kazansky . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

409

Cracking of 1,3,5-triisopropylbenzene over Deeply Dealuminated Y Zeolites- E.F. Sousa-Aguiar, M.L. Murta VaUe, E.V. Sobrinho and D. Cardoso . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

417

IX

P-7

P-11

P-14

P-15

P-16

P-17

P-18

P-19

P-20

P-21

P-25

P-26

P-27

Hydrogenation of Styrene and Hydrogenolysis of 2-Phenylethanol T. Sooknoi and J. Dwyer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

423

Catalyst Deactivation of High Silica Metallosilicates in Beckmann Rearrangement of Cyclohexanone Oxime - T. Takahashi, T. Kai and M.N.A. Nasution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

431

IR Studies on the Reduction of Nitric Oxide with Ammonia over MFIFerrisilicate - T. Komatsu, Md. A. Uddin and T. Yashima . . . . . . . . . . . . .

437

The Role of the Na and of the Ti on the Synthesis of ETS-4 Molecular Sieve - P. De Luca, S. Kuznicki and A. Nastro . . . . . . . . . . . . . . . . . . . .

443

On the Potential of Zeolites to Catalyse the Aromatic Acylation with Carboxylic Acids - E.A. Gunnewegh, R. Downing and H. van Bekkum

447

Elimination of Methanol from Dimethylacetal over Aluminophosphate Molecular Sieves and Zeolites - S.-M. Yang and K.-J. Wang . . . . . . . . . . .

453

Deuteration of Zeolitic Hydroxyl Groups in the Presence of Platinum Evidence for a Spillover Reaction Pathway - U. Roland, R. Salzer and L. SUmmchen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

459

IR Investigation of CO Adsorption at Low Temperature: A Key Tool to Characterize the Porosity of Matrix Embedded Zeolite Catalysts Z.M. Noronha, J.L.F. Monteiro and P. G61in . . . . . . . . . . . . . . . . . . . . . .

465

Identification of Active Ti Centers in TS-1 as Revealed by ESR Spectra of UV-Irradiated Samples - A. Ghorbel, A. Tuel, E. Jorda, Y. Ben TaCit and C. Naccache . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

471

Microwave Crystallization of Titanium-Containing Cloverite - R. Fricke, H.-L. Zubowa, J. Richter-Mendau, E. Schreier and U. Steinike . . . . . . . . . .

477

Aqueous Silicate Chemistry in Zeolite Synthesis - C.T.G. Knight, R.T. Syvitski and S.D. Kinrade . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

483

Effect of Hydrolysis Conditions of the Silicate Precursor on the Synthesis of Siliceous MCM-48 - J. Lujano, Y. Romero and J. Carrazza . . . . . . . . . .

489

Sorption of Light Alkanes on H-ZSM5 and H-Mordenite - F. Eder, M. Stockenhuber and J.A. Lercher . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

495

P-28

P-29

P-31

P-34

P-36

P-41

P-42

P-43

P-45

Acidity and Reactivity of Steamed HY Zeolites Obtained by Progressive Extraction of Extraframework A1 Species- L. Mariey, S. Khabtou, M. Marzin, J.C. Lavalley, A. Chambellan and T. Chevreau . . . . . . . . . . . .

501

L.

Adsorption of Nitrogen and Methane on Natural Clinoptilolite Predescu, F.H. Tezel and P. Stelmack . . . . . . . . . . . . . . . . . . . . . . . . .

507

Dealumination and Acidity Measurement of HEMT Zeolites Modified by Steaming and Leaching - O. Cairon, S. Sellem, C. Potvin, J.M. Manoli and T. Chevreau . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

513

The Synthesis of UTD-1, Ti-UTD-1 and Ti-UTD-8 Using CP*2CoOH as a Structure Directing Agent- K.J. Balkus Jr., A.G. Gabrielov and S.I. Zones . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

519

Zeolite Crystallization on Mullite Fibers - V. Valtchev, S. Mintova, B. Schoeman, L. Spasov and L. Konstantinov . . . . . . . . . . . . . . . . . . . . .

527

Synthesis and Characterization of V-Beta Zeolite - S.-H. Chien and J.-C. Ho . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

533

Titanium Boralites with MFI Structure Characterized Using XRD, IR, UV-Vis XANES and MAS-NMR Techniques - D. Trong On, M.P. Kapoor, S. Kaliaguine, L. Bonneviot and Z. Gabelica . . . . . . . . . . . .

535

Acidity and Structural State of Boron in Mesoporous Boron Silicate MCM-41 - D. Trong On, P.N. Joshi, G. Lemay and S. Kaliaguine . . . . . . .

543

The Role of Na and K on the Synthesis of Levyne-Type Zeolites C.V. Tuoto, J.B. Nagy and A. Nastro . . . . . . . . . . . . . . . . . . . . . . . . . . .

551

Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

557

Subject Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

561

Studies in Surface Science and Catalysis

567

................................

xI FOREWORD Once upon a time the catalytic use of zeolites was exclusively in the field of acid catalysis. Today zeolites also find applications as catalysts in a wide array of chemical reactions. These encompass base catalyzed reactions, Redox reactions and catalytic reactions on transition metals and their complexes in confined environments. The concepts of Bronsted or Lewis acid-base pairs are abundantly illustrated in the literature and better understood in terms of structural and electronic properties of zeolites. By contrast properties of chemically modified silicates, aluminosilicates and aluminophosphates are not yet fully explored. The list of oxydo-reduction reactions performed in the presence of these new materials is indeed continuously growing. For example the selective catalytic reduction of nitrogen oxides or the numerous oxidations employing hydrogen peroxide could be cited. In this context much effort is currently made in order to get a better insight into the nature of the sites involved. Thirdly, the zeolite lattice may be used as a host for encapsulated complexes or metallic clusters allowing to control the nuclearity of these active species and the steric constraints imposed on the reactants. The molecular sieve and shape selectivity effects have always constituted the most fascinating aspects of zeolite properties. The recent developments leading to increasingly large pore sizes with VPI-5, cloverite and more recently mesoporous molecular sieves have broadened the spectrum of these applications. Indeed larger and larger reactant and product molecules can be accommodated in these lattices. These new adsorbant/adsorbate systems create additional needs for experimental data and theoretical descriptions of transport properties, in particular of mono- and multicomponents diffusion coefficients in the zeolite pore lattice. All these questions represent the forefront and current trends of zeolite research. To various extends they deal with the specific factors of the zeolites which allow the fine tuning of the geometric and/or electronic properties of the active sites. It was indeed very rewarding for us, as organizers of this symposium, to realize that all these questions were actually discussed in the papers submitted to the selection committee and that they were widely represented in the selected papers. A feature general to most of these contributions is the combined use of a variety of analytical techniques. Some of these techniques are at the frontiers of the latest analytical developments such as multiple scattering EXAFS and bidimensional MAS-NMR. It is also worth mentioning that the on-going refinements of molecular modelling can now rely on more and more accurate quantum mechanics calculations such as the density functional theory (DFT) improved by introducing high level electron correlation. We wish to thank each and every one of the contributors to the Qu6bec International Zeolite Symposium, who gathered coming from not less than 27 countries to share their recent findings and ideas in a research field so liable to yield future fundamental developments as well as potential technical innovations.

Laurent Bonneviot and Serge Kaliaguine

XII PAPER SELECTION COMMITTEE

Y. BEN TA,M~T L. BONNEVIOT E.M. FLANIGEN Z. GABELICA M. GUISNET W.O. HAAG S. KALIAGUINE H.G. KARGE G.T. KOKOTAILO

L.B. McCUSKER W.J. MORTIER

J.M. NEWSAM T. TATSUMI H. van BEKKUM

Institut de Recherches sur la Catalyse, Villeurbanne, France Universit6 Laval, Qu6bec, Canada UOP, Tarrytown, NY, USA Facult6s Universitaires Notre-Dame de la Paix, Namur, Belgium Universit6 de Poitiers, Poitiers, France Mobil Research and Development Co., Princeton, NJ, USA Universit6 Laval, Qu6bec, Canada Fritz-Haber Institut der Max Planck, Gessleschaft, Berlin, Germany University of Pennsylvania, Philadelphia, PENN, USA and University of British Columbia, Vancouver, Canada Eidgentissische Technische Hochschule, Ztirich, Switzerland Exxon Chemical Europe Inc., Machelen, Belgium Biosym Technologies Inc., San Diego, CA, USA The University of Tokyo, Tokyo, Japan Delft University of Technology, Delft, The Netherlands

ACKNOWLEDGEMENTS

The organizers owe a huge debt of gratitude to Mrs. H61~ne Michel who contributed so efficiently to the Qu6bec Symposium organization as well as to the preparation of these proceedings.

Zeolites: A Refined Tool for Designing Catalytic Sites L. Bonneviot and S. Kaliaguine (editors) 9 1995 Elsevier Science B.V. All rights reserved.

O n e and t w o - d i m e n s i o n a l solid-state N M R investigations of the threed i m e n s i o n a l structures of zeolite-organic sorbate c o m p l e x e s C.A. Fyfe, H. Grondey, A.C. Diaz, G.T. Kokotailo, Y. Feng, Y. Huang, K.C. Wong-Moon, K.T. Mueller, H. Strobl and A.R. Lewis Department of Chemistry, University of British Columbia, 2036 Main Mall, Vancouver, B.C. V6T 1Z1, Canada

1. I N T R O D U C T I O N A most important characteristic of molecular sieve systems which is common to their applications as catalysts, sorbents and in gas separation is the size and shape selectivity toward adsorbed organic molecules conferred by the molecular dimensions of their channel and cage systems [ 1]. Because of their small crystallite dimensions, powder rather than single-crystal XRD techniques must be used. While it is possible to define framework topologies and structures powder X-ray diffraction, particularly if Rietveld analysis and synchrotron radiation are used [2,3], it is very difficult to reliably determine the structures of organic sorbate/framework complexes which would yield important information on the detailed nature of the interactions. Important exceptions in this regard are the single crystal XRD studies of van Koningsveld and co-workers who determined detailed high-quality structures of the highloaded forms of p-xylene and p-dichlorobenzene in zeolite ZSM-5. These are the only reliable zeolite/sorbate structures to date [4,5]. High resolution solid state NMR has emerged in recent years as an important complementary technique to XRD in the investigation of zeolite structures, being particularly sensitive to short to medium range geometries and orderings [6]. For some years we have worked to develop new approaches in the application of solidstate NMR techniques together with XRD studies to the investigation of zeolite structures with the aim of ultimately being able to determine the 3D structures of their complexes with sorbed organic molecules. In this paper, we outline the development of these techniques and their current standing.

2. RESULTS AND DISCUSSION In high-resolution solid state NMR, the widths of the signals from dilute spin-1/2 nuclei are determined by the degree of crystallinity and the perfection of the local ordering. This can be achieved in the case of zeolites by investigating high-quality, completely siliceous systems where there is only the Si(4Si) local environment present. As illustrated in Figures 1A and B for ZSM-12 [7] and ZSM-5 [8], respectively, sharp resonances are observed whose numbers and relative intensities reflect the number and occupancy of the crystallographically inequivalent T-sites in the unit cell. In the case of A1PO4 molecular sieves, there is exact alternation of A1 and P, giving completely and perfectly ordered frameworks in the as-synthesized materials, as shown in Figure 1C for the 31p spectrum of VPI-5 [9]. This result is quite general for perfectly

crystalline and ordered solids. These spectra may be used to monitor various structural transformations, for example, those induced by temperature as in the case of ZSM-5 [ 10], or in the case of A1PO4 materials by the hydration/dehydration of octahedral A1 sites. Of particular importance, they yield information on the interaction of organic sorbates with the molecular sieve framework. For example, Figure 2 shows the 29Si spectrum of ZSM-5 with p-xylene present at a loading of two

A

I

I

I

-107

I

I

I

I

PPM from TMS

-113

A

Cl

0 I

I

I

I

-108

I

I

I

I

I

I

B

c~

C

,r

I

-118

PPM from TMS

C

J

I

25

~

I

0 Frequency

A

I

-25

.JO

-75

(ppm from 83% H3PO4)

Figure 1. 29Si MAS spectra of (A) ZSM-12 and (B) ZSM-5. (C) 31p MAS spectrum of VPI-5 with spinning sidebands indicated by brackets.

--I

-105

!

I

I

I

I

I

I

I

PPM / TMS

I

I

I

I

I

-120

Figure 2. 29Si MAS spectra of ZSM-5 loaded with 2 molecules per u.c. of (A) p-dichlorobenzene, (B) p-chlorotoluene, and (C) p-xylene.

molecules per unit cell (u.c.). Comparison with Figure 1B shows that the number of T-sites has decreased from 24 to 12 indicating a change in symmetry from monoclinic to orthorhombic. Further, the similarities between the spectra in the presence of p-xylene and pdichlorobenzene and p-chlorotoluene (Figure 2) indicate that the interactions, at least in this case, are based on the size and shape of the organic molecule since the CH3 and C1 substituents have the same steric factors but the molecules differ in most other aspects [ 11]. The difficulty in using these spectra further is that the assignment of the resonances to the different T-sites is generally not known, although there may be some information from the intensities if the site occupancies are different. In the case of ZSM-12 and ZSM-5, all of the site occupancies are the same and no assignments are possible. This problem can be solved by using two dimensional homonuclear correlation experiments such as COSY and INADEQUATE to establish the three-dimensional (Si-O-Si) connectivity pattern with the framework when the topology is known [12]. Figure 3 shows such an experiment on ZSM-12 [7]. This yields the assignments of the resonances shown in the figures. The above experiments are based on the scalar Si/Si J-coupling which operates through the bonding network. Similar information may be obtained in the A1PO4 systems from CP and TEDOR experiments as shown in Figure 4 [9]. Although these are based on the heteronuclear dipolar coupling which is a through-space interaction, they are selective for 3tp_ O-2VA1 connectivities because the interaction is very strongly distance dependent and these P/A1 distances are the shortest. Knowledge of the assignments may now be used to gain additional information on the details of the structures and the various changes which they can undergo. For example, zeolite ZSM- 11 is found to undergo a temperature induced transition from a

T3

T+ Ts

T1 T7

T5 T2

, ,

// /J

/

// T1T2 / / /

//

--T+T --.--,.~-.4.' ST // T3T5

0~,,

,/ T3TI

"~

,," T4T6 /

It

I

I

I

I

I

-108.0-109.0 -110.0 -111.0-112"0 -113.0

PPM Figure 3.

29Si INADEQUATE

experiment on ZSM-12.

i

1-------7----.-

I

I

I

I I I I I I I I I I

"

tr

O

o7 "

E o E

q tT.

I I

'

i, . . . . . . .

I

o

9

I

,

'10 Frequency

I

-20

,

I

-30

9

I

,

.40

I,

i_

-so

(ppm from 85% H3PO4)

Figure 4. Two-dimensional 27A1 ~ alp TEDOR experiment on VPI-5. The connectivities are displayed in the dashed box and spinning sidebands are indicated by asterisks. high-temperature form with seven T-sites which matches the proposed framework topology (tetragonal, space group I~,m2) to a lower symmetry form at room temperature (twelve T-sites with equal occupancy). 2D INADEQUATE experiments on both forms yield the assignments of the resonances shown in Figure 5 (see ref. [13] for more details) and from the relationship between the two spectra it can be deduced that the space group of the room temperature form is I4 (tetragonal) and that the phase transition involves the loss of the mirror plane. Information of this type is useful for further investigation of the structure by diffraction techniques. The last step in the extension of these solid state NMR techniques is to apply them to the investigation of the three dimensional structures of zeolite-sorbate complexes. This can be done by using experiments such as cross-polarization and REDOR which are based on the through-space dipolar interaction. Because of the strong distance dependence, the distances between the T-sites in the framework (whose identifies are now known) and nuclei on suitably isotopically substituted substrates may be determined, yielding the 3D structure of the zeolitesorbate complex. To test the validity of this approach we have investigated a number of such experiments applied to the high-loaded form of zeolite ZSM-5 containing p-xylene where the answer is known from the high-quality single crystal structure of van Koningsveld and coworkers [4].

A

I

B

T=302 K 133'

T=342 K

4Z

I

65

3 7

4

2 I

I

I

I

I

-110

I

PPM

2

I

I

I

I

- 118

I

I

I

I

I

- 109

PPM

2

I

I

I

I

-117

Figure 5. 29Si MAS spectra of ZSM- 11 at (A) 302 K and (B) 342 K. Below each spectrum is the deconvolution in terms of Lorentzian curves. In the present paper, the application of the CP technique with protons as the source nuclei will be described as representative of this class of experiments. In order to localize the polarization source as much as possible, experiments were carried out with the two specifically deuterated p-xylenes (1) and (2). CH 3

CD3

CH 3

CD3

(1)

(2)

D

D

Since the CP process is greatly dependent on molecular motions, these must be well understood for the system being studied. In the present work, these were investigated by wide line deuterium NMR of the sorbed organics. It was found that at 6 molecules/u.c., the methyl groups in the organic substrate have rapid C3v rotational motion while the aromatic rings are essentially rigid but a proportion show some low frequency "ring-flips" around the 1,4-axis. The effect of the distance dependence can be seen qualitatively from a comparison of the CP spectra with that from a simple one-pulse experiment as shown in Figure 6. The structure is orthorhombic with 24 T-sites of equal occupancy and the assignments of the resonances come as previously from 2D INADEQUATE experiments [14]. In the CP spectrum some signals are obviously enhanced compared to the others. The resonances due to the T-sites 1, 2, 10, 12 and 16 are quite well resolved and these were used in the study.

CP/MAS

MAS 13,Z314'6'8 I /I 20,~SlIznll

:"),2~ 19,9

7

18

I

I

'

I

-110

'

$

I

-112

'

I

12 1716

'

I

'

3 4

I

-114 -116 -118 PPM from TMS

'

I

"

I

-120

Figure 6. (a) 29Si CP/MAS spectrum of the complex of p-xylene (2) in ZSM-5 at a loading of 6 molecules per u.c. The contact time was 5 ms, with a recycle delay of 5 s. (b) 29Si MAS spectrum of the complex of p-xylene in ZSM-5 at a loading of 8 molecules per u.c. The spin-dynamics of the cross-polarization process from I to S nuclei as a function of time are described by Equation (1) [ 15]. S(t)

= Sma x (1 - Tcp/T1p(H)) -1

(exp(-t/Tlotn)) - exp(-t/Tcv))

(1)

Smax represents the theoretical maximum signal intensity obtainable from the polarization transfer, T1oca) the proton TIp value and Tcp the cross-polarization time constant. Thus the S signal intensity as a function of time should consist of an exponential growth controlled by the cross polarization transfer and an exponential decay due the T1o process. Of particular interest, Tcp can be related to the second moment of the IS dipolar interaction, (Ao~2)Is, as in Equation (2) and is proportional to rls6 as in Equation (3), [16]. 1

Tce

-

C (A(o 2 )IS

(2)

~ ( A('02 )II 2

1

Tcp

o~ (A002 )IS oc

2

7i~s E ris6

(3)

(a)

Si 12 x

12

--~--

Si 3 10 _..~tlB

Si 17

9

'si ~o-Si 1 Si12

" " ' I t .... ----D.--

8

Si17 Si 16

-"1"-. ---o--e

m

I

M

c o

Si 3

'

r&,

i

i

SI 16

qmo

Si 1

c

m

Si 10

O~ 0

10

20

30

Contact

(b)

40

time

50

60

(ms)

Si 1

1.2

Si 17 Si 10 1.0

A

=~--...._ m ~ ' m ~ ' " " . ~

=

Si 3 Si 16

r

. . ~ . . . . - . =. ~ t . ~ . . - 2 ~ ~ . . . .- .

0.8

I=

Si 12

>., i m

c o

0.6

"-=-"

:' F

C m

0.4

,S

/

. . . . x .... "'"Q""

0.2

-"-P-

e~

......s~ ~0

II

Si 1 Si 12

Si 17 Si~6

---=---si3 0.0

o

~o

2o Contact

so time

4o

so

eo

(ms)

Figure 7. Variation of the intensity as a function of the contact time for (a) the complex of pxylene (2) in ZSM-5 at a loading of 6 molecules per u.c., and (b) the complex of p-xylene (1) in ZSM-5 at a loading of 6 molecules per u.c.

Using these equations, the cross-polarization results can be related to the sorbate-lattice distances in the complex. Figure 7 shows the experimental CP curves for the well-resolved resonances of Figure 6 for complexes of p-xylenes (1) and (2) [17]. Qualitatively, it can be seen that silicons 12, 3 and 17 are much more efficiently polarized and hence closer to the ring protons than silicons 16, 10, 1 (Figure 7a), and that silicons 10, 1, 17 and 3 are closer to the methyl protons than silicons 16 and 12 (Figure 7b). More quantitatively, the data can be compared directly with the XRD data. The structure is found to have two sites, one at the channel intersections and one in the zig-zag channels (Figure 8), and the lattice sorbate distances and related second moments can be calculated. The CP curves in Figure 7 can be fitted by deriving the Tip(H) values from the clearest decays and using these data to deduce the Tcp values are shown in Table 1. Figure 9 shows the plot of the experimental Tcp values versus the calculated second moment values, validating the general approach used.

Si-2 Si-3

Si-4

Si-5 Si-13

Si-6

/ C

C~ C--

'

,.c

/

Si-17

,

rc

i

ic c

i '

/,, C C

I

I

CH3

Figure 8. View approximately along [001] of the single crystal structure of p-xylene in ZSM-5 at a loading of 8 molecules per u.c. Oxygen and hydrogen atoms are omitted for clarity.

(a)

(b)

180 T

450 T

160--

Sil0

Sil

4 0 0 - |-

140 -120 -"7

350 --

Si 17 Si3.

300 "7

100 -13. O

80--

~ 13. O

"

Si12

~ 12

~16

~

250

-

200

60 -

150

40-

100

| si 1/ t - / " s, ,6

" 0

~ 250

0 500

750

1000

Calc. second moment (Hz 2)

0

10000

I 20000

I 30000

Calc. second moment (Hz 2)

Figure 9. Plot of experimental Tcp values versus the calculated second moment (Ao32 )IS values for (a) the complex of p-xylene (2) in ZSM-5 at a loading of 6 molecules per u.c., and (b) the complex of p-xylene (1) in ZSM-5 at a loading of 6 molecules per u.c. Thus, in the one case where the structure is well described, the CP technique yields a result in good agreement with the known structure. Similar results have been obtained using the REDOR [18] technique. It is thus felt that with proper precautions and with particular attention to the motions of different parts of the organic guest molecules, solid-state NMR spectroscopy will be a viable technique for the determination of the structures of zeolite/sorbate complexes and that it will yield valuable complementary information to that from diffraction studies of these systems. It should be possible to combine the results from solid-state NMR with molecular dynamics calculations in an interactive manner.

3. A C K N O W L E D G E M E N T S C.A.F. acknowledges the support of the NSERC of Canada in the form of operating and equipment grants, Y.H. and K.T.M. for Postdoctoral Fellowships, and K.C.W-M. and

10 H.S. for Postgraduate Scholarships. A.C.D. thanks INTEVEP S.A. for a Postgraduate Scholarship. REFERENCES

1.

2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14.

15. 16. 17. 18.

(a) Meier, W.M. In Molecular Sieves. SCI Mong., 1968. (b) Breck, D.W. Zeolite Molecular Sieves; Wiley Interscience: New York, 1974. (c) Smith, J.V. Zeolite Chemistry and Catalysis; Rabo, J.A., Ed. ACS Monograph Series 171, American Chemical Society: Washington, DC, 1976. (d) Barrer, R.M. Zeolites and Clay Minerals as Sorbents and Molecular Sieves; Academic Press: London, 1978. W.I.F.; Harrison, W.T.A.; Johnson, M.W. High Resolution Powder Diffraction. Materials Science Forum; Catlow, C.R.A., Ed.; Trans Tech. Publication: Andermannsdorf, Switzerland, 1986; Vol. 9, pp. 89-101. The use of synchrotron x-ray sources may permit single crystal studies on much smaller samples. Eisenberger, P.; Newsam, J.B.; Leonowicz, M.E.; Vaughn, D.E.W. Nature, 1984, 309, 45. van Koningsveld, H.; Tuinstra, F.; van Bekkum, H.; Jansen, J.C. Acta Cryst., 1989, B45, 423. van Koningsveld, H.; Jansen, J.C. Private communication, to be published. Fyfe, C.A.; Gobbi, G.C.; Murphy, W.J.; Ozubko, R.S.; Slack, D.A.J. Am. Chem. Soc., 1984, 106, 4435. Fyfe, C.A.; Feng, Y.; Gies, H.; Grondey, H.; Kokotailo, G.T.J. Am. Chem. Soc., 1990, 112, 3264. Fyfe, C.A.; Strobl, H.; Kokotailo, G.T.; Kennedy, G.J.; Barlow, G.E.J. Am. Chem. Soc., 1988, 110, 3373. Fyfe, C.A.; Mueller, K.T.; Grondey, H.; Wong-Moon, K.C.J. Phys. Chem., 1993, 97, 13484. (a) Fyfe, C.A.; Kennedy, G.J.; Kokotailo, G.T.; Lyerla, J.R.; Fleming, W.W.J. Chem. Soc., Chem. Commun., 1985, 740. (b) Hay, D.G.; Jaeger, H.; West, G.W.J. Phys. Chem., 1985, 89, 1070. Fyfe, C.A.; Strobl, H.; Gies, H.; Kokotailo, G.T. Can. J. Chem., 1988, 66, 1942. Fyfe, C.A.; Feng, Y.; Grondey, H.; Kokotailo, G.T.; Gies, H. Chem. Rev., 1991, 91, 1525. Fyfe, C.A.; Feng, Y.; Grondey, H.; Kokotailo, G.T.; Mar, A. J. Phys. Chem., 1991, 95, 3747. (a) Fyfe, C.A.; Grondey, H.; Feng, Y.; Kokotailo, G.T.J. Am. Chem. Soc. 1990, 112, 8812. (b) Fyfe, C.A.; Grondey, H.; Feng, Y.; Kokotailo, G.T. Chem. Phys. Lett., 1990, 173, 211. (c) Fyfe, C.A.; Feng, Y.; Grondey, H.; Kokotailo, G.T.J. Chem. Soc., Chem. Commun., 1990, 1224. Mehring, M. Principles of High Resolution NMR in Solids. Second edition, 1983, Springer, Berlin. Pines, A.; Gibby, M.G.; Waugh, J.S.J. Chem. Phys., 1973, 59, 569. Fyfe, C.A.; Diaz, A.C.; Grondey, H.; Fahie, B. To be published. (a) Guillon, T.; Schaefer, J. J. Magn. Reson., 1989, 81,196. (b) GuiUon, T.; Schaefer, J. Adv. Magn. Reson., 1989, 13, 55.

Zeolites: A Refined Tool for Designing Catalytic Sites L. Bonneviotand S. Kaliaguine (editors) 9 1995 Elsevier Science B.V. All rights reserved.

11

The use of small and weakly basic probe molecules for the investigation of Bronsted acid sites in zeolites by N M R spectroscopy Eike Brunner Universit/it Leipzig, Fakult/it for Physik und Geowissenschaften, Linn6str~e 5, D-04103 Leipzig, Germany Small and weakly basic probe molecules were used for the characterization of Bronsted acid sites in zeolites by variable-temperature 1H and 13C NMR spectroscopy. It could be shown that the induced ~H NMR chemical shift A& of free surface hydroxyl groups caused by the adsorption of probe molecules as, e.g., CO and C2C14reflects the strength of acidity of the hydroxyls. The deprotonation energy AEDp of surface hydroxyl groups can be determined from A& if the influence of rapid thermal motions and/or exchange processes is suppressed. Furthermore, the use of the ~H NMR chemical shift &H..M of surface hydroxyl groups influenced by adsorbed probe molecules M as a sensitive qualitative measure for the strength of acidity is suggested. The geometry of the adsorption complexes formed by CO molecules hydrogen bonded to Bronsted acid sites could be determined by NMR spectroscopy.

1. INTRODUCTION The catalytic behaviour of H-zeolites with respect to Bronsted-acid catalyzed reactions is determined by the strength of acidity, the concentration, and the accessibility of the bridging hydroxyl groups which are known to act as Bronsted acid sites [1,2]. The strength of gas phase acidity of a surface hydroxyl group TO-H is defined as the inverse value of the Gibbs free energy change n GDp of the deprotonation

TO-H

~. T O -

+ H +

(1)

It could be shown [3] that AGDp is the sum of the deprotonation energy AEDp (heterolytic dissociation energy) and a constant contribution for surface hydroxyl groups in zeolites. Therefore, AEDpis a convenient measure for the strength of gas phase acidity. However, the spectroscopic measurement of the deprotonation energy ~EDp is still a subject of discussion. Both, IR and 1H MAS NMR spectroscopy allow the direct investigation of bridging hydroxyl groups (see, e.g., [1,2]). Very often the potential energy of the O-H stretching vibrations is approximated by the Morse potential. A dissociation energy Do can be calculated (see, e.g., [4,5]) from the wavenumbers Yon and v02 of the fundamental stretching vibration and the corresponding first overtone, respectively, according to

12

hcv oa Do ~"

(2)

4X

with

1 2YouZ = ~

V02

(3)

3VoH - v02

h denotes Planck's constant and c the speed of light in empty space. It has however been demonstrated by Kazansky [4,5] that Do is close to the value for the homolytic dissociation energy of the surface hydroxyl groups which strongly deviates from AEDp. Jacobs and Mortier [6,7] have found that VOH of free bridging hydroxyl groups (i.e., bridging hydroxyl groups which are not influenced by additional electrostatic interactions with the zeolite framework) correlates well with the intermediate Sanderson electronegativity. A corresponding correlation could also be found between the 'H NMR chemical shift ~H of free bridging hydroxyl groups and the intermediate Sanderson electronegativity. Therefore, Pfeifer et al. [1,8,9] have suggested to make use of 8H as a measure for the strength of acidity of free surface hydroxyl groups. This suggestion was confirmed by quantum chemical calculations [10] which have shown that AEDp and bn are linearly correlated for hydroxyl groups bound to T-atoms (B, A1, Si or P) whose first coordination sphere consists of oxygen atoms only. The slope of AEDp amounts to - 84 _ 12 kJ mol-' ppm -1. Since chemical shift differences can be measured with an experimental error of __+0.1 ppm it is possible to determine differences in the deprotonation energy of free surface hydroxyl groups with an accuracy of + 8 kJ mo1-1. Recent quantum chemical calculations [ 11] revealed that the deprotonation energy of terminal SiOH groups amounts to 1400 ___25 kJ mol-~ which excellently agrees with the experimentally determined value of 1390 kJ mol -~ [12]. Since the ~H NMR chemical shift of terminal SiOH groups amounts to 2.0 + 0.1 ppm the deprotonation energy AEDp can be calculated [13] according to

AF-,Dp -

kJmo1-1

1570-

84

8H

.

(4)

ppm

Paukshtis and Yurchenko [12,14] have developed another method for the determination of differences in the deprotonation energy of TO-H groups. This method is based on the measurement of the induced wavenumber shift Av = VOH...M - VOH, where VOH...Mdenotes the wavenumber of the stretching vibration of the surface hydroxyl groups influenced by adsorbed probe molecules M. Provided that (-Av) ~ 400 cm-' (weak hydrogen bonding) the deprotonation energy can be calculated according to the formula

13

A EDp

A ~ SiOH =

kJmo1-1

"-'~DP

_ 1

kdmo1-1

log

A

A._____~_v

(5)

A v s~on

with A = 0.00226 [12] and AEDpSiOH = 1400 kJ mo1-1 (see above). The deprotonation energy of different types of surface hydroxyl groups in zeolites was determined successfully by this method by the use of CO as the probe molecule M [15,16]. On the other hand, it is known that the formation of hydrogen bonds leads to a considerable broadening of the stretching vibration bands of surface hydroxyl groups. This leads to a relatively large experimental error for Av which limits the accuracy of the measurement especially of small differences in the deprotonation energy. It is known that the interaction between surface hydroxyl groups and adsorbed probe molecules M causes an induced 1H NMR chemical shift A8 = 8H..M - 8H, where 8H.M denotes the chemical shift of the surface hydroxyl groups influenced by the probe molecules. It could be shown [ 17] that ~H and VoH are linearly correlated at least in limited ranges. The slope of ~H amounts to 0.0147 ppm/cm-' for surface hydroxyl groups in zeolites and to 0.0092 ppm/cm-' for hydrogen bonded protons in various solids. It can therefore be supposed that the correlation between A b and (-A v) is given by

a8

ppm

_

(-av) -1

cm

(6)

with B-values between 0.0092 and 0.0147. If this is true it should be possible to make use of A ~ instead of Av in eq. (5).

2. EXPERIMENTAL NMR and diffuse reflectance FTIR measurements have been carried out on identical samples which were prepared in the following manner: Glass tubes were filled with the hydrated zeolite (bed depth: ca. 8 ram) and heated up to 673 K with a heating rate of 10 K/h under permanent evacuation. At this temperature the samples were further evacuated for 24 h at a final pressure of 10.2 Pa. Then the samples were cooled to 77 K and loaded with definite amounts of probe molecules (CO or C2C14). 13C enriched substances were used for the 13C NMR spectroscopic investigation of the adsorption state of the probe molecules. After loading the samples were sealed. NMR spectroscopic investigations have been carried out on a Bruker MSL 500 spectrometer at low temperatures since small probe molecules as CO can exhibit rapid thermal motions at room temperature. The lowest temperature achievable under magic angle spinning conditions in our laboratory yet amounts to 123 K. In contrast, static 13C NMR investigations were carried out at temperatures down to 4.5 K. All NMR chemical shifts are given relative to tetramethylsilane (TMS).

14 3. RESULTS AND DISCUSSION 3.1. General remarks The influence of small and weakly basic probe molecules on the low-temperature ~H MAS NMR spectra of zeolites was firstly investigated by White et al. [18]. Table 1 summarizes induced chemical shift values for a variety of probe molecules. Table 1 Induced ~H NMR chemical shift A8 of bridging hydroxyl groups in zeolite H-ZSM-5 loaded with different amounts of probe molecules M per framework A1 atom (A1F) measured at a temperature of 123 K. The data are taken from ref. [18]. probe molecule M

N2

CO

A 8/ppm (1 M/A1 F)

-

1.8

AS/ppm (2 M/A1 F)

0.3

1.8

C2H6

0.6

C2H4

C2H2

2.7

3.5

2.7

3.9

Convenient probe molecules should exhibit the following properties: (i) The induced 1H NMR chemical shift A8 caused by the probe molecules should be high since its relative experimental error is then low. (ii) A 8 should remain constant for coverages higher than 1 probe molecule per surface hydroxyl group. Otherwise the dependence of n 8 on the coverage has to be taken into account which complicates the use of A8 as a measure for the strength of acidity. (iii) The molecules must be chemically stable even in the presence of acid sites in zeolites. Conditions (i) and (ii) are obviously fulfilled for CO and C2H4. However, it turned out that C2H4 rapidly chemically reacts at room temperature in H-zeolites which makes its use as a probe molecule difficult. Therefore, CO and C2C14 (instead of C2H4) were chosen as promising candidates. 3.2. The interaction of CO molecules with bridging OH groups in zeolites The data collected in Table 2 (see next page) show that the induced 1H NMR chemical shift A8 caused by the interaction of surface hydroxyl groups in H-ZSM-5 with CO molecules qualitatively reflects the strength of acidity of the hydroxyls. However, it is doubtful that the induced chemical shift of bridging hydroxyl groups in the large cavities of zeolite 0.3 HNa-Y is smaller than that of the AIOH groups on non-framework A1 species in H-ZSM-5. Furthermore, the induced wavenumber shift of the free bridging hydroxyl groups in H-ZSM-5 amounts to ca. 330 cm -~ which is in good agreement with the results reported in refs. [15,19,20]. According to eq. (6) one would then expect an induced ~H NMR chemical shift of ca. 3.0 - 4.9 ppm which is in contradiction to the experimentally observed value of only 2 ppm. It could be shown that the CO molecules exchange rapidly between the bridging hydroxyl groups even at 123 K [18,21] and that the induced chemical shift A 8 has not reached its maximum value [22] which explains this contradiction. Rigid 13CO molecules should exhibit a broad signal with the typical shape of a line dominated by the chemical shift anisotropy. The principal values of the chemical shift tensor of rigid ~3COare 81~ = 822 = 8~ = 305 ppm and 8 3 3 = 81 ~---- - 48 ppm [23], i.e., the chemical

15 shift anisotropy amonts to Aoc = 353 ppm. The corresponding line shape could be found [24] for the 13C NMR signal of 13CO molecules physisorbed on silicalite measured at 4.5 K. Table 2 Experimental data for the ~H NMR chemical shifts 8i~ and 8H..CO as well as the induced chemical shift A8 of surface hydroxyl groups in zeolites. The measurements were carried out at 123 K. It should be mentioned that bridging hydroxyl groups influenced by an additional electrostatic interaction with the zeolite framework are not considered here. zeolite

hydroxyl group structure

H-ZSM-5 hydrothermally treated

0.3 HNa-Y

8H/ppm

bH...co/ppm

A8/ppm

SiOH

2.0

2.0

0.0

A1OH (on non-framework A1)

2.9

3.9

1.0

SiOHA1 (free)

4.2

6.2

2.0

SiOHAI (in the large cavities)

3.9

4.8

0.9

The 13C NMR spectrum of ~3CO molecules adsorbed on a zeolite H-Y measured at 123 K consists of a single narrow line at ca. 185 ppm indicating that the CO molecules still move isotropically. At a temperature of 4.5 K the spectrum exhibits a width corresponding to that expected for rigid 13CO molecules (see Fig. 1). It is important to mention that the transition from the narrow line observed for 123 K to the broad signal shown in Fig. 1 takes place at temperatures between 60 and 80 K. The 1H MAS NMR spectroscopic investigation of the interaction between surface hydroxyl groups and CO molecules should therefore be carried out at temperatures below 60 K which will be possible in the near future.

__...

S / ppm

[

I

600

400

r

l

200

r

I

0

I

-200

1

-400

Figure 1. 13C NMR spectrum of 13CO adsorbed on zeolite H-Y (Si/A1 = 2.5) measured at 4.5 K. The sample was loaded with ca. 26 molecules 13CO per unit cell.

16 It is remarkable that the characteristic line shape shown in Fig. 1 can only be observed for CO coverages which are not higher than the concentration of accessible bridging hydroxyl groups. It could be shown by IR spectroscopic investigations that all CO molecules are then adsorbed on bridging hydroxyl groups. The characteristic features of the spectrum shown in Fig. 1, namely the two maxima at the left edge of the spectrum and the "step" at the right edge vanish for coverages higher than the concentration of accessible bridging hydroxyl groups. In the latter case IR spectroscopic investigations reveal the existence of physisorbed CO besides CO molecules adsorbed on bridging hydroxyl groups. Therefore, it has to be concluded that the characteristic shape of the spectrum shown in Fig. 1 is not caused by a superposition of two different signals due to 13CO molecules in different adsorption states. It is known from IR spectroscopic investigations as well as quantum chemical calculations [25-27] that the adsorption of CO molecules on bridging hydroxyl groups leads to the formation of linear adsorption complexes O-H CO where the C atom is preferentially attached to the proton. Therefore, the heteronuclear magnetic dipole interaction between the proton and the 13C nucleus of the ~3CO molecule has to be taken into account besides the chemical shift.anisotropy. The spectrum numerically calculated for this complex with a H-C hydrogen bond distance rH_c of 0.21 ___ 0.01 nm and AOc ~ 370 ppm excellently agrees [24] with the experimentally observed spectrum. This H-C hydrogen bond distance is in complete accordance with the predictions of quantum chemical calculations [25-27]. Furthermore, it is remarkable that AOc increases compared with physisorbed ~3CO. It can be stated that variable-temperature 13C NMR spectroscopy allows the study of the geometry and the thermal motions of the adsorption complexes formed by CO molecules hydrogen bonded to surface hydroxyl groups.

3.3. The influence of C2Cl 4 on the IH MAS NMR spectra of zeolites Fig. 2 shows the 1H MAS NMR spectrum of zeolite 0.3 HNa-Y loaded with 8 molecules C2CI4 per unit cell which is less than the actual concentration of ca. 17 bridging hydroxyl groups per unit cell. Obviously, a part of the signal at ~H = 3.9 ppm due to bridging hydroxyl groups in the large cavities is shifted to a position of ~H c2c~4 = 5.5 ppm.

Figure 2. 1H MAS NMR spectrum of zeolite 0.3 HNa-Y loaded with 8 molecules C2C14 per unit cell measured at 123 K.

. . . .

i

8

. . . .

i

7

. . . .

i

6

. . . . .

i

. . . .

5

i

. . . .

4

/ ppm

|

3

. . . .

i , , , , , i

2

. . . .

1

17 The spectrum measured at room temperature exhibits only one line at 4.5 ppm instead of the two signals at 3.9 and 5.5 ppm which shows that the C2C14molecules exchange rapidly between the bridging hydroxyl groups at 293 K. In contrast, the presence of the two well resolved signals at 123 K evidences that the exchange processes of the C2C14 molecules between the bridging hydroxyl groups are slow. Furthermore, it could be shown that the induced chemical shift becomes independent of the temperature for T < 150 K, i.e., it can be assumed that the maximum value of A8 is reached. Fig. 3 demonstrates that the correlation between A8 and (-Av) is given by a straight line with a slope of B = 0.01 [28] as it was predicted above (cf. eq. (6)). Therefore, the induced chemical shift A6 measured at 123 K for samples loaded with C2C14 is not only a qualitative measure for the strength of acidity of free surface hydroxyl groups. It can also be used for the calculation of the deprotonation energy according to eq. (5) since Av and n VSiOHcan be replaced by A~ and A~SiOH , respectively. An induced chemical shift of AbsioH = 0.75 ppm was determined for SiOH groups [28]. The induced chemical shift A 6 amounts to 1.6 ppm for bridging hydroxyl groups in the large cavities of 0.3 HNa-Y. From eq. (5) it follows that AEDp ~ 1250 kJ mo1-1 which is close to the value of ca. 1240 kJ mol -~ obtained with bH = 3.9 ppm by using eq. (4). Furthermore, it can be suggested to make use of 8H C2C14as a sensitive qualitative measure for the strength of acidity since both 8H and n 8 reflect the strength of acidity of free surface hydroxyl groups.

2,2

-

2,0

9

1,8

9

9

1,6

E Q. 1,4 ,,o 1,2 1,0 0,8 0,6 n

60

80

100

120

140

-Av / cm

160

180

200

220

-1

Figure 3. Correlation between the induced 1H NMR chemical shift A8 and the induced wavenumber shift nv caused by adsorbed C2C14 for free surface hydroxyl groups in zeolites.

18 ACKNOWLEDGEMENT The author wishes to express his gratitude to Prof. Dr. Dr. H. Pfeifer and Dr. B. Staudte for helpful discussions and to Mr M. Koch and Ms H. Sachsenr6der for excellent experimental assistance. Financial support by "Deutsche Forschungsgemeinschaft" (SFB 294 "MolekiJle in Wechselwirkung mit Grenzflfichen") is highly appreciated. REFERENCES .

2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23.

24. 25. 26. 27. 28.

H. Pfeifer, NMR Basic Principles and Progress, Vol. 31, Springer, Berlin 1994., p.31. H.G. Karge, Stud. Surf. Sci. Catal., 65 (1991) 133. J. Sauer, J. Mol. Catal., 54 (1989) 312. V.B. Kazansky, Kinetika i Kataliz, 21 (1980) 159. V.B. Kazansky, Stud. Surf. Sci. Catal., 85 (1994) 251. P.A. Jacobs, Catal. Rev.-Sci. Eng., 24 (1982) 415. P.A. Jacobs and W.J. Mortier, Zeolites, 2 (1982) 226. H. Pfeifer, D. Freude and M. Hunger, Zeolites, 5 (1985) 274. D. Freude, M. Hunger and H. Pfeifer, Z. Phys. Chemie NF, 152 (1987) 171. U. Fleischer, W. Kutzelnigg, A. Bleiber and J. Sauer, J. Am. Chem. Soc., 115 (1993) 7833. J. Sauer and J.-R. Hill, Chem. Phys. Lett., 218 (1994) 333. E.A. Paukshtis and E.N. Yurchenko, Usp. Khim., 52 (1983) 426. E. Brunner and H. Pfeifer, Z. Phys. Chemie, in press. E.A. Paukshtis and E.N. Yurchenko, React. Kinet. Catal. Lett., 16 (1981) 131. L. KubelkovS., S. Beran and J.A. Lercher, Zeolites, 9 (1989) 539. M.A. Makarova, A. Garforth, V.L. Zholobenko, J. Dwyer, G.J. Earl and D. Rawlence, Stud. Surf. Sci. Catal., 84 (1994) 365. E. Brunner, H.G. Karge and H. Pfeifer, Z. Phys. Chemie, 176 (1992) 173. J.L. White, L.W. Beck and J.F. Haw, J. Am. Chem. Soc., 114 (1992) 6182. L.M. Kustov, V.B. Kazansky, S. Beran, L. Kubelkovfi and P. Jir~, J. Phys. Chem., 91 (1987) 5247. I. Mirsojew, S. Ernst, J. Weitkamp and H. Kn6zinger, Catal. Lett., 24 (1994) 235. E. Brunner, J. Mol. Struct., in press. M. Koch, E. Brunner, D. Fenzke, H. Pfeifer and B. Staudte, Stud. Surf. Sci. Catal., 84 (1994) 709. A.J. Beeler, A.M. Orendt, D.M. Grant, P.W. Cutts, J. Michl, K.W. Zilm, J.W. Downing, J.C. Facelli, M.S. Schindler and W. Kutzelnigg, J. Am. Chem. Soc., 106 (1984) 7672. M. Koch, E. Brunner, H. Pfeifer and D. Zscherpel, Chem. Phys. Lett., 228 (1994) 501. S. Bates and J. Dwyer, J. Phys. Chem., 97 (1993) 5897. P. Geerlings, N. Tariel, A. Botrel, R. Lissillour and W.J. Mortier, J. Phys. Chem., 88 (1984) 5752. K.M. Neyman, P. Strodel, S.P. Ruzankin, N. Schlensog, H. Kn6zinger and N. ROsch, Catal. Lett., 31 (1995) 273. H. Sachsenr6der, E. Brunner, H. Pfeifer and B. Staudte, in preparation.


19

D e t e r m i n a t i o n o f the e n v i r o n m e n t o f titanium a t o m s in TS-1 silicalite by Ti Kedge X - r a y A b s o r p t i o n S p e c t r o s c o p y , 29Si and 1H N u c l e a r M a g n e t i c R e s o n a n c e . L. Le Noc a, C. Cartier dit Moulin a'b, S. Solomykina a, D. Trong On a, C. Lortie a, S. Lessard a and L. Bonneviot a aFacult6 des Sciences et G6nie, Universit6 Laval, G 1K 7P4 Sainte Foy (Qc), Canada bLaboratoire pour l'Utilisation du Rayonnement Electromagn6tique, CNRS-CEA-MENJS, B~timent 209d, 91405 Orsay Cedex, France The titanium environment and the modification engendered by the framework when this element is incorporated in a silicalite of orthorombic MFI structure were investigated using Ti K-edge X-ray Absorption, 298i and IH MAS NMR spectroscopy. A multiple scattering analysis of the EXAFS signal reveals that the titanium occupies a substitutional site characterized by Ti-O-Si bond angle, Ti-O and O-Si average distances consistent with a lattice expansion. The defects concentration remains constant and independent of the Ti loading. The proton NMR spectra of S-1 is characterized by narrow resonance lines whose relative intensities are modified in TS-I which indicate either the presence of new TiOH groups and/or a redistribution of the internal SiOH groups.

INTRODUCTION Microporous titanium silicalites of MFI or MEL structures, TS-1 and TS-2 respectively, have attracted many scientists owing to their unique capability to catalyze the oxyfunctionalization reactions of alkanes using H202 as oxidant. Despite the number of new titanium silicalites with larger and larger pore sizes such as Ti-beta, titanium mesoporous materials such as Ti-MCM-41, Ti-MCM-48 or high surface area amorphous TiO2-SiO2, no other materials were active for this demanding reaction [1-4]. The presence of micropores might be necessary for oxyfunctionalization to take place, though a specific local environment around titanium might as well be required for these reactions. These materials exhibit an IR band at 960 cm 1. It is assigned to a local vibration mode of the asymmetric stretching of a SiO4 unit linked to either a Ti 4+, a V 4+ or a Cr 3+ ion in amorphous or crystallyzed systems [57]. Consequently, this band is not an intrisic characteristic of titanium silicalite materials. The Ti isomorphous substitution for Si in these systems is a widely shared belief based on a conjunction of facts obtained from a panel of techniques such as XRD [8], IR, UV-visible, XPS [9], EXAFS [10], XANES [10] rather than on a direct proof. Recently, using the multiple scattering (MS) theory to analyze the second shell of neighbours around the titanium

20

in a TS-1, some of us calculated an average Ti-O-Si angle of about 160 ~ consistent with the isomorphous substitution proposal [~1]. Refined MS calculations on TS-1 free of Ti extraframework species, 29Si and H MAS NMR investigations of the framework modifications in presence of Ti are reported here. EXPERIMENTALS Three titanium silicalites with a Ti/(Ti+Si) atom ratio of 0.005, 0.010 and 0.015 were synthesized from a gel containing a mixture of TEOS, TEOT and TPAOH in propanol. Water was added. The hydrothermal crystallization was performed in a Teflon-lined stainless steel autoclave at 175~ for 4 days. The solid was filtered, washed and calcined at 5Q0~ The XRD recorded on a Rigaku D-Max IIIVC X-ray spectrometer revealed a crystallinity of about 95%. These samples exhibit features that characterizes TS-1 materials without extraframework Ti species, i) an MFI orthorhombic structure with a unit cell expansion of 25 A 3 per Ti, ii) an IR band at 960 cm -t and iii) no UV bands in the 270-400 nm UV range where the electronic transition of TiO2 anatase arises and iv) the characteristic X-ray near edge features of tetrahedral environment [ 10]. The X-ray Absorption measurements at the Ti K-edge were performed at the radiation synchrotron facility of the LURE (France). The white radiation was monochromatized using a Si(111) two crystals monochromator and harmonic rejection was obtained by detuning the crystals. The EXAFS signals were measured in transmission mode from 4800 to 5800 eV with 2 eV steps on samples pressed into pellets. The samples were dehydrated under vacuum at 500~ and transferred to a leak tight XAS cell under controlled atmosphere. The signal was extracted and normalized using a software package developed by Michalowicz [12]. The Fourier transforms were calculated after applying a parametrized Kaiser window function (t 3.7) between 3.16 and 14.20 A -1. The first shell was analyzed using the single scattering (SS) theory combined to an automated mean square minimization curve fitting process [12]. The second shell analysis that necessitated MS calculations were performed using the FEFF6 code from Rehr et al [ 13]. These calculations are usually performed on a priori known structures. Therefore the structures were generated using Cerius 2 TM, a molecular modeling package from Molecular Simulations. A trial and error approach was adopted to reach the structure that led to the best fit of the experimental data. The NMR measurements were also performed on TS-1 dehydrated at 500~ and transferred under controlled atmosphere into the rotors. No rehydration occured during the experiments as it was probed by tH NMR. The MAS NMR spectra were obtained at room temperature using a Bruker ASX 300. The rotors were spun at a rotation frequency of 4kHz. For 298i NMR, an onepulse sequence with a n/2 pulse of 5ps was used. The cycle time of 60s allows a full relaxation of the spins of all the silicon species. IH spectra were recorded following an echo sequence (n/2-x-n) where the echo time x is equal to the inverse of the spin rate, n/2 is 51as and the cycle time is fixed to 2.5s (full relaxation). The shifts are referenced to an external TMS standard.

21

~

:, , i , , . , i , , ~, i , ,', _ i r , , I , ,

9X - 0 . 4

_

:.i-

1.2

..'" ,,

,

i

4

'..

',..."r

i,

'..

....,- ......

,

6

I t i

8

,

12 I. ' ' . ' '

",..,. ....... 9.........

i

l,

10

,

I

~"

I '' ' ' I ....

~, ,.r,___

_

.

, , ...... i'1

12

....

~

14

0

.e

i

I

2

o

I . . . .

4

3

-

5

R(A)

k(A -i)

Figure 2. Module of forward transforms (uncorrected for phase the experimental (solid line) and FEFF6 (dots) model shown in fig.

Figure 1. Normalized EXAFS spectra of [1.5] TS-1 (solid line) compared to SS and MS-FEFF6 calculations using a [Ti(OSi)4] cluster with a geometry given in the text.

.4 pl i

Fourier shift) of the MS1.

........ , ....... .... ;

I II I I I'I I I'I I I I'"I I I'I I I I'I I l,lal i

f..t.,

"g o 2

4 1

2.s

3

RCA)

3.s

1

ltliltltlilll tt lit ill= ---Ii -

I

4

2.5

3

R C A ) 3.s

4

Figure 3. Imaginary part of forward Fourier transforms of the second shell of experimental (solid line) and MS-FEFF6 calculations (dots) for sets of Si-O distance and Ti-O-Si angle, a) 1.638 A and 161 ~ (as in fig. 1 and 2) and b) 1.62 A and 163 ~ respectively.

RESULTS

The X-ray near edge region for the three dehydrated TS-1 not shown here are identical and similar to those shown in the litterature for extraframework Ti free TS [10]. In addition, the similarity of the XANES of TS-1 and a model compound, the hexadecaphenyloctasiloxyspiro(9,9)titanium(IV) (HDPOSST), made of a tetrahedral [Ti(OSi)4] core strongly suggests that the titanium in the silicalite framework adopts the same symmetry and coordination compatible with a substitutional site [11]. The SS analysis of the first shell EXAFS

22

contribution led to a Ti-O distance of 1.80 +0.01 /~ also consistent with a tetrahedral environment. The multiple scattering calculation was proved to be necessary to reproduce the EXAFS signal of the HDPOSST [11]. This was attributed to widely open Ti-O-Si angles spanning in the 156-173 ~ range. A satisfying fit was obtained for calculations restricted to the tetrahedral [Ti(OSi)4 ] core of this molecule (Rmax = 4/~). A similar quality of fit was also obtained using average values 1.782/~ for Ti-O distance and 166 ~ for Ti-O-Si angle instead of the actual values. This demonstrated that the technique provides average values. A [Ti(OSi)4] cluster of atoms was also used to simulate the experimental EXAFS data of TS-1 samples (Fig. 1 and 2). To improve the match with the experimental data, the Ti-O distance and the Ti-O-Si angle were optimized resulting in a new set of 1.805 A and 163 ~ Further refinements were obtained by trial and error changing the O-Si distance as well. This led to a new O-Si distance of 1.638 ,~ and a Ti-O-Si angle of 161 ~ The accuracy of the process was estimated to be + 0.01 A on the distance and • 1~ on the angle. The Debye-Waller factors (DWF) that rely on the thermal agitation and structural disorder were found to be ol 2 = 0.00155 for the first shell and ~22 = 0.0045 for the second shell (S02 = 0.85) for the model compound [11]. For the TS-1 samples the DWF of the second shell had to be increased up to 0.0089. Keeping ol 2 and or22 the same as for the model compound would have led to 3 Si, assuming a [Ti(OSi)3(OH)] type of site. The 29Si MAS NMR was performed to identify the structural defects from vacancies of silicon implying nested OH or broken Si-O-Si bridges. The 29Si NMR spectra (fig.4) of the titanium free silicalite (S-l) and the TS-1 materials exhibit an intense and broad signal at -113 ppm assigned to Q4 [Si(OSi)4] species and a less intense signal at -103 ppm assigned to Q3 [Si(OSi)3OH ] species. No geminal silanol species were observed in the TS-1 or S-1 samples, even when cross polarization was used. The number of Q3 species is hardly constant and nearly equal to 8 per unit cell (8.2, 7.9, 7.2, 7.8 +1 for S-I, [0.5] TS-1, [1.0] TS-1 and [1.5] TS-1 respectively). Assuming a number of at least 3 Ti-O-Si bridges per Ti, the [Si(OSi)3(OTi)] species would already increase the Q3 signal by about 50% if they were to resonate at the same frequency. Consequently, the [Si(OSi)3(OTi)] species contribute to the Q4 signal intensity. The mformatlon taken from the Q signal is therefore exclusively related to the number of silanol species found at a nearly constant concentration whatever the titanium loading. The zeolites have been dehydrated during 12 hours at 500~ in vacuum. An analysis by mass spectrometry of the desorption under a flow of pure helium gas between 25 and 500~ (heating rate of 4~ demonstrates that all the physisorbed water is evacuated at 100~ and the water from the condensation of neighbouring silanols is eliminated after two hours at 500~ It can be pointed out that some trace of organic template is also observed and disappears at 400~ The structural defects observed by IH NMR correspond to terminal SiOH and possibly TiOH hydroxyls which are isolated or difficult to eliminate for structural reasons (lattice constraints, bad conformation of neighbouring hydroxyl groups,...). For the pure silicalite S-l, three components can be seen at 1.8 ppm (a), 2.2 ppm (b), and around 3.2 ppm (c) with the ratio 1/3 for each line (fig.5). A strong steaming (175~ in an autoclave in presence of water: conditions of crystallization) was performed leading to the monoclinic MFI structure. This has been confirmed by 29Si MAS NMR: 17 peaks or 9

9

3

9

23 10 TS11.5%

1.0

~TS10.5%

8

Q4 $1

0.8 A [, . ! . . . . [ ~ 5 t~ 0.6 ~ - 9 0 -100

56 .

4

3

2

,

,

-11

i""

r

y_=

L

e-

4

0.4

2

0.2 -

,,,,

0.0 0

!

5

4

3

2

1

chemical shift (ppm/TMS)

Figure 4. IH MAS-NMR spectra of the silicalites S-l, [0.5] TS-1 and [1.5] TS-I" spectra normalized on the weight of sample.

-80

-90

,

I

I

I

-100

-110

-120

-130

chemical shift (ppm/TMS)

Figure 5. 29Si MAS-NMR spectra of the silicalites S- 1 and [1.5] TS- 1.

shoulders with a total intensity of 24 are clearly visible instead of the Q4 broad line. The Q3 signal is absent. The receiver gain used for the IH NMR is 32 times greater than for the silicalite with 8% of defects. Surprisingly, these experiments show that the terminal silanols can resonate within a large window (1.5-4.5 ppm), far from the common value of 2 ppm. After one year of air exposure, the monoclinic silicalite progressively evolves toward the orthorhombic structure, containing 6% of Q3. The corresponding 'H NMR spectra exhibit the same components as the original silicalite with a higher proportion of the species giving rise to line c. It is unprobable that vacancies of silicon could have been created at room temperature in this structure, so the line c can not be assigned to strongly interacting nested OH around vacancies. Furthermore, it is worthwhile to notice that contamination by water lead to a well defined signal at 3.5 ppm, broader than those observed for the monoclinic silicalite and narrower than the line c for the orthorhombic silicalite. Finally, all the defects a, b and c seem to be natural defects of the structure. They probably correspond to terminal silanols in different conformations, arising from non-intact Si-O-Si bridges. The line a corresponds to defects which are easier to eliminate than the defects of type b and c. An ambiguity resides the high chemical shift of the line c (which could be due to strong interacting silanols), the difficulty to eliminate these defects and the fact that they do not seem to originate from silanol vacancies. The lH NMR spectra of the titanium containing silicalites show the same components as the pure silicalite. In this case, the proton can be a silanol proton or a titanol proton. The anatase TiO2 is characterized by two lines, one at 6.4 ppm from Ti-OH-Ti bridges and another

24 one from terminal TiOH groups at 2.3 ppm [16-17]. For anatase free TS-1, we only expect terminal TiOH species in the region of 2 ppm, their effective chemical shift depending on the structural environment as it is the case for the silanol protons. There is indeed no evidence of a signal from anatase clusters at about 6 ppm. A continuous increase of line a intensity with an increasing titanium content in the silicalite is observed (fig.5). Conversely and to a lesser extent, it seems that the line b decreases, while no significant change occurs for line c. The precision on the Q3 defects number does not allow to correlate the proton and Q3 signal intensities with the titanium loading. Moreover, these defects may be differently distributed among the type a, b and c, possibly explaining that the increase of line a intensity is not proportionnal to the Ti loading. Anyhow, the NMR data can be interpreted both ways as the result of formation of titanol species, or increase of type a silanol defects. It seems that no specific structural defects are produced in presence of titanium. DISCUSSION Within the range of 0.005-0.015 Ti/(Si+Ti) ratios, the TS-I investigated here exhibit the expected XRD, IR and UV-visible characteristics of TS-1 materials free of extraframework species. Their XANES spectra similar to the XANES spectrum of the model compound HDPOSST confirm this statement and bring further supports for the Ti isomorphous site occupancy [11]. The average Ti-O-Si angle decreases from 166 ~ in the molecular HDPOSST compound to 161 ~ in the TS-1, getting close to the average Si-O-Si angle of 154 ~ in the titanium free silicalite. This reveals the existence of a local pressure exerted on the Ti site by the lattice. The average Ti-Si distance of 3.38 A is however longer than the average Si-Si distance of 3.10 A and correspond to a local expansion of 37/~3 per Ti. The lattice is obviously relaxed since an actual 25 A 3 lattice expansion is measured. The 29Si NMR already reveals that the S-I framework relaxes from monoclinic to orthorhombic by SiO-Si bond breaking and narrowing of the Si-O-Si bond angle distribution. No further relaxation seems to occur when Ti is incorporated in the lattice according to the constant Q3 concentration. This situation could be explained by a local relaxation on the Ti site implying only a small rearrangement of the lattice. From this model, a [Ti(OSi)3(OH)] local (open) model is more probable because of the broken Ti-O-Si bond than the [Ti(OSi)4] (closed) model. In addition, according to the synthesis route that randomly distribute the titanium in the gel before cristallization, it is more likely that there is no site discrimination for the substitution. Unfortunately, since the coordination number and the DWF are correlated, it is impossible to differentiate between the closed and the open Ti site models from EXAFS. Moreover, the proton NMR does not bring any definitive proof of the existence of TiOH since a new SiOH distribution might as well explain the increase of line a in the spectra (fig.5).

25 CONCLUSION The MS calculations at the Ti K-edge EXAFS and the MAS NMR investigations at the IH nuclei lead to the conclusion that the titanium ions are located in framework substitutional sites of the TS-1 materials. It is more likely that Ti is randomly located in the structure. However, it seems that the relaxation of the lattice might occur by T-O-T bridge breaking preferably involving more Ti sites that are obviously more constrained than silicon sites with a roughly constant lattice defect concentration.

29Si and

1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17.

T. Blasco, M.A. Camblor, A. Corma and J. P6rez-Pariente. J. Am. Chem. Soc. 115 (1993) 11806. P.T. Tanev, M. Chibwe and T.J. Pinnavaia, Nature 368 (1994) 321. J.S. Reddy, A. Dicko and/~. Sayari, Prepr., Am. Chem. Soc., Div. Petrol. Chem. 40 (1995) 225. R. Hutter, D.C. Dutoit, T. Mallat, M. Schneider and A. Baiker, J. Chem. Soc., Chem. Commun. (1995) 163. M.R. Boccuti, K.M. Rao, A. Zecchina, G. Leofanti and G. Petrini, Stud. Surf. Sci. Catal. 48 (1989) 133. P.R. Hari Prasad Rao and A.V. Ramaswamy, J. Chem. Sot., Chem Commun. (1992) 1245. Y. Chapu, A. Tuel, Y. Ben Taarit and C. Naccache, Zeolites 14 (1994) 349. R. Millini, E. Massara Previde, G. Perego and G. Bellussi, J. Catal. 84 (1994) 501. D. Trong On, L. Bonneviot, A. Bittar, A. Sayari and S. Kaliaguine, J. Mol. Catal. 74 (1992) 223. S. Bordiga, S. Coluccia, C. Lamberti, L. Marchese, A. Zecchina, F. Buffa, F. Genoni, G. Leofanti, G. Petrini, G. Vlaic, J. Phys. Chem. 98 (1994) 4125. C. Cartier, C. Lortie, D. Trong On, H. Dexpert and L. Bonneviot, Physica B 208-209 (1995) 653. A. Michalowicz, Logiciels pour la chimie: EXAFS pour le Mac, Soci6t6 frangaise de chimie (1991) 116. P.A. O'Day, J.J. Rehr, S.I. Zabinsky and G.E. Brown Jr., J. Am. Chem. Soc. 116 (1994) 2938. G.E. Maciel, D.W. Sindorf, J. Am. Chem. Sot., 102 (1980) 7606. G. Engelhardt, D. Michel, High-Resolution Solid-State NMR of Silicates and Zeolites, Wiley Interscience, New York (1987). V.M. Mastikhin, A.V. Nosov, React. Kinet. Lett., 46 (1992) 123. S. Haukka, E.L. Lakomaa, A. Root, J. Phys. Chem., 97 (1993) 5085.



A n in situ

13C l~lmS N M R

27

study of toluene alkylation with m e t h a n o l over H -

ZSM-11 Irina I. Ivanova # and Avelino Corma Instituto de Technologia Quimica, UPV-CSIC, Avd. los Naranjos s/n, 46022 Valencia, Spain

13C MAS NMR has been performed in situ to investigate the reaction of toluene with methanol (MeOH)over H-ZSM-11 catalyst. The identification of the main reaction products and intermediates and the quantification of the NMR results provide new information concerning the nature of the methylating agents, the primary products of alkylation and the whole reaction network. During the initial stages of the reaction, methanol partially converts to dimethyl ether (DME).The reaction then proceeds via two parallel reaction pathways. The first pathway which is the major one includes toluene methylation with DME or MeOH to yield o- and p-xylenes in the statistical ratio of two to one. The primarily formed xylenes undergo isomerization and further alkylation to tri- and tetramethylbenzenes. The observation of diphenylmethane-like species as reaction intermediates suggests transalkylation of alkylaromatics. The second reaction pathway includes toluene alkylation with the fragments formed upon methanol conversion to hydrocarbons and thus leads to various alkyltoluenes. The latter, in turn, undergo dealkylation, fragmentation and transalkylation to yield mainly xylenes, ethyltoluenes and polymethylbenzenes.

1. INTRODUCTION Toluene alkylation with methanol over pentasil type catalysts has been thoroughly studied during the last decade [1-12]. Attention was focused mainly on the application of these catalysts in the selective production of p-xylene [1-7]. Mechanistic studies have been limited because the reaction pattern is affected significantly by diffusion [4], and the products observed at the outlet of convensional continuous flow reactors, do not reflect the real kinetic situation on the catalyst. Information on this situation can now be obtained by means of recently developed in situ spectroscopic techniques [8-12]. This study aims to clarify the mechanism of toluene alkylation with methanol using in situ 13C M.AS NMR technique. The attention is focused on the following main aspects: 1) identification of the reaction pathways operating under the NMR/batch conditions; 2) identification of the reaction intermediates which may play the role of alkylating agents; 3) identification of the primary and secondary reaction products.

# On leave from Moscow State University,Russia

28 2. EXPERIMENTAL 2.1. Materials The study was performed on H-ZSM-11 catalyst with Si/AI ratio equal to 23. The catalyst was characterized using XRD, electron microscopy, FTIR, solid state 27A1 and 29Si MAS NMR, sorption of n-hexane and NH3 TPD. Methanol 13C (99.9% - enriched) used as labelled reactant was obtained form ICON Services Iiac. Natural abundance 13C toluene and dimethyl ether were obtained from Aldrich Chemical Company and Linde Technische gase, respectively.

2.2. Controlled-atmosphere 13CMAS NMR measurements Controlled atmosphere NMR experiments were performed in sealed glass NMR cells containing a catalyst and an adsorbate, and fitting precisely into MAS rotors. In a typical experiment, the powdered catalyst was packed into the NMR tube (Wilmad, with constriction), and evacuated to a final pressure of 10 -5 Tort after heating for 10 h at 573 K. Thereafter, the catalyst was cooled down to 298 K and the reactants were subsequently adsorbed. Toluene and methanol were dosed gravimetrically, whereas dimethyl ether was dosed volumetrically. Quantitative adsorption was ensured by cooling the sample to 77 K. After loading the reactants, the NMR cells maintained at 77 K to avoid local overheating were carefully sealed to achieve proper balance and high spinning rates in the MAS NMR probe. 13C MAS NMR measurements were carried out on a Varian VXR-400S WB spectrometer operating at 100.6 MHz. Spinning rates were within 4.5 - 5 kHz. Quantitative conditions were achieved using high-power gated proton-decoupling with suppressed NOE effect (3 ~ts 90 ~ pulse, recycling delay 6 s). Some non-lH-decoupled spectra were recorded and the observed multiplet patterns were analyzed to identify reaction products and intermediates. The assignments of some signals were confirmed by the direct adsorption of model compounds on the same catalyst in a separate set of experiments (e.g. in case of xylene isomers). Variable-temperature (298 - 387 K) and cross-polarization (contact time 5 ms) experiments were performed to distinguish between the species with different mobilities. The reaction was carried out by heating the sealed NMR cells outside of the spectrometer in the temperature range of 373 - 673 K. The 13C MAS NMR measurements were performed at lower temperatures (298 or 387 K) after quenching of the sample cells. After collection of the NMR data, the NMR cells were returned to reaction conditions. Two different reaction protocols were used. In the first one, the temperature was increased in a stepwise manner from 298 to 673 K, the NMR spectra being recorded after heating for 10 min at each temperature step. In the other experiments, the sample was rapidly heated to a final temperature (433 K) and maintained at this temperature for various periods of time (2 - 1400 min). 13C NMR spectra were recorded over the time course of the reaction. The first experimental mode showed more clearly a sequence of products, while the second one facilitated a comparison of reaction kinetics. The list of experiments is reported in Table 1. Quantification of the NMR results requires a number of definitions which are listed below: Conversion of methanol 13C at time t (Xt)was determined as, Xt = (1 - Im,t/Im,0)'100 [%], where Im,t corresponds to the integral intensity of the NMR line assigned to methanol in the NMR spectrum observed after heating for t min, Ir,0 corresponds to the integral intensity of methanol resonance in the initial spectrum. Yield of reaction product p at time t (Yp,t) was determined as methanol 13C conversion into this product,

29

Yp,t = (Ip,t/Im,0)" 100 [%], where Ip,t corresponds to the integral intensity of the product p resonance in the NMR spectrum observed after heating for t min. Table 1 List of experiments Experi- Catalyst

Reactants (molec./u.c.) Toluene MeOH13C

Reaction protocol

DME

Tspectra

ment

(g)

registr. (K)

A

0.02

20

13

-

stepwise temp. rise: 298-673 K

298

B

0.03

14

-

7

stepwise temp. rise 298-483 K

298

C

0.03

17

9

-

progressive increase of reaction

298

D

0.02

17

9

-

progressive increase of reaction

time; Treact.=433 K 387

time; Treact.=433 K

3. RESULTS AND DISCUSSION 3.1. Identification of the reaction products and intermediates. Reaction network 13CMAS NMR spectrum observed after adsorption of reactants over H - Z S M - 1 1 at 293 K (Fig. 1, Sample A) shows 5 narrow resonance lines at ca. 21, 125.5, 128.4, 129.2 and 137.8 ppm corresponding to unlabelled toluene and a broad signal at ca. 51 ppm corresponding to methanol 13C [13]. After heating the NMR cell to 373 K, methanol starts to convert to DME, evidenced by the appearance of the broad NMR line at ca. 60.3 ppm, as was also observed for H - Z S M - 5 zeolites when methanol was the only reactant [14, 15]. The line corresponding to DME reaches its maximum intensity at 413 K and then remains unchanged until alkylation begins. Alkylation starts at 433 K. The primary products are o-xylene and p-xylene confirmed by the resonance lines at 19.3 and 20.7 ppm, respectively. In the aromatic part of the spectra, formation of o-xylene is evidenced by the appearance of weak resonances at 126 and 129.8 ppm; The lines corresponding to p-xylene overlap with toluene resonances. Interesting, that observation of the primary products depends on the reaction protocol used (Table 1). This is discussed further in w 3.3. Further heating to 483 K results in the observation of new resonances at ca. 41.3, 30, 24 and 14 ppm. The latter three lines are related to conversion of methanol or DME to hydrocarbons [14, 15]. The narrow resonance at 24 ppm is assigned to i-butane, the major product of methanol conversion. The broad lines at ca. 30 and 14 ppm correspond to very rigid species as confirmed by experiments with cross-polarization, and may be attributed to overlapping species stemming from various longer chain hydrocarbons. The NMR line at 41.3 ppm can not be associated with the products of methanol conversion and is more likely due to diphenylmethane-like species [13] formed from toluene and/or xylenes. After heating to 523 K methanol consumption is terminated. The lines corresponding to long chain hydrocarbons disappear, indicating their cracking. As a result, weak resonances corresponding to alkyltoluenes are observed, pointing on toluene alkylation with the fragments formed upon cracking [16]. The most intensive resonances are attributed to sec-

30 butyltoluenes (42.1, 31.6, 22.2, 12.3 ppm), n-propyltoluenes (38.3, 24.6, 14 ppm), isopropyltoluenes (34.3, 24.6 ppm) and ethyltoluenes (29, 16 ppm) [13]. Formation of alkyltoluenes is accompanied by growing of the resonances corresponding to i-butane (24 ppm) and propane (16-17 ppm).

Sample A

~.

,~

673

K

523

K. . . . . . . . .

433

K

/

21.2

" " '-

"-

" -'- -- , 4

\

S

___-/AX..__./x~

. . . . . . . . . . .

. . . . .

,,

373 .

'

140

'

"

"

l

130

'

'

"

"

I

120

.

.

S I '

,, , ,

.

K .

.

]9

~

.....

....

/ ~\

_0_' .

I

"

'

'

'

I

60

'

'

'

"

I

"

'

'

'

I

40

'

'

'

'

1

'

'

'

'

I

20

'

'

'

"

9

ppm

Figure 1. 13C MAS NMR spectra observed before and after reaction of methanol 13C and toluene over H-ZSM-11 at progressively increasing temperatures. *- denotes spinning sidebands. Isomerization of xylenes begins in the temperature range of 523 - 573 K as evidenced by the appearance of the growing NMR lines at ca. 21.2, 126.4 and 130 ppm, corresponding to m-xylene [13]. At this reaction step, the lines corresponding to trimethylbenzenes and tetramethylbenzenes mainly 1,2,4-trimethylbenzene, and 1,2,4,5-tetramethylbenzene (growing shoulder at ca. 19 ppm [13]) also begin to appear. Further heating to 623 and 673 K leads to an equilibrium mixture of xylenes (p:o:m=23.7:52.8:23.5), formation of more polymethylbenzenes, and to fragmentation and dealkylation of butyl- and propyltoluenes to give methyl- and ethylsubstituted aromatics. Meanwhile the line corresponding to diphenylmethane disappears. The reaction network presented in Figure 2 rationalizes our experimental observations. At the initial stages of the reaction, methanol readily condenses to give DME and water. Both

31

methanol and DME may be responsible for further reaction proceeding via two parallel pathways: 1) toluene methylation and 2) conversion to hydrocarbons. The first reaction pathway leads to o- and p-xylenes as the primary products. At higher temperatures, o- and p-xylene isomerize to m-xylene, to give an equilibrium mixture of xylenes. The timing of the appearance and disappearance of small amounts of diphenylmethane like species is indicative of their intermediary role in isomerization. This gives further evidence that some xylene formation on zeolites may occur via a bimolecular transalkylation mechanism [17-19]. Xylenes in their turn undergo further alkylation or transalkylation to yield tri- and tetramethylbenzenes.

J

/

IT~ 1~73~K MeOHI ~--"" ( DME (o-,p-XylenesI "',~ .~ 473K...~ t_ ~-..... 523-673K IHydrocarbons ) I Diphenylmethanes)-~~ ~ 523-673K 1 I Tri-andtelramethylbenzenes 1 IEthyl-rpropyl-andbutyltoluenes '~ 673K .~ I Xylenes(eq.mixt.),ethyltoluenesandpoly'nelhylbenzenes I Figure 2. Reaction network.

The second route leads to small amounts of i-butane and some longer chain hydrocarbons, which upon cracking may yield lower alkanes and carbenium ion like species. Toluene acts as a trap for these species, as was observed for benzene in the course of the cracking of propylene oligomers [16]. As a result, butyl-, propyl-, and ethyltoluenes are formed. At higher temperatures, alkyltoluenes undergo dealkylation, fragmentation and transalkylation to yield mainly xylenes, ethyltoluenes and polymethylbenzenes. The final products observed at high temperatures (673 K) in our NMRfoatch experiment correspond to those observed in continuous-flow reactor tests [3], suggesting that the reaction proceeds in a similar way under both experimental conditions. The advantage of the in situ NMR techniques is the possibility to tailor specifically the reaction protocol which enables one to minimize superimposing of the parallel and consecutive reaction steps. As a result, the individual steps of the reaction can be followed in separate experiments and the reaction intermediates can be identified. The optimal conditions for the investigation of the individual reaction steps were determined to be 433-453 K for alkylation, 473-523 K for methanol conversion to hydrocarbons, 523-573 K for isomerization and hydrocabons cracking, and above 573 K for transalkylation dealkylation and fragmentation. The alkylation step will be considered in more detail.

32

3.2. A l k y l a t i n g agents

It is currently accepted that toluene ring-alkylation with methanol proceed via chemisorption of methanol on the acid sites, followed by formation of surface active species such as methoxy groups [21-23] or methoxonium ions [10, 11, 20], which can further react with weakly adsorbed toluene. On the other hand, it is known that DME is always present in abundance on the onset of xylenes formation [9, 10, 21] and there are indications that observation of DME can be associated with ring alkylation [8, 24]. Our results (Figs. 1,2) are also in favour of DME intermediary role in alkylation. Two other experiments were performed to check this hypothesis. In the first one, DME was adsorbed together with toluene over H - Z S M - 1 1 catalyst and the reaction was carried out as in experiment A (Table 1, Sample B). The aliphatic regions of the 13C MAS NMR spectra are presented in Fig. 3. The initial spectrum obtained after adsorption of reactants shows a broad NMR line at ca. 60.3 ppm and narrow line at ca. 21 ppm, assigned previously to DME and methyl group of toluene, respectively. Heating to 453 K results in a partial conversion of DME to p - and o-xylenes confirmed by the resonances at 20.7 and 19.3 ppm, respectively. No resonance line corresponding to MeOH is detected, possibly because the latter is rapidly converted back to DME at 453 K, as was observed in experiment A. It must also be noted that the NMR line corresponding to small amounts of MeOH can be broadened to beyond detection limits. After heating to 483 K DME consumption is terminated. The final products observed at this temperature are p-, o-xylenes and i-butane (24 ppm). This experiment shows clearly that DME can also serve as an alkylating agent under experimental conditions studied. 80 i-butane

xylenes .,,-,

453

,/~

60-

483 K

40

-

K toluene

DME

DME

9

Sample B

20 1:3..

298 K ,

9 '

.... ,.... ,.... ,,,35 ....,....,....,....,....,....,....,....,....,...., 65 55 25 21 ppm ,

b

I 20

~

I 40

'

I 60

'

I 80

'

10

Methanol conversion (%)

,

Figure 3. 13C MAS NMR spectra observed before and after reaction of DME and toluene over H - Z S M - 1 1 .

Figure 4. Variation of products yields as a function of methanol conversion.

To verify further whether DME is responsible for xylenes formation in course of toluene reaction with methanol, a kinetic study was performed. For that, sample C was prepared and the reaction was carried out at 433 K using the second reaction protocol (Table 1). The NMR data were then quantified, as defined in the experimental section, and product yields were plotted versus conversion to give selectivity plots [25], as shown in Fig. 4. The shapes of the selectivity patterns suggest that at low reaction temperature the major reaction pathway leading to xylenes include DME as an intermediate. On the other hand, a

33 small part of the xylene species is formed directly from methanol, as evidenced by non-zero initial selectivity to xylenes determined from the initial slope of the corresponding selectivity plot. Therefore the following reaction scheme operating in coarse of toluene alkylation is concluded: MeOH ~--- ~ DME ,~ To Iu e y "

Xylenes It should be noted, however, that contributions of these pathways to the overall xylene formation may vary with the experimental conditions (batch/flow, pressure, temperature, etc.). NMR results presented in this paper do not permit the identification of the surface active species, formed from DME and MeOH engaged in these reaction pathways and, therefore, hinder the suggestion of a more detailed reaction mechanism. The surface active species will be discussed in a further contribution [26].

3.3. Primary products of aikylation Basic principles for electrophilic substitution in the aromatic ring [27] and the extensive investigations of the electrophilic aromatic substitution in monoalkylbenzenes in homogeneous systems [28, 29] conclude to initial orto/para orientation of methyl substituent in course of toluene alkylation with methanol. When the reaction takes place on zeolite catalysts, isomerization and transalkylation reactions are usually superimposed on the primary alkylation, leading to a mixture of all xylene isomers [1-7]. In addition, product diffusion limitations within zeolite constrains may affect the ratio of xylene isomers [3, 4], preventing therefore the determination of the primary reaction products. The question of the primary products can be best answered by means of in situ techniques, which do not suffer from the above limitations. An in situ 13C MAS NMR specroscopy appears to be a suitable technique, as it allows one to distinguish unambiguously between carbon atoms of methyl groups in xylene isomers (Fig. 1). However, care should be taken while quantification of the corresponding NMR lines, as molecular dynamics of xylene isomers within zeolite pores can profoundly affect spectral features, leading to significant line broadening or even loss of intensity of one or several xylene lines. Thus, our additional experiments with the model mixtures of p- and o-xylenes adsorbed on H-ZSM-11 (not shown) indicate that a part of p-xylene molecules, corresponding to 6 molecules per u.c. of zeolite, may not be observed at room temperature, while at temperature higher than 373 K all the molecules become 'visible'. A detailed discussion of these phenomenon will be presented elsewhere [30]. For the determination of the initial ratio of the primarily formed xylene isomers, NMR experiments were performed at 387 K (Table 1, Sample D). The estimated ratio of o- and pxylenes was of two to one, indicating statistical ortho/para orientation. In conclusion, alkylation of toluene with methanol at low reaction temperatures was found to obey general concepts of electrophilic aromatic substitution [27].

4. CONCLUSIONS 1. The main reaction pathways identified in the course of methanol reaction with toluene over H-ZSM-11 under the NMRAgatch conditions are: 1) toluene methylation and 2)

34 methanol conversion to hydrocarbons followed by toluene alkylation with the fragments subsequently formed. 2 Both, methanol and dimethyl ether formed in abundance at the initial stages of reaction, may play the role of alkylating agents in toluene methylation. Under the experimental conditions studied, dimethyl ether is the main alkylating agent. 3. The primary products of toluene methylation are o- and p-xylenes, formed in the statistical ratio of two to one. ACKNOWLEDGMENTS The authors thank CICYT (project MAT 94-0359-C02-01) for financial support. I.I.I. Ivanova thanks ITQ for research postdoctoral position. REFERENCES .

2. 3. .

5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30.

N.Y. Chen, W.W. Kaeding and F.G. Dwyer, J. Am. Chem. Soc., 101 (1979) 6783. P.B. Weisz, Pure Appl. Chem., 52 (1980) 2091. W.W. Kaeding, C. Chu, L.B. Young, B. Weinstein and S.A. Butter, J. Catal., 67 (1981) 159. L.B. Young, S.A. Butter and W.W. Kaeding, J. Catal., 76 (1982) 418. J. Nunan, J. Cronin and J. Cunningham, J. Catal., 87 (1984) 77. J.-H. Kim, S. Namba and T. Yashima, Zeolites, 11 (1991) 59. K. Beschmann, L. Riekert and U. Muller, J. Catal., 145 (1994) 243. M.D. Sefcik, J. Am. Chem. Soc., 101 (1979) 2164. A. Philippou and M.W. Anderson, J. Am. Chem. Soc., 116 (1994) 5774. G. Mirth and J.A. Lercher, J. Phys. Chem., 95 (1991) 3736. G. Mirth and J.A. Lercher, J. Catal., 132 (1991) 244. G. Mirth and J.A. Lercher, J. Catal., 147 (1994) 199. E. Breitmaier and W. Volter, Carbon-13 NMR Spectroscopy, VCH Verlag, Weinheim, 1987. M.W. Anderson and J. Klinowski, J. Am. Chem. Soc., 112 (1990) 10. E.J. Munson, A.A. Kheir, N.D. Lazo and J.F. Haw, J. Phys. Chem., 96 (1992) 7740. I.I. Ivanova, D. Brunel, J.B. Nagy and E.G. Derouane, J. Mol. Catal., 95 (1995) 243. D.H. Olson and W.O. Haag, ACS Symposium Series, 248 (1984) 275. M. Guisnet and N.S. Gnep, A.S.I. Ser. E. Nato, 80 (1984) 571. A. Corma and E. Sastre, J. Catal., 129 (1991) 177. H. Vinek, M. Derewinski, G. Mirth and J.A. Lercher, Appl. Catal., 68 (1991) 277. T.R. Forester and R.F. Howe, J. Am. Chem. Soc., 109 (1987) 5076. J. Rakoczy and T. Romotowski, Zeolites, 13 (1993) 256. A. Corma, G. Sastre and P. Viruela, Stud. Surf. Sci. Catal., 84 (1994) 2171. E. Dumitriu, V. Hulea and M. Palamaru, Bul. Inst. Politch. Iasi., Sect. 2: Chim. Ing. Chim., 39 (1993) 127. A.N. Ko and B.W. Wojciechowski, Progr. React. Kinet., 12 (1983) 201. I.I. Ivanova and A. Corma, to be published. A. Schriesheim, in "Friedel-Crafts and Related Reactions", G.A. Olah (Ed.), Vol II, Interscience, New York, 1964. R.H. Allen and L.D. Yats, J. Amer. Chem. Soc., 83 (1961) 2799. M.S. Stock and H.C. Brown, Advan. Phys. Org. Chem., 1 (1963) 35. I.I. Ivanova, D. Brunel and A. Corma, to be published.


35

S t r a t e g i e s for Zeolite S y n t h e s i s by D e s i g n M. E. Davis Chemical Engineering, California Institute of Technology, Pasadena, California 91125, United States

The thermodynamics, kinetics and a proposed synthesis mechanism for the crystallization of a prototypical high-silica zeolite, namely ZSM-5, are presented. This information is used to develop strategies for the design of zeolite syntheses and a few of these strategies are outlined. 1. I N T R O D U C T I O N

Zeolites; can they be synthesized by design? Currently, the answer to this question is no [1-3]. However, great strides are being made to reach this lofty goal of zeolite synthesis by design and I outline some of the progress below. Several factors complicate the design of zeolite synthesis. First the molecular-level of understanding of the self-assembly processes occurring during zeolite crystallizations is unknown. Second, since analogies to covalent organic synthesis [4] can not be made, the large number of organic reaction mechanisms provides little help for zeolite synthesis. With covalent organic synthesis, it is possible to systematically construct a final product by using numerous irreversible reaction steps involving different reagents and chemistries with separation between steps. Such is not the case with zeolite synthesis. With zeolites, the separation of desired from unwanted products is almost always difficult. Thus, a zeolite synthesis medium must be designed to spontaneously self-assemble into the correct architecture and atomic ordering in a single step with high yield. In view of this daunting task, the notion that zeolite syntheses can be accomplished by design has been questioned by many. I will demonstrate that this is not the case for zeolite synthesis, and that strategies for zeolite synthesis by design are currently being developed. The routes by which zeolites are crystallized from an amorphous oxide are complex and involve numerous simultaneous and interdependent equilibria and condensation steps. For this reason and others, zeolite syntheses are not well understood except that there does not appear to be a universal mechanism describing all crystallizations [1]. In this paper, emphasis ~s placed on high or pure-silica zeolite syntheses mediated by organic molecules [1,2,5]. These types of syntheses are the most amenable to design as will be described below. A typical synthesis of this kind involves the addition of water-soluble organic species into aqueous alkaline suspensions of silica. The silica dissolves and these inorganic species interact via noncovalent bonding with the organic

36 molecules to ultimately form the final organic-inorganic composite structures. The crystallization is a kinetically controlled process [1,4]. As stated above, a strategy for the design of this type of synthetic process is much different than those used in creating materials with complete covalent bonding. Whitesides et al. [4] have outlined the differences in producing new organic entities by either covalent or noncovalent synthesis. Table 1 is modified after Whitesides et al. and also contains information relevant to zeolite synthesis. The key additional factors with zeolite synthesis in comparison to totally organic preparations are the interactions between the organic and inorganic species and the condensation chemistry of the inorganic oxide. Table 1 Comparison of covalent and noncovalent syntheses (modified from ref. 4) Covalent

Noncovalent (zeolite)

Constituent bond types in the assembly process

covalent

ionic, hydrophobic, hydrogen (between organic and inorganic)

Bond strengths (kcal/mol)

25-200

0.1-5 (between organic and inorganic)

Stability of bonds in product

kinetically stable

kinetically reversible (includes silica chemistry)

Number of interactions in the assembly steps

few

many

Contributions to AG

usually dominated by AH

AH and AS can be comparable

Importance of solvent effects

secondary

primary

Other characteristics

cooperative behavior important

In this paper, I first review the salient results t h a t lead to a basic u n d e r s t a n d i n g of how high-silica zeolites are synthesized. Using this information, I then outline strategies for the synthesis of high-silica zeolites by design.

37 2. T O W A R D S A M E C H A N I S T I C U N D E R S T A N D I N G ZEOLITE SYNTHESIS

OF H I G H - S I L I C A

In the p a s t few y e a r s t h e r e has been significant a d v a n c e m e n t in the u n d e r s t a n d i n g of how the crystallization process of pure silica ZSM-5 (from now on denoted Si-ZSM-5) occurs. Here, I will use this example to highlight the g e r m a n e mechanistic issues for the synthesis of high-silica zeolites in general. In the original report on the synthesis of Si-ZSM-5, a mechanism of assembly was proposed and involved t e t r a p r o p y l a m m o n i u m cations (from now on denoted TPA) preorganizing silicate species to form the zeolite channel intersections [6]. This work and others led to the use of the term "template" to describe the role of the organic species. As pointed out by Davis and Lobo [1], the specificity between organic guest and zeolite host is as yet not sufficient to invoke true templating in the sense t h a t this term is used in biological contexts. R a t h e r we suggest t h a t the organics can act as structure-directing agents, i.e., they can dictate the final outcome of a zeolite synthesis. Figure 1 illustrates a schematic of w h a t I believe is a reasonable proposal for the TPA-mediated synthesis of Si-ZSM-5. The relevant experimental findings t h a t support this mechanistic picture are discussed below.

o

N

N. ) ~

H,

%"

H

,,I 2(" H

2

~ H " ~ I'HI

H H

Si Si

1 s

OI ~O--'Si SiI~"3"s,'Si

,,':4" ,,].w" -'-

H ~O "H 9 SO ~H

hydrophobichydration / .

.

.

.

,

,

y

i!iii:~!'ilil. ~"" :"

,

,

I

',F

J

.

o" ~ hb" T' ~ FFI~I-II ' " " "i h yorop ,,,,; .o lc ,l~-~;,i,i , ' h drat]on ~ ~ ' ' , spheres com~

,

7o~, I

,u

,

,

nucleation

70~.

,

spe~:

i

I

9

f Figure 1. Schematic Diagram of Proposed Mechanism

t agg:eW~ahibo y

38

2.1. Thermodynamics Recently, H e l m k a m p and Davis [2] have calculated e s t i m a t e s of the t h e r m o d y n a m i c p a r a m e t e r s for t h e T P A + F - m e d i a t e d s y n t h e s i s of TPAF-Si-ZSM-5 using experimental AH and AS data for quartz, amorphous silica (glass) and Si-ZSM-5 and the e x p e r i m e n t a l l y m e a s u r e d change in enthalpy for reaction (2) from Patarin et al. [7]. The synthesis of TPAF-Si-ZSM5 from quartz can be considered as follows (T=298K): quartz --~ Si-ZSM-5 AGI=AHI-(TAS)I = (5.5)-(1.4)=4.1 kJ (mol SiO2)-1

(1)

Si-ZSM-5 + TPA+F-(aq) --+ TPAF-Si-ZSM-5 + H20 AG2=AH2-(TAS)2 = (-6.2)-(1.7)=-7.9 kJ (mol SiO2)-1

(2)

quartz + TPA+F-(aq) ----> TPAF-Si-ZSM-5 + H20 AGtot=-3.8 kJ (mol SiO2)-1 When using glass instead of quartz as the starting silica, the glass-to-Si-ZSM-5 reaction yields AG3=-3.0 k J (mol SiO2) -1 and a AGtot=-10.9 k J (mol SiO2)-l. Since it is not possible to transform quartz into Si-ZSM-5, the presence of the organic structure-directing agent renders this synthesis possible (Si-ZSM-5 has been prepared from TPA and quartz [8]). For reaction (2), the favorable enthalpic term (AH_0) and reflects the release of ordered water molecules from the hydrophobic hydration spheres of the TPA molecules as they are encapsulated by silicate species. TPA in Si-ZSM-5 likely represents a nearly optimum case for the organic-inorganic van der Waals interactions in zeolite synthesis. When the enthalpic contribution to AG decreases, i.e., a "looser fit" of the organic guest in the zeolite host, it is possible t h a t the entropic factor arising from the release of ordered w a t e r from the hydrophobic h y d r a t i o n sphere of the organic species m a y play a more i m p o r t a n t role in driving the s y n t h e s i s process. The observation t h a t t e t r a e t h a n o l a m m o n i u m (approximately the same size as TPA but known to not have a hydrophobic hydration sphere due to the strong hydrogen bonding with water) does not produce a zeolite product at reaction conditions where TPA would quickly yield Si-ZSM-5 supports the above statements [9]. Zones and coworkers [10,11] as well as Davis and co-workers [9,12,13] have clearly demonstrated t h a t organic species that serve as structure-directing agents in the synthesis of high-silica zeolites are intermediate in hydrophobicity. That is, m o l e c u l e s t h a t are h y d r o p h i l i c do not elicit i n t e r a c t i o n s w i t h hydrophobically hydrated silica while those that are very hydrophobic tend to aggregate and phase separate in the aqueous synthesis medium [9]. One m u s t exercise c a u t i o n w h e n a t t e m p t i n g to e x t e n d the above thermodynamic analysis of zeolite synthesis involving F- to the more standard

39 ones t h a t use OH- and/or to the addition of heteroatoms (term used here to denote non-silicon t e t r a h e d r a l atoms) such as a l u m i n u m . Pure-silica syntheses with F- can result in TPAF ion pairs being occluded in the ZSM-5 voids w i t h o u t the f o r m a t i o n of siloxy groups. W h e n using O H - a s the mineralizer, only TPA § is e n c l a t h r a t e d and the charge balance is by either Si-O- or A1-O-Si g e n e r a t e d anionic sites [14]. Thus, in order to extend the a f o r e m e n t i o n e d t h e r m o d y n a m i c a n a l y s i s to zeolite s y n t h e s i s in general, coulombic interactions m u s t be accounted for as well. Recently, Koller et al. have provided a good working model for the structure of the Si-O- site in highsilica zeolites [14]. W h a t is now needed to generalize the t h e r m o d y n a m i c analysis given above are the interaction energies between TPA and Si-ZSM-5 and TPA-ZSM-5 (contains A1) analogous to t h a t m e a s u r e d by P a t a r i n et al. [7] for TPAF-Si-ZSM-5.

2.2. Proposed Si-ZSM-5 Synthesis Mechanism A reasonable model for the m e c h a n i s m of the TPA-mediated synthesis of Si-ZSM-5 is schematically illustrated in Figure 1. Initially, the hydrophobic h y d r a t i o n sphere formed around TPA is partially or completely replaced by silicate species when a sufficient amount of soluble silicate species is available [9,12,13]. Favorable van der Waals contacts between the alkyl groups of the organic species and the hydrophobic silicate species likely provide the enthalpic driving force while the release of ordered w a t e r to the bulk aqueous phase provides an additional entropic driving force for the process [2]. It is through t h e s e o r g a n i c - i n o r g a n i c i n t e r a c t i o n s t h a t the geometric c o r r e s p o n d e n c e between the structure-directing agent and the zeolite pore architecture t h a t is the h a l l m a r k of s t r u c t u r e direction arises [1,2,8-13,15,16]. These composite species have been identified by 1H-29Si CP NMR between the protons on the TPA and the encapsulating silicates in an otherwise d e u t e r a t e d synthesis medium [12]. Additionally, silica e n c l a t h r a t e d TPA has been t r a p p e d by silylation techniques [9], and small-angle n e u t r o n s c a t t e r i n g results suggesting the p r o m p t incorporation of TPA molecules into amorphous silicate s t r u c t u r e s when TPA and soluble silicate species are mixed together [17] are consistent with this postulated first step. The availability of soluble silicates influences the r a t e at which these composite species are formed [15]. The use of a monomeric silica source such as t e t r a e t h y l o r t h o s i l i c a t e or the presence of small amounts of alkali-metal cations in the synthesis mixture t h a t facilitate the dissolution of condensed silica sources, e.g., fumed silica, colloidal silica, precipitated silica, leads to an enhanced rate of nucleation [9]. After the formation of the silica enclathrated TPA species, these composites combine to form entities of size ~ 50-70 A. These units have been identified by in situ SAXS [17,18] and cryo-TEM [19] m e a s u r e m e n t s . Additionally, units revealing the ultimate s t r u c t u r e of the final zeolite but in the size range of 80-100 A have been observed by TEM for other zeolites [20,21]. Thus, these data suggest t h a t the nucleation centers for zeolite synthesis should be smaller t h a n 100 A in size. For Si-ZSM-5, I propose t h a t the 50-70 A entities observed by various techniques are in fact the nucleation sites. These sites may form in solution (homogeneous nucleation) and/or on the reactor wall and/or on the surface of inorganic oxide particles t h a t are in the s y n t h e s i s m e d i u m ( h e t e r o g e n e o u s n u c l e a t i o n ) d e p e n d i n g upon the s t a r t i n g r e a g e n t s and synthesis conditions. Additionally, Dokter et al. suggest from in situ SAXS

40 experiments that there is an intermediate step in the formation of the 50-70 ,~ units that involves the condensation of the TPA-silicate species into aggregates by a reaction-limited cluster-cluster mechanism leading to 70 A entities with mass fractal characteristics [18]. This step makes good physical sense and is consistent with other reaction limited growth processes and nucleation activation energies not being related to diffusion but r a t h e r chemical interactions of silicate species (-90-100 kJ/mole) [15]. After the formation of the mass fractal aggregates, there is a re-organization leading to densification of the -- 70 A entities (presumably to minimize surface energies). Both the aggregation and densification processes will involve silicate bond breaking and making steps mediated by the mineralizing agent and/or alkali-metal ions. It is interesting to note t h a t for crystal nucleation from aqueous electrolyte solutions, critical nuclei sizes have been measured, are normally 2-3 unit cells and are specified as microcrsstals with the same unit cells as the final bulk crystals [22]. Thus, a 50-70 A entity of ZSM-5 would be 2-3 unit cells and this size range correlates well with other critical nuclei sizes. Dokter et al. [18] and Regev et al. [19] both suggest that after the formation of the 50-70 ,~ species, ZSM-5 crystal formation occurs via the aggregation of the 50-70 /~ entities. Alternatively, we have proposed that the surfaces of the 50-70 A nuclei serve to "template" the crystal growth process by using the enclathrated TPA species as building units. Thus, the growth occurs in a layer-by-layer fashion [12,13]. Although aggregation of the 50-70 A entities into larger structures has been suggested from SAXS data, this crystal growth mechanism can not account for the layer-by-layer growth that is necessary for the formation of ZSM-5/ZSM-11 intergrowths [23] and could only occur with homogeneous nucleation. Thus, we suggest that crystal growth can occur in a layer-by-layer fashion [12,13]. This proposal for crystal growth may be more generally applicable to zeolite synthesis since layer-by-layer crystal growth is also implied by the intergrowth structures of zeolite beta [24], SSZ-26/SSZ33/CIT-1 [25-28], FAU/EMT [29-30], OFF/ERI [31], MAZ/MOR [31] and other non-zeolitic, crystalline oxides [32]. In fact, we have shown with two distinctly different zeolite systems, i.e., FAU/EMT [30] and SSZ-33/CIT-1 [27,28], that the control of the layer stacking sequence can be performed in a systematic and designed fashion t h r o u g h the purposeful m a n i p u l a t i o n of the organic structure-directing agents. If a layer-by-layer growth mechanism is feasible, then two questions immediately arise: (i) is the layer-by-layer growth truly single layer additions or are the building entities larger? and (ii) do the composite organic-inorganic species function as the building entities or do organic molecules adsorb onto the growing crystal surface from solution and then organize inorganic building units? To answer the first question, the upper bound on the number of layers possible in the growth step will be the minimum size of the observed stacking units in the crystal. Recently, Pan and co-workers have shown via high-resolution TEM images that the minimum size of stacking units can in fact be one layer for the n a t u r a l zeolite tschernichite (analogue of zeolite beta) [33]. Thus, single layer building units are possible. The answer to the second question is less clear. Vaughan has proposed that inorganic cations, e.g., Na § K § play a role in determining the s t r u c t u r e of the inorganic building units while the organics function to influence the orientation in which these structures assemble to create the longrange order [20]. For aluminosilicate materials like FAU/EMT where Si/A1

41 3.5-5.0, this type of mechanism appears plausible [29]. Recently, Stevens and Cox postulated a similar growth mechanism for zeolite beta by proposing that the organic molecules adsorb onto the growing crystal surface [34]. Since two equi-energetic orientations were possible, and since polymorphs A and B are postulated to have the same lattice energy [35], a high degree of faulting was predicted [34] (as is experimentally observed [24,33]). For high- and/or puresilica compositions it is hard to imagine otherwise that at least a portion of the units necessary for crystal growth would have to be silicate enclathrated organic species. These species are necessary for the formation of nuclei and it is most likely that they also participate in the growth process. Twomey et al. have shown that when Si-ZSM-5 is removed from its crystallization medium, nucleation and growth can be restarted [15]. These authors suggest that organic-inorganic entities are continually being produced and upon formation have to choose between nucleating new crystals or incorporating into those that already exist and are growing. From the experimentally measured crystal size distributions, Twomey et al. suggest that incorporation and growth are favored under typical conditions [15]. These results are consistent with silicate enclathrated organic species acting as building entities for crystal growth. However, these building units may be used in combination with "free" organic molecules that adsorb on the crystal surfaces and also participate in concert in the growth process. Evidence to support this conjecture is that ZSM-5 can be synthesized with occluded tetramethylammonium cations (TMA has been shown to not be enclathrated by silicate species [12 ]) if the synthesis mixture contains a sufficient amount of TPA (in addition to TMA) to induce nucleation and sustain crystal growth (Si/TPA=17; TPA/TMA>l)[36]. Thus, if this mechanism is close to the true molecular level assembly process then structure-direction appears to critically influence the ultimate fate of the crystallization at the nucleation stage. 2.3. K i n e t i c s

Since the kinetics of nucleation are typically slower than the kinetics of crystal growth in high- and/or pure-silica syntheses, the kinetics of nucleation play a significant role in determining the phase produced. Goepper et al. have shown that alkali-metal cations influence the rate of nucleation and crystal growth of pure-silica zeolites [37]. Additionally, Burkett and Davis have observed an isotope effect in pure-silica syntheses in the presence and absence of alkali-metal cations [9]. Pure-silica zeolites crystallize faster in H20 than D20 at low alkali-metal ion concentrations. This result supports the premise that O - - H (O--D) bond breaking is also important to zeolite crystallizations. With higher alkali-metal ion concentrations and/or the use of monomeric silica sources, the crystallization proceeds rapidly and the isotope effect becomes less obvious [9]. If the alkali-metal ion concentration is large, these cations compete with the organic species for interactions with the silicate species and ultimately form layered products [38]. This behavior can be rationalized using the bond valence approach of Brown [39]. That is, the best valence match with sodium ions is silicate ions that give layered structures rather than tectosilicates [39]. Since the nucleation step is typically the rate limiting process in highand/or pure-silica zeolite syntheses, the Ostwald ripening model would predict that increasing crystallization rates should be observed with increasing

42 stability of the crystallization material [8]. Recently, Harris and co-workers have shown that rate of crystallization (rate determining step is nucleation in this case) of nonasil does in fact increase as the calculated stabilization energy increases as predicted by the Ostwald ripening model [40]. It is suggested that as the organic molecular size more closely matches the nonasil cage size that there is an increase in van der Waals interactions t h a t lead to the energetic stabilization [40] (recall thermodynamic a r g u m e n t s described previously for TPAF-Si-ZSM-5). The one exception to the trend reported by Harris and coworkers was t h a t the largest organic molecule used (should have had the tightest fit according to the model calculations) had a slow crystallization rate. F u r t h e r work is necessary in order to u n d e r s t a n d this single anomalous experiment. 3. STRATEGIES F O R SYNTHESIS BY D E S I G N

If the thermodynamic, kinetic and mechanistic descriptions of high-silica syntheses of zeolites described above are phenomenologically correct, then this information can be used to develop strategies for zeolite synthesis by design. Two aspects of the design process are described below: (i) the design of pore architectures and (ii) the design of framework atom positioning. In order to develop a zeolite pore architecture by design, the proper choice of organic structure-directing agent and inorganic composition m u s t be made. The organic molecule m u s t have intermediate hydrophobicity and be able to organize w a t e r molecules as has been described by Zones, Davis and coworkers [9-13]. Next, the size and rigidity of the molecule m u s t be controlled. It is clear t h a t as the size of the organic molecule is increased, the specificity for producing zeolite pore architectures also increases [10]. Very recently, Zones has shown t h a t there is a good correlation between structure direction and the rigidity of the organic structure-directing agent as m e a s u r e d by the number of tertiary and quaternary connectivities in the molecule for a series of organics with C/N§ [11]. Thus, size, rigidity and hydrophobicity s i m u l t a n e o u s l y m u s t be correctly specified in order to p r e p a r e a good structure-directing agent by design. For the inorganic compositions, it is clear t h a t for pure-silica preparations a small a m o u n t of alkali-metal cations is necessary for practical kinetics [37]; however, large a m o u n t s will lead to layered compounds [2,38,39]. Since the Si-O-Si angle can vary greatly with little difference in energy [41], it is not surprising t h a t the enthalpies of formation for numerous pure-silica zeolites are all within 2RT (twice the thermal energy) at typical synthesis conditions [42]. Thus, numerous pure-silica structures are energetically feasible and the choice of the product formed is made by the kinetic p a t h w a y specified by the organic s t r u c t u r e - d i r e c t i n g agent. If heteroatoms, e.g., A1, B, Zn, are added into the inorganic mixture, they can strongly influence the s t r u c t u r e s produced even in a m o u n t s as small as SIO2/M203 = 50 [4,10]. We ascribe this behavior at least in part to the loss of flexibility in the Si-O-M angle as compared to the Si-O-Si bonding [4,10,11]. The loss of energetically feasible angles with the addition of heteroatoms creates a reduction in the possible atomic a r r a n g e m e n t s when compared to pure-silica structures. One fairly clear consequence of heteroatom addition is the shift from u n i d i m e n s i o n a l pore s y s t e m s (highly p r o b a b l e w i t h p u r e - s i l i c a

43 compositions) to multidimensional pore architectures [10]. A stellar example of the design of a zeolite pore architectural is by Zones in the synthesis of SSZ-26 [43]. Zones et al. prepared a polycyclic, rigid, diquaternary structure-directing agent to synthesize a multidimensional, large pore zeolite. Using this structure-directing agent and a low-alkali-metal cation c o n t a i n i n g a l u m i n o s i l i c a t e r e a c t i o n composition, a new multidimensional large-pore zeolite was synthesized [25,26,43]. This is the first example of a zeolite with a new pore architecture that was crystallized by the purposeful design of the structure-directing agent. For certain applications of zeolites, e.g., catalysis, it could be important to be able to place particular atoms into specific framework positions by design. Based on the aforementioned mechanism of high-silica zeolite crystallization, Li et al. synthesized ZSM-5 and ZSM-5/ll intergrowths using an organic structure-directing agent that contained a covalently attached silicon atom in order to test whether the silicon atom linked to the organic could be incorporated into a framework position [44]. Li et al. successfully implanted the silicon atom into a framework position as verified by solid-state NMR experiments. Thus, a strategy for designing atomic arrangements is now available in the attachment of target atoms onto specified positions of the organic structure-directing agent [44]. Currently, a completely designed synthesis of a zeolite has not been accomplished. However, the future is bright in this area. The design of structure-directing agents and synthesis compositions is occurring. A more complete design in the sense of the entire framework structure and atom positionings will require f u r t h e r molecular-level insights into the crystallization mechanism. RE~~CES 1. 2. 3. 4. 5. 6. 7. 8. ,

10. 11. 12. 13.

R.F. Lobo and M.E. Davis, Chem. Mater., 4 (1992) 756. M.M. Helmkamp and M.E. Davis, Annu. Rev. Mater. Sci., 25 (1995) 161. M.E. Davis, CHEMTECH, Sept. (1994) 22. G.M. Whitesides, E.E. Simanek, J.P. Mathias, C.T. Seta, D.N. Chin, M. Mammen and D.M. Gordon, Acc. Chem. Res., 28 (1995) 37. B.M. Lok, T.R. Cannan and C.A. Messina, Zeolites, 3 (1983) 282. E.M. Flanigen, J.M. Bennett, R.W. Grose, J.P. Cohen, R.L. Patton, R.M. Kirchner and J.V. Smith, Nature, 271 (1978) 512. J. Patarin, H. Kessler, M. Soulard and J.L. Guth, ACS Symp. Ser., 398 (1989) 221. R.A. van Santen, J. Keijsper, G. Ooms and A.G.T.G. Kortbeek, Stud. Sur. Sci. Catal., 28 (1986) 169. S.L. Burkett and M.E. Davis, Chem. Mater., submitted. R.F. Lobo, S.I. Zones and M.E. Davis, J. Incl. Phenom., in press. S.I. Zones and M.E. Davis, in Synth. Microporous Materials: Zeolites, Clays, Nanocomposites, (eds.) M. Occelli and H. Kessler, Marcel Dekker, NY, in press. S.L. Burkett and M.E. Davis, J. Phys. Chem., 98 (1994) 4647. S.L. Burkett and M.E. Davis, Chem. Mater., in press.

44 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36.

37. 38. 39. 40. 41. 42. 43. 44.

84

H. Koller, R.F. Lobo, S.L. Burkett and M.E. Davis, J. P h y s . Chem., submitted. T.A.M. Twomey, M. Mackay, H.P.C.E. Kuipers and R.W. Thompson, Zeolites, 14 (1994) 162. H. Gies and B. Marler, Zeolites, 12 (1992) 42. L.E. Iton, F. Trouw, T.O. Brun, J.E. Epperson, J.W. White and S.J. Henderson, Langmuir, 8 (1992) 1045. W.H. Dokter, H.F. Van Garderen, T.P.M. Beelen, R.A. van Santen and W. Bras, Angew. Chem. Int. Ed. Engl. 34 (1995) 73. O. Regev, Y. Cohen, E. Kahet and Y. Talmon, Zeolites, 14 (1994) 314. D.E.W. Vaughan, Stud. Sur. Sci. Catal., 65 (1991) 275. M. Tsapatsis, T. Okubo, M. Lovallo, and M.E. Davis, MRS Symp. Ser., in press. I.N. Tang and H.R. Munkelwitz, J. Coll. Inter. Sci., 98 (1984) 430. G.R. Millward, S. Ramdas, J.M. Thomas and M.T. Barlow, J. Chem. Soc., Faraday Trans., 79 (1983) 1075. M.M.J. Treacy and J.M. Newsam, Nature, 332 (1988) 249. R.F. Lobo, M. Pan, I. Chan, H.X. Li, R.C. Medrud, S.I. Zones, P.A. Crozier and M.E. Davis, Science, 262 (1993) 1543. R.F. Lobo, M. Pan, I. Chan, R.C. Medrud, S.I. Zones, P.A. Crozier and M.E. Davis, J. Phys. Chem., 98 (1994) 12040. R.F. Lobo, S.I. Zones and M.E. Davis, Stud. Sur. Sci. Catal., 84 (1994) 461. R.F. Lobo and M.E. Davis, J. Am. Chem. Soc., 117 (1995) 3766. S.L. Burkett and M.E. Davis, Microporous Mater., 1 (1993) 265. J.P. Arhancet and M:E. Davis, Chem. Mater., 3 (1991) 567. D.E.W. Vaughan in Multifunctional Mesoporous Inorganic Solids, (eds.) C.A.C. Sequeira, M.J. Hudson, Elsevier, Amsterdam (1993) 137. C.N.R. Rao and J.M. Thomas, Acc. Chem. Res., 18 (1985) 113. R. Szostak, M. Pan and K.P. Lillerud, J. Phys. Chem., 99 (1995) 2104. A.P. Stevens and P.A. Cox, J. Chem. Soc., Chem. Commun. (1995) 343. S.M. Tomlinson, R.A. Jackson and C.R.A. Catlow, J. Chem. Soc., Chem. Commun. (1990) 813. J.A. Rossin and M.E. Davis, Ind. J. Technol., 25 (1987) 621. M. Goepper, H.X. Li and M.E. Davis, J. Chem. Soc., Chem. Commun. (1992) 1665. S.I. Zones, Microporous Mater., 2 (1994) 281. I.D. Brown, in Structure and Bonding in Crystals, (eds.) M. O'Keefe and A. Navrotsky, Academic Press, NY (1981) 1. T.V. Harris and S.I. Zones, Stud. Sur. Sci. Catal., 84 (1994) 29. M.D. Newton and G.V. Gibbs, Phys. Chem. Miner., 6 (1980) 305. I. Petrovic, A. Navrotsky, M.E. Davis and S.I. Zones, Chem. Mater., 5 (1993) 1805. S.I. Zones, M.N. Olmstead and D.S. Santilli, J. Am. Chem. Soc., 164 (1992) 4195. H.X. Li, M.A. Camblor and M.E. Davis, Microporous Mater., 3 (1994) 117.


45

Use of Modified Zeolites as Reagents Influencing Nucleation in Zeolite Synthesis s. I. Zones and Y. Nakagawa Chevron Research and Technology Company Richmond, California 94802-0627 USA 1. INTRODUCTION New molecular sieve structures continue to be generated and the majority of these reactions rely on the participation of an organic component which is often found within the crystallized inorganic host lattice. There continues to be much debate regarding the role of various types of organic components in the crystallization process. Are there structure-directing roles that these organics fulfill in such areas as forming nuclei, redistributing solubilized silicate populations, aiding in crystal growing, or a number of other critical functions during the synthesis which are difficult to measure experimentally [ 1]? One of the more important results to impact this area of research has been the finding of Navrotsky and co-workers that the novel silicate zeolite structures formed in the presence of organic components do not differ much from each other in enthalpy of formation [2,3]. The enthalpy values are also not far removed from either very dense phases or amorphous silicates. A model of a guest/host complex that resides in an energy well as compared to a non-ordered gel state, while attractive in explaining the important role of the organics in zeolite synthesis, is not well supported by this calorimetry research. In fact, an energy balanced calculation for the formation of TPA*F-/MFI from quartz suggests that the entropic contributions may be the most important factors to consider in the crystallization process [4]. The burden of explaining the effectiveness of the organic reagents then shifts to the issue of zeolite synthesis kinetics and nucleation events. One of our recent studies addressed nucleation selectivity as related to a "goodness-of-fit" for nuclei which resemble the final crystalline product [5,6]. Bell and Chang also developed a model of zeolite crystallization from clathrated organic guests [7]. In this study we wish to continue this discussion on nucleation rate and phase selectivities in the presence of organic structure-directing agents, but with the focus on using highly porous zeolites as inorganic reactants. We have demonstrated that FAU materials are interesting reagents for delivery of A1 to a synthesis gel [8], and similarly, boron-beta zeolite has proven to be an even more impressive reagent for crystallizing novel borosilicate sieves [9]. We have expanded upon this work by modifying some of these reagents and examining the effects of modification of these reagents.

2. EXPERIMENTAL Two types of zeolite synthesis reactions were studied. In the first type, reactions were performed using calcined boron-beta zeolite as borosilicate source [ 10], a quaternary ammonium hydroxide compound, sodium hydroxide, water, and additional boron from sodium borate

46 decahydrate. In the second type of reaction, Cabosil M5 or "N" silicate was mixed into a solution of the two bases mentioned above and aluminum was provided as FAU zeolite [ 11 ], which in most instances was modified by ion-exchange with a transition metal. Care was taken to monitor the pH of the exchange to avoid precipitation from the hydrolysis of the hydrated cation. Table 1 describes the reactant ratios and reaction conditions used in the synthesis studies. The quaternary ammonium compounds used in this study are illustrated in Table 2. The zeolite synthesis reactions were run in Teflon cups for 23 mL Parr stainless steel reactors, and were heated with or without tumbling in Blue M convection ovens [ 12]. Crystallized products were analyzed by X-ray diffraction on a Siemens D-500 instrument. Elemental analyses were determined by ICP methods at Galbraith Laboratories, Knoxville, Tennessee (U.S.A.). Table 1 Reaction Conditions for Molecular Sieve Synthesis Reaction Type Conditions

Borosilicate

High OH A1

Low OH A1

OH/SiO2

0.25

0.90

0.30

Na/Si

0.10

0.90

0.12

Organic/Si

0.15

0.08

0.18

A1/Si

0.00

0.07

0.07

B/Si

0.05-0.10

0.00

0.00

H20/Si

44

32

28

Temp., ~

150

135

160

0

43

43

2-4

2-8

6-9

rpm Time, Days

Table 2 Organocations Used in the Study

Structure

Code

.3)~

T06

+

(CH3)3N~ (CH3)~

T20 ~1CH313

F40

Structure

Code B100

~ 3 I§ CH 3 ~

~

Me 3

M46

~ ~ F~1CH313 B15

B08 N(CH3)3

47

3. RESULTS AND DISCUSSION 3.1. Studies with boron-beta zeolite In our previous study we demonstrated that calcined boron-beta zeolite had limited stability under hydrothermal conditions and as such, could be converted to other zeolites in the presence of certain organocations [9]. Use of this reagent led to rapid nucleation in synthesis reactions, reducing crystallization times to a matter of hours as compared to weeks for synthesis of some zeolites, such as SSZ-24. No amorphous phase intermediate is detected in these conversion reactions. In fact, if the boron-beta alone is heated long enough in water, thereby becoming mostly amorphous, when it is used as a synthesis reagent, the rate and phase selectivity features are lost [13]. The reaction selectivity can sometimes be changed in the presence of additional borate. This demonstrates an interesting synergy between dissolved borate or borosilicates and the high surface area borosilicate zeolite. Is the re-dissolution of the boron-beta an important step in the rapid nucleation? Since borate has high solubility in basic solution, the role of the extra borate added may be to shift the equilibrium toward species which are needed in the synthesis of zeolites such as SSZ33, which appear to require some amount of lattice substitution in order to form [ 13]. We know that the boron-beta zeolite must eventually break down as part of the reaction process because products such as SSZ-24, which contain only 4- or 6-ring subunits, are obtained. The starting boron-beta zeolite contains 5-rings as well, therefore we do not believe that the conversions are a result of "local" transformations involving existing subunits at the surface of the beta-zeolite. Table 3 shows the results of conversion experiments run with and without extra borate. The most spectacular result is the shift in product from SSZ-24 to SSZ-33 upon changing the borate concentration in the presence of organocation T20. SSZ-24 is a large-pore, one dimensional zeolite, whereas SSZ-33 has a multidimensional, 12-10 pore system. F40, which was used to discover SSZ-26 [15] will make SSZ-33 provided there is sufficient boron available in the conversion reaction. Use of boron-beta zeolite is critical in this conversion, as evidenced by the failure of F40 to form SSZ-33 if more conventional, soluble sources of silica and boron are used. The organocation used by Lobo and Davis to discover CIT-1 (B 15) can affect, to some degree, conversion of boron-beta zeolite to CIT- 1, the pure polymorph B of the mixture which constitutes SSZ-33 [16]. B08, the organocation which crystallizes SSZ-33 from soluble systems, does not convert boron-beta into the expected product. We were at first surprised by the lack of reactivity observed in this reaction. Table 3 Conversion of Boron-Beta to Other Organo-Zeolites Organocation

Product

Product (B4Q'/Added)

T06

SSZ-24

SSZ-13

T20

SSZ-24

SSZ-33

F40

SSZ-33

SSZ-33

B08

Beta

Beta

B 100

SSZ-35

SSZ-35

B 15

CIT- 1/Beta

M46

Beta

Beta

48 Recently we showed that the boron-beta product from the reaction with B08 is not strictly "unreacted" [ 13]. We found that the pore system of the product was filled with B08 and that this was true even if the reaction was carried out at 85~ Our interpretation is that the organocation rapidly fills the beta pore system before nucleation to other structures can begin. We believe that the availability of the zeolite reactant surface is an important part of this transformation synthesis. This is similar to our earlier observations that FAU is not a viable reactant in zeolite synthesis if the pore access is blocked [ 11 ]. In Table 2, the entry M46 is similar in size to the adamantyl or tricyclodecane templates, but M46 is structurally much more flexible. As such, it very effectively inhibits conversion of boron-beta to other phases. In fact, using a mixed template system where one component is M46, all conversions were inhibited with the exception of the M46/TPA system, where we did observe formation of ZSM-5. Since calcined boron-beta zeolite has such attractive properties (enhanced rates, novel products) as a reagent for molecular sieve synthesis, we wondered if there were any other zeolites which would also behave in this capacity. Since we believe that high surface area materials are desirable, multidimensional materials would be the first likely candidates. We therefore considered using borosilicates SSZ-33 and B-ZSM-5. Neither of these materials show any reactivity in comparable conversion reactions. Part of the problem may be due to the stability of these zeolites under hydrothermal conditions. Both calcined SSZ-33 and B-ZSM-5 were found to be stable to prolonged aqueous heating, and are extremely steam stable. This again supports the hypothesis that the boron-beta zeolite breaks down under the reaction conditions prior to recrystallization of the new phase. 3.2. Reactions with modified Y zeolite (FAU) A number of years ago Bourgogne, Guth, and Wey demonstrated that Y zeolite can be used as an aluminum source in reactions to crystallize ZSM-5 [ 17]. More recently, we showed that Y zeolite can be used to prepare novel materials such as Al-rich beta which is difficult to make by other routes [ 18]. Other novel materials become accessible by using a very high Si/A1 version ofY [ 19]. Might the selectivity be changed further in our zeolite synthesis reactions if we ion-exchanged the Na in the Y zeolite for metals which would not easily re-exchange back out in basic solution? To test this, we prepared a series of ion-exchanged modified Y zeolites which are shown in Table 4. We envisioned that the rate of delivery of A1 (as AI(OH)4 or A1-O-Si-O units) might be slowed due to the blockage caused at the site by the potentially insoluble hydrolyzed metal. Table 4 Ion-Exchanged FAU as Reagent in Zeolite Synthesis Metal

% Exchange for Na

Solution pH

Surface Area m2_/g

Fe III (b)

80

2.10

346

Cr Ill (b)

75

3.10

532

Co II (c)

64

7.35

674

Ni (b)

69

6.60

664

Ag (b)

65

6.80

Zn (c)

71

6.40

600

Cu II (b)

75

4.38

656

(a) Exchange carried out at 23~ for 16 hours with M+z/A1 set at 5 for z=3 and 7.5 for z=2. (b) Nitrate salt (c) Acetate salt

49 Figure 1 shows a kinetic profile of crystallization of SSZ-13 (CHA) from T06 using the exchanged Y zeolites. We have shown that for this "high OH" type reaction (see Table 1), the pH increase above the buffering capacity of silicate is a good measure of extent of crystallization [8,11 ]. In this reaction the adamantyl organocation is found inside the chabazite cages and typical product SiO2/A1203 values are 10-12. 13.00 -

(~

.-.--~ ............

~_____.

,,~......................

,|

/

/ / ... / / /

I

lZSO-

_-

,' .~.~-

t

' ----'~

.....

~> .....

@--

*

iI / 12.00

-

@

~

@

~

~

To Point

_ _ ' " ' " ~ - ~ " " " ' " ' . ". . .". . ~. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

"@

,

0

25

50

75

100 200

Hours at 135~

Figure 1. Conversion of metal-exchanged Y zeolites (see Table 4) to SSZ-13. Rates are determined by pH increase during crystallization. The hydrothermal crystallization uses organocation T06 from Table 2 and "high OH" conditions in Table 1. Typically, no crystallization begins before about 10 hours and pH values remain at T o, the value before heating. Three groups of kinetic response were observed. Relative to the control (Na-Y), Co-Y or Cu-Y reaction rates were just as fast, indicating no retardation of crystallization. A second group consisting of Ni-Y, Ag-Y, and Zn-Y showed a slower growth rate. Here the pH-profile suggests that nucleation is not retarded, but that the overall crystallization is slower. These metals may affect the availability of A1 in solution. It is unclear as to why the Ag reactions plateau at a lower final pH. Fe and Cr, two trivalent metals, when exchanged into Y zeolite, completely inhibit the reaction! No transformation is observed, even at 200 hours ofreaction time. This may in part be due to the reduced surface areas for these latter two reagents (Table 4). Fe-Y shows the largest decrease of the prepared materials, and the fact that it has also lost 60% of its available micropore volume suggests that Fe has precipitated in the pores. In Table 1, a set of aluminosilicate reactant conditions at "low OH" is described. These conditions were used successfully in the synthesis of SSZ-26 [20], and the products of this type of reaction typically have SiO2/A1203 of 20-30. The reactant SiO2/A1203value is 35 (using Y zeolite with a value of 5 as a reagent), which is still in a range where SSZ-13 can be made from the adamantyl-type cations. We have shown that as the SiO2/A1203 reactant values change using T06, a progression of SSZ-13 (highest [A1]), SSZ-23, and SSZ-24 (lowest [A1]) can be produced [21].

50

In the same study it was pointed out that the use of methylene blue dye appeared to suppress the crystallization of SSZ-13 and allowed us to crystallize pure aluminosilicate SSZ-23. SSZ-23 is a novel (unsolved) structure which we believe has a constrained 10-ring pore system with sizable cavities (to accommodate the adamantyl cations) and higher surface area than other known intermediate pore zeolites (near 500 m2/g). If Co-Y is used under the "low OH" conditions, a Co-containing SSZ-13 is crystallized. However, if Cr-Y is used, Cr-SSZ-23 is crystallized in a region which would normally produce SSZ- 13 ! Based on the results in Table 1, we would not have expected to see any conversion at all. Therefore, in both "high" and"low" OH-type experiments, Cr-Y appears to be an inhibitor for SSZ13 formation (although the mechanism of inhibition might not be the same in the two reaction types). Since we were curious about how much of an inhibitory effect the Cr-Y had in these types of reactions, we ran a series of reactions where varying amounts of both Cr-Y and Co-Y were used as a source of aluminum. The results are graphed in Figure 2 and three product regions emerge. 0.80

0.70

0.60

0.50

| ~" 0.40 0 (.) 0.30

0.20

SSZ-24 Region

|

0.10 , 00

0.10

gi~ I

0.20

0.30

0.40

Cr+31AI

Figure 2. Zeolite synthesis from Co and Cr-exchanged Y zeolite as A1 source and under "low OH" conditions (Table 1). The amount of Co and Cr are inversely adjusted in the experiments and the mixture of zeolite products recovered is plotted. Organocation used was T20 (Table 2). SSZ- 13 (CHA) forms only in the presence of very minor amounts of Cr-Y; chromium seems to be a very effective inhibitor of its crystallization. When Co-Y is the major contributor, surprisingly, SSZ-24 forms. In this instance, the combined Co-Y and Cr-Y do not participate in the reaction; the SSZ-24 is crystallizing from the silica, added alkali and organocation. The metalexchanged Y materials are uninvolved "spectators" and FAU is detected in the isolated solids. (In an analogous experiment, this same behavior was observed for template B08 producing SSZ-31 and unreacted metal-Y). As Cr-Y becomes the major A1 source, SSZ-23 progressively dominates over SSZ-24 as the crystallization product. However, this SSZ-23 zeolite product contains both the chromium and aluminum from the starting reagent.

51 In the "high OH" reaction Cr-Y gave no product. The Cr-Y inhibits SSZ- 13 formation which needs some A1 to nucleate and grow, but the OH/Si level in these experiments is too high to nucleate all-Si SSZ-23 [22]. Under the "low OH" conditions, the SSZ-13 inhibition remains but SSZ-23 crystallizes. The fascinating change here is that the nucleation of SSZ-23 is not likely to be an allSi event favorable at these OH/Si ratios because SSZ-24 would be the more likely crystallization product. Therefore, relatively low amounts of A1 must be available at this early stage and by the end of the reaction, all of the Cr-Y is consumed, (only SSZ-23 is seen in the XRD), and A1 is transferred into the new structure. Catalytic activity is seen for the (chromium) aluminosilicate product, indicating incorporation of A1 into framework positions. Chromium is found associated with the SSZ-23 which is also recovered as a green solid! We do not have evidence at this time indicating the location of the chromium in the product, and there is no reason to believe that it is a framework component or in cationic sites in the organo-zeolite. 4. CONCLUSIONS Highly porous zeolites such as FAU and boron-beta are useful reagents for delivering A1 and B to growing organosilicate lattices during synthesis. The high surface areas of these reagents allow for rapid interaction with the synthesis solution providing for attractive nucleation rates. The way in which components are delivered to the nucleation site (by an unknown mechanism) can also alter the phase selectivity in the experiment. By ion-exchanging the FAU with transition metals whose hydrolysis products have limited solubility in basic solutions, the synthesis selectivity is altered even further. 23 Cr-Y was found to be a good inhibitor for SSZ-13 crystallization in the presence of adamantyl derivatives. The A1 could, however, be eventually transported into a growing SSZ-23 structure. This approach in zeolite synthesis may hold further promise for generating novel structures, compositions, or controlled lattice substitution in zeolite synthesis products. ACKNOWLEDGMENTS

We thank Lun Teh Yuen and Greg S. Lee for synthesis of organocations and help in running the zeolite syntheses. Chevron Research is gratefully thanked for permission to present this work and for continued research support. REFERENCES

1. 2. 3. 4. 5.

6. 7. 8. 9.

M.E. Davis, Chemtech (Sept. 1994), 22. I. Petrovic, A. Navrotsky, S. I. Zones, and M. E. Davis, Chem. Mater. 5 (1993) 1805. N.J. Henson, A. K. Cheetham, and J. D. Gale, Chem. Mater. 6 (1994). M.M. Helmkamp and M. E. Davis, Ann. Rev. Mater. Sci. (1995 in press). T.V. Harris and S. I. Zones, "Zeolites and Related Microporous Materials: State of the Art 1994," Ed. J. Weitkamp, H. G. Karge, H. Pfeifer, W. Holderich, (Elsevier, Amsterdam) (1994) 29. M.E. Davis and S. I. Zones, Third International Symp. on Synthesis of Microporous and Layered Mater. Ed. M. L. Occelli, H. Kessler in press. C.D. Chang and A. T. Bell, Catal. Lett. 8 (1991) 305. S.I. Zones, J. Chem. Soc. Farad. Trans. 86 (1990) 3467. S.I.Zones, L. T. Yuen, Y.Nakagawa, R.A. VanNordstrand, andS.D. Toto, inProc. 9thlnt'l. Zeol. Conf., Ed. R. von Ballmoos, J. B. Higgins, M. M. Treacy (Butterworth-Heinemann, Stoneham, Massachusetts) (1993) 163.

52 10. S.I. Zones, L. T. Yuen, and S. D. Toto, U.S. Patent 5,187,132 (1993). 11. S.I. Zones, J. Chem. Soc. Farad. Trans. 87 (1991) 3709. 12. Y. Nakagawa and S. I. Zones, "Molecular Sieves, Synthesis of Microporous Materials, Vol. 1," Ed. M. L. Occelli, H. E. Robson (Van Nostrand-Reinhold, New York) (1992) 222. 13. S.I. Zones and Y. Nakagawa, Microporous Mater 2 (1994) 543. 14. R.F. Lobo, M. Pan, I. Y. Chan, R. C. Medrud, S. I. Zones, P. A. Crozier, and M. E. Davis, J. Phys. Chem. 98 (1994) 12040. 15. S.I. Zones, M. M. Olmstead, and D. S. Santilli, J. Amer. Chem. Soc. 114 (1992) 4195. 16. R.F. Lobo, S. I. Zones, and M. E. Davis, "Zeolites and RelatedMicroporous Materials: State of the Art 1994," Ed. J. Weitkamp, H. G. Karge, H. Pfeifer, W. Holderich (Elsevier, Amsterdam) (1994) 461. 17. M. Bourgogne, G. L. Guth, and R. Wey, U.S. Patent 4,503,024 (1985). 18. S.I. Zones and Y. Nakagawa, U.S. Patent 5,340,563 (1994). 19. S.I. Zones and Y. Nakagawa, U.S. Patent 5,225,179 (1993). 20. S.I. Zones and D. S. Santilli, Proc. 9th Int'l. Zeol. Conf., Ed. R. von Ballmoos, J. B. Higgins, M. M. Treacy (Butterworth-Heinemann, Stoneham, Massachusetts) (1993) 171. 21. S. I. Zones, R. A. VanNordstrand, D. S. Santilli, D. M. Wilson, L. T. Yuen, and L. D. Scampavia, "Zeolites: Facts, Figures, Future, "Ed. P. A. Jacobs, R. A. vanSanten (Elsevier, Amsterdam) (1989) 299. 22. S.I. Zones, European Patent 213,018 (1987). 23. We have found that the ion-exchange properties of boron-beta zeolite are not as straightforward and hope to report on these findings in the near future.


53

Templating Studies Using 3,7-Diazabicyclo[3.3.1]nonane Derivatives: Discovery of New Large-Pore Zeolite SSZ-35 Y. Nakagawa Chevron Research and Technology Company Richmond, California 94802-0627 USA In our continuing studies of rigid, polycyclic organocations as structure-directing agents for zeolite synthesis, we have focused on the 3,7-diazabicyclo[3.3.1]nonane skeleton. Various derivatives of this family can be used as templates in zeolite synthesis reactions, leading to the formation of large-pore zeolites. For example, the N,N,N'-trimethyl-diazabicyclononane derivative, A, results in crystallization of the novel zeolite, SSZ-35, over a broad range of inorganic reaction conditions. We believe that SSZ-35 is a multi-dimensional zeolite possessing at least one 12-ring channel system. Larger derivatives of A resulted in the crystallization of unidimensional large-pore zeolite, SSZ-24. 1. INTRODUCTION Many of the recent discoveries of new high-silica zeolitic frameworks have been achieved through the use of novel organocationic templates as structure-directing agents. Large-pore, unidimensional zeolite SSZ-24 (AFI-type) was first discovered by Zones using trimethylammoniumsubstituted adamantanes. ~ This was followed by the discovery of large-pore multi-dimensional zeolite SSZ-26, which was made using a rationally designed propellane-based template. 2-4 ZSM18 was first discovered using an unusual tris-pyrrolidinium cation. 5,6 The success of this approach has encouraged us to search for new organic structure-directing agents which we hope will lead to the discovery of new zeolitic structures. We have reported on the use of Diels-Alder chemistry to efficiently prepare a series of structurally related polycyclic organic templates which were effective in crystallizing zeolites ZSM-12, SSZ-31, SSZ-33 and SSZ-37. 7,8 In this study, three large organic templating agents based on a 3,7-diazabicyclo[3.3.1 ]nonane skeleton were synthesized and their effectiveness as structure-directing agents were explored. One derivative in particular, Template A, has led to the discovery of a new multi-dimensional zeolite, S 5 Z - 3 5 . 9

2. EXPERIMENTAL Template Synthesis: Templates A and B were synthesized using the double-Mannich reactions described in references 9 through 11. Methylamine hydrochloride was heated in the presence of glacial acetic acid and paraformaldehyde, and the resultant mixture was added to the appropriately N-substituted 4-piperidone. After purification by extraction and fractional vacuum distillation, the intermediate 9-keto-3,7-diazabicyclononane was subjected to Wolff-Kishner reduction in a mixture of hydrazine, potassium hydroxide and triethylene glycol. The resulting 3,7-diazabicyclononane was treated with methyl iodide to afford the corresponding quaternary ammonium iodide salt, which was recrystallized prior to ion-exchange.

54 Sparteine-derived Template C was prepared by neutralization of commercially available (-)-sparteine sulfate pentahydrate (Aldrich) using aqueous NaOH, followed by extraction into dichloromethane. 12,13 The free diamine was quarternized in chloroform using methyl iodide, and the resulting mono-methylated salt was recrystallized from a mixture of acetone and diethyl ether with a small amount of added methanol. The iodide salts wer~ converted into the hydroxide form by ion-exchange using Bio-Rad AG lX8 (20-50 mesh, hydroxide form) in a minimum amount of water. The molarity of the resulting template solutions was determined by titration using phenolphthalein as an indicator.

Zeolite Synthesis: Zeolite synthesis reactions were performed in 23 mL Teflon cups for Parr 4745 reactors and were heated in Blue M ovens as previously described. 8 Reagents and ratios for the various synthesis reactions will be described in more detail in following sections. Products were analyzed by X-ray diffraction using a Siemens Model D500 diffractometer (using CuK ). SEM micrographs were taken on a Hitachi S-570 instrument. Elemental analyses of solids were performed at Galbraith Laboratories using ICP methods. C and N levels were determined on a Carlo Erba unit.

Template Structures: CH3

CH3~ N + ~ I

~N~ X-

CH 3

N~

CH3 CH3 ~ .N+ / CH 3

A

XB

CH3 CH3

N

X

C

3. RESULTS AND DISCUSSION Our previous studies using rigid, polycyclic organocations derived from Diels-Alder chemistry demonstrated their effectiveness in directing toward crystallization of interesting large-pore molecular sieves. 7,8 These studies also illustrated how other factors greatly affect the observed product selectivity: a given template may give rise to several different products depending on the starting gel composition. The three new templates discussed in this study were designed to capture similar space-filling characteristics of the successful Diels-Alder templates. We chose a large carbon skeleton to ensure that the organocations would not make clathrasil-type structures, and also incorporated two nitrogens in the template, thus keeping the C/N ratio within a reasonable range (< 8), in order to retain its water solubility.

3.1. SSZ-35 Zeolite Synthesis We were gratified to discover that the first template we synthesized, Template A, did in fact lead to the crystallization of a novel new phase, SSZ-35. SSZ-35 can be prepared under a variety of conditions to afford purely siliceous SSZ-35, as well as boro- and aluminosilicate compositions. This is in contrast with the tricyclo-undecene Template (D) from our Diels-Alder series which gave three different zeolites (SSZ-31, SSZ-33 and SSZ-37) from three different gel compositions) The selectivity for making SSZ-35 is quite high for Template A, and the framework structure of SSZ35 apparently tolerates the levels and types of substitution observed. Under more traditional high

55

aluminum and alkali hydroxide conditions (A1/Si = 0.06; NaOH/Si = 0.055), the frequently encountered SSZ- 13 (CHA) phase was obtained. Typical reaction conditions for the synthesis of SSZ-35 are given in Table 1. 9'14 Table 1 Reaction Conditions for Synthesis of SSZ-35 Using Template A

SiO2/W203 oo 50 33 100 50

W

A/Si

M+/Si

OH/Si

B B A1 A1

0.20 0.22 0.33 0.20 0.20

0.05 0.08 0.04 0.05 0.05

0.25 0.25 0.33 0.25 0.25

Temperature 160~ 160~ 160~ 170~ 170~

M=Na, K Silica sources include Cabosil M-5 fumed silica and Ludox AS-30 colloidal silica. Boron sources include Na2B4OT10H20 and boric acid; Reheis F2000 dried aluminum hydroxide gel was used as a source of aluminum. Typical H20/Si ratios were from 35-50. Reaction times ranged from 10-12 days, however, by seeding the reaction mixture with as little as 0.5 wt % SSZ-35 crystals, complete crystallization could be achieved in 5 days. A typical preparation of borosilicate SSZ-35 is described below: Template A (3.2 grams of a 0.70 M solution), 3.3 g water, and 0.45 mL of 1.0 N NaOH were added to a 23-mL Teflon cup of a Parr 4745 reactor. Sodium borate decahydrate (0.045 g) was added and the mixture was stirred until homogeneous. Ludox AS-30 (DuPont, 1.36 g) was added to give a starting gel with a SiOJB203 of 28. The reaction was heated at 160~ (static) for 12 days, after which the settled solid was collected and washed thoroughly. XRD indicated that the product was SSZ-35, and the product was found to have a SiO2]B203 of 53. Characterization The XRD patterns for both as-made and calcined all-silica SSZ-35 are shown in Figure 1. The material is stable to calcination in air to 595~ and the distinctive pattern is not reminiscent of any previously-reported zeolites, therefore we believe that SSZ-35 represents a new framework topology. Unit cell parameters and symmetry for calcined SSZ-35 were determined from synchrotron data: ~5

Symmetry:

triclinic _a = b = c =

13.9719(24) * 18.1747(28) 7.3734(16)

alpha beta gamma

= = =

90.87 ~ 98.93 ~ 90.55 ~

Scanning electron micrographs for all-silica and aluminosilicate SSZ-35 are shown in Figure 2. The difference in crystal size due to introduction of aluminum into the framework is readily apparent.

56

'

"

'

'

'

,

,

,

I

....

9

,

Calcined SSZ-35

rl r

I-- O ~

;~"

uJ t.z

III

-----.._~

o

~_~

7'.2

xb.4

tb.6

~.a

2b.o

TWO -

THETA

2w

2~.4

~.e

3~,.e

36.0

9(DEGREES)

As-made SSZ-35

,i Q.

z

uJ i-.

o

~

q

32B

t

TWO

-

THETA

{OEGREES)

Figure 1. Powder XRD Patterns of As-made and Calcined SSZ-35

360

57

~3CMASNMR spectra confirmed that Template A was still intact within the channels of SSZ35. Weight losses of 13-17 wt % were observed above 250~ using TGA on the as-made product. For one-dimensional large-pore zeolites such as SSZ-24 and ZSM-12, weight losses above 250~ are typically 10-11 wt %; therefore, we anticipated that there would be a large void volume in the SSZ-35 framework. The BET area for SSZ-35 was found to be -- 500 m2/g and adsorption data (shown in Table 2) are comparable to other multi-dimensional frameworks with different dimensionalities. H-SSZ-35 was shown to have a Constraint Index ~6 of < 0.8, indicative of the structure having at least one large-pore channel system. Table 2 Adsorption Data for SSZ-35 (in cc/g; P/P Q = 0.15) Zeolite

N2

n-hexane

cyclohexane

SSZ-35 Beta SSZ-24 ZSM-5 SSZ-37

0.20 0.26 0.12 0.14 0.18

0.15 0.24 0.10 0.14 0.16

0.09 0.25 0.12 0.04 0.14

a) All-Silica SSZ-35

Channel system

3D; 1D; 3D; 2D;

? 12R x 12R 12R 10R x 10R 10R x 12R

b) Aluminosilicate SSZ-35 (SiO2/A1203 = 65)

Figure 2. Scanning Electron Micrographs of All-Silica and Aluminosilicate SSZ-35

58

3.2 Related Templates While the N,N,N'-trimethyl-3,7-diazabicyclo[3.3.1]nonane Template A is very selective for new zeolite SSZ-35, the corresponding N'-isopropyl derivative, Template B, exhibits a high selectivity for the SSZ-24 framework (although in an all-silica reaction, both SSZ-31 and SSZ-24 are seen as products). The even larger methyl sparteine derivative (Template C) also strongly directs toward SSZ-24 crystallization. Representative results using this family of templates are shown in Table 3. Table 3 Results of Various Screening Reactions using 3,7-Diazabicyclononanes

Template A B C

Reaction Type Boron-Containingt

All-Silica SSZ-35 SSZ-31 +/or SSZ-24 SSZ-24

Aluminum-Containingw

SSZ-35 SSZ-24 SSZ-24

SSZ-35 SSZ-24 + Layered SSZ-24

t Starting SiO2/B203 = 33 - 50 (Cabosil/Na2B407 or H3BO3; B-beta zeolite) w Starting SiO2/A1203 > 200 (Cabosil/Reheis F200; Tosoh Y390 high silica FAU) The observation that these larger templates resulted in the formation of a one-dimensional zeolite was somewhat surprising at first. We originally thought that larger templates were better candidates for crystallizing multi-dimensional zeolites. In trying to understand why only Template D gave SSZ-37, while other closely related analogs do not, we have learned from modeling studies that it is extremely important to consider how the templates occupy the channel system. 7 In the case of SSZ-37, two template molecules can effectively fill the interconnecting 12-ring channel system. 17 If the template is any larger, two molecules no longer fit within the defined space, and one template alone cannot provide sufficient stabilization of the channel. Although we have not yet determined the structure of SSZ-35, it is possible that Templates B and C (and even only slightly larger N-Me,N-ethyl derivative E) fail to give SSZ-35 for the same reason. If the length ('T': see Figure 3) of a structure-directing agent is too great, one-dimensional zeolites are formed. We have observed this tendency with several of our"linear" templates where the observed product is ZSM12,18 and Casci has reported similar results using bis-trimethyl ammonium templating agents. ~9 In the latter case, the "default" products are medium-pore zeolites EU-2 (ZSM-48) and ZSM-23, whose channel systems are complementary to the width and height of the bisquat salts. ~ N ~ CH3

//_ H3C ~ CH3 O

I+ CH3 X-

CH3

E

59 1 v

Figure 3. Important Dimensions of Structure-Directing Agents The dimensions of the 3,7-diazabiocyclononane templates (calculated h -- 7.5 ]k; w -- 5.0/~k) 20 are too large to be accommodated in ZSM-12 channels (MTW pore dimensions: 6.2 x 5.5 x 5.2 A), but they fill the SSZ-24 pore system (d = 7.5 A) optimally. In fact, the selectivity for making SSZ24 using Templates B and C is so high that we are able to prepare B-SSZ-24 directly from Cabosil and soluble sources of boron (Na2B407 or H3BO3),rather than having to start with calcined B-beta as described in Zones' earlier work using trimethylammonium adamantane derivatives. ~2,2~ Davis and Lobo observed the same phenomenon, which they attributed to the hydrophobicity of the methyl sparteine template. 13,22,23Although hydrophobicity may be a contributing factor, our belief is that direct synthesis of B-SSZ-24 is possible because the diazabicyclononane derivatives B and C prevent the formation of a cage-type structure such as SSZ-13 (CHA), which is the favored product from the adamantyl derivatives as lattice substitution is introduced. Although the charged nitrogen to carbon ratio of the methyl sparteine template is high, the presence of the second nitrogen in the molecule reduces the overall C/N to 8, thereby making it less hydrophobic. We have also been able to incorporate a small amount of aluminum directly into the SSZ-24 structure using Template C and Cabosil and Reheis F2000 aluminum hydroxide dried gel to give a product with a final SiO2/A1203 of 270. In addition, from a high silica Y-zeolite (Tosoh Y390) as a starting source of both silica and alumina, 24we obtained an SSZ-24 product with a SiOJA120 3 of 400. Although we have demonstrated that it is possible to prepare A1-SSZ-24 directly using Template C, the level of aluminum incorporation which is possible is still quite low. Aluminum exchange of B-SSZ-24 as described by Zones et al. still affords a more highly substituted aluminum-containing product. 25 4. CONCLUSIONS A new zeolite, SSZ-35, was discovered using the 3,7-diazabicyclo[3.3.1 ]nonane Template A. This zeolite can be prepared over a range of compositions and appears to possess a multidimensional channel system having at least one large-pore component. Related Templates B and C do not lead to the crystallization of this new phase, but rather, exhibit a selectivity for SSZ-24. We believe that these results can be explained by the inability of the 12-ring component of the SSZ35 structure to accommodate the larger Templates B and C. These templates therefore induce the formation of one-dimensional, large-pore zeolites such as SSZ-31 or SSZ-24. B-SSZ-24 can be made directly from soluble reagents using these templates, and does not require the use of calcined B-beta zeolite as a starting material. When substituting elements are introduced into the starting gel, products such as CHA often are favored, however, the steric requirements of Templates B and C are such that they cannot be accommodated within the CHA cage. Thus, the direct systhesis of B-and A1-SSZ-24 becomes possible.

60 ACKNOWLEDGMENTS

We thank Gregory S. Lee for synthesizing the organic templates described in this study and for assistance in running the zeolite screening reactions, R. A. Van Nordstrand for preliminary XRD work and Dr. R. C. Medmd, Dr. G. Zhang and G. Mondo in collecting the synchrotron data on SSZ-35. We also acknowledge the work of Dr. Medmd in determining the symmetry and unit cell parameters for both as-made and calcined SSZ-35. Dr. T. V. Harris provided assistance in modeling of the templates and Dr. S. I. Zones was involved in many helpful discussions. We also thank Chevron Research and Technology Company for continuing support of this program and permission to publish this work. REFERENCES

1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21.

22. 23. 24. 25.

S.I. Zones, U.S. Patent No. 4 665 110 (1987). S.I. Zones, U.S. Patent No. 4 493 337 (1990). S.I. Zones, U.S. Patent No. 4 190 006 (1990). S.I. Zones, M. M. Olmstead, D. S. Santilli, J. Am. Chem. Soc., 114 (1992) 4195. S.L. Lawton, W. J. Rohrbaugh, Science, 247 (1990) 1319. J. Ciric, U.S. Patent No. 3 950 496 (1976). Y. Nakagawa, Studies in Surface Science and Catalysis, J. Weitkamp, H. G. Karge, H. Pfeifer, W. Holderich, eds., Part A, Elsevier (1994) 323. Y. Nakagawa, S. I. Zones, "Synthesis of Microporous Materials, Vol. 1," M. L. Occelli and H. Robson, eds., Van Nostrand Reinhold (1992) 222. Y. Nakagawa, U.S. Patent No. 5 316 753 (1994). S.A. Zisman, K. D. Berlin, F. K. Alavi, S. Sangiah, C. R. Clarke, B. J. Scherlag, J. Of Labeled Compounds and Radiopharmaceuticals, 27 (1989) 885. S.A. Zisman, K. D. Berlin, B. J. Scherlag, Org. Prep. and Proc. Int., 22 (199) 255. Y. Nakagawa, U. S. Patent No. 5 271 922 (1993). R.F. Lobo, M. E. Davis, Microporous Materials, 3 (1994) 61. We have since found other templates which make SSZ-35 and have broadened the synthesis ranges described in this paper: Y. Nakagawa, U.S. Patent No. 5 268 161 (1993) Y. Nakagawa, U.S. Patent No. 5 273 736 (1993). Synchrotron data was collected on X7A beam line, National Synchrotron Light Source, Brookhaven National Laboratory, which is supported by the U. S. Department of Energy, Division of Materials Sciences and Division of Chemical Sciences. V.J. Frilette, W. O. Haag, R. M. Lago, J. Catal., 67 (1981) 218. J.L. Casci, M. D. Shannon, P. A. Cox, S. J. Andrews, "Synthesis of Microporous Materials, Vol. 1," M. L. Occelli, H. Robson, eds., Van Nostrand Reinhold (1992) 359. S.I. Zones, Y. Nakagawa, J. W. Rosenthal, "Zeolites (Japan)," 11 (1994) 81. J.L. Casci, New Developments in Zeolite Science and Technology, Y. Murakami, A. Iijima and J. W. Ward, eds., Elsevier (1986) 215. Molecular dimensions were calculated using "Smallest cylinder program" which was developed at Chevron Research and Technology Co. by Paul Merz. S.I. Zones, L-T. Yuen, Y. Nakagawa, R. A. Van Nordstrand, S. D. Toto, "Proceedings from the Ninth International Zeolite Conference," R. Von Ballmoos, J. B. Higgins and M. M. J. Treacy, eds., Vol. 1, Van Nostrand Reinhold (1993)163. R. de Ruiter, J. C. Jansen, H. Van Bekkum, Zeolites, 12 (1992) 56. Although Davis and Lobo did not observe conversion of calcined B-beta zeolite to B-SSZ24 using Template C, we did obtain B-SSZ-24 from B-beta after 7 days at 150~ S.I. Zones, Y. Nakagawa, U. S. Patent No. 5 225 179 (1993). R.A. Van Nordstrand, D. S. Santilli, S. I. Zones, "Synthesis of Microporous Materials, Vol. 1," M. L. Occelli and H. Robson, eds., Van Nostrand Reinhold (1992) 373.


61

S y n t h e s i s of Z S M - 4 8 T y p e Zeolite in P r e s e n c e of Li, Na, K , R b and Cs Cations G. Giordano 1, A. Katovic 1, A. Fonseca 2 and J.B. Nagy 2 1Dipartimento di Ingegneria Chimica e dei Materiali, Universit~t della Calabria, 1-87030 RENDE (CS), Italy 2Laboratoire de Catalyse, Facult6s Universitaires Notre-Dame de la Paix, 61 rue de Bruxelles, B-5000 NAMUR, Belgium

SUMMARY

This study investigates the formation of ZSM-48 zeolite starting from a hydrogel with or without A1 as a function of the alkali cations (Li, Na, K, Rb and Cs). Crystallization kinetics data, chemical composition, NMR characterization, morphology and crystal size are presented in order to give additional information about the synthesis of this catalyst precursor. INTRODUCTION Recently, the preparation of ZSM-48 type zeolite has shown a new interest due to its catalytic properties as additive in FCC catalyst. The presence of ZSM-48 in FCC catalyst increases both conversion and selectivity in C3-C 4 products [1]. The ZSM-48 possesses a 10MR unidimensional channel system with pore opening 5.3 x 5.6 A and it can be synthesized starting from silica or alumino-silicate gels in presence of various organic compounds. The nature of inorganic and organic cations affects the formation of ZSM-48 and also influences morphology, crystal size and the amount of structural defects. The aim of this work is to evaluate the effect that the alkaline cations (Li, Na, K, Rb and Cs) have on the synthesis of ZSM-48 zeolite in a system that contains hexamethonium ions (HM++). ~ The data about crystallization kinetics, morphology, crystal size, chemical analysis and NMR characterization are discussed with the goal to optimize the synthesis of ZSM-48 in HM systems. EXPERIMENTAL The synthesis was carried out at 200~ under static condition and autogenous pressure in Teflon-lined Morey-type autoclaves starting from the following hydrogel: 5 M20 - 2.5 HMBr 2 - x A1203 - 60 S i t 2 - 3000 H20 where HMBr stands for hexamethonium bromide, x = 0 or 0.5 and M = Li, Na, K, Rb or Cs. The samples were filtered, washed, dried and characterized by usual procedures (X-ray, SEM, EDX, A.A. spectroscopy and NMR).

62 RESULTS AND DISCUSSION

First of all, it can be observed that in absence of A1 the systems with Li, Na and K show similar crystallization kinetics. After 24 hours of reaction the crystallinity is higher than 90%. In presence of Rb ions a reaction time of about 48 hours is required for a complete crystallization. The Cs ions disfavour the formation of ZSM-48, since the first crystals are detected, by X-ray, only after 30 hours while with the other cations the crystallization is almost complete at that time. Moreover, the maximum of crystallinity (60%) is obtained after 50 hours of reaction, and at longer times co-crystallization with cristobalite occurs. The presence of A1 in the starting hydrogel leads to a longer crystallization time. Similar behaviour is observed for other zeolites such as the MFI type. In this case all the inorganic cations, except Cs ions, show similar behavior. In fact the nucleation time is identical for all four samples and after 48 hours of reaction all samples show more than 85% crystallinity. Also in this case, the presence of Cs ions disfavours the ZSM-48 formation. Indeed, after 4 days of reaction the crystallinity achieved is only 35%. However, in this case the formation of cristobalite is not observed for longer reaction times. Chemical analyses show that the amount of HM ions is close to one per unit cell for all examined samples. In the synthesis in absence of A1, only traces of alkaline cations are detected. This suggests that the synthesis of the siliceous ZSM-48 can be obtained also in absence of inorganic cations, and hence only a component is required that solubilizes and mobilizes the siliceous species. On the other hand, the amount of organic cations detected is justified by the large amount of structural defect observed by 29Si-NMR analysis. In the samples synthesized in presence of A1, both inorganic and organic cations are found in the final products. The amount of inorganic cation observed is not sufficient to act as counteractions of A1 species, and this suggests that also the hexamethonium ions play a role in stabilizing the negative charges. Probably the organic cations stabilize the framework by a pore filling action and act as neutralizing agent of the negative charges and of the defect groups. The number of A1 per unit cell found in the final products is always less than 1, and this is in agreement with the framework suggested for the ZSM-48 zeolite. In fact when the amount of A1 in the initial reaction mixture increases, the system derives toward the formation of EUO type zeolite. The 27A1-NMR data show that the aluminium is incorporated in the ZSM-48 zeolite mainly in the tetrahedral coordination. The spectra of 29Si-NMR show that the amount of SiOM (M = organic and/or alkali cation) defect groups is higher in absence than in presence of A1. Indeed, in presence of Li, Na, K and Rb, some 18 SiOM/u.c. are detected without A1, and ca 12 SiOM/u.c. with A1 incorporated in the structure. These values are higher than those reported previously in the presence of Na and a high amount of HM §247ions [2]. In presence of Cs, the amount of SiOM/u.c. is smaller (13/u.c.) and the 29Si-NMR spectrum shows a higher resolution. REFERENCES 1. F. Di Renzo, G. Giordano, F. Fajula, P. Schultz and D. Anglerot, French Pat. 2.698,863 (1994). 2. G. Giordano, N. Dewaele, Z. Gabelica, J.B. Nagy, A. Nastro, R. Aiello and E.G. Derouane, Stud. Surf. Sci. Catal., 69 (1991) 157.


63

O u t - o f - p l a n e B e n d i n g Vibrations of Bridging OH G r o u p s in Zeolites: A New Characteristic of the G e o m e t r y and Acidity of a B r o e n s t e d Site L.M. Kustov*, E. Loeffler**, V.L. Zholobenko*, and V.B. Kazansky* *N.D. Zelinsky Institute of Organic Chemistry, Russian Academy of Sciences, Moscow, Leninsky prosp. 47, 117334 Russia **Adlershofer Umweldschudztechnik, Berlin, Rudower Chaussee 5, D-0-1199 Germany

Abstract Out-of-plane bending vibrations of bridging OH groups in zeolites were studied by diffuse-reflectance IR spectroscopy in the region of the combination bands of the out-of-plane bending and fundamental stretching vibrations (3800 - 4200 cm -1). Similar to the in-plane bending vibrations, the out-of-plane bending modes were found to be a very sensitive characteristic of the bridging OH groups in zeolites. These frequencies increase with increasing acidity of the OH groups and with formation of H-complexes with adsorbed molecules.

1. INTRODUCTION The observation of the bending, torsional, and other low-frequency vibrations of the structural hydroxyl groups in oxide adsorbents and catalysts encounters a number of problems because of the unfavorable background due to the framework vibrations in the region of 200 - 1000 cm-1, as well as due to the low concentration of OH groups. There are however two possible ways to overcome this difficulty: (i) the use of modern highly sensitive spectroscopic methods for studying OH groups, for instance, Fouriertransform IR, diffuse-reflectance IR, inelastic neutron scattering, etc. and (ii) the analysis of the combination bands. In our previous studies [1 - 3], we have analyzed the combination modes (v + (3) of the fundamental stretching and in-plane bending vibrations in the DRIR spectra of zeolites and other oxide materials using highly sensitive diffuse-reflectance IR spectroscopy. The bending vibrations of the bridging

64 OH groups were shown to be much more sensitive to the structural environment of hydroxyls as compared to the stretching vibrations. Thus, two types of the isolated bridging hydroxyl groups characterized by the frequency of the fundamental stretching vibration at 3660 cm -1 have been resolved in the region of the combination bands of the stretching and in-plane bending vibrations. The range of the stretching frequencies for various types of OH groups is quite narrow (3550 - 3800 cm -1), i.e., about 250 cm -1, whereas the corresponding range of the in-plane frequencies is much wider and extends from 600 cm-1 (for MeOH groups connected with rare-earth cations) to 1640 cm -1 (for water molecule). The in-plane bending frequencies were also proposed as an additional measure of the acidic properties of these OH groups. In studying structural OH groups in SAPOs [4], we succeeded in observing a combination band of the fundamental stretching and out-of-plane bending vibrations of the bridging OH groups. The combination band for SAPO-34 was found at 3940 cm -1, which was attributed to the structural OH groups characterized by the fundamental stretching frequency at 3600 cm -1 Thus, the out-of-plane bending frequency calculated as the difference of the two observed frequencies is equal to 340 cm-1. The quantum-chemical calculations [5, 6] agree fairly well with the experimental data and show that the bending modes are more sensitive to changes in the local structure at the Broensted site compared to the stretching modes. The calculations allowed the authors to estimate the outof-plane bending frequency for the bridging OH groups in zeolites (y - 400 cm-1). Recently, the fundamental in-plane bending and out-of-plane bending modes for the bridging OH groups in zeolites have been observed using inelastic neutron scattering spectroscopy at 1090 cm-1 and 420 cm-1, respectively [ 7 - 11], in perfect agreement with the theoretical prediction and our studies of the combination bands. In the present paper, we attempted to use the similar approach to investigate the combination bands of the stretching and out-of-plane bending vibrations of structural OH groups in different types of zeolites.

2. E X P E R I M E N T A L H-forms of the zeolites X (Si/AI = 1.25), Y (Si/AI = 2.35), erionite (Si/AI 3.5), mordenite (Si/AI = 21) and ZSM-5 (Si/AI = 21) were prepared by calcination of ammonium forms in a vacuum at 640 - 770 K depending on the thermal stability of the zeolites. The thermal-vacuum pretreatment at the final temperature was performed for 8 h, The rate of the temperature increase was 2 - 5 K/min. The diffuse-reflectance IR spectra were measured using a Perkin-Elmer 580B spectrophotometer as described in [1, 2]. Adsorption of N20, CO 2, n-

65 C8F18, and CF3H was carried out at room temperature pressures of 20 - 40 torr.

and adsorbate

3. R E S U L T S A N D D I S C U S S I O N DRIR spectra of the HNaY, HNaX, and HNa-ERI zeolites under study in the region of 3200 - 4000 cm -1 are presented in Fig. 1. The absorption bands in the region of 3550 - 3750 cm -1 have been reliably assigned to the fundamental stretching vibrations of the bridging hydroxyl groups [1, 2, 12]. However, in some cases, additional low-intensity lines at 3900 - 3950 cm-1 are also observed. These bands are present in the spectra of H-forms of X and Y zeolites and erionite, but they are absent in the spectra of the HZSM-5 zeolite and H-mordenite. Although X, Y, and erionite zeolites have so little in common as concern their structure, stability, and chemical composition, it is still possible to find at least one common feature distinguishing them from pentasils. Indeed, the socalled LF-bands in the low-frequency region (3520 - 3580 cm-1) are observed in the IR spectra of X, Y, and erionite zeolites. They have been attributed to the bridging OH groups, which are hydrogen-bonded to the neighboring oxygen atoms of the lattice in the double 6-membered rings (D6R structural elements) [1, 12]. For the pentasil-type zeolites, such as mordenite and ZSM5 zeolite, this type of H-bonding is not realized because of the absence of the D6R units in the structures of these zeolites. Hence, an assumption could be drawn that the bands near 3900 cm-1 are observed only in those cases when the bands at 3520 - 3580 cm-1 are present in the IR spectra. To confirm this hypothesis, HX and HY zeolites with different ion-exchange degree of Na + for NH4 -I- (o~) were investigated. As is seen in Fig. l b, for faujasites with 0c < 50%, unlike the samples with 0c > 70%, the intensity of the band at 3550 - 3580 cm-1 is much lower than that of the band at 3650 cm-1 Accordingly, the intensity of the high-frequency band at 3900 cm-1 decreases in a similar way. Thus, these data show that the band at 3900 - 3950 cm-1 is somehow connected with the bridging OH groups that are H-bonded to the oxygen atoms of the lattice in the D6R structural units (vOH - 3520 - 3580 cm-1). The high values of the frequencies corresponding to the observed maxima (-3900 cm-1), which are placed far beyond the region of the stretching vibrations of OH groups and the significant halfwidths of the bands (AH1/2=80 - 100 cm -1) allow us to assign them to a combination v + ~/, i.e., to a combination of the stretching vibration ( v ) o f the H-bonded acidic hydroxyl groups with a low-frequency vibration ~f. The y values derived by subtraction of the stretching frequency from the combination are presented in Table 1. They are ranged from 325 to 380 cm-1 depending on the zeolite. In our opinion,

66 3610

3640

3560 1 1

3555 ~~ 3660

,/ _,j! 3, 4 A ), and that, in this region, they fall very close to the q+Y2 curves. The X+-Y2 interaction is thus very well represented by the point charge model for distances sufficiently larger than the equilibrium distance. It appears that the reason why the Li+Y2 systems are thermodynamically more stable than the corresponding Na+Y2 systems is that the repulsive short range forces become significant at much shorter distances for Li + than for Na +, introducing larger electrostatic and induction energies in the total interaction energy. This explains the fact that, replacing Na + by Li + increases the binding energy with N2 (3.7 kcalmole -1) more than the binding energy with 0 2 (2.5 kcalmole-1), favoring the N2-O2 separation.

10

0

_

-$, ~-10 N

v

At p

v

LLI

.

o

/

o

/

W

r I

-20

o a

/

-10

9c

o a

d

A c

All

9

9 b

-30

I

1,5

'

-15

I

2,5 3,5 r(X-Y) (angs)

Figure 3. Potential energy curves of Na+N2 (a), Li+N2 (b) and q+N2 (c).

4,5

'

1,5

!

'

b !

2,5 3,5 r(X-Y) (angs)

'

4,5

Figure 4. Potential energy curves of Na+O2 (a), Li+O2 (b) and q+O2 (c).

113 Although N2 and 0 2 adsorption energies seem to differ essentially through the quadrupole dependent term, it is worth to investigate the relative weights of the electrostatic and induction contributions to their binding energies. With this purpose, we have considered the interaction of N2 and 0 2 with a Li+ dimer. The Y2 molecule can be situated either in a bridged geometry with each end bonded to a Li cation or in a top position, with one end bonded to a single cation. Several bridged structures have been optimized, for different Li-Li fixed distances. Their calculated geometries and binding energies (not corrected for BSSE) are presented in Table 1. These results show that the optimum Li-Li distance is 6.0 A for 0 2 and 5.5 or 6.0 ,a, for N2. In fact, these values correspond to distances between two cations located at site II and site III in the faujasite framework.

Table 1 Binding energies (B.E.) of N2 and 0 2 in bridged structures between two Li cations (distances in A, energies in kcal/mole) r(Li-Li)

r(Li-N)

B.E.(N2)

r(Li-O)

B.E.(O2)

5.3 5.5 6.0 6.5

2.10 2.20 2.45 2.70

12.3 13.2 13.1 11.5

2.05 2.15 2.40 2.65

2.8 4.0 5.1 4.8

The presence of the second Li cation, in these symmetrical structures, has a stabilizing effect, since it doubles the quadrupole energy term, but, at the same time, it has a destabilizing effect because the induced dipole created by symmetrically placed charges vanishes the induction term. A Y2 molecule on top on a Li + dimer has a preferred bent structure with a lowest energy for a (LiLlY) angle of around 160 ~ For a Li + - Li + distance of 6A, the corresponding binding energies are 15.2 and 10.9 kcal/mole for N2 and 02, respectively. The analysis of the binding energies with respect to different bending angles shows that these energy values are almost constant if the Y2 molecules approach the cation within a cone defined by a (LiLiY) angle of 110 ~ The binding energies will then differ from the above values by less than 0.5 kcal/mole. Comparison with the bridged structures shows that the top adsorption is more favored, especially for 0 2 9Larger binding energies for top structures means that, even for N2, the interaction energy with one Li cation includes a preponderant induction contribution (which is cancelled in the bridged geometries). Moreover, these results suggest that bridged geometries are probably not realistic structures for the N2 or 0 2 adsorption, except if the zeolite environment modifies differently the electrostatic and induction contributions.

114 3.3. I n t e r a c t i o n of N 2 and 02 with zeolite models 3.3.1 m ( L i +) a n d m ( N a +) Optimization of the structures of m(Li +) and m(Na +) clusters has led both Li + and Na + to form two equivalent XO bonds. The cation-(zeolite)O interaction is stronger in m(Li +) than in m(Na+), since the Li-O distance (1.88 A) is shorter than the Na-O distance (2.18 A). The potential energy curves (not corrected for BSSE) for the m(X+)Y2 clusters are depicted in Figure 5. These curves show that the binding energies for N2 are larger than those for 02 and that Li+-Y2 is still preferred to Na+-Y2, just as was found for q+-Y2 and X+-Y2 models. The binding energies corrected for BSSE indicate the same trend (3.7, 1.2, 2.6 and 0.9 kcal/mole for the Li+N2, Li+O2, Na+N2 and Na+O2 systems, respectively). Comparison with X+Y2 models shows that adsorption energies are smaller in the presence of the zeolite cluster but display the same relative ordering. The decrease of the calculated binding energies has to be related to the decreased positive charge attributed to the X cations, which is a consequence of charge transfer interactions between the aluminosilicate skeleton and its counterions. The net Mulliken charges are 0.79 for m(Li+)Y2 and 0.86 for m(Na+)Y2 , instead of a value of 0.95 for both Li+Y2 and Na+Y2 . These results indicate that the N 2 or 0 2 adsorption is, to first order, controlled by the Y2-cation interaction, whereas the zeolite framework plays a perturbative role, which can however modify the adsorption strength of these molecules.

0

-2

-2

E _~-4

E ---4

0

0

o

o .ar

v

v

LIJ

UJ

-6

-6 o -8

I

b I

a

v

c

e

b

v

d

v d I

2,5 3 3,5 r(X-Y) (angs) Figure 5. Potential energy curves of m(Li+)N2 (a), m(Li+)O2 (b),

m(Na+)N2 (c) and

o

m(Na+)O2 (d).

-8

I

2,5

I

I

3 3,5 r(Li-Y) (angs)

Figure 6. Potential energy curves of embedded Li+O2 a (site II), b (site [II) and Li+N2 c (site II), d (site III).

115

3.3.2 E m b e d d e d Li + and Na + m o d e l s The role of the small pentameric model of zeolite in N 2 or 02 adsorption can be depicted as a screening of the binding interaction with respect to free cations. This picture is however incomplete due to the small size of the cluster. Indeed, the long-range interactions between N 2 or 02 and the charge distribution of the zeolite framework are not negligible, since we have seen above that the electrostatic contributions decrease as 1/R 3. In order to take them into account in a further more realistic study, we have thus investigated how the previous X+Y2 descriptions are modified when they are placed in a representation where all surrounding atoms are replaced by appropriate point charges. Moreover, another purpose for this study was to delineate if this environment could modify specifically the adsorption properties of some cationic sites, with respect to N 2 or 02 molecules. In a very simple first attempt, conventional charges used in Molecular Mechanics simulations have been used for the zeolite atoms (values indicated in 3.2). Li cations have thus been placed at the coordinates of two neighboring cationic sites identified in faujasite as site II and site III, accessible to incoming gaseous molecules. These sites, located in a supercage, have been approached by N 2 or 02 molecules following a direction of lowest energy, roughly perpendicular to the zeolite framework around the cation, i.e. one six-membered ring for site II and two four-membered rings for site III. The binding energy curves for N 2 and 02 as a function of their distances from the Li cations in site II and site III are shown in Figure 6 (no BSSE corrections). The four curves have a minimum at around 2.3 A, which is longer than the equilibrium distances found for isolated cations and slightly shorter than the Li-Y distances evaluated for the m(Li+)Y2 models. The presence of the point charge network does not change the trend that N 2 is more strongly bonded to the cation than 02. Moreover, the binding energy difference between these two molecules, evaluated at 3.2 (site II) or 3.5 kcal/mole (site III) is similar to the value obtained from the pentameric model clusters (3.2 kcal/mole without BSSE correction). This difference was calculated at 4.4 for an isolated Li cation. If the presence of the point charge environment does not modify the relative ordering of N 2 and 02 adsorption energies, it clearly induces a difference between the adsorption property of cations in site II and site III. Indeed, the energy curves on Figure 6 show that site III is more favorable for adsorption than site II. Further calculations including larger clusters as well as more investigations on possible point charge sets are needed to confirm these preliminary results. 4. CONCLUSIONS These quantum mechanical calculations on small model clusters inluding Li and Na cations have shown that the different N2 and 0 2 adsorption properties in zeolites can be explained by simple arguments. Indeed, they have demonstrated

116 that a classical description involving electrostatic and induction energies, which depend on the quadrupole moment and dipole polarisability of these molecules repectively, is adequate to explain the basic reason for a stronger N2 adsorption. Moreover, the calculations show that a small cation like Li + behaves as a better attractor than a larger cation like Na +, the presence of the core electrons being the factor which limits the stabilizing contribution of the electrostatic and induction terms in the interaction energy. This conclusion justifies thus the use of classical simulations based on energy expressions including Lennard-Jones terms and also electrostatic (quadrupole) and induction contributions. The study of models including the zeolite as a cluster of atoms or as a network of point charges has shown that the zeolite framework has a perturbative role, screening the binding of both N2 and 0 2 , while leaving the main trends valid. A proper point charge embedding may help to rationalize the influence of the structure of the zeolite on the adsorption properties of specific sites.

REFERENCES

1. D.W. Breck, Zeolite Molecular Sieves, R.E. Krieger Publishing Company, Malabar (1984). 2. V.N. Choudary, R.V. Jasra, T.S.G. Bhat, Ind. Eng. Chem. Res., 32 (1993) 548. 3. C.G. Coe, J.F. Kirner, R. Pierantozzi, T.R. White, U.S. Patent No 5 152 813 (1992) and references therein. 4. D.H. Olson, J. Phys. Chem., 74 (1970) 2758. 5. I. P~pai, A. Goursot, F. Fajula, J. Weber, J. Phys. Chem., 98 (1994) 4654. 6. D.R. Salahub, R. Fournier, P. Mlynarski, I. P~ipai, A. St-Amant, J. Ushio, Density Functional Methods in Chemistry, J. Labanowski, J. Andzelm Eds., Springer-Verlag, New York, (1991) 77. 7. H. Sambe, R.H. Felton, J. Chem. Phys., 62 (1975) 1122. 8. B.I. Dunlap, J.W.D. Conolly, J.R. Sabin, J. Chem. Phys., 71 (1979) 3396. 9. A. St-Amant, D.R. Salahub, Chem. Phys. Lett., 169 (1990) 387. 10. A. St-Amant, Ph.D. Thesis, Universit~ de Montr6al, (1991). 11. C. Daul, A. Goursot, D.R. Salahub, Numerical Grid Methods and Their Application to Schrodinger's Equation, Cerjan, C., Ed.; NATO ASI Series, Kluwer Academic Press, 412 (1993) 153. 12. A.D. Becke, Phys. Rev. A, 38 (1988) 3098. 13. J.P. Perdew, Phys. Rev. B, 33 (1986) 8822; erratum in Phys. Rev. B, 38 (1986) 7406. 14. S.H. Vosko, L. Wilk, M. Nusair, Can. J. Phys., 58 (1980) 1200. 15. I. Phpai, A. Goursot, F. Fajula, D. Plee, J. Weber, J. Phys. Chem. to be published. 16. A.D. Buckingham, Intermolecular Interactions: From Diatomics to Biopolymers, Pullman, B.; Ed.; John Wiley & Sons, Chichester, (1978). 17. F. Visser, P.E.S. Wormer, W.P.J.H. Jacobs, J. Chem. Phys., 82 (1985) 3753.


117

L o a d i n g and l o c a t i o n of w a t e r m o l e c u l e s in the zeolite clinoptilolite Y.M. Channon a, C.R.A. Catlow a, R.A. Jackson b and S.L. Owens c. aDavy Faraday Research Laboratory, The Royal Institution of Great Britain, 21 Albemarle Street, London, W 1X 4BS. bDepartment of Chemistry, Keele University, Keele, Staffordshire ST5 5BG. cCompany Research Laboratory, British Nuclear Fuels P.L.C., Springfields Works, Salwick, Preston PR4 0XJ. 1. INTRODUCTION This study concerns the application of computer modelling techniques to the behaviour of water in the zeolite, clinoptilolite. Such techniques have been extensively applied to dehydrated zeolites [1]; but their application to hydrated systems has been limited. Clinoptilolite is a naturally occurring zeolite that is a member of the heulandite group of zeolites. It has a unit cell formula NaxCayAl(x+2y)Si36_(x+2y)O72.24H20, and is isostructural with heulandite having an Si/A1 ratio of approximately 4. Clinoptilolite has a 2-D microporous channel system which was first characterised for heulandite [2]. Two channels run parallel to each other and the c-axis: channel A, a 10-member ring and channel B, an 8member ring. Another 8-member ring, channel C, lies along the a-axis and intersects the A and B channels. Figure 1 clearly shows channel A and B.

Figure 1 Structure of clinoptilolite showing the channels A, the 10 - member ring and B, the 8-member ring.

118 Widely used in industry and employed in gas separation and ion exchange [3], the material has a strong affinity for aqueous caesium and strontium ions, even in the presence of other aqueous cations [4, 5]. This property is exploited via ion exchange in the nuclear industry, and clinoptilolite was used to decontaminate the surrounding area after the accidents at Chernobyl and Three Mile Island [6]. Investigation of the loading and location of water in the zeolite is of importance owing to the influence water molecules exert on the structural and chemical properties by binding certain extra framework cations at specific positions within the framework. Although the location of water and the strength with which it is bound to the structure have been studied experimentally [7, 8], the location of cations and water molecules, as well as the coordination of these species within the structure is still poorly understood. As noted there have been many computational studies on dehydrated zeolites [9, 10], but few on hydrated systems. In this paper we report the first attempts to model hydrated clinoptilolite, and focus on the networks of hydrogen bonded water molecules that form inside the siliceous and aluminium substituted material. 2. M E T H O D O L O G Y Our work is based on the use of 'force field' methods, i.e. techniques employing interatomic potentials. The models used for the latter are described before we summarise the main techniques involved in the study.

2.1 Interatomic Forces Interatomic potentials describe the total energy as a function of the nuclear co-ordinates. These potential functions can be analytic or numerical. There are two broad categories of potential models; first those based on the Born and other related models for ionic solids which describe the system in terms of ions interacting via short and long range terms. The second type is molecular mechanics forcefields; these describe explicitly the bond angles and torsions for each specific system with the energy being given as a function of bond angles, lengths, torsional planes and other cross terms and their associated deviations from equilibrium values. The present study uses the molecular mechanics forcefield cff91_czeo (distributed by BIOSYM Technologies [ 12]), which has been specifically designed to be used in zeolite simulations. This forcefield is used with the following simulation techniques. 2.2 Molecular Dynamics This technique iteratively solves Newton's classical equations of motion. As discussed by Allen and Tildesley [ 13] a standard Taylor series is used to update positions and velocities by a suitable algorithm such as those first described by Verlet [ 13, 14]. The integration time step At, is typically 10-15 seconds. After the solution of the Taylor series equation, the original co-ordinates are replaced by the new ones, the velocities are updated and the acceleration is corrected by calculating the gradients, with the new co-ordinates. The work carried out in this study used the DISCOVER code [ 12] which embodies the Verlet algorithm.

119

2.3 Monte Carlo Techniques These generate ensembles by searching the configurational space of a system by a series of random moves, following for example, methodology developed by Metropolis and Ulam [15]. In the present case a simpler approach is used in which molecules are inserted randomly into the zeolite, and configurations are accepted when their energies fall below a specified threshold. 2.4 Energy Minimisation This technique explores the potential energy surface to locate configurations of minimum energy, and as such yields only static information. Iterative numerical procedures are used to search for the minimum, employing both first and second derivatives of the energy function. The present study uses the code DISCOVER [ 12]. 2.5 Multi Docker Our aim is to locate low energy sites of sorbed water molecules in clinoptilolite. To achieve this the computational techniques use a combination of molecular dynamics, Monte Carlo and energy minimisation methods, all employing a molecular mechanics forcefield. Specifically we use the Docker code developed by BIOSYM Technologies [12] which employs the methodology described by Freeman et al [ 16]. A molecular dynamics simulation is carried out at 298K on a water molecule. This generates a library of possible conformers. A Monte Carlo algorithm is then employed to choose Table 1 Adsorption energy of water, for the lowest energy configuration, as a function of molecule number in system 1. Water Molecule Number Energy (kcal / mole) 1 -5.4 2 -11.4 3 -11.7 4 -5.5 5 -11.2 6 -11.5 7 -17.5 8 -12.3 9 -17.1 10 -12.2 11 -11.8 12 -12.3 13 -14.1 14 -12.5 15 -13.2 16 -12.1 17 -10.6 18 -12.5 19 -12.6 20 -10.3

120

randomly a conformation from the library and insert it at random within the zeolite framework. Each accepted configuration is subsequently energy minimised; the configuration with the lowest energy is then used as the starting structure and the process repeated until the zeolite contains the required number of sorbates. 3. R E S U L T S Our primary aim in the initial study was to identify the ways in which water molecules are accommodated within clinoptilolite and the nature of the resulting hydrogen bonding patterns. We present the results as applied to two systems.

3.1 System 1 Twenty water molecules of adsorption of water number and Table 1 lists molecules are adsorbed each other t

I

I

C~l

CO

.

.

~

were iteratively docked into this system. Figure 2 shows the energy (for the lowest energy configuration), as a function of molecule the adsorption energy for each water molecule. The first three water into the same channel B/channel C intersection, and interact with

I

I

I

t.ID

.

.

I

"4:)

.

.

I'~ .

r

I

I

I

O'.

.

I

I

I

(2:)

,-'-

Oq

cO

~

~

,"7"

,"7-

,"7-

~

I

I

I

I

I

143

',O

r',,,

co

O',

CD

~

'--

,"7"

~

"'-

C'q

-4 e o

E -6

-

-

,-

-

-,

.

.

.

.

,-

-

- , -

-

-~

-

-

7

-

-

J-

-

- ~ -

-

-,

-

-

7

-

-

,-

-

-,

-

-

-,

-

-

,-

-

- , -

-

-,

-

-

z

-

-

L_

ID ~

-8

,, -10

$I " -12

.

.

.

i i i i il i i i i ill i ill 'iiiiii' .

.

.

.

.

.

.

.

.

till

-14

-16 -18 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Water

Molecule

Number

Figure 2 Graph of adsorption energy for water molecules in system 1 and the framework via hydrogen bonds. The next three water molecules locate at the intersection of channel A and channel C, and interact with each other, and the framework but not with the first three (see figure 3). The addition of the seventh molecule joins the first six water molecules via a hydrogen bond, and the ninth water molecule locates in a position that creates a circular network of hydrogen bonded water molecules, figure 4. The location of the next ten water molecules is controlled by their interactions with the existing hydrogen bonded network. This is illustrated by figure 2, which shows very little variation in their relative adsorption energies. They all locate at positions near enough to the circular network to form small clusters with sorbates participating in the network as shown in figure 5.

121

Figure 3 Six water molecules in system 1

Figure 4 Large Circular Network of water molecules in system 1

Figure 5 Large network with small clusters of water molecules 3.2 System 2 Here we present preliminary results for a siliceous system containing an aluminium ion substituted for a silicon ion, with the framework charge balanced by an hydroxyl bridge oriented into channel B. Ten water molecules have been iteratively docked into this system. Figure 6 shows the minimum energy of adsorption of water as a function of molecule number, and table 2 lists the relative adsorption energy for each water molecule. The addition of the first seven water molecules takes place in the same intersection of channels B and C, forming linked clusters of hydrogen bonded water molecules that also interact with the framework via hydrogen bonds as illustrated by figure 7. At this point the channel intersection appears to be full and the location of the eighth water molecule is at the boundary of channel C. This is still close enough to interact with the large cluster via a hydrogen bond. The ninth and tenth water molecules adsorb into the adjacent C channel and begin to form a

122

0

"

t

2 -2

_.o 0

.

.

.

.

.

.

.

3 .

.

.

.

.

.

.

.

.

.

.

I

I

I

t

~,

5

6

7

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

I

8

.

.

.

.

.

.

9

.

.

.

.

.

.

.

]o .

.

-4

E i._

e

-6

. . . . .

'.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

o -~

-8

c

-10

tltl

-14

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Water Molecule Number

Figure 6 Graph of adsorption energy for water molecules in system 2 Table 2 Adsorption energy of water, for the lowest energy configuration, as a function of molecule number for system 2. water Molecule Number Energy (kcal / mole)

1

-5.6

2 -11.4 3 -10.7 4 -12.7 5 -10.5 6 -11.6 7 -13.5 8 -11.9 9 -11.4 10 -10.1 new cluster. They hydrogen bond to each other and the eighth sorbate, so creating a large network of hydrogen bonded small clusters as shown in figure 8. 4. DISCUSSION From figure 3 we can see how the first six water molecules have located, as two non interacting trimers in system 1. The adsorption energy of the fourth sorbate is almost the same as the first, reflecting a site similar to the first with no other water-water interaction. The adsorption energy of the seventh water molecule which links the two trimers has the minimum for this system. The low relative adsorption energy of the ninth water molecule also highlights the stability arising from linking the water molecules together through hydrogen bonded networks. At this point there is a large circular network of hydrogen bonded water molecules. Further addition of water molecules results in the formation of

123

Figure 7 Seven water molecules in system 2.

Figure 8 Ten water molecules in system 2.

small cyclic clusters interacting with each other via the large circular network. From these results it can be postulated that until the zeolite has been loaded with six water molecules, it is the confining effect of the structure that is the dominant influence; after this has been attained the dominant influence is the previous water structure, directing the next sorbate to form small trimer and tetramer clusters that are linked to each other via a large network of hydrogen bonds. Consider now system 2. Figure 6 has a totally different pattern from figure 2 and indeed until the addition of the sixth water molecule no distinguishable pattern of cluster formation is observed. The sixth water molecule locates at a position that forms a tetramer and trimer linked to each other by hydrogen bonds; but it is the seventh water molecule that has the lowest adsorption energy for this system. At this point two tetramers form, linked by a water molecule with hydrogen bonds forming a network of two clusters (see figure 7). The channel intersection is now full with water molecules, and the eighth water molecule must locate away from the intersection. It adsorbs at the boundary of channel C, and still along channel B, close enough to the large cluster to interact via a hydrogen bond. This water molecule acts as a link to a new cluster that begins to form on the addition of the ninth and tenth water molecules. From the results of system 1, it can be seen that when a new cluster begins to form there is a marked increase in the adsorption energy, although this increase does not occur if the initial cluster molecule is hydrogen bonded to another water molecule. Such is the case for the ninth molecule which begins a new cluster in the adjacent channel C, but is stabilised by the hydrogen bond to the eighth water molecule and thence the large cluster. The network of hydrogen bonds that has been built up for system 2 is quite different from that built up for system 1, which must be due to the influence of the aluminium and hydroxyl bridge. What both systems do have in common, however, is that the low adsorption energies

124 correspond to the formation of trimeric or tetrameric clusters. The system is even more stable if these small clusters can be linked together with hydrogen bonds to form large networks. The lack of interaction between water molecules and the hydroxyl bridge in system 2 is currently under investigation. It is thought that this may be due to the low loading of a small molecule into a structure containing channels of large dimensions, but it may of course, reflect inadequacies in the interatomic potentials. 5. CONCLUSIONS This study is only the first stage of developing an understanding of the behaviour of water molecules in clinoptilolite. The dominant influences in the location of each water molecule at low loading, and in zeolites in general, is the structure of the existing water network, although of course, the zeolite exerts an influence on the location of the sorbed molecule. These points are illustrated by the different hydrogen bonded water patterns built up, and the eight ring channels preferentially adsorbing water molecules over the ten ring channel. The formation of hydrogen bonded trimers and tetramers strongly influences the energy of the sorbed molecule. The linking of these small clusters also affects their stability. Future studies will be concerned with water behaviour at higher loadings, and different chemical potentials, using the Grand Canonical Monte Carlo methodology. Further studies of the interaction of water molecules with acid sites are also planned. 6. ACKNOWLEDGEMENTS We are grateful to British Nuclear Fuels p.l.c, for supporting this work. We are grateful to BIOSYM Technologies for supplying the Catalysis and Sorption software. 7. REFERENCES 1 C.R.A. Catlow (ed.), Modelling of structure and Reactivity in Zeolites, Academic Press, London, 1992. 2 A.B. Merkle and M. Slaughter, Am. Mineral. 53 (1968) 1120. 3 M.W. Ackley, R.F. Giese and R.T. Yang, Zeolites, 12 (1992) 780. 4 L.L. Ames, Jr. Am. Mineral., 45 (1960) 689. 5 L.L. Ames, Jr. Am. Mineral., 46 (1961) 1120. 6 L.J. King,, D.O. Campbell, E.D. Collins, J.B. Knauer and R.M. Wallice, in Proceedings of the 6th International Zeolite Conference, (eds. Olson, D. and Bisio, A.), Bt~tterworths, Guildford, (1984) 660. 7 K. Koyama and Y. Takeuchi, Z. Kristallogr., 145 (1977) 216. 8 R.L. Ward and H.L. McKague, J. Phys. Chem., 98 (1994) 1232. 9 C.J.J. den Ouden, R.A. Jackson, C.R.A. Catlow and M.F.M. Post, J. Phys. Chem., 94 (1990) 5286. 10 R.A. Jackson and C.R.A. Catlow, Mol. Sim., 1 (1988) 207. 11 M. Born and J.R. Oppenheimer, Ann. Physik., 84 (1927) 457. 12 BIOSYM Technologies Inc., 9685 Scranton Road, San Diego, CA 92121 U.S.A. 13 M.P. Allen and D.J. Tildesley, Computer Simulation of Liquids, Clarendon Press, Oxford, 1987. 14 L. Verlet, Phys. Rev., 159 (1967) 98. 15 N. Metropolis and S.J. Ulam, Am. Stat. Ass., 44 (1949) 335. 16 C.M. Freeman, C.R.A. Catlow, S. Brode and J.M. Thomas, Chem. Phys. Lett., 186 (1991) 23.

Zeolites: A Refined Tool for Designing Catalytic Sites L. Bonneviotand S. Kaliaguine(editors) 9 1995 Elsevier Science B.V. All rights reserved.

125

Withdrawal of Electron Density by Cations from Framework Aluminum in Y Zeolite Deterlnined by A1 XAFS Spectroscopy D. C. Koningsberger a and J. T. Millerb "Laboratory of Inorganic Chemistry and Catalysis, Debye Institute, Utrecht University, P.O. Box 80083, 3508 TB Utrecht, The Netherlands bAmoco Research Center, 150 W. Warrenville Rd., Naperville, IL 60566-7011, USA

The local AI structure and charge density in Y zeolites have been determined by low energy, AI XAFS spectroscopy. The whiteline intensity indicates that the AI charge density in Y zeolites decreases with increasing charge of the exchanged cation and correlates with the acidity of the zeolite. This result is consistent with the model that polyvalent cations withdraw electron density from the hydroxyl groups increasing their acidity.

1. INTRODUCTION For most hydrocarbon reactions, alkali metal zeolites, for example Na-Y, are relatively unreactive. Exchange of alkali ions by polyvalent cations like Ca, or La results in increased acidity and catalytic activity. The catalytic activity increases with the increasing charge on the cation [1-3]. Early explanations for the enhancement in activity by cations exchanged into Y proposed that strong electrostatic fields are present within the pores of the zeolite. Because of the rigid structure of the zeolite, the charge on the cation is not fully compensated by the negative charge localized on the aluminum-oxide tetrahedra [4]. Polyvalent cations in low coordination sites were proposed as acid centers [5]. An alternative explanation suggested that cations generate acidic hydroxyl groups by hydrolysis of coordinated 1-120 [6,7]. The acidic hydroxyl groups were confirmed by infrared spectroscopy [8]. Exchange of the alkali ions by ammonium ions with subsequent calcination to produce H-Y generates Bronsted acid sites and enhanced catalytic activity compared to Na-Y. In the absence of non-framework AI, however, the catalytic activity of H-Y is low. The activity of H-Y is greatly enhanced by exchange of a small amount of La ions [9,10]. The polyvalent cations, present in the 13 cages [11], are thought to withdraw electron density from the hydroxyl groups increasing their acidity [9,10,12]. The acid strength and catalytic activity of the Bronsted sites are determined by the polarizing effect of the cation and increases with increasing charge of the cation. In the present study, the effect of the cation on the charge density of the A1 ion and the AI-O bond distance has been determined by AI XAFS spectroscopy. As the charge on the cation increases from Na to H to Ca to La, the electron density on the A1 decreases. The

126 charge on the Al ion parallels the acidity of the catalysts. By contrast, the A1-O bond distance is independent of the type of exchanged cation and was about 1.7 A.

2. EXPERIMENTAL Na-Y was a commercial zeolite (LZY-54) purchased from UOP. The sample crystallinity was confirmed by XRD and had a unit cell dimension of 24.676 A, ca. 56 A1/unit cell. The preparation of H-Y has previously been given [13]. Briefly, Na-Y was repeatedly ammonium exchanged and carefully calcined to give H-Y. The crystallinity by XRD was 98% based on Na-Y as a standard, and the unit cell dimension was 24.643/~, or 52 Al/unit cell (11.5 wt% Al and 0.13 wt% Na). H-CaY (and similarly H-LaY) were prepared by ion exchange of Na-Y three times by a ten fold excess of 0.5 M Ca(NO3)2. The Ca-Y was wash, dried at 100C overnight and calcined at 300C for 3h. The Ca-Y was ion exchanged three times with a ten fold excess of 2 M NH4NO3, washed, dried and calcined at 300C for 5 h. The elemental analysis for H-CaY was 11.6 wt% Al, 1.5 wt% Ca, and 0.2 wt% Na, and for H-LaY was 12.4 wt% Al, 12.4% La, and 0.4 wt% Na. Standard methods were used to determine the Si and Al NMR, and the number of acid sites were determined by NI-I3 TPD [14]. EXAFS measurements were performed at the sot~ X-ray XAFS station 3.4 of the SRS at Daresbury (UK). This station is equipped with a quatrz, double crystal monochromator, and harmonic contamination of the X-ray beam was minimized by collimating mirrors. The estimated resolution was 1.5 eV at the A1 K-edge (1559 eV). The data were collected simultaneously with a fluorescence and an electron yield detector. Datum was collected with a k-space scan mode (start, 3 sec.; end, 30 sec.), and six scans were averaged in order to minimize both high and low frequency noise. Sample preparation, reference compounds, experimental conditions and standard procedures for analysis of XAFS data have previously been reported [13]. The near edge spectra were determined from the electron yield data, and the fluorescence data were used to generate the EXAFS function. The data were analyzed using the latest version of the Utrecht University XAFS Data Analysis Program (XDAP) which allows for fitting in r-space.

3. RESULTS X-ray diffraction and N2 micropore volumes indicate that the four Y zeolites are highly crystalline. The Si NMR also indicate that there is little change in the Si/AI ratio of the H-Y H-CaY and H-La-Y compared with Na-Y. Although the ratio of the peak intensities of HLaY are unchanged, the resonances are significantly broadened. The Si/A1 ratio of all catalysts is about 2.5 [ 15]. The Al NMR of Na-Y indicate that all of the Al is in tetrahedral coordination. For H-Y, 85% of the Al are in tetrahedral coordination with about 15% in octahedral coordination. The Al NMR of H-CaY indicate that 85% of the A1 is in tetrahedral coordination with 10% in octahedral coordination. The remaining 5% A1 seems to be in a distorted tetrahedral coordination which appears as a shoulder (about 55 ppm) of the main tetrahedral resonance. For H-LaY, 55% of the Al is in a symmetric tetrahedral coordination,

127

i.e., a resonance at 60 ppm. Approximately, 10% of the A1 is octahedral, 0 ppm, while the remaining 35% of the AI occurs as a broad resonance centered at 30 ppm. Since the Si NMR indicate little loss of structural A1, this 30 ppm peak may be due to structural, tetrahedral AI located near La ions resulting in a distorted electronic coordination about the AI. For each AI in the lattice, charge balance requires one equivalent of univalent cation. The elemental composition of H-Y, H-CaY and H-LaY indicate that there are 4.2, 4.3 and 4.6 mmol/g of AI, respectively. In addition, the univalent cation charge (due to Na, Ca or La) for H-Y, H-CaY and H-LaY are 0.06, 0.8 and 2.8 mmol/g. The number of protons required for charge balance for each, therefore, is estimated to be 4.1.3.5 and 1.8 mmoVg. The number of acid sites which strongly chemisorb NH3 in H-Y, H-CaY and H-LaY were 3.9, 3.6, and 1.9 mmol/g, respectively, and is equivalent to the number of protons in each catalyst. In H-CaY and H-LaY the cations, Ca +2 and La +3, do not chemisorb ammonia and are not strongly acidic. The x-ray absorption near edge spectra (XANES) of Na-Y (solid line), H-Y (dotted line), H-CaY (dashed line), and H-LaY (dotted-dashed line) are given in Figure 1. There is a clear distinction in the white line of the four samples with the intensity increasing in the order Na-Y < H-Y < H-CaY < H-LaY. A

N3 E

it;

L_

0

J t'.

~ ",

i..' \~ "-, ~." \ ' ; : ,

e,-

~c-2

/h

O e~

'.- .... ",,.

I

L_

I

... .......

.

J

-- .2"--:,'"

0

,',1 L_ !

X 0

.__j 0

-5

I

1

I

1

0

5

10

15

20

Energy (eV) Figure 1. X-ray Absorption K-edge of AI (Normalized) for Na-Y (solid line), H-Y (dotted line), H-CaY (dashed line) and H-LaY (dot-dash line). Figure 2a and 2b show the average EXAFS spectum and the Fourier transform of Na-Y (solid line) and H-CaY (dotted line), respectively. The high signal-to-noise ratio of the EXAFS data allows for the detection of higher coordination shells. The node positions for the two catalysts are nearly identical and indicate little difference in the A1-O bond distance which is confirmed in the Fourier transform. The lower amplitude of the first shell AI-O peak in the Fourier transform of H-CaY indicates a larger distortion in the AI-O coordination in

128

comparison to that in Na-Y. Similarly, Figure 3a and 3b compare the EXAFS spectrum and Fourier transform for H-CaY (solid) and H-LaY (dotted). Again, the node positions of the two catalysts indicate little difference in the Al-O bond distance. The amplitude of the first shell A1-O peak of H-LaY is slightly lower than that of H-CaY indicating a slightly larger distortion of the A1-O coordination.

0.4

0.3

0.3

0.2

0.2 ~ , 0.1

~ 0.1 =,?. I--0

0

'c. -0.1

-0.1

u. -0.2

"0"22

3

4

5

6

7

8

9

"0"30

k (l/A)

1

2

3

4

R (A)

Figure 2. a) Raw EXAFS data, and b) Fourier transform (k 1 : Ak = 2.7-8.2 A"1) of Na-Y (solid line) and H-CaY (dotted line).

0.4

0.3

0.3

~E 0 2

a

0.2

b

ol

c.9 -0.1

-o.1 " 0 "2 2

,,~ -o 2

3

4

5

6

k (l/A)

7

8

9

-0 301 - ' - ' ~ ~ ~ ~ ~ ~

~

R (A)

Figure 3. a) Raw EXAFS data, and b) Fourier transform (k ~ 9Ak = 2.7-8.2 A "~) of H-CaY (solid line) and H-LaY (dotted line). The first shell A1-O peak in the Fourier transform of the zeolite samples were fit in rspace (both magnitude and imaginary parts) using a non-linear multiple shell fitting routine [13]. All catalytsts were fit over the same data range (k ~ weighted, AR = 0.5 - 2.0 A, Ak = 2.7- 8.3 A'~). The number of independent parameters (N~p = 2*Ak*AR/x + 1) was 6.3, and the

129 degrees of freedom (Nfrce = N,,ap - Nat) were 2.3. The first shell fit in r-space is given in Figure 4a for H-LaY. The data and the fit were inverse Fourier filtered over the same r-range and are compared in Figure 4b. The quality of the fits is typical for all catalysts. The coordination parameters are given in the Table. Systematic errors are minimized by using the same background subtraction and normalization procedures for all data sets and by calibration with the reference phase shift and back-scattering amplitudes. Although fitting in r-space increases the useful data range, at present, it is not possible to calculate the limits of accuracy. For all Y zeolites, the AI-O coordination number is 4 in agreement with tetrahedral coordination. In addition, the AI-O bond distance for all catalysts is similar, around 1.7 A, also consistent with an A1-O tetrahedral coordination. The disorder (DebyeWaller factor) in the A1-O coordination increases in the order Na-Y < H-CaY < H-LaY.

Table Coordination Parameters Parameters

N

Ac~2

R

AEo

(xl 0 3) (,~2)

(A)

(eV)

Coordination: AI-O

E

Na-Y

4.0

-0.004

1.70

1.4

H-CaY

3.9

-0.002

1.71

-0.5

H-LaY

4.1

0.000

1.70

-0.3

1

0.3

o02

I-.

A ~

0

0

"~ -0.1 -0 2

-0.3 ' 0

1

1

9

i

1

2 3 R (A)

4

-0 12

3

4

5

6

7

8

9

k (llA)

Figure 4. Results of fit in r-space (k 1 9Ak = 2.7-8.2 A "l', AR "- 0.5-2.0 A) for H-LaY (data: solid line, and fit: dotted line), a) Fourier transform, and b) Fourier filtered data from 4a.

130 4. DISCUSSION As shown in Figure 1 the white line intensity, representing a ls to 3p transition, is sensitive to the type of cation exchanged into the zeolite. The white line intensity is the highest for H-LaY and increases in the order Na-Y < H-Y < H-CaY < H-LaY. Previously it was shown for H-Y and Na-Y that the whiteline intensity was higher for H-Y than Na-Y [13]. For these catalysts, the order of the electron densities derived from the white line intensities is in agreement with theoretical calculations for zeolite clusters. For example, the positive charge on the A1 in a protonated aluminosilicate ring, H § A1SiO3(OH)6", is 1.52 while on a symmetrically coordinated Na-aluminosilicate ring, Na § AISiO3(OH)6, the AI charge is 1.43, or 0.09 electrons less on the protonated cluster [16]. The whiteline intensities in this study, therefore, indicate that the electron density on the A1 is lowest, or positive charge of the AI ion is highest, for H-LaY and increases in the order H-LaY < H-CaY < H-Y < Na-Y. Structural determination of the A1-O bond distance in H-LaY, H-CaY and Na-Y indicate that the average distance is very similar, 1.70, 1.71, and 1.70 A, respectively. Previously, the AI-O bond distance in Na-Y was reported as 1.62 A [13]. The longer distance reported here is due to better data quality and the different fitting procedure, i.e., fitting in r-space. The current fluorescence data show a more linear response with increasing energy allowing for more accurate background subtraction. In addition, for AI in zeolites, the maximum data range for EXAFS analysis is limited up to about 8.5 A~ due to the overlap of the Si K-edge. Fourier filtering for fitting in k-space, however, limits the useful data range to about 3.5 A l [ 13]. By fitting in r-space, the useful data range can be increased by about 2 A "l improving the accuracy of the Al-O bond distance determination. Applying this new procedure to the previous data obtained on Na-Y results in an A1-O bond distance of 1.68/~ in general agreement with the results from this study. Reanalysis of the previous EXAFS data of H-Y results in an AI-O distance of 1.67/~. The current results indicate that within the limits of accuracy the AI-O bond distance is not sensitive to the type of cation exchanged into the zeolite. The whiteline intensity indicates that the cation has a dominating effect on the electron density of the aluminum ions. The order of the electron density parallels the acidity of the catalysts. That is, the electron density of the aluminum ion is the lowest and the whiteline intensity is highest for the most highly acidic zeolite, e.g., H-LaY. Since the cations are coordinated to lattice oxygen ions, it is likely that the cations are withdrawing aluminum electron density through these coordinated oxide and hydroxide ions. The low electron density on the aluminum ion also suggests that the electron density of the lattice oxide ions is lower due to coordination with the cation. Since the acid sites in H-CaY and H-LaY are due to the protons, the correlation of the whiteline intensity with the acidity suggests that a higher acid strength in ion-exchanged Y is due to withdrawal of electron density from the hydroxyl groups by the nearby polyvalent cations consistent with the model for enhanced acidity proposed by Lunsford [9,10,12].

131 5. CONCLUSION Low energy A1 XAFS spectroscopy is a powerful technique for directly measuring the local AI-O bond distance and charge on the AI ion. The average AI-O bond distance determined by EXAFS spectroscopy is independent of the type of exchanged cation in zeolite Y. On the other hand, the whiteline line intensity is very sensitive to changes in the charge on the AI ions induced by exchanged cations and correlates with the catalyst's acidity. As the charge on the cation increases, the charge on the A1 decreases suggesting that polyvalent cations withdraw electron density from the AI through the oxygen ions. The correlation of the whiteline intensity with the acidity suggests that the origin of the enhanced acid strength is due to the withdrawal of electron density from acidic hydroxyl groups by the nearby polycation.

REFERENCES 1. R.C. Hansford and J.W. Ward, J.Catal., 13, (1966) 316. 2. P.E. Eberly Jr. and C.N. Kimberlin, Adv. Chem. Ser., 102, (1971) 374. 3. M.L. Poustma, Zeolite Chemistry and Catalysis, ACS Mono. 171, J.A. Rabo (ed.), ACS, Washinton, D.C., (1976) 437. 4. J.A. Rabo, P.E. Pickett, D.N. Stamires, and J.E. Boyle, Proc. 2nd Int. Cong. Catal., Editions Technip, Paris, II ( 1961) 2055. 5. P.E. Pickett, J.A. Rabo, E. Dempsey and V. Schomaker, Proc. 3rd Int. Cong. Catal., W.M.H. Sachtler, G.C.A Schuit and P. Zwietering, eds., North-Holland Publishing Co., Amsterdam, I (1965) 714. 6. C.S. Plank,, Proc. 3rd Int. Cong. Catal., W.M.H. Sachtler, G.C.A Schuit and P. Zwietering, eds., North-Holland Publishing Co., Amsterdam, I (1965) 727. 7. P.A. Jacobs, J.B. Uytterhoven, J. Chem. Soc., Faraday, 69, (1973) 373. 8. J.W. Ward, J. Catal., 10, (1968) 34. 9. J.H. Lunsford, Fluid Catalytic Cracking II, Concepts in Catalyst Design, ACS Sym. Ser. 452, M.L. Occelli (ed.), ACS, Washington D.C., (1991) 1. 10. R. Carvajal, P.-J. Chu and J.H. Lunsford, J. Catal., 125, (1990) 123. 11. A.K. Cheetham, M.M. Eddy and J. M. Thomas, J. Chem. Soc., Chem. Commun., (1984) 1337. 12. P.O. Fritz and J.H. Lunsford, J. Catal., 119, (1989) 85. 13. D.C. Koningsberger and J.T. Miller, Catal. Lett., 29, (1994) 77. 14. B.L. Meyers, T.H. Fleisch, G.J. Ray, J.T. Miller, and J.B. Hall, J. Catal., 110, (1988) 82. 15. Engelhardt, G.; Michel, D. High Resolution Solid-State NMR of Silicates and Zeolitesi John Wiley and Sons, 150 (1987). 16. R.A. van Santen, B.W.M. van Beest and A.J.M. de Man, Guidelines for Mastering the Properties &Molecular Sieves, D. Barthomeuf (ed.), Plenum Press, New York, (1990) 201.


Zeolites: A Refined Tool for Designing Catalytic Sites L. Bonneviot and S. Kaliaguine (editors) 1995 Elsevier Science B.V.

133

Geometry of the Active Sites m Zeolites under Working Conditions F. Fajula Laboratoire de Matrriaux Catalytiques et Catalyse en Chimie Organique, URA 418 CNRS, ENSCM, 8, rue de l'Ecole Normale, 34053 Montpellier Cedex, France I. INTRODUCTION Due to the large use of zeolitic materials in major acid-catalyzed industrial processes, the characterization and description of the acid sites, particularly BrOnsted-type acid sites, in zeolites has received continuous attention. Among heterogeneous catalysts, zeolites constitute a unique class of solids because the acid centres are crystallographycally well defined. BrOnsted sites in zeolites can be attributed to acidic protons covalently bound to one of the oxygens linked to lattice aluminium atoms for which they provide charge compensation. In principle, the position of the active catalytic sites in the tridimensional network as well as their local geometry in terms of bond angles, bond lengths and environment might be fully determined and their acid strength and reactivity predicted. Considerable progress in the understanding of zeolite acidity has been gained from extensive spectroscopic, quantum chemical and structure modeling studies (see [1-3] for recent reviews), However, the catalytic significance of theoretical considerations has been established definitely only in a limited number of cases. The reasons for such a situation are multiple. The real links between structural acid sites and actual active sites is not straightforward. Reactivity depends indeed on intrinsic acidity but also on the properties of the conjugate ion generated upon protonation and on the contribution of the zeolite framework oxygen atoms in the stabilization of intermediates and transition states. These factors are not always easy to delineate and take into account in calculations and model studies. On the other hand, there are several evidences that H-form zeolites with their full complement of charge compensating protons exhibit little, if any, activity for hydrocarbon conversion reactions such as cumene and hexane cracking [4-6]. Secondary treatments generating extra-framework phases definitely enhance activity by creating new catalytic centers [6-8]. Similarly, although Lewis sites are almost systematically detected on activated zeolites, their direct or indirect implication in the catalytic events remains a matter of debate [9-11 ] and little is known regarding their nature, strength and location. Another main limitation arises from the fact that the actual working surface of a zeolite may be deeply modified by the presence of adsorbed and chemisorbed species, not

134 necessarily participating directly to the catalytic process. These species may behave as spectators, being chemisorbed on other sites, act as poisons or introduce additional diffusion limitations. In the last decade great efforts have been dedicated to adapt conventional spectroscopic techniques, among which infrared and NMR spectroscopy being the most popular, for studying catalytic reactions under conditions that recreate, or at least mimic, actual working conditions. The merits of this approach for the description of the reaction mechanisms at the molecular level and the detection of unrevealed reaction intermediates are well established and have been widely reviewed [12-14]. The in situ protocol should prove also well suited for the identification of the active sites and monitoring the changes of their local geometry during catalytic reactions or acidity measurements. It is my aim, at this symposium devoted to the , to consider briefly some recent examples of such studies and to stimulate discussion about them. The aspects concerning the fate of the bridging structural A1 -OH-Si groups being largely documented, they will not be considered here. The examples I choose deal with the direct spectroscopic observation of peculiar behaviours of zeolites in the presence of sorbates, such as the change in the coordination of aluminium in protonic zeolite Beta, the generation of very strong Lewis acidity on ZSM-5 upon adsorption of olefins and the manifestation of the synergistic effect betwen Bronsted and Lewis sites in Faujasite and Mazzite. 2. REVERSIBLE CHANGE OF THE COORDINATION OF ALUMINIUM Changes in the environment of silicon and aluminium atoms upon hydration-dehydration or adsorption of bases have been evidenced several times for aluminophosphates and silicoaluminophosphates [ 15-18]. The easy adjustment of the lattice in the presence of adsorbates favors the attack of the AIO4 tetrahedra by the bases and generates penta and hexacoordinated framework aluminium species. Such a mechanism is supposed to be at the origin of the Lewisaciditj in these materials [ 19]. A reversible AI'V/AIv' transformation in aluminosilicate zeolites has been suggested to occur in the protonic form of zeolite Beta [20]. 27A1 NMR studies of hydrated H-BEA revealed that 20% approximately of the aluminium atoms were octahedrally coordinated. Replacement of the protons by ammonium ions, by treatment of the zeolite with gaseous ammonia at 100~ or by sodium and/or pgtassium cations, by standard exchange procedures, led to the disappearance of the AlW NMR signal and to an increase of the intensity of the AIIv one (Fig. 1). Material balances demonstrated that no aluminium was extracted from the solid during these treatments. Density functional calculations performed on pentameric cluster models of zeolite Beta with optimized geometry [21] showed that, due to the strong electron affinity of protons, the structure of the AIO4 units was considerably distorted with Si-O and AI-O bonds of the Si-OH-AI bridges elongated and weakened. Moreover the least stable site was found to be the one associated with two four-membered tings in the framework. On these bases, and by analogy with what has been reported for

135

d

L

_

ppm Figure 1.27A1 MAS NMR spectra of zeolite B e t a a) hydrated H-form, b) sample a saturated with ammonia, c) sample b acid-exchanged, d) sample c potassium exchanged [20].

Figure 2 . 2 7

ml MAS

NMR

spectra of zeolite H-Beta. Sealed ampules : a) hydrated sample, b) sample dehydrated at 300~ c) sample b with ethanol adsorbed, d) sample c after reaction at 200~ [24].

bl

d)

It

'

pp.~

136 VPI-5 [22], the AlXaNMR signal in protonic zeolite Beta was thus attributed to framework aluminium atoms linked to 4 lattice oxygens, the oxygen of a hydronium ion and the oxygen of a water molecule. The unique behaviour of zeolite Beta, and particularly the formation of hexacoordinated 27 framework aluminium, has been further investigated by AI MAS NMR by following the changes in the environment of the aluminium nuclei during the dehydration reaction of ethanol into ethylene and water. The reaction was carried out in 5 mm o.d. sealed capsules which were placed directly in an aluminium-free home-made NMR probehead that could spin the capsule at speeds up to 4 kHz [23,24]. NMR spectra were recorded for the parent hydrated zeolite in the protonic form, for the zeolite dehydrated at 300~ under vacuum, for the dehydrated zeolite equilibrated with 10 Torr of ethanol or pyridine and finally, after heating the zeolite sample loaded with ethanol for 30 min at 200~ Figure 2a shows the characteristic 27A1MAS NMR spectrum of hydrated zeolite Beta [20] with signals at 55 and 0 ppm due to tetracoordinated and hexacoordinated aluminium atoms, respectively. The removal of adsorbed water molecules causes a strong distortion of the electric field symmetry around aluminium. The increase of the quadrupolar coupling constants leads to a broadening of the signals which are no longer detected under our conditions (Fig. 2b). Adsorption of ethanol (or pyridine) induces a relaxation of the structure (also evidenced by infrared spectroscopy) and allows to regain the resonance line of tetrahedral aluminium at 55 ppm (Fig. 2c) while that of octahedral aluminium is still not observed. The latter signal develops after ethanol had been dehydrated, producing nascent water molecules which coordinate distorted AIOaH units leading to a more stable octahedral environment (Fig. 2d). These observations are qualitatively consistent with our assignment of the O ppm line to framework hexacoordinated aluminium and confirm the role of adsorbed water in generating it. Additional work is needed to reinforce this hypothesis and provide more quantitative data. The formation of hexacoordinated aluminium has been also postulated in steam dealuminated mazzite after a complexation treatment with acetylacetone [25]. The above results imply some flexibility of the zeolite lattice [26] or, at least, the presence of crystallographic sites able to suffer major local distortions without loss of long range crystallinity. As regards implications for catalysis using Beta zeolite, the change in symmetry of some aluminic sites will certainly modify their chemical feature. This may be of little importance in the case of the conversion of hydrocarbons (at high temperatures and with a low surface coverage by water) but could become significant for reactions involving polar substrates in the presence of solvents. 3. GENERATION OF LEWIS ACIDITY BY ADSORPTION OF HYDROCARBONS Medin et al [27] were the first to report on infrared studies of adsorbed acetonitrile and ammonia on H-ZSM-5 showing that olefms pre-chemisorbed at room temperature generated strong Lewis sites which were not detected in the freshly pretreated zeolite. As illustrated in

137 Figure 3a, the infrared spectrum of CD3CN adsorbed on the clean surface in the frequency -1 range of CN vibrations reveals the sole presence, at 2295 cm , of molecules coordinated to acidic hydroxyl groups. In the presence of pre-chemisorbed propylene, a second -1 signal at 2370 cm develops (Fig. 3b) that the authors attributed to new Lewis sites strongly interacting with the base. The formation of the new sites was / completely reversible as they disappear / after desorption at 500~ and readsorption of base. A very similar 0~ effect was observed when using I ammonia as probe. This unusual l 2~70 manifestation of Lewis acidity has been (U I I t.~ 50 explained by the cleavage of AI-O ! b). bonds in bridged alkoxyl groups and I inversion of the AIO4 tetrahedra to which the base molecules coordinate T 2~5 (Fig. 4). Support to this hypothesis was provided by quantum-chemical 100 L I ', 1 2 2400 2~00 2200 21 O0 calculations showing that the process Wavenumbers depicted by figure 4 would be Figure 3. Infrared spectra of CD3CN adsorbed energetically unfavorable for R = H but on H-ZSM-5 ' a) fresly pretreated sample, would become possible when R is an b) sample containing pre-adsorbed propene [27]. alkyl group. Such an interpretation was disputed by Bystrov [28] who explained the above results by invoking the formation of nitrilium cations in which acetonitrile is complexed to carbenium ions, the latter acting as very strong Lewis sites. The hypothesis of Bystrov has been elegantly confirmed by Jolly et al [29] who investigated the surface acidity of ZSM-5 in the presence of a series of pre- (or co-) adsorbed olefins choosen so as to generate carbenium ions with different structures. As expected, different nitrilium complexes were obtained using methylcyclopentene (signal at 2376 cm , Fig. 5c, d) and cyclohexene (signal at 2385 cm , Fig. 6a), due to interaction of acetonitrile with tertiary and secondary carbenium ions, respectively. The method was also able to detect the two carbenium ions, cyclohexyl and methylcyclopentyl, involved in the cyclohexene isomerization reaction (Fig. 6b,c) as well as those involved in the rearrangement of adsorbed propene oligomers. Two main conclusions can be drawn from this study. Firstly, nitriles constitute very efficient probes for the identification of adsorbed carbenium ions which are the intermediates of most hydrocarbon acid-catalyzed reactions. Secondly, under working conditions, the acidity of the surface is significantly modified. The strong Bronsted acid sites of ZSM-5 do

138 R

I X

o / \ X

Si

/I

I

1.6

b s o

1.4;

;

/

/

\

o

-,

!

,o

0

/

/

0

X/ Si

/I

O O I \

I

\

O "o" / \

/~

~

2376

\

I ~ ~ ^

/~

l

X "AI...B

/7

OO

1

'.'---~

o

+ B,-

IX

\

A

I

0 AI

O O

Figure 4. Proposed mechanism for the generation of Lewis acidity according to Medin et al. [27].

R

d) c)

0.4

b)

0.2 0.0

4000

~

3500

~

3000 2500 Wavenumbers (cm-1)

"

a)

2000

Figure 5. Infrared spectra of methylcyclopentene adsorbed on H-ZSM-5 9 a) freshly pretreated sample, b) aider adsorption of CD3CN, c) immediately aider adsorption of methylcyclopentene, d)aRer 1 h [29]. 0.8 0.7-

2300

0.6-

c) b)

0.50.4-

a) 0.3. 0.2

2soo

24bo

-2~

22bo

2~00-

Wavenumbers (cm-1)

Figure 6. Isomerization of cyclohexene on ZSM-5. Infrared spectra of CD3CN adsorbed after a) 1 h at RT, b) 1 h at 50~ c) 5 min. at 200~ [29].

139 not exhibit sufficient strength to form nitrilium ions [29,30] whereas the intermediate carbenium ions they generate do. Moreover, not only the strength, but also the nature of the acidity is modified. Both factors should be considered when dealing with complex reactions such as, for example, coke formation or paraffin/olefin alkylation.

4. SYNERGY BETWEEN LEWIS AND BRONSTED SITES.

4.1. The 3600cm

-1

infrared signal in faujasites

Several infrared studies of steam dealuminated faujasite have pointed out the enhanced acid strength (they are able to protonate acetonitrile) of hydroxyl groups characterized by a -1 stretching vibration frequency of 3600 cm [5, 8, 11, 31 ]. They are supposed to correspond to OH groups pointing to the large cage, perturbed by cationic tetrahedral aluminium present as A1OH+ entities in the beta cages. This signal is not observed on faujasites dealuminated by isomorphous substitution, free from extra framework species. The higher efficiency of these sites for the conversion of hydrocarbons has been demonstrated by using an infrared cell as catalytic reactor, operated under dynamic conditions at a temperature of 400~ [6,1 l!i Activity for n-hexane cracking was definitely associated with the presence of the 3600 cm OH groups. Regular high frequency OH hydroxyls and Lewis sites were inactive. All the accessible OH groups were found, by contrast, active for the conversion of cyclohexene, a -1 less demanding reaction. A direct relationship betwen the intensity of the 3600 cm signal and n-hexane cracking activity did not exist however since poisoning experiments using 2,6-1 lutidine and pyridine revealed that the 3600 cm OH groups were heterogeneous, only some of them being acidic enough to initiate the reaction.

4.2. Formation of iminium ions on mazzite. A recent study of the surface acidity of dealuminated mazzite using infrared and ultraviolet spectroscopy of adsorbed pyridine provided a direct evidence for the presence of very strong acid sites associated with Br6nsted/Lewis pairs [32]. At very low coverages in base, infrared bands at 2914, -1 2848, 1496 and 1462 cm and an ultraviolet signal at 205 nm, Figure 7. Dihydropyridistinct from those attributable to A i ~ dinium ion formed pyridinium ions and coordinated Oby reaction of pyridine pyridine, were detected. These on a BrOnsted/Lewis spectral features were pair of sites [32]. characteristic of a conjugated system containing aliphatic CH 2 groups and have been thus assigned to iminium L (dihydropyridinium) ions (Fig. 7). Their formation results

140 from a nucleophilic attack and protonation of pyfidine molecules adsorbed on Lewis sites and requires, therefore, the presence of paired Lewis/BrOnsted sites. Since loss of the aromaticring resonate energy is an energetically demanding process we can expect that the sites on which iminium ions form exhibit a very strong acid character. Additional experiments using heat-flow calorimetry and X-ray photoelectron spectroscopy confirmed the presence of such very strong sites [33]. The formation of iminium ion is not specific to mazzite and does not require, apparently, a particular spatial arrangement of the acid sites. An infrared signal at 1462 cm-1 that could characterize adsorbed pyridinium species has also been observed after high-temperature desorption of pyridine on a series of zeolites with different structures [32], on amorphous silica-alumina [34] and on clays [35]. Proton mobility or migration of cationic species with strong Lewis character may thus also account for this phenomenon. NMR and modeling studies are underway to clarify this point. 5. CONCLUDING REMARK. Experimental approaches aiming at a description of the fate of the active sites of acidic zeolites in the presence of adsorbed reactants or probe molecules may provide a less conventional picture of the surface than the one traditionally emphasized through well established structural and physical aspects of zeolite chemistry. These experiments are certainly promising to refine our understanding of the interaction between reactants and products with the inorganic framework at a molecular scale. In the present state of the art, the information available in the literature is, however, rather limited and essentially qualitative. The technical difficulties associated with the characterization of actual catalysts under non-ideal spectroscopic environments constitute undoubtly, to day, a major limitation. Considerable progress can be nevertheless expected in the future on account of the extensive efforts dedicated to the design of spectroscopic probes simulating reactor conditions. REFERENCES 1 J. Sauer, Stud. Surf. Sci. Catal., 84 (1994) 2039. 2. R.A. van Santen, Stud. Surf. Sci. Catal., 85 (1994) 273. 3 V.B. Kazansky, Stud. Surf. Sci. Catal., 85 (1994) 251. 4. S.J. DeCanio, J.R. Sohn, P.O. Fritz and J.H. Lunsford, J. Catal., 101 (1986) 132. 5. R. Carvajal, P.J. Chu and J.H. Lunsford, J. Catal., 125 (1990) 123. 6. S. Jolly, J. Saussey, J.C. Lavalley, N. Zanier, E. Ben~zi and J.F. Joly, Ber. Bunsenges. Phys. Chem., 97 (1993) 313. 7. R.M. Lago, W.O. Haag, R.J. Mikovsky, D.H. Olson, S.D. Hellring, K.D. Schmitt and G.T. Kerr, Proceedings, 7th Intern. Zeolite Conf., Kodansha, Tokyo, p 677, 1986. 8. F. Lonyi and J.H. Lunsford, J. Catal., 136 (1992) 566. 9. Y. Hong, V. Gruver and J.J. Fripiat, J. Catal., 150 (1994) 421. 10. G.R. Bamwenda, Y.X. Zhao and B.W. Wojciechowski, J. Catal., 150 (1994) 243. 11. S. Jolly, J. Saussey and J.C. Lavalley, J. Mol. Catal., 86 (1994) 401.

141 12. J.A. Rabo and G.J. Gajda, Catal. Rev. Sci. Eng., 31 (1990) 385. 13. J.F. Haw, NMR Techniques in Catalysis, A.T. Bell and A. Pines, Eds., Dekker, New York, 1994, P 139. 14. I.I. Ivanova and E.G. Derouane, Stud. Surf. Sci. Catal., 85 (1994) 357. 15. L.M. Kustov, S.A. Zubkov, V.B. Kazansky and L.A. Bondar, Stud. Surf. Sci. Catal., 69 (1991) 303. 16. A. Stein, B. Wehrle and M. Jansen, Zeolites, 13 (1993) 291. 17. P.J. Grobet, J.A. Martens, I. Balakrishnan, M. Mertens and P.A. Jacobs, Appl. Catal., L21 (1989) 56. 18. R. Vomscheid, M. Briend, M.J. Peltre, P. Massiani, P.P. Man and D. Barthomeuf, J. Chem. Soc., Chem. Commun.,6 (1993) 544. 19. M. Derewinski, M. Briend, M.J. Peltre, P.P. Man and D. Barthomeuf, J. Phys. Chem., 97 (1993) 13730. 20. E. Bourgeat-Lami, P. Massiani, F. Di Renzo, P. Espiau, F. Fajula and T. Des Couri6res, Appl. Catal., 72 (1991) 139. 21. I. Papai, A. Goursot, F. Fajula and J. Weber, J. Phys. Chem., 98 (1994) 4654. 22. L.B. McCusker, Ch. Baerlocher, E. Jhan, and M. B01ow, Zeolites, 11 (1991) 308. 23. F. Rachdi, J. Reichenbach, L. Firlej, P. Bernier, M. Ribet, R. Aznar, G. Zimmer, M. Helme and M. Mehring, Solid State Commun., 87 (1993) 547. 24. W. Buckermann, L.C. De Menorval, F. Figueras and F. Fajula, J. Phys. Chem., Submitted. 25. W. Buckermann,B.H. Chiche, F. Fajula and C. Gueguen, Zeolites, 13 (1993) 448. 26. J.A. Rabo and G.J. Gajda, NATO ASI Ser. B, Physics, Vol. 221 (1990) 273. 27. A.S. Medin, V. Yu, V.B. Kazansky, A.G. Pelmentschikov and G.M. Zhidomirov, Zeolites, 10 (1990) 668. 28. D.S. Bystrov, Zeolites, 12 (1992) 328. 29. S. Jolly, J. Saussey and J.C. Lavalley, Catal. Let., 24 (1994) 141. 30. J.F.Haw, M.B. Hall, A.E. Alvaro-Swaisgood, E.J. Munson, Z. Lin, L.W. Beck and T. Howard, J. Amer. Chem. Soc., 116 (1994) 7308. 31. G. Garralon, A. Corma and V. Fornes, Zeolites, 9 (1989) 84. 32. B.H. Chiche, F. Fajula and E. Garrone, J. Catal., 146 (1994) 460. 33. D. McQueen, B.H. Chiche, F. Fajula, C. Guimon, A. Auroux, F. Fitoussi and Ph. Schulz, J. Catal., Submitted. 34. K.H. Bourne, F.R. Cannings and R.C. Pitkethly, J. Phys. Chem., 74 (1970) 2197. 35. S. Bodoardo, F. Figueras and E. Garrone, J. Catal., 147 (1994) 223.



143

Characterization of Hexagonal and Lamellar Mesoporous Silicas, Alumino- and Gallosilicates by Small-Angle X-Ray Scattering (SAXS) and Multinuclear Solid State N M R Zelimir Gabelica 1, Jean-Marc Clacens 1, Roger Sobry2 and Guy Van den Bossche 2 1Laboratoire de Catalyse, FacultEs Universitaires Notre-Dame de la Paix, 61 rue de Bruxelles, B-5000, Namur, Belgique 2Laboratoire de Physique Exp6rimentale, Universit6 de Liege, Institut de Physique, b~timent B-5, B-4000, Li/~ge, Belgique

SUMMARY

Various mesoporous silicas and the corresponding alumino- or gallosilicates have been synthesized using a series of literature or home made recipes. The efficiency of the A1 or Ga incorporation markedly depends on the trivalent source and the evolution (aging) of the so formed Si-M(III) gel-type phases at different starting pH values, prior to the addition of the surfactant-structuring compounds. Low temperature (e.g. < 100~ syntheses yielded mesoporous compounds with hexagonal topology (MCM-41 type), involving thin multilayered Si walls (29Si-NMR data), eventually partly substituted for by A1 or Ga, (27A1and 71Ga-NMR). Such structures do remain stable after calcination in air at 600~ (TG-DSC and sorption data). The calcination either results in the expulsion of the (too voluminous) Ga(III) ions from the framework, or to a structural rearrangement of the remaining A1 species. The resulting solids still exhibit quite strong Br6nsted type acidic sites (NH3-TPD data). When the same gels are crystallized at 150~ fro 2 days, lamellar frameworks (MCM50 type) are preferentially stabilized. They readily collapse upon heating but the final amorphous phase still involves both A1 and/or Ga cationic (acidic) species randomly distributed in various structural environments. Small-angle X-ray scattering (SAXS) proved to be the choice method to unambiguously identify different topologies. The actual roughness (smoothness) of the various as-synthesized or calcined phases could be directly related to the fractal dimension derived from SAXS spectra. The presence of peaks at very small diffraction angles allowed us to estimate that the channel lengths of the thermally stable mesoporous materials essentially extend in the 150-200 nm range. EXPERIMENTAL

The different synthesis procedures adopted in this study were either literature recipes [1-3], sometimes slightly or thoroughly changed by home performed modifications [4]. All

144 the as-synthesized and calcined phases were characterized by using a combination of conventional techniques: powder XRD (topology, purity), atomic absorption and/or EDX (extent of trivalent incorporation), 27A1-, 71Ga- and 29Si- high resolution solid state MAS NMR (coordination and structural quantitative repartition of Ga and A1 bearing species, as well as the various Qx configurations of the Si atoms), TG-DTA-DSC-DTG combined with in situ n-hexane sorption (thermal stability, template content, pore volume determination, crystallinity), BET (surface area) and NH3-TPD (acidity). SAXS data were collected with a Kratky camera (Cu Ks radiation) using an experimental setup specifically built in the Liege University physics department [5]. RESULTS AND DISCUSSION

Gels prepared according to recipes as described in the literature [2,3] and left to crystallize at high temperature (i.e. 150~ for at least 2 days, yield well crystalline mesoporous lamellar (L) type phases. They exhibit good crystallinity (XRD data, high template content), medium surface area (< 500 ma/g) but their lamellar structure collapses as soon as they are heated under conditions currently used to remove the templates from mesoporous materials (600~ air flow). Conversely, thermally stable phases exhibiting a hexagonal (H) topology and a high surface area (> 1000 m2/g) could be prepared by crystallizing the same gels under milder conditions (at temperatures < 100~ or by modifying the gel composition (addition of TMA and/or under a careful pH adjustment and control). In both H and L structures, the degree of A1 incorporation dramatically depends on the nature of the Si and A1 starting materials, their solubility and their final degree of (de)polymerization at the end of the aging process at controlled pH, prior to any template addition. For equivalent experimental conditions, gallium tends to incorporate the final Si framework more quantitatively than aluminum. However, when both trivalents are added simultaneously (as oxides in the presence of various solubilizing additives such as NH3), their degree of incorporation is equivalent (Table 1). In the as-synthesized products, Ga is only partly sited in tetrahedral positions in both the H and L structures, as ascertained by 71Ga-NMR [6]. However, part of it remains either as a dispersed amorphous phase intermixed with the mesoporous compound, or incorporated in its framework, but randomly partitioned throughout the less well defined coordination sites, at least not detectable by NMR. In contrast, all the AI(III) ions are found located in both tetrahedral (T) and octahedral (O) framework sites, in various proportions that could be quantified [4]. Their repartition was shown to strongly depend on the synthesis conditions. After template removal upon calcination, the aluminosilicate phases generally show a better thermal resistance than the corresponding Ga analogs, although a partial dealumination and a marked framework (O) and (T) A1 redistribution is observed by 27A1-NMR. When gallium bearing L type phases are calcined, some residual Ga sited in T positions is still detected by NMR. These phases involve both Q3 and Q4 silicon configurations in various proportions, as seen by 29Si NMR, thus suggesting that both terminal silanols and Si(4Si) structural entities are present. The relative amount of Q3 and Q4 configurations could be quantified, provided the appropriate repetition times between 2 NMR pulses are selected in each case (the Si nuclei in the as-synthesized and calcined phases do relax very differently

145 [4]. Assuming that Q3 essentially corresponds to the terminal hydroxylated surface while the Q4 configurations reflect the non defected bulky Si structure, the calculated Q3/Q4 ratios could be related to the actual thickness of the framework (essentially siliceous) walls. It could be concluded that the lattices of the L type phases involve at least 5 or more Si layers, in which Ga still can find a stable T coordination within the internal layer(s). Table 1A Synthesis conditions Code

Source of trivalent element

T(~

/ time

Synth. cond. [ref.] (1)

1/9 2/11

A1203 Na Aluminate

150 / 65 h 150 / 65 h

[3 a] [3 a]

3/12

---

150 / 65 h

[3 b]

4/14

Na Gallate

150 / 65 h

[3 a]

5/15 6/20

Ga203 Na Aluminate (NH 3)

150 / 65 h 150 / 65 h

[3 a] [3 c]

7/21

Ga203 (NH 3)

150 / 65 h

[3 c]

8/22

A1203 + Ga203 (NH 3)

150 / 65 h

[3 c]

15/7 9/17-1

A1203 (TMA) ---

150 / 65 h 150 / 65 h

[1,5] [4,5]

10/17-2

---

100 / 6 days

[4,5]

11/18-1

A1 Sulfate

150 / 65 h

[4,5]

12/18-2

A1 Sulfate

100 / 6 days

[4,5]

13/19-1

Ga Sulfate

150 / 65 h

[4,5]

14-19-2

Ga Sulfate

100 / 6 days

[4,5]

(1) Letters a, b and c refer to variants in the synthesis recipes described in reference [3]

146 Table 1B Some characteristics of the final mesoporous phases obtained by using various synthesis conditions Final product Code

Nature (1)

1/9 2/11

L L

% incorporation Si

A1

88 (nd) (2)

100 18

Si/M (at.) Ga 24 (nd)

3/12

L

87

---

4/14

L

100

---

81

5/15

L

100

---

84

6/20

L

(nd)

93

45 43 (nd)

7/21

L

100

---

94

26

8/22

L

100

80

76

20

15/7

H

100

41

9/17-1

L

(nd)

---

36

10/17-2

H

(nd)

---

11/18-1

L

100

16

12/18-2

H

88

100

13/19-1

L

(nd)

---

98

(nd)

14/19-2

H

87

---

96

90

311 44

(1) L: lamellar phase; H: hexagonal phase (2) (nd) not determined Similarly, all the as-synthesized H type phases also exhibit a quite broad NMR resonance that could be deconvoluted into two lines of variable intensity, one located a t - 100 ppm and better corresponding to Q3 configurations, the other a t - 110 ppm, assigned to Q4. It is actually not clear to which NMR line should be attributed the Si atoms being interacting with the bulky terminal -N(CH3) 3 groups of the CTABr templates and thus forming Si (3Si, IN) configurations, so that the estimation of the actual wall thickness from only the NMR data in the precursors is dubious, especially when the spectra are broad. After calcination, the spectra are better resolved and more Q3 configurations were detected, suggesting the presence of major silanol terminations in the final calcined hexagonal frameworks, in line with a Hoffman type degradation of the tetraalkyl ends of the template. An estimation of the Q3/Q4 ratios could be achieved. It still suggests the presence of multilayered walls of Si atoms, roughly composed of a maximum of 3-4 Si layers. Undoubtedly the walls of the H phases are thinner than those detected in the L type phases. Irrespectively to the smaller AI(III) ions, Ga(III) species are probably too bulky as to still

147 remain sited in the "peripheral" T positions of the hexagonal structure. Their expulsion is possibly more readily achieved because of the presence of important strains within the thin layers. Finally, the SAXS spectra of the various as-synthesized and calcined compounds strongly confirm the above detected structural features and, in most cases, bring more detailed structural characteristics about the thickness of the walls, the fractal slope, the smoothness of the external envelope and the length of the hexagonal channel systems. The presence of peaks at very small diffraction angles allowed us to evaluate the channel lengths of the thermally stable hexagonal mesoporous materials. Depending on experimental condition and on the presence of trivalent ions in the framework, the mean length essentially extends in the 150-200 nm range. The detailed SAXS study is discussed elsewhere [5]. REFERENCES

1. S.J. Beck, J.C. Vartuli, W.J. Roth, M.E. Leonovicz, C.T. Kresge, K.D. Schmitt, C.T.W. Chu, D.H. Olson, E.W. Sheppard, S.B. McCuller, J.B. Higgins and J.L. Schlenker, J. Amer. Chem. Soc., 114 (1992) 10834. 2. C.Y. Chen, H.X. Li and M.E. Davis, Microporous Materials, 2 (1993) 17. 3. K.M. Reddy, L. Moudrakovski and A. Sayari, J. Chem. Soc. Chem. Commun. (1994) 1059. 4. J.M. Clacens and Z. Gabelica, in preparation. 5. R. Sobry, G. Van den Bossche, J.M. Clacens and Z. Gabelica, in preparation. 6. Z. Gabelica, C. Mayenez, R. Monque, R. Galiasso and G. Giannetto, in: Synthesis of Microporous Materials: Vol. I: "Molecular Sieves" (M.L. Occelli and H. Robson, Eds.), Van Nostrand Reinhold, New York, 1992, 289.



149

CHARACTERISATION OF A CUBIC MESOPOROUS MCM-48 COMPARED TO A HEXAGONAL MCM-41 Raft Schmidta, b , Michael St6cker a and Ole Henrik Ellestada, b a SINTEF, P.O. Box 124 Blindem, N-0314 Oslo, Norway. b University of Oslo, Department of Chemistry, P.O. Box 1033 Blindem, N-0315 Oslo, Norway.

ABSTRACT A cubic MCM-48 and a hexagonal MCM-41 material were synthesised. The pore ordering was confirmed by X-ray powder diffraction and HREM studies. The pore size of the materials was determined to 2.9 nm for MCM-41 and 2.5 nm for MCM-48 by N 2 adsorption and by measuring the freezing depression of water enclosed in the pores by 1H NMR. The self diffusion coefficient of water confined in the pores of MCM-48, determined by 1H NMR spinecho measurements, was found to be significantly lar~er compared to that of MCM-41.29Si MAS NMR showed a significant higher number of Q J species (Si(3OSi)OH) for MCM-48 in the as-synthesised state compared to MCM-41. INTRODUCTION The discovery of a new family of mesoporous materials, denoted as M41S, was recently reported by researchers from the Mobil Oil Co-operation [ 1,2] and has dramatically expanded the range of defined pore sizes ( < 1.3nm) found in crystalline microporous materials (e.g. zeolites) into the mesopore regime ( > 2 nm). High hydrocarbon sorption capacity and high thermal stability are additional interesting properties of these materials. They are synthesised from gels, containing besides silica optionally different other metals (e.g. A1.). The structure of the porous solid is hereby defined by the aggregation of surfactant molecules (such as C16H33(CH3)3NBr) [1]. Already in the first publications [1,2] it was reported that purely siliceous materials with hexagonal ordered pores (MCM-41), layered materials (MCM50) and materials with a three dimensional pore system (MCM-48) could be synthesised by this mechanism, indicating the diversity of this new family of mesoporous materials. However, most of the academic interest following this new invention, was concentrated on MCM-41 [37] and only very few characterisation data of the cubic member of the M41S family (MCM48) have been published [5 - 7], even though the three dimensional pore system of MCM-48 may make this material even more interesting for industrial applications as MCM-41. Here we report on the synthesis of a MCM-48 and a MCM-41 material and the characterisation by

150 XRD, HREM, N 2 adsorption, 29Si MAS NMR and 1H NMR of water enclosed in the pores [8]. Additionally the self diffusion-coefficients of water enclosed in the pores of these mesoporous materials were determined by 1H NMR spin-echo measurements. EXPERIMENTAL A purely siliceous MCM-48 material was synthesised according to the following procedure (similar to the one described by Monnier et al. [5]): 0.7 g of NaOH was dissolved in 17 g of distilled water. Then 7.22 g of tetraethylortosilicate (TEOS) (98%, Jansen) was added to the gel while stirring. After 5 min 27.8 g of the template solution containing 25 wt% C16H33(CH3)3NC1 in distilled water (Fluka) was added. The molar composition of the gel was 1:0.65:62, respectively for TEOS : C16H33(CH3)3NC1 :H20. The resulting gel was then stirred for another 15 min, loaded into a stoppered Teflon bottle and heated without stirring at 100~ for 72 hr. After cooling to room temperature, the resulting solid was recovered by filtration, extensively washed with distilled water, and dried in air at ambient temperature. The template was removed by calcination at 540 ~ for 1 hr in flowing nitrogen followed by 6 hrs in flowing air with flow rates of 100 ml/min, respectively. A siliceous MCM-41 material was synthesised according to a procedure given by Beck et al. using C16H33(CH3)3NBr as template[2] and processed according to procedure described above. The obtained solid materials were characterised by X-ray powder diffraction (Siemens D5000 diffractometer, CuKo~ radiation) and High Resolution Electron Microscopy using a JEOL 2000 FX instrument applying primary magnifications between 68 000X and 210 000X. The TEM samples were prepared by crushing the material in ethanol in an agate mortar and dropping this dispersal of finely grounded material onto a holey carbon film. This produced free sheets of the material suitable for TEM work. Tilting of the specimen for accurate crystal alignment was usually not necessary. N 2 isotherms were measured at 77 K with a Carlo Erba instrument using a conventional volumetric technique. The samples were outgased at 200~ for 1 hr. The mean pore size was calculated from the N 2 isotherm using the Kelvin equation correcting for the mulfilayer thickness. The water-saturated samples were studied by 1H NMR in the temperature range 273 to 183 K using a Varian VXR 300 S NMR spectrometer, operating at 300 MHz proton resonance frequency [8]. All measurements were performed with a 90 ~ pulse of 16 Its. Each spectrum was composed of 16 transients. The samples were temperature equilibrated for 5 min at each temperature before any measurements were performed. The spectra were accumulated applying a temperature cycling, i.e., heating and cooling of the sample. Further T 2 measurements of water enclosed in the pores were performed at 268 K using Carr-Purcell-Meiboom-Gill (CPMG) pulse sequence with varying "cbetween 16 and 400 Its. The self diffusion coefficient of pore water was extracted from the obtained T 2 data using a model proposed by LeDoussal et al. [11]. 29Si MAS NMR spectra were recorded on the same spectrometer, equipped with a Jakobsen high spinning speed MAS probe using 7 mm zirconia rotors. Frequency of 59.6 MHz, sweep width: 1400Hz, pulse width 8 Its (90 ~ degrees pulse: 8.3 ~ts), repetition time: 300 sec, acquisition time: 1 sec, number of scans: 1000, MAS spinning speed: 4.5 kHz. The lines were referenced to the low field signal of ZSM-5 at -109.6 ppm (as a secondary reference).

151

RESULTS AND DISCUSSION

Well ordered MCM-48 and MCM-41 materials, confn'rned by XRD (Figures 1 and 2) and HREM (Figures 3and 4) were prepared.

t

d-value~ hki [4.~] 100 2.3 2.0

,--,

5 c6

d-values 4.19 3.59 2.73 2.48 2.25 2.15 2.06 1.98 1.84 1.63 1.56 1.47

110 200

21o II

1:53~ .

300 220

>, "~ r-"~ t

~

0

2

4

6

8

hid

[nm]

10

Degrees 2-theta

0

2

4

211 220 321 400 420 332 422 431 521 611/532 541 444 543

6

8

10

D~j rees 2-theta

Figure 1. X-ray powder diffraction data of the as-synthesised MCM-41

Figure 2 X-ray powder diffraction data of the as-synthesised MCM-48.

Figure 3. HREM image of as-synthesised MCM-41 in the direction of the pores.

Figure 4. HREM image of as-synthesised MCM-48 with pores on the 110 plane.

152 The X-ray diffractogram of the as-synthesised MCM-48 could be indexed according to a cubic space group with Ia3d symmetry, and a unit cell size of 102 A, which is in agreement with our HREM investigations. A representative electron micrograph of MCM-48 viewed along the [ 110] axis is shown in Figure 4. The unit cell size of the hexagonal MCM-41 was determined to 46/~ (Figure 1) and the hexagonal pore organisation was confirmed by HREM (Figure 3).The pore structure of MCM-48 was generally observed over the whole sheet investigated by HREM, even though small variations e.g. of the pore wall thickness could be registered. These variations result in the displacement of one single pore at the time, and not in the disturbance of the whole local structure. For MCM-41 local variations often result in a disturbance of the (hexagonal) pore organisation over a significant larger area than for MCM48. However, the pore structure of both materials is found to be quite unaffected by the removal of the template by a calcination at 540 deg C. The BET surface area was determined to be 1100 m2/g. and 1080 m2/g. for the MCM-48 and MCM-41, respectively. In accordance with the N 2 isotherms recorded for MCM-41 (Figure 5), the N 2 isotherm of MCM-48 showed a sharp increase in the adsorbed volume of N 2 due to capillary condensation of N 2 in the pores of this material (Figure 5).

~41~ I--v m

E

[

.~400 Figure 5. Nitrogen isotherm at 77 K on MCM-41 and MCM-48. Filled symbols denote adsorption, open symbols denote desorption.

0

E __=200 0 > |

0i 0

i

I

I

I

I

0,2

0,4

0,6

0,8

1

P/Po However, even though both materials were synthesised using C 16H33(CH3)3N+ as template the sharp increase in the adsorbed volume of N 2 due to capillary condensation occurs at different relative pressures P/P0; reflecting different pore sizes of the MCM-41 and MCM-48

153 material. Using the relative pressure (at the inflection point of the capillary condensation) of P/P0 = 0.35 for MCM-41 and P/P0 = 0.26 for MCM-48 and using the Kelvin equation correcting for multilayer thickness, the mean pore sizes were calculated to be 2.9 nm and 2.5 nm for MCM-41 and MCM-48, respectively. The pores of the MCM-41 and MCM-48 material were further characterised by 1H NMR temperature studies of water enclosed in the pores. The 1H NMR spectra of confined pore water of MCM-48 vs. temperature (from 265 to 183 K) are illustrated in Figure 6 and show clearly a sharp intensity decrease with decreasing temperature due to the freezing of pore water.

i

d >,

0

rr" Z "1, ..i. w

280

260

240

i

!

|

220

200

180

Temperature [ K] Figure 6. 1H NMR spectra of water enclosed in the pores of MCM-48 as a function of temperature. A mathematical model [8] was fitted to the 1H NMR signal intensity vs. temperature of water enclosed in MCM-41 and MCM-48 (Figure 7, solid line). Using this model a transition temperature of 224 K reflecting the freezing of the pore water was determined for MCM-48. For the MCM-41 material a transition was observed at 225 K while cooling and 228 K while wanning up. This observed hysteresis in freezing/melting transition (which is not observed for the MCM-48) is due to the larger pore size of the MCM-41 compared to MCM-48 and not due to the destruction of the pores during the freezing/melting process [9]. The effect of the pore size on the occurrence of freezing/melting transition is discussed in detail in a recent publication [10]. Using the freezing point depression (determined by warming up the water saturated samples form 183 K to 273 K, while recording the 1H NMR signal) the pore sizes of MCM-41 and MCM-48 were calculated to 2.9 + 0.08 nm and 2.7 + 0.16 nm, respectively. It was assumed that the observed freezing point depression (AT = 237.15 K - transition temperature) is correlated to the pore size by AT =

K/ R p - tl

9with Kf = 49.5 + 0.19 K*nm-1

154

and tf = 0.35 + 0.036 nm [9]. The pore sizes determined by N 2 adsorption and 1H NMR method are found to be in good agreement.

"L

MCM-41

"'7, :3

t'-

MC 48

Figure 7. 1H NMR signal intensity of pore water confined in MCM-41 and MCM-48. Filled symbols denote heating, open symbols denote cooling of the sample.

%

r

rr"

-

N

Z

"1-

280

260

240

220

200

180

T~rature [ K] The water saturated MCM-41 and MCM-48 samples were further studied by measuring the apparent spin-spin relaxation (T22) of the confined pore water at 268 K. The temperature of 268 K was chosen to ensure that water not enclosed in the pores is frozen and though does not contribute to the measured spin-spin relaxation time. Of particular interest is the dependence of the apparent relaxation rate, 1/T22, on the inter pulse time 2z. This behaviour has been discussed in the literature [11] and arises because of the motional diffusion of the water molecules through the magnetic field gradient created by the susceptibility difference between the solid matrix and the pore water. The apparent relaxation rate (1/T22) extracted from the observed CPMG echo envelope is therefore not only dependent on the inherent spinspin relaxation time of the enclosed pore water (T2), but also strongly affected by the fluid molecules moving (characterised by the diffusion coefficient D) in the inhomogeneous magnetic field within the pore: 1/T22 = 1/3 * ?2DG2z2 + 1/T 2, with ~/ the nuclear magnetogyric ratio, D the self diffusion coefficient of the pore fluid, G the magnetic field gradient G, and the inter pulse spacing z. For a porous medium this equation has to be modified to take into account the motional restrictions imposed by the pore walls. Applying the model derived by LeDoussal et al. [ 11] the self diffusion coefficient of the pore water can

155 be extracted from the measurements of the apparent spin spin relaxation times as a function of x. This is explained in detail for MCM-41 materials in a recent work by Hansen et al. [ 12]. The self diffusion coefficient for water enclosed in the pores was found to be 4.53 + 0.27 10-9 cm2/s for MCM-41 and 2.21 + 0.06 10-8 cm2/s for MCM-48 whereas the self diffusion coefficient (D) of bulk water at 300 K is approximately 2.5 10-5 cm2/s. This demonstrates a strong reduction of the diffusion coefficient of pore water compared to bulk water. In a previous study of MCM-41 materials D was found to decrease with decreasing pore size [12]. However, in this study a significantly higher self diffusion coefficient of pore water was calculated for MCM-48 compared to MCM-41 even though MCM-41 posses larger pores. This suggests a larger mobility of pore water in the three dimensional pore system of MCM-48 compared to the uni-dimensional pore system of MCM-41. However, other effects on D, as e.g. different particle sizes, validity of the model for the calculation of D, etc. may influence the result. This will be a subject of future studies. Through 29Si MAS NMR a significant larger number of Q3 species (Si(3OSi)OH) represented by the peak at around -100 ppm compared to Q4 (Si(4OSi) species represented by the peak around -110 ppm was observed for MCM-48 than for MCM-41 in their as synthesised state (Figure 8). A Q3/Q4 ratio of around 0.45 and 1 was found for MCM-41 and MCM-48, respectively.

-110 -100~"~

41

~ I

I

-50

-70

I

I

~-48 I

I

-90 -110 -130 -150 ppm

Figure 8.29Si MAS NMR spectra of MCM-41 and MCM-48. Resonances are assigned as follows" around - 100 ppm: Q3 _ species; around- 110 ppm" Q4_ species.

156 CONCLUSION Well ordered MCM-48 and MCM-41 were synthesised. The overall pore organisation of MCM-48 is much less affected by local variations (e.g. wall thickness) as for MCM-41. The pore size of the template free MCM-41 was found to be larger (around 2.9 nm) compared to MCM-48 (around 2.5 nm) synthesised with the same cationic template ( C 16H33(CH3)3N+)A good agreement between the determination of the pore size by N 2 adsorption and ~H NMR temperature studies of "pore water" was found. The self diffusion coefficient of water confined in the pores of MCM-48, determined by 1H NMR spin-echo measurements, was found to be significantly larger compared to that of MCM-41. A significant higher number of Q3 species (Si(3OSi)OH) was found for MCM-48 compared to MCM-41 in their assynthesised state. ACKNOWLEDGEMENTS A research grant from the Norwegian Research Council (Deminex program) is gratefully acknowledged by R.S.. The authors are indebted to E.H. TCrstad and A.Olsen for preparing the HREM micrographs and E. Hansen for recording the 1H NMR spectra and D. Akporiaye for fruitful discussions. REFERENCES 1. C.T. Kresge, M.E. Leonowicz, W.J. Roth, J.C. Vartuli, J.S.Beck, Nature 359 (1992) 710. 2. J.S. Beck, J.C. Vartuli, W.J. Roth, M.E. Leonowicz, C.T. Kresge, K.D. Schmitt, C.T-U. Chu, D.H. Olsen, E.W. Sheppard, S.B. McCullen, J.B. Higgins, J.L. Schlenker, J.Am. Chem. Soc., 114 (1992) 10834. 3. C.Y. Chen, H.X. Li and M.E. Davis, Microporous Materials 2 (1993) 17. 4. R. Schmidt, D. Akporiaye, M. St/Scker and O.H. Ellestad, J. Chem. Soc., Chem. Commun.(1994) 1493. 5. A. Monnier, F. Schiith, Q. Huo, D. Kumar, D. Margolese, R.S. Maxwell, G.D. Stucky, M. Krishnamurty, P.;Petroff, A. Firouzi, M. Janicke and B.F. Chemelka, Science 261 (1993) 1299. 6. J.C. Vartuli, K.D. Schmitt, C.T. Kresge, W.J. Roth, M.E. Leonowicz, S.B. McCullen, S.D. Hellring, J.S. Beck, J.L. Schlenker, D.H. Olsen and E.W. Sheppard, Chem. Mater. 6 (1994) 2317. 7. Q. Huo, D.I. Margolese, U. Ciesla, P. Feng, T.E. Gier, P. Sieger, R. Leon, P.M. Petroff, F. Schiith and G.D. Stucky, Nature 368 (1994) 317. 8. D. Akporiaye, E.W.Hansen, R. Schmidt and M. St/Scker. J. Phys. Chem., 98 (1994) 1926. 9. R. Schmidt, E.W. Hansen, M. St/Scker, D. Akporiaye and O.H. Ellestad, J. Am. Chem. Soc. in press. 10. R. Schmidt, M. St6cker, E.W. Hansen, D. Akporiaye and O.H. Ellestad, Microporous Materials 3 (1995) 443. 11. P. LeDoussal and P.N. Sen, Phys.Rev.B. 46 (1992) 3465 12. E.W. Hansen, R. Schmidt, M. St6cker and D. Akporiaye, Microporous Materials, submitted.


157

Synthesis and Characterization of Transition Metal Containing Mesoporous Silicas S. Gontier and A. Tuel Institut de Recherches sur la Catalyse. C.N.R.S. 2, av. A. Einstein 69626 Villeurbanne Cedex France Titanium and Vanadium mesoporous silicas (MS) have been synthesized at ambient temperature using a neutral templating route with primary alkyl amines as surfactant. Spectroscopic characterization of the samples showed that transition metal cations were highly dispersed in the silica framework. The metal content, the surfactant chain length and the amine/SiO 2 ratio greatly influenced the properties of the final product. These materials were found to be active as catalysts in oxidation reactions with alkyl peroxides at mild temperature. 1. I N T R O D U C T I O N There has been an extensive interest during the last years for the synthesis of transition metal containing molecular sieves, due to their remarkable properties as catalysts in oxidation reactions with organic peroxides [1,2]. A very beautiful example is TS-1, the titanium substituted silicalite-1, that catalyzes many oxidation reactions in the liquid phase with aqueous hydrogen peroxide. However, as far as zeolitic supports are concerned, reactions are limited to small substrates with a kinetic diameter smaller than 7~. Recently, the discovery of a novel family of silica-based mesoporous molecular sieves M41S by Mobil researchers [3] and a group from Waseda University [4] opens new perspectives in the field. Corma et al. [5]

were the first to synthesize a Ti

mesoporous silica analog to MCM-41 and to show that the material was active as catalyst in the oxidation of bulky substrates. In a similar manner, Franke et al. [6] also prepared Ti-MCM-41, but they did not report any catalytic data over these solids. Reddy et al. [7] also reported the possibility of preparing vanadium-containing MCM-41, active in oxidation reactions.

158 Very recently, Tanev et al. [8] have shown that mesoporous silicas (MS) could be prepared using a neutral templating route with primary alkyl amines as surfactant. A great advantage of the recipe was that the template could be removed from the mesopores by a solvent extraction, which is of great interest for environmental protection. We have followed a similar procedure to prepare Ti and V-containing mesoporous silicas. The influence of synthesis parameters, e.g. the metal content, the amine chain length or the synthesis time have been examined. These materials have shown very interesting properties as catalysts in oxidation reactions with organic peroxides as compared to zeolitic materials. 2. E X P E R I M E N T A L Ti or V-MS were prepared at ambient temperature by mixing a first solution containing tetraethyl orthosilicate (1 mole), ethanol (6.5 moles), isopropyl alcohol (1 mole) and the metal precursor (tetrabutyl orthotitanate or vanadyl acetylacetonate) to a second solution containing the alkylamine (0.3 mole) in water (36 moles). The resulting solution was homogeneized, stirred for about 30 min and aged for different periods at room temperature under static conditions. Solids were then filtered, washed several times with distilled water and air dried. Calcination of the samples was performed at 650 ~C in air for 6 h. When the organics were removed by a solvent extraction, 1 g of dried solid was dispersed in 100 ml of ethanol and the mixture refluxed for about 1 h. The solid was recovered by filtration, washed with cold ethanol and the procedure repeated once. A Ti-Beta sample was synthesized following the recipe of Camblor et al. [9] and contained about 1.5 wt % Ti and 0.4 wt % AI. A TS-1 sample was also synthesized following the patent literature [10]. Samples have been characterized using X-Ray diffraction (Philips PW 1710, CuKa radiation), IR spectroscopy (Perkin Elmer 580), U.V-Vis spectroscopy (Perkin Elmer Lambda 9) and EPR (Varian, E9). N 2 adsorption/desorption isotherms were carried out on a Catasorb apparatus. 3. R E S U L T S AND DISCUSSION

3.1. Synthesis and Characterization A series of samples have been synthesized with dodecylamine (n = 12 carbon atoms) and varying the amount of metal precursor in the gel. Gels were aged for 12 h

159 before recovering as-synthesized solids by filtration. As shown in Table i, the amount of metal incorporated in the silica matrix is very close to that introduced during the preparation. Table i Characterization of the different samples Sample Gel .

.

M~

oc

Si/Ti Product .

S(m2/g) .

m

.

973

r

V(cm3/g)

.

.

26.5

0.52

Ti-MS

i0()

85

i086

28

0.62

Ti-MS

50

45

i 066

29

0.65

I1-M~

30

3i

894

28

0.54

Ti-MS

20

i 9.5

667

25

0.36

V-Mb

i00

108

998

28

0.55

V-MS

50

62

i0 i5

28

0.64

V-MS

30

36

794

28.5

0.48

9 p()~) is the mean pore diameter V(cm3/g) is the porous volume measured at P/P0 = 0.5 in the N 2 isotherm. This suggests that all the Si and Me precursors were in the solid phase as the yield in Ti or V-MS was always very high (> 95 %). For relatively low Me contents (< 2 wt %), U.V-Vis as well as EPR spectroscopies showed that the cations were highly dispersed in the solid. Ti-MS materials exhibited a U.V. absorption band around 240 nm. The absorption edge decreased by about 10 nm to 230 nm upon calcination in air. For V-MS, EPR parameters were characteristic of vanadyl ions in an axially symmetric crystal field (A ![ = 190 G, g [] = 1.94, A__k_ = 72 G and gA_ = 2.00). The totality of the EPR signal disappeared upon calcination of V-MS samples in air. Dried calcined samples were white, but their color rapidly turned to bright yellow upon exposure to air, suggesting a change in the cation coordination. This was clearly evidenced in U.V-Vis and 51V NMR spectra. In dry samples, V 5 + cations are more likely in a tetrahedral environment but water molecules can easily enter their coordination sphere to give hexacoordinated cations. The process was perfectly reversible as original U.V-Vis and NMR spectra were restored after outgasing the samples at 200~

for 3 h. This clearly showed the relatively

160 high hydrophilic character of mesoporous silicas as compared to substituted zeolites like TS-1 or VS-1. All calcined samples exhibited relatively high surface areas, typically 1000-1200 m2/g and a mean pore diameter close to 28/~. Both slightly increased with the metal incorporation up to about 2 wt % of metal in the solid. Beyond that limit, for higher Me contents, the surface area as well as the pore diameter decreased. In the same time, U.VVis and NMR spectra revealed the presence of extrawall dispersed oxide species in the samples. However, the decrease in pore diameter could also arise from a loss of thermal stability of the structure for high Me incorporations. Similar observations have been made by Franke et al. [6] for Ti-MCM-41, and the authors interpreted the decrease in pore dimensions to the presence of extrawall species. A sample has been prepared with Si/Ti = 100 in the precursor gel and aged at room temperature for different periods ranging from a few minutes to 18 h. Indeed, Chen et al. [11] had reported that even though an X-Ray powder pattern typical of MCM-41 could be obtained after heating a silica-alumina gel at 70~

for 3 h, the

material was not thermally stable and collapsed upon calcination in air at 540 ~C. Table 2 clearly shows that using the present synthesis route, Ti-MS was obtained after a few minutes after mixing of all the reagents. There was no significant evidence for modifications occuring during the aging period ; the Ti content as well as the BET surface area and the pore dimension remained unchanged. The same observations could be made with vanadium-containing samples. Table 2 Characterization of Ti-MS samples recovered at different time intervals t(h)

Si/Ti

S(m2/g)

~p()~)

0

93

1081

28

0.58

1

85

1212

28

0.62

2

96

1091

28

0.60

3

94

1112

28

0.63

6

92

1174

28

0.62

18

85

1085

28

0.62

For the definition of ~p and V, see Table 1

V(cm3/g)

161 it has been widely reported that the pore dimension of MCM-4i synthesized with aikyitrimethyiammonium cations depended on the number of carbon atoms of the alkyi chain. For pure silica materials, data from Beck et ai. [3] showed an almost linear increase in the pore diameter from i8 ~ with octyiamine to about 37 /~ with hexadecyiamine. Very recently, Tanev et ai. [8] also reported that the pore dimensions of mesoporous silicas prepared using a neutral templating route increased with the amine chain length. For both Ti and V-substituted materials, we also observed that the dimension of the mesopores increased from about 25 ~ for n = i0 carbon atoms to 35 for n = i6. The corresponding isotherms are shown in Fig. i.

(c) o') A

I...03

1I~176

/

(b)

E "6 ..Q L_

o

Ca)

"13
60 wt. % tetralin) that zeolites exclude. The open structure and large pore size also result in high rates of adsorption and desorption. The superior adsorption properties of these new materials indicate their great potential for applications in adsorption and catalysis. Large-scale production of the mesosieve materials is expected to be straightforward and economical. INTRODUCTION There is a large and growing demand for adsorbent materials for use in various industries. The major commercial adsorbent materials on the market today are aluminas, molecular sieve zeolites (A, X, and mordenite), silica gels, carbons, and organic polymers. Compared with the other adsorbents, zeolite molecular sieves have significantly higher sorption capacities at low partial pressures. The maximum sorptive capacity of zeolite 13X which has the highest capacity among zeolites, is 25 to 36% [1]. A key determinant of the adsorption properties of a porous solid is its microstructure. A porous solid with high adsorption strength at low partial pressure, high capacity at high partial pressure and with rapid adsorption and desorption rates would be an ideal material for adsorption applications. CANMET has recently developed a novel method to make porous material with just these features [2]. The water sorption capacity of this new material is about 120 wt. %, which is 200 to 300 % higher than most of the commercial adsorbents currently on the market. The preparation and projected large scale manufacturing procedures are simple since only ambient

166 conditions are involved in the original preparations. In this paper, the adsorptive properties of the uniquely structured materials and some characterizations of their micro structures are presented. MATERIAL CHARACTERIZATION Researchers in Mobil R&D corporation had developed crystalline mesoporous molecular sieves designated as M41S using high temperature (100 to 200 ~ hydrothermal conditions (4,5). The present materials were all prepared using ambient conditions (0 to 25 ~ and atmospheric pressure) and short times (5 minutes to 4 hours) followed by a thermal treatment procedure. Silicate and aluminate sources common in zeolite synthesis and commercial surfactants such as Ammonyx KP (oleyl dimethyl benzyl ammonium chloride, Stepan Co., IN) were used in the preparations. The details of the preparation procedure and compositions are not described here. The major features of these materials are high porosity, low particle density, high surface area, and high pore volume. The surface area (S) and pore volume (V) of the materials, as well as the relative amounts of micropores and mesopores can be varied as required (from 100 to 1200 m2/g for S, and from 0.1 to 1.3 cm3/g for V) by controlling the original mixture composition. Commercial manufacturing of the material in a continuous mode is easily achievable because of the simple procedures required: ambient conditions, a short amount of time, and a final thermal treatment in a regular furnace. Figures 1 shows a SEM (Scanning Electron Microscopy) micrograph of a typical

Figure 1. A SEM micrograph of the mesoporous materials.

Figure 2. A TEM mesoporous particle.

image

of

167

sample of the new materials. It is clear that the material is composed of small particles with diameter of less than 0.5 micron. These small particles themselves are highly porous as indicated by their high surface area. Figure 2 is a TEM (Transmission Electron Microscopy) image of the same sample. It is evident that different from most zeolites and crystalline mesoporous sieves such as MCM-41 which has a well organized hexagonal pore structure, the present materials contain randomly distributed channels with uniform mesopore size (about 3 nm for this particular sample). The micropores are not visible on the TEM image but are believed to be in the walls of the mesopores. As can be seen from the pore size distribution from nitrogen adsorption in Figure 4, in the mesoporous range, most of the pore volume is contributed by pores with a diameter of 3 nm. Comparing the results of figures 2 and 4, there is a good agreement on the channel size of the material between the two independent characterization methods. This agreement implies that the new materials, which are not really crystalline, have randomly distributed channels with a uniform mesopore diameter of 3 nm. The name mesosieve is used for these new materials in the rest of the paper. In addition to the regular mesopore size, the materials also have very large surface area and large pore volume. Table 1 lists some typical physical properties of the materials compared with zeolite 13X. The framework density was measured using helium and the particle density (or mercury density) is calculated from the framework density and N 2 saturation capacity. The materials are stable under severe acidic and thermal conditions; there was no pore structure change and only a 3 % surface area decrease was observed after a sample was placed in 70% nitric acid for five days and then calcined at 510 ~ Calcination at 800 ~ for two hours would not significantly change the pore size distribution pattern although a less than 20% decrease in surface area was observed. Table 1. Some physical properties of a mesosieve and zeolite 13 X Sample ID

Skeletal density (g/cm3)

Particle density (g/cm3)

Surface area (m2/g)

Total pore volume (cm3/g)

Mesosieve 11195A

2.39

0.63

1075

1.36

Zeolite 13X

2.03

1.31

790

0.32

ADSORPTION P R O P E R T I E S OF THE MATERIALS Figure 3 compares an N2 adsorption isotherm of one sample of the new materials (Figure 3a) with that of zeolite 13 X (Figure 3b). As is evident from the figure, the new mesosieve exhibited a type IV isotherm and 13X a type I isotherm. At a relative pressure of 0.001, 85 % of 13X total capacity is occupied because of the strong adsorption potential in

168

1

i

i

i

l

i

+

I

T

................ i .......I........I....... I........t.......I........t.......T

!" - I :

'dt

~ 'i

........ ,........ ! ~l....... .+........ , , , + / ~ ~ ~ ,

+

!........ l..-,..+,.

++B} + ............... i +....... i ........ +....... +........ +....... i ........ l

....... !........ t ....... j .... //,,.. ..... l ........ l ....... +........ +....... +........ i ..............

I I ........ +!...... ~ ,

! ,P" i

i

~

l ~ ~

i

i

i

l

i

~

+....... Ii...,

'~"i

. . . . . . . .

....... +........ i....... + v o = ] _ p

'

'

;

"

~ S

I

I

i

i

l d,, = 28.

%

o

./-----i ........ i ....... i ........ i ....... i-~

:

,

%

, o.,

,

,

=

,

o

%

\~2 ~

,

, , ,

%

%

.

:.

I. . . . . . .

t+ .t....t .

--

:j!io,-.o,,~

I I"]:

%

"o

~v

,.,=--,--!

-.,

o

%

l

l

i

Zeolite 13X:i .... , ....

.......

+Vp

=

0.32

l

..+....... +s=798

r=o,,./

+'-'L--'-~--,., -o

+

l

i

L , o L I ! L ! ! ! ! %

l

'[ .......]

,

.... l---t.J...

.

-+ ....... +........ +....... i ........ +....

!

.... il

1235

=

,,

. ;7 ...,.. 7"i 711 IT 71i+.~ . ; . '

!

.....

".

.......................................

:

i

-,....

l ....... ] ........ +....... +........ +....... +........ l:~,..~i

........ ~

~I

.

,~.......j_..

%

+....

i

i do = 5 -

+

+

+

+

%

%

%

%

7{

+ ! 60 w. %). The materials also show greater rates of adsorption and desorption than zeolites. The unique pore structure of the materials indicate great potential for applications in adsorption and catalysis, particularly for applications that require 1) large pores and high surface area and 2) both micropores and mesopores, small mass transfer resistance and large pore volumes. REFERENCES

.

.

.

.

7.

R. H. Perry and D. H. Green (eds.), Perry's Chemical Engineer's Handbook, McGraw-Hill, Inc., 16-9, 1984. C. J. Guo and C. Fairbridge, CANMET division report WRC 93-48 (CF), November, 1993. C. T. Kresge, M. E. Leonowicz, W. J. Roth, J. C. Vartuli and J. S. Beck, Nature, 359 (1992) 710. J. S. Beck, C. T-W. Chu, I. D. Johnson, C. T. Kresge, M. E. Lex~nowicz, W. J. Roth and J. C. Vartuli, US Patent 5108725, April, 1992. G. Horvath and K. Kawozoe, J. Chem. Eng. Jpn., 16 (1983) 470. J. P. Olivier and W. B. Conklin, "The international Symposium on the effects of surface heterogeneity in adsorption and catalysis on solids", Kazimierz Dolny, Poland, July, 1992.


Zeolites: A Refined Tool for Designing Catalytic Sites L. Bonneviot and S. Kaliaguine (editors) 1995 Elsevier Science B.V.

173

M C M - 4 1 t y p e silicas as s u p p o r t s for i m m o b i l i z e d c a t a l y s t s Daniel Brunel*, Anne Cauvel, Francois Fajula and Francesco DiRenzo Laboratoire de Mat~riaux Catalytiques et Catalyse en Chimie Organique, CNRS-URA 418, Ecole Nationale Sup6rieure de Chimie, 8, rue de l'Ecole Normale, F-34053 - Montpellier C6dex 1 - Fax: + 33 - 67 14 43 49 Abstract MCM-41 type silicas were covalently grafted with various functional alkoxysilanes (RO)3Si(CH2)3 X with X = C1, NH(CH2)2NH2 and NHC(O)NSalpr. The functions of the first two organic moieties already attached were further transformed into respectively 2-(NHCH2)Pyr, 4-(NH(CH2)2NHCO)Pyr also known as organic ligands of transition metals. The grafted mesoporous silicas were c h a r a c t e r i z e d by IR, MAS-NMR s p e c t r o s c o p i e s , thermogravimetry, nitrogen adsorption and elemental analysis . The coupling reactions of organic ligands do not modify the covalent grafting bonds. Pore opening shrinks at increasing organic coverage. The diameter of the channels accessible to nitrogen varies from 33 to 13 /~. A total surface lining by concentrically oriented organic chains is suggested by the decrease in the nitrogen adsorption enthalpy as a function of the organic coverage. 1. INTRODUCTION MCM-41 molecular sieves are a new class of mesoporous aluminosilicates featuring cylindrical regular mesopores of monodispersed diameter w i t h potential applications in catalysis and adsorption [1,2]. They are obtained by precipitation of amorphous silica-alumina in the presence of cationic surfactants. The pores of the MCM-41 type materials are templated by the surfactant micelles. While the hydrocarbon chain length of the surfactant rules the pore diameter, the conditions of the hydrothermal synthesis, in particular the alkalinity of the parent hydrogel, influence the wall thickness [3]. An increase in the pore spacing leads to an improved thermal stability of MCM-41-type materials. Pure silica MCM-41s can also be prepared and they feature a better stability than their silica-alumina analogs. MCM-41-type silicas were obtained with a pore diameter of 33/~ and wall thickness higher than 10/~. MCM-41 type silicas are an ideal raw material for the design of new mesoporous solids having adjustable chemical and physicochemical properties [4] in view of applications in the field of adsorption and fine chemical catalysis. This objective can be reached by lining of the mesopore surface with covalently attached transition metal liganding moieties. The volume available inside the MCM-41 mesopores is much larger than in the channel volume of the usual

174 zeolite catalysts. The insertion and grafting of bulky, sterically-hindered functional molecules is hence possible, without a decrease of the accessibility of the catalytic groups. As the mesoporous surface of the MCM-41 type silicas presents the same surface silanol groups than the traditional amorphous silicas, it can be modified by applying usual methods of silica functionalization. They deal with the covalent linkage of organic groups to the silica surface by condensation of organotrialkoxysilanes with the silanol groups [5-22]. We report here the anchoring of organic groups to MCM-41 type silicas using functional propyltrialkoxysilanes. The organic moieties were f u r t h e r transformed by coupling reactions with suitable organic ligands of transition metals. Our goal in preparing these materials was to obtain mesoporous solidbound ligands that yield systems with potential for catalytic oxidation of organic substrates. As most of the monodentate ligands used to immobilize transition-metal catalysts suffer from the disadvantage of the metal complex leaching into the solution, there is considerable interest in covalently attaching multidentate chelating agent to a support surface. For this purpose, we selected the following l i g a n d s : (pyridino-2-methyl)amine, (2isonicotinamidoethyl)amine and pentadentate Schiff base as SalDPT [23,24]. In the case of the two first ligands, mild coupling organic reactions were carried out to perform the coupling of the liganding moieties while avoiding alteration of the covalent grafting bonds. The different modifications were controlled at each step using various characterization techniques. This study allowed the determination of the new adsorption properties of the modified mesopore surfaces. 2. E X P ~ A L 2.1. MCM-41 silica.

Pure MCM-41 mesoporous silica (MPS) was prepared in the presence of cetyltrimethylammonium hydroxyde according to the method described in the literature [1,2]. The MPS material consists in SiO2 containing 0.003 A1/(AI+Si) and 0.0014 Na/(AI+Si). X-ray powder diffraction pattern matches well with the patterns reported by J.S. Beck et al [2], with d loo = 40.0/~. The hexagonal lattice parameter is equal to 46.2 /~. The nitrogen sorption isotherm of MPS is a classical example of type IV isotherm [24], showing monolayer and multilayer adsorption on a mesoporous surface of very high area and a sharp, reversible step at P/P0 0.38, characteristic of capillary condensation within the mesopores (vide infra). No hysteresis loop is observed due to the low filling pressure of the small mesopores.

2.2. Functionalization procedure 3-chloropropylsilylated-MPS (C1-MPS), 3-(2-aminoethyl)aminopropylsilylated -MPS (NH2EtN-MPS) and 3-(bis-[3-(salicyliden-amino) propyl])carbamatopropyl silylated-MPS (SalDPT-amido-MPS). A suspension of freshly activated MCM-41 silica in toluene was refluxed and stirred for 1.5 hr. with the corresponding functional organoalkoxysilane under dry nitrogen atmosphere. After distillation in a Dean-Stark collector of a fraction of toluene containing volatile compounds, the mixture was again heated at toluene refluxing temperature for 1.5 hr. The distillation and heating sequences were repeated. After cooling, the solid was extracted in a soxhlet

175 apparatus overnight with ethyl ether and dichloromethane, then evacuated under vacuum at 200~ for 6 hr. Elemental analy~i~: C1-MPS: C 6.34%, C1 3,29%; NH2EtN-MPS" C 15,23%, N: 6.06%; SalDPT-amido-MPS: C 26.36%, N 5.17%.

2.3. Modification of the functional group 2.3.1.3-(2-pyridinomethyl)aminopropylsilylated-MPS (Pyr MeN-MPS) A suspension of activated C1-MPS material in toluene was refluxed and stirred in an excess of 2-(aminoethyl)pyridine for 6 hr. After separation, the modified solid was extracted according to the previous procedure, then dried under vacuum at 180~ overnight. Elemental analysis: C 14.62%, N 3.89% 2.3.2. 3-(2-isonicotinamidoethyl)aminopropylsilylated-MPS (IsoNicEtN-MPS) Activated NH2EtNH-MPS in suspension in dichloromethane was refluxed and stirred with an excess of a mixture (1/2) of isonicotinic acid and N,N'dicyclohexyl carbodiimine for 6 hr. The modified solid was then separated, washed in succession with water, ethanol, then treated according to the previous procedure. Elemental anolysi~: C 23.05%, N 6.98%. 3.3. Characterization Analyses of the modified solids were made using 13C MAS-NMR, Infrared and UV Spectroscopies, Thermogravimetry, Nitrogen Adsorption and Desorption Isotherm and Elemental Analyses. RESULTS AND DISCUSSION The grafting reactions on the MPS support were studied first by infrared and 13C MAS-NMR spectroscopies. Infrared spectra of the modified mesoporous silicas show a band at 2940 cm -1 characteristic of-CH2- stretching vibration whereas the band at 2970 cm-1 associated with the CH3 of the ethoxy group has a much lower intensity than in the grafting agent spectrum. Moreover the grafted MPS spectra exhibit lower silanol band intensity at 3741 cm -1 than the parent MPS. The release of ethanol, detected and characterized by GCMS during the process confirms the formation of the Si-O-Si linkage according to the following reaction scheme: ~-OH

,,, (EtO)3Si(CH2)3X

~

NN[_O\ S i / ~ / ~ X]---O/

X

+ 2 EtOH

X= CI, NH(CH2)2NH 2, NHC(O)SalDPT 13C MAS-NMR spectroscopy allows the determination of the structure of the grafted hydrocarbon chains. The assignments of the 13C NMR signals of the modified mesoporous silicas are reported in Table 1.

176 Table 1 13C CP-MAS NMR spectral features of the grafted mesoporous silicas Grafted species

Ca

C~

Cy

N(CH2)nN Aromatic C

-=SiCH2aCH2~CH2Y-CI

8.9

26

46

=_Si(CH2)3-NH(CH2)2-NH2

9.5 22.5

38

51

-Si(CH2)3-NHC(O)-SalDPT

9.5

22

40

45

118 131

-_-Si(CH2)3-NHCH2Pyr

10

19

38.5

52

125 138 149

=_Si(CH2)3-NH(CH2)2-NHC(O)Pyr 10

22

40

50

124

ppm

145

CO,CN

161

158

163

165

Thus, the eventual modification of the structure of the function can be monitored by the change in the 13C CP-MAS-NMR spectra resulting from the coupling reactions. This is examplified on Figure 1 showing the intensity of the signal assigned to the CH2-C1 group on the C1-MPS spectrum (Fig.la) at 47.2 ppm which is very low on the spectrum of the resulting Pyr MeNH-MPS (Fig. lb). Moreover this later spectrum exhibits other signals characteristic of the 2-pyridinomethylamino group. The disappearance of the residual ethyl arms during the coupling t r e a t m e n t probably results from nucleophilic assistance of siloxane coupling between adjacent chains.

=

S i ( O E t )

( C H 2 )

••-•• 3-CI

~.\ \ i ~ ~

-(CH1)3 - N I I - C I I .

1~o

.

_

,

.

.

1

-

,

6

--~

-

.

|

-

14o

.

.

,

- . -

,oo

,

-

.-

,

60

-

. - !

.

.

.

.

.

,

iO

PPM

Figure 1. 13C CP MAS-NMR of a)C1-MPS b) PyrMeN-MPS

PPM

Figure 2. a)13C CP MAS-NMR of SalDPT-amido-MPS. b) J-Mod. 13C NMR of SalDPT in CDC13 solution

On the other hand, Figure 2 illustrates the identification of the anchored

177 26,0 ~ SalDPT group attached with c a r b a m a t o p r o p y l silane chain (Fig 2a) by comparison with the s p e c t r u m of SalD PT zo.( material in CDC13 solution (Fig 2b). The diffuse r e f l e c t a n c e s p e c t r u m of SalD PT-amido-MPS presented on figure 3 is consistent with the s t r u c t u r e of the salen group linked by a carbamate function (~,= 254 14.( and 317 nm). The b a n d at ~,= 400 n m would K-M be c o n s i s t e n t w i t h excitonic t r a n s i t i o n r e s u l t i n g from a t i g h t ordered packing of 8. salen molecule. The i n f r a r e d s p e c t r u m of this solid is also in a good a g r e e m e n t with the proposed structure. This latter technique as well as 13C MAS-NMR spectroscopy also z. 200 300 ~0 500 indicate t h a t the coupling of organic ligands zi nm of t r a n s i t i o n m e t a l to the grafted molecules Figure 3. Diffuse reflectance spectrum does not a l t e r the covalent siloxane bonds of SalDPT-amido - MPS. between silica and organic moieties. Table 2 reports the organic content and coverage of the grafted mesoporous solids deduced from the elemental analysis and thermogravimetry.

Table 2. Organic coverage and content of modified MPS.

Materials Cl-MPS NH2EtNH-MPS PyrMN-MPS

Molar ratio C/Cl 6.1

chain nm 2. 0.6

10.3

-(ctl2) 3 -Nll-(Cll2)2-Ntt 2

C/N 2.7

1.3

23.5

"(C!]2)3 "NI]'CIt2 --~(,_))

C/N 4.4

0.9

26.7

C/N 3.8

1.0

33.5

C/N 5.9

0.6

40.7

Organic chain -(oil2)3 -ct

{or~) k SiO2

w

-(Cll2) 3 -NII-(CII2)2-Nlt

IsoNicEtN-MPS

SalDPT-amidoMPS

~,-~

oII -((;lt2)3 -Nn-C -N

llO

* surface area of the mineral support The organic coverage of the modified m a t e r i a l was c a l c u l a t e d from elemental analysis d a t a t a k i n g into account mass % chlorine for chloroalkyl

178 chain and mass % nitrogen for amino or amidoalkyl chain containing or not pyridino group. The organic content is obtained from thermogravimetric data as the mass loss at T > 200~ It is noticeable that the organic coverage is lower for C1-MPS than NH2EtNH-MPS. Probably the amino groups enhance the substitution reaction at the Si atom [25]. The low organic coverage of C1-MPS corresponds to a C/C1 ratio higher than the stoichometric value. It is likely that the C/C1 ratio is increased by the presence of EtO- groups linked to the Si atoms belonging to the grafted chain or to the surface. This hypothesis is consistent with the i.r. and 13C NMR results. It is noteworthy that C/N ratio confirms a complete coupling reaction whatever the previously grafted organic function. The nitrogen sorption isotherms of the functionalized MPS give informations on their texture and surface state. Figure 4 shows the isotherms of some materials grafted with chains of different lengths. Data derived from the sorption isotherms of all samples are reported in Table 3.

jf

8O0

n_

~=m600

j

v

w= 4 0 0 o

/

~

c 200 _1 o

.,,,..,.,.,,..~l~ ~ .

. -

450

400

300

l

200

100

0

,,~

o

~,

'

a.'2

]',, RS~..ATIV[

'

9

o'.6 PRESSURE

o'.~ ,

'

'

~

o

o

(J=/;=o)

.

0.2

.

.

o

u

Ri[LPTIV~

.

o

6

PRESSURE

o.~. ,

(P/Pro)

Fig.4. Nitrogen sorption isotherms of a) parent MPS b) (1) CI-, (2) PyrMN-, (3) SalDPTamido-, (4) IsoNicEtN-MPS. Table 3. Textural and thermodynamic parameters deduced from nitrogen sorption isotherms mesoporous mesoporous diameter BET Materials surface volume (cc/g) (A) parameter C (m2/g) MPS

920

0.76

33

100

C1-MPS

851

0.57

27

60

PyrMN-MPS

577

0.35

24

50

NH2EtNH-MPS

586

0.29

20

22

SalDPT-amido-MPS

608

0.31

20

19

isoNicEtN-MPS

436

0.22

13

28

179 The surface area and mesopore volume decrease with increasing length of the grafted chain, but the characteristic features of the MPS isotherms are essentially preserved. All isotherms are of type IV, indicating the preservation of the mesoporous system during the grafting reaction or the modification of the organic chain. However, the characteristic step of the sorption isotherm, corresponding to the Kelvin filling of the mesoporosity, becomes less sharp at increasing organic content, suggesting that the pore size distribution is widening. The pressure at which the step of the isotherm occurs decreases with increasing length of the organic molecules. The diameter of the organiclined mesopores can be evaluated by the ratio dmeso = 4V / S between mesopore volume and mesoporous surface area. The average pore diameter are reported in table 3. The pore opening shrinkage fits fairly well with the increase in the organic chain lengths. This trend is consistent with the variation of surface area and mesoporous volume. The application of the BET equation to the nitrogen sorption isotherms provides some useful information on the energy of the interaction between a probe molecule -nitrogen- and the surface. The BET parameter C is roughly connected to the nitrogen adsorption heat according to the following equation: C= exp[ (Eads- E1 ) /RT] where E1 is equal to liquefaction heat of nitrogen. Values of the parameter C for all samples are reported in Table 3. All organic-lined MPS feature much smaller values of the parameters C than the parent silica. The adsorption heat of nitrogen is indeed lower on organic surfaces than on hydroxyl-rich surfaces. As a consequence, the low C values for modified MPS can be considered as an indication of regular surface coverage by the organic groups. 4. CONCLUSION The MCM-41 type silicas provide suitable supports for anchoring organic moieties to the inner surface of a mesoporous system. The organic lined mesopores of regular, well-controlled diameter can be obtained by standard functionalization methods. The grafted function can be further transformed into well-defined transition metal liganding moities without loss of either the organic chain content or the regular mesoporous structure. Nitrogen sorption isotherms are useful tools to characterize the properties of such potentially hydrophobic materials. The strong decrease of the nitrogen adsorption heat with the organic coverage is a significant test of the modification of the surface properties. The decrease of the value of the C parameter of the BET equation indicates t h a t the mineral surface is no longer accessible to adsorbed molecules. These results confirm the unique properties of the composite materials prepared by grafting organic molecules to the inner surface of MCM-41 type silicas. Potential applications would be relevant to both specific oxidative catalysis induced by the anchored transition metal complexes and the hydrophobic property peculiar of the lined organic moieties. AI~OWLEDG~ The authors are grateful to ELF and CNRS for financial support. They thank Annie Finiels for her help in taking part of the 13C NMR spectra in

180 CDC13 solution. Anne Cauvel is indebted to ADEME (Agence de 1' Environnement et de la Maitrise de r Energie) for a doctoral grant. R~'ERENCI~ 1 J.C. Beck, C.T.-W. Chu, I.D. Johnson, C.T. Kresge, M.E. Leonowicz, W.J. Roth and J.C. Vartuli, to Mobil Oil Corporation, WO91/11390 (1991). 2 J.S. Beck, J.C. Vartuli, W.J. Roth, M.E. Leonowicz, C.T. Kresge, K.D. Schmitt, C.T.-W. Chu, D.H. Olson, E.W. Sheppard, S.B. McCuUen, J.B. Higgins and .L. Schlenker, J. Am. Chem. Soc., 114, 10834 (1992). 3 N. Coustel, F. Di Renzo and F. Fajula, J. Chem. Comm., Chem. Comm., 967 (1994). 4 A. Cauvel, D. Brunel, F. DiRenzo and F. Fajula, "Organic Coatings" 53rd International Meeting of Physical Chemistry, Paris, 2-6 jan. 1995. 5 W. Hertl, J. Phys. Chem., 72, 1248 (1968). 6 B. Arkles, Chemtech, 766 (1977). 7 D.R. Fruge, G.D. Fong and F.K. Fong, J. Am. Chem. Soc., i01, 3697 (1979). 8 L. Yu Fu, X. Yong-Xia, X. Don-Peng and L.J. Guang-Liang, J. Polym. Sci., 19, 3069 (1981). 9 D.W. Sindorf an G. Maciel, J. Am. Chem. Soc., 105, 3767 (1983). 10 E.J.R. SudhSlter, R. Huis, G.R. Hays and N.C.M. Alma, J. Coll. Interf. Sci., 103, 554 (1985). 11 R. Rosset, Bull. Soc. Chim. Fr., 1128 (1985). 12 W.H. Pirkle, T.C. Pochapsky, G.S. Mahler, D.E. Corey, D.S.Reno and D.M. Alessi, J. Org. Chem., 51, 4991 (1986). 13 J.W. De Haan, H.M. Van den Bogaert, J.J. Ponjed and L.J.M. Van de Ven, J. Coll. Interf. Sci., 110, 591 (1986). 14 U. Nagel and E. Kinsel, J. Chem. Soc. Chem. Comm., 1098 (1986). 15 K. Soai, M. Watanabe and A. Yamamoto, J. Org. Chem., 55. 4832 (1990). 16 E.I.S. Adreotti and Y. Gushikem, J. Coll. Interf. Sci., 142, 97 (1991). 17 H.U. Blaser, Tetrahedron Asymmetry, 2, 843 (1991). 18 P. Herman, C. del Pino and E. Ruitz-Hitsky, Chem. Mater., 4, 49 (1992). 19 B. Pugin and M. Miiller, in "Heterogeneous Catalysis and Fine Chemicals III", M.Guisnet et al. Eds., Elsevier Science Publishers, Stud. Surf. Sci. Catal., 78, 107 (1993). 20 Y.G. Akopyants, S.A. Borisenkova, O.L. Kalya, V.M. Derkacheva and E.A. Lukyanets, J. Mol. Catal., 83, 1 (1993). 21 M. McCann, E.M. Giolla and K. Maddock, J. Chem. Soc. Dalton Trans., 1489 (1994). 22 R.S. Drago, J. Gaul, A. Zombeck and D.K. Straub, J. Am. Chem. Soc., 102, 1033 (1980). 23 D.E. De Vos, F. Thibault-Starzyk and P.A. Jacobs, Angew. Chem. Int. Ed. Engl., 33, 431 (1994). 24 S. Brunauer, L.S. Deming, W.S. Deming and E. Teller, J. Am. Chem. Soc., 62, 1723 (1940). 25 A. Cauvel, D. Brunel, F. DiRenzo, P. Moreau and F. Fajula, in Proceeding of Zeocat'95, Stud. Surf. Sci. Catal. in press.

Zeolites: A Refined Tool for Designing Catalytic Sites L. Bonneviotand S. Kaliaguine (editors) 9 1995 Elsevier Science B.V. All rights reserved.

181

Synthesis of mesoporous manganosilicates Mn-MCM-41, Mn-MCM-48 and Mn-MCM-L at a low surfactant/Si ratio Dongyuan Zhao and Daniella Goldfarb Chemical Physics Department, Weizmann Institute of Science, Rehovot, Israel 76100

Mesoporous manganosilicate molecular sieve, Mn-M41 S, having hexagonal (Mn-MCM-41), cubic (Mn-MCM-48), lamellar (Mn-MCM-L) structure, were synthesized in low surfactant/Si ratio (0.12) and characterized by X-ray powder diffraction, transmission electron microscopy (TEM), themogravimetric analysis (TGA) and electron paramagnetic resonance (EPR). The phase transformations trends: hexagonal ~ lamellar ~ cubic -~ hexagonal or hexagonal hexagonal, lamellar mixture ~ cubic ~ lamellar were observed by variations of the base or acid content of the gel or reaction temperature respectively. The results show at low surfactant/Si ratio, Mn-MCM-41 can be synthesized both in acid and base medium, and in a wide range of temperature (21-100~ Mn-MCM-48 can be obtained only in basic medium and in high temperature (100-120~ lamellar phase is formed in high temperature (>130~ and caused the collapse of the structure after calcination at 540~ Addition of Mn ions induces the formation of cubic phase also at low surfactant/Si ratio (0.12). By variations of the base content of the gel, AI-MCM-48 also can be synthesized. EPR results suggest that Mn ions can be incorporated into the amorphous silica wall of M41S or in the interface region of the polar head groups of the template. After calcination, Mn ions in Mn-MCM-41 and Mn-MCM-48, have high mobility and are not located within the inorganic wall.

1. INTRODUCTION Recently a new family of mesoporous molecular sieves designated as M41S was synthesized in the laboratories of the Mobil Oil Company [ 1-3 ]. The M41S materials possess a regular array of uniform mesopores, which can be systematically varied in size from around 20 to 100 A [411 ]. These materials bridge the gap between microporous (zeolites) and macroporous materials (e.g. amorphous aluminosilicates). The members of the M41S family include MCM-41, having a hexagonal arrangement of pores, MCM-48, displaying a cubic structure and other members such as lamellar and cubic octamer, that are not as easily categorized [2-4,6]. The pore size of the M41S materials, their high surface areas (up to 1000 m2g-1), their distinct adsorption properties (pore condensation without hysteresis) [3b], and their thermal stability render them as prim candidates for industrial applications. Modifications in the composition of the silica based M41S materials were so far limited to the incorporation of aluminum, vanadium and titanium into MCM-41 [ 12,15,16]. Similar substitutions have not yet been reported for the silica base cubic phase, MCM-48, which have been

182 synthesized with a high surfactant/Si ratio (1-2) [2,4]. In this work we report on the synthesis of mesoporous manganosilicate materials Mn-M41 S, having hexagonal (Mn-MCM-41), cubic (Mn-MCM-48) and lamellar (Mn-MCM-L) structures, at a low surfactant/silica ratio (0.12), and present a preliminary account on the location of the Mn 2+ cations as obtained from Q-band EPR spectroscopy. We show that the addition of Mn ions induces the formation of the cubic phase also at a low surfactant/Si ratio (0.12) and that by the variation the base content of the gel, one can control the structure formed.

2. EXPERIMENTAL Materials: The synthesis mixture was prepared using sodium silicate, N brand, 27% silica, or tetraethyl orthosilicate (TEOS), Aldrich; Catapal alumina, Condea; Aluminum-isopropylate, Merck; AICI3, Merck; cetyltrimethylammonium chloride (25 wt.%) (CTAC) or bromide (CTAB), Aldrich. Other inorganic materials were MnCI 2 .6H20, Merck, HCI and NaOH. All chemical were used as received. Synthesis: Two Synthesis procedures, differing mainly in the silica source, were employed [2]. In the first, 15-30 ml of 2 M NaOH solution were added under constant stirring to 14.5 ml CTAC or CTAB. The mixture was then combined with 0.1-20 ml of a 0.4 M MnCI 2 solution and 20.5 ml tetraethyl orthosilicate. The composition of the final gel was SiO 2 .xMnO-yNa20 zCTAC(CTAB).wH20 9 , where 0.0004<x99.0%, Fluka)[9]) using Ga(NO3) 3 hydrate (Aldrich, 99.9 %; water content determined using oxine) and H3PO 4 ( 90 %, M.B)as starting materials. In the second method GaCI 3 solution was used to prepare GaOOH by titration with NaOH solution [ 10]. MCM-41 in pure silica and aluminium substituted forms were synthesised according to a modified method described by Ryoo [ 11 ]. Ludox HS-40 (DuPont)was used as silica source and sodium aluminate was used as source of aluminium. The template used was cetyltrimethylammonium chloride (CTACI) solution (25w% in water, Aldrich), with the gel composition: CTAC1 : 6 SiO 2 : 0.15 (NH4)20 : 1.5 Na20 : 250 H20. Calcinations were done by raising the temperature to 550C at a rate of 1 degree per minute in air, and holding at that temperature for 10 hours. Synthesized samples were characterised by XRD, FTIR and NMR. The sodium exchanged form was obtained by ion exchange of Al-MCM-41 (Si/Al =7 in mixed solution) with sodium chloride solution 0.02 M. Gravimetric measurements of carbonyl adsorption and decomposition were carried out on a vacuum microbalanCe. Samples, in pellet form, were activated up to 450 C, and held for 1 hour at this temperature under oxygen. After cooling to room temperature, the sample was exposed to metal carbonyl vapour, and the weight changes on subsequent decomposition monitored. EXAFS samples were prepared by activating the pelletised sample (~100rag dry weight/cm 2) under vacuum up to 450 C, then exposing to metal carbonyl vapour at room temperature for at least 6 hours to achieve saturation adsorption before evacuation. Samples were then heated to 200C or higher under dynamic vacuum to achieve complete decomposition of the adsorbed carbonyl (analogous FTIR experiments showed that this procedure completely removed carbonyl ligands). For multiple dosing, the sample was exposed to metal carbonyl vapour again at room temperature and all further steps were repeated. EXAFS measurements were carried out by transmission on the Australian National Beam Line Facility BL20B at the Photon Factory, Tsukuba, Japan, (except for multiple doting of molybdenum experiments, which used BL 10B). The collected data were analysed using the University of Washington UWXAFS 3.0 and FEFF5.05 programs[ 12].

3.RESULTS AND DISCUSSION

3.1 Bimetallic Clusters in Zeolite Y. Exposure of dehydrated NaY zeolite to Mo(CO) 6 vapor at room temperature causes rapid uptake of ca 2 molecules per supercage. The adsorbed carbonyl is irreversibly held, and heating to 200 C in vacuo causes complete decarbonylation, giving a material containing 14 w% Mo. A further 2 molecules of carbonyl can then be adsorbed, which on decarbonylation increases the Mo loading to ca 25 w%; this process can be continued further, although the capacity to adsorb Mo(CO)6 gradually decreases as the supercages fill with molybdenum. Yong and Howe

199 [ 13] prepared zeolites containing up to 53w% Mo by sequential adsorption of 14 doses of Mo(CO)6. Figure 1 shows Fourier transforms of Mo Kedge EXAFS data for MoY zeolites containing 1,2 and 4 successive doses of Mo(CO)6, decarbonylated at 200C. As reported previously[14], decomposition of Mo(CO)6 in NaY at 200C produces highly dispersed d. molybdenum species in which the average Mo-Mo ,P coordination number (the second peak in the c Fourier transform at r = 0.25nm uncorrected) is about 1.0, suggesting that the original loading of 2 Mo per supercage is retained on decomposition of , the carbonyl. The first peak in the Fourier 4 6 transform (0.16nm uncorrected) is due to Mo-O R, Angstroms bonding to the zeolite lattice (Mo-O coordination number = 2) [ 14]. The striking feature of the new data presented in Figure 1 for multiple doses of molybdenum is that the Fourier transform is Fig. 1 FT of Mo EXAFS: Mo(CO)6 in unchanged. The affinity of molybdenum for oxide NaY after decomposition at 200C; a. 1 ions of the zeolite lattice evidently overrides any dose; b. 2 doses; c. 3 doses; and d. 4 tendency to form metal clusters, and doses molybdenum in the multiple dosed samples appears to line the walls of the supercage. Some sintering as evidenced by an increase in the relative intensity of the Mo-Mo peak could be induced by heating to 400C. Gravimetric experiments showed that the stoichiometry of adsorption and decomposition of W(CO)6 in NaY is similar to that of "12 ~2 Mo(CO)6. In particular, multiple adsorption and g decomposition cycles could be used to build up o the tungsten content of the zeolite in steps of 2 FLL atoms per supercage. As shown in Figure 2 however, the W LIii-edge EXAFS data indicate that the dispersion of tungsten is very different from that of molybdenum. After decomposition o of a single dose of adsorbed W(CO)6 in NaY at 200C the dominant peak in the EXAFS Fourier 2 4 6 transform at about 0.27nm (uncorrected) is R, Angstroms attributed to a W-W interaction. Detailed fitting of the data to determine structural parameters is not yet complete, but from the relative intensities Fig.2 FT ofW EXAFS: W(CO)6 in NaY aider decomposition at 200C; a. 1 dose; b. of the 0.27nm and shorter distance peaks (which 2 doses; c. 4 doses and d. 4 doses at 400C are attributed to W-O interactions) it would appear that relatively large tungsten clusters have

~2

200 been formed. This conclusion is supported by the observation of a further peak in the Fourier transform at about 0.37nm(uncorrected) not seen in molybdenum zeolites. Evidently the interaction of tungsten with zeolite oxide ions is 3 insufficiently strong to prevent inter-supercage migration of W during decomposition of the a2 adsorbed W(CO)6 at 200C, forming what L appears to be a heterogenous distribution of larger tungsten clusters. Addition of further doses of W(CO)6 causes some changes in both relative intensities and positions of peaks; heating subsequently to 400C further enhances 0 2 4 6 the major W-W peak in the Fourier transform R, Angstroms relative to the W-O peaks, suggesting further sintering is taking place. The heterogeneity of this system will make structural analysis difficult. Fig.3 FT of a.Mo EXAFS and b.W Figure 3 shows Fourier transforms of the Mo EXAFS for NaY loaded with Mo(CO)6 K-edge and W Liii-edge EXAFS respectively for then W(CO)6 a NaY zeolite which was first loaded with 2 Mo per supercage then with one dose of W(CO)6 decomposed at 200C. Comparison of the Mo EXAFS with that obtained from one or two doses of Mo(CO) 6 alone (Figure 1) shows several differences. The relative intensity of the peak at r = 0.25nm (uncorrected) previously assigned to Mo-Mo is significantly enhanced, suggesting that this peak now contains a contribution from Mo-W interactions (which has yet to be confirmed by detailed fitting of the EXAFS). Also, the Mo-O peak at r = 0.16nm (uncorrected) is split into two components following addition of tungsten. The corresponding W EXAFS data show that "13 in comparison with the zeolite containing only -~ 2 W, addition of W to the Mo loaded zeolite produces a higher dispersion of W with a greatly reduced contribution from the W-W peak at r = 0.3 nm (uncorrected). Our provisional interpretation of these results is that pre-loading the zeolite with molybdenum ( 2 Mo atoms per supercage) A~/X,-x/'~rx produces a better dispersion of tungsten on 0 6 subsequent adsorption and decomposition of RoAngstroms W(CO)6 because of a direct Mo-W interaction in the supercage. If this is correct, Fig.4 FT of a.Mo EXAFS and b.W EXAFS of the opposite experiment in which tungsten is NaY loaded with W(CO)6 then Mo(CO)6 loaded first followed by molybdenum should lead to a different result, since the tungsten is not distributed homogeneously throughout

201 every supercage. Figure 4 shows Fourier transforms of Mo K-edge and W Liii-edge EXAFS from such an experiment( W(CO)6 adsorbed and decomposed at 200C followed by Mo(CO)6 ). The W data in this case are much more closely similar to those of the zeolite containing W alone, with major contributions from W-W peaks at r > 0.3nm (uncorrected). For Mo, on the other hand, the dispersion appears to be lower than in the zeolite containing only Mo, as judged by the enhanced contribution of the r = 0.25nm peak (uncorrected). This result is also consistent with a direct Mo-W interaction if the presence of a heterogeneous distribution of tungsten clusters in only some of the zeolite supercages is presumed to attract Mo to those particular supercages.

3.2 Molybdenum in Cloverite Since cloverite is unstable in air following removal of the template [ 15], infrared and gravimetfic experiments on Mo(CO)6 adsorption and decomposition were performed on samples calcined in-situ up to 480C in oxygen Weight loss measurements and infrared spectra both indicated that this calcination procedure removed more than 75% of the template. Nevertheless, the stoichiometry of Mo(CO)6 adsorption and decomposition were found to depend on which template was used in the cloverite synthesis Figure 5 shows the results of gravimetric experiments with both types of cloverite Prolonged exposure of cloverite derived from quinuclidine to Mo(CO)6 vapour at 18 room temperature gave a ~-- 16 iiiiiiiiii total uptake of only 4 14 iiilriiill wt%. Ozin et al [6] o 12 iiiiiiiiii report a total supercage ~ i ~ i H H H '~ 10 !@iiiili pore volume for calcined ~ 8 ::xx:::: .......... cloverite (as measured by oxygen adsorption) of ~ 4 iiiiiiiiii ca. 24 cm 3 per 100g. --- 2 iiiiiiiiii ~__~ ~ Complete filling of this 0 : : : : - : volume with Mo(CO)6 Ads Evac 100 C 200 C 400C would thus correspond quinuclidine template D piperidine template to an uptake of ca 47 wt%. The observed uptake for cloverite is only 8 % of this value, Fig. 5 Adsorption of Mo(CO)6 into cloverite suggesting that there is very little penetration of Mo(CO)6 into the supercages, and that adsorption occurs mainly on the external surface. The adsorption is only partly irreversible, the weight losses on subsequent heating in vacuo leave only 0.2 wt% Mo in the sample. In contrast, the uptake of Mo(CO)6 into cloverite derived from piperidine is 36 % of that expected for complete filling of the supercages, indicating that in this case supercage adsorption does occur. This adsorption is however also only partly irreversible. The 2 wt% Mo remaining after heating in vacuo to 200C or above corresponds to only 35% of the Mo initially adsorbed. An explanation for the differences between the quinuclidine and piperidine derived cloverite is offered in terms of the extent of dehydroxylation of the cloverite lattice during calcination. Ozin et al [6] undertook a detailed thermal analysis study of the decomposition of

202 the quinuclidine template, and identified three thermal events in the temperature range between 350 and 550C corresponding to desorption of untracked quinuclidine, pyrolysis of the quinuclidinium cation and evolution of HF and structural H20. These events did not cause any significant loss of structural integrity. The TGA-MS data reported in [6] show in fact a continuous evolution of water due to dehydroxylation between 300C and 800C, suggesting that the structural hydroxyl groups which partially block the supercage entrance in the as synthesized cloverite are still present to a significant extent after calcination at 480C. We have measured infrared spectra of cloverite samples calcined at 480C; these show a broad weak band in the region 3100-3300 cm -1 assigned by Ozin et al to the structural hydroxyl groups. Thermal analysis data are not available for the piperidine derived cloverite, but infrared spectra of this material after calcination at 480C show no evidence of the 3100-3300 cm -1 band, suggesting that dehydroxylation occurs more readily in this case. The partial desorption of Mo(CO)6 from cloverite on heating in vacuo is similar to that observed previously with AIPO-5 [16] . Ion exchanged cations have been shown to play an important role in the anchoring of adsorbed metal carbonyl complexes in zeolite Y [ 1]. The absence of such anchoring sites in aluminophosphate or gallophosphate molecular sieves restricts the applicability of the MOCVD method for these materials, at least for metal carbonyl precursors which require thermal decomposition at elevated temperatures. Mo K-edge EXAFS measurements on the Mo loaded cloverite show that the molybdenum is highly dispersed after decarbonylation, with no evidence of Mo-Mo interactions.

3.3 Molybdenum in MCM-41. Gravimetric measurements of Mo(CO)6 adsorption and decomposition were undertaken on 3 different MCM-41 materials: the all silica form, a sample containing aluminium substituted to a Si:A1 ratio of 7, and the same sample ion exchanged with sodium after calcination. Figure 6 summarizes the results of these experiments. The surface area of

~(':f~ ..-,

~~

Ifilifiii ~::;,i:~.::i~!]

i,iiiiii

;::h':: ....

e~ ~iii!i!i!i ,~] i i [ ! -

il

|

Ads Si-~l

i i I ! ~i'~I!~ "i!

Evac

:-."~

|

!

200 C ~ Si/AI= 7

Fig. 6 Adsorption of Mo(CO)6 into MCM-41

t ...............

..... ~;"

Reads

|

Evac

........

~

200C

~ Si/AI=7;Na-exdmaged

i |

203 MCM-41 after removal of the template is approximately 1000 m2 g-1. Monolayer coverage of this surface with Mo(CO)6 would correspond to an uptake of up to 150wt%. Exposure of silica MCM-41 to Mo(CO)6 at room temperature gives an actual uptake of around 20wt%, but this is almost completely removed by evacuation at room temperature, and the amount of molybdenum remaining after subsequent heating to 200C is less than 0. l wt % . Attempts to increase the molybdenum loading by adsorbing and decomposing further doses of Mo(CO)6 were unsuccessful. The aluminium substituted MCM-41 will contain acid protons after calcination of the cetyltrimethylammonium template, if the aluminium is substituting for silicon in the MCM-41 lattice. It is difficult to prove conclusively that all of the aluminium is in the lattice, but the 27A1 NMR spectrum of this material showed only tetrahedral aluminium. Aluminium substitution more than doubles the amount of Mo(CO)6 which is adsorbed on exposure of MCM-41 to the vapour at room temperature.This adsorption is however still almost completely reversible; the amount of molybdenum retained after heating to 200C in vacuo is ca. 0.5 %, and this figure was not increased by adding subsequent further doses of Mo(CO)6. Ion exchange of sodium into the aluminium substituted MCM-41 further increases the amount of Mo(CO)6 adsorbed at room temperature, to about one third of full monolayer coverage, but the amount of molybdenum retained on evacuation and heating to 200C remains extremely low. In this case, however, adsorbing and decomposing a second dose of Mo(CO)6 does significantly increase the Mo loading (to about 2.7wt% ). The surface chemistry of the all silica MCM-41 with regard to Mo(CO)6 adsorption closely resembles that of silica gel, which irreversibly adsorbs less than 1 wt% Mo [ 17]. The silanol groups lining the walls of the MCM-41 channels have a low affinity for Mo(CO)6 , and evacuation removes all of the adsorbed carbonyl. Infrared spectra have confirmed that physically adsorbed Mo(CO)6 is the only species present in this material. Substitution of aluminium into the silicate lattice will enhance the acidity of the hydroxyl groups (as in silica alumina gels), and this does appear to enhance the affinity of the MCM-41 for Mo(CO)6. The binding of Mo(CO)6 is however still much weaker than in the analogous zeolite systems (e.g.HY [18]), and desorption of the intact complex is still favoured over decarbonylation. Incorporation of sodium ions into the channels provides cation anchoring sites for the carbonyl complex which are likewise much weaker than the corresponding sites in NaY. The Na exchanged MCM-41 offers some hope of increasing the Mo content by many repeated cycles of adsorption and decomposition, but this will be a much less effective method than in NaY zeolite. The best chance of achieving high loadings of metal within the MCM-41 channel system is to decompose the adsorbed complex before it can escape the channels. We have recently attempted to do this by rapid heating of MCM-41 samples loaded with physically adsorbed Mo(CO)6 in sealed glass tubes of minimal volume. Preliminary XPS analyses of such materials show that high loadings of metal homogeneously dispersed through the MCM-41 pores are obtained, and further characterization is in progress. 4. C O N C L U S I O N S The results presented here have illustrated some of the chemistry involved in chemical vapour deposition of transition metals in micro- and mesoporous molecular sieves from carbonyl precursors. The loading and dispersion of the metal achieved depends on the size and

204 shape of the pore openings, the nature of adsorption sites within the pores, the volatility of the precursor complex and its ease of decomposition. Use of the MOCVD method to prepare novel catalysts or host-guest materials for other applications clearly warrants further investigation.

5. A C K N O W L E D G M E N T S We acknowledge Professor Ryong Ryoo for assistance with synthesis of MCM-41 and EXAFS measurements EXAFS measurements on the Australian National Beam Line Facility were made possible by a grant from the Access to Major Research Facilities Program funded by the Australian Government. Financial support also from the Australian Research Council and AusAID ( to SDD) is gratefully acknowledged.

6. R E F E R E N C E S 1. R.F.Howe in Tailored Metal Catalysts, ed. Y. Iwasawa, Reidel, Dordrecht, 1986 2. S. Ozkar, G.A. Ozin, K. Moiler and T. Bein, J. Am. Chem. Soc., 112 (1990), 9575 3. Y. Okamoto, T. Imanaka, K. Asakura and Y. Iwasawa, J. Phys. Chem. 95 (1991), 3700 4. G. Coudurier, P. Gallezot, H. Praliaud, M. Primer and B. Imelik, C.R. Acad. Sci. Ser. C., 282 (1976), 311 5. M. Estermann, L.B. McCusker, C. Baerlocher, A. Merrouche and H. Kessler, Nature, 352 (1991),320 6. R.L. Bedard, C.L. Bowes, N. Coombs, A.J. Holmes, T. Jiang, S.J. Kirkby, P.M. Macdonald, A.M. Malek, G.A.Ozin, S.Petrov, N. Plavac, R.A. Ramik, M.R.Steele, and D. Young, J.Am.Chem. Soc., 115 (1993), 2300 7. C.T.Kresge, M.E. Leonowicz, W.J.Roth, J.C. Vartuli, and J.S.Beck, Nature, 359 (1992), 710 and J.S. Beck, J.C. Vartuli, W.J. Roth, M.E. Leonowicz, C.T. Kresge, K.D. Schmitt, C.T-W. Chu, D.H. Olson, E W. Sheppard, S.B. McCullen, J.B.Higgins, and J.L. Schlenker, J. Am.Chem. Sot., 114 (1992), 10834 8. C.Y.Chen, H.X. Li and M. E. Davis, Microporous Mater., 2 (1993), 17 9. Q. Huo and R. Xu, J. Chem. Sot., Chem. Commun., (1992), 1391 10. S.Bradley, Thesis, Univ. Calgary, 1991 11. R.Ryoo and J.M.Kim, J.Chem. Sot., Chem.Commun., (1995), 711 12. E.A. Stem, M. Newville, B. Ravel, Y. Yacoby, and D. Haskel, preprint submitted to Elsevier Science, (1994) (UWXAFS); and J.J. Rehr, J. Mustre de Leon, S.I. Zabinsky and R.C. Albers, J. Am. Chem. Soc., 113 (1991), 5135 13. Yong,Y-S and R.F.Howe, J.Chem. Soc., Faraday Trans.l, 82 (1986), 2887 14. J.M.Coddington, R.F.Howe, Y-S.Yong, K. Asakura, and Y. Iwasawa, J.Chem. Soc. Faraday Trans., 86 (1990), 1015 15. S. M. Bradley, R.F. Howe, and J. V. Hanna, Solid State Nucl. Magn. Reson.,2 (1993), 37 16. R.F. Howe, M. Jiang, S.T Wong and J.H Zhu, Catalysis Today, 6(1989), 113 17. R.F. Howe, Inorg. Chem., 15 (1976), 486 18. S. Abdo and R.F. Howe, J. Phys. Chem, 87 (1983), 1713


205

Tailored synthesis, characterization and properties of ZnO, CdO and SnO 2 nano particles in zeolitic hosts

M.Wark 1, H.-J. Schwenn 2, M. Wamken 2, N.I. Jaeger 2 and B. Boddenberg I

1 Lehrstuhl fiir Physikalische Chemie II, Universit/it Dortmund, Otto-Hahn-Str.6, D-44227 Dortmund, Germany 2 Institut ftir Angewandte und Physikalische Chemie, Universit~it Bremen, D-28359 Bremen, Germany

ABSTRACT The successful incorporation ofnanometer-sized ZnO-, CdO- and SnO 2 particles in zeolites is presented. It is demonstrated that the particles are formed in mesopores which were created during the process. The sizes of the embedded particles are strictly depending on the extent of the formation of mesopores and can be adjusted by the experimental conditions such as the Si/Al ratio of the supporting zeolite. 119Sn-MAS-NMR is applied to probe the host-guest interactions of faujasite encaged SnO 2 nano particles. 1. I N T R O D U C T I O N While the encapsulation of sulfide clusters or nano particles in zeolitic hosts has been extensively studied in view of their potential application in opto-electronic devices [1], much less information is available on the specific preparation of oxide dispersions supported within zeolite matrices [2-4], although they are interesting as catalysts for oxidation reactions [4]. SnO 2 or ZnO thin films are furthermore widely used as sensor materials for reductive gas atmospheres or ammonia [5]. Highly dispersed oxide particles promise to show improved sensitivity as could be demonstrated in the case of zeolite stabilized CdO nano particles which show a drastically increased reactivity towards CO 2 [3]. In the following the influences of the zeolite type and the preparation conditions on the growth of ZnO and CdO nano particles in zeolitic matrices are studied systematically. 2. E X P E R I M E N T A L ZnO and CdO nano particles were prepared by conventional ion-exchange of A, X and Y zeolites in aqueous solutions of the metal acetates (c < 0.1 M), and in case of faujasites by subsequent treatment with NaOH solutions of different concentrations.

206 SnO2 loaded zeolites were generated either by ion exchange in aqueous SnC12 solutions of pH 3-4 for 4 days at 293 K, by repeated impregnation with small amounts of SnC12 solutions for 2 hours at 353 K, or by chemical vapour deposition (CVD) with SnC14. CVD was carded out as follows: after dehydration of the zeolite in a stream of dry nitrogen at 673 K, dry nitrogen loaded with SnC14 was led through the zeolite powder at temperatures between 373 and 673 I~ Subsequently, non chemisorbed SnC14 was removed by flushing with pure dry nitrogen. Finally, the zeolite was treated in a stream of wet nitrogen at the same temperatures at which the chemisorption was performed in order to hydrolyse the fixed SnC14 In a final step all samples were calcined under flowing oxygen at 673 K for 4-24 hours. The characterization was carded out by X-ray diffraction (Guinier-Haag camera), diffuse reflectance UV-VIS spectroscopy, transmission electron microscopy (Phillips EM 420, acceleration voltage: 120 kV), adsorption measurements, and solid state NMIL As UV-VIS spectrometer a Varian Cary 4, equipped with a diffiase reflectance attachment (Praying Mantis) and a specially designed gas cell was used. The absorption is expressed in units of the Kubelka-Munk function. Reflecting standards (LOT, reflection 75 and 99 %, resp.) were used to reference the absorption of the unloaded zeolites. The NMR experiments (27A1,298i and 119Sn) were recorded under MAS conditions at room temperature in a magnetic field of 9.4 T (Bruker MSL 400). After dehydration of the samples in vacuum (p < 1 Pa) at 673 K, adsorption and desorption isotherms were measured at room temperature using cyclopentane as probe molecule. Samples containing tin dioxide nano particles were additionally studied by impedance spectroscopy. The impedance spectra were recorded with a Hewlett Packard Impedance Analyzer HP 4192 A in a frequency range of 10-107 I-Iz under vacuum and different gas atmospheres (02, HE) using a high temperature measuring cell (up to 673 K). 3. R E S U L T S A N D D I S C U S S I O N 3.1. Zinc and cadmium oxide dispersions in zeolites The treatment of zeolite samples, in which 20 or 40 per cent of the sodium ions were exchanged by zinc or cadmium ions (table 1), with diluted sodium hydroxide solutions leads to the precipitation of the corresponding metal hydroxides in the zeolitic pores. The hydroxides can be converted into the oxides by calcination of the samples in a stream of oxygen or air at 673 K for 5 hours. The cadmium containing samples change their color from white to yellow during this treatment. 3.1.1.Diffuse reflectance UV-VIS spectroscopy Whereas zeolite samples containing zinc or cadmium only in the cationic form exhibit no absorption at wavelengths higher than 200 nm~ samples containing dispersions of zinc or cadmium oxide clusters start to absorb at wavelengths in the near UV- or visible region, respectively. In figure 1 the absorption behaviour of dispersions of zinc oxide in different zeolitic hosts is shown. With the exception of the reflectance spectrum obtained for the dispersion in A-zeolites which resembles more a molecule spectnun, all other spectra show the form typical for solid state semiconductors, but in comparison to bulk ZnO the absorption edges are blue-~ified for

207

F(R)

1-

'\

(e)

0.5-

x O_

200

250

300

350

400

wavelenght / Figure 1" Diffuse reflectance spectra of zinc oxide dispersions in different types of zeolites, (a) NaA (Si/Al = 1.0), (b) NaX (Si/A1 = 1.3), (c) NaY (Si/A1 = 2.9), (d) NaEMT (Si/AI = 3.8) and (e) bulk Zn0, physically diluted with NaY. all samples. This can, as a consequence of the well known quantum-size (Q-size) effect [6], be attributed to sizes of the zinc oxide particles smaller than about 6 nm (table 1). It is observable that the positions of the absorption edges resulting from the main diameters of the encapsulated nano-particles depend on the Si/AI ratio of the zeolitic hosts. In A-zeolites the absorption spectrum exhibits a relatively sharp maximum at 215 nm. By comparing this result with data published in the literature, the formation of extremely small clusters with the composition [Zn40] 6+ can be assumed. Kunkely and Vogler had found an absorption maximum at 216 nm for such cluster ions stabilized in aqueous solutions by acetate counterions [7]. The clusters are formed during the dehydration step even without a prior treatment of the samples with NaOH solution, because the zinc ions attempt to maintain a favorable coordination sphere by substituting the water molecules by oxygen atoms of the zeolite framework. The best possible coordination can be achieved by the formation of [Zn40] 6+ cluster ions in the sodalite cages. Only in A-zeolites (Si/AI = 1.0) there are enough A1 atoms in the sodalite units to balance the charge of the clusters. In faujasites, however, the treatment of the ion exchanged zeolites with NaOH solution is essential for the formation ofnano particles of zinc or cadmium oxide. The higher the Si/AI ratio of the zeolitic hosts the less pronounced is the blue-shift of the absorption edges and the bigger are the formed nano-oxide particles. The diameters of the zinc oxide nano particles can be calculated from the blue-shifts by an "effective mass approximation" established by Brus [8], to approximately 3 nm for the dispersion in zeolite X, and 4-6 nm for that in the zeolites Y and EMT. The results are in good agreement with particle sizes estimated from TEM micrographs and indicate that mesopores must be created during the formation of the nano oxide particles because they exceed the width of the supercages in

faujasites.

208

3.1.2.Adsorption isotherms Evidence for the growth of the oxide particles within the zeolite framework was obtained from the observation of hysteresis loops in the adsorption/desorption cycles of cyclopentane isotherms. The hysteresis loops range from relative pressures P/P0 ~ 0.3 to P/P0 ~ 0.85 corresponding to average pore size diameters of 2-10 nm [9]. The volumes of the micro- (Vmi) and mesopores (Vine) per gram of dry zeolite can be obtained from the total amount of adsorbed probe gas a 0 at the relative pressure P/P0 = 1, the amount aj at the junction of both branches of the hysteresis loop, and the molar volume of the liquid probe molecule V with the equations Vmi = a).Vand Vine= (a0-aj)V. For dispersions of zinc oxide nano particles in different zeolites the values listed in table 1 have been calculated. Table 1: Si/Al ratios, micropore (Vmi) and mesopore volumes (Vine) of the zeolites and diameters of the formed zinc oxide nano-particles of ZnO dispersions in different types of faujasites. Sample ZnNaX ZnO/NaX ZnNaY ZnO/NaY ZnNaEMT ZnO/NaEMT

degree of ion Si/Al ratio exchange/% 20 1.3 20 1.25 40 2.9 40 2.55 40 3.8 40 3.35

treatment with NaOH . . . . . 0.1M,0.3h,293K . . . . . 0.1M,0.3h,293K . . . . . 0.1M,0.3h,293K

particle size/nm . 1-2 . 3-4 . 4-6

E 9 E e /(cm~g"1) /(cm~ "1) 0.32 0.000 0.30 0.008 0.29 0.000 0.25 0.031 0.30 0.000 0.24 0.052

Additionally it was found by 29Si-MAS-NMRthat the Si/Al ratio of the zeolites decreases with the NaOH treatment. This observation indicates that in a first step of the mesopore formation the hydroxyl ions attack at the silicon centers of the zeolites. In this context it is important to mention that a parent zeolite containing no zinc or cadmium but only sodium ions showed no mesopore formation under the same conditions. This demonstrates that the simultaneous precipitation of metal hydroxides favours the formation of mesopores. A more detailed discussion of the mechanism of the mesopore formation will be presented in a further paper [ 10]. Table 2: Formation ofCdO nano particles and mesopores of the zeolite in dependence of the treatment of CdNaX zeolites (degree of ion exchange: 20 %) with NaOH solution. Sample

treatment with

NaOH

CdNaX CdO/NaX CdO/NaX CdO/NaX CdO/NaX H

particle

E 9

. . . . . . 0.1M,0.3h,293K 1-2 0.1M,0.3h,353K 7-9 0.1M,2h,293K 10-12 0.75M,0.3h,353K ,~ 30 ill

i

E e

size/am /(om 1) /(cm~;"1) 0.31 0.31 0.27 0.25 0.22 i

0.000 0.000 0.005 0.016 0.029 i

209 The extent of mesopore formation and the diameters of the formed oxide particles can furthermore be varied by the conditions of the ion-exchange (pH value) [3] and of the treatment with NaOH solution. The latter is shown as an example in table 2 for CdO dispersions in zeolite X. The higher the concentration of the NaOH solution, the longer the time of treatment or the higher the temperature the higher the extent of mesopore formation and consequently the larger the oxide particles formed.

3.2. Dispersion of SnO 2 particles in faujasites 3.2.1.Preparation The formation of S n O 2 llano particles depends both on the preparation method and the Sn loading. The modification of NaY zeolite (Si/AI = 2.9) with Sn (II) ions (SnC12 x 2 H20 ) by conventional ion exchange causes strong acidic conditions depending both on the solvent and the concentration of the solution. A suitable pH value is required to minimize the damage of the zeolite framework. If the zeolite is exposed to exchange solutions of pH > 3 only small losses in crystallinity, expressed by the BET surface, of 5-10 % were observed. But even under these conditions the removal of alumina l~om the zeolite framework leads to the appearance of a signal of extra framework aluminum in the 27A1-MAS-NMR (~i = -1 ppm, referenced to 0.1 m A1C13 solution) and the formation of mesopores, identified again by hysteresis loops in adsorption/desorption isotherms with cyclopentane as probe molecule. Due to these conditions at high loadings (> 4 wt %) SIlO2 nano particles with a broad size range of 2-20 nm are formed as can be inferred from TEM micrographs. The particles are inhomogeneously dispersed and numerous agglomerations are obvious probably on the outer surface of the zeolite. At lower loadings (< 4 wt %) the particle size range is somewhat smaller (2-10 nm) and the dispersions are more homogenous. In contrast to that the impregnation method reveals a relative homogenous SnO 2 dispersion with small particles of 2-5 n m By this preparation conditions the creation of large mesopores as well as the formation of large agglomerates is prevented. By CVD with SnC14 tin loadings up to 2 wt.% can be achieved trader a tolerable loss of crystallinity of 10-15 %. In the UV-VIS spectra of samples prepared by CVD the absorption of the tin species cannot unambigously be distinguished from the zeolite absorption leading to just a weak absorption below 2 = 250 nm~ and on TEM micrographs no particles were observed indicating that the tin species are very highly dispersed, i.e. that possibly formed SnO 2 nano particles are smaller than 1 nm in diameter. 3.2.2.Determination of particles sizes The determination of the sizes of highly dispersed tin dioxide nano particles is in principle possible from X-ray diffractograms, UV-VIS spectra, and transmission electron microscopy. X-ray difl~actograms offaujasites modified with SnO 2 allow the "fingerprint" identification of both the zeolite framework and the formed SnO 2. The SnO 2 reflections (110), (101) and (201) are detectable if the Sn loading is > 2 wt.% indicating that the particles consist of crystalline SnO2, but the reflections are partially superimposed by reflections of the zeolite lattice which prevents a particle size determination according to the Debye-Scherrer equation. Electron diffraction patterns reveal no single reflections of SnO 2, but rings which indicate that the particles must be randomly distributed.

210 F(R) d

200

300

400

500

w a v e l e n g t h / nm

Figure 2: Diffuse reflectance spectra of(a) bulk SnO2, physically diluted with NaY (30 wt.% Sn), and dispersions of SnO 2 in Y-zeolites, prepared by: (b) and (e) ion exchange (1.1 and 4.4 wt.% Sn, resp.), (d) impregnation (4.6 wt.% Sn) and (e) CVD (1.4

wt.% Sn). The analysis of diffuse reflectance UV spectra of SnO2 modified faujasites (figure 2) is possible only for loadings higher than about 1.2 wt.% and is moreover complicated by the absorption of the base zeolite materials which must be taken into account at wavelenghts smaller than 300 nm Therefore special precautions, e.g. use of suitable reference substances, must be taken to separate the absorption of the zeolite from that of the embedded material (in this case: SnO2) [ 11]. In contrast to the relative sharp absorption edge in the spectrum of bulk SnO2 which was physically diluted with the parent NaY zeolite (figure 2a), in the spectra of zeolite stabilized SnO 2 dispersions a blue-shitted absorption with a pronounced tailing is obvious due to either a superposition of the absorption edges of differently sized nano particles or to changes in the electronic structure which may result from changes in the crystal structure or from interactions with the host. The UV-VIS spectrum of the sample obtained by impregnation (figure 2d) which consists of particles with a rather narrow size range, shows a more pronounced blue-shill as well as a steeper slope of the absorption flank with less tailing supporting the interpretation, that a superposition of particles with different diameters influences the spectra. Furthermore there are no indications for changes in the crystal structure of the SnO 2 particles from X-ray or electron diffraction. But as an additional effect on the spectra interactions of the embedded particles with the host must be taken into account as revealed by ll9Sn-MAS-NMR spectroscopy (see below). Due to the uncertainty in the interpretation of the absorption edges of the SnO 2 nano dispersions, for this systems no size determination according to the "effective mass approximation" can be performed in contrast to the zinc or cadmium oxide dispersions. Therefore we conclude that for zeolite encapsulated SnO 2 nano dispersions the transmission electron microscopy is the only reliable method for particle size determinations. Consequently all statements in chapter 3.2.1. are based on TEM results.

211 3.2.3.119Sn-l~IAS-NMR The ll9Sn MAS-NM~ of bulk SnO 2 physically mixed with parent NaY zeolite (10 wt.% SnO2) and of a zeolite sample containing relatively large SnO 2 particles (2-20 nm) is shown in figure 3. The sample containing bulk SnO 2 exhibits the usual spectrum for SnO 2 with a central band at 8 - -602.8 ppm, referenced to (CH3)4Sn, and the typical rotation side-bands [12]. The width at half height for the central band is about 800 Hz.

"--500

'-600

' 70 O -

D pxltl

Figure 3: 119 Sn-MAS-N]VIR spectra of bulk SnO 2 physically mixed with zeolite Y (lower spectrum) and SnO 2 nano particles embedded in Y-zeolite (upper spectrum). For SnO 2 particles encapsulated in the zeolitic host the same signals are found but with a distinct decrease in the signal to noise ratio of the signals. Furthermore the width at half height of the central band is increased to about 1900 Hz. In the original spectrum of the dispersion a broad, non-structurated hump between -500 and -700 nm below the signal was found probably resulting from interactions of the smallest particles. It was surpressed by cutting off the first points in the FID (free induction decay) to clarify the effects on the linewidth. For particles smaller than 3-4 nm only the broad hump without a distinct signal of SnO 2 could be identified. Sheng et al. reported a broadening of the linewidths for SnO 2 which was co-precipitated with AI203 in comparison to non-diluted bulk tin dioxide. They assumed that the [19Sn spin-lattice relaxation time (T1) increases due to dilution effects or a strong interaction between Sn and A1 [13]. Our spectra of bulk SnO 2 physically mixed with the unloaded zeolite contradicts the assmnption that the degree of dilution is important. Moreover, our results confirm that both the number of defects and the interactions with the host which increases both with decreasing particle size of the encapsulated SnO 2 particles, influence the linewidth oft he NMR signal.

3.2.4.Impedance spectroscopy of tin dioxide dispersions Impedance spectroscopy reveals for SnO 2 modified zeolite samples with high loadings a conductivity process which is caused by charges hopping or tunneling between neighbouring SnO 2 nano particles. This conductivity process response is extremly sensitive to changes of the surrounding gas atmosphere. The mechanism of the conductivity will be discussed in detail in a forthcoming paper [14].

212 4. C O N C L U S I O N S 9It is possible to stabilize dispersions of zinc and cadmium oxide or tin dioxide nano particles in the pores offaujasites. The sizes of the encapsulated particles can be determined from UVVIS spectra or transmission electron micrographs. The first method fails for SnO2 dispersions, because the absorption of the embedded particles coincides with the absorption of the parent zeolite. 9The formation ofnano particles is combined with a formation ofmesopores in the zeolite and depends therefore on the pH values of the exchange solutions and the stability of the zeolite Si/AI ratio) against the used reactants, for example NaOH solution. 9The incorporated nano particles interact strongly with the zeolitic host, observable, for example, by an increase in the linewidths of I19Sn-MAS-NMR signals and with each other as indicated by impedance spectroscopy. 5. A C K N O W L E D G E M E N T S Dr. M. Wark is member of the graduate college "Dynamische Prozesse an FestkOrperoberfl~ichen" supported by the Deutsche Forschungsgemeinschait. We thank Drs. H. Wiggers and U. Simon (University of Essen, Germany) for recording impedance spectra and Drs. J. Rathousky and A. Zukal (Heyrowsky Institute Prague, Czech Republic) for the measurement of adsorption/desorption isotherms and Dr. 1L Grof~e (University of Dortmund, Germany) for discussions concerning the NMR spectroscopy. 6. R E F E R E N C E S 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14.

G.D. Stucky, Stud. Surf. Sci. Catal. 85 (1994), 115 and references therein S.0zkar, G.A. Ozin, IC MSller and T. Bein, J. Amer. Chem Soc. 112 (1990), 9575 M. Wark, H.-J. Schwenn, G. Schulz-Ekloffand N,I. Jaeger, Bet. Bunsenges. Phys. Chem. 96 (1992), 1727 F. Roessner, A. Hagen, U. Mroczek, H.G. Karge and I~L-H. Steinberg, Stud. Surf. Sci. Catal. 75 (1993), 1707 W. G6pel, J. Hesse and J.N. Zemel (eds.), Sensors, VCH Weinheim, 1989 H. Weller, Adv. Mater. 5 (1993), 88 H. Kunkely and A. Vogler, J. Chem Soc.; Chem. Commun. (1990), 1204 L.E. Brus, J. Phys. Chem 90 (1986), 2555 M. Wark, G. Schulz-Eklofl~ N.I. Jaeger and A. Zukal, Stud. Surf. Sci. Catal. 69 (1991), 189 M. Wark, H. Kessler and G. Schulz-Eklofl~ in preparation J. Klaas, K. Kulawik, G. Schulz-Ekloff and N.I. Jaeger, Stud. Surf. Sci. Catal. 84 (1994), 2261 N.J. Clayden, C.M. Dobson and A. Fern, J. Chem Soc.; Dalton Trans., (1989), 843 T.-C. Sheng, P. Kirszensztejn, T.N. Bell and I.D. Gay, Catal. Lett. 23 (1994), 119 H.-J. Schwenn, U. Simon, H. Wiggers, G. Schulz-Ekloff and N.I. Jaeger, in preparation.


213

Host- Guest Interactions in Zeolite Cavities. A. Zecchina, R. Buzzoni, S.Bordiga, F. Geobaldo w, D. Scarano, G. Ricchiardi, G. Spoto, Dipartimento di Chimica Inorganica, Chimica Fisica e Chimica dei Materiali, Universit~ di Torino, via P. Giuria 7, I-10125 Torino Italy. Abstract When molecules of low-medium proton affinity (B = N2, CO, C2H4, C2H2, propene, methylacetylene, acetonitrile and H20) are adsorbed on acidic zeolites, the resulting host-guest interactions in the cavities and in the channels are dominated by hydrogen bonding in neutral Z--H-..B adducts between the Bronsted groups Z---H (Z = zeolite framework) and the base. If the interaction is sufficiently strong and the protonated products stable (ethylene, acetylene, propene and methylacetylene) the hydrogen bonded precursors are slowly consumed with formation of olygomeric products entrapped into the zeolite cavities. Acetonitrile gives a stronger hydrogen bonding interaction not followed by protonation.Water initially gives neutral complexes; however when more than one water molecule per protonic site is present, HsO2§ species are formed. When molecules of high proton aff'mities are adsorbed (NH3 and Py), B--H§ - ionic pairs readily appear. The modifications induced on the spectrum of B - - H § by interaction with Z- can be used to probe the basicity of the negatively charged framework. At higher filling conditions BhH+...B dimers are also observed. The basic IR spectroscopy of all these hydrogen bonded systems is discussed in detail. 1. Introduction. At is well known, the zeolites cavities are often considered as special regions of the intracrystalline space where molecules are forced and guided to react together following special paths dictacted by: i) the forces acting inside the cavities; ii) the distribution of sites on the internal surfaces; iii) the spatial restrictions imposed by the dimension and shape of the cavities; iv) the defined organization of the intersecting channels imposing limitations to the chemical and diffusion paths. All these points can be collected under the synthethic definition of hostguest interactions so fruitfully used to understand many properties of supramolecular and enzymatic systems. Among the great variety of known zeolitic structures, in this contribution we shall restrict our analysis to two of the most acidic ones: H-ZSM-5 and H-MORD. This choice is dictated by the importance of these microporous solids in many acid-catalyzed reactions of great economic importance and by the abundance of data present in the specialized literature. However, we hope that many of the considerations contained in this paper can be sufficiently general to be usefully extended to other systems of different structure.

Dipartimento di Scienza dei Materiali ed Ingegneria Chimica Politecnico di Torino, corso Duca degli Abruzzi 24, 1-10129 Torino Italy

214 In particular we shall examine the following problems: i) the interaction of single molecules with increasing proton affmities with the acidic sites and the structure of the formed hydrogen bonded precursor species, ii) the subsequent proton tranfer process and the effect of the simultaneous presence of other molecules in the channels and in the cavities in favouring the proton tranfer (proton transfer as a cooperative process), iii) the role of cavities and channels dimensions in determining the structure of the protonated species, iv) the role of acidic sites concentration and v) the type of interaction occurring between the protonated species and the negatively charged walls of the zeolite cavities. The discussion of these items will be mainly based on the IR spectra of interacting species, not only because of our more specific experience in this field, but also because IR spectroscopy is an extremely sensitive tool for the study of the forces (even the weakest ones like those of the van der waals and hydrogen bonding type) acting among the molecules and between the molecules and the internal surfaces. 2. Bronsted acidity and hydrogen bonding: a short review of the perturbations induced in the IR spectra of the adsorbate and of the adsorbent.

When a base B interacts with a Bronsted acid H - - Z the formation of hydrogen bonded 1:1 adducts Z--H---B is constantly observed. If the internal modes of A and B moieties are (for the time being) not considered, only three relevant vibrational modes of the hydrogen atom (corresponding to the three degrees of freedom of the hydrogen mass) must be considered [ 1, 2]: v(Z--H...B) ~i(Z--H-..B )

Z - H stretch Z - H bending (in and out of plane)

7(Z-- H--.B) J When the Z moiety has internal structure, the two bending modes of the unperturbed Z - - H species can have distinct frequencies and the same happens after hydrogen bonds formation. A further important (low frequency) mode is the v( Z - - H---B) mode: Z....B stretch. With respect to the unperturbed v(Z--H) stretching frequency, the hydrogen bonding interaction causes: i) a downward shift Av ~: AH (interaction entalpy), ii) a parallel increase of the half-width (FWHM), because the band can be now better expressed as v( Z--H-.-B ) + nv (Z...B) and iii) a parallel increase of the intensity I [1,2]. Notice that it has been experimentally observed that the FWHM of the perturbed band is roughly 3/4 Av [ 1, 2]. The shifts Av can vary from few tens of cm -1 (extremely weak perturbations) to 300-400 cm 1 (weak-medium hydrogen bonds) to 1000-2000 cm-' (strong and very strong hydrogen bonds). Simultaneously the half width of the perturbed Z---H stretch can gradually increase from a few tens to 1500 cm -'. In this last case the band is so broad and covers a so large frequency interval that the possibility of mixing with other modes (either with fundamental or overtones and combination character) becomes very important, making the profile analysis a complicated matter [3]. It is useful to recall that the v(Z--H.--B) usually falls in a spectroscopic region where the zeolites are fully transmitting the IR radiation. The in and out of plane bending modes are shifted above the frequency of the unperturbed molecule (with broadening). The shift (and the associated broadening) is usually smaller than that observed for the stretching mode. A schematic illustration of the evolution of the v, ~i and ~/ modes with the increase of the strength of the hydrogen bonding interaction is illustrated in Fig. 1 (the frequencies of the unpertur-bed and perturbed modes are in rough agreement with

215 the literature data obtai-ned on zeolites and in homogeneous conditions) [1-4]. In this scheme, the complex effects on the A A AI band shape of the broad v(ZmH.--B) mode, deriving from resonance effects with the overtones and combinations of the 8 and 1, fundamen-tals are omitted. These effects should be particularly relevant in spectra d-e, be-cause the frequencies of the v(Z--H..-B) and of the 2x8 modes fall in the same range and consequently are suita-ble to give a strong Fermi-type reson-ance with appearance of an Evans win-dow. Similarly in spectra f-g, the effect of anharmonic mixing of the v(Z--H---B) with the corresponding 8 and 1, modes (now heavily superimposed) is not 2 oo lo'oo 4000 considered at all, although it is known wavenumber c m -1 that it deeply alters the band shape [3]. As a final point let us remark that all the Figure 1. Schematic illustration of the evolution of the considerations developed so far, did not v, 8 and "ymodes of the (ZH...B) groups with the increase consider the internal structure of the of the strength of the hydrogen bonding interaction. base B, with the associated (internal) overtones and fundamental modes which could either couple or direcOy mix with the v, 8 and ~/fundamentals and so contribute to make the band profile of the v(Z---H---B) mode more and more complicated. In conclusion the sequence of spectra reported in Fig.2 is highly simplified and must be simply considered as a useful frame for the understanding of the very basic manifestations of hydrogen bonding when the proton affinity of the base is gradually increased. On zeolites, the 8 and 7 modes cannot be always observed because they usually fall in the region where the skeletal modes of the zeolite are strongly absorbing. Moreover, due to their lower frequency, they can easily mix with these modes (zeolite framework stretching and bending modes). Finally the A.-.B stretch, usually in the 20-200 cm -1 interval [ 1-3], cannot be observed at all with the usual FI'IR spectrometers. As it is well known, the typical Bronsted sites of H-ZSM-5 and H-MORD [ZH = A1OHSi (Z = A1OSi )] are characterized by stretching bands at 3609 cm-' (H-ZSM-5) and 3612 cm -1 (HMORD) [5, 6]. Basic molecules entering the channels, form hydrogen bonded species with the acidic groups shifting their stretching frequency to lower values, the shift being proportional to the interaction energy and to the FHWM. As far as H-ZSM-5 is concerned, this is shown schematically in Fig. 2a for the series of bases of increasing proton affinities N2, CO, C2H2, C2H4, propene, methylacetylene, acetonitrile and H20 [7,8,9,10]. In this representation, the frequency and the FWHM of the peaks are in agreement with the experimental results.When the shift Av is reported against the FWHM, the diagrams reported in Fig.2b are obtained, which are extremely similar to those well known in homogeneous solutions. Entirely similar data have been obtained on H-MORD. The data reported in Fig 2a, 2b have been obtained after the completion of the reaction 2) Z w H + B ~ Z--H---B

A_v

8A YA

AA

aA

216

to

~ o. o o o.. o~ -,o • /ox

8 ~5

J

!

3500

3000

2500

wavenumber cm

2000

~

-1

Fig. 2a: Evolution of the v(Z---H...B) band caused by hydrogen bonds with bases (B) of increasing proton affinities. i.e.under conditions where the channels and cavities are far from being completely filled by the adsorbates. It must be recalled that the shape of the Z--H...B stretching bands of the CH3CN and H/O adducts are heavily modified by the presence of a prominent Evans window at 2630 cm ~ caused by a Fermi resonance with 900 the 8(Z--H-..B) mode localized a t - 1315 cm -1 (vide infra) as expected on the basis of 750 the previous discussion and as already demonstrated in ref [9,10] (two false bands being actually observed as shown schemati600 cally in Fig. 2a). The band corresponding to water is not the only one present in the "1450 stretching region: for Fig.2a we have simply LL selected the most intense, corresponding to 300 the strongest O--H...O hydrogen bond (vide

infra). 150

2~o

480

88o

~v

c m "1

88o

lo'oo

Figure 2b. Relation between the frequency shifts (Av) of the acidic OH groups and the FWHM.

Let us recall that the case of H20 has been the source of an interesting debate, since some autors think that beside hydrogen bonded species also the protonated ones can be present [4 and references therein]. The problem is not of simple solution either from the theoretical and from the spectroscopic point of view because the various types of hypothesized species have very similar energy [4] and have quite complicated IR spectra (owing to the presence of two OH

217 groups in the base) which can be distinguished only with some difficulty. 3. From the hydrogen bonded precursors to the protonated species: the ethene, propene and acetylene cases. After the formation of the hydrogen bonded precursors with the x-electrons of unsaturated hydrocarbons, a slow reaction is usually observed leading to the formation of protonated species. This species are: - ethene: saturated oligomers Z-CH2CH3, Z-CHzCH2CH2CH3 etc. The Z-C bond is partially covalent [8] - propene: saturated oligomers. Also in this case the Z-C bond is quite covalent [8]. - acetylene: Z-..[(CH=CH)n-CH=CH2] + polyacetylenic c h a i n s w i t h blue-violet color where the positive charge is delocalized on the whole chain [8]. On H-ZSM-5 the length of the chains is dictated by the distance between the intersecting channels (which is also the distance between he growing oligomeric chains initiated at different sites). We can clearly see here how the structure of the zeolite is determining also the structure of the reaction products. The protonation speed is faster in the case of propene, which is also forming the strongest hydrogen bonded precursor of the whole series of investigated hydrocarbons. The protonation reaction can be suppressed by lowering the temperature of the sample to - 200 K: under these conditions the hydrogen bonded precursors are stable. On H-MORD the situation is very similar with only one exception concerning the interaction with propene. In fact when the interaction is made at RT, the high number of olztgomefic chains (consequential to the high concentration of Bronsted sites present in our mordenite samples characterizes by Si/Al = 5), quickly formed at the channels mouths, is sufficient to block the penetration of propene in the internal spaces. The spectrum of the hydrogen bonded precursor cannot consequently be observed with sufficient intensity. This obstacle can be overcome by conducting the experiment at -200 K, i.e. at a temperature where the protonation reaction is suppressed: the schematic spectrum represented in fig 2a has been obtained in this way [ 11]. It is worth to consider that even with propene (which gives the fastest reaction at RT) the number of hydrocarbon molecules inserted into the Z - - H bond and into the growing chain is not exceeding (in average) the number o f - 1 per second [8]: this indicates that a high activation energy barrier is present for both protonation and insertion reactions. That the proton transfer reaction is slow, can be easily understood on the basis of the moderate strenght of hydrogen bonding interaction between the hydrocarbon and the acidic group in the precursor species. Despite this unfavourable kinetic factor, we can always observe the formation of a quantity of saturated and unsaturated oligomers sufficient to fill all the available internal space. These oligomeric species a r e thermodynamically very stable and cannot be depolymerized by decreasing the hydrocarbon pressure at RT. When the the Z--H---NCCH3 complex is considered, we notice that the involved hydrogen bonding interaction is stronger than that found before for the hydrocarbons: however no traces of protonated species Z--..*I-I--NCCH3 is observed at low Idling conditions [9]. We think that this is due to the low stability of the iminium species and to the insufficient stabilization of this positive species by hydrogen bonding with the weak Z- base. In principle the situation could change in presence of excess CH3CN (which is a base stronger than Z-) because the CH3CN--+H--NCCH3 species could be stabilized by a stronger hydrogen bonding interaction. We did not obtain definite proofs that this is really happening when the pores are completely f'dled by CH3CN. This does not means that nothing significant is happening in presence of an excess of

218 adsorbate: in fact the spectroscopic properties of the hydrogen bonded species are greatly influenced possibly because of the modification of the dielectric constant occurring during the pore filling [ 12].

4. On the formation of H2n+IOn + species through cooperative effects. The IR spectra of increasing doses of H 2 0 adsorbed on H-ZSM-5 are illustrated in Fig 3 (difference spectra). The full line spectra correspond closely to those already published [ 10] for H+/H20 ratios comprised in the 0-1 interval, the only difference being represented by appearance of a clear peak at 1315 cm -1 (not reported before). Another important feature HzO/H-ZSM-5 appearing upon water dosage (not J~ reported for sake of brevity), is ',i 'J' \ observed at 875 cm -1, i.e. in the small I, i| ; characteristic transmission window 1: ' h separating the two families of it j~11 , ' " .~..r r . i Ill stretching modes of the TOn building %1%~ units [12]. The negative peaks at 3610 and 3745 cm -~ correspond to consumed 'Ill ~ s'" '%'%# li.l 11 species (Bronsted sites and silanols ii,. s i ~ ** , | stretching modes). The reported specIll, %..,,.' , "~ t i t ~ 9 , , ,,~ tra are better explained on the basis of .oI~1 ~ .""* II~tl \ " " ' ' e, the dominant hydrogen bonded struc.." , -- :, ture: el t "14~ ~" " . " " ~# l, I II llt H I

.

...a

9

-',.,I

ia'"

OI

I

I

I

I

3500

3000

2500

2000

1500

i

w a v e n u m b e r c m "1

In particular: i) the narrow peaks in the 3680-3697 cm -~ interval belong to Figure 3. Background subtracted spectra of increasing the v(OH) of the external OH (a) doses of H20 adsorbed on H-ZSM-5 outgassed at 673 K. group of adsorbed water in two slightly different locations; ii) the broader absorption at - 3550-3530 cm -1 ( FWHM --- 200 cm-1), partially overlapped to the negative peak of the Bronsted sites, is the stretching mode of the second OH group of water (e) interacting with the negative oxygen group of the framework via a weak hydrogen bonding; iii) the two apparent bands at 2900 and 2450 cm -] belong to a nearly symmetric absorption centred at 2650 cm -1 with b'WHM - 800 cm -] (see Fig. 2a) and are generated by an Evans window at the centre of the adsorption caused by a Fermi resonance effect with the 8 mode of the [}(OH) group (Br~3nsted group) at 1315 cm -1 (following the empirical v(OH..)-do_o correlation well established for hydrogen bonding interactions in homogeneous phases [ 1-3], a do+ distance in the 2.6-2.65 A interval is inferred, in complete agreement with the structure proposed by Sauer et al. [ 14,4]); iv) the absorption at 1670-1620 cm -~ (highly asymmetric on the high frequency side)

219

corresponds to the bending mode of water; v) the band at 1315 cm -~ is the in plane bending ~5 mode of the Brrnsted group (at - 1050 cm -~ in vacuo) [15,16,4] upward shifted by the strong hydrogen bonding interaction; vi) the band at 875 cm -1 is the ? mode of the Br~3nsted group (at400 cm -1 in vacuo) upward shifted by the strong hydrogen bonding. When further doses of water are adsorbed (corresponding to the progressive filling of the channels and cavities and also with the formation of weak hydrogen bonding interactions with the external silanols" broken spectra) four main effects are observed, i.e." i) the disappearance of the 1315 cm -~ band; ii) the simultaneous disappearance of the Evans window; iii) the disappearance of the 875 cm -1 band; iv) the formation of a broad continuum in the 2250-1300 cm -1 interval, associated with an extremely broad absorption due to very strong hydrogen bonds typical of solvated protons [ 17]. Of course also the manifestations of liquid water (bands 3300 and 1620 cm -1) are clearly emerging. All these facts indicate that for H20/H + ratios > 1, a proton tranfer occurs through the agency of further water molecules and that the situation becomes similar to that of a concentrated acid solution, as it can be easily verified by comparing these spectra with those of concentrated HC1, H2SO4 etc [18]. The two background oscillations labelled with an asterisk, appear any time an adsorbate is interacting with the framework and do not depend upon the structure of the adsorbed molecule. A detailed discussion of their origin is given elsewhere [ 19]. A similar series of spectra has also been found on H-MORD" however in this case, due to the higher concentration of protons, the final spectrum obtained at high pore filling conditions is similar to that of an even more concentrated acid solution characterized by H20/H + =_ 3 [ 18]. Coming back to the H-ZSM-5/H20 system, it is worth to remark that the formation of the broad background in the 2000-1300 cm -~ range, already starts before the full disappearance of the free Bronsted v ( Z - - H ) band: this likely means that in the vicinity of the H20/H + - 1 stoichiometric ratio, free Brrnsted groups, monomeric hydrogen bonded species, dimeric H502 + and polymeric HsO2+-n(H20) species are in mutual equilibrium.

5. The proton transfer occurring with strong bases (NH3 and Py): the IR spectroscopy of NH4+-'-Z" and PyH§ - (Z = zeolite) ionic pairs. Strong bases like NH3 and Py interacting with Z - - H , immediatly gives the protonated NHn+--'Z- and PyH+---Z- ionic pairs. The strength of the interaction within the moieties of the ionic pairs will depend upon: i) the acidic character of the N-H groups of NH4 + and PyH+; ii) the basic strength of Z-. As Z - - H is a strong acid, the coniugated Z- base will be consequently weak; the same can be concluded about the acidity of NH4 + and PyH +. On this basis the hydrogen bonding interaction between the two moieties of the ionic pairs is expected to be rather weak. However as this interaction is sufficient to perturb the spectra of NH4 + and PyH +, we can use the NH4 + and PyH + species as probes of the framework basicity of Z- (exactly like we have used weak B bases as probes of the Brrnsted acidity). The spectra of increasing doses of NH3 adsorbed on H-MORD are illustrated in Fig. 4 (spectrum 7 is corresponding to NH3/ H + - 1). Similar spectra have been obtained on H-ZSM-5 (not illustrated for brevity). As far as the spectra 1-6 are concerned, we notice: i) the progressive consumption of the Brtinsted sites band at 3610 cm -1 (negative band); ii) the proportional growth of an intense peak with maximum at 1440 cm -1 (and a shoulder at -1420 cm -1) assigned to the v4 mode of the NH4 + (with T2 symmetry in the free ion); iii) the formation of a weak absorption at 1680 cm -1 assigned to the v2 mode of NH4 + (with E symmetry and only Raman active in the free ion) iv) the proportional growth of a complex band in the 3400-3150 cm -1 interval, with maximum at 3254 and

220 shoulders at 3370 and -3200 cm -1, corresponding to the v(N-H) of NH bonds of NH4 + not perturbed by hydrogen bonding interaction; v) the parallel growth of a broad and 1complex absorption in the 3150 2250 cm -1 interval with apparent maxima at -3000 and -2800 cm -1 (or minima at 3120 and 2880 cm -1) t.and a shoulder a t - 2650 cm-1; the O low frequency tail of this band t-" t~ seems to extend down to 2000 cm -1 t2 O where an apparent maximum at ffl ! r~ 0 2150 (or a minimum at 2250 cm -1) I, 1 I I is clearly observable. This broad and L._ v2 + V4 , ' complex absorption is the spectros2 XV 4 copic manifestation of the v ( N m H§ -) modes shifted to lower frequency by the hydrogen bonding perturbation. The complexity of this band is not necessarily the result of I I I I I 3000 2500 2000 1500 the presence of different types of interacting groups. In fact many of -I wavenumber cm the components previously mentioned, are clearly generated by Fermi resonance effects with low frequenFigure 4. Background subtracted spectra of increasing doses cy modes. For instance the first of NH3 adsorbed on H-Mord outgassed at 673 K. overtone of the mode at 1440 cm -1 corresponds closely with the apparent minimum at 2880 c m -1 (Evans window); similarly the other minimum at 3120 is explained by a second Evans resonance with the 1440 + 1680 c m -1 combination. For sake of brevity, in this paper we shall 00t go further in the interpretation of the NH4 § spectrum. Let us simply remark that the observed frequencies and the assignments closely correspond to theoretical calculations [4 and ref. therein) even if we are not yet in the position to choose among the different sitings of the NH4 § Other features (labelled with an asterisk) are also observable in the 2000-1750 c m -1 interval: these features are always observed upon adsorption of molecules and are consequently assigned to framework overtones modifications [19] not dependent upon the structure of the adsorbed molecule. With spectra 8 and 9 we start to clearly observe the modes of neutral NH3 (Vl: 3375, v3: 3315, v4:1625 cm -1) which is a base stronger than Z-. The presence of neutral NH3, not only means that the titration of the acid sites is reaching the final point, but also that the reaction

NH3/H-Mord

4)

Z-.-.NH4 + + NH3 ~ Z- + H3NmH +... NH3

is taking place with formation of new hydrogen bonded dimers (first step of NH4 § solvation with NH3). Also this problem will not be analyzed in detail for sake of brevity. This phenomenon is similar to that observed for water and once more suggest that the structure of the species present in the channels and cavities, is greatly infuenced by cooperative effects among the molecules.

221 As a final point, we illustrate the interaction of Py with H-ZSM-5 leading to the full consumption of Z - - H groups and stoichiometric formation of PyH +... Z- ionic pairs (Fig .5). For sake of brevity only one spec-trum obtained after exposure to excess Py and successive prolon-ged outgassing at RT is reported. Due to the absence of Py excess, the positive bands belong mainly to the PyH§ - ionic pair 0.8 only ( while the negative ones Py-H-ZSM-5 mainly correspond to consumed BrSnsted sites and to a small amount of silanols). We can divide the sprec-trum into two regions (3300-2000 and 1700-1300 cm-1). In the first region a very broad v(~t-w...z) band centred at --2800 cm -1 0.4(FWHM --_ 500-600 cm -1) is cundoubtely the v(NmH+---Z -) cstretching mode of the Py~H+-.-Z c~ ionic pair. As shemati-cally I I o reported in the figure, the bands ffl indicated with broken lines and modulating the v(Py~H+...Z -) stretch, are Fermi resonances with 0.0the combination and overtones of the ring (8a, 8b, 19a and 19b) and the ~5(Py--H+.--Z-) modes falling in the 1700-1300 cm -1 interval [20]. The detailed assignment will be I I I I I published elsewhere [19]. For the -1 time being, let us stress once more w a v e n u m b e r cm the importance of Fermi resonance effects, in explaining the high complexity of the spectra of Figure 5. Background subtracted spectrum of Py on Hstrongly hydrogen bonded systems. ZSM-5. The spectrum has been obtained after exposure Similarly to what observed with to Py excess and successive prolonged outgassing at RT. NH3, when higher filling conditions are investigated, PyH+...Py dimers are formed, characterized by a v (PymH+...Py) absorbing at lower frequency (because Py behaves as a base stronger than Z-). In a separate paper [19] we demonstrate that the band a t - 2200 cm -~ must be assigned to a small amount of the dimeric species. Conclusions. The host guest interactions in acidic zeolites are dominated by hydrogen bonding in neutral ZH...B complexes (B: weak base like N2, CO, C2H4, C2H2, propene, methylacetylene, acetonitrile and water) or in ionic Z-...+I-I~B pairs (B: strong base like NH3 and Py). The IR spectroscopy of these complexes is quite intriguing because of the presence of many types of resonance effects which modify the bands profile. In the case of H20, the cooperative effect of further water

222 molecules, favours the proton transfer and the formation of solvated HsOE+-nH20 clusters. The modifications induced on the spectrum of B--H + by interaction with Z- can be used to probe the basicity of the negatively charged framework.

Aknowledgements This work has been supported by the MURST and CNR (Progetto Strategico Tecnologie Chimiche Innovative)

References 1. G. C. Pimentel and A. L. McClellan, "The Hydrogen bond", W.H. Freeman and company Ed., London, 1960. 2. D. Hadzi and S. Bratos, in "The Hydrogen Bond/II Structure and Spectroscopy", P. Schuster, G. Zundel C. Sandorfy Eds., North-Holland Amsterdam, 1976. 3. C. Sandorfy, in "The Hydrogen bond/II Structure and Spectroscopy", P. Schuster,G. Zundel C. Sandorfy Eds., North-Holland Amsterdam, 1976. 4. J. Sauer, P. Ugliengo, E. Garrone and V.R. Saunders, Chem. Rev., 94 (1994) 2095. 5. A. Zecchina, S. Bordiga, G. Spoto, D. Scarano, G. Petrini, G. Leofanti, M. Padovan, C. Otero Are~in, J. Chem. Soc. Faraday Trans., 88 (1992) 2959. 6. S. Bordiga, C. Lamberti, F. Geobaldo, A. Zecchina, G. Turnes Palomino and C. Otero A r e s , Langmuir, 11 (1995) 527. 7. T. Yamazaki, I. Watanuki, S. Ozawa and Y. Ogino, Bull. Chem. Soc. Jpn., 61 (1988) 1039. 8. G. Spoto, S. Bordiga, G. Ricchiardi, D, Scarano, A. Zecchina and E. Borello, J. Chem. Soc. Faraday Trans., 90 (1994) 2827. and S. Bordiga, G. Ricchiardi, G. Spoto, D. Scarano, L. Camelli, A. Zecchina and C. Otero Are~in, J. Chem. Soc. Faraday Trans., 89 (1993) 1843. 9. A.G. Pelmenschikov, R. A van Santen, J. J~chen and E. Meijer., J. Phys Chem., 97 (1993) 11071. 10. A. Jentis and G. Warecka, M. Derwinski and J. A. Lercher, J. Phys. Chem., 93 (1989) 4837. 11. F. Geobaldo, C. Lamberti, D. Scarano, G. Spoto and A. Zecchina, in preparation. 12. S. Bordiga, R. Buzzoni, A. Zecchina and G. Spoto, in preparation. 13. D. Scarano, A. Zecchina, S. Bordiga, F. Geobaldo, G. Spoto, G. Petrini, M. Padovan and G. Tozzola, J. Chem. Soc. Faraday Trans., 89 (1993) 4123. 14. J. Sauer and M. M/iser, J. Phys. Chem., 98 (1994) 3083. 15. W.P.J.H. Jacobs, J.H.M.C. van Wolput and R.A. van Santen, Zeolites, 14 (1994) 117. 16. H. Jobic and M. Gzjzek, J. Phys Chem., 96 (1992) 1540. 17. J. M. Williams, in "The Hydrogen Bond/II Structure and Spectroscopy", P. Schuster,G. Zundel C. Sandorfy Eds., North-Holland Amsterdam, 1976. 18. R. Buzzoni, S. Bordiga, G. Ricchiardi, G. Spoto and A. Zecchina, J. Phys. Chem. submitted 19. S. Bordiga, R. Buzzoni, G. Ricchiardi, C. Lamberti, G. Bellussi, A. Zecchina, Langmuir to be submitted 20. V. P. Glazunov and S.E. Odinokov, Spectrochimica Acta, 38A (1982) 399.


223

Diffusion in Zeolites Douglas M. Ruthven Department of Chemical Engineering, University of Maine, Orono, ME, 04469

Abstract

Recent developments in the study of intracrystalline diffusion in zeolites by novel macroscopic methods and the results obtained by some of these methods are reviewed. For many systems there is a significant discrepancy between the macroscopic and microscopic (QENS, PFG NMR) diffusivity values. A possible explanation is suggested.

1. INTRODUCTION The earliest systematic studies of diffusion in zeolites (1) were reported in the 1930s yet, despite more than sixty years of study our understanding of this subject remains fragmentary and incomplete in many important aspects. The early expectation that, because of their structural regularity, zeolites would provide simple model systems for the study of micropore diffusion has been fulfilled to only a limited extent. While it is true that many of the broad features of the diffusional behaviour of these systems can be understood, in a general way, in terms of simple concepts (e.g. the size of the sorbate molecule relative to the channel diameter) the details are often confusing with many anomalies and inconsistencies. For many systems it is still unclear whether the apparent complexity is real or merely the result of poor experimentation and incorrect interpretation of the experimental data. For example, it is now well established that many of the earlier uptake rate measurements were corrupted by the intrusion of processes other than intracrystalline diffusion (external mass transfer resistance, heat transfer etc. (2-6)yet the reported intracrystalline diffusivity values continue to be quoted in the secondary literature. The "window effect", which is discussed in greater detail below, provides another example. The anomalous variation of diffusivity with chain length for linear alkanes in zeolite T reported by Gorring (7) has been quoted in many of the standard reference texts (8-11)and a detailed theory has been derived to account for this behaviour (12'13). Yet more recent experiments in a series of very similar zeolites (14) failed to reproduce the claimed effect, thus casting doubt on the validity of the interpretation of the original experiments. Diffusion in zeolites has been reviewed in detail by K/irger and Ruthven (is) and Rees 06). Therefore, in the present paper no attempt is made to provide an overall review; instead a number of recent topics related to the measurement and understanding of intracrystalline diffusion have been selected for more detailed discussion.

224 2. EXPERIMENTAL METHODS

A wide range of different methods have been developed for the measurement of intracrystalline diffusion. These can be divided into macroscopic methods (in which a transport flux is measured) and microscopic methods (in which the movement of the molecules is tracked directly)--see Table 1. The development of the PFG NMR technique in the mid-1970s 0719) represented a major milestone in that this approach provided the first reliable microscopic measurements of self-diffusivity. That the NMR values were orders of magnitude larger than most of the then accepted macroscopic values provided the first clear indication that the validity of much of the macroscopic data had to be questioned. Re-examination of the experimental conditions led to the conclusion that the impact of heat effects and extra-crystalline diffusion was much more significant than had been originally assumed C26). This stimulated the development of a range of more sophisticated macroscopic methods aimed at minimizing extraneous effects. Some of the recently reported results obtained with four such techniques are summarized below. Table 1

Summary of Experimental Techniques for Measuring Micropore Diffusion Macroscopic Methods /

Microscopic Methods NMR- Relaxation Times

Steady State

Transient

QENS

Membrane Permeation

Uptake Rate

PFG NMR

Effectiveness Factor

L

l

Chromatography DkeetMethods ZLC

IR Method Freq. Response

In comparing diffusivities measured by microscopic and macroscopic methods caution is needed. Most microscopic techniques measure the self-diffusivity ( ~ ) while most macroscopic techniques (except for tracer methods) measure the transport diffusivity (D). These quantities are different and for highly non-linear systems the difference may be more than an order of magnitude. A more meaningful comparison is between the self-diffusivity and the thermodynamically corrected diffusivity (Do), which is related to the transport diffusivity by the Darken expression: D = D O (denp/denC)T

(1)

where (denp/denC)T represents the gradient of the equilibrium isotherm in logarithmic

225 coordinates. For a Langmuir isotherm D = Do/(1-0 ) where 0 represents the fractional saturation, so the strong concentration dependence of D is evident. Although precise coincidence between and D O can be expected only in the low concentration limit, rough agreement (within a factor of 2 or 3) is to be expected over the entire concentration range (15). 2.1 Embedded Crystals Caro et al. (20) devised a technique to prepare oriented arrays of large silicalite crystals which were then coated with an impermeable copper film by a sputtering process. By careful abrasion, different faces of the crystal could be exposed, thus allowing diffusion in different directions to be measured. Diffusion (of n-hexane)in the transverse direction is much faster than diffusion along the crystal length, as is to be expected from the dimensiom of the crystals. However when diffusivities are estimated from the half-time of uptake the transverse diffusivity is found to be about three times the value for the longitudinal direction. These diffusivity values are close to the self-diffusivities obtained from PFG NMR measurements which show a similar degree of non-isotropy. However, this agreement is somewhat misleading since, for a proper comparison with NMR selfdiffusivities these values should be reduced by about an order of magnitude to allow for the Darken correction factor. Furthermore, the shape of the uptake curves does not conform well to the simple diffusion model, suggesting that processes other than simple diffusion may be significant. Experimental data showing the wide range of reported diffusivity values for this system are included in Table 2. Table 2 Diffusivity Data (~, D O xl06 cm2.s "1) for Linear Alkanes in Silicalite at 300K.

Method

~2H6

C3H 8

C6H14

Reference

Hernandez et al. (60) MD 6-7 --6 Caro et al. (37) PFG NMR 54 40 Heink et al. 21) 0.5 PFG NMR 1-12 Jobic et al. (22) 5 QENS 20 12 Kapteijn et al. (57) Membrane 2-6 0.65 Paravar and Hayhurst (25) 0.01 Membrane 0.02 0.07 van den Begin et al. (23'40) 0.2-0.3 FR 30 6 Hufton and Ruthven (24) 0.01 ZLC 1-1.5 Hufton and Danner (38) GC 1.5 0.4 Caro et al. (2~ 0.18 Uptake 10-310-7 Various authors (263~ Uptake M ) = Molecular dynamic calculations; PFG NMR = Pulsed field gradient nuclear magnetic resonance; QENS = Quasi-elastic neutron scattering; ZLC = Zero length column; GC = Gas chromatography; FR = Frequency response 2.2 ZLC (Zero Length Column) The ZLC method, introduced in the 1980s (31'32), is a very simple technique aimed at eliminating the intrusion of extraneous heat and mass transfer processes in a chromatographic flow system.

226 A very small quantity of sample (< 1 mg) is equilibrated with a low concentration of sorbate in an inert carrier (He or Ar) and then purged at a high flow rate under conditions such that the desorption rate is controlled by diffusion out of the crystal, rather than by convection. Since in a desorption experiment the baseline concentration is zero, with a sensitive detector (e.g. FID or mass spec.) the desorption rate can be measured accurately even at very low concentration levels. Consistency between limiting diffusivity values (Do) determined by the ZLC method and the NMR self-diffusivities has been demonstrated for several systems--see for example Figure 1. The original ZLC method is limited to the measurement of the limiting diffusivity at low sorbate concentrations (within the Henry region of the isotherm). However, in a recent modification, by the use of an isotopically tagged tracer (tracer ZLC)(a33) the method was extended to allow the measurement of tracer exchange (self) diffusivities over the entire concentration range. The sample is equilibrated with a known partial pressure of sorbate containing a proportion of the labelled species and the flow is then switched to a stream containing the same mole fraction of sorbate, all in the unlabelled form. The desorption of the labelled species is followed by mass spectrometry. For hydrocarbon sorbates the use of deuterated and non-deuterated forms is convenient although other kinds of isotopic labelling could also be used. A key advantage of this method is that thermal effects are eliminated although there remains the possibility of external diffusion resistance. Some of the recent results obtained by this method are summarized in Figure 2. For propane--5A (33) we find excellent agreement between the tracer ZLC data and the traditional ZLC values at low concentration. There is also reasonable agreement with the PFG NMR data--the trends being similar but with some difference in the absolute values. For methanol-NaX (35) the agreement with the NMR data is almost quantitative and the somewhat unusual maximum in the trend of diffusivity with loading is confirmed by both techniques. However, for propane and propene in NaX (34) there is a serious discrepancy, with the PFG NMR (36'37) data yielding much higher diffusivities, a much greater difference between paraffinic and olefinic sorbates and the opposite trend with loading. iO-s

Xe 0

zl0

ZLC A

CO 2

9

NMR

~

9 o

IO-B

Figure 1. Comparison of ZLC (Do) (44) and PFG NMR 05) ( ~ ) diffusivities for Xe and CO 2 in large crystals of 5A zeolite. From Ruthven (43). The recent FR data of Onyestyak et al. (4x) are also indicated ( - - - ) .

I I 10-7

10-8

I

2.0

3.0

,

I

4.0 103/T (K -) )

5.~

6.0

227

(b)

Ca) 1 E-07

'

3x10 -~

PFG

-I

r--:- - I - - ' -

-I

'

I

--'

I

lxlO -II ZLC ( t r a c e r )

.
= 3.3 at./cage < n > = 2.4 at./cage exchange of xenon atoms between adjacent a3 4 4 2 cages. Rates of exchange 5 " 3 6 i,'t, ,i'i F,, 2 1 can be obtained directly ~' t e-,, ____ ~ _ j ; "'u_,i ',,,I '~,,Q/.~___ /~,~ ~_,i" t i I '~_ by solving equation (3) together for all sites, t . = 1 O 0 ms ~. 0 t . = lOO taking into account the discrete cage populations [4]. We should emphasize that for this system with such a multisite exchange the complete set of rate constants can be estimated from a single F1 t . = 500 ms e~nt by using the t.:5OOms ~ ,~ 0 0 2D EXSY technique. Rate constants o 0 for xenon transfer between the cages of AgA zeolite depend on the number of xenon atoms in the cage, and for the cages with xenon occupancy lower than 6 230 180 130 ppm 230 180 130 are of the order of 1 s~ F2 (Fig. 6). The increase of the rate constants with Fig.5 129Xe2D EXSY NMR of xenon in AgA zeolite depending occupancy can be on the average loading and mixing time t~ explained as a result of decreasing sorption energy with rise of population [4]. From the variable temperature 2D exchange experiments the activation energy of xenon transfer between the different cages can be estimated as Ea = 45 + 10 kJ/mol, noticeably lower kn, s "1 than in NaA zeolite. Higher k, and lower "r" Ea for the intercage exchange in AgA than in NaA is most likely a consequence of higher hindrance at the intercage window w!,,0 by the sodium cations in the latter case. 2 4 6 8 n, ruTbe" d ) l n m a~-ns in o~age Another point of interest is that all the lines in the spectrum of xenon in AgA are Fig.6 Rate constants for xenon transfer shifted approximately 20 ppm to lower between the cages of AgA zeolite. field than for NaA.

ms

o

oaOo

o 0 9

00o

o000

249 4.4.

Adsorption of xenon in moleodar sieves with one-dimensional pore structure. To better tmdersta~ the relationship between lmXe NMR ~ ~ the pore sm/ctm'e it is irr~rtant to start with well-eaab~hed systen~. We have sadied with static, tragic angle spinning (MAS) and 2I~EXSY NMR the adsorption of xenon in ~ 5 , AI.J:~I 1, VPI-5, ~ 8 , ZSM-12 and SSZ-24. These rnok~a~ sieves are expected to be good nxxtel systems ~ of a) the absence of cations, whose presew.e can corr~licate the specwa b) their well established and relatively shrole crystal and pore structure. We have observed anisotropy of the ~29Xe c ~ shift which we relate to the structtwe of adsorption sites in ~ molecular sieves. The deper~nce of the anisotmpy on Xe loading in AI20-11 was analyzed in terms of a statistical dismbution of three types of xenon sites ppm (which have either 0, 1, or 2 neighboring sites occupied by other Xe, i.e., OXeO, XeXeO or XeXeXe), each with a characteristic shielding tensor, with fast exchange over 100 the three site types at room t e ~ [15]. The anisotropy is bading dependent for all the systen~ mxler 150 study, however detenmmtion of the anisotropies of the c o ~ n e n t s was possible only for AI20-11. The stall size of the channels in AI.PO-11 rmkes the n ~ of xenon types small, but when the size of the channels 150 100 50 ppm imxemses, more types of ~lsorption site can be realized and the analysis of anisotropy txx~ornes a t m c h bigger problem It is obvious, however, that the observed fig. 7 2I) EXSY spectrum ofhigh-baded anisotropy is not just a result of moulding of the xenon s m i l e of xenon in ZSM-12. electron cbud by the shape of the channels, as suggested Line in high field is from Xe in the previously by Springuel-Huet [16], ~ s e the interpart~le space anisotropy is observed even for xenon adsorbed in VPI5 - mo~ sieve with armch larger channet While exchange averaging within channels must be rapid in order to give the single anisotropic lines observed, 2D EXSY spectra show that slower exchange processes are occurring:

tm = 0 . 5 m s

tm = 3 0 m s

tm = 10 m s

FI 90 I00 ii0

''"il''''l''''l''''l'''' 110 100 90 ppm

Fag. 8

'

I

'

[

100

'

i

'

I

,

F2

i , , , i , , , , 1 , , , , i , , , , ' l , , , ,

~0

100

90

ppm

21~EXSY specam oflow-loaded sample of xenon in ZSM-12 with different mixing times.

250 a) Fig 7 shows a spectrum for a high loading of xenon in ZSM- 12. Intense cross-peaks between the anisotropic signal from xenon inside the channels and the high field line from xenon in the intercrystalline space indicate a significant degree of exchange between these two regions. The anisotropy is preserved in the cross-signal lineshape ( which again shows that xenon in each crystaUite has a single frequency due to rapid intra-channel exchange), b) There is also slow interparticle xenon exchange ( perhaps tlrAiated by the gas phase ). Fig. 8 shows spectra of a low xenon loading in ZSM-12 with different mixing times. At longer mixing times the intrazeolite signal develops off-diagonal intensity in a pattern which indicates exchange of xenon between crystallites with different orientations, i.e. exchange between components of the shielding tensor

5.

Conclusions

The resuks of the present study graphically demonstrate the utility of 2D-EXSY 129XeNMR in studies of adsorption and diffusion in molecular sieves. The technique is highly suitable for studying the porous structure of zeolites and permits a quantitative description of inter- and intra-crystallite mass transfer processes. Based on the experimentally rrr.asured rates of xenon exchange, it is possible to obtain inforrmtion on diffusion as well as the possible paths of exchange, and the distribution and availabih'ty of adsorption sites. Themxxtynamic characteristics of the adsorption system can also be detem'fined from variable temperature 2D-EXSY experiments. The experiments show that exchange processes are an important factor in xenon NMR spectra in porous which must be accounted for when correlations with the porous structure are made.

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11 12. 13. 14. 15. 16.

Ito, T., Fraissard, J." J.ChenxPhys. 7__6,5225 (1982) Ripm~ster, J.A." J.AnxChem.Soc. 104, 289 (1982) Jeener, J., Meier, B.H., Bachmann, P., Ernst, R.R.: J.ChenxPhys. 71, 4546 (1979) Larsen, R.G., Shore, J., Schmidt-Rohr, K., Emsley, L., Long, H., Pines A., Janicke M., Chn~lka B.F.: Chem.Phys.I~tt. 214, 220 (1993) Moudrakovski, I., Sayari, A., Ratcliffe, C., Ripme~ster, J., Preston, K.: J.Phys. Chem. 98, 10895 (1994) Drobny, G., Pines, A., Sinton, S., Weitkan~, D., Wemn~r,.D.: Faraday Syrup. Chem. So0,

XeVs,

(3)

j=l i.e. in vector notation

a t (l=Vx ~)Vx (1 ,

(4)

where ~ = { Dij }Ni,j

=

1

denotes the matrix of diffusion coefficients and t~= {a i }Ni= 1 the vector

of the local partial sorbate concentration. To describe the sorbate concentration ~s(t)=~

= ~ Y ( ~ ( t ) ) , t > O,

x, t)lx E

(5)

FVs

at the outer surface Fvs of sorbent particles, non-linear sorption isotherms 3: = fi i= 1 are permitted. The average concentration ~t is related to the local sorbate concentration by ~y,t)=

f

~x,t)dx,

t>0,

ye [0,L],

(6)

Vs~y where v s ~ y stands for the volume of all sorbent particles that are located in the column section at y ~ [ 0, L ]. To simulate mixture sorption kinetic experiments utilizing the above system of equations, the general solution of the transport equation (3) for variable boundary conditions is calculated from that for constant boundary conditions by solving systems of Volterra integral equations of the second kind. The solution for mixtures under constant boundary conditions is reduced to the knowledge of that for a single component and the solution of an eigenvalue problem as proposed in [ 14]. The technique allows the calculation of the sorbate concentrations c to fit

272 experimental ZLCC multi-component data as well as the determination of the vector ~ of average concentration in order to evaluate VrlR mixture kinetic data. 3. M O D E L

ASSUMPTIONS

If the experimental arrangement allows one to neglect the influence of the bed length L on the concentration vector ~, the gas phase concentration will be fully described by the values of the function C at the entrance and at the end of the column, i.e. by ~ 0, t ) and ~ L, t ) which are, of course, time-dependent functions. The concentration at the entrance of the column can easily be determined in blank experiments, i.e. C~0,t)=~(t),t>0.

(7)

Thus, ~ L , t ) remains as the only function that describes gas phase behavior. Assuming sufficiently high smoothness with respect to the space co-ordinate y, the Taylor expansion of the function C at y = L into a polynomial yields

u{ }=U~y~L,t)L e(L,t)-~(t)

2 Oy 2 e ( L ' t ) + O

(L2)

'

(8)

where O( L 2 ) stands for the quadratic Taylor remainder. Neglecting this remainder and setting e(t)-e(L,t),

~(t)-~L,t),

(9)

etc., in order to shorten the formulas furtheron, the mass balance equation (1) yields ot~t)+--~-~-~t)=13

6(t)-~(t)

, t>0,

(10)

where the parameter u [3=L=IV

V

~, I-IV

S

I

(11)

is introduced. Using such an approximation, the dispersion Dax is not neglected but it is equated to Lu Dax = 2

(12)

that makes the equations system stable. Equation (10) represents the model transition from a finite length column to that with zero length properties. Furthermore, the matrix of diffusion coefficients is assumed to be independent of sorbate concentration. This assumption is valid as far as the concentration change during an experimental kinetic run is sufficient enough. By ensuring small changes in concentration during kinetic experiments, the constancy of the diffusivity matrix fl~ can be assumed, i.e. equation (4) can be rewritten as b ~)t t~= ~ Axa.

(13)

The set of equations (5), (7), (10) and (13) has to be solved to simulate the sorption kinetic process under consideration.

273

4. MODEL SOLUTION The principle of superposition ([ 15] p. 180) as the basis of the Volterra integral equation approach, has its mathematical expression as a relationship between average sorbate concentration and corresponding surface concentration. The vector A of average partial concentrations is determined by t

~ t)- a0 : f~'(t- s) { as(S)- a0 } ds,

(14)

0 1

where ffC =

- ~Ni, j = 1 d e n o t e s Hij

the matrix of normalized solutions for constant boundary

conditions. The matrix function ffChas the following properties. The general solution ~onst for constant boundary conditions is expressed by ~i~c~

- (~_ = fie(t) { (i0+ -(10_ } .

(15)

The corresponding vector (l c~ of local concentrations fulfills the transport equation (3) under constant initial and boundary conditions: (tc~

. t ) = tl0_ if t < 0

(i'c~

. t ) = ~ + if t > 0 '

(16)

where tl0_ and tl0+ stand, respectively, for the constant surface concentration prior to and during a kinetic experiment. In the case of single gas sorption kinetics, the matrix function ~ degenerates into a scalar function that is represented by the normalized solution I:I for constant boundary conditions. The solutions fI are well known from literature (cf. [16, 17]). For simple geometric shapes of particles, e.g. of a microporous solid, IZI is given as cos( k n ) = 0 for 0 = 0 (plate),

oo

- 'I: ) = 1 - 2 (0 + 1) E H(

~ , J0 ( kn ) = 0 for 0 = 1 (cylinder), e-k~'l:

(17)

sin( k n ) = 0 for 0 = 2 (sphere). n=0

As shown in [14], in general, ffCis of the form fie( t ) = I- ~-I h( A t ) ~3

(18)

i= 1" T h e q u a n t i t i e s A = { A i } Ni= 1 a n d ~ = { Bij }N i, j= 1 are, where h( A t ) = diag { 1 - ft( Ai t ) }N respectively, the eigenvalues and the transposed eigenmatrix of the diffusivity matrix fl), i.e. ( ~ - A i I)B.,T = O , i = I(1)N, where B.i =

Bji

= 1 are the transposed eigenvectors of the diffusivity matrix fl~.

(19)

274 Integration of the mass balance equation (10) leads to t

) f{o ~ s ) - e B ( S ) } ds

e..(t)-eo+~

~ t ) - ~ 0 =13

'

t>O

(20)

9

Using (14), & can be substituted by the surface concentrations ~s and equation (20) yields t

t

1-13 e-(t) - dO + - - f ~ ' ( t - S ) { gs(S)- (~ } ds = 13f { e..(s) - ~ ( s ) } ds o o

'

t,O

9

(21)

Due to relation (5), equation (21) can be rewritten as t

t

e_(t , - ~ + - - 7 - - f :5c'( t - s ,{ 5(e,,( t , ) - 3 : ( ~ )} ds=[~ f { e,(s ,-eB( s , }ds, t > 0 . o o

(22)

This equation system represents a system of non-linear Volterra integral equations of the second kind with respect to the concentration vector ~ The corresponding vector 0~of average concentration can be calculated as a function of e utilizing relations (5) and (14), i.e. t

~ t , - (10 =

f :rc'tt-s ){ 5(e_,( t ,

) - 5 ( e 0 )} ds.

(23)

0 To solve non-linear integral equation systems, a broad variety of methods exists (cf [ 18]). Quadrature methods were successfully applied to simulate sorption kinetic curves [ 19, 20]. 5. S I M P L I F I C A T I O N S In cases where the gas phase concentration can be considered as being independent of the sorbed amount (e.g. if e-->1 {small sorbent volume} or [3--->~ {high flow rate}) and where the concentration change is small enough to ensure a sufficiently accurate linear approximation of sorption isotherm, formula (23) can directly be utilized to explicitly calculate ~. The rigour of this presumption becomes evident from equation (20). For example, utilization of exponential functions to express surface concentration ~s (as applied in [10, 11]), would lead, due to (20), to a representation of ~ in terms of exponential functions. Therefore, it is apparent, that the above presumption contradicts the validity of the column mass balance equation (10). Nevertheless, cases may exist for which such an approach is sufficiently accurate. A linear isotherm in the case of multi-component sorption processes is considered, i.e. N fi(•) = Z G j cj or 5 ( C ) = ~ C i=l

(24)

275 is assumed to be valid for an appropriate ~ =

{ Gij }Ni,j = 1" Furthermore,

the gas phase

concentration vs. time dependence might be given as (25)

r t )- C0 = ( 1 - e-at )(Coo- e0 ), t > 0 ,

where o~ > 0 is an appropriate constant and the vectors e0 and Coo denote the respective concentrations at the initial and at the equilibrium state. As far as the sorption phase concentration has no detectable influence on the gas phase concentration, this relation (25) with respect to the gas phase concentration should reflect all apparatus effects (as finite valve opening, delays due to finite flux, etc.) superimposed upon the intrinsic sorption kinetics. Besides, a piecewise linear approximation instead of (25) may also be utilized within this approach. Under above premises, the vector ~t of the average sorbate concentration can be calculated explicitly as

t 0.( t )- 120: f ( 1 - e-at ) ~ ' ( t- s )ds ~ (Coo- r ), t > 0 . 0

(26)

The integrand in (26) is a matrix function, i.e. the integral stands for the componentwise integration of this matrix. The matrix function ~; = Hij i, j = 1 can be calculated explicitly for mixtures up to ternary ones. For the most often considered case of a binary mixture ( c f [ 14] part II), one obtains

__1__1( c2 c 2H(A 2 t ) - c

ftll(t)=l-cl

llq(A it)

)

1

fi12 ( t )=

-cc1~

I5121( t ) =

c 1 c2 c~-c2 (I2I( A2 t )- I-I( A1 t ) )

1 (

H22 ( t ) = l + c l _ c 2

1

c 1 I2I(A 2 t ) - c 2I=I(A 1 t)

, t > O,

(27)

)

where A1/2=2 D l l + D 2 2 +

(Dll+D22)2+4D12D21

, Cl/2 =

D21

=A1/2-Dll

Thus, the integrals in formula (26) can be calculated by oo

t f0 I=I'(A( t - s ))(1-e-aS)ds = 1- 2 ( 0 + l )

oo

e 95%) are obtained using stoichiometric concentrations of reactants which almost suppress the successive Michael condensation between ethyl cyanocinnamate (IV) and ethyl cyanoacetate (11). The initial rates were determined in DMSO at 80~ with 0.3 mol/I concentrations of each reactant using 0.2 g of catalyst. Figures 4 and 5 show the variations of the initial rate and of the amount of desorbed CO2 up to 550~ v e r s u s the cesium loading for the two kinds of solids. Concerning the CsNaX n Cs solids we see that the initial rate increases up to 11 Cs atoms per unit cell. Then, a change in the slope and therefore, in the catalyst activity is observed. Initial rates with CsNaY n Cs solids are much smaller than with the former zeolites. Curves going through a maximum are obtained. 50

350

~o O X

E3

9

9 300 ~"

40

-~ -

o E

O 0

250 B 200 o

30

--

150

20

B

"O

:D

100

.Q 9 }"'-

o 03

10

r v

x

50

a 0

, 5

0

, , , 9 14 18 Encapsulated Cs (atoms / u. c.)

, 22

---" o(.o

0 27

Figure 4 Variation 9 of initial rates and amounts of desorbed CO 2 with cesium loading on CsNaX n Cs zeolites 9 Desorbed CO 2 up to 550~

4.

9 Initial rates

DISCUSSION

Previous results [1] showed that the 0 0 2 adsorption on the exchanged CsNaX and CsNaY zeolites occurs on the remaining sodium cations. Thus, the increase of this adsorption on the modified zeolites can be assigned first to the adsorption onto the basic oxides and moreover, on the basic function of these species. The peaks at 250~ and 150~ are devoted to the active cesium oxides with an intracristalline location whereas location of cesium oxide over the external surface of the zeolite leads to CO2 desorptions at temperatures higher than 550~ Effectively, Figure 2 shows that, for Ioadings higher than 16 Cs atoms per unit cell on the CsNaX n Cs solids, the amount of desorbed CO2 up to 550~ is nearly constant. However, a second desorption peak appears at high temperatures explained by an external

324

O

16

70

14

60

m

~

x 12 ~

0

E

5o 40

8

0 0

6

30

"0

~

4

o

2

0

.

g

10

v

..Q

_~.

20 10

3

0 m

3 = n

~

x 0

0

0

5

10

15

20

25

Encapsulated Cs (atoms / u. c.) Figure 5 Variation 9 of initial rates and amounts of desorbed CO 2 with cesium loading on CsNaY n Cs zeolites, 9 Desorbed CO 2 up to 550~

9 Initial rates

deposit of cesium oxide (Figure la). The unique peak described for the CsNaY n Cs zeolites (Figure l b) seems to indicate that all the oxide species are located inside the cages of these solids. Consequently, inside the (z-cages of CsNaX zeolite, the basic species are assumed to be cesium oxide Cs20 (slope = 0.44) which upon CO2 sorption would give a carbonate form, the structure of which has no longer been determined. Our results show clearly that another basic oxide is generated inside o~-cages of CsNaY zeolite, which is proposed to be a cluster (0s20)2 (slope = 0.23). Moreover, the variation of the desorption maximum from 250~ to 150~ for the modified CsNaX and CsNaY zeolites respectively, indicates that the encapsulated oxides (Cs20) generated inside the cages of the host CsNaX zeolite are more basic than those (0s20)2 inside the CsNaY zeolite. These results are confirmed by the volumetric measurements (Figure 3). The higher value of the micropore volume for CsNaY than for CsNaX can be explained, as for NaY and NaX zeolites, by the lower number of extraframework cations in the former than in the latter solid [1, 2, 6]. Micropore volumes decrease faster but remain higher for CsNaY n Cs than for CsNaX n Cs up to 9 cesium atoms per unit cell with increasing the cesium loading and decreasing the available void volume. For higher Ioadings the reverse is observed; the high decrease in available CsNaY n Cs pore volume and the conservation of cristallinity by XRD indicate a diffusional limitation. Thus, the variation of the micropore volume accounts for the formation of a larger cluster inside the Y than the X zeolite. The Knoevenagel condensation between ethyl cyanoacetate and benzaldehyde

325

(scheme 1) was chosen as a model reaction. In fact, the activation of the methylene group by electroattracting substituants enables soft conditions for the comparison of the solids [7]. The kinetic and mechanism studies in DMSO [8] show that the reaction is first order in each reactant and that the determining step is the condensation between the adsorbed reactants. The activity data are in good agreement with the TPD results (Figures 4 and 5). Modified CsNaX n Cs zeolites are much more efficient catalysts than CsNaY n Cs solids and therefore, as indicated by CO2 desorption temperatures, encapsulated cesium oxides in the former hosts are more basic than those in the latter ones. On CsNaX n Cs zeolites, base catalysis by intrazeolitic Cs20 leads to a linear initial rate increase until 5.5 species per unit cell. Above this value it is proposed that catalysis by species located on the external surface (Figure 1) takes place. On CsNaY n Cs zeolites no external species are formed. Thus activity (Figure 5) is related with the number of occluded oxides until 4.5 oxides per unit cell. Above, the same limitation as that shown by TPD and volumetric measurements is observed. 140 % "-X 120 r .m

E "~ O

E

10

x

04 O

x

100

8

e.w

E --

80

m

6

O

60

E

v

4

(D

40

...+-J~--~ 20 ..E c 0. 0

--~

2

c"

2'o

2s

CO2(mol/g ) x 105 Figure 6 Activity 9 of encapsulated cesium oxides

0

o

1'o 1'5 2'o 2's

30

Cs2CO3 (mole) x 104 Figure 7 Activity 9 of cesium carbonate

x CsNaX n Cs - 4- - C s N a Y n Cs Catalytic activities of intrazeolitic cesium oxides have been determined (Figure 6). Linear correlations are obtained between the initial rates and the amount of desorbed CO2 for Ioadings up to 11 and 9 cesium atoms per unit cell for CsNaX n Cs and CsNaY n Cs zeolites, respectively. These correlations lead to the calculation of the turn over frequency taking into account that the number of desorbed CO2 mole per gram reflects the number of basic sites : 340 and 110 mol / I / m i n for the intrazeolitic Cs20 or (Cs20)2 site, respectively. In the same conditions the activity of cesium carbonate is about 30 mol/I / min per mol (Figure 7). Our results show the effect of the composition of the host zeolite on the nature

326

and the basicity of the guest oxides which differ from that of the bulk oxide [9,2]. Taking into account that the number of exchanged cesium cations is nearly the same in CsNaX and CsNaY zeolites, the change arises from the number of sodium cations. It may result in somewhat different occupancies of cation sites II and III which affect the supercage environment and thus, the cage free volume, the electric field and the basic character [10]. The formation of the intrazeolitic oxide species in the activation step seems to be governed by the stability of the ionic oxide in interaction with the negatively charged oxygen framework and with the extraframework cations. The difference in basicity of the two species Cs20 and (0s20)2 can be explained by the coordination of the oxide [11]. Such a difference was not observed until now for intrazeolitic cesium oxides [2]. A curve similar to that described for CsNaY n Cs zeolites is explained rather by physicochemical changes in the occluded cesium oxide particules. The same conditions of activation of the solids allow us to propose the same oxidation state for the oxide species. 5.

CONCLUSION

A good agreement between the physical characterization of intrazeolitic cesium oxides by CO2 TPD and N2 volumetry and their catalytic activity in the model Knoevenagel condensation leads to the conclusion that the nature and the basicity of these oxides depend mainly on the composition of the host zeolite. Two different structures are proposed Cs20 and (0s20)2 inside the o~-cages of CsNaX and CsNaY, respectively, the former being more efficient than the latter. In relation with the cluster size, diffusional limitations can take place in the second case. REFERENCES

1. M. Lasp~ras, H. Cambon, D. Brunel, I. Rodriguez and P. Geneste, Microporous Mater., 1 (1993) 343; M. Lasp~ras, H. Cambon, D. Brunel, I. Rodriguez and P. Geneste, Microporous Mater., (1995) submitted. 2. P. E. Hathaway and M. E. Davis, J. Catal., 116 (1989) 263; P. E. Hathaway and M. E. Davis, J. Catal., 116 (1989) 279; J. C. Kim, H.-X. Li and M. E. Davis, Microporous Mater., 2 (1994) 413. 3. H. Tsuji, F. Yagi and H. Hattori, Chem. Lett., (1991) 1881; F. Yagi, N. Kanuka, H. Tsuji, H. Kita and H. Hattori, Stud. Surf. Sci. CataL, 90 (1994) 349. 4. S. Kawi, J. R. Chang and B. C. Gates, J. Am. Chem. Soc., 115 (1193) 4830. 5. I. Rodriguez, H. Cambon, D. Brunel, M. Lasp~ras and P. Geneste, Stud. Surf. Sci. CataL, 78 (1993) 623. 6. S. S. Tamhankar and V. P. Shiralkar, J. Inclusion Phenomena and Molecular Recognition in Chem., 17 (1994) 221. 7. A. Corma, V. Forn~s, R. M. Martin - Aranda, H. Garcia and J. Primo, Appl. Catal., 39 (1990) 237. 8. I. Rodriguez, Thesis~ Montpellier, U. S. T. L., (France), June 1995. 9. A. Jentys, R. W. Grimes, J. D. Gale and C. R. A. Catlow, J. Phys. Chem., 97 (1993) 13535. 10. D. C. Doetschman, D. W. Dwyer, J. D. Fox, C. K. Frederick, S. Scull, G. D. Thomas, S. G. Utterback and J. Wei, Chem. Physics, 185 (1994) 343. 11. G. Pacchioni, J. Amer. Chem. Soc., 116 (1994) 10152.


327

H Y D R O T H E R M A L AND A L K A L I N E STABILITY OF H I G H - S I L I C A Y Z E O L I T E S G E N E R A T E D BY C O M B I N I N G SUBSTITUTION AND STEAMING W. Lutz 1, E. L0ffier 2 and B. Zibrowius 3 1 Institut for Angewandte Chemie Berlin-Adlershof e. V., Rudower Chaussee 5, D- 12484 Berlin, Germany 2 Analytik - Umwelttechnik - Forschung - GmbH, Rudower Chaussee 5, D- 12484 Berlin, Germany 3 Institut for Brennstoffchemie und physikalisch-chemische Verfahrenstechnik der RWTH Aachen, Worringerweg 1, D-52074 Aachen, Germany Y-type zeolites dealuminated by steaming (DAY-T) are more stable in water and alkaline solutions than those dealuminated by substitution (DAY-S). However, DAY-S zeolites have a higher adsorption capacity and are more hydrophobic. A dealumination by substitution and subsequent steaming yields a zeolite DAY-ST that combines the high chemical stability of DAY-T with the high adsorption capacity and hydrophobicity of DAY-S. I. I N T R O D U C T I O N Dealuminated Y zeolites (DAY) are applied in technical processes as catalysts or carriers of catalytic active ingredients. Recently, they have become more and more interesting as incombustible adsorbents of organic pollutants from water and air. By steaming a NH4Y zeolite, a stable molecular sieve DAY-T is formed [1]. Mesopores as well as non-framework aluminium generated during this thermochemical process (T) reduce the sorption capacity of the zeolite. Since the degree of dealumination is limited to values of Si/A1 _ --- > --- > --- > --- > --- >

Co 2 + V4 + Fe 2 + Ti 3 + As 3 + Sn2 +

E (ev)

Reduction

H 2 0 2 decomp.

1.82 1.00 0.77 0.06 0.56 0.15

easy moderate moderate difficult moderate moderate

fast moderate moderate difficult moderate moderate

Catalytic oxidations with H 2 0 2 involve two types of H 2 0 2 cleavage / activation: (i) heterolytic (ionic) and (ii) homolytic (radical). In general, the heterolytic mechanism predominates in the case of those catalysts involvin~ metal ions (like TS-1) having low redox potential (difficult redox transition M n + ~ M~n-l) +/M(n-'2) + ), while the radical mechanism is predominantly operative i n cases (like V) where redox potentials are high with easy redox transition M n§ ~ M(n-1) +/M(n-2) 4. 4. ACKNOWLEDGEMENTS We thank members of Catalysis Division for providing many of the results. The work was partly funded by the European Commission (contract no. Cll-CT93-0361). REFERENCE

,

3. 4.

.

6. 7. 0

.

10. 11. 12. 13. 14. 15. 16.

W. H61derich, M. Hesse and F. Naumann, Angew. Chem., Int. Ed. Engl. 27 (1988) 226. W. H61derich and H. Van Bekkum, Stud. Surf. Sci. Catal., 58 (1991) 631. B. Notari, Stud. Surf. Sci. Catal., 37 (1988) 413. P. Kumar, R. Kumar and B. Pandey, J. Indian Inst. Sci., 74 (1994) 293; P. Kumar, R. Kumar and B. Pandey, Synlett., to appear as an Account in April/May issue of 1995. P. Ratnasamy and R. Kumar, Catal. Lett., 22 (1993) 227. A.V. Ramaswamy and S. Sivasanker, Catal. Lett., 22, (1993) 239. A.V. Ramaswamy, S. Sivasanker and P. Ratnasamy, Microporous Materials, 2 (1994) 451. N.K.Mal, V. Ramaswamy, S. Ganapathy and A.V. Ramaswamy, J. Chem. Soc. Chem. Commun., (1994) 1933. P. Ratnasamy and R. Kumar, Catal. Today, 6 (1991) 329. M. Taramasso, G. Perego and B. Notari, U.S. Pat. 4,410,501 (1983). A. Thangaraj, R. Kumar, S.P. Mirajkar and P. Ratnasamy, J. Catal., 130 (1991) 1. J.S. Reddy and R. Kumar, J. Catal. 130 (1991) 140. J.S. Reddy and R. Kumar, Zeolites, 12 (1992) 95. D.P. Serrano, H.-X. Li and M.E. Davis, J. Chem. Soc. Chem. Commun., (1992) 745. M.A. Camblor, A. Corma and J. Perez-Pariente, Zeolites, 13 91993) 82. A. Miyamoto, D. Medhanavyn and T. Inui, Appl. Catal. 28 (1990) 89.

376 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44.

M.S. Rigutto and H. Van Bekkum, Appl. Catal., 68 (1991) L1. P.R.H.P. Rao, A.V. Ramaswamy and P. Ratnasamy, J. Catal., 137 (1992) 225. P. R. H. P. Rao and A.V. Ramaswamy, Appl. Catal.A: 93 (1993) 123. A. Tuel and Y. Ben Taarit, Zeolites, 14 (1994) 18. A. Bhaumik and R. Kumar, Microporous Materials (communicated). K.M. Reddy, I. Moudrakovski and A. Sayari, J. Chem. Soc., Chem. Commun. (1994) 1491. K.R. Reddy, A.V. Ramaswamy and P. Ratnasamy, J. Chem. Soc., Chem. Commun. (1992) 1613. T. Sen, M. Chatterjee and S. Sivasanker, J. Chem. Soc. Chem. Commun., (1992) 207. N.K. Mal and A. V. Ramaswamy, Appl. Catal., in press N.K. Mal, A. Bhaumik, R. Kumar and A.V. Ramaswamy, Catal. Lett., in press A. Bhaumik and R. Kumar, J. Chem. Soc. Chem. Commun., (1995) in press. A. Bhaumik and R. Kumar, Indian Pat., 1994 (applied); A. Bhaumik and R. Kumar, Catal. Lett. (communicated). T. Tatsumi, M. Nakamura, K. Yuasa and H. Tominaga, Catal. Lett., 10 (1991) 259. T. Tatsumi, M. Yako, M. Nakamura, Y. Yuhara and H. Tominaga, J. Mol. Catal., 78 (1993) IA1. S.B. Kumar, S.P. Mirajkar, G.C.G. Pais, P. Kumar and R. Kumar, J. Catal. (accepted). A. Bhaumik, R. Kumar and P. Ratnasamy, Stud. Surf. Sci. Catal., 84C (1994) 1883. R. Kumar, G.C.G.Pais, P. Kumar and B. Pandey, J.Chem. Soc. Chem. Commun. (1995) in press P. Kumar, V.R. Hegde, B. Pandey and T. Ravindranathan, J. Chem. Soc., Chem. Commun., (1993) 1553. R. Joseph, A. Sudalai and T. Ravindranathan, Tetrahedron Lett., 35 (1994) 5493. S. Murata, M. Miura and N. Nomura, J. Org. Chem., 54 (1989) 4700. J.P. Snyder, V.T. Banduroo, F. Darack and H. Olsen, H. J. Am. Chem. Soc., 96 (1974) 5158. H.R. Sonawane, A.V. Pol, P.P. Moghe, S.S. Biswas and A. Sudalai, J. Chem. Soc., Chem. Commun., (1994) 1215. J.S. Reddy and P.A. Jacobs, J. Chem. Soc., Perkin Trans-I (1993) 2665. R.S. Reddy, J.S. Reddy, R. Kumar and P. Kumar, J. Chem. Soc., Chem. Commun., (1992) 84. R. Kumar, J.S. Reddy, R.S. Reddy and P. Kumar, in Selective Oxidation in Petrochemistry; M.Baerns and J.Weltkamp, (eds.), Goslar 1992, DGMK Tagungsbeicht No. 9204, pp 367. A. Bhaumik and R. Kumar, J. Chem. Soc. Chem. Commun., (1995) 349. P.R.H.P. Rao, A. Thangaraj and A.V. Ramaswamy, J. Chem. Soc. Chem. Commun., (1991) 1139. R.C. Weast (ed.), Hand Book of Chemistry and Physics, CRC Florida, 1980-81, pp. D155.


377

Novel model catalysts containing supported MFI-type zeolites N. van der Puil, E. C. Rodenburg, H. van Bekkum and J. C. Jansen Laboratory of Organic Chemistry and Catalysis, Delft University of Technology Julianalaan 136, 2628 BL Delft, The Netherlands.

Abstract The synthesis, characterization and catalytic testing of composite catalysts consisting of a supported catalytic phase and a silicalite-1 coating are reported. Si/Fe203/MFI and Si/Cr203/MFI composites were tested in the dehydrogenation of n-heptane at 500~ The composites showed selectivity for the formation of methylcyclohexane, while the uncoated metal oxide particles showed selectivityfor toluene. TiO2/Pt/silicalite1 composites were tested in the competitive hydrogenation of a mixture of 1-heptene and 3,3-dimethyl-1butene. The composite showed a high selectivity for the conversion of the linear olefin, which was not observed for a TiO2/Pt catalyst. These experiments indicate the feasibility and shape selectivity of the composite catalysts. 1. I N T R O D U C T I O N The synthesis of thin layers and coatings of different types of zeolites on supports such as aluminum, stainless steel, mullite, quartz has been reported recently [1]. The composites may be applicable as catalysts, membranes and as chemical interface in sensors. Coatings of silicalite-1 with a random orientation on sintered metal supports have been prepared and tested as membranes [2]. A prerequisite for optimization of the applications is the ability to prepare very thin and oriented layers of zeolites. In the case of MFI-type zeolites, various orientations of the supported crystals can be obtained under specific synthesis conditions [3]. Thin layers of laterally oriented crystals (b-direction perpendicular to the support surface) for sensor purposes were reported [4]. Axially oriented crystals of ZSM-5 (c-direction perpendicular to the support surface) on stainless steel provide a dust-free reactor set-up with advantageous heat transfer properties and a low pressure drop [5]. The objectives of the present study are the preparation and testing of well defined model catalysts, containing uniquely oriented crystallites of MFI on single crystal supports (silicon and rutile). In these composites the catalytic sites are present at the interface between support and zeolite, either resulting from the combination of two different oxide phases [6,7],or as a separate phase obtained by modification of the support surface. The support material serves to provide a well defined interface, which offers high stability for the catalytic sites. In this way it may be possible to stabilize catalytic sites exposed to the zeolite pores, which can not be obtained in the zeolite framework. Since the catalytic sites are created on the support before zeolite synthesis, no ion exchange capacity of the framework is needed. It is therefore possible to obtain a true monofunctional zeolite catalyst. Moreover, no catalytic sites will be present at the external surface of the zeolite phase. If the zeolite layer is not chemically bonded to the catalytic phase, only reactant selectivity is expected to occur. If however chemical bonding exists between the zeolite layer and the catalytic material, it is expected that additional transition state shape

378 selectivity will occur during reaction at the catalytic sites. In the ideal case, this might also result in an "end-on" effect by which only the ends of molecule chains are able to react. Here we report on the synthesis, characterization and catalytic performance of laterally oriented crystallites of silicalite-1 on different types of catalytic phases and/or supports. First composites consisting of a Si support, iron(III) and chromium(III) oxide particles and a silicalite-1 coating are reported, which were tested in the dehydrogenation/cyclization of n-heptane. Secondly, silicalite-1 coatings on TiO 2 supported particles of platinum and their performance in competitive hydrogenation reactions are described.

2. EXPERIMENTAL Silicon (100) wafers (10xl0x7 mm, one side polished, DIMES) and TiO 2 wafers (10xl0xl mm, both sides polished, Single Crystal Technology b.v.)were used as supports. Before use, the Si-wafers were treated by a special cleaning sequence in order to remove organics and metal contamination [3]. Metal oxide coatings on the polished side of the Si-wafers were obtained by a spincoating technique [8]. Ethanolic solutions of 0.1 wt% FeC13.6H20 (p.A. Janssen Chimica) and CrC13.6H20 (p.A. Janssen Chimica) were brought onto a rotating support, while flushing with nitrogen. The precursors were calcined in air at 450~ for 6 hours. The metal oxide coatings were studied by X-ray Photo-electron Spectroscopy (Phi 5400 spectrometer), ICP/AES (Varian SpectrAA 300)and Scanning Electron Microscopy (Philips XL20). Platinum particles on TiO 2 were obtained by sputtering on one side for 6-24 seconds at 300 mV and 160 Watt. Here TiO 2 was chosen as a support, since platinum does not stabilize on Si. At temperatures above 150~ platinum diffuses through the surface silicon oxide layer and forms a stable PtSi phase [9]. High Resolution SEM measurements of the sputtered samples were carried out with a Jeol JSM-6320F microscope. Zeolite synthesis conditions were refined to promote crystallization of the zeolite coating, and to avoid dissolution of the support phase. The silicalite-1 coatings were synthesized from TEOS (98%, Janssen Chimica), TPAOH (25%, Chemische Fabriek Zaltbommel) and deionized water. The molar oxide ratio in the gel was 6.6 SiO 2 : 1 TPA20 : 770 H20. Synthesis of the metal oxide composites took place at 150~ for 3 hours; the platinum composites were synthesized at 165 ~ for 5 hours and 15 minutes. Calcination of the metal oxide composites was carried at 500~ for 6 hours, while the platinum composites were calcined at 350~ for 12 hours or 500~ for 6 hours. After the low temperature calcination, the samples were treated in ozone at 120~ for one hour and subsequently reduced in H 2 at 100~ one hour. A cross-section of the metal oxide composites was analyzed with TEM (Philips CM-30 FEG). All model catalysts were tested in a batch reactor of 20 ml, suitable for analysis of amounts of products in the nano- and micro-range. Mixing of the products and reactants t o o k place by applying a small temperature difference over the reactor volume. The reactor set-up was fully automated. Analysis of the products was performed by online GC analysis on a Packard 438, equipped with a 25 m CP Sil5 column and a 25 m CP Wax 52 column in series. In all cases, the column temperature was kept constant at 40~ The Fe20 3 and Cr20 3 containing systems were tested in the non-oxidative

379

dehydrogenation of n-heptane at 500~ and atmospheric pressure. In each experiment five platelets of catalyst and 0.5 % n-heptane in N 2 were used. The platinum composites were tested in the (competitive) hydrogenation (1:1 volume ratio) of 1-heptene (99%, Aldrich) and 3,3-dimethyl-l-butene (95%, Aldrich). It is expected that for the zeolite coated catalysts the conversion of the linear molecule will be significantly higher than that of the branched molecule [12-14]. The reactions were carried out at 100~ atmospheric pressure in a mixture of H 2 and argon (1:9). The H2/hydrocarbon ratio was 17. In the experiments half a platelet was used of the TiO2/Pt catalyst which was sputtered for 12 seconds, as well as of the TiO2/Pt/silicalite-1 composite obtained from the same TiO2/Pt sample. After use the catalysts were analyzed by FTIR on a Bruker IFS-66 equipped with an A590 microscope.

3. RESULTS AND DISCUSSION

3.1. Metal oxide composite preparation and characterization From XPS analysis it is concluded that Fe203 and Cr203 was formed from the precursor materials on the surface of the silicon wafers. SEM measurements of the metal oxide layers indicate particle/cluster sizes of 50-500 nm for the Fe203 samples (see Fig.l) and 10-60 nm for the Cr203 samples. From these measurements it is concluded that the particles are hemi-spherical. Calculations based on the XPS results [10] and ICP/AES results showed that the surface coverage is approximately 8-10% for the Fe203 catalyst and 1-3.5% for the Cr203 catalysts. From XPS analysis before and after calcination of the chromium oxide catalysts, it was concluded that only part of the chromium precursor material was fixed to the support. Figure 2 shows a typical silicalite-1 layer which is grown on a metal oxide coated Si-wafer at 150~ 3 hours. The zeolite coating consists of more than a monolayer of crystals. The orientation of the first layer of crystals is lateral, and thus the b-direction is perpendicular to the support surface. On top of this layer, some axially oriented crystals are present. These crystals are not expected to influence the catalytic properties of the composites. From the thickness of the axial crystals, the layer thickness of the laterally oriented layer is estimated at 300 nm for each side of the wafer. Studies of the supported silicalite-1 systems proved the chemical bonding of the crystals to the support, which is assumed to take place by condensation between hydroxyl groups on the support surface and the zeolite crystals. Moreover, in this case it is possible that the crystals are bonded with the hydroxyl groups of the metal oxide particles. Calcination did not change the appearance of the layer or cause cracks. The total support coverage was estimated at a minimum of 95 %, thus leaving a limited amount of pin-holes. No evidence was found that at these holes uncovered support and catalytic material are exposed, and it is probable that a thin layer of zeolitic or amorphous SiO 2 is present. Since the wafers were suspended vertically in the synthesis mixture, it is expected that the lower horizontal side of the sandwich may contain particles which are not covered by the zeolite layer. The presence of the metal oxide particles after zeolite synthesis was confirmed by ICP/AES. Some of the metal oxide dissolved under synthesis conditions. XPS measurements of the external surface of the silicalite-1 coatings, however, did not show the presence of iron or chromium, which indicated that the dissolved metal oxide is not present in the MFI lattice [11]. For the Fe20 3 composites, TEM combined with elemental analysis showed the presence of iron at the interface between zeolite layer and

380

~i~i~!i!! ~i~ii~i '~i i~!!i~i 84 i i~!i!!i !i~!!ii~i~! ~i N i l~ ~ i~ ~ ~i iiiiiiii~!i~!~ii!i~!~!Wii~iiii~ ~i!!

;i!i iiiiii!!iii~iii i!! !ii~ '~'~'' i!i~ii!i!i~iiii!ii!i!!ii~ iii ~iii!!~~..........

~.,~ .......

Figure 1: Iron oxide particles on a Si support obtained by spin-coating and calcination.

Figure 2: Typical silicalite-1 layer on a Si support containing metal oxide particles. The average layer thickness is 300 nm.

Figure 3a: Platinum clusters on rutile (001) obtained after 12 s of sputtering. Average cluster size is 5 nm.

Figure 3b: Platinum clusters on rutile (001) obtained after 24 s of sputtering. Average cluster size is 20 nm.

381 support only. It is therefore concluded that the composites contain catalytic sites which are only accessible through the zeolite framework and are expected to show high reactant selectivity.

3.2. Platinum composite preparation and characterization Figures 3a and 3b show platinum species on the TiO 2 surface at different sputter times. The platinum cluster sizes increase from 2 to 20 nm, leaving parts of the support surface uncovered. At these parts the silicalite-1 layer is able to form chemical bonds with the surface OH-groups. From these pictures an estimation of the surface coverage by Pt of 85 % was made, from which the total amount of platinum sites was calculated. Zeolite coatings were grown on the TiO 2 supports which were sputtered during 6, 12 and 18 seconds. The silicalite-1 coatings have the same features as the coatings on the metal oxide particles, thus the bottom layer of crystals is almost continuous and has a lateral orientation. Since platinum does not have terminal functional groups, no chemical bonding or interaction with the zeolite layer is expected. XPS of the outer surface of the zeolite layer did not show any platinum.

3.3. Catalytic testing of the composites

3.3.1.Dehydrogenation of n-heptane over the metal oxide composites During dehydrogenation of n-heptane, heptenes, methylcyclohexane, toluene and various light components were formed. Figures 4 and 5 show the conversion of n-heptane as a function of time for the metal oxide particles and the metal oxides covered with silicalite-1. After a few cycles of heating to 500~ cooling to room temperature, part of the gold coating on the reactor wall had disappeared. This was accompanied by a sharp increase in the background conversion level, caused by the activity of the stainless steel reactor wall, responsible for a large amount of the cracking products and part of the dehydrogenation products. 4O

4O ~" 0

30

0

,,..4

20

~0

10

,~"

Fe203/siliealite-1

0

~O

30 20 10

f

Cr203/silicalite-1

0 0

10

20

30

40

50

time (min)

Figure 4: Conversion of n-heptane as a function of time for the Fe203 catalysts.

0

10

20

30

40

50

time (rain)

Figure 5: Conversion of n-heptane as a function of time for the Cr203 catalysts.

Table 1 gives the conversion levels, turnover numbers and product distribution of the catalysts at 30 minutes and at a conversion of 14.2%. The data have been corrected

382 for the background effects. It is shown that by coating of the catalyst sites, the conversion level decreased significantly. After 30 minutes and at equal conversions the composites gave a different product distribution than the metal oxide particles. For the composites the selectivity towards toluene had decreased, which was accompanied by an increased selectivity for the formation of methylcyclohexane. This selectivity is still not understood, and currently a matter of study.

Table 1. Dehydrogenation methylcylcohexane.

products of n-heptane

at 500~

after 30 min Catalyst

Fe20 3 Cr20 3 Fe203/MFI Cr203/MFI

TON (mole/mole/h-1) 51" 530** 20* 310"*

C7-" heptenes,

MCH:

at 14.2 % conversion

C7 (%)

MCH (%)

toluene (%)

C7 (%)

MCH (%)

toluene (%)

7.7 -

28.8 38.5 100.0 79.9

63.5 61.5 20.1

11.9 -

27.1 38.5 78.8 88.8

61.0 61.5 21.2 11.2

* calculations based on a particle size of 50 nm; ** calculations based on a particle size of 60 nm. After use the samples were analyzed by XPS, in order to find out if deactivation of the catalysts had taken place. It appeared that part of the chromium(III) species was reduced during use. This change towards Cr(II) or Cr(0) may enhance the activity of the catalyst [12]. No change in the oxidation state of the iron catalysts was observed.

3.3.2.Hydrogenation reactions by platinum composites The results of competitive hydrogenation of 1-heptene and 3,3-dimethyl-l-butene using the TiO2/platinum and TiO2/platinum/silicalite-1 catalysts are presented in Figures 7 and 8. Here the activity of the reactor wall was at a negligible level. Using the TiO2/Pt catalyst, both molecules were hydrogenated in competition at approximately the same rate. After coverage of platinum by silicalite-1, however, a large difference in conversion rates was observed. The same trends were observed for the hydrogenation of the single components. The results of the competitive hydrogenations are summarized in Table 2. The selectivities were calculated as the ratio of the initial reaction rates. From the conversion data of 1-heptene in a single component experiment, it appears that in the competition experiments the conversion of 1-heptene is not slowed down by the presence of 3,3-dimethyl-l-butene in the zeolite layer. Based on the kinetic diameter as well as the diffusivity of 3,3-dimethyl-l-butene, the adsorption of this molecule in MFI is low [13]. It is thus expected, in particular at this low partial pressure, that within the duration of these experiments no retardation of 1-heptene by 3,3-dimethyl1-butene is occurring. After use of the composite catalyst in single component experiments, it was observed by FTIR at 100~ that the zeolite layer was almost completely filled with 1-heptene and n-heptane, whereas only very small amounts of the branched molecules were found in the zeolite layer. It is therefore suggested that the conversion of the branched alkene takes place at pin-holes and not via the zeolite pores.

383 From Table 2 it is also apparent that the calcination procedure of the composites is an important variable. At high calcination temperatures, the activity increases and the selectivity decreases significantly, which is probably caused by cracks in the zeolite coating. Using mild calcination conditions, the selectivity remains high. 12

100

75 o

.,..~

~ 0

6

50 25

O o

0

1

2

3

3

0

time (h)

1

2

3

time (h)

Figure 8: Competitive hydrogenation of 1-heptene (o) and 3,3-dim ethyl-l-butene (A) over TiO2/Pt/MFI composite at 100~

Figure 7: Competitive hydrogenation of 1-heptene (o) and 3, 3-dimethyl-l-butene (A) over TiO2/Pt catalyst at 100~

Table 2. Competitive hydrogenation of 1-heptene and 3,3-dimethyl-l-butene at 100~ Catalyst

time ~ 1-heptene (min) (%)

~3,3-DMB-1 TON (%) (mol/mol/h -1)

TiO2/Pt TiO2/Pt/MFI(1)* TiO2/Pt/MFI(2)** TiO2/Pt/MFi(2) t TiO2/Pt/MFi(2) t

30 180 180 180 180

48.8 6.4 0.59 -1.10

47.2 49.4 10.0 10.7 --

* After 3 times of calcination at 500~ single component hydrogenation.

487 45 8.6 8.6 1.0

Selectivity

1.0 9-10 22-25 ---

6 hours;** after 1 time of calcination at 350~

12 hours; t

Separation of n-hexane and 2,2-dimethylbutane over a silicalite-1 membrane at 50~ resulted in a separation factor of 17.2 [14]. It was suggested that non-zeolitic micropores probably played a role, so that the separation factor in principle could be higher than the observed one. Competitive hydrogenation of linear and branched olefins over Pt/ZSM-5 catalysts was carried out for mixtures of 1-hexene/2,4,4-trimethylpentene-1 and 1-hexene/ 4,4-dimethylhexene-1 [15,16].In both cases a high selectivity for the conversion of the linear molecules was observed. In the first case a conversion ratio of approximately 12 was reported for reaction at 100~ the second case the selectivity was equal to 25 during conversion at 100~ composite catalysts thus show comparable selectivity. A difference in selectivity between a regular catalyst and the composite catalysts is expected on the basis of the particular catalyst configuration. Assuming comparable intrinsic reaction rates and sorption equilibria, the theoretical selectivity of a regular

384 catalyst containing homogeneously dispersed catalytic sites is equal to the square root of the ratio between the effective diffusivities [17]. In case of a shape selective diffusion barrier covering the catalytic site, as in the case of the composites, the theoretical selectivity is equal to the ratio of the effective diffusivities. Here the intra-zeolitic diffusivities at 100~ estimated at 10 -12 m2/s and 10 -18 m2/s for 1-heptene and 3,3dimethyl-l-butene, respectively. Thus if mutual interaction (retardation of 1-heptene by the branched alkene) would be absent, a theoretical selectivity of more than 104 might be possible. The relatively low selectivity can be due to pin-holes or sides of the sandwich, containing platinum directly exposed to the reactants or platinum covered with amorphous silica which only causes Knudsen diffusion effects. Additionally the presence of very small amounts of platinum on the external surface of the zeolite layer, which can not be detected by XPS may also be responsible. Adjustment of the zeolite synthesis procedure, resulting in less pin-holes, shorter synthesis times and a thinner zeolite layer are expected to further improve selectivity in catalysis.

4. C O N C L U S I O N S A novel type of catalyst containing supported oriented coatings of silicalite-1 has been achieved. With this concept, many types of catalytic material can be combined with the shape selectivity of a zeolite framework. This is demonstrated by the preparation of two types of MFI-covered supported catalysts. The first true monofunctional platinum/MFI catalyst is obtained from direct synthesis. The shape selectivity of the rutile-platinum-MFI composite was tested by competitive hydrogenation of a linear and a branched olefin. The composites give a high selectivity towards conversion of the linear olefin, which is comparable to or even higher than optimized regular zeolite catalysts.

Acknowledgements The authors thank Mr. Th. de Mooij from Jeol Europe and Dr. H. Otsuji from Jeol USA for the HRSEM measurements, and the Royal Dutch/Shell Laboratory, in particular Dr. Z. Chen and Dr. E.W. Kuipers for their advice in the development of the reactor and the spin-coating procedure. REFERENCES [1] J.C. Jansen, D. Kashchiev and A. Erdem-Senatalar, Stud. Surf. Sci. Catal., 85, (1994), 215-250. [2] E.R. Geus, H. van Bekkum, W.J.W. Bakker and J.A. Moulijn, Microporous Mat., 1, (1993), 131-147. [3] J.C. Jansen, W. Nugroho and H. van Bekkum, Proc. 9th Int. Zeolite Conf,, I, (1992), 247-254. [4] J.H. Koegler, H.W. Zandbergen, J.L.N.Harteveld, M.S. Nieuwenhuizen, J.C. Jansen and H. van Bekkum, Stud. Surf. Sci. Catal., 84, (1994), 307-314. [5] H.P. Calis, A.W. Gerritsen, C.M. van den Bleek, C. Legein, J.C. Jansen and H. van Bekkum, Can. J. Chem. Eng., 73(1), (1995), 120-128. [6] K. Tanabe, M. Misono, Y. Ono and H. Hattori, Stud. Surf. Sci. Catal., 51, (1989), 108-113. [7] M. Niwa, N. Katada and Y. Murakami, J. Catal., 134, (1992), 340-348. [8] E.W. Kuipers, C. Laszlo and W. Wieldraaijer, Cat. Lett., 17, (1993), 71-79. [9] S.P. Murarka, Silicides for VLSI Applications, Academic Press, (1983). [10] H.P.C.E. Kuipers, H.C.E. van Leuven and W.M. Visser, Surf. Interf. Analysis, 8, (1986), 235-242. [11] N. van der Puil, J.C. Jansen and H. van Bekkum, Stud. Surf. Sci. Catal., 84, (1994), 211-218. [12] T. Komatsu, J. Mol. Catal., 78, (1993), 57-66. [13] M.F.M.Post, J. van Amstel and H. van Kouwenhoven, Proc. 6th Int. Zeolite Conf., (1984), 517-527. [14] J.G. Tsikoyiannis and W.O. Haag, Zeolites, 12, (1992), 126-130. [15] J. Weitkamp, T. Kromminga and S. Ernst, Chem.-Ing.-Tech. 64 (12), (1992), 1112-1114. [16] R.M. Dessau, J. Catal., 89, (1984), 520-526. [17] W.O. Haag, R.M. Lago and P.B. Weisz, Disc. Faraday Soc., 72, (1981), 317-330.


385

O L I G O M E R I Z A T I O N OF B U T E N E S W I T H P A R T I A L L Y A L K A L I N E EARTH EXCHANGED NiNaY ZEOLITES B. Nkosi, F. T. T. Ng, and G. L. Rempel Department of Chemical Engineering, University of Waterloo, 200 University Avenue West, Waterloo, Ontario, Canada N2L 3G1 Summary A number of catalysts were prepared by partially exchanging NaY with alkalineearth cations followed by nickel promotion. These catalysts were used to oligomerize butenes in the liquid phase. For the catalysts with similar activities, the dimer selectivity was found to increase with increase in the cationic size of the alkaline-earth metals, but decreased with increase in the Sanderson's electronegativity function. This is attributed to the increased adsorption of olefins on the more acidic catalysts. Catalysts that showed good dimer selectivity were found to have better stability than those that had poor dimer selectivity, indicating that the deactivation was caused by fouling of pores by long chain oligomers. This improvement in catalyst stability imparted by the alkali-earth cations is probably due to the blocking of the sodalite cages of the NiNaY catalysts. The catalyst derived from partially exchanging sodium with barium cations was found to have high activity, good C8 selectivity and low deactivation rate. 1. Introduction

Fluid catalytic cracking of naphtha leads to the production of low molecular weight hydrocarbons. Low molecular weight alkenes in the C2-Ca range are a major product of this fluid catalytic cracking process. Oligomerization is used to convert these alkenes back to higher boiling liquid products. Butenes are oligomerized to octenes and higher alkenes for use in a variety of products. The octenes are useful for the synthesis of alcohols as well as for the blending of gasoline. For gasoline blending applications it is important that only dimers are formed from the oligomerization of butenes as higher molecular weight products have poor blending properties. These higher molecular weight products have also been reported to cause serious deactivation of NiNaY oligomerization catalysts(I). Other workers have also observed that the nickel exchanged zeolites undergo serious deactivation(2,3). In this study, an attempt has been made to improve the stability of these nickel exchanged catalysts by slowing down the deactivation of these oligomerization catalysts. The strategy used to address the deactivation phenomena was to partially exchange the sodium ions with alkali or alkali-earth metal ions. In this paper, only the results from the work on the partial exchange with alkaline-earth ions will be reported; the work on partial exchanging sodium ions with alkali ions will be reported elsewhere(4).

386

2. Experimental Unless otherwise stated, all catalysts were made by ion exchange procedures. All chemicals were purchased from Aldrich Chemicals and used without any further purifications. Linde Y (LZ-Y52, from Union Carbide) zeolite was ion exchanged, once, with 0.5M aqueous solutions of the appropriate alkali-earth chloride salts (e.g. MgCI 2, CaCI 2, SrCI2, BaCI2) at 80~ for 24 hours and then washed thoroughly with deionized water to get rid of residual ions. This was followed by a second exchange with 0.2M aqueous solutions of nickel chloride, at room temperature, for 16 hours. In the case of NiHY catalyst, LZ-Y82(from Union Carbide) was ion exchanged with a 0.2M aqueous solution of nickel chloride at 80~ for 24 hours. The loaded catalysts were subsequently washed with deionized water to get rid of occluded nickel salt. They were then dried, first in air, under ambient conditions followed by drying in the oven at 110~ for over 24 hours. Calcination was performed at 500~ under dry air for 16 hours. Fresh and spent catalysts were characterized for surface area by N2 adsorption using the BET method. An Autosorb-1 instrument from Quantachrome Corporation was used for the BET measurements. Nickel loadings were determined by X-ray fluorescence (XRF) using an Oxford Lab-X 1000 instrument. Sodium analysis was determined by atomic absorption analysis (AAS) using a Perkin Elmer 3100 instrument. Carbon content of spent catalysts was analyzed by the thermal gravimetric analysis/differential thermal analysis method (TGA/DTA) using an SDT 2960 from TA Instruments, Inc. Approximately 50 mg of spent catalyst samples were heated in a 5% 02 in helium flow up to 650~ at 20~ The compositions of the catalysts are shown in Table 1. Table 1: Composition of Alkaline-earth Exchanged NiNaY Catalysts Catalyst

Nickel Loading (wt%)

Sodium Content (wt%)

NiHY" NiNaY b NiNaY NiNaMgY b NiNaMgY NiNaCaY b NiNaCaY NiNaSrY NiNaBaY

3.97 0.76 4.85 0.90 4.67 0.86 2.72 3.09 0.87

< 0.14 7.19 2.23 5.62 1.37 3.27 1.73 1.72 1.25

catalyst prepared by ion exchange of nickel salt at 80~ for 36 hours. b prepared from a 0.02M of nickel chloride solution.

a

1-Butene oligomerization was performed in the liquid phase using a 300ml Parr autoclave. The autoclave was equipped with a cooling coil to control temperature

387 excursions such that temperature control was within 1~ of the desired set point. All the oligomerization reactions were performed in a batch reactor under the following standard conditions: Reactor temperature = 110~ pressure = 600psi; run time = 2 hours; isopentane : 1-butene wt ratio = 1:2.67. In a typical reaction, a liquid charge containing approximately 70 grams of 1-butene and 25 grams of isopentane is contacted with 3 grams of catalyst for 2 hours at 110~ and 600psi pressure. At the end of 2 hours the reaction products are cooled to 6~ The amount of products and unreacted butenes are weighed and subsequently analyzed by gas chromatography (GC). A 60 m DB-1 column was used to separate 1-butene and the reaction products. The oligomerization activity is determined from the GC results and the weight of recovered autoclave contents. The C8 selectivity is determined from the weight of all C8 isomers to the weight of all oligomers produced ( C4 isomers are not taken into account in ranking catalysts for activity and selectivity). 3. Results and Discussion

Product distributions for the oligomerization of butenes from a few selected catalysts are shown in Table 2. The results show that the C8 dimers produced are mainly branched isomers with negligible amounts of straight chain octenes present. These branched dimers would be suitable for gasoline blending purposes. Isomerization of 1butene to 2-butenes and trace amounts of isobutylene were observed. Close examination of the results reveal that the ratios of trans-2-butene:cis-2-butene(Tr-C4~: Cis-C4- ) decreases with increase in the size of the alkaline-earth cation( Trans:cis C4- ratio for Mg=l.95, Ca=l.90, Sr=1.82, and Ba=l.79). Trans -2-butene formation has been reported to be favoured over acidic catalysts compared with cis-2-butene (5). This trend indicates that the acidity of the catalysts is decreasing with cationic radius from Mg to Ba. Table 2: Typical Product Distribution obtained from Butene Oligomerization Reaction Using Selected NiY Catalysts under Standard Reaction Conditions. Catalyst

NiNaMgY NiNaCaY NiNaSrY NiNaBaY

C4

isomers (wt%)

C8 isomers (wt%)

>_C~

l-C4

Tr-C4

Cis-C~

DM-C6

M-C7

n-C~

_>C~

2.26 3.86 3.37 4.13

30.49 34.36 33.53 30.83

15.62 18.10 18.41 17.28

19.64 18.09 19.26 19.98

14.41 12.00 14.96 17.96

0.44 0.80 0.75 0.68

16.81 12.30 8.86 8.99

A summary of all the oligomerization data for a number of catalysts used in this study are shown in Table 3. The results show that generally for catalysts that have a nickel loading of less than 1% w/w, with the exception of the NiNaBaY catalyst, have lower activity than the catalysts with high nickel loading. This is attributed to the fact that when NaY is ion exchanged with Ni 2+ solutions, Ni 2+ ions preferentially exchange

388 with Na + ions in the sodalite and hexagonal prisms until about 36% of exchange capacity is reached (6). After all the exchange sites in these positions are exhausted, the Ni 2+ions begin to exchange for Na ~ cations in the supergages. It has been proposed that for Ni 2+ to be active for the oligomerization reaction, the cations have to be located in the supercages(6). It is probable that the low nickel loading for the zeolite exchanged with Ba 2~ could be due to the large size of the Ba 2~ cations which effectively block the hexagonal prism in the faujasite structure, confining the Ni 2+ cations to the supercages. Table 3: Oligomerization of Butenes Using Various NiY Exchanged with Alkali-earth Metal Chlorides a. Catalyst c

3.97NiHY 0.76NiNaY 4.85NiNaY 0.90NiNaMgY 4.67NiNaMgY 0.86NiNaC aY 2.72NiNaCaY 3.09NiNaSrY 0.87NiNaBaY

Activity (g /g. cat. /h)

C8 Selectivity b (%)

Coke Content (%)

Deactivation (%loss BET Area)

5.16 2.13 5.91 1.52 4.99 2.49 4.77 4.81 4.44

57.81 89.36 69.51 79.99 67.21 70.72 70.41 73.40 81.58

15.13 d 11.42 10.52 13.83 12.06 12.61 10.96 9.71

92.42 34.10 70.72 55.26 64.14 60.79 63.69 55.39 37.54

" Standard reaction conditions used. b as a % of all oligomers produced c The numbers designate the nickel % wt loading. d not measured Table 3 shows that for catalysts prepared from similar cationic types, but different nickel loadings, the dimer selectivity decreases with increase in activity. The catalysts prepared from NiNaCaY are exceptions to this generalization, it is not clear at this stage why the low nickel calcium catalyst has such a low selectivity. This decrease in selectivity with increase in activity is consistent with a consecutive reaction pathway, where dimers are the first intermediate products. Interestingly, the NiNaBaY catalyst has a good C8 selectivity even though its activity is reasonably high. Close examination of the data given in Table 3 for the catalysts prepared from the partially exchanged alkaline-earth metals at about similar activity (>4.44 g. olig./g.cat./h) reveal some interesting trends. The influence of cationic size on the dimer selectivity for catalysts with similar activities is given in Figure 1. The dimer selectivity of the catalysts increases with the cationic radius of the alkaline-earth metals, this is probably due to partial blocking of the sodalite cages and/or the hexagonal prisms by some of

389 of these bulky alkaline earth cations, preventing the formation of long chain oligomers. In fact, some of these cations have also been reported to have preference for occupying sodalite cages in these faujasite zeolites(7). The acidity of zeolites have been shown to be related to physicochemical properties, such as electrostatic field, electrostatic potential as well as electronegativity of the cations in the exchange sites (8,9). A plot of the dimer selectivity as a function of the Sanderson's electronegativity function of the alkali-earth cations in the catalyst is shown in Figure 2. These results show that the dimer selectivity is inversely related to the electronegativity of the alkali-earth metals, indicating that the more acidic the catalyst the less the dimer selectivity. To further probe the effect of acidity on dimer selectivity, a highly acidic NiHY catalyst was prepared. This catalyst was tested for the butene oligomerization reaction under standard conditions. Table 3 shows that this catalyst has the lowest selectivity of all the catalysts tested, thus confirming that high acidity leads to low C8 selectivity. This is ascribed to the enhanced olefin adsorption with increase in acidity and hence favour the formation of higher molecular products.

90

~ 85

B~

Ba

4--' >" .

_> ao +.~ ~

.E

65

C3

C3 60

0.60

BO

sr

L) (D 75 (I) CO 70 (1)

75

70 MQ ---~

_

o.ao

t 1.00

l 1.20

65 60

1.40

Cationic Radius(A)

Fig. I The Influence of Alkaline-earth Cationic size on Dimer Selectivity

0.60

I OZO

I i.O0

I 1~_0

i 1.40

1.60

Sanderson's e l e c t r o n e g a t i v i t y

Fig. 2 The Influence of the Sanderson's Electronegativity on Dimer Selectivity

Table 3 shows the deactivation of catalyts when used for the butene oligomerization reaction. Deactivation is defined as: Surface Area of Fresh Catalyst - Surface Area of Spent Catalyst x 100 Surface Area of Fresh Catalyst The results for surface area analysis of various catalysts before and after the oligomerization reaction are given in Table 4. These results show that all the catalysts are undergoing deactivation during the oligomerization reaction. It is noteworthy that the deactivation is very high for the highly acidic NiHY catalyst. The results in Table 4 show that the area inside micropores of spent catalysts is reduced quite dramatically, whilst that inside the mesopore does not change significantly indicating that the catalysts undergo deactivation by blocking of the micropores. Fig. 3 is a plot of the carbon content of spent catalysts as a function of

390 the Sanderson's electronegativity for catalysts with similar activity(_>4.44 g olig./g cat./h). Since the electronegativity function is related to the acidity of the zeolites (8,9), it appears, therefore, that the deactivation of these catalysts increases with increase in acidity. It has been generally accepted that high acidity leads to enhanced catalyst deactivation.

[Z o .Q

13

63 O

12

c-

11

jJ

J

~~ o~

g B

0.60

0.80

1.00

Sanderson's

1.20

1.40

1.60

electronegativity

Fig. 3 The Deactivation as a function of the Sanderson's Electronegativity.

One of the questions that need to be addressed is the cause and the nature of the carbon found in the spent catalysts. To address this question, a spent catalyst from NiNaCaY, which was previously used for 2 hours in a butene oligomerization reaction under standard reaction conditions, was subjected to a soxhlet extraction using hexane as solvent. This soxhlet extraction was performed over a 36 hour time duration. The Table 4 BET Surface Areas of Fresh and Spent Catalysts a BET Surface Area (m2/g)

Catalyst

Fresh Catalyst

3.97NiHY 4.85NiNaY 4.67NiNaMgY 0.86NiNaCaY 2.72NiNaC aY 3.09NiNaSrY 0.9NiNaBaY

Spent Catalyst

Total Area

Mesopore

Total Area

Mesopore

520.9 653.0 692.9 693.2 747.8 641.4 566.2

123.8 166.3 176.8 138.8 163.5 145.4 126.0

39.5 251.5 249.8 271.8 271.5 287.3 361.9

39.5 116.0 125.9 128.3 107.1 126.1 117.2

all samples were evacuated at 110~ for at least 5 hours; fresh catalysts were calcined at 500~ for 16 hours; spent catalysts were used in a batch reaction for 2 hrs under standard conditions; area is from a 5 point BET at P/Po E 0

0

21050

/

0:2

0:4

0:6

0,8

0 0,0

0:2

0:4

Fig. 5 Dependence of toluene conversion in toluene disproportionation on the number of OH groups for (Fe)ZSM-5 zeolites activated at 770 K ( 9 ) and for (Fe)ZSM-5D activated at higher temperatures ( o ).

._~

A

~

40

40

~0 0.

20

20

_= O F-

0

11)0

2()0

12

Fig.7 Dependence of toluene conversion in "steady state" in toluene alkylation with isopropanol on the number of OH groups; (AI)ZSM-5( e); (Fe)ZSM-5 ( 9).

60-

== 60t

0

1:0

80 A-"....jr.--d,~"

8o

.g E (!) > E O o q) E ~)

0:8

100

100

v i-

0:6

OH (mmol/g)

OH (mmol/g)

3()0

400

Time-on-stream (min)

Fig. 6 A,B Time-on-stream dependence of toluene conversion and propyltoluene selectivity; A - (AI)ZSM-5B B - (Fe)ZSM-5C

o

0

1~0

2~o

300

400

Time-on-stream (min)

toluene conversion isopropyltoluene n-propyltoluene sum of propyltoluenes

o 9 9 9

407 dehydrated at 690 K, the activity per one Broensted center for all the ferrisilicates investigated was higher than that of alumosilicates (Fig. 7). While with the AI analogs the toluene conversion exhibited nearly a constant value in the range of the concentration of OH groups investigated, for the Fe analogs the conversion slightly increased with the increasing number of the Si-OH-Fe sites. This clearly indicates that in addition to the intrinsic alkylation activity of the individual Si-OH-Me groups some other effect has to contribute substantially to the overall process. All these results imply that the process of toluene alkylation with isopropanol is not exclusively controlled by the intrinsic rate of the alkylation reaction on the Broensted sites, but on the contrary by the desorption/transport rate of the propyltoluene products from the inner molecular sieve channels. This is probably the reason why ferrisilicates exhibit a higher activity per Si-OH-Fe site compared to Si-OH-A1 as the desorption of propyltoluenes is easier from the less acidic former groups.

Table 2 Toluene disproportionation over H-(Fe)ZSM-5

Activation (K) Time-on-stream (min.) Conversion (%)

770 15 55 3.3 3.2

920 15 1.1

55 1.0

1050 15 55 0.8 0.6

Selectivity (%) p-Xylene m-Xylene o-Xylene

26.2 50.4 23.4

26.4 50.7 22.9

32.7 49.0 18.4

34.2 48.8 17.1

39.4 45.5 15.2

40.0 44.0 16.0

46.0 0.0 10.1 19.4 9.0 15.5

46.5 0.0 10.5 20.1 9.1 13.9

53.9 0.0 12.2 18.3 6.9 8.7

52.7 0.0 12.6 18.1 6.3 10.3

56.0 0.0 14.1 16.3 5.4 8.2

53.9 0.0 14.2 15.6 5.7 10.7

Selectivity (wt. %) Benzene Ethylb enzene p-Xylene m-Xylene o-Xylene C1-C4 aliphatics

408 4.0 CONCLUSIONS It can be summarised that (i) The reaction of toluene disproportionation is controlled by the number and acid strength of Si-OH-A1 or Si-OH-Fe groups, thus, by the intrinsic activity of the bridging OH groups. Extraframework A1 or Fe species, exhibiting Lewis acidity and simultaneously present with the Broensted sites in molecular sieves, contribute considerably to toluene dealkylation but not si~ificantly to toluene disproportionation. Thus the presence of these electron acceptor sites changes the selectivity of the toluene transformation. (ii) The reaction of toluene alkylation with isopropanol is controlled by the rate of desorption/transport of propyltoluenes from the active Si-OH-Me sites located in the inner channels of the molecular sieves. This is tmder~andable considering a relatively low reaction temperature, high boiling point of propyltoluenes and in addition the fact that propyltoluenes are bulky molecules. The desorption rate of propyltoluenes can be expected to be lower from Si-OH-A1 groups compared to Si-OH-Fe, however, this assumption should be independently experimentally evidenced. (iii) The effect of acid site strength on the desorption/transport rate of the products for reactions carrying out at relatively low temperature is shown here to be so dramatic that it can reverse the sequence of the activity of the sites of different acidity. If the reaction controlling step is this desorption/transport rate of the products, then the less acidic catalyst can be more efficient in the total conversion compared to that one exhibiting high acid site strength. ACKNOWLEDGEMENT

Financial support of the Grant Agency of the Academy of Sciences of the Czech Republic, project No. 440408, is highly acknowledged. REFERENCES

1. M. Tielen, M. Geelen, P.A. Jacobs, Acta Physica et Chemica Szegediensis, Proc. Int. Syrup. on Zeolite Catalysts, p. 1 (1985). 2. C.T.W. Chu, C.D. Chang, J. Phys. Chem. 89 (1985) 1569. 3. 1L Zahradnik, P. Hobza B. Wichteflovfi, J. Cejka, Coll. Czech. Chem. Commun. 58 (1993) 2474. 4. J. Cejka, A. Vondrovfi, B. Wichtedovfi, G. Vorbeck, R. Fricke, Zeolites 14 (1994) 147. 5. Y. Hong, V. Gruver, J.J. Fripiat, J. Catal. 150 (1994) 421. 6. J.1L Sohn, S.J. DeCanio, P.O. Fritz, J.H. Lunsford, J. Catal. 125 (1990) 123. 1L Carvajal, Po-Jen Chu, J.H. Lunsford, J. Catal. 125 (1990) 123. 7. I~M. Lago, W.O. Haag, R.J. Mikovsky, D.H. Olson, S.D. Hellring, K.D. Schmitt, G.T. Kerr, Proc. 7th Int. Zeol. Conf., (Eds. Y. Murakami et al.), Kodansha - Elsevier, p.677 (1986). 8. V.L. Zholobenko, L.M. Kustov, B.V. Kazansky, E. Loeffler, U. Lohse, G. Oehman, Zeolites 11 (1991) 132. 9. B. Wichterlov~, J. Nov~kovfi, L. Kubelkovfi, P. Jirfi, Proc. 5th Int. Zeol. Conf., Neapol, (Ed. L.V.C. Rees) Heyden, London, p. 373 (1980). 10. A. Corma, Stud. Surf. Sci. Catal. 49 (1989)49. 11. D. H. Lin, G. Coudurier, J.C. Vedrine, Stud. Surf. Sci. Catal. 49 (1989) 1431.


409

NOx Adsorption Complexes on Zeolites Containing Metal Cations and Strong Lewis Acid Sites and Their Reactivity in CO and CH4 Oxidation: A S p e c t r o s c o p i c Study L. M. Kustov, E. V. Smekalina, E. B. Uvarova, and V. B. Kazansky N. D. Zelinsky Institute of Organic Chemistry, Russian Academy of Sciences, Moscow, Leninsky prosp. 47, 117334 Russia ABSTRACT Complex formation and transformations of nitrogen oxides (NO, N20 ) on cations and strong Lewis acid sites in zeolites of the ZSM-5, MOR, LTA, and FAU types were studied by diffuse-reflectance IR spectroscopy. The NOD+ and N20 3 species yielded on cationic forms via NO disproportionation were shown to exhibit strong oxidizing properties with respect to CO and CH4 molecules. Unlike cationic forms, for the dehydroxylated zeolites or the zeolites containing strong Lewis acid sites, the latter were found to be responsible for the polarization of N20 (either produced from NO via disproportionation or adsorbed from the gas phase), which results in chemisorption of atomic oxygen that displays the oxidizing properties in the reactions of CO and CH4 oxidation to CO 2.

INTRODUCTION The exhaust gas purification from NOx is one of the important problems related to the ecology. The use of various heterogeneous catalysts for NOx abatement is reviewed, for instance, in [1]. Among the catalytic systems, the zeolite catalysts may be considered as an alternative to the available honeycomb ceramic oxide catalysts that are based on the process of selective catalytic reduction (SCR) of NOx using NH3. Carbon monoxide and hydrocarbons, such as methane and propane were proposed as efficient reducing agents for SCR of NO using zeolite catalysts (see, for example, [2 - 4]). Nitric and nitrous oxides are also often used as molecular probes for studying surface sites in oxide catalysts and supported metals by IR and ESR techniques [4, 5]. Cationic forms of zeolites are example of systems that are characterized by the formation of a variety of complexes with adsorption sites of different nature and by the occurrence of chemical transformations upon admission of NO [6, 7]. Such transformations can proceed not only on the

410

zeolites containing transition metals [8] and reduced noble metals [9] but also on the zeolites containing no d-elements [10]. It is known that NO disproportionation occurs on Na-chabasite, Na-faujasites, and CaY zeolites according to the overall reaction [11, 12]: 4 N O = N20 + N20 3 At the same time, there are indications that other products may be also formed, such as NO 2 and NO3". Metal cations were assumed to be the active sites for this reaction. However, the mechanism of this process and the influence of the zeolite structure and composition on this process were not yet virtually studied. There are some data on the SCR of NO using CO and methane or other hydrocarbons on zeolites (see, for instance, [2]), but the IR-spectroscopic data related to the investigation of these systems are scarce. In our previous papers [13, 14], we studied the interaction of nitrous oxide with strong Lewis acid sites in dehydroxylated zeolites and found that the latter are able to catalyze the reaction of N20 decomposition via the formation of a chemisorbed atomic oxygen species that is active in the oxidation of H 2, CO, CH 4, and some other molecules. The aim of this work was to study the complex formation and transformations of NO and N20 (disproportionation, reduction with CO and CH4) on cationic (alkaline and alkaline-earth) forms of zeolites that differ in their structure and chemical composition. The similar data were also obtained for the zeolites containing strong Lewis acid sites (dealuminated mordenite, dehydroxylated HZSM-5, and HZSM-5 modified with zinc oxide).

EXPERIMENTAL Na-, Ca-, and Mg-forms of A (Si/AI = 1.0), Y (Si/AI = 2.35), and MOR (Si/AI = 5.0) zeolites were studied. The ion-exchange degrees of Na for Ca or Mg were 50 - 90%. Ion exchange was carried out from aqueous solutions of nitrates. The samples were washed with distilled water, dried at 450 K, and pretreated under vacuum at 720 K for 4 h in IR cells. The rate of the temperature increase was 4 K/min. Also, HZSM-5 zeolite (Si/AI = 20), dehydroxylated at 1070 K for 5 h, H-MOR (Si/AI = 26), dealuminated with hydrochloric acid at 350 K, and HZSM-5 zeolite, modified by ZnO (2 wt %) or Ga (in the framework tetrahedral positions, Si/Ga = 30) were investigated. Diffuse-reflectance IR spectra were measured in the frequency range of 1600 - 4000 cm-1 using a Perkin-Elmer 580B spectrophotometer according to the procedure [15]. Nitrous and nitric oxides, methane, and carbon monoxide were adsorbed at 300 K and P = 1 - 100 torr.

411 RESULTS AND DISCUSSION 1. Adsorption of nitrous oxide Figure 1 shows IR spectra of nitrous oxide adsorbed on different zeolites. Adsorption results in the appearance of a set of bands attributed to the fundamental stretching vibration v as (2230 - 2245 cm-1), as well as to the com-

2240

l 2360

J

~

2590

2260 ~, 2220

2245 I

! 2820

A

I ~2455 590 2260 i 0

2230I .

.

.

.

[

~

"~2810

4.o,/r 2503

.

2800

Fig. 1. IR spectra of N 2 0 adsorbed onzeolites: 1 - NaMOR, 2 - NaA, 3 - NaY, 4 - CaY, 5 - dealuminated mordenite, 6 - the same, after admission of CO and heating at 470 K.

412

bination bands and overtones of adsorbed N20 molecules. The bands at 2420 2510 cm -1 are assigned to the combination of the symmetric stretching and overtone of the bending vibrations (vs0_l + 50-2), those at 2590 - 2600 - to the first overtone of the symmetric stretching vibration (vS0_2) and those at 2800 2820 cm -1 - to the combination of the antisymmetric stretching and bending vibrations (vas0_l + 50_]). The data show that N20 is adsorbed in the molecular form and the v as and (vs0_l + 50_1) frequencies barely change for the cationic forms, independently of the cation nature and type of a zeolite. More considerable changes are observed for the vS0_2 overtone frequency and the combination vs0_l + 50_ 2. Evidently, the metal cations are the centers responsible for the N20 adsorption. Then the distinctions in the IR spectra, especially in the region of 2400 - 2600 cm-1 ($0_2 + vS0 -1) may be explained by the presence of different localization sites for cations in the zeolites under study. For the NaA zeolite, localization in the centers of the 6-membered rings (8 of 12 sites) is most probable [16]. Accordingly, a single band is observed at 2455 cm -1 in the region of vs0_l + 50_2 and a single overtone band ( vS0_2 ) is revealed at 2590 cm -1. For the NaY and Na-MOR zeolites, the number of possible localization sites increases to 2 - 3 [16] (for instance, S I, S I, and SII for NaY, localization sites in the main and side channels of mordenite). In agreement with this site distribution, a more complex superposition of several bands is observed in the region of vs0_l + 80_2 and vS0_2. When passing from Naforms of Y and MOR zeolites to the alkaline-earth forms, the IR spectra in the region of 2400 - 2600 cm -1 become more simple, which may be accounted for by the more uniform distribution of the alkaline-earth cations in these zeolites, in agreement with the available data [16]. Thus, the frequencies of the symmetric stretching vibrations are most sensitive to the interaction with cations in zeolites, which may be explained by a particular, presumably, two-point geometry of the adsorption complex. In the case of the zeolites containing strong Lewis acid sites (ZnO/HZSM-5, dehydroxylated HZSM-5, HGaZSM-5, and dealuminated mordenite), adsorption of N20 gives rise to the shift of the v as frequency to higher frequencies (2280 2260 cm-1), compared to the gas phase and cationic forms. Further heating results in N20 decomposition with evolution of N 2 into the gas phase and chemisorption of atomic oxygen, as it was shown-previously [13, 14]. Admission of CO or CH 4 to the samples with preadsorbed N20 or chemisorbed oxygen, generated as described above, and heating to 400 - 450 K leads to the formation of CO 2 (v = 2360 cm -1). In the case of vel~' strong Lewis acid sites (ZnO/HZSM-5), the heating to 400 - 450 K is not necessary and oxydation occurs at room temperature. Adsorption of carbon monoxide or methane on the cationic forms of zeolites with preadsorbed N20 and further heating to 450 K does not result in the oxidation of these molecules, unlike the case of the zeolites containing strong Lewis acid sites.

413

2. NO a d s o r p t i o n The IR spectra of NO adsorbed at 300 K on CaY and Mg-MOR zeolites are presented in Fig. 2. Unlike NO adsorption on reduced metals or transition-metal ions, no IR bands attributed to NO adsorbed in the molecular form (v = 1900 1860 cm -1) were found for CaY and MgMOR. Instead, a set of bands with maxima ranging from 2260 to 1940 cm -1 was observed due to charging or disproportionation of adsorbed NO molecules [4, 5]. For MgMOR and other cationic forms of high-silica zeolites, the bands at 2160 - 2050 cm -1 predominate in the IR spectra. For CaY and other faujasites under study, the

"•221•0

223O

2280

230,

2360

1930

i1 -.

2060 120 k-t~^

\\\"

2260

k\

2120

j

2360 1 Ill I

255 2360 2585

Fig. 2. IR spectra of NO adsorbed on zeolites: 1 - NaY, 2 - MgMOR, 3 ZnO/HZSM-5. Solid lines: NO adsorption; dotted lines: after further admission of CH 4 and heating to 470 K.

414

bands at 1980 - 1940 cm -1 and those at 2260 - 2230 cm -1 are the most intense spectral features. The bands at 2260 - 2230 (and the corresponding vS0_2 band at 2505 cm -1) and 1980 - 1940 cm -1 can be assigned to N20 and N20 3 (or NO.NO2) species formed via disproportionation of NO, whereas the bands at 2160 - 2050 cm -1 could be attributed to positively charged NO + 5 species [3, 4]. According to [5, 6], other species, such as NO 3- and NO 2 are also formed as a result of NO disproportionation, but we failed to observe the corresponding bands in the DRIR spectra because of the unfavorable background below 1600 cm-1. Thus, for low-silica cationic forms, disproportionation of NO mainly takes place, whereas for the high-silica zeolites, positive charging of NO is more likely. The reason for this difference seems to be different cation density and pore structure of faujasites and pentasils (mordenite and ZSM-5). In the cavities of faujasites, the cation density is much higher than in the channels of the pentasil and short distances between neighboring Me(NO)x ensembles probably allow the easier NO transformation into N20 and N20 3 or NO 2. Further adsorption of ccarbon monoxide or methane at 300 K on CaY with preadsorbed NO results in a slow oxidation into CO 2 (v = 2360 cm -1) in both cases (Fig. 3). Simultaneously, the bands at 1980 - 1940 cm-1 disappear, whereas those at 2260 - 2060 cm -1 remain unchanged. Warming the samples with preadsorbed NO and methane up to 400 - 450 K accelerates CO 2 formation. Unlike the faujasite sample, MgMOR does not exhibit any oxidizing activity in the NO + CO or NO + CH 4 reactions. Slow oxidation starts only at T > 570 K. Hence, we may conclude that the oxidized products of NO disproportionation (in particular, N20 3 or NO2) are the active species responsible for the CO or CH 4 oxidation. In the case of the zeolites containing strong Lewis acid sites, NO adsorption does not result in the appearance of the N20 3 species (1950 - 1930 cm-1) and the N20 bands (2280 - 2230 cm -1) predominate in the IR spectra. Also, the bands assigned to NO c5+ species at 2120 -2150 cm -1 are observed. Probably, in this case, deeper oxidation of NO takes place yielding the species that cannot be observed in DRIR spectra (NO 3- and NO2). Further adsorption of CO, especially followed by heating at 370 - 400 K, causes the decrease in the intensity of the bands at 2280 - 2260 cm -1, attributed to N20 molecules coordinated to strong Lewis acid sites. Simultaneously, the band of CO 2 is revealed at 2360 cm -1. Further heating to 520 K results in the diminution of the band at 2120 - 2150 cm -1, assigned to NO ~5+, and in the parallel growth of the CO 2 band. We may assume that the oxidation of CO or CH 4 with NO (or, vice versa, the NO reduction with CO or CH4) on the zeolites containing strong Lewis acid sites proceeds as follows. First, NO disproportionation takes place that yields N20. Upon heating, N20 is decomposed with the chemisorption of atomic oxygen. The latter seems to be a species consumed for CO or methane oxidation to CO 2. It is possible that the strongly adsorbed N20 molecules also participate in the oxidation reactions.

415

CONCLUSION 1. It is found that N20 adsorption on alkaline and alkaline-earth forms of zeolites leads to weak molecular complexes that do not participate in the reaction of CO or methane oxidation. On the contrary, very strong adsorption complexes are formed upon N20 admission on the zeolites containing strong Lewis acid sites. These complexes are the precursors of the chemisorbed atomic oxygen species that are active in the oxidation processes. 2. Unlike N20, nitric oxide undergoes disproportionation on the cationic forms yielding various NOx species that differ in their stability and oxidizing activity, the most active entities being N20 3 and NO 5+. In the case of the zeolites containing strong Lewis acid sites, N20 formed as a result of disproportionation, and chemisorbed atomic oxygen are responsible for the oxidation activity of the zeolites.

REFERENCES 1. M. Iwamoto, Stud. Surf. Sci. Catal., 54 (1990) 121. 2. H. Bosch and F. Janssen, Catal. Today, 2 (1987) 369. 3. H. Hamada, Y. Kintaichi, and M. Sasaki, AppI. Catal., 64 (1990) L I. 4. A. A. Davydov, IR-Spectroscopy Applied to the Chemistry of Oxide Surfaces, Nauka, Novosibirsk, 1984. 5. Y. Kanno, Y. Matsui, and H. Imai, J. Incl. Phenom., 5 (1987) 385. 6. Y. Kanno, Y. Matsui, and H. Imai, J. Incl. Phenom., 3 (1985) 461. 7. A. A. Alekseev, V. N. Filimonov, and A. N. Terenin, Dokl. Akad. Nauk SSSR, 147 (1962) 1392. 8. M. Iwamoto and H. Yahiro, Chem. Lett., (1990) 1967. 9. H. Miessner and J. Burkhardt, J. Chem. Soc., Faraday Trans., 86 (1990) 2329. 10. C. C. Chao and J. H. Lunsford, J. Am. Chem. Soc., 93 (1971) 6794. 11. W. E. Adisson and R. M. Barrer, J. Chem. Soc., 70 (1955) 757. 12. C. C. Chao and J. H. Lunsford, J. Am. Chem. Soc., 97 (1971) 71. 13. V. L. Zholobenko, L. M. Kustov, and V. B. Kazansky, Proc. 9th Int. Zeolite Conf., Montreal, 1992, Butterworth-Heinemann, 2 (1993) 199. 14. V. L. Zholobenko, I. N. Senchenya, L. M. Kustov, and V. B. Kazansky, Kinet. Katal., 32 (1991) 151. 15. V. B. Kazansky, V. Yu. Borovkov, and L. M. Kustov, Proc. 8th Int. Congr. on Catalysis, Dechema, Weinheim, 3 (1984) 3. 16. W. J. Mortier, Compilation of Extraframework Sites in Zeolites, Guilford, 1982.



417

Cracking of 1 , 3 , 5 - triisopropylbenzene over deeply dealuminated Y zeolite E.Falabella S-Aguiar 1'2, M.L.Murta-Valle 1, E.V. Sobrinho 3, D. Cardoso 3. Petrobr/ls / CENPES - Ilha do Fund~o, Q7, 21949-900, Rio de Janeiro, Brazil. Fax: 55-21-5986626 2Escola de Quimica/UFRJ, Rio de Janeiro, Brazil, Tel: 55-21-5903192. 3DEQ / UFSCar, S~o Carlos, Brazil, Fax: 55-162-748266.

ABSTRACT Cracking of a rather volmninous molecule, 1,3,5-triisopropylbenzene, was carried out over several zeolites with different degrees of dealumination. Very crystalline zeolites were prepared via combined steam/acid leaching treatments, being afterwards characterized by various techniques. IR-OH region showed that highly condensed EFAL, rather than non-condensed EFAL, was removed during the acid leaching. Nevertheless, MAS/27A1 NMR clearly demonstrated that EFAL-free zeolites are never obtained, regardless of the pH of the acid leaching step. Mesopores surface areas, determined by t-plot increased with increasing number of treatments, as well as the strength of acid sites. Initial rate of cracking of 1,3,5-triisopropylbenzene was plotted against the number of A1 atoms per unit cell, a maximum being obtained for 11 A1/u.c.. Since this molecule has a kinetic diameter larger than 8.0 A, it will not penetrate the zeolitic micropores. After dealumination, mesopores are generated and the reactant is allowed to diffuse. Therefore, both accessibility and acidity seem to control the rate of reaction.

1. INTRODUCTION During the last three decades, extensive work has been carried out aiming at the correlation between the acidity of several dealuminated zeolites and their catalytic activity in the cracking of hydrocarbons. Various molecules such as 2,3dimethylbutane [1], n-pentane [2], n-hexane [3], n-heptane [4], isooctane [5], decalin [6] and cumene [7] have been used in model reactions. Nevertheless, the use of heavier feedstocks in FCC processes has greatly increased the interest for the cracking chemistry of more voluminous molecules. However, not much has been

418 published regarding model reactions with bulky molecules. Previous studies developed in our group [8] have emphasized the importance of external surface area of small crystallite zeolites in the cracking of molecules with kinetic diameter larger than 8.0A such as 1,3,5-triisopropylbenzene (TIPB), showing that acid strength and site density are not the only parameters controlling activity and selectivity during catalytic cracking. The aim of the present work is to demonstrate that the presence of mesopores generated by acid leaching may also play an important role in the cracking of the same molecule (TIPB).

2. EXPERIMENTAL Parent NaY (Si/AI=2.8) zeolite was ion-exchanged in a 11 wt% NH4C1 solution, at 70~ After the ion-exchange, steaming treatments took place in a cylindrical reactor containing 60 g of the zeolite, with 20 g/h flow rate of 100% saturated steam (200~ at 650~ for 90 minutes. Acid leachings were carried out at 70~ for 30 minutes, with pH between 1.0 and 2.5, generating DAZ samples. Treatments were repeated three times and the sodium values were reduced to values below 0.1% Na20. X-ray diffraction (XRD) took place in a Phillips PW1729 diffractometer with CuKo~ radiation and the relative crystallinity of the samples was estimated from the integrated areas of the peaks with Miller indexes 220, 311, 33 l, 333, 440, 533, 642, 660, 751 and 664. Chemical composition was determined by X-ray fluorescence (XRF) in a Phillips PW1407 spectrometer with CrKct radiation. X-ray photoelectron spectroscopy (XPS) was carried out in a VG-Scientific ESCALAB MK II spectrometer with MgKtx radiation, using bands of Al(2p), Si(2p) and Na(ls) to determine surface chemical composition. MAS/NMR spectra of 29Si and 27A1 were obtained on a Varian VXR 300 spectrometer working under a 7.05T magnetic field, being 27A1 spectra collected after impregnation with acetylacetone. Infrared measurements were done on a FTIR Nicolet 60 SXR spectrometer and the OH region (3400-3800 cm -1) was observed. Nitrogen adsorption isotherms were determined using a Micromeritics ASAP2400. From the isotherms, BET surface area, micropores volume (t-plot) and volume of mesopores (BJH) were calculated. Cracking of 1,3,5-triisopropylbenzene (FLUKA 92075) was carried out in a differential fixed-bed gas phase plug flow reactor, with a 15 mm bed containing 5 wt% zeolite diluted in kaolin. Reaction products were analyzed by means of GC-MS (HP5890 and HP5970-B), column HP-PONA (50 m, 0.2 mm, 0.5 pro) for liquid phase and A1203/KC1 column (50 m, 0.3 mm, 0.5 ~rn) for the gas phase.

3. RESULTS AND DISCUSSION Table 1 presents the main characteristics of the parent NaY zeolite and the dealuminated samples prepared thereof. It is clear that dealumination generated very crystalline samples, as a good indication that the process took place without extensive

419 destruction of the framework. It is also evident that progressive dealumination was performed along the cycles, since A o values decreased considerably whereas Si/A1 ratio in the framework (measured by MAS/29Si NMR) increased accordingly. The comparison between Si/A1 ratios obtained by XRF and XPS indicates that aluminum surface contents (XPS) in the samples obtained by acid leaching are lower than the values obtained by XRF and similar to those obtained by 29Si NMR (framework aluminum contents). This reveals a preferential outer surface leaching of the extra framework aluminum by acid treatment [9-11].

Table 1 Main characteristics of the NaY zeolite and dealuminated samples Sample XRD XRF XPS

29Si NMR

Cryst.(%)

Ao(A )

Si/Alg

Si/Als

Si/Alf

NaY

100

24.63

2.8

3.0

2.7

DAZ 1

110

24.50

3.8

4.0

5.6

DAZ 2

119

24.34

11.2

15.0

16.2

DAZ 3

126

24.29

27.8

35.0

33.5

*g=global, s=superficial, f=framework

Results of textural properties of the samples are depicted in table 2, as well as 27A1 NMR relative intensities of aluminum peaks after acetylacetone impregnation. Besides the increasing values of mesopore volume with the increasing dealumination conditions, one can also verify that micropore volume is growing, being larger than the value encountered for the NaY zeolite. This fact indicates that in addition to the characteristic micropores of the parent zeolite, dealuminated samples present supermicropores with diameter between 10 and 20A [9]. The 27A1 NMR spectra display two distinct peaks for tetrahedral (framework) and octahedral (non-framework) aluminum [9-12]. As showed in figure 1, all samples contain extraframework aluminum species (EFAL), regardless of the acid leaching step. Although acid leaching reduces EFAL contents, results of the table 2 indicate that dealuminated samples contain around 30% of EFAL, as an evidence that diffusional limitations in the removal of aluminum species are taking place. In fact, infrared spectra in the OH region after acid leaching revealed the disappearance of the band at 3690 cm l , which is ascribed to highly condensed EFAL located in the larger cavities [ 13]. However, the band at 3600 cm 1, related to low

420

Table 2 Textural properties and 27A1 NMR relative intensifies of the NaY zeolite and dealuminated samples 27A1NMR Sample MiPV a

MePV b

% tetrahedral

%octahedral

NaY

0.326

0.031

100

DAZ 1

0.326

0.092

68.5

31.5

DAZ 2

0.358

0.191

70.7

29.3

DAZ 3

0.360

0.188

64.8

35.2

a MiPV = volume of micropores (cm3/g) b MePV = volume ofmesopores (cm3/g)

DAZ 1

DAZ 2

DAZ 3

O

100

0

100

100

0

-100

100

0

-100

--9 p.p.m. Figure 1. MAS/27A1 NMR spectra for DAZ samples. (T) Tetrahedral, (O) Octahedral

condensation non-framework ahanina located at smaller cavities, remained almost intact. This confmn our assumption that non-accessible EFAL is hardly removed by acid treatment. Rates of 1,3,5-triisopropylbenzene cracking were calculated assuming differential behavior. Plots of rate of TIPB disappearance against time-on-stream were fit to the classical Voohries equation (r = r o t -n) [ 14], allowing one to estimate initial rates of reaction. Initial rates of TIPB disappearance were then plotted against the number of aluminum atoms per unit cell (A1/u.c.), according to NMR results. Such graph is

421 presented in figure 2, for three different temperatures (400, 425 e 450~ For all temperatures, a maximum is achieved for about 11 A1/u.c.. These results contradict previous published data which found a maximum for 30 A1/u.c. in the cracking of, nhexane [3],for instance. Although extensive published literature has often encountered a maximum of activity between 30 - 40 A1/u.c., it must be borne in mind that all model reactions employed in previous studies used small molecules such as linear paraffins (hexane, heptane) or light aromatics (cumene). Explanations for maximum occurrence were based upon changes in the strength, type and concentration of acid sites [3,9,15,16]. However, when voluminous molecules are used, the effect of the formation of mesopores (and thus changes in the external surface area of the zeolites) must not be disregarded. The increase in surface area would certainly be important for diffusion limited reactions, which is apparently the case of the cracking of TIPB. Thus, DAZ 2 and DAZ 3 zeolites possess similar values of micro and mesopore volumes, which, in turn, are higher than the values obtained for DAZ 1 (table 2). Nevertheless, DAZ 2 is more active than DAZ 3 because of the higher concentration of acid sites thereof. This fact confLrms that in the cracking of bulky molecules, both acidity and diffusion are playing an important role in the rate determining step.

3.60

3.20

2.80

~

2.40 -

2.00

1.60

~ 6

l 10

,

l f ~ 18 20 AI/u.c. (NMR)

~

1

l

28

30

Figure 2. Initial cracking rates of 1,3,5-TIPB as a function of A1/u.c. (NMR). (a) 400~ (o) 425~ and ( . ) 450~

422 4. CONCLUSIONS Cracking of TIPB was studied over several acid leached steam treated zeolites and initial rates of disappearance were plotted against the number of A1/u.c.. For three different temperatures, a maximum of activity was obtained for 11 A1/u.c., contradicting previous results. Apparently, as not yet observed in the cracking of smaller molecules, the accessibility of the larger molecule to the internal zeolitic pores is controlling the rate of reaction. Therefore, maxima are encountered for zeolites which display both high mesoporosity and considerable acid site concentration. In the case of acid leaching of ultra-stable samples this is a compromise between acidity and accessibility, since the combined treatments generate mesoporosity but also reduce the acid site concentration.

REFERENCES

1 2 3 4 5 6

G.R. Bamwenda, Y.X. Zhao, B.W. Wojciechowski. J. Catal. 150(1994)243. P.V. Shertukde, W.K. Hall, J.M. Dereppe, G. Marcelin. J. Catal. 139(1993)468. J.R. Sohn, S.J. DeCanio, P.O. Fritz, J.H. Lunsford. J. Phys. Chem. 90(1986)4847 A. Corma, J.Planelles, J.Sanches-Marin, F.Tomfis. J. Catal. 93(1985)30. R. Beamnont, D. Barthomeuf. J. Catal. 30(1973)288. E. Falabella S-Aguiar, M.P. Silva, M.L.Murta-Valle, D.F. Silva. ACS Division of Petroleum Chemistry, Inc. Preprints, 39-3(1994)356. 7 S.J. DeCanio, J.R. Sohn, P.O. Fritz, J.H. Lunsford. J. Catal. 101(1986)132. 8 E. Falabella S-Aguiar, M.L.Murta-Valle, M.P. Silva, D.F. Silva. Proc. of the 1st EUROPACAT, Montpellier, 1(1994) 104. 9 J. Scherzer in T.E. Whyte Jr.(Editor). The Preparation and Characterization of Aluminum Deficient Zeolites (ACS Symp.Series 248), New York, 1984, p.157. 10 G. Fleisch, B.L. Meyers, G.J. Ray, C.L. Marshall. J. Catal. 99(1986)117. 11 E.V. Sobrinho, D. Cardoso, E. Falabella S-Aguiar, J.G. Silva. XIV Simp. Iberoamericano Catal, Concepci6n, 1994, p.433. 12 G. Engelhardt in H.V. Bekkun (Editor). Solid State NMR Spectroscopy Applied to Zeolites, Elsevier, 1991, p.285. 13 R.D. Shannon, K.H. Gardner, R.H. Staley, G. Bergeret, P. Gallezot, A.Auroux. J. Phys. Chem. 89(1985)4778. 14 A. Voorhies. Ind. Eng. Chem. 37(1945)318. 15 D. Barthomeuf. Materials Chem. Phys. 17(1978)49. 16 H. Stach. Catal. Letters 13(1992)339.

Zeolites: A Refined Tool for Designing Catalytic Sites L. Bonneviot and S. Kaliaguine(editors) 9 1995 Elsevier Science B.V. All rights reserved.

423

H y d r o g e n a t i o n of Styrene and H y d r o g e n o l y s i s of 2- Phenylethanol

Mechanistic Study of the Side-Chain Alkylation of Toluene and Methanol Tawan Sooknoi and John Dwyer Centre for Microporous Materials, Department of Chemistry, UMIST, PO Box 88, Manchester, M60 1QD, United Kingdom

Introduction A mechanism reported previously for the side-chain alkylation of toluene with methanol suggests that formaldehyde is the alkylating agent, primarily forming styrene, which then undergoes hydrogenation to ethylbenzene with small amounts of hydrogen, created by decomposition of methanol. In the present work mechanisms involving hydrogenolysis of 2phenylethanol intermediates, or involving the direct alkylation of methanol, are suggested to be predominant routes to the formation of ethylbenzene. The latter mechanism appears to be especially promoted in the presence of excess of Cs over stoichiometric ion exchange requirement of zeolite X. It is proposed that the intermediate, 2-phenylethanol, is formed from the reaction between toluene and formaldehyde. The fact that it is not detected in the product mixture is attributed to its rapid dehydration to styrene and, as proposed in Figure 1, hydrogenolysis to ethylbenzene.

(1)

(U"

(2)

H

Figure 1 Formation of 2-phenylethanol and its reactions 1) Dehydration to styrene 2) Hydrogenolysis to ethylbenzene. In this work, the hydrogenation of styrene and the hydrogenolysis of 2-phenylethanol are demonstrated over Cs exchanged zeolite X under the same reaction conditions as used for the side-chain alkylation. We postulate the formation of ethylbenzene either from direct alkylation of toluene with methanol or from the reaction of both styrene and 2-phenylethanol with hydrogen produced from the decomposition of methanol. The overall mechanism is shown in Figure 2. The ethylbenzene/styrene ratio is used as a parameter to determine the relative reactivities of direct alkylation, hydrogenation and hydrogenolysis.

424

U

CH3

cH3oH MI

J

H2

"U"

T

H2

M5

M3

Figure 2 Overall mechanism for the side-chain alkylation reaction

Experimental procedure CsNaX was used in side-chain alkylation of toluene, hydrogenation of styrene and hydrogenolysis of 2-phenylethanol. Molecular sieve 13X (BDH | was ion-exchanged three times with 0.5 M CsC1 at 50 ~ and once with 0.5 M CsOH at room temperature. The solid material was separated into two portions. The first was washed several times with 0.5 M CsOH and left to dry at room temperature overnight. This material believed to contain clusters of CsOH (possibly CsHCO3/Cs2CO3) was defined as "CsNaX-I" catalyst. The second portion was washed with deionised water until no basicity was detected. The Cs cations in this material, designated as "CsNaX-2", were presumably exchanged ions. The reactions were carried out in a fixed bed down-flow reactor at 350 ~ C with helium as carrier gas. In the hydrogenation of styrene and the hydrogenolysis of 2-phenylethanol, hydrogen was also employed. Mixtures of toluene and methanol were continuously fed by syringe pump. In order to enhance the decomposition of methanol, and hence the amount of hydrogen, a toluene/methanol ratio of 0.2 was used. Styrene and 2-phenylethanol were fed as pure components. In addition, --15-20% styrene in toluene and --15-20% 2-phenylethanol in toluene were also employed to test the hydrogenation and hydrogenolysis reactions. Liquid products were collected in an acetone-dry ice bath every 30 minutes and separated using a Chromosorb 20M column at 140-180~ with helium flowrate 30 ml/min as carrier gas. Gas products were periodically detected by on-line gas chromatography using a 10 ft Molecular sieve 13X and a 6 ft Chromosorb 20M column at 40 ~ C. Helium was again used as carrier gas at the rate of 30 ml/min.

Results and Discussions Side-chain alkylation of toluene with methanol resulted in ethylbenzene, styrene and, in some case, traces of cumene and a-methylstyrene. The gas products contain carbon monoxide, carbon dioxide and small amounts of methane. No dimethylether was found which reflects the basicity of the catalysts. When CsNaX-1 was used, more ethylbenzene than styrene was produced and cumene, together with a trace of a-methylstyrene, was also found in the product mixtures. On the other hand, styrene was a major product over CsNaX-2 and additionally, no cumene or a-methylstyrene was observed. The conversion of toluene was 10% and 5% for CsNaX-1 and CsNaX-2, respectively, but the conversion of methanol over these catalysts was higher than that of toluene owing to the parallel decomposition of methanol to carbon monoxide. For hydrogenation of styrene 4% hydrogen was used in the carrier gas because, in side-chain alkylation, the hydrogen produced was found to be less than 4% of the total gas. The styrene hydrogenation over CsNaX-1 and CsNaX-2 catalysts using the same conditions gave small amounts of toluene, ethylbenzene and benzene. Most of the styrene remained unreacted. Although hydrogen was used as carrier gas, the yield of ethylbenzene was not significantly increased with both catalysts. However, when CsNaX-1 was used, there was an initial increase in hydrogenation products, followed by a rapid decrease in reactivity to a level similar to that observed with CsNaX-2. When feed mixtures of styrene and toluene were used (4% hydrogen), a dramatic increase in hydrogenation product was observed (Table 1) but the conversion of styrene was quite low.

425

Hydrogenolysis of 2-phenylethanol (4% hydrogen ), over both catalysts, gave product compositions similar to that obtained from hydrogenation of styrene but with a relatively higher yield of hydrogenolysis products. Interestingly, c~-methylstyrene, which results from the sidechain alkylation of e t h y l b e n z e n e with formaldehyde, and 3 - p h e n y l p r o p e n e were observed. About 10% of 2-phenylethanol is converted by direct reaction with hydrogen (Fig 1 scheme 2 and Fig 5 scheme a) and by decomposition to toluene (1), the rest is directly converted to styrene via dehydration (Fig 1 scheme 1). Even when hydrogen was used as carrier gas, the yield of hydrogenolysis products increased by only 2-3%. However, when a mixture of toluene and 2phenylethanol, at a composition roughly the same as that expected in the side-chain alkylation, was used (4% hydrogen), more than 90% of the 2-phenylethanol was h y d r o g e n o l y s e d at the beginning of the reaction subsequently the hydrogenolysis selectivity decreased and remained constant, at 50%, after 3 hours on stream. The product distributions are listed in Table 1. Table 1 Product distribution in side-chain alkylation of toluene, hydrogenation of styrene and hydrogenolysis of 2-phenylethanol (%mole).

Reaction S S Hs Hs Hts Hp Hp Htp

Catalyst H2* Conversion Benzene Toluene Ethylbenzene Styrene Cumene at-Methylstyrene CsNaX- 1 CsNaX-2 CsNaX-1 CsNaX-2 CsNaX- 1 CsNaX- 1 CsNaX-2 CsNaX- 1

4% 4% 4% 4% 4% 4%

11a 5a 1.6b 1.5b 8.4c 100b 100b 100c

trace 0.1 trace 0.5 0.1 trace 0.9

0.8 0.7 5.2 5.7 4.3 34.8

7.3 0.8 0.7 0.7 2.5 3.2 1.8 15.5

2.5 3.8 89.4 91.6 46.4

1.0 -

trace 1.5 1.4 1.9

Yields represent the average yield over 180 minutes time on stream, reaction temperature 350 ~ catalyst 1 g, contact time 0.015 min, W/F = 47 g.mol.h -1 (total feed), He carrier gas 25 ml/min S = Side-chain alkylation of toluene, H s = Hydrogenation of styrene, Hp = Hydrogenolysis of 2-phenylethanol, Hts = Hydrogenation of Styrene in the presence of toluene and Htp = Hydrogenolysis of 2-phenylethanol in the presence of toluene. aConversion based on Toluene b Conversion based on styrene or 2-phenylethanol CConversion based on styrene or 2-phenylethanol in the mixtures * %Hydrogen in Helium carrier gas (mol/mol)

Table 2 Ethylbenzene/Styrene ratio (mol/mol)

Reactions

Feed

Catalysts

Side-Chain alkylation Side-Chain alkylation Hydrogenation Hydrogenation Hydrogenation Hydrogenolysis Hydrogenolysis Hydrogenolysis

Toluene/methanol = 0.2 Toluene/methanol = 0.2 Styrene (pure) Styrene (pure) Toluene/styrene - 5 2-Phenylethanol (pure) 2-Phenylethanol (pure) Toluene/2-phenylethanol- 5

CsNaX-1 CsNaX-2 CsNaX-1 CsNaX-2 CsNaX-1 CsNaX-1 CsNaX-2 CsNaX-1

Hydrogen in cartier gas

-4% --4% --4% -4% --4% --4%

Ethylbenzene/ Styrene 3.16 0.20 --" 0.01 -

0.01

- 0.03 -0.04 -0.02 0.26

426 Table 2 shows the ratio of the yield of ethylbenzene to the yield of styrene in the product mixture for each reaction (averaged over the first 3 hours). In the hydrogenolysis reaction using mixtures of toluene and 2-phenylethanol and the hydrogenation reaction using mixtures of toluene and styrene, yields of the products are calculated, respectively, based on the amount of 2-phenylethanol and styrene in the feed mixtures. The ethylbenzene/styrene ratio for the side-chain alkylation is considerably higher for CsNaX-1 indicating that the formation of ethylbenzene is influenced by the Cs "clusters". It is suggested that the Cs "clusters" enhance either direct alkylation (Path M1, Fig 2 ) o r hydrogenation and hydrogenolysis (Path M4 and Ms, Fig 2). It seems clear that ethylbenzene is not produced, in the main, from hydrogenation of styrene but from the hydrogenolysis of the 2phenylethanol intermediate. A higher conversion of 2-phenylethanol to ethylbenzene by hydrogenolysis is obtained in the presence of toluene because the dehydration of 2phenylethanol, to give styrene, does not require the more active basic sites. The sites inducing only a weak electron donor-acceptor interaction can adequately promote dehydration of 2phenylethanol to styrene. For example, conversion of 2-phenylethanol over "silica-alumina" (1) results in styrene as product. In separate work (8), we also test dehydration of 2-phenylethanol over "silicalite-I" and over "silica gel" (Fig 3) -H20

H

H

Oxygen Bridge Sites (Silicalite) Si

Si

Si -H20 H

H

Na

-

/ O ~ Si

~ AI

/0 ~

H

/

Exchanged Cation Sites (NaX)

I

/0

OH

OH

0

I

I

I

-H20 -

Silanol Sites (Silica gel)

Silo / Silo / Silo

Figure 3 Three possible sites promoting the dehydration of 2-phenylethanol to styrene Presuming that 2-phenylethanol is more strongly adsorbed on the more basic sites, (Cs "clusters" or Cs exchanged sites) which promote hydrogenolysis to ethylbenzene, than on other electron donor-acceptor sites (e.g. oxygen bridges or exchangeable cations other than Cs), then, as the partial pressure of 2-phenylethanol is increased, relatively more conversion can take place on the more weakly adsorbing sites which, we presume, promote dehydration of 2phenylethanol to styrene, as observed with silicalite and silica gel. This may be the explanation for the reduced yields of ethylbenzene at the higher partial pressure of 2-phenylethanol (pure feed) as compared with the mixtures of 2-phenylethanol and toluene. This, of course, assumes that toluene acts largely as diluent. The assumption that 2-phenylethanol is more strongly adsorbed than toluene would result in inhibition of toluene sorption by the product (2phenylethanol) and also inhibition of the side-chain alkylation reaction rate.

427 Previous work (2) suggests that the side-chain alkylation can indeed be inhibited by the reaction products and we suggest that 2-phenylethanol is involved in this inhibition. Moreover, toluene, which is a reactant in the side-chain alkylation, is found to be the major product of 2phenylethanol hydrogenolysis (Table 1 and Fig 4), therefore, the side-chain alkylation of toluene seems unfavourable compared with ring alkylation to give xylene over acid catalysts.

CH3OH

Figure 4 Toluene and 2-phenylethanol cycle The observation that toluene is a major product from the hydrogenolysis of 2phenylethanol also provides evidence to support the effect of toluene/methanol ratio observed by Yashima (3) who reported that a higher toluene/methanol ratio (> 5) gave better results in sidechain alkylation than reactions using more methanol in the feed. We suggest that the lower conversion of toluene in the side-chain alkylation, when methanol concentration is high, is due to the competitive adsorption of methanol and the increase of hydrogen by decomposition of methanol which leads to more hydrogenolysis of 2-phenylethanol to give toluene and ethylbenzene. A high ratio of toluene/methanol in the side-chain alkylation reduces the concentration of methanol in the feed, diminishes the rate of direct alkylation and also reduces methanol decomposition to carbon monoxide and hydrogen so that less hydrogen is available for hydrogenolysis of 2-phenylethanol. This is reflected in a decrease in yield of ethylbenzene when feed methanol is decreased within the experimental range used (3) The mechanism for hydrogenolysis of 2-phenylethanol to toluene is proposed in Figure 5. Additionally, the presence of a-methylstyrene in products from hydrogenolysis of 2phenylethanol suggests that the catalysts are highly active in promoting side-chain alkylation of ethylbenzene with formaldehyde which results either from hydrogenolysis of 2-phenylethanol to toluene (Fig 5a) or from direct decomposition of 2-phenylethanol to give toluene as observed by Vivekanandan (1). The methanol from these reactions (Fig 5a) is rapidly decomposed to formaldehyde and to carbon monoxide which is also detected in small amounts during the reaction. The 2-phenylpropanol, which, we suggest, is an intermediate in the side-chain alkylation of ethylbenzene (Fig 5d), undergoes dehydration to a-methylstyrene as shown in Figure 5(e). No cumene (e.g. from hydrogenolysis of 2-phenylpropanol) is detected in the product mixtures, possibly due to its low concentration. However, direct alkylation to cumene is also not expected because the amounts of methanol produced by hydrogenolysis of 2phenylpropanol are small in proportion to the number of active basic sites which, presumably, results in rapid decomposition of methanol to formaldehyde and to carbon monoxide.The ethylbenzene/styrene ratio from the hydrogenolysis of 2-phenylethanol (4% hydrogen) in the presence of toluene using CsNaX-1 catalyst is similar to that for side-chain alkylation using CsNaX-2 catalyst whereas the ethylbenzene/styrene ratio from the hydrogenation of styrene (4% hydrogen) in the presence of toluene using CsNaX-1 catalyst is very much lower. This implies that, when Cs exchanged zeolites containing no "clusters" are used in side-chain alkylation, ethylbenzene is mainly produced from hydrogenolysis of the 2-phenylethanol (Path M4, Fig 2), and considerably less is produced from hydrogenation of styrene. This could also explain the lack of influence of transition metals in the catalysts, on selectivity of ethylbenzene and styrene (4,5,9) because the mechanism proposed here involves nucleophilic attack by a hydride anion generated from dissociation of hydrogen to H- and H + over the basic catalyst, and not over the transition metals.

428 | OH

HG / Basicsite

CH3

(a) .~ CH3OH +

~~'~2HH2 (d)

OH

~'-- ~ M . ~

(e) -HE0

2-phenylpropanol 120 Figure 5 Formation of ct-methylstyrene in hydrogenolysis of 2-phenylethanol According to the above reaction schemes (Fig 2, 3 and 5), it is impossible to obtain yields of ethylbenzene greater than yields of styrene from the side-chain alkylation of toluene with formaldehyde from the decomposition of methanol. Therefore, direct alkylation of toluene with methanol must take place in the reaction using catalysts containing Cs "clusters" in order to obtain the observed high yields of ethylbenzene (ethylbenzene/styrene ratio > 3). This is supported by the reaction over K exchanged zeolite X by Yashima (3) who used formaldehyde as alkylating agent and found that the ethylbenzene/styrene ratio is about 0.3 and is also supported by the work of Itoh (4) who found an ethylbenzene/styrene ratio less than 1 when KX was used as catalyst. In contrast, Hathaway (6) obtained higher ethylbenzene yields when CsNaY containing cesium acetate "clusters" were employed and, interestingly, Engelhardt (7) who used KX washed with KOH, as catalyst, obtained an ethylbenzene/styrene ratio higher than 10. In addition to this result, Engelhardt also found that the ethylbenzene/styrene ratio decreased with the number of washings of this catalyst with water. This strongly suggests that the "clusters" provide the active basic sites for the direct alkylation since extensive washing would remove K cations in excess of stoichiometry. In addition, the direct alkylation of toluene was also observed in the reaction between toluene and ethylene over Rb exchanged zeolite X.(3) Formation of cumene in the side-chain alkylation of toluene with methanol over CsNaX-1 is further evidence that direct alkylation of alkylbenzene with methanol takes place in the presence of Cs "clusters". Over strongly basic catalysts, ethylbenzene, produced from primary side-chain alkylation of toluene with methanol, undergoes secondary alkylation with methanol to form cumene directly (8). Some of the ethylbenzene can also react with formaldehyde, but in much smaller amounts, giving only traces of a-methylstyrene. We concluded that catalysts, for side-chain alkylation, perform two important roles; the first is the basicity which promotes proton abstraction of benzylic hydrogen to generate a "carbanion like" intermediate and the second is the stabilisation of this intermediate by the highly polar framework. Catalysts lacking the ability to stabilise this intermediate, will not be able to promote side-chain alkylation. For example, MgO which is a strongly basic solid catalyst cannot catalyse side-chain alkylation (6), presumably because the catalyst has no polar framework to stabilise the appropriated intermediates which could be generated by its basicity. For this reason zeolite X, which has a particularly high ion exchange capacity (the most polar framework), appears to be the most active catalyst for the side-chain alkylation. Cs exchanged zeolite X (CsNaX-2) seems to be sufficiently basic to generate benzyl "carbanion like" intermediates and to stabilise them within the highly polar framework. The negative charge of the intermediate is presumably delocalised into the benzene ring which, as suggested (10,11), is sitting on two (or more) Cs cations at the active site surrounding by a number of Cs and Na cations in the framework. This makes the benzyl carbanion sufficiently stable to facilitate reaction with more electrophilic species such as formaldehyde. Direct alkylation, therefore, is rarely promoted by Cs exchanged zeolites without Cs "clusters" and ethylbenzene is only produced from hydrogenolysis of 2-phenylethanol intermediates leading to

429 a low ethylbenzene/styrene product ratio. When Cs "clusters" are incorporated into Cs exchanged zeolite crystals (CsNaX-1), the benzyl carbanion is certainly generated by the strong basicity of the cluster, together with that of the negatively charged framework. Cs cations in "clusters" can stabilise "carbanion-like" intermediates by providing a stronger basic site localised on the "clusters". These species are sufficiently stable to facilitate proton abstraction of the benzylic hydrogen and localisation of the charge on the benzyl "carbanion" which is then more available to react with appropriate electrophilic species. In other words, Cs "clusters" in Cs exchanged zeolites generate more reactive benzyl "carbanion" intermediates than Cs exchanged zeolites without "clusters", and this results in the alkylation with both methanol and formaldehyde. Consequently, the yield of ethylbenzene in reactions using CsNaX-1 is higher than that of styrene since ethylbenzene is produced by both hydrogenolysis, and also by the direct alkylation with methanol which is promoted by incorporated Cs "clusters".

References 1. G. Vivekanandan, C. S. Swaminathan and V. Krishnasami, Ind. J. Chem., 32A, 215-220 (1993) 2. P. Beltrame, P. Fumagalli and G. Zuretti, Ind. Eng. Chem. Res., 32, 26-30 (1993). 3. T. Yashima, K. Sato, T. Hayasaka and N. Hara, J. Cat., 26, 303-312 (1972) 4. H. Itoh, A. Miyamoto and Y. Murakami, J. Cat., 64, 284-294 (1980). 5. C. Lacroix, A. Deluzarche, A. Kiennemann and A. Boyer, Zeolites, 4, 109-111 (1984) 6. P. Hathaway and M. Davis, J. Cat., 119, 497-507 (1989). 7. J. Englehardt, J. Szanyi and J. Volyon, J. Cat., 107, 296-306 (1987). 8. T. Sooknoi and J. Dwyer, Unpublished papers, Department of Chemistry, UMIST (1995). 9. C. Lacroix, A. Deluzarche, A. Kiennemann and A. Boyer, J. Chim. Phys., 81, 473479,481-485,487-490 (1984). 10. K. Tanabe, M. Misono, Y. Ono and H. Hattori, Stud. Surf. Sci. Cat., 51,233-234 (1989). 11. M. L. Unland and G.E. Baber, Catalysis of Organic Reaction, Marcel Delecker, 54 (1981).



431

Catalyst deactivation of high silica metallosilicates in Beckmann rearrangement of cyclohexanone oxime Takashige Takahashi, Takami Kai and M.N.A.Nasution Department of Applied Chemistry & Chemical Engineering, Faculty of Engineering, Kagoshima University, Kagoshima 890 Japan The vapor phase Beckmann rearrangemem was carried out over high silica ZSM-5 type metallosilicates to elucidate the effect of acid strength on ~ -caprolactam selectivity and catalyst deactivation rate. It was found that the indiosilicate which had the lowest acid strength was the best catalyst among the metallosilicates. When carbon dioxide and methanol were used as diluent gas and diluent solvent of cyclohexanone oxime, respectively, the deactivation rate decreased over indiosilicate. Furthermore, when the indiosilicate was modified with a precious metal, the catalyst deactivation significantly decreased. It was considered that the oxidation of coke on the surface was accelerated by the precious metal. 1. INTRODUCTION Although the vapor phase Beckmann rearrangement of cyclohexanone oxime (CHO) is known to be an attractive process to prepare ~ -caprolactam (CL) as the starting compound of nylon-6, the industrial process was not achieved due to the low selectivity to CL and rapid catalyst deactivation of the solid catalysts. Recently, some investigations were carried out to improve the CL selectivity. Sato et al. reported that a high silica HZSM-5 zeolite (SiO2/A120 3 ratio = 3200) was effective for the rearrangement reaction and the selectivity was 90 % or more [ 1]. They also reported that when an alcohol was used as the diluent solvent of CHO, the selectivity and catalyst life were significantly improved [2]. Furthermore, when the CHO vapor diluted with carbon dioxide was fed into the catalyst layer, the selectivity to the CL was higher than that diluted with nitrogen [3]. It was reported that the catalyst life and the selectivity to CL also increased by use of boria supported HZSM-5 zeolite [4, 5], MEL type zeolite [6], femerite zeolite [7] and bona deposited on an alumina or silica by a CVD method [8, 9]. The CL selectivity gradually increased by use of the new type catalysts, but the catalyst deactivation was not overcome so far. Many questions on the reaction mechamsm of the Beckmann rearrangement over the zeolites and on the deactivation mechanism of the zeolites still remained.

432 In the presem study, the rearrangement has been camed out over a ZSM-5 type metallosilicates including iron, gallium or radium in the crystal lattice to elucidate the effect of acid strength on the catalyst deactivation rate. We have also examined the effects of diluent solvent and diluent gas of CHO on the deactivation rate and CL selectivity. Furthermore, the rearrangement was camed out over an mdiosilicate modified with precious metals to decrease the catalyst deactivation. 2. EXPERIMENTAL

A high silica ZSM-5 type metallosilicate with gallium, iron or indium in the crystal lattice was synthesized by modifying the method described in a previous paper [ 10]. In the present study, the ratio of silica/metal changed from 500 to 3200. After the powder of the silicates was compressed into a small die, it was crushed to 32 -~ 48 mesh. The surface area was measured by a nitrogen adsorption method, in which Langmuir equation was used for calculation. The acidity and acid strength distribution were measured using ammonia temperature programmed desorption method. The impregnation method was used for the precious metal modification of zeolites with silica/metal ratio = 500. The vapor phase reaction of CHO has been carried out at atmospheric pressure in a flow type system. The reactor assembly was essentially similar to that reported in a previous paper [4]. 3.RESULTS AND DISCUSSION It has been reported that the catalytic activity of zeolites decreased with time on stream due to the deposition of coke on the strong acid sites in the reaction of CHO. Figure 1 shows the relationship between CHO conversion and time on stream over Z5(Ga)-500H and Z5(Ga)-1000H. Z5(Ga)-500H means proton exchanged gallosilicate whose Si/Ga ratio is 500. Since similar relationships between CHO conversion and time on stream were obtained over the metallosilicates, the effect of time on stream on CHO conversion is represemed by Equation (1). x(t)---x(O) 9exp(-b 9t)

(1)

where x(t) and x(0) are CHO conversion at any time on stream and initial conversion, respectively, b is deactivation coefficient and t is time on stream. The deactivation coefficients of various proton exchanged metallosilicates with the same Si/M ratio (500) are shown in Table 1. This table also shows the strong acid concentration and the maximum temperature of strong acid sites of the silicates measured by ammonia TPD method. The maximum

433

1.0 0.8 0.6

~

17 O tn

!

0.4 -

k.,

Reaction temp. = 623 K WlFA0 = 113 kg.s/mot Diluent gas = H2 Diluent solvent = Benzene

c 0.2

O Z5(Ga)-1OOOH 9Z5(Ga)- 500H

0 U

0.1 ~ 0

10

~

~ 20 30 40 50 Tim~ on stream [ m i n i

60

Figure 1. Catalyst deactivation of Z5(Ga)-500H and Z5(Ga)-1000H. Table 1 Strong acid concentration, deactivation factor(b) and maximum temperature of ammonia TPD Catalyst Z5(A1)500-H Z5(Fe)500-H Z5(Ga)500-H Z5(In)500-H

Strong acid conc. 1) [mmol/g] 0.058 0.046 0.044 0.039

Deact. factor [I/s] 2.7 x 10-4 2.4 x 10-4 1.8 • 10-4 4.0 • 10-5

Max. temp. [K] 758 749 733 706

1) Strong acid means the acid sites can retain ammonia at >680 K. Reaction temperature=623 K, Diluent gas=H 2, Diluent solvent=Benzene temperature of aluminosilicate is highest among the silicates. At the same time, the deactivation coefficiem is also the highest. On the other hand, the deactivation coefficient of the indiosilicate with the lowest maximum temperature is smallest among the silicates. This result indicates that the catalyst deactivation can decrease to control the acid strength of the strong acid sites m the metallosilicate. It was found that when the change of acid strength distribution of the used silicate was examined by ammonia TPD method, the strong acid concentration did not reduce as expected from the decrease in CHO conversion. Furthermore, the surface area of the used silicate was found to be almost the same as that of the fresh silicate. These results suggest that the coke on the surface should be removed on heating for 1 h at 773 K in a nitrogen stream or evacuating for 2 h at 473 K ~mder 1 torr. If the coke was removed from the

434

1.0 A

9

"

t

7_>_0.8 m 0.6 - Catalyst 9Z 5 ( I n ) - 500H =Reaction temp. = 623 K O Diluent gas = H2 solvent = Benzene O.4 -Diluent W/FAo = 120 kg .s/mo[ O ~

(J

0.2. 0

I 1

I 2

~- Evacuated at 473 K f o r 2h CHO Conversion

~

CL Selectivity

I I I I 5(0) I 2 3 4 T i m e on st ream [ h ]

I 3

Figure 2. Effect of evacuation on recovery of catalytic activity. silicate surface, the catalytic activity would be recovered by the evacuatton. Figure 2 demonstrates the effect of the evacuation of the used silicates on the CHO conversion. The catalytic activity decreases with the time on stream as shown in Figure 2. The reaction was terminated after 5 h of time on stream. The silicate taken out from a reactor was evacuated for 2 h at 473 K under 1 torr. After the silicate was recharged in the reactor, the reaction was started under the same conditions. Although the deactivation rate of the treated silicate is higher than that of the ti'esh silicate,the catalytic activity is recovered as expected. On the other hand, when the used mordenite was treated by the same conditions,the activity did not increased. This result indicates that the carbonaceous deposit on the metallosilicate should be low molecular weight. Recently, Ichihashi et al.[2] reported that when the solvent of CHO changed from benzene to methanol or other alcohols,the CL selectivity and the catalyst life of the high silica aluminosilicate were simultaneously improved. They also reported that when the carrier gas changed from hydrogen to carbon dioxide, the CL selectivity was improved [3]. The reaction of CHO was carried om under the same diluent gas and solvent over Z5(In)500-H. Figure 3 demonstrates the relationship between CHO conversion and CL selectivity and time on stream. The lactam selectivity and catalyst life are significantly improved under the reaction conditions. These results suggest that the oxygen atom in alcohol or carbon dioxide would be importm~t for increasing the catalyst performances. If the coke is removed with oxygen on the silicate surface, much higher selectivity and longer catalyst life will be expected on Beckmann rearrangement over the mdiosilicate modified by platinum. Figure 4 shows the relationship between CHO conversion and CL selectivity on time on stream over the

435

1.0

>" 0.95

.,=,..

O CHO Conversion

>

.m,.

u

9 CL Selectivity

-~ 0.90 U3

E O L_

O U

Catalyst : Z 5 ( I n ) - 5 O O R ~ O"'%M~O.." Reaction temp.= 623 K 0.85 -Diluent gas = CO2 Diluent solvent = n - Prol)anot WlFA0 ~ 118 kgis/mo{ 1 0.80 0 2 4 6 8 T i m e on stream [ h ]

10

Figure 3. Relationship between CHO conversion and CL selectivity over Z5(In)-500H.

1.00

0 CliO Conversion

I

,

i....i

>, 0.98

I

CL Selectivity

> o ~ .

u 0-96 151 m

- 0.94 -

Reaction temp. = 623 K Diluent gas = CO2 _ Diluent solvent = n - Propanot ~- 0.92 > W/FAo = 118 kg.s/mot

E O

. ~

E O

u 0.90

0

I

I

2

4

I

6 Time on st ream

I 8

10

[hi

Figure 4. Relationship between CHO conversion and CL selectwity over PtZ5(In)-500H. PtZ5(In)-500H. The selectivity to CL was constant throughout time on stream of 10 h. The CHO conversion gradually decreases with time on stream, but the deactivation coefficient calculated from Figure 4 is 1.29 • 10"6s. This value lS about two orders of

436 magnitude less than the aluminosilicate shown in Table 1. When CHO rearrangement was carded out over ruthenium or palladium modified indiosilicate under the same reaction conditions as shown in Figure 4, the CHO conversion and the selectivity to CL were almost same as the results shown in Figure 4. These results indicate that the precious metals with oxidation activity are effective to reduce the coke deposited on the metallosilicate surface. It is well known that an alcohol is easily dehydrated over acid catalysts. In this reaction system, since the metallosilicates with strong acid sites were used as the catalysts, the dehydration of n-propanol used as the diluent would occur. However, an ether or an olefin which would be produced by the dehydration was not detected by a gas chromatograph in this study, because the production rate would be too small to be measured. When CHO mixed with a small amount of water was fed into the reactor the catalyst deactivation rate decreased up to 0.3 wt% of water. This result suggests that water produced from n-propanol should play an important role in decreasing catalyst deactivation. The pentasil type indiosilicate with high silica ratio modified with precious metals could be effective catalyst for the vapor phase Beckmann rearrangement of CHO. Furthermore, when n-propanol and carbon dioxide were used as the diluent solvent and diluent gas, respectively, the catalyst deactivation rate was significantly depressed. The further investigations are required to develop a high CL selective catalyst with long catalyst life. REFERENCES

1. 2.

H. Sato, K. Hirose and M. Kitamura, Nippon Kagaku Kaishi, 1989 (1989) 548. M. Kitamura and H. Ichihashi, Prep. Acid-Base Catalysis II, The Organizing Committee of International Symposium Acid-Base, Sapporo (1993) p. 217. 3. H. Sato, N. Ishii, K. Hirose and S. Nakamura, Proc. 7th Int. Zeolite Conf., Y. Murakami, A. Iijima and J.W. Ward eds., Elsevier, Amsterdam, 1986, p. 755. 4. T. Takahashi, K. Ueno and T. Kai, Can. J. Chem. Eng., 69 (1991) 1096. 5. T. Takahashi, M. Nishi, Y. Tagawa and T. Kai, Microporous Materials, 3 (1995) 467. 6. J.S. Reddy, R. Ravishankar, S. Sivasanker and P. Ratnasamy, Catal. Lett., 17 (1993) 139. 7. K. Miura, T. Komatsu, S. Namba and T. Yashima, Prep. 64th Autumn Meeting of Chem. Soc. Japan (1992) p. 477. 8. H. Sato, S. Hasebe, H. Sakurai, K. Urabe and Y. Izumi, Appl. Catal., 29 (1987) 107. 9. H. Sato, K. Urabe and Y. Izumi, J. Catal., 102 (1986) 99. 10. T. Takahashi and X.Y. Yun, Research Report of Faculty of Engineering, Kagoshima University No. 26 (1984) 119.


437

IR studies on the reduction of nitric oxide with a m m o n i a over MFI-ferrisilicate

T. Komatsu, M.A. Uddin and T. Yashima D e p a r t m e n t of C h e m i s t r y , Tokyo I n s t i t u t e of Technology, Ookayama, Meguro-ku, Tokyo 152 Japan

IR studies were c a r r i e d out on H - f e r r i s i l i c a t e for the c a t a l y t i c r e d u c t i o n of nitric oxide with a m m o n i a in the p r e s e n c e of oxygen. Adsorbed NO 2 or n i t r a t e species was found to be the r e a c t i o n i n t e r m e d i a t e to form n i t r o g e n through the r e a c t i o n with adsorbed a m m o n i a species.

1. INTRODUCTION M e t a l c a t i o n s c o n s t i t u t i n g the f r a m e w o r k of m e t a l l o s i l i c a t e s are e x p e c t e d to have unique c a t a l y t i c p r o p e r t i e s b e c a u s e unlike pure m e t a l oxide t h e y have a t e t r a h e d r a l c o o r d i n a t i o n with [SiO 4] t e t r a h e d r a and are s e p a r a t e d from e a c h other. For f e r r i s i l i c a t e with MFI s t r u c t u r e , we have r e p o r t e d its unique c a t a lytic p r o p e r t i e s for the oxidation of CO [1,2] and the o x i d a t i v e d e h y d r o g e n a t i o n of alkanes [3,4] as c o m p a r e d with s u p p o r t e d iron oxide and F e 3 + - e x c h a n g e d ZSM-5. In this work, for the f u r t h e r u n d e r s t a n d i n g of the c a t a l y t i c p r o p e r t i e s of Fe 3+ in f e r r i s i l i c a t e , we applied f e r r i s i l i c a t e to the s e l e c t i v e c a t a l y t i c r e d u c t i o n (SCR) of nitric oxide with a m m o n i a in the p r e s e n c e of oxygen. R e c e n t l y , Fe ions in z e o l i t e s have been studied for this r e a c t i o n using F e - e x c h a n g e d Y zeolites [5]. We studied here on the r e a c t i o n i n t e r m e d i a t e in the SCR of NO with a m m o n i a on M F I - f e r r i s i l i c a t e , which we have found to be a c t i v e and s e l e c t i v e for this r e a c t i o n [6]. 2. EXPERIMENTAL M F I - f e r r i s i l i c a t e was p r e p a r e d by a usual h y d r o t h e r m a l synthesis [2] S i / F e a t o m i c r a t i o of 44 and e x c h a n g e d with an aqueous solution of followed by the c a l c i n a t i o n at 773 K in air to obtain H - f e r r i s i l i c a t e . e x c h a n g e d ZSM-5 (Fe-ZSM-5) was p r e p a r e d by an i o n - e x c h a n g e m e t h o d aqueous solution of FeC13 (Si/AI=29, Si/Fe=315). Iron oxide supported

to have NH4NO 3 Fe 3+with an on MFI-

438

Table 1 Reduction of nitric oxide with ammonia. Catalyst

NO conversion/%

NH 3 conversion/%

N 2 yield/%

56.3 76.5 -10.7 29.9

60.4 87.9 52.6 31.3

58.8 83.2 22.8 28.5

ill

H-ferrisilicate Fe-ZSM-5 FeOx/Sil H-ZSM-5 i

.ll

i,

React i o n t e m p e r a t u r e : 773 K. C a t a l y s t weight: 0.10 g (H-ferrisilicate and FeOx/Sil) , 0.70 g (Fe-ZSM-5 and H-ZSM-5). R e a c t a n t : NO (0.10%), NH 3 (0.10%) and 0 2 (2.0%) in helium. Total flow rate: 500 ml min -1.

silicalite (FeOx/Sil) was prepared by impregnating H-ZSM-5 with an aqueous solution of Fe(NO3) 3 followed by the calcination at 773 K in air. IR s p e c t r a were obtained at room t e m p e r a t u r e with self-supporting sample wafers of 11 mg cm -2 thickness placed in a quartz vacuum cell. Samples were e v a c u a t e d at 773 K for 2 h before the IR m e a s u r e m e n t .

3. R I ~ U L T S AND DISCUSSION 3.1. SCR of nitric oxide on H-ferrisilicate The c a t a l y t i c a c ti v i ty and s e l e c t i v i t y of H-ferrisilicate for SCR of nitric oxide with ammonia in the presence of excess oxygen were studied using a conventional fixed-bed flow-type r e a c t o r under atmospheric pressure co m p ar ed with those of Fe-ZSM-5, FeOx/Sil and H-ZSM-5. Table 1 shows the results obtained at 773 K a f t e r 1 h on stream. Iron co n t en t s of the c a t a l y s t s ex cep t for H-ZSM-5 was adjusted to be the same by changing the c a t a l y s t weight. Nitrogen was produced on all the catalysts. Though H-ferrisilicate gave lower a c t i v i t y than Fe-ZSM-5, the difference b e t w een conversions of NO and NH 3 on H - f e r r i s i l i c a t e was smaller than that on Fe-ZSM-5, indicating the high sel ect i v i ty of H - f e r r i s i l i c a t e for the formation of nitrogen. In the case of FeOx/Sil, however, the NO conversion was apparently negative, which resulted from the

formation of NO through the r e a c t i o n b e t w e e n NH 3 and 0 2 [6]. Kinetic studies were carried out at 723 K with 0.01 g of H - f e r r i s i l i c a t e to know the r e a c t i o n order with r e s p e c t to the partial pressures of the r e a c t a n t s . The r a t e of N 2 formation was found to be ca. 0.8 order with NO, ca. 0.3 order with 0 2 and nearly zero order with NH 3. This suggests that the r a t e - d e t e r m i n ing step involves a r e a c ti o n b e t w e e n NO and atomic oxygen. 3.2. Adsorption of NO or NH 3 The kinetic result that the r a t e of N 2 formation was zero-order with the pressure of ammonia suggests that ammonia is strongly adsorbed on the surface

439

of H - f e r r i s i l i c a t e under the r e a c t i o n conditions. T h e r e f o r e , IR s p e c t r a of adsorbed NO w e r e t a k e n first w i t h o u t i n t r o duction of a m m o n i a to o b s e r v e weakly adsorbed species origin a t e d from NO. F i g u r e 1, a shows the IR d i f f e r e n c e spectrum obtained after H-ferrisili-

1810 1843 I 1868

1

(a)

c a t e was exposed to NO (20 Torr) at 298 K for 5 min. (b) Four absorption bands were o b s e r v e d at 1810, 1843, 1868 and 1915 cm-lo A f t e r 4 h of t h e NO exposure (b), 1810 and (c) 1915 cm-1 bands i n c r e a s e d in (d) t h e i r i n t e n s i t y , while 1843 and I I I I I 1868 cm -1 bands did not 2000 1800 1600 change significantly. There e x i s t e d a linear relationship W a v e n u m b e r / cm -1 b e t w e e n t h e i n t e n s i t i e s of the f o r m e r bands during prolonged Figure 1. IR s p e c t r a of H - f e r r i s i l i c a t e adsorption, which indicates a f t e r NO (20 Torr) a d s o r p t i o n at 298 K t h a t t h e s e bands are a t t r i b u t e d for 5 min (a) and 4 h (b) and subsequent to a s y m m e t r i c and s y m m e t r i c e v a c u a t i o n at 298 K for 10 min (c) and at s t r e t c h i n g modes of a dinitro573 K for 30 rain (d). syl species. We may assign t h e o t h e r two bands at 1843 and 1868 cm -1 to m o n o n i t r o s y l species. Similar bands have been r e p o r t e d [7] for the NO adsorbed on F e - Y z e o l i t e . When NO was p u m p e d out at 298 K for 10 min (c), t h e i n t e n s i t i e s of the bands at 1810 and 1915 cm -1 d e c r e a s e d and a new band at 1766 cm -1 appeared. The new band still r e m a i n e d a f t e r the e v a c u a t i o n at 573 K for 30 min (d), while the dinitrosyl bands t o t a l l y disappeared. The m o n o n i t r o s y l bands w e r e r e l a t i v e l y s t r o n g a f t e r the e v a c u a t i o n . The new band at 1766 cm -1 might be due to a m o n o n i t r o s y l species f o r m e d from t h e dinitrosyl species.

3.3. Adsorption of NO-NH 3 In order to know the r e a c t i o n i n t e r m e d i a t e for t h e NO-NH3-O 2 r e a c t i o n , t h e r e a c t i v i t i e s of the adsorbed species w e r e studied. A f t e r the adsorption of NO (20 Torr) at 298 K for 1 h (Fig. 2, a), 1 Torr of a m m o n i a was i n t r o d u c e d at 298 K. As shown in Fig. 2, b, 5 min of the a m m o n i a exposure gave bands at 1465, 1620, 1766, 1830 and 1930 cm-1. The 1465 and 1620 cm-1 bands are due to the adsorbed a m m o n i a on Brdnsted and Lewis acid sites, r e s p e c t i v e l y . The r a t i o of the i n t e n s i t i e s of 1930 to 1830 cm -1 bands was a l m o s t the s a m e as t h a t of t h e

440 two dinitrosyl bands (1915 to 1815 cm -1 bands). Therefore, it is indicated that the 1930 and 1830 cm-1 bands result from the shift of the dinitrosyl bands. This shift may be caused by the interaction of NH4 + ions with framework iron as an adsorption site for the dinitrosyl species. The subsequent evacuation at 298 K for 30 rain (c) eliminated the 1830 and 1930 cm -1 bands, increased the 1766 cm-I band intensity, and generated a band at 1843 c m - 1 These two bands were already assigned to the mononitrosyl species. These results suggest that the mono- and dinitrosyl species do not r e a c t with ammonia in the absence of oxygen.

3.4. Adsorption of NO-O2-NH 3

1810

1915

(a) ii 1830 1465 1766

~ .X

1620

O

(b)

~o~L. ~

.

.

w l

.

o

o o,

.

A .

.

i

_

o

o

1 ....

z

2

o

o

9

o )Ill

_

....

o

o

o

o

O

o

c-

.5 0

.9~

~

515

c--

co

., (1) .~--

~

~

(-.

o

~d

,.

X

N

.,

o-re"

N~.o

t_

I--

(D

O

~L_

E

(1)

O

~E

9

(-

"o

i

oJ E c(D c- (~ c((1) (D (-

O

~.~

O

E

E ~

~E

(D c- ~_ E "7 Or)

LL

516

Since, the linear correlations are established using all the experimental results stemming from NMR and IR spectroscopies, they lead to a good evaluation of N(AIf) for every sample. In Table 1, the number N(Aif) is the average value of N(AIf) obtained for each solid, from the two equations given in Figure1. In the case of highly dealuminated zeolites (samples 5, 5', 6, 6'), this number N(AIf) allows the determination of Si/Alf ratio with a better accuracy. Table 1 and Figure 2 contain the ratios calculated by this method. Curve 2a shows the correlation between Si/AIf ratio and the steaming temperature with the steamed HY zeolites studied in reference [4, 5]. The present data obtained for hexagonal faujasites (hS(T) series) are plotted (curve 2b). The two curves are nearly superimposable. At a given steaming temperature, the framework dealuminations for cubic and hexagonal faujasites are almost equal, therefore, the dealumination level of HEMT steamed at various temperatures can readily be deduced from curve 2a.

3.2. Br6nsted acidity studies After steaming, the number of zeolitic hydroxyls is less than that of framework AI atoms because the aluminic cations formed counterbalance partly the negative lattice charges. FTIR spectroscopy allows quantitative study of Br6nsted acid sites. The v(OH) massif of HY zeolites steamed at 860 and 1010 K studied here has already been described [2, 5, 10]. Spectra of HY zeolites steamed at 720 K and HEMT steamed at 720, 860 and 1010 K are shown in Figure 3. In addition to the (HF)OH and (LF)OH groups and silanols, the IR spectra presents a band at c.a. 3600 cm -1 and sometimes a shoulder at 3525 cm -1, as already reported for steamed HY zeolites [1, 5, 11]. Regarding the HEMT zeolites steamed at 860 and 1010 K (curves 4, 4', 5, 5') and the corresponding steamed HY [2, 10], acid leaching increases the v(OH) massif and slightly increases the SiOH band. As expected from their low crystallinity after leaching, the zeolites steamed at low temperature are less stable towards acid attack, therefore, the intensity of their v(OH) massif decreases after leaching, whereas the SiOH band increases (Fig. 3, curves a,a' and 2,2'). The v(OH) massif of steamed zeolites is complex and results from the superposition of bands due to framework and extraframework OH groups. Nevertheless, the acidic framework hydroxyls can be counted through the number of pyridinium species formed by pyridine adsorption at 470 K [1]. Figure 4A shows the difference spectrum (before adsorption minus after py adsorption) i.e. the v(OH) bands of zeolitic acidic hydroxyls perturbed by pyridine adsorption at 470 K, after a brief evacuation. It appears that the series of steamed HEMT (except sample 4') hardly displays the (3600)OH band due to strongly acidic hydroxyls. This band shown on the direct spectra (Fig. 3) is mainly due to the nonacidic extraframework hydroxyls [1, 2]. The same observation can be noted concerning the HY solids steamed at low temperature (curves 4A, a, a') but this result is quite different from that observed for HY zeolites steamed at 860 K. In this latter case, the (3600)OH bands is well-defined and its intensity is maximum for the leaching L5, [2]. Knowing the molar extinction coefficient (EpyH+ = 1.8 l~mo1-1 cm, [1]), the number of pyH + species formed can be determined from the intensity of the characteristic band at 1543 cm -1 shown in Figure 4B. The results are given in Table 2 for the two series of hexagonal and cubic steamed faujasites. The data are expressed in l~mol

517

Table 2- BrSnsted acidity of various HEMT and HY zeolites

Zeolites 0 u C

i.._ 0

npyH+ / l~mol g-1 470K hex. cub.

HEMT(3.8) 852

42

HY(2.9)

0 ~

~'"

3800

3700

"

3600

J

3500

I

Desorpt. % 620K hex. cub.

735

69

S(720)L5 S(720)

442* 364 291" 439

53 62

42 37

S(860)L5 S(860)

218 155

283 102

53 59

59 59

S(1010)L5 S(1010)

82 58

83 42

70 72

3400

Figure 3. v(OH) bands ; a, 2, 4, 5 samples 9 cS(720) and hS(720, 860 or 1010) ; a', 2', 4', 5' 9the corresponding leached solids.

*Only one steaming was applied

,

i;,

} v/=, i

Q ~J

\ "w

I ~

c~ b,,

o w

.Q

" ::.:.:,:.:,:,'.-:.:.rex-. ...::...:,:,:.,...::,e.::.:.,~.

.... : .............

::::::::::::::::::::::::::::::::::::.". ~":.::~

:..%'~!!~NI! Ni :i~i:ii:i::i:;::: :,: . . . .

. -m.,

,. ,:::.:.:.:.,-::.... :.:,..... ::::::::::::::::::::::::::::::::: .::.: ::::::: : ",'.:-:..-

0

80

I--.-

_z

6o

.d

.. rr"

Al

40

6Na LZ-133 2K4Na 3K3Na

2O

4K2Na 6K I--" T . C ' a

0

g

~

,

1

5O

1

9 .

1

15O

100

a

2~0

20O

TIME, HOURS

Figure 2. Cristallization kinetics of levyne-type zeolite from system x N a 2 0 yK20-6MeQI-A]203-30SiO2-500H20 where x+y=6 at 170~

g58

oO

o

~)0 158

100

n n

50

m

al

15ooc

A, 0T

0

,

I

93

,

,

,

40

I

60

,

I

80

170~ a

!

100

K/(Na+K) % Figure 3. Induction time as a function of K/(Na+K) % ratio in the reaction batch at the t e m p e r a t u r e s of ( I ) 150~ and ( k ) 170~ The addition of potassium to the reaction batch prevents the formation of LZ133 zeolite as parasite phase at both temperatures. The increase of potassium

554

amount in the batch, at both temperatures of reaction, produces an increase of the induction time as shown in the Figures 1 and 2. The course of the nucleation time variation as function of the amount of sodium and potassium in the batch composition is reported in Figure 3. The increase of the ratio K/(Na+K) in the reaction mixture, produces a linear variation of the nucleation time. The system containing just potassium (K/(Na+K)=I) does not follow this course because the absence of sodium modifies completely the rate of the induction period, The results shown in F i b r e s 1-3 suggest that the levyne-type /zeolite can be obtained from both systems containing just sodium (K/(Na+K)=0) or potassium (K/(Na+K)=I). The kinetic parameters are different because the sodium is more important in forming the single six rings secondary building units and the potassium favours their assembling but does not increase the rate of crystal growth. In fact, the slope of the kinetic curves is practically constant with a value of 5.4+__0.2 %/hour and a larger size of the levyne crystals is observed increasing the potassium content in the batch composition. In the system containing only sodium at 170~ the LZ-133 zeolite is not just a parasite phase but it is a zeolite that co-crystallises with levyne (6-8). The

E X 0

E N D 0

00 ....

9I

. . . .

100

|

. . . .

200

i.

300

,,,

It.l.,

400

,,

i . . . .

500

I

,

600

,

,

,

I

,

700

, , ,

I

800

TEMPERATURE, ~ Figure 4. DSC thermograms of levyne-type zeolite obtained from batches with different K/(Na+K) ratios.

555 presence of potassium at 170~ does not favour the growth of the LZ-133 zeolite because the assembling of the single six rings secondary building units is favoured by the presence of potassium. This result confirms the hypothesis that the LZ-133 zeolite is not a zeolite with the same structure as levyne, as previously proposed (6-8). The behaviour of the sodium and the potassium ions, the first-ones are able to produce the nuclei and the second-ones are able to increase the crystal growth, has been also observed in the synthesis of other high silica zeolites (10). The variation of the course of the DSC thermograms reported in Figure 4 shows that the interaction of the methylquinuclidine ion with the framework of levyne-type zeolite is influenced by the synthesis conditions. In fact, the thermal decomposition of the MeQ is observed at different temperatures for levyne obtained from sodium (K/(Na+K)=0) or potassium (K/(Na+K)=I) system (Table 1). In agreement with the results obtained with other organo-zeolites (11), the peak a is attributed to organic compound partially occluded in the zeolitic channels. The peak b is attributed to the MeQ + balanced by an OHanion and the peak c is due to the MeQ + linked with the zeolitic negative charge. The levyne obtained from the systems containing both sodium and potasium shows the same behaviour probably because the ratio Na+/K + in the framework is not a function of the amount of potassium in the initial batch. In this case we suppose that during the crystallisation, an ion exchange balances the ratio Na+/K +. This hypothesis is suggested by the analysis reported elsewhere (5) where independently of the amount of potassium added to the initial mixture of reaction, a higher amount of potassium is detected in the crystals of levyne-type zeolite. More analytical tests are in progress to verify this hypothesis. In any cases the mass spectrometer analysis of the volatile phases due to decomposition of MeQ, has not shown the presence of iodine. This suggests that the MeQ + neutralises some of the negative charges of the levyne framework. Table. 1 Data computed from the DSC and TG characterisation of levyne-type zeolite obtained from systems at different K/(Na+K) content. TEMPERATURE, ~ I! MeQ+u.c.* weight losses, % II batch peak b peak c total peakb peak c total peakb p e a k c 6Na 458.2 589.5 23.0 11.3 8.7 6.7 3.8 2.9 4Na2K 463.1 590.0 21.8 8.1 11.5 6.4 2.7 3.8 3Na3K 456.2 585.3 22.0 8.3 11.4 6.5 2.7 3.8 2Na4K 460.0 586.0 22.6 8.9 10.3 6.4 2.9 3.4 6K 439.8 584.3 22.6 8.7 11.3 6.6 2.9 3.8 ,,,

,,

* Computed on the basis of a framework of 54 T atoms. CONCLUSIONS The above reported results show that: - The levyne-type zeolite can be obtained from systems containing potassium but with a longer induction time; - The presence of the potassium in the batch composition gives the possibility to synthesize the levyne-type zeolite free of parasite phases at a temperature higher than 150~

556 - The systems containing sodium and potassium produce a levyne-type zeolite with a yield of 80%; - The MeQ + is counter-ion of the zeolite framework ACKNOWI,EDGEMENT Work supported by Italian Research Council, CNR, "Progetto Strategico Tecnologie Chimiche Innovative" RE~'~CI~ 1. G.T. Kerr, US Patent 3 459 676, (1969). 2. R. Millini, A. Carati and G. Bellussi, Zeolites, 12, 265, 1992, and references therein. 3. T. R. Cannan, M. T. L. Brent and E. M. Flanigen, Eur. Patent Appl. 0091048 A1, 1983. 4. M. T. L. Brent, T. R. Cannan and E. M. Flanigen, Eur. Patent Appl. 0091049 A1, 1983. 5. E. J. Rosinsky and M.K. Rubin, Eur Patent Appl. 0107370, (1983) 6. C.V. Tuoto, F. Testa, R. Aiello and A. Nastro, Materials Engineering, 2, (1994), 175, and references therein. 7. C.V. Tuoto, F. Testa, R. Aiello and A. Nastro, Atti 2 ~ Conf. AIMAT, (Ass. It. Ing. dei Materiali) P. Giordano Orsini Ed., vol.1, Universit~ di Trento, (1994), 173. 8. C.V. Tuoto, F. Testa and A. Nastro, Proc. 10th Int. Zeolite Conf., Garmisch Partenkishen, July 1994, Recent research report, in press. 9. C.V. Tuoto, A. Regina, J. B.Nagy and A. Nastro, Proc. III It. Nat. Conf. "Scienza e Tecnolog~a delle Zeoliti", Cetraro, Sept. 1995, in press. 10. R. Aiello, F. Crea, A. Nastro and C. Pellegrino, Zeolites, 7, 549, 1987. 11. F. Crea, J. B.Nagy, A. Nastro, G. Giordano and R. Aiello, Thermochimica Acta, 135, (1988), 353.

557

AUTHOR INDEX

An, L.-D. Auroux, A. Azuma, N. B alkus Jr., K.J. Bandyopadhyay, R. Bates, S.P. Baur, W.H. Bell, R.G. Benazzi, E. Ben Ta~rit, Y. Boddenberg, B. Bonneviot, L. Bordiga, S. Brunel, D. Brunner, E. Btilow, M. Buzzoni, R. Cairon, O. Cambon, H. Cardoso, D. Carrazza, J. Cartier dit Moulin, C. Cauvel, A. Catlow, C.R.A. (~ejka, J. Chambellan, A. Channon, Y.M. Chevreau, T. Chien, S.-H. Ciambelli, P. Clacens, J.-M. Clark, L. Corbo, P. Corma, A. Curtiss, L.A. Davis, M.E. De Luca, P. Descorme, C.

F-l-3 Th-2-4 F-l-1 W-l-I, P-34 F-l-2 M-2-6 Th- 1-4 PL-2 F-2-3 P-20 W-l-3 M- 1-3, P-42 KL-3 Tu-2-5, Th-2-5 M-l-1 Th-l-5 KL-3 P-31 Th-2-5 P-6 P-26 M-l-3 Tu-2-5 Tu- 1-3, PL-2 F-2-4 P-28 Tu-l-3 P-28, P-31 P-41 Th-2-2 Tu-2-1 F-l-4 Th-2-2 M-l-4 Tu-l-1 KL-1 P-15 Th-2-1

Diaz, A.C. DiRenzo, F. Djajanti, S. Downing, R. Dwyer, J. Eder, F. Eder-Mirth, G. Eic, M. Eissa, M. Ellestad, O.H. Fajula, F. Feng, Y. Fonseca, A. Fricke, R. Fyfe, C.A. Gabelica, Z. Gabrielov, A.G. Gale, J.D. Gambino, M. Garforth, A. G61in, P. Geneste, P. Geobaldo, F. Ghorbel, A. Giordano, G. Goldfarb, D. Gontier, S. Goursot, A. Grondey, H. Guisnet, M. Gunnewegh, E.A. Guo, C.J. Hartmann, M. Ho, J.-C. Howe, R.F. Huang, M. Huang, Y.

PL-1 Tu-2-5 W-l-2 P-16 M-2-6, F-1-4, P-7 P-27 M-2-5 Th-l-3 W-l-1 Tu-2-2 Tu- 1-2, Tu-2-5, KL-2 PL-1 M-2-3 P-21 PL-1 Tu-2-1, P-42 P-34 PL-2 Th-2-2 F- 1-4 Th-2-1, P-19 Th-2-5 KL-3 P-20 M-2-3 Tu-2-6 Tu-2-3 Tu-l-2 PL-1 F-2-3 P-16 Tu-2-4 F-l-1 P-41 W-l-2 Th-2-4 PL-1

558 Inui, T. Iton, L.E. Ivanova, I.

KL-4 Tu-l-1 M-l-4

Jackson, R.A. Jaeger, N.I. Jansen, J.C. Jorda, E. Joshi, P.N.

Tu-l-3 W-l-3 F-2-1 P-20 P-43

P-11 Kai, T. Th-2-4, P-42, P-43 Kaliaguine, S. P-42 Kapoor, M.P. M-2-3 Katovic, A. M-2-4, P-2 Kazansky, V.B. F-l-1 Kevan, L. P-28 Khabtou, S. W-l-1 Khanmamedova, A. P-25 Kinrade, S.D. P-25 Knight, C.T.G. PL-1 Kokotailo, G.T. P-14 Komatsu, T. Tu-l-4 Koningsberger, D.C. P-36 Konstantinov, L. Th-l-4 Komatowski, J. KL-5 Kumar, R. M-2-4, P-2 Kustov, L.M. P-15 Kuznicki, S. Lasp6ras, M. Lavalley, J.C. L6cuyer, C. Lemay, G. Le Noc, L. Lercher, J.A. Lessard, S. Lewis, A.R. Lewis, D.W. Liao, B. Loeffler, E. Lortie, C. Lujano, J. Lutz, W. Makarova, M.A.

Th-2-5 P-28 Th-2-1 P-43 M-l-3 M-2-5, P-27 M-l-3 PL-1 PL-2 Th-l-3 M-2-4, Th-2-6 M-l-3 P-26 Th-l-4, Th-2-6 M-2-6

Manoli, J.M. Mariey, L. Marzin, M. Micke, A. Migliardini, F. Miller, J.T. Minelli, G. Mintova, S. Monteiro, J.L.F. Moretti, G. Moudrakovski, I.L. Mueller, K.T. MUller, G. Murta Vane, M.L. Naccache, C. Nagy, J.B. Nakagawa, Y. Nastro, A. Nasution, M.N.A. Ng, F.T.T. Nkosi, B. Noronha, Z.M.

P-31 P-28 P-28 Th-l-5 Th-2-2 Tu-l-4 Th-2-2 P-36 P-19 Th-2-2 Th-l-2 PL-1 M-2-5 P-6 P-20 M-2-3, P-45 M-2-1, M-2-2 P-15, P-45 P-11 F-2-2 F-2-2 P-19

Occelli, M.L. Owens, S.L.

Th-l-3 Tu-l-3

Papai, I. Porta, P. Potvin, C. Predescu, L. Primet, M.

Tu-l-2 Th-2-2 P-31 P-29 Th-2-1

Rao, B.S. Ratcliffe, C.I. Ratnasamy, P. Rees, L.V. Rempel, G.L. Ribeiro, F.R. Ribeiro, M.F. Ricchiardi, G. Richter-Mendau, J. Ripmeester, J. Rodenburg, E.C. Rodriguez, I.

F-l-2 Th-l-2 KL-5 Th-l-1 F-2-2 F-2-3 F-2-3 KL-3 P-21 Th-l-2 F-2-1 Th-2-5

559 Roland, U. Romero, Y. Rozwadowski, M. Ruthven, D.M. Saint-Just, J. Salzer, R. Scarano, D. Schmidt, R. Schoeman, B. Schreier, E. Schwenn, H.-J. Sellem, S. Shaikh, R.A. Shen, D. Silva, J.M. Singh, P.S. Smekalina, E.V. Sobrinho, E.V. Sobry, R. Solomykina, S. Sooknoi, T. Sousa-Aguiar, E.F. Spasov, L. Spoto, G. Steinike, U. Stelmack, P. Stockenhuber, M. StOcker, M. Strobl, H. Su, B.L. Stimmchen, L. Syvitski, R.T. Takahashi, T. Tezel, F.H.

P-18 P-26 Th-l-4 Th- 1-3, PL-4

Trong On, D. Tuel, A. Tuoto, C.V. Tvarfi~&ov~i, Z.

M-l-3, P-42, P-43 Tu-2-3, P-20 P-45 F-2-4

Th-2-1 P-18 KL-3 Tu-2-2 P-36 P-21 W-l-3 P-31 F-l-2 Th-l-1 F-2-3 F-l-2 P-2 P-6 Tu-2-1 M-l-3 P-7 P-6 P-36 KL-3 P-21 P-29 P-27 Tu-2-2 PL-1 Th-2-3 P-18 P-25

Uddin, Md. A. Uvarova, E.B.

P-14 P-2

P-11 P-29

Valtchev, V. van Bekkum, H. van den Bossche, G. van der Puil, N. Vasilyev, V. Wang, H.-L. Wang, K.-J. Wark, M. Wamken, M. Wichterlov~i, B. Wong-Moon, K.C.

P-36 F-2-1. P-16 Tu-2-1 F-2-1 Tu-I-2 F-l-3 P-17 W-l-3 W-l-3 F-2-4 PL-1

Xiao, T.-C.

F-l-3

Yang, S.-M. Yashima, T.

P-17 P-14

Zecchina, A. Zhao, D. Zholobenko, V.L. Zibrowius, B. Zilkov~i, N. Zones, S.I. Zubowa, H.-L. Zygmunt, S.A. v

KL-3 Tu-2-6 M-2-4, F- 1-4 Th-2-6 F-2-4 M-2-1, P-34 P-21 Tu-I-1


561

S U B J E C T

Th-2-4, P-36

A - zeolite Alcohols - sorption of

M-2-5

Alkane oxidation

W - l - l , KL-5

ALPO4-5

Th-l-2, F-l-4, P-17

ALPO4-8

Th-l-2

ALPO4-11

Th-l-2, F-l-2, F-1-4, P-17

Anisole acylation of

P-16

I N D E X

Brr

acid sites

M-l-l, M-2-4, M-2-5, M-2-6, Tu-l-1, F- 1-4, P-31, PL-2, KL-2, KL-3, KL-4

Butene-1 - isomerisation

F-l-4

Butenes - oligomerisation

F-2-2

Butylbenzene acylation

P-16

CH4 - oxidation by NO x

P-2

As -

KL-5

zeolites

Basic zeolites

Th-2-4, Th-2-5, P-7

Beckmann rearrangement

C2C14 - adsorption of

M-l-1

Chloroaniline oxidation of

Tu-2-3

P-11 Clays

Benzene T-h- 1-4, Th-2-3 - adsorption of P-28 - hydroconversion of BETA

M-2-2, P-16, P-17

BF3 - adsorption of B

M-2-6

-

pillared and expanded

Tu- 1-3, P-29

Clinoptilolite Cloverite

Th-l-3

M-2-5, W- 1-2, P-21

Clusters of cesium oxide - models

Th-2-5 Tu-l-1

oron silicalite - MCM-.' 1

M-2-1 P-42 P-43

Borosilicate sieves

M-2-1

-

CO oxidation by NO x - adsorption of

COAPO-5

P-2 M-l-l, P-18, P-29 F- 1-4

562 COAPO- 11

F- 1-2

Co - ZSM-5

Th-2-2, KL-4

of binary mixtures - of n-paraffins in clays

3,7-Diazabicyclononane templating studies

CP*2CoOH as template Cracking

P-6

Crystallization kinetics and mechanism -

M

-

2

-

3

Cyclohexane adsorption of

Tu-2-4

Cyclohexene oxidation of

Tu-2-3

-

Dimethylacetal elimination of methanol

P-17

DSC

P-45

-

,

KL-1

DTA/DTG

EMT

P-21, P-36, KL-4

M-2-6, Th-2-3, P-31

Encapsulated complexes

-

E

Cyclohexanone oxime

P

R

,

E

S

R

P-11

Cu

NaA - ZSM-5

M-2-2

-

P-34

-

-

Th-1-1 Th-l-3

-

KL-4 Th-2-2, KL-4

W-l-1

Tu-2-6, F-I-I, F-l-2 P-20, P-41

Erionite acid sites in

M-2-4

-

ESEM

F-l-1

DAF-1

PL-2

ETS-10

PL-2

Dehydrogenation of n-heptane

F-2-1

EU-1

PL-2

Density Functional Theory (DFT) Tu-l-1, Tu- 1-2

EXAFS A1 K-edge Mo K-edge - W Liii-edge Ti K-edge

Tu-l-4 W-l-2 W-l-2 M-1-3, P-42

-

-

Design of synthesis - strategies for

KL-1

Diffusion coefficient of n-butane, n-hexane and butenes PL-2 - coefficient of water in MCM-41 Tu-2-2 coefficient of water in MCM-48 Tu-2-2 measurements of- coefficients PL-4 -

-

FAU

M-2-1, KL-2

Fe -

ZSM-5

-

-

zeolites

F-2-4, P- 11, P- 14, PL-2, KL-4, KL-5 KL-5

-

Frequency response (FR)

Th- 1-1, PL-4

563

Friedel - Crafts

P- 16

Isomerisation - of C 8 aromatic cuts

F-2-3

Ga - ZSM-5 Gravimetry

Hartree - Fock theory H-D exchange

P-11, KL-4

Knoevenagel condensation

Th-2-5

L - zeolite

Th-2-3

P-27

M-2-6, Tu-1-1 P-18, P-20

P-45, PL-2

Levyne type zeolites Lewis acids

M-l-l,

M-2-5,

M-2-6,

Th-2-4, KL-2 Heat of adsorption Host-guest interaction HREM

Th-2-4, P-27 PL-2, KL-3

Light alkanes - sorption of

P-27

Tu-2-2

Hydrogenation - competitive - of 1-heptene and 3-3 dimethyl-1butene F-2-1

MAPO- 11

F- 1-2

Mazzite

KL-2

MCM-41

Tu-2-1, Tu-2-2, Tu-2-5, W- 1-2, P-26

MCM-48

Tu-2-2, Tu-2-6, P-26

In -

ZSM-5

P-11

IR

-

-

-

-

-

diffuse reflectance M-2-4, P-2 framework P-42 of alkanes P-27 of ammonia M-2-5, KL-3 of benzene Th-2-3 of CD3CN KL-2 of CO P-19 of cyclohexene KL-2 of methanol F-1-3 of methylcyclopentene KL-2 of NO Th-2-1, P-14 of OH (and OD) M-2-6, P- 18, P-28, P-31, KL-3 of pyridine M-2-5, F-l-4, P-27, P-31, P-41, P-43, KL-2, KL-3 of pyrrole Th-2-4

MCM-L

Tu-2-1, P-26

MeAPO

F- 1-2, F- 1-4

Mesoporous materials Tu-2-1, Tu-2-2, Tu-2-3, Tu-2-4, Tu-2-5, Tu-2-6, P-26, P-43 Methane adsorption of P-29 Methanol to olefins (MTO)

F-l-3

MgAPO- 11

F- 1-4

Microcalorimetry

Th-2-4

Milling

Th-2-4

564

Mn - MCM-41 - MCM-48 - MCM-L

Tu-2-6 Tu-2-6 Tu-2-6

MOCVD

W-l-2

Mo(CO) 6

W-l-2

Modelling - of adsorption properties '

Mr

N

Tu-l-2, Tu-l-3

Plesset perturbation (MP2) M-2-6, Tu- 1-1

Mordenite acid sites in M-2-4 dealumination of F-2-3, P-19 - H-form P-19, P-27, KL-3 - Na Th-l-2, P-17

2

- adsorption of

Tu-l-2, Tu-2-4, Tu-2-5, Th- 1-4, Th-2-6, P-29

Ni - SAPO-5 and SAPO-11

F-l-1

NO x - adsorption complexes reduction with ammonia reduction with methane - reduction with propane - decomposition

P-2 P-14 Th-2-1 Th-2-2 KL-4

-

Nano-particles of ZnO, CdO and SnO 2 W-l-3

-

Mullite fibers

Nu-3

PL-2

Nucleation rate

M-2-1

P-36 02

- adsorption of NMR - 27A1 - liB -

13C

-

2 D

- 1H - 71Ga 31p

_

M-1-3, Tu-2-1, Th-2-1, P-6, P-19, P-41, PL-1, KL-2 P-42, P-43 M - l - l , M-l-4, Tu-2-5, F-l-3 PL-1 M - l - l , M-l-3, Tu-2-2 Tu-2-1 PL-1

- pulsed field gradient NMR PL-4 - 29Si M-1-3, M-2-3, Tu-2-1, Tu-2-2, Th-2-1, P-6, P-19, P-25, P-31, P-41, PL-1 ll7Sn, 119Sn W-l-3, P-25 - two dimensional Th-l-2, P-25 PL-1 - 129Xe Th-l-2, F-1-3

Tu-l-2 KL-5

Oxidation reactions with H202

Pd - exchanged zeolites Perfluorotributylamine

Th-2-1, F-1-3 Th-l-3

2-Phenylethanol hydrogenolysis -

P-7

Quasi-elastic neutron scattering (QENS) PL-4

-

Ruthenium perfluorophtalocyanines W-1-1

565 SAPO-5

F-l-l, F-l-3, P-17

Ti boralite cloverite MCM-41 MCM-48 - NCL-1 - 13 - UTD-1, UTD-8 ZSM-48

P-42 P-21 Tu-2-3 Tu-2-3 KL-5 Tu-2-3, KL-5 P-34 KL-5

-

SAPO-11

F-l-l, F-l-2, F-l-4, P-17

-

-

-

SAPO-34

P-17

SAXS

Tu-2-1

-

Tu-2-4, F- 1-2, F-2-1, P-21, P-34, P-36, P-41

SEM

Toluene acylation P-16 - alkylation M-l-4, F-2-4, P-7 disproportionation F-2-4 -

Silanes- alkoxy grafting

Tu-2-5

-

Silicalite- 1

M-l-3, Th-l-1, Th-l-4, F-2-1, P-36, P-42, PL-4

Silicate speciation

P-25

-

TPD F- 1-3, F- 1-4, P- 11 Tu-2-1 - of carbon dioxide Th-2-5 of methanol M-2-5

-

of ammonia

-

Sn -

zeolites

KL-5

Sorption kinetics

Th- 1-5

Spillover

F-2-3

Triisopropylbenzene - cracking of

P-6

P- 18

SSZ-24

M-2-2, Th- 1-2

SSZ-31

M-2-2

SSZ-35

M-2-2

SSZ-37

M-2-2

Styrene hydrogenation of -

TEM

Transalkylation

TS- 1

M- 1-3, Tu-2-3, P-20, P-42, KL-5

TS-2

KL-5

UV irradiation of TS-1

P-20

UV-visible - diffuse reflectance P-7

Tu-2-5, W- 1-3, Th-2-2, F- 1-2, P-34, P-42

Tu-2-4, Tu-2-6, Th-2-6, F-2-1 BETA MCM-41 - NCL-1 - VAPO- 11 -

Tetralin - adsorption of

-

Tu-2-4

P-41, KL-5 Tu-2-3 KL-5 F-l-2

566 - ZSM-12

KL-5

VPI-5

Th- 1-2, PL- 1

VS- 1

Tu-2-3, KL-5

VS-2

- CsNa - dealuminated -

H

-

L a

-

N a

KL-5 -

Water - adsorption of

Tu-l-1, Tu- 1-3, Tu-2-4, Th-l-4, P-18

W(CO)6

W-l-2

-

Th-2-4, Th-2-5 Th-2-6, P-6, P-28, P-31, Tu-l-4 Tu-l-4 Tu-l-4, W-l-2, Th-l-2, Th-2-3, P-17, P-18, PL-4

NiNa - alkaline-earth PtNa

F-2-2 P-18

Zero length column (ZLC)

Th-l-3, Th- 1-5, PL-4

M-2-2, Tu-l-1, F-2-1, F-2-4, P-17, P-27, PL- 1, PL-2, KL- 1, KL-2, KL-3 M-2-4 acid sites in Th-l-4, F-2-3 - dealuminated P-11, KL-1 high silica

ZSM-5

X acid sites in - alkali exchanged -

- CsNa

M-2-4 W- 1-1, Th-l-2, Th-2-3, Th-2-4, PL-4 Th-2-4, Th-2-5, P-7

-

-

XPS Xylenes disproportionation

Th-2-2, Th-2-4

Y - acid sites in - alkali exchanged Ca -

ZSM-11

M-l-4, PL-1, PL-2

ZSM-12

Th-l-2, PL-1

ZSM-18

PL-2

ZSM-48

M-2-3

F-2-3

M-2-4 Th-2-4 Tu-l-4

567

STUDIES IN SURFACE SCIENCE AND CATALYSIS Advisory Editors: B. Delmon, Universite Catholique de Louvain, Louvain-la-Neuve, Belgium J.T. Yates, University of Pittsburgh, Pittsburgh, PA, U.S.A.

Volume

1

Volume 2

Volume 3

Volume 4

Volume 5

Volume 6

Volume 7

Volume 8 Volume 9

Volume 10

Volume 11

Volume 12

Volume 13 Volume 14

Preparation of Catalysts I.Scientific Bases for the Preparation of Heterogeneous Catalysts. Proceedings of the First International Symposium, Brussels, October 14-17,1975 edited by B. Delmon, P.A.Jacobs and G. Poncelet The Control of the Reactivity of Solids. A Critical Survey of the Factors that Influence the Reactivity of Solids, with Special Emphasis on the Control of the Chemical Processes in Relation to Practical Applications by V.V. Boldyrev, M. Bulens and B. Delmon Preparation of Catalysts I1. Scientific Bases for the Preparation of Heterogeneous Catalysts. Proceedings of the Second International Symposium, Louvain-la-Neuve, September 4-7, 1978 edited by B. Delmon, P. Grange, P.Jacobs and G. Poncelet Growth and Properties of Metal Clusters. Applications to Catalysis and the Photographic Process. Proceedings ofthe 32nd International Meeting ofthe Societe de Chirnie Physique, Villeurbanne, September 24-28, 1979 edited by J. Bourdon Catalysis by Zeolites. Proceedings of an International Symposium, Ecully (Lyon), September 9-11, 1980 edited by B. Imelik, C. Naccache, Y. Ben Taarit, J.C. Vedrine, G. Coudurier and H. Praliaud Catalyst Deactivation. Proceedings of an International Symposium, Antwerp, October 13-15,1980 edited by B. Delmon and G.F. Froment New Horizons in Catalysis. Proceedings of the 7th International Congress on Catalysis, Tokyo, June 30-July4, 1980. Parts A and B edited by T. Seiyama and K. Tanabe Catalysis by Supported Complexes by Yu.l. Yermakov, B.N. Kuznetsov and V.A. Zakharov Physics of Solid Surfaces. Proceedings of a Symposium, Bechyhe, September 29-October 3,1980 edited by M. Lazni~ka Adsorption at the Gas-Solid and Liquid-Solid Interface. Proceedings of an International Symposium, Aix-en-Provence, September 21-23, 1981 edited by J. Rouquerol and K.S.W. Sing Metal-Support andMetaI-Additive Effects in Catalysis. Proceedings of an International Symposium, Ecully (Lyon), September 14-16, 1982 edited by B. Imelik, C. Naccache, G. Coudurier, H. Praliaud, P. Meriaudeau, P. Gallezot, G.A. Martin and J.C. Vedrine Metal Microstructures in Zeolites. Preparation - Properties- Applications. Proceedings of a Workshop, Bremen, September 22-24, 1982 edited by P.A. Jacobs, N.I. Jaeger, P.Jin3 and G. Schulz-Ekloff Adsorption on Metal Surfaces. An Integrated Approach edited by J. Benard Vibrations at Surfaces. Proceedings of the Third International Conference, Asilomar, CA, September 1-4, 1982 edited by C.R. Brundle and H. Morawitz

568 Volume 15 Volume 16

Volume 17

Volume 18

Volume 19

Volume 20

Volume 21

Volume 22 Volume 23 Volume 24

Volume 25

Volume 26

Volume 27 Volume 28

Volume 29 Volume 30

Volume 31


Heterogeneous Catalytic Reactions Involving Molecular Oxygen by G.I. Golodets Preparation of Catalysts III. Scientific Bases for the Preparation of Heterogeneous Catalysts. Proceedings of the Third International Symposium, Louvain-la-Neuve, September 6-9, 1982 edited by G. Poncelet, P.Grange and P.A. Jacobs Spillover of Adsorbed Species. Proceedings of an International Symposium, Lyon-Villeurbanne, September 12-16, 1983 edited by G.M. Pajonk, S.J. Teichner and J.E. Germain Structure and Reactivity of Modified Zeolites. Proceedings of an International Conference, Prague, July 9-13, 1984 edited by RA. Jacobs, N.I. Jaeger, R Jin3, V.B. Kazansky and G. Schulz-Ekloff Catalysis on the Energy Scene. Proceedings of the 9th Canadian Symposium on Catalysis, Quebec, P.Q., September 30-October 3, 1984 edited by S. Kaliaguine and A. Mahay Catalysis by Acids and Bases. Proceedings of an International Symposium, Villeurbanne (Lyon), September 25-27, 1984 edited by B. Imelik, C. Naccache, G. Coudurier, Y. Ben Taarit and J.C. Vedrine Adsorption and Catalysis on Oxide Surfaces. Proceedings of a Symposium, Uxbridge, June 28-29, 1984 edited by M. Che and G.C. Bond Unsteady Processes in Catalytic Reactors by Yu.Sh. Matros Physics of Solid Surfaces 1984 edited by J. Koukal Zeolites: Synthesis, Structure, Technology and Application. Proceedings of an International Symposium, Portoro~-Portorose, September 3-8, 1984 edited by B. Dr~aj, S. Ho~evar and S. Pejovnik Catalytic Polymerization of Olefins. Proceedings of the International Symposium on Future Aspects of Olefin Polymerization, Tokyo, July 4-6, 1985 edited by T. Keii and K. Soga Vibrations at Surfaces 1985. Proceedings of the Fourth International Conference, Bowness-on-Windermere, September 15-19, 1985 edited by D.A. King, N.V. Richardson and S. Holloway Catalytic Hydrogenation edited by L. Cerven# New Developments in Zeolite Science and Technology. Proceedings of the 7th International Zeolite Conference, Tokyo, August 17-22, 1986 edited by Y. Murakami, A. lijima and J.W. Ward Metal Clusters in Catalysis edited by B.C. Gates, L. Guczi and H. Kn6zinger Catalysis and Automotive Pollution Control. Proceedings of the First International Symposium, Brussels, September 8-11, 1986 edited by A. Crucq and A. Frennet Preparation of Catalysts IV. Scientific Bases for the Preparation of Heterogeneous Catalysts. Proceedings of the Fourth International Symposium, Louvain-la-Neuve, September 1-4, 1986 edited by B. Delmon, P. Grange, P.A. Jacobs and G. Poncelet Thin Metal Films and Gas Chemisorption edited by P. Wissmann Synthesis of High-silica Aluminosilicate Zeolites edited by RA. Jacobs and J.A. Martens Catalyst Deactivation 1987. Proceedings of the 4th International Symposium, Antwerp, September 29-October 1, 1987 edited by B. Delmon and G.F. Froment

569 Volume 35 Volume 36

Volume 37

Volume 38

Volume 39

Volume 40

Volume 41


Volume 45 Volume 46

Volume 47 Volume 48

Volume 49

Volume 50

Volume 51 Volume 52

Volume 53

Keynotes in Energy-Related Catalysis edited by S. Kaliaguine Methane Conversion. Proceedings of a Symposium on the Production of Fuels and Chemicals from Natural Gas, Auckland, April 27-30, 1987 edited by D.M. Bibby, C.D. Chang, R.F. Howe and S. Yurchak Innovation in Zeolite Materials Science. Proceedings of an International Symposium, Nieuwpoort, September 13-17, 1987 edited by P.J. Grobet, W.J. Mortier, E.F.Vansant and G. Schulz-Ekloff Catalysis 1987. Proceedings ofthe 10th North American Meeting ofthe Catalysis Society, San Diego, CA, May 17-22, 1987 edited by J.W. Ward Characterization of Porous Solids. Proceedings of the IUPAC Symposium (COPS I), Bad Soden a. Ts., April 26-29,1987 edited by K.K. Unger, J. Rouquerol, K.S.W. Sing and H. Kral Physics of Solid Surfaces 1987. Proceedings of the Fourth Symposium on Surface Physics, Bechyne Castle, September 7-11, 1987 edited by J. Koukal Heterogeneous Catalysis and Fine Chemicals. Proceedings of an International Symposium, Poitiers, March 15-17, 1988 edited by M. Guisnet, J. Barrault, C. Bouchoule, D. Duprez, C. Montassier and G. Perot Laboratory Studies of Heterogeneous Catalytic Processes by E.G. Christoffel, revised and edited by Z. Paal Catalytic Processes under Unsteady-State Conditions by Yu. Sh. Matros Successful Design of Catalysts. Future Requirements and Development. Proceedings ofthe Worldwide Catalysis Seminars, July, 1988, on the Occasion of the 30th Anniversary of the Catalysis Society of Japan edited by T. Inui Transition Metal Oxides. Surface Chemistry and Catalysis byH.H. Kung Zeolites as Catalysts, Sorbents and Detergent Builders. Applications and Innovations. Proceedings of an International Symposium, W~rzburg, September 4-8,1988 edited by H.G. Karge and J. Weitkamp Photochemistry on Solid Surfaces edited by M. Anpo and T. Matsuura Structure and Reactivity of Surfaces. Proceedings of a European Conference, Trieste, September 13-16, 1988 edited by C. Morterra, A. Zecchina and G. Costa Zeolites: Facts, Figures, Future. Proceedings of the 8th International Zeolite Conference, Amsterdam, July 10-14, 1989. Parts A and B edited by P.A. Jacobs and R.A. van Santen Hydrotreating Catalysts. Preparation, Characterization and Performance. Proceedings ofthe Annual International AIChE Meeting, Washington, DC, November 27-December 2, 1988 edited by M;L. Occelli and R.G. Anthony New Solid Acids and Bases. Their Catalytic Properties by K. Tanabe, M. Misono, Y. Ono and H. Hattori Recent Advances in Zeolite Science. Proceedings of the 1989 Meeting of the British Zeolite Association, Cambridge, April 17-19, 1989 edited by J. Klinowsky and P.J. Barrie Catalyst in Petroleum Refining 1989. Proceedings of the First International Conference on Catalysts in Petroleum Refining, Kuwait, March 5-8, 1989 edited by D.L. Trimm, S. Akashah, M. Absi-Halabi and A. Bishara

570

Future Opportunities in Catalytic and Separation Technology edited by M. Misono, Y. Moro-oka and S. Kimura New Developments in Selective Oxidation. Proceedings of an International Volume 55 Symposium, Rimini, Italy, September 18-22, 1989 edited by G. Centi and F.Trifiro Olefin Polymerization Catalysts. Proceedings of the International Symposium Volume 56 on Recent Developments in Olefin Polymerization Catalysts, Tokyo, October 23-25, 1989 edited by T. Keii and K. Soga Volume 57A Spectroscopic Analysis of Heterogeneous Catalysts. Part A: Methods of Surface Analysis edited by J.L.G. Fierro Volume 57B Spectroscopic Analysis of Heterogeneous Catalysts. Part B: Chemisorption of Probe Molecules edited by JoL.G. Fierro Introduction to Zeolite Science and Practice Volume 58 edited by H. van Bekkum, E.M. Flanigen and J.C. Jansen Heterogeneous Catalysis and Fine Chemicals II. Proceedings of the 2nd Volume 59 International Symposium, Poitiers, October 2-6, 1990 edited by M. Guisnet, J. Barrault, C. Bouchoule, D. Duprez, G. Perot, R. Maurel and C. Montassier Chemistry of Microporous Crystals. Proceedings of the International Symposium Volume 60 on Chemistry of Microporous Crystals, Tokyo, June 26-29, 1990 edited by T. Inui, S. Namba and T. Tatsumi Natural Gas Conversion. Proceedings of the Symposium on Natural Gas Volume 61 Conversion, Oslo, August 12-17, 1990 edited by A. Holmen, K.-J. Jens and S. Kolboe Characterization of Porous Solids II. Proceedings of the IUPAC Symposium Volume 62 (COPS II), Alicante, May 6-9, 1990 edited by F. Rodriguez-Reinoso, J. Rouquerol, K.S.W. Sing and K.K. Unger Preparation of Catalysts V. Scientific Bases for the Preparation of Heterogeneous Volume 63 Catalysts. Proceedings of the Fifth International Symposium, Louvain-la-Neuve, September 3-6, 1990 edited by G. Poncelet, P.A. Jacobs, R Grange and B. Delmon New Trends in CO Activation Volume 64 edited by L. Guczi Catalysis and Adsorption by Zeolites. Proceedings of ZEOCAT 90, Leipzig, Volume 65 August 20-23, 1990 edited by G. (~hlmann, H. Pfeifer and R. Fricke Dioxygen Activation and Homogeneous Catalytic Oxidation. Proceedings of the Volume 66 Fourth International Symposium on Dioxygen Activation and Homogeneous Catalytic Oxidation, Balatonf~red, September 10-14, 1990 edited by L.I. Simandi Structure-Activity and Selectivity Relationships in Heterogeneous Catalysis. Volume 67 Proceedings of the ACS Symposium on Structure-Activity Relationships in Heterogeneous Catalysis, Boston, MA, April 22-27, 1990 edited by R.K. Grasselli and A.W. Sleight Catalyst Deactivation 1991. Proceedings of the Fifth International Symposium, Volume 68 Evanston, IL, June 24-26, 1991 edited by C.H. Bartholomew and J.B. Butt Zeolite Chemistry and Catalysis. Proceedings of an International Symposium, Volume 69 Prague, Czechoslovakia, September 8-13, 1991 edited by RA. Jacobs, N.I. Jaeger, L. Kubelkova and B. Wichterlova Volume 54

571 Volume 70

Volume 71

Volume 72

Volume 73

Volume 74 Volume 75

Volume 76 Volume 77

Volume78

Volume79

Volume80

Volume81

Volume82

Volume 83

Volume84

Volume 85

Poisoning and Promotion in Catalysis based on Surface Science Concepts and Experiments by M. Kiskinova Catalysis and Automotive Pollution Control II. Proceedings of the 2nd International Symposium (CAPoC 2), Brussels, Belgium, September 10-13, 1990 edited by A. Crucq New Developments in Selective Oxidation by Heterogeneous Catalysis. Proceedings of the 3rd European Workshop Meeting on New Developments in Selective Oxidation by Heterogeneous Catalysis, Louvain-la-Neuve, Belgium, April 8-10, 1991 edited by P. Ruiz and B. Delmon Progress in Catalysis. Proceedings of the 12th Canadian Symposium on Catalysis, Banff, Alberta, Canada, May 25-28, 1992 edited by K.J. Smith and E.C. Sanford Angle-Resolved Photoemission. Theory and Current Applications edited by S.D. Kevan New Frontiers in Catalysis, Parts A-C. Proceedings of the 10th International Congress on Catalysis, Budapest, Hungary, 19-24 July, 1992 edited by L. Guczi, F. Solymosi and P.Tetenyi Fluid Catalytic Cracking: Science and Technology edited by J.S. Magee and M.M. Mitchell, Jr. New Aspects of Spillover Effect in Catalysis. For Development of Highly Active Catalysts. Proceedings ofthe Third International Conference on Spillover, Kyoto, Japan, August 17-20, 1993 edited by T. Inui, K. Fujimoto, T. Uchijima and M. Masai Heterogeneous Catalysis and Fine Chemicals III. Proceedings ofthe 3rd International Symposium, Poitiers, April 5-8, 1993 edited by M. Guisnet, J. Barbier, J. Barrault, C. Bouchoule, D. Duprez, G. Perot and C. Montassier Catalysis: An Integrated Approach to Homogeneous, Heterogeneous and Industrial Catalysis edited by J.A. Moulijn, P.W.N.M. van Leeuwen and R.A. van Santen Fundamentals of Adsorption. Proceedings of the Fourth International Conference on Fundamentals of Adsorption, Kyoto, Japan, May 17-22, 1992 edited by M. Suzuki Natural Gas Conversion II. Proceedings ofthe Third Natural Gas Conversion Symposium, Sydney, July 4-9, 1993 edited by H.E. Curry-Hyde and R.F. Howe New Developments in Selective Oxidation II. Proceedings of the Second World Congress and Fourth European Workshop Meeting, Benalm~dena, Spain, September 20-24, 1993 edited by V. Cortes Corberan and S. Vic Bellon Zeolites and Microporous Crystals. Proceedings of the International Symposium on Zeolites and Microporous Crystals, Nagoya, Japan, August 22-25, 1993 edited by T. Hattori and T. Yashima Zeolites and Related Microporous Materials: State of the Art 1994. Proceedings ofthe 10th International Zeolite Conference, Garmisch-Partenkirchen, Germany, July 17-22, 1994 edited by J. Weitkamp, H.G. Karge, H. Pfeifer and W. H61derich Advanced Zeolite Science and Applications edited by J.C. Jansen, M. St6cker, H.G. Karge and J.Weitkamp

572 Volume86 Volume 87

Volume88

Volume 89

Volume90

Volume91

Volume92

Volume 93 Volume94

Volume 95 Volume96

Volume97

Oscillating Heterogeneous Catalytic Systems by M.M. Slin'ko and N.I. Jaeger Characterization of Porous Solids III. Proceedings of the IUPAC Symposium (COPS III), Marseille, France, May 9-12, 1993 edited by J.Rouquerol, F. Rodriguez-Reinoso, K.S.W. Sing and K.K. Unger Catalyst Deactivation 1994. Proceedings of the 6th International Symposium, Ostend, Belgium, October 3-5, 1994 edited by B. Delmon and G.F. Froment Catalyst Design for Tailor-made Polyolefins. Proceedings of the International Symposium on Catalyst Design for Tailor-made Polyolefins, Kanazawa, Japan, March 10-12, 1994 edited by K. Soga and M. Terano Acid-Base Catalysis II. Proceedings of the International Symposium on Acid-Base Catalysis II, Sapporo, Japan, December 2-4, 1993 edited by H. Hattori, M. Misono and Y. Ono Preparation of Catalysts VI. Scientific Bases for the Preparation of Heterogeneous Catalysts. Proceedings of the Sixth International Symposium, Louvain-La-Neuve, September 5-8, 1994 edited by G. Poncelet, J. Martens, B. Delmon, P.A. Jacobs and P.Grange Science and Technology in Catalysis 1994. Proceedings of the Second Tokyo Conference on Advanced Catalytic Science and Technology, Tokyo, August 21-26, 1994 edited by Y. Izumi, H. Arai and M. Iwamoto Characterization and Chemical Modification of the Silica Surface by E.F.Vansant, R Van Der Voort and K.C. Vrancken Catalysis by Microporous Materials. Proceedings of 7EOCAT'95, Szornbathely, Hungary, July 9-13, 1995 edited by H.K. Beyer, H.G.Karge, I. Kiricsi and J.B. Nagy Catalysis by Metals and Alloys by V. Ponec and G.C. Bond Catalysis and Automotive Pollution III. Proceedings of the Third International Symposium (CAPoC 3), Brussels, Belgium, April 20-22, 1994 edited by A. Frennet and J.-M. Bastin Zeolites: A Refined Tool for Designing Catalytic Sites. Proceedings of the International Zeolite Symposium, Quebec, Canada, October 15-20, 1995 edited by L. Bonneviot and S. Kaliaguine

Zeolites: A Refined Tool for Designing Catalytic Sites

Zeolites for Cleaner Technologies (Catalytic Science Series, 3)

Groundwater Geophysics: A Tool for Hydrogeology

Wavelets: A Mathematical Tool for Signal Analysis

Wavelets: a mathematical tool for signal processing

Like Gold Refined

Clifford algebra: a computational tool for physicists

Clifford algebra: a computational tool for physicists

Imaging spectrometry, a tool for environmental observations

Clifford Algebra: A Computational Tool for Physicists

Wavelets: A Mathematical Tool for Signal Analysis

Molecular Sieve Zeolites II

Designing for Emotion

Designing Sound for Animation

Designing Sound for Animation

Designing for Emotion

Designing Sound for Animation

Designing Sound for Animation

Designing for XOOPS

Designing for Emotion

Designing for XOOPS

Catalytic Antibodies

Designing a Market Basket for NAEP

Designing a Market Basket for NAEP

Designing Language Courses: A Guide for Teachers

Catalytic Heterofunctionalization

Catalytic Kinetics

Convert!: Designing Web Sites to Increase Traffic and Conversion

Convert!: Designing Web Sites to Increase Traffic and Conversion

Ict for Social Welfare: A Tool Kit for Managers

Zeolites: A Refined Tool for Designing Catalytic Sites

Zeolites for Cleaner Technologies (Catalytic Science Series, 3)

Groundwater Geophysics: A Tool for Hydrogeology

Wavelets: A Mathematical Tool for Signal Analysis

Wavelets: a mathematical tool for signal processing

Like Gold Refined

Clifford algebra: a computational tool for physicists

Clifford algebra: a computational tool for physicists

Imaging spectrometry, a tool for environmental observations

Clifford Algebra: A Computational Tool for Physicists

Wavelets: A Mathematical Tool for Signal Analysis

Molecular Sieve Zeolites II

Designing for Emotion

Designing Sound for Animation

Designing Sound for Animation

Designing for Emotion

Designing Sound for Animation

Designing Sound for Animation

Designing for XOOPS

Designing for Emotion

Designing for XOOPS

Catalytic Antibodies

Designing a Market Basket for NAEP

Designing a Market Basket for NAEP

Designing Language Courses: A Guide for Teachers

Catalytic Heterofunctionalization

Catalytic Kinetics

Convert!: Designing Web Sites to Increase Traffic and Conversion

Convert!: Designing Web Sites to Increase Traffic and Conversion

Ict for Social Welfare: A Tool Kit for Managers

Recommend Documents