advances in catalysis, volume 48
TRANSCRIPT
ADVANCES IN CATALYSIS
VOLUME 48
Advisory Board
M. CHE
Paris, FranceD.D. ELEY
Nottingham, EnglandG. ERTL
Berlin/Dahlem, Germany
V.B. KAZANSKY
Moscow, RussiaW.M.H. SACHTLER
Evanston, Illinois, USAR.A. VAN SANTEN
Eindhoven, The Netherlands
K. TAMARU
Tokyo, JapanJ.M. THOMAS
London/Cambridge, England
H. TOPSØE
Lyngby, DenmarkP.B. WEISZ
State College, Pennsylvania, USA
ADVANCES INCATALYSIS
VOLUME 48
Edited by
BRUCE C. GATES
University of California
Davis, California, USA
HELMUT KNOZINGER
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.else-
vier.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
W1 The paper used in this publication meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper). Printed in
The Netherlands.
Contents
CONTRIBUTORS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii
PREFACE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv
ROBERT L. BURWELL, Jr. (1912–2003) . . . . . . . . . . . . . . . . . . . . . . . . . xix
Active Sites and Reactive Intermediates in Titanium Silicate
Molecular Sieves
P. Ratnasamy, D. Srinivas and H. Knozinger
I. Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
II. Active Sites. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
II.A. State and Framework Coordination of Ti . . . . . . . . . . . . . . . . . 9
II.A.1. Diffraction Techniques . . . . . . . . . . . . . . . . . . . . . . . . 10
II.A.2. Influence of Particle Size . . . . . . . . . . . . . . . . . . . . . . . 12
II.A.3. UV–Visible Spectroscopy . . . . . . . . . . . . . . . . . . . . . . 12
II.A.4. Photoluminescence Spectroscopy . . . . . . . . . . . . . . . . . 15
II.A.5. X-Ray Absorption Spectroscopy . . . . . . . . . . . . . . . . . 15
II.A.6. Vibrational Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . 18
II.A.7. EPR Spectroscopy. . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
II.B. Surface Acidity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
II.B.1. Brønsted Acid Sites . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
II.B.2. Lewis Acid Sites and Expansion of
Coordination Sphere . . . . . . . . . . . . . . . . . . . . . . . . . . 28
III. Oxo-Titanium Species and Reactive Intermediates . . . . . . . . . . . . . . 33
III.A. UV–Visible Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
III.B. Vibrational Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
III.C. X-Ray Absorption Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . 39
III.D. Cyclic Voltametry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
III.E. EPR Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
IV. Computational Investigations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
V. Catalytic Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
V.A. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
V.B. Reactions Using H2O2 as Oxidant . . . . . . . . . . . . . . . . . . . . . . 56
V.B.1. General Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
v
V.B.2. H2O2-Catalyzed Reactions in the Homogeneous
Phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
V.C. Epoxidation on Titanium Silicate Molecular Sieves . . . . . . . . . 60
V.C.1. General Features of Epoxidations. . . . . . . . . . . . . . . . . 60
V.C.2. Yields and Stereospecificities. . . . . . . . . . . . . . . . . . . . 62
V.C.3. Diffusional Constraints . . . . . . . . . . . . . . . . . . . . . . . . 62
V.C.4. Influence of Ti-Silicate Structure . . . . . . . . . . . . . . . . . 65
V.C.5. Epoxidation Catalyzed by Mesoporous Titanium
Silicates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
V.C.6. Influence of Alkene Structure . . . . . . . . . . . . . . . . . . . 70
V.C.7. Dialkenes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
V.C.8. Epoxidation in the Presence of Other Oxidizable
Functional Groups. . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
V.C.9. Hydroxyl-Assisted Epoxidation . . . . . . . . . . . . . . . . . . 72
V.C.10. Diastereoselectivity in Epoxidations . . . . . . . . . . . . . . 74
V.C.11. Side Reactions During Epoxidation . . . . . . . . . . . . . . 75
V.C.12. Influence of pH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
V.C.13. Epoxidation with Alkyl Hydroperoxides . . . . . . . . . . . 80
V.C.14. Epoxidation of Alkenes Containing Carbonyl Groups . . 81
V.C.15. Epoxidation Using Urea–H2O2 Adduct . . . . . . . . . . . 82
V.C.16. Epoxidation Using Dioxygen . . . . . . . . . . . . . . . . . . . 83
V.D. Hydroxylations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
V.D.1. General Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
V.D.2. Hydroxylation of Aliphatic Compounds . . . . . . . . . . . . 85
V.D.3. Hydroxylation of Aromatic Compounds . . . . . . . . . . . . 89
V.E. Oxidation of Nitrogen-Containing Compounds . . . . . . . . . . . . . 90
V.F. Oxidation of Sulfur-Containing Compounds . . . . . . . . . . . . . . . 93
V.G. Oxidation of Oxygen-Containing Compounds . . . . . . . . . . . . 100
V.G.1. Alcohols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
V.G.2. Ethers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
V.G.3. Phenols. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
V.G.4. Ketones, the Baeyer–Villiger Oxidation. . . . . . . . . . . 102
V.H. CyN Cleavage Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
V.I. Acid-Catalyzed Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
V.I.1. Beckmann Rearrangement . . . . . . . . . . . . . . . . . . . . . . 106
V.I.2. Synthesis of Polycarbonate Precursors. . . . . . . . . . . . . . 106
V.I.3. Transesterification of Esters . . . . . . . . . . . . . . . . . . . . . 110
V.I.4. Carbon–Carbon Bond Formation Reactions . . . . . . . . . 110
Contentsvi
V.I.5. Formation of Pinacols . . . . . . . . . . . . . . . . . . . . . . . . . 114
V.I.6. Oxidative Dehydrogenation . . . . . . . . . . . . . . . . . . . . . 115
V.J. Photocatalysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
V.J.1. Photocatalytic Degradation of Pollutants . . . . . . . . . . . . 116
V.J.2. Photocatalytic Synthesis . . . . . . . . . . . . . . . . . . . . . . . . 120
V.J.3. deNOx Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
V.K. Influence of Solvents. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
V.L. Influence of Silylation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
VI. Structure-Activity Correlations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
VI.A. Structure of Titanium Species and Activity . . . . . . . . . . . . . . 127
VI.B. Titanium-Oxo Species and Activity . . . . . . . . . . . . . . . . . . . 128
VII. O–O Bond Cleavage and Product Selectivity. . . . . . . . . . . . . . . . . 137
VII.A. General. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
VII.B. Epoxidation of Alkenes . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
VIII. Conclusions and Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
Appendix A. Fingerprint Features for Ti Isomorphous Substitution
in TS-1 Titanosilicates . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
Appendix B. Characteristics of the Oxo-Titanium Species Generated
on TS-1 on Contact with Aqueous H2O2 . . . . . . . . . . . . . . 143
Appendix C. Synthesis of Titanium Silicate Molecular Sieves . . . . . . . . 143
C.1. TS-1, TS-2, Ti-ZSM-48, Ti-MWW, and Ti-MMM-1. . 144
C.2. Ti-Beta Zeolite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
C.3. Ti-Containing HMS, MCM-41, and MCM-48. . . . . . . 147
C.4. Ti-SBA-15 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
C.5. Ti-TUD-1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
Electron Microscopy and the Materials Chemistry of Solid Catalysts
John Meurig Thomas and Pratibha L. Gai
I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
II. Electron Microscopy (EM) Methods . . . . . . . . . . . . . . . . . . . . . . . . 176
II.A. Electron Microscopy in Catalysis . . . . . . . . . . . . . . . . . . . . . . 177
II.B. Imaging in the Electron Microscope . . . . . . . . . . . . . . . . . . . . 178
II.C. TEM Imaging Method Using Diffraction Contrast. . . . . . . . . . 179
II.D. Theoretical Procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
III. High-Resolution Transmission Electron Microscopy . . . . . . . . . . . . 181
Contents vii
III.A. Conditions Required for Optimizing HRTEM Images . . . . . 182
III.B. Development of HRTEM. . . . . . . . . . . . . . . . . . . . . . . . . . 184
III.C. Elucidation of the Structures of Meso- and Microporous
Catalysts by HRTEM. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
III.C.1. L-Type Zeolite Catalysts . . . . . . . . . . . . . . . . . . . . 185
III.C.2. Metal-Substituted Aluminum Phosphate
(MAPO-36) Microporous Catalysts . . . . . . . . . . . . 186
III.C.3. High-Silica Microporous SSZ-48 Catalysts . . . . . . 187
III.C.4. Intergrowths in Zeolite Catalysts: Coherent,
Recurrent, and Random. . . . . . . . . . . . . . . . . . . . . 188
IV. Chemical Composition Analysis with the Analytical
Electron Microscope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
V. Scanning Transmission Electron Microscopy . . . . . . . . . . . . . . . . 193
VI. Recent Advances in Ultra-High Resolution, Low-Voltage
Field Emission Scanning Electron Microscopy and
Extreme FESEM in Catalysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
VII. Cathodoluminescence Imaging for Elucidation of
Electronic Structures of Catalysts. . . . . . . . . . . . . . . . . . . . . . . . . 195
VIII. Recent Advances in In Situ Atomic Resolution-Environmental
Transmission Electron Microscopy (ETEM) Under Controlled
Reaction Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
VIII.A. In Situ Investigations of Gas–Solid Reactions and
Active Sites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
VIII.B. Illusrative Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
VIII.B.1. In Situ Gas–Catalyst Reactions
at the Atomic Level . . . . . . . . . . . . . . . . . . . . . 201
VIII.B.2. Atomic-Resolution ETEM of Butane
Oxidation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
VIII.B.3. Atomic-Resolution ETEM of Nanorods . . . . . . 210
VIII.C. Advances in In Situ Wet-Electron Microscopy
Technique (Wet-ETEM) for Probing Solid Catalysts
Under Liquid Environments . . . . . . . . . . . . . . . . . . . . . . 210
IX. Environmental Scanning Electron Microscopy . . . . . . . . . . . . . . . 212
X. Electron Tomography: Three-Dimensional Electron
Microscopy Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
X.A. The Topography and Location of Nanoparticles
in Supported Catalysts; BSE and HAADF . . . . . . . . . . . . . . 213
X.B. Pinpointing the Location of Nanoparticles Supported
on Nanoporous Solids. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
Contentsviii
XI. Energy Filtered Transmission Electron Microscopy
and Elemental Maps of Solid Catalysts Using EFTEM . . . . . . . . . 218
XII. Other Significant Trends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
XIII. Critical Evaluations of the Methods and Challenges . . . . . . . . . . . 220
XIV. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
Chemistry and Technology of Isobutane/Alkene Alkylation Catalyzed
by Liquid and Solid Acids
Andreas Feller and Johannes A. Lercher
I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
II. Alkylation Mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
II.A. Overall Product Distribution . . . . . . . . . . . . . . . . . . . . . . . . . 234
II.B. Initiation Steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
II.C. Alkene Addition and Isomerization . . . . . . . . . . . . . . . . . . . . 239
II.D. Hydride Transfer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
II.E. Oligomerization and Cracking . . . . . . . . . . . . . . . . . . . . . . . . 247
II.F. Self-Alkylation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
II.G. Product and Acid Degradation . . . . . . . . . . . . . . . . . . . . . . . . 251
II.H. Pathways to Allylic and Cyclic Compounds . . . . . . . . . . . . . . 251
II.I. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252
III. Physical–Chemical Phenomena Influencing the Reaction. . . . . . . . . 252
III.A. Properties of Liquid Acid Alkylation Catalysts . . . . . . . . . . . 253
III.B. Properties of Zeolitic Alkylation Catalysts . . . . . . . . . . . . . . 255
III.B.1. Adsorption and Diffusion of Hydrocarbons . . . . . . . . 255
III.B.2. Brønsted Acid Sites . . . . . . . . . . . . . . . . . . . . . . . . . 256
III.B.3. Lewis Acid Sites and Extra-Framework Aluminum . . 260
III.B.4. Silicon/Aluminum Ratio . . . . . . . . . . . . . . . . . . . . . 261
III.B.5. Metal Ions in Ion-Exchange Positions. . . . . . . . . . . . 263
III.B.6. Structure Types of Zeolites . . . . . . . . . . . . . . . . . . . 264
III.C. Other Solid Acids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
III.C.1. Sulfated Zirconia and Related Materials . . . . . . . . . . 267
III.C.2. Heteropolyacids. . . . . . . . . . . . . . . . . . . . . . . . . . . . 268
III.C.3. Acidic Organic Polymers . . . . . . . . . . . . . . . . . . . . . 269
III.C.4. Supported Metal Halides . . . . . . . . . . . . . . . . . . . . . 270
III.D. The Influence of Process Conditions . . . . . . . . . . . . . . . . . . . 271
III.D.1. Reaction Temperature . . . . . . . . . . . . . . . . . . . . . . . 272
Contents ix
III.D.2. Alkane/Alkene Ratio and Alkene Space Velocity . . . 274
III.D.3. Alkene Feed Composition . . . . . . . . . . . . . . . . . . . . 276
IV. Industrial Processes and Process Developments. . . . . . . . . . . . . . . . 278
IV.A. Liquid Acid-Catalyzed Processes . . . . . . . . . . . . . . . . . . . . . 278
IV.A.1. Sulfuric Acid-Catalyzed Processes . . . . . . . . . . . . . . 278
IV.A.2. Hydrofluoric Acid-Catalyzed Processes . . . . . . . . . . 281
IV.B. Solid Acid-Catalyzed Processes . . . . . . . . . . . . . . . . . . . . . . 283
IV.B.1. UOP Alkylenee Process . . . . . . . . . . . . . . . . . . . . . 285
IV.B.2. Akzo Nobel/ABB Lummus AlkyCleane Process . . . 286
IV.B.3. LURGI EUROFUELw Process. . . . . . . . . . . . . . . . . 286
IV.B.4. Haldor Topsøe FBAe Process . . . . . . . . . . . . . . . . . 287
V. Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 289
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 289
Catalytic Conversion of Methane to Synthesis Gas by
Partial Oxidation and CO2 Reforming
Yun Hang Hu and Eli Ruckenstein
I. Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 298
II. Partial Oxidation of Methane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301
II.A. Hot Spots in Catalyst Beds . . . . . . . . . . . . . . . . . . . . . . . . . . 301
II.B. Minimizing O2 Purification Costs. . . . . . . . . . . . . . . . . . . . . . 306
II.C. Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312
II.D. Reaction Pathways . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 314
II.D.1. Changes in Catalyst During Reaction . . . . . . . . . . . . . 315
II.D.2. Which is the Primary Product, CO or CO2? . . . . . . . . 316
II.D.3. CHx Species and Rate-Determining Steps . . . . . . . . . . 318
II.D.4. Comparison of Reactions on Reduced and
Unreduced Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . 320
III. CO2 Reforming of Methane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 321
III.A. Carbon Formation on Metal Surfaces . . . . . . . . . . . . . . . . . . 321
III.B. Critical Issues Related to Carbon Deposition . . . . . . . . . . . . . 322
III.C. Supported Noble Metal Catalysts . . . . . . . . . . . . . . . . . . . . . 323
III.D. Non-Noble Metal Supported Catalysts . . . . . . . . . . . . . . . . . 324
III.D.1. Ni/Al2O3 Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . 325
III.D.2. Ni/SiO2 Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . 327
III.D.3. Ni/La2O3 Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . 328
III.D.4. Ni/ZrO2 Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . 330
III.D.5. Other Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . 331
Contentsx
III.E. MgO-Containing Solid-Solution Catalysts . . . . . . . . . . . . . . . 332
III.E.1. Characteristics of MgO-Containing Solid-Solution
Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332
III.E.2. Highly Effective MgO-Containing Solid-Solution
Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333
IV. Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 338
INDEX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347
Contents xi
This Page Intentionally Left Blank
Contributors
Numbers in parentheses indicate the pages on which the authors’ contributions begin.
ANDREAS FELLER, Institut fur Technische Chemie, Technische Universitat
Munchen, 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. KNOZINGER, Department Chemie-Physikalische Chemie, Universitat
Munchen, Butenandt Strasse, 5-13, Haus E, D-81377 Munchen, Germany (1)
JOHANNES A. LERCHER, Institut fur Technische Chemie, Technische Universitat
Munchen, 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 isobutane-
alkene alkylation and partial oxidation and CO2 reforming of methane to
synthesis gas.
Ratnasamy, Srinivas, and Knozinger 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 comp-
lements 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 nitrogen-
and oxygen-containing organic compounds, and acid-catalyzed and photocata-
lytic 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 high-
resolution 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
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
Prefacexvi
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
Preface xvii
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
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)xx
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
Robert L. Burwell, Jr. (1912–2003) xxi
Active Sites and Reactive
Intermediates in Titanium Silicate
Molecular Sieves
P. RATNASAMY* and D. SRINIVAS
National Chemical Laboratory, Pune 411008, India
and
H. KNOZINGER*
Department Chemie-Physikalische Chemie, Universitat Munchen, Butenandt Strasse,
5-13, Haus E, D-81377 Munchen, Germany
I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
II. Active Sites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
II.A. State and Framework Coordination of Ti . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
II.A.1. Diffraction Techniques. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
II.A.1.1. X-Ray Diffraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
II.A.1.2. Neutron Diffraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
II.A.2. Influence of Particle Size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
II.A.3. UV–Visible Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
II.A.4. Photoluminescence Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
II.A.5. X-Ray Absorption Spectroscopy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
II.A.6. Vibrational Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
II.A.7. EPR Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
II.B. Surface Acidity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
II.B.1. Brønsted Acid Sites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
II.B.2. Lewis Acid Sites and Expansion of Coordination Sphere . . . . . . . . . . . . . . 28
III. Oxo-Titanium Species and Reactive Intermediates . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
III.A. UV–Visible Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
III.B. Vibrational Spectroscopy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
III.C. X-Ray Absorption Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
III.D. Cyclic Voltametry. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
III.E. EPR Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
IV. Computational Investigations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
ADVANCES IN CATALYSIS, VOLUME 48 Copyright q 2004 Elsevier Inc.ISSN: 0360-0564 DOI 10.1016/S0360-0564(04)48001-8 All rights reserved
*Corresponding author.
E-mail address: [email protected] (P. Ratnasamy); [email protected]
(H. Knozinger).
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169
V. Catalytic Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
V.A. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
V.B. Reactions Using H2O2 as Oxidant . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
V.B.1. General Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
V.B.2. H2O2-Catalyzed Reactions in the Homogeneous Phase . . . . . . . . . . . . . . 58
V.C. Epoxidation on Titanium Silicate Molecular Sieves. . . . . . . . . . . . . . . . . . . . . . . 60
V.C.1. General Features of Epoxidations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
V.C.2. Yields and Stereospecificities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
V.C.3. Diffusional Constraints. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
V.C.4. Influence of Ti-Silicate Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
V.C.5. Epoxidation Catalyzed by Mesoporous Titanium Silicates . . . . . . . . . . . . 67
V.C.6. Influence of Alkene Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
V.C.7. Dialkenes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
V.C.8. Epoxidation in the Presence of Other Oxidizable Functional Groups . . . . 72
V.C.8.1. Alkenes and Alcohol Functions . . . . . . . . . . . . . . . . . . . . . . . . 72
V.C.8.2. Alkenes and Alkanes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
V.C.9. Hydroxyl-Assisted Epoxidation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
V.C.10. Diastereoselectivity in Epoxidations . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
V.C.11. Side Reactions During Epoxidation. . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
V.C.12. Influence of pH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
V.C.13. Epoxidation with Alkyl Hydroperoxides . . . . . . . . . . . . . . . . . . . . . . . . 80
V.C.14. Epoxidation of Alkenes Containing Carbonyl Groups . . . . . . . . . . . . . . 81
V.C.15. Epoxidation Using Urea–H2O2 Adduct . . . . . . . . . . . . . . . . . . . . . . . . . 82
V.C.16. Epoxidation Using Dioxygen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
V.D. Hydroxylations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
V.D.1. General Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
V.D.2. Hydroxylation of Aliphatic Compounds . . . . . . . . . . . . . . . . . . . . . . . . . 85
V.D.3. Hydroxylation of Aromatic Compounds . . . . . . . . . . . . . . . . . . . . . . . . . 89
V.E. Oxidation of Nitrogen-Containing Compounds . . . . . . . . . . . . . . . . . . . . . . . . . . 90
V.F. Oxidation of Sulfur-Containing Compounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
V.G. Oxidation of Oxygen-Containing Compounds . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
V.G.1. Alcohols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
V.G.2. Ethers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
V.G.3. Phenols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
V.G.4. Ketones, the Baeyer–Villiger Oxidation . . . . . . . . . . . . . . . . . . . . . . . . . 102
V.H. CyN Cleavage Reactions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
V.I. Acid-Catalyzed Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
V.I.1. Beckmann Rearrangement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
V.I.2. Synthesis of Polycarbonate Precursors. . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
V.I.3. Transesterification of Esters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
V.I.4. Carbon–Carbon Bond Formation Reactions. . . . . . . . . . . . . . . . . . . . . . . . 110
V.I.5. Formation of Pinacols. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
V.I.6. Oxidative Dehydrogenation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
V.J. Photocatalysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
V.J.1. Photocatalytic Degradation of Pollutants . . . . . . . . . . . . . . . . . . . . . . . . . . 116
V.J.2. Photocatalytic Synthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
V.J.3. deNOx Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
V.K. Influence of Solvents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
V.L. Influence of Silylation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
VI. Structure-Activity Correlations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–1692
VI.A. Structure of Titanium Species and Activity . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
VI.B. Titanium-Oxo Species and Activity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
VII. O–O Bond Cleavage and Product Selectivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
VII.A. General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
VII.B. Epoxidation of Alkenes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
VIII. Conclusions and Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
Appendix A. Fingerprint Features for Ti Isomorphous Substitution
in TS-1 Titanosilicates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
Appendix B. Characteristics of the Oxo-Titanium Species Generated on TS-1
on Contact with Aqueous H2O2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
Appendix C. Synthesis of Titanium Silicate Molecular Sieves . . . . . . . . . . . . . . . . . . . . . . . 143
C.1. TS-1, TS-2, Ti-ZSM-48, Ti-MWW, and Ti-MMM-1. . . . . . . . . . . . . . . . . 144
C.2. Ti-Beta Zeolite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
C.3. Ti-Containing HMS, MCM-41, and MCM-48. . . . . . . . . . . . . . . . . . . . . . 147
C.4. Ti-SBA-15 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
C.5. Ti-TUD-1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 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(O2z2), 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 microporous titanium silicate molecular sieves with MFI
and MEL structures, respectively
Ti-beta (Ti-b) large-pore titanium silicate with BEA structure
Ti-MCM-41 and Ti-MCM-48 titanium-containing Mobil composite materials/mesopor-
ous composite materials of type 41 (hexagonal array of
pores) and 48 (cubic array of pores)
Ti " MCM-41 titanium grafted on MCM-41
Ti-HMS titanium containing hexagonal mesoporous silica material
Ti-ZSM-48 titanium containing one-dimensional 10-ring zeolite
Ti-ZSM-12 Mobil Corporation’s one-dimensional large-pore
(12-membered ring) zeolite
ZSM-5 Zeolite Socony Mobil constructed from five-membered-
ring building units
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 3
Ti-SBA-15 mesoporous, titanium-containing silica self-assembly-15
(with uniform, hexagonal, tubular channels) synthesized by
using a triblock organic copolymer as a template
Ti-MMM titanium-containing microporous mesoporous material
Ti-MWW titanium silicate with MWW structure
Ti-TUD-1 three-dimensionally randomly connected mesoporous silica
ETS-4 and ETS-10 Engelhard Corporation titanium silicate molecular sieves
MST amorphous mesoporous silica-titania
VS-2 vanadium-containing silicalite with MEL topology
Sil silicalite
XRD X-ray diffraction
UV–visible ultraviolet–visible
DRUV diffuse reflectance ultraviolet
FTIR Fourier transform infrared
NIR near infrared
EXAFS extended X-ray absorption fine structure
XANES X-ray absorption near-edge structure
XAS X-ray absorption spectroscopy
XAFS X-ray absorption fine structure
EPR electron paramagnetic resonance (also known as electron
spin resonance (ESR))
NMR nuclear magnetic resonance
LMCT ligand to metal charge transfer
DFT density functional theory
HP aqueous H2O2
TBHP tert-butyl hydroperoxide
UHP urea–H2O2 (1:1) adduct
FCC fluidized catalytic cracking
TEOS tetraethyl orthosilicate
TEOT tetraethyl orthotitanate
TBOT tetrabutyl orthotitanate
TMAOH tetramethylammonium hydroxide
TEAOH tetraethylammonium hydroxide
TPAOH tetrapropylammonium hydroxide
TPABr tetrapropylammonium bromide
DDA dodecylamine
TEA triethanolamine
CTABr cetyltrimethylammonium bromide
DH0 gas-phase dissociation enthalpy
DE energy required for gas-phase heterolytic cleavage
TOF turnover frequency (moles of reactant converted per mole
of active catalyst species per unit time)
SEM scanning electron microscopy
TPD temperature programmed desorption
n frequency
Dn shift in peak position in frequency units
l spin–orbit coupling constant
D energy gap between pxg and p
yg orbitals of oxygen
E energy separation between 3sg and 1pgx orbitals of oxygen
gxx; gyy; and gzz principal g-values
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–1694
pKa negative logarithm of acidity constant
rip ion pair separation
DEipðsolventÞ energy of heterolytic cleavage in a solvent
DEsolv solvation energy
1 dielectric constant
m dipole moment of the solvent
a radius of a spherical cavity formed by solvent molecules
surrounding an ion pair
e 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), Ti-
MWW (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 predomi-
nant 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
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 5
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 mid-
1990s, 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 tech-
niques 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. Knozinger / Adv. Catal. 48 (2004) 1–1696
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
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 7
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 con-
centrations 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. Knozinger / Adv. Catal. 48 (2004) 1–1698
(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 inter-
mediates 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 frame-
work positions in the silicate structure as its ionic radius is too large. Titanium
compounds with tetrahedral geometry are scarce, as highly stable hexacoordi-
nated 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.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 9
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 þ4 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. Knozinger / Adv. Catal. 48 (2004) 1–16910
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 crystallographi-
cally 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).
Fig. 1. The structure of orthorhombic form of silicalite-1 (MFI type) showing the 12 crystallo-
graphically distinct T sites. The oxygen atoms are omitted for clarity [Reprinted from Henry et al. (85)
with permission. Copyright (2001) American Chemical Society].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 11
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 con-
sequence, 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. Knozinger / Adv. Catal. 48 (2004) 1–16912
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)
206 (85) 228 (8) 258 (6) 293 (1) Nil
TS-1
(Sample 2)
203 (72) 228 (10) 255 (8) 288 (5) 328 (5)
TS-1
(Sample 3)
206 (78) 229 (11) 260 (7) 293 (4) Nil
TS-2 201 (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.
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].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 13
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.
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)].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–16914
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 discrim-
inate 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 photo-
luminescence, 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 correspond-
ing 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
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 15
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
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].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–16916
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). Well-
prepared, 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/
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].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 17
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 character-
istic 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. Knozinger / Adv. Catal. 48 (2004) 1–16918
Boccuti et al. (91) interpreted the 960 cm21 band on the basis of a consider-
ation 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 wave-
numbers 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 photo-
luminescence 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
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 19
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 dehy-
drated 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. Knozinger / Adv. Catal. 48 (2004) 1–16920
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 vibra-
tional excitations. Quantum chemical calculations (41) of the vibrational freque-
ncies 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].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 21
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
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].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–16922
temperatures, the Ti ions are reduced from a diamagnetic þ 4 (3d0) to a param-
agnetic, 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) investi-
gated 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 investi-
gation 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
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].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 23
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-MCM-
41 . 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
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)].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–16924
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; gyyÞ; 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 stabili-
zation 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
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)
Species gk g’ gzz gxx gyy
ETS-10 873 1.969 1.942
673 1.966 1.941
ETS-10(Cs) 673 1.869 1.944 1.959
ETS-10(Cs)-Pt 673 1.870 1.943 1.959
ETS-4 673 1.863 1.930
ETS-4-Pt 673 I 1.870 1.920
II 1.863 1.930
TS-1 873 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 1.894 1.938 1.974
Ti-MCM-41-Pt 873 I 1.906 1.958
II 1.894 1.938 1.974
Adapted from Bal et al. (130).
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 25
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 impreg-
nated 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
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)].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–16926
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 char-
acterized 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 capro-
lactam (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).
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 27
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 spectro-
scopy 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 inter-
action 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 acid-
catalyzed 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 out-
gassing 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. Knozinger / Adv. Catal. 48 (2004) 1–16928
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 char-
acterizes 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 wave-
number 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 coordina-
tion 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
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 29
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 hydrogen-
bonded 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,
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].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–16930
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
Fig. 13. FTIR spectra of pyridine adsorbed on dehydrated TS-1 and Ti-MCM-41 [from Srinivas
et al. (152)].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 31
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.
Fig. 14. Temperature programmed desorption of NH3 profiles of TS-1 and Ti-MCM-41 [from
Srinivas et al. (152)].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–16932
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 inter-
action 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.
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)].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 33
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
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, phase-
uncorrected Fourier transforms of the corresponding EXAFS spectrum [Reprinted from Ricchiardi
et al. (41) with permission. Copyright (2001) American Chemical Society].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–16934
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 O2z2 species have been detected on titanium-containing silicalites
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].
TABLE III
IR spectroscopy of peroxo and superoxo species
Compound Dioxygen species n(O–O) n(M–O)s,as Oxygen source Reference
(Pic)2TiO2HMPA O222 895 575, 615 H2O2 (155)
(OEP)TiO2 O222 895 595, 635 H2O2, O2 (155)
Ca12Al10Si4O35 O222 895 O2 (156)
Ca12Al10Si4O35 O2z2 1075 O2 (156)
Pic, pyridine-2-carboxylate; HMPA, hexamethylphosphoric triamide; OEP, octaethylporphyrin.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 35
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; end-
on 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 H216O2 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 H218O2 was loaded in
TS-1 instead of H216O2 (Fig. 18). No bands were observed at 837 or 3400 cm21
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 frame-
work 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).
Fig. 18. Infrared difference spectra before and after loading of H216O2 (curve a) and H2
18O2 (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. Knozinger / Adv. Catal. 48 (2004) 1–16936
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).
Scheme 1.
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].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 37
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, photo-
dissociation 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 restora-
tion 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
(NH4þ)3(TiF5O2)32 and TS-1/H2O2 systems, Bordiga et al. (42) concluded that
Fig. 20. Infrared difference spectra before and after 20 min irradiation (with 355 nm light
(45 mW cm22)) of aqueous H216O2 loaded TS-1 molecular sieves [Reprinted from Lin and Frei (133)
with permission. Copyright (2002) American Chemical Society].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–16938
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
of side-on (O2) attachment in the TiF5(O2) molecular unit of (NH4þ)3(TiF5O2)32,
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
distance of 2.01 A. Presumably, the formation of this complex is accompanied
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].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 39
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-
parable to those in the pristine surface structure (about 1.81 A), in the reactive
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
indicated that there are at least three Ti–O distances close to 1.83 A (a slight
expansion compared to the “free” surface), and two of the other three oxygen
distances were between 2.2 and 2.4 A. From among different models of the
titanium oxo species investigated, the authors concluded that the Ti-h2-OOR
Scheme 2.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–16940
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
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].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 41
a redox couple at 20.6 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, O2z2, 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 non-
framework 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(O2z2)) 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. Knozinger / Adv. Catal. 48 (2004) 1–16942
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(O2z2) 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(O2z2) 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
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)].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 43
H2O2 or non-aqueous urea–H2O2 adducts (Fig. 25). Bands I and II were
attributed to the charge transfer transitions associated with Ti(O2z2) 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(O2z2)) 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-TS-
1 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(O2z2)
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 “experi-
mental” in Fig. 26 shows that the intensity of the EPR signal varies linearly with
Fig. 24. EPR spectra of Ti(O2z2) and Ti3þ ions at 80 K. (a) Pd(2)-TS-2 þ H2O2; (b) Pt(0.015)-
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) TS-
1 þ H2 þ O2 (treated at 673 K). For clarity, spectra (c) and (d) are shown at four and five times the
actual gain. Spectral regions corresponding to Ti(O2z2) and Ti3þ ions are marked [from Shetti et al. (54)].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–16944
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 Species gzz gyy gxx D (cm21)b
TS-1 þ HP A 2.0264 2.0090 2.0023 11203
B 2.0238 2.0090 2.0023 12558
Ti-MCM-41 þ HP B 2.0244 2.0095 2.0031 12217
Pd(2)-TS-1 þ HP A0 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
TS-1 þ UHP 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
Pd(2)-TS-1 þ H2 þ O2 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
Pt(0.015)-TS-1 þ H2 þ O2 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
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 pg
x and pgy orbitals.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 45
Fig. 25. DRUV–visible spectra of TS-1, TS-1 þ H2O2, TS-1 þ urea–H2O2, and Pd(2)-TS-1þ
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. Knozinger / Adv. Catal. 48 (2004) 1–16946
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 para-
magnetic 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 experi-
mental 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(O2z2) 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 magneto-
meter (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:
H2O2 ! 2HOz ð2Þ
Ti–OOH þ HOz ! TiðOz22 Þ þ H2O ð3Þ
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 47
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.
Oz22 þ Oz2
2 ! O222 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,
Ti4þ(O222) to Ti3þ(O2
z2)), as proposed by Notari (33). Formation of Ti3þ
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(O2z2) 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(O2z2), 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
C from the defect sites. The free O2z2 radical, with a 2P ground state, has a
(1sg)2(1su)2(2sg)2(2su)2(3sg)2(1pu)4(1pg)3 electronic configuration. Interaction
with Ti removes the degeneracy of the HOMO pg into pgx and pg
y orbitals with
an energy gap of D: Neglecting the second-order terms, the g value expressions
(when l , Dp E) may be written as follows (159):
gzz ¼ ge þ 2l=D ð5Þ
gyy ¼ ge þ 2l=E; ð6Þ
andgxx ø ge; ð7Þ
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–16948
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–(O2z2) 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 p
yg orbitals ðDÞ
(Table III). The gzz values of various (Ti–(O2z2)) species decrease in the order
A . B . C. The D (O2z2) 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 investi-
gations 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
Scheme 3.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 49
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 (TiO2z2)
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 Ti-
peroxo 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 electron-
rich alkene double bond will preferentially interact with the most electrophilic
oxygen atom, which was identified to be the one closest to Ti in the hydro-
peroxo 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
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–16950
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
Cluster Ti–O distance
(A)
Ti–Si distance
(A)
Ti–O–Si (Ti–O–OH)
angle (8)
Eformation
(Calculated)a
(kJ/mol)
R-factor
(EXAFS)
Calculateda EXAFS parameter Calculateda EXAFS parameter Calculateda EXAFS parameter
Ti-h 2(OOH) 1.92 1.91 3.35 3.38 151.3 160 245 16.02
2.25P 2.20 3.32 3.30 145.5 148
1.83Si 1.83Si 3.28 3.21 143.0 139
1.80Si 1.83Si (81.9) (80)
1.80Si 1.83Si
2.26W 2.43
Ti-h 1(OOH) 2.24 2.20 3.31 3.28 145.3 144 2102 16.18
1.97P 1.97 3.34 3.38 146.8 152
1.81Si 1.83Si 3.34 3.39 151.5 163
1.84Si 1.83Si (117.3) (120)
1.81Si 1.83Si
2.35W 2.43
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.
Ratn
asamy
,D
.S
riniv
asan
dH
.K
nozin
ger
/A
dv
.C
atal.4
8(2
00
4)
1–
16
95
1
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. 27. Ti-peroxo species in TBHP/Ti " MCM-41 catalysts. All distances (DFT calculated values
and experimental parameters (in parentheses)) shown are in A [Reproduced from Barker et al. (48) by
permission of the PCCP Owner Societies].
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. Knozinger / Adv. Catal. 48 (2004) 1–16952
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, substi-
tution 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
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)].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 53
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
Fig. 30. Calculated initial and final states for the interaction of “ethene” with (a) theh1-, (b) theh2-,
and (c) h1-O2H2 Ti-peroxo intermediate [from Cora et al. (59)].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–16954
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 ortho-
rhombic 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].
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].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 55
(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 anti-
cipated. In fact, no experimental evidence exists for a purely Brønsted acid-
catalyzed 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 H2O
2as 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:
H2O2 þ HþO H2O 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
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].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–16956
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, DH0; for O–O homolytic cleavage in H2O2 is 205 kJ/mol
(171). The O–O bond is considerably weakened if H is replaced by electron-
donating 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)
DEipðsolventÞ 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 14:39ðð12 1Þ=ð21þ 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.
With a ¼ 3:5 �A (173), the solvation energy of a typical hydrocarbon solvent
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 dis-
tribution 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.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 57
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 Cu2þðOHÞ: 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 stereo-
specific, 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. Knozinger / Adv. Catal. 48 (2004) 1–16958
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 H2O2 decompostion
Co(III) þ e ! Co(II) þ1.82 Easy Fast
V(V) þ e ! V(IV) þ1.00 Moderate Moderate
Fe(III) þ e ! Fe(II) þ0.77 Moderate Moderate
Ti(IV) þ e ! Ti(III) 20.37 Difficult Difficult
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 59
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
Scheme 4.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–16960
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 hydro-
peroxide. 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–O-
ligands. 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–CH2Cl; ð17Þ
CH3 –CHOH–CH2Cl þ base ! CH3CHðOÞCH2 þ baseðHClÞ: ð18Þ
Scheme 5.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 61
The chlorhydrin route is also used in the manufacture of epichlorohydrin from
allyl chloride (187):
CH2yCH–CH2Cl þ HOCl ! CH2ðOÞCH–CH2Cl 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 amor-
phous 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). Metal-
loaded 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).
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–16962
TABLE VII
Diffusional constraints in selective oxidation of alkenes over Euro-TS-1 and Ti-Beta
Alkene Catalyst Reaction
time (h)
Turnover
(mol/mol Ti)
H2O2 Product selectivity
(%)
Glycol
ethers
Conversion (%) Selectivity (%) Epoxide Glycol
Hex-1-enea TS-1b 3 50 98 80 96 – 4
Ti-betac 3 12 80 80 12 8 80
Cyclohexenea TS-1 3 1 – – 100 – –
Ti-betad 3.5 14 80 83 – – 100
Dodec-1-enee TS-1 3.5 110 83 68 77 23 –
Ti-betaf 3.5 87 80 87 – 100 –
cyclododecenee TS-1 4 5 26 26 66 34 –
Ti-betaf 3.5 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.
Ratn
asamy
,D
.S
riniv
asan
dH
.K
nozin
ger
/A
dv
.C
atal.4
8(2
00
4)
1–
16
96
3
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 tetrapropyl-
ammonium 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 tetramethylortho-
silicate. 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 R0b Conversionc
(%)
Selectivityd
(%)
Yielde
(%)
1 0.2 Cubic 10.2 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. Knozinger / Adv. Catal. 48 (2004) 1–16964
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).
Fig. 34. Ratio of product concentrations [sum of epoxide and secondary products; (a) from oct-1-
ene 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].
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)
TS-1 323 ndb 76 (24) (193)
a Numbers in parentheses indicate the selectivities to the corresponding unsaturated carbonyl
compounds.b nd, no data available.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 65
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. Knozinger / Adv. Catal. 48 (2004) 1–16966
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 character-
istics 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.
TABLE XI
Epoxidation of oleic acid over Ti-beta prepared in fluoride (F) and alkali (OH) medium
Catalyst 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
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.
Scheme 6.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 67
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 titanium-
grafted 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
TABLE XII
Ti composition and textural characteristics of titanium silicates
Sample Method of preparation Ti loading,
calcined
samples
(wt%)
Specific
surface
area
(m2/g)
Total pore
volume
(mL/g)
Mean pore
diameter
(nm)
(A) Ti-MCM-41 Ordered, Ti-grafted, mesoporous
silica
1.88 861 0.53 2.4
(B) Ti-SiO2 Amorphous, Ti-grafted, porous
silica (Grace Davison 62)
1.75 303 1.10 12.8
(C) Ti-SiO2 Ti-grafted silica (Aerosil 380,
Degussa)
1.78 268 nd nd
(D) MST Amorphous, mesoporous titania-
silica (co-precipitation)
1.84 454 0.38 4.6
(E) TiO2-SiO2 Commercial, amorphous, porous
mixed oxide (Grace)
1.40 303 1.16 12.7
Adapted from Guidotti et al. (189); nd, not determined.
Scheme 7.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–16968
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
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)
5 40 (73)c 45 (84) 43 (83) 30 (83) 52 (75)
6 30 (90)c 33 (89) 32 (92) 19 (85) 25 (75)
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 (%).
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 69
(xSiO)2TiOTi(OO-t-Bu)4 species, prepared by the grafting route, in the epoxida-
tion 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).
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)].
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%)
92.0 81.2 72.0 56.4 28.8 40.2
Epoxide selectivity
(%)
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. Knozinger / Adv. Catal. 48 (2004) 1–16970
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 phenom-
enon 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 con-
centrations, 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.
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).a 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.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 71
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 titanosi-
licates 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-ene-
1-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 com-
peting 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, coordi-
nating 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
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.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–16972
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
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.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 73
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 erythro-
epoxides (Schemes 11 and 12). Control of the relative amounts of the two isomers
Scheme 8.
Scheme 9.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–16974
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 phenylacetalde-
hyde (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 reac-
tion in the presence of acetone, alcoholysis occurred to a great extent (45%) in the
presence of methanol solvent, producing mono glycol ethers.
Scheme 10.
Scheme 11.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 75
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 influ-
ence 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
Scheme 12.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–16976
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)
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 77
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
order: Liþ , NH4þ , Naþ , Kþ , Csþ and Mg2þ , Ca2þ , Ba2þ. Epoxide
selectivity followed the reverse order; Csþ exhibited 100% allyl alcohol con-
version 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 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).
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–16978
TABLE XIX
Epoxidation of allyl alcohol and allyl chloride—influence of pH
Run no. Catalyst Reactant pH TOF Olefin conversion
(mol%)
H2O2
efficiency
Epoxide selectivity
(mol%)
Initial/ before
H2O2 addition
After H2O2
addition
At the end
of the reaction
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 92.8
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 TS-1 in runs 3
and 4 reused
AA 7.8 5.5 5.7 14.8 74.4 97 96.8
8 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
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 97 73.8
13 TS-1 AC 7.0 3.8 3.8 19.8 100 100 79.7
14 TS-1 AC 8.0 4.3 4.2 19.8 100 100 81.6
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.
Ratn
asamy
,D
.S
riniv
asan
dH
.K
nozin
ger
/A
dv
.C
atal.4
8(2
00
4)
1–
16
97
9
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.
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 NH4þ 9.7 48.6 100
3 Naþ 17.0 85.8 91.7
4 Kþ 18.7 94.4 79.1
5 Csþ 19.8 100 76.7
6 Mg2þ 12.5 63.0 100
7 Ca2þ 18.8 94.7 88.5
8 Ba2þ 18.6 94.1 75.0
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; H2O2 (50%), 0.9 mL; H2O2/
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.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–16980
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 con-
jugated with the alkene double bond, retard the epoxidation as the delocalization
TABLE XXII
Epoxidation of allylalcohols with titanosilicates and H2O2
Reactant Catalyst T (K) Conversion
(%)
Epoxide
selectivitya (%)
Productivityb
(mmol/g/h)
Ti-beta 343 42 89 (11) 28
TS-1 323 nd 76 (24) 19
Ti-beta 343 nd .90 3.9c
TS-1 333 nd 96 21
Ti-beta 343 nd .90 2.5c
TS-1 323 nd 100 3
Ti-beta 343 nd 85 (11) 8.5c
TS-1 323 nd 90 (10) 3.3
Ti-beta 343 nd 96 25
TS-1 323 nd 82 (18) 9
Ti-beta 343 nd 74 (26) 12
TS-1 303 54 100 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.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 81
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 Catalyst Oxidanta Conversion
(mol%)
TONb Product distribution (mol%)a
EP PAD BD Diols
Styrene TS-1 HP 56 13.4 5 44 29 22
U þ HP 65 15.6 81 8 7 4
UHP 71 17.0 87 5 7 1
TS-2 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
Allylbenzene TS-1 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, 3-
phenyl-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. Knozinger / Adv. Catal. 48 (2004) 1–16982
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
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 83
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
The second type is the hydroxylation by a triplet oxygen atom (3O) 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 CH3z 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 RCOz
3 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. Knozinger / Adv. Catal. 48 (2004) 1–16984
the reduction of H2O2:
H2O ! H2Oþ ! H3Oþ þ HOz ð25Þ
H2O2 þ Fe2þ ! HO2 þ HOz þ Fe3þ ð26Þ
H2O2 þ Ti3þ ! HO2 þ HOz þ Ti4þ ð27Þ
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 4-
position. 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 hydroxy-
lation of an equimolar mixture of n-hexane and another alkane (alkane II)
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.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 85
TABLE XXV
Comparative activity of Ti-beta and Euro-TS-1 for selective oxidation of different alkanes
Alkane Catalysta Turnover
(mol/mol Ti)
H2O2 Product
selectivity (%)
Conv. (%) Sel. (%)b -ol -one
n-Hexane TS-1 48.5 77 100 91.5 8.5
Ti-beta 0.5 11 32 55.0 45.0
3-Methylpentane TS-1 0.7 6 19 88.9 11.1
Ti-beta 0.8 17 29 84.8 15.2
Cyclohexane TS-1 –b – – – –
Ti-beta 2.3 22 51 98.9 1.1
Methylcyclohexane TS-1 –b – – – –
Ti-beta 5.2 29 88 92.8 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
conversion
n-Hexane Alkane II
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. Knozinger / Adv. Catal. 48 (2004) 1–16986
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 hydroxy-
lation 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
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 87
(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
Scheme 14.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–16988
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
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
n-Octane
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
Ti-MCM-41 2.9 21.5 52.7 25.8 0.41
Adapted from Poladi and Landry (223).a Conversion ¼ (moles of alkane converted/total moles of alkane) £ 100.b Includes diols and diones.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 89
ethyl phenols (40%), acetophenone (56%), and 2-phenyl ethanol (4%). Mono-
substituted 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
TABLE XXVIII
Hydroxylation of aromatics over TS-2 and VS-2 molecular sieves
Benzene Toluene
TS-2 VS-2 TS-2 VS-2
Conversion (mol%) 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.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–16990
(both aliphatic and aromatic) are not reactive in the TS-1/H2O2 system; nitro-
benzene, 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 H2O2 ! NH2OH þ H2O ð28Þ
Scheme 15.
TABLE XXIX
Oxidation of primary amines catalyzed by TS-1
Amine Solvent Conversion Oxime selectivity H2O2 efficiency
CH3NH2 CH3OH 40 88 90
CH3NH2 CH3OHa 3 0 0
n-C3H7NH2 CH3OH 32 73 86
i-C3H7NH2 CH3OH 38 77 88
i-C3H7NH2 t-BuOHb 29 74 85
i-C3H7NH2 t-BuOHc 31 84 90
C6H11NH2 CH3OH 3 33 8
C6H11NH2 t-BuOH 3 32 8
C6H5CH2NH2 CH3OH 20 82 55
Adapted from Reddy and Jacobs (228).a Reaction without catalyst.b t-BuOH, tert-butyl alcohol.c Reaction over TS-2.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 91
R1R2CH–NH2 ! R1R2CH–NHOH ! R1R2CH–NO ! R1R2CyNOH ð29Þ
R1R2CH–NHR3 ! R1R2CyNðOÞR3 ð30Þ
R3N ! R3NO ð31Þ
C6H5NH2 ! C6H5NðOÞyN–C6H5 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.
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.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–16992
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 oxida-
tion of the primary product (sulfoxide) becomes more competitive, leading to
increased formation of the corresponding sulfone (Table XXXI):
R2S ! R2SO ! R2SðOÞ2 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).
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 93
TABLE XXXII
Structural and textural parameters of Ti-SBA-15 catalysts
Sample no. Ti content
(wt %)
Si/Ti
(atomic ratio)
pHa Structural parameters Textural parameters
Unit cell
parameter (nm)
FWHMb Specific surface area
(m2/g)
Specific mesopore
volume (cm3/g)
Mesopore
diameter (nm)
Wall
thicknessc
Mesopore External
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
3 7.17 10 2.78 12.84 0.033 514 44 1.10 10.9 2.0
Ti-MMMd 1.89 39 9.00 4.23 0.110 1260 29 0.90 3.45 0.8
Adapted from Trukhan et al. (234).a 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.
P.
Ratn
asamy
,D
.S
riniv
asan
dH
.K
nozin
ger
/A
dv
.C
atal.4
8(2
00
4)
1–
16
99
4
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)].
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)].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 95
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.
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)].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–16996
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)].
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
3 1 29 71 29
Ti-MMMa 0.5 100 76 24
TS-1 (Ti, 2.54 wt%) 1 29 79 21
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.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 97
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 oxida-
tion 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-dimethyldibenzo-
thiophene (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. Knozinger / Adv. Catal. 48 (2004) 1–16998
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 sulf-
oxides 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 2500
Ti-beta Acetonitrile 5 Kerosene 80 94.0
Ti-beta Acetonitrile 5 Acetonitrile 2300
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 1800
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.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 99
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. Knozinger / Adv. Catal. 48 (2004) 1–169100
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 benzo-
quinones and products derived from them. The benzoquinones of polyalkyl-
benzenes 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
Scheme 17.
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).a Isolated yield and the rest is essentially unreacted ether.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 101
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.
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.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169102
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.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 103
TABLE XXXVIII
Baeyer–Villiger rearrangement and hydroxylation of acetophenone catalyzed by TS-1/H2O2 system
System Phasea Conv.
(%)
Product selectivities (mol%)b
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
TS-1/H2O2/Hþ Bi 6.1 61.0 – – 4.6 10.8 4.4 19.0
TS-1/H2O2 Bi – – – – – – – –
Blank/H2O2/Hþ Bi 5.5 31.8 6.9 - 24.9 2.8 3.5 30.1
Blank/H2O2/Hþ Mono – – – – – – – –
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 co-
solvent).b PA, phenyl acetate; o-HAP, o-hydroxy acetophenone; p-HAP, p-hydroxy acetophenone; PH,
phenol; CA, catechol; HQ, hydroquinone; AA, acetic acid.
TABLE XXXVII
Oxidation of cyclohexanone catalyzed by TS-1
System Phasea Conv.
(mol%)
Product selectivity (mol%)
1-Capro-
lactone
Hydroxy-
ketone
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 co-
solvent).
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169104
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, 1,3-
cyclohexadiene, 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 19.
Scheme 20.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 105
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 epichloro-
hydrin, 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.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169106
TABLE XXXIX
Synthesis of cyclic carbonates from epoxides and CO2
Run no. Catalyst Co-catalyst Temperature
(8C)
Run time
(h)
Epoxide Conv. of epoxide
(mol%)
TOF Selectivity for cyclic
carbonate (mol%)
1 TS-1 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 100 595 82.0
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.
Ratn
asamy
,D
.S
riniv
asan
dH
.K
nozin
ger
/A
dv
.C
atal.4
8(2
00
4)
1–
16
91
07
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 transesterifica-
tion 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).
TABLE XL
Synthesis of cyclic carbonates from alkenes: epoxidation-cum-cycloaddition
Catalyst Alkene Oxidizingagent
Stage 1: alkeneto epoxide
Stage 2: epoxide tocyclic carbonate
Alkeneconversion toepoxide (%)
Epoxideselectivity
(%)
Epoxideconversion
(%)
Cyclic carbonateselectivity
(%)
TS-1 Allyl chloride H2O2 54.6 100.0 92.5 55.6
TS-1 Styrene H2O2 50.4 89.0 49.2 26.0
TiMCM-41 Allyl chloride TBHP 13.3 100 100 100
TiMCM-41 Styrene TBHP 44.0 93.1 97.2 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).
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169108
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
DPC selectivity
(mol%)a
Chloropropylene carbonate CH3OH 26.5 86.2
Propylene carbonate CH3OH 5.1
Propylene carbonate C6H5OH 58.9 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.
Scheme 22.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 109
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) co-
synthesis 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, dimethylaminopyr-
idine, 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 trans-
esterification 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 (ethyl-
acetoacetate 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
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169110
TABLE XLII
Transesterification of ethylacetoacetate with various alcohols (ROH) over TS-1
Entry ROH Transester product Conv. (mol%) Product yield (%)
1 95.6 92.9
2 100 87.1
3 97.6 90.7
4 99.2 85.0
5 96.2 84.3
(Continued)
P.
Ratn
asamy
,D
.S
riniv
asan
dH
.K
nozin
ger
/A
dv
.C
atal.4
8(2
00
4)
1–
16
91
11
TABLE XLII
Continued
Entry ROH Transester product Conv. (mol%) Product yield (%)
6 96.4 95.3
7 86.4 69.5
8 83.1 66.9
9 CH3(CH2)7CHyCH(CH2)7CH2OH 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.
Ratn
asamy
,D
.S
riniv
asan
dH
.K
nozin
ger
/A
dv
.C
atal.4
8(2
00
4)
1–
16
91
12
(silyl enol ether) with benzaldehyde (Scheme 23) catalyzed by various metallo-
silicates, 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).
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
n-Butanolb 95.8 100 46.4 53.6
n-Butanol (recycle I)a 95.0 100 45.9 54.1
n-Butanol (recycle II)a 94.4 100 46.6 53.4
n-Hexanol 99.7 100 10.9 89.1
n-Octanol 100 100 82.6 17.4
Isobutanol 95.6 96.3 58.0 42.0
Cyclohexanol 100 100 34.4 65.6
Benzyl alcohol 84.2 88.8 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.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 113
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 cata-
lyzed 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).
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 Allylalcohol
Ester conversion
(mol%)
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.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169114
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).
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
Amorphous TiO2-SiO2 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
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.
Scheme 23.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 115
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.
TABLE XLVI
Activities of various metallosilicates for Mukaiyama aldol reaction of benzaldehyde
with silyl enol ether
Catalyst Si/M ratio
(product)
Particle size
(mm)
Micropore volume
(mL/g)
Yield
(%)
TONa
Ti-ZSM-5 (TS-1)b 33.5 0.1–0.2 0.138 85.0 9.2
Sn-ZSM-5b 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
H-ZSM-5b 40.0 0.3–0.5 0.153 - -
Ti-betab 43.0 0.2–0.3 0.269 87.0 11.6
Al-betab 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
Na–Y 2.5–3.0 0.5–0.7 0.343 - -
La–Yc 2.5–3.0 0.5–0.7 0.292 37.0 6.5d
Re–Yc 2.5–3.0 0.5–0.7 0.270 50.0 7.3d
Zn/ZSM-5c 40.0 0.3–0.5 0.141 34.0 4.4e
Adapted from Sasidharan and Kumar (257). Reaction conditions: catalyst, 150 mg; methyl trimethyl-
silyl 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.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169116
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 4-
chlorophenol (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
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.
Scheme 24.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 117
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 photo-
reactivity was attributed to the increased adsorption of 4-CP resulting from
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.
Scheme 25.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169118
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 4-
CP aqueous solutions in the presence of TiO2, TS-2, and TS-2h catalysts in suspension [from Kang
et al. (262)].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 119
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 three-
dimensional in the latter and one-dimensional in the former. Addition of Pt
onto Ti-MCM-48 increased its photocatalytic activity. However, only the for-
mation 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. Knozinger / Adv. Catal. 48 (2004) 1–169120
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 hole-
trapped 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
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)].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 121
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
CH3OH:H2O
(wt%)
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
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. Knozinger / Adv. Catal. 48 (2004) 1–169122
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 alcohol-
water 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
Fig. 46. Influence of solvent dielectric constant (logarithm (ln) values) on (a) phenol hydroxyla-
tion [data taken from Thangaraj et al. (266)] and (b) epoxidation of allyl alcohol catalyzed by TS-1
[data from Wu and Tatsumi (229)].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 123
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
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169124
TABLE L
Epoxidation of ally alcohol with H2O2 in various solvents
Solvent Ti-MWW (Si/Ti ¼ 46) (mol%) TS-1 (Si/Ti ¼ 36) (mol%) Ti-beta (Si/Ti ¼ 42) (mol%)
AA conv. Prod. Sel.a H2O2 AA conv. Prod. Sel.a H2O2 AA conv. Prod. Sel.a H2O2
Gly. Others Conv. Eff. Gly. Others Conv. Eff. Gly. Others Conv. Eff.
MeCN 87.0 99.9 0.1 87.9 99.0 26.8 82.6 17.3 28.5 94.1 13.9 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.
Ratn
asamy
,D
.S
riniv
asan
dH
.K
nozin
ger
/A
dv
.C
atal.4
8(2
00
4)
1–
16
91
25
TABLE LI
Effect of trimethylsilylation on catalytic activity of Ti-containing mesoporous molecular sieves
Catalysta Conv.
(mol% of max.)
TON
(mol/(mol Ti))
Selectivity (%) H2O2 decom
(%)
Alcohol Ketone Epoxide Diol
Cyclohexene
Ti-MCM-41 (nonsil)b 0.72 5.4 30.0 15.2 0 54.7 57.6
Ti-MCM-48 (nonsil)c 2.1 6.1 26.7 32.8 4.7 35.7 61.9
Ti-MCM-41(sil)b 13.3 112.1 14.4 21.0 13.9 50.7 0
Ti-MCM-48 (sil)b 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 – – – – 74.7
Ti-MCM-41 (sil)c 0.06 0.5 40.5 59.5 0.0 0.0 0.0
Ti-MCM-41 (sil)d 0.2 6.8 22.0 22.9 31.0 24.1 97.3
Ti-MCM-48 (nonsil)c 0 0 – – – – 75.0
Ti-MCM-48 (sil)b 0.17 0.52 45.4 54.6 0 0 20.2
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); Ti-
MCM-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.
Ratn
asamy
,D
.S
riniv
asan
dH
.K
nozin
ger
/A
dv
.C
atal.4
8(2
00
4)
1–
16
91
26
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 inter-
action 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 concentra-
tions of the various surface groups and titanium oxo complexes, on the one hand,
TABLE LII
Influence of silynation on epoxidation of cyclohexene with TBHP over Ti-SBA-15
Sample Si/Ti Conversion
(mol%)
TON
(mol/mol Ti)
Product selectivity (mol%)
Oxide Diolsa Allylicb
Ti-SBA1 80 1.9 76 83.9 6.5 9.7
Ti-SBA2 27 2.8 38 86.9 6.8 6.3
Ti-SBA1-silc 87 21.1 843 97.0 2.0 1.9
Ti-SBA2-silc 28 29.8 437 96.0 2.4 1.6
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.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 127
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. Knozinger / Adv. Catal. 48 (2004) 1–169128
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 imme-
diately 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
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)].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 129
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. 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].
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. Knozinger / Adv. Catal. 48 (2004) 1–169130
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 1409-
and 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 H218O2) was used. The epoxide
product, C2H418O, 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].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 131
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 irradia-
tion. 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 oxida-
tion (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 predomi-
nant 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 hydroper-
oxide 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. Knozinger / Adv. Catal. 48 (2004) 1–169132
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.
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)].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 133
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 epoxi-
dation (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
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)].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169134
TABLE LIII
Catalytic activities of TS-1 samples (Si/Ti ¼ 33; particle size ¼ 0.2–0.3 mm) prepared by different methods
Catalyst Epoxidation of Allyl Alcohol (AA)a Phenol hydroxylationb,c
AA conversion
(mol%)
Product selectivity
(mol%)
Phenol conversion
(mol%)
Product selectivity
(mol%)
Glycidol -diol Catechol Hydroquinone
TS-1 (without anatase) 96.1 96.3 3.7 12.9 (16.2) 43.7 (24.9) 56.4 (75.2)
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.b 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.
P.
Ratn
asamy
,D
.S
riniv
asan
dH
.K
nozin
ger
/A
dv
.C
atal.4
8(2
00
4)
1–
16
91
35
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).
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.
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. Knozinger / Adv. Catal. 48 (2004) 1–169136
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 dis-
cussed 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 forma-
tion. Hence, systems having the tripodal Ti(OSi)3(OH) sites in preponderance
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)].
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 137
(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 hydroper-
oxides 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 analo-
gues 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. Knozinger / Adv. Catal. 48 (2004) 1–169138
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 electro-
philic 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.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 139
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-MCM-
41. 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 Ti-
MCM-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(O2z2)
on these catalysts.
These titanium oxo species oxidize various organic reactants. Direct con-
firmations 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. Knozinger / Adv. Catal. 48 (2004) 1–169140
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 H2O2: 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 con-
straints), 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 charac-
terizing modifying their physicochemical and catalytic properties in the coming
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 141
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. Knozinger / Adv. Catal. 48 (2004) 1–169142
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 develop-
ments in the synthesis of micro and mesoporous titanosilicate molecular sieves.
TABLE B1
Technique Characteristic feature
Visual appearance
(color)
Yellow
Diffuse reflectance
UV–visible
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
Vibrational spectroscopy
(infrared and
Raman/resonance Raman)
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.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 143
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 peroxo-
titanate 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 ( pre-
hydrolysis 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 tetraethylpho-
sphonium hydroxides), (tetrapropylammonium bromide and ammonia, water,
hexanediamine, n-butylamine, diethylamine, ethylenediamine, or triethanola-
mine) 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 extraframe-
work 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. Knozinger / Adv. Catal. 48 (2004) 1–169144
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 conven-
tional 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 pre-
hydrolysis 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 impreg-
nation 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 hydro-
thermal 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
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 145
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 (tetrapropylammo-
nium 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 Ti-
beta 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. Knozinger / Adv. Catal. 48 (2004) 1–169146
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 (Ti-
HMS) 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 Ti-
MCM-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 postsyn-
thetic 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
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 147
TABLE C1
Synthesis of titanosilicate molecular sieves
Titanosilicate Synthesis methodology, composition and
improvements
Si/Ti Crystallite size (nm)/
morphology
References
TS-1 (MFI) 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
90–30 Parallelepipeds with
rounded edges
(5)
Dissolved titanium method. Hydrothermal syn-
thesis 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
90–30 Parallelepipeds with
rounded edges
(5)
Preparation using TiCl2, 14% aqueous TPAOH,
30% colloidal silica, and demineralized water
Microspheres of
diameter 5–1000 mm
(284)
Preparation at low pH using fluoride ions as
mineralizing agent
(295,296)
Wetness impregnation method (282,283)
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:xTiO2:0.36-
TPA:35H2O ðx ¼ 0–0:10Þ; the synthesis time is
reduced considerably (1–5 days at 433 K com-
pared to 6–30 days at 448 K, as reported in the
original patent (5))
$10 Cuboid (,1 mm) (275,281)
P.
Ratn
asamy
,D
.S
riniv
asan
dH
.K
nozin
ger
/A
dv
.C
atal.4
8(2
00
4)
1–
16
91
48
Prehydrolysis method. Synthesis using binary
mixtures of tetrabutylammonium and tetraethy-
lammonium hydroxides instead of TPAOH
(285)
Influence of TPAOH/TEAOH and TPAOH/NH4-
OH ratio on the rate of crystallization and
crystallite size investigated
(286)
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
composition, SiO2:xTiO2:0.4 (x0TEPOH þ (1 2
x0)TBPOH):30H2O ðx ¼ 0–0:02Þ;
temperature ¼ 443 K and synthesis time ¼ 4 days
Ovate shaped crystals
(when x0 ¼ 0); hexago-
nal prisms (when
x0 ¼ 0:25–0:5) (2–
3 mm)
(287)
Influence of nature of silicon and titanium
alkoxides on the incorporation of Ti
(288)
Wetness impregnation method (325,326)
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: SiO2:0.01TiO2:0.3C6-
DN:0.1TPABr:50H2O
24–76 Elongated prisms
,7 £ 2.5 £ 0.5 mm
(289)
Prehydrolysis method. Synthesis using SiO2
instead of silica alkoxides. Gel composition:
SiO2:xTiO2:0.4TPAOH:35H2O; 0 , x , 0:03
50–86 Hexagonal prisms/t-
winned conffin shaped
particles (10–19 mm)
(290)
(Continued)
P.
Ratn
asamy
,D
.S
riniv
asan
dH
.K
nozin
ger
/A
dv
.C
atal.4
8(2
00
4)
1–
16
91
49
TABLE C1
Continued
Titanosilicate Synthesis methodology, composition and
improvements
Si/Ti Crystallite size (nm)/
morphology
References
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)
30–80 0.1–0.2 mm (298–300)
Synthesis using TPABr as structure-directing
agent and ammonia, water, hexanediamine, n-
butylamine, diethylamine, ethylenediamine, or
triethanolamine as base (seeds of TS-1 were added
to get smaller crystallites and 100% crystallinity)
(327)
Synthesis of “fibrous” titanosilicate 2.5 mm length and
aspect ratio
(length/diameter) ¼ 50–
70
(328)
Synthesis using TiF4 (as the source of Ti), TEOS,
TPAOH, and distilled water. Gel composition:
SiO2:xTiO2:0.4TPA:30H2O, 0 , x , 0:05
45–90 Round shaped particles
(0.3 mm diameter)
(291,292)
Prehydrolysis method. Crystallization without
evaporating the alcohol in the conventional
synthesis (7,8)
(293)
Preparation by gas–solid isomorphous substitution
of Ti4þ for Si4þ and hydrothermal crystallization
using TPABr as template
(329)
Preparation using TiCl3 as source of Ti: influence
of pH (11.6–9.7)
Crystallite size 0.1–
4 mm
(330)
P.
Ratn
asamy
,D
.S
riniv
asan
dH
.K
nozin
ger
/A
dv
.C
atal.4
8(2
00
4)
1–
16
91
50
Mixed alkoxide method. Synthesis using TPAOH
and HF and wetness-impregnation method using
TPABr and NH4F
(297)
Mixed alkoxide method. Preparation using ethyl-
silicate-40 (ES-40) as the cheaper, cost-effective
source of Si. Gel composition: SiO2:0.03-
TiO2:0.33TPA:35H2O
33 0.1–0.2 mm (294)
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 environ-
ment. Crystalline product dried at 383 K and
calcined at 823 K for 5 h (crystal yield .90%)
50 Round shaped particles
,0.5 mm
(301)
Prehydrolysis method. Synthesis under microwave
irradiation; gel composition: SiO2:xTiO2:0.36-
TPAOH: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
10–33 0.3–1.2 mm (302,303)
Prehydrolysis method. Synthesis using small
amount of TPAOH template in the presence of
Tween 20, a non-ionic surfactant. Gel compo-
sition: 0.03TiO2:SiO2:0.12TPAOH:0.0009Tween
20:0.88IPA:14.45H2O. Crystallized at 433 K for
18 h under autogeneous pressure
33 0.15 mm (304)
TS-2 (MEL) Mixed alkoxide method using TBAOH as structure
directing agent
(6)
(Continued)
P.
Ratn
asamy
,D
.S
riniv
asan
dH
.K
nozin
ger
/A
dv
.C
atal.4
8(2
00
4)
1–
16
91
51
TABLE C1
Continued
Titanosilicate Synthesis methodology, composition and
improvements
Si/Ti Crystallite size (nm)/
morphology
References
Prehydrolysis method using tetraethylorthosilicate,
titanium tetrabutoxide, and tetrabutylammonium
hydroxide. Gel composition: SiO2:xTiO2:0.2-
TBAOH:20H2O, x ¼ 0:14–0:0055; 443 K, 2–7
days
(7,305)
Synthesis using TBPOH as templating agent. Only
a maximum of 1.1 Ti/unit cell can be incorporated
in the framework
(286)
Wetness-impregnated SiO2–TiO2 xerogels Elliptical particles
(,1 mm)
(307)
Synthesis based on hydrolyzed titanium alkoxides
with H2O2. Gel composition: SiO2:xTiO2:0.88-
TBAOH:99H2O:25x H2O2. Crystallization at
449 K
25 Ovate type crystals
(2 mm)
(8,308)
Ti-ZSM-48 Prehydrolysis method. TEOS, TBOT, hexametho-
nium hydroxide template; 473 K, 7 days, crystal-
lization by rotation (40 rpm)
36–60 0.2–0.3 mm Spherical
random agglomerates of
small needle shaped
crystals 5–15 mm diam-
eter containing needles
of 0.2–1 mm long and
diameter 0.1 mm
(309,310)
Hydrolyzed titanium oxide method using fumed
silica as Si source
24–111
Synthesis using hexamethonium hydroxide 49 (311)
P.
Ratn
asamy
,D
.S
riniv
asan
dH
.K
nozin
ger
/A
dv
.C
atal.4
8(2
00
4)
1–
16
91
52
Ti-Beta (BEA;) [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%.
(9,10,110)
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
(312,313)
Seeding technique. Al-free Ti-beta obtained by use
of dealuminated zeolite-beta seeds
(12)
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)
50 (13,314)
Al-free Ti-beta: Direct synthesis (196,315,316)
(Continued)
P.
Ratn
asamy
,D
.S
riniv
asan
dH
.K
nozin
ger
/A
dv
.C
atal.4
8(2
00
4)
1–
16
91
53
TABLE C1
Continued
Titanosilicate Synthesis methodology, composition and
improvements
Si/Ti Crystallite size (nm)/
morphology
References
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:Na2-
O: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
,30 (318)
ETS-10/-4
(Zorite structure)
Synthesis with TiCl3 and without any organic
template
(331–336)
P.
Ratn
asamy
,D
.S
riniv
asan
dH
.K
nozin
ger
/A
dv
.C
atal.4
8(2
00
4)
1–
16
91
54
Influence of various organic bases (R ¼
pyrrolidine, tetramethylammonium chloride, tet-
raethylammonium chloride, tetrapropylammonium
chloride, 1,2-diaminoethane, and 1,2-diaminohex-
ane) on crystallization of ETS-10. Synthesis using
Na2SiO3·n H2O, TiCl4, NaOH, KOH, and distilled
water. Gel composition: 40R:52Na2O:42K2-
O:20TiO2:100SiO2:7030H2O. pH ¼ 10.5–12.9.
Crystallization at 473 K for 2–30 days
(337,338)
ETS-10: Hydrothermal synthesis using TiO2 (P25,
200 A particle size, Degussa), 40% colloidal silica,
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
0.6–1 mm (339)
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
(340)
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
Cuboid or wheat-shaped
agglomerated crystals of
2–4 mm
(341)
(Continued)
P.
Ratn
asamy
,D
.S
riniv
asan
dH
.K
nozin
ger
/A
dv
.C
atal.4
8(2
00
4)
1–
16
91
55
TABLE C1
Continued
Titanosilicate Synthesis methodology, composition and
improvements
Si/Ti Crystallite size (nm)/
morphology
References
Synthesis of ETS-10 using TiCl3 and crystalline
TiO2 (anatase) as Ti sources. Gel composition:
4Na2O:1.5K2O:TiO2:5.5SiO2:125H2O. Crystalli-
zation at 503 K for 24 h
,25 mm crystals (342)
ETS-10 synthesis from gels containing TiF4 and
TiO2
(343)
Ti-MCM-41 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:TMA2-
O ¼ 0.67, (TMA)2O:SiO2 ¼ 0.13,
H2O:(TMA)2O ¼ 188
60 Pore size ¼ 2 nm; sur-
face area ¼
936 m2/g
(17)
Silylation of surface of Ti-MCM-41. Synthesis gel
composition: SiO2:0.015 TEOT:0.26
CTABr:0.26TMAOH:24.3 H2O
66 (321)
Trimethylsilylation: Ti-MCM-41 prepared from
TEOS, TBOT, and CTMACl with molar gel
composition SiO2:0.01TiO2:0.6CTMA:0.3NMe4-
OH:60H2O was silylated with Me3SiCl and
(Me3Si)2O
139 (123 before
silylation)
Pore diameter ¼ 1.9 nm
(2.32 nm before silyal-
tion); pore
volume ¼ 0.82 mL/g
(0.88 mL/g before sily-
lation, surface
area ¼ 139 m2/g
(123 m2/g before silyla-
tion)
(273)
P.
Ratn
asamy
,D
.S
riniv
asan
dH
.K
nozin
ger
/A
dv
.C
atal.4
8(2
00
4)
1–
16
91
56
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(OC2-
H5)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
60–133 (320)
Ti-HMS Synthesis by acid hydrolysis in alcohol solution of
mixture of TEOS and Ti(iso-OC3H7)4 in dodecy-
lamine
100 Pore diameter ,2.8 nm (19)
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)
Ti-MCM-48 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 cetyl-
benzyl dimethylammonium chloride (CBDAC) as
a template. In the grafting method, silicious MCM-
48 first prepared and then the dry surface grafted
with titanium isopropoxide
Pore diameter ¼ 2.6 nm,
Surface area ¼ 1296
m2/g (1093 m2/g for
grafted material)
(25)
(Continued)
P.
Ratn
asamy
,D
.S
riniv
asan
dH
.K
nozin
ger
/A
dv
.C
atal.4
8(2
00
4)
1–
16
91
57
TABLE C1
Continued
Titanosilicate Synthesis methodology, composition and
improvements
Si/Ti Crystallite size (nm)/
morphology
References
Ti-SBA-15 Grafting method. SBA-15 prepared first using the
amphiphilic triblock copolymer poly(ethyleneox-
ide)–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
Pore diameter ¼ 6.3 nm,
surface area ¼ 518
m2/g, pore
volume ¼ 0.68
(25)
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
5–40 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
(25)
Incipient wetness method. For every 1 g of SBA-
15, 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.
0.6–36
(XPS)
Pore size ¼ 4.2–5.1 nm,
specific surface
area ¼ 690–997 m2/g,
volume ¼ 0.81–
1.17 cm3/g
(27)
P.
Ratn
asamy
,D
.S
riniv
asan
dH
.K
nozin
ger
/A
dv
.C
atal.4
8(2
00
4)
1–
16
91
58
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. Further-
more, 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 titano-
silicate (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). Ti-
TUD-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 titan-
ium 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 Perezpariente, J., Zeolites 13, 82 (1993).
11. Corma, A., Camblor, M.A., Esteve, P., Martinez, A., and Perezpariente, 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).
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 159
13. Blasco, T., Camblor, M.A., Corma, A., Esteve, P., Guil, J.M., Martinez, A., Perdigon-
Melon, 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 Perezpariente, J., J. Chem. Soc., Chem. Commun. 147
(1994).
18. Blasco, A.T., Corma, A., Navarro, M.T., and Perezpariente, 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. Knozinger / Adv. Catal. 48 (2004) 1–169160
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., Sanchez, 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. Knozinger 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 Holderich, W.F., J. Catal. 176, 376 (1998).
70. Meiers, R., and Holderich, 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).
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 161
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. Wilkenhoner, 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., Beland, 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. Knozinger / Adv. Catal. 48 (2004) 1–169162
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 Perez-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 Spectro-
scopies.” 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).
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 163
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. Knozinger, H., (G. Ertl, H. Knozinger 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. Knozinger, 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. Knozinger, 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. Knozinger / Adv. Catal. 48 (2004) 1–169164
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., Martınez, 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).
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 165
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 Holderich, 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. Tetenyi, 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 Petro-
chemistry” (J. Weitkamp, Ed.), DGMK Conference, p. 367, 1992.
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169166
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 Martınez, 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).
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 167
270. Sauer, M.V.C., and Edwards, J.O., J. Phys. Chem. 75, 3004 (1971).
271. Bach, R.D., and Knigh, J.W., Org. Synth. 60, 63 (1981).
272. McIsaac, J.E., Ball, R.E., and Behrman, E.J., J. Org. Chem. 36, 3048 (1971).
273. Tatsumi, T., Koyano, K.A., and Igarashi, N., Chem. Commun. 325 (1998).
274. Wu, P., Tatsumi, T., Komatsu, T., and Yashima, T., Chem. Mater. 14, 1657 (2002).
275. Thangaraj, A., Kumar, R., Mirajkar, S.P., and Ratnasamy, P., J. Catal. 130, 1 (1991).
276. Mantegazza, M.A., Petrini, G., Spano, G., Bagatin, R., and Rivetti, F., J. Mol. Catal. A:
Chem. 146, 223 (1999).
277. Chaudhari, K., Bal, R., Srinivas, D., Chandwadkar, A.J., and Sivasanker, S., Micropor.
Mesopor. Mater. 50, 209 (2001).
278. Adam, W., Harold, M., Hill, C.L., Saha-Moller, C.R., and Eur, J. Org. Chem. 941 (2002).
279. Corma, A., Chem. Rev. 97, 2373 (1997).
280. Biz, S., and Occelli, M.L., Catal. Rev. Sci. Engng 40, 329 (1998).
281. Thangaraj, A., Eapen, M.J., Sivasanker, S., and Ratnasamy, P., J. Catal. 12, 943 (1992).
282. Padovan, M., Leofanti, G., Roffia, P., EP Patent No. 311983 (1989).
283. Padovan, M., Genoni, F., Leofanti, G., Petrini, G., Trezza, G., and Zecchina, A., Stud.
Surf. Sci. Catal. 63, 431 (1991).
284. Bellussi, G., Buonomo, F., Esposito, A., Clerici, M., Romano, U., Notari, B., US Patent
No. 4, 701,428 (1987).
285. Tuel, A., Ben Taarit, Y., and Naccache, C., Zeolites 13, 454 (1993).
286. Tuel, A., and Ben Taarit, Y., Micropor. Mater. 1, 179 (1993).
287. Tuel, A., and Ben Taarit, Y., Zeolites 14, 272 (1994).
288. Tuel, A., and Ben Taarit, Y., Appl. Catal. A. Gen. 110, 137 (1994).
289. Tuel, A., Zeolites 16, 108 (1996).
290. Gontier, S., and Tuel, A., Zeolites 16, 184 (1996).
291. Tuel, A., Teissier, R., EP Patent Appl. No. 665188 A 1 (1995).
292. Jorda, E., Tuel, A., Treissier, R., and Kervennal, J., Zeolites 19, 238 (1997).
293. Tuel, A., Catal. Lett. 51, 59 (1998).
294. Sabde, D.P., Hegde, S.G., and Dongare, M.K., J. Mater. Chem. 10, 1365 (2000).
295. 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).
296. Delmotte, L., Soulard, M., Guth, F., Seive, A., Lopez, A., and Guth, J.L., Zeolites 10, 778
(1990).
297. Grieneisen, H., Kessler, H., Fache, E., and Le Govic, A.M., Micropor. Mesopor. Mater.
37, 379 (2000).
298. Kumar, R., Bhaumik, A., Ahedi, R.K., and Ganapathy, S., Nature 381, 298 (1996).
299. Bhaumik, A., Ganapathy, S., and Kumar, R., Stud. Surf. Sci. Catal. 113, 225 (1998).
300. Kumar, R., Mukherjee, P., Pandey, R.K., Rajmohanan, P., and Bhaumik, A., Micropor.
Mesopor. Mater. 22, 23 (1998).
301. Ahn, W.S., Kang, K.K., and Kim, K.Y., Catal. Lett. 72, 229 (2001).
302. 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).
303. 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).
304. Khomane, R.B., Kulkarni, B.D., Paraskar, A., and Sainker, S.R., Mater. Chem. Phys. 76,
99 (2002).
305. Reddy, J.S., and Kumar, R., J. Catal. 130, 440 (1991).
306. Uguina, M.A., Serrano, D.P., Ovejero, G., van Griekson, R., and Camacho, M., Zeolites
18, 368 (1997).
307. de Lucas, A., Rodrıguez, L., and Sanchez, P., Appl. Catal. A: Gen. 180, 375 (1999).
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169168
308. de Lucas, A., Rodrıguez, L., and Sanchez, 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., Perez-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).
P. Ratnasamy, D. Srinivas and H. Knozinger / Adv. Catal. 48 (2004) 1–169 169
This Page Intentionally Left Blank
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 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
II. Electron Microscopy (EM) Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
II.A. Electron Microscopy in Catalysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
II.B. Imaging in the Electron Microscope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
II.C. TEM Imaging Method Using Diffraction Contrast . . . . . . . . . . . . . . . . . . . . . . . 179
II.D. Theoretical Procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
III. High-Resolution Transmission Electron Microscopy. . . . . . . . . . . . . . . . . . . . . . . . . . 181
III.A. Conditions Required for Optimizing HRTEM Images . . . . . . . . . . . . . . . . . . . 182
III.B. Development of HRTEM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
III.C. Elucidation of the Structures of Meso- and Microporous
Catalysts by HRTEM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
III.C.1. L-Type Zeolite Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
III.C.2. Metal-Substituted Aluminum Phosphate (MAPO-36)
Microporous Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
III.C.3. High-Silica Microporous SSZ-48 Catalysts . . . . . . . . . . . . . . . . . . . . . 187
III.C.4. Intergrowths in Zeolite Catalysts: Coherent,
Recurrent, and Random . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
IV. Chemical Composition Analysis with the Analytical Electron Microscope. . . . . . . . . . 191
V. Scanning Transmission Electron Microscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
ADVANCES IN CATALYSIS, VOLUME 48 Copyright q 2004 Elsevier Inc.ISSN: 0360-0564 DOI 10.1016/S0360-0564(04)48002-X All rights reserved
*Corresponding Addresses.
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 . . . . . . . . . . . . . . . . 195
VII. Cathodoluminescence Imaging for Elucidation of Electronic
Structures of Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
VIII. Recent Advances in In Situ Atomic Resolution-Environmental
Transmission Electron Microscopy (ETEM) Under Controlled Reaction Conditions. . . 196
VIII.A. In Situ Investigations of Gas–Solid Reactions and Active Sites . . . . . . . . . . . 196
VIII.B. Illustrative Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
VIII.B.1. In Situ Gas–Catalyst Reactions at the Atomic Level . . . . . . . . . . . 201
VIII.B.2. Atomic-Resolution ETEM of Butane Oxidation . . . . . . . . . . . . . . . 203
VIII.B.3. Atomic-Resolution ETEM of Nanorods . . . . . . . . . . . . . . . . . . . . . 210
VIII.C. Advances in In Situ Wet-Electron Microscopy Technique (Wet-ETEM)
for Probing Solid Catalysts Under Liquid Environments . . . . . . . . . . . . . . . . 210
IX. Environmental Scanning Electron Microscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
X. Electron Tomography: Three-Dimensional Electron Microscopy Imaging . . . . . . . . . . 212
X.A. The Topography and Location of Nanoparticles in Supported
Catalysts; BSE and HAADF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
X.B. Pinpointing the Location of Nanoparticles Supported on Nanoporous Solids. . . . 218
XI. Energy Filtered Transmission Electron Microscopy and Elemental Maps of
Solid Catalysts Using EFTEM. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
XII. Other Significant Trends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
XIII. Critical Evaluations of the Methods and Challenges . . . . . . . . . . . . . . . . . . . . . . . . . . 220
XIV. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 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) pin-
pointing 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) identi-
fying 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 absorption
a unit cell dimension of crystal along a-axis
ADF annular dark field
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227172
AEM analytical electron microscopy
A angstrom units
b unit cell dimension of crystal along b-axis
BET Brunauer, Emmett, Teller surface area
b angle between a- and c-axes in the crystal unit cell
BF bright field
c unit cell dimension of crystal along c-axis
CA concentration of element A in a compound AB
CB concentration of element B in a compound AB
CBDP convergent beam electron diffraction
CCD charge-coupled device
Cs coefficient of spherical aberration of the electron microscope objective lens
CTF contrast transfer function
Df objective lens defocus value
Df ðSÞ Scherzer defocus value
e electron charge
E electron energy
ECELL environmental cell
ED electron diffraction
EDX energy dispersive X-ray spectroscopy
EELS electron energy loss spectroscopy
EFTEM energy-filtered transmission electron microscopy
ELNES electron energy loss near-edge structure
EM electron microscopy (or microscope)
EPMA electron probe microanalysis
ESEM environmental scanning electron microscopy
ETEM environmental-TEM
EXELFS extended energy loss fine structure
F fluorescence
FESEM field emission scanning EM
FðuÞ electron envelope function
FT Fourier transform
g gram
GIF Gatan imaging filter
L Green’s function
HAADF high-angle ADF
HREM or HRTEM, high-resolution TEM
HRSTEM high-resolution scanning TEM
HVEM high-voltage EM
IA background-subtracted peak intensity of element A
IB background-subtracted peak intensity of element B
Iðx; yÞ image intensity of sample in the image plane, k, rate constant
KAB sensitivity factor in analysis for elements A and B in compound AB
kcal kilo calories
l wavelength of electrons
LVSEM low-voltage SEM
m electron mass
MA maleic anhydride
m meter
mm micrometers
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227 173
mbar millibar
mm millimeters
mrad milliradians
mol mole
nm nanometers
PEELS parallel EELS
f0 amplitude of electron wave incident on sample
fg amplitude of scattered electron wave
SEM scanning EM
s interaction constant
sinðxÞ contrast transfer function (CTF)
Cðx; yÞ electron wave function at exit face of sample, with incident electrons
along z-direction
CðrÞ electron wave function at the spatial coordinate, r
SMSI strong metal–support interactions
STM scanning tunneling microscopy
t sample thickness
ti time
TEM transmission electron microscopy
u scattering angle of electrons in radians
XAFS X-ray absorption fine structure
XRD X-ray diffraction
XRE X-ray emission
V volume of crystal unit cell
Vðx; yÞ thickness-projected crystal potential
VðrÞ crystal potential at the spatial coordinate, r
WDS wavelength dispersive X-ray spectroscopy
WPO weak phase object
Z 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 high-
area 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), sum-
frequency 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) 171–227174
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 accommo-
dation 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 spectro-
scopic 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
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
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227 175
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 energy-
filtered 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) 171–227176
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
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227 177
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 develop-
ment 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 , 0:64C1=4s l3=4 ð1Þ
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227178
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.
Fig. 1. Schematic of the information from elastically and inelastically scattered electrons during
the electron beam–sample interactions.
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227 179
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 deter-
mining the dot product of the vectors g and b (called the g·b product analysis).
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.
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227180
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 Schrodinger equation:
72CðrÞ þ 8p2meðE 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
ðfgÞ is a small fraction of the amplitude of the incident wave ðf0Þ 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
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227 181
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) 171–227182
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 phase-
contrast imaging performance of an HRTEM is controlled by sin x; which
contains the basic phase-contrast sinusoidal terms modified by an envelope
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227 183
function, FðuÞ; which is due to the partial coherence of the electron beam:
sin x ¼ FðuÞsin½ð2p=lÞðDfu 2=2 2 Csu4=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
Df ðSÞ ¼ 2C1=2s 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
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).
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227184
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
best home-built HRTEM instruments (,2 A) at a small fraction of the cost, and
it came in a user-friendly package, achieving the full theoretical performance
routinely while fitting in a standard laboratory and requiring no special
buildings. Incremental improvements in resolution (,1.3–1.6 A) were
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–
2.3 A and ,1.3–1.6 A at operating voltages of 200 and 400 keV, respectively,
providing information at the atomic level. New high-voltage (1 MeV) and high-
resolution commercial instruments have also been built, and a point resolution
of ,1 A has been reported (37). Aberration-corrected commercial HRTEM
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 simu-
lations 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.
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227 185
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
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.
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227186
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:
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).
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227 187
a ¼ 11:19 �A; b ¼ 4:99 �A; c ¼ 13:65 �A; and b ¼ 100:78 ðV ¼ 748:6 �A3Þ: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).
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).
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227188
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.
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227 189
The stacking is shown in Fig. 9. This is the situation pertaining to ZSM-3, ZSM-
20, 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. 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).
Fig. 9. Schematic of FAU and EMT intergrowth structures.
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227190
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 microdiffrac-
tion) 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 inner-
shell 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 KABIA=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.
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227 191
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) 171–227192
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. 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.
Fig. 12. Schematic of the information from HRSTEM, DF, and high-angle annular dark-field
(HAADF) microscopy.
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227 193
(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 Z2;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) 171–227194
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, non-
conducting 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.
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227 195
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 Resolution-
Environmental 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 high-
vacuum 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) 171–227196
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) perfor-
mance. 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.
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227 197
In the development of Gai and Boyes (87,88,90), the ECELL, atomic-
resolution (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 four-
aperture 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 low-
electron 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 non-
invasive 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) 171–227198
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.
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.
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227 199
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).
Fig. 15. (a) Novel atomic resolution-ETEM (87) and (b) schematic of various components for
imaging, chemical analysis and diffraction under catalyst operating conditions.
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227200
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
Fig. 15. Continued
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227 201
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
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).
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227202
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 three-
dimensional 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
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227 203
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) 171–227204
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.
Fig. 18. (a) Atomic structure image of VPO and (b) electron diffraction (ED) at room temperature.
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227 205
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
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).
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227206
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.
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227 207
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 solid-
state 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) 171–227208
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 ðtiÞ; 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)
1.5% butane/air; % selectivity at 1.5% butane/N2;
maleic capacity
(micromol/g catalyst)20%
conversion
40%
conversion
60%
conversion
1 2.75 0.110 78 75 71 2.22
2 3.07 0.134 79 78 73 1.23
3 3.35 77 77 74 6.06
4 3.39 ^ 0.22 0.113 82 80 77 4.95 ^ 0.58
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.
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227 209
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) 171–227210
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
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.
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227 211
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, low-
temperature 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-atomic-
number) 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 distri-
bution 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) 171–227212
Electron tomography has been used in biology (115) to investigate the three-
dimensional 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 two-
dimensional 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 two-
dimensional 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
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227 213
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
Fig. 23. Schematic diagram illustrating the acquisition of a series of titled projections and
reconstruction of the three-dimensional object (118,119).
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227214
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 þ60 to 2548 (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,
Fig. 24. Schematic of BSE and HAADF detector geometry.
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227 215
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 three-
dimensional nanoparticle distributions (119). These studies show the consider-
able 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.
Fig. 25. (a) SEM-BSE image and (b) STEM-HAADF image of palladium nanocatalysts on a
carbon support (119).
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227216
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).
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227 217
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 prepara-
tion 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,
individual nanoparticles (of 10–15 A diameter) may be anchored to the inner
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 Z2
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 oxyfunctio-
nalization 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) 171–227218
(125). In the case of silica-encapsulated bimetallic catalysts, one can establish
from precisely coincident element maps taken with Ru K-emission and Pd K-
emission 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).
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)).
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227 219
XII. Other Significant Trends
The electron crystallography method (21) has been used to characterize three-
dimensional structures of siliceous mesoporous catalyst materials, and the three-
dimensional 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 Bronsted 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) 171–227220
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 compu-
tational 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
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227 221
methods are also helpful in gleaning structural information from supported
nanocatalysts. Calculations carried out by Gai et al. (40) show that in spherical-
aberration ðCsÞ 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 determi-
nation 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 maintain-
ing 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) 171–227222
Aberration-corrected ETEM/STEM (130) is expected to offer superior (sub-
atomic) 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 three-
dimensional 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 micro-
analysis 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
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227 223
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 (low-
coordinated 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) 171–227224
18. Gai, P.L., Top. Catal. 21, 161 (2002).
19. Heinemann, H., and Poppa, H., Ultramicroscopy 17, 213 (1985).
20. Ohnishi, N., and Hiraga, K., J. Electron Microsc. 45, 85 (1996).
21. Carlsson, A., Kaneda, M., Sakamoto, Y., Terasaki, O., Ryoo, R., and Joo, S.H., J. Electron
Microsc. 48, 795 (1999).
22. Wagner, P., Terasaki, O., Ritsch, S., Nery, J.G., Zones, S., Davis, M.E., and Hiraga, K.,
J. Phys. Chem. B103, 8245 (1999).
23. Hirsch, P.B., Howie, A., Nocholson, R.B., Pashley, D., and Whelan, M.J., “Electron
Microscopy of Thin Crystals.” Kruger, Florida, 1977.
24. Gai, P.L., and Howie, A., Phil. Mag. A30, 939 (1974).
25. Gai, P.L., and McCarron, E., Science 247, 553 (1990).
26. Menter, R., Proc. R. Soc. A236, 119 (1952).
27. Cowley, J.M., and Moodie, A.F., Acta Cryst. 10, 609 (1957).
28. Cowley, J.M., and Iijima, S., Z. Naturforsch. 27a, 445 (1972).
29. Anderson, J.S., J. Chem. Soc. Dalton Trans. 1107 (1973).
30. Thomas, J.M., Phil. Trans. R. Soc. A277, 251 (1974).
31. Cowley, J.M., “Diffraction Physics.” North-Holland, Amsterdam, 1981.
32. Scherzer, O., J. Appl. Phys. 20, 20 (1949).
33. 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).
34. Thomas, J.M., Jefferson, D.A., and Millward, G.R., JEOL News 23E, 7 (1985).
35. Jefferson, D.A., Thomas, J.M., Millward, G.R., Harrison, A., and Brydson, R.D., Nature
323, 428 (1986).
36. Boyes, E.D., Hutchison, J., and Watanabe, E., JEOL News 24E (1986).
37. Ruhle, M., Ultramicroscopy 56, 1 (1994).
38. Jia, C.L., Lentzen, M., and Urban, K., Science 299, 870 (2003).
39. O’Keefe, M.A., Acta. Cryst. A29, 389 (1973).
40. Gai, P.L., Goringe, M.J., and Barry, J.C., J. Microsc. 142, 9 (1986).
41. Datye, A., and Smith, D.J., Catal. Rev.—Sci. Engng 34 (1992).
42. Yacaman, M.J., and Avalos-Borja, M., Catal. Rev.—Sci. Engng 34 (1992).
43. 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).
44. Wright, P.A., Natarajan, S., Thomas, J.M., Gai-Boyes, P.L., Jones, R.H., and Chen, J.,
Angew. Chem. 31, 1472 (1992).
45. Thomas, J.M., Raja, R., Sankar, G., and Bell, R.G., Acc. Chem. Res. 34, 191 (2001).
46. Millward, G.R., J. Chem. Soc. Faraday Trans. II 79, 1075 (1983).
47. Millward, G.R., Ramdas, S., and Thomas, J.M., Proc. R. Soc. A399, 57 (1985).
48. Thomas, J.M., Audier, M., Klinowski, J., Bursill, L.A., and Ramdas, S., Faraday Discuss.
72, 345 (1981).
49. Terasaki, O., Ohsuma, T., Alfredsson, V., Bovin, J., Watanabe, D., and Anderson, M.W.,
Chem. Mater. 5, 452 (1993).
50. 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.
51. Thomas, J.M., Millward, G.R., White, D., and Ramdas, S., J. Chem. Soc. Chem. Commun.
434 (1988).
52. Gai, P.L., Bart, J.C., and Boyes, E.D., Phil. Mag. A45, 531 (1982).
53. Gai, P.L., and Boyes, E.D., IOP (London) Conf. Ser. 61, 215 (1981).
54. Boyes, E.D., and Gai, P.L., in “Analytical Electron Microscopy” (R. Geiss, Ed.), p. 71.
San Francisco Press, San Francisco, 1981.
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227 225
55. Thomas, J.M., Ultramicroscopy 8, 13 (1982).
56. Boyes, E.D., Goringe, M.J., and Muggridge, B., J. Microsc. 127, 321 (1982).
57. Cliff, I., and Lorimer, G., J. Microsc. 103, 203 (1975).
58. Crewe, A.V., Issacson, M., and Johnson, D.E., Rev. Sci. Inst. 42, 411 (1971).
59. Treacy, M.M.J., and Rice, S.B., Ultramicroscopy 7, 11 (1981).
60. Howie, A., J. Microsc. 117, 11 (1979).
61. Treacy, M.M.J., Howie, A., and Pennycook, S.J., Inst. Phys. Conf. Ser. 52, 261 (1980).
62. Pennycook, S.J., J. Mol. 20, 345 (1983).
63. Ozkaya, D., Zhou, W., Thomas, J.M., Midgley, P.A., Keast, V., and Hermans, S., Catal.
Lett. 60, 113 (1999).
64. (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).
65. Boyes, E.D., Microsc. Microanal. 6, 307 (2000).
66. Boyes, E.D., Adv. Mater. 10, 1277 (1998).
67. Boyes, E.D., Gai, P.L., and Warwick, C., Nature 313, 666 (1985).
68. Hashimoto, H., Naiki, T., Eto, T., and Fujiwara, K., Jpn. J. Appl. Phys. 7, 946 (1968).
69. Swann, P., and Tighe, N., Jernkont. Ann. 155, 251 (1971).
70. Butler, P., and Hale, F., “Dynamic Experiments.” North-Holland, Amsterdam, 1981.
71. Doube, D.D., Mater. Sci. Engng 12, 29 (1973).
72. Parsons, D.F., Adv. Biol. Med. Phys. 15, 161 (1974).
73. Gai, P.L., Phil. Mag. 43, 841 (1981).
74. Gai, P.L., Kristall. Technik 14, 1385 (1979).
75. Gai, P.L., Phil. Mag. 48, 359 (1983).
76. Gai, P.L., J. Solid State Chem. 49, 25 (1983).
77. Gai, P.L., and Labun, P.A., J. Catal. 94, 79 (1985).
78. Gai, P.L., Smith, B.C., and Owen, G., Nature 348, 430 (1990).
79. Doole, R.C., IOP Conf. Ser. 119, 161 (1991).
80. Crozier, P., Sharma, A., and Datye, A., Microsc. Microanal. 4, 228 (1998).
81. Goringe, M.J., Bithell, E., and Doole, R.C., Faraday Discuss. 105, 85 (1996).
82. Parkinson, G.R., and White, D., Proc. XI Int. Congr. Electron Microsc. 331 (1986).
83. Collozo-Davila, C., Marks, L., and Landree, E., Microsc. Microanal. 1, 219 (1995).
84. Baker, R.T., and Chludzinski, J., Carbon 19, 75 (1981).
85. Gai, P.L., and Kourtakis, K., Science 267, 661 (1995).
86. Boyes, E.D., and Gai, P.L., Mater. Res. Soc. 404, 53 (1996).
87. Boyes, E.D., and Gai, P.L., Ultramicroscopy 67, 219 (1997).
88. Gai, P.L., and Boyes, E.D., in “In situ Microscopy in Materials Research” (P.L. Gai, Ed.).
Kluwer, Boston, 1997.
89. Gai, P.L., Acta Cryst. B53, 346 (1997).
90. Gai, P.L., Adv. Mater. 10, 125 (1998).
91. Haggin, J., C&E News (Am. Chem. Soc.) 73, 39 (1995).
92. Jacoby, M., C&E News (Am. Chem. Soc.) 80, 26 (2002).
93. Hansen, T.W., Wagner, J., Hansen, P.L., Dahl, S., Topsøe, H., and Jacobsen, C., Science
294, 1508 (2001).
94. Oleshko, V., Crozier, P., Cantrell, R., and Westwood, A., J. Electron Microsc. 51, S27 (2002).
95. Gai, P.L., Top. Catal. 8, 97 (1999).
96. Tauster, S., Fung, S., and Garten, J.L., J. Am. Chem. Soc. 100, 170 (1978).
97. Gai, P.L., and Smith, B.C., Ultramicroscopy 34, 17 (1990).
98. Sinfelt, J.H., “Bimetallic Catalysts.” Wiley, New York, 1985.
99. Centi, G., Catal. Today 16(1) (1993).
100. Gorbunova, Yu., and Linde, S., Dokl. Acad. SSSR 245, 584 (1979).
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227226
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 three-
dimensional 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.
J. M. Thomas and P. L. Gai / Adv. Catal. 48 (2004) 171–227 227
This Page Intentionally Left Blank
Chemistry and Technology of
Isobutane/Alkene Alkylation
Catalyzed by Liquid and Solid Acids
ANDREAS FELLER1 and JOHANNES A. LERCHER
Institut fur Technische Chemie, Technische Universitat Munchen,
D-85747 Garching, Germany
I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
II. Alkylation Mechanism. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
II.A. Overall Product Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
II.B. Initiation Steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
II.C. Alkene Addition and Isomerization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239
II.D. Hydride Transfer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
II.E. Oligomerization and Cracking. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247
II.F. Self-Alkylation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
II.G. Product and Acid Degradation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251
II.H. Pathways to Allylic and Cyclic Compounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251
II.I. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252
III. Physical–Chemical Phenomena Influencing the Reaction . . . . . . . . . . . . . . . . . . . . . . . 252
III.A. Properties of Liquid Acid Alkylation Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . 253
III.B. Properties of Zeolitic Alkylation Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255
III.B.1. Adsorption and Diffusion of Hydrocarbons . . . . . . . . . . . . . . . . . . . . . . 255
III.B.2. Brønsted Acid Sites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256
III.B.3. Lewis Acid Sites and Extra-Framework Aluminum . . . . . . . . . . . . . . . . 260
III.B.4. Silicon/Aluminum Ratio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
III.B.5. Metal Ions in Ion-Exchange Positions . . . . . . . . . . . . . . . . . . . . . . . . . . 263
III.B.6. Structure Types of Zeolites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264
III.C. Other Solid Acids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
III.C.1. Sulfated Zirconia and Related Materials . . . . . . . . . . . . . . . . . . . . . . . . 267
III.C.2. Heteropolyacids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268
III.C.3. Acidic Organic Polymers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
III.C.4. Supported Metal Halides. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270
III.D. The Influence of Process Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
III.D.1. Reaction Temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
ADVANCES IN CATALYSIS, VOLUME 48 Copyright q 2004 Elsevier Inc.ISSN: 0360-0564 DOI 10.1016/S0360-0564(04)48003-1 All rights reserved
E-mail address: [email protected] address: CS CLEAN SYSTEMS AG, Fraunhoferstr. 4, 85732 Ismaning, Germany.
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295
III.D.2. Alkane/Alkene Ratio and Alkene Space Velocity. . . . . . . . . . . . . . . . . . 274
III.D.3. Alkene Feed Composition. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276
IV. Industrial Processes and Process Developments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
IV.A. Liquid Acid-Catalyzed Processes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
IV.A.1. Sulfuric Acid-Catalyzed Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
IV.A.2. Hydrofluoric Acid-Catalyzed Processes . . . . . . . . . . . . . . . . . . . . . . . . . 281
IV.B. Solid Acid-Catalyzed Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
IV.B.1. UOP Alkylenee Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285
IV.B.2. Akzo Nobel/ABB Lummus AlkyCleane Process. . . . . . . . . . . . . . . . . . 286
IV.B.3. LURGI EUROFUELw Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 286
IV.B.4. Haldor Topsøe FBAe Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
V. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 289
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 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 acid-soluble oil
DMH dimethylhexane
EFAL extra-framework aluminum
H0 Hammett acidity function
k rate constant
LHSV liquid-hourly space velocity (m3olefin/(m3
catalyst h))
OSV olefin space velocity (kgolefin/(kgcatalyst h))
P/O paraffin/olefin ratio (mol/mol)
r reaction rate
RE rare earth
RON research octane number
T temperature (K)
TMP trimethylpentane
TOS time on stream
TS transition state
WHSV 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) 229–295230
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 approxi-
mately 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 methyl-
tertiary-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 commercia-
lized (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,
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295 231
Fig. 1. Process units in a modern refinery.
A.
Feller
and
J.A
.L
ercher
/A
dv
.C
atal.4
8(2
00
4)
22
9–
29
52
32
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 acid-
catalyzed 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-to-
date 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.
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295 233
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 high-
lighted. 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, oligomeriza-
tion, 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.
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295234
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
H2SO4
(T ¼ 528 K, P=O ¼ 5)
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)
Isopentane 93.0 1.2 1.8 6.8 24.0
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
(Continued)
A.
Feller
and
J.A
.L
ercher
/A
dv
.C
atal.4
8(2
00
4)
22
9–
29
52
35
TABLE I
Continued
Component
(wt%)
Research octane number Catalyst
H2SO4
(T ¼ 528 K, P=O ¼ 5)
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)
2,2,4-Trimethylpentane 100 30.2 48.7 23.8 25.5
n-Heptane 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 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
2,3,3-Trimethylpentane 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
Octenes .90 0 0 0.3 1.3
C9þ <80–85 5.4 2.9 7.5 3.3
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
.L
ercher
/A
dv
.C
atal.4
8(2
00
4)
22
9–
29
52
36
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
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295 237
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, Engel-
hardt 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
Fig. 2. Potential energy profile and structure of final alkoxide for the adsorption of isobutylene on a
high-silica zeolite according to Ref. (29).
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295238
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
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295 239
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 2,2,3-
TMP 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 com-
positions, 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) 229–295240
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 2,2,3,3-
tetramethylbutane 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.
Fig. 3. Possible hydride- (<H) and methyl-shifts (<CH3) between the individual TMP isomers.
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295 241
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) 229–295242
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 ion-
like 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/tert-
butoxide (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 zeolite-
catalyzed 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
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295 243
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
Fig. 4. Potential energy profiles for the isobutane/t-butyl cation hydride transfer reaction in various
media (25,64).
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295244
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 1-
butene 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
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295 245
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 obser-
vation 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 2,2,3-
TMP 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 H-
BEA 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) 229–295246
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 high-
quality 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 rCÞ 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).
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295 247
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 informa-
tion characterizing the b-scission step. On the basis of this work, a classification
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295248
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 character-
ized 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
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295 249
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 self-
alkylation 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-C4H10 þ CnH2n ! i-C8H18 þ CnH2nþ2 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) 229–295250
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 conden-
sation and rearrangement steps. Aromatic compounds are usually not formed
under such mild conditions (95). Extending these results to typical alkylation
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295 251
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 alkyl-
substituted 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. Dimeriza-
tion 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) 229–295252
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
Fig. 8. Concerted alkylation mechanism including alkylation, “self-alkylation”, cracking,
dimerization, and hydride transfer via isobutane and via conjunct polymers.
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295 253
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 hydro-
fluoric 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 concen-
trations 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 214.1 for H2SO4
and 212.1 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 28.0 and
28.5 (109). If such low-acidity values occur in an alkylation reactor, oligomer-
ization 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 acid-
catalyzed 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) 229–295254
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 oligomeriza-
tion. 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 oligo-
merization 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.
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295 255
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) 229–295256
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).
A.
Feller
and
J.A
.L
ercher
/A
dv
.C
atal.4
8(2
00
4)
22
9–
29
52
57
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 H-
SAPO-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 above-
presented investigations, it can be concluded that stronger sites are necessary to
effectively catalyze hydride transfer. Stocker 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) 229–295258
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 H-
SAPO-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 acid-
catalyzed 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).
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295 259
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) 229–295260
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 concen-
trations 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
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295 261
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 perfor-
mance 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) 229–295262
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
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295 263
(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 dimethyl-
hexanes 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 inter-
mediate 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) 229–295264
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 2,2,4-
TMP 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,
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295 265
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 large-
pore 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 H-
and 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 alumino-
silicate 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) 229–295266
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 well-
demonstrated 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-on-
stream 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 SO422/
ZrO2 . SO422/TiO2 . SO4
22/SnO2, which was reflected in the cracking activities
of the samples. All the oxides showed considerable sensitivity towards the modi-
fication procedure, each with a different optimum. All the samples deactivated
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295 267
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 gas-
phase 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 oxide-
impregnated 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 NH4þ, 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) 229–295268
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 contra-
dicts 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 hetero-
polyacid/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
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295 269
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 Friedel-
Crafts 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) 229–295270
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 high-
quality 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
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295 271
effective regeneration set high hurdles for a competition with sulfuric acid-
catalyzed 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.
TABLE II
Typical values of important process parameters for alkylation
Catalyst
HF H2SO4 Zeolites
Reaction temperature (K) 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
Catalyst productivity (kgAlkylate/kgAcid) 1000–2500 6–18 4–10
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.
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295272
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 per-
formance 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
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295 273
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 con-
ventional 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 super-
critical conditions generally does not improve the alkylation performance.
A temperature-programmed-oxidation analysis of samples coked under super-
critical 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 appro-
ximately 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) 229–295274
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 instan-
taneously 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 acid-
catalyzed 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.
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295 275
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. Self-
alkylation 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 2-
butene . 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) 229–295276
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).
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
C5
Isopentane 1.0 3.8 0.5 10.0 0.3 4.2 1.0 4.7
C6
Dimethylbutanes 0.3 0.8 0.7 0.8
Methylheptanes 04.2
0.25.2
0.24.6
0.34.4
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
C8
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 0 1.9 0.9
2,3,4-Trimethylpentane 7.5 12.8 22.2 19.1
2,3,3-Trimethylpentane 4.03.7
7.123.1
12.941.6
9.739.1
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
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295 277
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 nitrogen-
containing 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) 229–295278
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).
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295 279
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) 229–295280
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 high-
octane 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
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295 281
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
Fig. 13. ConocoPhillips HF alkylation reactor (236).
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295282
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
Fig. 14. UOP HF alkylation reactor.
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295 283
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. Super-
critical 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 long-
chain 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) 229–295284
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).
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295 285
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 nitrogen-
containing 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 Sud-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) 229–295286
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
Fig. 16. LURGI EUROFUELw solid acid-catalyzed alkylation process (240).
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295 287
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 by-
product. 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. 17. Reaction zone in Haldor Topsøe’s FBAe alkylation process (206).
Fig. 18. Haldor Topsøe’s FBAe alkylation process. Adapted from Ref. (206).
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295288
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. Stell, J., Oil Gas J. 99(52), 74 (2001).
2. Anonymous, Oil Gas J. 98(13) (2000).
3. Anonymous, Oil Gas J. 98(9) (2000).
4. Ipatieff, V.N., and Grosse, A.V., J. Am. Chem. Soc. 57, 1616 (1935).
5. Ipatieff, V.N., and Pines, H., US Patent 2,122,847 (1938).
6. 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.
7. Pines, H., Chemtech. March, 150 (1982).
8. Albright, L.F., Chemtech. June, 40 (1998).
9. Albright, L.F., Oil Gas J. 26, 70 (1990).
10. Furimsky, E., Catal. Today 30, 223 (1996).
11. Albright, L.F., Chemtech. July, 46 (1998).
12. Kinnear, S., “STRATCO Alkylation Seminar,” Phoenix, AZ, 1998.
13. Garwood, W.E., and Venuto, P.B., J. Catal. 11, 175 (1968).
14. Kirsch, F.W., Potts, J.D., and Barmby, D.S., J. Catal. 27, 142 (1972).
15. Corma, A., and Martinez, A., Catal. Rev.—Sci. Engng 35, 483 (1993).
16. Weitkamp, J., and Traa, Y., in “Handbook of Heterogeneous Catalysis” (G. Ertl,
H. Knozinger and J. Weitkamp, Eds.), Vol. 4, p. 2039. VCH, Weinheim, 1997.
17. Schmerling, L., J. Am. Chem. Soc. 67, 1778 (1945).
18. Schmerling, L., J. Am. Chem. Soc. 68, 275 (1946).
19. Schmerling, L., Ind. Engng Chem. 45, 1447 (1953).
20. Albright, L.F., Spalding, M.A., Faunce, J., and Eckert, R.E., Ind. Engng Chem. Res. 27,
391 (1988).
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295 289
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. Knozinger 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) 229–295290
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. Schollner, R., and Holzel, 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. Schollner, R., and Holzel, 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).
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295 291
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., Torok, B., Wang, Q., Molnar, A., and Surya
Prakash, G.K., Appl. Catal. A 146, 107 (1996).
111. Olah, G.A., Marinez, E., Torok, B., and Prakash, G.K.S., Catal. Lett. 61, 105 (1999).
112. Holderich, 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. Janchen, 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., Stocker, M., Karlsson, A., and Rørvik, T., Appl. Catal. A 144, 305 (1996).
123. Stocker, M., Mostad, H., and Rørvik, T., Catal. Lett. 28, 203 (1994).
124. Stocker, 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) 229–295292
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 Orchilles, 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., Stocker, 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. Tetenyi, Eds.), Vol. B, p. 1735. Akademiai Kiado, Budapest, 1993.
171. Chu, C.T.W., Husain, A., Keville, K.M., and Lissy, D.N., US Patent 5,516,962 (1996).
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295 293
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 Stocker, 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) 229–295294
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., “Proceed-
ings of the DGMK-Conference Catalysis on Solid Acids and Bases,” Berlin 1996, p. 175.
A. Feller and J. A. Lercher / Adv. Catal. 48 (2004) 229–295 295
This Page Intentionally Left Blank
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 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 298
II. Partial Oxidation of Methane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301
II.A. Hot Spots in Catalyst Beds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301
II.B. Minimizing O2 Purification Costs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 306
II.C. Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312
II.D. Reaction Pathways . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 314
II.D.1. Changes in Catalyst During Reaction . . . . . . . . . . . . . . . . . . . . . . . . . . . 315
II.D.2. Which is the Primary Product, CO or CO2? . . . . . . . . . . . . . . . . . . . . . . . 316
II.D.3. CHx Species and Rate-Determining Steps . . . . . . . . . . . . . . . . . . . . . . . . 318
II.D.4. Comparison of Reactions on Reduced and Unreduced Catalysts . . . . . . . . 320
III. CO2 Reforming of Methane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 321
III.A. Carbon Formation on Metal Surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 321
III.B. Critical Issues Related to Carbon Deposition . . . . . . . . . . . . . . . . . . . . . . . . . . . 322
III.C. Supported Noble Metal Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 323
III.D. Non-Noble Metal Supported Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 324
III.D.1. Ni/Al2O3 Catalysts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 325
III.D.2. Ni/SiO2 Catalysts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327
III.D.3. Ni/La2O3 Catalysts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 328
III.D.4. Ni/ZrO2 Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 330
III.D.5. Other Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 331
III.E. MgO-Containing Solid-Solution Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332
III.E.1. Characteristics of MgO-Containing Solid-Solution Catalysts. . . . . . . . . . 332
III.E.2. Highly Effective MgO-Containing Solid-Solution Catalysts . . . . . . . . . . 333
IV. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 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 Copyright q 2004 Elsevier Inc.ISSN: 0360-0564 DOI 10.1016/S0360-0564(04)48004-3 All rights reserved
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 diffuse reflectance infrared Fourier transform
EDS energy dispersive X-ray spectrometer
FTIR Fourier transform infrared
GHSV gas hourly space velocity
MIEC mixed ionic/electronic conductors
MS mass spectrometer
R reaction rate
SPARG sulfur passivated reforming
TEM transmission electron microscopy
TG/DTG thermal gravimetric/differential thermal gravimetric
TPD temperature-programmed decomposition
TPH temperature-programmed hydrogenation
TPO temperature-programmed oxidation
TPR temperature-programmed reduction
TPSR temperature-programmed surface reaction
WHSV weight hourly space velocity
XRD X-ray diffraction
w/o water-in-oil
XPS 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 H2O ! CO þ 3H2; DH0298 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) 297–345298
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 þ12
O2 ! CO þ 2H2; DH0298 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; DH0298 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 long-
chain 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
Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345 299
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).
TABLE I
Gibbs free energy change, DG0; for methane transformation reactions (10)
Reaction DG0 (kcal mol21)
400 K 600 K 800 K 1000 K
CH4 þ12
O2 !12
C2H4 þ H2O 234.6 235.1 235.8 236.4
CH4 þ14
O2 !12
C2H6 þ12
H2O 218.4 217.1 215.8 214.5
CH4 þ12
O2 ! CH3OH 225.4 223.0 220.5 218.0
CH4 þ O2 ! HCHO þ H2O 269.0 270.0 270.8 271.2
CH4 þ 2O2 ! CO2 þ 2H2O 2191.3 2191.3 2191.3 2191.3
CH4 þ12
O2 ! CO þ 2H2225.0 233.9 243.1 252.5
CH4 þ H2O ! CO þ 3H2 28.6 17.3 5.5 26.5
Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345300
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,
Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345 301
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) 297–345302
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.
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).
Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345 303
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 fluidized-
bed 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.
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).
Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345304
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. Further-
more, 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%.
Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345 305
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) 297–345306
of electronic charge carriers across the MIEC membrane constitute a charge-
compensation 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
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.
Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345 307
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
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).
Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345308
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
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).
Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345 309
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
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).
Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345310
cost-shared project) made up of Air Products, Arco (now part of BP), Argonne
National Laboratory, Babcock and Wilcox, Ceramate (Salt Lake City), Chevron-
Texaco, 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)
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).
Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345 311
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, pre-
reduction, and feed ratio on the reaction of CH4/O2 mixtures catalyzed by NiO/a-
Al2O3 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-contain-
ing 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 non-
noble 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) 297–345312
TABLE III
Noble metal catalysts for partial oxidation of methane
Metal Support References
Rh Al2O3 (89,94,102,104,106,126)
SiO2 (101)
MgO (101)
a-Al2O3 monolith (16,105)
Pt Al2O3 (89,91,99,126,148)
CeO2/Al2O3 (91)
MgO (95)
ZrO2 (148)
CeO2 (99)
CeO2/ZrO2 (148)
a-Al2O3 monolith (16)
Pt sponge (103)
Pd Al2O3 (89,126)
a-Al2O3 monolith (16)
Ir TiO2 (90,92,96)
a-Al2O3 monolith (16)
Al2O3 (126)
Eu2O3 (126)
Ru SiO2 (43,93)
Al2O3 (10,93,126,127)
YSZ (yttria-stabilized zirconia) (93)
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)
Re a-Al2O3 monolith (16)
Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345 313
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 commercializa-
tion 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) 297–345314
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 carbon-
containing 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
Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345 315
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
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).
Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345316
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
Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345 317
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 dis-
sociation 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) 297–345318
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 rate-
determining 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,
Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345 319
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 con-
ditions. 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:
CH4ðgÞ þ 4NiO ! CO2ðgÞ þ 2H2OðgÞ þ 4Ni ð4Þ
2Ni þ O2ðgÞ! 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 nickel-
containing 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) 297–345320
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; DH0298 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
Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345 321
(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 commercia-
lized 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) 297–345322
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).
Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345 323
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) 297–345324
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
Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345 325
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) 297–345326
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 carbon-
containing 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
Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345 327
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) 297–345328
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.
Ruckenstein and Hu (230) investigated the role of the anions NO32 or Cl2
(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).
Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345 329
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) 297–345330
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 solid-
solution 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
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).
Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345 331
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) 297–345332
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).
Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345 333
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
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).
Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345334
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 hydrogen-
ation (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
Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345 335
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 solid-
solution 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) 297–345336
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-to-
liquid process, can account for at least 60% of the integrated cost of the total gas-
to-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 industrial-
scale 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
Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345 337
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. Byrne, J.P.J., Gohr, R.J., and Haslam, R.T., Ind. Engng Chem. 24, 1129 (1932).
2. Rostrup-Nielsen, J.R., Sehested, J., and Nørskov, J.K., Adv. Catal. 47, 65 (2002).
3. Prettre, M., Eichner, C., and Perrin, M., Trans. Faraday Soc. 43, 335 (1946).
4. Lang, J., Z. Phys. Chem. 2, 161 (1888).
5. Gadalla, A.M., and Bower, B., Chem. Engng Sci. 43, 3049 (1988).
6. 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.
7. Puskas, I., CHEMTECH 24, 43 (1995).
8. (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).
9. (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).
10. Kiennemann, A., Petrol. Technol. 29, 355 (1990).
11. Guczi, L., van Santen, R.A., and Sarma, K.V., Catal. Rev.—Sci. Engng 38, 249 (1996).
12. Pasquon, I., Plenary lecture at EUOPACAT-1, Montpellier, France, September 12–17,
1993.
13. 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).
14. Hickman, D.A., and Schmidt, L.D., J. Catal. 138, 267 (1992).
15. Hickman, D.A., and Schmidt, L.D., Science 259, 343 (1993).
Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345338
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).
Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345 339
60. Wiley, J.B., and Poeppelmeier, K.R., J. Solid State Chem. 88, 250 (1990).
61. Eror, N.G., and Balachandran, U., J. Solid State Chem. 40, 85 (1981).
62. Balachandran, U., and Eror, N.G., J. Phys. Chem. Solids 44, 231 (1983).
63. Teraoka, Y., Nobunaga, T., Okamoto, K., Miura, N., and Yamazoe, N., Solid State Ion. 48,
207 (1991).
64. Balachandran, U., Dusek, J.T., Maiya, P.S., Mieville, R.L., Kleefisch, M.S., and Udovich,
C.A., Catal. Today 133, 19 (1997).
65. 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).
66. 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).
67. Mazanec, T.J., Cable, T.L., Frye, J.G., and Kliewer, W.R., US Patent 5,306,411 (1994).
68. Schwartz, M., While, J.H., Myers, M.G., Deych, S., and Sammells, A.F., Prepr. ACS Div.
Fuel Chem. 42, 596 (1997).
69. Tsai, C.Y., Dixon, A.G., Moser, W.R., and Ma, Y.H., AIChE J. 43, 2741 (1997).
70. Tsai, C.Y., Dixon, A.G., Ma, Y.H., Moser, W.R., and Pascucci, M.R., J. Am. Ceram. Soc.
81, 1437 (1998).
71. Li, G.T., Shao, Z.P., Xiong, G.X., Dong, H., Cong, Y., and Yang, W.S., J. Inorg. Mater.
17, 1041 (2002).
72. Dong, H., Shao, Z.P., Xiong, G.X., Tong, J.H., Sheng, S.S., and Yang, W.S., Catal. Today
67, 3 (2001).
73. Tong, J.H., Yang, W.S., Cai, R., Zhu, B.C., and Lin, L.W., Catal. Lett. 78, 129 (2002).
74. DOE Techline, May 20,1997.
75. Oil & Gas Journal, Page 62, May 3, 1999.
76. 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.
77. 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.
78. Teraoka, Y., Nobunaga, T., Okamoto, K., Miura, N., and Yamazoe, N., Solid State Ion. 48,
207 (1991).
79. Teraoka, Y., Zhang, H.M., Okamoto, K., and Yamazoe, N., Mater. Res. Bull. 23, 51 (1988).
80. Mazanec, T.J., Solid State Ion. 70/71, 11 (1994).
81. Steele, B.C.H., Mater. Sci. Engng B 13, 79 (1992).
82. Steele, B.C.H., Curr. Opin. Solid State Mater. Sci. 1, 684 (1996).
83. Lewis, W.K., Gilliland, E.R., and Reed, W.A., Ind. Engng Chem. 41, 1227 (1949).
84. Lewis, W.K., and Gilliland, E.R., US Patent 2,603,608 (1952).
85. Lewis, W.K., US Patent 2,631,934 (1953).
86. Read, W.A., DSc Dissertation, Massachusetts Institute of Technology, 1948.
87. Huszar, K., Racz, G., and Szekely, G., Acta Chim. Acad. Sci. Hungary 70, 287 (1971).
88. Gavalas, G.R., Phichitcul, C., and Voecks, G.E., J. Catal. 88, 54 (1984).
89. Kikuchi, E., and Chen, Y., Natural gas conversion V. Stud. Surf. Sci. Catal. 119, 441
(1998).
90. Elmasides, C., Ioannides, T., and Verykios, X.E., Natural gas conversion V. Stud. Surf.
Sci. Catal. 119, 801 (1998).
91. 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).
92. 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) 297–345340
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).
Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345 341
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) 297–345342
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., Knozinger, 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).
Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345 343
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) 297–345344
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).
Y. H. Hu and E. Ruckenstein / Adv. Catal. 48 (2004) 297–345 345
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 com-
pounds, 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
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 isomeriza-
tion, 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
Index348
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
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
Index 349
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
Index350
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
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
Index 351
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
Index352
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
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 trans-
mission 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
Index 353
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
Index354
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
Partial dislocations, 181
Partial oxidation of methane
catalyst changes during reaction,
315–316
catalyst composition and tempera-
ture 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
Index 355
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
Index356
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
mixedalkoxidemethod,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
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
Index 357
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
Index358
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
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
Index 359
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
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
Index360