HomogeneousCatalysts

Piet W. N. M. van Leeuwen and John C. Chadwick

Activity Stability Deactivation

57268File AttachmentCover.jpg

Piet W.N.M. van Leeuwen

and John C. Chadwick

Homogeneous Catalysts

Further reading

Arpe, H.-J.

Industrial Organic ChemistryFifth, Completely Revised Edition

2010

ISBN: 978-3-527-32002-8

Behr, A. / Neubert, P.

Applied Homogeneous Catalysis2012

ISBN: 978-3-527-32641-9 (Hardcover)

ISBN: 978-3-527-32633-4 (Softcover)

Roberts, S. M.

Catalysts for Fine Chemical SynthesisVolumes 1-5. Set

2007

ISBN: 978-0-470-51605-8

Platz, M. S., Moss, R. A., Jones, M. (eds.)

Reviews of Reactive Intermediate Chemistry2007

ISBN: 978-0-471-73166-5

Sheldon, R. A., Arends, I., Hanefeld, U.

Green Chemistry and Catalysis2007

ISBN: 978-3-527-30715-9

Dalko, P. I. (ed.)

Enantioselective OrganocatalysisReactions and Experimental Procedures

2007

ISBN: 978-3-527-31522-2

Piet W.N.M. van Leeuwen and John C. Chadwick

Homogeneous Catalysts

Activity Stability Deactivation

The Authors

Prof. Dr. Piet W.N.M. van LeeuwenInstitute of Chemical Researchof Catalonia (ICIQ)Av. Paisos Catalans 1643007 TarragonaSpain

Dr. John C. ChadwickUniversity of EindhovenChemical Engineering & ChemistryP.O. Box 5135600 MB EindhovenThe Netherlands

All books published by Wiley-VCH are carefullyproduced. Nevertheless, authors, editors, andpublisher do not warrant the information containedin these books, including this book, to be free oferrors. Readers are advised to keep in mind thatstatements, data, illustrations, procedural details orother items may inadvertently be inaccurate.

Library of Congress Card No.: applied for

British Library Cataloguing-in-Publication DataA catalogue record for this book is available from theBritish Library.

Bibliographic information published bythe Deutsche NationalbibliothekThe Deutsche Nationalbibliothek lists thispublication in the Deutsche Nationalbibliografie;detailed bibliographic data are available on theInternet at http://dnb.d-nb.de.

# 2011 Wiley-VCH Verlag & Co. KGaA,Boschstr. 12, 69469 Weinheim, Germany

All rights reserved (including those of translation intoother languages). No part of this book may bereproduced in any form by photoprinting,microfilm, or any other means nor transmitted ortranslated into a machine language without writtenpermission from the publishers. Registered names,trademarks, etc. used in this book, even when notspecifically marked as such, are not to be consideredunprotected by law.

Typesetting Thomson Digital, Noida, IndiaPrinting and BindingCover Design Adam Design, Weinheim

Printed in SingaporePrinted on acid-free paper

Print ISBN: 978-3-527-32329-6

Contents

Preface XI

1 Elementary Steps 11.1 Introduction 11.2 Metal Deposition 21.2.1 Ligand Loss 21.2.2 Loss of H, Reductive Elimination of HX 21.2.3 Reductive Elimination of C-, N-, O-Donor Fragments 51.2.4 Metallic Nanoparticles 61.3 Ligand Decomposition by Oxidation 71.3.1 General 71.3.2 Oxidation 71.3.2.1 Catalysis Using O2 71.3.2.2 Catalysis Using Hydroperoxides 81.4 Phosphines 81.4.1 Introduction 81.4.2 Oxidation of Phosphines 91.4.3 Oxidative Addition of a PC Bond to a Low-Valent Metal 111.4.4 Nucleophilic Attack at Phosphorus 161.4.5 Aryl Exchange Via Phosphonium Intermediates 191.4.6 Aryl Exchange Via Metallophosphoranes 211.5 Phosphites 231.6 Imines and Pyridines 261.7 Carbenes 271.7.1 Introduction to NHCs as Ligands 271.7.2 Reductive Elimination of NHCs 281.7.3 Carbene Decomposition in Metathesis Catalysts 311.8 Reactions of MetalCarbon and MetalHydride Bonds 361.8.1 Reactions with Protic Reagents 361.8.2 Reactions of Zirconium and Titanium Alkyl Catalysts 371.9 Reactions Blocking the Active Sites 381.9.1 Polar Impurities 381.9.2 Dimer Formation 39

V

1.9.3 Ligand Metallation 40References 41

2 Early Transition Metal Catalysts for Olefin Polymerization 512.1 ZieglerNatta Catalysts 512.1.1 Introduction 512.1.2 Effect of Catalyst Poisons 522.1.3 TiCl3 Catalysts 532.1.4 MgCl2-supported Catalysts 542.1.4.1 MgCl2/TiCl4/Ethyl Benzoate Catalysts 542.1.4.2 MgCl2/TiCl4/Diester Catalysts 562.1.4.3 MgCl2/TiCl4/Diether Catalysts 572.1.5 Ethene Polymerization 572.2 Metallocenes 582.2.1 Introduction 582.2.2 Metallocene/MAO Systems 622.2.3 Metallocene/Borate Systems 662.3 Other Single-Center Catalysts 692.3.1 Constrained Geometry and Half-Sandwich Complexes 692.3.2 Octahedral Complexes 732.3.3 Diamide and Other Complexes 752.4 Vanadium-Based Catalysts 762.5 Chromium-Based Catalysts 802.6 Conclusions 82

References 83

3 Late Transition Metal Catalysts for Olefin Polymerization 913.1 Nickel- and Palladium-based Catalysts 913.1.1 Diimine Complexes 913.1.2 Neutral Nickel(II) Complexes 943.1.3 Other Nickel(II) and Palladium(II) Complexes 983.2 Iron- and Cobalt-based Catalysts 983.2.1 Bis(imino)Pyridyl Complexes 983.3 Conclusions 101

References 102

4 Effects of Immobilization of Catalysts for OlefinPolymerization 105

4.1 Introduction 1054.2 Metallocenes and Related Complexes 1064.2.1 Immobilized MAO/Metallocene Systems 1064.2.2 Immobilized Borane and Borate Activators 1094.2.3 Superacidic Supports 1104.2.4 MgCl2-Supported Systems 1104.3 Other Titanium and Zirconium Complexes 113

VI Contents

4.3.1 Constrained Geometry Complexes 1134.3.2 Octahedral Complexes 1154.4 Vanadium Complexes 1174.5 Chromium Complexes 1214.6 Nickel Complexes 1224.7 Iron Complexes 1244.8 Conclusions 125

References 126

5 Dormant Species in Transition Metal-CatalyzedOlefin Polymerization 131

5.1 Introduction 1315.2 ZieglerNatta Catalysts 1325.2.1 Ethene Polymerization 1325.2.2 Propene Polymerization 1325.3 Metallocenes and Related Early Transition Metal

Catalysts 1345.3.1 CationAnion Interactions 1345.3.2 Effects of AlMe3 1365.3.3 Effects of 2,1-insertion in Propene Polymerization 1375.3.4 Effects of Z3-allylic Species in Propene Polymerization 1405.3.5 Chain Epimerization in Propene Polymerization 1415.3.6 Effects of Dormant Site Formation on Polymerization

Kinetics 1425.4 Late Transition Metal Catalysts 1435.4.1 Resting States in Nickel Diimine-Catalyzed

Polymerization 1435.4.2 Effects of Hydrogen in Bis(iminopyridyl) Iron-Catalyzed

Polymerization 1435.5 Reversible Chain Transfer in Olefin Polymerization 1455.6 Conclusions 147

References 148

6 Transition Metal Catalyzed Olefin Oligomerization 1516.1 Introduction 1516.2 Zirconium Catalysts 1526.3 Titanium Catalysts 1536.4 Tantalum Catalysts 1566.5 Chromium Catalysts 1576.5.1 Chromium-catalyzed Trimerization 1576.5.2 Chromium-catalyzed Tetramerization of Ethene 1606.5.3 Chromium-Catalyzed Oligomerization 1626.5.4 Single-component Chromium Catalysts 1646.6 Nickel Catalysts 1666.7 Iron Catalysts 168

Contents VII

6.8 Tandem Catalysis involving Oligomerization andPolymerization 170

6.9 Conclusions 171References 172

7 Asymmetric Hydrogenation 1777.1 Introduction 1777.2 Incubation by Dienes in Rhodium Diene

Precursors 1797.3 Inhibition by Substrates, Solvents, Polar Additives,

and Impurities 1817.3.1 Inhibition by Substrates: Iridium 1817.3.2 Inhibition by Substrates, Additives: Rhodium 1827.3.3 Inhibition by Substrates: Ruthenium 1877.4 Inhibition by Formation of Bridged Species 1907.4.1 Inhibition by Formation of Bridged Species:

Iridium 1917.4.2 Inhibition by Formation of Bridged Species:

Rhodium 1957.5 Inhibition by Ligand Decomposition 1987.6 Inhibition by the Product 1997.6.1 Inhibition by the Product: Rhodium 1997.6.2 Ruthenium 2007.7 Inhibition by Metal Formation; Heterogeneous

Catalysis by Metals 2017.8 Selective Activation and Deactivation of Enantiomeric

Catalysts 2047.9 Conclusions 206

References 207

8 Carbonylation Reactions 2138.1 Introduction 2138.2 Cobalt-Catalyzed Hydroformylation 2148.3 Rhodium-Catalyzed Hydroformylation 2178.3.1 Introduction of Rhodium-Catalyzed Hydroformylation 2178.3.2 Catalyst Formation 2218.3.3 Incubation by Impurities: Dormant Sites 2238.3.4 Decomposition of Phosphines 2278.3.5 Decomposition of Phosphites 2318.3.6 Decomposition of NHCs 2358.3.7 Two-Phase Hydroformylation 2388.3.8 Hydroformylation by Nanoparticle Precursors 2448.4 Palladium-Catalyzed AlkeneCO Reactions 2448.4.1 Introduction 2448.4.2 Brief Mechanistic Overview 246

VIII Contents

8.4.3 Early Reports on Decomposition and Reactivation 2488.4.4 Copolymerization 2508.4.5 Methoxy- and Hydroxy-carbonylation 2538.5 Methanol Carbonylation 2598.5.1 Introduction 2598.5.2 Mechanism and Side Reactions of the Monsanto

Rhodium-Based Process 2608.5.3 The Mechanism of the Acetic Anhydride Process Using

Rhodium as a Catalyst 2618.5.4 Phosphine-Modified Rhodium Catalysts 2638.5.5 Iridium Catalysts 2658.6 Conclusions 268

References 269

9 Metal-Catalyzed Cross-Coupling Reactions 2799.1 Introduction; A Few Historic Notes 2799.2 On the Mechanism of Initiation and Precursors 2839.2.1 Initiation via Oxidative Addition to Pd(0) 2839.2.2 Hydrocarbyl Pd Halide Initiators 2909.2.3 Metallated Hydrocarbyl Pd Halide Initiators 2939.3 Transmetallation 2999.4 Reductive Elimination 3039.4.1 Monodentate vs Bidentate Phosphines and Reductive

Elimination 3039.4.2 Reductive Elimination of CF Bonds 3139.5 Phosphine Decomposition 3169.5.1 Phosphine Oxidation 3169.5.2 PC Cleavage of Ligands 3179.6 Metal Impurities 3229.7 Metal Nanoparticles and Supported Metal

Catalysts 3279.7.1 Supported Metal Catalysts 3279.7.2 Metal Nanoparticles as Catalysts 3309.7.3 Metal Precipitation 3349.8 Conclusions 334

References 335

10 Alkene Metathesis 34710.1 Introduction 34710.2 Molybdenum and Tungsten Catalysts 34910.2.1 Decomposition Routes of Alkene Metathesis Catalysts 34910.2.2 Regeneration of Active Alkylidenes Species 35610.2.3 Decomposition Routes of Alkyne Metathesis Catalysts 35910.3 Rhenium Catalysts 36310.3.1 Introduction 363

Contents IX

10.3.2 Catalyst Initiation and Decomposition 36510.4 Ruthenium Catalysts 37010.4.1 Introduction 37010.4.2 Initiation and Incubation Phenomena 37110.4.3 Decomposition of the Alkylidene Fragment 37610.4.4 Reactions Involving the NHC Ligand 37910.4.5 Reactions Involving Oxygenates 38110.4.6 Tandem Metathesis/Hydrogenation Reactions 38510.5 Conclusions 388

References 390

Index 397

X Contents

Preface

Homogeneous catalysts have played a key role in the production of petrochemicalsand coal-derived chemicals since the 1960s. In the last two decades, transition metalcatalysts have revolutionized synthetic organic chemistry, both in the laboratory andin industrial production. The use of homogeneous catalysts in polyolefin synthesisstarted in the 1980s and triggered enormous R&D efforts, leading to hithertoinaccessible polymers and to greatly improved control over polymer structure andproperties. The introduction of new processes and catalysts continues in bulkchemical production, as exemplified by new routes that have recently come onstream for the production of 1-octene and methyl methacrylate.

For all catalysts, selectivity and rates of reactions are crucial parameters and in thelaboratory even the rate may not concern us that much, as catalyst loadings of 5% ormore are often applied. For industrial applications, however, high turnover numbersare required for economic reasons, whichmay bemore complex than simply catalystcosts. For the bulk chemical applications, studies of catalyst activation, activity,stability, deactivation, recycling and regeneration have always formed an integralpart of catalysis research. A considerable research effort has been devoted to this,mainly in industry, but explicit publications are scarce, although some stabilityissues can be deduced from the patent literature. Catalyst stability has been a highlyimportant factor in transforming advances in catalysis into practical applications,notably in the areas of polymer synthesis, cross-coupling chemistry, hydrogenationcatalysis, carbonylation reactions and metathesis chemistry.

In heterogeneous catalysis, the study of activation, deactivation, and regenerationof catalysts has always been a major research activity. These topics have beenaddressed in many articles, books and conferences, and literature searches withthese keywords givemany relevant results. For homogeneous catalysts this is not thecase, with the possible exception of metathesis. A wealth of knowledge can be foundin a vast number of publications, but this is not easily retrieved. The approach inhomogeneous catalysis is entirely different to that of heterogeneous catalysis,especially before industrial applications come into sight; in homogeneous catalysis,the general approach to improving the catalyst performance is variation of one of thecatalytic components, without much attention being paid to the question of whyother catalyst systems failed.

XI

In this book, we address a number of important homogeneous catalysts, focusingon activity, stability and deactivation, including the important issue of how deactiva-tion pathways can be avoided. Key concepts of activation and deactivation, togetherwith typical catalyst decomposition pathways, are outlined in the first chapter.Chapters 26 cover homogeneous catalysts for olefin polymerization and oligomer-ization, including the effects of catalyst immobilization and polymerization ratelimitation as a result of dormant site formation. The following sections of the book,Chapters 710, describe catalyst activity and stability in asymmetric hydrogenation,hydroformylation, alkene-CO reactions, methanol carbonylation, metal-catalyzedcross-coupling catalysis and, finally, alkene metathesis.We hope that the contents of this book will be valuable to many scientists working

in the field of homogeneous catalysis and that the inclusion of a broad range oftopics, ranging from polymerization catalysis to the synthesis of speciality and bulkchemicals, can lead to useful cross-fertilization of ideas.We would like to acknowledge useful comments and contributions from

Rob Duchateau, Peter Budzelaar and Nick Clment. We thank Marta Moya andMara Jos Gutirrez for polishing the final draft. We are also indebted to ManfredKohl from Wiley-VCH, for convincing one of us at the 10th InternationalSymposium on Catalyst Deactivation to embark on this book project. We thankhim, Lesley Belfit and their team for the perfect support provided throughout.

February 2011 Piet W.N.M. van LeeuwenJohn C. Chadwick

XII Preface

Piet van Leeuwen (1942) is group leader at the Institute ofChemical Research of Catalonia in Tarragona, Spain, since2004. After receiving his PhD degree at Leyden University in1967 he joined Shell Research in 1968. Until 1994 he headed aresearch group at Shell Research in Amsterdam, studying manyaspects of homogeneous catalysis. He was Professor of Homo-geneous Catalysis at the University of Amsterdam from 1989until 2007. He has coauthored 350 publications, 30 patents,and many book chapters, and is author of the book Homoge-

neous Catalysis: Understanding the Art. He (co)directed 45 PhD theses. In 2005 he wasawarded the Holleman Prize for organic chemistry by the Royal Netherlands Academy. In2009 he received a doctorate honoris causa from the University Rovira I Virgili, Tarragona,and he was awarded a European Research Council Advanced Grant.

John Chadwick was born in 1950 in Manchester, England andreceived his B.Sc. and Ph.D. degrees from the University ofBristol, after which he moved to The Netherlands, joining ShellResearch in Amsterdam in 1974. He has been involved inpolyolefin catalysis since the mid 1980s and in 1995 transferredfrom Shell to the Montell (later Basell) research center inFerrara, Italy, where he was involved in fundamental ZieglerNatta catalyst R&D. From 2001 to 2009, he was at EindhovenUniversity of Technology on full-time secondment from Basell

(now LyondellBasell Industries) to the Dutch Polymer Institute (DPI), becoming DPIProgramme Coordinator for Polymer Catalysis and Immobilization. Until his retirementin 2010, his main research interests involved olefin polymerization catalysis, including theimmobilization of homogeneous systems, and the relationship between catalyst and poly-mer structure. He is author or co-author of more than 60 publications and 11 patentapplications.

Preface XIII

1Elementary Steps

1.1Introduction

Catalyst performance plays a central role in the literature on catalysis and is expressedin terms of selectivity, activity and turnover number.Most often catalyst stability is notaddressed directly by studying why catalysts perform poorly, but by varying condi-tions, ligands, additives, andmetal, in order to find a better catalyst. One approach tofinding a suitable catalyst concerns the screening of ligands, or libraries of ligands [1]using robotics; especially, supramolecular catalysis [24] allows the fast generation ofmany new catalyst systems. Another approach is to study the decompositionmechanism or the state the catalyst precursor is in and why it is not forming anactive species. For several important reactions such studies have been conducted, butthey are low in number. As stated in the preface, in homogeneous catalysis there hasalways been less attention given to catalyst stability [5] than there is in heterogeneouscatalysis [6]. We favor a combined approach of understanding and exploration,without claiming that this is more efficient. In the long term this approach maybe the winner for a reaction that we have got to know in much detail. For reactions,catalysts, or substrates that are relatively novel a screening approach is much moreefficient, as shown bymany examples during the last decade; we are not able to studyall catalysts in the detail required to arrive at a level at which our knowledge will allowus to make predictions. We can reduce the huge number of potential catalysts(ligands, metals, co-catalysts) for a desired reaction by taking into account what weknow about the decomposition reactions of our coordination or organometalliccomplexes and their ligands. Free phosphines can be easily oxidized and phosphitescan be hydrolyzed and thus these simple combinations of ligands and conditions canbe excluded from our broad screening program. In addition we can make sophis-ticated guesses as to what elsemight happen in the reaction with catalysts that we areabout to test and we can reduce our screening effort further. To obtain a betterunderstanding we usually break down the catalytic reaction under study intoelementary steps, which we often know in detail from (model) organometallic ororganic chemistry. Asmany books do,we can collect elementary steps and reverse theprocess and try to design new catalytic cycles. We can do the same for decomposition

Homogeneous Catalysts: Activity Stability Deactivation, First Edition. Piet W.N.M. van Leeuwenand John C. Chadwick. 2011 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2011 by Wiley-VCH Verlag GmbH & Co. KGaA.

j1

processes and first look at their elementary steps [7]; the process may be a single stepor more complex, and even autocatalytic. In this chapter we will summarize theelementary reactions leading to the decomposition of the metal complexes andthe ligands, limiting ourselves to the catalysis that will be dealt with in the chaptersthat follow.

1.2Metal Deposition

Formation of ametallic precipitate is the simplest andmost commonmechanism fordecomposition of a homogeneous catalyst. This is not surprising, since reducingagents such as dihydrogen, metal alkyls, alkenes, and carbon monoxide are thereagents often used. A zerovalent metal may occur as one of the intermediates ofthe catalytic cycle, which might precipitate as metal unless stabilizing ligands arepresent. Precipitation of the metal may be preceded by ligand decomposition.

1.2.1Ligand Loss

A typical example is the loss of carbon monoxide and dihydrogen from a cobalthydrido carbonyl, the classic hydroformylation catalyst (Scheme 1.1).

The resting state of the catalyst is either HCo(CO)4 or RC(O)Co(CO)4, and bothmust lose onemolecule of CObefore further reaction can take place. Thus, loss of COis an intricate part of the catalytic cycle, which includes the danger of complete loss ofthe ligands giving precipitation of the cobalt metal. Addition of a phosphine ligandstabilizes the cobalt carbonyl species forming HCo(CO)3(PR3) and, consequently,higher temperatures and lower pressures are required for this catalyst in thehydroformylation reaction.

A well-known example of metal precipitation in the laboratory is the formation ofpalladium black, during cross coupling or carbonylation catalysis with the useof palladium complexes. Usually phosphorus-based ligands are used to stabilizepalladium(0) and to prevent this reaction.

1.2.2Loss of H , Reductive Elimination of HX

The loss of protons from a cationicmetal species, formally a reductive elimination, isa common way to form zerovalent metal species, which, in the absence of stabilizing

HCo(CO)2 4H- 2

Co2(CO)8CO-

metalCo

Scheme 1.1 Precipitation of cobalt metal.

2j 1 Elementary Steps

ligands, will lead to metal deposition. Such reactions have been described for metalssuch as Ru, Ni, Pd, and Pt (Scheme 1.2). The reverse reaction is a common way toregenerate ametal hydride of the late transitionmetals and clearly the position of thisequilibrium will depend on the acidity of the system.

Too strongly acidic media may also lead to decomposition of the active hydridespecies via formation of dihydrogen and a di-positively charged metal complex(reaction (2), Scheme 1.2). All these reactions are reversible and their course dependson the conditions.

As shown in the reaction schemes for certain alkene hydrogenation reactions andmost alkene oligomerization reactions (Schemes 1.3 and 1.4), the metal maintainsthe divalent state throughout, and the reductive elimination is not an indissolublepart of the reaction sequence.

The species LnMH are stabilized by phosphine donor ligands, as in the Shell

Higher Olefins Process (MNi) [8] and in palladium-catalyzed carbonylationreactions [9].

We mention two types of reactions for which the equilibrium, shown inScheme 1.2, between MH and M H is part of the reaction sequence, theaddition of HX to a double bond and the Wacker reaction. As an example of an HXaddition we will take hydrosilylation, as for HCN addition the major decomposition

P

Pd

P S 2+

P

Pd

P S

H

+

BH+ (2)

B

+

H

H

P

Pd

P S

H

+

BBH+

+

P

Pd

P

S(1)

P

Pd

P S

S

2+-H2

Scheme 1.2 Reactions involving protons and metal hydrides.

P

Ru

P CH2

P

Ru

P

H

+

H

H

P

Ru

P CH2+

H2

P

Ru

P

H

+

CH2=CH2

CH3

+

CH3

C- 2H6

Scheme 1.3 Simplified scheme for heterolytic hydrogenation.

1.2 Metal Deposition j3

reaction is a different one, aswewill see later. Thehydrosilylation reaction is shown inScheme 1.5 [10].

In the Wacker reaction, elimination of HCl from PdHCl leads to formation ofpalladium zero [11] and the precipitation of palladiummetal is often observed in theWacker reaction or related reactions [12]. In the Wacker process palladium(II)oxidizes ethene to ethanal (Scheme 1.6) and, since the re-oxidation of palladiumbymolecular oxygen is too slow, copper(II) is used as the oxidizing agent. Phosphineligands cannot be added as stabilizers for palladium zero, because they would beoxidized. In addition, phosphine ligandswouldmake palladium less electrophilic, animportant property of palladium in the Wacker reaction.

In the palladium-catalyzed Heck reaction (Scheme 1.7), as in other cross couplingreactions, the palladium zero intermediate should undergo oxidative addition beforeprecipitation of themetal can occur. Alternatively, Pd(0) can be protected by ligandspresent, as in the example of Scheme 1.7, but this requires another dissociation stepbefore oxidative addition can occur. Both effective ligand-free systems [13] and ligand-containing systemshave been reported [14]. Apolarmediumaccelerates the oxidativeaddition. The second approach involves the use of bulky ligands, which give rise tolow coordination numbers and hence electronic unsaturation and more reactivespecies. Turnover numbers of millions have been reported [15].

P

Pd

P (C2H4)n+1

P

Pd

P

H

+

P

Pd

P CH2+

P

Pd

P

H

+

CH2=CH2

CH3

+

alkenes-

CHn 2=CH2

H

Scheme 1.4 Alkene oligomerization.

CH3Cl3Si

H2C

CH3

SiCl3

H2C

Pt

CH2CH2

SiCl3

HPt

CH2CH2

SiCl3

HPt

HSiCl3"Pt"clusters,metal

Scheme 1.5 Simplified mechanism for hydrosilylation.

4j 1 Elementary Steps

1.2.3Reductive Elimination of C-, N-, O-Donor Fragments

Many cross-coupling reactions have been reported in the last decades. The palladiumand nickel-catalyzed formation of CC, CN, CO, and CP bonds has become animportant tool for organic syntheses [16]. The general mechanism for CC bond

Pd metal

Cl OH2Cl

Pd

Cl

Cl Cl

Cl

Pd

Cl

Cl OH2Pd

Cl

OHH2C

Cl OH2

CH2

Pd

Cl

CH2H2C

H

OH3C

CH2H2C

Pd2 Cl-H3O

+

2 Cu++

2 Cu+ ,H+

1/2 O2

(-)

(-)

2-

H2O

H+

H2OCl-

Cl-

OH-

Scheme 1.6 Ethanal formation from ethene via a Wacker oxidation reaction.

CO2R

H

Br

L

L

Pd

RO2C

Br

L

L

Pd

CO2R

BrL

L

PdPd

L

L

L

L

L2

basebase-HBr+PdL4+

L2++

BrL

L

Pd

Br

Scheme 1.7 The mechanism of the Heck reaction using excess phosphine.

1.2 Metal Deposition j5

formation is depicted in Scheme 1.8. Again, the zerovalent state of the metal is anintrinsic part of themechanism.As in theHeck reaction, the phosphine ligandsmustprevent metal deposition and/or oxidative addition of a hydrocarbyl halide should befaster than metal deposition.

1.2.4Metallic Nanoparticles

Formation of metal agglomerates probably starts with dimer and trimer formationand occasionally this has been observed via mass spectroscopy or EXAFS; the lattercase concerns a palladium diphosphine catalyst forming first dimeric and trimericspecies before clusters were observed [17]. On the way frommetal complexes to bulkmetal particles the system passes undoubtedly through the stage of metal nanopar-ticles (MNPs). Often this can be deduced from the intermediate yellow, green andblue solutions before a black precipitate is observed. MNPs can also be synthesizedon purpose and used as a catalyst [18]. The selective formation of giant clusters orMNPs [19] can be regulated by the conditions, metal to ligand ratios, stabilizingagents [20] such as polymers, solid surfaces [21], ionic liquids [22], surfactants [23],and dendrimers [24]. MNPs are soluble, recyclable species, which may present anintermediate between homogeneous and heterogeneous catalysts [25]. Reactionstypical of heterogeneous catalysts, such the hydrogenation of alkenes and aromatics,may take place on the surface of theMNPs, andmost likely they remain intact [26]. Inoxidation catalysis palladiumgiant clusters (called such at the time) have been knownfor quite some time [27], but the nature of the actual catalyst is not known. Reactionswith PdNP catalysts that strongly resemble homogeneous catalytic processes, such ascross coupling, theHeck reaction, and allylic alkylationhave been the subject ofmuchdiscussion as to whether the PdNP serves as the catalyst or as a sink/precursor formonometallic complexes [28]. Ligand-free palladium atoms (solvated, though) areprobably very active catalysts in CC coupling reactions and this may explain whynanoparticles can lead to active catalysts, and even to efficient recycling, as onlya very small amount of the catalyst precursor is consumed in each cycle. Evenasymmetric MNP catalysts have been reported, and examples include Pt-catalyzedhydrogenation of ethyl pyruvate [29], Pd-catalyzed hydrosilylation of styrene [30], and

L

L

Pd

L Br

PhLPd

L

MgBr

PhL

Pd

L

MgBr2 L

L

Pd

L

-L

eliminationreductive

transmetallationoxidative addition

+

++L

+

PhBr

Scheme 1.8 General mechanism for cross-coupling reactions.

6j 1 Elementary Steps

Pd-catalyzed allylic alkylation of racemic substrates [31]. Modification of surfaceswith chiral molecules has been known for several decades to give rise to enantio-selective catalysis [32], but the similarity of ligands used in homogeneous and MNP-based enantioselective catalysis seems suspect. Evidence is growing that the latterreactions are catalyzed by homogeneous complexes [28, 33].

In the pyridine-palladium acetate catalyzed oxidation of alcohols the formation ofPdNPs was observed by transmission electron microscopy (TEM) measure-ments [34], but, by using dendritic pyridine ligands containing a 2,3,4,5-tetraphe-nylphenyl substituent at the 3-position of the pyridine ring, this was suppressedsuccessfully by Tsuji and coworkers [35].

A key issue for homogeneous catalysis is that MNPs can form in a reversiblemanner, while the formation of larger metal particles is usually irreversible, boththermodynamically and/or kinetically. MNPs still hold promise for new reactions tobe discovered and as precursors for molecular catalysts they have shown advantages,but the control of their size during catalysis seems an intrinsic problem not to besolved easily.

1.3Ligand Decomposition by Oxidation

1.3.1General

The main tool for catalyst modification in homogeneous, catalytic processes ismodification of the ligands. By changing the ligand properties we try to obtainbetter selectivity and activity. Decomposition of the ligands and their complexes hasa large influence on the performance of the system. Catalysts based on late transitionmetals often contain phosphites and phosphines as modifying ligands. The decom-position routes of these ligands have received a great deal of attention. They aresensitive to many reactions as we will see. Nitrogen-based ligands, such as amines,imines and pyridines, are much more robust in general, but they are much lesseffective as ligands for low-valent late transition metals, such as rhodium(I) inrhodium-catalyzed hydroformylation. In ionic complexes though, we have seen anenormous increase in the use of nitrogen donor ligands in catalytic reactions thathave become highly efficient and selective.

1.3.2Oxidation

1.3.2.1 Catalysis Using O2Homogeneous catalysts for oxidation reactions using O2 do not contain modifyingsoft ligands, but they are ionic species solvated by water, acetic acid, and the like.Examples include the Wacker process for making ethanal (palladium in water) andthe oxidation of p-xylene to terephthalic acid (cobalt in acetic acid) [36].

1.3 Ligand Decomposition by Oxidation j7

Ligands based on nitrogen donor atoms are the ligands of choice; they stabilizehigh-valent metal ions and are not as sensitive to oxidation as phosphorus- or sulfur-based ligands. For example, phenanthroline ligands were used for the palladium-catalyzed oxidation of alcohols to ketones or aldehydes [37], and diamines areeffective ligands for the copper-catalyzed oxidative coupling of phenols in thesynthesis of polyphenylene ether [38]. We are not aware of commercial processesutilizing polydentate nitrogen ligands yet, although many interesting new oxidationcatalysts have been reported in recent years [39]. Oxidation of the ligand backbonemay be a concern as even porphyrins should be used in halogenated form in order toenhance their stability in oxidation reactions [40].

1.3.2.2 Catalysis Using HydroperoxidesThe commercial processes using hydroperoxides (t-butyl hydroperoxide and 1-phenylethyl hydroperoxide) for the epoxidation of propene involve ligand-freemetals such as titanium alkoxides and ligand oxidation is not an issue for theseprocesses [41]. For the asymmetric epoxidation using Sharpless catalyst, ligandoxidation is also not a major issue [42].

Phosphorus ligands are very prone to oxidation. Therefore, oxygen and hydro-peroxides have to be thoroughly removed from our reagents and solvents beforestarting our catalysis. In spite of this common knowledge, oxidation of phosphineligands has frequently obscured the catalytic results.

When phosphines are bonded strongly to a transition metal such that no or littledissociation occurs, their oxidation by hydroperoxides will not take place. This is thecase for the platinum-catalyzed epoxidation reaction of alkenes by hydrogen peroxidedeveloped by Strukul [43]. The bidentate phosphine ligands survive the hydro-peroxidic conditions and asymmetric and regioselective epoxidations have beenachieved, proving that the chiral ligands remain intact and coordinated to platinum.Typically, turnover numbers are 50 to100, and while the use of hydrogen peroxide isattractive froma green chemistry point of view, thesemodest numbers have so far notled to industrial applications. Clearly, from a cost point of view, the Sharpless catalystseems more attractive.

1.4Phosphines

1.4.1Introduction

Phosphines and diphosphines are widely used as the ligand component in homo-geneous catalysts. Large-scale processes include rhodium-catalyzed hydroformyla-tion for propene, butene, and heptene, ethene oligomerization, cobalt-catalyzedhydroformylation for internal higher alkenes, and butadiene dimerization. Small-scale operations include asymmetric hydrogenation of enamides and substitutedacrylic acids, asymmetric isomerization to make menthol, alkoxycarbonylation

8j 1 Elementary Steps

(ibuprofen), and Group 10 metal-catalyzed CC bond formation (Heck reaction,Suzuki reaction). Future possibilities may comprise selective trimerization andtetramerization of ethene [44], more alkoxycarbonylations (large-scale methyl meth-acrylate, more pharmaceuticals), hydroxycarbonylations, a large variety of newCC [45] and CN coupling reactions, asymmetric alkene dimerizations, alkenemetathesis, and new hydroformylation reactions.

Garrou [46] reviewed the decomposition of phosphorus ligands in relation tohomogeneous catalysis many years ago. Many interesting studies on phosphorusligand decomposition have appeared since, but Garrous review is still a usefulcollection of highly relevant reactions. At the time, the formation of phosphidospecies seemed to be the most common fate of our phosphine-containing catalysts,but in the last twodecadesmanymore reaction types have been added, aswe reviewedin 2001 [7]. In a recent reviewParkins discusses reactions taking place in the ligand inthe coordination sphere, a large number of them being examples of phosphines [47].The reactions in Parkins review are ordered by metal. Exchange of substituents atphosphorus in metal complexes has been reviewed by Macgregor recently and hisreview shows that many new reactions have been discovered since the review byGarrou was published [48].

1.4.2Oxidation of Phosphines

Oxidation of free phosphines was mentioned above (Section 1.3.2.2) as a reactionleading to phosphine loss. Phosphines are used extensively in a large number oforganic synthetic reactions in which they usually end up as the phosphine oxide, thatis, they are used as reducing agents. Well known examples are the Mitsunobureaction to generate esters from alcohols and carboxylic acids under very mildconditions, and the Appel reaction to convert alcohols into alkyl halides, and theWittig reaction. Therefore it is not surprising that, in catalysis, oxidation of phos-phines is a common way to deactivate catalytic systems. Common oxidizing agentsare dioxygen and hydroperoxides. High-valent metals may also function as theoxidizing agent. Sometimes this reaction is utilized on purpose and the reducingfunction of phosphines is used to activate the catalyst; for example palladium(II)acetate can be reduced by an excess of phosphine ligand (see Scheme 1.9, thirdreaction).

PR3 + H2O H2 + O=PR3

CO + O=PR3PR3 + CO2

PR3 + Pd(OAc)2 + H2O Pd(0) + 2 HOAc + O=PR3

PR3 + 1/2 O2 O=PR3

Scheme 1.9 Examples for oxidation of phosphines.

1.4 Phosphines j9

For instance, in cross coupling chemistry a palladium(II) precursor is reduced insitu by phosphine to generate Pd(0), the active species (the oxygen atom is providedby water). Molybdenum(VI), tungsten(VI) and water have been reported as oxidizingagents of phosphines [49]. Rh(III) carbonate oxidizes triphenylphosphine formingCO2 and Rh(I) [50]. Rh(III) in water was found to oxidize tppts yielding tppts-O andRh(I) [51]. Thermodynamics show that even water and carbon dioxide may oxidizephosphines to the corresponding oxides. These reactions may be catalyzed by thetransition metal in the system, for example, Rh for CO2 [52]. Water oxidizesphosphines using palladium as a catalyst and palladium has to be thoroughlyremoved after its use in a PCcross coupling synthesis [53]. It should bementioned,however, that, in view of the many successful applications of water and sCO2 assolvents in homogeneous catalysis, these oxidation reactions are relatively rare.

Oxidation, or partial oxidation of phosphine can also be turned into a usefulreaction if an excess of phosphine retards the catalytic reaction. Above we havementioned that phosphine-free palladium compounds may be very active catalystsfor cross coupling reactions, and, thus, intentional or accidental ingress of oxygenmay be advantageous for the catalysis. Another example is the oxidation of one of thephosphine molecules in the Grubbs I metathesis catalyst.

Nitro and nitroso compounds are strongly oxidizing agents and, for instance, theyhave been reported to oxidize PH3 [54]. Thus, nitrobenzene and phosphine giveazoxybenzene and phosphorus acids under harsh conditions. In the palladium-catalyzed reductive carbonylation of nitrobenzene it was found that phosphineligands are not suitable as they are oxidized to phosphine oxides [55]. Nitrosobenzeneand isocyanate complexes of zerovalent Group 10 metals will transfer oxygen totriphenylphosphine and also form azoxybenzene [56]. Nitrosobenzene ismuchmorereactive than nitrobenzene towards phosphines as it will oxidize arylphosphines inthe absence ofmetal catalysts, forming azoxybenzenes at ambient temperature in thepresence of base [57].

Many sulfur-containing compounds will also oxidize phosphines, and either formphosphine sulfide or, when water is present, phosphine oxides. This reaction hasbeen known since 1935 [58] and, especially with water-soluble tris(2-carboxyethyl)phosphine (TCEP), it is of interest in biochemical systems [59]. It has been studieda couple of times over the years, but only in the last decade has it become extremelypopular in biochemistry and molecular biology to reduce protein disulfide bonds(Scheme 1.10), for example in labeling studies, and as a preparatory step for gelelectrophoresis [60].

HO P OH

O

HO O

O

RS

SR

OHH

+HO P OH

O

HO O

O

RSH

HSR+

O

Scheme 1.10 Reduction of disulfides by TCEP.

10j 1 Elementary Steps

TCEP has also been employed for the reduction of sulfoxides, sulfonylchlorides,N-oxides, and azides (Staudinger reaction), thus showing that these compounds alsopresent a potential hazard for phosphines in catalytic systems [61].

1.4.3Oxidative Addition of a PC Bond to a Low-Valent Metal

In the next four sections we will discuss four additional ways of phosphinedecomposition: oxidative addition of phosphines to low-valent metal complexes,nucleophilic attack on coordinated phosphines, aryl exchange via phosphoniumspecies, and substituent exchange via metallophosphorane formation. Interestingly,in all cases the metal serves as the catalyst for the decomposition reaction!

In his review [46a] Garrou emphasizes the first mechanism, oxidative additionof the phosphoruscarbon bond to low-valentmetal complexes (or reductive cleavageof PC bonds) and formation of phosphido species. In the last two decadesexperimental support for the other three mechanisms has been reported(Sections 1.4.41.4.6). In Scheme 1.11 the four mechanisms are briefly outlined.

Reductive cleavage of the phosphoruscarbon bond in triaryl- or diarylalkylpho-sphines is an important tool formaking new phosphines [62]. Metals used to this endin the laboratory are lithium, sodium (or sodium naphthalide), and potassium.Cleavage of triphenylphosphine with sodium in liquid ammonia to give Ph2P

,benzene, and NaNH2 is carried out on an industrial scale for the synthesis of theligand of the SHOPprocess, obtained via reaction of sodiumdiphenylphosphidewitho-chlorobenzoic acid [63]. The cleavage reaction works well for phenyl groups andmethyl and several methoxy-substituted phenyl groups; most other substitutionpatterns lead to a Birch reaction or cleavage of the functional group [62b]. It is notsurprising, therefore, that low-valent transition metals will also show reductivecleavage of the PC bond, although mechanistically it involves interaction of themetal with the carbon and phosphorus atoms rather than an initial electron transferas is the case for the alkali metals. The reaction with transition metals is usuallyreferred to as an oxidative addition of the R0PR2 molecule to the metal complex.

Ph PhPh

M

P

Ph Ph

P

MPh

Ph

PhPh

ORM

POR

PhPh

P

MPh

Ar

PhPh

Pd

PPh

Ar

PhPhPPhPd

Phosphonium ion formation

+

+

+

Oxidative addition Nucleophilic attack

Nu

PhPh

M

PPh Nu

Ph

Ph

M

PPh

Phosphorane formation

Scheme 1.11 Four mechanisms for phosphine decomposition leading to PC bond cleavage.

1.4 Phosphines j11

Oxidative addition of CBr or CCl bonds is an important reaction in cross-coupling type catalysis, and the reaction of a PC bond is very similar, although thebreaking of carbonphosphorus bonds is not a useful reaction in homogeneouscatalysis. It is an undesirable side-reaction that occurs in systems containingtransition metals and phosphine ligands, leading to deactivation of the catalysts.Indeed, the oxidative addition of a phosphine to a low valent transition metal can bemost easily understood by comparing the Ph2P fragment with a chloro- or bromo-substituent at the phenyl ring; electronically they are very akin, see Hammettparameters and the like. The phosphido anion formed during this reaction willusually lead to bridged bimetallic structures, which are extremely stable. Thedecomposition of ligands during hydroformylation, which has been reported bothfor rhodium and cobalt catalysts [64] may serve as an example.

Thermal decomposition of RhH(CO)(PPh3)3, the well known hydroformylationcatalyst, in the absence of H2 and CO leads to a stable cluster, shown in Figure 1.1,containing m2-PPh2 fragments, as was studied by Pruetts group at Union Carbide(now Dow Chemical) [65]. It is not known whether PC cleavage takes place ona cluster or whether it starts with a monometallic species (see the reactions belowtaking place in clusters).

After heating, the corresponding iridium compound led to the formation ofa dimer containing two bridging phosphido bridges. The phenyl group has beeneliminated (as benzene or diphenyl), see Scheme1.12. In viewof the short IrIr bondthe authors suggested a double bond [66].

Several authors have proposed a mechanism involving orthometallation as a firststep in the degradation of phosphine ligands, especially in the older literature.Orthometallation does take place, as can be inferred from deuteration studies, but itremains uncertain whether this is a reaction necessarily preceding the oxidativeaddition (Scheme 1.13).

Ph2P

Ph3P

CO

PPhCO 3

CO

Ph2P

PPh2Rh

Rh

Rh

Figure 1.1 Rhodium cluster that may form in hydroformylation mixture work-up.

Ir(PPh3)3(CO)Hdecalin, T

Ph2P

Ir

PPh2

IrCO

PPh3

Ph3P

OC

Scheme 1.12 Iridium dimer that forms in the absence of syn gas.

12j 1 Elementary Steps

Subsequently the phosphoruscarbon bond is broken and the benzyne interme-diate inserts into the metal hydride bond. Although this mechanism has beenpopular withmany chemists there aremany experiments that contradict it. A simplepara-substitution of the phenyl group answers the question whether orthometalla-tion was involved, as is shown in Scheme 1.14.

Decomposition products of p-tolylphosphines should contain methyl substituentsin the meta position if the orthometallation mechanism were operative. For palla-dium-catalyzed decomposition of triarylphosphines this was found not to be thecase [67]. Using rhodium-containing solutions of tri-o-, tri-m-, and tri-p-tolylpho-sphines Abatjoglou et al. found that only one isomeric tolualdehyde is formedfrom each phosphine [68]. Thus, the tolualdehydes produced are those resultingfrom intermediates formed by direct carbonphosphorus bond cleavage. LikewiseCo, and Ru hydroformylation catalysts give aryl derivatives not involving the earlierproposed ortho-metalation mechanism [69].

Several rhodium complexes catalyze the exchange of aryl substituents of triar-ylphosphines at elevated temperatures (130 C) [68]:

R03PR3PRh!R0R2PR02RPAbatjoglou et al. proposed as the mechanism for this reaction a reversible

oxidative addition of the arylphosphido fragments to a low valent rhodiumspecies.A facile aryl exchange has been described for complexes Pd(PPh3)2(C6H4CH3)I [70]. These authors also suggested a pathway involving oxidative additions andreductive eliminations. The mechanisms outlined below in the following sections,however, can also explain the results of these two studies.

Phosphido formation has been observed for many transition metal phosphinecomplexes [43].Uponprolongedheating, andunder an atmosphere ofCOand/orH2,palladium and platinum also tend to give stable phosphido-bridged dimers orclusters [71].

H

M

P

H

M

P

Scheme 1.13 Orthometallation of a phosphine.

H3CH

M

P

H3CH

M

P

H3CX

H3C X

X

Scheme 1.14 Disproving of orthometallation as a decomposition pathway.

1.4 Phosphines j13

A prototype of an oxidative addition with concomitant PC bond cleavage is thereaction of 1 (Scheme 1.15) with Pd2(dba)3, which gives the addition of an aryl groupto palladium and formation of phosphido bridges [72]. The interesting feature of thisexample is that the aryl group is a pentafluorophenyl group, for which only very fewexamples of this reaction have been reported. Hydrogen analogs dppe and dppp intheir reaction with low valent metals, for example, Pt(0) give metalation instead ofPC bond-cleavage (2, Scheme 1.15) [73].

Bridging diphosphine metal complexes have been characterized that may be enroute to PC cleavage, such as shown in Figure 1.2 [74].

During the studies of the isomerization of butenyl cyanides, relevant to thehydrocyanation of butadiene to give adiponitrile, the intermediate (TRIPHOS)Ni(CN)H complexes were found to decompose to benzene and highly stablem-phosphido-bridged dimers (only one isomer shown) that deactivate the catalyticprocess. Oxidative addition as themechanismwould invokeNi(IV) and therefore oneof the mechanisms to be discussed later, nucleophilic attack or phosphoraneintermediates, may be operative (Scheme 1.16).

Cluster or bimetallic reactions have also been proposed in addition to monome-tallic oxidative addition reactions. For instance, trishydridoruthenium dimers willcleave PC bonds in aryl and alkyl phosphines to give phosphido-bridged hydrides.Alkylphosphines give alkenes as the co-product, but phenylphosphines give benzene.For phenylphosphines the intermediate containing a bridging phenyl grouphas beenobserved, thus showing that that the reaction is an oxidative addition of thePCbondalong the rutheniumdimer, andnot a nucleophilic attack of a hydride at a phosphorusatom. The alkene products of the alkylphosphines are in accordwith thismechanism,as they are formed via b-elimination of the intermediate alkylruthenium groups(Scheme 1.17) [75].

Pd2(dba)3 2+

P

P

Ar

Ar

Ar

Ar

C=Ar 6F5

P

Pd

P

Pd

P

P

Ar

Ar

Ar

Ar

Ar

Ar Ar

Ar

1

Pt

P

Pt

P

PPPh2

Ph2

Ph

Ph2

Scheme 1.15 Reactions of diphosphines with zerovalent palladium and platinum complexes.

P

Pd

PR2

Pd LL

RR H

Ph2 PdP

PhP

Pd

PPh

PPh2

+ 2+

Figure 1.2 Interactions en route to PC cleavage.

14j 1 Elementary Steps