Mechanisms in Homogeneous Catalysis: A Spectroscopic Approach

Mechanisms in Homogeneous Catalysis Edited by Brian Heaton Mechanisms in Homogeneous Catalysis. A Spectroscopic Approa...

Author: Brian Heaton

119 downloads 1337 Views 6MB Size Report

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

DOWNLOAD PDF

Mechanisms in Homogeneous Catalysis Edited by Brian Heaton

Mechanisms in Homogeneous Catalysis. A Spectroscopic Approach. Edited by Brian Heaton Copyright c 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN 3-527-31025-8

Further Titles of Interest B. Cornils, W. A. Herrmann, R. Schlgl, C.-H. Wong (Eds.)

Catalysis from A to Z A Concise Encyclopedia 2nd Edition 2003 ISBN 3-527-30373-1

I. Chorkendorff, J. W. Niemantsverdriet

Concepts of Modern Catalysis and Kinetics 2003 ISBN 3-527-30574 -2

J. W. Niemantsverdriet

Spectroscopy in Catalysis 2nd Edition 2000 ISBN 3-527-30200-X

Mechanisms in Homogeneous Catalysis A Spectroscopic Approach Edited by Brian Heaton

Edited by Prof. Dr. Brian Heaton University of Liverpool Department of Chemistry Liverpool L69 7ZD UK

All books published by Wiley-VCH are carefully produced. Nevertheless, authors, editors, and publisher do not warrant the information contained in these books, including this book, to be free of errors. Readers are advised to keep in nimd that statements, data, illustrations, procedural details or other items may inadvertently be inaccurate. Library of Congress Card No.: applied for British Library Cataloguing-in-Publication Data: A catalogue record for this book is available from the British Library. Bibliographic information published by Die Deutsche Bibliothek Die Deutsche Bibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data is available in the Internet at < http://dnb.ddb.de > c 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim All rights reserved (including those of translation into other languages). No part of this book may be reproduced in any form – by photoprinting, microfilm, or any other means – nor transmitted or translated into a machine language without written permission from the publishers. Registered names, trademarks, etc. used in this book, even when not specifically marked as such, are not to be considered unprotected by law. Printed in the Federal Republic of Germany. Printed on acid-free paper. Typesetting hagedorn kommunikation, Viernheim Printing betz-druck GmbH, Darmstadt Bookbinding Litges & Dopf Buchbinderei GmbH, Heppenheim ISBN-13: ISBN-10:

978-3-527-31025-8 3-527-31025-8

Preface This volume brings together leading international authors who have made important contributions to developing/applying NMR and IR spectroscopic methods to the study of a wide variety of industrial, homogeneous transition metal catalysed reactions used for the manufacture of high tonnage products (eg. aldehydes and alcohols) to lower volume, speciality chemicals. The spectroscopic identification of catalytic intermediates in the elucidation of the catalytic cycle, together with the rates and mechanisms of the individual steps, have long been of interest to both academic and industrial chemists. A better understanding of the catalytic cycle, using the two most widely applicable spectroscopic techniques-IR and NMR-in this area, has allowed improved overall rates of the catalytic reaction (sometimes it has been possible to even measure and improve the rates for individual steps in the catalytic cycle), selectivities and a reduction in by-product formation to be achieved through systematically varying ligand design, the metal, promoters and reaction conditions. In this way, many processes have been improved and new ones developed. NMR is one of the most powerful methods for structural identification and for obtaining information about both the type and rate of inter- and intra-exchange processes; recent developments also allow information about diffusion and ionpairing to be obtained. Detailed structural information in solution is possible from NMR measurements, since most of the elements in the Periodic Table can be used for NMR measurements. NMR no longer relies solely on variable temperature 1-D multinuclear measurements for the identification of catalytic species but a variety of 2-D NMR methods, using either 1-, 2- or 3-bond coupling constants, allow data to be obtained in a much more efficient manner. Thus, although rhodium complexes have long been known to be active catalysts, 103Rh NMR data, hitherto, have been difficult to obtain because of the low sensitivity of 103Rh; now, 103Rh NMR data can be readily obtained, using 2-D HMQC methods, which rely upon nJ(Rh-X) (X ¼ 1H, 13C, 31P, etc; n ¼ 1, 2, 3). This, together with the vast armoury of NMR methods now applicable to the study of homogeneous catalysis, are dealt with in chapter 1 and, hopefully, this will allow the non-specialist NMR person to select and use the appropriate method to solve their particular problem.


VI

Preface

One inherent problem with NMR is the relatively low sensitivity of the technique. Thus, for a reasonable S : N/collection time, it is necessary to use solutions of 10-100 mM, which is well above the concentrations (often I1 mM) used in catalytic experiments. As a result, species present in the catalytic solution can sometimes be different at the higher concentrations used for NMR measurements. However, for reactions involving H2, the use of p-H2 circumvents these problems and detection of species at very low concentrations is possible (chapters 1, 6 and 9). IR spectroscopy is complementary to NMR and is especially useful for reactions involving M-CO’s; it is suitable for the study of catalytic solutions, at the catalytic concentrations used, and has long been used to study catalytic reactions to identify species and obtain rates e. g. Forster’s work at Monsanto in the 1970’s when he successfully clarified the nature of the catalytic cycle for the rhodium-catalysed conversion of methanol to acetic acid. For M-CO’s, it is much easier and quicker to obtain IR spectra containing y(CO) bands than to obtain NMR spectra of even the most sensitive nucleus, 1H, even at the highest magnetic fields now available. However, since the dispersion of y(CO) bands is not very great, deconvolution of IR spectra and identification of species present in catalytic solutions has in the past been somewhat difficult. Garland has now introduced a powerful method (chapter 4) for the reconstruction of individual pure component IR spectra from complex component catalytic mixtures-the Band Target Entropy Minimisation (BTEM) protocol; this is an extremely powerful computational method, which presently allows recovery of pure component IR spectra of unknown species when present at very low concentrations. This method seems to be generally applicable and is being presently extended to include NMR, X-ray powder diffraction etc. Many catalytic reactions require high pressures of reactant gases. Thus, an indepth understanding of such catalytic systems requires truly in situ NMR and IR measurements and it has been necessary to develop appropriate High Pressure-spectroscopic cells; the development and use of HP-NMR and HP-IR cells are reviewed in chapters 2 and 3 respectively. The use of both of these complementary methods/HP-techniques is probably best illustrated in chapters 5 – carbonylation reactions, chapter 6 – hydroformylation and chapter 7 – alkene/CO copolymerisation, which deal with the recent advances in each of these important areas. Over the last 20 years, there has been an enormous increase in the use of NMR spectroscopy in metallocene-based polymerisation catalysis and these studies have provided an unprecedented increase in our understanding; this has allowed the properties of homo- and co-polymers to be tailored and transition state energies to be lowered by 1-2 kcal mol–1, which makes all the difference between a poor and a highly successful catalyst. The increased knowledge about bonding, reactivity in organometallic chemistry has greatly contributed to our understanding of the possible mechanisms of catalysis and this, together with the advances/applications in NMR and IR techniques/ cells described in this volume, has allowed catalytic mechanisms to be much better understood. I hope this volume will transfer some important aspects of NMR and IR, including the use of HP-spectroscopic cells for measurements under actual reaction con-

ditions, to the homogeneous community. I would like to thank all the authors of the chapters for their valiant contributions in this area. However, despite much international effort, the complete spectroscopic identification of all the intermediates in any catalytic cycle has only so far been achieved for two reactions:- the hydrogenation described in chapter 8, which surprisingly involves an intermediate with an agostic-H, and the Pd-catalysed methoxycarbonylation of ethene, which we reported recently, described in chapter 1. So, there is still plenty of opportunity for identification of new mechanisms by people working in this area! Liverpool, January 2004

Brian Heaton

Contents 1

1.1 1.2 1.2.1 1.2.2 1.3 1.3.1 1.3.2 1.3.3 1.4 1.4.1 1.4.2 1.4.3 1.5 1.5.1 1.5.2 1.5.3 1.5.4 1.6 1.6.1 1.6.2 1.6.2.1 1.6.2.2 1.6.3 1.6.3.1 1.6.3.2 1.6.4

2

2.1

NMR Spectroscopy and Homogeneous Catalysis 1 Elosa Martnez Viviente, Paul S. Pregosin, and Daniele Schott Introduction 1 Reaction Mechanisms via Reaction Monitoring 3 Detecting Intermediates 3 Reaction Kinetics via NMR 9 Structural Tools 14 Chemical Shifts 14 Coupling Constants 21 NOE Spectroscopy and 3-D Structure 23 Isotopes in Catalysis 27 Kinetic Isotope Effect (KIE) 28 Structural Effects 29 An Active Site Counting Method 31 Dynamic NMR Spectroscopy 33 Variable Temperature Studies 33 Line Shape Analysis 38 Magnetization Transfer 42 NOESY/EXSY/Hidden Signals 43 Special Topics 50 T1 and Molecular H2Complexes 50 Parahydrogen Induced Polarization (PHIP) 51 Hydrogenation Mechanism Studies 52 Parahydrogen as a Magnetic Probe 55 High Pressure NMR 56 Introduction 56 Applications 56

Diffusion and Pulsed Gradient Spin Echo Measurements 65 References 71 High Pressure NMR Cells 81 Gabor Laurenczy and Lothar Helm Introduction 81


X

Contents

2.2 2.2.1 2.2.2 2.3 2.3.1 2.3.2 2.4 2.4.1 2.4.2 2.5

High Pressure NMR of Liquids 83 High Pressure, High Resolution Probes 83 Glass and Quartz Capillaries 88 High Pressure NMR of Supercritical Fluids 90 High Pressure, High Temperature NMR Probes 90 Toroid Probes for High Pressure NMR 93 High Pressure NMR of Gases Dissolved in Liquids 96 Sapphire Tubes 96 High Pressure Probes for Pressurized Gases 100 Conclusions, Perspectives 102 Acknowledgments 104 References 105

3

The Use of High Pressure Infrared Spectroscopy to Study Catalytic Mechanisms 107 Anthony Haynes Introduction 107 Cell Design 108 Transmission Cells 109 Amsterdam Flow Cell 110 Low-temperature HP IR Cells 111 HP IR Cells for Flash Photolysis 112 Reflectance Cells 114 Mechanistic Studies using High Pressure IR Spectroscopy 117 In situ Studies under Catalytic Conditions 117 Methanol Carbonylation 117 Hydroformylation 123 Other Reactions of Carbon Monoxide 130

3.1 3.2 3.2.1 3.2.1.1 3.2.1.2 3.2.1.3 3.2.2 3.3 3.3.1 3.3.1.1 3.3.1.2 3.3.1.3 3.3.2 3.3.2.1 3.3.2.2 3.3.2.3 3.3.2.4 3.3.2.5 3.3.2.6 3.3.2.7 3.4

4

4.1 4.2 4.2.1 4.2.2

Kinetic and Mechanistic Studies of Stoichiometric Reaction Steps 133 Migratory CO Insertion Reactions of Metal Alkyls 133 Substitution and Exchange Reactions of CO Ligands 138 Exchange between Rh–D and H2 140 Hydrogenolysis of M–C Bonds 140 Mechanistic Studies in Polymer Matrices 141 Noble Gas and H2 Complexes 142 Alkane Complexes and C–H Activation Reactions 144 Conclusions 146 References 147 Processing Spectroscopic Data 151 Marc Garland Introduction 151 The Catalytic System 154 Recycle CSTR with Analytics 154 Physical System 156

Contents

4.2.3 4.3 4.3.1 4.3.2 4.3.2.1 4.3.2.2 4.3.3 4.3.4 4.3.4.1 4.3.4.2 4.3.4.3 4.3.4.4 4.3.5 4.4 4.4.1 4.4.2 4.4.3 4.4.4 4.4.5 4.5 4.5.1 4.5.2 4.5.2.1 4.5.2.2 4.5.3 4.5.3.1 4.5.3.2 4.5.3.3 4.5.3.4 4.5.4 4.6

Chemical Description 157 Experimental Design 159 Transport Time-scales 159 Reaction Time-scales 160 Spectroscopic Measurements 161 Time-scales for Spectroscopic Measurements 162 The Meaning of “In Situ” Studies 163 The Planning of Experiments 164 Batch and Semi-batch 164 Choice of Spectrometers 164 Groups of Experiments 166 Range of Experiments 168 Well-Posedness and Ill-Posedness 168 Data Pre-processing 169 Data Filtering and Outliers 169 Solvent and Reagent Pure Component Spectra 170 Pre-conditioning 171 Track Finding 173 Curve Fitting 174 Spectral Reconstruction 176 Historical Context 176 Entropy Minimization 176 Organometallics 177 Catalysis 179 Band-target Entropy Minimization 180 Multiple-run, Preconditioned, Monometallic Catalytic Data 181 Multiple-run, Preconditioned, Bimetallic Catalytic Data 182 Semi-batch, Non-preconditioned, Monometallic Catalytic Data 184 Semi-batch, Non-preconditioned Data: HRh(CO)4 187 Additional Notes on BTEM 187 Conclusions and Future Directions 188 Acknowledgment 189 References 190

5

Carbonylation of Methanol to Acetic Acid and Methyl Acetate to Acetic Anhydride 195 George Morris Introduction: Evolution of Carbonylation Processes 195

5.1 5.2 5.3 5.4 5.4.1 5.4.2

Some Important Features of Carbonylation Process Chemistry 196 Key Steps in the Mechanism of Carbonylation Processes 199 Information from HP IR and HP NMR for Carbonylation Reaction Studies 201 Studying Carbonylation Mechanisms with IR and HP IR 201 Studying Carbonylation Mechanisms with NMR and HP NMR 204

XI

XII

Contents

5.5 5.5.1 5.5.2 5.5.3 5.6 5.6.1 5.6.2 5.6.3 5.7 5.7.1 5.7.2 5.7.3 5.8 5.8.1 5.8.2 5.8.3 5.9

6

6.1 6.2 6.3 6.4 6.5 6.6 6.7 6.8 6.8.1 6.8.2 6.8.3 6.9

7

7.1

Spectroscopic Studies of Model Reaction Steps of the Rh Carbonylation Cycle 205 Model Studies of Oxidative Addition in the Rh system 206 Model Studies of Migratory Insertion in the Rh System 207 Model Studies of Reductive Elimination in the Rh System 208 Spectroscopic Studies of the Model Reaction Steps of the Ir Carbonylation Cycle 209 Model Studies of Oxidative Addition in the Ir System 209 Model Studies of Migratory Insertion in the Ir System 210 Model Studies of Reductive Elimination in the Ir System 211 Spectroscopic Studies of the Organic Cycles of Carbonylation Reactions 212 NMR Studies of Ac2O and AcI Hydrolysis 213 HP NMR Studies of the Reaction of AcI with MeOAc in Anhydrous Media 214 HP NMR Studies of the Reaction of HI with MeOAc in Aqueous Media 218 Spectroscopic Studies of Working Carbonylation Reactions 222 Rh Catalysed Carbonylation of MeOAc to Ac2O 223 Rh Catalysed Carbonylation of MeOH to AcOH 224 Ir Catalysed Carbonylation of MeOH to AcOH 226 Conclusions: Spectroscopy and Understanding Carbonylation Mechanisms 228 Acknowledgments 228 References 229 Rhodium Catalyzed Hydroformylation 231 Paul C. J. Kamer, Joost N. H. Reek, and Piet W. N. M. van Leeuwen Introduction 231 Study of Catalytic Resting States 233 IR studies on Ligand-free Rhodium Carbonyl Catalysts 237 Phosphite Ligands 239 Diphosphite Ligands 244 Dimer Formation 250 Study of Bulky Phosphorus Diamide Ligands 252 Study of the Elementary Steps of the Catalytic Cycle 260 CO-dissociation 260 Exchange between RhD and H2 262 Hydride Migration 265 Conclusions 267 References 267 Alkene/CO Copolymerisation 271 Claudio Bianchini and Andrea Meli Introduction 271

Contents

Catalytic Cycles of Alkene/CO Copolymerisation 274 Mechanism of Ethene/CO Copolymerisation 274 Methanol and Other Protic Solvents 274 Aprotic Solvents 276 Formation of Active PdII Sites and Initiation of Ethene/CO Copolymerisation 277 7.2.1.4 Chain Propagation of Ethene/CO Copolymerisation 279 7.2.1.5 In Situ High Pressure NMR Studies of Ethene/CO Copolymerisation in Protic Solvents 280 7.2.1.6 In Situ High Pressure NMR Studies of CO/Ethene Copolymerisation in Aprotic Solvents 282 7.2.1.7 In Situ High Pressure IR Studies of Ethene/CO Copolymerisation 284 7.2.1.8 Model Synthetic Studies of Ethene/CO Copolymerisation 286 7.2.1.9 Kinetic and Thermodynamic Studies of Ethene/CO Copolymerisation in Aprotic Solvents 288 7.2.1.10 Chain Transfer in Ethene/CO Copolymerisation 292 7.2.2 Mechanism of Styrene/CO Copolymerisation 297 7.2.3 Mechanism of Propene/CO Copolymerisation 301 7.2.4 Mechanism of Cyclic Alkenes/CO Copolymerisation 302 7.2.5 Mechanism of Polar Alkenes/CO Copolymerisation 304 7.2.6 Catalyst Deactivation Paths 305 References 306 7.2 7.2.1 7.2.1.1 7.2.1.2 7.2.1.3

8

8.1 8.2 8.3 8.4 8.5 8.6 8.7 8.8 8.9 8.10

9

9.1 9.2 9.3 9.3.1 9.3.2

The Use of Spectroscopy in Metallocene-based Polymerisation Catalysis 311 Manfred Bochmann Introduction 311

Identification of the Catalytically Active Species: The Chemistry of Group 4 Metal Methyl Species 313 Activation of Alternative Group 4 Catalyst Precursors with B(C6F5)3 319 Olefin Coordination to d0 Metal Centers 323 Ion Pair Dynamics in Metallocene Catalysts 328 Monomer Coordination 333 Polymerisation Kinetics 335 Spectroscopic Studies on Complex Systems 339 Spectroscopy of Poly(1-alkenes): Polypropylene 344 Conclusion 352 References 353 Hydrogenation 359 Ralf Giernoth Introduction 359

The Dihydrogen Molecule 360 Rhodium-catalyzed Homogeneous Hydrogenation, the Basics 360 The Achiral Case 360 The Chiral Case 361

XIII

XIV

Contents

9.3.2.1 9.3.2.2 9.4 9.4.1 9.4.2 9.4.2.1 9.4.2.2 9.4.2.3 9.4.2.4 9.5 9.5.1 9.5.2 9.5.3 9.5.4 9.6

Catalysts 361 Hydride Route versus Unsaturate Route 362 Spectroscopic Methods 363 “Standard” NMR Spectroscopy 363 PHIP-NMR-Spectroscopy 365 Ortho- and Parahydrogen 365 Parahydrogen Induced Polarization and In Situ Spectroscopy 366 ALTADENA and PASADENA 367 Applications of the PHIP Method 370 Studying the Catalytic Mechanism of Enantioselective Hydrogenation with NMR Spectroscopic Methods 371 Early NMR Experiments: Wilkinson’s Catalyst Revisited 371 The Mechanistic Model for the Chiral Case 371 Detection of Intermediates with Standard NMR Spectroscopy 373 The “Breakthrough”: Characterization of the Key Intermediates with PHIP NMR Spectroscopy 374 Conclusion and Outlook 377 References 377 Index 379

List of Contributors Claudio Bianchini Institute of Chemistry of Organometallic Compounds, ICCCOM-CNR Via J. Nardi 39 50132 Firenze Italy Manfred Bochmann Wolfson Materials and Catalysis Centre School of Chemical Sciences and Pharmacy University of East Anglia Norwich, NR4 7TJ United Kingdom Marc Garland Department of Chemical and Biomolecular Engineering 4 Engineering Dr 4 National University of Singapore Singapore 117576 Singapore Ralf Giernoth Institute of Organic Chemistry University of Cologne Greinstr. 4 50939 Kln Germany

Anthony Haynes Department of Chemistry University of Sheffield Dainton Building Brook Hill Sheffield, S3 7HF United Kingdom Lothar Helm Institute of Chemical Sciences and Engineering, ISIC Ecole polytechnique fdrale de Lausanne BCH 1015 Lausanne Switzerland Paul C. J. Kamer van‘t Hoff Institute for Molecular Sciences University of Amsterdam Nieuwe Achtergracht 166 1018 WV Amsterdam The Netherlands Gbor Laurenczy Institute of Chemical Sciences and Engineering, ISIC Ecole polytechnique fdrale de Lausanne BCH 1015 Lausanne Switzerland


XVI

List of Contributors

Piet W. N.M. van Leeuwen van t Hoff Institute for Molecular Sciences University of Amsterdam Nieuwe Achtergracht 166 1018 WV Amsterdam The Netherlands

Paul C. J. Reek van t Hoff Institute for Molecular Sciences University of Amsterdam Nieuwe Achtergracht 166 1018 WV Amsterdam The Netherlands

Andrea Meli Institute of Chemistry of Organometallic Compounds, ICCCOM-CNR Via J. Nardi 39 50132 Firenze Italy

Daniele Schott Laboratory of Inorganic Chemistry ETHZ HCI Hnggerberg 8093 Zurich Switzerland

George Morris BP-Amoco Chemicals Saltend Hull HU12 8DS United Kingdom Paul S. Pregosin Laboratory of Inorganic Chemistry ETHZ HCI Hnggerberg 8093 Zurich Switzerland

Elosa Martnez Viviente Laboratory of Inorganic Chemistry ETHZ HCI Hnggerberg 8093 Zurich Switzerland

1 NMR Spectroscopy and Homogeneous Catalysis Elosa Martnez Viviente, Paul S. Pregosin, and Daniele Schott

1.1

Introduction

NMR spectroscopy continues to develop and refine new techniques and, consequently, has become an indispensable tool in connection with solution studies in the area of homogeneous catalysis. Since the soluble catalyst precursors in transition metal catalyzed reactions often contain a variety of atoms from differing parts of the Periodic Table, it is not surprising that a multinuclear NMR approach plays an important role. Although 1H NMR is still prevalent, 13C (for metal carbonyls, metal carbenes, metal acyls…etc.), 31P (for phosphine-based catalysts), 19F (for fluorous catalysts), and occasionally even the metal center itself, are all routinely measured. Admittedly, the NMR approach to obtaining these data no longer relies solely on 1-D measurements. A variety of two-dimensional methods, using either one, two or three-bond coupling constants allow the data to be accessed in a much more efficient manner. As examples, Figure 1.1 shows: (a) part of the 13C,1H correlation for the Ru(ii) dialkyl, p-arene compound, 1 [1], showing the relatively low frequency positions of the three-coordinated biaryl CH-resonances; (b) a section of the long-range carbon-proton correlation for the zirconium polymerization catalyst precursor, 2, using 3J(13C,1H) [2]. The observed d values suggest some p-interaction from the five-membered ring; (c) the use of 3J(31P,1H) in the chiral Pd(ii) Duphos complex, 3 [3]; and (d) the 103 Rh resonance, for the Biphemp-based hydrogenation catalyst precursor, 4, via 3 103 J( Rh,1H) [4]. In addition to chemical shifts, d, and coupling constants, J, relaxation time data, e. g., T1, and, increasingly, diffusion constants, D, are being used to help solve selected structural problems. Although line widths in connection with variable temperature studies are still used to calculate rate constants for processes involving catalysts or intermediates which are dynamic on the NMR time scale, magnetization transfer and, especially, phase sensitive nuclear Overhauser effect (NOE) methods are often the method(s) of choice. These NMR techniques are somewhat more demanding; nevertheless, they are finding increasing acceptance. Further, additional Mechanisms in Homogeneous Catalysis. A Spectroscopic Approach. Edited by Brian Heaton Copyright c 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN 3-527-31025-8

2

1 NMR Spectroscopy and Homogeneous Catalysis

C,1H correlation for 1, R1 ¼ CH3 , R ¼ 3,5-di-tertbutylphenyl, showing the relatively low frequency positions of the three coordinated biaryl CHresonances. The open and closed cross-peaks reflect the phases. (b) Section of the 13C,1H longrange correlation for Zr ansa-fluorenyl-complex, 2, showing the assignment of the fluorenyl carbons, C-7 and C-12. Each of these carbons shows two correlations stemming from the values 3 13 1 J( C, H). (c) 31P,1H correlation showing the cross-peaks which help to identify the two terminal allyl protons of [Pd(h3 -PhCHCHCHPh)(Me-Duphos)](OTf), 3. These terminal allyl protons, which correlate to their respective pseudo-trans P-atoms, appear as triplets (similar 3J(31P,1H) and 3 1 J( H,1H) values) further split by long-range proton–proton and proton–phosphorus interactions. (d) The 103Rh,1H correlation for 4, showing selective contacts to the two olefinic protons at d 4.31, and d 4.72 (one stronger than the other) and an aliphatic proton of the 1,5-COD (there is also a very weak aliphatic second contact). The multiplicity in the rhodium dimension arises due to the two equivalent 31P atoms coordinated to the rhodium. Figure 1.1 (a) Section of the phase-sensitive

13

1.2 Reaction Mechanisms via Reaction Monitoring

tools such as: (a) parahydrogen-induced polarisation (PHIP), which may allow one to detect species present in solution at relatively low concentration, (b) high pressure measurements, which simulate catalytic conditions, and (c) NOESY measurements, which allow the determination of 3-D solution structures, are all slowly moving from the hands of the NMR specialist to the practising catalyst chemist. In the following pages we will try to illustrate and summarise some of the more relevant applications of all of these methods.

1.2

Reaction Mechanisms via Reaction Monitoring

Following the course of a reaction by NMR remains one of the most popular applications of this technique in homogeneous catalysis. The resulting kinetic information and/or the detection and identification of intermediates are important sources of mechanistic information. Often, isotopic labeling with 2H [5–12] or 13C [13–15] facilitates the acquisition and interpretation of the resulting NMR spectra. 1.2.1

Detecting Intermediates

A number of compounds can be recognized in the hydrogenation of 13C labeled MAC (methyl-(Z)-a-acetamidocinnamate, 50 % 13C in the a-olefin carbon) using the model Rh catalyst, [Rh(diphos)(MeOH)2]þ, 5 [16, 17]. Figure 1.2 shows the proposed catalytic cycle and the most relevant sections of the various NMR spectra of 6 and 7. The 31P spectrum of 6 was measured at 233 K and shows the 13C satellites for PA (the larger 31P–13C coupling is associated with the trans-geometry). The 13C NMR spectrum of the a-carbon of 7 (intercepted at –78 oC) clearly reveals that H transfer during the migratory insertion step occurs at the b-carbon atom of the C¼C bond, leaving the a-carbon atom bonded to the Rh (1J(103Rh,13C(a)) ¼ 21 Hz). The increased 13C S/N and additional spin–spin interactions provided by the 13C labeling are important for the assignment. Monitoring studies on the last step of the cycle via 31P and 1H NMR allowed the determination of a firstorder rate law and the activation parameters. In a related study using the chelating phosphine chiraphos, several species, 8–10, were recognized by 31P NMR (see Figure 1.3) [18]. Only a single diastereomer, 10, forms, indicating that the binding is stereospecific. The catalytic cycle for the Rh-catalyzed 1,4-addition of phenylboronic acid to an a,b-unsaturated ketone could be nicely described by in situ 31P NMR (see Figure 1.4) [19]. The three {S}-Binap species RhPh(PPh3)(Binap), 11, Rh(oxaallyl)(Binap), 12 and [Rh(OH)(Binap)]2, 13, have all been detected. Complex 11 affords a modestly complicated spectrum (see spectrum A), due to the ABMX spin system. The oxa-allyl complex, 12, in spectrum B exists in two diastereomeric forms (with overlapping signals between 48 and 49 ppm). The resonances for the bridging hydro-

3

4


Figure 1.2 Catalytic cycle for the Rh-catalyzed hydrogenation of methyl-(Z)-a-acetamidocinna-

mate, (50 %

13

C in the a-C, denoted by *) in MeOH.

Figure 1.3 Stable inter-

mediates in the enamide hydrogenation by (S,S)-transbis(2,3-diphenylphosphinobutane)rhodium, detected by 31P NMR. The various multiplicities arise from 1 103 J( Rh,31P) and 3J(31P,31P).


Figure 1.4 Mechanistic aspects of the Rh-catalysed 1,4-addition of phenylboronic acid to an a,b-unsaturated ketone, monitored by 31P NMR.

xide 13, with equivalent P-atoms, are observed, in spectrum C, at ca. 55 ppm. The starting PPh3 complex, 11, is regenerated in spectrum D. In the zirconocene-catalyzed polymerization of alkenes, Landis and coworkers [20] have reported in situ observation of a Zr-polymeryl species, 15, at 233 K (Figure 1.5). Complex 15 is formed by partial reaction of 14 with excess 1-hexene. Derivatives 16 and 17 are generated quantitatively from 15 by addition of ca. 10 equiv. of propene and ethene, respectively. No other intermediates, such as alkene complexes, secondary alkyls, diasteromers of 15 or 16, or termination products, accumulate to detectable levels. These NMR studies permit direct monitoring of the initiation, propagation and termination processes, and provide a definitive distinction between intermittent and continuous propagation behavior. Espinet and coworkers [21] have captured an NMR “snapshot” of a catalytic cycle for the Stille reaction, involving compounds 18–22 (see Figure 1.6). The vinylic region of the 1H and 1H{31P} NMR spectra of 19–22 is shown. Both Pd(ii) Pd(vinyl)R(PP), 20 and Pd(0) Pd(RCH¼CH2)(PP), 21 species were identified. Brown and coworkers [22] have studied the Pd-catalyzed Heck arylation of methyl acrylate via 31P and 13C NMR (see Figure 1.7). Reaction of the aryl iodide complex 23 with AgOTf (THF, 195 K) gives the THF and aquo-complexes 24 and 25, respectively, which were detected via 31P NMR below 203 K. Addition of H2O to the sample shifts the equilibrium towards 25, pointing to an existing fast exchange between solvates 24 and 25. Reaction of 24/25 with 3-{13C} labeled methyl acrylate (20-fold excess, 213 K) affords the insertion product 26. Warming to 233 K leads to the formation of 27, which is in turn converted into 29, stable to

5

6


Figure 1.5 In situ Detection of Zr-polymeryl species by 1H NMR (233 K, d8 -toluene). Only the region between 0 and -1.3 ppm is shown, in which resolved resonances for the diastereotopic Zr–CH2 –POL protons and the Zr–Me–B groups are detected.

273 K. The rearrangement from 27 to the more stable, primary alkyl regioisomer 29 was shown to be intermolecular. This is thought to occur via the unobserved hydride 28, since addition of unlabeled methyl acrylate leads to the equilibrium distribution of 13C label in 29 (28 exchanges acrylate ligands with the acrylate pool). Using 31P NMR the same authors have identified several intermediates in the asymmetric Heck arylation of dihydrofuran [23, 24]. Reaction of the Binap salt, 30, with 2,3-dihydrofuran below 233 K gave salt 31 as the single species (see Figure 1.8). A parallel reaction between 30 and [2- 2H] 2,3-dihydrofuran confirmed the structure. At 243 K, 31 slowly decomposed to form 32 and 32’ with concomitant release of the coupling product 33 (91 % ee).


Figure 1.6 Catalytic cycle for a Stille reaction showing the vinylic regions of the 1H and 1H{31P} NMR spectra of the products detected in situ (in d8 -THF).

Figure 1.7 NMR investigation on the Heck reaction. The 31P NMR spectra on the right show the low-frequency region.

7

8


31 P NMR spectra of the reaction sequence between 30 and 2,3-dihydrofurane. Spectrum A: partial conversion of 30 to 31 at 223 K. Spectrum B: After complete formation of 31 at 233 K. Spectrum C: nearly complete decomposition of 31 at 243 K to form 32 and 32’. The signals at higher frequency correspond to Pa.

Figure 1.8

For the industrially important Pd-catalyzed methoxycarbonylation of ethene to methyl propanoate, all the intermediates of the cycle have been identified. Starting from 34, 13CH2¼CH2 and 13CO, the process has been shown to proceed via a hydride rather than a methoxycarbonyl cycle (Scheme 1.1) [25][26]. Figure 1.9 shows the 31P NMR spectrum at 193 K of a 1:1 mixture of the two isotopomers 35a and 35b, formed in the reaction of 34 with 13CH2¼12CH2. The presence of an agostic interaction is supported by the 13C chemical shifts (d(CH2) 31, and d(CH3) 8), which are reversed with respect to classical Pd-ethyl complexes.

O OMe (tBu2) P

Pd H

OTf

34

H

Pd O P (tBu2) H

MeOH

Me Pd

CO

Pd OMe

Pd

Pd 35

O

OMe O

OMe Pd O

34

CO Hydride Cycle Scheme 1.1

Methoxy Cycle

The two possible mechanisms for the Pd-catalysed methoxycarbonylation of ethene.


31

P NMR spectrum at 193 K of a 1:1 mixture of the two isotopomers [Pd(L–L)(CH213CH3)]þ, 35a, and [Pd(L–L)(13CH2CH3)]þ, 35b, formed in the reaction of 34 with 13CH2¼12CH2. For 35a, the phosphorus trans to the ethyl group (PB d 36.1) couples with the cis-phosphorus PA [d 68, 2J(31PA,31PB) 31 Hz], while the 2J(31PB,13C) is not observed due to the low natural abundance of 13C. In 35b, however, the coupling of PB with both PA and the labelled trans carbon atom [2J(31PB,13C) 38 Hz] can be observed, resulting in a doublet of doublets, which is superimposed with the doublet originating from 35a. Figure 1.9

In situ NMR studies on analogous Pt catalysts for the methoxycarbonylation reaction reveal CO trapping at every step in the catalytic cycle of the active intermediates (Figure 1.10) [27]. This explains the observed slow kinetics. Thus, 36 reacts with 13CO in CH2Cl2 at 193 K to form only [Pt(L–L)(C2H5)(13CO)]þ, 37, which upon warming to ambient temperature in the presence of excess CO affords [Pt(L–L)(13C(O)Et)(13CO)]þ, 38. This transformation is reversible, and both compounds have been detected by in situ 13C{1H} NMR spectroscopy. 1.2.2

Reaction Kinetics via NMR

The previous section focused on the detection of intermediates in a catalytic reaction, thereby affording an “NMR picture” of the several steps involved in the mechanism. Occasionally, NMR can be a convenient tool for monitoring reaction rates provided that the reaction is slow enough for a series of 1D spectra to be acquired during its course.

9

10


Figure 1.10 Pt-catalysed methoxycarbonylation of ethene, studied by

13

C NMR.

Figure 1.11 provides an example of 1H NMR monitoring in the Pd-catalyzed cycloisomerization of dimethyl diallyl malonate, 39 [28]. The kinetic profile reveals a pronounced induction period after which the exocyclic alkene 40a is formed predominantly as the kinetic product. A hydropalladation mechanism was proposed on the basis of NMR experiments, and the transient species 41, formed by allylpalladation of the coordinated diene, could be detected and identified with the help of 2 H and 13C labeling. The hydride Pd catalyst, 42, would be generated from 41 by water-promoted b-hydride elimination. The observed induction period is associated with the formation of the Pd-hydride 42. In the oxidative addition of a fluorinated aryl iodide, 43, to “Pd(PPh3)2” (Figure 1.12) [29], 19F NMR has been used to follow the cis-to-trans isomerization of the cis-bis-phosphine product, 44, to the trans-isomer, 45. The 19F NMR kinetic study reveals a first order dependence for the rate of isomerization on the concentration of 44. An application of a 19F NMR kinetic study to the evaluation of the


Figure 1.11 Pd-catalysed cycloisomerisation of dimethyl diallyl malonate. Kinetic profile based on 1

H NMR, and proposed reaction mechanism.

Figure 1.12 19F NMR study of the cis-to-trans isomerization of 44 to 45. Only the ortho 19F resonances are shown. In 44, the coupling with two inequivalent 31P atoms affords a doublet of doublets. For 45, the spectrum consists of a triplet.

11

12


Pd(OAc)2 R NaOR + HNRR'

N R'

P

Y

Br

Pd P 46

Y

Y Pd R'N R

P

P

P

P

NaBr+ HOR Scheme 1.2

P

Y

P

Pd Br

NaOR+ HNRR'

Pd-catalysed amination of aryl halides using the chiral ligand Binap.

factors contributing to the “copper effect” in the Stille reaction has also been reported [30]. In the Pd-catalyzed amination of aryl halides using Binap, the Pd(0) complex Pd(Binap)2, 46, has been identified by 31P NMR as the resting state in the catalytic cycle (Scheme 1.2) [31]. The zero-order dependence of the reaction rate on the amine concentration has been confirmed via a 1H NMR study with primary amines (Figure 1.13, left). For secondary amines, however, a first-order dependence on amine was apparent (Figure 1.13, right), suggesting a change in the resting state of the catalyst to one that would react with the amine. 31P monitoring of the catalyst concentration (Figure 1.13, center) showed a gradual consumption of 46 in the reaction with the secondary amine, but not with the primary, explaining the different kinetic behavior. We note that there are NMR-based kinetic studies on zirconocene-catalyzed propene polymerization [32], Rh-catalyzed asymmetric hydrogenation of olefins [33], titanocene-catalyzed hydroboration of alkenes and alkynes [34], Pd-catalyzed olefin polymerizations [35], ethylene and CO copolymerization [36] and phosphine dissociation from a Ru-carbene metathesis catalyst [37], just to mention a few. Finally, an example of reaction monitoring with a “rare” nucleus: Figure 1.14 reproduces three sequences of 11B NMR spectra of the Zr-catalysts 47-49/MAO (MAO ¼ methylaluminoxane) during the polymerization of ethylene [38]. No changes are detected in the systems 47/MAO (a) and 49/MAO (c) during the course of reaction; however, for 48/MAO in (b), a new 11B signal appears. This is attributed to an exchange of the boron benzyloxy substituent of 48 with the methyl from the MAO, effectively transforming 48 into 49. This transformation of the catalyst is thought to explain why the selectivity of the 48/MAO system


for a-olefin production is intermediate between that of the similar 47/MAO (more than 99 % a-olefins) and 49/MAO (which reacts with the 1-alkenes and converts them to other isomers).

Decay of the concentration of a primary amine (hexylamine, left) and a secondary amine (N-methylaniline, right) during the reaction with BrC6H5 catalyzed by 46, followed by 1 H NMR. The reaction is zero-order in the primary amine (linear concentration decay) and firstorder in the secondary amine (nonlinear concentration decay). Center: 31P NMR spectra acquired during the reaction of aniline (left) and N-methylaniline (right) with BrPh catalyzed by 46. P(o-Tol)3 is used as internal standard. Figure 1.13

Figure 1.14

A

11

B NMR study on the Zr-catalyzed polymerization of ethylene.

13

14


1.3

Structural Tools

In the studies on detecting intermediates many of the structural conclusions drawn were based on our (often empirical) understanding of chemical shifts and spin–spin interactions. In the following two sections we show a selection of these in connection with (mostly) catalytically relevant organometallic molecules. 1.3.1

Chemical Shifts

The 1H signals for transition metal hydrides, MH(L)n, M ¼ Ru, Rh, Ir, Pt etc., afford very low frequency 1H resonance positions, usually in the range d ca. –5 to –30. Several values for palladium hydrides, 50, (often postulated in catalysis, but rarely observed) are shown below [39].

L

H Pd L

Cl

L

PEt3

-13.6

PCy3

-14.4

PPh3

-13.2

50

[RhH(Cp*)(Binap)](SbF6), a presumed intermediate in the hydrosilylation of phenyl acetylene [40], shows the hydride resonance at d –10.39. Hydride resonances in Ru(ii) phosphine complexes are often found in the same region [41–43]. Agostic interactions, i. e., the three-center bonds related to structure 51 [26, 44– 49], were noted earlier by Green and Brookhart and have been cited above in the methoxycarbonylation chemistry (Figure 1.9). These bonds are often characterized by low frequency (hydride-like) proton chemical shifts, and/or substantially reduced 1J(13C,1H) values. Often, it is necessary to cool the NMR sample in order to “freeze” the equilibrium. Complex 52 represents a nice example of an agostic C–H bond, with relevance to polymerization chemistry [47].

H C

M

51 M = transition metal

1.3 Structural Tools

BArF

R N

N R Pd

H

H

H3C

broad triplet H

H

-8.00 1

J(13C,1H) = 65 Hz at 158 K

52

The 1H signals of carbene-based metathesis catalysts can be found at relatively high frequency, e. g., d 19.55 in 53 [50], or d 12.76 in one isomer of 54 [51]. 19.55 H C Cl

IMes Ru

MesI

Cl 53

IMes = NN-bis(mesityl)imidazol-2-ylidene

t-Bu

i-Pr N

O

i-Pr

Mo O t-Bu

54

Ph

C H Me

Me

12.76

With respect to 1H chemical shifts, two additional general points are worth noting: (a) due to local anisotropic effects, protons in cis-position to pyridine or phenyl phosphine ligands afford low frequency 1H signals (Structures 55 and 56), and (b) not so well recognized, but still useful, is the fact that square planar metal complexes have very anisotropic regions above and below the coordination plane, e. g., in 57, the proton moves to higher frequency. This positioning can result in a weak bond to the metal [52, 53].

15

16


H

P

M N

M

L

M

L

H

L

H

55

56

57

For the heavier nuclei, such as 13C and 31P, where both the nature of the bonding and the corresponding energy level considerations determine the chemical shifts [54–56], the chemical shift range (and variation with structure) is much larger than for protons. The 13C resonances for the Ru(ii) carbene atoms in 58 [57] appear at very high frequency. Interestingly, for the Ir(iii) bis-carbene complex 59 [58], thought to be involved in transfer hydrogenation chemistry, the observed 13C position for the carbene atom is only ca. 127 ppm. Cl

NR

P

P

O

Ru P

N N

P

C

Ir O

CHR 58

I

NR

carbene 327-358

I 59

carbene 127

P P = Ph2PCH2CH2PPh2

The carbon of complexed CO, i. e., M–CO, can appear at either a lower or higher frequency than CO itself, depending on the metal. A useful list of 13CO chemical shifts can be found in a study describing mechanistic aspects of the Rh- and Ir-catalyzed carbonylation of methanol [59]. Additional 13C NMR data on Rh-acyl intermediates, derived from the Rh-catalyzed carbonylation of ethene, e. g., 60, have been reported [60]. I

CO I

Rh

Et(O)C

I

Rh

I I

C(O)Et I

CO 60

δ 214.8 (C(O)Et) and 182.7 (CO)

An aliphatic metal–carbon sigma bond, e. g., h1 M–CH3, affords a fairly small 13C chemical shift, sometimes at negative d values. The relatively rare h1 allyl derivative, PdCl(h1 allyl)(PHOX), 61 [61], which is related to allylic alkylation chemistry, shows


a normal low frequency sigma bound methylene carbon resonance, thereby clearly indicating that the bonding is not of the usual h3 type.

O N

PPh 2 Pd

i-Pr

Cl Ph Ph

δ 25. 8 61

The rather novel hydroxy MOP-derived Pd(ii) complex 63 [62], derived from the MOP chiral auxiliary 62, reveals an unexpected s-bond with the fully substituted coordinated carbon at d 70.5, thus indicating that this carbon is no longer aromatic.

OH

O Pd(acac)

δ 70. 5

PPh2

PPh2

62 OH-MOP

63

In a related fashion, the 13C resonance at d 79.4 from the Zr(ansa-Fluorenyl) polymerization catalyst precursor 64, is consistent with substantial sp3 character at this carbon [2].

Ph Ph

ZrCl2

δ 79.4 64

However, the 13C chemical shift for a s-bound M–aryl bond appears at high frequency, often between 130 and 180 ppm. Both the cyclometallated nitrosoamine compound 65 [63–65] and the Pd(ii)(Duphos) complex 66 (an intermediate in the enantioselective hydroarylation of norbornene) [66] represent typical examples.

17

18

1 NMR Spectroscopy and Homogeneous Catalysis 2

J(31P,13C) = 138 Hz

CH3 Br

Me N N O Pd

P H3C

Pd CN P

δ 151.5

δ 174

CH3

H3C

65

66

The 13C signals for coordinated olefin ligands are often (but not always) markedly shifted to low frequency. The position of these formerly sp2 carbons depends strongly on the p-back-bonding characteristics of the metal, together with the donor characteristics of the remaining ligands in the coordination sphere. The Ru(ii) compound 67, in which the MeO–Biphep is acting as a 6e donor, shows the complexed olefinic 13C signals at d 66.4 and d 86.5, in the region expected for a coordinated double bond (despite the relatively long Ru–C distances found in the solid state) [67].

Ph2 P + Ru

MeO MeO

Cp

P Ph2

67 δ (C=C) 66.4 and 86.5

The two isomeric Pd(0)(dba)(phosphino-oxazoline) Heck catalyst precursors 68 and 69 show the two olefinic resonances at d 56.0 and d 69.3, plus d 56.0 and d 67.3 [68].

O O

Ph i-Pr

N P Ar2

Pd O

N

i-Pr

P Pd Ar2

Ph

Ph 68

O

69 Ph


C resonances for complexed h6 -arene moieties, e. g. Ru(ii)(h6 -p-cymene or h6 benzene) are often found between 60 and 100 ppm [69–77]. 31 P represents a favored NMR nucleus when phosphine precursors are used in catalysis. For tertiary phosphine complexes the normal chemical shift range is several hundred ppm. There are a few useful empiricisms: 13

1. The coordination chemical shift, Dd ¼ d(complex) –d(ligand), is often fairly large and positive. Via integration of the 31P signal it is possible to determine the number of complexed ligands in a catalyst precursor. In their enantioselective C–C bond-making catalytic studies, Tomioka and coworkers [78] have used this approach to show that Rh(acac)(C2H4)2 reacts sequentially with one, two and three equivalents of 70 to form different materials. The first equivalent affords a P,O-chelate complex, the second a bis-phosphine derivative and the third equivalent of phosphine is not complexed to rhodium.

N PPh2 t-Bu

O 70

2. Binding strongly electron withdrawing groups to the P-atoms, e. g., several P–F or P–O bonds, usually shifts the 31P signal to higher frequency. Occasionally this can happen when P–C bonds are cleaved, as in Scheme 1.3, where the coordinated PFPh2 ligand appears at d 181.7, more than 100 ppm away from the aryl phosphine [79]. [Ru(p-cymene)(Binap)](SbF6)2 Bu4NF δ 50.9 3 31

J( P,19F) = 13 Hz

PPh2 Ru(PFPh2)(p-cymene)(SbF6) δ 181.7 1 31

J( P,19F) = 936 Hz

Shift to higher frequency of the P chemical shift in a PFPh2 ligand with respect to an aryl phosphine.

Scheme 1.3 31

3. Including the P-atom in one or more five-membered rings (chelation or cyclometallation) strongly moves the 31P signal to high frequency. The Ru(ii) acetalization catalyst 71 [80] demonstrates the chelation principle in that the central

19

20


P-donor appears at much higher frequency, 96.9 ppm, than the two equivalent terminal P-ligand atoms. There are many more useful details and the reader would do well to consult a suitable review [81]. δ 96.9 (OTf)2

P NCMe

P δ 59.2

Ru NCMe

P NCMe 71

P P = PhP(CH2CH2P{p-FC6H4)2})2

Metal chemical shifts have not found extensive use in relation to structural problems in catalysis. This is partially due to the relatively poor sensitivity of many (but not all) spin I ¼ 1/2 metals. The most interesting exception concerns 195Pt, which is 33.7 % abundant and possesses a relatively large magnetic moment. Platinum chemistry often serves as a model for the catalytically more useful palladium. Additionally, 195Pt NMR, has been used in connection with the hydrosilylation and hydroformylation reactions. In the former area, Roy and Taylor [82] have prepared the catalysts Pt(SiCl2Me)2(1,5-COD) and [Pt(m-Cl)(SiCl2Me)(h2 -1,5-COD)]2 and used 195Pt methods (plus 29Si and 13C NMR) to characterize these and related compounds. These represent the first stable alkene platinum silyl complexes and their reactions are thought to support the often-cited Chalk–Harrod hydrosilylation mechanism. Philipsborn and coworkers [83] have successfully used the 59Co signals in the substituted Co(i)Cp complexes 72, in connection with understanding the mechanism of pyridine/acetylene trimerization reactions. The metal resonance was found to vary strongly with the catalyst structure and a correlation of d 59Co with reactivity was observed. R Co(1,5-COD) 72

The groups of Philipsborn [84], Heaton [85–88] and Mann [89–91] have used Rh NMR extensively to elucidate structural and mechanistic aspects of a wide variety of metal carbonyl and metal cluster complexes. Further, Zamaraevl [92] has shown that NMR studies on several quadrupolar nuclei, e. g. 95Mo, help with the characterization of the alkyl peroxo-complexes, which are thought to be inter103


mediates in the course of the homogeneous epoxidation of cyclohexene and oxidation of cyclohexane. Clearly, metal NMR in catalysis remains promising, but relatively undeveloped. 1.3.2

Coupling Constants

All of the many one, two and three-bond interactions in organic molecules, e. g., the various Karplus relations involving 3J(X,Y) as a function of dihedral angle, or the 3J(13C,1H) interaction used to find fully substituted 13C signals in Figure 1.1 (b), find wide ranging applications. However, if we center on those spin–spin interactions that directly involve the transition metal, then 1J(M,L) and 2J(L1–M–L2) (L ¼ a donor atom) have proven, generally, to be the most helpful in terms of defining the local coordination sphere. In square planar and octahedral complexes, 73, 1 J(M,L) depends on the trans-ligand, with stronger donors reducing the1J(M,L) value. L Pt or Rh trans ligand 73 L = 1H or 13C or 31P etc.

This empiricism (which concerns the concept of trans-influence) [93] is derived from theory and the Fermi contact term. Since the s-component of the M–L bond determines the magnitude of 1J(M,L), bonding considerations which decrease the s-component, e. g., a relatively strong s-bond in the trans-position, decrease 1 J(M,L). In Wilkinson’s catalyst, RhCl(PPh3)3, (and the p-tolyl analog) the two 1 103 J( Rh,31P) values are quite different: 189 Hz (P trans to Cl) and 142 Hz (P trans to P) [94]. The larger 1J-value arises from the P-atom trans to the weaker donor. For the second and third transition series, the values 2J(L1-M-L2) depend strongly on geometry, as indicated in 74 and 75, with the trans-interactions being normally much larger than those for the corresponding cis-compounds.

M

L

L

31

M

31

P

P

74 2J(31P-M-L trans)

75 >

2J(31P-M-L ) cis

L = 1H or 13C or 15N or 31P

21

22


A specific example can be found in the Rh-chemistry mentioned in connection with Figure 1.4. The complex RhPh(PPh3)(Binap), 11, shows very different 2J(31P,31P) values with the trans-interaction much larger than the cis. Taken together, the three different 31P chemical shifts and the various 1J(103Rh,31P) and 2J(31P,31P) values are all important indicators of the correct structure for this complex.

P

2 31

J( P,31P) trans = 324 Hz

P Rh

2 31

J( P,31P) cis = 30 Hz

PPh3

Ph 11

The cationic Ru(ii) phosphine“pincer” complex 76 [95], which contains an agostic Ru–H–C bond, possesses four different P-atoms and thus demonstrates the geometric dependence of 2J(31P,31P). The fairly small 1J(13C,1H) value, 112 Hz, helps in the recognition of the agostic bond and is 46 Hz smaller than in the free ligand. Ph2 P Ph2 P Ru H

OTf

PP Ph2 Ph

2

76 31

2

31

J( P, P)trans = 345.7 Hz and 233.3 Hz

2

J(31P,31P)cis = 34.6 Hz, 26.7 Hz and 27.1 Hz

We note that in the Palladium salt 35 (Figure 1.9) a large 2J(31P,13C)trans value is reported. An important consequence of the often relatively large 2J(31P,31P) value is that 13C spectra of bis-phosphine complexes are often second order, e. g., the ¼CH signals for the two isomers of [Pd(NCCH¼CHCN)(Me-Duphos)], 77, see Figure 1.15 [3]. The figure shows that these absorptions appear as complicated multiplets (i. e., the X-part of an ABX spin system). The cationic Pt(ii) hydride 78 [27], a model for the analogous Pd intermediate which is thought to be involved in the methoxycarbonylation of ethene (see Scheme 1.1 and Figure 1.10), also shows the expected markedly different 2J(31P,1H) values.

P

+ Pt

HOMe H

P 78

P P = 1,2-C6H4(CH2PBut)2 2

J(31P,1H) = 176 Hz (trans) and 18 Hz (cis)


Figure 1.15 Section of the 13C NMR spectrum of 77, showing the second-order ABX character for

the two isomers of the two olefinic 13CH¼ resonances of the fumaronitrile ligand. The separation of the two most intense lines represents 2J(31P,13C)cis þ 2J(31P,13C)trans.

Rounding off this section, the Pt(ii)(diimine) complex 79 [96], a possible intermediate in C,H-activation chemistry, shows a 23.5 Hz 2J(195Pt,1H) coupling to the (averaged) h2 -C6H6 protons, thus helping to support the h2 olefin structure. The examples above comprise only a small fraction of the catalytically relevant NMR literature; however, they are representative. CH3

H3C ArN

+ Pt

NAr CH3

79

1.3.3

NOE Spectroscopy and 3-D Structure

The increasing interest in enantioselective homogeneous catalysis has led to questions with respect to how the auxiliaries transfer their chiral information to the coordinated substrate. In solution, the position of the chiral auxiliary relative to the complexed organic ligand is best determined via 1H-1H NOESY studies. Although this methodology enjoys a long history in biochemistry, there are still relatively few applications involving chiral organometallic complexes [22, 61, 97–128]. Early NOE studies on cationic Pd(ii) allyl complexes used “reporter ligands” [97, 98], i. e. simple bidentate nitrogen ligands, such as bipy or phenanthroline, whose

23

24


Figure 1.16 1H,1H NOESY spectrum of 80. The cross peaks indicated by the arrows arise from selective interligand NOE effects, i. e., the ortho protons of the bipyridyl recognize the b-pinene protons Ha and Hd’, but not Hb or Hc.

, or less, of the allyl ligand protons, in three-dimensional space, come within 3 A (see 80). These chelate ligand protons were then able to “report” on allyl rotations or other molecular distortions via NOEs. Although several allyl ligands were tested, the most useful proved to be that derived from b-pinene. A typical spectrum is given in Figure 1.16 [97]. This NOE idea was then extended to Pd(ii) allyl complexes with bidentate phosphine auxiliaries [99–111], with the ortho P-phenyl protons acting as the reporters (see 81). Figure 1.17 shows a section of the 1H,1H NOESY for [Pd(b-pinene allyl) (Chiraphos)](OTf) (Chiraphos ¼ Ph2PCH(CH3)CH(CH3)PPh2), 81 [129], and reveals the numerous contacts from the chiral phenyl array to the allyl ligand. Whereas the simple bidentate nitrogen ligands proved to be rather limited, the frequent occurrence of a set of four P–phenyl or alkyl substituents, e. g., in coordinated Binap, MeO-Biphep, Josiphos or Duphos (shown, from left to right in Scheme 1.4), offered many more “reporters”. In this way, one can develop a more detailed NOE picture of how the complexed substrate interacts with the chiral pocket offered by these auxiliaries. From these NOE studies [97, 98] it can be shown that the atropisomeric bidentate ligands Binap and MeO–Biphep tend to have fairly classical axial and equatorial P–phenyl substituents.


Section of the 1H,1H NOESY showing the contacts from the ortho P-phenyl protons to various b-pinene protons. Chiraphos ¼ Ph2PCH(CH3)CH(CH3)PPh2. Figure 1.17

PPh2 Me

Me PPh2 MeO

PPh2

MeO

PPh2

PPh2

Fe

PCy2

P

P Me

Me Binap

MeO-Biphep

Josiphos

Duphos

Bidentate chiral phosphorus ligands usually used in catalysis. The four P-phenyl or alkyl substituents are useful NOE reporters. Scheme 1.4

An NOE study of the intermediate [Pd(h3 -PhCHCHCHPh)(Binap)]þ, 82, thought to be involved in the Pd-catalyzed allylic alkylation of a 1,3-diphenylpropene, revealed that two phenyl rings, one from the auxiliary, D, and one from the substrate, F, are forced to take up parallel positions, i.e, they are p-stacked, as shown in 83 [103]. Since the p-stacking is repulsive, and thus selectively weakens one of the two Pd–C(allyl) bonds, the reaction becomes stereoselective. The D and F rings do not show inter-ligand NOEs.

25

26


A Ph P

P

+ Pd

= Binap

P

P2 D

P

B

Pd

P1

nucleophile

C

F

E

Ph aromatic stacking

82

83

Daley and Bergens [124] have used this approach to characterize the 3-D structure of the intermediate [Ru(ii)(alkoxide)(CH3CN)(Binap)]BF4, 84, in their study of an enantioselective catalytic hydrogenation of a ketone. The ortho-protons of the equatorial and axial P–phenyl rings provide the reporter protons.

NCCH3

Phax Pheq Pheq

P

O Ru

O O(i-Pr)

P O Phax

Me Me

O(i-Pr)

84 P P = Binap

The bidentate oxazoline ligands 85 and 86 (and derivatives thereof) are excellent reporter ligands, and several studies have used NOEs to determine the nature of their chiral pockets [61, 113, 114, 126]. NOESY studies on the cations [Ir(1,5COD)(86)]þ and several cationic tri-nuclear Ir(iii)(hydrido) compounds [110], e. g. [Ir3(m3 -H)(H)6(86)3]2þ, 87, in connection with their hydrogenation activity, allowed their 3-D solution structures to be determined. In addition to the ortho P–phenyl protons, the protons of the oxazoline alkyl group R2 are helpful in assigning the 3-D structure of both the catalyst precursors and the inactive tri-nuclear clusters. Specifically, for one of these tri-nuclear Ir(iii) complexes, 87 [110], with terminal hydride ligands at d –17.84 and d –21.32 (and a triply bridging hydride at d – 7.07), the P–phenyl and oxazoline reporters define their relative positions, as shown in Scheme 1.5. These NOE studies teach us that many successful P (or N…etc) auxiliaries possess a relatively rigid and intrusive chiral pocket [105, 107]. The shape of this pocket is a function of the individual chelate ligand, i. e., there is no one successful shape.

1.4 Isotopes in Catalysis

O

O N

P

N

2

R 85

H

86

N IrH2 P

N H2Ir

N

= 86 with R1 = Me R2 = t-Bu

N IrH2

P 87

δ (hydride) -21.3

N R2

2

2

R

O

R1

P

P

87

δ (hydride) -17.8

Scheme 1.5 Determination of the 3-D solution structure of 87 via NOEs between the hydride ligands and the P-phenyl and oxazoline reporters.

1.4

Isotopes in Catalysis

Isotopic labels (and especially enriched materials) have proven crucial in the investigation of the mechanisms of homogeneously catalyzed reactions [130]. Further, isotope effects on the rate or the equilibrium constant of a reaction can be diagnostic, and structural information can be provided by isotope-induced changes in the chemical shifts of neighbouring nuclei, and/or alterations in the coupling pattern of the detected spectra. The isotope- and position-specific information inherent to NMR techniques are ideally suited for the analysis of isotope effects in catalysis [131].

27

28


1.4.1

Kinetic Isotope Effect (KIE)

Proton/deuterium isotope effects on reaction rates are useful mechanistic probes. In the zirconocene-catalyzed alkene polymerization, the observed values of ka-1H/ka-2H determined by 1H NMR fall in the range of 1.2–1.3 and support a transition state in which there is an a-agostic interaction (see 88) [132].

H

H Cα

M

P

88

Landis and coworkers [133] have reported the first determination of a 12C/13C KIE in the [(rac-C2H4(1-indenyl)2)ZrMe2]–catalyzed polymerization of 1-hexene (Eq. (1)). It is suggested that the transition state, in which the alkene is irreversibly fixed into the growing polymer, does not change significantly as a function of the cocatalyst.

Me Zr Me

Cocatalyst Toluene 0 oC

n

In contrast to the isotope effects observed in d0 olefin polymerization catalysts, Brookhart and coworkers [134] have detected an inverse 1H/2H isotope effect of 0.59 in the Co(iii)-catalyzed polymerization of ethylene. This inverse effect was ascribed to a b-agostic CH bond which is stronger in the ground state than the “free” C–H bond during or just prior to the insertion step (see Scheme 1.6). In the hydrogenation of aldehydes catalyzed by the Ru complex 89, Casey and coworkers [135] have established the individual (kO1H/kO2H ¼ 2.2 and kRu1H/kRu2H ¼ 1.5) and overall (k1H/k2H ¼ 3.6) KIEs. As the product of the two individual KIEs (1.5 x 2.2 ¼3.3) is in good agreement with the experimentally measured value (3.6 e 0.3 relative to the RuD-OD analog), a reaction mechanism with concerted delivery of both H– and Hþ to the carbonyl is proposed.

1.4 Isotopes in Catalysis Me5

(MeO)3P

Me5

+ Co (MeO)3P

H

29

Me5 migratory

+ Co

H

n C2H4, etc + Co

insertion (MeO)3P

H

Me5

H

+ Co (MeO)3P Et

H

P = alkyl chain Scheme 1.6

Proposed b-agostic CH bond in the Co(iii)-catalyzed polymerization of ethylene.

Ph Tol

O H

Tol

Ph

Ru OC

O

H CO

C H

Ph

89

A direct kinetic isotope effect of k1H/k2H ¼ 3.1 e 0.1 is found in the Rh-catalyzed enantioselective transfer hydrogenation of ab-unsaturated carboxylic acids using HCOOH/Et3N vs. DCOOH/Et3N as the hydrogen source (Eq. (2)) [136]. This observation clearly indicates that the cleavage of the C–H bond of formic acid is the ratelimiting step, and rules out the participation of free molecular hydrogen in the reaction [136]. COOH

HCOOH / Et3N (5:2)

R

COOH R

CH3OH, 300 K

R'

R'

[Rh(nbd)(dppb)]OTf nbd = norbornadiene dppb = 1,4-bis(diphenylphosphinobutane)

1.4.2

Structural Effects

The introduction of isotopes into a compound alters the coupling pattern and the chemical shifts of the observed spectrum. As shown in Figure 1.18, deuterium-induced 13C chemical shift variations have allowed the estimation of the ratio of isomers 90a–d formed in Eq. (2) when R ¼ Ph, R’ ¼ CH2COOH, and DCOOD/ Et3N is used for the hydrogen transfer [136]. The three sp3 -carbons C1, C2 and C3 each afford a distinct singlet for the four possible isotopomers 90a–d (replacement of 1 H by 2H shifts the resonances of the adjacent carbon nuclei to lower frequency) [137].

P

30


Figure 1.18 The three high-field signals, C1–C3, in the 13C{1H,2H} NMR spectrum of the product obtained by transfer hydrogenation of PhCH¼C(COOH) –CH2COOH with 2HCOOH/NEt3, at about 25 % conversion. The chemical shifts for the undeuterated isotopomer 90a are d C1 44.3, d C2 38.6 and d C3 35.9, respectively.

In Figure 1.19, the small difference in d 1H between the CH3 and CH2D ligands permits the recognition of the exchange between RhD(CH3)(Tp’)(CNCH2But) and RhH(CH2D)(Tp’)(CNCH2But)Tp’Rh(L), (Tp’ ¼ tris-3,5-dimethylpyrazolylborate, L ¼ CNCH2CMe3), as a function of time [138]. The methyl signal of the CH2D ligand is found at slightly lower frequency.

Figure 1.19 Methyl region as a function of time (minutes) of the 1H NMR spectrum from the rearrangement of RhD(CH3)(Tp’)(CNCH2But) to RhH(CH2D)(Tp’)(CNCH2But) in d6 -benzene at 295 K.

1.4 Isotopes in Catalysis

Figure 1.20 Methyl region of the 1H NMR spectra arising from a solution of Ir(OTf)Me(Cp*)(PMe3): Top: in CD2Cl2 under 2 bar of 13CH4 immediately after mixing (298 K). Bottom: after heating at 318 K for 6 h.

And in a last example, facile Ir(iii) C–H activation is proven via Figure 1.20, where [Ir(CF3SO3)(CH3)(Cp*)(PMe3)] is allowed to react with 13CH4. After warming for 6 h, incorporation of 13C-labelled methane to give [Ir(CF3SO3)(13CH3)(Cp*)(PMe3)] and CH4 had occurred to an extent of 50 %, as observed from the new signals deriving from the large 1J(13C,1H) spin–spin coupling [139]. 1.4.3

An Active Site Counting Method

Landis and coworkers [140] have developed an active-site counting method based on 2H-labelling, for the metallocene-catalyzed alkene polymerization. After quenching the reaction by addition of methanol, the polymer is analyzed by 2H NMR, which allows the quantification of Zr-alkyl sites. A typical 2H NMR of quenched polymer is shown in Scheme 1.7 (label is found at terminal positions only). This technique has been applied to the polymerization of 1-hexene catalyzed by [Zr(rac-C2H4(1-indenyl)2)Me][MeB(C6F5)3], 91. As shown in Scheme 1.7, there are two possible approaches: (A)Quenching with CH3OD, and analysis of the polymer by 2H NMR yields the number of active sites at the time of quench. (B) Carrying out the reaction with the labeled catalyst d6 -91 [Zr(rac-C2H4(1-indenyl)2)CD3][CD3B(C6F5)3] and quenching with unlabelled CH3OH yields the number of sites that were active at any point before the quench.

31

32


A) Label at quench timed reaction interval

CH3B(C6F5)3-

CH3B(C6F5)3-

Zr

Zr

R

Me

D MeOD

R 91

quench

R

n

R

R n

91*

count 2H-labeled polymers with 2H NMR

B) Label before quench timed reaction interval

CD3B(C6F5)3-

CH3B(C6F5)3-

Zr

Zr

R

CD3

CD3 MeOH R

d6-91

CD3

quench

H R

n

R

R n

d6-91*

CDCl3 Internal Standard -CH2D End

H NMR

Groups

[C*]/[1]

0

2 Natural Abundance C6H5D

10

9

8

7

6

5

4

3

2

time (s) 2

Scheme 1.7 Active-site counting method based on H-labelling, in the zirconocene-catalyzed polymerization of 1-hexene. Lower left: typical 2H NMR of the quenched polymer according to method A. The integrals allow the quantification of the Zr–alkyl active sites. All labels are found in the terminal position. Lower right: comparison of fractional active-site counts using Method A (open circles, o) or Method B (diamonds k).

The plot shown in Scheme 1.7 shows that both labeling methods yield similar active-site counts. From this observation it can be concluded that the catalyst 91 does not deactivate during the time scale of the experiments. This active site counting methodology has been applied to the determination of initiation, propagation and termination rate laws and activation parameters for the polymerization of 1-hexene [141] catalyzed by 91 in toluene solution.

1.5 Dynamic NMR Spectroscopy

1.5

Dynamic NMR Spectroscopy

Since NMR spectroscopy involves the storage of information relating to the precession frequencies of individual nuclei over periods up to and including a few minutes, it provides the opportunity to study equilibrium and chemical exchange effects during this time [107, 142–145]. Suitable chemical exchange processes may be inter- or intramolecular, and correspond to internal rotations, conformational changes and even tautomerism. The time scale of NMR is such that firstorder rate constants in the range of 10 –2 – 106 s–1 can be measured. Both irreversible and reversible processes can be affected by temperature changes. The use of NMR spectroscopy in the study of irreversible changes has been illustrated in Section 1.2. Reversible intramolecular and intermolecular exchange processes may be slow, intermediate or fast, according to the frequency difference between the two (or more) exchanging sites [146]. 1.5.1

Variable Temperature Studies

Macchioni and coworkers [147] have investigated the effect of the counterion on CO/Styrene copolymerization catalyzed by [Pd(h1-h2 -C8H12OMe)bipy]X, (92, Figure 1.21) where X is the counterion. The bipyridine ligand shows dynamic behavior and a series of 1H NMR spectra recorded between 204 and 302 K (Figure 1.21) pro-

Figure 1.21 Variable temperature 1H NMR spectra in CD2Cl2, showing the passage from a slow to fast exchange for the two halves of the bipyridine ring in 92.

33

34


vide data for the determination of the activation parameters for the exchange of the two halves of the bipyridine ligand. At 204 K, the NMR spectrum in Figure 1.21 clearly exhibits well separated and well defined signals such as 5 and 5’, which coalesce at 251 K, and finally resharpen to provide a broad triplet. The situation is similar for the protons 3 and 3’, and 4 and 4’. For 6 and 6’ it is difficult to distinguish the averaged peak at high temperature. These four sets of protons resonances, 3, 4, 5, 6 and 3’, 4’, 5’ and 6’, belonging to the bypiridine ligand, interchange positions via the mechanism shown is Scheme 1.8.

X

X-

-

N2

1

+ Pd

N

+ Pd

N2

N1

92 O

O

X

X

N1

Pd

N2

O

X

X 2

N

+

Pd

N1 O

N1

Pd

O

Pd

N2

N1 N2

O

Mechanism of the reversible exchange of the bipyridine ligand in 92.

Scheme 1.8

Brookhart et al. [148] have reported a variable temperature NMR study for a Pd CO/olefin copolymerization catalyst. They studied the bite angle effect of the bidendate phosphine ligand in a series of complexes [Pd(CH3)(OEt2)(P–P)](BAr4), on the rate of migratory insertion of both ethylene and carbon monoxide, and isolated the resting state for the catalytic cycle involving C2H4. With the ligand (1,3-diisopropylphosphino)propane (dippp), upon insertion of ethylene, the b-agostic ethyl complex [Pd(CH2CH3)(dippp)]þ, 93, is the resting state for the catalytic cycle. This complex exhibits two dynamic processes, described as the interchange of Ca and Cb (pathway II in Scheme 1.9) and rotation of the agostic methyl group (pathway I in Scheme 1.9).


* P

+ Pd

*

P

Hb Hc

II

P

P

Ha

II

+ Pd Ha

P

* + Pd Ha 93 I

I P

Hb Hc

P

93

35

P + Pd

P Scheme 1.9

*

Ha Hb

Hc

P

+ Pd

* Ha Hb Hc

Fluxional processes observed for 93.

In a quest to increase the efficiency of olefin polymerization catalysts and their selectivity in the orientation of the polymerization, the highly effective Group IV metallocene catalysts, M(Cp)2(L)2, have been studied, since they all display high fluxionality. Following methide abstraction, the metallocene catalysts of general formula M(Cp-derivatives)2(CH3)2 (M¼ Ti, Zr, Hf), were turned into highly reactive Mþ–CH3 cationic species. The activation parameters for the methide abstraction, derived from variable temperature NMR experiments, establish a correlation between the enthalpies of methide abstraction, the 13C chemical shift in the resulting cation, and the ethylene polymerization activities [149]. At ambient temperature the 1H NMR spectrum of Zr(C7H9NCH3)2Cl2, 94, shows a simple pattern, characteristic of a molecule with C2v symmetry (Figure 1.22) [150]. However, upon cooling, the signals for the three allylic and two bridgehead resonances broaden, and at 183 K they separate into 10 different signals. The coalescence temperature for the central allyl 1H resonance is 225 K. The alkyne hydrogenation catalyst [Rh(7-SPh-8-Ph-7,8-C2B9H10)(COD)], 95, (and derivatives thereof) shows dynamic behavior at ambient temperature (Figure 1.23) [151]. The carborane is bonded to the rhodium center via a thiolate anion and a bridging B–H–Rh bond. Complex 95 undergoes B–H/B–H–Rh interchange coupled with an apparent rotation of the Rh(COD)þ fragment. Variable temperature 1H NMR spectra were recorded between 293 and 179 K. The apparent rotation of 1, 5-COD is found in a rather large number of Rh-complexes [152]. The zwitterionic rhodium complex Rh(h6 -PhBPh3)(COD) is an effective catalyst for regioselective hydroformylation, silylformylation and hydrogenation reactions in the presence of dppb (1,2-bis(diphenylphosphino)butane) [153]. Alper and coworkers [153] have studied the interconversion of the zwitterionic Rh(h6 PhBPh3)(diene)n(dppb)1-n and cationic [Rh(diene)n(dppb)2-n](BPh4) (n ¼ 0 or 1) complexes in solution and showed fluxional behavior for the h6 -tetraphenylborate-coordinated complexes. The 31P variable temperature NMR spectra in Figure 1.24 show the solution behavior of the zwiterrion Rh(h6 -PhBPh3)(dppb),

36


Figure 1.22 Variable temperature 1H NMR spectra of 94.

Figure 1.23 Variable temperature spectra in the olefinic and B–H regions of the 1H{11B} NMR spectra, showing nonequivalent four 1,5-COD protons and the B–H/B–H–Rh interchange in 95.


Figure 1.24 Variable temperature 31P NMR spectra of 96 in CD2Cl2/d8 -toluene, and the proposed dynamic process responsible for the exchange of the 31P nuclei.

96. At ambient temperature, equivalent P atoms are observed (1J(103Rh,31P) ¼ 200 Hz), due to a rapid exchange of the two 31P nuclei. As the temperature decreases the signal broadens and decoalesces to provide two rather broad doublets at 168 K (athough 1J(103Rh,31P) is resolved, 2J(31P,31P) is not). This indicates that, although the presumed rotation is slowed significantly, some exchange is still occurring at 168 K (the lowest temperature reachable). Hindered rotation of the h6 -coordinated phenyl ring is believed to be responsible for the observed dynamics. The Ru(iv) complex Ru(SiMe3)2(H)2(PMe3)3 is a model for dehydrogenative C–Si bond formation chemistry. Berry and coworkers have shown that, at low temperature, this complex undergoes fast reversible SiH reductive elimination on the NMR time scale (Eq. (3)), and slow H2 reductive elimination as a minor process [154].

T < 250 K

PMe 3 Me 3Si

PMe3 Ru

H

+ HSiMe3 – HSiMe3

PMe 3

PMe3 Me 3 P

H Ru

Me3 Si

SiMe3 H

PMe3

37

38


1.5.2

Line Shape Analysis [142, 146, 155–159]

In solution, rate constants and activation parameters for dynamic processes can be estimated by direct analysis of the change of the NMR signal shape as a function of temperature. This technique is called line shape analysis (LSA) and it is best suited when the rate of exchange ranges from ca. 10 to 103 s–1 [142, 159]. If the exchange rate, kex, is much smaller than the frequency difference between the signals for the exchanging sites, then the NMR spectrum will exhibit well separated peaks for these resonances. Based on the Bloch equation [146], it is possible to find relationships connecting the shape of the NMR signal, Dn1/2, the lifetime, t, for a nucleus in different positions of a molecule and the rate constant, k. The lifetime is related to the rate constant by Eq. (4). τ =

1 k

At values where k equals the difference in resonance frequencies, the peaks coalesce. The rate of interchange at coalescence, kcoal, can be determined via Eq. (5) [155], for exchange between two equally populated sites that do not exhibit scalar coupling [160]. kcoal = 2π∆

Although this equation only applies when the coalescence point is reached, rate constants for the exchange between two or more exchanging sites are accessible by analysis of line widths at half height, Dn1/2, and shift differences, Dn, in Hz. The comparison between the experimental spectrum and the spectrum calculated by use of a simulation package for line shapes provides the mechanism for determining the rate constant of exchange [161, 162]. NCN-pincer ligands show synthetic potential as catalysts [163, 164]. Chung and coworkers [165] have developed an efficient pincer catalyst for the Heck reaction, by a judicious modification of the classical NCN ligands (see 97). Interestingly, at elevated temperatures the methyl signals of the Pd–NMe of 97 coalesce with those for the noncoordinated N–CH3 arms, suggesting a ligand exchange reaction (Figure 1.25). The entropy of activation derived from the temperature dependent rate constants supports an associative mechanism. Casey et al. have studied complex 98 as a model for intermediates in metallocene catalyzed alkene polymerization, by means of LSA [166]. At 195 K, the resonances for the Cp ligands, silyl methyl groups and the methylene fragment in 98 exhibit diastereotopic resonances in CD2Cl2. As the temperature is increased, coalescence of the pairs of diastereotopic resonances occurs at different temperatures (Cp rings at 245 K, silyl methyl groups at 240 K). All the tem-


39

Experimental (solid line) and calculated (dotted line) variable temperature 1H NMR spectrum of the N-methyl resonances of 97, in d7-DMF. This series of spectra shows the exchange of the methyl group A with C, and B with D. The resonances for A and B, and for C and D overlap at all temperatures considered. Figure 1.25

B(C6F5)4

H3C Cp

H

Si

CH3

Zr Cp

H2C 98

perature-dependent changes observed are reversible up to 248 K, and decomposition begins at 253 K. An LSA study affords an activation barrier of 12.8 kcal mol–1 at 248 K for alkene dissociation. The proposed explanation for the fluxional behavior of the model is shown in Scheme 1.10. Alkene dissociation, followed by rotation around the Zr–C bond and subsequent recoordination to the opposite face of the alkene are required. The study also shows that the resonances of the terminal vinyl protons are not affected by temperature, thereby ruling out a mechanism involving intramolecular insertion followed by ring-opening.

H3C H

+

Zr

Si

CH3 CH3

+

Zr

Si

CH3

Hb Ha C +

Zr

Hb

Ha 98 Scheme 1.10

Si CH 3

C Ha

Hb C

Mechanism for the fluxional behaviour of the model Zr complex 98.

H3C 98

40


The dynamic behavior of the model intermediate rhodium-phosphine 99, for the asymmetric hydrogenation of dimethyl itaconate by cationic rhodium complexes, has been studied by variable temperature 31P NMR LSA [167]. The line shape analysis provides rates of exchange and activation parameters in favor of an intermolecular process, in agreement with the mechanism already described for bis(phosphinite) chelates by Brown and coworkers [168]. These authors describe a dynamic behavior where two diastereoisomeric enamide complexes exchange via olefin dissociation, subsequent rotation about the N–C(olefinic) bond and recoordination. These studies provide insight into the electronic and steric factors that affect the activity and stereoselectivity for the asymmetric hydrogenation of amino acid precursors.

H2C

N PPh 2 +

CH H2C O

Rh

O PPh2

O

99

Complexes based on seven-membered bis(phosphane) chelate ligands show very high catalytic activity in hydrogenation chemistry. Kadyrov and coworkers [169] have studied the temperature-dependent 31P NMR spectra for [Rh(R,R-diop) (COD)](BF4), 100. At 240 K (Figure 1.26, top), the 31P NMR spectrum of this complex exhibits a sharp doublet. As the temperature is decreased, the signals broaden and a second component is observed. At 123 K, a complex with two non-equivalent 31 P signals is present. The signal for the major isomer is thought to reflect a rigid conformation with C2 symmetry, due to the sharpening of the signal at low temperature. This major species is believed to contain a seven-membered ring in a twist chair conformation. The minor species was described as a distorted boat. Line shape analysis performed using WIN-DYNAMICS [170] provided activation parameters for the interchange between these conformations. The results are interpreted as arising from a temperature-dependent equilibrium of at least two different seven-membered ring conformers. In an attempt to evaluate the parameters that influence the production of methyl propanoate versus CO–ethylene copolymer, Doherty and coworkers [171] have probed the dynamic processes connected with the catalyst precursors [PdCl(CH3)(P–P)], 101, where P–P is a bidentate phosphine bearing a cyclohexane or norbornane group, containing a C4 -backbone. The variable temperature 31P


Figure 1.26 Variable temperature 31P NMR study for 100. The second 31P resonance for the minor component lies under the signal for the major resonance.

NMR spectra (Figure 1.27) revealed an equilibrium. The analysis of the variation of the shape and position of these 31P resonances provided the activation parameters for these processes. In a similar fashion, the dynamic behavior of Ni(b-agostic alkyl)(a-diimines) cationic complexes, e. g. 102, which are models for the intermediates involved in Ni catalyzed polymerization of ethylene, has been studied by LSA and provides insight into the mechanism of ethylene polymerization [44]. Eq. (6), shows the exchange between two of these species. 31P

NMR

308 K

Ph2 P

Ph2 P Pd

298 K

Me Cl

273 K

101

253 K

Cl Ph2 P Pd

Me

233 K

PPh2 213 K 50

Figure 1.27

31

40

30

Variable temperature P NMR spectra of 101, in THF.

20

10

0

41

42


BAr'4 N

BAr'4 N

Ni

Ni

H

N

H

N 102

These few examples illustrate how the shape of NMR signals is affected by dynamic phenomena within the molecule. The analysis of these effects provides a useful tool for both a qualitative (localisation of exchanging sites) and quantitative (kinetic data) understanding of fluxionality within metal complexes. 1.5.3

Magnetization Transfer

Magnetization transfer techniques [146, 172] are used to study systems under conditions of slow exchange, where variation of the temperature is either undesirable or not relevant. If the exchange rate is comparable to T1 (the spin–lattice relaxation time), rate constants can be determined by these double resonance experiments. By selectively irradiating one of the exchanging sites, observable transfer of magnetization occurs from the irradiated site to all sites directly involved in the exchange. The intensity of the signals after transfer of magnetization has a characteristic time dependence, from which the rate constant for the exchange can be derived [173]. In this respect, magnetization transfer techniques can be viewed as a quantitative method. One can imagine two protons, HA and HB, being part of the same molecule and undergoing chemical exchange, at a rate kHH. When HA is irradiated, it “remembers” the new condition and transfers this information to HB as a result of the chemical exchange. The newly arrived HB proton does not contribute to the normal amount of signal intensity in the final NMR spectrum. If the initial intensity of HB is Io, and the final intensity for HB as a result of irradiation of HA is If, then the rate of exchange kHH is defined by Eq. (7), where T1 refers to the spin–lattice relaxation time. Io T 1( HB) −1 = I F kHH + T 1( HB) −1

Brown and coworkers [128] have studied the exchange process for the enamide complex 103, using magnetization transfer techniques (Figure 1.28). Compound 103 represents the catalytic resting state in the asymmetric homogeneous hydroge-


43

nation of dehydroamino acid derivatives and exists as a mixture of two diastereoisomers, 103a and 103b. Figure 1.28 shows the exchange between the saturated 31 P nucleus trans to amide in the major diastereoisomer, 103a, and the 31P nucleus trans to amide in the minor diastereoisomer, 103b, (exchanging resonances indicated by arrows). The use of this NMR technique, coupled to other studies, allows extrapolation to the mechanism shown in Figure 1.28. The exchange pathway involves dissociation of the olefin with subsequent rotation about the N–C(olefin) bond and recoordination of the olefin to the metal center. The amide oxygen atom remains coordinated to the rhodium during the entire process. An understanding of the mechanism of this diastereomeric interchange is of importance since the minor diastereoisomer, 103b, is believed to carry the flux of catalysis. Saturated resonance

31

P NMR

103a MeOPh

Ph P

Ph

+ PhRh

P PhOMe O

RO2C

N H

R'

103a

MeOPh Ph RO2C

Ph P

+ Rh NH

Site of magnetisation transfer 103b

P PhOMe O

Ph R' 103b

MeOPh

Ph P

Ph

+ Rh

P PhOMe

O HN

R' Ph

RO2C

Figure 1.28 31P NMR spectra obtained after saturation transfer of the 31P nucleus trans to amide in the [Rh(dipamp)(enamide)]þ diastereoisomer 103a. Direct exchange of magnetisation is observed between the 31P atoms trans to amide in the diastereomers 103a and 103b. The arrows pointing upwards indicate the most affected resonance. The proposed mechanism of intramolecular equilibration of 103a and 103b is shown.

1.5.4

NOESY/EXSY/Hidden Signals [146, 174]

A number of 2D NMR experiments have been developed which allow the study of slow exchange phenomena. The most common (Exchange Spectroscopy, EXSY), is based on the standard pulse sequence 90hx–t1–90hx–tm–90hx – FID(t2), where t1 is an evolution delay, tm is the mixing time, and t2 is the detection period [175, 176].

44


The first 90hx pulse produces transverse magnetization, My, which develops during the evolution time, t1, by precessing around B0 with the different Larmor frequencies. The second 90hx pulse rotates the magnetization into the z direction. The longitudinal magnetization, Mz, evolves further under two effects, spin–lattice relaxation (T1) and chemical exchange. The latter effect induces a transfer of magnetization between the exchanging sites and leads to off-diagonal cross-peaks. This effect is read-out by the third 90hx pulse, which creates the magnetization in the xy plane that produces the FID during t2. The amplitude of the transverse magnetization depends directly on the mixing time, tm, and the efficiency of the magnetization transfer. Whereas the time t1 is incremented sequentially to produce the second chemical shift axis, the acquisition time, t2, is equal to the acquisition time in a normal 1D NMR experiment. Typically, tm ranges from 0.03 to 1 s. Double Fourier transformation of the time domains t1 and t2 results in a 2D NMR spectrum [145]. From these data, the rate of magnetization transfer during the mixing time, tm, can be estimated. For the extreme narrowing condition, these cross peaks have the same phase as the diagonal, whereas the NOE cross-peaks possess an opposite phase. EXSY is a very effective method where no insight with respect to the exchange pathway is available, e. g. for systems undergoing multiple site exchange (e. g. three or more nuclei exchanging their positions at the same or at different rates of exchange), and/or where signals are broad or hidden. Returning to the enantioselective hydrogenation chemistry with rhodium [177, 178], Landis and Halpern [33] have proposed an intermolecular pathway based on their 31P NMR spectra, while Brown and coworkers [128, 168] have proposed an intramolecular pathway, according to the dynamic data obtained from DANTE inversion-recovery NMR experiments. In a later publication, Philipsborn and coworkers have studied the dynamic behavior of the model intermediate [Rh(S,S-chiraphos)(MAC)]þ, 105, (Eq. (8)) [179]. Their new data supported both the intra- and intermolecular exchange mechanisms, and showed that there is a unique pathway that can possibly account for both suggestions. Figure 1.29(a) shows the 31P EXSY spectrum of 105 in the presence of excess olefin substrate. The phosphorus nuclei of the major and minor diastereoisomers exchange with partial retention of the configuration. For instance, PA (trans to the olefin in the major diastereoisomer) exchanges with Pa (trans to the olefin in the minor diastereoisomer, exchange cross-peaks circled). Further, PB exchanges with Pb (both trans Ph2 Pa

+ Rh

Pb Ph2 MeO2C

105 minor

Ph2 PA

O

+ Rh

Ph NH

O

PB Ph2 MeO2C

NH Ph

105 Major


Figure 1.29 31P EXSY spectrum of [Rh(S,S-chiraphos)(mac)]þ in CD3OD with (a) a mixing time of 200 ms in the presence of mac, (b) a mixing time of 40 ms and a defiency of mac.

to the amide oxygen). These observed exchange processes are consistent with an intramolecular exchange pathway. Nevertheless, an exchange of the 31P nuclei between the two diastereoisomers without retention of configuration was also observed (e. g. Pa interchanges with PB, as indicated by the diamond) which points to an intermolecular pathway. Moreover, a 31P EXSY spectrum of a sample of 105 with a deficiency in MAC (Figure 1.29(b)) shows a similar exchange pattern, but also exchange with a bis-solvated species, [Rh(S,S-chiraphos)(solvent)2]þ (indicated by an arrow). Scheme 1.11 shows the proposed exchange pathway for 105, involving

Ph2PA

+ Rh

O

Ph 2PB MeO2C

105

H

Ph 2PA NH

S

+ Rh

Ph2PB

S Ph 2 PA

Ph

= S,S-chiraphos

Major

Ph2PB

Ph2PA

+ Rh

O

Ph 2Pa NH

Ph 2PB MeO2C

Ph2Pb

+ Rh

O

H

Ph NH Exchange pathway for the interchange of diastereoisomers of [Rh(R,Rdipamp)mac]þ.

Scheme 1.11

Ph H

MeO2C

105 minor

45

46


the bis-solvent complex. This mechanism accounts for intra- and intermolecular exchange. This study illustrates the substantial advantage of using EXSY NMR when exchange occurs between more than two sites on one or more molecules, since it provides a complete mapping of the dynamics. A mixture of the zwitterionic rhodium complex Rh(h6 -PhBPh3)(1,5-COD), 106, and dppb affords the mono- and dinuclear complexes 107 and 108 (Eq. (9)) [153]. Alper and coworkers [153] have used 2D EXSY spectroscopy in order to map the exchange pathway between 31P nuclei in these two rhodium complexes (Figure 1.30) [153]. The mixture shows four 31P resonances for each cation, marked ADGM for 107 and adgm for 108, and the exchange is shown to be both inter- and intramolecular in nature. Each species undergoes slow selective intramolecular exchange of its four 31P nuclei. Within the monomeric species, PD undergoes intramolecular exchange with PA (circled) and with PM (indicated by an arrow). Further, PG exchanges position with Pm and Pa (squared), indicating an intermolecular exchange. The intensity of these peaks is directly related to the rate constant of the exchange considered.

PG

+ Rh

PM

PD

Pg

PA

Pm

+ Rh

Pd

P

Pa

P

P + Rh P

107 108

B

+ Rh

106

Perutz and coworkers [180] have used 2D EXSY to study the dynamic behavior of RhCp(PMe3)(h2 -naphthalene), 109, which is thought to be a model intermediate for the oxidative addition of arenes to a metal center. In this complex, there are two processes taking place. The first involves an equilibrium between the h2 -naphthalene complex, 109, and the naphthyl hydride complex, 110. The second process involves an intramolecular [1,3]-shift which moves the coordination site of the naphthalene ring from one side of the ring to the other (Scheme 1.12).


31

P EXSY spectrum (CD2Cl2, 176 K) showing cross-peaks for the intra- and intermolecular exchange between 31P nuclei of the monomeric, 107, and dimeric, 108, forms of [Rh(dppb)2](BPh4). Figure 1.30

1,3 shift Rh

Rh

Me3P

Me3P 109

Rh Me3P 110

H Scheme 1.12

109.

Dynamic processes in

47

48


Figure 1.31 Section of the 1H EXSY NMR spectrum showing selective exchange between the species a and c (indicated by arrows), and the species a and the minor allyl compound e (circled), in 111.

Pd(ii) allyl phosphino-oxazoline complexes can be intermediates in allylic alkylation chemistry [181–184]. [Pd(h3 -PhCHCHCHPh)(phosphino-oxazoline)]þ, 111, reveals a mixture of four species a, b, c, d with “a” dominating [185]. Figure 1.31 shows a section of the 1H EXSY spectrum in the region of the terminal allyl protons. The species a and c, corresponding to the syn/syn endo and syn/syn exo diastereoisomers, are clearly exchanging. However, a major feature of the spectrum is the intense cross-peaks between species a and a fifth complex, e. Complex “e”, present at very low concentration, was not detected by conventional 1D NMR spectroscopy. This species was later identified as a syn/anti allyl compound, using low temperature NMR techniques. This application represents a classic example of the detection of a hidden species using 2D EXSY. 2D EXSY has also proven to be a useful tool for the study of the dynamic behavior of transition metal carbonyl clusters [186]. These complexes have diverse applications in homogenous catalysis, including carbonylation, hydrogenation and hydroformylation reactions [187–189]. The dynamic behavior of such compounds is often viewed in terms of the migration of the CO ligand about the surface of the metallic skeleton [190]. Figure 1.32 shows the 13C EXSY spectrum recorded at 240 K for Rh4(CO)6(PPh2)4, 112. Two pairs of 13C nuclei are undergoing interchange, while a set of 31P EXSY spectra (not shown) also showed that the two 31 P nuclei are exchanging their positions. The evaluation of the activation parameters for these processes and the analysis of the possible mechanism led the


13

C EXSY spectrum of 112 in CD2Cl2 at 240 K, showing pairwise exchange of nuclei (circled). Figure 1.32

13

C

authors to propose an exchange pathway where the strongly bridging PPh2 ligand actually hops between rhodium atoms. On the more exotic side, we note an example of EXSY using 195Pt and 125Te as probes. Orrell and coworkers [174, 191] have studied the dynamic behavior of a series of trimethylplatinum(iv) iodide complexes with ditelluroether bridging ligands, PtIMe3(L-L) [(L-L) ¼ MeTe(CH2)3TeMe and PhTe(CH2)3TePh], as shown in Figure 1.33. At ambient temperature, the complexes exist as distinct dl isomers that

Possible exchange pathways for the inversion at Te in complexes of the type [PtIMe3L-L], (L-L ¼ MeTe(CH2)3TeMe and PhTe(CH2)3TePh). 125Te EXSY spectrum of [PtIMe3{MeTe(CH2)3TeMe}] in CDCl3 at 313 K. The circles indicate the position of the exchange cross peaks between the DL and the meso isomers. No exchange between the two meso isomers is observed in this compound. Figure 1.33

49

50


undergo slow pyramidal inversion at tellurium on the NMR time scale. 195Pt (not shown) and 125Te EXSY spectra (Figure 1.33) afforded both qualitative and quantitative information with respect to the exchange observed between the DL and the meso species. Note that the 195Pt satellites do not exhibit cross-peaks between the various species. This absence of 195Pt to 195Pt exchange processes supports an intramolecular process.

1.6

Special Topics 1.6.1

T1 and Molecular H2 Complexes

The longitudinal relaxation time (often called the spin–lattice relaxation time), T1, is concerned with the rate at which nuclei in a molecule exchange energy with their surroundings (the lattice). This time constant can vary from 10 –3 to 102 s and is directly related to the efficiency of the coupling between the nuclear spin and the lattice [142, 143, 192]. The measurement of spin–lattice relaxation times, T1, has been proposed as a method of distinguishing between hydride and dihydrogen complexes [193]. Dihydrogen complexes [194] are recognized as intermediates in the oxidative addition of dihydrogen to a metal center. Crabtree has proposed that T1 values are much shorter for h2 -H2 complexes (a few milliseconds) than for hydride complexes (several hundred milliseconds) [193]. The short H–H distances in h2 -dihydrogen ligands, typically between 0.82 and 1 A, lead to efficient dipole–dipole relaxation. The corresponding H–H separations in hydride complexes are larger than 1.6 A. Other authors have urged caution in using this approach, arguing that the ranges of relaxation rates for the two types of complex actually overlap [195]. However, recent calculations have shown that the method is valid, and that it can be rationalised considering that both slow (static) and fast (rotation) motions for the dihydrogen ligand are of importance in solution [196–198]. Protonation of a hydride can readily lead to an, e. g., Ru-(H2) complex [199]. Since a correlation between 1J(2H,1H) and d(H–D) has been suggested (Eq. (10)), the three parameters, T1 values, coupling constants 1J(2H,1H), and d(H–H) in hydrogen complexes are related. d(H–D) ≈ – 0.0167 1J(2D, 1H) + 1.42

Morris and coworkers have studied trans-[OsH(H2)(depe)2]þ (depe ¼ diethylphosphinoethane) using several spectroscopic methods, and in particular by T1(H2) inversion–recovery methods in solution between 190 and 300 K [200]. They conclude that if the motion of the H2 ligand is much faster than the tumbling frequency, Eq. (11) applies. If the motion of the H2 ligand is slower than the tumbling frequency,

1.6 Special Topics

d(H–H) = 4.611(T 1(min)/ν)1/6

then Eq. (12) is applicable (n is the frequency of the spectrometer and the H–H distance is given in A). d(H–H) = 5.815(T 1(min)/ν)1/6

In a later publication, Morris and Wittebort re-examined the 1J(2H,1H), T1 and d(H–H) correlation for 73 complexes already reported in the literature [196]. For most complexes the calculated H–H distances, based on T1, lie between those found for the slow and fast motion conditions for the h2 -H2, relative to the motion of the molecule. Dihydrogen fast rotation was proposed for 32 complexes, whereas six complexes appeared to have a H2 ligand with slow internal motion. Further, torsional oscillation of the H2 ligand, or fast 90h hopping, may still play a significant role in the T1 relaxation value of the remaining 35 complexes. It would seem that employing T1 values as a measure for distances in complexed H2 is still not straightforward. 1.6.2

Parahydrogen Induced Polarization (PHIP) [201, 202]

The study of detailed chemical reaction mechanisms in homogeneous catalysis requires the identification and characterization of reaction intermediates. However, limitations arise due to both the short life time (transient type) and the low concentration of such species [203]. In 1986, Bowers and Weitekamp demonstrated the existence of hydrogen in its para spin state, which opened yet another possibility for intermediate detection [204]. Molecular hydrogen exists in two isomeric forms, with its two spins aligned either parallel, orthohydrogen, with the possible spin state combinations aa, ab þ ba and bb (nuclear triplet state), or antiparallel, parahydrogen, bearing the ab-ba spin state (nuclear singlet state). Interconversion between these two isomers is spin-forbidden. Under equilibrium conditions at ambient temperature, dihydrogen contains 25 % parahydrogen and 75 % orthohydrogen. At low temperature, parahydrogen is preferred, and the mixture can be enriched in this isomer by use of a paramagnetic catalyst (e. g. 50 % para-enriched hydrogen is obtained at 77 K) [201, 205]. Since ortho–para interconversion is slow, it is possible to separate and store the para-enriched hydrogen for subsequent hydrogenation reactions. Parahydrogen is not directly detectable by NMR, as it has no magnetic moment. To achieve an NMR signal, the two hydrogen atoms of the para-enriched hydrogen molecule must (a) be delivered to the substrate in pairs (so that the original nuclear spin state is retained), and (b) form a product with two magnetically distinct protons (so that the spin symmetry is broken during the transfer) [204]. The product will experience non-Boltzman spin populations and, hence, yield substantially en-

51

52


Figure 1.34 Energy level diagram for a two spin AX system, where A and X are inequivalent. Their corresponding NMR patterns are shown. (a) regular distribution (b) parahydrogen-derived distributions.

hanced and phase distorted NMR signals. Thus, this technique permits the detection of minor reaction products and intermediates, while simultaneously providing a signature for the transfer of hydrogen atoms from the same hydrogen molecule to a specific substrate. Figure 1.34 illustrates this effect by reference to the formation of a M(H)2 complex where the two hydride ligands are inequivalent [206]. The thicker lines associated with the ab and ba levels on Figure 1.34(b) correspond to parahydrogen-derived population changes, while those on the left correspond to the normal (Boltzmann) distribution. For a successful observation of PHIP signals, the overall process must be faster than thermal spin relaxation.

Hydrogenation Mechanism Studies [201–203, 207] One of the first molecules studied using parahydrogen techniques involved Vaska’s complex, trans-Ir(Cl)(CO)(PPh3)2. Oxidative addition of H2 was thought to proceed solely over the CO–Ir–Cl axis, yielding cis-cis-trans-IrCl(CO)(H)2(PPh3)2 [208, 209]. However, calculations showed that H2 addition over the P–Ir–P axis, forming the all-cis-IrCl(CO)(H)2(PPh3)2, should be energically accessible [210]. When para-enriched hydrogen was added to Vaska’s complex at 295 K (Figure 1.35), the expected PHIP resonances were observed for cis-cis-trans-IrCl(CO)(H)2(PPh3)2, but a minor species could also be detected. This minor product was identified as the all-cis1.6.2.1

1.6 Special Topics

1

H NMR spectrum obtained upon addition of parahydrogen to Vaska’s complex trans-IrCl(CO)(PPh3)2 in C6D6 at 295 K, showing the extra species resulting from addition of H2 over the P–Ir–P axis. Figure 1.35

IrCl(CO)(H)2(PPh3)2, with two chemically equivalent but magnetically inequivalent hydrides [203, 211, 212]. Catalytic hydrogenation of alkynes by a monomeric transition metal complex was thought to yield exclusively the Z-alkene. However, Bargon and coworkers [213, 214], who have been active in PHIP research, observed the formation of E-alkenes using [RuCp*(alkene)]þ as catalyst. Figure 1.36 shows the 1H NMR spectrum after

Olefinic region of the 1H NMR spectrum for the product of the hydrogenation reaction of 3-hexyne-1-ol in presence of [RuCp*(alkene)]þ (alkene ¼ 3-hexenoic acid) and paraenriched hydrogen, under mild reaction conditions (300 K, 1 bar of H2) and low conversion rate. Figure 1.36

53

54


the hydrogenation of 3-hexyne-1-ol using para-enriched hydrogen and in the presence of [RuCp*(alkene)]þ (alkene ¼ 3-hexenoic acid). This spectrum is the result of a pairwise transfer of para-H2 to the substrate, and shows a 3J(1H,1H) of 15 Hz, typical of a trans coupling. The formation of E-alkene by trans hydrogenation of the substrate is undisputable, and unlikely with a single metal center catalyst. Initial formation of a Z-alkene was not detected, even at low temperature. Moreover, the Z-alkene was found not to isomerise in the presence of [RuCp*(alkene)]þ. A mechanism involving two ruthenium centers, as shown in Scheme 1.13, was therefore proposed. OTf

COOH

Ru

H2

R

H

H

R

3-hexenoic acid

+

R

Ru

H2

R

Ru

R

R H

R

Ru

Ru

H

R H

Ru

H R

R

Ru Ru

R

R H

Ru H

H H + Ru

Proposed mechanism for the hydrogenation of alkynes in the presence of [RuCp*(alkene)]þ (alkene ¼ 3-hexenoic acid), involving two ruthenium centers.

Scheme 1.13

1.6 Special Topics

* * H H Rh + P P

* * H H

P + Rh P

hydrogenation

S 1) polarization transfer COD

S

S -

P + S Rh P S

*

*

*

*

2) hydrogenation *

P + S* Rh * P *

S

* P + S* Rh P * * H H

-

*

*

*

*

, - H2 P P

S

* * H * + H* Rh S

Scheme 1.14 Possible mechanisms for the transfer of polarisation from parahydrogen onto cyclooctene via [Rh(COD)(dppb)]þ : (a) and (b) are the possible intermediate dihydride species responsible for the polarisation transfer at cyclooctene.

In the previous example, the addition of para-hydrogen at a transition metal has led to the observation of PHIP at the substrate [213, 214]. In a similar manner, Aime and coworkers [215] have studied the hydrogenation of cyclooctadiene from [Rh(COD)(dppb)]þ in the presence of parahydrogen, and observed strongly polarized hydrogen resonances at both the hydrogenated sites and the olefinic region of the 1H spectrum of the free cyclooctene formed. In view of these data, they suggested two possible mechanisms (Scheme 1.14), and showed that the enhanced absorption for the olefinic protons of cyclooctene is due to NOE transfer both within the para-hydrogenated substrate and the transition metal hydride complex formed during hydrogenation.

Parahydrogen as a Magnetic Probe The electronic spin state requirement for observation of PHIP at a transition metal center makes it a very good magnetic probe [202, 203, 216]. Ru(CO)2H2(dppe) (dppe ¼ Ph2PCH2CH2PPh2) is highly fluxional and undergoes a pairwise interchange of the two hydrides, the two P-atoms, and the two CO ligands at the same rate. H2 readily eliminates from the ruthenium center at high temperature (> 330 K) [216]. Para-H2 was used to investigate this reductive elimination, pointing to a diamagnetic intermediate, Ru(CO)2(dppe). Indeed, the 1H spectrum observed upon addition of parahydrogen to Ru(CO)2(dppe) showed strongly polarized 1.6.2.2

55

56


1

H NMR spectrum of the product of oxidative addition of H2 to Ru(CO)2(dppe), (a) with normal hydrogen, (b) with para-enriched hydrogen Figure 1.37

signals in the hydride region of the 1H spectrum, (Figure 1.37(b)). Therefore, Ru(CO)2(dppe) exists in a singlet electronic spin state (a paramagnetic intermediate would have quenched the enhancement through magnetic anisotropy-induced relaxation and no polarised resonance would have been observed). Further, this observation is consistent with an exchange mechanism involving a trigonal bipyramid transition state containing an h2 -coordinated hydrogen. 1.6.3

High Pressure NMR Introduction High gas pressures are widely used in many homogeneous catalytic processes, such as hydrogenations, hydroformylations or polymerizations. The use of pressure has a number of advantages: (a) increasing the concentration of the reactant gas (usually CO or H2) in solution, thereby achieving faster reaction rates; (b) controlling dynamic equilibria; (c) suppressing the boiling of a solvent at high temperature; or (d) avoiding decomposition of the catalyst. Although NMR is slow and insensitive compared to the well-established high pressure infrared spectroscopy [217], NMR under pressure is advantageous in that it is non-invasive and provides detailed structural information. Several reviews on the subject have been published, and the reader is advised to consult Chapter 2 for instrumental details [218–222]. 1.6.3.1

Applications The main applications of high pressure NMR (HP NMR) to homogeneous catalysis include [219–221]: 1.6.3.2

1. 2. 3. 4.

Monitoring reactions under conditions similar to the catalytic reaction. Stabilization and identification of intermediates. Measurement of kinetic and thermodynamic parameters. Investigation of reactions in supercritical fluids.

1.6 Special Topics

P HP NMR spectrum of 114. d 31P 138. The observed second-order spin system, AA’XX’, is consistent with the complete conversion of 113 into 114. Figure 1.38

31

Some of these applications will be illustrated in the next paragraphs, and many others can be found in the literature [223–225]. The first application of HP NMR to homogeneous catalysis was reported by Heaton, Jonas and coworkers [226, 227], who measured 13C under pressure of the Rh carbonyl clusters [Rh12(CO)30]2– and [Rh5(CO)15]–, involved in the catalytic synthesis of ethylene glycol from CO and H2. In a related application, the olefin hydroformylation catalyst precursors [Rh4(CO)12-x{P(OPh)3}x] (x ¼ 1–4) were studied using high pressures of CO and CO/H2, while monitoring 13C and 31P [228]. Figure 1.38 shows the 31P HP NMR spectrum for 114, formed when [Rh4(CO)8{P(OPh)3}4], 113, is submitted to 400 bar CO at 260 K. The dinuclear Rh complex, rac-[Rh2H2(m-CO)2(CO)2(et,ph,-P4)]2þ, 115, (et,ph-P4 ¼ Et2PCH2CH2)P(Ph)CH2P(Ph)CH2CH2PEt2), was identified by 1H HP NMR as the active catalyst in the regioselective Rh-catalyzed hydroformylation of 1-alkenes [229]. Figure 1.39 shows the hydride region of the 1H and 1H{31P} HP NMR spectra

Figure 1.39

catalyst 115.

Hydride region of the 1H and 1H{31P} HP NMR spectra of the hydroformylation

57

58


17 O, 13C and 1H HP NMR spectra of aqueous solutions of 116 and 117, (in 10 % [17O]water) under 60 bar ethylene at 298 K. For 116, the 17O NMR spectrum shows signals at d –160.1 and d –46.9, in ratio 4:1, corresponding to the equatorial and axial water oxygens, respectively. For 117 the 17O NMR spectrum shows two signals with 1:1 intensity ratio at d –49.52 and –132.1, which can be assigned to the cis,trans and cis,cis water oxygens, respectively.

Figure 1.40

of 115, acquired under 13.8 bar H2/CO, at 295 K. An extremely large 1J(109Rh,1H) coupling constant (164 Hz) was reported [229]. Additional applications involving CO [220, 230–238] and CO2 [239–241], plus an elegant combination of 103Rh NMR [242] together with high pressure have been reported. In the [Ru(H2O)6]2þ -catalyzed dimerisation of ethylene at 60 bar [243] 17O labeling and a multinuclear NMR approach (17O, 13C and 1H, see Figure 1.40) have allowed the identification of the reaction intermediates [Ru(CH2¼CH2)(H2O)5](tos)2, 116, and [Ru(CH2¼CH2)2(H2O)4](tos)2, 117, (tos ¼ toluene-p-sulfonate). As noted above, molecular hydrogen complexes are important intermediates in homogeneous catalytic hydrogenation reactions. When a solution of [Cr(CO)3(PCy3)2], 118, reacts with 28 bar H2 at ambient temperature, [Cr(h2 -H2)(CO)3(PCy3)2], 119, is formed (Eq. (13)) [244]. The rate of h2 -H2 elimination from 119 was determined by 1 H HP NMR inversion recovery experiments in which both the bound and dissolved H2 were measured (Figure 1.41) [244]. Although this is not a generally applicable method for obtaining rate information, it works in this case due to the large difference in the intrinsic relaxation rates of the two H2 species. The calculated rate constants are found to be independent of the H2 pressure, showing that the elimination is a first-order process. Cr(CO)3[P(C6H11)3]2 + H2 118

(η2-H2)Cr(CO)3[P(C6H11)3]2 119

1.6 Special Topics

Figure 1.41 1H HP NMR inversion-recovery experiments on dissolved and bound H2 in a d8 -toluene solution of [Cr(h2 -H2)(CO)3(PCy3)2], 119.

Iggo and coworkers have recently developed a high pressure NMR flow cell for the study of homogeneous reactions and reported several interesting applications [245, 246]. In the reaction of [RuCp(m-CO)2(m-dcpm)RhCl2] (dcpm ¼ (C6H11)2PCH2P(C6H11)2], 120, with CO, the intermediate bimetallic complex 121 could be detected by in situ 31P HP NMR (Figure 1.42) [247]. Previous attemps to detect 121 by high pressure infrared spectroscopy had proven unsuccessful [248].

Figure 1.42

31

P HP NMR investigation of the reaction of 120 with CO.

59

60


The toroidal pressure probe, introduced in 1989 by Rathke and coworkers [249, 250], has been modified by Woelk and coworkers [249, 251], who have used a toroid cavity NMR autoclave for high pressure PHIP NMR experiments. Figure 1.43 shows the PHIP spectrum of the [Rh(norbornadiene)(PPh3)2]PF6 -catalyzed hydrogenation of 1,4-diphenylbutadiyne with 40 bar of 50 % enriched para-H2 [252]. The spectrum from the same reaction at ambient H2 pressure is shown in the inset [253]. The two absorption/emission PHIP patterns in both spectra indicate that para-H2 is transferred pairwise during the catalytic cycle. A special area of HP NMR in catalysis involves supercritical fluids, which have drawn substantial attention in both industrial applications and basic research [249, 254, 255]. Reactions in supercritical fluids involve only one phase, thereby circumventing the usual liquid/gas mixing problems that can occur in conventional solvents. Further advantages of these media concern their higher diffusivities and lower viscosities [219]. The most commonly used supercritical phase for metal-catalyzed processes is supercritical CO2 (scCO2), due to its favorable properties [256– 260], i. e., nontoxicity, availability, cost, environmental benefits, low critical temperature and moderate critical pressure, as well as facile separation of reactants, catalysts and products after the reaction.

Figure 1.43 HP PHIP NMR spectrum of the [Rh(norbornadiene)(PPh3)2](PF6)-catalyzed hydrogenation of 1,4-diphenylbutadiyne with 40 bar of 50 % enriched p-H2, plus the spectrum from the same reaction at ambient H2 pressure (in the inset).

1.6 Special Topics

Figure 1.44 (a) 59Co NMR spectra of a 0.04 M solution of Co2(CO)8 in scCO2 (303 K) and liquid C6D6 (298 K) solutions (59Co quadrupole moment: Q ¼ 0.4 x 10 –28 cm2). (b) 187Re NMR spectrum of a solution of Re2(CO)10 in scCO2 (473 K, 145 bar) (187Re quadrupole moment: Q ¼ 2.6 x 10 –28 cm2).

Supercritical fluids possess several advantages from the NMR point of view [249]. The very low viscosity of these solvents produces a beneficial line-narrowing effect on quadrupolar nuclei such as 59Co, 14N, 53Cr, 91Zr, 95Mo, 55Mn and 103Re, due to their increased transverse relaxation times, T2 [219, 254, 261–264]. The improvement in observed line-widths can be clearly appreciated in Figure 1.44(a), which shows 59Co NMR spectra of the olefin hydroformylation catalyst Co2(CO)8 [249, 265]. While in C6D6 at 298 K the half-height line-width (Dn1/2) equals 30.0 kHz, in scCO2 at 305 K Dn1/2(59Co) is only 5.1 kHz. Hence, the distinction between cobalt complexes with similar 59Co chemical shifts, e. g. Co(C3H7C(O))(CO)4 and Co2(CO)8, is facilitated in the supercritical medium. Figure 1.44(b) reproduces the 187Re NMR spectrum from Re2(CO)10, which, due to the high quadrupole moment of 187Re, cannot be observed at all under routine conditions in normal solvents [249]. An even more useful property of supercritical fluids involves the near temperature-independence of the solvent viscosity and, consequently, of the line-widths of quadrupolar nuclei. In conventional solvents the line-widths of e. g. 59Co decrease with increasing temperature, due to the strong temperature-dependence of the viscosity of the liquid. These line-width variations often obscure chemical exchange processes. In supercritical fluids, chemical exchange processes are easily identified and measured [249]. As an example, Figure 1.45 shows 59Co line-widths of Co2(CO)8 in scCO2 for different temperatures. Above 160 oC, the line-broadening due to the dissociation of Co2(CO)8 to Co(CO)4 can be easily discerned [249]. There are an increasing number of applications of high pressure NMR in supercritical fluids to homogeneous catalysis [266]. Using their toroidal pressure probe, Rathke and coworkers [249, 267–269] have extensively studied the Co2(CO)8 -catalyzed hydroformylation of olefins in scCO2 (Eq. (14)). The hydrogenation of Co2(CO)8 (Eq. (15)) is a key step in this reaction.

RCH=CH2 + H2 + CO

Cat: Co2(CO)8

2 RCH2CH2CHO

61

62


Co2(CO)8 + H2

2 CoH(CO)4

Figure 1.45 Temperature dependence of the 59Co NMR line widths for a 4.8 mM solution of Co2(CO)8 in scCO2. 59

Co variable temperature HP NMR (Figure 1.46(a)) revealed an equilibrium reaction which exchanges the cobalt centers in Co2(CO)8 and CoH(CO)4, in the higher temperature region. This process does not broaden the 59Co signal of Co4(CO)12, even at 473 K (Figure 1.46(b)), nor the 1H signals of H2 or CoH(CO)4 at 453 K (Figure 1.46(c)). These observations point to a process involving dissociation of Co2(CO)8 into two Co(CO)4 radicals, followed by hydrogen atom transfer

Figure 1.46 (a) 59Co VT-NMR spectra for the reaction of Co2(CO)8 with H2 in scCO2. (b) 59Co NMR spectrum of a mixture containing Co4(CO)12, Co2(CO)8 and CoH(CO)4 at 473 K in scCO2. (c) 1H NMR spectrum at 453 K. Conditions as in (a).

1.6 Special Topics

from CoH(CO)4 to the radical Co(CO)4 (Eqs. (16) and (17)). Such a mechanism would interconvert the 59Co signals of Co2(CO)8 and HCo(CO)4 without affecting the 1H signals [249, 267, 268]. Co2(CO)8

2 Co(CO)4

CoH(CO)4 + Co(CO)4

Co(CO)4 + CoH(CO)4

In a related study, the hydroformylation of propene to n- and iso-butyraldehydes, catalyzed by Co2(CO)8, was followed by 1H (Figure 1.47(a)) and 59Co HP NMR (Figure 1.47(c)) [249, 265]. A typical 59Co spectrum at 353 K and 219 bar is shown in Figure 1.47(b) and reveals Co(C3H7C(O))(CO)4, Co2(CO)8 and CoH(CO)4. As shown in Figure 1.47(c), the cobalt complexes reach a nearly steady-state condition early in the hydroformylation, which persists during the 15 h period that the olefin is still present at significant levels. Woelk and coworkers [252, 270] have provided a detailed view into the activation and transfer of the dihydrogen molecule during hydrogenations in scCO2, using PHIP and their toroid cavity NMR autoclave. For the asymmetric hydrogenation

Figure 1.47 (a) Conversion of propylene (0.14 M) to n- and iso-butyraldehydes in scCO2 in the presence of Co2(CO)8 (0.017 M) at 353 K, using H2 and CO pressures of 42 bar each, followed by 1 H NMR. (b) In situ 59Co NMR spectrum showing the catalytic intermediates, near the steady state, during hydroformylation of propylene in scCO2 at 353 K. (c) Concentrations of catalytic intermediates during propylene hydroformylation in scCO2 at 353 K, followed by 59Co NMR

63

64


Figure 1.48 Left: standard 1H spectrum of a reaction mixture containing 122 and 123 (1:100) in scCO2 at 318 K and 150 bar. Right: PHIP spectrum of the same mixture after increasing the pressure to 180 bar with para H2. Chemical interactions between CO2 and reactive intermediates of the catalytic pathway can be excluded as the source of the different catalytic behavior in the supercritical medium with respect to usual solvents.

of 123 to yield 124, catalyzed by the rhodium complex 122 (Figure 1.48), it is concluded that the major catalytic pathway in scCO2 is very similar, if not identical, to that found in a nonprotic organic solvent of low polarity [270]. The Rh-catalyzed cyclization of the amine 125 in scCO2 affords the cyclic amine 127 as the major product (path A in Scheme 1.15), whereas in conventional solO path A H N [Rh]-H

125

[Rh] N H

O CO

[Rh] N H 126

Rh

+ P

(CH2)2(CF2)5CF3

- H2O

path B N in conventional

O O

+ [Rh]-H

O

F3C Cat:

in scCO2

+ H2 H N

3

solvents

+ [Rh]-H

128

F3C

Scheme 1.15 Different result of the Rh-catalyzed cyclization of 125 in scCO2 (path A) and in conventional solvents (path B).

N 127

1.6 Special Topics

Figure 1.49 Left: 1H NMR spectra of 125 in d8 -THF (298 K, left) and in scCO2/d8 -THF (180 bar, 323 K, right). Right: 14N NMR spectra of 129 in d6 -acetone (298 K, left) and scCO2/d6 -acetone (110 bar, 313 K, right; a smaller signal of a second species is visible under these conditions).

vents the cyclic amide 128 is formed preferentially (path B) [271]. HP 1H and 14N NMR measurements in scCO2 (Figure 1.49) have allowed the detection of an intermediate carbamic acid, 129, in this solvent. In the 1H NMR spectra, the broad signal of the NH proton of 125 at d 1.4 is replaced by a new signal at d 12.4, corresponding to 129, and the resonances of both CH2 protons directly adjacent to the N center of 125 exhibit a significant high frequency shift of 0.6 ppm in 129. In the 14N NMR spectra, an induced shift of 64 ppm in the major resonance indicates the formation of the carbamic acid as the major species. This reversible formation of a carbamic acid in scCO2 would reduce the tendency for intramolecular ring closure at the Rh-acyl intermediate 126, and thus suppress path B in Scheme 1.15, explaining the low formation of cyclic amides in the supercritical solvent. Supercritical CO2 would act simultaneously as solvent and temporary protecting group in this reaction [271]. 1.6.4

Diffusion and Pulsed Gradient Spin Echo Measurements

PGSE (Pulsed Gradient Spin Echo) NMR diffusion methods were introduced in 1965 by Stejskal and Tanner [272]. The basic pulse sequence is based on a spinecho with two incorporated pulsed field gradients (Figure 1.50). The effect of the two gradients is first, to defocus and, then, refocus the magnetization. However, if during the time D, the molecules diffuse from their positions after the first gradient, the effective magnetic field experienced will be different during both gradients. This results in incomplete refocusing of the spins and a decrease in the intensity of the detected NMR signals. Repetition of the experiment with increasing

65

66

1 NMR Spectroscopy and Homogeneous Catalysis Figure 1.50 Basic pulse sequence for PGSE diffusion measurements. G ¼ gradient strength, D ¼ delay between the midpoints of the gradients, D ¼ diffusion coefficient, d ¼ gradient length

gradient strengths, G, affords a set of signals from which the diffusion coefficient, D, can be obtained, according to Eq. (18). A typical plot is shown in Figure 1.51, for measurements on several different anions of a Ru(ii) Binap complex [273]. Larger molecules will diffuse more slowly than smaller molecules, and thus afford smaller slopes, as in Figure 1.51. More elaborate pulse sequences have been proposed [274–279]. From the diffusion coefficients a hydrodynamic radius, rh, and thus the molecular volume can be estimated, via the Stokes–Einstein equation (Eq. (19)). ln

I =– I0

2

G2

3

D

=

19 F PGSE diffusion measurements on several salts of a Ru(ii) p-cymene Binap complex. The larger the anion, the smaller the absolute value of the slope.

Figure 1.51 Plots of

1.6 Special Topics

67

The PGSE methodology presents several advantages with respect to other methods for estimating molecular size and diffusion coefficients [280]. PGSE measurements are fast, noninvasive and require only small samples. They allow a reasonably accurate determination of D values over several orders of magnitude, without the need to set up and maintain a concentration gradient. Several components of a mixture can be measured simultaneously (as long as they afford resolvable signals), which makes the technique especially valuable for materials which are not readily isolable, or mixtures of special interest. PGSE measurements have been widely applied in the investigation of small organic molecules [281], polymers [282, 283], surfactants [284], “container molecules” [285–288], supramolecular reagents [289], dendrimers [290, 291] and biomolecules [292–294]. There are also some few examples concerned with molecules in heterogeneous environments (e. g. in porous silica [295] and zeolites [296]). Recently, PGSE studies have been extended to the field coordination and/or organometallic chemistry, to address problems such as the formation of polynuclear complexes [76], ion pairing [4, 297, 298], hydrogen-bonding ligands [299], and otherwise aggregated species [115, 300, 301]. This subject has been reviewed [302–304]. Apart from nuclei with high receptivity (1H and 19F), PGSE measurements on inorganic and organometallic compounds can be successfully carried out with nuclei such as 31 P [298], 35Cl [298] or 7Li [305]. The first application of NMR diffusion measurements to determine the aggregation state of a transition metal catalyst concerned the chiral, tetranuclear Cu(i) catalysts 130–132, used in the conjugate addition reactions of anions to a,b-unsaturated cyclic ketones. Compounds 130–132 react with isonitriles to form 133–135, and do not degrade to lower molecular weight species (see Eq. (20)) [109]. Britzinger and coworkers [306] have studied the polymerization catalyst MAO/ ZrCp2Me2 in C6D6. The calculated effective hydrodynamic radius of 12.2–12.5 A

O

O

Ph

Ph Ph

Ph

Ph X

O

X

S

Cu

Ph Cu

Ph Ph O

S

S O

N

Ph

Cu

Cu Ph Ph X

S

Ph

Ph Ph

t

C

Ph X

BuCN

O

Cu

S X Ph Ph

O

d8 -THF

O

Cu Ph Ph C

Ph N O

O

X = OH (130), OMe (131), NMe2 (132)

X = OH (133), OMe (134), NMe2 (135)

68


at different zirconocene and MAO concentrations indicates that the ion pair 136 remains associated even at the lowest concentrations studied. At elevated concentrations, aggregation to ion quadrupoles or higher aggregates is indicated by an apparent size increase [306]. MeMAO Me

Me Al

Zr Me

Me 136

Interestingly, for the series of zirconocene catalysts [Zr]þ[MeB(C6F5)3]– Marks and coworkers [307] have found no evidence of significant aggregation to ion quadrupoles, as in Eq. (21). These authors have found that the tendency to form aggregates of higher nuclearity than simple ion-pairs is dependent on whether the anion is in the inner or outer coordination sphere of the metallocenium cation [308]. Me Me

C6D6

Cp 2Zr A−

Cp2Zr A−

A− ZrCp2 Me

The PGSE methodology has also been applied to study the dependence of enantioselectivity on the distribution of the chiral Rh-hydrogenation catalyst 137 between an aqueous and micellar phase. The observed increase in enantioselectivity when amphiphiles are added to the water is associated with an aggregation of the catalyst to the micelles [309]. H

Ph P 2

H

P Ph2

O

OTf

Rh(COD) OH

O 137

PGSE diffusion measurements have proved very valuable in studying ion-pairs. A relatively large number of cationic compounds are currently in use in homogeneous catalysis and/or organic synthesis. It has been shown that the counterion

1.6 Special Topics

may influence the rate and/or product distribution of some of these reactions [147, 310–315], as well as the stability of the compounds [316]. In principle, one can determine the diffusion coefficients for the cation and anion separately, and thus gain insight into whether they move together as a single unit (tight ion-pair) or separately. For anions such as PF6 –, BF4 –, OTf– or BArF–, 19F represents both an alternative and a complement to 1H PGSE methods. HOESY (Heteronuclear Overhauser Spectroscopy), and especially 1H,19F HOESY measurements, also help to localize the position of anions such as PF6 – or BArF–, relative to a catalytically active transition metal cation [116, 147, 304, 317–319]. The Ir(i) catalyst precursors 138a (X ¼ BArF–) and 138b (X ¼ B(C6F5)4 –) have been shown to hydrogenate tri-substituted olefins in CH2Cl2 with excellent enantioselectivity. With the smaller anions X ¼ PF6 –, BF4 – and OTf– (138c–d) the rate of reaction is much lower [320]. PGSE diffusion measurements on 138a–d in CDCl3 afford very similar D-values for cation and anion in each compound, pointing to a complete ion-pairing in this solvent (e. g., for 138c in CDCl3, D(cation) ¼ 7.13 x 10 –10 m2 s–1 and D(PF6 –) ¼ 7.21 x 10 –10 m2 s–1) [321]. In CD3OD, the cation and anion move separately, and in CD2Cl2, there seems to be a partial, but not complete, ion-pairing [321]. Figure 1.52 shows the results from PGSE measurements on the cations of 138a–d in CD2Cl2. For 138a and 138b (white circles) the cations move significantly more slowly (lower slope) than for 138c–d (black circles).

Figure 1.52 Plot of ln(I/Io) vs. arbitrary units proportional to the square of the gradient amplitude for 1H PGSE diffusion measurements on the cation of the five Ir(i) catalyst precursors 138a–d, in CD2Cl2.

H

N IrH2 P

N H2Ir

N =

N IrH2

P 87

P

P

O

Me P 2

N t-Bu

69

70


Clearly, the (partial) ion-pairing with the larger boron anions, BArF– and B(C6F5)4 –, decreases the mobility of the cation in 138a,b with respect to 138c–e, where the anions are smaller. It is known that mononuclear Ir complexes such as 138 react under hydrogen in solution to afford trinuclear hydrido cluster complexes such as 87, which have been shown not to be catalytically active in hydrogenation chemistry [110]. If the mechanism of the formation of these inactive Ir3 clusters requires that two fairly large species associate, and subsequently add yet another large moiety, then, in CH2Cl2, the lower mobility of 138a,b compared to 138c–e might explain the faster deactivation of the latter. PGSE diffusion measurements can also be presented as a “2D spectrum” where the chemical shift is displayed in the first dimension and the diffusion coefficient in the second one. Such an experiment is called DOSY (Diffusion Ordered Spectroscopy) [274, 322, 323] and has also been referred to as “NMR chromatography”, for its ability to facilitate and visualize the resolution and assignment of complex mixtures. Although used in several areas of chemistry, such as micelles [324], polymers [325–328], resins [329], biochemistry [330–332] and organic chemistry [333–336],

Figure 1.53 13C INEPT DOSY spectrum obtained during the reaction of 140 with 13CO2 at 195 K in d8 -THF. The sections show the signals of 141 (d 114.6 (Cp) and d 101.7 (CH2)) and 142 (d 114.9 (Cp) and d 63.5 (OMe)).

References

DOSY measurements have attracted very little attention in organometallic chemistry [337]. 3D versions of DOSY, where the diffusion dimension is added to a 2D experiment [338, 339], as well as heteronuclear detection, e. g. with 13C [340– 342], 31P [335], 29Si [343], and 7Li [337] have been reported recently. In a very recent example (Figure 1.53), 13C INEPT DOSY has been applied to confirm the dinuclear nature of the unstable Zirconium intermediate 141 in the reaction of 13CO with [ZrHCl(Cp)], 140 [344], a model reaction for the heterogeneously catalyzed hydrogenation of CO2 to methanol [345]. As was expected for a binuclear compound, the diffusion coefficient of intermediate 141 is smaller than for the mononuclear 142. Comment: Whether it is used for simple monitoring, or to prove the existence of traces of a key intermediate (EXSY or PHIP), NMR spectroscopy has clearly developed into one of the catalytic community’s most valuable analytical tools. NOE data will never afford a molecular picture which is quite as structurally exact as that from X-ray crystallography; however, this latter method cannot determine kinetic parameters, recognize equilibria, mimic catalytic conditions (HP) or recognize when ion-pairing is important (PGSE). NMR is not a very sensitive method; indeed, most methods are far more amenable to quantitative results. Nevertheless, its proven flexibility makes it indispensable.

References [1] N. Feiken, P. S. Pregosin, G. Trabe[2] [3] [4]

[5]

[6]

[7]

[8]

singer, Organometallics 1997, 16, 3735. D. Drago, P. S. Pregosin, A. Razavi, Organometallics 2000, 19, 1802. D. Drago, P. S. Pregosin, J. Chem. Soc., Dalton Trans. 2000, 3191. P. G. Anil Kumar, P. S. Pregosin, T. Schmid, G. Gonsiglio, Magn. Reson. Chem. 2004, 42, in press. J. Pawlas, Y. Nakao, M. Kawatsura, J. F. Hartwig, J. Am. Chem. Soc. 2002, 124, 3669. R. A. M. Robertson, A. D. Poole, M. J. Payne, D. J. Cole-Hamilton, J. Chem. Soc., Dalton Trans. 2000, 1817. G. C. Lloyd-Jones, S. W. Krska, D. L. Hughes, L. Gouriou, V. D. Bonnet, K. Jack, Y. Sun, R. A. Reamer, J. Am. Chem. Soc. 2004, 126, 702. T. Agapie, S. J. Schofer, J. A. Labinger, J. E. Bercaw, J. Am. Chem. Soc. 2004, 126, 1304.

[9] S. C. van der Slot, P. C. J. Kamer,

[10]

[11] [12]

[13]

[14]

[15] [16] [17]

P. W. N. M. van Leeuwen, Organometallics 2001, 20, 1079. T. Hayashi, K. Inoue, N. Taniguchi, M. Ogasawara, J. Am. Chem. Soc. 2001, 123, 9918. G. C. Lloyd-Jones, Org. Biomol. Chem. 2003, 1, 215. G. R. Eastham, R. P. Tooze, M. Kilner, D. F. Foster, D. J. Cole-Hamilton, J. Chem. Soc., Dalton Trans. 2002, 1613. R. Whyman, A. P. Wright, J. A. Iggo, B. T. Heaton, J. Chem. Soc., Dalton Trans. 2002, 771. H. Steinhagen, M. Reggelin, G. Helmchen, Angew. Chem. 1997, 109, 2199. J. M. Brown, P. A. Chaloner, J. Chem. Soc. Chem. Commun. 1980, 344. A. S. C. Chan, J. J. Pluth, J. Halpern, Inorg. Chim. Acta 1979, 37, L477. A. S. C. Chan, J. Halpern, J. Am. Chem. Soc. 1980, 102, 838.

71

72

1 NMR Spectroscopy and Homogeneous Catalysis [18] J. M. Brown, P. A. Chaloner, J. Am. [19]

[20]

[21]

[22] [23]

[24]

[25]

[26]

[27]

[28]

[29] [30] [31]

[32] [33] [34] [35]

Chem. Soc. 1980, 102, 3040. T. Hayashi, M. Takahashi, Y. Takaya, M. Ogasawara, J. Am. Chem. Soc. 2002, 124, 5052. C. R. Landis, K. A. Rosaaen, D. R. Sillars, J. Am. Chem. Soc. 2003, 125, 1710. A. L. Casado, P. Espinet, A. M. Gallego, J. M. Martnez-Harduya, Chem. Commun. 2001, 339. J. M. Brown, K. K. Hii, Angew. Chem. 1996, 108, 679. K. K. Hii, T. D. W. Claridge, J. M. Brown, Angew. Chem. Int. Ed. Engl. 1997, 36, 984. K. K. Hii, T. D. W. Claridge, J. M. Brown, A. Smith, R. J. Deeth, Helv. Chim. Acta 2001, 84, 3043. G. R. Eastham, B. T. Heaton, J. A. Iggo, R. P. Tooze, R. Whyman, S. Zacchini, Chem. Commun. 2000, 609; see J. Liu, B. T. Heaton, J. A. Iggo, R. Whyman, Angew. Chem. Int. Ed. Engl. 2004, 43, 90 for a comment on the carbomethoxy cycle. W. Clegg, G. R. Eastham, M. R. J. Elsegood, B. T. Heaton, J. A. Iggo, R. P. Tooze, R. Whyman, S. Zacchini, Organometallics 2002, 21, 1832. J. Wolowska, G. R. Eastham, B. T. Heaton, J. A. Iggo, C. Jacob, R. Whyman, Chem. Commun. 2002, 2784. K. L. Bray, J. P. H. Charmant, I. J. S. Fairlamb, G. C. Lloyd-Jones, Chem.-Eur. J. 2001, 7, 4205. A. L. Casado, P. Espinet, Organometallics 1998, 17, 954. A. L. Casado, P. Espinet, Organometallics 2002, 2003, 1305. L. M. Alcazar-Roman, J. F. Hartwig, A. L. Rheingold, L. M. Liable-Sands, I. A. Guzei, J. Am. Chem. Soc. 2000, 122, 4618. D. R. Sillars, C. R. Landis, J. Am. Chem. Soc. 2003, 125, 9894. C. R. Landis, J. Halpern, J. Am. Chem Soc. 1987, 109, 1746. J. F. Hartwig, C. N. Muhoro, Organometallics 2000, 19, 30. D. J. Tempel, L. K. Johnson, R. L. Huff, P. S. White, M. Brookhart, J. Am. Chem. Soc. 2000, 122, 6686.

[36] C. S. Shultz, J. Ledford, J. M. DeSi-

[37]

[38]

[39]

[40] [41] [42]

[43] [44]

[45]

[46] [47] [48]

[49] [50] [51]

[52] [53] [54]

[55]

mone, M. Brookhart, J. Am. Chem. Soc. 2000, 122, 6351. M. S. Sanford, J. A. Love, R. H. Grubbs, J. Am. Chem. Soc. 2001, 123, 6543. Z. J. A. Komon, J. S. Rogers, G. C. Bazan, Organometallics 2002, 21, 3189. B. T. Heaton, S. P. A. Hebert, J. A. Iggo, F. Metz, R. Whyman, J. Chem. Soc., Dalton Trans. 1993, 3081. J. W. Faller, D. G. DlAlliessi, Organometallics 2002, 21, 1743. J. A. Wiles, S. H. Bergens, Organometallics 1998, 17, 2228. J. A. Wiles, S. H. Bergens, K. P. M. Vanhessche, D. A. Dobbs, V. Rautenstrauch, Angew. Chem. Int. Ed. Engl. 2001, 40, 914. T. J. Geldbach, P. S. Pregosin, Helv. Chim. Acta 2002, 85, 3937. M. D. Leatherman, S. A. Svejda, L. K. Johnson, M. Brookhart, J. Am. Chem. Soc. 2003, 125, 3068. L. H. Shultz, D. J. Tempel, M. Brookhart, J. Am. Chem. Soc. 2001, 123, 11539. L. H. Shultz, M. Brookhart, Organometallics 2001, 20, 3975. D. J. Tempel, M. Brookhart, Organometallics 1998, 17, 2290. R. J. Deeth, A. Smith, K. K. Hii, J. M. Brown, Tetrahedron Lett. 1998, 39, 3229. M. Rubina, V. Gevorgyan, J. Am. Chem. Soc. 2001, 123, 11107. J. C. Conrad, G. P. A. Yap, D. E. Fogg, Organometallics 2003, 22, 1986. S. Aeilts, D. R. Cefalo, P. J. Bonitatebus, J. H. Houser, A. H. Hoveyda, R. R. Schrock, Angew. Chem. Int. Ed. Engl. 2001, 40, 1452. A. Albinati, P. S. Pregosin, F. Wombacher, Inorg. Chem. 1990, 29, 1812. P. S. Pregosin, F. Wombacher, Magn. Reson. Chem. 1991, 29, S106. P. S. Pregosin, in Encyclopedia of Nuclear Magnetic Resonance, Ed. B. Lever, Vol. 4, John Wiley & Sons Ltd., New York, 1996, p. 2549. P. S. Pregosin, ”Phosphorus-31 NMR Spectrol Properties in Compound

References

[56]

[57]

[58]

[59]

[60]

[61] [62]

[63]

[64]

[65] [66] [67]

[68]

[69]

[70] [71] [72] [73]

Characterization and Structural Analysis“, Vol. 8, Eds. J. G. Verkade, L. D. Quin, VCH, Deerfield Beach, Fl, 1987, p. 465. P. S. Pregosin, R. W. Kunz, 31P and 13 C NMR of Transition Metal Phosphine Complexes, Vol. 16, Springer Verlag, Berlin, 1979. D. Touchard, P. Haquette, S. Guesmi, L. LePichon, A. Daridor, L. Toupet, P. H. Dixneuf, Organometallics 1997, 16, 3640. M. Albrecht, J. R. Miecznikowski, A. Samuel, J. W. Faller, R. H. Crabtree, Organometallics 2002, 21, 3596. A. Haynes, B. E. Mann, G. E. Morris, P. Maitlis, J. Am. Chem. Soc. 1993, 115, 4093. D. C. Roe, R. E. Sheridan, E. E. Bunel, J. Am. Chem. Soc. 1994, 116, 1163. M. Kollmar, G. Helmchen, Organometallics 2002, 21, 4771. P. Dotta, P. G. A. Kumar, P. S. Pregosin, A. Albinati, S. Rizzato, Organometallics 2003, 22, 5345. E. Martinez-Viviente, P. S. Pregosin, M. Tschoerner, Magn. Reson. Chem. 2000, 38, 23. M. Tschoerner, R. W. Kunz, P. S. Pregosin, Magn. Reson. Chem. 1999, 37, 91. M. Tschoerner, P. S. Pregosin, Inorg. Chim. Acta 1999, 290, 95. D. Drago, P. S. Pregosin, Organometallics 2002, 21, 1208. N. Feiken, P. S. Pregosin, G. Trabesinger, A. Albinati, G. L. Evoli, Organometallics 1997, 16, 5756. P. Dotta, A. Magistrato, U. Rothlisberger, P. S. Pregosin, A. Albinati, Organometallics 2002, 21, 3033. M. A. Bennett, I. J. McMahon, S. Pelling, M. Brookhart, D. M. Lincoln, Organometallics 1992, 11, 127. M. Bennett, A. K. Smith, J. Chem. Soc., Dalton Trans. 1974, 233. J. W. Faller, D. G. DlAlliessi, Organometallics 2003, 22, 2749. J. W. Faller, J. Parr, Organometallics 2000, 19, 1829. B. Therrien, T. R. Ward, M. Pilington, C. Hofmann, f. Gilardoni, J. Weber, Organometallics 1998, 17, 330.

[74] P. J. Fagan, M. D. Ward, J. C. Ca-

[75]

[76]

[77]

[78]

[79]

[80] [81]

[82] [83] [84] [85]

[86] [87]

[88]

[89] [90] [91]

[92]

labrese, J. Am. Chem. Soc. 1989, 111, 1698. C. W. Cyr, S. J. Rettig, B. O. Patrick, B. R. James, Organometallics 2002, 21, 4672. T. J. Geldbach, P. S. Pregosin, A. Albinati, F. Rominger, Organometallics 2001, 20, 1932. Y. Chen, M. Valentini, P. S. Pregosin, A. Albinati, Inorg. Chim. Acta 2002, 327, 4. M. Kuriyama, K. Nagai, K. Yamada, Y. Miwa, T. Taga, K. Tomioka, J. Am. Chem. Soc. 2002, 124, 8932. C. J. den Reijer, P. Dotta, P. S. Pregosin, A. Albinati, Can. J. Chem. 2001, 79, 693. Q. Jaing, H. Regger, L. M. Venanzi, Inorg. Chim. Acta 1999, 290, 64. J. G. Verkade, J. A. Mosbo, in Phosphorus-31 NMR Spectroscopy in StereoChemical Analysis, Vol. 8, Eds. J. G. Verkade, L. D. Quin, VCH, Deefield Beach, Fl, 1987, p. 425. A. K. Roy, R. B. Taylor, J. Am. Chem. Soc. 2002, 124, 9510. W. von Philipsborn, Pure Appl. Chem. 1986, 58, 513. W. von Philipsborn, Chem. Soc. Rev. 1999, 28, 95. J. S. Z. Sabounchei, B. T. Heaton, J. A. Iggo, C. Jacob, I. S. Podkorytov, J. Cluster Sci. 2001, 12, 339. T. Eguchi, B. T. Heaton, J. Chem. Soc., Dalton Trans. 1999, 3523. B. T. Heaton, J. A. Iggo, I. S. Podkorytov, D. J. Smawfield, S. P. Tunik, R. Whyman, J. Chem. Soc., Dalton Trans. 1999, 1917. S. P. Tunik, I. S. Podkorytov, B. T. Heaton, J. A. Iggo, J. Sampanthar, J. Organomet. Chem., 1998, 550, 221. B. E. Mann, C. Spencer, Inorg. Chim. Acta 1982, 65, L57. B. E. Mann, C. Spencer, Inorg. Chim. Acta 1983, 76, L65. J. Ruiz, P. O. Bentz, B. E. Mann, C. Spencer, B. F. Taylor, P. Maitlis, J. Chem. Soc., Dalton Trans. 1987, 2709. K. Zamaraevl, J. Mol. Catal. 1993, 82, 275.

73

74

1 NMR Spectroscopy and Homogeneous Catalysis [93] For a useful study relating allyl

[109] A. Pichota, P. S. Pregosin, M. Valen-

13

[94]

[95]

[96]

[97]

[98]

[99]

[100]

[101]

[102]

[103]

[104]

[105] [106]

[107] [108]

C chemical shifts in Pd(ii) complexes to trans influence, see B. Akermark, B. S. H. Krakenberger, A. Vitagliano, Organometallics 1987, 6, 620. C. A. Tolman, P. Z. Meakin, D. L. Lindner, J. P. Jesson, J. Am. Chem. Soc. 1974, 96, 2762. P. Dani, M. A. M. Toorneman, G. P. M. van Klink, G. van Koten, Organometallics 2000, 19, 5287. L. Johansson, M. Tilset, J. A. Labinger, J. E. Bercaw, J. Am. Chem. Soc. 2000, 122, 10846. A. Albinati, C. Ammann, P. S. Pregosin, H. Ruegger, Organometallics 1990, 9, 1826. A. Albinati, R. W. Kunz, C. J. Ammann, P. S. Pregosin, Organometallics 1991, 10, 1800. H. Ruegger, R. W. Kunz, C. J. Ammann, P. S. Pregosin, Magn. Reson. Chem. 1991, 29, 197. R. E. Blumer, F. Lianza, P. S. Pregosin, H. Ruegger, A. Togni, Inorg. Chem. 1993, 32, 2663. P. S. Pregosin, H. Ruegger, R. Salzmann, A. Albinati, F. Lianza, R. W. Kunz, Organometallics 1994, 13, 5040. P. von Matt, G. C. Lloyd-Jones, A. B. E. Minidis, A. Pfaltz, L. Macko, M. Neuburger, M. Zehnder, H. Regger, P. S. Pregosin, Helv. Chim. Acta 1995, 78, 265. P. Barbaro, P. S. Pregosin, R. Salzmann, A. Albinati, R. W. Kunz, Organometallics 1995, 14, 5160. P. S. Pregosin, R. Salzmann, A. Togni, Organometallics 1995, 14, 842. P. S. Pregosin, R. Salzmann, Coord. Chem. Rev. 1996, 155, 35. G. Trabesinger, A. Albinati, N. Feiken, R. W. Kunz, P. S. Pregosin, M. Tschoerner, J. Am. Chem. Soc. 1997, 119, 6315. P. S. Pregosin, G. Trabesinger, J. Chem. Soc., Dalton Trans. 1998, 727. P. S. Pregosin, in Advances in BioChirality, Eds. G. Palyi, C. Zucchi, L. Caglioti, Elsevier, Amsterdam, 1999, pp. 335-345.

[110]

[111]

[112]

[113]

[114]

[115]

[116] [117]

[118]

[119]

[120] [121]

[122]

[123]

[124] [125] [126]

[127]

tini, M. Wrle, D. Seebach, Angew. Chem. Int. Ed. Engl. 2000, 112, 153. S. P. Smidt, A. Pfaltz, E. MartinezViviente, P. S. Pregosin, A. Albinati, Organometallics 2003, 22, 1000. T. J. Geldbach, H. Ruegger, P. S. Pregosin, Magn. Reson. Chem. 2003, 41, 703. J. S. Giovannetti, C. M. Kelly, C. R. Landis, J. Am. Chem. Soc. 1993, 115, 4040. M. Kollmar, B. Goldfuss, M. Reggelin, F. Rominger, G. Helmchen, Chem. Eur. J. 2001, 7, 4913. H. Steinhagen, M. Reggelin, G. Helmchen, Angew. Chem. Int. Ed. Engl. 1997, 36, 2108. D. Zuccaccia, S. Sabatini, G. Bellachioma, G. Cardaci, E. Clot, A. Macchioni, Inorg. Chem. 2003, 42, 5465. A. Macchioni, Eur. J. Inorg. Chem. 2003, 195. C. Zuccaccia, G. Bellachioma, G. Cardaci, A. Macchioni, J. Am. Chem. Soc. 2001, 123, 11020. A. Macchioni, G. Bellachioma, G. Cardaci, G. Cruciani, E. Foresti, S. P., C. Zuccaccia, Organometallics 1998, 17, 5549. A. Macchioni, G. Bellachioma, G. Cardaci, V. Gramlich, H. Ruegger, S. Terenzi, L. M. Venanzi, Organometallics 1997, 16, 2139. J. M. Brown, Chem. Soc. Rev. 1993, 25. J. M. Brown, D. I. Hulmes, P. J. Guiry, Tetrahedron 1994, 50, 4493. J. M. Brown, J. Perez-Torrente, N. Alcock, H. J. Clase, Organometallics 1995, 14, 207. J. M. Brown, J. J. Perez-Torrente, N. W. Alcock, Organometallics 1995, 14, 1195. C. J. A. Daley, S. H. Bergens, J. Am. Chem. Soc. 2002, 124, 3680. J. A. Wiles, S. H. Bergens, Organometallics 1999, 18, 3709. J. M. Canal, M. Gomez, F. Jimenez, M. Rocomora, G. Muller, E. Dunach, D. Franco, A. Jimenez, F. H. Canos, Organometallics 2000, 19, 966. S. K. Mandal, G. A. N. Gowda, S. S. Krishnamurthy, C. Zheng,

References

[128]

[129]

[130]

[131] [132] [133] [134]

[135]

[136] [137] [138]

[139] [140] [141]

[142] [143]

[144]

[145] [146]

S. J. Li, N. S. Hosmane, Eur. J. Inorg. Chem. 2002, 2047. J. M. Brown, P. A. Chaloner, G. A. Morris, J. Chem. Soc., Perkin Trans. 2 1987, 1583. P. S. Pregosin, H. Ruegger, R. Salzmann, A. Albinati, F. Lianza, R. W. Kunz, Organometallics 1994, 13, 83. R. M. Bullock, B. K. Bender, in Encyclopedia of Catalysis, Ed. I. T. Horvath, Wiley, New York, 2002. D. A. Singleton, A. A. Thomas, J. Am. Chem. Soc. 1995, 117, 9357. R. H. Grubbs, G. W. Coates, Acc. Chem. Res. 1996, 29, 85. C. R. Landis, K. A. Rosaaen, J. Uddin, J. Am. Chem. Soc. 2002, 124, 12062. M. J. Tanner, M. Brookhart, J. M. DeSimone, J. Am. Chem. Soc. 1997, 119, 7617. C. P. Casey, S. W. Singer, D. R. Powell, R. K. Hayashi, M. Kavana, J. Am. Chem Soc. 2001, 123, 1090. W. Leitner, J. M. Brown, H. Brunner, J. Am. Chem. Soc. 1993, 115, 152. P.-E. Hansen, Annu. Rep. NMR Spectrosc. 1984, 15, 105. D. D. Wick, K. A. Reynolds, W. D. Jones, J. Am. Chem. Soc. 1999, 121, 3974. P. Burger, R. G. Bergman, J. Am. Chem. Soc. 1993, 115, 10462. Z. X. Liu, E. Somsook, C. R. Landis, J. Am. Chem. Soc. 2001, 123, 2915. Z. X. Liu, E. Somsook, C. B. White, K. A. Rosaaen, C. R. Landis, J. Am. Chem. Soc. 2001, 123, 11193. J. Sandstrm, Dynamic NMR Spectroscopy, Academic Press, London, 1982. N. Chandrakumar, S. Subramanian, Modern Techniques in High Resolution FT-NMR, Springer-Verlag, New York, 1987. H. Kessler, M. Gehrke, C. Griesinger, Angew. Chem. Int. Ed. Engl. 1988, 27, 490. C. L. Perrin, T. J. Dwyer, Chem. Rev. 1990, 90, 935. H. Gnther, NMR Spectroscopy: Basic Principles, Concepts, & Applications in Chemistry, 2nd edn., John Wiley & Sons, New York 1995.

[147] A. Macchioni, G. Bellachioma,

[148]

[149] [150]

[151]

[152]

[153]

[154]

[155] [156]

[157] [158] [159]

[160]

[161]

[162] [163]

G. Cardaci, M. Travaglia, C. Zuccaccia, B. Milani, G. Corso, E. Zangrando, G. Mestroni, C. Carfagna, M. Formica, Organometallics 1999, 18, 3061. J. Ledford, C. S. Shultz, D. P. Gates, P. S. White, J. M. DeSimone, M. Brookhart, Organometallics 2001, 20, 5266. P. A. Deck, C. L. Beswick, T. J. Marks, J. Am. Chem Soc. 1998, 120, 1772. S. J. Skoog, C. Mateo, G. G. Lavoie, F. J. Hollander, R. G. Bergman, Organometallics 2000, 19, 1406. F. Teixidor, M. A. Flores, C. Vinas, R. Sillanpaeae, R. Kivekaes, J. Am. Chem Soc. 2000, 122, 1963. H. Valentini, K. Selvakumar, M. Worle, P. S. Pregosin, J. Organomet. Chem. 1999, 587, 244. Z. Zhou, G. Facey, B. R. James, H. Alper, Organometallics 1996, 15, 2496. V. K. Dioumaev, B. R. Yoo, L. J. Procopio, P. J. Carroll, D. H. Berry, J. Am. Chem Soc. 2003, 125, 8936. M. L. H. Green, L. L. Wong, A. Sella, Organometallics 1992, 11, 2660. S. A. Baker, W. E. Hill, C. A. McAuliffe, J. Chem. Soc., Dalton Trans. 1986, 129, 7. D. Seebach, J. F. M. Oth, J. Heinzer, Helv. Chim. Acta 1985, 8, 1848. A. D. Bain, G. J. Duns, J. Magn. Reson. Ser. A 1995, 112, 258. L. M. Jackman, F. A. Cotton, Dynamic Nuclear Magnetic Resonance Spectroscopy, Academic Press, New York, 1975. J. A. Iggo, NMR Spectroscopy in Inorganic Chemistry, Oxford University Press, Oxford, 2000. D. S. Stephenson, G. Binsh, DNMR5, Quantum Chemistry Program Exchange; QCPE 365. Modified by: Le Master, C. B.; Le Master, C. L.; True, N. S., Quantum Chemistry Program Exchange; QCPE 059. Ivory-Soft, g-NMR v. 5. 0. NMR simulation package 2001. M. Q. Slagt, R. J. M. Klein Gebbink, M. Lutz, A. L. Spek, G. van Koten,

75

76


[164]

[165]

[166] [167]

[168]

[169] [170]

[171]

[172]

[173]

[174] [175] [176]

[177] [178]

[179]

[180] [181]

J. Chem Soc., Dalton Trans. 2002, 2591. J. A. M. Brandts, E. Kruiswijk, J. Boersma, A. L. Spek, G. van Koten, J. Organomet. Chem. 1999, 585, 93. I. G. Jung, S. U. Son, K. H. Park, K.-C. Chung, J. W. Lee, Y. K. Chung, Organometallics 2003, 22, 4715. C. P. Casey, D. W. Carpenetti, H. Sakurai, Organometallics 2001, 20, 4262. D. Heller, R. Kadyrov, M. Michalik, T. Freier, U. Schmidt, H. W. Krause, Tetrahedron: Asymmetry 1996, 7, 3025. J. M. Brown, P. A. Chaloner, G. A. Morris, J. Chem. Soc., Chem. Commun. 1983, 664. R. Kadyrov, A. Borner, R. Selke, Eur. J. Inorg. Chem. 1999, 705. O. N. Bruker-Franzen Analytik Gmbh and K. Illyasov, WIN-DYNAMICS 1.0 Release 951220, NMR Dynamic Spectra Simulation and Iteration. J. G. Knight, S. Doherty, A. Harriman, E. G. Robins, M. Betham, G. R. Eastham, R. P. Tooze, M. R. J. Elsegood, P. Champkin, W. Clegg, Organometallics 2000, 19, 4957. J. K. M. Sanders, B. K. Hunter, Modern NMR Spectroscopy, Oxford University Press, Oxford, 1987. J. W. Faller, in The Determination of Organic Structures by Physical Methods, Vol. 5, Chap. 2, Academic Press, New York, 1973. K. G. Orrell, in Annual Reports on NMR Spectroscopy, Vol. 37, 1999, pp. 1. S. Schaeublin, A. Wokaun, R. R. Ernst, J. Magn. Reson. 1977, 27, 273. S. Schaeublin, A. Hoehener, R. R. Ernst, J. Magn. Reson. 1974, 13, 196. J. Halpern, Science 1982, 217, 401. H. B. Kagan, G. Wilkinson, F. G. A. Stone, in Comprehensive Organometallic Chemistry, Vol. 8, Pergamon Press, Oxford, 1982, p. 463. H. Bircher, B. R. Bender, W. v. Philipsborn, Magn. Reson. Chem. 1993, 31, 293. L. Cronin, C. L. Higgitt, R. N. Perutz, Organometallics 2000, 19, 672. A. Togni, Angew. Chem. Int. Ed. Engl. 1996, 108, 1581.

[182] G. Pioda, A. Togni, Tetrahedron

Asymmetry 1998, 9, 3903. [183] T. K. Woo, G. Pioda, U. Rothlisberger,

[184]

[185]

[186] [187] [188]

[189]

[190]

[191]

[192]

[193] [194]

[195]

[196] [197] [198]

[199]

A. Togni, Organometallics 2000, 19, 2144. J. Herrmann, P. S. Pregosin, R. Salzmann, A. Albinati, Organometallics 1995, 14, 3311. P. Dotta, P. G. A. Kumar, P. S. Pregosin, Magn. Reson. Chem. 2002, 40, 653. J. L. Farrugia, J. Chem. Soc., Dalton Trans. 1997, 1783. W. Hieber, Adv. Organomet. Chem. 1970, 8, 1. N. Chatani, Y. Ishii, Y. Ie, F. Kakituchi, S. Murai, J. Org. Chem. 1998, 63, 5129. G. Gervasio, R. Giordano, D. Marabello, E. Sappa, J. Organomet. Chem. 1999, 588, 83. U. Frey, L. Helm, A. E. Merbach, R. Roulet, Advanced Applications of NMR to Organometallic Chemistry, J. Wiley & Sons Ltd, Chichester, 1996. E. Abel, K. G. Orrell, S. P. Scanlan, D. Stephenson, T. Kemmit, W. J. Levason, J. Chem. Soc., Dalton Trans. 1991, 591. D. Shaw, Fourier transform NMR spectroscopy, Elsevier, Amsterdam, 1976. R. H. Crabtree, Acc. Chem. Res. 1990, 23, 95. G. J. Kubas, R. R. Ryan, B. I. Swanson, P. J. Vergamini, H. J. Wasserman, J. Am. Chem. Soc. 1984, 106, 451. P. J. Desrosiers, L. Cai, Z. Lin, R. Richards, J. Halpern, J. Am. Chem. Soc. 1991, 113, 4173. R. H. Morris, R. J. Wittebort, Magn. Reson. Chem. 1997, 35, 243. V. I. Bakhmutov, Magn. Reson. Chem. 2004, 42, 66. V. I. Bakhmutov, C. Bianchini, F. Maseras, A. Lledos, M. Peruzzini, E. V. Vorontsov, Chem.-Eur. J. 1999, 5, 3318. J. A. Ayllon, S. Gervaux, S. Sabo-Etienne, B. Chaudret, Organometallics 1997, 16, 2000.

References [200] K. A. Earl, G. Jia, P. A. Maltby,

[201]

[202] [203] [204] [205] [206] [207]

[208] [209] [210] [211]

[212]

[213] [214] [215]

[216]

[217]

[218]

R. H. Morris, J. Am. Chem Soc. 1991, 113, 3027. S. B. Duckett, C. L. Newell, R. Eisenberg, J. Am. Chem Soc. 1994, 116, 10548. S. B. Duckett, C. J. Sleigh, Prog. Nucl. Magn. Reson. Spectrosc. 1999, 34, 71. S. B. Duckett, D. Blazina, Eur. J. Inorg. Chem. 2003, 2901. C. R. Bowers, D. P. Weitekamp, J. Am. Chem Soc. 1987, 109, 5541. K. F. Bonhoeffer, P. Harteck, Z. Phys. Chem-Leipzig 1929, B4, 113. J. Natterer, J. Bargon, Prog. Nucl. Magn. Reson. Spectrosc. 1997, 31, 293. P. Hubler, R. Giernoth, G. Kummerle, J. Bargon, J. Am. Chem Soc. 1999, 121, 5311. L. Vaska, M. F. Werneke, Trans. N. Y. Acad. Sci. 1971, 33, 70. R. Ugo, S. Cenini, A. Fusi, A. Pasini, J. Am. Chem Soc. 1972, 94, 7364. A. L. Sargent, M. B. Hall, Inorg. Chem. 1992, 31, 317. S. K. Hasnip, S. B. Duckett, C. J. Sleigh, D. R. Taylor, G. K. Barlow, M. J. Taylor, Chem. Commun. 1999, 1717. S. K. Hasnip, S. A. Colebrooke, C. J. Sleigh, S. B. Duckett, D. R. Taylor, G. K. Barlow, M. J. Taylor, J. Chem. Soc., Dalton Trans. 2002, 743. D. Schleyer, H. G. Niessen, J. Bargon, New J. Chem. 2001, 25, 423. M. Haake, J. Natterer, J. Bargon, J. Am. Chem Soc. 1996, 118, 8688. S. Aime, D. Canet, W. Dastru, R. Gobetto, F. Reineri, A. Viale, J. Phys. Chem. A 2001, 105, 6305. D. Schott, C. J. Sleigh, J. P. Lowe, S. B. Duckett, R. J. Mawby, M. G. Partridge, Inorg. Chem. 2002, 41, 2960. R. Whyman, in Laboratory Methods in Vibrational Spectroscopy, Eds. H. A. Willis, J. H. Van der Mass, R. G. J. Miller, Wiley, New York, 1987. J. E. Jonas, in “High Pressure NMR in NMR Basic Principles and Progress”, Vol. 24, p. 85, Eds. P. Diehl, E. Fluck, H. Gnther, R. Kosfeld, J. Selig Vol. 24, Springer, Berlin, 1991.

[219] I. T. Horvath, J. M. Millar, Chem. Rev.

1991, 91, 1339. [220] C. J. Elsevier, J. Mol. Catal. 1994, 92,

285. [221] D. C. Roe, P. M. Kating, P. J. Krusic,

B. E. Smart, Top. Catal. 1998, 5, 133. [222] D. C. Roe, in Homogeneous Transition

[223] [224]

[225]

[226]

[227]

[228]

[229]

[230] [231]

[232] [233] [234]

[235]

[236]

[237]

Metal Catalyzed Reactions, Eds. W. R. Moser, D. W. Slocum, American Chemical Society, Washington, 1992. T. B. Marder, D. C. Roe, D. Milstein, Organometallics 1988, 7, 1451. M. Brookhart, A. F. J. Volpe, D. M. Lincoln, I. T. Horvath, J. M. Millar, J. Am. Chem. Soc. 1990, 112, 5634. K. T. Moore, I. T. Horvath, M. J. Therien, J. Am. Chem. Soc. 1997, 119, 1791. B. T. Heaton, L. Strona, J. Jonas, T. Eguchi, G. A. Hoffmann, J. Chem. Soc., Dalton Trans. 1982, 1159. B. T. Heaton, J. Jonas, T. Eguchi, G. A. Hoffmann, J. Chem. Soc., Chem. Commun. 1981, 331. D. T. Brown, T. Eguchi, B. T. Heaton, J. A. Iggo, R. Whyman, J. Chem. Soc., Dalton Trans. 1991, 677. R. C. Matthews, D. K. Howell, P. W., S. G. Train, W. D. Treleaven, G. G. Stanley, Angew. Chem. Int. Ed. Engl. 1996, 35, 2253. D. C. Roe, Organometallics 1987, 6, 942. I. T. Horvath, R. V. Kastrup, A. A. Oswald, E. J. Mozeleski, Catal. Lett. 1989, 2, 85. I. Toth, T. Kegel, C. J. Elsevier, L. Kollar, Inorg. Chem. 1994, 33, 5708. I. Toth, C. J. Elsevier, J. Am. Chem. Soc. 1993, 115, 10388. G. Dekker, C. J. Elsevier, K. Vrieze, P. Vanleeuwen, Organometallics 1992, 11, 1598. J. M. Millar, R. V. Kastrup, S. Harris, I. T. Horvath, Angew. Chem. Int. Ed. Engl. 1990, 194. M. Poyatos, P. Uriz, J. A. Mata, C. Claver, E. Fernandez, E. Peris, Organometallics 2003, 22, 440. C. Bianchini, H. M. Lee, A. Meli, F. Vizza, Organometallics 2000, 19, 849.

77

78

1 NMR Spectroscopy and Homogeneous Catalysis [238] K. Nozaki, T. Hiyama, S. Kacker,

[239]

[240]

[241]

[242]

[243] [244]

[245]

[246]

[247] [248]

[249]

[250] [251] [252]

[253] [254]

I. T. Horvath, Organometallics 2000, 19, 2031. D. J. Darensbourg, H. P. Wiegreffe, P. W. Wiegreffe, J. Am. Chem. Soc. 1990, 112, 9252. D. J. Darensbourg, P. Wiegreffe, C. G. Riordan, J. Am. Chem. Soc. 1990, 112, 5759. D. J. Darensbourg, B. L. Mueller, C. J. Bischoff, S. S. Chojnacki, J. H. Reibenspies, Inorg. Chem. 1991, 30, 2418. F. R. Bregman, J. M. Ernsting, F. Muller, M. D. K. Boele, L. A. van der Veen, C. J. Elsevier, J. Organomet. Chem. 1999, 592, 306. G. Laurenczy, A. Merbach, J. Chem. Soc., Chem. Commun. 1993, 187. J. M. Millar, R. V. Kastrup, M. T. Melchior, I. T. Horvath, C. D. Hoff, R. H. Crabtree, J. Am. Chem. Soc. 1990, 112, 9643. J. A. Iggo, Y. Kawashima, J. Liu, T. Hiyama, K. Nozaki, Organometallics 2003, 22, 5418. S. C. van der Slot, P. C. J. Kamer, P. van Leeuwen, J. A. Iggo, B. T. Heaton, Organometallics 2001, 20, 430. J. A. Iggo, D. Shirley, N. C. Tong, New J. Chem. 1998, 22, 1043. P. S. Bearman, A. K. Smith, N. C. Tong, R. Whyman, Chem. Commun. 1996, 2061. J. W. Rathke, R. J. Klingler, R. E. Gerald, K. W. Kramarz, K. Woelk, Prog. Nucl. Magn. Reson. Spectrosc. 1997, 30, 209. J. W. Rathke, J. Magn. Reson. 1989, 85, 150. K. Woelk, J. W. Rathke, R. J. Klingler, J. Magn. Reson. Ser. A 1994, 109, 137. H. G. Niessen, P. Trautner, S. Wiemann, J. Bargon, K. Woelk, Rev. Sci. Instrum. 2002, 73, 1259. J. Bargon, J. Kandels, K. Woelk, Z. Phys. Chem. 1993, 180, 65. J. W. Rathke, R. J. Klingler, R. E. Gerald, D. E. Fremgen, K. Woelk, S. Gaemers, C. J. Elsevier, in Chemical Synthesis Using Supercritical Fluids, Eds. P. G. Jessop,

[255] [256] [257]

[258] [259] [260]

[261] [262]

[263] [264] [265]

[266]

[267] [268]

[269]

[270]

[271]

[272] [273]

W. Leitner, Wiley-VCH, Weinheim, 1999, p. 165. P. G. Jessop, T. Ikariya, R. Noyori, Chem. Rev. 1999, 99, 475. S. L. Wells, J. M. DeSimone, Angew. Chem. Int. Ed. Engl. 2001, 40, 518. Chemical Synthesis Using Supercritical Fluids, Eds. P. G. Jessop, W. Leitner, Wiley-VCH, Weinheim, 1999. P. G. Jessop, T. Ikariya, R. Noyori, Science 1995, 269, 1065. P. G. Jessop, T. Ikariya, R. Noyori, Nature 1994, 368, 231. W. Leitner, in Modern Solvents in Organic Synthesis, in Topics in Current Chemistry, Ed. P. Knochel, SpringerVerlag, Berlin, Vol. 206, 1999, p. 107. J. M. Robert, R. F. Evilia, J. Am. Chem. Soc. 1985, 107, 3733. D. M. Lamb, S. T. Adamy, K. W. Woo, J. Jonas, J. Phys. Chem. 1989, 93, 5002. S. Gaemers, J. Groenevelt, C. J. Elsevier, Eur. J. Inorg. Chem. 2001, 829. S. Gaemers, C. J. Elsevier, Chem. Soc. Rev. 1999, 28, 135. J. W. Rathke, R. J. Klingler, T. R. Krause, Organometallics 1991, 10, 1350. H. G. Niessen, A. Eichhorn, K. Woelk, J. Bargon, J. Mol. Catal. A 2002, 182–183, 463. R. J. Klingler, J. W. Rathke, J. Am. Chem. Soc. 1994, 116, 4772. J. W. Rathke, R. J. Klingler, T. R. Krause, Organometallics 1992, 11, 585. K. W. Kramarz, R. J. Klingler, D. E. Fremgen, J. W. Rathke, Catal. Today 1999, 49, 339. S. Lange, A. Brinkmann, P. Trautner, K. Woelk, J. Bargon, W. Leitner, Chirality 2000, 12, 450. K. Wittmann, W. Wisniewski, R. Mynott, W. Leitner, C. L. Kranemann, T. Rische, P. Eilbracht, S. Kluwer, J. M. Ernsting, C. L. Elsevier, Chem.Eur. J. 2001, 7, 4584. E. O. Stejskal, J. E. Tanner, J. Chem. Phys. 1965, 42, 288. P. G. Anil Kumar, P. S. Pregosin, J. M. Goicoechea, M. K. Whittlesey, Organometallics 2003, 22, 2956.

References [274] C. S. J. Johnson, Prog. NMR Spectrosc. [275] [276]

[277] [278] [279] [280]

[281] [282] [283] [284] [285]

[286] [287] [288]

[289]

[290]

[291]

[292] [293]

[294]

[295]

1999, 34, 203. P. Stilbs, Prog. NMR Spectrosc. 1987, 19, 1. C. S. J. Johnson, in Encyclopedia of Nuclear Magnetic Resonance, Eds. D. M. Grant, R. K. Harris, Wiley, Chichester, 1996, pp. 1626. W. S. Price, Concepts Magn. Reson. 1998, 10, 197. W. S. Price, Concepts Magn. Reson. 1997, 9, 299. J. Krger, H. Pfeifer, W. Heink, Adv. Magn. Reson. 1988, 12, 1. H. J. V. Tyrrell, K. R. Harris, Diffusion in Liquids Butterworths, London, 1984. P. Stilbs, Anal. Chem. 1981, 53, 2135. E. E. van Meerwal, Adv. Polym. Sci. 1984, 54, 1. W. Brown, P. Stilbs, Polymer 1982, 24, 188. P. Griffiths, P. Stilbs, Curr. Opin. Colloid Interface Sci. 2002, 7, 249. L. Frish, S. E. Matthews, V. Bhmer, Y. Cohen, J. Chem. Soc., Perkin Trans. 2 1999, 669. K. S. Cameron, L. Fielding, J. Org. Chem., 2001, 66, 6891. L. Avram, Y. Cohen, J. Am. Chem Soc. 2003, 125, 16180. L. Frish, F. Sansone, A. Casnati, R. Ungaro, Y. Cohen, J. Org. Chem. 2000, 65, 5026. B. Olenyuk, M. D. Lovin, J. A. Whiteford, P. J. J. Stang, J. Am. Chem. Soc. 1999, 121, 10434. J. K. Young, G. R. Baker, G. R. Newkome, K. F. Morris, C. S. Johnson, Macromolecules 1994, 27, 3464. R. C. van Duijvenbode, I. B. Rietvelt, G. J. M. Koper, Langmuir 2000, 16, 7720. S. G. Yao, G. J. Howlett, R. S. Norton, J. Biomol. NMR 2000, 16, 109. K. R. Deaton, E. A. Feyen, H. J. Nkulabi, K. F. Morris, Magn. Reson. Chem. 2001, 39, 276. Y. Wan, J. K. Angleson, A. G. Kutateladze, J. Am. Chem. Soc. 2002, 124, 5610. C. Frste, W. Heink, J. Krger, H. Pfeifer, N. N. Feokistova, S. P. Zhdanov, Zeolites 1989, 9, 299.

[296] G. H. Sorland, D. Aksnes, L. Gjerda-

ker, J. Magn. Reson. 1999, 137, 397. [297] T. J. Geldbach, F. Breher, V. Gramlich,

[298]

[299] [300]

[301]

[302] [303]

[304]

[305]

[306] [307]

[308]

[309]

[310] [311]

[312]

[313]

P. G. Anil Kumar, P. S. Pregosin, 2004, 43, 1920. E. Martinez-Viviente, H. Ruegger, P. S. Pregosin, J. Lopez-Serrano, Organometallics 2002, 21, 5841. E. Martinez-Viviente, P. Pregosin, Helv. Chim. Acta 2003, 86, 2364. A. Burini, J. P. Fackler, R. Galassi, A. Macchioni, M. A. Omary, M. A. Rawashdeh-Omary, B. R. Pietroni, S. Sabatini, C. Zuccaccia, J. Am. Chem. Soc. 2002, 124, 4570. A. Macchioni, A. Romani, C. Zuccaccia, G. Guglielmetti, C. Querci, Organometallics 2003, 22, 1526. M. Valentini, H. Ruegger, P. S. Pregosin, Helv. Chim. Acta 2001, 84, 2833. P. S. Pregosin, E. Martinez-Viviente, P. G. A. Kumar, Dalton Trans. 2003, 4007. B. Binotti, A. Macchioni, C. Zuccaccia, D. Zuccaccia, Comments Inorg. Chem. 2002, 23, 417. R. E. Hoffman, E. Shabtai, M. Rabinovitz, V. S. Iyer, K. Mllen, A. K. Rai, E. Bayrd, L. T. Scott, J. Chem. Soc., Perkin Trans. 2 1998, 1659. D. E. Babushkin, H. Brintzinger, J. Am. Chem. Soc. 2002, 124, 12869. N. G. Stahl, C. Zuccaccia, T. R. Jensen, T. J. Marks, J. Am. Chem. Soc. 2003, 125, 5256. C. Zuccaccia, N. G. Stahl, A. Macchioni, M.-C. Chen, R. J. A., T. J. Marks, J. Am. Chem. Soc. 2004, 126, 1448. M. Ludwig, R. Kadyrov, H. Fiedler, K. Haage, R. Selke, Chem.-Eur. J. 2001, 7, 3298. K. Fagnou, M. Lautens, Angew. Chem. Int. Ed. Engl. 2002, 41, 27. Y.-X. Chen, M. V. Metz, L. Li, C. L. Stern, T. J. Marks, J. Am. Chem. Soc. 1998, 120, 6287. E. P. Kndig, C. M. Saudan, G. Bernardinelli, Angew. Chem. 1999, 111, 1298. E. P. Kndig, C. M. Saudan, F. Viton, Adv. Synth. Catal. 2001, 343, 51.

79

80

1 NMR Spectroscopy and Homogeneous Catalysis [314] D. A. Evans, J. A. Murray, P. von Matt,

[315]

[316]

[317]

[318]

[319]

[320]

[321] [322] [323]

[324]

[325] [326]

[327]

R. D. Norcross, S. J. Miller, Angew. Chem. 1995, 107, 864. U. Burckhardt, M. Baumann, A. Togni, Tetrahedron Asymmetry 1997, 8, 155. C. Bianchini, C. Mealli, M. Peruzzini, F. Zanobini, J. Am. Chem. Soc. 1992, 114, 5905. K. Gruet, E. Clot, O. Eisenstein, D. H. Lee, B. Patel, A. Macchioni, R. H. Crabtree, New J. Chem. 2003, 27, 80. A. Macchioni, A. Magistrato, I. Orabona, F. Ruffo, U. Rothlisberger, C. Zuccaccia, New J. Chem. 2003, 27, 455. B. Binotti, C. Carfagna, E. Foresti, A. Macchioni, P. Sabatino, C. Zuccaccia, D. Zuccaccia, J. Organomet. Chem., 2004, 689, 647. A. Pfaltz, J. Blankenstein, R. Hilgraf, E. Hormann, S. McIntyre, F. Menges, M. Schonleber, S. P. Smidt, B. Wustenberg, N. Zimmermann, Adv. Syn. Catal. 2003, 345, 33. E. Martinez-Viviente, P. S. Pregosin, Inorg. Chem. 2003, 42, 2209. K. F. Morris, C. S. J. Johnson, J. Am. Chem. Soc. 1992, 114, 3139. M. D. Pelta, H. Barjat, G. A. Morris, A. L. Davis, S. J. Hammond, Magn. Reson. Chem. 1998, 36, 706. K. F. Morris, P. Stilbs, C. S. J. Johnson, Anal. Chem. 1994, 66, 211. A. Jerschow, N. Mller, Macromolecules 1998, 31, 6573. D. A. Jayawickrama, C. K. Larive, E. F. McCord, D. C. Roe, Magn. Reson. Chem. 1998, 36, 755. T. Zhao, H. W. Beckham, Macromolecules 2003, 36, 9859.

[328] T. Zhao, H. W. Beckham, H. W. Gib-

son, Macromolecules 2003, 36, 4833. [329] C. Gambs, T. J. Dickerson, S. Maha-

[330]

[331] [332] [333] [334] [335] [336]

[337] [338] [339] [340] [341] [342] [343]

[344]

[345]

jan, L. B. Pasternack, K. D. Janda, J. Org. Chem. 2003, 68, 3673. S. Viel, D. Capitani, L. Mannina, A. Segre, Biomacromolecules 2003, 4, 1843. M. D. Diaz, S. Berger, Carbohydr. Res. 2000, 329, 1. D. P. Hinton, C. S. Johnson, Jr., J. Phys. Chem. 1993, 97, 9064. E. J. Cabrita, S. Berger, Magn. Reson. Chem. 2001, 39, S142. S. Viel, L. Mannina, A. Segre, Tetrahedron Lett. 2002, 43, 2515. G. S. Kapur, E. J. Cabrita, S. Berger, Tetrahedron Lett. 2000, 41, 7181. A. Hori, K. Kumazawa, T. Kusukawa, D. K. Chand, M. Fujita, S. Sakamoto, K. Yamaguchi, Chem. Eur. J. 2001, 7, 4142. I. Keresztes, P. G. Williard, J. Am. Chem. Soc. 2000, 122, 10228. H. Barjat, G. A. Morris, A. G. Swanson, J. Magn. Reson. 1998, 131, 131. E. Gozansky, D. G. Gorenstein, J. Magn. Reson. Ser. B 1996, 111, 94. G. S. Kapur, M. Findeisen, S. Berger, Fuel 2000, 79, 1347. D. Wu, A. Chen, C. S. Johnson, Jr., J. Magn. Reson. Ser. A 1996, 123, 215. D. Wu, A. Chen, C. S. Johnson, Jr., J. Magn. Reson. Ser. A 1996, 121, 88. R. K. Harris, K. A. Kinnear, G. A. Morris, M. J. Stchedroff, A. Samadi-Maybadi, Chem. Commun. 2001, 2422. N. E. Schlrer, E. J. Cabrita, S. Berger, Angew. Chem. Int. Ed. Engl. 2002, 41, 107. N. E. Schlrer, S. Berger, Organometallics 2001, 20, 1703.

2 High Pressure NMR Cells Gabor Laurenczy and Lothar Helm

2.1

Introduction

With the technical development achieved in the last 30 years, pressure has become a common variable in several chemical and biochemical laboratories. In addition to temperature, concentration, pH, solvent, ionic strength, etc., it helps provide a better understanding of structures and reactions in chemical, biochemical, catalyticmechanistic studies and industrial applications. Two of the first industrial examples of the effect of pressure on reactions are the Haber process for the synthesis of ammonia and the conversion of carbon to diamond. The production of NH3 and synthetic diamonds illustrate completely different fields of use of high pressures: the first application concerns reactions involving pressurized gases and the second deals with the effect of very high hydrostatic pressure on chemical reactions. High pressure analytical techniques have been developed for the majority of the physicochemical methods (spectroscopies: e. g. NMR, IR, UV–visible; and electrochemistry, flow methods, etc.). High pressure has proven to be a useful tool in biochemistry for the study of a number of cell-mediated processes, the most important being the effect on gene expression. The pressure effect on the stability of enzymes and biopolymers is a topic of general interest that may generate a number of possible applications in the area of food science. Biochemistry and biophysics continue to attract many new research groups, especially in the field of protein chemistry. Pressure may be a tool to obtain unique textures and provide biochemical products with new properties. Many interesting chemical reactions and catalytic processes require the use of both high pressures and temperatures, and therefore pressure is an important and increasingly available parameter to study reaction mechanisms and to synthesise new compounds. Under high pressures, the concentration of dissolved gases (H2, CO, CH2¼CH2, CO2, O2, N2, etc) can be increased greatly according to Henry’s law, dramatically modifying equilibria or kinetics involving these gases. Catalytic processes carried out under pressure or in supercritical fluids are becoming Mechanisms in Homogeneous Catalysis. A Spectroscopic Approach. Edited by Brian Heaton Copyright c 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN 3-527-31025-8

82

2 High Pressure NMR Cells

more and more numerous. To study these processes requires in situ and nondestructive analytical methods: multinuclear NMR spectroscopy is one of the most versatile techniques, the number of publications describing different types of high pressure, high temperature NMR cells working under special conditions (pressure, temperature, mixing, solid, liquid, gas samples, etc.) is still increasing. Because these requirements are sometimes conflicting, it is difficult to build a general purpose, variable temperature, high pressure NMR cell. It is quite common that, for a given task, whether it be a high temperature, high pressure industrial synthesis, a special analytical method or mechanistic studies of a given catalytic process, a special high pressure cell must be used. High pressure analytical equipment is not readily available and has in general to be built by the research groups. Due to the inherent risk, using high pressure techniques requires a good understanding of the potential safety issues. Multinuclear NMR spectroscopy is being used increasingly to study catalytic reactions. In order for such studies to be possible, all contact between the sample solutions and the metal part of the cell body has to be avoided because of possible catalytic effects. All materials in contact with the sample solution have to be chemically inert and nonmetallic: quartz, glass, PTFE, Vespel (polyimide polymer, Dupont de Nemours), PCTFE (poly-chlorotrifluoroethylene), Viton seals, etc. The crucial point of the high pressure NMR cell design for catalytic reactions involving gases is the choice of the mixing facilities. To study the reactions of dissolved gases, one has to guarantee saturation or a constant homogenous concentration distribution. The provision of adequate mixing under high gas pressures presents serious technical difficulties. One can drive the mixer from outside the pressurised cell, in which case one needs to avoid any leakage around the moving parts going through the pressure vessel wall. The other solution is to put the whole mixing unit into the pressurised cell, in this case careful design is necessary to avoid increasing the whole gas pressurised volume excessively. Some interesting solutions, e. g. using the high magnetic field to drive the mixer motor, bubbling the gas through the solutions, etc., will be described later. Nuclear magnetic resonance (NMR) spectroscopy is one of the most versatile techniques to study catalytic systems, organic and organometallic compounds, in both nonaqueous and aqueous solutions. To explore all the possibilities of 1H NMR one has either to work in deuterated solvents or use a solvent signal suppression technique. Proton chemical shifts can give information about the structure. Generally, protons bound to carbons coordinated to a metal center show a down field shift, about 1 to 4 ppm, compared to the metal-free environment. Metal hydrides usually have upfield chemical shifts, sometimes up to –40 ppm. In a paramagnetic environment chemical shifts can be very large, the resonances are broadened and sometimes are even not detectable. Substituting some of the protons by deuterium can have a drastic effect on the rest of the 1H NMR spectrum: since 2H nuclei have a spin I¼1, the multiplicity of the resonances increases, although the J(2H) coupling constants are significantly smaller. 13C is naturally one of the most widely used nuclei to elucidate structure and dynamic behavior of organic, organometallic compounds and catalytic systems. 13C NMR is applied both when

2.2 High Pressure NMR of Liquids

in natural abundance and for compounds enriched with 13C during the synthesis. 13 C nuclei have a spin I¼1/2 and 1J(C,H) couplings give direct information on the number of H atoms linked to the carbon: the CH signal is a doublet, the CH2 is a triplet and the CH3 gives a quartet in a 13C NMR spectrum. The 1J(C,H) coupling constants are about 250, 160 and 125 Hz in alkynes, alkenes and alkanes respectively. Protons can be decoupled to simplify 13C spectra at the expense of loss of information. In aqueous solution, 17O NMR gives the possibility to study solvent exchange and dynamic behavior of the aqua-complexes. Having a spin I¼5/2, the 17O signals are broad and the observed chemical shifts are between 1500 and –250 ppm referenced to bulk water. The presence of other NMR active nuclei in the compounds (31P and 19F having a spin 1/2, 14N and 2H having a spin 1) and sometimes the metal atoms themselves (103Rh, 59Co, 195Pt, etc.) extend the information available through NMR spectroscopy. To move to a higher magnetic field can simplify spectra with strong scalar coupling, because chemical shifts in Hz increase with the magnetic field but the coupling constants stay the same (in Hz). Measuring spin–lattice or longitudinal relaxation times, T1, can help to distinguish between classical dihydrides and molecular hydrogen complexes, the latter having much shorter T1 values. To resolve difficult molecular structures two-dimensional NMR or special pulse sequences can sometimes be useful. Intra- and intermolecular exchange processes can broaden NMR signals or, in the case of rapid exchange, lead to spectra where only coalesced signals of exchanging sites can be observed. Variable temperature NMR is commonly used to measure activation enthalpies, DH ‡ , and entropies, DS‡ , of these dynamic processes. Variable pressure NMR gives access to the activation volume, DV ‡, a very useful quantity in the determination of reaction mechanisms [1, 2]. In the following, we will present a selected choice of high-pressure cells developed for high-resolution NMR. We have divided the applications of high-pressure NMR into three sections. First, we will describe high pressure equipment used to study liquids under hydrostatic pressure up to 1000 MPa; then, we present some special approaches used to study supercritical fluids which, besides moderate pressure, often need high temperatures. Finally, we discuss NMR cells to study solutions under gas pressure, a situation which is quite common in catalysis.

2.2

High Pressure NMR of Liquids 2.2.1

High Pressure, High Resolution Probes

In chemistry and biochemistry NMR is, in general, applied to liquid samples. The quality of the experimental data is mainly influenced by the spectral resolution and by the signal-to-noise ratio obtained. Building high pressure, high resolution NMR probes for liquid samples has to take into account these constraints. Special probes for variable pressure experiments differ from commercial probes by the presence

83

84


of a pressure vessel, which is usually constructed from a cylinder closed by plugs on the upper and lower ends. Connections for the pressure-transmitting fluid and the electrical feedthroughs for radiofrequency (RF) are built into these plugs. All parts of the pressure vessel, which will be placed into superconducting NMR magnets, have to be machined from highly nonmagnetic materials with high tensile strength, such as titanium alloys or beryllium–copper alloys (see Table 2.1). The NMR coil sitting inside the pressure vessel is normally a saddle-shaped coil and is connected to the tuning and matching electronic circuit, which is located at normal pressure outside the pressure vessel. To allow locking of the magnetic field, this circuit has to be double tuned to the NMR observation frequency (1H, 13C or others) and to the lock frequency, which normally uses the resonance of 2H. To obtain the best signal-to-noise ratio from a given coil volume, the RF tuning and matching electronics should be situated as close as possible to the coil, putting strong limitations on the design of the closing plug and the RF-feedthrough. The liquid sample is confined in a sample container, separating it from the pressure transmitting fluid. Different techniques like bellows or moving pistons have been proposed to transmit the pressure to the sample. For optimum sensitivity a thin-walled container such as a standard NMR tube should be used to obtain a good filling factor of the NMR coil. The NMR coil is surrounded by the pressure transmitting fluid, which should therefore not contain any of the nuclei that are to be observed in the NMR experiment. For 1H NMR, fully halogenated hydrocarbons like C2Cl4 or CS2 are commonly used. A liquid, rather than a gas like argon is preferred for safety reasons. At very high temperatures, however, corrosion may be a problem with halogen-containing molecules so that inert gases may have to be used. At low temperatures, attention has to be paid to the strong increase in the freezing point with pressure. To measure the effect of pressure on structures and dynamics, a good stability and homogeneity of the sample temperature has to be guaranteed. Therefore, variable pressure probes for high resolution NMR are in general thermostatted by

Table 2.1 Nonmagnetic materials for high pressure NMR (from Ref. [3]).

Material

Tensile strength, MPa

Beryllium–copper alloy Berylco-25

1200–1400

Phosphorus bronze

690

Titanium alloys IMIa 680 RMIb 6Al-4V RMIb 6Al-2Sn-4Zr-6Mo RMIb 3Al-8V-6Cr-4Zr-4Mo

1250 970 1170 1380

Stainless steel Type 316 a

1030 b

IMI: Imperial Metal Industries, UK. RMI: RMI Titanium Co, Niles, Ohio, USA.


pumping liquid from a temperature regulated bath through the pressure vessel. The temperature is measured by a thermocouple or a Pt-100 resistor placed either inside the pressure vessel, where it is exposed to varying pressure, or inside the wall of the vessel where care has to be taken not to weaken the container. In the following, we will present two typical NMR probes for high resolution high pressure NMR of liquid samples working in a temperature range from –50 to þ120 hC. Structure and dynamic studies in physical chemistry, catalysis, and pressure denaturation of proteins in biochemistry, generally need relatively high pressures, between 500 and 1000 MPa. NMR probes for such relatively high hydrostatic pressures are therefore designed to fit into so-called wide-bore cryo-magnets with a usable bore of H ¼ 69 mm (inner diameter of room temperature shim coils). The first high pressure probes for NMR cryo-magnets were built as soon as wide-bore magnets became available [4–6]. An example of such a high resolution NMR probe for pressures up to 1000 MPa is shown in Figure 2.1. The vessel body is machined from the high-strength beta alloy of titanium, 3AL-8V-6Cr-4Zr4Mo (see Table 2.1). The top plug driver and the bottom plug are built from Berylco-25 which was heat-treated for 3 h at 315 hC before final machining [7]. The probe was tested up to a pressure of 925 MPa. Temperature can be adjusted between –30 and þ80 hC [8] by pumping a thermostatted liquid through etched grooves in the exterior of the vessel. A first RF-circuit uses a single double-tuned saddle-coil and is double tuned to 300 and 46 MHz, corresponding to the reso-

Figure 2.1 Schematic drawing of a high pressure NMR probe designed for wide-bore cryomagnets and a sample cell [10].

85

86


Figure 2.2 High frequency RF-coil for high pressure NMR probe [8].

nance frequencies of 1H and 2H in a 7.04 T magnet [7]. Spectral resolution achieved is about 1 Hz, independent of pressure up to 800 MPa. Different types of sample cells have been used with this probe [3]: a piston sample cell consisting of a thinwalled 8 mm NMR tube connected to a precision-bore Pyrex tube through a glass capillary [9] or better, but more expensive, a bellows design, consisting of a glass (Pyrex 7740) or quartz sample chamber connected to stainless steel (SS 316) tubing and bellows to accommodate the volume change of the liquid due to compression (see Figure 2.1). A novel RF-coil design (Figure 2.2) placed into the same pressure vessel allows the higher frequency of the double-tuned RF-circuit to be increased to 500 MHz [8]. The spectral resolution obtained with this coil is 0.7 Hz and 1.5 Hz at 300 MHz and 500 MHz, respectively (Figure 2.3). The signal-to-noise ratio measured at 500 MHz using the 8 mm piston-closed sample cell is about five times that observed at 300 MHz. The high resonance frequency and the high maximum pressure makes this probe ideal to study, for example, pressure induced cold denaturation of proteins. Examples of applications of this type of probe to high pressure NMR studies of protein can be found in the reviews of Jonas [10, 11].

Figure 2.3 CHCl3 peak of the line-shape sample (1 % v/v CHCl3 in acetone-d6, 300 MHz; 1 % v/v CHCl3 in chloroform-d, 500 MHz) measured at 1 bar in an 8 mm tube using the new RF coil probe at 300 MHz (a) and 500 MHz (b) [8].


Variable pressure studies of chemical equilibria and rate constants of chemical reactions generally require lower maximum pressures up to “only” 200 MPa. As a consequence, the walls of the pressure vessel can be less thick. High pressure, high resolution NMR probes with an external diameter of Hexternal ¼ 40 mm can therefore be built for normal-bore cryo-magnets. An advantage of these smaller probes, besides the fact that normal-bore magnets are much more commonly available than wide-bore magnets, is the reduced weight, making them much easier to manipulate. The first high resolution, high pressure NMR probe for normal-bore cryo-magnets was built in 1994, working at a 1H NMR resonance frequency of 200 MHz [12]. A new design of high pressure probes for narrow-bore magnets was published in 1997 (Figure 2.4) [13]: the RF tuning and matching circuit was closer to the NMR coil, which allowed an increase in the tuning frequency to 400 MHz. The high pressure vessel of this probe is machined from Berylco-25 and has an outer diameter of 27 mm, an inner diameter of 17 mm, and an inner length of 170 mm. A double helix (Figure 2.4F) is cut into the outside of

Figure 2.4 High pressure high resolution NMR probe for a 9.4 T narrow-bore cryo-magnet: (A) pressurizing fluid inlet; (B) thermostatting inlet and outlet; (C) thermal insulation by water circulation; (D) aluminum support; (E) Berylco plug; (F) double helix for themostatting; (G) sample holder; (H) 5 mm o. d. NMR tube with Macor seal; (I) NMR coil; (J) Pt 100 V resistor; (K) RF leadthrough; (L) matching/tuning capacitors; and (M) screwdrivers for matching/tuning; maximum working pressure 200 MPa; temperature range –50 to þ120 hC.

87

88


Figure 2.5 Sample container for normal bore high pressure NMR probe.

the vessel for circulation of the thermostatting liquid which can thus accommodate a temperature range between –40 and þ150 hC. The temperature gradient over the sample region was found to be less than 0.3 hC and the temperature stability to be better than 0.2 hC over the whole pressure range. An electronic tuning and matching circuit was built for 1H (400 MHz) observation and 2H (61.4 MHz) field lock. A spectral resolution of I0.5 Hz was obtained at 400 MHz. A second probe with an identical pressure vessel was built for multinuclear NMR. It is equipped with two NMR coils: an observation coil which is broad-band tunable from 120 MHz to 20 MHz for 13C, 17O and 14N NMR, for example, and a second coil, tuned to 1H (400 MHz) for spin decoupling and 2H (61 MHz) for field locking. The container for liquid samples is shown in Figure 2.5. It consists of a normal high precision 5 mm NMR tube cut to a length of 60 mm and closed with a piston and a cap made from the machinable ceramic Macor. These probes were successfully used to study solvent exchange on solvated metal ions and metal ion complexes [14, 15]. 2.2.2

Glass and Quartz Capillaries

An alternative technique to perform high resolution NMR at high pressure uses glass or quartz capillaries. In this method, which was pioneered by Yamada [16] a pressure resisting cell is used as a sample container and introduced into a commercial 5 or 10 mm NMR probe. To withstand the force exerted by the compressed liquid, the cell has to be made of nonmagnetic material of high tensile strength like reinforced glass, quartz, sapphire or ceramics. A high pressure NMR cell made from a glass capillary is shown in Figure 2.6 [17–19]. The high pressure ca-


Figure 2.6 High pressure glass capillary and autoclave system as developed by Ldemann [19] on the basic design of Yamada. The dimensions of the high pressure capillary are: inner diameter 1 mm, outer diameter 5 mm, length 35 to 40 mm.

89

90


Figure 2.7 High frequency part of 600 MHz 1

H NMR spectra of 5 mM Gly-Gly-Ar-Ala in 50 mM phosphate buffer at various pressures at 305 K. The total recording times was 19 min [19].

pillary has an outer diameter of 5 mm and an inner diameter of only 1 mm. Some capillaries withstand pressures up to 400 MPa, but normal working pressure is up to 200 MPa. To protect the commercial 10 mm NMR probe from damage in the case of an exploding capillary, the capillary is placed into a short polished Teflon tube (5 mm inner diameter) and then into a thick-walled 10 mm glass tube [19]. The field homogeneity obtained with such an assembly is better than 2 Hz at 500 or 600 MHz (Figure 2.7). A high pressure NMR cell made from a quartz capillary is shown in Figure 2.8 [20]. This cell has been tested for pressures up to 300 MPa in a commercial 5 mm NMR probe inside a 17.6 T (750 MHz 1H NMR) cryo-magnet. The spectral resolution in the 750 MHz 1H NMR probe is better than 3q10 –9, corresponding to about 2 Hz. The main disadvantage of the glass and quartz capillary design is the low effective sample volume and the bad filling factor of the NMR coil [21]. This results mainly in a low sensitivity. An advantage is, however, that the glass or quartz container is nonmagnetic and transparent to RF irradiation. Furthermore, commercial NMR probes can be used without modification up to the highest magnetic fields available today. This makes the high pressure cells an interesting tool for studies of biological samples such as proteins [19, 22, 23].

2.3

High Pressure NMR of Supercritical Fluids 2.3.1

High Pressure, High Temperature NMR Probes

NMR investigations of supercritical fluids or of molecules dissolved in supercritical solvents have become more and more common. These studies normally need pressures of several MPa and often temperatures well above 100 hC (Table 2.2).

2.3 High Pressure NMR of Supercritical Fluids

Figure 2.8 (A) Schematic illustration of the entire set-up of the high pressure high resolution sample cell for a 750 MHz NMR system [20]. (B) Detailed view of the high pressure cell assembly.

Table 2.2

Critical values of some solvents [24].

Solvent

Tc / K

Pc / MPa

rc g cm–3

H2O CO2 NH3 CHCl3 CH4 CH3OH CH3CH2OH CH3COCH3 C6H5CH3 Xe

647.1 304.1 405.5 536.4 190.5 512.6 513.9 508.1 591.8 289.7

22.06 7.375 11.35 5.47 4.604 8.092 6.137 4.700 4.104 5.84

0.322 0.468 0.235 0.500 0.162 0.272 0.276 0.278 0.292 1.11

91

92


A NMR probe for very high temperature and high pressure was published by Jonas in 1979 [5]. The probe had a relatively low spectral resolution but allowed NMR relaxation measurements at temperatures up to 700 hC. The pressure vessel was made from the IMI-680 titanium alloy and could withstand pressures up to 200 MPa. Because of the high temperatures attained with the probe, gaseous argon was used as the pressure transmitting medium. A first 1H NMR relaxation study of supercritical water performed with this probe was published in 1981 [25]. More recently the group of Conradi built a probe to study supercritical water with a working pressure up to 40 MPa and a maximum temperature of 600 hC (Figure 2.9) [26]. The pressure vessel, made of a Ti-Al6-V4 alloy, has an outer diameter of 38 mm and an inner diameter of 22 mm. The probe has a weight of about 6 kg and fits into a 4.5 T (186.6 MHz 1H) superconducting magnet of 98 mm bore within room temperature shims. The sample container (Figure 2.9, right), which has to be impervious to supercritical water, is made from a ceramic tube. The pres-

Figure 2.9 Left: Overall view of a high pressure, high temperature probe for NMR studies of supercritical water. Right: detail of the ceramic sample tube and free piston pressure balancing arrangement. The stainless-steel cylinder and piston are far from the heated region and remain cool.


sure transmission medium is compressed argon. Full width at half-height, independent of applied pressure and temperature, is about 20 Hz. The group of Conradi used this probe to study, for example, proton exchange in supercritical media [27]. 2.3.2

Toroid Probes for High Pressure NMR

A different probe design for pressure NMR studies of fluids like compressed gases or supercritical fluids involves the use of toroids [28, 29]. Due to the internal confinement of the magnetic flux within a torus (Figure 2.10), toroids are intrinsically more sensitive than other coil types [30] and can be operated close to the internal wall of a metallic pressure vessel. A typical probe design using a four-turn torus is shown in Figure 2.11. The pressure vessel has an internal volume of 8 cm3 and can be used at pressures up to 170 MPa [29]. The four-turn torus is double tuned to 300 MHz (1H) and 71 MHz (13C)–75 (59Co). Probe heating is accomplished by means of a chilled-water-jacketed electrical furnace for temperatures up to 250 hC. Spectral resolution, as determined from the 2H signal of C6D6, is Z2q10 –8 [28]. Another class of toroidal NMR probes is formed by toroid cavity detectors (TCD) [31]. These TCD have been mounted into cylindrical metallic autoclaves to study spin relaxation effects of gases under pressure [31]. If the toroid cavity detector is the metallic pressure vessel itself, it is called a toroid cavity autoclave (TCA) probe [32, 33]. These probes can be tuned to higher resonance frequencies than TCDs and also show better spectral resolution [29]. Figure 2.12 shows the design of a TCA where the autoclave body is built from phosphorus bronze [33] keeping the field distortions of the magnetic field B0, induced by susceptibility mismatches

Figure 2.10

Magnetic flux in different coil types used in NMR probes (from Ref. [29]).

93

94


Figure 2.11 High pressure NMR probe using a toroid detector (from Ref. [29]). SP-1: Vespel

(DuPont).

between sample and autoclave, to a reasonable minimum. Current pressure and temperature capabilities are 0–30 MPa and 20–150 hC, respectively [32]. Signalto-noise ratios comparable to standard NMR probes are achieved at 200 MHz 1H with a spectral resolution of 0.55 Hz. Gas–liquid reactions can be studied in modified TCA having two compartments separated by a contact disk placed inside the autoclave body [32, 33]. Woelk and Bargon have studied the catalytic hydrogenation of alkynes in scCO2 [34], using catalysts based on transition metal colloids, carried out in the TCA setup. Under mild experimental conditions (15 bar hydrogen partial pressure, at 323 K) they have found extremely high TOFs (as high as 4 q 106 h–1). They have synthesized and applied a new “CO2 -philic” chiral rhodium diphosphinite complex as catalyst precursor in the asymmetric hydrogenation of di-Me itaconate in scCO2, scC2H6 and various organic solvents [35]. Deuterium labelling


Figure 2.12 Exploded (left) and assembled (right) view of a toroid cavity autoclave probe for in situ investigations under high gas pressures or in supercritical fluids. Autoclave base (A) and autoclave body (P-bronze, B); thermocouple (C); coaxial heater (D); PTFE ring (E); central conductor (Cu/Be ring, F); nonmagnetic pin from male coaxial connection (G); RF feedthrough (from Rathke [28], H); base-plate (MACOR, I); fixing screws (P-bronze, J); PEEK capillary (K); ceramic ball (Si3N4, L); PTFE seal (M).

studies and parahydrogen-induced polarization (PHIP) NMR experiments were used to provide detailed mechanistic insight into the activation and transfer of the dihydrogen molecule during hydrogenation in scCO2. They have excluded as a possible explanation, the chemical interactions between CO2 and reactive intermediates of the catalytic pathway, according to the experimental difference in the catalytic behavior in scCO2 and hexane.

95

96


2.4

High Pressure NMR of Gases Dissolved in Liquids 2.4.1

Sapphire Tubes

To study catalytic reactions very high pressures or temperatures are often not needed. For pressures below 100 MPa and temperatures within the rage of commercial NMR probes (below 150 hC) no special probes built with metallic pressure vessels are needed. In 1985 Roe [36] introduced NMR tubes made of a sapphire crystal to perform pressure studies in homogeneous catalysis. The original design used a 5 mm outer diameter sapphire tube with a 0.8 mm thick wall to which a Ti-Al6-V4 alloy valve was glued (Figure 2.13). This assembly allowed measurements up to about 14 MPa. Sapphire NMR tubes are very convenient to work with: they can be filled in a glove box, they can be shaken for faster equilibrium in the case of liquid samples pressurized with gases (see below) and they are transparent and therefore allow color changes or phase separations to be monitored. The original design was adapted by different groups to work for example with 10 mm sapphire tubes [37, 38] or to record pressure continuously [39]. To give special attention to safety, a protection device should always be used with sapphire tubes. Figure 2.14 shows a 10 mm outer diameter sapphire tube inside a protection device, made of acrylic glass [40, 41]. This device can be put directly on the top of the cryo-magnet and the sample goes down into the NMR probe using the standard sample lift of the spectrometer. Special effort was given to make the closing valve centrosymmetric and light to minimize spinning sidebands. Several olefin complexes of the Ru(II) aqua-ion [42–44] and of the other Ru complexes [45, 46] have been synthesized and characterized in the ring opening metathesis polymerization (ROMP) or olefin isomerization reactions. The simplest olefin complexes of Ru were also observed, isolated and characterized under ethy-

Figure 2.13 Schematic drawing of Roe’s titanium-alloy valve glued to a 5 mm outer diameter sapphire tube.

2.4 High Pressure NMR of Gases Dissolved in Liquids

Figure 2.14

Schematic drawing of the sapphire tube/Ti-valve assembly (expanded) and the safety

device.

lene pressure [47]. An aqueous solution of [Ru(H2O)6]2þ in a 10 mm sapphire tube was pressurized with 6.0 MPa of ethylene and mixed. The reactions: [Ru(H2O)6]2þ þ CH2CH2 p [Ru(CH2CH2)(H2O)5]2þ þ H2O and [Ru(CH2CH2)(H2O)5]2þ þ CH2CH2 p Ru(CH2CH2)2(H2O)4]2þ þ H2O were followed by 1H, 13C and 17O NMR spectroscopy. The synthesis [48] of one of the simplest organometallic aquaions, [Ru(CO)(H2O)5]2þ, was carried out under 5.0 MPa CO pressure:

97

98


[Ru(H2O)6]2þ þ CO p [Ru(CO)(H2O)5]2þ þ H2O in a 10 mm sapphire NMR tube [36, 40]. (The FT-IR spectra of [Ru(CO)(H2O)5]2þ, measured in a high pressure IR cell [49], shows the characteristic stretching frequency at 1971 cm–1 due to the coordinated CO.) Further substitution of H2O by CO in [Ru(CO)(H2O)5]2þ is very slow [50]. The water exchange on the resulting complexes, cis-[Ru(CO)2(H2Oeq)2(H2Oax)2](tos)2, [Ru(CO)3(H2O)3](ClO4)2 and on [Ru(CO)(H2Oeq)4(H2Oax)](tos)2, as well as the 17O exchange between the bulk water and the carbonyl oxygens have been studied by 17O NMR spectroscopy. The water soluble tertiary phosphine complex of ruthenium(II), [RuCl2(PTA)4], (PTA¼1,3,5-triaza-7-phosphaadamantane) was used as catalyst precursor for the hydrogenation of CO2 and bicarbonate in aqueous solution [51]. For this relatively

Figure 2.15 Heating device for medium pressure sapphire NMR tubes.


slow reaction, followed by 1H, 2H, 13C and 31P NMR spectroscopy, the sample was thermostatted by the NMR heating facilities (when spectra were taken) or by a heating device for medium pressure sapphire NMR tubes (Figure 2.15). The CO2 and H2 pressure were controlled and monitored during the reduction by an electronic pressure gauge (Figure 2.16).

Figure 2.16 Electronic pressure gauge to follow/control the pressure in sapphire NMR tubes. The small dead volume allows the pressure in the tube to be measured precisely and also allows the pressure variation to be monitored with time.

99

100


2.4.2

High Pressure Probes for Pressurized Gases

To study liquid samples pressurized by gases to higher pressures NMR probes using high pressure vessels have to be used. The first experiments performed by Heaton and Jonas [52, 53] showed the need for special high pressure NMR probes. After pressurizing the sample with a gas like H2, N2 or CO, equilibrium is established by diffusion of the gas molecules across the gas–liquid interface into the liquid. This process depends on the area of the gas–liquid interface and is normally very slow. To accelerate the establishment of the equilibrium the gas/liquid sample has to be shaken, as normally done outside the NMR magnet with sapphire tubes described above, or a stirrer has to be used inside the pressure vessel, which is located in the NMR magnet. The first high pressure probe equipped with a stirrer was built by Jonas in 1987 [54] (Figure 2.17). Temperature control (–50 to 150 hC) is provided by circulating liquid through copper tubing coiled on a metal cylinder tightly fitted to the pressure vessel. The pressure vessel is made of Berylco-25, chosen for its resistance to hydrogen embrittlement. The maximum working pressure of the probe is 200 MPa. The stirring assembly (Figure 2.17, left) is enclosed in a second pressure vessel located outside the NMR magnet and connected to the probe by 90 cm long stainless steel tubing. The probe fits into a 4.2 T cryo-magnet with a bore of 130 mm and an optimal resolution of 2 Hz at 180 MHz 1H frequency [9]. To avoid the long tubing connecting the stirring assembly to the pressure vessel a special gas–liquid mixing unit working inside the cryo-magnet was built for the normal bore high pressure probe presented in Figure 2.4. The mixing motor (Figure 2.18, right) which is directly connected to the pressure vessel (Figure 2.18, left) is located at the upper end of the superconducting coil inside the NMR magnet. The strong magnetic field is very inhomogeneous at this point and serves as the stator of the mixing motor [55]. A sinusoidal current passing through the driving coil leads to a see-saw movement of a perforated mixing disk. The frequency and amplitude of the movement can be controlled easily by changing the amplitude and frequency of the applied current. Good mixing is achieved in less than 1 min. The high pressure probe is impervious up to pressures of 100 MPa, as tested with H2 [40]. Stirring facilities have also been built for high pressure toroid cavity detectors [57, 58]. Figure 2.19 shows such a probe built for a 9.4 T wide-bore magnet [57]. Stirring is achieved by passing a small AC current, with a frequency of about 80 Hz, through the stir coil which leads to a rocking motion of the coil. The gas–liquid equilibration half life for CH4 dissolving into D2O has been lowered from 15 h to 10 min. A different, convenient high pressure NMR flow cell for in situ studies of homogeneous catalysis has been developed by Iggo and coworkers [59]. To overcome the slow gas–liquid mixing, the inlet gas passes from the bottom through the reaction solution providing a constant stream of gas bubbles through the solution (Figure 2.20). Provided that a positive pressure with respect to the probe is maintained at the base of the probe, leakage of solvent down the gas inlet is not problematic.


Figure 2.17

Schematic of a high pressure NMR probe equipped with stirring [41, 54].

The gas flow rates are adjusted to ensure a constant pressure inside the cell and to minimize degradation of the spectral resolution by the gas bubbles. Typical linewidths at half height are I4 Hz on 1H NMR (200 MHz) and I2 Hz on 31P (80.3 MHz). The probe fits into the room temperature shims of a wide-bore NMR magnet and the pressure and temperature range of the high pressure NMR flow are 0.1 to 20 MPa and –40 to 175 hC, respectively. Iggo and coworkers have studied the asymmetric copolymerisation of styrene with CO catalyzed by Pd-(R,S)-BINAPHOS complexes under both diffusion- and

101

102


Figure 2.18 High pressure, high resolution normal bore NMR probe (left) equipped with a mixing facility (right) capable of use for studies under gas pressures up to 100 MPa [56].

reaction-controlled conditions using in situ NMR spectroscopy [60]. The use of a high-pressure NMR flow cell allowed them to observe potential catalytic intermediates [58]. The formation of Pd-alkyl complexes, via 1,2-insertion of styrene into Pd-acyl complexes, has been confirmed to be the most active catalytic pathway. The 2,1-insertion complexes were found to be quite inactive to further insertion and remarkably stable toward b-hydride elimination, in contrast to our previous expectations. This type of high pressure NMR flow cell is specially convenient for mechanistic studies in gas–liquid reactions [61–69].

2.5

Conclusions, Perspectives

High pressure, high resolution NMR has been very successfully applied over the last 30 years to solve problems in the physical chemistry of liquids, chemical kinetics, catalysis and more recently in biochemistry. For these purposes, research

2.5 Conclusions, Perspectives

Figure 2.19

High pressure NMR probe and toroid detector equipped with stir coil [57].

groups have built a variety of high pressure probes and cells, adapted to their special needs. Probably, the variety of probe designs prevents commercial construction of high pressure NMR equipment. An important step in the development of probes with metallic pressure vessels was the design of normal bore probes, making the probe handling much more convenient and allowing, in principle, measurements at higher magnetic fields. The maximum frequency for this kind of probe is probably 500–600 MHz, limited by the RF feedthrough and the small distance between the RF coil and the metallic walls of the vessel. Such probes could however be used at very high magnetic field (i14 T) for multinuclear NMR like 13C, 31P or 17O. The use of glass and quartz capillaries is, in principle, not restricted to what nowadays are medium frequencies (400–600 MHz). This type of high pressure NMR equipment will certainly be further improved, mainly to study biological samples by high-field 1H NMR. To overcome the serious sensitivity problem inher-

103

104


Figure 2.20

Exploded (left) and assembled (right) view of the high pressure NMR flow cell [59].

ent with this probe design the use of high-pressure capillaries in commercial cryoprobes will open new possibilities in the future. High-pressure NMR studies for catalysis and with supercritical fluids will lead to a much broader application of sapphire NMR cells and to special applications of toroidal probes. The sapphire tube technique can today be considered as a standard, cheap and easily applicable technique to study samples under medium gas pressures, up to 100 MPa.

Acknowledgments

The authors gratefully acknowledge financial support from the Swiss National Science Foundation and the Swiss Federal Office for Education and Science (COST Program). Furthermore, we wish to thank the large number of people who have contributed to the work performed in Lausanne.

References

References [1] R. van Eldik, F.-G. Klrner, High

[2]

[3] [4] [5] [6]

[7] [8] [9]

[10] [11] [12]

[13]

[14]

[15]

[16]

Pressure Chemistry, Wiley-VCH, Weinheim, 2002. Inorganic Reaction Mechanisms, Eds. R. van Eldik, C. D. Hubbard, Academic Press, Amsterdam, 2003, Vol. 54. P. Koziol, C. Reiner, J. Jonas, Appl. Magn. Reson. 1996, 11, 19–29. J. Jonas, Rev. Sci. Instrum. 1972, 43, 643–649. T. DeFries, J. Jonas, J. Magn. Reson. 1979, 35, 111–119. D. L. Pisaniello, L. Helm, P. F. Meier, A. E. Merbach, J. Am. Chem. Soc. 1983, 105, 4528–4536. L. Ballard, C. Reiner, J. Jonas, J. Magn. Reson. A 1996, 123, 86. L. Ballard, C. Reiner, J. Jonas, J. Magn. Reson. 1998, 133, 190–193. J. Jonas, High Pressure NMR Studies of the Dynamics in Liquids and Complex Systems, Ed. J. Jonas, Springer Verlag, Berlin, 1991, Vol. 24, pp. 85–128. J. Jonas, Biochim. Biophys. Acta 2002, 1595, 145–159. J. Jonas, L. Ballard, D. Nash, Biophys. J. 1998, 75, 445–452. B. Moullet, U. Frey, L. Helm, R. Tschanz, R. Ith, A. E. Merbach, High Pressure Res. 1994, 12, 285–289. A. Cusanelli, L. Nicula-Dadci, U. Frey, A. E. Merbach, Inorg. Chem. 1997, 36, 2211–2217. F. A. Dunand, L. Helm, A. E. Merbach, Adv. Inorg. Chem. 2003, 54, 1–69. L. Helm, A. E. Merbach, in Water Exchange on Metal Ions: The Effect of Pressure, Eds. R. v. Eldik, F.-G. Klrner, Wiley-VCH, Weinheim, 2002, pp. 131–160. H. Yamada, in Glass Cell Method for High-Pressure, High-Resolution NMR Measurements. Applications to the Studies of Pressure Effects on Molecular Conformation and Structure, Ed. J. Jonas, Springer Verlag, Berlin, 1991, Vol. 24, pp. 235–263.

[17] E. W. Lang, H.-D. Ldemann, Ber.

[18] [19]

[20]

[21]

[22] [23]

[24] [25] [26] [27] [28] [29]

[30] [31] [32]

[33]

[34]

Bunsenges. Phys. Chem. 1980, 84, 462–470. E. W. Lang, H.-D. Ldemann, Progr. NMR Spectrosc. 1993, 25, 507–633. M. R. Arnold, W. Kremer, H.-D. Ldemann, H. R. Kalbitzer, Biophys. Chem. 2002, 96, 129–140. H. Yamada, K. Nishikawa, M. Honda, T. Shimura, K. Akasaka, K. Tabayashi, Rev. Sci.Instrum. 2001, 72, 1463– 1471. W. E. Price, H.-D. Ldemann, NMR and Diaphragm Cell Techniques for the Study of Molecular Dynamics in Fluids, Eds. W. B. Holzapfel, N. S. Isaacs, Oxford University Press, Oxford, 1997, Vol. 2, p. 225. K. Akasaka, Pure Appl. Chem. 2003, 75, 927–936. M. W. Lassalle, H. Li, H. Yamada, K. Akasaka, C. Redfield, Protein Science 2003, 12, 66–72. Handbook of Chemistry and Physics, CRC Press, Boca Raton, 1993, Vol. 73. W. J. Lamb, J. Jonas, J. Chem. Phys. 1981, 74, 913–921. M. M. Hoffmann, M. S. Conradi, Rev. Sci. Instrum. 1997, 68, 159–164. M. M. Hoffmann, M. S. Conradi, J. Supercrit. Fluids 1998, 14, 31–40. J. W. Rathke, J. Magn. Reson. 1989, 85, 150–155. J. W. Rathke, R. J. Klingler, R. E. Gerald, K. W. Kramarz, K. Woelk, Prog. NMR Spectrosc. 1997, 30, 209– 253. T. E. Glass, H. C. Dorn, J. Magn. Reson. 1983, 51, 527–530. K. Woelk, J. W. Rathke, R. J. Klingler, J. Magn. Reson. A 1994, 109, 137–146. H. G. Niessen, P. Trautner, S. Wiemann, J. Bargon, K. Woelk, Rev. Sci.Instrum. 2002, 73, 1259–1266. H. G. Niessen, P. Trautner, R. Backhausen, K. Woelk, Concepts Magn. Reson. 2003, 16B, 15–21. H. G. Niessen, A. Eichhorn, K. Woelk, J. Bargon, J. Mol. Catal. A: Chemical 2002, 182–183, 463–470.

105

106

2 High Pressure NMR Cells [35] S. Lange, A. Brinkmann, P. Trautner,

[36] [37] [38] [39]

[40]

[41]

[42] [43] [44] [45]

[46]

[47]

[48]

[49] [50]

[51] [52]

[53]

K. Woelk, J. Bargon, W. Leitner, Chirality 2000, 12, 450–457. D. C. Roe, J. Magn. Reson. 1985, 63, 388–391. I. T. Horvth, E. C. Ponce, Rev. Sci. Instrum. 1991, 62, 1104–1105. I. T. Horvth, J. M. Millar, Chem. Rev. 1991, 91, 1339–1351. S. Gaemers, H. Luyten; J. M. Ernsting, C. J. Elsevier, Magn. Reson. Chem. 1999, 37, 25–31. A. Cusanelli, U. Frey, D. T. Richens, A. E. Merbach, J. Am. Chem. Soc. 1996, 118, 5265–5271. A. Cusanelli, U. Frey, D. Marek, A. E. Merbach, Spectrosc. Eur. 1997, 9, 22–27. B. M. Novak, R. H. Grubbs, J. Am. Chem. Soc. 1988, 110, 7542–7543. D. V. McGrath, R. H. Grubbs, J. Am. Chem. Soc. 1991, 113, 3611–3613. T. Karlen, A. Ludi, Helv. Chim. Acta 1992, 75, 1604–1606. S. T. Nguyen, L. K. Johnson, R. H. Grubbs, J. W. Ziller, J. Am. Chem. Soc. 1992, 114, 3974–3975. S. T. Nguyen, R. H. Grubbs, J. W. Ziller, J. Am. Chem. Soc. 1993, 115, 9858–9859. G. Laurenczy, A. E. Merbach, J. Chem. Soc., Chem. Commun. 1993, 1993, 187–189. G. Laurenczy, L. Helm, A. Ludi, A. E. Merbach, Helv. Chim. Acta 1991, 74, 1236–1238. G. Laurenczy, F. Lukcz, R. Roulet, Anal. Chim. Acta 1998, 359, 275–281. U. C. Meier, R. Scopelliti, E. Solari, A. E. Merbach, Inorg. Chem. 2000, 39, 3816–3822. G. Laurenczy, F. Jo, L. Ndasdi, Inorg. Chem. 2000, 39, 5083–5088. B. T. Heaton, J. Jonas, T. Eguchi, G. A. Hoffman, J. Chem. Soc., Chem. Commun. 1981, 331–332. B. T. Heaton, L. Strona, J. Jonas, T. Eguchi, G. A. Hoffmann, J. Chem. Soc., Dalton Trans. 1982, 1159–1164.

[54] D. Vander Velde, J. Jonas, J. Magn.

Reson. 1987, 71, 480–487. [55] P. Favre, U. Frey, D. Marek, F. Metz,

US Pat. 5545998, 1996. [56] L. Helm, A. E. Merbach, J. Chem.

Soc., Dalton Trans. 2002, 633–641. [57] D. E. Fremgen, E. S. Smotkin, R. E.

[58]

[59] [60]

[61]

[62]

[63]

[64]

[65]

[66]

[67]

[68]

[69]

Gerald, R. J. Klingler, J. W. Rathke, J. Supercrit. Fluids 2001, 19, 287–298. K. W. Kramarz, R. J. Klingler, D. E. Fremgen, J. W. Rathke, Catal. Today 1999, 49, 339–352. J. A. Iggo, D. Shirley, N. C. Tong, New J. Chem. 1998, 1043–1045. J. A. Iggo, Y. Kawashima, J. Liu, T. Hiyama, K. Nozaki, Organometallics 2003, 22, 5418–5422. D. T. Brown, T. Eguchi, B. T. Heaton, J. A. Iggo, R. Whyman, J. Chem. Soc., Dalton Trans. 1991, 677–683. B. T. Heaton, S. P.A. Hebert, J. A. Iggo, F. Metz, R. Whyman, J. Chem. Soc., Dalton Trans. 1993, 3081–3084. G. R. Eastham, B. T. Heaton, J. A. Iggo, R. P. Tooze, R. Whyman, S. Zacchini, J. Chem. Soc., Chem. Commun. 2000, 609–610. S. C. van der Slot, P. C. Kamer, P. W.N. M. van Leeuwen, J. A. Iggo, B. T. Heaton, Organometallics 2001, 20, 430–441. G. R.Eastham, W.Clegg, M. R.J. Elsegood, B. T.Heaton, J. A. Iggo, R. P. Tooze, R. Whyman, S. Zacchini, Organometallics 2002, 21, 1832–1840. W. Clegg, G. R. Eastham, M. R.J. Elsegood, B. T. Heaton. J. A. Iggo, P. R. Tooze, R. Whyman, S. Zacchini, J. Chem. Soc., Dalton Trans. 2002, 3300–3308. R. Whyman, A. P. Wright, J. A. Iggo, B. T. Heaton, J. Chem. Soc., Dalton Trans. 2002, 771–777. J. Wolowska, G. R. Eastham, B. T. Heaton, J. A. Iggo, C. Jacob, R. Whyman, Chem. Comm. 2002, 2784–2785. J. Liu, B. T. Heaton, J. A. Iggo, R. Whyman, Angew. Chem. Int. Ed. Engl. 2004, 43, 90–94.

3 The Use of High Pressure Infrared Spectroscopy to Study Catalytic Mechanisms Anthony Haynes

3.1

Introduction

High pressure infrared (HP IR) spectroscopy has now been used for over 30 years for the study of homogeneous transition metal catalysed processes. The technique is particularly useful for reactions involving carbon monoxide, for which transition metal carbonyl complexes are key intermediates in the catalytic mechanisms. Such complexes have one or more strong n(CO) absorptions, the frequencies and relative intensities of which provide information about the geometry and electronic character of the metal center. As well as probing the metal species, HP IR spectroscopy can also be used to monitor the depletion and formation of organic reactants and products if they have appropriate IR absorptions. The principal catalytic carbonylation reactions are: x x x x x

alkene hydroformylation methanol carbonylation alkene methoxycarbonylation alcohol homologation CO/alkene co-polymerization

These reactions typically employ high pressures of CO or syn-gas (CO/H2) and high temperatures, so in situ spectroscopy under catalytic conditions requires a cell of appropriately robust design. Such in situ studies, however, will often only provide information about the dominant metal complexes, or catalyst resting states. By definition, many important intermediates in a catalytic mechanism are very reactive and short-lived, making them exceedingly difficult to observe directly under working catalytic conditions. Therefore a number of complementary strategies must also be employed to probe reaction mechanisms in more detail, for example: x x x

kinetic studies of stoichiometric reaction steps spectroscopic identification of reactive intermediates at low temperature rapid detection of intermediates generated photochemically


108

3 The Use of High Pressure Infrared Spectroscopy to Study Catalytic Mechanisms

HP IR spectroscopy has been coupled with each of these approaches, leading to a number of recent advances in both cell-design and mechanistic understanding. Another important area of current interest is the use of supercritical fluids as reaction media for catalytic reactions. By their very nature, supercritical fluids require high pressures, and HP IR has also played a significant role in this field. This chapter begins, after this brief Introduction, by considering the different designs of HP IR cell, with particular emphasis on more recent developments. Applications of HP IR spectroscopy to mechanistic studies of catalytic reactions will then be discussed, illustrated by examples of both in situ catalytic investigations and model stoichiometric reactions. The chapter will concentrate on homogeneous catalytic processes. The reader is referred elsewhere for coverage of in situ IR spectroscopic methods in heterogeneous catalysis [1].

3.2

Cell Design

HP IR cells need to exhibit high mechanical strength and resistance to corrosion by solvents and reagents. They are often fabricated from austenitic steels (e. g. type 316) which are satisfactory for relatively mild temperatures and pressures but can be corroded by acid or form [Fe(CO)5] and [Ni(CO)4] by reaction with CO. Alternative materials for construction include some titanium alloys (which can be vulnerable to primary alcohols at high temperature) and nickel–molybdenum–chromium alloys (e. g. Hastelloy C-276, Hastelloy B2) which are highly resistant to reducing, oxidising and acidic conditions. HP IR cells can be categorised into two types, based on their optical configurations, namely transmission cells and reflectance cells. These are represented schematically in Figure 3.1. A transmission cell employs IR transparent windows of high mechanical strength (e. g. CaF2, ZnS) between which the sample solution is contained, such that the IR beam from a spectrometer can pass directly through the sample. High pressure transmission cells can be regarded as more robust versions of the liquid cells routinely used in many laboratories for measuring solution IR spectra under ambient conditions. In reflectance cells, in contrast, the IR beam is directed through an IR-transmitting crystal which has a surface which is in intimate contact with the sample solution. The system is arranged such that one or more internal reflections of the IR beam occur at the interface of the crystal and sample. At each reflection, an evanescent wave is generated which penetrates a short distance into the sample medium,

Figure 3.1 Schematic representation of transmission and reflectance methods.

3.2 Cell Design

109

resulting in attenuation of IR intensity at frequencies absorbed by the solvent or solutes. The essential details and some recent developments in HP IR cells based on both transmission and reflectance approaches are discussed in the following sections. 3.2.1

Transmission Cells

HP IR transmission cells can be divided into two broad categories, namely (i) where the contents of the high pressure vessel are observed directly through IR transparent windows and (ii) where the reaction solution is circulated from the autoclave to an auxiliary observation cell. The first type is exemplified by the cell shown in Figure 3.2, developed by Whyman at ICI [2, 3]. The stirred reaction solution surrounds the cell windows in an arrangement that minimises the problem of

: anti-extrusion rings

: cell o-rings : spring

: Hastelloy-Sheathed soft iron core : solenoid : window holder : window holder o-ring : window : screw cap : stirrer : entrance port : band heater : cell body : exit port

Figure 3.2 The transmission cell design of Whyman (from Ref. [3], reproduced by permission of IOP Publishing Ltd).

110


stagnant volume between the windows. Exchange of solution between the bulk and the observation area is also aided by castellations on the ends of the screw caps holding the windows. This type of cell can be regarded as affording a truly in situ measurement of the reaction solution. Cells of the second type were initially developed by Tinker and Morris at Monsanto [4] and subsequently by Penninger [5]. In these systems, the reaction solution is circulated from the autoclave through an external IR cell of relatively small volume. This arrangement means that the cell can be isolated from the main reaction vessel relatively easily (for example in the event of window failure) thus protecting the spectrometer. Cells of this sort can, in principle, be fitted to plants or pilot plants to monitor liquid streams. However, the circulation of solution from the main reaction vessel through an external cell introduces some potential problems. A pressure drop in the circulation system can lead to release of dissolved gas, which may accumulate between the cell windows and interfere with the spectroscopic measurement. A change in pressure may also influence the catalyst speciation, such that the observed spectra may not be truly representative of the bulk reaction solution. In recent years, a number of modifications to HP IR transmission cells have been developed for applications to particular problems. Some of these are considered in detail below.

Amsterdam Flow Cell For HP IR systems in which the reaction solution from the main autoclave is circulated through an external spectroscopic cell, the residence time in the circulation system can be problematic. This issue has been addressed by a cell design developed by the group of van Leeuwen in Amsterdam, with the aim of monitoring relatively fast reactions [6, 7]. A novel aspect of this design, shown in Figure 3.3, is that rotation of the stirrer blades is used to force the solution from the lower chamber of the reaction vessel through the spectroscopic cell and back into the upper chamber. Whilst the total volume of the autoclave is 50 cm3, the volume of liquid in the circulation loop is only 0.35 cm3 and circulation velocities of 1.4–7.7 cm3 s–1 can be achieved (with a 0.4 mm window spacing). This means that under typical operating conditions (2200 rpm stirring) the reaction solution reaches the center of the IR beam 56 ms after leaving the main vessel, and this time can be lowered to 33 ms at the maximum stirring rate. The total volume is kept relatively low (15 cm3) to ensure efficient mixing between upper and lower chambers of the main autoclave unit. The cell has been used, together with rapid FTIR scanning, to study the species present during hydroformylation reactions using a phosphite modified rhodium catalyst (see Section 3.3.1.2). It has also been employed to investigate elementary reaction steps in the hydroformylation mechanism, such as CO ligand dissociation and exchange between rhodium hydrides and H2 (see Sections 3.3.2.2 and 3.3.2.3) [6, 8]. 3.2.1.1

3.2 Cell Design

Figure 3.3 Side- and top-view of the in situ infrared autoclave. A: IR window; B: Turbine rotor; C: Reagent addition; D: Thermocouple; E: Opening between upper and lower chamber; F: Electrical heaters; G: Kalrezr O-rings (from Ref. [6], reproduced by permission of Elsevier).

Low-temperature HP IR Cells At low temperatures, the lifetime of reactive species can be extended for spectroscopic observation and characterisation. This approach has been used in combination with liquid noble gases (LNGs) as solvents for identification of reactive transition metal complexes. LNGs are potentially attractive inert solvents for spectroscopy, due to their transparency in the IR and UV/Vis regions of the spectrum. This enables relatively long pathlength cells to be employed, allowing detection of species at quite low concentration. Although the noble gases can be liquefied at cryogenic temperatures and atmospheric pressure, their liquid ranges are rather restricted (e. g. Ar –189.3 to –185.8 hC; Kr –157.4 to –153.2 hC; Xe –111.7 to –108 hC). However, the liquid range can be extended substantially at higher pressures, for example, at 22 bar, liquid xenon can be used as a solvent over the temperature range ca. –113 to –25 hC. In combination, argon, krypton and xenon can be used as liquid solvents over the entire temperature range between –196 and –33 hC at 15 bar. HP IR cells for this purpose were originally designed by the group of Turner and Poliakoff in Nottingham in collaboration with Maier [9, 10]. The cell, shown in Figure 3.4 consists of a machined copper block which sits inside an evacuated jacket (to prevent condensation on the cooled part of the cell). The cell configuration allows both photolysis (to generate the unstable species of interest) and IR spectroscopy. To this end, it has two sets of windows. One opposing pair is made of KRS-5 to transmit IR light with a cell pathlength of ca. 2.5 cm, whilst the other pair is made of CaF2 or quartz for photochemical excitation with UV or 3.2.1.2

111

112


Figure 3.4 Low temperature cell for HP IR spectroscopy in liquid noble gases (from Ref. [10], reproduced by permission of The Royal Society of Chemistry).

visible light. Sealing is by means of lead gaskets and clamping plates. Cooling of the cell is achieved by pulsing liquid nitrogen through a chamber in the cell body. Investigations of photochemical reactions of metal carbonyl species in LNGs has enabled many unstable products to be characterised by IR spectroscopy, notably including complexes containing coordinated noble gas atoms (e. g. [Cr(CO)5Xe] [11]) or molecular dihydrogen (Section 3.3.2.6) [12–14]. Intermediates in catalytic hydrogenation reactions have also been investigated [15]. Similar cells have been used by other groups for IR spectroscopy in LNGs. Notably, Bergman and co-workers have studied photochemical C–H activation reactions (Section 3.3.2.7) [16–20].

HP IR Cells for Flash Photolysis Ford and co-workers have developed a HP IR cell for use in flash photolysis experiments with time-resolved IR (TR IR) spectroscopic monitoring [21, 22]. The apparatus, shown in Figure 3.5, comprises a conventional Parr autoclave from which reaction solution, equilibrated at the desired pressure and temperature, is passed into a transmission cell adapted from a design of Noack [23]. CaF2 windows are used to allow transmission of UV/Vis as well as IR radiation. Flash photolysis of the cell contents is achieved with a XeCl Excimer laser (308 nm) or with a Nd/ YAG laser operating at the second (532 nm), third (355 nm) or fourth (266 nm) harmonic. Detection of transients is achieved by a tuneable IR laser source and fast rise time detector. Since signal averaging is generally required to achieve satisfactory signal/noise ratios in these experiments, multiple laser pulses of the sample are necessary. The sample solution must therefore flow through the cell at a 3.2.1.3

3.2 Cell Design

Figure 3.5

Ford’s HP IR cell for flash photolysis (from Ref. [22], reproduced by permission of

Elsevier).

rate which is slow relative to the lifetime of transients but fast enough to renew the sample between laser pulses. This is achieved by a double metering valve downstream of the IR cell or by maintaining the capture vessel at an appropriate pressure. The system has been used to probe the reactive intermediates in migratory CO insertion reactions of manganese and cobalt complexes (see Section 3.3.2.1). HP IR spectroscopy has also been coupled successfully with flash photolysis by George and co-workers for experiments in supercritical fluids. These studies were aimed at quantifying the reactivity of noble gas and alkane complexes close to ambient temperature, rather than in cryogenic LNGs (Sections 3.3.2.6 and 3.3.2.7) [24, 25]. Supercritical fluids exhibit properties which are hybrid between those of liquids and gases. For example, they can dissolve a range of compounds, including metal complexes, but are also completely miscible with gases, allowing high concentrations of gases like H2 or CO to be used. Xenon has a critical point of TC ¼ 16.8 hC and PC ¼ 58 bar while carbon dioxide has TC ¼ 31.6 hC and PC ¼ 73 bar. The TR IR experiments in supercritical fluids used a high pressure IR transmission cell with CaF2 windows (Figure 3.6) [26, 27]. The TR IR measurements utilised an Excimer or Nd/YAG laser for flash photolysis and a tuneable IR diode laser or a step-scan FT IR instrument [28] to detect transients.

113

114


Figure 3.6 Cross-section (a) and external view (b) of stainless steel HP IR cell used for experiments supercritical fluids. P ¼ fill ports; E ¼ epoxy resin, used to seal CaF2 windows. (From Ref. [26], reproduced by permission of Wiley-VCH).

3.2.2

Reflectance Cells

The main alternative to conventional transmission cells for HP IR spectroscopy is attenuated total reflectance (ATR) spectroscopy. In this technique, the sample is held in contact with the surface of an IR transmitting crystalline material of relatively high refractive index, such as silicon or zinc selenide. The IR beam is directed through the crystal such that it undergoes a number of internal reflections at the crystal–sample interface. At each internal reflection, an evanescent wave is generated, which penetrates a short distance (of the order of microns) into the surrounding medium. The intensity of the evanescent wave decays exponentially with distance from the surface of the ATR crystal as illustrated in Figure 3.7. The energy of the evanescent wave is attenuated in regions of the IR spectrum containing absorptions of the sample (for example the solvent and any dissolved species). Thus the reflected radiation carries away with it the absorption spectrum of the sample. The depth of penetration (at which the intensity of the evanescent wave has decayed to 1/e of its original value) is given by Eq. (1): l

depth w 2pn1

(1)

2 !12 n2 2 sin u – n1

Figure 3.7 The evanescent wave in ATR spectroscopy.

3.2 Cell Design

where l is the wavelength of IR radiation, n1 is the refractive index of the ATR crystal, n2 is the refractive index of the sample and u is the angle of incidence. The effective penetrations are actually different for parallel and perpendicular polarizations, with more complex mathematical descriptions. However, the simpler equation given above serves to illustrate that penetration increases with IR wavelength (l) and with decreasing refractive index of the ATR crystal (n1). The ATR technique is now routinely used for IR spectroscopy as it allows measurement of spectra for a variety of sample types with minimal preparation. The crystals employed are generally prismatic in shape, allowing contact of a flat surface with the sample. The ATR method was first adapted for HP IR spectroscopy by Moser [29–33], who realised that a conventional autoclave could easily be adapted for in situ IR spectroscopy by fitting an ATR crystal of cylindrical cross section. The technique developed by Moser is thus known as cylindrical internal reflectance (CIR) spectroscopy and high pressure cells based upon the CIR method have been commercialised by Spectra-Tech. A typical CIR cell is illustrated in Figure 3.8. The CIR crystal (silicon in this case) is inserted through holes in the body of the autoclave, and sealed by means of tight fitting Teflon O-rings held in place by retaining clamps. The IR light from the spectrometer is directed onto one of the conically shaped ends of the crystal, by using either mirrors or optical fibers [34]. The IR beam then follows a zigzag pathway through the crystal via multiple internal reflections, a number of which interact with the reaction solution at its interface with the CIR crystal. Since the crystal is immersed in the reaction solution, this is a truly in situ technique, with efficient stirring and gas–liquid mixing being straightforward. The overall effective pathlength of a CIR cell is relatively short (ca. 10 mm) which has both advantages and disadvantages. The short pathlength means that spectra can be measured even on highly absorbing samples, such as aqueous solutions, or (as in the case of methanol carbonylation reactions) acetic acid–water mixtures. The downside, however, is that IR absorptions of the solute species of interest

Figure 3.8 Schematic of a CIR-HP IR cell (adapted from Ref. [35] by permission of The Royal Society of Chemistry).

115

116


(e. g. metal complexes) are also weak, sometimes requiring the use of relatively high metal concentrations. This drawback can, however, be overcome to a certain extent using highly sensitive, liquid nitrogen cooled IR detectors (e. g. mercury-cadmium-telluride, or MCT) which are commonplace on many modern FTIR spectrometers. Thus, very high signal/noise ratios can be attained, allowing even weak absorptions to be observed reliably. Examples of the use of CIR cells of this type include studies of hydroformylation [30, 31, 36], methanol carbonylation [32, 33, 35, 37–39], bromobenzene carbonylation [40], nitroaromatic carbonylation [41] and hydrocarbon cracking [42]. As well as the CIR-type cells discussed above, the ATR technique has also been employed in other types of HP IR cell. An alternative arrangement to a cylindrical crystal inserted through the autoclave body (Figure 3.8) is to embed the ATR crystal in the base or wall of the autoclave. This approach was used in the HP IR cell developed by Wolf et al., illustrated in Figure 3.9 [43]. An ATR crystal is mounted in the bottom of the reactor, which can operate at pressures up to 200 bar and can be adapted to give a high pressure flow-cell.

Figure 3.9 ATR-HP IR cell design of Wolf et al. (from Ref. [43], reproduced by permission of Elsevier).

High pressure ATR cells have been commercialised by Mettler-Toledo, in their ReactIR reaction monitoring systems. In the so-called “Comp Probes”, the ATR crystal is a thin wafer of silicon (Si-COMPä) or diamond (Di-COMPä), although zirconia and zinc sulfide are other possibilities. The wafer is sealed with gold into a Hastelloy housing at the tip of the probe. Immediately adjacent to the ATR crystal is a zinc selenide element which focuses the IR radiation onto the wafer, as illustrated in Figure 3.10. The probe can be mated with a suitably engineered autoclave in a variety of configurations (e. g. through the reactor wall or base, in a flow loop or via a dip tube). ReactIR probes of this type have been used, for example, in studies of methanol carbonylation [39], catalytic ethylene carbonylation [44], enantioselective hydrogenation of imines in supercritical carbon dioxide (scCO2) [45] and co-polymerization of CO and epoxides [46]. Schneider et al. have reported a cell for HP IR spectroscopy which combines both transmission and reflectance methods for in situ investigations of multiphase reactions [47]. The upper part of the cell can be monitored by variable pathlength

3.3 Mechanistic Studies using High Pressure IR Spectroscopy

Figure 3.10 Cut-away schematic of ReactIR ATR probe (reproduced with permission of Mettler-Toledo).

transmission IR spectroscopy, while the bottom of the cell can be probed by ATR spectroscopy. The ATR crystal can be coated with a catalyst film, such that both phases of heterogeneous reactions can be monitored quasi-simultaneously. The utility of the cell was demonstrated in studies of the heterogeneous catalytic hydrogenation of ethyl pyruvate over Pt/alumina in scC2H6, and in the homogeneous catalytic formylation of morpholine with scCO2/H2.

3.3

Mechanistic Studies using High Pressure IR Spectroscopy

This section will describe the various applications of HP IR spectroscopy to determine reaction mechanisms of transition metal catalysed reactions. It will begin by looking at truly in situ studies, carried out under catalytic conditions, and then consider investigations of stoichiometric reaction steps and characterisation of reactive intermediates. 3.3.1

In situ Studies under Catalytic Conditions 3.3.1.1

Methanol Carbonylation

Rhodium catalysts

In situ HP IR studies of methanol carbonylation reactions were originally undertaken by Forster and co-workers at Monsanto. Using a transmission cell, it was shown that for the rhodium catalyst, the dominant species (at relatively high

117

118


[H2O]) is [Rh(CO)2I2]– [48]. The same Rh(I) complex was observed during carbonylation of higher alcohols [49], leading to the conclusion that oxidative addition of alkyl iodide is rate determining. Under conditions of lower [H2O] build up of [Rh(CO)2I4]– occurs [50]. This Rh(III) tetraiodide is an inactive species which is an intermediate in the water-gas shift (WGS) reaction, an important side reaction to methanol carbonylation. For the related carbonylation of methyl acetate to acetic anhydride, Schrod and Luft observed both [Rh(CO)2I2]– and [Rh(CO)2I4]– by HP IR [51]. Further spectroscopic studies by researchers at Eastman demonstrated that the inactive [Rh(CO)2I4]– could be converted back into active [Rh(CO)2I2]– by introducing 5 % hydrogen into the CO gas feed [52]. Keeping the catalyst in its Rh(I) form has the effect of removing an induction period and maintaining high activity. HP IR has also shown that [Rh(CO)2I2]– is the only observable Rh species during the rhodium iodide catalysed carbonylation of methyl formate into acetaldehyde or methyl acetate [53]. Ligand-modified rhodium catalysts have also been probed by HP IR. For example, the bidentate P,S donor ligand dppms (1) gives a very active (although not particularly robust) methanol carbonylation catalyst. During the period of high activity the sole Rh carbonyl species observed was the Rh(I) chelate complex, [Rh(CO)(dppms)I], indicating that MeI oxidative addition remains rate determining [54]. This was confirmed by subsequent mechanistic and kinetic studies [55, 56]. A similar P,O donor ligand, dppeo (2), has been found to promote methanol carbonylation under quite mild conditions, but in that case the potentially bidentate ligand was found to be hemi-labile [57]. The sole complex observed at 80 hC and 3.3 bar CO was cis-[Rh(CO)2(dppeo)I] with a monodentate P-coordinated dppeo ligand. By contrast, dppmo (3) forms a more robust 5-membered chelate ring and only a trace of the dechelated cis-[Rh(CO)2(dppmo)I] was detected by IR under 3 bar CO. The Rh(I) complex [Cp*Rh(CO)2] was found to be active for carbonylation of methyl acetate to acetic anhydride, and HP IR spectroscopy indicated conversion to [Cp*Rh(CO)I2] on reaction with MeI under 20–100 bar CO [58]. PPh2 Ph2P

S

PPh2

PPh2 Ph2P

O

Ph2P

O

1

2

3

dppms

dppeo

dppmo

Iridium catalysts

Forster also reported HP IR measurements on iridium catalysed reactions [59]. It was noted that the iridium speciation is dependent on reaction conditions, with three different regimes being distinguishable. At intermediate [H2O], the dominant Ir species are [MeIr(CO)2I3]– and [Ir(CO)2I4]–. The anionic methyl complex is regarded as the active form of the catalyst in a cycle analogous to the Rh system, with carbonylation of [MeIr(CO)2I3]– being rate determining. The Ir(III) tetraiodide


119

(a WGS intermediate) is an inactive form. At high [H2O] significant amounts of a hydride, [HIr(CO)2I3]– were also present, and WGS activity was substantial. By contrast, at low [H2O] the iridium speciation shifts towards neutral tricarbonyl complexes, [Ir(CO)3I] and [Ir(CO)3I3]. In this regime oxidative addition of MeI to Ir(I) is thought to be rate determining, and this is inhibited by increasing the CO pressure. On the basis of these observations, Forster proposed a mechanism for iridium catalysed carbonylation that comprised two linked cycles based on neutral or anionic complexes (Scheme 3.1). The dominant cycle depends on the process conditions, particularly the concentrations of water and iodide. There has been a recent resurgence of interest in iridium catalysed methanol carbonylation, arising from the commercialisation by BP Chemicals of the Cativa process. This uses a promoted iridium catalyst and has now superseded the rhodium catalyst on a number of plants. Its success relies on the discovery of promoters which increase catalytic activity, particularly at commercially desirable low water concentrations. HP IR spectroscopy has been used to investigate the behavior of

H2

HI + CO

[Ir(CO) 3I3]

HI

H2 [Ir(CO) 2I4] -

[HIr(CO) 2I3] -

[HIr(CO) 2I2]

H2O

H2O + CO

CO2 + 2 HI

HI

HI I-

[Ir(CO) 3I]

CO

CO

MeI

MeCOI

[(MeCO)Ir(CO) 3I2]

[Ir(CO) 2I]

CO MeI

CO2 + 2 HI

[Ir(CO) 2I2] -

Neutral Cycle

[MeIr(CO) 2I3] -

[(MeCO)Ir(CO) 2I3] I-

[(MeCO)Ir(CO) 2I2]

Anionic Cycle

[MeIr(CO) 2I2]

[MeIr(CO) 2I2] [MeIr(CO) 3I2]

CO

CO

Organic reactions:

I-

MeOH + HI MeCOI + H2O

MeI + H2O MeCO2H+ HI

Scheme 3.1 Anionic and neutral cycles proposed by Forster for iridium catalysed methanol carbonylation and WGS reaction (adapted from Ref. [59] by permission of The Royal Society of Chemistry).

120


both the iridium catalyst and the promoters [39, 60]. Initial studies were carried out using a transmission cell but analogous results were also obtained using an autoclave equipped for ATR-IR spectroscopy. At 190 hC and 22 bar pressure, the catalyst precursor, H2[IrCl6], is rapidly converted into iridium iodocarbonyl species. For [H2O] j ca. 5 % w/w, the predominant iridium species is initially [MeIr(CO)2I3]– and a small amount of [Ir(CO)2I2]– is also observed. As the catalytic reaction proceeds, conversion of [MeIr(CO)2I3]– into [Ir(CO)2I4]– is observed (Figure 3.11(a)).

Figure 3.11 In situ ATR HP IR spectra from batch carbonylation runs with (a) an Ir-only catalyst (b) a Ru promoted Ir catalyst (2:1 Ru: Ir) (adapted from Ref. [39], Supplementary Information, by permission of the American Chemical Society).


This is due to the batch nature of the experiment in which the substrate is added as methyl acetate to mimic the actual process where methanol exists largely as its esterified form. The net carbonylation reaction can be represented by Eq. (2). 2 MeOAc þ H2O þ CO p 2 AcOH

(2)

Thus, [H2O] decreases during the reaction, which slows the rate of reduction of [Ir(CO)2I4]– to [Ir(CO)2I2]–. At 50 % substrate conversion, the ratio of [MeIr(CO)2I3]– and [Ir(CO)2I4]– is ca. 1:1. On addition of LiI or Bu4NI the Ir catalyst is predominantly [MeIr(CO)2I3]–; thus iodide salts appear to lower the activity of [MeIr(CO)2I3]– rather than convert it into a different, inactive form. In experiments at low [H2O] (I ca. 4 % w/w), bands due to [Ir(CO)3I] and [Ir(CO)3I3] indicate operation of the “neutral cycle” consistent with Forster’s earlier studies. Two classes of promoter have been identified for iridium catalysed carbonylation: (i) transition metal carbonyls or halocarbonyls; (ii) simple group 12 and 13 iodides. Increased rates of catalysis are achieved on addition of 1–10 mole equivalents (per Ir) of the promoter. An example from each class was chosen for spectroscopic study. An InI3 promoter provides a relatively simple system since the main group metal does not tend to form carbonyl complexes which can interfere with the observation of iridium species by IR. In situ HP IR studies showed that an indium promoter (InI3 :Ir ¼ 2:1) did not greatly affect the iridium speciation, with [MeIr(CO)2I3]– being converted into [Ir(CO)2I4]– as the batch reaction progressed, as in the absence of promoter. With a ruthenium promoter (added as [Ru(CO)4I2]), n(CO) bands due to Ru iodocarbonyls dominated the spectrum, precluding the easy observation of iridium species. Before injection of the Ir catalyst, absorptions due to [Ru(CO)2I2(sol)2], [Ru(CO)3I2(sol)] and [Ru(CO)3I3]– are present. After injection of the iridium catalyst (Ru:Ir ¼ 2:1), [Ru(CO)3I3]– becomes the dominant Ru species (Figure 3.11(b)). The observations indicate that the Ru(II) promoter has a high affinity for iodide and scavenges HI(aq) as H3Oþ[Ru(CO)3I3]–. An indium promoter is believed to behave in a similar manner to form H3Oþ[InI4]–. These promoter species also catalyse the reaction of HI(aq) with methyl acetate (Eq. (3)), which is an important organic step in the overall process. H3Oþ þ I– þ MeOAc w MeI þ AcOH þ H2O

(3)

A key role of the promoter is to moderate the concentration of “free” iodide. This has a strong influence on catalytic rate, as carbonylation of [MeIr(CO)2I3]– is poisoned by iodide. The mechanism in Scheme 3.2 shows how loss of iodide from [MeIr(CO)2I3]– is required to generate neutral species which lie on the main pathway for catalytic turnover. Studies on the stoichiometric reaction steps (see Section 3.3.2.1) show that the methyl migration in the neutral tricarbonyl [MeIr(CO)3I2] is greatly favored over the anionic resting state. The mechanism also explains the influence of [H2O] on the Ir speciation. The equilibrium in Eq. (3) means that concentrations of water and I– are coupled. Thus, high [H2O] leads to high [I–], and

121


122

AcOH

[Ir(CO)2I2]– H+ H+ I–

AcOH

MeOH

MeI promoter

[Ir(CO)2I] MeOAc

CO

H 2O

H+ I –

[Ir(CO)3I] H2O [AcIr(CO)3I2] CO

[MeIr(CO)3I2] CO

[AcIr(CO)2I2]

[MeIr(CO)2I2] H+ I–

H+ I–

[MeIr(CO)2I3]– H+

[AcIr(CO)2I3]– H+

CO

[AcIr(CO)I3]–

Mechanism for promoted iridium catalysed methanol carbonylation. The red arrows indicate the dominant pathway for catalytic turnover (Ac ¼ C(O)Me). (Adapted from Ref. [39] by permission of the American Chemical Society).

Scheme 3.2

favors an anionic form of the Ir catalyst, [MeIr(CO)2I3]–. At low [H2O], and correspondingly low [I–], neutral Ir species are favored and the equilibria at the top of Scheme 3.2 lead to build up of the catalyst as [Ir(CO3I]. Nickel catalysts

Moser et al. investigated PPh3 -promoted nickel/iodide catalysts for methanol carbonylation by HP IR using a CIR cell [33]. Under conditions optimised for catalytic activity (160 hC, 60 bar) anionic Ni carbonyls (e. g. [Ni(CO)I3]–) were not observed and only a trace of [Ni(CO)4] was detected at steady state. The concentration of the tetracarbonyl increased during an initial induction period and after all of the methanol substrate had been consumed. It was concluded that the dominant Ni species during the period of catalytic activity contains no Ni–CO bonds, and that conditions required for optimum rates are the same as those needed to minimise the concentration of [Ni(CO)4]. A mechanism with [Ni(PPh3)2] as the active species was proposed, with rate determining (and reversible) addition of MeI. The HP IR spectra exhibited weak overlapping absorptions due to MePPh3þI–, PPh3 and [Ni(PPh3)n] in the region of 1580 cm–1. The anion [Ni(CO)I3]– (previously suggested as the active catalyst) was only detected at low concentration under non-optimum conditions, and no relationship was found between this species and carbonylation activity. It is possible that the major Ni-containing species present is the Ni(II)


complex, [NiI4]2–. The necessity for H2 to maintain high catalytic activity is presumably to reduce Ni(II) to Ni(0), which coordinates phosphine ligands to generate the active [Ni(PPh3)2]. In a subsequent paper [32], Moser et al. studied the effect of changing the steric and electronic properties of the phosphine ligand. HP IR was used to estimate the proportion of PR3 ligand in its unquaternised from (i. e. PR3 and [Ni(PR3)n]) relative to MePR3þI–. Whilst the phosphine was found to be substantially converted into its phosphonium salt by reaction with MeI, the in situ spectroscopic studies indicated the presence of “free” phosphine, and catalytic rates increased linearly with concentration of free PR3 in solution. This is consistent with the proposal that [Ni(PR3)2] is the active species.

Hydroformylation The essential features of the alkene hydroformylation mechanism proposed by Heck and Breslow [61] remain intact, after many investigations using a variety of techniques. The cycle shown in Scheme 3.3 is that for the unmodified, cobalt 3.3.1.2

- H2 HCo(CO)4

R

+ H2

O

O

H

H

+CO

-CO

Co2(CO)8

R

R HCo(CO)3 H2 O R

Co(CO)3

O -CO R

R O HCo(CO)3

Co(CO)3

Co(CO)4

R

+CO O Co(CO)4 R R

Co(CO)4

R

Co(CO)3

Co(CO)4

Co(CO)3

R

Alkene hydroformylation mechanism for an unmodified cobalt catalyst.

R

Scheme 3.3

CO

123

124


catalysed reaction, but the mechanisms of rhodium catalysed hydroformylation, and of ligand modified systems are thought to follow the same basic steps. Modification by P-donor ligands, however, means that one or more of the CO ligands may be replaced at each stage of the cycle, which also leads to the possibility of geometrical isomers. The rate determining step and the dominant observable species are dependent on the metal (Co or Rh) and the modifying ligands used. HP IR has played a key role in these studies, often complemented by 31P HP NMR for the P-ligand modified systems. Early HP IR studies of cobalt catalysed alkene hydroformylation were carried out by Whyman [62]. It was shown that under catalytic conditions (but in the absence of alkene substrate), the dimeric catalyst precursors [Co2(CO)8] and [Co2(CO)6(PBu3)2] are converted into the hydrides [HCo(CO)4] and [HCo(CO)3(PBu3)], respectively. In the absence of phosphine, a mixture of [Co(C(O)R)(CO)4] (R ¼ C8H17) and [Co2(CO)8] was observed during hydroformylation of 1-octene (the proportion of acyl complex increasing with pCO) whereas for internal alkene substrates, only [HCo(CO)4] and [Co2(CO)8] were observed. Hydrogenolysis of the acyl (requiring loss of CO before H2 activation) was suggested as the rate determining step. Contrary to this, Penninger reported that only very small concentrations of acyl complexes could be detected, and suggested that aldehyde formation occurred exclusively via a bimolecular reaction between [Co(C(O)R)(CO)4] and [HCo(CO)4] [63]. However, subsequent HP IR studies by Mirbach [64] and by Pino et al. [65] found hydrogenolysis of the cobalt acyl to be dominant, a conclusion supported by the kinetic studies of Kovcs [66]. In the phosphine modified system, acyl species are not observed during 1-octene hydroformylation and [HCo(CO)3(PBu3)] is the dominant species, while [HCo(CO)2 (PBu3)2] and [Co2(CO)7(PBu3)] are formed, respectively, at high and low phosphine concentrations [62]. The rate determining step was proposed to be reaction of alkene with [HCo(CO)3(PBu3)], via initial dissociation of a CO ligand. Equilibrium constants and thermodynamic data for some of the reactions of cobalt complexes with H2, CO and PBu3 were determined by Penninger and co-workers in heptane over the range 100–150 hC [67]. HP IR measurements have recently been reported by workers at Sasol for cobaltcatalysed 1-dodecene hydroformylation reactions using bicyclic phosphines (4) derived from (R)-(þ)-limonene [68]. Using Fourier deconvolution to separate absorptions due to [HCo(CO)4] and [Co2(CO)7(phosphine)], it was possible to estimate the ratio of “modified” [HCo(CO)3(phosphine)] to “un-modified” [HCo(CO)4] in the catalytic mixture, using peak areas. Values of this ratio ranged from ca. 2–20, depend-

P

R

4


ing on the nature of the R group in the phosphine ligand, and showed a correlation with catalyst activity and selectivity. A higher proportion of “modified” [HCo(CO)3 (phosphine)] was found for catalysts which were less active but more selective towards the desired linear products. In a recent study of a Co/triphenylphosphite catalyst system, HP IR indicated [HCo(CO)3(P(OPh)3)] at ca. 110 hC, but at higher temperatures absorption bands corresponding to [HCo(CO)4] were observed. At higher P(OPh)3 concentrations, the less active dicarbonyl [HCo(CO)2{P(OPh)3}2] was also present [69]. Rhodium-based hydroformylation catalysts have been the subject of numerous investigations by HP IR spectroscopy. Unmodified catalysts have received substantial study by Garland and co-workers. Under catalytic conditions, rhodium precursors such as [Rh4(CO)12], [Rh6(CO)16], [Rh(CO)2Cl]2 and RhCl3, are converted into the catalyst resting state, [Rh(C(O)R)(CO)4], as are mixed-metal precursors such as [CoRh(CO)7] and [Co2Rh2(CO)12].[70] For hydroformylation of 3,3-dimethylbut1-ene, it was shown that catalytic rate is proportional to the concentration of [Rh(C(O)R)(CO)4] (independent of the precursor), as judged by HP IR spectroscopy. On this basis, the apparent higher activity of Co–Rh catalysts was ascribed to facile fragmentation of the mixed-metal precursors rather than cluster catalysis or a binuclear elimination mechanism. Acyl complexes [Rh(C(O)R)(CO)4] were also observed for a range of alkene types (terminal alkenes, cyclic alkenes, symmetrical linear internal alkenes, methylene cycloalkenes) [71]. It was concluded that the different catalytic rates for different classes of alkene were primarily dependent on the extent of conversion of [Rh4(CO)12] into [Rh(C(O)R)(CO)4] rather than turnover frequency within the active cycle. Ethylene hydroformylation was treated as a separate case, as difficulties arise from dramatic changes in the IR spectrum of dissolved ethylene as a function of its partial pressure. This was overcome using the method of band-target entropy minimisation (BTEM, see Chapter 4) to recover the pure component spectra of all observable species and their concentrations [72]. As well as the conventional acyl tetracarbonyl, [Rh(C(O)Et)(CO)4], evidence was obtained for [Rh(C(O)Et)(CO)3(C2H4)], containing coordinated ethylene. The presence of this species indicates that ethylene can compete with H2 for the unsaturated [Rh(C(O)Et)(CO)3]. The ketone and polyketone side products of Rh-catalysed ethylene hydroformylation arise from insertion of coordinated ethylene into the Rh-acyl bond in [Rh(C(O)Et)(CO)3(C2H4)] (or multiple CO/C2H4 insertions in the case of polyketone). Garland’s studies showed that, in general, catalytic hydroformylation rate is proportional to the concentration of [Rh(C(O)R)(CO)4], and kinetic data for hydrogenolysis of this acyl to liberate acetaldehyde have been determined (see Section 3.3.2.4) The potential participation of an alternative route, involving a binuclear elimination reaction between a metal-acyl and a metal-hydride has also been probed [73]. In Rh-catalysed cyclohexene hydroformylation, both [Rh4(CO)12] and [Rh(C(O)R)(CO)4] are observed by HP IR at steady state, the cluster species being a potential source of [HRh(CO)4] by reaction with syn-gas. The kinetic data for aldehyde formation indicated no statistically significant contribution from binuclear elimination, with hydrogenolysis of the acyl complex dominant. For a mixed Rh–Mn system,

125

126


however, the binuclear elimination route (Eq. (4)) is greatly enhanced, and it was estimated that [HMn(CO)5] is ca. 170 times more efficient than H2 for the hydrogenolysis of the Rh–acyl bond [74]. The mechanism proposed for binuclear reductive elimination involves loss of CO from [Rh(C(O)R)(CO)4] prior to reaction with the Mn hydride. [Rh(C(O)R)(CO)4] þ [HMn(CO)5] p RCHO þ [RhMn(CO)8] þ CO

(4)

Garland and co-workers have used HP IR spectroscopy to detect the elusive [HRh(CO)4] [75]. This hydride is thought to participate in a variety of rhodium catalysed reactions, including hydrogenation, hydroformylation and hydrosilation of alkenes, hydrogenation of aldehydes and alkoxycarbonylation of alkenes, but prior evidence for its observation was scarce. Garland’s spectroscopic measurements used [Rh4(CO)12] as the precursor, subjected to 10–50 bar CO, 10–50 bar H2 (or 5–20 bar D2) in the absence of alkene. Spectral deconvolution was achieved using the BTEM method, which revealed a minor component (ca. 0.15 %) having three n(CO) absorptions and a n(Rh–H) band, comparable with the spectrum of the known [HCo(CO)4]. The minor species was assigned as trigonal bipyramidal [HRh(CO)4] with an axial hydride. Consistent with this, when D2 was used in place of H2, the Rh–H band was lost and two of the three n(CO) bands (those of a1 symmetry) shifted to lower frequency whilst the e mode remained unshifted. Early studies on phosphine-modified Rh hydroformylation systems carried out by Morris and Tinker [76] and by Wilkinson and co-workers [77] used transmission cells where the reacting solution was periodically flowed from a high pressure autoclave into an externally located IR cell. These studies were hampered by the inability to probe the steady state reactor composition. Moser et al. used a CIR cell to gain truly in situ spectroscopic information on triaryl phosphine modified Rh catalysts. For L ¼ PPh3, the major complex observed during hydroformylation of 1-hexene was [HRh(CO)2L2] (known from NMR studies to exists as two rapidly exchanging trigonal bipyramidal isomers with the phosphine ligands occupying bis(equatorial) or equatorial and axial positions [78]). Dissociation of CO from [HRh(CO)2L2], prior to coordination of alkene, was identified as the rate limiting step. Moser et al. also investigated the catalyst deactivation mechanisms for these systems [31]. Conversion of [HRh(CO)2L2] into less active dimeric species was observed, this being faster for more electron-donating phosphine ligands, L. This was followed by formation of an inactive binuclear complex with a bridging phosphido ligand. A PEt3 -modified Rh catalyst in scCO2 solvent was studied by HP IR spectroscopy in the absence of 1-hexene substrate [79]. The catalyst precursor, [Rh2(OAc)4], was found to be converted into a mixture of phosphine-containing complexes, thought to include [HRh(CO)2L2], [HRh(CO)L3] and [Rh2(CO)2L6]. On replacing PEt3 with P(n-octyl)3, no metal carbonyl species were observed in the HP IR spectrum, indicating insolubility of complexes of this ligand in scCO2. Only weak peaks were observed for a PPh3 system, again indicating low solubility, but introduction of paraSiMe3 groups gave a more significant n(CO) peak. The different solubilities were reflected in the hydroformylation activities observed.


Catalyst solubilty in scCO2 can be enhanced by using phosphine ligands containing perfluoroalkyl moieties. The behavior of Rh-PAr3F complexes (ArF ¼ 3,5-(CF3)C6H3) was probed using in situ HP IR spectroscopy by Haji and Erkey [80]. Although the precursor complex, [HRh(CO)(PAr3F)3] did not dissociate phosphine ligand in scCO2 at 50 hC, addition of 4.5 bar CO led to formation of [HRh(CO)2(PAr3F)2]. In the presence of CO and ethylene, a number of bands between 1645 and 1695 cm–1, associated with Rh-acyl species, were observed; these were also detected during a working ethylene hydroformylation reaction. For hydroformylation of cyclooctene using a Rh/tris(o-tert-butylphenyl)phosphite catalyst, van Leeuwen et al. found that the dominant species was the tricarbonyl hydride, [HRh(CO)3L] [81]. H/D isotopic labelling experiments allowed the aldehyde-forming step in the cycle to be identified as reaction of a Rh-acyl complex with H2 (or D2) rather than with a Rh hydride. More recently a study of the Rh/ tris(2-tert-butyl-4-methylphenyl)phosphite catalyst system employed rapid scan IR measurements to elucidate changes in catalyst speciation immediately after introduction of the alkene substrate [6]. The hydride complex [HRh(CO)3L] was generated from [Rh(acac)(CO)2] and the phosphite under CO/H2. On injection of cyclohexene (an internal alkene of relatively low reactivity), the IR spectrum was unchanged apart from the growth of an aldehyde product band. However, on injection of more reactive 1-alkenes, the n(CO) bands rapidly shifted to low frequency relative to those of [HRh(CO)3L] and became broad or shouldered. Using the rapid scan method (up to 80 scans s–1), a Rh-acyl absorption was detected at 1690 cm–1 within 1 s of alkene addition, before this region of the spectrum was overwhelmed by the aldehyde product band. The observations were interpreted on the basis of conversion of [HRh(CO)3L] to [Rh(C(O)R)(CO)3L] as the catalyst resting state for 1-alkene hydroformylation. This is consistent with rate inhibition by CO, since hydrogenolysis of the acyl complex requires loss of CO. Similar experiments were reported for rhodium catalysts with bulky phosphorus diamide ligands [7]. For these systems, mixtures of [HRh(CO)3L] and [HRh(CO)2L2] were observed in the absence of alkene. On addition of 1-octene, rapid scan IR spectroscopy revealed partial conversion to a mixture of isomeric rhodium acyl complexes, confirmed by HP NMR. It has been suggested that bis(equatorial) (ee) coordination of bidentate P-donor ligands in [HRh(CO)2(diphosphine)] (structure 5) can favor high linear/branched product ratios relative to systems which prefer equatorial-axial (ea) chelation (structure 6). For the bulky bis-phosphite 7 shown in Scheme 3.4, HP IR in conjunction H

H P

OC

Rh C O

CO P

Rh

P

5

ee isomer

CO P 6

ea isomer

127

128


t

Bu O

MeO

O O tBu

MeO

O

P

O O

P

PPh2

PPh2 8

O

xantphos 7

S

SO3Na

NaO3S

O

O

PPh2

PPh2

PPh2 9

PPh2 10

thixantphos

sulfoxantphos

O P

O P

P

P

O

Scheme 3.4

O

11

12

PCP xantphos

POP xantphos

Selected bidentate phosphorus ligands.

with HP NMR studies showed the major complex during hydroformylation of 1-octene to be [HRh(CO)2(7)], with the bis(phosphite) ligand occupying two equatorial sites in the trigonal bipyramid [82]. A family of “wide bite angle” diphosphine ligands based upon xantphos (8) has been developed and tested in rhodium catalysed hydroformylation. van Leeuwen and co-workers conducted HP IR measurements on a range of Rh/thixantphos


(9) catalyst systems and were able to demonstrate the existence of dynamic equilibria between ee and ea species. Each isomer has two n(CO) bands and these were assigned with the help of D-labelling of the hydride ligand. Only the n(CO) bands of the ee isomer show a measurable shift on deuteration, due to vibrational coupling of trans Rh–H and C–O stretching modes. HP IR spectra of [HRh(CO)2 (diphosphine)] complexes recorded at 80 hC and 20 bar CO/H2 showed that the ee/ea ratio varied with phosphine basicity, with the ee isomer being favored by more electron-withdrawing phosphine aryl substituents. Similar spectroscopic observations were made for working hydroformylation systems, with no other Rh carbonyl species being detected. However, the selectivity for linear aldehyde formation from 1-octene was essentially unaffected by ligand basicity, suggesting that the chelation mode in the [HRh(CO)2(diphosphine)] is not the key parameter controlling regioselectivity A rate determining step early in the catalytic cycle (i. e. CO dissociation or alkene coordination) was implicated. A HP IR investigation of related phosphacyclic xantphos derivatives (e. g. PCP xantphos (11) in rhodium catalysed hydroformylation showed that these ligands have an enhanced preference for ee coordination relative to the parent xantphos ligand [83]. Dynamic equilibria between ee and ea isomers were also found in HP IR measurements on a range of other xantphos derivatives, with the ee:ea ratio being sensitive to small changes in ligand structure [84]. Similar behavior was found for the rhodium-sulfoxantphos (10) catalysed hydroformylation of 1-octene in an ionic liquid solvent (1-n-butyl-3-methylimidazolium hexafluorophosphate), the ratio of the ee and ea isomers being influenced by both the temperature and the syn-gas pressure [85]. HP IR showed the presence of [HRh(CO)2(dpppts)] under 14 bar CO/H2 in a basic water/methanol mixture, using the water soluble tetrasulfonated dpppts ligand, Ar2P(CH2)3PAr2 (Ar ¼ 3-NaSO3C6H4) [86]. Two n(CO) bands at 1955 and 1990 cm–1 were assigned to the complex with an ea coordinated diphosphine, while a third band at 2034 cm–1 could be due to n(Rh–H) or a n(CO) mode of the ee isomer. By contrast, some chiral phosphine-phosphite ligands, used for enantioselective hydroformylation of styrene, were shown to coordinate exclusively in the ea mode with the phosphine moiety apical, giving only two n(CO) bands [87]. Systems in which dinuclear Rh complexes with bridging thiolate ligands [Rh(m-SR)(COD)]2 were used (along with PPh3 or P(OPh)3) as the catalyst precursors have been probed by HP IR spectroscopy [88]. The results demonstrated that both dinuclear and mononuclear complexes are present under 1-hexene hydroformylation conditions, but that mononuclear hydrides (ee and ea isomers of [HRh(CO)2L2]) are the active catalytic species. Stanley and co-workers used in situ HP IR and HP NMR spectroscopy to investigate the nature of the active species in a highly active dinuclear Rh hydroformylation catalyst, containing the tetradentate phosphine ligand, Et2PCH2CH2P(Ph)CH2P(Ph)CH2CH2PEt2 [36]. Under 6.2 bar CO at 60 hC, complex 13 (Scheme 3.5) is converted into the hexacarbonyl 14. Under 6.2 bar CO/H2 at the same temperature, an equilibrium mixture is generated containing 14 (92 %) and several new hydrido species. On the basis of 31 P NMR data, the structure 15 was proposed for the active catalyst.

129

130


Rh

CO

Rh P

OC

2+

PEt2

Et2P

OC

P

Ph

CO

Ph

13

CO

Rh OC

2+

PEt2

Et2P

OC

CO

Rh P

CO

P

OC

CO Ph

Ph

14

CO/H2

Et2 H P Rh OC

P Ph

O C

2+

H CO Rh

C O

PEt2 P Ph

Formation of active species in Stanley’s dinuclear hydroformylation catalyst [36].

Scheme 3.5

15

Other Reactions of Carbon Monoxide The carbonylation of allylic halides and prop-2-en-1-ol using a Rh/PEt3 catalyst was studied by HP IR (using a CIR cell) by Payne and Cole-Hamilton [89]. The initial reaction of [Rh(CO)(PEt3)2Cl] with allyl chloride in the absence of CO gave the oxidative addition product, [Rh(CO)(PEt3)2Cl2(CH2CH¼CH2)]. On treatment with 40 bar CO at 120 hC, the spectra gave evidence for an acyl product, [Rh(C(O)CH2CH¼CH2)(CO)(PEt3)2Cl2] along with some of the reductive elimination product, but-3-enoyl chloride. HP IR spectroscopy has been applied to study the [Mo(CO)6] catalysed carbonylation of ethylene to propionic acid/anhydride [44]. Like methanol carbonylation, an alkyl halide promoter (usually ethyl iodide) is required, and a halide salt (e. g. (Bu4P)I) is also generally added. Under reaction conditions (e. g. 160 hC, 27 bar CO), a propionic acid solution of (Bu4P)I and the parent hexacarbonyl generated an equilibrium mixture of [Mo(CO)6] and [Mo(CO)5I]– which were observed using an ATR-type HP IR cell. On addition of ethyl iodide, an additional species, assigned as [Mo(CO)4I3]– was formed resulting from oxidation of Mo(0) to Mo(II). In the presence of ethylene, the same Mo complexes were still present, but in different relative proportions. By measuring the changes in the IR spectrum 3.3.1.3


as a function of CO pressure, it was possible to estimate equilibrium constants for Eq. (5). [Mo(CO)6] þ I– w [Mo(CO)5I]– þ CO

(5)

From the temperature variation of the equilibrium constant, thermodynamic parameters for the reaction were also obtained. The extent of formation of [Mo(CO)5I]– was found to be cation-dependent, and while equilibrium constants of 39 and 21 atm L mol–1 were obtained for Bu4Pþ and pyHþ, none of the anionic iodide complex was observed for Naþ. Despite this variation, there seemed to be no correlation between the concentration of [Mo(CO)5I]– and the rate of the catalytic carbonylation reaction. It was proposed that [Mo(CO)6] and [Mo(CO)5I]– are spectator species, with the catalysis being initiated by [Mo(CO)5]. Based on the in situ spectroscopic results and kinetic data, a catalytic mechanism was suggested, involving radicals formed by inner sphere electron transfer between EtI and [Mo(CO)5]. A number of ruthenium-based catalysts for syn-gas reactions have been probed by HP IR spectroscopy. For example, Braca and co-workers observed the presence of [Ru(CO)3I3]–, [HRu3(CO)11]– and [HRu(CO)4]– in various relative amounts during the reactions of alkenes and alcohols with CO/H2 [90]. The hydrido ruthenium species were found to be active in alkene hydroformylation and hydrogenation of the resulting aldehydes, but were inactive for alcohol carbonylation. By contrast, [Ru(CO)3I3]– was active in the carbonylation of alcohols, glycols, ethers and esters and in the hydrogenation of alkenes and oxygenates. The complexes [Ru(CO)3I3]– and [HRu3(CO)11]– were also observed by Dombek in an HP IR study of the [Ru3(CO)12]/iodide catalysed hydrogenation of CO to methanol, ethanol ethylene glycol and their derivatives [91]. On the basis that neither observed complex is active in the absence of the other, a catalytic mechanism involving interaction between hydrido Ru(0) and iodo Ru(II) species was proposed. Addition of rhodium to the ruthenium catalyst did not affect the total activity but increased the selectivity towards ethylene glycol [92]. The same Ru species, [Ru(CO)3I3]– and [HRu3(CO)11]– were observed by HP IR and the Rh component existed as [Rh(CO)2I2]–. It was suggested that the change in selectivity arose from hydride transfer from [HRu(CO)4]– to a Rh(III) iodocarbonyl complex such as [Rh(CO)2I4]–, to give a Rh-formyl intermediate. Ruthenium and rhodium catalysts for CO hydrogenation have also been studied by Whyman using HP IR spectroscopy [93]. These systems used Et3N rather than an iodide promoter. For a Ru-only system (either with or without Et3N) the major species observed was [Ru(CO)3(OAc)]2 along with traces of [Ru(CO)]5 and [Ru3(CO)12]. For a Rh-only catalyst, the main species was [Rh6(CO)16] but [Rh6(CO)16X] (X ¼ H, OAc) was also formed on addition of Et3N. For a mixed metal catalyst (Ru:Rh ¼ 10:1), the principal band at 2040 cm–1 was assigned to [HRu(CO)n(OAc)] (n ¼ 3 or 4). In the ruthenium catalysed carbonylation of piperidine (60 hC, 10 bar CO) the catalyst precursor, [Ru3(CO)12] was found to be converted mainly to [Ru(CO)5], although IR absorptions due to other minor species were also observed [94]. A catalytic mechanism was tentatively proposed, which involved [RuCO)4] as the active

131

132


species, undergoing oxidative addition of the N–H bond of piperidine followed by carbonylation and reductive elimination of the product N-formyl amine. In the [Ru(CO)3(dppe)] catalysed carbonylation of para-nitrotoluene (Eq. (6)) HP IR spectroscopy indicated conversion of the Ru(0) complex into an oxidised species with n(CO) bands at higher frequency [41]. A mechanism involving single-electrontransfer from the nitroaromatic to the Ru complex was proposed. ArNO2 þ 3CO þ MeOH p ArNHC(O)OMe þ 2CO2

(6)

An HP IR study of the platinum catalysed carbonylation of methanol to methyl formate, revealed that the catalyst precursor, cis-[Pt(PEt3)2Cl2] is converted into cis[Pt(PEt3)2(CO)2] along with a cluster species, [Pt3(PEt3)3(CO)n] (n ¼ 3 or 4) [95]. A mechanism involving oxidative addition of methanol to Pt(0) followed by CO insertion into the Pt–OMe bond was suggested. Another catalytic carbonylation reaction of considerable recent interest is the alternating co-polymerization of CO and alkenes to give polyketones, normally achieved with Pd(II) catalysts and chelating P,P or N,N ligands. Despite the large number of mechanistic studies, HP IR spectroscopy has not been widely used to probe this reaction. In a study by Luo et al., the [Pd(dppp)(OTs)2] precursor generated species with bands at 1616 and 1638 cm–1 in 2-ethylhexanol under CO (4 bar) and C2H4 (40 bar) at 85 hC [96]. These absorptions may be attributed to chelate complexes of the type 16 or 17, which are thought to play an important role in governing the perfectly alternating CO and alkene insertion steps. Strong bands in the region 1670–1700 cm–1 attributable to carbonyl groups in the polyketone chain were also observed. A study of solvent-free CO/C2H4 copolymerization using a microcrystalline [Pd(dppp)Me(OTf)] catalyst used polarisation modulation reflection absorption infrared spectroscopy (PM-RAIRS), although pressures above 1 bar were not employed [97]. An absorption at 1625 cm–1 was attributed to Pd-chelated carbonyl groups in structures of type 16 or 17, and a band at 1690 cm–1 arose from the Pd-acyl moiety in 17. Polymer growth could be followed by the appearance of a band at 1705 cm–1. polymer Ph2 P

O

polymer

Ph2 P

O Pd

Pd P Ph2

P Ph2 16

17

O

ATR-HP IR spectroscopy has also been used to follow the cobalt-catalysed carbonylation of epoxides to give lactones or polyesters [46]. Addition of excess propylene oxide to [HCo(CO)4] (generated in situ by protonation of [Co(CO)4]–) under 20 bar CO was found to give an acyl complex, [Co(C(O)CH2CH(OH)Me)(CO)4]. Depending


upon the reaction conditions employed, subsequent reaction was found to lead to formation of b-butyrolactone or polyesters of the form H{OC(Me)CH2C(O)}nOH, arising from alternating co-polymerization of the epoxide and CO. 3.3.2

Kinetic and Mechanistic Studies of Stoichiometric Reaction Steps Migratory CO Insertion Reactions of Metal Alkyls Migration of an alkyl ligand onto metal-coordinated CO to give the acyl ligand is a key C–C bond forming step of many catalytic carbonylation processes. The reverse reaction can also play a role, for example in the decarbonylation of aldehydes. In this section, the use of HP IR to probe the mechanism of migratory CO insertion reactions will be exemplified by recent model studies on iridium catalysed methanol carbonylation. The identification of reactive intermediates in migratory insertion using HP IR coupled with flash photolysis will also be discussed (see below). 3.3.2.1

Model studies of iridium catalysed methanol carbonylation [37–39]

As described above (Section 3.3.1.1), in situ HP IR measurements under catalytic conditions identified the anion [MeIr(CO)2I3]– as the catalyst resting state in the Cativa process. The rate controlling step in the catalytic cycle was proposed to be carbonylation of [MeIr(CO)2I3]– (Eq. (7)). [MeIr(CO)2I3]– þ CO p [Ir(C(O)Me)(CO)2I3]–

(7)

In weakly polar aprotic solvents, the reaction proceeds slowly, even at elevated temperatures (i80 hC). Kinetic studies were carried out using a CIR-HP IR cell and a typical series of spectra is shown in Figure 3.12 from an experiment carried out in PhCl. The two n(CO) bands of [MeIr(CO)2I3]– decay and are replaced by new absorptions due to [Ir(C(O)Me)(CO)2I3]– as well as a small amount of [Ir(CO)2I2]–, formed by reductive elimination of MeI. Addition of methanol (and other protic solvents) was found to facilitate carbonylation of [MeIr(CO)2I3]– at milder temperatures (ca. 30–50 hC) [37, 39]. A typical set of IR spectra for the reaction in PhCl–MeOH is shown in Figure 3.13. It is noticeable that the two terminal n(CO) bands of [Ir(C(O)Me)(CO)2I3]– have different relative intensities in the two solvent systems. In the presence of methanol (Figure 3.13), each band has similar intensity, consistent with a cis dicarbonyl geometry, analogous to that of the precursor, [MeIr(CO)2I3]– [98]. At higher temperature in neat PhCl, however, the high and low frequency n(CO) bands of [Ir(C(O)Me)(CO)2I3]– have an intensity ratio of ca. 1:2 (Figure 3.12). This is explained by the formation of both cis and trans dicarbonyl isomers of [Ir(C(O)Me)(CO)2I3]– (shown in Scheme 3.6), with the single strong n(CO) band of the trans isomer being coincident with the low frequency absorption of the cis isomer. The effect of solvent on the isomer ratio for [Ir(C(O)Me)(CO)2I3]– was confirmed by high pressure 13C NMR experiments and an X-ray crystal structure was obtained for the fac,cis isomer [39].

133

134


Figure 3.12 Series of in situ HP IR spectra for the reaction of [MeIr(CO)2I3]– with CO (5.5 bar) in PhCl at 93 hC (adapted from Ref. [39] by permission of the American Chemical Society).

Figure 3.13 Series of in situ HP IR spectra for the reaction of [MeIr(CO)2I3]– with CO (5.5 bar) in PhCl–MeOH (3:1 v/v) at 33 hC (adapted from Ref. [39] by permission of the American Chemical Society).


– Me I CO Ir I CO I

Me

C

– I–

Me I CO Ir I CO

Me I CO Ir I CO C O

CO

O

Me

OC I Ir I CO CO I–

Me

C

O

OC I Ir I CO I

–

C

O

I CO Ir I CO Me

C

O

OC I Ir OC CO I

CO

I–

Me

C

O

I CO Ir I CO I

–

Mechanism for carbonylation of [MeIr(CO)2I3]– (adapted from Ref. [39] by permission of the American Chemical Society). Scheme 3.6

Pseudo-first-order rate constants for carbonylation of [MeIr(CO)2I3]– were obtained from the exponential decay of its high frequency n(CO) band. In PhCl, the reaction rate was found to be independent of CO pressure above a threshold of ca. 3.5 bar. Variable temperature kinetic data (80–122 hC) gave activation parameters DH‡ 152 (e6) kJ mol–1 and DS‡ 82 (e17) J mol–1 K–1. The acceleration on addition of methanol is dramatic (e. g. by an estimated factor of 104 at 33 hC for 1 % MeOH) and the activation parameters (DH‡ 33 (e2) kJ mol–1 and DS‡ –197 (e8) J mol–1 K–1 at 25 % MeOH) are very different. Added iodide salts cause substantial inhibition and the results are interpreted in terms of the mechanism shown in Scheme 3.6 where the alcohol aids dissociation of iodide from [MeIr(CO)2I3]–. This enables coordination of CO to give the tricarbonyl, [MeIr(CO)3I2] which undergoes more facile methyl migration (see below). The behavior of the model reaction closely resembles the kinetics of the catalytic carbonylation system. Similar promotion by methanol has also been observed by HP IR for carbonylation of [MeIr(CO)2Cl3]– [99]. In the same study it was reported that [MeIr(CO)2Cl3]– reductively eliminates MeCl ca. 30 times slower than elimination of MeI from [MeIr(CO)2I3]– (at 93–132 hC in PhCl). Consistent with the mechanism shown in Scheme 3.6, additives capable of acting as iodide acceptors were found to accelerate carbonylation of [MeIr(CO)2I3]– [39]. Neutral Ru iodocarbonyls, such as [Ru(CO)3I2]2 and [Ru(CO)4I2], give rate enhancements by factors of 15–20 for a Ru:Ir ratio of 1:13 at 93 hC, the effect being approximately proportional to [Ru]. Simple Group 12 and 13 iodides (e. g. ZnI2, GaI3 and InI3) gave comparable rate enhancements. IR experiments conducted in the absence of CO demonstrated stoichiometric iodide transfer from [MeIr(CO)2I3]– to the neutral promoter (giving, for example, [Ru(CO)3I3]– or [InI4]–) and the neutral dimer [MeIr(CO)2I2]2 was isolated. The model studies showed that promoters

135

136


which are effective in the catalytic system (see Section 3.3.1.1) can accelerate stoichiometric carbonylation of [MeIr(CO)2I3]– and confirmed that their role is to act as iodide acceptors. Control of the iodide concentration is crucial to achieve high activity for an Ir-based catalyst. A key species in the proposed catalytic mechanism is the neutral tricarbonyl, [MeIr(CO)3I2], which was never observed in Forster’s studies. Ghaffar et al. used HP IR and HP NMR spectroscopy to show that [MeIr(CO)3I2] is generated by reaction of the dimer, [MeIr(CO)2I2]2, with CO [38, 39]. Less than 10 % conversion occurred at 1 bar CO, but essentially complete conversion was achieved using 27 bar CO pressure. The spectroscopic data, including 13CO isotoipic labelling experiments, indicated a fac,cis geometry for [MeIr(CO)3I2] and showed that the incoming CO ligand coordinates reversibly in the site trans to methyl, without scrambling into the cis positions. The same tricarbonyl was also observed by HP NMR on reaction of [MeIr(CO)2I3]– with [Ru(CO)2I4] under 6 bar CO, the ruthenium species being converted into [Ru(CO)3I3]– via an iodide bridged intermediate [100]. HP IR spectroscopy was used to monitor the kinetics of carbonylation of [MeIr(CO)3I2] into [Ir(C(O)Me)(CO)3I2] [38, 39]. At 85 hC in PhCl, [MeIr(CO)3I2] is carbonylated more than 700 times faster than the anion, [MeIr(CO)2I3]–, representing a lowering of DG‡ by 20 kJ mol–1. Activation parameters (DH‡ 89 (e3) kJ mol–1 and DS‡ –36 (e8) J mol–1 K–1), calculated from variable temperature measurements, enabled rates to be extrapolated at higher temperatures, such that at 180 hC, [MeIr(CO)3I2] is predicted to react an order of magnitude faster than [MeIr(CO)2I3]–. The proposed mechanism involving methyl migration in a neutral complex is therefore supported by the model stoichiometric studies. The faster methyl migration in [MeIr(CO)3I2] is ascribed to the coordination of an additional p acceptor CO ligand, which reduces backdonation to the other carbonyl ligands, thus increasing their electrophilicity. Carbonylation of group 9 metal complexes containing phosphine ligands

Alkyl migration to CO in Rh(alkyl)(carbonyl) complexes is often fast, even in the absence of added CO. The reaction can, however be retarded by strong donor co-ligands, an example being the PEt3 system studied by Cole-Hamilton and co-workers [101]. The methyl complex [MeRh(CO)(PEt3)2I2] is stable and isolable and the kinetics of its reaction with CO to give [Rh(C(O)Me)(CO)(PEt3)2I2] were followed in a CIR cell. At 22.8 hC in CH2Cl2, pseudo-first-order rate constants of 7.1 q 10 –5 and 1.1 q 10 –4 –1 were measured at 27 and 40 bar CO respectively, suggesting a first-order dependence on pCO. Heating the acetyl product to 70 hC under 40 bar CO gave rise to formation of some acetyl iodide, identified by its n(CO) band in the HP IR spectrum. In the same study, oxidative addition of methyl iodide was found to be ca. 57 times faster for [Rh(CO)(PEt3)2I] than for [Rh(CO)2I2]– (at 25 hC), ascribable to the greater nucleophilicity bestowed by the phosphine ligands. Despite oxidative addition being faster and migratory insertion being slower than in the [Rh(CO)2I2]– system, MeI addition is thought to remain the rate determining step in the PEt3 modified catalytic cycle for methanol carbonylation.


A CIR cell was also used by Gonsalvi et al. to monitor carbonylation of the Ir(iii) methyl complexes, [MeIr(CO)(L-L)I2] (where L-L ¼ dppe or dppms) [56]. Under 10 bar CO, in PhCl containing 1 % methanol, a rate constant of 7.3 q 10 –4 s–1 was measured for the dppms complex. Under the same conditions, by contrast, the dppe complex was unreactive, reflecting the different rates of methyl migration found for the analogous Rh complexes. Both steric and electronic effects can account for these ligand effects. Flash-photochemical studies

Ford and co-workers have applied HP IR spectroscopy in conjunction with flashphotochemical methods to investigate reactive intermediates in alkyl migration reactions [22, 102]. Transient, coordinatively unsaturated transition metal complexes are not normally amenable to direct study in a working catalytic system, but can sometimes be generated photochemically and probed by rapid time-resolved spectroscopy, in this case time-resolved infrared (TR IR). In catalytic carbonylation reactions (e. g. hydroformylation, methanol carbonylation), a key step in the catalytic cycle is generally alkyl migration onto a carbonyl. The primary product of such a migration will be coordinatively unsaturated with a vacant site created by the combination of alkyl and CO ligands to give an acyl. Under normal circumstances, the vacant site will be rapidly filled by coordination of CO or another available ligand (e. g. solvent). Ford’s approach to studying the unsaturated intermediate was to use stable 18-electron acyl species (e. g. [Mn(C(O)R)(CO)5] or [Co(C(O)R)(CO)4]) as the precursor for the flash photolysis experiment. In principle, photochemical ejection of a CO ligand from such complexes will generate formally 16-electron intermediates identical (or closely related) to the species formed in the forward alkyl migration reaction. Flash photolysis of [M(C(O)Me)(CO)n] (M ¼ Mn, Co; n ¼ 5, 4) was shown to generate transient species of the general formula [M(C(O)Me)(CO)n-1]. Truly coordinatively unsaturated complexes are unlikely to be observed on the ns–ms timescale of these experiments, and the vacant coordination site generated by CO loss is thought to be stabilised in some manner. Alternative structures suggested (Scheme 3.7) involve h2 -acyl coordination (18), coordination of solvent (19) and a b-agostic C–H interaction with the metal center (20). The intermediate observed is thought to be an equilibrium ensemble of all of these species, the proportions depending on the metal and the solvent. For Co, the evidence supports an h2 -acyl species as the dominant structure, since the rate of trapping by CO to give back [Co(C(O)Me)(CO)4] is Me C (OC)n-1M 18

O

Me

O C

O

(OC)n-1M 19

C sol

(OC)n-1M

CH2 H

20

Alternative structures for “coordinatively unsaturated” species generated by CO loss from [M(C(O)Me)(CO)n]. Scheme 3.7

137

138


hν –CO [M(C(O)Me)(CO)n]

kM [M(C(O)Me)(CO)n-1]

[MeM(CO)n]

kCO[CO] Scheme 3.8

Photochemistry of metal acetyl complexes.

insensitive to the solvent coordinating strength. For Mn, however, in THF, the solvent coordinated species [Mn(C(O)Me)(CO)4(THF)] dominates. For both metals, the bimolecular reaction with CO (rate constant kCO, Scheme 3.8) to regenerate the saturated 18 electron precursor competes with methyl migration (kM) from CO to metal (or migratory deinsertion) to give the corresponding alkyl complex, [MeM(CO)n]. Second-order kinetics for the reaction with CO were confirmed by measuring the rate as a function of CO pressure. For manganese, kCO varies from ca. 102 –104 M–1 s–1 depending on the solvent (with higher values in weakly coordinating media) and kM varies from 1–50 s–1 (at 23 hC). For cobalt, values of both rate constants were considerably larger (kCO ca. 107 M–1 s–1 and kM 105 –106 s–1), consistent with cobalt being the more favored carbonylation catalyst. Analogous experiments for the iron acetyl complex, [CpFe(CO)2(C(O)Me)], were also reported [21], in that case the photoproduct was assigned a structure with a b-agostic interaction between the Fe center and the acetyl methyl group. At low CO pressures (J1 atm) trapping of [CpFe(CO)(C(O)Me)] by CO is not competitive with methyl migration (kM 6.3 q 104 s–1, 25 hC) to give [CpFe(CO)2Me]. However, the apparatus used by Ford gave access to higher CO pressures (J28 bar) and a kCO of 2.6 q 105 M–1 s–1 for the bimolecular reaction of [CpFe(CO)(C(O)Me)] with CO was measured.

Substitution and Exchange Reactions of CO Ligands Related to catalytic methanol carbonylation reactions, substitution of iodide by CO in [M(CO)2I2]– (M ¼ Rh, Ir) has been studied by HP IR spectroscopy [103]. Under CO pressure, in CHCl3, both complexes were found to generate the corresponding neutral tricarbonyl, [M(CO)3I]. The Rh species had already been reported by Morris and Tinker in the reaction of [Rh(CO)2I]2 with CO [104] and the Ir species is observed under certain conditions during Ir-catalysed methanol carbonylation [39, 60]. While, for M ¼ Ir, 35 % conversion was attained at 150 bar CO, for Rh less than 4 % conversion occurred. 3.3.2.2

[M(CO)2I2]– þ CO w [M(CO)3I] þ I–

(8)

Equilibrium constants for Eq. (8) were determined as KIr ¼ 1.8 q 10 –3 and KRh ¼ 4 q 10 –5 from the HP IR spectroscopic data, indicating a higher affinity for CO (relative to I–) for Ir(I) than Rh(I). The same study also reported the HP IR observation of two isomers of [H2Ir(CO)2I2]–, from the oxidative addition of H2 to [Ir(CO)2I2]–. In combination with HP NMR data, it was found that the initially


formed all-cis isomer rearranged to a more stable isomer with trans carbonyl ligands. No similar hydride products were observed for the [Rh(CO)2I2]– system, although these have been detected by the sensitive parahydrogen NMR method [105]. van Leeuwen and co-workers [83, 84] have used HP IR spectroscopy to follow CO ligand exchange in complexes of the type [HRh(CO)2(diphosphine)], which are observed as the major Rh species during hydroformylation catalysis. Dissociation of CO from the catalyst resting state is believed to be a key step, prior to alkene substrate coordination, and the influence of ligand structure on CO dissociation is therefore of clear importance for catalytic activity. The relative lability of the CO ligands in these species necessitated the use of rapid-scan HP IR experiments, using the flow transmission cell described in Section 3.2.1.1. The kinetics of CO ligand exchange were monitored by exposing the 13CO labelled complex to a large excess of 12CO. The n(CO) bands of the labelled and unlabelled complexes are separated sufficiently to allow individual observation of each species by IR spectroscopy. Representative spectra are illustrated in Chapter 6 (Figure 6.14) and show that decay of the n(13CO) bands of [HRh(13CO)2(diphosphine)] is mirrored by the appearance of n(12CO) absorptions of [HRh(12CO)2(diphosphine)] at higher frequency. The presumed intermediate mixed isotopomer [HRh(12CO)(13CO)2(diphosphine)] is not observed. Simple first-order kinetics, independent of pCO, are consistent with a dissociative mechanism for CO exchange. The individual CO dissociation rates for the ee and ea isomers could not be distinguished due to the dynamic equilibrium between them, which occurs on a timescale faster than CO dissociation. Good time resolution was obtained with spectra being recorded at ca. 0.2 s intervals. Firstorder rate constants for a range of ligands from the xantphos family were found to be of the order of 200 h–1 at 40 hC in cyclohexane, corresponding to half-lives of the order of 10–15 s. This is orders of magnitude faster than the hydroformylation rate and no correlation with ligand bite-angle was found. The CO exchange rate was found to be even faster (ca. 1200 h–1, half-life 2 s) for the ligand POP-xantphos (12), which exhibited very high activity and selectivity for hydroformylation of internal octenes to linear nonanal. Cationic Group 11 carbonyl complexes have been implicated in catalytic carbonylation reactions of alkenes, arenes, alcohols, saturated hydrocarbons and aldehydes under acidic conditions [106]. While the mono- and di-carbonyls [M(CO)]þ and [M(CO)2]þ (M ¼ Cu, Ag, Au) can be formed at atmospheric pressure of CO, only Cu(I) forms a tricarbonyl cation under such conditions [107]. Strauss and co-workers reported the observation of [Ag(CO)3]þ when a Fluorolube mull of Ag[Nb(OTeF5)6] was subjected to 13 bar CO in an HP IR transmission cell [108]. The gold analog, [Au(CO)3]þ is reported to be formed (with an [Sb2F11]– counterion) only under 100 bar CO [107]. The n(CO) bands of these species occur significantly above the stretching frequency of free CO (2143 cm–1); for example the silver and gold tricarbonyl cations have IR absorptions at 2192 and 2212 cm–1 respectively. This is taken to indicate that there is little or no metal p CO p-backdonation in these species, which are termed non-classical metal carbonyls [109].

139

140


Exchange between Rh–D and H2 Deuterium labelling can, in principle, provide mechanistic information on hydroformylation reactions, but complications may arise if scrambling of the D-label between a metal hydride and D2 is fast. The isotopic exchange reaction shown in Eq. (9) was probed by rapid scan HP IR spectroscopy (1.3 scans s–1) where 21 is the bidentate pyrrolyl-based phosphorus amidite ligand illustrated [8]. 3.3.2.3

[DRh(CO)2(21)] þ H2 w [HRh(CO)2(21)] þ HD

N

P N

O

O

P

(9)

N

N 21

The bidentate ligand coordinates in an ee mode such that a CO ligand is trans to the Rh–H/D bond, giving rise to significant vibrational coupling between Rh–H and C–O stretches. The n(CO) frequencies are therefore sensitive to H/D substitution, allowing the exchange process to be monitored by IR spectroscopy. A firstorder rate constant of 1140 h–1 was measured at 80 hC in cyclohexane, which, although fast, is still an order of magnitude slower than the hydroformylation rate for this catalyst. Therefore, the RhH/D2 exchange reaction is only significant at high substrate conversion.

Hydrogenolysis of M–C Bonds The aldehyde formation step in catalytic hydroformylation is thought to involve hydrogenolysis of a metal acyl complex (although an alternative pathway involving reaction of the metal acyl with a metal hydride is implicated in some cases). Garland and Pino used HP IR spectroscopy to measure the kinetics of hydrogenolysis of [Rh(C(O)R)(CO)4] (R ¼ CH2CH2CMe3) [110]. The experiments involved initial conversion of [Rh4(CO)12] into [Rh(C(O)R)(CO)4] by reaction with 3,3-dimethylbut-1-ene and syn-gas. Rate measurements at different reactant concentrations gave a rate law for this conversion of rate ¼ k[Rh4(CO)12]1[CO]1.8[H2]0.7[alkene]0.1 and activation parameters DH‡ 74e12 kJ mol–1 and DS‡ –19e42 J mol–1 K–1. These data suggested a pre-equilibrium between [Rh4(CO)12] and an intermediate [Rh4(CO)14], followed by rate determining activation of H2 by [Rh4(CO)14]. Formation of aldehyde product was then found to proceed with a rate law, rate ¼ k[Rh(C(O)R)(CO)4]1[CO]–1.1[H2]1[alkene]0.1 and activation parameters DH‡ 49e4 kJ mol–1 and DS‡ 121e14 J mol–1 K–1. Similar measurements were also reported for analogous acyl complexes derived from insertion of cyclohexene [111] 3.3.2.4


O

O

R CO

OC

CO

H H2, CO

– CO OC

Rh C O

R C

C

Rh

CO

CO OC

C O

Rh C O

CO

RCHO Scheme 3.9

Hydrogenolysis of [Rh(C(O)R)(CO)4].

or styrene [112], the linear and branched isomers being distinguished in the latter case due to their different acyl n(C¼O) values. In each case, the kinetic data are consistent with rate determining dissociation of CO from [Rh(C(O)R)(CO)4] followed by reaction of the 16 electron [Rh(C(O)R)(CO)3] with H2 and release of the aldehyde (Scheme 3.9). The kinetics of hydrogenolysis of a metal-alkyl have been monitored by HP IR spectroscopy for [MeIr(CO)2I3]–, the resting state in the cycle for iridium catalysed methanol carbonylation [113]. On treatment with H2 at elevated temperatures, the n(CO) bands of [MeIr(CO)2I3]– decayed and were replaced by new n(CO) bands at slightly higher frequency and a n(Ir–H) absorption, corresponding to Eq. (10). [MeIr(CO)2I3]– þ H2 p [HIr(CO)2I3]– þ CH4

(10)

This represents one pathway to the formation of methane, a known by-product in iridium catalysed methanol carbonylation. The hydrogenolysis reaction was severely retarded by the presence of excess CO, indicating a mechanism involving initial dissociation of CO from [MeIr(CO)2I3]– , prior to activation of H2. The mechanism therefore resembles that for hydrogenolysis of Rh acetyl complexes in hydroformylation.

Mechanistic Studies in Polymer Matrices Matrix isolation has been used for many years for the spectroscopic study of unstable metal carbonyl species. The technique generally involves trapping an unstable molecular fragment in an inert matrix such as a frozen noble gas, an organic glass or a polymer. In an extension of these techniques, Poliakoff, George and coworkers have developed a high pressure, low temperature copper cell for infrared spectroscopy of species embedded in polymer films. In a typical experiment, a polyethylene film is impregnated with the metal complex by immersion of the film in a solution of the complex. The polymer film is then clipped into the cell, which is pressurised with a gas (e. g. H2 or CO), in order to initiate reactions. The technique was used initially to study photochemical reactions between Group 6 metal carbonyls with N2 or H2, providing evidence for new N2 and H2 complexes and establishing the parameters which influence the penetration of gases into polyethylene films [114]. 3.3.2.5

141

142


The approach has also been used to study catalytic hydrogenation and hydroformylation processes. The iron catalysed hydrogenation of dimethylfumarate (DF) was modelled by studying the photochemistry of [Fe(CO)4(h2 -DF)] in a polyethylene film under H2 pressure [115]. IR spectroscopy indicated the initial photochemical loss of CO at –123 hC to give [Fe(CO)3(h4 -DF)]. In the presence of H2 (60 bar), warming to –13 hC resulted in the formation of a species assigned as [Fe(CO)3(h2 DF)(h2 -H2)]. At room temperature, decomposition of this species and formation of the hydrogenated product, dimethyl succinate was observed, thus demonstrating part of the catalytic cycle. In the same paper, hydrogenation of norbornadiene (nbd) by Group 6 metal carbonyls was also investigated. On UV photolysis of [M(CO)4(h4 -nbd)] and excess nbd in a polyethylene film under D2, IR spectroscopy showed n(C-D) bands for the deuterated products, norbornene and nortricyclene, the product ratio being dependent on the metal (Cr or Mo). For the molybdenum system, n(CO) bands due to fac and mer isomers of [M(CO)3(h4 -nbd)(h2 -nbd)] were observed, corresponding to the catalyst resting state seen in fluid solution [15]. Reactions of the rhodium b-diketonate complex [Rh(CO)2(acac)] with alkenes and hydrogen in polyethylene films have been studied [116]. In the absence of H2, thermal ligand substitution occurs to give [Rh(CO)(h2 -alkene)(acac)]. With 96 bar H2, catalytic alkene hydrogenation occurs within the polymer matrix. A catalytic cycle involving H2 oxidative addition to [Rh(CO)(h2 -alkene)(acac)] was suggested. Hydroformylation mechanisms were also probed using [Rh(CO)2(acac)] and [Rh(CO)(PPh3)(acac)] precursors in polyethylene films [117]. For the un-modified catalyst, hydroformylation of 1-octene, 1-butene and propene was achieved at 25 hC and 134 bar syn-gas (confirmed by GC analysis of products extracted from the film). The presence of CO gas in the cell space surrounding the film gives a broad IR absorption (2030–2230 cm–1) which hinders observation of metal carbonyl species with bands in this region. This was overcome by venting the syn-gas at intervals to record the IR spectrum. For each 1-alkene substrate, the corresponding acyl complex [Rh(C(O)R)(CO)4] was observed (probably as a mixture of linear and branched isomers), as found in homogeneous solution by Garland [71]. The behavior for ethylene was different, with a new species, assigned as [Rh(C(O)Et)(CO)3(h2 -C2H4)], being observed at high partial pressure of C2H4. This acyl-ethylene complex was also detected by Garland [72]. For the PPh3 modified catalyst and the same 1-alkenes, Rh acyl complexes were again observed, assigned as [Rh(C(O)R)(CO)3(PPh3)] with acyl and PPh3 ligands occupying axial positions of a trigonal bipyramid. On venting the syn-gas, the bands of the acyl complexes decayed over 1–2 h.

Noble Gas and H2 Complexes As described earlier, high pressure cells have been developed for the use of noble gases as solvents for IR spectroscopic studies, either at low temperature, or at ambient temperature where the supercritical phase exists. A particular focus of this work was the study of reactive complexes containing coordinated noble gas atoms or molecular H2, the latter being particularly relevant to hydrogenation reactions. 3.3.2.6


UV photolysis of [Cr(CO)6] in liquid Xe at –98 hC was found to give [Cr(CO)5Xe] [11]. Weiller investigated the kinetics of the reverse reactions of [M(CO)5L] (M ¼ Cr, W; L ¼ Kr, Xe) with CO in the liquid noble gases, and from the temperature dependent rates, estimated a W–Xe bond energy of 8.4 e 0.2 kcal mol–1 [118]. The same complexes have also been detected at room temperature. For example, flash photolysis of Group 6 hexacarbonyls has been employed to generate [W(CO)5Ar], [M(CO)5Kr] and [M(CO)5Xe] (M ¼ Cr, Mo, W) in scAr, scKr and scXe respectively [24]. These noble gas complexes have lifetimes of the order of microseconds at 24 hC. Kinetic data from TR IR measurements showed that the order of reactivity toward CO is Cr z Mo i W, and for a particular metal the Kr complex is more reactive than the Xe complex. The reactivity of [M(CO)5Xe] was found to be similar to that of [M(CO)5(CO2)], formed in the same way in scCO2. Flash photolysis of [Cr(h6 -C6H6)(CO)3] in scXe (100 bar) at 20 hC generates [Cr(h6 -C6H6)(CO)2Xe] which is slightly more reactive than [Cr(CO)5Xe] [25]. Rhenium complexes [CpRe(CO)2Kr] and [CpRe(CO)2Xe] were found to be less reactive than the corresponding [W(CO)5L] species [119]. TR IR measurements in liquid xenon and krypton identified [Cp*Rh(CO)L] (L ¼ Kr, Xe) as transient species which react rapidly with CO, even at low temperature, to regenerate the precursor, [Cp*Rh(CO)2] [16, 17]. These Rh complexes have been the subject of intense interest due to their propensity for C–H activation of alkanes (Section 3.3.2.7). The noble gas complexes [CpRh(CO)L] and [Cp*Rh(CO)L] (L ¼ Kr, Xe) have also been studied in supercritical fluid solution at room temperature [120]. For both Kr and Xe, the Cp* complex is ca. 20–30 times more reactive towards CO than the Cp analogue. Kinetic data and activation parameters indicated an associative mechanism for substitution of Xe by CO, in contrast to Group 7 complexes, [CpM(CO)2Xe] for which evidence supports a dissociative mechanism. Photolysis of Group 6 hexacarbonyls in LNGs doped with H2 gave rise to unstable complexes of the type [M(CO)5(h2 -H2)], and in the case of M ¼ Cr and W, cis-[M(CO)4(h2 -H2)2] complexes were also observed [13]. As well as n(CO) bands in the IR spectrum, weak absorptions due to n(H–H) can be detected in the region 2700–3100 cm–1, aided by the long pathlengths attainable using IR transparent LNGs as solvents. These dihydrogen complexes are more reactive analogs of the stable complexes first isolated by Kubas [121], and have clear relevance to catalytic hydrogenation reactions. The thermal stability of [M(CO)5(h2 -H2)] is in the order Mo II Cr I W. [Cr(CO)5(h2 -H2)] reacts thermally with D2 to form [Cr(CO)5(h2 -D2)] but not [Cr(CO)5(h2 -HD)], although the HD complex can be formed by a proposed intramolecular H/D exchange in [Cr(CO)4(h2 -H2)(h2 -D2)]. Photolysis of [Fe(CO)2(NO)2] and [Co(CO)3(NO)] in liquid xenon also gave h2 -H2 complexes by substitution of a CO ligand [14]. The photocatalytic hydrogenation of alkenes and dienes by Group 6 metal carbonyls has been investigated in LNG solvents [15]. Photolysis of trans[M(CO)4(C2H4)2] (M ¼ Cr, Mo, W) in liquid xenon doped with H2 leads to formation of mer-[M(CO)3(C2H4)2(h2 -H2)] and cis-[M(CO)4(C2H4)(h2 -H2)]. The h2 -H2 complexes for M ¼ Cr and Mo are much less stable than those for M ¼ W. The evidence supported h2 -coordination of H2 rather than oxidative addition to give dihy-

143

144


O C CO Mo

H

C O

H 23

22 H H

CO

Cr CO C O 24

25

Scheme 3.10 Formation of nbd hydrogenation products from h2 -H2 complexes.

drides. Similar experiments with [Mo(CO)4(h4 -nbd)] (nbd ¼ norbornadiene) gave fac and mer-[Mo(CO)3(h4 -nbd)(h2 -H2)] as well as [Mo(CO)4(h2 -nbd)(h2 -H2)] as products. Whereas fac-[Mo(CO)3(h4 -nbd)(h2 -H2)] decays by loss of H2, the mer isomer (22, Scheme 3.10) undergoes intramolecular addition of H2 to one of the double bonds of the nbd ligand, leading to the hydrogenation product, norbornene (23). For the analogous Cr system it is proposed that intramolecular hydrogen transfer in fac-[Cr(CO)3(h4 -nbd)(h2 -H2)] (24) gives the alternate hydrogenation product, nortricyclene (25).

Alkane Complexes and C–H Activation Reactions Activation of C–H bonds of hydrocarbons is an important goal in organometallic chemistry and homogeneous catalysis, and has been the subject of numerous studies in recent years. One of the most significant results was the discovery that photochemically generated Rh and Ir complexes, [(h5 -C5R5)ML] (L ¼ CO, PMe3) can activate the C–H bonds of saturated hydrocarbons, including methane [122]. HP IR spectroscopy has played an important role in probing the mechanistic details of these reactions, particularly in the case of L ¼ CO, for obvious reasons. In 1989 Weiller et al. reported the results of low temperature UV flash photolysis of [Cp*Rh(CO)2] in liquid krypton doped with cyclohexane [16, 19]. TR IR measurements demonstrated that the transient complex [Cp*Rh(CO)Kr] reacts with cyclohexane to form the C–H activation product, [Cp*Rh(CO)(C6H11)H]. The pseudofirst-order rate constant increased with [cyclohexane] in a non-linear manner, tending to a limiting value at high [cyclohexane], but giving a substantially smaller limiting rate constant when C6H12 was replaced by C6D12. A mechanism consistent with these data was proposed where an initial pre-equilibrium substitution of Kr by alkane in the Rh coordination sphere is followed by the C–H bond cleavage (Scheme 3.11). 3.3.2.7


R–H Rh OC Scheme 3.11

Rh

Rh Kr

Kr

OC

(RH)

OC

H R

C–H activation mechanism in liquid Kr.

This mechanism clearly implicated alkane complexes as precursors to C–H activation but the IR absorptions of [Cp*Rh(CO)Kr] and [Cp*Rh(CO)(C6H12)] were not resolved and were presumed to be coincident. The temperature dependent data gave values of DH‡ ¼ 18 (or 22) kJ mol–1 for the unimolecular C–H (or C–D) activation step representing a normal kinetic isotope effect, kH/kD z 10. However, an inverse equilibrium isotope effect (KH/KD z 0.1) was found for the slightly exothermic pre-equilibrium displacement of Kr by C0H12/C6D12 implying that C6D12 binds more strongly to the rhodium center than does C6H12. A subsequent study using neopentane as the alkane substrate gave evidence in support of the same mechanism, and also allowed resolution of near-coincident n(CO) absorptions due to [Cp*Rh(CO)Kr] (1946 cm–1) and [Cp*Rh(CO)(d12 -neopentane)] (1947 cm–1) [18]. Further studies were able to quantify the reactivity of [Cp*Rh(CO)Kr] towards a range of alkanes [20]. It was found that binding of the alkane to Rh becomes more favorable, thermodynamically, as the alkane size is increased, but that the rate of the C–H oxidative addition step shows less variation with linear alkane chain length. No reaction with methane was observed, which was explained by the ineffective binding of methane (relative to excess Kr) to Rh. Alkane complexes, of the sort described above, have been studied in a variety of reaction media since the early 1970s. For example, [M(CO)5(CH4)] (M ¼ Cr, Mo, W) were first identified in low temperature methane matrices [123]. The reactivity of related [M(CO)5(alkane)] complexes has since been probed in alkane solvents under ambient conditions [124] and in the gas phase [125]. Their reactivity (with respect to displacement of the alkane) has been shown to decrease both from left to right and down Groups 5, 6 and 7 with [CpRe(CO)2(alkane)] being the most stable class of alkane complex identified to date [25, 119, 126]. The use of HP IR spectroscopy has enabled the extension of these studies from liquid alkanes (e. g. n-heptane, cyclopentane) to the normally gaseous ethane [25]. It was found that, in liquid or supercritical ethane, [CpRe(CO)2(C2H6)] reacts with CO faster than the analogous n-heptane and cyclopentane complexes, indicating that ethane is less strongly bound than the higher alkanes. This reflects the same trend as found for [Cp*Rh(CO)(alkane)] [20] and [W(CO)5(alkane)] [125].

145

146


3.4

Conclusions

The examples described in this Chapter illustrate how the breadth of applications of HP IR spectroscopy has grown in recent years. The pioneering studies in the field were principally targeted at defining the dominant species present under catalytic conditions and relating this to the observed catalytic activity and selectivity. Whilst clearly an important piece of information, this cannot provide a complete description of the mechanism of a catalytic process. New techniques for analysis of spectral data (e. g. BTEM) have enabled the identification of significant metal complexes (e. g. [Rh(C(O)R)(CO)4]) at the sub-1 % level. Faster sampling, enabled by new cell-designs, has also provided some noteworthy results. The study of individual stoichiometric reaction steps, under conditions somewhat removed from those of the catalytic process, has an important role to play. This is exemplified by the effects of iodide abstracting promoters on the carbonylation of [MeIr(CO)2I3]–, which elucidated the role of these promoters in Ir-catalysed methanol carbonylation. Some catalytic intermediates, of course, are too reactive and short-lived to be directly observed by “conventional” HP IR spectroscopy, and new strategies have been developed to address this. Formally unsaturated 16 electron complexes are commonly generated by photochemical ejection of a CO ligand, and both low temperature (e. g. liquid noble gas) and room temperature (e. g. supercritical fluid) HP IR techniques have been developed to probe these intriguing complexes. The results generally indicate that the formal unsaturation of the metal center is relieved by an interaction of the metal with solvent or ligand. Rapid detection methods (TR IR) have quantified the reactivity of such species and enabled a detailed picture of processes such as C–H activation to emerge. Another clear conclusion is that a complete mechanistic description of a catalytic process generally requires the use of a number of experimental methods. This is particularly obvious in transition metal/phosphine catalysed reactions, where HP IR is ideal to probe the carbonyl ligands, but HP NMR is unrivalled for studying the phosphine coordination. Although not covered specifically in this chapter, recent years have also witnessed a growing use of theoretical (e. g. density functional theory (DFT)) methods to model catalytic reactions. As computing power and the accuracy of these methods grows, the prediction of spectroscopic properties (e. g. vibrational frequencies, chemical shift) is becoming a valuable tool to complement experimental spectroscopic studies.

References

References [1] R. F. Howe, in In-situ Spectroscopy in

[2] [3]

[4] [5] [6]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

[14]

[15]

Heterogeneous Catalysis, Ed. J. F. Haw, Wiley-VCH, Weinheim, 2002, pp. 139–177. S. Rigby, R. Whyman, K. Wilding, J. Phys. E: Sci. Instrum. 1970, 3, 572. R. Whyman, K. A. Hunt, R. W. Page, S. Rigby, J. Phys. E: Sci. Instrum. 1984, 17, 559–561. H. B. Tinker, D. E. Morris, Rev. Sci. Instrum. 1972, 43, 1024–1026. J. M. L. Penninger, J. Catal. 1979, 56, 287–289. P. C. J. Kamer, A. van Rooy, G. C. Schoemaker, P. W. N. M. van Leeuwen, Coord. Chem. Rev., 2004, 248, 2409–2424. S. C. van der Slot, P. C. J. Kamer, P. W. N. M. van Leeuwen, J. A. Iggo, B. T. Heaton, Organometallics 2001, 20, 430–441. S. C. van der Slot, J. Duran, J. Luten, P. C. J. Kamer, P. W. N. M. van Leeuwen, Organometallics 2002, 21, 3873–3883. W. B. Maier, M. Poliakoff, M. B. Simpson, J. J. Turner, J. Mol. Struct. 1982, 80, 83–86. J. J. Turner, M. Poliakoff, S. M. Howdle, S. A. Jackson, J. G. McLaughlin, Faraday Discuss. Chem. Soc. 1988, 86, 271–284. M. B. Simpson, M. Poliakoff, J. J. Turner, W. B. Maier, J. G. McLaughlin, J. Chem. Soc., Chem. Commun. 1983, 1355–1357. R. K. Upmacis, G. E. Gadd, M. Poliakoff, M. B. Simpson, J. J. Turner, R. Whyman, A. F. Simpson, J. Chem. Soc., Chem. Commun. 1985, 27–30. R. K. Upmacis, M. Poliakoff, J. J. Turner, J. Am. Chem. Soc. 1986, 108, 3645–2651. G. E. Gadd, R. K. Upmacis, M. Poliakoff, J. J. Turner, J. Am. Chem. Soc. 1986, 108, 2547–2552. S. A. Jackson, P. M. Hodges, M. Poliakoff, J. J. Turner, F. W. Grevels, J. Am. Chem. Soc. 1990, 112, 1221–1233.

[16] B. H. Weiller, E. P. Wasserman,

[17]

[18]

[19]

[20]

[21] [22] [23] [24]

[25]

[26]

[27]

[28]

[29]

R. G. Bergman, C. B. Moore, G. C. Pimentel, J. Am. Chem. Soc. 1989, 8288–8290. B. H. Weiller, E. P. Wasserman, C. B. Moore, R. G. Bergman, J. Am. Chem. Soc. 1993, 115, 4326–4330. A. A. Bengali, R. H. Schultz, C. B. Moore, R. G. Bergman, J. Am. Chem. Soc. 1994, 116, 9585–9589. R. H. Schultz, A. A. Bengali, M. J. Tauber, B. H. Weiller, E. P. Wasserman, K. R. Kyle, C. B. Moore, R. G. Bergman, J. Am. Chem. Soc. 1994, 116, 7369–7377. B. K. McNamara, J. S. Yeston, R. G. Bergman, C. B. Moore, J. Am. Chem. Soc. 1999, 121, 6437–6443. S. M. Massick, P. C. Ford, Organometallics 1999, 18, 4362–4366. P. C. Ford, S. Massick, Coord. Chem. Rev. 2002, 226, 39–49. K. Noack, Spectrochim. Acta 1968, 43, 1024–1026. X.-Z. Sun, M. W. George, S. G. Kazarian, S. M. Nikiforov, M. Poliakoff, J. Am. Chem. Soc. 1996, 118, 10525– 10532. M. K. Kuimova, W. Z. Alsindi, J. Dyer, D. C. Grills, O. S. Jina, P. Matousek, A. W. Parker, P. Portius, X. Z. Sun, M. Towrie, C. Wilson, J. X. Yang, M. W. George, Dalton Trans. 2003, 3996–4006. M. Poliakoff, S. M. Howdle, S. G. Kazarian, Angew. Chem. Int. Ed. Engl. 1995, 34, 1275–1295. S. M. Howdle, M. Poliakoff, in Supercritical Fluids - Fundamentals and Applications, Vol. 273, Eds. E. Kiran, J. M. Levelt Sengers, Kluwer, Dordrecht, The Netherlands, 1993, pp. 527–537. X. Z. Sun, S. M. Nikiforov, J. Yang, C. S. Colley, M. W. George, Appl. Spectrosc. 2002, 56, 31–39. W. R. Moser, J. E. Cnossen, A. W. Wang, S. A. Krouse, J. Catal. 1985, 95, 21–32.

147

148

3 The Use of High Pressure Infrared Spectroscopy to Study Catalytic Mechanisms [30] W. R. Moser, C. J. Papile, D. A.

[31]

[32]

[33]

[34]

[35]

[36]

[37]

[38]

[39]

[40]

[41]

[42] [43] [44]

Brannon, R. A. Duwell, S. J. Weininger, J. Mol. Catal. 1987, 41, 271–292. W. R. Moser, C. J. Papile, S. J. Weininger, J. Mol. Catal. 1987, 41, 293–302. W. R. Moser, B. J. Marshik-Guerts, S. J. Okrasinski, J. Mol. Catal. A: Chem. 1999, 143, 71–83. W. R. Moser, B. J. Marshik-Guerts, S. J. Okrasinski, J. Mol. Catal. A: Chem. 1999, 143, 57–69. W. R. Moser, in Homogeneous Transition Metal Catalyzed Reactions, ACS Advances in Chemistry Series, Vol. 230, Eds. W. Moser, D. W. Slocum, American Chemical Society, Washington, DC, 1992, pp. 3–18. P. M. Maitlis, A. Haynes, G. J. Sunley, M. J. Howard, J. Chem. Soc., Dalton Trans. 1996, 2187–2196. R. C. Matthews, D. K. Howell, W.-J. Peng, S. G. Train, W. D. Treleaven, G. G. Stanley, Angew. Chem. Int. Ed. Engl. 1996, 35, 2253–2256. J. M. Pearson, A. Haynes, G. E. Morris, G. J. Sunley, P. M. Maitlis, J. Chem. Soc., Chem. Commun. 1995, 1045–1046. T. Ghaffar, H. Adams, P. M. Maitlis, A. Haynes, G. J. Sunley, M. J. Baker, Chem. Commun. 1998, 1023–1024. A. Haynes, P. M. Maitlis, G. E. Morris, G. J. Sunley, H. Adams, P. W. Badger, C. M. Bowers, D. B. Cook, P. I. P. Elliott, T. Ghaffar, H. Green, T. R. Griffin, M. Payne, J. M. Pearson, M. J. Taylor, P. W. Vickers, R. J. Watt, J. Am. Chem. Soc. 2004, 126, 2847–2861. W. R. Moser, A. W. Wang, N. K. Kildahl, J. Am. Chem. Soc. 1988, 110, 2816–2820. A. J. Kunin, M. D. Noirot, W. L. Gladfelter, J. Am. Chem. Soc. 1989, 111, 2739–2741. Z. Dardas, M. G. Ser, Y. H. Ma, W. R. Moser, J. Catal. 1996, 159, 204–211. U. Wolf, R. Leiberich, J. Seeba, Catal. Today 1999, 49, 411–418. J. R. Zoeller, E. M. Blakely, R. M. Moncier, T. J. Dickson, Catal. Today 1997, 36, 227–241; J. R. Zoeller, N. L. Buchanan, T. J. Dickson, R. K. K, Catal. Today 1999, 49, 431–440.

[45] S. Kainz, A. Brinkmann, W. Leitner,

[46]

[47]

[48] [49] [50] [51] [52]

[53]

[54]

[55]

[56]

[57]

[58]

[59] [60]

[61]

A. Pfaltz, J. Am. Chem. Soc. 1999, 121, 6421–6429. M. Allmendinger, R. Eberhardt, G. Luinstra, B. Rieger, J. Am. Chem. Soc. 2002, 124, 5646–5647; M. Allmendinger, M. Zintl, R. Eberhardt, G. A. Luinstra, F. Molnar, B. Rieger, J. Organomet. Chem. 2004, 689, 971–979. M. S. Schneider, J. D. Grunwaldt, T. Burgi, A. Baiker, Rev. Sci. Instrum. 2003, 74, 4121–4128. D. Forster, Ann. N. Y. Acad. Sci. 1977, 295, 79–82. T. W. Dekleva, D. Forster, J. Am. Chem. Soc. 1985, 107, 3565–3567. R. T. Eby, T. C. Singleton, Appl. Ind. Catal. 1983, 1, 275–299. M. Schrod, G. Luft, J. Grobe, J. Mol. Catal. 1983, 22, 169–178. J. R. Zoeller, V. H. Agreda, S. L. Cook, N. L. Lafferty, S. W. Polichnowski, D. M. Pond, Catal. Today 1992, 13, 73–91. M. Fontaine, Y. Castanet, A. Mortreux, F. Petit, J. Catal. 1997, 167, 324–336. M. J. Baker, M. F. Giles, A. G. Orpen, M. J. Taylor, R. J. Watt, J. Chem. Soc., Chem. Commun. 1995, 197–198. L. Gonsalvi, H. Adams, G. J. Sunley, E. Ditzel, A. Haynes, J. Am. Chem. Soc. 1999, 121, 11233–11234. L. Gonsalvi, H. Adams, G. J. Sunley, E. Ditzel, A. Haynes, J. Am. Chem. Soc. 2002, 124, 13597–13612. R. W. Wegman, A. G. Abatjoglou, A. M. Harrison, J. Chem. Soc., Chem. Commun. 1987, 1891–1892. A. C. Marr, P. Lightfoot, E. J. Ditzel, A. D. Poole, G. P. Schwarz, D. F. Foster, D. J. Cole-Hamilton, Inorg. Chem. Commun. 2000, 3, 617–619. D. Forster, J. Chem. Soc., Dalton Trans. 1979, 1639–1645. M. J. Howard, G. J. Sunley, A. D. Poole, R. J. Watt, B. K. Sharma, Stud. Surf. Sci. Catal. 1999, 121, 61–68; G. J. Sunley, D. J. Watson, Catal. Today 2000, 58, 293–307. R. F. Heck, D. S. Breslow, J. Am. Chem. Soc. 1961, 83, 4023–4027.

References [62] R. Whyman, J. Organomet. Chem.

[63]

[64] [65]

[66] [67]

[68]

[69]

[70] [71] [72] [73] [74] [75]

[76] [77]

[78] [79]

1974, 66, C23-C25; R. Whyman, J. Organomet. Chem. 1974, 81, 97–106. M. van Boven, N. H. Alemdaroglu, J. M. L. Penninger, Ind. Eng. Chem. Prod. Res. Dev. 1975, 14, 259–264. M. F. Mirbach, J. Organomet. Chem. 1984, 265, 205–213. P. Pino, A. Major, F. Spindler, R. Tannenbaum, G. Bor, I. T. Horvath, J. Organomet. Chem. 1991, 417, 65–76. K. Kovcs, F. Ungvry, L. Mark, Organometallics 1986, 5, 209–215. M. van Boven, N. Alemdaroglu, J. M. L. Penninger, J. Organomet. Chem. 1975, 84, 65–74. C. Crause, L. Bennie, L. Damoense, C. L. Dwyer, C. Grove, N. Grimmer, W. Janse van Rensburg, M. M. Kirk, K. M. Mokheseng, S. Otto, P. Steynberg, Dalton Trans. 2003, 2036–2042; C. L. Dwyer, H. Assumption, J. Coetzee, C. Crause, L. Damoense, M. Kirk, Coord. Chem. Rev. 2004, 248, 653– 669; L. Damoense, M. Datt, M. Green, C. Steenkamp, Coord. Chem. Rev. 2004, 248, 2393–2407. M. Haumann, R. Meijboom, J. R. Moss, A. Roodt, Dalton Trans. 2004, 1679–1686. M. Garland, Organometallics 1993, 12, 535–543. G. Liu, R. Volken, M. Garland, Organometallics 1999, 18, 3429–3436. C. Li, L. Guo, M. Garland, Organometallics 2004, 23, 2201–2204. J. Feng, M. Garland, Organometallics 1999, 18, 1542–1546. C. Li, E. Widjaja, M. Garland, J. Am. Chem. Soc. 2003, 125, 5540–5548. C. Li, E. Widjaja, W. Chew, M. Garland, Angew. Chem. Int. Ed. Engl. 2002, 41, 3785–3789. D. E. Morris, H. B. Tinker, Chemtech 1972, 554–559. G. Yagupsky, C. K. Brown, G. Wilkinson, J. Chem. Soc., A 1970, 1392– 1401. J. M. Brown, A. G. Kent, J. Chem. Soc., Perkin Trans. 2 1987, 1597–1607. M. F. Sellin, I. Bach, J. M. Webster, F. Montilla, V. Rosa, T. Aviles, M. Poliakoff, D. J. Cole-Hamilton,

[80] [81]

[82]

[83]

[84]

[85]

[86]

[87]

[88]

[89]

[90]

[91] [92] [93]

J. Chem. Soc., Dalton Trans. 2002, 4569–4576. S. Haji, C. Erkey, Tetrahedron 2002, 58, 3929–3941. T. Jongsma, G. Challa, P. W. N. M. van Leeuwen, J. Organomet. Chem. 1991, 421, 121–128. B. Moasser, W. L. Gladfelter, Organometallics 1995, 14, 3832–3838; A. van Rooy, P. C. J. Kamer, P. W. N. M. van Leeuwen, K. Goubitz, J. Fraanje, N. Veldman, A. L. Spek, Organometallics 1996, 15, 835–847. L. A. van der Veen, P. C. J. Kamer, P. W. N. M. van Leeuwen, Organometallics 1999, 18, 4765–4777. L. A. van der Veen, P. H. Keeven, G. C. Schoemaker, J. N. H. Reek, P. C. J. Kamer, P. W. N. M. van Leeuwen, M. Lutz, A. L. Spek, Organometallics 2000, 19, 872–883. S. M. Silva, R. P. J. Bronger, Z. Freixa, J. Dupont, P. W. N. M. van Leeuwen, New J. Chem. 2003, 27, 1294–1296. A. Aghmiz, A. Orejon, M. Dieguez, M. D. Miquel-Serrano, C. Claver, A. M. Masdeu-Bulto, D. Sinou, G. Laurenczy, J. Mol. Catal. A: Chem. 2003, 195, 113–124. S. Deerenberg, P. C. J. Kamer, P. W. N. M. Van Leeuwen, Organometallics 2000, 19, 2065–2072. M. Dieguez, C. Claver, A. M. MasdeuBulto, A. Ruiz, P. W. N. M. van Leeuwen, G. C. Schoemaker, Organometallics 1999, 18, 2107–2115. M. Payne, D. J. Cole-Hamilton, J. Chem. Soc., Dalton Trans. 1997, 3167–3175. G. Braca, A. M. R. Galletti, G. Sbrana, E. Trabuco, J. Mol. Catal. 1989, 55, 184–198. B. D. Dombek, J. Organomet. Chem. 1983, 250, 467–483. B. D. Dombek, Organometallics 1985, 4, 1707–1712. R. Whyman, J. Chem. Soc., Chem. Commun. 1983, 1439–1441; R. Whyman, K. Gilhooley, S. Rigby, D. Winstanley, ACS Symp. Ser. 1987, 328, 108–124; R. Whyman, in Homogeneous Transition Metal Catalyzed Reactions, ACS Advances in Chemistry Series, Vol. 230, Eds. W. Moser, D. W.

149

150


[94]

[95] [96] [97]

[98]

[99]

[100]

[101]

[102]

[103]

[104] [105]

[106] [107] [108]

[109]

Slocum, American Chemical Society, Washington, DC, 1992, pp. 19–31. G. Liu, M. Hakimifard, M. Garland, J. Mol. Catal. A: Chem. 2001, 168, 33–37. R. A. Head, M. I. Tabb, J. Mol. Catal. 1984, 26, 149–158. H. Luo, Y. Kou, X. Wang, D. Li, J. Mol. Catal. A: Chem. 2000, 151, 91–113. W. P. Mul, H. Oosterbeek, G. A. Beitel, G.-J. Kramer, E. Drent, Angew. Chem. Int. Ed. Engl. 2000, 39, 1848– 1851. S. Gautron, R. Giordano, C. Le Berre, J. Jaud, J.-C. Daran, P. Serp, P. Kalck, Inorg. Chem. 2003, 42, 5523–5530. P. W. Vickers, J. M. Pearson, T. Ghaffar, H. Adams, A. Haynes, J. Phys. Org. Chem. 2004, 17, 1007– 1016. R. Whyman, A. P. Wright, J. A. Iggo, B. T. Heaton, J. Chem. Soc., Dalton Trans. 2002, 771. J. Rankin, A. C. Benyei, A. D. Poole, D. J. Cole-Hamilton, J. Chem. Soc., Dalton Trans. 1999, 3771–3782. S. M. Massick, J. G. Rabor, S. Elbers, J. Marhenke, S. Bernhard, J. R. Schoonover, P. C. Ford, Inorg. Chem. 2000, 39, 3098–3106; S. M. Massick, T. C. Buttner, P. C. Ford, Inorg. Chem. 2003, 42, 575–580; W. T. Boese, P. C. Ford, J. Am. Chem. Soc. 1995, 117, 8381–8391. R. Churlaud, U. Frey, F. Metz, A. E. Merbach, Inorg. Chem. 2000, 39, 4137–4142. D. E. Morris, H. B. Tinker, J. Organomet. Chem. 1973, 49, C53–C56. S. Hasnip, S. B. Duckett, D. R. Taylor, G. K. Barlow, M. J. Taylor, Chem. Commun. 1999, 889–890. Q. Xu, Y. Souma, Topics Catal. 1998, 6, 17–26. J. J. Rack, S. H. Strauss, Catal. Today 1997, 36, 99–106. J. J. Rack, B. Moasser, J. D. Gargulak, W. L. Gladfelter, H. D. Hochheimer, S. H. Strauss, J. Chem. Soc., Chem. Commun. 1994, 685–686. P. K. Hurlburt, J. J. Rack, J. S. Luck, S. F. Dec, J. D. Webb, O. P. Anderson, S. H. Strauss, J. Am. Chem. Soc. 1994, 116, 10003–10014.

[110] M. Garland, P. Pino, Organometallics

1991, 10, 1693–1704. [111] C. Fyhr, M. Garland, Organometallics

1993, 12, 1753–1764. [112] J. Feng, M. Garland, Organometallics

1999, 18, 417–427. [113] T. Ghaffar, J. P. H. Charmant,

[114]

[115]

[116]

[117] [118] [119]

[120] [121] [122]

[123] [124]

[125]

[126]

G. J. Sunley, G. E. Morris, A. Haynes, P. M. Maitlis, Inorg. Chem. Commun. 2000, 3, 11–12. S. E. J. Goff, T. F. Nolan, M. W. George, M. Poliakoff, Organometallics 1998, 17, 2730–2737. G. I. Childs, A. I. Cooper, T. F. Nolan, M. J. Carrott, M. W. George, M. Poliakoff, J. Am. Chem. Soc. 2001, 123, 6857–6866. J. Zhang, X.-Z. Sun, M. Poliakoff, M. W. George, J. Organomet. Chem. 2003, 678, 128–133. J. Zhang, M. Poliakoff, M. W. George, Organometallics 2003, 22, 1612–1618. B. H. Weiller, J. Am. Chem. Soc. 1992, 114, 10910–10915. X. Z. Sun, D. C. Grills, S. M. Nikiforov, M. Poliakoff, M. W. George, J. Am. Chem. Soc. 1997, 119, 7521–7525. O. S. Jina, X. Z. Sun, M. W. George, Dalton Trans. 2003, 1773–1778. G. J. Kubas, Acc. Chem. Res. 1988, 21, 120–128. J. K. Hoyano, W. A. G. Graham, J. Am. Chem. Soc. 1982, 104, 3723–3725; A. H. Janowicz, R. G. Bergman, J. Am. Chem. Soc. 1982, 104, 352–354; A. J. Rest, I. Whitwell, W. A. G. Graham, J. K. Hoyano, A. D. McMaster, J. Chem. Soc., Chem. Commun. 1984, 624–626. R. N. Perutz, J. J. Turner, J. Am. Chem. Soc. 1975, 97, 4791–4800. P. M. Hodges, S. A. Jackson, J. Jacke, M. Poliakoff, J. J. Turner, F. W. Grevels, J. Am. Chem. Soc. 1990, 112, 1234–1244. C. E. Brown, Y. Ishikawa, P. A. Hackett, D. M. Rayner, J. Am. Chem. Soc. 1990, 112, 2530–2536. S. Geftakis, G. E. Ball, J. Am. Chem. Soc. 1998, 120, 9953–9954; G. I. Childs, C. S. Colley, J. Dyer, D. C. Grills, X. Z. Sun, J. X. Yang, M. W. George, J. Chem. Soc., Dalton Trans. 2000, 1901–1906.

4 Processing Spectroscopic Data Marc Garland

4.1

Introduction

Signal processing plays an important part in electrical engineering, communication theory, spectroscopy and many other branches of science and engineering [1]. Across many of these fields, generic applications of digital signal processing include filtering, phase analysis, signal characterization and transformation i. e. fast Fourier transformation (fft). In the context of this book, a generic discussion of signal processing is simply inappropriate. Indeed, the emphasis of this chapter should and will be to highlight “specialized aspects” of signal processing which have been recently developed and to indicate explicitly and by means of example, how these algorithms can be used to solve mechanistic problems in homogeneous catalysis. Having said this, a broader context for the problem statement as well as detailed experimental considerations are essential for a well rounded exposition on this complex subject. Accordingly, this introduction begins with an applied mathematics perspective and ends with a concise spectroscopic problem statement where the crucial role of signal processing is specified. Every attempt will be made to keep the mathematics to a bare but essential minimum, and the exposition readable for the practising catalytic chemist. The primary aim is to make these ideas as accessible as possible to the catalytic community and hence expedite their wider usage. In many branches of the physical sciences and engineering, the term “inverse problem” holds a very important and precise meaning [2, 3]. Inverse problems are crucial and highly specialized applied mathematical problems which frequently define the focus of a field of study. Extremely well known examples of inverse problems include (i) the inversion of seismic data for oil exploration [4], (ii) the inversion of attenuated X-ray data for medical imaging [5], and (iii) the inversion of diffracted X-ray data for 3D crystal structure determinations [6]. In each case, copious experimental data are needed and the researcher seeks a working model of the physical system. In a slightly more mathematical representation (Eq. (1)), a set of observations y (data) is given and the function f(x) (model) is determined. The mathematical operation of going from y to f(x) is the inverse problem. In contrast, the Mechanisms in Homogeneous Catalysis. A Spectroscopic Approach. Edited by Brian Heaton Copyright c 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN 3-527-31025-8

152

4 Processing Spectroscopic Data

direct or so-called “forward problem” is by far the most commonly encountered mathematics problem, and in this case, the researcher knows or specifies a priori a model f(x) and then simulates the results y. y p f(x)

(1)

Many physically motivated inverse problems are of daunting complexity, their accessible range of measurements is frequently incomplete, and considerable experimental design and planning are needed in order to guarantee, as far as possible, uniqueness of the final solution. The solution of an inverse problem frequently necessitates the use of one or more sophisticated mathematical tools [7], and these must often be applied in a very novel manner. Many but not all inverse problems are “ill-posed” [8]. The term ill-posed explicitly denotes inverse problems where incomplete information and/or uniqueness issues are of paramount concern [9, 10]. Most difficult physical science related inverse problems are by their very nature ill-posed. The initial mathematical solution of a long-outstanding inverse problem in the physical sciences frequently marks a turning point in the field and may usher in the birth of an associated specialization, as was the case with examples (i)–(iii). The previously mentioned inverse problems, (i)–(iii), provide examples of frequently used methodologies and issues to be addressed. The measured signals are seismic intensity versus time and terrestrial cartesian surface coordinates i. e. I(t,x,y), intensity versus distance and angle i. e. I(r,u) and X-ray intensity versus position and angle I(z,u) respectively. In other words, the physical objects cannot be taken apart and the individual parts sampled directly. The whole object must be probed as a whole by external stimulation, and then the response of the whole system must be detected. The raw detector signals may be filtered and numerically adjusted /corrected in some manner – this is part of the signal processing procedure. The signals are then processed further to invert the information to get a model of the physical system. The primary numerical tools used for examples (i)–(iii) are the z-transform [11], the Radon transform [12, 13] and the Fourier transform, respectively [14]. The noise inherent in the measurements together with the signal processing steps may, and probably will, result in some minor or even non-negligible artifacts in the final solution. In this regard, attention to experimental design can greatly help to minimize these effects. Researcher experience is often crucial in evaluating the limitations of the solutions obtained. As already emphasized, the initial physical problem and successful solution is often referred to as the inverse problem in the applied mathematics literature. In the engineering literature, the subsequent repeated application of a developed numerical approach to other examples from the same general class of problems is frequently referred to as system identification. Let us now turn our attention to a transition-metal homogeneous catalytic organic synthesis conducted in a liquid phase. Initially, solvent, soluble organic reagents, organometallic precursor, ligand, and promoter are brought together in

4.1 Introduction

some order. We explicitly acknowledge the presence of impurities. After mixing of the compounds present, various reactions occur. These reactions include, but are not restricted to, (i) the transformation of organometallic precursors to intermediates, (ii) the transformation of organic reagents to organic products via one or more sets of organometallic intermediates, (iii) the degradation of organometallic precursors and intermediates to inactive species and (iv) uncatalyzed stoichiometric reactions among the organic reactants. The collective action of the intermediates resulting in product formation is the phenomenon we call catalysis. Clearly, the physical system is very complex. There are many species and many reactions involved, and there is, in general, a superposition of the induction kinetics, product formation kinetics associated with the turnover frequency, and deactivation kinetics. In situ spectroscopic measurements of a catalytic system provide a considerable opportunity to determine the chemical species present under reactive conditions. FTIR and NMR have been the two most frequently used in situ spectroscopic methods (see Chapters 2 and 3). They have been successfully used to identify labile, non-isolatable transient species believed to be involved in the catalytic product formation. Furthermore, efforts have been made to use this information in order to obtain more detailed kinetics, by decoupling induction, product formation, and deactivation. Thus, in situ spectroscopic techniques have the potential for considerably advancing mechanistic studies in homogeneous catalysis. Returning to the general liquid phase catalytic system, assume that you have chosen an appropriate spectroscopy to investigate the system under reaction conditions. The spectroscopy provides spectra, i. e. absorbance A(t), at specific intervals in time. If {S} denotes the complete set of all species that exist at any time in the physical system, then {S}obs is the subset of all observable species obtained using the in situ spectroscopy. This requires that the pure component spectra {a1..as}obs are obtainable from the multi-component solution spectra A(t) without separation of constituents, and without recourse to spectral libraries or any other type of a priori information. Once reliable spectroscopic information concerning the species present under reaction are available, down to very low concentrations, further issues such as the concentrations of species present, the reactions present, and reaction kinetics can be addressed. In other words, more detailed aspects of mechanistic enquiry can be posed. The experienced catalytic chemist or chemical reaction engineer will immediately recognize that the study of a new catalytic reaction system using an in situ spectroscopy, has a great deal in common with the concepts of inverse problems and system identification. First, there is a physical system which cannot be physically disassembled, and the researcher seeks to identify a model for the chemistry involved. The inverse in situ spectroscopic problem can be denoted by Eq. (2). Secondly, the physical system evolves in time and spectroscopic measurements as a function of time are a must. There are realistic limitations to the spectroscopic measurements performed. For this reason as well as for various other reasons, the inverse problem is ill-posed (see Section 4.3.6). Third, signal processing will be needed to filter and correct the raw data, and to obtain a model of the system. The ability to have the individual pure component spectra of the species present in

153

154


a catalytic synthesis, including the intermediates, opens enormous possibilities for mechanistic understanding.

A p {a1..as}obs

(2)

The scope of this chapter is devoted to solving Eq. (2) and the implications that thereby arise. Accordingly, a description of the physical system is presented (Section 4.2), experimental design is discussed at length (Section 4.3), data pre-processing is addressed (Section 4.4) the current status of pure component spectral recovery is reviewed along with the salient mathematical issues (Section 4.5), and future directions are indicated (Section 4.6). Although Eq. (2) is the necessary starting point for determining the moles of all intermediates, the algebraic structure of the catalytic system, and the kinetics of induction/turnover frequency/deactivation – these goals lie far beyond the scope of the present contribution.

4.2

The Catalytic System

In general, the homogeneous catalyzed synthesis will take place in a physical system consisting of multiple phases and interfaces. The chemical constituents of the physical system will include many species, and these species are involved in a variety of reactions. Some of these reactions are directly associated with the catalytic synthesis of the desired organic product and some are not. In addition to the homogeneous catalytic reaction of interest, other competing stoichiometric, homogeneous catalytic and even heterogeneous catalytic reactions may, and probably will, occur simultaneously. Given the complexity of the system, descriptions of the physical and chemical aspects are needed, and here a common continuum mechanics approach is adopted [15]. 4.2.1

Recycle CSTR with Analytics

Although many experimental configurations are possible, one configuration is by far the easiest for most catalytic chemists to implement.1) Accordingly, we assume that the in situ spectroscopic study is conducted with a continuous stirred tank reaction (CSTR) vessel with recycle. In general, the liquid phase from the CSTR flows through a metering pump, through analytical instruments, and then back to the CSTR. In modern laboratories, a large variety of sensors and spectrometers 1) We recognize that immersion probes for direct

introduction into a vessel are available for some sensors and a very limited number of spectroscopies. Therefore immersion probes represent a more specialized and limited case.

In addition, since attenuated total reflection (ATR) probes belong to the category of absorbance–reflection spectroscopies, artifacts are known to arise [16].

4.2 The Catalytic System

are available, and these instruments can either be purchased for or adapted to flowthrough measurements. We can assume that the placement of the in-line measurements is most convenient along the recycle stream. For our purpose, it is convenient to classify the measurements according to the format of the data produced. Sensors provide scalar valued quantities of the bulk fluid i. e. density r(t), refractive index n(t), viscosity h(t), dielectric constant e(t) and speed of sound vs(t). Spectrometers provide vector valued quantities of the bulk fluid. Good examples include absorption spectra A(t) associated with (1) far-, mid- and near-infrared FIR, MIR, NIR, (2) ultraviolet and visible UV–VIS, (3) nuclear magnetic resonance NMR, (4) electron paramagnetic resonance EPR, (5) vibrational circular dichroism VCD and (6) electronic circular dichroism ECD. Vector valued quantities are also obtained from fluorescence I(t) and the Raman effect J(t). Some spectrometers produce matrix valued quantities M(t) of the bulk fluid. Here 2D-NMR spectra, 2D-EPR and 2D-flourescence spectra are noteworthy. A schematic representation of a very general experimental configuration is shown in Figure 4.1 where t is the recycle time for the system.

Figure 4.1 Schematic diagram of a general-purpose CSTR recycle system for in situ investigations of liquid-phase homogeneous catalyzed reactions. The blocks represent in-line instruments and their signals, namely, (i) sets of scalar valued measurements (sensors), (ii) sets of vector valued measurements (1D spectroscopies) and (iii) sets of matrix valued measurements (2D spectroscopies). The recycle time for the system is given by t.

155

156


4.2.2

Physical System

Within the experimental apparatus, one or more phases {P1..PP}, and one or more interfaces {If1…Ifif } are present. Most importantly, one of these phases is a liquid phase PL where the homogeneous catalytic reaction of interest occurs.1) Consistent with Figure 4.1, the liquid phase PL has interfaces with the experimental apparatus i. e. CSTR, pump, sample cells and tubing, as well as any gas or solid phases present. If mass transfer occurs across any interface with PL, then PL is considered open. If no mass transfer occurs, then PL is closed. The physical system V consists of the phases and the interfaces contained within the experimental apparatus. It is convenient to define a boundary @ for the system and this will be taken to be the surface of the experimental apparatus. The phases {P1..PL} in the system may or may not be entirely uniform in temperature due to the need for heating/cooling, the presence of reaction, the degree of insulation etc.2) Also, the total thermodynamic pressure P in the system may not be entirely constant throughout the system even if low viscosity solvents are used in the syntheses.3) Energy transfer with the physical system V occurs. For our purpose we will consider thermal, mechanical and radiational terms. The first is characterized by the heat transfer at the boundary @. The second is characterized by the stirring and pumping of the fluid and by intense forced periodic mechanical energy (sound) if part of the system is sonicated i. e. to induce sono-homogeneous catalysis [22]. Radiational energy transfer arises from three primary sources, namely, stray natural and artificial light, the spectroscopic measurements, and deliberate irradiation to induce reaction. The latter is exemplified by UV–VIS exposure and intense microwave exposure using a photoreaction source or microwave oven leading to photo-homogeneous catalysis [23] and microwave-homogeneous catalysis [24]. For radiation induced chemical reaction, a distinction is often made between single-photon and multiple-photon events. The differentiation is based on the intensity (flux) of the photon source. For single photon events, the maximum energy of mid-IR photons is ca. 2.4 kJ mole–1 and near-IR photons ca. 48 kJ mole–1 [25, 26]. Therefore, single photon mid-IR irradiation is normally considered non-destructive. However, intense irradiation and hence multiple photon absorption in mid-IR is known to promote chemical transformations [27, 28]. As an example of NIR proprocess fluid and atmosphere, a ca. 0.5 hC m–1 [17, 18] and interfacial homogeneous catalysis or less change in the process fluid can be expected. [18, 19] are known. The associated signal processing treatment for the latter is beyond 3) The primary contributions to pressure drop, in approximately decreasing order, will be (1) inthe scope of the present chapter. 2) Simple heat transfer calculations i. e. using line filters (2) needle valves (3) check valves (4) sources like Ref. [20], indicate that even unthe spectroscopic cells and (5) capillary tubing. insulated lines are not a severe problem in Methods for detailed calculations can be found most spectroscopic applications. With insuin Ref. [21]. Our experience is that total recycle pressure drop is a small fraction of a bar with lated 1/8, 1/16, and 1/32 inch stainless steel and PEEK tubing, with a flow velocity of ca. normal flow rates. 0.1–1m s–1 and driving force of 50 hC between 1) We recognize that phase transfer catalysis

4.2 The Catalytic System

moted reaction, photons of energy of only ca. 35 kJ mole–1 (ca. 820 nm) can lead to the photolysis of organic peroxide bonds and hence induce chemical reaction [29]. Conventional spectrometers are typically single photon event sources. This gives rise to the following two useful working definitions for this chapter. Dark Reactions: If a homogeneous catalytic system is studied with the total exclusion of (1) stray light, (2) intense microwave irradiation or photo irradiation, and (3) spectrometer photon sources with hv j NIR radiation, then the reactive system will be considered to be thermally activated. Only dark reactions occur. Irradiated Reactions: If a homogeneous catalytic system is studied with either (1) introduction of stray light, (2) intense microwave irradiation or photo irradiation, or (3) spectrometer photon sources with hv j NIR radiation, then the reactive system will be considered to be both thermally and potentially radiationally activated. Both dark reactions and potentially photoinduced reactions occur. The phases {P1..PL} may or may not be thoroughly mixed at any particular time t. The phases are certainly not at equilibrium, (i. e. vapor–liquid, liquid–liquid, vapor– liquid–liquid equilibrium), until all reactions have proceeded to completion. Therefore, mass transfer can be assumed to occur across the interfaces for all time t. Since mass transfer between phases occurs for all time t, the mass and volumes of each phase are non-constant in the interval between initial and final reaction times [toItI tf ]. Volumetric effects can and will influence the spectroscopic results and data analysis, particularly when calculating moles present. 4.2.3

Chemical Description

The homogeneous catalytic reaction is initiated by introduction of the reagents to the CSTR. If no further additions of reagents occur, then the homogeneous catalytic reaction is performed in a batch manner. If further additions of reagents occur, for example at various intervals, then the homogeneous catalytic reaction is performed in a semi-batch manner. Each further addition of reagents can be considered as a perturbation of the system. In both the batch and semi-batch cases, the homogeneous catalytic system exists for a finite interval of time [to, tf ]. The solvent and all reagents must be brought together in some manner. Many of the reagents are solids in their natural state, i. e. high molecular weight substrates, organometallics and ligands. Furthermore, some catalyst precursors used in transition metal homogeneous catalysis are virtually insoluble in frequently used organic solvents. Good examples include inorganic metal complexes, halides and even finely dispersed metal. This initial startup or preparation of the catalytic system can be classified into two categories, where the first is by far the preferable starting point for in situ spectroscopic studies.

157

158


Type A Systems, All Fluid Components: The state of the system is a fluid or a mixture of fluids at to. For example, the vessel may contain solvent, dissolved organics, and dissolved organometallics, ligands, inorganics (pre-dissolve if the natural state of the solute is solid). Type B Systems, Some Solid Components: The state of the system is a fluid/solid mixture at to. For example, the vessel may have been filled with solvent and gases, but one or more of the organics, organometallics, ligands, or inorganics are added as an amorphous or microcrystalline solid. The homogeneous catalytic reaction occurs in the multi-component liquid phase PL. The chemical constituents of the liquid phase PL include Hþ, e–, atoms, ions, and molecules etc. which are dissolved/solvated in one or more molecular or ionic solvents. Primary examples of the ions and molecules present are the dissolved organic and organometallic reagents, intermediates and products. By definition, all the molecular and ionic species involved directly in the homogeneous catalysis are soluble in this liquid phase PL. The set of all dissolved species in the phase will be denoted by Eq. (3). S ¼ {S1….Ss}L

(3)

By definition, it is necessary to recognize geometric isomers, isotopomers and stereoisomers as distinct species. Moreover, there is the pragmatic issue that regio-selectivity, isotopic labeling and stereo-chemical investigations are three very important avenues of mechanistic enquiry. The in situ spectroscopies and the signal processing have limitations. Therefore, the set of observable species Sobs is a proper subset of all liquid phase species S. The validity of Eq. (4), namely, that the number of observable species is less than the number of species, is easily verified. Regardless of the instrument, the sensitivity is finite, and some dilute and most trace species must be lost in the experimental noise. In addition, numerous experimental design shortcomings further contribute to the validity of Eq. (4). Sobs ¼ {S1….Ss}L,obs C S

(4)

4.3 Experimental Design

4.3

Experimental Design

The classic texts on the design of experiments in scientific and engineering studies emphasize (1) measurement and instrumentation, (2) sources of error, (3) factor design etc. [30, 31] This section addresses step-by-step many of these issues for in situ studies, and does so by integrating relevant chemical and chemical engineering concerns and concepts. This section attempts to provide a very useful short-list of design considerations for the experimentalist so that Eq. (2) can be solved. 4.3.1

Transport Time-scales

In computational science and engineering, researchers frequently speak of multiple time-scale problems. These are problems where the characteristic times for events span many orders of magnitude. More often than not, the full range of desired time-scales is not experimentally accessible. The combined experimental and numerical approach must settle for analysis of a narrow window of time-scales. For most of our purposes, the half-life of an event will prove to be a useful definition of characteristic time-scale In a multiple phase CSTR, gas–liquid, liquid–liquid and solid–liquid mass transfer must be considered. The characteristic time-scale for gas–liquid mass transfer t gl, [32, 33] and liquid–liquid mass transfer t ll [34] in well agitated vessels is ca. 100 s or less. Furthermore, even for Type A Systems, liquid–liquid mixing must be achieved before reaction can proceed. In a diffusion-reaction context [35], segregation denotes the initial state of two miscible fluids and micro-mixedness denotes the final intimately mixed state of the species [36]. The characteristic time-scale for liquid–liquid mixing t mix, i. e. starting from a segregated mixture and going to micro-mixing, is also ca. 100 s or less in typical CSTRs [37, 38]. In summary, 100 s is a fair first approximation for the characteristic time-scales for mass transfer considerations in a CSTR. For convenience we will denote this tcstr z 100 s. A more detailed exposition containing an experimental checklist for transport effects in homogeneous catalysis can be found elsewhere [39]. The characteristic time-scales mentioned above take into account some but not all practical considerations. For example, really intense stirring (rpm i 500) in the CSTR is not recommended for in situ studies since a deep vortex will be formed in the liquid, gas will be entrained, and two-phase flow will occur in the recycle line. Also, two-phase flow will generally cause cavitation in a mechanical pump (possibly stopping flow) and induce irreproducible spectroscopic measurements. In Figure 4.1, the CSTR is connected to a recycle loop and measurement cells. If the cells and recycle loop have a volume V and the pump has a volumetric pumping speed of vp then the characteristic residence time is t rt ¼ V/vp. With our various laboratory set-ups, the characteristic residence time in the recycle loop is usually ca. 50 s. However, back mixing in the spectroscopic cells causes considerable disper-

159

160


sion, and the residence time distribution of a non-reactive tracer is seriously affected [40]. Much longer times can be observed when large volume cells and low pumping speeds are used. It must also be noted that the CSTR and recycle streams with cells are configured in series and that such a configuration can be approximated as a cascade of CSTRs with recycle or a CSTR with plug-flow and recycle [41]. In summary, a fair characteristic time-scale t system for the transport of a tracer throughout the liquid phase of a well-designed CSTR-recycle system is ca. 100 s.1) The characteristic time- scale mentioned takes into account that laminar flow (Reynolds numbers less than ca. 2000) and not turbulent flow may be desired, particularly in many spectroscopic cells. It must be emphasized that the above considerations were made in the absence of reaction. Interfacial mass transfer followed by reaction requires further consideration. The Hatta regimes classify transfer-reaction situations into infinitely slow transport compared to reaction (Hatta category A) to infinitely fast transport compared to reaction (Hatta category H) [42]. In the first case all reaction occurs at the interface and in the second all reaction occurs in the bulk fluid. Homogenous catalytic hydrogenations, carbonylations etc. require consideration of this issue. An extreme example of the severity of mass transport effects on reactivity and selectivity in hydroformylation has been provided by Chaudari [43]. Further general discussions for homogeneous catalysis can be found elsewhere [39]. Taken together, the transport and reaction considerations have two broad implications for dark reactions in well-designed CSTR-recycle systems: Initial Times: Immediately after the introduction of a reagent into a CSTR-recycle system, and for the first ca. 100 s, the compositions of the fluid elements throughout the liquid phase homogeneous catalytic system are measurably different. Longer Reaction Times: For homogeneous catalytic systems without reactants in a second phase (PG or PL’), the fluid elements in the CSTR, recycle loop, and cells are quite similar at times greater than ca. 100 s. Furthermore, for homogeneous catalytic systems with reactants in a second phase (PG or PL’), the fluid elements in the CSTR, recycle loop, and cells are quite similar at times greater than ca. 100 s if transport is fast compared to reaction i. e. Hatta regimes F–H. 4.3.2

Reaction Time-scales

In stark contrast to the transport considerations, typical time-scales t react for elementary and non-elementary steps in organometallic chemistry and homogeneous catalysis range from vibrational motions of ca. 10 –10 s or less [44], to ca. hours or even days for the formation of products and side products. Moreover, the three pri1) The experimentalist should treat this opti-

mistic time with caution. Our group’s first attempts at CSTR recycle with in-line cells resulted in characteristic times of the order of

300–500 s in the presence of g-l mass transfer. The presence of dead volumes should not be casually dismissed.


mary overall reactions can be considered (i) the transformation of organometallic precursors to intermediates, (ii) the transformation of organic reagents to organic products via one or more sets of organometallic intermediates, and (iii) the degradation of organometallic precursors and intermediates to inactive species. These phenomenon give rise to three characteristic time-scales, namely t ind, t TOF ¼ TOF–1 and t deact for induction, turnover frequency and deactivation, respectively.1) At initial reaction times, i. e. for the first ca. 100 s, all three phenomena should be controlled by transport considerations. If the induction kinetics are intrinsically fast compared to transport, then the evolution of the system is transport controlled, and most of the precursor cannot be converted to intermediates before 100 s is reached. Furthermore, if both induction kinetics and turnover frequency are intrinsically fast compared to transport, the system may experience only ca. one turnover within the first 100 s. Finally, if deactivation kinetics are intrinsically fast compared to transport, a significant fraction of precursor has been degraded to inactive species within the first 100 s. The net effect, for better or worse, is that transport effects bias the in situ observations and hence the accessible set of observable species in Eq. (4).

Spectroscopic Measurements Although more than one generic type of spectroscopic measurement was mentioned in the introduction, i. e. A(t), I(t), J(t) and M(t), only A(t) will be discussed in detail. Most of the issues surrounding A(t) are readily transferred to the other types of measurements. The few important exceptions will be mentioned. The absorbance spectroscopy covers a range of frequency measurements and each measured frequency will be called a data channel. A spectrum with 1024 measured frequencies will have 1024 channels of data. A spectrum is a vector A1qn. In any particular synthesis, the experimentalist will measure a few preliminary spectra n. These preliminary spectra might be backgrounds of various sorts, the cell with solvent, as well as a spectrum or two made after the addition of each reagent in order to prepare the catalytic system. If a reagent is added, it is advantageous to wait the required ca. 100 s for system homogeneity before taking the spectrum. The consolidated block of preliminary spectra form a matrix Anqn. Upon addition of the last reagent to the CSTR, the catalytic system is initiated. During the synthesis, k spectra will be taken. These k spectra form a matrix Akqn. The sum total of spectra for a synthesis form a matrix A(nþk) qn. If e experiments (syntheses) are performed, the matrix of raw spectroscopic data is a matrix A(nþk)eqn. 4.3.2.1

1) There is one significant, but frequently over-

looked implicit assumption when discussing mixing and reaction. The assumption is that only one solution to the diffusion-reaction equation exists. This does not have to be the case. Already in 1952, the mathematician Turning predicted the existence of exotic solutions in 3D space [45]. These give rise to

spatio and spatio-temporal concentration patterns in reacting fluids. The existence of such patterns for some reactive systems has now been confirmed [46], although to the best of my knowledge, experimental evidence in homogeneous catalytic systems for organic syntheses has not yet been obtained.

161

162


Typical numerical values of the order of n¼10, k¼100–1000, e¼20 and n¼5000 are currently being used by our group for FTIR studies. Although the term “pure component spectrum” is a very useful concept, it is somewhat misleading, regardless of the spectroscopy used. The pure component spectrum of a species, i. e. S2, is not identical at some early time and some later time, i. e. k¼46 versus k¼817. In general, the band positions will move to another frequency and the band shapes will change. The pure component spectrum of S2 is an example of a non-stationary signal. Therefore, the Lambert–Beer–Bourguer law is not strictly valid over extended regions of composition space, T and P. Let {a1..as}obs denote the average pure component spectra of the observable species over the entire set of measurements. Then a model for the spectroscopic data can be constructed (Eq. (5)) where c is the concentration and e represents both experimental error and model error (non-linearities) [47].1)

A(nþk)eqn ¼ c(nþk)eqs asqn þ e(nþk)eqn

(5)

The experimental error can often be minimized by (i) using fixed dedicated spectroscopic cells/probes, (ii) avoiding situations that require Type B Systems to start the catalytic reaction and (iii) avoiding situations which generate colloidal metal, or give rise to particulates or metal plating. The model error can often be minimized by choosing modest regions in composition space, T and P for the experimental plans and analysis (see Section 4.3.5.4).

Time-scales for Spectroscopic Measurements The quality of the system identification results is strongly dependent on the manner in which the spectroscopic measurements are made. In this regard, the timescale of the individual spectral measurements t spect is crucial. Many good resolution FTIR, Raman, UV–VIS, fluorescence and 1H,19F,31P NMR spectra can be obtained in 100 s or less.2) Also, many VCD, ECD, and 2D NMR spectra can be obtained in 1000 s or less. The time-dependence of the concentrations of the observable species {S1…Ss}L, obs provide estimates for the half-lives {t 1…t s}. Although these half-lives are not known during the data acquisition period, they can be obtained during data analysis. Let min{t 1…t s} denote the smallest value. It is important to check that Eq. (6) is valid. If Eq. (6) does not hold, then one or more of the pure component spectra possess some temporal induced distortion and are therefore inaccurate. 4.3.2.2

t spect II min{t 1…t s} 1) We note in passing that many more spectra

will be accumulated compared to the observable species present. This represents an overdetermined problem, one of many qualities present in real physical ill-posed problems.

(6) 2) Spin relaxation times are a very important

issue in flow-through NMR, and may preclude some types of measurements.


One can and should enquire about the time-scale of the spectroscopic measurements and the reaction time-scales. In general, there will be a few observable species i. e. organometallics, associated with the induction kinetics, and the deactivation kinetics. Therefore, the kinetic time-scales are similar to the half-lives of these species. If t spect is short compared to the half-lives of these species, both the induction and deactivation kinetics can be modeled accurately. The turnover frequency is a special case. It is the rate of product formation normalized by the instantaneous moles of intermediates. The time-scale of the turnover frequency t TOF ¼ TOF–1 of the cycle can be equal to or even less than t spect without serious consequences. The experimentalist simply needs to ensure that t spect is much less than the half-lives of the observable species associated with both the organics and organometallic intermediates involved in the cycle studied. 4.3.3

The Meaning of “In Situ” Studies

Some controversy surrounds the usage of the term in situ. Some researchers even go so far as to suggest that unless a reactor and spectroscopic cell/probe are one and the same unit, the measurement cannot be in situ. The results of Section 4.3.1 suggest otherwise. If the fluid elements in the cell are compositionally similar to the fluid elements in the CSTR and are at similar temperature and pressure, then they are indistinguishable. The measurements are in situ. With proper care regarding transport effects, and reaction considerations, an experimental apparatus with a configuration like Figure 4.1 provides in situ spectroscopic capability for dark reactions. The experimental configuration in Figure 4.1 does not provide in situ capability for sono-, photo- and microwave-induced homogeneous catalysis. The chemistry occurring in the fluid elements of an irradiated or sonicated CSTR are not similar to the fluid elements in the recycle stream and sample cells. However, Figure 4.1 maintains the potential for on-line measurement capability, particularly for product, and for some side-products. The issue concerning intermediates is more complex. Experimental systems without irradiation of the reactor, but with (i) Raman spectroscopy with UV or VIS sources (ii) UV–VIS and ECD spectroscopy and (iii) fluorescence spectroscopy, represent special cases. It is possible, even perhaps probable, that the chemistry observed in the sample cell is dissimilar to the chemistry in the CSTR fluid elements. This is particularly worrisome in the cases were UV exposure occurs. Having said that, a test can be performed using exact replicate catalytic runs. For example, in the first run, the sample cell can be irradiated for the full duration [to I t I tf ]. In the second and third case, the sample cell can be irradiated for 50 % and 10 % of the duration. If the same set of observable species are found, and the time dependences of all the observable species are the same between runs, then, to a first approximation, the spectroscopy has not affected the system. In conclusion, cases (i) to (iii) should be treated with caution by the experimentalist.

163

164


4.3.4

The Planning of Experiments

The planning of good experiments for an in situ spectroscopic homogeneous catalytic study is a non-trivial matter. Indeed, the number and quality of the pure component spectra recovered are strongly dependent on the experimental planning.

Batch and Semi-batch For our purpose, there are two types of catalytic syntheses, namely, batch runs and semi-batch runs. In a batch run, each reagent is added only once. The chances are high that one batch run will fail to give satisfactory results during analysis to solve Eq. (2). This is due to (i) the high probability of co-linearities in a single data set arising from one initial condition, and (ii) closure [48, 49]. Consequently the pure component spectra cannot be untangled from one another. If experimental design is planned with only batch runs, then many batch runs with different initial conditions need to be planned. In semi-batch runs, each reagent is added a number of times, at discrete intervals. Once the reaction is initiated, each further addition of reagents perturbs the system. Multiple perturbations in a single semi-batch run go a long way towards avoidance of co-linearities and closure. Moreover, a single semi-batch run is far more frugal with the resources needed for experimental work. These include the spectrometer time, and the expensive organometallic reagents, ligands and substrates. It will be shown later (see Section 4.5.3.3), that a single very well planned half-day semi-batch exploratory study can provide sufficient high quality data for solution of Eq. (2). 4.3.4.1

Choice of Spectrometers The information content of a particular spectroscopic method, its sensitivity, and its ease of implementation are all important considerations. The utility of combining more than one complementary spectroscopic method is also an issue that requires serious consideration. 4.3.4.2

Vibrational spectroscopies Any polyatomic molecule has at least one vibration in the infrared region, and in general there are 3N–6 fundamental modes in a polyatomic molecule with N atoms [50]. This characteristic, together with the localized signal characteristics, translates to a high information content for vibrational spectroscopy. Furthermore, vibrational spectroscopy is quite sensitive to low solute concentrations, and in many cases quantitative vibrational work can be performed at the ppm level. In general, well designed experiments, covering both FIR and MIR, can readily distin-


guish geometric isomers, and isotopomers of both the organics and organometallics present.1) Serious consideration must be given to the use of more than one vibrational spectrometer. The list of currently available commercial vibrational spectrometers include: NIR and NIR-CD, MIR and MIR-CD, FIR, Raman and Raman Optical Activity (RAO). Diasteriomers have distinct non-circularly polarized vibrational spectra. Enantiomers are only distinguishable by CD and RAO [51, 52]. If a synthesis is performed where enantiomers are present, then a proper experimental design is needed to solve Eq. (2), and this necessitates the use of CD/ROA. If CD/ROA is not used, then one or more of the pure component spectra a1..asobs are lumped parameters i. e. they represent both enantiomers present. The limitations of vibrational spectroscopy include the inability to detect monoatomic molecules and ions. Noble gases and halide ions are therefore problematic. Oligomeric and polymeric species are also problematic. If only a very few oligomers are present, and if their concentrations change in the data set, then Eq. (2) is still potentially solvable. However, polymeric materials, where a large distribution of molecules is present, must be treated as a lumped parameter. It has been demonstrated that the solution of Eq. (2) using vibrational spectroscopy is possible with modest concentrations of suspended solids in the liquid phase (see Section 4.5.3), but not high concentrations. Polyatomic organic solvents are the norm in homogeneous catalysis and the intensity of the C–H region 3200– 2800 cm–1 is a complication. Substitution of deuterated solvents can be very advantageous since the region from 2400 to 2200 cm–1 (C–D vibrations) is normally of limited use in homogeneous catalysis. Finally, Raman using UV and VIS excitation sources conflict with the concerns expressed in Section 4.2.2. Fortunately, commercial Raman instruments using NIR excitation are available. Magnetic resonance NMR is by far the most useful spectroscopy in the chemical sciences and nearly every element in the periodic table has at least one isotope with a magnetic moment [53, 54]. Since nearly every element is accessible, very small chemical shifts are discernible and the signals are highly localized, the information content in NMR spectroscopy is very high. 2D NMR extends the information content considerably. For many applications, the sensitivity of NMR is adequate, at least for the major species present. In general, well designed experiments should readily distinguish geometric isomers of both the organics and organometallics present. However, problems with isotopomers can arise if some isotopes are not magnetically active. NMR is not limited to the detection of polyatomic molecules and ions. Monoatomic molecules and ions are also accessible. If only a very few oligomers 1) With regard to isotopomers, two sub-cate-

gories arise. The first is the natural abundance isotopomer distribution of the organics and organometallics. The second is specific isotopic labeling. Normally, the pure component spectra a1..asobs recovered are lumped para-

meters involving all the isotopomers present; this is a consequence of co-linearity. However, if deliberate isotopic labeling occurs, co-linearity is broken and the pure component spectra of the isotopomers are revealed.

165

166


are present, and if their concentrations change in the data set, then Eq. (2) is still potentially solvable. The limitations of NMR include the inability to directly distinquish enantiomers; the addition of a chiral shift reagent or chiral solvating agent is necessary. Also, polymeric materials, where a large distribution of molecules are present, must be treated as a lumped parameter. At the moment, the biggest limitation of NMR seems to be that that the chemical shifts are perhaps too sensitive to local environment. Consequently, the signals are too non-stationary in position, and Eq. (2) becomes intractable in many, but not all, cases. EPR is a powerful tool for unpaired electron/radical studies [55]. Therefore, EPR has the potential to fill a unique niche in the solution of Eq. (2). Only the subset of species that are radicals are accessible (dim Sobs II dim S). EPR is also promising because it produces localized signals rather than broad signals. UV–VIS spectroscopies and fluorescence These spectroscopies are relatively sensitive and provide good signal to noise spectra but they require the presence of a UV–VIS chromophore somewhere on the molecule. Consequently, only a subset of species are accessible (dim Sobs II dim S). The signals are typically broad, so the information content is low. Also, the UV and VIS excitation source conflicts with concerns expressed in Section 4.3.3.1. However, having said this, these spectroscopies still have potential use as an aid in supplementing information from other spectroscopies i. e. vibrational spectroscopies. The reasons are three-fold. First, many typical ligands used in homogeneous catalysis possess chromophores. Second, many organometallics are susceptible to metal–ligand charge transfer and are therefore colored. Third, ECD is more accessible than MIR-CD for the study of enantiomers. Spectrometer linearity and quantitative work All too frequently, researchers perform experiments beyond the means of the spectrometer used. The most common problem involves concentrations exceeding the linear response limits of the spectrometer. When this happens, solution of Eq. (2) becomes intractable. The most common mistake in infrared spectroscopy is when samples with absorbance greater than ca. 2–3 are measured with a DGTS detector [56]. At this point, the response of the spectrometer is no longer linear, and quantitative information is no longer accessible. The same type of situation exists in NMR, except it is perhaps a little more severe. The difficulties encountered when performing quantitative NMR work are well known [57, 58]. Most other types of spectrometers have similar limitations.

Groups of Experiments It is quite useful to structure the experiments used to acquire the spectroscopic data set A(nþk)eqn. Foremost, the researcher should keep in mind that the overall observation arises from the three individual phenomena: induction, product forma4.3.4.3


tion and deactivation. Therefore, experiments can usually be devised which are more appropriate for investigating each phenomenon. Chemicals Ideally, the source(s) of chemicals should be varied. Thus, commercial solvents, organic substrates and gases should be purchased from more than one source. When possible, both carefully purified and not-so-carefully purified solvents, substrates and gases should be used in different experiments. The purpose is to vary, as far as possible, the components present. The moles of solvents, substrates and gases present are often in considerable excess compared to the transition metal, and therefore the minor impurities can influence the chemistry considerably.1) Typically, less freedom is available with regard to variations in the sources of the transition metal complexes and ligands used. Induction, precursor transformation One set of experiments should be conducted at low substrate to precursor ratio e. g. ca. 1:1–10:1. The chronological order for the introduction of the solvent, soluble organic reagents, organometallic precursor, ligand, and promoter should be varied as far as possible. These experiments may well give rise to observably different pre-catalytic transformations. Catalytic product formation Homogeneous catalysis is, in the first instance, a synthetic approach. Therefore, a high turnover number is highly desirable for pragmatic and economic reasons. Nevertheless, experiments should be conducted at rather low substrate to precursor ratio as well as rather high substrate to precursor ratio. This type of approach is particularly well suited to semi-batch investigations. A wide variation in substrate to precursor ratio is readily performed with a minimum of resources. It may even be useful to provide perturbations in product, temperature and pressure. Again it may be desirable to vary the chronological order for the introduction of the solvent, soluble organic reagents, organometallic precursor, ligand, and promoter. Dramatic variations in catalytic activity may occur, going from very active systems to systems that possess more-or-less no activity and are therefore fully deactivated. In at least one case, the variability of catalytic activity as a function of start-up procedure has been studied by in situ FTIR. The term “path-dependent” catalysis has been used to emphasize the importance of the start-up procedure [60].

1) In our group, we typically vary the gases used.

In this manner, we have observed both Fe(CO)5 and Ni(CO)4 in some commercially available CO delivered in steel cylinders. With regard to substrates, the ultra-purification of

cyclohexene leads to clean rhodium hydroformylations, while cyclohexene with a trace of 1,3 cyclohexadiene leads to detectable poisoning of the system and formation of new rhodium complexes [59].

167

168


Deactivation Normal impurities for homogeneous catalysis, i. e. oxygen and water, should be deliberately introduced starting at levels much lower than 1 equivalent to metal. Impurities, which are common to the substrates used, should also be introduced. For example, if an alkene is the substrate, the corresponding diene(s) which are common impurities may also be introduced in small amounts. If phosphines are used as ligands, it may be useful to prepare some corresponding phosphine oxide and deliberately add this. The goal of this group of experiments is to better resolve spectra, which are probably already present but are perhaps deeply embedded in the noise, and which correspond to degraded complexes and ligands.

Range of Experiments Each group of experiments should be pre-planned. Perhaps multiple batch experiments, semi-batch or both can be conveniently used. The initial conditions for the batch runs and the steps of the semi-batch runs can be chosen randomly or in a judicious manner. The really important issue is that the mole ratios of any two reagents should never be held constant in a group; this constitutes a co-linearity induced by the experimentalist. It can be noted that each perturbation in a semibatch experiment limits the future accessible region of composition space so planning is advisable. At first glance, the experimentalist will be tempted to do the following in a single group of experiments; (i) vary substrate from 1 to 50 mol % or more, (ii) start with a high concentration of substrate and achieve very high conversions of substrate and (iii) cover large ranges in temperature e.g 20–100 hC. This is not advisable due to the non-stationary characteristics of the pure component spectra. If really significant liquid phase concentration changes occur in a group, i. e. ca. 10–15 % and more, or if really significant temperature changes occur, i. e. ca. 35 hC or more, the data set may not be invertible and hence Eq. (2) cannot be realized. For a further discussion of this non-stationary signal issue as it relates to Eq. (2) the reader is referred to Ref. [39]. The experimentalist can, and should, cover the entire range of composition, temperature and pressure desired, but in grouped sets of experiments. The overall range is bounded by the composition, temperature and pressure limitations of the system in Figure 4.1. For example, there will be lower and upper limits for temperature due to the cooling and heating capacity. The upper limit for pressure may be fixed by the CSTR, pump or spectroscopic cells. The composition range is frequently limited by the materials of construction, particularly the O-ring seals. 4.3.4.4

4.3.5

Well-Posedness and Ill-Posedness

A solution to Eq. (2) can be considered, if and only if, the physical science and engineering issues surrounding the experiments are correctly addressed in advance. Good experimental decisions are prerequisites to meaningful analysis. The cur-

4.4 Data Pre-processing

rently known limitations inherent in various facets of the experimental design were emphasized in Section 4.3 in order that other catalytic groups can more readily and successfully implement this chapter’s signal processing concepts in the future. The CSTR-recycle in Figure 4.1 has limitations and many are associated with mass transfer and time-scale issues. A CSTR contacting pattern is not perfect as input to Eq. (2). Each spectrometer associated with Figure 4.1 has its limitations, and many are associated with the class of species detectable and the sensitivity. No single spectrometer is the perfect input to Eq. (2). Each experimental plan has limitations and many of these are associated with accessible compositions, temperatures and pressures. In addition, there is the issue of having an overdetermined system as remarked previously. Taken together, we arrive at the conclusion: Given adequate consideration of the experimental system, the spectroscopies used, and the experimental design, a good but not perfect data set A(nþk)eqn can be measured. Although the data set A(nþk)eqn is necessarily incomplete, (like any truly difficult physical inverse problem), it provides sufficient input for a first approximate solution to Eq. (2).

4.4

Data Pre-processing

Data pre-processing is commonly used prior to the solution of an inverse problem. In this section, five data pre-processing procedures are discussed. None of these procedures are absolutely necessary to obtain a solution to Eq. (2). However, all have been implemented, at one time or another by our group, in order to improve the inversion of in situ spectroscopic measurements of catalytic systems.1) 4.4.1

Data Filtering and Outliers

The filtering of 1D signals is commonly achieved using fft, spline functions, wavelets etc. [61]. The goal of filtering is to increase signal to noise. However, in spectroscopy, loss of resolution and spectral distortion can arise due to over-filtering. Each batch experiment or each step of a semi-batch experiment results in a data array Ak xn. This matrix is often rather smooth in the time direction, especially if rather fast spectra are taken compared to the evolving chemistry. If one follows any particular channel of data in the time direction, it may slowly increase or de1) All our group’s algorithms have been imple-

mented in MatLabTM on PCs or workstations. MatLab is a very convenient platform for importing, manipulating and exporting spectroscopic data. The MatLab library of functions is

very extensive and many frequent operations like filtering, are one-line commands. We are slowly releasing most of our programs, with full documentation after testing, on our freeware site.

169

170


crease, or go through a maximum or minimum. One can filter in the wavenumber direction or time direction. The 2D property can be used to increase filtering efficiently [62]. We have filtered FTIR data from the homogeneous catalyzed rhodium hydroformylation of alkenes using a variety of 1D and 2D filters. On blocks of 100–1000 spectra, the 1D filters i. e. SG, fft, cubic spline, can reduce noise by ca. 10–50 %, but the 2D filters, i. e. 2D fft, can reduce the noise level even further, to ca. 85 %þ [63]. The procedure for each block of spectroscopic data can be viewed as Eq. (7)

Akqn p Afilt kqn

(7)

4.4.2

Solvent and Reagent Pure Component Spectra

The arrays of preliminary, pre-catalytic spectral data Anqn can be processed in order to get pure component spectra which are valid in the vicinity of the initial reaction conditions. For infrared spectroscopy, this requires proper subtraction of the background moisture and carbon dioxide to get a good approximation of solvent (in the cell). These can then be used to obtain a good approximation of solute A, followed by solute B etc. This process can be done manually by trial and error,1) or it can be automated for optimal subtraction. The output is a set of n reference spectra. ref Aexp nqn p Anqn

(8)

An automated and optimal subtraction procedure was previously developed for Eq. (8). The algorithm takes one spectrum after another and optimally subtracts previous signals. The criterion used is minimization of the signal entropy (see Section 4.5.2). The recursive equation used appears in Eq. (9), where Aexp is an experimental spectrum, Aref is a set of reference spectra, and x are scalar coefficients. The values of x are easily determined. exp ref Aref n ¼ An – Sxn–1An–1

(9)

The automatic procedure for reference spectra generation was first demonstrated for the start-up of a homogeneous catalyzed rhodium hydroformylation of cyclooctene using Rh4(CO)12 as precursor, n-hexane as solvent and FTIR as the in situ spectroscopy at 298 K [63]. The first n spectra were: (i) empty spectrometer compartment (background), (ii) n-hexane at 0.2 MPa in a high pressure thermostatically controlled cell fitted with CaF2 windows (iii) system equilibrated with 2.0 MPa CO, (iv) system upon addition of cyclo-octene, and (v) system upon addition of Rh4(CO)12. The n¼1 reference spectrum, which contained atmospheric 1) Many commercial spectrometers, particularly

IR spectrometers, have software capability for automated subtraction of a pair of experi-

mental spectra. The technology behind these pair-wise subtractions appears in most cases to be proprietary


Figure 4.2 The pure component reference spectra of hexane (a), dissolved carbon monoxide (b), dissolved cyclooctene (c) and dissolved Rh4(CO)12 obtained after entropy minimization. The maximal absorbance in each case is ca. 0.15 AU, 0.60 AU, 0.15 AU and 0.4 AU, respectively. (L. Chen, M. Garland, Appl. Spectrosc., 2002, 56, 1422–1428.)

moisture and CO2 is a special case, it was partitioned into two decoupled individual spectra. The remaining reference spectra for hexane, dissolved CO, dissolved cyclooctene and dissolved Rh4(CO)12 were readily generated by the recursive formula Eq. (9). These reference spectra are shown in Figure 4.2. Although these reference spectra look really good, they are not truly representative of the absorptivities at the reaction conditions. This issue will be found to be critical later. 4.4.3

Pre-conditioning

The time-series array of catalytic spectral data Akqn can be processed in order to remove the reference spectra, and thereby simplify the reaction spectra. For infrared spectroscopy, this requires proper subtraction of the background moisture and carbon dioxide, the solvent, and some of the reagents. This process can be done manually by trial and error, or it can automated for optimal subtraction. The output is a set of k simplified reaction spectra, Eq. (10).

171

172


Aexpkqn p Aprekqn

(10)

An automated and optimal subtraction procedure was previous developed for Eq. (10). The algorithm takes one reaction spectrum after another and optimally subtracts reference signals. Again, the criterion used is minimization of the signal entropy (see Section 4.5.2). The recursive equation used appears in Eq. (11), where again Aexp is an experimental spectrum, Aref is a reference spectrum, and y are scalar coefficients. The values of y are easily determined. exp Aref – SyiAref k ¼ Ak i

(11)

The automatic procedure for time-series reference spectra generation was first demonstrated for the homogeneous catalyzed rhodium hydroformylation of cyclooctene using Rh4(CO)12 as precursor, n-hexane as solvent and FTIR as the in situ spectroscopy at 298 K [64]. Upon addition of hydrogen to the system, hydroformylation is initiated. A typical reaction spectrum (k¼7) and the pre-conditioned

Figure 4.3 The seventh sequential experimental reaction spectrum Ak¼7exp (a) and the seventh preconditioned reaction spectrum obtained by entropy minimization Ak¼7pre (b). The maximal absorbance in each case is ca. 1.0 AU and 0.4 AU, respectively. (L. Chen, M. Garland, Appl. Spectrosc., 2002, 56, 1422–1428.)


spectrum after optimal removal of hexane, CO, cyclo-octene, Rh4(CO)12 signals is shown in Figure 4.3. The quality of the pre-conditioning is rather good. The product aldehyde is easily seen at ca. 1732 cm–1, the organometallic intermediate RC(O)Rh(CO)4 [65] at ca. 1703, 2020, 2038, 2065, 2112 cm–1 and the impurity Rh6(CO)16 at 2071 cm–1 [66]. The nominal rhodium loading in this experiment is only 100 ppm. The factors y enable the researcher to easily extract the mole fraction concentrations of cyclooctene and Rh4(CO)12 as a function of time. The imperfect baseline is due in part to the reference spectra, which were not measured at exactly the reaction conditions. The automated procedure outlined by Eq. (11) has been executed for data sets as large as ca. k¼1000. 4.4.4

Track Finding

The number of bands present in a spectral window and their centers of symmetry are pre-requisites to other signal processing procedures i. e. curve fitting. For in situ spectroscopic reaction studies, a set of tracks can be assigned which specify the centers of symmetry for all the bands. Since the bands move as a function of composition, the tracks in a matrix Aexpkqn or Aprekqn drift. The second derivative of a spectrum has been frequently used to identify individual bands underlying composite spectral features [67]. The local minima of the second derivative provide excellent approximations of band center positions. The presence of spectroscopic noise is the primary difficulty encountered with this approach. Noise interferes with the enumeration of bands as well as the accurate identification of the band center positions. A procedure has been developed for Aexpkqn (Aprekqn) which starts with second derivative information, provides a sequence of more and more refined maps, and results in a final estimate of the 2D tracks [68]. The input to the procedure can be either Aexpkqn or Afiltkqn, although the latter is preferred. The second derivative of each spectrum is taken and all local minima are identified. This results in a map M2. This map is subsequently sorted to give Msort and filtered to give a map Mfilt. The filtering procedure employs a few criteria in order to eliminate erroneous peak center positions. Finally, a polynomial fit is employed for data reconciliation e. g. missing peak center positions (interpolation) and to extrapolate to k¼1 or k, resulting in a map Mpoly which contains all the peak center positions up to a given user-specified level of tolerance. M2 p Msort p Mfilt p Mpoly

(12)

The procedure has been successfully tested on synthetic data and applied to the homogeneous rhodium catalyzed hydroformylation reaction, where the 2D tracks for the peak center positions of both organometallic intermediates and products were identified. If the data Aexpkqn is collected in a very small region of composition space, the peak center positions are virtually constant. If the data Aexpkqn are col-

173

174


lected in a large region of composition space, the peak center positions may drift significantly. 4.4.5

Curve Fitting

Curve fitting is an important tool for obtaining band shape parameters and integrated areas. Spectroscopic bands are typically modeled as Lorenzian distributions in one extreme and Gaussian distributions in the other extreme [69]. Since many observable spectroscopic features lie in between, often due to instrument induced signal convolution, distributions such as the Voight and Pearson VII have been developed [70]. Many reviews of curve fitting procedures can be found in the literature [71]. It should be noted that for many spectroscopies, the integrated peak area rather than peak height remains proportional to the moles of solute present, even over large regions of composition space. This quantitative feature of integrated areas has a precise analytical form in vibrational spectroscopy [72, 73]. The Pearson VII model contains four adjustable parameters and is particularly well suited for the curve fitting of large spectral windows containing numerous spectral features. The adjustable parameters a, p, q and no correspond to the amplitude, line width, shape factor and band center respectively. As q p1 the band reduces to a Lorenzian distribution and as q approaches ca. 50, a more-or-less Gaussian distribution is obtained. If there are b bands in a data set Aexpkqn, and each band has an integrated area Aint, then the Pearson model for each experimental spectrum can be represented as Eq. (13) where the parameters p and q are embedded in the matrix G [74].

Aintbq1 ¼ Gbqb abq1

(13)

The peak map Mpoly from Section 4.4.4 helps to jump start the calculations since it fixes the value of b and provides an excellent first approximation of the band center positions no. With the first approximation of no, the three remaining unknowns a, p, q can be optimized. At this point, the value of each no can be relaxed within a small tolerance, and then all a, p, q and no can be re-determined. The execution of Eq. (13) for a full series of k sequential spectra in Aexpkqn provides not only a set of integrated areas, but also a precise understanding of how the band shape changes as a function of composition. This procedure has been implemented for the homogeneous rhodium catalyzed hydroformylation of cyclo-octene. An example of the curve fitting for both the organometallic species present and the organic product is shown in Figure 4.4.1) 1) Briefly, it can be mentioned that Eq. (2) can be

written in an integral rather than decadic form, and that one can solve for integrated

pure component spectra. Such an approach has been realized [63].


Figure 4.4 (a) A comparison of the 10th curve fitted spectrum (dotted curve) with the 10th experimental spectrum (solid curve) after re-optimization of the band parameters. (b) Expanded view of the metal-carbonyl mid-infrared region. (L. Chen, M. Garland, Appl. Spectrosc., 2003, 57, 331–337.)

175

176


4.5

Spectral Reconstruction

In this section, the historical context for Eq. (2) is formulated, the concept of signal entropy is discussed, and a solution for Eq. (2) is shown. The repeated application of Eq. (2) to organometallic and homogeneous catalytic systems is shown. 4.5.1

Historical Context

Modern pure component spectral reconstruction from multi-component spectroscopic data has its origins in a 1973 paper by Lawton and Sylvester [75] in which a known two component system was studied, and the pure component spectra were determined. The term curve resolution became intimately associated with pure component spectral reconstruction in the chemometrics literature [76]. A number of more sophisticated curve resolution procedures for multi-component mixtures have been developed over the years. The most well-known of these procedures are: IPCA [77], SIMPLISMA [78] and OPA-ALS [79]. All rely on an a priori estimate of the number of observable species present, most if not all rely on the concept of a pure wavelength i. e. a spectral window where only one component is present, and all have difficulties with trace components i. e. components whose spectral features are embedded in the signal noise. In practice, both the prior estimate of the number of observable species present and the need for pure wavelengths are very serious restrictions, especially for catalytic systems where so many species are simultaneously present.1) 4.5.2

Entropy Minimization

To a significant extent, the theoretical basis of modern communication theory arose from the work of Claude Shannon at Bell Labs. [80]. In these seminal works, the concept of the information entropy associated with an arbitrary signal arose. In 1981, Watanabe realised the close association between entropy minimization and pattern recognition [81]. An association between entropy minimization and the principle of simplicity is also recognized [82]. The basic mathematical form of signal 1) PCA techniques have been applied to various

data sets. From the open literature, I have been unable to find reference to their application to homogeneous catalytic systems. Having said that, it is important to note that the new ReactIR system by Mettler Toledo contains a program ConcIRT. It is a proprie-

tory tool which attempts to generate pure component spectra. Upon inspection, it appears to belong to the PCA class of tools. The ReactIR system is an ATR type spectroscopy that has been used in about one dozen publications on homogeneously catalyzed systems.

4.5 Spectral Reconstruction

entropy is given by Eq. (14) where H is the signal entropy, and p is a probability distribution. H ¼ p log p

(14)

In 1983, Sasaki et al. obtained rough first approximations of the mid-infrared spectra of o-xylene, p-xylene and m-xylene from multi-component mixtures using entropy minimization [83–85] However, in order to do so, an a priori estimate of the number S of observable species present was again needed. The basic idea behind the approach was (i) the determination of the basis functions/eigenvectors Vsqn associated with the data (three solutions were prepared) and (ii) the transformation of basis vectors into pure component spectral estimates by determining the elements of a transformation matrix Tsqs. The simplex optimization method was used to optimize the nine elements of T3q3 to achieve entropy minimization, and the normalized second derivative of the spectra was used as a measure of the probability distribution.

asqn ¼ Tsqs Vsqn

(15)

At least two complications are immediately apparent. First, the separation of highly overlapping pure component spectra may be problematic. Second, for systems with more than three components, many local solutions (minima) should exist for Eq. (15).

Organometallics It has been shown that the first problem can be entirely overcome by changing the form of the term p [86], and by taking large spectral windows with data intervals (resolution) much finer than the features present [87]. The latter experimental example involved eight physical multi-component solutions containing two metal carbonyls Co2(CO)8 and Co4(CO)12 in n-hexane. MIR measurements were made at 0.2 cm–1 intervals over the range 1800–2200 cm–1. Figure 4.5 shows the first two basis functions /right singular vectors from VT8q2251 obtained by taking the singular value decomposition of the pre-conditioned data Apre8q2251.1) The first two vectors are taken and the 2x2 problem associated with Eq. (15) was solved. The resulting MIR spectral reconstructions are very good (Figure 4. 6), even though six vectors of information were discarded prior to reconstruction. 4.5.2.1

1) The singular value decomposition (SVD) is the

preferred method for determining basis functions for rectangular matrices. The SVD formula is given by M¼USVT where M is any data matrix, U contains the left singular vectors, S is the matrix of singular values and the matrix VT contains the right singular vectors

[88]. In the case where the data matrix M is square, the SVD reduces to the classic eigenvector problem. It is important to note that the vectors in VT are ordered in descending order of their contribution to the total signal variance of the system.

177

178


Figure 4.5 The two significant right singular vectors obtained after SVD of the matrix A. Note that all vectors in this figure have maximal absorbance of the order of 1 AU. (Y Pan, L. Susithra, M. Garland, J. Chemomet., 2000, 14, 63–77.)

Figure 4.6 Pure component spectral estimates of Co2(CO)8 and Co4(CO)12 obtained from the local minimum of the fourth derivative, unweighted, full-spectrum minimizations. (Y Pan, L. Susithra, M. Garland, J, Chemomet., 2000, 14, 63–77.)


Figure 4.7 The next six right singular vectors, corresponding to noise, obtained after SVD of the matrix A. Note that the vectors arises from the small shift in band maxima and band shapes in the pure component spectra when solutions of different composition are prepared, and that all vectors in this figure have very low absorbance (ca. 10 –2 AU or less). (Y Pan, L. Susithra, M. Garland, J. Chemomet., 2000, 14, 63–77.)

The remaining 6 vectors contained some localized signals. From Figure 4.7 it is rather straightforward to show that the pure component spectra of each solute are non-stationary i. e. the position and band shapes change.1) Indeed, real and localized signal repeat throughout the vectors. By discarding these vectors, real signal is lost.

Catalysis The second problem can be overcome by using a stochastic search. Thus Corana’s simulating annealing SA [89], a sophisticated global search method arising from the Metropolis algorithm [90] was employed. A synthetic seven species problem was constructed and the elements in T7 X7 were correctly determined using SA. The recovered spectra are essentially identical to the synthetic 7 pure component spectra. [91] The homogeneous rhodium catalyzed hydroformylation of isoprene was the first application of SA to spectral reconstruction of a real catalytic system [92]. The MIR 4.5.2.2

1) From linear algebra and systems theory, it is

well known that a linear system with n degrees of freedom has only n basis vectors. If the signals of Co2(CO)8 and Co4(CO)12 were

stationary, the system would be linear, and there would be only two right singular vectors containing useful information. This is clearly not the case.

179

180


data were pre-conditioned according to Section 4.4.3. A total of six solute pure component spectra were recovered after determining the global optimal solution for 36 unknowns in T6q6. Two of the recovered pure component spectra were outstanding matches for the catalyst precursor Rh4(CO)12 and the organic substrate isoprene. At this point, the applicability of Eq. (15) appears exhausted. Although some problems are solvable, at least three significant problems remain, namely (i) an a priori estimate of the number S of observable species present is still needed; (ii) for systems with large S, there will be severe limitations on the simultaneous optimization of SqS unknowns and (iii) recovery of trace component pure component spectra appears out-of-reach. 4.5.3

Band-target Entropy Minimization

The concept of band-target entropy minimization (BTEM) arose from the desire to retain entropy minimization as the core concept for spectral reconstruction, but to modify Eq. (15) into a more tractable form. The key to all the three aforementioned outstanding problems proved to be (i) good experimental design such that kiiS was achieved; (ii) choice of j basis vectors rather than s vectors, where jiis and (iii) user targeting1) of interesting localized spectral features to achieve one-at-a-time spectral reconstructions. The primary equation for BTEM is Eq. (16). Eq. (16) is executed repeatedly and exhaustively on all the localized signals in the data set until there are no more pure component spectra to recover. By using jiis vectors, little real signal is lost, and k-j “noise” vectors are discarded, leading to significant signal to noise enhancement.

a1qn ¼ T1qj VTjqn

(16)

The first application of BTEM and Eq. (16) was to very dilute solutions containing the organometallic clusters Rh4(CO)12 and Rh6(CO)16 in n-hexane [93]. Twelve solutions were prepared and MIR spectra recorded. The data intervals were 0.2 cm–1 (compared to the 3 cm–1 line widths), and 2251 channels of data were used. The data A12q2251 was preconditioned according to Section 4.4.3 to remove the solvent and atmospheric moisture and CO2 absorbance. SVD was performed to obtain VT12q2251 and local maxima and minima were exhaustively targeted. Although the maximum absorbance of the data set was 0.04 AU, outstanding pure component spectra of both dissolved Rh4(CO)12 and Rh6(CO)16 were recovered. The vector inner products of these estimates compared to authentic references were considerably greater than 0.998, indicating superb reconstruction. In addition, the broad absorbance signature of non-soluble and suspended colloidal Rh6(CO)16 was recov1) Targeting is the term we have adopted

whereby the user specifies a small spectral window, i. e. e 1 cm–1, in the vicinity of a local maximum or minimum. This constrains the

optimization to retain a feature at this position. The program then reconstructs all f(x) to the left and right of the chosen point.


ered! Since the pure component spectra are mean representations, their integrated areas are slightly greater than the reference spectra. This contribution also explains in detail the insensitivity of the final solutions with regards to the choice of j. The primary issue is that j should be chosen much greater than s. This is easily done by taking all vectors which have any localized signal and do not appear to be noise.

Multiple-run, Preconditioned, Monometallic Catalytic Data The Rh4(CO)12 initiated homogeneous catalyzed hydroformylation of 3,3-dimethyl1-butene was conducted at 298 K [94]. A set of e¼3 experiments with kz63 spectra and n¼4751 MIR spectral channels was used. The spectra from the 3 hydroformylation experiments were pre-conditioned and assembled into one matrix A188q4751. This data matrix was then subjected to singular-value decomposition to obtain the right singular vectors VT188q4751. The first ca. 50 right singular vectors contained localized signals, the rest were essentially Gaussian distributed white noise. Thus, j was set to 50. The first six right singular vectors are shown in Figure 4.8. The local maxima and minima were exhaustively targeted. In all, seven pure component spectra were recovered including: the precursor Rh4(CO)12, the omnipresent cluster Rh6(CO)16, the intermediate RC(O)Rh(CO)4, the alkene, the 2 regioisomeric aldehydes, and a new species Rh4(s-CO)12. The mean rhodium loading in 4.5.3.1

Figure 4.8 The first six right singular vectors of the VT matrix: (1) first vector; (2) second vector; (3) third vector; (4) fourth vector; (5) fifth vector; (6) sixth vector. (C. Li, E. Widjaja, M. Garland, J. Catal., 2003, 213, 126–134.)

181

182


Figure 4.9 The reconstructed pure component spectra of major components: (1) 3,3-dimethylbut-1-ene, (2) Rh4(s-CO)9(m-CO)3, (3) RC(O)Rh(CO)4, (4) 4,4-dimethylpentanal. (C. Li, E. Widjaja, M. Garland, J. Catal., 2003, 213, 126–134.)

these experiments was ca. 75 ppm and the mean concentration of the most dilute organometallic species was ca. 1 ppm. The four most prominent species are shown in Figure 4.9. These species were generated by targeted features denoted as 1–4 in Figure 4.9. The progression from the target (#3 in Figure 4.8), to the realization of Rh4(CO)12 in Figure 4.9 can be followed. Snapshots of the SA optimization for recovery of the pure component spectrum of Rh4(CO)12 are shown in Figure 4.10. Further refinements are made to the estimate as the optimization proceeds. For j¼50 and k¼5000, a pure component spectral recovery can require a few hours cpu time on a good workstation. SA typically encounters and evaluates 104 –107 local minima before the global minimum is achieved.

Multiple-run, Preconditioned, Bimetallic Catalytic Data The same type of procedure was applied to the Rh4(CO)12 / HMn(CO)5 initiated homogeneous catalyzed hydroformylation of 3,3-dimethyl-1-butene at 298 K [94]. A set of e¼21 experiments with kz26 spectra and n¼4751 MIR spectral channels were used. The spectra from the 21 hydroformylation experiments were preconditioned and assembled into one matrix A713q2951. This data matrix was then 4.5.3.2


Figure 4.10 Spectral reconstruction progression of Rh4 (s-CO)9(m-CO)3 during entropy minimization process by simulated annealing optimization; (1) 1st temperature with entropy value of resolved spectrum ¼ 6.90; (2) 10th temperature with entropy value ¼ 6.64; (3) 20th temperature with entropy value ¼ 5.86; (4) final temperature with entropy value ¼ 5.66. (C. Li, E. Widjaja, M. Garland, J. Catal., 2003, 213, 126–134.)

subjected to singular-value decomposition to obtain the right singular vectors VT713q2951. The first ca. 50 right singular vectors contained localized signals, the rest were essentially Gaussian distributed noise. Thus, j was set to 50. Local maxima and minima were exhaustively targeted. In all, seven pure component spectra were recovered including: the precursor Rh4(CO)12, the omnipresent cluster Rh6(CO)16, the intermediate RC(O)Rh(CO)4, the alkene, the primary aldehyde, HMn(CO)5 and Mn2(CO)10. The most dilute species Rh6(CO)16 had a mean concentration of ca. 1.5 ppm. Figure 4.11 shows a few of the right singular vectors including the 713th which is essentially pure white noise. By fixing a small window of wavenumbers at each extremum in Figure 4.11, the user effectively targets a species. This consequently constrains the evolution of the SA optimization. By targeting a feature, the feature is retained and the simplest continuation to the right and left are realized. A pure component spectral estimate is obtained. The spectroscopic and kinetic data from this reaction indicated the existence of a long sought catalytic reaction topology, bimetallic catalytic binuclear elimination. The kinetic data provided a linear–bilinear form in organometallics [95]. One term represented the classic unicyclic rhodium catalyzed hydroformylation and the other represented the attack of manganese hydride carbonyl on an acyl rhodium tetracarbonyl species. A representation of the interconnected topology is shown in Figure 4.12.

183

184


Figure 4.11 A singular value decomposition of the preconditioned in situ spectroscopic data showing the 1st, 4th, 5th, 7th significant vectors and the 713th vector. The marked extrema are those which were used to recover the organometallic pure component spectra as well as alkene and aldehyde by BTEM. Atmospheric moisture and CO2, hexane, and dissolved CO were removed from the experimental data during preconditioning. (C. Li, E. Widjaja, M. Garland, J. Am. Chem. Soc., 2003, 125, 5540–5548.)

Semi-batch, Non-preconditioned, Monometallic Catalytic Data The need for preconditioning is not only somewhat bothersome, but it also has the potential of introducing minor artifacts since the reference spectra are not exact for each reaction condition. Thus, the direct application of Eq. (2) without preconditioning was undertaken. In addition, instead of performing multiple reaction runs, a semi-batch approach was adopted, whereby a reaction is started and then perturbations are performed throughout the duration. The semi-batch approach is potentially much more frugal with experimental resources. A similar semi-batch approach was taken with the unmodified rhodium catalyzed in order to confirm its utility [96]. The Rh4(CO)12 initiated homogeneous catalyzed hydroformylation of 3,3-dimethyl-1-butene was conducted at 298 K. The MIR spectra were assembled into one matrix A250q4751. This data matrix was then subjected to singular-value decomposition to obtain the right singular vectors VT250q4751. The first ca. 50 right singular vectors contained localized signals, the 4.5.3.3


Figure 4.12 The proposed reaction topology for the simultaneous interconnected unicyclic Rh and bimetallic Rh–Mn CBER hydroformylation reactions. (C. Li, E. Widjaja, M. Garland, J. Am. Chem. Soc., 2003, 125, 5540–5548).

rest were essentially Gaussian distributed noise. Thus, j was set to 50. Local maxima and minima were exhaustively targeted. In all, 10 pure component spectra were recovered including: atmospheric moisture, atmospheric CO2, n-hexane, dissolved CO, the precursor Rh4(CO)12, the omnipresent cluster Rh6(CO)16, the intermediate RC(O)Rh(CO)4, the alkene, the primary aldehyde, and Rh4(s-CO)12. The mean rhodium loading in these experiments was ca. 75 ppm and the mean concentration of the most dilute organometallic species was ca. 1 ppm. Figure 4.13 shows the spectral reconstructions for the organic and organometallic solutes. The extreme overlap of spectral bands is readily apparent. It is instructive to integrate all the pure component signals and compare this to the integrated reaction spectra. This provides a useful indication of the amount of signal recovered. Table 4.1 shows this result. The sum of the individual contributions is ca. 103 % of the integrated reaction spectra. This result simple re-affirms a

185

186


Figure 4.13 The reconstructed pure component spectra of minor components. (e) 4,4-dimethylpentanal (f) 3,3-dimethylbut-1-ene (g) Rh4(s-CO)9(m-CO)3 (h) RC(O)Rh(CO)4 (i) Rh4(s-CO)12 (j) Rh6(CO)16. (E. Widjaja, C. Li, M. Garland, Organometallics, 2002, 21, 1991– 1997).

Percentage of reconstructed integrated absorbance of each component compared to the total original experimental data. (E. Widjaja, C. Li, M. Garland, Organometallics, 2002, 21, 1991–1997.)

Table 4.1

Component n-C6H14 CO CO2 H2O 44DMP 33DMB Rh4(s-CO)9(m-CO)3 RCORh(CO)4 Rh4(s-CO)12 Rh6(CO)16 Total

Reconstructed integrated absorbance compared to the total original experimental data, % 61.392 26.819 0.8379 4.7850 3.1008 2.2386 1.2554 1.8870 0.3247 0.2379 102.88


previous issue: each recovered pure component spectrum is a mean representation. It is a convolution of all the states for that species within the group of spectroscopic observations. The bands are slightly broader than any individual spectroscopic observation and hence the integrated area is also a little greater.

Semi-batch, Non-preconditioned Data: HRh(CO)4 The simple metal carbonyl hydride species HRh(CO)4 has been sought by many researchers including Hieber [97], Chini [98], Vidal [99] and Whyman [100]. Vidal was able to obtain evidence for new MIR bands when Rh4(CO)12 was subjected to ca. 1400 bar CO/H2 pressure, and Whyman was able to corroborate this finding at ca. 400 bar CO/H2 pressure. Solutions of Rh4(CO)12 in n-hexane were subjected to mixtures of CO/H2 and CO/D2 in a semi-batch manner by perturbing the mole fractions of n-hexane, Rh4(CO)12, CO, H2, and D2 at ca. 298 K [101] .The maximum pressure used was 50 bar. In situ MIR measurements were collected into matrices A473q4751 and A200q4751. To the unaided eye, no new spectral features can be seen. SVD was performed and local extrema targeted. Eight easily identifiable pure component spectra were recovered including; atmospheric moisture, atmospheric CO2, n-hexane, dissolved CO, Rh4(CO)12, Rh6(CO)16, Rh4(s-CO)12, and Rh2(CO)8 [102,103]. In addition, two new spectra were obtained corresponding to HRh(CO)4 and DRh(CO)4 with signal to noise ratios of ca. 50:1, in spite of the fact that their signals constitute only ca. 0.15 % of the experimental data. The equilibrium constant was also obtained. Figure 4.14 shows the results. 4.5.3.4

4.5.4

Additional Notes on BTEM

In Section 4.5.3, discussion was restricted to the application of BTEM to organometallic and homogeneous catalytic systems using MIR spectroscopy. However, it should also be noted that BTEM has been successfully applied to other problems in the chemical sciences using FTIR [104], RAMAN, FTIR and RAMAN [105, 106], as well as MS [107, 108]. A one-to-one comparison between BTEM and SIMPLISMA, OPA-ALS, has appeared [109].

187

188


Figure 4.14 The reconstructed pure-component spectra of the organometallic carbonyl hydrides: (1) HCo(CO)4, (2) HRh(CO)4, (3) DRh(CO)4. (C. Li, E. Widjaja, W. Chew, M. Garland, Angew. Chem. Int. Ed. Engl., 2002, 41, 3785–3789.)

4.6

Conclusions and Future Directions

This chapter has focused on formulating the context for Eq. (2) and demonstrating that a solution exists. The results to date show that a genuine and robust solution to Eq. (2) is available in the absence of a priori information. The solution, BTEM, is a very powerful tool in the investigation of homogeneous catalytic systems using in situ spectroscopies. Future directions can be divided into experimental and numerical. The two most obvious experimental directions are (i) extensions to CD spectroscopies so that comparable system identification of stereo-selective systems can be realized and (ii) extension to NMR since it is the most widespread analytical technique in the chemical sciences. Efforts in both directions are currently underway in our laboratory.

4.6 Conclusions and Future Directions

Solution of Eq. (2) is only part of the overall problem in mechanistic studies. Indeed, Eq. (2) represents only the first of many sequential inverse problems which we commonly lump together and which represent the overall inverse problem in chemical kinetics [110–112]. The next problem is the determination of the moles of all species, the elemental stoichiometries of species, the number of observable reactions, and the balanced stoichiometries for all observable reactions. This fixes the algebraic structure of the system. Reaction kinetic analysis naturally follows, including in situ evaluation of TOF [113]. The overall set of inverse problems can be represented by Figure 4.15. The development of the latter steps is still in progress. Parts of the algebraic and kinetic analyses can be found in the BTEM references contained in Section 5.4, Refs. [114,115], and three Ph. D. theses [116–118].

A ke × ν ø S obs , a s × ν ø N kexs , R obs , ξ ke × r , ν r × s ø Classic Rate Problem Figure 4.15 Representation of the three primary steps for the generic inverse problem in chemical kinetics including homogeneous catalysis. In situ spectroscopic data is represented by Akeqn. The inverse spectroscopic problem (Eq. (2)), which is the focus of this chapter, is represented by Sobs, asqn. The inverse problem associated with stoichiometries and reaction topology is represented by Nkexs, Robs, j keqr, n rqs for moles, reactions, extents of reaction and reaction stoichiometries. Having determined the algebraic structure of the system, the differential structure i. e. the classic rate problem, follows directly.

Acknowledgment

The development of the techniques reported in this chapter represents an arduous 15 year undertaking, requiring the continuous support of two academic institutions and colleagues as well as the dedicated work of many researchers. Accordingly, I wish to thank (i) the Eidgenossische Technische Hochschule (ETH-Zurich) for the funding and initial freedom which gave rise to many ideas represented by these developments, (ii) Prof Dr Piero Pino (Institute for Polymers), Prof Dr Luigi Venanzi (Inorganic Chemistry) and Prof Dr D. W.T Rippin (Chemical Engineering) for providing me with many research opportunities, important discussions and persistent encouragement and (iii) my students at ETH, particularly Romeo Volken, Christain Fyhr and Erik Visser. I also wish to thank the National University of Singapore (NUS) for funding and my present and past students at NUS, particularly Drs Chen Li, Effendi Widjaja, Li Chuanzhao and Chew Wee. It was the dedicated effort of this latter group of individuals which led to BTEM.

189

190


References [1] Handbook for Digital Signal Processing,

[2] [3]

[4]

[5]

[6]

[7] [8] [9]

[10]

[11]

[12]

[13] [14]

[15] [16] [17]

Eds. S. K. Mitra, J. F. Kaiser, Wiley, New York 1993. P. C. Sabatier, Applied Inverse Problems, Springer, Berlin 1978. Inverse Problems: Principles and Applications in Geophysics, Technology and Medicine, Eds. G. Anger, R. Gorenflo, H. Jochmann, H. Moritz, W. Weber, Akademie Verlag, Berlin 1993. A. Sinvhal, H. Sinvhal, Seismic Modelling and Pattern Recognition in Oil Exploration, Kluwer, Dordrecht 1992. Basic Methods of Tomography and Inverse Problems, Eds. G.T Hermann, H. K. Tuy, K. J. P. C. Langenberg, P. C. Sabatier, IOP Publishing, Bristol 1987. J. D. Dunitz, X-Ray Analysis and the Structure of Organic Molecules, Cornell University Press, Ithaca 1979. P. C. Sabatier, Inverse Problems, 1985, 1, i–vi. J. Hadamard, Princeton University Bulletin, 1902, 13, 49–52. A. B. Bakushinski, A. U. Goncharskii, Ill-posed Problems: Theory and Application. Kluwer, Dordrecht 1994. M. M. Lavrentiev, Some Improperly Posed Problems of Mathematical Physics, Springer, Berlin 1967. E. A. Robinson, S.Treitel, Geophysical Signal Processing, Prentice Hall, Englewood Cliffs 1980. J. Radon, Ber. Verh. Saechs. Acad. Wis. Leipzip. Math-Phys. Kl. 1917, 69, 262–277. S. Helgason, The Radon Transform, Birkhauser, Boston 1980. H. Lipson, C. A. Taylor, Fourier Transforms and X-Ray Diffraction, G. Bell and Sons, London 1958. A. J. M. Spencer, Continuum Mechanics, Longman, London 1980. K. Yamamoto, H. Ishida, Vibr. Spectrosc., 1994, 8, 1–36. Phase-Transfer Catalysis: Mechanisms and Syntheses, Ed. M. Halpern, American Chemical Society,

[18]

[19] [20]

[21]

[22]

[23]

[24]

[25]

[26] [27] [28]

[29]

[30] [31]

[32]

Washington, D. C. 1996; Handbook of Phase-Transfer Catalysis, Eds. Y. Sasson, R. Neumann, Blackie Academic, London 1997. R. V. Chaudhari, B. M. Bhanage, R. M. Deshpande, H. Delmas, Nature, 1995, 373, 501–503. Interfacial Catalysis, Ed. A. G. Volkov, Dekker, New York 2003. W. L. McCabe, J. C. Smith, P. Harriott, Unit Operations of Chemical Engineering, McGraw-Hill, Boston 2001. D. A. Crowl, J. F. Louvar, Chemical Process Safety: Fundamentals with Applications, Prentice Hall PTR, Upper Saddle River, N. J. 2002. J. Herwig, R. Trainor, in Applied Homogeneous Catalysis with Organometallics, Eds. B. Cornils, W. Herrmann, Wiley-VCH, New York 1998. A. Heuman, M. Chanon, in Applied Homogeneous Catalysis with Organometallics, Eds. B. Cornils, W. Herrmann, Wiley-VCH, New York 1998. D. Peters, in Applied Homogeneous Catalysis with Organometallics, Eds. B. Cornils, W. Herrmann, WileyVCH, New York 1998. M. Mousseron-Canet, J. C. Mani, Photochemistry and Molecular Reactions, Keter, Jerusalem 1972. W. Stiller, Nichtthermisch Aktivierte Chemie, Birkhauser, Basel 1987. R. V. Ambartzumaian, V. S. Letokhov, Acc. Chem. Res., 1977, 10, 61–67. D. M. Golden, M. J.Rossi, A. C.Baldwin, J. R.Barker, Acc. Chem. Res., 1981, 14, 56–62. J. Kagan, Organic Photochemistry. Principles and Applications, Academic Press, London 1993. D. R. Cox, Planning of Experiments, Wiley, New York 1958. D. C. Montgomery, Design and Analysis of Experiments, Wiley, New York 2000. T. I. Mizan, J. Li, B. I. Morsi, M. Y. Chang, E. Maier, C. P.P. Singh, Chem. Eng. Sci., 1994, 49, 821–830.

References [33] Z. Tekie, J. Li, B. I. Morsi,

[34]

[35]

[36]

[37] [38] [39]

[40]

[41] [42] [43] [44] [45] [46]

[47]

[48]

[49]

[50]

[51]

[52]

M. Y. Chang, Chem. Eng. Sci., 1997, 52, 1541–1551. S. Hiraoka, Y. Tada, H. Suzuki, H. Mori, T. Aragaki, I. Yamada J. Chem. Eng. Jpn., 1990, 23, 468–474. G. Caristi, E. Mutudieri, Reaction Diffusion Systems, Marcel Dekker, New York 1998. J. Baldyga, J. R. Bourne, Turbulent Mixing and Chemical Reactions, Wiley, Chichester 1999. D. C. Peters, J. M. Smith, Can. J. Chem. Eng., 1969, 47, 268. Y. Sano, H Usui, J. Chem. Eng. Jpn., 1985, 18, 47. M. Garland, in Encyclopedia of Catalysis, Ed. I. T.Horvath (Ed) Wiley, New York, 2003. C. Y. Wen, L. T.Fan, Models for Flow Systems and Chemical Reactions, Dekker, New York 1975. D. W. T. Rippin, Ind. Eng. Chem. Fundam., 1967, 6, 488–492. O. Levenspiel, Chemical Reaction Engineering, Wiley, New York 1999. A. Bhattacharya, R. V. Chaudhari, Ind. Eng. Chem. Res., 1987, 26, 1168–1173. E. L. Muetterties, Inorg. Chem., 1965, 4, 769–771. A. M. Turning, Philos. Trans. R. Soc. London, Ser. B, 1952, 327, 37. V. Castets, E Dulos, J. Boissonade, P. De Depper, Phys. Rev. Lett., 1990, 64, 2953–2956. M. Garland, E. Visser, P. Terwiesch, D. W. T. Rippin, Anal. Chim. Acta, 1997, 351, 337–358. J. Aitchison, The Statistical Analysis of Composition Data, Chapman and Hall, London 1986. J. A. Cornell, Experiments with Mixtures: Design Models and the Analysis of Mixture Data, Wiley, New York 2002. H. Gunzler, H. U. Gremlich, Infrared Spectroscopy, Wiley-VCH, New York 2002. P. L. Polavarapu, Vibrational Spectroscopy: Principles and Applications with Emphasis on Optical Activity, Elsevier, New York 1998. Circular Dichroism, Eds. N. Berova, K. Nakamishi, R. Woody, Wiley-VCH, New York 2000.

[53] J. A. Iggo, NMR Spectroscopy in Inor-

[54]

[55] [56] [57]

[58] [59] [60] [61] [62] [63]

[64] [65] [66] [67] [68] [69] [70] [71]

[72] [73]

ganic Chemistry, Oxford University Press, Oxford 1999. J. W. Akitt, B. E. Mann. NMR and Chemistry: An Introduction to Modern NMR Spectroscopy, Stanley Thornes, Cheltenham 2000. J. M. Perkins, Radical Chemistry, Ellis Horwood, New York 1994. A. L. Smith, Applied Infrared Spectroscopy, Wiley, New York 1979. F. Kasler, Quantitative Analysis by NMR Spectroscopy, Academic Press, New York 1973. C. F. Tirendi, J. F. Martin. J. Magn. Reson., 1989, 85, 162–169. C. Fyhr, M. Garland, Organometallics, 1993, 12, 1753–1764. G. Liu, M. Hakimifard, M. Garland, J. Mol. Catal., 2001, 168, 33–37. L. B. Jackson, Digital Filters and Signal Processing, Kluwer, Boston 1996. W. S. Lu, 2-Dimensional Digital Filters, Dekker, New York 1992. L. Chen, The Exploratory Chemometric Studies of the Homogeneous Rhodium Catalyzed Hydroformylation of Isoprene, PhD Thesis, National University of Singapore, 2001. L. Chen, M. Garland, Appl. Spectrosc., 2002, 56, 1422–1428. M. Garland, G. Bor, Inorg. Chem., 1989, 28, 410–413. P. Chini, J. Chem. Soc., Chem. Commun., 1967, 440–441. W. F. Maddams, Appl. Spectrosc., 1980, 34, 245–265. L. Chen, M. Garland Appl. Spectrosc., 2003, 57, 323–330. J. Pitha, R. N. Jones, Can. J. Chem., 1966, 44, 3031. B. G. M. Vandeginste, L. De Galan, Anal. Chem., 1975, 47, 2124. P. R. Griffiths, J. A. Pierce, G. Hongjin, in Computer-Enhanced Analytical Spectroscopy, Eds. H. L. C. Meuzelaar, T. L. Isenhour, Plenum Press, New York 1987. T. L. Brown, Chem. Rev., 1958, 58, 581–608. W. M. A. Smit, in Laboratory Methods in Vibrational Spectroscopy, Eds. H. A. Willis, J. H. van der Maas, J. H. Miller, Wiley, New York 1989.

191

192

4 Processing Spectroscopic Data [74] L. Chen, M. Garland, Appl. Spectrosc., [75] [76]

[77] [78] [79]

[80] [81] [82]

[83] [84] [85]

[86] [87] [88]

[89]

[90] [91] [92] [93] [94] [95] [96] [97]

2003, 57, 331–337. W. H. Lawton, E. A. Sylvestre, Technometrics, 1971, 13, 617–633. E. A. Sylvestre, W. H. Lawton, M. S. Maggio, Technometrics, 1973, 45, 353–368. D. S. Bu, C. W. Brown, Appl. Spectrosc., 2000, 54, 1214–1221. W. Windig, J. Guilmat, Anal. Chem., 1991, 63, 1425–1432. F. C. Sanchez, J, Toft, B. Van den Bogaert, D. L. Massart, D. L. Anal. Chem., 1996, 68, 79–85. C. S. Shannon, Bell Syst. Technol. J., 1948, 3, 379–423. S. Watanabe, Pattern Recognit., 1981, 13, 381–387. J. N. Kanpur, Maximum-Entropy Models in Science and Engineering, Wiley, New York 1993. K. Sasaki, S. Kawata, S. Minami, Appl. Opt., 1983, 22, 3599–3603. K. Sasaki, S. Kawata, S. Minami, Appl. Opt., 1984, 23, 1955–1959. S. Kawata, K. Sasaki, S. Minami, H. Komeda, Appl. Spectrosc., 1985, 39, 610–614. Y. Zeng, M. Garland, Anal. Chim. Acta, 1998, 359, 303–310. Y. Pan, L. Susithra, M. Garland, J. Chemomet., 2000, 14, 63–77. G. H. Golub, C. F. van Loan, Matrix Computations, Johns Hopkins University Press, Baltimore 1989. A. Corana, M. Marchesi, C. Martini, S. Ridella, ACM Trans. Math. Software, 1987, 13, 263–280. S. Kirkpatrick, C. Gelatt Jr., M. Vecchi, Science, 1983, 220, 671–680. E. Widjaja, M. Garland, J. Comput. Chem. 2002, 23, 911. L. Chen, M. Garland, Appl. Spectrosc., 2003, 57, 491–498. W. Chew, E, Widjaja, M. Garland, Organometallics, 2002, 21, 1982–1990. C. Li, E. Widjaja, M. Garland, J. Catal., 2003, 213, 126–134. C. Li, E. Widjaja, M. Garland, J. Am. Chem. Soc., 2003, 125, 5540–5548. E. Widjaja, C. Li, M. Garland, Organometallics, 2003, 21, 1991–1997. W. Hieber, H. Lagally, Z. Anorg. Allg. Chem., 1943, 251, 96–113.

[98] P. Chini, S. Martinengo, Inorg. Chim.

Acta, 1969, 3, 21–24. [99] J. L. Vidal, W. E. Walker, Inorg. Chem.,

1981, 20, 249–254. [100] R. Whyman in Homogeneous Transi-

[101]

[102] [103] [104]

[105]

[106]

[107]

[108]

[109]

[110]

[111]

[112]

[113]

tion Metal Catalyzed Reactions, Eds. W. R. Moser, D. W. Slocum, American Chemical Society, Washington, D. C. 1992. C. Li, E. Widjaja, W. Chew, M. Garland, Angew. Chem. Int. Ed. Engl., 2002, 41, 3785–3789. R. Whyman, J. Chem. Soc., Dalton Trans., 1972, 13, 1375–1381. F. Oldani, G. Bor, J. Organomet. Chem., 1983, 246, 309–324. E. Widjaja, M. Garland in Recent Advances in Computational Science and Engineering, Eds. H. P. Lee and K. Kumar, Imperial College Press, London 2002, 62–66. L. R. Ong, E. Widjaja, R. Stanforth, M. Garland, J. Raman Spectrosc., 2003, 34, 282–289. S. Y. Sin, E. Widjaja, L. E. Yu, M. Garland, J. Raman Spectrosc., 2003, 34, 795–805. H. Zhang, M. Garland, Y. Zeng, W. Ping, in Recent Advances in Computational Science and Engineering, Eds. H. P. Lee and K. Kumar, Imperial College Press, London 2002, 49–53. H. Zhang, M. Garland, Y. Zeng, P. Wu, J. Am. Soc. Mass Spectrom., 2003, 14, 1295–1305. E. Widjaja, C. Li, W. Chew, M. Garland, Anal. Chem., 2003, 75, 4499–4507. Modeling of Chemically Reactive Systems, Eds. K. H. Ebert, P. Deulfhard, W. Jager, Springer, Berlin 1981. Complex Chemically Reactive Systems: Mathematical Modeling and Simulation, Eds. J. Warnatz, W. Jager, Springer, Berlin 1987. Numerical Treatment of Inverse Problems in Differential and Integral Equations, Eds. P. Deulfhard, E. Hairer, Birkhauser, Boston 1983. R. Shirt, M. Garland, D. W. T. Rippin, Anal. Chim. Acta, 1998, 374, 67–91.

References [114] E. Widjaja, C. Li, M. Garland in Re-

[117] C. Li, Bimetallic Catalytic Binuclear

cent Advances in Computational Science and Engineering, Eds. H. P. Lee, K.Kumar, Imperial College Press, London 2002, 36–40. [115] E. Widjaja, C. Li, M. Garland, J. Catal., 2004, 223, 278–289. [116] E. Widjaja, Development of Band-Target Entropy Minimization (BTEM) and Associated Software Tools, PhD Thesis, National University of Singapore, 2002.

Elimination Reaction. Experimental, Spectroscopic and Kinetic Elucidation, PhD Thesis, National University of Singapore, 2003. [118] W. Chew Exploratory Chemometric Studies of Unmodified Rhodium Catalyzed Conjugate Diene Hydroformylations via in situ FTIR Spectroscopy, PhD Thesis, National University of Singapore, 2003.

193

5 Carbonylation of Methanol to Acetic Acid and Methyl Acetate to Acetic Anhydride George Morris

5.1

Introduction: Evolution of Carbonylation Processes

The carbonylation of MeOH to AcOH was developed to pilot plant scale during the 1930s [1] (Eq. (1)). MeOH þ CO p AcOH

(1)

The first catalyst systems were Cu promoted by phosphoric acid and then first row transition metals promoted by iodide. By the 1950s, the related reaction of carbonylation of MeOAc to Ac2O was also known (Eq. (2)). MeOAc þ CO w Ac2O

(2)

However, even with the use of high temperatures and pressures, low productivity and poor selectivity did not make MeOH carbonylation to AcOH attractive commercially compared with processes such as oxidation of butane or naptha fractions or the Pd catalysed oxidation of ethylene (Wacker). Since then MeOH carbonylation to AcOH has been developed to commercial scale so successfully that it has displaced other technologies for the manufacture of AcOH. The first of these processes, operated from the 1960s, was that of BASF using Co promoted by iodide [2]. However, Monsanto introduced their Rh and MeI catalysed process in 1968, which operated at lower temperatures and pressures and with much improved selectivity, and this immediately became the technology of choice, being licensed to a number of other companies [3]. In the 1970s, Halcon discovered that MeOAc carbonylation to Ac2O could be carried out at industrially attractive rates and selectivities by using a Rh catalysed process promoted with MeI and an iodide salt [4]. This was developed into a process operated by Eastman [5]. Meanwhile, the original Monsanto reaction could still be improved, notably with respect to catalyst stability, which led to other constraints on operation. Celanese in Mechanisms in Homogeneous Catalysis. A Spectroscopic Approach. Edited by Brian Heaton Copyright c 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN 3-527-31025-8

196

5 Carbonylation of Methanol to Acetic Acid and Methyl Acetate to Acetic Anhydride

particular further developed it by the addition of iodide salts, which not only improved Rh stability, but also relaxed constraints on operating conditions leading, amongst other benefits, to improved rates [6]. At about the same time, BP Chemicals, who were also licensors of the Monsanto process, developed their own process for carbonylation of a mixed MeOH/MeOAc/ H2O feed to AcOH and Ac2O using a Rh and MeI catalysed process promoted with a quaternary ammonium iodide salt ([QAS]I), [7]. In 1986, BP Chemicals became the owners of the Monsanto technology. They subsequently also developed their own Cativaä process, announced in 1996, carbonylation of MeOH to AcOH catalysed by Ir and MeI and promoted with specific metal iodides [8]. As with the improvements in the original Monsanto Rh process, Cativaä had benefits such as improved catalyst stability and more favorable operating conditions [9]. Throughout the development and improvements of these processes, several groups published important work on their mechanisms and spectroscopy has played a major role in studying and understanding the reactions. Forster and coworkers at Monsanto set out the basic reaction schemes both for the Rh [10] and Ir [11] catalysed carbonylation of MeOH to AcOH. They also contributed to the understanding of the cycle leading to inactive catalyst and the by-product water gas shift chemistry [12]. Halcon [13] and Eastman [5] explained the role of iodide salt promoters in MeOAc carbonylation to Ac2O. Workers at Celanese investigated and explained the effect of iodide salt promoters in MeOH carbonylation to AcOH [14–16]. More recently, BP Chemicals and the groups of Maitlis and Haynes at Sheffield University have looked at details of both the Rh [17] and Ir [18] systems.

5.2

Some Important Features of Carbonylation Process Chemistry

Before looking in detail at the mechanisms of the various reactions and processes it is worth summarizing some relevant observations about the carbonylation processes as a whole that need to be explained by these mechanisms. It should be noted that, although the overall feedstock is MeOH for an AcOH process, for practical reasons it is usual to operate with relatively high [AcOH] in the reactor and that most of the substrate is present as MeOAc through esterification (Eq. (3)). MeOH þ AcOH w MeOAc þ H2O

(3)

It is usual therefore to use [MeOAc] as the measure of substrate present in reactor compositions when describing either carbonylation of MeOH to AcOH or MeOAc to Ac2O. The important features of the carbonylation processes described below are summarised in Table 5.1. Although the stoichiometry of the original Monsanto process (Eq. (1)) does not require H2O, it operates typically at 15 % w/w H2O in the reactor, and a low stand-

5.2 Some Important Features of Carbonylation Process Chemistry Some features of the principal Rh or Ir technologies for MeOH or MeOAc carbonylation in commercial operation.

Table 5.1

Process

High water Rh

Low water Rh

Anhydrous Rh

Low water Ir

Original Operators

Monsanto

Celanese

Eastman BP

BP Cativaä

Feedstock

MeOH

MeOH

MeOAca MeOAc/H2Ob

MeOH

Catalyst

Rh/MeI

Rh/MeI

Rh/MeI

Ir/MeI

Co-promoter

None

Effect of I salts

a

LiI

LiI [QAS]Ib

[Ru(CO)xIy]

Promoter/stabiliser

Promoter/stabiliser

Poison

H2O

i 15 % w/w

I 10 % w/w

Anhydrous

I 10 % w/w

MeOAc

1–2 % w/w

ii 2 % w/w

ii 10 % w/w

i 10 % w/w

Water gas shift

CO2/H2

CO2/H2

C2 by-products

EtCOOH

EtCOOH/AcH

a

Eastman.

b

CO2/CH4 EDA

EtCOOH

BP.

ing concentration, typically 1 to 2 % w/w, of substrate MeOAc. At lower [H2O], it may not be possible to sustain catalyst activity, either in the short term as the catalyst becomes less active, or in the long term when it may deposit from the process. This deposition is as an insoluble form of RhI3 and tends to occur in areas of the plant such as separation units where the CO partial pressure is low [6]. Although higher rates can be achieved by increasing reactor [MeOAc], this also tends to lead to long term catalyst deposition. At higher [MeOAc], the overall carbonylation rate tends to a limiting value, is first order in [Rh] and [MeI] and is independent of [H2O], [MeOAc] and [CO]. In the high [H2O], low [MeOAc] regime used in commercial operation the carbonylation rate is less than the limiting value and a significant side reaction is the water gas shift [12] (Eq. (4)). CO þ H2O w CO2 þ H2

(4)

In Rh catalysed MeOH carbonylation, the water gas shift reaction is catalysed by Rh carbonyl iodides and the presence of HI. It leads to a loss of yield with respect to CO, both because it is being consumed by the water gas shift and because some CO has to be vented along with H2 and CO2 to prevent them building up in the reactor. In this regime, the trend is for the water gas shift rate to increase relative to the carbonylation rate as [MeOAc] and [H2O] decrease. The presence of H2 tends also to be associated with the formation of small amounts of EtCOOH, which may be sufficient to require separation (Eq. (5)). MeOH þ 2CO þ 2H2 p EtCOOH þ H2O

(5)

197

198


The major Rh species observed in MeOH carbonylation to AcOH at higher [H2O] under commercial operating conditions are [Rh(CO)2I2]– and [Rh(CO)2I4]– [12]. In the carbonylation of MeOAc to Ac2O, little or no reaction corresponding to the water gas shift takes place. Indeed it may be advantageous to feed small amounts of H2 to the process to increase catalyst activity [5]. This is also associated with the formation of ethylidene diacetate, CH3CH(OAc)2, (EDA) (Eq. (6)). 2Ac2O þ H2 p EDA þ AcOH

(6)

Some CO2 is also formed via a decarboxylation of Ac2O to acetone (Me2CO) (Eq. (7)). Ac2O p Me2CO þ CO2

(7)

The process is operated at higher reactor [MeOAc] than the original Monsanto MeOH carbonylation and there are no issues of catalyst stability. Reaction rate tends to increase with reactor [Rh], [MeI], [MeOAc], [CO] partial pressure and promoter [I–]. Indeed the reaction shows a classical shift between two limiting cases of kinetic behavior [19]. At high [MeOAc] the rate is independent of [MeOAc] and tends to first order in [Rh] and [MeI]. At high [Rh], the reaction tends to first order in [MeOAc]. Another difference between carbonylation of MeOAc to Ac2O and carbonylation of MeOH to AcOH is that the first reaction can go to completion, (Eq. (1)), at typical operating pressures of CO but the second will reach equilibrium (Eq. (2)). The major Rh species observed in MeOAc carbonylation to Ac2O in anhydrous media under commercial operating conditions is [Rh(CO)2I2]– [5]. The carbonylation of MeOH catalysed by Rh and MeI and promoted with iodide salts can be operated at lower reactor [H2O] and higher [MeOAc] than were originally used in commercial plant. The iodide salt overcomes stability issues and higher reaction rates and lower water gas shift rates are obtained. Some formation of reduced C2 species still takes place both as EtCOOH but also acetaldehyde (AcH). The addition of the iodide salt alone extends the region where the overall rate depends only on [Rh] and [MeI] to lower [MeOAc] and [H2O]. The major Rh species observed in MeOH carbonylation to AcOH in the presence of iodide salts at lower [H2O] under commercial operating conditions are [Rh(CO)2I2]– and [Rh(CO)2I4]–, though [Rh(CO)2I2]– may be favored compared with working in the absence of iodide salts at lower [MeOAc] [16]. The carbonylation of MeOH catalysed by Ir and MeI can also be operated at lower reactor [H2O] and higher [MeOAc] than the original Monsanto process and without issues of catalyst stability. Commercially acceptable rates can be achieved at lower [MeI] concentrations by using promoters such as carbonyl iodide complexes of Ru and Os or covalent iodides such as InI3 or ZnI2 [9]. Ionic iodide salts are potent poisons for the Ir catalysed reaction [11]. In contrast with the Rh catalysed systems, CH4 and not H2 is co-produced as a gaseous by-product (Eq. (8)). MeOH þ CO p CH4 þ CO2

(8)

5.3 Key Steps in the Mechanism of Carbonylation Processes

The principal reduced C2 by-product is EtCOOH. In Ir catalysed reactions, the carbonylation rate increases with [Ir], [MeOAc], [MeI] and [CO]. The rate with respect to [H2O] passes through a maximum. In contrast to Rh systems, the water gas shift in Ir catalysed MeOH carbonylation tends to be a relatively constant fraction of carbonylation rate as the reactor composition is varied, though it tends to decrease with [MeI] and [MeOAc] but increase if ionic iodides are added [9]. The major Ir species observed in MeOH carbonylation to AcOH in the Cativaä process under commercial operating conditions are [IrMe(CO)2I3]– and [Ir(CO)2I4]– [18]. Many other modifications, particularly of the Rh and MeI catalysed carbonylation of MeOH, have been proposed and some of these have been operated commercially or may have been tested at significant pilot plant scale. These include, for example, the use of phosphine oxide species such as PPh3O [20] as promoters and systems involving immobilizing the Rh on ion exchange resins [21]. Numerous examples of ligand modified catalysts have been described, particularly for Rh, though relatively few complexes have been shown to have any extended lifetime at typical process conditions and none are reported in commercial use [22, 23]. The carbonyl iodides of Ru and Os mentioned above in the context of the Cativaä process are also promoters for Rh catalysed carbonylation of MeOH to AcOH [24].

5.3

Key Steps in the Mechanism of Carbonylation Processes

In common with many catalysed reactions, the important features of carbonylation process chemistry may be associated with different aspects of the catalytic cycle. Broadly, process activity may vary either because (i) more of the catalyst is present in the active form, (ii) the activity of the catalyst in the active form is enhanced or inhibited or, less commonly, (iii) the rate controlling step does not involve the catalyst. The process selectivity may vary because of side reactions (i) occurring through the active catalyst cycle, (ii) involving inactive catalyst, or (iii) taking place because of the organic chemistry of the systems. Examples of all these contributions to overall process efficiency are found in the various commercial carbonylation processes. The main steps in the catalytic MeOH carbonylation cycle which were proposed for the Co catalysed process [2] have served, with some modification perhaps in the carbonylation of MeOAc to Ac2O, to the present day and are familiar as a classic example of a metal catalysed reaction. These steps are shown in Figure 5.1. They are of course, (i) the oxidative addition of MeI to a metal center to form a metal methyl species, (ii) the migratory insertion reaction which generates a metal acyl from the metal methyl and coordinated CO and (iii) reductive elimination or other evolution of the metal acyl species to products. Broadly, as will be discussed in more detail later, the other ligands in the metal environment are CO and iodide. To balance the overall chemistry a molecule of CO must also enter the cycle.

199

200


Figure 5.1 The principal steps of MeOH carbonylation reactions.

Loss of selectivity during the active catalyst cycle can occur, for example, at the level of the metal methyl, leading to CH4 or the metal acyl, leading to C2 byproducts. The sum of the metal catalysed steps leads to the overall primary carbonylation reactions (Eq. (9)) and (Eq. (11)). To complete the cycle of the organic components, for every mole of AcOH or Ac2O generated a mole of either HI or AcI must be converted to MeI (Eq. (10)) and (Eq. (12)). MeI þ CO þ H2O p AcOH þ HI HI þ MeOH w MeI þ H2O MeOH þ CO p AcOH

(9) (10) (1)

MeI þ CO w AcI

(11)

AcI þ MeOAc w MeI þ Ac2O

(12)

MeOAc þ CO w Ac2O

(2)

This last step, recycling HI or AcI, has perhaps received less attention than the metal catalysed steps but it will be seen to have considerable practical importance in the overall behavior of the processes. Finally, the most important nonproductive cycle of carbonylation catalysts, which leads to a loss of activity through formation of an inactive higher valent catalyst species, as well as loss of selectivity through the water gas shift reaction, is summed up by the following steps, (Eq. (13)) and Eq. (14)). M(n) þ 2HI w M(nþ2)I2 þ H2

(13)

M(nþ2)I2 þ CO þ H2O w M(n) þ CO2 þ 2HI

(14)

CO þ H2O w CO2 þ H2

(4)

5.4 Information from HP IR and HP NMR for Carbonylation Reaction Studies

5.4

Information from HP IR and HP NMR for Carbonylation Reaction Studies

MeOH and MeOAc carbonylation reactions lend themselves particularly well to mechanistic investigation by spectroscopic methods, particularly IR and NMR. There are a number of reasons for this. Firstly the species involved, both the bulk reactants and the catalysts, are structurally relatively simple. This means that there is a reasonable possibility of resolving, identifying and quantifying components by either or both methods. 5.4.1

Studying Carbonylation Mechanisms with IR and HP IR

Fortunately, the mid-IR stretches of the terminal carbonyls of simple carbonyl iodide complexes of Rh and Ir occur in the region 1950 to 2150 cm–1. Their extinction coefficients are sufficiently strong that even in aqueous AcOH, which would not be a solvent of choice for mid-IR spectroscopy, at concentrations in the range of 100s of ppm w/w, the simple Rh or Ir carbonyl iodides can be detected by FTIR with a modest acquisition time. Indeed much of the original IR work to study both Rh and Ir catalysed carbonylation by workers at Monsanto [11, 25] and by Schrod et al. [26] appears to have been carried out using continuous wavelength machines. Figure 5.2 shows a typical IR spectrum for a Rh system in aqueous AcOH obtained from a transmission cell at 180 hC and pressurized to about 30 bar with CO. Peaks due to [Rh(CO)2I2]– are indicated.

Figure 5.2 HP IR spectrum from measurement of water gas shift rate and Rh speciation in a transmission cell showing Rh species and CO2 in AcOH/H2O/HI at 180 hC and 30 bar.

201

202


The cell used for this work was a modified version of the design described by Whyman [27]. This cell is in effect a stirred autoclave with a pair of CaF2 windows mounted in it. The path length needs to be as short as possible to give sufficient transmission through the sample but not so short as to lead to the restriction of movement of liquid between the windows. The original design was modified to seat the windows on Kalrezä O-rings, an elastomer reliably resistant to the very aggressive reaction medium. In fact, the spectrum in Figure 5.2 was obtained during an experiment to measure the Rh and HI catalysed water gas shift rate in situ by also following the generation of CO2. The CO2 peak is also indicated in Figure 5.2. It will be recognized that little useful information can be obtained from the rest of the spectrum. More recently, cells using total internal reflectance (TIR) probes have become readily available. The spectrum in Figure 5.3 was obtained, again at temperature and pressure, from an Ir catalysed carbonylation of methanol in a Parr autoclave modified to take a Sentinelä IR probe. The effective path length with the TIR cell is much shorter and composition information can be abstracted from the fingerprint region as well as direct observation of metal carbonyl species and CO2. However, the shorter path length can make detection and quantification of low concentrations of catalyst species more difficult. Using either a transmission cell or a TIR cell as described above to study MeOH carbonylations means that the IR information is taken under the same conditions as any normal batch autoclave experiment in carbonylation. Typically the reactor is charged with most of the liquid components for a run, such as MeOAc, H2O, MeI and some AcOH as solvent. A small portion of the liquid components and the Rh or Ir catalyst is charged to an injector system. The main cell is pressurized to a few bar with CO and heated to reaction tempera-

HP IR spectrum from measurement of carbonylation rate, water gas shift rate and Ir speciation in TIR cell showing Ir species and CO2 in AcOH/H2O/MeOAc/MeI at 190 hC and 30 bar.

Figure 5.3

5.4 Information from HP IR and HP NMR for Carbonylation Reaction Studies

ture. At this point a background spectrum of the initial reaction mixture is generally taken. Once operating temperature has been reached, the catalyst is injected into the cell by feeding CO from a ballast vessel, up to the required operating pressure, through the injector. The reaction starts more or less immediately. As CO is consumed, CO is fed from the ballast vessel via a pressure regulator to maintain the reactor pressure and the pressure drop in the ballast vessel is recorded against time. Since MeOH or MeOAc carbonylation is generally a very selective reaction, the reactor composition at any time throughout the reaction can be calculated from the amount of CO consumed. From these measurements the relationship between reaction rate and reactor composition can be established. By obtaining IR data at the same time, the nature and amount of catalyst species present can be measured to relate to rate and reaction composition. At the same time, useful data about the water gas shift reaction can be obtained from the increase in the CO2 peak. A set of spectra from the carbonylation of MeOH catalysed by Ir is shown in Figure 5.4. It will be seen that background to the catalyst species is changing throughout the reaction as MeOAc and H2O are consumed to produce AcOH in the liquid phase. To obtain spectra of the catalyst, which can be compared with each other, and where the species can be reliably quantified, may require that further background spectra of the reaction composition be available. The spectra of a mid-point and an end-point composition are usually sufficient for this. Model studies of the reactions of catalyst species by IR are generally carried out under conditions of temperature, pressure and solvent to isolate individual reaction

Figure 5.4 Detail of HP IR spectra from measurement of water gas shift rate and Ir speciation in TIR cell showing Ir species and CO2 in AcOH/H2O/MeOAc/MeI at 190 hC and 30 bar during a batch carbonylation reaction.

203

204


Figure 5.5 Detail of HP IR spectra from measurement of migratory insertion of [IrMe(CO)2I3]– in Circleä cell showing Ir species in PhCl at 93 hC and 5.5 bar.

steps and look at conversion of one catalyst species in the cycle to another. Figure 5.5 shows spectra from an experiment to measure the rate of the migratory insertion reaction in the Ir system in PhCl at moderate temperature and pressure. The conditions are far less demanding from the point of view of the spectroscopy. In general, these reactions do not involve gross changes in the composition of the reaction medium, requiring less, or sometimes no, further manipulation of the spectra. At the same time, increased concentrations of the species may be used to improve the signal to noise ratio. The overall effect is to generate spectra from which good quality kinetic data can readily be obtained. Through working on a smaller scale, it is often feasible to use isotopically labeled compounds such as 13CO, 13 C-MeI or 2H species as a further probe of structure and kinetics. 5.4.2

Studying Carbonylation Mechanisms with NMR and HP NMR 1

H and 13C NMR have been used extensively in studies relating to MeOH and MeOAc carbonylation. For the most part, this has been to identify the structure of catalyst complexes and to probe the mechanism by following the incorporation of labeled species such as 13CO and 13C-MeI. Some kinetic studies of metal complexes using NMR as well as studies related to the organic iodide species in the catalytic cycle are described later. Many of these studies have been carried out at temperature and pressure using high pressure sapphire tubes with Ti metal valves of a variety of designs but drawing on the original work of Roe et al. at DuPont [28]. In general, a belief may have arisen that high pressure NMR experiments can only give very limited information about reactions where significant gas uptake is occurring, such as working catalytic carbonylations, because of poor gas/liquid mixing. While this may often be the case and there is also no provision for topping

5.5 Spectroscopic Studies of Model Reaction Steps of the Rh Carbonylation Cycle

1 H HP NMR data (sapphire tube) for carbonylation of MeOAc catalysed by Rh/MeI: (i) aqueous and (ii) anhydrous, measured by 1H NMR at 180 hC and 35 bar (initial) in AcOH/ MeI/[QAS]I.

Figure 5.6

up CO as it is consumed, it seems that under some circumstances the reaction can be followed and useful information obtained [29]. 1H NMR spectra of a Rh/MeI catalysed carbonylation of MeOAc to Ac2O in the presence of H2 at 180 hC were measured. This reaction was carried out with a substantially higher concentration of Rh than would have been used in a batch autoclave experiment. Data for the decrease in MeOAc and increase in Ac2O and EDA during this experiment are shown in Figure 5,6 and compared with data for decrease in MeOAc in a similar experiment but in the presence of H2O. The data mirror those for gas uptake in batch autoclave experiments, which suggest a faster and near zero-order rate of MeOAc consumption in the aqueous reaction and a gradual decline in rate towards an equilibrium composition in the anhydrous system. However, the HP NMR experiment generates information on the rate of formation of minor by-products such as EDA, which cannot be inferred from batch autoclave gas uptake measurements or IR experiments. It was concluded from these results that, at high temperature, there was significant gas liquid mixing in tubes of this design, presumably through refluxing and convection, though there is no bulk boiling of the solution under pressure.

5.5

Spectroscopic Studies of Model Reaction Steps of the Rh Carbonylation Cycle

The individual reaction steps of both Rh or Ir and MeI catalysed MeOH or MeOAc carbonylation: oxidative addition of MeI to the metal center, migratory insertion to generate a metal acyl species and elimination from the metal acyl to generate the

205

206


acetyl product and return the catalyst to the catalytic cycle; have now been studied to a greater or lesser extent under model conditions, chiefly using IR and NMR spectroscopy. In addition some work has been carried out to investigate reactions that may contribute to by-products such as the water gas shift and CH4 formation. 5.5.1

Model Studies of Oxidative Addition in the Rh system

The original work by Monsanto identified [Rh(CO)2I2]– as the major Rh species under their process conditions and reaction of MeI with this complex as rate controlling for the process [3]. However, the proposed primary product of oxidative addition of MeI to [Rh(CO)2I2]–, [RhMe(CO)2I3]–, was not observed in early work. Forster [10] studied the reaction of MeI with [Rh(CO)2I2]– by IR and found that it gave the acyl complex [Rh(C(O)Me)(CO)I3]–, which was also isolated and characterized in the solid state as a dimer. The reaction could be followed quantitatively but the observed rate constant would be anticipated to be a composite of the rate constants for the formation of the intermediate [RhMe(CO)2I3]– and its further reaction to [Rh(C(O)Me)(CO)I3]– (Eq. (15)) and (Eq. (16)). [Rh(CO)2I2]– þ MeI w [RhMe(CO)2I3]–

(15)

[RhMe(CO)2I3]– w [Rh(C(O)Me)(CO)I3]–

(16)

The first quantitative studies of the reaction of [Rh(CO)2I2]– with MeI were reported by Maitlis and Hickey [30] who used IR to follow the formation of [Rh(C(O)Me)(CO)I3]– in aprotic media and in the presence of iodide salts. They proposed that the promotion of the overall reaction rate was due to the formation of a species such as [Rh(CO)2I3]2–, which would be a more potent nucleophile towards MeI than [Rh(CO)2I2]–, (Eq. (17–18)). [Rh(CO)2I2]– þ I– w [Rh(CO)2I3]2–

(17)

[Rh(CO)2I3]2– þ MeI w [RhMe(CO)2I3]– þ I–

(18)

Similar observations on the promotion of the reaction of [Rh(CO)2I2] – with MeI in protic media were published by workers at Celanese [14, 15]. They investigated by IR the promotional effect of both iodide and acetate salts on the reaction, since for metals such as Li, a preferred promoter for carbonylation, substantial amounts of LiOAc, formed in situ (Eq. (19)), might be present in the working carbonylation media. LiI þ MeOAc w LiOAc þ MeI

(19)

LiI and LiOAc both promoted the reaction of [Rh(CO)2I2]– with MeI in protic media and again the formation of a more nucleophilic [Rh(CO)2I2X]2– (X ¼ I– or OAc–)

5.5 Spectroscopic Studies of Model Reaction Steps of the Rh Carbonylation Cycle

species was put forward as an explanation of the results. At the same time, the authors studied the influence of solvent properties, notably composition/polarity, on the combined oxidative addition/migratory insertion step and reported them not to greatly influence the rate, though the baseline for this conclusion was an aqueous system. It was reasoned that the likelihood of observing the [RhMe(CO)2I3]– species would be favored by working at as high a concentration of MeI as possible, because this would enhance the formation of [RhMe(CO)2I3]– from [Rh(CO)2I2]– and, being a relatively nonpolar medium, would retard its subsequent reaction to [Rh(C(O)Me)(CO)I3]–. When neat MeI was used in the reaction with [Rh(CO)2I2]–, a very small peak was initially identified by IR, which would be consistent with [RhMe(CO)2I3]– [31, 32]. Further careful analysis of the IR spectra showed a second peak associated with this species, which would be expected for a cis-dicarbonyl complex. Using 13CO and 13C-MeI, 13C NMR measurements showed the presence of Me and CO resonances which are also consistent with the presence of [RhMe(CO)2I3]–. Taken together, the IR and NMR data showed the complex to be the fac, cis isomer. From a quantitative analysis of the data in the IR spectra, it was possible to obtain the rate constant for formation of [RhMe(CO)2I3]– as it grew in. The rate of reaction of 2H MeI with [Rh(CO)2I2]– was measured and compared with the rate of reaction of MeI and the kinetic isotope effect (KIE) was shown, through ab initio molecular orbital calculations, to be consistent with an SN2 mechanism proceeding through a linear (Rh–C–I ) transition state [33]. 5.5.2

Model Studies of Migratory Insertion in the Rh System

Since [RhMe(CO)2I3]– could be quantified, it was also possible to estimate the rate constant for the formation of [Rh(C(O)Me)(CO)I3]– from [RhMe(CO)2I3]–. [Rh(C(O) Me)(CO)I3]– had also been shown to eliminate MeI to regenerate [Rh(CO)2I2]– and the reactions are therefore reversible. Using rate data from IR experiments on the overall elimination reaction from [Rh(C(O)Me)(CO)I3]– in the absence of MeI and the equilibration of [Rh(13C(O) Me)(CO)I3]– with [Rh(C(O)Me)(13CO)I3]– in neat MeI it was possible to estimate the two further rate constants that defined the system. Putting these together showed that [RhMe(CO)2I3]– is unstable both with respect to reductive elimination and migratory insertion at 35 hC in MeI/CH2Cl2 [32]. k15 ¼ 6.8 x 10 –5 M–1 s–1 [Rh(CO)2I2]– þ MeI w [RhMe(CO)2I3]– k–15 ¼ 1.5 x 10 –2 s–1

(15)

k16 ¼ 1.3 x 10 –1 s–1 [RhMe(CO)2I3]– w [Rh(C(O)Me)(CO)I3]– k–16 ¼ 4.2 x 10 –5 s–1

(16)

207

208


The effect of solvent polarity on the rate of the individual steps was also deduced from a comparison of the kinetics determined by IR. It was concluded that, comparing MeOH/MeI (80:20 v/v) with CH2Cl2/MeI (80:20 v/v), the overall increase in rate of reaction of [Rh(CO)2I2]– with MeI to give [Rh(C(O)Me)(13CO)I]– included contributions due to enhancement of the forward rates of both oxidative addition (ca. 50 %) and migratory insertion (ca. 100 %). 5.5.3

Model Studies of Reductive Elimination in the Rh System

The final steps to complete the catalytic cycle in Rh catalysed MeOH carbonylation involve coordination of CO to [Rh(C(O)Me)(CO)I3]– and elimination of the elements of AcI from [Rh(C(O)Me)(CO)I3]– to regenerate [Rh(CO)2I2]– (Eq.(20)) and (Eq.(21)). Rh(C(O)Me)(CO)I3]– þ CO w [Rh(C(O)Me)(CO)2I3]–

(20)

[Rh(C(O)Me)(CO)2I3]– w [Rh(CO)2I2]– þ AcI

(21)

The reductive elimination has been less studied than the oxidative addition and migratory insertion steps. Forster reported that [Rh(C(O)Me)(CO)I3]– decomposed to AcI and [Rh(CO)2I2]– in several solvents without any marked dependence of the rate on the solvent. AcI was detected in aprotic media from reaction of MeI with CO in the presence of [Rh(CO)2I2]–. However, it was also concluded that oxidative addition of AcX to [Rh(CO)2X2]– (X ¼ Cl, Br) was not thermodynamically favorable [34]. The rate of decomposition of [Rh(C(O)Me)(CO)2I3]– in CH2Cl2 was subsequently measured by IR (t1/2 ¼ 12 h at 25 hC) [17]. Meanwhile, it had been shown, by following the reaction in situ by NMR [35], that the elimination of AcI from [Rh(C(O)Me)(CO)2I3]– was reversible and, in fact, that AcI would add to [Rh(CO)2I2]–. The reaction of [Rh(C(O)Me)(CO)2I3]– with various nucleophiles has been studied by IR [17]. Acetate ion gave immediate formation of Ac2O. Amines gave amides, the more nucleophilic dialkylamines reacting more rapidly than methyl anilines. The kinetics of these reactions were interpreted as consistent with two pathways, one being elimination of AcI from [Rh(C(O)Me)(CO)I3]– and the other direct nucleophilic attack at the acyl (Eq. (22)). [Rh(C(O)Me)(CO)2I3]– þ MNu p [Rh(CO)2I2]– þ AcNu þ MI

(22)

5.6 Spectroscopic Studies of the Model Reaction Steps of the Ir Carbonylation Cycle

5.6

Spectroscopic Studies of the Model Reaction Steps of the Ir Carbonylation Cycle

As with Rh catalysed carbonylation of MeOH, much enduring work on the mechanism of Ir catalysed carbonylation was carried out and reported by Forster at Monsanto [11]. Combining data from catalytic reactions, synthesis and reactions of intermediates and IR studies of reaction solutions, three regimes or cycles, designated I, II and III, were identified, compared with the one for Rh. A key reactor composition variable determining which regime is operating is [H2O]. Regime II, at intermediate [H2O], referred to as the ionic cycle, parallels the Rh cycle and involves complexes such as [IrMe(CO)2I3]–, [Ir(CO)2I4]–, Ir(C(O)Me) (CO)2I3]– and [Ir(CO)2I2]–, with the first two of these being the predominant species in solution. At low [H2O], which it will be seen later favors low [HI], neutral species such as [Ir(CO)3I], [IrMe(CO)2I2] and [Ir(C(O)Me)(CO)2I2] were observed or proposed to be on the main catalytic pathway. In the third regime, at high [H2O], where Forster noted high water gas shift activity as well as carbonylation, the ionic species now included substantial amounts of the ionic hydride complex [IrH(CO)2I3]–. Forster reported significantly different kinetic behavior in the neutral and ionic regimes. In the neutral regime reaction of MeI with [Ir(CO)3I] was relatively slow and inhibited by increased CO. In contrast, in the ionic regime the oxidative addition of MeI to [Ir(CO)2I2]– to generate [IrMe(CO)2I3]– was fast. The subsequent conversion of [IrMe(CO)2I3]– to [Ir(C(O)Me)(CO)2I3]– was rate controlling, favored by increased CO and powerfully inhibited by I–. Of the three regimes described by Forster, it is the second, the ionic regime, which has proved the most attractive for commercial operation and subsequent quantitative model studies have concentrated on interconversion of complexes in that cycle. As with model studies of Rh catalysts, IR and NMR spectroscopy have been the main techniques used to follow kinetics and characterize complexes. 5.6.1

Model Studies of Oxidative Addition in the Ir System

The reaction of [Ir(CO)2I2]– with MeI was followed by IR at 25 hC in CH2Cl2 [36] (Eq. (23)). [Ir(CO)2I2]– þ MeI w [IrMe(CO)2I3]–

(23)

The rate constant for the second order reaction was found to be at least two orders of magnitude greater than that for the corresponding Rh system. As with the corresponding Rh reaction, a modest kinetic isotope effect was observed when 2H-MeI was used, again consistent with an SN2 mechanism for this step [33].

209

210


5.6.2

Model Studies of Migratory Insertion in the Ir System

As Forster had noted, migratory insertion in [IrMe(CO)2I3]– in nonpolar solvents such as PhCl was slow, even at elevated temperature in the presence of CO, while the corresponding Rh species undergoes migratory insertion spontaneously at room temperature, (Eq. (24)). [IrMe(CO)2I3]– þ CO w [Ir(C(O)Me)(CO)2I3]–

(24)

The rate for the Ir reaction was measured at temperatures up to 120 hC and CO pressures up to 6 bar in PhCl by IR using a cylindrical internal reflectance cell [17]. Projecting the rates observed for the Rh complex to the temperature at which the Ir reaction was measured suggested that the rate constant for Rh would be five orders of magnitude greater than the Ir one. However, while the effect, as was described above, of adding a polar co-solvent to the Rh system is modest, for the Ir system it is dramatic. The kinetics of the migratory insertion of [IrMe(CO)2I3]– can be followed at room temperature in CH2Cl2/MeOH. The measured activation parameters are substantially different from those in neat PhCl, consistent with a dissociative rate-controlling step in more polar solvents and an associative one in less polar media [17, 37]. The same reactions were followed by HP NMR using selective 13CO and 13C-MeI labeling. The HPNMR results taken with the IR data suggest that it is the fac, cis isomer of [Ir(C(O)Me)(CO)2I3]– which forms initially at 25 hC in CH2Cl2/MeOH and that this partially isomerizes to the mer,trans isomer on heating. Both isomers are formed in broadly equal amounts in the higher temperature reaction in PhCl [18]. In PhCl/MeOH, the migratory insertion of [IrMe(CO)2I3]– is faster with increasing CO and slower with increasing [I–]. In the presence of [I–], although the overall rate is slower, it still increases with CO. Overall the results of these studies are consistent with co-solvents such as MeOH accelerating the reaction by promoting iodide dissociation from [IrMe(CO)2I3]– to give [IrMe(CO)2I2], (Eq. (25)), which adds CO, (Eq. (26)), undergoes migratory insertion (Eq. (27)) then recombines with I– to give [Ir(C(O)Me)(CO)2I3]– (Eq. (28)). [IrMe(CO)2I3]– w [IrMe(CO)2I2] þ I–

(25)

[IrMe(CO)2I2] þ CO w [IrMe(CO)3I2]

(26)

[IrMe(CO)3I2] w [Ir(C(O)Me)(CO)2I2]

(27)

[Ir(C(O)Me)(CO)2I2] þ I– w [Ir(C(O)Me)(CO)2I3]–

(28)

The results above are consistent with the original proposal of Forster on the importance of the formation of intermediate neutral, possibly reduced coordination num-

5.6 Spectroscopic Studies of the Model Reaction Steps of the Ir Carbonylation Cycle

ber, species such as [IrMe(CO)2I2] in the Ir catalytic carbonylation cycle. Further work has been carried out on the formation and reactivity of these species, particularly through the intervention of compounds which also function as co-promoters in Ir catalysed carbonylation, such as Ru carbonyl iodides and InI3 [18]. The reaction of InI3 with [IrMe(CO)2I3]– was found to be a convenient route to the synthesis of [IrMe(CO)2I2] whereby the pure compound could be isolated as a dimer [IrMe(CO)2I2]2, (Eq. (29)). [IrMe(CO)2I3]– þ InI3 w [IrMe(CO)2I2] þ InI4 –

(29)

In PhCl, reaction of [IrMe(CO)2I2]2 with CO to give [IrMe(CO)3I2] and onward reaction to give [Ir(C(O)Me)(CO)3I2] was followed by IR at temperatures up to 85 hC. The reaction is 700 times faster than the migratory carbonylation of the corresponding ionic species, [IrMe(CO)2I3]–. A further enhancement of rate for the neutral species is observed in MeOH/PhCl. When the migratory carbonylation of [IrMe(CO)2I3]– was carried out in PhCl in the presence of InI3 and followed quantitatively by IR, it was found to be catalysed by InI3 and the rate increased with InI3. Overall, the results are consistent with the InI3 generating [IrMe(CO)2I2] and migratory carbonylation occurring through this intermediate. Neutral Ru carbonyl iodide complexes such as [Ru(CO)4I2] and [Ru(CO)3I2]2, capable of I– abstraction, promoted the migratory carbonylation of [IrMe(CO)2I3]– in PhCl in a similar way to InI3, while ionic species such as [Ru(CO)3I3]– were inactive. Reactions followed quantitatively by IR showed the molar activity of these Ru species to be similar to that of InI3. Whereas the stoichiometric reaction of [IrMe(CO)2I3]– with InI3 could be used preparatively, the reaction with neutral Ru complexes appeared to give rise to Ir–Ru species (Eq. (30)). [IrMe(CO)2I3]– þ [Ru(CO)4I2] w [Ru(CO)3I2(m-I)IrMe(CO)2I2]–

(30)

These species were also observed by NMR and it was shown that under CO they could be cleaved to [IrMe(CO)3I2] and [Ru(CO)3I3]– [38]. 5.6.3

Model Studies of Reductive Elimination in the Ir System

As in the Rh system, the final step in the catalytic carbonylation cycle, the elimination of AcI or the direct attack on the acetyl ligand to generate products, has received less attention than oxidative addition of MeI or migratory carbonylation. Forster noted the elimination of AcI, identified by IR, in model reactions in the neutral regime for Ir. However, for the time being, elimination reactions in the presence of protic solvents, which lead to HI which goes on to react with the metal complexes, have tended to be regarded as confounding the studies of migratory insertion rather than adding to the overall understanding of the carbonylation cycle [18].

211

212


5.7

Spectroscopic Studies of the Organic Cycles of Carbonylation Reactions

As has already been noted, the catalytic cycles for carbonylation of MeOH to AcOH and MeOAc to Ac2O involve conversion of one or both of the iodide species HI or AcI back to MeI. This step was explicitly mentioned in the description of the BASF Co catalysed process, for example, though it does not seem to have merited any further comment [2]. The reaction of HI with MeOH to regenerate MeI, as well as the esterification of MeOH and AcOH were recognized as interrelated equilibria by the team at Monsanto who first investigated the Rh catalysed carbonylation of MeOH, (Eq. (3)), (Eq. (10)), (Eq. (31)) MeOH þ AcOH w MeOAc þ H2O

(3)

HI þ MeOH w MeI þ H2O

(10)

MeOAc þ HI w MeI þ AcOH

(31)

They suggested that these equilibria were established “instantaneously” at reaction conditions [12]. The overall effect of total MeOH and MeOAc in broadly determining the standing [HI] and the relationship with the behavior of Rh and Ir was also described. In particular, Forster related the major Ir species at different MeI/ MeOAc/MeOH/H2O compositions to [HI]. However, in general, there was no indication that the kinetics of the reaction of HI generated in the carbonylation cycle might have a bearing on the overall carbonylation reaction. In Rh catalysed systems, where the metal acyl species also clearly contained iodide, a further possibility was introduced, compared with the mechanism postulated by BASF for their Co systems, that elimination of AcI could occur. The earliest publications from Monsanto which described the proposed mechanism noted that they could not distinguish between a final step involving (i) reductive elimination of AcI followed by hydrolysis of AcI (Eq. (32)), (Eq. (33)) and (ii) hydrolysis at the metal center followed by some other initially unspecified mechanism of recycling HI to MeI (Eq. (34)) [3]. M(C(O)Me)(CO)xIy w M(CO)xIy-1 þ AcI

(32)

AcI þ H2O p AcOH þ HI

(33)

M(C(O)Me)(CO)xIy þ H2O p M(CO)xIy-1 þ AcOH þ HI

(34)

In fact, Monsanto opted to favor the elimination of AcI and Forster demonstrated the elimination of AcI from the complex [Rh(C(O)Me)(CO)2I3]–, though under anhydrous conditions. Since the hydrolysis of AcI (Eq. (33)) is expected to be a very fast reaction, detecting it under the conditions of a working MeOH carbonylation process, that is one

5.7 Spectroscopic Studies of the Organic Cycles of Carbonylation Reactions

in the presence of H2O, and thus demonstrating in this way that elimination of AcI from the metal acyl may be occurring, would be difficult and has not been reported. In contrast, the presence of AcI in the carbonylation of MeOAc to Ac2O is well established. Indeed a patent to Halcon [13] describes an experiment where AcI was claimed to have been distilled from an anhydrous carbonylation reaction mixture. 5.7.1

NMR Studies of Ac2O and AcI Hydrolysis

Interest in BP in the chemistry and rates of reactions of AcI and HI in carbonylation processes was initially directed to AcI in anhydrous carbonylation during the development of the [QAS]I promoted Rh/MeI catalysed process for co-producing AcOH and Ac2O. The feed to the process was actually MeOH/MeOAc/H2O. A question had arisen as to whether H2O and MeOH would have any significant standing concentration anywhere in a stirred tank reactor which was substantially anhydrous at the outlet, containing as it did a significant amount of Ac2O. The literature suggested that the hydrolysis of Ac2O, (Eq. (35)), was catalysed by halogen acids such as HI but intriguingly that, under some conditions, acetyl halides were not intermediates in the hydrolysis [39]. This seemed at odds with the observation that halogen acids react with anhydride to give acyl halides (Eq. (36)) and the plausible route for catalysed hydrolysis. AcI þ H2O w AcOH þ HI

(33)

Ac2O þ HI w AcI þ AcOH

(36)

Ac2O þ H2O w 2AcOH

(35)

AcI can be generated readily in AcOH/Ac2O by reaction with HI and 1H NMR spectra show that it is easily distinguished from them. The first experiments on hydrolysis were carried out with relatively large amounts of HI using a stopped-flow 1H NMR apparatus in which aqueous HI in AcOH was added and mixed rapidly with Ac2O/AcOH [40]. These clearly demonstrated the formation of AcI. [AcI] remained steady as Ac2O was hydrolyzed to AcOH. Having recognized the presence of AcI, it was possible to repeat the reaction at much lower initial [HI] and follow it without the need for stopped-flow methods. The original work in the literature from which the suggestion was made that the catalysed hydrolysis of Ac2O did not pass through the formation of acetyl halides used a measurement of temperature change to follow the reaction. It is possible that it was not appreciated that the formation of the acetyl halide from Ac2O, (Eq. (36)), is more exothermic than its hydrolysis, (Eq. (33)) and that the temperature changes reported could therefore be consistent with hydrolysis via the acetyl halide.

213

214


5.7.2

HP NMR Studies of the Reaction of AcI with MeOAc in Anhydrous Media

Once it was realized that AcI could be readily detected at low levels by 1H NMR in AcOH, work began to investigate the equilibria involving AcI in MeOAc carbonylation systems and the implications for the process chemistry. The work of Halcon [13] and subsequent publications by Eastman [5], demonstrated the formation of AcI (Eq. (11)) and that the overall reaction of carbonylation of MeOAc to Ac2O was an equilibrium (Eq. (2)). The principle of microscopic reversibility suggested that overall, when the reaction reached equilibrium, AcI should be present to mutually satisfy these equilibria. Furthermore, at the concentrations of MeOAc/MeI/ Ac2O at equilibrium for a given CO pressure, it should be possible to generate the equilibrium AcI by reaction of the organic species, (Eq. (12)). MeI þ CO w AcI

(11)


(12)

MeOAc þ CO w Ac2O

(2)

To maximize the level of AcI, MeI and Ac2O in AcOH were simply heated in an HP NMR tube and 1H NMR was used to identify and quantify the AcI and MeOAc formed. A room temperature 1H NMR spectrum from such an experiment is shown in Figure 5.7 since AcI is not readily detected at temperature and may be in exchange with other species. Based on the amounts of AcI and MeOAc formed from the reaction of MeI and Ac2O, the equilibrium constant for the reaction (Eq. (12)) is of the order of 1000.

1 H HP NMR spectra (sapphire tube) showing formation of AcI and MeOAc after heating MeI/Ac2O/AcOH/NaI to 180 hC.

Figure 5.7


These HP NMR experiments also demonstrated that decomposition of Ac2O was occurring. It was subsequently shown that this decomposition increased with [MeI] and [Ac2O] and thus with [AcI]. An explanation of this decomposition is that it arises through acylation of Ac2O by AcI to give the a–keto anhydride which decarboxylates to Me2CO. The original Halcon patent [4] on carbonylation of MeOAc to Ac2O demonstrated that LiI catalysed the reaction of AcI with MeOAc. An explanation of the activity of LiI had been put forward by workers at Eastman that involved the presence of LiOAc in the system through reaction with MeOAc, (Eq. (19)) then reaction of LiOAc with AcI (Eq. (37)) [5]. LiI þ MeOAc w LiOAc þ MeI

(19)

LiOAc þ AcI w LiI þ Ac2O

(37)


(12)

It was also proposed that a parallel but slower cycle was operating involving AcOH (Eq. (31)), (Eq. (38)) and (Eq. (12)) MeOAc þ HI w MeI þ AcOH

(31)

AcI þ AcOH w Ac2O þ HI

(38)


(12)

However, it was known that a range of other ionic iodide species such as NaI, KI and the [QAS]I salts were also effective as promoters for the Rh and MeI catalysed carbonylation of MeOAc to Ac2O [5, 41]. Further, workers at Rhone-Poulenc reported that some effective promoters such as phosphonium iodides generated only very small amounts of the corresponding acetates [42] and it was unclear that direct reaction of AcI with MeOAc in the presence of a range of iodides would be able to distinguish subtle variations in promoter activity. The effect of various salts on the reaction of Eq. (12) was investigated instead using 13C-MeOAc and following the equilibration of the label between Me species by 1H NMR at temperatures comparable to typical process conditions using HP NMR. In a typical experiment, equimolar amounts of 12C-MeI and 13C-MeOAc were heated to 180 hC in AcOH/Ac2O. The 1H NMR spectrum showed that the Me doublet of 13C-MeOAc decreases and is partly replaced by a singlet due to 12 C-MeOAc. At the same time, a doublet for 13C-MeI appears and the singlet due to 12C-MeI decreases. The 1H NMR spectra of the Me resonances of MeOAc during the course of an experiment are shown in Figure 5.8. The fractional amount of 12C-MeOAc in a similar experiment is plotted against time in Figure 5.9. A very good fit for a first order approach to equilibrium can be obtained for this plot. It will be seen that the half life for equilibration is of the order of 2 h in the example given.

215

216


1 H HP NMR spectra (sapphire tube) of MeOAc; Me resonances during reaction of C-MeOAc (7 % w/w) with 12C-MeI (15 % w/w) at 160 hC in AcOH/H2O.

Figure 5.8 13

Figure 5.9 1H HP NMR data (sapphire tube) of extent of 12C incorporation in MeOAc at 160 hC in AcOH/H2O starting from 13C-MeOAc and 12C-MeI.

If the dynamic equilibrium of Eq. (12), follows a simple rate law expression then the forward and backward reactions, having the same rate, would be described by the following relationship (Eq. (39)). k12[MeOAc][AcI] ¼ k–12 [MeI][Ac2O]

(39)

The observed rate constant for the approach to equilibrium for the label in the MeOAc, kMeeq, is given by the sum of the forward and back terms, (Eq. (40)), [43]. kMeeq ¼ k12[AcI]þ k–12[Ac2O]

(40)


Combining Eq. (39) and Eq. (40) suggests that the exchange rate should increase with [Ac2O] (Eq. (41)). kMeeq ¼ k12 ([MeI]/[MeOAc])[Ac2O] þ k–12 [Ac2O]

(41)

However, if the reaction takes place not directly but through HI as described above, (Eq. (31)) and (Eq. (38)), it can be shown that the exchange rate should increase not with [Ac2O] but with [AcOH], (Eq. (42)). kMeeq ¼ k31([MeI]/MeOAc) [AcOH] þ k–31 [AcOH]

(42)

If the position of the equilibrium is not significantly affected by the presence of the added salt, then the forward and back reactions are influenced to the same extent by the added salt and any change in the rate of equilibration of the label can be directly related to a change in forward and back rates. Thus, an increase in the rate of equilibration of the labelled species by an added species, such as an iodide salt, can be related to the capacity of the salt to increase the rate of reaction of AcI with MeOAc. The observed rate constants determined for the exchange of 13C-MeOAc with 12 C-MeI at 180 hC in Ac2O/AcOH over a range of concentrations of LiI, NaI, and [QAS]I are shown in Figure 5.10 [29]. It can be seen that the rate increases progressively with concentration for all the salts, and while the effect of NaI or [QAS]I is very similar, LiI is up to twice as effective on a molar basis. Rates were also measured for the [QAS]I system at different Ac2O/AcOH concentrations. These data are shown in Figure 5.11 plotted against [AcOH]. Clearly, the exchange rate is increasing with AcOH and not Ac2O in these experiments, consis-

Figure 5.10 1H HP NMR data (sapphire tube) of effect of iodide salts on rate of 12C incorporation in MeOAc at 180 hC in AcOH/Ac2O (20 % w/w) starting from 13C-MeOAc and 12C-MeI.

217

218


1

H HP NMR data (sapphire tube) of effect of [AcOH] on rate of 12C incorporation in MeOAc at 180 hC in AcOH/Ac2O/[QAS]I starting from 13C-MeOAc and 12C-MeI. Figure 5.11

tent with the mechanism involving the reaction of HI with MeOAc in the rate controlling step rather than the reaction of AcI with MeOAc. Overall, the effect of iodide salts on the reaction of AcI with MeOAc can be rationalized by a mechanism where there is a general I– catalysis of the HI route following protonation of MeOAc, (Eq. (43)), (Eq. (44)) [44]. MeOAc þ Hþ w MeO(H)Acþ

(43)

MeO(H)Acþ þ I– w MeI þ AcOH

(44)

This operates in the same way for LiI, NaI and [QAS]I but there is further catalysis through the acetate route for Li. If the difference between exchange rates for LiI and NaI or [QAS]I is attributed to the acetate route then the major part of the reaction may still be occurring through the HI route. 5.7.3

HP NMR Studies of the Reaction of HI with MeOAc in Aqueous Media

The role of iodide salts in the reaction of HI with MeOH or MeOAc under aqueous conditions has also been studied by several groups. In particular, the effect of LiI was investigated and reported by Celanese [14–16]. This work identified LiOAc again as a potentially important intermediate in aqueous systems and similar arguments were put forward as those used by Eastman for the anhydrous system. Before considering these further, it is relevant to look at the effect of H2O on equilibria in the aqueous system, (Eq. (3)), (Eq. (10)) and (Eq. (31)).


MeOAc þ H2O w AcOH þ MeOH

(3)

HI þ MeOH w MeI þ H2O

(10)


(31)

Monsanto measured and reported the Hammett acidity of solutions of HI in AcOH/H2O [12]. In this experiment the extent of protonation of a weak base such as nitroaniline is determined by UV/VIS spectroscopy at a wavelength where the free base absorbs but the protonated form does not. As [H2O] is decreased below about 15 % w/w in AcOH the Hammett acidity of HI increases. The accepted explanation of this phenomenon is that there is in effect a competition between the base B and H2O for the proton (Eq. (45)). BHþ þ H2O w B þ H3Oþ

(45)

At lower [H2O], the same quantity of HI becomes, in effect, a more potent protonating agent. There were indications when [HI] was determined at different [H2O] in process media for MeOH carbonylation to AcOH that [HI] varied as a function of [H2O]. Equilibrium [HI] was measured or inferred therefore over a range of [H2O] by heating MeI in AcOH/H2O [45]. Some data are shown in Figure 5.12. In the autoclave experiments [HI] was determined using an ion selective electrode method and in the NMR experiments [HI] was inferred from [MeOAc]. There is a variation of several orders of magnitude in the equilibrium constant on going from 15 % w/w H2O towards anhydrous AcOH as the amount of HI and MeOAc generated at equilibrium falls at lower [H2O]. However, when the Hammett acidity of the equilibrium [HI] was measured it was higher at lower [H2O]. On going to lower [H2O], the acidity of the equilibrium composition increases but [HI] decreases. Later we will see how these changes influence the overall behavior of carbonylation systems.

Equilibrium constants for MeOAc/HI/MeI/ AcOH system as a function of [H2O] determined from 1H HP NMR data and batch reaction analysis.

Figure 5.12

219

220


As with anhydrous systems, it was found that a systematic study of reactions such as HI with MeOAc or LiOAc with MeI to give accurate kinetic information, particularly at temperatures comparable with process conditions, was difficult. However, it was possible to obtain some useful qualitative information initially by 1H NMR. For example, as [H2O] was decreased HI was found to react more rapidly with MeOAc in aqueous AcOH. MeOMe was shown to react more rapidly than MeOAc with HI. For the LiI/LiOAc systems, where a significant amount of the LiOAc salt is formed at equilibrium, particularly at low [H2O], it was possible to follow the approach to equilibrium at temperature by 1H NMR using HP NMR. However, the dead time of these experiments was a major drawback and the reaction had taken place to a significant extent by the time the target temperature was reached [29]. The exchange reaction of labeled 13C-MeOAc with 12C-MeI in aqueous AcOH was used to study the effect of [H2O] and iodide species on the rates of reactions involving HI and MeOAc, (Eq. (3)), (Eq. (10)) and (Eq. (31)). A plot of exchange rate against [H2O] at 160 hC is shown in Figure 5.13 from which it can be seen that the exchange rate increases with [H2O]. The increase in the exchange rate with [H2O] is consistent with the rate controlling steps being the reactions of MeOH with HI and MeI with H2O (Eq. (10)). However, because both the equilibrium [HI] and the activity of HI are a function of [H2O], as described above, a number of factors are contributing to the variation of exchange rate with [H2O] and it does not appear to be a simple, linear relationship. As can be seen from Figure 5.13, the effect of LiI is greater than NaI at 2 % w/w [H2O] and the relative effect of [I–] on exchange rate is greater at lower [H2O]. At constant [H2O], the exchange rate increases with [I–] and depends also on the cation, as in the anhydrous system. Some data at 2 % w/w H2O for LiI, NaI and [QAS]I are shown in Figure 5.14. Again LiI is more effective than NaI or [QAS]I.

Figure 5.13 1H HP NMR data (sapphire tube) of effect of [H2O] on rate of 12C incorporation in MeOAc at 160 hC in AcOH starting from 13C-MeOAc and 12C-MeI.


1 H HP NMR data (sapphire tube) of effect of iodide salts on rate of 12C incorporation in MeOAc at 180 hC in AcOH/H2O (2 % w/w) starting from 13 C-MeOAc and 12C-MeI.

Figure 5.14

The effect of these simple iodide salts on the exchange rate can be explained by a similar mechanism to that described in detail above for the same promoters in anhydrous systems. This comprises catalysis of the reaction of HI with MeOAc, probably via MeOH, overall by a common iodide salt effect for all iodide salts and a second pathway for metals such as Li involving the parallel MI/MOAc cycle so that LiI is the most effective catalyst. The 13C-MeOAc/MeI exchange reaction has also been used to probe the interaction of Ru carbonyl iodides with MeOH carbonylation systems. In parallel 13C NMR was used to identify the Ru species present as a function of [H2O] and CO pressure [18]. Considering Ru speciation first, [Ru(CO)4I2] was heated to 160 hC under 13CO before recording the spectra and quantifying the principal Ru species, [Ru(CO)4I2] and [Ru(CO)3I3]– present. Some results are shown in Table 5.2. In addition to these two complexes, it will be noted that under certain conditions, particularly at lower CO pressure, significant amounts of other Ru species are observed. It can be seen that increasing [MeI] and [H2O] favor [Ru(CO)3I3]–, while higher CO pressure favors [Ru(CO)4I2]. The results are consistent with increasing [HI] at higher [H2O] and [MeI]. [Ru(CO)4I2] þ HI w [ Ru(CO)3I3]– þ Hþ þ CO

(46)

Preliminary experiments on the 13C-MeOAc/MeI exchange reaction in the presence of [Ru(CO)4I2] under CO showed that there was catalysis of the exchange though sometimes with large variations in the observed rate. It was recognized that the variable causing this was the CO partial pressure. Results are also shown in Table 5.2 from which it will be seen that the exchange rate decreases as CO pressure increases. As with ionic iodide salts the relative effect of these Ru species is greater at lower [H2O]. The Ru species are effective at relatively low concentrations compared with ionic iodide salts. At 2 % H2O, 0.04 M Ru gives an exchange rate that would require at

221

222

5 Carbonylation of Methanol to Acetic Acid and Methyl Acetate to Acetic Anhydride Table 5.2 Ru speciation and

13

C-MeOAc/12C-MeOAc exchange data from 1H HP NMR experi-

ments.a T, hC 160d 160d 160d 160d 160d 160d 160d 160d 160d 160d 160d 160d 168e 168e 168e 168e 168e a e

MeI, % w/w

MeOAc, % w/w

H2O, % w/w

14 0 0 16 16 5 5 5 5 5 15 15 15 15 15 15 15

0 0 0 0 0 0 0 15 15 15 7 7 7 7 7 7 7

1 7 7 8 8 7 7 7 7 7 2 2 2 2 4 7 7

LiI:Ru

1.4

1.1 b

CO, bar 36 36 5 36 5 32 4 3 10 40 44 7 31 31 28 28 28

Rub % Rutot

Ruc % Rutot

82 95 75 0 0 0 0 16 40 75

13 5 8 100 100 100 100 38 60 25

84 3 59 24 1

Exchange experiments 4000 ppm Ru. [Ru(CO)4I2]. c [Ru(CO)3I3]–. Independently determined temperature.

13 91 35 76 99 d

105 kobs, s–1

16.7 33.3 45.6 3.6 45.6 21.2 13.1

Nominal temperature.

least 0.5 M LiI, the most effective ionic iodide salt. This suggests that a further mechanism may be operating with promoters such as the Ru carbonyl iodides compared with the ionic iodides. Hþ[Ru(CO)3I3]– generated in this system is itself reported to be a strong acid and it may be the acidity of this species which promotes the exchange reaction through increased protonation of MeOAc in the rate controlling step (Eq. (43)). This explanation is supported by the observation that Liþ[Ru(CO)3I3]– does not catalyse MeOAc/MeI exchange as seen from the data reported in Table 5.2 [18].

5.8

Spectroscopic Studies of Working Carbonylation Reactions

We are now in a position, having considered the various model studies on individual reaction steps, to look more closely at what is observed in working catalytic MeOH and MeOAc carbonylation reactions. In some ways the anhydrous carbonylation of MeOAc to Ac2O is easier to describe because, in the absence of H2O, there is little or no water gas shift chemistry competing with the main carbonylation cycle.

5.8 Spectroscopic Studies of Working Carbonylation Reactions

5.8.1

Rh Catalysed Carbonylation of MeOAc to Ac2O

One of the first studies of a working system to be published was that by Schrod and Luft who observed the catalyst species in Rh catalysed MeOAc carbonylation to Ac2O by HP IR using a short path length transmission cell attached to an autoclave [26]. They recognized that even with potential ligands such as PPh3 present in the initial solutions the observed carbonylation rate was closely linked with the formation of [Rh(CO)2I2]–. However, at that time it was not always generally understood that most monodentate ligands such as PPh3 would be quaternised by MeI and reaction schemes involving complexes of such ligands were proposed for systems under catalytic conditions [26, 30]. Nevertheless, this quaternisation was implicit in many of the reactions described in patents on the Rh catalysed carbonylation of MeOAc to Ac2O and the observations of Monsanto that in aqueous systems almost any Rh precursor was converted to [Rh(CO)2I2]– under catalytic conditions [3]. Schrod and Luft also recognized the importance of reduction of higher valent Rh precursors to generate [Rh(CO)2I2]– in their batch reactions, achieving this with Cr as a reductant [46]. In a later publication, workers at Eastman demonstrated more clearly the role of reductants in maintaining catalyst activity in MeOAc carbonylation to Ac2O. They also used a HP IR transmission cell external to an autoclave to show that a portion of the Rh was present as inactive [Rh(CO)2I4]– in batch reactions and that by adding H2 to the reaction it was converted to [Rh(CO)2I2]–, thus increasing the catalytic activity, (Eq. (47)) [5]. [Rh(CO)2I4]– þ H2 w [Rh(CO)2I2]– þ 2HI

(47)

Further work at BP using HP IR suggested that the extent of reduction was a function of H2 and, with sufficient H2, Rh could be maintained as the active species, [Rh(CO)2I2]–, throughout a batch reaction [47]. However, in a working process this gain in activity must be traded off with the formation of EDA from Ac2O hydrogenation. In related HP IR work, it was also shown that increasing CO could contribute to maintaining Rh as [Rh(CO)2I2]– under catalysis conditions. As already described, commercially useful rates in Rh catalysed MeOAc carbonylation to Ac2O can only be achieved in the presence of substantial amounts of I–. The kinetic studies of MeOAc carbonylation to Ac2O coupled with the various spectroscopic observations and model reactions show that overall the observed carbonylation rate is controlled by the kinetics of formation and reaction of AcI as indicated in the reaction sequence below, (Eq. (48)), (Eq. (21)) and (Eq. (12)). [Rh(CO)2I2]– þ MeI þ CO w [Rh(C(O)Me)(CO)2I3]–

(48)

[Rh(C(O)Me)(CO)2I3]– w [Rh(CO)2I2]– þ AcI

(21)


(12)

223

224


The Eastman group clearly described how this was due to promotion of reaction of intermediate AcI with MeOAc, (Eq. (12)). The HP NMR experiments described above show a good correlation between activity in MeI/MeOAc exchange and promotional activity in catalytic carbonylation. At the same time, the model studies of I– promotion of oxidative addition suggest that the formation of AcI may also be accelerated under process conditions (Eq. (48)). Thus, I– salts can contribute to accelerating both steps that can be rate controlling in the carbonylation of MeOAc to Ac2O. Finally, we have also seen from the studies of AcI chemistry and its reaction with Rh complexes that, at all times in working Ac2O processes, AcI will be present and available to reform Rh(C(O)Me)-species. This can explain the HP NMR observation that formation of EDA by hydrogenation of Ac2O increases as batch reactions proceed. It is toward the completion of carbonylation of MeOAc to Ac2O, or the approach to equilibrium to be more precise, that [AcI] will be greatest and thus also presumably the Rh(C(O)Me)-species, which are hydrogenated to EDA. For these reasons all the reaction steps are shown as reversible equilibria. 5.8.2

Rh Catalysed Carbonylation of MeOH to AcOH

The Rh catalysed carbonylation of MeOH to AcOH was studied at Monsanto by HP IR under working reaction conditions using a short path length transmission cell coupled to a stirred reactor [12]. The presence of [Rh(CO)2I2]– as the principal Rh species was generally noted. Consistent with the model studies and the kinetics of the carbonylation reaction, which tended to first order in total Rh and MeI, the rate controlling step was of course the reaction of [Rh(CO)2I2]– with MeI. Later Eisenberg reported that the water gas shift reaction catalysed by Rh and HI involved [Rh(CO)2I2] – and [Rh(CO)2I4] –, (Eq. (48)) and (Eq. (49)), though the original publication was not substantially based on spectroscopic studies [48, 49]. [Rh(CO)2I2]– þ 2HI w [Rh(CO)2I4]– þ H2

(48)

[Rh(CO)2I4]– þ CO þ H2O w [Rh(CO)2I2]– þ CO2 þ 2HI

(49)

Subsequently Eby and Singleton of Monsanto investigated the interaction of the water gas shift and carbonylation reactions during Rh catalysed MeOH carbonylation and much of their mechanism was based on IR studies of the reactions [12]. They described the principal observations from in situ studies of batch carbonylations of MeOH to AcOH at the high [H2O] (i 15 % w/w) originally favored by Monsanto for operation of the commercial process. For much of the reaction the Rh is present as [Rh(CO)2I2]– and the reaction rate is constant. When the substrate has largely been consumed, in the later stage of the run, the rate falls and the Rh is progressively converted to [Rh(CO)2I4]–. At the same time the water gas shift rate, judged by the build up of CO2, is also increasing. The explanation of this behavior is that as MeOH/MeOAc are consumed [HI] increases (Eq. (31)).



(31)

This increased [HI] leads to the formation of [Rh(CO)2I4]–, a decrease in [Rh(CO)2I2] – and thus a decrease in carbonylation rate. Carrying out the carbonylation of MeOH to AcOH at lower [H2O] leads to decreased activity. At the same [MeOAc] and lower [H2O] more [Rh(CO)2I4] – is present. Eby and Singleton [12] described the factors influencing the water gas shift reaction rate which controls the amount of active [Rh(CO)2I2]– present during MeOH carbonylation. The key points were that increasing [HI] led to more [Rh(CO)2I4]– and the water gas shift rate increased while the rate controlling step was the Rh oxidation step, (Eq. (48)) above. However, on further increasing [HI] and [Rh(CO)2I4] – the water gas shift rate decreased again as the rate controlling step shifted to the reduction of [Rh(CO)2I4]–, (Eq. (49)) above. At lower [H2O], the same behavior was implied to occur but with the maximum water gas shift rate now at lower [HI]. In order to better understand and quantify these reactions, the water gas shift reaction catalysed by Rh and HI was studied by HP IR at BP. Some data for relative water gas shift rate and Rh speciation for a range of [HI] and two [H2O] are shown in Figure 5.15 [45]. Just as intermediates such as [RhMe(CO)2I3] – are not generally observed under the usual working conditions of MeOH carbonylation to AcOH, no intermediate species are observed directly for Rh catalysed water gas shift. However, the first step in oxidation of [Rh(CO)2I2]– to [Rh(CO)2I4]–, and thus a key step in the Rh/HI catalysed water gas shift reaction, the formation of [RhH(CO)2I3] –, has been reported by Bunel, who followed the reaction by NMR (Eq. (50)) [50]. [Rh(CO)2I2]– þ HI w [RhH(CO)2I3]–

(50)

Figure 5.15 HP IR data (transmission cell) for Rh speciation and water gas shift rate as a function of [HI] in AcOH at 180 hC and 30 bar. (i) v Rh(i) 10 % w/w H2O, (ii) y Rh(i) 15 % w/w H2O, (iii) k Water gas shift rate 10 % w/w H2O, (iv) x Water gas shift rate 15 % w/w H2O.

225

226


The overall decrease in carbonylation activity in Rh catalysed MeOH carbonylation at lower [H2O] can be explained both by an increase in the rate of formation of [Rh(CO)2I4]– due to the increased acidity of HI at lower [H2O] and a decrease in the rate of reduction of [Rh(CO)2I4]– by H2O and CO at lower [H2O]. Promoters such as iodide salts have a relatively greater effect on increasing the carbonylation rate at lower [H2O]. Celanese interpreted the increased carbonylation rate as being due to an increase in the Rh present in the active [Rh(CO)2I2] – form. They suggested an explanation based on the LiOAc/LiI cycle in aqueous media analogous to that proposed before by Eastman in the anhydrous carbonylation of MeOAc to Ac2O. The key effect of the metal species was to decrease the standing [HI] in the reaction medium, leading to a decrease in the rate of oxidation of [Rh(CO)2I2]– to [Rh(CO)2I4] – and thus, at the same time, in the water gas shift rate [14–16]. The same effect has been demonstrated with salts such as quaternary ammonium iodides in HP IR experiments carried out at BP [51] even though no substantial amounts of acetate salts are observed. Millenium have reported that phosphine oxides similarly lead to increased [Rh(CO)2I2]– at low H2O using HP IR to quantify Rh speciation [20]. All the results of Rh catalysed carbonylation of MeOH to AcOH from batch autoclave experiments, including HP IR studies, taken along with the HP NMR MeI/ MeOAc exchange data described above, are consistent with the promoters used at lower [H2O] acting to decrease the standing [HI] in the reaction medium. This occurs through the promoter catalysing the reaction of HI with MeOAc to regenerate MeI. Some promoters such as Li salts may be more effective through the presence of the LiOAc/LiI cycle, but any I– salt has a substantial effect. The I– may also have a promotional influence on the Rh/MeI oxidative addition, as shown in model studies. Lastly, although there is no HP IR evidence for how Ru carbonyl iodides work in Rh systems, their behavior in the HP NMR experiment is also consistent with a kinetic decrease of [HI] in the carbonylation system. In the Rh systems for MeOH carbonylation to AcOH, it is the property of HI as an acid which is most important in its influence on overall carbonylation activity, through its role in the conversion of [Rh(CO)2I2]– to [Rh(CO)2I4]–. 5.8.3

Ir Catalysed Carbonylation of MeOH to AcOH

In Ir catalysed carbonylation of MeOH to AcOH, the original mechanistic work of Forster was based extensively on HP IR studies. Overall, a greater number of species are seen compared with Rh systems. This is mainly due to one or two predominant species from the various catalytic cycles identified by Forster being present in certain regimes. The regime most closely studied more recently at BP has been that most relevant to Cativaä, referred to as Regime II by Forster, the ionic cycle. This is best illustrated by reference to the HP IR spectra from a batch experiment shown in Figure 5.4. In the earlier spectra, after the reaction has commenced and [H2O] and [MeOAc] are high, more of the catalyst is present as [IrMe(CO)2I3]–


than [Ir(CO)2I4]–. As the reaction proceeds and MeOAc and H2O are consumed, [IrMe(CO)2I3]– decreases to be replaced by [Ir(CO)2I4]–. At lower initial [H2O] the Ir species [Ir(CO)3I3] and [Ir(CO)3I] characteristic of the “neutral” Regime I of Forster are also observed as the reaction progresses, consistent with lower [HI] at lower [H2O] [11, 18]. Compared with Rh systems, where the two principal species are well resolved, it can be seen that as well as more Ir species the bands also overlap, making quantification more difficult. Qualitatively some conclusions can be drawn from the spectra. Forster identified for example that in the presence of I–, a potent catalyst poison, much of the Ir could still be present as [IrMe(CO)2I3]–. Similarly, as [H2O] is increased the carbonylation rate falls. This is consistent with increased [I–] since equilibrium [HI] increases with [H2O] as described above, inhibiting the migratory insertion reaction of [IrMe(CO)2I3]–. When Ru carbonyl iodides are used as promoters for Ir catalysed carbonylation of MeOH to AcOH, the accurate quantification of Ir species becomes more difficult because the bands of the Ru species overlap those of the Ir species and they have larger extinction coefficients, so dominating the spectra. In batch reactions followed by HP IR, the Ru bands present initially before Ir is added and carbonylation commences have been assigned to neutral Ru carbonyl iodides and [Ru(CO)3I3]– [18]. On addition of Ir, the carbonylation cycle liberates HI and much of the Ru is then present as [Ru(CO)3I3]–. Spectra of samples from continuously operating pilot plant using Ir and Ru recorded offline in a total internal reflectance cell at lower temperatures and pressures are more tractable and both [IrMe(CO)2I3]– and [Ir(CO)2I4]– can be identified after careful subtraction of [Ru(CO)3I3]–bands [52]. In this respect, offline Ir catalyst system samples seem to be less prone to change with time and more representative of reactor composition than Rh catalyst system samples which can undergo significant changes during handling [53]. InI3, which has a similar promotional effect to Ru carbonyl iodides for Ir catalysed MeOH carbonylation to AcOH, of course has no carbonyl bands to obscure the Ir species. Both [Ir(CO)2I4]– and [IrMe(CO)2I3]– are observed under carbonylation conditions by HP IR, as in the absence of InI3 [18]. The various spectroscopic studies on Ir catalysed carbonylation of MeOH to AcOH suggest that the overall carbonylation activity is strongly influenced by the levels of HI or other sources of I– in the reaction medium. Related to this is the concept that key steps of the cycle with Ir may pass through neutral Ir species, even in the so-called ionic regime where the Ir is often largely present as [IrMe(CO)2I3]– [18]. As with Rh systems, certain promoters such as Ru carbonyl iodides, which can be shown by HP NMR to promote the reaction of HI with MeOAc, can also therefore be promoters for the Ir system. However, in contrast to Rh processes it is not only the property of HI as an acid which is important in determining the activity of Ir processes under commercial conditions but the property of being a source of I–, since the rate controlling step in Ir catalysed MeOH carbonylation to AcOH involves the migratory insertion reaction, which is strongly influenced by I–.

227

228


5.9

Conclusions: Spectroscopy and Understanding Carbonylation Mechanisms

The present understanding of the mechanism of Rh or Ir catalysed carbonylation of MeOH to AcOH and Rh catalysed carbonylation of MeOAc to Ac2O is due in large part to the application of spectroscopy, particularly IR and NMR. As well as characterizing complexes involved in the main catalyst cycles, spectroscopy has contributed to the measurement of the kinetics of these cycles and to byproduct reactions. The major catalyst species present under working conditions of the catalyst systems have been identified for all the systems. Individual reaction steps involving interconversion of catalyst complexes have been isolated and studied in model reactions. IR has been very important in these studies with metal carbonyl species, including the identification of Ru promoter species in MeOH carbonylation. At the same time NMR has proved particularly useful in determining factors influencing the important reaction of HI or AcI with MeOAc and understanding how different classes of promoters can affect the standing concentrations of these intermediates under process conditions. The result of these studies has been to show how the differences between these apparently very similar processes arise. In the Rh catalysed carbonylation of MeOH to AcOH, it is the control of [HI] which determines how much of the catalyst is present in the active form as well as the relative rate of the competing water gas shift cycle and it is the property of HI as an acid, which is important. In the Ir catalysed carbonylation of MeOH to AcOH, it is again the control of [HI] which is important, not so much because of the shift between active and inactive forms of the catalyst as with Rh but because of the inhibition of the carbonylation cycle by I– and thus because of the property of HI as an iodide rather than as an acid. Finally, in the Rh catalysed carbonylation of MeOAc to Ac2O it is the control of [AcI] rather than [HI] which determines the overall carbonylation rate because AcI is involved in the rate controlling reaction with MeOAc. While the main carbonylation cycles are now understood in considerable detail for these apparently simple catalytic systems, there will undoubtedly be considerably more work done on these and related sytems to understand the factors influencing the principal steps of oxidative addition, migratory insertion and reductive elimination and, in particular, further work to understand the unwanted reactions that lead to by-products.

Acknowledgments

I gratefully acknowledge the permission of BP Chemicals Ltd to publish this work and the contribution of all the colleagues at BP who have carried out the studies of all aspects of carbonylation chemistry over many years. In particular, from the outset of use of the techniques in BP, Jane Boyle and Mike Taylor have been involved

References

with HP NMR, and Rob Watt with HP IR. The overall mechanistic interpretation draws greatly on discussions with Glenn Sunley and our collaborators at Sheffield University, Tony Haynes and Peter Maitlis.

References [1] Weymouth, F. J.; Millidge, A. F. [2]

[3]

[4] [5] [6]

[7] [8] [9] [10] [11] [12] [13] [14]

[15]

[16]

[17]

[18]

Chem. Ind. 1966, 887. Hohenschutz, H.; von Kutepow, N.; Himmele, W. Hydrocarbon Process. 1966, 37, 383. Roth, J. F.; Craddock, J. H.; Hershman, A.; Paulik, F. E. Chemtech 1971, 600. Hewlett, C. GB Pat. 1468940, 1974. Polichnowski, S. W. J. Chem. Educ. 1986, 63, 204. Smith, B. L.; Torrence, G. P.; Aguilo, A.; Alder, J. S. Eur. Pat. 0161874, 1985. Cooper, J. B. Eur. Pat. 0087870, 1985. Chem. Ind. (London) 1996, 483. Sunley, G. J.; Watson, D. J. Catal. Today 2000, 58, 293. Forster, D. J. Am. Chem. Soc. 1976, 98, 846. Forster, D. J. Chem. Soc., Dalton Trans. 1979, 1639. Eby, R. T.; Singleton, T. C. Appl. Ind. Catal. 1983, 1, 275. Hewlett, C. US Pat. 4559183, 1985. Murphy, M. A.; Smith, B. L.; Torrence, G. P.; Aguilo, A. Inorg. Chim. Acta. 1985, 101, L47. Murphy, M. A.; Smith, B. L.; Torrence, G. P.; Aguilo, A. J. Organomet. Chem. 1986, 303, 257. Smith, B. L.; Torrence, G. P.; Murphy, M. A.; Aguilo, A. J. Mol. Catal. 1987, 39, 115. Maitlis, P. M.; Haynes, A.; Sunley, G. J.; Howard, M. J. J. Chem. Soc., Dalton Trans. 1996, 2187. Haynes, A.; Maitlis, P. M.; Morris, G. E.; Sunley, G. J.; Adams, H.; Badger, P. W.; Bowers, C. M.; Cook, D. B.; Elliott, P. I. P.; Ghaffar, T.; Green, H.; Griffin, T. R.; Payne, M.; Pearson, J. M.; Taylor, M. J.; Vickers, P. W.;

[19]

[20] [21]

[22]

[23]

[24]

[25] [26] [27] [28] [29]

[30] [31]

[32]

[33]

Watt, R. J. J. Am. Chem. Soc. 2004, 126, 2847. Howard, M. J.; Jones, M. D.; Roberts, M. S.; Taylor, S. A. Catal. Today 1993, 18, 325. Hallinan, N.; Hinnenkamp, J. A. US Pat. 6103934, 2000. Yoneda, N.; Shiroto, Y.; Hamato, K.; Asaoka, S.; Maejima,T. US Pat. 5334755, 1994. Wegman, R. W.; Abatjoglou, A. G.; Harrison, A. M. J. Chem. Soc., Chem. Commun. 1987, 1891. Baker, M. J.; Giles, M. F.; Orpen, A. G.; Taylor, M. J.; Watt, R. J. J. Chem. Soc., Chem. Commun. 1995, 197. Howard, M. J.; Sunley, G. J.; Poole, A. D.; Watt, R. J.; Sharma, B. K. Stud. Surf. Sci. Catal. 1999, 121, 61. Morris, D. E.; Tinker, B. H. Chemtech 1972, 554. Schrod, M.; Luft, G. J Mol. Catal. 1983, 22, 169. Rigby, W.; Whyman, R.; Wilding, K. J. Phys. (E) Sci. Instrum. 1970, 3, 572. Roe, D. C. J. Magn. Reson. 1985, 63, 388. Boyle, J. D.; Morris, G. E.; Roberts, D.; Taylor, M. J. BP Chemicals, unpublished work. Hickey, C. E.; Maitlis, P. M. J. Chem. Soc., Chem. Commun. 1984, 1609. Haynes, A.; Mann, B. E.; Gulliver, D. J.; Morris, G. E.; Maitlis, P. M. J. Am. Chem. Soc. 1991, 113, 8567. Haynes, A.; Mann, B. E.; Morris, G. E.; Maitlis, P. M. J. Am. Chem. Soc. 1993, 115, 4093. Griffin, T. R.; Cook, D. B.; Haynes, A.; Pearson, J. M.; Monti, D.; Morris, G. E. J. Am. Chem. Soc. 1996, 118, 3029.

229

230

5 Carbonylation of Methanol to Acetic Acid and Methyl Acetate to Acetic Anhydride [34] Forster, D. Adv. Organomet. Chem. [35] [36]

[37]

[38]

[39] [40]

[41] [42]

1979, 17, 255. Howe, L. Bunel, E. E. Polyhedron 1995, 14, 167. Bassetti, M.; Monti, D.; Haynes, A.; Pearson, J. M.; Stanbridge, I. A.; Maitlis, P. M. Gazz. Chim. Ital. 1992, 122, 391. Pearson, J. M.; Haynes, A.; Morris, G. E.; Sunley, G. J.; Maitlis, P. M. J. Chem. Soc., Chem. Commun. 1995, 1045. Whyman, R.; Wright, A. P.; Iggo, J. A.; Heaton, B. T. J. Chem. Soc., Dalton Trans. 2002, 771. Gold, V.; Hilton, J. J. Chem. Soc. 1955, 838. Moore, P.; Morris, G. E. Warwick University and BP Chemicals, unpublished work. Rizkalla, N. GB Pat. 1538783, 1979 Gauthier-Lafaye, J.; Perron, R. Methanol and Carbonylation EditionsTechnip, Paris 1987, p. 152.

[43] Atkins, P. W. Physical Chemistry

Freeman, New York 1994, p. 874. [44] March, J. Advanced Organic Chemistry

Wiley, NewYork 1985, p. 334. [45] Morris, G. E.; Watt, R. J. BP Chemi-

cals, unpublished work. [46] Schrod, M.; Luft, G. Ind. Eng. Chem.

Prod. Res. Dev. 1981, 20, 649. [47] Giles, M. F.; Watt, R. J. BP Chemicals,

unpublished work. [48] Cheng, C-H.; Hendriksen, D. E.;

[49]

[50]

[51] [52] [53]

Eisenberg, R. J. Am. Chem. Soc. 1977, 99, 2791. Baker, E. C.; Hendriksen, D. E.; Eisenberg, R. J. Am. Chem. Soc. 1980, 102, 1020. Roe, C. D.; Sheridan, R. E.; Bunel, E. E. J. Am. Chem. Soc. 1994, 116, 1163. Watt, R. J. BP Chemicals, unpublished work. Williams, B. L. BP Chemicals, unpublished work. Jones, J. H.; Morris, G. E. BP Chemicals, unpublished work.

6 Rhodium Catalyzed Hydroformylation Paul C. J. Kamer, Joost N. H. Reek, and Piet W. N. M. van Leeuwen

6.1

Introduction

The catalytic formation of aldehydes from alkenes, carbon monoxide and hydrogen is a reaction with an extremely high atom economy, even though the reaction is almost seventy years old. Rhodium catalysts can give high selectivities to aldehydes while cobalt catalysts lead to alkanes as the byproduct, which represents a cost factor, but not an environmental factor, as the alkanes can be added conveniently to fuel. To date the hydroformylation reaction is one of the most important reactions employing a homogeneous catalyst, covering an annual production of almost eight million tons of aldehydes and alcohols [1]. Consequently, hydroformylation is one of the most extensively studied homogeneous catalytic processes. Improvement of rates and selectivities by ligand design and mechanistic aspects have received much attention. Although the reaction kinetics have been investigated a number of times, relatively little has been reported on the characterization of the intermediates present during the reaction. Nowadays high-pressure (HP) spectroscopic techniques can be applied routinely to identify organometallic compounds present under high pressures. HP-IR is most often applied to identify in situ hydroformylation intermediates, but only information about carbonyl ligands is obtained and complete characterization can often not be achieved. HP-NMR spectroscopy is a powerful technique for the characterization of complexes formed under high pressures. However, in situ characterization of complexes present during catalysis cannot be performed using a standard HP-NMR tube because of limited gas diffusion. Slow diffusion causes a shortage of the gaseous reactants and a concomitant change in the species occurring. Continuous supply of gas and optimal mixing is achieved in an HP-NMR flow cell [2] and therefore this is an elegant tool for the characterization of hydroformylation reaction intermediates. Several solutions for the problem of mass transfer limitation in NMR studies have been reported in the literature [3, 4]. The combination of in situ HP-IR and HP-NMR spectroscopy can lead to complete structural analysis of the proposed intermediates in the catalytic cycle. Mechanisms in Homogeneous Catalysis. A Spectroscopic Approach. Edited by Brian Heaton Copyright c 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN 3-527-31025-8

232

6 Rhodium Catalyzed Hydroformylation

In addition to characterization of intermediates, in situ spectroscopic techniques can be applied in kinetic studies, providing additional mechanistic insight. Also, isotopic labeling studies have proven very useful, especially when studying the individual steps of a catalytic cycle [5, 6] The large impact of phosphine ligands became evident with the important discovery of the Wilkinson hydrogenation catalyst, RhCl(PPh3)3 [7]. Substitution at the aromatic ring of the ligand revealed an electronic effect on the reaction rate, illustrating the distinct ligand effect on the reactivity of a transition metal complex. It should be noted that for both this reaction and many others, the question remains whether the effect is an intrinsic effect on the catalyst’s properties or an effect on equilibria in which the species are involved. Nevertheless, an impressive number of phosphine ligands have been applied in many catalytic reactions and it has become obvious that the steric and electronic properties of the ligands have an enormous effect on the reactivity of metal complexes. The huge effects that were observed called for a systematic classification of ligand properties. Strohmeier showed that the IR carbonyl frequencies of metal complexes could be used as a measure of the electronic properties of the ligands [8]. Tolman introduced a systematic approach to describe electronic and steric ligand effects [9]. The electronic parameter x is based on the difference in the IR frequencies of Ni(CO)3L and the reference compound Ni(CO)3(PtBu3), similar to the method introduced by Strohmeier. For phosphorus ligands the cone angle u is defined as the apex angle of a cylindrical cone, centered at 2.28 A from the center of the P atom, which touches the outermost atoms of the model. The concepts of the natural bite angle and the flexibility range for diphosphine ligands were introduced by Casey and Whiteker and are a means of predicting chelational preferences of bidentate ligands [10]. The natural bite angle (bn) is defined as the preferred chelation angle determined only by the ligand backbone and not by the metal valence angles. The flexibility range is defined as the accessible range of bite angles that induce less than 3 kcal mol–1 excess strain energy compared to the calculated natural bite angle. The development of organotransition metal chemistry has contributed greatly to the enormous growth of homogeneous catalysis [11]. Knowledge of bonding and reactivity in organometallic chemistry has been of great support to catalysis studies [12]. The reactivity of organotransition metal complexes is dependent on the ligand environment of the metal. By changing the ligands, the performance of the catalyst can be directed and sometimes the effects can even be predicted. In transition metal catalysis, extensive research has been devoted to fine-tuning the selectivity and activity of catalysts by means of ligand modification, simply by looking at electronic and steric effects. In this way many processes have been improved and several new ones have been developed. An in-depth understanding of a catalytic reaction, however, requires mechanistic studies under actual catalytic conditions. Characterization of reaction intermediates during the rhodium-catalyzed hydroformylation has only rarely been performed. This is probably due to the difficulties arising when spectroscopic studies need to be carried out under high pressure and temperature, examples being mass transfer lim-

6.2 Study of Catalytic Resting States

itations in combination with low concentrations of the catalytically active metal complex. In this review, we will discuss the use of in situ spectroscopic techniques, in combination with kinetic and isotopic labeling studies, to obtain a detailed mechanistic insight of the rhodium catalyzed hydroformylation.

6.2

Study of Catalytic Resting States

NMR is perhaps the most powerful technique for the characterization of catalytic reaction intermediates. NMR data for the various nuclei (1H, 13C, 15N, 31P, 103Rh, 195 Pt etc.) can be combined, sometimes leading to a full structural characterization of the complexes involved. Although this technique is very informative, it also has some drawbacks. The concentration range necessary for reasonable signal to noise ratios (10–100 mM) is well above the concentration region used for catalysis experiments (often I1 mM). Metal–ligand equilibria may shift considerably upon concentrating the solutions, which hampers direct comparison of the data observed in catalysis and in the NMR experiments. Most reactions studied are fast because of the high catalyst concentrations needed. Reagents in the form of dissolved gases will be rapidly consumed and diffusion and mass transfer become rate limiting. As a result, the observed complexes will be different from the intermediates under actual catalytic conditions. The power of in situ NMR as an analytical tool for the characterization of homogeneous catalysts is evident from the study of rhodium complexes in hydroformylation. The actual catalyst is a hydridorhodiumdicarbonyldiphosphine, HRh(CO)2P2, which has a trigonal bipyramidal structure. Two isomers are possible with both phosphines (ee) or with both carbonyls (ea) in the equatorial plane. The position and multiplicity of the resonances and the coupling constants observed in the NMR spectra are characteristic of the structure of the complexes. The position of the hydride resonance in the 1H NMR spectrum varies between –8 and –12 ppm depending on the electronic properties of the other ligands coordinated to the rhodium atom. The value of 1JRhH observed for the ee isomer is usually approximately 3 Hz, while it is larger for the ea isomer. The value of 2JPH gives information on both the coordination mode of the phosphorus ligands and the degree of distortion of the trigonal bipyramidal structure. Large values of 2JPH, ca. 100 Hz, are characteristic of a trans-phosphine-hydride coordination mode; a trans-phosphite-hydride coordination mode can give values of 2JPH up to 180 Hz. Small values of 2JPH

H

H

OC

P Rh

P

CO

Rh P

P ea

CO ee

CO

233

234


H

O

O C Rh C O

H P

P

C O

P

b

a Figure 6.1

O

Ph3P Rh

P

Bis-equatorial coordination of xantphos.

(z 3 Hz) are observed for a cis-phosphorus-hydride coordination mode. If there is fast exchange of the phosphorus ligands between the equatorial and apical positions, an averaged resonance and coupling constant are observed for the equilibrium mixture of the ee and ea isomers. Several complexes containing xantphos-type ligands, which induce a wide biteangle, have been isolated and identified [13]. They contain the xantphos ligand co-ordinating mainly in the bis-equatorial fashion as shown in Figure 6.1. Detailed analysis has shown that, in many cases, the co-ordination was not entirely as planned. The tris-phosphine complex (see b in Figure 6.1) shows an interesting 31 P NMR spectrum; the values of 2JPP are relatively large for two equatorial phosphorus ligands (120–150 Hz) since, when two phosphorus atoms are in cis-positions, the values of 2JPP are usually small (0–80 Hz). A trans-P–P geometry leads to large coupling constants (several hundred Hz). Table 6.1 shows data for rhodium complexes containing xantphos ligands in which the electronic properties were systematically varied [13]. With few exceptions, we see that the rate of the hydroformylation reaction increases when the rhodium center is more electron deficient. This is as expected, since dissociation of

Table 6.1

Electronic effects in xantphos ligands. H PR2 OC

Rh

O

S

X

R=

PR2 C O

X

–dRh

1

CF3 Cl H F Me MeO NMe2

850 840

4.4 5.9 6.6 6.3 7.3 7.3 8.8

835 831 825 814

JHRha

1

JPRha

135 132 128 131 126 125 122

2

JPHa

3.6 8.4 15 11 18 21 28

Rateb

% Isomc

l : bd

158 68 107 75 78 45 29

7 7 5 6 5 6 5

89 68 50 52 44 37 45

J in Hz. b Rate in mol–1 h–1. c % isom ¼ selectivity for 2-octene. d l : b ¼ linear to branched ratio of aldehyde product.

a

6.2 Study of Catalytic Resting States tBu H O

C

P(OR)2

H P

O O

Rh

C

O C a

O

Figure 6.2

C tBu

b

O

O C

O C P

P(OR)2

H

O

Rh

Rh P

O

P c

O

O

Diphosphite rhodium hydride complexes.

carbon monoxide is easier and alkene complexation becomes stronger. As we have seen before, in most catalytic systems this enhances the rate of reaction. The phosphite complex, a in Figure 6.2, is a typical example of the new generation catalysts (UCC has commercialised a process using a diphosphite catalyst). The NMR characteristics are of course the same for a and b: x x

a small coupling constant, 2JPH, I 30 Hz if the two phosphorus atoms are inequivalent, a large value of 2JPP, 250 Hz.

Complex c in Figure 6.2 shows the opposite: x x

one large and one small value of 2JPH, 180 and –30 Hz if the two phosphorus atoms are inequivalent, a small value of 2JPP I 80 Hz.

At room temperature, the complexes are extremely fluxional and in fact the proton spectrum shows a doublet (due to coupling with 103Rh, spin 1/2) of triplets due to the averaged coupling constants of the two phosphorus nuclei. The details of chemical shifts and coupling constants are now well known and can be used for the determination of the solution structures [14]. When ligand exchange is slow on the NMR time scale, the structure can be easily deduced from the spectra. A bidentate ligand such as PPh2NMeCHMeCHPhOPPh2 [15] (we neglect here the asymmetric nature of the ligand), having two very distinct resonances for the amido phosphorus and the phosphinite phosphorus nuclei, gives 1H and 31P spectra that are readily interpreted (see Figure 6.3 for structure). The hydride reonance is a doublet ca. –8.7 ppm with a large coupling to the trans-phosphorus (116 Hz) and smaller couplings with rhodium (10.1 Hz) and the phosphinite phosphorus nucleus (9.9 Hz). In the phosphorus spectra, we find the expected couplings with rhodium, the mutual phosphorus coupling, and in the proton-coupled spectra we see the proton coupling to the transphosphorus.

H Ph Ph CO P Rh CO O Ph P Ph N Ph Me Me

Figure 6.3 Rhodium complex containing a bidentate ligand with two different donor atoms.

235

236


R P P

Rh

P

R

R P

CO

Rh

Rh

P

CO CO

P

CO

CO a

P b

c

Ph Ph

O CO Rh

P

P

CO

P

C8H17

O CO Rh

P

CO

P

Rh

CO

P

d e Intermediates identified by Brown [16].

Figure 6.4

O CO

f

Brown and Kent [16] studied the species involved in the hydroformylation of styrene and 1-decene at ambient pressure and temperatures of –80 to 25 hC using the catalyst precursor, HRh(PPh3)3(CO). Under an atmosphere of syn-gas, HRh(PPh3)3(CO) (a in Figure 6.4) transforms into HRh(PPh3)2(CO)2 (b and c in Figure 6.4) and free PPh3. Typical concentrations are 20 mmol l–1, which is considerably higher than the concentrations used in typical catalytic experiments (I 0.5 mmol l–1). At room temperature, a rapid exchange with free phosphine and CO is observed. Two conformations are observed for the bis-phosphine complex, isomer b having two equatorial phosphines and isomer c having one equatorial and one axial phosphine (Figure 6.4). The bis-equatorial isomer, b, is the dominant species. When b/c react with styrene, the branched acyl complex, d (Figure 6.4), is formed; now the equatorial/axial isomer is the major species and this is presumably due to steric factors. The branched isomer, d, slowly isomerises to the linear acyl complex, e

O H P

2

CO

Rh

P

CO

P

Rh

Rh

+ H2

P

CO P

CO

P O

O

O P

CO

P

Rh

Rh

P

P

CO

P

P Rh

P

P

O a orange Figure 6.5

Formation of inactive dimers.

+ 2 CO

Rh O

b red

6.3 IR studies on Ligand-free Rhodium Carbonyl Catalysts

(Figure 6.4). This shows that the kinetic product in the initial reaction with styrene is the branched styryl intermediate. With 1-octene, only the linear acyl complex, f (Figure 6.4), is observed. Interestingly, a fast intramolecular exchange of the two distinct phosphines in f can be observed and, at higher temperatures, there is also a rapid exchange with free CO and PPh3. Formation of dimeric species, a and b (Figure 6.5), is a general feature of these studies [17], but the subsequent reaction mechanism of these dimers with dihydrogen could not be elucidated in this study.

6.3

IR studies on Ligand-free Rhodium Carbonyl Catalysts

Recently, thorough studies providing great detail on the rhodium carbonyl catalysts have been reported by Pino and Garland [18]. They studied the reaction of 3,3-dimethyl-1-butene, CO, H2, using Rh4(CO)12 at 4–14 hC. This substrate was chosen because it lacks the possibility to isomerise, while still giving an acceptable rate of hydroformylation. The cluster carbonyl reacts slowly with the syn-gas and alkene to give the monomeric acyl complex RC(O)Rh(CO)4. Clearly, the reaction of the acyl complex with dihydrogen is the rate-determining step in this catalytic system. Thus, this confirms early results obtained by Mark [19]. The rate equation for the reaction with hydrogen under 5–30 bar of H2 and 10– 40 bar of CO) is shown in Eq. (1): v ¼ k [RC(O)Rh(CO)4]1[CO]–1.1[H2]1[3,3-DMB]0.1

(1)

The rate of hydroformylation was proportional to the concentration of the acyl complex. The apparent activation parameters were DH‡ ¼ 49.3 kJ mol–1 and DS‡ ¼ 121 J mol–1 K–1. Both the activation parameters and the reaction order are consistent with the hydrogenolysis reaction being rate determining. The low order of 0.1 in alkene suggests that the “rate-determining step” is not purely the reaction with hydrogen and that either a pre-equilibrium also contributes or one of the earlier steps in the cycle is also somewhat slower. Assuming that all the rhodium occurs as rhodium acyl species, the “ideal” rate equation, applying the steady state approximation, is shown in Eq. (2): v=

ks6 k7 [Rh][H2 ] ks6 Sk6 [CO] + k7 [H2 ]

(2)

Garland et al. have developed a powerful method for the reconstruction of individual pure component spectra from complex catalytic mixtures [20]. Using this band-target entropy minimization (BTEM) protocol, he was able to identify the mononuclear rhodium acyl intermediate in the hydroformylation reaction of 3,3-dimethylbut-1-ene starting from Rh4(s-CO)9(m-CO)3 as catalyst precursor [21]. In addition to the catalyst precursor and the more stable decomposition product

237

238


Rh6(CO)16 they revealed the presence of an unidentified complex which was suggested to be the previously unknown species Rh4(s-CO)12. The BTEM protocol is an extremely powerful technique to recover pure component spectra of unknown species, even when present at very low concentrations. This was illustrated by a detailed mechanistic study of the promoting effect of HMn(CO)5 on the Rh4(CO)12 catalyzed hydroformylation of 3,3-dimethylbut-1ene [22]. A dramatic increase in the hydroformylation rate was found when both metals were used simultaneously. Detailed in situ FTIR measurements using the BTEM protocol indicated the presence of homometallic complexes only during catalysis. The metal complexes that were identified under catalytic conditions were RC(O)Rh(CO)4, Rh4(CO)12, Rh6(CO)16, HMn(CO)5, and Mn2(CO)10 (see Figure 6.6). The kinetics of product formation showed an overall product formation rate, Eq. (3): vtotal ¼ k1[RC(O)Rh(CO)4][CO]–1[H2] þ k2[RC(O)Rh(CO)4][HMn(CO)5][CO]–1.5 (3) The first term represents the classic unicyclic rhodium catalysis, while the second indicates a hydride attack on an acyl species. These spectroscopic and kinetic results strongly suggested the presence of bimetallic catalytic binuclear elimination as the origin of synergism of both metals rather than cluster catalysis. This detailed evidence for such a catalytic mechanism, and its implications for selectivity and nonlinear catalytic activity illustrate the important mechanistic knowledge that can be revealed by this powerful in situ spectroscopic technique.

Figure 6.6 Recovered pure component spectra of the organometallic species using BTEM: (a) RC(O)Rh(CO)4, (b) Rh4(CO)12, (c) Rh6(CO)16, (d) HMn(CO)5, and (e) Mn2(CO)10. Reproduced from Ref. [22] with permission.

6.4 Phosphite Ligands

6.4

Phosphite Ligands

In the sixties, Pruett and Smith already recognized the complicated effects of ligand structure and process conditions on the product distribution and the rate of the catalytic reaction [23]. The presence of several catalytically active rhodium complexes in the reaction mixture often obscures systematic studies of ligand effects on the hydroformylation reaction (see Scheme 6.1). These complexes, containing different numbers of phosphorus ligands, are in equilibrium, and complexes containing up to three phosphorus ligands can all be active as hydroformylation catalysts. The composition of the equilibrium mixture is dependent on many reaction parameters, such as type of ligand, concentration, temperature and pressure. L RhH(CO)4 Scheme 6.1

L

L RhH(CO)3L

RhH(CO)2L2 R

L hH(CO)L3

RhHL4

CO CO CO CO Actual rhodium catalysts containing various phosphorus and carbonyl ligands.

The complexes in Scheme 6.1 react at different rates and have different selectivities in the hydroformylation of 1-alkenes. In general, phosphorus ligands are stronger s-donors and weaker p-acceptors than carbonyl ligands and introduce more steric hindrance around the metal. Both stronger CO bonding and steric hindrance hamper the alkene addition. Therefore, the overall rate of the hydroformylation reaction decreases with the number of phosphorus ligands coordinating to the rhodium, whereas, simultaneously, the selectivity increases. For small phosphites, an excess of phosphite results in the formation of an inactive rhodium complex. In laboratory experiments, this can be used to quench reaction samples by addition of a large excess of a non-bulky ligand such as tributyl phosphite. Pruett and Smith studied a wide variety of phosphite and phosphine ligands [23]. The general trend they found is an increasing selectivity for the linear aldehyde when the electron withdrawing properties of the ligand increase. A donating substituent like 4-methoxy resulted in a decrease of the linear to branched ratio, whereas the 4-chloro substituted phenyl phosphite gave a relatively high l : b ratio of 13 (see Table 6.2). The expected higher reaction rates with the stronger p-acceptor phosphite ligands are less obvious from their results and the amounts of isomerized alkenes were not reported either. The use of ortho-substituted aryl phosphites gave lower selectivity for the linear product, whereas no remarkable effects on the rate of the reaction were reported. Generally, one would expect that increasing steric hindrance in the catalytically active rhodium complex would result in lower reaction rates. In this respect, the results of Van Leeuwen and Roobeek seemed at first to be contradictory. They used the very bulky tris(ortho tert-butylphenyl)phosphite 1a (Chart 6.1) as a ligand and found high reaction rates in the rhodium catalyzed hydroformylation of other-

239

240

6 Rhodium Catalyzed Hydroformylation Table 6.2 Hydroformylation using rhodium bulky monophosphite catalysts.a

T, hC

pCO

pH2

Alkene

% Isom.

Rate,b mol. (mol Rh)–1 h–1

l:b

P(OPh)3c 90 P(OC6H4 -p-Cl)3c 90 P(OC6H4 -p-OMe)3c 90 80 PPh3d 1ae 90 80 1bf 2e 95 2e 120 80 1ae

3 3 3 10 10 10 4.8 4.8 7

3 3 3 10 10 10 4.8 4.8 7

n. d. n. d. n. d. 1.5 n. d. 13 15 – –

– – – 2200 7100 40,000 300 1000

6.1 13.3 4.9 2.8 3.3 1.9 19 1.9 i100

1bf 1bf 1ae

10 11 7

10 11 7

1-octene 1-octene 1-octene 1-octene 1-heptene 1-octene 1-heptene 2-heptene 2-methyl-1hexene c-hexene styrene limonene

– – –

500 10,000 1700

– – i100

Ligand

80 70 80

Conditions: 0.1–1 mM Rh, L/Rh ¼ 10–20, [alkene] ¼ 0.5–1 M in benzene or toluene. b Initial rate. n. d. ¼ not determined. c Data taken from Ref. [23]. d Data taken from Ref. [58]. e Data taken from Ref. [24]. f Data taken from Refs. [25] and [27]. a

R t-Bu O P O O

t-Bu t-Bu

t-Bu

O P O O

CF3 R

t-Bu

CF3 O P O O

CF3 t-Bu

t-Bu

R 1aR=H 2 3 b R = Me Chart 6.1 Structure of bulky phosphites, 1 and 3, and an electron-poor phosphite, 2.

wise unreactive alkenes eg. 1,2- and 2,2-dialkylalkenes (see Table 6.2) [24]. The high reactivity was explained by the exclusive formation of monoligated rhodium phosphite complexes, which was corroborated by in situ IR and NMR studies [25]. On addition of tris(2-tert-butyl-4-methylphenyl)phosphite to the rhodium precursor, Rh(CO)2acac, a spectrum is observed immediately in which the two CO vibrations (2080, 2012 cm–1) are replaced by one at 2015 cm–1, as a result of substitution of one CO by a phosphite, to give Rhacac(CO)P (P ¼ phosphite). Preferential formation of monoligated complexes in the presence of CO is due to the large cone angle of the bulky phosphite ligand (u ¼ 190h). Coordination of two ligands does occur in related Rh chloro carbonyl complexes, in which the ligands can occupy mutually trans positions [26]. Pressurizing a HP-IR cell containing Rhacac(CO)P with 10 bars of syn-gas slowly reversed the reaction to Rh(CO)2acac. Heating

6.4 Phosphite Ligands Table 6.3 Observed IR frequencies in 2200–1600 cm–1 region after addition of substrate to HRh(CO)3P (P ¼ tris(2-tert-butyl-4-methylphenyl) phosphite.

Substrate

Observed frequencies (cm-1)

none cyclohexene 1-octene styrene pentafluorostyrene

2093, 2093, 2080, 2080, 2079,

a

2043, 2043, 2019, 2022, 2027,

2013 2013 1996, 1690a 1998 1998

When performed with the rapid-scan method.

this stirred mixture for about 2 h at 40 hC resulted in complete conversion to the active hydroformylation catalyst, HRh(CO)3P, as was previously observed by Jongsma et al. [25] (nCO at 2093, 2043 and 2013 cm–1). An additional amount of a decomposition product (2064, 2031 cm–1) is also detected. After addition of the less reactive, internal alkene, cyclohexene, the carbonyl region of the IR spectrum remained unchanged. The infrared absorptions in the 2200–1600 cm–1 region, that appear after injection of the substrate, are given in Table 6.3. During the catalytic reaction, the absorptions remain the same with an additional band growing rapidly at approximately 1730 cm–1 originating from the product aldehydes. The data in the table show that the IR spectrum in the presence of cyclohexene is identical to that of HRh(CO)3P, indicating that this is the resting state during the catalytic reaction. This is in accordance with rate determining alkene addition or hydride migration, as could be inferred from kinetic studies, which showed a first order dependence on cyclohexene concentration, a negative order in CO, and zero order in hydrogen [27, 28]. It should be noted that the study of intermediates during fast reactions requires an efficient flow cell to avoid mass transfer limitations [28], see also Chapter 3. For 1-alkenes as substrate, the IR spectra have nCO bands that are shifted somewhat to lower energies, but the pattern in the carbonyl region is similar to that of the spectrum of HRh(CO)3P; all absorptions are broad or contain a shoulder. The metal–acyl frequency could not be observed with this spectroscopic technique, because of the immediate appearance of a strong overlapping absorption of the aldehyde product. Therefore, the experiment for 1-octene was repeated using the rapidscan option where a sequence of spectra of up to 80 scans s–1 could be obtained. Figures 6.7 and 6.8 show a selection of the spectra of the two important frequency regions. Figure 6.7 shows the section where the acyl, alkene and bridging CO vibrations are located, whereas Figure 6.8 shows the metal terminal CO region. The difference in time between the first and the last spectrum is 0.82 s. An absorption at 1690 cm–1 is detected within 1 s and initially even two bands can be distinguished (Figure 6.7). When the reaction proceeds, the aldehyde band appearing at 1734 cm–1 rapidly conceals both bands. After all the substrate had been consumed, the spectrum belonging to the hydrido rhodium complex reappeared. In Figure 6.8, the rapid conversion of the starting hydrido rhodium complex into the predominant species is depicted. The absorption patterns are shifted to lower frequen-

241

242


Acyl region of IR spectra recorded with rapid-scan method after addition of 1-octene to HRh(CO)3P (P ¼ tris(2-tert-butyl-4-methylphenyl) phosphite. * ¼ 1-octene. Reproduced from Ref. [28] with permission. Figure 6.7

Figure 6.8 Terminal CO region of the IR spectra recorded with the rapid-scan method after addition of 1-octene to HRh(CO)3P (P ¼ tris(2-tert-butyl-4-methylphenyl) phosphite. Reproduced from Ref. [28] with permission.

cies, but similar to those of HRh(CO)3P, indicating the presence of a complex with a structure analogous to the hydrido rhodium complex. The frequencies differ for the various substrates, which suggests that a substrate-derived fragment is now linked to the rhodium complex. Migration of the hydride, subsequent to coordina-

6.4 Phosphite Ligands

OC

H

R

OC

Rh

R'

OC

Rh

P

Rh

P CO

CO

4b O OC

R'

4c O OC

Rh P

R'

Rh P

CO 4d

CO

P CO

4a

R'

CO 4e

CO

Intermediates in the tris(2-tert-butyl-4-methylphenyl) phosphite modified Rhodium catalyzed hydroformylation.

Chart 6.2

tion of the alkene, can occur in two ways; one giving rise to the linear and the other to the branched alkyl rhodium intermediate. These, and their derived compounds, probably give rise to slightly different IR absorptions. This explains the broadness of the signals of the new species formed. The possible substrate bonded intermediates of the rhodium-catalyzed hydroformylation are depicted in Chart 6.2: the p-coordinated alkene, 4a, the alkyl complexes, RhR’(CO)2P, 4b, and RhR’(CO)3P, 4c, and the acyl complexes, RhC(O)R’(CO)2P, 4d, and RhC(O)R’(CO)3P, 4e, (R’ ¼ styryl, octyl or pentafluorostyryl) [18]. The complexes 4b and 4d are coordinatively unsaturated and therefore unlikely to have sufficiently long lifetime to be observed. Furthermore, three CO frequencies indicate that at least three CO molecules are coordinated to the rhodium center, excluding 4b and 4d as the observed species. Rhodium carbonyl complexes without phosphorus ligands coordinated, have their nCOs at substantially higher frequencies [18] excluding the phosphite as the substituted ligand. These considerations lead towards the coordinatively saturated complexes 4c and 4e possibly being the most abundant species. Moreover, the additional acyl absorption at 1690 cm–1 points to 4e. A band around 1700 cm–1 is characteristic of the stretching absorption of a keto-group [29]. Metal–acyl absorptions are usually very weak and broad, in accordance with the observations. Considering the steric hindrance present in the alkyl rhodium complex, 4c, a larger effect on the CO frequencies would be expected. Also, the absorption of the CO that is trans to the alkyl group would be strongly affected by the electronic influence of the different alkyl groups. The absence of this large substrate dependence indicates that the acyl complex, 4e, is the most likely species. The presence of 4e as the predominant species during the catalysis is also in accord with the observed kinetic behavior of this catalyst with 1-octene and styrene as the substrates. The observation of this saturated acyl rhodium complex is in line with the positive dependence of the reaction rate on the hydrogen concentration and the zero order in alkene concentration. It was concluded previously that this saturated acyl complex is an unreactive “resting state” [18]. Before the final hydrogenolysis reaction step can occur, a CO molecule has to dissociate in order to form

243

244


the coordinatively unsaturated complex, 4d. This means that 4d and 4e are in equilibrium and that their relative concentrations depend on the concentration of CO present in the reaction mixture and the reaction rate has a negative order in CO pressure.

6.5

Diphosphite Ligands

Phosphite ligands and especially bulky phosphites are very useful in rhodium catalyzed hydroformylation because of the higher reaction rates obtained when compared to triphenylphosphine. An important drawback, however, is the loss of selectivity. Where rhodium triphenylphosphine systems can provide selectivities of up to 92 % for the linear aldehyde, albeit at low rates, for bulky phosphite ligands the selectivity is reduced to 70 %. One way to improve the selectivity is to change to diphosphite systems. It was only after the first reports from Bryant and coworkers that diphosphites were recognized as a new generation of promising ligands in rhodium catalyzed hydroformylation of alkenes [30]. This initial report was followed by numerous patents from UCC (now Dow) and triggered a large research effort in academia and industry. Table 6.4 Hydroformylation using rhodium bulky diphosphite catalysts.a

Ligand

T, hC

p CO/H2 bar

Ratio CO : H2

Alkene

5a c 5a c 9a c 7c 8c 7c 6a c 10a c 10b c 10c c 11 c 12 c 11 c 12 c 13a d 13b d 6a d 6b d 5b d 9b d

70 71 70 70 70 70 74 90 90 90 90 90 90 90 80 80 80 80 80 80

2.5 6.7 7 7 4.3 4.3 4.5 7.1 7.1 7.1 7.1 7.1 7.1 7.1 20 20 20 20 20 20

1.:.2 1.:.2 1.:.2 1.:.2 1.:.1 1.:.1 1.:.1 1.:.1 1.:.1 1.:.1 1.:.1 1.:.1 1.:.1 1.:.1 1.:.1 1.:.1 1.:.1 1.:.1 1.:.1 1.:.1

1-butene “ “ “ propene “ “ 1-butene “ “ “ “ 2-butene 2-butene 1-octene 1-octene 1-octene 1-octene 1-octene 1-octene

Isom. %

Rate,b mol. (mol Rh)–1 h–1

l: b

n. d. 20 18 27 n. d. 13

2400 730 1480 160 280 20 402 1620 1320 1070 3660 1650 1140 65 11,100 1550 3600 6,120 3375 520

50 35 3.2 6.3 1.2 2.1 53 2.3 3.8 2.2 2.0 9.9 0.5 2.8 1.6 2.2 i100 51 19 1.2

Conditions: 0.1–1 mM Rh, L.:.Rh ¼ 10–20.:.1, [alkene] ¼ 0.5–1 M in toluene. b Initial rate. c Taken from Refs. [30] and [31]. d Ref. [32]. n. d. ¼ not detected.

a

6.5 Diphosphite Ligands

The selectivity for the linear aldehyde in the rhodium catalyzed hydroformylation of 1-alkenes increases tremendously upon changing from bulky monophosphite to bulky diphosphite ligands based on a bisphenol linker [30, 31]. The reaction rates are in general much lower than that of the bulky monophosphite system, but still relatively high compared to the triphenylphosphine based catalyst. The selectivity was found to be very dependent on the exact ligand structure (see Charts 6.3 and 6.4) [32]. Selectivities higher than 95 % were obtained and, depending on the ligand, small amounts of either the branched aldehyde, isomerized alkenes or both were observed (see Table 6.4). The bridge length of the diphosphites has a large influence on the selectivity for the linear aldehyde. For aliphatic bridges, as in ligands 10, the optimum selectivity was found for a three-carbon bridge, 10b, derived from 1,3-propanediol. Remarkably, the same preference for a three-carbon bridge was observed for the asymmetric hydroformylation of styrene using chiral diphosphite ligands [33, 34]. Although this was not recognized at first, the bite angle of the diphosphite ligands is probably an important parameter determining the selectivity of the hydroformylation reaction, as was also found for diphosphine ligands [35, 36]. The highest selectivities were, however, achieved using bisphenol bridges.

R

R R

R

t-Bu

O O t-Bu P OAr t-Bu P O OAr O t-Bu

t-Bu

O O t-Bu P O O P O O

R OO

R 5 a R = OMe, (OAr)2 =

6 a, R = OCH3 b, R = t-C4H9,

b R = t-Bu, OAr = 4-C6H4-Cl MeO

MeO

OMe

t-Bu t-Bu O P O

O O

t-Bu t-Bu P O O

t-Bu OMe

MeO

OMe 7 Chart 6.3 Bulky diphosphites developed by Bryant at UCC.

P O O

P O

t-Bu t-Bu MeO

O O

O

t-Bu MeO

8

245


246

t-Bu

MeO t-Bu R

t-Bu

O O

O

P O O

P O

R

t-Bu

MeO

t-Bu

MeO

t-Bu

O P O O

O P

t-Bu

MeO

t-Bu

OMe

10 a, n = 2 b, n = 3 c, n = 4

R

9a, R = OCH3 b, R = t-C4H9

OMe

O O P O (CH2)n O P O O

R

t-Bu t-Bu

t-Bu

OMe

O O

t-Bu

OMe

11 R

R R

R

O O P

R

O P O O

R R

R

R

O

O O P O O

O P O R

R R

O O P O (CH2)n O P O O R

13 a, n = 2 b, n = 3

R

R

R 12, R = t-C4H9

Chart 6.4

Bulky diphosphite ligands.

Van Leeuwen et al. studied the kinetics of the hydroformylation of 1-octene using the bulky diphosphite 5b [32]. The reaction rate was almost independent of the hydrogen pressure and showed a first order dependence on the alkene concentration and a negative order in CO pressure. All data indicate a rate-determining step early in the catalytic cycle. The kinetic data are very similar to those obtained by Cavalieri d’Oro et al. for the triphenylphosphine based catalyst [37]. Instead of a negative order in CO they found a negative order in triphenylphosphine, probably because of the more facile ligand dissociation of the phosphine from the putative resting state, HRh(PPh3)3(CO), compared to the HRh(CO)2(diphosphite).


The selectivity to the linear product nonanal was strongly dependent on the CO pressure (see Table 6.4). The linear to branched ratio drops from 29 at PCO ¼ 5 bar to 4 at PCO ¼ 40 bar. Part of this selectivity change can be ascribed to enhanced isomerization at lower CO pressure, but faster b-hydride elimination cannot account completely for the increased formation of the branched aldehyde. The reduced selectivity was attributed to partial ligand dissociation resulting in less selective rhodium monophosphites and/or ligand-free complexes. Under syn-gas pressure, the rhodium acac precursors were converted to the catalytically active hydride complexes, HRh(CO)2(L–L). The complexes are generally assumed to have a trigonal bipyramidal structure and two isomeric structures of these complexes are possible, containing the diphosphite coordinated in a bisequatorial (ee) or an equatorial–apical (ea) fashion. The structure of the complexes can be elucidated from HP-IR and HP-NMR data (Table 6.5). In the carbonyl region of the infrared spectrum, the vibrations of the ee and ea complexes can be easily distinguished. The ee complexes typically show absorptions around 2015 and 2075 cm–1 [32, 33, 38, 39] whereas the ea complexes exhibit carbonyl vibrations around 1990 and 2030 cm–1 [33, 39]. The rhodium hydride vibration lies in the same region as the carbonyl ligands but is often very weak. In fact, additional carbonyl signals in the case of mixtures of complexes are often erroneously assigned to the rhodium hydride stretching frequencies. The IR spectrum of HRh(5b)(CO)2, as a Nujol mull, showed three absorptions in the CO-stretching region instead of two which were observed in acetone solution (see Table 6.5). Upon measuring DRh(5b)(CO)2, as a Nujol mull, only two absorptions remained (2058, 2012 cm–1), which were shifted compared to HRh(5b)(CO)2 (2035, 1998, 1990 cm–1). This implies that the three frequencies in the hydrido complex are a combination of two CO stretching frequencies and one rhodium hydrido stretching frequency. The rhodium-hydride vibration disappears upon deuteration of the complex as the rhodium-deuteride vibration appears in the fingerprint region. The large frequency shift of the highest energy absorption is indicative of a trans-CO geometry [40]. In solution IR, the rhodium hydride vibration and the lowest energy CO vibration overlap, which results in only two absorptions. Table 6.5

NMR and IR data for HRh(P–P)(CO)2 complexes.a

Ligand

d

31

d 1H

1

1

2

1

JPP

n Rh-CO

5b

160.5 159.8

–10.4

237, 226

3.5

–19, 70

170

2049 1966b

6a

174.5

–10.8

239

3.5

4

6b

173.0

–10.3

234

3

9

14

167.4 156.9

–9.60

238, 154

n. r.

190

58

15

168.1 165.5

–10.11

244, 225

I2

I2

247

a

P

JRhP

JRhH

Data taken from Refs. [32] and [34]. b In acetone solution.

JPH

2074 2013 2035 1994

247

248


Additional information can be obtained from high pressure NMR data. Phosphorus donors coordinated in the equatorial plane generally have small coupling constants of phosphorus to hydrogen. The coupling constant of a trans-coordinated phosphite shows a large value of 2JPH ca. 180–200 Hz. The value of 2JPP is much larger for the ee (ca. 250 Hz) than for the ea complex (usually I70 Hz). Some representative NMR and IR data of rhodium-diphosphite complexes are given in Table 6.5. When C2 -symmetric ligands are employed both phosphorus donor atoms become inequivalent, no matter whether the ee or ea complex is formed. Buisman prepared complexes of ligands 14 and 15 (Chart 6.5) that exhibited exclusive ea and ee coordination respectively. These complexes showed fluxional behavior at room temperature [39]. Both the ee and the ea complexes have inequivalent phosphorus donor atoms which are in fast exchange on the NMR time scale. For the ea complex of 14, an averaged phosphorus to hydrogen coupling constant of 80 Hz was observed. At low temperature, the exchange process stopped and a large coupling constant due to the phosphorus being trans to the hydride was observed, whereas the cis-coupling constant remained very small and unresolved [39]. Variable temperature NMR studies allowed the thermodynamic activation parameters to be determined by comparison of the calculated and experimental NMR spectra (see Figure 6.9). The value of DH‡ , calculated from the Eyring equation, was 35 kJ mol–1 and DS‡ was 6 J K–1 mol–1. Both DH‡ and DS‡ were significantly higher for the ee rhodium complex containing ligand 15, (62 kJ mol–1 and 46 J K–1 mol–1 respectively). From the relatively low entropy of activation it was concluded that the exchange process was an intramolecular process and did not involve ligand dissociation. Remarkably, it was observed that the exchange rate of the ea complex was an order of magnitude higher than that of the ee complex. The exchange process of the phosphorus donor atoms of the ea and ea complexes was suggested to proceed by the low energy rearrangement mechanism, described by Meakin (see Figure 6.10) [41]. A simultaneous bending motion of the t-Bu t-Bu R

R t-Bu O t-Bu EtO

O P

O

O

O

P O

OEt t-Bu O t-Bu

Chart 6.5

P O O

O P O

R

R

O

t-Bu

t-Bu

t-Bu 15 R = Sit-BuMe2

t-Bu 14

O

t-Bu

O

t-Bu

Chiral diphosphites forming exclusively ea, 14, or ee, 15, rhodium complexes.


Figure 6.9 Calculated and observed variable-temperature 1H (300 MHz) NMR spectra for HRh(14)(CO)2. Reproduced from Ref. [39] with permission.

hydride and carbonyl ligands takes place in the hydridorhodium diphosphite dicarbonyl complexes containing ea coordinating diphosphites (Figure 6.10A). For ee coordinating ligands, however, a motion of the hydride in the equatorial phosphite functions is responsible for the exchange (Figure 6.10B). The latter process is expected to be more difficult than the former, which explains the higher fluxionality of the ea complexes.

249

250


A

H P1' OC

P1'

OC

CO

H P2'

P2'

B

H CO(1) P1'

CO

Rh

Rh

CO(1)

P1'

Rh

Rh P2'

CO(2)

P2'

H CO(2)

Figure 6.10 A: Equatorial–apical phosphorus exchange. B: Equatorial– equatorial phosphorus exchange. Reproduced from Ref. [39] with permission.

Summarizing, the relationship between ligand and complex structure and selectivity of the hydroformylation reaction is not always straightforward. In general, bis-equatorial coordination of the ligand is required in order to obtain a high preference for the formation of linear aldehyde but bis-equatorial coordination does not always lead to high selectivity, as the preference for the linear aldehyde is very dependent on the exact ligand structure. Best results are obtained with bridging backbones based on bisphenol; aliphatic bridges often give poor results. An important feature of the ligands is that they must create sufficient steric bulk around the rhodium center. The steric bulk must be located on the bridging bisphenol and, at most, on one of the terminal end-groups. Introduction of both electron-donating and electron-withdrawing substituents on the dissymmetric bulky bisphenols induced higher selectivity for the linear aldehyde. This shows that the effect of electronic variations of the system on the catalytic performance is also not very straightforward [32]. When aliphatic bridges are introduced instead of bisphenol, the resulting selectivity is dependent on the length of the bridge; the optimum is found when a three-carbon linker is applied. A similar feature was observed in the asymmetric hydroformylation of styrene using chiral diphosphites [33]. The obtained linearity is generally much lower than that of the successful bisphenol bridged bidentate phosphites.

6.6

Dimer Formation

In many instances, the formation of inactive dimers from active, monomeric catalytic species is observed during catalysis. When weak or unstable ligands are used, even larger rhodium carbonyl clusters like Rh4(CO)12 and Rh6(CO)16 can be observed [42–44]. The formation of dimers is often a reversible equilibrium (Scheme 6.2). This only leads to a reduction in the amount of catalyst available and does not kill the catalyst. One of the first examples was the formation of the so-called orange dimer from HRh(PPh3)3CO, already reported by Wilkinson [45]

6.6 Dimer Formation

H

CO O

CO - H2

P Rh CO

P Rh

P

P

P Rh

O

CO

P Dimer formation for hydroformylation catalysts.

Scheme 6.2

and characterized by Chan (see Chart 6.6) [46]. They also observed the formation of red dimers after losing two carbon monoxide molecules. Since the hydroformylation reaction for most substrates shows a first order dependence on the concentration of rhodium hydride, the reaction becomes slower when considerable amounts of rhodium are tied up in dimers. This will occur at low pressures of hydrogen and high rhodium concentrations. Dimer formation has mainly been reported for phosphine ligands [17, 42, 45], but similar dimeric rhodium complexes from monophosphites [47] and diphosphites [33, 39] have been reported. The orange side product obtained from HRh(15)(CO)2 was characterized as the carbonyl bridged, dimeric rhodium species Rh2(15)2(CO)2 [39]. Gladfelter et al. reported the presence of dimeric complexes during the rhodium catalyzed hydroformylation of 1-octene using ligand 6a [38]. The appearance of dimers was mainly observed at the final stage of the reaction at low hydrogen pressure. No effect of the formation of dimers on the actual catalytic reaction was reported. By reducing the steric bulk of the ligand, 16, (Chart 6.7), they even isolated a tetranuclear rhodium complex, illustrating the importance of steric hindrance in the ligand structure [48].

O CO C L L Rh Rh L CO C CO O

O O

O P P O

16

Chart 6.6

Rhodium dimer characterized by Chan (L ¼ PPh3)

[46].

O O

Chart 6.7

Diphosphite ligand studied by Gladfelter [38].

251

252


6.7

Study of Bulky Phosphorus Diamide Ligands

N-acyl phosphorus diamides, 17 in Chart 6.8, provide a related class of bulky and strongly p-accepting ligands, which show similar behavior in the rhodium catalyzed hydroformylation to that found with bulky phosphites. Similar to bulky phosphites, the combination of electronic and steric ligand properties leads to an increase in the hydroformylation activity compared to phosphine-based catalyst systems [49]. The mechanism of the hydroformylation reaction employing this class of ligands has been investigated thoroughly by in situ identification of intermediate rhodium complexes and kinetic studies [50]. The rate-limiting step in the hydroformylation reaction of 1-alkenes using ligands 17 in Chart 6.8, was investigated and the solution structure of the resting state of the catalyst was studied using in situ high-pressure spectroscopic techniques. Ph N

O Ph

N

P Ph

17a Chart 6.8

O N

Ph

Et N

O Et

N

P Ph

N

17b

O

O

Et

Ph

Ph N N

O

N P Ph OPh 17c

Et N

O Et

N

O

N P Et OPh 17d

Phosphorus diamide ligands studied by van Leeuwen et al.

The rhodium hydride, as actual catalyst precursor, was formed from Rh(acac)(CO)2 in the presence of various concentrations of ligand under 20 bar of CO : H2 (1 : 1). The reaction was monitored using HP-NMR and HP-IR spectroscopy in order to identify the structures of the complexes present in situ. In all cases, a mixture of HRhL(CO)3 and HRhL2(CO)2 was observed with the ratio between HRhL2(CO)2 and HRhL(CO)3 depending both on the ligand concentration and on the steric and electronic ligand properties. The spectroscopic data of the hydride complexes HRhLx(CO)4–x (x ¼ 1, 2, 3; L ¼ 17a–d) are presented in Table 6.6. A detailed kinetic study of the rhodium catalyzed hydroformylation of 1-octene using ligand 17b resulted in the rate equation (4): v ¼ k [1-octene]0.3 [17a]0.3 [CO]1 [Rh]1 [H2]0.8

(4)

From the kinetic data, it was concluded that the rate-determining step of the hydroformylation reaction using HRh(17b)2(CO)2 (18b in Scheme 6.3) was not one single elementary step of the hydroformylation mechanism in the range of the conditions studied. Several reaction steps in the proposed mechanism are involved in the rate control of the reaction. The early steps, alkene coordination/hydride migration, and hydrogenolysis, which occurs late in the catalytic cycle, have similar rates and the overall reaction rate is strongly dependent on the conditions used. At high hydro-

6.7 Study of Bulky Phosphorus Diamide Ligands

253

NMR and IR Data for HRhLx(CO)4–x (x ¼ 1, 2, 3; L¼17a–17d).

Table 6.6

Compound

d(31P) (ppm)a

JRhP (Hz)

d(1H) (ppm)a

n CO (cm–1)b

HRh(17a)(CO)3 HRh(17a)2(CO)2

110 114

186 196

–10.6 (d, JPH¼ 6 Hz) –11.4 (t, JPH¼12 Hz)

2094, 2047, 2017 2079, 2023

HRh(17b)(CO)3 HRh(17b)2(CO)2 HRh(17b)3CO

104 110 117

177 181 169

–10.3 (d,JPH¼15 Hz) –10.6 (s, broad) –10.7 (s, broad)

2095, 2045 , 2008 2070, 2018 2019

HRh(17c)(CO)3 HRh(17c)2(CO)2

not observed 126

221

not observed –10.6 (s, broad)

2096, 2047, 2020 2071, 2003

HRh(17d)(CO)3 HRh(17d)2(CO)2

116 118

220 224

not observed –10.5 (s, broad)

2098, 2043, 2014 2076, 2020

a Measured in toluene-d8. b Complexes with L ¼ 17a, 17c measured in 2-methyl tetrahydrofuran, complexes with L ¼ 17b, 17d measured in cyclohexane.

H OC

L

Rh

L

R CO

O

18b OC

L

Rh

step 1

L CO

25b step 7

CO R

CO H

product

L

Rh

L

R

CO

O H2

19b

Rh

H

step 2

step 6 L

R

L

L

CO

CO

24b

20b R

L

R

step 5

step 3 L L

step 4 Rh CO

23b

Scheme 6.3

Rh

L

CO

CO

Hydroformylation reaction cycle.

Rh CO

22b

L

L

Rh CO

R

21b

L

branched product + 2-alkenes

254


gen pressure, the alkene coordination/hydride migration determines the overall rate of the reaction and consequently the rhodium hydride is expected to be the most abundant species. When the alkene concentration is increased at relatively low hydrogen pressure the rate-determining step shifts to the hydrogenolysis and, as a result, the rhodium acyl complex will be predominantly present. Depending on the conditions, both rhodium-hydride and rhodium-acyl complexes are expected to be present as potential “resting” states during the hydroformylation reaction. The complexes present were investigated using in situ spectroscopic techniques applying an IR autoclave. The results obtained in the in situ HP-IR experiments are presented in Table 6.7. The hydride complex 18b was formed in situ and after addition of 1-octene the hydroformylation reaction was monitored using a rapid scan IR technique (7 scans s–1). The difference spectra obtained (Figure 6.11) show negative absorption bands for the carbonyl frequencies of complexes that are converted (in part) to other complexes after addition of the substrate. Positive absorption bands are obtained for carbonyl frequencies of complexes that are formed during the hydroformylation reaction. The rhodium hydride complex 18b (nCO ¼ 2070, 2018 cm–1, Table 6.7) is the only rhodium complex present in all experiments before addition and after complete conversion of 1-octene. The difference IR spectra show absorption bands in the terminal carbonyl region only. No absorptions of bridging carbonyls belonging to inactive rhodium dimers or clusters are observed, which confirms that only mononuclear rhodium complexes are present during the reaction, as already concluded from Table 6.7 IR absorptions obtained from in situ IR experiments.

pCO (bar)

pH2 (bar)

[1-octene] (M)

n CO (cm–1)a Disappearingb

7

7

0.2

2070 (18b) 2018 (18b)

7

32

0.2

–c

7

7

0.6

2070 (18b) 2018 (18b)

Appearingb 2085 2077 2028 2010 2001 1991 1967 –c

(w) (sh) (w) (sh) (m) (m) (w)

2085 2077 2028 2010 2001 1991 1967

(w) (sh) (w) (sh) (m) (m) (w)

All experiments were performed in cyclohexane at 40 hC. b. Appearing absorption bands are obtained for carbonyl frequencies of complexes that are formed during the hydroformylation reaction, whereas disappearing absorption bands are obtained for carbonyl frequencies of complexes that are converted to other complexes. c.No change in IR spectrum was observed.

a

6.7 Study of Bulky Phosphorus Diamide Ligands

Figure 6.11

Difference spectrum of the in situ IR experiment. Reproduced from Ref. [28] with

permission.

the first order rate-dependence on the rhodium concentration observed in kinetic studies. Immediately after addition of 1-octene, the strong absorption band indicative of aldehyde formation (1734 cm–1) appeared in the IR spectrum, proving that the hydroformylation reaction had started. The amount of 18b dropped considerably upon addition of 1-octene, but it did not disappear completely during the hydroformylation reaction. Seven new absorptions appeared in the terminal carbonyl region indicating that complex 18b was converted (in part) to several new carbonyl-containing rhodium complexes. The kinetic experiments showed that the hydrogenolysis (step 6, Scheme 6.3) is a relatively slow step in the hydroformylation reaction and rhodium-acyl complexes should be present during the hydroformylation reaction. Several possible structures of rhodium-acyl complexes are depicted in Chart 6.9. Since the catalyst does not show high selectivity towards the linear aldehyde both the linear and branched rhodium-acyl complexes should be formed (25b, 26b). Coordination of one of the phosphorus atoms at an apical position leading to ea coordinated complexes is also plausible. The new carbonyl bands obtained in the IR spectrum probably belong to several of the rhodium-acyl complexes depicted in Chart 6.9. The strong amide bands of the ligand in this region obscured the expected rhodium-acyl absorption band [18b] around 1600–1700 cm–1. Therefore, in contrast to the bulky phosphite system, the rhodium-acyl vibration could not be observed in the IR spectra. The rhodium hydride complex, 18b, was the only rhodium complex observed during the hydroformylation reaction at increased partial hydrogen pressure of 32 bar, as concluded from the absence of carbonyl signals in the IR difference spectra. The spectrum of the rhodium hydride resting state is taken as background at

255

256


R O

O

R L

Rh

CO

L

Rh

CO

L

L

CO

CO

26b

25b R O

O

R OC

Rh

L

OC

Rh

L

OC

OC L

L

27b

28b

Possible structures of rhodium acyl complexes.

Chart 6.9

t ¼ 0 (Table 6.7). The difference IR spectra show no terminal carbonyl bands, indicating the presence of no other complex other than the rhodium hydride, 18b, under these conditions. Higher hydrogen pressures facilitate the hydrogenolysis step and one of the early steps of the catalytic cycle becomes rate limiting. When the alkene concentration is raised, the amount of 18b decreases and the intensity of the additional carbonyl bands increases. This indicates that the role of the hydrogenolysis reaction as rate-controlling step increases at higher alkene concentration. The shift of rate control between different steps, depending on the conditions, nicely illustrates the general feature in catalysis that not one single step determines the reaction rate. A combination of HP-NMR and HP-IR studies was performed to reveal the occurrence of rhodium-acyl intermediates. The hydroformylation reaction was repeated in a stepwise manner by the subsequent reactions of 18b with alkene, CO, and H2 in an attempt to characterize the complexes formed during the hydroformylation reaction. In the absence of H2, complex 18b, alkene and CO can undergo all the hydroformylation reaction steps except for the hydrogenolysis step and rhodium-acyl complexes are expected to be formed. The IR data obtained after the subsequent reaction steps are presented in Table 6.8. The carbonyl frequencies obtained after the reaction of 18b, 1-octene and CO are identical to those obtained in the in situ IR experiments obtained during the hydroformylation reaction, except for the carbonyl frequency due to the aldehyde. Subsequent addition of hydrogen showed aldehyde formation and, when all the 1-octene was converted, 18b was again the only complex present. The complexes formed in the stoichiometric reaction of 18b, alkene and CO were assigned to be equal to the most abundant complexes present during the hydroformylation reaction at low H2 pressure. This indicated that the stoichiometric reaction is an elegant and reliable

6.7 Study of Bulky Phosphorus Diamide Ligands Table 6.8

IR frequencies obtained in the stoichiometric reaction of 18b (L ¼ 17b), 1-octene, CO

and H2. Conditionsa

n CO (cm–1)

18b under 5 bar of CO

2080, 2017

18b þ 25 equiv. of 1-octene under 5 bar of CO

2085 (w), 2077 (sh), 2028 (w), 2010 (sh), 2001 (m), 1991 (m) 1967 (w)

18b þ 25 equiv. of 1-octene under 10 bar of CO/H2¼1/1

2080, 2017

a

All reactions were performed in cyclohexane at room temperature.

method to study the intermediates of the hydroformylation reaction in a stationary system. In addition, the stoichiometric hydroformylation of 1-hexene using NMR spectroscopy was investigated using the same procedure as that for the IR experiments. The experiments were performed in a high pressure NMR flow cell as described by Iggo and co-workers [2]. The advantages of a HP flow cell instead of a HP-NMR tube [51] are the continuous supply of reactants and optimal mixing of the reactants (minimization of diffusion problems). Homogeneously catalyzed reactions can be monitored using this flow cell and stable intermediates can be characterized using different NMR techniques. The rhodium complex, 18b, was prepared in situ in the NMR cell from Rh(acac)(CO)2 and 5 equivalents of ligand 17b at 80 hC under 20 bar of CO/H2 ¼ 1/1 (spectrum a, Figure 6.12). Complex 18b is the only rhodium complex obtained according to the 31P NMR spectrum. After removal of hydrogen gas by bubbling CO through the solution for 30 min approximately 25 equivalents of 1-hexene were injected into the NMR cell at 253 K. Upon warming to room temperature, the 31P NMR spectrum started to broaden and both the 1H NMR and 31P NMR spectrum showed that the resonance due to the hydride complex decreased in intensity. After decreasing the temperature to 253 K, the 31P NMR spectrum showed an additional (broad) doublet (d ¼ 109.5 ppm, 1JRhP ¼ 205 Hz) upfield from the doublet due to 18b (spectrum b, Figure 6.12). Complex 18b was converted almost completely to the new compounds by warming the solution to room temperature again (spectrum c, Figure 6.12). A broad resonance appeared downfield from the doublet, indicating that probably more than one complex is present. The 31P NMR spectrum at room temperature showed a very broad peak at the position of the free ligand (64 ppm), indicating that the coordinated phosphorus ligands were in exchange with the free phosphorus ligand. The hydride resonance in the 1 H NMR spectrum had disappeared and no aldehyde resonance was observed. Changing the gas flow to 5 bar of CO/H2 (1/1) resulted in reformation of the hydride complex 18b (spectrum d, Figure 6.12). The 1H NMR spectrum showed the reappearance of the hydride resonance at –10.6 ppm together with an aldehyde resonance at 9.3 ppm. Repeating this sequential procedure with this NMR sample

257

258

6 Rhodium Catalyzed Hydroformylation *

*

* HRhL2(CO)2 a *

*

b

c *

*

d

113

111

109

107

ppm

Figure 6.12 Overview of the P{ H} NMR spectra of the stepwise reaction of 18b, (L ¼ 17b), 1-hexene and CO and H2. All spectra were recorded at 253 K, (a) 31P{1H} NMR spectrum of 18b under 5 bar of CO; (b) 31P{1H} NMR spectrum of 18b in presence of 5 bar of CO and 25 equivalents of 1-hexene. Approximately 50 % of 18b was converted to a new rhodium complex. (c) 31P{1H} NMR spectrum after complete conversion of 18b to a new rhodium complex; (d) 31P{1H} NMR spectrum after addition of 5 bar of CO/H2 (1/1) to the solution. Reproduced from Ref. [50] with permission. 31

1

gave the same results, confirming that the complex formed in the absence of H2 is an intermediate of the hydroformylation reaction. The reaction was also performed with 13CO to enable identification of the rhodium-acyl resonance in 13C NMR. All resonances obtained in the 13C NMR spectrum are very broad at room temperature. The resonances sharpened when the temperature was decreased (Figure 6.13). The 13C NMR spectrum at 223 K (Spectrum 3a, Figure 6.13) showed two rhodium-acyl resonances (d ¼ 230.0, 227.2 ppm) and three different rhodium carbonyl resonances (d ¼ 194.7, 193.7 and 190.6 ppm), indicating that we are dealing with two different rhodium-acyl complexes. The structure of the two rhodium-acyl complexes was elucidated using 13C COSY90 spectra, selective decoupling of the phosphorus resonances, and 103 Rh–13C HMQC spectra. Spectrum 3d (Figure 6.13) shows the complete simulation of the 13C{1H} NMR spectrum obtained at 223 K. The 13C{1H} NMR spectrum contains both rhodium-acyl complexes 25b/26b and 27b/28b in a ratio 1 to 0.4. Both rhodium-acyl complexes have a trigonal bipyramidal structure with the acyl moiety coordinated at an apical position. Both complexes contain two phosphorus ligands and two carbonyl ligands. The major complex formed contains two phosphorus ligands coordinated in the equatorial plane of the trigonal bipyramid, the minor complex probably has one phosphorus atom coordinated at an equatorial position and one at an apical position of the trigonal bipyramid. We are not able to distinguish between the linear and branched acyl structures based on the NMR data obtained.

6.8 Study of the Elementary Steps of the Catalytic Cycle

T = 253 K

259

T = 223 K *

1a

*

13CO

*

1b 232

230

228

226

196

192

188

* * *

3a .

184

3b

PPM

T = 233 K * *

3c

2a

1 .

.

1 .

2b 232

230

228

226

196

192

188

3d 232

184

230

228

226

196

192

PPM 13

1

Variable Temperature C{ H} NMR spectra obtained after the reaction of 18b (L ¼ 17b), 1-hexene and CO. (1a) 13C{1H} NMR spectrum obtained at 253 K; (1b) Simulation of 13 C{1H} NMR spectrum 1a; (2a) 13C{1H} NMR spectrum obtained at 223 K; (2b) Simulation of 13 C{1H} NMR spectrum 2a; (3a) 13C{1H} NMR spectrum obtained at 223 K; (3b) simulation of the 13 C{1H} NMR spectrum of 25b/26b at 223 K; (3c) simulation of the 13C{1H} NMR spectrum of 27b/28b at 223 K; (3d) Simulation 13C{1H} NMR spectrum obtained at 223 K. * Impurity. Reproduced from Ref. [50] with permission. Figure 6.13

6.8

Study of the Elementary Steps of the Catalytic Cycle

6.8.1

CO-dissociation

Casey and Whiteker introduced the concept of the natural bite angle to calculate the preferred chelation angle of diphosphine ligands using molecular mechanics [10]. They were the first to show that the bite angle has a great influence on the regioselectivity in the rhodium-catalyzed hydroformylation [35]. Following this approach, we have developed a series of diphosphine ligands based on the xanthene-type backbones (Chart 6.10) [36, 52] in this series of ligands, a clear trend of increasing selectivity for linear aldehyde formation with increasing natural bite angle was observed. We discovered that not only the variation of the ligand backbone but also substitution of the standard diphenylphosphine moieties has a direct influence on the ligand bite angle. We have developed new phosphacyclic xantphos derivatives that have shown an increased preference for ee coordination and display a higher activity in the hydroformylation of 1-octene than the non-cyclic parent ligand. Moreover, DBP-xantphos, 30, and POP-xantphos, 33, give very active

188

184

PPM


260

O PPh2

O PPh2

P

O

Ph P

P

Ph P

Ph Ph

Ph

Ph

Ph 29, Xantphos

30, DBP-xantphos

O P

31, TPP-xantphos

O

O P

P

H2C

CH2

P

P

O

32, PCP-xantphos

O

S

34, PSP-xantphos

O

O P

O

35, POP-DPEphos Chart 6.10

P

S

33, POP-xantphos

P

Ph

PPh2

PPh2

O

36, Isopropxantphos

Xantphos family.

and selective catalysts for the hydroformylation of internal octenes to linear nonanal [53, 54]. To investigate the origin of the very high hydroformylation and isomerization activity of ligand 33, we measured the rate of CO dissociation from the HRh(diphosphine)(CO)2 complex using 13CO labeling in rapid-scan IR experiments [54]. The CO dissociation rate constants, k1, can be obtained by exchanging 13CO for 12CO in the HRh(diphosphine)(13CO)2 complexes [52].The CO dissociation proceeds via a dissociative mechanism and consequently obeys simple first-order kinetics. The rate constants k1 can, therefore, be derived from Eqs. (5) and (6). –d[HRh(diphosphine)(13CO)2]/dt ¼ k1[HRh(diphosphine)(13CO)2]

(5)

ln[HRh(diphosphine)(13CO)2] ¼ –k1t þ ln[HRh(diphosphine)(13CO)2]0

(6)

6.8 Study of the Elementary Steps of the Catalytic Cycle Table 6.9

Kinetics of

13

CO dissociation in HRh(diphosphine)(13CO)2 complexes.a

Diphosphine Ligand

Phosphacycle

p (CO) (bar)

r2

k1 (h–1)

32 32 33 33 36 36 36

phosphorine

25 25 25 25 20 25 30

0.987 0.987 0.992 0.990 0.999 1.000 0.999

–288 e 8 –266 e 7 –1188 e 29 –1171 e 23 182 e 1 185 e 1 181 e 1

phenoxaphosphine Isopropxantphos

Reaction conditions: [HRh(diphosphine)(13CO)2] ¼ 2.00 mM in cyclohexane, p(13CO) ¼ 1 bar, p(H2) ¼ 4 bar, T ¼ 40 hC, diphosphine : Rh ¼ 5 :1. Values for k1 are least-squares fit of lines from ln[Rh] vs. time over the first 15 s (ligand 32) or over the first 4 s (ligand 33).

a

The HRh(diphosphine)(13CO)2 complexes were prepared in situ from Rh(acac)(CO)2 and diphosphine under an atmosphere of 13CO/H2 (1 : 4). The exchange of 13CO for 12 CO in the HRh(diphosphine)(13CO)2 complexes was monitored by rapid-scan HPIR spectroscopy at 40 hC. The 13CO/12CO exchange was initiated by adding a large excess of 12CO after release of the pressure. Free lying carbonyl absorptions of the complexes at approximately 1945 cm–1 (1948 cm–1 for ligand 33, see Figure 6.14a) were taken to calculate the concentrations of the complexes. These absorption bands are assigned to one of the CO vibrations of the ee isomer of the complexes. Representative kinetic data of the experiments with ligand 32, 33 and 36 are shown in Figure 6.14, and the observed rate constants, k1, are listed in Table 6.9. The experiments with ligand 36 show that the rate of CO exchange is indeed independent of the 12CO pressure. The representative difference IR spectrum displayed in Figure 6.14a for one of the experiments with HRh(33)(13CO)2 (37-13CO) clearly shows the conversion of the 13CO labeled complex to its 12CO analogue 37-12CO. In the IR spectrum, only the ee complex isomer is visible (vide supra). The exponential decay of the intensity of the carbonyl absorption at 1948 cm–1 with time is displayed in Figure 6.14c. The negative slope of this line is the first-order rate constant k1 (Eq. (6)). The decay of the carbonyl bands of the HRh(diphosphine)(13CO)2 complexes with time follows simple first-order kinetics in all experiments. Plots of ln[HRh (diphosphine)(13CO)2] vs. time are linear for at least two half-lives. Comparison of the rate constants, k1, obtained for ligands 32 and 33 [54] with those obtained for other xantphos ligands [52] shows that the CO dissociation rate for ligand 32 is in the same range as other ligands. The CO dissociation rate for ligand 33, however, proves to be four to six times higher. Furthermore, the rate is also independent of the concentration of HRh(diphosphine)(13CO)2, as demonstrated by the experiments with ligand 32. It can therefore be concluded that the CO dissociation for these complexes proceeds by a purely dissociative mechanism and obeys a first-order rate-law. The observed k1 values for the wide bite angle ligands revealed that the rates of CO dissociation,

261


262

(a)

(22)Rh(13CO)2H (25-13CO)

.1

Absorbance

.05

0

-.05

-.1

-.15

(22)Rh(12CO)2H (25-12CO) -.2 2060

2040

2020

2000

1980

1960

1940

1920

1900

1880

-1

Wavenumber (cm ) 1

0.8

(c)

(b) 0.75

0.775

0.25 ln[Rh]

Absorbance

0.5 0.75 0.725

0

-0.25

0.7

-0.5 0.675 0.65

-0.75 0

5

10

t (s)

15

20

25

-1

0

1

2

t (s)

3

4

5 13

Figure 6.14 Representative difference IR spectrum (a) and kinetic data (b and c) for the CO dissociation from HRh(33)(13CO)2 (37-13CO) in the presence of unlabeled CO at 40 hC. Reproduced from Ref. [54] with permission.

measured at 40 hC, are higher than the hydroformylation rates at 80 hC [36, 53]. Since reaction rates increase by approximately an order of magnitude with an increase in temperature of 20 hC, the CO dissociation rate at 80 hC is about 100 times faster than the hydroformylation reaction. 6.8.2

Exchange between RhD and H2

Important mechanistic information on the hydroformylation reaction can be obtained by deuterium labeling studies. The outcome of these labeling studies, however, will be influenced if fast exchange between rhodium hydrides (or deuterides)


and bulk H2 (or D2) occurs. To gain insight into the interference of this side reaction, we studied the H/D exchange rate of a catalyst system based on a p-acidic phosphorusdiamide ligand, 38, in the absence of substrate, using rapid-scan IR spectroscopy [55].

N

P N

O

O

P N

N

38

The IR absorptions of the carbonyl ligands of a rhodium-deuteride and a rhodium-hydride complex differ significantly for rhodium complexes containing two phosphorus ligands in the equatorial plane (Table 6.10). For the investigation of the H/D exchange rate, we synthesized DRh(38)(CO)2 in situ from Rh(acac)(CO)2 and 1.5 equivalents of ligand 38 under 20 bar of CO/D2 (1/1). After complete conversion to the rhodium deuteride complex, the D2 gas was removed by flushing several times with 5 bar of carbon monoxide. The H/D exchange was initiated by adding 10 bar of hydrogen to DRh(38)(CO)2 under 10 bar of carbon monoxide. The exchange process was monitored using rapid-scan HP-IR spectroscopy at 80 hC (1.3 spectra s–1). The difference IR spectrum presented in Figure 6.15 shows that the rhodium-deuteride complexes DRh(38)(CO)2 and DRh(38)2(CO) (negative peaks at nCO¼ 2067, 2041 and 2020 cm–1) are quantitatively converted into the rhodium-hydride complex HRh(38)(CO)2 (positive peaks at nCO¼ 2078 and 2026 cm–1).

Table 6.10

Spectroscopic data of HRh(L)x(CO)4–x complexes (x ¼ 2, 3 and 4).

Complex

d (1H) (ppm)a

d (31P) (ppm)a

JHP (Hz)

JHRh (Hz)

JRhP (Hz)

n CO (cm–1)b

HRh(38)(CO)2 DRh(38)(CO)2

–10.7 (broad)

138

n. d.

n. d.

218

2078, 2026 2067, 2020

HRh(38)2(CO)

–10.5 (broad q)c

135 (m) 111 (s)d

6

J3

n. d.e

2069 2041

DRh(38)2(CO) 115 (s)d HRh(38)2 a

–10.9 (d qui) –10.3 (d qui)

129 130

36 36

6 6

203 203

– –

Measured in toluene-d8 or benzene-d6. b Measured in cyclohexane. c All phosphorus atoms have similar 2JHP coupling constants. q ¼ quartet, qui ¼ quintet. d This chemical shift belongs to the noncoordinated phosphorus atom of (L\L’). e Not determined.

263

264


The presence of small amounts of HRh(38)2(CO) could not be determined because of overlap with one of the absorptions of the deuteride complex. The exponential decay of the strongest carbonyl absorption is presented in Figure 6.16A. The natural logarithm of the relative absorption of the carbonyl frequency at 2020 cm–1 versus time is presented in Figure 6.16B, indicative of first-order kinetics in the rhodium concentration and the pressure of H2 and the H/D exchange rate was 1140 h–1. Although the H/D exchange reaction was very

Absorbance

2026 cm-1

0.10

2078 cm

-1

0.05

0.00

*

-0.05

2067 cm

-1

-0,10

2020 cm-1

-0,15 2200

2100

2000

1900

1800

1700 cm-1

Difference IR spectrum obtained after addition of H2 to a solution of DRh(38)(CO)2 and DRh(38)2(CO) (indicated by *) under carbon monoxide pressure at 80 hC. Reproduced from Ref. [55] with permission. Figure 6.15

Figure 6.16 Kinetic data of the H/D exchange. A. Exponential decay of nCO at 2020 cm–1 vs. time. B. Logarithmic plot of the decay of the relative absorption at 2020 cm–1 vs. time. Reproduced from Ref. [55] with permission.


265

fast indeed, the initial rate of hydroformylation of 10 x 103 mol aldehyde (mol Rh) –1 h–1 with this catalyst was still an order of magnitude higher. This implied that up to 70 % conversion of alkene and RhH/D2 exchange is not significant with this catalyst system. At high conversion, the rate of hydroformylation is lower, because of the first-order dependence in the alkene concentration, and therefore some H/D exchange will occur. 6.8.3

Hydride Migration

The important step determining the regioselectivity of the hydroformylation reaction is the hydride migration to the coordinated alkene substrate. As the reverse reaction, b-hydride elimination, is a very common process in low valent late transition metals, the reversibility of the migratory insertion is an important point of consideration. If hydride migration is relatively slow and followed by fast subsequent carbonylation, the regioselectivity is determined kinetically by the relative rates of hydride migration to C1 or C2. If hydride migration is fast and reversible the regioselectivity is determined by the thermodynamic ratio of the branched and linear rhodium alkyl intermediates. The reversibility of the hydride migration in unmodified rhodium catalysts has been studied intensively by Lazzaroni et al. [56]. Reversible hydride migration will result in aldehydes containing deuterium at the a-carbon, whereas irreversible hydride migration will result in exclusive deuteration of the aldehyde and b-carbon. Mutual reversible alkene coordination and hydride migration will result in the formation of deuterated alkenes. Reversible formation of the branched rhodium alkyl will place deuterium at C1 and reversible formation of the linear alkyl complex will provide deuterium at C2 (Scheme 6.4). D L Rh

CO

L CO L

D

D

Rh

L

R

CO

L CO

R D

L

H

R

L CO

R

D D

D

Rh

L Rh

R

CO

L

Rh

D

Scheme 6.4

L

L

CO D

D

R

R

R CDO

R

CDO

D

Rh

CO

CO

D CDO

H L

L

CO

R

R

L

Rh

CDO D

R

266


Lazzaroni showed that the reversibility of the hydride migration to coordinated vinyl ethers is highly dependent on the temperature for the unmodified rhodium catalyst (Scheme 6.4, L ¼ CO) [56a]. Both the branched and linear rhodium alkyl intermediates showed considerable b-hydride elimination at 100 hC, as indicated by significant amounts of both alkenes deuterated at C1 or C2 (Figure 6.17). Deuter-

Figure 6.17 46 MHz deuterium-NMR spectrum (C6D6) of the crude mixture resulting from deuterioformylation of (ethyl)vinyl ether at (a) 20 hC and (b) 100 hC. Reproduced from Ref. [5] with permission.

References

ated alkenes were not detected when the reaction was performed at 20 hC, indicating that hydride migration was irreversible under these conditions. It was also observed that b-hydride elimination was more pronounced for the branched alkyl rhodium intermediate than the linear one. When styrene was used as substrate b-hydride elimination occurred exclusively in the branched alkyl intermediate; formation of the linear rhodium alkyl was irreversible [56b]. In a study of the effect of the ligand-induced bite angle on the regioselectivity of the rhodium-catalyzed hydroformylation, Casey and Petrovich investigated the reversibility of the hydride migration in a similar way by deuterioformylation [57]. They found very little deuterium incorporation in recovered hexenes, indicating that the hydride migration to coordinated hexene is virtually irreversible. Their results clearly established that the regioselectivity of the aldehyde formation is set by the kinetic ratio of formation of branched and linear rhodium alkyls, which both exclusively react further to the rhodium acyl and subsequently result in aldehyde formation.

6.9

Conclusions

The above examples illustrate clearly the impact of HP-NMR and HP-IR on our insight into the mechanism of the rhodium catalyzed hydroformylation. In situ IR can be applied on a routine basis to probe the resting state of the system, provided that a cell that is not prone to diffusion limitations is used. The majority of ligands lead to rhodium hydrides as the resting state and only a few systems, viz those rich in carbonyls or very electron poor, show acyl species as the resting state. A combination of IR and NMR measurements is needed to obtain full characterization of all species.

References [1] Rhodium Catalyzed Hydroformylation,

[2] [3]

[4]

[5]

Eds. van Leeuwen, P. W. N. M.; Claver, C., Kluwer, Dordrecht 2000. Iggo, J. A.; Shirley, D.; Tong, N. C., New J. Chem. 1998, 1043. Niessen, H. G.; Trautner, P.; Wiemann, S.; Bargon, J.; Woelk, K. Rev. Sci. Instrum. 2002, 73, 1259. Cusanelli, T.; Frey, U.; Marek, D.; Merbach, A. E. Spectrosc. Eur. 1997, 9/3, 22. Lazzaroni, R.; Settambolo, R.; Caiazzo, A., in Rhodium Catalyzed

Hydroformylation, Eds. van Leeuwen, P. W. N. M.; Claver, C., Kluwer, Dordrecht 2000, Ch. 2, pp. 15–33. [6] van Leeuwen, P. W. N. M.; Casey, C. P.; Whiteker, G. T. in Rhodium Catalyzed Hydroformylation, Eds. van Leeuwen, P. W. N. M., Claver, C., Kluwer, Dordrecht 2000, Ch. 4, pp. 63–105. [7] Young, J. F.; Osborn, J. A.; Jardine, F. A.; Wilkinson, G. Hydride intermediates in homogeneous hydrogenation reactions of olefins and acety-

267

268


[8] [9] [10] [11]

[12]

[13]

[14]

[15] [16] [17]

[18]

[19] [20] [21] [22] [23]

lenes using rhodium catalysts, J. Chem. Soc., Chem. Commun. 1965, 131. Strohmeier, W.; Mller, F. J. Chem. Ber. 1967, 100, 2812. Tolman, C. A. Chem. Rev. 1977, 77, 313. Casey, C. P.; Whiteker, G. T. Isr. J. Chem. 1990, 30, 299. Parshall, G. W. Homogeneous Catalysis: the Applications and Chemistry of Catalysis by Soluble Transition Metal Complexes, Wiley, New York 1980. (a) Theoretical Aspects of Homogeneous Catalysis; Applications of Ab Initio Molecular Orbital Theory, Eds. van Leeuwen, P. W.N. M.; Morokuma, K.; van Lenthe, J. H., Kluwer, Dordrecht 1995; (b) Computational Modeling of Homogeneous Catalysis, Eds. Maseras, F.; Lleds, A., Kluwer, Dordrecht 2002. Van der Veen, L. A.; Boele, M. D. K.; Bregman, F. R.; Kamer, P. C. J.; van Leeuwen, P. W. N. M.; Goubitz, K.; Fraanje, J.; Schenk, H.; Bo, C. J. Am. Chem. Soc. 1998, 120, 11616. van Leeuwen, P. W. N. M.; Buisman, G. J. H.; van Rooy, A.; Kamer, P. C. J. Rec. Trav. Chim. Pays-Bas, 1994, 113, 61. Pottier, Y.; Mortreux, A.; Petit, F. J. Organomet. Chem., 1989, 370, 333. Brown, J. M.; Kent, A. G. J. Chem. Soc., Perkin Trans. II, 1987, 1597. Castellanos-Pez, A.; Castilln, S.; Claver, C.; van Leeuwen, P. W. N. M.; de Lange, W. G. J. Organometallics, 1998, 17, 2543. (a) Garland, M.; Pino, P. Organometallics 1991, 10, 1693; (b) Feng, J.; Garland, M., Organometallics, 1999, 18, 417. Csontos, G.; Heil, B.; Mark, L. Ann. NY Acad. Sci. 1974, 239, 47. Chew, W.; Widjaja, E.; Garland, M. Organometallics 2002, 21, 1982. Widjaja, E.; Li, C.; Garland, M. Organometallics 2002, 21, 1991. Li, C.; Widjaja, E.; Garland, M. J. Am. Chem. Soc. 2003, 125, 5540. Pruett, R. L.; Smith, J. A. J. Org. Chem. 1969, 34, 327; (b) Pruett, R. L.; Smith, J. A. S. African Pat. 6804937,

[24]

[25]

[26]

[27]

[28]

[29] [30]

[31]

[32]

[33]

[34]

[35]

[36]

[37]

1968 (to Union Carbide Coorporation); Chem. Abs. 1969, 71, 90819. (a) Van Leeuwen, P. W. N. M.; Roobeek, C. F. J. Organomet. Chem. 1983, 258, 343; (b) Van Leeuwen, P. W. N. M.; Roobeek, C. F. Brit. Pat. 2 068 377, 1980 (to Shell); Chem. Abs. 1984, 101, 191142. Jongsma, T.; Challa, G.; van Leeuwen, P. W. N. M. J. Organomet. Chem. 1991, 421, 121. Fernandez, E.; Claver, C.; Castilln, S.; Polo, A.; Piniella, J. F.; AlvarezLarena, A. Organometallics 1998, 17, 2857. Van Rooy, A.; Orij, E.; Kamer, P. C. J.; Van Leeuwen, P. W. N. M. Organometallics 1995, 14, 34. Kamer, P. C. J.; van Rooy, A.; Schoemaker, G.; Van Leeuwen, P. W. N. M. Coord. Chem. Rev. 2004, 248, 2409. Noack, K. Spectrochim. Acta, Part A, 1968, 24, 1917. Billig, E.; Abatjoglou, A. G.;. Bryant, D. R. U. S. Pat. 4,668,651; Eur. Pat. Appl. 213,639, 1987 (to Union Carbide); Chem. Abs. 1987, 107, 7392. Billig, E.; Abatjoglou, A. G.; Bryant, D. R.; Murray, R. E.; Maher, J. M. U. S. Pat. 4,599,206, 1986 (to Union carbide); Chem. Abs. 1988, 109, 233177. van Rooy, A.; Kamer, P. C. J.; van Leeuwen, P. W. N. M.; Goubitz, K.; Fraanje, J.; Veldman, N.; Spek, A. L. Organometallics, 1996, 15, 835. Babin, J. E.; Whiteker, G. T. W. O. 93 03839, U. S. Pat. 911 518, 1992 (to Union Carbide Corporation); Chem. Abs. 1993, 119, 159872. Buisman, G. J. H.; Vos, E.; Kamer, P. C. J.; Van Leeuwen, P. W. N. M. J. Chem. Soc., Dalton Trans., 1995, 409. Casey, C. P.; Whiteker, G. T.; Melville, M. G.; Petrovich, L. M.; Gavney Jr., J. A.; Powell, D. R. J. Am. Chem. Soc. 1992, 114, 5535. Kranenburg, M.; van der Burgt, Y. E. M.; Kamer, P. C. J.; van Leeuwen, P. W. N. M. Organometallics, 1995, 14, 3081. (a) Cavalieri d’Oro, P.; Raimondt, L.; Pagani, G.; Montrasi, G.; Gregorio,

References

[38] [39]

[40] [41]

[42] [43]

[44] [45] [46]

[47]

G.; Oliveri del Castillo, G. F.; Andreeta, A. Symposium on Rhodium in Homogeneous Catalysis, Veszprem, 1978, pp. 76–83; (b) Gregorio, G.; Montrasi, G.; Tampieri, M.; Cavalieri dlOro, P.; Pagani, G.; Andreetta, A. Chim. Ind. 1980, 62(5), 389; (c) Cavalieri dlOro, P.; Raimondi, L.; Pagani, G.; Montrasi, G.; Gregorio, G.; Andreetta, A. Chim. Ind. 1980, 62, 572. Moasser, B.; Gladfelter, W. L.; Roe, C. D. Organometallics 1995, 14, 3832. Buisman, G. J. H.; van der Veen, L. A.; Kamer, P. C. J.; van Leeuwen, P. W. N. M. Organometallics 1997, 16, 5681. Vaska, L. J. Am. Chem. Soc. 1966, 88, 4100. (a) Meakin, P.; Jesson, J. P.; Tebbe, F. N.; Muetterties, E. L., J. Am. Chem. Soc., 1971, 93, 1797; (b) Meakin, P.; Muetterties, E. L.; Jesson, J. P., J. Am. Chem. Soc., 1972, 94, 5271. Whyman, R., J. Chem. Soc., Chem. Commun., 1970, 230–231. Booth, B. L.; Else, M. J.; Fields, R.; Haszeldine, R. N., J. Organomet. Chem., 1971, 27, 119–131. Cattermole, P. E.; Osborne, A. G., Inorg. Synth., 1977, 17, 115. Evans, D.; Yagupsky, G.; Wilkinson, G. J. Chem. Soc. A 1968, 2660. Chan, A. S. C.; Shieh, H. S.; Hill, J. R. J. Chem. Soc., Chem. Commun. 1983, 688. James, B. R.; Mahajan, D.; Rettig, S. J.; Williams, G. M. Organometallics, 1983, 2, 1452.

[48] Moasser, B.; Gladfelter, W. L. Inorg.

Chim. Acta 1996, 242, 125. [49] van der Slot, S. C.; Kamer, P. C. J.;

[50]

[51] [52]

[53]

[54]

[55]

[56]

[57] [58]

van Leeuwen, P. W. N. M.; Fraanje, J.; Goubitz, K.; Lutz, M.; Spek, A. L. Organometallics 2000, 19, 2504. van der Slot, S. C.; Kamer, P. C. J.; van Leeuwen, P. W. N. M.; Iggo, J. A.; Heaton, B. T. Organometallics 2001, 20, 430. Elsevier, C. J. J. Mol. Catal. 1994, 92, 285. van der Veen, L. A.; Keeven, P. H.; Schoemaker, G. C.; Reek, J. N. H.; Kamer, P. C. J.; van Leeuwen, P. W. N. M.; Lutz, M.; Spek, A. L. Organometallics 2000, 19, 872. van der Veen, L. A.; Kamer, P. C. J.; van Leeuwen, P. W. N. M. Angew. Chem. Int. Ed. Engl. 1999, 38, 336. van der Veen, L. A.; Kamer, P. C. J.; van Leeuwen, P. W. N. M. Organometallics 1999, 18, 4765. van der Slot, S. C.; Duran, Josep; Luten, Jordy; Kamer, P. C. J.; van Leeuwen, P. W. N. M. Organometallics 2002, 21, 3873. (a) Lazzaroni, R.; Settambolo, R.; Uccello-Barretta, G. Organometallics 1995, 14, 4644; (b) Lazzaroni, R.; Raffaelli, R.; Settambolo, R.; Bertozzi, S.; Vitulli, G. J. Mol. Cat. 1989, 50, 1. Casey, C. P.; Petrovich, L. M. J. Am. Chem. Soc., 1995, 117, 6007. Buhling, A.; Kamer, P. C. J.; van Leeuwen, P. W. N. M. J. Mol. Catal. A: Chemical, 1995, 98, 69.

269

7 Alkene/CO Copolymerisation Claudio Bianchini and Andrea Meli

7.1

Introduction

Alkenes and carbon monoxide are currently copolymerised in the presence of homogeneous PdII catalysts to give thermoplastic materials with a perfectly alternating structure (Scheme 7.1a) [1, 2]. The non-perfect alternation of monomers (Scheme 7.1b) has been uniquely observed for ethene/CO copolymerisation reactions catalysed by PdII precursors with anionic P–O ligands [3]. 13C {1H} NMR spectroscopy provides a reliable and immediate tool to discriminate between perfect (Figure 7.1a) and non-perfect alternation (Figure 7.1b). The active species in the catalytic cycles are square-planar PdII complexes of the formula [PdII(X)(S)(L–L)]Y where L–L is a chelating ligand with the same or different donor atoms among P, N, O and S; X is the growing polyketone chain or hydride; S may be a solvent molecule, a co-monomer, or a keto group from the chain. Finally, Y is a counter-anion of weak nucleophilicity in order to avoid competition with the co-monomer for coordination to palladium (Scheme 7.1). O

R

O

cat + CO L cat =

X

Y

R

n

(a

perfect alternation

x

(b n

non-perfect alternation

Pd L’

S

L = L’ or L ≠ L’ = P, N, O,S

S = MeOH,H2O, CO, alkene, O=C(R)-

X = acyl, alkyl, H

Y = OTs, CF3CO2, PF6, BAr4


Scheme 7.1

272

7 Alkene/CO Copolymerisation

Figure 7.1 13C{1H} NMR spectra (methylene region) of (a) alternating (hexafluoroisopropanol/ C6D6) and (b) non-alternating (hexafluoroisopropanol/CDCl3) ethene/CO copolymers. (From E. Drent, R. van Dijk, R. van Ginkel, B. van Oort, R. I. Pugh, Chem. Commun. 2002, 964.)

Either protic (alcohols, preferentially methanol) or aprotic solvents (toluene, dichloromethane, THF) can be used, depending on the structure of the metal precursors that can generate the catalysts by a number of pathways. Metals other than palladium, for example nickel [4], can form active catalysts for alkene/CO copolymerisation, yet with largely lower productivities as compared to structurally similar palladium precursors [1]. For this reason, only PdII-catalysed alkene/CO copolymerisation reactions are reviewed and commented in the present chapter. Over the last ten years, remarkable progress has been achieved in the understanding of the mechanism of the alkene/CO copolymerisation [1j], which has allowed the design of increasingly efficient catalysts, capable of producing more than 50 kg of copolymer ((g Pd)–1 h–1) with only ppm quantities of residual palladium [1]. A number of ex situ spectroscopic techniques, multinuclear NMR, IR, EXAFS, UV–vis, have contributed to rationalise the overall mechanism of the copolymerisation as well as specific aspects related to the nature of the unsaturated monomer (ethene, 1-alkenes, vinyl aromatics, cyclic alkenes, allenes). Valuable information on the initiation, propagation and termination steps has been provided by endgroup analysis of the polyketone products, by labelling experiments of the catalyst precursors and solvents either with deuterated compounds or with easily identifiable functional groups, by X-ray diffraction analysis of precursors, model compounds and products, and by kinetic and thermodynamic studies of model reactions. The structure of some catalysis resting states and several catalyst deactivation paths have been traced. There is little doubt, however, that the most spectacular mechanistic breakthroughs have been obtained from in situ spectroscopic studies,

7.1 Introduction

primarily based on NMR [5–8] and IR [9, 10] spectroscopies, especially since the advent of fast spinning high pressure NMR sapphire tubes (Figure 7.2) [5–8] and flow cells (Figure 7.3) [6a,b] as well as IR cells at low and high pressure (Figure 7.4) [9, 10]. It is just with the application of in situ NMR and IR spectroscopy to the study of the alternating alkene/CO copolymerisation mechanisms that this chapter is largely concerned.

Figure 7.2 Fast spinning, high pressure NMR sapphire tube with safety and charging device (constructed at ICCOMCNR, 2003).

Figure 7.3 Exploded view of the high pressure in situ NMR flow cell. (From J. A. Iggo, D. Shirley, N. C. Tong, New J. Chem. 1998, 1043.)

Figure 7.4 High pressure IR cell connected to an autoclave (constructed at ICCOM-CNR, 2003).

273

274


7.2

Catalytic Cycles of Alkene/CO Copolymerisation

The most common alkenes employed in the Pd-catalysed synthesis of alternating polyketones are ethene, styrene, propene and cyclic alkenes such as norbornene and norbornadiene. Even though the mechanism does not vary substantially with the alkene, the reactions of the various co-monomers are here reported and commented on separately, starting with the ethene/CO copolymerisation, which is still the most studied process. As a general scheme, the proposed catalytic cycles are presented first, then the spectroscopic experiments that have allowed one to elucidate each single mechanistic step. 7.2.1

Mechanism of Ethene/CO Copolymerisation Methanol and Other Protic Solvents Scheme 7.2 summarises the principal steps of the alternating ethene/CO copolymerisation in MeOH by PdII catalysts modified with bidentate ligands [1b,c]. Two competing cycles, connected by two cross termination steps, are contemporaneously at work for the production of the alternating polyketone, the prevalence of either cycle depending on the experimental conditions. Cycle B initiates (I) with the insertion of ethene into a Pd–H bond that can be generated in a variety of ways (see below). Insertion of CO into the resulting alkyl complex is reversible and faster than ethene insertion, while CO insertion into the Pd–acyl is thermodynamically disfavoured. Since ethene insertion into the Pd–acyl is rapid and irreversible, the propagation (P) can occur by alternating insertion of CO and ethene. The copolymer produced by this cycle contains either keto-ester or diketone end groups, depending on the termination path: the keto-ester end structure is obtained via methanolysis (M) of a Pd–acyl bond, while the diketone structure occurs by protonolysis (H) of a Pd–alkyl intermediate. A copolymer with keto-ester end groups can be produced also by protonolysis of a Pd–alkyl bond formed during the propagation in the competing cycle A that starts with the insertion of CO into a Pd–OMe bond to give a Pd carbomethoxy complex. The diester structure is obtained via methanolysis of a Pd–acyl coming from cycle A. The factors that control the strictly alternating copolymer chain with no detectable errors (e. g., microstructures involving double insertion of ethene) have been the object of detailed studies since the discovery of the first PdII catalysts for the alternating alkene/CO copolymerisation [11]. Sen was the first to demonstrate that double carbonylation is thermodynamically unfavorable and to suggest that the higher binding affinity of PdII for CO relative to ethene inhibits multiple ethene insertions, even in the presence of very low concentrations of CO [12]. Therefore, once a palladium alkyl is formed, CO coordination ensures that the next monomer will be a CO molecule to generate the acyl complex. Replacing MeOH with water does not significantly affect the catalytic cycle of copolymerisation, which involves the usual steps of insertion of ethene into 7.2.1.1

7.2 Catalytic Cycles of Alkene/CO Copolymerisation

β-chelates

+ Pd

H

O O

n-1

MeOH O H

H

P

O + Pd

nH

275

C2H4

nH

Diketone O

+ Pd OMe

CO

+ Pd

I

A

nH

MeOH

O H

P

M

OMe n

B

O + Pd

+ Pd

Keto-ester

OMe

(n) CO (n-1) C2H4 H

H MeOH

(n) C2H4 (n-1) CO

P

O + Pd

+ Pd H

OMe n O

+ Pd CO

P

Scheme 7.2

O O

O + Pd

C2H4

I

OMe n

M MeOH

OMe O

γ-chelates

O

MeO O Diester

n-1

Pd–H to give Pd–ethyl (initiation), CO insertion to give Pd–acyl, nC2H4/nCO insertions (propagation), and hydrolysis to give Pd–OH (termination). However, depending on the structure of the catalysts, polyketones with diketone and/or acid end groups may be obtained. The mechanism reported in Scheme 7.3 describes the reaction catalysed by [Pd(TFA)2(Na2DPPPDS)]/TsOH (Na2DPPPDS ¼ Na2(O3S(C6H4)CH2)2C(CH2PPh2)2; TFA ¼ trifluoroacetic acid; TsOH ¼ p-toluenesulfonic acid) [5a]. The exclusive formation of copolymer with diketone structure and the evolution of CO2 are clearly indicative of the occurrence of the water-gasshift reaction (WGS). A slightly more complicated mechanism has been proposed by Sheldon for copolymerisation reactions assisted by PdII complexes containing diphosphine ligands sulfonated at the phenyl substituents [13]. The formation of some copolymer bearing acid end groups in fact requires a termination step involving hydrolysis of Pd–acyl [13a]. Therefore, it was suggested that both b- and g-chelates are formed, each with its own chain transfer mechanisms, as shown in Scheme 7.2.

OMe n

276


CO2

[Pd]-COOH

[Pd]-OH H-CH2CH2

[Pd]-H =

O CCH2CH2

C2H4

[Pd]-H

[Pd]-CH2CH2-H CO O [Pd] C-CH2CH2-H n C2H4/n CO

H n

H2O O [Pd]-CH2CH 2 CCH2CH 2

H n

P + H Pd P

-O

3S -O S 3

Scheme 7.3

In particular, hydrolysis of the g-chelate would give acid end groups and Pd–H, while the protonation of the enolate, in equilibrium with the b metallacycle via hydride shift (vide infra) would give keto end groups and Pd–OH (Scheme 7.4) [13c]. O P H + C2H4, CO,C2H4 Pd P L

P Pd P

+ O

C2H4

P

P Pd P

+

+Na-O S 3

P = P

O P

- CO2 CO + [PdOH(P-P)] + CH3CH2C(O)- P

CO

hydride shift H H2O

CH3 P Pd P

+

H2O HO(O)CCH2CH2C(O)-

+Na-O S 3

P P

SO3-Na+ SO3-Na+

P

+

O

P

P H + Pd P L

Scheme 7.4

Aprotic Solvents The mechanism of ethene/CO copolymerisation in aprotic solvents is essentially limited to cycle B in Scheme 7.2. The main differences from the reactions in alcohols are in the precursors, that already contain an alkyl ligand to initiate the copolymerisation, and the termination step involving b-H elimination, unless water or a protic acid is present in the reaction in which case protonolysis may also occur. Scheme 7.5 illustrates the main steps for a typical (P–P)Pd–alkyl precursor. The presence of water in the organic solvent, even in very low concentration, may lead to chain transfer by protonolysis with formation of mononuclear hydroxo complexes that may convert to m-OH binuclear species. Depending on the chelating ligand, these m-OH compounds may be a dead-end for catalysis or resting states capable of re-initiating the copolymerisation process (Scheme 7.5) [5e,f ]. 7.2.1.2


P

+

OH Pd

O H

P

O

P Pd P

monomers + R S

P

H+/ S

Pd

+

P Pd P

P

O

P

R

R = alkyl; S = solvent, monomer or anion

P

O

P P

P

S 2+ S

O β-H

Scheme 7.5

2+

H2O

R

P

P Pd

Pd

S

P

H O

P

277

R

Pd P

R

H + S

Formation of Active PdII Sites and Initiation of Ethene/CO Copolymerisation The catalysts for the ethene/CO copolymerisation in protic solvents are generally generated by neutral [Pd(X)2(L–L)] or cationic [Pd(S)2(L–L)]X2 precursors that can also be prepared in situ by reaction of a PdII salt (commonly Pd(OAc)2 or [Pd(NCMe)4](BF4)2) with a chelating diphosphine or dinitrogen ligand [1]. In either case, a slight excess of a strong protic acid (commonly trifluoroacetic acid or p-toluenesulfonic acid) is added to the reaction mixture to neutralise anionic nucleophiles (e. g., acetate ions) that may compete with MeOH in the activation of the precursor and with the co-monomers in the propagation step. The use of phosphine-modified catalysts prepared in situ may be highly risky. Indeed, recent studies have shown that, depending on the chelating diphosphine, the preparation of the catalyst precursor in situ may give much lower productivities as compared to reactions where a preformed PdII complex is used. For example, 1,3-bis(diphenylphosphino)propane (dppp) can be used in either preformed or in situ generated systems in MeOH without affecting the results, whereas 1,2-bis(diphenylphosphino)ethane (dppe), 1,2-bis(diphenylphosphino)benzene (dppbz) and 1,3-bis[di-(o-methoxyphenyl)phosphino]propane (bdompp) give much worse results when used in situ [14, 15]. The very low activity of the dppe-PdII catalyst prepared in situ in MeOH has been accounted for by a combined 31P{1H} NMR and kinetic study [14, 15]. It is now apparent that dppe reacts with Pd(OAc)2 in MeOH at room temperature to give the kinetic bis-chelate product [Pd(dppe)2](OAc)2 (1), which slowly reacts (t1/2 ¼ 4 h at room temperature) with the residual Pd(OAc)2 converting to the thermodynamic, and catalytically active, mono-chelate complex [Pd(OAc)2(dppe)] (2) (Scheme 7.6). The characteristic of Pd(OAc)2 to exist in MeOH in various forms, from monomers to aggregates, depending on the tem7.2.1.3

P 2 Pd(OAc) 2 + 2 P Scheme 7.6

MeOH fast

P

P

2 OAc- + Pd(OAc)2

Pd P

2+

P

(1)

MeOH slow

P

OAc Pd

2 P

OAc (2)

278


Figure 7.5 Variable-temperature 31P{1H} NMR study (CD2Cl2, 81.01 MHz) of the reaction of Pd(OAc)2 with dppe: (a) at –70 hC after dissolving Pd(OAc)2 and dppe in CD2Cl2 at –70 hC; (b) at –20 hC after 50 min; (c) at room temperature after 10 min.

perature and concentration was shown to be responsible for the initial shortage of PdII and hence for the formation of the bis-chelate complex. In CH2Cl2, where the reaction of the bis-chelate complex (1) with Pd(OAc)2 to give (2) is much faster than in MeOH (Figure 7.5), no significant difference between dppe or dppp PdII catalysts has been observed in CO/ethene copolymerisation [5e,f ]. The activation of (P–P)PdII promoters in MeOH proceeds via formation of PdII–OMe (Eq. (1)) that can straightforwardly initiate the catalysis cycle or generate PdII–H via b-H elimination, yielding formaldehyde (Eq. (2)) [16]. The fast kinetics under real copolymerisation conditions do not allow for the spectroscopic detection of Pd–H initiators. However, their formation has been unambiguously assessed by end-group analysis, isotopic labelling experiments and model reactions [1]. Pd2þ þ MeOH p Pd–OMeþ þ Hþ

(1)

Pd–OMeþ p Pd–Hþ þ HC(O)H

(2)

Tertiary alcohols are rather inefficient in alkene/CO copolymerisation as the formation of alkoxy palladium complexes is less favored for them than for primary and secondary alcohols, while the lack of b hydrogens does not allow the formation of Pd–H. Some primary alcohols activated by electron-withdrawing substituents, e. g., CF3CH2OH, are equally unable to form Pd–H due to their low propensity to oxidation [6c]. In the presence of water, Pd–H moieties are generated by WGS (Eq. (3)), while the occurrence of a Wacker-type reaction has also been suggested (Eq. (4)) [1]. Pd2þ þ H2O p Pd–OHþ þ Hþ Pd–OHþ þ CO p Pd–C(O)OHþ Pd–C(O)OHþ p Pd–Hþ þ CO2


or Pd2þ þ CO p Pd–CO2þ Pd–CO2þ þ H2O p Pd–C(O)OHþ þ Hþ Pd–C(O)OHþ p Pd–Hþ þ CO2

(3)

Pd2þ þ C2H4 þ MeOHp Pd–Hþ þ CH2¼CHOMe þ Hþ

(4)

Formic acid, (Eq. (5)) and protic acids may generate Pd–H initiators by oxidative attack on Pd0 species [1]. Pd–OAcþ þ HC(O)OH p Pd–O–C(O)Hþ þ HOAc Pd–O–C(O)Hþ p Pd–Hþ þ CO2

(5)

To achieve high productivities, commonly expressed as kg of polyketone (g Pd)–1, the reactions are performed in the presence of an excess of either organic (e. g., 1,4-benzoquinone (BQ), 1,4-naphthoquinone) or inorganic (e. g., CuII salts) oxidants. BQ is the most common oxidant and its presence in the reaction mixture may increase the overall productivity by more than an order of magnitude [1b]. In general, the oxidant does not affect the copolymer molecular weight, rather it favors the formation of ester end groups due to the occurrence of the reaction shown in Eq. (6). Therefore, the oxidant influences neither the propagation nor the termination; it simply serves to maintain a high concentration of active PdII sites in the reaction mixture. Pd–Hþ þ BQ þ MeOH þ CO p Pd–C(O)OMeþ þ BQH2

(6)

As previously mentioned, the catalyst precursors in aprotic solvents generally contain a Pd–alkyl moiety for immediate insertion of CO without the need of activators. In some cases, bis-halide PdII precursors have been employed in conjunction with activators such as methyl aluminoxane (MAO), which is able to replace one halide ligand with a methyl group and to create a free coordination site at the metal [17].

7.2.1.4

Chain Propagation of Ethene/CO Copolymerisation Irrespective of the solvent, the chain propagation comprises two alternating migratory insertion steps involving Pd(alkyl)(CO) and Pd(acyl)(ethene) moieties. The Pd(alkyl)(ethene) mis-insertions are virtually absent (ca. one double ethene insertion for every ca. 105 CO insertions into Pd-alky) [18]. Early studies under actual copolymerisation conditions in MeOH showed a dependence of the copolymerisation rate on the ethene pressure, and therefore it was concluded that the insertion of ethene is rate-determining [1b]. Recent investigations have proved that the overall rate-limiting step is mechanistically more complicated than a simple migratory

279

280


insertion of Pd(acyl)(ethene), and apparently involves a competitive coordination of ethene with the other ligands present in the reaction mixture, such as CO, MeOH and keto groups of the propagating chain (vide infra). Unfortunately, under real catalytic conditions (typically, MeOH, 80–85 hC, 30–40 bar 1:1 CO/C2H4, excess protic acid), the chain transfers by protonolysis of b-chelates and by methanolysis of g-chelates (Scheme 7.2) are generally too fast to allow the spectroscopic detection of these intermediates. Only with the use of some mediocre PdII catalysts under mild experimental conditions, have b-chelates and other intermediates or resting states relevant to catalysis been successfully intercepted by NMR. Better results have been achieved by means of in situ IR spectroscopy because of its faster time-scale.

In Situ High Pressure NMR Studies of Ethene/CO Copolymerisation in Protic Solvents High pressure 31P{1H} NMR spectra (HP NMR) taken with fast spinning sapphire 10 mm OD sapphire tubes for ethene/CO copolymerisation reactions catalysed by (P–P)PdII catalysts in MeOH or other alcohols at 60–90 hC are generally poorly informative, showing single 31P signals attributed to species with no direct correlation to the catalysis cycle. It has been established that these resonances are generated by “(P–P)Pd2þ” moieties in rapid exchange with solvent molecules and/or counter-anions. These PdII species would constitute a sort of reservoir of active palladium, a small aliquot of which is delivered into the catalysis cycle through various activation processes (for example, any of reactions (1)–(4)) [5a–c]. The 31P{1H} HP NMR picture of CO/ethene copolymerisation in MeOH is exemplified by the sequence of variable-temperature spectra shown in Figure 7.6 relative to a reaction catalysed by [Pd(TFA)2(dppp)]. It has been observed that the intensity of the 31P NMR signal decreases with time, which is apparently due to the irreversible reductive degradation of PdII species to Pd0 metal [5b, c]. A quite similar picture has been reported for copolymerisation reactions catalysed by [Pd(TFA)2(Na2DPPPDS)] in water in the presence of an excess of TsOH (Figure 7.7) [5a]. Neither CO adducts nor ethene adducts were observed. Instead, a broad featureless resonance appeared at room temperature, which was assigned to several species containing trifluoroacetate, p-toluenesulfonate (the reaction was performed in the presence of a slight excess of TsOH), H2O, hydroxo and/or m-hydroxo species, eventually in equilibrium with each other (Figure 7.7a). In contrast, 7.2.1.5

Figure 7.6 Selected 31P{1H} NMR spectra recorded during a CO/ethene copolymerization assisted by [Pd(TFA)2(dppp)] (10 mm sapphire tube, MeOH-d4, 20–85 hC, 81.01 MHz). (a) Spectrum at room temperature; (b) after addition of TsOH (5 equiv) and pressurization with 600 psi 1:1 CO/ethene; (c) at 85 hC during copolymer synthesis.

7.2 Catalytic Cycles of Alkene/CO Copolymerisation Figure 7.7 Selected 31P{1H} NMR spectra recorded during a CO/ethene copolymerization assisted by [Pd(TFA)2(Na2DPPPDS)] in the presence of 20 equiv of TsOH and a 1:1 CO/C2H4 pressure of 600 psi (10 mm sapphire tube, D2O, 20–85 hC, 81.01 MHz). (a) Spectrum at room temperature; (b) heating to 85 hC; (c) cooling to room temperature; (d) heating again at 85 hC.

a reversible single sharp signal was observed at 85 hC when the polymer began to separate as an off-white microcrystalline solid and accumulated at the water–gas interface (Figure 7.7b). Since this sharp signal was displayed even in the absence of CO and/or ethene, it was concluded that most of the palladium is incorporated in species that reside out of the catalysis cycle. 31 P{1H} HP NMR spectra showing a single signal due to the precursor or related species have been observed even in the course of copolymerisation reactions in 1,1,1,3,3,3-hexafluoro-2-propan-2-ol-d2 (HFIP-d2) where the processes are truly homogeneous [5a–c]. Only with less efficient catalysts and at low temperature, have b-chelate intermediates been intercepted by 31P{1H} HP NMR spectroscopy in the course of copolymerisations in MeOH-d4 [5g]. The unambiguous detection of b-chelates has been observed in a reaction catalysed by the 1,1’-bis(diphenylphosphino)ferrocene complex [Pd(H2O)2(dppf)](OTs)2 (3) at room temperature (Scheme 7.7) [5g]. As shown in the sequence of 31P{1H} NMR spectra reported in Figure 7.8, the b-chelate intermediates 4 disappeared already at 50 hC. A parallel model study confirmed the formation and the structure of the dppf b-chelates and also provided information of more elusive intermediates (see Section 7.2.1.8) [19].

P Fe

OH2 Pd OH2

P (3) Scheme 7.7

OTs 40 bar C2H4 /CO MeOH-d4, RT

OTs

P Fe

Pd P

R O

4

O 4 R =C2H4D(H), OCD3

Notably, this HP NMR investigation showed the formation of a transient binuclear m-H-m-CO complex, [Pd2(m-H)(m-CO)(dppf)2]OTs (5), and of the termination product [Pd(m-OH)(dppf)]2(OTs)2 (6) (Chart 7.1). Based on the in situ study, these compounds could be isolated, characterised and used to catalyse copolymerisation reactions. Both complexes proved to be active in batch copolymerisation reactions. However, the productivities in polyketone were significantly lower than those

281

282


P Fe

Pd P

Pd C O

OTs

P

H

Fe P

H O

P Fe

Pd

P Pd

O H

P

(5)

(OTs)2 Fe

P (6)

Variable-temperature P{1H} NMR study (sapphire tube, MeOH-d4, 81.01 MHz) of the carbonylation reaction of ethene catalyzed by (3): (a) Dissolving (3) in MeOH-d4 under nitrogen at room temperature; (b) after the tube was pressurized with 40 bar of CO/C2H4 (1:1) at room temperature; (c) after 10 min at 50 hC; (d) after 10 min at 70 hC; (e) after the tube was cooled to room temperature. Figure 7.8 31

obtained with the mononuclear precursor, [Pd(H2O)2(dppf)](OTs)2, due to a long induction period [5g].

In Situ High Pressure NMR Studies of CO/Ethene Copolymerisation in Aprotic Solvents In aprotic solvents, chain transfer occurs exclusively by b-H elimination, unless a protic acid or water is present in the reaction mixture, in which case protonolysis may occur. Indirect evidence (for example, Mw and Mn measurements) proves that b-H chain transfer in aprotic solvents is slower than methanolysis in protic solvents with comparable structures of the PdII catalyst [5f, 17, 20, 21]. This effect and the possibility of using well-defined catalysts have remarkably favored the use of in situ NMR spectroscopy for the detection of intermediates during CO/copolymerisation in organic solvents. Figure 7.9 shows a sequence of 31P{1H} HP NMR spectra in CD2Cl2 acquired at different times and temperatures in the course of an ethene/CO copolymerisation catalysed by [Pd(Me)(NCMe)(dppp)]PF6 [5f ]. Similar spectral features were observed for several PdII methyl complexes of the formula [Pd(Me)(NCMe)(P–P)]PF6 [P–P ¼ 1,2-(bis-diphenylphosphino)ethane (dppe); meso- and rac-2,4-bis(diphenylphosphino)pentane (meso- and rac-bdpp); meso- and rac-2,3-bis(diphenylphosphino)butane 7.2.1.6


Variable-temperature 31P{1H} NMR study (sapphire tube, CD2Cl2, 81.01 MHz) of ethene/CO copolymerisation: (a) Dissolving [Pd(Me)(NCMe)(dppp)]PF6 in CD2Cl2 under nitrogen at room temperature; (b) after the tube was pressurized with 40 bar of ethene/CO (1:1) at room temperature; (c) after 15 min at 50 hC; (d) after 10 min at 70 hC; (e) after 60 min at 70 hC; (f) after the tube was depressurized at room temperature. Figure 7.9

(meso- and rac-dppb)]. In all cases, the spectra were taken with the use of fast spinning 10 mm OD sapphire tubes [5e] that reduce remarkably gas diffusion problems, as shown by control experiments in which the gases were directly bubbled into the reaction mixtures prior to NMR acquisition. Besides proving the formation of b-chelates [Pd(CH2CH2C(O)Me)(P–P)]þ at room temperature, the spectra showed the occurrence of chain-transfer by protonolysis with adventitious water to give the m-hydroxo compounds cis/trans [Pd(m-OH)(P–P)]22þ as well as the conversion of the latter compounds into cis/trans bis-chelates [Pd(P–P)2]2þ (Chart 7.2) [5f ]. Independent experiments with isolated compounds showed that the m-OH and bis-chelate complexes are not dead ends, and can reenter the catalysis cycle to give alternating polyketones. H O

P Pd

P

H O

P Pd

Pd O H

P

2+

P

P

2+

P

P P

2+

Pd

Pd P

Pd O H

P

P

2+

P

P

trans-structures

P

P

cis-structures

In situ NMR experiments in anhydrous CD2Cl2 confirmed unequivocally that the m-OH complexes are just generated by reaction of water with the b-chelates [5f ]. For example, Figure 7.10 illustrates a reaction catalysed by [Pd(Me)(NCMe)(mesodppb)]PF6 in anhydrous CD2Cl2 during which no m-OH complex was formed, whereas the bis-chelates were equally produced through a mechanism involving partial catalyst deactivation (Section 7.2.6).

283

284


Figure 7.10 31P{1H} NMR study (sapphire tube, CD2Cl2, 81.01 MHz) of CO/C2H4 copolymerization: (a) Dissolving [Pd(Me)(NCMe)(meso-dppb)]PF6 in CD2Cl2 under nitrogen at room temperature; (b) after the tube was pressurized with 40 bar of CO/C2H4 (1:1) at room temperature; (c) at 70 hC (first spectrum); (d) after 45 min at 70 hC.

In Situ High Pressure IR Studies of Ethene/CO Copolymerisation Unlike NMR, IR spectroscopy in actual catalytic conditions has provided experimental evidence for intermediate species, even at high temperature. Luo et al. have studied the copolymerisation of CO and ethene assisted by (dppp)PdII catalysts (MeOH, TFA, 60 bar 1:1 CO/C2H4) via 1H NMR, 31P NMR, IR and EXAFS techniques [9]. While 31P{1H} and 1H HP NMR were uninformative as usual, a series of Pd-K-edge EXAFS experiments showed the presence of five-coordinate PdII species in the course of polyketone formation. Bands at 1638 and 1616 cm–1 attributable to g- and b-chelates, respectively, were observed at 85 hC by means of high pressure IR (HP IR) spectroscopy in the course of an ethene/ CO copolymerisation catalysed by [Pd(OTs)2(dppp)] in 2-ethyl hexanol (40 bar C2H4 and 4 bar CO) (Figure 7.11). A band at 1970 cm–1 was assigned to a Pd(alkyl)CO intermediate whose intensity increased with the partial pressure of CO at the expense of those of the other species. The formation of stable PdII b-chelates by alkene insertion into Pd–acyl bonds has led several authors to conclude that the greater binding affinity of CO to PdII ions as compared to ethene and the thermodynamics preventing double CO insertion are not the only factors accounting for the perfectly alternating structure and the lack of termination by b-elimination [1b, 22, 23]. Indeed, the internal coordination of the growing polymer chain to the metal center as in the b-chelates would play a relevant role in promoting strict alternation and also in inhibiting b-elimination paths. The displacement of the chelate carbonyl by ethene is not an easy process, essentially for steric reasons, and may contribute to raising the energy barrier to ethene insertion in Pd–alkyl under polyketone synthesis conditions. Carbon monoxide is smaller and exhibits a greater binding affinity for palladium than ethene. It was suggested, therefore, that b-chelate opening is brought about by CO to generate a six-membered metallacycle in which the chelate carbonyl group is located g with respect to palladium. As six-membered rings are less stable 7.2.1.7


Figure 7.11 High pressure in situ IR spectra under real reaction conditions with [Pd(OTs)2(dppp)] as catalyst. [Pd(OTs)2(dppp)] ¼ 1 mmol, 2-ethyl hexanol ¼ 80 ml. (a) 80 hC, 40 bar C2H4. (b) 85 hC, introduce 4 bar CO. (c) 85 hC, 42 bar, after 40 min. (d) 85 hC, 38 bar, after 120 min. (e) 85 hC, 36 bar, after 240 min. (f) 85 hC, 29 bar, introduce 10 bar CO and after 20 min. (g) 90 hC, 28 bar, introduce 13 bar CO and after 30 min. (From H.-K. Luo, Y. Kou, X.-W. Wang, D.-G. Li, J. Mol. Catal. A. 2000, 151, 91.)

than five-membered ones, it has also been proposed that the propagation occurs by displacement of the g carbonyl group by ethene, followed by formation of the next five-membered ring [1b]. This mechanistic view of the propagation cycle involving Pd-diphosphine catalysts is still largely accepted. Recently, however, a different mechanism has been reported by Drent on the basis of gas-phase ethene/CO copolymerisation experiments in the presence of solid catalysts [10]. Polarization modulation reflection absorption infrared spectroscopy (PM-RAIRS) was employed to follow the reaction of CO, C2H4 and CO/C2H4 with microcrystalline [Pd(Me)(OTf)(dppp)] deposited onto a gold coated wafer. Single insertion steps were observed by alternately exposing the catalyst precursor to low CO (500– 333 mbar) and ethene (333 mbar) pressures (Figure 7.12). The PM-RAIRS study clearly showed that the b- and g-chelate complexes [Pd(CH2CH2C(O)Me)(dppp)]þ and [Pd(C(O)CH2CH2C(O)Me)(dppp)]þ are resting-

285

286


Figure 7.12 In situ PM-RAIRS spectra of a microcrystalline sample of [Pd(Me)(OTf)(dppp)]. (a) At room temperature; (b) under 500 mbar CO; (c) under 2 mbar CO and 333 mbar of ethene; (d) under subsequent exposure to 750 mbar CO; (e) during subsequent polymerisation under 666 mbar of ethene/CO (evolution of the spectrum at 15 min intervals). (From W. P. Mul, H. Oosterbeek, G. A. Beitel, G.-J. Kramer, E. Drent, Angew. Chem. Int. Ed. Engl. 2000, 39, 1848.)

states in equilibrium under catalytic conditions (666 mbar ethene/CO, room temperature). Most importantly, it was discovered that ethene insertion into the Pd–acyl bond of the g-chelate complex [Pd(C(O)CH2CH2C(O)Me)(dppp)]þ does not occur unless some CO is present. On the basis of the PM-RAIRS evidence, Drent has proposed a catalytic cycle (Scheme 7.8) where ethene insertion in the propagation step is CO-assisted and the substitution of the chelating ketone in c by ethene would proceed in two consecutive steps: associative substitution of the chelating ketone by CO (c p d p e), followed by associative substitution of CO by ethene (e p f p i). The disruption of the chelate structure of c would be more facile for CO than for ethene for steric reasons (end-on vs. side-on approach).

Model Synthetic Studies of Ethene/CO Copolymerisation Over the last ten years, several model studies of the ethene/CO copolymerisation have been carried out, aimed at investigating the reactions of isolated compounds with CO, ethene and other reactive components of the catalytic mixtures. Due to space limitations and the presence in the literature of excellent reviews on this sub7.2.1.8


P P

CO

+

CO

P

Pd

P

O

+

Pd O

R

R

β-chelate +

O + P P

P P

Pd O

CO

+

O

e

R

Pd

R

R

P

- CO

O Pd

P

R

O

P

+ P P

Pd

P

CO

γ-chelate

Pd

d

+

O O

a

O O

CO R

+ P P

Scheme 7.8

c

+

CO Pd O O R

CO O P Pd P O b

R

ject [1], only a few examples are commented upon here; this will give the reader an idea of the synthetic strategies involved, which have made large use of NMR and IR spectroscopy. By using the dissymmetric diphosphine 1-diphenylphosphino-2-tert-butyl-3-dicyclohexylphosphinoprop-1-ene, van Leeuwen has demonstrated that the insertion of CO into Pd–Me proceeds via migration of the hydrocarbyl group to the CO ligand [24]. The early stages in the chain growth process have been mimicked by Braunstein with the use of a Pd–Me complex supported by an acetamido-derived P–O ligand. Four catalytic intermediates were intercepted by the sequential addition of CO–ethene–CO–ethene, and the occurrence of reversible and irreversible steps was established (Scheme 7.9). Unlike diphosphine ligands [10], the insertion of ethene into the g-chelate acyl complex was found to be a facile process occurring without the intervention of CO [25]. Monitoring, by multinuclear NMR spectroscopy, the sequential addition of CO and ethene to a CD2Cl2 solution of the b-chelate complex [Pd(CH2CH2C(O)Me)(dppf)]OTf allowed Bianchini to confirm the mechanism of the alternating propagation as well as demonstrate that the b-chelates are true catalysis resting states (Scheme 7.10) [5g].

287

288


O H

P

N

Me

Me

H

+ CO

N

Pd

Pd NCMe

O

1h

H

O

Me

N

Me

+

P Me NCMe

1.5 h

+ Pd

O

O

- CO

CO 1.5 h

H

C2H4 Me

O

1h

N

Me

Pd O

O

P

CO, 1 bar

Fe

OTf CO

Pd

Pd P

-40 °C

O

P P

O

Me

OTf

P

OTf Pd

Fe

+

P

O

P Fe

Me

O

Pd

Scheme 7.9

P

N

Me

+

P O

H

C 2H4

C2H4

O

C2H 4, 1 bar -60 °C

P Fe

OTf Pd

P

O

O

CO

O O 2

Scheme 7.10

Kinetic and Thermodynamic Studies of Ethene/CO Copolymerisation in Aprotic Solvents Besides the isolation and characterisation of several catalytically relevant intermediates, model reaction studies, generally based on variable-temperature NMR experiments in CD2Cl2, with isolated PdII complexes have provided valuable kinetic and thermodynamic information on the mechanism of the alternating ethene/CO copolymerisation. The first study of the energy barriers associated with the migratory insertion reactions occurring in the initiation and propagation steps was carried out with PdII complexes stabilised by structurally rigid dinitrogen ligands such as bipyridine (bipy), phenanthroline (phen) and bis(arylimino)acenaphthene (Ar-BIAN) [22, 26, 27]. Detailed mechanistic studies have been carried out with the phen complexes [Pd(R)(L)(phen)][3,5-(CF3)2C6H3)] where the R/L combination may be Me/CO; Me/C2H4; C(O)Me/C2H4; C(O)Me/CO; CH2CH2C(O)Me/C2H4; CH2CH2C(O)Me/ CO; C(O)CH2CH2C(O)Me/C2H4; C(O)CH2CH2C(O)Me/CO. A low-temperature NMR study of the intramolecular migratory insertions involving these R/L combinations provided the barriers for insertion which increase in the order: DG‡ (RpCO) z 15 kcal mol–1 (–66 hC) I DG‡ (AcpC2H4) z 17 kcal mol–1 (ca. –45 hC) 7.2.1.9


I DG‡ (RpC2H4) z 19 kcal mol–1 (–25 hC). The relative ground state stabilities of several species, including the b-chelate [Pd(CH2CH2C(O)Me)(phen)]þ and the g-chelate [Pd(C(O)CH2CH2C(O)Me)(phen)]þ were determined through a combination of competitive and relative equilibria studies. Based on this multiform investigation, a complete mechanistic cycle of chain propagation was proposed (Scheme 7.11) [26]. O

+

(phen)Pd O +

O

CO

CO

(phen)Pd

O

P

+

(phen)Pd

P

CO C2H4

O

+

CO (phen)Pd Scheme 7.11

P

(phen)Pd

+

CO

+

O

C2H4

P

(phen)Pd

C2H4

R

CO

C2H4

P O

Kinetic studies of migratory insertion reactions of the ligands that are involved as (P–P)PdII fragments in either the propagation cycle of ethene/CO copolymerisation or ethene dimerisation to butenes have been reported by Brookhart [28] and Bianchini [5e, f ]. First-order kinetics has been established by Brookhart for the insertion reactions, Eq. (7)–(10), whose free energies of activation were determined experimentally by in situ 1H NMR spectroscopy (Scheme 7.12). For reactions (7) and (8), the DH‡ [15.2(7) and 14.8(7) kcal mol–1, respectively] and DS‡ [-6.2(2.9) and 0.1(3.5) cal mol–1 K–1, respectively] parameters were calculated from Eyring plots [28]. In contrast to theoretical results reported by Morokuma [29] and Ziegler [30], as well as previous studies with PdII-phen model compounds [26], the lowest experimental energy barrier was found for the migratory insertion of the acyl(ethene) complex (Eq. (10)). The relative rates of alkyl to CO and alkyl to ethene migratory insertion reactions allowed one to estimate that sequential ethene insertions occur once for every ca. 105 insertions of CO into the Pd–alkyl bond [18]. Using in situ NMR spectroscopy, Brookhart has also studied the activation barriers for the migratory insertion steps corresponding to chain growth in ethene/ CO copolymerisation catalysed by dppe-derived nickel(II) complexes [4a]. Activa-

289

290


P P

+

CH3 Pd

+ C2H4 C3H6

P P

+

CH2CH3

(7)

Pd

+ +

∆G (-45.6 °C) = 16.6(1) kcal mol-1 P P

CH3 Pd

O

+

k1

CO

P P

k -1

+

O

k2 P

CO

Pd

P

Pd

+

(8) CO

+ +

∆G ( -81.7 °C) = 14.8(1) kcal mol-1 P P

CH2CH3 Pd

+

O

+ CO

P

CO

P

Pd

(9) CO

+

∆G+ (-94.2 °C) = 13.4(1) kcal mol-1 O + P P

Pd

+ P P

(10)

Pd O

+

∆G+(-103.1 °C) = 12.3(1) kcal mol-1 Scheme 7.12

tion barriers remarkably lower than those of analogous dppe-palladium catalysts were obtained (I 10 kcal mol–1), which contrasts with the much lower copolymerisation activity of nickel as compared to palladium. The isolation of several fourand five-coordinate intermediates relevant to the alternating copolymerisation process suggests that nickel is much less active than palladium due to the formation of a strongly stabilised catalyst resting state relative to the transition state for the turnover-limiting step [4a]. In situ NMR analysis has also been used to determine the kinetic barriers for the migratory insertions of methyl carbonyl complexes [Pd(CO)(Me)(PPh2(CH2)nPPh2)]þ (n ¼ 2–4) relevant to propagation in ethene/CO copolymerisation. It was found that the steric bulk of the diphosphine has a significant effect on the insertion barriers with the most bulky ligand having the lowest barrier. Bianchini has reported that the migratory insertion reactions of [Pd(R)(CO)(P–P)]þ complexes (R ¼ Me, Et) are reversible and follow first-order kinetics irrespective of the chelating diphosphine (P–P ¼ dppp, dppe, meso-dppb, rac-dppb, meso-bdpp, rac-bdpp) [5e, f ]. The free energies of activation for these reactions were calculated from the half-life times (t1/2) obtained by 31P{1H} HP NMR spectroscopy as all the rates of conversion of the methyl carbonyl complexes were independent of the CO pressure. Therefore, the DG‡ values associated with the migratory insertion of the methyl carbonyl complexes could be straightforwardly calculated from the t1/2 values using the equation DG‡ ¼ RT(ln kr-ln kT/h) with kr ¼ ln 2/t1/2. First-order kinetics were also observed for the irreversible migratory insertion of [Pd(C(O)Me)(C2H4)(P–P)], yielding b-chelate complexes with a DG‡ value J 12 kcal mol–1, irrespective of the diphosphine ligand [5e, f ].


Displacement of the chelate carbonyl from palladium by ethene has never been observed in model studies, which accounts for the virtual absence of double ethene insertions in actual copolymerisation reactions. Indeed, b-chelate opening is actually brought about by CO to generate a six-membered metallacycle (g-chelate), while b-chelates of catalytically active systems generally react with CO to yield carbonyl acyl complexes, even at very low temperature. For the systems investigated by Bianchini [5e, f ], the activation barriers for the conversion of the b-chelates [Pd(CH2CH2C(O)Me)(P-P)]þ to the corresponding carbonyl acyl complexes could not be calculated straightforwardly from the t1/2 values as the reaction rates were found to be dependent on the CO pressure in the range investigated (15–25 bar) (Scheme 7.13). However, the rates of conversion of the b-chelates to the carbonyl acyl complexes were evaluated as half-life times obtained from the decay (and increase) of the phosphorus NMR resonances at appropriate temperatures. The t1/2 values and the experimental temperature at which the carbonyl acyl species begin to form were used to estimate the energy barrier to b-chelate opening by CO. The results obtained indicated that these energy barriers decrease in the ligand order: dppe (15 min at 20 hC) i rac-dppb (480 min at –40 hC) i meso-dppb (97 min at –40 hC) i meso-bdpp (200 min. at –70 hC) i dppp (84 min at –70 hC) i rac-bdpp (13 min at –70 hC). On the basis of the CO dependence of the transformation of the b-chelates into the carbonyl acyl compounds, it was proposed that the rate-limiting step in the overall conversion of the b-chelates to carbonyl acyl complexes is related to the opening of the metallacycle by CO (steps a and b in Scheme 7.13) rather than to the following migratory insertion of the alkyl carbonyl complex that is independent of the CO pressure (step c). The discovery that the keto chelates (especially the b ones) are the species controlling the strict alternation of the monomers and the intrinsic copolymerisation rate in ethene/CO propagation has stimulated much research aimed at designing catalytic systems where the keto chelates can be readily opened by CO. These studies have allowed the development of a new generation of more efficient PdII copolymerisation catalysts based on diphosphines with o-methoxyphenyl groups on the phosphorus atoms [13b, 31–34]. The beneficial effect of the o-methoxy groups seems to be both steric and electronic in nature: these groups do not seem to influence the phosphine basicity, rather

CO

Pd P

Pd P

a

O

+

P CO

+

P

O

slow b

P

CO

O

O

P

c

Pd P

+

O

+

Pd

CO

P fast

CO Scheme 7.13

291

292


P

P CO

Pd

Pd

MeO t 1/2 = 15 min at 20 °C MeO

P

P P

O

CO

P

O

P

+

O

+

P

OMe t 1/2 = 90 min at -40 °C OMe Scheme 7.14

they interact with the metal center in a hemilabile manner, so as to stabilize coordinatively unsaturated PdII complexes [34]. Recent in situ 31P{1H} NMR studies have shown that o-methoxy groups in chelating diphosphine ligands destabilise the b-chelate metalla-ring, favoring its opening by CO. Evidence of this has been obtained for the dppe and 1,2-bis[di(o-methoxyphenyl)phosphino]ethane (o-MeO-dppe) ligands (Scheme 7.14) [35].

Chain Transfer in Ethene/CO Copolymerisation NMR analysis of the polyketone end groups and in situ NMR studies have shown that two transfer mechanisms in MeOH may occur simultaneously: (a) methanolysis of Pd–acyl and (b) protonolysis of Pd–alkyl (Scheme 7.15). The eventual presence of water in MeOH, even in trace amounts, gives rise to two similar terminations, yielding different end groups (–COOH) and metal re-initiator (Pd–OH) [5e-g, 13, 36]. Termination by b-H transfer (c) is typical for reactions performed in organic solvents. 7.2.1.10

O

+

P

ROH/H2O Pd

R = Me, H +

P

ROH/H2O R = Me, H

O

S

P

OR

alcoholysis

O b)

+

P

H Pd

O

P

P

S

P

Pd P

+

Pd

a)

RO

protonolysis

P +

P

O

+ +

P

Pd P

H Pd

P S

P

P

P

+

O +

S

P Scheme 7.15

S = solvent, co-monomer

P

β-H transfer

c)


The percentage of diester, keto-ester and diketone end groups in the copolymers depends on the relative rates of alcoholysis and protonolysis. For example, when these two termination processes occur at a comparable rate, the diester, ketoester and diketone end groups are in a 1:2:1 ratio, respectively [11]. The use of oxidant co-reagents may increase the amount of ester end groups [1b], while the chain-transfer mechanism controls the nature of the termination metal product (Pd–OR, Pd–OH and Pd–H). A. Chain transfer by alcoholysis/hydrolysis

The chain transfer by alcoholysis involves the attack of ROH on a propagating Pd–acyl to give a free chain, bearing at least an ester-end group, and a Pd-hydride species, which re-initiates the chain growth by insertion of ethene (Scheme 7.15a). The alcoholysis rate decreases by either increasing the steric bulk of the alcohol or decreasing its nucleophilicity (i. e., MeOH i EtOH i iPrOH i tBuOH z 2,2,2-trifluoroethanol). Parallel to the decrease of the chain-transfer rate, the molecular weight of the copolymer increases. An effective role of water as hydrolysis agent in alcoholic media appears very unlikely as HOOC-terminated polyketone or oligoketone have never been observed. Obviously, in neat water, hydrolysis is a feasible chain-transfer path, leading to acid-terminated materials [13]. Most of the mechanistic studies of chain transfer have involved MeOH and (P–P)PdII precursors: it has been found that the methanolysis rate depends remarkably on the structure and electronic properties of the chelating diphosphine [1]. In principle, MeOH can bring about a nucleophilic attack at the Pd–acyl carbon in either intermolecular or intramolecular fashion. The intermolecular mechanism, illustrated in Scheme 7.16, does not require a precise bonding mode of the diphosphine ligand: PdII acyl complexes with either cis-chelating or trans-chelating diphosphines should equally undergo intermolecular methanolysis. In fact, a recent study by van Leeuwen, however, suggests that an intermolecular attack by MeOH on [Pd(acyl)(P–P)]þ complexes is highly unlikely. Indeed, the acyl complex [Pd(C(O)Me)(SPANphos)]þ, containing a truly trans-diphosphine ligand, was found to be indefinitely stable in MeOH, and therefore it was concluded that a cis-configuration of the phosphorus atoms in PdII acyls is required for methanolysis [37]. Experimental evidence pointing to intermolecular methanolysis, or at least, not excluding this termination path has been reported as well. An in situ NMR investigation showed in fact that the PdII acyl complexes [Pd(C(O)Me)(NCMe)(dppf)]OTs and [Pd(C(O)Me)(dppomf)]OTs, containing dihapto and trihapto ligands, respectively, undergo fast methanolysis at room temperature (dppomf ¼ 1,1’-bis(diphe+

O

P Pd P

P Pd

P S

O Me

H

H

P

+ S

O

+ MeO

P Scheme 7.16

293

294


O

P Fe

+

P MeOH

Pd C Me

P

immediate reaction at room temperature

C2H4 (1 bar)

H

+

Pd

Fe

+ CH3C(O)OMe

P

no reaction P Fe

MeOH

Pd P

P

C(O)Me

H

+

Pd

+ CH3C(O)OMe

P

+

+

P C2H4, 1 bar

Pd P

Fe

immediate reaction at room temperature

NCMe

Fe

P

+

C(O)Me

NCMe

Pd

Fe P

-20 °C

O

Scheme 7.17

nylphosphino)octamethylferrocene) (Scheme 7.17) [5g]. However, only the acyl complex with the dihapto cis-ligand dppf reacts with C2H4 to give the corresponding b-chelate. On this basis, it was proposed that ethene insertion requires a free coordination site at palladium, whereas methanolysis does not. The chain transfer by intramolecular methanolysis is illustrated in Scheme 7.18. A cis-coordinating ligand is apparently required to bind and activate MeOH so that a methoxy group is transferred to the polyketone chain and a hydride remains on palladium. Two mechanisms are possible for this reaction: (i) nucleophilic attack by the oxygen at the acyl carbonyl with concerted formation of Pd–H; (ii) formation of a Pd(acyl)(methoxy) complex and Hþ, followed by reductive elimination and subsequent proton attack on a Pd0 center. No experimental evidence favoring either mechanism in ethene/CO copolymerisation has been provided so far. O

+

P

P Pd

P

O H Me

H Pd

P P

+ S

O

+ MeO

P Scheme 7.18

B. Protonolysis

The chain transfer by protonolysis represents the predominant termination step in homogeneous ethene/CO copolymerisation, and involves the reaction between a propagating Pd–alkyl species and MeOH or adventitious water (Scheme 7.15a). As a result, the propagation is terminated with formation of a polymeric chain with a ketone-end group and Pd–OMe (or Pd–OH) species, which can re-enter the catalytic cycle by CO insertion.


+ β-H

P

P

P

P Pd O

P

Pd O

R = MeOH, H2O

P

P

CH3-CH2-C(O)

P

Pd

P

H CH3 + C2 Pd C P O P

OR + ROH

P

+

H

Scheme 7.19

By means of in situ NMR spectroscopy combined with deuterium incorporation experiments, van Leeuwen has elucidated the mechanism of termination by protonolysis, showing that the b-chelates are in equilibrium with their enolate form by a b-H elimination/hydride migration process (Scheme 7.19). The enolate intermediates are regioselectively protonated at the C2 carbon atom by either MeOH or H2O to give Pd–OMe or Pd–OH and keto terminated copolymer. The enolate formation has been reported to be rate determining in the chain transfer [19]. In anhydrous organic solvents, ethene/CO copolymerisation termination occurs exclusively by b-H transfer to give vinyl terminated polyketone and Pd–H (Scheme 7.15c). On the other hand, traces of water are very difficult to eliminate and consequently chain transfer by protonolysis is often observed, together with b-H transfer. Experimental evidence in this sense has been straightforwardly obtained by an in situ NMR study of the chemical stability of the b-chelate [Pd(CH2CH2C(O)Me)(dppe)]PF6 (7) in wet and anhydrous CD2Cl2 [5e]. Figure 7.13 reports a sequence of 31P{1H} NMR spectra taken after dissolution of the b-chelate in the wet solvent: already the first spectrum at room temperature showed the formation of the m-hydroxo binuclear complex [Pd(OH)(dppe)]2(PF6)2 (8), that was the only detectable species after 15 h. Parallel 1H NMR and GC/MS measurements showed the exclusive formation of methyl ethyl ketone. In anhydrous CD2Cl2, the b-chelate (7) underwent slow degradation with time, yielding predominantly vinyl methyl ketone as termination organic product, together with several palladium species and also Pd metal (Figure 7.14). Scheme 7.20 summarises the results of these two experiments. Upon b-H transfer from the acyl ligand, an unstable PdII–H complex is formed, which decomposes to give Pd0 and free ligand in the absence of co-monomers. The free ligand reacts with residual PdII species forming various compounds, including the bis-chelate [Pd(dppe)2]2þ [5e].

295

296


P{1H} NMR study (sapphire tube, CD2Cl2, 20 hC, 81.01 MHz) on the stability of the b-chelate 7 in wet CD2Cl2 under nitrogen: (a) after dissolving (7) in wet CD2Cl2; (b) after 1.5 h; (c) after 3.5 h; (d) after 15 h. Figure 7.13

31

Figure 7.14 31P{1H} NMR study (sapphire tube, CD2Cl2, 20 hC, 81.01 MHz) on the stability of the b-chelate (7) in CD2Cl2 under nitrogen: (a) after dissolving (7) in CD2Cl2; (b) after 1.5 h; (c) after 3.5 h; (d) after 15 h.


+ +

P

β-H

Pd P

+

P

P

H

Pd

Pd

termination by β-H transfer

O

P

P

O

C2

termination by protonolysis

O

H2O - CH3-CH2C(O)CH3

- CH2=CHC(O)CH3 P

P

+ H+ + Pd 0 P

P

H

+

H O

P Pd

Pd P

P

2+

Pd O H

P

Scheme 7.20

7.2.2

Mechanism of Styrene/CO Copolymerisation

The alternating copolymerisation of CO with styrene and related vinyl aromatics is effectively catalysed in both protic and aprotic solvents by PdII complexes bearing stereochemically rigid dinitrogen ligands such as bipy and phen [1]. Chelating diphosphines and phosphine–phosphite ligands may form active catalysts as well. However, the phosphorus ligands favor the oxidative carbonylation of vinyl aromatics to esters and diesters rather than copolymerisation, which has been attributed to the major propensity of (P–P)Pd–alkyl to undergo b-H transfer and to the higher stability of (P–P)Pd–p-benzyl complexes as compared to (N–N)Pd systems [1b]. The use of oxidant co-reagents, commonly BQ, is of mandatory importance to form active catalysts for the carbonylation of vinyl aromatics, because of the fast and unavoidable degradation of the PdII catalysts to inactive Pd0 compounds and even palladium metal. Spectroscopic evidence has been provided showing that the reaction of Pd–H with MeOH and CO to give Pd–C(O)OMe occurs through a series of equilibriums involving the initial formation of unstable Pd0 complexes that are then oxidized by BQ [1]. The general mechanistic features of the ethene/CO copolymerisation cycle (Scheme 7.2) are substantially valid also for styrene. In particular, the propagation steps are similar for both alkenes and consist of subsequent alternated migratory insertions of alkyl to CO and of acyl to olefin, with b-chelate and g-chelate resting states. The structures of the first intermediates in the syndiotactic copolymerisation of styrene derivatives with CO have been determined by an in situ NMR study using [(PriDAB)Pd(Me)(NCMe)]BAr4 as precursor (Scheme 7.21) [38]. Secondary (1,2) insertion of 4-methylstyrene into the Pd–acyl bond gave the bchelate, which transformed into the propagating g-chelate upon CO insertion. Unlike ethene, the largely predominant path to chain termination in styrene/CO copolymerisation consists of a fast b-hydrogen elimination from the last inserted

297

298


N = N

Me Me

N

N

Me

N

Me

N

Me Pd NCMe CO

O

N N

Pd

N

Me CO

N

CO

Me Pd CO -10

Me

Me O H N N

H

Pd O

CO

N

-10 °C

N

H Me

20 °C Me H

H Pd

H O

Me Scheme 7.21

styrene unit. Only in the presence of a large excess of organic oxidant may methanolysis compete with b-H transfer. Thus, irrespective of the termination step and the solvent, only Pd–H species are produced at the end of any catalytic cycle. Due to the presence of the excess oxidant, the copolymerisation of styrene does not initiate with migratory insertion PdH(styrene) because the Pd–H moieties are rapidly oxidised to Pd–OMe. Hence, the first reaction is CO insertion to give Pd–acyl. Catalysts may also be generated in situ through any of the reactions detailed in Section 7.2.1.3. Otherwise, precursors with preformed Pd–acyl or Pd–alkyl bonds can be employed [1]. The copolymerisation of CO with vinyl aromatics (and in general with a-olefins) is complicated by the prochiral nature of the substrates that may insert in either 1,2 or 2,1 fashion into the propagating g-chelate through either enantioface [1e]. Accordingly, the copolymers [CH(Ar)CH2CO]n show regio- and stereoisomerism due to the presence of truly stereogenic centers in the polymer backbone [39]. Depending on the regioselectivity of the monomer insertion into the propagating Pd–acyl, three different arrangements are possible: tail-to-tail, head-to-tail and head-to-head (Chart 7.3). Obviously, the occurrence of the same insertion mode during the chain growth leads only to head-to-tail units. In turn, the sequence of the absolute configuration of the stereogenic centres in the backbone determines the formation of atactic (stereoirregular), syndiotactic (RSRSRS sequence) or isotactic (RRRRRR or SSSSSS sequence) copolymers (Chart 7.3). stereochemistry

regiochemistry O R head to tail

O

O

R R

R

O

R

R

head to head

S Ar

tail to tail

O

Ar R O

syndiotactic

R n

Ar R Ar O

isotactic

m


299

The control of the regio- and enantioselectivity of vinyl aromatics insertion can be achieved through a suitable choice of the supporting ligands. In general, styrene insertion in (N–N)Pd–acyl takes place exclusively in the 2,1-fashion, leading to head-to-tail regioselectivity. Achiral ligands produce syndiotactic materials due to the chain end-controlled enantioface selection of the enantioface of the incoming styrene by the chiral center of the last-incorporated styrene. Chiral catalysts can switch the chain-end control to enantiomorphic site control, yielding regioregular optically active highly isotactic polyketones, the enantioface of the incoming styrene being now selected by the chiral ligand and not by the chain-end [1b, e]. An almost isotactic material has been obtained in CH2Cl2 with the palladium catalyst [Pd(Me)(NCMe)((R,S)-BINAPHOS)]BAr4 (Ar ¼ 3,5-bis(trifluoromethyl)phenyl) containing the unsymmetrical chiral bidentate phosphine-phosphite (R,S)-BINAPHOS ¼ [(R)-2-(diphenylphosphino)-(S)-1,1’-binaphthalene-2,2’-diyl phosphite] [6b, 7a, d, e, g]. An ex situ NMR study in CDCl3 under 1 bar CO elucidated the regioselectivity of styrene insertion into [Pd(C(O)Me)(NCMe)((R,S)-BINAPHOS)]þ (9) to give b-chelate products resulting from both 1,2 and 2,1 insertion. The b-chelates were converted to a Pd–p-benzyl complex via b-H elimination, followed by styrene insertion (Scheme 7.22). A more recent in situ 31P{1H} HP NMR study of styrene/CO copolymerisation by the same [(R,S)-BINAPHOS]Pd–acyl complex applying both diffusion-controlled (non-spinning sapphire NMR tube) and reaction-controlled (flow NMR cell) conditions has provided evidence for other intermediate species as well as information on the relative rates of CO insertion into isomeric Pd–acyls [6b, 7g]. Figure 7.15

NCMe +

P Pd OP

Me

(9) O

P

O

Ph

+

P

Pd OP

+

OP

Ph +

O

(10’)

(11)

P

O

P

OP

Ph

(12)

20 bar, 1h

P

P

O

OP

P = (R,S)-BINAPHOS OP

Ph

P = O Scheme 7.22

n

PPh2 OPO O

(13) P

+

bubble CO 20 bar, 3 h

+

OP Ph +

L

Pd OP (14) O

+

P Pd

Pd Ph

+

Pd

Ph bubble CO

Pd OP

P

Pd

(10)

P

O

P

(15) +

(11) + (13)

300


High pressure 31P{1H} NMR studies (U ¼ 202.5 MHz) under diffusion-controlled conditions: (a) on treatment of Pd–acetyl complex, (9), (0.02 mmol) with styrene (1.0 mmol) in 0.6 ml of CDCl3 at 23 hC; (b) after 7 h under 20 bar of CO. (From J. A. Iggo, Y. Kawashima, J. Liu, T. Hiyama, K. Nozaki, Organometallics 2003, 22, 5418.)

Figure 7.15

shows a sequence of 31P{1H} HP NMR spectra relative to the copolymerisation of CO (20 bar) and styrene at 23 hC under diffusion-controlled conditions. The b-chelates 10 and 11 were initially formed by 1,2 and 2,1 insertion of styrene in 9, respectively (Figure 7.15, trace a). Under CO, (10) or (11) gave, together with the termination p-benzyl complex (15), the poly(styrene-alt-CO)-attached b-chelates (12) and (13), respectively. At the end of the reaction (7 h), all of the initial 1,2 complex (10) was consumed, while some initial 2,1 complex (11) was still present (Figure 7.15, trace b). There

Figure 7.16 High pressure 31P{1H} NMR studies (U ¼ 81.0 MHz) under reaction-controlled conditions, where Pd–acetyl complex, (9), (0.2 mmol) was treated with styrene (10 mmol) in 5.2 ml of CDCl3 : (a) 19 hC, 1 bar of CO, t ¼ 0 min; (b) 19 hC, 10–14 bar of CO, t ¼ 55 min; (c) 23 hC, 20 bar of CO, t ¼ 98 min; (d) 23 hC, 20 bar of CO, t ¼ 127 min. (From J. A. Iggo, Y. Kawashima, J. Liu, T. Hiyama, K. Nozaki, Organometallics 2003, 22, 5418.)


was no trace of the h1-acyl complex, (14). The latter species was intercepted during the propagation only under reaction-controlled conditions (Figure 7.16). Besides confirming that chain propagation involves 1,2 insertion, the simultaneous observation of acyl and alkyl intermediates in the flow cell has shown that both intermediates are relatively long-lived species under catalytic conditions and have also comparable lifetimes in the presence of both CO and styrene [6b]. Matrix-assisted laser desorption/ionization (MALDI) mass spectrometry has contributed remarkably to unravelling the termination and initiation steps of the styrene/CO copolymerisation catalysed by the highly active bis-chelated complex [Pd(bipy)2](PF6)2 in TFE [40]. Chain-end group analysis of the material produced in the absence of BQ showed that the termination by b-H elimination is accompanied by three different initiators: two palladium alkyls from Pd–H formed by reaction of the precursor with CO and water (a and b) and a palladium carboalkoxy species formed by reaction of the precursor with the fluorinated alcohol and CO (c) (Chart 7.4). The suppression of the chain-transfer by alcoholysis was proposed to be responsible for the enhanced stability of the palladium acyl intermediates and hence for the high molecular weight of the copolymers produced.

O

O

Ph-CH=CH--C-CH-CH 2--C-CH2-CH2-Ph n Ph O

(a)

O

Ph-CH=CH--C-CH-CH 2--C-CH-CH3 n Ph Ph O

(b)

O

Ph-CH=CH--C-CH-CH 2--C-O-CH2-CF3 Ph

(c)

n

7.2.3

Mechanism of Propene/CO Copolymerisation

The catalysts for the alternating copolymerisation of propene and carbon monoxide are of the type [L2PdX2]nþ in which X represents a weakly or non-coordinating anionic ligand or a neutral ligand and L2 is a chelating bidentate phosphine. Brønsted or Lewis acids and/or oxidants are generally added to the catalytic mixtures in order to increase the activity of the catalyst [1]. The reactions are generally carried out in solvent mixtures containing at least one alcoholic component (THF/ MeOH, NO2Me/MeOH, tBuOH/MeOH/toluene, CH2Cl2/MeOH). Depending on the reaction conditions, the copolymer product can be isolated in the form of either polyketone as poly(1-oxo-2-methyltrimethylene) (Chart 7.5a) or polyspiroketal as poly[spiro-2,5-(3-methyltetrahydrofuran)] (Chart 7.5b). This latter can be transformed into the thermodynamically more stable polyketone form, either thermally or by dissolution in HFIP [1, 41, 42].

301

302


O

Me (a Me

O

Me

Me

n Me Me

O

O

O

(b O

n

The mechanism of propene/CO copolymerisation by palladium catalysis is essentially analogous to those of ethene and styrene (i. e., chain propagation proceeds via alternating insertions of CO into Pd–alkyl and alkene into Pd–acyl controlled by b-chelates) [1]. Relevant mechanistic aspects of the asymmetric copolymerisation of propene with CO have been disclosed by an NMR study of subsequent reactions of the complex [Pd(Me)(NCMe)((R,S)-BINAPHOS)]BAr4 with CO and propene (Scheme 7.23) [7a–f ]. Reaction of (9) with CO gave the Pd–acyl, (16), which under propene gives a 4:1 mixture of the alkyl palladium complexes (17) and (17’). Subsequent exposures of the mixture to CO and propene produced second- and third-generation complexes containing – [CH2CH(Me)C(O)]Me or – [CH2CH(Me)C(O)]2Me groups, respectively. The alkyl complexes (9) and (17) were found to be the predominant species under low-pressure conditions. In contrast, in analogous NMR experiments carried out under high pressure of co-monomers, both Pd-acyl and Pd-alkyl species analogous to (16) and (17), respectively, were found to constitute the two major resting states of the catalysis [7f ]. P

+ Pd

NCMe + CO

P Pd OP

Pd Me

OP

observed (9)

+

L

P

Me O

observed

OP

O

P

(16)

O

Me (17)

CO

Me +

Pd OP

L

P Pd OP

O observed

(17’) observed

L = CO, NCMe

Me

+ polyketone

O

Scheme 7.23

7.2.4

Mechanism of Cyclic Alkenes/CO Copolymerisation

The copolymerisation of CO and strained olefins such as norbornene and norbornadiene is effectively catalysed by PdII complexes with both phosphines and chelating diphosphines [43–49]. As previously mentioned, many model studies aimed at elucidating the elementary steps of alkene/CO copolymerisation have made use of cyclic alkenes in aprotic media because the products resulting from the insertion of


O Cl

Me

N

N

CO

Pd N

Cl

Me

N

O

Pd N

NCMe

Cl

Pd N

CO O

Me

Me

Cl

CO

Ar N

N

N

O

N

Pd

N

N

Me Cl

N

Pd

=

O

N O

Ar Scheme 7.24

such alkenes do not contain metal-accessible b-hydrogens and therefore are generally stable and isolable. The formation of b-chelates on the way to polyketone chain growth was proved by the isolation of such products from the reaction of [Pd(C(O)Me)(NCMe)(PPh3)2]BF4 with norbornene [50] and of cis-[Pd(C(O)Me)(NCMe)(dppe)]BF4 with cyclopentene or cycloheptene [23]. The insertions of CO into Pd–Me and of various strained alkenes into Pd–COR supported by the dinitrogen ligand Ar-BIAN have been studied by Vrieze and Elsevier by means of ex situ NMR spectroscopy [22, 51]. For example, the sequential insertions of CO and norbornadiene into Pd–alkyl and Pd–acyl bonds allowed the detection and isolation of several intermediates such as acyl(carbonyl) and bchelate (Scheme 7.24). The Pd complexes isolated after alkene insertion were found to have a structure arising from cis addition of Pd–C(O)R to the exo face of the olefin. Further studies of the insertion of CO and various norbornenes into Pd–alkyl and Pd–acyl bonds, respectively, have been reported by Boersma and van Koten for PdII complexes stabilised by bipy [52]. These authors confirmed that the cis,

O

N

Me

OTf

Pd

Me

N

CO, NaI

Pd

O

I

N

N

O AgOTf O

Me

N

O N

O

N = N

Me OTf

O

N

Pd N

N

O Pd

I

N Scheme 7.25

303

304


exo insertion of the alkene is both stereo- and chemoselective and showed the first example of reversible alkene insertion in an isolated PdII complex as well as the first isolated alkene/CO co-oligomer connected to a PdII centre (Scheme 7.25). The insertion reactions of chiral strained alkenes into Pd–acyl bonds have also been studied by Braunstein with the use of heterodinuclear Fe–Pd complexes containing either dppm or bis(diphenylphosphino)amine (dppa) bridging ligands [53]. Remarkably, the insertions were found to occur with high stereo- and regioselectivity, thus indicating that the growth of polyketones containing a chiral center is a feasible goal. 7.2.5

Mechanism of Polar Alkenes/CO Copolymerisation

The migratory insertion reactions of Pd(acyl)(alkyl acrylate) and Pd(acyl)(vinyl acetate) have been investigated by several authors and valuable mechanistic information on this important elementary step of the chain growth process has been obtained by NMR spectroscopy [1i, j]. Of particular relevance are recent NMR studies by Braunstein dealing with the stereochemistry of insertion of methyl acrylate into Pd–acyl bonds derived from the reaction of CO into Pd–Me or Pd–CH2CH2C(O)Me. The insertion of CO in the latter Pd–alkyl bond has provided original information on the mechanism and stereochemistry of alternating copolymer blocks incorporating a polar alkene (Scheme 7.26) [25]. +

O H

N

P Me

Pd

NCMe

O

Me

C2H4

H Me

N

P

O

+

CO

Pd O

O

H

- CO

Me

CO2Me

O H Me

N

+

OMe C

Pd

Scheme 7.26

Pd O

O

Me

CO2Me

P O

N

Me

+

P

O

H

H H Me

O H Me

N

P

H

Pd O

+

OMe C

H H H H Me

O H

H

O


7.2.6

Catalyst Deactivation Paths

Irrespective of the alkene and the experimental conditions, all PdII catalysts for alkene/CO copolymerisation have a short lifetime and no catalytic activity is generally observed after 8–10 h. As shown by Drent, the ethene/CO copolymerisation, initially truly homogeneous, becomes partially heterogeneous as a result of the formation of insoluble Pd–polyketone moieties [54]. Besides this phase change, the rate also slows because the number of catalytically active PdII sites decreases steadily during the reaction in consequence of several unfavorable events. In almost all reactions investigated, Pd metal separates from the solution, although in trace amount when compared to the overall concentration of Pd in the precursors. In situ NMR experiments have shown that PdII catalysts may undergo degradation or conversion into less active species by a number of paths [5]. The most dramatic deactivation involves the formation of Pd metal as a consequence of the inherent instability of the [PdH(L–L)]þ moieties as shown in Scheme 7.27 (L ¼ P, N). Therefore, irrespective of the supporting chelating ligand, the number of active species decreases with time. In the case of diphosphines, the presence of free ligand in the reaction mixture has the additional drawback of causing the formation of bis-chelates (Figures 7.9 and 7.10), which, depending on the ligand structure, may (for example dppp [5f ]) or may not (for example dppe [5e]) re-enter the catalysis cycle by reaction with CO and water (Scheme 7.27). The formation of bis-chelates with dinitrogen ligand is not a problem as they are still active catalyst precursors [55]. The formation of m-hydroxo diphosphine complexes by protonolysis of the b-keto chelates with water (Figure 7.9) is another factor that contributes to a decrease in the copolymerisation activity (Scheme 7.28) [5f ]. In general, the m-hydroxo complexes can re-enter the copolymerisation cycle by reaction with CO that breaks the binuclear structure to give Pd–H via Pd–C(O)OH [5a, 13, 36]. The contribution

2+

P L

L + H+ + Pd 0 L

+

H

Pd

Pd L

P

P

2+

P Pd

P

P - P-P

P

COOH Pd

+ HO 2

P

P

H +

P

C2H4

Pd P Scheme 7.27

CO

2+

CO

P

2+ Pd

Pd P

- CO 2

CO

P

305

306


+ P P

Pd

P

CH3-CH2-C(O)

H O

P

Pd

O

P

OH +

P

H2O Pd

O H

P

2+

P Pd P

P CO

C2H4

P

CO

H

+ - CO2

P

COOH

+

P

Pd

Pd P

P

OH

+

Pd P

CO

Scheme 7.28

of the m-hydroxo complexes to the overall productivity in polyketone is generally higher than that of the bis-chelates, and depends on the diphosphine ligand. As a general trend, the stability to carbonylation of the m-hydroxo complexes decreases with the steric rigidity of the ligand backbone [5f ]. Finally, NMR evidence has been provided for the occurrence of catalyst degradation by intramolecular phosphine oxidation of the Amatore type in alkene/CO copolymerisation catalysed in organic solvents by PdII-diphosphine catalysts containing oxygen co-ligands (for example acetate, as exemplified in a dppp complex, Scheme 7.29) [56].

Pd(OAc) 2 + dppp Ph2

OAc

P Pd P

OAc

fast

Pd(OAc) 2(dppp) Ph2

slow

+

P OAc

P

Pd0 OAc

Ph2

Ph2

+ H2O, -H+ (O)PPh2-(CH2)3-PPh2-Pd0(OAc)- + AcOH

Scheme 7.29

References [1] (a) A. Sen, Acc. Chem. Res. 1993, 26,

303; (b) E. Drent, P. H. M. Budzelaar, Chem. Rev. 1996, 96, 663; (c) E. Drent, J. A. M. van Broekhoven, P. H. M. Budzelaar in Applied Homogeneous Catalysis with Organometallics Compounds, Eds. B. Cornils, W. A. Hermann, Wiley-VCH, Weinheim 1996,

p. 333; (d) A. Sommazzi, F. Garbassi, Progr. Polym. Sci. 1997, 22, 1547; (e) K. Nozaki, T. Hijama, J. Organomet. Chem. 1999, 576, 248; (f) E. Drent, J. A. M. van Broekhoven, P. H. M. Budzelaar, in Applied Homogeneous Catalysis with Organometallics Compounds, Eds. B. Cornils,

References

[2]

[3]

[4]

[5]

W. A. Hermann, Wiley-VCH, Weinheim 2001, p. 344; (g) G. P. Belov, Russ. Chem. Bull. Int. Ed. 2002, 51, 1605; (h) R. A. M. Robertson, D. J. C. Cole-Hamilton, Coord. Chem. Rev. 2002, 225, 67; (i) C. Bianchini, A. Meli, Coord. Chem. Rev. 2002, 225, 35; (j) C. Bianchini, A. Meli, W. Oberhauser, Dalton Trans. 2003, 2627. (a) B. J. Lommerts, E. A. Klop, J. Aerts, J. Polym. Sci., Polym. Phys. B 1993, 31, 1319; (b) M. A. Del Nobile, G. Mensitieri, L. Nicolais, A. Sommazi, F. Garbassi, J. Appl. Polym. Sci. 1993, 50, 1261; (c) Y. Xu Frank, C. W. Jame, Macromolecules 1993, 26, 3485; (d) M. D. E. Forbes, Macromolecules 1994, 27, 1020; (e) C. E. Ash, J. Mater. Educ. 1994, 16, 1; (f) C. E. Ash, Int. J. Polym. Mater. 1995, 30, 1; (g) S. De Vito, F. Ciardelli, G. Ruggeri, O, Chiantore, A. Moro, Polym. Int. 1998, 45, 353; (h) A. S. Abu-Surrah, B. Rieger, Top. Catal. 1999, 7, 165. E. Drent, R. van Dijk, R. van Ginkel, B. van Oort, R. I. Pugh, Chem. Commun. 2002, 964. (a) S. Shultz, J. M. DeSimone, M. Brookhart, Organometallics 2001, 20, 16; (b) S. Shultz, J. M. DeSimone, M. Brookhart, J. Am. Chem. Soc. 2001, 123, 9172; (c) F. De Angelis, A. Sgamellotti, Organometallics 2002, 21, 2036. (a) C. Bianchini, H. M. Lee, A. Meli, S. Moneti, V. Patinec, G. Petrucci, F. Vizza, Macromolecules 1999, 32, 3859; (b) C. Bianchini, H. M. Lee, A. Meli, S. Moneti, F. Vizza, M. Fontani, P. Zanello, Macromolecules 1999, 32, 4183; (c) C. Bianchini, H. M. Lee, P. Barbaro, A. Meli, S. Moneti, F. Vizza, New. J. Chem. 1999, 23, 929; (d) C. Bianchini, G. Mantovani, A. Meli, W. Oberhauser, P. Bruggeller, T. Stampfl, J. Chem. Soc., Dalton Trans. 2001 690; (e) C. Bianchini, H. M. Lee, A. Meli, W. Oberhauser, M. Peruzzini, F. Vizza, Organometallics 2002, 21, 16; (f) C. Bianchini, A. Meli, G. Mller, W. Oberhauser, E. Passaglia, Organometallics 2002, 21, 4965; (g) C. Bianchini, A. Meli, W. Oberhauser, P. W. N. M. van

[6]

[7]

[8] [9] [10]

[11]

[12]

[13]

Leeuwen, M. A. Zuideveld, Z. Freixa, P. C. J. Kamer, A. L. Spek, O. V. Gusev, A. M. Kal’sin, Organometallics 2003, 22, 2409; (h) C. Bianchini, A. Meli, A. Traversi, Ital. Pat. FI A000025, 1997. (a) J. A. Iggo, D. Shirley, N. C. Tong, New J. Chem. 1998, 22, 1043; (b) J. A. Iggo, Y. Kawashima, J. Liu, T. Hiyama, K. Nozaki, Organometallics 2003, 22, 5418; (c) W. Clegg, G. R. Eastman, M. R. J. Elsegood, B. T. Heaton, J. A. Iggo, R. P. Tooze, R. Whyman, S. Zacchini, J. Chem. Soc., Dalton Trans. 2002, 3300. (a) K. Nozaki, N. Sato, H. Takaya, J. Am. Chem. Soc. 1995, 117, 9911; (b) K. Nozaki, N. Sato, Y. Tonomura, M. Yasutoni, H. Takaya, T. Hiyama, T. Matsubara, N. Koga, J. Am. Chem. Soc. 1997, 119, 12779; (c) K. Nozaki, M. Yasutomi, K. Nakamoto, T. Hijama, Polyhedron 1998, 17, 1159; (d) K. Nozaki, Y. Kawashima, K. Nakamoto, T. Hijama, Macromolecules 1999, 32, 5168; (e) K. Nozaki, Y. Kawashima, K. Nakamoto, T. Hijama, Polym. J. 1999, 31, 1057; (f) K. Nozaki, T. Hijama, S. Kacker, I. T. Horvath, Organometallics 2000, 19, 2031; (g) K. Nozaki, H. Komaki, Y. Kawashima, T. Hiyama, T. Matsubara, J. Am. Chem. Soc. 2001, 123, 534. L. Helm, A. E. Merbach, J. Chem. Soc., Dalton Trans. 2002, 633. H.-K. Luo, Y. Kou, X.-W. Wang, D.-G. Li, J. Mol. Catal. 2000, 151, 91. W. P. Mul, H. Oosterbeck, G. A. Betel, G.-J. Kramer, E. Drent, Angew. Chem. Int. Ed. Engl. 2000, 39, 1848. (a) E. Drent, Eur. Pat. 121 965, 1984; (b) E. Drent., J. A. M. van Broekhoven, M. Doyle, J. Organomet. Chem. 1991, 417, 235. (a) J.-T. Chen, A. Sen, J. Am. Chem. Soc. 1984, 106, 1506; (b) A. Sen, J.-T. Chen, W. M. Vetter, R. Whittle, J. Am. Chem. Soc. 1987, 109, 148. (a) G. Verspui, G. Papadogianakis, R. A. Sheldon, J. Chem. Soc., Chem. Commun. 1998, 401; (b) G. Verspui, F. Schanssema, R. A. Sheldon, Angew. Chem. Int. Ed. Engl. 2000, 39, 804; (c) G. Verspui, F. Schanssema,

307

308


[14] [15] [16] [17] [18]

[19]

[20]

[21]

[22]

[23]

[24]

[25]

[26] [27]

[28]

[29] [30]

[31]

R. A. Sheldon, Appl. Catal. 2000, 198, 5. A. Marson, A. B. van Oort, W. P. Mul, Eur. J. Inorg. Chem. 2002, 3028. C. Bianchini, A. Meli, W. Oberhauser, Organometallics 2003, 22, 4281 H. E. Bryndza, W. Tam, Chem. Rev. 1988, 88, 1163. Y. Koide, S. G. Bott, A. R. Barron, Organometallics 1996, 15, 2213. S. Shultz, J. Ledford, J. M. DeSimone, M. Brookhart, J. Am. Chem. Soc. 2000, 122, 6351. M. A. Zuideveld, P. C. J. Kamer, P. W. N. M. van Leeuwen, P. A. A. Klusener, H. A. Stil, C. F. Roobeek, J. Am. Chem. Soc. 1998, 120, 7977. J. Schwarz, E. Herdtweck, W. A. Herrmann, Organometallics 2000, 19, 3154. G. K. Barlow, J. D. Boyle, N. A. Cooley, T. Ghaffar, D. F. Wass, Organometallics 2000, 19, 1470. R. Van Asselt, E. G. C. Gielens, E. R. Rulke, K. Vrieze, C. J. Elsevier, J. Am. Chem. Soc. 1994, 116, 977. F. Ozawa, T. Hayashi, H. Koide, A. Yamamoto, J. Chem. Soc., Chem. Commun. 1991, 1469. W. N. M. van Leeuwen, C. F. Roobeck, H. van der Heijden, J. Am. Chem. Soc. 1994, 116, 12117. P. Braunstein, C. Frison, X. Morise, Angew. Chem. Int. Ed. Engl. 2000, 39, 2867. C. Rix, M. Brookhart, P. S. White, J. Am. Chem. Soc. 1996, 118, 4746. B. A. Markiens, D. Kruis, M. H. P. Rietveld, K. A. N. Verkerk, J. Boersma, H. Kooijman, M. T. Lakin, A. L. Spek, G. van Koten, J. Am. Chem. Soc. 1995, 117, 5263. S. Shultz, J. Ledford, J. M. DeSimone, M. Brookhart, J. Am. Chem. Soc. 2000, 122, 6351. M. Svensson, T. Matsubara, K. Morokuma, Organometallics 1996, 15, 5568. (a) P. Margl, T. Ziegler, J. Am. Chem. Soc. 1996, 118, 7337; (b) P. Margl, T. Ziegler, Organometallics 1996, 15, 5519. T. M. Shryne, H. V. Holler, US Pat. 3984388, 1976.

[32] (a) E. Drent, M. C. T. De Kock,

[33] [34] [35] [36] [37]

[38]

[39]

[40]

[41]

[42] [43] [44]

[45] [46]

[47]

[48] [49]

US Pat. 5688909, 1997; (b) E. Drent, US Pat. 4835250, 1989. E. Drent, L. Wife, Eur. Pat. Appl. B222454, 1987. K. R. Dumbar, J.-S. Sun, J. Chem. Soc., Chem. Commun. 1994, 2387. C. Bianchini, A. Meli, W. Oberhauser, manuscript in preparation. Z. Jiang, A. Sen, Macromolecules 1994, 27, 7215. P. W. N. M. van Leeuwen, M. A. Zuideveld, B. H. G. Swennenhuis, Z. Freixa, P. C. J. Kamer, K. Goubitz, J. Fraanje, M. Lutz, A. L. Spek, J. Am. Chem. Soc. 2003, 125, 5523. (a) C. Carfagna, M. Formica, G. Gatti, A. Musco, A. Pierleoni, Chem. Commun. 1998, 1113; (b) B. Binotti, C. Carfagna, G. Gatti, D. Martini, L. Mosca, C. Pettinari, Organometallics 2003, 22, 1115. S. Bruckner, C. De Rosa, P. Corradini, W. Porzio, A. Musco, Macromolecules 1996, 29, 1535. B. Milani, G. Corso, G. Mestroni, C. Carfagna, M. Formica, R. Seraglia, Organometallics 2000, 19, 3435. (a) A. Batistini, G. Consiglio, Organometallics 1992, 11, 1766; (b) S. Bronco, G. Consiglio, R. Utter, A. Batistini, U. W. Suter, Macromolecules 1994, 27, 4436. Z. Jiang, A. Sen, J. Am. Chem. Soc. 1995, 117, 4455. S. Kacker, A. Sen, J. Am. Chem. Soc. 1995, 117, 10591. J. Tsuji, J. Hosaka, J. Polym. Sci. 1965, 3, 703; M. Graziani, G. Carturan, U. Bellico, Chim. Ind. (Milan) 1971, 53, 939. A. Sen, W. Lai, J. Am. Chem. Soc. 1982, 104, 3520. T. Kawaguchi, M. Kanno, T. Yanagihara, Y. Inoue, J. Mol. Catal. A 1999, 143, 253. D. Roberto, M. Castellani, G. P. Chiusoli, Tetrahedron. Lett. 1988, 29, 2115. A. L. Safir, B. M. Novak, J. Am. Chem. Soc. 1998, 120, 643. D.-J. Liaw, J. Polym. Sci. 1993, 31, 309.

References [50] J. S. Brumbaugh, R. R. Whittle,

[53] P. Braunstein, J. Cossy, M. Knorr,

M. Parvez, A. Sen, Organometallics 1990, 9, 1735. [51] J. H. Groen, M. J. M. Vlaar, P. W. N. M. van Leeuwen, K. Vrieze, H. Kooijman, A. L. Spek, J. Organomet. Chem. 1998, 551, 67. [52] B. A. Markies, D. Kruis, M. H. P. Rietveld, K. A. N. Ververk, J. Boersma, H. Kooijman, M. T. Lakin, A. L. Spek, G. van Koten, J. Am. Chem. Soc. 1995, 117, 5263.

C. Strohmann, P. Vogel, New. J. Chem. 1999, 2, 1215. [54] W. P. Mul, E. Drent, P. J. Jansens, A. H. Kramer, M. H. W. Sonnemans, J. Am. Chem. Soc. 2001, 123, 5350. [55] B. Milani, L. Vicentini, A. Sommazzi, F. Garbassi, E. Chiarparin, E. Zangrando, G. Mestroni, J. Chem. Soc., Dalton Trans.1996, 3139. [56] C. Amatore, A. Jutand, A. Thuilliez, Organometallics 2001, 20, 3241.

309

8 The Use of Spectroscopy in Metallocene-based Polymerisation Catalysis Manfred Bochmann

8.1

Introduction

Some 50 years after the discovery of alkene polymerisation catalysts based on titanium halide/aluminium alkyl mixtures by Ziegler and Natta [1–3], polyolefin materials have developed into a wide spectrum of materials and continue to witness impressive production figures and growth rates. In 2001, an estimated 79.6 million tons of polyethylene and polypropylene were produced worldwide, ca. 55 % of total plastics production. Most polymerisation processes continue to employ heterogeneous catalysts; indeed, for the production of highly stereo-regular isotactic polypropylene the latest generation of Ti/MgCl2 catalyst is extremely active. Similarly, for rigid high molecular weight polyethylene, Cr/SiO2 catalysts are excellent and unlikely to be replaced in the foreseeable future. However, whereas such heterogeneous catalysts are highly effective for the production of specific products, metallocenes are gaining ground, mainly because of their flexibility: the properties of polymers obtained with metallocenes can easily be tailored, from rigid to elastomeric and stretchable, with adjustable melting points and varying amounts of co-monomer incorporation. This gives rise to a much wider spectrum of homo- and copolymers than was hitherto possible. Suitable tailoring of metallocene ligand structures and new methods of catalyst activation are capable of generating catalysts with unprecedentedly high activities and stereoselectivities. In particular, their ability to incorporate co-monomers evenly along the polymer chain and the uniformity of their active sites cannot be matched by heterogeneous catalysts. It was also quickly recognised that these soluble, diamagnetic systems had the potential of providing detailed mechanistic information about fundamental steps of the polymerisation process. Most of these studies involve NMR spectroscopy. Group 4 metallocene complexes were reported by Wilkinson in early 1953 [4], a few months before Ziegler’s seminal discovery that mixtures of TiCl4 and AlEt3 catalysed the polymerisation of ethene. It was not long before the new compounds were tested as potential ethene polymerisation catalysts, not least because these Mechanisms in Homogeneous Catalysis. A Spectroscopic Approach. Edited by Brian Heaton Copyright c 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN 3-527-31025-8

312

8 The Use of Spectroscopy in Metallocene-based Polymerisation Catalysis

homogeneous catalysts were more amenable to mechanistic investigation and were likely to throw some light on the processes operative in Ziegler catalysts [5, 6]. Studies by Breslow and Newburg soon realized some key mechanistic features, such as: 1. The need for Lewis acidic main group alkyls as activators, which can form complexes with the inactive catalyst precursor and are capable of transferring an alkyl ligand [7]; 2. oxidation state IV for the active species and reduction as a deactivating side reaction [7, 8]; 3. polymer chain growth by monomer insertion into a Ti–alkyl bond [8]; 4. the equilibrium nature of the reactions leading to the active species; 5. reversible alkyl ligand transfer to aluminum, and 6. an intermittent chain growth behavior, with an equilibrium between active and dormant states [9]. These initial investigations also included kinetic studies on catalyst initiation, polymer chain propagation and termination rates and the time dependence of the number-average molecular weight, using isotopic labelling with (14CH3)2AlCl [10].

d+

Cl + AlClEt 2

Cp2Ti

Et

Cp2Ti

Cl

Cl

d-

AlCl 2Et

Et

d+

d-

Cp2Ti

AlCl 2Et Cl

C4H9 Cp2Ti

Scheme 8.1

Cl

AlCl 2Et

Breslow suggested a mechanistic scheme (Scheme 8.1) involving a zwitterionic metal alkyl species, with a tetrahedral geometry of the titanium center. However, it was not possible at the time to provide direct structural evidence for these species. Alternative suggestions, e. g. by Oliv et al. [11], included octahedral intermediates (Scheme 8.2), while an early proposal by Shilov et al. [12], based on electrochemical studies in dichloromethane, that the active species was likely to be

R

Cl Cp2Ti

Cl R

Scheme 8.2

AlRCl 2

Ti

Cl R

C2H4R AlCl 2

Cp2Ti Cl

AlCl 2R

8.2 Identification of the Catalytically Active Species: The Chemistry of Group 4 Metal Methyl Species Cl Cp2Ti Cl

Scheme 8.3

AlClR 2

R Cp2Ti Cl

AlCl 2R

Cp2Ti R

+

AlRCl 3-

polymerisation

cationic (Scheme 8.3) received little attention until the advent of well-defined ionic polymerisation catalysts in the mid-1980s [13–16]. The activities of these early metallocene systems were far lower than those of industrial heterogeneous catalysts, and they were prone to deactivation by reduction. This changed with the discovery by Sinn and Kaminsky that methylaluminoxane (MAO) gave high-activity metallocene catalysts [17]. Methylaluminoxane is the product of partial hydrolysis of trimethylaluminum, an amorphous glassy substance of approximate composition [AlMe1.3–1.5O0.8–0.75]n and a molecular weight typically of 900–1200. The structure is unknown but consists most probably of Al–O chains, rings and cages of varying sizes. It usually contains some free Al2Me6; most commercially available MAO grades contain substantial amounts of this (ca. 30 wt. %). MAO is a highly effective activator and scavenger and is widely employed in industry, although its ill-defined nature allows little scope for spectroscopic and mechanistic studies. Finally, Brintzinger’s synthesis of stereo-rigid ansa-metallocenes [18, 19] provided the basis for modern high-activity metallocene catalysts and greatly increased their economic potential. However, although MAO activation and stereo-rigid structures of the new metallocene systems were extremely successful, the complexity of these catalytic systems was such that precise information about the structure of the active species remained elusive. This brief review attempts to summarise some recent advances in the mechanistic understanding of metallocene polymerisation catalysts and the role of NMR spectroscopy in these endeavors. For further information the reader is referred to a series of excellent recent reviews covering various aspects of the chemistry of metallocene polymerisation catalysts, for example Refs. [20–28].

8.2

Identification of the Catalytically Active Species: The Chemistry of Group 4 Metal Methyl Species

The species shown in Schemes 8.1 and 8.2 do not contain vacant coordination sites suitable for binding weakly donating ligands such as alkenes. Even in Breslow’s zwitterionic intermediate (Scheme 8.1) the nature of the metal–ethene interaction is unclear; alkenes do not bind to the LUMO of 16-electron complexes Cp2MCl2 (M ¼ Ti, Zr, Hf) or their alkyl derivatives. The isolation by Eisch in 1985 of a cationic titanium vinyl complex [Cp2TiC(Ph)¼C(Me)SiMe3]þ, apparently formed by insertion of an alkyne into a putative [Cp2TiMe]þ intermediate [29], raised the

313

314

8 The Use of Spectroscopy in Metallocene-based Polymerisation Catalysis Me

R

Lewis acid or cation-generating agent

Cp2M

Me

Me Cp2M

Cp2M

R

Me R

R

R Me

Cp2M

Scheme 8.4

polymer chain growth R

possibility of the existence of a reactive 14-electron species [Cp2MR]þ. Such a species would certainly have a vacant acceptor orbital of suitable symmetry for the coordination of an alkene. The polymer chain growth sequence could then be envisaged as shown in Scheme 8.4. It is now well recognised that the active species is a cationic complex, or more precisely a solvent-separated or tight ion pair, the structure of which depends on the mode of catalyst activation. Early spectroscopic and synthetic studies on metallocene dimethyl precursors helped to outline the principal reaction pathways, these have been reviewed [16, 21, 23]. Some of this chemistry is briefly summarised here since it presents the background for the understanding of later studies on methylaluminoxane (MAO) systems. The first examples of “base-free” metallocenium cations, i. e. cationic species not stabilised by N- or O-donor ligands, were prepared by protolysis of L2MMe2 with [HNMe2Ph][BPh4] in non-coordinating solvents at low temperature (M ¼ Ti, Zr; L ¼ indenyl, Cp); the counteranion was the poorly coordinating (for that time) [BPh4]– anion [30]. The reaction is conveniently monitored by low-temperature NMR spectroscopy. At –40 hC in CD2Cl2 the [Ind2TiMe]þ cation is probably solvated and characterised by Ti–Me resonances at d –0.15 (1H) and 70.8 (13C). The catalytic activity of these species was demonstrated by the injection of ethene into the NMR samples. There was an immediate formation of polyethylene, even though the Ti–Me did not decrease perceptibly, indicative of slow initiation followed by fast chain propagation. This behavior was in agreement with earlier kinetic results which suggested that the rate of ethene insertion into a Ti–Me bond was i100 times slower than into a Ti–Et bond [31]. These cations also polymerised propene. More detailed VT-NMR examinations showed that the reaction proceeded in stages, with initial formation of a methyl-bridged binuclear species [(L2TiMe)2(m-Me)]þ (L ¼ Cp, Ind), characterised by a high-field m-CH3 resonance, which reacts more slowly with further cation-generating agents to give the mononuclear cation (Scheme 8.5) [32]. The formation of methyl-bridged binuclear cations of this Me [HNMe2Ph]BPh4

L2Ti Me

[CPh3][B(C6F5)4]

L2Ti

R

+ polymerisation

(solv) Me

Me L2Ti

Me

TiL2

+ [CPh3][B(C6F5)4]

Scheme 8.5

8.2 Identification of the Catalytically Active Species: The Chemistry of Group 4 Metal Methyl Species

315

type turned out to be a general feature of the chemistry of cationic Group 4 metal methyl compounds. Whether or not such electrophilic organometallic species can be identified, or indeed isolated, depends primarily on the stability of the counteranion. The perfluorophenyl boron compounds B(C6F5)3 and [B(C6F5)4]–, first prepared by Stone and co-workers in 1963 [33], proved particularly useful in this respect. Their use in metallocene polymerisation catalysis [34, 35] led to significantly more active catalysts and “well-defined” catalyst systems that proved mechanistically informative. These results have then enabled similar species to be detected in the more complex MAO-activated catalyst systems (vide infra). The strong Lewis acid B(C6F5)3 reacts with L2ZrMe2 under methide abstraction, to give zwitterionic products of molecular structure, L2ZrMe(m-Me)B(C6F5)3 ((B), Scheme 8.6). According to X-ray crystal structures on isolated examples, (B) is stabilised by agostic interactions between zirconium and the methylborate [35a, 36, 37]. The difference in methyl binding is reflected in the Zr–C bond lengths of ca. 2.25 and 2.55 A to the terminal and bridging methyl, respectively. The spectra of zwitterions of type B are characterised by quadrupolar broadening of the CH3B resonances and low-field Zr–13CH3 signals; the latter show normal JC–H coupling constants (e. g. in Cp2ZrMe(m-Me)B(C6F5)3, JCH ¼ 122.6 Hz [35a]). By contrast, the bridging methyl in cations of type A is trigonal-bipyramidal [38] and shows a much larger C–H coupling constant; e. g. in [(Cp2ZrMe)2(m-Me)]þ, JCH ¼ 135.5 Hz [39]. The methyl group of the free [MeB(C6F5)3]– anion tends to give NMR signals at d z 0.5 (1H) and d z 10 (13C), whereas the methyl-bridged anion is found at about d z 1 (1H) and d z 15 (13C). However, 1H and 13C data do not necessarily give unambiguous information about the bonding mode of [MeB(C6F5)3]–. Better indicators are the chemical shift differences of the p-F and m-F 19F NMR resonances, with DdF ¼ D(dp-F) – (dm-F) z 3–6 ppm being typical of anion coordination to the metal, while DdF I3 ppm suggests a free anion [40].

B(C6F5)3

Me

Me

L

Cp2Zr

Zr Me

L

H H Me

L

CPh3+B(C6F5)4-

Me

Cp2ZrMe2

H

Zr H Me

B(C6F5)4 A2

L

C H

dB(C6F5)3

H B

A1

L

C

Zr L

H

d+ Me H Zr

L

[MeB(C6F5)3] fast

L

Zr

C H

L

B(C6F5)3

L

CPh3+B(C6F5)4-

L

Me Zr

slow L

B(C6F5)4 C

Scheme 8.6

316


Although zwitterions of structure (B) are typical for Group 4 elements, alternative bonding modes of the [MeB(C6F5)3]– anion have been found. For example, in L2Y(Me)B(C6F5)3 (1, L ¼ Cp, C5H4SiMe3) the metal is primarily coordinated to one of the ortho-F atoms. In this case, 1H, 13C and 19F NMR spectra are comparatively uninformative; the methyl group gives rise to broadened singlets at d ca. 1 (1H) and d 14–15 (13C), and even at –80 hC the 19F spectrum shows three equivalent C6F5 substituents; the data are in agreement with a tight ion pair structure in toluene or CD2Cl2. However, the crystal structure revealed that, at least in the solid state, the anion is coordinated via a relatively short Y–F contact (ca. 2.37 A), with additional agostic interactions to two of the three methyl-hydrogen atoms, while the distance to carbon is substantially longer (2.85 A) than in the analogous zirconium compound (B) [41]. Me3Si

H

H C H

Y

C6F5 B C6F5

F Me3Si

F F

1 F

F

Unlike the titanocene cations shown in Scheme 8.5, zirconocenes and hafnocenes react with [CPh3][B(C6F5)4] to give relatively stable and isolable homobinuclear cations [(L2MMe)2(m-Me)]þ (A2) [39]. In the case of ansa-zirconocenes, e. g. L2 ¼ rac-Me2Si(Ind)2 ¼ SBI), two diastereomers are observed (Scheme 8.7). For the major isomer in CD2Cl2 at –60 hC, the bridging methyl group is found at typically low field, d –2.75, compared to d –0.89 for the terminal Zr–Me. Here, too, the trigonal-bipyramidal environment of the bridging C atom is indicated by a large C–H coupling constant, 136 Hz [39]. The formation of (A2) can be regarded as instantaneous and complete within a fraction of a second, even at low temperatures [39], while the generation of (C) with an excess of [CPh3][B(C6F5)4] is comparatively slow, depending on the ligand sys-

Me

Scheme 8.7

Me

Me

Me

Me

Me

(R, S)

(R, R)

8.2 Identification of the Catalytically Active Species: The Chemistry of Group 4 Metal Methyl Species

Me Cp2Zr

CPh3+B(C6F5)4-

Me

L

Scheme 8.8

L

Me D

Me

Me

- AlMe3 Zr

B(C6F5)4

Al

Zr Me

L

Me

317

+ AlMe3

B(C6F5)4

L C

tem. With L2 ¼ SBI in benzene-d6 at 298 K, the conversion of (A2) into (C) under pseudo-first order conditions proceeded with a rate k ¼ 3 q 10 –4 s–1 [42]. As the formation of adducts with neutral zirconocene dimethyls shows, the 14electron species [L2MMe]þ will react with whatever is a suitably strong donor. In the presence of sterically unhindered aluminium alkyls such as Al2Me6, heterobinuclear adducts (D) result (Scheme 8.8) [39]. These form isolable oils or crystalline solids and are more stable in solution than cations (A), although so far they have resisted crystallographic characterisation. The NMR data indicate that the bonding mode of the methyl ligands in (D) is similar to that in Al2Me6 itself; thus in [Cp2Zr(m-Me)2AlMe2]þ, the resonances for the terminal Al–Me and the m-CH3 signals show very similar JC–H coupling constants of 115 and 114 Hz. The identification of these species is of obvious relevance to the catalyst speciation in systems activated by MAO, which typically contains substantial amounts of Al2Me6 (vide infra). Cations of type (A) and (D) are catalytically highly active, even though a neutral metal alkyl (i. e. L2ZrMe2 and AlMe3, respectively) occupies the coordination site necessary for alkene binding. One has to assume, therefore, that both these species are part of dissociation equilibria with the “naked” ion pair (C). If so, the addition of incremental amounts of AlMe3 should have a negative effect on catalyst activity; this was indeed the case. Dissociation becomes more favorable with increasing bulkiness of AlR3; for example, it has been shown that AlEt3 adducts of [Cp2HfEt]þ dissociate more readily and are consequently catalytically more active than AlMe3 complexes [43]. It is questionable whether even bulkier aluminum alkyls form similar adducts. The importance of solution equilibria was further demonstrated by NMR studies on the system Cp2ZrMe2/B(C6F5)3, where it could be shown that in the presence of excess Cp2ZrMe2 the methyl-bridged binuclear species (A1) dominates [44]. The structural and reaction chemistry outlined for metallocenes cannot necessarily be transferred to half-sandwich complexes. In these more open systems solvent coordination becomes more important and may even outweigh anion coordination. For example, while Cp*TiMe3 reacts with B(C6F5)3 in toluene to give the (thermally unstable) methyl-bridged product 2, the zirconium and hafnium analogues form ionic toluene complexes of type 3 [45, 46]. Arene coordination is indicated by low-field shift of the signals for coordinated toluene; with the Hf compound 3, there does not seem to be an exchange with free toluene in solution. The crystal structures of the hafnium C5H3(SiMe3)2 and Cp* compounds confirm h6 -toluene coordination and the absence of any close contacts with the anion. Toluene coordination is thought to be responsible for the inferior polymerisation activity of 3

318


compared to 2 [47, 48]. A similar arene complex is formed in the case of the “constrained-geometry” zirconium complex 4 after activation of CGCZrMe2 with [CPh3][B(C6F5)4] [49].

Arene coordination has been exploited for the synthesis of bridged sandwich complexes of types 5 [50] and 6 (Scheme 8.9). Arene coordination is easily detected by the low-field 1H shifts of the m- and p-phenyl signals, to d 8.4–8.7. Arene complexation reduces the catalytic activity of these complexes and also reduces the polymer molecular weight [50b]. Although in the case of zirconocenes there is no evidence of significant coordination of aromatic solvents, pendant benzyl substituents tethered to Cp ligands in [(C5H4CH2Ph)2ZrMe][MeB(C6F5)3] bind to the metal sufficiently strongly to displace the anion from the coordination sphere and give an equilibrium between isomers with free [d(1H) 0.48, d(13C) 10] and coordinated [d(1H) 0.20, d(13C) 24] [MeB(C6F5)3]– anion [51a]. Methyl-substituted benzyl substituents, as in 7, lead to stronger coordination [51b, c]. Similar titanium Cp-arene sandwich complexes 8 catalyse the selective trimerisation of ethene to 1-hexene (Scheme 8.10) [52].

Scheme 8.9

8.3 Activation of Alternative Group 4 Catalyst Precursors with B(C6F5)3

319

R' R2C

Ti

[Ti]

[Ti]

H Rn

8

R' = H, SiMe3, CMe3, CMe2Ph

Scheme 8.10

8.3

Activation of Alternative Group 4 Catalyst Precursors with B(C6F5)3

Apart from the formation of compounds of type (B), several other modes of activating metallocene complexes with B(C6F5)3 have been demonstrated. These avoid the complications of methyl-bridged binuclear products discussed above. For example, B(C6F5)3 reacts with 1,3-butadiene complexes by attacking a terminal diene-CH2 moiety. If the diene is present in an s-trans conformation, as in complex 9 (Scheme 8.11), a zwitterionic zirconocene h3 -allyl complex of type 10 results which is stabilised by a Zr p p p F interaction. The complex is fluxional and undergoes rapid h3 – h1– h3 allyl interchange; from the coalescence point at 40 hC, DG‡ ¼ 19.8 kcal mol–1. There is also rapid exchange between the three B–C6F5 substituents. The Zr p p p o-F coordination can be detected by the 19F high-field shift of the unique o-F atom at low temperature: while five of the o-F signals are found in the expected region (d–126 to –137), the sixth o-F signal is found at d –213. From the temperature dependent NMR spectrum, a Zr–F dissociation energy of 8.1e0.4 kcal mol–1 was estimated [53]. Complexes of type 10 have proved useful

B(C6F5)3

R

C6F5 Cp2Zr

Cp2Zr

B C6F5

F 9

F

B(C6F5)3 Cp2Zr R

F F

F 10 R

12

Cp2Zr

L1

L1 Zr

B(C6F5)3

H

H

B(C6F5)3

Zr

L2

B(C6F5)3 13

Scheme 8.11

L2

14

11

320


F

CpR

CpR

H

B(C6F5)3

Zr

H

(C6F5)3B

F

F Zr

CpR

F

F

C6F5

B

-

Zr

C6F5 15

17

16

R

Cp = C5H5, C5H4Me, C5H4SiMe3

toluene, 50 oC

Et2O

- C6F5H - C6F5H

Me3Si SiMe3

Zr B

OEt2 F F F

F F Scheme 8.12

18

F

F F

CpR

C6F5 F F

CpR

F

Zr C6F5

Zr

F

F

F F

B B C6F5 19

for studying the energetics of monomer insertion, since at low temperature they react with ethene or propene to give a mono-insertion product 12, probably via a s-allyl alkene complex 11. Alkene coordination takes place in a pre-equilibrium step, followed by rate-limiting alkene insertion, which tends to be about 2 kcal mol–1 higher than the dissociation activation energy [54]. If the coordinated diene exists in the s-cis conformation as in 13, B(C6F5)3 addition leads to a zwitterionic structure with almost linear Zr–C–B arrangements, stabilised by agostic Zr p p p HC interactions but without Zr p p p F coordination, as in 14 (Scheme 8.11) [53]. A similar reaction sequence with allylzirconium diene complexes of type 15 has resulted in the surprisingly facile transfer of the C6F5 substituents from boron to zirconium, combined with C–H activations to give 17 (Scheme 8.12) [55]. Further C6F5 transfer reactions can lead to catalytically active borole complexes 18, or inactive borole-bridged tripledeckers 19 [56]. Such reactions provide information about possible catalyst deactivation pathways in B(C6F5)3 -activated zirconium half-sandwich complexes. As was seen for 10, the decomposition products are stabilised by Zr–F coordination, indicated by 19F NMR high-field shifts, although in these cases fluorine bonding to the metal center is significantly stronger; in the tripledecker 19 it persists even at elevated temperature (DG‡ ¼ 13.2 kcal mol–1 at 68 hC). Further examples of zwitterionic products are formed by B(C6F5)3 attack on “tuck-in” complexes to give 20 [57–59] and 21 (Scheme 8.13); the latter is a single-component polymerisation catalyst which deactivates under rearrangement to an allyl complex [60]. Attack by B(C6F5)3 on the coordinated ethene in zirconium(II) complexes gives zwitterions of type 22 which act as single-component polymerisation catalysts [58].

C6F5

8.3 Activation of Alternative Group 4 Catalyst Precursors with B(C6F5)3 d- B(C F ) 6 5 3

B(C6F5)3 Zr

+

Zr d

Ph

Ph

Cp*

Cp*

20

dB(C6F5)3

+

d Zr

CH2CMe3

Cp*

+

d Zr

-

Me

dB(C6F5)3

Cp*

21

L

d+ Cp2Zr

B(C6F5)3

Cp2Zr

Scheme 8.13

L

H H dB(C6F5)3

22

Metal benzyl complexes are frequently convenient catalyst precursors. Because the benzyl ligand can adopt a variety of coordination modes to satisfy the electron demand of the metal center, binuclear alkyl-bridged species are not usually found. For example, the reaction of zirconocenes L2Zr(CH2Ph)2 with [CPh3][B(C6F5)4] leads to cationic complexes 23 (Scheme 8.14). The h2 -coordination of the benzyl ligand is readily recognised by the low-field shift of the CH2 resonance (1H). In C2 -symmetric complexes such as L2 ¼ rac-C2H4(Ind)2 where the benzyl-CH2 group is diastereotopic, this is accompanied by a reduction of the geminal 2JHH coupling constant from 11.3 Hz to 7.8 Hz. In the 13C NMR spectra, the change to h2 -coordination is reflected in a high-field shift of the ipso-C atom by almost 30 ppm (from d 150–153 to d z124), coupled with an increase in JCH of the benzylic CH2 signal to ca. 130–135 Hz as the M–C–C(phenyl) angle becomes more acute. Such cationic

CH2Ph

[CPh3][B(C6F5)4]

B(C6F5)4

L2Zr

L2Zr CH2Ph

23

B(C6F5)3 Zr(CH2Ph)4

Scheme 8.14

B(C6F5)3

Ph Zr Ph

24

321

322


benzyl complexes are excellent single-component catalysts for ethene and propene polymerisations [37, 61]. Since tri- and tetrabenzyl complexes such as CpZr(CH2Ph)3 and Zr(CH2Ph)4 are comparatively stable and readily accessible, their activation, particularly with B(C6F5)3, is now widely employed. The phenyl ring of the resulting [PhCH2B(C6F5)3]– anion is frequently found coordinated to the metal center, as in 24 [62] and 25 [63], indicated by high-field 1H NMR chemical shifts of the phenyl ring, although there may be an equilibrium between free and coordinated anions [64], and anion coordination may be prevented by steric hindrance, e. g. 26 [65]. In mixed-ligand systems such as CpRM(1,3-diene)(CH2Ph) (M ¼ Zr, Hf), B(C6F5)3 attacks the benzyl ligand preferentially, as in 27 [66].

Reactions of B(C6F5)3 with higher metal alkyls have only rarely been used to generate cationic metallocene catalysts. Given the choice, the borane reacts selectively with methyl. Thus, (C5R5)2Zr(Me)CH2CMe3 (R ¼ H, Me) and B(C6F5)3 in toluene or bromobenzene proceed with (reversible) isobutene elimination to give 28; a Zr-neopentyl cation could not be found (Scheme 8.15) [67]. On the other hand, abstraction of bulky alkyls is possible; for example, Cp*Ti(CH2SiMe3)2 and B(C6F5)4 give spectroscopically identified [Cp*Ti(CH2SiMe3)3][Me3SiCH2B(C6F5)3], a highly active catalyst for the syndiospecific polymerisation of styrene [68]. The mixed alkyls Cp’2Zr(Me)(R) (Cp’ ¼ 1,2-C5H3Me2) react with B(C6F5)3 to give exclusively Cp’2Zr(R)(m-Me)B(C6F5)3; variable temperature NMR spectroscopy showed that in these compounds the mobility of the anion increases sharply with increasing bulk of R ¼ Me I CH2SiMe3 II CH(SiMe3)2. Again, at i–15 hC the analogous CH2CMe3 complex was prone to b-Me elimination, and VT-NMR studies allowed the determination of the activation parameters for this process [69]. Similarly, the reaction of (SBI)Zr(Me)CH2SiMe3 with B(C6F5)3 or [CPh3][B(C6F5)4] is completely selective for methyl abstraction; toluene solutions of the resulting [(SBI)ZrCH2SiMe3][B(C6F5)4] salts are stable for days at room temperature and

Me

B(C6F5)3, -

L2Zr +

Scheme 8.15

Me L2Zr

0 oC

Me B(C6F5)3 28

8.4 Olefin Coordination to d0 Metal Centers H

H Cp'2Zr H

[CPh3][B(C6F5)4]

ZrCp'2

- CHPh3

H Me H

H

ZrCp'2

[B(C6F5)4]

H H 29

H Me

ZrCp'2

Cp'2Zr

H Cp'2Zr

X Cp'2Zr

H 30

31

Scheme 8.16

are particularly suitable for studying site epimerisation and anion exchange processes (vide infra) [70]. Metallocene hydride complexes have been little explored as precursors for alkene polymerisation catalysts, although the insertion of alkenes into M–H bonds is facile. The complex [Cp’2ZrH(m-H)]2 (Cp’ ¼ C5H4SiMe3) reacts with [CPh3][B(C6F5)4] to give [Cp’4Zr2H(m-H)2][B(C6F5)4] 29 (Scheme 8.16). The compound is fluxional; in CD2Cl2 the slow exchange limit is reached at –60 hC where the 1H NMR spectrum shows two bridging (d–2.02 and –2.66) and one terminal (d 4.55) hydrides. This zirconocene cation acts as a very potent initiator for the polymerisation of isobutene via a carbocationic mechanism. Detailed mechanistic NMR studies have shown that the mode of initiation of the polymerisation reaction is complex and involves insertion of isobutene into the Zr–H bonds, followed by hydrogenation to isobutane. Following the reaction by 13C NMR spectroscopy revealed a low-field signal at d 196.0, due possibly to a cationic Zr vinyl 30 or an h1-isobutene adduct of type 31 [71].

8.4

Olefin Coordination to d0 Metal Centers

Having generated suitable (partially) cationic, Lewis acidic metal centers, several factors need to be considered to understand the progress of the alkene polymerisation reaction: the coordination of the monomer, and the role (if any) of the counteranion on catalyst activity and, possibly, on the stereoselectivity of monomer enchainment. Since in d0 metal systems there is no back-bonding, the formation of alkene complexes relies entirely on the rather weak donor properties of these ligands. In catalytic systems complexes of the type [L2M(R)(alkene)]þ cannot be detected and constitute structures more closely related to the transition state rather than intermediates or resting states. Information about metal–alkene interactions, bond distances and energetics comes from model studies and a combination of spectroscopic and kinetic techniques. A first indication of the likely geometry of zirconium(IV) alkene complexes was found in the crystallographically characterised alkyne insertion product 32 [72],

323


324

which showed Zr–C bond distances to the coordinated diene unit of 2.6–2.76 A, i. e. at the outer limit of what would unequivocally be interpreted as a coordinative interaction. In line with this weak interaction, the C¼C bond distances differ little from that of a free olefin. For comparison, the Zr–C bond lengths in the Zr(II) complex Cp2Zr(C2H4)(PMe3) are 2.354(3) and 2.332(4) A, and the C¼C bond is elongated [73]. The labile vanadium(V) ethene and propene complexes 33 (R ¼ H, Me) were detected by low-temperature NMR; the C¼C bond is apparently aligned parallel to V¼N [74]. Similarly, weak Zr–alkene interactions were detected spectroscopically in the products of ethene and propene insertion into zirconium butenylborate complexes (cf. compound 12, Scheme 8.11) [53, 54]. NMR evidence suggests that pendant allyl substituents, as in 34, bind to the metal center in preference to the [PhCH2B(C6F5)3]– or [B(C6F5)4]– anions [75]. Me

H SiMe2

Me Me

N

Ph

V

(ButCp)2Zr

N

Zr

R Cp

Me 32

34

33

A first structural model of a Zr(IV) alkene (as opposed to diene) complex was provided by Jordan with the structurally characterised pentenoxy complex 35 (Scheme 8.17) [76]. The chelate ring size is important; the smaller butenoxy ligand does not chelate, larger ligands chelate more weakly. Obviously, the chelate formation restricts the orientation of the alkene, and it can be argued that the oxygen, which is capable of donating electron density to Zr via its lone pairs, reduces the Lewis acidity of the metal center compared to that of a zirconocene alkyl cation [Cp2Zr–R]þ. Nevertheless, a combination of NMR and X-ray diffraction studies has shown that this chelate possesses several of the features expected for a coordinated alkene ready to undergo migratory insertion. Thus, in 35 the terminal vinyl 13C resonance is shifted upfield, to d 94.3, while the signal of the neighboring C-2 exCp

Cp

Cp

O Zr

- MeCPh3

H

Zr

Zr Me

Cp

O

O

[CPh3][B(C6F5)4]

2

Cp 1

[B(C6F5)4]

Cp

H 35

Cp

Cp

O Zr 2.89

Cp 2.68

O 35A

Zr H

Scheme 8.17

Cp

H

Zr

O

36

8.4 Olefin Coordination to d0 Metal Centers

H

Cp*

B(C6F5)3

H Zr

B(C6F5)3

H

Cp* +

Cp*

B(C6F5)3

Zr Cp*

Cp*2Z r

325

H

38a

38b

37 [HNMePh2][B(C6F5)4]

H

Cp* H Zr Cp*

Scheme 8.18

H

Cp* Me

+

Me Zr

Cp* 39a

H 39b

periences a strong downfield shift of 20 ppm compared to the unchelated ligand, to d 158.8. The 1H signals of the vinyl group are similarly shifted, most notably, the signal for the internal CH is shifted downfield, from d 5.86 to d 7.50. It is evident that alkene coordination results in polarisation of the C¼C bond and accumulation of positive charge on the internal carbon atom (structure 35A), exactly the situation required for a subsequent nucleophilic attack on C-2 by a polymeryl chain in cisposition during the chain growth step in a metallocene catalyst. The chelate ring formation makes the two Cp ligands in 35 diastereotopic; this is resolved on cooling to –80 hC. There is however rapid interchange between the two possible olefin coordination modes; variable-temperature NMR studies on 35 and its rac-C2H4(Ind)2 analogue 36 suggest that this olefin face-interchange proceeds via a dissociative mechanism, with minimal participation of the anion, the solvent or s-complex intermediates. Line shape analysis and spectra simulation gives a value for the free energy of activation of the alkene-face exchange process as 10.7(2) kcal mol–1 for 35, and a slightly higher value for 36, DG‡ ¼ 15.4(4) kcal mol–1 [76b]. Casey was able to prepare related zirconocene alkenyl complexes according to Scheme 8.18. Alkene coordination was established by a number of NMR techniques. While zwitterionic compounds 38 allowed the determination of the alkene dissociation energy, DG‡ ¼ 10.5 kcal mol–1, very similar to that of 35. Thermally more stable complexes were obtained by protonation of 37 with [HNMePh2][B(C6F5)4]. Dynamic NMR spectroscopy and line shape analysis allowed the measurement of the barriers of alkene dissociation (DG‡ ¼ 10.7 and 11.1 kcal mol–1), as well as for the site epimerisation (“chain skipping”) at the zirconium center (DG‡ ¼ 14.4 kcal mol–1) (Scheme 8.19) [77]. Activation of silapentenyl zirconocenes with either B(C6F5)3 or [CPh3][B(C6F5)4] gave a similar series of cationic alkenyl complexes 40–43. Alkene bonding was readily recognisable by the large chemical shift separation of 3.6 ppm for the terminal vinyl protons and the downfield shift of the internal vinyl-H by 2 ppm to d 8.6, in line with resonance structures akin to 35A. Variable-temperature NMR in CD2Cl2 showed similar free energies of activation for alkene dissociation for com-


326

DG‡ (kcal/mol) 3.2

14.3

10.7

11.1

Cp*2Zr

Cp*2Zr

H Me alkene dissociation H

H

Cp*

H Zr

Me

H

0.4 H Me

Cp*2Zr

Me

H

Cp*

Cp* Cp*2Zr

Zr

Me

site epimerisation

H

Cp*

Me

H H

Scheme 8.19 (adapted from ref. [77a])

pounds 40–42, DG‡ z 13 kcal mol–1 (248 K), i. e. little influence of the ligand structure or the counteranion on alkene dissociation. Site epimerisation (i. e. inversion at Zr) occurs simultaneously with alkene dissociation. On the other hand, the doubly-bridged complex 43 gave no indication of olefin dissociation until 273 K; in C6D5Br at 283 K, DG‡ ¼ 15.6 kcal mol–1. There was no detectable anion influence on this process. These complexes proved surprisingly stereochemically stable, with no sign of site epimerisation even on heating to 75 hC [78]. R

Zr

SiMe2

Me2Si

Zr

SiMe2

Me2Si

Zr

SiMe2

Me2Si R 40, R = H; 41, R = But

42

43

Neutral lanthanide complexes are convenient models for the cationic zirconocene systems and avoid complications due to the presence of counteranions and the limited solubility of ionic compounds. Dynamic NMR studies on yttrium complexes 44–46 has allowed the determination of the alkene binding enthalpy, the activation enthalpy of alkene dissociation, and the relative rates of dissociation and alkyl site exchange (site epimerisation) (Scheme 8.20). Compared to the Zr

8.4 Olefin Coordination to d0 Metal Centers H

Cp*

Cp*

Cp* Y

Y

Y

Cp*

Cp*

Cp*

Me

Cp*

H

Cp*

Cp*

H

Y

Y

Y

Cp*

Cp* 45-on

H 44-on

44-off

44-on

Cp* 45-off

46-on

Scheme 8.20

chelate 39, the barrier to alkene dissociation in the yttrium complexes is ca. 2 kcal mol–1 lower. The dissociation of the monosubstituted alkene from chelate 46 in methylcyclohexane-d14/pentane-d12 (1:1) was measured at I–100 hC to determine the activation parameters (DG‡ ¼ 7.5 kcal mol–1, DH‡ ¼ 9.3 kcal mol–1). For the disubstituted alkene complex 45, direct observation of the equilibrium between the chelate (45-on) and non-chelate structures (45-off) was observed, giving a low binding enthalpy DH0 ¼ -2.6 kcal mol–1. The reason for the difference between this binding enthalpy and the alkene dissociation enthalpy may be the stabilisation of the open form by agostic interactions (e. g. 45-off). Binding energies for monosubstituted alkenes are higher, e. g. DH0 ¼ –4.0 kcal mol–1 for 46. The rate of alkene dissociation was found to be much faster than the rate of reversible alkene insertion to give 47 (Scheme 8.21), although the fact that the latter is destabilised by ring strain limits the general validity of such findings to alkene polymerisation systems [79]. Cp*

Cp*

Y

Y Cp*

DG‡ = 14.4 kcal/mol 46-on

Cp* 47

Scheme 8.21

The bonding of non-chelated alkenes to yttrium and their insertion into the Y–C bond has also been observed as a convenient model of the chain growth step in Group 4 catalysts. Yttrium alkyls are accessible from the corresponding alkene and [Cp*2Y(m-H)]2 at –50 hC in methylcyclohexane-d14. Primary yttrium alkyls of type 48 show b-agostic bonding in the ground state, as indicated by the observation of a small yttrium–carbon coupling to b-C (JCY ¼ 36 Hz), the high-field shift of the b-protons and deuteration studies (Scheme 8.22). This is also reflected in the larger than expected Ja-CH ¼ 124 Hz and the small Jb-CH ¼ 110 Hz; these values return to “normal” on addition of THF. The secondary alkyl 49 also shows a b-agostic inter-

327

328


H Cp*2Y

H

H

Cp*2Y H

H H

H

49

48

H

Cp*2Y 50

R

R R

R H

Cp*2Y H

H

Cp*2Y 51

H

52

Scheme 8.22

action with the methyl group, whereas compounds Cp*2Y–CH2CHR2 (50) adopt an a-agostic state, with a large JYC ¼ 51 Hz. The equilibrium constants for the coordination of propene, 1-butene and 1-hexene were determined in pentane-d12/methylcyclohexane-d14 (1:1) at –90 to –150 hC. Since the barrier of alkene dissociation is estimated at ca. 5 kcal mol–1, there is a rapid equilibrium between free and coordinated alkene at temperatures where insertion is slow. The 1H NMR chemical shift separation of the terminal E- and Z-protons (Dd ¼ dHZ – dHE) was used as a probe for alkene binding; Dd is small for free alkene but on coordination can increase to ca. 1.5 ppm. Equilibrium constants for propene binding to n-, g-substituted, b-substituted and sec-alkyl complexes were very similar. On the other hand, the rate of alkyl chain migration to propene depended strongly on the structure of the alkyl group and decreased in the series n-alkyl z g-substituted ii b-substituted ii a-substituted alkyl. Thus the insertion of propene (at –100 hC) into 48 to give 51 is 200 times faster than into 50, due to the kinetically slow migration of the b-substituted alkyl rather than to weak alkene binding. Rate constants for propene at –100 hC are 15 q 10 –4 L mol–1 s–1 for 48 and 0.08 q 10 –4 L mol–1s–1 for 50. Ethene insertion is rapid in all cases even at –130 hC [80]. These Group 3 models allow studies on fundamental aspects of alkene insertion that have so far eluded direct study with ionic (and much less soluble) Group 4 metallocene catalysts.

8.5

Ion Pair Dynamics in Metallocene Catalysts

Returning to ion-pair zirconocene catalysts, the initiation of the polymerisation process requires the displacement of the anion so that the alkene can be coordinated. The mobility of the anion is therefore an important factor and has become the focus of a number of detailed investigations. The original mechanistic scheme of alkene insertion and polymer chain growth (Scheme 8.4) implied dissociation of the anion and formation of a 14-electron cationic intermediate, which then reacted

8.5 Ion Pair Dynamics in Metallocene Catalysts (C2H4)n-R path A R L2Zr

R L2Zr

X

L2Zr

X

(C2H4)n+1-R L2Zr X

X R

d+ L2Zr

R L2Zr X

path B

associative

d-

R

path C

R L2Zr

L2Zr

dissociative

329

X

X (C2H4)n-R

d+ L2Zr

(C2H4)n+1-R L2Zr X

-

Xd

R

X

L2Zr path D

X

Scheme 8.23

with alkene. In this model the propagation process is essentially independent of the anion. But is this model correct? Anion exchange may be faster or slower than monomer binding and may be associative of dissociative (Scheme 8.23). The potential presence of the counteranion in the coordination sphere of the metal may also have consequences for the stereochemistry of polymer chain growth; if so, the role of the anion remains to be elucidated. Several scenarios are possible: 1. The anion dissociates, and the coordinatively unsaturated metal center then picks up a monomer molecule for subsequent enchainment. This dissociative model has been favored in the past [16, 21–23, 27–28] since it allows a convenient explanation of the observed polymer stereochemistry by considering only the roles of the ligand and the alkyl chain in the cationic metallocene complex. However, anion dissociation opposes the electrostatic attraction between cation and anion and is therefore energetically expensive. So does it operate at all? 2. Even if anion dissociation can be shown to be prevalent, there will be a strong tendency for anion binding. If anion re-association is slow, many monomer units may be consumed and polymerised before anion binding interrupts the process. On the other hand, if enchainment is slow, anion recombination can be expected to occur after each insertion step, to form a possibly long-lived resting state. Which mechanism applies? Evidence for anion mobility comes from the exchange process shown in Scheme 8.24, observed by dynamic NMR spectroscopy [81–84]. In principle, in such a system there are productive and non-productive exchange processes, i. e. those where R and X change positions and those where the starting complex is re-formed; only the former, ion pair symmetrisation pathway is spectroscopically observable of course and described by the ion-pair symmetrisation rate, kips The process can be conveniently monitored using suitable Cp substituents or, if present, the bridge-SiMe2 resonances as reporter signals.

etc.

330


X

X

R

R kips

Zr

Zr

Si

Si Me

Me

Me

Scheme 8.24

Me

In the specific case of [MeB(C6F5)3]–, a second exchange pathway is possible via borane dissociation and recombination. This, too, will lead to the permutation of Cp and bridge signals, though in addition Zr–Me and B–Me signals will exchange at the same rate (Scheme 8.25). The lowest energy exchange pathway in methylborate complexes L2ZrMe(m-Me)B(C6F5)3 is ion pair symmetrisation (anion exchange), with an activation energy of ca. 14–15 kJ mol–1, while DG‡ for borane exchange (Scheme 8.25) is slightly higher, 15–19 kJ mol–1. The evidence suggested that ion pairs are held within a solvent cage [82].

Si

Me* Zr MeB(C6F5)3

Si

Me* + B(C6F5)3

Zr Me

Si

Me*B(C6F5)3 Zr Me

Scheme 8.25

Brintzinger et al. have recently reinvestigated these anion exchange processes in complexes 53–58 of the more strongly coordinating [MeB(C6F5)3]– anion, as well as the analogous, more ionic [B(C6F5)4]– compounds, using NOESY experiments with mixing times t m ¼ 50–500 ms [84]. Measurements were carried out in C6D6 solutions at room temperature. For 53, 55 and 56 it was not possible to distinguish between anion and borane exchange processes, while for 54, borane rather than anion exchange predominated. On the other hand, in 57 and 58 only anion exchange was observed. For most of the ansa-metallocenes apparent first-order rate constants kapp of the order of 1–3 s–1 were found, although those of 58 are an order of magnitude higher. Generally rates increased with increasing zirconium concentration over the observed concentration range from 2–20 mmol L–1, as well as on addition of Li[MeB(C6F5)3]. The exchange rates for the analogous [B(C6F5)4]– ion pairs were about two orders of magnitude higher (recalculations from the given activation parameters lead to kapp z 200–300 s–1 at 300 K). Activation parameters were evaluated over the temperature range from 252–260 K (in toluene-d8) and 300–310 K (C6D6); although this range is somewhat narrow, the data indicated little difference in activation enthalpies DH‡ between the [MeB(C6F5)3]– and [B(C6F5)4]– systems, the difference in rates being almost exclusively due to entropy.

8.5 Ion Pair Dynamics in Metallocene Catalysts

Me Zr

Me

Si

Zr

MeB(C6F5)3

53

MeB(C6F5)3

54

Me

Si

Me

Si

Zr

MeB(C6F5)3

Si

Zr

55

Me Zr

MeB(C6F5)3

MeB(C6F5)3

57

56

Si

Me Zr MeB(C6F5)3

58

In effect, the polarised but essentially molecular complexes L2ZrMe(m-Me)B(C6F5)3 require much higher concentrations to display ionic-like exchange rates than [B(C6F5)4]– salts. This difference in polarity and exchange behavior is of course reflected in the well-known differences in solubility of the two systems and in their tendency to phase separation, i. e. [MeB(C6F5)3]– compounds show good solubility in hydrocarbon solvents and give homogeneous solutions even at high concentrations, whereas [B(C6F5)4]– systems in the same solvents tend to lead to microdroplets, phase separation and finally oily precipitates. The authors suggested the formation of ion aggregates, i. e. ion quadruples or higher assemblies, rather than ion pairs, at least under the observed concentrations. Such aggregates would facilitate anion exchange processes (Scheme 8.26) since any energy costs incurred by elongation of the Mþ p p p X– distance of one ion pair would be compensated by an energy gain as X– approaches the neighboring metal center of another ion pair, i. e. ion exchange becomes a concerted process. The suggestion of ion quadruple formation was further supported by diffusion coefficient measurements, using pulsed field-gradient NMR methods, which suggested the existence in benzene solution of ion aggregates larger than simple ion pairs [85]. These suggestions have not been without criticism, the objection being that these studies were carried out at significantly higher concentrations than are typically employed under catalytic conditions and are therefore probably not relevant to

X-

M+

M+

X-

331

X-

M+

M+

X-

Scheme 8.26

332


the actual catalytic process. Cryoscopic and pulsed field-gradient spin-echo (PGSE) NMR studies [86] did indeed confirm that methylborate complexes L2ZrMe(m-Me) B(C6F5)3, all of which are zwitterionic molecules in the solid state, as well as [Cp*2ThMe][B(C6F5)4], exist in solution as simple tight ion pairs [87]. Zirconocenium [B(C6F5)4]– ion pairs (Scheme 8.4, type (C)) are less accessible and were not measured. On the other hand, it could be shown by a combination of nuclear Overhauser effect (NOE) and PGSE measurements that outer-sphere ion pairs, where direct contact between cation and anion is prevented by firmly bonded donor ligands, as in 59–61, lose specific anion–cation interactions in solution at higher concentration and have a tendency to form ion quadruples or even hextuples, independent of the nature of the anion. As NOE measurements show, in the ion pairs the anion resides preferentially on the side opposite the Zr–Me moiety [88]. Since similar structures will be formed when an alkene is coordinated, the solution structures of these ion pairs are informative, although the overall trend seems to suggest that at catalytic concentrations (e. g. [Zr] ¼ 10 –5 –10 –6 M) ion pairs rather than ion aggregates prevail.

These investigations were carried out using zirconium–methyl complexes. However, as is evident from the chemistry depicted in Schemes 8.5 and 8.6, methyl ligands give a number of rather specific reactions and structures and are therefore poor models of the growing polymer chain. Thus it could be shown that, in the series (1,2-Me2Cp)2Zr(R)(m-Me)B(C6F5)3 (62), the rates of ion pair symmetrisation (expressed as reorganisation enthalpies) increase sharply with increasing bulkiness of the alkyl ligand, R ¼ Me I CH2But I CH2SiMe3 II CH(SiMe3)2 [69]. For the ansa-zirconocenes 63, exchange rates for R ¼ CH2SiMe3 are approximately four times higher than for R ¼ Me, with values of ca. 20 s–1 for X ¼ [MeB(C6F5)3]– and 800 s–1 for X ¼ [B(C6F5)4]–. PGSE studies on 63 (R ¼ CH2SiMe3, X ¼ [B(C6F5)4]–)

8.6 Monomer Coordination

Me3Si

Ha

Hb

X

Hb

X

Ha

SiMe3

Scheme 8.27

in toluene-d8 solutions with [Zr] ¼ 2.5 mmol L–1 show a degree of aggregation close to 3, i. e. the prevalent formation of ion hextuples. In such aggregates, it is difficult to envisage the site exchange processes observed by dynamic NMR (Scheme 8.24) as “ion pair symmetrisation” since the anion position cannot be unambiguously identified; they may constitute chain swinging in the presence of loosely associated counteranions. Site exchange in 63 is evident from the temperature dependence of the Me signals of the SiMe2 -bridge as well as for the indenyl-C5 signals, whereas the diastereotopic methylene-hydrogens of the CH2SiMe3 ligand do not interchange. This observation illustrates that the “chain swinging” event (site epimerisation) involves a 180h rotation of the alkyl ligand (Scheme 8.27), as indeed required by symmetry in catalysts with a C2 -symmetric ligand framework [89].

8.6

Monomer Coordination

The alkene monomer can approach the metallocene catalyst from various directions, head-on (Scheme 8.28, path A), from the same side as the anion (path B), or from a position trans to the anion, with side exchange of the alkyl ligand R (path C).

A

L

L

R

R Zr

Zr X

L L L

R Zr

L

L

R

R

Zr

B

Zr

X

L

X

X

L

X

L

L

L

C

R

Zr

Zr L

Scheme 8.28

X

R

L X

333

334


In contrast to alkene binding in models (Section 8.4), these processes are not accessible to direct spectroscopic investigations. However, in an effort to provide evidence of the stereochemical course, rates and equilibrium positions of anion displacement, Brintzinger et al. have used 1D and 2D NMR methods to investigate the reactions of a series of zirconocene methylborates, including Cp2ZrMe(m-Me) B(C6F5)3 as well as 54 and 57, with weak donors L, notably N, N-dimethylaniline (DMA) and dibutyl ether (DBE) [90]. Anion displacement with these donors is reversible. Indenyl and benzindenyl complexes did not react with DMA. Surprisingly, the equilibrium constants K ¼ [ZrMe(L)þX–]/[ZrMe(X)] p [L] change by only a factor of 5 from the most open (54) to the most bulky ligand system (57). However, the second-order displacement rate constants change by five orders of magnitude, from 0.03 L mol–1 s–1 for 57 to 4000 L mol–1 s–1 for 54 (at 300 K). The rate of anion displacement proved particularly sensitive to substituents in the 2-position on the cyclopentadienyl ring. The anion substitution reaction shows several features that are likely to be important for analogous reactions in polymerisation catalysis: (i) The reaction proceeds via an associative, SN2-type mechanism, via fivecoordinate intermediates; (ii) anion substitution is stereospecific, with attack by L on the same side as the anion; (iii) on addition of L, the anion is retained in an outer-sphere complex; solvated, separated ions are not formed in hydrocarbon solvents; (iv) exchange of outer-sphere and inner-sphere anions shows the same stereospecificity as anion displacement by the donor ligand; this process is much faster than methyl site exchange. The rates of anion displacement depended strongly on the ligand framework and was found to be particularly slow for those ligands known to give the most effective catalysts for the polymerisation of propene, i. e. for 55, k ¼ 8 L mol–1 s–1, while the value for 57 is even lower, k ¼ 0.03 L mol–1 s–1. To the extent that these data are applicable to catalytic systems, the results point towards a very slow initiation, although this is not backed up by kinetic studies. The reverse reaction, the rate of displacement of DBE by [MeB(C6F5)3]–, is even slower: 0.005 s–1 and 2.7 q 10 –5 s–1 for 55 and 57, respectively. Polymerisation rates with such catalysts are typically many orders of magnitude higher. Even if it can be expected that the anion can compete much more effectively with alkene ligands than with DBE, these slow rates of anion re-association have led the authors to suggest that anion re-association is an infrequent event that constitutes a chain termination reaction. This would hold particularly for the even less nucleophilic [B(C6F5)4]–. Chain growth would therefore be initiated by anion displacement, followed by rapid monomer uptake and slow anion re-coordination. These suggestions gain support by the observation that the molecular weights of polypropenes generated with “constrained-geometry” titanium catalysts are 10 times higher if the counteranion is [B(C6F5)4]– rather than the more strongly coordinating [MeB(C6F5)3]– [42]. On the other hand, the idea of a “spectator” role for the anion during the polymerisation process was contradicted by catalyst activity and kinetic studies. Efforts have been made to determine the approach of monomer onto the catalysts by theoretical modelling. DFT calculations on the ion pair [(1,2-

8.7 Polymerisation Kinetics

Me

Me Si

+ Zr

Me

Si X

Me

Me + Zr

Me

Me

Me L

Si

+ Zr

Me

L

X X

Scheme 8.29

Me2C5H3)2ZrRþ p p p B(C6F5)4 –] (R ¼ Me, Et) have suggested that ethene-sandwiched p-complexes are formed via pathways A and B, and a non-sandwiched p-complex via C (see Scheme 8.28). Semi-empirical and DFT calculations by the Brintzinger group suggested a front-attack by L on the metal center, which readily explains the observed retention of configuration during the substitution process (Scheme 8.29). The same structural type was also favored for L ¼ ethene or propene [90]. On the other hand, DFT calculations by Ziegler et al. found the product of path C, arising from C2H4 attack on the metal center trans to the anion in an SN2type reaction, to be the most stable by 3.5 kcal mol–1 [91]. Calculations by Lanza and Fragal on a simplified model of the constrained-geometry catalyst, [(C5H4SiH2NMe)TiMe][MeB(C6F5)3], ruled out paths A and B and suggested the monomer approach via path C, i. e. from the side opposite to the anion, as the favored insertion pathway [92]. This reaction sequence minimises cation–anion separation and allows facile anion coordination once the insertion step is complete. By contrast, calculations by Nifant’ev et al. on Cp2ZrEt(X) þ C2H4 suggested path A as the preferred route for anion substitution by the olefin [93].

8.7

Polymerisation Kinetics

The use of 13C NMR spectroscopy to probe for intermediates in the metallocene catalysed ethene polymerisation dates back to early work by Fink et al. who studied the system Cp2TiRCl/AlR2Cl/ethene (R ¼ Me, Et) [31, 94]. The equilibrium constants for the reaction Cp2TiRCl þ AlR2Cl p Cp2TiR(m-Cl)AlR2Cl were determined from the chemical shift changes of the Ti–CH2 and Cp signals; e. g. for Cp2TiEtCl þ AlEtCl2, K j 5000 at 240 K[94b]. The use of 13C2H4 in this comparatively slow catalyst system allowed the assignment and quantification of the Ti–n-propyl, Ti–n-pentyl etc. oligomers and the determination of the relative rates of the first few ethene insertion into Ti–alkyl bonds: C1 p C3

kMe ¼ 1 L mol–1 s–1

C2 p C4

kEt ¼ 120 L mol–1 s–1

C3 p C5

kpropyl ¼ 96 L mol–1 s–1

C4 p C6

kbutyl ¼ 62 L mol–1 s–1

335

336


The data showed (i) that initiation was significantly slower than propagation, i. e. that polymer chains were formed before most of the catalyst had undergone the first insertion; (ii) that the primary complex between the titanocene and the aluminum alkyl activator was not itself the active species; (iii) that the active species was in equilibrium with an observable stabilised species, i. e. a resting state, and (iv) that a titanium ethene p-complex (cf. Schemes 8.4 and 8.23), although mechanistically necessary, was not present in detectable quantities. These observations formed the basis of a general mechanism where polymer chain growth proceeded stepwise and could be interrupted at any stage by reversible formation of a resting state, the “intermittent growth” model [95]. More recently, Landis et al. studied the polymerisation kinetics of 1-hexene with (EBI)ZrMe(m-Me)B(C6F5)3 64 as catalyst in toluene [EBI ¼ rac-C2H4(Ind)2]. Catalyst initiation was defined as the first insertion of monomer into the Zr–Me bond, 65 (Scheme 8.30). Deuterium quenching with MeOD was used to determine the number of catalytically active sites by 2H NMR. The time dependence of the deuterium label in the polymer was taken as a measure of the rate of catalyst initiation. This method also provides information of the type of bonding of the growing polymer chain to zirconium, as n-or sec-alkyl, allyl etc. Hexene polymerisation is comparatively slow, with high regio- and stereoselectivity; there was no accumulation of secondary zirconium alkyls as dormant states [96].

Me Zr

MeB(C6F5)3 L2Zr

1-hexene

MeB(C6F5)3 L2Zr Me

MeB(C6F5)3

Bu

64

Bu

Bu

Me 65

66

MeOD

Me

D

Scheme 8.30

n

Bu

Bu

n

Under the definition given above, the active species concentration grows exponentially towards 100 % (neglecting catalyst deactivation by trace impurities). On the other hand, isotopic quenching measures the totality of Zr–alkyl bonds, whether active or not, and as discussed above, in species like 65 the coordination site required for alkene binding is occupied by the anion, i. e. 65 could be regarded as akin to a resting state. By contrast, Song et al. determined the active species concentration in propene polymerisations with related ansa-zirconocene catalysts, using a combination of quenched-flow kinetic techniques and the time-dependence of the number-average polymer molecular weight. Under this definition only those complexes that are actively engaged in polymer chain growth at any one time are classed as “active species”; in this case this amounted to ca. 8 % of the total zirconocene [97].

8.7 Polymerisation Kinetics

Following the reaction of 1-hexene with 64 at –40 hC by 1H NMR spectroscopy indicated a partial (50–65 %) conversion of 64 into the Zr–poly(hexene) (Zr–PH) species 66 [98]. The first insertion step was 400 times slower than subsequent insertions; propagation is 4 q 104 times faster than termination. The polymerisation was living under these conditions, and polymerisation of 1-13C-hexene gave a Zr–13CH2 -polymeryl species. The rate of disappearance of the Zr–13CH2 signal on addition of 0.4–2 equivalents of unlabelled 1-hexene led to a kinetic model where the propagation rate was equal to catalyst re-initiation, i. e. each monomer insertion step was followed by anion re-coordination. This “continuous” insertion mechanism is in contradiction to Brintzinger’s suggestions [90] of slow anion re-coordination and Fink’s “intermittent” model [95]. Since the propagation rate of hexene polymerisation is low, kp z 6 L mol–1 s–1 at 20 hC [96b], the re-association of a comparatively strongly coordinating anion such as [MeB(C6F5)3]– is plausible. Whether the same kinetic regime prevails in much faster ethene or propene reactions remains to be determined. Since at –40 hC the hexene polymerisation is living, the addition of 10 equivalents of propene to species 66 led to the complete consumption of propene and formation of a Zr–PP–b–PH block (without further conversion of any unreacted Zr–Me precursor). The identification of the Zr–(propene)x–(hexene)y polymer was confirmed by isotopic labelling (1-13C-propene, 1,1’-D-2,3-13C-propene) and 1H, 13C, 19 F and 11B NMR spectroscopy. The rate of propene polymerisation with this system was only three times faster than that of 1-hexene. This slow rate contributes to the high regioselectivity of the polymerisation: no 2,1-propene misinsertions were detected. 1H and 13C NMR spectroscopy also provided information about the chain termination mechanism; here this occurred by b-H elimination in a first-order process. Polymer chain-end epimerisation, i. e. chirality inversion at the b-carbon of the polymer chain (Scheme 8.31), proceeded via a zirconium tertalkyl (rather than p-allyl) intermediate [96c]. X

*

X

X

H

Zr

Zr

POL Me

H

Zr

X H

POL Me

Zr *

POL

Me Me *

Me

*

POL X Zr

X H

Zr

X

H Me * POL

POL Me *

Zr

H Me * POL

Scheme 8.31

Application of Singleton’s method of using 13C NMR to determine 12C/13C kinetic isotope effects [99] to hexene polymerisations with 64 showed a higher KIE for C2 than for C1 but none for C3 –C5. The results, in comparison with computed KIEs, are in agreement with an alkene binding equilibrium, followed by an irreversible alkene insertion step (Scheme 8.32).

337

338


X Cp2Zr

+

k1 R

k- 1

Cp2Zr

R

X

k2

Cp2Zr X

Scheme 8.32

Since the same isotope effect was observed for different activators (B(C6F5)3, Al(C6F5)3, [HNMe2Ph][B(C6F5)4], MAO), in spite of widely varying catalyst productivities, it was concluded that the transition state must have very similar structures in all cases [100]. Although this might seem to question the influence of the counteranion on the transition state energy, the results suggest, in essence, that the transition state involves the same step, i. e. the transfer of an alkyl ligand to monomer. By contrast to the polymerisation of hexene with 64, which can be followed conveniently by variable-temperature NMR, the polymerisation of smaller monomers like ethene and propene illustrate the limitations of spectroscopic methods since with most metallocene catalysts they are too fast. The kinetic behavior of (SBI)ZrMe2/AlBui3/[CPh3][CN{B(C6F5)3}2] at 25 hC was therefore investigated by quenched-flow techniques to estimate the rates of initiation, chain propagation and chain termination [SBI ¼ rac-Me2Si(Ind)2] [97]. The results are summarised here for comparison with the results on 1-hexene polymerisation discussed above. The polymerisation is first-order in [C3H6] and [Zr]. Two propagation rate constants were determined: the time dependence of polymer mass gives an “apparent” rate kpapp, since the analysis assumes that 100 % of the initial zirconocene concentration has become catalytically active; kp z (1.3–1.9 e 0.1) q 103 L mol–1 s–1, with the lower values found for the higher catalyst concentration where monomer depletion effects are felt. At [C3H6] ¼ 0.59 M this corresponds to an observed first-order rate kobs ¼ 760–1100 s–1. By contrast, the rate of polymer chain growth was an order of magnitude faster, and from the determinations of the number-average molecular weight Mn an observed propagation rate kpobs¼ 10,000 s–1 and a secondorder rate constant kp ¼ (17.2 e 1.4) q103 L mol–1 s–1 were determined, i 2000 times faster than hexene polymerisations with 64. The ratio kp/kpapp was taken as a measure of the mole fraction of total [Zr] that was actively engaged in chain growth at any one time, in this case kp/kpapp ¼ 0.08. About 90 % of the total zirconocene added was in a reversibly deactivated (dormant) state. The nature of the dormant state was determined by 1H NMR end-group analysis. Two types of terminal unsaturations were found, vinylidene end groups indicative of b-H elimination from 1,2-inserted polymeryl chains, and cis-butenyl end groups, arising from 2,1-misinsertions (Scheme 8.33); the latter were dominant (66 %). 13 C NMR analysis of the polymer also detected low levels of stereo-errors due to enchained 2,1-misinsertions (cf. Section 8.9), about 1 in 500. The data suggested that 2,1-insertion is slow but is responsible for the accumulation of dormant states carrying Zr–sec-alkyl chains.

8.8 Spectroscopic Studies on Complex Systems

[Zr]

P

P

vinylidene

cis-butenyl

[Zr]

P

P

Scheme 8.33

Very similar kinetics were observed with (SBI)ZrCl2/MAO: although this catalyst is an order of magnitude less active, the concentration of active species turned out to be almost identical to the borate system: kp/kpapp ¼ 0.08, with a similar accumulation of dormant states. Evidently the counteranion ([CN{B(C6F5)3}2]– vs. [Me–MAO]–) modulates the energetics of the chain growth cycle, i. e. the anion is intimately involved in the transition state, but does not influence the distribution between active and dormant states (Scheme 8.34) [97].

CH2R'

2,1 slow

CH2R'

L2Zr

CH2R'

L2Zr X

L2Zr

slow

fast X

Resting state

X

Resting state 1,2 fast

H L2Zr

CHR'

chain growth cycle

CH2R' L2Zr

X

X

Active species

Scheme 8.34

8.8

Spectroscopic Studies on Complex Systems

Most of the spectroscopic investigations discussed above were carried out on “welldefined” metallocene systems, either isolated species or those generated from a well-defined metallocene alkyl precursor activated with one equivalent of a borane or borate activator. Most practical polymerisation catalysts, on the other hand, include a scavenger, usually an aluminum alkyl, and may contain ill-defined activators such as methylaluminoxane (MAO), usually at high MAO/Zr ratios. Such systems are less amenable to quantitative studies; nevertheless, the identifications of species such as those depicted in Schemes 8.5–8.8 has enabled similar compounds to be identified in more complex mixtures. An idea of the possible mode of action

339

340


of MAO and the Lewis acidity of some of its aluminum sites is provided by the reaction between Cp2ZrMe2 and structurally characterised tert-butylaluminoxane cages such as [ButAl(m3 -O)]6, which leads to methyl transfer to Al and formation of 67 [101]. d-

But

Al Me

d

Zr

t

O Bu Me

O

O

Bu +

But

Al Al

Al t

Al O

But

O Al O

But

67

In early studies Tritto et al. investigated the system Cp2TiMe2 and Cp2TiMeCl with MAO at moderate Al/Ti ratios of 10:1 to 40:1 using 13C NMR. A number of resonances were observed which were tentatively assigned to the ion pairs [Cp2TiMe]þ[Cl–MAO] and [Cp2TiMe]þ[Me-MAO] [102]. Bryliakov et al. recently re-investigated this system, using Cp2TiCl2/MAO at Al/Ti ratios of 5:1 to 300:1 and 13C-enriched MAO. The 1H and 13C NMR spectra in toluene showed that seven different types of Ti(IV) species were formed: the methyl complexes Cp2TiMeCl (mainly as adducts with AlMe3 and MAO) and Cp2TiMe2, the homobinuclear ion pairs [Cp2TiMe(m-Cl)Cp2TiCl]þ[Me-MAO] (68a), [Cp2TiMe(m-Cl)Cp2TiMe]þ[MeMAO] (68b) and [Cp2TiMe(m-Me)Cp2TiMe]þ[Me-MAO] (68c), the heterobinuclear ion pair [Cp2Ti(m-Me)2AlMe2]þ[Me-MAO] (69) and a ‘zwitterion-like’ intermediate Cp2TiMe(þ)nMeAl(–)aMAO (70). In spite of models like 67, the interaction with MAO is thought to involve a methyl rather than an oxide bridge. Species 68a could only be detected at very low Al/Ti ratios, while at higher ratios (20–40) the homobinuclear compounds 68b and 68c appeared. The chemical shifts of 68a and 68b were only slightly affected by the outer-sphere anion, as shown by comparison with the ion pairs [Cp2TiMe(m-Cl)Cp2TiR]þ[B(C6F5)4] (R ¼ Cl, Me). With increasing Al/Ti ratio, 68b is replaced by methyl-bridged 68c. At Al/Ti ¼ 300, complexes 69 and 70 dominated in the reaction mixture; the assignment of the signals for 69 was confirmed by the in situ synthesis of [Cp2Ti(m-Me)2AlMe2]þ[B(C6F5)4] [103]. Using 13C-labelled Cp2Zr(13CH3)2 in combination with MAO at Al/Zr ratios of 10:1 to 40:1, as well as B(C6F5)3 for comparison, Tritto et al. identified compounds 71 and 72, analogous to species (A1) and (B) in Scheme 8.6 [104]. Only the C–H coupling constant for the terminal Zr–Me in 71 was reported, JCH ¼ 120.3 Hz (cf. Section 8.2). Species 72 has an even smaller coupling constant, JCH ¼ 100 Hz, although the signal was too broad for an accurate determination. Since the MAO used had been dried under vacuum to remove toluene and excess AlMe3, at the given Al/Zr ratio of 20 the heterobinuclear cation 73 was only detected in very low amounts. The scrambling of 13CH3 into MAO indicates that


methyl exchange reactions proceed rapidly in this system. A similar 13CH3 scrambling between Cp2ZrMe2 and MAO was found by Siedle [105]. Raising the temperature from 0 to 25 hC and increasing the Al/Zr ratio led to a relative increase in the mononuclear species 72. Low Al/Zr ratios of 10–40 are far removed from catalytic conditions where usually Al/Zr ratios of 1000:1 and higher are required before good catalytic activity is observed. Using 13C-enriched MAO, Babushkin and co-workers extended the 13C NMR studies on the Cp2ZrMe2/MAO system to Al/Zr ratios of up to 4000 and zirconocene concentrations down to 0.5 mmol L–1 [106]. In addition to 71–73, a weak adduct between Cp2ZrMe2 and MAO was also identified, though its Cp and Me resonances were broad; this was thought to represent a different type of donor–acceptor interaction from 72. At Al/Zr i 1000 this adduct as well as 71 became undetectable, while with decreasing [Zr] and increasing Al/Zr ratio species 73 became dominant. The spectroscopic parameters correspond well to those determined with [B(C6F5)4] – as counteranion [39]. Some exchange broadening of 73 was explained by an equilibrium between contact and solvent-separated ion pairs. Species 72 gives inhomogeneously broadened 1H and 13C NMR signals, thought to arise by slow exchange of the [Cp2ZrMe]þ moiety with various Al–Me sites in MAO. A more comprehensive 1H and 13C NMR study on a wide range of zirconocene catalysts showed a similar picture. The NMR data and ethene polymerisation activities of the (Cp-R)2ZrCl2/MAO system over Al/Zr ratios of 50–1000 were compared with (Cp-R)2ZrCl2/AlMe3/[CPh3][B(C6F5)4]. At low Al/Zr ratios of 50–200, species of type 72 with different [Me–MAO] – counteranions were identified (R ¼ Bun, But), while at Al/Zr ratios of 500–1000 Zr–Al heterobinuclear ions of type 73 dominated in all MAO-activated catalysts. For the systems (Cp-R)2ZrCl2/MAO (R ¼ H, Me, Bun, But), the ethene polymerisation activity depended strongly on the Al/Zr ratio, whereas the activity was virtually constant over the same Al/Zr range if

341

342


R ¼ 1,2-Me2, 1,2,3-Me3, 1,2,4-Me3, Me4 or (Cp-R)2 ¼ SBI. Polymerisations were conducted in toluene under 2–5 bar ethene pressure and found to be of the order of 20–30 q 106 gPE (mol Zr) –1 bar–1 h–1 [107]. The size of the [Me–MAO]– anion in 73 at high Al/Zr ratios typical for polymerisation systems was estimated by diffusion coefficient measurements using pulsed field-gradient (PFG) NMR methods. The MAO used was freed of excess Al2Me6 by evacuation at room temperature and used in Al/Zr ratios of 60–15,000. At Al/Zr i 200 only products 72 and 73 are detected, of which 73 showed sufficiently sharp signals for PFG measurements. Both species are in a concentrationdependent equilibrium. A mean effective hydrodynamic radius of the ion pair in C6D6 of 12.2–12.5 A indicated tight ion pairing even at the lowest [Zr]; the volume of the [Me–MAO]– anion suggested that it contained ca. 150–200 Al atoms [108]. NMR studies on the half-sandwich titanium complexes Cp*TiMe3/MAO and Cp*TiCl3/MAO (Cp* ¼ C5Me5) in toluene and chlorobenzene in the temperature range 253–293 K using 13C-enriched MAO showed up significant differences between half-sandwich and bis-Cp complexes [109]. Titanium half-sandwich complexes have been under discussion as catalysts for the polymerisation of styrene, and there has been some debate whether such catalysts are Ti(IV) or Ti(III) [110, 111]. The NMR studies showed that the activation of Cp*TiMe3 with MAO gives mainly a zwitterionic product of type 74. In the system Cp*TiCl3/MAO three products were identified, Cp*TiMeCl2, Cp*TiMe2Cl and 74. The latter is the dominant product at high Al/Ti ratios of i200; heteronuclear Ti–AlMe3 adducts akin to 73 could not be found. EPR measurements showed that in the Cp*TiCl3/MAO system at Al/Ti ¼ 35 the proportion of EPR-active Ti(III) is small, I1 %, which rises to 10 % at Al/Ti ¼ 700 [109].

d+ Ti Me Me

Me

dAl MAO 74

Deck and co-workers used 19F NMR spectroscopy to probe the nature of MAO-activated species, using the C6F5 -substituents in Cp(C6F5 -Cp)ZrCl2 and (C6F5 Cp)2ZrCl2 as probes. The 19F NMR spectra were very similar to the those of (C6F5 -Cp)2ZrMe(m-Me)B(C6F5)3; the relative amount of L2ZrMe2 and [L2ZrMe]þ[Me-MAO]– was a function of the MAO concentration. The results point towards MAO having a relatively low concentration of sites that are Lewisacidic enough to abstract a methide from the zirconocene [112]. Some of the most active catalysts are generated by activating metallocene dichlorides with the [CPh3]þ or [HNMe2Ph]þ salts of non-coordinating anions in the presence of AlBui3 as scavenger and alkylating agent. AlBui3 is far more effective in this respect than AlMe3 or AlEt3 [113]. For example, while (SBI)ZrCl2 activated


H

H [CPh3][B(C6F5)4]

Cp2Zr H

- HCPh3

Cp2Zr

Cp2Zr

+

HCMe3

75

Scheme 8.35

with MAO described in the previous section showed activities of ca. 20–30 q 106 g PE (mol Zr) –1 bar–1 h–1, the activity of the same catalyst precursor increases by a factor of 30–40 when activated with [CPh3][CN{B(C6F5)3}2] in the presence of AlBui3 [114]. The chemistry of metallocene systems containing AlBui3 is however little known. Low-temperature NMR studies showed that the reaction of Cp2ZrBui2, prepared from Cp2ZrCl2 and BuiMgCl at I 0 hC, with [CPh3][B(C6F5)4] leads to fast C–H activation and formation of [Cp2Zr(h3 -methallyl)]þ 75, without evidence for cationic Zr–Bui intermediates (Scheme 8.35) [115]. Allylic activation in zirconocene chemistry is a well-recognised phenomenon [60, 67, 116]. The reactions of Cp2ZrCl2 and Ph2C(Cp)(Flu)ZrCl2 76 (Flu ¼ fluorenyl) with AlBui3 in the presence and absence of [HNMe2Ph][B(C6F5)4] was investigated by variable-temperature NMR spectroscopy [117]. The reaction of Cp2ZrCl2 with AlBui3 required more than 2 equivalents of the aluminum alkyl for total conversion. The product, obtained in C6D6 at room temperature, is not a zirconocene alkyl but the hydride complex 77, resulting from b-H eliminations and characterised by resonances at d –0.88 (triplet, Zr–H–Al) and –1.95 (Zr–H–Zr) (Scheme 8.36). Similar AlMe3 hydride adducts have been known for some time [118]. Reacting Cp2ZrCl2 with AlBui3 in C6D6 followed by [HNMe2Ph][B(C6F5)4] gave a complex mixture of zirconocene hydride species, probably including zirconium dimethylaniline adducts. The reaction of 76 with AlBui3 under similar conditions gave exclusively the mono-alkylated product, Ph2C(Cp)(Flu)Zr(Cl)Bui 78, even at Al/Zr ratios up to 50; a ratio of 10 is required for total consumption of 76. The CH2 protons in 78 are diastereotopic and give rise to an ABX pattern (2JAB ¼ 13 Hz). The 13C chemical shift of the methylene group changed from d 73.6 to 85.4 with increasing Al/Zr ratio, most probably due to an equilibrium with a chloro-bridged adduct of AlBui3 or AlBui2Cl. Adding [HNMe2Ph][B(C6F5)4] to such a mixture gave an oily precipitate which allowed the organic by-products to be removed. The residue was identified by a combination of NMR techniques as the unusual hydrido-allyl complex 79; the concentration of this species increases with increasing Al/Zr ratio. At Al/Zr ¼ 100 compound 79 was the only observable product. Whether such products, obtained at high [Zr] after lengthy work-up at room temperature, are characteristic of the species responsible for the high activity of AlBui3 containing systems under catalytic conditions, i. e. in situ generation at very low [Zr], is not clear at present. However, Bryliakov et al. recently observed the forma-

343

344

8 The Use of Spectroscopy in Metallocene-based Polymerisation Catalysis AlBui3

Cp2ZrCl2

excessAlBui3

H

H Cp2Zr

-

H

ZrCp2

77

H

Bui3Al

ZrCl2

Ph2C

excessAlBui3

AlBui2X LZr

LZr

C6D6

Cl

Cl 78

76

excessAlBui3

AlBui2X

X = Bui, Cl

[HNMe2Ph][B(C6F5)4]

Bui Bui H

Al

B(C6F5)4

LZr

Scheme 8.36

79

tion of products similar to 79 in the system (SBI)ZrCl2/AlBui3/[CPh3][B(C6F5)4] [119]. The identification of a zirconium-bound polymeryl chain was achieved by following the polymerisation of 13C2H4 by Cp2Zr(13CH3)2 in the presence of either MAO or B(C6F5)3 [120]. Reactions with MAO were conducted at –20 hC in toluene-d8 with a low Al/Zr ratio of 20:1. Species 71–73 are formed under these conditions, together with the appearance of signals at d 55 and 65 for Zr–CH2 –P moieties, observed as doublets due to coupling with one neighbouring 13C (JCC ¼ 29.7 Hz). Related Cp2Ti–CH2 –P species had earlier been detected by 13C NMR spectroscopy in Cp2TiCl2/AlEt2Cl systems at low Al/Ti ratios [94, 95]. Products of chain transfer to aluminum were also identified, indicating that this process is facile even at fairly low Al/Zr ratios at low temperatures. By comparison with the B(C6F5)3 -activated system, the polymer-chain carrying species were identified as [Cp2ZrCH2 –P][Me–MAO] and Cp2Zr(Me)CH2 –P which are in equilibrium with each other [120].

8.9

Spectroscopy of Poly(1-alkenes): Polypropylene

A particularly important application of NMR spectroscopy is the structural characterisation of polymers. While initial attempts were made using 1H NMR, the wide

8.9 Spectroscopy of Poly(1-alkenes): Polypropylene

chemical shift range of 13C NMR is ideal for this purpose and was used almost as soon as it became more generally available [121]. The determination of the microstructure of vinyl polymers is not merely a characterisation tool. Each polymer molecule is unique, and each polymer chain is a record of the history of its formation, including mis-insertions, rearrangements, the incorporation of co-monomers, and the mode of its termination. NMR analysis of polymers can therefore be used to provide detailed mechanistic and kinetic information. This approach has been applied particularly successfully to the microstructure, i. e. the sequence distribution of monomer insertions, of polypropylene, giving rise to a wealth of studies far too numerous to cover here. Progress in this area has recently been summarised in two excellent and very comprehensive review articles [122, 123]. Here we will cover only the most fundamental aspects of stereoselective polymerisations. In principle, each enchainment of a prochiral monomer such as propene generates a chiral carbon center. The relative orientation of two such centers can therefore give rise to meso-diads or racemo-diads. Three monomers consecutively enchained in meso-fashion give rise to a mm triad, and so forth (Scheme 8.37).

Me

H Me

CH2

CH2 meso-diad

H

Me CH2

H

H

CH2

racemo-diad

Me

m

m

mm-triad

Scheme 8.37

In principle, therefore, the microstructure of such a polymer could be described by a sequence of m and r diads. However, the limits of 13C NMR resolution is such that the microstructure can only be determined over a limited section of the polymer, usually no longer than 5–7 monomer units. NMR spectra on comparatively insoluble polymers such as polyethylene and polypropylene are usually recorded in 1,2,4-trichlorobenzene or tetrachloroethane-d2 at elevated temperatures (110–150 hC). There are three principal stereochemical types of poly(1-alkene)s, illustrated in Scheme 8.38 for polypropylene. In isotactic polypropylene 80 (i-PP) all methyl substituents have the same relative orientation (m). The scheme shows the stereochemistry with the usual Fischer projection underneath. In syndiotactic PP (81, s-PP) every second CHMe unit has the opposite stereochemistry to the first, while in atactic PP (82, a-PP) the orientation of the methyl substituents is random. In some polymers there is partial order, i. e. only every second monomer orientation is random (83, hemi-isotactic PP). The 13C NMR chemical shift of a given CH3 group is influenced by its neighbors. Most modern NMR instruments will allow resolution at the “pentad” level, i. e. the resonance of a methyl group is determined by the orientation of the two monomer

345

346


m

m

m

m

r

m

r 81

80 mmmm mmmr

82

mmrm mrmr

83

Scheme 8.38

units on either side. This gives rise to different signals for mmmm, mmmr, mmrm etc. pentads. The 13C NMR chemical shifts of PP pentads are given in Table 8.1. A significant improvement in resolution and information content of NMR spectra was achieved by high-field (150 MHz) 13C NMR which allows resolution at heptad or even higher levels. At present, this level of resolution is still the preserve of the specialist NMR practitioner. In any case, for most purposes of structural assignments resolution at pentad level is satisfactory. For chemical shifts of high-resolution spectra see Ref. [123].

Experimental 13C NMR chemical shifts for the methyl pentads of regioregular polypropylene (in 1,2,4-trichlorobenzene, 137 hC). Table 8.1

Pentad

d (13C)/ppm

mmmm mmmr rmmr mmrr mmrm ¼ rmrr rmrm rrrr rrrm mrrm

21.78 21.55 21.33 21.01 20.85 20.71 20.31 20.17 20.04

8.9 Spectroscopy of Poly(1-alkenes): Polypropylene 1st insertion

2nd insertion H

Me

H

Me

Me

path A

M

Me H

H

Me

R

H

R

R M

347

Me R

M

Me

isotactic

M

path B

H

Me

H Me

Me Me

R

R

H

Me H

M

H

Me R

M

M

Scheme 8.39

A catalyst will generate an isotactic polymer as long as consecutive propene monomers bind to the metal with the same p-face (Scheme 8.39 path A), as illustrated in the scheme by black and white p-orbitals of the alkene. On the other hand, if the second monomer binds with the opposite p-face to the first (path B), a syndiotactic polymer will result. There are, in principle, two main pathways by which stereoselectivity can be achieved. Since the carbon in the b-position to the metal center is chiral, its chirality can control the preferred orientation of the next incoming monomer, i. e. the polymer chain end controls the polymer structure. This is the case with classical heterogeneous catalysts as well as with metallocenes without stereo-directing ligands. For example, in early work Ewen found that a Cp2TiCl2/MAO catalyst at low temperature (–60 hC) gave i-PP due to chain-end control; however, stereoregularity was lost at higher temperatures [124]. On the other hand, using Brintzingers’ rac-C2H4(Ind)2TiCl2 [19], an isotactic polymer was obtained, the stereoregularity of which was the result of the ligand structure. This control process is termed enantiomorphic site control. Chain-end control and site control pathways can be differentiated by 13C NMR. In the chain-end control process, a stereo-error will be propagated further, to give a predominantly iosotactic polymer with a block-type structure; the stereochemistry will only change back to the original one if a second stereo-error occurs (Scheme 8.40). In the NMR spectrum this is indicated by the occurrence of mmrm (but not mmrr) pentads, i. e. r diads are isolated. In the case of enantiomorphic site control, the error will be corrected by the chiral ligand environment, leading to two successive insertions with relative r stereochemistry and the observation of mmrr pentads. Successful chain end control at higher temperatures would require a very rigid ligand environment of the catalytically active center, as on the edges of solid-

syndiotactic

348


Chain-end control: error m

m

m

m

error m

r

r m

m

m

m

m

m

m

m

r

m

m

m

mmrm Site control: m

Scheme 8.40

m

error m

m

m

mmrr

r

error r

m

m

m

m

m

r

mmrr

state catalysts. Metallocenes, on the other hand, are particularly suited for the construction of ligands that generate chiral coordination pockets, akin to the active sites in enzymes, such that the polymerisation is conducted under enantiomorphic site control. There is now a huge family of metallocene complexes that give a greater or lesser control of polymer regio- and stereochemistry. Complexes 84– 88 are typical examples. The C2 -symmetric compounds 84–87 produce i-PP, with the increasing complexity of the substitution pattern ensuring the correct stereochemistry and suppression of termination processes. Polymerisation with C2 -symmetric catalysts proceeds as depicted in Scheme 8.39, path A. By contrast, the Cssymmetric ligand environment of 88a enforces consecutive approaches of the monomer with opposite p-faces and hence syndiotacticity [125]. C2 - and Cs-symmetry are, however, not in themselves necessary prerequisites for stereo-control; for example, introducing a methyl substituent, as in 88b, disturbs the selectivity of every second insertion step to give a polymer with approximate hemiisotactic structure, while 88c (R ¼ But) generates iso-rich polymers. Calculations show that the most stable combination of re- or si-mode of alkene attachment and ligand structure determines the preferred transition state [126]. An analysis of polymer end groups provided insight into the mechanism of stereo-control in such catalysts. The first polymerisation step, where propene inserts into a Zr–Me bond, is in fact not stereoselective, while the insertion into a Zr–iso-butyl bond proceeds with high enantioselectivity. Ligand stereo-control operates therefore by an indirect mechanism: the ligand determines the conformation of the polymeryl chain, and this in turn influences the preferred orientation of the incoming alkene [127], as illustrated in structure 89 for a syndiospecific case. The stereo-regularity of a polymer can be reduced by a variety of defects. For example, regio-errors arise when a 2,1-monomer insertion occurs instead of a 1,2-insertion; the latter is usually 102 –103 times faster. Misinsertions can lead to chain termination and formation of butenyl end groups, they can be incorporated into


ZrCl2

ZrCl2

84

85

Me2Si

ZrCl2

Me2Si

349

ZrCl2

87

86

R P ZrCl2

Me2C

X Me2E

+ Zr

Zr Me

89

Me

90, E = C, Si

88a, R = H 88b, R = Me 88c, R = But

the growing polymer chain as 2,1-errors, or they can isomerise before further monomer insertion takes place to give 1,3-enchainments (Scheme 8.41). A list of 13C NMR-detectable end groups in polypropylene is shown in Scheme 8.42 [122]. In addition, internal unsaturations may be found; these arise most probably via C–H activation and zirconium allyls. A mechanism for their formation is shown in Scheme 8.43; this also explains the formation of H2 in such polymerisation systems [128].

Zr

P

2,1-insertion

P

Zr

1,2-insertion

P

Zr 2,1-regioerror

isomerisation P

Zr H

Zr

P 1,2-insertion P

Zr Scheme 8.41

1,3-enchainment


350

P

Zr

P

isomerisation

P

vinylidene

isobutenyl

P

Zr

P

P

Zr

P

Zr

cis-2-butenyl

b-methyl transfer

+ Me AlR2

P

Zr

Me

+

allyl

P

R2Al

H+

P

isobutyl P Zr

Me

n-propyl

Zr Scheme 8.42

P Zr

P Zr

H H

Zr - H2

P

Zr

P

Scheme 8.43

Other sources of stereo-errors in polymers are site epimerisation (chain skipping, cf. Scheme 8.24) and chain end epimerisation, as illustrated in Scheme 8.31. Since chain epimerisation involves processes that are orders of magnitude slower than the rate of chain propagation, it is most prevalent at low monomer concentrations where propagation is slow. This process has been the subject of several mechanistic studies. Busico demonstrated that with falling monomer concentration the stereoselectivity of C2 -symmetric ansa-zirconoces decreases sharply, due to slow epimerisation involving a tert-alkylzirconium intermediate, Zr–CMe2 –P [129]. This mechanism has been supported by a number of isotopic labelling studies. Brintzinger et al. used the polymerisation of E- and Z-CHD¼CHMe with 85/ MAO to determine the rate of chain end epimerisation by observing the deuterium incorporation into the methyl side chains (Scheme 8.44). They showed that in sterically more constricted catalysts, like 87, this isomerisation process is unable to compete with chain propagation, hence the high iso-specificity of these catalysts


D or

Me L2ZrCl2 / MAO

CH3 Zr

H H

H Zr

P H

D

Zr

H CH3 P

D

Me

H

D CH2D

CH2D Zr

P CH3

351

H

Zr H

P H

Scheme 8.44

[130]. Bercaw and Yoder obtained similar results using doubly labelled CH2¼CD–13CH3 in polymerisations with 84 and 85 at 50–75 hC and low propene concentrations. There was b-D elimination and olefin rotation, leading to migration of the 13C label from the methyl substituent into the main chain, but no evidence for this process involving a p-allyl intermediate [131]. The data discussed in Sections 8.5 and 8.6 make it clear that in the low-dielectric media typically employed for polymerisation reactions, the counteranions in metallocene ion pair catalysts are closely associated with the cationic complex as either inner-sphere or outer-sphere ligands. If anions are coordinated in the transition state, they must be expected to exert a significant influence on the stereochemistry of alkene polymerisation, even though the formation of syndiotactic and isotactic 1-alkenes have been readily explained by considering only the cationic metallocenium species and their ligand structure [21, 23, 122, 132, 133]. There is no detectable effect of the counteranion on polypropenes produced with C2 -symmetric ansa-zirconocenes [134]. For the Cs-symmetric complex 88a, on the other hand, there is a pronounced increase in stereoselectivity with tighter ion pairing. In early work, Fink et al. [135] showed that the activity of MAO-activated 88a increased linearly with the dichloromethane content of the solvent, i. e. with the dielectric constant (e ¼ 8.9); at the same time the intensities of the rrrr pentads decreased sharply, while the rmrr pentads increased. This was interpreted in terms of the formation of solvated ion pairs at high dichoromethane concentration, which would allow more facile site epimerisation (“chain swinging”, cf. Scheme 8.24) than tight ion pairs. Similar observations were made by Deffieux et al. for the 1-hexene polymerisation with 88a/MAO [136]. The same effect was found by Busico et al. in the 88a/AlBui3/[HNMe2Ph][B(C6F5)4] system; these authors estimated that with an insertion activation energy in toluene of 10–15 kcal mol–1, the DG‡ for site epimerisation z 20–25 kcal mol–1, with both values reduced by 5–10 kcal mol–1 in the more polar solvent bromobenzene [137]. Marks et al. studied a family of derivatives of 88a with different anions, i. e. [Me2C(Cp)(Flu)Zr(Me) p p p X] (X ¼ [MeB(C6F5)3]–, [MeB(2-C6F5C6F4)3]–, [B(C6F5)4]– and [FAl(2-C6F5C6F4)3]–), using EXSY and dynamic NMR [138]. They found that [MeB(C6F5)3]– is considerably more mobile than [FAl(2-C6F5C6F4)3]–; the latter

P CH3

D

352


X

X

X Pn

Pn

X Pn

Scheme 8.45

gives the least active catalyst but also produces the highest syndiospecificity in toluene. On the other hand, in 1,3-dichlorobenzene the stereoselectivity of all catalysts collapsed. Clearly anion coordination suppresses the formation of stereo-defects due to fast site epimerisation (chain swinging). Remarkably, there is essentially no anion effect with catalyst 90 where inversion (chain swinging) is always much faster than propagation [139]. It is clear therefore that close proximity of the anion is required to suppress chain swinging and to ensure stereo-regular chain propagation, at least in syndiospecific catalysts. Equally, syndiospecific propagation can only be explained if the monomer does not attack from the side opposite the anion, the electrostatically favored route, but from a position cis to the anion (Scheme 8.45). It seems therefore that the highly stereoselective sequence: anion substitution–insertion/chain migration–anion association to the opposite site requires a remarkable degree of molecular acrobatics.

8.10

Conclusion

Over the last 20 years NMR methods have provided unprecedented insight into the workings of metallocene polymerisation catalysts. While 13C NMR spectroscopy has for some time been well established in determining the microstructure of 1-alkene polymers, high field NMR has now achieved a degree of resolution that allows the identification and statistical analysis of structural blocks some 13 carbon atoms in length and sometimes longer. NMR studies on catalyst systems have refined mechanistic thinking and moved from simple ion models to ion pairs and the detection of higher ion aggregates, the preferred relative anion and cation positions in solution, and measurements of the size of such ill-defined anions as those generated from MAO. These studies combine to give a vivid illustration of the variety of factors that enable transition state energies to be lowered by those crucial 1–2 kcal mol–1 that make the difference between a poor and a highly successful catalyst. Such understanding will lay the foundations for rational catalyst design in the future.

Pn+1

References

References [1] (a) K. Ziegler, H. Breil, E. Holzkamp,

[2]

[3]

[4]

[5]

[6]

[7] [8] [9] [10] [11] [12]

[13]

[14]

H. Martin, German Pat. DBP973626 (18.11.1953); (b) K. Ziegler, E. Holzkamp, H. Breil, H. Martin, Angew. Chem. 1955, 67, 426; (c) K. Ziegler, E. Holzkamp, H. Breil, H. Martin, Angew. Chem. 1955, 67, 541. (a) G. Natta, P. Pino, E. Mantica, F. Danusso, G. Mazzanti, M. Peraldo, Chim. Ind. 1956, 38, 124; (b) G. Natta, Angew. Chem. 1956, 68, 393. (a) K. Ziegler, Angew. Chem. 1964, 76, 545; (b) G. Natta, Angew. Chem. 1964, 76, 553. (a) G. Wilkinson, P. L. Pauson, J. M. Birmingham, F. A. Cotton, J. Am. Chem. Soc. 1953, 75, 1011; (b) G. Wilkinson, J. M. Birmingham, J. Am. Chem. Soc. 1954, 76, 4281. (a) G. Natta, P. Pino, G. Mazzanti, U. Giannini, J. Am. Chem. Soc. 1957, 79, 2975; (b) G. Natta, P. Pino, G. Mazzanti, U. Giannini, E. Mantica, M. Peraldo, J. Polym. Sci. 1957, 26, 120. (a) D. S. Breslow, US Pat. 537039 (1955); (b) D. S. Breslow, N. R. Newburg, J. Am. Chem. Soc. 1957, 79, 5072. W. P. Long, D. S. Breslow, J. Am. Chem. Soc. 1960, 82, 1953. D. S. Breslow, N. R. Newburg, J. Am. Chem. Soc. 1959, 81, 81. W. P. Long, J. Am. Chem. Soc. 1959, 81, 5312. J. C. W. Chien, J. Am. Chem. Soc. 1959, 81, 86. G. Henrici-Oliv and S. Oliv, Angew. Chem. Int. Ed. Engl. 1967, 6, 790. (a) A. K. Zefirova, A. E. Shilov, Dokl. Akad. Nauk. SSSR 1961, 136, 599; (b) F. S. Dyachkovskii, A. K. Shilova, A. E. Shilov, J. Polym. Sci. Part C 1967, 16, 2333. R. F. Jordan, C. S. Bajgur, R. Willet, B. Scout, J. Am. Chem. Soc. 1986, 108, 7410. M. Bochmann, L. M. Wilson, J. Chem. Soc., Chem. Commun. 1986, 1610.

[15] (a) M. Bochmann, L. M. Wilson,

[16] [17] [18]

[19]

[20] [21]

[22] [23] [24]

[25] [26]

[27] [28] [29]

[30]

M. B. Hursthouse, R. L. Short, Organometallics 1987, 6, 2556; (b) M. Bochmann, A. J. Jaggar, L. M. Wilson, M. B. Hursthouse, M. Motevalli, Polyhedron 1989, 8, 1838. R. F. Jordan, Adv. Organomet. Chem. 1991, 32, 325. H. Sinn, W. Kaminsky, Adv. Organomet. Chem. 1980, 18, 99. H. Schutenhaus, H. H. Brintzinger, Angew. Chem. Int. Ed. Engl. 1979, 18, 777. F. R. W. P. Wild, L. Zsolnai, G. Huttner, H. H. Brintzinger, J. Organomet. Chem. 1982, 232, 233. P. C. Mhring, N. J. Coville, J. Organomet. Chem. 1994, 479, 1. H. H. Brintzinger, D. Fischer, R. Mlhaupt, B. Rieger, R. M. Waymouth, Angew. Chem. Int. Ed. Engl. 1995, 34, 1143. R. H. Grubbs, G. W. Coates, Acc. Chem. Res. 1996, 29, 85. M. Bochmann, J. Chem. Soc., Dalton Trans. 1996, 255. (a) W. Kaminsky, A. Hopf, M. ArndtRosenau, Macromol. Symp. 2003, 201, 301; (b) W. Kaminsky, A. Laban, Appl. Catal. 2001, 222, 47; (b) W. Kaminsky, J. Chem. Soc., Dalton Trans. 1998, 1413. (c) W. Kaminsky, M. Arndt, Adv. Polym. Sci. 1997, 127, 144. A. L. McKnight, R. M. Waymouth, Chem. Rev. 1998, 98, 2587. Metalorganic Catalysts for Synthesis and Polymerization: Recent Results by Ziegler-Natta and Metallocene Investigations, Ed. W. Kaminsky, Springer, Berlin 1999. T. J. Marks, J. C. Stevens (eds.), Top. Catal. 1999, 7, 1. J. A Gladysz (ed.), Chem Rev. 2000, 100, 1167. J. J. Eisch, A. M. Piotrovski, S. K. Brownstein, E. J. Gabe, F. L. Lee, J. Am. Chem. Soc. 1985, 107, 7219. M. Bochmann, A. J. Jaggar, J. C. Nicholls, Angew. Chem. Int. Ed. Engl. 1990, 29, 780.

353

354

8 The Use of Spectroscopy in Metallocene-based Polymerisation Catalysis [31] R. Mynott, G. Fink, W. Fenzl, Angew. [32] [33]

[34] [35]

[36] [37]

[38]

[39] [40]

[41] [42]

[43] [44]

[45]

[46]

Makromol. Chem. 1987, 154, 1. M. Bochmann, S. J. Lancaster, J. Organomet. Chem. 1992, 434, C1. (a) A. G. Massey, A. J. Park, F. G. A. Stone, Proc. Chem. Soc., London 1963, 212; (b) A. G. Massey, A. J. Park, J. Organomet. Chem. 1964, 2, 245; (c) J. L. W. Pohlmann, F. E. Brinckmann, Z. Naturforsch. B, 1965, 20, 5. H. W. Turner, Eur. Pat. 277004, 1988. (a) X. Yang, C. L. Stern, T. J. Marks, J. Am. Chem. Soc. 1991, 113, 3623; (b) J. A. Ewen, M. J. Elder, Eur. Pat. 427696 (1991). X. Yang, C. L. Stern, T. J. Marks, J. Am. Chem. Soc. 1994, 116, 10015. M. Bochmann, S. J. Lancaster, M. B. Hursthouse, K. M.A. Malik, Organometallics 1994, 13, 2235. (a) Y. X. Chen, C. L. Stern, S. Yang and T. J. Marks, J. Am. Chem. Soc. 1996, 118, 12451; (b) Y. X. Chen, M. V.; Metz, L. Li, C. L. Stern, T. J. Marks, J. Am. Chem. Soc. 1998, 120, 6287. M. Bochmann, S. J. Lancaster, Angew. Chem. Int. Ed. Engl. 1994, 33, 1634. A. D. Horton, J. de With, A. J. van der Linden, H. van de Weg, Organometallics 1996, 15, 2672. X. Song, M. Thornton-Pett, M. Bochmann, Organometallics 1998, 17, 1004. J. Zhou, S. J. Lancaster, D. A. Walker, S. Beck, M. Thornton-Pett, M. Bochmann, J. Am. Chem. Soc. 2001, 123, 223. M. Bochmann, S. J. Lancaster, J. Organomet. Chem. 1995, 497, 55. (a) S. Beck, M. H. Prosenc, H. H. Brintzinger, R. Goretzki, N. Herfert, G. Fink, J. Mol. Catal. A: Chemical 1996, 111, 67; (b) T. Haselwander, S. Beck, H. H. Brintzinger, in Ziegler Catalysts, Eds. G. Fink, R. Mlhaupt, H. H. Brintzinger, Springer, Berlin 1995, p. 181. M. Bochmann, O. B. Robinson, S. J. Lancaster, M. B. Hursthouse, S. J. Coles, Organometallics, 1995, 14, 2456. (a) D. J. Gillis, R. Quyoum, M. J. Tudoret, Q. Wang, D. Jeremic, A. W. Roszak, M. C. Baird, Organo-

[47]

[48]

[49]

[50]

[51]

[52]

[53] [54]

[55]

metallics 1996, 15, 3600; (b) D. J. Gillis, M. J. Tudoret, M. C. Baird, J. Am. Chem. Soc. 1993, 115, 2543. J. Saßmannshausen, M. Bochmann, J. Rsch, D. Lilge, J. Organomet. Chem. 1997, 548, 23. (a) M. C. Baird, Can. J. Chem. 2003, 81, 330; (b) S. W. Ewart, M. C. Baird, Topics in Catalysis 1999, 7, 1. L. Jia, X. Yang, C. L. Stern and T. J. Marks, Organometallics 1997, 16, 842. (a) J. C. Flores, J. S. Wood, J. C. W. Chien, M. D. Rausch, Organometallics 1996, 15, 4944; (b) J. Saßmannshausen, A. K. Powell, C. E. Anson, S. Wocadlo, M. Bochmann, J. Organomet. Chem. 1999, 592, 84. (a) M. Bochmann, M. L. H. Green, A. K. Powell, J. Saßmannshausen, M. U. Triller, S. Wocadlo, J. Chem. Soc., Dalton Trans. 1999, 43; (b) L. H. Doerrer, M. L. H. Green, D. Hußinger, J. Saßmannshausen, J. Chem. Soc., Dalton Trans. 1999, 2111; (c) M. L. H. Green, J. Saßmannshausen, Chem. Commun. 1999, 115. (a) P. J. W. Deckers, B. Hessen, J. H. Teuben, Angew. Chem. Int. Ed. 2001, 40, 2516; (b) P. J. W. Deckers, B. Hessen, J. H. Teuben, Organometallics 2002, 21, 5122. G. Erker, Acc. Chem. Res. 2001, 34, 309. (a) M. Dahlmann, G. Erker, K. Bergander, J. Am. Chem. Soc. 2000, 122, 7986; (b) J. Karl, M. Dahlmann, G. Erker, K. Bergander, J. Am. Chem. Soc. 1998, 120, 5643. (a) G. Jimnez-Pindado, M. Thornton-Pett, M. Bouwkamp, A. Meetsma, B. Hessen, M. Bochmann, Angew. Chem. Int. Ed. Engl. 1997, 36, 2358; (b) G. Jimnez-Pindado, M. Thornton-Pett, M. Bochmann, J. Chem. Soc., Dalton Trans. 1997, 3115; (c) G. Jimnez Pindado, S. J. Lancaster, M. Thornton-Pett, M. Bochmann, J. Am. Chem. Soc. 1998, 120, 6816; (d) M. M. Corradi, G. Jimenez-Pindado, M. J. Sarsfield, M. ThorntonPett and M. Bochmann, Organometallics 2000, 19, 1150; (e) A. H. Cow-

References

[56]

[57] [58]

[59] [60]

[61]

[62]

[63]

[64]

[65]

[66]

[67] [68]

ley, G. S. Hair, B. G. McBurnett, R. A. Jones, Chem. Commun. 1999, 437. T. J. Woodman, M. Thornton-Pett, M. Bochmann, Chem. Commun. 2001, 329; T. J. Woodman, M. ThorntonPett, M. Bochmann, Organometallics 2001, 20, 4080. X. Song and M. Bochmann, J. Organomet. Chem. 1997, 545-546, 597. (a) Y. Sun, R. E. v. H. Spence, W. E. Piers, M. Parvez and G. P. A. Yap, J. Am. Chem. Soc. 1997, 119, 5132; (b) Y. M. Sun, W. E. Piers, S. J. Rettig, Chem. Commun. 1998, 127; (c) L. M. W. Lee, W. E. Piers, M. Parvez, S. J. Rettig, V. G. Young, Organometallics 1999, 18, 3904. Review: W. E. Piers, Chem. Eur. J. 1998, 4, 13. H. van der Heijden, B. Hessen, A. G. Orpen, J. Am. Chem. Soc. 1998, 120, 1112. (a) M. Bochmann, S. J. Lancaster, Organometallics 1993, 12, 633; (b) R. F. Jordan, R. E. LaPointe, C. S. Bajgur, S. F. Echols, R. Willet, J. Am. Chem. Soc. 1987, 109, 4111. (a) C. Pellecchia, A. Grassi, A. Immirzi, J. Am. Chem. Soc. 1993, 115, 1160; (b) M. Bochmann, G. Karger, A. J. Jaggar, J. Chem. Soc., Chem. Commun. 1990, 1038. C. Pellecchia, A. Immirzi, A. Grassi, A. Zambelli, Organometallics 1993, 12, 4473. See for example P. J. Sinnema, K. Liekelema, O. K. B. Staal, B. Hessen, J. H. Teuben, J. Mol. Catal. 1998, 128, 143. (a) C. Pellecchia, A. Immirzi, D. Pappalardo, A. Peluso, Organometallics 1994, 13, 3773; (b) J. I. Amor, T. Cuenca, M. Galakhov, P. Royo, J. Organomet. Chem. 1995, 497, 127. G. Jimnez Pindado, M. ThorntonPett, M. B. Hursthouse, S. J. Coles, M. Bochmann, J. Chem. Soc., Dalton Trans. 1999, 1663. A. D. Horton, Organometallics 1996, 15, 2675. G. X. Wu, Macromolecules 1998, 31, 586.

[69] (a) C. L. Beswick, T. J. Marks,

[70] [71]

[72] [73]

[74] [75] [76]

[77]

[78]

[79]

[80]

[81]

[82]

[83]

Organometallics 1999, 18, 2410; (b) C. L. Beswick, T. J. Marks, J. Am. Chem. Soc. 2000, 122, 10358. F. Song, R. D. Cannon and M. Bochmann, Chem. Commun. 2004, 542. S. Garratt, A. G. Carr, G. Langstein, M. Bochmann, Macromolecules 2003, 36, 4276. A. D. Horton, A. G. Orpen, Organometallics 1992, 11, 8. P. Binger, P. Mller, R. Benn, A. Rufinska, B. Gabor, C. Krger, P. Betz, Chem. Ber. 1989, 122, 1035. P. T. Witte, A. Meetsma, B. Hessen, J. Am. Chem. Soc. 1997, 119, 10561. M. V. Galakhov, G. Heinz, P. Royo, Chem. Commun. 1998, 17. (a) Z. Wu, R. F. Jordan, J. L. Petersen, J. Am. Chem. Soc. 1995, 117, 5867. (b) J. F. Carpentier, Z. Wu, C. W. Lee, S. Strmberg, J. N. Christopher, R. F. Jordan, J. Am. Chem. Soc. 2000, 122, 7750. (a) C. P. Casey, D. W. Carpenetti, Organometallics 2000, 19, 3970; (b) C. P. Casey, D. W. Carpenetti, H. Sakurai, Organometallics 2001, 20, 4262. C. G. Brandow, A. Mendiratta, J. E. Bercaw, Organometallics 2001, 20, 4253. C. P. Casey, J. Fisher Klein, M. A. Fagan, J. Am. Chem. Soc. 2000, 122, 4320. (a) C. P. Casey, J. A. Tunge, T. Y. Lee, M. A. Fagan, J. Am. Chem. Soc. 2003, 125, 2641; (b) C. P. Casey, T. Y. Lee, J. A. Tunge, D. W. Carpenetti, J. Am. Chem. Soc. 2001, 123, 10762. (a) P. A. Deck and T. J. Marks, J. Am. Chem. Soc. 1995, 117, 6128; (b) P. A. Deck, C. L. Beswick, T. J. Marks, J. Am. Chem. Soc. 1998, 120, 1772. (a) A. R. Siedle and R. A. Newmark, J. Organomet. Chem. 1995, 497, 119; (b) A. R. Siedle, B. Hanggi, R. A. Newmark, K. R. Mann and T. Wilson, Macromol. Symp. 1995, 89, 299. E. Y. X. Chen, T. J. Marks, Chem. Rev. 2000, 100, 1391.

355

356

8 The Use of Spectroscopy in Metallocene-based Polymerisation Catalysis [84] S. Beck, S. Lieber, F. Schaper,

[85] [86]

[87]

[88]

[89]

[90] [91] [92]

[93]

[94]

[95] [96]

[97]

[98]

[99]

A. Geyer, H. H. Brintzinger, J. Am. Chem. Soc. 2001, 123, 1483. S. Beck, A. Geyer, H. H. Brintzinger, Chem. Commun. 1999, 2477. B. Binotti, A. Macchioni, C. Zuccaccia, D. Zuccaccia, Comments Inorg. Chem. 2002, 6, 417. N. G. Stahl, C. Zuccaccia, T. R. Jensen, T. J. Marks, J. Am. Chem. Soc. 2003, 125, 5256. C. Zuccaccia, N. G. Stahl, A. Macchioni, M. C. Chen, J. A. Roberts, T. J. Marks, J. Am. Chem. Soc. 2004, 126, 1448. F. Song, S. J. Lancaster, C. Zuccaccia, A. Macchioni, M. Bochmann, results to be published. F. Schaper, A. Geyer, H. H. Brintzinger, Organometallics 2002, 21, 473. K. Vanka, T. Ziegler, Organometallics 2001, 20, 905. (a) G. Lanza and I. Fragal, Top. Catal. 1999, 7, 45; (b) G. Lanza, I. L. Fragal, T. J. Marks, Organometallics 2002, 21, 5594. I. E. Nifant’ev, L. Y. Ustynyuk, D. N. Laikov, Organometallics 2001, 20, 5375. (a) G. Fink, R. Rottler, Angew. Makromol. Chem. 1981, 94, 25; (b) G. Fink, R. Rottler, C. G. Kreiter, Angew. Makromol. Chem. 1981, 96, 1. G. Fink, W. Fenzl and R. Mynott, Z. Naturforsch. B, 1985, 40, 158. (a) Z. Liu, E. Somsook, C. R. Landis, J. Am. Chem. Soc. 2001, 123, 2915; (b) Z. Liu, E. Somsook, C. B. White, K. A. Rosaaen, C. R. Landis, J. Am. Chem. Soc. 2001, 123, 11193; (c) D. R. Sillars, C. R. Landis, J. Am. Chem. Soc. 2003, 125, 9894. F. Song, R. D. Cannon, M. Bochmann, J. Am. Chem. Soc. 2003, 125, 7641. C. R. Landis, K. A. Rosaaen, D. R. Sillars, J. Am. Chem. Soc. 2003, 125, 1710. D. A. Singleton, A, A, Thomas, J. Am. Chem. Soc. 1995, 117, 9357; (b) D. E. Frantz, D. A. Singleton, J. P. Snyder, J. Am. Chem. Soc. 1997, 119, 3383.

[100] C. R. Landis, K. A. Rosaaen, J. Uddin,

J. Am. Chem. Soc. 2002, 124, 12062. [101] C. J. Harlan, S. G. Bott, A. R. Barron,

J. Am. Chem. Soc. 1995, 117, 6465. [102] (a) I. Tritto, M. Sacchi, S. Li, Macro-

[103] [104]

[105]

[106]

[107]

[108] [109]

[110]

[111]

[112]

mol. Rapid Commun. 1994, 15, 217; (b) I. Tritto, M. Sacchi, P. Locatelli, S. Li, Macromol. Symp. 1995, 89, 289. K. P. Bryliakov, E. P. Talsi, M. Bochmann, Organometallics 2004, 23, 149. I. Tritto, R. Donetti, M. C. Sacchi, P. Locatelli, G. Zannoni, Macromolecules 1997, 30, 1247. A. R. Siedle, R. A. Newmark, W. M. Lamanna, J. N. Schroepfer, Polyhedron 1990, 9, 301. D. E. Babushkin, N. V. Semikolenova, V. A. Zakharov, E. P. Talsi, Macromol. Chem. Phys. 2000, 201, 558. (a) K. P. Bryliakov, N. V. Semikolenova, D. V. Yudaev, V. A. Zakharov, H. H. Brintzinger, M. Ystenes, E. Rytter, E. P. Talsi, J. Organomet. Chem. 2003, 683, 92; (b) K. P. Bryliakov, N. V. Semikolenova, D. V. Yudaev, M. Ystenes, E. Rytter, V. A. Zakharov, E. P. Talsi, Macromol. Chem. Phys. 2003, 204, 1110; (c) E. P. Talsi, K. P. Bryliakov, N. V. Semekolenova, V. A. Zakharov, Mendeleev. Commun. 2003, 48. D. E. Babushkin, H. H. Brintzinger, J. Am. Chem. Soc. 2002, 124, 12869. K. P. Bryliakov, N. V. Semikolenova, V. A. Zakharov, E. P. Talsi, J. Organomet. Chem. 2003, 683, 23. (a) S. W. Ewart, M. J. Sarsfield, D. Jeremic, T. L. Tremblay, E. F. Williams, M. C. Baird, Organometallics 1998, 17, 1502; (b) S. W. Ewart, M. J. Sarsfield, E. F. Williams, M. C. Baird, J. Organomet. Chem. 1999, 579, 106; (c) E. F. Williams, M. C. Murray, M. C. Baird, Macromolecules 2000, 33, 261. (a) A. Grassi, A. Zambelli, F. Laschi, Organometallics 1996, 15, 480; (b) A. Grassi, S. Saccheo, A. Zambelli, Macromolecules 1998, 31, 5588; (c) C. Pellecchia, A. Grassi, Topics in Cataysis 1999, 7, 125. E. J. Hawrelak, P. A. Deck, Organometallics 2003, 22, 3558.

References [113] (a) J. C. W. Chien, B. Xu, Makromol.

[114]

[115]

[116]

[117] [118] [119]

[120]

[121]

[122]

[123] [124]

Chem. Rapid Commun. 1993, 14, 109; (b) J. C. W. Chien, W. M. Tsai, Makromol. Chem., Macromol. Symp. 1993, 66, 141; (c) Y. X. Chen, M. D. Rausch, J. C. W. Chien, Organometallics 1994, 13, 748. S. J. Lancaster, D. A. Walker, M. Thornton-Pett and M. Bochmann, Chem. Commun. 1999, 1533. A. G. Carr, D. M. Dawson, M. Thornton-Pett, M. Bochmann, Organometallics 1999, 18, 2933. (a) J. H. Teuben, A. Meetsma, Y. T. Yong, J. J. W. Eshuis, J. Renkema, G. G. Evans, Organometallics 1992, 22, 362; (b) H. H. Brintzinger, S. Lieber, M. H. Prosenc, Organometallics 2000, 19, 377; (c) G. Moscardi, L. Resconi, Organometallics 2001, 20, 1918. C. Gtz, A. Rau, G. Luft, J. Mol. Catal. A: Chemical 2002, 184, 95. P. C. Wailes, H. Weigold, A. P. Bell, J. Organomet. Chem. 1972, 43, C29. K. P. Bryliakov, S. J. Lancaster, F. Song, M. Bochmann, E. P. Talsi, to be published. I. Tritto, R. Donetti, M. C. Sacchi, P. Locatelli, G. Zannoni, Macromolecules 1999, 32, 264. See for example: (a) J. C. Randall, Polymer Sequence Determination: 13C NMR Method, Academic Press, New York 1977, (b) F. A. Bovey, Chain Structure and Conformation of Macromolecules, Academic Press New York, 1982; (c) A. E. Tonelli, NMR Spectroscopy and Polymer Microstructure: The Conformational Connection, VCH, Deers Field 1989. L. Resconi, L. Cavallo, A. Fait, F. Piemontesi, Chem. Rev. 2000, 100, 1253. V. Busico, R. Cipullo, Progr. Polym. Sci. 2001, 26, 443. J. A. Ewen, J. Am. Chem. Soc. 1984, 106, 6355.

[125] J. A. Ewen, R. L. Jones, A. Razavi,

[126]

[127]

[128] [129] [130] [131] [132] [133] [134]

[135] [136]

[137]

[138]

[139]

J. Ferrara, J. Am. Chem. Soc. 1988, 110, 6255. (a) Y. van der Leek, K. Angermund, M. Reffke, R. Kleinschmidt, R. Goretzki, G. Fink, Chem. Eur. J. 1997, 3, 585; (b) L. Cavallo, G. Guerra, M. Vacatello, P. Corradini, Macromolecules 1991, 24, 1784. (a) P. Longo, A. Grassi, C. Pellecchia, A. Zambelli, Macromolecules 1987, 20, 1015; (b) M. Dahlmann, G. Erker, M. Nissinen, R. Frhlich, J. Am. Chem. Soc. 1999, 121, 2820. P. M. Margl, T. K. Woo, T. Ziegler, Organometallics 1998, 100, 307. V. Busico, R. Cipullo, J. Am. Chem. Soc. 1994, 116, 9329. M. K. Leclerk, H. H. Brintzinger, J. Am. Chem. Soc. 1996, 118, 9024. J. C. Yoder, J. E. Bercaw, J. Am. Chem. Soc. 2002, 124, 2548. G. W. Coates, Chem. Rev. 2000, 100, 1223. A. K. Rapp, W. M. Skiff, C. J. Casewit, Chem. Rev. 2000, 100, 1435. (a) S. Hahn, G. Fink, Macromol. Rapid Commun. 1997, 18, 117; (b) A. Rodriguez-Delgado, M. D. Hannant, S. J. Lancaster, M. Bochmann, Macromol. Chem. Phys. 2004, 205, 334. N. Herfert, G. Fink, Macromol. Chem. 1992, 193, 773. D. Coevoet, H. Cramail, A. Deffieux, Macromol. Chem. Phys. 1999, 200, 1208. V. Busico, R. Cipullo, F. Cutillo, M. Vacatello, V. van Axel Castelli, Macromolecules 2003, 36, 4258. (a) M. C. Chen, J. A. S. Roberts, T. J. Marks, J. Am. Chem. Soc. 2004, 126, 4605; (b) M. C. Chen, T. J. Marks, J. Am. Chem. Soc. 2001, 123, 11803. M. Mohammed, M. Nele, A. Al-Humydi, S. Xin, R. A. Stapleton, S. Collins, J. Am. Chem. Soc. 2003, 125, 7930.

357

9 Hydrogenation Ralf Giernoth

9.1

Introduction

Hydrogenation is one of the most basic chemical transformations that one can think of: the smallest molecule in existence, the dihydrogen molecule, is transferred to the double or triple bond of a second molecule (Figure 9.1). The beauty of this is that this reaction is extremely atom economic and it is possible, provided that the “right” catalyst system is at hand, to run this reaction in a highly enantiopure fashion. Obviously, olefins and acetylenes are among the basic building blocks for the chemical industry and therefore hydrogenation is one of the most important industrial reactions. In particular, the enantioselective hydrogenation of prochiral enamines, leading to chiral amino acid derivatives, is extremely valuable for the synthesis of pharmaceuticals, agrochemicals and natural products. The focus of this chapter will be on the rhodium-catalyzed homogeneous hydrogenation of prochiral enamides, for this reaction is the most widely studied organometallic reaction in terms of mechanism. Of course, a few other examples will be discussed where appropriate, but predominantly for illustration purposes. We do not aim to give a complete and exhaustive review of homogeneous hydrogenation and its detailed mechanisms; rather, we want to focus on the specific spectroscopic techniques and the mechanistic information that has been obtained by applying them. For a thorough overview of the reaction itself, its detailed history, its applications, as well as ruthenium chemistry, please refer to the chemical literature already published [1, 2].

R

R3 +

R2

R4

Figure 9.1

H

H

catalyst

H

H R3

R R2

R4

General scheme for the homogeneous hydrogenation of olefins.


360

9 Hydrogenation

9.2

The Dihydrogen Molecule

The dihydrogen molecule is the smallest molecule in existence. It has a strong covalent bond with a dissociation energy of 103 kcal mol–1 [1]. In a hydrogenation reaction, this bond has to be broken and two new C–H bonds are formed, one of the simplest forms of chemical reaction. But for organic chemists the hydrogen molecule is famous for another reason: 1 H is the second best isotope for NMR spectroscopy in terms of sensitivity. It comes in a natural abundance of 99.989 %, has a spin of I ¼ 1⁄2 and its gyromagnetic ratio of g ¼ 26.752 is only topped by that of tritium (3H, g ¼ 28.535), which is of limited use in routine chemistry because of its radioactivity. Proton NMR spectroscopy has become the standard analytical technique; its information content is invaluable, not only for organic chemists. This dominance might also be a reason why homogeneous hydrogenation today is the most thoroughly studied chemical transformation.

9.3

Rhodium-catalyzed Homogeneous Hydrogenation, the Basics 9.3.1

The Achiral Case

The history of homogeneous hydrogenation with a transition metal catalyst really started in 1966 with the development of Wilkinson’s catalyst (Figure 9.2). This rhodium complex was the first that allowed the controlled reduction of unsaturated carbon–carbon bonds under mild conditions [3]. The mechanism of this remarkable reaction was elucidated first by the groups of Halpern [4] and Tolman [5], and later by Brown [6] and finally Eisenberg [7] with the help of intensive kinetic studies and NMR spectroscopy, especially 31P and PHIP NMR (see below). It is depicted in Figure 9.3. First, one of the three phosphine ligands of the precatalyst (of course, “Wilkinson’s catalyst” is in fact a precatalyst, since the catalytically active species is only generated in solution just prior to catalytic turnover) dissociates from the rhodium center in such a way that the remaining two phosphines are cis to each other. The resulting catalyst bears an “open” sphere to which the substrates (dihydrogen and the unsaturated substrate) can coordinate. The first step in the catalytic cycle is the oxidative cis-addition of dihydrogen to the rhodium center, followed by association of the unsaturated substrate (via its double or triple bond). The reactants are now located close to each

PPh3 Ph3P Rh Cl Ph3P (Wilkinson, 1966 [3])

Figure 9.2 Structure of Wikinson’s catalyst, the first successful homogeneous hydrogenation catalyst.

9.3 Rhodium-catalyzed Homogeneous Hydrogenation, the Basics

PPh3 Ph3P Rh Cl Ph3P fast R Ph3P

R’ Elimination

Rh

PPh3 Cl

H2

cis

Addition

H H Ph3P Rh Ph3P Cl

R

Ph3P Ph3P

R’

H Rh H Cl

H Insertion

Ph3P Ph3P

Rh

H

Association

Cl R’

Figure 9.3

R

R R’

The mechanism of homogeneous hydrogenation using Wilkinson’s catalyst.

other, the dihydrogen bond is already broken and the unsaturated bond is weakened. Next, the unsaturated bond inserts into one of the Rh–H bonds and thus the first hydrogen is transferred to the product molecule. Transfer of the second hydrogen leads to reductive elimination of the product molecule and the catalyst is ready for the next cycle. The general catalytic cycle thus typically consists of the four steps addition, association, insertion, and elimination. 9.3.2

The Chiral Case Catalysts The possibility of asymmetric catalysis in homogeneous hydrogenation was first demonstrated by Horner [8] and Knowles [9] in 1968, for which the latter received the Nobel Prize in 2001 [10] (together with Noyori and Sharpless), and later developed up to a practical point in 1972 by Kagan [11] and again by Knowles [12]. Kagan was able to demonstrate the importance of chiral chelating diphosphine ligands, while Knowles’ group at Monsanto established the first industrial process for the production of L-DOPA. In the meantime, a rather large and multifaceted number of chiral hydrogenation catalysts based on rhodium have entered the chemical literature. In Figure 9.3.2.1

361

362

9 Hydrogenation

O H 3C

CH3

Ph2P

H 3C H 3C

PPh2

(S,S)-CHIRAPHOS (Bosnich, 1977)

O

PPh2

O

PPh2

(R,R)-DIOP (Kagan, 1972)

N

Ph2P

PPh2 PPh2

(S,S)-BPPM Ph P 2 (Achiwa, 1976)

(R)-BINAP (Noyori, 1976)

CH 3

P

P

PPh2 CH 3

OCH3 H 3CO

P

P

CH 3 PPh2

(R,R)-DIPAMP (Knowles, 1972)

CH 3 (S,S)-MeDUPHOS (Burk, 1990)

(R,R)-PHANEPHOS (Pye, 1997)

Figure 9.4 A selection of successful chiral bisphosphine ligands for enantioselective hydrogenation.

9.4 you can see a selection of the more prominent and powerful examples of chiral phosphine ligands which, coordinated to a rhodium center, form homogeneous hydrogenation catalysts. In contrast to Wilkinson’s catalyst, all of these are cationic rather than neutral. Typically, weakly-coordination anions such as tetrafluoroborate, hexafluorophosphate or triflate are used here as the counter-ions.

Hydride Route versus Unsaturate Route In principle, the mechanism of homogeneous hydrogenation, in the chiral as well as in the achiral case, can follow two pathways (Figure 9.5). These involve either dihydrogen addition, followed by olefin association (“hydride route”, as described in detail for Wilkinson’s catalyst, vide supra) or initial association of the olefin to the rhodium center, which is then followed by dihydrogen addition (“unsaturate route”). As a rule of thumb, the “hydride route” is typical for neutral, Wilkinson-type catalysts whereas the catalytic mechanism for cationic complexes containing diphosphine chelate ligands seems to be dominated by the “unsaturate route” [1]. 9.3.2.2

9.4 Spectroscopic Methods

"unsaturate route" L 2Rh+

H

+ H2 L 2Rh+

H2

+

L 2Rh+ H

H L 2Rh+ H

"dihydride route"

H L 2Rh+ H Figure 9.5

The mechanism of homogeneous hydrogenation: “unsaturate route” versus “hydride

route”.

9.4

Spectroscopic Methods 9.4.1

“Standard” NMR Spectroscopy

Obviously (and as already mentioned in Section 9.2), NMR spectroscopy is the most valuable tool for the elucidation of mechanistic aspects in homogeneous hydrogenation reactions. The 1H nucleus is among the most sensitive NMR nuclei, and proton NMR is an absolutely standard technique in every chemistry department. From an NMR point of view, the dihydrogen molecule is an A2 system, i. e. it consists of two identical nuclei with the same chemical shift. The energy scheme for an A2 (i. e. H2) system is depicted in Figure 9.6: outside a magnetic field, the two spins of 1⁄2 couple to four degenerate energy levels. In the strong field of an NMR magnet, these split into three triplet (T–1, T0, Tþ1) states and one singlet (S0) state (this is called the “Zeeman effect”). The energy levels are populated according to the Boltzmann distribution, that is, the population of a lower level is just slightly higher than that of the one above. Irradiation of the system with a radio frequency pulse of matching energy (resonance) gives rise to only two transitions (singlet–triplet transitions are symmetry forbidden) of identical energy. Thus, the NMR spectrum of the dihydrogen molecule, as shown schematically in Figure 9.6, comprises just one single signal. Sing-

363

364

9 Hydrogenation

Figure 9.6 Molecular hydrogen and NMR spectroscopy. (a) Energy levels of an A2 spin system depending on the strength of the magnetic field (Zeeman effect). (b) NMR transitions and resulting NMR spectrum (shown schematically).

let hydrogen (i. e. hydrogen molecules in which the two nuclear spins are antiparallel, vide infra) is “NMR-silent”. In a hydrogenation reaction, the two molecules of the highly symmetrical dihydrogen molecule are transferred into a new chemical environment (the hydrogenated product molecule) and thus the two hydrogen molecules are desymmetrized. For an ideal model case, the original A2 spin system is thus transformed into an AX system (which in NMR terms means that the two hydrogen atoms, HA and HX, are strongly coupled and show a chemical shift difference of at least ten times the corresponding coupling constant [13]). The energy scheme for an AX spin system is shown in Figure 9.7: the original singlet and triplet energy levels of the A2 system (cf. Figure 9.6) are transformed into the linear combinations of the a (–1⁄2) and the b (þ1⁄2) states of the two individual coupled spins. The two middlemost levels (ab and ba), which stem from the original S0 and T0 levels in the A2 case, are now “mixing”, which means that they no longer comprise strict singlet or triplet states. Therefore, now four transitions are possible and allowed, giving rise to an NMR spectrum which consists of two doublets at two different chemical shifts (Figure 9.7). This is nothing else but the simplest case of a standard NMR experiment.


Figure 9.7 The NMR spectrum of an AX spin system (shown schematically). Left: NMR transitions. Right: Resulting NMR spectrum.

9.4.2

PHIP-NMR-Spectroscopy

The PHIP effect was first predicted by Bowers and Weitekamp in 1986 [14] and experimentally proven in the following year [15, 16]. The acronym stands for ParaHydrogen Induced Polarization. PHIP NMR spectroscopy is an in situ method for the elucidation of mechanistic details in homogeneous hydrogenation reactions. It is dynamic, thus allowing one to get insight into hydrogenation reactions at real turnover conditions. The NMR signals of the hydrogen molecule are enhanced significantly (due to a polarization effect, vide infra) which allows the detection and characterization of reaction intermediates that are present in very low concentration. The theoretical background for the PHIP effect is discussed below.

Ortho- and Parahydrogen As already mentioned, molecular hydrogen consists of two spin isomers, one of which has a total spin of I¼0 (singlet), and the other a total spin of I¼1 (triplet). The first is named parahydrogen, the latter orthohydrogen (Figure 9.8). 9.4.2.1

Figure 9.8 The distribution of ortho- and para-hydrogen depending on the temperature. As one can see, at room temperature molecular hydrogen consists of 25 % para and 75 % ortho.

365

366

9 Hydrogenation

For PHIP NMR spectroscopy, parahydrogen (or at least para-enriched hydrogen) is needed. To prepare para-enriched hydrogen, several techniques are known. According to one concept, it is necessary to separate the two atoms of the dihydrogen molecule, allowing them to recombine thereafter at low temperature. The easiest way to achieve this is to pass hydrogen through a pipe filled with activated charcoal, which acts as the required catalyst. The tube containing the charcoal can be cooled, for example, to the temperature of liquid nitrogen (77 K). In this fashion, a constant stream of hydrogen can be obtained that is “para-enriched” such that it contains about 50 % of each spin isomer [17] (cf. Figure 9.8). This enrichment results from the fact that the singlet and triplet states of the dihydrogen molecule are directly coupled to the corresponding rotational states. Due to its symmetry, the energetically lowest rotational state (0) is reserved for parahydrogen; therefore, this spin isomer becomes enriched at low temperature.

Parahydrogen Induced Polarization and In Situ Spectroscopy When breaking the symmetry of the parahydrogen molecule via a chemical reaction, for example, by hydrogenating an unsaturated substrate, this process converts the initial A2 -type spin system of parahydrogen into an AX-spin system in the resulting hydrogenation product. Correspondingly, the original S and T0 states of the dihydrogen mix in such a fashion that in the final hydrogenation product the energy levels associated with the former parahydrogen nuclei eventually contain both a triplet and a singlet portion. Therefore, when using exclusively parahydrogen, only these two states become populated, proportional to their individual singlet character. Provided that the NMR spectra are recorded in situ, the corresponding resonances of the hydrogenation products show up in antiphase, that is, the resonances of the former parahydrogen nuclei appear as strongly enhanced signals (typically by a factor of about 1000) whereby an equal number of lines occurs in emission as does in absorption (Figure 9.9). Consequently, in PHIP spectra both doublets and triplets only exhibit two lines, since the central line of the triplet has an intensity of zero. This phenomenon has to be taken into account appropriately when analyzing PHIP spectra, since the size of the J-coupling constants might otherwise easily be misjudged by a factor of two! 9.4.2.2

Figure 9.9 The PHIP-NMR spectrum (PASADENA) of an AX spin system (shown schematically). Left: NMR transitions. Right: Resulting NMR spectrum.


In addition, it is important to note that the phase behavior of PHIP NMR signals differs substantially from that in standard NMR experiments. For normal NMR spectroscopy, a 90h pulse renders the maximum possible signal intensity, while a 180h pulse gives no intensity at all. This is not true for PHIP spectra!! In PHIP NMR, the maximum signal intensity is achieved by a 45h pulse; a 90h pulse is here the zero-crossing (like 180h in standard NMR, yielding minimum intensity). Since in NMR spectroscopy in particular all populations that differ from the conventional Boltzmann distribution are termed “polarization”, this antiphase enhancement effect has previously been named “ParaHydrogen Induced Polarization” (PHIP) [15]. This is, however, strictly speaking incorrect, since according to a traditional convention in magnetic resonance spectroscopy the expression “polarization” is to be reserved for such cases where, following Boltzmann’s law, a spin temperature, albeit even a negative one, can be defined. Typically, for four energy level systems, and higher ones, this is not possible and hence is not the case here. Therefore, in all cases studied here the expression “alignment” should be used or at least preferred, as is evident in the long form of the expression for which the acronym “ALTADENA” stands (vide infra). Nevertheless, since “PHIP” has been coined in the past, and since this acronym is derived from “polarization”, this “misnomer” is used here as well (albeit “illegally” so). The occurrence of polarization signals during homogeneous hydrogenations can be regarded as proof of the transfer of both atoms of the former parahydrogen molecule, that is, they become transferred as a pair or in a “pair-wise” manner. This process does not necessarily have to be simultaneous, but the two atoms of the dihydrogen molecule must retain a spin–spin coupling throughout the whole process. Conversely, if they are not transferred pair-wise (i. e., if the transferred hydrogen atoms stem from different dihydrogen molecules or if they lose their coupling in the course of the process), no polarization can be detected.

ALTADENA and PASADENA In principle, there are two ways of conducting a PHIP experiment: One way is to hydrogenate the premixed sample outside any sizeable magnetic field, that is, outside the spectrometer, and then transport the tube rapidly into the NMR magnet for the subsequent spectroscopic analysis. This type of experiment has been termed “ALTADENA” (Adiabatic Longitudinal Transport After Dissociation Engenders Nuclear Alignment) [18] and is referred to in the following as such. By contrast, the other mode is to hydrogenate the reaction mixture directly inside the NMR magnet. It has become convenient to term this latter mode “PASADENA” (Parahydrogen And Synthesis Allow Dramatically Enhanced Nuclear Alignment), although this acronym originally meant the same as “PHIP” according to its inventors [16]. We nowadays use “PHIP” generically to describe the whole phenomenon, whereas “ALTADENA” and “PASADENA” are used to denote the individual mode via which the actual experiment is conducted, that is, to differentiate, whether a PHIP experiment is carried out inside or outside the magnet of the NMR spectrometer. 9.4.2.3

367

368

9 Hydrogenation

It is in fact important to differentiate between ALTADENA and PASADENA experiments, because the appearance of the resulting NMR spectra differs substantially (see Figure 9.10). In PASADENA experiments, the polarization is, in most cases, limited to the two hydrogen atoms stemming from the parahydrogen molecule. Accordingly, it is quite simple to visualize the pair-wise transferred hydrogen atoms in the resulting product molecule (“magnetic labelling”). In this fashion, it is possible to quickly differentiate between the geometric isomers formed when hydrogenating an alkyne to an alkene, namely whether the hydrogenation yields the E- or the Z-form of the alkene. Likewise, when hydrogenating a multiply unsaturated substrate, it is easy to identify via PHIP experiments which individual double or triple bond is being hydrogenated (including their ratios). Even a mixture of products can thus be analyzed simultaneously within seconds after the hydrogenation has started. In theory, ALTADENA experiments yield spectra as depicted in Figure 9.10. When hydrogenating the samples outside the NMR magnet, however, only that very state that has evolved from the former singlet state, namely that of the parahydrogen, should become populated. Nevertheless, due to a multitude of possible processes including the NOE (nuclear Overhauser effect) and isotropic mixing outside the magnet, experimentally obtained ALTADENA spectra are more complex than initially expected. Furthermore, the duration, that is, the width of the detection pulses used for recording the spectra, has a significant influence on their appearance [19]. Their optimal values may well differ from the traditional optimum for recording conventional NMR spectra, depending on the mode via which the PHIP experiment is conducted. In the simplest case, however, just one of the resonance lines of the signal stemming from one hydrogen nucleus shows up in enhanced absorption, while just one line of the other resonance caused by a second hydrogen atom will occur in emission. Likewise, whole resonances may exhibit exclusively emission lines, whereas other ones show up totally in enhanced absorption (“netto effect”). Most of the time, however, the actual polarization phenomenon will also extend to other resonances in the product molecule, making it difficult to decide, which of the hydrogen atoms have been transferred (Figure 9.11). This effect is mainly due to isotropic mixing occurring at zero magnetic field, that is, while the frequently hydrogenated product is still outside the NMR magnet,

Figure 9.10 The theoretical ALTADENA-NMR spectrum of an AX spin system (shown schematically). Left: NMR transitions. Right: Resulting NMR spectrum.


Figure 9.11 The difference between PASADENA- and ALTADENA-PHIP-NMR spectra: hydrogenation of diphenylbutadiene. (a) PASADENA spectrum. (b) ALTADENA spectrum [38].

where all nuclear states are more or less energetically degenerate. In these cases, the application of a computer program especially designed for simulating PHIP spectra [20] can help enormously. It is still rather difficult to exactly reproduce an experimentally obtained ALTADENA spectrum via a simulation. This is because the supplemental processes, which cause a spreading of the polarization across the whole spin system of the product molecule, are difficult to determine or to even guess correctly. The method is, therefore, most suitable and most efficient for cases where qualitative results are sufficient. For all the reasons mentioned above, it is more desirable to obtain PASADENA spectra when applying the PHIP method, at least initially. Nevertheless, in certain cases ALTADENA spectra can be very helpful, too, especially for a quick screening, because these experiments can be performed without any specialized NMR probes. Rather, in these cases, standard NMR equipment will suffice totally. Furthermore, in other cases, the (normally unwanted) isotropic mixing can be used advantageously to obtain information about other additional aspects of the system. A typical such case results when it is desirable to transfer the original proton polar-

369

370

9 Hydrogenation

ization to the hetero nuclei of the hydrogenation product [21]. For further details pertaining to the theory of the PHIP phenomenon see Ref. [19] and references therein.

Applications of the PHIP Method PHIP NMR spectroscopy is applicable to a large variety of problems arising in homogeneous hydrogenation and in seemingly unrelated fields as well. First, it is possible to continuously monitor a reaction from the very beginning to the very end, that is, during the whole conversion or during a complete catalytic cycle, thereby exploring many of the key steps, especially in asymmetric homogeneous catalysis. Figure 9.12 shows schematically the catalytic hydrogenation of styrene. Apart from the substrate complex 1, all steps in the catalytic cycle have been pinned down exhibiting polarization signals [22]. It is also possible to screen catalysts according to their selectivities (regio- and stereoselectivity), their stabilities, and rates, since kinetic investigations can be carried out as well [23]. In addition, the application of especially designed NMR pulse sequences further extends the possibilities of the PHIP method [19]. For all of these applications, the main advantage of PHIP is the remarkable signal enhancement which is combined with a spectroscopic in situ method of observation. With the basic (albeit unconventional) tools at hand, we will focus in the following on the detection of key intermediates with standard NMR and the application of the PHIP method to mechanistic studies of enantioselective enamide hydrogenations. 9.4.2.4

H

H

L H

R’

H

R

L

R

1

R’

H

R’

H

R

H

H2

H H

L

L L

H2

M+

M+

R’

H

R

HH

L L

M+

L

H

M+

H

H2

R R’

H H

H

R’

H

H H

H

L

L

M

H R’

HH

L

+

R

R

H

L

M+

H

R R’

Reaction monitoring and investigation of catalytic mechanisms using PHIP-NMR spectroscopy (see text) [38]. Figure 9.12

9.5 Studying the Catalytic Mechanism of Enantioselective Hydrogenation

9.5

Studying the Catalytic Mechanism of Enantioselective Hydrogenation with NMR Spectroscopic Methods 9.5.1

Early NMR Experiments: Wilkinson’s Catalyst Revisited

The catalytic cycle of homogeneous hydrogenation using Wilkinson’s catalyst has already been discussed in Section 9.3.1. Three techniques haven been exploited for the elucidation of this mechanism. First, intensive kinetic studies have been performed [24], but since we are going to focus on spectroscopic measurement, these are beyond the scope of this book. Anyway, in most cases where a detailed reaction mechanism is needed, kinetics alone cannot prove the mechanistic model completely. Therefore, it is vital that at least the key intermediates (if not all) of the proposed mechanism are detected and characterized in situ spectroscopically. The second technique, used by the group of Tolman in 1974 [5], was 31 P NMR spectroscopy. They were able to visualize the exchange process between the Wilkinson complex and the active catalyst (which has lost one triphenylphosphine ligand, cf. Figure 9.3). The stereochemistry of this complex, cis versus trans, was described first by Brown [25], who demonstrated the inactivity of the trans complex in hydrogenation reactions, and later by Eisenberg [7] with the help of PHIP NMR spectroscopy. Eisenberg was able to directly detect the corresponding rhodium/alkene/dihydride intermediate, which was only possible with the help of the strong signal enhancement using the PHIP technique, the concentration of the complex in question, as being common with catalytic intermediates, being too low to be detectable with standard NMR techniques. The structure could be determined and was found to contain one hydrogen atom trans to one of the phosphine ligands and cis to the other phosphine; the group (olefin or chloride) trans to the second hydride ligand could not be established unambiguously, both possibilities remain (Figure 9.13). 9.5.2

The Mechanistic Model for the Chiral Case

For the enantioselective homogeneous hydrogenation of prochiral enamides, many experiments following the same principle (a combination of careful kinetic work and standard 1H- and 31P-NMR experiments), dominated by the groups of Kagan [26], Schrock and Osborn [27], Halpern [28], and Brown [29], led to the mechanistic

H H

PPh3 PPh3 Rh Cl

PPh3 PPh3 Rh H Cl H

R

R

Figure 9.13 Dihydride complexes of Wilkinson’s catalyst that were characterized using PHIP [7].

371

9 Hydrogenation

372

picture shown in Figure 9.14. Consequently, the mechanism follows the “unsaturate route” through addition of the olefin prior to dihydrogen. The enamide binds to the rhodium center in a chelate fashion through the double bond on the one hand and the amide carbonyl oxygen on the other. In this reaction, two diastereoisomeric pathways are possible; the catalyst, because of the chiral diphosphine ligand, can coordinate to the enamide in two diastereoisomeric ways. As a result, the two substrate complexes exhibit different chemical reactivity. One of the complexes is quite stable and relatively unreactive while the other is highly reactive towards molecular hydrogen. The high reactivity of the latter leads to a high enantiomeric excess of the one enantiomeric product generated by this complex. The two diastereoisomers have been termed major and minor by Halpern [30] and the rule of thumb here is that “minor gives major” and vice versa. This mechanistic behavior has two important consequences. On the one hand, it explains the observation that enantioselection in reactions of this type responds significantly to the hydrogen pressure used for the hydrogenation: high pressures give low ee and vice versa. This can be rationalized by the fact that the higher the hydrogen pressure, the higher the hydrogen concentration, and thus more of the more stable (but less reactive) dihydrido complex is formed, which ultimately gives the unwanted enantiomer (low ee). On the other hand, the major–minor behavior has consquences for the probability of detecting the catalytically active

O P

*

O

+

H N

O

H O

NH

Rh NH

P

*

O

P

+

H O

Rh

O

P O

O

"major complex" H

minor product O

*

P

+ Rh

P

H N

S

O

+

O

S

postulated dihydrides - never observed!

O H

*

P

O

+ Rh

P

NH

*

P

+

H

H

O O

Rh P

O

H N

O O

NH O

O

"minor complex"

Figure 9.14

The mechanism of enantioselective enamide hydrogenation.

major product


373

species: a detectable reactive intermediate will most probably be the major diastereoisomer and thus the one which does not form the dominant enantiomer. All of the intermediates in the mechanistic picture of Figure 9.14, i.e the catalyst, both substrate complexes [29], and both monohydrides [30–32], have been characterized by NMR spectroscopy, except one vital key intermediate: the catalyst-substrate-dihydrido complex. Without characterization of this molecule, the whole mechanistic model remains uncertain. In the following, the “hunt for the dihydride” is described. 9.5.3

Detection of Intermediates with Standard NMR Spectroscopy

So, what was known at this point about the missing dihydride complex? For direct comparison, Brown et al. studied iridium complexes with NMR spectroscopy [33, 34], since iridium dihydrides are easier to synthesize and much more stable than rhodium dihydrides. They were able to characterize both enamide substrate complexes of the iridium DIPAMP complex; upon hydrogenation, though, only the monohydride could be detected (interestingly, with an oxygen atom trans to the remaining hydrogen). The same group succeeded later in also characterizing the rhodium equivalent of this monohydride [32]. The structures of the three fully characterized monohydride complexes in the rhodium case are depicted in Figure 9.15 together with the spectroscopic data for the hydrido NMR signals. In all three cases, there is always an oxygen atom trans to the remaining hydride. From these complexes, we know about the chemical shift region of this hydrogen atom as well as the 2JHP coupling constants (which are all cis). For the hydrogenation of a related cyclopentenyl amide substrate, Brown was able to characterize all steps of the proposed mechanism including the first alkene dihydrides of the “missing” type using the iridium DIPAMP catalyst [33]

+

R Ar2 H P O Rh O P CH 2Ph Ar2 O NH Ph

Ph

CH3

+

O

Ph2 H OCH3 P NH Rh P O Ph2 O D CD3

O Ph H3COOC

PH Rh HN H Ph

PPh2 O CH3

δ = -19.3 ppm

δ = -20.9 ppm

δ = -20.2 ppm

2J = 11, 23 Hz HP 1J HRh = 34 Hz

2J = 14, 36 Hz HP 1J HRh = 27 Hz

2J = 13.5, 21 HP 1J HRh = 36 Hz

(Brown, 1980 [31])

(Halpern, 1980 [30])

(Brown, 1995 [32])

Rhodium monohydrides in enantioselective enamide hydrogenation that were characterized by conventional NMR spectroscopy. Figure 9.15

Hz

+

9 Hydrogenation

374

N(CH3)2 Ph

Ph H3CO

P P

+ Ir

H2

H3CO

-70 °C

O

Ph

N(CH3)2

Ph H3CO

P P

+ Ir H

H O

H3CO -45 °C N(CH3)2

Ph

H3CO

Ph O P + H Ir P H

H3CO δH = -30, -7.9 ppm

δH = -9.2, -8.7 ppm

N(CH3)2 H3CO -10 °C Ph

Ph O P + Ir P H

H3CO 0 °C

Ph CH 3 H N P + Ir P O O Ph CH 3

H3CO δH = -28.7 ppm

δH = -28.4 ppm

Figure 9.16 A series of iridium dihydrides that were fully characterized using standard NMR techniques [33].

(Figure 9.16). In this case, at low temperatures (–70 to –25 hC) one hydride atom was coordinated trans to the olefinic double bond. At higher temperatures (–40 to –10 hC) a second species with the hydride trans to the amide oxygen was detected. Warming the sample above –10 hC gave the monohydride. From the sum of these NMR investigations, the chemical shift and the corresponding coupling constants for a hypothetical rhodium-substrate-dihydride complex can be rationalized. One of the two hydrogen signals should fit the scheme of the well-known monohydride complexes (d z –20 ppm being trans to an oxygen atom with two cis 2JHP couplings). The other one should be coordinated either trans to a phosphorus atom or trans to the olefinic double bond. Therefore, one would expect a chemical shift of around –8 to –12 ppm (–8 to –10 trans to P, –12 to –14 trans to olefin [33]) and a large trans 2JHP coupling (100–180 Hz) in the case of P being trans. Up to 1998 this was the state-of-the-art and the limit of direct observation with standard spectroscopic techniques [2]. Still, the dihydrido complex in the rhodium case was undetected. 9.5.4

The “Breakthrough”: Characterization of the Key Intermediates with PHIP NMR Spectroscopy

So far, we have demonstrated that two of the three hypothetical intermediates in asymmetric enamide hydrogenation (i. e. the substrate complex and the monohy-


dride) can be observed directly by NMR, while the third and crucial one (the dihydride) is far too reactive to be detected. Just as a reminder, PHIP NMR spectroscopy is a unique tool for the investigation of homogeneous hydrogenation reactions due to a huge gain in signal intensity as compared to standard NMR spectroscopy and because this technique is truly dynamic and in situ. Thus, it is possible to detect intermediates (i) at very low concentration and (ii) under real turnover conditions. Therefore, the groups of Bargon and Brown chose this method in the search for the “missing” (i. e., formerly undetected) dihydride. After a long and discouraging period of time, the breakthrough finally came by using a new type of diphosphine ligand: PHANEPHOS (cf. Figure 9.4). This ligand was first described by Pye and Rossen at Merck/USA [35]. The use of PHANEPHOS as a ligand for rhodium-catalyzed homogeneous hydrogenation yielded a catalyst that was unusually reactive at low temperatures. The authors reported hydrogenation activity at –40 hC in methanol. We (the groups of Brown and Bargon including myself) were extremely lucky to receive an invaluable gift from Merck of the PHANEPHOS ligand at a very early stage when it was still not commercially available. The combination of the Rh(PHANEPHOS)þ catalyst together with PHIP NMR spectroscopy yielded strong polarization signals, as depicted in Figure 9.17 [36].

Figure 9.17 The first rhodium dihydrides in enantioselective enamide hydrogenation that were detected using PHIP NMR spectroscopy. A: Experimental spectrum. B: Computer simulation [36].

375

376

9 Hydrogenation

Analysis of the spectrum in detail shows that there are two antiphase hydride signals, one around –19 ppm and one around –2 ppm. From all we know so far (vide supra), the hydride signal at –19 ppm must be situated trans to an oxygen atom, which either stems from the substrate or from the solvent (methanol). It should comprise two cis 2JHP couplings (z 11 Hz), a 2JHH coupling (–4.3 Hz), and a 1JHRh coupling (23 Hz), and this is indeed what we find. (Note: The sign of the 2JHH coupling, positive or negative, can easily be determined by PHIP, since a change of sign results in inverted antiphase signals.) For the hydride signal at –2 ppm we have a serious explanatory problem. As described earlier, we would rather expect a signal in the range of –8 to –14 ppm, depending on the hydride being trans to P or to the double bond of the substrate. Additionally, the coupling constants for this signal read: –4.3, 6.5, 15, and 36 Hz , no obvious trans 2JHP coupling here. Additionally, people who are familiar with PHIP realize that the lines of this signal are significantly broadened, which might hint at either a limited lifetime of this species or a dynamic exchange process. Nevertheless, is was clear that these signals could only stem from a real dihydrido intermediate because (i) the PHIP effect is only detectable in cases where the two atoms of the dihydrogen molecule are transferred in a pair-wise manner and where a coupling between them is retained throughout the process, and (ii) selective decoupling experiments demonstrated unambiguously that they are coupled to each other. At this stage, being very uncertain about the structure of our dihydride, or even if the substrate was involved at all, we decided to use 13C labeled substrate with which an additional H–C coupling should be detectable (Figure 9.18). And indeed, for the b-labeled enamide the signal at –2 ppm changed significantly due to the additional coupling (86 Hz) [36]. A coupling of this size is not detectable for the a-labeled substrate. Furthermore, different substrates with different electronic properties (through an additional substituent in the phenyl ring of the substrate) resulted in a significant shift of both the dihydride signals [37, 38]. Consideration of all this NMR information, together with a DFT (density functional theory) calculation of the most reasonable structure with the highest probability, enabled the structure of this dihydride to be concluded [36]. One of the hydride atoms (the one with the signal at –2 ppm) is agostic; that is, it is situated in a position between being a classical hydride (coordinated only to Rh trans to O) and being transferred already to the b-carbon of the substrate (cf. Figure 9.19). With this structure, all the coupling constants (including the very strong 13C cou-

Ph

Ph

13C

O

O N H O

para-H2, [Rh(PHANEPHOS)] methanol

+

O

13C

H H O

N H H O

Hydrogenation of a b-13C-labeled enamide for the characterization of the resulting rhodium dihydrido species. Figure 9.18

References

CH3 Ph2 P

O

H

Rh H P Ph2

Ph NH The structure of the “agostic” rhodium dihydrido complex that was fully characterized with PHIP NMR spectroscopy [36].

Figure 9.19

H O

OCH3

pling) can be explained, as well as the chemical shift of this signal (being somewhere in the middle between the classical dihydride at –9 ppm and the product hydrogen at 3 ppm) and the much higher sensitivity of chemical shift of this signal towards the choice of substrate as opposed to the signal at –19 ppm. With this system, we finally succeeded in characterizing the first rhodium dihydride species in the asymmetric hydrogenation of enamides. Additionally, we succeeded afterwards in the characterization of all the possible catalyst dihydride species [39]. In subsequent work, now knowing what to look for and where to look, all transient complexes in the asymmetric enamide hydrogenation with the Rh(PHANEPHOS) catalyst could also be observed with classical NMR techniques [37].

9.6

Conclusion and Outlook

For the rational design of transition metal catalyzed reactions, as well as for finetuning, it is vital to know about the catalytic mechanism in as much detail as possible. Apart from kinetic measurements, the only way to learn about mechanistic details is direct spectroscopic observation of reactive intermediates. In this chapter, we have demonstrated that NMR spectroscopy is an invaluable tool in this respect. In combination with other physicochemical effects (such as parahydrogen induced nuclear polarization) even reactive intermediates, which are present at only very low concentrations, can be observed and fully characterized. Therefore, it might be worthwhile not only to apply “standard” experiments, but to go and exploit some of the more exotic techniques that are now available and ready to use. The successful story of homogeneous hydrogenation with rhodium catalysts demonstrates impressively that this really might be worth the effort.

References [1] H. Brunner, in Applied Homogeneous

Catalysis with Organometallic Compounds, Vol. 1, Eds. B. Cornils, W. A. Herrmann, VCH, Weinheim 1996, p. 201.

[2] J. M. Brown, in Comprehensive Asym-

metric Catalysis, Vol. 1, Eds. E. N. Jacobsen, A. Pfaltz, H. Yamamoto, Springer Verlag, Berlin 1999, p. 121.

377

378

9 Hydrogenation [3] J. A. Osborn, F. H. Jardine,

[4] [5]

[6]

[7]

[8] [9] [10] [11] [12]

[13] [14] [15]

[16] [17] [18] [19] [20] [21] [22]

G. W.Young, J. Chem. Soc. A 1966, 1711. J. Halpern, C. S. Wong, Chem. Commun. 1973, 629. C. A. Tolman, P. Z. Meakin, D. L. Lindner, J. P. Jesson, J. Am. Chem. Soc. 1974, 96, 2762. J. M. Brown, P. L. Evans, A. R. Lucy, J. Chem. Soc., Perkin Trans. 1987, 2, 1589. S. B. Duckett, C. L. Newell, R. Eisenberg, J. Am. Chem. Soc. 1994, 116, 10548; S. B. Duckett, C. L. Newell, R. Eisenberg, J. Am. Chem. Soc. 1997, 119, 2068. L. Horner, H. Buthe, H. Siegel, Tetrahedron Lett. 1968, 4023. W. S. Knowles, M. J. Sabacky, Chem. Commun. 1969, 1445. W. S. Knowles, Adv. Synth. Catal. 2003, 342, 3. T. P. Dang, H. B. Kagan, J. Am. Chem. Soc. 1972, 94, 6429. W. S. Knowles, M. J. Sabacky, B. D. Vineyard, Chem. Commun. 1972, 10. H. Gnther, NMR Spectroscopy, Thieme Verlag, Stuttgart 1992. C. R. Bowers, D. P. Weitekamp, Phys. Rev. Lett. 1986, 57, 2645. T. C. Eisenschmid, R. U. Kirss, P. P. Deutsch, S. I. Hommeltoft, R. Eisenberg, J. Bargon, R. G. Lawler, A. L. Balch, J. Am. Chem. Soc. 1987, 109, 8089. C. R. Bowers, D. P. Weitekamp, J. Am. Chem. Soc. 1987, 109, 5541. K. Woelk, J. Bargon, Z. Phys. Chem. 1993, 180, 65. M. G. Pravica, D. P. Weitekamp, Chem. Phys. Lett. 1988, 145, 255. J. Natterer, J. Bargon, Progr. NMR Spectrosc. 1997, 31, 293. T. Greve, Ph. D. Thesis, University of Bonn, Bonn, Germany, 1996. J. Barkemeyer, M. Haake, J. Bargon, J. Am. Chem. Soc. 1995, 117, 2927. A. Harthun, R. Giernoth, C. J. Elsevier, J. Bargon, J. Chem. Soc., Chem. Commun. 1996, 2483; R. Giernoth, P. Huebler, J. Bargon, Angew. Chem. Int. Ed. Engl. 1998, 37, 2473.

[23] P. Huebler, R. Giernoth, G. Kuem-

[24] [25]

[26] [27]

[28]

[29]

[30] [31] [32]

[33] [34]

[35]

[36]

[37] [38] [39]

merle, J. Bargon, J. Am. Chem. Soc. 1999, 121, 5311; S. Wildschutz, P. Huebler, J. Bargon, ChemPhysChem 2001, 2, 328. H. Arai, J. Halpern, J. Chem. Soc. D. 1971, 1571. J. M. Brown, L. R. Canning, A. R. Lucy, J. Chem. Soc., Chem. Commun. 1984, 914; J. M. Brown, A. R. Lucy, J. Chem. Soc., Chem. Commun. 1984, 915. H. B. Kagan, N. Langlois, T. P. Dang, J. Organomet. Chem. 1975, 90, 353. R. R. Schrock, J. A. Osborn, J. Am. Chem. Soc. 1976, 98, 2143; R. R. Schrock, J. A. Osborn, J. Am. Chem. Soc. 1976, 98, 2134. J. Halpern, D. P. Ripley, A. S. C. Chan, J. J. Pluth, J. Am. Chem. Soc. 1977, 99, 8055. J. M. Brown, P. A. Chaloner, Tetrahedron Lett. 1978, 19, 1877; J. M. Brown, P. A. Chaloner, J. Chem. Soc., Chem. Commun. 1978, 321. J. Halpern, A. S. C. Chan, J. Am. Chem. Soc. 1980, 102, 838. J. M. Brown, P. A. Chaloner, J. Chem. Soc., Chem. Commun. 1980, 344. J. A. Ramsden, T. D. W. Claridge, J. M. Brown, J. Chem. Soc., Chem. Commun. 1995, 2496. J. M. Brown, P. J. Maddox, J. Chem. Soc., Chem. Commun. 1987, 1278. N. W. Alcock, J. M. Brown, A. E. Derome, A. R. Lucy, J. Chem. Soc., Chem. Commun. 1985, 575; N. W. Alcock, J. M. Brown, P. J. Maddox, J. Chem. Soc., Chem. Commun. 1986, 1532. P. J. Pye, K. Rossen, R. A. Reamer, N.N. Tsou, R. P. Volante, P. J. Reider, J. Am. Chem. Soc. 1997, 119, 6207. R. Giernoth, H. Heinrich, N. J. Adams, R. J. Deeth, J. Bargon, J. M. Brown, J. Am. Chem. Soc. 2000, 122, 12381. H. Heinrich, Ph. D. Thesis, University of Bonn (Bonn, Germany), 2002. R. Giernoth, Ph. D. Thesis, University of Bonn (Bonn, Germany), 1999. H. Heinrich, R. Giernoth, J. Bargon, J. M. Brown, J. Chem. Soc., Chem. Commun. 2001, 1296.

Index A AAlXXl system 57 ABMX system 3 Absorbance spectroscopy 161 ABX system 23 Acetic acid 195 Acetic anhydride 195 Acetyl iodide equilibrium 214 Acetyl iodide hydrolysis 213 Achiral hydrogenation catalysts 360 Activation mechanism 145 Activation parameters 136 Active site counting method 31 Active state 339 Adiabatic Longitudinal Transport After Dissociation Engenders Nuclear Alignment (ALTADENA) 367–369 Agostic interaction 8, 14–15, 22, 28–29, 34, 41, 137, 318, 377 Alcoholysis 293 Aldehydes 231 Alkane complexes 144–145 Alkene/CO copolymerization 271 Alkene/CO copolymerisation, catalytic cycles 274 Alkene dissociation 39, 327 Alkene, exocyclic 10 Alkene hydroformylation 123, 239, 244 Alkene insertion 327 Alkene polymerisation reaction 323 Alkene polymerization, metallocene catalyzed 31, 382 Alkenes 271 Alkyl migration 136 Alkyl peroxo-complex 20 Alkyne hydrogenation 54 Alkyne hydrogenation catalyst 35 Alkyne insertion 323 Allylic halides carbonylation 130

Allylzirconium diene complexes 320 ALTADENA, see Adiabatic Longitudinal Transport After Dissociation Engenders Nuclear Alignment Amine cyclization, rhodium catalysed 64 Amsterdam HP-IR flow cell 110 Anhydrous carbonylation 226 Aprotic solvent 276, 282, 288 Ar-BIAN, see Bis(arylimino)acenaphthene Aryl halides amination 12 Aryl iodide 10 Aryl phosphine 19 Asymmetric catalysis 245, 361 Asymmetric copolymerisation 302 Atactic copolymer 298 Atactic polypropylene 345 ATR, see Attenuated total reflectance spectroscopy Atropisomeric 24 Attenuated total reflectance (ATR) spectroscopy 114 Autoclave HP-NMR system 89, 93 AX system 52, 366

B Band-target entropy minimisation (BTEM) 125, 180, 237 Batch polymerisation 281 bdompp, see Bis(di(o-methoxyphenyl)phosphino)propane bdpp, see Bis(diphenylphosphino)pentane Benzyl pi complex 297, 299, 300, 322 Benzyl sigma complex 321-2 Benzoquinone 279 Berylco, see Beryllium-copper alloy Beryllium-copper alloy (Berylco) 25, 84–85, 87,100 Bicyclic phosphines 124 Bidentate nitrogen ligand 24


380

Index Bidentate phosphine ligand 23, 25, 34, 40, 128 Bidentate phosphite ligand 244–245, 248 Bidentate pyrrolyl-based phosphorus amidite ligand 140 Bimetallic catalytic data 182 Bimetallic complex 59, 185 Binap 6, 8, 12, 14, 19, 22, 25, 362 BINAPHOS, see 2-(diphenylphosphino)binaphthalene-diyl phosphite bipy, see Bipyridine Bipyridine (bipy) 23–24, 33–34, 288, 301 Bis(arylimino)acenaphthene (Ar-BIAN) 288, 303 Bis(dicyclohexylphosphino)methane 59 Bis[di(o-methoxyphenyl)phosphino]ethane (MeO-dppe) 277, 292 Bis(di(o-methoxyphenyl)phosphino)propane (bdompp) 277 Bis(diphenylphosphino)amine (dppa) ligand 304 Bis(diphenylphosphino)benzene (dppbz) 277 Bis 1,4-(diphenylphosphino)butane (dppb) 23, 29, 35, 46–47, 55, 282–283, 290–291 Bis(2,3-diphenylphosphinobutane)rhodium 4 Bis(diphenylphosphino)ethane (dppe) 55–56, 132, 137, 277–278, 282, 289–290, 295, 303, 305 Bis(diphenylphosphino)ethane oxide, (dppeo) 12 Bis(diphenylphophino)ferrocene (dppf) complex 281, 282, 287, 293–294 Bis(diphenylphosphino)methane, (dppm) 304 Bis(diphenylphosphino)methane oxide, (dppmo) 12 Bis(diphenylphosphino)methane sulfide, (dppms) 12 Bis(diphenylphosphino)octamethylferrocene (dppomf) 293 Bis(diphenylphosphino)pentane, (bdpp) 282 Bis 3,4-(diphenylphosphino)pentane (bdpp) 12, 20–21, 282 Bis(diphenylphosphino)propane (dppp) 131, 277, 280, 282, 284–286, 290–291, 305–306 Bis(ditertiarybutylphosphino)xylyl 10, 35 Bis(mesityl)imidazol-ylidene 15 Bis(phosphinite) chelate 40 Bis(trifluoromethyl)phenyl 299 Bloch equation 38 Borane 322, 330, 339 Borole-bridged tripledecker 320 Bridged-sandwich complexes 318

Bromobenzene carbonylation 116 BTEM, see Band-target entropy minimisation Bulky diphosphite 245–246 Bulky mono-phosphite 240 Bulky phosphorus diamide 127, 252 Butene equilibrium constants for coordination to Y 328 Butene hydroformylation 142, 244 Butylaluminoxane cages 340 Butyl-methylimidazolium hexafluorophosphate 129 Butyraldehydes 63

C Carbon monoxide 271 Carbon monoxide dissociation 260 Carbon monoxide dissociation kinetics 257 Carbon monoxide reactions 130 Carbonyl cluster 20, 26, 48, 57, 70, 125, 132, 180, 185, 237, 250 Carbonylation activity 227 Carbonylation catalysts 118, 138, 200 Carbonylation mechanism 122, 135, 138, 199, 201, 204 Carbonylation mechanism with HP-IR 117, 120, 201 Carbonylation mechanism with HPNMR 204 Carbonylation process chemistry 195–196 Carbonylation process mechanism 199 Carbonylation processes 195–196 Carbonylation reaction 200 Carbonylation reaction, organic cycles of 212 Carbonylation of methanol, iridium catalysed 226 Carbonylation of methanol, rhodium catalysed 223 Carborane 35 Catalysis resting states 12, 34, 42, 107, 121, 125, 127, 133, 139, 141, 142, 233, 241, 272, 287, 297, 336 Catalyst deactivation 70, 126, 164, 168, 283, 305 Catalyst poison 227 Catalyst precursors 18, 26, 40, 57, 69, 98, 124, 129, 131–132, 236–237, 321, 319 Catalytic carbonylation 131 Catalytic hydroformylation 125, 140, 231 Catalytic product formation 167 Catalytic synthesis, types 164 Catalytic system 154 Catalytically active species, identification 313 Cativa process 19, 133, 196–197, 226 Celanese process 197

Index Chain-end control 299, 347–348 Chain-end epimerisation 337 Chain epimerisation 350 Chain propagation 279, 289, 302, 338 Chain transfer 275, 292–294, 301, 344 Chalk-Harrod hydrosilylation 20 Chelation 19, 127, 129 Chemical shift variation 29 Chemical shifts 14, 346 CIR, see Cylindrical internal reflectance spectroscpoy Chiral biphosphine ligands 362 Chiral diphosphites 248 Chiral hydrogeneation catalysts 361 Chiral rhodium diphosphinite complex 94 Chiraphos 3, 24 Chromium carbonyl complex 58, 112, 142–143, 145 Circle cell 204 Cis coupling constant 9, 21–22, 234, 248, 373, 376 C NMR 3, 7, 10, 16, 20, 23, 133, 207, 231, 259, 321, 323, 325, 337–338, 340–341, 344–347 Coalescence point 38, 319 Cobalt acyl complex 61, 63, 123–124, 132, 137–138 Cobalt alkyl complex 123, 138 Cobalt carbonyl cluster 62, 144–145, 179 Cobalt carbonyl complex 132, 142 Cobalt carbonyl dimer 61, 63, 123–124, 145, 179 Cobalt catalysts 123, 231 Cobalt complex, modified hydroformylation catalyst 124 Cobalt complex, unmodified hydroformylation catalyst 124 Cobalt hydride 62–63, 123–125, 132, 188 Cobalt/rhodium carbonyl cluster 125 Cobalt/rhodium carbonyl dimer 125 Co NMR 61–63 CO/ethene copolymerisation 282 CO migration 48 ConcIRT 76 Comp probe 116 Continuous insertion mechanism 337 Continuous stirred tank reactor (CSTR) 154 Continuous stirred tank reactor recycle 155, 160, 169 Copolymerisation, catalytic cycle 274–275 Copolymerisation reactions 271 Copper carbonyl 139 Copper catalysts 67 Copper effect 12

Coranals simulating annealing 179 COSY 259 Coupling constant 21 Cross peaks 2, 24, 44, 47–49 Cryomagnet 100 Cryomagnet, wide-bore 85 CSTR, see Continuous stirred tank reactor Curve fitted spectrum 175 Curve fitting 174 Curve resolution 176 Cyclic alkenes/CO copolymerisation 274, 302 Cyclohexene hydroformylation, rhodium catalysed 125 Cyclometallation 19 Cyclooctene 55 Cyclooctene hydroformylation 127 Cyclopentane complexes 145 Cylindrical internal reflectance HP-IR cell 115 Cylindrical internal reflectance crystal 115 Cylindrical internal reflectance (CIR) spectroscopy 115

D DANTE inverse recovery 44 Dark reaction 157 Data filtering 169 Data pre-processing 169 DBE, see Dibutylether dcpm, see Bis(dicyclohexylphospino)methane dtbpx, see Bis(ditertiarybutylphosphino)xylyl Decene hydroformylation 124, 236 Density functional theory (DFT) 146 Density functional theory calculation 334, 376 Deuterium labelling 94, 129, 140, 262 Deuterium quenching 336 DGTS detector 166 Dialklylalkenes 240 Dibutylether (DBE) 334 Dichlorobenzene 352 Dichloromethane 312 Difference IR spectrum 262, 264 Diffusion ordered spectroscopy (DOSY) 70 Dihydrofuran 6 Dihydrogen complexes 50, 58, 59, 143–144 Di(isopropylphospino)propane (dippp) 34 Dimeric complexes, see – Cobalt carbonyl dimer, – Cobalt/rhodium carbonyl dimer, – Iridium/ruthenium dimer, – Manganese carbonyl dimer, – Manganese/rhodium carbonyl dimer, – Palladium dimer,

381

382

Index – Rhodium carbonyl dimer, Dimer formation 250 Dimethylaniline (DMA) 334 Dimethylbutene 186, 237 Dimethylbutene hydroformylation 125, 181 Dimethyl diallyl malonate cycloisomerisation 11 Dimethylfumarate 142 Dimethylpentanal 182, 186 DIOP 40, 362 Diphenylbutadiene 369 Diphenylbutadiyne 60 (Diphenylphosphino)binaphthalene-diyl phosphite (BINAPHOS) 101, 299, 302 Diphenylphosphino-tert-butyl-3-dicyclohexylphosphinopropene 287 Diphosphine ligands, wide bite angle 128, 234 Diphosphite catalysts 244 Diphosphite ligands 244–246, 251 Diphosphite rhodium hydride complex 235 Dipole-dipole relaxation 50 dippp, see Diisopropylphospino propane Dissymmetric diphosphine 287 Ditelluroether 49 DMA, see Dimethylaniline 30–32, 336 2D NMR experiment 43, 46, 48, 70, 334 Dodecene hydroformylation, cobalt catalysed 124 Dormant state 338 DOSY, see Diffusion ordered spectroscopy Double ethene insertion 279 dcpm, see Bis(dicyclohexylphosphino)methane dppa, see Bis(diphenylphosphino)amine dppb, see Bis(diphenylphosphino)butane dppbz, see Bis(diphenylphosphino)benzene dppe, see Bis(diphenylphosphino)ethane dppeo, see Bis(diphenylphosphino)ethane oxide dppf, see Bis(diphenylphophino)ferrocene dppm, see Bis(diphenylphosphino)methane dppmo, see Bis(diphenylphosphino)methane oxide dppms see Bis(diphenylphosphino)methane sulphide dppomf, see Bis(diphenylphosphino)octamethylferrocene dtbpx, see Bis(ditertiarybutylphosphino)xylyl 2D spectrum 43, 70 3D solution structure determination 27, 43–45, 362 Duphos 2, 17, 23, 24 Dynamic NMR spectroscopy 33, 325, 329, 351

E Eastman process 197 ECD, see Electronic circular dichroism E-isomer 53–54, 351, 368 Electrochemical studies 312 Electron paramagnetic resonance (EPR) 155, 342 Electronic circular dichroism (ECD) 155, 163 Electronic spin state 56 Enantioselective enamide hydrogenation 372 Enantioselective hydrogenation 40, 42, 63, 94, 361–362, 370–374, 377 Enantioselective hydroformylation 245, 250 Entropy minimisation 176 Epoxide carbonylation, cobalt catalysed 132 EPR, see Electron paramagnetic resonance Equilibrium constants 124, 131, 138, 219, 325, 331–332 Ester end group 279, 293 Ethene/CO copolymerisation 274, 284, 286 Ethene/CO copolymerisation kinetics 288 Ethene/CO copolymerisation model synthetic studies 286 Ethene hydroformylation 125 Ethene insertion Ethene methoxycarbonylation 8, 10 Ethene polymerisation 29, 311, 335, 341 Ethene polymerisation, nickel catalysed 41 Ethyl hexanol 132, 284 Ethylidene diacetate 198 Evanescent wave 114 Evolution time 44 Exchange rate 38, 42, 220 Exchange spectroscopy (EXSY) 43 EXSY spectrum 43, 47–49 Extinction coefficient 201

F Fast fourier transformation (fft) 151 Fast spinning high pressure NMR sapphire tube 273 Fermi contact term 21 fft, see Fast fourier transformation Flash photolysis 112, 133, 137, 143–144 Flexibility range 232 Fluorescence spectroscopy 163, 166 Fluxional processes 35, 39, 55, 235, 248, 319, 323 F NMR 11, 315, 347 Fordls HP-IR cell 113 Formic acid 29, 279 Fourier deconvolution 124 Fourier transform 152

Index

G Gas/liquid mixing 204 Gaussian distribution 174 Glass capillary 88 Gold carbonyl 139

H Hafnocene 316 Half lives 162 Half-sandwich complexes 317, 342 Half-sandwich titanium complexes 342 Hammett acidity 219 Hastelloy B2, 108 Hastelloy C 108, 276 Hatta regimes 160 Heck arylation 5–7 Heck catalyst 18 Heck reaction 38 Hemi-isotactic polypropylene 345 Heptane 145 Heterogeneous catalysts 108, 117, 311 Heteronuclear Overhauser spectroscopy (HOESY) 69 Hexafluoro-2-propanol (HFIP) 281 Hexene 28, 31–32, 257–259 Hexene equilibrium constants 328 Hexene polymerisation 32, 336 Hexenoic acid 53–54 Hexynol 53 High pressure ATR IR cell 116 High pressure glass capillary 89 High pressure high resolution NMR probe 87 High pressure IR spectroscpy (HP IR) 107 High pressure IR cell 108, 117 High pressure IR spectroscopy, applications 107, 117 High pressure NMR cell 85–104 High pressure NMR flow cell, application 100, 231 High pressure NMR, nonmagnetic materials for 84 High pressure NMR of dissolved gases 96 High pressure NMR of liquids 83 High pressure NMR in supercritical fluids 60–61, 64–65, 81 High pressure NMR of supercritical fluid 90 High pressure NMR probe 94, 100 High pressure NMR probe with stirring/ mixing 100–104 High pressure NMR spectroscopy (HP NMR) 56, 218 HMQC spectra 259

HOESY, see Heteronuclear Overhauser spectroscopy Homogeneous catalytic hydrogenation reaction 1, 3, 4, 26, 28–30, 35, 40, 44, 48, 53–55, 58, 60–61, 63, 68, 98, 360–361, 369, 372–373, 375 Homogeneous hydrogenation 360, 361 Homometallic complex 238 HP NMR, see High pressure NMR spectroscopy 6, 10–11, 13, 30–31, 33, 36, 39, 53, 56, 62–63, 65, 90, 205, 215–216, 289–290, 323 Hydride migration 241–242, 252, 254, 265, 295 Hydride resonances 14, 233, 258 Hydride route 8, 362 Hydrido rhodium complex 14, 30, 242 Hydrido rhodium dicarbonyl diphosphine 233 Hydridoruthenium species 131 Hydroformylation 35, 57, 123, 232 Hydroformylation, cobalt catalysed 61, 63, 124 Hydroformylation, rhodium catalysed 35, 57, 124, 231 Hydroformylation catalysts 125, 240, 251 Hydroformylation reaction cycle 253 Hydrogenation 359 Hydrogenation mechanism study 52 Hydrogenolysis 140, 255 Hydrogenolysis, kinetics of 141 Hydrolysis 293 Hydropalladation mechanism 10

I Ill-posedness 168 Indenyl complexes 28, 31, 314, 335 Indium promoter 121 Induction-precursor transformation 167 Infrared (IR) autoclave 111 Infrared (IR) frequencies from stoichiometric reaction 257 Infrared (IR) laser 112 Infrared (IR) spectroscopy, kinetics 139 In situ HP IR 107–145, 201, 237, 264, 284 In situ HP NMR 81–102, 280, 282 In situ NMR spectroscopy 290 Intermittent growth model 337 Intramolecular exchange 43–46, 50, 143, 237, 248, 288, 294, 306 Intrinsic relaxation rates 58 Iodoruthenium species 121–122, 132, 136–137, 198, 212, 222, 228 Ionic liquids 129

383

384

Index Ion-pair dynamics 68–70, 316–318, 328, 333–334, 341, 343 Ion-pair symmetrisation rate 330, 332 IPCA 176 Iridium acyl complex 119, 122, 133, 136, 209–211 Iridium carbene complex 16 Iridium carbonyl complex 118–122, 133, 138, 199, 209, 227 Iridium carbonylation cycle 122, 209, 211, 227 Iridium dihydrides 27, 52–53, 139, 142, 374 Iridium iodocarbonyl species 120, 129–130, 132, 200, 210 Iridium methyl complex 31, 118–122, 129, 131, 133–137, 141, 195, 199, 200, 204–207, 209–211, 222–223, 226–227 Iridium monohydride 52, 119, 138, 141, 209, 319 Iridium/ruthenium dimer 211 Iridium technology 197 Iron acetyl complex 138 Iron alkyl complex 138 Iron carbonyl complex 142 Irradiated reaction 157 Isobutane 323 Isobutene 323 Isobutene elimination 322 Isoprene hydroformylation 179 Isotactic copolymer 298 Isotactic polypropylene 345 Isotopic labelling 27, 89, 127, 139, 204, 232, 278, 312, 337, 350

J Josiphos 24

K Kalrez O-ring 111, 202 Karplus relation 21 Keto-ester end group 274 Kinetic isotope effect (KIE) 28–29, 145, 207 Kinetics, first order 289 Kinetics of carbonylation 134 Kinetics, pseudo first order 135–136 Krypton complex 143, 145

L Lactones 132 Lambert-Beer-Bourguer-law 162 Lanthanide complexes 326 Larmor frequency 444 Limonene 124 Line shape analysis (LSA) 38

Liquid noble gas (LNG) 111 Longitudinal magnetization 44, 50, 83 Lorenzian distribution 174 Low temperature HP-IR cell 111 LSA, see Line shape analysis Lumped parameters 165

M MAC, see Methyl-acetamidocinnamate Macor 88 MALDI, see Matrix-assisted laser desorption/ ionization MAO, see Methylaluminoxane Magnetic flux 93 Magnetic labelling 368 Magnetization transfer technique 42 Manganese acyl complex 137–138 Manganese alkyl complex 138 Manganese carbonyl dimer 238 Manganese hydride complex 126, 182, 185, 238 Manganese/rhodium carbonyl dimer 126,185 Matrix-assisted laser desorption/ionization (MALDI) mass spectrometry 301 MCT, see Mercury-cadmium-telluride detector MeO-Biphep 24 MeO-dppe, see Bis[di(o-methoxyphenyl)phosphino]ethane Mercury-cadmium-telluride (MCT) detector 116 Meso isomers 49-50, 282-4, 290-1 Meso-2,3-bis(diphenylphosphino)butane, (meso-dppb) 282 Meso-2,4-bis(diphenylphosphino) pentane, (meso-bdpp) 282 Metal chemical shift 20, 61 Metal cluster complex 20, 26, 48, 57, 70, 125, 132, 159–160, 237, 250 Metallacycle 276, 284, 291 Metallocene-based polymerisation catalysis 311 Metallocene catalysts 35, 313, 328 Metallocene hydride complexes 323 Metallocene polymerisation catalysis 315 Metallocene systems, spectroscopic studies 339 Methanol carbonylation 117, 134, 139, 142, 195 Methanol carbonylation, iridium catalysed 119, 122, 133, 138, 141, 146, 226 Methanol carbonylation, platinum catalysed 132 Methanol carbonylation, rhodium catalysed 117–119, 138, 224

Index Methanolysis, intramolecular 293 Methyl-acetamidocinnamate (MAC) 3, 44–45 Methylacetate carbonylation 118, 195 Methylaluminoxane (MAO) 12–13, 67–68, 279, 338–344, 347, 351–352, 313, 338–344, 347, 351–352 Methylaniline 13, 335 Methylborate 315, 331 Methyl ethyl ketone 295 Methyl migration 29, 30–32, 135, 287, 328 Methyl pentads 346–347, 351 Methyl propanoate 8, 40 Methylstyrene insertion 297 Metropolis algorithm 179 Microwave-homogeneous catalysis 156 Migratory CO insertion reaction 34, 133, 204, 207, 211–212, 316 Migratory deinsertion 138 Migratory ethene insertion reaction 29, 34, 267, 279, 288–290, 304 Migratory insertion 207, 210, 289 Model reaction study 288 Molecular mechanics 260 Mo NMR 20 Molybdenum carbonyl complex 130–131, 142–143, 145 Monomer coordination 333 Monometallic catalytic data 181, 184 Monsanto process 196 Multiple phase CSTR 159

N Naphthoquinone 279 Naphthyl hydride complex 46 Natural bite angle 232 Nickel carbonyls 122, 232 Nickeo(diimine) complex 41 Nickel catalysts 122 Nitrotoluene 132 N NMR 65, 88 NMR application 204 NMR chromatography 70 NMR diffusion method 65–71 NMR flow cell 257 NMR heating facility 99 NMR relaxation measurement 92 NMR silent 364 NMR spectroscopy 12 NMR spectroscopy, reaction kinetics 9, 335 NMR spectroscopy of polyalkenes 344 Noble gas complexes 142–143 NOE, see Nuclear Overhauser effect NOESY measurement 3 NOESY studies 23

Non-Bolzmann spin 51 Non-classical metal carbonyls 139 Norbornadiene 142, 302 Norbornene 144, 302 Norbornene enantioselective hydroarylation 17 Nortricyclene 142, 144 Nuclear Overhauser effect (NOE) 1, 23, 332, 368

O Octene 256 Octene hydroformylation 124, 142, 244, 251 Olefin dissociation 326 Olefin hydroformylation catalyst 61 Olefin hydroformylation catalyst precursor 57 Olefins, homogeneous hydrogenation 359 O NMR 58, 83, 97–98 OPA-ALS 176 Orthohydrogen 365 Oxidative addition 10, 46, 50, 52, 56, 118– 119, 130, 132, 136, 138, 142–145, 205–209

P Palladium acyl complex 276, 279, 286–288, 289–292, 294, 298–299, 302–304 Palladium alkyl complex 277, 279, 281–291, 294, 298–299, 302–304 Palladium allyl complex 2, 10–11, 16, 25, 48, 295, 297, 299, 306 Palladium allyl phosphino-oxazoline complex 48 Palladium catalysts 5, 7–8, 11–12, 20, 270–306 Palladium dimer 277, 282–283, 297, 306 Palladium hydride 7–9, 14, 292–295, 297– 298, 305–306 Palladium methyl complexes 282 Palladium salt 22 Palladium sites, active 277 Parahydrogen 51–55, 365–368 Parahydrogen And Synthesis Allow Dramatically Enhanced Nuclear Alignment (PASADENA) 366–369 Parahydrogen-induced polarisation (PHIP) 3, 51–53, 55, 60, 63–64, 95, 365—370, 374–377 Parr autoclave 112, 202 PASADENA, see Parahydrogen And Synthesis Allow Dramatically Enhanced Nuclear Alignment Path dependant catalysis 167 PCA, see Principle component analysis Pearson VII model 174

385

386

Index Pentad level 345 Perfluorophenyl boron compounds 315 PFG, see Pulsed field-gradient PGSE, see Pulsed gradient spin echo Phenanthroline (phen) 23, 288 Phenyl acetylene hydrosilylation 14 Phenylboronic acid 3, 5 PHIP, see Parahydrogen-induced polarisation Phosphorus bronze 84, 93, 95 Phosphorus diamide ligands 252, 263 Photo-homogeneous catalysis 156 Pincer catalyst 38 Pincer ligand complex 22 Pinene 24 Piperidine carbonylation, ruthenium catalysed 131–132 Platinum carbonyl cluster 132 Platinum carbonyl complex 132 Platinum catalyst 20 Platinum(diimine) complex 23 Platinum hydride 10, 22 Platinum silyl complex 20 PM-RAIRS, see Polarisation modulation reflection absorption IR spectroscopy P NMR 4–5, 7–9, 13, 37, 41, 43, 234, 258, 278, 280–4, 296, 300 Pocket angle 25 Polar alkenes/CO copolymerisation 304 Polarisation modulation reflection absorption IR spectroscopy (PM-RAIRS) 132, 285 Polyesters 132 Polyethylene film 141–143 Polyketone end group 275, 292 Polyketones 132, 274, 284, 301–302 Polymer chain growth 314, 328 Polymerisation catalysts 311 Polymerisation kinetics 335 Polymer matrices, mechanical studies 141 Poly(1-oxo-2-methyltrimethylene) 301 Polypropylene 311, 344–346 Poly[spiro-2, 5-(3-methyltetrahydrofuran)] 301 Polyspiroketal 301 Pressure gauge, electronic 99 Principle component analysis (PCA) 176 Principle of simplicity 176 Probability distribution 177 Promoter 119–121, 135, 196–199, 211, 215 Propene/CO copolymerisation mechanism 301 Propene equilibrium constants 328 Propene hydroformylation 142, 244 Propene insertion 324 Propene polymerisation 334, 336

Propene polymerization, zirconocene catalyzed 12, 344–351 Propenol 130 Propylene 63 Propylene hydroformylation 63, 311 Protic solvent 274, 280 Protonolysis 283, 294 PTA, see 1,3,5-Triaza-7-phosphaadamantane 71 Pt NMR 20 Pulse gradient spin echo (PGSE) 65, 332 Pulsed field-gradient (PFG) NMR methods 342 Pure component spectrum 125, 162, 237 Pyrex tube 86

Q Quartz capillary 88, 91 Quaternary ammonium iodide salt [QAS]I 196, 197, 205, 215–218, 220 Quenching 31–32, 239, 336, 338

R Rac isomers 28, 31, 57, 282-3, 290-1 Rac-2,3-bis(diphenylphosphino)butane, (rac-dppb) 283, 290-1 Radicals 62-3, 131, 166 Radiofrequency (RF) feedthrough 84 Radon transform 152 Raman optical activity (ROA) 165 Raman spectroscopy 163 Rate constant 38, 42, 136–138, 140, 207, 209–210, 217 Rate of hydroformylation 237 Re NMR 61 Reaction time scale 160 ReactIR system 116 Reagent pure component spectra 170 Reductive elimination 37, 55, 126, 133, 207–208, 211–212, 361 Reductive elimination in the iridium methanol carbonylation system 126, 133, 211 Reductive elimination in the rhodium methanol carbonylation system 207–208, 211–212 Reference spectra 171 Reflectance IR cell 114 Regioisomerism 181, 298 Regioselectivity 260, 265, 304 Reporter ligand 23, 26 Resonance frequencies 38 RF-coil design 86 RF-feedthrough 87, 95, 104 Rhenium carbonyl complex 142

Index Rh NMR 20, 58 Rhenium complex 37, 143 Rhodium-acyl complex 16, 125–127, 130, 140–142, 173, 182–183, 185–186, 202–203, 206–208, 212, 223, 254–255 Rhodium-acyl resonances 258 Rhodium-acyl vibration 255 Rhodium alkyl complex 136, 140, 185, 206–208, 236, 243 Rhodium allyl complex 3 Rhodium based hydroformylation catalyst 125 Rhodium bulky diphosphite catalysts 244 Rhodium carbonyl catalyst, ligand-free 237 Rhodium carbonyl cluster 48, 57, 125, 131, 140, 171–173, 182–183, 187, 238 Rhodium carbonyl complex 118, 130, 136, 138, 142, 144–145, 198, 201, 206–208, 223–226, 240 Rhodium carbonyl dimer 226 Rhodium carbonylation cycle 205 Rhodium catalysed, modified hydroformylation 243 Rhodium catalysed homogeneous hydrogenation 4, 40, 55, 60, 64, 68, 359–377 Rhodium catalysts 3–4, 40, 55, 60, 64, 68, 110, 117 Rhodium catalysts selectivity 29, 35, 40, 44, 57, 68, 231, 231, 234, 244–245 Rhodium-deuteride vibration 247 Rhodium dihydrides 371, 375 Rhodium diketonate complex 142 Rhodium dimer 57, 126, 187, 251 Rhodium-hydrides vibration 247 Rhodium hydroformylation catalysts, unmodified 125, 184 Rhodium hydroformylation catalysts, phosphine modified 126, 243 Rhodium hydrogenation catalysts 68 Rhodium monohydrides 126–129, 139–141, 185–188, 225, 236, 239, 241, 243, 246–247, 249, 251–253, 257–258, 261–264, 373 Rhodium technology 197 Ring opening metathesis polymerization (ROMP) 96 Roels sapphire HPNMR tube 96 ROMP, see Ring opening metathesis polymerization Ruthenium carbonyl complex 55–56, 97–98, 121, 131–132, 135–136, 211, 221–222, 227 Ruthenium carbonyl iodides 135, 211, 221 Ruthenium catalysts 54 Ruthenium hydride complex 55–56, 119, 131

Ruthenium phosphine pincer complex 22 Ruthenium promoter 121

S Sandwich complexes 318 Sapphire tube 96 Sapphire tube electronic pressure gauge 99 Sapphire tube heating device 98 Sapphire tube schematic 97 Sentinel IR probe 202 Silapentenyl zirconocenes 325 Silver carbonyl 139 Silylformylation 35 SIMPLICMA 176 Singular value decomposition (SVD) 177 Site epimerisation 325, 350 Site exchange 327, 333 solvent polarity 207–208 Solvent pre-catalytic spectral data 170 Solvents, critical values of 91 Sono-homogeneous catalysis 156 Spectral reconstruction 176 Spectrometer choices 164 Spectrometer linearity 166 Spectroscopic measurement 161–162 Spin lattice relaxation 44, 50 Stainless steel 84 Stanleyls dinuclear hydroformylation catalysts 130 Step-scan FT IR instrument 113 Stereoisomerism 298 Stereoselectivity 40, 304, 347 Stille reaction 5, 12 Stoichiometric carbonylation 136 Stoichiometric hydroformylation 257 Stoichiometric reactions 256 Stokes-Einstein equation 66 Structural effect 29 Styrene/CO copolymerisation 297 Styrene hydroformylation 129, 236 Styrene, syndiospecific polymerisation 322 Superconducting NMR magnet 84 Supercritical fluids 60–61, 64, 90, 92–93, 95 Syn gas 142 Syn gas reaction 131 Syndiotactic copolymer 298 Syndiotactic copolymerisation 297–299 Syndiotactic polypropylene 345 SVD, see Singular value decomposition

T TCA, see Toroid cavity autoclave TCD, see Toroid cavity detector Teflon tube 90

387

388

Index Temperature effect 34 Tetrachloroethane 345 Tetrahydrofuran (THF) complex 322 THF, see Tetrahydrofuran Time-resolved IR spectroscopy (TR IR) 112 Timescales for spectroscopic measurements 162 TIR, see Total internal reflection Titanium alloy 84, 92 Titanium catalysts 334 Titanocene 316 Toluene 322 Toroid cavity autoclave (TCA) 93, 95 Toroid cavity detector (TCD) 93, 100 Toroid detector with stir coil 100 Toroid probes 93 Total internal reflection (TIR) 202 Track finding 173 Trans coupling constant 212 Transmission IR cell 202 Transport time scale 159 Transverse magnetisation 44 1,3,5-Triaza-7-phosphaadamantane (PTA) 98 Tributylphosphite 239 Trichlorobenzene 345 Trimethylplatinum iodide complex 49 Trinuclear cluster 26, 70, 131–132 TR IR, see Time-resolved IR spectroscopy Tris(tert-butyl-4-methylphenyl)phosphite 240 Tris(tert-butylphenyl)phosphite 239 Tuck-in complexes 320 Tunable IR diode laser 113 Tungsten carbonyl complex 142, 145 Turnover frequency 94, 163

U UV-VIS spectroscopy 166

V Vanadium ethene complexes 324 Vanadium propene complexes 324 Variable temperature NMR spectra 33–41, 62, 248–249, 259, 278, 282–283, 322, 325, 338, 343 Vaska complex 52 VCD, see Vibrational circular dichroism Vespel 82 Vibrational circular dichroism (VCD) 155 Vinyl aromatics carbonylation 297 Vinyl methyl ketone 295 Viton 82

W Wacker type reaction 278 Water gas shift (WGS) reaction 118, 119, 196–199, 204, 224–227, 275 Well-posedness 168 WGS, see Water gas shift reaction Wilkinson catalyst 21, 371 Wilkinson catalyst, structure 360 Wilkinson hydrogenation catalysts 232 WIN DYNAMICS 40

X Xantphos 128 Xantphos family 128, 260 Xantphos ligands, electronic effects Xenon complexes 112, 143

234

Y Yttrium complexes 326

Z Zeeman effect 363 Ziegler catalysts 312 Ziegler-Natta catalysts 311 Zirconium 315 Zirconium complex 35, 319–320, 324, 343, 351 Zirconium alkene complexes 323 Zirconium (ansa-fluorenyl) complexes 2 Zirconium (ansa-fluorenyl) polymerization 17 Zirconium butenylborate complexes 324 Zirconium catalysts 32 Zirconium complex fluxional behaviour 39 Zirconium complexes 320 Zirconium dimethylaniline adduct 343 Zirconocene 316 Zirconocene alkenyl complexes 325 Zirconocene catalysts 68 Zirconocene catalysts, ion pair 328 Zirconocene-catalyzed polymerization 5 Z-isomer 53–54, 328, 351, 368, Zirconocene hydride species 343 Zwitterionic metal alkyl species 312 Zwitterionic rhodium complex 35, 46 Zwitterionic zirconocene 319 Zwitterions 315

Mechanisms in Homogeneous Catalysis: A Spectroscopic Approach

Mechanisms in Homogeneous and Heterogeneous Epoxidation Catalysis

Mechanisms in Homogeneous and Heterogeneous Epoxidation Catalysis

Homogeneous Catalysis: Mechanisms and Industrial Applications

Homogeneous catalysis: mechanisms and industrial applications

Homogeneous Catalysis II

Homogeneous Catalysis: Understanding the Art

Some studies on homogeneous catalysis

Homogeneous Catalysis: Understanding the Art

Industrial Catalysis. A Practical Approach

Mechanisms in Classical Conditioning: A Computational Approach

Computational Modeling for Homogeneous and Enzymatic Catalysis

Mechanical Catalysis: Methods of Enzymatic, Homogeneous, and Heterogeneous Catalysis

Computational Modeling of Homogeneous Catalysis (Catalysis by Metal Complexes)

Homogeneous Catalysis. Industrial Applications and Implications

Fuel cell catalysis: a surface science approach

Spectroscopic measurement

Catalysis: An Integrated Approach to Homogenous, Heterogenous and Industrial Catalysis (Studies in Surface Science and Catalysis)

Catalysis An Integrated Approach to Homogenous Heterogenous and Industrial Catalysis

Selective Nanocatalysts and Nanoscience: Concepts for Heterogeneous and Homogeneous Catalysis

Spectroscopic Techniques in Industrial Hygiene

Spectroscopic Measurement

Catalysis

Catalysis

Advances in Catalysis

Trends in Catalysis Research

Supported metals in catalysis

The Enzymes, Vol XX: Mechanisms of Catalysis, 3rd Edition

The Enzymes, Vol XIX: Mechanisms of Catalysis, 3rd Edition

Hydrolysis of Cellulose Mechanisms of Enzymatic and Acid Catalysis

The Enzymes, Vol XX: Mechanisms of Catalysis, 3rd Edition

Mechanisms in Homogeneous Catalysis: A Spectroscopic Approach

Mechanisms in Homogeneous and Heterogeneous Epoxidation Catalysis

Mechanisms in Homogeneous and Heterogeneous Epoxidation Catalysis

Homogeneous Catalysis: Mechanisms and Industrial Applications

Homogeneous catalysis: mechanisms and industrial applications

Homogeneous Catalysis II

Homogeneous Catalysis: Understanding the Art

Some studies on homogeneous catalysis

Homogeneous Catalysis: Understanding the Art

Industrial Catalysis. A Practical Approach

Mechanisms in Classical Conditioning: A Computational Approach

Computational Modeling for Homogeneous and Enzymatic Catalysis

Mechanical Catalysis: Methods of Enzymatic, Homogeneous, and Heterogeneous Catalysis

Computational Modeling of Homogeneous Catalysis (Catalysis by Metal Complexes)

Homogeneous Catalysis. Industrial Applications and Implications

Fuel cell catalysis: a surface science approach

Spectroscopic measurement

Catalysis: An Integrated Approach to Homogenous, Heterogenous and Industrial Catalysis (Studies in Surface Science and Catalysis)

Catalysis An Integrated Approach to Homogenous Heterogenous and Industrial Catalysis

Selective Nanocatalysts and Nanoscience: Concepts for Heterogeneous and Homogeneous Catalysis

Spectroscopic Techniques in Industrial Hygiene

Spectroscopic Measurement

Catalysis

Catalysis

Advances in Catalysis

Trends in Catalysis Research

Supported metals in catalysis

The Enzymes, Vol XX: Mechanisms of Catalysis, 3rd Edition

The Enzymes, Vol XIX: Mechanisms of Catalysis, 3rd Edition

Hydrolysis of Cellulose Mechanisms of Enzymatic and Acid Catalysis

The Enzymes, Vol XX: Mechanisms of Catalysis, 3rd Edition

Recommend Documents