Advances in Catalysis, Volume 48

ADVANCES IN CATALYSIS VOLUME 48 Advisory Board M. CHE D.D. ELEY G. ERTL Paris, France Nottingham, England Berlin...

Author: H. Knözinger | Bruce C. Gates

85 downloads 1289 Views 7MB 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

ADVANCES IN CATALYSIS VOLUME 48

Advisory Board M. CHE

D.D. ELEY

G. ERTL

Paris, France

Nottingham, England

Berlin/Dahlem, Germany

V.B. KAZANSKY

W.M.H. SACHTLER

Moscow, Russia

Evanston, Illinois, USA

R.A.

VAN

SANTEN

Eindhoven, The Netherlands

K. TAMARU

J.M. THOMAS

Tokyo, Japan

London/Cambridge, England

H. TOPSØE

P.B. WEISZ

Lyngby, Denmark

State College, Pennsylvania, USA

ADVANCES IN CATALYSIS VOLUME 48

Edited by BRUCE C. GATES

HELMUT KNO¨ ZINGER

University of California Davis, California, USA

University of Munich Munich, Germany

Amsterdam · Boston · Heidelberg · London · New York · Oxford Paris · San Diego · San Francisco · Singapore · Sydney · Tokyo

ELSEVIER B.V. Sara Burgerhartstraat 25 P.O. Box 211, 1000 AE Amsterdam, The Netherlands

ELSEVIER Inc. 525 B Street, Suite 1900 San Diego, CA 92101-4495 USA

ELSEVIER Ltd The Boulevard, Langford Lane Kidlington, Oxford OX5 1GB UK

ELSEVIER Ltd 84 Theobalds Road London WC1X 8RR UK

q 2004 Elsevier Inc. All rights reserved. This work is protected under copyright by Elsevier Inc., and the following terms and conditions apply to its use: Photocopying Single photocopies of single chapters may be made for personal use as allowed by national copyright laws. Permission of the Publisher and payment of a fee is required for all other photocopying, including multiple or systematic copying, copying for advertising or promotional purposes, resale, and all forms of document delivery. Special rates are available for educational institutions that wish to make photocopies for non-profit educational classroom use. Permissions may be sought directly from Elsevier’s Rights Department in Oxford, UK: phone (+44) 1865 843830, fax (+44) 1865 853333, e-mail: [email protected]. Requests may also be completed on-line via the Elsevier homepage (http://www.elsevier.com/locate/permissions). In the USA, users may clear permissions and make payments through the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA; phone: (+1) (978) 7508400, fax: (+1) (978) 7504744, and in the UK through the Copyright Licensing Agency Rapid Clearance Service (CLARCS), 90 Tottenham Court Road, London W1P 0LP, UK; phone: (+44) 20 7631 5555; fax: (+44) 20 7631 5500. Other countries may have a local reprographic rights agency for payments. Derivative Works Tables of contents may be reproduced for internal circulation, but permission of the Publisher is required for external resale or distribution of such material. Permission of the Publisher is required for all other derivative works, including compilations and translations. Electronic Storage or Usage Permission of the Publisher is required to store or use electronically any material contained in this work, including any chapter or part of a chapter. Except as outlined above, no part of this work may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior written permission of the Publisher. Address permissions requests to: Elsevier’s Rights Department, at the fax and e-mail addresses noted above. Notice 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. Because of rapid advances in the medical sciences, in particular, independent verification of diagnoses and drug dosages should be made. First edition 2004 Library of Congress Cataloging in Publication Data A catalog record is available from the Library of Congress. British Library Cataloguing in Publication Data A catalogue record is available from the British Library.

ISBN: 0-12-007848-1 ISSN (Series): 0360-0564 1 The paper used in this publication meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper). Printed in W The Netherlands.

Contents CONTRIBUTORS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . PREFACE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ROBERT L. BURWELL , Jr. (1912 – 2003) . . . . . . . . . . . . . . . . . . . . . . . . .

xiii xv xix

Active Sites and Reactive Intermediates in Titanium Silicate Molecular Sieves P. Ratnasamy, D. Srinivas and H. Kno¨zinger I. Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . II. Active Sites. . . . . . . . . . . . . . . . . . . . . . . . . . . . . II.A. State and Framework Coordination of Ti . . . II.A.1. Diffraction Techniques . . . . . . . . . . II.A.2. Influence of Particle Size . . . . . . . . . II.A.3. UV – Visible Spectroscopy . . . . . . . . II.A.4. Photoluminescence Spectroscopy . . . II.A.5. X-Ray Absorption Spectroscopy . . . II.A.6. Vibrational Spectroscopy . . . . . . . . . II.A.7. EPR Spectroscopy . . . . . . . . . . . . . . II.B. Surface Acidity . . . . . . . . . . . . . . . . . . . . . . II.B.1. Brønsted Acid Sites . . . . . . . . . . . . . II.B.2. Lewis Acid Sites and Expansion of Coordination Sphere . . . . . . . . . . . . III. Oxo-Titanium Species and Reactive Intermediates III.A. UV –Visible Spectroscopy . . . . . . . . . . . . . III.B. Vibrational Spectroscopy . . . . . . . . . . . . . . III.C. X-Ray Absorption Spectroscopy . . . . . . . . . III.D. Cyclic Voltametry . . . . . . . . . . . . . . . . . . . III.E. EPR Spectroscopy . . . . . . . . . . . . . . . . . . . IV. Computational Investigations . . . . . . . . . . . . . . . . V. Catalytic Properties . . . . . . . . . . . . . . . . . . . . . . . V.A. Introduction . . . . . . . . . . . . . . . . . . . . . . . . V.B. Reactions Using H2O2 as Oxidant . . . . . . . . V.B.1. General Features . . . . . . . . . . . . . . . v

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

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

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

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

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

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

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

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

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

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

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

5 9 9 10 12 12 15 15 18 22 26 26

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

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

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

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

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

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

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

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

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

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

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

28 33 34 34 39 41 42 49 55 55 56 56

vi

Contents

V.B.2. H2O2-Catalyzed Reactions in the Homogeneous Phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.C. Epoxidation on Titanium Silicate Molecular Sieves . . . . . . . . . V.C.1. General Features of Epoxidations. . . . . . . . . . . . . . . . . V.C.2. Yields and Stereospecificities. . . . . . . . . . . . . . . . . . . . V.C.3. Diffusional Constraints . . . . . . . . . . . . . . . . . . . . . . . . V.C.4. Influence of Ti-Silicate Structure . . . . . . . . . . . . . . . . . V.C.5. Epoxidation Catalyzed by Mesoporous Titanium Silicates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.C.6. Influence of Alkene Structure . . . . . . . . . . . . . . . . . . . V.C.7. Dialkenes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.C.8. Epoxidation in the Presence of Other Oxidizable Functional Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.C.9. Hydroxyl-Assisted Epoxidation . . . . . . . . . . . . . . . . . . V.C.10. Diastereoselectivity in Epoxidations . . . . . . . . . . . . . . V.C.11. Side Reactions During Epoxidation . . . . . . . . . . . . . . V.C.12. Influence of pH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.C.13. Epoxidation with Alkyl Hydroperoxides . . . . . . . . . . . V.C.14. Epoxidation of Alkenes Containing Carbonyl Groups . . V.C.15. Epoxidation Using Urea –H2O2 Adduct . . . . . . . . . . . V.C.16. Epoxidation Using Dioxygen . . . . . . . . . . . . . . . . . . . V.D. Hydroxylations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.D.1. General Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.D.2. Hydroxylation of Aliphatic Compounds . . . . . . . . . . . . V.D.3. Hydroxylation of Aromatic Compounds . . . . . . . . . . . . V.E. Oxidation of Nitrogen-Containing Compounds . . . . . . . . . . . . . V.F. Oxidation of Sulfur-Containing Compounds . . . . . . . . . . . . . . . V.G. Oxidation of Oxygen-Containing Compounds . . . . . . . . . . . . V.G.1. Alcohols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.G.2. Ethers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.G.3. Phenols. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.G.4. Ketones, the Baeyer – Villiger Oxidation. . . . . . . . . . . V.H. CyN Cleavage Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . V.I. Acid-Catalyzed Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.I.1. Beckmann Rearrangement . . . . . . . . . . . . . . . . . . . . . . V.I.2. Synthesis of Polycarbonate Precursors. . . . . . . . . . . . . . V.I.3. Transesterification of Esters . . . . . . . . . . . . . . . . . . . . . V.I.4. Carbon – Carbon Bond Formation Reactions . . . . . . . . .

58 60 60 62 62 65 67 70 71 72 72 74 75 76 80 81 82 83 83 83 85 89 90 93 100 100 100 101 102 105 105 106 106 110 110

Contents

vii

V.I.5. Formation of Pinacols . . . . . . . . . . . . . . . . . . . . . . . . . V.I.6. Oxidative Dehydrogenation . . . . . . . . . . . . . . . . . . . . . V.J. Photocatalysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.J.1. Photocatalytic Degradation of Pollutants . . . . . . . . . . . . V.J.2. Photocatalytic Synthesis . . . . . . . . . . . . . . . . . . . . . . . . V.J.3. deNOx Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.K. Influence of Solvents. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.L. Influence of Silylation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI. Structure-Activity Correlations . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI.A. Structure of Titanium Species and Activity . . . . . . . . . . . . . . VI.B. Titanium-Oxo Species and Activity . . . . . . . . . . . . . . . . . . . VII. O – O Bond Cleavage and Product Selectivity. . . . . . . . . . . . . . . . . VII.A. General. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VII.B. Epoxidation of Alkenes . . . . . . . . . . . . . . . . . . . . . . . . . . . VIII. Conclusions and Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix A. Fingerprint Features for Ti Isomorphous Substitution in TS-1 Titanosilicates . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix B. Characteristics of the Oxo-Titanium Species Generated on TS-1 on Contact with Aqueous H2O2 . . . . . . . . . . . . . . Appendix C. Synthesis of Titanium Silicate Molecular Sieves . . . . . . . . C.1. TS-1, TS-2, Ti-ZSM-48, Ti-MWW, and Ti-MMM-1. . C.2. Ti-Beta Zeolite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C.3. Ti-Containing HMS, MCM-41, and MCM-48. . . . . . . C.4. Ti-SBA-15 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C.5. Ti-TUD-1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

114 115 116 116 120 121 122 124 127 127 128 137 137 138 140 142 142 143 143 144 146 147 147 159 159

Electron Microscopy and the Materials Chemistry of Solid Catalysts John Meurig Thomas and Pratibha L. Gai I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II. Electron Microscopy (EM) Methods . . . . . . . . . . . . . . . II.A. Electron Microscopy in Catalysis . . . . . . . . . . . . . II.B. Imaging in the Electron Microscope . . . . . . . . . . . II.C. TEM Imaging Method Using Diffraction Contrast. II.D. Theoretical Procedures . . . . . . . . . . . . . . . . . . . . III. High-Resolution Transmission Electron Microscopy . . .

. . . . . . .

. . . . . . .

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

. . . . . . .

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

. . . . . . .

. . . . . . .

174 176 177 178 179 181 181

viii

IV. V. VI.

VII. VIII.

IX. X.

Contents

III.A. Conditions Required for Optimizing HRTEM Images . . . III.B. Development of HRTEM . . . . . . . . . . . . . . . . . . . . . . . . III.C. Elucidation of the Structures of Meso- and Microporous Catalysts by HRTEM. . . . . . . . . . . . . . . . . . . . . . . . . . . III.C.1. L-Type Zeolite Catalysts . . . . . . . . . . . . . . . . . . III.C.2. Metal-Substituted Aluminum Phosphate (MAPO-36) Microporous Catalysts . . . . . . . . . . III.C.3. High-Silica Microporous SSZ-48 Catalysts . . . . III.C.4. Intergrowths in Zeolite Catalysts: Coherent, Recurrent, and Random. . . . . . . . . . . . . . . . . . . Chemical Composition Analysis with the Analytical Electron Microscope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Scanning Transmission Electron Microscopy . . . . . . . . . . . . . . Recent Advances in Ultra-High Resolution, Low-Voltage Field Emission Scanning Electron Microscopy and Extreme FESEM in Catalysis. . . . . . . . . . . . . . . . . . . . . . . . . . Cathodoluminescence Imaging for Elucidation of Electronic Structures of Catalysts. . . . . . . . . . . . . . . . . . . . . . . Recent Advances in In Situ Atomic Resolution-Environmental Transmission Electron Microscopy (ETEM) Under Controlled Reaction Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VIII.A. In Situ Investigations of Gas – Solid Reactions and Active Sites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VIII.B. Illusrative Examples . . . . . . . . . . . . . . . . . . . . . . . . . . VIII.B.1. In Situ Gas – Catalyst Reactions at the Atomic Level . . . . . . . . . . . . . . . . . . . VIII.B.2. Atomic-Resolution ETEM of Butane Oxidation. . . . . . . . . . . . . . . . . . . . . . . . . . . VIII.B.3. Atomic-Resolution ETEM of Nanorods . . . . VIII.C. Advances in In Situ Wet-Electron Microscopy Technique (Wet-ETEM) for Probing Solid Catalysts Under Liquid Environments . . . . . . . . . . . . . . . . . . . . Environmental Scanning Electron Microscopy . . . . . . . . . . . . . Electron Tomography: Three-Dimensional Electron Microscopy Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . X.A. The Topography and Location of Nanoparticles in Supported Catalysts; BSE and HAADF . . . . . . . . . . . . X.B. Pinpointing the Location of Nanoparticles Supported on Nanoporous Solids. . . . . . . . . . . . . . . . . . . . . . . . . . .

.. ..

182 184

.. ..

185 185

.. ..

186 187

..

188

.. ..

191 193

..

195

..

195

..

196

.. ..

196 201

..

201

.. ..

203 210

.. ..

210 212

..

212

..

213

..

218

Contents

XI. Energy Filtered Transmission Electron Microscopy and Elemental Maps of Solid Catalysts Using EFTEM . . . . . . . . . XII. Other Significant Trends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XIII. Critical Evaluations of the Methods and Challenges . . . . . . . . . . . XIV. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

ix

218 220 220 223 224 224

Chemistry and Technology of Isobutane/Alkene Alkylation Catalyzed by Liquid and Solid Acids Andreas Feller and Johannes A. Lercher I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II. Alkylation Mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II.A. Overall Product Distribution . . . . . . . . . . . . . . . . . . . . . . . . . II.B. Initiation Steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II.C. Alkene Addition and Isomerization . . . . . . . . . . . . . . . . . . . . II.D. Hydride Transfer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II.E. Oligomerization and Cracking . . . . . . . . . . . . . . . . . . . . . . . . II.F. Self-Alkylation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II.G. Product and Acid Degradation . . . . . . . . . . . . . . . . . . . . . . . . II.H. Pathways to Allylic and Cyclic Compounds . . . . . . . . . . . . . . II.I. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . III. Physical – Chemical Phenomena Influencing the Reaction. . . . . . . . . III.A. Properties of Liquid Acid Alkylation Catalysts . . . . . . . . . . . III.B. Properties of Zeolitic Alkylation Catalysts . . . . . . . . . . . . . . III.B.1. Adsorption and Diffusion of Hydrocarbons . . . . . . . . III.B.2. Brønsted Acid Sites . . . . . . . . . . . . . . . . . . . . . . . . . III.B.3. Lewis Acid Sites and Extra-Framework Aluminum . . III.B.4. Silicon/Aluminum Ratio . . . . . . . . . . . . . . . . . . . . . III.B.5. Metal Ions in Ion-Exchange Positions. . . . . . . . . . . . III.B.6. Structure Types of Zeolites . . . . . . . . . . . . . . . . . . . III.C. Other Solid Acids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . III.C.1. Sulfated Zirconia and Related Materials . . . . . . . . . . III.C.2. Heteropolyacids. . . . . . . . . . . . . . . . . . . . . . . . . . . . III.C.3. Acidic Organic Polymers . . . . . . . . . . . . . . . . . . . . . III.C.4. Supported Metal Halides . . . . . . . . . . . . . . . . . . . . . III.D. The Influence of Process Conditions . . . . . . . . . . . . . . . . . . . III.D.1. Reaction Temperature . . . . . . . . . . . . . . . . . . . . . . .

230 234 234 237 239 242 247 249 251 251 252 252 253 255 255 256 260 261 263 264 267 267 268 269 270 271 272

x

Contents

III.D.2. Alkane/Alkene Ratio and Alkene Space Velocity . . . III.D.3. Alkene Feed Composition . . . . . . . . . . . . . . . . . . . . IV. Industrial Processes and Process Developments. . . . . . . . . . . . . . . . IV.A. Liquid Acid-Catalyzed Processes . . . . . . . . . . . . . . . . . . . . . IV.A.1. Sulfuric Acid-Catalyzed Processes . . . . . . . . . . . . . . IV.A.2. Hydrofluoric Acid-Catalyzed Processes . . . . . . . . . . IV.B. Solid Acid-Catalyzed Processes . . . . . . . . . . . . . . . . . . . . . . IV.B.1. UOP Alkylenee Process . . . . . . . . . . . . . . . . . . . . . IV.B.2. Akzo Nobel/ABB Lummus AlkyCleane Process . . . IV.B.3. LURGI EUROFUELw Process. . . . . . . . . . . . . . . . . IV.B.4. Haldor Topsøe FBAe Process . . . . . . . . . . . . . . . . . V. Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

274 276 278 278 278 281 283 285 286 286 287 289 289

Catalytic Conversion of Methane to Synthesis Gas by Partial Oxidation and CO2 Reforming Yun Hang Hu and Eli Ruckenstein I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II. Partial Oxidation of Methane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II.A. Hot Spots in Catalyst Beds . . . . . . . . . . . . . . . . . . . . . . . . . . II.B. Minimizing O2 Purification Costs. . . . . . . . . . . . . . . . . . . . . . II.C. Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II.D. Reaction Pathways . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II.D.1. Changes in Catalyst During Reaction . . . . . . . . . . . . . II.D.2. Which is the Primary Product, CO or CO2? . . . . . . . . II.D.3. CHx Species and Rate-Determining Steps . . . . . . . . . . II.D.4. Comparison of Reactions on Reduced and Unreduced Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . III. CO2 Reforming of Methane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . III.A. Carbon Formation on Metal Surfaces . . . . . . . . . . . . . . . . . . III.B. Critical Issues Related to Carbon Deposition . . . . . . . . . . . . . III.C. Supported Noble Metal Catalysts . . . . . . . . . . . . . . . . . . . . . III.D. Non-Noble Metal Supported Catalysts . . . . . . . . . . . . . . . . . III.D.1. Ni/Al2O3 Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . III.D.2. Ni/SiO2 Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . III.D.3. Ni/La2O3 Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . III.D.4. Ni/ZrO2 Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . III.D.5. Other Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . .

298 301 301 306 312 314 315 316 318 320 321 321 322 323 324 325 327 328 330 331

Contents

III.E. MgO-Containing Solid-Solution Catalysts . . . . . . . . . . . . . III.E.1. Characteristics of MgO-Containing Solid-Solution Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . III.E.2. Highly Effective MgO-Containing Solid-Solution Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV. Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

xi

..

332

..

332

.. .. ..

333 337 338

INDEX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

347

This Page Intentionally Left Blank

Contributors Numbers in parentheses indicate the pages on which the authors’ contributions begin.

ANDREAS FELLER , Institut fu¨r Technische Chemie, Technische Universita¨t Mu¨nchen, D-85747 Garching, Germany (229) PRATIBHA L. GAI , DuPont, Central Research and Development Laboratories, Experimental Station, Wilmington, DE 19880-0356, USA and also at Department of Materials Science, University of Delaware, Newark, DE 19716, USA (171) YUN HANG HU , Department of Chemical Engineering, State University of New York at Buffalo, Buffalo, NY 14260, USA (297) H. KNO¨ ZINGER , Department Chemie-Physikalische Chemie, Universita¨t Mu¨nchen, Butenandt Strasse, 5-13, Haus E, D-81377 Mu¨nchen, Germany (1) JOHANNES A. LERCHER , Institut fu¨r Technische Chemie, Technische Universita¨t Mu¨nchen, D-85747 Garching, Germany (229) P. RATNASAMY , National Chemical Laboratory, Pune 411008, India (1) ELI RUCKENSTEIN , Department of Chemical Engineering, State University of New York at Buffalo, Buffalo, NY 14260, USA (297) D. SRINIVAS , National Chemical Laboratory, Pune 411008, India (1) JOHN MEURIG THOMAS , Davy Faraday Research Laboratory, The Royal Institution of Great Britain, 21 Albemarle Street, London, United Kingdom and also at Department of Materials Science, Cambridge CB2 1QY, UK (171)

xiii

This Page Intentionally Left Blank

Preface The forty-eighth volume of Advances in Catalysis includes a description of a new and increasingly well understood class of catalysts (titanosilicates), a review of transmission electron microscopy and related methods applied to catalyst characterization, and summaries of the chemistry and processes of isobutanealkene alkylation and partial oxidation and CO2 reforming of methane to synthesis gas. Ratnasamy, Srinivas, and Kno¨zinger provide an incisive review of recent advances in the understanding of titanosilicate catalysts, which have generated intensive research activity and already found industrial application for hydroxylation of phenol to hydroquinone and catechol. This chapter complements one by Notari in Volume 41 of Advances in Catalysis. The application of physical and computational methods has resulted in a detailed understanding of the nature and coordination state of titanium ions and functional groups such as OH on dehydrated titanosilicate molecular sieves. Tetrapodal (Ti(OSi)4) and tripodal (Ti(OSi)3OH) structures have been identified, and the interactions of these active sites with oxidant/reactant molecules during catalysis lead to the formation of oxo intermediates. The authors analyze the properties of the catalysts that influence the activity and selectivity of these sites and the reaction intermediates, showing, for example, that O – O bond cleavage can occur heterolytically or homolytically, with the relative rates determining product selectivities. The review includes a compilation of reactions catalyzed by titanosilicates, including epoxidations, hydroxylations, oxidations of nitrogenand oxygen-containing organic compounds, and acid-catalyzed and photocatalytic reactions. The results lead to correlations between catalyst structure and activity of titanium sites and reactivity of oxo-titanium intermediates. Thomas and Gai contribute an exhaustive review of advanced methods of electron microscopy, highlighting the techniques that provide the most insight into the understanding of solid catalysts. The techniques comprise highresolution real-space imaging, electron crystallography, powerful scanning probe methods, and electron energy loss spectroscopy. Recent developments in electron tomography permit the three-dimensional imaging of catalytic materials at the nano scale, and environmental cells make possible the direct in-situ probing of the dynamics of catalytic reactions at the atomic scale. The authors emphasize the complementarity of electron microscopy and other physical characterization tools (including sum frequency generation, scanning tunneling microscopy, and X-ray absorption spectroscopy) and the accompanying capabilities for xv

xvi

Preface

elucidation of the nature of solid catalysts in the electron microscope, including determination of the number and nature of crystallographic phases; electronic properties such as oxidation states of particular atoms and the electronic structure of the solid; coordination of atoms to neighboring atoms; locations of active sites; mechanisms of the release of structural oxygen and of the creation of defects; and the accommodation of catalyst non-stoichiometry. Feller and Lercher present a critical and insightful assessment of alkylation of isobutane with light alkenes, summarizing both the chemistry and processes. Alkylation is gaining in importance as aromatics and methyl-tertiary-butyl ether in motor fuels are limited by environmental concerns. Increasingly, the branched alkane products of alkylation are regarded as superior gasoline components. The authors build from the well-known chemistry of acid-catalyzed hydrocarbon conversion, using concepts such as those of carbenium ion stability and reactivity to elucidate patterns of the complex parallel and consecutive reactions. Considering both liquid-phase alkylation catalyzed by hydrofluoric acid and sulfuric acid, they draw contrasts between the two classes of processes and assess the interplay between the chemistry and effects of physical properites such as viscosity and the solubility of hydrocarbons in acid phases, which illuminate issues such as mixing and dispersion in the reactors, where the reactions occur near liquid-liquid interfaces. Feller and Lercher also consider solid-catalyzed alkylation, providing a critical review of process developments and the role of zeolite catalysts. The fundamental chemistry of zeolite-catalyzed alkylation is essentially identical to that occurring in acidic solutions, but key differences between liquid and solid catalysts result from differences in individual reaction steps originating from the variety of possible structures and distributions of acid sites in the solid catalysts; the sensitivity to a particular parameter depends strongly on the catalyst. All the acids deactivate by the formation of unsaturated polymers, which are strongly bound to the acid. Liquid acid-catalyzed alkylation is a mature technology, but solid acid-catalyzed alkylation now has been developed to a point where it eliminates most of the drawbacks of the liquid acid processes and can compete with them economically. Catalyst regeneration by hydrogen treatment is the method of choice for the solid catalysts. Hu and Ruckenstein present a review of the catalytic production of synthesis gas from methane by partial oxidation and CO2 reforming. This chapter complements that by Rostrup-Nielsen et al. in Volume 47 of the Advances, which provides an in-depth review of the chemistry and technology of steam reforming of hydrocarbons, with some information about CO2 reforming as well. Hu and Ruckenstein present results of catalyst testing experiments, chemical reaction engineering analysis, and determination of reaction networks, addressing the issue of whether CO2 and H2O are the primary products and whether CO is formed from CO2 or H2O and CH4 or directly from CH4 and O2. The rapid heat generation that results when the partial oxidation of methane produces some CO2

Preface

xvii

leads to hot-spot formation in fixed-bed reactors and potentially hazardous operation and difficulty in process control. Process options include the application of fluidized bed reactors to flatten the temperature gradients and processes that eliminate hot spots by combining the exothermic partial oxidation with the endothermic CO2 reforming or steam reforming. The partial oxidation requires an air separation unit, and a major research goal is to make the process a commercial reality by reducing the cost of air separation, for example, by using O2-permeable ceramic membrane reactors in which air could be used without pre-separation. CO2 reforming of methane is in prospect an attractive technology because it converts two greenhouse gases into useful chemicals. Catalyst deactivation, a consequence of carbon deposition, constitutes the greatest challenge in this process. Although noble metal catalysts are less sensitive to carbon deposition, Ni-containing catalysts have attracted the most research interest, and some are reported to have both high activity and stability. A solid solution catalyst offers high activity, selectivity, and stability by inhibiting carbon deposition and catalyst sintering. B.C. GATES

This Page Intentionally Left Blank

Robert L. Burwell, Jr. 1912 –2003 Robert L. Burwell, Jr., Ipatieff Professor Emeritus of Chemistry at Northwestern University, passed away at his home in Williamsburg, VA, on May 15, 2003. He will be remembered by his many friends, colleagues, and students as a learned gentleman of high moral standard, a dedicated educator, a thorough and brilliant researcher in heterogeneous catalysis, and a leading figure in guiding the catalysis community. Robert Burwell was born May 6, 1912. He graduated from St. John’s College in 1932 and received his Ph.D. in 1936 from Princeton University under the guidance of Sir Hugh Taylor. After three years as a chemistry instructor at Trinity College, in 1939 he joined the Chemistry Department at Northwestern University. During World War II, having enlisted, he worked at the Naval Research Laboratory (1942-1945). After the war, he returned to the chemistry faculty at Northwestern where he served until his retirement in 1980. He was Chair of the Chemistry Department from 1952 until 1957 and in 1970 succeeded Herman Pines as Ipatieff Professor, holding this position until his retirement. Later, as Ipatieff Professor Emeritus, he continued his research and intellectual activities for another decade. In 1994, he moved to Virginia with Elise, his wife of more than sixty years. xix

xx

Robert L. Burwell, Jr. (1912–2003)

To those who knew him personally, Burwell was not only an imposing intellect, but a warm, deeply caring, pleasant person, and a complicated individual with many facets. For instance, while wise and judicious, he nevertheless conducted himself with a great sense of humor and wit. Any whom he favored soon realized he could engage in lively conversation on practically any subject. Many of his coworkers also remembered him for his perceptive scientific advice and suggestions. Often in seminars, students felt that they learned more about a subject from Burwell’s probing questions than from the seminar itself. His family remembered him also as a caretaker extraordinaire. His devotion to his beloved Elise, particularly during the last year of her life, will be remembered by all. During his career, Robert Burwell published more than 170 original research articles. He was among the first scientists who understood the critical connection between general chemistry and catalysis. He introduced and popularized concepts that are now familiar and even commonplace within the entire catalysis community. His research themes centered around elucidation of reaction mechanisms, the nature of surface intermediates, and characterization of active sites of solid catalysts. He was well known for the use of H-D exchange for such studies. Using this technique, he identified the importance of 1,2-diadsorbed alkane on noble metal surfaces in the exchange and the hydrogenation reaction, and the irreversibility in the adsorption of alkene during hydrogenation. [J. Amer. Chem. Soc. 148, 6272 (1960); Acc. Chem. Res. 2, 289 (1969); Catal. Rev.-Sci. Eng. 7, 25 (1972)]. He established the “rollover” mechanism for cyclic hydrocarbons in these reactions [J. Amer. Chem. Soc. 79, 5142 (1957)], and the term “surface organometallic zoo.” He carefully documented the importance of surface coordination unsaturation in catalysis by metal oxides [Adv. Catal. 20, 1 (1969)] and developed new catalysts of unusual activities by deposition of organometallic complexes on alumina and silica, and by modifying silica surfaces [J. Amer. Chem. Soc. 97, 5125 (1975); J. Catal. 52, 353 (1978); J. Amer. Chem. Soc. 107, 641 (1985)]. Together with colleagues John Butt and Jerome Cohen, he completed one of the most comprehensive series of characterizations of supported noble metal catalysts, starting with the paper J. Catal. 50, 464 (1977) and concluding with the paper J. Catal. 99, 184 (1986). Burwell’s contributions to the scientific community include service on the governing body of the North American Catalysis Society from 1964 to 1977 as Director, Vice President, and, from 1973 until 1977, President. From 1955 until 1984 he served the International Congress on Catalysis, as a member of the Board of Directors; as U.S Representative; Vice President; and President (1980-84). He chaired the Gordon Research Conference on Catalysis in 1957 and was Associate Editor (1984-88) and a member of the Editorial Board of Journal of Catalysis. He served on National Research Council committees, IUPAC committees, the Petroleum Research Fund Advisory Board, the National Science Foundation

Robert L. Burwell, Jr. (1912–2003)

xxi

Chemistry Advisory Board, and others. Professor Burwell was a long-time consultant for Amoco Oil Company and was a consultant for the World Book Encyclopedia. His many scientific contributions and their industrial applications were recognized by the awards and honors he received. They include the American Chemical Society Kendall Award in Colloid and Surface Chemistry in 1973, the American Chemical Society Lubrizol Award in Petroleum Chemistry in 1983, and the Alexander von Humboldt Senior Scientist Award. The Robert L. Burwell Lectureship Award of the North American Catalysis Society was established in recognition of his outstanding contributions to catalysis. Professor Burwell was also known for the first short course in heterogeneous catalysis, which he taught for several years with Michel Boudart. Robert Burwell’s influence on the catalysis community goes beyond his science to his sharing of his many cultural interests with his colleagues, friends, and post-doctoral and graduate students. Harold Kung Kathleen Taylor Gary Haller Polly Burwell Haynes Lou Allred

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Active Sites and Reactive Intermediates in Titanium Silicate Molecular Sieves P. RATNASAMY* and D. SRINIVAS National Chemical Laboratory, Pune 411008, India

and ¨ ZINGER* H. KNO Department Chemie-Physikalische Chemie, Universita¨t Mu¨nchen, Butenandt Strasse, 5-13, Haus E, D-81377 Mu¨nchen, Germany

I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II. Active Sites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II.A. State and Framework Coordination of Ti . . . . . . . . . . . . . . . . . . . . . II.A.1. Diffraction Techniques. . . . . . . . . . . . . . . . . . . . . . . . . . . . . II.A.1.1. X-Ray Diffraction . . . . . . . . . . . . . . . . . . . . . . . . . II.A.1.2. Neutron Diffraction . . . . . . . . . . . . . . . . . . . . . . . . II.A.2. Influence of Particle Size . . . . . . . . . . . . . . . . . . . . . . . . . . . II.A.3. UV–Visible Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . II.A.4. Photoluminescence Spectroscopy . . . . . . . . . . . . . . . . . . . . . II.A.5. X-Ray Absorption Spectroscopy. . . . . . . . . . . . . . . . . . . . . . II.A.6. Vibrational Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . II.A.7. EPR Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II.B. Surface Acidity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II.B.1. Brønsted Acid Sites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II.B.2. Lewis Acid Sites and Expansion of Coordination Sphere . . . . III. Oxo-Titanium Species and Reactive Intermediates . . . . . . . . . . . . . . . . . . III.A. UV– Visible Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . III.B. Vibrational Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . III.C. X-Ray Absorption Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . III.D. Cyclic Voltametry. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . III.E. EPR Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV. Computational Investigations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

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

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

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

5 9 9 10 10 10 12 12 15 15 18 22 26 26 28 33 34 34 39 41 42 49

*Corresponding author. E-mail address: [email protected] (P. Ratnasamy); [email protected] (H. Kno¨zinger). ADVANCES IN CATALYSIS, VOLUME 48 ISSN: 0360-0564 DOI 10.1016/S0360-0564(04)48001-8

Copyright q 2004 Elsevier Inc. All rights reserved

2

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

V. Catalytic Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.A. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.B. Reactions Using H2O2 as Oxidant . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.B.1. General Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.B.2. H2O2-Catalyzed Reactions in the Homogeneous Phase . . . . . . . . . . . V.C. Epoxidation on Titanium Silicate Molecular Sieves. . . . . . . . . . . . . . . . . . . . V.C.1. General Features of Epoxidations . . . . . . . . . . . . . . . . . . . . . . . . . . . V.C.2. Yields and Stereospecificities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.C.3. Diffusional Constraints. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.C.4. Influence of Ti-Silicate Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . V.C.5. Epoxidation Catalyzed by Mesoporous Titanium Silicates . . . . . . . . . V.C.6. Influence of Alkene Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.C.7. Dialkenes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.C.8. Epoxidation in the Presence of Other Oxidizable Functional Groups . V.C.8.1. Alkenes and Alcohol Functions . . . . . . . . . . . . . . . . . . . . . V.C.8.2. Alkenes and Alkanes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.C.9. Hydroxyl-Assisted Epoxidation . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.C.10. Diastereoselectivity in Epoxidations . . . . . . . . . . . . . . . . . . . . . . . . V.C.11. Side Reactions During Epoxidation. . . . . . . . . . . . . . . . . . . . . . . . . V.C.12. Influence of pH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.C.13. Epoxidation with Alkyl Hydroperoxides . . . . . . . . . . . . . . . . . . . . . V.C.14. Epoxidation of Alkenes Containing Carbonyl Groups . . . . . . . . . . . V.C.15. Epoxidation Using Urea–H2O2 Adduct. . . . . . . . . . . . . . . . . . . . . . V.C.16. Epoxidation Using Dioxygen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.D. Hydroxylations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.D.1. General Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.D.2. Hydroxylation of Aliphatic Compounds . . . . . . . . . . . . . . . . . . . . . . V.D.3. Hydroxylation of Aromatic Compounds . . . . . . . . . . . . . . . . . . . . . . V.E. Oxidation of Nitrogen-Containing Compounds . . . . . . . . . . . . . . . . . . . . . . . V.F. Oxidation of Sulfur-Containing Compounds . . . . . . . . . . . . . . . . . . . . . . . . . V.G. Oxidation of Oxygen-Containing Compounds . . . . . . . . . . . . . . . . . . . . . . . . V.G.1. Alcohols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.G.2. Ethers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.G.3. Phenols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.G.4. Ketones, the Baeyer–Villiger Oxidation . . . . . . . . . . . . . . . . . . . . . . V.H. CyN Cleavage Reactions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.I. Acid-Catalyzed Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.I.1. Beckmann Rearrangement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.I.2. Synthesis of Polycarbonate Precursors. . . . . . . . . . . . . . . . . . . . . . . . . V.I.3. Transesterification of Esters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.I.4. Carbon–Carbon Bond Formation Reactions. . . . . . . . . . . . . . . . . . . . . V.I.5. Formation of Pinacols. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.I.6. Oxidative Dehydrogenation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.J. Photocatalysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.J.1. Photocatalytic Degradation of Pollutants . . . . . . . . . . . . . . . . . . . . . . . V.J.2. Photocatalytic Synthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.J.3. deNOx Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.K. Influence of Solvents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.L. Influence of Silylation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI. Structure-Activity Correlations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

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

55 55 56 56 58 60 60 62 62 65 67 70 71 72 72 72 72 74 75 76 80 81 82 83 83 83 85 89 90 93 100 100 100 101 102 105 105 106 106 110 110 114 115 116 116 120 121 122 124 127

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169 VI.A. Structure of Titanium Species and Activity . . . . . . . . . . . . . . . . . . VI.B. Titanium-Oxo Species and Activity. . . . . . . . . . . . . . . . . . . . . . . . VII. O –O Bond Cleavage and Product Selectivity . . . . . . . . . . . . . . . . . . . . . VII.A. General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VII.B. Epoxidation of Alkenes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VIII. Conclusions and Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix A. Fingerprint Features for Ti Isomorphous Substitution in TS-1 Titanosilicates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix B. Characteristics of the Oxo-Titanium Species Generated on TS-1 on Contact with Aqueous H2O2. . . . . . . . . . . . . . . . . . . . . . . . . . Appendix C. Synthesis of Titanium Silicate Molecular Sieves . . . . . . . . . . . . . C.1. TS-1, TS-2, Ti-ZSM-48, Ti-MWW, and Ti-MMM-1. . . . . . . C.2. Ti-Beta Zeolite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C.3. Ti-Containing HMS, MCM-41, and MCM-48. . . . . . . . . . . . C.4. Ti-SBA-15 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C.5. Ti-TUD-1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

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

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

3 . . . . . . .

127 128 137 137 138 140 142

..........

142

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

143 143 144 146 147 147 159 159

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

. . . . . . . .

This review is a summary and critical analysis of recent advances in the understanding of (a) the nature and coordination state of Ti ions and other functional groups (such as OH) on dehydrated titanium silicate molecular sieves, (b) the type and structure of the oxo intermediates generated by the interaction of these active sites with oxidant/reactant molecules during catalytic reactions, and (c) the factors that influence the reactivity and selectivity of these active sites and reaction intermediates. In the dehydrated state, most of the Ti4þ ions have the tetrapodal (Ti(OSi)4) or the tripodal (Ti(OSi)3OH) structure. On contact with H2O2, titanium oxo species, Ti(O2H) and Ti(Oz2 2 ), respectively, are formed. On reaction with organic reactants, O –O bond cleavage in these titanium oxo species occurs in a hetero- or homolytic manner. Product selectivity is determined by the relative importance of these two modes of O –O cleavage. Factors such as the coordinative environment of titanium, substituents on the O –O bond (H or alkyl), temperature, solvent, nature of the organic reactant, etc. influence the mode of O –O cleavage. Correlations between the structure and catalytic activity of titanium sites and oxo-titanium intermediates are also described. q 2004 Elsevier Inc.

Abbreviations TS-1 and TS-2 Ti-beta (Ti-b) Ti-MCM-41 and Ti-MCM-48

Ti " MCM-41 Ti-HMS Ti-ZSM-48 Ti-ZSM-12 ZSM-5

microporous titanium silicate molecular sieves with MFI and MEL structures, respectively large-pore titanium silicate with BEA structure titanium-containing Mobil composite materials/mesoporous composite materials of type 41 (hexagonal array of pores) and 48 (cubic array of pores) titanium grafted on MCM-41 titanium containing hexagonal mesoporous silica material titanium containing one-dimensional 10-ring zeolite Mobil Corporation’s one-dimensional large-pore (12-membered ring) zeolite Zeolite Socony Mobil constructed from five-memberedring building units

4

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169 Ti-SBA-15

Ti-MMM Ti-MWW Ti-TUD-1 ETS-4 and ETS-10 MST VS-2 Sil XRD UV –visible DRUV FTIR NIR EXAFS XANES XAS XAFS EPR NMR LMCT DFT HP TBHP UHP FCC TEOS TEOT TBOT TMAOH TEAOH TPAOH TPABr DDA TEA CTABr DH 0 DE TOF SEM TPD n Dn l D E gxx ; gyy ; and gzz

mesoporous, titanium-containing silica self-assembly-15 (with uniform, hexagonal, tubular channels) synthesized by using a triblock organic copolymer as a template titanium-containing microporous mesoporous material titanium silicate with MWW structure three-dimensionally randomly connected mesoporous silica Engelhard Corporation titanium silicate molecular sieves amorphous mesoporous silica-titania vanadium-containing silicalite with MEL topology silicalite X-ray diffraction ultraviolet–visible diffuse reflectance ultraviolet Fourier transform infrared near infrared extended X-ray absorption fine structure X-ray absorption near-edge structure X-ray absorption spectroscopy X-ray absorption fine structure electron paramagnetic resonance (also known as electron spin resonance (ESR)) nuclear magnetic resonance ligand to metal charge transfer density functional theory aqueous H2O2 tert-butyl hydroperoxide urea –H2O2 (1:1) adduct fluidized catalytic cracking tetraethyl orthosilicate tetraethyl orthotitanate tetrabutyl orthotitanate tetramethylammonium hydroxide tetraethylammonium hydroxide tetrapropylammonium hydroxide tetrapropylammonium bromide dodecylamine triethanolamine cetyltrimethylammonium bromide gas-phase dissociation enthalpy energy required for gas-phase heterolytic cleavage turnover frequency (moles of reactant converted per mole of active catalyst species per unit time) scanning electron microscopy temperature programmed desorption frequency shift in peak position in frequency units spin–orbit coupling constant energy gap between pxg and pyg orbitals of oxygen energy separation between 3sg and 1pgx orbitals of oxygen principal g-values

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169 pKa rip DEip ðsolventÞ DEsolv 1 m a e

5

negative logarithm of acidity constant ion pair separation energy of heterolytic cleavage in a solvent solvation energy dielectric constant dipole moment of the solvent radius of a spherical cavity formed by solvent molecules surrounding an ion pair charge of the electron

I. Introduction Hugh Taylor’s landmark postulate in 1925 that particular atoms or groups of atoms on the surfaces of solids are the active sites responsible for the catalytic activity and selectivity laid the foundation for catalysis by design (1,2). Once the active sites for a particular reaction are identified, one can, in principle, design and prepare an optimal catalyst wherein the constituents of the active sites are laid out to meet the needs of that reaction. The design and preparation of aluminosilicate-containing zeolite catalysts wherein the Al ions (the active sites) are located in different shape-selective channels and cavities (as per the needs of the reaction) is an illustration of the further development and beneficial consequences of Taylor’s postulate (1,2) in the area of acid-catalyzed reactions. Similarly, solid catalysts containing supported bimetallic nanoparticles that are highly active and selective for the hydrogenation of specific organic functional groups can now be tailor made (3,4). The discovery by Taramasso et al. (5), in 1983, of a titanosilicate zeolite with the MFI structure (titanium silicate-1, TS-1), active in oxidation reactions, raised hopes of a similar achievement in the catalysis of oxidation reactions by solids. Since 1983, many titanosilicate molecular sieves containing Ti ions in various structural and geometric locations have been synthesized and their physical, chemical and catalytic properties investigated (TS-2 (6,7), Ti-ZSM-48 (8), Ti-beta (9 –15), Ti-ZSM-12 (16), Ti-MCM-41 (17 – 19), Ti-HMS (19– 21), Ti-MCM-48 (22), Ti-MSU (23,24), Ti-SBA-15 (25 –27), Ti-MMM (28– 30), TiMWW (31) and Ti-TUD-1 (32)). TS-1 was one of the earliest classes of molecular sieves containing a transition metal cation (Ti4þ) in framework positions and possessing remarkable activity and selectivity for partial oxidation of organic reactants by aqueous H2O2. Such molecular sieves containing a redox metal cation (such as Ti4þ, Fe3þ, or V3þ) in framework positions have an enormous potential in shape-selective oxidation reactions, similar to the predominant role of their aluminosilicate analogs in acid-catalyzed reactions. However, in comparison with the enormous literature on the structure and dynamics of the acidic active sites in aluminosilicate zeolites (both Brønsted and Lewis acid

6

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

sites), our knowledge of the identity and structure of the active sites on these titanosilicates, the configuration of the reaction intermediates formed by their interaction with the oxidant/reactant molecules, and the reaction mechanism is far from adequate. An excellent overview of the early work (up to 1995) by Notari (33) and a discussion of the state and coordination of titanium ions in titanium silicates by Vayssilov (34) are already available. During the 1980s and 1990s, the main technical issues that dominated the research in this area were the confirmation of the isomorphous substitution of titanium in the MFI lattice of TS-1 and the development of fingerprints for distinguishing samples of TS-1 with good catalytic activity. These were characterized by the crystalline MFI XRD pattern; small (, 0.5 mm) particles; infrared/Raman bands at 960 and 1125 cm21; sharp peaks at 210 nm in the UV region; the absence of significant absorption in the 250– 400 cm21 region; the absence of other elements (such as Fe, Al, B, etc.); and intense yellow color upon addition of aqueous H2O2. Substitution of Ti for Si in other molecular sieve frameworks (both silicate and phosphate) and the discovery of new catalytic applications were other areas of worldwide research. Since the reviews of this area by Notari (33) and Vayssilov (34) in the mid1990s, significant advances have been made in the charaterization of these materials by use of FTIR and resonance Raman vibrational spectroscopies (35 –45), EXAFS and XANES (35,43,46– 49), EPR (50 – 54), NMR (55) and UV – visible (55 –57) spectroscopies as well as computational chemistry (41,48, 58,59,61 – 63). An informative review of the molecular structural characteristics and physical chemical properties of titania –silica catalysts was published by Gao and Wachs (64). There is a consensus now that tetrahedrally coordinated, isolated Ti4þ ions in the MFI framework of TS-1 zeolite are the precursors of the active sites for many selective oxidations. Although a coherent picture of the identity and structure of the surface groups on TS-1, TS-2, and, to some extent, Ti-MCM-41 is slowly emerging, the function and role of these surface Ti and OH groups during catalytic oxidation reactions is far from clear. Active sites are usually formed by the interaction of the solid surface with the reactant molecules during the catalytic reaction (1,2). This is especially true in oxidation catalysis with H2O2. Do the tetrahedrally coordinated Ti ions present on the “free” surface preserve their tetrahedral coordination on interaction with H2O2? In recent years, advances in in situ spectroscopic techniques have added considerably to our knowledge of the structure of the active sites and the nature of reaction intermediates on TS-1 and Ti-MCM-41 during catalysis (35 – 57). Results of these investigations suggest that the coordination number of the Ti ions expands from tetra- to penta- and 6-fold coordination on contact with H2O, H2O2, reactant, and solvent molecules. The latter are probably more relevant in the quest for the active site. Related to the nature of the titanium species present during the catalytic reaction is the structure of the oxo

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

7

intermediate formed from H2O2 on contact with the titanium ion. Here again, in situ EPR spectroscopic investigations carried out recently (51,52,54) in the presence of H2O2, (H2 þ O2), H2O, NH3, and organic reactants (such as alkenes, alcohols, and aromatic compounds) have revealed significant information about the peroxo- and superoxo-species that are probably the reactive intermediates that influence selectivity in the various oxidation reactions. In contrast to the significant progress that has been made in the structural and scientific investigations of TS-1 during the past two decades, and, notwithstanding the enormous potential of such a novel class of selective oxidation catalysts in the chemical and petrochemical industry, their commercial utilization in industrial plants has been rather disappointing. This is especially so when the applications are compared with the major commercial process breakthroughs and dozens of industrial plants using the Al-MFI analogs during a similar period after their discovery (applications include hydrodewaxing of petroleum fractions, production of ethyl benzene, xylene isomerization, methanol to gasoline conversion, use as FCC additives for production of alkenes, etc.) (65). Only one world-scale commercial plant (for hydroxylation of phenol to dihydroxy benzenes) (66) and a large pilot plant (for the ammoximation of cyclohexanone) using TS-1 are reported to be in operation so far (67,68). Apart from the higher cost of manufacture of TS-1 (the current price is about US $100/ kg), another major constraint has been the necessity to use H2O2 in stoichiometric quantities, rather than molecular oxygen, as the oxidant. Because H2O2 itself is rather expensive, its use can be commercially justified only for the manufacture of high-value products (say, those costing more than US $2/kg), thereby excluding the majority of bulk and petrochemicals. High-valued fine chemicals (used in the pharmaceutical, agrochemical, flavors, and perfumery industries) are, however, usually complex molecules too large to enter the pores of the MFI structure in TS-1. This was one of the driving forces for attempts, worldwide, to synthesize titanosilicate and titanophosphate molecular sieves with large and mesoporous structures. Such materials (such as Ti-beta, Ti-MCM-41, and Ti-SBA-15, for example) do not have the geometric constraints of TS-1. Unfortunately, even though significant success has been attained in the synthesis of such materials, they are not found to be as chemoselective as TS-1 in oxidation reactions using aqueous H2O2 as the oxidant. Their structural stability is also less (especially with regard to leaching of the Ti ions). They are more suitable when alkyl hydroperoxides are used as the oxidant, thereby lacking the advantages of inherent process simplicity and environmental advantages that ensue when aqueous H2O2 is used. Why is TS-1 more chemoselective than Ti-beta and Ti-MCM-41 (17– 19) even though Ti4þ ions are isolated and in near-tetrahedral locations in all of them? Are differences in hydrophobicity/hydrophilicity between TS-1 and the large/mesoporous material the only factors responsible for the lower

8

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

chemoselectivity of the latter? During the past few years, in situ XAFS investigations (46 – 48) have revealed that although Ti4þ ions have 4-fold coordination, in TS-1 and Ti-MCM-41, most of the Ti ions in the former have a closed tetrapodal Ti(OSi)4 structure, whereas those in the latter have an open tripodal Ti(OSi)3(OH) structure. Parallel diffuse reflectance UV (DRUV) and EPR spectroscopic investigations (51,52,54) have provided evidence that the nature of the oxo intermediates formed on contact with H2O2 depends on the intrinsic local structure and environment of the Ti ions. The tetrapodal structures seem to generate oxo species the concentrations of which correlate with selectivity in the epoxidation of alkenes. The structure of the titanium peroxo and superoxo species formed on the surface during the catalytic reaction influences the scission of the O –O bond in H2O2 (homolytic vs. heterolytic). The oxo ion/radical formed during such scission, in turn, determines the selectivity in oxidation reactions. Recent XAFS (46 –48) and Raman (39,42) spectroscopic investigations indicate that a side-on bound O2 species is formed on interaction of H2O2 with TS-1. In situ UV and EPR spectroscopic measurements also suggest (51,52) that at least some of them exist as titanium superoxide ion radicals. Such species can initiate a radical reaction pathway for the oxidation reaction. It is possible that, depending on the type of oxo species and the consequent O –O bond scission, two different mechanisms may be operative on TS-1: one involving the heterolytic O – O bond dissociation, acting, for instance, in the epoxidation of alkenes, and a second involving the homolytic O –O bond dissociation, acting in the oxidation of alkanes and side chains in alkyl aromatics (66). Although attempts have been made to replace the aqueous H2O2 oxidant with a mixture of H2 and O2 in the presence of metals such as palladium and gold (69 –74), the observed catalytic activities are much lower. But selectivities of 99% for propene oxide formation from propene were observed by Haruta and coworkers (73) with Au-containing TS-1 catalysts. In situ EPR investigations (54) have shown that similar oxo species are generated in reactions using H2 þ O2 instead of H2O2, thereby suggesting the exciting feasibility of designing efficient Ti-silicate-containing partial oxidation catalysts which can use H2 þ O2 instead of the more expensive H2O2 as the oxidant. The main objective of this review is to summarize and critically analyze recent advances made in the characterization and catalytic properties of titanium silicate molecular sieves after the reviews of Notari (33) and Vayssilov (34) in 1996 and 1997, respectively. Of special interest are (1) the nature and coordination state of Ti ions and other functional groups (such as OH) on the “free” surface of titanosilicates, (2) the type and structure of the active sites and oxo intermediates generated by interaction of these surface groups with oxidant/reactant molecules during catalysis, and

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

9

(3) the factors that influence the reactivity and selectivity of these active sites and reaction intermediates. It is hoped that the better understanding of the active sites and reaction intermediates will lead to the design of superior solid titanium-containing selective oxidation catalysts.

II. Active Sites Although many micro- and mesoporous titanosilicate-containing oxidation catalysts have been synthesized and their catalytic properties studied extensively since 1983, detailed information about surface structure and active sites is available mainly for TS-1 and, to a limited extent, Ti-MCM-41. The surface structures of titanosilicates can be described in terms of (i) the state and framework coordination of Ti and (ii) surface – OH groups present in the form of silanols and titanols. All these structural characteristics together influence the catalytic activity and selectivity. In this section, the various parameters affecting the surface structure and the methodologies adopted to quantify and distinguish the surface properties of the titanosilicate molecular sieves are discussed. The reviews by Notari (33) and Vayssilov (34) give excellent accounts of the early structural work done up to about 1995. During this period, the main subjects of investigation were (i) the state and extent of framework coordination of Ti ions, (ii) the presence, nature, and influence of extra-framework titanium, (iii) the influence of impurities (such as Al, Fe, B, etc.), (iv) the types of surface acidic sites, (v) the influence of surface hydrophobicity/hydrophilicity on catalytic activity and selectivity, and (vi) the dependence of product distribution on crystal size.

II.A. State and Framework Coordination of Ti According to Pauling’s criterion, Ti4þ cannot normally be included in framework positions in the silicate structure as its ionic radius is too large. Titanium compounds with tetrahedral geometry are scarce, as highly stable hexacoordinated complexes are more stable. However, the flexibility of the MFI framework (for example, for the reversible orthorhombic $ monoclinic transformation) or the fact that it tolerates the trigonally coordinated B atom in B-substituted ZSM-5, allows for such a substitution (5). But because of the differences in the ionic radii, the coordination about Ti cannot be perfect tetrahedral, but instead is pseudotetrahedral. Moreover, in small crystals of dimension of about 0.1 mm, of TS-1, for example, even the silicate lattice will contain many defects (Si – OH groups) and, hence, can accommodate some additional strain in accepting the larger Ti ions in tetrahedral positions.

10

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

As Ti is incorporated in the silicate lattice, the volume of the unit cell expands (consistent with the flexible geometry of the ZSM-5 lattice) (75), but beyond a certain limit, it cannot expand further, and Ti is ejected from the framework, forming extraframework Ti species. Although no theoretical value exists for such a maximum limit in such small crystals, it depends on the type of silicate structure (MFI, beta, MCM, mordenite, Y, etc.) and the extent of defects therein, the latter depending to a limited extent on the preparation procedure. Because of the metastable positions of Ti ions in such locations, they can expand their geometry and coordination number when required (for example, in the presence of adsorbates such as H2O, NH3, H2O2, etc.). Such an expansion in coordination number has, indeed, been observed recently (see Section II.B.2). The strain imposed on such 5- and 6-fold coordinated Ti ions by the demand of the framework for four bonds with tetrahedral orientation may possibly account for their remarkable catalytic properties. In fact, the protein moiety in certain metalloproteins imposes such a strain on the active metal center leading to their extraordinary catalytic properties (76). II.A.1. Diffraction Techniques II.A.1.1. X-Ray Diffraction. The X-ray patterns of silicalite-1 and TS-1 demonstrate a change from the monoclinic structure of the former to orthorhombic when Ti4þ is introduced into the silicalite framework (5). The Rietveld analysis of Millini et al. (75) demonstrates a linear dependence of the lattice parameters and unit cell volume on the extent of Ti substitution in silicalite-1 and constitutes confirmatory evidence for the location of Ti in framework positions. Millini and Perego (77) concluded that the upper limit for incorporation of Ti in the TS-1 framework is about 2.5%. XPS (78 – 80) and XANES (81 –84) data indicate that in the as-synthesized and calcined state all the Ti ions in titanosilicates are in the þ 4 oxidation state. II.A.1.2. Neutron Diffraction. There are 12 crystallographically distinct T sites in the orthorhombic structure of silicalite (MFI type), as illustrated in Fig. 1. The exact location of the Ti atoms in TS-1 could not be determined unambiguously by X-ray diffraction, even on the basis of high-quality synchrotron data (85 –87). The first evidence for non-random siting of Ti atoms was obtained by neutron diffraction (85,87,88). It is complicated to determine the preferred Ti substitution sites in TS-1 because of the low concentration of titanium (less than 2.5 Ti atoms per unit cell) and the presence of silicon vacancies. Although the neutron scattering length of titanium is quite different from that of silicon, it remains difficult to determine a multiple Ti site substitution among the 12 possible ones. Hijas et al. (87) concluded from their neutron diffraction results that Ti is distributed among only four or five of the 12 sites, with Ti occupying T3(0.30),

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

11

Fig. 1. The structure of orthorhombic form of silicalite-1 (MFI type) showing the 12 crystallographically distinct T sites. The oxygen atoms are omitted for clarity [Reprinted from Henry et al. (85) with permission. Copyright (2001) American Chemical Society].

T7(0.34), T8(0.92), T10(0.41), and T12(0.50), where the numbers in parentheses represent the estimated site occupancies for the 2.57 total Ti atoms per unit cell of the particular sample. Investigating a TS-1 sample with a Si:Ti atomic ratio of 39:1, Henry et al. (85) applied a combination of single and multiple data set Rietveld analyses exploiting the scattering length contrast between the different titanium isotopes and silicon. They succeeded in determining the silicon vacancy and titanium site substitution distribution. Both distributions were found to be non-random, with Ti preferentially substituting three of the 12 crystallographically independent framework sites, namely, T8, T10, and T3 (in the order of decreasing Ti content), and silicon vacancies being located at two framework sites, T1 and T5. Although not identical with that reported by Hijas et al. (87), this titanium siting agrees reasonably well with it. In contrast, Lamberti et al. (88) concluded from their neutron diffraction data that T6, T7, and T11 are the sites most populated by Ti. The debate about the origin of the discrepancies in these results is ongoing (85,88). Very likely, the preparation procedures of TS-1 have a significant influence on the Ti site distribution, and it was argued that kinetics rather than thermodynamics controls the framework formation and stability (85,87) (Section V.C.3).

12

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

II.A.2. Influence of Particle Size A useful “fingerprint” of an active TS-1 catalyst is the particle size of the titanosilicate (, 0.4 mm). Although the particle size influences the catalytic activity of all molecular sieves, it is especially so in the case of TS-1 and due care should be exercised in comparing samples varying in particle size (89,90). II.A.3. UV –Visible Spectroscopy Additional evidence of isolated Ti ions in tetrahedral locations in the silicate lattice comes from the diffuse reflectance UV band indicative of a charge transfer process in isolated Ti(OSi)4 or Ti(OSi)3(OH) units from the ligand oxygen to an unoccupied orbital of the central Ti ion (82,84,91). This band occurs at 210 nm for TS-1 and TS-2, at 220 nm for Ti-MCM-41 (51,52), and at 205 – 220 nm for Ti-beta(F) that was synthesized in a fluoride medium (13). TS-1 (and other titanosilicates) sometimes also contain Ti ions in other coordination states (usually six) and in non-framework locations. The latter exhibit a broad absorption in the region about 270 –290 nm. If the Ti content is high, a separate titania phase is also observed. Large anatase particles have an absorption maximum at 330 nm, and rutile absorbs at about 400 nm. Amorphous TiO2 – SiO2 shows a band at 290 nm (possibly penta- or hexacoordinated Ti). The blue shift from 330 nm (anatase) to 210 nm (TS-1) is due to isolation of the Ti ion in the silicate matrix and the change in coordination (from 6 to 4). These spectral differences among Ti ions in various environments can be related to different Ti –O –Si bond angles at the Ti sites (92). An increase of the angle will shift the bridging oxygen hybridization from sp3 to sp2 and eventually to sp, favoring a p-electron donation into the empty orbitals of Ti in Td symmetry. As a consequence, the non-bonding “e” level of Td will split into a bonding “ep” level and an empty anti-bonding “epp” level (LUMO). Because it is this LUMO that is involved in the ligand-to-metal charge transfer (LMCT) responsible for the UV band, the enlargement of the Ti –O – Si angle (as a result of a change from 6-fold to 4-fold coordination, for example) will lead to a blue shift of the LMCT band, as indeed has been observed experimentally. On the basis of XANES data, Gleeson et al. (47) inferred two types of tetrapodal structures, one having three 1408 Ti –O –Si angles and one 1608 Ti –O – Si angle and the other having only two 1408 Ti – O – Si angles but two 1608 Ti – O –Si angles (Fig. 2). These structures should, in principle, show LMCT transitions at two different positions. Except for TS-1, data representing these angles for other titanosilicates are not available. Such data would be useful in determining the influence of the Ti – O –Si angle on the ease of hydrolysis of the Ti – O – Si bond, which is crucially important for the stability and, hence, utility of the material in catalytic applications.

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

13

Fig. 2. Schematic representations of the two different tetrapodal environments: Model A, characterized by 3 Ti –O –Si angles of 1408 and 1 at 1608; Model B, characterized by 2 Ti –O–Si angles of 1408 and 2 at 1608 [Reproduced from Gleeson et al. (47) by permission of the PCCP Owner Societies].

Table I illustrates the utility of DRUV –visible data in determining the surface structures involving Ti. Samples of TS-1 were prepared by three different methods or treatments. Samples 1 and 2 were prepared by conventional hydrothermal synthesis and sample 3 by synthesis in a fluoride medium. TS-2 was synthesized as reported (7). At least five bands could be discerned by deconvolution (Fig. 3), at 205, 228, 258, 290, and 330 nm. Band 1 at 205 nm is assigned to tetrahedral, tetrapodal Ti present in TS-1, TS-2, and Ti-beta. Band 5 at 330 nm is assigned to an

TABLE I Diffuse reflectance UV-visible data of titanosilicate samples Titanosilicatea

Deconvoluted bands and assignments: lmax , nm (relative intensity, %) Band 1 (Ti(OSi)4)

Band 2 (Ti(OH)(OSi)3)

Band 3 (Ti(OH)(H2O)(OSi)3)

Band 4 (Ti(OH)2 (H2O)2(OSi)2)

Band 5 (Anatase-like)

TS-1 (Sample 1) TS-1 (Sample 2) TS-1 (Sample 3) TS-2

206 (85)

228 (8)

258 (6)

293 (1)

Nil

203 (72)

228 (10)

255 (8)

288 (5)

328 (5)

206 (78)

229 (11)

260 (7)

293 (4)

Nil

201 (58)

229 (13)

255 (24)

288 (5)

Nil

Ti-MCM-41

207 (27)

227 (49)

263 (8)

290 (16)

Nil

Adapted from Shetti et al. (93). a All the titanosilicates (TS-1 (Si/Ti ¼ 33), TS-2 (Si/Ti ¼ 30) and Ti-MCM-41 (Si/Ti ¼ 35)) except TS-1 (sample 3) were synthesized by the conventional pre-hydrolysis method (see Appendix C). Sample 3 was synthesized in the fluoride medium.

14

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Fig. 3. Experimental and deconvoluted DRUV–visible spectra of TS-1 ðSi=Ti ¼ 33Þ and TS-2 ðSi=Ti ¼ 30Þ samples prepared by various methods/treatments. Deconvoluted bands are representated by 1–5 [from Shetti et al. (93)].

anatase—such as phase. Band 2 at 228 nm is probably best assigned to tetrahedral, tripodal Ti (present in all the samples, with the maximum amount in Ti-MCM-41). Bands 3 and 4 are probably best attributed to penta- and hexacoordinated open Ti structures in which Ti is attached to ligands such as H2O.

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

15

II.A.4. Photoluminescence Spectroscopy Because of the high sensitivity of Ti-containing luminescence centers to their local environments, photoluminescence spectroscopy can be applied to discriminate between various kinds of tetrahedral or near-tetrahedral titanium sites, such as perfectly “closed” Ti(OSi)4 and defective “open” Ti(OSi)3(OH) units. Lamberti et al. (49) reported an emission spectrum of TS-1 with a dominant band at 495 nm, with a shoulder at 430 nm when the sample was excited at 250 nm. When the excitation wavelength was 300 nm, the emission spectrum was characterized by a dominant band at 430 nm with a shoulder at 495 nm. These spectra and their dependence on the excitation wavelength clearly indicate the presence of two slightly different families of luminescent Ti species, which differ in their local environments, in agreement with EXAFS measurements carried out on the same samples. When photoluminescence spectra were recorded for a Ti(OSi(CH3)3)4 model compound, upon excitation at 250 nm only one emission band was detected (at 500 nm), which was assigned to a perfect “closed” Ti(OSi)4 site. The excitation of these species is considered to be a LMCT transition, O22Ti4þ ! (O2Ti3þ)p, and the emission is described as a radiative decay process from the charge transfer state to the ground state, O2Ti3þ ! O22Ti4þ. Soult et al. (94) also observed an emission band at 499 nm, which they attributed to the presence of a long-lived phosphorescent excited state. The emission band at 430 nm of TS-1 was tentatively assigned to a defective “open” Ti(OSi)3(OH) site (49). Ti-beta at 77 K exhibits a photoluminescence spectrum at about 465 nm (95). The excitation was at 260 nm. Addition of H2O and CO2 quenches the photoluminescence, H2O being more effective than CO2 (Fig. 4). The lifetime of the charge transfer excited state was also shortened by such additions, indicating that H2O and CO2 interact with the Ti4þ ions in both the ground and excited states. Recently, Gianotti et al. (96) reported photoluminescence and DRUV spectra of pure siliceous MCM-41 and Ti-MCM-41 containing Ti4þ species anchored to the inner walls of the siliceous MCM-41. They observed complex luminescence signals and concluded that these could be used for a clear distinction of the emission of tetrahedral Ti4þ ions from those of silica surface centers. II.A.5. X-Ray Absorption Spectroscopy A distinctive feature of Ti4þ ions in tetrahedral coordination is the intense XANES peak at 4969 eV (39,97). The position and intensity of the pre-Ti K edge peaks can throw significant light on the coordination number and corresponding concentrations of surface Ti ions. The pre-edge intensity arising from the transition between the core level (in this case 1s) to an unoccupied or a partially occupied level (3d, which is unoccupied, because Ti4þ is a d0 system) is known

16

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Fig. 4. (a) The photoluminescence spectrum of Ti-beta(OH) and the effects of the addition of CO2 ((b) 0.5 mmol CO2/g) and H2O ((c) and (d) 0.1 and 0.5 mmol H2O/g, respectively) molecules on the photoluminescence spectrum. Measurements were made at room temperature with excitation at 260 nm [Reproduced from Yamashita et al. (95) with kind permission of Kluwer Academic Publishers].

to be sensitive to the symmetry of the coordination environment. Ti4þ ions in octahedral positions show low intensity (because the corresponding A1g ! T2g and A1g ! Eg transitions are symmetry-forbidden) and those in tetrahedral positions show the maximum intensity. Penta-coordinated Ti4þ ions (square pyramidal, for example) exhibit intermediate values. Rutile and anatase, in which all the Ti ions are in 6-fold coordination, exhibit three low-intensity peaks. Titanium complexes, some of which are known from single crystal XRD data to incorporate Ti4þ ions in Td positions, or well-synthesized samples of TS-1 exhibit an intense peak in the pre-edge region (Fig. 5), the intensity of which should be proportional to the Ti content of the sample. When the intensities of pre-edge peaks of samples containing varying amounts of Ti are normalized to the absorption edge jump (i.e., to the respective total amount of absorbing Ti atoms contained in the sample), the resulting values are invariant, as shown in the inset in Fig. 5, thus demonstrating the proportionality between pre-edge peak intensity and the amount of Ti in a given sample. Difficulties may arise when a sample contains Ti ions in more than one type of location (the usual case). An intense peak representative of tetrahedral Ti (the majority species) can then also include contributions from minor quantities of Ti in 5- and 6-fold coordination (34). In particular, such species are observed if the samples are not fully dehydrated or contain larger amounts of Ti. EXAFS investigations of TS-1 (98,99) and TS-2 (81,100) indeed showed the presence of

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

17

Fig. 5. XANES spectrum of a typical TS-1 sample in vacuum. Inset: intensity of the pre-edge peak (spectra normalized to the edge jump) for samples with various Ti contents. Because the height of the edge jump is proportional to the Ti content, the intensity of the normalized pre-edge is invariant (within experimental uncertainty) with Ti concentration [Reprinted from Ricchiardi et al. (41) with permission. Copyright (2001) American Chemical Society].

6-coordinated Ti in addition to the tetracoordinated Ti species. This technique is not sensitive enough to discriminate between mixtures of this predominant species with other oxidic tetrahedral species (101,102). DFT calculations (103) indicated the possible coexistence of various oxidic tetrahedral structures, as the difference in energy between them was very small (about 20 kJ/mol). Wellprepared, dehydrated, titanocene grafted on MCM-41 (104) and TS-1 (47,49) catalysts contained mainly the tetrahedral, tripodal (in Ti-MCM-41) and tetrapodal structures (in TS-1) as the most plausible of the averaged structures. Blasco et al. (13) observed single sharp and intense pre-edge peaks for calcined dehydrated Ti-beta silicates which were synthesized in either an OH2 (Ti-beta(OH)) or F2 (Ti-beta(F)) medium, suggesting the uniformity of the tetrahedral Ti species in these materials. Rehydration affected the pre-edge peak, resulting in a decrease of the intensity, a shift of the peak position to higher energy, and a peak broadening. The effects of rehydration were more noticeable for samples synthesized in an OH2 medium, and it was concluded that the degree of interaction of titanium with water was strongly influenced by the hydrophobic/

18

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

hydrophilic character of the zeolitic framework. The XANES spectra of hydrated Ti-beta(F) were consistent with the presence of Ti in either 4 or 5-fold coordination, indicating the strong adsorption of one water molecule per Ti atom. This result was confirmed independently by adsorption measurements. In contrast, the XANES spectrum of hydrated Ti-beta(OH) was consistent with a mixture of 5 and 6-fold coordinated Ti atoms, suggesting the preferred adsorption of one or two water molecules per Ti atom, as supported by independent adsorption measurements. II.A.6. Vibrational Spectroscopy In addition to the characteristic XRD patterns and photoluminescence, UV – visible and X-ray absorption spectra, another fingerprint thought to indicate lattice substitution of titanium sites was the vibrational band at 960 cm21, which has been recorded by infrared and Raman spectroscopy (33,34). Although there is some controversy about the origin of this band, its presence is usually characteristic of a “good” TS-1 catalyst, although it turned out to be experimentally extremely difficult to establish quantitative correlations between the intensity of the 960 cm21 band and the Ti content of a Ti silicate and/or its catalytic activity. The band at 960 cm21 was already reported in the original TS-1 patent (5) and attributed to the presence of isomorphously substituted Ti in the silicate lattice. It was shown later that an analogous band in the 960– 970 cm21 range also characterizes other Ti silicates, namely, TS-2, Ti-ZSM-48, Ti-beta, and Ti-MCM-41 (34). This band was attributed in early work to a Si – O stretching vibration in a Si – O –Ti group (91) and later to a titanyl TiyO group (105). The attribution of the band to the presence of Ti in the silicate matrix was based on the argument that Ti-free silicates would not show any vibrational modes in the 950– 970 cm21 region. However, this reasoning is not entirely valid, because the presence of bands in this region, although they are weak, has been reported for the Raman spectra of pure silicalite-1 (106) and for the infrared spectra of crushed silica, alkali silicates, and silica gels (107 – 109). Therefore, Camblor et al. (110) assigned the band at 960 cm21 to the stretching vibration of Si – O2 groups. An analogous band was also observed in the spectra of zeolites with high concentrations of defects. The observation of an oxygen isotope effect (9,10,111) and the absence of a hydrogen isotope effect were considered consistent with this band assignment. However, it was recently demonstrated that in Ti-beta synthesized by the fluoride route there is no noticeable hydrolysis of Ti – O –Si bonds (13). Consequently, bands near 960 cm21 cannot be attributed to Si – OH defects, which are essentially absent from these zeolites. It was, therefore, concluded (112) that the stretching of Si –O bonds in Si –O – Ti groups is the major contribution to the absorption in this region in Ti silicates, in agreement with previously reported results (91,112,113).

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

19

Boccuti et al. (91) interpreted the 960 cm21 band on the basis of a consideration of the effect of a TiO4 unit on the vibrational modes of a neighboring SiO4 tetrahedron. The Si –O stretching mode was expected to shift to lower wavenumbers because of the higher ionicity of the Ti – O bond (Si – Od2zzzTidþ). The quantum chemical (SCF) calculations of de Man and Sauer (62) suggested that the 960 cm21 band can be interpreted as an antisymmetric stretching mode of the Si –O – Ti bridge in a Ti(OSi(OH)3)4 unit in which Ti is tetracoordinated. Ricchiardi et al. (41) pointed out that these band assignments may be considered as coincident because they describe the same physical mode on the basis of different building units. Su et al. (114), in an investigation of a wide variety of silicotitanates by Raman spectroscopy, concluded that for titanosilicates containing isolated TiO6 units, a strong band at 960 cm21 indicative of the [(O3Si – O)]d2 – [(TiO5)]dþ stretching mode will dominate the spectra. In contrast, Smirnov and van de Graaf (115), applying molecular dynamics techniques, calculated the vibrational spectrum of a periodic model of TS-1 containing TiO4 tetrahedra and supported the localized Ti – O – Si nature of the 960 cm21 vibration. They also emphasized that the Si –O and Ti – O bands are not equivalent and that the Si –O stretching makes the greater contribution to the vibration, consistent with previous conclusions (41,91). Further support for the direct relationship of the 960 cm21 band to the presence of 4-coordinated Ti atoms in the framework of TS-1 came from the photoluminescence investigations of Soult et al. (94). At 12 K, an emission band was observed at 490 nm, which was unequivocally attributed to titanium (Section II.A.4). This band showed a resolved vibrational structure of 966 ^ 24 cm21, which clearly demonstrates that Ti is involved in the corresponding vibrational mode. This relationship was recently questioned by Li et al. (40,116) when they reported the observation of bands at 490, 530 and 1125 cm21 in the UV-excited (244 nm) Raman spectra of TS-1. Bands at 1085 and 1110 cm21 were also observed for Ti – SiO2 prepared by chemical grafting (117) and for Ti-MCM-41 (118), respectively. Raman bands near 1120 cm21 in addition to the 960 cm21 band had been reported earlier for TS-1 by Scarano et al. (113) and Deo et al. (119), who used conventional Raman spectroscopy (NIR excitation), and later by Bordiga et al. (39), who used UV –visible- and NIR-excitation. Li et al. (40,116) were the first to show that the bands of TS-1 at 490, 530, and 1125 cm21 and the corresponding bands of Ti-MCM-41 were resonance-enhanced when the Raman spectra were excited in the UV (244 nm) in the wavelength region of the O22Ti4þ ! O2Ti3þ LMCT absorption (band at 220 nm; see Section II.A.3), whereas the 960 cm21 band was not resonance-enhanced. On the basis of this observation, the authors concluded that the oscillator responsible for the 960 cm21 band cannot be located in the immediate vicinity of the Ti atom. Consequently, they also proposed that the three resonance-enhanced bands at

20

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

490, 530, and 1120 cm21 were the real fingerprint for the presence of Ti in the framework. The three bands were assigned to the bending, symmetric, and antisymmetric stretching modes of a Ti –O – Si unit (116). Unfortunately, the different selection rules that apply to resonant and normal Raman scattering were not taken into account in this spectral interpretation. In the following, it is shown that the conclusions and assignments mentioned above have to be modified on the basis of symmetry considerations as discussed by Ricchiardi et al. (41). Figure 6 reproduces the Raman spectra in the region 800 –1200 cm21 reported by these authors for pure silicalite (sample 1) and for two TS-1 samples, 3 and 5, which contain 1.4 and 3.0 wt% TiO2. The spectra shown in Fig. 6a were recorded with a Fourier transfrom (FT) Raman spectrometer at an excitation wavelength of lexc ¼ 1064 nm (9398 cm21), whereas those shown in Fig. 6b were excited with a UV – laser line at lexc ¼ 244 nm (40,984 cm21). With each excitation wavelength, the pure silicalite gives rise to weak bands at 975 and 1085 cm21 and a complex band centered near 800 cm21. In the FT-Raman spectra of the dehydrated TS-1 samples (Fig. 6a), a band is clearly visible at 960 cm21, the intensity of which increases with TiO2 content. This band is not to be confused with the silicalite band that is observed at 975 cm21. In addition, a band appears at 1125 cm21, the intensity of which, although relatively low, also grows with the TiO2 content. Hence, both bands

Fig. 6. Raman spectra of sample 1 (Ti-free silicalite), and samples 3, and 5 (TS-1 with TiO2 wt% being 2 and 3, respectively). (a) Spectra collected with a l ¼ 1064 nm (9398 cm21) excitation. (b) Spectra collected with a l ¼ 224 nm (40,984 cm21) excitation. Inset: UV–DRS spectrum of sample 5. Vertical line indicates the position of the excitation wavelength l used for collecting the sample reported in part (b). Vertical dotted lines are placed at 960 cm21. Spectra of both parts have been vertically shifted for clarity [Reprinted from Ricchiardi et al. (41) with permission. Copyright (2001) American Chemical Society].

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

21

may be considered as fingerprints of the Ti incorporation into the silicalite framework. In contrast, the UV –excited Raman spectra (Fig. 6b) show a weak band at 960 cm21 and a very strong band at 1125 cm21, suggesting a resonance enhancement of this vibration, but not of the 960 cm21 band, consistent with the observations reported by the group of Li (40,116 – 118). The requirements for Raman resonance that must be fulfilled are the following (120,121): (a) total symmetry of the vibrations with respect to the absorbing center, and (b) same molecular deformation induced by the electronic and vibrational excitations. Quantum chemical calculations (41) of the vibrational frequencies and the electronic structure of shell-3 cluster models allowed the assignment of the main vibrational features, as shown in Fig. 7. The 1125 cm21 band is unequivocally assigned to the symmetric stretching of the TiO4 tetrahedron. Vibrations of the TiO4 tetrahedron, achieved via in-phase, anti-symmetric stretching vibrations of the four-connected Ti –O –Si oscillators, are outlined in Fig. 8b. Considering the electronic structure of the Ti moiety and the symmetry of this mode, it is the only vibration that fulfills the resonance Raman selection rules (a) and (b) above. This vibrational mode can be described equivalently as the in-phase stretching of the four Si –O bonds surrounding Ti. The 960 cm21 band is assigned to the antisymmetric stretching mode of the TiO4 unit, which can

Fig. 7. Calculated vibrational frequencies for the Ti[OSi(OH)3]4 model, classified following the symmetries of the T–O –T unit (upper part) or according to the symmetries of the TO4 unit (lower part) [Reprinted from Ricchiardi et al. (41) with permission. Copyright (2001) American Chemical Society].

22

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Fig. 8. (a) Definition of symmetric and antisymmetric stretching modes of the T– O–T bridges. (b) Symmetric stretching of the central tetrahedron, achieved through in-phase antisymmetric stretching of the four connected Ti–O –Si bridges. (c) One of the antisymmetric stretching modes of the central tetrahedron, achieved through out-of-phase antisymmetric stretching of the Ti –O– Si bridges [Reprinted from Ricchiardi et al. (41) with permission. Copyright (2001) American Chemical Society].

be described as the out-of-phase-antisymmetric stretching of the four connected Ti –O –Si oscillations or as the out-of-phase stretching of the four Si –O bonds surrounding the Ti atom (Fig. 8c). This vibrational mode does not fulfill the resonance Raman selection rules (a) and (b) above and is, therefore, not expected to be resonance-enhanced, consistent with the experimental results (Fig. 6). On the basis of these assignments, the two bands must be associated with the presence of isolated Ti atoms in tetrahedral coordination within the silicalite framework. Consequently, a quantitative linear correlation between the TiO2 content and the intensities of both the infrared and Raman bands at 960 cm21 is expected—and this is indeed observed, as shown in Fig. 9b. Furthermore, both the resonant (Fig. 6b) and non-resonant (Fig. 6a) Raman spectra give a constant value for the ratio of the intensity of IR band at 1125 cm21 to that at 960 cm21 ðIð1125Þ=Ið960ÞÞ ratio of 0.25 and 11, respectively, for samples with varying TiO2 contents. This result suggests that the two bands should be related to two different spectroscopic manifestations of the same phenomenon, namely, incorporation of Ti in the silicalite framework (41). II.A.7. EPR Spectroscopy Electron paramagnetic resonance (EPR) spectroscopy is yet another diagnostic tool for the detection of isomorphous substitution of Ti. Its sensitivity is very high, and investigations can be performed with samples even with very low contents of paramagnetic species. The spectra and g parameters are sensitive to the local structure and associated molecular distortions. Hence, it is an ideal tool to characterize Ti in titanosilicates. Ti in the þ 4 oxidation state in titanosilicates is diamagnetic and hence EPR-silent. Upon contacting with CO or H2 at elevated

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

23

Fig. 9. (a) Infrared spectra of outgassed thin pellets of Ti-free silicalite (curve 1) and TS-1 with increasing Ti content x (curves 2 –5). Spectra were normalized by means of the overtone bands between 1500 and 2000 cm21 (not shown) and vertically shifted for clarity. The thick horizontal line represents the fwhm of the 960 cm21 band for sample 2. By assuming that this band has a constant fwhm for any x; the absorbance W obtained is plotted as the ordinate in panel b, where the band has the same fwhm as in curve 2 (horizontal thin lines). (b) Intensity W of the 960 cm21 infrared band (normalized absorbance units) as a function of x (full squares) and corresponding Raman counts (open squares) [Reprinted from Ricchiardi et al. (41) with permission. Copyright (2001) American Chemical Society].

temperatures, the Ti ions are reduced from a diamagnetic þ 4 (3d0) to a paramagnetic, EPR-active þ 3 (3d1) oxidation state. Tuel et al. (122) and Zecchina et al. (123) used this technique to differentiate Ti3þ ions from framework and extraframework precursors. Later, Kevan and co-workers (124 –129) investigated TS-1 and Ti-MCM-41 reduced with g-radiation. This method is, however, valid only if the reduced structure retains a structure memory of the precursor. Recently, Srinivas and Ratnasamy (130,131) reported a detailed EPR investigation of Ti3þ in titanosilicate molecular sieves, TS-1, Ti-MCM-41, ETS-4, and ETS-10 (Fig. 10). Ti4þ was reduced to Ti3þ by dry hydrogen. Only one type of Ti3þ species (I) was identified when the sample was reduced at 673 K. However, reduction at 873 K revealed two non-equivalent Ti3þ ions (species I and II) in TS-1 and Ti-MCM-41 (Table II). ETS-4 and ETS-10 contained only one type of Ti3þ ion in octahedral positions. In agreement with the other spectroscopic investigations (XAS and UV), EPR gave evidence for the presence of two types of tetrahedral Ti (tetrapodal and tripodal) structures in TS-1 and Ti-MCM-41, differing in their reducibility (130,131). The EPR g-parameters (Table II) indicate that Ti3þ ions in TS-1 and Ti-MCM-41 have a tetragonally elongated Td

24

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Fig. 10. EPR spectra (at 77 K) of Ti3þ generated by contacting TS-1, Ti-MCM-41, and ETS-10 with dry H2 at 873 K, and ETS-4 at 673 K. Signals denoted by an asterisk correspond to superoxo radical species generated by further reaction of Ti3þ with O2 [from Bal et al. (130)].

geometry whereas those in ETS-4 and ETS-10 have a tetragonally compressed Oh geometry. The reducibility of Ti (monitored by formation of Ti3þ) varied with the type of silicate structure. The spectra normalized (with respect to the Ti atoms in TS-1) indicate that the overall signal intensity of Ti3þ ions decreases in the following order: ETS-10 . ETS-4 q TS-1 at 673 K and ETS-4 . ETS-10 . Ti-MCM41 . TS-1 at 873 K. Apparently, it is more difficult to reduce Ti in a tetrahedral coordination geometry (as in Ti-MCM-41 and TS-1) than in an octahedral geometry (as in ETS-10 and ETS-4). The intensity of the Ti3þ signals increased with an increase in the reduction temperature (673 –873 K). The g-values are sensitive to the silicate structure (Table II). Whereas both the Ti3þ species (I and II) in TS-1 are characterized by axial symmetry, species I has axial symmetry, and II has rhombic symmetry in Ti-MCM-41. In each structure, gk , g’ . In the case of ETS-10, gk . g’ ; and for ETS-4, gk , g’ : The investigations also showed that counterions and additives also influence the redox properties. ETS-10 samples were exchanged with Csþ ions to examine

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

25

TABLE II EPR spin Hamiltonian parameters (at 77 K) of Ti3þ in titanosilicate molecular sieves generated by reduction with dry hydrogen Sample

Reduction temperature (K)

ETS-10

Species

gk

g’

873

1.969

1.942

673

1.966

1.941

gzz

gxx

gyy

ETS-10(Cs)

673

1.869

1.944

1.959

ETS-10(Cs)-Pt

673

1.870

1.943

1.959

1.894

1.938

1.974

1.894

1.938

1.974

ETS-4

673

ETS-4-Pt

673

TS-1

873

1.863

1.930

I

1.870

1.920

II

1.863

1.930

I

1.930

1.956

II

1.916

1.956

673

I

1.930

1.956

TS-1-Pt

673

I

1.931

1.955

Ti-MCM-41

873

I

1.902

1.958

II Ti-MCM-41-Pt

873

I II

1.906

1.958

Adapted from Bal et al. (130).

the interaction of extraframework ions with titanium. The exchanged samples (ETS-10(Cs)) were then reduced with dry H2 at 673 K. The spectrum of ETS-10 containing Naþ/Kþ ions is characterized by axial g values with gk . g’ : After exchange of the cations with Csþ, the spectrum corresponded to rhombic g-values with gzz , gxx ; gyy (Fig. 11) and the overall Ti3þ signal intensity decreased by a factor of about three. A platinum (0.05 wt%)-impregnated ETS-10(Cs) sample showed spectra similar to that of ETS-10(Cs) ðgzz , gxx ; gyy Þ; except that the Ti3þ signal intensity increased by a factor of about 2.4 compared with that of the ETS-10(Cs) sample. Although the reduction in Ti3þ intensity by Cs is attributed to greater stabilization of Ti4þ ions by the more basic and larger Cs atoms, the increase in the intensity induced by platinum is attributed to better activation of the reductant molecules (H2) by platinum and the consequently greater reduction of Ti4þ to Ti3þ. In other words, both cesium and platinum influence the reducibility

26

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Fig. 11. EPR spectra of Ti3þ (at 77 K) showing the influence of Cs exchange and platinum impregnation on the intensity and g-parameters of Ti3þ signals in ETS-10 reduced in dry H2 at 673 K (signals denoted by an asterisk correspond to superoxide radical species generated by secondary reactions by Ti3þ interaction with O2) [from Bal et al. (130)].

of Ti. Similar enhancements in Ti3þ signal intensity of TS-1 (by 3 times) and Ti-MCM-41 (by 1.35 times) were observed when the titanosilicates were impregnated with platinum. II.B. Surface Acidity II.B.1. Brønsted Acid Sites In addition to the Ti, hydroxyl groups constitute a second class of surface functional groups on dehydrated samples that can be of importance in catalytic reactions. The presence of a large number of Si – OH groups on the surfaces of all the titanosilicates is apparent from the intense absorption in the 3200 – 3800 cm21 region of the infrared spectra. The experimental evidence of surface

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

27

Ti – OH groups, on the other hand, is scarce. Titanol groups on Ti-grafted MCM-48 (132) and TS-1 (133) have been claimed to absorb at about 3676 cm21. In the case of TS-1, the 3676 cm21 band was not observed (133) on the free dehydrated surface, but instead only as a result of contact with H2O2 and photoirradiation. TS-1 typically contains a high density of framework defects (Si vacancies) generating internal, hydrogen-bonded hydroxyl groups (silanols as well as possibly titanols acting as potential weak Brønsted acid sites) (49,134). The infrared spectra in the O –H stretching region of dehydrated TS-1 and pure silicalite are, therefore, very similar to each other and characterized by broad bands, which do not allow an easy discrimination between titanols and silinols (43,44,135,136). The presence of acidity in TS-1 was inferred from typical acid-catalyzed reactions, such as the formation of diols in epoxidation reactions (137), rearrangement of cyclohexanone oxime to caprolactam (138,139), and the cycloaddition of CO2 to epoxides (140), the latter two not involving the use of H2O2 during the reaction. Although there is no doubt about the presence of functional acid sites on dehydrated TS-1 (and other titanosilicates), their type (Brønsted or Lewis), structure and concentration have not yet been conclusively established. Of course, acidity can be generated, in situ, during oxidation reactions in the presence of H2O2, because the peroxide proton-donor group, generated by coordination of H2O2 to the titanium sites, can be quite acidic (111). But, as noted earlier, there is evidence for the occurrence of acid-catalyzed reactions on TS-1 even in the absence of H2O2 (138– 140). However, results of earlier investigations of the acidity of TS-1 have to be viewed with caution because of inadequate appreciation of the influence of impurities (such as Fe, Al, B, etc.) and non-framework Ti ions in generating surface Brønsted acidity on these materials. The Brønsted acid strength of the hydroxyl groups on dehydrated TS-1 was tested by measuring the wavenumber shift DnOH of the O – H stretching bond induced by hydrogen bonding with probe molecules (141,142), viz., CO (135,143), acetonitrile (100,136,141), tert-butylnitrile (141), and pyridine (44). The O – H stretching spectra of TS-1 and pure silicalite resulting from the adsorption of the probe molecules were practically identical for all probes. For example, the O –H stretching band was found at 3390 cm21 for silicalite-1 and at 3400 cm21 for TS-1 upon contact with acetonitrile. The corresponding wavenumber shift is very close to the shifts of 300 – 330 cm21 reported for amorphous silica after adsorption of acetonitrile (64,144). Brief outgassing caused the almost complete disappearance of the band due to hydrogen bonding, without leaving evidence of the presence of other components indicating that the OH groups on TS-1 were not more acidic than those on silicalite-1. The main conclusion was that the presence of Ti in the silicalite lattice does not generate new OH groups or does not induce detectable Brønsted acidity in the Si– OH groups of the silicalite (135,139).

28

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Conclusions, some of them contrary to the above, were reached more recently by Zhuang et al. (145) from a combination of 31P and 1H MAS NMR spectroscopy of adsorbed trimethylphosphine. These authors found not only Lewis acid sites (vide infra), but also Brønsted acid sites in TS-1 (145). They claimed that the 1H, 29Si MAS NMR spectra and the resonance related to Brønsted acid sites in the 31P MAS NMR demonstrated clearly that the “presence of Ti in the framework results in the formation of a new OH group, titanols, which is more acidic than the silanols of silicalite-1 (145)”. The peak at 4.3 ppm in the 31P MAS NMR spectra was assigned to a ((CH3)3P – H)þ complex arising from the interaction of (CH3)3P with Brønsted acid sites present on TS-1. The origin of this proton is not clear at present, especially because the 1H MAS NMR spectra of the same TS-1 samples did not differ significantly from those of silicalite-1 (145); the latter, when free from impurities, is not known to be a Brønsted acid. In conclusion, dehydrated TS-1 (and presumably other titanosilicates) most likely does not have Brønsted acid centers. The observed activity for acidcatalyzed reactions that yield undesired side products is, therefore, inferred to be created under reaction conditions in the presence of aqueous H2O2 (vide infra). II.B.2. Lewis Acid Sites and Expansion of Coordination Sphere Although there are doubts about the existence of Brønsted acid sites on TS-1 and related materials, there is strong evidence that Lewis acid sites are present on the surface of dehydrated TS-1. The significant activity of TS-1 and of Ti-MCM-41 in the cycloaddition of CO2 to epoxides to give cyclic carbonates (140), a reaction typically catalyzed by Lewis acids such as AlCl3, SbF5, etc., lends strong support to the inference of the existence of Lewis acid sites on their surfaces. Infrared spectroscopic evidence of Lewis acidity comes from recent spectra of CH3CN adsorbed on TS-1 (136). In the liquid state, the C –N stretching vibration is characterized by a doublet at 2294 and 2254 cm21, which is caused by Fermi resonance (144). Upon interaction with electron-withdrawing groups, these frequencies are shifted to higher values (146– 148). When CH3CN is adsorbed on silicalite-1, the bands shift to 2297 and 2263 cm21. The slight shift to higher energy was attributed to hydrogen bonding with the silanol groups that act as weak electron-withdrawing centers from the nitrile nitrogen lone pair. In the case of TS-1, two doublets were observed, the first at 2313 and 2291 cm21 and the second at 2290 and 2256 cm21. The band at 2256 cm21 and one of the bands in the 2290 cm21 region decrease in intensity faster than the others upon outgassing as a result of the desorption of hydrogen bonded acetonitrile from the Si –OH sites. The positions of the other (more stable) doublet (2313 and 2290 cm21) is similar to that found in the spectrum of anatase, TiO2, on which two Lewis-bonded species, characterized by two doublets at 2315 and 2290 cm21 and 2304 and 2274 cm21 were observed earlier (149) and assigned to CH3CN

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

29

attached to tetra- and penta-coordinated Ti4þ ions. The observation of a similar doublet (at about 2313 and 2290 cm21) in the case of both TS-1 and anatase on adsorption of CH3CN suggests that Ti4þ ions in TS-1 also possess Lewis acidity similar to that in anatase. The detailed interpretation of the C –N stretching region of CH3CN is relatively complex because of the Fermi resonance between the C – N stretching fundamental mode n2 and the combination mode of the C – C stretching and symmetric CH3 deformation modes that leads to the doublet mentioned above (146). Therefore, the use of CD3CN as a probe molecule is preferred, as this has only a single C –N stretching band (at 2259 cm21) in the free molecule. This band shifts to higher wavenumbers when the molecule forms a coordination bond (146). Bonino et al. (44), therefore, tested the Lewis acid centers in TS-1 with the infrared spectra of adsorbed CD3CN in comparison with those observed for CD3CN on pure silicalite-1. The corresponding spectra are shown in Fig. 12. On adsorption of CD3CN on silicalite-1, a C – N stretching band grows at 2276 cm21 at low equilibrium pressure followed by a second band at 2265 cm21 as the pressure increases. These bands are attributed to CD3CN that is hydrogen bonded to SiOH groups (Section II.B.1) and physically adsorbed molecules, respectively. The same two bands are detected when CD3CN is adsorbed on TS-1, together with an additional band at 2302 cm21 which characterizes the most stable adsorbed species. The high C – N stretching frequency signals the highest adsorption bond energy, with the CD3CN molecule being coordinated to a Ti4þ ion:

ð1Þ

The inset in Fig. 12 shows the effect of the adsorption of CD3CN on the 960 cm21 framework band of TS-1, which clearly shifts to higher wavenumber with increasing CD3CN loading. This observation is a strong evidence of the tetrahedral Ti4þ ions in the silicalite framework acting as Lewis acid sites, which can undergo an expansion of their coordination sphere from a coordination number of four to a coordination number of five, as indicated in Eq. (1). Bonino et al. (44) reported supporting evidence for the Lewis acid character of the tetrahedral Ti4þ ions by using pyridine as an alternative probe. Furthermore, quantum chemical calculations were fully consistent with the conclusions drawn from the infrared spectra of the adsorbed probe molecules. Zecchina et al. (135) were unable to detect coordination of CO on Ti4þ centers at 77 K. A possible explanation for the apparent discrepancy between this result and those stated above may be the steric shielding of the tetrahedral Ti4þ by

30

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Fig. 12. Background-substracted spectra at increasing coverage of CD3CN on TS-1 (top) and silicalite-1 (bottom), n(CN) region. The spectra obtained at high CD3CN coverages are reported with the bold line. The inset reports the perturbative effect of CD3CN on the 960 cm21 band; the pure TS-1 spectrum is reported with a dotted line, although the bold line reports the spectrum obtained at high CD3CN coverage [Reprinted from Bonino et al. (44) with permission. Copyright (2003) American Chemical Society].

the oxygen ligands despite the larger size of Ti4þ relative to Si4þ. At 77 K, the vibrational motions of the TiO4 moiety are likely frozen, and the oxygen ligands may, therefore, not allow a close approach of the very weak base CO to the Ti4þ center. In contrast, the stronger bases acetonitrile and pyridine may overcome the steric barrier at the temperature of the experiments (room temperature). Infrared spectra of pyridine adsorbed on dehydrated TS-1 and Ti-MCM-41 of comparable Ti content indicated the presence of only Lewis acid sites (Fig. 13). The infrared absorptions at 1595 and 1445 cm21 are attributed to hydrogenbonded pyridine (Si/Ti – OHzzzpyridine) and those at 1580 and 1485 cm21 to pyridine bonded to weak Lewis acid sites (Fig. 12). Brønsted sites, if present,

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

31

Fig. 13. FTIR spectra of pyridine adsorbed on dehydrated TS-1 and Ti-MCM-41 [from Srinivas et al. (152)].

should show pyridinium ion peaks at 1639 and 1546 cm21, and strong Lewis acid sites should give rise to bands at 1623 and 1455 cm21 (141,150,151). The infrared bands disappeared as temperatures were increased beyond 398 K for TS-1 and 523 K for Ti-MCM-41, indicating higher acid strength in the latter than in the former titanosilicate. Furthermore, the number of acid sites (estimated from infrared peak intensities) is higher on Ti-MCM-41 than on TS-1. The temperature-programmed desorption of NH3 from these samples showed a desorption peak maximum at 448 K (Fig. 14). The peak is broader and more asymmetric when the sample is Ti-MCM-41. The amount of NH3 desorbed is 1.3 times higher for Ti-MCM-41 than for TS-1. With the Ti4þ ions acting as Lewis acid centers, a strong interaction with ammonia and water with these centers is expected. There is in fact abundant spectroscopic evidence for the coordination of NH3 and H2O molecules to tetrahedral Ti4þ centers and for the corresponding expansion of their coordination spheres. Figure 15 shows the modification in the UV – visible spectra of TS-1, initially in vacuo, upon interaction with H2O (152). Evidence of the interaction of NH3, a stronger base, is also shown. The LMCT band (mentioned in Section II.A.3) undergoes a red shift of the edge as a result of the increase of the coordination sphere about Ti4þ ions. In TiO2, in which Ti is surrounded octahedrally by six O atoms in its first coordination sphere, the Ti4þO22 ! Ti3þO2 LMCT is also red shifted to lower wavenumbers (32,000 cm21). A stronger perturbation is obtained upon dosing of NH3, but the line shape of the UV –visible curve is

32

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Fig. 14. Temperature programmed desorption of NH3 profiles of TS-1 and Ti-MCM-41 [from Srinivas et al. (152)].

similar. It was, therefore, concluded (152) that the four-coordinated, framework Ti species in dehydrated samples of TS-1 increase their coordination number (to 5 or 6) on interaction with H2O (or NH3), thus forming Ti(H2O)xO4 (or Ti(NH3)xO4) species with x ¼ 1 or 2. Bolis et al. (43) reported volumetric data characterizing NH3 adsorption on TS-1 that demonstrate that the number of NH3 molecules adsorbed per Ti atom under saturation conditions was close to two, suggesting that virtually all Ti atoms are involved in the adsorption and have completed a 6-fold coordination: Ti(NH3)2O4. The reduction of the tetrahedral symmetry of Ti4þ ions in the silicalite framework upon adsorption of NH3 or H2O is also documented by a blue shift of the Ti-sensitive stretching band at 960 cm21 (43,45,134), by a decrease of the intensity of the XANES pre-edge peak at 4967 eV (41,43,134), and by the extinction of the resonance Raman enhancement of the 1125 cm21 band in UV –Raman spectra (39,41). As an example, spectra in Figs. 15 and 16 show the effect of adsorbed water on the UV – visible (Fig. 15), XANES (Fig. 16a), and UV – Raman (Fig. 16b) spectra of TS-1. Appendix A summarizes what we believe to be the basic “fingerprint” features for the isomorphous substitution of Ti in silicate-1 lattice.

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

33

Fig. 15. UV –visible spectra of a TS-1 catalyst in vacuo (solid line) and upon interaction from the gas phase with H2O (dashed line) and NH3 (dotted line) [from Armaroli et al. (136)].

III. Oxo-Titanium Species and Reactive Intermediates Although the identification of tetrahedrally coordinated, tetra- and tripodal Ti4þ ions on the surface of titanosilicates, as the likely active sites in reactions that require Lewis acidity, seems convincing, the structure and role of the sites active in catalytic oxidation, presumably oxo-titanium species, formed by the interaction of H2O2 (or H2 þ O2) with these surface Ti ions, are not clear. In recent years, this problem has been investigated by FTIR (133), Raman (39,40), XANES (46 –48), electronic (54 – 57), and EPR (51– 54) spectroscopies. This is one of the areas in which major progress has been made since the reviews of Notari (33) and Vayssilov (34). Zecchina et al. (153) recently summarized some of the salient features of this progress.

34

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Fig. 16. Effect of soaking TS-1 with water on the XANES (a) and UV–Raman (b) spectra: dried TS-1 (solid line); soaked TS-1 (dotted line). The inset in part (a) reports the k3 -weighted, phaseuncorrected Fourier transforms of the corresponding EXAFS spectrum [Reprinted from Ricchiardi et al. (41) with permission. Copyright (2001) American Chemical Society].

III.A. UV– Visible Spectroscopy The color of an aqueous solution of Ti4þ in H2O2 depends on the pH, being orange in acidic solutions, yellow in neutral solutions, and colorless in strongly alkaline solutions. The yellow species contains one peroxy group for each Ti ion (154). The formation of a yellow color when TS-1 is brought in contact with H2O2 and its disappearance during the hydrocarbon oxidations has been known for a long time. DRUV – visible spectroscopy has confirmed the formation, upon contact of TS-1 with H2O2/H2O solutions, of a new LMCT band at about 385 nm (26,000 cm21, Fig. 17) corresponding to a charge transfer from the peroxide moiety to the Ti center (42). Hence, this UV – visible light-absorbing species (a peroxo moiety interacting with framework Ti ions) must be involved in the oxidation reaction. The yellow color produced by aqueous H2O2 progressively loses its color with time (153) (Fig. 17). The intensity, however, is nearly restored upon addition of pure H2O to the system, and this observation highlights the cooperative role of water in the stabilization of the Ti(O2) complex.

III.B. Vibrational Spectroscopy Vibrational frequencies of some titanium peroxo complexes and of solids containing peroxo and/or superoxo species are summarized in Table III. The three infrared vibrations of the triangular peroxo group in the C2v structure

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

35

Fig. 17. Evolution of the UV –visible spectra of a TS-1 catalyst brought in contact with an aqueous solution of H2O2 as a function of time: 1 min, 4, and 8 h (curves 1, 2, and 3, respectively). Curve 4 shows the effect of H2O dosage on the catalyst sample after acquisition of spectrum 3 [Reproduced from Zecchina et al. (153) with kind permission of Kluwer Academic Publishers].

typically appear in the regions 800– 950 cm21 (n(O– O)), and 500 –650 cm21 (n(M –O) symmetric and antisymmetric stretching) (155,156), the exact band positions being strongly dependent on the nature of the central atom. The O –O stretching mode of superoxo groups has been detected in the range of 1020 – 1220 cm21 for the typical end-on configuration on CoO – MgO solid solutions (157). Although Oz2 2 species have been detected on titanium-containing silicalites TABLE III IR spectroscopy of peroxo and superoxo species Dioxygen species

n (O–O)

n (M–O)s,as

(Pic)2TiO2HMPA

O22 2

895

575, 615

H2O2

(155)

(OEP)TiO2

O22 2

895

595, 635

H2O2, O2

(155)

Ca12Al10Si4O35

O22 2

895

O2

(156)

Ca12Al10Si4O35

Oz2 2

1075

O2

(156)

Compound

Oxygen source

Reference

Pic, pyridine-2-carboxylate; HMPA, hexamethylphosphoric triamide; OEP, octaethylporphyrin.

36

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

by EPR spectroscopy (Section III.E), the corresponding O – O stretching vibration has, to the best of our knowledge, never been reported. The lack of such reports may possibly be a consequence of the low sensitivity of infrared and Raman spectroscopy and an overlap of the O –O stretching band with the 1125 cm21 band of TiO4 tetrahedra. Infrared absorption of an unstable hydroperoxo species had been observed at 230 K by Tozzola et al. (63). A peak at 886 cm21, strongly overlapping the peak at 877 cm21 attributed to physisorbed H2O2, was attributed to TiOOH (h1 ; endon coordination), although a band at 837 cm21 was assigned to anionic triangular Ti(O2) (side-on coordination). Lin and Frei (133), upon loading of aqueous H16 2 O2 into TS-1 and removal of the solvent by evacuation, detected a peroxidic O – O stretch absorption at 837 cm21 and a broad band at 3400 cm21 by infrared difference spectroscopy. The former absorption shifted to 793 cm21 when aqueous H18 2 O2 was loaded in 21 TS-1 instead of H16 2 O2 (Fig. 18). No bands were observed at 837 or 3400 cm with the same loading of H2O2 on silicalite-1. Lin and Frei (133) assigned the 3400-cm21 band (Fig. 18) to hydrogen-bonded OH groups of TiOOH, and the two infrared bands were suggested to originate from a side-on hydroperoxo species (h2 -Ti(O2H) interacting with framework Ti (Scheme 1). The large red shift of the O –O stretching band (from 877 cm21 for physisorbed H2O2 to 837 cm21 for the strongly attached species) was claimed to be a result of the hydroperoxo group’s being covalently linked to the Ti center (133). This h2 -Ti(O2H) group was found to be indefinitely stable at room temperature. It was suggested that the exposure of dehydrated TS-1 to H2O2 led (133) to the conversion of the tetrapodal framework Ti to (SiO)3TiOOH (Scheme 1).

18 Fig. 18. Infrared difference spectra before and after loading of H16 2 O2 (curve a) and H2 O2 (curve b) into TS-1 followed by 12 h evacuation (1025 mbar) [Reprinted from Lin and Frei (133) with permission. Copyright (2002) American Chemical Society].

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

37

Scheme 1.

The very large bandwidth and red shift of nOH of the hydroperoxo group was postulated to be evidence of hydrogen bonding to the oxygen of the Si – OH moiety formed by cleavage of the Ti – O – Si linkage (Scheme 1). In the case of the tripodal framework (SiO)3Ti – OH centers, substitution of OH by OOH rather than opening of Si –O – Ti bridges was thought to occur. Hence, independent of whether H2O2 reacts with tetra- or tripodal framework Ti, the result is the same, namely, the formation of a TiOOH moiety adjacent to a Si –OH group. When the DRUV difference spectrum of the H2O2-loaded TS-1 sample was recorded after photolysis at 355 nm, it showed clearly the growth of a LMCT band with a maximum at about 360 nm and a tail extending to 550 nm (Fig. 19).

Fig. 19. Diffuse reflectance difference spectrum of the LMCT absorption upon 355 nm photolysis of TS-1/TiOOH molecular sieve (20 min at 45 mW cm22) [Reprinted from Lin and Frei (133) with permission. Copyright (2002) American Chemical Society].

38

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

The red shift of this band from its position in the dehydrated sample (Section II.A.3) is attributed to the increase of the coordination sphere about Ti4þ ions and is similar to the changes observed on adsorption of H2O and of NH3 (153). The simultaneous observation of the 837 and 3400 cm21 bands in the infrared region (attributed to peroxidic O –O, Fig. 18) and the 360 nm band in the DRUV spectra (attributed to octahedrally coordinated Ti4þ ions, Fig. 19) further confirms that the Ti4þ ions in the side-bound Ti(O2H) species are indeed 6-fold coordinated. When the H2O2-loaded TS-1 sample was irradiated with 355-nm light of a Nd:YAG laser or the visible emission of a conventional tungsten source, photodissociation of TiOOH was observed (133). The 837 and 3400 cm21 bands (and the corresponding 18O substitutes) diminished in intensity (Fig. 20). The loss of the 837 and 3400 cm21 bands was accompanied by the growth of bands at 3676 cm21 (assigned to O –H), 1629 cm21 (assigned to the bending mode of H2O), and 960 cm21 (assigned to Si – O – Ti), indicating at least partial restoration of the original coordination environment of the metal center (Scheme 1). The net result of the photodissociation is the disproportionation of TiOOH to TiOH and O and the further condensation of this TiOH with adjacent SiOH to regenerate Ti –O –Si and H2O. The lack of Ti leaching in TS-1 during catalytic oxidations was attributed to such recondensation of the Ti – O – Si linkages. The structure of the peroxide species in the TS-1 catalyst was also investigated by resonance Raman spectroscopy (39,42). Interaction with H2O2 caused (i) a reduction and blue shift (to 976 cm21) of the 960-cm21 band, (ii) a quenching of the 1125 cm21 band in the UV – Raman spectrum as a result of the breakdown of the tetrahedral symmetry, (iii) the appearance of a strong and sharp band at 875 cm21 (attributed to O –O stretching in physically adsorbed H2O2), and (iv) the appearance of a strong and complex new feature centered at 618 cm21. The 618 cm21 band was assigned to a resonance Raman enhanced vibration mode of the titanium peroxo complex. On the basis of the similarity between the spectroscopic features in both the UV –visible and Raman spectra of 32 (NHþ and TS-1/H2O2 systems, Bordiga et al. (42) concluded that 4 )3(TiF5O2)

Fig. 20. Infrared difference spectra before and after 20 min irradiation (with 355 nm light (45 mW cm22)) of aqueous H16 2 O2 loaded TS-1 molecular sieves [Reprinted from Lin and Frei (133) with permission. Copyright (2002) American Chemical Society].

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

39

the species responsible for the 385 nm LMCT band is a side-on titanium peroxo species which is also characterized by a Raman mode at 618 cm21. The presence 32 of side-on (O2) attachment in the TiF5(O2) molecular unit of (NHþ , 4 )3(TiF5O2) in particular the Ti(O2) fragment, is known (42).

III.C. X-Ray Absorption Spectroscopy XANES and EXAFS spectroscopies were applied by Zecchina et al. (153) to investigate the changes in coordination of the framework Ti ions in TS-1 on contact with H2O, NH3, and a mixture of H2O þ H2O2 (Fig. 21). There is a progressive reduction in the pre-edge intensity on going from H2O to NH3 to H2O þ H2O2, indicating the transition from four to six coordination (Section II.B.2). Their EXAFS results suggested the formation of a strongly adsorbed side-on peroxo complex in which both the O atoms are located at a Ti – O ˚ . Presumably, the formation of this complex is accompanied distance of 2.01 A

Fig. 21. XANES spectra of TS-1 catalyst in vacuo and upon interaction with H2O (from the liquid phase), NH3 (from the gas phase), and H2O/H2O2 (liquid solution) [Reproduced from Zecchina et al. (153) with kind permission of Kluwer Academic Publishers].

40

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Scheme 2.

by the hydrolysis of one or even two Ti – O – Si bonds and the total deprotonation of H2O2 (153). Ti(O2) and Ti(O2H) species formed on Ti " MCM-41 during reaction were studied by using XANES and EXAFS measurements and density functional theory (DFT) (36,46,48,104). Investigating the nature of titanium sites on catalysts obtained by grafting titanocene dichloride on MCM-41 (Ti " MCM-41), the authors found that in the “free”, dehydrated state, these sites consist mostly of Ti4þ –OH groups tripodally anchored to the silica via covalent bonds to oxygen. In addition to these tripodal, single-site, titanol centers, there were also bipodal Ti4þ centers present in the as-prepared Ti " MCM-41 catalysts. Their proposed models of the tetrahedral tri- and bipodal species are illustrated in Scheme 2. There were no signs of Ti –O –Ti linkages, nor of any titanyl (TiyO) groups, nor of a three-, five-, or six-coordinated species. Under reaction conditions when cyclohexene and tert-butylhydroperoxide (TBHP) were brought in contact with these catalysts, there was a decrease in the pre-edge intensity of the XANES, in comparison with the intensities characterizing the calcined and dehydrated catalysts, indicating that the coordination about the Ti ions increases on contact with the oxidant/reactant. Considering both the intensity and position of the pre-edge peak (the energy position of the peak after interaction with the TBHP was slightly higher), the authors ruled out the presence of a five-coordinated Ti species. The expansion in coordination was from four to six. Furthermore, whereas four of the surrounding oxygen atoms are at distances strictly com˚ ), in the reactive parable to those in the pristine surface structure (about 1.81 A ˚ ). state there are two additional oxygen atoms situated farther away (2.2 – 2.4 A The EXAFS data characterizing the (catalyst þ TBHP þ alkene) system also ˚ (a slight indicated that there are at least three Ti – O distances close to 1.83 A expansion compared to the “free” surface), and two of the other three oxygen ˚ . From among different models of the distances were between 2.2 and 2.4 A titanium oxo species investigated, the authors concluded that the Ti-h2 -OOR

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

41

Fig. 22. Best fit between experimental results and computed EXAFS employing the full multiple scattering method. The model is depicted in the bottom right figure [Reprinted from Thomas and Sankar (104) with permission. Copyright (2001) American Chemical Society].

and Ti-h1 -OOR structures (where R is H or alkyl) gave the best fits between the experimental and computed EXAFS data (Fig. 22). III.D. Cyclic Voltametry The presence of two types of titanium sites in TS-1 (tetra- and tripodal) was also suggested by the cyclic voltametry experiments of Bodoardo et al. (158). The tripodal Ti(OSi)3(OH) showed a redox couple at 0 V and the tetrapodal Ti(OSi)4

42

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

a redox couple at 2 0.6 V, indicating that the electron density is higher in the tripodal than in the tetrapodal structure. The higher electron density at Ti, in turn, will increase the electron density at the O –O bond attached to it, facilitating the cleavage of the latter. The ease of cleavage of the O –O bond will influence the mode of its cleavage, homo- or heterolytic. Product selectivity in H2O2-catalyzed reactions of course depends strongly on the mode of cleavage (homo or heterolytic) of the O – O bond, as discussed in detail in Section VI.

III.E. EPR Spectroscopy Superoxide species, Oz2 2 , were observed by Zhao et al. (50) by EPR spectroscopy on contact of TS-1 with H2O2. Two types of superoxides were identified, a major species with gzz ¼ 2:0236; gyy ¼ 2:0100; and gxx ¼ 2:0091; and a minor species differing only in its gzz value which was 2.0270 in contrast to 2.0236. The major signal was assigned to superoxides on framework titanium sites and the weaker signal to those on dispersed, extra-framework titanium sites. The superoxide attached to the framework Ti was also less stable, decomposing completely within a few hours. The second signal, assigned to the superoxide on nonframework Ti, was more stable. When a drop of phenol in acetone solution was wetting TS-1, the lines of the superoxide species on framework Ti disappeared and a new intense signal attributed to phenoxy radicals appeared. It was suggested that the appearance of the phenoxy radical along with the disappearance of the superoxide on framework titanium sites provided direct support for a free radical mechanism of oxidation. The formation of paramagnetic oxygen species as a result of interaction of H2O2 or H2 þ O2 with titanosilicates was also investigated by Ratnasamy et al. (51,52,54) using a combination of UV – visible and EPR spectroscopies. The diamagnetic peroxo/hydroperoxo species (TiO2H) could be discerned by their UV – visible spectra, and the concentration of the paramagnetic superoxo species (Ti(Oz2 2 )) was independently estimated from their EPR spectra. Two types of Ti4þ-superoxo species, A and B (A being preponderant), were detected in TS-1 and Ti-beta. Ti-MCM-41 contained mainly species B (Fig. 23). An additional species, C, was detected upon interaction of TS-1 with the (H2O2 þ urea) adduct or palladium impregnated TS-1 (Pd-TS-1) with H2O2. EPR spectroscopy also provided evidence, for the first time, for the in situ generation of similar oxo species in reactions using H2 þ O2 instead of H2O2 as the oxidant. The titanium sites adjacent to Pd ions (in Pd-TS-1) behave magnetically differently from the other Ti ions, generating a greater variety of superoxo species. Pd (as expected) was found to facilitate the reducibility of Ti4þ ions and promoted the formation of the diverse titanium oxo species at lower temperatures (about 323 K). In the absence of H2, exposure of TS-1, Ti-MCM-41, Pd-TS-1, or Pt-TS-1 to O2 alone

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

43

Fig. 23. EPR spectra (at 210 K) of titanosilicates interacting with aqueous H2O2; the gzz region at higher gain (£ 5) is shown. The peaks corresponding to A0 , A, and B-type Ti-superoxo species are indicated [(from Srinivas et al. (52)].

does not generate the superoxo species. When Pd(Pt)-TS-1 samples were brought in contact with H2 þ O2, Ti4þ was reduced to Ti3þ by H2 (Fig. 24). The Ti3þ ion (characterized by its typical EPR spectrum) generates Ti(Oz2 2 ) species on interaction with O2. This reduction and reoxidation of Ti ions, which requires 473 K or higher temperatures in TS-1, is facilitated by Pd or Pt and even occurs at 323 K (Fig. 24). The superoxo species generated are more of A (and A0 ) types (Table IV and Fig. 24). The extent of Ti4þ reduction and Ti(Oz2 2 ) formation depends on the Pd content, with the concentration of the paramagnetic titanium oxo species reaching maximal values at 2 wt% Pd (54). There has been an attempt to estimate the relative concentrations of the two superoxo and hydroperoxo species (54) by deconvolution into two bands of the broad UV –visible band observed on reaction of titanosilicates with aqueous

44

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

3þ Fig. 24. EPR spectra of Ti(Oz2 ions at 80 K. (a) Pd(2)-TS-2 þ H2O2; (b) Pt(0.015)2 ) and Ti TS-1 þ H2 þ O2 (treated at 673 K); (c) Pd(2)-TS-1 þ H2 þ O2 (treated at 323 K); and (d) TS1 þ H2 þ O2 (treated at 673 K). For clarity, spectra (c) and (d) are shown at four and five times the 3þ actual gain. Spectral regions corresponding to Ti(Oz2 ions are marked [from Shetti et al. (54)]. 2 ) and Ti

H2O2 or non-aqueous urea – H2O2 adducts (Fig. 25). Bands I and II were attributed to the charge transfer transitions associated with Ti(Oz2 2 ) superoxide and Ti(O2H) hydroperoxo/peroxo species, respectively. The position and relative intensity of these two bands are different in TS-1 and Pd-TS-1. The intensity ratio (Ti(O2H))/Ti(Oz2 2 )) was higher for Pd-TS-1 than TS-1. In the spectrum of Ti-MCM-41, these bands overlapped with those assigned to the H2O2-free solid. The conversion energy for the hydroperoxo –superoxo transformation was estimated from the DRUV – visible band positions in (TS-1 þ H2O2), (Pd-TS1 þ H2O2), and (TS-1 þ (urea þ H2O2)) to be 38.8, 46.0, and 56.4 kJ/mol, respectively. At 298 K, for the (TS-1 þ H2O2) system, the Ti(O2H)/Ti(Oz2 2 ) ratio was found to be 0.66. A comparative value of this ratio was also computed from EPR measurements (52). The line labeled “theoretical” passing through the origin in Fig. 26 was computed on the assumption that all the Ti ions in the sample react with H2O2 forming only the paramagnetic superoxo species. The line labeled “experimental” in Fig. 26 shows that the intensity of the EPR signal varies linearly with

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

45

TABLE IV EPR parameters (at 77 K) for the superoxo-Ti(IV) species generated on titanosilicates by contacting with aqueous H2O2 (HP), urea-H2O2 adduct (UHP) and (H2 þ O2) Systema TS-1 þ HP

Ti-MCM-41 þ HP Pd(2)-TS-1 þ HP

gzz

gyy

gxx

D (cm21)b

A

2.0264

2.0090

2.0023

11203

B

2.0238

2.0090

2.0023

12558

2.0244

2.0095

2.0031

12217

Species

B 0

2.0309

2.0100

2.0350

9440

A

2.0276

2.0100

2.0350

10672

A00

2.0265

2.0100

2.0350

11157

B0

2.0255

2.0100

2.0350

11638

B

2.0245

2.0100

2.0350

12162

C

2.0220

2.0100

2.0350

13705

A0

2.0300

2.0101

2.0035

9747

A

2.0275

2.0101

2.0035

10715

B

2.0242

2.0101

2.0035

12329

C

2.0206

2.0101

2.0035

14754

Ti-MCM-41 þ UHP

B

2.0232

2.0096

2.0046

12919

TS-1 þ H2 þ O2

A

2.0265

2.0080

2.0010

11157

Ti3þ

1.930

1.956

1.956

A0

2.0340

2.0092

2.0022

8517

A00

2.0295

2.0092

2.0022

9926

B

2.0241

2.0092

2.0022

12385

Ti3þ

1.928

1.953

1.953

A0

2.0300

2.0080

2.0012

9747

A000

2.0295

2.0080

2.0012

9890

B

2.0241

2.0080

2.0012

12385

Ti3þ

1.931

1.955

1.955

TS-1 þ UHP

Pd(2)-TS-1 þ H2 þ O2

Pt(0.015)-TS-1 þ H2 þ O2

A

Adapted from Shetti et al. (54). a Pd(2)-TS-1 and Pt(0.015)-TS-1 correspond to TS-1 samples impregnated with 2 wt% Pd and 0.015 wt% Pt, respectively. b D is the energy separation between the oxygen pxg and pyg orbitals.

46

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Fig. 25. DRUV–visible spectra of TS-1, TS-1 þ H2O2, TS-1 þ urea –H2O2, and Pd(2)-TS-1 þ H2O2. Bands characterizing superoxo (I) and hydroperoxo (II) species are marked. Experimental (—), simulated ( – – –), and deconvoluted oxo-titanium bands ( –· –·–) are shown [from Shetti et al. (54)].

Fig. 26. Total EPR signal intensity as a function of Ti content in TS-1 samples [Srinivas et al. (52)].

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

47

the Ti content in the various TS-1 samples. This line, however, does not pass through the origin (Fig. 26). If all the Ti ions in TS-1 had formed the paramagnetic Ti-superoxo species, the experimental line would have passed through the origin and coincided with the theoretical line. All the Ti ions in the chosen samples (Si/Ti ¼ 30, 60, and 80) were isolated and in framework positions (as shown by XRD, FTIR, and UV – visible analyses). Thus, they are expected to interact with H2O2 and form either paramagnetic superoxo or diamagnetic peroxo-Ti species. Consequently, it is concluded that only a fraction of the Ti ions form paramagnetic superoxo-Ti species and the rest form diamagnetic hydroperoxo/peroxo-Ti species. From the difference in the theoretical and experimental EPR intensity values (Fig. 26), the amounts of Ti-hydroperoxo and Ti-superoxo species were estimated to be 45 and 55%, respectively, at 80 K. This estimate of the (Ti(O2H)/Ti(Oz2 2 ) ratio ¼ 45/55 ¼ 0.82 is in reasonable agreement with the value of 0.66 based on DRUV data. An additional, independent estimate of the concentration of paramagnetic superoxo and diamagnetic hydroperoxo-/peroxo-titanium species was made from magnetic susceptibility measurements using a Lewis coil force magnetometer (52). The gram-susceptibility of Ti in TS-1 þ H2O2 was estimated to be 5.5 £ 1026 emu/g, which corresponds to an effective magnetic moment of 0.79 B.M. If all the Ti ions in the sample had formed superoxo species upon interaction with H2O2, the effective magnetic moment should have been 1.73 –1.78 B.M. The concentration of superoxo-Ti species is, thus, about 45% of the total Ti, comparable to the values found by EPR (55%) and electronic spectroscopies. The remaining fraction is, presumably, the diamagnetic hydroperoxo-/peroxo-Ti species. H2O2 can be a potential source of many radicals (e.g., OH, O2H, etc.). However, EPR spectroscopy did not reveal the presence of any of these radicals, indicating that their concentrations are not very significant. They may be highly unstable. Thus, their contribution to the total magnetic susceptibility is apparently negligible. The conversion of hydroperoxide/peroxide to superoxide is a one-electron redox reaction and requires the presence of transition metals having accessible multiple oxidation states as in biological iron or manganese clusters (e.g., Fe(II, III, IV) clusters of monooxygenase or the Mn(II, III, IV) clusters of photosystems). Ti is usually not reduced at ambient temperatures. The various possibilities that could facilitate the transformation of hydroperoxo/peroxo to superoxo species are as follows: 1. Homolysis of H2O2 to HOz radicals, which react with hydroperoxo-Ti species to form superoxo-Ti and H2O: H2 O2 ! 2HOz z

Ti – OOH þ HO !

TiðOz2 2 Þ

ð2Þ þ H2 O

ð3Þ

48

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Formation of HOz radicals by decomposition of H2O2 on contact with titanium silicates increases with temperature. At 77 K, this decomposition is less probable. 2. The second possibility is the dismutation of two superoxo ions to yield the peroxo species. z2 22 Oz2 2 þ O2 ! O2 þ O2

ð4Þ

Again, even if mobile superoxide ions were present in the material, they would not be able to diffuse at the low temperatures used for the EPR experiments (190 –77 K). 3. The third possibility for the conversion of the superoxide to the peroxide is the homolytic opening of a cyclic peroxo species (more precisely, 3þ 3þ z2 Ti4þ(O22 2 ) to Ti (O2 )), as proposed by Notari (33). Formation of Ti species was indeed observed in the presence of a base, such as NaOH (spectrum not shown), but in neutral or acidic conditions, the Ti3þ species was not observed. Either their concentration, if they were formed, was very low or they were short-lived. 4. The concentration of the Ti(Oz2 2 ) species is solvent dependent. Thus, the solvent (or H2O) may play the role of a redox partner. The HOz radicals, generated from the decomposition of H2O2, perhaps cause the hydroperoxo/peroxo to superoxo conversion. The superoxo species (with the O –O stretching absorption near 1120 – 1150 cm21) could not be seen in the FTIR spectrum (63), perhaps because of the dominant stretching and bending modes of water in the same region. Although the Ti(O2H) hydroperoxide may be reasonably identified with the corresponding species derived from infrared – Raman and XAFS spectroscopies mentioned above, the nature of the paramagnetic superoxide ion-radical, Ti(Oz2 2 ), seen in the EPR spectra, merits more elaboration. Shetti et al. (54) proposed tentative structures A, B, and C arising from the tetrahedral TiO4 units upon interaction of the sample with H2O2 (Scheme 3). Species A was postulated to arise from the framework substitutional sites in the MFI lattice and B and 2 C from the defect sites. The free Oz2 2 radical, with a P ground state, has a 2 2 2 2 2 4 3 (1sg) (1su) (2sg) (2su) (3sg) (1pu) (1pg) electronic configuration. Interaction with Ti removes the degeneracy of the HOMO pg into pxg and pyg orbitals with an energy gap of D: Neglecting the second-order terms, the g value expressions (when l , D p E) may be written as follows (159): gzz ¼ ge þ 2l=D

ð5Þ

gyy ¼ ge þ 2l=E;

ð6Þ

gxx ø ge ;

ð7Þ

and

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

49

Scheme 3.

where ge ¼ 2:0023; l is the spin –orbit coupling constant (135 cm21 for oxygen), and E is the energy separation between 3sg and 1pgx orbitals. The gzz value of the superoxo anion is sensitive to the oxidation state, coordination number, and local geometry of the cation to which it is coordinated. (Ti –(Oz2 2 ) distances also influence the gzz parameter. The stronger the Ti – O bond, the lower the gz value of the superoxo anion. Using the above expressions and the experimental gzz value, Shetti et al. (54) estimated the separation between the pxg and pyg orbitals ðDÞ (Table III). The gzz values of various (Ti – (Oz2 2 )) species decrease in the order A . B . C. The D (Oz2 2 ) values for the A type species lie in the range of 8520 – 11,200 cm21. Accordingly, the electron density in the O – O bond increases in the order A , B , C. Because this electron is added into the antibonding orbital, the strength of the O – O bond may be expected to decrease in the order A . B . C (Scheme 3). The O –O bond strength (in the oxo-Ti intermediate) is expected to play a significant role in influencing the nature of its cleavage (homolytic vs. heterolytic). Appendix B is a list of some of the major characteristics of the titanium oxo species generated on TS-1 as a result of contact with H2O2.

IV. Computational Investigations Significant progress has been made in the last few years in theoretical investigations of the geometry and coordination number of Ti ions in TS-1 and Ti-MCM-41, both in the dehydrated state and after interaction with H2O2 or TBHP (48,59– 63,103). When such investigations are combined with X-ray

50

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

absorption, infrared, UV – visible, Raman, and other spectroscopic results described in Sections II and III, an integrated picture of the structural identity of the active sites and reactive intermediates involved in the catalytic reactions of titanosilicates emerges. The various spectroscopic techniques had revealed that Ti4þ ions in TS-1, Ti-beta and, Ti-MCM-41 are 4-coordinate in the dehydrated state. Tetrapodal Ti(OSi)4 and tripodal Ti(OH)(OSi)3 are the main Ti species. Upon exposure to H2O, NH3, H2O2, or TBHP, they increase their coordination number to 5 or 6. On samples in which the Ti4þ has been grafted onto the silica (referred to as Ti " MCM-41), a dipodal Ti species (Ti(OH)2(OSi)2) may also be present. As a result of interaction with the oxidant ROOH (R ¼ H, alkyl), the formation of h1- and h2 -peroxo (Ti –O – O2), hydroperoxo (Ti – OOH), and superoxo (TiOz2 2 ) species has been observed experimentally (Section III). A linear correlation between the concentration of the h2 -hydroperoxo species and the catalytic activity for propene epoxidation has also been noted from vibration spectroscopy (133). Computational methods, especially DFT, have been used to elucidate the structure of the oxo-titanium species and their interactions with reactants such as ethene and NH3 (48,60). From a combined DFT and EXAFS investigation, Barker et al. (48) recently proposed that 6-coordinate hydrated Ti(h1 -OOR) and (h2 -OOR) complexes, where R ¼ H or tert-butyl, are the oxygen-donating species in peroxide/Ti " MCM-41 mixtures. The computed structural features of the h1 - and h2 -species are given in Table V. A schematic illustration of the two structures in the case of TBHP/Ti-MCM-41 is given in Fig. 27. Figure 28 shows the calculated energetic pathways from the bare active site and isolated peroxide to the h1 and h2 reactive oxo-intermediates. The calculated activation barriers are in each case about 40 kJ/mol. In addition to the monodentate h1 -Ti – OOH and bidentate h2 -Ti(O2H) complexes, a third type of oxo-intermediate h1 -Ti(O2H2) complex was also calculated to be feasible. The structures of these three Tiperoxo intermediates are shown in Fig. 29. (The calculations were done starting from the model of the tripodal Ti (Ti(OH)(OSi)3), as this was the predominant species in Ti " MCM-41. Similar calculations, more realistic for TS-1 and Ti-beta, starting from the tetrapodal Ti(OSi)4 will be of interest.) If the h1 - and h2 -hydroperoxo species are the oxygen-donating entities, the mode of their interaction with reactants such as alkenes is of interest. Cora` et al. (59) claimed, on the basis of a Mullikan population analysis, that the electronrich alkene double bond will preferentially interact with the most electrophilic oxygen atom, which was identified to be the one closest to Ti in the hydroperoxo species (h1 -TiOOH), because it has a lower net negative charge. Following a frontier orbital approach and comparing the energies of the HOMO and LUMO of the oxo intermediates with that of ethene, the authors found that for both h1 and h2 structures, the interaction between the LUMO of the catalyst and the HOMO of the alkene was, as expected, energetically more favorable

TABLE V Calculated and refined EXAFS parameters for six-coordinate Ti-h 2(OOH) and Ti-h 1(OOH) species in peroxide/surface grafted Ti " MCM-41 mixtures Ti–O distance ˚) (A Calculateda Ti-h 2(OOH)

1.92 P

2.25

Si

Ti-h 1(OOH)

Ti –Si distance ˚) (A

EXAFS parameter 1.91 2.20 Si

1.83

1.83

1.80Si

1.83Si

1.80Si

1.83Si

2.26W

2.43

2.24

2.20

P

1.97

Si

1.81

Si

1.97 Si

1.83

Si

1.84

1.83

1.81Si

1.83Si

2.35W

2.43

Ti –O–Si (Ti– O– OH) angle (8)

Calculateda

EXAFS parameter

Calculateda

3.35

3.38

151.3

160

3.32

3.30

145.5

148

3.28

3.21

143.0

139

(81.9)

(80)

R-factor (EXAFS)

EXAFS parameter

3.31

3.28

145.3

144

3.34

3.38

146.8

152

3.34

3.39

151.5

163

(117.3)

Eformation (Calculated)a (kJ/mol)

245

16.02

2102

16.18

(120)

Adapted from Barker et al. (48). Superscript characters: P, Ti –peroxide bond length; Si, Ti –OSi bond length; W, The Ti –O bond distance of Ti to water molecule. Calculated by the BP86/DZVP procedure employing a larger model cluster extending three-coordination spheres from the central Ti ion. a Eformation ¼ Etotal (“extended” Ti-h 1(OOH) þ other products) 2 Etotal (“extended” tripodal TiIV cluster þ H2O2 þ 2H2O).

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Cluster

51

52

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Fig. 27. Ti-peroxo species in TBHP/Ti " MCM-41 catalysts. All distances (DFT calculated values ˚ [Reproduced from Barker et al. (48) by and experimental parameters (in parentheses)) shown are in A permission of the PCCP Owner Societies].

than the inverse interaction of the LUMO of the alkene and the HOMO of the catalyst. Further, the LUMO –HOMO gap for propene was approximately 50 kJ/mol lower than for ethene, suggesting a higher reactivity of propene, as indeed was observed experimentally (Section V). Figure 30 illustrates this interaction for the three p-peroxo species. In each case the starting geometries for modeling were obtained by orienting the ethene molecule so that its HOMO overlaps with the LUMO of the catalyst. The interaction of ethene with all the three peroxo species is exothermic. In the case of the side-bound h2 intermediate, the interaction was initiated (in the calculations) by positioning the double bond parallel to the peroxide

Fig. 28. Calculated energetic pathways from the bare active site and isolated peroxide to the h1 - (left) and h2 - intermediates (right) [from Cora` et al. (59)].

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

53

Fig. 29. Geometry-optimized structure of the three stable Ti-peroxo intermediates: (a) h1 monodentate complex, (b) h2 -bidentate complex, and (c) h1 -O2H2 complex [from Cora` et al. (59)].

molecule, because the OH ligand hinders other directions of attack of ethene molecule on the peroxidic oxygen closest to Ti. Optimization of this structure leads to an alcohol-type functionality (Fig. 30b), which the authors suggested (59) to be possibly responsible for the formation of the diol products observed experimentally (Section V). The Ti4þ distribution in TS-1 has also been studied by computational methods (34,62,160 –163). The actual location of the Ti atoms in the framework of titanosilicates is difficult to determine experimentally because of the low Ti content (Section II), and information obtained from theoretical methods is, therefore, of considerable interest. In the orthorhombic MFI structure, substitution can take place at 12 crystallographically different tetrahedral (T) sites (T1 –T12) (Fig. 1 and Section II.A.1.b). In the monoclinic MFI framework, the mirror symmetry is lost and 24 crystallographically different T sites can be distinguished (Fig. 31) (160). Although all computational investigations that have been reported confirm that Ti atoms are incorporated in the framework at regular Ti-sites, there is still controversy about the exact siting of the Ti atoms in the MFI structure. De Man and Sauer (62) by ab initio investigations found only small subsitution energy differences among the various T sites, and this result implies that Ti atoms are distributed over all the lattice positions rather than being located at one preferred T-site. Using a combination of Metropolis Monte Carlo method and molecular mechanics calculations, Njo et al. (160) concluded that the Ti atoms are indeed distributed over all the crystallographically different lattice positions rather than located at one preferred site. The distribution, however, is not equal or random. In Fig. 32 the Ti occupancies per unit cell for the orthorhombic and monoclinic structure are shown (160). In the orthorhombic structure, T12 is preferred, whereas in the monoclinic structures T2 is preferred. The framework symmetry (orthorhombic/monoclinic) is apparently related to both the location of the Ti atoms and the Ti loading. Njo et al. (160) also computed the occupancy of the different T sites at different loadings (Fig. 33). At all Ti loadings up

54

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Fig. 30. Calculated initial and final states for the interaction of “ethene” with (a) the h1 -, (b) the h2 -, and (c) h1 -O2H2 Ti-peroxo intermediate [from Cora` et al. (59)].

to 2.5 Ti atoms per unit cell, the experimentally determined upper limit for incorporation of Ti in lattice positions, the T2 and T12 sites were preferred. A 200-atom cluster study of Ti-siting in TS-1 by Atoguchi and Yao (162) using the ONIOM method (164), however, suggested that the most stable Ti substituted

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

55

Fig. 31. Crystallographically different T-sites in MFI. T1 (T2,…,T12) and T13 (T14,…,T24) are related by a mirror plane in orthorhombic MFI [Reprinted from Njo et al. (160) with permission. Copyright (1997) American Chemical Society].

T sites were T9 and T10 sites—if thermodynamics controls the structure of Ti-containing MFI zeolite. The stability sequence of T sites was found to be T9 . T10 . T12 . T1 . T6 . T5 . T3. The exact location of Ti ions in TS-1 is still controversial. There are no similar investigations for other Ti silicates.

V. Catalytic Properties V.A. Introduction The catalytic activity of the titanosilicate molecular sieves, especially those of TS-1, TS-2, Ti-beta and Ti-MCM-41 has been investigated extensively

Fig. 32. Ti distribution per unit cell over the crystallographically different T-sites for the orthorhombic structure (T1–T12, white) and monoclinic (T1– T12, stripes; T13–T24, black) structures [Reprinted from Njo et al. (160) with permission. Copyright (1997) American Chemical Society].

56

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Fig. 33. Population of crystallographically different T-sites for various Ti loadings: (i) one Ti atom per unit cell (white), (ii) one Ti atom per double unit cell (dotted), and (iii) eight Ti atoms per unit cell (striped) [Reprinted from Njo et al. (160) with permission. Copyright (1997) American Chemical Society].

(33,165 – 169). When a tetravalent ion, such as Ti4þ, replaces, the Si4þ in a silicate lattice isomorphously, the generation of Brønsted acidity is not anticipated. In fact, no experimental evidence exists for a purely Brønsted acidcatalyzed reaction in a well-synthesized and pure sample of TS-1 and in the absence of H2O2. Lewis acid-catalyzed reactions can, of course, occur because of the coordinatively unsaturated Ti ions, as mentioned above (Section II.B). The enormous interest in these materials is, however, due to their remarkable catalytic activities in oxidation reactions using the environmentally benign aqueous H2O2 as the oxidant. V.B. Reactions Using H2O2 as Oxidant V.B.1. General Features Oxidations of organic reactants using H2O2 as an oxidant have been known for a long time (170). Although H2O2 is a weak acid ðpKa ¼ 11:6Þ and a mild oxidant, a small amount of HOþ may be present in equilibrium with H2O2 solutions, especially at low pH: H2 O2 þ Hþ O H2 O þ HOþ

ð8Þ

The major use of H2O2 as an oxidant arises from its ability to insert an oxygen atom in an organic molecule (alkene, alkane, aromatic hydrocarbon, etc.) in the presence of some catalysts. In reactions using H2O2 as an oxidant, the type of

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

57

cleavage of the O –O bond (in H2O2) plays a crucial role in determining the product distribution. A homolytic cleavage generating radicals (such as HOz) usually leads to a product distribution different from the one that arises by heterolytic cleavage (generating HOþ and HO2, for example). The gas-phase dissociation enthalpy, DH 0 ; for O –O homolytic cleavage in H2O2 is 205 kJ/mol (171). The O – O bond is considerably weakened if H is replaced by electrondonating alkyl groups as in ROOH (R ¼ alkyl), the bond dissociation enthalpy being only 180 kJ/mol for the homolytic cleavage of the O – O bond in CH3OOH (171). A heterolytic cleavage of the O – O bond, HOOH ! HOþ þ HO2, requires a considerably higher dissociation enthalpy if the emerging ions are not stabilized. The enthalpy for the heterolytic O – O cleavage of H2O2 into HOþ and HO2 is 1252 kJ/mol (171) in the gas phase. The corresponding value for CH3O – OH ! CH3Oþ and HO2 is 775 kJ/mol. The situation is, however, different in solution. Heterolytic cleavage requires less energy if the dissociated ions form an ion pair in solution at a distance less than rip ; separating the effective charge centers. Then, the energy of heterolytic cleavage in a solvent, DEip (solvent) is given (171) by Eq. (9) DEip ðsolventÞ ¼ DE 2 e2 =rip 2 DEsolv ;

ð9Þ

where DE is the energy required for gas-phase heterolytic cleavage, rip ¼ 2:65 A (172,173), and DEsolv is the solvation energy given by DEsolv ø 14:39ðð1 2 1Þ=ð21 þ 1ÞÞm2 =a3 ;

ð10Þ

1 is the dielectric constant and m the dipole moment of the solvent, and, a is the radius of a spherical cavity formed by solvent molecules surrounding the ion pair. (173), the solvation energy of a typical hydrocarbon solvent With a ¼ 3:5 A ð1 ¼ 2Þ is about 45 kJ/mol (171). This energy will increase if the dielectric constant of the solvent is higher. Hence, as the dielectric constant/dipole moment of the solvent is progressively increased, the heterolytic fission of the O – O bond (in H2O2, TBHP, etc.) will be favored over homolytic fission. Because the latter generates radical intermediates and the heterolytic fission produces ionic products, it is likely that the oxidation reaction mechanism and product distribution will depend to some extent on the choice of the solvent, as indeed has been observed experimentally (vide infra). Homolytic decomposition increases at higher temperatures, especially temperatures above about 333 K. Radical pathways, hence, play a greater role in influencing product selectivity at higher temperatures and in non-polar solvents.

58

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

V.B.2. H2O2-Catalyzed Reactions in the Homogeneous Phase Reactions with H2O2 may be divided into two classes arising from the homolytic vs. heterolytic cleavage of the O –O bond (173). In homolytic catalysis, the oxygen-centered radicals are intermediates; the participation of concerted processes in heterolytic catalysis precludes paramagnetic intermediates. Product selectivity is usually higher in the latter class. Transition metal cations in low oxidation states, such as Cu1þ, Ti3þ, V2þ, Cr2þ, and Fe2þ, catalyze the homolytic route, although those in higher oxidation states, such as Mo6þ, W6þ, V5þ, and Ti4þ, catalyze the heterolytic cleavage. The one-equivalent, homolytic scission of peroxides may be either reductive (Eq. (11)) or oxidative (Eq. (12)): HOOH þ Mnþ ! HOz þ HO2 þ Mðnþ1Þþ

ð11Þ

HOOH þ Mnþ ! HOOz þ Hþ þ Mðn21Þþ

ð12Þ

An alternate homolytic cleavage is the following: HOOH ! 2HOz :

ð13Þ

The reductive cleavage (Eq. (11)) is more common. TBHP can also undergo preferential reductive cleavage to the alkoxyl radical: ROOH þ Cu1þ ! ROz þ Cu2þ ðOHÞ:

ð14Þ

The oxidative cleavage may be illustrated as follows: ROOH þ Co3þ ! ROOz þ Co2þ þ Hþ :

ð15Þ

Hydroxy radicals are intermediates in the reaction of Ti3þ and H2O2 (175). This system is also capable of hydroxylation of aromatics and alkanes but, in contrast to reactions with Fenton’s reagent (Fe2þ þ H2O2, reductive, homolytic cleavage, Eq. (11)), only non-chain processes are possible, because Ti4þ is not usually an oxidant. Hence, relatively high selectivities are feasible. Heterolytic catalysis is promoted by W6þ, Ti4þ, Cr3þ, V5þ, and many Mo6þ complexes. These complexes do not normally react with peroxides. However, in the presence of electron-rich molecules, such as alkenes, amines, sulfides, etc., oxygen insertion in the reactant occurs. For example, M – ðROOHÞ þ alkene ! ROH þ epoxide;

M ¼ Mo; Cr; V; Ti; W

ð16Þ

These catalytic reactions are distinguished from the homolytic reactions in that no evidence exists for paramagnetic intermediates. The epoxidation is stereospecific, trans- and cis-alkenes yielding trans- and cis-epoxides, respectively. Under the same conditions, complexes of Cu, Mn, and Fe give no yields or

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

59

poor yields of epoxides because they decompose ROOH rapidly into radicals. High yields of epoxides and, especially, the stereospecificity of the reaction are compatible only with a heterolytic mechanism in which the active epoxidizing agent delivers an electrophilic oxygen species from a hydroperoxide-metal complex to the reactant in a concerted manner; there is no free rotation of the C – C bond during this process. The high yields of epoxides in one case (Mo6þ, V5þ, Cr6þ, and Ti4þ) and the low yields in the other case (Fe2þ, Cu1þ, Co2þ, Cr2þ) suggest that the epoxidation of the alkene by heterolytic cleavage and oxygen insertion and the homolytic decomposition of ROOH (R ¼ H, alkyl) are competing processes (176). The selectivity to epoxide is determined by the relative rates of reaction of the catalyst-hydroperoxide complex with the alkene (Eq. (16)) in competition with its homolytic decomposition (Eq. (12)). The oxidation potential of the metal ion (in the complex) and its Lewis acidity may be expected to influence the relative rates of Eqs. (12) and (16). The redox potentials of some transition metals are given in Table VI; the heterolytic pathway is likely to be preferred for reaction on Ti4þ-silicalite. In the epoxidation step (Eq. (16)), the main function of the catalyst is to withdraw electrons and reduce the electron density at the peroxide O – O bond, making it more susceptible to attack by nucleophiles such as alkenes. In this process, the M ion acts as a Lewis acid. Active epoxidation catalysts are usually strong Lewis acids and relatively weak oxidants in their highest oxidation state (to avoid one-electron oxidative decomposition of the peroxide as per Eq. (13). (177). The Lewis acidity of M, in turn, is influenced by its coordinating ligands. The hetero- vs. homolytic O – O cleavage is also affected by the substituent on the hydroperoxide; electron-donating tert-alkyl groups on the peroxide moiety tend to favor the homolytic cleavage of the O – O bond, whereas electron-withdrawing substituents such as acyl groups facilitate O –O bond heterolysis. In other words, homolytic O – O bond cleavage is facilitated when more electron density resides on the O – O bond of the M– OOR (R ¼ H, alkyl) intermediate.

TABLE VI Redox potentials of transition metal ions in aqueous solutions Reaction

E0 (V)

Reduction

Co(III) þ e ! Co(II)

þ 1.82

Easy

Fast

V(V) þ e ! V(IV)

þ 1.00

Moderate

Moderate

Fe(III) þ e ! Fe(II)

þ 0.77

Moderate

Moderate

Ti(IV) þ e ! Ti(III)

20.37

Difficult

Difficult

H2O2 decompostion

60

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Scheme 4.

In the field of enzyme catalysis, heme-proteins such as cytochrome P450, for example, exhibit both types of O – O bond cleavages in organic hydroperoxides and peroxy acids (178). Heterolytic cleavage of HOOH/ROOH yields H2O or the corresponding alcohol, ROH and a ferryl-oxo intermediate (Scheme 4). Homolytic O – O bond cleavage results in the formation of a hydroxyl (HOz) or an alkoxyl (ROz) radical and an iron-bound hydroxyl radical. V.C. Epoxidation on Titanium Silicate Molecular Sieves V.C.1. General Features of Epoxidations Epoxidation reactions in the liquid phase have been reviewed by Sawaki (179) and more recently by Arends and Sheldon (180), and those occurring in the presence of solid catalysts by Dusi et al. (181). Because H2O2 is only a mild oxidant, its use in alkene epoxidation requires the application of appropriate catalysts. The catalytic epoxidation using H2O2 and tungstic acid, for example, proceeds via the formation of peroxytungstic acid. Aqueous conditions are usually not appropriate for epoxidations, because epoxides are prone to undergo acid-catalyzed hydrolysis. In alkene epoxidation with alkyl hydroperoxides catalyzed by various metal complexes of Ti, Mo, and V in the liquid phase, two alternate pathways, A and B in Scheme 5, each involving a metal alkyl peroxide complex, have been accepted in the literature (182). Mechanism A involves an electrophilic O transfer to alkene. Mechanism B involves a five-membered dioxametallocyclopentane. For the particular case of vanadium, the alkylperoxy

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

61

Scheme 5.

complexes were isolated and pathway B was supported by the fact that the relative rates were correlated with the coordinating ability of alkenes. The operating pathway seems, however, to change as a result of changes in the metals, ligands, and solvents (182). Early transition metals, such as Ti, for example, seem to prefer path A (182). Prior to the discovery of TS-1, amorphous Ti – SiO2 was the best known solid catalyst for the epoxidation of propene (183) using alkyl hydroperoxides, offering an alternate route to the homogeneous catalytic Halcon/ARCO process (184). However, the catalyst was unstable in the presence of H2O. In the overall reaction, ethylbenzene and propene are converted with oxygen to styrene, propene oxide, and H2O. The epoxidizing agent is ethylbenzene hydroperoxide. Sheldon et al. (185) attributed the catalytic activity to site isolation of Ti4þ on the silica surface, preventing the formation of TiO2 domains, and to the enhanced Lewis acidity of Ti4þ resulting from electron withdrawal by the Si –Oligands. The reaction mechanism is assumed to involve the Ti-alkyl peroxo groups (Ti – OOR). Propene oxide is also manufactured by the chlorhydrin route (186): CH3 – CHyCH2 þ HOCl ! CH3 – CHðOHÞ – CH2 Cl;

ð17Þ

CH3 – CHOH – CH2 Cl þ base ! CH3 CHðOÞCH2 þ baseðHClÞ:

ð18Þ

62

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

The chlorhydrin route is also used in the manufacture of epichlorohydrin from allyl chloride (187): CH2 yCH – CH2 Cl þ HOCl ! CH2 ðOÞCH – CH2 Cl þ HCl:

ð19Þ

The direct conversion of propene to its epoxide, in near quantitative yields, with aqueous H2O2 will be environmentally more benign. One of the unique features of TS-1 as a solid oxidation catalyst is its ability to utilize aqueous H2O2 as the oxidant for such conversions. This ability of TS-1 derives from the fact that silicalite-1 is hydrophobic, in contrast to the hydrophilic amorphous Ti – SiO2. Consequently, hydrophobic reactants, such as alkenes, are preferentially adsorbed by TS-1, thus precluding the strong inhibition by H2O observed with amorphous Ti –SiO2. Unfortunately, for economic reasons and in the absence of compelling environmental legislation, the process for manufacture of propene oxide using TS-1 and H2O2 is not very attractive and is not yet in commercial practice. Worldwide efforts are underway to develop this process by using H2O2 generated in situ (from H2 þ O2) or (secondary/tertiary alcohol þ O2). Metalloaded TS-1 structures are the likely catalysts (Section V.C.16). Titanosilicate molecular sieves, especially those with large pores and mesopores, however, offer great potential in the fine chemicals industry (for manufacture of drug intermediates, fragrances, agrochemicals, etc.), as the reactant molecules are larger and the economics allows the use of the more expensive H2O2 as the oxidant. Most of these large-pore and mesoporous materials need to use the alkyl hydroperoxides (such as TBHP) rather than aqueous H2O2 as the oxidant (see, however, Section V.F). V.C.2. Yields and Stereospecificities Lower alkenes such as ethene, propene, and butenes are epoxidized in high yields (. 95%) in the presence of TS-1 catalyst by aqueous H2O2 (33). The stereochemical configuration is retained in the case of butenes; cis-but-2-ene gives exclusively the cis-epoxide, and trans-but-2-ene gives exclusively the trans-epoxide. These high epoxide yields and retention of stereochemical configuration argue against the homolytic decomposition of the O – O bond of the Ti(O2H) intermediate and support a heterolytic mechanism. V.C.3. Diffusional Constraints As expected, although TS-1 is more active and selective in the epoxidation of linear alkenes (such as hex-1-ene and dodec-1-ene), the large-pore Ti-beta is more active in the case of the bulkier cyclohexene (TON of 14 vs. 1 for TS-1) and cyclododecene (TON of 20 vs. 5; Table VII) (11).

TABLE VII Diffusional constraints in selective oxidation of alkenes over Euro-TS-1 and Ti-Beta Catalyst

Hex-1-enea

TS-1b Ti-beta a

Cyclohexene

Dodec-1-enee

c

cyclododecene

H2O2

Turnover (mol/mol Ti)

Product selectivity (%)

Conversion (%)

Selectivity (%)

Epoxide

Glycol ethers

Glycol

3

50

98

80

96

–

4

3

12

80

80

12

8

80

TS-1

3

1

–

–

100

–

–

Ti-betad

3.5

14

80

83

–

–

100

TS-1

3.5

110

83

68

77

23

–

3.5

87

80

87

–

100

–

Ti-beta e

Reaction time (h)

f

TS-1

4

Ti-betaf

3.5

5

26

26

66

34

–

20

47

71

80

20

–

Adapted from Corma et al. (11). a Reaction condition: catalyst, 0.2 g; alkene, 33 mmol; H2O2/alkene (mol) ¼ 0.082; solvent (methanol), 23.57 g; temperature ¼ 333 K, tr ¼ 4 h. b Euro-TS-1 (1.7 wt% of Ti given as TiO2). c Ti-beta (Ti/(Ti þ Si) ¼ 0.044, TiO2 (wt%) ¼ 5.7, TiO2/Al2O3 ¼ 244). d Ti-beta (Ti/(Ti þ Si) ¼ 0.040, TiO2 (wt%) ¼ 5.2, TiO2/Al2O3 ¼ 210). e Reaction condition: catalyst, 0.2 g; alkene, 33 mmol; H2O2/alkene (mol) ¼ 0.258; solvent (ethanol), 23.57 g; temperature ¼ 353 K; tr ¼ 4 h. Some oxidation of ethanol was observed at these reaction conditions, which was taken into account to calculate H2O2 conversion and selectivity. H2O2 selectivity(%) ¼ (mol alkene oxidized/mol H2O2 converted) £ 100. f Ti-beta (Ti/(Ti þ Si) ¼ 0.018, TiO2 (wt%) ¼ 2.4, TiO2/Al2O3 ¼ 111).

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Alkene

63

64

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

The influence of catalyst particle size and morphology in phenol hydroxylation is shown in Table VIII and confirms the diffusional constraints in this reaction also. A novel strategy for overcoming the diffusional limitations associated with the pore size of TS-1 without sacrificing the advantages of its hydrophobicity was demonstrated by Schmidt et al. (188). These authors impregnated a sample of carbon black (particle diameter 18 nm) with a clear solution of tetrapropylammonium hydroxide, water, and ethanol. After evaporation of the ethanol, the carbon particles were impregnated with a 20% excess (relative to the incipient wetness value) of a mixture of tetraethyl orthotitanate and tetramethylorthosilicate. The composition of the resultant synthesis gel was 20 TPA2O:TiO2: 100SiO2:200H2O, and the resultant zeolite concentration was about 20%. TS-1 was then obtained by conventional hydrothermal synthesis from this inorganic gel– carbon matrix system. Finally, carbon was removed by calcination at 823 K. The resulting sample of TS-1 had a Si/Ti atomic ratio of 110, a high crystallinity, and an average crystallite size of about 1.5 mm, and it exhibited mesoporosity (about 20 nm in diameter dispersed throughout the crystal). The advantage of this “mesoporous” TS-1 over samples prepared by the conventional route is illustrated in Fig. 34. The two samples behave similarly for the oxidation of linear reactant oct-1-ene. But a marked difference was observed for the oxidation of bulkier cyclohexene. Because of the absence of diffusional constraints, the catalytic epoxidation activity in the “mesoporous” TS-1 enhanced by almost an order of magnitude for the oxidation of the bulkier cyclohexene.

TABLE VIII Influence of textural properties of TS-1 samples on phenol hydroxylation activity Sample

Average particle sizea (mm)

Morphologya

R0 b

Conversionc (%)

Selectivityd (%)

Yielde (%)

10.2

1

0.2

Cubic

50

95

93

2

0.3

Cauliflower

9.00

44

93

92

3

5.0

Coffins

1.07

6

15

40

4

10.0

Coffins

0.46

2.5

8

18

Adapted from van der Pol et al. (89). Reaction conditions: catalyst, 0.5 g; phenol, 10 g; solvent (acetone), 10 mL; 35 wt% H2O2, 2 mL (added at the beginning of the reaction); temperature ¼ 353 K. a Estimated using SEM. b R0 ¼ initial reaction rate of dihydroxy benzene formation (mol/m3 s). c Conversion ¼ H2O2 conversion at t ¼ 1 h. d Selectivity ¼ (moles dihydroxybenzene/moles of reacted H2O2) £ 100% at t ¼ 1 h. e Yield ¼ (moles dihydroxy benzene/moles of H2O2 added) £ 100% at complete H2O2 conversion.

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

65

Fig. 34. Ratio of product concentrations [sum of epoxide and secondary products; (a) from oct-1ene and (b) from cyclohexene] obtained with mesoporous and conventional TS-1 as a function of the contact time. The results show that the mesoporous TS-1 has a similar activity for oct-1-ene epoxidation as conventional TS-1. However, the mesoporous TS-1 is significantly more active for cyclohexene epoxidation [Reproduced from Schmidt et al. (188) by permission of the Royal Society of Chemistry].

V.C.4. Influence of Ti-Silicate Structure The greater activity of Ti-beta (vs. TS-1) in the oxidation of the bulky cyclohexane was noted in the previous section. Table IX provides a comparison of the conversion and epoxide selectivity in the reaction catalyzed by TS-1 and three large-pore/mesoporous Ti-silicates in the epoxidation of a single, linear allyl alcohol (pentenol). TABLE IX Influence of titanosilicate structure on epoxidation of pentenol with H2O2 Catalyst

Temperature (K)

Pentenol conversion (%)

Epoxide selectivity (%)a

Reference

Ti-MCM-41

323

32

19 (81)

(273)

Ti-MCM-48

323

32

21 (79)

(273)

Ti-beta

343

42

89 (11)

(195)

323

b

76 (24)

(193)

TS-1 a

nd

Numbers in parentheses indicate the selectivities to the corresponding unsaturated carbonyl compounds. b nd, no data available.

66

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

The higher conversion in the presence of Ti-beta is probably a result of the higher temperature (343 vs. 323 K). Diffusional constraints cannot account for the observed differences in selectivity. Ti-beta and TS-1 are distinctly more selective than the mesoporous material. Recalling that tetrapodal titanium sites are more predominant in the former two molecular sieves although tripodal titanium sites are the major surface species over the latter mesoporous material (Section II), we infer that the data indicate that high epoxidation selectivity is probably correlated with the presence of tetrapodal structures in these two molecular sieves. This correlation is discussed in Section VI. The epoxidation of hex-1-ene catalyzed by Ti-beta samples synthesized in the conventional, basic medium (Ti-beta(OH)) is compared in Table X with that catalyzed by a sample synthesized in a fluoride-containing medium (Ti-beta(F)) (13). The latter was more hydrophobic. Results for the reaction catalyzed by TS-1 are also included in Table X. Ti-beta(F) is superior to TS-1 for reaction in acetonitrile solvent. The most significant difference between Ti-beta(F) and Ti-beta(OH) is in their selectivities. Although the selectivity to the epoxide for reaction in acetonitrile is always very high, regardless of the zeolite; for reaction in methanol, Ti-beta(F) is more selective than Ti-beta(OH) (76.6 vs. 54.9%, Table X). Both Ti-beta samples are, however, less selective than TS-1 for reaction in methanol. The lower activity of Ti-beta(OH) in the epoxidation of an alkene containing a polar head (oleic acid, Table XI) was attributed by Blasco et al. (13) to the different adsorption properties of the two catalysts. A strong adsorption of oleic acid through the polar head on the relatively more hydrophilic Ti-beta(OH) TABLE X Epoxidation of hex-1-ene catalyzed by Ti-containing zeolites: influence of method of preparation Catalyst

TiO2 (wt%)

Solvent

Hex-1-ene conversiona

Epoxide selectivity (%)

H2O2 selectivity (%)

TONb

Ti-beta(F)

2.86

CH3CN

41.2

100

99.7

43.1

Ti-beta(OH)

2.78

CH3CN

40.3

100

76.6

53.4

TS-1

2.18

CH3CN

25.5

100

76.5

39.1

Ti-beta(F)

2.86

CH3OH

26.8

76.6

97.9

30.7

Ti-beta(OH)

2.78

CH3OH

25.4

54.9

90.1

24.2

TS-1

2.18

CH3OH

46.6

97.6

96.7

94.5

Adapted from Blasco et al. (13). Reaction conditions: catalyst, 0.1 g; hex-1-ene, 16.5 mmol; solvent, 11.8 g; H2O2, 4.1 mmol; temperature ¼ 323 K; time ¼ 2 h. a Percentage of maximum. b Initial turnover number (moles of converted alkene/moles of Ti £ hours).

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

67

TABLE XI Epoxidation of oleic acid over Ti-beta prepared in fluoride (F) and alkali (OH) medium TiO2 wt %

Acid conversiona

Epoxide selectivity

H2O2 selectivity

Ti-beta(F)

2.52

31.2

100

67.6

Ti-beta(OH)

2.78

20.2

100

24.8

Catalyst

Adapted from Blasco et al. (13). Reaction conditions: catalyst, 30 mg; oleic acid, 1 mmol; CH3CN, 2 mL; H2O2, 0.25 mmol; temperature, 323 K; time, 8 h. a Percentage of maximum.

would make the oxidation of the double bond in the middle of the hydrocarbon chain more difficult. V.C.5. Epoxidation Catalyzed by Mesoporous Titanium Silicates Although the mesoporous materials, such as Ti-MCM-41, have lower intrinsic epoxidation selectivity than TS-1 and Ti-beta, they must nevertheless be used as catalysts for reactions of large molecules typical in the fine chemicals industry. It is, therefore, interesting to elucidate how these ordered mesoporous materials compare with the earlier generation of amorphous titania – silica catalysts. Guidotti et al. (189) recently compared Ti-MCM-41 with a series of amorphous titania – silica catalysts for the epoxidation of six terpene molecules of interest in the perfumery industry (Scheme 6). Anhydrous TBHP was used as the oxidant because the catalytic materials are unstable in water. The physical characteristics of these catalysts are compared in Table XII. It was observed that no leaching of Ti occurs during the catalytic reaction in the anhydrous medium. The acidity of the catalysts (which gave rise to many side products) was evaluated by a comparison of their reaction rates in the acid-catalyzed conversion of citronellol into isopulegol (Scheme 7). The acidity of the catalysts decreased in the following order: A . C . D . B ø E. The catalytic activity and epoxidation selectivities are compared in Table XIII.

Scheme 6.

68

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169 TABLE XII Ti composition and textural characteristics of titanium silicates

Sample

Method of preparation

Ti loading, Specific Total pore Mean pore calcined surface volume diameter samples area (mL/g) (nm) (wt%) (m2/g)

(A) Ti-MCM-41 Ordered, Ti-grafted, mesoporous silica Amorphous, Ti-grafted, porous (B) Ti-SiO2 silica (Grace Davison 62) (C) Ti-SiO2 Ti-grafted silica (Aerosil 380, Degussa) (D) MST Amorphous, mesoporous titaniasilica (co-precipitation) (E) TiO2-SiO2 Commercial, amorphous, porous mixed oxide (Grace)

1.88

861

0.53

2.4

1.75

303

1.10

12.8

1.78

268

nd

nd

1.84

454

0.38

4.6

1.40

303

1.16

12.7

Adapted from Guidotti et al. (189); nd, not determined.

The results led to the following conclusions: 1. With regard to the specific activity, the mixed oxide catalyst, E, showed the best performance of all reactants, 1 and 6 being exceptions. For the latter, A and B performed better. 2. Epoxidation of alkeneic reactants is faster on titanium-grafted silicates (such as A, B and C) than on the coprecipitated titanosilicates (such as D and E). This difference was attributed to the fact that on extra-framework titaniumgrafted silicates, the catalytically active sites are virtually all exposed and accessible, whereas on the coprecipitated material some of them may be buried within the silicate walls and, thus, cannot adsorb reactant molecules. 3. Most of the side product formation was caused by the oxidation of the alcohol function, as expected. 4. When the OH group in the reactant is absent or far from the double bond (reactants 6 and 1, respectively), the Ti-grafted materials displayed the best activity values. When the OH group is in the proximity of the CyC

Scheme 7.

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

69

TABLE XIII Comparative catalytic activities (turnover numbers and selectivity (in parentheses)) of ordered Ti-MCM-41 (A) and amorphous titania –silica (B–E) catalysts in the epoxidation of unsaturated cyclic terpenes (1–6) using anhydrous TBHP Terpenes

Catalyst A

B

C

D

E

1

44a (51)b

37 (60)

29 (57)

22 (58)

28 (53)

2

38 (61)

37 (80)

31 (88)

23 (65)

44 (90)

3

43 (64)

44 (84)

40 (81)

40 (74)

59 (71)

4

36 (80)

38 (82)

32 (88)

19 (84)

45 (89)

c

45 (84)

43 (83)

30 (83)

52 (75)

c

33 (89)

32 (92)

19 (85)

25 (75)

5 6

40 (73) 30 (90)

Adapted from Guidotti et al. (189). Reaction conditions: catalyst, 50 mg; substrate, 1 mmol; TBHP: terpene (mol) ¼ 1:1; solvent, CH3CN; VTOT mix., 10 mL; temperature, 363 K; time, 24 h; magnetic stirring (ca. 800 rpm). Textural properties of the catalysts (A– E) are given in Table XII. Structures of the substrates (1 –6) are shown in Scheme 6. a TON, turnover number after 24 h ([mol converted terpene]/[mol Ti]). b Selectivity to monoepoxide after 24 h (%). c Selectivity to endocyclic monoepoxide after 24 h (%).

bond, the promotion effect of the OH group (hydroxyl-assisted epoxidation, see Section V.C.9) prevails and the differences in activities between the various catalysts become smaller. 5. The epoxide selectivity did not depend noticeably on the gross structural features of the catalyst. For instance, the selectivity in the epoxidation of 4 is about 85% on all solids (Table XIII). 6. As long as the pore diameters are large enough for easy entry and exit of reactant and product molecules, the catalyst porosity features do not have a significant influence on the epoxidation activity. In a comparison between two epoxidation catalysts obtained by grafting Ti(iso-PrOi)4 on MCM-41 and an amorphous silica gel, respectively, the former showed a lower activity (189). 7. A significant absorption band in the 300– 350 nm region of the DRUV spectra indicated that samples B and C, which contained significant amounts of Ti –O – Ti oligomeric sites in octahedral coordination (Fig. 35), have good catalytic activity. The authors postulated that on these materials “complete site isolation is not mandatory in order to have active and selective titania – silica epoxidation catalysts”. The 100% selectivity of the dinuclear, silica-supported

70

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Fig. 35. Diffuse reflectance UV–visible spectra of Ti-MCM-41 (A), Ti –SiO2 Davison (B), Ti – SiO2 Aerosil (C), MST (D), and TiO2 –SiO2 Grace (E) [from Guidotti et al. (189)].

(xSiO)2TiOTi(OO-t-Bu)4 species, prepared by the grafting route, in the epoxidation of cyclohexene (190) was cited as additional support for the above argument. V.C.6. Influence of Alkene Structure Epoxidation of alkenes with terminal CyC bonds is faster than that of alkenes with internal CyC bonds when the reaction is catalyzed by TS-2 (Table XIV). TABLE XIV Epoxidation of various alkenes over TS-2: influence of alkene structure Hex-1-ene Hex-2-ene Hex-3-ene Oct-1-ene Dodec-1-ene Cyclohexene Conversion (mol%) Epoxide selectivity (%)

92.0

81.2

72.0

56.4

28.8

40.2

73.5

69.0

76.5

66.3

50.0

54.3

Adapted from Kumar et al. (165). Reaction condition: catalyst (TS-2; Si/Ti ¼ 29), 0.1 g; reactant, 1.0 g; H2O2/substrate ¼ 1.1; solvent (CH3CN), 10 g; temperature, 333 K; time, 6 h.

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

71

TABLE XV Relative reaction rates for epoxidation between different alkenes and hex-1-ene on Ti-beta with H2O2 and TBHP Oxidant

Oct-1-ene

Dec-1-ene

4-m-Pent-1-ene

1-m-Cyclohex-1-ene

H2O2a

0.70

0.60

1.45

1.22

TBHPb

0.52

0.37

0.49

1.09

Adapted from Corma et al. (191). Reaction conditions (H2O2 oxidant): catalyst, 0.2 g; Alkene, 33 mmol; H2O2 (35 wt%), 0.8 g; solvent (CH3OH), 23.6 g; temperature, 323 K; time, 2 h. b Reaction conditions (TBHP oxidant): catalyst, 0.3 g; alkene, 25 mmol; TBHP, 6.25 mmol; solvent (CH3CN), 10 g; temperature, 323 K; time, 5 h. a

The rate also decreases with an increase in the chain length of the alkene molecule (hex-1-ene . oct-1-ene . dodec-1-ene). Although the latter phenomenon is attributed mainly to diffusion constraints for longer molecules in the MFI pores, the former (enhanced reactivity of terminal alkenes) is interesting, especially because the reactivity in epoxidations by organometallic complexes in solution is usually determined by the electron density at the double bond, which increases with alkyl substitution. On this basis, hex-3-ene and hex-2-ene would be expected to be more reactive than the terminal alkene hex-1-ene. The reverse sequence shown in Table XIV is a consequence of the steric hindrance in the neighborhood of the double bond, which hinders adsorption on the electrophilic oxo-titanium species on the surface. This observation highlights the fact that in reactions catalyzed by solids, adsorption constraints are superimposed on the inherent reactivity features of the chemical reaction as well as the diffusional constraints. The epoxidation rates of various alkenes relative to hex-1-ene on Ti-beta with H2O2 and TBHP are summarized in Table XV. In the absence of diffusional constraints, the branched alkenes are more reactive than the linear ones (see also Section V.C.13). V.C.7. Dialkenes Selective epoxidation of one of the double bonds in dialkenes is of practical interest (Table XVI). Although monoepoxides predominate at low H2O2 concentrations, the diepoxides are also formed at higher concentrations. The diallyl epoxides of bisphenol A are major intermediates in the adhesives industry, and their synthesis in solid-catalyzed reactions in an eco-friendly manner remains a challenge.

72

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169 TABLE XVI Epoxidation of dialkenes catalyzed by TS-1

Alkene

Solvent

T (K)

H2O2 conversion

Yield based on H2O2 (%)

Epoxide selectivity (%) Mono

Di

Butadiene

tert-Butyl alcohol

293

98

85

85

15

Diallyl carbonate

Methanol

338

95

50

93

5

Diallyl ether

Methanol

338

96

60

90

4

Adapted from Romano et al. (192); diene/H2O2 ¼ 2.5.

V.C.8. Epoxidation in the Presence of Other Oxidizable Functional Groups V.C.8.1. Alkenes and Alcohol Functions. Although TS-1 and other titanosilicates oxidize alcohols to the corresponding aldehydes and ketones, the rates are suppressed in the presence of compounds containing CyC bonds. CH3OH, for example, is not oxidized at all during epoxidations of alkene reactants. Higher alcohols, however, are partially oxidized. The oxidation of unsaturated alcohols in the presence of TS-1 is shown in Table XVII (193). When the double bond has no substituents, as in allyl alcohol, but-3-ene1-ol, or 2-methylbut-3-ene-1-ol, only the epoxide is formed; but when the double bond has substituents, the epoxidation rate is decreased and ketone and aldehyde products are formed from the oxidation of the OH group. This effect is more pronounced with a greater degree of substitution of the reactant. Because the double bond and the OH group are part of the same molecule, this difference must arise from the different abilities of the functional groups to coordinate and react at the Ti center. The terminal double bond, sterically less hindered, interacts strongly with titanium, preventing coordination of the competing OH group. Because of steric hindrance, this interaction is weaker in substituted alkenes, allowing the OH group to undergo oxidation (190). V.C.8.2. Alkenes and Alkanes. When oct-1-ene was oxidized by H2O2/TS-1 in the presence of n-hexane, under conditions that would lead to the oxidation of each if it were used separately, epoxidation occurred preferentially (103). This result is probably an evidence of the greater nucleophilicity and, hence, coordinating ability of the alkene. V.C.9. Hydroxyl-Assisted Epoxidation Hydroxyl-assisted epoxidation using TS-1/H2O2 is chemo- and stereoselective (165). Thus, when cyclopent-2-en-1-ol or cyclohex-2-en-1-ol was treated with

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

73

TABLE XVII Oxidation of unsaturated alcohols in the presence of TS-1: effect of alkene structure on selectivity Reactant

Product yield (mol/mol of Ti) Ketone/aldehyde

Epoxide

0

19

0

16

0

30

31

95

37

4

7

27

43

65

44

141

98

94

18

10

75

17

Adapted from Tatsumi et al. (193). Reaction conditions: TS-1 (Si/Ti ¼ 52), 0.01 g; reactant, 2.5 mL; H2O2 (30% aq. solution), 2.5 mL; temperature, 323 K; time, 3 h.

aqueous H2O2/TS-1, the corresponding epoxides were obtained in 75– 80% yields. Cyclohexenol gave the cis I as the major product (90%) (where epoxide and OH are cis to each other), and the trans II as a minor product (10%) (where the epoxide and OH are trans to each other) (Scheme 8). Cyclopentenol also

74

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Scheme 8.

behaved similarly. In addition to the epoxide, other products resulting from oxidation of the OH group and cleavage of the epoxide were also detected. As a further example of a hydroxyl-assisted epoxidation, geraniol and nerol bearing two isolated CyC double bonds were regioselectively epoxidized with TS-1 at the 2-position (near the OH group), as reported by Kumar et al. (195). On the basis of these results, Kumar et al. (195) proposed that the transition state of the epoxidation of allylic alcohols involves coordination of the alcoholic functional group to the Ti active site and that the double bond interacts with one of the peroxidic oxygen atoms, not with the titanium site (Scheme 9). The epoxidation of a bulky reactant such as alpha-terpineol was accomplished with Ti-beta as the catalyst. The initially formed epoxide was rearranged to cineol alcohol, as shown in Scheme 10 (18,196,197). Even as large a molecule as cholesterol was epoxidized in the presence of Ti-MCM-41 catalyst (198). An epoxide selectivity of 53% at 48% conversion was achieved. The oxidation of the OH group and allylic oxidations were important side reactions. V.C.10. Diastereoselectivity in Epoxidations Epoxidation of allyl alcohols can generate two isomers, the threo- and erythroepoxides (Schemes 11 and 12). Control of the relative amounts of the two isomers

Scheme 9.

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

75

Scheme 10.

is crucial in the synthesis of many compounds of interest in the fine chemicals industry. The results of Adam et al. (199,200) for reactions catalyzed by TS-1 and Ti-beta are summarized in Table XVIII. As expected, TS-1 was not active for the bulky reactants (Table XVIII, entries 9 –14). The diastereoisomeric ratios evidencing catalysis by TS-1 and Ti-beta are broadly similar to those of the homogeneous system Ti(OPr)4-TBHP and chloroperbenzoic acid. A transition state for the active species analogous to the structure of peracid epoxidations was, therefore, suggested (199), involving interaction of the alcoholic functional group with the peroxo oxygen atom by hydrogen bonding. V.C.11. Side Reactions During Epoxidation On titanosilicate molecular sieves, especially non-TS-1 materials, the epoxides formed react further to form glycols, glycol ethers, and even products arising from the further rearrangemnent of the epoxide. Thus, in the epoxidation of styrene by H2O2/TS-1, the epoxide rearranged efficiently into phenylacetaldehyde (165). No or very little acetophenone was produced, phenylacetaldehyde being the sole or major product. The high regioselectivity for phenylacetaldehyde was attributed to the stabilization of the benzyl cation (165). Although high (epoxide þ phenylacetaldehyde) selectivities (85 – 90%) were obtained for reaction in the presence of acetone, alcoholysis occurred to a great extent (45%) in the presence of methanol solvent, producing mono glycol ethers.

Scheme 11.

76

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Scheme 12.

Allyl alcohol epoxidation with TS-1/H2O2 and the subsequent epoxide ring opening reaction by water or the organic solvent was investigated thoroughly by Hutchings et al. (201 – 203). Although very high selectivities to epoxides were observed at low conversions and temperatures, ether diols, resulting from the nucleophilic epoxide ring opening by the alcohol, were the major products at temperatures above 338 K. Scheme 13 was proposed for epoxide ring opening by polar solvent molecules. It was shown (201– 203) that the Ti-peroxo complex is more acidic than TS-1 alone (without H2O2), and it is mainly this complex that catalyzes the solvolysis reaction. When Brønsted acid sites were deliberately introduced into TS-1 (by partial introduction of Al3þ in the framework), the epoxide was not found among the reaction products because it was rapidly converted to the ether diol solvolysis products. V.C.12. Influence of pH Since acidity (Lewis or Brønsted) impacts adversely on the yield of epoxides, Clerici and Ingallina (204) added basic compounds in low concentrations to TS-1 catalysts during epoxidation of alkenes to inhibit the oxirane ring opening and enhanced the epoxide yields. A comprehensive investigation of the influence of pH on product selectivity in epoxidation of allylalcohol, allylchloride, and styrene catalyzed by various titanosilicates was reported recently by Shetti et al. (205). Although conversion of allyl alcohol catalyzed by TS-1 decreased from 95.3% (at pH ¼ 3.5) to 22.2% (at pH ¼ 8.5), epoxide selectivity increased from 86.8 to 100% (Table XIX). The H2O2 efficiency decreased markedly at high pH. Most of the H2O2 probably decomposed to H2O and O2 at high pH. Ti-MCM-41 exhibited lower activity than TS-1. Changes in pH did not affect conversions significantly when reaction was catalyzed by Ti-MCM-41. To investigate the

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169 TABLE XVIII Diastereoselective epoxidation of allylalcohols Reactant

Catalyst/oxidant/solvent (TS-1/UHP/CH3COCH3)

Ti-beta/H2O2 (85%)/ CH3CN

Diastereomeric ratio (threo: erythro) 60:35

62:38

55:45

56:44

65:35

64:36

87:13

91:9

81:19

89:11

95:5

95:5

90:10

80:20

93:7

No epoxide

58:42

No reaction

95:5

No reaction

70:30

(Continued)

77

78

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169 TABLE XVIII Continued Reactant

Catalyst/oxidant/solvent (TS-1/UHP/CH3COCH3)

Ti-beta/H2O2 (85%)/ CH3CN

No reaction

88:12

No reaction

15:85

No reaction

70:30

Adapted from Adam et al. (199, 200).

influence of cations present in solution, the epoxidation of allyl alcohol was carried out with TS-1 catalyst at pH ¼ 8 in the presence of various alkali metal and alkaline earth compounds (205). Catalytic activity increased in the following þ þ þ order: Liþ , NHþ and Mg2þ , Ca2þ , Ba2þ. Epoxide 4 , Na , K , Cs selectivity followed the reverse order; Csþ exhibited 100% allyl alcohol conversion but only 76.7% epoxide selectivity (Table XX, Run number 5). The influence of pH on epoxidation of styrene with aqueous H2O2 catalyzed by TS-1 was also investigated. Conversion of styrene decreased, and styrene oxide selectivity increased marginally at high pH values (Table XXI).

Scheme 13.

TABLE XIX Epoxidation of allyl alcohol and allyl chloride—influence of pH Run no.

Catalyst

Reactant

pH After H2O2 addition

At the end of the reaction

Olefin conversion (mol%)

H2O2 efficiency

Epoxide selectivity (mol%)

1

TS-1

AA

4.5

3.5

4.2

18.9

95.3

100

86.8

2

TS-1

AA

5.5

3.5

4.2

18.7

94.4

100

87.4

3

TS-1

AA

7.0

5.5

5.8

17.7

89.2

95

4

TS-1

AA

8.0

5.7

5.9

16.0

80.7

87

100

5

TS-1

AA

9.0

5.9

6.2

7.1

35.6

46

100

6

TS-1

AA

10.0

8.5

8.0

4.4

22.2

21

100

7

AA

7.8

5.5

5.7

14.8

74.4

97

8

TS-1 in runs 3 and 4 reused TS-1-Na(8)

AA

8.0

5.7

5.9

16.2

81.6

89

100

9

TS-1-Na(10)

AA

10.0

7.8

8.0

5.9

29.7

20

100

92.8

96.8

10

Ti-MCM-41

AA

6.8

–

–

2.1

10.4

100

11

Ti-MCM-41

AA

8.0

–

–

2.3

11.4

100

12

TS-1

AC

6.8

3.2

3.0

19.3

97.2

13

TS-1

AC

7.0

3.8

3.8

19.8

14

TS-1

AC

8.0

4.3

4.2

19.8

97

73.8

100

100

79.7

100

100

81.6

79

Adapted from Shetti et al. (205). Reaction conditions: catalyst (TS-1: Si/Ti ¼ 33; Ti-MCM-41: Si/Ti ¼ 52), 0.1 g; reactant, 0.5 g; CH3OH, 10 g; H2O2 (50% aqueous), 0.9 mL; H2O2/allylalcohol, 2.0; temperature, 333 K; time, 8 h; AA, allyl alcohol; AC, allyl chloride. TOF, moles reactant converted per mol of Ti per hour. Catalysts used in run nos. 8 and 9 were prepared by impregnating TS-1 with Naþ ions with initial pH being 8 and 10, respectively.

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Initial/ before H2O2 addition

TOF

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

80

TABLE XX Effect of alkali and alkaline ions on the epoxidation of allylalcohol Run no.

Alkali/Alkaline earth ions

TOF

Conversion (mol%)

Epoxide selectivity (mol%)

1

Liþ

2.4

11.9

100

2

NHþ 4

9.7

48.6

100

þ

3

Na

17.0

85.8

91.7

4

þ

18.7

94.4

79.1

K

5

þ

Cs

19.8

6

Mg2þ

12.5

63.0

7

Ca2þ

18.8

94.7

88.5

8

2þ

18.6

94.1

75.0

Ba

100

76.7 100

Adapted from Shetti et al. (205). Reaction conditions: catalyst (TS-1; Si/Ti ¼ 33), 100 mg; allyl alcohol, 0.5 g; CH3OH, 10 g; H2 O2 (50%), 0.9 mL; H2 O2 / allylalcohol ¼ 2.0; temperature, 333 K; run time, 8 h; pH, 8.0. TOF, moles of allyl alcohol converted per mol of Ti per hour.

V.C.13. Epoxidation with Alkyl Hydroperoxides Although aqueous H2O2 is an efficient oxidant with TS-1 and Ti-beta, catalyst stability and conversion are not as good when Ti-MCM-41 or other hydrophilic, mesoporous Ti-silicate molecular sieves are used as catalysts. The behavior of the mesoporous materials resembles the Shell catalyst, amorphous Ti –SiO2. TBHP is a better oxidizing agent than H2O2 in this case. Although mesoporous materials do not match the epoxide selectivity and H2O2 efficiency of TS-1 for small TABLE XXI Influence of pH on styrene epoxidation over TS-1 Run no.

pH

Conv. (mol%)

Styrene oxide

Methylated diol

Diol

Benzaldehyde

Phenyl acetaldehyde

Others

1

6.8

39.9

35.9

46.0

1.1

13.8

0.3

2.9

2

7.0

42.3

35.0

45.5

0.8

15.1

0.5

3.1

3

8.0

35.2

40.9

40.7

1.1

15.5

0.3

1.5

4

9.0

27.3

45.0

37.4

0.8

16.8

0.0

0.0

5

11.0

4.4

66.6

5.9

0.0

18.3

9.2

0.0

Adapted from Shetti et al. (205). Reaction conditions: TS-1: 0.1 g; styrene, 0.898 g; CH3OH, 10 g; H2O2 (50%), 0.9 mL; H2O2/styrene, 2; temperature, 333 K; time, 8 h.

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

81

molecules, they are superior to it in the epoxidation of bulky alkenes (11,13,193, 196,199). Ti-beta, in contrast to TS-1, is considerably more active in the epoxidation of allylic alcohols highly substituted at the CyC bond (compare reactants 3 and 4 in Table XXII). The accessibility of the CyC bond to the titanium oxo centers is apparently not seriously hindered by the alkyl substituents in reaction catalyzed by Ti-beta. V.C.14. Epoxidation of Alkenes Containing Carbonyl Groups In homogeneous systems, electron-withdrawing groups such as CyO, when conjugated with the alkene double bond, retard the epoxidation as the delocalization

TABLE XXII Epoxidation of allylalcohols with titanosilicates and H2O2 Reactant

Productivityb (mmol/g/h)

Catalyst

T (K)

Conversion (%)

Epoxide selectivitya (%)

Ti-beta

343

42

89 (11)

28

TS-1

323

nd

76 (24)

19

Ti-beta

343

nd

.90

TS-1

333

nd

96

Ti-beta

343

nd

.90

TS-1

323

nd

Ti-beta

343

nd

85 (11)

8.5c

TS-1

323

nd

90 (10)

3.3

Ti-beta

343

nd

96

TS-1

323

nd

82 (18)

9

Ti-beta

343

nd

74 (26)

12

TS-1

303

54

100

100

3.9c 21 2.5c 3

25

73

Adapted from Dusi et al. (181). Note: nd, no data available. a Numbers in parentheses indicate the selectivities to the corresponding unsaturated carbonyl compounds. b Amount of oxygenated products, related to unit amount of catalyst and unit time. c Based on epoxide and triol formed.

82

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

of the p-electrons reduces the electron density at the double bond. Ratnasamy and Kumar (206) found that the main products formed in the oxidation of acrolein and methacrolein were the corresponding acids from the reaction at the carbonyl end (. 90%); only little epoxide was obtained. V.C.15. Epoxidation Using Urea –H2O2 Adduct The epoxide selectivity in the TS-1/aqueous H2O2 system is reduced because of the formation of isomerized and/or cleaved secondary products because the oxirane ring is quite prone to hydrolysis in the presence of water. To circumvent this problem, an anhydrous source of H2O2, namely, urea –H2O2 adduct (UHP), which slowly releases anhydrous H2O2 into the solution, was successfully employed by Laha and Kumar (207) to enhance epoxide selectivities, even in the difficult case of styrene to styrene oxide (Table XXIII). The formation of side products, diols (by hydrolysis), phenylacetaldehyde (by rearrangement of the epoxide), and benzaldehyde (by C –C bond cleavage), have all been significantly reduced when UHP was used as the oxidant. TABLE XXIII Effect of different oxidants on epoxidation of styrene and allylbenzene catalyzed by TS-1 and TS-2 Reactant

Styrene

Catalyst

TS-1

TS-2

Allylbenzene

TS-1

Oxidanta

HP

Conversion (mol%)

56

TONb

13.4

Product distribution (mol%)a EP

PAD

BD

Diols

5

44

29

22

U þ HP

65

15.6

81

8

7

4

UHP

71

17.0

87

5

7

1

HP

57

14.1

7

42

28

23

U þ HP

62

15.8

80

8

8

4

UHP

67

17.3

85

6

7

2

HP

60

12.7

58

–

–

42

U þ HP

68

14.4

95

–

–

5

UHP

70

14.8

98

–

–

2

Adapted from Laha and Kumar (207). Reaction conditions: reactant:oxidant (mol) ¼ 4; solvent, acetone; reactant:acetone (wt/wt) ¼ 1; reaction time (h) ¼ 12 h; catalyst wt ¼ 20 wt% of the reactant; T ¼ 313 K. a EP, epoxy allylbenzene or styrene oxide; PAD, phenylacetaldehyde; BD, benzaldehyde; Diols, 3phenyl-1,2-propanediol or styrene diol, including some high-boiling products; HP, H2O2 (45 wt% aqueous); U þ HP, urea and H2O2 mixture (1:1 mol ratio); UHP, urea–H2O2 adduct. b TON, moles of H2O2 converted for producing epoxide þ secondary products per mole of Ti.

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

83

V.C.16. Epoxidation Using Dioxygen One of the major developments in the preceding decade in the area of epoxidation catalyzed by titanosilicates is the attempt to generate H2O2 in situ by a mixture of H2 þ O2 catalyzed by Pd/Pt-TS-1 (69– 71,208 –210) or Au-TS-1 (74). The strategy was to generate H2O2 from H2 þ O2 catalyzed by the noble metals and react it with the alkenes (especially propene) in the presence of TS-1 catalyst to produce the epoxide. Intimate contact between the metal and TS-1 and consequently a high dispersion of the metal on the hydrophobic TS-1 surface is needed. The latter is difficult to achieve and especially to maintain. Catalyst deactivation was a major problem (71). In addition to propene epoxide, the by-products included methyl formate (from the methanol solvent), acetone, acrolein, acrylic acid, and methylated glycols (71). An interesting observation in most of the investigations (69– 71,74,208) was that although the yields were low, the propene selectivity to the epoxide was . 99%; the yields were low as a consequence of the low hydrogen and oxygen efficiencies in the production of H2O2 (74). The in situ generation of H2O2 at the precious metal site is probably rate-determining in this reaction (208). Catalyst deactivation was also a problem. Meiers et al. (69) found that the formation of propene oxide in the presence of Pd – Pt-TS-1 was favored when a high fraction of palladium is present as Pd2þ species and small palladium clusters, whereas fully reduced palladium and large clusters favored propene reduction to propane. The fraction of Pd2þ was increased by autoreduction of the complex incorporating tetramine ligands [(Pd(NH2)4]2þ was the precursor for Pd-TS-1) in the absence of hydrogen in the reduction medium; calcination of the dried sample in N2 at 523 K was adequate to reduce the Pd ions. Reaction temperatures . 423 K or calcinations in air led to palladium cluster agglomeration on the external TS-1 surface and thus to decreasing epoxide yields and selectivities. Addition of minor amounts of platinum also drastically increased the fraction of Pd2þ species in comparison to the Pd0 species (69). Although no epoxidation of propene occurred in the catalysis by TS-1 with H2 þ O2 as the oxidant, a 5.3% yield of propene oxide was obtained with a 1% wt Pd – 0.1%wt Pt-TS-1 catalyst under the same conditions. Of course, the yield was much higher (39%) in the TS-1/H2O2 system (70). Although higher yields have been reported (up to 12% propene oxide (69)), they are still much lower than those obtained with H2O2.

V.D. Hydroxylations V.D.1. General Features Titanosilicate molecular sieves, especially TS-1, are active in the hydroxylation of both alkanes and aromatic compounds (33,165) when H2O2 is used as

84

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

the oxidant. The manufacture of hydroquinone and catechol, in nearly equal quantities, from phenol and H2O2 with TS-1 catalyst is in commercial practice. The substitution or insertion of an oxygen atom into C – H bonds is not easy, and the applied reagents have to be strongly electrophilic oxidants or radical species. C– H hydroxylations can be classified broadly into two reaction types (179): the first type is the insertion of a singlet oxygen atom (1O) into C – H bonds from electrophilic oxidants as in Eq. (20).

ð20Þ 3

The second type is the hydroxylation by a triplet oxygen atom ( O) and involves radical intermediates via H abstraction: ð21Þ The hydroxylation of C – H bonds by radicals, in contrast to the case of electrophilic oxidants, leads to alcohols without retention of stereochemical configuration. H2O2, activated by strong acids (superacids (211), HF – BF3 (212), AlCl3 (213), and CF3COOH (214)) have been used for the hydroxylation of aromatic compounds. These acid-catalyzed hydroxylations cannot be applied for aliphatic reactants because the hydroxylated products are more reactive than the starting compounds and, hence, they are oxidized further. Radical hydroxylation of hydrocarbons by autooxidation yields alcohols (major products), ketones, and acids (minor products). Cyclohexanol, for example, is formed in 90% yield from cyclohexane and peroxyacetic acid (215). The high -ol/-one ratio at low conversions can sometimes be used as a partial diagnostic tool to distinguish between the radical and electrophilic pathways. The predominant reaction of electrophilic radicals, such as HOz, ROOz, and CHz3 is H-atom abstraction from reactants (S –H) or peracids, as exemplified by the following: Xz þ S – H ! XH þ Sz

ð22Þ

Sz þ HOz ! SOH

ð23Þ

Xz þ H – OOCOR ! XH þ RCOz3

ð24Þ

Thus, the generation of these radicals leads to the hydroxylation of S – H. The reactive hydroxyl radicals can be produced by the radiolysis of water or

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

85

TABLE XXIV Oxidation of n-alkanes in 95% methanol Hydrocarbon

Selectivity based on H2O2 (%)a

2/3 ratiob

Product distribution (mol%) 2-ol

3-ol

4-ol

2-one

3-one

4-one

Propane

35

66.2

33.8

n-Butane

69

55.0

45.0

n-Pentane

82

4.5

34.3

16.1

47.4

2.1

n-Hexane

86

2.6

32.1

25.9

39.8

2.0

n-Heptane

75

1.9

33.7

29.2

6.2

28.1

2.8

Trace

n-Octane

63

2.6

30.1

20.5

12.5

32.8

3.0

1.0

n-Decane

56

1.1

11.5

20.5

36.2

16.5

4.5

10.8

From Notari (33). a Represents the moles of oxygenated products obtained per 100 moles H2O2 reacted. b Ratio between 2- and 3-compounds.

the reduction of H2O2: H2 O ! H2 Oþ ! H3 Oþ þ HOz 2

ð25Þ 3þ

ð26Þ

H2 O2 þ Ti3þ ! HO2 þ HOz þ Ti4þ

ð27Þ

H2 O2 þ Fe

2þ

z

! HO þ HO þ Fe

V.D.2. Hydroxylation of Aliphatic Compounds Linear alkanes have been hydroxylated in the 2-, 3-, and 4-positions to give secondary alcohols and ketones in the presence of TS-1 catalyst (216,217) with good selectivities based on alkanes and H2O2 (Table XXIV). The alcohols are intermediates in the formation of ketones. Isomerization of the products is not observed. Hydroxylation at the 2-position is favored over that at the 3-position, and the latter is preferred over hydroxylation at the 4position. Solubility and concentration in the reaction medium, intrazeolite diffusion of the reactants, steric hindrance at the reactive carbon center, and C – H bond strength influence the reactivity and H2O2 selectivity (Table XXIV). The advantage of the large-pore Ti-beta over TS-1 in the oxidation of bulky alkane molecules is shown by the results in Table XXV. Table XXVI shows the results of a competitive experiment in which hydroxylation of an equimolar mixture of n-hexane and another alkane (alkane II)

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

86

TABLE XXV Comparative activity of Ti-beta and Euro-TS-1 for selective oxidation of different alkanes Catalysta

Alkane

n-Hexane

Turnover (mol/mol Ti)

TS-1

3-Methylpentane

Cyclohexane

Methylcyclohexane

H2O2

Product selectivity (%)

Conv. (%)

Sel. (%)b

-ol

-one

48.5

77

100

91.5

8.5

Ti-beta

0.5

11

32

55.0

45.0

TS-1

0.7

6

19

88.9

11.1

Ti-beta

0.8

17

29

84.8

15.2

TS-1

b

–

–

Ti-beta

2.3

TS-1

b

–

Ti-beta

5.2

–

–

–

22

51

98.9

–

–

–

29

88

92.8

1.1 – 0.9

Adapted from Corma et al. (11). Reaction condition: catalyst, 0.2 g; alkane, 33 mmol; solvent (CH3OH), 23.57 g; H2O2/alkane ¼ 0.082 mol/mol; temperature, 333 K; reaction time ¼ 4 h. Catalyst: Euro-TS-1 (1.7 wt% TiO2); Ti-beta (5.2 wt% TiO2, TiO2/Al2O3 ¼ 210; Ti/(Ti þ Si) ¼ 0.040). a H2O2 selectivity (%) ¼ (mol alkane oxidized/mol H2O2 converted) £ 100. b Activity below detection limit.

TABLE XXVI Competitive oxidation of equimolar mixtures of n-hexane and another alkane (alkane II) over TS-2 using H2O2 as oxidant Alkane II

Critical diameter (nm)

Conversion (mol%) n-Hexane

Alkane II

n-Hexane/alkane II conversion

3-MP

0.55

7.8

2.8

2.8

2,2-DMB

0.61

8.2

1.7

4.8

Cyclohexane

0.60

12.3

1.8

6.8

n-Hexane

0.43

18.9

–

–

From Kumar et al. (165). Reaction conditions: catalyst (TS-2; Si/Ti ¼ 77); reactant, 1 g; reactant/H2O2 (mol), 3; solvent (CH3CN), 10 g; temperature, 353 K; time, 8 h; 3-MP, 3-methyl pentane; 2,2-DMB, 2,2-dimethyl butane.

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

87

with varying critical diameter, selected from 3-methylpentane (3-MP) or 2,2-dimethylbutane (2,2-DMB), or cyclohexane was carried out in the presence of TS-2 with dilute H2O2. As the size of the competing alkane II increases, its relative conversion (vis-a`-vis n-hexane) decreases, the reactivity order being n-hexane . 3-MP . 2,2-DMB . cyclohexane. From the point of view of chemical reactivity in unconstrained or homogeneous catalytic systems, the reverse trend is expected. Further, although the critical diameters of 2,2-DMB and cyclohexane are comparable (0.60 and 0.61 nm, respectively), 2,2-DMB competes better with n-hexane than with cyclohexane. Apparently, not only the size but also the shape and/or conformation of the reactants may play a role in competitive hydroxylations; the results highlight the importance of steric factors in the adsorption process. Similar results were obtained with TS-1 catalyst. In contrast to their vanadosilicate analogues, the titanosilicate molecular sieves do not hydroxylate the terminal primary carbon in n-alkanes. Ramaswamy et al. (218,219) found that when n-hexane was hydroxylated under identical conditions in the presence of TS-2 or VS-2 (VS-2 is a vanadium analogue of TS-2), the distribution of products was as follows: TS-2 : hexan-2-ol ð52%Þ . hexan-3-ol ð48%Þ ðno activation at 1-positionÞ VS-2 : hexan-2-ol ð45%Þ . hexan-3-ol ð42%Þ . hexan-1-ol ð13%Þ Furthermore, the -ol/-one ratio was also higher when the catalyst was TS-2 (0.77) than when it was VS-2 (0.36). The pathways for reaction catalyzed by the titano- and vanadosilicates are probably different. The absence of hydroxylation of the primary C – H bond and the higher -ol/-one ratio when the catalyst is the titanosilicate is significant. Because the homolytic bond dissociation energies decrease in the order primary C – H . secondary C – H . tertiary C –H bonds, radical pathways involving C – H bond homolysis almost always show a marked preference for the functionalization of tertiary and secondary C– H bonds (220). The preference for secondary C – H bonds and the high -ol/-one ratios when the catalyst is TS-2 suggest that radical pathways are involved in the hydroxylation of alkanes with TS-2. In fact, Khouw et al. (221) had earlier proposed a possible mechanism for alkane hydroxylation catalyzed by TS-1 which proceeds via homolytic Hz abstraction from R – H by a Ti(O2H) group which may have some superoxo-like character (Scheme 14). This Hz abstraction generates an alkyl radical, Rz, and is accompanied by reduction of Ti4þ to Ti3þ. A subsequent homolytic O –O bond cleavage occurs to form the C –O bond. In support of the above mechanism, the following results may be mentioned: (i) superoxo radicals have indeed been observed in oxidation reactions catalyzed by titanosilicates (51,52,54,131,205,222); (ii) Ti4þ ions are reduced to Ti3þ in the presence of reducing agents such as CO (122), H2, and hydrocarbons

88

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Scheme 14.

(51,52,130,131), or at high pH (205); (iii) the preference of secondary over primary C – H bonds in the hydroxylation of alkanes; and (iv) the high -ol/-one ratios in the oxidation of cyclohexane. The titanyl group (TiyO) proposed by Khouw et al. (221), has, so far, not been observed experimentally during oxidation catalyzed by titanosilicates. The hydroxylation of octane and cyclohexane catalyzed by Ti-MMM-1, a mixed- phase material (TS-1 and Ti-MCM-41) containing both micro- and mesopores, with aqueous H2O2 was reported by Poladi et al. (223). Ti-MMM-1 was found to be more active and selective in these hydroxylations than either Ti-MCM-41 or TS-1; the yield of alcohol was higher (Table XXVII). The detailed crystallographic and textural structure of this mixed phase material is not clear. It seems likely that the higher activity (conversion) is a consequence of the presence of mesopores (of the MCM-41 phase) leading into the micropores (of the MFI phase); these mesopores would enhance the diffusion of the reactants deep into the crystallites while simultaneously preserving the advantages of the microporous MFI phase (such as higher intrinsic activity and selectivity). In the absence of the mesopores of the MCM-41 phase, a significant portion of the interior of the crystallite would have been inaccessible to the reactants. Similarly, the high selectivity for alcohols, the primary oxidation product, is a consequence of their faster diffusion out of the solid crystallite through the mesopores. In the absence of the mesopores, the alcohol molecules diffusing more slowly through the pores of the MFI phase would undergo further oxidation to the ketone before emerging from a catalyst particle. The advantages of a mixed phase catalyst are thus evident. One major advantage of Ti-MMM-1 is that it allows the application of aqueous H2O2 as the oxidant. Apparently most

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

89

TABLE XXVII Comparative activity of mixed phase Ti-MMM-1 with TS-1 and Ti-MCM-41 for the oxidation of cyclohexane and n-octane Conversiona (mol%)

Ketone(s) (mol%)

Alcohol(s) (mol%)

Othersb

Ketone:alcohol

Cyclohexane Ti-MMM-1

9.2

35.1

54.7

13.2

0.64

TS-1

4.2

26.4

27.6

46.0

0.96

Ti-MCM-41

1.9

9.8

17.0

73.2

0.58

Ti-MMM-1

19.8

14.5

80.8

4.7

0.18

TS-1

13.3

10.3

80.3

9.4

0.13

2.9

21.5

52.7

25.8

0.41

n-Octane

Ti-MCM-41

Adapted from Poladi and Landry (223). a Conversion ¼ (moles of alkane converted/total moles of alkane) £ 100. b Includes diols and diones.

of the catalysis occurs in the TS-1 phase, which, being hydrophobic, is quite stable in aqueous media. The role of the MCM-41 phase is mainly to facilitate the transport of reactants and products to and from the active sites of TS-1. Other mesoporous titanosilicates suffer from their instability in an aqueous medium, and therefore, have to be used with TBHP or other alkyl hydroperoxides, with the attendant environmental problems. Hence, if the hydrothermal stability, absence of titanium leaching, and catalytic superiority of this mixed phase material is validated thoroughly, it will be a significant addition to the family of titanosilicate-containing oxidation catalysts. V.D.3. Hydroxylation of Aromatic Compounds The selective hydroxylation, in the presence of aqueous H2O2, of aromatic hydrocarbons such as benzene, toluene, and xylene to phenol, cresols, and xylenols, respectively, occurs easily on TS-1 (33,165,224). Again, a significant contrast between TS-2 and VS-2 in the oxidation of toluene is that when the catalyst is the former, only aromatic ring hydroxylation takes place, although when the catalyst is VS-2, the side chain C –H bonds are also hydroxylated (165, 218,219,225,226) (Table XXVIII). When the alkyl substituent contains secondary C – H bonds, both ring and side chain oxidation at the secondary C – H bond occur. Thus, ethylbenzene gives

90

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169 TABLE XXVIII Hydroxylation of aromatics over TS-2 and VS-2 molecular sieves Benzene

Conversion (mol%)

Toluene

TS-2

VS-2

TS-2

VS-2

51.3

21.6

39.6

35.1

Products (mol%) Phenol

88.0

90.0

–

–

p-Benzoquinone

9.0

7.0

–

–

o-Cresol

–

–

36.0

20.0

p-Cresol

–

–

59.0

17.0

Benzyl alcohol

–

–

–

8.0

Benzaldehyde

–

–

–

52.0

Others

3.0

3.0

5.0

3.0

Adapted from Kumar et al. (165). Reaction conditions: catalyst (TS-2: Si/Ti ¼ 77; VS-2: Si/V ¼ 79), 0.1 g; reactant, 1 g; reactant/H2O2 (mol) ¼ 3.0; solvent (CH3CN), 10 g; temperature, 333 K; time ¼ 8 h.

ethyl phenols (40%), acetophenone (56%), and 2-phenyl ethanol (4%). Monosubstituted benzenes with electron-donating groups (such as phenol, toluene, etc.) undergo rapid hydroxylation (mainly in the ortho and para positions), although those containing electron-withdrawing groups (such as Cl, NO2, etc.) do not react so facilely (165). Similarly, bulky substituents, such as tert-butyl, retard the reaction because of the steric restriction imposed by the pore size of the TS-1. An increased selectivity for phenol in the oxidation of benzene by H2O2 with TS-1 catalyst in sulfolane solvent was attributed to the formation of a bulky sulfolane – phenol adduct which cannot enter the pores of TS-1. Further oxidation of phenol to give quinones, tar, etc. is thus avoided. Removal of Ti ions from the surface regions of TS-1 crystals by treatment with NH4HF2 and H2O2 was also found to improve the activity and selectivity (227). The beneficial effects of removal of surface Al ions on the catalytic performance of zeolite catalysts for acid-catalyzed reactions have been known for a long time.

V.E. Oxidation of Nitrogen-Containing Compounds As expected from the Lewis acidity of Ti4þ, the titanosilicates strongly adsorb and oxidize basic nitrogen-containing compounds with a lone pair of electrons localized on the N atom. By contrast, nitrogen oxides (NOx) and nitro compounds

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

91

TABLE XXIX Oxidation of primary amines catalyzed by TS-1 Amine

Solvent

Conversion

Oxime selectivity

H2O2 efficiency

CH3NH2

CH3OH

40

88

90

a

CH3NH2

CH3OH

3

0

0

n-C3H7NH2

CH3OH

32

73

86

i-C3H7NH2

CH3OH

38

77

88

b

29

74

85

c

31

84

90

i-C3H7NH2

t-BuOH

i-C3H7NH2

t-BuOH

C6H11NH2

CH3OH

3

33

8

C6H11NH2

t-BuOH

3

32

8

C6H5CH2NH2

CH3OH

20

82

55

Adapted from Reddy and Jacobs (228). Reaction without catalyst. b t-BuOH, tert-butyl alcohol. c Reaction over TS-2. a

(both aliphatic and aromatic) are not reactive in the TS-1/H2O2 system; nitrobenzene, for example, is not oxidized to nitrophenols. The following oxidations occur: (i) NH3 to NH2OH (14); (ii) primary amines to oximes (Table XXIX, Scheme 15) (228); (iii) secondary amines to nitrones (229); (iv) tert-amines to the corresponding nitrogen oxides (33); and (v) anilines to azoxybenzenes (230): NH3 þ H2 O2 ! NH2 OH þ H2 O

Scheme 15.

ð28Þ

92

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169 TABLE XXX Ammoxidation of carbonyls over TS-1 (Si/Ti ¼ 29) Reactant

Time (h)

Conversion (mol%)

Selectivity (%)

Acetone

6.0

79.7

98.1

Hex-3-one

4.0

70.4

98.1

Methylisobutyl ketone

3.0

98.0

99.5

Cyclohexanone

4.0

98.2

96.4

p-Tolualdehyde

2.0

97.0

97.7

Benzaldehyde

2.5

97.0

99.4

Adapted from Kumar et al. (165). Reaction conditions: catalyst (TS-1; Si/Ti ¼ 29), 1.5 g; reactant, 10 g; reactant: H2O2:NH3 ¼ 1:1.2:2.0; solvent (tert-butanol), 40 g; temperature, 343 K.

R1 R2 CH – NH2 ! R1 R2 CH – NHOH ! R1 R2 CH – NO ! R1 R2 CyNOH

ð29Þ

R1 R2 CH – NHR3 ! R1 R2 CyNðOÞR3

ð30Þ

R3 N ! R3 NO

ð31Þ

C6 H5 NH2 ! C6 H5 NðOÞyN – C6 H5

ð32Þ

The oxidation of NH3 to NH2OH forms the basis of a process for the ammoximation of cyclohexanone to the oxime because the NH2OH formed in solution readily reacts with the ketone (non-catalytically) to give the oxime (231). Table XXX (165) illustrates the conversions and selectivites obtained for a few typical ketones and aldehydes. The ammoximation of aldehydes is faster than that of ketones. The oxime selectivity is also higher. The ammoximation of cyclohexanone by this method offers a more eco-friendly alternative route to the cyclohexanone oxime intermediate for the production of Nylon-6. The current route coproduces large quantities of ammonium sulfate and involves the use of hazardous chemicals such as oleum, halides, and oxides of nitrogen. One of the major problems in all the ammoximation processes using aqueous H2O2 þ TS-1 with NH3 is that, under the basic conditions (pH $ 10) prevailing during the reaction, some of the lattice Si ions of the zeolite structure in TS-1 are leached into solution, leading to catalyst destruction. This leaching is a common characteristic of all silicates. Innovative catalyst formulations and process modifications are needed to overcome this problem.

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

93

V.F. Oxidation of Sulfur-Containing Compounds Similar to nitrogen compounds, electron-rich sulfur compounds, such as the sulfides, with the lone pair of electrons on the sulfur atom, are oxidized to sulfoxides and, further, to sulfones by the H2O2/titanosilicate sytem (218,232, 233). Table XXXI (232) illustrates typical conversions and product selectivities for various sulfides for the reactions catalyzed by TS-1. Bulky sulfides such as alkyl, phenyl sulfides are relatively unreactive because of their steric exclusion from the pores of TS-1. Diphenyl sulfide could not be oxidized at all. As the diffusivity and, hence, the conversion of the sulfide decreases, the further oxidation of the primary product (sulfoxide) becomes more competitive, leading to increased formation of the corresponding sulfone (Table XXXI): R2 S ! R2 SO ! R2 SðOÞ2

ð33Þ

Promising results in the oxidation of sulfides with mesoporous SBA-15 type titanium silicates with hydrolytic stability in aqueous H2O2 were obtained by Trukhan et al. (233). Their structural and textural parameters are given in Table XXXII along with those of Ti-MMM, a mesoporous, mesophase material of the MCM-41 type (29,229). The oxidation of methylphenyl sulfide (MPS) was chosen as a test reaction. The SBA-15 samples had a highly ordered hexagonal arrangement of mesopores (with a diameter about 11 nm). XPS, XANES, and DRUV spectra indicated (234) that most of the Ti4þ ions in the Ti-SBA-15 (Fig. 36) and Ti-MMM samples are in an octahedral environment. Ti ions in Ti-SBA-15 are present both as oligomerized titanium-oxygen species and as segregated TiO2 (anatase) particles. The presence of anatase in Ti-SBA-15 containing 7.17 wt% Ti was also confirmed by Raman spectroscopy (Fig. 37) by the strong peak at 145 cm21 characteristic of anatase. The absence of this Raman peak in the spectrum of Ti-MMM (containing 1.9 wt% Ti) indicated that the Ti ions in it are more dispersed than those in Ti-SBA-15. One difference between TABLE XXXI Oxidation of sulfides with H2O2 catalyzed by TS-2 Reactant

Conversion (%)

Selectivity (%) Sulfoxide

Sulfone

CH3 –S–CH3

100

97

3

C2H5 –S–C2H5

100

85

15

C6H5 –S–CH3

98

78

22

C6H5 –S–C2H5

70

75

15

Adapted from Reddy et al. (232).

94

Sample no.

Ti content (wt %)

Si/Ti (atomic ratio)

pHa

Structural parameters Unit cell parameter (nm)

FWHMb

Textural parameters

Specific surface area (m2/g) Mesopore

External

Specific mesopore volume (cm3/g)

Mesopore diameter (nm)

Wall thicknessc

1

2.05

38

3.18

12.25

0.054

573

30

1.34

10.6

1.7

2

4.00

19

2.61

12.46

0.061

619

37

1.40

10.9

1.6

7.17

10

2.78

12.84

0.033

514

44

1.10

10.9

2.0

1.89

39

9.00

4.23

0.110

1260

29

0.90

3 d

Ti-MMM

Adapted from Trukhan et al. (234). pH in the final mixture. b FWHM, full width at half maximum of the (100) reflection. c Calculated from the equation unit cell parameter ¼ mesopore diameter þ wall thickness. d Mesoporous mesophase material of the MCM-41 type. a

3.45

0.8

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

TABLE XXXII Structural and textural parameters of Ti-SBA-15 catalysts

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

95

Fig. 36. UV –visible diffuse reflectance spectra and elemental analysis data for Ti-SBA-15: (1) sample 1; (2) sample 2; (3) sample 3; and (4) sample 1 after treatment with 30% H2O2 [from Trukhan et al. (234)].

the Ti-MMM and Ti-SBA-15 samples is that, as a consequence of the greater wall thickness in the latter (1.6 – 2.0 vs. 0.8 nm, Table XXXII), a greater fraction of the Ti ions in Ti-SBA-15 are inaccessible to the reactants, as was confirmed by infrared spectra of CO adsorbed on these samples (Fig. 38). Three types of bands

Fig. 37. Ambient-temperature Raman spectra of Ti-MMM, Ti-SBA-15 (samples 1–3), and TiO2 (anatase); p , plasma line [from Trukhan et al. (234)].

96

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Fig. 38. Infrared spectra of adsorbed CO for samples with similar titanium contents: (1) Ti-MMM and (2) Ti-SBA-15 (sample 1) [from Trukhan et al. (234)].

were observed (Fig. 38), at 2137 cm21 (physically adsorbed CO), 2153 cm21 (complexes of CO with Si – OH groups), and 2179 cm21 (CO on Ti4þ). The 2179 cm21 band is clearly seen only for Ti-MMM, indicating that the surface concentration of Ti4þ is considerably higher for Ti-MMM than for Ti-SBA-15 with the same Ti content. The catalytic activities of Ti-MMM, Ti-SBA-15, and TS-1 are compared in Table XXXIII (234). The activities of these titanoslicates for MPS oxidation are in the order Ti-MMM . Ti-SBA-15 . TS-1. The catalytic activity was found to correlate with the rate of H2O2 decomposition in the absence of the organic reactant (Fig. 39). Ti-MMM on which H2O2 decomposed (to H2O and O2) faster (curve b) was also more active in the oxidation of the sulfur-containing compounds (Table XXXIII). Among the Ti-SBA-15 samples, the activity decreased in the order, sample 1 . sample 2 . sample 3. The intensity of the broad band in the 200 – 350 nm DRUV spectra of these samples also follows the same order (Fig. 36) and is a rough measure of the dispersion of Ti in the sample. The higher catalytic activity of Ti-MMM was ascribed to its greater surface Ti concentration.

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

97

TABLE XXXIII Thioanisole (MPS) oxidation with 30% aqueous H2O2 over Ti-SBA-15 and other Ti, Si-catalysts Catalyst

Time (h)

MPS conversion (%)

Product distribution (%) Sulfoxide

Sulfone

None

1

6

–

–

1

1

48

72

28

1 (second cycle)

1

47

73

27

1 (third cycle)

1

50

72

28

2

1

26

73

27

2

3

53

65

35

1

29

71

29

100

76

24

29

79

21

3 a

Ti-MMM

0.5

TS-1 (Ti, 2.54 wt%)

1

Adapted from Trukhan et al. (234). Reaction condition: MPS, 0.1 M; [MPS]/[H2O2] ¼ 1/1.1; CH3CN, 3 mL; Ti, 6 £ 1023 mmol, 292 K. Structural and textural properties of Ti-SBA-15 (1–3) and Ti-MMM catalysts are given in Table XXXII. a [MPS]/[H2O2] ¼ 1/1.3.

Fig. 39. H2O2 conversion profiles: (a) for reaction catalyzed by Ti-SBA-15 (sample 1, 30 mg) and (b) by Ti-MMM (33 mg). Reaction conditions: H2O2, 1.29 mmol; Ti, 0.013 mmol; MeCN, 3 ml; T ¼ 353 K [from Trukhan et al. (234)].

98

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Contrary to what was observed with Ti-MMM (29), no loss of catalytic activity was observed after the recycling of the Ti-SBA-15 catalyst (Table XXXIII), a result that confirms the hydrolytic stability of the Ti-SBA-15 materials. It was verified that there was no leaching of Ti during the catalytic reaction (by hot filtration of catalyst and testing the filtrate for catalytic activity) (29). Elemental analysis after the catalytic runs confirmed that the total Ti content remained the same for Ti-MMM (236), Ti-SBA-15, and TiO2 –SiO2 mixed oxides (30). A comparison of the DRUV spectra recorded before and after the treatment with aqueous H2O2 indicated that, in contrast to the observations for Ti-MMM (236), Ti-MCM-41 (237), TiO2 – SiO2 mixed oxides (30), and TS-1 (228), there was no change in the Ti-SBA-15 (221). The higher hydrolytic stability could not be attributed to a lower hydrophilicity of Ti-SBA-15 because the specific H2O adsorption capacity was similar for both Ti-MMM and Ti-SBA-15 (Fig. 40). We emphasize that the above results have been observed only in the oxidation of sulfides and phenols, reactions known to follow radical mechanisms. A thorough investigation of the catalytic potential of the materials in other oxidation reactions (epoxidation, hydroxylations, etc.) is warranted. One of the major challenges in the petroleum industry today is the removal of sulfur compounds, especially refractive ones such as 4,6-dimethyldibenzothiophene (DMDBT), from petroleum fractions such as diesel to concentrations , 5– 10 ppm from the current values of 50 –500 ppm. The current technology is hydrodesulfurization catalyzed by cobalt –nickel –molybdenum sulfides at high pressures. Reducing sulfur concentratios in diesel fuels below 5 – 10 ppm

Fig. 40. Water adsorption on Ti-SBA-15 (180 mg) and Ti-MMM (202 mg) [from Trukhan et al. (234)].

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

99

will impose a heavy economic penalty as a consequence of the high H2 partial pressures that will be required to remove the DMDBTs. Hulea et al. (238) demonstrated the ability of Ti-beta and Ti-HMS to oxidize the thiophenic compounds to their corresponding sulfoxides and sulfones (with H2O2 as the oxidant), which are then removed by conventional liquid – liquid separation technology. The use of high-pressure equipment and the consumption of large quantities of H2 can be avoided by this route. TS-1, as expected, exhibits low activity as a consequence of the restricted access of DMDBT to the active sites. Both Ti-beta and Ti-HMS catalysts exhibited high activities for the removal of sulfur compounds from kerosene by mild oxidation with H2O2 (238). The best results were obtained with acetonitrile as the polar solvent, because the oxidized compounds (sulfoxides and sulfones) were fully soluble in this solvent (and they are only partially soluble in ethanol and water) (Table XXXIV). During the chemical treatment, the oxidized organic sulfur compounds (such as the sulfoxides and sulfones of dibenzothiophene and DMDBT) transfer completely to the polar solvent, which is immiscible with kerosene. The oxidized product is then recovered from the solvent, and the latter is recycled to the oxidation reactor. TABLE XXXIV Influence of catalyst and nature of solvent on the sulfur removal from kerosene (T ¼ 343 K) Catalyst

Solvent

Reaction time (h)

Phase

Sulfur (ppm)

Sulfur removal (%)

-

Acetonitrile

Extraction

Kerosene

1220

7.0

Ti-HMS

Acetonitrile

9

Kerosene

190

85.5

Ti-HMS

Acetonitrile

9

Acetonitrile

Ti-beta

Acetonitrile

5

Kerosene

Ti-beta

Acetonitrile

5

Acetonitrile

Ti-beta

Ethanol

5

Kerosene

390

70.2

Ti-beta

Ethanol

10

Kerosene

300

77.0

Ti-beta

Ethanol

24

Kerosene

250

81.0

Ti-beta

Ethanol

24

Kerosenea

80

94.0

Ti-beta

Ethanol

24

Ethanol

Ti-beta

Water

10

Kerosene

840

36.0

Ti-beta

Water

10

Kerosene

300

77.1

Ti-beta

Water

10

Water

450

Adapted from Hulea et al. (238). a Kerosene washed with acetonitrile.

2500 80

94.0

2300

1800

100

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

V.G. Oxidation of Oxygen-Containing Compounds V.G.1. Alcohols The oxidation of primary alcohols to aldehydes and secondary alcohols to ketones proceeds smoothly on TS-1 and Ti-beta. On TS-1, because of diffusion constraints, the oxidation rate decreases with reactant chain length, and linear alcohols are oxidized faster than branched and cyclic alcohols, contrary to the trends observed in homogeneous systems (198). By analogy with transition metal complexes, it has been supposed (111) that intermediates such as that illustrated in Scheme 16 can be responsible for the oxidation of alcohols with H2O2. In the absence of diffusional constraints, Ti-beta exhibits (240) activity and selectivity trends similar to those observed in homogeneous systems. Rates increase with chain length, and cyclic/branched alcohols are more reactive than linear alcohols. When alkyl substituents are introduced near the carbon atom bearing the OH group, the reactivity of the molecule decreases, the decrease being more pronounced when the number of such alkyl groups is increased. These results are in agreement with the cyclic intermediate proposed in Scheme 16 and reflect the importance of the steric restrictions to form the transition state complex at the Ti sites on the reactivity of molecular sieves. The apparent activation energy was the same (70 kJ/mol) for both TS-1 and Ti-beta, indicating that the oxidation of alcohol proceeds on both catalysts through similar cyclic intermediates (239,240). V.G.2. Ethers The oxidation of both linear and cyclic ethers to the corresponding acids and lactones by aqueous H2O2 as catalyzed by TS-1 and TS-2 was reported by Sasidharan et al. (241) (Scheme 17 and Table XXXV). The titanosilicates exhibited significantly better activity (about 55% conversion) and selectivity (98%) than chromium silicates, although vanadium silicates totally failed to catalyze the reaction. Such conversions are usually accomplished using either stoichiometric amounts of chromium trioxide, lead tetraacetate, or ruthenium tetroxide as oxidants (242) or catalytic amounts of RuO4 in the presence of

Scheme 16.

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

101

Scheme 17.

hypochlorite or periodate (243). The use of solid catalysts such as TS-1 has significant environmental and economic advantages. V.G.3. Phenols When aromatic compounds containing a phenolic OH group are brought in contact with titanosilicates in the presence of H2O2, two reactions are possible: the first is the hydroxylation of the aromatic ring to give diphenols (Section V.D). When the electron density in the ring is high (as in polyalkyl phenols) and the ortho- and/or para position (with respect to the OH group) is vacant, the formation of ortho- or para-benzoquinone also occurs. Indeed, in the hydroxylation of phenol to catechol and hydroquinone, one of the major side products (and the main cause of the tar formation) is the formation of benzoquinones and products derived from them. The benzoquinones of polyalkylbenzenes are starting materials for many products in the photographic and fine chemicals industries. Trukhan et al. (234) reported the oxidation of 2,3,6-trimethylphenol (TMP) to trimethylbenzoquinone (TMBQ) catalyzed by Ti-SBA-15, Ti-MMM, or TS-1 with aqueous H2O2 used as a reactant (Table XXXVI). The Ti-SBA-15 samples with higher Si/Ti ratios, which according to their diffuse reflectance UV spectra have higher dispersions of

TABLE XXXV TS-1 catalyzed oxidation of various ethers with 30% H2O2 Reactant

Product

Yield (%)a

Dibutyl ether

Butyric acid

54

Benzyl methyl ether

Benzoic acid

65

Tetrahydrofutan

g-Butyrolactone

55

Tetrahydropyran

d-Valerolactone

42

Dihydropyran

d-Valerolactone

40

1,4-Dioxan

Keto-1,4-dioxane

5

Sasidharan et al. (241). Isolated yield and the rest is essentially unreacted ether.

a

102

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169 TABLE XXXVI 2,3,6-Trimethylphenol (TMP) oxidation with 30% aqueous H2O2 Catalyst

Time (h)

TMP conv. (%)

TMBQ yield (%)

None

6

0

0

Ti-SBA-15 (38)

6

57

43

Ti-SBA-15 (19)

6

43

29

Ti-SBA (10)

6

31

30

Ti-MMM (39)

0.4

100

77

TS-1 (33.4)

6

14

8

Adapted from Trukhan et al. (234). Reaction conditions: TMP, 0.1 mol; TMP/H2O2, 0.28; CH3CN, 3 mL; temperature, 353 K; Ti, 1.3 £ 1022 mmol. Values in parentheses refer to Si/Ti ratios. TMBQ, trimethylbenzoquinone.

titanium species, exhibited a higher catalytic activity. The higher catalytic activity of Ti-MMM was also thought to arise from the higher dispersion of Ti in Ti-MMM. Apart from TMBQ, the main byproduct was the C – C coupling dimer, 2,20 ,3,30 ,6,60 -hexamethyl-4,40 -biphenol. A small amount of the C – O coupling dimer was also found. Experiments with fast catalyst filtration at the reaction temperature confirmed (Fig. 41) that no further reactant conversion occurred in the filtrate after catalyst removal, indicating that the oxidation takes place on the catalyst surface and is a true heterogeneous process. V.G.4. Ketones, the Baeyer – Villiger Oxidation Baeyer –Villiger (BV) oxidation, induced by a peroxy acid or a H2O2/Lewis acid system, organometallics, and metalloenzymes is an important reaction for synthesizing lactones or esters from ketones. Bhaumik et al. (244) reported that TS-1 is an efficient catalyst for BV oxidation of cyclic and aromatic ketones (such as cyclohexanone and acetophenone, respectively) (Scheme 18, Tables XXXVII and XXXVIII). Conversions and yields were higher in the absence of any solvent in the triphase (solid catalyst along with two immiscible liquid reactants (ketone þ aqueous H2O2). The addition of a few drops of H2SO4 increased the yield of the BV products. The titanium peroxo species, a Brønsted acid stabilized by the presence of protic solvent was proposed by the authors to be responsible for the BV reaction. In accordance with this proposal, Wang et al. (245) later found that the Brønsted acid HZSM-5(Al) was also more active than TS-1 in BV oxidation of cyclopentanone to d-valerolactane. The conversions of the ketone and yield of the lactone were 47 and 15% for HZSM-5 vs. 35 and 10% for TS-1.

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

103

Fig. 41. TMP oxidation catalyzed by Ti-SBA-15, after filtration of the catalyst (full squares) and without filtration (open squares) [from Trukhan et al. (234)].

Scheme 18.

104

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169 TABLE XXXVII Oxidation of cyclohexanone catalyzed by TS-1 Phasea

System

Conv. (mol%)

Product selectivity (mol%)

1-Caprolactone

Hydroxyketone

Diketone

Cyclohexene þ Epoxide

TS-1/H2O2/Hþ

Tri

64.0

45.2

17.0

14.0

23.8

TS-1/H2O2/Hþ

Bi

30.2

28.4

25.5

31.0

15.1

TS-1/H2O2

Tri

31.0

19.6

31.3

33.6

15.5

TS-1/H2O2

Bi

5.0

–

64.0

36.0

–

Adapted from Bhaumik et al. (244). Reaction conditions: reactant:H2O2 ¼ 1:1; catalyst (TS-1, Si/Ti ¼ 29), 20 wt% with respect to reactant; temperature, 353 K. a Tri: solid catalyst þ two immisible liquid phases (organic reactant þ H2O2 in water); bi: solid catalyst þ one homogeneous liquid phase (organic reactant þ aqueous H2O2 þ CH3CN as cosolvent).

TABLE XXXVIII Baeyer– Villiger rearrangement and hydroxylation of acetophenone catalyzed by TS-1/H2O2 system Phasea

System

Product selectivities (mol%)b

Conv. (%) PA

o-HAP

p-HAP

PH

CA

HQ

AA

TS-1/H2O2/Hþ

Tri

31.0

49.7

16.6

16.0

7.0

1.0

1.1

8.6

TS-1/H2O2

Tri

7.0

27.0

2.8

5.6

12.6

7.4

12.3

32.3

61.0

–

–

4.6

10.8

4.4

19.0

–

–

–

–

6.9

-

24.9

–

–

–

þ

TS-1/H2O2/H

Bi

6.1

TS-1/H2O2

Bi

–

þ

Bi

5.5

þ

Mono

–

Blank/H2O2/H Blank/H2O2/H

31.8 –

–

–

2.8

3.5

–

–

– 30.1 –

Adapted from Bhaumik et al. (244). Reaction conditions: reaction time, 12 h; reactant:H2O2 ¼ 1:1; catalyst (TS-1, Si/Ti ¼ 29), 20 wt% with respect to reactant; temperature, 353 K. a Tri: solid catalyst þ two immisible liquid phases (organic reactant þ H2O2 in water); bi: solid catalyst þ one homogeneous liquid phase (organic reactant þ aqueous H2O2 þ CH3CN as cosolvent). b PA, phenyl acetate; o-HAP, o-hydroxy acetophenone; p-HAP, p-hydroxy acetophenone; PH, phenol; CA, catechol; HQ, hydroquinone; AA, acetic acid.

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

105

Scheme 19.

V.H. CyN Cleavage Reactions Titanium silicate molecular sieves not only catalyze the oxidation of CyC double bonds but can be successfully employed for the oxidative cleavage of carbon – nitrogen double bonds as well. Tosylhydrazones and imines are oxidized to their corresponding carbonyl compounds (243) (Scheme 19). Similarly, oximes can be cleaved to their corresponding carbonyl compounds (165). The conversion of cyclic dienes into hydroxyl ketones or lactones is a novel reaction reported by Kumar et al. (165) (Scheme 20). Thus, when cyclopentadienes, 1,3cyclohexadiene, or furan is treated with aqueous H2O2 in acetone at reflux temperatures for 6 h in the presence of TS-1, the corresponding hydroxyl ketone or lactone is obtained in moderate to good yields (208).

V.I. Acid-Catalyzed Reactions Acid catalysis by titanium silicate molecular sieves another area characterized by recent major progress. Whereas only two categories of acid-catalyzed reactions (the Beckmann rearrangement and MTBE synthesis) were included in the review by Notari in 1996 (33), the list has grown significantly since then. In view of the presence of weak Lewis acid sites on the surfaces of these catalysts, they can be used for reactions that require such weak acidity.

Scheme 20.

106

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

V.I.1. Beckmann Rearrangement The transformation of oximes to lactams (the Beckmann rearrangement) was one of the earliest such acid-catalyzed reactions to be reported with TS-1 (138) and TS-2 (247) catalysts. The rearrangement of cyclohexanone oxime to 1-caprolactam proceeds with high selectivity in the presence of TS-1, with high catalyst stability (138,247). V.I.2. Synthesis of Polycarbonate Precursors Recently, Srivastava et al. (248, 249) reported the novel application of TS-1 and Ti-MCM-41 in the synthesis of polycarbonate precursors such as cyclic carbonates and dimethyl/diphenyl carbonates, avoiding toxic chemicals such as phosgene or CO. With either TS-1 or Ti-MCM-41, cyclic carbonates were prepared in high yields by cycloaddition of CO2 to epoxides such as epichlorohydrin, propene oxide, and styrene oxide at low temperatures and pressures (Scheme 21, Table XXXIX). Although TS-1 and Ti-MCM-41 showed similar activity for epoxides of smaller dimensions (such as epichlorohydrin and propene oxide) (compare runs 1 and 3 and 5 and 7, Table XXXIX), Ti-MCM-41 was more active for cycloaddition of CO2 to the larger styrene epoxide (compare runs 9 and 10, Table XXXIX). Although most of the experiments reported in Table XXXIX were conducted with CH2Cl2 as solvent, similar (or better) yields were obtained, even in the absence of any solvent (runs 3, 6, and 10, Table XXXIX). However, the product was slightly colored. At higher temperatures/pressures/ reaction periods (e.g., 413 K, 24 bar, and 24 h), HPLC analyses showed the formation of methanol-insoluble solid aliphatic polycarbonates. Apparently the cyclic carbonate monomer had polymerized to give polycarbonates under the influence of the weak acidity of the TS-1 system. In addition to the main cyclic carbonate, the side products in the case of epichlorohydrin included 3-chloro-1, 2-propanediol, and 3-chloropropanaldehyde. The cyclic carbonate could also be synthesized directly from the alkenes in the same reactor by reacting the alkenes in the presence of Ti-MCM-41 with a mixture of an epoxidizing agent (such as H2O2 or tert-butyl hydroperoxide) and

Scheme 21.

TABLE XXXIX Synthesis of cyclic carbonates from epoxides and CO2

1

Catalyst

TS-1

Co-catalyst

Temperature (8C)

Run time (h)

Epoxide

Conv. of epoxide (mol%)

TOF

Selectivity for cyclic carbonate (mol%)

DMAP

120

4

EC

85.4

790

92.6

160

4

EC

94.2

872

97.0

2

TS-1 (3rd recycle)

DMAP

120

4

EC

77.0

713

90.4

3

TS-1 (no solvent)

DMAP

120

4

EC

89.6

829

97.5

4

TiMCM-41

DMAP

120

4

EC

78.8

938

84.0

5

TS-1

DMAP

120

6

PO

66.8

412

84.6

160

6

PO

94.0

580

83.0

6

TS-1 (no solvent)

DMAP

120

6

PO

77.6

719

88.1

7

TiMCM-41

DMAP

120

6

PO

63.7

758

91.2

8

TS-1

DMAP

120

6

BO

76.6

354

70.9

9

TS-1

DMAP

120

8

SO

44.7

166

45.5

10

TiMCM-41

DMAP

140

10

SO

98.1

584

73.1

11

TiMCM-41(no solvent)

DMAP

140

10

SO

595

82.0

100

From Srivatsava et al. (248). Reaction conditions: catalyst (TS-1: Si/Ti ¼ 36, Ti-MCM-41: Si/Ti ¼ 46), 100 mg; co-catalyst, 0.0072 mmol; epoxide, 18 mmol; CH2Cl2, 20 mL; CO2, 6.9 bar. DMAP: N,N-dimethylaminopyridine; EC: epichlorohydrin; PO: propylene oxide; SO: stytene oxide; BO: a-butylene oxide; TOF: turnover frequency (moles epoxide converted per mole of Ti per hour.

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Run no.

107

108

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169 TABLE XL Synthesis of cyclic carbonates from alkenes: epoxidation-cum-cycloaddition

Catalyst

Alkene

Oxidizing agent

Stage 1: alkene to epoxide

Stage 2: epoxide to cyclic carbonate

Alkene conversion to epoxide (%)

Epoxide selectivity (%)

Epoxide conversion (%)

Cyclic carbonate selectivity (%) 55.6

TS-1

Allyl chloride

H2O2

54.6

100.0

92.5

TS-1

Styrene

H2O2

50.4

89.0

49.2

TiMCM-41

Allyl chloride

TBHP

13.3

TiMCM-41

Styrene

TBHP

44.0

100 93.1

100 97.2

26.0 100 83.4

Adapted from Srivatsava et al. (248). Runs with TS-1 (Si/Ti ¼ 36; 400 mg) were carried out with 26.2 mmol alkene, 0.0072 mmol DMAP, 14.7 mmol 50% H2O2 and CO2 (6.9 bar) in acetone (20 mL). Runs with TiMCM-41 (Si/Ti ¼ 46; 100 mg) were carried out with 8 mmol alkene, 0.0036 mmol DMAP, 8 mmol 40% TBHP in CH2Cl2 and CO2 (6.9 bar) in acetonitrile (6.4 g).

CO2 (Table XL). A conversion of 54.6% and cyclic carbonate selectivity of 55.6% were obtained when allylchloride was the reactant. Some ring-hydrolyzed products were also detected. With styrene, a conversion of 50.4% and cyclic carbonate selectivity of 26% were obtained. When the reaction was conducted with TiMCM-41 as the catalyst and TBHP as the oxidizing agent, the conversions of alkenes to epoxides (stage 1) were lower (allylchloride conversion ¼ 13.3% and styrene conversion ¼ 44%), but the further conversion of epoxide formed during the reaction to cyclic carbonate (stage 2) was almost 100% (Table XL). As expected, TiMCM-41, with its larger pore diameter, was more active and selective than TS-1 for the cycloaddition of CO2 to the epoxide (stage 2, rows 2 and 4, Table XL). Aromatic polycarbonates are currently manufactured either by the interfacial polycondensation of the sodium salt of diphenols such as bisphenol A with phosgene (Reaction 1, Scheme 22) or by transesterification of diphenyl carbonate (DPC) with diphenols in the presence of homogeneous catalysts (Reaction 2, Scheme 22). DPC is made by the oxidative carbonylation of dimethyl carbonate. If DPC can be made from cyclic carbonates by transesterification with solid catalysts, then an environmentally friendlier route to polycarbonates using CO2 (instead of COCl2/CO) can be established. Transesterifications are catalyzed by a variety of materials: K2CO3, KOH, Mg-containing smectites, and oxides supported on silica (250). Recently, Ma et al. (251) reported the transesterification of dimethyl oxalate with phenol catalyzed by Sn-TS-1 samples calcined at various temperatures. The activity was related to the weak Lewis acidity of Sn-TS-1 (251).

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

109

Scheme 22.

The transesterifications of chloropropene carbonate and propene carbonate with methanol and phenol catalyzed by TS-1, Ti-MCM-41, and TiO2 (Table XLI) have been reported (248). Neither TiO2 nor TS-1 showed any activity in the transesterification reactions. Ti-MCM-41 catalyzed the reaction with a high selectivity for DMC (86%). Ti-MCM-41 also catalyzes the transesterification of cyclic carbonates with phenols (Table XLI).

TABLE XLI Transesterification of cyclic carbonates with CH3OH and phenol catalyzed by Ti-MCM-41 Cyclic carbonate

ROH

Cyclic carbonate conversion (mol%)

DMC selectivity (mol%)a

Chloropropylene carbonate

CH3OH

26.5

86.2

Propylene carbonate

CH3OH

5.1

Propylene carbonate

C6H5OH

58.9

DPC selectivity (mol%)a

24.4

Adapted from Srivatsava et al. (248). Reaction conditions: for reactions with methanol (3.2 g)— catalyst (TiMCM-41: Si/Ti ¼ 46), 400 mg; cyclic carbonate, 1.36 g; temperature, 393 K, reaction time ¼ 2 h. For reactions with phenol (4.7 g) reaction time ¼ 17 h and rest all are the same. a Balance is phenyl ether.

110

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

V.I.3. Transesterification of Esters Transesterification is a crucial step in several industrial processes such as (i) production of higher acrylates from methylmethacrylate (for applications in resins and paints), (ii) polyethene terephthalate (PET) production from dimethyl terephthalate (DMT) and ethene glycol (in polyester manufacturing), (iii) intramolecular transesterifications leading to lactones and macrocycles, (iv) formation of alkoxy esters (biodiesel) from vegetable oils, and (v) cosynthesis of dimethyl carbonate (an alkylating agent, octane booster, and precursor for polycarbonates) and ethene glycol from ethene carbonate and methanol (252,253). Other than mineral acids and bases, compounds such as metal alkoxides (aluminum isopropoxide, tetraalkoxytitanium, (RO)Cu(PPh3)n, PdMe(OCHCF3 Ph(dpe)), organotin alkoxides, etc.), non-ionic bases (amines, dimethylaminopyridine, guanidines, etc.), and lipase enzymes also catalyze these transformations (252). Tatsumi et al. (254) reported the synthesis of dimethylcarbonates from ethene carbonate and methanol using K-TS-1 as a solid base catalyst. The transesterification of dimethyl oxalate with phenol has also been reported recently (251). TS-1 and Ti-MCM-41 catalyze transesterification reactions of aliphatic esters selectively (152). Acidity measurements (infrared spectra of adsorbed pyridine and TPD of NH3) had revealed the presence of only weak Lewis acid sites on these samples. Catalytic activity was found to parallel the acid strength. Both increased in the order TS-1 , Ti-MCM-41 , amorphous TiO2 –SiO2. TS-1 catalyzed the transesterifications (Tables XLII and XLIII) of linear esters (ethylacetoacetate and diethylmalonate), but failed for cyclic esters such as propene carbonate. Ti-MCM-41 and amorphous TiO2 –SiO2 were found to be superior for the cyclic esters (Tables XLIV and XLV). The catalysts could be recycled without any loss in activity/selectivity. V.I.4. Carbon –Carbon Bond Formation Reactions The Mukaiyama-type aldol reactions (255) between silyl enol ethers and aldehydes to give b-hydroxy esters/aldols provide a facile method for C – C bond formation. They are facilitated by a variety of Lewis acids, including TiCl4, SnCl4, and ZnCl4, used in either stoichiometric or catalytic amounts under homogeneous conditions. A few solid catalysts, such as Nafion-117, zeolite Ca – Y, montmorillonite clay, and SiO2 – Al2O3, have also been reported to be active for these reactions (256). Sasidharan and Kumar (257) recently investigated the Mukaiyama-type reactions with a variety of metallosilicates including TS-1 and Al-free Ti-beta. Michael addition reactions of silyl enol ethers with various a,b-unsaturated carbonyl compounds were also investigated with these catalysts. In the Mukaiyama aldol reaction of methyl trimethylsilyl dimethylketene acetal

Entry 1

2

ROH

Transester product

Conv. (mol%)

Product yield (%)

95.6

92.9

100

87.1

3

97.6

90.7

4

99.2

85.0

5

96.2

84.3

(Continued)

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

TABLE XLII Transesterification of ethylacetoacetate with various alcohols (ROH) over TS-1

111

112

Entry

Conv. (mol%)

Product yield (%)

6

96.4

95.3

7

86.4

69.5

8

83.1

66.9

9

ROH

CH3(CH2)7CHyCH(CH2)7CH2OH

Transester product

87.6a

From Srinivas et al. (152). Reaction conditions: catalyst (TS-1; Si/Ti ¼ 33), 130 mg; ethylacetoacetate, 5 mmol; ROH, 15 mmol; temperature ¼ 383 K, run time ¼ 4 h. a Isolated yield.

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

TABLE XLII Continued

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

113

TABLE XLIII Transesterification of diethyl malonate with various alcohols catalyzed by TS-1

ROH

Conversion (%)

Selectivity (%)

Products distribution (%) Mono

Di

n-Propanol

97.5

98.4

26.6

73.4

n-Butanol

99.3

97.0

16.0

84.0

b

n-Butanol

95.8

100

46.4

53.6

n-Butanol (recycle I)a

95.0

100

45.9

54.1

n-Butanol (recycle II)

94.4

100

46.6

53.4

n-Hexanol

99.7

100

10.9

89.1

a

n-Octanol Isobutanol Cyclohexanol Benzyl alcohol

100 95.6 100 84.2

100 96.3 100 88.8

82.6

17.4

58.0

42.0

34.4

65.6

39.0

61.0

From Srinivas et al. (152). Reaction conditions: catalyst (TS-1; Si/Ti ¼ 33), 130 mg; diethyl malonate, 5 mmol (0.8 g); ROH, 15 mmol; temperature, 383 K; run time, 12 h. a Reaction conditions are same except the temperature, 353 K.

(silyl enol ether) with benzaldehyde (Scheme 23) catalyzed by various metallosilicates, TS-1 and Ti-beta gave the highest yields (85 – 87%) of the product b-hydroxy ester (aldol) (Table XLVI). The number of turnovers for different isomorphously substituted metallosilicates followed the order Ti . Sn . V . Al. Table XLVII illustrates the 1,4-Michael addition of various a,b-unsaturated carbonyl compounds with silyl enol ether (Scheme 24). The reactions were carried out in the absence of H2O or H2O2. The product yields mentioned in Tables XLVI and XLVII are isolated yields; the selectivity for the aldols as well as the Michael addition products was always 100%, regardless of conversion, and no side products were observed. Among the various solvents investigated, tetrahydrofuran was found to be the best. The authors attributed the excellent activity of TS-1 and Ti-beta in the aldol condensation and Michael addition reactions to the “oxophilic Lewis acidity” of Ti4þ ions (257).

114

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

TABLE XLIV Comparative activity of TS-1, Ti-MCM-41 and amorphous TiO2 – SiO2 in transesterification of (a) ethylacetoacetate with benzyl alcohol and allylalcohol and (b) diethylmalonate with allylalcohol Catalyst

Benzyl alcohol Ester conversion (mol%)

Allylalcohol

Transester yield (%) Ester conversion (mol%)

Transester yield (%)

(a) Ester– ethylacetoacetate (Run time ¼ 4 h) TS-1

86.4

69.5

83.1

69.8

Ti-MCM-41

93.7

90.2

85.2

84.5

Amorphous TiO2-SiO2

95.2

91.9

87.3

86.1

Monotransester selectivity (%)

Ditransester selectivity (%)

(b) Ester –diethylmalonate (Run time ¼ 12 h); alcohol–n-butanol TS-1

58.2

57.4

59.0

41.0

Ti-MCM-41

66.5

65.6

87.2

12.8

Amorphous TiO2-SiO2

68.6

66.7

89.2

10.8

Adapted from Srinivas et al. (152). Reaction conditions: ester, 5 mmol; alcohol, 15 mmol; catalyst, 130 mg; temperature, 383 K.

V.I.5. Formation of Pinacols The name “pinacol” denotes vicinal diols with four alkyl groups; when all the alkyls are methyl, it is called pinacol (CH3)2C(OH) –C(OH)(CH3)2. These compounds are the starting materials for the manufacture of many pesticides, pharmaceuticals, fragrances, photographic chemicals, and crop protection chemicals. They are usually made by dihydroxylation of alkenes by OsO4 or KMnO4. Both of these toxic reagents are used in stoichiometric quantities. Another strategy to make these 1,2-diols is reduction of aldehydes and ketones with reactive metals such as Na, Mg, or Al. But many side products are formed as a result of coupling reactions. Sasidharan et al. (258) reported the formation of pinacols from alkenes catalyzed by various titanosilicates. Aluminum-free Ti-beta exhibited better activity and pinacol selectivity than TS-1, TS-2, Ti-MCM-22, or mesoporous Ti-MCM-41 (Table XLVIII). The side products included the pinacolone, alcohol, and oligomers. The epoxide was the initial product, which underwent acid-catalyzed nucleophilic ring-opening by H2O molecules to give the pinacol (Scheme 25).

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

115

TABLE XLV Transesterification of cyclic propylenecarbonate with different alcohols and phenol catalyzed by titanosilicates Titanosilicate

ROH

Reaction time (h)

Conversion of propylene carbonate (mol%)

Selectivity of transester product (mol%)a –

TS-1

Methanol

2

Nil

TiMCM-41

Methanol

2

5.1

Phenol

8

58.9

24.4

Methanol

4

71.4

48.2

8

86.0

51.2

Ethanol

8

73.0

61.8

Propanol

12

86.3

69.4

n-Butanol

12

85.0

73.4

n-Hexanol

12

49.0

61.5

Amorphous TiO2-SiO2

Adapted from Srinivas et al. (152). Reaction conditions: catalyst (TS-1 or TiMCM-41), 400 mg; propylene carbonate, 1.36 g, ROH (alcohol, 3.2 g; phenol, 4.7 g); temperature, 393 K. Reaction conditions: catalyst (amorphous titanosilicate), 400 mg; propylene carbonate, 1.02 g (0.01 mol); ROH, 0.1 mol; temperature, 423 K. a Balance is the corresponding ether.

V.I.6. Oxidative Dehydrogenation The oxidative dehydrogenation of propane to give propene catalyzed by TS-1, Ti-beta, Ti-MCM-41, TiO2-silicalite-1, or others was investigated by Schuster et al. (259). TS-1 was the best catalyst, with a selectivity of 82% for propene at a propane conversion of 11% (Fig. 42). Sulfation of TS-1 by H2SO4 prior to the reaction increased the conversion to 17%, with a selectivity of about 74%. Although conversion of propane was higher on Ti-beta and Ti-MCM-41, selectivity for propene was much lower; CO2 was the main product. Lewis acid sites were considered to be the major active sites (259).

Scheme 23.

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

116

TABLE XLVI Activities of various metallosilicates for Mukaiyama aldol reaction of benzaldehyde with silyl enol ether Si/M ratio (product)

Particle size (mm)

Micropore volume (mL/g)

Yield (%)

TONa

33.5

0.1–0.2

0.138

85.0

9.2

Sn-ZSM-5

73.5

0.2–0.4

0.132

25.0

7.35

V-ZSM-5b

86.0

0.4–0.6

0.135

10.0

1.9

b

Catalyst Ti-ZSM-5 (TS-1)b b

40.0

0.3–0.5

0.153

-

-

b

43.0

0.2–0.3

0.269

87.0

11.6

b

Al-beta

26.7

0.3–0.4

0.274

29.0

1.3

Sn-ZSM-12b

78.0

1–2

0.169

15.0

4.5

2.5 –3.0

0.5–0.7

0.343

-

-

2.5 –3.0

0.5–0.7

0.292

37.0

6.5d

2.5 –3.0

0.5–0.7

0.270

50.0

7.3d

40.0

0.3–0.5

0.141

34.0

4.4e

H-ZSM-5 Ti-beta

Na–Y c

La– Y

Re–Yc c

Zn/ZSM-5

Adapted from Sasidharan and Kumar (257). Reaction conditions: catalyst, 150 mg; methyl trimethylsilyl dimethylketene acetal (silyl enol ether), 10 mmol; benzaldehyde, 10 mmol; dry THF as dispersion medium, 10 mL; temperature, 333 K; reaction time, 18 h. Yield refers to the isolated product yield. a Moles of product per mole of metal per hour. b The metal atom is substituted in the tetrahedral position. c La ¼ 2.3 wt%; combination of all the rare-earth metals ¼ 2.85 wt% and Zn ¼ 2.63 wt%. d TON based on rare-earth metals. e TON based on Zn.

V.J. Photocatalysis V.J.1. Photocatalytic Degradation of Pollutants The oxidation of small concentrations of aromatic compounds in industrial effluents using UV radiation and catalysts such as TiO2 is gaining in importance (260). The hydroxyl radicals generated on TiO2 under UV irradiation are the agents of photodegradation. To increase the efficiency of the process, the TiO2 has been dispersed on SiO2 (261). Titanosilicates such as TS-1 and Ti-beta have two inherent advantages as photodegradation catalysts: (i) they are hydrophobic and, hence, adsorb selectively the aromatic pollutants from aqueous effluents, thereby facilitating the photocatalytic efficiency for charge transfer from the catalyst to the pollutants; (ii) the high surface area and atomic dispersion of Ti enable an efficient use of the metal. Kang et al. (262) compared TiO2 and two samples of TS-2 (TS-2 and TS-2h) in the photodegradation of various phenols.

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

117

TABLE XLVII Michael addition of various a,b-unsaturated carbonyl compounds to silyl enol ether catalyzed by Ti-beta and TS-1 a,b-Unsaturated carbonyl compounds

Product

Product yield (%)a Ti-beta

TS-1

Methyl methacrylate, (2a)

3a

53.0

47.0

Ethyl methacrylate, (2b)

3b

41.0

35.0

2-Ethylhexyl acrylate, (2c)

3c

39.0

36.0

2-Hydroxyethyl methacrylate, (2d)

3d

41.0

39.0

Methyl vinyl ketone, (2e)

3e

45.0

49.0

Cyclohexenone, (2f)

3f

39.0

36.0

2-Methylcyclohexenone, (2g)

3g

35.0

33.0

Adapted from Sasidharan and Kumar (257). Reaction conditions: catalyst, 150 mg; methyl trimethylsilyl dimethylketene acetal (silyl enol ether), 10 mmol; a,b-unsaturated carbonyl compounds, 10 mmol; dry THF, 10 mmol; reaction temperature, 333 K; reaction time, 14 h. Structures of a,b-unsaturated carbonyl compounds (2a–2g) and products (3a–3g) are shown in Scheme 24. a Isolated yield by column chromatography and the rest is unconverted starting material.

The surface areas of the three catalysts were 58 (TiO2), 360 (TS-2), and 550 (TS-2h) m2/g, respectively. UV-irradiation of solutions (1024 M) containing 4chlorophenol (4-CP) in the presence of suspended TiO2, TS-2, or TS-2h yielded time-dependent spectra from which the concentration of unconverted 4-CP was estimated. Figure 43 is a plot of the relative concentration of 4-CP as a function

Scheme 24.

118

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169 TABLE XLVIII Formation of pinacol over various titanium-silicates

Catalysta

Conv. (mol%)

H2O2 selectivity (%)

Product selectivity (%) Epoxide

Pinacol

Pinacolone

DMBb

Othersb

Ti-beta (43)

55.3

80.1

1.3

92.9

1.9

4.3

0.5

Ti-Al-beta (40)

51.2

76.5

1.1

82.6

3.7

15.6

0.4

TS-1 (33)

39.2

61.5

3.9

88.0

1.3

1.0

5.9

TS-2 (46)

21.2

57.0

4.0

83.6

1.2

1.9

9.0

Ti-MCM-22 (51)

22.6

54.5

3.4

86.0

2.0

5.4

5.1

Ti-MCM-41 (50)

48.2

65.0

1.6

96.3

1.1

0.7

0.3

Adapted from Sasidharan and Kumar (258). Reaction conditions: 2,3-dimethyl-2-butene, 10 mmol; H2O2 (31 wt% aqueous solution), 10 mmol; catalyst, 20 wt% with respect to substrate; water (as dispersion medium), 5 mL; temperature, 333 K; reaction time, 6 h. a The figures in the parentheses represent the Si/Ti ratios. b DMB, 2, 3-dimethyl-2-butanol and “others” include oligomers.

of irradiation time for the three catalysts. The activity decreases in the order TS-2h . TS-2 . TiO2. Notwithstanding the lower surface Ti concentration (by about 19%) and the larger band gap of the TS-2 catalysts relative to TiO2, the photodecomposition rate is enhanced on TS-2 and TS-2h. The greater photoreactivity was attributed to the increased adsorption of 4-CP resulting from

Scheme 25.

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

119

Fig. 42. Catalyst screening for the oxidative dehydrogenation of propane to propene. T ¼ 823 K; molar ratios C3H8/O2/N2/H2O ¼ 5/25/25/45; GHSV ¼ 1300 h21; mcat ¼ 1:4 2 8:0 g; vcat ¼ 5 ml [from Schuster et al. (259)].

Fig. 43. Time dependence of the relative concentration of 4-CP at 225 nm of illuminated 4CP aqueous solutions in the presence of TiO2, TS-2, and TS-2h catalysts in suspension [from Kang et al. (262)].

120

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

the greater hydrophobic surface areas of TS-2 and TS-2h as well as their greater total surface areas relative to TiO2. V.J.2. Photocatalytic Synthesis Reduction of CO2 with H2O to give useful chemicals using sunlight is one of the holy grails in solar energy-to-fuels and chemicals conversion. Towards this goal, Anpo et al. (263) used Hg lamp radiation (l . 280 nm) to reduce CO2 with H2O to CH4 and CH3OH at 328 K using titanosilicate molecular sieves, TS-1, Ti-MCM-41, and Ti-MCM-48 (Fig. 44). The order of reactivity was Ti-MCM-48 . Ti-MCM-41 . TS-1 . TiO2. The Ti-containing zeolites led to the formation of considerable amounts of the CH3OH, although the formation of the CH4 was found to be the major reaction on bulk TiO2 (Fig. 44). Although both Ti-MCM-41 and Ti-MCM-48 are mesoporous, the pore geometry is threedimensional in the latter and one-dimensional in the former. Addition of Pt onto Ti-MCM-48 increased its photocatalytic activity. However, only the formation of CH4 is promoted, being accompanied by a decrease in the CH3OH yields (Fig. 45). Anpo et al. (263) proposed that CO2 is reduced to CO and subsequently to C radicals although H2O photodecomposes to H and OH radicals. Reaction of OH and H with the carbon species yields CH3OH and CH4, respectively (263). The mechanism of CO2 photoreduction in TS-1 with methanol as the electron donor was also investigated by Ulagappan and Frei (264), who used in situ FTIR

Fig. 44. Yields of CH4 and CH3OH in the photocatalytic reduction of CO2 with H2O on TiO2 powder: (a) TS-1; (b) Ti-MCM-41; (c) Ti-MCM-48; and (d) zeolite catalysts [from Anpo et al. (263)].

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

121

Fig. 45. The effects on Pt-loading on the yields of CH4 and CH3OH in the photocatalytic reduction of CO2 with H2O on Ti-MCM-48 zeolite catalyst: (a) Ti-MCM-48; (b) Pt-loaded Ti-MCM-48 (0.1 wt% Pt); and (c) Pt-loaded Ti-MCM-48 (1.0 wt% Pt) [from Anpo et al. (263)].

spectroscopy. The reaction was induced by 266-nm excitation of the Ti4þ – O22 ! Ti3þ –O2 ligand-to-metal charge transfer transition of the framework center. HCO2H, CO, and HCO2CH3 were the observed products. The CO originates from secondary photolysis of HCO2H, although HCO2CH3 is formed by the spontaneous Tischenko reaction of HCHO, which is the initial oxidation product of methanol. HCO2H is the primary 2-electron reduction product of CO2 at the Ti centers, a result that suggests that C – H bond formation occurs in the initial steps of CO2 activation. V.J.3. deNOx Reactions TS-2 exhibited high photocatalytic activity (with a 75-W high-pressure Hg lamp) for the direct decomposition of NO into N2 and O2 and N2O at 275 K (265), with a high selectivity (76%) for the formation of N2. The yields (in mmol/g of TiO2 h) of N2 and N2O were 12 and 4, respectively. In the case of isolated Ti ions in 4-fold coordination present in TS-2, charge transfer excited complexes (Ti3þ –O2)p are formed under UV irradiation. Electron transfer from Ti3þ to the p-antibonding orbital of NO takes place, and simultaneously the electron transfer from the p-bonding orbital of another NO into the holetrapped center (O2) occurs. These electron-transfer processes lead to the direct decomposition of two sets of NO on the (Ti3þ – O2) species, to selectively form N2 and O2 (265). On the other hand, when Ti ions are present in an aggregated

122

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

form (as in anatase), the photoformed holes and electrons are rapidly separated from each other. This separation prevents the simultaneous activation of two NO molecules on the same active site, resulting in the formation of N2O and NO2 instead of N2 and O2 (265).

V.K. Influence of Solvents Solvents are usually used to keep both reactants and products in a single phase. Apart from enabling the proper mixing of the reactants, solvents can also affect conversions and product selectivities through interaction with the active sites and the transition state. The influence of the dielectric constant of the solvent on the mode of cleavage of the O – O bond in H2O2 (hetero- vs. homolytic cleavage) and consequently on product distribution was mentioned above (Section V.B). The influence of solvents on oxidation reactions catalyzed by TS-1 had been investigated by both experimental (111,266,267) and theoretical (63,268,269) methods. Atoguchi and Yao (267) examined the effect of solvents (various mixtures of H2O and CH3OH) on the oxidation of phenol catalyzed by TS-1 both experimentally (Table XLIX) and by DFT calculations for cluster models made up of the Ti center having the tetrahedral structure, Ti(OSiH3)4, a H2O2, and a solvent molecule. Water addition to methanol increases the dielectric constant of the reaction medium and accelerates the catalytic oxidation of phenol (increasing the conversion from 43.6 to 70.2%). The amount of dihydroxy benzenes increases from 4.3 to 6.6 mmol (Table XLIX). Additional results of the enhancement in phenol conversion (to dihydroxy benzenes) and oxidation of allyl alcohol (to glycidol and allylic oxidation products) catalyzed by TS-1 in various solvents are illustrated in Fig. 46. In solvents with high dielectric constants, the heterolytic cleavage of the O –O bond

TABLE XLIX Phenol oxidation over TS-1 in H2O and methanol mixture solvent Dielectric constant (1)

Phenol conversion (%)

Hydroquinone þ catechol (mmol)

Hydroquinone/ catechol

Selectivity (%)a

85.6: 14.4

39.2

43.64

4.33

1.99

91.97

77.0: 23.0

43.1

51.67

5.06

1.82

91.76

42.8: 57.2

58.9

70.16

6.58

1.44

87.24

CH3OH:H2O (wt%)

Adapted from Atoguchi and Yao (267). Reaction conditions: catalyst, 0.2 g; phenol, 1.0 g; 30% aqueous H2O2, 1.2 g; solvent, 5 g; temperature, 349 K; time, 3 h. a Selectivity (%) ¼ {(produced hydroquinone þ catechol)/(consumed phenol)} £ 100 (mol/mol).

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

123

Fig. 46. Influence of solvent dielectric constant (logarithm (ln) values) on (a) phenol hydroxylation [data taken from Thangaraj et al. (266)] and (b) epoxidation of allyl alcohol catalyzed by TS-1 [data from Wu and Tatsumi (229)].

is probably dominant when the partially or totally ionic intermediates are stabilized by the solvent. Solvents can influence the reactivity of the hydroperoxo titanium species Ti(O2H) either through changes in the dielectric constant of the reaction medium (as discussed in Section V.B.1) or by specific coordination to the Ti center. Care should be taken to distinguish between the two effects in investigations of solvent effects. Furthermore, molecules such as acetone and acetonitrile are oxidized by H2O2, forming 2-hydroxy-2-hydroperoxy propane and peroxyimidic acid CH3 – CyNH(OOH), respectively (269– 270), affecting the rate of the oxidation and H2O2 selectivity. An interesting observation reported in Table XLIX is the increase in the hydroquinone/catechol ratio from 1.44 to 1.99 when the dielectric constant of the medium is decreased from 58.9 to 39.2 by addition of methanol to water. A similar increase in the hydroquinone/catechol ratios was also observed in phenol hydroxylation catalyzed by TS-1 (266) in dioxane-water and tert-butyl alcoholwater mixtures. The para/ortho ratio increased nearly 10-fold when 10% dioxane was added to water. Similarly, the para/ortho ratio more than doubled (1.3 – 3.0) when 10% tert-butyl alcohol was added to water. An opposite trend, namely, a decrease in the para/ortho ratio from 1.4 to 0.6, was observed when 10% formamide ð1 ¼ 108Þ was added to water. Because of geometric constraints in the MFI pores, catechol is expected to be formed more easily on the external surface of TS-1 crystallites than in the pores (91). Hydroquinone, less spatially demanding, can form in the TS-1 channels. A greater coverage of the hydrophobic

124

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

external surface of the TS-1 crystals by the less polar solvents having lower dielectric constants (alcohol and dioxane in alcohol – water and dioxane – water mixtures, respectively) will suppress the formation of catechol. The influence of solvents of varying polarity in the epoxidation of allyl alcohol catalyzed by TS-1, Ti-beta, and Ti-MWW is shown in Table L (229). The surface of TS-1 is hydrophobic. Hydrophilic molecules such as H2O do not compete with the reactant, allyl alcohol, for either diffusion in the pores or coordination at the Ti site. On the contrary, hydrophobic solvent molecules do compete with reactant molecules for adsorption on the hydrophobic TS-1 surface. Hence, the conversion of the reactant is higher in hydrophilic H2O than in hydrophobic isopropanol. No steric constraints in the pores are anticipated for any of the above solvents. Ti-beta is relatively more hydrophilic and, hence, it is not surprising that H2O inhibits the conversion strongly. In general, large-pore and mesoporous Ti-silicates behave similarly to amorphous Ti – SiO2 catalysts in this respect. Hence, TBHP is a better oxidant than aqueous H2O2 for such materials. Ti-MWW is even more active and selective. Conversion follows a trend similar to that observed for TS-1, except that it is the highest when CH3CN is the solvent. Selectivity to the epoxide is high for reaction catalyzed by either TS-1 or Ti-MWW. The selectivity of Ti-beta is low, especially in solvents of high coordinating ability such as the higher alcohols.

V.L. Influence of Silylation One of the reasons for the low selectivity of the mesoporous Ti silicates is their surface hydrophilicity, which is caused by the presence of a large number of surface Si – OH and Ti – OH groups. Because these mesoporous materials are better suited than TS-1 to the oxidation of large, bulky molecules, the passivation of these OH groups (e.g., by silylation) may improve catalyst activity and selectivity. Attempts have been made to reduce the concentrations of such OH groups by silylating them with various alkyl silanes (Table LI) (273). The treatment leads to a significant improvement in alkene conversion in cyclohexene epoxidation in the case of Ti-MCM-41 and Ti-MCM-48 (273). Although epoxide selectivity improved in the former case, there was a decrease in the latter. In the case of hexane oxidation, silylation did not improve the conversion. An enhancement in the number of turnovers and selectivity for the epoxide on silylation was also observed in the cyclohexene epoxidation with TBHP catalyzed by Ti-SBA-15 (Table LII) (274). Ti-SBA-15 was claimed to be thermally more stable than Ti-MCM-41. Ti leaching was absent. A better understanding of the changes in surface structure during silylation is needed before the potential advantages of silylation of these mesoporous materials are realized. A potential pitfall in silylation reactions is the silylation of

Solvent

Ti-MWW (Si/Ti ¼ 46) (mol%) AA conv.

MeCN

87.0

a

Prod. Sel.

TS-1 (Si/Ti ¼ 36) (mol%)

H2 O2

AA conv.

Gly.

Others

Conv.

Eff.

99.9

0.1

87.9

99.0

26.8

a

Prod. Sel.

Ti-beta (Si/Ti ¼ 42) (mol%)

H2O2

AA conv.

Gly.

Others

Conv.

Eff.

82.6

17.3

28.5

94.1

13.9

Prod. Sel.a

H2O2

Gly.

Others

Conv.

Eff.

75.4

24.6

18.4

75.5

Water

82.3

99.9

0.1

84.3

97.6

34.6

96.0

4.0

36.6

94.5

2.8

92.6

7.4

9.6

29.2

MeOH

34.5

75.7

24.3

35.9

96.1

34.2

86.6

13.4

36.2

94.5

16.7

42.0

58.0

21.6

77.3

EtOH

32.5

91.0

9.0

33.0

98.5

24.4

94.6

5.4

29.8

81.8

15.1

59.5

40.5

28.6

52.8

1-PrOH

30.1

96.0

4.0

37.5

80.3

12.6

95.6

4.4

16.1

78.6

–

–

–

–

–

Acetone

41.5

96.7

3.3

42.5

97.6

31.0

92.8

7.2

36.6

84.7

11.9

41.4

58.6

26.3

45.2

Dioxane

27.8

96.0

4.0

28.6

97.2

–

–

–

–

–

5.2

78.3

21.7

6.5

80.0

Adapted from Wu and Tatsumi (229). Reaction conditions: catalyst, 70 mg; allyl alcohol (AA), 10 mmol; H2O2, 10 mmol; solvent, 5 mL; temperature, 333 K; time, 0.5 h. a Gly, glycidol; others, solvolysis products, glycerol and alkyl glycerol ethers, etc.

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

TABLE L Epoxidation of ally alcohol with H2O2 in various solvents

125

Catalysta

TON (mol/(mol Ti))

Selectivity (%)

H2O2 decom (%)

Alcohol

Ketone

Epoxide

Diol

Cyclohexene Ti-MCM-41 (nonsil)b c

Ti-MCM-48 (nonsil) Ti-MCM-41(sil)

b

Ti-MCM-48 (sil)b

0.72

5.4

30.0

15.2

0

54.7

57.6

2.1

6.1

26.7

32.8

4.7

35.7

61.9

13.3

112.1

14.4

21.0

13.9

50.7

0

38.5

120.9

21.3

17.0

2.2

59.4

0

2-ol

3-ol

2-one

3-one

Hexane oxidation Ti-MCM-41 (nonsil)c

0

0

–

–

–

–

Ti-MCM-41 (sil)c

0.06

0.5

40.5

59.5

0.0

0.0

0.0

0.2

6.8

22.0

22.9

31.0

24.1

97.3

0

0

–

–

–

–

75.0

0.17

0.52

45.4

54.6

0

0

20.2

Ti-MCM-41 (sil)

d c

Ti-MCM-48 (nonsil) Ti-MCM-48 (sil)

b

74.7

Adapted from Tatsumi et al. (273). a Ti-MCM-41 (non-sil) (Si/Ti ¼ 123, SBET ¼ 1015 m2/g, pore diameter ¼ 2.32 nm, pore volume ¼ 0.88); Ti-MCM-41 (sil) (Si/Ti ¼ 139, SBET ¼ 879 m2/g, pore diameter ¼ 1.90 nm, pore volume ¼ 0.82); Ti-MCM-48 (non-sil) (Si/Ti ¼ 47, SBET ¼ 1048 m2/g, pore diameter ¼ 2.32 nm, pore volume ¼ 0.91); TiMCM-48 (sil) (Si/Ti ¼ 51, SBET ¼ 839 m2/g, pore diameter ¼ 1.90 nm, pore volume ¼ 0.71). b Reaction conditions: catalyst, 50 mg; reactant, 25 mmol; H2O2, 5 mmol; temperature, 323 K; time, 3 h. c Catalyst, 50 mg; reactant, 25 mmol; H2O2, 5 mmol; temperature, 323 K; time, 2 h. d Catalyst, 50 mg; reactant, 100 mmol; H2O2, 20 mmol; temperature, 353 K; time, 16 h.

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Conv. (mol% of max.)

126

TABLE LI Effect of trimethylsilylation on catalytic activity of Ti-containing mesoporous molecular sieves

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

127

TABLE LII Influence of silynation on epoxidation of cyclohexene with TBHP over Ti-SBA-15 Sample

Si/Ti

Ti-SBA1 Ti-SBA2

Conversion (mol%)

TON (mol/mol Ti)

Product selectivity (mol%) Oxide

Diolsa

Allylicb

80

1.9

76

83.9

6.5

9.7

27

2.8

38

86.9

6.8

6.3

c

87

21.1

843

97.0

2.0

1.9

Ti-SBA2-silc

28

29.8

437

96.0

2.4

1.6

Ti-SBA1-sil

Adapted from Wu et al. (274). Reaction conditions: catalyst, 0.05 g; cyclohexene, 30 mmol, TBHP (70%), 30 mmol; CH3CN, 10 mL; temperature, 333 K; time, 2 h. a 1,2-Cyclohexanediols. b Products of allylic oxidation: 2-cyclohexen-1-ol and 2-cyclohexen-1-one. c Trimethylsilylated by refluxing in hexamethyldisilazane/toluene for 2 h.

the Ti –OH groups in the tripodal Ti centers. Such Ti – OH groups play an essential role in the formation and reactivity of the titanium oxo groups (Sections III and IV). Their elimination by silynation will lead to a reduction in the number of active sites. Elimination of the Si – OH groups without affecting the Ti – OH groups is difficult and may account for some of the conflicting results of silynation reported in the literature.

VI. Structure-Activity Correlations The majority of the titanium ions in titanosilicate molecular sieves in the dehydrated state are present in two types of structures, the framework tetrapodal and tripodal structures. The tetrapodal species dominate in TS-1 and Ti-beta, and the tripodals are more prevalent in Ti-MCM-41 and other mesoporous materials. The coordinatively unsaturated Ti ions in these structures exhibit Lewis acidity and strongly adsorb molecules such as H2O, NH3, H2O2, alkenes, etc. On interaction with H2O2, H2 þ O2, or alkyl hydroperoxides, the Ti ions expand their coordination number to 5 or 6 and form side-on Ti-peroxo and superoxo complexes which catalyze the many oxidation reactions of NH3 and organic molecules.

VI.A. Structure of Titanium Species and Activity Attempts have been made to find correlations between the types and concentrations of the various surface groups and titanium oxo complexes, on the one hand,

128

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

and their catalytic activity and selectivity on the other. The concentration of framework Ti ions was the earliest structural parameter to be related to the catalytic activity of TS-1. The original patent of Taramasso et al. (5) itself claimed that the intensity of the 960-cm21 band, indicative of the concentration of Ti in framework positions, is related to the catalytic activity. Later, Thangaraj et al. (275) observed that the catalytic activity of TS-1 in phenol conversion was proportional to the molar ratios x ( ¼ Ti/(Si þ Ti)) at low Ti concentrations ðx , 0:02Þ and suggested that these Ti ions are responsible for the observed catalytic activity. A similar conclusion was also reached by Mantegazza et al. (276), who observed that at low Ti concentrations the activity of TS-1 (represented by the turnover number) in ammonia oxidation, cyclohexanone ammoximation, and propene epoxidation was proportional to the mole fraction of Ti in the framework (276). There is, hence, a consensus that, on TS-1 (and probably Ti-beta), tetrapodal Ti ions in framework tetrahedral positions are responsible for the catalytic activity. On Ti-MCM-41 (and probably other similar mesoporous materials), the XANES/XAFS investigations of Thomas and Sankar (104) show that the tripodal Ti centers are responsible for catalytic activity in the conversion of cyclohexene to its epoxide with TBHP as the oxidant. From their in situ XAFS data, these authors concluded that during the catalytic reaction the original four-coordinated Ti4þ centers in the tripodal species expand their coordination sphere to six (Section V.C.5). Chaudhari et al. (277) had observed a linear dependence of H2O2 selectivity on Ti content in Ti-MCM-41 in the hydroxylation of 1-naphthol to 1,2-dihydroxy naphthalene with aqueous H2O2 (Fig. 47). Both XAS and EPR results had indicated the presence of mainly the tripodal titanium sites on Ti-MCM-41. As a consequence of the large surface area of the material, these sites are well dispersed, leading to the linear dependence of catalytic activity on Ti content. Such detailed structural information about surface Ti species is not available for other Ti –SiO2 mesoporous materials. The results of Guidotti et al. (189) (Section V.C.5) indicate that catalytic reactions on these materials involving peroxide are complex processes and other titanium oxo species may also be involved.

VI.B. Titanium-Oxo Species and Activity If the tetra- and tripodal Ti structures and the titanium oxo species derived from these structures in the presence of ROOH (R ¼ H, alkyl) are involved as active sites and reaction intermediates, the next step beyond their identification is to seek correlations between the structure and concentrations of these titanium oxo species and catalytic activity and selectivity. Clerici and Ingallina (204) were the first to propose the Ti(O2H) group as the active site of alkene epoxidation by

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

129

Fig. 47. Catalytic selectivity as a function of Ti content in Ti-MCM-41 for 1-naphthol hydroxylation with aqueous H2O2. H2O2 selectivity (mol%) ¼ (number of moles of H2O2 utilized in product (1, 4-naphthoquinone, 1,4-dihydroxynaphthalene and 1,2-dihydroxynaphthalene) formation/number of moles of H2O2 fed) £ 100 [data from Chaudhari et al. (277)].

H2O2 in TS-1. On the basis of the observed solvent and acid/base effects on the kinetics and yield in alkene epoxidation in various alcohols, an end-on (1) group with a simultaneously coordinated alcohol group was envisioned as the reactive intermediate. A direct correlation between the concentration of the titanium oxo species and epoxidation activity was proposed by Lin and Frei (133). Loading TS-1/H2O2 with propene after evacuation, they observed by FTIR difference spectroscopy the loss of the bands characterizing propene (at 1646 cm21) and TiOOH (at 837 and 3400 cm21). Figure 48 is the infrared difference spectrum recorded immediately after loading the propene on TS-1/H2O2; Fig. 49 includes the spectra recorded 80 and 320 min later. The disappearance of the propene bands was not noticed when H2O2 (and consequently TiOOH) was not present. After 80 min, the product spectrum included bands at 830, 895, 1372, 1409, 1452, 1460 and 1493 cm21. The product spectrum was similar to that obtained when a sample of propene oxide was loaded onto TS-1. The rate of decay of the 837-cm21 absorption (O– O vibration of TiOOH) was accompanied by the growth of the infrared bands of the product. These observations led Lin and Frei to conclude that the TiOOH group was

130

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Fig. 48. Infrared difference spectrum recorded immediately after loading of 6.5 mbar propene gas into TS-1 molecular sieve containing TiOOH. Although the main peaks originate from adsorbed C3H6, the small shoulders of the bands at 1443, 1646, 2980, and 3081 cm21 are attributed to gas-phase propene [Reprinted from Lin and Frei (133) with permission. Copyright (2002) American Chemical Society].

the active species in alkene epoxidation catalyzed by TS-1. When propene oxide was brought in contact with a sample of TS-1 containing the TiOOH species, propionaldehyde was formed by rearrangement. No such rearrangement of the epoxide occurred (133) in the absence of the TiOOH, indicating that it is the protonic acidity of TiOOH and not the Lewis acidity of the Ti ions in TS-1 that is responsible for this acid-catalyzed rearrangement. Although dehydrated TS-1 does not contain Brønsted acid sites, such sites are apparently created during its interaction with H2O2. The Lewis acid sites on TS-1 are probably deactivated by the water present in the reaction medium.

Fig. 49. FTIR difference spectrum recorded 80 min (trace a) and 320 min (trace b) after loading of TS-1/TiOOH molecular sieve with 6.5 mbar of propene at room temperature [Reprinted from Lin and Frei (133) with permission. Copyright (2002) American Chemical Society].

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

131

In an attempt to quantify the relationship between the TiOOH groups and the yield of propene oxide from the extinction coefficients of the latter’s 1409and 1493-cm21 bands, it was determined that 0.6 mol of the epoxide formed per mole of framework Ti center in the molecular sieve. That is, at least 60% of all framework Ti (80% of the surface-exposed Ti) is converted to TiOOH upon reaction with H2O2. The consumption of the TiOOH species during the oxygen insertion into propene was also independently confirmed by the loss in intensity of its LMCT band at 360 nm when the catalyst was brought in contact with propene at room temperature (Fig. 50). In contrast to propene, ethene, with its less electron-rich CyC bond, did not react at room temperature in the dark with TS-1 and instead required excitation of the UV –visible LMCT absorption at 360 nm to activate the TiOOH group for electrophilic oxygen transfer to form the epoxide. Again, the formation of the products, ethene oxide (at 871 cm21) and acetaldehyde (at 1353 and 1724 cm21) was accompanied by the loss of the TiOOH peaks at 837 and 3400 cm21 and the concurrent growth of the 3676- and 1629-cm21 bands assigned to Ti –OH and H2O, respectively (133). Direct evidence for O transfer from TiOOH to ethene was sought from the 18O isotope frequency shifts of ethene epoxide when a Ti18O18OH moiety (generated from TS-1 and H18 2 O2) was used. The epoxide product, C2H18 4 O, was isotopically pure, confirming that the oxygen atom in the epoxide indeed originated from the TiOOH species.

Fig. 50. Diffuse reflectance spectra recorded (a) before and (b) after 20 min of thermal reaction of propene in TS-1/TiOOH molecular sieve at room temperature [Reprinted from Lin and Frei (133) with permission. Copyright (2002) American Chemical Society].

132

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

To probe the origin of acetaldehyde in ethene oxidation, ethene oxide was admitted to the (TS-1/H2O2) system containing TiOOH groups. The formation of acetaldehyde was negligible even under the influence of UV –visible irradiation. Hence, the significant amount (10%) of acetaldehyde formed in the reaction of ethene with TS-1/H2O2 could not have been the product of the further reaction of ethene oxide. It is rather a primary product of oxidation at the vinylic carbon atom. Zhao et al. (50), on the basis of the appearance of the phenoxy radical (detected by EPR spectroscopy) simultaneously with the disappearance of the framework Ti-superoxide species resulting from contact of phenol with TS-1/ H2O2, correlated the concentration of the superoxide with catalytic activity for phenol oxidation (Section III.E). Srinivas et al. (52) recently attempted to correlate the relative EPR intensities of individual Ti-superoxides (A0 , A, B, and C) in the various titanosilicates with their chemoselectivities in styrene oxidation (Sections II.A.7 and III.E). The relative concentration of A0 þ A was related to styrene oxide (SO) selectivity (Fig. 51). Both the intensity of (A0 þ A) Ti-superoxo signals and the selectivity for styrene oxide (SO) were higher in the case of TS-1 than Ti-beta (Fig. 51). The yield of non-selective products (phenyl acetaldehyde and benzaldehyde) correlates with the concentration of the (B þ C) oxo species. Similarly, the concentration of the (B þ C) oxo species is higher in methanol solvent than in acetonitrile, in parallel with the greater formation of the non-selective products in the former than in the latter. It was also found that the styrene epoxide concentration was higher when the total EPR signal intensity was lower. On the basis of these results, Srinivas et al. (52) suggested that EPR-inactive hydroperoxo/peroxo titanium species are probably responsible for epoxidation, although superoxo-titanium is responsible for the side reactions. The predominant formation of the epoxide at low temperatures and the non-selective products observed when the temperature was raised were ascribed to the greater stability of the hydroperoxo/peroxo-titanium species at lower temperatures and the relatively high stability of the superoxo species at elevated temperatures. Additional support for the greater involvement of the hydroperoxide in epoxidation comes from investigations of the Pd-TS-1 system. The hydroperoxo/superoxo ratio (0.73) observed when Pd-TS-1 is brought in contact with H2O2 was noted in Section III.E (Fig. 25). Correspondingly, the selectivity for the epoxide in the oxidation of propene catalyzed by Pd-TS-1 with H2O2 generated in situ from H2 and O2 was also high (99%) (Section V.C.16). The EPR signal intensity of the titanium oxo species in Ti-MCM-41 was lower (52) when tert-butyl hydroperoxide in n-decane (rather than aqueous H2O2) was used as the oxidant, suggesting that a majority of the oxo-titanium is in the EPR-silent hydroproxo/peroxo form when reaction occurs in n-decane solvent.

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

133

Fig. 51. Correlation between the intensity of Ti-superoxo ([A0 þ A] and [B þ C]) signals and selectivity for styrene oxide and non-selective products in the styrene epoxidation reaction. The effects of titanosilicates, oxidants, and solvent on the correlation are depicted [from Srinivas et al. (52)].

A similar conclusion was also reached by Sankar et al. (46), who used EXAFS/ DFT techniques. From the selective decrease in the EPR intensity of the A type superoxo species during the epoxidation of styrene and allyl alcohol (Fig. 52), Srinivas et al. (52) concluded that these types of oxo species are preferentially consumed during the reaction. The correlation between the concentration of the superoxide species, A and B, and catalytic activity is further illustrated in Tables LIII and LIV. A TS-1 sample (without any trace of anatase) as well as another one containing some anatase were prepared by the method of Thangaraj et al. (138) (with some minor modifications). A sample of TS-1 (fluoride) was prepared in a fluoride medium.

134

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Fig. 52. EPR spectra recorded at 90 K. (a) TS-1 þ aqueous H2O2. (b)–(d) TS-1 þ H2O2 þ styrene reacted at 333 K for 5, 10, and 20 min, respectively, and (e) TS-1 þ H2O2 þ allyl alcohol reacted at 333 K for 25 min. Asterisk represents signal caused by a styrene-derived radical formed during the reaction [from Shetti et al. (93)].

The three TS-1 catalysts with similar Ti contents have cuboidal morphology with comparable particle sizes of 0.2– 0.3 mm (as shown in SEM pictures, Fig. 53). The EPR spectra of the samples in contact with aqueous H2O2 (46%) (Fig. 54) indicate that the ratio of the A to B superoxo species in various TS-1 samples increases in the order TS-1 (fluoride) , TS-1 (with anatase) , TS-1 (without anatase). Catalytic activity for phenol hydroxylation and allyl alcohol epoxidation (Table LIII) was found to parallel the A/B ratio of the oxo-Ti species (TS-1(fluoride) , TS-1 (with anatase) , TS-1 (without anatase)). Catalytic activity in benzene hydroxylation (Table LIV), on the other hand, followed the total concentration of the various superoxo species, which increased in the order TS-1 (with anatase) , TS-1 (without anatase) , TS-1 (fluoride). The total concentration of the superoxo species was obtained from the integrated intensity of all the EPR signals representing superoxo species. This intensity in various solvents increases in the order acetone , methanol p water. The picture that emerges from the results summarized above is the following: H2O2 reacts with the titanium centers on TS-1 and other titanosilicates to generate the titanium oxo species (hydroperoxo and superoxo). At room temperature and

Catalyst

Epoxidation of Allyl Alcohol (AA)a AA conversion (mol%)

Phenol hydroxylationb,c

Product selectivity (mol%) Glycidol

-diol

Phenol conversion (mol%)

Product selectivity (mol%) Catechol

Hydroquinone

43.7 (24.9)

56.4 (75.2)

TS-1 (without anatase)

96.1

96.3

3.7

12.9 (16.2)

TS-1 (with anatase)

89.6

97.5

2.5

11.1 (17.3)

49.0 (23.6)

51.0 (76.4)

TS-1 (fluoride)

32.9

97.1

2.9

3.3 (13.6)

40.7 (22.4)

59.4 (77.6)

a

Reaction conditions (epoxidation of AA): catalyst, 100 mg; AA, 8.6 mmol; H2O2 (aq. 46%), 17.2 mmol; acetone, 10 g; temperature, 333 K; time, 8 h. Reaction conditions (phenol hydroxylation): catalyst, 100 mg; phenol, 10 mmol; H2O2 (aq. 32.8%), 3.33 mmol; solvent (acetone or methanol), 4.2 mL; temperature, 333 K; H2O2 addition over 1.5 h; reaction time, 5.25 h (after H2O2 addition). c Values in parentheses correspond to the results in methanol solvent. b

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

TABLE LIII Catalytic activities of TS-1 samples (Si/Ti ¼ 33; particle size ¼ 0.2–0.3 mm) prepared by different methods

135

136

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169 TABLE LIV Catalytic activity in benzene hydroxylation of TS-1 samples (Si/Ti ¼ 33; particle size ¼ 0.2 –0.3 mm) prepared by different methods Catalyst

Benzene conversion (mol%)

Product selectivity (mol%) Phenol

Catechol

Hydroquinone

TS-1 (without anatase)

33.8

64.1

16.0

19.9

TS-1 (with anatase)

23.4

72.4

12.9

14.8

TS-1 (fluoride)

36.1

63.0

16.3

20.8

Reaction conditions: catalyst, 100 mg; benzene, 19.2 mmol; H2O2 (aq. 32.8%), 9.6 mmol; solvent (water), 7.5 g; temperature, 333 K; H2O2 addition in one lot; reaction time, 2 h. Benzene conversion in methanol, acetone and acetonitrile solvents is negligible.

higher temperatures, there is an interconversion of the two types of oxo-species (Section III.E). In alkene epoxidation the hydroperoxide reacts with the alkene to give the epoxide (133). In view of the direct correlation observed between the concentration of the (A þ A0 ) superoxo species and selectivity for the styrene epoxide (Fig. 51), these two types of superoxides (A and A0 , respectively) are perhaps transformed more easily into the hydroperoxides than the others (B and C, respectively). The side products probably arise from the reaction of either or both of the B and C groups of superoxides. The more recent calculations and in situ EPR results of Shetti et al. (54) suggest that the A and A0 superoxides are attached to tetrapodal Ti, although the B and C species are coordinated to tripodal titanium sites. The formation of TiOOH by both the tetra- and tripodal Ti is also supported by the FTIR spectroscopic results of Lin and Frei (133).

Fig. 53. SEM photographs of TS-1 samples, without anatase (left); with a trace amount of anatase (center); and from a fluoride medium (right) [from Shetti et al. (93)].

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

137

Fig. 54. EPR spectra showing the differences in the types of superoxo species generated on various TS-1 samples prepared by different methods after contacting with aqueous H2O2 [from Shetti et al. (93)].

VII. O– O Bond Cleavage and Product Selectivity VII.A. General It is known from the homogeneous catalytic oxidations by metal complexes and biological oxidations by metalloenzymes that the type of cleavage of the O – O bond in the active oxygenated metal species formed during the oxidation reactions plays a crucial role in determining the product pattern. The breaking of the O –O bond in Ti – O – O or the various other titanium oxo species discussed in Section III will also be determined by similar structural considerations and influence product selectivities. Electron-donating or withdrawing ligands either on the Ti atom (such as OSi, OH, or H2O) or the peroxo moiety (such as the alkyl group in TBHP) can influence the scission of the O –O bond. In other words, the type of Ti site (tetra-, tripodal, etc.) or oxidant (H2O2, TBHP) influences the homolytic vs. heterolytic cleavage. The open structured, tripodal titanium sites form penta- or hexa-coordinated species such as Ti(OSi)3 (H2O)2(OH) more easily than the closed tetrapodal Ti structures (vide supra). The coordinated water and OH groups enhance electron density at Ti center and the O –O bond, favoring homolytic O – O bond cleavage and zOH radical formation. Hence, systems having the tripodal Ti(OSi)3(OH) sites in preponderance

138

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

(such as Ti-MCM-41 and possibly other Ti-mesoporous material) are likely to cleave the O – O homolytically and generate greater amounts of radicals than catalysts with predominantly tetrapodal Ti(OSi)4 sites (such as TS-1). Similarly, alkyl hydroperoxides, when used as oxidants, are more likely to cleave the O – O bond in TiOOR complexes homolytically than H2O2 in TiOOH. This may be one of the reasons for the greater selectivities observed with TS-1 that uses aqueous H2O2 as the oxidant than with Ti-MCM-41 that uses alkyl hydroperoxides as the oxidants.

VII.B. Epoxidation of Alkenes In the epoxidation of alkenes, as was discussed above (Section V.C), TS-1 produces mainly the epoxide, although Ti-MCM-41 and similar mesoporous materials produce, in addition, significant amounts of side products including those derived from allylic CH activation. Adam et al. (278), exploring the factors that influence the allylic CH oxidation vs. epoxidation in the oxidation of 2-cyclohexenol by Cr- and Mn-salen complexes in the liquid phase, found that although manganese salens were selective for epoxidation, the chromium analogues selectively gave allylic CH oxidation. Iodosobenzene was the oxygen source. The authors interpreted the chemoselectivity in terms of the electron transfer (for manganese salens) vs. the hydrogen abstraction mechanisms (for the chromium salens) (Scheme 26). When the reactant is cyclohexene, in the first step of Scheme 26, the direct hydrogen abstraction for the allylic oxidation (path 1) competes with the electron transfer (from the alkene to the M-oxo complex) for the epoxidation (path 2). Because the manganese complex is more readily reduced than the chromium

Scheme 26.

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

139

complex, the authors speculated that the higher reduction potential of the manganese complex (relative to the chromium complex) favors electron transfer from the cyclohexene reactant to the metal catalyst and thus allows competetive epoxide formation to take place. Conversely, for the more difficult to reduce, electron-rich chromium complex, allylic oxidation by hydrogen abstraction (path 1) is favored. In the titanosilicate system, cyclic voltametric measurements had indicated (Section III.D) that the electron density at the tripodal sites is higher than at the tetrapodal sites. Hence, by analogy with the chromium and manganese complexes, we may expect the tripodal sites to favor hydrogen abstraction and allylic CH oxidation, although electron transfer and epoxidation occur preferentially on the tetrapodal sites. A tentative mechanism involving the heterolytic cleavage of the O – O bond along with electron transfer from the alkene to the electrophilic oxygen of the Ti(O2H) complex is shown in Scheme 27. In the envisaged titanium oxo complex, the Ti atom is side-bound to the peroxy moiety (O2H), consistent with all the spectroscopic results mentioned in Section III; in Scheme 27, between the two O atoms that are side-bound to Ti4þ, the O atom attached to both the Ti and H atoms is expected to be more electrophilic than the O atom attached to only the Ti atom and is likely to be the site of nucleophilic attack by the alkene double bond. The formation of the Ti – OH group (and not the titanyl, TiyO, as proposed by Khouw et al. (221)) after the epoxidation and its subsequent condensation with Si –OH to regenerate the Ti – O – Si links had been observed (Section III.B) by FTIR spectroscopy by Lin and Frei (133). Because this is a concerted heterolytic cleavage of the O – O bond, high epoxide selectivity and retention of stereochemistry may be expected, as indeed has been observed experimentally (204). The transition state in the above scheme differs from the cyclic titanium peroxo complex proposed earlier (217). In the earlier mechanism, any of the two peroxo oxygens in the Ti –O – O – H (bound end-on) could have been inserted into the CyC bond, and accordingly two isomers would be possible. They have never

Scheme 27.

140

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

been observed (33). In Scheme 27, on the other hand, the oxygen attached to both Ti4þ and the proton will be relatively more electrophilic to accept the electron from the CyC bond. Our mechanism bears similarities to those proposed in the homogeneous catalysis literature (170) for reactions catalyzed by peroxyacids, RC(O)OOH. The Ti4þ replaces, formally, the acylium cation, RCOþ. When instead of H2O2 an alkyl hydroperoxide (such as tert-butyl hydroperoxide) is used, the titanium oxo species that is generated may be Ti(O2R) (R ¼ alkyl). As a consequence of the electron-donating effect of R, it is unlikely that the oxygen atom attached to it acquires an electrophilic character. Hence, it is the other oxygen atom attached to the metal that is more electrophilic and is, therefore, attached to the CyC bond forming epoxide, as shown in Scheme 5. We emphasize that the above mechanism is strictly valid only for H2O2 and alkyl hydroperoxide epoxidations of alkenes catalyzed by TS-1 and Ti-MCM41. In view of the observation of similar titanium oxo species when H2 þ O2 are brought in contact with TS-1 or Ti-MCM-41 (54), similar conclusions may be drawn for that system as well. A radical mechanism involving the TiyO groups had been proposed earlier by Khouw et al. (221) for the hydroxylation of alkanes. No spectroscopic investigation of the TS-1/H2O2/alkane has yet been reported.

VIII. Conclusions and Outlook Significant progress has been achieved in the preceding few years in the study of titanosilicate molecular sieves, especially TS-1, TS-2, Ti-beta, and TiMCM-41. In the dehydrated, pristine state most of the Ti4þ ions on the surfaces of these materials are tetrahedrally coordinated, being present in either one of two structures: a tetrapodal (Ti(OSi)4) or a tripodal (Ti(OSi)3OH) structure. The former predominates in TS-1, TS-2, and Ti-beta, and the latter is prominent in Ti-MCM-41. The Ti ions are coordinatively unsaturated and act as Lewis acid sites that coordinatively bind molecules such as H2O, NH3, CH3CN, and H2O2. Upon interaction with H2O2 or H2 þ O2, the Ti ions form titanium oxo species. Spectroscopic techniques have been used to identify side-bound hydroperoxo species such as Ti(O2H) and superoxo structures such as Ti(Oz2 2 ) on these catalysts. These titanium oxo species oxidize various organic reactants. Direct confirmations of the participation of these titanium oxo species in the oxidation reactions have been obtained by infrared and EPR spectroscopies (54,133). The infrared absorption (133) or EPR (54) signal intensity of the titanium oxo species decreased simultaneously with an increase in the infrared or EPR signal intensities characterizing reaction products.

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

141

Although TS-1 in its dehydrated state is not a Brønsted acid, the hydroperoxo species Ti(O2H) generated as a result of its interaction with H2O2 has Brønsted acidity and catalyzes reactions such as the isomerization of epoxides to aldehydes (for example, propene oxide to propionaldehyde). Hence, although oxidation by H2O2 is the predominant reaction catalyzed by these materials, side reactions attributed to the Brønsted acidity of the Ti(O2H) group can also occur, decreasing the selectivity for the desired oxidation product. In the absence of H2O2, these titanium silicates are weak Lewis acids and catalyze reactions such as the rearrangement of cyclohexanone oxime to 1-caprolactam or the cycloaddition of CO2 to epoxides to yield cyclic carbonates. A large number of oxidation reactions of a variety of reactants have been reported to be catalyzed by titanosilicate molecular sieves (Section V). The transition from the laboratory to the factory will undoubtedly happen in some of the cases. Because of the high price of H2O2, most of the novel applications are likely to be in the area of fine chemicals rather than commodity or bulk materials. Attempts have already been made to find substitutes for H2O2 or to generate H2O2 in situ from H2 þ O2 or alcohol þ O2. Metals such as platinum, palladium, gold, etc. supported on TS-1 have been explored as catalysts. The strategy was to synthesize the H2O2 on the metal and use it in turn to catalyze the oxidation reaction on the titanosilicate. The main difficulty has been the efficient synthesis of H2O2; only low H2 and O2 efficiencies have been encountered in the synthesis of H2O2, rendering the process economically unviable. An alternate approach is to generate H2O2 in situ from the oxidation of alcohols (such as isopropanol or anthraquinol) with O2: Alcohol þ O2 ! ketone þ H2 O2 :

ð34Þ

The ketone can be hydrogenated in a separate reactor and recycled. This is the current route for the manufacture of H2O2 using anthraquinone –anthraquinol. The technological and economic advantages of combining the two processes (H2O2 synthesis and oxidation of organic reactants) in one reaction zone are not clear. To overcome the limitations of the MFI pore structure of TS-1 in oxidizing large molecules, Ti-beta, Ti-MCM-41, and other large and mesoporous materials have been investigated. The results have been mixed. Although the rates of the oxidation reaction have been enhanced (by the absence of diffusional constraints), attaining high selectivity for the desired oxidation product has been more elusive. Identifying, designing, and synthesizing the appropriate titanium oxo species on the surface of large-pore or mesoporous Ti-silicates while simultaneously increasing their hydrophobicity will be necessary to obtain the high selectivity characteristic of TS-1. There will be an increasing focus on the standardization of the synthesis procedures of these novel materials and characterizing modifying their physicochemical and catalytic properties in the coming

142

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

years. Appendix C includes a list of some of the recent advances and publications regarding the synthesis of titanium silicate molecular sieves.

Acknowledgements PR thanks the Alexander von Humboldt Foundation for a visiting Fellowship to Munich.

Appendix A. Fingerprint Features for Ti Isomorphous Substitution in TS-1 Titanosilicates See Table A1.

TABLE A1 Characterization technique

Fingerprint feature

XRD

MFI structure; orthorhombic (Pnma space group at room temperature) to monoclinic (P21 =n space group at low temperatures) structural phase transition

UV (diffuse reflectance)

Intense band at 210–220 nm (O(2p) ! Ti(3d) charge transfer transition)

XAS

Intense Ti pre-edge peak (1s ! 3d) at about 4969 eV

EPR

No signal (diamagnetic þ 4 oxidation state of Ti); contact with CO or H2 (at elevated temperatures (773 K)) generates paramagnetic Ti3þ species

UV resonant Raman

Strong bands at 490, 530 and 1125 cm21 (due to bending, symmetric stretching and asymmetric stretching vibrations of Ti –O –Si, respectively) when excited at 244 nm

UV photoluminescence

Emission bands at 495 and 430 nm with the corresponding excitation bands at 250 and 300 nm, respectively

XPS

Ti2p core level spectrum at 460.0 ^ 0.2 eV (due to þ 4 oxidation state of tetrahedral Ti; higher energy shift in binding energy by ,1.5 eV compared to TiO2 anatase) (caution: highly dispersed Ti in silica matrices (Ti . 2%) can produce a similar high energy shift; this shift is also claimed to depend on the large number of Si atoms in the second coordination shell of Ti)

Infrared and Raman

Band at 960 cm21 assigned to Ti–O–Si vibration (Caution: Si–OH and defect sites in silicalites also show this feature).

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

143

Appendix B. Characteristics of the Oxo-Titanium Species Generated on TS-1 on Contact with Aqueous H2O2 See Table B1.

Appendix C. Synthesis of Titanium Silicate Molecular Sieves The review of Notari (33) covers the synthesis methodologies of titanium silicate molecular sieves available up to 1996. The reviews of Corma (279) and subsequently of Biz and Occelli (280) describe the synthesis of mesoporous molecular sieves. An informative article on the preparation of TS-1 was reported recently by Perego et al. (68). In this section we list some of the recent developments in the synthesis of micro and mesoporous titanosilicate molecular sieves.

TABLE B1 Technique Visual appearance (color) Diffuse reflectance UV –visible Vibrational spectroscopy (infrared and Raman/resonance Raman)

Characteristic feature Yellow A labile charge transfer band at about 385 nm (25,800 cm21) in neutral H2O2 solutions and a relatively more stable band at 350 nm (28,500 cm21) in alkaline H2O2 solutions Reduction and blue shift of characteristic Si–O–Ti band (at 960 cm21) to 976 cm21 and quenching of 1125 cm21 band in resonance Raman spectrum when excited with 442 and 1064 nm laser radiation Strong, complex feature at 618 cm21 in resonance Raman spectrum when excited with 442 nm radiation Infrared-weak and Raman-intense absorption at about 880–890 cm21 in neutral H2O2 and at about 840 cm21 in alkaline H2O2 solutions Large bandwidth, red-shifted infrared band corresponding to hydrogen bonded OH groups at 3400 cm21.

XAS

Significant reduction in the pre-edge intensity indicating increase in the coordination number of Ti

EPR

Labile, rhombic type spectrum corresponding to Ti-superoxo species; spectral features sensitive to the type of silicate structure, temperature, solvent and pH

Magnetism

Partly paramagnetic.

144

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

C.1. TS-1, TS-2, Ti-ZSM-48, Ti-MWW, and Ti-MMM-1 Taramasso et al. (5) had originally reported two methods for the hydrothermal synthesis of TS-1. The first method (mixed alkoxide method) involves the preparation of a solution of mixed alkoxides of titanium and silica (preferably ethoxides) followed by hydrolysis with alkali-free solution of tetrapropylammonium hydroxide (TPAOH), distillation of the alcohol and crystallization of the resulting gel at 448 K. In the second method (dissolved or hydrolyzed titanium method) a soluble tetrapropylammonium peroxotitanate species was prepared initially and then colloidal SiO2 (Ludox AS-40) was added. This entire operation had to be carried out at 278 K. The TS-1 samples obtained by these two synthesis routes differed, particularly because of the presence of impurities such as Al3þ usually present in colloidal silica (33). Later, Thangaraj et al. (275,281) developed a novel, improved route ( prehydrolysis method) for the preparation of good quality TS-1 samples. In this method the silica source (tetraethyl orthosilicate; TEOS) in iso-propanol was first hydrolyzed with 20% aqueous TPAOH solution prior to the (dropwise) addition of titanium butoxide in dry iso-propanol under vigorous stirring. Crystallization was done statically at 443 K for 1– 5 days and the solid was calcined at 823 K for 10 h. The TS-1 samples thus obtained exhibited high catalytic activity in hydroxylation reactions. Another method (known as the wetness impregnation method) originally reported by Padovan et al. (282,283) used a SiO2 –TiO2 coprecipitated dry gel which was impregnated with an aqueous solution of TPAOH and crystallized under autogeneous pressure. At a high concentration of the base, dissolution of the oxides occurs, followed by crystallization in the presence of TPAOH. This method offers the advantage of requiring relatively small amount of TPAOH. But the catalyst obtained was poorly active as a consequence of the impurities present in the starting material. In an attempt to produce TS-1 at low cost, alternative, cheaper sources of Ti and Si and other bases such as binary mixtures of (tetrabutylammonium and tetraethylammonium hydroxides), (tetrabutylphosphonium and tetraethylphosphonium hydroxides), (tetrapropylammonium bromide and ammonia, water, hexanediamine, n-butylamine, diethylamine, ethylenediamine, or triethanolamine) in place of TPAOH have been used (284– 294). TS-1 was synthesized in the presence of fluoride ions but the material thus formed contained extraframework Ti species (295 – 297). Kumar et al. (298 – 300) reported a method wherein the crystallization time is significantly reduced. They found that addition of a small amount of oxyanion (e.g., H3PO4) to the TS-1 synthesis gel enhances the nucleation and crystallization

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

145

rates. By this promoter-induced synthesis method the overall crystallization time was reduced by about five times. Ahn et al. (301) and subsequently Prasad et al. (302,303) reported the rapid synthesis of highly crystalline TS-1 by microwave irradiation technique with yields exceeding 90%. The synthesis, which requires 1– 2 days by the conventional heating methods of Taramasso et al. (5) and Thangaraj et al. (275,281), was achieved within 30 min. In the synthesis reported by Ahn et al. (301), a SiO2 –TiO2 cogel ðSi=Ti ¼ 50Þ prepared by a two-step acid/base sol –gel process was dried overnight at 383 K and subsequently ground to give a fine powder which was dry impregnated by adding TPAOH solution. The impregnated gel was then heated with microwaves (500 W; 443 K) to obtain the crystalline powder. Prasad et al. (303) prepared the gel ðSi=Ti ¼ 10Þ following the prehydrolysis synthesis method and then heated by microwaves (800 W; 448 K). Approximately 12 –14 bar autogeneous pressure was developed during the synthesis. The catalysts prepared by the microwave technique showed activity similar to those prepared by the conventional heating methods. In an attempt to reduce the amount of expensive TPAOH template, Khomane et al. (304) used a non-ionic surfactant, Tween 20, in the TS-1 synthesis. Their method required only a small amount of TPAOH. Highly crystalline TS-1 samples (0.15 mm size) showing good activity for octane epoxidation were obtained. Similar procedures adopted for the synthesis of TS-1 (the mixed alkoxide method, dissolved titanium method, pre-hydrolysis method, wetness impregnation method, and promoter induced synthesis method) were also used for the synthesis of TS-2. Tetrabutylammonium hydroxide (TBAOH) instead of TPAOH was used as the template (6,7,305 – 308). Ti-ZSM-48 was prepared by the dissolved titanium method using fumed silica (Cabosil), TBOT, H2O2, and diaminooctane (309 –310). Ti-ZSM-48 was also prepared using hexamethonium hydroxide base and by the pre-hydrolysis method (311). A titanosilicate with MWW structure (Ti-MWW) reported by Wu and Tatsumi (228) was claimed to be more active than TS-1 in the epoxidation of linear alkanes. Ti-MWW was synthesized in two steps. The first step consists of hydrothermal synthesis of Ti-containing MWW lamellar precursors using piperidine as a structure-directing agent and boric acid as a crystallization support agent. The second step was to treat the precursors in HNO3 or H2SO4 solutions under reflux for removing the extraframework titanium species together with a part of the framework boron. The diffusional properties of TS-1 catalysts could be modified by the synthesis of nanosized TS-1 (by the recently developed confined space synthesis method), but the separation of the finely crystalline catalyst from the product mixture is difficult. The procedure of Jacobsen and co-workers (188) for the synthesis of a mesoporous TS-1 overcomes this problem. In a typical synthesis of mesopous

146

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

TS-1 (mesoporosity , 20 nm, 0.3 – 1.2 mm size), carbon black pearls 700w (Carbot Corp., average particle diameter ¼ 18 nm (ASTM D-3249)) were impregnated by the incipient wetness method with a clear solution of TPAOH, water, and ethanol. After evaporation of ethanol, the carbon particles were impregnated with 20% excess (relative to incipient wetness) of a mixture of TEOT and TEOS. Aging for a minimum of 3 h at room temperature and heating at 453 K for 72 h yielded the solid product, which was isolated, and the carbon black was removed by controlled combustion in air at 523 K for 8 h. A similar development in this direction is the synthesis of a mixed-phase material containing both micro- and mesopores (Ti-MMM-1) (223). This material was synthesized by the addition of organic templates for mesopores (cetyltrimethylammonium bromide, CTABr) and micropores (tetrapropylammonium bromide, TPABr) at staggered times and the variation of the temperature of a single reaction mixture. Ti-MMM-1 is more selective (for oxidation of cyclohexane and of n-octane) than either Ti-MCM-41 or TS-1. The powder X-ray diffraction pattern indicates that the material contains both MCM-41 and MFI structures. The mixed phase contains framework Ti species and more atomic order within its walls than Ti-doped MCM-41.

C.2. Ti-Beta Zeolite Large-pore Ti-beta (pore diameter , 0.4 – 1 nm) was synthesized by direct hydrothermal synthesis, wetness impregnation, and by secondary synthesis methods (9,10,12,14,196,312 –318). It was thought initially that cations such as Al3þ are essential for the crystallization of beta-zeolite. Most of the early methods gave low zeolite yields, together with inefficient use of the expensive structure-directing agent (tetraethylammonium cation). Futhermore, the intrinsic activity of these materials was lower than that of TS-1 for small reactant molecules. The lower activity was found to be caused by Al3þ ions, a high density of connectivity defects (resulting in extreme hydrophilic properties), and a higher acidity of framework Ti species. Although Al-free Ti-beta zeolite could be synthesized by the use of dealuminated zeolite-beta seeds at high pH, the product (Ti-beta(OH)) contained a high density of Si – OH groups with a hydrophilic surface (12,13). Blasco et al. (12,13) developed a novel method for the synthesis of Al-free Tibeta zeolite in a fluoride medium. The Ti-beta zeolite thus obtained (Ti-beta(F)) was free of connectivity defects and was hydrophobic. The typical unseeded synthesis of Al-free Ti-beta zeolite (Ti-beta(F)) involves hydrolysis of TEOS in aqueous solutions of TEAOH (35%) and H2O2, followed by hydrolysis of TEOT and evaporation of ethanol and water. The water lost in the evaporation and

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

147

an appropriate amount of HF (48%) are then added and the reaction mixture crystallized while tumbling the autoclaves (60 rpm) at 413 K.

C.3. Ti-Containing HMS, MCM-41, and MCM-48 Tanev et al. (19) prepared titanium-substituted hexagonal mesoporous silica (TiHMS) by adding Ti(iso-OC3H7)4 and Si(OC2H5)4 dissolved in a mixture of ethanol – isopropanol to an aqueous solution of dodecylamine (DDA) and HCl. Aging of the resulting gel for 18 h at ambient temperatures afforded the crystalline as-synthesized Ti-HMS sample, which was then calcined in air at 923 K for 4 h. Ti-MCM-41 was prepared in a similar manner except for using quaternary ammonium ion template [C16H33N(CH3)3]þ (CTMAþ) (with counterion Br2) as a replacement of DDA (19). Corma et al. (17) reported the preparation of Ti-MCM-41 by use of amorphous silica (Aerosil 200 Degussa), an aqueous solution of tetramethylammonium hydroxide (25% TMAOH, K þ Na , 5 ppm,), an aqueous solution of hexadecyltrimethylammonium bromide (CTABr), and titanium isopropoxide at 408 K under static conditions (14 h). Maschmeyer et al. (319) prepared Ti-containing MCM-41 by grafting titanocene to the surface of silica walls (Ti " MCM-41). In contrast to the situation in TiMCM-41, the Ti ions in Ti " MCM-41 are at the surface, mostly having the tripodal tetrahedral structure. In Ti-MCM-41, part of the Ti is substituted in the silica lattice and resides within the walls. In an improved procedure, Corma et al. (320) reported that the structural order of MCM-41 is superior when Si(OCH3)4 is used as the silica source in place of Si(OC2H5)4. Ti-MCM-41 prepared by the above methods exhibited a lower efficiency in the utilization of H2O2 (for formation of the epoxide) in alkene oxidation than either TS-1 or Ti-beta. The hydrophilic/hydrophobic properties of Ti zeolites influence their catalytic activity and selectivity. The activity of Ti-MCM-41 catalysts was enhanced by silylation of the surface (273,321,322). Ti-MCM-48 (surface area ¼ 1000 –1450 m2/g, pore volume ¼ 0.8 – 1.1 cm3/g, pore diameter ¼ 2.4 – 2.7 nm) was synthesized by hydrothermal and postsynthetic grafting techniques from cationic alkylammonium surfactants (22,25,323).

C.4. Ti-SBA-15 Morey et al. (25) synthesized Ti-SBA-15 with uniform tubular channels (surface area ¼ 600– 900 m2/g, pore volume ¼ 0.6 – 1.3 cm3/g, average pore diameter ¼ 6.9 nm) by direct and postsynthesis methods by using triblock copolymers, poly(ethylene oxide)-poly(propylene oxide)-poly(ethylene oxide) in

148

TABLE C1 Synthesis of titanosilicate molecular sieves Titanosilicate

Si/Ti

Crystallite size (nm)/ morphology

Mixed alkoxide method. Hydrothermal synthesis using tetraethylorthosilicate (TEOS) as the source of Si, tetraethyltitanate (TEOT) as the source of Ti, tetrapropylammonium hydroxide (TPAOH) as structure directing agent (template), base and distilled water Dissolved titanium method. Hydrothermal synthesis using tetrapropylammonium peroxytitanate (prepared from TEOT, distilled water, 30% aqueous H2O2, and 25% aqueous TPAOH) as the source of Ti and colloidal silica (Ludox AS-40) as the source of Si and TPAOH as template. All additions done at 278 K Preparation using TiCl2, 14% aqueous TPAOH, 30% colloidal silica, and demineralized water Preparation at low pH using fluoride ions as mineralizing agent

90–30

Parallelepipeds with rounded edges

(5)

90–30

Parallelepipeds with rounded edges

(5)

Microspheres of diameter 5–1000 mm

(284) (295,296) (282,283)

Wetness impregnation method Prehydrolysis method. The Si source (TEOS) in dry iso-propyl alcohol is hydrolyzed with 20% aqueous TPAOH prior to addition of Ti source, Ti(OBu)4. Gel composition: SiO2:xTiO2:0.36TPA:35H2O ðx ¼ 0 – 0:10Þ; the synthesis time is reduced considerably (1–5 days at 433 K compared to 6– 30 days at 448 K, as reported in the original patent (5))

References

$10

Cuboid (,1 mm)

(275,281)

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

TS-1 (MFI)

Synthesis methodology, composition and improvements

(285)

(286)

Ovate shaped crystals

x0 ¼ 0:25 – 0:5) (2– 3 mm)

composition, SiO2:xTiO2:0.4 (x0 TEPOH þ (1 2 x0 )TBPOH):30H2O ðx ¼ 0 – 0:02Þ; temperature ¼ 443 K and synthesis time ¼ 4 days Influence of nature of silicon and titanium alkoxides on the incorporation of Ti Wetness impregnation method Prehydrolysis method. Synthesis under stirring (250 rpm; 453 K, 5 days) using TPABr and hexanediamine instead of TPAOH and other alkali media, TEOS and TBOT are sources of Si and Ti. Gel composition: SiO2:0.01TiO2:0.3C6DN:0.1TPABr:50H2O Prehydrolysis method. Synthesis using SiO2 instead of silica alkoxides. Gel composition: SiO2:xTiO2:0.4TPAOH:35H2O; 0 , x , 0:03

(287)

(when x0 ¼ 0); hexagonal prisms (when

(288) (325,326) 24–76

Elongated prisms ,7 £ 2.5 £ 0.5 mm

(289)

50–86

Hexagonal prisms/twinned conffin shaped particles (10–19 mm)

(290)

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Prehydrolysis method. Synthesis using binary mixtures of tetrabutylammonium and tetraethylammonium hydroxides instead of TPAOH Influence of TPAOH/TEAOH and TPAOH/NH4OH ratio on the rate of crystallization and crystallite size investigated Prehydrolysis method. Synthesis using binary mixtures of tetrabutylphosphonium hydroxide and tetraethylphosphonium hydroxide instead of TPAOH as base and template; TEOS and TBOT are sources of Si and Ti, respectively. Molar gel

(Continued) 149

150

TABLE C1 Continued Titanosilicate

Si/Ti

Prehydrolysis method. Investigation of influence of added oxyanions such as phosphate, perchlorate, arsenate, chlorate, bromate, etc. on rate of crystallization. The overall crystallization time in the presence of additives reduced by about five times compared to the conventional prehydrolysis method (7,8) Synthesis using TPABr as structure-directing agent and ammonia, water, hexanediamine, nbutylamine, diethylamine, ethylenediamine, or triethanolamine as base (seeds of TS-1 were added to get smaller crystallites and 100% crystallinity) Synthesis of “fibrous” titanosilicate

30–80

Synthesis using TiF4 (as the source of Ti), TEOS, TPAOH, and distilled water. Gel composition: SiO2:xTiO2:0.4TPA:30H2O, 0 , x , 0:05 Prehydrolysis method. Crystallization without evaporating the alcohol in the conventional synthesis (7,8) Preparation by gas–solid isomorphous substitution of Ti4þ for Si4þ and hydrothermal crystallization using TPABr as template Preparation using TiCl3 as source of Ti: influence of pH (11.6 –9.7)

45–90

Crystallite size (nm)/ morphology 0.1 –0.2 mm

References

(298–300)

(327)

2.5 mm length and aspect ratio (length/diameter) ¼ 50– 70 Round shaped particles (0.3 mm diameter)

(328)

(291,292)

(293)

(329)

Crystallite size 0.1– 4 mm

(330)

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Synthesis methodology, composition and improvements

(297)

33

0.1 –0.2 mm

(294)

50

Round shaped particles ,0.5 mm

(301)

0.3 –1.2 mm

(302,303)

0.15 mm

(304)

10 –33

33

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

TS-2 (MEL)

Mixed alkoxide method. Synthesis using TPAOH and HF and wetness-impregnation method using TPABr and NH4F Mixed alkoxide method. Preparation using ethylsilicate-40 (ES-40) as the cheaper, cost-effective source of Si. Gel composition: SiO2:0.03TiO2:0.33TPA:35H2O Template-impregnated SiO2 –TiO2 xerogels. SiO2 –TiO2 cogel prepared via a two-step acid/base sol –gel process. Gel obtained dried overnight 383 K, ground to fine powder and dry impregnated by adding 1.6 g of TPAOH (20% aq. solution) per 1 g of xerogel and heated in microwave environment. Crystalline product dried at 383 K and calcined at 823 K for 5 h (crystal yield .90%) Prehydrolysis method. Synthesis under microwave irradiation; gel composition: SiO2:xTiO2:0.36TPAOH:35H2O, x ¼ 0:03 – 0:11; reaction temperature ¼ 448 K, power input ¼ 800 W, 12– 14 bar autogeneous pressure, crystallization time ¼ 20–90 min Prehydrolysis method. Synthesis using small amount of TPAOH template in the presence of Tween 20, a non-ionic surfactant. Gel composition: 0.03TiO2:SiO2:0.12TPAOH:0.0009Tween 20:0.88IPA:14.45H2O. Crystallized at 433 K for 18 h under autogeneous pressure Mixed alkoxide method using TBAOH as structure directing agent

(6) 151

(Continued)

152

TABLE C1 Continued Titanosilicate

Si/Ti

Ti-ZSM-48

Hydrolyzed titanium oxide method using fumed silica as Si source Synthesis using hexamethonium hydroxide

References

(7,305)

Prehydrolysis method using tetraethylorthosilicate, titanium tetrabutoxide, and tetrabutylammonium hydroxide. Gel composition: SiO2:xTiO2:0.2TBAOH:20H2O, x ¼ 0:14 – 0:0055; 443 K, 2–7 days Synthesis using TBPOH as templating agent. Only a maximum of 1.1 Ti/unit cell can be incorporated in the framework Wetness-impregnated SiO2 –TiO2 xerogels Synthesis based on hydrolyzed titanium alkoxides with H2O2. Gel composition: SiO2:xTiO2:0.88TBAOH:99H2O:25x H2O2. Crystallization at 449 K Prehydrolysis method. TEOS, TBOT, hexamethonium hydroxide template; 473 K, 7 days, crystallization by rotation (40 rpm)

Crystallite size (nm)/ morphology

(286)

25

36–60

Elliptical particles (,1 mm) Ovate type crystals (2 mm)

(307)

0.2 –0.3 mm Spherical random agglomerates of small needle shaped crystals 5–15 mm diameter containing needles of 0.2 –1 mm long and diameter 0.1 mm

(309,310)

(8,308)

24–111 49

(311)

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Synthesis methodology, composition and improvements

Ti-Beta (BEA;)

(9,10,110)

(312,313)

(12) 50

(13,314)

(196,315,316) (Continued)

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

[Ti –Al]-beta (Si/Al # 150): Prehydrolysis method–conventional method using amorphous silica (Arosol 200), tetraethyl titanate, sodium aluminate/aluminium nitrate as sources of Si, Ti, and Al, respectively. Crystallization at 408 K by rotation (60 rpm); zeolite yield #7%. Cogel method by impregnating TiO2 –SiO2 cogel with TEAOH solution in the presence of some amount of aluminium ions. Crystallization at 408 K while tumbling the autoclave (60 rpm). Zeolite yields ,29%; Si/Al ¼ 300. Requires lesser amount of TEAþ ions than classical prehydrolysis method Seeding technique. Al-free Ti-beta obtained by use of dealuminated zeolite-beta seeds Fluoride method. Al-free Ti-beta: synthesis from a reaction mixture containing TEAOH and fluoride ions (HF) at near-neutral pH. Gel composition: TiO2: 60SiO2:32.9NEt4OH:32.9HF:20H2O:457.5 H2O. Crystallization at 413 K with rotation of the autoclave (60 rpm) Al-free Ti-beta: Direct synthesis

153

154

TABLE C1 Continued Titanosilicate

Si/Ti

Dry gel conversion method. 0.58 g of TBOT suspended in distilled water (4.0 g) to which was added 2 g of H2O2 (31 wt%). Mixture was stirred for 1 h, leading to solution A. Solution B prepared by dissolving anhydrous NaAlO2 (0.0124 g) and 0.015 g of NaOH in 8 g of TEAOH (40 wt% in water) and stirred for 1 h. Solution B added to solution A, stirred during heating at 353 K to dryness. Dried powder with composition SiO2: TiO2:Al2O3:Na2O:TEAOH ¼ 304:10:0.46:1.55:132.5) transferred to an autoclave where water as a source of steam was pored into the bottom. Crystallization carried out in steam first at 403 K (96 h) and then at 448 K (18 h) under autogeneous pressure. The recovered product was washed, dried (308 K, 10 h), and calcined (793 K, 10 h). The resulting Ti-beta was treated with 1-M H2SO4 at room temperature (12 h), washed, dried, and again calcined at 793 K for 5 h in the flowing air. By using colloidal silica (ST-40, 40 wt% SiO2, Nissan) instead of fumed silica, Ti-beta with higher crystallinity was synthesized. The molar composition of the gel was SiO2:TiO2:Al2O3:Na2O: TEAOH ¼ 310:10:0.52:12:135 Synthesis with TiCl3 and without any organic template

,30

Crystallite size (nm)/ morphology

References

(318)

(331–336)

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

ETS-10/-4 (Zorite structure)

Synthesis methodology, composition and improvements

(337,338)

0.6 –1 mm

(339)

(340)

Cuboid or wheat-shaped agglomerated crystals of 2 –4 mm

(341)

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Influence of various organic bases (R ¼ pyrrolidine, tetramethylammonium chloride, tetraethylammonium chloride, tetrapropylammonium chloride, 1,2-diaminoethane, and 1,2-diaminohexane) on crystallization of ETS-10. Synthesis using Na2SiO3·n H2O, TiCl4, NaOH, KOH, and distilled water. Gel composition: 40R:52Na2O:42K2O:20TiO2:100SiO2:7030H2O. pH ¼ 10.5– 12.9. Crystallization at 473 K for 2 –30 days ETS-10: Hydrothermal synthesis using TiO2 (P25, ˚ particle size, Degussa), 40% colloidal silica, 200 A KF, and NaOH and crystallization at 473 K for 2 days. Gel composition: 1.0 M2O:TiO2: 2– 8SiO2:5–50H2O. ETS-4: hydrothermal synthesis using NaF instead of KF (used in ETS-10) and crystallization at 473 K for 44 h. Gel composition: 1.0 M2O:TiO2:1.2– 6SiO2:5– 50H2O Synthesis of ETS-10, both in the presence and in the absence of seeds of ETS-4 and using TiCl4 as the source of Ti ETS-10 synthesis using organic templating agents (R) such as choline chloride and bromide salt of hexaethyl diquat-5, sodium silicate, TiCl3 (15% solution in HCl), NaOH, and KF·2H2O. Gel composition: 1.14R2O:3.7Na2O:0.95K2O: TiO2:5.71SiO2:171.9 or 256.9 H2O. Crystallization at 473 K for 5–7 days

(Continued) 155

156

TABLE C1 Continued Titanosilicate

Synthesis of ETS-10 using TiCl3 and crystalline TiO2 (anatase) as Ti sources. Gel composition: 4Na2O:1.5K2O:TiO2:5.5SiO2:125H2O. Crystallization at 503 K for 24 h ETS-10 synthesis from gels containing TiF4 and TiO2 Synthesis mixtures prepared using amorphous silica (Aerosil 200, Degussa), 25% aq. TMAOH, aqueous solution of hydroxide and bromide of hexadeciltrimethylammonium. Source of Ti was TEOT. Gels with following molar compositions were prepared: Si/Ti ¼ 60, (CTMA)2O:TMA2O ¼ 0.67, (TMA)2O:SiO2 ¼ 0.13, H2O:(TMA)2O ¼ 188 Silylation of surface of Ti-MCM-41. Synthesis gel composition: SiO2:0.015 TEOT:0.26 CTABr:0.26TMAOH:24.3 H2O Trimethylsilylation: Ti-MCM-41 prepared from TEOS, TBOT, and CTMACl with molar gel composition SiO2:0.01TiO2:0.6CTMA:0.3NMe4OH:60H2O was silylated with Me3SiCl and (Me3Si)2O

Si/Ti

Crystallite size (nm)/ morphology ,25 mm crystals

References

(342)

(343) 60

Pore size ¼ 2 nm; surface area ¼ 936 m2/g

66

139 (123 before silylation)

(17)

(321)

Pore diameter ¼ 1.9 nm (2.32 nm before silyaltion); pore volume ¼ 0.82 mL/g (0.88 mL/g before silylation, surface area ¼ 139 m2/g (123 m2/g before silylation)

(273)

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Ti-MCM-41

Synthesis methodology, composition and improvements

Ti-MCM-48

60–133

20–160

(20)

Synthesis using Gemini surfactant (bromide salt of [C18H37(CH3)2N–C12H24 –N(CH3)2C18H37]2þ

14.3 and 33.3

(21)

50 and 100

(22)

Direct hydrothermal synthesis. Prepared using titanium isopropoxide (triethanolaminato) and TEOS as the sources of Ti and Si, respectively, and the Gemini-type surfactant 18– 12–18 or cetylbenzyl dimethylammonium chloride (CBDAC) as a template. In the grafting method, silicious MCM48 first prepared and then the dry surface grafted with titanium isopropoxide

100

(320)

Pore diameter ,2.8 nm

Pore diameter ¼ 2.6 nm, Surface area ¼ 1296 m2/g (1093 m2/g for grafted material)

(19)

(25)

(Continued)

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Ti-HMS

One-step synthesis with methylated silicons: synthesis of organo-silica containing Ti-MCM-41 carried out with gels having following molar compositions: (1 2 x)Si(OCH3)4:xCH3Si(OC2H5)3:0.26TMAOH:0.15CTABr:24.3 H2O:yTEOT, where x ¼ 0:15 – 0:35 and y ¼ 0:0166 – 0:0075: After crystallization, the solid was first treated with 0.05-M H2SO4 in ethanol and then with 0.15-M HNO3 in heptane-ethanol Synthesis by acid hydrolysis in alcohol solution of mixture of TEOS and Ti(iso-OC3H7)4 in dodecylamine

157

158

TABLE C1 Continued Titanosilicate

Grafting method. SBA-15 prepared first using the amphiphilic triblock copolymer poly(ethyleneoxide)– poly(propyleneoxide)– poly(ethyleneoxide) (EO–PO– EO) as template and TEOS as Si source. The composition was 2 g copolymer:0.021 mol TEOS: 0.12 mol HCl:3.33 mol H2O. The solid was calcined at 600 K for 4 h to remove the copolymer. Ti in the form of titanium isopropoxide was grafted onto the dehydrated surface of SBA-15 Direct synthesis under microwave heating. Ti-substituted SBA-15 prepared using TEOS and TiCl4 as sources of Si and Ti and the triblock copolymer EO–PO– EO as structure-directing agent. The gel was crystallized during heating in a microwave environment

Incipient wetness method. For every 1 g of SBA15, varying amounts of titanium isopropoxide in 10 g of ethanol were used for impregnation. The titanium concentration in the solution varies from 0.05 to 5 M, depending on the desired titanium loading. The impregnated material was dried and calcined at 723 K for 5 h.

Si/Ti

5–40

0.6 –36 (XPS)

Crystallite size (nm)/ morphology

References

Pore diameter ¼ 6.3 nm, surface area ¼ 518 m2/g, pore volume ¼ 0.68

(25)

Mesopore size ¼ 7.3– 7.6 nm, specific surface area ¼ 767–844 m2/g, external surface area ¼ 15–26 m2/g), mesopore volume ¼ 0.78– 0.95 cm3/g Pore size ¼ 4.2–5.1 nm, specific surface area ¼ 690–997 m2/g, volume ¼ 0.81– 1.17 cm3/g

(25)

(27)

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Ti-SBA-15

Synthesis methodology, composition and improvements

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

159

an acidic medium. The direct synthesis of Ti-SBA-15 molecular sieves under microwave-hydrothermal conditions has considerably reduced the crystallization times (27). Kevan and co-workers (26,324) prepared SBA-15 incorporating Ti by incipient-wetness impregnation with titanium isopropoxide in ethanol followed by calcination.

C.5. Ti-TUD-1 The mesoporous materials reported above are usually prepared from relatively expensive surfactants. Some of them have poor hydrothermal stability. Furthermore, the MCM-41 host structure has a one-dimensional pore system with consequent pore blockage and diffusion limitations. Shan et al. (32) reported the synthesis of a three-dimensional and randomly connected mesoporous titanosilicate (Ti-TUD-1, mesopore wall thickness ¼ 2.5 –4 nm, surface area , 700 – 1000 m2/g, tunable pore size , 4.5 –5.7 nm) from triethanolamine (TEA). TiTUD-1 showed higher activity (about 5.6 times) for cyclohexene epoxidation than the framework-substituted Ti-MCM-41. Its activity was similar to that of the Ti-grafted MCM-41(32). Compositions of the synthesis gel and other physical characteristics of titanium silicate materials obtained in various synthesis methodologies are listed in Table C1.

References 1. Taylor, H.S., Proc. R. Soc. A 108, 105 (1925). 2. Taylor, H.S., J. Phys. Chem. 30, 145 (1926). 3. Thomas, J.M., Johnson, B.F.G., Raja, R., Sankar, G., and Midgley, P.A., Acc. Chem. Res. 36, 20 (2003). 4. Thomas, J.M., Raja, R., Johnson, B.F.G., O’Connell, T.J., Sankar, G., and Khimyak, T., Chem. Commun. 1126 (2003). 5. Taramasso, M., Perego, G., Notari, B., US Patent No. 4,410,501 (1983) to Snamprogetti S. p. A. 6. Belussi, G., Carati, A., Clerici, M.G., Esposito, A., Millini, R., Buonomo, F., Belg, Patent No.1,001,038 (1989) to Eniricerche S. p. A., Snamprogetti S. p. A., EniChem. S. p. A. 7. Reddy, J.S., Kumar, R., and Ratnasamy, P., Appl. Catal. 58, L1 (1990). 8. Serrano, D.P., Hong-Xin, L., and Davis, M.E., J. Chem. Soc., Chem. Commun. 745 (1992). 9. Camblor, M.A., Corma, A., Martinez, A., and Perezpariente, J., J. Chem. Soc., Chem. Commun. 589 (1992). 10. Camblor, M.A., Corma, A., and Pe´rezpariente, J., Zeolites 13, 82 (1993). 11. Corma, A., Camblor, M.A., Esteve, P., Martinez, A., and Pe´rezpariente, J., J. Catal. 145, 151 (1994). 12. Camblor, M.A., Costantini, M., Corma, A., Gilbert, L., Esteve, P., Martinez, A., and Valencia, S., Chem. Commun. 1339 (1996).

160

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

13. Blasco, T., Camblor, M.A., Corma, A., Esteve, P., Guil, J.M., Martinez, A., Perdigo´nMelon, J.A., and Valencia, S., J. Phys. Chem. B 102, 75 (1998). 14. Davis, R.J., Liu, Z., Tabora, J.E., and Wieland, W.S., Catal. Lett. 34, 101 (1995). 15. van der Waal, J.C., Kooyman, P.J., Jansen, J.C., and van Bekkum, H., Micropor. Mesopor. Mater. 25, 43 (1998). 16. Tuel, A., Zeolites 15, 236 (1995). 17. Corma, A., Navarro, M.T., and Pe´rezpariente, J., J. Chem. Soc., Chem. Commun. 147 (1994). 18. Blasco, A.T., Corma, A., Navarro, M.T., and Pe´rezpariente, J., J. Catal. 156, 65 (1995). 19. Tanev, P.T., Chibwe, M., and Pinnavaia, T.J., Nature 368, 321 (1994). 20. Gontier, S., and Tuel, A., Zeolites 15, 601 (1995). 21. Sudhakar Reddy, J., Dicko, A., and Sayari, A., in “Synthesis of Porous Materials: Zeolites, Clays and Nanostructures” (M. Occelli and H. Kessler, Eds.), p. 405– 415. Marcel Dekker, New York, 1997. 22. Koyano, K.A., and Tatsumi, T., Chem. Commun. 145 (1996). 23. Bagshaw, S.A., Prouzet, E., and Pinnavaia, T.J., Science 269, 1242 (1995). 24. Bagshaw, S.A., Di Renzo, F., and Fajula, F., Chem. Commun. 2209 (1996). 25. Morey, M.S., O’Brien, S., Schwarz, S., and Stucky, G.D., Chem. Mater. 12, 898 (2000). 26. Luan, Z., and Kevan, L., Micropor. Mesopor. Mater. 44/45, 337 (2001). 27. Newalkar, B.L., Olanrewaju, J., and Komarneni, S., Chem. Mater. 13, 552 (2001). 28. Kholdeeva, O.A., Derevyankin, A.Yu., Shmakov, A.N., Trukhan, N.N., Paukshtis, E.A., Tuel, A., and Romannikov, V.N., J. Mol. Catal. A: Chem. 158, 417 (2000). 29. Trukhan, N.N., Derevyankin, A.Yu., Shmakov, A.N., Paukshtis, E.A., Kholdeeva, O.A., and Romannikov, V.N., Micropor. Mesopor. Mater. 44/45, 603 (2001). 30. Trukhan, N.N., Romannikov, V.N., Paukshtis, E.A., Shmakov, A.N., and Kholdeeva, O.A., J. Catal. 202, 110 (2001). 31. Wu, P., Tatsumi, T., Komatsu, T., and Yashima, T., J. Phys. Chem. B 105, 2897 (2001). 32. Shan, Z., Jansen, J.C., Marchese, L., and Maschmeyer, Th., Micropor. Mesopor. Mater. 48, 181 (2001). 33. Notari, B., Adv. Catal. 41, 253 (1996). 34. Vayssilov, G.N., Catal. Rev.—Sci. Engng 39, 209 (1997). 35. Bolis, V., Bordiga, S., Lamberti, C., Zecchina, A., Carati, A., Rivetti, F., Spano`, G., and Petrini, G., Micropor. Mesopor. Mater. 30, 67 (1999). 36. Ricchiardi, G., de Man, A.J.M., and Sauer, J., Phys. Chem. Chem. Phys. 2, 2195 (2000). 37. Damin, A., Bordiga, S., Zecchina, A., and Lamberti, C., J. Chem. Phys. 117, 226 (2002). 38. Damin, A., Bordiga, A., Zecchina, A., Doll, K., and Lamberti, C., J. Chem. Phys. 118, 10183 (2003). 39. Bordiga, S., Damin, A., Bonino, F., Ricchiardi, G., Zecchina, A., Tagliapietra, R., and Lamberti, C., Phys. Chem. Chem. Phys. 5, 4390 (2003). 40. Li, C., Xiong, G., Liu, J., Ying, P., Xin, Q., and Feng, Z., J. Phys. Chem. B 105, 2993 (2001). 41. Ricchiardi, G., Damin, A., Bordiga, S., Lamberti, C., Spano`, G., Rivetti, F., and Zecchina, A., J. Am. Chem. Soc. 123, 11409 (2001). 42. Bordiga, S., Damin, A., Bonino, F., Ricchiardi, G., Lamberti, C., and Zecchina, A., Angew. Chem., Int. Ed. 41, 4734 (2002). 43. Bolis, V., Bordiga, S., Lamberti, C., Zecchina, A., Carati, A., Rivetti, F., Spano`, G., and Petrini, G., Langmuir 15, 5753 (1999). 44. Bonino, F., Damin, A., Bordiga, S., Lamberti, C., and Zecchina, A., Langmuir 19, 2155 (2003).

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

161

45. Damin, A., Bonino, F., Ricchiardi, G., Bordiga, S., Zecchina, A., and Lamberti, C., J. Phys. Chem. B 106, 7524 (2002). 46. Sankar, G., Thomas, J.M., Catlow, C.R.A., Barker, C.M., Gleeson, D., and Kaltsoyannis, N., J. Phys. Chem. B 105, 9028 (2001). 47. Gleeson, D., Sankar, G., Catlow, C.R.A., Thomas, J.M., Spano`, G., Bordiga, S., Zecchina, A., and Lamberti, C., Phys. Chem. Chem. Phys. 2, 4812 (2000). 48. Barker, C.M., Gleeson, D., Kaltsoyannis, N., Catlow, C.R.A., Sankar, G., and Thomas, J.M., Phys. Chem. Chem. Phys. 4, 1228 (2002). 49. Lamberti, C., Bordiga, S., Arduino, D., Zecchina, A., Geobaldo, F., Spano`, G., Genoni, F., Petrini, G., Carati, A., Villain, F., and Vlaic, G., J. Phys. Chem. B 102, 6382 (1998). 50. Zhao, Q., Bao, X., Wang, Y., Lin, L., Li, G., Guo, X., and Wang, X., J. Mol. Catal. A: Chem. 157, 265 (2000). 51. Chaudhari, K., Srinivas, D., and Ratnasamy, P., J. Catal. 203, 25 (2001). 52. Srinivas, D., Manikandan, P., Laha, S.C., Kumar, R., and Ratnasamy, P., J. Catal. 217, 160 (2003). 53. Bonoldi, L., Busetto, C., Congiu, A., Marra, G., Ranghino, G., Salvalaggio, M., Spano`, G., and Giamello, E., Spectrochim. Acta 58A, 1143 (2002). 54. Shetti, V.N., Manikandan, P., Srinivas, D., and Ratnasamy, P., J. Catal. 216, 461 (2003). 55. Carati, A., Flego, C., Massara, E.P., Millini, R., Carluccio, L., Parker, W. Jr., and Bellussi, G., Micropor. Mesopor. Mater. 30, 137 (1999). 56. Krijnen, S., Sa´nchez, P., Jakobs, B.T.F., and van Hooff, J.H.C., Micropor. Mesopor. Mater. 31, 163 (1999). 57. Grieneisen, J.L., Kessler, H., Fache, E., and Le Govic, A.M., Micropor. Mesopor. Mater. 37, 379 (2000). 58. Zicovich-Wilson, C., and Dovesi, R., J. Mol. Catal. A: Chem. 119, 449 (1997). 59. Cora`, F., Alfredsson, M., Barker, C.M., Bell, R.G., Foster, M.D., Saadoune, I., Simperlar, A., and Catlow, C.R.A., J. Solid State Chem. 176, 496 (2003). 60. Munakata, H., Oumi, Y., and Miyamoto, A., J. Phys. Chem. B 105, 3493 (2001). 61. Vayssilov, G.N., and van Santen, R.A., J. Catal. 175, 170 (1998). 62. de Man, A.J.M., and Sauer, J., J. Phys. Chem. 100, 5025 (1996). 63. Tozzola, G., Mantegazza, M.A., Ranghino, G., Petrini, G., Bordiga, S., Ricchiardi, G., Lamberti, C., Zulian, R., and Zecchina, A., J. Catal. 179, 64 (1998). 64. Gao, X., and Wachs, I.E., Catal. Today 51, 233 (1999). 65. Roberts, M.W., Catal. Lett. 67, 65 (2000). 66. Belussi, G., and Perego, C., in “Handbook of Heterogeneous Catalysis” (G. Ertl, H. Kno¨zinger and J. Weitkamp, Eds.), Vol. 5, p. 2329. Wiley, New York, 1997. 67. Petrini, G., and Leofanti, G., Mantegazza, M.A., and Pignataro, F., in “Green Chemistry: Designing Chemistry for the Environment” (P.T. Anastas and T.C. Williamson, Eds.), ACS Symposium Series No. 626, p. 33. American Chemical Society, Washington, DC, 1996. 68. Perego, C., Carati, A., Ingallina, P., Mantegazza, M.A., and Bellussi, G., Appl. Catal. A: Chem. 221, 63 (2001). 69. Meiers, R., Dingerdissen, U., and Ho¨lderich, W.F., J. Catal. 176, 376 (1998). 70. Meiers, R., and Ho¨lderich, W.F., Catal. Lett. 59, 161 (1999). 71. Jenzer, G., Mallat, T., Maciejewski, M., Eigenmann, F., and Baiker, A., Appl. Catal. A: Gen. 208, 125 (2001). 72. Stangland, E.E., Stevens, K.B., Andres, R.P., and Delgass, W.N., J. Catal. 191, 332 (2000). 73. Qi, C., Akita, T., Okumura, M., and Haruta, M., Appl. Catal. A: Gen. 218, 81 (2001).

162

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

74. Nijhuis, T.A., Huizinga, B.J., Makkee, M., and Moulijn, J.A., Ind. Engng Chem. Res. 38, 884 (1999). 75. Millini, R., Massara, E.P., Perego, G., and Bellussi, G., J. Catal. 133, 220 (1992), and 137, 497 (1992). 76. Holm, R.H., Kennepohl, P., and Solomon, E.I., Chem. Rev. 96, 2239 (1996). 77. Millini, R., and Perego, G., Gazz. Chim. Ital. 126, 133 (1996). 78. Reddy, J.S., and Sayari, A., Stud. Surf. Sci. Catal. 94, 309 (1995). 79. Hasegawa, Y., and Ayame, A., Catal. Today 71, 177 (2001). 80. Contarini, S., van der Heide, P.A.W., Prakash, A.M., and Kevan, L., J. Electron Spectrosc. Relat. Phenom. 125, 25 (2002). 81. Trong On, D., Bonneviot, L., Bittar, A., Sayari, A., and Kaliaguine, S., J. Mol. Catal. 74, 233 (1992). 82. Zecchina, A., Bordiga, S., Lamberti, C., Ricchiardi, G., Scarano, D., Petrini, G., Leofanti, G., and Mantegazza, M., Catal. Today 32, 97 (1996). 83. Bordiga, S., Coluccia, S., Lamberti, C., Marchese, L., Zecchina, A., Boscherini, F., Buffa, F., Genoni, F., Leofanti, G., Petrini, G., and Vlaic, G., J. Phys. Chem. 98, 4125 (1994). 84. Bordiga, S., Boscherini, F., Coluccia, S., Genoni, F., Lamberti, C., Leofanti, G., Marchese, L., Petrini, G., Vlaic, G., and Zecchina, A., Catal. Lett. 26, 195 (1994). 85. Henry, P.F., Weller, M.T., and Wilson, C.C., J. Phys. Chem. B 105, 7452 (2001). 86. Lamberti, C., Bordiga, S., Zecchina, A., Carati, A., Fitch, A.N., Artioli, G., Petrini, G., Salvalaggio, M., and Marra, G.L., J. Catal. 183, 222 (1999). 87. Hijas, C.A., Jacubinas, R.M., Eckert, J., Henson, N.J., Hay, P.J., and Ott, K.C., J. Phys. Chem. B 104, 12157 (2000). 88. Lamberti, C., Bordiga, S., Zecchina, A., Artioli, G., Marra, G., and Spano`, G., J. Am. Chem. Soc. 123, 2204 (2001). 89. van der Pol, A.J.H.P., Verduyn, A.J., and van Hooff, J.H.C., Appl. Catal. A: Gen. 92, 113 (1992). 90. Wilkenho¨ner, U., Langhendries, G., van Laar, F., Baron, G.V., Gammon, D.W., Jacobs, P.A., and van Steen, E., J. Catal. 203, 201 (2001). 91. Boccuti, M.R., Rao, K.M., Zecchina, A., Leofanti, G., and Petrini, G., Stud. Surf. Sci. Catal. 48, 133 (1989). 92. Le Noc, L., Trong On, D., Solomykina, S., Echchahed, B., Be´land, F., Cartier di Moulin, C., and Bonneviot, L., Stud. Surf. Sci. Catal. 101A, 611 (1996). 93. Shetti, V.N., Srinivas, D., Ratnasamy, P., Unpublished results. 94. Soult, A.S., Poore´, D.D., Mayo, E.I., and Stiegman, A.E., J. Phys. Chem. B 105, 2687 (2001). 95. Yamashita, H., Ikeue, K., Takewaki, T., and Anpo, M., Top. Catal. 18, 95 (2002). 96. Gianotti, E., Yoshida, H., Dellarocca, V., Marchese, L., Martra, G., and Coluccia, S., Res. Chem. Intermed. 29, 681 (2003). 97. Lamberti, C., Bordiga, S., Zecchina, A., Vlaic, G., Tozzola, G., Petrini, G., and Carati, A., J. Phys. 7, C2– 851 (1997). 98. Behrens, P., Felsche, J., Vetter, S., Schulz-Ekloff, G., Jaeger, N.I., and Niemann, W., J. Chem. Soc., Chem. Commun. 678 (1991). 99. Bonneviot, L., Trong On, D., and Lopez, A., J. Chem. Soc., Chem. Commun. 685 (1993). 100. Trong On, D., Bittar, A., Sayari, A., Kaliaguine, S., and Bonneviot, L., Catal. Lett. 16, 85 (1992). 101. Marchese, L., Maschmeyer, T., Gianotti, E., Coluccia, S., and Thomas, J.M., J. Phys. Chem. B 101, 8836 (1997). 102. Marchese, L., Gianotti, E., Dellarocca, V., Maschmeyer, T., Rey, F., Coluccia, S., and Thomas, J.M., Phys. Chem. Chem. Phys. 1, 585 (1999).

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

163

103. Sinclair, P.E., Sankar, G., Catlow, C.R.A., Thomas, J.M., and Maschmeyer, T., J. Phys. Chem. 4232 (1997). 104. Thomas, J.M., and Sankar, G., Acc. Chem. Res. 34, 571 (2001). 105. Huybrechts, D.R.C., Buskens, P.L., and Jacobs, P.A., J. Mol. Catal. 71, 129 (1992). 106. Jacobs, P.A., “Proceedings of DGMK Conference on Selective Oxidations in Petrochemistry, Germany” 1992, p. 171. 107. Decottiguier, M., Phalippai, J., and Zarzycki, J., J. Mater. Sci. 13, 7605 (1978). 108. Soda, R., Bull. Chem. Soc. Jpn 34, 1491 (1961). 109. Duran, A., Sesna, C., Fornes, V., and Fernandez-Navarro, J.M., J. Non-Cryst. Solids 82, 69 (1986). 110. Camblor, M.A., Corma, A., and Pe´rez-Pariente, J., J. Chem. Soc., Chem. Commun. 557 (1993). 111. Bellussi, G., Carati, A., Clerici, M.G., Maddinelli, G., and Millini, R., J. Catal. 133, 220 (1992). 112. Bellusi, G., and Rigutto, M.S., Stud. Surf. Sci. Catal. 85, 177 (1994). 113. Scarano, D., Zecchina, A., Bordiga, S., Geobaldo, F., Spoto, G., Petrini, G., Leofanti, G., Padovan, M., and Tozzola, G., J. Chem. Soc., Faraday Trans. 89, 4123 (1993). 114. Su, Y., Balnics, M.L., and Bunker, B.C., J. Phys. Chem. B 104, 160 (2000). 115. Smirnov, K.S., and van de Graaf, B., Micropor. Mater. 7, 133 (1996). 116. Li, C., Xiong, G., Xin, Q., Lin, J., Ying, P., Feng, Z., Li, J., Yang, W., Wang, Y., Wang, G., Liu, X., Lin, M., Wang, X., and Min, E., Angew. Chem., Int. Ed. 38, 2220 (1999). 117. Yang, Q., Wang, S., Lu, J., Xiong, G., Feng, Z., Xin, Q., and Li, C., Appl. Catal. A: Gen. 194/195, 507 (2000). 118. Yin, J., Feng, Z., Xu, L., Li, M., Xin, Q., Lin, Y., and Li, C., Mater. Chem. 13, 1994 (2001). 119. Deo, G., Turex, A.M., Wachs, I.E., Huybrechts, D.R.C., and Jacobs, P.A., Zeolites 13, 365 (1993). 120. Nishimura, Y., Hirakawa, A.Y., and Tsuboi, M., in “Advances in Infrared and Raman Spectroscopy” (R.J.H. Clark and R.E. Hester, Eds.). Heydon & Sons, London, 1978. 121. Carey, P.R., “Biochemical Applications of Raman and Resonance Raman Spectroscopies.” Academic Press, New York, 1982. 122. Tuel, A., Diab, J., Gelin, P., Dufaux, M., Dutel, J.-F., and Ben Taarit, Y., J. Mol. Catal. 63, 95 (1990). 123. Zecchina, A., Spoto, G., Bordiga, S., Ferrero, A., Leofanti, G.G., and Padovan, M., Stud. Surf. Sci. Catal. 69, 251 (1991). 124. Prakash, A.M., and Kevan, L., J. Catal. 178, 588 (1998). 125. Luan, Z., and Kevan, L., J. Phys. Chem. B 101, 2020 (1997). 126. Prakash, A.M., Kurshev, V., and Kevan, L., J. Phys. Chem. B 101, 9794 (1997). 127. Prakash, A.M., Sung-Suh, H.M., and Kevan, L., J. Phys. Chem. B 102, 857 (1998). 128. Prakash, A.M., Kevan, L., Zahedi-Niaki, M.H., and Kaliaguine, S., J. Phys. Chem. B 103, 831 (1999). 129. Zhu, Z., Hartmann, M., Macs, E.M., Czernuszewicz, R.S., and Kevan, L., J. Phys. Chem. B 104, 4690 (2000). 130. Bal, R., Chaudhari, K., Srinivas, D., Sivasanker, S., and Ratnasamy, P., J. Mol. Catal. A: Chem. 162, 199 (2000). 131. Chaudhari, K., Bal, R., Srinivas, D., Chandwadkar, A.J., and Sivasanker, S., Micropor. Mesopor. Mater. 50, 209 (2001). 132. Morey, M.S., O’Brian, S., Schwarz, S., and Stucky, G.D., Chem. Mater. 12, 898 (2000). 133. Lin, W., and Frei, H., J. Am. Chem. Soc. 124, 9292 (2002).

164

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

134. Bordiga, S., Damin, A., Bonino, F., Zecchina, A., Spano`, G., Rivetti, F., Bolis, V., Prestipino, C., and Lambert, C., J. Phys. Chem. B. 106, 9892 (2002). 135. Zecchina, A., Spoto, G., Bordiga, S., Padovan, M., Leofanti, G., and Petrini, G., Stud. Surf. Sci. Catal. 65, 671 (1991). 136. Armaroli, T., Milella, F., Notari, B., Willey, R.J., and Busca, G., Top. Catal. 15, 63 (2001). 137. Huybrechts, D.R.C., Vaesen, I., Li, H.X., and Jacobs, P.A., Catal. Lett. 8, 237 (1991). 138. Thangaraj, A., Sivasanker, S., and Ratnasamy, P., J. Catal. 137, 252 (1992). 139. Ichihashi, H., and Sato, H., Appl. Catal. A: Gen. 221, 359 (2001). 140. Srivastava, R., Srinivas, D., and Ratnasamy, P., Catal. Lett. 91, 133 (2003). 141. Kno¨zinger, H., (G. Ertl, H. Kno¨zinger and J. Weitkamp, Eds.), Handbook of Heterogeneous Catalysis, Vol. 2, p. 707. Wiley-VCH, Weinheim, 1997. 142. Paushkits, E.A., and Yurchenko, E.N., Russ. Chem. Rev. 52, 42 (1983). 143. Manoilova, O.V., Dakka, J., Sheldon, R.A., and Tsyganenko, A.A., Stud. Surf. Sci. Catal. 94, 163 (1995). 144. Kno¨zinger, H., in “The Hydrogen Bond” (P. Schuster, G. Zundel and C. Sandorfy, Eds.), Vol. III, p. 1265. North-Holland, Amsterdam, 1976. 145. Zhuang, J., Yan, Z., Liu, X., Liu, X., Han, X., Bao, X., and Mueller, U., Catal. Lett. 83, 87 (2002). 146. Kno¨zinger, H., and Krietenbrink, H., J. Chem. Soc., Faraday Trans. 71, 2421 (1975). 147. Redijk, J., Zaur, A.P., and Groeneveld, W.L., Rec. Trav. Chim. 86, 1127 (1967). 148. Bertran, J.F., La Serna, B., Doerffel, K., Dathe, K., and Kabish, G., J. Mol. Struct. 95, 1 (1982). 149. Busca, G., Saussey, H., Saur, O., Lavalley, J.C., and Lorenzelli, V., Appl. Catal. 14, 245 (1985). 150. Barzetti, T., Selli, E., Moscotti, D., and Forni, L., J. Chem. Soc., Faraday Trans. 92, 1401 (1996). 151. Farnoth, W.E., and Gorte, R.J., Chem. Rev. 95, 615 (1995). 152. Srinivas, D., Srivastava, R., Ratnasamy, P., Catal. Today (2004) (in press). 153. Zecchina, A., Bordiga, S., Spoto, G., Damin, A., Berlier, G., Bonino, F., Prestipino, C., and Lamberti, C., Top. Catal. 21, 67 (2002). 154. Connor, J.A., and Ebsworth, E.A.V., in “Advances in Inorganic Chemistry and Radiochemistry” (H.J. Emeleus and A.G. Sharpe, Eds.), p. 287. Academic Press, New York, 1964. 155. Mimoun, H., in “The Chemistry of Functional Groups, Peroxides” (S.J. Patai, Ed.), p. 463. Wiley, New York, 1983. 156. Fujita, S., Suzuki, K., Ohkawa, M., Mori, T., Jida, Y., Miwa, Y., Masuda, H., and Shimada, S., Chem. Mater. 15, 255 (2003). 157. Giamello, E., Sojka, Z., Che, M., and Zecchina, A., J. Phys. Chem. 90, 6084 (1986). 158. Bodoardo, S., Geobaldo, F., Penazzi, N., Arrabito, M., Rivetti, F., Spano`, G., Lamberti, C., and Zecchina, A., Electrochem. Commun. 2, 349 (2000). 159. Che, M., and Tench, A., Adv. Catal. 32, 1 (1983). 160. Njo, S.L., van Koningsveld, H., and van de Graaf, B., J. Phys. Chem. B 101, 10065 (1997). 161. Oumi, Y., Matsuba, K., Kubo, M., Inui, T., and Miyamoto, A., Micropor. Mesopor. Mater. 4, 53 (1995). 162. Atoguchi, T., and Yao, S., J. Mol. Catal. A: Chem. 191, 281 (2003). 163. Zentys, A., and Catlow, C.R.A., Catal. Lett. 22, 251 (1993). 164. Maseras, F., Chem. Commun. 1821 (2000), and references therein. 165. Kumar, P., Kumar, R., and Pandey, B., Synlett 289 (1995). 166. Ramaswamy, M., and Roesky, H.W., Angew. Chem., Int. Ed. Engl. 36, 477 (1997).

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

165

167. Sheldon, R.A., Arends, I.W.C.E., and Lempers, H.E.B., Heterogeneous catalysts for liquid phase oxidations (supported reagents and catalysis in chemistry). Spec. Publ. R. Soc. Chem. 216, 37 (1998). 168. Sheldon, R.A., Arends, I.W.C.E., and Lempers, H.E.B., Collect. Czech. Chem. Commun. 63, 1724 (1998). 169. Clerici, M.G., in “Fine Chemicals through Heterogeneous Catalysis” (R.A. Shelden and H. van Bekkum, Eds.), p. 538, Wiley-VCH, Weinheim, Germany, 2001. 170. Patai, S. (Ed.), “The Chemistry of Peroxides.” Wiley, New York, 1983. 171. Cremer, D., in “The Chemistry of Peroxides” (S. Patai, Ed.), p. 43. Wiley, New York, 1983, Chapter 43. 172. Nangia, P.S., and Benson, S.W., J. Phys. Chem. 83, 1138 (1979). 173. Benson, S.W., and Nangia, P.S., Acc. Chem. Res. 12, 223 (1979). 174. Kochi, J.K., “Organometallic Mechanisms and Catalysis.” Academic Press, New York, 1978, p. 50. 175. Gunther, K., Filby, W.G., and Eiben, K., Tetrahedron Lett. 251 (1971). 176. Sheldon, R.A., and van Doorn, J.A., J. Catal. 34, 242 (1974). 177. Sharpless, K.B., and Akashi, K., J. Am. Chem. Soc. 98, 1986 (1976). 178. Nam, W., Han, H.J., Oh, S.-Y., Lee, Y.J., Choi, M.-M., Han, S.-Y., Kim, C., Woo, S.K., and Shin, W., J. Am. Chem. Soc. 122, 8677 (2000). 179. Sawaki, Y., in “The Chemistry of Hydroxyl, Ether and Peroxide Groups” (S. Patai, Ed.), Suppl. E, Vol. 2, p. 587. Wiley, New York, 1993. 180. Arends, I.W.C.E., and Sheldon, R., Topics Catal. 19, 133 (2002). 181. Dusi, M., Mallat, T., and Baiker, A., Catal. Rev.—Sci. Engng 42, 213 (2000). 182. Bartok, M., and Schneider, G., in “The Chemistry of Double-Bonded Functional Groups, Part 2” (S. Patai, Ed.), p. 1229. Wiley, New York, 1997. 183. Sheldon, R.A., Van Doorn, J.A., Schran, C.W.A., and De Jong, A.J., J. Catal. 31, 438 (1973). 184. Sheldon, R.A., in “Applied Homogeneous Catalysis with Organometallic Compounds” (B. Cornils and W.A. Hartman, Eds.), Vol. 1, p. 411. VCH, Weinheim, 1996. 185. Sheldon, R.A., Arends, I.W.C.E., and Lempers, H.E.B., in “Supported Reagents and Catalysts in Chemistry” (B.K. Hodnett, A.P. Kybett, J.H. Clark and K. Smith, Eds.), p. 37. Royal Society of Chemistry, Limerick, 1997. 186. Sheldon, R.A., and van Santen, R.A. (Eds.), “Catalytic Oxidation: Principles and Applications.” World Scientific, Singapore, 1995. 187. “Industrial Organic Chemicals: Starting Materials and Intermediates: An Ullmann’s Encyclopedia,” Vol. 7, p. 4135, Wiley-VCH, Weinheim, Germany, 1999. 188. Schmidt, I., Krogh, A., Wienberg, K., Carlsson, A., Brorson, M., and Jacobsen, C.J.H., Chem. Commun. 2157 (2000). 189. Guidotti, M., Ravasio, N., Psaro, R., Ferraris, G., and Moretti, G., J. Catal. 214, 242 (2003). 190. Bouh, A.O., Rice, G.L., and Scott, S.L., J. Am. Chem. Soc. 121, 7201 (1999). 191. Corma, A., Esteve, P., Martı´nez, A., and Valencia, S., J. Catal. 152, 18 (1995). 192. Romano, U., Esposito, A., Maspero, F., Neri, C., and Clerici, M.G., Chim. Ind. 72, 610 (1990). 193. Tatsumi, T., Yako, M., Nakamura, M., Yuhara, Y., and Tominaga, H., J. Mol. Catal. 78, L41 (1993). 194. Bhaumik, A., Kumar, R., and Ratnasamy, P., Stud. Surf. Sci. Catal. 84C, 1883 (1994). 195. Kumar, R., Pais, G.C.G., Pandey, B., and Kumar, P., J. Chem. Soc., Chem. Commun. 1315 (1995). 196. van der Waal, J.C., Rigutto, M.S., and van Bekkum, H., Appl. Catal. A: Gen. 167, 331 (1998).

166

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

197. Corma, A., Navarro, M.T., Perez-Pariente, J., and Sanchez, F., Stud. Surf. Sci. Catal. 84, 69 (1994). 198. Vercruysse, K.A., Klingeleers, D.M., Colling, T., and Jacobs, P.A., Stud. Surf. Sci. Catal. 117, 469 (1998). 199. Adam, W., Corma, A., Reddy, T.I., and Renz, M., J. Org. Chem. 62, 3631 (1997). 200. Adam, W., Kumar, R., Reddy, T.I., and Renz, M., Angew. Chem., Int. Ed. Engl. 35, 880 (1996). 201. Hutchings, G.J., and Lee, D.F., J. Chem. Soc., Chem. Commun. 1095 (1994). 202. Hutchings, G.J., Lee, D.F., and Minihan, A.R., Catal. Lett. 33, 369 (1995). 203. Hutchings, G.J., Lee, D.F., and Minihan, A.R., Catal. Lett. 39, 83 (1996). 204. Clerici, M.G., and Ingallina, P., J. Catal. 140, 71 (1993). 205. Shetti, V.N., Srinivas, D., and Ratnasamy, P., J. Mol. Catal. A: Chem. 210, 171 (2004). 206. Ratnasamy, P., and Kumar, R., Stud. Surf. Sci. Catal. 97, 367 (1995). 207. Laha, S.C., and Kumar, R., J. Catal. 208, 339 (2002). 208. Laufer, W., Meiers, R., and Ho¨lderich, W.F., J. Mol. Catal. A: Chem. 141, 215 (1999). 209. Sato, A., Oguri, M., Tokumaru, S., Miyake, T., JP Patent 08269029A (1996). 210. Miller, U., Lingelbach, P., Bassler, P., Harder, W., Eller, K., Kohl, V., Dembowski, J., Rieber, N., Fischer, M., EP Patent 7,724,91 A1 (1995). 211. Olah, G.A., and Ohnishi, R., J. Org. Chem. 43, 865 (1978). 212. Olah, G.A., Fung, A.P., and Keumi, T., J. Org. Chem. 46, 4305 (1981). 213. Kurz, M.E., and Johnson, G., J. Org. Chem. 36, 3184 (1971). 214. Olah, G.A., and Ernst, T.D., J. Org. Chem. 54, 1204 (1989). 215. Ogata, Y., and Tomizawa, K., J. Org. Chem. 43, 261 (1978). 216. Clerici, M.G., Appl. Catal. 68, 249 (1991). 217. Huybrechts, D.R.C., Parton, R.F., and Jacobs, P.A., Stud. Surf. Sci. Catal. 65, 225 (1991). 218. Ramaswamy, A.V., and Sivasanker, S., Catal. Lett. 22, 239 (1993). 219. Ramaswamy, A.V., Sivasanker, S., and Ratnasamy, P., Micropor. Mater. 2, 451 (1994). 220. Sen, A., in “Applied Homogeneous Catalysis with Organometallic Compounds,” 2nd Edition. p. 1226 Wiley, New York, 2002. 221. Khouw, C.B., Dartt, C.B., Labinger, J.A., and Davis, M.E., J. Catal. 149, 195 (1995). 222. Geobaldo, F., Bordiga, S., Zecchina, A., Giamello, E., Leofanti, G., and Petrini, G., Catal. Lett. 16, 109 (1992). 223. Poladi, R.H.P.R., and Landry, C.C., Micropor. Mesopor. Mater. 52, 11 (2002). 224. Thangaraj, A., Kumar, R., and Ratnasamy, P., Appl. Catal. 55, L1 (1990). 225. Rao, P.R.H.P., Ramaswamy, A.V., and Ratnasamy, P., J. Catal. 137, 225 (1992). 226. Rao, P.R.H.P., and Ramaswamy, A.V., Appl. Catal. A: Gen. 93, 123 (1993). 227. Bianchi, D., Balducci, L., Bortolo, R., Aloisio, D., Ricci, M., Tassinari, R., and Ungarelli, R., Angew. Chem., Int. Ed. 42, 4937 (2003). 228. Reddy, J.S., and Jacobs, P.A., J. Chem. Soc., Perkin. Trans. 1, 2665 (1993). 229. Wu, P., and Tatsumi, T., J. Catal. 214, 317 (2003). 230. Sonawane, H.R., Pol, A.V., Moghe, P.P., Biswas, S.S., and Sudalai, A., J. Chem. Soc., Chem. Commun. 1215 (1994). 231. Zecchina, A., Spoto, G., Bordiga, S., Geobaldo, F., Petrini, G., Leofanti, G., Padova, M., Mantegazza, M., and Roffia, P., in “Proceedings of the 10th International Congress on Catalysis” (L. Guczi, F. Solymosi and P. Te´te´nyi, Eds.), p. 719. Akademiai Kiado, Budapest, 1993; Stud. Surf. Sci. Catal. 75, 719 (1993). 232. Reddy, R.S., Reddy, J.S., Kumar, R., and Kumar, P., J. Chem. Soc., Chem. Commun. 84 (1992). 233. Kumar, R., Reddy, J.S., Reddy, R.S., and Kumar, P., in “Selective Oxidation in Petrochemistry” (J. Weitkamp, Ed.), DGMK Conference, p. 367, 1992.

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

167

234. Trukhan, N.N., Romannikov, V.N., Shmakov, A.N., Vanina, M.P., Paukshtis, E.A., Bukhtiyarov, V.I., Kriventsov, V.V., Danilov, I.Y., and Kholdeeva, O.A., Micropor. Mesopor. Mater. 59, 73 (2003). 235. Hutter, R., Mallat, T., and Baiker, A., J. Catal. 153, 177 (1995). 236. Kholdeeva, O.A., Trukhan, N.N., Vanina, M.P., Romanikov, V.N., Parmon, V.N., Bialon, J.M., and Jerzebski, A.B., Catal. Today 75, 203 (2002). 237. Hagen, A., Schueler, K., and Roessner, F., Micropor. Mesopor. Mater. 51, 23 (2002). 238. Hulea, V., Fajula, F., and Bousquet, J., J. Catal. 198, 179 (2001). 239. van der Pol, A.J.H.P., and van Hooff, J.H.C., J. Appl. Catal. A: Gen. 106, 97 (1993). 240. Corma, A., Esteve, P., and Martı´nez, A., Appl. Catal. A: Gen. 143, 87 (1996). 241. Sasidharan, M., Suresh, S., and Sudalai, A., Tetrahedron Lett. 36, 9071 (1995). 242. Harrison, I.T., and Harrison, S., J. Chem. Soc., Chem. Commun. 752 (1966). 243. Clarlsen, P.H.J., Katsuki, T., Martin, V.S., and Sharpless, K.B., J. Org. Chem. 46, 3936 (1981). 244. Bhaumik, A., Kumar, P., and Kumar, R., Catal. Lett. 40, 47 (1996). 245. Wang, Z.B., Mizusaki, T., Sano, T., and Kawakami, Y., Bull. Chem. Soc. Jpn 70, 2567 (1997). 246. Joseph, R., Sudalai, A., and Ravindranathan, T., Tetrahedron Lett. 35, 5493 (1994). 247. Reddy, J.S., Ravishanker, R., Sivasanker, S., and Ratnasamy, P., Catal. Lett. 17, 139 (1993). 248. Srivastava, R., Srinivas, D., and Ratnasamy, P., Catal. Lett. 91, 133 (2003). 249. Srivasta, R., Srinivas, D., and Ratnasamy, P., in “Proceedings 14th International Zeolite Conference” (E.W.J. van Steen, L.H. Callanan, M. Claeys and C.T. O’Connor, Eds.), p. 2703, Produced by: Document Transformation Technologies, Cape Town, South Africa (2004). 250. Bhanage, B.M., Fujita, S.I., Ikushima, Y., Torri, K., and Arai, M., Green Chem. 5, 71 (2003). 251. Ma, X., Guo, H., Wang, S., and Sun, Y., Fuel Proc. Technol. 83, 275 (2003). 252. Otera, J., Chem. Rev. 93, 1449 (1993). 253. Schuchardt, U., Sercheli, R., and Vargas, R.M., J. Braz. Chem. Soc. 9 (1998). 254. Tatsumi, T., Watanabe, Y., and Koyano, K.A., Chem. Commun. 2281 (1996). 255. Mukaiyama, T., Bano, K., and Narasaka, K., J. Am. Chem. Soc. 96, 7503 (1974). 256. Kawai, M., Onaka, M., and Izumi, B., Bull. Chem. Soc. Jpn 61, 1237 (1988). 257. Sasidharan, M., and Kumar, R., J. Catal. 220, 326 (2003). 258. Sasidharan, M., Wu, P., and Tatsumi, T., J. Catal. 209, 260 (2002). 259. Schuster, W., Niederer, J.P.M., and Hoelderich, W.F., Appl. Catal. A: Gen. 209, 131 (2001). 260. Hanmann, M.M., in “Photodegradation of Water Pollutants”. CRC Press, New York, 1995. 261. Anpo, M., Yamashita, H., Ichihashi, Y., Fujii, Y., and Honda, M., J. Phys. Chem. B 101, 2632 (1997). 262. Kang, M.G., Park, H.S., and Kim, K.-J., J. Photochem. Photobiol. A: Chem. 149, 175 (2002). 263. Anpo, M., Yamashita, H., Ikeue, K., Fujii, Y., Zhang, S.G., Ichihashi, Y., Park, D.R., Suzuki, Y., Koyano, K., and Tatsumi, T., Catal. Today 44, 327 (1998). 264. Ulagappan, N., and Frei, H., J. Phys. Chem. A 104, 7834 (2000). 265. Yamashita, H., Ichihashi, Y., Zhang, S.G., Matsumura, Y., Souma, Y., Tatsumi, T., and Anpo, M., Appl. Surf. Sci. 121/122, 305 (1997). 266. Thangaraj, A., Puthoor, L., and Sivasanker, S., Indian J. Chem. A 33, 255 (1994). 267. Atoguchi, T., and Yao, S., J. Mol. Catal. A: Chem. 176, 173 (2001). 268. Tantanak, D., Vincent, M.A., and Hillier, I.H., J. Chem. Soc., Chem. Commun. 1031 (1998). 269. Sinclair, P.E., and Catlow, C.R.A., J. Phys. Chem. B 103, 1084 (1999).

168

270. 271. 272. 273. 274. 275. 276. 277. 278. 279. 280. 281. 282. 283. 284. 285. 286. 287. 288. 289. 290. 291. 292. 293. 294. 295. 296. 297. 298. 299. 300. 301. 302. 303. 304. 305. 306. 307.

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

Sauer, M.V.C., and Edwards, J.O., J. Phys. Chem. 75, 3004 (1971). Bach, R.D., and Knigh, J.W., Org. Synth. 60, 63 (1981). McIsaac, J.E., Ball, R.E., and Behrman, E.J., J. Org. Chem. 36, 3048 (1971). Tatsumi, T., Koyano, K.A., and Igarashi, N., Chem. Commun. 325 (1998). Wu, P., Tatsumi, T., Komatsu, T., and Yashima, T., Chem. Mater. 14, 1657 (2002). Thangaraj, A., Kumar, R., Mirajkar, S.P., and Ratnasamy, P., J. Catal. 130, 1 (1991). Mantegazza, M.A., Petrini, G., Spano, G., Bagatin, R., and Rivetti, F., J. Mol. Catal. A: Chem. 146, 223 (1999). Chaudhari, K., Bal, R., Srinivas, D., Chandwadkar, A.J., and Sivasanker, S., Micropor. Mesopor. Mater. 50, 209 (2001). Adam, W., Harold, M., Hill, C.L., Saha-Mo¨ller, C.R., and Eur, J. Org. Chem. 941 (2002). Corma, A., Chem. Rev. 97, 2373 (1997). Biz, S., and Occelli, M.L., Catal. Rev. Sci. Engng 40, 329 (1998). Thangaraj, A., Eapen, M.J., Sivasanker, S., and Ratnasamy, P., J. Catal. 12, 943 (1992). Padovan, M., Leofanti, G., Roffia, P., EP Patent No. 311983 (1989). Padovan, M., Genoni, F., Leofanti, G., Petrini, G., Trezza, G., and Zecchina, A., Stud. Surf. Sci. Catal. 63, 431 (1991). Bellussi, G., Buonomo, F., Esposito, A., Clerici, M., Romano, U., Notari, B., US Patent No. 4, 701,428 (1987). Tuel, A., Ben Taarit, Y., and Naccache, C., Zeolites 13, 454 (1993). Tuel, A., and Ben Taarit, Y., Micropor. Mater. 1, 179 (1993). Tuel, A., and Ben Taarit, Y., Zeolites 14, 272 (1994). Tuel, A., and Ben Taarit, Y., Appl. Catal. A. Gen. 110, 137 (1994). Tuel, A., Zeolites 16, 108 (1996). Gontier, S., and Tuel, A., Zeolites 16, 184 (1996). Tuel, A., Teissier, R., EP Patent Appl. No. 665188 A 1 (1995). Jorda, E., Tuel, A., Treissier, R., and Kervennal, J., Zeolites 19, 238 (1997). Tuel, A., Catal. Lett. 51, 59 (1998). Sabde, D.P., Hegde, S.G., and Dongare, M.K., J. Mater. Chem. 10, 1365 (2000). Guth, J.L., Kessler, H., Higel, J.M., Lamblin, J.M., Patarin, J., Seive, A., Chezeou, J.M., and Way, A., ACS Symp. Ser. 398, 176 (1989). Delmotte, L., Soulard, M., Guth, F., Seive, A., Lopez, A., and Guth, J.L., Zeolites 10, 778 (1990). Grieneisen, H., Kessler, H., Fache, E., and Le Govic, A.M., Micropor. Mesopor. Mater. 37, 379 (2000). Kumar, R., Bhaumik, A., Ahedi, R.K., and Ganapathy, S., Nature 381, 298 (1996). Bhaumik, A., Ganapathy, S., and Kumar, R., Stud. Surf. Sci. Catal. 113, 225 (1998). Kumar, R., Mukherjee, P., Pandey, R.K., Rajmohanan, P., and Bhaumik, A., Micropor. Mesopor. Mater. 22, 23 (1998). Ahn, W.S., Kang, K.K., and Kim, K.Y., Catal. Lett. 72, 229 (2001). Kulkarni, S.K., Prasad, M.R., Kamalakar, G., Raghavan, K.V., Prasad, P.S.S., Rao, K.N., US Patent No. 6,387,348 B1 (2002). Prasad, M.R., Kamalakar, G., Kulkarni, S.J., Raghavan, K.V., Rao, K.V., Sai Prasad, P.S., and Madhavendra, S.S., Catal. Commun. 3, 399 (2002). Khomane, R.B., Kulkarni, B.D., Paraskar, A., and Sainker, S.R., Mater. Chem. Phys. 76, 99 (2002). Reddy, J.S., and Kumar, R., J. Catal. 130, 440 (1991). Uguina, M.A., Serrano, D.P., Ovejero, G., van Griekson, R., and Camacho, M., Zeolites 18, 368 (1997). de Lucas, A., Rodrı´guez, L., and Sa´nchez, P., Appl. Catal. A: Gen. 180, 375 (1999).

P. Ratnasamy, D. Srinivas and H. Kno¨zinger / Adv. Catal. 48 (2004) 1–169

169

308. de Lucas, A., Rodrı´guez, L., and Sa´nchez, P., Chem. Engng Res. Des. 78, 136 (2000). 309. Reddy, K.M., Kaliaguine, S., and Sayari, A., Catal. Lett. 23, 169 (1994). 310. Reddy, K.M., Kaliaguine, S., Sayari, A., Ramaswamy, A.V., Reddy, V.S., and Bonneviot, L., Catal. Lett. 23, 174 (1994). 311. Tuel, A., and Ben Taarit, Y., Zeolites 15, 164 (1995). 312. Sexton, R.J., Crocco, G.L., Zajacek, J.G., Wijesekea, K.S., EP Patent No. 659685 (1994). 313. Camblor, M.A., Costaitini, M., Corma, A., Gilbert, L., Esteve, P., Martinez, A., and Valentia, S., Appl. Catal. 133, L185 (1995). 314. Blasco, T., Camblor, M.A., Corma, A., Esteve, P., Martinez, A., Prieto, C., and Valencia, S., Chem. Commun. 2367 (1996). 315. van der Waal, J.C., Lin, P., Rigutto, M.S., and van Bekkum, H., Stud. Surf. Sci. Catal. 150, 1093 (1997). 316. Reddy, J.S., and Sayari, A., J. Chem. Soc., Chem. Commun. 23 (1995). 317. Dartt, C.B., and Davis, M.E., Appl. Catal. A: Gen. 143, 53 (1996). 318. Jappar, N., Xia, Q., and Tatsumi, T., J. Catal. 180, 132 (1998). 319. Maschmeyer, T., Rey, F., Sankar, G., and Thomas, J.M., Nature 378, 159 (1995). 320. Corma, A., Jorda´, J.L., Navarro, M.T., and Rey, F., Chem. Commun. 1899 (1998). 321. Corma, A., Domine, M., Gaona, J.A., Jorda´, J.L., Navarro, M.T., Rey, F., Pe´rez-Pariente, J., Tsuji, J., McCulloch, B., and Nemeth, L.T., Chem. Commun. 2211 (1998). 322. D’Amore, M.B., and Schwarz, S., Chem. Commun. 121 (1999). 323. Morey, M., Davidson, A., and Stucky, G.D., Micropor. Mesopor. Mater. 6, 99 (1996). 324. Luan, Z., Maes, E.M., van der Heide, P.A.W., Zhao, D., Czernuszewicz, R.S., and Kevan, L., Chem. Mater. 11, 3680 (1999). 325. Uguina, M.A., Ovejero, G., van Griechen, R., Serrano, D.P., and Camacho, M., J. Chem. Soc., Chem. Commun. 27 (1994). 326. Uguina, M.A., Serrano, D.P., Ovejero, G., van Grieken, R., and Camacho, M., Appl. Catal. A: Gen. 124, 391 (1995). 327. Guo, G.L., Zhang, X., and Wang, X., Stud. Surf. Sci. Catal. 112, 499 (1997). 328. Jung, K.T., Hyum, J.H., Shul, Y.G., and Kim, D.S., Zeolites 19, 161 (1997). 329. Wang, X.S., and Guo, X.W., Catal. Today 51, 45 (1999). 330. Gao, H., Lu, W., and Chen, Q., Micropor. Mesopor. Mater. 34, 307 (2000). 331. Kuznicki, S.M., US Patent No. 4,938,939 (1990). 332. Kuznicki, S.M., Thrush, A.K., EP Patent No. 0405978 A1 (1990). 333. Kuznicki, S.M., US Patent No. 4,853,202 (1989). 334. Chapman, D.M., and Roe, A.L., Zeolite 10, 730 (1990). 335. Anderson, M.W., Terasaki, O., Ohsuna, T., Phillippou, A., Mackay, S.P., Ferreira, A., Rochca, S., and Lidin, S., Nature 367, 347 (1994). 336. Anderson, M.W., Terasaki, O., Ohsuna, T., Malley, P.J.O., Phillippou, A., Mackay, S.P., Ferreira, A., Rochca, J., and Lidin, S., Phil. Mag. B 71, 813 (1995). 337. Valtchev, V.P., J. Chem. Soc., Chem. Commun. 1435 (1994). 338. Valtchev, V.P., and Mintova, S., Zeolites 14, 697 (1994). 339. Liu, X., and Thomas, J.K., Chem. Commun. 1435 (1996). 340. Das, T.Kr., Chandwadkar, A.J., Budhkar, A.P., Belhekar, A.A., and Sivasanker, S., Micropor. Mesopor. Mater. 4, 195 (1995). 341. Das, T.Kr., Chandwadkar, A.J., Budhkar, A.P., and Sivasanker, S., Micropor. Mesopor. Mater. 5, 401 (1996). 342. Rocha, J., Ferreira, A., Lin, Z., and Anderson, M.W., Micropor. Mesopor. Mater. 23, 253 (1998). 343. Yang, X., Paillaud, J.-L., van Bruekelen, H.F.W., Kesser, H., and Duprey, E., Micropor. Mesopor. Mater. 46, 1 (2001).

This Page Intentionally Left Blank

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

Electron Microscopy and the Materials Chemistry of Solid Catalysts JOHN MEURIG THOMAS *Davy Faraday Research Laboratory, The Royal Institution of Great Britain, 21 Albemarle Street, London W1S 4BS, UK and also at Department of Materials Science, Cambridge CB2 1QY, UK

and PRATIBHA L. GAI *DuPont, Central Research and Development Laboratories, Experimental Station, Wilmington, DE 19880-0356, USA and also at Department of Materials Science, University of Delaware, Newark, DE 19716, USA

I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II. Electron Microscopy (EM) Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II.A. Electron Microscopy in Catalysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II.B. Imaging in the Electron Microscope . . . . . . . . . . . . . . . . . . . . . . . . . . . . II.C. TEM Imaging Method Using Diffraction Contrast . . . . . . . . . . . . . . . . . . II.D. Theoretical Procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . III. High-Resolution Transmission Electron Microscopy. . . . . . . . . . . . . . . . . . . . . III.A. Conditions Required for Optimizing HRTEM Images . . . . . . . . . . . . . . III.B. Development of HRTEM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . III.C. Elucidation of the Structures of Meso- and Microporous Catalysts by HRTEM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . III.C.1. L-Type Zeolite Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . III.C.2. Metal-Substituted Aluminum Phosphate (MAPO-36) Microporous Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . III.C.3. High-Silica Microporous SSZ-48 Catalysts . . . . . . . . . . . . . . . . III.C.4. Intergrowths in Zeolite Catalysts: Coherent, Recurrent, and Random . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV. Chemical Composition Analysis with the Analytical Electron Microscope. . . . . V. Scanning Transmission Electron Microscopy . . . . . . . . . . . . . . . . . . . . . . . . . .

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

. . . . . . . . .

174 176 177 178 179 181 181 182 184

..... .....

185 185

..... .....

186 187

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

188 191 193

*Corresponding Addresses.

ADVANCES IN CATALYSIS, VOLUME 48 ISSN: 0360-0564 DOI 10.1016/S0360-0564(04)48002-X

Copyright q 2004 Elsevier Inc. All rights reserved

172

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

VI. Recent Advances in Ultra-High Resolution, Low-Voltage Field Emission Scanning Electron Microscopy and Extreme FESEM in Catalysis . . . . . . . . . . . . . . . . VII. Cathodoluminescence Imaging for Elucidation of Electronic Structures of Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VIII. Recent Advances in In Situ Atomic Resolution-Environmental Transmission Electron Microscopy (ETEM) Under Controlled Reaction Conditions. . . VIII.A. In Situ Investigations of Gas –Solid Reactions and Active Sites . . . . . . . . . . . VIII.B. Illustrative Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VIII.B.1. In Situ Gas –Catalyst Reactions at the Atomic Level . . . . . . . . . . . VIII.B.2. Atomic-Resolution ETEM of Butane Oxidation . . . . . . . . . . . . . . . VIII.B.3. Atomic-Resolution ETEM of Nanorods . . . . . . . . . . . . . . . . . . . . . VIII.C. Advances in In Situ Wet-Electron Microscopy Technique (Wet-ETEM) for Probing Solid Catalysts Under Liquid Environments . . . . . . . . . . . . . . . . IX. Environmental Scanning Electron Microscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . X. Electron Tomography: Three-Dimensional Electron Microscopy Imaging . . . . . . . . . . X.A. The Topography and Location of Nanoparticles in Supported Catalysts; BSE and HAADF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . X.B. Pinpointing the Location of Nanoparticles Supported on Nanoporous Solids. . . . XI. Energy Filtered Transmission Electron Microscopy and Elemental Maps of Solid Catalysts Using EFTEM. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XII. Other Significant Trends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XIII. Critical Evaluations of the Methods and Challenges . . . . . . . . . . . . . . . . . . . . . . . . . . XIV. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

195 195 196 196 201 201 203 210 210 212 212 213 218 218 220 220 223 224 224

No other method rivals electron microscopy (EM) in the wealth of structural (atomic, nanoscopic, microscopic, and mesoscopic), topographic, and electronic information that it provides in the characterization of solid catalysts such as those used commercially, for laboratory trials or model studies: EM provides deep insights into the structure of solid catalysts—their precursors, active sites, and expired or regenerated forms—as well as vital clues to their mode of operation. In some important instances it serves as the only trustworthy means of determining the structure and composition of a catalyst. After a brief update on the significance of recent advances in EM techniques which allow (i) the probing of catalysts at atomic resolution, (ii) electron crystallography, and (iii) the determination of the chemical compositions of catalysts, we illustrate these achievements with specific examples. These include (a) pinpointing the location and topography of nanoparticle catalysts; (b) constructing elemental maps (and compositional distributions) of solid catalysts; (c) in situ investigations of active sites and reaction processes at the atomic level; (d) elucidating the nature of intergrowths (coherent, recurrent, and random) of closely similar structures within a supposed new catalyst; (e) identifying atoms (or small groups of atoms) of high atomic number supported on high-area solids; and (f) characterizing nanoparticles on uneven supports. In (a) and (e) the recently developed technique of electron tomography plays a crucial role. q 2004 Elsevier Inc.

Abbreviations A a ADF

absorption unit cell dimension of crystal along a-axis annular dark field

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227 AEM ˚ A b BET b BF c CA CB CBDP CCD Cs CTF Df Df ðSÞ e E ECELL ED EDX EELS EFTEM ELNES EM EPMA ESEM ETEM EXELFS F FESEM FðuÞ FT g GIF L HAADF HREM HRSTEM HVEM IA IB Iðx; yÞ KAB kcal l LVSEM m MA m mm

analytical electron microscopy angstrom units unit cell dimension of crystal along b-axis Brunauer, Emmett, Teller surface area angle between a- and c-axes in the crystal unit cell bright field unit cell dimension of crystal along c-axis concentration of element A in a compound AB concentration of element B in a compound AB convergent beam electron diffraction charge-coupled device coefficient of spherical aberration of the electron microscope objective lens contrast transfer function objective lens defocus value Scherzer defocus value electron charge electron energy environmental cell electron diffraction energy dispersive X-ray spectroscopy electron energy loss spectroscopy energy-filtered transmission electron microscopy electron energy loss near-edge structure electron microscopy (or microscope) electron probe microanalysis environmental scanning electron microscopy environmental-TEM extended energy loss fine structure fluorescence field emission scanning EM electron envelope function Fourier transform gram Gatan imaging filter Green’s function high-angle ADF or HRTEM, high-resolution TEM high-resolution scanning TEM high-voltage EM background-subtracted peak intensity of element A background-subtracted peak intensity of element B image intensity of sample in the image plane, k, rate constant sensitivity factor in analysis for elements A and B in compound AB kilo calories wavelength of electrons low-voltage SEM electron mass maleic anhydride meter micrometers

173

174 mbar mm mrad mol nm PEELS f0 fg SEM s sinðxÞ Cðx; yÞ

CðrÞ SMSI STM t ti TEM u XAFS XRD XRE V Vðx; yÞ VðrÞ WDS WPO Z

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227 millibar millimeters milliradians mole nanometers parallel EELS amplitude of electron wave incident on sample amplitude of scattered electron wave scanning EM interaction constant contrast transfer function (CTF) electron wave function at exit face of sample, with incident electrons along z-direction electron wave function at the spatial coordinate, r strong metal–support interactions scanning tunneling microscopy sample thickness time transmission electron microscopy scattering angle of electrons in radians X-ray absorption fine structure X-ray diffraction X-ray emission volume of crystal unit cell thickness-projected crystal potential crystal potential at the spatial coordinate, r wavelength dispersive X-ray spectroscopy weak phase object atomic number

I. Introduction Most commercial catalysts are powdered solids that consist of one or two distinct phases (or polyphasic aggregates) or of supported metallic components on higharea supports of a quite different composition (such as oxides, chacolgenides or halides). Table I is a list of elements present in typical catalysts. A wide range of techniques has been developed (1 –4) to characterize the composition and structure of surfaces of model catalysts, such as single-crystals of metals, alloys or oxides. These techniques include low-energy electron diffraction (ED), sumfrequency generation, and polarized reflection – absorption infrared spectroscopy and others that are usually inapplicable in the characterization of commercial catalysts and of no value in determining structural, electronic, or compositional information for functioning catalysts. Insofar as most solid catalysts are concerned, characterization entails, inter alia, the determination of surface composition; the number and nature of distinct

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

175

TABLE I A selection of typical commercial and viable new solid catalysts Elements present in the catalyst

Process catalyzed

Fe, K (Al, Si, O)

Synthesis of ammonia

Mo (W), S, Co (Ni)

Hydrodesulfurization

V, P, O

Selective oxidation of butane

Co (Mn), Al, P, O

Oxyfunctionalization of alkanes

La, Pt, Al, Si, O

Cracking of hydrocarbons

Pt, Re (Ir), Al (Si), O

Naphtha reforming

Ti, Si, O

Alkene epoxidation

Si (P), Mo (W), O, Cs (Na), Co, Al, P, O

Dehydration of alkanols

crystallographic phases; electronic properties of the catalyst (encompassing such information as the oxidation states of particular atoms, especially those at active sites) and their coordination to surrounding atoms; the location of active sites; reaction mechanisms; the mode of release of structural oxygen; and accommodation of the catalyst non-stoichiometry (3,4). In the growing field of nanoporous solids (used as catalysts or catalyst supports), the atomic structure of the framework (5), as well as the nature of its nanoporosity, needs to be determined. For the elucidation of these properties, electron microscopy (EM), used in one or more of its many modern variants — high-resolution (real-space) imaging, or as a means of effecting electron crystallography, or as a powerful scanning probe instrument, or as an electron energy loss spectroscopic (EELS) tool — is of unrivalled value. No other single tool yields such a wealth of diverse information concerning solid catalysts and their surfaces. The sophistication, reliability, and ease of operation of electron microscopes have increased enormously since their early applications, which included channelling of metallic particles across the surfaces of graphite (6), and a range of physico chemical problems have been solved (7– 10). In contrast, mass spectrometers, for example, are very powerful tools, but the information they yield is largely compositional. Likewise, laser-based spectroscopic tools (such as laser-induced Raman or infrared (IR) spectroscopy) yield insights that are largely related to bonding and site environment. Scanning probe methods, especially STM, provide great detail and high resolution concerning atomic arrangements at surfaces (even under in situ conditions), but they yield essentially no information about atomic composition and diffraction. In addition to the information enumerated above that is important in the characterization of catalysts, we also require as much knowledge as possible

176

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

about the electronic states of individual atoms, the electronic (band) structure of the solid and — for specific active sites in, say, oxide catalysts — the statics and dynamics of bonding of the atoms that constitute these sites. So far as the last named desideratum is concerned, X-ray absorption fine structure (XAFS) is the prime technique of choice (11– 13). But such is the progress that has recently been made (12,14) in electron energy loss near-edge structure (ELNES) analysis using electron microscopes equipped with the appropriate electron spectrometers that there are real prospects for retrieval of information equivalent to that which XAFS (15) yields from micro- and nano-regions of a catalyst, in EM studies. In the following sections, we summarize some of the most advanced and novel EM methods that are playing pivotal roles in the understanding of solid catalysts. We then proceed to demonstrate the veracity of the claims made above about the unique power of EM in catalyst characterization. The reader is also directed to a series of up-to-date authoritative reviews pertaining to EM and catalysis contained in Ref. (16). In particular, there are comprehensive reviews of energyfiltered TEM (EFTEM), which has advantages in constructing element-image maps of specimens under consideration, including solids of catalytic interest such as carbon nanotubes (17) and the development of in situ atomic resolution-ETEM for direct probing of dynamic catalytic reactions at the atomic scale (18).

II. Electron Microscopy (EM) Methods The use of EM (except in the special case of SEM) demands that the catalyst, whether mono-or multi-phasic, be thin enough to be electron transparent. But, as we show below, this seemingly severe condition by no means restricts its applicability to the study of metals, alloys, oxides, sulfides, halides, carbons, and a wide variety of other materials. Most catalyst powder preparations and supported metallic catalysts, provided that representative thin regions are selected for characterization, are found to be electron transparent and thus amenable to study by EM without the need for further sample preparation. In recent years, increasing use has been made of in situ methods in EM—as is true of other techniques of catalyst characterization such as IR, Raman, and NMR spectroscopy, or X-ray diffraction. Although the low mean-free path of electrons prevents EM from being used when model catalysts are exposed to pressures comparable to those prevailing in industrial processes, Gai and Boyes (4) reported early investigations of in situ EM with atomic resolution under controlled reaction conditions to probe the dynamics of catalytic reactions. Direct in situ investigation permits extrapolation to conditions under which practical catalysts operate, as described in Section VIII. Most applications of EM to catalysis take advantage of high-resolution transmission EM (HRTEM) instruments, and the structures of an ever-increasing

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

177

number of molecular sieve catalysts have been determined by HRTEM. Scanning transmission EM (STEM) instruments, however, as well as sophisticated variants of conventional SEM, which are ideal for determining both the morphology and the composition of exterior layers of solid catalysts in a spatially highly resolved fashion, plays a significant role in the characterization of catalysts and related materials such as precursor gels or supports. The modern-day analytical EM (AEM) is capable of achieving a multiplicity of functions: information pertaining to structure and/or phase purity comes via ED patterns and real-space images; composition on the other hand emerges from the electron-stimulated X-ray emission (XRE) peaks or from EELS. And because of advances in the technology of energy dispersive detectors for XRE spectra and in parallel processing of EEL spectra, commercial EMs are now routinely equipped with those two powerful analytical capabilities. They are also equipped with more sensitive means of recording, digitizing (and processing by, for example, Fourier transform (FT) and various filtering procedures) transmission images of a sample. One of the significant instrumental advances has been in the field of detection and recording of diffracted or focused electrons. One of the difficulties is the occurrence of electron-beam damage in EMs (19), and low electron-dose imaging methods are required to eliminate it. The traditional electron microscope quality film first gave way to TV recording (with an improved sensitivity and a slightly inferior dynamic range). But then came the image plate (IP) and the slow scan charge-coupled device (CCD) (20), each possessing very high sensitivities (2 £ 10214 and 5 £ 10214 C cm22) and dynamic ranges of 4.0 and 2.5 orders of magnitude, respectively. This revolutionary improvement in detection now makes it possible to deploy novel electron crystallography (21,22) to solve the crystal structures of microscopic samples such as siliceous mesoporous materials in a manner analogous to conventional X-ray crystallography using direct methods. The key difference, however, apart from the inability of the X-ray crystallography to cope with the minute specimens now solvable by electron crystallography, is that, with so-called mesoporous (open-structure) solids such as the SSZ-48 silica family, it is in principle impossible to determine the details of the pore topology (when the pore diameters are in the range 1 –20 nm) using X-ray crystallography, when the framework silica that constitutes the filigree arrangement of pores is structurally disordered (as demonstrated by solid-state NMR).

II.A. Electron Microscopy in Catalysis Traditional approaches to explore catalysts are generally based on indirect chemical and spectroscopic methods. Constructions of structural or mechanistic models of reactions on the surfaces of complex catalysts based on such methods often provide

178

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

incomplete or inadequate pictures of the processes involved. EM is providing important insights into changes in the atomic structure and chemistry of reactions that profoundly influence catalytic properties. These have prompted the development of new catalytic materials, the solution of complex structures, and also the optimization of catalytic properties by delicate control of the solid structures. In this chapter, we outline some of the most significant recent developments in EM methods, including in situ EM techniques for probing catalysis and active sites at the atomic level, the imaging conditions required to obtain the local fine structure, and the chemistry of the catalysts. We also briefly discuss limitations and future trends.

II.B. Imaging in the Electron Microscope Electrons undergo scattering as a result of the beam-sample interactions. An essential feature of EM is diffraction. Crystals (samples) diffract electrons according to Bragg’s law. The diffraction pattern thus formed may be regarded as the FT of the crystal, and hence an inverse FT in the objective lens forms the image. With high-energy electrons ($ 100 kV) incident on a sample, a number of signals are emitted, which can be used for structural and chemical analyses (Fig. 1). These signals result from elastically and inelastically scattered transmitted electrons, characteristic X-rays, and back-scattered and secondary electrons (SEs). In the operation of a conventional transmission electron microscope (CTEM), the electron beam generated by a filament passes through a condenser lens system, and the collimated beam is then incident upon the sample. Scattered rays from the same point are brought to focus in the image formed by the objective lens (Fig. 2a). The associated signals are illustrated schematically in Fig. 2b. The characteristics of the objective lens (its spherical aberration coefficient, Cs ; and the accelerating voltage (wavelength of electrons, l) determine the image resolution. Parallel electron beams interfere in the back focal plane (bfp) of the objective lens to form a diffraction pattern. The information in the image is present in the diffraction pattern originating from the same region of the sample. The relationship between the image and the diffraction pattern is that of direct (real) and reciprocal space. These are mutually complementary in the interpretation of structural characteristics of the sample. The intermediate lens can bring into focus either the image or the electron diffraction (ED) pattern (through a change in its focal length) onto the focal plane of the projector lens system, which magnifies the image on the screen. The point (or interpretable) resolution, d; depends on the wavelength of the electron beam, l; and the spherical aberration coefficient Cs and is given by a simple relationship:

d , 0:64Cs1=4 l3=4

ð1Þ

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

179

Fig. 1. Schematic of the information from elastically and inelastically scattered electrons during the electron beam–sample interactions.

To improve the resolution, one, therefore, minimizes Cs (with aberration correction almost to zero) and increases the electron energy. However, electron energy spread and stability issues are also critical as the resolution is improved.

II.C. TEM Imaging Method Using Diffraction Contrast In TEM diffraction contrast imaging, the Bragg condition is satisfied for a single diffracted beam (23). The interpretable resolution depends on the size of the objective aperture (i.e., it is diffraction limited) and can be of the order of 1 nm. If the objective aperture includes only the diffracted beam corresponding to the incident electron beam direction (primary beam containing the direct transmitted electrons), a bright-field (BF) image is obtained. The contrast is produced as a consequence of differences in electron intensities scattered into Bragg reflections from different areas of a thin sample. In imaging, if only scattered electrons are included, a dark-field (DF) image is formed.

180

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

Fig. 2. (a) Ray diagram in the electron microscope under imaging (microscopy) conditions. E: electron source; C: condenser lens; S: sample; O: objective lens; bfp: back focal plane of O; I: intermediate lens; P: projector lens. (b) Structural imaging, diffraction and compositional functionalities of TEM.

The diffraction contrast technique is very useful in determining the nature of defects or lattice imperfections of catalysts. The technique can be used to analyze dislocations in catalysts by determining components of their displacement vector (called the Burgers or shear vector, b) in the three crystallographic dimensions and to define the three-dimensional geometry of defects. (In the HRTEM method described below, which gives a planar image, calculations may be necessary to ascertain the component of the displacement vector of the defect normal to the plane of projection.) Defects such as dislocations play a key role in governing the properties of catalysts, and understanding their nature is critical in the optimization of catalytic properties. There are established criteria for obtaining b by using diffraction contrast (23). Briefly, the dislocation intensity (contrast) is mapped in several Bragg reflections (denoted by vector, g) by tilting the crystal to different reflections and determining the dot product of the vectors g and b (called the g·b product analysis).

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

181

The reflections include a particular g in which the dislocation is invisible (i.e., g·b ¼ 0 when b is normal to the reflecting plane). With these criteria in diffraction contrast, one can determine the character of the defect, e.g., screw (where b is parallel to the screw dislocation line or axis), edge (with b normal to the line), or partial (incomplete) dislocations. The dislocations are termed screw or edge, because in the former the displacement vector forms a helix and in the latter the circuit around the dislocation exhibits its most characteristic feature, the half-plane edge. By definition, a partial dislocation has a stacking fault on one side of it, and the fault is terminated by the dislocation (23– 25). The nature of dislocations is important in understanding how defects form and grow at a catalyst surface, as well as their critical role in catalysis (3,4). We now briefly review some theoretical aspects of transmission ED using high-energy electrons based on an electron wave mechanical formulation of the dynamical theory of contrast.

II.D. Theoretical Procedures The steady-state wave function CðrÞ describing electrons with energy E moving in a crystal potential VðrÞ obeys the Schro¨dinger equation: 72 CðrÞ þ 8p2 meðE þ VðrÞÞCðrÞ ¼ 0

ð2Þ

Where m and e are the electronic mass and charge, respectively, and r is the spatial coordinate (23). To interpret electron micrographs and diffraction patterns, it is essential to understand electron scattering mechanisms occurring through the crystal. In kinematical theory of ED contrast, the amplitude of a scattered electron wave ðfg Þ is a small fraction of the amplitude of the incident wave ðf0 Þ and the kinematical theory is valid only for thin crystals. In thicker crystals, kinematical theory breaks down because of multiple scattering effects, and the dynamical theory incorporating Bloch wave functions should be used instead. Intensity (contrast) calculations for specific defects located at a particular depth in a crystal of thickness t can be performed by using the two-beam approximation in the dynamical theory of ED, or more accurately by using the many-beam theory for thicker crystals, with the inclusion of absorption effects.

III. High-Resolution Transmission Electron Microscopy One of the most powerful methods of direct structural analysis of solids is provided by HRTEM, whereby two or more Bragg reflections are used for imaging. Following Menter’s first images of crystal lattice periodicity (26) and

182

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

the early theoretical work by Cowley and Moodie (27), the power of experimental HRTEM in the experimental determination of real-space structures of complex inorganic solids that were not amenable to conventional techniques of structure determination (e.g., X-ray and neutron methods) was elucidated by Cowley and Iijima (28), Anderson (29), and Thomas (30). In contrast to conventional diffraction techniques, HRTEM provides localized real-space information—at the atomic level—concerning the bulk and surface properties of solids, as well as the corresponding chemical information and the ED information in reciprocal space. Because atomic scattering amplitudes for electrons are approximately 104 –105 times as large as they are for X-rays and neutrons, it follows that, with electrons as probes, structural information may be obtained from single crystals of almost nanoscale dimensions. To illustrate this point, we note that the best attainable X-ray performance (with synchrotron sources) requires crystal dimensions of 2 £ 2 £ 2 mm3. Because of the strong interactions between the electron beam and atoms in a sample, only some 104 unit cells of sample (corresponding to masses of sample as little as 10218 g) are required to yield significant HRTEM images and diffraction patterns. In HRTEM, very thin samples can be treated as weak-phase objects (WPOs) whereby the image intensity can be correlated with the projected electrostatic potential of crystals, leading to atomic structural information. Furthermore, the detection of electron-stimulated XRE in the electron microscope (energy dispersive X-ray spectroscopy, or EDX, discussed in the following sections) permits simultaneous determination of chemical compositions of catalysts to the sub-nanometer level. Both the surface and bulk structures of catalysts can be investigated. The micrograph or the image obtained on an EM screen, photographic film, or (more commonly today) a CCD is the result of two processes: the interaction of the incident electron wave function with the crystal potential and the interaction of this resulting wave function with the EM parameters which incorporate lens aberrations. In the wave theory of electrons, during the propagation of electrons through the sample, the incident wave function is modulated by its interaction with the sample, and the structural information is transferred to the wave function, which is then further modified by the transfer function of the EM.

III.A. Conditions Required for Optimizing HRTEM Images The HRTEM requires samples that are electron transparent (normally, a few tens of nanometers in thickness). As described in preceding section, during the interaction of the electron beam with a crystal specimen, electrons are

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

183

scattered by the interactions with the inner potential of the crystal. The objective lens of a microscope serves as a kind of Fourier transformer. The diffraction pattern formed at the bfp of the objective lens is further Fourier transformed to yield the image of the specimen. The theory of HRTEM tells us that, because the objective lens is imperfect (being characterized by size, aberration (spherical and chromatic), and defocus effects, some fundamental information about the specimen structure is lost. Electron-sample interactions result in phase and amplitude changes in the electron wave. The contrast of images in HRTEM (for example, in atomic-scale imaging) is a result of phase contrast caused by phase shifts (changes) of diffracted electron beams by the scattering, in combination with the objective lens effects. Amplitude changes are small. For a thin enough crystal, the WPO approximation is used, which is based on the assumption that the electron wave is modulated only in phase (phase contrast) and not in amplitude. The image intensity is then linearly related to the projected potential distribution of the sample (similar to the charge density) along the direction of incidence of the electron beam and can be expressed in terms of the crystal structure. The phase contrast is produced by the phase modulation of the incident electron wave when it is transmitted through the crystal potential of the sample. The propagation of a plane electron wave traversing a thin sample is thus treated as a weak (scattering) phase object. The wave function Cðx; yÞ at the exit face of a thin sample can be written as follows:

Cðx; yÞ ¼ expðisVðx; yÞÞ

ð3Þ

and for a very thin crystal, Eq. (3) can be approximated as

Cðx; yÞ ¼ 1 þ isVðx; yÞ

ð4Þ

where Vðx; yÞ is the thickness projected crystal potential and s is the interaction constant, which is a function of the electron wavelength and energy (31). The image intensity, Iðx; yÞ at the image plane of the objective lens results from two-dimensional Fourier synthesis of the diffracted beams (square of the FT of the waves at the exit face of the crystal), modified by a phase contrast transfer function factor (CTF or sin x which is dependent on the objective lens parameters and incident electrons. These are given by Scherzer (32) in Eqs. (5) and (6) as follows: Iðx; yÞ , 1 2 2sVðx; yÞ p FTðsin xÞ

ð5Þ

Where p is a convolution integral and FT is the Fourier transform. The phasecontrast imaging performance of an HRTEM is controlled by sin x; which contains the basic phase-contrast sinusoidal terms modified by an envelope

184

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

Fig. 3. HRTEM atomic structure image of germanium silicalite (GeSiO4) in which there are channels of aperture diameter 0.55 nm running along the [010] direction. Inset shows the 5- and 6-membered smaller apertures that are circumjacent to larger (0.55 nm) channels (5).

function, FðuÞ; which is due to the partial coherence of the electron beam: sin x ¼ FðuÞsin½ð2p=lÞðDf u 2 =2 2 Cs u 4 =4Þ

ð6Þ

where u is radial scattering angle, Df is objective lens defocus value, and F depends on the coherence conditions of the incident beam. CTF is a quantitative measure of the trustworthiness of the lens in recording a reliable image. Directly interpretable structure images are recorded near the Scherzer defocus, defined as 1=2 Df ðSÞ ¼ 2Cs l1=2 : At this defocus, the image can be directly related to the two-dimensional projected potential of the specimen, with dark regions corresponding to columns of heavier atoms. This is illustrated for a Ge-silicalite (GeSiO4) in Fig. 3. Beyond the point resolution, calculations to match experiments are required. In the following section we discuss the progress in HRTEM instrumentation.

III.B. Development of HRTEM To improve the point resolution, a number of important home-built instruments operated at higher voltages (, 500– 600 keV) were developed during the 1970s. However, these were in-house, highly specialized instruments that experienced some difficulties in operation. (Some were built at substantial cost and had difficulty meeting the theoretical resolution limit specifications, and some lacked a proper goniometer stage for tilting the samples.) The breakthrough in wide applications of HRTEM came with the development of the first state-of-the-art medium-voltage (200 kV) HRTEM by

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

185

Boyes, Gai and coworkers at the University of Oxford (in association with JEOL Ltd) (33) and by Thomas and coworkers at the University of Cambridge (also in association with JEOL Ltd) (34,35). The key points of these developments were that the instrument had a resolution similar to that of the ˚ ) at a small fraction of the cost, and best home-built HRTEM instruments (, 2 A it came in a user-friendly package, achieving the full theoretical performance routinely while fitting in a standard laboratory and requiring no special ˚ ) were buildings. Incremental improvements in resolution (, 1.3 – 1.6 A achieved later with the development of a 400-kV HRTEM (36). The state-of-the-art HRTEM has achieved very high resolutions of , 1.7 – ˚ and , 1.3 –1.6 A ˚ at operating voltages of 200 and 400 keV, respectively, 2.3 A providing information at the atomic level. New high-voltage (1 MeV) and highresolution commercial instruments have also been built, and a point resolution ˚ has been reported (37). Aberration-corrected commercial HRTEM of , 1 A instruments are becoming available (38). On-line digital processing techniques are also available to quantify HRTEM images. Quantification of the HRTEM image interpretation is checked by matching experimental images with complementary multi-slice image simulations using the n-beam dynamical theory of ED (27,39). Variations in image detail can be computed as a function of sample thickness, electron wavelength, and lens characteristics (spherical and chromatic aberrations and focusing conditions) (3,4,40 – 42).

III.C. Elucidation of the Structures of Meso- and Microporous Catalysts by HRTEM As described in the preceding section, there are fundamental and practical difficulties that require great caution in the interpretation of HRTEM images. The electron beam-sample interactions lead to multiple scattering (dynamical) effects that are quite complex but can be simulated. These are especially important in understanding the structures and shapes of nanoscale catalysts on supports (40). Furthermore, the image information is limited by electron lens aberrations. Efforts are in progress to minimize or eliminate corruption of the image by spherical aberration and chromatic aberration by aberration-free EMs and energy filtering; these are described in Sections XI and XIII.

III.C.1. L-Type Zeolite Catalysts A convenient approach in HRTEM is to record a series of images at different settings of the objective lens defocus and as a function of sample thickness.

186

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

Fig. 4. (a) HRTEM image of zeolite LTL along the [001] direcction. On the extreme (top) left is a schematic drawing of the framework of the idealized LTL structure. Next to it is a computed image, which is almost indistinguishable from the observed HREM image. This comparison demonstrates that the extent of structural distortion at the surface and immediate sub-surface region of the zeolite is less than about 5%. The cancrinite cages (Fig. 2(b)) are clearly visible at the outermost surface (side wall) (5). (b) Schematic diagram of a cancrinite cage, which is a major structural component of zeolite LTL.

A trustworthy result can be obtained by comparing the observed image to a simulated image, as illustrated in Fig. 4a and b for an L-type (LTL) zeolite catalyst (5). HRTEM shows that the structure remains unchanged from the surface (outermost layer) to the bulk. The simulated image (inset) of the surface of the LTL zeolite—terminating with cancrinite cages which are major structural components (Fig. 4b)—matches the observed HRTEM image. There is little evidence of contraction normal to the catalyst surface. III.C.2. Metal-Substituted Aluminum Phosphate (MAPO-36) Microporous Catalysts Microporous catalysts such as MAPO-36 (43,44), which are excellent for selective oxidation of hydrocarbons (45), are highly beam-sensitive. Yet HRTEM

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

187

Fig. 5. HRTEM image of MAPO-36 showing well-defined large apertures. The inset shows the computed image where the outline of the 12-rings is clearly visible (top and bottom are with and without taking beam damage into account) (5,43).

yielded images that even show high-symmetry crystallographic directions that unmistakably reveal (Fig. 5, with the computed image inset) 12-ring channel systems (similar to those in zeolite LTL (and also ALPO-5), the structure of which was solved by X-ray diffraction (44)). The results are consistent with results of gas adsorption measurements. The crystal symmetry and approximate values of the unit dimensions of the MAPO-36 catalyst were determined by HRTEM and ED patterns (43). These data provided a plausible structural model; the resulting simulated XRD pattern closely resembles the experimental pattern measured at high temperatures. The structure was then refined by the use of distance least-squared and energy minimization techniques, and excellent agreement was obtained between the experimental and simulated XRD patterns at both high and low temperatures (44). III.C.3. High-Silica Microporous SSZ-48 Catalysts ED intensity data collected by using a HRTEM and CCD detector reveal a monoclinic crystal structure having the following unit-cell dimensions:

188

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

Fig. 6. Structural model of the SSZ-48 crystal structure, showing the projected positions of the organic template within the pores, of SSZ-48 (5).

b ¼ 4:99 A; c ¼ 13:65 A; and b ¼ 100:78 ðV ¼ 748:6 A 3 Þ: a ¼ 11:19 A; Reflections with normalized structure factors between 0.65 and 10 were used in the structural solution by the direct methods (5). The phases obtained were used to generate a three-dimensional potential map that readily revealed the seven tetrahedrally coordinated silicon atoms in the asymmetric unit and five of the 14 oxygen atoms. The resulting structure is shown in Fig. 6. III.C.4. Intergrowths in Zeolite Catalysts: Coherent, Recurrent, and Random One of the earliest direct bonuses of imaging zeolitic catalysts by HRTEM was the discovery (10) that the nominally phase-pure ZSM-5 (structure code MFI) contained sub-unit-cell coherent intergrowths of ZSM-11 (MEL). It soon became apparent (46) that, depending on the mode of synthesis of these and other pentasil (zeolitic) catalysts, some specimens of ZSM-5 contained recurrent (regular) intergrowths of ZSM-11. It also emerged that intergrowths of offretite and erionite are features of both nominally phase-pure erionite and of pure offretite and of many members of the so-called ABC-6 family of zeolites (47).

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

189

All this served as a prelude to the clarifying work (48– 50) that showed that faujasite (FAU) and its hexagonal analogue (EMT) (Fig. 7) exhibit a strong tendency to form coherent intergrowths (5). And so, by HREM direct imaging, many hitherto puzzling problems concerning the structure of zeolitic catalysts were unambiguously resolved. For example, some zeolites claimed to be new on the basis of powder X-ray diffractograms (and usually published in the patent literature) turned out not to be new structures but rather intergrowths (of various kinds) of FAU (cubic) and EMT (hexagonal), as revealed by HRTEM (Fig. 8).

Fig. 7. Diagram illustrating the building units and structural relationship between the FAU and EMT frameworks. (a) Two (111) layers type K (A) and L (B) in twin orientation. (b) Hexagonal [100], (or cubic [1 2 1 0]) views of A and B. (c) Cubic FAU framework occurs when only type A or B stack. (d) Hexagonal EMT framework which occurs when A and B stack in alternation.

190

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

Fig. 8. HRTEM image of FAU/EMT intergrowths viewed along the [110] direction. The stackings ABC… and AB… correspond to the FAU and EMT end-member structures, respectively (5).

The stacking is shown in Fig. 9. This is the situation pertaining to ZSM-3, ZSM20, and ECR-30, for example. High-resolution EM also showed that the synthetic zeolitic catalyst ZSM-23 (MTT) is a recurrently twinned version of the synthetic zeolite theta-1 (TON) (51). It is noteworthy that the elucidation of the structures of zeolite beta, for a long time an enigma and problematic for X-ray crystallographers, came only through the application of HRTEM (50).

Fig. 9. Schematic of FAU and EMT intergrowth structures.

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

191

IV. Chemical Composition Analysis with the Analytical Electron Microscope Chemists who study solids are aware of the fact that microstructures of solids profoundly influence and control their properties. AEM at high resolution permits both the analysis of the elemental composition of a solid and its structure under high-resolution conditions. HRTEM (with high spatial resolution microdiffraction) provides high-resolution structure images, including structural defects such as dislocations or internal boundaries, in parallel with direct experimental measurements of local chemical composition from small areas—especially for heterogeneous solids (52– 56). Microcomposition analysis in the EM using electron-stimulated characteristic X-rays is a well-known technique, and EFTEM serves a very similar purpose. EDX, in which X-ray intensities are measured as a function of the X-ray energy, is the common method for chemical composition analysis in the electron microscope. In EDX, interaction of a beam of high-energy electrons with an inner-shell electron of the sample atom results in the ejection of a bound innershell electron from the attractive field of the nucleus in the sample atom, leaving the atom in an excited state with an electron shell vacancy. De-excitation by transition from an outer shell involving a change in the energy state of the atom between sharply defined levels produces X-rays (or Auger electrons), characteristic of elements in the sample. Stoichiometric variations in compositions of a material and of surface layers can be revealed by AEM. Because a relatively small amount of scattering occurs through a thin HRTEM specimen, X-rays are generated from a volume that is considerably less than in the case of electron microprobe analysis (EPMA). For quantitative microanalysis, a ratio method for thin crystals (57) is used, given by the equation: CA =CB ¼ KAB IA =IB

ð7Þ

where CA and CB are the concentrations of the elements A and B and IA and IB are the background-subtracted peak intensities for A and B, respectively; typically, a few dozen crystals are analyzed. The sensitivity factor KAB is determined by using appropriate standards. For bulk materials, more complex correction procedures are required and account is taken of the atomic number ðZÞ; absorption ðAÞ and X-ray fluorescence ðFÞ: Thus, AEM provides real-space imaging and crystallographic and microcompositional information on a very fine scale. Furthermore, AEM can be used to obtain partial occupancies of cation sites (and, under some conditions, anion sites). In cases for which elemental peaks overlap, wavelength dispersive X-ray spectroscopy (WDS) may be used to advantage.

192

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

Spatial mapping of the distribution of particular elements in catalysts typified by those listed in Table I is readily conducted by EM. Chemical variations of entire crystals in a sample can be obtained by analyzing X-ray intensities from elements across a line or over an area in the sample. The latter (two-dimensional scanning) is known as X-ray elemental mapping. Elemental maps recorded in an analytical HRTEM from MAPO catalysts (e.g., Zn – aluminum phosphate, Fig. 10) indicate a uniform distribution of the elements. Similarly, Fig. 11 shows an X-ray elemental map for GeSiO4 silicalite (Fig. 3), indicating a uniform distribution of Ge and Si. Quantification of intensities in X-ray maps can provide relative amounts of the elements (but care is required when peak-overlaps occur). Examples of elemental mapping of transition metal ion distributions in framework-substituted ALPO catalysts determined by EFTEM are described in Section X.

Fig. 10. X-ray elemental map in the electron microscope of metal-substituted aluminophosphate (MAPO-36 (with M ¼ Zn)) catalyst. The map shows a uniform distribution of the elements in the sample.

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

193

Fig. 11. X-ray elemental map in the electron microscope of GeSiO4 catalyst (shown in Fig. 3). The map illustrates a uniform distribution of Ge and Si.

V. Scanning Transmission Electron Microscopy Crewe et al. (58) pioneered STEM as a structural and an analytical tool. STEM, which is capable of acquiring signals that are difficult to obtain by other methods, is essentially a combination of SEM and TEM. In STEM, electrons are focused on a spot with a diameter less than 0.8 – 1 nm by a “probe-forming” lens (Fig. 12). The STEM detector collects scattered electrons and generates picture points by scanning the focused electron spot on the sample via a pair of deflection coils, and the resulting signal variation constitutes the image. Noteworthy is the excellent microanalytical capability of high-resolution STEM (HRSTEM)

Fig. 12. Schematic of the information from HRSTEM, DF, and high-angle annular dark-field (HAADF) microscopy.

194

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

(including modern TEM/STEM instruments) equipped with a field-emission gun (FEG-STEM), especially in the context of catalyst characterization. By use of sub-nanometer electron probes with high-electron currents, chemical analyses of catalysts (in addition to high-resolution imaging and element mapping) may be effected at the sub-attogram (10218 g) level. An important aspect of HRSTEM is Z-contrast (or atomic number) imaging. It exploits the fact that electrons scattered at high angles (. 30 mrad) obey Rutherford’s scattering law; the scattering cross-section is proportional to Z 2 ; where Z is the atomic number. Moreover, the scattered electron wave is predominantly incoherent, so that images formed by using a high-angle annular dark-field detector (HAADF) (or “Rutherford” detector) do not show the complicating contrast changes associated with coherent scattering, as occurs in BF images (formed from Bragg-scattered electrons). HAADF images are directly interpretable, and the technique is tailor-made for detecting clusters of catalytically active metals such as Pt, Pd, or Ru clusters (including bimetallics) on light supports such as zeolites. Isolated atoms or small cluster of heavy atoms (such as Pt) have been clearly identified by HRSTEM (59) as shown in Fig. 13. Since Crewe’s work, there have been significant advances in Z-contrast imaging, following suggestions by Howie and coworkers (60– 62). For example, Treacy et al. (61) and Pennycook et al. (62) imaged very small particles in catalysts by using high-angle Rutherford scattering contrast. Using the HAADF imaging technique in the STEM, low concentrations of dopants (, 1 at.%) in semiconductors and in zeolites have been demonstrated (63). Other spectroscopies

Fig. 13. Uniform bifunctional platinum-loaded zeolite catalyst. Large white dots (Pt) are ,0.5 nm in diameter.

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

195

are also possible in the STEM. In Section X, we illustrate recent advances in three-dimensional mapping of nanocatalysts using HAADF.

VI. Recent Advances in Ultra-High Resolution, Low-Voltage Field Emission Scanning Electron Microscopy and Extreme FESEM in Catalysis A new ultra-high-resolution low-voltage field emission scanning electron microscopy (HR-LV FESEM) instrument with a 0.5-nm probe at 30 kV (and 2.5 nm at 1 kV), integrated with high-sensitivity EDX, was designed by Boyes (64) to explore high-resolution imaging and chemical microanalysis in reflection from bulk samples. The instrument is equipped with an optimized high-brightness cold-FEG, combined with a very low-aberration condenser objective final lens. The low voltages allow investigations of uncoated, nonconducting samples (e.g., ceramic catalyst supports). Low-voltage electron probes (, 5 kV) generally yield inherently better SE images, making HRLVSEM a powerful tool in catalytic science. This advance is particularly important because it has made possible high-resolution surface analysis from bulk catalyst samples, and the resolution gap that previously existed between the SEM and many of the STEM applications has been bridged. Furthermore, a novel approach to FESEM design by Boyes (65,66) integrates new levels of low-voltage image resolution (, 1 nm at 1 keV) with greatly improved sensitivity for EDX elemental microanalysis; chemical imaging at new levels of spatial resolution down to , 100 nm; and, in favorable cases, resolution limits of 1– 10 nm, while retaining the advances of robust and representative bulk samples (mm in extent). These powerful capabilities are markedly improving our understanding of catalysts (4).

VII. Cathodoluminescence Imaging for Elucidation of Electronic Structures of Catalysts Cathodoluminescence imaging uses photons emitted from a sample area irradiated by a scanning electron beam for understanding point defect concentrations and promoter distributions in working catalysts (67). When an energetic electron scatters inelastically, an electron from the (filled) valence band can be promoted to the (empty) conduction band, creating an electron/hole pair. On recombination, the excess energy is released as a photon, the wavelength of which is well defined by the band-gap transition. The cathodoluminescence technique is powerful for determining the local electronic structures of catalysts.

196

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

It is diagnostic of electronic/chemical state, is sensitive to point defects, and can be used to probe the distribution of promoters in catalytic oxides (67). Examples include effects of the distribution of antimony in Sb – SnO2 catalysts (used for selective hydrocarbon oxidation) on the electronic structure of the catalyst and mapping of point defects in titania catalysts.

VIII. Recent Advances in In Situ Atomic ResolutionEnvironmental Transmission Electron Microscopy (ETEM) Under Controlled Reaction Conditions VIII.A. In Situ Investigations of Gas –Solid Reactions and Active Sites Catalysis is a dynamic process, and deeper insights into its phenomenology are extractable from in situ measurements than from characterizations of catalysts before and after catalysis. A number of notable in situ experiments have relied on modifications of standard TEM operations under vacuum. The main functions of the EM depend on a high-vacuum environment, and the pressure in a TEM is usually of the order of 1027 –1026 mbar. Because the influence of the reaction environment on the structure and activity of a catalyst is critical (3), the highvacuum environment of a conventional EM is inappropriate for investigating a catalytic reaction, as are characterizations of catalysts in post-reaction environments (e.g., when the catalyst has been taken out of the reaction environment and cooled to room temperature). With the gas reaction cell or an environmental cell (ECELL), controlled chemically reducing atmospheres and oxidizing atmospheres can be maintained in the EM, and a wide range of gases and vapors can be used. The development of the methods is described in the following sections. Early in situ ETEM experiments have been well documented by Hashimoto et al. (68), Swann and Tighe (69), and Butler and Hale (70). In the development of ECELLs, window cells have been used to contain gases, solvent vapors, and hydrated samples (71,72). However, these cells present problems in reliably sustaining a large pressure difference across a window that is thin enough to permit electron penetration. Generally, window cells are not suitable for heating systems. Below, we describe alternative methods used recently to investigate gas –solid and solution – solid reactions in the ETEM. The complications of windows can be avoided by substituting small apertures above and below the sample to restrict the diffusion of gas molecules while allowing penetration of the electron beam. Typically, pairs of apertures are added above and below the sample, with differential pumping lines attached between them. In the early in situ experimentation, an ECELL system (69) could be inserted inside the EM column vacuum between the objective lens pole pieces.

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

197

The gas reaction chamber and the objective aperture assembly occupied the gap between the upper and lower objective pole pieces, leading to a gas reservoir around the sample. Such ECELL systems were a major step forward in scientific capability, being used by Gai et al. (3,73– 78), Doole et al. (79), Crozier et al. (80), and Goringe et al. (81) to characterize catalysis. Other developments for catalytic studies include an ex situ reaction chamber attached externally to the column of a TEM, for example, by Parkinson and White (82) and Colloso-Davila et al. (83). Reactions were carried out in the ex situ chamber (and not in situ), and the sample was cooled to room temperature and inserted into the column of the TEM (without exposure to the atmosphere) under vacuum. Baker et al. (84) used ETEM at gas pressures of a few mbar with limited resolution, and, in these experiments, representative higher gas pressures were not employed. Gai (73) developed in situ high-voltage EM (HVEM) to meet the demands of realistically high gas pressures and temperatures (up to 1273 K) for catalysis, performing the first in situ investigations of selective hydrocarbon oxidation reactions catalyzed by metal oxides at high pressures (, 1 bar) and operating temperatures. The results provided insights into the fundamental role of defects at the catalyst surface in selective oxidation catalysis. With this system, image resolution was improved from , 1– 2 nm at . 100 mbar to 0.5 nm at lower gas pressures of , 30 mbar. This in situ HVEM development thus laid the foundation for the development of in situ atomic-resolution ETEM (85– 90). The quest to probe gas molecule – solid catalyst reactions directly at the atomic level resulted in the pioneering development of in situ atomic resolution-ETEM by Gai and Boyes (87 –90), who demonstrated that catalysis at atomic resolution was possible under controlled dynamic reaction conditions of gas pressure of a few mbar and elevated temperatures (91,92). In this development, a new approach was taken to design the ETEM instrument, which is dedicated to ECELL operations; the ECELL is permanently mounted and integrated with the HRTEM. The design is based on a computer-controlled Philips CM30T TEM/ STEM system with a proven high-resolution (crystal lattice imaging) performance. Furthermore, the whole EM column, and not just the region around the sample, was redesigned for the ECELL functionality, and a custom set of objective lens pole pieces incorporating radial holes was designed for the first stage of differential pumping (with no deleterious effect on imaging). In the atomic-resolution ETEM, the alignment and excellent atomic resolution (0.2 nm) of the microscope were maintained with the ECELL facilities, even with sample temperatures exceeding 973 K and small amounts of gas (at mbar pressures) flowing through the ECELL. The relatively large apertures in the cell provide useful angles of diffraction and allow some convergent beam diffraction pattern (CBDP) analysis with a dynamic STEM probe. The regular, smaller objective apertures can be used inside the ECELL for diffraction contrast experiments to determine the nature of defects.

198

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

In the development of Gai and Boyes (87,88,90), the ECELL, atomicresolution (HRTEM), STEM, hot stage and PEELS/Gatan imaging filter (GIF) functionalities were combined in a single instrument. The combination is required to aid simultaneous dynamic structure and composition of the reactor contents. ETEM is thus used as a “nanolaboratory” with multi-probe measurements. Design of novel reactions and nanosynthesis are possible. The structure and chemistry of dynamic catalysts are revealed by atomic imaging, ED, and chemical analysis (via PEELS/GIF), while the sample is immersed in controlled gas atmospheres at the operating temperature. The analysis of oxidation state in intermediate phases of the reaction and, in principle, EXELFS studies are possible. In many applications, the size and subsurface location of particles require the use of the dynamic STEM system (integrated with ETEM), with complementary methods for chemical and crystallographic analyses. The basic geometry of the novel atomic resolution-ETEM design is a fouraperture system, in pairs above and below the sample, but the apertures are now mounted inside the bores of the objective lens pole pieces (rather than between them, as in previous designs). Regular microscope apertures are mounted in bushes in each pole piece. The controlled-environment ECELL volume is the regular sample chamber of the microscope. Differential pumping between the apertures is achieved by using molecular drag pumps (MDP) and turbo-molecular pumps (TMP). This design permits high gas pressures in the ECELL sample region while maintaining high vacuum in the rest of the ETEM (Fig. 14). A conventional reactor-type gas manifold system enables the inflow of gases into the ECELL of ETEM, and a sample hot stage allows samples to be heated. A mass spectrometer is included for gas analysis. For dynamic atomic resolution, a few millibars of gas pressure are used in the ECELL. Higher gas pressures (up to a substantial fraction of a bar) are possible, but they compromise the resolution (as a consequence of multiple scattering effects of the electron beam through thicker gas layers). A video system connected to the ETEM facilitates digital image processing and real-time recording of dynamic events. The instrument and a schematic of the accessories are shown in Fig. 15a and b, respectively. In in situ ETEM experiments, very lowelectron dose techniques (with doses well below the threshold for structural damage) are used. The signal is amplified by a low light-level TV camera. The in situ data are checked in a parallel blank calibration experiment, with the beam switched off for this in situ reaction and the sample exposed to the beam only to record the reaction end point. In situ experiments are then confirmed by comparisons with data from calibration experiments. The aim is completely noninvasive characterization under benign conditions. Electronic image shift and drift compensation help to stabilize high-resolution images for data recording on film or with real-time digitally processed video. Under carefully simulated

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

199

Fig. 14. Schematic of the basic geometry of the aperture system and objective lens pole pieces incorporating radial holes for differential pumping system in the novel atomic resolution-ETEM design of Gai and Boyes (85 –90) to probe catalysis at the atomic level.

conditions close to those in practical reactors, data from in situ ETEM can be directly related to structure-activity relationships in technological processes. Because of the small amounts of solid reactant in the microscope sample, analyses of reaction products are performed with larger samples in a microreactor operating under similar conditions, and these are used for microstructural correlations. Several conditions must be met for successful ETEM investigations. Thin, electron-transparent samples are necessary—this requirement can usually be met with most catalyst powders. Ultrahigh-purity heater materials and sample grids capable of withstanding elevated temperature and gases are required (such as those made of stainless steel or molybdenum). The complex nature of catalysis with gas environments and elevated temperatures requires a stable design of the ETEM instrument to simulate realistic conditions at atomic resolution.

200

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

Fig. 15. (a) Novel atomic resolution-ETEM (87) and (b) schematic of various components for imaging, chemical analysis and diffraction under catalyst operating conditions.

The design of in situ atomic-resolution environmental cell TEM under controlled reaction conditions pioneered by Gai and Boyes (87,89) has been adopted by commercial TEM manufacturers, and latter versions of this in situ instrument have been installed in a number of laboratories. In situ atomic resolution-ETEM data demonstrated by Gai et al. (85– 90) have now been reproduced by researchers in laboratories using commercial instruments; examples include investigations of promoted ruthenium and copper catalysts in various gas environments (93) and detailed investigations of Ziegler – Natta catalysts (94).

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

201

Fig. 15. Continued

VIII.B. Illustrative Examples VIII.B.1. In Situ Gas – Catalyst Reactions at the Atomic Level Nanophase platinum catalysts supported on rutile TiO2 are of technological interest in environmental pollution control and methane reforming (95). Strong metal – support interactions of a reacting metal particle can lead to catalyst deactivation (96). Such phenomena can be examined in atomic-resolution ETEM. An ETEM investigation of sintering of Pt/TiO2 prepared by impregnation of TiO2 with a solution of a platinum salt, is shown in Fig. 16; Fig. 16a shows the catalyst containing finely dispersed platinum on TiO2; Fig. 16b shows in situ ETEM of dynamic catalyst activation in H2 at 573 K, and Fig. 16c shows the same particle of plainum (P) under dynamic conditions in H2 at , 723 K. The 0.23-nm atomic lattice spacings are clearly resolved in the platinum particle (P) in H2 at the elevated temperatures. The dynamic image (Fig. 16c) shows that the particle is faceted; SMSI deactivation with a growth of an amorphous titanium oxide monolayer on the particle is observed (indicated at the area marked

202

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

Fig. 16. Nanophase Pt/TiO2 catalysts: (a) finely dispersed Pt/TiO2 at room temperature. (b) In situ dynamic catalyst activation in hydrogen imaged at 3008C. The (111) lattice atomic spacings (0.23 nm) are clearly resolved in the platinum metal particle (P) under the controlled reaction conditions. (c) The same particle of platinum (P) imaged at 4508C, also in H2. Catalyst deactivation with growth of the support oxide monolayer indicated by a larger arrow, and the development of nm-scale single-crystal clusters of platinum metal (which show no coating as they emerge) with ,0.2-nm lattice spacings indicated by smaller arrow (87).

by a larger arrow), along with the development of nanometer-scale single-crystal clusters of platinum with , 0.2-nm lattice spacings, without overlayers (indicated by a smaller arrow in Fig. 16c). The H2 is a key contributor to this process. The results provide insights into the platinum dispersion, and the role of

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

203

temperature and particle size in the strong metal support interactions. A range of conditions and the dynamic rearrangement of the microstructure can be followed in each in situ experiment. In situ dynamic surface structural changes of catalyst particles in response to variations in gas environments were examined by ETEM by Gai et al. (78,97). In studies of copper catalysts on alumina, which are of interest for the water gas shift reaction, bulk diffusion of metal particles through the support in oxygen atmospheres was shown (78). The discovery of this new catalyst diffusion process required a radical revision of the understanding of regeneration processes in catalysis. Bimetallic (98) and alloy catalysts (97), of interest for hydrogenation reactions, have been investigated in in situ characterizations of methanol synthesis from CO and H2 in the presence of novel Cu –Pd alloy catalysts supported on carbon; the results show surface segregation of palladium on the catalyst particles in CO atmospheres, but surfaces with equal amounts of copper and palladium when the atmosphere is H2 (97). VIII.B.2. Atomic-Resolution ETEM of Butane Oxidation The selective oxidation of n-butane to give maleic anhydride (MA) catalyzed by vanadium phosphorus oxides is an important commercial process (99). MA is subsequently used in catalytic processes to make tetrahydrofurans and agricultural chemicals. The active phase in the selective butane oxidation catalyst is identified as vanadyl pyrophosphate, (VO)2P2O7, referred to as VPO. The threedimensional structure of orthorhombic VPO, consisting of vanadyl octahedra and phosphate tetrahedra, is shown in Fig. 17, with a ¼ 1:6594 nm, b ¼ 0:776 nm, and c ¼ 0:958 nm (100), with (010) as the active plane (99). Conventional crystallographic notations of round brackets ( ), and triangular point brackets k l, are used to denote a crystal plane and crystallographic directions in the VPO structure, respectively. The latter refers to symmetrically equivalent directions present in a crystal. In situ ETEM has met the formidable goal of revealing atomic structures of active sites; a mechanism for the release of catalyst structural oxygen; and the means for accommodation of anion deficiency in the butane oxidation catalyst (85,89). In situ ETEM and parallel chemical reactivity tests of calcined and activated VPO catalysts ((010) face), carried out with a continuous fixed-bed as well as with a pulse microreactor (101), were performed with the catalyst in butane, and alternatively in N2, or steam and 1.5% butane in air. Figures 18a and b show the (010) lattice image of the well-ordered VPO at room temperature and the corresponding ED, respectively. The structural model is superimposed, with dark regions corresponding to the heavier atoms. The ED shows some of the Bragg reflections. Figure 19a and b illustrates a sequence of in situ ETEM images

204

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

Fig. 17. Structure of complex (VO)2P2O7 in (010), viewed down the b-axis. Vanadium octahedra and phosphate tetrahedra link together forming a three-dimensional network. Front (bold) and back (faint) layers are shown.

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

205

Fig. 18. (a) Atomic structure image of VPO and (b) electron diffraction (ED) at room temperature.

with the catalyst in 20% butane in He (5 mbar) at room temperature and at the operating temperature of , 663 K, respectively. The dynamic surface structural development (a consequence of the catalyst anion loss) in butane with the formation of extended defects along the k201l direction is illustrated in Fig. 19b. The corresponding ED (inset, Fig. 19b) shows streaking along the k201l direction. The image in Fig. 19b is enlarged in Fig. 19c, showing a dislocated lattice with terminating lattice planes and the presence of partial dislocations (defects) in (201) lattice planes. The two partial dislocations, P1 and P2 (arrowed), are close to the catalyst surface (shown at S in profile, with the projection of the structure along the electron beam direction), bounding a stacking fault associated with them.

206

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

Fig. 19. (a) In situ atomic resolution ETEM image of (010) VPO in n-butane at room temperature with electron diffraction (ED); (201) lattice plane (0.63 nm) spacings and other lattice planes are resolved (201 reflection is arrowed). (b) In situ direct imaging of dynamic atomic motion of reacting VPO in n-butane at ,3908C. (c) Enlarged image of (b). The (201) lattice displacements (disturbing the periodicity) due to the reaction are close to the surface S. The resulting defects P1 and P2 are formed by novel glide shear and the lattice is not collapsed. The corresponding ED (inset) shows diffuse streaks along k201l (arrowed) (4).

The streaking in the ED provides important evidence of the structural disorder attributed to the defects in (201) planes. This means that anions in (201) planes, located between vanadyl octahedra and phosphate tetrahedral, are involved in the alkane oxidation reaction. The disorder attributed to the catalyst anion loss is revealed only in (201) lattice planes, thus excluding all other planes in the crystal structure. These findings, coupled with the results of detailed diffraction contrast experiments (85,89), show that the defects are formed by glide shear; the lattice is

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

207

not collapsed (Fig. 20). The sheared (transformed) structure creates regions of extended glide plane defects. The defect regions (at P1 and P2 in Fig. 19c) lead to structural regions akin to metaphosphate (PO3)n groups. The dynamic atomic studies show that only a few monolayers of the catalyst are involved in butane catalysis (89). The results showing disorder along the k201l direction illustrate that in the catalyst –adsorbate interaction, lattice oxygen loss leads to the formation of

Fig. 20. (a) Active sites observed by in situ atomic-resolution ETEM: structural modification of VPO in n-butane along k201l indicates the presence of in-plane anion vacancies (active sites in the butane oxidation) between vanadyl octahedra and phosphate tetrahedra. (b) Projection of (010) VPO (top) and generation of anion vacancies along k201l in n-butane. V and P are denoted. Bottom: model of novel glide shear mechanism for butane oxidation catalysis; the atom arrowed (e.g., front layer) moves to the vacant site leading to the structure shown at the bottom.

208

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

coplanar anion vacancies between vanadyl octahedra and phosphate tetrahedra (Fig. 21a). Extended defects are introduced along the k201l direction. They show that the release of structural oxygen in the oxidation catalysis is accompanied by a novel glide shear mechanism in which a few surface layers of the oxide undergo a structural transformation by glide shear to accommodate the surface misfit resulting from anion vacancies formed during the reaction (shown schematically in Fig. 21b). This mechanism explains the release of structural oxygen and the preservation of active Lewis acid sites at the surface without changes in the bulk structure of the catalyst. This mechanism is of fundamental importance in the understanding of solidstate heterogeneous catalytic oxidation processes. The glide defect regions are

Fig. 21. (a) The nature of the glide shear plane defects in three-dimensional projection and (b) in one layer of idealized structure, showing the novel glide shear process and the formation of glide shear plane defects. Filled circles are anion vacancies. (c) Schematic of glide shear. Glide defects accommodate the misfit at the interface between catalyst surface layers with anion vacancies (filled circles) and the underlying bulk (85,89).

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

209

not readily revealed in XRD because of overlap of the peaks from the defective regions and the VPO matrix; atomic-resolution ETEM has been crucial to unraveling the reaction mechanism. The positively charged anion vacancy sites preserved by glide shear at the catalyst surface can be readily available for alkane activation (by accepting electrons) and for exchange with gaseous oxygen. In partial oxidation in 1.5% butane/air, the alkane catalysis and the catalyst regeneration are possible, as validated by parallel reaction chemistry, shown in Table II (101). Pseudo-first-order rate constants ðkÞ for the disappearance of butane were measured with a microreactor and a larger amount of the catalyst (, 1 g) at 633 K. The constants are normalized to T ¼ 633 K assuming an activation energy of 25 kcal/mol and are shown in the second column of Table II. By varying the volumetric flow-rate of gas and constant times ðti Þ; k is obtained by fitting the reactor data to the classical first-order rate expression (101), dðbutaneÞ=dti ¼ 2kðbutaneÞ

ð8Þ

The conversion of butane is based on the difference in the moles between the feed and the products. Intrinsic rate constants, shown in the third column of Table II, are based on BET surface areas (m2/g) measurements (101). Samples 1 and 2 correspond to VPO treated in steam for 92 and 312 h, respectively. Samples 3 and 4 are N2-treated and activated base VPO catalysts, respectively. MA capacities represent the total amount of MA liberated by reduction in 1.5% butane/N2 at the reaction temperature. Table II shows that the base and N2-treated catalyst have nearly equal activities in the presence of air in the reactant stream and continue to operate.

TABLE II Continuous fixed-bed microreactor measurements Sample

k (rate) (s21)

k (intrinsic) (g/m2 s)

1

2.75

2

3.07

3

3.35

4

3.39 ^ 0.22

1.5% butane/air; % selectivity at

1.5% butane/N2; maleic capacity (micromol/g catalyst)

20% conversion

40% conversion

60% conversion

0.110

78

75

71

2.22

0.134

79

78

73

1.23

77

77

74

6.06

82

80

77

4.95 ^ 0.58

0.113

Samples 1 –4 correspond to VPO treated in steam for 92, 312 h, in N2 and activated base catalysts, respectively. k; are pseudo-first-order rate constants for the disappearance of butane. The constants are measured in a microreactor on a larger amount (,1 g) of catalyst at 633 K. k (intrinsic) are based on the BET surface area.

210

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

The novel glide shear mechanism revealed by ETEM and correlations with activity (89,101) show that glide shear is a key to effective butane oxidation catalysis. Investigations of the reduction of other oxide catalysts have also shown that the glide shear mechanism and temperature regimes where glide shear operates are beneficial for optimal catalyst performance (3,4). Catalysts can accommodate anion deficiencies without collapse of the crystal lattice and continue to operate, lengthening the catalyst life under optimized butane/air ratios. The work has led to the development of improved catalysts for the butane oxidation process, by incorporation of promoters to induce selective glide transformations (89,101). Earlier in situ EM investigations correlated with reaction chemistry (3,4,52,102 – 105) have shown that crystallographic shear plane defects produced by the well-known crystallographic shear mechanism, which eliminate super-saturation of anion vacancies (resulting from the reduction of oxides) by shear and lattice collapse, are secondary to catalysis. That is, crystallographic shear planes are consequences of oxide reactions and not the origins of catalytic activity (3).

VIII.B.3. Atomic-Resolution ETEM of Nanorods Nanowires and nanorods with high-aspect ratios have generated interest because of their potential applications in the next generation of nano and molecular electronics and in catalysis (106). They are being developed as potential supports for organic molecules (for applications in molecular electronics) and catalysts. Investigations of surface atomic structure by HRTEM and ED from single gold nanorods have provided the first direct evidence of the stabilization of the highly unstable (110) surface by surfactant molecules of cetyl trimethylammonium bromide (107). In situ heating experiments in an atomic-resolution ETEM in an atmosphere of N2 (Fig. 22) demonstrated that the rods are stable at elevated temperatures (18).

VIII.C. Advances in In Situ Wet-Electron Microscopy Technique (Wet-ETEM) for Probing Solid Catalysts Under Liquid Environments Many hydrogenation and polymerization reactions in the chemical industry are carried out with liquid-phase reactants. An example is the hydrogenation of aliphatic dinitriles to produce diamines (108,109), which are subsequently converted with adipic acid in solution and polymerized to produce linear polyamides, including nylon 6,6. Recently, the development of wet-environmental transmission electron microscopy (wet-ETEM) for direct nanoscale probing of

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

211

Fig. 22. ETEM at 1808C in N2, illustrating the stability of gold nanorods, for nanoelectronics and catalysis applications. Gold atomic layers and surface atomic structures are visible. Surface of gold nanorod at room temperature showing twin defect lamellae on the atomic scale. They indicate interaction of the surfactant with the (110) surface forming twins to accommodate the shape misfit between the two.

reactions between solid catalysts and reactants in the liquid phase—at reaction temperatures—has been reported (110). Using a liquid-feed holder with an injector system (similar to those used in chromatography), it is possible to inject pulses of the liquid into the ECELL under appropriate gaseous environments. The gas

212

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

manifold of the ETEM allows the flow of gases in the ECELL, and catalytic hydrogenation and polymerization reactions can be followed at operating temperatures. The wet-ETEM has been used to discover alternative, lowtemperature routes for the heterogeneous hydrogenation of liquid-phase adiponitrile using novel nanocatalysts consisting of Co –Ru on TiO2 followed by polymerization (110). The approach used in these experiments is different from that with window cells, which are generally not compatible with heating (71,72). The advances in characterization with liquid-phase reactants may lead to new opportunities for high-resolution imaging of a wide range of solution– solid and solution –gas – solid reactions in the chemical and biological sciences.

IX. Environmental Scanning Electron Microscopy Following early ETEM investigations using environmental cells, environmental scanning electron microscopy (ESEM) has been developed for characterization of surface effects of “bulk” SEM samples in the presence of gaseous or wet environments (111 –114). The method has been applied to the examination of food, wool fibers (111), and polymers (112) and in the conservation of cultural properties (113). Recently, fuel cell catalysts have been characterized using a low-voltage ESEM with a resolution capability of , 2 nm (114).

X. Electron Tomography: Three-Dimensional Electron Microscopy Imaging There is a growing need for ultra-sensitive methods for determining the size, elemental composition, precise location, spatial distribution, and detailed morphology of nanoparticles anchored to high-area supports. In catalysis and fuel cell technology, many different high-area (and generally low-atomicnumber) supports are employed, such as silica, alumina, and magnesia, as well as graphitic, amorphous, or adamantine carbons and thermally stable polymers. Furthermore, in many other areas of nanotechnology and biology, information about three-dimensional morphology and understanding of the spatial distribution and composition of nanoparticles are important. As shown above, the size and distribution of minute particles are conveniently investigated by high-resolution STEM with a HAADF detector (60,63). The intensity in HAADF images is a monotonic function of the sample thickness and atomic number, a pre-requisite for the electron tomography experiments described below.

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

213

Electron tomography has been used in biology (115) to investigate the threedimensional structure of macromolecules and cells. Recently, the approach has been applied to zeolites using conventional BF-TEM (116). Whereas conventional transmission electron microscopic images are essentially twodimensional projections of the object—the structural features are superimposed upon one another in the direction of the electron beam—in tomography, by contrast, one acquires projections of the object as viewed from different directions, and then one merges them computationally into a three-dimensional reconstruction, the tomogram. For electron tomography, a series of images must be recorded at successive tilt angles using a signal which must be a monotonic function of the projected thickness of the sample (115,117). A schematic diagram illustrating the acquisition of a series of tilted projections and the reconstruction of a three-dimensional object (a magnetite nanocrystal from a magnetotactic bacterium) is shown in Fig. 23 (116,117). The novel use of HAADF-STEM to determine the three-dimensional structure of a supported metal nanocatalyst at a spatial resolution of , 1 nm has been demonstrated for Pd –Ru nanocatalysts supported on mesoporous silica (117,118). The goal was achieved by tilting the sample to a series of different and finely spaced angles of two-dimensional projection. In the same way as was used with the established X-ray tomography methods, the information in the series of individual twodimensional images is analyzed to yield a detailed three-dimensional construction of the structure, with the full resolution of the process (in this case, 0.8 nm and potentially even higher resolution). The images obtained by the use of HAADF-STEM signal are directly interpretable.

X.A. The Topography and Location of Nanoparticles in Supported Catalysts; BSE and HAADF Many catalysts consist of heavy (high-Z) atoms such as platinum, palladium, ruthenium, or alloys (binary or ternary) and bimetallic variants of these elements, supported on low-Z, high-area solids such as carbon, alumina, silica, or magnesia. The metal particles are rendered readily visible by HAADF imaging, as described above, and when a series of two-dimensional images is recorded (117,118) at a succession of closely spaced tilt angles, tomographic information is retrieved. Moreover, by using back-scattered (Rutherford) imaging, as pioneered by Gai and Boyes (4), even more refined information may be gleaned about the spatial distribution and topography of such nanocatalysts. Back-scattered electrons (BSE), i.e., those scattered to angles greater than 908, also yield sharp images of nanoparticles containing . 100 atoms of high-Z materials distributed over low-Z supports, again because they obey Rutherford’s scattering law. BSE scattering may be thought of as reverse Rutherford

214

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

Fig. 23. Schematic diagram illustrating the acquisition of a series of titled projections and reconstruction of the three-dimensional object (118,119).

scattering, although the exact form of the experimental BSE scattering is modified by the high (. 908) scattering angle and by the bulk specimen environment. High-angle scattered electrons recorded using a STEM equipped with a HAADF detector and an SEM equipped with a BSE detector (Fig. 24) offer an essentially incoherent signal, and images are monotonically dependent on the atomic number of the sample and its thickness. Typical examples of Rutherford-scattered imaging of nanoparticles of a commercially important Pd/C catalyst recorded with (a) a BSE detector in a field emission scanning electron microscope as well as (b) a STEM HAADF image of the same 5% Pd/C sample, recorded in the same instrument, are shown

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

215

Fig. 24. Schematic of BSE and HAADF detector geometry.

in Fig. 25a and b, respectively (119). The strong Z-dependence is apparent in the images, which show enhanced contrast from the nanoparticles. It is clear that high-spatial resolution (, 1 nm) is achievable in the FESEM, here operated at 30 kV, and similar images are obtained in either BSE or forward scattered (HAADF) mode. It is a simple matter to identify small particles in thin sections by EDX methods. With a bulk (electron opaque) sample, the sensitivity of the BSE method in the nanometer range (and of EDX on the sub-micron scale) increases at medium to low voltages (with some limit set by instrumental parameters). The mixing of the SEM-BSE signal, primarily for higher-Z particle imaging—with a component of SE imaging, for lower Z support topography— together with the use of medium to low beam energies, may prove to be the optimum combination in the SEM (65,109). We now illustrate the HAADF images of the Pd/C catalyst. Figure 26a shows a single image from a series of successive tilt angles from þ 60 to 2 548 (119). Figure 26b shows the images of the same sample where each image represents the projection of the reconstructed three-dimensional structure (119,120). In these images, the reconstruction was obtained using a back-projection approach, shown schematically in Fig. 23. The data of Fig. 26 demonstrate the power of the technique for monitoring changes in the three-dimensional distribution of supported nanocatalysts. The examples shown in the preceding paragraphs illustrate that combined use of HAADF imaging and BSE imaging, both using Rutherford-scattered electrons,

216

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

Fig. 25. (a) SEM-BSE image and (b) STEM-HAADF image of palladium nanocatalysts on a carbon support (119).

is powerful in recording images of nanoparticle catalysts supported on irregular and thick carbonaceous supports. The incoherent scattering process ensures that images are ideal for electron tomography and the reconstruction of threedimensional nanoparticle distributions (119). These studies show the considerable potential of the method in the analysis of nonuniform catalysts and similar nanostructured systems. The images also illustrate that the HAADF and BSE approach (in which images are directly interpretable) may be superior to conventional BF-TEM and BF-STEM methods for catalysts, because of reduced exposure of the samples to the electron beam. In conventional TEM, for example, the large beam currents used can quickly damage the sample. BSE imaging can also be simple and effective in the study of surface-loaded nanocatalysts on bulk supports (employed in many industrial reactions), compared to conventional TEM or STEM analysis, which requires electron-transparent samples.

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

217

Fig. 26. (a) STEM-HAADF image acquired from the Pd/C sample shown in Fig. 25. (b) Animation of the three-dimensional reconstruction of the object in (a) (119).

218

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

X.B. Pinpointing the Location of Nanoparticles Supported on Nanoporous Solids An exciting area of modern heterogeneous catalysis involves the production of highly dispersed bimetallic nanoparticles (such as Cu4Ru12, Pd6Ru6, Ru10Pt2, and Ru6Sn) distributed over the interior surfaces of mesoporous silica (the pore diameter of which may be determined in the range of 2 –20 nm by the preparation conditions). Such highly dispersed nanoparticles function effectively as catalysts for a variety of solvent-free reactions, especially the hydrogenation of organic molecules (121,122). Provided due care is taken in their preparation, ˚ diameter) may be anchored to the inner individual nanoparticles (of 10 –15 A walls of the porous silica (Fig. 27). Figure 27a shows an HRTEM image of a hexagonal array of nanopores in silica and Fig. 27b shows a schematic of the interior of the single pore of silica. Evidence that the individual nanoparticles are situated as depicted in Fig. 27d emerges from images such as the HAADF image of Fig. 27c, which, as described in the preceding section, is an example of Z-contrast imaging whereby elements of high-atomic number ðZÞ show up readily against a background of low-Z elements. Indeed, because of a Z 2 dependence on electron scattering cross-section of elements (described in Section V), one Pt atom scatters as strongly as about 100 oxygen atoms or 32 silicon atoms (in conformity to the Rutherford scattering law). Images such as that of Fig. 27c, coupled with electron tomography (123), show that the nanoparticles are indeed anchored to the walls of the pores, and with the pore diameter being so large there is ample room for reactant and product molecules to diffuse in the pores.

XI. Energy Filtered Transmission Electron Microscopy and Elemental Maps of Solid Catalysts Using EFTEM Recent advances in elemental mapping of solid catalysts have been accomplished by the use of EFTEM (124), as exemplified by the distribution of transition metal ions in framework-substituted aluminophosphate, which are good shape-selective and regio-selective oxidation catalysts (43,44,121). With up to about 4 at.% of the Al3þ ions isomorphously replaced by either Co3þ or Mn3þ, giving oxyfunctionalization catalysts for alkanes (122), it is important to know how uniformly these ions (the active sites) are distributed. This is rapidly done by using a solid-state detector to record the electron-stimulated XRE spectra characteristic of the ion, as shown in the example of Fig. 10. Energy-filtered (EF) EM in various modes yields the element distribution maps for light as well as heavy elements (124). Even mixed-valence states in catalytic solids may be charted by electron-filtered EM

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

219

Fig. 27. (A) HRTEM micrograph of a typical hexagonal array of nanopores in silica (diameter 10 nm). (B) Computer graphic representation of the interior of a single pore of the silica showing pendant silanol groups. (C) HAADF (see text) showing the distribution of anchored Ru6Sn nanoparticles within the nanopores of the siliceous host. (D) Computer graphic illustration of the Ru6Sn nanoparticles superimposed on an enlargement of the electron micrograph shown in (C). (After Ref. (122b)).

(125). In the case of silica-encapsulated bimetallic catalysts, one can establish from precisely coincident element maps taken with Ru K-emission and Pd Kemission X-rays that the individual nanoparticles retain their structural integrity and are indeed nanoparticles such as Ru6Pd6 (or Ru12Cu4, (63)). EF images (for example, those obtained using oxygen K-loss peaks or nitrogen K-loss peaks, which are centered around 530 and 400 eV, respectively, or even plasmon-loss peaks) are also instructive in revealing the distributions of light elements in catalytic solids (14).

220

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

XII. Other Signiﬁcant Trends The electron crystallography method (21) has been used to characterize threedimensional structures of siliceous mesoporous catalyst materials, and the threedimensional structural solutions of MCM-48 (mentioned above) and of SBA-1, -6, and -16. The method gives a unique structural solution through the Fourier sum of the three-dimensional structure factors, both amplitude and phases, obtained from Fourier analysis of a set of HRTEM images. The topological nature of the siliceous walls that define the pore structure of MCM-48 is shown in Fig. 28.

XIII. Critical Evaluations of the Methods and Challenges The advanced EM methods described in this chapter are critical to the fundamental understanding of the nanostructure and chemistry of chemically and physically complex solid catalysts. These methods uniquely determine the nature, atomic structure and crystallography of defect structures (disorder) at catalyst surfaces in the reaction. These include whether defects result from vacancies or interstitials, the nature of point defects associated with surface Lewis or Bro¨nsted acidity or basicity, their diffusion in the catalytic reaction, growth of extended defects, and specific crystallographic planes and lattice displacements (Burgers vector) involved in these processes. The nature of defects is, therefore, of critical importance to the catalyst performance, in the hydrocarbon activation and catalyst regeneration processes. Bulk diffraction methods such as X-ray diffraction simply average data from larger areas, and scanning probe methods (for which chemical composition and diffraction information are difficult and deficient, respectively) require specialized sample preparations and are not

Fig. 28. Schematic illustration of the siliceous wall and channel structure of the mesoporous solid known as MCM-48 (based on the results given in Ref. (122b)).

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

221

readily applicable to commercial catalysts. In EM, careful experimentation is required along with understanding of the ED phenomena. We now address some of the challenges and opportunities in the methods described in the chapter. Conventional HRTEM operates at ambient temperature in high vacuum and directly images the local structure of a catalyst at the atomic level, in real space. In HRTEM, as-prepared catalyst powders can be used without additional sample preparation. The method does not normally require special treatment of thin catalyst samples. In HRTEM, very thin samples can be treated as WPOs, whereby the image intensity can be correlated with the projected electrostatic potential of the crystal, leading to the atomic structural information characterizing the sample. Furthermore, the detection of electron-stimulated XRE in the EM permits simultaneous determination of the chemical composition of the catalyst. Both the surface and sub-surface regions of catalysts can be investigated. However, care must be taken to use a very low-dose electron beam to avoid beam damage to the sample. This is especially important in molecular sieve and zeolite catalysts, which have extraordinary tendency to become amorphous under prolonged exposure to the electron beam. This limitation has been overcome by using high electron accelerating voltages in the EM (e.g., 200– 300 kV instead of 100 kV), to minimize the inelastic collisions that are primarily responsible for the structural degradation, along with better vacuum in the EM. For the new class of ALPO catalysts, high-resolution CCD, because of their ability to record digital images with very low-incident electron doses, are becoming increasingly common to image catalysts and avoid sample damage. In HRTEM of complex structures, image simulations are necessary to correlate an experimental image with theory. Calculations are especially needed for images from thicker samples, from the latest FEG HRTEMs and very high-voltage electron microscopes. Electron crystallography, incorporating HRTEM, ED, and computational methods are powerful in determining the three-dimensional structure of complex zeolites and molecular sieve structures which are not amenable to X-ray diffraction. The approach offers opportunities in identifying the fine structure of zeolite catalysts and metal promoters in particular positions in the catalysts. Challenges include the determination of the three-dimensional structures of point and extended defects on the surfaces of these materials during catalysis. In supported catalysts, particle visibility may be a challenge if the support thickness exceeds a certain value. This statement is applicable to both amorphous and crystalline supports. Particles can be viewed in plan view or in the surface profile mode. In the former, the contrast from nanoparticles can be obscured by the support contrast (40). Surface profile imaging can be employed for thicker industrial supported catalysts in which particles are visible only when they are near the edge of the support. Investigations can provide insights into the structure and shape of the nanoparticles even when the fraction of the particles near the edges of the support is small. Out-of-focus imaging and image processing

222

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

methods are also helpful in gleaning structural information from supported nanocatalysts. Calculations carried out by Gai et al. (40) show that in sphericalaberration ðCs Þ corrected (ideal) electron microscopes, the particle visibility is dramatically improved. Lens aberrations (imperfections) yield limited spatial and spectral resolution in EM. Sample thickness also affects the achievable resolution. HRTEM with selected-area ED is especially useful in providing insights into the disorder and ordering of anion vacancies in oxide catalysts in oxidation catalysis. To image oxygen atom columns in an oxide using conventional HRTEMs with Cs ; thin samples are oriented down the exact crystallographic zone axis, and the imaging requires appropriate defocus conditions. For example, imaging of oxygen atom columns in high-temperature cuprates has been demonstrated (126,127). Challenges for EM technology are, therefore, to achieve the development of spherical (and chromatic) aberration-free electron microscopes to improve the spatial and analytical resolution. Abberation-corrected HRTEM and STEM instruments have been reported (128,129). Recent work using Cs -corrected EM shows oxygen atom column imaging in perovskite ceramics (38). Thus, aberration-corrected EMs are becoming routinely available. The aforementioned development of in situ atomic-resolution environmental TEM (ETEM) as a multifunctional “nanolaboratory” has enabled the determination of the structure and chemistry of catalysts including active site configurations by atomic imaging, ED, and chemical analyses during catalysis. Low-electron beam currents (well below the threshold for sample damage) are employed, and the signal is amplified and recorded via a low-light level television camera and a video system. In addition, blank experiments are performed without the electron beam, and the beam is switched on for only a few seconds to record the final state of the material. The results are then compared with those of in situ experiments performed with very low electron doses to confirm the validity of the in situ experiments. Under these controlled experimental conditions, beam damage to the catalyst is not observed, and ETEM data can be directly related to structure – property relationships and reaction kinetics in technological processes. Time- and temperature-resolved experiments can be carried out. In situ ETEM thus helps to reduce the time and costs involved in scaling up laboratory experiments to industrial conditions. Because the method operates under dynamic catalyst operating conditions, caution should be exercised in maintaining the reaction environment, temperature regimes, and imaging. At present, atomic-resolution in ETEM is possible with a few mbar of gas pressures. Higher gas pressures (up to 1 bar) are possible, but the resolution is compromised at higher pressures because of the absorption of electrons by thicker gas layers. In gas –catalyst experiments, the coverage of the catalyst with the reactant-derived species is crucial, and this is more important than the presence of high gas pressures in the ECELL (gas reaction cell or microreactor), or around the sample.

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

223

Aberration-corrected ETEM/STEM (130) is expected to offer superior (subatomic) resolution under catalytic reaction conditions; furthermore, it will provide improved flexibility for tilting the sample to different crystallographic orientations to allow understanding of the geometry of surface structural changes, enable the use of complex sample stages, and perhaps higher gas pressures. STEM uses a very small probe scanned in a raster across the sample. The method provides many analytical signals, including HAADF and EELS, and offers several advantages over conventional TEM. In HAADF, highly incoherent high-angle scattering electrons are employed (Rutherford scattering), and the method is sensitive to the atomic number of the atoms (Z-contrast). The HAADF signal removes the complexity of conventional bright-field scattering in TEM and associated diffraction complications and allows the direct interpretation of results. Three-dimensional electron tomography using HAADF-STEM (Z-contrast) is powerful in determining the structures of supported nanocatalysts. The results are achieved by tilting the catalyst sample to a series of different and finely spaced angles, and the images are reconstructed. Current challenges of STEM include resolution; delivery of adequate current in the 0.2-nm probe in EDX chemical analysis at the atomic level; beam damage to the sample; and sample stability. Pulsing the electron beam onto the sample can be helpful in increasing the sample stability. Aberration-corrected STEM can be helpful in obtaining high probe currents for chemical analysis. In three-dimensional electron tomography, it may be challenging to obtain enough tilt for the sample and reconstruction of threedimensional images of nanoparticles on irregular (and thick) supports. Wide gap lenses with aberration correction may be able to provide adequate tilt range and resolution. Electron beam damage to the sample is a fundamental issue in STEM, and careful experimentation to ensure the stability of the sample is required. In low-voltage, high-resolution SEM (LVSEM) of catalysts, a spatial resolution 0.5 nm at 1 kV and more current in electron probes for high-precision microchemical analysis are being sought. Challenges in LVSEM of catalysts include control of the sample charging and preservation of sample stability. In ESEM, challenges and opportunities include improved resolution and microanalysis with better sensitivity and accuracy.

XIV. Conclusions Several general conclusions are drawn concerning the status of EM as a supremely versatile tool in the study of the materials chemistry of catalysts. First, it is no longer necessary to regard EM as a tool for model studies (131– 133). The triumphant exploitation of the environmental cell in HRTEM marks the dawn of a new era in probing dynamic catalysis (4,87– 95). Second, EM techniques, as has

224

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

recently been illustrated by Rupprechter (134), may be smoothly integrated with parallel investigations (e.g., of polycrystalline, nanoparticle platinum, palladium, and rhodium) by vibrational (sum frequency generation) spectroscopy and scanning tunnelling microscopy. Thus, for example, with alumina-supported rhodium nanoparticles, it was explicitly demonstrated that high-index faces (lowcoordinated sites) are preferred for hydrogenolysis catalysis. Extrapolating Rupprechter’s results and recognizing the vast new possibilities that are now possible (thanks to the arrival of intense near-IR femtosecond laser pulses) in time-resolved in situ measurements (135,136), one may reasonably expect further major advances in studies of polycrystalline rather than just single-crystal surfaces. Finally, electron crystallography (21) and electron tomography (117 –119) are important new developments in the study of catalysts.

Acknowledgements We thank our colleagues Osamu Terasaki, Edward Boyes, Paul Midgley, Robert Raja, Frank Gooding, Leland Hanna, Kostantinos Kourtakis, Gopinath Sankar, Matthew Weyland, and Brian Johnson for their friendly cooperation.

References 1. Thomas, J.M., and Lambert, R.M. (Eds.), “Characterization of Catalysts.” Wiley, Chichester, 1980. 2. Thomas, J.M., and Thomas, W.J., Principles and Practice of Heterogeneous Catalysis. Wiley-VCH, Weinheim, 1997. 3. Gai, P.L. (Gai-Boyes, P.L.), Catal. Rev.—Sci. Engng 34, 1 (1992). 4. Gai, P.L., and Boyes, E.D., “Electron Microscopy in Heterogeneous Catalysis.” Insitute of Physics Publishing, Bristol, UK, 2003. 5. Thomas, J.M., Terasaki, O., Gai, P.L., Zhou, W., and Gonzalez-Calbet, J., Acc. Chem. Res. 34, 583 (2001). 6. Thomas, J.M., Evans, E.L., and Bahl, O.P., Surf. Sci. 8, 473 (1967). 7. Thomas, J.M., Inorganic Chemistry: Towards the 21st Century, 211, p. 445. ACS Publication, 1983. 8. Thomas, J.M., Millward, G.R., and Bursill, L.A., Phil. Trans. R. Soc. A300, 43 (1981). 9. Terasaki, O., Millward, G.R., and Thomas, J.M., Proc. R. Soc. A395, 153 (1984). 10. Thomas, J.M., and Millward, G.R., J. Chem. Soc. Chem. Commun. 1380 (1982). 11. Stern, E.A., and Siegel, R.W., Curr. Opin. Solid State Mater. Sci. 4, 321 (1999). 12. (a) Redlich, P., Curr. Opin. Solid State Mater. Sci. 4, 325 (1999); (b) Thomas, J.M., Jefferson, D.A., and Egerton, R.F., Chem. Brit. 17, 514 (1981). 13. Sankar, G., and Thomas, J.M., Top. Catal. 8, 1 (1999). 14. Egerton, R.F., Top. Catal. 21, 185 (2002). 15. Thomas, J.M., and Sankar, G., Acc. Chem. Res. 34, 571 (2001). 16. Gai, P.L., and Thomas, J.M. (Eds.), Top. Catal. 21, 107 (2002). 17. Kirkland, A.J., and Sloan, J., Top. Catal. 21, 139 (2002).

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

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. 45. 46. 47. 48. 49. 50.

51. 52. 53. 54.

225

Gai, P.L., Top. Catal. 21, 161 (2002). Heinemann, H., and Poppa, H., Ultramicroscopy 17, 213 (1985). Ohnishi, N., and Hiraga, K., J. Electron Microsc. 45, 85 (1996). Carlsson, A., Kaneda, M., Sakamoto, Y., Terasaki, O., Ryoo, R., and Joo, S.H., J. Electron Microsc. 48, 795 (1999). Wagner, P., Terasaki, O., Ritsch, S., Nery, J.G., Zones, S., Davis, M.E., and Hiraga, K., J. Phys. Chem. B103, 8245 (1999). Hirsch, P.B., Howie, A., Nocholson, R.B., Pashley, D., and Whelan, M.J., “Electron Microscopy of Thin Crystals.” Kruger, Florida, 1977. Gai, P.L., and Howie, A., Phil. Mag. A30, 939 (1974). Gai, P.L., and McCarron, E., Science 247, 553 (1990). Menter, R., Proc. R. Soc. A236, 119 (1952). Cowley, J.M., and Moodie, A.F., Acta Cryst. 10, 609 (1957). Cowley, J.M., and Iijima, S., Z. Naturforsch. 27a, 445 (1972). Anderson, J.S., J. Chem. Soc. Dalton Trans. 1107 (1973). Thomas, J.M., Phil. Trans. R. Soc. A277, 251 (1974). Cowley, J.M., “Diffraction Physics.” North-Holland, Amsterdam, 1981. Scherzer, O., J. Appl. Phys. 20, 20 (1949). Boyes, E.D., Watanabe, E., Gai, P.L., Naruse, M., Jenkins, M.L., Hutchison, J.L., and Skarnulis, A., Inst. Phys. (IOP, London, UK), Conf. Ser. 52, 445 (1980). Thomas, J.M., Jefferson, D.A., and Millward, G.R., JEOL News 23E, 7 (1985). Jefferson, D.A., Thomas, J.M., Millward, G.R., Harrison, A., and Brydson, R.D., Nature 323, 428 (1986). Boyes, E.D., Hutchison, J., and Watanabe, E., JEOL News 24E (1986). Ruhle, M., Ultramicroscopy 56, 1 (1994). Jia, C.L., Lentzen, M., and Urban, K., Science 299, 870 (2003). O’Keefe, M.A., Acta. Cryst. A29, 389 (1973). Gai, P.L., Goringe, M.J., and Barry, J.C., J. Microsc. 142, 9 (1986). Datye, A., and Smith, D.J., Catal. Rev.—Sci. Engng 34 (1992). Yacaman, M.J., and Avalos-Borja, M., Catal. Rev.—Sci. Engng 34 (1992). Gai, P.L. (Gai-Boyes, P.L.), Thomas, J.M., Wright, P.A., Jones, R.H., Natarajan, S., Chen, J., and Xu, R., J. Phys. Chem. 96, 8206 (1992). Wright, P.A., Natarajan, S., Thomas, J.M., Gai-Boyes, P.L., Jones, R.H., and Chen, J., Angew. Chem. 31, 1472 (1992). Thomas, J.M., Raja, R., Sankar, G., and Bell, R.G., Acc. Chem. Res. 34, 191 (2001). Millward, G.R., J. Chem. Soc. Faraday Trans. II 79, 1075 (1983). Millward, G.R., Ramdas, S., and Thomas, J.M., Proc. R. Soc. A399, 57 (1985). Thomas, J.M., Audier, M., Klinowski, J., Bursill, L.A., and Ramdas, S., Faraday Discuss. 72, 345 (1981). Terasaki, O., Ohsuma, T., Alfredsson, V., Bovin, J., Watanabe, D., and Anderson, M.W., Chem. Mater. 5, 452 (1993). Vaughan, D., Treacy, M.M.J., and Newsam, J.M., in “Guidelines for Mastering the Properties of Molecular Sieves” (D. Barthomeu, Ed.), p. 96. Pleasure Press, New York, 1999. Thomas, J.M., Millward, G.R., White, D., and Ramdas, S., J. Chem. Soc. Chem. Commun. 434 (1988). Gai, P.L., Bart, J.C., and Boyes, E.D., Phil. Mag. A45, 531 (1982). Gai, P.L., and Boyes, E.D., IOP (London) Conf. Ser. 61, 215 (1981). Boyes, E.D., and Gai, P.L., in “Analytical Electron Microscopy” (R. Geiss, Ed.), p. 71. San Francisco Press, San Francisco, 1981.

226

55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100.

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

Thomas, J.M., Ultramicroscopy 8, 13 (1982). Boyes, E.D., Goringe, M.J., and Muggridge, B., J. Microsc. 127, 321 (1982). Cliff, I., and Lorimer, G., J. Microsc. 103, 203 (1975). Crewe, A.V., Issacson, M., and Johnson, D.E., Rev. Sci. Inst. 42, 411 (1971). Treacy, M.M.J., and Rice, S.B., Ultramicroscopy 7, 11 (1981). Howie, A., J. Microsc. 117, 11 (1979). Treacy, M.M.J., Howie, A., and Pennycook, S.J., Inst. Phys. Conf. Ser. 52, 261 (1980). Pennycook, S.J., J. Mol. 20, 345 (1983). Ozkaya, D., Zhou, W., Thomas, J.M., Midgley, P.A., Keast, V., and Hermans, S., Catal. Lett. 60, 113 (1999). (a) Boyes, E.D., Proc. ICEM-13 (Paris) 1, 51 (1994); (b) Boyes, E.D., Inst. Phys. (IOP, London, UK) Conf. Ser. 168, 115 (2001). Boyes, E.D., Microsc. Microanal. 6, 307 (2000). Boyes, E.D., Adv. Mater. 10, 1277 (1998). Boyes, E.D., Gai, P.L., and Warwick, C., Nature 313, 666 (1985). Hashimoto, H., Naiki, T., Eto, T., and Fujiwara, K., Jpn. J. Appl. Phys. 7, 946 (1968). Swann, P., and Tighe, N., Jernkont. Ann. 155, 251 (1971). Butler, P., and Hale, F., “Dynamic Experiments.” North-Holland, Amsterdam, 1981. Doube, D.D., Mater. Sci. Engng 12, 29 (1973). Parsons, D.F., Adv. Biol. Med. Phys. 15, 161 (1974). Gai, P.L., Phil. Mag. 43, 841 (1981). Gai, P.L., Kristall. Technik 14, 1385 (1979). Gai, P.L., Phil. Mag. 48, 359 (1983). Gai, P.L., J. Solid State Chem. 49, 25 (1983). Gai, P.L., and Labun, P.A., J. Catal. 94, 79 (1985). Gai, P.L., Smith, B.C., and Owen, G., Nature 348, 430 (1990). Doole, R.C., IOP Conf. Ser. 119, 161 (1991). Crozier, P., Sharma, A., and Datye, A., Microsc. Microanal. 4, 228 (1998). Goringe, M.J., Bithell, E., and Doole, R.C., Faraday Discuss. 105, 85 (1996). Parkinson, G.R., and White, D., Proc. XI Int. Congr. Electron Microsc. 331 (1986). Collozo-Davila, C., Marks, L., and Landree, E., Microsc. Microanal. 1, 219 (1995). Baker, R.T., and Chludzinski, J., Carbon 19, 75 (1981). Gai, P.L., and Kourtakis, K., Science 267, 661 (1995). Boyes, E.D., and Gai, P.L., Mater. Res. Soc. 404, 53 (1996). Boyes, E.D., and Gai, P.L., Ultramicroscopy 67, 219 (1997). Gai, P.L., and Boyes, E.D., in “In situ Microscopy in Materials Research” (P.L. Gai, Ed.). Kluwer, Boston, 1997. Gai, P.L., Acta Cryst. B53, 346 (1997). Gai, P.L., Adv. Mater. 10, 125 (1998). Haggin, J., C&E News (Am. Chem. Soc.) 73, 39 (1995). Jacoby, M., C&E News (Am. Chem. Soc.) 80, 26 (2002). Hansen, T.W., Wagner, J., Hansen, P.L., Dahl, S., Topsøe, H., and Jacobsen, C., Science 294, 1508 (2001). Oleshko, V., Crozier, P., Cantrell, R., and Westwood, A., J. Electron Microsc. 51, S27 (2002). Gai, P.L., Top. Catal. 8, 97 (1999). Tauster, S., Fung, S., and Garten, J.L., J. Am. Chem. Soc. 100, 170 (1978). Gai, P.L., and Smith, B.C., Ultramicroscopy 34, 17 (1990). Sinfelt, J.H., “Bimetallic Catalysts.” Wiley, New York, 1985. Centi, G., Catal. Today 16(1) (1993). Gorbunova, Yu., and Linde, S., Dokl. Acad. SSSR 245, 584 (1979).

J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227

227

101. Gai, P.L., Kourtakis, K., Coulson, R., and Sonnichsen, G.C., J. Phys. Chem. 101, 9916 (1997). 102. Gai, P.L., J. Solid State Chem. 104, 119 (1993). 103. Gai, P.L., Curr. Opin. Solid State Mater. Sci. 4, 63 (1999). 104. Gai, P.L., in “Catalyst Materials for High Temperature Processes” (K. Ramesh, M. Misono and P.L. Gai, Eds.), p. 127. American Ceramic Society, Westerville, OH, USA, 1997. 105. Gai, P.L., Curr. Opin. Solid State Mater. Sci. 5, 371 (2001). 106. Service, R.F., Science 294, 2442 (2001). 107. Gai, P.L., and Harmer, M.A., Nano Lett. 2, 771 (2002). 108. De Bellefon, C., and Foilloux, P., Catal. Rev.—Sci. Engng 36, 459 (1994). 109. Gai, P.L., Kourtakis, K., and Ziemecki, S., Microsc. Microanal. 6, 335 (2000). 110. Gai, P.L., Microsc. Microanal. 8, 21 (2002). 111. Danilotos, G., in “In situ Microscopy in Materials Research” (P.L. Gai, Ed.), p. 13. Kluwer, Boston, 1997. 112. Donald, A., Proc. EUREM 12, 1 (2000). 113. Doehne, E., in “In situ Microscopy in Materials Research” (P.L. Gai, Ed.), p. 45. Kluwer, Boston, 1997. 114. Boyes, E.D., “Frontiers in Electron Microscopy Conference,” Berkeley, CA, USA, 2003. 115. Crowther, R.A., de Rossier, D., and Klug, A., Proc. R. Soc. A317, 319 (1970). 116. Koster, A.J., and de Jong, K.P., J. Phys. Chem. B104, 9368 (2000). 117. Midgley, P.A., Weyland, M., Thomas, J.M., and Johnson, B.F.G., Chem. Commun. 907 (2001). 118. Weyland, M., Midgley, P.A., and Thomas, J.M., J. Phys. Chem. B 105, 7882 (2001). 119. Midgley, P.A., Weyland, M., Thomas, J.M., Gai, P.L., and Boyes, E.D., Angew. Chem. Int. Ed. 41, 3804 (2002). 120. For further details an animated description of electron tomographic work on threedimensional nanostructure analysis based on Rutherford scattering (117– 119) see http:// www.hrem.msm.cam.uk/~mw259/Work/Tomo.html. 121. Raja, R., Angew. Chem Int. Ed. 40, 4638 (2001). 122. (a) Thomas, J.M., Raja, R., Sankar, G., Johnson, B.F.G., and Lewis, D.W., Chem. Eur. J. 7, 2973 (2001); (b) Thomas, J.M., and Terasaki, O., Top. Catal. 21, 155 (2002). 123. Weyland, M., Top. Catal. 22, 175 (2002). 124. Thomas, P., and Midgley, P.A., Top. Catal. 22, 107 (2002). 125. Wang, Z., Bentley, J., and Evans, N., J. Phys. Chem. B 103, 751 (1999). 126. Gai, P.L., Subramanian, M.A., Gopalakrishnan, J., and Boyes, E.D., Physica C 159, 801 (1989). 127. Gai, P.L., and Thomas, J.M., Supercond. Rev. 1, 1 (1991). 128. Haider, M., Kabius, K., Uhelmann, S., Urban, K., and Rose, H., Microsc. Microanal. 7, 902 (2001). 129. Batson, P., and Krivanek, O., Nature 423, 211 (2003). 130. Gai, P.L., and Boyes, E.D., “Frontiers in Electron Microscopy Conference,” Berkeley, CA, USA, 2003. 131. Freund, H.-J., Baumer, M., and Kuhlenbeck, H., Adv. Catal. 45, 412 (2000). 132. Poppa, H., Catal. Rev.—Sci. Engng 35, 359 (1993). 133. Henry, C.R., Surf. Sci. Rep. 31, 235 (1998). 134. Rupprechter, G., Phys. Chem. Chem. Phys. 3, 4621 (2001). 135. Hess, C., Wolf, M., and Bonn, M., Phys. Rev. Lett. 85, 4321 (2000). 136. Thomas, J.M., Faraday Discuss. 122, 395 (2002), and references cited therein.

This Page Intentionally Left Blank

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

Chemistry and Technology of Isobutane/Alkene Alkylation Catalyzed by Liquid and Solid Acids ANDREAS FELLER1 and JOHANNES A. LERCHER Institut fu¨r Technische Chemie, Technische Universita¨t Mu¨nchen, D-85747 Garching, Germany

I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II. Alkylation Mechanism. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II.A. Overall Product Distribution . . . . . . . . . . . . . . . . . . . . . . . . . II.B. Initiation Steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II.C. Alkene Addition and Isomerization . . . . . . . . . . . . . . . . . . . . II.D. Hydride Transfer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II.E. Oligomerization and Cracking. . . . . . . . . . . . . . . . . . . . . . . . II.F. Self-Alkylation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II.G. Product and Acid Degradation . . . . . . . . . . . . . . . . . . . . . . . II.H. Pathways to Allylic and Cyclic Compounds . . . . . . . . . . . . . II.I. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . III. Physical–Chemical Phenomena Influencing the Reaction . . . . . . . . III.A. Properties of Liquid Acid Alkylation Catalysts . . . . . . . . . . . III.B. Properties of Zeolitic Alkylation Catalysts . . . . . . . . . . . . . . III.B.1. Adsorption and Diffusion of Hydrocarbons . . . . . . . III.B.2. Brønsted Acid Sites . . . . . . . . . . . . . . . . . . . . . . . . III.B.3. Lewis Acid Sites and Extra-Framework Aluminum . III.B.4. Silicon/Aluminum Ratio . . . . . . . . . . . . . . . . . . . . . III.B.5. Metal Ions in Ion-Exchange Positions . . . . . . . . . . . III.B.6. Structure Types of Zeolites . . . . . . . . . . . . . . . . . . . III.C. Other Solid Acids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . III.C.1. Sulfated Zirconia and Related Materials . . . . . . . . . III.C.2. Heteropolyacids . . . . . . . . . . . . . . . . . . . . . . . . . . . III.C.3. Acidic Organic Polymers . . . . . . . . . . . . . . . . . . . . III.C.4. Supported Metal Halides. . . . . . . . . . . . . . . . . . . . . III.D. The Influence of Process Conditions . . . . . . . . . . . . . . . . . . III.D.1. Reaction Temperature . . . . . . . . . . . . . . . . . . . . . .

1

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

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

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

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

230 234 234 237 239 242 247 249 251 251 252 252 253 255 255 256 260 261 263 264 267 267 268 269 270 271 272

E-mail address: [email protected] Present address: CS CLEAN SYSTEMS AG, Fraunhoferstr. 4, 85732 Ismaning, Germany.

ADVANCES IN CATALYSIS, VOLUME 48 ISSN: 0360-0564 DOI 10.1016/S0360-0564(04)48003-1

Copyright q 2004 Elsevier Inc. All rights reserved

230

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

III.D.2. Alkane/Alkene Ratio and Alkene Space Velocity. . III.D.3. Alkene Feed Composition. . . . . . . . . . . . . . . . . . . IV. Industrial Processes and Process Developments . . . . . . . . . . . . . . IV.A. Liquid Acid-Catalyzed Processes. . . . . . . . . . . . . . . . . . . . IV.A.1. Sulfuric Acid-Catalyzed Processes . . . . . . . . . . . . IV.A.2. Hydrofluoric Acid-Catalyzed Processes . . . . . . . . . IV.B. Solid Acid-Catalyzed Processes . . . . . . . . . . . . . . . . . . . . . IV.B.1. UOP Alkylenee Process . . . . . . . . . . . . . . . . . . . IV.B.2. Akzo Nobel/ABB Lummus AlkyCleane Process. . IV.B.3. LURGI EUROFUELw Process . . . . . . . . . . . . . . . IV.B.4. Haldor Topsøe FBAe Process . . . . . . . . . . . . . . . V. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

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

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

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

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

274 276 278 278 278 281 283 285 286 286 287 289 289

This contribution is an in-depth review of chemical and technological aspects of the alkylation of isobutane with light alkenes, focused on the mechanisms operative with both liquid and solid acid catalysts. The differences in importance of the individual mechanistic steps are discussed in terms of the physical–chemical properties of specific catalysts. The impact of important process parameters on alkylation performance is deduced from the mechanism. The established industrial processes based on the application of liquid acids and recent process developments involving solid acid catalysts are described briefly. q 2004 Elsevier Inc.

Abbreviations ASO DMH EFAL H0 k LHSV OSV P/O r RE RON T TMP TOS TS WHSV

acid-soluble oil dimethylhexane extra-framework aluminum Hammett acidity function rate constant liquid-hourly space velocity (m3olefin/(m3catalyst h)) olefin space velocity (kgolefin/(kgcatalyst h)) paraffin/olefin ratio (mol/mol) reaction rate rare earth research octane number temperature (K) trimethylpentane time on stream transition state weight-hourly space velocity (kgolefin/(kgcatalyst h))

I. Introduction Alkylation of isobutane with C3 – C5 alkenes in the presence of strong acids leads to the formation of complex mixtures of branched alkanes, called alkylate, which are excellent blending components for gasoline. Alkylate has a high octane

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

231

number and a low Reid vapor pressure, and is free of aromatics, alkenes, and sulfur. The worldwide production capacity as of the end of 2001 was approximately 74 million tons/year (1). Because of increasing strictness of the clean air regulations in the EU and the USA and restrictions of the contents of alkenes, sulfur, and aromatics (particularly benzene) in gasoline, the production of alkylate is expected to increase. Furthermore, the planned phase-out of methyltertiary-butyl ether (MTBE), a high-octane-number oxygenate, will boost the demand for alkylate to meet the requirements for reformulated gasoline (2). Alcohols such as ethanol, that could conceivably replace the ethers, suffer from a very high blending vapor pressure when mixed into gasoline, thus limiting their usefulness. Therefore, it is expected that the demand for alkylation catalysts will increase by 5% per year up to the year 2003, with an estimated total catalyst value for 2003 of $340 million (3). The alkylation unit in a petroleum refinery is situated downstream of the fluid catalytic cracking (FCC) units. The C4 cut from the FCC unit contains linear butenes, isobutylene, n-butane, and isobutane. In some refineries, isobutylene is converted with methanol into MTBE. A typical modern refinery flow scheme showing the position of the alkylation together with an acid regeneration unit is displayed in Fig. 1. In the 1930s, Ipatieff’s group at Universal Oil Products discovered that isoalkanes react with alkenes in the presence of strong acids to give saturated hydrocarbons under relatively mild conditions. The acids initially tested were AlCl3/HCl and BF3/HF (4). Soon, the first processes were commercialized (5). The early alkylation plants utilized sulfuric acid, but the need for high-octane-number aviation gasoline spurred by World War II led to the construction of plants based on HF as catalyst, which are more flexible regarding the feed alkenes. The first HF alkylation process units were built in 1942 by Phillips as wartime emergency units (6). The importance of alkylate increased steeply, and the daily production of alkylate then reached 5 million gallons; during the Korean War in 1952 the production rate was already 14 million gallons/day, and in the beginning of the 1980s, with the phase out of leaded gasoline in the USA, it increased to an estimated 50 million gallons/ day (7). From the 1960s to about 1986, the relative importance of plants using HF increased relative to those using H2SO4 (8). Now, nearly equal amounts of alkylate are produced on a worldwide basis by each of the two processes (1). Both H2SO4 and HF catalysts suffer from substantial drawbacks. Anhydrous HF is a corrosive and highly toxic liquid with a boiling point close to room temperature. Tests in the Nevada desert showed that, if released into the atmosphere, HF forms stable aerosols, which drift downwind at ground level for several kilometers. In 1987, the accidental release of gaseous HF in Texas City resulted in emergency treatment for several hundred people (9). Therefore,

232

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

Fig. 1. Process units in a modern refinery.

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

233

refineries with HF alkylation plants are under pressure to install expensive mitigation systems minimizing the dangers of HF leaks. Moreover, authorities in many industrialized countries have ceased to license new HF alkylation plants. Sulfuric acid is also a corrosive liquid, but not volatile, making its handling easier. Its major disadvantage is the high acid consumption in the alkylation process, which can be as much as 70 – 100 kg of acid/ton of alkylate. The spent acid contains water and heavy hydrocarbons and has to be regenerated, usually by burning. The cost of such a regenerated acid is about 2 – 3 times the market price for sulfuric acid (10). About one-third of the total operating costs of alkylation units using H2SO4 can be attributed to acid consumption (11). The sulfuric acidcatalyzed process is more sensitive than the other towards the feed alkenes; C3 and C5 alkenes generally lead to higher acid consumption and lower octane numbers of the alkylate (12). Equipment corrosion, transport, and handling hazards and environmental liability associated with the disposal of spent acid are disadvantages of both the processes. For more than 30 years, these issues have provided the driving force for research in industry and academia to find suitable replacements for the existing liquid acid catalysts. Zeolites, being non-corrosive, non-toxic, and rather inexpensive, seemed to be promising candidates, especially after they were successfully installed as cracking catalysts. In the late 1960s, two groups, those of Garwood and Venuto of Mobil Oil (13) and Kirsch, Potts, and Barmby of Sun Oil (14), did pioneering work on rare earth exchanged faujasitic zeolites. Later, other zeolites were also examined. In general, all large-pore zeolites are active alkylation catalysts, giving product distributions similar to those characteristic of the liquid acids, but their unacceptably rapid deactivation was and still is the obstacle to commercialization. Other materials that have been investigated include sulfated zirconia, Brønsted and Lewis acids promoted on various supports, heteropolyacids, and organic resins, both supported and unsupported. On the whole, these materials also deactivate rapidly, and some of them also exhibit environmental and health hazards. The technology and chemistry of isoalkane – alkene alkylation have been thoroughly reviewed for both liquid and solid acid catalysts (15) and for solid acid catalysts alone (16). The intention of this review is to provide an up-todate overview of the alkylation reaction with both liquid and solid acids as catalysts. The focus is on the similarities and differences between the liquid acid catalysts on one hand and solid acid catalysts, especially zeolites, on the other. Thus, the reaction mechanism, the physical properties of the individual catalysts, and their consequences for successful operation are reviewed. The final section is an overview of existing processes and new process developments utilizing solid acids.

234

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

II. Alkylation Mechanism Since the discovery of alkylation, the elucidation of its mechanism has attracted great interest. The early findings are associated with Schmerling (17– 19), who successfully applied a carbenium ion mechanism with a set of consecutive and simultaneous reaction steps to describe the observed reaction kinetics. Later, most of the mechanistic information about sulfuric acid-catalyzed processes was provided by Albright. Much less information is available about hydrofluoric acid as catalyst. In the following, a consolidated view of the alkylation mechanism is presented. Similarities and dissimilarities between zeolites as representatives of solid acid alkylation catalysts and HF and H2SO4 as liquid catalysts are highlighted. Experimental results are compared with quantum-chemical calculations of the individual reaction steps in various media.

II.A. Overall Product Distribution Table I gives the compositions of alkylates produced with various acidic catalysts. The product distribution is similar for a variety of acidic catalysts, both solid and liquid, and over a wide range of process conditions. Typically, alkylate is a mixture of methyl-branched alkanes with a high content of isooctanes. Almost all the compounds have tertiary carbon atoms; only very few have quaternary carbon atoms or are non-branched. Alkylate contains not only the primary products, trimethylpentanes, but also dimethylhexanes, sometimes methylheptanes, and a considerable amount of isopentane, isohexanes, isoheptanes and hydrocarbons with nine or more carbon atoms. The complexity of the product illustrates that no simple and straightforward single-step mechanism is operative; rather, the reaction involves a set of parallel and consecutive reaction steps, with the importance of the individual steps differing markedly from one catalyst to another. To arrive at this complex product distribution from two simple molecules such as isobutane and butene, reaction steps such as isomerization, oligomerization, b-scission, and hydride transfer have to be involved. The distributions of products within a certain carbon number fraction are far from equilibrium. In the C8-fraction, for example, the dimethylhexanes would be thermodynamically favored over the trimethylpentanes, but the latter are predominant. The distribution within the trimethylpentanes is also not equilibrated. 2,2,4-TMP would prevail at equilibrium over the other TMPs, constituting 60– 70% of the product, depending on the temperature. Furthermore, 2,2,3-TMP as the primary product is found in less than equilibrium amounts. Qualitatively, the same statement is valid for the other carbon number distributions. Products with a tertiary carbon atom in the 2-position dominate over other isomers in all fractions.

TABLE I RON values of various alkanes and the C5þ composition of isobutane/butene alkylates produced with various acids in laboratory scale/pilot-plant scale reactors Component (wt%)

Research octane number

Catalyst HF (T ¼ ?; P=O ¼ 12)

RE-FAU (T ¼ 348 K, P=O ¼ 7)

Sulfated zirconia (T ¼ 275 K, P=O ¼ 15) 24.0

Isopentane

93.0

1.2

1.8

6.8

n-Pentane

61.8

0

0.1

0

0

2,2-Dimethylbutane

91.8

0

0

0

0.8

2,3-Dimethylbutane

104.3

1.5

1.4

4.8

4.3

2-Methylpentane

73.4

0.2

1.4

4.8

3.5

3-Methylpentane

74.5

0.1

0.1

0.7

1.7

n-Hexane

24.8

0

0

0

0

2,2-Dimethylpentane

92.8

0

1.3

0

0.1

2,4-Dimethylpentane

83.1

0.6

1.3

3.5

5.5

2,2,3-Trimethylbutane

112.1

0.1

0

0.2

0.3

3,3-Dimethylpentane

80.8

0

0

0

0.3

2,3-Dimethylpentane

91.1

0.6

0.6

1.7

1.8

2-Methylhexane

42.4

0

0.1

1.7

1.0

3-Methylhexane

52.0

0

0.2

0.3

0.7 235

(Continued)

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

H2SO4 (T ¼ 528 K, P=O ¼ 5)

236

TABLE I Continued Component (wt%)

Catalyst

Research octane number HF (T ¼ ?; P=O ¼ 12)

RE-FAU (T ¼ 348 K, P=O ¼ 7)

Sulfated zirconia (T ¼ 275 K, P=O ¼ 15)

30.2

48.7

23.8

25.5

0

0

0

0

0

2,2-Dimethylhexane

72.5

0

0

0

0.4

2,4-Dimethylhexane

65.2

1.2

2.9

1.1

0.8

2,5-Dimethylhexane

55.5

2.0

2.1

10.1

0

2,2,3-Trimethylpentane

109.6

0.8

1.1

10.1

11.0

2,3,4-Trimethylpentane

2,2,4-Trimethylpentane n-Heptane

100

102.7

33.9

21.4

13.6

5.0

2,3-Dimethylhexane

71.3

1.7

2.1

3.0

0.9

2-Methylheptane

21.7

0

0

0

0

106.1

20.4

12.9

21.8

7.4

3,4-Dimethylhexane

76.3

0.2

0.2

1.0

0.4

3-Methylheptane

26.8

0

0

0

0

.90

0

0

0.3

1.3

5.4

2.9

7.5

3.3

2,3,3-Trimethylpentane

Octenes C9þ

<80–85

Data taken from Ref. (20) for H2SO4, Ref. (21) for HF, and Ref. (22) for sulfated zirconia; RE-FAU, unpublished data.

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

H2SO4 (T ¼ 528 K, P=O ¼ 5)

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

237

The overall reaction is highly exothermic. Depending on the product composition, 82– 93 kJ/mol of reacted isobutane are liberated (23).

II.B. Initiation Steps The alkylation reaction is initiated by the activation of the alkene. With liquid acids, the alkene forms the corresponding ester. This reaction follows Markovnikov’s rule, so that the acid is added to the most highly substituted carbon atom. With H2SO4, mono- and di-alkyl sulfates are produced, and with HF alkyl fluorides are produced. Triflic acid (CF3SO2OH) behaves in the same way and forms alkyl triflates (24). These esters are stable at low temperatures and low acid/hydrocarbon ratios. With a large excess of acid, the esters may also be stabilized in the form of free carbenium ions and anions (Reaction (1)). ð1Þ The esters differ from each other in stability. To decompose the isopropyl ester, higher temperatures and higher acid strengths are needed than for decomposition of the s-butyl ester. It is claimed that the resulting carbenium ions are stabilized by solvation through the acid (25– 27). Branched alkenes do not form esters. It is believed that they are easily protonated and polymerized (28). In zeolites, the adsorption of an alkene will lead to a surface alkoxide and not to an adsorbed carbenium ion. The alkene is “solvated” by the basic surface oxygen atoms of the zeolite, and the solvation is similar to that by water in aqueous solutions. Depending on the basicity of the surface oxygen atoms, proton transfer to adsorbed alkenes results in the formation of more or less covalent surface alkoxides rather than carbenium ions (29,30). Ab initio quantum-chemical calculations representing a cluster modeling the zeolitic acid site (29,31) showed that the alkene first forms a p-complex with the acidic site. This transforms via a carbenium ion-like transition state into the alkoxide. The transition state has a much higher positive charge than the alkoxide, and it forms a cyclic species with both oxygen atoms and an aluminum atom of the zeolite. The final alkoxide will not bind to the oxygen to which the hydrogen was bonded but instead to one neighboring it. The involvement of both oxygen atoms and the “switching” between them is characteristic of hydrocarbon transformations on zeolitic acid sites (32). An illustrative energy diagram is depicted for the isobutylene protonation in Fig. 2. More recent calculations representing propene chemisorption, however, showed the sensitivity of the system to the surrounding zeolite structure. The calculated energies were found to depend strongly on the relaxation of the zeolite

238

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

Fig. 2. Potential energy profile and structure of final alkoxide for the adsorption of isobutylene on a high-silica zeolite according to Ref. (29).

unit cell size and its shape (33). Experimentally, monomeric alkoxides are difficult to investigate. Because of their high reactivities, alkenes tend to oligomerize, so that mainly dimerized species were detected upon adsorption of isobutylene and of n-butenes on zeolites (34,35). In their experiments with perdeuterioisobutane on various zeolites, Engelhardt and Hall (36) found the carbenium ions to be metastable reaction intermediates. The lifetime of an intermediate was concluded to depend on the acid strength. The direct protonation of isobutane, via a pentacoordinated carbonium ion, is not likely under typical alkylation conditions. This reaction would give either a tertiary butyl cation (trimethylcarbenium ion) and hydrogen, or a secondary propyl cation (dimethylcarbenium ion) and methane (37 –39). With zeolites, this reaction starts to be significant only at temperatures higher than 473 K. At lower temperatures, the reaction has to be initiated by an alkene (40). In general, all hydrocarbon transformations at low temperatures start with the adsorption of the much more reactive alkenes, and alkanes enter the reaction cycles exclusively through hydride transfer (see Section II.D). When n-butenes are used, the initiation produces a secondary carbenium ion/butoxide. This species may isomerize via a methyl shift (Reaction (2)) or accept a hydride from isobutane to form the tertiary butyl cation (Reaction (3)). Isobutylene forms the tertiary cation directly.

ð2Þ

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

239

ð3Þ The skeletal rearrangement needed in reaction (2) has to go through a transition state, which resembles a primary carbenium ion, for which an activation energy of about 130 kJ/mol has been calculated (41). In zeolites and presumably in the liquid acids also, this reaction does not proceed under alkylation conditions. Another possibility is the addition of a butene molecule to the secondary butyl cation, giving a 3,4-dimethylhexyl cation, which can be freed via hydride transfer from isobutane and form the tertiary butyl cation in this way. This route seems to play only a minor role, as no significantly higher dimethylhexane selectivities during the initial reaction phase have been reported. At the same time, n-butane is formed in substantial amounts at this stage, confirming the importance of this initiating step. In the reaction with sulfuric acid and n-butenes or propene, only minor amounts of n-butane or propane are observed. Only little isobutane is consumed in the initial phase, whereas the alkenes react immediately (42). In this case, the alkenes first oligomerize to form conjunct polymers. These polymers are also called acid-soluble oil (ASO) or red oil, because they are found in the acid phase and exhibit a dark red color. This oil is a complex mixture of highly branched hydrocarbons with single and conjugated double bonds and rings containing five and six carbon atoms. The individual compounds have molecular weights in the range of 265– 360 (43). They can abstract a hydride from isobutane, forming a tertiary carbenium ion (8,44). When the reaction is started with sulfuric acid that already contains some ASO, a better alkylate is produced than with fresh acid (45), and the initiation period, which is characterized by low yield and product quality, is markedly reduced (46). The importance of the conjunct polymers is discussed below.

II.C. Alkene Addition and Isomerization Once the tertiary cations have been formed, they can undergo electrophilic addition to alkene molecules (Reaction (4)). The addition is exothermic and contributes most of all the reaction steps to the overall heat of reaction. It has been proposed (24) that instead of the alkenes, the corresponding esters are added to the carbenium ions, restoring the acid in this way (Reaction (5)). The products of both potential steps are the same. ð4Þ

240

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

ð5Þ In the case of the butene isomers, the addition will lead to different isooctyl cations, depending on the isomer and the type of carbenium ion. The reactions involving s-butyl ions are likely to be negligible for liquid acid catalysts and of minor importance for zeolites.

2-Butene as the feed alkene would thus—after hydride transfer—give 2,2,3TMP as the primary product. However, with nearly all the examined acids, this isomer has been observed only in very small amounts. Usually the main components of the TMP-fraction are 2,3,3-, 2,3,4-, and 2,2,4-TMP, with the selectivity depending on the catalyst and reaction conditions. Consequently, a fast isomerization of the primary TMP-cation has to occur. Isomerization through hydride- and methyl-shifts is a facile reaction. Although the equilibrium composition is not reached, long residence times favor these rearrangements (47). The isomerization pathways for the TMP isomers are shown schematically in Fig. 3. Using 1-butene as the feed alkene in most cases does not lead to dimethylhexanes as expected, but also to a mixture of TMPs. These are formed in a rapid isomerization of the linear butenes, almost to equilibrium compositions, in which the 2-butenes are strongly favored. On the other hand, some of the DMH-isomers produced in 2-butene alkylation also stem from a rapid isomerization of the feed. Not all acids are equally active isomerization catalysts. With zeolite H-BEA, nearly identical selectivities are achieved when the feed is 1-butene instead of 2-butene (48). In general, even mildly acidic zeolites are excellent catalysts for double-bond shift isomerization. Sulfuric acid also produces nearly identical

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

241

Fig. 3. Possible hydride- (
alkylates with 1- and 2-butene (45,49). Hydrofluoric acid, on the other hand, is known to produce substantial amounts of DMHs from 1-butene (21,50). Aluminum chloride also shows low rates of butene isomerization (18,51). It seems unlikely that under mild alkylation conditions skeletal rearrangements, which could isomerize TMP-cations into DMH-cations (and methylheptyl cations), occur to a large extent. This type of isomerization has a much higher true activation energy than hydride and methyl shifts. Theoretically, even the direct alkylation of carbenium ions with isobutane is feasible. The reaction of isobutane with a t-butyl cation would lead to 2,2,3,3tetramethylbutane as the primary product. With liquid superacids under controlled conditions, this has been observed (52), but under typical alkylation conditions 2,2,3,3-TMB is not produced. Kazansky et al. (26,27) proposed the direct alkylation of isopentane with propene in a two-step alkylation process. In this process, the alkene first forms the ester, which in the second step reacts with the isoalkane. Isopentane was found to add directly to the isopropyl ester via intermediate formation of (non-classical) carbonium ions. In this way, the carbenium ions are freed as the corresponding alkanes without hydride transfer (see Section II.D). This conclusion was inferred from the virtual absence of propane in the product mixture. Whether this reaction path is of significance in conventional alkylation processes is unclear at present. HF produces substantial amounts of propane in isobutane/propene alkylation. The lack of 2,2,4-TMP in the product, which is formed in almost all alkylates regardless of the feed (53), implies that the mechanism in the two-step alkylation process is different from that of conventional alkylation.

242

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

II.D. Hydride Transfer Intermolecular hydride transfer (Reaction (6)), typically from isobutane to an alkyl-carbenium ion, transforms the ions into the corresponding alkanes and regenerates the t-butyl cation to continue the chain sequence in both liquid acids and zeolites. ð6Þ Hydride transfer is the crucial step in the reaction sequence. It ensures the perpetuation of the catalytic cycle and leads to the exclusive formation of saturated compounds. In general, the hydride transfer between alkanes and alkyl cations is the elementary step responsible for chain propagation of acid-catalyzed transformations of hydrocarbons (54). Hydride transfer between tertiary carbon atoms is much faster than that between secondary carbon atoms. Although hydride transfers involving secondary alkyl cations take place in aluminum halide systems (55), they are too slow to be observed in sulfuric acid (56). In general, hydride transfer is accelerated by neighboring groups, which encourage the stabilization of the resulting ion (57). Investigations of hydride transfer in the gas phase (58– 61) showed that the reaction proceeds without a substantial activation energy. Its reaction rate was found to exhibit two regimes, i.e., fast kinetics at low temperatures and slow kinetics at high temperatures. This behavior was explained by a consecutive mechanism proceeding through two reaction steps. It involved the formation of a loose complex between the ion and the neutral alkane, which reacts to form a tight complex having a bridging hydride between the two fragments. The rates of different hydride transfer reactions between different carbenium ions and different alkanes were found to depend on the reaction enthalpy and steric factors involving van der Waals interactions between the approaching ion and hydrogen and methyl groups on the adjacent carbon atom next to the tertiary carbon atom. Steric hindrance in tertiary– tertiary hydride transfer reactions was also established in the liquid phase employing superacidic catalysts (62). These steric restrictions are presumably responsible for the low selectivity to the primary product 2,2,3-TMP observed with all acids. Hydride- or methyl-shifts are much more likely than hydride transfer to a difficult-to-access carbon atom bearing the positive charge. Note that the precursor carbenium ions of the most abundant TMPs have their charge centers next to the chain end at a tertiary carbon atom (Fig. 3). There are substantial differences between gas-phase and liquid-phase hydride transfer reactions. In the latter, the hydride transfer occurs with a low activation energy of 13 – 17 kJ/mol, and no carbonium ions have been detected as intermediates when secondary or tertiary carbenium ions were present (25).

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

243

These differences were explained by solvation effects in the liquid phase. The carbenium ions are more efficiently stabilized by solvation than carbonium ions, because the former have unsaturated trivalent carbon atoms. In this way, the energy barrier between the (solvated) carbenium ion and the carbonium ion transition state increases. In zeolites, this barrier is even higher. As discussed in Section II.B, the lower acid strength and the interaction between the zeolitic oxygen atoms and the hydrocarbon fragments lead to the formation of alkoxides rather than carbenium ions. Thus, extra energy is needed to transform these esters into carbonium ionlike transition states. Quantum-chemical calculations of hydride transfer between C2 – C4 adsorbed alkenes and free alkanes on clusters representing zeolitic acid sites led to activation energies of approximately 200 kJ/mol for isobutane/tertbutoxide (29), 230 –305 kJ/mol for propane/sec-propoxide, and 240 kJ/mol for isobutane/tert-butoxide (32), 130 – 150 kJ/mol for ethane/ethene (63), 95 – 105 kJ/mol for propane/propene, 88– 109 kJ/mol for isobutane/isobutylene, and 110 –118 kJ/mol for propane/isobutylene (64). In the last two references, the carbonium ions were not found to be transition states but instead to be energetically high-lying reaction intermediates. The authors claimed that these carbonium ions exist as intermediates when the charge is delocalized and not accessible to framework oxygen. The carbonium ions decompose directly into the alkene and alkane, without forming alkoxides. Thus, the activation energies are about a 100 kJ/mol lower than those calculated in the other mentioned references, because covalent bonds do not have to be broken to reach the transition state. Note that the activation energy is lowest in tertiary– tertiary hydride transfer. In a study by Nowak et al. (65) activation energies for hydride transfer between isobutane and tertiary and secondary acceptor cations were compared with activation energies of isomerization steps between tertiary carbenium ion species. The energy for tertiary –tertiary hydride transfer was comparable to the energy of the isomerization, whereas the energy for tertiary – secondary hydride transfer was almost twice as high. Another study of ethane/ethene hydride transfer was performed to investigate the influence of the Si/Al ratio and different levels of coverage of the acid sites (66). The zeolite was modeled to represent the chabazite structure. It was found that the electrostatic effects increase with decreasing Si/Al ratio, but they are important only when the interaction between the zeolite and the adsorbed species is clearly ionic. High coverage led to a destabilizing effect on the carbonium ions due to repulsion between neighboring ions. The authors inferred that the electrostatic forces are just one of many effects being of importance in zeolitecatalyzed hydrocarbon reactions. Figure 4 summarizes the different calculated potential energy profiles for the hydride transfer reaction in different media. Experimental results characterizing hydride transfer in zeolites are scarce, as it is a secondary reaction, which cannot be observed directly. Data from kinetics

244

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

Fig. 4. Potential energy profiles for the isobutane/t-butyl cation hydride transfer reaction in various media (25,64).

measurements of cracking reactions of 2,2,4-TMP on USY zeolite gave values for the apparent energies of activation of 47 kJ/mol lumped for all the hydride transfer reactions that were occurring (67). A more detailed investigation of isobutane cracking gave values of 64 kJ/mol for isobutane/propyl, 76 kJ/mol for isobutane/n-butyl, and 62 kJ/mol for isobutane/isopentyl hydride transfer (40). An earlier investigation by the same group led to higher values, namely, 81 kJ/ mol for isobutane/propyl, 67 kJ/mol for isobutane/n-butyl, and 125 kJ/mol for isobutane/isopentyl hydride transfer (68). Even when average heats of adsorption (ca. 40 kJ/mol) are added to the measured apparent energies to estimate the true activation energies, these values are lower than the calculated values. Clearly, the theoretical calculations overestimate the energy barrier. The overestimation is speculated to be a consequence of incorrect modeling of the acid strength (deprotonation energy, basicity of the lattice oxygen atoms) in the zeolitic cluster used for the calculation. It has been proposed that hydride transfer in zeolites requires the presence of two adjacent Brønsted acid sites (69). In light of the above-mentioned theoretical examinations and also adsorption isotherms of 1-butene and n-butane on USY zeolites with various aluminum content (70), this proposition seems unlikely. The reaction enthalpy of the hydride transfer step usually has a low absolute value. Whether hydride transfer is exo- or endothermic depends on the stability (evidenced by the heat of formation) of the involved carbenium ions. Branched carbenium ions are more stable than linear ones. Longer carbenium ions are more stable than shorter ones. Replacement of a long-chain carbenium ion by

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

245

a short-chain alkane to give a short-chain ion is endothermic, as exemplified by the transfer of a hydride from isobutane to C8 carbenium ions. With both liquid acid catalysts, but presumably to a higher degree with sulfuric acid, hydrides are not transferred exclusively to the carbenium ions from isobutane, but also from the conjunct polymers (44,46,71). Sulfuric acid containing 4 – 6 wt% of conjunct polymers produces a much higher quality alkylate than acids without ASOs (45). Cyclic and unsaturated compounds, which are both present in conjunct polymers, are known to be hydride donors (72). As was mentioned in Section II.B, these species can abstract a hydride from isobutane to form the t-butyl cation, and they can give a hydride to a carbenium ion, producing the corresponding alkane, for example the TMPs, as shown in reactions (7) and (8). ð7Þ ð8Þ In this way, the conjunct polymers serve as a reservoir of hydride ions. Under some conditions, the polymers are a source of hydride ions, but they accept these ions under other conditions. Substantial amounts of the saturated products are supposedly formed via this route with sulfuric acid. In zeolites, species similar to conjunct polymers also form. The heavy hydrocarbon molecules, which deactivate the catalyst by pore blocking or by site blocking, are generally termed “soft coke” or “low-temperature coke”, because of the absence of aromatic species. Only scant information is available about the influence of coke formation on the alkylation mechanism. It has been proposed that, similar to the conjunct polymers in liquid acids, heavy unsaturated molecules participate in hydride transfer reactions. However, no direct evidence was given for this proposition (69). In another study, the hydride transfer from unsaturated cyclic hydrocarbons was deduced from an initiation period in the activity of NaHY zeolites; complete conversion of butene was achieved only after sufficient formation of such compounds (73). In a series of investigations of the cracking of alkanes and alkenes on Y zeolites (74,75), the effect of coke formation on the conversion was examined. The coke that formed was found to exhibit considerable hydride transfer activity. For some time, this activity can compensate for the deactivating effect of the coke. On the basis of dimerization and cracking experiments with labeled 1butene on zeolite Y (76), it is known that substantial amounts of alkanes are formed, which are saturated by hydride transfer from surface polymers. In both liquid and solid acid catalysts, hydride transfer from isoalkanes larger than

246

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

isobutane may occur, especially from isopentane, which sometimes is used as feedstock. However, no data are available providing information about the significance of hydride transfer reactions with higher hydrocarbons. Hydride transfer from alkenes was also proposed to occur during sulfuric acid-catalyzed alkylation modified with anthracene (77). Then the butene loses a hydride and forms a cyclic carbocation intermediate, yielding—on reaction with isobutene—trimethylpentyl cations. This conclusion was drawn from the observation of a sharp decrease in 2,2,3-TMP selectivity upon addition of anthracene to the acid. Fast hydride transfer reduces the lifetime of the isooctyl cations. The molecules have less time to isomerize and, consequently, the observed product spectrum should be closer to the primary products and further from equilibrium. This has indeed been observed when adamantane, an efficient hydride donor, was mixed with zeolite H-BEA as the catalyst (78). When 2-butene/isobutane was used as the feed, the increased hydride transfer activity led to considerably higher 2,2,3TMP and lower 2,2,4-TMP selectivities, as shown in Fig. 5.

Fig. 5. Changes in TMP selectivities with the use of adamantane (5 wt%) as an additive in a HBEA catalyst at 30 min TOS (P=O ¼ 10; OSV ¼ 0:2 h21, T ¼ 348 K) (78).

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

247

II.E. Oligomerization and Cracking The overall product distribution is governed by the relative rates of alkene addition and hydride transfer. With all acids, alkene addition is a much more facile reaction than hydride transfer. With sulfuric acid, n-butene oligomerization was found to be four times faster than hydride transfer (79). With zeolites, de Jong et al. (80) reported oligomerization to be two orders of magnitude faster than hydride transfer, whereas Simpson et al. (81) reported it to be three orders of magnitude faster. With too low internal alkane/alkene ratios the alkenes will oligomerize before they can be removed via hydride transfer. This is the key problem in solid acid-catalyzed alkylation. A polymer will build up, which will finally block the acid sites. With liquid acids, the conjunct polymers help in maintaining a high hydride transfer activity. However, when the concentration reaches a critical level, the acid strength will be too low for producing highquality alkylate. For this reason, in a continuous process, a stream of used acid has to be constantly replaced by fresh acid to maintain the optimum level of acid strength. The route to oligomerization products (sometimes also called multiple alkylate) is depicted in Fig. 6. The rate constant kA defines the rate of alkene addition, kB the hydride transfer rate, and kC the rate of deprotonation. The rate ratio rB =ðrA þ rC Þ is the critical parameter that determines whether the catalyst will effectively catalyze alkylation or deactivate quickly through multiple alkylation/oligomerization reactions. High ratios can be achieved with low alkene concentrations (as would be achieved in a backmixed reactor) and maximized hydride transfer rates (a property of the catalyst). Hydrocarbons with up to 16 carbon atoms are detected in a typical alkylate (82). With the liquid acids, it was found that the oligomerization rate is higher for isoalkenes than for linear alkenes (49). The same is true for solid acids (14,83). Because of their tertiary carbon atoms, isobutylene and isopentene obviously react more easily with carbenium ions. This point can be inferred from the reverse reaction, b-scission (see below), which is fastest for reactions of tertiary cations to give tertiary cations. In oligomerization experiments, the following pattern of

Fig. 6. Pathway to oligomerization products with the corresponding rate constants. Adapted from Ref. (81).

248

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

reactivity of alkenes was found: isobutylene q n-butenes . propene . ethene. This order can be readily explained by the relative stabilities of the carbenium ions involved (84). Not only are products with carbon numbers that are multiples of four are formed, but so also are C5 –C7 and C9, C10, and higher hydrocarbons. Cracking is invariably associated with oligomerization. The heavy cations formed by oligomerization have a tendency to fragment, forming C4 –C16 cations and alkenes, according to the b-scission rule, as depicted schematically in Reaction (9) for a dodecyl cation cracking to give an isopentene and a heptyl cation.

ð9Þ

The isopentene produced will either be protonated or be added to another carbenium ion. With a butyl cation, this would lead to a nonyl cation. The resultant carbenium ion fragment can accept a hydride and form a product heptane, or it can possibly add a butene to form a C11 cation. With hydride transfer, another alkane with an odd number of carbon atoms is produced. Just this example is sufficient to show the huge variety of possible reactions. By means of gas chromatographic analysis, Albright and Wood (82) found about 100– 200 peaks in the C9 –C16 region, regardless of the alkene and acid employed. A similar number of products can be observed for solid acid-catalyzed alkylation. In general, oligomerization and cracking products exhibit lower octane numbers than the TMPs. Average research octane number (RON) values of 92 – 93 for C5 – C7 and of 80 – 85 for C9 –C16 have been reported (8). Parts of the octane fraction also stem from oligomerization/cracking reactions. It is believed that substantial amounts of the dimethylhexanes are produced via this route (79), especially when isobutylene is the feed alkene (71). Isobutylene tends to oligomerize quickly. Hence, it produces higher amounts of light and heavy ends and cannot isomerize to 1-butene to produce DMHs in this way. Some of the TMPs also will be produced through oligomerization/cracking pathways (20). Concentrations of more than 20 wt% of TMPs in the C6þ fraction have been observed in isopentane/2-pentene alkylation (53). The TMPs cannot be produced via simple alkylation or self-alkylation with this feed. It has been proposed that oligomerization/cracking constitutes the main route to alkylation products (16), but this proposition fails to explain the usually high selectivity to the TMPs. To form trimethylpentanes, some specific precursors would have to build up in high concentrations, which is rather unlikely. Hydrocracking experiments under ideal conditions provided kinetics information characterizing the b-scission step. On the basis of this work, a classification

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

249

of various types of b-scission has been introduced (85). Fragmentations starting from a tertiary carbenium ion and giving a tertiary cation (type A) are very rapid. Fragmentations involving secondary and tertiary ions (type B) are slower than tertiary– tertiary b-scissions, but faster than secondary –secondary b-scissions (type C). The slowest mode is the cracking of a secondary ion to give a primary ion (type D). From the typical low reaction temperatures and the product composition of a typical alkylate, which consists almost exclusively of branched hydrocarbons, it can be concluded that only type A b-scissions occur at significant rates. Furthermore, protolytic cracking of alkanes via a carbonium ion mechanism is highly unlikely under typical alkylation conditions. Hydrogen or methane, which are characteristic products of such cracking, are not found in the alkylate. At low temperatures, the cracking of alkanes is initiated by traces of alkenes in the feed (also see Section II.B). In general, oligomerization is an exothermic reaction (and therefore the reverse, b-scission, is an endothermic reaction). Quantum-chemical calculations of the b-scission step on a zeolite represented by a cluster model were performed to estimate activation energies. For tertiary – secondary fragmentations, values in the range of 234 – 284 kJ/mol and for secondary – secondary values in the range of 288 – 314 kJ/mol (32) and 217 –275 kJ/mol (86) were calculated. Here, the activation energy of the reverse reaction was reported to be 71 kJ/mol less than that of the forward reaction. Evaluation of alkane conversion experiments with USY zeolite as catalyst, in general, provided much lower values than these (40, 67); average apparent activation energies for secondary –tertiary and secondary – secondary b-scission steps were estimated to be approximately 115 kJ/mol. The values for tertiary– tertiary b-scission given in the two references differed between 66 and 102 kJ/mol. In an older study by the same authors (68), values for b-scission and oligomerization were given; tertiary –tertiary b-scission was characterized by an activation energy of 184 kJ/mol and the reverse reaction by a value of 105 kJ/mol. Tertiary –secondary b-scission was found to be characterized by an activation energy of 84 kJ/mol and the reverse reaction by a value of 71 kJ/mol. The corresponding values for secondary –secondary b-scission were found to be 130 and 33 kJ/mol. As for hydride transfer, the calculated values are significantly greater than the measured values (plus the heat of adsorption), presumably as a consequence of an underestimation of the acid strength.

II.F. Self-Alkylation With hydrofluoric acid (23,50), and to a lesser degree also with zeolites (14,81, 87 –89), a significant fraction of the product stems from self-alkylation, which is sometimes also termed hydrogen transfer. The importance of this mechanism depends on the acid, the alkene, and the reaction temperature. Self-alkylation

250

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

reactivity increases with molecular weight and the degree of branching of the feed alkene (90). Generally, sulfuric acid is less active for self-alkylation than hydrofluoric acid. Only when pentenes or higher alkenes are used is selfalkylation significant with sulfuric acid (49,91). In Fig. 7, the mechanism is displayed for an exemplary isobutane/2-butene feed. The crucial step in self-alkylation is decomposition of the butoxy group into a free Brønsted acid site and isobutylene (proton transfer from the t-butyl cation to the zeolite). Isobutylene will react with another t-butyl cation to form an isooctyl cation. At the same time, a feed alkene repeats the initiation step to form a secondary alkyl cation, which after accepting a hydride gives the t-butyl cation and an n-alkane. The overall reaction with a linear alkene CnH2n as the feed is summarized in reaction (10): 2i-C4 H10 þ Cn H2n ! i-C8 H18 þ Cn H2nþ2

ð10Þ

With propene, n-butene, and n-pentene, the alkanes formed are propane, n-butane, and n-pentane (plus isopentane), respectively. The production of considerable amounts of light n-alkanes is a disadvantage of this reaction route. Furthermore, the yield of the desired alkylate is reduced relative to isobutane and alkene consumption (8). For example, propene alkylation with HF can give more than 15 vol% yield of propane (21). Aluminum chloride– ether complexes also catalyze self-alkylation. However, when acidity is moderated with metal chlorides, the self-alkylation activity is drastically reduced. Intuitively, the formation of isobutylene via proton transfer from an isobutyl cation should be more pronounced at a weaker acidity, but the opposite has been found (92). Other properties besides acidity may contribute to the self-alkylation activity. Earlier publications concerned with zeolites claimed this mechanism to be a source of hydrogen for saturating cracking products or dimerization products (69,93). However, as shown in reaction (10), only the feed alkene will be saturated, and dehydrogenation does not take place.

Fig. 7. Self-alkylation mechanism, depicted with 2-butene as the feed alkene.

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

251

II.G. Product and Acid Degradation It has been found that C7 – C9 isoalkanes react with strong acids to produce a low-quality alkylate and conjunct polymers (94). In the presence of conjunct polymers, highly branched isoalkanes might re-enter the reaction cycle by the reverse of reaction (8). Oligomerization/cracking will then lead to inferior products. This problem affects alkylation by both HF and H2SO4. It is unclear whether this side reaction is of importance with zeolites under alkylation conditions. On the zeolite H-FAU at temperatures as low as 373 K, 2,2,4-TMP undergoes cracking into isobutane and isobutylene, with significant coke formation (95). A problem that is characteristic of sulfuric acid-catalyzed alkylation is its capability to oxidize hydrocarbons. H2SO4 decomposes in the presence of isoalkanes to form water, SO2, and alkenes. This is a slow process, and so it occurs predominantly when the acid is in contact with hydrocarbons for a longer period. Higher temperatures favor the formation of SO2 (10). Some irreversible reactions between acid and hydrocarbons also take place during alkylation. Sulfone, sulfonic acid, and hydroxy groups have been detected in conjunct polymers produced with H2SO4 as the catalyst (8,96). Kramer (97) reported that 2,3,4-TMP, after an induction period, is converted into a mixture of lower alkanes (with a high fraction of isobutane) and isomerized octanes. The reaction was initiated by the reduction of sulfuric acid to SO2 with the formation of carbenium ions. In a subsequent paper by Kramer (98), more information about the reaction of selected branched alkanes with sulfuric acid led to the conclusion that SO2 is produced only during the initiation reaction. All subsequent reactions are conventional carbenium ion type reactions. Alkanes with a higher degree of branching show higher rates of degradation. Pure isobutane was found to react with sulfuric acid at 298 K. The acid was slowly reduced to SO2, with isobutane forming carbenium ions undergoing subsequent reactions. With traces of alkenes in the feed, however, acid reduction was not observed (99).

II.H. Pathways to Allylic and Cyclic Compounds The conjunct polymers formed during liquid-phase alkylation contain single and conjugated double bonds and five- and six-rings. The residue on zeolitic catalysts is highly branched, containing double bonds and conjugated double bonds and possibly also five- and six-rings (73,88,100,101). The H/C ratio is about 1.8 (102), similar to that of conjunct polymers. In general, it is believed that at temperatures below 473 K, coking of acidic catalysts mainly involves condensation and rearrangement steps. Aromatic compounds are usually not formed under such mild conditions (95). Extending these results to typical alkylation

252

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

reaction conditions, we expect that several alkene molecules will oligomerize and crack or deprotonate to form a large and branched alkene. This alkene might transfer a hydride to another carbenium ion and thus form an alkenyl carbenium ion, which can desorb via proton transfer as a diene (Reaction (11)). Further hydride transfer leads to a dienylic cation, which easily rearranges into an alkylsubstituted ring (Reaction (12)) via a 1,5-cyclization and subsequent hydride and methyl shifts. ð11Þ

ð12Þ

The resultant cycloalkenyl carbenium ions, especially the cyclopentenyl cations, are very stable (103,104) and can even be observed as free cations in zeolites (105,106). These ions can oligomerize further and, within zeolites, irreversibly block the acidic hydroxyl groups. With liquid acids, the oligomers will dilute the acid and thus lower its acid strength.

II.I. Summary Figure 8 summarizes the main reactions occurring during alkylation. Dimerization and oligomerization reactions are more important with zeolitic catalysts on acidic sites with lower acid strengths (Section III.B.2) or with severely diluted liquid acids (Section III.A). Hydride transfer from conjunct polymers is more important with sulfuric acid, and self-alkylation activity is more significant with hydrofluoric acid. Repeatedly going through the alkylation cycle without hydride transfer (multiple alkylation) and through the dimerization cycle without proton transfer (oligomerization) leads to the formation of heavy compounds, which will react further via cracking, hydride or proton transfer, and cyclization. As long as the catalyst shows sufficient hydride transfer activity, all alkenes will react, and only saturated products will leave the reaction cycles.

III. Physical–Chemical Phenomena Inﬂuencing the Reaction As was pointed out, the chemistry of the alkylation reaction can be explained by a set of mechanistic steps that are similar and in some cases the same for all the different acids examined. However, the importance of each step varies with

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

253

Fig. 8. Concerted alkylation mechanism including alkylation, “self-alkylation”, cracking, dimerization, and hydride transfer via isobutane and via conjunct polymers.

the catalyst and reaction conditions. The understanding of these parameters is thus of utmost importance. This is especially true for the solid acid catalysts. They can be synthesized and modified in a nearly infinite number of ways to influence in a complex and subtle manner the alkylation performance. In Section III.A, the chemical and physical properties of the individual alkylation catalysts and how they affect the mechanism are reviewed, and concomitantly the influence of process parameters, such as temperature, alkane/alkene ratio, and residence time on the reaction is assessed.

III.A. Properties of Liquid Acid Alkylation Catalysts In the liquid acid-catalyzed processes, the hydrocarbon phase and the acid phase are only slightly soluble in each other; in the two-phase stirred reactor, the hydrocarbon phase is dispersed as droplets in the continuous acid phase. The reaction takes place at or close to the interface between the hydrocarbon and the acid phase. The overall reaction rate depends on the area of the interface. Larger interfacial areas promote more rapid alkylation reactions and generally result in higher quality products. The alkene is transported through the hydrocarbon phase to the interface, and, upon contact with the acid, forms an acid-soluble ester, which slowly decomposes in the acid phase to give a solvated

254

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

carbenium ion or the alkene. Isobutane can react at the interface or be transported into the acid phase and react there. The most important parameters determining the ease of formation of a large reaction zone are the viscosity and the solubility of hydrocarbons in the acid. These properties differ substantially for sulfuric and hydrofluoric acid. Under typical alkylation conditions, the viscosity of sulfuric acid is two orders of magnitude higher than that of hydrofluoric acid, and the solubility of isobutane is approximately 30 times lower. The relatively high solubility of isobutane in HF, together with a high interfacial area, ensures high isobutane/alkene ratios in the acid and, thus, high hydride transfer rates and relatively low selectivity for the formation of undesired products from oligomerization/cracking and isomerization and for the formation of conjunct polymers. Consequently, sulfuric acid/hydrocarbon phases have to be mixed much more vigorously than hydrofluoric acid/hydrocarbon phases to obtain a high-quality alkylate. For the same reason, hydrofluoric acid-catalyzed processes can operate at lower residence times and higher temperatures than sulfuric acid-catalyzed processes. Using sulfuric acid with isobutane/2-butene in a laboratory reactor, Li et al. (107) found that increasing the agitator speed from 1000 to 3000 rpm increased the product RON from 86 to 94. Albright (11) discerned a minimum of four types of droplets in acid/hydrocarbon dispersions. The droplets differ in size and in the concentrations of reactants and products. The formation (and the separation) of acid/ hydrocarbon emulsions depends on the temperature, the composition of the acid, and the acid/hydrocarbon ratio (108). Sulfuric acid is a somewhat stronger acid than hydrofluoric acid. The values of the Hammett acidity function H0 for the water-free acids are 2 14.1 for H2SO4 and 2 12.1 for HF. It is, however, interesting to note that the maximum alkylate quality obtained with sulfuric acid is not achieved with the highest acidity, but with acid containing 1– 1.5 wt% water and 4– 5 wt% ASOs (96). Water reduces the acidity to a greater extent than hydrocarbon diluents. Besides their hydride transfer capabilities, the ASOs act as surfactants, increasing the interfacial area. When the concentration of diluents exceeds a certain level, the acid strength is too low to produce a high-quality alkylate. Sulfuric acid of 60 –80 wt% concentration catalyzes only alkene oligomerization. The acid strength is too low for catalysis of the more demanding reactions hydride transfer and b-scission (27). A relatively sharp transition between oligomerization and alkylation activity has been measured with sulfuric acid at H0 values between 2 8.0 and 2 8.5 (109). If such low-acidity values occur in an alkylation reactor, oligomerization reactions become so predominant that the acid strength cannot be maintained and the plant is said to be in an acid runaway condition. The same principles regarding the acidity can be applied to hydrofluoric acidcatalyzed alkylation, which is more sensitive towards water, so that the feed must be thoroughly dried before entering the reactor. Furthermore, the acid dilution by

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

255

hydrocarbons is greater as a consequence of their higher solubility in HF (15). Employing triflic acid modified with water or trifluoroacetic acid, Olah et al. (110) found the best alkylation conditions at an acid strength of about H0 ¼ 210:7 for both systems. Pure triflic acid (with H0 ¼ 214:1) produced mainly cracked compounds. Diluted triflic acid with H0 . 210:7 favored oligomerization. The same research group tested different liquid acids diluted with liquid carbon dioxide. Although very strong acids such as triflic acid produce higher quality alkylate upon dilution with CO2, sulfuric acid (being less strongly acidic than triflic acid) performed better without CO2 (111). The different H0 values observed for the transition from alkylation to oligomerization with sulfuric and triflic acid suggest that the acid strength is not the only factor determining the reactivity of the carbenium ions.

III.B. Properties of Zeolitic Alkylation Catalysts Zeolite molecular sieves are widely used as solid acid catalysts or catalyst components in areas ranging from petroleum refining to the synthesis of intermediates and fine chemicals (112,113). An important reason for their widespread use is the flexibility they offer regarding the tailoring of the concentration and nature of catalytically active sites and their immediate environments. We note that discrimination between chemical and structural aspects works well at a conceptual level, but one faces quite severe limitations as soon as one tries to separate the contributions of the two effects. The complexity arises because the chemical properties of a particular molecular sieve are connected with its framework density. III.B.1. Adsorption and Diffusion of Hydrocarbons One of the major characteristics of acidic zeolites that sets them apart from the liquid acids is their selective and strong chemisorption of unsaturated compounds. Because of the high polarity of the zeolitic surface, especially in aluminum-rich zeolites, polar molecules will be preferentially adsorbed. This property is clearly demonstrated by the high water uptake capacity of zeolite X, which exceeds 25 wt%. Furthermore, the electrostatic field in the zeolite pores enhances the adsorption of polarizable molecules (114). Thus, although the concentration of alkenes in the liquid phase might be low, they will preferentially adsorb in the zeolite pores, so that in the pore system the alkene concentration will be considerably higher resulting in much higher relative rates of oligomerization vs. hydride transfer, as was discussed in Section II.E. This selective adsorption is the major reason why zeolites deactivate rapidly if no special measures are taken to minimize the alkene concentration close to the acidic sites.

256

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

Nevertheless, the adsorption of alkenes can differ substantially from one zeolite to another, even for one type of zeolite, depending on the concentration of framework aluminum and the modification procedure (70). Also typical for molecular sieves is the increasing heat of adsorption of hydrocarbons with increasing chain length (115). Each carbon atom contributes equally to the total heat of adsorption. This value depends on the zeolite pore size and shape, so that different adsorption enthalpies are measured for different zeolites. Increasing framework density (number of T-atoms per unit volume, where T refers to Si or Al) leads to increased heats of adsorption (116,117). Protons add another constant value (which depends on the chemical composition) to the overall heat of adsorption, as represented in Fig. 9A and B. This phenomenon is responsible for different apparent activation energies for a given reaction type found with hydrocarbons of different chain lengths. The actual intrinsic activation energies (as well as the corresponding pre-exponential factors) are nearly independent of chain length (118). Assuming the relationship between chain length and adsorption enthalpy to be linear over a wide range, relative desorption rates for various hydrocarbons can be calculated for a given temperature. Thus, using the data for H-FAU and a temperature of 348 K, the desorption of a C12 molecule is four orders of magnitude slower than that of an C8 molecule, and that of a C16 is eight orders of magnitude slower and that of a C20 12 orders of magnitude slower (Fig. 9C). These huge differences give one a sense of the difficulties of removing heavy products from the zeolite surface using purely adsorption/desorption arguments. Once such a heavy molecule has formed, it is unlikely to desorb.

III.B.2. Brønsted Acid Sites Zeolites exhibit a considerably lower proton (acid site) concentration than liquid acids. For example, 1 g of H2SO4 contains 20 £ 1023 moles of protons, whereas 1 g of zeolite HY, with a Si/Al atomic ratio of five, contain no more than 3 £ 1023 moles of protons. (Note that this is a crude approximation of the acidic sites available for catalysis, because it assumes that with both materials all protons are available and catalytically active.) Moreover, 1 g of H2SO4 occupies far less volume (i.e., 0.5 cm3) than the equivalent mass of zeolite (4 –6 cm3). In contrast to liquid acids, zeolites encompass different populations of sites that differ substantially in their nature and strength. Liquid acids with a given composition have a well-defined acid strength. This is not the case for zeolites. Depending on the type of zeolite, its aluminum content, and the exchange procedure used in its preparation, Brønsted and Lewis acid sites with a wide range of strengths and concentrations are present. To summarize the effects of all parameters influencing the acidity of zeolites is beyond the scope of this review.

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

Fig. 9. Effect of the chain length of hydrocarbons on the adsorption enthalpy and rates of desorption. (A) Hydrocarbon in interaction with zeolite framework. Methyl groups interact with the framework oxygen; protons exhibit an additional attractive force. (B) Heat of adsorption as a function of carbon number for zeolites MFI and FAU in the acidic and non-acidic form. (C) Relative desorption rates of a C12, C16, and C20 alkane compared to octane at 348 K. Values calculated from the linear extrapolation of the heat of adsorption values shown in (B).

257

258

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

The different reaction steps in alkylation require different minimum acid strengths to be effectively catalyzed. Double bond isomerization is catalyzed even by weak acid sites. Even a fully deactivated zeolite retains some activity for isomerization of butenes (119,120). Dimerization/oligomerization also does not require strong acidity, as was concluded from a study of a series of USY zeolites with different unit cell sizes. Correlations between the acidity and the alkylation performance revealed that the acid strength required for performing the different reactions occurring during alkylation decreases in the order cracking . alkylation (addition of butene to a tertiary butyl) . dimerization (addition of a butene to a secondary butyl) (121). A comparison between the iso-structural HSAPO-37 and H-FAU as alkylation catalysts showed that the H-FAU has a much higher relative concentration of strong acid sites than the H-SAPO-37. Therefore, the H-SAPO-37 mainly catalyzed dimerization, with a small amount of 3,4-DMH as the most abundant saturated compound, whereas the H-FAU produced mainly TMPs (122). The lifetime of a zeolitic alkylation catalyst depends on the concentration of Brønsted acid sites. This has been shown by Nivarthy et al. (78), who used a series of zeolites H-BEA with varied concentrations of back-exchanged sodium ions. The sodium decreased the concentration of Brønsted acid centers, which led to a concomitant decrease in the measured catalyst lifetime during alkylation. However, there are contrasting opinions about the acid strength required for optimum alkylation performance with zeolites. Hydride transfer is the step that determines the product quality and the catalyst lifetime. Thus, it is crucial to know which conditions favor a high hydride transfer rate. From the abovepresented investigations, it can be concluded that stronger sites are necessary to effectively catalyze hydride transfer. Sto¨cker et al. (123) synthesized and tested EMT and FAU samples with enhanced Si/Al ratios of 3.5 (made by use of crown ethers as templates). They explained the better performance of H-EMT relative to that of H-FAU by its higher ratio of strong-to-weak Brønsted acid sites. Dealumination of the H-FAU led to better results because of additional small numbers of very strong acid sites. No direct demonstration was given to support this opinion (124). La-exchange of H-EMT led to a slightly better performance than that of H-EMT. This improvement was also attributed to a higher ratio of strong-to-weak Brønsted acid sites (125). A similar conclusion was derived by Corma et al. (126), who compared USY, MOR, BEA, ZSM-5, and MCM-22. The relative decrease of activity for the formation of TMPs with time on stream was observed to depend on the concentration of strong Brønsted acid sites in the fresh zeolite. Diaz-Mendoza et al. (127) studied commercial REY, USY, and BEA samples. In contrast to the aforementioned groups, they found Brønsted acid sites with intermediate strengths to be the appropriate sites for maintaining good alkylation performance.

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

259

It is well established that with time on stream the sites strong enough for catalyzing hydride transfer deactivate first. In the first phase of operation and deactivation, the catalyst produces a mixture of saturated isoalkanes, usually with (nearly) complete butene conversion, and in the second phase, alkenes, mainly octenes, are produced at a substantially lower butene conversion. The product in this second phase resembles the product observed with the weakly acidic HSAPO-37. The mixture of butene isomers found in the product stream in the second phase is close to the equilibrium composition. A typical example is depicted in Fig. 10. Several investigations of zeolite- and other solid acidcatalyzed alkylation obviously have been performed in the oligomerization regime (129– 134). As a consequence of insufficient acidity or an inappropriate choice of reaction conditions, the catalysts examined in these investigations produced mainly oligomerization products and only small amounts of true alkylate. Unequivocal conclusions can be drawn neither about the alkylation mechanism nor on the true alkylation activity of the tested materials under these circumstances. The characterization method employed in nearly all the above-mentioned investigations for measuring the strength of acid sites was pyridine adsorption/ desorption monitored by IR spectroscopy. Pyridine forms the pyridinium ion on Brønsted acid sites and binds to Lewis acid sites by forming coordination complexes. Heating the sample with the adsorbed pyridine gives characteristic desorption curves for pyridine bound to Brønsted or Lewis acid centers. From such data, Brønsted acid/Lewis acid site ratios at a given temperature and “strong” to “weak” acid site ratios can be calculated and correlated with

Fig. 10. Typical time on stream behavior of a CeY zeolite alkylated in a fixed-bed reactor (128).

260

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

the catalytic performance. Strong and weak acid sites here are defined by the amount of pyridine that remains adsorbed at a certain temperature—the higher the temperature, the stronger the bonding and the stronger the acid site. These temperatures are chosen in a rather arbitrary manner, the upper limit of which is typically restricted by the IR cell and the zeolite itself (673 – 823 K). The strong acid sites are not of uniform strength, likely exhibiting a broad distribution, up to a minute amount of “very strong” acid sites, which are difficult to detect because of their low concentrations. Whether such very strong acid sites are responsible for an enhanced alkylation activity is not yet determined. III.B.3. Lewis Acid Sites and Extra-Framework Aluminum Lewis acid sites in zeolites originate from a partial destruction of the framework. During the modification procedure, which is necessary to transform the parent material into its acidic form, part of the aluminum present in the framework is removed from its positions in the crystalline framework (usually during calcination in a water-containing atmosphere, i.e., high-temperature hydrolysis of Si – O – Al bonds) to give extra-framework aluminum species (EFAL). Some of the species formed in this way exhibit Lewis acidity. Another source of Lewis acid sites is metal ions in cation-exchange positions. However, most of these metals exhibit weaker Lewis acidity than aluminum species. Lewis acid sites do not catalyze the alkylation reaction, but their presence undoubtedly influences the performance of zeolitic catalysts in alkylation. It has been claimed that the presence of strong Lewis acid sites promotes the formation of unsaturated compounds (127). The favored production of unsaturated carbenium ions with increased Lewis acidity was also evidenced by Flego et al. (100) investigating the deactivation of a La – H-FAU zeolite in isobutane/1-butene alkylation. Increasing the catalyst activation temperature led to higher Lewis acid site concentrations, which increased the formation of mono- and di-enylic carbenium ions. Besides the ability to increase the rate of formation of unsaturated compounds, Lewis acid sites have been claimed to increase the alkane/alkene ratio close to the Brønsted acid sites, through the adsorption/desorption equilibrium of the alkenes on the Lewis acid sites. The increased alkene concentration accelerates oligomerization and leads to premature catalyst deactivation (78). EFAL species also influence the acidity of neighboring Brønsted acid sites. Corma et al. (135) examined zeolite H-BEA, which they had exposed to several post-synthesis treatments to change the framework and extra-framework composition. On the basis of the combined reaction and characterization data, the authors concluded that some cationic EFAL species compensate the framework charge, and other condensed EFAL species block a fraction of the Brønsted acid sites, thus lowering their concentration. On the other hand, these authors suggested a synergistic effect of dispersed cationic EFAL species and framework hydroxyls to form Brønsted acid

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

261

sites of enhanced strength. A further study by the same group (136) showed that in samples with a high framework aluminum concentration, the removal of EFAL was detrimental to the catalytic performance, whereas in the samples with low framework aluminum content the catalytic activity increased as a result of the removal of EFAL. The fact that mild steaming enhances the strength of Brønsted acid sites is known from other hydrocarbon reactions, such as cracking and isomerization of alkanes and disproportionation of toluene. Selective poisoning of Brønsted acid sites with cesium has shown that only a minute amount of very strong sites is present in mildly steamed samples. However, these sites are responsible for a drastic increase in activity (118). Residual sodium also exhibits a poisoning effect on very strong Brønsted acid sites. Small amounts of sodium were found to eliminate highly acidic centers created by the interaction of EFAL with protonic sites (137). III.B.4. Silicon/Aluminum Ratio The influence of the Si/Al ratio on the catalytic performance is discussed primarily in terms of effects of changes in the concentration and acid strength of the protonic sites. The electrostatic forces induced by the presence of framework aluminum are often neglected. With increasing aluminum concentration in the framework (i.e., with lower Si/Al ratio), the total concentration of protonic acid sites increases. On the other hand, it is believed that the strengths of the acid sites decrease with increasing aluminum concentration. At high aluminum concentrations the thermal stability of the zeolites in their protonic forms is also reduced, facilitating the formation of extra-framework species (138). Examining a series of ultrastable Y zeolites, Corma et al. (121) found the catalyst with the lowest Si/ Al ratio to be best in time-on-stream behavior and TMP selectivity. With a decreasing Si/Al ratio, the ratio of stronger to weaker acid sites increased and was correlated with the alkylation/oligomerization selectivity ratio measured with the samples. The same trend was found by de Jong et al. (80), who also tested a series of ultrastable Y zeolites in a semi-batch reactor. These authors also tested a zeolite BEA with a Si/Al ratio of 15 that performed better than the Y zeolites. They postulated that a decrease of the Si/Al ratio in BEA also should lead to a superior catalyst associated with a higher Brønsted acid site concentration. Weitkamp and Traa (139) also accentuated this hypothesis. Some investigations have focused on the influence of the Si/Al ratio in zeolite BEA. Corma et al. (140) used various BEA samples synthesized with different Si/Al ratios and found a higher thermal stability towards dealumination with increasing Si/Al ratio. The most stable catalyst was also the most active one. Weitkamp et al. (141) compared the selectivities of four H-BEA samples with Si/Al ratios ranging from 12 to 90. The octane number selectivities ran through a maximum at a Si/Al ratio of 19, whereas the TMP/DMH ratio decreased

262

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

continuously with the Si/Al ratio. Loenders et al. (142) tested BEA samples with framework Si/Al ratios ranging from 13 to 77, reporting that the individual acid sites perform an identical number of catalytic turnovers before deactivation, independent of the acid site density. They claimed that the only way to enhance the activity and stability of zeolite BEA for isobutane alkylation is to increase the aluminum content of BEA nanoparticles. The only reported investigation of zeolite BEA with a Si/Al ratio lower than nine was performed by Yoo and Smirniotis (143), utilizing H-BEA synthesized with Si/Al ratios between 6 and 30. In contrast to what was postulated earlier, the zeolites exhibited a maximum in the catalytic lifetime when the Si/Al ratio was between 8.5 and 15. The hydrogen transfer activities measured separately with n-hexane as the reactant were comparable for all the samples up to a Si/Al ratio of 15. The authors concluded that the BEA with the highest aluminum content performed worse than the other samples with the same hydride transfer activity as a consequence of the lower crystallinity and micropore volume, which were inherent to the synthesis procedure for aluminum-rich zeolite BEA. In a patent assigned to Mobil Oil (144), three BEA samples, with Si/Al ratios of 7.3, 16.0, and 18.5, were compared. The only detail given about the alkylation performance was the TMP/(C8-TMP) ratio, which was observed to increase with decreasing Si/Al ratio, which is suggestive of the superiority of the material with the low Si/Al ratio. Notwithstanding some obviously contradictory results in the literature, the data summarized above can be summarized as follows: the general trend is that high aluminum contents are beneficial for the alkylation performance. This inference is supported by results from cracking experiments with zeolites having various Si/Al ratios. The bimolecular hydride transfer step is favored in materials with low Si/Al ratios (54,145,146). Thus, zeolites with low Si/Al ratios should exhibit better time-on-stream behavior than those with high Si/Al ratios. Zeolite X is the large-pore zeolite with the highest aluminum content possible. The first investigations of zeolite-catalyzed alkylation were done on this material (13,147). Weitkamp, comparing highly cerium-exchanged Y and X zeolites, found the CeX zeolite to exhibit twice the lifetime of CeY zeolite as a consequence of the higher concentration of acid sites (148). In light of these findings, it is surprising that only a small number of investigations have been devoted to this material. As the purely protonic form of zeolite X is unstable, polyvalent metal ions have to be introduced to induce acidity (Section III.B.5). A variety of di- and tri-valent metals have been examined, with and without additional ammonium exchange (149– 151). Rare earth elements, especially lanthanum, obviously are best suited to the goal, producing highly acidic and thermally stable catalysts. LaCaX zeolite has also been proposed as an excellent isobutane/ethene alkylation catalyst (152,153). Falsely, the authors

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

263

attributed the excellent performance to superacidic centers with a narrow acidity distribution. III.B.5. Metal Ions in Ion-Exchange Positions Rare earth exchanged faujasites (REHY and REUSY) are widely used in the FCC process (138). Aqueous ion exchange with rare earth salts in faujasites leads to removal of ions in the supercage only, because a bulky hydration sphere around the ions is larger than the six-membered ring of the sodalite cage, so that the ions do not enter these cages. Calcination removes the waters of hydration, and the naked cation is able to move into the sodalite cage, forming cationic polynuclear hydroxy complexes (154– 156). These species impart thermal and hydrothermal stability to the material. Rare earth exchanged zeolites exhibit considerable Brønsted acidity resulting from hydrolysis of the hydrated rare earth ions (157,158). This principle works with most polyvalent metals, and the rare earth elements induce the highest acidity and best stability (159– 161). Besides zeolite X (discussed in Section III.B.4), zeolite Y is the one that has been the subject of most investigations of cation exchange. Researchers at Sun Oil Company extensively explored rare earth exchanged Y zeolites (14). On the basis of their work described in patents (151,162– 166), it can be concluded that partially rare earth exchanged faujasites are more active catalysts than the purely protonic forms. The importance of quantitative removal of sodium from zeolite was demonstrated. Chu and Chester (119,120) compared variously modified Y zeolites; REHY zeolite (RE is rare earth) gave the highest yield and the best product quality. Dealumination of REHY zeolite did not improve its performance. USY and REUSY zeolites were both characterized by low conversion and yield, and there were not significant differences between the two. In their work on EMT and FAU zeolites, the SINTEF group (125,167) compared H- and La-exchanged samples and showed that a partially La-exchanged catalyst is superior to both fully La-exchanged and pure H-form samples. H-EMT contains the highest total number of Brønsted acid sites as measured with pyridine adsorbed at 423 K. The partially La-exchanged sample (51% exchanged) has twice as high a concentration of strong Brønsted acid sites (as measured by the pyridine retained at 823 K) as the pure H-EMT and also a lower concentration of Lewis acid sites. The increase in acid strength has been rationalized by a withdrawal of electrons from the Lewis-base framework oxygen atoms through polyvalent lanthanum cations in the sodalite cages. This electron withdrawal effect is supposed to be similar to the action of EFAL species in steamed zeolites. The abstraction of electrons weakens the O – H bond and thus increases the proton-donor strength of the OH group (156). In a patent assigned to Mobil Oil (168), rare earths exchanged into zeolite ZSM-20

264

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

(intergrowth between FAU and EMT) are claimed to improve conversion and selectivities to TMPs. REZSM-20 was also claimed to perform better than REHY. In another Mobil patent (169), REY and REUSY zeolites were compared; the REY zeolite exhibited a slightly higher alkylate quality, whereas the REUSY zeolite gave a slightly higher conversion. The subtle differences in the reports are, however, difficult to evaluate as detailed characterizations of the materials are lacking most of the time. In light of the strong influence of the concentration of Brønsted and Lewis acid sites, a judgment about which of the materials has the best properties is not possible.

III.B.6. Structure Types of Zeolites Only large-pore zeolites exhibit sufficient activity and selectivity for the alkylation reaction. Chu and Chester (119) found ZSM-5, a typical medium-pore zeolite, to be inactive under typical alkylation conditions. This observation was explained by diffusion limitations in the pores. Corma et al. (126) tested HZSM-5 and HMCM-22 samples at 323 K, finding that the ZSM-5 exhibited a very low activity with a rapid and complete deactivation and produced mainly dimethylhexanes and dimethylhexenes. The authors claimed that alkylation takes place mainly at the external surface of the zeolite, whereas dimerization, which is less sterically demanding, proceeds within the pore system. Weitkamp and Jacobs (170) found ZSM-5 and ZSM-11 to be active at temperatures above 423 K. The product distribution was very different from that of a typical alkylate; it contained much more cracked products; trimethylpentanes were absent; and considerable amounts of monomethyl isomers, n-alkanes, and cyclic hydrocarbons were present. This behavior was explained by steric restrictions that prevented the formation of highly branched carbenium ions. Reactions with the less branched or non-branched carbenium ions require higher activation energies, so that higher temperatures are necessary. MCM-22, with a larger pore volume than ZSM-5, revealed behavior intermediate between what was observed for large- and medium-pore zeolites (126). Unverricht et al. (141) also examined MCM-22; at 353 and 393 K, it was found to produce mainly cracked products and dimethylhexanes and to deactivate rapidly. MCM-36 gained considerable interest that is evidenced by the patent literature (171 –174). MCM-36 is a pillared zeolite based on the structure of MCM-22. Ideally, it should contain mesopores between layers of MCM-22 crystallites. This structure was found to be much more active and stable than MCM-22 (175). Alkane cracking experiments with zeolites having various pore dimensions evidenced the preference of monomolecular over sterically more demanding bimolecular pathways, such as hydride transfer, in small- and medium-pore zeolites (146).

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

265

In contrast to the product distributions observed for medium-pore zeolites, the product distributions observed for large-pore zeolites resemble those of typical alkylates. However, within the distribution, significant differences are observed. It is difficult to separate the influence on the alkylation reaction of the structure from the influences of other properties, mainly the acid site strength and concentration. Undisputable results may be achieved only if all but one parameters are held constant. Yoo et al. (176) compared USY, BEA, MOR, LTL, and ZSM-12 zeolites with Si/Al ratios between 20 and 34 (achieved either by direct synthesis or by various leaching techniques) and acid site densities between 0.4 and 0.7 mmol/g. These structure types were chosen because they represent three-, two-, and one-dimensional zeolites. The authors claimed that the influence of most properties influencing the performance—besides the structure—was minimized. Zeolite BEA exhibited the best time-on-stream behavior with respect to lifetime and TMP selectivity. ZSM-12 also showed a long lifetime, but it catalyzed oligomerization instead of alkylation. USY, MOR, and LTL were found to deactivate quickly, with LTL retaining a surprisingly stable TMP selectivity at low conversions. No heavy coke molecules were found in zeolites BEA and ZSM-12. The authors concluded that zeolites without periodic expansions (i.e., without larger voids that connect channels) do not allow extensive coke formation and hence deactivate relatively slowly. Unfortunately, no details about the concentrations and strengths of the acid sites in the samples were given. The finding that zeolite BEA does not produce significant amounts of coke is at variance with results of other research groups. For example, Nivarthy et al. (48) calculated values of about 14 wt% of deposit formed on H-BEA zeolites. In the aforementioned investigation by Corma et al. (126), USY, BEA, and MOR were compared with ZSM-5 and MCM-22. The three large-pore zeolites exhibited similar C8-selectivities but different behavior with time on stream. The differences were attributed to differences in the acidities of the samples. In a comparative investigation of the acidity of zeolites with low Si/Al ratios (zeolites BEA, ZSM-20, Y, and dealuminated USY), the acid strength was found to decrease in the following order: H-BEA . H-USY . H-ZSM-20 . H-Y (177). In another article by Corma et al. (178), ITQ-7, a three-dimensional large-pore zeolite, was tested as an alkylation catalyst and compared with a BEA sample of comparable Si/Al ratio and crystal size. The ratio of the selectivities to 2,2,4TMP and 2,2,3-TMP, which have the largest kinetic diameter of the TMPs, and 2,3,3-TMP and 2,3,4-TMP, which have the lowest kinetic diameter, was used as a measure of the influence of the pore structure. Lower (2,2,4-TMP þ 2,2,3-TMP)/ (2,3,3-TMP þ 2,3,4-TMP) ratios in ITQ-7 were attributed to its smaller pore diameter. The bulky isomers have more spacious transition states, so that their formation in narrow pores is hindered; moreover, their diffusion is slower. The hydride transfer activity, estimated by the dimethylhexane/dimethylhexene ratio,

266

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

was found to be lower in ITQ-7 than in H-BEA. This observation was also explained by the smaller pore diameter, because the acidities of the two different zeolites were found to be similar. Nivarthy et al. (179) compared the three largepore zeolites H-BEA, H-FAU, and H-EMT; the lifetimes of the zeolites were found to depend on the Brønsted acid site concentration. H-BEA, with the lowest Brønsted acid site concentration, was characterized by the shortest lifetime and H-EMT, with the highest concentration, the longest lifetime. Significant differences were observed in the TMP distribution. H-BEA exhibited a very high 2,2,4-TMP selectivity, which was attributed to a lower rate of hydride transfer vs. isomerization of the precursor carbenium ions. An exceptionally high 2,2,4-TMP selectivity is characteristic of zeolite BEA. Although with most other zeolites the selectivities vary depending on the conditions employed, BEA always produces high yields of 2,2,4-TMP. The research group at SINTEF (123,124,180) dedicated a series of papers to the examination of FAU and EMT zeolites, comparing them in their Hand La-exchanged form with and without dealumination. EMT was always superior to FAU. The alkylate yield, expressed as mass of alkylate produced divided by the catalyst mass, was higher for the EMT samples. EMT also produced a greater amount of trimethylpentanes than the FAU samples. The differences between the two materials were discussed in terms of the slightly larger supercage in EMT, which is claimed to reduce the steric constraints on the bulky transition states for hydride transfer, and in terms of acidity, with EMT samples exhibiting a higher concentration of Brønsted acid sites retaining pyridine at high temperatures. A comparison of La-EMT, La-FAU, and La-BEA revealed that the La-BEA performed worse than the two other materials, both in terms of alkylate yield and selectivity (167), but the lack of information about the acidity of the samples prevents a detailed evaluation of this report. Recently, mesoporous aluminosilicates with strong acidity and high hydrothermal stability have been synthesized via self-assembly of aluminosilicate nanoclusters with templating micelles. The materials were found to contain both micro- and mesopores, and the pore walls consist of primary and secondary building units, which might be responsible for the acidity and stability (181). These materials were tested in isobutane/n-butene alkylation at 298 K, showing a similar time-on-stream behavior to that of zeolite BEA. No details of the product distribution were given. The patent literature discloses alkylation performances of several additional structure types. A Mobil patent (182) describes the use of VTM-A, a pillared titanosilicate of the MCM-27 family. The catalyst produced about 80 wt% of octanes under relatively mild conditions (OSV ¼ 0:05 h21, P=O ratio ¼ 20). A number of patents describe the use of MCM-36. MCM-49, which is closely related to MCM-22, has also been tested as an alkylation catalyst. In general,

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

267

these materials require a relatively high reaction temperature to be sufficiently active, which inevitably leads to high cracking and high DMH selectivities (172, 183 –187).

III.C. Other Solid Acids III.C.1. Sulfated Zirconia and Related Materials A variety of solid acids besides zeolites have been tested as alkylation catalysts. Sulfated zirconia and related materials have drawn considerable attention because of what was initially thought to be their superacidic nature and their welldemonstrated ability to isomerize short linear alkanes at temperatures below 423 K. Corma et al. (188) compared sulfated zirconia and zeolite BEA at reaction temperatures of 273 and 323 K in isobutane/2-butene alkylation. While BEA catalyzed mainly dimerization at 273 K, the sulfated zirconia exhibited a high selectivity to TMPs. At 323 K, on the other hand, zeolite BEA produced more TMPs than sulfated zirconia, which under these conditions produced mainly cracked products with 65 wt% selectivity. The TMP/DMH ratio was always higher for the sulfated zirconia sample. These distinctive differences in the product distribution were attributed to the much stronger acid sites in sulfated zirconia than in zeolite BEA, but today one would question this suggestion because of evidence that the sulfated zirconia catalyst is not strongly acidic, being active for alkane isomerization because of a combination of acidic character and redox properties that help initiate hydrocarbon conversions (189). The time-onstream behavior was more favorable for BEA, which deactivated at a lower rate than sulfated zirconia. Whether differences in the adsorption of the feed and product molecules influenced the performance was not discussed. In a subsequent publication (22), two sulfated zirconia samples with different sulfate loadings were reported as alkylation catalysts with isobutane/2-butene feed at temperatures between 263 and 323 K. The sample with the higher sulfur loading was slightly more active in the initial reaction phase, and the rates of deactivation were similar for the two catalysts. The alkylation/cracking ratio increased with decreasing reaction temperature. 2,2,4-TMP was the dominant octane isomer under all conditions and less dimethylhexanes and octenes were produced than with the zeolitic catalysts. In another investigation by the same authors, sulfate-doped ZrO2, TiO2, and SnO2 were prepared by various sulfation and activation procedures. The acidity decreased in the order SO22 4 / 22 /TiO . SO /SnO , which was reflected in the cracking activities ZrO2 . SO22 4 2 4 2 of the samples. All the oxides showed considerable sensitivity towards the modification procedure, each with a different optimum. All the samples deactivated

268

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

rapidly and additionally lost significant fractions of the sulfur that was originally present (190). Satoh et al. (191) also compared several sulfated metal oxide catalysts, which were tested with gas-phase reactants at 273 K. This is an inappropriate procedure, however, because most of the products are liquid under these conditions and remain in the catalyst pores. The authors of an investigation with pulsed gasphase reactants for alkylation with sulfated zirconia catalyst also concluded that at temperatures below 323 K the TMPs could not desorb from the pores. Raising the temperature to just 373 K led to dehydrogenation of isobutane (192,193). Other investigations of alkylation of gas-phase reactants with sulfated zirconia were reported by Das and Chakrabarty (194) and Guo et al. (195,196). Working with liquid-phase reactants and employing relatively mild conditions, Xiao et al. (197) were able to extend the lifetime of a sulfated zirconia catalyst to more than 70 h. In the initial phase, the catalyst produced 80 wt% cracked products, but the value fell to less than 20 wt% after 30 h TOS with an increase in TMP selectivity to more than 60 wt%. Within the TMPs, 2,2,4-TMP selectivities became higher than 60 wt%. Platinum-promoted sulfated zirconia and tungstated zirconia were found to be much less active alkylation catalysts. An interesting variation on sulfated metal oxide type catalysts was presented by Sun et al. (198), who impregnated a dealuminated zeolite BEA with titanium and iron salts and subsequently sulfated the material. The samples exhibited a better time-on-stream behavior in the isobutane/1-butene alkylation (the reaction temperature was not given) than H-BEA and a mixture of sulfated zirconia and H-BEA. The product distribution was also better for the sulfated metal oxideimpregnated BEA samples. These results were explained by the higher concentration of strong Brønsted acid sites of the composite materials than in H-BEA. III.C.2. Heteropolyacids Heteropolyacids are strongly acidic non-porous solids. Salts of these acids containing large cations, such as Csþ, Kþ, Rbþ, and NHþ 4 , exhibit surface areas in the order of 150 m2/g. Supporting heteropolyacids on highly porous carriers provide a method to increase the surface area. This was done by Blasco et al. (199), who used 12-tungstophosphoric acid on silica, on a high-surface-area amorphous aluminosilicate, and on all-silica mesoporous MCM-41. These materials were tested for isobutane/2-butene alkylation at 306 K. The acid supported on silica performed best, with high initial activity and selectivity to trimethylpentanes. Heteropolyacids supported on the aluminosilicate interacted strongly with the support, which decreased the acidity, thus leading to lower activity and selectivity. Heteropolyacids on MCM-41 were observed to partially block the pores of the support, so that a fraction of the acid was inaccessible

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

269

to the reactants. This effect of pore blocking could be decreased by the use of a MCM-41 with a larger pore diameter. All the materials deactivated rapidly. Gayraud et al. (200) tested potassium salts of 12-tungstophosphoric acid with various potassium loadings to modify acidity and porosity. The samples were tested at sub- and super-critical conditions. Samples with high potassium content exhibited better time-on-stream behavior and selectivities than others. The authors claimed that high acid site density was detrimental for the alkylation reaction, leading to increased oligomerization activity. This conclusion contradicts the generally accepted notion that high acid site densities enhance the alkylation activity. The results can be better explained by the decrease in surface area with decreasing potassium content, which was found to vary from 156 to 50 m2/g. Cesium salts of 12-tungstophosphoric acid have been compared to the pure acid and to a sulfated zirconia sample for isobutane/1-butene alkylation at room temperature. The salt was found to be much more active than either the acid or sulfated zirconia (201). Heteropolyacids have also been supported on sulfated zirconia catalysts. The combination was found to be superior to heteropolyacid supported on pure zirconia and on zirconia and other supports that had been treated with a variety of mineral acids (202). Solutions of heteropolyacids (containing phosphorus or silicon) in acetic acid were tested as alkylation catalysts at 323 K by Zhao et al. (203). The system was sensitive to the heteropolyacid/acetic acid ratio and the amount of crystalline water. As observed in the alkylation with conventional liquid acids, a polymer was formed, which enhanced the catalytic activity. III.C.3. Acidic Organic Polymers Nafion-H, a perfluorinated sulfonic acid resin, is another strongly acidic solid that has been explored as alkylation catalyst. Rørvik et al. (204) examined unsupported Nafion-H with a nominal surface area of 0.2 m2/g (surface area of a swellable polymer is difficult to define) in isobutane/2-butene alkylation at 353 K and compared it with a CeY zeolite. The zeolite gave a better alkylate and higher conversion than Nafion-H, which produced significant amounts of octenes and heavy-end products. The low surface area of the resin and questions about the accessibility of the sulfonic acid groups probably make the comparison inadequate. To increase the surface area, the resin can be supported on porous carriers, or it can be directly incorporated into silica by a sol –gel preparation technique. Both methods have been used by Botella et al. (205), who compared several composite Nafion/silica samples with varying surface areas and Nafion loadings for isobutane/2-butene alkylation at 353 K. Furthermore, supported and unsupported Nafion samples were used. As expected, the unsupported resin with its low

270

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

surface area performed only poorly. The composite materials showed an optimum performance at intermediate surface areas, which was explained by an interaction between the sulfonic groups of the resin and the silanol groups of the silica, decreasing the acid strength of the resin. The supported resin showed activity and selectivity similar to that of the composite material of the same Nafion content. Changes in the temperature from 305 to 353 K showed that the material produces oligomers at low temperatures and saturated products at higher temperatures. This led the authors to claim the catalyst to have an acid strength in the range characteristic of zeolite BEA and lower than that of sulfated zirconia. Such a ranking, however, seems obsolete because sulfated zirconia has been shown to exhibit only a moderate acid strength (Section III.C.1). III.C.4. Supported Metal Halides Supported metal halides gained considerable attention as candidate alkylation catalysts, and at least three companies tested them in pilot plants (206). Chlorinated alumina, obtained by reacting alumina with hydrogen chloride, is a highly Brønsted-acidic porous solid. This material is related to the FriedelCrafts catalyst aluminum chloride, which was one of the first catalysts tested in alkylation. Similar catalysts are used in commercial alkane isomerization plants. A series of chlorinated alumina samples modified with Liþ and Naþ ions was prepared and tested by Clet et al. (207) for isobutane/2-butene alkylation at 273 K. The purpose of the cation addition was to moderate the acidity of the material. It was shown that the cations prevent excessive cracking, and the time-on-stream behavior is superior to that of the unmodified sample. The improvement was attributed to a selective annihilation of very strong acid sites by the cations. The degradation of 2,2,4-TMP on these catalysts at 273 K was also investigated. 2,2,4-TMP was found to be surprisingly reactive under these conditions and gave a product resembling an alkylate—but with more dimethylhexanes and light- and heavy-end products. Emphasis was placed on the explanation of the rearrangement steps for producing dimethylhexanes and cracked products, but the initiation of the 2,2,4-TMP degradation was not discussed. These catalysts are also described in a patent application (208). A similar type of catalyst including a supported noble metal for regeneration was described extensively in a series of patents assigned to UOP (209 –214). The catalysts were prepared by the sublimation of metal halides, especially aluminum chloride and boron trifluoride, onto an alumina carrier modified with alkali or rare earth-alkali metal ions. The noble metal was preferably deposited in an eggshell concentration profile. An earlier patent assigned to Texaco (215) describes the use of chlorinated alumina in the isobutane alkylation with higher alkenes, especially hexenes. TMPs were supposed to form via self-alkylation. Fluorinated alumina and silica samples were also tested in isobutane alkylation,

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

271

but were found to produce mainly heavy-end products under the conditions employed (216). Patents assigned to Mobil (217) describe the use of boron trifluoride supported on several porous carriers. BF3 supported on silica was found to exhibit a slightly higher performance with added water in the alkylation of a mixed alkene feed at 273 K. It was also shown that self-alkylation activity was considerably lower than that with HF as catalyst. Another patent (218) describes the use of a pillared layered silicate, MCM-25, promoted with BF3 to give a high-quality alkylate at temperatures of about 273 K. BF3 was also supported on zeolite BEA, with adsorbed water still present (219). This composite catalyst exhibited low butene isomerization activity, which was evident from the inferior results obtained with 1-butene. At low reaction temperatures, the product quality was superior to that of HF alkylate. Triflic acid has also been supported on a porous silica carrier (220). The authors emphasized the importance of a strong interaction between the acid and the support to prevent leaching of the acid. In pulsed liquid-phase isobutane/ 1-butene alkylation experiments at 298 K, the catalysts produced a very highquality alkylate, made up almost exclusively of isooctanes. With silanol groups on the silica surface or with added water, triflic acid was found to form a monohydrate that was firmly grafted to the silica surface.

III.D. The Influence of Process Conditions The choice of appropriate reaction conditions is crucial for optimized performance in alkylation. The most important parameters are the reaction temperature, the feed alkane/alkene ratio, the alkene space velocity, the alkene feed composition, and the reactor design. Changing these parameters will induce similar effects for any alkylation catalyst, but the sensitivity to changes varies from catalyst to catalyst. Table II is a summary of the most important parameters employed in industrial operations for different acids. The values given for zeolites represent best estimates of data available from laboratory and pilot-scale experiments. Two points are emphasized: (i) zeolites can be successfully operated at the same or higher severities (with respect to P/O (feed alkane/alkene) ratio and OSV (alkene space velocity)) than the liquid acids; (ii) the productivities of zeolite catalysts (i.e., the total amount of alkylate produced per mass of catalyst) are roughly the same as of that of sulfuric acid. If the intrinsic activities of zeolites (which have 0.5 –3 mmol of acid sites per gram) are compared with that of sulfuric acid (which has 20 mmol of acid sites per gram), zeolites outperform sulfuric acid. Nevertheless, the price of a zeolite catalyst and the high costs of

272

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295 TABLE II Typical values of important process parameters for alkylation Catalyst HF

H2SO4

Zeolites

289–313

277–291

323–373

Feed alkane/alkene ratio (mol/mol)

11–14

7–10

6–15

Alkene space velocity (kgAlkene/kgAcid h)

0.1–0.6

0.03–0.2

0.2 –1.0

Exit acid strength (wt%)

83–92

89–93

–

Acid per reaction volume (vol%)

25–80

40–60

20 –30

1000–2500

6–18

4–10

Reaction temperature (K)

Catalyst productivity (kgAlkylate/kgAcid)

The numbers for the liquid acids are taken from Refs. (12,23,221). As zeolites are not used in industrial alkylation process, the given values represent the judgment of the authors extracted from laboratory and pilot scale data obtained in a slurry reactor.

effective regeneration set high hurdles for a competition with sulfuric acidcatalyzed processes. III.D.1. Reaction Temperature The reaction temperature affects both the chemistry of alkylation through the activation energies of the individual reaction steps and the solubility/adsorption and diffusion of products and reactants. With sulfuric acid, the viscosity is also strongly influenced by the temperature. Dispersion effects, such as too low interfacial areas between acid and hydrocarbons, thus set the lower temperature limit with sulfuric acid. Temperatures below 277 K inhibit the separation of acid from the hydrocarbon phase and lead to acid carryover from the acid settler downstream of the reactor. At temperatures exceeding 291 K, polymerization reactions dominate, leading to increased acid consumption and low octane numbers (12). The higher solubility of isobutane in HF and its lower viscosity allow higher isobutane consumption rates to be applied with HF. Therefore, HF can be operated at higher temperatures, resulting in higher reaction rates. This operation also reduces the refrigeration costs. Instead of a true refrigeration system, cooling water can be used. Nevertheless, the product quality is higher when the operation is at the lower temperature limit. With increasing temperature, the rates of side reactions increase. Oligomerization/cracking is of greater importance at higher temperatures, reducing the selectivity to trimethylpentanes.

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

273

Zeolites, in principle, operate at significantly higher reaction temperatures than the liquid acids. The need for higher temperatures is attributed to the lower acid strengths of zeolites or the lack of solvation, resulting in higher activation energies for the individual reaction steps. Efficient mobility in the zeolite micropores also requires higher temperatures. The optimum temperature is in the range 323 – 373 K, with the exact value likely depending on the individual sample. The problem of the optimum reaction temperature is often overlooked in test for comparison of various catalysts. Testing catalysts at sub-optimum temperatures leads to false conclusions about the true alkylation performance. Nivarthy et al. (48) found an optimum for zeolite H-BEA at 348 K, at which temperature the highest octane selectivity and the highest TMP/DMH ratio were achieved. At lower temperatures, oligomerization dominated, and at higher temperatures, cracking reactions dominated. Kirsch et al. (14) tested various rare earth exchanged Y zeolites at temperatures 298– 373 K. A sample with 0.2 wt% residual sodium had a optimum temperature around 313 K, and a sample with 1.0 wt% sodium performed best at 353 K. Taylor and Sherwood (222) examined the influence of several process parameters on the performance of a USY zeolite. The catalyst was tested at 311, 339, and 367 K. The TMP selectivity decreased steadily with increasing temperature, and the longest lifetime was achieved at 339 K. Pronounced effects on the product selectivities were also observed by Corma et al. (140), who used a H-BEA catalyst at 323 and 353 K. At the higher temperature, the activity was higher, as indicated by the increased conversion. The selectivity to cracked products increased drastically, and the C9þ selectivity also increased with temperature. Within the TMP fraction, 2,2,4-TMP increased significantly with temperature. Feller et al. (89) performed a detailed investigation of the influence of the reaction temperature in the range of 313 –403 K on the performance of a LaX zeolite. The catalyst lifetime was found to depend strongly on the reaction temperature, with an optimum at 348 K. The product quality was highest at low temperatures; with increasing temperatures, increasingly more cracked and heavy compounds were produced. The TMP/DMH ratio declined with temperature. The selectivity phenomena can be explained by the relative rates of the individual reaction steps. b-Scission (and presumably also alkene addition) are characterized by higher activation energies than hydride transfer. Increases in temperature consequently lead to higher relative rates of secondary products from multiple alkylation and cracking. Cracked products are favored over multiple alkylation products, because the activation energy is higher for b-scission than for alkene addition, which is the (exothermic) reverse reaction. The bad performance of zeolites at low reaction temperatures is most likely a consequence of the hindered diffusion of bulky molecules under such conditions. The catalyst will be prematurely deactivated by pore blocking. These diffusion

274

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

problems are the reason why several research groups tried to overcome a buildup of heavy molecules in the catalyst pores by employing supercritical conditions. A supercritical reaction medium should combine liquid-like density with high oligomer solubilities and gas-like transport properties. Under such conditions, the bulky molecules that otherwise would deactivate the catalyst are supposedly more efficiently removed from the catalyst pores. The feed itself can be employed as a supercritical medium, but the critical point of isobutane is 408 K and 36.5 bar. Performing the alkylation reaction under these conditions leads to excessive cracking. The catalyst stays active for longer times than in the conventional operation, but it produces cracked products and especially substantial amounts of alkeneic products (130,131). To overcome the problems associated with the high critical temperature of isobutane, carbon dioxide has been used as a diluent to reduce the critical temperature. The results presented by Clark and Subramaniam (223) show that a stable conversion can indeed be maintained with a 10-fold excess of carbon dioxide at 323 K and 155 bar. However, the conversion was very low (, 20 wt%), and the product contained only minor amounts of trimethylpentanes. Similar results were reported by Santana and Akgerman (224). Ginosar et al. (225), testing a variety of supercritical solvents and a variety of solid acids, came to the conclusion that working under supercritical conditions generally does not improve the alkylation performance. A temperature-programmed-oxidation analysis of samples coked under supercritical conditions revealed that the carbonaceous deposits are very similar in concentration and oxidizability to coke produced under liquid-phase reaction conditions. The slight changes were related to a smaller amount of coke on the outer surface of the zeolite (226).

III.D.2. Alkane/Alkene Ratio and Alkene Space Velocity Rates of reaction are influenced by reactant concentrations, determined by the feed composition, and temperature. The crucial parameter that determines a high alkylate quality and a low acid consumption is the ratio of rates of hydride transfer and oligomerization. This ratio should be as high as possible. Increasing the isobutane concentration minimizes undesired reactions and acid consumption by increasing the probability that the carbenium ion will react with an isobutane molecule to form the desired product via hydride transfer rather than undergoing oligomerization with other alkenes. The ratio of rates of hydride transfer to oligomerization is primarily influenced by two process parameters: the feed alkane/alkene (P/O) ratio and the alkene space velocity (OSV, which is approximately proportional to the reciprocal of the average residence time). The P/O ratio determines the concentration of isobutane in the reactor and thereby the rate

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

275

of hydride transfer. The P/O ratio also sets the product concentration, which affects the rates of the product degradation reactions. Another point might be of importance, although no quantitative data are available to assess it. Ideally, feed entering the reactor should be instantaneously mixed with the acid. The conversion of the alkene in the reactor is usually complete, so that the internal P/O ratio might be 1000:1 or even higher. In the case of incomplete mixing, the alkene concentration will be higher at some positions in the reactor and consequently lead to higher rates of oligomerization and acid consumption than would occur if mixing were perfect. With high feed P/O ratios, the detriment of incomplete mixing will be minimized. Thus, increasing the P/O ratio increases alkylate quality and yield and decreases acid consumption. On the other hand, at high P/O ratios, more isobutane has to be recycled, which leads to increased separation costs. A balance has to be found to optimize the economic performance of the unit. The OSV determines the production rate of alkylate, so that high OSV would be economically favored, but this is limited by high acid consumption, low octane number, and high rates of formation of heavy-end products at high values of OSV. When the catalyst is sulfuric acid, more esters are introduced into the products than when the catalyst is hydrofluoric acid, and these products corrode down-stream equipment (221). As a first approximation for sulfuric acidcatalyzed n-butene alkylation, an increase in OSV of 0.1 vol/(vol £ h) leads to a decrease in RON of about 1, and an increase in the P/O from 8 to 9 leads to an increase in RON of 0.15 (12). The above-mentioned higher solubility of isobutane in HF allows higher space velocities in HF plants, although they are usually operated at higher P/O ratios. In principle, the same rules hold true when zeolitic alkylation catalysts are used. A detailed study of the influence of PO and OSV on the performance of zeolite H-BEA in a backmix reactor was reported by de Jong et al. (80). The authors developed a simple model of the kinetics, which predicted catalyst lifetimes as a function of P/O and OSV. Catalyst lifetime (which is equivalent to the catalyst productivity, the reciprocal of acid consumption) increased with increasing P/O ratio and decreasing OSV. Furthermore, the authors persuasively demonstrated the superiority of a backmix reactor over a plug flow reactor. Qualitatively similar results were obtained by Taylor and Sherwood (222) employing a USY zeolite catalyst in a backmix reactor. The authors stressed the detrimental effect of unreacted alkene on the catalyst lifetime and product quality. Feller et al. (89) tested LaX zeolites in a backmix reactor and found the catalyst productivity to be nearly independent of the OSV within the examined OSV range. At higher values of OSV, the catalyst life was shorter, but in this shorter time the same total amount of product was produced. The P/O ratio had only a moderate influence on the catalyst performance.

276

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

III.D.3. Alkene Feed Composition Propene, 1-butene, 2-butene, isobutylene, and normal- and isopentenes can be used as feedstocks in alkylation. Depending on the catalyst, they give different alkylate qualities and yields with differing acid consumptions. Only linear butenes give a fairly low acid consumption in sulfuric acid-catalyzed processes. All the other alkenes lead up to three times higher acid consumption (12). Hydrofluoric acid consumption is nearly independent of the feed alkene (227). The low double bond isomerization activity of HF leads to higher production of dimethylhexanes when 1-butene is the feed alkene. The high self-alkylation activity of HF is responsible for a high fraction of TMPs in the alkylate when alkenes other than butenes are used. Table III provides a comparison of alkylate compositions for both the liquid acid-catalyzed reactions with various feed alkenes. The data show that H2SO4 produces a better alkylate with 1-butene, whereas HF gives better results with propene or isobutylene. The products from 2-butene and also from pentenes (not shown in Table III) are nearly the same with either acid. Zeolites have also been tested with feed alkenes other than butenes. Daage and Fajula (229) reported an investigation of isobutane/propene alkylation with a CeY zeolite with 13C-labeled feed molecules. The products could be grouped into those formed by three classes of reactions: dimerization leading to C6 products, alkylation leading to C7 products, and self-alkylation leading to C8 and also to C7 products. Investigations by Guisnet et al. (69,93) comparing 2-butene and propene as feed alkenes with a USY zeolite catalyst gave similar results. Selfalkylation was slower by a factor of two than the alkylation of isobutane with propene and faster by a factor of two than the dimerization of propene. The conversion in isobutane/propene alkylation was considerably lower than in isobutane/2-butene alkylation. A comparative study of zeolite H-BEA-catalyzed alkylation of isobutane with 2-butene, propene, and ethene was published by Nivarthy et al. (230). The reactivity of the alkenes decreased in the order 2butene . propene . ethene. Here, the products could also be grouped into those formed by dimerization, alkylation, and self-alkylation. Dimerization is especially important with ethene, forming n-butenes, which react in the normal way to give octanes. The distribution within the C8 fraction was almost the same when ethene was used instead of 2-butene. Ethene exhibits a low reactivity because it can form only primary carbenium ions, which requires high activation energies. Ethene is reactive with AlCl3/HCl, but not with sulfuric acid or hydrofluoric acid. Early investigations of zeolite REHX as a catalyst were done with ethene as the feed alkene (13). At 300 K, the product was mainly hexanes, whereas at temperatures as high as 422 K, isopentane dominated, with hexanes and octanes being the other main products. KTI developed a process which utilizes ethene

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

277

TABLE III Compositions of alkylates obtained with various feed alkenes and various acid catalysts (50,228) Component (wt%)

Feed alkene and employed acid catalyst Propene

Isobutylene

2-Butene

1-Butene

HF

H2SO4

HF

H2SO4

HF

H2SO4

HF

H2SO4

1.0

3.8

0.5

10.0

0.3

4.2

1.0

4.7

C5 Isopentane C6 Dimethylbutanes

0.3

0.8 4.2

Methylheptanes

0

0.7 5.2

0.2

0.8 4.6

0.2

4.4 0.3

C7 2,3-Dimethylpentane

29.5

50.4

2.0

2.6

1.5

1.4

1.2

1.5

2,4-Dimethylpentane

14.3

20.8

0

3.9

0

2.4

0

2.6

2,2,4-Trimethylpentane

36.3

4.4

66.2

28.7

48.6

30.6

38.5

30.5

2,2,3-Trimethylpentane

0

2,3,4-Trimethylpentane

7.5

C8

0

1.9

12.8 3.7

0.9

22.2 23.1

7.1

19.1 41.6

12.9

39.1

2,3,3-Trimethylpentane

4.0

9.7

Dimethylhexanes

3.2

1.7

3.4

9.5

6.9

9.0

22.1

11.0

C9þ products

3.7

11.0

5.3

17.1

4.1

6.3

5.7

6.2

from FCC off-gases to produce alkylate with a zeolite catalyst having a “dimerization function” (231). The catalyst disclosed consists of a RECaX zeolite impregnated with palladium as the “dimerization function” (152). Operated at temperatures in the range of 323 –343 K, the catalyst produces a high yield of octanes and almost no hexanes (153,232,233). Chlorided alumina was also tested as a catalyst for isobutane/ethene alkylation at temperatures between 273 and 373 K. Catalyst stability was better at low temperatures than at high temperatures. Hexanes constituted the main product fraction, especially at high P/O ratios (234). Thermodynamically, hexanes are strongly favored over octanes and higher molecular weight products (235).

278

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

IV. Industrial Processes and Process Developments This section is a review of alkylation process technology. The processes in which liquid acids are used are all mature technologies and described briefly here. Information about process developments with solid acid catalysts is also presented.

IV.A. Liquid Acid-Catalyzed Processes All the processes require intensive mixing of acid and hydrocarbon phases to form emulsions. The droplets have to be small enough to give a sufficiently large phase boundary area, but they also have to ensure a quick separation in the settler downstream of the reactor to prevent degradation reactions. Because of the high viscosity of sulfuric acid, mixing is more of a problem with sulfuric acid than with hydrofluoric acid. In all sulfuric acid-catalyzed processes impellers have to be employed. In hydrofluoric acid processes the hydrocarbons are typically injected through nozzles, which are sufficient for effective dispersion. Because the alkylation reaction is exothermic, a considerable amount of process heat has to be removed. As HF-catalyzed processes operate at temperatures between 289 and 313 K, the reactors can be cooled with water. H2SO4-catalyzed processes operate at temperatures between 277 and 291 K (Table II) and therefore require more complex cooling systems, which typically utilize the processed hydrocarbon stream itself. The feed hydrocarbons, which come from the FCC or from the etherification unit of a petroleum refinery, usually have to be treated before entering the alkylation unit. They contain water, butadienes, and sulfur- and nitrogencontaining compounds and—when coming from an etherification unit—traces of oxygenates. The general treatment of the hydrocarbon stream leaving the alkylation reactor is similar in all processes. First, the acid and hydrocarbon phases have to be separated in a settler. The hydrocarbon stream is fractionated in one or more columns to separate the alkylate from recycle isobutane as well as from propane, n-butane, and (sometimes) isopentane. Because HF processes operate at higher isobutane/alkene ratios than H2SO4 processes, they require larger separation units. All hydrocarbon streams have to be treated to remove impurity acids and esters. IV.A.1. Sulfuric Acid-Catalyzed Processes Two licensors now offer sulfuric acid alkylation units. The one with the higher market share is Stratco, with its Effluent Refrigerated Sulfuric Acid Alkylation

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

279

Process (12). The reactor is a horizontal pressure vessel called Contactore and containing an inner circulation tube, a heat exchanger tube bundle to remove the heat of reaction, and a mixing impeller in one end. The hydrocarbon feed and recycle acid enter on the suction side of the impeller inside the circulation tube. This design ensures the formation of a fine acid-continuous emulsion. The high circulation rate prevents significant temperature differences within the reactor. The reactor is shown schematically in Fig. 11. A portion of the emulsion flows to the settler, where the hydrocarbon phase is separated from the acid phase. The hydrocarbon phase is expanded and partially evaporated. The cold two-phase hydrocarbon effluent is passed through the cooling coils of the contactor reactor and takes up the heat of reaction as it undergoes evaporation. To increase the efficiency of the cooling system, propane is co-fed to the reactor. The gaseous hydrocarbons are sent to a refrigerant compressor and separated from excess propane in a depropanizer column. The acid leaving the settler is recycled into the reactor, with a small stream of fresh acid continuously replacing the equivalent stream of spent acid. To increase product quality and reduce acid consumption, the reactor can be staged with respect to the acid flow; the acid can be passed through up to four contactor reactors with each reactor being fed with fresh hydrocarbons. The spent acid strength is maintained at about 90 wt% H2SO4. The molar isobutane/alkene feed ratio ranges from 7:1 to 10:1. Typical operating alkene space velocities (LHSV) range from 0.2 to 0.6 h21 (corresponding to WHSVs from 0.06 to 0.19 h21). The optimum reaction temperatures range from 279 to 283 K, but some units are operated at temperatures up to 291 K.

Fig. 11. Stratcow Contactore reactor used in sulfuric acid-catalyzed alkylation (12).

280

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

The second licensor of sulfuric acid-catalyzed alkylation processes is ExxonMobil, with the stirred auto-refrigerated process (221), a technology formerly licensed by Kellogg. In this process, the reactor consists of a large horizontal vessel divided into a series of reaction zones, each equipped with a stirrer (Fig. 12). The alkene feed is premixed with recycle isobutane and fed in parallel to all mixing zones, and the acid and additional isobutane enter only the first zone and cascade internally to the other zones. The heat of the reaction is removed by evaporating isobutane plus added propane from the reaction zones. Thus, no cooling coils are necessary in this type of process. To minimize any increase in temperature along the reaction zones, the vessel is divided into two pressure stages, with the second stage operating at a lower pressure to decrease the boiling point of the hydrocarbon mixture. The vapors are sent to the refrigeration section, where they are compressed, condensed, and returned to the reactor as recycle refrigerant. To prevent a buildup of propane in the refrigeration section, a slipstream is withdrawn and separated in a depropanizer. The liquid stream is separated in a settler, from which the acid phase is recycled into the reactor. Because of its large reactor volume, the auto-refrigerated process can operate at very low alkene space velocities of about 0.1 h21 LHSV (WHSV ca. 0.03 h21). This design helps in increasing the octane number of the product and lowering acid consumption. The reaction temperature is maintained at about 278 K to minimize side reactions. Spent acid is withdrawn as 90 – 92 wt% acid. The isobutane concentration in the hydrocarbon phase is kept between 50 and 70 vol%.

Fig. 12. ExxonMobil auto-refrigerated alkylation process. Adapted from Ref. (221).

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

281

Stratco offers a process called Alkysafee, proposing the conversion of an existing HF alkylation unit to use H2SO4 for approximately the same cost as installing an effective HF mitigation system. The process reuses the reaction and distillation sections from the existing unit. Refrigeration is carried out with a closed-loop packaged propane refrigeration section. Emulsion pumps and static mixers have to be installed to provide the required mixing. Stratco claims the production of similar or even increased quality alkylate as compared to that of the former HF plant. The process flow of the converted unit is similar to that of the time tank units built between 1938 and 1958 (15,227). In this process, the hydrocarbons are brought in contact with the acid in a large non-cooled pipe close to the entrance of a centrifugal pump, which provides mixing and emulsification. The heat of reaction is removed in a chiller (utilizing propane as a refrigerant) situated downstream of the pump. The emulsion then flows into the time tank, which is a large vertical vessel containing baffles. Although these units produced a highoctane alkylate, they were successively shut down or changed over to different types because of high costs of operation. IV.A.2. Hydrofluoric Acid-Catalyzed Processes ConocoPhillips offers a process using a non-cooled riser-type reactor (Fig. 13). The hydrocarbon mixture is introduced through nozzles at the bottom and along the length of the riser (236). The acid is injected at the bottom. The reactor contains perforated trays, which help to maintain a high dispersion of the hydrocarbons in the acid phase. The reaction mixture enters the settler, from which the acid is withdrawn at the bottom and then cooled in a heat exchanger with cooling water to remove the heat of the reaction. The cold acid is then fed back into the reactor. The acid flow is driven by gravity. The hydrocarbons in the settler are routed to the fractionation section, with an overhead stream of propane and HF, a side stream of isobutane, another side stream of n-butane, and a bottoms stream of alkylate leaving the section. The HF is separated from propane in an HF stripper. The acid is regenerated by distillation to remove ASO and water. Typical process parameters are temperatures of about 297 K, molar isobutane/alkene ratios of about 14 –15, and acid concentrations of 86– 92 wt%. At the heart of the UOP HF alkylation unit is a vertical reactor-heat exchanger, shown in Fig. 14. The isobutane –alkene mixture enters the shell of the reactor through several nozzles, and HF enters at the bottom of the reactor. The reaction heat is removed by cooling water, which flows through cooling coils inside the reactor. After phase separation in the settler, the acid is recycled to the reactor. The hydrocarbon phase together with a slipstream of used acid and makeup isobutane is sent to the “isostripper”, where the alkylate product, n-butane, and isobutane are separated. The isobutane is recycled to the reactor. During normal

282

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

Fig. 13. ConocoPhillips HF alkylation reactor (236).

operation, the acid is distilled with the product, so that no external regeneration is necessary. An additional acid regeneration column is still needed, however, for startup, or when feed contamination occurs. As a reaction to the pressure imposed on refiners operating HF processes, licensors developed safety systems to reduce the inherent risks. Among the mitigation systems are high-volume water sprays to “knock down” an acid cloud, a low acid inventory, and a rapid acid de-inventory system. HF modifiers, which reduce the volatility and the aerosol-forming tendency of HF, are also offered. ConocoPhillips together with Mobil developed an HF modifier technology named ReVape to reduce the volatility of the acid. It is claimed that a 60 –90% reduction in airborne acid release relative to that of the unmodified acid is

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

283

Fig. 14. UOP HF alkylation reactor.

achieved. The modifier does not undergo a chemical reaction with the acid. The additive is separated from the alkylate by extraction and recycled within the alkylation unit. Furthermore, the ASO has to be separated from the additive. The additive most likely is based upon sulfones. ConocoPhillips claims that when using the additive the acid concentration can be lowered to 60 wt%. UOP in a joint venture with ChevronTexaco developed an additive technology named Alkade. The additive is based on HF salts of amines, which form liquid “onium” polyhydrogen fluoride complexes with HF, reducing the vapor pressure of the catalyst; 65% to more than 80% aerosol reduction is claimed with this additive. As in the ReVape technology, additional separation columns have to be installed. Both additives are claimed to increase the product octane number, especially when propene, isobutylene, and pentenes are employed in the feedstock.

IV.B. Solid Acid-Catalyzed Processes Processes based on solid acids are not operated on an industrial scale. However, several companies are developing processes or already offering technology for licensing. The overall process scheme is similar to that of a liquid acid-based process, except for the regeneration section, which is necessary with all solid acid

284

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

catalysts. In principle, three regeneration methods have been examined closely: (1) As in FCC, the hydrocarbons can be burned off the catalyst surface. This requires a catalyst with extreme temperature stability, which only ultrastable zeolites achieve. Moreover, as the alkylation process is exothermic and conducted at low to moderate temperatures, large amounts of process heat have to be removed. (2) The catalyst can be treated with a solvent to extract hydrocarbon deposits. The most straightforward solvent to use is isobutane, which has been shown to restore catalytic activity only partially. Supercritical solvents have been tested, but they also lead to only partial restoration of the activity. Supercritical alkylation to remove the deposits in situ has been shown in Section III.D.1 to be less effective. It is unlikely that this method of operation will lead to a competitive process. (3) The most promising regeneration method and the one that is used in all true solid acid-catalyzed process developments is a hydrogen treatment at both reaction and elevated temperature. This typically requires the incorporation of a hydrogenation function, for example a noble metal, in the catalyst. The regeneration mechanism depends on the temperature: at low temperatures (, 373 K), highly unsaturated species, which block the acid sites but not the pores, are hydrogenated. At higher temperatures, hydrocracking of longchain alkanes and other hydrocarbons that are too bulky to leave the pores is the predominant reaction. The fragments formed in this process easily desorb and leave the pore system. Although substantial research was devoted to plug-flow reactors, they are not a good choice for large-scale operation. To achieve a high internal isobutane/ alkene ratio (. 200), an enormous amount of isobutane has to be recycled. Nevertheless, a plug-flow reactor remains attractive because of the simplicity of its design and operation. When the alkene feed is introduced over the whole length of the reactor, very low isobutane/alkene ratios can be avoided. However, in a true fixed-bed reactor the inlet zones would nevertheless suffer from the higher alkene concentration and deactivate prematurely. A more appropriate type of reactor would be a backmixed slurry reactor, with the catalyst suspended in the liquid. Such a system, however, also has obvious disadvantages, such as the more complex design necessary for suspending the solid in the liquid and for solid/liquid separation. These disadvantages may be compensated by intrinsically higher isobutane/alkene ratios (a consequence of the backmixing), which lower catalyst consumption. Another advantage of a slurry reactor is the possibility to withdraw spent catalyst for regeneration. In fixed-bed reactors, the bed can only be regenerated as a whole, so that multiple swing reactors are necessary for uninterrupted production. Moreover, the attainment of isothermal operation in slurry reactors is better than in fixed-bed reactors.

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

285

IV.B.1. UOP Alkylenee Process UOP offers the Alkylenee process (237) utilizing a vertical riser reactor. A process scheme is shown in Fig. 15. The pretreated alkene feed is mixed with recycle isobutane and injected into the riser together with freshly reactivated catalyst. Both flow concurrently upward in the riser, where the reaction occurs. At the top of the riser the catalyst particles are disengaged and sink down into the reactivation zone. The hydrocarbons flow out through the top of the reactor vessel to the fractionation section, where they are separated into alkylate, n-butane, isobutane, and light ends including hydrogen. The recycle isobutane is cooled before re-entering the riser. The reactivation zone is a packed bed with the catalyst slowly moving downward in a low-temperature stream of isobutane saturated with hydrogen. Unsaturated molecules on the catalyst are claimed to be hydrogenated and desorbed from the catalyst surface. The reactivation zone leads to the bottom of the riser, where the cycle starts again. The catalyst reactivation is not complete, and so a small slipstream of catalyst is withdrawn and directed to a reactivation vessel, where the catalyst is regenerated in a semi-batch or batch mode at elevated temperature in a circulating hydrogen stream. The composition of the catalyst, which UOP refers to as HAL-100e, has not been disclosed. In several patents the use of an alumina-supported AlCl3 catalyst modified with alkali metal ions and a Ni, Pd, or Pt hydrogenation function is mentioned (see, for example, Ref. (214)). Obviously, traces of halogen compounds are leached out

Fig. 15. UOP Alkylenee solid acid-catalyzed alkylation process (237).

286

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

of the catalyst, because a product treatment section is necessary. This would additionally imply that a makeup halogen source is required. The alkene feed has to be extensively treated to remove di-alkenes, sulfur-, oxygen-, and nitrogencontaining compounds. The process operates at temperatures of 283 –313 K and at a molar isobutane/alkene ratio of 6– 15. No information is available concerning the alkene space velocity. It is interesting that typical alkene conversions are between 93– 100%, which most likely is a consequence of very short contact times in the riser reactor. The alkylate RON is claimed to be as high as what is attained with the existing technology. IV.B.2. Akzo Nobel/ABB Lummus AlkyCleane Process Akzo Nobel and ABB Lummus recently started a solid acid-catalyzed alkylation demonstration plant at a Fortum refinery in Finland (238). The reactor type used in the so-called AlkyCleane process has not been disclosed. However, the process utilizes serial reaction stages with distributed alkene feed injection for high internal isobutane/alkene ratios. The reactor type is claimed to achieve a high degree of mixing to reduce alkene concentration gradients throughout the reactor. Multiple reactors are used, which swing between reaction and regeneration. As in the Alkylenee process, two regeneration phases with different severities are employed. A mild regeneration at reaction temperature and pressure with hydrogen dissolved in isobutane is performed frequently (far before the end of the theoretical catalyst lifetime). When necessary, the catalyst is fully regenerated at 523 K in a stream of gas-phase hydrogen. Presumably, each reactor is in (mild) regeneration mode far longer than in reaction mode. The catalyst is reported to be a “true solid acid” without halogen ion addition. In the patent describing the process (239), a Pt/USY zeolite with an alumina binder is employed. It was claimed that the catalyst is rather insensitive to feed impurities and feedstock composition, so that feed pretreatment can be less stringent than in conventional liquid acid-catalyzed processes. The process is operated at temperatures of 323– 363 K, so that the cooling requirements are less than those of lower temperature processes. The molar isobutane/alkene feed ratio is kept between 8 and 10. Alkene space velocities are not reported. Akzo claims that the alkylate quality is identical to or higher than that attained with the liquid acid-catalyzed processes. IV.B.3. LURGI EUROFUELw Process LURGI and Su¨d-Chemie AG are developing a solid acid-catalyzed alkylation process termed LURGI EUROFUELw. The reactor is derived from tray distillation towers. Isobutane and suspended catalyst enter at the top of the

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

287

Fig. 16. LURGI EUROFUELw solid acid-catalyzed alkylation process (240).

tower, and the alkene with premixed isobutane is introduced in stages (Fig. 16). The evolved heat of reaction is most likely dissipated by the evaporation of the reaction mixture. Thus, the temperature is controlled by the overall pressure and the composition of the liquid. The catalyst – reactant mixture is agitated by the boiling mixture of alkylate and isobutane. At the bottom of the column, the catalyst is separated, and the majority of the alkylate/isobutane mixture is fed into the separation section. Isobutane is recycled and mixed with the catalyst, which is fed into the top of the reaction column. Intermittently, the catalyst is exposed to hydrogen-rich operating conditions to minimize accumulation of unsaturated compounds on its surface. Infrequent regeneration occurs in a proprietary section at elevated temperatures. The catalyst is faujasite derived, with a high concentration of sufficiently strong Brønsted acid sites and a minimized concentration of Lewis acid sites. It also contains a hydrogenation function. The process operates at temperatures of about 323– 373 K with a molar isobutane/alkene ratio between 6 and 12 and a higher alkene space velocity than in the liquid acid-catalyzed processes. Preliminary details of the process concept have been described (240). IV.B.4. Haldor Topsøe FBAe Process Haldor Topsøe’s fixed-bed alkylation (FBAe) technology is a compromise between liquid and solid acid-based processes. It applies a supported liquid-phase catalyst in which liquid triflic (trifluoromethanesulfonic) acid is supported on a porous material (206,241). The acid in the bed is concentrated in a well-defined catalyst zone, in which all the alkylation chemistry takes place: at the upstream

288

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

Fig. 17. Reaction zone in Haldor Topsøe’s FBAe alkylation process (206).

end of the catalyst zone, ester intermediates are formed, which are soluble in the hydrocarbons and are transported into the acid zone. Here, they react to form the products and free acid. Thus, the active zone slowly migrates through the bed in the direction of the hydrocarbon flow, as shown in Fig. 17. The spent acid can be withdrawn from the reactor without interrupting the production. The acid is regenerated in a proprietary acid recovery unit, which produces some oil as a byproduct. The products have to be treated to remove trace amounts of acid. Reaction temperatures are in the range of 273 –293 K. The reactor is operated adiabatically, and the reaction heat is removed by a cooled reactor effluent recycle (Fig. 18). The process is claimed to be robust against feed impurities. Feed drying, however, is recommended.

Fig. 18. Haldor Topsøe’s FBAe alkylation process. Adapted from Ref. (206).

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

289

V. Conclusions The foregoing review of the alkylation mechanism and the influence of the catalyst type and reaction conditions show that, in essence, the chemistry is identical with all the examined acid catalysts, liquid and solid. Differences in the importance of individual reaction steps originate from the variety of possible structures and distributions of acid sites of solid catalysts. Changing process parameters induces similar effects with each of the catalysts; however, the sensitivity to a particular parameter depends strongly on the catalyst. All the acids deactivate by the formation of unsaturated polymers, which are strongly bound to the acid. Liquid acid-catalyzed processes are mature technologies, which are not expected to undergo dramatic changes in the near future. Solid acid-catalyzed alkylation now has been developed to a point where the technology can compete with the existing processes. Catalyst regeneration by hydrogen treatment is the method of choice in all the process developments. Some of the process developments eliminate most if not all the drawbacks of the liquid acid processes. The verdict about whether solid acid-catalyzed processes will be applied in the near future will be determined primarily by economic issues.

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

Stell, J., Oil Gas J. 99(52), 74 (2001). Anonymous, Oil Gas J. 98(13) (2000). Anonymous, Oil Gas J. 98(9) (2000). Ipatieff, V.N., and Grosse, A.V., J. Am. Chem. Soc. 57, 1616 (1935). Ipatieff, V.N., and Pines, H., US Patent 2,122,847 (1938). Hutson, T. Jr., and McCarthy, W.C., in “Handbook of Petroleum Refining Processes” (R.A. Meyers, Ed.), p. 1/23. McGraw-Hill, New York, 1986. Pines, H., Chemtech. March, 150 (1982). Albright, L.F., Chemtech. June, 40 (1998). Albright, L.F., Oil Gas J. 26, 70 (1990). Furimsky, E., Catal. Today 30, 223 (1996). Albright, L.F., Chemtech. July, 46 (1998). Kinnear, S., “STRATCO Alkylation Seminar,” Phoenix, AZ, 1998. Garwood, W.E., and Venuto, P.B., J. Catal. 11, 175 (1968). Kirsch, F.W., Potts, J.D., and Barmby, D.S., J. Catal. 27, 142 (1972). Corma, A., and Martinez, A., Catal. Rev.—Sci. Engng 35, 483 (1993). Weitkamp, J., and Traa, Y., in “Handbook of Heterogeneous Catalysis” (G. Ertl, H. Kno¨zinger and J. Weitkamp, Eds.), Vol. 4, p. 2039. VCH, Weinheim, 1997. Schmerling, L., J. Am. Chem. Soc. 67, 1778 (1945). Schmerling, L., J. Am. Chem. Soc. 68, 275 (1946). Schmerling, L., Ind. Engng Chem. 45, 1447 (1953). Albright, L.F., Spalding, M.A., Faunce, J., and Eckert, R.E., Ind. Engng Chem. Res. 27, 391 (1988).

290

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

21. Hutson, T. Jr., and Logan, R.S., Hydrocarbon Process. September, 107 (1975). 22. Corma, A., Martinez, A., and Martinez, C., J. Catal. 149, 52 (1994). 23. Gary, J.H., and Handwerk, G.E., in “Petroleum Refining—Technology and Economics” (L.F. Albright, R.N. Maddox and J.J. McKetta, Eds.), p. 142. Marcel Dekker, New York, 1979. 24. Hommeltoft, S.I., Ekelund, O., and Zavilla, J., Ind. Engng Chem. Res. 36, 3491 (1997). 25. Frash, M.V., Solkan, V.N., and Kazansky, V.B., J. Chem. Soc. Faraday Trans. 93, 515 (1997). 26. Kazansky, V.B., Abbenhuis, H.C.L., van Santen, R.A., and Vorstenbosch, M.L.V., Catal. Lett. 69, 51 (2000). 27. Kazansky, V.B., Catal. Rev.—Sci. Engng 43, 199 (2001). 28. Albright, L.F., Spalding, M.A., Nowinski, J.A., Ybarra, R.M., and Eckert, R.E., Ind. Engng Chem. Res. 27, 381 (1988). 29. Kazansky, V.B., Catal. Today 51, 419 (1999). 30. Gorte, R.J., and White, D., Topics Catal. 4, 57 (1997). 31. Kazansky, V.B., in “Handbook of Heterogeneous Catalysis” (G. Ertl, H. Kno¨zinger and J. Weitkamp, Eds.), Vol. 2, p. 740. VCH, Weinheim, 1997. 32. Rigby, A.M., Kramer, G.J., and van Santen, R.A., J. Catal. 170, 1 (1997). 33. Rozanska, X., Demuth, T., Hutschka, F., Hafner, J., and van Santen, R.A., J. Phys. Chem. B 106, 3248 (2002). 34. Ishikawa, H., Yoda, E., Kondo, J.N., Wakabayashi, F., and Domen, K., J. Phys. Chem. B 103, 5681 (1999). 35. Kondo, J.N., Ishikawa, H., Yoda, E., Wakabayashi, F., and Domen, K., J. Phys. Chem. B 103, 8538 (1999). 36. Engelhardt, J., and Hall, W.K., J. Catal. 151, 1 (1995). 37. Mota, C.J.A., Esteves, P.M., Ramirez-Solis, A., and Hernandez-Lamoneda, R., J. Am. Chem. Soc. 119, 5193 (1997). 38. Hogeveen, H., Gaasbeek, C.J., and Bickel, A.F., Rec. Trav. Chim. 88, 703 (1969). 39. Olah, G.A., and Olah, J.A., in “Carbonium Ions” (G.A. Olah and P.R. Schleyer, Eds.), Vol. 2, p. 715. Interscience, New York, 1970. 40. Sanchez-Castillo, M.A., Agarwal, N., Miller, C., Cortright, R.D., Madon, R.J., and Dumesic, J.A., J. Catal. 205, 67 (2002). 41. Boronat, M., Viruela, P., and Corma, A., J. Phys. Chem. A 102, 982 (1998). 42. Stewart, T.D., and Calkins, W.H., J. Am. Chem. Soc. 70, 1006 (1948). 43. Miron, L., and Lee, R.J., J. Chem. Engng Data 8, 150 (1963). 44. Albright, L.F., and Li, K.W., Ind. Engng Chem. Process Des. Dev. 9, 447 (1970). 45. Li, K.W., Eckert, R.E., and Albright, L.F., Ind. Engng Chem. Process Des. Dev. 9, 441 (1970). 46. Hofmann, J.E., and Schriesheim, A., J. Am. Chem. Soc. 84, 953 (1962). 47. Gorin, M.H., Kuhn, C.S., and Miles, C.B., Ind. Engng Chem. 38, 795 (1946). 48. Nivarthy, G.S., He, Y., Seshan, K., and Lercher, J.A., J. Catal. 176, 192 (1998). 49. Albright, L.F., and Kranz, K.E., Ind. Engng Chem. Res. 31, 475 (1992). 50. Shah, B.R., in “Handbook of Petroleum Refining Processes” (R.A. Meyers, Ed.), p. 1/3. McGraw-Hill, New York, 1986. 51. Pines, H., Grosse, A.V., and Ipatieff, V.N., J. Am. Chem. Soc. 64, 33 (1942). 52. Olah, G.A., and Prakash, G.K.S., in “The Chemistry of Alkanes and Cycloalkanes” (S. Patai and Z. Rappoport, Eds.), p. 609. Wiley, London, 1992. 53. Durrett, L.R., Taylor, L.M., Wantland, C.F., and Dvoretzky, I., Anal. Chem. 35, 637 (1963). 54. Cumming, K.A., and Wojciechowski, B.W., Catal. Rev.—Sci. Engng 38, 101 (1996).

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

291

55. Bartlett, P.D., Condon, F.E., and Schneider, A., J. Am. Chem. Soc. 66, 1531 (1944). 56. Deno, N.C., Peterson, H.J., and Saines, G.S., Chem. Rev. 60, 7 (1960). 57. Wojciechowski, B.W., and Corma, A., “Catalytic Cracking—Catalysts, Chemistry, and Kinetics.” Marcel Dekker, New York, 1986, p. 5. 58. Ausloos, P., and Lias, S.G., J. Am. Chem. Soc. 92, 5037 (1970). 59. Meot-Ner, M., and Field, F.H., J. Chem. Phys. 64, 277 (1976). 60. Meot-Ner, M., J. Am. Chem. Soc. 109, 7947 (1987). 61. Sunner, J.A., Hirao, K., and Kebarle, P., J. Phys. Chem. 93, 4010 (1989). 62. Olah, G.A., Mo, Y.K., and Olah, J.A., J. Am. Chem. Soc. 95, 4939 (1973). 63. Boronat, M., Viruela, P., and Corma, A., Phys. Chem. Chem. Phys. 2, 3327 (2000). 64. Boronat, M., Viruela, P., and Corma, A., J. Phys. Chem. A 102, 9863 (1998). 65. Nowak, A.K., Mesters, C.M.A.M., Rigby, A.M., and Schulze, D., Preprints. Div. Petrol. Chem., Am. Chem. Soc. 41, 668 (1996). 66. Boronat, M., Zicovich-Wilson, C.M., Corma, A., and Viruela, P., Phys. Chem. Chem. Phys. 1, 537 (1999). 67. Beirnaert, H.C., Alleman, J.R., and Marin, G.B., Ind. Engng Chem. Res. 40, 1337 (2001). 68. Yaluris, G., Rekoske, J.E., Aparicio, L.M., Madon, R.J., and Dumesic, J.A., J. Catal. 153, 54 (1995). 69. Guisnet, M., and Gnep, N.S., Appl. Catal. A 146, 33 (1996). 70. Corma, A., Faraldos, M., Martinez, A., and Mifsud, A., J. Catal. 122, 230 (1990). 71. Hofmann, J.E., and Schriesheim, A., J. Am. Chem. Soc. 84, 957 (1962). 72. Nenitzescu, C.D., in “Carbonium ions” (G.A. Olah and P.R. Schleyer, Eds.), Vol. 2, p. 463. Interscience, New York, 1970. 73. Scho¨llner, R., and Ho¨lzel, H., Z. Chem. 15, 469 (1975). 74. Reyniers, M.-F., Beirnaert, H., and Marin, G.B., Appl. Catal. A 202, 49 (2000). 75. Reyniers, M.-F., Tang, Y., and Marin, G.B., Appl. Catal. A 202, 65 (2000). 76. Weeks, T.J. Jr., and Bolton, A.P., J. Chem. Soc. Faraday Trans. 170, 1676 (1974). 77. Lutsyk, A.I., and Suikov, S.Y., Petrol. Chem. 35, 412 (1995). 78. Nivarthy, G.S., Seshan, K., and Lercher, J.A., Micropor. Mesopor. Mater. 22, 379 (1998). 79. Lee, L., and Harriott, P., Ind. Engng Chem. Process Des. Dev. 16, 282 (1977). 80. de Jong, K.P., Mesters, C.M.A.M., Peferoen, D.G.R., van Brugge, P.T.M., and de Groot, C., Chem. Engng Sci. 51, 2053 (1996). 81. Simpson, M.F., Wei, J., and Sundaresan, S., Ind. Engng Chem. Res. 35, 3861 (1996). 82. Albright, L.F., and Wood, K.V., Ind. Engng Chem. Res. 36, 2110 (1997). 83. Scho¨llner, R., and Ho¨lzel, H., J. Prakt. Chem. 317, 694 (1975). 84. Martens, J.A., and Jacobs, P.A., Stud. Surf. Sci. Catal. 137, 633 (2001). 85. Weitkamp, J., and Ernst, S., Stud. Surf. Sci. Catal. 38, 367 (1988). 86. Frash, M.V., and van Santen, R.A., Topics Catal. 9, 191 (1999). 87. Daage, M., and Fajula, F., Bull. Soc. Chim. Fr. 5/6, 153 (1984). 88. Pater, J., Cardona, F., Canaff, C., Gnep, N.S., Szabo, G., and Guisnet, M., Ind. Engng Chem. Res. 38, 3822 (1999). 89. Feller, A., Guzman, A., Zuazo, I., and Lercher, J.A., J. Catal. 224, 80 (2004). 90. Hofmann, J.E., J. Org. Chem. 29, 1497 (1964). 91. Graves, D.C., Kranz, K., and Millard, J., Prepr. Div. Petrol. Chem., Am. Chem. Soc. 39, 398 (1994). 92. Roebuck, A.K., and Evering, B.L., Ind. Engng Chem. Prod. Res. Dev. 9, 76 (1970). 93. Cardona, F., Gnep, N.S., Guisnet, M., Szabo, G., and Nascimento, P., Appl. Catal. A 128, 243 (1995). 94. am Ende, D.J., and Albright, L.F., Ind. Engng Chem. Res. 33, 840 (1994). 95. Guisnet, M., and Magnoux, P., Appl. Catal. A 212, 83 (2001).

292

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

96. Albright, L.F., Spalding, M.A., Kopser, C.G., and Eckert, R.E., Ind. Engng Chem. Res. 27, 386 (1988). 97. Kramer, G.M., J. Org. Chem. 32, 920 (1967). 98. Kramer, G.M., J. Org. Chem. 32, 1916 (1967). 99. Otvos, J.W., Stevenson, D.P., Wagner, C.D., and Beeck, O., J. Am. Chem. Soc. 73, 5741 (1951). 100. Flego, C., Kiricsi, I., Parker, W.O. Jr., and Clerici, M.G., Appl. Catal. A 124, 107 (1995). 101. Feller, A., Guzman, A., Zuazo, I., Barth, J.O., and Lercher, J.A., J. Catal. 216, 313 (2003). 102. Weitkamp, J., and Maixner, S., Zeolites 7, 6 (1987). 103. Deno, N.C., in “Carbonium Ions” (G.A. Olah and P.R. Schleyer, Eds.), Vol. 2, p. 783. Interscience, New York, 1970. 104. Sorensen, T.S., in “Carbonium Ions” (G.A. Olah and P.R. Schleyer, Eds.), Vol. 2, p. 807. Interscience, New York, 1970. 105. Nicholas, J.B., and Haw, J.F., J. Am. Chem. Soc. 120, 11804 (1998). 106. Yang, S., Kondo, J.N., and Domen, K., Catal. Today 73, 113 (2002). 107. Li, K.W., Eckert, R.E., and Albright, L.F., Ind. Engng Chem. Process Des. Dev. 9, 434 (1970). 108. Albright, L.F., Ind. Engng Chem. Res. 40, 4032 (2001). 109. He, M.Y., and Min, E., Catal. Today 63, 113 (2000). 110. Olah, G.A., Batamack, P., Deffieux, D., To¨ro¨k, B., Wang, Q., Molnar, A., and Surya Prakash, G.K., Appl. Catal. A 146, 107 (1996). 111. Olah, G.A., Marinez, E., To¨ro¨k, B., and Prakash, G.K.S., Catal. Lett. 61, 105 (1999). 112. Ho¨lderich, W., and Jacobs, P.A., Stud. Surf. Sci. Catal. 137, 821 (2001). 113. Corma, A., and Garcia, H., Catal. Today 38, 257 (1997). 114. Ramachandran, S., Lenz, T.G., Skiff, W.M., and Rappe´, A.K., J. Phys. Chem. 100, 5898 (1996). 115. Eder, F., and Lercher, J.A., Zeolites 18, 75 (1997). 116. Ja¨nchen, J., Stach, H., Uytterhoeven, L., and Mortier, W.J., J. Phys. Chem. 100, 12489 (1996). 117. Eder, F., and Lercher, J.A., J. Phys. Chem. 100, 16460 (1996). 118. Haag, W.O., Stud. Surf. Sci. Catal. 84, 1375 (1994). 119. Chu, Y.F., and Chester, A.W., Zeolites 6, 195 (1986). 120. Weitkamp, J., in “Proceedings of the Fifth International Zeolite Conference” (L.V.C. Rees, Ed.), p. 858. Heyden, London, 1980. 121. Corma, A., Martinez, A., and Martinez, C., J. Catal. 146, 185 (1994). 122. Mostad, H.B., Sto¨cker, M., Karlsson, A., and Rørvik, T., Appl. Catal. A 144, 305 (1996). 123. Sto¨cker, M., Mostad, H., and Rørvik, T., Catal. Lett. 28, 203 (1994). 124. Sto¨cker, M., Mostad, H., Karlsson, A., Junggreen, H., and Hustad, B., Catal. Lett. 40, 51 (1996). 125. Rørvik, T., Mostad, H.B., Karlsson, A., and Ellestad, O.H., Appl. Catal. A 156, 267 (1997). 126. Corma, A., Martinez, A., and Martinez, C., Catal. Lett. 28, 187 (1994). 127. Diaz-Mendoza, F.A., Pernett-Bolano, L., and Cardona-Martinez, N., Thermochim. Acta 312, 47 (1998). 128. Weitkamp, J., and Ernst, S., “Proceedings of the 13th World Petroleum Congress,” 1992, p. 315. 129. Zhuang, Y., and Ng, F.T.T., Appl. Catal. A 190, 137 (2000). 130. Nakamura, I., Ishida, S., and Fujimoto, K., Preprints. Div. Petrol. Chem., Am. Chem. Soc. 40, 512 (1995).

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

293

131. Fan, L., Nakamura, I., Ishida, S., and Fujimoto, K., Ind. Engng Chem. Res. 36, 1458 (1997). 132. Chu, W., Zhao, Z., Sun, W., Ye, X., and Wu, Y., Catal. Lett. 55, 57 (1998). 133. Baronetti, G., Thomas, H., and Querini, C.A., Appl. Catal. A 217, 131 (2001). 134. Ramos-Galvan, C.E., Dominguez, J.M., Sandoval-Robles, G., Mantilla, A., and Ferrat, G., Catal. Today 65, 391 (2001). 135. Corma, A., Martinez, A., Arroyo, P.A., Monteiro, J.L.F., and Sousa-Aguiar, E.F., Appl. Catal. A 142, 139 (1996). 136. Corma, A., Martinez, A., and Martinez, C., Appl. Catal. A 134, 169 (1996). 137. Fritz, P.O., and Lunsford, J.H., J. Catal. 118, 85 (1989). 138. Scherzer, J., Catal. Rev.—Sci. Engng 31, 215 (1989). 139. Weitkamp, J., and Traa, Y., Catal. Today 49, 193 (1999). 140. Corma, A., Gomez, V., and Martinez, A., Appl. Catal. A 119, 83 (1994). 141. Unverricht, S., Ernst, S., and Weitkamp, J., Stud. Surf. Sci. Catal. 84, 1693 (1994). 142. Loenders, R., Jacobs, P.A., and Martens, J.A., J. Catal. 176, 545 (1998). 143. Yoo, K., and Smirniotis, P.G., Appl. Catal. A 227, 171 (2002). 144. Agaskar, P.A., and Huang, T.J., US Patent 5,824,835 (1998). 145. Corma, A., and Orchille´s, A.V., Microp. Mesop. Mater. 35-36, 21 (2000). 146. Wielers, A.F.H., Vaarkamp, M., and Post, M.F.M., J. Catal. 127, 51 (1991). 147. Garwood, W.E., Leaman, W.K., and Plank, C.J., US Patent 3,251,902 (1966). 148. Weitkamp, J., Stud. Surf. Sci. Catal. 5, 65 (1980). 149. Kashkovskii, V.I., Galich, P.N., Patrilyak, K.I., and Royev, L.M., Petrol. Chem. 31, 480 (1991). 150. Galich, P.N., Tsupryk, I.N., and Vasil’yev, A.N., Petrol. Chem. 37, 332 (1997). 151. Kirsch, F.W., Barmby, D.S., and Potts, J.D., US Patent 4,300,015 (1981). 152. Mortikov, E.S., Plakhotnic, V.A., and Dolinsky, S.E., German Patent DE 197 45 548 A1 (1999). 153. Dolinsky, S.E., Plakhotnik, V.A., Mortikov, E.S., and Bachurikhin, A.L., Prepr. Div. Petrol. Chem., Am. Chem. Soc. 46, 246 (2001). 154. Park, H.S., and Seff, K., J. Phys. Chem. B 104, 2224 (2000). 155. Scherzer, J., Bass, J.L., and Hunter, F.D., J. Phys. Chem. 79, 1194 (1975). 156. Carvajal, R., Chu, P.J., and Lunsford, J.H., J. Catal. 125, 123 (1990). 157. Bolton, A.P., J. Catal. 22, 9 (1971). 158. Cheetham, A.K., Eddy, M.M., and Thomas, J.M., J. Chem. Soc., Chem. Commun. 1337 (1984). 159. Ward, J.W., J. Catal. 13, 321 (1969). 160. Ward, J.W., J. Colloid Interf. Sci. 28, 269 (1968). 161. Ward, J.W., J. Catal. 14, 365 (1969). 162. Kirsch, F.W., Barmby, D.S., and Potts, J.D., US Patent 3,624,173 (1971). 163. Kirsch, F.W., Barmby, D.S., and Potts, J.D., US Patent 3,706,814 (1972). 164. Kirsch, F.W., Barmby, D.S., and Potts, J.D., US Patent 3,839,228 (1974). 165. Kirsch, F.W., Barmby, D.S., and Potts, J.D., US Patent 3,803,256 (1974). 166. Kirsch, F.W., Barmby, D.S., and Potts, J.D., US Patent 3,865,894 (1975). 167. Mostad, H., Sto¨cker, M., Karlsson, A., Junggreen, H., and Hustad, B., Stud. Surf. Sci. Catal. 105, 1413 (1997). 168. Chester, A.W., and Chu, Y.F., US Patent 4,377,721 (1983). 169. Buchanan, J.S., and Huang, T.J., PCT Patent WO 97/20787 (1997). 170. Weitkamp, J., and Jacobs, P.A., in “New Frontiers in Catalysis” (L. Guczi, F. Solymosi and P. Te´te´nyi, Eds.), Vol. B, p. 1735. Akade´miai Kiado´, Budapest, 1993. 171. Chu, C.T.W., Husain, A., Keville, K.M., and Lissy, D.N., US Patent 5,516,962 (1996).

294

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

172. Husain, A., PCT Patent WO 94/03415 (1994). 173. Child, J.E., Huss Jr., A., Krambeck, F.J., Ragonese, F.P., Thomson, R.T., and Yurchak, S., US Patent 5,073,665 (1991). 174. Huss Jr., A., Kirker, G.W., Keville, K.M., and Thomson, T.J., US Patent 4,992,615 (1991). 175. He, J.Y., Nivarthy, G.S., Eder, F., Seshan, K., and Lercher, J.A., Microp. Mesop. Mater. 25, 207 (1998). 176. Yoo, K., Burckle, E.C., and Smirniotis, P.G., Catal. Lett. 74, 85 (2001). 177. Borade, R.B., and Clearfield, A., J. Phys. Chem. 96, 6729 (1992). 178. Corma, A., Diaz-Cabanas, M.J., Martinez, C., and Valencia, S., “Proceedings of the 13th International Zeolite Conference, Montpellier,” 2001, p. 275. 179. Nivarthy, G.S., Feller, A., Seshan, K., and Lercher, J.A., Stud. Surf. Sci. Catal. 130, 2561 (2000). 180. Rørvik, T., Mostad, H., Ellestad, O.H., and Sto¨cker, M., Appl. Catal. A 137, 235 (1996). 181. Zhang, Z., Han, Y., Xiao, F.S., Qiu, S., Zhu, L., Wang, R., Yu, Y., Zhang, Z., Zou, B., Wang, Y., Sun, H., Zhao, D., and Wei, Y., J. Am. Chem. Soc. 123, 5014 (2001). 182. Huang, T.J., and Kresge, C.T., US Patent 5,498,817 (1996). 183. Husain, A., Huss, A., Klocke, and D.J., Timken, H.K.C., PCT Patent WO 93/07106 (1993). 184. Chu, C.T., Husain, A., Huss Jr., A., Kresge, C.T., and Roth, W.J., US Patent 5,258,569 (1993). 185. Harandi, M.N., Beech Jr., J.H., Huss Jr., A., Ware, R.A., and Husain, A., US Patent 5,625, 113 (1997). 186. Husain, A., Huss Jr., A., and Rahmim, I.I., US Patent 5,475,175 (1995). 187. Huss Jr., A., Le, Q.N., and Thomson, R.T., US Patent 5,326,922 (1994). 188. Corma, A., Juan-Rajadell, M.I., Lopez-Nieto, J.M., Martinez, A., and Martinez, C., Appl. Catal. A 111, 175 (1994). 189. Farcasiu, D., Ghenciu, A., and Li, J.Q., J. Catal. 158, 116 (1996). 190. Corma, A., Martinez, A., and Martinez, C., Appl. Catal. A 144, 249 (1996). 191. Satoh, K., Matsuhashi, H., and Arata, K., Appl. Catal. A 189, 35 (1999). 192. Gore, R.B., and Thomson, W.J., Appl. Catal. A 168, 23 (1998). 193. Chellappa, A.S., Miller, R.C., and Thomson, W.J., Appl. Catal. A 209, 359 (2001). 194. Das, D., and Chakrabarty, D.K., Energy Fuels 12, 109 (1998). 195. Guo, C., Yao, S., Cao, J., and Qian, Z., Appl. Catal. A 107, 229 (1994). 196. Guo, C., Liao, S., Qian, Z., and Tanabe, K., Appl. Catal. A 107, 239 (1994). 197. Xiao, X., Tierney, J.W., and Wender, I., Appl. Catal. A 183, 209 (1999). 198. Sun, M., Sun, J., and Li, Q., Chem. Lett. (1998), 519 (1998). 199. Blasco, T., Corma, A., Martinez, A., and Martinez-Escolano, P., J. Catal. 177, 306 (1998). 200. Gayraud, P.Y., Stewart, I.H., Derouane-Abd Hamid, S.B., Essayem, N., Derouane, E.G., and Vedrine, J.C., Catal. Today 63, 223 (2000). 201. Okuhara, T., Yamashita, M., Na, K., and Misono, M., Chem. Lett. (1994), 1451 (1994). 202. de Angelis, A., Ingallina, P., Berti, D., Montanari, L., and Clerici, M.G., Catal. Lett. 61, 45 (1999). 203. Zhao, Z., Sun, W., Yang, X., Ye, X., and Wu, Y., Catal. Lett. 65, 115 (2000). 204. Rørvik, T., Dahl, I.M., Mostad, H.B., and Ellestad, O.H., Catal. Lett. 33, 127 (1995). 205. Botella, P., Corma, A., and Lopez-Nieto, J.M., J. Catal. 185, 371 (1999). 206. Hommeltoft, S.I., Appl. Catal. A 221, 421 (2001). 207. Clet, G., Goupil, J.M., Szabo, G., and Cornet, D., Appl. Catal. A 202, 37 (2000). 208. Nascimento, P., Szabo, G., and Milan, A., US Patent 6,225,517 (2001). 209. Kojima, M., and Kocal, J.A., US Patent 5,391,527 (1995).

A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295

295

210. Funk, G.A., Hobbs, S.A., Oroskar, A.R., Gembicki, S.A., and Kocal, J.A., US Patent 5, 523,503 (1996). 211. Zhang, S.Y.F., Gosling, C.D., Sechrist, P.A., and Funk, G.A., US Patent 5,675,048 (1997). 212. McBride, T.K., Bricker, M.L., and Steigleder, K.Z., US Patent 5,744,682 (1998). 213. Kocal, J.A., and Oroskar, A.R., US Patent 5,849,977 (1998). 214. McBride, T.K., Bricker, M.L., and Steigleder, K.Z., US Patent 5,883,039 (1999). 215. Herbstman, S., Cole, E.L., and Estes, J.H., US Patent 4,138,444 (1979). 216. Mantilla-Ramirez, A., Ferrat-Torres, G., Dominguez, J.M., Aldana-Rivero, C., and Bernal, M., Appl. Catal. A 143, 203 (1996). 217. Chou, T.S., Huss Jr., A., Kennedy, C.R., Kurtas, R.S., and Tabak, S.A., PCT Patent WO 90/00534 (1990). 218. Huss Jr., A., and Johnson, I.D., US Patent 5,221,777 (1993). 219. Chou, T.S., Huss, Jr., A., Kennedy, C.R., and Kurtas, R.S., PCT Patent WO 90/00533 (1990). 220. de Angelis, A., Flego, C., Ingallina, P., Montanari, L., Clerici, M.G., Carati, C., and Perego, C., Catal. Today 65, 363 (2001). 221. Ackerman, S., Chitnis, G.K., and McCaffrey, D.S. Jr., Prepr. Div. Petrol. Chem., Am. Chem. Soc. 46, 241 (2001). 222. Taylor, R., and Sherwood, D.E. Jr., Appl. Catal. A 155, 195 (1997). 223. Clark, M.C., and Subramaniam, B., Ind. Engng Chem. Res. 37, 1243 (1998). 224. Santana, G.M., and Akgerman, A., Ind. Engng Chem. Res. 40, 3879 (2001). 225. Ginosar, D.M., Thompson, D.N., Coates, K., and Zalewski, D.J., Ind. Engng Chem. Res. 41, 2864 (2002). 226. Querini, C.A., Catal. Today 62, 135 (2000). 227. Jones, E.K., Adv. Catal. 10, 165 (1958). 228. Cupit, C.R., Gwyn, J.E., and Jernigan, E.C., Petrol. Chem. Engng 33, 47 (1961). 229. Daage, M., and Fajula, F., Bull. Soc. Chim. Fr. 5/6, 160 (1984). 230. Nivarthy, G.S., Feller, A., Seshan, K., and Lercher, J.A., Microp. Mesop. Mater. 35/36, 75 (2000). 231. Schweitzer, E.J.A., Barendregt, S., and van der Oosterkamp, P.F., “Proceedings of the Petrotech 98,” Bahrain, 81, 1998. 232. Dolinsky, S.E., and Plakhotnik, V.A., “Proceedings of the 13th International Zeolite Conference, Montpellier,” 2001, p. 272. 233. Dolinskii, S.E., Subbotin, A.N., Lishchiner, I.I., Plakhotnik, V.A., and Mortikov, Y.S., Petrol. Chem. 37, 447 (1997). 234. Cornet, D., Goupil, J.M., Szabo, G., Poirier, J.L., and Clet, G., Appl. Catal. A 141, 193 (1996). 235. Goupil, J.M., Poirier, J.L., and Cornet, D., Ind. Engng Chem. Res. 33, 712 (1994). 236. http://www.fuelstechnology.com/soft_processoverview.htm, 2000. 237. Black, S.M., Gosling, C.D., Steigleder, K.Z., and Shields, D.J., “Proceedings of the NPRA Annual Meeting, San Antonio, TX, USA,” 2000. 238. D’Amico, V.J., van Broekhoven, E.H., Nat, P.J., Nousiainen, H., and Jakkula, J., “Proceedings of the NPRA Annual Meeting, San Antonio, TX, USA” 2002. 239. van Broekhoven, E.H., Mas Cabre, F.R., Bogaard, P., Klaver, G., and Vonhof, M., US Patent 5,986,158 (1999). 240. Buchold, H., Dropsch, H., and Eberhardt, J., “Proceedings of the 17th World Petroleum Congress, Brazil, 2002,” Vol. 3, p.189, Institute of Petroleum, London, 2003. 241. Sarup, B., Hommeltoft, S.I., Sylvest-Johansen, M., and Søgaard-Andersen, P., “Proceedings of the DGMK-Conference Catalysis on Solid Acids and Bases,” Berlin 1996, p. 175.

This Page Intentionally Left Blank

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

Catalytic Conversion of Methane to Synthesis Gas by Partial Oxidation and CO2 Reforming YUN HANG HU and ELI RUCKENSTEIN Department of Chemical Engineering, State University of New York at Buffalo, Buffalo, NY 14260, USA

I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II. Partial Oxidation of Methane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II.A. Hot Spots in Catalyst Beds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II.B. Minimizing O2 Purification Costs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II.C. Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II.D. Reaction Pathways . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II.D.1. Changes in Catalyst During Reaction . . . . . . . . . . . . . . . . . . . . . . II.D.2. Which is the Primary Product, CO or CO2? . . . . . . . . . . . . . . . . . . II.D.3. CHx Species and Rate-Determining Steps . . . . . . . . . . . . . . . . . . . II.D.4. Comparison of Reactions on Reduced and Unreduced Catalysts . . . III. CO2 Reforming of Methane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . III.A. Carbon Formation on Metal Surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . III.B. Critical Issues Related to Carbon Deposition . . . . . . . . . . . . . . . . . . . . . . III.C. Supported Noble Metal Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . III.D. Non-Noble Metal Supported Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . III.D.1. Ni/Al2O3 Catalysts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . III.D.2. Ni/SiO2 Catalysts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . III.D.3. Ni/La2O3 Catalysts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . III.D.4. Ni/ZrO2 Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . III.D.5. Other Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . III.E. MgO-Containing Solid-Solution Catalysts . . . . . . . . . . . . . . . . . . . . . . . . III.E.1. Characteristics of MgO-Containing Solid-Solution Catalysts. . . . . III.E.2. Highly Effective MgO-Containing Solid-Solution Catalysts . . . . . IV. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

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

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

298 301 301 306 312 314 315 316 318 320 321 321 322 323 324 325 327 328 330 331 332 332 333 337 338

The preparation of synthesis gas from natural gas, which is the most important step in the gas-to-liquid transformation, has attracted increasing attention in the last decade. Steam reforming, partial oxidation, and CO2 reforming are the three major processes that can be employed to prepare synthesis gas. Because steam reforming was reviewed recently in this series [Adv. Catal. 47 (2002) 65], this chapter deals only with the latter two processes.

ADVANCES IN CATALYSIS, VOLUME 48 ISSN: 0360-0564 DOI 10.1016/S0360-0564(04)48004-3

Copyright q 2004 Elsevier Inc. All rights reserved

298

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

The history of the development of methane conversion to synthesis gas is summarized as an introduction to the partial oxidation of methane, which is reviewed with emphasis on hot spots in reactors, major developments in the reduction of O2 separation costs, and reaction mechanisms. The various catalysts employed in CO2 reforming are examined, with emphasis on inhibition of carbon deposition. q 2004 Elsevier Inc.

Abbreviations DRIFT EDS FTIR GHSV MIEC MS R SPARG TEM TG/DTG TPD TPH TPO TPR TPSR WHSV XRD w/o XPS

diffuse reflectance infrared Fourier transform energy dispersive X-ray spectrometer Fourier transform infrared gas hourly space velocity mixed ionic/electronic conductors mass spectrometer reaction rate sulfur passivated reforming transmission electron microscopy thermal gravimetric/differential thermal gravimetric temperature-programmed decomposition temperature-programmed hydrogenation temperature-programmed oxidation temperature-programmed reduction temperature-programmed surface reaction weight hourly space velocity X-ray diffraction water-in-oil X-ray photoelectron spectroscopy

I. Introduction In the 1930s, Standard Oil of New Jersey (1) was the first company to employ on a commercial scale the indirect conversion of methane, the main component of natural gas, via steam reforming to give synthesis gas, which is a mixture of H2 and CO, with the H2/CO ratio depending on the reactant composition. CO2 is also formed in synthesis gas production, and sulfur compounds are present as impurities. Synthesis gas can be used as a feedstock for numerous chemicals and fuels and as a source of pure hydrogen or carbon monoxide. The steam reforming process is widely employed today (2). The reaction CH4 þ H2 O ! CO þ 3H2 ;

0 DH298 ¼ 206 kJ mol21

ð1Þ

is expensive because of its endothermic nature, the requirement for low space velocities, and the high H2/CO ratio (3/1), which is unsuitable for synthesis of methanol or the long-chain hydrocarbons made in the Fischer – Tropsch process.

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

299

The other two main processes for conversion of methane into synthesis gas are partial oxidation and CO2 reforming. In the 1940s, Prettre et al. (3) first reported the formation of synthesis gas by the catalytic partial oxidation of CH4 CH4 þ

1 2

O2 ! CO þ 2H2 ;

0 DH298 ¼ 236 kJ mol21

ð2Þ

They used a Ni-containing catalyst. In contrast to steam reforming of methane, methane partial oxidation is exothermic. However, the partial oxidation requires pure oxygen, which is produced in expensive air separation units that are responsible for up to 40% of the cost of a synthesis gas plant (2) (in contrast, the steam reforming process does not require pure oxygen). Therefore, the catalytic partial oxidation of methane did not attract much interest for nearly half a century, and steam reforming of methane remained the main commercial process for synthesis gas manufacture. CO2 reforming, CH4 þ CO2 ! 2CO þ 2H2 ;

0 DH298 ¼ 247 kJ mol21

ð3Þ

was investigated as early as 1888 (4). Although this process, like steam reforming, is also endothermic, it produces synthesis gas with a lower H2/CO ratio than steam reforming, and is, therefore, suitable for the Fischer – Tropsch synthesis of longchain hydrocarbons (5). Furthermore, it can be carried out with natural gas from fields containing large amounts of CO2, without the pre-separation of CO2 from the feed. Because CO2 is a greenhouse gas that causes warming of the earth and climate change, there are incentives for reducing its concentration in the atmosphere (6). CO2 reforming of methane may provide a practical method for consumption of the two greenhouse gases—CH4 and CO2. Unfortunately, no industrial technology for CO2 reforming of methane has yet been developed, because no effective, economic catalysts have been discovered (7); furthermore, high energy costs may be another drawback preventing commercialization. When the conventional Ni-containing catalyst for steam reforming was used for CO2 reforming, carbon deposits formed on the catalyst, which deactivated rapidly, at least in the absence of steam. A high molar ratio of CO2 to CH4 ($ 3) could be used to reduce the carbon deposition by inhibiting CO disproportionation, but the selectivity to synthesis gas was found to become much lower than that for the stoichiometric CO2 reforming (CO2/CH4 ¼ 1, molar). Therefore, the inhibition of carbon deposition without extra cost and loss of catalyst performance constitutes a major challenge for CO2 reforming of methane. In the 1980s, the oxidative coupling of methane to give ethylene and ethane was reported by Keller and Bhasin (8), whose discovery prompted numerous attempts to convert methane directly—and not only to ethylene and ethane (8), but also to methanol and formaldehyde (9) (Table I). Research on oxidative coupling of methane was motivated by results showing that the methane was

300

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345 TABLE I Gibbs free energy change, DG0 ; for methane transformation reactions (10) Reaction

DG0 (kcal mol21) 400 K

600 K

800 K

1000 K

234.6

235.1

235.8

236.4

218.4

217.1

215.8

214.5

CH4 þ 12 O2 ! CH3 OH

225.4

223.0

220.5

218.0

CH4 þ O2 ! HCHO þ H2 O

269.0

270.0

270.8

271.2

CH4 þ 2O2 ! CO2 þ 2H2 O

2191.3

2191.3

2191.3

2191.3

! CO þ 2H2

225.0

233.9

243.1

252.5

CH4 þ H2 O ! CO þ 3H2

28.6

17.3

5.5

26.5

CH4 þ 12 O2 ! 12 C2 H4 þ H2 O CH4 þ

CH4 þ

1 4 O2

1 2 O2

!

1 2 C 2 H6

þ

1 2 H2 O

converted into hydrocarbons with higher boiling points, which can be more economically transported than methane; the transportation issue is important because substantial reserves of methane are located in remote places. The reported results show that Li/MgO, with or without promoters, is the best known catalyst (11). However, Pasquon (12) reported that the best result obtained in long-run tests has been a C2þ yield of only 15% for methane conversions of 15– 40%, at 1270 – 1370 K and a pressure of 1 –2 bar, when a 5 –10 CH4/O2 molar ratio was used. In the early 1990s, a consensus emerged that it would be very difficult to achieve a significantly better result than that mentioned above for the oxidative coupling to become an economical industrial process. The reason is that the formation of CO2, rather than of more desirable products (ethylene, ethane, methanol, and formaldehyde), is favored thermodynamically (Table I) when the reactions of methane and oxygen become fast enough to be of practical interest (typically at temperatures exceeding 973 K). Consequently, in the early 1990s, interest in the direct processes decreased markedly, and the emphasis in research on CH4 conversion returned to the indirect processes giving synthesis gas (13). In 1990, Ashcroft et al. (13) reported some effective noble metal catalysts for the reaction; about 90% conversion of methane and more than 90% selectivity to CO and H2 were achieved with a lanthanide ruthenium oxide catalyst (L2Ru2O7, where L ¼ Pr, Eu, Gd, Dy, Yb or Lu) at a temperature of about 1048 K, atmospheric pressure, and a GHSV of 4 £ 104 mL (mL catalyst)21 h21. This space velocity is much higher than that employed by Prettre et al. (3). Schmidt et al. (14 –16) and Choudhary et al. (17) used even higher space velocities (with reactor residence times close to 1023 s).

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

301

An advantage of the high space velocities is the relatively low mass-transfer resistances associated with them. The catalytic partial oxidation of methane to CO is exothermic, and even a low conversion to CO2 generates a large amount of heat, which leads to significant temperature gradients (hot spots) in the reactor; the temperature may vary by several hundred kelvin over a distance of only 1 mm from the hot spot. Because the partial oxidation is a fast reaction, it is difficult to remove the heat from the reactor as fast as it is generated, particularly from a large-scale reactor. As a result, the process is potentially hazardous and can lead to explosions. The partial oxidation process requires a pure oxygen feed and, therefore, a unit to prepare oxygen by air separation. Therefore, one of the major research goals for making the catalytic partial oxidation a commercial reality is to reduce the cost of the air separation. The reaction pathways for the partial oxidation reaction are still debated. According to one interpretation, CO2 and H2O are the primary products, and CO is formed by the reaction of CO2 or H2O with CH4; according to another interpretation, CO is produced directly by the reaction of CH4 with O2. In summary, major challenges in the partial oxidation of methane are: (1) designs to avoid excessive thermal gradients (hot spots) in the catalyst bed; (2) reduction of the cost of O2 separation; and (3) elucidation of the reaction pathways as a step toward improved catalyst design. The purpose of this chapter is to provide a critical assessment of the literature regarding the partial oxidation of methane and the CO2 reforming of methane, with emphasis on the following challenging areas: hot spots, O2 separation cost, and the issues of reaction pathways and catalyst selection; we also address the issue of carbon deposition in the CO2 reforming of methane. The reason why we review these two reactions together is that they have many common characteristics, including the catalysts, the products, and CH4 as reactant.

II. Partial Oxidation of Methane II.A. Hot Spots in Catalyst Beds In the early 1990s, several papers (17 –20) reported that one can reach CO and H2 concentrations in excess of those expected at thermodynamic equilibrium by operating the CH4 oxidation reaction at exceptionally high space velocities (GHSV ¼ 52,000 mL (g catalyst)21 h21) in a fixed-bed reactor. The following catalysts were employed: Ni/Yb2O3 (18), Co/rare earth oxide (19), Co/MgO (20), and Ni/Al2O3 (17). However, the actual reaction temperatures (21) could have been much higher than those reported (17– 20). By using an optical pyrometer, Lunsford et al. (22) found that,

302

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

during the catalytic oxidation of methane to CO and H2, the combination of a high space velocity, an exothermic reaction, and an active catalyst (Ni/ Yb2O3) gave rise to steep temperature gradients (hot spots). Furthermore, the temperature of the hot spot was greater by as much as 573 K than the temperature measured with a thermocouple located at a distance of only 1 mm from the hot spot in the catalyst bed. If a temperature lower than that of the hot spot is used to calculate the equilibrium concentrations of CH4, CO, CO2, and H2, one can draw the conclusion that the concentrations of CO and H2 exceeded their thermodynamic equilibrium values. However, if the true maximum (hot spot) temperature is used in the calculation, the observed concentrations are found to be somewhat less than those predicted at equilibrium. Indeed, using a careful temperature measurement method, in which a thermocouple end contacted just the top surface of the catalyst bed, Hu et al. (23,24) found that the CH4 conversion in the presence of Ni/Al2O3 catalyst was less than that predicted by thermodynamic equilibrium. Furthermore, Hu and Ruckenstein (25) observed hot layers (thinner than 1 mm) in NiO/MgO solid solution catalysts and in NiO/Al2O3 and NiO/SiO2 catalysts during the partial oxidation of methane in a fixed-bed reactor. The hottest layers were located at the top of the bed of the NiO/MgO and NiO/Al2O3 catalysts, but they were observed to move down and then up for the NiO/SiO2 catalyst bed. The down-and-up movement resulted in an oscillatory temperature of the NiO/SiO2 catalyst at a given position in the bed (Fig. 1), which was absent when the catalyst was NiO/MgO or NiO/Al2O3 (Fig. 2). The different temperature behaviors of the three catalysts were attributed to the different strengths of the interactions between the metal oxide and the support. Temperature-programmed reduction (TPR) experiments with 4% H2 in argon indicated that the initial reduction temperature was about 3308C for 13.6 wt% NiO/SiO2, which is near that of pure NiO (about 3008C) (26). In contrast, for 13.6 wt% NiO/Al2O3 the initial reduction temperature was high (6708C) and no marked reduction peak could be detected even at 8008C for 13.6 wt% NiO/MgO. These results clearly indicate that there are weak interactions between NiO and SiO2 and much stronger interactions between NiO and Al2O3 and between NiO and MgO. The weak interactions in Ni/SiO2 might have been responsible for the temperature oscillation by allowing a facile redox behavior of the active nickel sites, namely, the oxidation of Ni0 to NiO by O2 and the reduction of NiO to Ni0 by CH4. The strong interactions characteristic of NiO/Al2O3 and NiO/MgO were inferred (25,26) to inhibit in part the redox behavior of the nickel sites. In the case of NiO/SiO2, according to this interpretation, the freshly reduced NiO located at the inlet of the bed became highly active, causing a hot layer to be generated. The high temperature of this hot layer resulted in sintering of the nickel particles, which led to the loss of activity. Therefore, the reaction is inferred to have taken

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

303

Fig. 1. Relationship between catalyst temperature and reaction time in methane partial oxidation catalyzed by Ni/SiO2 (temperature of the gas phase: (a) 1019 K, (b) 899 K, (c) 809 K, (d) 625 K). The reaction was carried out in a fixed-bed reactor (a quartz tube of 2 mm inside diameter) at atmospheric pressure. Before reaction, the feed gas was allowed to flow through the catalyst undergoing heating of the reactor from room temperature to 1073 K at a rate of 25 K min21 to ignite the reaction, and then the reactant gas temperature was decreased to the selected value. Reaction conditions: pressure, 1 atm; catalyst mass, 0.04 g; feed gas molar ratio, CH4/O2 ¼ 2/1; GHSV, 90,000 mL (g catalyst)21 h21) (25).

place in the neighboring section of the catalyst. As a result, a hot layer propagated downward in the reactor. However, the sintered nickel particles were re-dispersed on the SiO2 support when they were reoxidized by O2, because the oxygen concentration is high when the reaction of CH4 with O2 does not take place. After a certain time, the reoxidized layer near the entrance was again reduced by CH4 and became active again, resulting in a hot layer. The following part of reoxidized nickel on SiO2 can be reduced rapidly by H2 and CO generated near the entrance of the reactor. The redox of the Ni/SiO2 catalyst constitutes a cycle of deactivation and reactivation in each part of the catalyst. The hot layer moved downward in the bed during the time required for the reduction of the entrance layer. Consequently, the time scale of the oscillations was determined by the time scale of the reduction – oxidation process. Recently, such a temperature oscillation was also observed by Zhang et al. (27,28) with nickel foils. Furthermore, Basile et al. (29) used IR thermography to monitor the surface temperature of the nickel foil during the methane partial oxidation reaction by following its changes with the residence time and reactant concentration. Their results demonstrate that the surface temperature profile was strongly dependent on the catalyst composition and the tendency of nickel to be oxidized. Simulations of the kinetics (30) indicated that the effective thermal conductivity of the catalyst bed influences the hot-spot temperature.

304

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

Fig. 2. Relationship between catalyst temperature and reaction time for reaction catalyzed by Ni/Al2O3(- - -) and Ni-MgO solid solutions (—); temperature (K) of the gas phase: (a) 1019; (b) 899: (c) 809; (d) 625. The reaction was carried out in a fixed-bed reactor (a quartz tube of 2 mm inside diameter) at atmospheric pressure. Before reaction, the feed gas was allowed to flow through the catalyst undergoing heating of the reactor from room temperature to 1073 K at a rate of 25 K min21 to ignite the reaction, and then the reactant gas temperature was decreased to the selected value. Reaction conditions: pressure, 1 atm; catalyst mass, 0.04 g; feed gas molar ratio, CH4/O2 ¼ 2/1; GHSV, 90,000 mL (g catalyst)21 h21) (25).

Researchers have attempted to minimize thermal gradients, for example, by using fluidized-bed reactors (31 –33). Olsbye et al. (31) investigated methane partial oxidation in a fixed bed and in a fluidized-bed reactor with a 1.5 wt% Ni/ Al2O3 catalyst operated at 973 K, with a feed flow rate of about 400 (STP) mL min21 (CH4/O2/N2/H2O ¼ 2/1/2/0.5, molar) and a catalyst volume of 17 mL. They observed that the maximum temperature difference was only 282 K in the fluidized bed, but 423 K in the fixed bed, indicating that the fluidizedbed reactor is a good heat exchanger because of the rapid mixing of the fluid and the catalyst. Another way to minimize the temperature gradient (34 –40) is to combine the exothermic partial oxidation with an endothermic reaction. Ioannides and Verykios (34) developed a novel reactor consisting of a ceramic tube with metal catalyst films deposited on the inner and outer surfaces. The CH4/O2 feed enters into the tube, and a large fraction of the heat generated by the methane combustion reactor is transferred through the tube wall towards the outer catalyst film, where an endothermic reforming reaction takes place. With this design, the temperature in the combustion zone is controlled and the hot spots are significantly reduced in magnitude.

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

305

Coupling of the endothermic CO2 reforming of methane with the exothermic catalytic partial oxidation of methane can, in addition to overcoming the hazard of overheating, also provide a control of the H2/CO ratio and thus the selectivity for various Fischer – Tropsch synthesis products. Aschcroft et al. (35) carried out this combination of reactions with an Ir/Al2O3 catalyst, obtaining synthesis gas yields of up to 90% (Table II). However, they found that when nickel-containing catalysts were used, carbon deposits were formed rapidly, except when an excess of CO2 was used. Choudhary et al. (41,42) reported that a NiO – CaO catalyst for 15 h exhibited a conversion . 95%, with 100% CO selectivity and . 90% H2 selectivity, without catalyst deactivation caused by carbon deposition. Furthermore, Ruckenstein and Hu (37) found that the reduced NiO/MgO catalyst provided a high activity and selectivity, as well as excellent stability in the combination process, even when no excess of CO2 was used. They carried out the combined reaction catalyzed by each of the following: a NiO/MgO solid solution, NiO/Al2O3, and NiO/SiO2. A CH4 conversion of about 90% and selectivities to CO and H2 of about 98% were achieved at 1063 K and a GHSV of 90,000 mL (g catalyst)21 h21 (O2/CO2/CH4 ¼ 14.5/26.9/58.6) when a reduced NiO/MgO solid-solution catalyst was used. Almost no change in activity or selectivity occurred during 50 h of reaction. Compared with the reduced NiO/ MgO, the reduced NiO/SiO2 and NiO/Al2O3 catalysts provided lower activities and stabilities. Furthermore, Ruckenstein and Hu (37) observed a decrease in the CH4 conversion with increasing space velocity, whereas during the partial oxidation alone, because of the hot spots, it would have increased (43). This observation implies that the coupling can, indeed, control the thermal behavior of

TABLE II Results of catalytic reactions with mixtures of CH4, O2, and CO2 of different compositions in the presence of 1 wt% Ir/Al2O3 at 1050 K (35) Feed composition (mol%)

CH4 converted (%)

CO2 converted (%)

H2 yield (%)

CO yield (%)

CH4

CO2

O2

64.4

3.5

32.1

92

9

89

86

59.4

20.0

20.6

87

83

81

86

58.3

23.7

18.0

84

83

81

84

58.0

28.0

14.0

83

90

79

85

49.8

48.8

1.4

91

87

91

89

Total gas hourly space velocity, 2 £ 104 mL (mL catalyst)21 h21; pressure, 1 atm. In all the cases, the oxygen conversions were .99.7%.

306

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

the reactor. Ruckenstein and Wang (39) found that the Co/MgO solid solution is also an effective catalyst for the combined reaction. Steam reforming of methane, which like CO2 reforming is endothermic, has also been combined with the exothermic partial oxidation of methane (44– 47). This combination process is usually called “autothermal reforming”, because no heat addition is required for the reforming reaction. For example, ExxonMobil (44– 46, 48– 50) extended its experience with fluidized-bed catalytic cracking to the synthesis gas production, developing a process in which the steam reforming was combined with partial oxidation of the natural gas in a single fluidized-bed reactor.

II.B. Minimizing O2 Purification Costs Although the partial oxidation of methane with air as the oxidant would at first seem to be a potential alternative to the steam-reforming process, the downstream processing requirements in the conventional process do not tolerate nitrogen (because the cost of compression of synthesis gas diluted by nitrogen to pressures . 20 atm, which is necessary for downstream industrial processes, is high), and, therefore, pure oxygen must be used. An important advance in the direction of making air a feedstock resulted from the use of an inorganic membrane reactor (51 –75). The reactor consists of a dense ceramic membrane (made from mixtures of ionic and electronic conductors, such as SrFeCo0.5Ox (51)) that is permeable only to oxygen; application of this reactor can, in principle, reduce the entire synthesis gas process to a single step, allowing elimination of the oxygen plant and decreasing the total cost of the synthesis gas production by 25 –40%. For this reason, this inorganic membrane process has attracted significant commercial interest. Solid electrolytes are materials that exhibit high ionic conductivities (76). If a solid electrolyte is a pure ionic conductor, the transference number for ions is two or more orders of magnitude greater than that for electrons. Yttria-stabilized zirconia, a pure ionic conductor, is the classical solid electrolyte for solid-state transport of oxygen. However, a system based on a classical solid electrolyte for ionic oxygen transport requires electrodes to transfer the electrons to the reduction interface from the oxidation interface (Fig. 3a). In contrast, the perovskites of the ABO3 type (with the CaTiO3 structure) with dopants in the A and/or B sites, called mixed ionic/electronic conductors (MIEC), provide high conductivities for both oxygen ions and electrons (54– 62) (Fig. 3b). The MIEC membrane can be used for the O2 separation without electrodes. The driving force for the overall oxygen transport is the gradient of the oxygen partial pressure across the membrane (77). The dissociation and ionization of oxygen to generate oxygen ions, by capturing the electrons provided by accessible surface electronic states, occur at the oxide surface at the high-pressure feedside. The flux of oxygen ions and the reverse flux

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

307

Fig. 3. Oxygen transport in solids. O2 is dissociated and ionized at the reduction interface to give O22 ions, which are transferred across the solid to the oxidation interface, at which they lose the electrons to return back to O2 molecules that are released to the stream. (a) In the solid electrolyte cell based on a classical solid electrolyte, the ionic oxygen transport requires electrodes and external circuitry to transfer the electrons from the oxidation interface to the reduction interface; (b) in the mixed conducting oxide membrane, the ionic oxygen transport does not require electrodes and external circuitry to transfer the electrons to the reduction interface from the oxidation interface, because the mixed conductor oxide provides high conductivities for both oxygen ions and electrons.

of electronic charge carriers across the MIEC membrane constitute a chargecompensation process. The individual oxygen ions from the high-pressure feedside separate from their electrons and recombine again, at the low-pressure permeateside, to form O2 molecules that are released into the permeate stream. Therefore, because of its ability to conduct both oxygen ions and electrons, the MIEC membrane can operate without electrodes attached to the oxide surface and without external circuitry. Extensive research has been carried out with the acceptor-doped perovskite oxides with the generic formula La12xAxCo12yByO32d (where A ¼ Sr, Ba, or Ca and B ¼ Fe, Cu, or Ni) (77). Teraoka et al. (54,55,63) were the first to report very high oxygen fluxes through the cobalt-rich perovskites that can become highly oxygen anion defective at elevated temperatures and reduced oxygen partial pressures. The oxygen-ion conductivity in these perovskites can be 1 – 2 orders of magnitude greater than those of stabilized zirconias at elevated temperatures, although in the usual ranges of temperature and oxygen partial pressure, the electronic conduction of the perovskite remains predominant (78,79). In the early 1990s, Balachandran et al. (51,64,65) of the Argonne National Laboratory, in collaboration with Amoco (now part of BP), investigated the partial oxidation of methane using membrane materials consisting of Sr –Fe –Co – O mixed oxides with the perovskite structure, which have high oxygen permeabilities. In their experiments (51,66), the membrane tubes, which were

308

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

Fig. 4. Configuration of a ceramic membrane reactor for partial oxidation of methane. The membrane tube, with an outside diameter of about 6.5 mm and a length of up to about 30 cm and a wall thickness of 0.25–1.20 mm, was prepared from an electronic/ionic conductor powder (Sr–Fe– Co–O) by a plastic extrusion technique. The quartz reactor supports the ceramic membrane tube through hot Pyrex seals. A Rh-containing reforming catalyst was located adjacent to the tube (51).

prepared from an electronic/ionic conductor powder (Sr – Fe– Co –O) by a plastic extrusion technique, were investigated for their performance in the quartz reactor sketched in Fig. 4. The quartz reactor supports the ceramic membrane tube with hot Pyrex seals. A Rh-containing reforming catalyst was located adjacent to the tube. In this reactor, air could be used directly, because the membrane itself carried out the separation of oxygen from air. The electrons of the membrane combine with the oxygen from air to generate oxygen anions. The ions migrate through the membrane, from the air side to the methane side. At the methane side, the electrons are stripped from the ions, which are thus converted into oxygen atoms that combine with methane to form the synthesis gas. The freed electrons migrate back to the air side of the membrane, generating fresh oxygen anions, and so on. The experimental results show that the performance of the membrane was strongly dependent on the composition of the material. The most promising material had the composition SrFeCo0.5Ox. This membrane operated in a partial oxidation reactor for more than 1000 h at 1123 K (Fig. 5), whereas other mixed-oxide membranes fractured rapidly. A methane conversion of 98% with a 90% CO selectivity was thus achieved. Another advantage of the membrane reactor is that the process does not involve the handling of potentially explosive CH4/O2 mixtures. Other early contributions to the membrane processes for partial oxidation of methane include the following: (a) the La0.2Sr0.8Fe0.8Cr0.2Ox membrane of Standard Oil Company at Ohio (now part of BPAmoco) (67), which remained stable for more than 1000 operating hours at 1373 K, and (b) a brownmillerite membrane with the general composition A2B2O5 (where A and B were not disclosed), consisting of a layer of BO6 octahedra sharing vertices with a layer of BO4 tetrahedra (68), which was tested for more than 3000 operating hours at 1173 K and 1 atm with a CO selectivity . 96% and a CH4 conversion . 80%. A group at Worcester Polytechnic Institute (69) also investigated the partial

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

309

Fig. 5. Methane conversion and oxygen flux during partial oxidation of methane in a ceramic membrane reactor. Reaction conditions: pressure, 1 atm; temperature, 1173 K, feed gas molar ratio, CH4/Ar ¼ 80/20; feed flow rate, 20 mL min21 (NTP); catalyst mass, 1.5 g; membrane surface area, 8.4 cm2 (51).

oxidation of methane to give synthesis gas using a mixed-conducting La(12x)AxFe0.8Co0.2O32d perovskite dense membrane reactor at 1123 K, in which the oxygen was separated from air and simultaneously fed into the methane stream. The steady-state oxygen permeation rates for membranes in non-reacting air/ helium experiments were in the sequence La0.2Ba0.8Fe0.8Co0.2O32d . La0.4 Ba0.6Fe0.8Co0.2O32d . La0.4Ca0.6Fe0.8Co0.2O32d . La0.4Sr0.6Fe0.8Co0.2O32d. By packing a 5% Ni/Al2O3 catalyst directly on the reaction-side surface of the membrane, the researchers obtained a fivefold increase in O2 permeation and a fourfold increase in CH4 conversion. The oxygen, which was continuously transported from the air side, appeared to stabilize the membrane interior, and the reactor could be operated for up to 850 h (69,70). Recently, Li et al. (71) demonstrated a promising application of a Ba0.5Sr0.5 Co0.8Fe0.2O32d membrane for oxygen separation characterized by a high permeation flux (1.1 mL cm22 min21 at 1123 K) and stability (leak-free during partial oxidation). A membrane reactor, prepared from a Ba0.5Sr0.5Co0.8Fe0.2 O32d membrane (Fig. 6), was applied successfully to the partial oxidation of methane (with LiLaNiOx/g-Al2O3 containing 10 wt% Ni as catalyst, located on the top of the membrane) at 1148 K for about 500 h without failure, with a methane conversion . 97% and a CO selectivity . 95% (Fig. 7) (72). A novel dense catalytic membrane reactor, prepared from the stable conducting perovskite BaCo0.4Fe0.4Zr0.2O32d and the catalyst LiLaNiO/g-Al2O3 also

310

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

Fig. 6. Configuration of a ceramic membrane reactor for partial oxidation of methane. The membrane disk was prepared by pressing Ba0.5Sr0.5Co0.8Fe0.2O32d oxide powder in a stainless steel module (17 mm inside diameter) under a pressure of (1.3–1.9) £ 109 Pa. The effective area of the membrane disk exposed to the feed gas (CH4) was 1.0 cm2 (72).

demonstrated excellent performance for partial oxidation (73). This membrane reactor was characterized by a short induction period (2 h), high CH4 conversion (98%) and CO selectivity (about 99%), and excellent stability (more than 2200 operating hours) at 1123 K. Since 1997, to accelerate the membrane technology towards commercialization, two major alliances have been formed, one comprising Amoco (now part of BP), Praxair, Statoil, Sasol, and Philips, and the other (a US Department of Energy

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

311

Fig. 7. Methane conversion, CO selectivity, and oxygen flux through the ceramic membrane during the partial oxidation of methane in a ceramic membrane reactor (see Fig. 6). Reaction conditions: temperature, 1148 K; catalyst, 300 mg of LiLaNiOx/g-Al2O3; air flow rate, 300 mL min21 (NTP); feed gas molar ratio, CH4/He ¼ 1/1; feed flow rate, 42.8 mL min21 (NTP) (72).

cost-shared project) made up of Air Products, Arco (now part of BP), Argonne National Laboratory, Babcock and Wilcox, Ceramate (Salt Lake City), ChevronTexaco, Eltron Research, Norsk Hydro, Pacific Northwest National Laboratory, Pennsylvania State University, and the University of Pennsylvania (74). Notwithstanding the extensive research, there are still hurdles to overcome (80– 82). Although the mixed conducting membranes offer high oxygen fluxes, they are mechanically and chemically less stable than the traditional stabilized zirconias. Furthermore, the integration of a ceramic membrane into large-scale production units will be difficult, because the ceramics break easily and are not easily manufactured without microscopic voids and fractures. It is also difficult to connect them to other, more flexible materials such as steel pipes. These critical issues represent major challenges to the commercialization of MIEC membrane reactors for the partial oxidation of methane. Therefore, a team, led by the University of Alaska-Fairbanks, was formed to study these practical issues (75), including the composition of the ceramic membrane, seals that would join the ceramic and metal materials, membrane performance, and development of a ceramic that would resist warping and fracturing at the high temperatures of the conversion process. Another way to eliminate the oxygen plant is to react a metal oxide with methane to yield the synthesis gas in a fluidized-bed reactor (83 – 86). Experiments have shown that copper oxide readily oxidizes methane to carbon monoxide and hydrogen with high selectivity at a temperature of about 1200 K and that the reduced CuO can be reoxidized with air. Lewis et al. (83– 86)

312

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

proposed a process using two interconnected fluidized beds—a reactor for the hydrocarbon oxidation by the metal oxides (Step 1) and a regenerator for the reoxidation of the reduced metal oxide by air (Step 2). The major advantage of this process is that air can be used directly without pre-separation. A high conversion of about 95% and a selectivity of 90% were thus achieved (83– 86). However, metal oxide sintering during the reduction –oxidation cycles could be a difficulty.

II.C. Catalysts In the 1940s, Prettre et al. (3) reported the formation of synthesis gas via the catalytic partial oxidation of CH4 catalyzed by a 10 wt% refractory supported nickel, at temperatures between 973 and 1173 K. Thermodynamic equilibrium corresponding to the catalyst bed exit temperature was achieved under all conditions investigated. In 1970, Huszar et al. (87) examined the effect of diffusion on methane partial oxidation catalyzed by a single grain of Ni/mullite catalyst in the temperature range of 1033 – 1173 K and examined the ignition and extinction characteristics of this catalyst. They observed that the nickel catalyst deactivated in an oxidative environment but could recover on reduction. In 1984, Gavalas et al. (88) investigated the effects of the calcination temperature, prereduction, and feed ratio on the reaction of CH4/O2 mixtures catalyzed by NiO/aAl2O3 at 843– 1033 K. However, under their experimental conditions, the main products were CO2 and H2O. Since 1990, researchers (89 – 148) have continued to examine nickel-containing catalysts for the partial oxidation of methane, and they also started to use noble metals as catalysts. In 1990, Ashcroft et al. (13) reported a methane conversion of about 90% and more than 90% selectivity to CO and H2 at 1043 K, atmospheric pressure, and at the high GHSV of 4 £ 104 mL (mL catalyst)21 h21 for a reaction catalyzed by lanthanide ruthenium oxides, such as Pr2Ru2O7, Eu2Ru2O7, Gd2Ru2O7, Dy2Ru2O7, or Lu2Ru2O7. In 1992, Hickman and Schmidt (14) used platinum monoliths to achieve high selectivities to CO and H2 in the partial oxidation of methane. In the following 10 years, various noble metal catalysts have been examined (Table III) (89 –106). Compared with the nonnoble metal catalysts, the noble metals exhibit high stability with excellent activities and selectivities. The major drawback of the noble metal catalysts is their high cost, which restricts their potential use in industrial processes. Non-noble metal catalysts, particularly those containing nickel, have also been investigated extensively since 1990. Lunsford et al. (107) examined a 25 wt% Ni/Al2O3 catalyst in the temperature range 723– 1173 K. Carbon monoxide selectivities approaching 95% and virtually complete conversion of the methane were achieved at temperatures above 973 K. The authors observed that, under their operating conditions, the calcined catalyst bed consisted of

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345 TABLE III Noble metal catalysts for partial oxidation of methane Metal Rh

Pt

Support

References

Al2O3

(89,94,102,104,106,126)

SiO2

(101)

MgO

(101)

a-Al2O3 monolith

(16,105)

Al2O3

(89,91,99,126,148)

CeO2/Al2O3

(91)

MgO

(95)

ZrO2

(148)

CeO2

(99)

CeO2/ZrO2

(148)

a-Al2O3 monolith

(16)

Pd

Al2O3

(89,126)

a-Al2O3 monolith

(16)

Ir

TiO2

(90,92,96)

a-Al2O3 monolith

(16)

Al2O3

(126)

Eu2O3

(126)

SiO2

(43,93)

Al2O3

(10,93,126,127)

YSZ (yttria-stabilized zirconia)

(93)

Pt sponge

Ru

Re

(103)

TiO2

(93,97,100)

a-Al2O3 monolith

(16)

Pr2O3

(13)

Sm2O3

(13)

Eu2O3

(13)

Gd2O3

(13)

Tb2O3

(13)

Dy2O3

(13)

Tm2O3

(13)

Yb2O3

(13)

Lu2O3

(13)

a-Al2O3 monolith

(16)

313

314

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

three regions, NiAl2O4 (upstream, section), NiO þ Al2O3 (middle section), and reduced Ni/Al2O3 (downstream section). In the upstream section of the reactor, the CH4/O2/He feed contacted NiAl2O4, which exhibited only a moderate activity for the complete oxidation of methane to CO2 and H2O. The next section of the reactor contained NiO þ Al2O3, which catalyzed the complete exothermic oxidation of methane to CO2. Because of the complete consumption of O2 in the second section, the third (downstream) section of the catalyst bed consisted of a reduced Ni/Al2O3. The formation of the CO and H2 products, corresponding to thermodynamic equilibrium at the temperature of the bed exit, occurred in this section, as a result of the reforming reactions of CH4 with CO2 and H2O produced during the complete oxidation reaction catalyzed by the NiO/Al2O3. Choudhary et al. reported a high conversion of CH4 and high selectivities to CO and H2 with Ni/CaO (17a), Ni/Al2O3 (17b), NiO-rare earth oxide (108), and Co/rare earth oxide catalysts (19). Hu et al. (23) used a Ni/Al2O3 catalyst for the adiabatic partial oxidation of methane. The nickel- or cobalt-containing catalysts exhibited high activities and selectivities to synthesis gas from CH4/O2 mixtures. The major problem encountered with these non-noble metal catalysts is their relatively low stability (109– 111). The main causes of the deactivation of the catalysts are carbon deposition and metal sintering in the catalyst. Nevertheless, numerous effective nickel-containing catalysts have been developed by incorporation in suitable supports (111– 116), such as La2O3 (111), MgO (112, 113), SrTiO3 (114), and CeO2 (115); effective promoters (117 – 119), including La2O3 (117,118), Li2O (118), and iron oxide (119); and novel preparation methods (120 –125), such as a solid phase crystallization method (120), a sol – gel method (122), and a citrate method (125). However, because the high stabilities reported for these effective nickel-containing catalysts were based on short-term tests (, 100 h), it is unclear how stable these catalysts will be in long term tests (. 1000 h), which is the first step that any candidate catalyst for commercialization must pass.

II.D. Reaction Pathways In the last decade, numerous attempts have been made to understand the mechanism of the partial oxidation of methane (3,13 – 15,17,25,37,97,107, 128– 137,142 –148). Mechanistic investigations of the partial oxidation are still challenging, because this exothermic reaction is very fast and causes extremely high catalyst temperature rises, so that the usual methods of investigation are unsuitable. Two kinds of pathways have been suggested: (i) a combustion-reforming pathway, in which CO2 and H2O are the primary products, and CO and H2 are

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

315

formed by their reactions with CH4 (3,14,97,107,128,132,133); and (ii) a pyrolysis pathway, in which CO is the primary product formed by the pyrolysis of methane, CH4 ! CHx þ ð2 2 12 xÞH2 ; followed by the oxidation of carboncontaining species to give CO without the pre-formation of CO2 (17,129,130, 134 –137). Thus, the major questions regarding these reactions are: (1) is CO or CO2 the primary product? (2) What is the rate-determining step? (3) What are the intermediate species? (4) How does the state of the catalyst change during reaction? We review these issues in the following sections. II.D.1. Changes in Catalyst During Reaction The catalyst surface structure depends on the reactants in contact with it. During steady-state experiments, the catalyst surface may reach an equilibrium with the reactants at various positions in the reactor, and so steady-state methods provide little information about the surface state of the catalyst. On the other hand, pulse methods, in which a small amount of reactants is injected into the reactant stream, do not affect the surface of the catalyst significantly during a single pulse. Therefore, during the first pulse, the reaction can be attributed to the original state of the catalyst. As additional pulses are introduced, the catalyst surface gradually changes. Therefore, changes in the selectivities and conversions as a function of the number of pulses are indicative of the changes in the catalyst. Thus, Hu and Ruckenstein (134) determined the selectivities and conversions as a function of the number of pulses of CH4 and O2 using mass spectrometry to analyze the products. The catalyst was an unreduced NiO/La2O3 or one reduced in H2. As shown in Fig. 8, when the catalyst was unreduced, the CH4 conversion increased gradually with the number of pulses, reaching the constant value of about 18% after the ninth pulse. When the catalyst was initially reduced, the CH4 conversion was the greatest for the first pulse and after the ninth pulse reached the same constant value as for the initially unreduced catalyst. The change of the CO selectivity with the number of pulses of CH4 and O2 was found to be similar to that observed for the CH4 conversion. This comparison indicates that the initial oxide and reduced states of the catalyst changed towards the same working state as the number of CH4/O2 pulses increased. In other words, the oxide state of the catalyst was partially reduced during catalysis, and the reduced catalyst was partially oxidized during catalysis. Presumably, a redox equilibrium was finally attained between the catalyst and the reactant stream. Furthermore, the curves of oxygen coverage on a reduced Ni/La2O3 catalyst with time during a pulse of CH4/O2 indicated that oxygen-containing species were easily generated on the reduced catalyst, and that, after a pulse of CH4 and O2 had reacted completely, oxygen-containing species were still present on the catalyst surface, hence that the reduced catalyst had been partially oxidized (129). This inference is consistent with the X-ray diffraction data of Lunsford et al. (107), which showed

316

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

Fig. 8. CH4 conversion as a function of the number of CH4/O2 pulses for partial oxidation of CH4 catalyzed by Ni/La2O3. Reaction conditions: temperature, 873 K; catalyst, 20 mg of 20 wt% Ni/La2O3 loaded in a fixed-bed flow reactor; feed gas, 0.9 mL CH4/O2 (molar ratio 2/1) in each pulse; carrier gas, helium (flow rate,100 mL min21) (134).

that both reduced and oxidized nickel were present in a Ni/Al2O3 catalyst used for CH4 oxidation. Because the reactor was a fixed bed, the change in the catalyst resulting from the interactions between the catalyst and the stream containing reactants and products were non-uniformly distributed along the catalyst bed (138). As the catalyst was reduced, the CH4 conversion increased. This result implies that the reduced nickel is more active than the oxidized nickel for CH4 activation, and that in the reaction between CH4 and the lattice oxygen of NiO, the CH4 conversion increased when NiO was partially reduced (139,140). Campbell et al. (141) reported that the reaction probability of methane on NiO films is significantly lower than that on a clean Ni(100) surface. Furthermore, results of experiments with deuterium-methane pulses showed that CH4 easily dissociates into CHx and H on a reduced nickel-containing catalyst, whereas such a dissociation cannot take place on the catalyst in the oxidized state (131). One can, therefore, conclude that the reduced nickel, which might be a zero valent nickel, constitutes the principal active site for the partial oxidation of methane.

II.D.2. Which is the Primary Product, CO or CO2? To discriminate between the combustion-reforming mechanism and the pyrolysis mechanism, one must clarify whether CO or CO2 is the primary product (or whether both are). Typically, to discriminate between primary and secondary

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

317

products in a conventional continuous flow reactor, one changes the reactant – catalyst contact time by changing the space velocity of the reactant stream, expecting the primary products but not the secondary products to be observed in the limit, as the contact time approaches zero. However, this method is not straightforwardly suitable for the partial oxidation of methane, because when the space velocity of the reactants is markedly changed, it is difficult to maintain the catalyst temperature unchanged as a consequence of the large differences in heats of reaction at different space velocities. Therefore, pulse methods were used to determine the dependences of the CO and CO2 selectivities on the residence time, by changing the carrier gas flow rate. Because the amount of reactant in a pulse was small, no significant differences in the catalyst temperature resulted from injection of the pulses. Pulse reaction experiments with the Ni/La2O3 catalyst (134) showed that when the space velocity of the carrier gas was changed from 257,000 to 400,000 mL (g catalyst)21 h21, the selectivity to CO increased gradually from 41 to 46% at 873 K, whereas that for CO2 decreased. This comparison implies that CO formation is favored by short residence times, consistent with the suggestion that CO is the primary product and CO2 a secondary product, formed from CO. This observation supports the pyrolysis mechanism by which CO is generated by the oxidation of C, formed via the CH4 dissociation on the Ni/La2O3 catalyst. Furthermore, pulse transient response experiments (129) show that the initial time at which CO was detected is shorter than that of CO2 by about 0.2 s for a CH4/O2 (2/1) pulse at 773 K, in the presence of a reduced Ni/La2O3 catalyst. This result indicates that either the CO generation occurs earlier, or the CO desorption is faster than that of CO2. However, the response curve to a pulse of pure CO and the response curve to a pulse of pure CO2 were similar when a reduced Ni/La2O3 catalyst was used. Consequently, the delay of the CO2 generation relative to that of CO is not caused by desorption, but by the earlier generation of CO. Therefore, the combination of experiments demonstrates that CO is the primary product. Shen et al. (142) used an isotopic transient technique and XPS to investigate the partial oxidation of CH4 to synthesis gas on a Ni/Al2O3 catalyst at 973 K. The results show that CH4 can decompose easily and quickly to give H2 and NixC on the reduced catalyst, and that NixC can react rapidly with NiO, formed by the oxidation of nickel by O2 to give CO or CO2, depending on the relative concentration of NixC around NiO on the catalyst surface. The conclusion drawn by the authors (142) was not only that H2 and CO are primary products in the partial oxidation of CH4, but also that most of the CO2 is also the primary product of the surface reaction between NixC and NiO. In contrast, the kinetics results of Verykios et al. (143) indicated that the reaction on the Ni/La2O3 catalyst mainly takes place via the sequence of total oxidation to CO2 and H2O, followed by

318

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

the reforming reactions to give synthesis gas, whereas CO formation by the direct route was observed at very low oxygen partial pressures. The partial oxidation of methane to synthesis gas on a Ru/TiO2 catalyst was examined by combining non-steady-state and steady-state isotopic transient experiments with in situ DRIFT spectroscopy (144). The authors showed that the primary product of the reaction was CO, which resulted from the surface reaction between carbon and adsorbed atomic oxygen on metallic Ru sites, with CO2 being formed by the oxidation of CO on the oxidized sites. The unique ability of the Ru/TiO2 to catalyze the direct formation of H2 and CO was attributed to its high resistance to oxidation under the conditions of partial oxidation of methane. However, Weng et al. (145) proposed that on Ru/SiO2 the dominant pathway to synthesis gas is via the sequence of total oxidation of CH4 followed by the reforming of the unconverted CH4 by CO2 and H2O; the prevalence of this pathway can be attributed to the high oxygen affinity of ruthenium. It was reported (146) that the pathway of partial oxidation of methane on a rhodium-containing catalyst depends strongly on the support material. On the basis of a pulsed reaction and temperature-jump measurements, Nakagawa et al. (146) proposed that, on Rh/TiO2 and Rh/Al2O3 catalysts, the endothermic decomposition of CH4 to H2 and deposited carbon or CHx first takes place at the upstream end of the catalyst bed, followed by the oxidation of the deposited carbon or the CHx species to COx. However, on Rh/SiO2, synthesis gas was produced by a two-step pathway consisting of a highly exothermic complete oxidation of methane to H2O and CO2, followed by the endothermic reforming of methane by H2O and CO2. In contrast, in situ time-resolved IR spectroscopy showed that on the Rh/SiO2 catalyst, the synthesis gas was formed principally by the direct oxidation of CH4—hence CO was the primary product (145). CO and H2 produced as primary products were also observed for the reaction catalyzed by a rhodium sponge at temperatures from 873 to 1023 K (147). It was suggested that the partial oxidation of methane on Ir/TiO2 (146), Pt/ Al2O3, Pt/ZrO2, and Pt/Ce – ZrO2 (98) takes place by a two-step pathway, consisting of a highly exothermic complete oxidation of methane to H2O and CO2 followed by the endothermic reforming of methane with H2O and CO2. Thus, in summary, we infer that there is not a simple answer to the question “which is the primary product, CO or CO2?”. The complexity might suggest that the reaction pathway depends not only on the catalyst composition but also on the reaction conditions. II.D.3. CHx Species and Rate-Determining Steps Although numerous authors (17,129,130,134 –137) suggested that the dissociation of methane constitutes the first step in the methane oxidation by the pyrolysis mechanism, it is important to provide direct evidence for the formation

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

319

of the CHx species. Deuterium-methane pulses were used to obtain such evidence (131). The experiments showed that only CH4 and CD4 free of CHxDy (x þ y ¼ 4; x . 0; y . 0) were present in the product obtained in the presence of the unreduced nickel-containing catalyst, whereas besides CH4 and CD4, CHxDy was detected in the product obtained with the reduced catalyst (3% CH4, 31% CH3D, 38% CH2D2, 20% CHD3, and 8% CD4) (131). Whereas 28% of the methane (CH4 and CD4) in the feed gas was converted to CO and CO2, a larger amount (65%) of the methane participated in the isotopic exchange reaction on the reduced catalyst. This result demonstrates that on the reduced catalyst the isotopic exchange reaction was faster than the partial oxidation. In contrast, on the unreduced catalyst, no isotopic exchange reaction occurred. Hence, CH, CH2, and CH3 species are inferred to form on the reduced catalyst, demonstrating that the reaction follows the pyrolysis mechanism. Furthermore, on the reduced catalyst, the amount of methane involved in the exchange between CH4 and CD4 was found to be greater than that involved in the conversion to CO and CO2. This observation demonstrates that the exchange between CH4 and CD4 was faster than the conversion of methane to CO and CO2, and hence that the methane cleavage cannot be the rate-determining step. It is inferred that, instead, the ratedetermining step is the reaction of the CHx species with oxygen. Considering the oxidation of surface carbon as the rate-determining step, Hu and Ruckenstein (129) obtained an activation energy of 30.5 kcal mol21 for the partial oxidation from the concentration curves of the C and O species on the nickel-containing catalyst against reaction time. This activation energy is consistent with the theoretical value of 33 kcal mol21 obtained for the CðsÞ þ OðsÞ ! COðsÞ reaction on Ni(111) (149), and this result provides additional support that the oxidation of the surface carbon species is rate-determining. Deuterium effects were also used to identify the rate-determining step for the partial oxidation (137,150 –152). By replacing the reactants CH4 þ O2 with CD4 þ O2, Tang et al. (150) examined the deuterium isotope effect in the partial oxidation of methane on Pt/a-Al2O3 in the temperature range of 823 –923 K by using pulses of reactant and analysis of the product by mass spectrometry. No deuterium isotope effect was observed for CH4 conversion, whereas the CO formation exhibited a normal deuterium isotope effect, indicating that the surface reaction between the adsorbed hydrocarbon species and adsorbed oxygen species to give CO may be a relatively slow step. In contrast, on the rhodium-containing catalysts at 973 K, normal deuterium isotope effects were observed for methane conversion and CO yield, but no effect on the CO selectivity was detected (137). For the partial oxidation of methane to synthesis gas catalyzed by Ru/TiO2 at 903 K, Elmasides and Verykios (144) also found deuterium isotope effects on the CH4 consumption rate ðRH =RD ¼ 1:6Þ and CO formation rate ðRH =RD ¼ 1:9Þ; but no effect on CO2 formation. Therefore, they suggested that the processes of CH4 consumption and CO production are affected by the C – H cleavage of CH4 and,

320

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

consequently, are slow or rate-determining steps. However, it is worthwhile to note that, to identify the rate-determining steps, one must ensure that the deuterium isotope effects are obtained under non-equilibrium reaction conditions. This restriction applies because the equilibrium partial oxidation of methane also involves deuterium isotope effects on CH4 conversion and CO formation (153), which do not depend on the rate-determining steps. II.D.4. Comparison of Reactions on Reduced and Unreduced Catalysts Results of isotopic pulse experiments showed that different mechanisms occurred on the reduced and unreduced nickel catalysts (130). Hu and Ruckenstein (130) demonstrated that the reaction on the unreduced catalyst involved gas-phase or weakly adsorbed CH4 and strongly adsorbed or lattice oxygen (an Eley – Rideal mechanism). Furthermore, these authors found that the methane conversion on the unreduced catalyst took place predominantly (18%) by its reaction with the lattice oxygen of the catalyst and also by its reaction with the oxygen from the gas feed stream (12%). This comparison implies that the reaction with the lattice oxygen was more facile than that with the gas-phase oxygen. According to the above results, it is reasonable to suggest that CH4 is oxidized mainly by the oxygen of the lattice, which is replenished by the oxygen from the gas phase. This mechanism can be expressed as follows: CH4 ðgÞ þ 4NiO ! CO2 ðgÞ þ 2H2 OðgÞ þ 4Ni

ð4Þ

2Ni þ O2 ðgÞ ! 2NiO

ð5Þ

Using a temperature-programmed surface reaction (TPSR) technique, Li et al. (154) showed that this complete oxidation of methane took place on the NiO catalyst during the CH4/O2 reaction. Weng et al. (145) used in situ microprobe Raman and in situ time-resolved IR spectroscopies to obtain a relationship between the state of the catalyst and the reaction mechanism. These authors showed that RuO2 in the Ru/SiO2 catalyst formed easily at 873 K in the presence of a CH4/O2/Ar (2/1/45, molar) mixture and that the dominant pathway to synthesis gas was by the sequence of total oxidation of CH4 followed by reforming of the unconverted CH4 by CO2 and H2O. Thus, these results indicate that the oxidation of methane takes place principally by the combustion mechanism on the oxidized form of this catalyst. Hu and Ruckenstein’s results (130) showed that on the reduced nickelcontaining catalyst, the reaction took place by a Langmuir– Hinshelwood mechanism involving adsorbed CH4 and oxygen species. Furthermore, they indicated that a slow dynamic redox process consisting of lattice oxygen formation and its reduction by carbon species was at least partly responsible for the CO formation.

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

321

In summary, one can conclude that even for one catalyst, the oxidation of methane follows different reaction pathways on the reduced and oxidized forms. Because the state of the metal in the catalyst depends on reaction conditions and the reactant and catalyst compositions (155), the oxidation of methane might follow different mechanisms under different conditions even for the same type of catalyst (156). Wang and Ruckenstein (152) indeed found that the mechanisms of the partial oxidation of methane to synthesis gas on rhodium-containing catalysts were dependent on both the metal loading and temperature. At low metal loadings (e.g., 0.05 wt%), the combustion-reforming mechanism was responsible for the reaction, whereas at high loadings (e.g., 1.0 wt%) a combination of the combustion-reforming and pyrolysis –oxidation mechanisms predominated at low temperatures (# 773 K), and the pyrolysis –oxidation mechanism became dominant at high temperatures ($ 923 K). Froment et al. (157) also found that, on Rh/Al2O3, the oxidation products under oxidative conditions were CO2 and H2O, whereas the selectivities towards CO and H2 rose to almost 100% as the conditions became more reductive.

III. CO2 Reforming of Methane Deactivation of supported metal catalysts by carbon or coke formation, which has its origin in the CH4 dissociation and/or CO disproportionation, is the most serious problem hindering the application of the CO2 reforming of methane. Attempts to overcome this limitation have focused on the development of improved catalysts. III.A. Carbon Formation on Metal Surfaces In the CO2 reforming of methane, carbon formation can occur via two possible pathways: CH4 decomposition and CO disproportionation (the Boudouard reaction). Carbon formation by CH4 decomposition is a structure-sensitive reaction (158,159). Specifically, the Ni(100) and Ni(110) surfaces are more active in the decomposition of CH4 to carbon than the Ni(111) surface (158). The CO disproportionation, 2CO ¼ C þ CO2 ;

0 DH298 ¼ 2172 kJ mol21

ð6Þ

is an exothermic reaction favored at temperatures below 973 K. Measurable rates of carbon deposition occur in the presence of cobalt, iron, and nickel catalysts at temperatures above 623 K (159). The form of carbon on metal surfaces generated during this reaction depends on the reaction conditions; amorphous and filamentous carbons predominate in the lower temperature range of 623– 873 K

322

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

(160 – 163), and a graphitic structure predominates at 973 K or higher temperatures (160,164 –166). The diffusion and segregation of carbon are also dependent on the metal surface structure. For example, the carbon on Ni(110) can diffuse more readily into the bulk than that on Ni(100) (159). Furthermore, the carbon adsorbed on the smaller metal particles diffuses with more difficulty than that on the larger particles (167). The structure-sensitivity of carbon formation provides the possibility for inhibition of the carbon deposition by modification of the catalyst surface structure.

III.B. Critical Issues Related to Carbon Deposition Thermodynamic considerations (5,168,169) suggest operation at high CO2/CH4 ratios (. 1) and high temperatures to minimize carbon formation in the CO2 reforming of methane. However, from an industrial viewpoint, it is desirable to operate at lower temperatures and with a CO2/CH4 (or H2O/CH4) ratio near unity. Such an operation requires a catalyst that kinetically inhibits the carbon formation under conditions that are thermodynamically favorable for carbon deposition. The noble metals and nickel were found to be highly active catalysts (170). Although the noble metals are characterized by much less carbon deposition than others (35), their high cost makes them unsuitable for large-scale applications. In terms of cost, nickel appears to be the most suitable catalyst. However, thermodynamic investigations indicated that the nickel-containing catalysts are prone to carbon deposition in CO2 reforming, resulting in catalyst deactivation (5). The inhibition of carbon deposition on the catalyst constitutes the greatest challenge in CO2 reforming. Two main properties of a catalyst affect the carbon deposition: surface structure and surface acidity (171,172). Evidence that the structure has a strong influence on carbon formation is provided by data showing that carbon formation is more difficult on Ni(111) than on Ni(100) or Ni(110) (159). One method of inhibiting carbon deposition is to control the size of the ensembles of metal atoms on the surface, because the ensembles necessary for carbon formation are larger than those needed for CH4 reforming (173). Thus, by controlling the nickel particle size, one can control the carbon deposition. For example, strong adsorption of sulfur can be used to influence ensemble size, and the suppression of carbon deposition on nickel catalysts by sulfur passivation was commercialized in the SPARG process (174,175). Sulfur passivation is attributed to the control of the size of the active metal ensembles because sulfur preferentially eliminates the larger ensembles. It has also been noted that carbon deposition can be attenuated or even suppressed when the metal is supported on a metal oxide with a strong Lewis basicity (176 – 179). This suppression occurs because the high Lewis basicity of the support increases the ability of the catalyst to

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

323

chemisorb CO2 in the CO2 reforming of methane and H2O in the steam reforming of methane, and these species react with carbon to form CO, resulting in decreased net carbon formation.

III.C. Supported Noble Metal Catalysts Inui (180) and Rostrup-Nielsen et al. (175) reported that the amount of carbon deposited on metal catalysts decreases in the order Ni q Rh . Ir ¼ Ru . Pt ø Pd at 773 K and Ni . Pd ¼ Rh . Ir . Pt q Ru at 923 K. Thus, the noble metals exhibit higher selectivities for a carbon-free operation than nickel. Nevertheless, carbon deposition does also occur on noble metals. The above sequence also depends on the nature of the support (35,175, 181 –184). ZrO2 has been widely used as support for platinum because of the lower rate of carbon formation than with other supports (185 –190). Bitter et al. (188) observed that the rate of carbon formation decreased in the sequence Pt/Al2O3 q Pt/TiO2 . Pt/ZrO2. Furthermore, the authors found that carbon formation (most likely from methane) rather than sintering is the main cause of the deactivation of the platinum-containing catalyst (Fig. 9). The high stability of the zirconia-containing catalysts is probably associated with the strong Pt – Zrnþ interactions, which reduce the carbon formation during reaction by promoting the CO2 dissociation (189). It was suggested that the catalytic activity is determined

Fig. 9. CO2 conversions in the CO2 reforming of CH4 catalyzed by Pt/ZrO2 (V), Pt/TiO2 (B), and Pt/g-Al2O3 (O). Each catalyst contained 0.5 wt% Pt. Before reaction, the catalyst was reduced in flowing H2 at 1125 K for 1 h. Reaction conditions: temperature, 875 K; feed gas molar ratios, CO2 =CH4 =Ar=N2 ¼ 4:2=4:2=7:5=1:0; GHSV, 32,000 mL (g catalyst)21 h21 (188).

324

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

by the available Pt –ZrO2 perimeter (186). On Pt/ZrO2, methane is decomposed on the metal to give CHx (the average value of x is 2) and H2. The principal pathway to CO2 reduction occurs by the initial formation of a carbonate close to the metal-support boundary. The carbon on the metal reduces the carbonate to formate, which decomposes rapidly to CO and surface hydroxyl groups. The hydroxyl groups recombine to form water or react further with methane to generate CO and hydrogen (steam reforming). When the rate of methane decomposition and carbonate reduction are in balance, the catalytic activity remains stable. In contrast, the activity of supported rhodium catalysts is determined principally by the concentration of accessible surface Rh atoms, which catalyze methane decomposition, followed by CO2 reduction (186). As a result, the support plays a minimal role in the rhodium-containing catalysts. The promoters also have a significant effect on carbon deposition. It was found that the bimetallic Pt – Au/SiO2, Pt – Sn/SiO2, and Pt –Sn/ZrO2 catalysts exhibited less carbon deposition during CO2 reforming of CH4 than the respective monometallic platinum catalysts (191), probably because of the formation of alloys. Vanadium oxide also plays a promoting role in the Rh/SiO2 catalyst at temperatures of 723– 773 K (192). Vanadium oxide enhances the catalytic activity of Rh/SiO2 and decreases the carbon deposition. This benefit was attributed to the formation of a partial VOx overlayer on the rhodium surface, which decreases the sizes of the accessible ensembles of Rh atoms, making some of them too small for coke formation; new sites at the Rh –VOx interface that are considered to activate CO2 dissociation were also created. The addition of cerium or lanthanum resulted in a significant improvement in the stability of Pt/ZrO2, with no decrease in either CH4 or CO2 conversion (193). Temperature-programmed oxidation (TPO) data showed that although the total amount of carbon deposited on the Ce-promoted Pt/ZrO2 catalyst was not less than that on the unpromoted catalyst, these deposits were eliminated at much lower temperatures, indicating the ability of the catalyst to self-clean its active sites. The La-promoted catalyst also exhibited a much lower carbon deposition than the unpromoted catalyst.

III.D. Non-Noble Metal Supported Catalysts In CO2 reforming, most of the reported research has been focused on non-noble metal catalysts, particularly nickel, because nickel has activity and selectivity comparable to those of noble metals, at much less cost. However, thermodynamic investigations indicated that the nickel-containing catalysts are prone to carbon deposition in CO2 reforming, resulting in catalyst deactivation (5). Therefore, an important challenge is to increase the resistance of nickel-containing catalysts to deactivation by carbon deposition.

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

325

III.D.1. Ni/Al2O3 Catalysts Alumina is one of the most commonly used supports for nickel catalysts (111, 178,194 –204). Ni/Al2O3 exhibits carbon deposition (180) that depends on the catalyst structure, composition, and preparation conditions. Chen and Ren (205) observed that the carbon deposition was markedly suppressed if NiAl2O4 was formed during pretreatment. This suppression might be the result of a strengthening of the Ni – O bond in NiAl2O4 when compared to that in the NiO crystal (206). The stronger Ni –O bond increases the difficulty of reduction of Ni2þ to Ni0, resulting in smaller nickel crystallites on the catalyst surface. These nickel crystallites, which are smaller than the size necessary for carbon deposition, decrease the carbon formation (195). Kim et al. (194,207) noted that, in comparison with the alumina-supported nickel catalyst prepared by the conventional impregnation method, Ni/Al2O3 catalysts prepared from aerogel alumina exhibited remarkably low coking rates, which the authors associated with the high dispersion of the metal particles. A similar observation was made by Osaki et al. (208). The authors suggested that the Ni –O – Al bonds formed in aerogels, which resulted in fine nickel particles after H2 reduction, contributed to both the high activity and low carbon deposition (208). A water-in-oil (w/o) microemulsion method was also effective in the preparation of a Ni/Al2O3 catalyst with good stability and low carbon deposition (209). Hayashi et al. (209) demonstrated that, although their conventionally impregnated catalyst deactivated with time-on-stream as a result of severe coking, the catalyst prepared by a w/o microemulsion method maintained its activity for 50 h, generating little coke for a CO2/CH4 molar ratio . 1.4. Furthermore, it was found that, at 1088 K and 21 atm pressure, a fresh nickel- and magnesium-containing hydrotalcite clay-derived catalyst provided the same performance as the commercial Ni/Al2O3 or Ni/MgAl2O4 catalysts, whereas under more severe operating conditions, the clay-derived catalysts exhibited superior activity and stability (210). A dependence of the amount of carbon deposition on the nickel loading was observed for Ni/Al2O3 catalysts (197). For example, a 1-wt% Ni/Al2O3 exhibited much less carbon deposition than a 13.6 wt% Ni/Al2O3 catalyst (197). Many promoters have been used to improve the performance of Ni/Al2O3 catalysts. The effect of the basic oxides of Na, K, Mg, and Ca on Ni/Al2O3 was examined by a number of authors (178,203,211 –213). They found that these added oxides markedly decrease the carbon deposition. The kinetics results showed that the added metal oxides changed the reaction order in CH4 from negative to positive and that in CO2 from positive to negative. This observation implies that the surface of a nickel catalyst incorporating basic metal oxides is abundant in adsorbed CO2, whereas the surfaces devoid of these oxides are abundant in adsorbed CH4 (178). The coverage of nickel with CO2 is most likely unfavorable to CH4 decomposition

326

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

and, as a result, the carbon deposition is decreased. Wang and Lu (214) also observed that Na2O or MgO promoters decreased the carbon deposition on Ni/ Al2O3 catalysts (Fig. 10). However, these promoted Ni/Al2O3 catalysts were characterized by lower activities and significant deactivation. Hence, it is inferred that the deactivation of the NaO- or MgO-promoted Ni/Al2O3 catalysts was not principally caused by carbon deposition. Choi et al. (215) examined the effect of Co, Cu, Zr, and Mn as promoters of Ni/ Al2O3 catalysts. They found that, in comparison with the unmodified Ni/Al2O3 catalysts, those modified with Co, Cu, and Zr exhibited slightly improved activities, whereas other promoters reduced the activity. The Mn-promoted catalyst provided a remarkable reduction in coke deposition with only a small reduction in catalytic activity. Furthermore, Seok et al. (216) noted that the manganese addition to Ni/Al2O3 led to a partial coverage of the surface of nickel by patches of MnOx, which promoted the adsorption of CO2. Both the partial coverage of the nickel surface with MnOx and the promoted CO2 adsorption appear to be responsible for the decreased carbon deposition on Ni/MnO – Al2O3 catalysts. Mo can also improve the stability of Ni/Al2O3 by reducing the carbon

Fig. 10. Carbon deposition on nickel-containing catalysts at 973 K as determined by TGA. Before reaction, the catalysts were reduced at 1073 K for 3 h. Reaction conditions: temperature, 973 K; feed gas molar ratio, CO2 =CH4 ¼ 1=1; GHSV, 144,000 mL (g catalyst)21 h21 (214).

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

327

deposition (217). Noble metal (Ru or Pd) addition to supported nickel catalysts resulted in a marked improvement in both activity and stability (218). Rare earth metals have also been used to promote Ni/Al2O3 catalysts. Slagtern et al. (219) tested Ni/Ln/Al2O3 (Ln ¼ rare earth mixture) catalysts containing 0.15 wt% Ni for their lifetimes (60 – 600 h) in a fluidized-bed reactor at 1073 K and 1 atm. The catalyst with a rare earth content of 1.7 wt% Ln was more active and stable than the unpromoted catalyst, and more active than a catalyst containing 8.5 wt% Ln. Furthermore, it was found that nickel sintering was initially the major cause of deactivation, with coking becoming increasingly important at longer times on stream (. 60 h). The catalyst with 1.7% Ln had a higher initial nickel dispersion than the catalyst devoid of Ln. However, the higher activity of the promoted catalyst than of the unpromoted catalyst could not be fully explained by this difference. Neodymium also promotes Ni/Al2O3 catalysts, by reducing the carbon deposition (220). CeO2 was also found to have an effect on the Ni/Al2O3 catalyst (221,222). Although CeO2 is not a suitable support for nickel because of the strong metal – support interaction, which reduces the catalytic activity, it can have a positive effect on the catalytic activity, stability, and suppression of carbon deposition when used as a promoter of Ni/ Al2O3 catalysts (221,222). A loading of 1 –5 wt% CeO2 was found to be the optimum. The use of CeO2 as a promoter for the nickel catalysts decreases the strength of the interactions between the nickel oxide and support, resulting in an increase in the reducibility of the nickel oxide and a higher nickel dispersion. The stability and reduced coking characteristic of CeO2-promoted catalysts can be attributed to the redox properties of CeO2, which can react directly with carboncontaining species to generate CO and CeOx, followed by the reoxidation of CeOx by CO2 back to CeO2 (221). III.D.2. Ni/SiO2 Catalysts The deactivation of Ni/SiO2 catalysts during the CO2 reforming of methane was examined as a function of various operating parameters (223). The two principal causes of catalyst deactivation, nickel sintering and carbon deposition, were shown to depend strongly on the pretreatment conditions. Kroll et al. (224) noted that for the Ni/SiO2 catalyst, nickel carbide-like layers, formed during the very initial period of the run, provided the active phase for CO2 reforming. However, when the carbon formation, which takes place at equilibrium with gaseous CH4, became faster than the oxidation of the carbon with the oxygen adspecies formed by carbon dioxide activation, carbon deposition occurred. The carbon deposition depended strongly on the nickel loading (197). It was found that a 13.6 wt% Ni/ SiO2 catalyst exhibited a greater carbon deposition than a 1 wt% Ni/SiO2. A physical mixture of SiO2 and nickel minimized the amount of deposited carbon (225), and a physical mixture of Al2O3 and nickel generated a greater amount of

328

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

carbon deposition (197). This comparison indicates that the Al2O3 surface promotes carbon deposition. CaO also affects Ni/SiO2 catalysts by decreasing the dispersion of the nickel phase (226). III.D.3. Ni/La2O3 Catalysts Zhang and Verykios (227) reported a Ni/La2O3 catalyst which exhibited a higher activity and higher long-term stability for CO2 reforming of methane to synthesis gas than Ni/Al2O3 and Ni/CaO catalysts. As shown in Fig. 11, although the initial rate of reaction on Ni/g-Al2O3 was higher than that on Ni/CaO, probably as a consequence of the higher dispersion of nickel in the former catalyst, the deactivation rate of Ni/g-Al2O3 was higher than that of Ni/CaO. In contrast, the rate of reaction on a Ni/La2O3 catalyst increased significantly with time on stream during the initial 2 –5 h of reaction, and then tended to remain unchanged with time

Fig. 11. CO formation rates determined from reactant conversions and product selectivities in a fixed-bed flow reactor for CO2 reforming of CH4. The catalysts were nickel supported on La2O3, g-Al2O3, or CaO. Each catalyst contained 17 wt% Ni. Before reaction, the catalyst was reduced in flowing H2 at 773 K for at least 5 h and then at 1023 K for 2 h. Reaction conditions: pressure, 1.0 atm; temperature, 1023 K; feed gas molar ratio, CH4 =CO2 =He ¼ 2=2=6; GHSV, 1,800,000 mL (g catalyst)21 h21 (227).

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

329

on stream for 100 operating hours. In these experiments, low conversions of CH4 and CO2 were observed at a very high space velocity (227,228). However, when higher CH4 and CO2 conversions (about 75 and 80%, respectively) were obtained by reducing the space velocity, the Ni/La2O3 catalyst exhibited deactivation (Fig. 12) (228). Other researchers also observed the deactivation of Ni/La2O3 catalysts at high CH4 and CO2 conversions (229). The higher stability of the catalyst at low reactant conversions might have occurred because high concentrations of unreacted CO2 inhibited carbon deposition by the reaction CO2 þ C ¼ 2CO. 2 Ruckenstein and Hu (230) investigated the role of the anions NO2 3 or Cl (used in the catalyst preparation by impregnation of the unreduced Ni/La2O3) in carbon deposition on Ni/La2O3 catalysts. The unreduced Ni/La2O3 catalyst, prepared from nickel nitrate, was characterized by a high initial CO yield but a low stability; in contrast, the unreduced Ni/La2O3 catalyst, prepared with chloride, had a high stability. This stabilization probably occurred because a stable lanthanum chloride inhibited the formation of large ensembles of nickel atoms, which are necessary for carbon deposition. The preparation method also affects the Ni/La2O3 catalysts (231). The conversions of CH4 and CO2 in the CO2 reforming of CH4 catalyzed by Ni/La2O3

Fig. 12. Conversions of CH4 and CO2 and selectivities for formation of CO and H2 as a function of time on stream for CO2 reforming of CH4 catalyzed by 17 wt% Ni/La2O3. Before reaction, the catalyst was reduced in flowing H2 at 773 K for at least 5 h and then at 1023 K for 2 h. Reaction conditions: pressure, 1 atm; temperature, 1023 K; feed gas molar ratio, CH4 =CO2 ¼ 1=1; GHSV is unknown (228).

330

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

prepared by a sol – gel technique were significantly higher than those catalyzed by Ni/La2O3 prepared by wet impregnation. TG/DTG experiments confirmed that the amount of carbon deposited in the former case was smaller than in the latter case. It is inferred that the difference can be attributed to the uniform dispersion of nanoscale nickel particles in the sol – gel-generated Ni/La2O3 catalyst. III.D.4. Ni/ZrO2 Catalysts The suitability of zirconia-supported nickel catalysts for the CO2 reforming reaction was investigated with emphasis on the stability of the catalysts under conditions favorable for carbon formation (232). It was found that at temperatures between 993 and 1053 K, the ZrO2-supported catalysts with lower nickel loadings (, 2 wt%) were more stable than those with higher nickel loadings for a stoichiometric CO2/CH4 ratio. Furthermore, two forms of deposited carbon were observed in the less stable catalysts, and only one form was observed in the more stable ones. Carbon deposits were formed on the reduced catalyst at a very high rate during the TPSR (233). The amount of deposited carbon remained constant on the catalyst during reaction at 973 K (233), consistent with the inference that the initially formed carbon acted as a reaction intermediate that transformed CO2 into CO. Even with catalysts having high nickel loadings, catalyst lives without significant deactivation were achieved for 30 h at 1023 K and for 20 h at 1123 K (234). Li et al. (235) found that promoters can affect the Ni/ZrO2 catalysts. Among the Ni/ZrO2 catalysts promoted with oxides of lanthanum, cerium, or manganese, Ni/La – ZrO2 exhibited the highest activity, whereas Ni/Ce – ZrO2 and Ni/Mn – ZrO2 were characterized by low carbon depositions during reaction. Furthermore, Ni– Mg/ZrO2 exhibited the highest activity and stability. It was inferred that the promotion by magnesium can be attributed to increasing dispersion of nickel and to an enhancement in the interaction between CO2 and the catalyst. Lercher et al. (236) reported that Ni/ZrO2 catalysts with small sizes of metal particles (2 –3 nm) exhibited high stability. The small particles prevented the formation of carbon filaments. The stabilities of Ni/ZrO2 catalysts were also dependent on the preparation method. Wei et al. (237) reported that the Ni/ZrO2 catalyst prepared from large Zr(OH)4 particles deactivated rapidly. In contrast, a catalyst with a high metal loading of nickel (27 wt%), obtained by impregnating ultra-fine Zr(OH)4 particles (6 nm) with nickel nitrate, exhibited a high and stable activity for CO2 reforming without deactivation by carbon deposition. The activity of this catalyst for CO2 reforming of CH4 at 1030 K, with a CH4/ CO2 ¼ 1:1 molar feed rate of 24,000 mL (g catalyst)21 h21 did not deactivate for 600 h, but exhibited oscillations in the CH4 conversion between 80 and 85% (Fig. 13). Comparing their best Ni/ZrO2 catalyst with the NiO/MgO

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

331

Fig. 13. CH4 conversion in the CO2 reforming of CH4 catalyzed by Ni/ZrO2. Before reaction, the catalyst was reduced in flowing H2/N2 (1/9, molar ratio) at 973 K for 3 h. Reaction conditions: pressure, 1 atm; temperature, 1030 K; feed gas molar ratio, CH4 =CO2 ¼ 1=1; GHSV, 24,000 mL (g catalyst)21 h21 (237).

solid-solution catalyst of Fujimoto et al. (238), Wei et al. concluded that their Ni/ZrO2 catalyst exhibited higher activity. But the activity of their best Ni/ZrO2 catalyst is much lower than that of Ruckenstein and Hu’s NiO/MgO solidsolution catalysts (239) (see Section III.E for more details). III.D.5. Other Catalysts Carbon deposition is much greater on Co/Al2O3 catalysts than on Ni/Al2O3 (240). The presence of MgO markedly decreased the carbon deposition on the surface of the cobalt catalyst (241). The role of MgO may be attributed to the formation of strongly adsorbed CO2 species, which can easily react with the deposited carbon, thus preventing catalyst deactivation (241). Osaki et al. (242) compared the catalytic performance of MoS2 and WS2 with that of Ni/SiO2. The CO2 reforming of methane on MoS2 or WS2 catalysts was characterized by much lower reaction rates than that on the nickel catalyst, although the sulfides prevented carbon deposition during the reforming reaction. Completely different rate equations were obtained for the metal disulfide and nickel catalysts. The positive reaction order in CH4 partial pressure and the negative order in CO2 partial pressure characteristic of the sulfide catalyst are

332

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

contrasted with the negative order in CH4 partial pressure and positive order in CO2 partial pressure characteristic of the nickel catalyst. These observations suggest that the surface of the sulfide catalyst was abundant in adsorbed CO2, whereas the surface of the nickel catalyst was abundant in adsorbed CH4. The coverage with CO2 can be considered to be the principal cause of the suppression of carbon deposition on the sulfide catalysts. The Haldor Topsøe firm developed the SPARG process for the CO2/CH4 reforming, in which the conventional nickel-containing steam reforming catalyst was modified to reduce its coke-forming propensity, by the continuous addition of small amounts of sulfur to the feed gas during operation (174,175,243). However, the passivation process led to a lower catalytic activity and required high operating temperatures as a consequence of the sulfur poisoning of the active sites. To develop effective catalysts for the CO2 reforming of methane, other supports were also used for nickel catalysts, including perovskite (244), Y zeolite (245,246), 5A zeolite (247), high-silica ZSM-5 zeolite (248), and AlPO4 (tridymite) (249). In summary, the development of non-noble metal catalysts has been focused on nickel-containing catalysts (because nickel has an activity and a selectivity comparable to those of noble metals at much less cost); on finding effective promoters, selecting suitable supports; and on improving preparation methods. Although some nickel-containing catalysts appear to be effective in short-term tests for CO2 reforming, their long-term stability and tolerance for impurities, which are important in industrial applications, are not yet clear.

III.E. MgO-Containing Solid-Solution Catalysts III.E.1. Characteristics of MgO-Containing Solid-Solution Catalysts MgO is a basic metal oxide and has the same crystal structure as NiO. As a result, the combination of MgO and NiO results in a solid-solution catalyst with a basic surface (171,172), and both characteristics are helpful in inhibiting carbon deposition (171,172,239). The basic surface increases CO2 adsorption, which reduces or inhibits carbon-deposition (Section III.B). The NiO – MgO solid solution can control the nickel particle sizes in the catalyst. This control occurs because in the solid solution NiO has strong interactions with MgO and, as indicated by TPR data (26), the former oxide can no longer be easily reduced. Consequently, only a small amount of NiO is expected to be reduced, and thus small nickel particles are formed on the surface of the solid solution, smaller than the size necessary for coke formation. Indeed, the nickel particles on a reduced 16.7 wt% NiO/MgO solid-solution catalyst were too small to be observed by TEM (171). Furthermore, two additional important qualities stimulated the selection of MgO as a support: its high thermal stability and low cost (250,251).

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

333

Like NiO, CoO and FeO are characterized by the same crystal structure as MgO and have comparable lattice parameters, and, hence, can form CoO/MgO and FeO/MgO solid solutions. Therefore, it was expected that CoO/MgO and FeO/MgO would inhibit carbon deposition and metal sintering, just as Ni/MgO does, resulting in high stability (171).

III.E.2. Highly Effective MgO-Containing Solid-Solution Catalysts In 1989, Gadalla and Sommer (252) reported that a solid-solution NiO/MgO (1:1.35) catalyst prepared by precipitation can inhibit the carbon deposition in the CO2 reforming of methane; however, they obtained a low CO2 conversion (66%), a low H2 selectivity (79%), and a low CO selectivity (77%), even at the very low WHSV of 3714 cm3 (g catalyst)21 h21 with a CH4/CO2 (1/1, molar) feed gas and the high temperature of 1200 K. Their relatively high CH4 conversion was partly a consequence of homogeneous gas-phase reactions that occurred under their conditions. Indeed, the authors found extensive carbon deposits plugging the reactor upstream and downstream of the reaction zone. In 1992, Fujimoto et al. (176) reported results for the CO2 reforming of methane catalyzed by NiO/MgO prepared by coprecipitation of the hydroxides from aqueous solutions of nickel acetate and magnesium acetate with K2CO3 at 333 K; the coprecipitates were dried at 393 K for 12 h and calcined at 1223 K for 20 h. Although they did not mention that the NiO/MgO, which had the composition Ni0.03Mg0.97O, was a solid solution, it was surely a solid solution because it was calcined at a high temperature, as the authors (253) later reported. Fujimoto et al. (176) observed that the NiO/MgO catalyst had a low stability, suggested to be a consequence of carbon deposition. Although they added CaO to the NiO/MgO to increase the stability, this addition decreased the activity tremendously. Takayasu et al. (254,255) also noted a deactivation of the NiO/ MgO catalysts, caused by the formation of carbonaceous deposits. In 1994, Swaan et al. (256) reported that a 3 wt% Ni/MgO catalyst had a low activity. They suggested that the stabilization of Ni2þ ions in the MgO matrix was responsible for the limited reducibility of the nickel observed experimentally and for the formation of an active phase for the reforming reaction. Because of these results, NiO –MgO solid-solution catalysts did not attract much interest at that time. In 1995, Ruckenstein and Hu reported a highly efficient 16.7 wt% NiO/MgO solid-solution catalyst for CO2 reforming of methane, which was prepared by impregnation and was calcined at 1073 K for 1.5 h (239). It exhibited almost 100% conversion of CO2, . 91% conversion of CH4, and . 95% selectivities to CO and H2 at 1063 K, atmospheric pressure, and the very high space velocity of 60,000 mL (g catalyst)21 h21 for a CH4/CO2 molar ratio of 1 (Fig. 14) (239).

334

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

Fig. 14. CH4 conversion (a) and CO yield (b) in the CO2 reforming of CH4 catalyzed by reduced 16.7-wt% NiO/alkaline earth metal oxides. Before reaction, each catalyst was reduced in flowing H2 at 773 K for 14 h. Reaction conditions: pressure, 1 atm; temperature, 1063 K; feed gas molar ratio, CH4 =CO2 ¼ 1=1; GHSV, 60,000 mL (g catalyst)21 h21 (239).

The conversions and selectivities remained unchanged during the entire reaction time employed (120 h), indicating that the reduced NiO/MgO catalyst had a high stability (Fig. 14). In contrast to MgO, the other alkaline-earth oxides, such as CaO, SrO, and BaO, were found to be poor supports for NiO, as they provided catalysts with low activities, selectivities, or stabilities (Fig. 14) (239). Although the reduced NiO/Al2O3 catalyst provided high initial conversions (CH4, 91%; CO2, 98%) and selectivities (. 95% for both CO and H2), it was characterized by the fastest carbon deposition, which led to the complete plugging of the reactor after only 6 h of reaction (197). The reduced Ni/TiO2 catalyst gave relatively low initial

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

335

conversions of CH4 and CO2 (41 and 67%, respectively), which decreased with increasing reaction time (197). It seems reasonable to conclude that the excellent catalytic performance of NiO/MgO should be attributed to the formation of a solid solution (257). The conversions and selectivities characteristic of NiO/MgO solid-solution catalysts were found to be dependent on their composition, preparation conditions, and even the properties of the MgO (257 –259). Furthermore, the authors found that high and stable CO yields (. 95%) occurred with NiO/MgO catalysts having NiO contents between 9.2 and 28.6 wt% (258). No activity was observed, however, for a NiO content of 4.8 wt%. At the high NiO content of 50 wt%, the CO yield decreased from 91 to 53% after 40 h, and the catalyst became black, because of carbon deposition, after about 50 h of reaction. In contrast, the other NiO/MgO solid-solution catalysts maintained their initial color, and no carbon deposition was detected by TEM even after 120 h of reaction (171). It was, therefore, inferred that too small amounts of NiO in the NiO/MgO catalysts provided too-small numbers of Ni sites, and too-high amounts provided numerous nickel metal particles that could easily sinter, generating large particles that facilitated carbon deposition. Furthermore, the MgO surface area, pore size distribution, and lattice parameters were observed to affect significantly the performance of NiO/MgO solid-solution catalysts (259). An unsuitable MgO can lead to a low initial conversion and a long induction time (259). In 1997, Fujimoto et al. (260 –262) reported new results for the CO2 reforming of methane catalyzed by the Ni0.03Mg0.97O solid solution used by them in 1992 (176) and by bimetallics containing in addition small amounts of platinum, palladium, or rhodium (molar ratio of M/(Ni þ Mg) was varied between 0.7 £ 1024 and 3.2 £ 1024, where M ¼ Pt, Pd, or Rh) (260). The Ni0.03Mg0.97O solid-solution catalyst provided a low CO yield (about 215 mmol (g catalyst)21 s21, i.e., 38%) at a space velocity of 44,800 mL (g catalyst)21 h21 even at 1123 K. However, the addition of a noble metal promoted both the activity and the stability at 773 K. The optimum noble metal loading was obtained for M/ (Ni þ Mg) < 2.1 £ 1024 (molar ratio). Temperature-programmed hydrogenation (TPH) of the carbonaceous species formed during the catalytic reaction indicated that the resistance of the Ni0.03Mg0.97O solid-solution catalyst to carbon deposition was retained by the bimetallic catalysts as well (260). Furthermore, TEM and EDS observations provided evidence of the formation of Pt – Ni alloy particles (260). Temperature-programmed decomposition (TPD) data obtained with CH4 suggested that CH4 decomposition was the rate-determining step on Ni0.03Mg0.97O and that the CH4 decomposition was accelerated by alloy formation (260). The improved stability of the catalyst was attributed to the increased catalyst reducibility caused by noble metal promotion. Fujimoto et al. (253) also found that the water treatment of the Ni0.03Mg0.97O solid-solution catalyst increased the catalytic activity and stability for CO2

336

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

reforming of CH4. This promoting effect was inferred to be the consequence of a structural rearrangement of the solid solution by the formation of nickel and magnesium hydroxides (253). Furthermore, Fujimoto et al. (261– 263) reported that even Ni0.03Mg0.97O has a fairly good stability in the CO2 reforming of methane. The excellent anti-coking performance of the reduced NiO/MgO solid solution catalyst can be attributed to the high dispersion of the reduced nickel species, the basicity of the support surface, and the nickel-support interactions. From the above results, one can conclude that different NiO/MgO solidsolution catalysts can have very different catalytic performances. For example, Fujimoto et al.’s Ni0.03Mg0.97O solid-solution catalyst exhibited relatively low activities. To reach about 82% conversion of CH4 in the presence of this Ni0.03Mg0.97O catalyst, the space velocity had to be reduced to 18,670 mL (g catalyst)21 h21 at 1123 K (Fig. 15) (238). In contrast, Ruckenstein and Hu’s NiO/MgO catalysts have very high activities (. 91% conversion of CH4 and . 95% selectivities of CO and H2 at the space velocity of 60,000 mL (g catalyst)21 h21 at 1063 K) (Fig. 14) (239). Hu and Ruckenstein (239,257,259) noted that the properties of the MgO, such as its surface area, pore size distribution, and crystal structure, have important effects on the NiO/MgO solid-solution catalysts. They found that the MgO supplied by Aldrich, which has

Fig. 15. CH4 conversions in the CO2 reforming of CH4 in the presence of nickel-containing catalysts. Before reaction, the catalyst was reduced in flowing H2 at 1123 K for 14 h. Reaction conditions: pressure, 1 atm; temperature, 1123 K; feed gas molar ratio, CH4 =CO2 ¼ 1=1; GHSV, 18,670 mL (g catalyst)21 h21 (238).

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

337

a surface area of about 50 m2 g21 with nano-pores (10 – 100 nm) and nano-crystals (about 20 nm) (171,172,257), was a suitable support material for MgO-containing solid-solution catalysts with very high activity and selectivity as well as high stability. Recently, Ruckenstein and Wang (264 –266) also successfully developed excellent CoO/MgO solid-solution catalysts for CO2 reforming of methane. They reported that Co/MgO exhibited a good catalytic performance with a CO yield of 93% and a H2 yield of 90% at the high space velocity of 60,000 mL (g catalysts)21 h21 and 1163 K, which remained unchanged during 50 h of investigation (264). In contrast, Co/CaO, Co/SrO, and Co/BaO each provided low CO yields, and Co/CaO also had a low stability. The results indicate that the CoO/MgO catalysts are characterized by performances similar to those of NiO/MgO. In summary, the basicity and the strong NiO –MgO interactions in binary NiO/MgO solid solution catalysts, which inhibit carbon deposition and catalyst sintering, result in an excellent catalytic performance for CO2 reforming. The characteristics of MgO play an important role in the performance of a highly efficient NiO/MgO solid-solution catalyst. Moreover, the NiO/MgO catalyst performance is sensitive to the NiO content: a too-small amount of NiO in the solid solution leads to a low activity, and a too-high amount of NiO to a low stability. CoO/MgO solid solutions have catalytic performances similar to those of NiO/MgO solid solutions, but require higher reaction temperatures. So far, no experimental information is available regarding the use of a FeO/MgO solid solution for CH4 conversion to synthesis gas.

IV. Conclusions Synthesis gas production from natural gas, the most important step in the gas-toliquid process, can account for at least 60% of the integrated cost of the total gasto-liquid plant. The catalytic partial oxidation of methane provides a fast process for the synthesis gas production. However, several challenges still remain regarding this process. Large temperature gradients in the reactors (hot spots), which are the result of a combination of a high space velocity and an exothermic reaction, could make the process hazardous and difficult to control in industrialscale operations. Current technical options to solve this problem include fluidized-bed reactors, in which the temperature of the mixed catalyst is almost uniform, and combined processes that eliminate hot spots by combining the exothermic partial oxidation with the endothermic CO2 reforming (or steam reforming). High O2 separation costs represent the greatest challenge facing the partial oxidation process. The main focus of research aimed at overcoming this

338

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

limitation is on O2-permeable ceramic membrane reactor processes, in which air can be used directly. The membrane processes, which obviate the O2 separation plant, could reduce the process cost by 25 – 40%. However, the O2 permeation and stability of such membranes still need improvement. The complexity of the reaction mechanisms, which depend on the catalyst composition and degree of reduction, as well as on reaction conditions, can cause great difficulties in the process design and control. CO2 reforming of methane is an attractive technology because it converts two greenhouse gases into useful chemicals. The deactivation of the catalyst, caused by carbon deposition, constitutes the greatest challenge in this process. Although noble metal catalysts are less subject to carbon deposition, nickel-containing catalysts have attracted the most research interest, in part because of the relatively low cost of nickel. In the preceding 10 years, several types of nickel-containing catalysts with high activities and stabilities have been reported. For example, nickel-containing solid solution catalysts have very high activity, selectivity, and stability; and they inhibit carbon deposition and catalyst sintering. Because CO2 reforming of methane is a strongly endothermic process, the development of new methods to provide less expensive energy constitutes attractive goal for future research related to CO2 reforming.

References 1. 2. 3. 4. 5. 6. 7. 8.

9.

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

Byrne, J.P.J., Gohr, R.J., and Haslam, R.T., Ind. Engng Chem. 24, 1129 (1932). Rostrup-Nielsen, J.R., Sehested, J., and Nørskov, J.K., Adv. Catal. 47, 65 (2002). Prettre, M., Eichner, C., and Perrin, M., Trans. Faraday Soc. 43, 335 (1946). Lang, J., Z. Phys. Chem. 2, 161 (1888). Gadalla, A.M., and Bower, B., Chem. Engng Sci. 43, 3049 (1988). Dyrssen, D., and Turner, D.R., in “Carbon Dioxide Chemistry: Environmental Issues” (J. Paul and C.M. Pradier, Eds.), p. 317. Athenaeum Press, Cambridge, 1994. Puskas, I., CHEMTECH 24, 43 (1995). (a) Keller, G.E., and Bhasin, M.M., J. Catal. 73, 9 (1982); (b) Lee, J.S., and Oyama, S.T., Catal. Rev.—Sci. Engng 30, 249 (1988); (c) Amenomiya, Y., Birss, V.I., Goledzinowski, M., Galuszka, J., and Sanger, A.R., Catal. Rev.—Sci. Engng 33, 163 (1990). (a) Pitchai, R., and Klier, K., Catal. Rev.—Sci. Engng 28, 13 (1986); (b) Foster, N.R., Appl. Catal. A: Gen. 19, 1 (1985); (c) Hall, T.J., Hargreaves, J.S.J., Hutchings, G.J., Joyner, R.W., and Taylor, S.H., Fuel Process. Technol. 42, 151 (1995). Kiennemann, A., Petrol. Technol. 29, 355 (1990). Guczi, L., van Santen, R.A., and Sarma, K.V., Catal. Rev.—Sci. Engng 38, 249 (1996). Pasquon, I., Plenary lecture at EUOPACAT-1, Montpellier, France, September 12 –17, 1993. Ashcroft, A.T., Cheetham, A.K., Foord, J.S., Green, M.L.H., Grey, C.P., Murrell, A.J., and Vernon, P.D.F., Nature 344, 319 (1990). Hickman, D.A., and Schmidt, L.D., J. Catal. 138, 267 (1992). Hickman, D.A., and Schmidt, L.D., Science 259, 343 (1993).

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

339

16. Torniainen, P., Chu, X., and Schmidt, L.D., J. Catal. 146, 1 (1994). 17. (a) Choudhary, V.R., Rajput, A.M., and Prabhakar, B., Catal. Lett. 15, 363 (1992); (b) Choudhary, V.R., Rajput, A.M., and Prabhakar, B., J. Catal. 139, 326 (1993). 18. Choudhary, V.R., Rajput, A.M., and Rane, V.H., J. Phys. Chem. 96, 8686 (1992). 19. Choudhary, V.R., Rajput, A.M., and Rane, V.H., Catal. Lett. 16, 269 (1992). 20. Choudhary, V.R., Sansare, S.D., and Mamman, A.S., Appl. Catal. A: Gen. 90, L1 (1992). 21. Chang, Y.F., and Heinemann, H., Catal. Lett. 21, 215 (1993). 22. Dissannayake, D., Rosynek, M.P., and Lunsford, J.H., J. Phys. Chem. 97, 3644 (1993). 23. Hu, Y.H., Au, C.T., and Wan, H.L., Chin. Sci. Bull. 40, 303 (1995). 24. Hu, Y.H., Au, C.T., and Wan, H.L., Kexue Tongbao 40, 430 (1995). 25. Hu, Y.H., and Ruckenstein, E., Ind. Engng Chem. Res. 37, 2333 (1998). 26. Hu, Y.H., and Ruckenstein, E., Ind. Engng Chem. Res. 38, 1742 (1999). 27. Zhang, X.L., Hayward, D.O., and Mingos, D.M.P., Catal. Lett. 83, 149 (2002). 28. Zhang, X.L., Mingos, D.M.P., and Hayward, D.O., Catal. Lett. 72, 147 (2001). 29. Basile, F., Fornasari, G., Trifiro, F., and Vaccari, A., Catal. Today 64, 21 (2001). 30. Wolf, D., Hohenberger, M., and Baerns, M., Ind. Engng Chem. Res. 36, 3345 (1997). 31. Olsbye, U., Tangstad, E., and Dahl, I.M., in “Natural gas conversion II.” p. 303, Elsevier, Amsterdam, 1994. 32. Bharadwaj, S.S., and Schmidt, L.D., J. Catal. 146, 11 (1994). 33. Roberts, M.J., AIChE 1997 Spring National Meeting, Houston, TX, March 9 –13, 1997. 34. Ioannides, T., and Verykios, X.E., Catal. Lett. 47, 183 (1997). 35. Ashcroft, A.T., Cheetham, A.K., Green, M.L.H., and Vernon, P.D.F., Nature 352, 225 (1991). 36. Inui, T., Saigo, K., Fujii, Y., and Fujioka, K., Catal. Today 26, 295 (1995). 37. Ruckenstein, E., and Hu, Y.H., Ind. Engng Chem. Res. 37, 1744 (1998). 38. O’Connor, A.M., and Ross, J.R.H., Catal. Today 46, 203 (1998). 39. Ruckenstein, E., and Wang, H.Y., Catal. Lett. 73, 99 (2001). 40. Mo, L.Y., Zheng, X.M., Huang, C.J., and Fei, J.H., Catal. Lett. 80, 165 (2002). 41. Choudhary, V.R., Rajput, A.M., and Prabhakar, B., Catal. Lett. 32, 391 (1995). 42. Choudhary, V.R., Rajput, A.M., and Prabhakar, B., Angew. Chem. Int. Ed. 33, 2104 (1994). 43. Matsumura, Y., and Moffat, J.B., Catal. Lett. 24, 59 (1994). 44. Goetsch, D.A., and Say, G.R., US Patent 4877550 (1989). 45. Eberly, P.E., Goetsch, D.A., Say, G.R., Vargas, J.M., and Argas, J.M., US Patent 4888131 (1989). 46. Fiato, R.A., Long, D.C., Say, G.R., Taylor, J.H., Say, G., and Long, D., US Patent 5143647 (1992). 47. Choudhary, V.R., and Mamman, A.S., Appl. Energy 66, 161 (2000). 48. Clavenna, L.R., Davis, S.M., Fiato, R.A., and Say, G.R., US Patent 5348717 (1994). 49. Say, G.R., and Tayor, J.H., US Patent 5421840 (1995). 50. Long, D.C., EP Patent 673877 (1997). 51. Balachandran, U., Dusek, J.T., Mieville, R.L., Poeppel, R.B., Kleefisch, M.S., Pei, S., Kobylinski, T.P., Udovich, C.A., and Bose, A.C., Appl. Catal. A: Gen. 133, 19 (1995). 52. Jin, W., Gu, X., Li, S., Huang, P., Xu, N., and Shi, J., Chem. Engng Sci. 55, 2617 (2000). 53. Jin, W.Q., Li, S.G., Huang, P., Xu, N.P., Shi, J., and Lin, Y.S., J. Membr. Sci. 166, 13 (2000). 54. Teraoka, Y., Zhang, H.M., Furukawa, S., and Yamozoe, N., Chem. Lett. 1743 (1985). 55. Teraoka, Y., Nobunaga, T., and Yamazoe, N., Chem. Lett. 503 (1988). 56. Mazanec, T.J., Cable, T.L., and Frye, J.G. Jr., Solid State Ion. 53111 (1992). 57. Gur, T.M., Belzner, A., and Huggins, R.A., J. Membr. Sci. 75, 151 (1992). 58. Cable, T.L., European Patent 0 399 833 A1 (1990). 59. Cable, T.L., European Patent 0 438 902 A2 (1991).

340

60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77.

78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92.

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

Wiley, J.B., and Poeppelmeier, K.R., J. Solid State Chem. 88, 250 (1990). Eror, N.G., and Balachandran, U., J. Solid State Chem. 40, 85 (1981). Balachandran, U., and Eror, N.G., J. Phys. Chem. Solids 44, 231 (1983). Teraoka, Y., Nobunaga, T., Okamoto, K., Miura, N., and Yamazoe, N., Solid State Ion. 48, 207 (1991). Balachandran, U., Dusek, J.T., Maiya, P.S., Mieville, R.L., Kleefisch, M.S., and Udovich, C.A., Catal. Today 133, 19 (1997). Balachandran, U., Dusek, J.T., Picciolo, J.J., Maiya, P.S., Ma, B., Mieville, R.L., Kleefisch, M.S., and Udovich, C.A., Prepr. ACS Div. Fuel Chem. 42, 591 (1997). Balachandran, U., Ma, B., Maiya, P.S., Mieville, R.L., Dusek, J.T., Picciolo, J.J., Guan, J., Dorris, S.E., and Liu, M., Solid Sate Ion. 108, 363 (1998). Mazanec, T.J., Cable, T.L., Frye, J.G., and Kliewer, W.R., US Patent 5,306,411 (1994). Schwartz, M., While, J.H., Myers, M.G., Deych, S., and Sammells, A.F., Prepr. ACS Div. Fuel Chem. 42, 596 (1997). Tsai, C.Y., Dixon, A.G., Moser, W.R., and Ma, Y.H., AIChE J. 43, 2741 (1997). Tsai, C.Y., Dixon, A.G., Ma, Y.H., Moser, W.R., and Pascucci, M.R., J. Am. Ceram. Soc. 81, 1437 (1998). Li, G.T., Shao, Z.P., Xiong, G.X., Dong, H., Cong, Y., and Yang, W.S., J. Inorg. Mater. 17, 1041 (2002). Dong, H., Shao, Z.P., Xiong, G.X., Tong, J.H., Sheng, S.S., and Yang, W.S., Catal. Today 67, 3 (2001). Tong, J.H., Yang, W.S., Cai, R., Zhu, B.C., and Lin, L.W., Catal. Lett. 78, 129 (2002). DOE Techline, May 20,1997. Oil & Gas Journal, Page 62, May 3, 1999. Riess, I., in “The CRC Handbook of Solid State Electrochemistry” (P.J. Gellings and H.J.M. Bouwmeester, Eds.), p. 223. CRC Press, Boca Raton, FL, 1997. Bouwmeester, H.J.M., and Burggraaf, A.J., in “Fundamentals of Inorganic Membrane Science and Technology” (A.J. Burggraaf and L. Cot, Eds.), p. 435. Elsevier, Amsterdam, 1996. Teraoka, Y., Nobunaga, T., Okamoto, K., Miura, N., and Yamazoe, N., Solid State Ion. 48, 207 (1991). Teraoka, Y., Zhang, H.M., Okamoto, K., and Yamazoe, N., Mater. Res. Bull. 23, 51 (1988). Mazanec, T.J., Solid State Ion. 70/71, 11 (1994). Steele, B.C.H., Mater. Sci. Engng B 13, 79 (1992). Steele, B.C.H., Curr. Opin. Solid State Mater. Sci. 1, 684 (1996). Lewis, W.K., Gilliland, E.R., and Reed, W.A., Ind. Engng Chem. 41, 1227 (1949). Lewis, W.K., and Gilliland, E.R., US Patent 2,603,608 (1952). Lewis, W.K., US Patent 2,631,934 (1953). Read, W.A., DSc Dissertation, Massachusetts Institute of Technology, 1948. Huszar, K., Racz, G., and Szekely, G., Acta Chim. Acad. Sci. Hungary 70, 287 (1971). Gavalas, G.R., Phichitcul, C., and Voecks, G.E., J. Catal. 88, 54 (1984). Kikuchi, E., and Chen, Y., Natural gas conversion V. Stud. Surf. Sci. Catal. 119, 441 (1998). Elmasides, C., Ioannides, T., and Verykios, X.E., Natural gas conversion V. Stud. Surf. Sci. Catal. 119, 801 (1998). Yan, Q.G., Chu, W., Gao, L.Z., Yu, Z.L., and Yuan, S.Y., Natural gas conversion V. Stud. Surf. Sci. Catal. 119, 855 (1998). Nakagawa, K., Ikenaga, N., Suzuki, T., Kobayashi, T., and Haruta, M., Appl. Catal. A: Gen. 169, 281 (1998).

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

341

93. Boucouvalas, J., Efstathiou, A.M., Zhang, Z.L., and Verykios, X.E., Natural gas conversion IV. Stud. Surf. Sci. Catal. 107, 435 (1997). 94. Laa, J.C., Berger, R.J., and Marin, G.B., Catal. Lett. 43, 63 (1997). 95. Soick, M., Buyevskaya, O., Hohenberger, M., and Wolf, D., Catal. Today 32, 163 (1996). 96. Nakagawa, K., Suzuki, T., Kobayashi, T., and Haruta, M., Chem. Lett. 1029 (1996). 97. Boucouvalas, Y., Zhang, Z.L., and Verykios, X.E., Catal. Lett. 40, 189 (1996). 98. Mattos, L.V., Oliveira, E.R., Resende, P.D., Noronha, F.B., and Passos, F.B., Catal. Today 77, 245 (2002). 99. Pantu, P., and Gavalas, G.R., Appl. Catal. A: Gen. 223, 253 (2002). 100. Elmasides, C., Kondarides, D.I., Neophytides, S.G., and Verykios, X.E., J. Catal. 198, 195 (2001). 101. Wang, H.Y., and Ruckenstein, E., J. Catal. 186, 181 (1999). 102. Wang, H.Y., and Ruckenstein, E., Catal. Lett. 59, 121 (1999). 103. Mallens, E.P.J., Hoebink, J.H.B.J., and Marin, G.B., Catal. Lett. 33, 291 (1995). 104. Wang, D., Dewaele, O., Groote, A.M., and Froment, G.F., J. Catal. 159, 418 (1996). 105. Hickman, D.A., and Schmidt, L.D., Catal. Lett. 17, 223 (1993). 106. Hofstad, K.H., Hoebink, J.H.B.J., Holmen, A., and Marin, G.B., Catal. Today 40, 157 (1998). 107. Dissanayake, D., Rosynek, M.P., and Lunsford, J.H., J. Catal. 132, 117 (1991). 108. Choudhary, V.R., Rane, V.H., and Rajput, A.M., Catal. Lett. 22, 289 (1993). 109. Drago, R.S., Jurczyk, K., Kob, N., Bhattacharyya, A., and Masin, J., Catal. Lett. 51, 177 (1998). 110. Claridge, J.B., Green, M.L.H., Tsang, S.C., York, A.P.E., Ashcroft, A.T., and Battle, P.D., Catal. Lett. 22, 299 (1993). 111. Tsipouriari, V.A., Zhang, Z., and Verykios, X.E., J. Catal. 179, 283 (1998). 112. Wang, J.K., Hu, Y.H., Weng, W.Z., Zheng, B.L., and Wan, H.L., Acta Chim. Sinica 54, 869 (1996). 113. Tang, S., Lin, J., and Tan, K.L., Catal. Lett. 51, 169 (1998). 114. Takehira, K., Shishido, T., and Kondo, M., J. Catal. 207, 307 (2002). 115. Zhu, T.L., and Flytzani-Stephanopoulos, M., Appl. Catal. A: Gen. 208, 403 (2001). 116. Lago, R., Bini, G., Pena, M.A., and Fierro, J.L.G., Third world congress on oxidation catalysis. Stud. Surf. Sci. Catal. 110, 721 (1997). 117. Cao, L., Chen, Y., and Li, W., Natural gas conversion IV. Stud. Surf. Sci. Catal. 107, 467 (1997). 118. Liu, S.L., Xiong, G.X., Sheng, S.S., Miao, Q., and Yang, W.S., Natural gas conversion V. Stud. Surf. Sci. Catal. 119, 747 (1998). 119. Wang, J.G., Liu, C.J., Zhang, Y.P., Zhu, X.L., Zou, J.J., Yu, K.L., and Eliasson, B., Chem. Lett. 1068 (2002). 120. Shiozaki, R., Andersen, A.G., Hayakawa, T., Hamakawa, S., Suzuki, K., Shimizu, M., and Takehira, K., Third world congress on oxidation catalysis. Stud. Surf. Sci. Catal. 110, 701 (1997). 121. Lezaun, J., Gomez, J.P., Blanco, M.D., Cabrera, I., Pena, M.A., and Fierro, J.L.G., Natural gas conversion V. Stud. Surf. Sci. Catal. 119, 729 (1998). 122. Zhang, Y.H., Xiong, G.X., Sheng, S.S., and Yang, W.S., Catal. Today 63, 517 (2000). 123. Nichio, N., Casella, M., Ferretti, O., Gonzalez, M., Nicot, C., Moraweck, B., and Frety, R., Catal. Lett. 42, 65 (1996). 124. Lee, K.M., and Lee, W.Y., Catal. Lett. 83, 65 (2002). 125. Hayakawa, T., Harihara, H., Andersen, A.G., Suzuki, K., Yasuda, H., Tsunoda, T., Hamakawa, S., York, A.P.E., Yoon, Y.S., Shimizu, M., and Takehira, K., Appl. Catal. A: Gen. 149, 391 (1997).

342

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

126. Claridge, J.B., Green, M.L.H., Tsang, S.C., York, A.P.E., Ashcroft, A.T., and Battle, P.D., Catal. Lett. 22, 299 (1993). 127. Poirier, M.G., Trudel, J., and Guary, D., Catal. Lett. 21, 99 (1993). 128. Boucouvalas, Y., Zhang, Z.L., and Verykios, X.E., Catal. Lett. 27, 131 (1994). 129. Hu, Y.H., and Ruckenstein, E., J. Catal. 158, 260 (1996). 130. (a) Hu, Y.H., and Ruckenstein, E., J. Phys. Chem. B 102, 230 (1998); (b) Hu, Y.H., and Ruckenstein, E., Acc. Chem. Res. 36, 791 (2003). 131. Hu, Y.H., and Ruckenstein, E., J. Phys. Chem. A 102(1), 10568 (1998). 132. Au, C.T., Liao, M.S., and Ng, C.F., J. Phys. Chem. A 102, 3959 (1998). 133. Papp, H., Schuler, P., and Zhuang, Q., Catal. Lett. 3, 299 (1996). 134. Hu, Y.H., and Ruckenstein, E., Catal. Lett. 34, 41 (1995). 135. Au, C.T., Liao, M.S., and Ng, C.F., Chem. Phys. Lett. 267, 44 (1997). 136. Buyevskaya, O.V., Walter, K., Wolf, D., and Baerns, M., Catal. Lett. 38, 81 (1996). 137. Au, C.T., and Wang, H.Y., J. Catal. 167, 337 (1997). 138. VanLooij, F., and Geus, J.W., J. Catal. 168, 154 (1997). 139. Au, C.T., Hu, Y.H., and Wan, H.L., Catal. Lett. 27, 199 (1994). 140. Au, C.T., Hu, Y.H., and Wan, H.L., Catal. Lett. 36, 159 (1996). 141. Campbell, R.A., Lenz, J.P., and Goodman, D.W., Catal. Lett. 17, 39 (1993). 142. Li, C., Yu, C.C., and Shen, S.K., Catal. Lett. 75, 183 (2001). 143. Tsipouriari, V.A., Zhang, Z., and Verykios, X.E., J. Catal. 179, 283 (1998). 144. Elmasides, C., and Verykios, X.E., J. Catal. 203, 477 (2001). 145. Weng, W.Z., Yan, Q.G., Luo, C.R., Liao, Y.Y., and Wan, H.L., Catal. Lett. 74, 37 (2001). 146. Nakagawa, K., Ikenaga, N., Teng, Y.H., Kobayashi, T., and Suzuki, T., J. Catal. 186, 405 (1999). 147. Mallens, E.P.J., Hoebink, J.H.B.J., and Marin, G.B., J. Catal. 167, 43 (1997). 148. Au, C.T., Ng, C.F., and Liao, M.S., J. Catal. 185, 12 (1999). 149. Shustorovich, E., Adv. Catal. 37, 101 (1990). 150. Tang, S., Lin, J., and Tan, K.L., Catal. Lett. 55, 83 (1998). 151. Hu, Y.H., and Ruckenstein, E., Catal. Lett. 57, 167 (1999). 152. Wang, H.Y., and Ruckenstein, E., J. Phys. Chem. 103, 11327 (1999). 153. Hu, Y.H., and Feeley, J.S., AIChE J. 49, 3253 (2003). 154. Ji, Y.Y., Li, W.Z., Xu, H.Y., and Chen, Y.X., Catal. Lett. 71, 45 (2001). 155. Hayakawa, T., Andersen, A.G., Shimizu, M., Suzuki, K., and Takehira, K., Catal. Lett. 22, 307 (1993). 156. Ruckenstein, E., and Hu, Y.H., Appl. Catal. A: Gen. 183, 85 (1999). 157. Wang, D.Z., Dewaele, O., DeGroote, A.M., and Froment, G.F., J. Catal. 159, 418 (1996). 158. Schouten, S.C., Gijzeman, O.L.J., and Bootsma, G.A., Bull. Soc. Chim. Belg. 88, 541 (1979). 159. Bartholomew, C.H., Catal. Rev.—Sci. Engng 24, 67 (1982). 160. Baker, R.T.K., and Harris, P.S., in “Chemistry and Physics of Carbon” (P.L. Walker Jr., Ed.), Vol. 14, p. 83. Marcel Dekker, New York, 1979. 161. Baukloh, W., Chatterjee, B., and Das, P.P., Trans. Indian Inst. Metals 4, 271 (1950). 162. Kehrer, V.J., and Leidheiser, H. Jr., J. Phys. Chem. 58, 550 (1954). 163. Rostrup-Nielsen, J.R., J. Catal. 27, 343 (1972). 164. Renshaw, G.D., Roscoe, C., and Walker, P.L., J. Catal. 22, 394 (1971). 165. Grenga, H.E., and Lawless, K.R., J. Appl. Phys. 43, 1508 (1972). 166. Renshaw, G.D., Roscoe, C., and Walker, P.L., J. Catal. 18, 164 (1970). 167. Eizenberg, M., and Blakely, J.M., Surf. Sci. 82, 228 (1979). 168. Reitmeier, R.E., Atwood, K., Bennet, H.A. Jr., and Baugh, H.M., Ind. Engng Chem. 40, 620 (1948).

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

343

169. White, G.A., Roszkowski, T.R., and Stanbridge, D.W., Hydrocarbon Process. 54, 130 (1975). 170. Tokunaga, O., and Ogasawara, S., React. Kinet. Catal. Lett. 39, 69 (1989). 171. (a) Ruckenstein, E., and Hu, Y.H., Chem. Innovat. 30, 39 (2000); (b) Hu, Y.H., and Ruckenstein, E., J. Catal. 184, 298 (1999); (c) Hu, Y.H., and Ruckenstein, E., J. Catal. 163, 306 (1996); (d) Hu, Y.H., and Ruckenstein, E., Langmuir 13, 2055 (1997). 172. Hu, Y.H., and Ruckenstein, E., Catal. Rev.—Sci. Engng 44, 423 (2002). 173. Rostrup-Nielsen, J.R., Stud. Surf. Sci. Catal. 68, 85 (1991). 174. (a) Udengaar, N.R., Hansen, J.-H.B., Hanson, D.C., and Stal, J.A., Oil Gas J. 90, 62 (1992); (b) Dibbern, H.C., Olesen, P., Rostrup-Nielsen, J.R., Totrrup, P.B., and Udengaard, N.R., Hydrocarbon Process. 65, 71 (1986). 175. Rostrup-Nielsen, J.R., and Hansen, J.H.B., J. Catal. 144, 38 (1993). 176. Yamazaki, O., Nozaki, T., Omata, K., and Fujimoto, K., Chem. Lett. 1953 (1992). 177. Zhang, Z.L., and Verykios, X.E., Catal. Today 21, 589 (1994). 178. Horiuchi, T., Sakuma, K., Fukui, T., Kubo, Y., Osaki, T., and Mori, T., Appl. Catal. A: Gen. 144, 111 (1996). 179. Kim, G.J., Cho, D.-S., Kim, K.-H., and Kim, J.-H., Catal. Lett. 28, 41 (1994). 180. Inui, T., Catalysis, Vol. 16, p. 133, The Royal Society of Chemistry, Cambridge, 2002. 181. Perera, J.S.H., Couves, J.W., Sankar, G., and Thomas, J.M., Catal. Lett. 11, 219 (1991). 182. Qin, D., and Lapszewicz, J., Catal. Today 21, 551 (1994). 183. Solymosi, F., Kustan, Gy., and Erohelyi, A., Catal. Lett. 11, 149 (1991). 184. Erdohelyi, A., Fodor, K., and Solymosi, F., Stud. Surf. Sci. Catal. 107, 525 (1997). 185. Bitter, J.H., Hally, W., Seshan, K., van Ommen, J.G., and Lercher, J.A., Catal. Today 29, 349 (1996). 186. Bitter, J.H., Seshan, K., and Lercher, J.A., J. Catal. 176, 93 (1998). 187. O’Connor, A.M., and Ross, J.R.H., Catal. Today 46, 203 (1998). 188. Bitter, J.H., Seshan, K., and Lercher, J.A., J. Catal. 183, 336 (1999). 189. Souza, M.M.V.M., Aranda, D.A.G., and Schmal, M., Ind. Engng Chem. Res. 41, 4681 (2002). 190. Chen, Y.Z., Liaw, B.J., and Lai, W.H., Appl. Catal. A: Gen. 230, 73 (2002). 191. Stagg, S.M., and Resaco, D.E., Stud. Surf. Sci. Catal. 111, 543 (1997). 192. Sigl, M., Bradford, M.C.J., Kno¨zinger, H., and Vannice, M.A., Top. Catal. 8, 211 (1999). 193. Stagg-Williams, S.M., Noronha, F.B., Fendley, G., and Resasco, D.E., J. Catal. 194, 240 (2000). 194. Kim, J.H., Suh, D.J., Park, T.J., and Kim, K.L., Appl. Catal. A: Gen. 197, 191 (2000). 195. Bhattacharyya, A., and Chang, V.W., Stud. Surf. Sci. Catal. 88, 207 (1994). 196. Halliche, D., Bouarab, R., Cherifi, O., and Bettahar, M.M., Catal. Today 29, 373 (1996). 197. (a) Ruckenstein, E., and Hu, Y.H., J. Catal. 162, 230 (1996); (b) Hu, Y.H., and Ruckenstein, E., J. Phys. Chem. B 101, 7563 (1999). 198. Wang, S., and Lu, G.O.M., Appl. Catal. B: Environ. 16, 269 (1998). 199. Yan, Z.-F., Ding, R.-G., Song, L.-H., and Qian, L., Energy Fuels 12, 1114 (1998). 200. Wang, S., and Lu, G.Q., Appl. Catal. A: Gen. 169, 271 (1998). 201. Wang, S., and Lu, G.Q., Ind. Engng Chem. Res. 38, 2615 (1999). 202. Ito, M., Tagawa, T., and Goto, S., Appl. Catal. A: Gen. 177, 15 (1999). 203. Xu, Z., Li, Y.M., Zhang, J.Y., Chang, L., Zhou, R.Q., and Duan, Z.T., Appl. Catal. A: Gen. 210, 45 (2001). 204. Wang, S.B., and Lu, G.Q., Energy Fuels 12, 1235 (1998). 205. Chen, Y.G., and Ren, J., Catal. Lett. 29, 39 (1994). 206. Sridhar, S., Sichen, D., and Seetharaman, S., Z. Metallkd. 85, 9 (1994).

344

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

207. Kim, J.H., Suh, D.J., Park, T.J., and Kim, K.L., Natural gas conversion V. Stud. Surf. Sci. Catal. 119, 771 (1998). 208. Osaki, T., Horiuchi, T., Sugiyama, T., Suzuki, K., and Mori, T., Catal. Lett. 52, 171 (1998). 209. Hayashi, H., Murata, S., Tago, T., Kishida, M., and Wakabayashi, K., Sekiyu Gakkaishi— J. Jpn Petrol. Inst. 44, 334 (2001). 210. Bhattacharyya, A., Chang, V.W., and Schumacher, D.J., Appl. Clay Sci. 13, 317 (1998). 211. Mori, T., Osaki, T., Horiuchi, T., Sugiyama, T., and Suzuki, K., Catalyst deactivation 1999. Stud. Surf. Sci. Catal. 365 (1999). 212. Lemonidou, A.A., Goula, M.A., and Vasalos, I.A., Catal. Today 46, 175 (1998). 213. Osaki, T., and Mori, T., J. Catal. 204, 89 (2001). 214. Wang, S.B., and Lu, G.Q.M., J. Chem. Technol. Biotechnol. 75, 589 (2000). 215. Choi, J.S., Moon, K.I., Kim, Y.G., Lee, J.S., Kim, C.H., and Trimm, D.L., Catal. Lett. 52, 43 (1998). 216. Seok, S.H., Han, S.H., and Lee, J.S., Appl. Catal. A: Gen. 215, 31 (2001). 217. Quincoces, C.E., de Vargas, S.P., Grange, P., and Gonzalez, M.G., Mater. Lett. 56, 698 (2002). 218. Crisafulli, C., Scire, S., Maggiore, R., Minico, S., and Galvagno, S., Catal. Lett. 59, 21 (1999). 219. Slagtern, A., Olsbye, U., Blom, R., Dahl, I.M., and Fjellvag, H., Appl. Catal. A: Gen. 165, 379 (1997). 220. Li, W.Y., Feng, J., and Xie, K.C., React. Kinet. Catal. Lett. 64, 381 (1998). 221. Wang, S.B., and Lu, G.Q., Appl. Catal. B: Environ. 19, 267 (1998). 222. Xu, G.L., Shi, K.Y., Gao, Y., Xu, H.Y., and Wei, Y.D., J. Mol. Catal. A: Chem. 147, 47 (1999). 223. Kroll, V.C.H., Swaan, H.M., and Mirodatos, C., J. Catal. 161, 409 (1996). 224. Kroll, V.C.H., Delichure, P., and Mirodatos, C., Kinet. Catal. 37, 698 (1996). 225. Takayasu, O., Takegahara, Y., and Matsuura, I., Energy Conserv. Mgmt 36, 597 (1995). 226. Quincoces, C.E., de Vargas, S.P., Diaz, A., Montes, M., and Gonzalez, M.G., Natural gas conversion V. Stud. Surf. Sci. Catal. 119, 837 (1998). 227. Zhang, Z.L., and Verykios, X.E., J. Chem. Soc., Chem. Commun. 71 (1995). 228. Zhang, Z., and Verykios, X.E., Appl. Catal. A: Gen. 138, 109 (1996). 229. Wang, S.B., and Lu, G.Q., Energy Fuels 12, 248 (1998). 230. Ruckenstein, E., and Hu, Y.H., J. Catal. 161, 55 (1996). 231. Liu, B.S., and Au, C.T., Catal. Lett. 85, 165 (2003). 232. Hally, W., Bitter, J.H., Seshan, K., Lercher, J.A., and Ross, J.R.H., Catalyst deactivation 1994. Stud. Surf. Sci. Catal. 167 (1994). 233. Li, X.S., Chang, J.S., Lee, F.K., and Park, S.E., React. Kinet. Catal. Lett. 67, 383 (1999). 234. Li, X.S., Chang, J.S., and Park, S.E., React. Kinet. Catal. Lett. 67, 375 (1999). 235. Li, X.S., Chang, J.S., Tian, M.Y., and Park, S.E., Appl. Organomet. Chem. 15, 109 (2001). 236. Lercher, J.A., Bitter, J.H., Halley, W., Niessen, W., and Seshan, K., in “Proceedings of the 11th International Congress on Catalysis” (J.W. Hightower, W.N. Delgass, E. Iglesia and A.T. Bell, Eds.), Stud. Surf. Sci. Catal., 101, p. 463. Elsevier, Amsterdam, 1996. 237. Wei, J.M., Xu, B.Q., Li, J.L., Cheng, Z.X., and Zhu, Q.M., Appl. Catal. A: Gen. 196, L167 (2000). 238. Tomishige, K., Yamazaki, O., Chen, Y., Yokoyama, K., Li, X., and Fujimoto, K., Catal. Today 45, 35 (1998). 239. Ruckenstein, E., and Hu, Y.H., Appl. Catal. A: Gen. 133, 149 (1995). 240. Osaki, T., Masuda, H., Horiuchi, T., and Mori, T., Catal. Lett. 34, 59 (1995).

Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345

345

241. Guerrero-Ruiz, A., Sepulveda-Escribano, A., and Rodriguez-Ramos, I., Catal. Today 21, 545 (1994). 242. Osaki, T., Horiuchi, T., Suzuki, H., and Mori, T., Appl. Catal. A: Gen. 155, 229 (1997). 243. Dibbern, H.C., Olesen, P., Rostrup-Nielsen, J.R., Tottrup, P.B., and Udengaard, N.R., Hydrocarbon Process. 65, 71 (1986). 244. Hayakawa, T., Suzuki, S., Nakamura, J., Uchijima, T., Hamakawa, S., Suzuki, K., Shishido, T., and Takehira, K., Appl. Catal. A: Gen. 183, 273 (1999). 245. Chang, J.S., Park, S.E., and Lee, K.W., Stud. Surf. Sci. Catal. 84, 1587 (1994). 246. Park, S.E., Nam, S.S., Choi, M.J., and Lee, K.W., Energy Conserv. Mgmt 36, 573 (1995). 247. Luo, J.Z., Gao, L.Z., Ng, C.F., and Au, C.T., Catal. Lett. 62, 153 (1999). 248. Chang, J.S., Park, S.-E., and Chon, H., Appl. Catal. A: Gen. 145, 111 (1996). 249. Choudhary, V.R., Uphade, B.S., and Mamman, A.S., Micropor. Mesopor. Mater. 23, 61 (1998). 250. Suzuki, I., Okajima, S., and Seya, K., J. Phys. Earth 27, 69 (1979). 251. Arai, H., and Machida, M., Catal. Today 10, 81 (1991). 252. Gadalla, A.M., and Sommer, M.E., J. Am. Ceram. Soc. 72, 683 (1989). 253. Chen, Y., Tomishige, K., and Fujimoto, K., Chem. Lett. 999 (1997). 254. Takayasu, O., Soman, C., Takegahara, Y., and Matsuura, I., Stud. Surf. Sci. Catal. 88, 281 (1994). 255. Takayasu, O., Hongo, N., and Matsuura, I., Stud. Surf. Sci. Catal. 77, 305 (1993). 256. Swaan, H.M., Kroll, V.C.H., Martin, G.A., and Mirodatos, C., Catal. Today 21, 545 (1994). 257. (a) Hu, Y.H., and Ruckenstein, E., Catal. Lett. 43, 71 (1997); (b) Ruckenstein, E., and Hu, Y.H., Catal. Lett. 51, 183 (1998). 258. Hu, Y.H., and Ruckenstein, E., Catal. Lett. 36, 145 (1996). 259. Ruckenstein, E., and Hu, Y.H., Appl. Catal. A: Gen. 154, 185 (1997). 260. Chen, Y., Tomishige, K., Yokoyama, K., and Fujimoto, K., Appl. Catal. A: Gen. 165, 335 (1997). 261. Tomishige, K., Yamazaki, O., Chen, Y., Yokoyama, K., Li, X., and Fujimoto, K., Catal. Today 45, 35 (1998). 262. Tomishige, K., Himeno, Y., Matsuo, Y., Yoshinaga, Y., and Fujimoto, K., Ind. Engng Chem. Res. 39, 1891 (2000). 263. Chen, Y.G., Tomishige, K., and Fujimoto, K., Appl. Catal. A: Gen. 161, L11 (1997). 264. Ruckenstein, E., and Wang, H.Y., Appl. Catal. A: Gen. 204, 257 (2000). 265. Ruckenstein, E., and Wang, H.Y., J. Catal. 205, 289 (2002). 266. Ruckenstein, E., and Wang, H.Y., Catal. Lett. 73, 99 (2001).

This Page Intentionally Left Blank

Index 1-butene, and alkylation isomerization, 240 2-butene, and alkylation isomerization, 240 4,6-Dimethyldibenzothiophene (DMDBT), 98– 99 5A Zeolite, 332 A Acceptor-doped perovskite oxides, 307 –308 Acid catalysis, 105 – 116 Acid runaway condition, 254 Acid-soluble oil (ASO), 239 Acid strength and alkylation reaction, 256 – 260 in TS-1, 26 –28 Activation energy and adsorption by zeolites, 256 of cracking, 249 and hydride transfer, 243 Active sites. see Titanosilicate surface structures Adamantane, 246 AEM (analytical EM), 177 Alcohol oxidation, 100 Aliphatic compound hydroxylation, 85 –89 Alkanes, cracking, 249 Alkene epoxidation, 70 and O –O bond cleavage, 138 –140 Alkene space velocity (OSV), 274 –275 AlkyCleanTM, 286 Alkyl hydroperoxides, 80 –81

Alkylation, 235– 236t about, 230 – 233 and alkane/alkene ratio, 274– 275 alkene addition, 239 – 241 and alkene feed composition, 276 – 278 and alkene space velocity, 274–275 and aluminum content, 262 of carbenium ions with isobutane, 241 and coke formation, 245 –246 cracking, 247– 249 hydride transfer, 242 –246 initiation steps, 237– 239 isobutylene protonation, 238f isomerization, 239– 241 isomerization catalysts, 240 – 241 isomerization for TMP isomers, 241f oligomerization, 247– 249 pathways to allylic/cyclic compounds, 251 –252 and process parameters, 272t product distribution, 234 –237 reaction rate and interface area, 253 – 254 reaction temperature, 272– 274 refinery process unit, 232f self-alkylation, 249 – 250 solid-acid processes, 283 – 288 sulfated zirconia, 267 – 271 Alkylation as industrial process ConocoPhillips HF-catalyzed process, 281 – 282

347

348

Index

ExxonMobil auto-refrigerated alkylation, 280 Haldor Topsøe FBATM process, 287– 288 LURGI EUROFUEL, 286 –287 ReVapTM alkylation process, 282 Stratco Alkysafe, 281 sulfuric-acid as catalyst, 278 – 281 UOP Alkad process, 283 UOP AlkyleneTM, 285 – 286 Alkylation catalysts. see also Zeolites acid strength, 257– 258 heteropolyacids, 268– 269 liquid acid properties, 253– 255 Nafion-H, 269 –270 rare earth exchanged zeolites, 263– 264 and silicon/aluminum ratio, 261–263 sulfated metal oxides, 268 supported metal halides, 270– 271 VTM-A, 266 ALPO catalysts, 221 AlPO4, 332 Aluminum, and alkylation, 260 – 261 Aluminum chloride, and isomerization, 241 Ammoximation, 92 Analytical electron microscope (AEM), 191 – 192 Aromatic compound, hydroxylation, 89– 90 Aromatic polycarbonates, 108 Atomic resolution ETEM, 197 – 200 of butane oxidation, 203 –210 of nanorods, 210 Autothermal reforming of methane, 306 B Ba0.5Sr0.5Co0.8Fe0.2O32x membrane reactor, 309

Back-scattered electrons (BSE), 213– 217 Baeyer-Villiger (BV) oxidation, 102, 104t BEA and silicon/aluminum ratio, 261– 262 time-on-stream behavior, 265 Beckmann rearrangement, 106 Benzene hydroxylation, 134– 135 Benzoquinones, 101 b-scission, 247, 248– 249 Boron trifluoride, 271 Boudouard reaction, 321 Bright-field (BF) image, 179 Brønsted acid sites and catalytic action of titanosilicate molecular sieves, 56 creation during TS-1 and H2O2 interaction, 130 in TS-1, 26– 28 in zeolites, 256– 260 Brownmillerite membrane, 308 BSE and HAADF detector geometry, 215f Butane oxidation, 203– 210 C CvN cleavage reactions, 105 Carbon deposition, with Ni/Al2O3 catalysts, 325 – 327 Carbon dioxide as product of partial oxidation of methane, 316 – 318 Carbon monoxide as product of partial oxidation of methane, 316 – 318 Carbon monoxide disproportionation, 321

Index

Catalysts. See also reactions by name to locate references to catalysts for these reactions, 177 –178 deactivation by metal-support interactions, 201 defects and TEM analysis, 180 dislocation types, 181 glide shear plane mechanism, 208 –210 high-silica microporous SSZ-48 type, 187 –188 HRTEM analysis of L-type zeolites, 185 –186 HRTEM analysis of MAPO-36 microporous type, 186 – 189 intergrowths in zeolites, 188 –190 in situ analysis, 196 –200 solid in liquid environments, 210 –212 titanosilicate molecular sieves, 56, 132 –136 Catalysts (for conversion of methane) and carbon deposition in CO2 reforming, 322 –323 changes during partial oxidation of methane, 315– 316 Co/Al2O3, 331 MgO-containing catalysts, 332–337 MoS2, 331 Ni/Al2O3, 324 Ni/ZrO2, 330– 331 noble metal types for CO2 reforming, 323 –324 for partial oxidation of methane, 312 –314 reduced and oxidized for partial oxidation of methane, 320 – 321 WS2, 331– 332 Cathodoluminescence imaging, 195 –196 Ceramic membrane reactor, 308f

349

configuration for partial oxidation of methane, 310f Chlorided alumina, and alkene feed composition, 277 Chlorinated alumina, 270 Co/Al2O3 catalyst, 331 CO2 photoreduction, 120 –121 CO2 reforming of methane carbon deposition, 299 carbon deposition pathways, 321 – 322 catalyst and carbon deposition, 322 – 323 development of processes, 299 MgO-containing catalysts, 332 – 337 Ni/Al2O3 catalysts, 325– 327 noble metal catalysts, 323 –324 SPARG process, 332 Cobalt catalysts for partial oxidation of methane, 314 ConocoPhillips HF-catalyzed process, 281 – 282 Coordination sphere, expansion of TS-1, 31 –32, 38 Copper catalysts, 200, 203 Cracking and alkylation, 248 – 249 and reaction temperature, 272 –273 CTEM (conventional transmission electron microscope), 178 Cyclic carbonates, 107t –108t transesterification, 109 Cyclic voltametry of oxo-titanium species, 41 –42 Cyclohexane hydroxylation, 88 Cytochrome P450, 60 D Dark field (DF) image, 179, 193f DFT, and titanosilicate surface structure, 50

350

Index

Dialkene epoxidation, 71– 72t Diamagnetic peroxo/hydroperoxo species, 42 Diethyl malonate, 113t Diffuse reflectance UV-visible spectroscopy and concentrations of superoxo and hydroperoxo species, 44 for oxo-titanium species, 37 TS-1 (titanium silicate-1), 46f titanosilicate, 13t titanosilicate surface structure, 14f Dimethyl terephthalate (DMT), 110 Dimethylcarbonates, 110 Dimethylhexanes, 234, 248 Dioxygen, 83 Diphenyl carbonate (DPC), 108 Dislocation types, 181 DMDBT, 98– 99 E ECELLs, 196 – 200, 222 Edge dislocations, 181 EDX (energy dispersive X-ray spectroscopy), 182, 191 EFAL (extra-framework aluminum species), 260– 261 Effluent Refrigerated Sulfuric Acid Alkylation Process (Stratco), 278– 279 EFTEM (energy-filtered TEM), 176 Electron crystallography, 220 Electron diffraction (ED), 174 Electron microscopy (EM) aberration correction, 222 –223 analytical electron microscopy (AEM), 191 – 192 application to catalysis, 176– 177 beam damage, 222 in catalysis, 177 –178

cathodoluminescence imaging, 195– 196 challenges, 220– 223 charge-coupled device, 177 diffraction patterns, 178 electron-beam damage, 177 electron tomography, 212 –218 energy-filtered TEM, 218 –219 environmental scanning electron microscopy (ESEM), 212 ETEM, 196– 200 HRLVSEM, 195 image plate (IP), 177 imaging, 178– 179 lens aberrations, 222 methods, 176 –181 point resolution, 178 ray diagram, 180f resolution, 222 sample preparation, 176 spatial mapping, 192 STEM (scanning transmission EM), 177, 193– 195, 222 TEM imaging methods, 179– 181 theoretical procedures, 181 wet-ETEM, 210 – 212 Electron tomography about, 212– 218 BSE and HAADF, 213– 217 nanoparticle location, 218 ELNES (electron energy loss near-edge structure), 176 EMT acid strength and catalyst role in alkylation, 258 intergrowths in, 189 and rare earth exchange, 263 –264 SINTEF research, 266 Energy Gibbs free energy for methane transformations, 300t

Index

and hydride transfer, 243 potential energy profiles for hydride transfer, 244f Energy-filtered transmission electron microscopy, 218 – 219 Enthalpy of adsorption on zeolites, 256 –257f of hydride transfer step in alkylation, 244 of O – O cleavage in H2O2, 57 Environmental scanning electron microscopy, 212 Environmental transmission electron microscopy, 196 – 200 Epichlorohydrin, 62 Epoxidation and alkene structure, 70 –71 alkenes and alcohol functions, 72 –73t alkenes and alkanes, 72 alkenes and O – O bond cleavage, 138 –140 alkenes containing carbonyl groups, 81 –82 with alkyl hydroperoxides, 80 – 81 allyl alcohol, 80t, 125t catalyzed by mesoporous titanium silicates, 67 –70 and concentration of titanium oxo species, 129 of cyclohexene and silylation, 127t dialkenes, 71 –72t diastereoselectivity, 74– 75, 77 –78t diffusional constraints, 62-63t general features, 60 – 62 of hex-1-ene, 66t hydroperoxide involvement, 132 hydroxyl-assisted, 72 –74 of oleic acid, 67t and pH, 78– 80t reaction rates for alkenes, 71t side reactions, 75– 76

351

stereospecificities, 62 and Ti-silicate structure, 65 –67 of unsaturated cyclic terpenes, 69t using dioxygen, 83 using urea-H2O2, 82 yields, 62 EPR spectroscopy, oxo-titanium species, 42 –49 Ester transesterification, 110 ETEM, 196– 200 Ethane/ethene hydride transfer, 243 Ethene oxidation, 131 –132 Ether oxidation, 100– 101 Ethylacetoacetate, 111 –112, 114t ETS-10 synthesis, 154 – 156 vibrational spectroscopy, 25 Euro-TS-1, diffusional constraints, 63t EXAFS oxo-titanium species, 39– 41 and titanosilicate surface structure, 50 – 51t ExxonMobil auto-refrigerated alkylation process, 280 F FAU zeolites and rare earth exchange, 263– 264 SINTEF research, 266 Faujasite, 189, 263 –264 FEG HRTEM, 221 FEG-STEM, 194 Fluorinated alumina, 270 G g·b product analysis, 180 –181 Glide shear mechanism, 208 – 210 H H-BEA and Brønsted acid sites, 258

352

Index

and EFAL species, 260 – 261 and hydride transfer, 246 and silicon/aluminum ratio, 261– 262 H-EMT, acid strength and catalyst role in alkylation, 258 H-FAU, acid strength and catalyst role in alkylation, 258 H-SAPO-37, 257 H2O2, and stabilization of Ti(O2) complex, 34 H2O2 anhydrous source, 82 as catalyst in homogeneous phase, 58– 60 conversion using Ti-SBA-15 and Ti-MMM, 97t as oxidant, 56 – 57 and oxo-titanium species, 33 replacement in TS-1, 8 HAADF, 193f, 194 combined with BSE, 215 STEM-HAADF image, 217f and topography of nanoparticles, 213– 217 Haldor Topsøe FBATM process, 287– 288 Heterolytic catalysis, 58, 137– 138 Heteropolyacids, 268– 269 High-angle annular dark-field (HAADF) miscroscopy, 193f –194 High-resolution STEM (HRSTEM), 193– 194 High-voltage EM, 197 HMCM-22, 264 Homolytic catalysis, 58, 137 – 138 HRLVSEM, 195 HRTEM (high-resolution transmission EM), 176 –177 aberration-corrected, 222 description, 181 –182

development, 184– 185 germanium silicate, 184f intergrowths in zeolites, 188– 190 L-type zeolite catalysts, 185 –186 MAPO-36 microporous catalysts, 186– 189 of MAPO catalysts, 192 nanopores in silica, 219f optimizing images, 182 – 183 sample preparation, 221 samples as weak phase objects (WPO), 182 SSZ-48 catalysts, 187 – 188 Hydride transfer in alkylation process, 242– 246 and acid strength, 258 from alkenes, 246 energy barrier, 243 of ethane/ethene, 243 gas-phase and liquid-phase, 242– 243 potential energy profiles, 244f reaction enthalpy, 244 – 245 in zeolites, 243– 244 Hydrocracking, 248– 249 Hydrofluoric acid (HF) and alkylation initiation, 237 drawbacks as catalyst, 231, 233, 251 as isomerization catalyst, 241 strength and alkylation product quality, 254 –255 Hydrogen transfer, 249 Hydroperoxo Ti species, 36– 37 transformation to superoxo species, 47– 49 Hydroxyl-assisted epoxidation, 72– 74 Hydroxylation advantages of mixed-phase catalyst, 88 of aliphatic compounds, 85– 89

Index

aromatic compounds, 89– 90 general features, 83 – 85 I In situ ETEM, 203 gas – catalyst reactions, 201 – 203 gas – solid reactions, 196 – 200 VPO in n-butane, 206f Inorganic membrane reactor, 306 Ir/Al2O3 catalyst, 305 IR spectroscopy and irradiation of TS-1 (H2O2 loaded), 38– 39 and Lewis acidity in TS-1, 28 oxo-titanium species, 34 – 39 Ti-MMM, 96f Ti-SBA-15, 96f Isoalkanes, and sulfuric acid as alkylation catalyst, 251 Isobutane, and hydride transfer, 242 Isobutylene oligomerization, 248 and self-alkylation, 250 Isobutylene protonation, 238f Isopentane alkylation, 241 ITQ-7, 265 –266 K Ketones, 102 L La-promoted catalysts, 324 La0.2Sr0.8Fe0.8Cr0.2Ox membrane, 307 –308 La(12x)AxFe0.8Co0.2O32x perovskite dense membrane reactor, 309 Lanthanide ruthenium oxide, 300 Laser-based spectroscopy, 175 Lewis acid sites

353

and catalytic action of titanosilicate molecular sieves, 56 deactivation by water, 130 in TS-1, 28 –32 in VPO, 208 in zeolites, 256, 260 –261 Lewis basicity, and carbon deposition in CO2 reforming, 322 Li/MgO, 300 Ligand-to-metal charge transfer (LMCT) and coordination sphere expansion in TS-1, 31 – 32 in DRUV of oxo-titanium species, 37f and titanosilicate surface structure, 12 Low-temperature coke, 245 Low-voltage, high resolution SEM, 223 LURGI EUROFUEL, 286 –287 M MAPO-36 microporous catalysts, 186 – 189 MAPO catalysts HRTEM (high-resolution transmission EM) analysis, 192 Mass spectrometry, 175 MCM-22, 264 –265 MCM-25, 271 MCM-36, 264, 266 MCM-41, 268 –269 MCM-48, 220f MCM-49, 266 Mesoporous aluminosilicates, 266 Mesoporous titanium silicates, epoxidation reactions, 67– 70 Mesoporous TS-1, 64 influence of silylation, 124, 126t

354

Index

Metal-substituted aluminum phosphate (MAPO-36) microporous catalysts, 186 –189 HRTEM (high-resolution transmission EM) analysis, 192 Methane to synthesis gas. see also CO2 reforming of methane; partial oxidation of methane development of processes, 298– 301 Gibbs free energy for reactions, 300t Methyl-tertiary-butyl ether (MTBE), 231 Methylheptanes, 234 Metropolis Monte Carlo method, and Ti4þ distribution, 53 MgO-containing catalysts, 332– 337 Michael addition reactions, 110, 113, 117t MIEC (mixed ionic/electronic conductors) membrane, 306 –311 Molecular sieves analysis by EM, 221 early investigations, 5 MTBE, 231 Mukaiyama-type aldol reactions, 110, 116t Mulliken population analysis, and titanosilicate surface structure, 50 Multiple alkylate, 247 N n-butenes, and alkylation initiation, 238– 239 Nafion-H, 269 –270 Nanoparticles location in nanoporous solids, 218 Nanophase Pt/TiO2 catalysts, 202f Nanoporous solids, 175 Nanorods, 210 – 211f Ni/Al2O3 catalysts, 325 –327 Ni/La2O3 catalysts, 328 –330

Ni/SiO2 catalysts, 327 –328 Ni/ZrO2 catalysts, 330 –331 Nickel catalyst changes during partial oxidation of methane, 316 catalysts for partial oxidation of methane, 312 – 314 particle size and carbon deposition, 322 reduced and oxidized for partial oxidation of methane, 320 –321 and thermal gradients in partial oxidation of methane, 302, 305 NiO/Al2O3 catalyst, 305 NiO/MgO catalyst, 305 NiO/SiO2 catalyst, 305 Nitrogen-containing compound oxidation, 90 –93 NO decomposition, 121 – 122 O O –O bond cleavage in alkene epoxidation, 138 –140 Octane hydroxylation, 88 Octane number, 235 –236 for oligomerization and cracking products, 248 Oligomerization and alkylation, 247 –249 and reaction temperature, 272– 273 and strength of acid catalyst, 258 and triflic acid, 255 ONIOM method, and titanosilicate surface structure, 54 –55 OSV (alkene space velocity), 274– 275 Oxidation of alcohols, 100 Baeyer-Villiger (BV) oxidation, 102 cyclohexanone, 104t

Index

of ethers, 100– 101 influence of solvents, 122– 124 of n-alkanes, 85t nitrogen-containing compounds, 90 –93 of phenols, 101 – 102 sulfur-containing compounds, 93 –99 of TMP, 103f Oxidative coupling of methane, 299 –300 Oxidative dehydrogenation, 115, 119t Oxo-titanium species characteristics on TS-1 with aqueous H2O2, 143t concentrations of superoxo and hydroperoxo species, 43– 44, 47 cyclic voltametry, 41– 42 EPR spectroscopy, 42– 49 formation, 33 free radical oxidation mechanism, 42 H2 þ O2 as oxidant, 42 O –O stretch, 36 paramagnetic oxygen species, 42 peroxide species structure, 38 –39 role in epoxidation reactions, 132 –133 structure and activity, 128– 136 transformation to superoxo species, 47 –49 UV-visible spectroscopy, 34 –35f vibrational spectroscopy, 34– 39 X-ray absorption spectroscopy, 39 –41 Oxygen separation methods, 306 –312 P P/O (paraffin/olefin) ratio, 274 –275 Paramagnetic superoxo-titanium species, 47

355

Partial dislocations, 181 Partial oxidation of methane catalyst changes during reaction, 315 – 316 catalyst composition and temperature profile, 303 catalysts, 312– 314 ceramic membrane reactor, 308f, 310f CHx species formation, 318 –320 deuterium isotope effects, 319 development of processes, 299 fluidized-bed reactor, 310– 311 hot spots in catalyst beds, 301 – 306 inorganic membrane process, 306 La0.2Sr0.8Fe0.8Cr0.2Ox membrane, 307 – 308 lanthanide ruthenium oxide catalyst, 300 – 301 major challenges, 301 minimizing thermal gradients, 304 Ni/La2O3 catalysts, 328– 330 Ni/SiO2 catalysts, 327 –328 oxygen purification methods, 306 – 312 primary product determination, 316 – 318 rate-determining steps, 318 –320 reaction pathways, 301, 314– 315 reaction temperatures, 301 with reduced and unreduced catalysts, 320 –321 temperature and catalyst composition, 303 using reduced and oxidized catalysts, 320 –321 Pd/C catalyst, 214 Perdeuterioisobutane, 238 Perovskite-based oxygen transport, 306 – 311 Peroxo-titanium species, 34

356

Index

structure, 52f Phenol oxidation, 101 – 102 Photocatalysis and degradation of pollutants, 116 –120 Photocatalytic synthesis, 120 –121 Pillared layered silicate, 271 Pinacols, 114, 118t Platinum-containing catalysts, 323 Polycarbonate precursors, 106 Polyethene terephthalate (PET), 110 Propane, in two-step alkylation, 241 Propene epoxidation, 129 –131 Propene oxide, 61 –62 Pyridine, and acid strength measurement, 259 R Raman spectra selection rules, 21– 22 Ti-MMM, 95t and TS-1 peroxide species structure, 38– 39 Rare earth exchanged faujasites (REHY, REUSY), 263 –264 Rare earth metals, as promoters for Ni/Al2O3 catalysts, 327 Red oil, 239 Redox potentials of transition metal ions, 59t REHX, 276 ReVapTM alkylation process, 282 Rhodium catalysts, 324 RON (research octane number) for oligomerization and cracking products, 248 values of alkanes, 235 – 236t Ruthenium catalysts, 200 Rutherford detector, 194

S Sb –SnO2 catalysts, 196 SBA-15 type titanium silicates, 93 Scanning probe methods, 175 Screw dislocations, 181 Self-alkylation, 249 – 250 Silica nanopores, 219f Silicalite-1 orthorhombic structure, 11f Silicon/aluminum ratio, 261 –263 Silylation, 124 – 127 SMSI deactivation, 201 Soft coke, 245 SPARG process, 322, 332 Spectroscopic analysis of titanosilicate surface structures EPR spectroscopy, 22 – 26 and Lewis acidity in TS-1, 28 –32 photoluminescence spectroscopy, 15 UV-visible spectroscopy, 12– 14 vibrational spectroscopy, 18 – 22 X-ray absorption spectroscopy, 15– 18 Sr– Fe– Co – O mixed oxide membranes, 307– 308 SSZ-48 catalysts, 187– 188 Steam reforming of methane, 298 combined with partial oxidation, 306 STEM-HAADF image, 217f STEM (scanning transmission EM), 177, 193– 195 aberration-corrected, 222 Stoichiometry, and AEM, 191 Stratco Alkysafe process, 281 Stratco Contactor reactor, 279f Sulfated zirconia, 233, 267 –271 Sulfur compound removal, 98 – 99 Sulfuric acid (H2SO4) and alkylation initiation, 237 drawbacks as catalyst, 231, 233, 251

Index

in industrial processes, 278 –281 as isomerization catalysts, 240 and oligomerization, 247 and self-alkylation, 250 strength and alkylation product quality, 254 Superoxo titanium, 34 and catalytic activity, 132 – 133 EPR data, 45t EPR spectroscopy, 42 transformation from hydroperoxo/ peroxo species, 47 –49 Synthesis of titanium silicate molecular sieves confined space method, 145 dissolved (hydrolyzed) titanium method, 144, 148 ETS-10, 154 –156 microwave irradiation technique, 145 mixed alkoxide method, 144, 148, 151 prehydrolysis method, 144, 149–150 reduced crystallization time, 144 Ti-beta, 153 –154 Ti-HMS, 147, 157 Ti-MCM-41, 147, 154– 156 Ti-MCM-48, 147, 157 Ti-SBA-15, 147, 158 Ti-ZSM-48, 152 TS-2, 151– 152 using TiF4, 150 wetness impregnation method, 144, 148 T Tar formation, 101 TEM diffraction contrast technique, 180 energy-filtered, 218 – 219 imaging methods, 179– 181 sample damage, 216

357

Temperature and alkylation reaction, 272– 274 and partial oxidation of methane, 301, 303 Thioanisole (MPS) oxidation, 97t Ti-beta and alcohol oxidation, 100 diffusional constraints, 62, 63t relative selectivity, 65 – 67 synthesis, 146 – 147, 153– 154 Ti-HMS synthesis, 147, 157 Ti-MCM-41, 67 catalytic activity, 69t catalytic selectivity and Ti content, 129f influence of silylation, 124 photocatalytic synthesis, 120 – 121 structure and activity, 128 synthesis, 147, 154 –156 Ti composition and textural characteristics, 68t and transesterification reactions, 110 Ti-MCM-48, 124 synthesis, 147, 157 Ti-MMM-1, 88 and H2O2 conversion, 97t IR spectroscopy, 96f Raman spectra, 95t synthesis, 146 Ti-MWW synthesis, 145 Ti-SBA-15, 93 and H2O2 conversion, 97t IR spectroscopy, 96f Raman spectra, 95t structural and textural parameters, 94t synthesis, 147 – 148, 158 UV-visible spectroscopy, 95t Ti-ZSM-48 synthesis, 145, 152 Ti4þ coordination number, 50

358

Index

distribution in TS-1, 53 Time on stream behavior of BEA, 265 behavior of CeY zeolite, 259f and catalyst temperature in partial oxidation of methane, 304f for CO formation in CO2 reforming of methane, 328f –329f for partial oxidation of methane, 303f Ti(O2H) activity, 128 –129 Titanium peroxo species, 8 Titanium silicate molecular sieves acid-catalyzed reactions, 105–116 active sites, 6 –7, 9 –33 CvN cleavage reactions, 105 catalytic properties, 56, 132 –136 commercial application, 7, 62 computational investigations, 49– 55 early investigations, 5– 6 epoxidation. see Epoxidation hydroxylations. see Hydroxylations neutron diffraction, 10 –11 NO decomposition, 121– 122 O – O bond cleavage, 137– 138 oxidation reactions. see Oxidation photodegradation of pollutants, 116– 120 silylation, 124 – 127 structure and activity, 127 – 128 synthesis, 143– 146 Ti composition and textural characteristics, 68t Titanium superoxo species, 8 Titanosilicate surface structures EPR spectroscopy, 22 –26 particle size, 12 photoluminescence spectroscopy, 15 Ti tetrahedral geometry, 9 UV-visible spectroscopy, 12– 14

vibrational spectroscopy, 18 – 22 X-ray absorption spectroscopy, 15– 18 X-ray diffraction, 10 TMP (trimethylpentane), 101, 234 isomerization pathways, 241f oligomerization, 248 Toluene oxidation, 89 – 90 Tomography, 212 –218 using HAADF-STEM, 223 Transesterification of cyclic carbonates, 109 of diethyl malonate, 113t of esters, 110 of ethylacetoacetate, 111 –112, 114t Triflic acid as alkylation catalyst, 271 and oligomerization, 255 Trimethylpentanes. see TMP (trimethylpentanes) TS-1 (titanium silicate-1) Brønsted acid sites, 26 –28 chemoselectivity, 7 coordination sphere expansion, 31– 32 diffusional constraints, 64 discovery, 5 DRUV data, 13t – 14f fingerprint features for Ti isomorphous substitution, 142t and oxidation of amines, 91t photocatalytic synthesis, 120– 121 Raman spectra, 20f relative selectivity, 65 –67 SEM photographs, 136f and transesterification reactions, 110 XANES spectrum, 17f TS-2 (titanium silicate-2) aliphatic compound hydroxylation, 87 epoxidation of alkenes, 70t oxidation of sulfides, 93t

Index

synthesis, 145, 151 –152 and toluene oxidation, 89 – 90 U Ultra-high resolution low-voltage field emission scanning electron microscopy (HRLVSEM), 195 UOP Alkad process, 283 UOP AlkyleneTM process, 285 –286 USY zeolite and alkene feed composition, 276 cracking, 249 and reaction temperature, 273 UV-visible spectroscopy and adsorbed water on TS-1, 32– 33f and Lewis acidity in TS-1, 31 – 32 oxo-titanium species, 34 – 35f and surface structure of TS-1, 12 – 14 for TiSBA-15, 95t V Vanadium oxide, 324 Vanadyl pyrophosphate (VO)2P2O7, 203 active sites, 207f electron diffraction, 205f glide shear plane defects, 208 –209 structure, 204f Vibrational spectroscopy, oxo-titanium species, 34– 39 VPO, 203 active sites, 207f electron diffraction, 205f solid-state heterogeneous catalytic oxidation processes, 208 –209 structure, 204f VS-2 n-haxane hydroxylation, 87 and toluene oxidation, 89 – 90 VTM-A, 266

359

W Wavelength dispersive X-ray spectroscopy (WDS), 191 Weak phase objects (WPO), 182 –183 Wet-ETEM, 210– 212 WS2 catalyst, 331 –332 X X-ray absorption spectroscopy, 39 – 41 X-ray elemental mapping, 192 XAFS (X-ray absorption fine structure), 176 XANES spectrum and adsorbed water on TS-1, 33f oxo-titanium species, 39– 41 Z Z-contrast imaging, 194 Zeolite X, 262– 263 Zeolite Y, 263– 264 and CO2 reforming of methane, 332 Zeolites acid strength and catalyst role in alkylation, 258 advantages in alkylation, 233 and alkene feed composition, 276 and alkylation initiation, 237, 239 analysis by EM, 221 Brønsted acid sites, 256– 260 characteristics as molecular sieve, 255 – 256 electron microscopic image, 194f enthalpy of adsorption, 257f HRTEM analysis of L-type, 185 – 186 and hydride transfer, 243– 244 intergrowth analysis by HRTEM, 188 – 190 as isomerization catalysts, 240

360

Lewis acid sites, 260 – 261 and reaction temperature, 273 and self-alkylation, 250 structure and types, 264– 267 vibrational spectroscopy, 18 Ziegler-Natta catalysts, 200 Zirconia-based oxygen transport, 306 Zirconia-containing catalysts, 323

Index

Zirconia-supported nickel catalyst, 330– 331 ZSM-11, 264 ZSM-12, 264 ZSM-20, and rare earth exchange, 263– 264 ZSM-5, 264– 265 ZSM-5 zeolite, 332