Mechanisms in Homogeneous Catalysis A Spectroscopi PDF

Catalysis by Metal Complexes
Volume 22
Editors:
B. R. James, The University of British Columbia, Vancouver, Canada
P. W. N. M. van Leeuwen, University of Amsterdam, The Netherlands
Advisory Board:
I. Horvth, Exxon Corporate Research Laboratory, Annandale, NJ, U.S.A.
S. D. Ittel, E. I. du Pont de Nemours Co., Inc., Wilmington, Del., U.S.A.
A. Nakamura, Osaka University, Osaka, Japan
W. H. Orme-Johnson, M.I.T, Cambridge, Mass., U.S.A.
R. L. Richards, John Innes Centre, Norwich, U.K.
A. Yamamoto, Waseda University, Tokyo, Japan
The titles published in this series are listed at the end of this volume.
RHODIUM CATALYZED
HYDROFORMYLATION
Edited by
PIET W.N.M. VAN LEEUWEN

Institute of Molecular Chemistry,
University of Amsterdam,
Amsterdam, The Netherlands
and
CARMEN CLAVER
Department de Quimica Fsica i Inorgnica,
Universitat Rovira i Virgili,
Tarragona, Spain
KLUWER ACADEMIC PUBLISHERS

NEW YORK / BOSTON / DORDRECHT / LONDON / MOSCOW
eBook ISBN: 0-306-46947-2
Print ISBN: 0-792-36551-8
2002 Kluwer Academic Publishers

New York, Boston, Dordrecht, London, Moscow
All rights reserved
No part of this eBook may be reproduced or transmitted in any form or by any means, electronic,
mechanical, recording, or otherwise, without written consent from the Publisher
Created in the United States of America
Visit Kluwer Online at: http://www.kluweronline.com

and Kluwer's eBookstore at: http://www.ebooks.kluweronline.com
Preface
This book covers the developments in rhodium catalyzed

hydroformylation of the last decade, one of the most important reactions in
industry catalyzed by homogeneous catalysts. The work includes many of
the advances that have been made by academic and industrial researchers.
The field has undergone drastic changes, both in its industrial applications
and in our understanding. Clearly, the new advances pose new problems and
set new targets for future research.
In spite of the importance of the field, the last reviews covering a broad
area in hydroformylation are outdated (Falbe 1980, Pruett 1977) and it was
felt timely to bring together the recent developments. Only in the area of
aqueous biphasic hydroformylation there are several exhausting reviews
available. This is the first monograph on hydroformylation of this type and
for other processes there not many examples.
The aim of the book is to review the mainstream of the activities in the
field and not to present a complete coverage of the literature, not even the
recent literature. Several thousands of papers and patents deal with rhodium-
catalyzed hydroformylation and a complete review would be impossible. We
have chosen for a more didactic approach, in which we have tried to avoid
one-liners about publications. In the book one will find typical examples
about kinetics, applications in organic chemistry, industrial processes,
mechanistic understanding, etc. In the mainstream activities we have tried to
include industrial developments. We may have missed new catalyst systems
that are as yet small but may turn out to be of major importance later, but
that can hardly be avoided. New and important developments involving
other metals, such as cobalt, platinum, and palladium will also be absent.
While writing we had a broad audience in mind: chemists and engineers
in industry and academia with an interest in homogeneous catalysis, whose
backgrounds may be as varied as those of the present authors: inorganic,
organic, organometallic, catalytic, chemical engineering. It is hoped that
specialists in one area will read with interest the chapters on the
neighbouring expertise. The book is also meant for PhD-students and
advanced students interested in this area.
The combination of topics we have chosen is rather unique, connecting
studies on ligand effects, catalyst characterization, industrial requirements
regarding stability and separation, catalyst decomposition, and applications
xi
xii Preface
in fine and bulk chemistry. The reader will notice the importance of one
discipline for the other. In many cases these relationships have already been
established, but for other cases the book might assist future developments.
The key roles that ligands may play in selectivity may be an eye-opener for
organic chemists and it will further enhance the large number of new
applications and reactions that are being discovered. The comments in
several chapters on catalyst preparation and feed purification may be useful
for scientists who are not specialized in homogeneous catalysis using
transition metal complexes.
Hydroformylation is also a model reaction system in homogeneous
catalysis as it contains so many aspects such as ligand effects (electronic,
steric, bite angle), in situ studies, complicated kinetics, and effects of
conditions and impurities. All this, combined with its practical value, makes
it an ideal topic in education.
The editors are very grateful to the authors for the good work they did
and the prompt responses. The writing took only a few months, as did the
production by the publisher. Writing the book has been rewarding, because
we learnt many things. Most of all perhaps, we obtained a clearer view on
what we still dont fully understand.
Amsterdam, Tarragona
Piet van Leeuwen, Carmen Claver
TABLE OF CONTENTS
Preface xi
1 Introduction to hydroformylation 1
Piet W. N. M. van Leeuwen
1.1 History of phosphorus ligand effects 1
1.2 Hydroformylation 6
1.3 Ligand parameters 8
2 Hydroformylation with unmodified rhodium catalysts 15

Raffaello Lazzaroni, Roberta Settambolo and Aldo Caiazzo
2.1 Introduction 15
2.2 Regioselectivity in the rhodium-catalyzed
hydroformylation of vinyl and vinylidenic substrates 16
2.2.1 Catalyst precursors 17
2.2.2 Influence of the alkene structure on the
regioselectivity 17
2.2.3 Influence of temperature 21
2.2.4 Influence of CO and H2 partial pressures 22
2.3 Mechanism of the hydroformylation of vinyl and
vinylidenic alkenes 22
2.3.1 Activation of the catalyst precursor 24
2.3.2 Behavior of the isomeric alkyl-metal intermediates
via deuterioformylation 24
2.3.3 In situ IR investigation of the formation and
reactivity of acylrhodium intermediates 28
2.4 Origin of the regioselectivity 29
2.4.1 Influence of the nature of the substrate 29
2.4.2 Influence of the reaction parameters 31
3 Rhodium phosphite catalysts 35

Paul C. J. Kamer, Joost N. H. Reek, and Piet W. N. M. van Leeuwen
3.1 Introduction 35
v
vi Table of contents
3.2 Monophosphites 37
3.2.1 Catalysis 37
3.2.2 Mechanistic and kinetic studies 40
3.3 Diphosphites 44
3.3.1 Catalysis 44
3.3.2 Mechanistic and kinetic studies 48
3.4 Hydroformylation of internal alkenes 55
3.4.1 Hydroformylation of less reactive internal and
functionalized alkenes 55
3.4.2 Formation of linear aldehydes starting from
internal alkenes 57
3.5 Calixarene based phosphites 59
4 Phosphines as ligands 63
Piet W. N. M. van Leeuwen, Charles P. Casey, and Gregory T. Whiteker
4.1 Monophosphines as ligands 63
4.1.1 Introduction 63
4.1.2 The mechanism 64
4.1.3 Ligand effects 66
4.1.4 In situ studies 68
4.1.5 Kinetics 69
4.1.6 Regioselectivity 72
4.1.7 Conclusion 75
4.2 Diphosphines as ligands 76
4.2.2 Ferrocene based diphosphine ligands 78
4.2.3 BISBI ligands and the natural bite angle 82
4.2.4 Xantphos ligands: tunable bite angles 87
4.2.5 The mechanism, regioselectivity, and
the bite angle. Concluding remarks 96
5 Asymmetric hydroformylation 107

Carmen Claver and Piet W.N.M. van Leeuwen
5.1 Introduction 107
5.2 Rhodium systems with chiral diphosphite ligands 109
5.2.1 C2 Symmetric chiral diphosphite ligands 109
5.2.2 Catalyst preparation and hydroformylation 111
5.2.3 Characterisation of [RhH(L)(CO)2] intermediates.
Solution structures of hydroformylation
catalysts 113
5.2.4 Structure versus stability and enantioselectivity 115
Table of contents vii
5.2.5 Chiral cooperativity and effect of substituents in

diastereomeric diphosphite ligands 116
5.2.6 C1 Sugar backbone derivatives. Diphosphinite and
diphosphite ligands 121
5.3 Phosphine-phosphite rhodium catalysts 124
5.3.2 Rhodium complexes with BINAPHOS and
related ligands 124
5.3.3 [RhH(CO) 2 (BINAPHOS)] complexes; models for
enantioselectivity 127
5.3.4 Separation studies for the BINAPHOS system 129
5.3.5 Chiral phosphine-phosphite ligands containing a
stereocenter in the backbone 129
5.4 Diphosphine rhodium catalysts 131
5.4.2 C1 Diphosphines as chiral ligands 131
5.4.3 C2 Diphosphines as chiral ligands 132
5.4.4 The Rh/BDPP system. HPNMR and HPIR
studies under hydroformylation conditions 136
5.5 Mechanistic considerations 138
5.5.1 Regioselectivity 138
5.5.2 Enantioselectivity and conclusions 140
6 Hydroformylation in organic synthesis 145

Sergio Castilln and Elena Fernndez
6.2 Hydroformylation of unfunctionalized alkenes 146
6.3 Hydroformylation of functionalized alkenes 149
6.4 Substrate directed stereoselectivity 155
6.5 Control of the regio- and stereoselectivity by heteroatom-
directed hydroformylation 160
6.6 Consecutive processes under hydroformylation
conditions 164
6.6.1 Hydroformylation-acetalization (intramolecular) 165
6.6.2 Hydroformylation-acetalization (intermolecular) 166
6.6.3 Hydroformylation-amination (intramolecular) 168
6.6.4 Hydroformylation-amination-reduction.
Hydroaminomethylation 172
6.6.5 Consecutive hydroformylation-aldol reaction 175
6.6.6 Consecutive hydroformylation-Wittig reaction 177
6.7 Alkyne hydroformylation 178
6.8 Concluding remarks 182
viii Table of contents
7 Aqueous biphasic hydroformylation 189

Jrgen Herwig and Richard Fischer
7.1 Principles of biphasic reactions inwater 189
7.1.1 Why two-phase catalysis?
Scope and Limitations 189
7.1.2 Concepts for two-phase hydroformylation 190
7.2 Hydroformylation of propene and butene 191
7.2.1 Historic overview of two-phase
hydroformylation technology 191
7.2.2 Ligand developments 191
7.2.3 Kinetics and catalyst pre-formation 193
7.2.4 Process description 196
7.2.5 Status of the operated plants 197
7.2.6 Economics 198
7.3 Reaction of various alkenes 199
7.3.1 Ethylene to propanal: why not applied? 199
7.3.2 Long-chain alkenes 200
8 Process aspects of rhodium-catalyzed hydroformylation 203

Peter Arnoldy
8.2 Economics 204
8.3 Catalyst selectivity and activity 206
8.3.1 Catalyst selectivity 206
8.3.2 Catalyst activity 207
8.4 Catalyst stability; degradation routes, losses and
recovery 208
8.4.1 Rhodium loss routes 208
8.4.2 Ligand loss routes 209
8.4.3 Catalyst recovery processes 210
8.5 Process concepts 211
8.5.1 Type I: Stripping reactor process/Rh
containment in reactor 212
8.5.2 Type II: Liquid recycle process/use of
distillative separation 213
8.5.3 Type III: Two-phase reaction/extraction process 2 15
8.5.4 Type IV: Extraction after one-phase reaction 216
8.6 Survey of commercialized processes and new
developments 220
8.6.1 Hydroformylation of butenes 220
8.6.2 Branched higher alkenes to mainly plasticizer
alcohols 223
Table of contents ix
8.6.3 Linear higher alkenes to mainly detergent

alcohols 223
8.6.4 1,4-Butanediol 225
8.6.5 Nylon monomers 226
9 Catalyst preparation and decomposition 233

9.2 Catalyst preparation 233
9.3 Catalyst decomposition 235
9.3.1 Metal plating or cluster formation 235
9.3.2 Oxidation of phosphorus ligands 235
9.3.3 Phosphorus-carbon bond breaking in phosphines 237
9.3.4 Decomposition of phosphites 243
9.3.5 Formation of dormant sites 247
9.4 Concluding remarks 249
10 Novel developments in hydroformylation 253

Joost N. H. Reek, Paul C. J. Kamer, and Piet W. N. M. van Leeuwen
10.2 New bimetallic catalysts 253
10.3 Novel developments in catalyst separation 256
10.3.1 Micellar catalysis 256
10.3.2 Supported aqueous phase catalysis (SAPC) 260
10.3.3 Hydroformylation in supercritical fluids 262
10.3.4 Fluorous Biphase catalysis 265
10.3.5 Dendrimer supported catalysts 267
10.3.6 Novel developments in polymer supported catalyst
269
10.4 Supramolecular catalysis 274
10.5 Conclusions 277
Index 281
Chapter 1
Introduction to hydroformylation
Phosphorus ligands in homogeneous catalysis

Institute of Molecular Chemistry, University of Amsterdam, Nieuwe Achtergrucht 166, 1018 WV,
1.1 History of phosphorus ligand effects
In this chapter we will briefly review phosphorus ligand effects in

homogeneous catalysis and hydroformylation more in particular. First we
will have a look at a few historical landmarks in homogeneous catalysis
concerned with the use of phosphorus ligands, then focus on the history of
rhodium catalyzed hydroformylation, and subsequently summarize a few
basic concepts. Since phosphorus ligands are the only ligands used in
hydroformylation in addition to carbon monoxide, we will not discuss
ligands containing other donor atoms. In later chapters we will see that in
hydroformylation, as it is today, bidentate phosphorus ligands are of great
importance. In the introduction we show that in the early history the positive
effect of bidentates on selectivities and rates of catalytic reactions was not
fully recognized [ 1].
The favorable effects of phosphine ligands in catalysis have been known
for more than half a century. One of the first reports involves the use of
triphenylphosphine in the Reppe chemistry, the reactions of alkynes,
alcohols and carbon monoxide to make acrylic esters [2]. An early example
of a phosphine-modified catalytic process is the Shell process for alkene
hydroformylation using a cobalt catalyst containing an alkylphoshine [3].
Hydrocyanation as applied by Du Pont is another early example of an
industrially applied catalytic reaction employing ligands [4]. The nickel
catalyzed reaction uses aryl phosphite ligands for the production of
adiponitrile from butadiene and hydrogen cyanide. The development of this
process has played a key-role in the introduction of the now common study
1
P.W.N.M. van Leeuwen and C. Claver (eds.), Rhodium Catalyzed Hydroformylation. 113.
2000 Kluwer Academic Publishers. Printed in the Netherlands.
2 Chapter 1
of ligand effects in the field of homogeneous catalysis using

organometallic complexes [5].
Both academia and industries made important contributions to the new
field in the early sixties with the appearance of the first phosphine modified
and other hydrogenation catalysts. An early example of a phosphine-free
ruthenium catalyst was published by Halpern [6]. In 1963 Cramer (Du Pont)
reported a triphenylphosphine-modified platinum-tin catalyst for the
hydrogenation of alkenes [7]. In the same year Breslow (Hercules) included
a few phosphine complexes of late transition metals in a hydrogenation study
employing metal salts reduced by aluminum alkyls, but interestingly the
systems containing phosphine were less active [8]!
Rhodium catalyzed hydrogenation was discovered in the mid-sixties by
Wilkinson and coworkers [9]. The mechanism of this reaction using
RhCl(PPh3)3 as the catalyst was studied in great detail. These studies by
Wilkinson and many others have been a major stimulant for workers in this
area. Substitution at the aromatic ring revealed an electronic effect on the
reaction rate, electron donors giving higher rates [10]. A few months later
Vaska published his first work on the rhodium and iridium catalyzed
hydrogenation of alkenes [ 11].
Rhodium-catalyzed hydroformylation using catalysts modified with
alkylphosphines and arylphosphines was reported by Wilkinsons group
[ 12]. Phosphine ligand variation hardly affected the rate and selectivity
under the circumstances used (70 C and 100 bar). Pruett (Union Carbide
Corporation) found that phosphites can also be used, and the type of
phosphite had a profound effect on rates and selectivities [13].
Figure 1. Structures of dppe, SHOP ligand, DIOP, and DIPAMP
Bidentate ligands have played an important role in the development of

the chemistry of metal organic complexes. The synthesis of dppe was
reported as early as 1959 [14]. Chatt and Hieber [15] explored the
coordination chemistry of several diphosphines with an ethane bridge, but it
took a while before diphosphines became routinely included in catalysis
studies. In the early sixties diphosphines were mentioned in patents, but
specific advantages are not apparent. In their exploration of carbonyl
chemistry of cobalt related to carbonylation catalysis, Heck and Breslow
1. Introduction to hydroformylation 3
[16] reported that HCo(CO)4 gave unidentifiable complexes with dppe. The
use of dppe in cobalt catalyzed hydroformylation was reported by Slaugh
[17], but compared to PBu3 it had little effect on the rate and the selectivity
of the cobalt carbonyl catalyst. Copolymerization of butadiene and
propylene oxide using nickel bromide and dppe was published in 1965 [18].
In the late sixties at Shell Development, Keim and coworkers discovered
that certain bidentates containing an oxygen and a phosphorus donor atom
formed excellent catalysts with nickel for the oligomerization of ethene [19].
Typical ligands are diphenylphosphinoacetic acid or 2-diphenyl-
phosphinobenzoic acid (SHOP ligand, Figure 1). The ligand required a
relatively laborious ligand synthesis for those days. In addition it was the
first process utilizing the concept of two-phase catalysis. This discovery led
to the Shell Higher Olefins Process that came on stream in 1977.
Hata [20] reported a phosphine-free iron catalyst for the codimerization
of butadiene and ethene in 1964. A year later this was followed by
phosphine-free rhodium catalysts [21]. The oldest publication describing
advantageous results for diphosphines we found is by Iwamoto and Yuguchi
(1 966) who studied the same reaction using iron catalysts containing a range
of diphosphines varying in bridge lengths [22]. In many instances the
activity of catalysts containing dppe instead of PPh3 is lower. For example,
the hydrogenation of styrene using rhodium(I) chloride and dppe is 70 times
slower as compared to the PPh3 based system [23]. The strong chelating
power of the diphosphine was held responsible for this. Thus, initially the
use of dppe and other bidentate phosphines in catalysis found little support
as they were supposed to lead mostly to more stable complexes, rather than
more active or selective catalysts.
Theoretical work of Thorn and Hoffmann [24] explained why migration
reactions in complexes containing for instance dppe were slow. The
constrained P-M-P angle would slow down the migration reaction, since
ideally the phosphine ligand coordinated in the position cis to the migrating
group, would have a tendency to widen the P-M-P angle in the process to
pursue the migrating group.
Figure 2. Structures of Phenyl--GLUP, BINAP, and DuPhos

4 Chapter 1
A beneficial use of bidentate diphosphines was discovered in 1971 by

Kagan [25] who reported the use of DIOP modified rhodium for the
hydrogenation of N-acetylphenylalanine. Monophosphines for asymmetric
hydrogenation of similar substrates were reported by Knowles [26]. His
discovery of the P-chiral diphosphine DIPAMP, also a bidentate ligand, led
to the commercial application of the asymmetric hydrogenation of the
Levodopa precursor. For the same process Selke developed another ligand, a
sugar based bisphosphonite Phenyl--GLUP [27]. The company VEB-Isis
applied this ligand for many years in Germany.
In the area of asymmetric hydrogenation chiral dighosphines have played
a center role since and many applications have been developed. Important
new ligands that have been introduced comprise Noyoris BINAP [28],
DuPhos (Burk) [29], Takayas BINAPHOS [30], and C1-symmetric
ferrocene-based ligands introduced by Togni [3 1]. Industrial products, of
which the synthesis uses enantioselective phosphine-derived metal-catalysts
are for instance menthol, metolachlor, biotin, and several alcohols, e.g. (R)-
1,2-propanedioI, For details about the applications the reader is referred to
reviews and references therein [32, 33]. Substituents and backbones have an
enormous influence on the performance of the ligands, but usually
rationalizations are lacking.
Figure 3. Asymmetric phosphine ligands BINAPHOS and Josiphos
In carbonylation chemistry using phosphine or phosphite complexes of

palladium or rhodium a number of breakthroughs achieved in the seventies
and eighties should be mentioned; hydrofomylation will be reviewed in
section 1.2. Here we will concentrate on those that have found industrial
application or may find application in the near future. In the early eighties
Sen [34] and Drent [35] discovered that ethene and carbon monoxide can
polymerized in an alternating fashion leading to polyketones. The catalyst is
a palladium complex containing phosphines and non-coordinating anions in
methanol as the solvent. Drents bidentate phosphine containing catalysts
proved by far to be the fastest ones. Especially diphosphines having a
propane bridge give a fast reaction to high molecular weight products.
Shells process-related patents often use the propane bridged 1,3-bis-(di-2-

anisylphosphino)propane as the ligand (dapp) [36]. Carilon, Shells trade
name for the terpolymer of ethene, CO and propene - added for lowering the
processing temperature of the product - has been in commercial production
on a relatively small scale in the late nineties.
Figure 4 Ligands for bulk chemical processes
Another impressive ligand effect reported by Drent and coworkers [37]

concerns the methoxycarbonylation of propyne to form methyl methacrylate.
Triphenylphosphine modified palladium catalysts give low rates, but using
2-pyridyldiphenylphosphine instead gives very high rates and selectivities.
The mechanism is still a matter of debate [38].
Tris-m-sulfonatophenylphosphine (tppts) plays an important role in the
history of homogeneous catalysis [39], mainly due to its use in the
Ruhrchemie/Rhne-Poulenc hydroformylation process [40], now operated
by Celanese (see 1.2 and Chapter 7). It is also used in a number of fine
chemical processes, such as selective hydrogenation with ruthenium [41],
carbon-carbon bond formation with rhodium [42], and the Heck reaction
[43]. Monosulfonated triphenylphosphine (tppms) is used for the preparation
of nonadienol [44] (see Figure 5).
In C-C, C-O, and C-N bond formation reactions catalyzed by palladium
and nickel, ligand effects have been explored in an extremely wide area [33].
The data available on ligand effects for these reactions are numerous. In
asymmetric allylic alkylation the embracing effect of the bidentate ligand
explains the efficacy of the ligand in many instances [45]; the longer the
backbone, - i.e. the larger the bite angle (vide infra) - and the more effective
the ligand interacts with the substrate. For the Heck reaction the ligand size
seems to be a dominant factor, as bulky phosphines [46], phosphites [47],
and amidites [48] were found to lead to highly effective catalysts. For
amidites it was shown that the bulky ligands lead to mono-ligand complexes
which are effectively more prone to substrate coordination than bis-ligand
complexes. This effect was first observed by us for the same bulky
phosphites in rhodium catalyzed hydroformylation [49] (Figure 5).
6 Chapter 1
1.2 Hydroformylation
The first generation of hydroformylation catalysts was based on cobalt

carbonyl without phosphine ligand [50]. The conditions were harsh, as the
reactivity of cobalt is low. The process was used both for lower as for higher
alkenes, and notably also internal alkenes give mainly linear product
aldehyde. Initially rhodium catalyzed reaction seemed slow, because the
formation of rhodium hydrides requires high pressures of hydrogen [51]. An
early commercial application of phosphine-free rhodium was by Mitsubishi
for the hydroformylation of higher 1-alkenes in 1970. The kinetics of
rhodium carbonyl catalyzed hydroformylation were studied for the first time
in the sixties, but in the last decade the studies by Lazzaroni and Garland
have revealed interesting aspects that will be dealt with in Chapter 2.
Since Shell's report on the use of phosphines in this process [3], many
industries started applying phosphine ligands in the rhodium process as well
[52]. While alkylphosphines are the ligands of choice for cobalt, they lead to
slow catalysis when applied in rhodium catalysis. In the mid-sixties the work
of Wilkinson showed that arylphosphines should be used for rhodium and
that even at very mild conditions very active catalysts can be obtained [9].
tppms "bulky" phosphite UCC ligand
Figure 5. Structures of ttpms, van Leeuwen's "bulky phosphite", and a highly stable, bulky
phosphite from UCC
The second generation processes use rhodium as the metal and the first
ligand-modified process came on stream in 1974 (Celanese) and more were
to follow in 1976 (Union Carbide Corporation) and in 1978 (Mitsubishi
Chemical Corporation), all using triphenylphosphine (tpp). The UCC
process has been licensed to many other users and it is often referred to as
the LPO process. Not only are rhodium catalysts much faster - which is
translated into milder reaction conditions -, but also their feedstock
utilization is much better than that of cobalt catalysts. For example, the
cobalt-alkylphosphine catalyst may give as much as 10% of alkane
formation. Since the mid-seventies the rhodium catalysts started to replace

the cobalt catalysts in propene and butene hydroformylation. For detergent
alcohol production though, even today, the cobalt systems are still in use,
because there is no good alternative yet for the hydroformylation of internal
higher alkenes.
The third generation process concerns the Ruhrchemic-RhonePoulene
process utilizing a two-phase system containing water-soluble rhodium-tppts
in one phase and the product butanal in the organic phase. The process has
been in operation in Oberhausen since 1984 by Celanese, as the company is
called today. The. system will be discussed in Chapter 7. Since 1995 this
process is also used for the hydroformylation of 1 -butene.
In the late sixties phosphites have also been considered as candidate
ligands for rhodium hydroformylation, but tpp turned out to be the ligand of
choice. A renewed interest in phosphites started in the eighties after we had
discovered the peculiar effect of bulky monophosphites giving very high
rates [49]. Bryant and coworkers at Union Carbide expanded this work
enormously, first by making more stable bulky monophosphites [53], later
by focusing on diphosphites [54]. There is only one relatively small
commercial application of bulky monophosphite by Kuraray for the
hydroformylation of 3-methylbut-3-en- l-ol [55]. A large amount of research
has been devoted to diphosphites in the last decade aiming at a variety of
applications. The results will be discussed in Chapter 3.
Diphosphines have also become very popular ligands since the late
eighties in rhodium hydroformylation, e.g. Eastmans BISBI. Chapter 4
focuses on diphosphines. The two-phase system has undergone considerable
improvements involving diphosphines [39].
Figure 6. Eastmans BISBI and typical diphosphites from Union Carbide Corporation
In recent years the interest for hydroformylating higher alkenes with

catalysts other than cobalt has increased. Platinum and palladium based
catalysts have been studied and the results of the latter [56] seem very
8 Chapter 1
promising. Platinum has been known for many years to have a high
preference for the formation of linear products, but ligand decomposition
hampers applications [57]. Palladium and platinum will not be discussed, but
recent advances for rhodium have been collected in Chapter 10.
A ligand with great potential for hydroformylation of higher alkenes in a
one-phase system that is worked up by adding water to separate catalyst and
product afterwards is the monosulfonated triphenylphosphine, tppms, that
was studied by Abatjoglou, also at Union Carbide [58] (Chapter 8).
The fourth generation process for large-scale application still has to be
selected from the potential processes that have been nominated. In the
chapters to follow several of these candidates will be discussed. The fourth
generation will concern higher alkenes only, since for propene
hydroformylation there are hardly wishes, if any, left [59] (a cheaper catalyst
would be on my shopping list!). Many new phosphite-based catalysts have
been reported that will convert internal alkenes to terminal products and
recently also a new diphosphines have been reported that will do this [60].
The most interesting ligand discovered for asymmetric hydroformylation
is undoubtedly BINAPHOS, introduced by Takaya [30]. Diphosphites have
also been studied to this end by UUC [61] and by us [62]; Babin and
Whiteker reported the first successful ones [61]. Asymmetric rhodium
catalysts are discussed in Chapter 5.
Ligand design for fine chemical applications has been very limited and
usually the ligands designed for large-scale applications are also tested for
more complicated organic molecules. Tpp has been the workhorse in fine
chemicals hydroformylation since Wilkinsons first examples [63, 64], but
also bulky phosphite [49], tppts and tppms [41-43] turned out to be very
useful, and recently diphosphites have been studied [65] (see Chapter 6).
1.3 Ligand parameters
Tolman reviewed ligand effects for the first time [5]. Prior to his studies
[66] the effects of phosphorus ligands on reactions or properties of metal
complexes were rationalized mainly in terms of electronic effects.
Systematic studies had shown, however, that steric effects are at least as
important as electronic effects, and in terms of stability of complexes can
even be dominant. Since then, numerous studies have appeared using both
the electronic parameter and the steric parameter for the cone angle, .
The electronic parameter is a measure for the overall effect of electron
donating and accepting properties of the phosphorus ligand L. It is measured
as the symmetric stretching frequency of the carbonyls in Ni(CO)3L, similar
to the methods proposed by Strohmeier and Horrocks [67]. High -values
stand for strong -acceptors and low -values stand for strong -donor
ligands. An expansion of Tolmans list of -values was published by Bartik

[68]. NMR methods for the measurement of substituent effects on
phosphorus have been reported [69]. i values are used for one substituent.
Figure 7. A simple picture of Tolmans -value. On the left strong back donation to CO
leading to a low stretching frequency. On the right weak back-donation to CO leading to a
high stretching frequency [5]
For monodentate phosphorus ligands the cone angle is defined as the

apex angle of a cylindrical cone, centered at 2.28 from the center of the P
atom, which touches the outermost atoms of the model (actually, this is done
using CPK models). For phosphines containing different substituents an
average for the three substituents is taken. Crystal structure determinations
have shown that the actual angles in the complexes are smaller than those
expressed by the -value due to an intermeshing of the Substituents. The
relative order of cone angles parallels with many properties that have been
measured, such as equilibrium constants and rate constants [70].
Figure 8. Tolmans cone angle [5]. The distance M-P amounts to 2.24 . Caseys natural
bite angle n as calculated by Molecular Mechanics [79, 80].
Several studies have been conducted aiming ai the separation of steric

and electronic effects [71-73]. For a single step process such as an oxidative
addition or one-electron change in electrochemical processes this may be
useful, but for multi-step reactions as we are dealing with in catalysis, this
technique will encounter many problems. There will be different effects on
the distinct steps and linear free-energy relationships will be an exception
rather than the rule. When for instance both an oxidative addition step and a
reductive elimination step are involved volcano curves must be expected
for reactivity versus a ligand property, as in a series of metal oxides when
10 Chapter 1
plotting oxidation activity versus position in the periodic table. In addition,

multiple metal-ligand equilibria will affect the schemes. Enthalpy studies
can contribute to the solution of these complex matters [74].
During the seventies the attention for bidentate phosphines as ligands in
catalysis had been growing and so did the need for including them into the
electronic and steric mapping. Tolman extended the cone angle for
monophosphines to diphosphines, which was defined as the average cone
angle measured for the two substituents and the angle between the M-P bond
and the bisector of the P-M-P angle. Even today, this looks like a good
approximation for defining a cone angle of a bidentate. Other parameters for
bidentate ligands have been reported, such as the solid angle [75], pocket
angle [76], repulsive energy [77], and the accessible molecular surface [78].
In the present study we take a different approach, which is less elaborate
than the methods mentioned above. The steric properties of diphosphines are
determined by the four substituents at the two phosphorus atoms and the
length of the bridge. In general, the most stable complexes are obtained
when a five-membered ring can form, i.e. when the bridge between the two
phosphorus donor atoms consist of two carbon atoms as in dppe. This is true
for octahedral and square-planar complexes in which the metal-preferred
[79] P-M-P angle is ~ 90. The vast majority of chelate complexes have been
synthesized from bidentate ligands possessing relatively short, bridging
backbones. Tetrahedral complexes will prefer P-M-P angles of 109, and
bis-equatorial coordination in a trigonal bipyramid requires an angle of 120.
During catalytic processes transitions between different coordination modes
may be needed. The natural preference of a ligand for a certain
coordination mode can influence a reaction of a catalytic cycle in several
ways: stabilization or destabilization of the initial, transition, or final state. In
addition the flexibility of a bidentate ligand may be important in order to
accelerate certain transitions. In a one-step reaction the effect of the bite
angle may be very clear-cut, but a catalytic cycle involves more steps and
equilibria and in many instances the effect on catalysis may not be
characteristic of the bite angle.
A means to predict the ligand-preferred [79] P-M-P angle using
molecular mechanics has been developed by Casey and Whiteker [80]. They
introduced the concepts of natural bite angle (n) and flexibility range for
diphosphine ligands. Computer modeled geometries can be used to estimate
ligand bite angles. The calculations can even be performed before ligands
are synthesized. If computer modeling is employed to design new ligands, it
is more important to calculate a correct trend rather than perfect geometries.
The bite angle not only induces steric effects, but also electronic effects,
as the P-M-P angles clearly affect the binding in the complex or intermediate
states [81].
References
1 van Leeuwen, P. W. N. M.; Kamer, P. C. J.; Reek, J. N. H.; Dierkes, P. Chem. Rev. 2000,
100, in press.
2 Reppe, W.; Schweckendiek, W. J. Annalen 1948, 104, 560.
3 Slaugh, L. H.; Mullineaux, R. D. U.S. Pat. 3,239,569 and 3,239,570, 1966 (to Shell);
Chem. Abstr. 1964, 64, 15745 and 19420. J. Organornet. Chem. 1968,13,469.
4 (a) Brown, E. S. Aspects of Homogeneous Catalysis; Ugo, R., Ed.; Reidel: 1974; Vol. 2;
pp 57-78. (b) Drinkard, W. C.; Lindsey, R. V. U. S. Pat. 3,655,723, 1970 (to Du Pont);
Chem. Abstr. 1972, 77, 4986.
5 Tolman, C. A. Chem. Rev. 1977, 77, 313.
6 Halpem, J.; James, B. R. J. Am. Chem. Soc. 1961, 83, 753.
7 Cramer, R. D.; Jenner, E. L.; Lindsey, R. Y.; Stolberg, U. G.; J. Am. Chem. Soc. 1963,
85, 1691.
8 Sloan, M. F.; Matlack, A. S.; Breslow, D. S. J. Am. Chem. SOC. 1963, 85,4015.
9 Young, J. F.; Osbom, J. A,; Jardine, F. A.; Wilkinson, G. J. Chem. SOC. , Chem. Comm.
1965, 131.
10 O'Connor, C.; Wilkinson, G., Tetrahedron Lett. 1969,18, 1375.
11 Yaska, L.; Rhodes, R. E. J. Am. Chem. SOC. 1965, 87, 4970.
12 (a) Jardine, F. H.; Osborn, J. A.; Wilkinson, G.; Young, J. F. Chem. and Ind. (London)
1965, 560. (b) Evans, D.; Osborn, J. A.; Wilkinson, G. J. Chem. Soc. A. 1968, 3133.
13 Pruett, R. L.; Smith, J. A. J. Org. Chem. 1969, 34, 327.
14 Issleib, K.; Muller, D. W. Chem. Ber. 1959, 92, 3175.
15 (a) Chatt J.; Hart, W. J. Chem. Soc. 1960, 1378. (b) Hieber, W.; Freyer, W. Chem.Ber.
1960, 93, 462.
16 Heck, R. F.; Breslow, D. S. J. Am. Chem. SOC. 1962, 84, 2499.
17 Cannel, L. G.; Slaugh, L. H.; Mullineaux, R. D. Ger. Pat. 1,186,455 (priority date 1960,
to Shell); Chem. Abstr. 1965, 62, 16054.
18 United States Rubber Co, Neth. Appl. 6,507,223; Chem. Abstr. 1965, 62, 16013.
19 Bauer, R. S.; Glockner, P. W.; Keim, W.; van Zwet, H.; Chung, H. U.S. Pat. 3,644,563,
1972.
20 Hata, G. J. Am. Chem. Soc. 1964, 86 ,3903.
21 Cramer, R. J. Am. Chem. SOC. 1965, 87 , 4717 and 1967, 89, 1633.
22 Iwamoto, M.; Yuguchi, S. J. Org. Chem. 1966, 31, 4290.
23 Dang, T. P.; Kagan, H. B. Chem. Commun. 1971, 481 and J. Am. Chem. Soc. 1972, 92,
6429.
24 Thorn D. L.; Hoffmann R. J. Am. Chem. SOC. 1978, 100, 2079.
25 Poulin, J. C.; Dang, T. P.; Kagan, H. B. J. Organometal. Chem. 1975, 84, 87.
26 (a) Knowles, W. S.; Sabacky, M. J. J. Chem. SOC., Chem. Commun. 1971, 481 (b)
Knowles, W. S.; Sabacky, M. J.; Vineyard, B. D., and Weinkauff, J. J. Am. Chem. SOC.
1975 97, 2567.
27 Selke, R. and Pracejus, H.; J. Mol. Catal., 1986, 37, 213.
28 Miyashita, A.; Yasuda, A.; Tanaka, H.; Toriumi, K.; Ito, T.; Souchi, T.; Noyori, R. J. Am.
Chem. Soc., 1980, 102,7932,
29 Burk, M. J. J. Am. Chem. SOC. 1991, 113, 8518.
30 Sakai, N.; Mano, S.; Nozaki, K.; Takaya, H. J. Am. Chem. Soc. 1993, 15, 7033.
31 Togni, A. Angew. Chem. 1996, 108, 1581.
32 Transition Metals for Organic Synthesis; Building Blocks and Fine Chemicals, ed. M.
Beller and C. Bolm, Wiley-VCH, 1998, Vol 1 and 2.
12 Chapter 1
33 Metal-catalyzed Cross-coupling Reactions, eds. F. Diederich and P.J. Stang, Wiley-

VCH, Weinheim, 1998.
34 Sen, A.; Lai, T. W. J. Am. Chem. SOC. 1982, 104, 3520.
35 (a) Drent, E. Eur. Pat. Appl. 121,965, 1984 (to Shell); Chem. Abstr. 1985, 102, 46423. (b)
Drent, E.; Budzelaar. P. H. M. Chem. Rev. 1996, 96, 663.
36 (a) Mastenbroek, B.; Petrus, L.; Rosenbrand, G. G. Can. Pat. Appl., 2,030,596, 1990 (to
Shell); Chem. Abstr. 1991, 115, 184575. (b) Mastenbroek, B.; Tummers, C. H. M.; De
Smedt, P. J. M. M. Brit. Pat. Appl., 2,278,366, 1994 (to Shell); Chem. Abstr. 1995, 123,
170553.
37 Drent, E.; Amoldy, P.; Budzelaar, P. H. M. J. Orgunornet. Chem. 1994, 475, 57.
38 Scrivanti, A.; Beghetto, V.: Campagna, E.; Zanato, M.; Mattcoli, U. Organometallics,
1998, 17, 630.
39 Aqueous Phase Organometallic Catalysis-Concepts and Applications, eds. B. Cornils and
W. A. Herrmann. Wiley-VCH, Weinheim, 1998.
40 Cornils, B.; Kuntz, E. G. J. Organomet. Chem. 502, 1995, 177.
41 Grosselin, J. M.; Mercier, C.; Allmang, G.; Grass, F. Organometallics, 1991, 10, 2126.
42 Mercier, C.; Chabardes, P. Pure Appl. Chem., 1994. 66, 1509.
43 Haber, S.; Kleiner, H.-J. WO 9705104, 1997 (to Hoechst); Chem. Abstr. 1997, 126,
199349.
44 Yoshimura, N.; Tokito, Y. Eur. Pat. Appl. 1987, 223,103 (to Kuraray); Chem. Abstr.
1987, 107, 154896.
45 Trost, B. M.; Murphy, D. J. Organometallics 1985, 4, 1143.
46 Beller, M.; Riermeijer, T. H.; Haber, S.; Kleiner, H. J.; Herrmann, W.A. Chem. Ber. 1996,
129, 1259.
47 Albisson, D. A.; Bedford, R. B.; Lawrence, S. E.; Scully, P. N. J. Chem. soc. Chem.
Commun., 1998, 2095.
48 van Strijdonck, G. P. F. ; Boele, M. D. K.: Kamer, P. C. J.; de Vries, J. G.: van Leeuwen,
P. W. N. M. Eur. J. Inorg. Chem. 1999, 1073.
49 (a) van Leeuwen, P. W. N. M.; Roobeek, C. F. J. Organometal. Chem. 1983, 258, 343; Brit.
Pat. 2,068,377, US Pat. 4,467,116 (to Shell Oil); Chem. Abstr. 1984, 101, 191142. (b)
Jongsma, T.: Challa, G.; van Leeuwen, P. W. N. M. J. Organometal. Chem. 1991, 421, 121.
(c) van Rooy, A.; Orij, E. N.; Kamer, P. C. J.; van Leeuwen, P. W. N. M. Organometallics,
1995, 14, 34.
50 Frohning, C. D.; Kohlpaintner, C. W. page 29, in Applied Homogeneous Catalysis with
Organometallic Compounds, eds. Comils B.; Herrmann, W. A., VCH, Weinheim, 1996,
Vol 1 and 2.
51 Vidal, J. L.; Walker, W. E. Inorg. Chem.1981, 20, 249.
52 Onoda, T. ChemTech. 1993, Sept. 34.
53 Billig, E.; Abatjoglou, A. G.; Bryant, D. R.; Murray. R. E.; Maher, J. M. U.S.Pat. 4,599,206
1986 (to Union Carbide Corp.); Chem. Abstr. 1989, 109, 233 177.
54 Billig, E.; Abatjoglou, A. G.; Bryant, D. R. U.S. Pat. 4,769,498, U.S. Pat. 4,668,651;
U.S. Pat. 4748261, 1987 (to Union Carbide Corp.); Chem. Abstr. 1987, 107, 7392r.
55 Yoshinura, N.; Tokito, Y. Eur. Pat. 223,103, 1987 (to Kuraray).
56 Drent, E.; Jager, W. W. U.S. Pat. 5,780,684, 1998 (to Shell); Chem. Abstr. 199x, 129,
137605.
57 Hayashi T.; Kawabata Y.; Isoyama T.; Ogata, I. Bull. Chem. soc. Jpn. 1981, 54, 3438.
58 Abatjoglou, A. G.; Bryant, D. R.; Peterson, R. R. U.S. Pat. 5,180,854) 1993 (to Union
Carbide Corp.; Chem. Abstr. 1991, 113, 80944.
59 Wiebus, E.; Cornils, E. Chem. Ing. Tech. 1994, 66, 916; Comils, B.; Wiebus, E.
ChemTech, 1995, 25, 33,; Comils, B.; Wiebus, E. Recl. Trav. Chim. Pays-Bas, 1996,
115,211.
60 van der Veen. L. A.; Kamer, P. C. J.; van Leeuwen, P. W. N. M. Angew. Chem. Int. Ed.
Engl. 1999, 38, 336.
61 Babin, J. E.; Whiteker, G. T. U.S. Pat. 5,360,938 1994 (to Union Carbide Corp.); Chem.
Abstr. 1995, 122, 186609.
62 Buisman, G. J. H.; Kamer, P. C. J.; van Leeuwen, P. W. N. M. Tetrahedron:
Asymmetry, 1993, 4, 1625.
63 Brown, C. K.; Wilkinson, G. .J. Chem. Soc. (A) 1970, 2753.
64 Eilbracht, P.; Brfacker, L.; Buss, C.; Hollmann, C.; Kitsos-Rzychon, B. E.; Kranemann,
C. L.; Rische, T.; Roggenbuck, R.; Schmidt A. Chem. Rev. 1999, 99, 3329.
65 Cuny, G. D.; Buchwald, S. L. J. Am. Chem. Soc. 1993, 115, 2066.
66 Tolman, C. A. J. Am. Chem. Soc. 1970, 92, 2953 and 2956.
67 (a) Strohmeier, W.; Muller, F. J. Chem. Ber. 1967, 100, 2812. (b) Horrocks Jr., W. D.;
Taylor, R. C. Inorg. Chem. 1963, 2,723.
68 Bartik, T.; Himler, T.; Schulte, H.-G.; Seevogel, K. J. Organomet. Chem. 1984, 272, 29.
69 Grim, S.O.; Yankowsky, A. W. J. Org. Chem. 1977, 42, 716.
70 Trogler, W. C.; Marzilli, J. G. J. Am. Chem. Soc. 1974, 96, 7589.
71 Heimbach, P.; Kluth, J.; Schlenkluhn, H.; Weimann, B. Angew. Chem. 1980, 92, 567.
72 Femandez A.; Reyes C.; Wilson M. R.; Woska D. C.; Prock A.; Giering, M. P.
Organometallics, 1997, 16, 342.
73 Joerg S.; Drago R. S.; Sales J. Organometallics, 1998, 17, 589.
74 Haar, C. M.; Nolan, S. P.; Marshall, W. J.; Moloy, K. G.; Prock, A.; Giering, W. P.
Oganometallics, 1999, 18,474.
75 (a) Hirota M.; Sakakibara K.; Komatsuzaki T.; Akai, I. Comp. Chem. 1991, 15, 241. (b)
White D.; Taverner, B. C.; Coville, N. J.; Wade, P. W. J. Organomet. Chem. 1995, 495,
41. (c) White D.; Taverner, B. C.; Leach, P. G. L.; Coville N. J. J. Organomet. Chem.
1994, 478, 205.
76 Koide, Y.; Bott, S. G . ; Barron, A. R. Organometallics, 1996, 15, 2213.
77 a) Brown, T. L. Inorg. Chem. 1992, 31, 1286. b) Choi, M.-G.; White, D.; Brown, T. L.
Inorg. Chem. 1993, 32, 5591.
78 Angemund, K.; Baumann, W.; Dinjus, E.; Fornika, R.; Gorls, H.; Kessler, M.; Krger,
C.; Leitner, W.; Lutz, F. Chem. Eur. J. 1997, 3, 755.
79 Dierkes, P.; van Leeuwen, P. W. N. M. J. Chem. Soc. Dalton Trans. 1999, 1519.
80 Casey, C. P.; Whiteker, G. T. Isr. J. Chem. 1990, 30, 299.
81 Kranenburg, M.; Kamer, P. C. J.; van Leeuwen, P. W. N. M.; Vogt, D.; Keim, W. J. Chem.
Soc. Chem. Commun. 1995, 2 177; Marcone, J. E.; Moloy, K. G. J. Am. Chem. Soc. 1998,
120,8527.
Chapter 2
Hydroformylation with unmodified rhodium catalysts

Reaction mechanism and origin of regioselectivity
Raffaello Lazzaroni, Roberta Settambolo* and Aldo Caiazzo

Dipartitnento di Chimica e Chimica Industriale, Via Risorgimento 35, 56126 Pisa, Italy;
*Istituto di Chimica Quantistica ed Energetica Molecolare del CNR, Area della Ricerca, Via
Alfieri l, 56010 Ghezzano (PI), Italy.
2.1 Introduction
The first investigations on rhodium-catalyzed hydroformylation were

carried out at the end of 1950s [1], about 20 years after the discovery by
Roelen of the cobalt-catalyzed oxo reaction [2]. Initially, simple catalyst
precursors, such as RhCl3 and Rh/A12O3, were employed. Even at the
beginning it was clear that the rhodium-based catalysts were much more
active than the cobalt based ones and were much more tolerant of the
presence of other functional groups in the unsaturated substrates [3]. The
synthesis and the spectroscopic characterization of rhodium-hydride
complexes containing triphenylphosphine by Wilkinsons group [4] and their
use in the hydrogenation and hydroformylation processes opened the way to
the research on phosphine modified rhodium catalysts [5]. There has been an
enormous amount of research on the synthesis and use of phosphorus- and
sulfur-containing ligands with various steric and electronic characteristics
[6], including optically active ones for use in enantioselective processes [7].
So the phosphorus modified catalysts have been used much more extensively
than the corresponding unmodified ones [5a, 8].
Nevertheless, unmodified Rh catalytic precursors such as Rh(CO)2(acac),
[Rh(COD)(OAc)]2 and Rh4(CO)12 are still the subject of detailed
investigations. As recently reported in the fundamental review of Cornils
(1995): This is due to their easy availability, their well-known properties
and their rather unproblematic handling. Additionally they serve as much
simpler models than modified catalysts. But the main reason for their
15
P. W.N.M. van Leeuwen and C. Claver (eds.), Rhodium Catalyzed Hydroformylation, 15-33.
16 Chapter 2
persistent topicality is the fact that the mechanism of hydrofomylation is far

from being well understood [9].
In this regard we can observe that:
1) Hydroformylations with unmodified catalyst precursors can be carried
out under mild reaction conditions, some of the above catalysts being
active even at room temperature.
2) With most substrates the reaction is completely chemoselective for the
formation of aldehydes. Under mild reaction conditions, the other typical
side processes, such as hydrogenation, isomerization, polymerization and
aldol condensation, are negligible.
3) The above catalyst precursors allow an extensive investigation of the
influence of the reaction parameters as well as of steric and electronic
factors connected to the alkene structure on the regioselectivity of the
reaction, avoiding any interference from phosphine ligands.
As far as the investigation of reaction mechanism is concerned it is to be
noted that:
4) Unmodified catalyst precursors, especially Rh4(CO)12, have been
successfully employed by Garland in fundamental IR spectroscopic
investigations carried out under typical hydroformylation conditions.
Information on the kinetic aspects of the process and on the reactivity of
the acyl-rhodium intermediates has been obtained [10].
5) The same catalyst precursor, Rh4(CO)12, has been extensively employed
in deuterioformylation experiments at partial substrate conversion. 2H-
NMR analyses carried out directly on the crude reaction mixture provide
information on the behavior of alkyl-rhodium intermediates under oxo
conditions [11].
This chapter will discuss the factors affecting the reaction regioselectivity of
several alkenyl substrates and the related mechanistic aspects arising from in
situ IR spectroscopic studies and deuterioformylation experiments.
2.2 Regioselectivity in the rhodium-catalyzed

hydroformylation of vinyl and vinylidenic substrates
It is well known that the main goal in rhodium-catalyzed
hydroformylation of unsaturated, especially vinyl, substrates, concerns the
control of reaction regioselectivity, i.e. of the regioisomeric ratio b:1
(branched to linear) between the isomeric aldehydes produced.
2. Hydroformylation with unmodified rhodium catalysts 17
Figure 1 Regioisomeric aldehydes obtained from the rhodium-catalyzed hydroformylation of

vinyl alkenes
The influence of substrate structure as well as of reaction parameters on

the reaction regioselectivity, in the presence of unmodified rhodium
catalysts, has been extensively investigated. The main results will be
discussed in the following sections.
2.2.1 Catalyst precursors
The following unmodified catalyst precursors are those most frequently

employed in the hydroformylation of vinyl and vinylidenic substrates:
Rh 4(CO) 12 [12], [Rh(CO) 2(C1)] 2 [13], [Rh(CO) 2(acac)] 2 [14],
[Rh(cod)(OAc)]2 [ 15], and zwitterionic complexes such as [BPh 4]-
[Rh(COD)]+ (Rhzw) [16]. These last ones, extensively used by Alper and
coworkers, can be included in the above group, since they are generally used
without any ancillary ligand.
As reported by several authors [8, 10], Rh4(CO)12 generates, under
hydroformylation conditions, a rhodium-carbonyl hydride [HRh(CO)3],
which constitutes the catalytic active species of the reaction. A similar
process probably occurs with the other catalyst precursors. No information
on the course of reaction with Rhzw systems has been reported [16].
Nevertheless it is likely that an analogous hydride species is formed also
with this precursor.
2.2.2 Influence of the alkene structure on the regioselectivity
The rhodium-catalyzed hydroformylation of many simple and

functionalized substrates in the presence of phosphorus ligands has been
extensively investigated. The comparative results are reported in several
reviews [7c, 9]. By contrast the results for alkenyl substrates
hydroformylated with unmodified rhodium-based precursors are relatively
few, the mechanistic aspects of the reaction and not the synthetic goal being
the main purpose of these studies.
These substrates can be divided into three different classes, on the basis
of their structure:
a) vinyl substrates; b) allyl substrates; c) vinylidenic substrates.
18 Chapter 2
In order to examine the influence of the alkene structure on the

regioselectivity, a comparison of the results obtained with different
substrates and different catalyst precursors under mild reaction conditions
(20-50 C, 50-160 bar of CO/H2 , 1:1) will be made.
a) Vinyl substrates
These substrates cannot isomerize into internal alkenes, due to the

absence of hydrogens in a position with respect to the double bond.
The most investigated alkene belonging to this group is styrene. The
hydroformylation of this substrate under mild temperature always provides a
large predominance of the branched isomer over the linear one (b:l = 95/5-
98/2), as shown in Table 1. The sole branched isomer is obtained also in the
hydroformylation of fluorinated alkenes, namely, fluoroethene, 3,3,3-
trifluoropropene and pentafluorostyrene in the presence of Rh4(CO) 12 [ 12b].
Heteroaromatic vinyl substrates, such as vinylfurans [17], vinylthiophenes
[18], vinylpyrroles [19] and vinylpyridines [ 14, 20], behave in a very similar
manner to styrene, giving a high prevalence of the branched aldehyde.
Isomeric vinylpyridines show a quite singular behavior: whilst 3-
vinylpyridine show the typical trend of vinylaromatic alkenes, 4-
vinylpyridine gives rise to the hydrogenation product 4-ethylpyridine with
more than 96 % selectivity [20]. As far as oxygen containing substrates are
concerned, vinyl acetate exclusively gives the branched isomer (b:l 99/1)
[ 15], (vinyl)alkyl ethers provide a regioisomeric ratio b:l = 80/20 [ 1 Id],
whilst the presence of a phenyl group directly bonded to the oxygen
enhances the regioselectivity towards the branched aldehyde (b:l = 95/5 for
PhOCH=CH2) [21]. ,-Unsaturated aldehydes and ketones give
hydrogenation of the double bond only under oxo conditions [5a]. By
contrast, alkyl alkenes containing a tertiary carbon atom bonded to the vinyl
group, such as 3,3-dimethylbut- 1 -ene, show the opposite regioselectivity,
giving the linear aldehydic isomer almost exclusively [15].
b) Allyl substrates
It is well known that alkenes with a hydrogens (linear 1-alkenes,

allylethers, etc.) can give more than two regioisomers when migration of the
double bond into internal positions occurs and hydroformylation of the
resulting internal alkene takes place [22]. However, under mild reaction
conditions, particularly at room temperature, there is no isomerization of the
starting substrate, hence only two regioisomers are observed in all cases.
Very similar amounts of branched and linear aldehydes are obtained for
all the linear 1-alkenes (i.e. b:l = 48/52 for 1-hexene at 20 C and 100 bar
CO/H2, 1:1) [22], while an oxygen in b position to the vinyl group favors the
formation of the branched isomers as observed for (allyl)ethyl ether (b:l =
70/30) and similar substrates [11d]. It should be noted that the linear isomer
largely predominate over the branched ones in the hydroformylation of 3-
alkyl substituted allyl alkenes (i.e. 3-methylbut-1-ene) [5a].
c) Vinylidenic alkenes
(R, R' = Ar, alkl )
The hydroformylation rate in the case of vinylidenic alkenes is very low

at room temperature, so the reaction is usually carried out at temperatures
higher than 80 C. Whatever kind of Z substituent is present (dialkyl,
arylalkyl, diaryl) the linear isomer is almost exclusively produced [11f, 23].
Only when one of the substituents is a 2-pyridyl group does the branched
isomer predominates over the linear one; in these cases a high amount of
hydrogenation products is obtained [24].
In conclusion, unsaturated vinyl substrates can give opposite
regioselectivities depending on the steric and electronic nature of the
substituent bonded to the alkenyl moiety. When this substituent is a phenyl
or an oxygen the branched aldehydic isomers predominates. By contrast,
bulky groups favor the formation of the linear aldehyde, which is observed
also in the case of vinylidenic substrates.
20 Chapter 2
Table I. Isomeric composition of aldehydic products arising from styrene hydroformylation

carried out in the presence of unmodified rhodium-based precursorsa.
C
Catalyst T P Reaction times b:1 Reference
precursor (C) (bar)b (h)
Rh4(CO)12 60 150 5 95/5 12c
Rh4(CO)12 25 150 9.5 98/2 12c
[Rh(COD)(OAc)]2 25 50 16 96/4 15
[Rh(COD)(Cl)]2 80 40 1.5 95/5 13

zw d
[Rh ] 47 14 22 98/2 16
Rh(CO)2(acac) 30 90 16 97/3 14
a
At complete substrate conversion
b
CO/H2= 1:l
c
d
Regioselectivity
Zwitterionic rhodium complex [BPh4]-[Rh(COD)]+
Table 2. Selected values of regioisomeric ratio in the hydroformylation of unsaturated

substrates in the presence of unmodified rhodium precursorsa.
Substrates T P Reaction times b:l c References

(C) (bar)b (h)
FIuoroethene 80 110 6 100/0 12b
3,3,3-Trifluoropropene 80 110 6 97/3 12b
Substituted styrenes 20 60 16 95/5-98/2 15
(Vinyl)ethylether 20 100 9 83/17 11d
3,3-Dimethylbutene 20 60 16 0/100 15
Vinyl acetate 20 60 16 > 99/1 15
1-Hexene 15 100 6 52/48 22b
(Allyl)ethyl ether 20 100 6 70/30 11d
2-phenyl propene 100 100 3 < 1/99 11f
1,1-Diphenylethene 100 100 20 < 1/99 11f
2-Methylpropene 100 100 1 0/100 11f
a
At complete substrate conversion
b
CO/H2= 1:l
c
Regioselectivity
2.2.3 Influence of temperature
Systematic studies on the influence of temperature on the regioselectivity

in the hydroformylation of vinyl substrates in the presence of unmodified
rhodium-based precursors have been carried out only in few cases. In
particular the investigations reported in literature concern the
hydroformylation of styrene, ethyl- and allyl ethers and 1-hexene, with
Rh4(CO)12 over temperatures, ranging from 20 C to 100 C.
In the case of styrene a strong increase of linear aldehydic isomer with
increasing temperature is observed (b:1 = 98/2 at 20 C to 64/36 at 130 C)
[12c]. For (ethyl)vinyl ether the above increase is lower, the percentage of
linear aldehyde ranging from 12% at 20 C to 24% to 100 C [11d]. In the
case of (phenyl)vinyl ether, which shows a high a-regioselectivity at 20 C
(b:l = 95/5), a negligible temperature effect is obtained [21]. In all these
cases no variation of the regioisomeric ratio with increasing substrate
conversion is observed.
Figure 2. Influence of temperature on the hydroformylation regioselectivity of selected

substrates in the presence of Rh4(CO)12 as catalyst precursor
The hydroformylation of allylic alkenes is characterized, at high

temperatures, by competing isomerization [22]. Significant regioselectivity
ratios for the hydroformylation of the sole terminal double bond have been
obtained by carrying out the reaction at partial substrate conversion, when
the starting terminal alkene is still present in excess relative to that resulting
from isomerization [22b]. In 1 -hexene hydroformylation the amount of
22 Chapter 2
linear isomer increases with increasing temperature, ranging from 52% at 20

C to 72% at 100 C. As internal alkenes are converted into aldehydes only
when all the terminal alkene has reacted, it is possible to estimate, at partial
substrate conversion, the chemoselectivity of the reaction, i. e. the amount of
1 -hexene converted into aldehydes (b+l) with respect to that isomerized to
internal alkene (E-2). Chemoselectivity to aldehydes decreases with
increasing temperature, (b+l)/E-2 reaching the value 60/40 at 120 C. The
effect of temperature is greater in the case of (allyl)ethyl ether, for which the
percentage of linear isomer increases from 30% at 20 C to 60 % at 100 C.
A slight increase of linear aldehyde with increasing temperature is observed
also for propylene, which cannot give any internal alkenyl product.
In the case of vinylidenic alkenes the linear aldehydic isomer is obtained
with a complete selectivity at any temperature [11f, 23].
2.2.4 Influence of CO and H2 partial pressures
In the case of styrene the CO and H2 partial pressures affect the reaction
regioselectivity only when the reaction is carried out at high temperatures. In
particular it has been observed that a decrease of carbon monoxide or
hydrogen partial pressure causes an increase of linear aldehydic isomer, this
effect being more evident at higher temperatures (100 C). So in the case of
styrene hydroformylation at 100 C the b:l ratio ranges from 80/20 at 170
bar of CO/H2 (1:1) to 56/44 at PH2 = 6 bar, PCO = 85 bar or to 60/40 at PH2 =
85 bar, PCo = 6 bar [12c]. In the case of 1-hexene gas pressure does not
affect the regioselectivity of the reaction either at room temperature or at
high temperature. By contrast the chemoselectivity to aldehydes increases
with increasing temperature, (b+l)/E-2 being 44/56 at 40 bar and 77/23 at
140 bar [22b].
2.3 Mechanism of the hydroformylation of vinyl and

vinylidenic alkenes
As described above, both the nature of the substrate and the reaction
conditions strongly influence the regioselectivity in the hydroformylation of
vinyl substrates. The above results clearly demonstrate that, by raising the
reaction temperature, and decreasing the CO and H2 partial pressures, the
amount of linear aldehydes increases. Indeed, this is a general trend in the
hydroformylation of different substrates and constitutes a fundamental
starting point for a rationalization of the influence of experimental
parameters on the reaction selectivity.
In this context we are going to examine the above results in the light of
the generally accepted mechanism for hydroformylation, taking into account
the more recent findings on the behavior under reaction conditions of the
main intermediate species, namely alkyl- and acyl-rhodium complexes. A
simplified scheme for the hydroformylation of a typical vinyl substrate is
shown in Figure 3.
The rhodium hydride tricarbonyl species easily coordinates the vinyl
substrate generating the complex (1), which is converted into the alkyl-
rhodium intermediates (2) through insertion of the alkene into the Rh-H
bond. Migratory insertion of the alkyl moiety on to a CO molecule
coordinated to the metal center provides the acyl-rhodium species 3, which,
at the end of the catalytic cycle, interacts with hydrogen via an oxidative
addition, giving rise to aldehydic products and regenerating the rhodium-
hydride species.
Figure 3. Generally accepted mechanism for the rhodium-catalyzed hydroformylation

24 Chapter 2
2.3.1 Activation of the catalyst precursor
The mechanism of fragmentation of the rhodium cluster Rh4(CO)12

under oxo conditions has been extensively studied during recent years; in the
course of these investigations the presence of low nuclearity species has
been proposed [25]. Only since Garland and co-workers began studying the
fragmentation of Rh4(CO)12 in the presence of alkenyl substrates via in situ
IR spectroscopy the mechanism of transformation of rhodium-clusters under
oxo condition become clearer. The disappearance of the typical bands due to
Rh4(CO)12 and the appearance of those due to the acyl-rhodium
intermediates were investigated under different experimental conditions in
order to determine a kinetic expression for the fragmentation process [10].
In a paper by Garland [10a] concerning the hydroformylation of 3,3-
dimethylbut- 1 -ene (33DMB) with Rh4(CO)12, the cluster fragmentation was
investigated and a mechanism proposed, according to the following kinetic
expression
rate (I) = k0(I) [Rh4(CO)12][CO].8[H2]0.7[33DMB]0.
This equation is consistent with i) a preequilibrium between Rh4(CO)12,

CO and a second unstable cluster, Rh4(CO)14 and ii) a rate-limiting step
involving the activation of the latter cluster by H2.
Figure 4. Rh4(CO)12 cluster conversion into acyl-metal intermediate
As reported in the same paper, it is likely that this unstable rhodium

cluster is converted into the mononuclear rhodium-hydride species
HRh(CO)x (x = 3,4), which are usually considered as the true catalyst system
in the reaction mixture. These compounds represent extremely unstable
intermediates, which would certainly recombine to form higher nuclearity
rhodium species if alkene is not present in the reaction mixture. This
mechanism is proposed for all the hydroformylation experiments carried out
in the presence of Rh4(CO)12.
2.3.2 Behavior of the isomeric alkyl-metal intermediates via

deuterioformylation experiments
The crucial step that determines regioselectivity is the alkene insertion

into the Rh-H bond which gives rise to the alkyl-metal intermediates. This
step can be reversible or irreversible, depending on the reaction conditions.
Deuterioformylation experiments carried out at partial substrate conversion

has proved to be the best way to investigate the reversibility of the above
step [11]. As shown in Figure 5, when a deuterated alkyl species undergoes a
-hydride elimination process, the elimination of Rh-H is favored over that
of Rh-D one, because of the well documented kinetic isotope effect observed
in this kind of process [26]. Thus -hydride elimination from the linear alkyl
species gives rise to an alkene deuterated at the carbon atom in position 2,
whilst the analogous process for the branched alkyl intermediate generates
an alkene deuterated at the terminal position of the double bond.
Examination by 2H-NMR spectroscopy of the crude deuterioformylation
mixture at partial substrate conversion gives direct information, both
qualitative and quantitative, on the occurrence of a -elimination process, i.e.
on the reversibility of formation of the alkyl intermediates.
As a typical example, the H-NMR spectra of a mixture resulting from
deuterioformylation of (ethyl)vinyl ether at 20 C and 100 C and 30%
substrate conversion are shown in Figure 6 [11d].
At 100 C -hydride elimination occurs for both the alkyl-rhodium
species; this is evident from the presence of Et-O-CH=CHD (1-d) (signals at
4.10 and 3.97 ppm) derived from the branched isomer and by the presence of
Et-O-CD=CH2 (2-d) (signal at 6.44 ppm) derived from the linear one.
R = OEt
Figure 5. Rhodium-catalyzed deuterioformylation of (ethyl)vinyl ether

26 Chapter 2
By contrast, at 20 C, no signals due to the deuterated alkene are present,

indicating that no -elimination process takes place and hence the formation
of the alkyl intermediates is not a reversible step. It is noteworthy that
analogous experiments carried out at room temperature with styrene as well
as 1-hexene [1la] and (allyl)ethyl ether [11d], clearly show that no -hydride
elimination occurs under these mild conditions. This means that the alkene
insertion into the Rh-H bond is not reversible and hence the isomeric alkyl-
rhodium intermediates are directly transformed into the acyl species to give
isomeric aldehydes. Deuterioformylation experiments also give other
interesting information. As far as hydrofomylation of (etlyl)vinyl ether at
100 C is concerned, comparing the relative intensities of alkenyl absorption
and formyl group at 9.46 ppm, it clear that the ratio between the above
absorptions [( alkenyl deuteriums)/(formyl deuterium)] increases as the
temperature is raised. It has also been observed that -hydride elimination
occurs more readily for the branched alkyl intermediate than for the linear
one, as shown by the lower intensity of the signal at 6.44 ppm with respect to
those at 3.97-4.14 ppm. Analogous behavior has been observed also for
allylic substrates, e.g. 1-hexene and (allyl)ethyl ether. In these cases the -
hydride elimination from the branched alkyl gives mainly the isomeric
internal alkene deuterated at the terminal position (Z-CH=CH-CH2D)
(Figure 5).
By contrast, in the case of styrene [12c] and other aromatic substrates
[ 19a] -hydride elimination occurs selectively for the branched intermediate,
at least at temperatures lower than 100 C, since the signal at low fields,
2
typical of 2-d-alkene, is absent from the H NMR spectra.
Deuterioformylation at 80- 100 C of vinylidenic alkenes containing one
or two phenyl groups on the double bond [1l e-f] clearly shows, in addition
to the linear aldehyde, the formation of alkenes deuterated in the terminal
position, arising from -elimination from the tertiary alkyl-metal
intermediates. However, there are no traces of the quaternary aldehyde
deriving from migratory insertion of this alkyl group to CO and subsequent
oxidative addition of H2. In contrast, deuterioformylation experiments
carried out on vinylidenic alkenes containing two alkyl groups [1If] show
that the formation of a tertiary alkyl-metal intermediate occurs only to a very
low extent (2-methylpropene) or that it does not occur at all (2,3,3-
trimethylbutene).
Figure 6. H-NMR spectrum (46 MHz, 25 C, C6D6 as external standard) of the crude
mixture resulting from deuterioformylation of (ethyl)vinyl ether at (a) 20 C and (b) 100 C
28 Chapter 2
2.3.3 In situ IR investigation of the formation and reactivity of acyl-

rhodium intermediates
Acyl intermediates are the only detectable reaction intermediates under

typical hydroformylation conditions; they can be relatively stable under CO
atmosphere and at mild temperatures. Using in situ IR spectroscopy, Garland
and co-workers have made an important contribution in making clear several
kinetic and mechanistic aspects of the behavior of acyl intermediates [10].
The most interesting aspect of the reactivity of such species is the oxidative
addition of H2 and the following reductive elimination to provide the final
aldehydic products, with the regeneration of the metal-hydride complex. The
first studies were carried out with alkenes generating only an aldehydic
isomer, such as 33DMB [10a] and cyclohexene [lob]. In the case of 33DMB
the existence of a preequilibrium between the rather stable and detectable
intermediate RCORh(CO)4, CO and the unsaturated reactive complex
RCORh(CO) 3, is observed; the latter undergoes oxidative addition of H2 to
provide the final aldehydic products, which represents the rate-limiting step
of the reaction.
Figure 7. Hydrogenolysis of acyl-metal intermediates
The whole process can be described by the following kinetic expression
rate (11) = k0(II)[RCORh(CO)4]1[CO]-. [H 2][33DMB]0.1
Investigations carried out on several alkenyl substrates indicate that in all

the cases the acyl-rhodium species formed under hydroformylation
conditions have a bipyramidal geometry, with Cs symmetry, the acyl group
taking an axial position (Figure 8) [10e].
Figure 8. Structure of acyl-metal intermediates

Recently a very detailed investigation of the mechanism of the interaction

between the isomeric acyl-metal species and H2 under different reaction
conditions in the case of styrene was carried out [ 10d]. Both the isomeric
acyl-rhodium intemediates were observed, and their hydrogenolysis to give
aldehydic products and the relative kinetics analyzed under different reaction
conditions. The kinetic expression derived for the whole process is
rate=k0 [RCORh(CO)4]ss1.0[CO]-1.0[H2]1.0[C8H8]0.0
Of particular interest is the effect of reaction temperature on the reaction

rate and regioselectivity. The most remarkable results from the experiments
carried out at P(CO) = 50 bar and P(H2) = 5 bar, in the range 298-313 K, are
summarized in Table 3.
The hydrogenolysis rate of the linear acyl-metal intermediate is higher
than the one of the branched isomer, the difference being much more marked
at 40 C than at 25 C. The regioisomeric ratio between the acyl
intermediates at 40 C is higher than the one between the corresponding
aldehydes; by contrast, at 25 C the two values are quite similar.
Table 3. Selected values of kinetic constants and regioisomeric ratios for styrene
hydroformylation in the presence of Rh4(CO)12 as catalyst precursor, at 25 C and 40 C.
Values Temperature
25 C 40 C
kb 0.93 4.18
kn 1.27 10.3
3b:31a 97.5/2.5 87.51/12.5
b:1 96.7/3.3 66/34
a
Regioisomeric ratio between branched (3b) and linear (31) acyl-rhodium intermediates.
2.4 Origin of the regioselectivity
2.4.1 Influence of the nature of the substrate
As previously shown by deuterioformylation experiments, when the

reaction is carried out at low temperature, the formation of alkyl-metal
intermediates is not a reversible step. Under these conditions the
regioselectivity observed for aldehydic isomers is directly determined in the
step at which the alkyl metal intermediates are formed. Taking into account
30 Chapter 2
the structure of the linear and branched alkyl-metal intermediates it is

possible to explain why the branched aldehydes are strongly favored in the
case of styrene or other functionalized substrates, whereas in the case of
simple 1 -alkenes approximately equal amounts of aldehydic isomers are
formed.
As shown in Figure 9, the metal-carbon bond in the alkyl-rhodium
intermediates is polarized with a partial positive charge on the metal and a
partial negative charge on the carbon atom. When this carbon atom is
bonded to a strongly polarizable group (i.e. -C6H5) or to an electron
withdrawing group (i.e. -F, -OR, -CH2OR, -CF3), the partial negative charge
on the carbon atom is better delocalized owing to the inductive effect in the
branched isomer 2b than in the linear one 21 [12b-c].
When R is an electron donor group (n-alkyl group), no delocalization of
the partial negative charge occurs for either isomer. As a consequence the
branched and linear alkyl intermediates are formed in similar amounts, hence
so are the corresponding aldehydes (Figure 9).
However, when the alkyl group bonded to the vinyl moiety has a
secondary or tertiary structure, steric hindrance plays in crucial role on the
regioselectivity, causing the linear aldehyde to predominate.
Figure 9. Stabilization of alkyl-rhodium intermediates arising from the hydroformylation

of different alkenes
As far as the vinylidenic substrates are concerned, deuterioformylation of

phenyl substituted vinylidenic alkenes gives interesting information about
the formation of a tertiary alkyl intermediate under reaction conditions.
Indeed, the formation of vinylidenic alkenes deuterated at the terminal
position to a larger extent than the linear aldehyde, demonstrates that the
branched alkyl predominates over the linear one. As previously mentioned,
this is due to the higher stabilization induced by the two phenyl groups
adjacent to the carbon-rhodium bond. However the migratory insertion on to
the CO coordinated to the metal in the case of tertiary alkyls is prevented by
steric reasons. Thus it seems evident that the behavior of the two isomeric
alkyl-rhodium intermediates is completely different: while the primary one is
converted into the linear aldehyde, the tertiary one exclusively undergoes -
hydride elimination, regenerating the starting alkene [1le].
2
In conclusion, the H-NMR analysis of crude deuterioformylation
products derived from vinyl or vinylidenic aromatic substrates is a direct and
simple way to detect the different behavior of a primary, secondary and
tertiary alkyl-metal intermediate, related to the -hydride elimination process
under typical hydroformylation conditions.
2.4.2 Influence of the reaction parameters
As far as the influence of reaction parameters, observed for vinyl and

allyl substrates, is concerned, the increase of linear aldehyde with increasing
temperature can be easily explained on the basis of the different behavior of
the alkyl-rhodium intermediates under the reaction conditions. Thus the
linear alkyl mainly undergoes the migratory insertion process and, hence,
gives the linear aldehyde. In contrast, the branched one undergoes
carbonylation only partially, mainly providing -hydride elimination. It is to
be noted that the complex derived from the above elimination process
regenerates both the linear and the branched alkyls. Thus the whole process
brings about a partial isomerization of the branched alkyl isomer to the linear
one and hence determines an increase of linear aldehyde. The different
behavior of isomeric alkyl-rhodium intermediates could account also for the
increase of linear aldehyde with decreasing CO and H2 pressure. At high gas
pressure both the intermediate alkyls are forced to take part in the
carbonylation to provide the aldehydic products. At low pressure, the -
elimination process becomes competitive with the acyl formation and with
the subsequent oxidative addition of H2. Because the above elimination
process is favored in the case of branched alkyl-metal species, the final result
will be an increase of linear aldehyde.
On this basis it is possible to explain the results obtained by Garland in
the hydroformylation of styrene under relatively mild reaction conditions. It
is plausible that the b:l ratio = 66/34 obtained at 40 C, which is lower than
the one observed for the acyl-rhodium species (3b/31 = 87.5/12.5), is due to
the -elimination process which is much more favorable for the branched
alkyl intermediate than for the linear one.
Phosphine ligands, when employed in excess with respect to rhodium,
generally block the -elimination process, as shown by deuterioformylation
32 Chapter 2
experiments carried out by Casey [27] and Takaya [7a], thus accounting for
the low variation of regioselectivity with temperature obtained in the
presence of phosphine-modified precursors [20b].
It is to remark that in the hydroformylation of styrene, the most
investigated vinyl aromatic substrate, the predominance of the branched
aldehyde at room temperature is higher with unmodified rhodium precursors
than with phosphine-modified ones [4, 7c, 28]. In this context, when
hydroformylation of styrene with chiral phosphines occurs without
asymmetric induction and with a large prevalence of the branched aldehyde
(> 96%), it is likely that unmodified rhodium-catalysts are also present in the
reaction mixture [ 14, 29, 30].
References
1 (a) Schiller, G. Ger. Pat. 965,605 1956 (to Chem. Verwertungsges. Oberhausen); Chem
Abstr. 1959, 53, 11226. (b) Hughes, V. L. Br. Pat. 801,734 1958 (to Esso Res. Eng
Comp.); Chem Abstr. 1959, 53, 7014.
2 Roelen, O .Ger. Pat. 849,548 1938; Chem. Zentr. 1953, 927.
3 Organometallic Chemistry of Transition Elements, ed. F. P. Pruchnik, Plenum Press,
New York, 1990, p 691.
4 Evans, D.; Osborn, J. A.; Wilkinson, G. J. Chem. Soc., A 1968. 3133.
5 (a) Organic Syntheses via Metal Carbonyls, eds. I. Wender and P. Pino, Wiley-
Interscience, New York, 1977, Vol. 2. (b) Homogeneous Catalysis with Metal
Phosphine Complexes, ed. L. H. Pignolet, Plenum Press, New York, 1983. (c) Applied
Homogeneous Catalysis with Organometallic Compounds, eds. B. Cornils and W. A.
Herrmann, VCH, Weinheim, 1996.
6 (a) Casey, C. P.; Paulsen, E. L.; Beuttenmueller, E. W.; Proft, B. R.; Matter, B. A.;
Powell, D. R. J. Am. Chem. Soc. 1999, 121, 63. (b) Casey, C. P.; Paulsen, E. L. ;
Beuttenmueller, E. W.; Proft, B. R.; Petrovich, L. M.; Matter, B. A.; Powell, D. R. J.
Am. Chem. Soc. 1997, 119, 11817. (c) van der Van, L. A.; Kamer, P. C. J.; Van
Leeuwen, P. W. N. M. Organometallics 1999, 18, 3765. (d) van Rooy, A.; Kamer, P. C.
J.; Van Leeuwen, P. W. N. M.; Goubitz, K.; Fraanje, J.; Veldman, N.; Spek, A.;
Organometallics 1996, 15, 835.
7 (a) Nozaki, K.; Sakai, N.; Nanno, T.; Higashijima, T.; Mano,, S.; Horiuchi, T.; Takaya,
H. J. Am. Chem. Soc. 1997, 119, 4413. (b) Nozaki, K.; Nanno, T.; Takaya, H. J.
Organomet. Chem. 1997, 527, 103. (c) Gladiali, S.; Bayn, J. C.; Claver, C.
Tetrahedron: Asymm. 1995, 6, 1453. (d) Bayn, J. C.; Claver, C.; Masdeu-Bulto, A. M.
Coord. Chem. Rev. 1999, 195, 73. (e) Miquel-Serrano, M. D.; Masdeu-Bulto, A. M.;
Claver, C; Sinou, D. J. Mol. Cut., A: Chemical 1999, 143, 49.
8 (a) Thatchenko, I. Comprehensive Organometallic Chemisty, eds. G . Wilkinson, F. G.
A. Stone and E. W. Abel, Pergamon, Oxford, 1982, Vol 8, pp. 101. (b) Transition
metals for Organic Synthesis, eds. M. Beller and C. Bolm, Wiley VCH, 1999.
9 Beller, M.; Cornils, B.; Frohning, C. D.; Kohlpaintner, C. W. J. Mol. Cat. A: Chem.
1995, 104, 17.
10 (a) Garland, M.; Pino, P. Organometallics 1991, 10, 1693. (b) Garland, M.
Organometallics 1993, 12, 535. (c) Fyhr, C.; Garland, M. Organometallics 1993, 12,
1753. (d) Feng, J.; Garland, M. Organometallics 1999, 18 ,417. (e) Guowei, L.; Volken,
R.; Garland, M Organometallics 1999, 18, 3429.
11 (a) Lazzaroni, R.; Uccello-Barretta, G.; Benetti, M. Organometallics 1989, 8, 2323. (b)
Raffaelli, A.; Pucci, S.; Settambolo, R.; Uccello-Barretta, G.; Lazzaroni, R.
Organometallics 1991, 10, 3892. (c) Uccello-Barretta, G.; Lazzaroni, R.; Settambolo,
R.; Salvadori, P. J. Organomet. Chem. 1991, 417, 111. (d) Lazzaroni, R.; Settambolo,
R.; Uccello-Barretta, G. Organometallics 1995, 14, 4644. (e) Lazzaroni, R.; Uccello-
Barretta, G.; Scamuzzi, S.; Settambolo, R.; Caiazzo, A. Organometallics 1996, 15,
4657. (f) Lazzaroni, R.; Settambolo, R.; Uccello-Barretta, G.; Caiazzo, A.; Scamuzzi, S.
J. Mol. Cat., A: Chemical 1999, 143, 123.
12 (a) Pino, P.; Oldani, F.; Consiglio, G. J. Organomet. Chem. 1983, 250, 491, (b) Ojima,
I. Chem. Rev. 1988, 88, 1011. (c) Lazzaroni, R.; Raffaelli, R.; Settambolo, R.; Bertozzi,
S.; Vitulli, G. J. Mol. Cat. 1989, 50, 1.
I3 Botteghi, C.; Paganelli, S.; Bigini, L.; Marchetti, M. J. Mol. Cat. 1994, 93, 279.
14 Basoli, C.; Botteghi, C.; Cabras, M. A.; Chelucci, G.; Marchetti, M. J. Organomet.
Chem. 1995, 488, C20.
15 Doyle, M. P.; Shanklin, M. S.; Zlokazov, M. V. Synlett 1994, 615.
16 Amer, I.; Alper, H. J. Am. Chem. Soc. 1990, 112, 3674.
17 (a) Kalck, P.; Serein-Spiran, F. New J. Chem. 1989, 13, 515. (b) Lapidus, A. L.; Rodin,
A. P.; Pruidze, I. G.; Ugrak, B. I. Izv. Akad. Nauk SSSR, Ser. Khim. 1990, 7, 1661.
18 Browning, A. F.; Bacon, A. D.; White, C.; Milner, D. J. J. Mol. Catal. 1993, 83, L11.
19 (a) Caiazzo, A.; Settambolo, R.; Uccello-Barretta, G.; Lazzaroni, R. J. Organomet.
Chem. 1997, 548, 279. (b) Lazzaroni, R.; Settambolo, R.; Mariani, M.; Caiazzo, A. J.
Orgamomet. Chem. 1999, 592, 69.
20 (a) Settambolo, R.; Scamuzzi, S.; Caiazzo, A.; Lazzaroni, R. Organometallics 1998, 17,
2127. (b) Caiazzo, A.; Settambolo, R.; Pontorno, L.; Lazzaroni, R. J. Organomet.
Chem. 2000, in press.
21 Lazzaroni, R.; Bertozzi, S.; Pocai, P.; Troiani, F.; Salvadori, P. J. Organomet. Chem.
1985, 295, 371.
22 (a) Hanson, B. E.; Davis, N. E. J. Chem. Educ. 1987, 64, 928. (b) Lazzaroni, R.; Pertici,
P.; Bertozzi, S.; Fabrizi, G. J. Mol. Catal. 1990, 58, 75.
23 Botteghi, C.; Cazzolato, L.; Marchetti, M.; Paganelli, S. J. Org. Chem. 1995, 60, 6612.
24 Botteghi, C.; Marchetti, M.; Paganelli, S.; Sechi, B. J. Mol. Catal. A: Chem. 1997, 118,
173.
25 (a) Vidal. J. L.; Walker, W. E. Inorg. Chem. 1981, 20, 249 (b) Oldani, F.; Bor, G. J.
Organomet. Chem. 1983, 246, 309.
26 Evans, J.; Schwartz, J.; Urquhart, P. W. J. Organomet. Chem. 1974, 81, C37.
27 Casey, C. P.; Petrovich, L. M. J. Am. Chem. Soc. 1995, 117, 6007.
28 Sakai, N; Mano, S.; Nozaki, K.; Takaya, H. J. Am. Chem. Soc. 1993, 115, 7033.
29 Brown, J. M.; Cook, S. J.; Khan, R. Tetrahedron 1986,42, 5105.
30 Eckl, R. W.; Priermeier, T.; Herrmann, W. A. J. Organornet. Chem. 1997, 532, 243.
Chapter 3
Rhodium Phosphite Catalysts
Paul C. J. Kamer, Joost N. H. Reek, and Piet W. N. M. van Leeuwen

Institute of Molecular Chemistry. University of Amsterdam, Nieuwe Achtergracht 166, 1018 WV
3.1 Introduction
The important discovery by Wilkinson [1] that rhodium afforded active

and selective hydroformylation catalysts under mild conditions in the
presence of triphenylphosphine as a ligand triggered a lot of research on
hydroformylation, especially on ligand effects and mechanistic aspects. It is
commonly accepted that the mechanism for the cobalt catalyzed
hydroformylation as postulated by Heck and Breslow [2] can be applied to
phosphine modified rhodium carbonyl as well. Kinetic studies of the
rhodium triphenylphosphine catalyst have shown that the addition of the
alkene to the hydrido rhodium complex and/or the hydride migration step is
probably rate-limiting [3] (Chapter 4). In most phosphine modified systems
an inverse reaction rate dependency on phosphine ligand concentration or
carbon monoxide pressure is observed [4].
Since p-back bonding contributes significantly to the strength of the
metal-to-ligand bond, especially for carbonyl ligands, it is not surprising that
electron-withdrawing substituents on the ligands increase the reaction rate as
a result of the more facile CO dissociation and stronger alkene association
[5, 6]. Because phosphites are better -acceptors than phosphines they have
great potential in hydroformylation catalysis. An additional advantage of
phosphites is that they can be more easily prepared than phosphines and are
less sensitive to sulfur compounds and oxidizing agents. On the other hand
phosphites are more sensitive to side reactions such as hydrolysis,
alcoholysis, and the Arbuzov reaction. Shortly after the discovery of
Wilkinson, Pruett and Smith of UCC reported the beneficial effect of
35
P.W.N.M. van Leeuwen and C. Claver (eds.), Rhodium Catalyzed Hydroformylation, 35-62.
36 Chapter 3
phosphite ligands in the rhodium catalyzed hydroformylation of 1 -octene

and methyl methacrylate [7]. Their results indicated that electron-
withdrawing ligands showed a higher tendency to form linear aldehyde using
1-octene as substrate. Although the relative stability of primary and
secondary alkyl complexes is determined mainly by steric factors, electronic
effects can also be important. Strongly electron-withdrawing ligands will
create a higher positive charge on the metal, which might favor the linear
metal alkyl complex formation. For diphosphine ligands electron-
withdrawing substituents resulted in an increased preference for
bisequatorial coordination of the phosphorus donors [6]. Since phosphites
have a much higher value than phosphines, a high preference for
bisequatorial coordination is anticipated. The effect of the coordination
mode on the selectivity, however, is not always clear. For
triphenylphosphine Brown and Kent concluded that coordination of two
phosphines in the equatorial plane resulted in the highest linearity [13],
whereas the study of structurally related diphosphines showed that the
selectivity for the linear aldehyde was independent of the relative amount of
bisequatorially coordinating complex [6].
In their early reports 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. Systematic studies of
ligand effects on the hydroformylation reaction are often obscured by the
presence of several catalytically active rhodium complexes in the reaction
mixture (see Figure 1). These complexes containing different numbers of
phosphorus ligands are in equilibrium and up to three phosphorus ligands
they can all be active as hydrofomiylation catalyst. The composition of the
equilibrium mixture is dependent on many reaction parameters, such as type
of ligand, concentration, temperature and pressure.
Figure 1. Actual rhodium catalyst containing various phosphorus and carbonyl ligands
The complexes of Figure 1 will show different rates and selectivities in

the hydroformylation of 1 -alkenes. In general, phosphorus ligands are
stronger -donors and weaker -acceptors than carbonyl ligands. As a
consequence of the larger size, phosphines and phosphites create more steric
bulk in the rhodium complex. 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 phosphites results in the
3. Rhodium phosphite catalysts 37
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.
3.2 Monophosphites
3.2.1 Catalysis
As stated above, the first example of the use of phosphite ligands in

rhodium catalyzed hydroformylation of 1 -alkenes was reported by Pruett and
Smith of Union Carbide Corporation. They studied a wide variety of
phosphite and phosphine ligands [7]. The general trend that can be found in
their results is an increasing selectivity for the linear aldehyde when the
electron withdrawing properties of the ligand increase. The donating 4-
methoxyphenyl substituent resulted in a decrease of the linear to branched
ratio, whereas the 4-chloro substituted phenyl phosphite gave a relatively
high 1:b ratio of 13 (see Table 1). The expected higher reaction rates with the
stronger -acceptor phosphite ligands are less obvious from their results. 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.
Figure 2. Structure of bulky phosphites (1, 3) and electron poor phosphite (2)
As one would expect that increasing steric hindrance in the catalytically

active rhodium complex will result in lower reaction rates, the results of Van
Leeuwen and Roobeek seemed contradictory at first. They used the very
bulky tris(ortho tert-butylphenyl)phosphite 1a (Figure 2) as a ligand and
found high reaction rates in the rhodium catalyzed hydroformylation of
otherwise unreactive alkenes like 1,2-and 2,2-dialkyalkenes (see table 1)
[8]. The high reactivity was explained by the exclusive formation of
38 Chapter 3
monoligated rhodium phosphite complexes, which was confirmed by in-situ

IR and NMR studies [9].
The high catalytic activity of the rhodium bulky phosphite system was
also evident in the hydroformylation of 1-alkenes [11,12]. A rhodium
complex containing tris(2- tert-butyl-4-methylphenyl) phosphite (1b) as
ligand gave extremely high rates up to 161,000 mol.mol Rh-1.h-1 in the
hydroformylation of 1-octene with moderate selectivity for the linear
aldehyde.
Table 1. Hydroformylation using rhodium bulky mono phosphite catalystsa

Ligand T, pCO pH2 Alkene Isom. Rate, b mol. l:b
-1 -1
C % (mol Rh )h
c
P(OPh)3 90 3 3 1 octene
- n.d. - 6.1
c
P(OC6H4-p-C1)3 90 3 3 1-octene n.d. - 13.3
c
P(OC6H4-p-OMe)3 90 3 3 1-octene n.d. - 4.9
d
PPh3 80 10 10 1-octene 1.5 2200 2.8
e
1a 90 10 10 1-heptene n.d. 7100 3.3
f
1b 80 10 10 1 o- ctene 13 40,000 1.9
e
2 95 4.8 4.8 1-heptene 15 300 19
2
e
120 4.8 4.8 2 -heptene - 1.9
e
la 80 7 7 2-methyl-1- - 1000 >100
hexene
1 bf 80 10 10 c-hexene - 500 -
f -
1b 70 11 11 styrene - 10,000
e
la 80 7 7 limonene - 1700 >100
g
lb 80 10 10 methyl - 400 -
oleate
a b
Conditions: 0.1-1 mM Rh, L/Rh = 10-20, [alkene] = 0.5-1 M in benzene or toluene.
c d e f g
Initial rate. N.d. = not determined. [7]. [61]. [8]. [9, 1 1, 12]. [20].
f
Next to the high reactivity induced by sterically demanding ligands Van
Leeuwen and Roobeek found remarkable reaction rates using strongly
electron withdrawing ligands, even for the hydroformylation of less reactive
internal alkenes. It was found that the selectivity for the linear product
increased using electron withdrawing ligands. By applying tris(2,2,2-
trifluoroethyl) phosphite (2) they obtained 96% linear aldehyde starting from
1-alkenes and 66% linearity with internal alkenes as substrate (Table 1) [8b].
To obtain a high linearity in the hydroformylation of 1-heptene a high ligand
concentration was required. Probably at low ligand to rhodium ratios the
phosphite ligand is partially replaced by a carbonyl resulting in higher rates

but lower selectivity. The explanation for the high rates and selectivities is
mainly based on electronic factors. The strongly electron withdrawing
phosphite ligand induces fast replacement of a carbonyl ligand by the alkene
substrate, resulting in high reaction rates. Increased isomerization to internal
alkenes is a side effect of the high rates induced by electron-poor rhodium
catalysts.
Zilkowski and Trzeciak have extensively studied the use of phosphite
ligands in the rhodium catalyzed hydroformylation of alkenes [ 16]. As a
consequence of the higher -values compared to triphenylphosphine, an
inactive hydroformylation catalyst, RhHL4, is obtained using triphenyl
phosphite at high ligandhhodium ratio's (L/Rh). This is in contrast to the
PPh3 catalyst; here at high ligand concentrations when all the rhodium is
maximally coordinated by the phosphines, the rate is very low and
eventually becomes independent of L/Rh. Phosphites with moderate cone
angles and higher -values give rise to acceptable 1:b ratios. Good results
were obtained by tri(4-chlorophenyl) phosphite, contrary to tri(2,6-
dimethylphenyl) phosphite, having a cone angle of 190. Unfortunately, the
increase in regioselectivity for the former is accompanied by an additional
increase of isomerization products. The high isomerization rates could be a
consequence of the low pressures used; most experiments were performed at
1 bar, which can easily lead to CO depletion.
One of the commercial applications of bulky phosphites is the production
of 3-methylpentane- 11,5-diol by hydroformylation of 3-methylbut-3-en- 1-01
by Kuraray [17]. Furthermore, they reported the use of bulky phosphite in
the hydroformylation of vinyl acetate and 7-octenal, the latter providing an
intermediate for the preparation of nonanediol. The high reactivity induced
by bulky phosphite ligands has also led to the application of
hydroformylation in the functionalization of natural product derivatives that
are otherwise hardly reactive. Syntheses of important intermediates to fine-
chemicals have been reported by hydroformylation of dihydrofuran [18],
glucal derivatives [19] and methyl oleate [20] (see also chapter 6 for further
details).
Bryant from UCC reported a very elegant application of bulky phosphite
ligands. He utilized the high activity but moderate selectivity of bulky
phosphites as an indicator for ligand depletion during the hydroformylation
reaction using selective bulky diphosphite systems [59]. When the
diphosphite ligand concentration becomes too low, as a consequence of
ligand decomposition, the bulky phosphite ligand 3 will coordinate to
rhodium preventing rhodium plating or cluster formation. Furthermore, the
hydroformylation rate of this bulky phosphite ligand based system is orders
of magnitude higher than that of the diphosphite catalyst. Because of the
40 Chapter 3
lower selectivity and much higher rate, diphosphite depletion will be

monitored as a drop in overall product selectivity and new diphosphite can
be added.
Easy recovery of the rhodium bulky phosphite catalyst could be achieved
by immobilization of the catalyst [21]. A perfectly random copolymer of
styrene and 2,2-bis(4,6-di-tert-butylphenyl)-4-styryl phosphite was used to
form rhodium complexes. The structure of the active catalyst was found to
be dependent on the loading of the phosphite monomer. At very high
phosphite loadings the less active rhodium bis phosphite complexes were
formed whereas low phosphite loadings provided a high activity in the
hydroformylation of cyclooctene, showing rates comparable to the low
molecular weight analogue [21a]. Separation of the catalyst from the
reaction product was facilitated by using a silica-grafted polymer bound
bulky phosphite [21b]. Additionally this resulted in a very high stability of
the immobilized catalyst.
3.2.2 Mechanistic and kinetic studies
The extremely large cone angle of 180 of 1a prevents coordination of a

second bulky phosphite under hydroformylation conditions [10]. The effect
is twofold. First of all the overall steric hindrance at the rhodium metal is
low because three relatively small carbonyl ligands are coordinated next to
the bulky phosphite (structure A, Figure 3). Secondly the rhodium center
containing only one weak phosphite donor and strongly electron
withdrawing carbonyl ligands is electron poor. As a result the carbonyl
ligands are very loosely bound and the fast dissociation of CO (structure B)
and subsequent alkene addition (structure C) results in high reaction rates.
Subsequent hydride migration results in the formation of the linear (D) or
branched (H) rhodium alkyl complex.
The extreme steric bulk of the tris(ortho tert-butylphenyl) phosphite (1a)
ligand not only prevents coordination of two ligands but in fact requires a
high concentration of the ligand (up to 60 mM [9]) to ensure complete
formation of the rhodium phosphite complex. It should be noted that the
complex fomiation is a function of both ligand and rhodium concentration.
Since the rhodium catalyst based on bulky phosphite is very active, low
rhodium concentrations down to 0.1 mM Rh are often employed. Therefore,
a high ligand to rhodium ratio of 50 can be necessary to prevent the
formation of unsubstituted rhodium carbonyl complexes; the latter would
result in low selectivity for the linear aldehyde and high isomerization rates.
Figure 3. Mechanism of rhodium catalyzed hydroformylation using bulky monophosphites
The isomerization reaction is often ignored in catalytic studies but it is

important with respect to the selectivity for the linear aldehyde. Under
reaction conditions the rhodium alkyl complexes D and H (see figure 3) can
give either migratory insertion forming the rhodium acyl complex or give -
hydride elimination. Starting from the primary rhodium alkyl D, -hydride
elimination is not productive as 1-alkene is reformed. The secondary
rhodium alkyl complex H, however, can give both the starting 1-alkene and
the internal 2-alkenes by -hydride elimination. For both complexes the -
hydride elimination is competing with the migration reactions that lead to the
product aldehydes. Since -hydride elimination is faster for the secondary
rhodium alkyl than for the primary rhodium alkyl complex high
isomerization rates will reduce the formation of the branched rhodium acyl
to a larger extent than the linear rhodium acyl. As a consequence the ratio of
the linear to branched aldehyde will increase when the isomerization rate
increases. The total selectivity for the linear aldehyde of the
42 Chapter 3
hydroformylation reaction decreases when the selectivity for isomerized

internal alkenes increases.
Since the mechanism of alkene isomerization involves -hydride
elimination this reaction requires the creation of a vacant site [15].
Therefore, isomerization rates can be suppressed by using low temperatures,
high CO pressures and high ligand concentrations, provided that the ligand is
not very bulky. As stated above, a change in linear to branched ratio can
often be explained, at least partially, by a change in isomerization rate.
Therefore, the reported effects of pressure, temperature and ligand
concentration on the selectivity for the linear aldehyde should be considered
with care.
Kinetic studies showed that. the catalyst based on bulky phosphites 1
differed from systems using phosphines as ligands (see chapter 4) and
resembled the unmodified rhodium carbonyl catalysts (see chapter 2).
,
Cavalieri d Oro et al. showed that for the hydroformylation of propene with
rhodium triphenylphosphine as catalyst the rate of the reaction is determined
by the first steps in the catalytic cycle [3]. The kinetic studies by Van
Leeuwen et al showed completely different behavior of the bulky phosphite
modified rhodium catalyst [11, 12]. Using 1-octene as substrate the reaction
rate has a zeroth order dependency on the alkene concentration. The rate of
the hydroformylation reaction was first order in hydrogen and rhodium
concentration and showed an inverse first order in carbon monoxide
pressure. Increasing the H2 : CO ratio to 7 at 80 C resulted in extremely
high reaction rates of 161,000 mol aldehyde (mol Rh)-1 h-1 [11]. The reaction
even has a runaway character; at low CO pressure the reaction becomes so
fast that CO transfer to the solution becomes rate limiting and the reaction
rate increases further until CO has been depleted. A negative order in the
concentration of one of the reactants has important implications in process
design (see chapter 8).
In situ IR studies showed that using the more reactive 1-octene as
substrate the predominant species under reaction conditions was a rhodium
acyl complex (structure G, figure 3), probably containing one phosphite
ligand and three carbonyl ligands [ 14]. Before addition of the substrate under
reaction conditions the rhodium hydride (A) containing one ligand was
observed. Rapid scan IR shows that immediately after addition of the
substrate a small signal at 1690 cm-1 is observed, while the signals of the
hydride A in the carbonyl region are disappearing (see figure 4). The new
signal at 1690 cm-1 is in the region of the carbonyl vibration of rhodium acyl
complexes. This signal is only visible for a very short time as later it is
concealed by the carbonyl band of the product aldehyde. In the metal-
carbonyl region of the spectrum three new bands are observed during the
reaction (see figure 4). It was concluded that during the reaction the acyl
complex G was the only observable species. When all alkene was consumed
the rhodium hydride A was formed back. Both the spectroscopic and the
kinetic studies show that the rate limiting step for the hydroformylation of 1-
alkenes using the bulky phosphite modified rhodium catalyst is the
hydrogenolysis of the rhodium acyl intermediate.
wavenumber (cm-1)
Figure 4. Metal-CO region of IR spectra recorded with rapid-scan method after addition of 1-
octene to RhH(CO)31b
When less reactive substrates such as internal or 2,2-disubstituted alkenes

are used as the substrate the kinetics of the system resemble those of the
rhodium triphenylphosphine catalyst system. The alkene addition or hydride
migration is rate determining again, as evidenced by kinetic studies [12] and
in-situ spectroscopy [9]. NMR and HR spectroscopic studies using
cyclooctene as substrate revealed that the resting state of the catalytic
reaction was the monoligated hydride HRh(CO)3 1a (structure A, Figure 3)
[9]. The structure of the catalyst is independent of the ligand concentration;
even when a hundred fold excess of ligand is used, only one bulky phosphite
is coordinated to rhodium.
Remarkably, Claver et al showed that in a square planar rhodium
carbonyl chloride complex two bulky phosphite ligands 1b were able to
coordinate in a trans orientation [19b]. This complex was isolated from the
reaction mixture after performing hydroformylation in chlorinated solvent.
The steric hindrance of the bulky ligands is less in the square planar trans
complex than in the trigonal bipyramidal rhodium hydride. In the absence of
CO pressure one of the carbonyl ligands can be replaced by a bulky
44 Chapter 3
phosphite ligand. Probably, the phosphite ligand can exchange with carbon
monoxide under higher CO pressure as is also observed in the formation of
rhodium complexes of bulky diphosphites (see section 3.3.2).
Catalyst and ligand stability are important features in catalysis (for more
details see chapter 9). The sensitivity towards hydrolytic reactions and/or
solvolysis is strongly dependent on the structure of the phosphites. For
instance, UCC reported that the hydrolytic stability of bulky phosphites
could be improved by the use of bulky bisphenols in the ligand structure
(structure 3, figure 2). Ligand hydrolysis and other destructive side reactions
like Michaelis-Arbuzov rearrangement [62] are among the problems that can
be encountered when employing phosphites in catalysis (see figure 5). Many
examples of metal catalyzed Arbuzov-type decomposition reactions have
been reported [48]. The Arbuzov reaction is restricted to alkyl phosphites,
which is probably the reason that almost exclusively aryl phosphite ligands
are used in catalysis.
Figure 5. Classical (A) and metal catalyzed (B) Michaelis-Arbuzov reactions
3. 3 Diphosphites
3.3.1 Catalysis
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 was changing 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 [22]. This initial report was followed

by numerous patents from UCC and triggered a large research effort at
academia and industries.
Figure 6 Bulky diphosphites developed by Bryant at UCC
The change from bulky monophosphite to bulky diphosphite ligands by

using a bisphenol linker resulted, in several cases, in a tremendous increase
of the selectivity for linear aldehyde in the rhodium catalyzed
hydroformylation of 1-alkenes [22, 23]. The reaction rates were 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 figures 6 and
7) [24]. Selectivities higher than 95% were obtained and depending on the
ligand small amounts of the branched aldehyde, isomerized alkenes or both
were observed (see Table 2).
The bridge length of the diphosphites has a large influence on the
selectivity for the linear aldehyde. For aliphatic bridges as in ligands 9 the
optimum selectivity was found for a three-carbon bridge (9b), 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 [25,26]. Although this was not recognized at first, the
46 Chapter 3
a
Table 2. Hydroformylation using rhodium bulky diphosphite catalysts
Ligand T, C p CO/H2 Ratio Alkene Isom. Rate,b mol. l:b

bar CO/H2 % (mol Rh)-. h-
4aC 70 2.5 1.2 1-Butene 2400 50
C
4a 71 6.7 1 :2 730 35
8aC 70 7 1 :2 1480 3.2
C
6 70 7 1 :2 160 6.3
c
7 70 4.3 1:1 Propene 280 1.2
c
6 70 4.3 1:1 20 2.1
5aC 74 4.5 1:1 402 53
9ac 90 7.1 1:1 1-Butene 1620 2.3
c
9b 90 7.1 1:1 1320 3.8
9cC 90 7.1 1:1 1070 2.2
c
10 90 7.1 1:1 3660 2.0
c
11 90 7.1 1:1 1650 9.9
c
10 90 7.1 1:1 2- Butene 1140 0.5
c
11 90 7.1 1:1 2-Butene 65 2.8
12ad 80 20 1:1 1-octene n.d. 11,100 1.6
d
12b 80 20 1:1 1-octene 20 1550 2.2
d
5a 8O 20 1:1 1-octene 18 3600 >100
5bd 80 20 1:1 1-octene 27 6,120 51
d
4b 80 20 1:1 1-octene n.d. 3375 19
8bd 80 20 1:1 1-octene 13 520 1.2
a b c
Conditions: 0.1-1 mM Rh, L/Rh = 10-20, [alkene] = 0.5-1 M in toluene. Initial rate.
d
[22,23]. [24]. N.d. = not detected.
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 [27-29]. The highest selectivities were,
however, achieved using bisphenol bridges.
A crucial feature for obtaining high selectivities for the linear aldehyde
seems to be large steric bulk at the bridging bisphenol like in ligands 4 and 5
[22-24]. Ligand 6 is probably too sterically hindered as can be concluded
from the relatively low hydroformylation rates that are observed. In contrast
ligand 7 is apparently too small to induce the desired high selectivity for the
linear aldehyde. Interestingly, ligand 8 is structurally related to 5 and the
steric hindrance is comparable to that of 4. Nevertheless the selectivity in the
hydroformylation of 1 -octene is very low compared to the successful ligands

4 and 5 and even lower than obtained when using (bulky) monophosphites.
Similar to the ligands 9 containing the aliphatic bridges, the backbone of
ligand 8 is lacking steric bulk. Although ligand 10 seems to have large steric
bulk, the obtained selectivity is low. Probably this ligand cannot coordinate
in a bidentate fashion; the distance between both phosphorus donors is
estimated to be more than 7 . The activity is high, however, which is
reminiscent of bulky monophosphites (see section 3.2.1). The behavior of
ligand 10 as bulky monophosphite is even more evident in the
hydroformylation of the internal substrate 2-butene. Relatively high rates are
obtained albeit with moderate selectivity for the linear aldehyde.
Figure 7. Bulky diphosphite ligands

48 Chapter 3
In contrast to bulky monophosphite modified systems, rhodium

diphosphite catalysts showed very low activity for internal alkenes like
cyclohexene. The required monodentate coordination of the ligand is
prevented by the chelate effect.
Bulky diphosphite based catalysts are active in the hydroformylation of
styrene [24]. The rates of the reaction are generally a factor of ten lower than
those for aliphatic 1-alkenes. Styrene is a substrate that has an intrinsic
preference for the formation of the branched aldehyde, probably caused by
stabilization of the branched rhodium alkyl by the phenyl ring [30]. The high
selectivity for the formation of linear aldehydes of the bulky diphosphite
catalyst was also observed with styrene as a substrate. Typically linear to
branched ratios around 1 are observed while the selectivity for the branched
aldehyde is generally higher than 90% when monodentate ligands are
employed. The selectivity for the linear aldehyde can be enhanced by
influencing the isomerization rate. Secondary rhodium alkyl complexes
are more prone to -hydride elimination than primary rhodium alkyls (see
section 3.2.2). For styrene this effect is even more pronounced because of
the formation of relatively stable branched rhodium styryl complexes. Since
b-hydride elimination can only result in the reformation of the substrate, this
reaction results in the preferential removal of the branched rhodium alkyl
and, therefore, a relatively slow formation of the branched aldehyde.
Lazzaroni et al. came to the same conclusion when they studied the
deuterioforniylation of styrene (see chapter 2) [57]. Employing reaction
conditions that promote high isomerization rates, i.e. high temperature and
low CO pressure, will result in a higher selectivity for the linear aldehyde.
The high hydrogen pressure that was used resulted in the formation of a
substantial amount of the hydrogenation product, ethylbenzene.
3.3.2 Mechanistic and kinetic studies
Van Leeuwen et al. studied the kinetics of the hydroformylation of 1-

octene using the bulky diphosphite 4b [24]. The reaction exhibited a first
order dependency of the alkene concentration. The reaction rate was almost
independent of the hydrogen pressure and showed 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 dOro et al.
for the triphenylphosphine based catalyst (see chapter 4) [3]. 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)3CO) compared to the
HRh(CO)2( diphosphite).
The selectivity to the linear product nonanal was strongly dependent on

the CO pressure (see Table 2). The linear to branched ratio drops from 29 at
Pco = 5 bar to 4 at Pco = 40 bar. Part of this selectivity change can be
ascribed to enhanced isomerization at lower CO pressure (vide supra), but
faster -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.
Rhodium diphosphite catalysts were studied under actual
hydroformylation conditions by the groups of van Leeuwen [24] and
Gladfelter [3 1]. The active catalyst was prepared from Rh(CO)2acac and
excess diphosphite ligand. By monitoring the reaction mixture by NMR, it
was shown that the first intemiediate was a monoligated rhodium acac
carbonyl complex [24]. Only in the absence of an excess of CO the bis-
coordinated Rh(P-P)acac was observed; this complex was also reported as an
intermediate by Gladfelter [31]. An X-ray crystal structure of the Rh(P-
P)acac complex of ligand 4b was reported by van Rooy [24], indicating that
despite the steric bulk of the diphosphite ligand cis-coordination in the Rh(1)
complex was feasible.
Figure 8. Equilibrium between ee and ea coordinating rhodium diphosphite 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 by (high pressure)
IR and NMR data (Table 3). In the carbonyl region of the infrared spectrum
the vibrations of the ee and ea complex can be easily distinguished. The ee
complexes typically show absorptions around 20 15 and 2075 cm-1 [24, 26,
3 1, 32], whereas the ea complexes exhibit carbonyl vibrations around 1990
and 2030 cm-1 [26,32]. The rhodium hydride vibration lies in the same
region as the carbonyl ligands but is often very weak. In fact additional
carbonyl signals in case of mixtures of complexes are often erroneously
assigned to the rhodium hydride stretching frequencies. Van Leeuwen et al.
were able to isolate some complexes as powders and the IR spectra were
measured as nujol mulls (see for IR data Table 3). For complex
50 Chapter 3
HKh(CO)2 (4b) it was found that three absorptions were present in the CO-
stretching region instead of two observed with IR in solution. Upon
measuring DRh(CO)2(4b), only two absorptions remained (2058, 201 2 cm-
1), which were shifted compared to HRh(CO)2(4b) (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
stretch. The rhodium-hydride vibration disappears upon deuteration of the
complex as the rhodium-deuteride vibration is situated in the fingerprint
region. The large frequency shift of the highest energy absorption is
indicative of a trans hydrido-CO relation [46]. In solution IR, the rhodium
hydride vibration and the lowest energy CO vibration overlap, which results
in two absorptions.
Additional information can be obtained from NMR data. Phosphorus
donors coordinated in the equatorial plane have generally small coupling
constants of phosphorus to hydrogen. The coupling constant of a trans
coordinated phosphite shows a large phosphorus to hydrogen coupling
constant of 180-200 Hz. The phosphorus to phosphorus coupling constant is
much larger for the ee (around 250 Hz) than for the ea complex (usually <70
Hz). Some representative NMR and IR data of rhodium-diphosphite
complexes are given in Table 3.
Table 3. NMR and IR data for HRh(P-P)(CO)2 complexesa

31 1 1J 1J 2J 1J
Ligand P H Rh-P Rh-H P-H P -P
V
Rh-CO
4b 160.5 -10.4 237,226 3.5 -19, 70 170 2049

159.8 1966
5a 174.5 -10.8 239 3.5 4 2074
2013
5b 173.0 -10.3 234 3 9
13 167.4 -9.60 238, 154 n.r. 190 58 2035
156.9 1994
14 168.1 -10.11 244,225 <2 <2 247
165.5
a
Data taken from references 24 and 26.
When C2-symmetric ligands are employed both phosphorus donor atoms

become inequivalent no matter if the ee or ea complex is formed. Buisman
prepared complexes of ligands 13 and 14 (figure 9) that exhibited exclusive
ea or ee coordination respectively. These complexes showed fluxional
behavior at room temperature [32]. 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 13 an averaged phosphorus to
hydrogen coupling constant of 80 Hz was observed. At low temperature the
exchange process was halted and the large trans coupling constant of the
phosphorus to the hydride was observed, whereas the cis coupling constant
remained very small and unresolved [32]. By variable temperature NMR
studies the thermodynamic activation parameters could be determined by
comparison of the calculated and experimental NMR spectra (see figure 10).
The H, calculated from the Eyring equation, was 35 kJ.mol and the S
-1
was 6 J.K-1.mol-1. Both H and S were significantly higher for the ee

-1
rhodium complex derived from ligand 14, 62 kJ.mol-1 and 46 J.K .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.
Figure 9. Chiral diphosphites forming exclusively ea ( 1 3 ) or ee (14) rhodium complexes
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 [47] (see figure 11). A simultaneous
bending motion of the hydride and carbonyl ligands takes place in the
hydridorhodium diphosphite dicarbonyl complexes containing ea
coordinating diphosphites (figure 11A). For ee coordinating ligands,
however, a motion of the hydride the equatorial phosphite functions is
responsible for the exchange (figure 11b). The latter process is expected to
be more difficult than the former, which explains the higher fluxionality of
the ea coordinating complexes.
For ligand 4b an X-ray crystal structure of the hydrido rhodium
dicarbonyl complex could be obtained [24]. The P-Rh-P angle in
HRh(CO)2( 4b) is 116 and the complex is indeed a trigonal bipyramid with
the diphosphite in the equatorial plane. The structure is somewhat distorted;
the larger ligands in the equatorial plane bend toward the small hydride. This
explains why the coupling constants between phosphorus and hydrogen are
52 Chapter 3
sometimes larger (up to 70 Hz, see table 3) than might be expected in a

complex showing a pure cis-relationship between hydride and phosphorus
ligand (< 10 Hz). This crystal structure confirms the NMR analysis, but so
far it is the only example of an X-ray crystal structure of a diphosphite
rhodium catalyst.
1
Figure 10. Calculated and observed variable-temperature H (300 MHz) NMR spectra for
HRh(CO)213
Figure 11. A: Equatorial-apical phosphorus exchange. B: Equatorial-equatorial phosphorus

exchange
Summarizing, the relationship between ligand and complex structure and

selectivity of the hydroformylation reaction is not always Straightforward. In
general bisequatorial coordination of the ligand is required for obtaining a
high preference for the formation of linear aldehyde. Bisequatorial
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 one of the
terminal end-groups. Introduction of electron donating and withdrawing
substituents on the dissymmetric bulky bisphenols both induced higher
selectivity for the linear aldehyde. This shows that the effect of electronic
variations of the system on the catalytic performance is not very
straightforward either [24]. When aliphatic bridges are introduced instead of
bisphenol, the obtained 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 [26] (see also chapter 5). The obtained linearity is generally
much lower than for the successful bisphenol bridged bidentate phosphites.
In many instances the formation of inactive dimers next to active,
monomeric catalyst species is observed during catalysis. When weak or
instable ligands are used even larger rhodium carbonyl clusters like
Rh4(CO)12 and Rh6(CO)16 can be observed [49, 50, 51]. The formation of
dimers is often an equilibrium which is reversible. This only leads to a
reduction of the amount of catalyst available and it does not kill the catalyst.
One of the first examples is the fomiation of the so-called orange dimer from
HRh(PPh3)3CO, already reported by Wilkinson [1c] and characterized by
54 Chapter 3
Chan (see figure 12) [58]. They also observed the formation of red dimers
after losing two carbon monoxide molecules.
Figure 12. Rhodium dimer characterized by Chan (L=PPh3)
Since the hydroformylation reaction for most substrates shows a first

order dependence in 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 (see chapter 4). Dimer formation was mainly reported for
phosphine ligands [1c, 49, 52], but similar dimeric rhodium complexes from
monophosphites [53] and diphosphites [26,32] have been reported. The
orange side product obtained from HRh l4 (CO)2 was characterized as the
carbonyl bridged dimeric rhodium species Rh2 (14) 2(CO)2 [32].
Figure 13. Diphosphite ligand studied by Gladfelter
Gladfelter et al. reported the presence of dimeric complexes during the

rhodium catalyzed hydroformylation of 1-octene using ligand 5a [3 1]. The
appearance of the 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 using ligand 15 they even isolated a tetranuclear rhodium complex,
illustrating the importance of steric hindrance in the ligand structure [54].
In addition to dimerization and clustering to multinuclear rhodium
carbonyl complexes, active catalyst can also be removed from the reaction
mixture by orthometallation. Several reports have appeared on
orthometallation of rhodium triphenyl phosphite complexes in literature but
structural proof for the orthometallated species has not always been
convincing. Parshall reported the ortho-deuterium incorporation in phosphite
ligands [56]. In agreement with this, Coolen showed that the hydrogenation
catalyst HRh[P(OPh)3]4 provided ortho-deuterated product when stirred
under deuterium atmosphere [55]. Both researchers suggested that the
deuterium exchange proceeded via orthometallation. The intermediate
orthometallated complex was characterized by 31P NMR [55].
3.4 Hydroformylation of internal alkenes.
3.4.1 Hydroformylation of less reactive internal and functionalized

alkenes.
The great advantage of phosphite ligands compared to phosphines is the

higher reactivity of the corresponding rhodium catalysts. As a result, a much
larger variety of substrates can be converted at acceptable rates. An
additional advantage of the phosphite modified rhodium catalysts is that the
hydroformy lation can be performed under very mild conditions. Because of
these mild reaction conditions many functionalized alkenes can be used as
substrate; the functional group coinpatability is excellent, which renders the
hydroformylation reaction also suitable for natural product synthesis [ 18-20,
38-41] (see chapter 6).
Bulky monophosphite ligands proved to be very useful for the
functionalization of very unreactive substrates. Already in their first study
van Leeuwen and Roobeek obtained relatively high rates for the
hydroformylation of substrates such as cyclohexene and limonene. [8]. Van
Rooy performed a systematic study to the rhodium catalyzed
hydroformylation of substituted alkenes and compared the reaction rates
with the triphenylphosphine system [42]. The bulky monophosphite derived
catalyst was up to two orders of magnitude faster and gave acceptable rates
using substrates for which the Wilkinson hydrofomylation catalyst gave
hardly any activity.
Fatty acid derivatives containing an additional functional group are
important high-value-added oleochemicals. Therefore, hydroformylation of
unsaturated fatty acids and their esters can be a very profitable process.
Since the substrates are again internal alkenes and generally will contain
impurities that are known to retard the catalytic reaction, a very active
catalyst is required. Muilwijk studied the hydroformylation of pure and
technical grade methyl oleate using the rhodium bulky phosphite catalyst
[20]. The double bond of methyl oleate of 99% purity was hydroformylated
56 Chapter 3
without any problem at relatively high rates. Next to formation of the desired
product methyl formylstearate (as a mixture of isomers) the rapid formation
of the trans isomer of methyl oleate, methyl elaidate, was observed. As
expected the hydroformylation of the trans isomer was slower than that of
the cis isomer (see figure 14). Comparison of the bulky phosphite modified
system with the rhodium triphenylphosphine catalyst revealed that the
reaction rates of the former were much higher, as expected for internal
alkene substrates.
time/h
Figure 14. Hydroformylation of pure methyl oleate (3.64 mmol), Pinitially = 20 bar, CO/H2 =
1:1, T=100 C, Rh=4.10-3 mmol, Ligand:Rh =25:1, methyl oleate,
methyl elaidate,
methyl fomylstearate, 0 methyl formylstearate (isomers)
Technical grade methyl oleate, however, contains about 14% of the diene
methyl linoleate that easily isomerizes to a conjugated diene. These
conjugated dienes strongly retard the hydrofomylation reaction, probably by
the formation of stable -allyl type rhodium complexes [43]. The initial rate
of the hydroformylation reaction is much lower than for the pure methyl
oleate substrate that does not contain significant amounts of dienes. Only
after the methyl linoleate has been converted to other hydrogenation and
hydroformylation products the rate of the reaction increases. Again the
reaction mixture shows isomerization of methyl oleate to the trans product
methyl elaidate during the hydroformylation process (see figure 15). Since
hydrogenation of the diene methyl linoleate to mono alkenes was required
prior to hydroformylation, the reaction rate increased significantly when
higher hydrogen pressures were applied.
Figure 15. Hydroformylation of technical grade methyl oleate (3.64 mmol), Pinitially = 20 bar,
CO/H2 = I: 1, T=100 "C, Rh = 4.10-3 mmol, Ligand : Rh =25: 1, A methyl oleate, methyl
elaidate, methyl linoleate, o methyl linoleate (isomers), methyl formylstearate, methyl
formylstearate (isomers)
3.4.2 Formation of linear aldehydes starting from internal alkenes.
The selective formation of linear aldehydes starting from internal alkenes

is still one of the greater challenges in hydroformylation chemistry. To
achieve this goal a catalyst is required that obviously has a high
isomerization activity since the thermodynamically less stable terminal
alkene has to be formed prior to hydroformylation. Furthermore, the
selectivity for the formation of the linear rhodium alkyl complex has to be
high. Both isomerization and hydroformylation products are formed from the
same rhodium alkyl intermediate. Therefore, a very active catalysts is
required to obtain addition of the unreactive internal alkene and subsequent
hydride migration. Moreover the branched rhodium alkyl complex must
have a large preference for -hydride elimination compared to migration to a
carbonyl ligand, which would result in the formation of branched aldehydes.
If isomerization rates are very high compared to hydroformylation rates, the
catalyst can maintain the small concentration of 1-alkenes in the alkene
mixture governed by thermodynamics. Once the terminal alkene has been
58 Chapter 3
formed it will react with the rhodium complex because the alkene addition is
much faster than for internal alkenes. If the catalyst is very selective and
gives a high linear to branched ratio in combination with fast isomerization,
only the linear rhodium alkyl intermediate will show preferential
carbonylation. As a consequence the linear aldehyde will be the main
product, even using internal alkenes as starting material. So far only few
catalysts are known that show these properties.
The strongly electron withdrawing monophosphite 2 was one of the first
examples that provided rhodium catalyst that gave good selectivity for the
formation of linear aldehyde starting from internal alkenes (see table 1) [8].
The high -value of the phosphite induces enhanced selectivity for the
formation of the rhodium alkyl complex and high isomerization rates,
resulting in a high overall selectivity for the linear aldehyde.
For lower alkenes such as 2-butenes UCC has achieved high contents of
linear products (see Table 2). Bryant reported 74% selectivity for the
formation of linear pentanal by hydroformylation of 2-butene using the
rhodium bulky diphosphite catalyst [22, 23].
Figure 16. DSM and Mitsubishi diphosphite ligands
Du Pont and DSM have patented the use of ligand 16 (Figure 16) for the
hydroformylation of methyl 3-pentenoate to linear product [33]. Instead of t-
butyl groups on the bridge they use electron-withdrawing ester groups, while
the remaining substituents are monophenols [34, 35], rather than diols [36,
37] or bisphenols. The necessary isomerization to the linear alkene prior to
hydroformylation reaction is probably promoted by the electron withdrawing
ester groups. They were able to obtain high selectivity for the linear
aldehyde, providing a useful intermediate to nylon-6 feedstocks. Comparable
selectivity up to 97% was reported for the hydroformylation of 2-hexene.
The hydroformylation of butadiene to the linear 1,6-hexanedial proved to
proceed less satisfactory; the main products were pentanal and 3-pentenal
and only small amounts of the desired 1,6-hexanedial were formed.
Similar results were obtained using the UCC systems; in several patents
the reported yields for dialdehyde in the hydroformylation of butadiene were
relatively low [60]. Moderate yields for the linear aldehyde were generally
obtained when methyl 3-pentenoate was hydroformylated with the bulky
diphosphite modified UCC catalyst [60]
The efficacy of monophenols containing bulky substituents is described
in the patents from Mitsubishi Chemical Corporation [34,35] (17, Figure 16).
They report high yields of aldehyde (> 90%) for the hydroformylation of 1-
alkenes with high linearities. The 1:b ratios are generally above twenty using
this ligand.
3.5 Calixarene based phosphites.
Recently several reports appeared of phosphite ligands based on well-

defined supramolecular backbones like calixarenes and diphenylglycoluril.
Although ligands based on these types of structures have been known for
quite some time very few examples of actual applications in catalysis have
appeared. One of the first examples of an application of a calix[4]arene
based diphosphite stems from BASF [44]. They used a calix[4]arene
substituted at the 1 and 3 position with bulky phosphite substituents in the
rhodium catalyzed hydroformylation of 1 -alkenes (Figure 17). Although they
were able to obtain high linear to branched ratios the catalytic reaction gave
large amounts of isomerized internal alkenes.
Figure 17. Calix[4]arene based diphosphite by BASF [44] and a monophosphite employed in
hydroformylation [45]
Parlevliet has shown that calix[4]arene based monophosphites can exist

in different conformations [45]. By using the different conformations as
ligands in the rhodium catalyzed hydroformylation of 1-octene he showed
that the exact conformation influenced the performance of the catalyst. Some
of the conformations behaved more like triphenyl phosphite, whereas others
showed catalytic results like bulky monophosphites, giving high rates with
moderate selectivity for the linear aldehyde.
60 Chapter 3
References
1 Young, J. F.; Osbom, J. A.; Jardine, F. A.; Wilkinson, G. J. Chem. Soc., Chem. Comm.
1965, 131. (b) Evans, D.; Osborn, J. A.; Wilkinson, G . J. Chem. Soc. (A) , 1968, 3133. (c)
Evans, D.; Yagupsky, G.; Wilkinson, G. J. Chem. Soc. A 1968, 2660. (d) Brown, C. K.;
Wilkinson, G. J. Chem. Soc. A 1970, 2753.
2 Heck, R. F; Breslow, D. S. J. Am. Chem. Soc. 1961, 83, 4023.
3 Cavalieri d'Oro, P.; Raimondt, L.; Pagani, G.; Montrasi, G.; Gregorio, 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 d'Oro:
P.; Pagani, G.; Andreetta, A. Chim. Ind. 1980, 62(5), 389; Cavalieri d'Oro, P.; Raimondi,
L.; Pagani, G.; Montrasi, G.; Gregorio, G.; Andreetta, A. Chim. Ind. 1980, 62, 572
4 Van Leeuwen, P. W. N. M.; Van Koten, G. Catalysis. An Integrated approach to
Homogeneous, Heterogeneous and Industrial Catalysis, Moulijn, J. A.; Van Leeuwen, P.
W. N. M.; Van Santen, R. A. Eds., Elsevier: Amsterdam. London, New York, Tokyo,
2nd ed. 1995, pp 199-222.
5 Unruh, J. D.; Christenson, J. R. J. Mol. Catal. 1982, 14, 19.
6 L. A. van der Veen, M. D. K. Boele, F. R. Bregman, P. C. J. Kamer, P. W. N. M. van
Leeuwen, K. Goubitz, J. Fraanje, H. Schenk and C. Bo, J. Am. Chem. Soc., 1998, 120,
11616.
7 Pruett, R. L.; Smith, J. A. J. Org. Chem. 1969, 34, 327. (b) Pruett, R. L.; Smith, J. A. S.
African Pat. 6804937, 1968 (to Union Carbide Coorporation); Chem. Abstr. 1969, 71,
90819.
8 (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.
Abstr-. 1984, 101, 191 142.
9 Jongsma, T.; Challa, G.; van Leeuwen, P. W. N. M. J. Organonmetal. Chem. 1991, 421,
121.
10 Tolman. C. A. Chem. Rev. 1977, 77, 313.
11 Van Rooy, A.; Orij, E. N., Kamer, P. C. J.; Van den Aardweg, F.; Van Leeuwen, P. W.
N. M. J. Chem. Soc., Chem. Commun. 1991, 1096.
12 Van Rooy, A.; Orij, E. N.; Kanier, P. C. J.; Van Leeuwen, P. W. N. M. Organometallics,
1995, 14, 34.
13 Brown, J. M.; Kent, A. G. J. Chem. Soc. Perkin Trans. II 1987, 1597.
14 Van Rooy, A. PhD thesis, University of Amsterdam 1995.
15 Collman, J. P.; Hegedus, L. S.; Norton, J. R.: Finke, R. G. Principles and Applications of
Organotransition Metal Chemistry University Science Books: Mill Valley, CA 2nd ed.
1987.
16 Trzeciak, A. M.; Zilkowski, J. J.; Aygen, S.; Van Eldik, R. J. Mol. Cutal. 1986, 34, 337.
(b) Trzeciak, A. M.; Zilkowski, J. J. Trans. Met. Chem. 1987, 12 , 408. (c) Trzeciak, A.
M.; Zilkowski, J. J. Inorg. Chim. Acta Lett. 1982, 64, L267. (d) Trzeciak, A. M.;
Zilkowski, J. J. J. Mol. Catal 1986,34, 213. (e) Janecko, H.; Trzeciak, A. M.;
Zilkowski, J. J. J. Mol. Catal. 1984, 26, 355. (f) Trzeciak, A. M.; Zilkowski, J. J. J.
Mol. Cutal. 1987, 43, 13.
17 Yoshinura, N.; Tokito, Y. Eur. Pat. 223 103, 1987 (to Kuraray); Chem. Abstr 1987, 107,
154896. (b) Omatsu, T. Eur. Pat. 0 303 060, 1989 (to Kuraray); Chem. Abstr. 1989, 111,
38870.
18 Polo, A.; Real, J.; Claver, C.; Castilln, S.; Bayn, J. C. J. Chem. Soc. Chem. Commnun.
1990, 600.
19 (a) Polo, A.; Femandez, E.; Claver, C.; Castilln, S. J. Chem. Soc. Chem. Commun.
1992, 639. (b) Fernandez, E.; Claver, C.; Castilln, S.; Polo, A.; Piniella, J. F.; Alvarez-
Larena, A. Organometallics 1998, 17, 2857.
20 Muilwijk, K. F.; Kamer, P. C. J.; van Leeuwen, P. W. N. M. J. Am. Oil Chem. Soc. 1997,
74, 223.
21 Jongsma, T.; Fossen, M.; Challa, G.; van Leeuwen, P. W. N. M. J. Mol. Catal. 1993, 83,
17. (b) Jongsma, T.; van Aert, H; Fossen, M.; Challa, G.; van Leeuwen, P. W. N. M. J.
Mol. Catal. 1993,83, 37.
22 Billig, E.; Abatjoglou, A. G.;. Bryant, D. R. U.S. Pat. 4,668,651; Eur. Pat. Appl.
213,639, 1987 (to Union Carbide); Chem. Abstr. 1987, 107, 7392.
23 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. Abstr. 1988, 109, 233177.
24 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, 535.
25 Babin, J. E.; Whiteker, G. T. W.O. 93 03839, US. Pat. 91 1 518, 1992 (to Union Carbide
Corporation); Chem. Abstr. 1993, 119, 159872.
26 Buisman, G. J. H.; Vos, E.; Kamer, P. C. J.; Van Leeuwen, P. W. N. M. J. Chem. Soc.,
Dalton Trans., 1995, 409.
27 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.
28 Kranenburg, M.; van der Burgt, Y. E. M.; Kamer, P. C. J.; van Leeuwen, P. W. N. M.
Organometallics. 1995, 14, 3081.
29 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.
30 Tolman, C. A.; Faller, J. W. Homogeneous Catalysisis with Metal Phosphine Complexes,
Pignolet, L. H. Ed., Plenum Press: New York and London, 1983; Chapter 2, pp 88-89.
31 Moasser, B.; Gladfelter, W. L.; Roe, C. D. Organometallics 1995, 14, 3832.
32 Buisman, G. J. H.; van der Veen, L. A.; Kanier, P. C. J.; van Leeuwen, P. W. N. M.
33 Burke, P. M.; Garner, J. M.; Tam, W.; Kreutzer, K. A.; Teunissen, A. J. J. M. WO
97133854, 1997, (to DSM/Du Pont); Chem. Abstr. 1997, 127, 294939.
33 Sato, K.; Karawagi, J.; Tanihari, Y. Jpn. Kokai Tokkyo Koho JP 07,278,040 (to
Mitsubishi); Chem. Abstr. 1996, 124, 231851.
35 Sato, K.; Kawaragi, Y.; Takai, M.; Ookoshi, T. US Pat. 5,235,113, EP 518241 (to
Mitsubishi); Chem. Abstr. 1993, 118, 191183.
36 Rper, M.; Lorz, P. M.; Koeffer, D. Ger. Offen. DE 4,204,808 (to BASF); Chem. Abstr.
1994,120, 133862.
37 Lorz, P. M.; Bertleff, W.; Roper, M.; Koeffer, D. EP 472,071 (to BASF); Chem. Abstr.
1992,117,34513.
38 Cuny, G. D.; Buchwald, S. L. J. Am. Chem. Soc. 1993, 115, 2066.
39 Johnson, J. R.; Cuny, G. D.; Buchwald, S. L. Angew. Chem. Int. Ed. Engl. 1995, 34,
1760.
40 Van den Hoven, B. G.; Alper, H. J. Org. Chem. 1999, 64, 3964.
41 Van den Hoven, B. G.; Alper, H. J. Org. Chem. 1999, 64, 9640.
42 van Rooy, A.; de Bruijn, J. N. H.; Roobeek, K. F.; Kanier, P. C. J.; van Leeuwen , P. W.
N. M. J. Organomet. Chem. 1995, 507, 69.
43 (a) Van Leeuwen, P. W. N. M.; Roobeek, C. F. Brit. Patent Appl. 33,554, 1981 (to Shell
Resareh); Chem. Abstr. 1982, 96, 6174. (b) Van Leeuwen. P. W. N. M.; Roobeek, C. F.
J. Mol. Catal., 1985, 31, 345.
62 Chapter 3
44 Paciello, R.; Rper, M.; Kneuper, H.-J.; Langguth, E.; Peter, M. DE 4321194, 1995,
(BASF AG); Chem. Abstr. 1995, 122, 160937. (b) Paciello, R.; Siggel, L.;Rper, M.
Angew. Chem. Int. Ed. Engl. 1999, 38, 1920. (c) Paciello, R.; Siggel, L.; Walker, N.;
Rper, M. J. Mol. Catal. A 1999, 143, 85.
45 Parlevliet, F. J.; Kiener, C.; Fraanje, J.; Goubitz, K.; Lutz, M.; Spek, A. L.; Kamer, P. C.
J.; van Leeuwen, P. W. N. M. J. Chem. Soc. Dalton Trans. 2000, 1113.
46 Vaska, L. J. Am. Chem. Soc. 1966, 88 , 4100.
47 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
48 Brill, T. B.; Landon, S. J. Chem. Rev. 1984, 84,577. (b) Werener, H.; Feser, R. Z Anorg.
Allg. Chem. 1979, 458, 30 1.
49 Whyman, R., J. Chem. Soc.. Chem. Commun., 1970, 230-231.
50 Booth, B. L.; Else, M. J.; Fields, R.; Haszeldine, R. N., J. Organomet. Chem., 1971, 27,
119-131.
51 Cattermole, P. E.; Osborne, A. G., Inorg. Synth., 1977, 17, 115.
52 Castellanos-Pez, A.; Castilln, S.; Claver, C.; van Leeuwen, P. W. N. M.; de Lange, W.
G. J. Organometallics, 1998, 17, 2543.
53 James, B. R.; Mahajan, D.; Rettig, S.J .; Williams, G. M. Organometallics, 1983, 2, 1452
54 Moasser, B.; Gladfelter, W. L. Inorg. Chim. Acta 1996, 242, 125.
55 Coolen, H. K. A. C.; van Leeuwen, P. W. N. M.; Nolte, R. J. M. J. Organornet. Chem.
1995, 496, 159.
56 Parshall, G. W.; Knoth, W. H.; Schunn, R. A. J. Am. Chem. Soc. 1969, 91,4990.
57 Lazzaroni, R.; Rafaelli, A.; Settambolo, R.; Bertozzi, S.; Vitulli, G. J. Mol. Catal., 1989,
50, 1. (b) Lazzaroni, R.; Settambolo, R.; Rafaelli, A.; Pucci, S.; Vitulli, G. J. Organomet.
Chem. 1988, 339, 357.
58 Chan, A. S. C.; Shieh, H. S.; Hill, J. R. J. Chem. Soc., Chem. Commun. 1983, 688.
59 Bryant, D. R; Leung, T. W. US patent 5,741,943, 1998, (to UCC); Chem. Abstr. 1998,
128, 283881.
60 Packett, D. L.; Briggs, J. R.; Bryant, D. R.; Phillips, A. G. U.S Patent 5,886,237 (to
UCC); Chem. Abstr. 1998, 128, 14354.
61 Buhling, A.; Kamer, P. C. J.; van Leeuwen, P. W. N. M. J. Mol. Catal. A: Chemical,
1995, 98, 69.
62 Bhattacharya, A. K.; Thyagarjan, G. Chem. Rev. 1981, 81, 415.
Chapter 4
Phosphines as ligands
Bite angle effects for diphosphines
Piet W. N. M. van Leeuwen, Charles P. Casey, Gregory T. Whiteker#

Institute of Molecular Chemistry, University ofofAmsterdam. Nieuwe Achtergracht 166, 1018
WV, Amsterdam, the Netherlands: Department of Chemistry, University of Wisconsin-Madison,
1 101 University Avenue, Madison. WI 53706, U.S.A.; # Union Carbide Corporation, P.O. Box
8361, South Charleston WV 25303, U.S.A.
4.1 Monophosphines as ligands
4.1.1 Introduction
The most famous rhodium catalyst precursor for hydroformylation is

undoubtedly RhH(PPh3)3CO. It was first reported by Vaska in 1963 [1], but
its activity for hydroformylation was discovered by Wilkinson and co-
workers a few years later [2, 3]. Initially the activity was assigned to chloro-
rhodium species [4]. The catalysis and the organometallic chemistry
involved were studied in great detail, such as kinetics, substrate activity,
metal-ligand equilibria, equilibria between monomeric and dimeric
complexes, reaction mechanisms, spectroscopic properties, etc. Soon
thereafter the first commercial applications were reported [Chapter 1]. The
chemistry reported in the late sixties and early seventies is still relevant for
the hydroformylation studies of today. The chemistry and catalysis of
rhodium-PPh3 has been reviewed many times [3, 5]. By way of introduction
for the catalysts based on diphosphines a few salient features of the PPh3-
based catalysts will be summarized. These will include mechanistic studies,
in situ spectroscopic studies, kinetics, and phosphine ligand effects. Special
consideration will be given to the rate-determining step under common
circumstances (vide infra); we will see that it is not the oxidative addition of
hydrogen as most textbooks erroneously say [6].
63
64 Chapter 4
4.1.2 The mechanism
In Figure 1 the well-known mechanism, as first proposed by Heck [7],

has been depicted. It corresponds to Wilkinsons so-called dissociative
mechanism [2]. The associative mechanism involving 20-electron
intermediates will not be considered. For PPh3 as the ligand a common
starting complex is RhH(PPh3)3CO, complex 1, which under 1 bar of carbon
monoxide forms the complexes 2ee and 2ae, containing the phosphine
ligands in equatorial positions (denoted ee throughout the scheme) or one in
an apical position and the other ligand in an equatorial position (complexes
denoted ae). Brown [8] found a preference for the ee isomer in the hydride.
Dissociation of either equatorial L or equatorial CO from 1 or 2 leads to the
square-planar intermediates 3c and 3t, which have phosphines in cis or trans
configurations respectively. Preferential dissociation of equatorial ligands
from trigonal bipyramids is normally observed. Complexes 3 associate with
ethene to give complexes 4, again in two
Figure I. Simplified mechanism for hydrofomylation of ethene (L = PPh3)

4. Phosphines as ligands 65
isomeric forms ae and ee, having a hydride in an apical position and ethene
coordinating in the equatorial plane. While four-coordinate species rhodium
3c (cis) have never been observed, the analogous four-coordinate complex
for 3t trans-Rh(PCy3)2(CO)H has been isolated and structurally
characterized [56]. In addition, magnetization transfer experiments indicate
that both 1 and 2 are in equilibrium with free PPh3 [8]. These results are
highly suggestive of the formation of the four-coordinate complex 3 by
phosphine dissociation from 1 and 2. The complexes never observed
experimentally are shown in brackets in Figure 1.
Although the complexes shown in Figure 1 contain at least two
coordinated PPh3 ligands, a large body of indirect evidence suggests the
influence of equilibria involving catalytically active species containing a
single coordinated phosphine ligand. Such monophosphine species were
initially invoked to explain the positive effect on hydroformylation
regioselectivity that is typically seen with increasing P:Rh ratios (see 4.1.6).
The 31P NMR magnetization transfer experiments described by Brown [8]
also indicate that PPh3 dissociation from the RhL2 complex 2 occurs, albeit
at a significantly slower rate than the corresponding PPh3 dissociation from
tris-phosphine complex, 1. In addition, IR spectra have revealed the
presence of Rh[P(OAr)3](CO)3H with a very bulky coordinated phosphite
ligand [48].
It has not been established experimentally whether alkene complexation
is reversible or not; in the scheme we have drawn all steps but the
hydrogenolysis at the end as reversible. Brown observed that complex 1
catalyzes the isomerization of cis-1,2-dideuteriostyrene under nitrogen [8].
Although under these conditions the isomerization of this labeled alkene is
slower than the rate of hydroformylation using 1, this result suggests that
alkene coordination and insertion into the Rh-H bond can be reversible
processes. Complex 4 undergoes a migratory insertion to give square-planar
alkyl complexes 5c and 5t, which are respectively cis or trans. In the
absence of CO, the 16-electron complexes of this type can be isolated as
mixtures of cis and trans isomers for aryl groups or alkyl groups not
containing -hydrogen atoms, [9]. Complex 5 can undergo -hydride
elimination, thus leading to isomerization when higher alkenes are used, or it
can react with CO to form trigonal bipyramidal complexes 6. Thus, under
low pressure of CO more isomerization may be expected. At low
temperatures (< 70 C) and a sufficiently high pressure of CO (> 10 bar) the
insertion reaction is usually irreversible and thus at this point also the
regioselectivity of hydroformylation of -alkenes is determined.
Complexes 6 undergo the second migratory insertion in this scheme to
form the acyl complexes 7. Complexes 7 can react either with CO to give the
saturated acyl intermediates 8, which have been observed spectroscopically,
66 Chapter 4
or with H2 to give the aldehyde product and the unsaturated intermediates 3.

The reaction with H2 involves presumably oxidative addition and reductive
elimination, but for rhodium no trivalent intermediates have been observed.
For iridium the trivalent intermediate acyl dihydrides have been observed
[ 10]. The Rh-acyl intermediates 8 have also been observed by Brown [8] and
due to the influence of the more bulky acyl group, as compared to the
hydride atom in 2e and 2a, isomer 8ae is the most abundant species.
At low hydrogen and high rhodium concentrations, formation of
dirhodium species such as 9 becomes significant. Since several of the studies
of the Wilkinson group were carried out under such conditions, formation of
dirhodium species was very relevant to their chemistry. Not only did they
report the formation of the orange dirhodium species of type 9, they also
observed the formation of red dirhodium species containing two carbon
monoxide molecules less than 9. A dirhodium complex 9 containing three
molecules of PPh3 was fully characterized by Chan [11]. Regeneration of
rhodium hydrides from dormant rhodium species formed by impurities, is
another cause for the positive rate response to raising the H2 pressure. Note
that in Figure 1 not all reagents have been depicted in the scheme for the
sake of simplicity.
Usually, the full mechanistic scheme will be more complicated than that
shown in Figure 1 :
1. Species containing one to four molecules of L will occur, depending
on the concentrations of L and CO, and the nature of L (vide infra),
2. When alkenes such as propene and higher are used the regio-isomers
of linear and branched alkyl and acyl groups should be included,
3. Alkyl species 5 containing 2-butyl species or higher may lead to
internal alkenes via -hydride elimination.
The scheme shown in Figure 1 is typical of NMR studies using PPh3,
carried out at room temperature, ~1 bar of CO (and H2), and a complex
concentration of 10-1 to 10-2 M [8]. It is also characteristic of catalytic
conditions, having as typical values 70-120 C, 10-30 bar, rhodium
concentration of =10-3 M, and a hundred-fold excess of aryl phosphines. In
situ IR spectroscopy will show whether RhL1 or RhL 2 or RhL3 (L = PPh3)
species are present. NMR studies at high concentrations and low pressures
and temperatures will prove the formation of dimers 9. Quantitative data for
monomer-dimer equilibria have been obtained for bidentate phosphines [ 12].
4.1.3 Ligand effects
Steric and electronic properties of a ligand can drastically influence the

rate and selectivity of the hydroformylation reaction (see Chapter 3,
Phosphites as ligands). A systematic study though is absent for several
reasons. For monodentate ligands a systematic study is impeded by the

variety of complexes that may be involved at the various stages of the
catalytic cycle. Secondly, RhH(PPh3)3CO was used as the precursor in
several studies, while aiming at the study of the effect of ligands added. Use
of this procedure without removal of the liberated PPh3 leads to possible
mixed phosphine complexes, which complicate interpretations. Thirdly,
complexes of type 1 were often used without the addition of free ligand,
which is needed to ensure a high equilibrium concentration of 1 or 2. Other
complications arise from the use of different experimental conditions. As
we shall see in section 4.1.5. changes in temperature, concentrations and
pressure can significantly influence rate and regioselectivity. For these
reasons, studies which are performed under conditions outside those
typically used for hydroformylation (10-30 bar, temperature 70-120 C, [Rh]
= 10-3 M, [alkene] = 0.1-2 M, [L] depending on complex stability) can
result in conclusions which are not particularly relevant. Impurities in one of
the feedstocks often led to oxidation of the phosphine ligand. Another source
of erroneous results is the use of rhodium chloride (and the like) as the
precursor, while even in the late sixties it was known that this gave only
partial conversion to hydrido species [2]. When discussing selectivities
isomerization has often not been taken into account. In addition, many
interesting effects have been published only in patents and these results are
less accessible in common literature searches.
Thus, we are facing an impossible task to summarize the ligand effect
for monodentate phosphines (and phosphorus ligands) and to give credit to
the numerous contributions. For the sake of didactics we will present a few
rules of thumb, which at this point will not be fully exemplified by literature
data, but which will be supported when diphosphines are discussed in
section 4.2.
Electronic effects. Electron donating ligands, such as alkylphosphines,
lead to slower catalysts and as result they require higher temperatures [13].
In recent years it has been shown that triethylphosphine gives turnover
numbers of 700 mol.mol-1.h-1 for 1-hexene and very high turnover numbers
for ethene at 120 C and 40 bar. As a secondary solvent-dependent reaction
hydrogenation to alcohols takes place [ 13c]. Arylphosphines containing
electron-withdrawing substituents give a faster catalytic reaction than
triphenylphosphine [ 14-16]. Phosphites can give faster catalysts than
phosphines [17, 18], but this is certainly not true for all cases. A recent study
in supercritical CO2 also shows that electron withdrawing arylphosphines
form more active catalysts [19], but the low concentrations of RhHL3(CO),
in the absence of extra L added, and the relatively high concentration of CO
in the supercritical medium may lead to excessive dissociation of the
phosphine ligand. Dibenzophospholes are more electron-withdrawing than
68 Chapter 4
diphenylphosphino groups and without exception the former lead to faster

catalysts [S, 20, 43b, 59]. In general, ligand effect studies are hard to
interpret, because for a particular ligand the rates may differ by at least an
order of magnitude, depending on the concentration of rhodium, ligand and
co.
Most of the authors cited agree on the explanation; electron-withdrawing
ligands lead to a decrease of the back-donation to carbon monoxide and thus
to a weaker binding of the carbonyls. This will affect the formation of
species 3 and 7, such that their rate of formation, or their equilibrium
concentration increases. Alkene complexation, giving complexes 4, may also
be accelerated or become more favored thermodynamically. Migratory
insertions are not particularly sensitive to electronic properties of the ligand
[21 and references therein], but it is important to remember that oxidative
addition will slow down when electron-withdrawing ligands are used.
Steric effects. More sterically demanding ligands will favor the formation
of species containing a low number of ligands L and therefore more CO
ligands. A high proportion of CO ligands also leads to electron poor rhodium
species and thus to enhanced dissociation of CO. For phosphites this effect
has been clearly observed [22] as was discussed in Chapter 3.
4.1.4 In situ studies
Moser has studied monophosphine systems of substituted arylphosphines

by in situ IR reflectance spectroscopy [14]. During 1-hexene
hydroformylation, at 70 C, 11-24 bar of syn gas, [Rh]=10 mM, L/Rh = 6,
the species observed in most instances was RhH(CO)2L2 , 2. The three bands
observed were assigned to two carbonyl stretching frequencies and a
rhodium-hydride stretching frequency. Claver and Van Leeuwen studied the
system containing PPh3 as the ligand during the hydroformylation of 1-
octene (60-100 C, 5-20 bar, [Rh] = 1 mM, PPh3/Rh = 5) using in situ IR
transmission spectroscopy [23]. They observed four bands (2042, 1992,
1981, 1947 cm-1), which were assigned to the mixture of RhH(CO)2(PPh3)2
isomers 2ae and ee as reported in the NMR studies by Brown [8], without
any other species being present. The rhodium-hydride bands are very weak
and usually only the four carbonyl stretches are observed. The spectral
assignments are derived from the spectra of the diphosphine complexes
(section 4.2). The concentration ranges applied in the IR and catalysis
studies are similar. Thus, both studies infer that under the conditions of
practical catalysis species 2ae and 2ee are the resting state of the catalyst.
NMR spectroscopy is less useful for in situ studies as the concentrations
of metal and ligand are often two orders of magnitude higher, which
influences the positions of the equilibria considerably. Secondly, in most
NMR devices replenishment of the consumed gases would be slow

compared to the rate of reaction, which would lead to the formation of
species that differ from the ones under actual catalytic conditions.
Using the same ligand to metal ratios (4-9:1) as in the IR studies above
[23] and pressures up to 4 bar but significantly higher ligand concentrations
of 10 mM, Oswald [13b] found in NMR studies that the predominant
species is now the tris-triphenylphosphine complex 1. Dissolving 1 in
toluene at 21 mM at ambient conditions (1 bar of CO, ratio 3:1) gives a
mixture of free PPh3, 1 and 2, the latter being the major species as was
reported by Brown [8]. His studies show that ligand exchange is much faster
than the catalytic processes. These examples clearly illustrate the importance
reporting concentrations rather than metal to ligand ratios, a common but
erroneous way of describing catalyst composition.
Upon addition of alkene to solutions containing 1 and 2 under a CO
atmosphere, Brown observed the formation of five-coordinate acyl
complexes, which were characterized by NMR spectroscopy.
4.1.5 Kinetics
In spite of the industrial importance of the rhodium-PPh3 catalyst, very

few data have been published on the kinetics of the hydroformylation
reaction and the data known are often contradictory. Early mechanistic work
by Wilkinson (at < 1 bar) demonstrated the inhibitory effect of increased CO
and PPh3 concentrations on 1 -hexene hydroformylation rate [2c]. Increased
Rh and alkene concentrations were observed to lead to higher catalytic rates.
Wilkinson also observed an accelerated hydroformy lation rate caused by
increasing H2 pressure, which is most likely, an artifact of inactive
dirhodium species formed under non-standard, low-pressure conditions.
We will argue that under standard catalytic conditions the reaction is first
order in the concentration of alkene, first order in the rhodium concentration,
zero order in hydrogen, and the reaction shows a negative order in ligand
concentration (phosphine or carbon monoxide, or both). The most detailed,
early study is the report by Cavalieri dOro et al.[24] from Montedison. They
found the following expression for the rate:
v=k[C3H6] [PPh3]-0.7[CO]-0.1 [Rh] [H2]0

0.6 1
(conditions 90-1 10 C, p(CO) = 1-25 bar, p(H2) = 1-45 bar, [PPh3] = 0.05-5 M,
[Rh] = (0.5-7). 10-3 M, PPh3/Rh = 300:1 to 7:1, [propene]t=0 = 2-7 M)
They reported an overall activation energy of 84 kJ.mol-1 for the process.
The important features of this kinetic study are the zero order dependence in
dihydrogen, the negative order in PPh3 ligand (and CO), and the positive
70 Chapter 4
order in alkene concentration. Under standard conditions, i.e. the

industrial operating conditions, chosen by many workers in the field, this
rate equation seems a good starting point for arylphosphine modified
catalysts. These standard conditions# (T = 70-120 C, CO = 5-25 bar, H 2 =
5-25 bar, Rh 1 mM, alkene = 0.1 - 2 M) are assumed in later sections
unless otherwise specified.
The observed order in propene concentration is less than one, which
might point to saturation kinetics. Indeed, high concentrations were used, but
perhaps the non-ideal behavior of propene (critical temperature 94 C) plays
a role in this. Under similar conditions for 1-hexene and 1 -octene a neat first
order behavior in alkene has been observed using Rh-PPh3 catalysts [18,
25a].
At high PPh3 concentrations, where the catalyst resting state is
(PPh3)3Rh(CO)H, phosphine dissociation must occur to form the
coordinatively unsaturated intermediates 3c and 3t. This dissociation is
suppressed by increased PPh3 concentration, which serves to reduce the
concentration of active Rh species in the catalytic cycle.
At lower PPh3 concentrations where the predominant resting state
observed by in situ studies is (PPh3)2Rh(CO)2H, species 3c and 3t are formed
by CO dissociation, which is likewise inhibited by increased CO
concentration. Consistent with this mechanism is the recent determination
that dissociation/association of CO is reversible and faster than
hydroformylation for arylphosphines (see 4.2). An inverse order in CO
pressure was reported by Moser [14]. Strohmeier [25a] also noted that the
rate dropped by 30 % when the total pressure was raised from 10 to 15 bar,
as did Wilkinson [2a] when the pressure was raised from 50 to 80 bar. Both
observations are in accord with a negative order in CO pressure and a zero
order dependency on H2 pressure. At very low pressures of CO or H2 (a few
tenths of a bar at 25 C) the reaction will slow down as a sufficient
concentration of the reagents is controlled by mass transfer limitations.
# Hydroformylation reactions are extremely sensitive to experimental conditions. As detailed

in the following sections, various equilibria exist which influence catalytic rates. These
equilibria are controlled by concentrations of Rh, CO, H2, alkene and added phosphine
ligand. For a given catalyst, hydroformylation rate and regioselectivity are, therefore,
often dependent upon the concentrations of these reactants. In addition, the concentrations
of gaseous reactants can become controlled by mass transfer limitations and result in
increased amounts of hydrogenation and isomerization products. Variations in added
phosphine concentration can also dramatically impact catalyst performance. Under typical
industrial conditions, additional phosphine ligand is added, however some studies utilize a
low P/Rh ratio of 3, especially when chelating diphosphines are used. Unfortunately,
experimental conditions reported in the literature for Rh-phosphine catalysts vary widely
and make comparisons difficult. For this reason, we will define a window of standard
conditions that cover the ranges typically utilized in industrial applications.
Under standard conditions, we propose that the best starting point for
the kinetics is an equation of the type:
A [alkene][Rh]
Rate(type - I ) =
B +[L ]
(the constants A, B do not refer to specific rate constants;

[L] is proportional to [PPh3] and [CO])
Two possible scenarios exist which are consistent with this rate equation.
Rate determining alkene coordination by 3c and 3t followed by rapid alkene
insertion into the Rh-H bond is one possibility. An equally valid alternative
explanation for the observed kinetics involves rate determining migratory
insertion of alkene into Rh-H preceded by fast, reversible alkene
coordination. In both cases, the concentrations of coordinatively unsaturated
3c and 3t are influenced by PPh3 and CO concentrations.
The kinetics at moderate PPh3 concentrations are in agreement with a
catalyst resting state of RhH(PPh3)2(CO)2 and a transition state of
composition RhH(PPh3)2(CO)(alkene). At very high PPh3 concentrations, the
likely resting state is RhH(PPh3),(CO) and a transition state of composition
RhH(PPh3)2(CO)(alkene) is again likely. The set of rate-limiting reactions
are dissociation of CO (or PPh3), complexation of alkene, and migratory
insertion. Dissociation/association of CO is reversible and faster than
hydroformylation for arylphosphines (see 4.2). Complexation of alkene is
most likely reversible, although there are no experimental data on this
process. Theoretical studies [26] indicate that complexation of CO or ethene
to species 3 (Figure 1) has a low barrier. For migratory insertion of ethene
into the metal-hydride an early transition state was found, involving a re-
organization of the complex. This means that steric hindrance will play a
crucial role and especially the rotation of the alkene from the in-plane
coordination to a perpendicular coordination mode contributes to the barrier.
There is a strong body of evidence in the literature in disagreement with
the kinetics presented above [2-4, 27, 28]. Ever since the first statements of
the Wilkinson group [2a, not repeated though in 2c!] the oxidative addition
of hydrogen has been considered as the rate-determining step [5, 6]. The first
suggestion was based on an analogy with the hydrogenation reaction
employing Wilkinsons catalyst RhC1(PPh3)3. In the textbooks [6] and
reviews [5] oxidative addition of dihydrogen is still mentioned as the most
likely rate-determining step. The theoretical work cited shows the highest
barriers for the migration reactions [21, 26], but yet supports the common
point of view!
72 Chapter 4
The scheme shown in Figure 1 can explain why in many instances a

different rate equation has been obtained, having a positive order in
dihydrogen concentration. Without experimental effort we cannot re-
interpret data of specific publications and we will confine ourselves to
general statements. Clearly, at low pressures a different rate equation should
be obtained, even though rhodium and ligand concentrations are within our
standard regime. At low dihydrogen pressures, high rhodium
concentrations, and low temperatures di-rhodium species of the type 9 will
form, which will lead to a positive order in the pressure of dihydrogen due to
the H2 requirement to cleave these di-rhodium species and re-enter the
catalytic cycle [12, 27b].
These basic findings were reported by Wilkinson, who observed at
pressures below 1 bar the presence of dimers, the dependency of the rate on
hydrogen pressure, and yet the positive order in alkene concentration. If the
oxidative addition were rate-determining, the rate of reaction would be zero
order in alkene (unless rapid pre-equilibration would occur all the way from
1 through 8).
As has been discussed in Chapters 2 and 3, for carbonyl systems and
bulky phosphite systems, the resting state of these catalysts (for 1-alkenes) is
the acylrhodium species of type 8 and thus, the reaction with dihydrogen is
rate limiting. At low Rh concentration, using RhH(PPh3)3CO as the
precursor without addition of additional PPh3, substantial dissociation will
occur, giving rise to the kinetics of mono-phosphine or phosphine-free
catalysts, i.e. a type II rate equation. Under these conditions, indeed the
acylrhodium complex 8 is the resting state and oxidative addition is rate-
determining [29]. We would like to stress that Type-II kinetics is the
exception rather than the rule.
Rate(type - II) = C[H2][Rh]

D + [CO]
4.1.6 Regioselectivity
The regioselectivity of the catalysts based on PPh3 has been extensively

studied (see reviews [5]). The regioselectivity of 1 -alkene hydroformylation
varies from 70 to 92 % for the linear aldehyde. The highest selectivity is
obtained at high concentrations of PPh3 [5, 25], or even liquid PPh 3 [2, 30],
and low pressures of CO. Isomerization should be considered as well, as this
forms an escape route for the branched alkyl intermediate leading to
internal alkene isomers instead of branched aldehyde. Since the activity of
most phosphine catalysts for internal alkenes is low, the apparent linearity of
the aldehyde product is high, but the overall selectivity to linear aldehyde
may be low. This will be discussed further in 4.2.
Clearly, the number of phosphines coordinated to rhodium, along with
the stereochemistry at Rh, determines the regioselectivity. At high PPh3
concentrations, the resting state of the catalyst is 1 (Figure 1, 2), which
undergoes PPh3 dissociation to form selectively 4-coordinate intermediate
3t. At lower PPh3 concentrations, the resting state is an equilibrating mixture
of 2ae and 2ee which undergo either CO dissociation to form 3c and 3t or
PPh3 dissociation to form 10c and 10t. According to NMR studies,
conversion between axial and equatorial phosphines in 2ae and 2ee is many
orders of magnitude faster than the rate of hydroformylation [8]. The rate of
hydroformylation is only two orders of magnitude slower than that of carbon
monoxide exchange of species of type 2 [31]. The relative concentrations of
these four-coordinate diphosphine (3) and monophosphine (10)
intermediates are controlled by the PPh3 and CO concentrations. Wilkinson
[2] and Andreetta [24b] suggested that species 3, formed at high PPh3
Figure 2. Intermediates leading to linear and branched products

74 Chapter 4
concentration, lead to higher linear aldehyde selectivity (1:b = 20: 1], and that
species 10, containing only one phosphine, lead to a lower selectivity for
linear aldehyde (1:b = 4: 1) (see Figure 2).
Dissociation of PPh3 from 1 gives 3t, dissociation of CO from 2ee also
gives 3t, while dissociation of CO from 2ae will give 3c. Dissociation of
PPh3 from 2 gives the isomers 10t and 10c. The use of diphosphines such as
dppe and dppp gives modest linear-branched ratios and their putative
intermediates 3 must have cis structure 3c. Therefore, 3c is not a likely
intermediate for the selective hydroformylation. For linear-branched ratios of
4:1 we don't have to invoke mono-phosphine species of type 10, as already
structure 3c suffices for the production of moderate I:b ratios (3:1).
Hydroformylation regioselectivity can be established by either four-
coordinate or five-coordinate intermediates. In one possible mechanism, the
regioselectivity is controlled by the relative concentrations of four-
coordinate intermediates, 3c, 3t, 10c and 10t, which undergo irreversible
formation of alkene hydride complexes, 4. Accordingly, this mechanism
requires that interconversion of isomeric pairs 4ee/4ae and 4e/4a be slow
relative to irreversible migratory insertion of alkene.
Alternatively, the regioselectivity can be determined during migratory
insertion of alkene into the Rh-H bond in five-coordinate intermediates, 4.
The concentrations of these five-coordinate alkene hydride complexes are
controlled by the relative concentrations of PPh3 and CO. In addition, these
stereoisomers (4ee/4ae, 4e/4a), which differ in the nature of the apical
ligand, presumably undergo rapid interconversion via a pseudorotation
mechanism, as has been observed for 2ee/2ae [8].
In both scenarios, the dependence of I:b on PPh3 concentration is
attributed to changes in the extent of phosphine ligation, and the
stereochemistry at Rh is also proposed to influence regioselectivity. Studies
using diphosphine ligands with well-defined coordination number and
geometry (section 4.2) have provided much insight into these factors that
control hydro formylation regioselectivity.
Substituted alkenes will be discussed in Chapters 5 and 6. From a
mechanistic viewpoint the results obtained with substituted 1 -alkenes are
worth mentioning as they stress the importance of steric interactions. Simple
1-alkenes carrying methyl substituents at the carbon atoms 3 or 4 give
progressively more linear product for a higher degree of substitution closer
to the alkene bond, without losing much activity as is shown in Table 1[ 18].
The table shows that in most instances the rate of hydroformylation to the
terminal product is retained and that a remarkably high linearity can be
obtained when steric hindrance is provided at carbon-3. Substitution at
carbon-2 under these conditions gives rates of only 100 mol.mol-1.h-1, and
Table 1. Hydroformylation of methyl-substituted 1 -alkenes
Alkene Rate, mol.mol-1.h-1 Linearity, %

1 -pentene 11,300 78.4
4-Me-1-pentene 9,300 78.0
4,4-Me2- 1 -pentene 5,300 85.0
3-Me-1-pentene 9,600 91.0
3,3-Me2- 1-pentene 7,600 99.0
Conditions: 90 C, p(CO/H2) = 20 bar, [Rh]=0.5 mM, [PPh3]=5 mM, [alkene]=0.5 M,
initial rates at <20% conversion, no isomerization was observed [18].
only linear product is formed. Internal alkenes also show rates that are one to
two orders of magnitude lower than those of 1-alkenes [2].
Cis-2-alkenes react a few times faster than trans-2-alkenes. Steric
influences, both from the complex part and the alkene, play a dominant role
in determining the fate of hydroformylation of alkenes not containing
functional groups. For alkenes such as styrene, 3,3,3-trifluoropropene,
acrylates, and vinyl esters, the electronic properties of the alkene determine
to a large extent the regioselectivity.
4.1.7 Conclusion
Monodentate ligands, be they phosphines, phosphites, or phosphor-

amidites, are very versatile ligands for rhodium catalyzed hydroformylation.
Their performance as ligands though is strongly obscured by the equilibria in
which they are involved. Literature articles, reviews and textbooks disagree
on the rate-determining step, whether this is the insertion (or complexation)
of the alkene (type-I) or the oxidative addition of dihydrogen (type-11). We
have argued that under standard conditions type-I kinetics prevail. We
have mentioned a number of pitfalls that might obscure the kinetic analysis
and we will return to that in Chapters 8 and 9. The differences in the rates of
the respective steps are small, and it is not surprising that different rate
equations have been postulated due to differences in experimental
conditions. Steps eligible as rate-determining are complexation of the
alkene, hydride migration, migration of the alkyl group forming the acyl
group, and oxidative addition of dihydrogen. We propose that grosso modo
in phosphine catalyst systems migratory insertion of the alkene into Rh-H is
the rate-determining step under standard, industrial process conditions. The
concentrations of the reagents, metal and ligands may disturb the equilibria,
causing deviations from our rule of thumb. As we will see in section 4.2 the
uncertainties arising from the use of monodentate ligands can be reduced by
the use of bidentate phosphines.
76 Chapter 4
4.2 Diphosphines as ligands
4.2.1 Introduction
The key role of intermediate species 3 containing two triphenyl-

phosphines for obtaining a high selectivity to linear products has been
recognized since 1970 [2]. The way to guarantee that these intermediates are
formed in excess with respect to the less selective monophosphine
intermediates 10, was the use of a high excess of PPh3. The first
commercializations (1 974, Celanese, 1976, UCC) utilize this finding and
linearities greater than 90 % can be achieved. It would seem obvious, in
hindsight, that the use of suitable diphosphines might greatly facilitate the
formation of relatively stable L2Rh complexes. Other benefits later derived
from bidentate diphosphines include control of metal geometry and efficient
asymmetric induction. Favorable effects of diphosphines in catalysis were
reported as early as 1966 by Iwamoto [32], who found that dppp was the best
ligand for the codimerization of butadiene and ethene using iron catalysts.
During the seventies the advantages of the use of bidentates were especially
exploited in the field of asymmetric hydrogenation using chiral diphosphines
such as DIOP and DIPAMP [33], but it would take many years before
suitable bidentates were evaluated for selective hydroformylation to linear
aldehydes.
Sanger [34] studied the effect that the addition of common diphosphines
exerts on catalyst 1 (2 mM in benzene) at room temperature and 1 bar. While
the amount of ligand added strongly influenced the rate, the 1:b ratio of the
aldehyde product ranged from 3.2 to 3.9 for all diphosphines (dppe, dppp,
dppb, DIOP). With the exception of dppb, which forms also dimeric species,
it is expected that these diphosphines will form chelated intermediates of the
structures 2ae and 3c, due to the small bite angle of the phosphines. The
behavior in such mixtures of ligands (PPh3 and diphosphine) remains hard to
disentangle.
Pittman [37b] also studied the effect of added diphosphines (dppe, dppp,
and dppb) to catalyst 1 (10 mM) using higher temperatures and pressures
(60-120 C, 7 bar) on the hydroformylation of 1-pentene. He found low I:b
ratios of only 1. He concluded that apparently a cis coordination as in 3c
gives low I:b ratios. Thus, under these standard conditions the I:b ratio is
much lowcr than that found by Sanger at ambient conditions. The presence
of PPh3 in both sets of experiments would make an explanation speculative.
Later studies have shown that dppp and other ligands containing a
propane bridge smoothly form complexes 11a [18, 35]. It was noted by
Brown that dppe formed less clean solutions. The small bite angle of 84 is
apparently not attractive for the formation of such complexes. Bayn and
coworkers observed that constrained I ,2-diphosphines form dimeric
complexes in addition to the expected monomeric ones [36].
Studies using dppe or dppp under standard conditions, in the absence
of PPh3, have not been published in the seventies, but it would seem likely
that several industries have conducted research in this field. As described in
4.1.1, two explanations for the low regioselectivities observed with such
diphosphines can be brought forward. Resting state 11a can lead to 3c by
dissociation of CO. Since the small bite angle imposed by chelates such as
dppe and dppp cannot accommodate the trans orientation in 3t, the resultant
cis complex 3c leads to low regioselectivity. Alternatively, 3c can reversibly
add alkene to form the five-coordinate alkene hydride complex, which
contains an aical-equatorial diphosphine chelate. This ae isomer,
analogously to 4ae, leads to low regioselectivity. In addition to these two
possibilities, 11a may also undergo dissociation of one phosphine donor
giving 12t, the analog of unselective 10t. The kinetics would be different for
the two possibilities, but unfortunately these have not been reported.
Figure 3. Reaction scheme cis-bidentates
In the late seventies enhanced rates for platinum hydroformylation

catalysts were reported for diphosphines containing a bridge that was longer
than the common two or three carbon atoms [38]. The selectivities to linear
product are very high using these platinum catalysts (>>95%). One rhodium
system has been reported giving linearities > 99% using dppe as the ligand
[39], but this uses 1,3-butadiene as the substrate and most likely the
mechanism involves allylic species and does not concern our discussion of
the mechanism in Figures 1-3.
DIOP was first used as a bidentate ligand added to 1 as early as 1973 by
Consiglio while aiming at asymmetric catalysis. For 1 -alkenes they observed
high I:b ratios of 13 at 25 C and 1 bar [37a].
In 1981 Hughes and Unruh (Celanese) were the first to publish on
diphosphines, which unlike the more common dppe and dppp led to higher
1:b ratios under standard conditions (see Figure 4) [40] (100 C, 8 bar, [Rh] =
2 mM). High I:b (5-8) ratios were obtained in systems consisting of 1 and
added DIOP or trans-dppm-cyb, both having rigid trans conformations. Low
1:b ratios ( 1-3) were obtained for systems containing trans-dppm-cyh, cis-
78 Chapter 4
dppm-cyb, and non-rigid diphosphines dppe, dppp, and dppb. According to a

later report [ 15c] the use of trans-2,3-dppm-nor (Figure 4) (containing
trifluoromethyl substituents) also gave a high ratio (1:b=7) at a high rate.
Steric bulk and the chelate effect are brought forward as an explanation.
Pre-empting our later discussion, we might propose that ligands with small
chelation angles such as dppe and dppp give intermediates 2c which lead to
low I:b ratios, while ligands with wider angles behave differently.
Figure 4. Diphosphine ligands used by Unruh
In the following we will present the results obtained with ferrocene based
ligands and a variety of bidentate phosphine ligands containing bite angles
larger than 100. Since the catalysis for these ligands has been carried out at
somewhat different conditions we will present them separately and discuss
the results in three parts. Finally we will speculate about a general view in an
attempt to find a global explanation.
4.2.2 Ferrocene based diphosphine ligands
Unruh and Christenson [15a,b] studied the use of ferrocene diphosphine

ligands in 1982. Their study was interesting in that the results were different
from the Wilkinson work in several ways. We will quote their own
interpretations, but we cannot avoid adding more recent viewpoints that may
be relevant. Dppf is a flexible ligand and we now know from crystal
structures that the phosphorus-metal-phosphorus angle in complexes of dppf
can vary from 92 up to 120, with a strong preference for values around
972 [46]. Rotation of the cyclopentadienyl fragments with respect to one
another should help to accommodate a variety of structures. From this point
of view, dppf is an interesting choice. In addition to dppf they studied three
electronically modified ferrocenyl-based ligands, each containing electron
withdrawing groups, see Table 2 and Figure 5.
Figure 5. Unruhs substituted dppf ligands

Table 2. Hydroformylation using substituted dppf ligands
Ar = i-value 1:b ratio Relative rate 2-hexene, %

Ph 4.3 5.4 7.2 3.1
4-ClC6H4 5.6 6.8 9.3 5.0
3-FC6H4 6.0 8.0 13.7 5.0
4-CF3C6H4 6.4 11.4 13.8 6.4
Conditions: 110 C, p(CO/H2) = 7.9 bar, [Rh] = 0.6 mM, ligand/Rh = 3, for ligand
structure see Figure 5. i-value, see Chapter 1.
The table shows that with increasing accepting ability, i.e. increasing -
value (the authors used Hammett constants), the rate of the reaction
increases and so does the 1:b ratio. For phosphites a similar trend had been
observed, see Chapter 3. The rate increase can be easily understood, as
dissociation of CO will be more facile when -back-donation decreases.
Assuming that steric properties are the same coordination of alkene may also
be enhanced by a more electrophilic rhodium center. Rates of migration
reactions are usually not affected by electronic changes according to MO
studies [21], thus we observe an overall rate increase with electron
withdrawing phosphines. Steric properties of the ligands are the same,
except that the electronic changes do have an influence on the mode of
coordination as we shall see later, but this was not yet known at the time.
Substituting trifluoromethyl groups for hydrogens in other arylphosphines
(trans-2,3-dppm-nor) also has a favorable effect on rate and selectivity
[15c].
The I:b ratio of the product aldehyde increases even though electron-
withdrawing phosphines give rise to increasing amounts of alkene
isomerization. The total preference for linear alkyl formation still increases
when the 2-hexene is included in the total amount of intermediary 2-
hexylrhodium that must have been formed. Thus, we can say that the
intrinsic preference for 1 -alkyl formation increases with increasing -values
of the ligand. Figure 6 explains the influence of 2-hexene formation on the
observed 1:b ratio in the aldehyde product in a system in which the internal
alkene is not converted into (branched) aldehyde. The higher I:b ratio for
more positively charged metal centers was supposed to originate from a
higher preference for 1 -alkyl formation for such centers. An increasing rate
of isomerization is a general observation, also for phosphite calalysts. Note
that none of the species in Figure 6 has been observed experimentally.
The authors note that under their conditions, involving relatively high
temperatures and low pressures, many steps of the cycle have similar rates,
as the kinetics are rather complex. The best results were obtained when an
excess of phosphine was employed. A ligand to metal ratio of 1.5 was
preferred, as at higher ratios the selectivity did not improve further.
80 Chapter 4
Figure 6 Influence of 2-hexene formation on observed 1:b ratio
This led Unruh and Christenson to the conclusion that the most selective
species was one containing more than two phosphorus ligands, while the
current view was, Wilkinsons view, that complexes containing two
phosphines were the ones leading to high 1:b ratios of 20. They propose that
the most selective species contains three phosphorus atoms, one dppf as a
bidentate on one metal and one dppf bridging between two metals. The two
metals are to function independently in the catalysis.
Complexes of this type have indeed been identified [41] in 31P NMR
studies. A whole range of diphosphines was studied including those of
Figure 4, Figure 5, and the common diphosphines dppe and dppp. The study
involved the addition of various amounts of the diphosphine to complex 1.
The spectra were all interpreted in terms of the species shown in Figure 7.
13 14 15
Figure 7. Hughess phosphine complexes 13, 14, and 15
Unfortunately no 1H NMR data for the hydride signals were reported.

The Jp-p coupling constants of the 31P atoms in the equatorial positions,
2
typically ranging from 100 to 130 Hz, clearly show that in most cases tris-
phosphine complexes are present having all phosphorus atoms in the
equatorial plane. In a few instances the coupling constants are smaller (50 -
100 Hz) indicating that the phosphines occupy in part an apical position,
probably involving a fast exchange. These results support the catalytic
results suggesting that three phosphine ligands are coordinated to the catalyst
in its resting state. However, it should be borne in mind that the
concentrations for the catalysis experiments and the NMR experiments differ
considerably. In catalysis the free ligand concentration is at most 1.2 mM
([Rh] = 0.6 mM) and p(CO) = 4 bar, while in the NMR experiments the
concentrations are one order of magnitude, maybe two, higher (27 mM for
[Rh] and 130 mM for phosphine in one example) and no CO is added. Thus
one might wonder whether indeed tris-phosphine species are present under
the reaction conditions of the catalytic experiments.
Re-examination of the dppf-Rh system has shown that the coordination
behavior under syn-gas is different [42]. Both NMR spectra and in situ IR
([Rh] = 8 mM, [dppf] = 17 mM, p(CO) = 20 bar, 40 C) spectra show that
the only species under these conditions is RhH(dppf)(CO)2 being a mixture
of rapidly equilibrating 2ae (approximately 80 %) and 2ee (20 %). The 1:b
Figure 8. Complexes for dppf and 16
ratio for 1-octene hydroformylation is 5.4, in accord with the results by

Unruh. Interestingly, replacement of two phenyl groups by more bulky 1-
naphtyl groups, rendering a S,S-chiral diphosphine 16 (Figure 8), gives a
much higher proportion of a structure of type 2ee, viz. 85 %, but the linearity
only slightly improves as the I:b ratio is 7.3. Thus, by increasing the steric
bulk at phosphorus, and perhaps by introducing the C2 chirality, the
ferrocene ligand can adopt a larger (averaged) bite angle. Ferrocene remains
an interesting backbone and perhaps still better results may be achieved.
82 Chapter 4
4.2.3 BISBI ligands and the natural bite angle
Figure 9. BISBI, 17, and its complexes 18, and 19, and ligands 20 and 21
In 1987 Devon and co-workers from Texas Eastman reported on a new

diphosphine ligand, BISBI 17 (see Figure 9), that shows a very high
regioselectivity for the formation of linear aldehydes [43a]. Propene was
hydroformylated using BISBI under standard conditions in a vapor
stripped reactor. The aldehyde product shows a high linearity (1:b = 30). For
comparison, other ligands known hitherto gave much lower linearities as
shown in Table 3, taken from their work.
Table 3. Hydroformylation results using BISBI and other ligands [43a]
Ligand Rate (mol.mol-1.h-1) Ratio 1:b

BISBI 3650 25.1
trans-dppm-cyb 3200 4.4
trans-2,3 -dppm-nor 2550 4.3
dPPf 3800 3.6
DIOP 3250 4.0
dPPP 600 0.8
PPh3a 6000 2.4
Conditions: syn gas pressure 16 bar, [Rh] = 1.5 mM, Iigand/Rh = 2.4, 95-125 C, solvent
2,2,4-trimethyl pentane- 1,3-diolmonoisobutyrate (Texanol), p(propene) 5 bar.
a
[Rh] = 0.7 mM, ligand/Rh = 124.
More recently improved variations were described in a later patent [43b].

Instead of diphenylphosphino groups dibenzophosphole groups were used,
which often leads to an increase in rate and selectivity. In addition, BISBI
was substituted with tertiary butyl groups to increase the solubility of the
ligand. Rates were in the same order of magnihide as those mentioned in
Table 3. It was also mentioned that the rate of reaction was accelerated by
high propene pressures, hinting at the common rate equation for
hydroformylation under standard conditions. Under optimized conditions
(i.a. low CO pressure) 1:b ratios as high as 288 were obtained for propene.
Examination of molecular models of metal complexes of BISBI indicates
that the chelate bite angle is much greater than 90, the most common bite
angle for bidentate ligands used in coordination chemistry and

organometallic chemistry [44]. The work quoted so far, and the results
shown in Table 3 led to the speculation that ligands with wider bite angles
may perhaps have a fortuitous effect on the 1:b ratio in the hydroformylation
of 1-alkenes. However, at this time few chelating diphosphines had been
proven by X-ray crystallography to be capable of adopting large bite angles.
To identify ligands with wide bite angles Casey and Whiteker [45]
introduced the concept of the "natural bite angle", which was estimated
using molecular mechanics. It is defined as the preferred chelation angle of
the bidentate ligand, as determined only by ligand backbone constraints and
not by metal valence angles. The neglect of the electronic preferences of the
metal for particular P-M-P angles facilitates the calculation, as the force
fields for the organic part provide a high degree of accuracy. The natural bite
angle is calculated by minimizing the strain energy of the diphosphine-M
fragment with a P-M-P bending force constant of 0 keal.mol-1.rad-2, for
certain M-P distances. In addition, by fixing the bite angle at various angles
by using a large bending force constant and then calculating the strain
energy, potential diagrams can be constructed, which allow an estimation of
the flexibility of the ligand. The outcome will be different for different
programs used, even if the same program and ligand parameters were the
same, since dihedral force constants and stretching modes involving the
metal may be treated differently. Comparisons with crystal structures are
useful, but one should not expect the same outcome [46].
The two wide bite angle ligands studied in the publication of 1992 were
BISBI and trans-bisdiphenylphosphinocyclopropane (trans-dppm-cyp, 20).
Using these ligands the complexes 18 and 19 were synthesized and
structurally characterized by X-ray crystallography. This study
demonstrated for the first time the unequivocal bis-equatorial coordination of
a bidentate diphosphine. The measured bite angle for BISBI in 18 turned out
to be 124, compared to the calculated natural bite angle in this study of
112. When complexes 18 are subjected to 1 bar of CO the ligand PPh3 is
replaced by CO and 19 is formed. NMR and IR spectroscopy proved that the
structure of complexes 19 indeed contained the two phosphorus atoms in the
equatorial plane and hydrogen in one of the apical positions.
Yamamoto and Casey separately reported that 2,5-dppm-nor 21 (Figure
9) which has a wide natural bite angle (126) [47a,b], gave an unexpectedly
low 1:b ratio of only 2.6, which was attributed to its inability to form stable
chelates. Indeed, no monomeric metal complexes of 21 have been reported.
Molecular mechanics calculations show that the Rh 21 chelate has 12 kcal
more strain than the free ligand; in contrast, BISBI and DPPE chelates are
less than 3 kcal more strained than the free ligands. In designing new wide
84 Chapter 4
bite angle ligands, it is suggested that the extra strain imposed by chelation
be calculated to determine whether the new ligand will from a stable chelate.
Hydroformylation of 1-hexene under mild conditions (34 C, 6 bar of syn
gas, 4mM solution of 19 )gave a 1:b ratio of 66.5 for BISBI and for trans-
dppm-cyp it was 12. DIOP and dppe afforded values of 8.5 and 2.1. No
isomerization was observed. It was concluded that apparently a bis-
equatorial coordination of a diphosphine lead to high 1:b ratios, although the
reason for this was not understood in detail.
Figure 10. Chart showing bite angles and selectivities
The correlation between natural bite angle and 1:b ratio is very interesting
and was an important finding for future work. Thus, complexes 2ee (Figure
1) lead to higher 1:b values than complexes having structure 2ae, or
alternatively ligands with larger bite angles lead to more effective steric bulk
of the diphosphine and a relative destabilization of branched alkyl
intermediates.
Before developing detailed explanations of the control of regiochemistry
based on steric or electronic differences between bis-equatorial and
equatorial-apical diphosphine complexes it was considered crucial to test
whether alkene insertion to produce 1 -alkyl and 2-alkyl rhodium species was
irreversible [47c]. For instance, previous work on the deuterioformylation of
cyclohexene using bulky phosphite had shown that H-D exchange was at
least ten times faster than formylation (40-80 C, < 10 bar) [48], but for 1-
alkenes insertion is irreversible, provided that the CO concentration is
maintained at a sufficiently high level [49]. Similar results have been
reported for unmodified carbonyl catalysts (see Chapter 2). In the alternative
case of fast alkene exchange, the regioselectivity would depend upon the
equilibrium of 1- and 2-alkylrhodium species and their respective rates of
CO insertion. Deuterioformylation of 1-hexene at 6 bar and relatively low
temperature (32 C) has shown that insertion of the 1-alkene is an
irreversible process as very little deuterium was incorporated in recovered
alkenes. The small amounts of secondary alkyl chains that are formed lead
mainly to (deuterated) 1- or 2-hexene and only 25 % proceeds to branched
4. Phosphines asligands 85
aldehyde. Little deuteration a to the carbonyl is observed indicating that

alkene dissociation occurs after -hydrogen elimination. These results
establish that the regiochemistry of aldehyde formation is set largely by the
rhodium hydride addition to coordinated 1 -alkene. Plausible steric
explanations were considered and a whole range of intermediates for
diequatorial and equatorial-apical resting states were discussed, but
molecular mechanics calculations failed to support such steric explanations.
We will return to this in section 4.2.5.
Since steric arguments failed to give a satisfactory explanation for the
much higher 1:b aldehyde ratios observed from diequatorial chelates like
BISBI compared with apical-equatorial chelates like dppe, it was considered
whether the electronic differences between diequatorial chelates and apicai-
equatorial chelates are responsible for the differences in regioselectivity.
Because the electronic interaction between two apical ligands or between
two equatorial ligands is stronger than the interaction between apical and
equatorial ligands, electronic differences can be expected between
diequatorial chelated complexes and apical-equatorial chelated complexes.
For example, back bonding from rhodium to the alkene ligand in the
equatorial plane would be expected to be stronger for the BISBI complex
with two strong donor phosphines in the equatorial plane than for the dppe
complex with only a single donor phosphine in the equatorial plane. Also,
the apical hydride of the BISBI complex is trans to a CO ligand and would
be expected to be more acidic than the hydride of the dppe complex which is
trans to a phosphine. Previously, Unruh (see 4.2.2) reported favorable effects
of CF3 substitution at the aryl rings on phosphorus.
Figure 11. Ligands having electron withdrawing substituents
To investigate electronic effects of equatorial and apical phosphines,

analogs of chelating diphosphines of BISBH, trans -dppm-cyp, and dppe
having electron-withdrawing substituents on the aryl rings were synthesized
and used in hydroformylation [50] (see Figure 11, 22-24). The introduction
of electron withdrawing substituents on the aryl rings of the diequatorial
chelate 22 leads to an increase in linear aldehyde selectivity as well as rate.
In contrast, introduction of electron withdrawing substituents on the aryl
rings of the apical-equatorial chelate 24 resulted in a decrease in linear
aldehyde selectivity when compared with the phenyl-substituted dppe.
86 Chapter 4
Table 4. Regiochemistry of Hydroformylation of 1 -Hexene with Rhodium Diphosphine

Catalysts
diphosphine ee:aea 1:bb Ratec

BISBI 100:0 66.5 12.4
22 100:0 123 61.9
Trans-dppm-cyp 37:63 12.1 3.7
23 90: 10 17.7 13.7
dPPe 0: 100 2.6 3.5
24 0:100 1.3 4.3
a
Ratio of diequatorial:apical-equatorial chelates of (diphosphine)Ir(CO)2H at room
c
temperature.b Mol heptanal:mol 2-methylhexanal. Turnover rate = [mol aldehyde] x [mol
-1 -1
Rh] h- .
Interestingly, electron-withdrawing groups in equatorial and apical

positions had opposite effects on 1:b regioselectivity. Hydroformylation
results with the electron withdrawing dieguatorial coordinated ligands 22
and 23 showed an increase in the n:i ratio, while apical-equatorial
coordinated ligand 24 showed a decrease in the 1:b ratio. Thus, electron
withdrawing aryl substituents in the equatorial position lead to higher 1:b
regioselectivity and in the apical position to lower regioselectivity.
It was hypothesized that a dissymmetric dppe derivative
(Ar2PCH2CH2PAr'2) could lead to greater 1:b selectivity than either of the
related symmetric dppe derivatives. A chelate designed to have an electron
withdrawing phosphine in the equatorial position and a electron-donating
phosphine in the apical position should give higher 1:b regioselectivity than
either of the parent symmetric dppe compounds.
The new ligand 25 was synthesized and both its coordination behavior
and selectivity in catalysis were studied [51]. Ligand 25 shows a 1:b ratio of
4.2, higher than either of the symmetric ligands dppe or 24, supporting our
hypothesis that creating an electronically dissymmetric environment about
our hydro formylation catalysts enhances linear aldehyde selectivity. An
electron-withdrawing substituent on an equatorial phosphine increases the
1:b ratio while an electron-withdrawing substituent on an apical phosphine
decreases the 1:b ratio. Therefore. the best dppe catalysts place an electron-
Figure 12. Dissymmetric dppe derivative

withdrawing substituted phosphine in the equatorial position and an electron-

donating substituted phosphine in the apical position. Since ligand 24 gave a
lower 1/b than dppe, the effect of the electron-withdrawing phosphine in the
apical position to decrease 1/b must override the increase in 1/b caused by the
electron-withdrawing phosphine in the equatorial position.
However, an inconsistency emerges if the macroscopic differences
between BISBI and dppe catalysts are considered. The macroscopic changes
in going from H(alkene)Rh(CO)(BISBI) to H(alkene)Rh(CO)(dppe) involve
replacement of an equatorial phosphine (one equatorial BISBI arm) with a
strongly electron accepting CO ligand and replacement of an apical CO
ligand by a strongly electron donating phosphine ligand (apical dppe arm).
The second arm of each diphosphine, in the equatorial position, remains
unchanged. Overall, the replacement of BISBI by dppe places a better
acceptor ligand in the equatorial position and a better donor ligand in the
apical position. By analogy to the effect of electron donor aryl substituents,
both of these changes should have resulted in higher regioselectivity not in
the observed lower regioselectivity seen for dppe. Clearly, the effects on
regioselectivity for equatorial electron-withdrawing phosphines and -
accepting CO ligands are not identical.
4.2.4 Xantphos ligands: tunable bite angles
In 1990 a program was started at the University of Amsterdam aiming at

the synthesis of ligands having bite angles outside the common range of 75
to 99. It was thought that ligands having bite angles wider than this might
induce interesting catalysis by stabilizing or destabilizing initial states,
transition states or final states of catalytic reactions catalyzed by square-
planar, tetrahedral, or trigonal-bipyramidal metal-ligand complexes.
Molecular mechanics methods, as developed in Shell [52], were used for the
design of the ligands, but after the publication by Casey and Whiteker [45]
the latter method was adopted. A variety of backbones have been designed
supporting wider bite angles, but often the synthesis was tedious.
Readily available backbones turned out to be those based on xanthene
and related compounds (Figure 13). The diphosphines derived from these
cover a range of angles from 102 to 121. Key feature is the oxygen ether
atom in the bridge that prevents metallation and that in many instances will
not participate in the coordination to the metal. By changing the fragment in
the back of the molecule (C1, C2, N, S) the bite angle can be easily tuned to
other values. The calculated natural bite angles are shown in Figure 13 and
Table 5. In cationic metal complexes the oxygen will often participate in the
coordination to the metal, which leads to P-M-P bite angles around 165.
88 Chapter 4
R = H, Nixantphos (32) 114.1 Benzoxantphos (34) 120.6 DPEphos (35) 102

R = Bn, Benzylnixantphos (33) 114.2
Figure 13. Xantphos type ligands and their calculated n values
Hydroformylation. Changing the bite angle does usually affect the

catalytic reaction, be it the rate or the selectivity, and in several cases a
favorable effect was reported [46] for Xantphos ligands. For rhodium
catalyzed hydroformylation the first results were published in 1995 by
Kranenburg et al. [53]. In Table 5 we have combined his results (26-34) with
those of Van der Veen to complete the list of Xantphos ligands tested to date
[54]. From the table one can see that the selectivity for linear aldehyde
increases with wider bite angles. For ligands 26-30 the 1:b ratio increases,
but at higher bite angles the correlation breaks down. The rate of reaction
increases with increasing bite angle (with the exception of ligands 31-33) as
one might expect since two phosphines with a small bite angle are a better
donor pair than two phosphines having a wider bite angle. Ligand 34, which
will adopt a flattened confirmation compared with the other ligands, gives an
interesting result in that a low isomerization rate is observed. Ligand 26
assumes a local C2 type symmetry, in contrast to the other ligands all having
C, symmetry. Ligand 35, the diphosphine made from diphenyl ether, gives a
moderate linear branched ratio of 6.7 but no isomerization [53] at a pressure
of 10 bar.
Table 5. Hydroformylation of 1 -octene using Xantphos ligandsa
Ligan n, deg ee:aed 1:b ratiob % lin. ald.b % isomer.b tofb,c

d
26 102.0 3:7 8.5 88.2 1.4 37
21 107.9 7:3 14.6 89.7 4.2 74
28 108.5 6:4 34.6 94.3 3.0 81
29 109.6 7:3 50.0 93.2 4.9 110
30 111.4 7:3 52.2 94.5 3.6 187
31 113.2 8:2 49.8 94.3 3.8 162
32 114.1 7:3 50.6 94.3 3.9 154
33 114.2 8:2 69.4 94.9 3.7 160
34 120.6 6:4 50.2 96.5 1.6 343
a
Conditions: CO/H2 = 1, P(CO/H2) = 20 bar, ligand/Rh = 5, substrate/Rh = 637, [Rh] = 1.00
b
mM, number of experiments = 3. In none of the experiments was hydrogenation observed.
Linear over branched ratio, percentage linear aldehyde, percent isomerization to 2-octene, and
c
turnover frequency were determined at 20% alkene conversion. Turnover frequency = (mol
d
of aldehyde) (mol of Rh)-1 h-1. Calculated from JP-H NMR data.
The X-ray crystal structures of (33) Rh(CO)H(PPh 3) and

(34) Rh(CO)H(PPh3) were determined and both structures reveal distorted
trigonal bipyramidal complex geometries with all the phosphines occuping
equatorial sites. The observed P-Rh-P bite angles for ligands 33 and 34 in
the crystal structures are 110.2 and 109.2, respectively. The observed bite
angles are within the flexibility ranges calculated for the ligands and close to
the bite angle observed in the complex containing ligand 29 (111.7) [55],
but especially the angle for 34 is smaller than the calculated one (120.6).
The bite angle of approximately 110 found in all three crystal structures, is
probably the result of stereoelectronic interactions between the diphosphine
and the large triphenylphosphine ligand. When this PPh3 ligand is replaced
by CO and subsequently by alkene during the catalytic cycle, the P-Rh-P
angle presumably can increase as a result of decreased steric congestion.
Furthermore, the crystal structures show several possibilities for --
stacking interactions between the diffrent phenyl moieties, which could also
influence the final complex geometry.
Electronic effects. To study the nature of the electronic effect in the
rhodium diphosphine catalyzed hydroformylation, a series of thixantphos 29
ligands with varying basicity was synthesized 36-44 (Figure 14). In this
series of ligands, steric differences are minimal so purely electronic effects
could be investigated. To investigate the influence of phosphine basicity on
chelation behavior, the solution structures of the
(diphosphine)Rh(CO)H(PPh3) complexes and the (diphosphine)Rh(H)(CO)2
complexes were studied in great detail for all ligands. Displacement of PPh3
in (PPh3)3Rh(CO)H by the diphosphines 29 and 36-41 resulted in formation
of the (diphosphine)Rh(CO)H(PPh3) complexes. Crystals of the
90 Chapter 4
(diphosphine)Rh(CO)H(PPh3) complexes suitable for X-ray structure

determination were obtained for ligands 29 and 37.
R=N(CH3)2 36
OCH3 37
CH3 38
H 29
F 39
CI 40
CF3 41
Figure 14. Substituted thixantphos ligands
The X-ray crystal structures of (29)Rh(CO)H(PPh3) and

(37)Rh(CO)H(PPh3) are very similar. The P-Rh-P bite angles are 11 1.7 for
the thixantphos ligand 29 and 109.3 for the p-CH3O-thixantphos ligand 37.
The small difference in observed bite angles confirms the assumption based
on molecular mechanics that p-substitution on the aryl rings has only minor
steric effects. The better donor prefers a smaller bite angle. A noticeable
difference between comparable bond lengths in the complexes is found in
the Rh-CO distance. The decreased Rh-CO bond length for the p-CH3O-
thixantphos complex (1.86 ) compared to the thixantphos complex (1.89
) is a result of increased back-bonding of rhodium to the carbonyl ligand.
In situ IR spectra. Unambiguous evidence for the presence of ee and ae
complex isomers was supplied by HP-IR spectroscopy. The spectra of the
(diphosphine)Rh(CO)2H complexes all showed four absorption bands in the
carbonyl region (Figure 15). In an effort to assign the bands to ee and ae
isomers the (thixantphos)Rh(CO)2D complex was measured for comparison.
Upon HID exchange, only v 1 and v 3 shift to lower wavenumbers
(respectively 18 and 14 cm-1), and therefore, these two bands are assigned to
the carbonyls of the ee complex. The two bands that do not shift, v 2 and v 4,
belong to the ae complex. From the disappearance of a low-frequency
shoulder upon H/D exchange, it can be concluded that one of the rhodium
hydride vibrations is partly hidden under v 4. DFT calculations were fully in
line with the observed shifts and wavenumbers, even for the Rh-H stretching
frequencies. The shift in isomer composition found in the NMR spectra (vide
infra) was confirmed when comparing the IR spectra of the series of
complexes (Figure 15).
During hydroformylation the IR spectra of the rhodium species
(diphosphine)Rh(CO)2H for ligands 37, 29, and 41 do not change. The
respective ee:ae equilibria are not influenced, and no other carbonyl
complexes are observed.
Figure 15. IR spectra of complexes RhH(ligand)CO (ligand = 36-41,29)
NMR spectra. For all diphosphine ligands, the formation of the

(diphosphine)Rh(CO)2H complexes was evidenced by 31P NMR spectra and
an apparent triplet of doublets for the rhodium hydride in the 1H NMR
spectra. Decreasing phosphine basicity gives decreasing rhodium-proton
(1J(Rh,H)) and phosphorus-proton (2J(P,H)) coupling constants but
1
increasing rhodium-phosphorus ( J(Rh,P)) coupling constants. The rhodium-
proton and phosphorus-proton coupling constants and their dependence on
phosphine basicity also show that these complexes exist as mixtures of ee
and ae isomers that are in rapid, dynamic equilibrium.
DFT calculations. The enhanced preference of the basic phosphines for
apical coordination and the preference of the nonbasic phosphines for
equatorial coordination was demonstrated by DFT calculations on the ee and
92 Chapter 4
ae isomers of complexes (PH3)2Rh(CO)2H and (PF3)2Rh(CO)2H The +
energies of the ee and ae isomer of the complex modified with PH3 as model
-1
ligand for a basic phosphine were the same within 0.1 keal mol . Uisng PF3
as a model ligand for a nonbasic phosphine the ee isomer was found to be
favored by 2.4 keal mol-1. This corresponds to an ee:ae equilibrium ratio of
approximately 50:1 at room temperature. This trend is consistent with the
experiments.
Hydroformylation results in Table 6 show that, with the exception of
ligands 39 and 40, the rate of the reaction increases with decreasing
phosphine basicity. An explanation for the deviant behavior of 39 and 40 can
be incomplete catalyst formation or deactivation of the catalyst. Decreasing
phosphine basicity facilitates CO dissociation from the
(diphosphine)Rh(CO)2H complex and enhances alkene coordination to form
the (diphosphine)Rh(CO)H(alkene) complex, and therefore, the reaction rate
increases.
Table 6. Electronic effects in hydroformylation using thixantphos derivatives
ligand i ee:ae 1:b 2-octene, % nonanal, % t.o.f.

36 1.7 47:53 44.6 4.8 93.1 28
37 3.4 59:41 36.9 5.3 92.1 45
38 3.5 66:34 44.4 4.7 93.2 78
29 4.3 72:28 50.0 4.9 93.2 110
39 5.0 79:21 51.5 5.7 92.5 75
40 5.6 85:15 67.5 6.9 91.7 66
41 6.4 92:8 86.5 6.8 92.1 158
1-octene, 80 C, 20 bar of 1:1 CO/H2, 1.0 mM rhodium (diphosphine/Rh = 5)
While the 1:b ratio increases with the -value, no electronic effect ofthe
ligands on the selectivity for hear aldehyde is observed. The selectivities
for linear aldehyde are all between 92 and 93%. The increase in 1:b ratio with
decreasing phosphine basicity can be attributed completely to an increased
tendency of the branched alkyl rhodium species to form 2-octene instead of
branched aldehyde (Figure 6). The increasing electrophilicity of the rhodium
center leads to a higher reactivity of the rhodium alkyl species toward CO
dissociation and -hydrogen elimination [ 15a]. The total amount of all other
products (branched aldehyde and 2-octene) is 7-8% for all ligands.
The constant selectivity for linear aldehyde in the hydroformylation of 1-
octene implies that for the basic ligands the 1:b ratio reflects the
regioselectivity of the formation of the rhodium alkyl species. For the less
basic ligands the increase in 1:b ratio results from the different behavior of
branched and linear rhodium alkyls toward -hydrogen elimination, as was
already reported by Lazzaroni and co-workers for the deuteriofomylation of
1-hexene [57]. The linear alkyl is mainly converted to linear aldehyde, while
the branched alkyl partially generates 2-octene. Since 2-octene is far less
reactive in the hydroformylation, its formation is irreversible in these
experiments. We conclude that for this catalytic system the ratio of linear to
branched rhodium alkyl species formed is determined not by phosphine
basicity but by steric constraints only.
13
Kinetics of CO dissociation from (Diph osphine)Rh ( 13 CO) 2H Complexes.
The rate determining step in the hydroformylation of 1-octene with
xantphos-type ligands is in an early stage in the catalytic cycle. Dissociation
of CO, alkene coordination, and migratory insertion are relevant for the
determination of the rate. So far, only the first step has been studied
separately [54]. An experimental set-up was introduced enabling the
measurement of the rate constants for CO dissociation from the
(diphosphine)Rh(CO)2H complexes. By exchanging the 12CO ligands for
isotopically pure 13CO ligands, the carbonyl absorptions in the IR spectra of
the (diphosphine)Rh(CO)2H complexes shift 30-40 cm-1 to lower energy.
Figure 16. Diphosphines used for CO dissociation studies
The rate constants k1 for the dissociation of 13CO from the

(diphosphine)Rh(13CO)2H complexes were determined by monitoring the
exchange of 13CO for 12CO by exposing the I3CO labeled complexes to a
large excess of I3CO so any dissociated I3CO will be replaced quantitatively
by 12CO. The ligands are shown in Figure 16.
To investigate the possible effect of the natural bite angle on the rate of
CO dissociation from the (diphosphine)Rh(CO)2H complexes I3CO
exchange was measured for complexes containing ligands 27, 29 and 31.
The exchange of 13CO by 12CO in the (diphosphine)Rh(13CO)2H complexes
was monitored by rapid-scan HP IR spectroscopy at 40 C. The 13CO/12CO
exchange was initiated by adding a large excess of 12CO. The difference
between the reactivity of the ee and ae isomers of the
(diphosphine)Rh(l3CO)2H complexes cannot be determined, since the
exchange between ee and ae isomers is several orders of magnitude faster
13
than the rates of CO dissociation, as is evident from the NMR spectra.
94 Chapter 4
13
Table 7. Kinetics of CO dissociation in (diphosphine)Rh(13CO)2H complexesa
12 -1
Ligand Bite angle, deg. P( CO), bar k1, h Rate, t.o.f b
27 107.9 25 177 74
29 109.6 25 206 110
29c 25 200
31 113.2 20 182 162
31 25 185
31 30 181
43d 123.1 25 1200 1560
a 13 13
Reaction conditions: [(diphosphine)Rh( CO)2H] = 2.00 mM in cyclohexane, P( CO) = 1
bar, P(H2) = 4 bar, T = 40 C, diphosphine/Rh = 5. b In mol.mol-1.h-1 at 80 C and 20 bar
c
CO/H2 [(diphosphine)Rh(13CO)2H] = 3.00 mM.d See Figure 17 below.
The observed rate constants, k1, are listed in Table 7. It is commonly

accepted that CO exchange in (diphosphine)Rh(CO)2H complexes proceeds
via the dissociative pathway. The decay of the carbonyl resonances of the
(diphosphine)Rh(13CO)2H complexes indeed followed simple first-order
kinetics. The experiments with ligand 31 at different 2CO partial pressure
1
show that the rate of CO displacement is independent of the CO pressure.

Furthermore, the rate is also independent of the (diphosphine)Rh(13CO)2H
complex concentration, as demonstrated by the experiments with ligand 29.
It can therefore be concluded that CO dissociation for these complexes obeys
a first-order rate-law and proceeds by a purely dissociative mechanism.
Table 7 shows that there is no correlation between the rate of CO
dissociation and the natural bite angle. The rate constants k1 for ligands 27,
29, and 31 do not differ significantly and cannot explain the trend in
observed hydroformylation activities. The comparison of the k1 values for
ligand 27, 29, and 31 with the turnover frequencies depicted in Table 5,
reveal that the rates of CO dissociation, measured at 40 C, are higher than
the hydroformylation rates at 80 C. Since reaction rates increase
approximately an order of magnitude with a temperature rise of 20 degrees,
the CO dissociation rate at 80 C is about 100 times as fast as the
hydroformylation reaction.
In this study no influence of the natural bite angle on the rate of
formation of the square planar (diphosphine)Rh(CO)H intermediate (k1) was
found, implying that the activation energy for the formation of these
complexes is not affected significantly. Therefore, the increase in
hydroformylation rate with increasing bite angle might originate either from
an increase in the concentration of these four-coordinate complexes, or from
a decrease in the activation energy for alkene coordination (if this step were
rate-determining), or a faster migratory insertion of the alkene.
It is not likely that increasing the bite angle would lower the activation
energy for alkene coordination. Increasing the bite angle results in increased
steric congestion around the rhodium center and consequently in more steric
hindrance for the alkene entering the coordination sphere. What kind of
electronic effect the widening of the bite angle has on the activation energy
for alkene coordination is unclear, since it depends on the bonding mode of
the alkene. Rhodium to alkene back-donation is promoted by narrow bite
angles, while alkene to rhodium donation is enhanced by wide bite angles
[62].
Widening the bite angle of a cis bidentate in a square planar complex
would certainly accelerate a migration reaction, but is is not clear how this
would work out in trigonal bipyramid having the diphosphine as a bis-
equatorial ligand. We will return to this in section 4.2.5.
Figure 17. New phosphole and phenoxaphosphino xantphos ligands
Internal alkenes. Dibenzophosphole- and phenoxaphosphino substituted

xantphos ligands 42 and 43 were reported by van der Veen et al. [58, 59]
(Figure 17). They show a high activity and selectivity in the rhodium
catalyzed linear hydroformylation of 1-octene (1:b = > 60). More
importantly, ligands 42 and 43 exhibit an unprecedented high activity and
selectivity in the hydroformylation of trans 2- and 4-octene to linear nonanal.
Table 8. Hydroformylation of 2- and 4-octene a,d

c
Ligand Substrate 1:b ratiob % linear ald t.0.f .
PPh3 2-octene 0.9 46 39
42 9.5 90 65
43 9.2 90 112
PPh3 4-octene 0.3 23 2
42 6.1 86 15
43 4.4 81 20
a c
[Rh] = 1.0 mM, Rh:P:octene = 1:10:673. b 1:b Ratio includes all branched aldehydes.
-1 -1
Turnover frequencies were calculated as (mol aldehyde)(mol catalyst) (h) at 20-30%
d
conversion. 120 C and p = 2 bar.
96 Chapter 4
They constitute the first rhodium phosphine modified catalysts for such a
selective linear hydroformylation of internal alkenes. The extraordinary high
activity of 43 even places it among the most active diphosphines known.
Since large steric differences in the catalyst complexes of these two ligands
are not anticipated, the higher activity of 43 compared to 42 might be
ascribed to very subtle bite angle effects or electronic characteristics of the
phosphorus heterocycles.
The high 1:b ratios of ligands 42 and 43 are in good agreement with
earlier observations that diphosphines with natural bite angles close to 120
give very selective catalysts. The calculated bite angles are largely due to the
interference of the rigid substituent rings at phosphorus. Table 8 shows the
acitivities and selectivities achieved for internal alkenes. To enhance the rate
of isomerization and to prevent hydroformylation of the internal alkenes, a
relatively high temperature and low pressure were used. The dissociation
rate of CO from the resting state (43)RhH(Co)2 was very fast, several times
faster than that found for diphenylphosphino Xantphos ligands (see Table 7).
Thus larger (natural) bite angles promote the formation of four-coordinate
species, as may be expected. For the same reason they will enhance
isomerization, since this involves a competition between CO association and
-hydrogen abstraction by the vacant site.
4.2.5 The mechanism, regioselectivity, and the bite angle

Concluding remarks
Summary of ligand effects. A wider bite angle leads to higher 1:b ratios
and to a higher proportion of bis-equatorial hydride resting states. Above a
certain value of n or the ee:ae ratio the effect on the 1:b ratio levels off. It
might be that in a region of 104 to 113 a change in mechanism occurs.
When the bite angle of the diphosphine is changed with large increments
from 78 (dppe), to 107 (trans-dppm-cyp), to 112 (BISBI or Xantphos),
the trend is clear; a higher 1:b ratio is found. The same applies to electronic
effects in symmetrically substituted diphosphines; a higher -value leads to
higher 1:b ratios. However, the latter was not true for the Xantphos series
shown in Figure 14 and Table 6 when the selectivity is considered instead of
the 1:b ratio. Wider bite angles and higher -values both lead to faster
catalysts and a relatively higher rate of isomerization. For propene the
isomerization is immaterial and as a consequence very high linearities can be
obtained using ligands displaying these properties. When isomerization
becomes sufficiently fast, even internal alkenes can be converted to linear
products, albeit at moderate rates (ligands 42 and 43).
Bis-equatorial coordination as in 2ee is not a prerequisite for high
linearity, as selective catalysts show a substantial amount of 2ae. On the
other hand, ligands showing purely equatorial-apical coordination give the

lowest 1:b ratios, even lower than monophosphines as PPh3. Monodentate
phosphines can always give trans-like intermediates, while ligands having a
small angle cannot (intermediates 4ae and 5c in Figure 19).
Ligands having high -values prefer equatorial sites and lead to higher
1:b ratios. Dissymmetrically substituted dppe ligands give, indeed, the
acceptor arm of the bidentate in the equatorial plane together with an
electron-accepting CO ligand, but the unexpected result in hydroformylation
is a lower 1:b ratio.
After dissociation of carbon monoxide (or a phosphine) the unsaturated
intermediates 3c, 3t, 10c, and 10t may form, which will subsequently
undergo association with alkene to give species 4.
Figure 18. Migratory insertion in tbp-like structures for mono and bidentate dppp
Figure 19. Structure of key intermediate 3wide-t
Bidentate ligands having wide bite angles will lead to species similar to
3t and 4ee. Ligands with wide bite, such as BISBI and Xantphos, should
impose little deformation of the trans P-Rh-P angle of 4t. Indeed, in
RhH(CO)(PCy3)2, the P-Rh-P angle is 164 [56], which seems well within
reach of the ligands having wide bite angles such as BISBI and Xantphos
[45, 46]. This species 3wide-t associates with the alkene to give 4ee. The
overall large steric hindrance in such a complex could be sufficient
98 Chapter 4
explanation for the selective formation of a primary rhodium alkyl species.

Ligands with bite angles in excess of 125, such as 21 [58] and 44 [57] (see
Figure 20), lead to low linearities [53]. They do not form the desired species
2ee [47].
Figure 20. Structure of dibenzofuran based ligand with n = 131
Transition state geometry. As we have seen above, a better knowledge of

the geometry of the transition state is of the utmost importance before we
can arrive at a better understanding of the bite angle effect. Two
theoretical studies have been devoted to the mechanism of the
hydroformylation reaction, one by Morokuma [60] and the other by Gleich
[61]. Using DFT methods Morokuma finds that the trigonal bipyramid
containing an apical hydride and ethene in the equatorial plane has to
rearrange to a square pyramid having both hydride and ethene in the basal
plane prior to migration. This leads to a structure that has an orbital of dx2-y2
character as the empty orbital, a situation very similar to insertion processes
in palladium(II) complexes.
In Figure 2 1 we have depicted the square pyramidal intermediate containing
the diphosphine in the basal plane, together with the hydride and ethene
ligand. A carbonyl is drawn at the apical position, as was found to be the
most stable geometry. The directing effect of the bidentate is now very
similar to that in square planar complexes. The alkyl substituent of the
alkene will experience steric hindrance from the diphosphine, in the basal
plane and the most favorable position for the alkyl group will be away from
the ligand, leading to linear alkyl formation when migration takes place.
During migration the carbonyl moves to the vacancy left by the hydride. The
wider the bite angle, the stronger the steric hindrance forcing the alkene in
the desired position. Morokuma reported that the rearrangement from 4ee or
4ae involves most of the activation energy, rather than the bond breaking,
which points to an early transition state in which non-bonding interactions
are most important. For wide bite angles one would expect a trans
disposition of the diphosphine and as a result there is no cis relation between
hydride and alkene. For bidentate ligands the movements involved seem
unlikely and therefore we reject this transition state.
Gleich states that the geometry is less well defined and that just as well
one can describe the transition state as a tbp structure in which the alkene
assumes a position perpendicular to the trigonal plane in the transition state.
Figure 21. Square pyramidal transition state
The study of Gleich et al. involves a combination of quantum mechanics

to calculate the geometry of the transition state and molecular mechanics to
calculate the steric effects of the ligands. The tbp case is shown above in
Figure 19. A wider bite angle will create more steric hindrance around the
metal and a linear alkyl will result, but this could not be substantiated by
molecular mechanics calculations using an estimated transition state [47c].
Thus according to the latter study dppe gives a calculated 1:b ratio even
higher than BISBI, as we hypothesized above for 4ae, which might mean
that there is another intermediate involved having a lower barrier and giving
a low 1:b ratio.
Table 9. Experimental versus calculated regioselectivities [61]

Ligand 1:b, exp. 1:b, calc.
BISBI 17.0 8 .0
dppe 2.2 3.4
PPh3 3.5 3.5
Gleich also assumes that the migratory insertion of alkene into rhodium
hydride determines both the rate and the regioselectivity. In their
calculations they start from a common alkene complex that leads to both
linear and branched alkyl rhodium. The 1:b ratios calculated for BISBI, dppe
and PPh3 agree quite well with those obtained from the experiment.
For PPh3 the outcome seems surprising, because the bis-equatorial
complex was used for the calculation for which one would expect a high 1:b
ratio (~12) according to experiments at high phosphine concentrations. For
equatorial-apical coordination modes for both bidentates and PPh3 they also
calculate very higher 1:b ratios, but indeed the transition state energies are
100 Chapter 4
much higher by 7 kca/mol and therefore these pathways do not contribute to

the product formation.
Energetics. While a lot of progress has been made both experimentally
and theoretically, the kinetics and energetics of the hydroformylation
sequence remain unclear, even if we only consider the type-I kinetics.
Above we discussed the following scheme, now modified for propene:
RhHL2(CO)2 (2 ) RhHL2 (CO) (3) + CO (1)
RhHL2 (CO)2 (3)+C3 H6 RhHL2(CO)(C3H6) (4) (2)
RhL2 (CO)(C3 H6) (4) Rh(1-C 3H7)L2(CO) (51) (3-linear)
RhHL2(CO)(C 3H 6) (4) Rh(2-C3H7)L2(CO) (5b) (3-branched)
Rh(2-C3H7)L2(CO) (5b) + CO Rh(2-C3H7)L2(CO)2 (6) (4)
In this scheme we have assumed that insertion to give the linear alkyl
rhodium species is irreversible and rate-determining and rapid pre-equilibria
exist for CO and propene association and dissociation. From deuteration
studies we know that the conditions can be chosen in such a way that this is
valid. At high temperatures and low pressures this is no longer true. The
free-energy diagram for the kinetics showing irreversible insertion have been
drawn in Figure 22, reactions (1), (2), and (3-linear).
For the formation of branched alkyls either from I-alkenes or internal
alkenes this scheme must be modified, because now often the insertion is
reversible, dependent on conditions and concentrations. A potential scheme
showing the competition between the backward reaction (3-branched) and
the complexation of CO reaction (4) is shown in Figure 23. At low
temperatures and sufficiently high pressures the formation of 2-alkyl species
from 1-alkenes can also be irreversible. We have drawn the barriers for the
backwards reaction (-elimination) and forward reaction (CO complexation)
at about the same height to indicate the competition between the two steps.
The 2-alkyl species shown in Figure 23 can also be formed from 2-
alkenes and this species should have the same energy. We know that in most
catalytic systems the hydroformylation of internal alkenes is one or two
orders of magnitude slower. From the viewpoint of microscopic reversibility
this leads to an interesting point. It would seem therefore that in most
systems a relatively fast hydrogen exchange must take place for internal
alkenes while no or little hydroformylation occurs.
Figure 22. Energy diagram Type-I kinetics in which hydride migration is rate-determining for
1 -alkenes
Figure 23. Reversible migratory insertion of alkene to 2-alkylrhodium
For bulky phosphite (Chapter 3) systems it is known that cyclooctene

undergoes deuterium exchange at a rate an order of magnitude higher than
that of hydroformylation [48]. The pathway for insertion of 1 -butene to give
2-butylrhodium might well be very similar to the pathway leading to 1-
butylrhodiuin, but likely the complexation of CO is more cumbersome in the
branched alkyl species. B-elimination may lead to 2-butene, which forms a
102 Chapter 4
less stable complex than 1 -butene (not shown). Clearly, energy differences
are very small and entropy and concentrations will have large influence on
the actual kinetics.
The few data known on CO dissociation from the starting
pentacoordinate rhodium hydrides show that the rate of these processes are
only two orders of magnitude higher than those of the overall
hydroformylation reaction. For this reason this barrier was drawn as it is in
Figures 22 and 23. One might speculate that coordination of the alkene is
rate-determining. A simplified scheme is shown in Figure 24.
Figure 24. Type-I kinetics for rate-determining alkene complexation
The regioselectivity can be determined in two ways, by the formation of

distinct stereoisomeric alkene complexes, assuming slow rotation, or by the
migration reactions leading to linear and branched alkyl species. The rate-
determining and selectivity determining steps do not coincide in the latter
case. The kinetic equation does not distinguish between the mechanism of
Figure 22 and that of Figure 24.
In spite of all recent findings we conclude that there is still a lot to be
done before the kinetics, the mechanism, and ligand effects in
hydro formylation will be fully understood.
References
1 Bath, S. S.; Vaska, L. J. Am. Chem. SOC. 1963, 85, 3500.

2 (a) Evans, D.; Osborn, J. A.; Wilkinson, G. J. Chem. SOC. (A), 1968, 3133. (b) Evans, D.;
Yagupsky, G.; Wilkinson, G. J. Chem. SOC. A 1968, 2660. (c) Brown, C. K.; Wilkinson,
G. .J Chem. SOC. A 1970 ,2753.
3 Jardine, F. H. Polyhedron, 1982, 1, 569.
4 Jardine, F. H.; Osbom, J. A.; Wilkinson, G.; Young, J. F. Chem. and Ind. (London)
1965, 560.
5 (a) Pruett, R. L. Adv. Organornet. Chem. 1979 , 17, 1. (b) Comils, B. in New syntheses
with Carbon Monoxide, (Ed. Falbe, J.), Springer, Berlin, 1980 , Chapter 1. (c) Trzreciak,
A. M.; Zilkowski, J. J. Coord. Chem. Rev. 1999 , 190-2, 883. (d) Pino, P. J. Organornet.
Chem. 1980, 200,223. (e) Falbe, J. (Ed.), New Syntheses with Carbon Monoxide,
Springer-Verlag, New York, 1980. (f) Leeuwen, P. W .N. M.; van Koten, G. Catalysis,
an Integrated Approach, (Ed. van Santen, R. A.; van Leeuwen, P. W. N. M.; Moulijn, J.
A.; Averill, B. A.) Elsevierscience Publishers, Amsterdam, 3rd Ed. 1999, Chapter 6. (g)
Frohning, C. D.; Kohlpaintner, C. W. in Cornils, B.; Herrmann, W. A. Eds. Applied
Homogeneous Catalysis with Organometallic Compounds, VCH-Wiley: New York,
1996; Vol. 1, pp 3-25. (h) Tolman, C. A.; Faller, J. W. in Homogeneous Catalysis with
Metal Phosphine Complexes, Ed. Pignolet, L. H. Chapter 2, Plenum Press, New York,
1983, pp 81-109.
6 (a) Hegedus, (b) Cotton, Wilkinson, (c) Masters, (d) Parshall G. W.; Ittel, S. D.
Homogeneous Catalysis, 2nd Ed. 1992, Wiley, New York.
7 (a) Heck, R. F.; Breslow, D. S. J. Am. Chem. Soc. 1961, 83 ,4023. (b) Heck, R. F.;
Breslow, D. S. J. Am. Chem. SOC. 1962, 84, 2499. (c) Heck, R. F. Acc. Chem. Res. 1969,
2, 10.
8 Brown, J. M., Kent, A.G. J. Chem. Chem. SOC. Perkin Trans. II, 1987, 1597.
9 Dahlenburg, L.; Mirzaei, F.; Yardimeioglu, A. Z. Naturforsch . l982, 37b, 310.
10 Deutsch, P. P.; Eisenberg, R. Organometallics, 1990, 9, 709.
11 Chan, A. S. C.; Shieh, H. S.; Hill, J. R. J. Chem. Soc., Chem. Commun. 1983, 688.
12 Castellanos-Paz, A.; Castilln, S.; Claver, C.; van Leeuwen, P. W. N. M.; de Lange, W.
13 (a) Oswald, A. A.; Hendrikse, D. E.; Kastrup, R.V.; Irikura, K.; Mozeleski, E. M.;
Young, D. A. Phosphorus and Sulfur, 1987, 30, 237. (b) Oswald, A. A.; Merola, J. S.;
Mozeleski, E. J.; Kastrup, R.V.; Reisch, J. C. Adv. Chem. Series 1981, 104, 503. (c)
MacDougall, J .K.; Simpson, M. C.; Green, M. J.; Cole-Hamilton, D. J. J. Chem. SOC.
Dalton Trans. 1996, 1161. (d) Lawrenson, M. J. (to B.P.) UK Patent, 1,254,222 and
1,284,615; Chem.Abstr. 1972, 76, 33787 and 1972, 77, 125982.
14 Moser, W. R.; Papite, C. J.; Brannon, D. A.; Duwell, R. A.; Weininger, S. J. J. Mol Cat.
1987, 41, 271.
15 (a) Unruh, J. D.; Christenson, J. R. J. Mol Cat. 1982, 14, 19. (b) Unruh, J. D.; Wells, W.
J. Belgian patent 840,906 1976. (c) Unruh, J. D.; Segmuller, B. E.; Chupa. G. R.; Pryor,
K. E. (to Hoechst Celanese) U.S. Patent 5,567,856; Chem. Abstr. 1998, 125, 328099.
16 Buhling, A.; Kamer, P. C. J.; van Leeuwen, P. W. N. M. J. Mol. Catal. A: Chemical,
1995, 98, 69.
17 Janecko, H.; Trzeciak, A. M.; Ziolkowski, J. J. J. Mol. Cat. 1984, 26, 355.
18 van Rooy, A.; de Bruijn, J. N. H.; Roobeek, C. F.; Kamer, P. C. 1.; van Leeuwen, P. W.
N. M. J. Organomet. Chem. 1996, 507, 69.
19 Palo, D. R.; Erkey, C. Organometallics, 2000, 19, 81.
104 Chapter 4
20 (a) Hayashi, T.; Tanaka, M.; Ogata, I. J. Mol. Catal., 1979, 6, 1. (b) Hjortkjaer, J.;
Toromanova-Petrova, P. J. Mol. Cat. 1989, 50, 203. (c) Neibecker, D.; Rau, R. J. Mol.
Catal., 1989, 53, 219.
21 Torrent, M.; Sol, M.; Frenking G. Chem. Rev. 2000, 100, 439.
22 Van Leeuwen, P. W. N. M.; Roobeek, C. F. J. Organomet. Chem. 1983, 258, 343.
23 Diguez, M.; Claver, C.; Masdeu-Bult, A. M.; Ruiz, A.; van Leeuwen, P. W. N. M.;
Schoemaker, G. C. Organometallics, 1999, 18, 2107
24 (a) Cavalieri dOro, P.; Raimondo, L.; Pagani, G.; Montrasi, G.; Gregorio, G.; Andreetta,
A. Chim. Ind. (Milan), 1980, 62, 572. (b) Gregorio, G.; Montrasi, G.; Tanipieri, M.;
Cavalieri dOro, P.; Pagani, G.; Andreetta, A. Chim. Ind. (Milan), 1980, 62, 389.
25 (a) Strohnieier, W.; Michel, M. Z. Phys. Chem.Neue Folge, 1981, 124, 23 (in German).
(b) Hjortkjaer, J. J. Mol. Cat. 1979, 5, 377.
26 (a) Musaev, D. C.; Morokuma, K. Adv. Chem. Phys. 1996, 95, 61. (b) Musaev, D. G.;
Matsubara, T.; Mebel, A. M.; Koga, N.; Morokuma, K. Pure Appl. Chem. 1995, 67, 257.
27 (a) Nair, V. S.; Mathew, S. P.; Chaudhari, R. V. J. Mol. Catal., A 1999, 143, 99. (b)
Divekar, S. S.; Deshpande, R. M.; Chaudhari, R. V. Catal. Lett.. 1993, 21, 191. (c)
Deshpande, R. M.; Chaudhari, R. V. Ind. Eng. Chem. Res. 1988, 27, 1996.
28 Palo, D. R.; Erkey, C. Ind. Eng. Chem. Res. 2000, 38, 3786.
29 (a) Heil, B.; Marko, L. Chem. Ber. 1968, 101, 2209. (b) Feng, J.; Garland, M.
30 Gerritsen, L. A.; Klut, W.; Vreugdenhil, M. H.; Scholten, J. J. F. J. Mol. Catal. 1980, 9,
265.
31 Van der Veen, L. A.; Kamer, P. C. J.; van Leeuwen, P. W. N. M.; Organometallics,
2000, 19, ***.
32 lwamoto, M.; Yuguchi, S. J. Org. Chem. 1966, 31, 4290.
33 (a) Dang, T. P.; Kagan, H. B. J. Chem Soc., Chem. Commun. 1971, 481. (b) Knowles,
W. S.; Sabacky, M. J.; Vineyard, B. D.; Weinkauff, D.J. J. Am. Chem. Soc.. 1975, 97,
2567.
34 (a) Sanger, A. R. J. Mol. Catal. 1977/8, 3, 221. (b) Sanger, A. R.; Schallig, L. R. J. Mol.
Catal. 1977/8, 3, 10 1.
35 Castellanos-Pez, A.; Castilln, S.; Claver, C.; van Leeuwen P. W. N. M.; de Lange, W.
36 Freixa, Z.; Pereira. M. M.; Pais, A. A. C. C.; Bayn, J. C. J. Chem. Soc., Dalton Trans.
1999, 3245.
37 (a) Consiglio, G.; Botteghi, C.; Salomon, C.; Pino, P. Angew. Chem. 1973, 85, 665. (b)
Pittman Jr., C. U.; Hirao, A. J. Org. Chem. 1978, 43, 640.
38 Kawabata, Y.;Hayashi, T.; and Ogata I. J. Chem. SOC., Chem. Commun., 1979, 462.
39 Van Leeuwen. P. W. N. M.; Roobeek, C. F. Brit. Patent Appl. 33,554, 1981 (to Shell
Resarch); Chem. Abstr. 1982, 96, 6174. Van Leeuwen, P. W. N. M.; Roobeek, C. F. J.
Mol. Catal., 1985, 31, 345-53.
40 Hughes, O. R.; Unruh, J. D. J. Mol. Catal. 1981, 12, 71.
41 Hughes, 0. R.; Young, D. J. Am. Chem. Soc. 1981, 103, 6636.
42 (a) Nettekoven, U.; Kamer, P. C. J.; Widhalm, M.; van Leeuwen, P. W. N. M.
Organometallics, 2000, 19, ***. (b) The first reexamination is believed to be in error,
since it places hydride in an equatorial position: Fung, D. C. M.;James, B. R. Gazz.
Chem. Ital. 1992, 122, 329.
43 (a) Devon, T. J.; Phillips, G. W.; Puckette, T. A.; Stavinoha, J. L.; Vanderbilt, J. J. (to
Texas Eastman) U.S. Patent 4,694,109, 1987; Chem. Abstr. 1988, 108, 7890. (b) Devon,
T. J.; Phillips, G. W.; Puckette, T. A.; Stavinoha, J. L.; Vanderbilt, J. J. (to Texas
Eastman) US. Patent 5,332,846, 1994; Chem. Abstr. 1994, 121, 280,879.
44 Casey, C. P.; Whiteker, G. T.; Melville: M. G.; Petrovich, L. M.; Gavney, J. A., Jr.;
45 Casey, C. P.; Whiteker, C. T. Isr. J. Chem 1990, 30, 299.
46 Dierkes, P.; van Leeuwen, P. W. N. M. J. Chem. Soc. Dalton Transnctions, 1999, 1519.
47 (a) Yamamoto, K.; Momose, S.; Funahashi. M.; Miyazawa, M. Synlett. 1990, 711. (b)
Casey, C. P.; Whiteker, G. T. J. Org. Chem. 1990, 55, 1394. (c) Casey, C. P.; Petrovich,
L. M. J. Am. Chem. Soc., 1995, 117, 6007.
48 Jongsma, T.; Challa, G.; van Leeuwen, P. W. N. M. J. Organometal. Chem. 1991, 421,
121.
49 van Rooy, A.; Orij, E. N.; Kamer, P. C. J.; van Leeuwen. P. W. N. M. Organometallics,
1995, 14, 34 and unpublished.
50 Casey, C. P.; Paulsen, E. L.; Beuttenmueller, E. W.; Proft, B. R.: Petrovich, L. M.;
Matter, B. A.; Powell, D. R. J. Am. Chem. Soc. 1997, 119, 11817.
51 Casey, C. P.; Paulsen, E. L.; Beuttenmueller, E. W.; Proft, B. R.; Matter, B. A.; Powell,
D. R. J. Am. Chem. Soc. 1999, 21, 63.
52 Van der Made, A. W.; van Leeuwen, P. W. N. M., unpublished.
53 Kranenburg, M.; van der Burgt, Y. E. M.; Kamer, P. C. J,; van Leeuwen, P. W. N. M.
54 van der Veen, L. A.; Keeven, P. H.; Schoemaker, C. C.; Reek, J. N. H.; Kamer, P. C. J.;
van Leeuwen, P. W. N. M.; Lutz, M.; Spek A. L. Organometallics, 2000, 19, 872.
55 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.
56 Freeman, M. A.; Young, D. A. Inorg. Chem. 1986, 25, 1556.
57 Lazzaroni, R.; Raffaelli, A.; Settambolo, R.; Bertozzi, S.; Vitulli, G. J. Mol. Catal. 1989,
50, 1.
58 van der Veen, L. A.; Kamer, P. C. J.; van Leeuwen, P. W. N. M. Orgainometallics, 1999,
18, 4765.
59 van der Veen, L. A.; Kamer, P. C. J.; van Leeuwen, P. W. N. M. Angew. Chem. Int. Ed.
Engl. 1999, 38, 336.
60 (a) Koga, N.; Qian Jin, S.; Morokuma, K. J. Am. Chem. Soc. 1988, 110, 3417. (b)
Matsubara, T.; Koga, N.; Ding, Y.; Musaev, D. G.; Morokuma, K. Organometallics,
1997. 16, 1065.
61 Gleich, D.; Schmid, R.; Hemnann,W. A. Qrganometallics, 1998, 17, 4828.
62 Tatsumi, K.; Hoffmann, R.; Yamamoto, A.; Stille, J. K. Bull. Chem. Soc. Jpn. 1981, 54,
1857.
Chapter 5
Asymmetric hydroformylation
Carmen Claver and Piet W.N.M. van Leeuwen

Departament de Qumica Fsica i Inorgnica. Universitat Rovira i Virgili, Plaza Imperial
Tarraco, 43005, Tarragona, Spain: Institute ofMolecular Chemistry, University ofAmsterdam.
Nieuwe Achtergracht 166, 1018 WV, Amsterdam, The Netherlands
5.1 Introduction
In this chapter we will concentrate on the asymmetric hydroformylation

of vinyl aromatics, which are model substrates of interest to the
pharmaceutical industry [ 1]. In 1993 and 1995 some reports were published
which described the state of the art in hydroformylation with both rhodium
and platinum systems [2, 3]. A report about carbonylation appeared in 1999
[4].
Figure 1. Hydroformylation of styrene
Highly enantioselective hydroformylation catalyzed by chiral metal

complexes has been obtained with only a few catalytic systems. Many chiral
phosphorus ligands have been used in Pt(II) and Rh(I) systems in the
asymmetric hydroformylation of styrene. The first highly enantioselective
examples of asymmetric hydroformy lation of styrene were reported by
Consiglio et al. in 1991 and used Pt-Sn systems [5, 6]. They achieved an ee
of 86%. Platinum catalysts have several disadvantages: they have low
reaction rates, they hydrogenate the substrate and their regioselectivity to the
107
108 Chapter 5
branched aldehyde is low. Pt-diphosphit/SnCl2 systems also have low

9
selectivity. When the appropriate diphosphite has been used, ee s have been
as high as 90% [7].
Modified Rh systems show considerable activity in the hydroformylation
of styrene to the branched aldehydes. Chiral diphosphine rhodium
complexes that lead to high activity and high regioselectivity in branched
aldehyde have been studied. These systems do not provide high
enantioselectivity [2, 3]. Stanley reported that a tetraphosphine ligand can be
used to form a bimetallic complex [8] and provide ees up to 85% using
vinyl esters.
The first reports on asymmetric hydroformylation using diphosphite
ligands revealed no asymmetric induction [9, 10]. In 1992, Takaya et al.
published the results of the asymmetric hydroformylation of vinyl acetate
(ee=50%) with chiral diphosphites [ 11]. In 1992, an important breakthrough
appeared in the patent literature when Babin and Whiteker at Union Carbide
reported the asymmetric hydroformylation of various alkenes with ees up to
90%, using bulky diphosphites derived from homochiral (2R,4R)-pentane-2,
4-diol, UC-PP * (see Figure 2). Van Leeuwen et al. studied the influence of
the bridge length, bulky substituents and cooperativity of chiral centers on
the performance of the catalyst [12, 13].
Figure 2. Chiral diphosphites and phosphine-phosphite ligands
The catalyst discovered by Takaya et al. using (R,S) BINAPHOS, a

phosphine-phosphite ligand of C1 symmetry [14] has provided ees as high
as 96% as well as total conversions and high regioselectivities. The origin of
the stereodifferentiation in rhodium catalyzed hydroformylation has been
discussed in theoretical reports [ 15, 16].
5. Asymm etric hydroformy la ti on 109
5.2 Rhodium systems with chiral diphosphite ligands
5.2.1 C2 Symmetric chiral diphosphite ligands
Optically active diols are useful building blocks for the synthesis of chiral
diphosphite ligands. Chiral diphosphites based on commercialIy available
optically active 1,2 and I ,4-diols, I ,2:5,5-diisopropylidene-D-maiinitol, L-
,,,-tetramethyl-1,3-dioxalan-4,5-dimethanol and L-diethyl tartrate, 1-3,
were first used in the asymmetric hydroformylation of styrene [ 17].
2 (2R, 3R) R= H
Figure 3. Chiral diphosphites used in rhodium asymmetric hydroformylation
Ee's reached 20 % and regioselectivities to the branched aldehyde were

high with the most bulky ligand under mild reaction conditions. The less
bulky and more flexible diphosphites hardly produce enantioselectivity.
Chiral diphosphites, 4-7 with C2 symmetry based on (2R,3R)-butane-2,
3-diol, (2R,4R)-pentane-2,4-diol and (2S,5S)-hexane-2,5-diol have been
synthesized in order to study the effect of the length of the bridge (see Figure
4). In this series, tert-butyl and methoxy substituents were introduced at the
110 Chapter 5
ortho and para positions of the biaryl moieties to vary the properties of the
ligands [13].
Diphosphite ligands containing biaryl moieties have low energy barrier
for interconversion in atropisomers, which could lead to the formation of
several diastereomers (see Figure 5). However, the rotation around the biaryl
axis of the diphosphite ligands containing bulky substituents in the biaryl
moieties is hindered. The advantage of using bisnaphthol derivatives is that
interconversion around the bisnaphthol bond is energetically highly
unfavorable, so stable diastereomeric diphosphites were obtained in a pure
form. Because bulky substituents in the aromatic bisphenol positions
significantly affect catalyst performance, derivatives based on 2,4-
pentanediol with ortho substituents, having increasing steric bulk trimethyl-
triethyl and tert-butyl dimethylsilyl have also been used, 8 and 9 [ 18].
Figure 4. Variation of the length of the bridge in diphosphite ligands

5. Asymmetric hydroformy la tion 111
Figure 6. Bisphenol and bisnaphthol pentane-2,4-diol diphosphites
The trimethylsilyl group is the substituent that provides the best

enantioselectivity, and is used to prepare the optically pure bisnaphthol
substituents in diphosphites 9 [18]. One of the interesting aspects of this
study is that it analyses the solution structures of the rhodium diphosphite
system with NMR and IR spectroscopy. The [RhH(CO)2(diphosphite)]
complexes form a mixture of isomers containing an equatorial-equatorial
(ee) or apical-equatorial (ae) coordination of the diphosphite in a trigonal
bipyramidal structure. Owing to the C2 symmetry, the phosphorus atoms are
equivalent in the free ligand. However, the trigonal bipyramidal
hydridorhodium dicarbonyl [RhH(CO)2(diphosphite)] complex, the putative
resting state of the hydroformylation catalyst, has C1 symmetry [ 13, 18, 19].
5.2.2 Catalyst preparation and hydroformylation results
The chiral diphosphites 4-9 have been used in the rhodium catalyzed
asymmetric hydroformylation of styrene. The catalysts are prepared in situ
by adding the diphosphite L to [Rh(acac)(CO)2] (acac = acetylacetonate) as a
catalyst precursor. Under typical hydroformylation conditions the active
catalyst precursor [RhH(diphosphite)(CO)2] is formed (See 5.2.3). Two
important features have been found for the preparation of efficient catalyst
using these diphosphites: (1) An excess of diphosphite is used to exclude the
112 Chapter 5
formation of [RhH(CO)4] which is a highly active achiral hydroformylation

catalyst. (2) The [RhH(diphosphite)(CO)2] can form from [Rh(acac)(CO)2]
and the diphosphite in an incubation time of about 15h under
hydroformylation conditions. Enantioselectivities were high under relatively
mild reaction conditions (Tables I and 2). The regioselectivity for branched
aldehyde (2-phenylpropanal) always exceeded 90%. When the reaction
temperature was decreased from 40 to 25 C rates were lower but the
enantiomeric excess increased considerably (Table 1).
Table 1. Hydroformylation of styrene with chiral rhodium diphosphite catalysts

c
Diphosphite T C Turnover Conversion % 2-phenyl- % ee (Absolute
frequencyb % propanal Configuration)
4a 40 66 74 93 19 (S)
4b 40 177 d 99 92 25 (S)
4b 25 31 40 93 34 (S)
5a 40 117 81 80 11 (R)
5b 40 113 89 96 50 (S)
5c 40 207d 98 94 67 (S)
7a 40 19 26 93 1 (R)
7b 40 19 26 92 7 (R)
6 40 165 99 90 47 (S)
6 25 40 45 95 62 (S)
All reactions were carried out with 13.3 mmol of styrene.CO/H2 = 10 bar. Styrene:catalyst
b
molar ratio is 421, P:Rh molar ratio is 2.5. In mol styrene (mol Rh-1)h-1 determined after 2 h
c d
ofreaction by GC. Of styrene after 5 h. After 1 h of reaction.
The influence of the structure of various diphosphites on the activity and

regioselectivity in the rhodium catalytic hydroformylation has been studied
by modifying the diphosphites previously reported by Union Carbide. In the
hydroformylation of 1 -octene, a bulky diphosphite coordinating in a bis-
equatorial fashion is required if the regioselectivity is to be high in linear
aldehyde. A kinetic study was made to determine several parameters
affecting the reaction rates and regioselectivity in the hydroformylation of 1 -
octene and styrene [20].
The catalytic systems formed with the diphosphites based on (2R,4R)-
pentane-2,4-diol in which the bisphenol moiety is substituted with methoxy
groups at the para position, 5c and 6, give the highest activity. The catalyst
based on (2S,5S)-hexane-2,5-diols show lower catalytic activity than those
based on (2R,4R)-pentane-2,4-diol (Table 1).
The highest regioselectivity was obtained with trimethylsilyl and ter-
butyl substituents at the ortho position. The reaction rates decreases as the
steric bulk of the ortho substituents C(CH3)3 < Si(CH3)3 < Si(t-Bu)(CH3)2 <
Si(CH2CH3)3 increases. The bulky groups Si(t-Bu)(CH3)2 and Si(CH 2CH3)3
5. Asymmetric hydroformylation 113
dont improve the enantiomeric excess. The steric bulk in the ortho positions
is optimal with the trimethylsilyl substituent, 9a [ 18].
An interesting connection is found between the enantiomeric excess and
the structure of the diphosphite. Enantiomeric excesses are highest for the
backbones of the diphosphites based on (2R,4R)-pentane-2,4-diol, 5 and 6,
which form eight-membered rings in the catalyst. Ees are low for the
diphosphites based on (2S,5S)-hexane-2,5-diol 7 and moderate for the
backbones of the diphosphites based on (2R,3R)-butane-2,3-diol, 4 (Table
1). In most cases when the ligands are based on (R,R) diols the (S)-aldehyde
is formed predominantly. If the configuration is inverted at the chiral carbon
atom C-2 and C-5 in the (2S,5S)-hexane-diol (R)-aldehyde is predominantly
formed [13,18].
5.2.3 Characterisation of [RhH(L)(CO)2] intermediates. Solution

structures of hydroformylation catalysts
The solution structures of the trigonal bipyramidal hydridorhodium

complexes containing diphosphite ligands, [HRh(L-L)(CO)2], which are the
resting states in the hydroformylation reaction, have been analyzed in detail
[ 13, 18-21] (see also Chapter 3).
31
P and 1H NMR spectroscopy at variable temperature, together with IR
studies, allow to establish a relation between the results from catalysis, the
coordination mode of the diphosphites to the rhodium, and the stability of
the hydride rhodium diphosphite intermediates.
Hydride rhodium diphosphites have been prepared and analyzed in situ
by adding one equivalent of diphosphite to the catalyst precursor
[Rh(acac)(CO)2] under standard hydroformylation reaction conditions.
Relatively long reaction times are required for a complete conversion into
[RhH(L-L)(CO)2], which explains the incubation times in the
hydroformylation experiments. The complexes showed broadened doublets
31 1
at room temperature in the P-{ H) NMR spectra, caused by a rhodium to
phosphorus coupling and exchange phenomena. In the case of ligands 4a and
4b, which form a seven-membered ring with the metal, the 1JRh-p coupling
1
constants are relatively small, about 210 Hz. Furthermore, both 31P and H
2
NMR spectra showed a large JP-H coupling constant, about 100 Hz, which
suggests a time-averaged cis, trans relationship between the phosphorus and
the hydrogen atoms bonded to the rhodium (see Figure 7). Large 2Jp-H.
coupling constants are characteristic of phosphorus and hydrogen atoms
2
bonded trans to rhodium [19, 22, 23, 24]. The JP-H coupling constants for a
cis relationship between the phosphorus and the hydrogen atom bonded to
the rhodium are small (< 3Hz).
114 Chapter 5
Figure 7. Ee and ea rhodium hydride dicarbonyl diphosphite
The mode of coordination depends on the length of the bridge [13, 18,
19]. For ligands 5-7 derived from 2,4-pentanediol and 2,5-hexanediol, which
1
form an eight- or nine-membered ring with the metal, the JRh-p coupling
constants are in the range 231-237 Hz. Stable hydride rhodium diphosphite
dicarbonyl complexes are formed. Typically, coupling constants of rhodium
to an equatorial phosphorus are in the range of 220-246Hz [ 13,19,25]. The
2J
P-H coupling constants are < 3Hz, which is also indicative of a cis
disposition between the phosphorus and the hydrogen atom bonded to the
rhodium (Figure 7b). The two phosphorus atoms are diastereotopic in the ee
complexes but at room temperature they are in fast exchange. Evidence for
the rapid exchange process was obtained from variable temperature NMR
experiments between 293 and 203 K. At temperatures between 293 and 253
line broadening occurred. On further cooling to 223K an ABX double
doublet appeared, indicative of two chemically inequivalent phosphorus
atoms P1 and P2 in the C1-symmetric complex.
All complexes of these diphosphite ligands showed fluxional behavior on
the NMR time scale, although the exchange parameters depended on the
mode of coordination (ea or ee). Ligand exchange in trigonal bipyramidal
HML4 complexes has been explained by so-called Berry-type rotations [26],
in one step two apical ligands exchange positions with two equatorial
ligands. However, a Berry mechanism is unlikely to be operative in RhH(L-
L)(CO)2 complexes containing bidentate diphosphite, since phosphorus
exchange requires two successive Berry-type interconversions via a high-
energy intermediate which has a hydride ligand equatorially coordinated to
rhodium. There are no known examples of trigonal bipyramidal HML4
complexes, which have an equatorial coordinated hydride ligand. X-ray and
infrared studies on such complexes always show the hydride in an apical
position. It is also unlikely that phosphorus-rhodium-phosphorus bite angles
can vary freely between 90 and 180 without considerable energy strain.
The low-energy turn-stile rearrangement postulated by Meakin et al. [22,
23] for monophosphites has been proposed to explain the fluxional behavior
of these complexes containing flexible diphosphite. It is believed that the
phosphorus atoms can easily exchange without changing the phosphorus-
rhodium-phosphorus bite angle appreciably.
Figure 8 shows a simultaneous bending motion of the hydride and the
carbon monoxide ligands in the hydride rhodium phosphite dicarbonyl
complexes containing equatorial-apical coordinating phosphite, (A) and the
motion of the hydride and phosphite functions for bis-equatorially

coordinating diphosphite, (B) respectively. For diphosphites complexes,
containing a bridge between the two phosphorus atoms, the last hydride
movement may not be facile, as the hydride must pass the bridge.
The energy barrier for the fluxional processes is lower when the steric
bulkiness of the ortho substituent of the ligand decreased. The lowest energy
barriers correspond to equatorial-axial phosphorus exchange in complexes
containing seven-membered chelate rings.
Figure 8. (A) Equatorial-axial phosphorus exchange. (B) Equatorial-equatorial phosphorus

exchange
This seems logical since this process involves the movement of two
carbonyl ligands, which should be easier than the movement of the two
phosphite moieties that is required for equatorial-equatorial exchange.
Furthermore, the energy barrier for phosphorus-phosphorus exchange is
lower for the nine-membered bis-equatorial chelate rings than for the eight-
membered ones, as the former are more flexible owing to the reduced steric
congestion.
5.2.4 Structure versus stability and enantioselectivity
Both the structure and stability of the hydridorhodium dicarbonyl

diphosphite catalysts play a role in asymmetric induction. As described
above, short-bridged diphosphites forming seven-member rings coordinate
in an equatorial-apical fashion, while larger bidentate ligands forming eight-
or nine-member chelate rings show preferentially bis-equatorial
116 Chapter 5
coordination. As well as differences in coordination of the diphosphites to

the rhodium, there are also differences in the stability of
[RhH(diphosphite)(CO)2] complexes. The complexes formed with ligands
coordinating in a seven or nine membered ring are rather unstable. The IR
spectra of the characteristic orange solution in toluene often show the
presence of [Rh4(CO)12] and carbonyl-bridged dirhodium complexes
containing diphosphite. Signals also appear in the 31P NMR spectrum, which
may be ascribed to the hydrolysis of the diphosphite ligand (see Chapter 9).
The complexes containing ligands that form an eight-membered chelate ring
give yeIlow solutions in toluene and the IR and NMR spectra show that
decomposition is considerably less than in the complexes containing ligands
forming seven or nine membered rings. IR absorptions of rhodium-carbonyl
cluster are practically absent. Enantiomeric excesses are highest with the
relatively stable [RhH(diphosphite)(CO)2] complexes in which the
phosphorus donor atoms coordinate both equatorially to the rhodium,
forming an eight-membered ring. Low enantiomeric excesses are obtained
with the rather unstable [RhH(diphosphite)(CO)2] complexes which contain
a nine-membered ring having bis-equatorial phosphites (20%), or those
containing seven-membered rings and axial-equatorial phosphite
coordination [ 13].
It should be noted that coordination in an equatorial-axial manner leads
to two intermediate species b and c, having effectively opposite absolute
configurations (see Figure 9), thus explaining the low enantioselectivities
[13, 16]. As we will see later when discussing BINAPHOS, this is not a
general rule for enantioselectivity.
Figure 9. The intermediates for equatorial-equatorial and equatorial-apical coordination
5.2.5 Chiral cooperativity and effect of terminal substituents in

diastereomeric diphosphite ligands
Diastereomeric diphosphites based on homochiral pentane-2,4-diol ligand

backbones substituted with bulky 2,2-bisphenols and bulky, 2,2-bisnaphtol
with different configurations 8-13 have been prepared to study the effect of
chiral cooperativity on the asymmetric hydroformylation of styrene (see
Figures 4 and 9). Stable diastereomeric diphosphites were obtained in a pure

form using the bisnaphthol derivatives [ 18].
The absolute configuration of the diastereomers is given as (S,2R,4R,S)
(S,2S,4S,S), (R,2R,4R,R), (R,2R,4R,S) 10-13 (see Figure 10). The indicators
S and R refer to the absolute configuration of the chiral axis, while the
indicators, 2R, 4R and 2S, 4s refer to the absolute configuration of the
carbon atoms C-2 and C-4 in the pentane-2,4-diol backbone. The structures
with the configurations (S,2S,4S,S) and (R,2R,4R,R) are mirror images.
10: (S, 2R, 4R, S)

11: (S, 2S, 4S, S)
12: (R, 2R, 4R, R)
13: (R, 2R, 4R, S)
Figure 10. Diastereomeric diphosphite pentane-2,4-diol derivatives
The fact that the diphosphite ligands behave differently in the formation
of the RhH(diphosphite)(CO)2 complexes demonstrates that both the
absolute configuration of the 2,4-pentanediol ligand backbone and the chiral
bisnaphthol substituents determine the stability and catalytic of the rhodium
complexes. The ligands with configuration (S,2S,4S,S) 11 and (R,2R,4R,R)
12, lead to undefined mixtures of complexes and ligand decomposition.
However the ligands (S,2R,4R,S) 8, 10, and (R,2R,4R,S) 13 form well
defined, stable complexes. When there was an excess of the statistical
mixture of diastereomers 9, only the complexes derived from 10 and 13
formed. When both the pentane backbone and the bisnaphthol substituents
have all S (11) or all R (12) configurations the ee is low. In situ NMR studies
show that the [RhH(CO)2(L-L)] complexes of these ligands cannot be
synthesized selectively.
The low enantioselectivity and relatively high reaction rates observed
have been attributed to the coexistence of highly active rhodium species in
which the ligands coordinate as monodentates. In contrast, the ligands with
different configurations at the pentane bridge backbone and such bisnaphthol
substituents, as ligands 10 and 13, both form stable rhodium complexes, but
only diastereomer 10 gives a high ee (Table 2). These results have been
118 Chapter 5
attributed to an effect known as chiral cooperativity between the central

backbone and the biaryl substituents.
Curiously, the results of rigid 10 are the same as for 8a, the bisphenol
derivative in which the ligand can adopt freely several diasteromeric
conformations. These results may indicate that the bisphenol derivative leads
predominantly to the (S,2R,4R,S) conformation in the
[RhH(CO)2(diphosphite)] complex, as can indeed also be inferred from a
comparison of the low temperature 31P NMR spectra.
Table 2. Hydroformylation of styrene with RhH(diphosphite)(CO)2

Diphosphite T Turnover Conversionc % 2-phenyl- % ee (Absolute
C frequencyb % propanal Configuration)
8a 40 45 21 89 67 (S)
8a 25 9 26d 93 87 (S)
9a 25 14 20 91 47 (S)
10 50 133 43 83 58 (S)
10 25 17 3 8d 88 69 (S)
10 15 11 12d 92 86 (S)
11 40 259 89 91 18 (R)
11 25 45 21 94 40 (R)
12 25 28 18 95 38 (S)
13 25 4 2 91 23 (S)
Styrene:catalyst molar ratio is 1000. CO/H2 = 20 bar. P:Rh molar ratio is 2.2. b TOF in mol
-1 c d
styrene.(mol RK-1)h determined after 1 h by GC. Of styrene after 5 h. After 24 h of
reaction.
The ee is highest, 86%, when the diastereomer 10 is used and it is in the

same range as for 8, which has rotational freedom around the biaryl axis.
When the diphosphite ligands are bisnaphthol derivatives 9a and 10, the
three diastereomers of 9a resulted in low asymmetric inductions in
comparison with the single isomer 10.
The formation of the [RhH(CO)2(diphosphite)] complexes using one
equivalent of ligand 9a results in the formation of a statistical mixture of the
diastereoisomeric complexes 10, 12 and 13. The turnover frequency in the
hydroformylation reaction corresponds to the average of this statistical
mixture [18]. Also, the observed enantiomeric excess corresponds to the
average value, when the different reaction rates of the diastereomeric
31
complexes are considered. P NMR analysis of the rhodium complexes
derived from the ligand 9a showed that the most stable complex is formed
by ligand 10, the one that also induces the highest enantioselectivity. Using
an excess of 9a, a mixture of rhodium hydride complexes derived only from
10 and some 13 is formed. This explains the increased ee when an excess of
9a is used.
Figure 11. Diphosphite ligands containing chiral chelate backbones and chiral
dioxaphosphorinane moieties
Kollar et al. prepared diastereomers of diphosphites which were

synthesized by combining different modifications in the bridge and the
phosphorus substituents [27, 28]1. These ligands lead to active rhodium
systems, and are considerably affected by chiral cooperativity. Some of these
diastereomers prepared from (2S,4S) 2,4-pentenediol and (1 S,3S) diphenyl
propane 1,3-diol form rhodium complexes in which the ligand is coordinated
in an eight-membered ring, 14, 15, but the enantioselectivity obtained in the
experiments described are no higher than 24% [27].
Diastereomers derived from atropisomeric backbones, such as the
bisnaphthol derivatives, form nine-membered rings 16-19 (see Figure 12).
The diphosphites 17 and 19 are sterically less demanding than the
diphosphites 16 and 18. The diastereomers differ only in the axial
configuration in the backbone; the configurations in the terminal groups are
the same. The results obtained in asymmetric hydroformylation of styrene
suggest that catalysts containing 16 and 18 have a higher preference for the
formation of the branched alkylrhodium intermediate than those containing
17 and 19, which is probably due to their different steric constraints. The
enantioselectivities were always low or moderate, never exceeding 24% [28,
29]. The systems having the same absolute configuration in the chelate
backbone and the terminal groups, R-bis(4R,6R) 17b, and S-bis(S) 19a,
provide slightly higher enantiomeric excesses than the corresponding S-
bis(4R,6R), and R-bis(S).
120 Chapter 5
Figure 12. Chiral diphosphites forming nine membered ring coordinated to the metal
Systematically varying the chirality at both the chelate backbone and the
terminal groups affected the enantioselectivity of the catalyst. The axial
chirality on the bridge determines the product configuration with a
cooperative effect from the terminal groups. When both the bisnaphthol
backbone and the terminal dioxaphosphepine units have the same chiral
denominator, ee's are lower than when they have different configurations, as
can be seen from the systems discussed above [18, 28].
Chiral diphosphite ligands containing a spiro backbone in the bridge and
forming a complex with an eight-membered ring have also been used as
catalysts in the asymmetric hydroformylation of styrene and derivatives
(Figure 13) [30]. At 10 bar of CO/H2 and 40 C regioselectivity to 2-
phenylpropanal is 97% and enantioselectivity 69% using the (1 S,5S,6R)-
(cis,trans)-diol phosphite shown.
Figure 13. Chiral diphosphites derived from spiro[4,4]nonane- 1,6-diol
Decreasing the temperature from 40 to 25 C led to a slight increase of

enantiomeric excess at the cost of a much lower conversion. Diphosphites
with the same spiro backbone, and 1 ,1'-bisnaphthol substituents of different
chirality, afforded the opposite prevailing enantiomer in the product. This
shows that the terminal bisnaphthol groups dictate enantioselection. The
catalytic efficacy of these diphosphite systems is mainly controlled by the
large bulky substituents. No cooperative effect between central and axial
chirality was observed in this case [30].
5.2.6 C1 Sugar backbone derivatives. Diphosphinite and diphosphite

ligands
Sugar derivatives have been used as ligands in homogeneous catalysis and

are among the most easily accessible chiral ligands. They have provided
high enantiomeric excesses for several asymmetric processes. Their
versatility and the presence of several stereogenic centers allow the
syntheses of several series of ligands [31]. C2 chiral ligands are extensively
used in asymmetric homogeneous catalysis and can be considered the best
choice, because their symmetry diminishes the number of diastereomeric
transition states, but C1 chiral ligands have proved to be of value in
enantioselective catalysis as well. C1 symmetric ligands can cause
stereochemical restrictions because of such electronic effects as the trans-
influence, differentiation in the metal environmental and enantioface
discrimination towards the coordination of the prochiral substrate. C1 ligands
also cause differences in diastereomeric transition states [ 15, 16]. In the
rhodium-catalyzed hydroformylation of styrene and related substrates, the
first carbohydrate ligands were tetrahalose derivative diphosphines. When
these systems were used under ambient conditions regioselectivities were
high for branched aldehydes, but zero ee's were obtained [32]. Later, Rajan
Babu applied in asymmetric hydroformylation diphosphinite ligands, which
had previously been used in asymmetric hydrocyanation [33].
122 Chapter 5
Figure 14. Diphosphinite carbohydrate derivatives
Vicinal carbohydrate diphosphinites 20 provide high regioselectivities for

the branched aldehyde, but the enantioselectivities are moderate. In the case
of 6-methoxy-2-vinylnaphthalene, however, the ee increased from 10% to
38% when electron deficient aryl groups containing two fluorine atoms or
two trifluoromethyl substituents are introduced on the aryl groups (20c-d).
The enantioselectivity further improves to 5 1 and 72% when apolar solvents
like hexane and triethylsilane respectively are used. These ligands are quite
sensitive to the alkene structure and lead to large differences in
enantioselectivity for similar substrates. The highest ee for 2-
vinylnaphthalene does not exceed 30% and asymmetric induction was even
lower for other alkenes such as styrene and vinyl acetate [34].
Diphosphites derived from 1,2-O-isopropylidene-a-xylohranose and the
corresponding ribose derivatives, differently modified are effective C1 chiral
ligands in the rhodium hydroformylation of styrene. As described for C2
diphosphites in the section 5.2.2, the diphosphite ligands based on 1,3-diols
give rise to hydridorhodium diphosphite hydroformylation catalysts, which
are relatively stable in comparison to those based on 1,2 and 1,4-diols, and
enantioselectivi ties are higher. 1,2-O-isopropylidene--xylofuranose, which
has a three-carbon atom bridge between the hydroxy groups, is an
appropriate starting material for a series of useful diphosphites [35].
Figure 15. 1,2-O-isopropylidcne-D-xylofuranose 21 a and ribofuranose 21b,

Because the primary hydroxy at C-5 is more reactive towards

phosphorochloridites than the secondary hydroxy group at C-3, different
substituents can be introduced into the molecule. Using bisphenol
phosphorochloridites having various steric bulk at the ortho and para
positions, several xylose and ribose derivative ligands were synthesized and
used in rhodium asymmetric hydroformylation [35,36].
Figure 16. Xylose (22) and ribose (23) diphosphite derivatives
Regioselectivities, as high as 97% to the branched aldehydes were found

under relatively mild conditions, T= 25-40 C, 9-45 bar of CO/H2 pressure.
As expected, the enantioselectivity depends on the substitution of the ligand.
The highest enantioselectivity was found for the ligands having bulky tert-
butyl substituents at the ortho position of the bisphenol moieties.
Enantioselectivities up to 64% have been obtained with these substituted
ligands [35]. Ligands based on 1,2-O-isopropylidene--xylofuranose gave
(S)-aldehydes, while the related ligands with a ribofuranose backbone lead to
the (R)-aldehydes with similar rates and regioselectivities and
enantioselectivities of the same magnitude for each enantiomer. The
inversion of the asymmetric has been rationalized by the proposal that the
absolute configuration at the stereogenic carbon atom C-3 produces two
diastereomeric rhodium complexes that are effectively mirror images [3 6].
The solution structures for bisphenol derivatives are consistent with the
presence of two diastereomers for the [RhH(diphosphite)(CO)2]
intermediates, diphosphites of which coordinate bisequatorially to the
rhodium in a fast exchange on the NMR time scale [36].
Selke et al. prepared sugar diphosphites derived from glucopyranose.
They had different cyclic substituents on the phosphorus atoms and they
124 Chapter 5
were tested in the rhodium hydroformylation of vinyl acetate, allyl acetate

and p -methoxystyrene. The highest ee obtained for allyl acetate is 32% and
for p-methoxystyrene 20%. The conclusions are in accord with the general
features reported by van Leeuwen for diphosphites derived from vicinal
diols [17, 37].
5.3 Phosphine-phosphite rhodium catalysts
5.3.1 Introduction
In 1992, Takaya et al. reported on the use of a chiral diphosphite derived

from bisnaphthol in the asymmetric hydroformylation of vinyl acetate [ 11],
but the enantioselectivity achieved was only 50%. They noted that
diphosphites led to more stable hydroformylation catalysts than
diphosphines. This observation prompted Takaya, Nozaki et al. to synthesize
the chiral phosphine-phosphite ligands (R,S)- and (R,R)-BINAPHOS (24,
figure 17), which were expected to combine the high enantioselectivity
obtained with diphosphines such as BINAP in asymmetric hydrogenation,
with the apparently efficient coordination of the phosphite moiety [38].
Indeed, the Rh(I) complex of C1-symmetric (R,S)-BINAPHOS provided
much higher enantioselectivities than either C2 symmetric diphosphine
ligands or diphosphite ligands, viz. more than 90% ee for a wide variety of
both functionalized and internal alkenes [38, 39, 40, 41].
Ever since the successful results obtained for (R,S)-BINAPHOS, the
preparation of other unsymmetrical ligands to be used in the asymmetric
hydroformylation has become a recurrent theme in the literature [3, 4, 36,
42]. The following section will discuss those ligands, which are structurally
related to BINAPHOS, while other phosphine-phosphite systems will be
discussed separately in section 5.3.5.
5.3.2 Rhodium complexes with BINAPHOS and related ligands.
The chirality and the steric properties of BINAPHOS were varied

systematically by synthesizing of 24-27 shown in Figures 17 and 18 [14].
The flexible ligand (R)-26 derived from bisphenol provides a single species,
for which the NMR data indicate that there is a fast exchange between the
two diastereomers (R,S) and (R,R) via a rotation around the bisphenol axis.
(S,R) -28a and (R,R) -28b were obtained as a mixture of two diastereomers,
owing to the rotation around the biphenyl axis. According to the NMR
spectra a 55:45 mixture (S,R vs. R,R) is obtained that cannot be separated by
chromatography. This suggests that the exchange is too fast at room

temperature on the laboratory time scale, but slow on the NMR time scale.
The first R/S indicator refers to the binaphthyl (biphenyl) bridge and the
second to the bisnaphthol (bisphenol) moiety. Thus for 26, which lacks the
bisphenol chirality, we write in the table (R,--), and for 28, which contains a
non-chiral biphenyl bridge, we write (--,R).
26 (R) 27a (S,R)

27b (R,R)
Figure 17. (R,S) BINAPHOS and related ligands

126 Chapter 5
Figure 18. (S,R) and (R,R) equilibrium
The [Rh(acac)(phosphine-phosphite] complexes were used as precursors

for the asymmetric hydroformylation of styrene after adding 3 equivalents of
free ligand. The results are shown in Table 3. The ee of the product
decreases with increasing temperature. The ratio of CO to H2 does not affect
the regio- and enantioselectivity.
For all ligands the absolute configuration of the product is the same as
that of the absolute configuration of the binaphthyl (biphenyl) bridge. Thus,
it seems that the binaphthyl bridge directs the position of the
diphenylphosphino fragment. Remarkably, the best combination is formed
when the absolute configurations for the binaphthyl and bisnaphthol moieties
are opposites (i.e. S,R or R,S diastereomers). Enantioselectivities were much
lower with (R,R)-ligands. R,R diastereomer 24 still gives the R-aldehyde but
the ee is low (26%).
Table 3. Asymmetric hydroformylation of styrene using ligands 24-28
Ligand TC CO/H2 Time h Conversion b/1 % ee

bar/bar % (Absolute
configuration)
24(S,R) 60 50/50 43 >99 88/12 94 (S)
24 (R,S) 80 50/50 16 >99 86/14 89 (R)
24(R,S) 60 10/90 40 >99 88/12 92 (R)
24(R,R) 60 50/50 3s >99 86114 25 (R)
25(R,S) 60 50/50 37 >99 90/10 85 (R)
26 (R,--) 60 50/50 43 >99 91/9 83 (R)
27(S,R) 60 50/5 0 42 >99 90/10 94 (S)
27(R,R) 60 50150 40 95 92/8 16 (R)
28 (--,R) 60 50/50 40 98 89/11 69 (S)
Styrene 20 mmols in benzene. [Rh] = 0.010mmol. L = 0.040 mmol. The ees were determined
by GLC analysis of the corresponding 2-arylpropionic acids derived by Jones oxidation of the
products
The results obtained for ligands 26 and 28, which contain only one fixed
stereocenter, are interesting and very infomative about the system. Ligand
26-(R,--), in which only the binaphthyl bridge has a predetermined absolute
configuration R, leads to an ee of 83% (R-aldehyde), which is quite close to
the value of 94% for (R,S)-BINAPHOS. This suggests that in the formation
of the complex the binaphthyl bridge controls the conformation of the
bisphenol moiety so that the final conformation is (R,S). Likewise, 28 -(--,R)
gives 69% ee of the S-aldehyde, which suggests that the complex formed
assumes an (S,R) conformation, since now the R-bisnaphthol induces the
formation of an S-diphenyl bridge. The control by the binaphthyl bridge in
26 is somewhat more efficient than that of the bisnaphthol moiety in 28.
In the free ligand 28 we saw a slight preference for the S,R diastereomer,
but in the complexes there is a strong preference for the S,R or R,S
conformation. The NMR spectra of the complexes of the two non-rigid
ligands 26 and 28 led to this conclusion but other data also support this
observation. In the free ligands with R,S configurations the through-space
coupling constants between the two phosphorus nuclei are always larger than
those in R,R ligands. This shows that in R,S ligands the ligand has a natural
tendency to bring the donor atoms into close proximity. Secondly, in
competition experiments between R,R and R,S ligands the latter give the
most stable complexes and are prevalently formed in the mixtures. In
diphosphites, the bisphenol configurations were also observed to be
controlled by the backbone and the most stable configurations lead to the
highest ee's in hydroformylation catalysis [4, 14, 18, 38].
5.3.3 [RhH(CO)2(BINAPHOS)] complexes; models for

enantioselectivity
Thus, the sense of the enantioselection is predominantly controlled by the

phosphine moiety and opposite configurations of the binaphthyl and
bisphenol moieties give the highest ee. Complexes [RhH(CO)2(L-L)] (L-L=
24 -(R,S), 24 -(R,R), 26 -(R,--), 27a -(S,R), 27b -(R,R) have been characterized
by 31P and 1H NMR spectroscopy. In all cases a single species was observed
in which the phosphine occupies an equatorial position and the phosphite an
apical position [43]. The NMR data indicate that the hydride is oriented cis
to phosphine and trans to phosphite. The 2JP-H coupling constant (phosphite -
hydride) is ~160 Hz which corresponds to an apical hydride trans to
phosphite in a trigonal bipyramidal rhodium complex [ 14,241, except for 24-
(R,R) for which it is smaller. All data point to a high preference for a
coordination mode with equatorial phosphine and apical phosphite. This is
unexpected, since for a d8 electron configuration one would expect the
strongest acceptor ligand to be in the equatorial plane. As yet there is no
explanation for this phenomenon.
128 Chapter 5
Figure 19. RhH(CO)2(BINAPHOS) complex
The spectroscopic data of phosphine-phosphite complexes show slight

differences between the (R,S) and (S,R) complexes on the one hand and the
(R,R) and (S,S) complexes on the other. In the latter compounds the data are
less clear-cut, which could be due to structural deviations of the
monohydride complexes from an ideal trigonal bipyramid or to equilibria
between isomers. The two ea isomers cannot be distinguished by IR
spectroscopy as their absorptions tend to coincide [49].
31
P NMR spectroscopy shows that the complex [RhH(CO)2(R,S -24)] is a
single species at 60 C and 25 C in toluene-d8 under 1 bar of CO. No
apical-equatorial interchange was observed at any temperature. The unique
dissymmetric environment in a single catalytically active species created by
the phosphine-phosphite seems to be an important factor in the high
enantioselectivity obtained.
Mechanistic studies have been made of the asymmetric hydrofomylation
of the BINAPHOS system [4, 44]. Deuterioformylation experiments
(Chapter 2) have been used to study the alkene insertion step [45]. Studies
using [RhH(CO)2(R,S-BINAPHOS)] indicate that the alkene-insertion is
irreversible for styrene at CO/H2 total pressures between 20 and 100 bar, but
the insertion is partially irreversible at a total pressure of 1 bar [44].
The models which explain the enantiofacial selection in the Rh-
BINAPHOS systems assume that the alkene insertion in the Rh-H bond is
irreversible. The old model proposed by Consiglio and Pino [46] has been
used by Takaya [ 14]. The model involves partitioning the space around the
metal center into four quadrants and for each of these the steric encumbrance
is estimated. This simple model differentiates between the enantiofaces of
the alkene and it has often been successfully applied. For RhH(CO)2(R,S-
BINAPHOS) the aldehyde enantiomer expected from this model was indeed
obtained [ 15].
While the simplicity of this model is attractive, the many studies on
asymmetric hydrogenation have shown that such models can be deceptive
and that more advanced studies are needed. Gleich et al used a semi-
quantitative theoretical model that combined quantum mechanics and
molecular mechanics. Their findings matched previously reported
experimental data; the aldehyde found experimentally had the lowest barrier
of formation and the (R,R)-BINAPHOS complex should give much lower
ees than the (R,S)-BINAPHOS complex [14, 15].
5.3.4 Separation studies for the BINAPHOS system
The importance of recovering of the catalyst used in hydroformylation is

described in chapters 7 and 8. Chiral phosphine-phosphite Rh(I) complexes
have been immobilized on polystyrene matrices so that they can be used in
the hydroformylation of alkenes. After introducing a vinyl group into the
phosphine-phosphite 24, the resulting ligand can be copolymerized with
styrene derivatives to provide a highly cross-linked polymer matrix. The
high levels of catalytic activity and selectivity obtained in the homogeneous
hydroformylation were reproduced in the immobilized system [47].
The BINAPHOS system has also been modified by using the
perfluoroalkyl-substituted derivative of (R,S)-BINAPHOS (see Chapter 10)
in the rhodium catalyzed hydroformylation of vinyl arenes in liquid or
supercritical carbon dioxide as solvent This system leads to similar catalytic
activity and the same level of enantiocontrol as the rhodium-(R,S)-
BINAPHOS system in organic solvents [48].
5.3.5 Chiral Phosphine-Phosphite Ligands containing a stereocenter

in the backbone
Van Leeuwen et al designed a series of phosphine-phosphite ligands that

contain a phosphine with a stereogenic phosphorus atom, a biaryl phosphite
and a backbone with a stereocenter generated from chiral epoxides. The
moieties combine to form diastereomers 29 and 30 (Figure 20).
Systematically varying the ligands shows how the stereocenters affect the
enantioselectivity. The substituents on the chiral phosphine are in close
vicinity to the metal. The stereocenter in the backbone is close to the
phosphite moiety.
The behavior of the substituted bisphenol 29 and bisnaphthol 30
derivatives show what influence the biaryl moiety has. Complexes
[RhH(CO)2(phosphine-phosphite)] have been used as catalysts in the
asymnietric hydroformylation of styrene.
130 Chapter 5
29a (Sp ,Sc) R1=l-Naphthyl, R2=Ph 30a (Sp,Sc,R) R1=l-Naphthyl, R2=Ph

29b (Sp ,Sc) R1=2-Anisyl, R2=Ph 30b (Sp,Sc,S) R1=l-Naphthyl, R2=Ph
29c (Sp,Rc) R1=l-Naphthyl, R2=Ph
29d (--,Sc) R1=Ph1 R2=Ph
Figure 20. Examples of phosphine-phosphite ligands
Table 4. Asymmetric hydroformylation of styrene catalyzed by Rh/phosphine-phosphite

ligands
Ligand Conversion % (h) b/l a TOF b ee%c

29a SpSc 55 (24) 11 45 63 (s)
29b SpSc 24 (25) 20 18 57 (S)
29c SpRc 7 (6) - 24 33 (R)
29d --Sc 69 (25) 2 56 18 (S)
30a SpScR 95 (18) 9 111 40 (S)
30b SpScS 26 (19) 8 25 4 (S)
a
Styrene:Rh= 2000. Ligand to rhodium ratio is 4. T C = 50. Branched to linear aldehydes
ratio.b TOF in mol styrene (mol Rh)-1h-1 determined at the end of the reaction time by GC.
c
Enantiomeric excess. Absolute configuration is given in parentheses.
The results in Table 4 show regioselectivities for the branched aldehyde

larger than 92% and ees as high as 63% at 20 bar and 50 oC. A ligand to
rhodium ratio of ~4 is required to obtain enantiomeric excesses. The results
of the hydroformylation reaction using bisphenol and bisnaphthol derivatives
show the different effects of the substituents in the backbone, the
substituents at the phosphine, and the phosphite moieties. A phosphine with
a stereogenic phosphorus atom must be present if enantioselectivity is to be
high. The 1-naphthyl group of ligand 29a and the o-anisyl group of ligand
29b both have a considerable effect on the enantioselectivity.
The configuration of the stereocenter in the backbone controls the
absolute configuration of the major enantiomer, although the enantiomeric
excess for 29d containing the phenyl group in the bridge is small. The
configuration of the bisnaphthol moiety in the phosphite has also an
important influence on the ee. The diastereomer containing the (S)-
bisnaphthol moiety 30b provides only 4% ee, while ligand 30a with a (R)-
bisnaphthol moiety results in an ee of 40%. The results obtained with the
ligands 29a, 30b, 30a, show that ligand 29a leads predominantly to the same
conformation as ligand 30a. The conformation of the bisphenol moiety at the

phosphite is controlled by the substituent at the stereocenter in the backbone,
which also controls the configuration of the product.
The coordination mode of the ligands in the [RhH(CO)2(phosphine-
phosphite)] complexes has been studied using IR and NMR spectroscopy
under 20 bar of CO/H2. The HP-IR spectrum shows the presence of a single
species and HP-NMR spectroscopy shows that the phosphine and the
phosphite occupy an apical and equatorial position respectively, as would
have been expected from the donor and acceptor properties. This is in
contrast with the coordination mode found for the [RhH(CO)2(phosphine-
phosphite)] complex of BINAPHOS in which the phosphine occupies an
equatorial position and the phosphite an apical position [49].
The results of hydroformylation show that both the configuration of the
phosphite moiety, coordinated at an equatorial position, and the chiral
phosphine at the apical position are important for obtaining high
enantioselectivity. The experiments done using a ligand containing a
stereogenic bisnaphthol group shows a cooperative effect of the bisnaphthol
moiety and the group attached to the stereocenter in the backbone [49]. In
the case of the BINAPHOS catalyst, the equatorially coordinating phosphine
(or rather the binaphthyl bridge) has a slightly larger influence on the ee than
the apically coordinating phosphite, although the results show that a strict
separation is impossible.
5.4 Diphosphine rhodium catalysts
5.4.1 Introduction
Diphosphines are among the most efficient chiral ligands in

homogeneous catalysis. Their characteristics and how they are used as
ligands in the hydroformylation reaction are described in chapter 4. Many
chiral diphosphines have been used as ligands in rhodium systems for
asymmetric hydroformylation. Despite all the effort made to apply
diphosphines in asymmetric hydroformylation, the enantioselectivities of
rhodium-diphosphine systems are not as high those of rhodium-diphosphite
or rhodium phosphine-phosphite systems.
5.4.2 C1 diphosphines as chiral ligands
C1 diphosphines have been much less extensively studied than C 2 chiral

diphosphines. As an example of a C1 diphosphine we mention BPPM, 31,
132 Chapter 5
which had previously shown a high activity and enantioselectivity in Pt/Sn

based hydroformylation catalysts giving ees up to 96%, albeit with a low
chemo- and regioselectivity, characteristic of the Pt/Sn systems. In the
catalytic systems based on rhodium, BPPM was used in the asymmetric
hydroformylation of several substrates such as N-vinylphthalimide,
tricarbonylchromium complexes and arylalkenes, but the ees were always
low at 10-14% [2-4].
31aBPPM R=Ph 32 EPHOS

31b BPPM-DBP R2= DBP
Figure 21. C1 chiral diphosphines
Mortreux et al. prepared diphosphine 32, EPHOS, derived from (S,R)-

ephedrine. This phosphinite-phosphors amide was applied in the
hydroformylation of styrene in the presence of a rhodium chloride precursor
as well as an electrochemically generated rhodium catalyst. The
enantioselectivity increased to 31 % (96% regioselectivity) when
electrochemical reduction was applied. This is the first non-phosphine
system in which the structures of the resting state of the catalyst were
1 31
investigated under 1 bar of CO/H2. H and P NMR data show that there is a
major hydrido rhodium complex [RhH(CO)2(diphosphine)] where the ligand
coordinates in an ea mode, in equilibrium with the [Rh(-CO)(CO)(S,R-
EPHOS)]2 dinuclear complex [50]. A bidentate ligand such as EPHOS,
having two very distinct absorptions for the amido phosphorus and the
phosphinite phosphorus nuclei gives NMR spectra that are readily
interpreted. The hydride resonance shows a large coupling constant (116 Hz)
with the amidite phosphorus atom in trans position and smaller coupling
constants with rhodium (10.1 Hz) and the phosphinite phosphorus nucleus
(9.9 Hz). The phosphorus spectra are in accord with this.
5.4.3 C2 diphosphines as chiral ligands
Many chiral C2 diphosphines have been used as ligands in rhodium

systems for asymmetric hydroformylation. It can be observed that the chiral
diphosphines providing the highest enantioselectivities in asymmetric
hydrogenation lead to low chiral induction in hydroformylation. The C2-
symmetric diphosphines, which were developed for square planar
coordination in hydrogenation catalysts, will coordinate in an apical-

equatorial manner. The intermediate four-coordinate species can undergo an
approach from the alkene at either side of the complex. As suggested for
phosphites [10], these two sides of the complex may lead to opposite,
enantiospecific binding modes of the alkene. The scheme presented in 5.2.4
for diphosphites illustrates this.
DIOP and related ligands. The first reports on asymmetric
hydroformylation using DIOP as an asymmetric diphosphine date back to
1973 [51]. Throughout the seventies and eighties platinum and rhodium
systems with DIOP were extensively studied. Several modifications have
been reported, as for instance the introduction of different substituents at the
phosphorus atoms, the application to different substrates, the study of the
reaction conditions. DIOP and several DIOP type ligands were used in the
asymmetric hydro formylation of unsaturated nitrogen compounds, aliphatic
alkenes, N-acyl-1 -aminoacrylic acid derivatives, unsaturated carboxylic
acids and esters (see Figure 23). In the hydroformylation of alkyl -N-
acylaminoacrylates using rhodium complexes of DIOP, 33, and a related
ligand DIOCOL, 34, the reaction takes place with complete selectivity for
the branched aldehyde and an ee of 60%.
It has been said that the regiospecific addition of carbon monoxide to the
most substituted carbon is due to the polarization of the double bond and to
the acylamide groups chelation control [52]. Although the enantioselectivity
is not exceptional, it is higher than normal in the rhodium hydroformylation
of vinylidenic alkenes.
Figure 22. DIOP derivatives and DIOCOL diphosphines

134 Chapter 5
[Rh]
CO2Me diphosphine
NHCOMe CO/H2
Figure 23. Hydroformylation of acrylic acid derivatives
Many C2 diphosphines have been used in the asymmetric

hydroformylation reaction. The enantioselectivities, however, are always
low or moderate, generally no higher than 30% [2-4].
Diphosphines with small bite angles. BDPP and related diphosphines.

In 1996 a rhodium system was described which provided ees as high as
60% and which had a regioselectivity for 2-phenylpropanal of 95% despite
having a diphosphine as simple as BDPP, 35 (see Figure 24). The ligand to
rhodium ratio had to be 2 and the temperature as low as 65 0C to obtain this
enantioselectivity [53]. Other diphosphines such as DIOP, 33, BINAP, or
CHIRAPHOS, 36, used under the same conditions provide lower ees (12-
25%) [53] (see Table 5). Diphosphines such as DIOP or BINAP form seven
membered rings when coordinating to rhodium. On the basis of previous
findings described in chapter 4, one might consider the formation of both
bis-equatorial and equatorial-apical intermediates, especially for the more
flexible DIOP. As outlined in sections 5.2 and 5.3 the occurrence of a variety
of species does not seem to be a good starting point for obtaining high ees.
On the other hand, CHIRAPHOS or BDPP with smaller bite angles between
80-90 [54] probably form equatorial-apical bipyramidal intermediates only,
which limits the number of possible structures. In section 5.4.4 it will be
shown that BDPP indeed forms only ae species [55]. It was also argued in
5.2 that ae coordination leads to four-coordinate intermediates, the two faces
of which may be effectively mirror images leading to opposite enantiomeric
products.
35 BDPP 36 CHIRAPHOS 37 DEGUPHOS
Figure 24. BDPP, CHIRAPHOS and DEGUPHOS
For the rhodium catalytic systems containing BDPP or CHIRAPHOS the

regioselectivities depend in the same way on pressure and temperature. The
amount of branched aldehyde increases when the pressure is raised and/or
when the temperature decreases. The same behavior had previously been
observed in the hydroformylation of styrene with the unmodified rhodium
catalyst (chapter 2) [45, 56].
Table 5. Asymmetric hydroformylation of styrene with C2 chiral diphosphines
Ligand P (bar) T C L/Rha Conversion b/l ee % c

b%
33 (SS) 10 65 2 94 61/39 12 (S)
d
35 (S,S) 8 65 2 33 95/5 43 (S)
35 (S,S) 8 80 1 76 74/26 7 (S)
35 (S,S) 8 80 2 76 79/21 51 (S)
35 (S,S) 30 80 2 45 94/6 58 (S)
36 (R,R) 8 80 1 98 85/15 15 (R)
36 (R,R) 8 80 2 99 97/3 25 (R)
36 (R,R) 30 80 2 100 95/5 21 (R)
d
37 (R,R) 8 65 2 97 94/6 >2 (S)
a
10 mmol of styrene and 0.0125 mmol of [Rh2(-(OMe)2(cod)2] in toluene Ligand to
c d
rhodium ratio. Enantiomeric excess by GC. Absolute configuration is given in parentheses.
Time is 22 h.
The enantioselectivities for the system Rh/CHIRAPHOS are in the range

of 15-28% ee and there was not significant change in the enantioselectivity
when the ligand to rhodium molar ratio was increased. However, the ees of
the catalytic system with BDPP the ee depends on the ligand/rhodium molar
ratio, and can vary from 10% at a diphosphine-rhodium ratio of 1 until 58%
for a diphosphine/rhodium ratio of 2 [53, 56].
A related bis(diphenylphosphino)pyrrolidine ligand, DEGUPHOS 37, has
been used under the same conditions as those for BDPP and CHIRAPHOS.
The enantioselectivity however was nearly null. An NMR study showed that
there were five-coordinated rhodium species, in which the ligand is
coordinated in both bridging and chelating modes. The bridging coordination
mode also involves monodentate species and is considered to be the cause of
the lack of enantioselectivity [57].
Several diaryl phosphines have been sulphonated to form a rhodium
catalytic precursor for the hydroformylation of styrene in an aqueous
solution. The enantioselectivities reported for (S,S)-BDPPTS are in the range
of 14-17% only [58].
Diphosphines having wide bite angles. BISBI and related diphosphines
Herrmann et al. reported how chiral diphosphines with wider natural bite
angles were applied in the asymmetric hydroformylation of styrene. The
diphosphines related to the BISBI ligand, 38-40 coordinate in a bis-
equatorial mode to the metal. NAPHOS 39 and a 2,2',6,6'-tetrasubstitued
derivative of BISBI named BIPHOLOPHOS 40 have been prepared and
sulphonated to be used in mono- and biphasic asymmetric hydroformylation.
136 Chapter 5
Figure 25. BISBI and related ligands
The rhodium system formed using (-)BIPHOLOPHOS provides an ee of

15% at 40 C, whereas only racemic product was detected at 80 C. The
results have been best at 100 bar and 40 "C for this ligand: the ee was 34%
with a regioselectivity in branched aldehyde of 96%. The binaphthyl
backbone in NAPHOS 39 seems to produce a better enantiodiscrimination
than the substituted biphenyl backbone [59, 60]. The NMR characterization
of the pentacoordinate [RhH(NAPHOS)(CO)2] complex show ee
coordination of these diphosphines [59].
As a continuation of an earlier study on biphasic hydroformylation [61]
Herrmann et al. prepared chiral ligands which were highly soluble in water
by direct sulfonation of the enantiopure diphosphines NAPHOS and
BIPHOLOPHOS. They then applied them in the aqueous/organic biphasic
hydroformylation of styrene [60]. The enantioselectivities obtained in the
two-phase hydroformylation of styrene are in general lower than those
obtained by the corresponding homogeneous catalysts in organic solvents.
The highest ee achieved in a two-phase hydroformylation of styrene is 18%,
using the sulphonated NAPHOS ligand [59]. This trend has been observed in
many other catalytic asymmetric reactions; application of water-soluble
catalysts normally decreases the optical induction.
5.4.4 The Rh/BDPP system. HPNMR and HPIR studies under

hydroformylation conditions.
Characterization of the intermediates. The solution structure of the

species formed during the reaction of [Rh(-OMe)(cod)]2 with (-)-BDPP
under hydroformylation conditions has been determined at 20 bar CO/H2 at
room temperature [55].
Figure 26. Mononuclear and dinuclear species in equilibrium
31
The P NMR spectrum reveals the presence of two species in
equilibrium, the [RhM(BDPP)(CO)2] hydride complex and the dimer
[Rh(BDPP)(-CO)(CO)]2. The mononuclear [RhH(BDPP)(CO)2] complex
1
shows only one doublet at 29.8 ppm ( JRh-P)=112 Hz and the 1H NMR
1 2
spectrum revealed a double triplet in the hydride region JRh-H= 11, JH-P= 57
Hz, indicative of a coupling of the hydride with the rhodium nucleus and two
degenerated phosphorus nuclei. Small cis phosphorus-hydride coupling
constants are reported for complexes with ee coordinating diphosphine
ligands [62] and relatively large phosphorus-hydride coupling constants are
characteristic of complexes containing trans P-Rh-H arrangements [22, 24].
These studies revealed that the diphosphine BBPP is equatorially-axially
coordinated, in accord with the calculated bite angle of 91 [54], and that a
fast exchange of the two phosphine groups occurs. The turn-stile
mechanism is the same as that described for diphosphites (5.2) [21].
The monomer-dimer equilibrium. Since the work of Wilkinson it has been
known that complexes of the type [RhH(CO)2(L-L)] and [Rh(L-L)(-
CO)(CO)]2 are in equilibrium with one another. The equilibrium constant
depends on the ligand used and the concentrations of rhodium and hydrogen.
Many authors have studied on the formation of these dinuclear species, but
no quantitative data had been reported before the study involving BDPP [50,
62]. In the Rh-BDPP system, both species can be observed and the
equilibrium constant is measured by 31P NMR. The data indicate that at
room temperature, relatively high catalyst concentrations, and partial H2
pressures of below 5 bar, rhodium will be present predominantly as the
dimeric complex. More monomer is formed when the H2 pressure is raised,
and hence, the rate of hydroformylation increases with the pressure,
regardless the mechanism of the rhodium hydride catalyst. The equilibrium,
therefore, is of great importance to the overall kinetics observed during
hydro formy la tion.
In addition the 31P NMR data show the presence of other species
containing more than two phosphorus atoms, in which the diphosphine acts
as monodentate. Hughes et al. [63] have described related complexes (see
chapter 4).
138 Chapter 5
In situ HPIR asymmetric hydroformylation. Under actual catalytic

conditions, high pressures, low concentrations, at 65C and 10 bar, the IR
spectra were measured. They revealed strong carbonyl absorptions at 1988
-1
and 1944 cm , characteristic of the mononuclear hydride dicarbonyl
complex, and only weak carbonyl absorptions at 1745 and 1724 cm-1, due to
a small quantity of the dinuclear complex. The substrate was then added and
the hydroformylation reaction was monitored. The absorptions
corresponding to RhH(BDPP)(CO)2 are retained throughout the reaction
indicating that the hydride is the resting state under these standard
hydroformylation conditions (see Chapter 4), at least for BDPP [55].
5.5 Mechanistic considerations
5.5.1 Regioselectivity
Regioselectivity is a key issue in hydroformylation. While for simple 1-

alkenes the emphasis is on the formation of linear aldehydes (see previous
chapters), in asymmetric hydroformylation the branched, chiral aldehyde is
the desired one. Many functional alkenes have a high propensity for the
formation of the branched product. In this chapter we have concentrated on
styrene, but before discussing the regioselectivity, it is instructive to have a
look at other alkenes (see Figure 27).
Figure 27. Functionalized alkenes
3,3,3-Trifluoropropene was studied by Ojima [64] and rhodium catalysts

gave a highly branched product at very high rates. If we assume that alkene
insertion into the rhodium-hydride bond is irreversible, the regioselectivity is
determined in this step. Theoretical studies point to an early transition state
for this reaction [15]. What would determine this early transition state? Both
frontier orbital combinations would lead to a secondary rhodium alkyl bond.
The electron withdrawing CF3-group would also stabilize the secondary
alkyl anion and a late transition state would lead to the same alkyl species.
Thus the branched product would form provided that no equilibration occurs
between the linear and the branched species. If there were to be
equilibration, the result would change drastically, because the primary alkyl
group would migrate much faster than the electronically stabilized secondary
alkyl group [65]. These are the electronic considerations. For steric reasons
one would expect a preference for the linear alkyl.
Ethyl acrylate, the hydroformylation of which has been studied many
times, is an interesting case. Tanaka et al. [66] found that the ratio of 1:b
could be varied from 100:0 (80 C and 1 bar) to 1:99 (40 C and 30 bar) by
using a rhodium catalyst and triphenyl phosphite as the ligand. Electronic
arguments lead to the same result as for trifluoropropene because the frontier
orbitals on ethyl acrylate are the same. Thus, high pressures lead to the
expected result.
Migration of the stabilized ester enolate, however, is expected to be much
1
slower than that of the linear alkyl, irrespective of whether it is 3 or
bonded. At low pressure the catalyst rests in the enolate state of the cycle
for the branched intermediate and the formation of the enolate can be
reversed. The product-forming route is now via the linear alkyl, which
undergoes a relatively rapid migration to form the acyl species (see Figure
28). The role of CO may be either to form a species containing more CO
molecules or to trap the acyl species after migration. In the absence of more
data this remains speculation.
Steric control in the figure is an oversimplification, because this route
is also taken when migration of the ester enolate is too slow, irrespective of
steric control as outlined above.
Figure 28. Hydroformylation of ethyl acrylate
In a broad range of conditions the rate laws for the rate of formation of
the linear and the branched product are different, which explains the
enormous influence that these have on the product distribution. At low CO
pressures the enolate-rhodium species may be the resting state of the
catalyst. Species of this type often undergo hydrogenation instead of
hydro formy lation.
For styrene the reactivity pattern is less obvious. The assumption that
there is an early transition state does not lead to a simple conclusion, because
the coefficients of both the LUMO and the HOMO of styrene are higher on
the terminal carbon atom. If the interaction of the HOMO on the Rh-H
140 Chapter 5
located on the hydrogen with the LUMO of styrene were to dominate the
branched alkyl would be formed. Clearly the resonance stabilized branched
alkyl is more stable than the linear phenylethyl species. Thus, electronically
the branched species is preferred as most authors have argued (see also
Chapter 2). In addition the branched alkyl species may be stabilized by 3
coordination, but the homo-allylic linear species is also stabilized by alkene-
rhodium interaction.
Insertion of styrene does not always lead to branched alkyl species.
Insertion of styrene into a cationic palladium hydride species may give
purely linear alkyl species, but perhaps not for steric reasons [67]. An early
transition state for this process may involve the interaction of the LUMO for
PdH+, which has the highest coefficient on palladium, with the HOMO of
styrene, which has the highest coefficient on the terminal carbon atom.
Figure 29 Hydroformylation of styrene
The influence of the conditions on the product distribution is the same as

for ethyl acrylate (see Figure 29). High temperatures and low pressures give
linear product (up to 1:b=3: 1), and low temperatures and high pressures lead
to branched products (1:b=1:99). For carbonyl based catalysts, phosphite
based catalysts, and phosphine based catalysts this has been well
documented (Chapters 2, 3, and 4). The reversibility of the branched alkyl
formation is also accounted for.
5.5.2 Enantioselectivity and conclusions
Two types of ligands have been successfully applied in asymmetric

hydroformylation using rhodium catalysts: diphosphite ligands and
phosphine-phosphite ligands. For both ligands, complexes with the formula
RhH(L-L)(CO)2 have been characterized as the resting state of the catalyst.
The formation of only one diastereomeric complex and alkene adduct is
thought to be crucial if enantioselectivity is to be high.
In the case of the diphosphites, compounds containing seven-, eight-, and

nine-membered rings have been studied. The bis-equatorial coordination
mode that involves an eight-membered ring provides the highest
enantioselectivities. Chiral cooperativity between the central backbone and
the aryl substituents as well as the influence of the terminal groups has been
observed for several systems.
In RhH(BINAPHOS)(CO)2 the ligand coordinates in an equatorial-apical
fashion. Both apical-equatorial isomers of the phosphine-phosphite
bidentates have been identified. For BINAPHOS and related systems, the
unique dissymmetric environment created by phosphine-phosphite seems to
be determinant for enantioselectivities to be high. As for diphosphites,
phosphine-phosphite systems require a well-tuned cooperation between the
chiral centers in the ligand.
High enantioselectivities have not still been found for rhodium chiral
diphosphine systems in asymmetric hydroformylation.
As pointed out in Chapters 3 and 4 the intermediate alkene adducts of the
diphosphite catalysts probably contain bis-equatorial bidentate ligands and
the intermediates of BINAPHOS contain an equatorial-apical phosphine-
phosphite ligand. Thus the structures involved in the migratory insertion are
different for the two systems.
The models used for rationalizing the enantioselectivities have been
discussed in section 5.3.3. The simple four-quadrant model was cited [46],
which gave good results for many substrates, but often failed for styrene.
Gleich and Herrmann [ 15,161 have applied computational methods to
hydroformylation using a combination of MM and QM. The predictive
power of this approach will certainly increase in the near future.
References
1 (a) Sheldon, R. A. Chemistry and Industry, 1990, 212. (b) Botteghi, C.; Paganelli, S;.
Schionato, A; Marchetti, M. Chirality, 1991, 3, 355.
2 Consiglio, G. Catalytic Asymmetric Syntheses, I. Ojima editor VCH Publishers, New
York, 1993.
3 (a) Agbossou, F.; Catpentier, J. F.; Mortreux, A. Chem Rev. 1995, 95, 2485. (b)
Gladiali, S.; Bayn, J .C.; Claver. C. Tetrahedron Asymmetry, 1995, 7, 1453.
4 Nozaki, K. Comprehensive Asymmetric Catalysis ; Jacobson E. N.; Pfaltz, A.,
Yamamoto, H. Ed.; Springer Heidelberg, 1999; Vol.1 , pp 381-409.
5 Stille, J. K.; Su H.; Brechot, P.; Paninello, G.; Hegedus, L. S. Organometallics, 1991,
10, 1183.
6 a) Consiglio, G., Nefkens, S. C. A, Tetrahedron Asymmetry, 1990, I 417. b) Consiglio,
G.; Nefkens, S. C. A, Borer, A. Organometallics, 1991, 10, 2046.
7 Cserpi-Szcs, S. Bakos, J. Chem. Commun. 1997, 635.
142 Chapter 5
8 Stanley, G. G. Catalysis of Organic Reactions, ed. Scaros, M. G.: Prunier, M. L.:

Marcel Dekker Inc. New York, 1995, p. 363.
9 Clark jun. C. E.: Diss Abstr.Int. B, 1973, 36, 2220.
10 Wink, J. D.; Kwok ,T. J. ; Yee A. Inorg. Chem. 1990, 29, 5007
11 Sakai, N.; Nozaki, K.; Mashima, K.; Takaya, H.Tetrahedron . Asymmetry, 1992, 3, 583.
12 Babin, J. E.; Whiteker, G. T. Pat, WO 93103830, 1992.
13 Buisman G. J. H., Vos, E. J.; Kamer, P. C. J.; van Leeuwen, P. W. N. M. J. Chem. Soc.
Dalton Trans, 1995, 409.
14 (a) Sakai, N.; Mano, S.; Nozaki, K.; Takaya, H. J. Am. Chem.Soc. 1993, 115, 7033. (b)
Nozaki K.; Sakai, N.; Nanno, T.; Higashijima, T.; Mano, S.; Horiuchi, T.; Takaya, H. J.
Am. Chem. Soc. 1997, 119, 4413.
15 Gleich, D.; Schmid, R.; Herrmann, W. A.; Organometallics, 1998, 17, 2141
16 Gleich, D.; Herrmann, W. A.; Organometallics, 1999, 18, 4354
17 Buisman, G. J. H.; Kamer, P. C. J.; van Leeuwen, P. W. N. M.; Tetrahedron
Asymmetry, 1993, 4, 1625.
18 Buisman, G. J. H.; van der Veen, L. A.; Klootwijk,: de Lange, W. G. J.; Kamer, P. C. J.;
van Leeuwen, P. W. N. M.; Vogt D. Organometallics, 1997, 16, 2929.
19 van Leeuwen, P. W. N. M.; Buisman , G. J. H.; van Rooy, A.; Kamer, P. C. J., Recl.
Trav. Chim. Pays-Bas, 1994, 113, 61.
20 van Rooy, A.; Kamer, P. C. J.; van Leeuwen, P. W. N. M.; Goubitz, K. L Fraanje, J.;
Veldman, N.; Spek, A. L.; Organometallics, 1996, 15, 83
21 Buisman, G. J. H.; van der Veen, L. A.; Kamer, P. C. J.; van Leeuwen, P. W. N. M.;
22 Meakin, P.; Muetterties, E. L.; Jesson, J. P. J. Am. Chem. SOC. 1972, 94, 5271.
23 Meakin, P.; Jesson, J. P.; Tebbe, F. N.; Muetterties, E. L. J. Am. Chem. Soc. 1971, 93,
1797.
24 Hyde, E. M.; Swain, J. R.; Verkade, J. G.; Meakin, P.; J. Chem. Soc. Dalton Trans,
1976, 1169.
25 van Rooy, A.; Kamer, P. C. J.; van Leeuwen, P. W. N. M.; Veldman, N.; Spek, A. L. J.
Organometal. Chem. 1995, 494, C15.
26 (a) Berry, R. S.; J. Chem. Phys.; 1960, 32, 933. (b) Muetterties, E. L. Acc.Chem.
Res., 1970, 3, 266.
27 Cserepi-Szucs, S.; Tth, I.; Prkny, L. Bakos, J.; Tetrahedron: Asymmetry, 1998, 9,
313.
28 Cserepi-Szucs, S.; Huttner, G.; Zsolnai, L. Bakos, J. J. Organomet. Chem., 1999, 586,
70.
29 Cserepi-Szucs, S.; Huttner, G.; Zsolnai, L.; A. Szlsy, A.; Hegeds, C.; Bakos J.
Inorganica Chimica Acta, 1999, 296, 222.
30 Jiang ,Y.;. Xue, S.; Li, Z.; Deng, J.; Mi, A.; Chan, A. S. C. Tetrahedron: Asymmetry,
1998, 9, 3185.
31 (a) Reetz, M. T.; Neugebauer, T. Angew. Chem. Int. Ed. 1999, 38, 179. (b) RajanBabu,
T. V.; Radetich, B.; You, K.; Ayers, T. A.; Casalnuovo, A.L.; Calabrese, J. C.; J. Org.
Chem. 1999, 64, 3429 and references cited therein.
32 Brown, J. M.; Cook, S. F., Tetrahedron, 1986, 42, 5105.
33 RajanBabu, T. V.; Casalnuovo, A. L. Pure and Appl. Chem. 1994, 66 1535.
34 RajanBabu, T. V.; Ayers T.A. Tetrahedron Letters 1994, 35, 4295.
35 Buisman, J. H. G.; Martin, M. E.; Vos, E. J.; Kamer, P. C. J.; van Leeuwen, P. W. N. M
Tetrahedron Asymmetry, 1995, 6, 7 19.
36 Pmies, O.; Net, G.; Ruiz, A.; Claver, C. Tetrahedron Asymmetry, 2000, 11, 1097.
37 Kadyrov R.; Heller. D.; Selke, R. Tetrahedron Asymmetry. 1998, 9, 329.
38 Sakai, N.; Mano, S.; Nozaki, K.; Takaya, H.; J. Am. Chem. SOC.; 1993, 115, 7033.
39 Sakai, N.: Nozaki, K.; Takaya, H. J. Chem Soc.Chem. Comm. 1994, 395.
40 Higashijima, T.; Sakai, N.; Nozaki, K.; Takaya, H. Tetruhedron Letters, 1994, 35, 2023.
41 a) Horiuchi, T.; Ohta, T.; Nozaki, K.; Takaya, H. Chem Comm. 1996, 155. (b) Horiuchi
T.; Ohta, T.; Shirakawa, E.; Nozaki, K.; Takaya, H. J. Org. Chem.; 1997, 62, 4285.
42 (a)Naieli ,S.; Carpentier, J-F.; Agbossou, F.; Mortreux, A.; Nowogrocki ,G.; Wignacout
J. P. Organometallics, 1995, 14, 401 (b) Kless, A.; Holz, J.; Heller, D.; Kadyrov, R.;
Selke, R.; Fischer, C.; Bmer, A.; Tetrahedron Asymmetry, l996, 7, 33. (c) Basoli, C.;
Botteghi, C.; Cabras, M. A.; Chelucci, G.; Marchetti, M. J. Organomet. Chem. 1995,
488, C20. (d).; Chelucci, G.; Cabras, M. A.; Botteghi, C., Basoli, C., Marchetti, M.
Tetrahedron Aymmetry, 1996, 7, 33.
43 Vaska, L. J. J. Am. Chem.Soc. 1996, 88, 4100.
44 Horiuchi, T.; Ohta, T.; Shirakawa, E.; Nozaki. K., Takaya, H. Organometallics, 1997,
16, 2981.
45 (a) Lazzaroni, R.; Raffaelli, A.; Settambolo, R.; Bertozzi, S.; Vitulli ,G.; J. Mol. Cat.
1989, 50, 1(b) ) Lazzaroni, R; Ucello Barretta, G.; Bennetti, M.; Organometallics, 1989,
8, 2323 (c) Raffaelli A.; Pucci, S.; Settambolo, R.; Ucello Barretta, G.; Lazzaroni, R.;
Organometallics, 1991, 10, 3892, (d) Lazzaroni, R.; Settambolo, R.; Ucello Barretta,
G.; Organometallics, 1995, 14, 4644 (e) Lazzaroni, R.; Ucello Barretta G., Scamuzzi S.;
Settambolo, R.; Caiazzo, A.; Organometallics, 1996, 15, 4657.
46 Consiglio, G.; Pino, P.; Top Curr. Chem., 1982, 105, 77.
47 Nozaki, K.; Itoi, Y.; Sibahara, F.; Shirakawa, E.; Ohta, T.; Takaya, H.; Hiyama, T. J.
Am. Chem. Soc. 1998, 120, 4051.
48 Francio, G.; Leitner, W. Chem. Commun. 1999, 1663.
49 Deerenberg, S.; Kamer, P. C. J.; van Leeuwen, P. W. N. M. Organometallics, 2000, 19,
2065.
50 Pottier, Y.; Mortreux, A.; Petit, F.; J. Organometal. Chem. 1989, 370, 333.
51 Salomon, C.; Consiglio, G.; Botteghi, C.; Pino P. Chimia, 1973, 27, 215.
52 Gladiali, S.; Pinna, L. Tetrahedron: Asymmetry, 1991, 2, 263.
53 Masdeu-Bult, A. M.; Orejon. A.; Castellanos, A.; Castillon, S.; Claver, C. Tetrahedron
Asymmetry, 1996, 7, 1829.
54 Dierkes, P.; van Leeuwen, P. W. N. M. J. Chem. SOC. Dalton Trans. 1999, 1519.
55 Castellanos-Pez, A.; Castillon, S.; Claver, C.; van Leeuwen, P. W. N. M.; de Lange,
W. G. J. Organometallics, 1998, 17, 2523.
56 Diguez, M.; Pereira, M. M.; Masdeu-Bult, A. M.; Claver, C.; Bayn, J. C. J. Mol.
Cat. 1999, 143, I11.
57 Freixa, Z.; Pereira, M. M.; Pais, A. A. C. C.; Bayon, J. C. J. Chem. Soc. Dalton Trans.
1999, 3245.
58 Miquel-Serrano, M. D., Masdeu-Bult, A. M.; Claver, C.; Sinou, D. J. Mol. Cat. 1999,
143, 49.
59 Eckl, R. W.: Primereir, T.; Herrmann, W. A. J. Organometal. Chem. 1997, 532, 243.
60 Rampf, F. A.; Splieger, M.; Herrmann, W. A. J. Organometal. Chem. 1999, 582, 204
61 Cornils, B.;. Herrmann, W. A (eds) Applied Homogeneous Catalysis with
Organometallic Complexes, VCH, Weinheim, 1996. Chap 4.1.
62 Casey, C. P.; Whiteker, G. T.; Melville, M. G.; Petrovich, L. M.; Gavney, J. A.;Jr.;
144 Chapter 5
63 Hughes, 0. R.; Young, D. A. J. Am. Chem. Soc. 1981, 103, 6636.

64 Ojima, I.; Kato, K.; Okabe, M.; Fuchikami, T. J. Am. Chem. SOC. 1987, 109, 7714.
65 Cotton, J. D.; Dunstan, P. R. Inorg. Chim. Acta, 1984, 88. 223.
66 Yamashita, H.; Roan, B. L.; Sakakura, T.; Tanaka, M. J. Mol. Catal. 1993 ,81, 255.
67 Pisano, C.; Mezzetti, A.; Consiglio, G. Organometallics, 1992, 11, 20.
Chapter 6
Hydroformylation in Organic Synthesis

Sergio Castilln and Elena Fernndez
Departament de Qumica Analitica i Qumicu Orgnica. Departament de Quimica Fsica i

Inorgnica. Universitat Rovira i Virgili. Placa Imperial Trracol, 43005 Tarragona, Spain.
6.1 Introduction
The discovery of new catalytic systems that enabled the regio- and
steroselectivity of hydroformylation to be controlled has meant that this
reaction has emerged as a flexible and important tool in organic synthesis.
Some interesting reviews on asymmetric hydroformylation (see chapter 5 in
this book) [ 1a-h], hydroformylation tandem reactions, [2,3] other general
aspects of the reaction and how it can be applied in the synthesis of-fine
chemicals [4a-g] give an account of the impressive progress of
hydroformylation over the last decade. Rhodium-catalyzed hydroformylation
takes place under mild conditions and it is compatible with the most
common functional groups present in an olefinic substrate [4a], and it is,
therefore, a synthetically useful tool for the preparation of organic
compounds [2].
This chapter discusses some general aspects of the hydroformylation of
alkenes in organic synthesis. It focuses mainly on regio- and stereoselective
processes, and analyzes the influence of the substrate and the catalysts.
Practical methods which provide high yields and selectivities, and short-cuts
compared to classical organic routes will be described. Particular attention
will be paid to recent advances that have helped to enlarge the synthetic
application of this reaction. Section 6.7 deals with the hydroformylation of
alkynes, and such key aspects as hydroformylation in water-gas shift
conditions and silylformylation, particularly efficient catalytic systems and
the application of hydroformylation in organic synthesis.
145
146 Chapter 6
6.2 Hydroformylation of unfunctionalized alkenes
Control of selectivity, chemo-, regio-, and stereoselectivity, is the most

important problem in the hydroformylation reaction. As far as
chemoselectivity is concerned such competitive reactions as isomerization,
double bond hydrogenation and aldehyde hydrogenation occur under
hydroformylation conditions.
In general, rhodium catalysts do not provide hydrogenation products, and
factors favoring isomerization are well established (see discussion below for
factors favoring -elimination).
It is generally accepted that in rhodium-catalyzed hydroformylation the
formation of the most stable linear and branched alkyl rhodium complexes is
assumed to be the step that determines the regioselectivity when there are no
-elimination processes (Figure 1), (see chapter 2). But the reverse process,
the dissociation of the metal-alkyl via -hydride elimination, plays an
important role in the regioselectivity of the reaction, because it helps to
equilibrate the alkyl species and can produce isomerization. In this case the
regioselectivity is determined by the relative migratory ability of both metal-
alkyl intermediates.
-Elimination is favored by the temperature [5,6] and disfavored by the
pressure. However, temperatures cannot be low in many cases because the
activity of catalyst is low. This is the case of highly hindered olefins,
because the coordination of the olefin to the alkene is the rate limiting step
and high temperatures are required which result in secondary isomerization
processes.
The regioselectivity is determined by the steric and electronic properties
of the alkenes, both the metal and the ligand of the catalytic system and the
reaction conditions, mainly temperature, pressure and the ratio of carbon
monoxide to hydrogen.
The substituents in the double bond strongly determine the
regioselectivity, and in general the formyl group is introduced into the less
substituted alkene carbon. Thus, regioselectivity is complete in disubstituted
terminal alkenes 2, (Figure 2) to give the linear aldehyde. The
monosubstituted alkenes 1 give mainly lineal aldehydes, but different
percentages of the branched aldehydes are always obtained. The
disubstituted alkenes such as 3 give mixtures, except when are
symmetrically substituted. The main problem of alkene 4 is its low
reactivity, which decreases with the substitution. As a result more drastic
reaction conditions are required and side reactions such as isomerization or
hydrogenation also usually take place [7,8].
Alkenes 1 and 2 are the synthetically most useful. Alkene 3 can also be
useful in substrates containing structural elements which can control the
6. Hydroformylation in organic synthesis 147
regio- and, eventually, the stereoselectivity (i.e. heteroatoms which can

provide chelation).
Figure 1. Competitive processes in the alkene hydroformylation
Hydroformylation of terminal alkenes. The hydroformylation of terminal

alkenes has been widely investigated in an attempt to find systems that can
control the regioselectivity. Although linear/branched mixtures are usually
obtained, an excess of monophosphorus ligand helps to obtain preferentially
the linear isomer. Recently, new ligands have been reported which overcome
the regioselectivity problems.
Formyl group
introduction
Reactivity
Figure 2. Selectivity and reactivity in the hydroformylation of differently substituted alkenes
Thus, monosubstituted terminal olefins are selectively hydroformylated

to give the lineal aldehyde by using diphosphite ligands such BIPHEPHOS
(5) (also known as UCC ligand (Figure 3) [9,10,11], or diphosphines such as
BISBI (6) [12] or XANTPHOS (7) [13]. By combining these ligands with
simple rhodium complexes such as [Rh(acac)(CO)2] or [Rh(dipivaloyl-
methanoate)(CO)2], regioselectivities can be as high as 100%.
Since differently substituted alkenes are hydroformylated at different
rates, mono- or disubstituted alkenes can be selectively hydroformylated in
the presence of di- or trisubstituted ones, respectively [4b].
As has been explained above, isomerization is a competitive process that
can be diminished, or suppressed, by choosing low reaction temperatures
and an appropriate catalyst. However, isomerization can be advantageously
used to hydroformylate internal olefins to give the linear aldehyde.
148 Chapter 6
5 BiPHEPHOS 6 BlSBl 7
Figure 3. Phosphite and phosphine ligands for regioselective hydroformylation
Thus, by using XANTPHOS analogues [ 14] and appropriate reaction

conditions such as high temperature and low pressure, 2- and 4-octene are
hydroformylated to give a remarkable 90% of the lineal aldehyde (Table 1)
(Figure 4) (see chapter 4). Under these conditions, the isomerization takes
place very quickly and all the possible alkenes are present in equilibrium in
the reaction medium, and the terminal alkene is hydroformylated faster than
the internal ones. Moreover, the catalytic system proved to be remarkably
regioselective towards the linear aldehyde.
8 9 10 11
Figure 4. Isomerization-hydroformylation of internal alkenes
Table 1. Isomerization-hydroformylation of alkenes

Substrate Catalyst P(bar) T C Conv % l/b %
2-octene Aa 2 120 22 9.2
4-octene Aa,b 2 120 67 3.4
2-butene BC 48 I20 5 1.5
a
A:[Rh(CO) 2)dipivaloylmcthanoate)]/7, [Rh]= 1 mM, ratio [Rh]/P/substrate= 1: 10:673, CO/H2=1, toluene,
c
1h. b Sameconditionsthana, l7h. B: Rh(OAc)3/phosphine-phosphiteligand/Ru3(CO)12CO/H2=1.
toluene,17h.
Linear aldehydes are mainly obtained from 2-butene by using a dual

catalytic system in which one metal catalyses the isomerization reaction and
another metal the consecutive hydroformylation [ 15]. Although the results
are still far from being synthetically useful, this work opens up new
perspectives for this process (Figure 4).
As will be seen below, when coordinating atoms are present in the
substrate chelation is an important factor in regioselectivity control. Linear
or branched aldehydes can be obtained depending on the size of the chelate

ring (see section 6.5).
Hydroformylation of 1,1-disubstituted alkenes. As has been mentioned

above, 1,1 -disubstituted alkenes are regioselectively hydroformylated to give
3,3-disubstituted propanal derivatives [ 16,171. They are valuable
intermediates in the preparation of various drugs [ 18,191.
Functional groups such as silyl ether, ester, acetal and halide, are
compatible with hydroformylation conditions. Thus, -functionalized 3-
substituted alkanals are synthesized from 2-alkyl-1 -alkenes, and this
methodology has been applied to the synthesis of muscone (14) (Figure 5).
[20].
Figure 5. Selective hydroformylation of -alkenes. Synthesis of muscone
Acrolein acetals 15 bearing an alkyl or aryl substituent in the 2-position

are regiospecifically hydroformylated to the corresponding succinaldehyde
monoacetals 16 with rhodium catalysts (Figure 6) [21].
Figure 6. Hydroformylation of acrolein dimethylacetal. Synthesis of partially protected 1,4-

dialdehydes
6.3 Hydroformylation of functionalized alkenes

The formyl group generated by rhodium (I) catalyzed hydroformylation
is generally incorporated at the carbon atom contiguous to the heteroatom.
This is the case of 1-monosubstituted and 1,2-disubstituted alkenes.
150 Chapter 6
However, in 1,1-disubstituted alkenes the steric factors predominate and the

formyl group usually goes to the less substituted alkene carbon (Figure 7)
although there are many exceptions.
Figure 7. General trends in the hydroformylation of heteroatom substituted alkenes
Consequently, functionalized 1 -alkenes such as vinyl esters, vinyl

phthalimides, vinyl fluorides and vinyl ethers can lead to -acyloxy-
aldehydes 17 [22], -phthalimido-aldehydes 18 [23], -halo-aldehydes 20
[24], and -alkoxy-aldehydes 21, [5, 25], respectively, (Figure 8). Chelation
usually helps to increase the regioselectivity, although the reactivity
decreases significantly, as is the case of vinyl acetate [26].
a c
[Rh4(CO)12, 217 bar, CO/H2= 1,60C b [RhHCO(PPh3]3 100 bar CO/H2= 1, 100C Rh6(CO)16, 120
d e
bar, CO/H2= 1,70C Rh 4(CO)12, 110 bar, CO/H2=1,8OC Rh4(CO)12, 100 bar,CO/H2 (1:1), 20C
Figure 8. Hydroformylation of substituted alkenes
Usually unsaturated functionalized cyclic alkenes such as dihydrofurans

[27] and N-acyl-2-pyrrolines [23] are also very reactive towards rhodium-
catalyzed hydroformylation. They principally give the expected -formyl
6. Hydrofo rmyla tion in organic synthesis 151
product, (Figure 9). Dihydropyrans [27] behave similarly but are much less
reactive.
As has been seen above, ligands have a considerable influence on the
control of regioselectivity because they stabilize a preferred M-alkyl
intermediate or accelerate or avoid the -elimination process. An example of
this is the hydroformylation of 2,3- and 2,5-dihydrofuran [27]. Thus, the
t
hydroformylation of both dihydrofurans 22 and 23, using P(O- o- BuC6H4)3
gives practically the same ratio of products 24:25 (Figure 9) (entries 2 and 4
Table 2).
The evolution of the reaction shows that 23 is isomerized into 22 before
the hydroformylation reaction starts, because the M--alkyl intermediate is
b-eliminated faster than CO is inserted. However, if the reaction is started
from 23 and the ligands have small cone angles, such as P(OMe)3,
practically the only aldehyde detected is the 3-formyl derivative (entry 3). If
it is started from 22, the 3-formyl derivative is the main aldehyde detected
(entry 1). This suggests that under these conditions there is no -elimination
process or that it is very slow.
These results show that the regioselectivity of the process can be
controlled if the ligand and reaction conditions are selected. Thus, PPh3 can
be used to quantitatively convert 23 into 25. The results are similar to when
P(OMe)3 is used in a high P/Rh ratio, at moderate temperatures and high CO
pressures but the catalyst is more active using PPh3. When P(O-o-tBuC6H4)3
is used at high temperatures and low CO pressure, however, the main
aldehyde 24 is obtained.
However, 3,4-di hydro-2H-pyran and 5,6-dihydro-2H-pyran required
more drastic conditions to be hydroformylated and when only P(O-o-
t
BuC6H4)3 was used as the ligand conversions were as high as 80% and the
selectivity in 2-/3-formyl derivatives was 68/32 [27].
Figure 9. Hydroformylation of dihydrofurans
Reactive unsaturated halides, such as allyl chloride, destroy the catalytic

activity of rhodium catalysts, probably because very stable inactive
complexes are formed [ 7 ] .
152 Chapter 6
a
Table 2. Hydroformylation of dihydrofuran 24 and 25 using different phosphorus ligands.
Entry Substrate Ligand Conversion(%) Ratio 24/25

1 22 P(OMe) 3 13 29:71
2 22 P(O-o-tBuC6H4) 3 99 76:24
3 23 P(OMe)3 43 1:99
4 23 P(O-o-tBuC6H4) 3 97 65:35
a
0.25 mol/% of [Rh2(-CH2)3NMe2)2(cod)2], Ligand[Rh]= 10, CO/H2= 1,5 bar, 80C,
1,2-dichloroethane.
The hydroformylation of vinyl arenes 26 has been widely studied and the
branched aldehyde 28 is the main product of the reaction (Figure 10). When
rhodium catalysts are used and pressures are high, regioselectivities are
usually between 80 and 98%. The major regioisomer is the inverse of the
one for alkyl monosubstituted olefins. The formation of -complex 27
3
explains the regioselectivity and this has already been discussed in the
literature (Figure 10) [24, 28, 29]. The regioselectivity can vary according to
the substituents in the ring [24b, 30].
Figure 10. Hydroformylation of vinylarenes
In the rhodium catalyzed hydroformylation of 1,1 -diphenyl alkenes the

linear aldehyde is preferred. However, the branched metal-alkyl is formed
faster, although the CO insertion is quicker in the linear metal-alkyl
intermediate. On the contrary, in 1 -phenyl-1 -(2-pyridyl)-alkenes such as 30
the branched aldehyde 31 is obtained instead of the linear one 29 [3 1,321.
i) [RhH(CO)(PP3)3], substrate/[Rh]= 80, CO/H2= 1, 100 bar, 80C, benzene, 99%.
Figure 11. Hydroformylation of 1 -phenyl- 1 -(2-pyridyl)-ethene

6. Hydroformy lation in organic synthesis 153
The chemoselectivity of 3- and 4-vinylpyridines when Rh4(CO)12 is used

as the catalyst is very different (Figure 12). While the 3-vinylpyridine 32 is
fully hydroformylated to give the branched aldehyde 33 (b:1 =96:4), the 4-
vinylpyridine 34 affords only the hydrogenation product 35. This is because
of the different nucleophilicity of the alkyl carbon atom in the rhodium
intemediate, which in the case of 4-vinylpyridine is not able to give the
migratory insertion [33]. However, when Rh4(CO)12 is modified with
phosphorus ligands, the branched aldehyde is obtained [34]. In these
conditions the carbon bonded to the metal becomes sufficiently nucleophilic
to give the migratory insertion of CO, hence evolving to the acyl
intermediate. Vinyl pyrrole also give the branched aldehyde, but temperature
has a greater influence on 2-vinylpyrrole and the nitrogen needs to be
protected if conversions are to be good [35]. Thus, 1 -tosyl-2-vinyl-pyrrole
and 1-tosyl-3-vinyl-pyrrole are hydroformylated in good yields and with a
branched /linear ratio >94:6.
I) [Rh4(CO)12], substrate/[Rh]= 300, CO/ H2= 1, 100 bar, 60-1 10 C. benzene.
Figure 12. Hydroformylation of vinylpyridines
Electron-poor alkenes are less reactive than common alkenes and

hydroformylation using rhodium catalysts gives mainly the branched
aldehyde (Figure 13). Interestingly, cobalt catalyst lead to the hear
aldehyde, while for the electron-rich alkenes (Figure 8) both catalysts
provide the same regioselectivity. Compared to simple olefins, the
regioselectivity in the branched aldehyde is higher in the rhodium
hydroformylation of electron acceptor-substituted olefins such as
trifluoropropene 36 (X=GF3 [24]), dimethoxyaerolein 37 (X=(AcO)2CH
[23]) vinylsulphones 38 (PhSO2 [36]), acrylonitrile 39 (X=CN [37]). This
reflects also the higher stability of branched alkyl-metal intermediates
(Figure 1). The regioselectivity shown by cobalt catalysts can be explained
because branched alkyl-cobalt intermediates tend to isomerize more than the
analogous rhodium catalysts.
154 Chapter 6
36 39
a C + 6 -
[Rh6(CO)16], CO/H 2 =1, 110 bar, 80C. b [Rh], 200 bar CO/H2=1, 100C, [Rh ( C6H5B Ph3)(cod)]
/dppb, CO/H2= 1,40br, 75C.
Figure 13. Hydroformylation of electron-deficient alkenes
The chemo- and regioselectivity of rhodium catalyzed hydroformylation

of these substrates depends heavily on the phosphorus ligand and on the
reaction conditions. Thus, if P(OPh)3 is used in hydroformylation of ethyl
acrylate at low temperatures and high pressures, the branched aldehyde is
almost exclusively obtained (Table 3). Under these conditions, no
isomerization takes place and the regioselectivity reflects that the branched
alkyl-metal intermediate is preferred. If the temperature is increased and
pressure decreased, the regioselectivity is completely reversed and only the
linear aldehyde is obtained. Significant amounts of hydrogenation product
are also observed. Under these conditions, the isomerization is faster than
hydroformylation, and the linear and branched alkyl-metal intermediates are
equilibrated. In this context, the linear intermediate migrates faster than the
branched intermediate and therefore generates the linear aldehyde. In the
presence of electron-donating phosphines such as P(OiPr)3 hydrogenation
takes place and gives ethyl propanoate, probably because the rhodium(III)
intermediate stabilizes.
Table 3. Influence of phosphorus ligands and reaction conditions on the hydroformylation of

a
ethyl acrylate.
Ligand T (C)Pressure Conversion Branched Linea(%) Hydrog.

(bar) (% (%) (%)
P(OPh)3 40 30 100 96 1 1
P(OPH)3 80 1 70 0 27.7 38
P(OiPr)3 40 1 96.4 7.9 0.7 87.8
a
0.005 mol of [Rh(acac)(CO)2] per mol of substrate. L/Rh= 5, CO/H2=1, 17 h.
Since most hydroformylation experiments are performed at high

pressures, the reported examples of hydroformylation of ethyl acrylate
mainly gives the branched aldehyde. However, if sterically hindered and
electron withdrawing phosphite ligands are used with high ligand/Rh ratios
and high concentrations of catalysts at high pressures and low temperatures,
the linear aldehyde can be preferentially obtained [38].
When the alkene is 1,1-disubstituted, the linear aldehyde is obtained
(Figure 14) [28]. The behavior of amide derivatives, however, depends on
the amino group substitution. Thus, hydroformylation of the 2-methyl-
acrylamide 42 (R=H) gives aldehyde 46 and by subsequent cyclization with
the formation of unsaturated lactams 47, [39, 40] (Figure 14). Alternatively
when R=Et, lactone 45 is formed by consecutive hydroformylation,
reduction and cyclization (see section 6.6 for other consecutive processes).
i)[Rh4(CO)12],substrate/[Rh]=300,CO/H 2=1,300bar,130C,26h.toluene ii)[ Rh4(CO)12],CO/H2=1.

80bar,120C.72h, toluene
Figure 14. Hydroformylation of vinylamides
1,1 -Disubstituted functionalized alkenes can give the branched derivative

when they are sufficiently activated and the conditions are appropriate, (see
for instance figure 22 in chapter 5).
6.4 Substrate directed stereoselectivity
This section deals with substrate-controlled stereoselective

hydroformylation, since asymmetric hydro formylation is covered in chapter
5. The stereoselectivity of the hydroformylation reaction is the result of the
cis addition of the proton and the formyl group to the less hindered face of
the double bond [41]. The presence of heteroatoms in the substrate causes
chelation, so the stereoselectivity can be controlled, (see section 6.5).
There are various ways of generating stereocenters by hydroformylation.
In monosubstituted terminal alkenes, a stereocenter is generated when the
156 Chapter 6
branched aldehyde is obtained (Figure 15, a). This is the case when R is a
phenyl group or a heteroatom. The regioselectivity in 1,1-disubstituted
alkenes gives the linear aldehyde but a stererocenter is generated in the
1
branched aldehyde when R is different from R2 (Figure 15, b). Two possible
regioisomers can be achieved in the hydroformylation of 1,2-disubstituted
alkenes so the formation of a stereocenter is only interesting in symmetric
alkenes or when exist elements controlling the regioselectivity (Figure 15,
c). And in trisubstituted alkenes, two stereocenters are generated as can be
seen in Figure 15, d.
Figure 15. Stereocentres generated in the hydroformylation reaction
Rhodium catalysts can be used to hydroformylate differently substituted

endocyclic alkenes such as -pinene (46) from the less hindered face of the
double bond to give 10-formylpinane (48) [42, 43]. Two stereocenters are
created. Interestingly, a cobalt catalyst gives aldehye 49 as a consequence of
a skeletal rearrangement followed by hydroformylation. The whole process
takes place with no loss in optical activity. However, the related exocyclic
alkene -pinene (47) is stereoselectively hydroformylated to give cis 50 or
trans 51 derivatives depending on the catalyst used [44] (Figure 16).
-3
i) [Rh]= 6.10 M, CO/H2=1, 300 bar, 70C. ii) [Co]= 51.10-2 M, CO/H2=1, 200 bar, 120 C.
- -
iii) [Rh]= 5.10 3 M, PPh3/Rh =100, CO/H2=1, 60 bar, 100 C. iv) [M]= 5.10 3 M, CO/H 2=1,
60 bar, 100 C.
Figure 16. Stereoselective hydroformylation of - and -pinene

i) [Rh(acac)(CO)2] / PPh 3 (ratio 1:4), H2/CO=1, 60 bar, 85C, 16h, THF.
Figure 17. Stereoselective hydroformylation of exocyclic alkenes
3,5-Dihydroxy aldehydes 54 are also stereoselectively prepared by

i t
hydroformylation of enol ether 52 (R1= H, Me, Pr, BnO(CH2)2, R2= Bu,
Me). This suppose an alternative procedure to the aldol reaction, (Figure 17)
[45, 46].
The reaction is highly stereoselective and only the syn isomer 54 is
observed. The process is kinetically controlled, and the olefin insertion is
apparently the rate-determining step.
i) [Rh(acac)(CO)2] / PCy3 (from 4), PPh3 (from 5), ratio Rh/P= 1.4, H2/CO=1, 60 bar, 120C,THF
Figure 18. Conformational control of the hydroformylation reaction

158 Chapter 6
In these exocyclic alkenes the chemo- and regioselectivity of the process

depends heavily on the conformation [46]. Thus, compound 55, which in
chair conformation has an axial methyl close to the coordinating point, led to
the anti isomer 59 after hydroformylation, together with significant amounts
of the isomerization compound 63. Elevated catalyst loading and drastic
conditions were required, and the results were best when PCy3 was the
auxiliary ligand.
Both hydroformylation and isomerization were found to occur from the
same diastereoface of enol ether 55. The fact that isomerization must take
place through a tertiary rhodium-alkyl which has a 1,3-diaxial interaction in
chair conformation seems to indicate that the intermediate has a boat
conformation.
Compound 56, the chair conformation of which is destabilized because of
a 1,3-diaxial interaction, gives exclusively the isomerization product 64, and
no aldehydes are detected. To explain these results it is suggested that
compounds with a preference for a chair-like conformation show good
regioselectivity for the primary rhodium-alkyl, and substrates with a
preference for a twist-boat conformation (substrate 56) show good
regioselectivity for the tertiary rhodium-alkyl leading to the isomerization
products (compound 64). Further, substrates without an overwhelming
conformational bias (substrate 55) react with lower regioselectivity.
The hydroformylation of glucal derivatives is a potential method for
synthesizing 2-deoxy-C-glycosides. When the cobalt catalyst [CO2(CO)8]
was used at high pressures and temperatures, glucal derivatives were
hydroformylated to give a mixture / of aldehydes 68 (Figure 19) [47,48].
Figure 19. Hydroformylation of glucal derivatives
On the other hand, rhodium catalysts were only able to hydroformylate

glycals in the presence of P(O-o-tBuC6H4)3. Independently of protecting
groups, the main product was the -2-formyl derivative 66 (sugar
numbering) [49]. When the protecting group is acetate significant amounts
of the elimination product 67 are also obtained, but when it is an ether-type
group the elimination product is not formed and increasing amounts of
compound 68 are obtained (Table 4).
Table 4. Hydroformylation of glucal derivatives 65.

b b b
R Catalyst P(bar) T(C) Conv. 66 67 68
a
(%)
Ac CO2(CO)g 300 200 >90
Ac [Rh]/P c 70 100 91 58 19 --
c
Bn [Rh]/P 50 100 99 68 19
a
Percentage of transformed product. b Percentage of products identified related to the
c t
products of the reaction detected by GC. [Rh]/P= [Rh2(-OMe)2(cod)2]/P(O-o- BuC6H4)3
Although glucal derivatives can exist in a conformational equilibrium,

the hydroformylation reaction is stereoselective and prefers the attack from
the face. The stereoselectivity is much better when the formyl group is in
C2 and not in C1. Apparently, the stereoselectivity in the former case is
controlled by the substituent at C3, which adopts a pseudoaxial arrangement
(vinylogous effect).
The regioselectivity obtained in the hydroformylation of 65 is the inverse
of the regioselectivity for the hydroformylation of 3,4-dihydro-2H-pyran
with the same catalytic system [27], or for the hydroformylation of 65 with
cobalt catalyst (Table 4).
The regioselectivity in dihydro-pyran and glucals may be different when
the same catalytic system is used because C2-metal intermediates form more
quickly than C 1-metal intermediates and because isomerization requires a
conformational arrangement in the molecule (Figure 20). This arrangement
is more difficult in substituted rings, such as glucals, than in non substituted
rings such as dihydropyrans.
Figure 20. Conformational equilibrium for -elimination in Rh-alkylglucals
Double bonds not directly bonded to cycles can also be stereoselectively

hydroformylated, but there must be an efficient sterical discrimination of
both faces of the double bond must. This is the case of 1-methylvinyl-C--
glucoside 71 which is hydroformylated to give a 2-substituted aldehyde 72 in
excellent yield and diastereoselectivity of 99% (Figure 21) [50] .The bulky
substituent at position 2 of sugar blocks the conformation shown in 71 and
forces the rhodium to be coordinated from the back face of the double bond.
160 Chapter 6
Similarly, the hydroformylation of 2-phenyl-4-(prop-2-enyl)- 1,3-dioxane

73 affords an all-anti-stereotriad 75 which becomes a valuable intermediate
in the total synthesis of the macrolide antibiotic bafilomycin A [51] (Figure
22). The conformation for intermediate 74 was calculated to be the most
stable and the equatorial methyl group appears to cause the high
stereocontrol observed. As can be deduced from these representative
examples six-member rings seem to provide good stereoselectivities.
i) [Rh(CO)2acac], toluene, H2/CO=1 :1, 80bar, 80C. 48h.
Figure 21. Stereoselective hydroformylation of vinyl-C-glycosides
i) [Rh(CO)2(acac)]/4P(OPh)3,toluene, H2/CO=1,20 bar, 70C.
Figure 22. Stereoselective hydroformylation of exocyclic alkenes
6.5 Control of the regio- and stereoselectivity by

heteroatom-directed hydroformylation
Remote substituted groups in the substrate, which are able to chelate to

the metal catalyst, can be used to control the regio- and stereoselectivity of
organic reactions. The sections above have discussed some examples of
carbonyl coordination, vinyl acetate for instance, and their influence on
reactivity and selectivity. Coordination of nitrogen in aminoalkenes to
rhodium was demonstrated by isolation of complexes with the double bond
i) [Rh(OAc)2]2, alkene/Rh= 50:1, PhH, CO/H2= 1, 37bar, 100C, benzene, 44h. ii) LiA]H4, Et2O.
Figure 23. Phosphite directed hydroformylation of homoallylic cyclic alkenes
i) [Rh(OAc)2]2, alkene/Rh= 50:1, PhH. CO/H2= 1, 37 bar, 50C. benzene, 5-22h. ii) LiAH4 Et2O.
Figure 24. Phosphite directed hydroformylation of homoallylic linear alkenes
Figure 25. Phosphino-alkenes for directed hydroformylation
and the amino group coordinated to rhodium [52]. Other illustrative

examples of carbonyl or nitrogen coordination will be discussed below (see
for instance Figures 35, 36 and 39)
Introducing phosphorus-containing groups in the substrate has also
proved to be a flexible and efficient procedure for controlling the regio- and
stereoselectivity of the reaction. Thus, homoallylic alcohols have been
transformed into homoallylic phosphites, and the influence of the
coordinating heteroatom on the regio- and stereoselectivity of cyclohexenyl
derivatives has been studied [53].
By using an unmodified system such as [Rh(OAc)2]2, the homoallylic
phosphite 76 was converted into the formyl derivative 77 with total regio-
and stereocontrol (Figure 23). In the open chain homoallylic phosphites 79,
even though the stereoselectivity was low (60:40 for R= Me and 70:30 for
R=Ph) regiocontrol was high which led to the formation of the branched
162 Chapter 6
aldehydes 80 (Figure 24). In both cases a six-member chelating ring was

formed. This ring was shown to exist by treating the phosphate derivative of
79 under similar hydroformylation conditions that led to an equimolar
mixture of branched and terminal aldehydes.
Some chelation was also observed to form larger rings. Thus, the
hydroformylation of n-pentenylphosphites gave a mixture with a branched-
linear ratio of 87:13.
Treatment of formyl-phosphites 77 and 80 with lithium aluminum
hydride gave respectively the 1,4-diols 78 and 81 in good yields.
Similarly, the alkene 82 was hydroformylated with complete regio- and
stereocontrol by using a Rh/phosphine catalytic system (Figure 25). Under
the same conditions 83 was hydroformylated with good regioselectivity but
low stereoselectivity (ratio cis/trans=4: 1) [54, 55]. Regioselectivity was also
complete in the hydroformylation of the open-chain phosphine 84 although
in this case the stereoselectivity was negligible.
When the phosphino derivatives 82, 83 were hydroformylated at higher
temperatures (90C) alcohols were formed as a consequence of consecutive
hydroformylation-reduction. This is an interesting result since alcohols are
not usually obtained with modified rhodium catalysts.
The regio-and stereoselective introduction of a formyl group into the
cyclohexene ring of 85 was a key step in the synthesis of phyllantocin
(Figure 26) [56, 57]. The first experiments in the hydroformylation of 85 led
to the undesired regioisomer.
An attempt was then made to anchor the phosphorus ligand to the
substrate and this required an easily removable group incorporated in the
phosphino group. The diphenylphosphinobenzoate group fulfilled these
requirements, and the ester 86 was prepared by reacting 85 with p-
(Ph2P)C6H4COOH using DCC as coupling reagent. Compound 86 gave a
very low yield in hydroformylation, but its phosphine oxide derivative gave
moderate yields of a mixture of formyl derivatives. It was suggested that this
was because the spacer was too long, and pushed the catalysts beyond the
olefin. Coupling the m-(Ph2P)C6H4COOH to the alcohol 85 gave compound
87 in an 88% yield, which was hydroformylated to give a mixture of
aldehydes in a 72% yield, presumably via the intermediate 88. The
selectivity in the major isomer 89 was 77 %. This result is the first and
overwhelming example of a long distance regio- and stereocontrolled
hydro formylation reaction.
The formyl group in 89 was then epimerized to obtain the compound 90,
in which the formyl group had the same configuration as the natural product.
When treated in basic medium the meta -(diphenylphosphino)benzoic acid
directing group, was recovered.
i) 8 mol % [Rh(OAc)(cod)]2, CO:H2= 1, 55 bar, 85C. benzene, 3h.
Figure 26. Hydroformylation step in the synthesis of phyllantocine
The introduction of the ortho -diphenyl-phosphinobenzoate ( o -DPPB)

group led to a considerable improvement in the directed diastereoselective
hydroformylation of methallylic [58, 59, 60] and homo-metallylic alcohols
[61, 62]. It also supposes the formation of a big chelate ring. In these
processes high 1,2- and 1,3-asymmetric induction is obtained.
Protecting groups such as t-BuPh2Si or the free OH do not provide good
stereoselectivity. This proves the absence of a directing effect in these cases.
The fact that the coordinating properties of the hydroxyl are lower than those
of CO may explain this result.
The hydroformylation of 91 (Figure 27) using [Rh(CO)2(acac)]/4
P(OPh)3 as catalyst quantitatively furnished the aldehydes 92, 93 with
diastereoselectivities of up to 92:8. An additional phosphorus ligand
(monophosphine) was necessary to ensure the presence of two phosphorus
i) 0.7 mol/% [Rh(CO)2(acac)]/4 P(OPh)3, CO/H2= 1, 20 bar, 90C, toluene, 24 h.
Figure 27. Stereoselective hydroformylation of metallylic alkenes controlled by the group

o-DPPB
164 Chapter 6
(o-DPPB)= ortho-diphenylphosphinobenzoate
i) 0.7 mol/% [Rh(CO)2(acac)]/4 P(OPh)3, CO/H2= 1, 20 bar, 50C, toluene, 72 h.
Figure 28. Stereoselective hydroformylation of homometallylic alkenes controlled by the o -

DPPB group
ligands coordinated to rhodium. Ligands such as PPh3, P(OPh)3, P(O-2,6-di-

t
Bu-C6H3)3, P(OEt)3 and P(N-pyrrolyl)3 were tested, and the results were best
with P(OPh)3. Temperatures higher or lower than 90 C decreased the
stereoselectivity.
The scope of the procedure was tested in a variety of substrates (different
R) and diastereoselectivity was highest in 1 ,I-disubstituted alkenes. The
directing group was compatible with the presence of several groups such as
ester, acid, phenyl, etc., and there was no loss in enantioselectivity when the
starting compounds were enantiomerically pure.
Compound 94 was also quantitatively converted into aldehydes 95, 96
with diastereoselectivity up to 91 :9 when the same catalytic system was used
(Figure 28). In the absence of o-DPPB group some diastereoselectivity is
also observed, although the disastereomer cis is mainly obtained in this case.
This result confirms that the o-DPPB group acts as directing group by
reversible coordination to the catalyst. A significant dependence of the
diastereoselectivity on the reaction temperature was observed. The
diastereoselectivity was best at 50C, which mean longer reaction times.
6.6 Consecutive processes under hydroformylation

conditions
In the presence of alcohols or amines, the aldehydes generated in the

hydroformylaton reaction give acetals or imines. Depending on the
hydroformylation catalyst used an additional acid catalyst is required. When the
process is intramolecular, (i.e. when alcohols, amines or amides are present in
the starting material), it is spontaneous and gives hemiacetals or imines,
especially if five or six member rings can be formed [4b]. Moreover the presence
6. Hydroformy la tion in organic synthesis 165
of coordinating atoms such as nitrogen leads to a chelate which controls the

regioselectivity of the process.
6.6.1 Hydroformylation-acetalization (intramolecular)
The hydroformylation of allylic and homoallylic alcohols has been widely

studied in this process because hydroxy-aldehydes can easily lead to hemiacetals
with five- or six- member rings. Likewise, the hydroformylation of allylic
alcohol to give 4-hydroxybutanol and 1,4-dihydroxybutanoI is an important
industrial process [63].
The hydroformylation of allylic alcohol 97 usually a mixture of regioisomers
98, 99, and propanal (for R=H) which is formed by the isomerization of the
double bond and is subsequent tautomerism (Figure 29). The regioselectivity is
mainly determined by the ligand. The substitution of the double bond also has a
strong influence on the selectivity and reactivity.
The general trend is that reactivity is low when double bond substitution is
high, and the formyl group is also introduced into the less substituted carbon in
this case.
Aldehyde 99 spontaneously cycles to give the hemiacetal 100. This leads to
the acetal 102 if the reaction takes place in the presence of an alcohol and the
catalytic system [Rh(CO)2-zeolite]/PEt3 [64], or in the presence of a carboxylic
acid and a rhodium catalyst [65]. Lactone 101 is easily obtained from hemiacetal
100 by oxidation.
Figure 29. Hydroformylation-intramolecular acetalization of allylic alcohols
An enantioselective version of this reaction has been reported [66]. The

hydroformylation of 97 (R=Ph) with the catalytic system [Rh(acac)(CO)2]/ 4
BINAPHOS gives the hemiacetal 100 (R=Ph) with a yield up to 99%, which
after oxidation provides the lactone 102 (R=Ph) with 88% ee.
The hydroformylation of the substituted allylic alcohols 103-106 (Figure 30)
leads mainly or exclusively to the linear aldehydes which evolve upon
cyclization to the respective hemiacetals type 102 [67, 68, 69].
166 Chapter 6
Figure 30. Different allylic alcohols studied in hydroformylation-intramolecular acetalization
A stereoselective version of the hydroformylation of 106 by heteroatom-

directing hydroformylation has been discussed above (see Figure 27).
Homoallylic alcohols such as 107 (R=H) give mixtures of regioisomers 108
and 109. Both can be cycled to provide tetrahydrofuran or tetrahydropyran
derivatives (Figure 31). As in the case of allylic alcohols, substitution in the
double bond or in the allylic carbon favors the formation of the linear aldehyde.
The linear aldehyde 109 provides the hemiacetal 110, from which acetals [64],
lactones [65, 77] and also enolethers such as 111, can be formed [70, 71, 72, 73].
Figure 31. Hydroformylation-acetalization of homoallylic alcohols
6.6.2 Hydroformylation-acetalization (intermolecular)
When alkenes are hydroformylated in the presence of alcohols,

dialkoxyacetals or orthoesters, aldehydes are mainly obtained, and there are only
small amounts of acetals. Normally, acetals can only be formed in presence of an
acid catalyst. Since the catalyst for the hydroformylation process requires the
presence of basic phosphine ligands, and the acetalization reaction requires an
acid catalyst, the main goal of the process was to find two compatible catalytic
systems. The solution was to use weak acids, such as phosphonium or ammoniun
salts, or carboxylic acids, in the presence of phosphorus ligands. Strong acids
shift favors the equilibrium to the protonated form of the phosphine, thus
inhibiting the hydroformylation reaction, and can also react with Rh-H to give
dihydrogen and salts (see chapter 9).
If platinum catalysts are used no additives are required, probably because of

the Lewis acidity of platinum [74].
Using phosphonium or pyridinium salts, 2,5-dihydrofuran (113) is selectively
hydroformylated with the [Rh(-OMe)(cod)2]/PPh3/PPTS (PPTS =pyridinium p-
toluensulfonate) catalytic system and 2,2-dimethoxypropane as the solvent, to
give the dimethoxyacetal 114 in quantitative yield [75]. Styrene 115 is converted
into dimethoxyacetal 116, with high yield and selectivity. The presence of the
weak acid has no influence on the regioselectivity. Interestingly, the aldehyde
generated during the hydroformylation of the keto-alkene 117 was selectively
acetalized in the presence of a ketone group to give the dimethoxyacetal 118,
(Figure 32).
Catalytic systems such as [Rh(-S(CH)2NMe2)(cod)]/PPh3, which are
anchored to a sulphonic exchange resin by protonation of the amino group,
converts 115 into 116 under hydroformylation conditions and with methanol as
the solvent [76]. In a similar reaction the catalytic system [Rh(-
SCH2NH3)(COOH]2 (OTf)2, in the presence of ethyl orthoformate, converts
terpenes into acetals [77].
i) 1 mol/% [Rh(-OMe)(cod)]2/ 10 t
PPh3/PPTS, CO/H 2= 1. 50 bar, 60C 2,2-dimethoxypropane, 24 h.
ii) Identical to i but using P(O-o - BUC6H3)3.
Figure 32. Consecutive hydroformylation-intermolecular acetalization with the catalytic

systems [Rh(-OMe)(cod)2]/PPh3/PPTS
As has been shown above (see section 6.4), the hydroformylation of 3,4,6-tri-
O-acetylglucal gives considerable amounts of the elimination product 121,
which must be obtained by eliminating of acetic acid from 120 (see also Figure
19) [49]. The hydroformylation of allylic esters to give ,-unsaturated
aldehydes by hydroformylation and acid elimination is a well documented
process [4e]. In situ acetal formation partially avoids this process. Thus, using
[Rh(-OMe)(cod)]2/P(O-o-tBuC6H4)3/PPTS as hydroformylation-acetalyzation
catalysts, 3,4,6-tri- O -acetyl-D-glucal 119 is converted into the dimethoxy acetal
168 Chapter 6
120 (Figure 34). Only a small amount of the elimination product 121 is formed
178].
Unexpectedly, the 3,4,6-tri-O-benzyl-D-glucal gave only the methyl -
glycoside 121 under the hydroformylation-acetalization conditions. In fact, there
are two electrophilic reagents in competition for the nucleophilic alkene, the
rhodium complex and the proton. When the alkene is deactivated (R=Ac) the
coordination of rhodium is preferred and the hydroformylation-acetalization
takes place. But when it is not deactivated (R=Bn) the acidic proton reacts faster
than rhodium and methanol is added.
t
i) 1 mol/% [Rh(-OMe)(cod)]2/ 10P(O-o- BuC6H4)3 / PPTS, CO/H2=1, 50 bar, 100C,
2,2-dimethoxypropane, 48h.
Figure 33. Hydroformylation-acetalization of glucal derivatives
6.6.3 Hydroformylation-amination (intramolecular)
When an amino group is present in the substrate (123, 125, 127), various
processes can take place consecutively under hydroformylation conditions to
afford, cyclic N,O- (124, 126) [79, 80] or N,N-acetals (128) [81] (Figure 34).
Imines and enamines can also be formed. The formation of acetals, imines or
enamines depends on alcohols being present in the substrate or the solvent,
additional amino groups being present in the substrate, the substitution of the
amino group, and the reaction conditions. Moreover the presence of coordinative
atoms such as nitrogen allows a chelate to be formed which control the
regioselectivity of the process.
Because of the chelate control of the process, allylamides such as 129 react
under hydroformylation conditions to give mainly the branched aldehyde 131,
together with cyclic derivatives 132 and 133 (Table 5) [82, 83]. Products 132
and 133 are formed from the linear aldehyde through a sequence of reactions
involving cyclization to give the enamide 134, followed by hydrogenation or
hydroformylation, respectively [84].
i) 1 mol/% [Rh(acac)(CO)2]/BlPHEPHOS, COH2= 1, 4 bar, 65C, EtOH. ii) [Rh(OAc)2]2/3 PPh3,

COH2= 1,28 bar, 80C.
Figure 34. Consecutive hydroformylation-intramolecular aminoacetalization
Hydroformylation of substrate 130 does not show a similar directing effect

and, depending on the catalyst used, gives the terminal aldehyde from which 131
or 132 are formed.
The reaction also works with cyclic amides to afford bicyclic heterocycles
[84]. The 3-butenamide 135 (n=1, R=H) also undergoes a consecutive process
which affords compounds 136 or the dimer 138 depending on the ligand used
(Table 6) [84, 85]. Large excess of PPh3 affords almost exclusively the linear
aldehyde and further to the pyrrolidone 136 by cyclization, water elimination to
give 140, and double bond isomerization.
Cross-coupling reaction between pyrrolidone 137 and the intermediate 140
would give the dimer 138. In the 4-pentenamide 135 (n=2) the regioselectivity
which gives the branched aldehyde is also very high and apparently is controlled
by the formation of a chelate. Then the branched aldehyde cyclizes and
eliminates water to give the lactam 139.
Figure 35. Hydroformylation of N-allylamides

170 Chapter 6
Table 5. Hydroformylation of N-allylamides 129 and 130.
Substrate Catalyst Yield(%) Ratio 131:132:133

129 [Rh( dpp b)(NBD)]CIO4a 87 7 1 :5:24
129 Co2Rh2(CO)12a 80 79:21 :O
130 [ R h(dpp b)(NBD)] C1O4b 87 --:--:>99.5
130 Co2Rh2(CO)12c 8.5 --: 100:-
a
CO/H2=1, 90 bar, 80C, THF, 18h. b CO/H2=1 1, 140 bar, 100C, THF, 18h. c CO/H2=0.3,
92 bar, 100C THF, 18h
The bulkiness of the substituent on the amide nitrogen virtually does not
have any effect on the regioselectivity, but it exerts a marked effect on the
cyclic/acyclic ratio of the products. The effect of the bulky N-substituents is
particularly pronounced in the trityl derivative (R=Tr, n=l), since no formation
of pyridone 136 was observed. A mixture of open chain linear aldehyde and
pyrrolin-2-one 137 was obtained.
Figure 36. Consecutive processes in the hydroformylation of unsaturated amides.
Table 6. Hydroformylation of unsaturated amide 135.
n Catalyst Yield (%) Ratio 136:137:138 139

a
1 [RhCI(PPh3)3 /20PPh3 100 92:8:0
1 [RhCI(PPh3)3 /10P(OPh)3a 90 3:3:94
2 [Rh4(CO)12]b 92 100
a
1 mol % [Rh], CO/H2= 3,90 bar, 100C, THF, 40h. b 1 mol % [Rh], CO/H2= 1,
90 bar, 100C THF, 18h.
Treating the protected amine 141 in the presence of the Rh-BIPHEPHOS

catalyst (see Figure 3) under hydroformylation conditions, leads to enamide 142.
The consecutive reactions are hydroformylation to give the linear aldehyde,
cyclization to give aminoacetal and the elimination of water. Compound 142 is a
key intermediate in the synthesis ofprosopinine 143 [86].
i) 1 mol/% Rh(acac)(CO)2/ 2 BIPHEPHOS, CO/H2, 4 bar, 65C 96%.
Figure 37. Consecutive hydroformylation-acetalyzation-elimination processes in the synthesis

of (+)-prosopinine
The hydroformylation of the amidodiene 144 catalyzed by Rh-BIPHEPHOS

under standard hydroformylation conditions gave the dehydropiperidine-
aldehyde 146 as the sole product. The reaction is extremely chemo- and
regioselective. The hydroformylation takes place at the homoallylic olefin
moiety exclusively, and yields only the linear aldehyde intermediate 145 [87] .
i) 1 mol % [Rh(acac)(CO)2] / 2 BIPHEPHOS. CO/H 2,4 bar, 65C.
Figure 38. Consecutive processes in the hydroformylation of diene 147
ortho-Propenebenzeneamines 147 are hydroformylated with Rh-PPh3

catalysts and the resulting product depends on the substitution in the allylic
carbon. Thus, when R1=OH, R2=H, alkyl, the benzazepine 149 is obtained with
regioselectivities up to 9 1 % because of the preferred formation of linear
aldehyde and the subsequent formation of the imine. However, when R1=R=H,
only dihydroquinoline 148 is formed, shows that the amino group has a
remarkable directing effect since the hydro formylation of 3-phenylpropene gives
a ratio b/1 70:30 [88].
N-Alkenyl-1,2-diaminobenzenes are hydroformylated with the Rh-PPh 3
i) 0.5 mol/% [Rh(OAc)2]2 ,4 PPh3, CO/H2= 1, 30 bar, 60C. EtOAc.
Figure 39. Consecutive processes in the hydroformylation of alkene-amine 150.

172 Chapter 6
catalyst to give fused benzimidazoles [89, 90].

Alkenyl acyclic diamines such as N-allyl-1,3-diaminopropanes (150, n=1,
R=H) are hydroformylated with Rh-BIPHEPHOS catalysts to give a quantitative
yield of product 154, which is formed from the linear aldehyde 151, through the
intermediate 152. The homoallylic derivative gives a similar result (150, n=2).
By changing the auxiliary ligand the reaction can be controlled to give
dihydropyrrolidone derivative [9 1]. When n=4-9, diazabicyclicoalkanes 153
with a macrocycle of 8-13 member rings are formed in yields up to 95% in
conditions that do not require high dilution. These aminals can be reduced with
DIBAL-H to give macrocycles 154 in good yields [92, 93].
i) 0.5 mol % [Rh(OAC)2]2 /4 BIPHEPHOS, CO/H2= 1, 30 bar, 80C, benzene, 20 h.
Figure 40. Consecutive processes in the hydroformylation of alkene diamines
6.6.4 Consecutive Hydroformylation-amination-reduction.

Hydroaminomethylation
The hydroaminomethylation of alkenes was originally discovered by Reppe

[94] and consists of the hydroformylation of an alkene, followed by reaction of
Figure 41. Consecutive processes in the hydroaminomethylation of alkenes.

the intermediate aldehyde with a primary or secondary amine to form an imine

or enamine, and a final hydrogenation to give a secondary or tertiary amine
(Figure 42) [4b, 95].
The hydroaminomethylation of 1 -octene with morpholine in the presence of
[Rhu(-S-tBu)(CO)2(PPh3)2]/PPh3 gives a mixture of amines 161 and 162,
together with other secondary products (figure 42) [96, 97]. Using an
unmodified rhodium catalyst, high pressures and a CO/H2 ratio of 4.5, the same
mixture of amines is obtained in quantitative yields that depend on the alkene
[96]. No byproducts were detected. The high CO/H2 ratio was chosen to
suppress substrate hydrogenation and support catalyst stability.
R=Hexyl
t
[Rh(-S-Bu )2(CO)2(PPh3)2]/PPh3,8bar,80C 42 13%
[RhCl(cod)]2/ Pco= 90 bar, PH2= 20 bar, 80C 46 36%
Figure 42. Hydroaminomethylation of 1-octene in the presence of morpholine
This reaction is general and preserves the regioselectivity observed under

normal hydroformylation conditions, although variations are observed as a
function of the amine. Thus, hydroaminomethylation of styrene leads to the
branched derivative 164 with both secondary (R1= R2=alkyl) and primary
amines (R1=alkyl, R2=H), although in this case more drastic conditions are
needed. The piperazines give the iso,iso-diamine 165 as the major product in
90% yield.
i) 1 mol/% [RhCl(cod)]2, CO/H 2= 1, 11 0 bar, 80C, dioxane.
Figure 43. Hydroaminomethylation of styrene in the presence of secondary amines and

piperazinc.
174 Chapter 6
The reaction is compatible with the presence of carbonyl groups in the

molecule, which remains unaltered, and only the aldehyde function reacts in the
reaction conditions (see also Figure 32).
In the last few years Eilbracht et col. have made a lengthy study of how to
apply this methodology to prepare a great number of differently functionalized
organic compounds using [RhCl(cod]2 as catalyst [4b]. No phosphorus ligands
were used as co-catalysts. Dialkenes 166 are treated under hydroformylation
conditions in the presence of secondary amines to give the diamines 167 (Figure
1
44). Mixtures n,n, n,iso, and iso/iso are obtained when R = R2= H. Lineal
2
aldehydes are the only products obtained when R1and R are alkyl groups. The
use of [Rh(acac)(CO)2] and BIPHEPHQS exclusively leads to n,n products. The
group X also has some influence on the regioselectivity but mainly on the
reaction that takes place. Thus, when X=O double bond isomerization gives the
enol, and when X=NH pyrrolidone is formed. Consequently, in order to obtain
triamines the NH function should be protected [98].
Diamines are also obtained from allyl chlorides by initial halogen
substitution to give the allylamine and subsequent hydroaminomethylation [99].
With primary amines or diamines this method can also be used in the synthesis
of heterocyclic systems.
Different - and d-aminofunctionalized ethers, amines and silanes 169 are
obtained from alkenes 168. Branched derivative is the main product obtained
when X=OR, NR. The alkylation of the primary amines leading to secondary or
tertiary amines can be controlled by the alkene/amine ratio [100]. Nitro
derivatives can be used instead of amines, since they are reduced to amines in
the reaction conditions [101]. Selective monoalkylation or dialkylation of nitro
compounds is achieved depending on the alkene/nitro compound ratio.
Enamines and enamides such as 170 mainly give the iso aldehyde, which
leads to the 1,2-diamines 171 by forming and then reducing intermediate
enamines [ 102].
Hydroformylation of 1,4-dialkenes 172 in the presence of primary amines
affords pyrroles 174 or the eight-member heterocycles 173, depending on the
alkene substitution pattern [103]. 1,4-Pentadienes, if substituted only in the 3,3-
1
position (R =Me, R=H), undergo intramolecular pyrrole formation to give the
bicylic systems 174. Pyrrole is generated via hydrocarbonylative cyclization
leading to the alkyl intermediate 175. This rhodium intermediate undergoes
further insertion of carbon monoxide to form an acyl intermediate, from which
pyrrole is obtained by a Pall-Knorr synthesis. In contrast, 1,4-pentadienes with
substituents in the double bond (R=alkyl) give an eight-membered heterocycle.
This reaction has been used in the preparation of a series of
pharmacologically active amines [ 104] and diamines [ 105]. A stereoselective
version of this reaction by using the directing group o-
diphenylphosphinobenzoic acid has been reported [ 106] (Figure 27).
Figure 44 Hydroaminomethylation of differently functionalized alkenes.
6.6.5 Consecutive hydroformylation-aldol reaction
Aldehydes generated in the hydroformylation reaction in the presence of silyl

enol ethers, enamines or enolates undergo consecutive aldol reaction. Thus,
under usual hydroaminomethylation conditions the siladiene 176 leads to the
sylacyclohexane 182. The process is make up of a sequence of reactions and the
proposed mechanism is the following: a) hydroformylation of 176 to give the
dialdehyde 177, b) 177 forms the rhodium enolate 178 (X=O) or the rhodium
metalloenamine (X= NR), c) further aldol reaction to give ,-unsaturated
aldehyde 179, d) double bond reduction to give the aldehyde 180, from which e)
enamine 181 is formed. This enamine is in equilibrium with 181. Subsequent
hydrogenation takes probably place through conformer 181 to give 182 which
then evolves to the conformationally more stable 182. The diastereoselectivity of
the process is determined in the last hydrogenation step. The regioselectivity in
176 Chapter 6
the 2,5-disubsituted sylacyclohexane is very high in unsymmetrically

disubstituted divinylsilanes, and the trans isomer is exclusively formed when
Figure 45. Consecutive hydroformylation-aldol reaction.
R2=Ph. The reaction provides better yields with cyclic secondary amines than
with acyclic ones [ 107].
Silylenol ethers such as 184 also undergo the hydroformylation-aldol reaction
to give the silylated aldol adducts 185 in good yields through a sequence of
reactions involving the hydroformylation of the alkene and the intramolecular
Mukaiyama type aldol reaction. [108]. Best results were achieved using the
trimethylsilyl group.
Different reaction products are obtained if the hydroformylation of the
unsturated ketone 183 is carried out in the presence of amines. With secondary
amines the hydroaminomethylation of the double bond is observed, leaving the
carbonyl group unaffected.
In the presence of benzylamine, ketone 183 is converted into the
aminoketone 186 by alkene hydroformylation, imine formation and aldol
reaction. Under more drastic hydroformylation conditions the reaction of 183
with benzylamine leads to the amine 187, which results from a mechanism
similar to those above including reductive amination of the ketone moiety.
185 186 187 188
i) [RhCl(cod)]2,CO/H2=1,80bar,90C,CH2Cl2. ii) [Rh(acac)(CO2)]/BlPHEPHOS.BnNH2,

CO/H2=2,30 bar, 60C.dioxane. iii)[RhCl(cod)]2/P(OPh)3, BnNH2. CO/H2=2, 90 bar, 120C,
dioxane.iv)[RhCl(cod)]2/P(OPh)3, iPrNH2, CO/H2=2,90bar,120C,dioxane.
Figure 46. Consecutive hydroformylation-aldol reactions in the presence of amines
Using bulkier primary amines such as isopropyl- or cyclohexylamine, no

aldol reaction is observed and, instead, heterocycles of type 188 are generated
via alkene hydroformylation, double amine condensation and reductive
amination of both carbonyl functions.
6.6.6 Consecutive hydroformylation-Wittig reaction
Consecutive hydroformylation-Witttig reaction is the last reported

consecutive reaction involving hydroformylation. As in the case of
hydroaminomethylation and hydroformylation-aldol reaction, the last step is also
a hydrogenation reaction [109].
The reaction is limited to stabilized ylides, because non stabilizided ylides
are too basic and induce rhodium inactivation.
Under hydroformylation conditions and in the presence of Ph3P=CHCOR,
compound 189 leads to the oxo derivative 190. The process involves a sequence
of reactions that includes initial hydroformylation to give the aldehyde 191 in a
stereoselective way (see also Figure 26), Wittig olefination to give the trans
conjugated alkene I92 and hydrogenation.
isubstituted ylides (i.e. PPh3=C(Me)COR) do not undergo hydrogenation and
in consequence ,-unsaturated ketones or esters are obtained. Ylides including
the ester function provide low yields. The stereoselectivity of the process Is
determined by the chelating (directing) group o-DPPB, and stereochemistries
all- syn, anti-syn, and all -anti can be obtained.
178 Chapter 6
191 192
Figure 47. Consecutive hydroformylation-Wittig reaction
6.7 Alkyne hydroformylation
While the hydroformylation of alkenes is an important industrial process

which also has enormous potential in organic syntheses, (see previous
section in this chapter), the hydroformylation of alkynes has been much less
investigated. This is because of the lack of general catalytic systems and the
low selectivity of the process, which leads to the formation of undesired
products [7, 110]. However, acetylenes are versatile intermediates in organic
synthesis, and to improve the efficiency of the hydroformylation should be
the main objective of research into this reaction.
The problem of the chemoselectivity is associated with the drastic
reactions conditions usually employed in the hydroformylation reaction.
Thus, the consecutive hydrogenation of unsaturated aldehyde gives
aldehydes or alcohols [111, [ 112] (Figure 48). No products of double
hydroformylation are obtained.
The regioselectivity is low except for the hydroformylation of acetylenic
bonds bearing bulky substituents on one side of sp carbons.
For years, the problem of hydroformylating alkynes was circumvented by
performing the reaction under water-gas shift conditions or by
silylformylation. However, some authors have recently effectively used
synthesis gas in alkyne hydroformylation.
Using CO/H2. Internal alkynes 197 are hydroformylated at room

temperature and 1 bar CO/H2 with the catalytic system [Rh]/BIPHEPHOS to
give excellent yields of ,-unsaturated aldehydes 198 [ 113].
Figure 49. Hydroformylation of acetylenes with CO/H2.
The hydroformylation of compounds which have a triple bond

conjugated with a double bond (199) takes place, contrary to what might be
expected, in the triple bond to give formyl dienes [ 114, 115]. The catalytic
system [RhH(PPh3)3] provides a mixture of diene 200 and cyclopentanone.
zw
However, a zwitterionic rhodium complex Rh /P(OPh)3 mainly provides
the formyl diene 200, together with the nonconjugated unsaturated aldehyde
201 as by product. This phosphite provides similar yields to BIPHEPHOS,
which, however, gives a more active catalyst.
199 R yield in 200=50-60% 200 201

+ 6 .
I)4 mol % [Rh (h -C6H5-B Ph3)(cod)]/4 P(OPh) 3, CO/H2=1, 12 bar, 60C, Ch2CI2,36-48h
Figure 50. Hydroformylation of alkene-acetylenes.
Terminal acetylenes give complex mixtures. The reaction also takes place
in the acetylene that shows that for these substrates the triple bond is more
reactive than the double bond.
The hyfroformylation of propargylamines 202 in the presence of the
classical catalytic system [Rh(OAc)2]2/PPh3 gives 2,4-disubstituted pyrroles
204 in excellent yields, through consecutive hydroformylation, cyclization
and double bond isomerization [ 116, 117].
180 Chapter 6
i) [Rh(OAc)2]2/PPh3, CO/H2=1, 30 bar, 70C. 20h.
Figure 51. Synthesis of pyrrole by hydroformylation of amino-acetylenes.
- Using CO/H2O (water-gas shift). The rhodium-catalyzed carbonylation
of alkynes under water-gas shift reaction conditions proceeds through a

different mechanism (type of reaction) than that under synthesis gas. Under
these conditions alkynes give a regioselective mixture of furanones [ 1 18,
119, 120, 121, 122]. Rhodium carbonyl compounds in the presence of
triethylamine are used as the catalyst. Triethylamine seems to be necessary
for the reaction to initiate because when it was absent the reaction did not
occur.
Of special interest is the hydroformylation of functionalized alkynes with
amino [123], formyl [124], phenol [125], benzyl alcohol [126] or
carboxylate groups [ 127] which leads to differently fuctionalized
heterocycles.
i) Rh6(CO)16NEt3, PCO= 100 bar, 180C, water-dioxane.
Figure 52. Synthesis of heterocycles by hydroformylation of phenylacetylenes.
The reaction gives compounds 206, 209, and 210, and proceeds through
the ,-unsaturated lactones which undergoes an in situ hydrogenation to
give the saturated lactones.
-Silylformylation. Carbonylation in the presence of hydrosilanes is

conceptually comparable to carbonylation under hydrogen pressure [ 128].
The simultaneous incorporation of a triorganosilyl group and a formyl group
into an acetylenic bond via rhodium catalyzed silylformylation [ 129], is an
excellent procedure for synthesizing -silylenals with high regio- and
chemoselectivity [ 130, 13 1, 132, 133]. -Silylenals are precursors of silyl-
substituted dienes [ 134], dienones [ 135] and ,-unsaturated ketones [ 136].
Rhodium carbonyl complexes in the presence of amines are used as
catalytic systems, although zwitterionic rhodium complex has also provided
excellent yields.
The regioselectivity in the silylformylation of terminal and internal
acetylenes seems to be governed in general by the steric bulkiness of the
substituents. Thus the bulky silyl group goes to the least hindered carbon,
and consequently the formyl group is introduced into the most substituted
one [137]. The electronic effect is the major factor that determines the
regiochemistry of propiolate derivatives. Hydrosilylation is a competitive
reaction.
The substituents of alkynes and hydrosilanes affect the reaction rate. The
reactivity of the alkynes decreased in the order phenylacetylene > 1-hexyne
> (trimethylsilyl)acetylene >> 3,3-dimethyl- 1 -butyne, whereas for the
hydrosilane was MePh2SiH, Me2PhSiH >Et2MeSiH, Et3SiH > tBuMe 2SiH
>> iPr3SiH. Thus, the rate of silylformylation largely depends on the
combination of alkynes and hydrosilanes.
R= H 90% (Z:E= 8:2)

R= Me 74% 8%
i) Rh4(CO)12/Et3N. PhMe2SiH,25C, benzene, 15h.
Figure 53. Silylformylation of phenylacetylenes.
Regioselectivity in acetylenes with a dimethylsilyloxy moiety is inverted

and high, since the transfer of the silylgroup is intramolecular [138]. The
reaction works with open chain as well as with cyclic compounds (Figure
54). Thus, differently -(dimethylsilyloxy)alkynes reacts under
hydroformylation conditions to give the corresponding 3 -exo -(formyl-
methylene)oxasilacycloalkanes in excellents yields.
182 Chapter 6
n=0 69% (only product)

n=1 73% (only product)
Figure 54. Intramolecular silylformylation.
The reaction of 3-(dimethylsilyloxy)- 1 -propyne does not proceed at all.

But the reaction of of 3-(dimethylsilyloxy)-2-heptyne a 3-exo-( 1 -
formylpentylid- 1 -ene) 1 -oxa-2-silacyclobutane. Oxa-silacycles of 4-7-
member ring were reported. When starting from cyclic compounds bicyclic
compounds are obtained.
Alkynes can also undergoes consecutive process under silylformylation
conditions. Thus, rhodium catalyzed silylformylation of alkyne 217 in the
presence of primary or secondary amines leads directly to the azadiene 221
by silylformylation and enamine formation [ 139]. These azadienes
undergoes Diels-Alder reaction with dimethyl acetylendicarboxilate to give
dihydropyridines 222.
i) 1 mol % [Rh(cod)2]BPh 4,R'NH 2.R3SiH, T=60C, toluene, 22h.
Figure 55. Silylformylation-amination of acetylenes.
6.8 Concluding remarks
The intense investigation of recent years in hydroformylation has allowed

the best catalytic precursors to be selected, and new ligands have been found
6. Hydroformy la t ion in organic synthesis 183
that provides highly active and selective catalysts. The incorporation of a

ligand in the substrate now means that there are new ways of controlling the
regio- and the stereoselectivity of the reaction. The compatibility of
hydroformylation with many organic functionalities and the possibility of
performing consecutive reactions with high selectivity also add to the value
of the reaction. However, some of the selected examples in this chapter do
not use the best catalytic precursors and ligands. Readers should, therefore,
take into account all the information that has been accumulated in recent
years, since at the present moment appropriate catalyst and reaction
conditions can be chosen for hydroformylation of each substrate. We highly
recommend that you read the various chapters in this book about the role of
ligands and other criteria in selectivity before selecting a catalyst. Although
new discoveries are expected in the different aspects of hydroformylation in
the coming years, the present situation of knowledge of this reaction allows
its use as an efficient synthetic tool in organic synthesis.
References
1 a) Gladiali, S.; Bayn, J. C.; Claver, C. Tetrahedron: Asymmetry 1995, 6, 1453. b)

Agbossou, F.; Carpentier, J. F.; Mortreux, A. Chem Rev. 1995, 95, 2458. c) Botteghi, C.;
Paganelli. S.; Schionato, A.: Marchetti, M. Chirality 1991, 3, 355. d) Eilbracht, P. in
Methoden der Organischen Chemie Stereoselective Synthesis, Vol. E21c, sub vol 4.
(Eds. Helmchen, G.; Hoffmann, R.W.; Mulzer, I.; Schaumann, E.) Houben-Weyl,
Thieme, Sttutgart, 1996, p. 2488. e) Nozaki, K.; Takaya, M.; Hiyama, T. Topics in
Catalysis 1997, 4, 175. f) Nozaki, K. in Comprehensive Asymmetric Catalysis, vol 1,
(Eds. Jacobsen, E. N.; Pfaltz, A.; Yamamoto, H.), Springer, Berlin, 1999, p. 381. g)
Consiglio, G. in Catalytic Asymmetric Synthesis, (Ed. Ojima, I.), VCH, Winheim, 1993,
p. 273. h) Noyori, R. Asymmetric Catalysis in Organic Synthesis, Wiley, New York,
1994.
2 Eilbracht, P.; Barfacker, L.; Buss, C.; Hollmann, C.; Kitsos-Rzychon, B. E.; Kranemann,
C. L.; Rische, T.; Roggenbuck, R.; Schmidt, A. Chem. Rev. 1999, 99, 3329.
3 Ojima, Y.; Tzamarioudaki, M.; Li, Z.; Donovan, R. J. Chem. Rew. 1996, 96, 635.
4 a) Botteghi, C.; Ganzerla, R.; Lenarda. M.; Moretti, G. J. Mol. Catal. 1987, 40, 129. b)
Siegel, H.; Hinimele, W. Angew. Chem. Int. Ed. Engl. 1980, 19, 178. c) Comils, B.;
Henmann, W. A. Applied Homogeneous Catalysis with Organometallic Compounds,
VCH, Weinheim, 1996. d) Beller, M.; Cornils, B.; Frohming, C. D.; Kohlpaintner, C.
W. J. Mol. Catal. 1995, 104, 17. e) Cornils, B.; Herrmann, W. A.; Kohlpaintner, C. W.
Angew. Chem. Int. Ed. Engl. 1994, 33, 2144. f) Botteghi, C.; Marchetti, M.; Paganelli, S.
in Transition Metals for Organic Synthesis, (Eds. Beller, M.; Bolm, C.), Wiley-VCH,
1998, p. 25. g) Comils B. in New Synthesis with carbon monoxide, (Ed. Falbe, J.)
Springler, Berlin, 1980, p. 1.
5 Lazzaroni, R.; Settambolo, L. R.; Uccello-Barreta, G. Organometallics 1995, 14,4644.
6 Botteghi, C.; Branca, M.; Marchetti, M.; Saba, A. J. Organomet. Chem. 1978, 197, 161.
184 Chapter 6
7 Brown, C. K.; Wilkinson, G. J. Chem. Soc., A. 1970, 2753.

8 van Leeuwen, P. W. N. M.; Roobeek, C. F. J. Organomet. Chem. 1983, 258, 343.
9 Billing, E.; Abatjoglou, A. G.; Bryant, D. R. U.S. Patent 4769 498, 1988.
10 Cuny, C. D.; Buchwald, S. L. J. Am. Chem. Soc. 1993, 115, 2066.
11 Cuny, C. D.; Buchwald, S. L. Synlett 1995, 519.
12 Dewar, T. J.; Philips, G.W.; Puckette, T.A.; Stavinoha, J. L.; Vanderbilt, J.J. Eastman
Kodak Company, US 4694109,1987.
13 Kranenburg, M.; Kamer, P. C. J.; van Leeuwen, P. W. N. M. Eur. J. Inor. Chem. 1998,
155.
14 van der Veen, L. A.; Kamer, P. C. J.; van Leeuwen, P. W. N. M. Angew. Chem. Int. Ed.
Engl. 1999, 38, 336.
15 Beller, M.; Zimmermann, B.; Geissler, H. Chem. Eur. J. 1999, 5, 1301.
16 Matsui, Y.; Orchin, M. J. Organomet. Chem. 1983, 57, 246.
17 a) Botteghi, C.; Cazzolato, L.; Marchetti, M.; Paganelli, S. J. Org. Chem. 1995, 60,
6612. b) Botteghi, C.; Marchetti, M.; Paganelli, S.; Sechi, B. J. Mol. Catal. A 1997, 118,
173.
18 a) Kleemann, A.; Engel, J. Pharmazeutische Wirkstoffe, George Thieme Verlag,
Stuttgart, 1987. b) Elks, J.; Ganellin, G. R.; Dictionary of Drugs, Chapman Hall,
London 1990.
19 von Kleemann, A. Chem. Ztg., 1977, 101, 389.
20 Takahashi, T.; Machida, K.; Kido, Y.; Nagashima, K.; Ebata, S.; Doi, T. Chem. Lett.
1997, 1291.
21 Botteghi, C.; Soccolini, F. Synthesis 1985, 592.
22 a) Fell, B.; Barl, M. J. Mol. Catal. 1977, 2, 301. b) Abatjoglou, A. B.; Bryant, D. R.;
DEsposito, L. C. J. Mol. Catal. 1983, 18,381.
23 a) Beker, Y.; Eisenstadt, A.; Stille, J. K. J. Org. Chem. 1980, 45, 2145. b) Delogu, G.;
Faedda, G.; Gladiali, S. J. Orgunornet. Chem. 1984, 268, 167.
24 a) Ojima, I. Chem. Rev. 1988, 88, 101 1. b) Fuchikami, T.; Ojima, I. J. Am. Chem. Soc.
1982, 104, 3527.
25 Crameri, Y.; Ochsner, P. A.; Schudel, P. Eur. Pat. Appl. EP52775, 1982
26 Matsumoto, M.; Tamura, M. J. Mol. Catal. 1982, 16, 195.
27 Polo, A.; Claver, C.; Castillon, C.; Ruiz, A.; Bayn, J. C.; Real, J.; Mealli, C.; Massi, D.
Organometallics 1992, 1 I, 3525.
28 Kawabata, Y.; Suzuki, T. M., Ogata, L. Chem. Lett. 1978, 361.
29 Hayashi, T.; Tanaka, M.; Ogata, L. J. Mol. Catal. 1981, 13, 323.
30 Lai, R.; Ucciani, E. J. Mol. Catal. 1978, 4,401.
31 Botteghi, C.; Paganelli, S.; Bigini L.; Marchetti, M. J. Mol. Catal. 1994, 93, 279.
32 Botteghi, C.; Chelucci, C.; del Ponte, G.; Marchetti, M.; Paganelli, S. J. Org. Chem.,
1994, 59, 7125.
33 Settambolo, R.; Scamuzzi, S.; Caiazzo, A.; Lazzaroni, R. Organometallics 1996, 17,
2127.
34 Caiazzo, A.; Settambolo, R.; Pontomo, L.; Lazzaroni, R. In press
35 Caiazzo, A.; Settambolo, R.; Uccello-Barreta, G.; Lazzaroni, R. J. Organomet. Chem.
1997, 548 ,279.
36 Totland, K.; Alper, H. J. Org. Chem. 1993, 58, 3328.
37 Kurkov, V. P.; U. S. Patent, 4344896 1982, Chem. Abst. 1982, 97, 181731.
38 Yamashita, H.; Roan, B. L.; Sakakura, T.; Tanaka, M. J. Mol. Catal. 1993, 83, 255.
39 Consiglio, G.; Kollr, L.; Klliker, R. J. Organomet. Chem. 1990, 396, 375.
40 Kollr, L.; Consiglio, G.; Pino, P. J. Organomet. Chem. 1990, 386, 389.
41 Stefani, A.; Consiglio, G.; Botteghi, C.; Pino, P. J. Am. Chem. Soc. 1973, 95, 6504.
42 Himmele, W.; Siegel, H. Tetrahedron Lett. 1976 ,907.
43 Siegel, H.; Himmele, W. Angew. Chem. Int. Ed. Engl. 1980, 19, 178.
44 Azzaroni, F.; Biscarini, P.; Bordoni, S.; Longoni, G.; Venturini, E. J. Organomet. Chem.
1996, 508, 5958.
45 Leighton, J. L.; ONeil, D. N. J. Am. Chem. Soc. 1997, 119, 11118.
46 Sarraf, S. T.; Leighton, J. L. Tetrahedron Lett. 1998, 39, 6423.
47 Rosenthal, A. Adv. Carbohydr. Chem. 1969, 24, 59.
48 Rosenthal, A.; Abson, D.; Field, T. D.; Koch, H. J.; Mitchel, R. E. Can. J. Chem. 1967,
45, 1525.
49 Ferndez, E.; Ruiz, A.; Claver, C.; Castilln, S.; Polo, A.; Piniella, J. F.; Alvarez-
Larena, A. Organometallics 1998, 17, 2857.
50 Takahashi, T.; Ebata, S; Yamada, H. Synlett. 1998, 38 1.
51 Breit, B.; Zahn, S. K. Tetrahedron Lett. 1998, 39, 1901.
52 a) Kraft, E. M.; Wilson, L. J. Organometallics 1988, 7, 2528. b) Kraft, E. M.
Tetrahedron Lett. 1989, 301, 539.
53 Jackson, W. R.; Perlmutter, P.; Tasdelen, E. E. J. Chem. Soc., Chem. Commun. 1990,
763
54 Jackson, W. R.; Perlmutter, P.; Tasdelen, E. E. Tetrahedron Lett. 1990, 31, 2461.
55 Jackson, W. R.; Perlmutter, P.; Suh, G. H. J. Chem. Soc., Chem. Commun. 1987, 724.
56 Burke, S. D.; Cobb, J. E. Tetrahedron Lett. 1986, 27,4237.
57 Burke, S. D.; Cobb, J. E.; Takeuchi, K. J. Org. Chem. 1990, 55,2138.
58 Breit, B. Angew. Chem. Int. Ed. Engl. 1996, 35, 2835.
59 Breit, B. Liehigs Ann. Recueil 1997, 1841.
60 Breit, B.; Dauber, M.; Harms, K. Chem. Eur. J. 1999, 5, 2819.
61 Breit, B. Chem. Commun. 1997, 591.
62 Breit, B. Eur. J. Org. Chem. 1998, 1123.
63 Pittman, C. U. Jr.; Honnick, W. D. J. Org. Chem 1980, 45, 2132.
64 Wuts, P. G. M.; Obrzut, M. L.; Thompson, P. A. Tetrahedron Lett. 1984, 25, 4051.
65 Smith, W. E.; Chambers, G. R.; Lindberg, R. G.; Cawse, J. N.; Dennis, A. J.; Harrison,
G. E.; Bryant, D. R. In Catalysis of Organic Reactions; (Ed. Augustine, R. L.) Dekker,
New York, 1985, p.151.
66 Nozaki, K.; Li. W.; Horiuchi, T.; Takaya, H. Tetrahedron Lett. 1997, 38, 4611.
67 Trzeciak, A. M.; Wolszcak, E.; Ziolkwsky, J. New J. Chem. 1996, 20, 365.
68 Anastasiou, D.; Jackson, W. R. Aust. J. Chem. 1992, 45, 21.
69 Anastasiou, D.; Jackson, W. R.; McCubbin, Q. J.; Tmacek, A. E. Aust. J. Chem. 1993,
46, 1623.
70 Baker, R.; Herbert, R. H. Nat. Prod. Rep. 1984, 43, 3309.
71 Jaramillo, C.; Knapp, S. Synthesis 1994, 1.
72 Sirol, S.; Kalck, Ph. New. J. Chem. 1997, 21, 1129.
73 Kitsos-Rzychon, B.; Eilbracht, P. Tetrahedron 1998, 54, 10721.
74 Parriello, G.; Stille, J. K. J. Am. Chem. Soc. 1987, 109, 7122.
75 Fernndez, E.; Castilln, S. Tetrahedron Lett. 1994, 35, 2361.
76 Balu, J.; Bayn, J. C. J. Mol. Catal. 1999, 137, 193.
77 Soulantica, K.; Sirol, S.; Koinis, S.; Pneumatikakis, G.; Kalck, Ph. J. Organomet. Chem.
1995, 498, C 10.
78 Fernndez, E.; Polo, A.; Ruiz, A.; Claver, C.; Castilln, S. Chem. Commun. 1998, 1803.
186 Chapter 6
79 Ojima, I.; Tzamarioudaki, M.; Eguchi, M. J. Org. Chem. 1995, 60, 7078.
80 Campi, E. N.; Jackson, W. R.; McCubbin, Q. J.; Tmacek, A. E. Aust. J. Chem. 1996, 49,
219.
81 Campi, E. N.; Habsuda, J.; Jackson, W. R.; Jonasson, C. A. M.; McCubbin, Q. Aust. J.
Chem. 1995, 48, 2023.
82 Ojima, I.; Zhang, Z. J. Org. Chem. 1988, 53, 4425.
83 Ojima, I.; Zhang, Z. J. Organomet. Chem. 1991, 417, 253.
84 Ojima, 1.; Korda, A. Tetrahedron Lett. 1989, 30, 6283.
85 Ojima, I.; Korda, A.; Shay, W. R. J. Org. Chem. 1991, 56, 2024.
86 Ojima, 1.; Vidal, E. S. J. Org. Chem. 1998, 63, 7999.
87 Ojima, I.; Iula, D. M.; Tzamarioudaki, M. Tetrahedron Lett. 1998. 39, 4599.
88 Anastasiou, D.; Jackson, W. R. J. Chem. Soc., Chem. Commun. 1990, 1205.
89 Anastasiou, D.; Chaouk, H.; Jackson, W. R. Tetrahedron Left. 1991, 32, 2499.
90 Anastasiou, D.; Campi, E. M.; Chaouk, H.; Jackson, W. R. Tetrahedron 1992, 48, 7467.
91 Bergmann, D. J.; Campi, E. M.; Jackson, W. R.; Cubin, Q. J.; Patti, A. F. Tetrahedron Lett.
1997, 38, 4315.
92 Bergmann, D. J.; Campi, E. M.; Jackson, W. R.; Patti, .A. F. Sylik, D. Tetrahedron Lett.
1999, 40 5597.
93 Bergmann, D. J.; Campi, E. M.; Roy Jackson, W.; Patti, A. F. Chem Commun. 1999. 1279.
94 a) Reppe, W. Experientia 1949, 5, 93. b) Reppe, W.; Vetter, H. Liebigs Ann. Chem. 1953,
582, 133.
95 Rische, T.; Eilbracht, P. Synthesis 1997, 1331.
96 Baig, T.; Kalc, Ph. J. Chem. Soc., Chem. Commun. 1992, 1373.
97 Baig, T.; Molinier, J.; Kalc, Ph. J. Organomet. Chem. 1993, 455, 219.
98 Eilbracht, P.; Kranemann, C. L.; Brfacker, L. Eur. J. Org. Chem 1999, 1907.
99 Rische, T.; Eilbracht, P. Tetrahedron 1999, 55, 3917.
100 Rische, T.; Barfacker, L.; Eilbracht, P. Eur. J. Org. Chem. 1999, 653.
102 Barfacker, L.; Rische, T.; Eilbracht, P. Tetrahedron 1999, 55, 7177.
103 Kranemann, C. L.; Kitsos-Rzychon, B.; Eilbracht, P. Tetrahedron 1999, 55, 4721.
105 Rische, T.; Muller, K-S.; Eilbracht. P. Tetrahedron 1999, 55, 9801.
106 Breit, B. Tetrahedron Lett. 1998, 39, 5163.
107 Barfacker, L.; El Tom, D.; Eilbracht, P. Tetrahedron Lett. 1999, 40, 4031.
108 Hollman, C.; Eilbracht, P. Tetrahedron Lett. 1999, 40, 4313.
109 Breit, B.; Zahn, S. K. Angew. Chem. Int. Ed. 1999, 38, 969.
110 Pino, P.; Barca, G. in Organic Synthesis via Metal Carbonyls, vol 2, (Eds. Wender, I.;
Pino, P,), Wiley-Interscience, New York, 1977, p. 419.
111 Tkatehenko, I. in Comprehensive Organometallic Chemistry. vol. 8, (Eds. Wilkinson,
Sir G. Stone, F. G. A.; Abel, E.), Pergamon Press, Oxford, 1982.
112 Botteghi, C.; Salomon, Ch. Tetrahedron Lett. 1974, 4285.
113 Johnson, J. R.; Cuny, G. D.; Buschwald, S. L. Angew. Chem. Int. Ed. Engl. 1995, 34,
1760.
114 Doyama, K.; Joh, T.; Takahashi, S. Tetrahedron Lett. 1986, 27, 4497.
115 Doyama, K.; Joh, T.; Shiohara, T.; Takahashi, S. Bull Chem. Soc. Jpn. 1988, 61, 4353.
116 Campi, E. M.; Jackson, W. R. Aust. J. Chem. 1989, 42, 471.
117 Campi, E. M.; Jackson, W. R.; Nilsson, Y. Tetrahedron Lett. 1991, 32, 1093.
1 I8 Joh, T.; Doyama, K.; Onitsuka, T.; Shiohara, S.; Takahashi, S. Organometallics 1991,
10, 2493.
119 Laine, R. M.; Crawford, E. J. J. Mol. Catal. 1988, 44, 357.
120 Cavinato, G.; Toniolo, L. J. Mol. Catal. 1996, 105, 9.
121 Joh, T.; Fujiwara, K.; Takahashi, S. Bull. Chem. Soc. Jpn. 1993, 66, 978.
122 Joh, T.; Doyama, K.; Fujiwara, K.; Maeshima, K.; Takahashi, S. Organometallics 1991,
10, 508.
123 Hirao, K.; Morii, N.; Joh, T.; Takhashi, S. Tetrahedron Lett. 1995, 36, 6243.
124 Sugioka, T.; Zhang, S.-W.; Mori, N.; Joh, T.; Takahashi, S. Chem. Lett. 1996, 249.
125 Yoneda, E.; Sugioka, T.; Hirao, K.; Zhang, S.-W.; Takahashi, S. J. Chem. Soc., Perkin
Trans. I 1998, 477.
126 Yoneda, E.; Kaneko, T.; Zhang, S-W.; Takahashi, S. Tetrahedron Lett. 1998, 39, 5061.
127 Sugioka, T.; Yoneda, E.; Onitsuka, K.; Zhang, S-W.; Takahashi, S. Tetrahedron Lett.
1997, 38 , 4989.
128 a) Murai, S.; Sonoda, N. Angew. Chem. Int. Ed. Engl. 1979, 18, 837. b) Chatani, N.;
Kajiwata, Y.; Nishimura, H.; Murai, S. Organometallics 1991, 10, 21. c) Chatani, N.;
Murai, S. Synlett 1996, 414.
129 Matsuda, I.; Ogiso, A.; Sato, S.; Isumi, Y. J. Am. Chem. Soc. 1989, 111, 2332.
130 a) Monteil, F.; Matsuda, I.; Alper, H. J. Am. Chem. SOC. 1995, 117, 4419. b) Ojima, I.,
Vidal, E.; Tzamarioudaki, M.; Matsuda, I. J. Am. Chem. SOC. 1995, 117, 6797.
131 Ojima, I.; Fracchiolla, D. A.; Donovan, R. J.; Banerji, P. J. Org. Chem. 1994, 59, 7594.
132 a) Doyle, M. P.; Shanklin, M. S. Organometallics 1993, 12, 11. b) Doyle, M. P.;
Shanklin, S. Organometallics 1994, 13, 1081.
133 Zhou, J.-Q.; Alper, H. Organometallics 1994, 13, 1586.
134 a) Jung, M. E.; Gaede, B. Tetrahedron 1979, 35, 621. b) Carter, M. J.; Fleming, I.;
Percival, A. J. Chem. Soc., Perkin Trans. I 1981, 2415.
135 Pillot, J.; Dunogues. J.; Calas, R. J. J. Chem. Res. Synop. 1977, 268.
136 Fleming, I.; Perry, D. A. Tetrahedron 1981, 37, 4027.
137 Matsuda, I.; Fukuta, Y.; Tsuchihashi, T.; Nagashima, H.; Itoh, K. Organometallics
1997, 16, 4327.
138 Ojima, I.; Vidal, E.; Tzamarioudaki, M.; Matsuda, J. J. Am. Chem. Soc. 1995, 117,
6797.
139 Brfacker, L.; Hollmann, Ch.; Eilbracht, P. Tetrahedron 1998, 54, 4493.
Chapter 7
Aqueous biphasic hydroformylation
Jrgen Herwig and Richard Fischer

Celanese GmbH, Werk Ruhrchemie. Research & Technology, D-46128 Oberhausen, Germany
7.1 Principles of biphasic reactions in water
7.1.1 Why two-phase catalysis? Scope and Limitations
What has to be achieved to combine the major important advantages of

homogeneous and heterogeneous catalysis? It seems to be a non-complex
problem with a simple solution: find a way of separating the catalyst and the
reaction products under mild conditions that is ecologically as well as
economically efficient. However, to get there is not that simple. Normally
the organometallic catalysts applied in solution are not very stable against
high temperatures, water or oxygen. Usually, reaction products, especially
bulk chemicals, are separated in two ways. Firstly, by gas phase separation
technologies like distillation as applied for e.g. the synthesis of acetic acid
by the carbonylation of methanol or the hydroformy lation of alkenes to
aldehydes. Secondly, by reaction conditions that keep the reaction partners,
starting materials, and the products in the gas phase as e.g. for the gas phase
process for the production of phthalic anhydride. In the case of the
production of terephthalic acid the product is insoluble in the reaction
medium acetic acid, in which the Co and Mn based oxidation catalyst are
dissolved. This way it is possible to separate the product from the catalyst
solution by simple filtration. However, such beneficial circumstances occur
rather seldom, as mixtures of liquid reaction media and liquid products are
the rule. Thus it is a major step forward if catalytic reactions can be
performed in liquid multi-phase systems with clean and fast separation of
189
P. W.N.M. van Leeuwen and C. Claver (eds.), Rhodium Catalyzed Hydroformylation, 189-202.
190 Chapter 7
reactant and product phase. See Chapter 8 for an overview of the other
separation techniques in use for hydroformylation.
In contrast to heterogeneous catalysts, their homogeneous counterparts
don't resist high temperatures or harsh work-up conditions mostly due to
damage of weak carbon-metal bonds or the loss of ligand stabilization of the
central metal atom of the complexes. The avoidance of (traces) of water or
air at such conditions is also often a parameter that is critical to the success
of the systems. Thus the conditions of the separation of the reaction product
from the solution, in which the catalyst has been dissolved, must be as mild
as possible. A textbook example of such a modern technology is the
application of two- or multi-phase reaction systems. One of the most
important developments in 1980-2000 in the area of homogeneous catalysis
is the successful development of water-stable as well as highly water-soluble
catalyst systems and the consecutive introduction of the aqueous two-phase
technology.
What limitations have to be accounted for if multi-phase systems are
applied is shown in the following listing of possible disadvantages:
- low reaction rates and the resulting low space-time yields due to
insufficient solubility of the reactants or complex interface processes such
as mass-transport limitations,
- surfactant effects that might favor the enrichment of the reaction product
or side products in the reaction media,
- slow or incomplete phase separation,
- the need of high reaction volumes,
- corrosion effects,
- decrease of the chemoselectivity (e.g. pH-controlled side-reactions, etc.).
7.1.2 Concepts for two-phase hydroformylation
Basis for two-phase catalysis is the presence of two immiscible phases.

These phases can consist of:
- water (+ co-solvent) - organic liquid
- polar organic liquid - non polar organic liquid
- ionic liquid - organic liquid
- fluorous organic phase - organic liquid
The first concept has been realized in the Ruhrchemie-Rhne Poulenc
process for propene and butene hydroformylation. This process will be
discussed in detail below. The catalyst is retained in the aqueous phase by
applying a water-soluble phosphine ligand. The product constitites the
organic solvent.
The second concept is realized in the Shell Higher Olefins Process for the
oligomerization of ethylene to linear higher alkenes. Butane-1,4-diol is used
7. Aqueous biphasic hydroformylation 191
as the solvent for the Ni based catalyst. The linear higher alkenes are not
soluble in the catalyst phase and can thus easily be separated. This concept is
not applied in hydro formylation.
The third concept is used in an IFP process for the dimerisation of
alkenes. Hydroformylation has been successfully conducted in ionic liquids
in the laboratory [ 1].
Fluorinated organic liquids, concept four, have not been applied in
industry, because of technical and economic disadvantages (expensive
ligands and solvents). Hydroformylation has been investigated in fluorinated
solvents [2] (see Chapter 10).
7.2 Hydroformylation of propene and butene
7.2.1 Historic overview of two-phase hydroformylation technology
In 1982, Rhne Poulenc and Ruhrchemie AG (now part of Celanese AG)

joined to develop a continuous aqueous biphasic hydroformylation process.
The work was driven by the fact that Rh was much more efficient in
hydroformylation of short chain alkenes than cobalt; Rhodium gives lower
raw material consumption, lower cost for catalyst recycling, and a higher 1:b
ratio. In order to stay competitive Ruhrchemie as a leading oxo producer
decided to switch to a Rh based system. Furthermore, a two-phase system
could offer several advantages over a homogeneous system e.g. lower
catalyst loss, lower heat stress of the catalyst, easy catalyst separation from
the products and higher energy efficiency.
Following the laboratory results of several, biphasic catalytic reactions
(hydroformylation, hydrocyanation, and diene conversion) performed by E.
Kuntz and patented by Rhne Poulenc [3] the process work started. In only
two years time these laboratory results were transferred to a 100,000 t/a
plant in Oberhausen by a team managed by B. Cornils. The first aqueous
biphasic propene hydroformylation reactor came on stream in 1984.
7.2.2 Ligand developments
The water-soluble ligand used by Kuntz and Cornils is TPPTS [tri(m-

sulfonyl)triphenylphosphine trisodium salt], the properties of which are very
similar to the parent compound tpp. Sulfonation of triphenylphosphine,
neutralization, followed by pH-controlled selective extraction and re-
extraction procedures, yields an aqueous solution of the trisodium salt of
tris(m-sulfonyl)triphenylphosphine in good yields. The catalyst used is a
192 Chapter 7
rhodium complex of TPPTS, which is highly water-soluble as in the order of 1

kg of the ligand "dissolves" in 1 kg of water. The ligand forms complexes with
rhodium that are most likely similar to the ordinary triphenylphosphine
complexes (i.e. RhH(CO)(PPh3)3) [7].
Figure 1. The RCH/RP ligand for two-phase hydroformylation
Until today the ligand synthesis is object of permanent improvement and

investigation. Described by a quite simple reaction scheme (see below), the
sulfonation technology and the consecutive work up procedure require
significant know-how [17].
P(C6H5)3 + 3 SO3 P(C6H4SO3H)3
Almost two decades of experience with the technically applied system

demonstrated that it is of great importance to reduce by-product formation
during the ligand synthesis: not only the improved yields but even more so
the elimination of specific reaction impurities bear a large potential for
further economic savings. Ligand decomposition can lead to the formation of
inactive rhodium species. Ligand decomposition can take place via P-C-
bond cleavage, followed by formation of propyldi(sulfonylphenyl)-
phosphine, which acts as strong electron donor reducing the activivity of the
rhodium catalyst. To prevent fast catalyst deactivation [18] it turned out be
beneficial to control the P(III)/Rh ratio very carefully and never let it drop
below 60/1. In addition it was found that an almost continuously performed
addition of fresh ligand solution increases the catalyst life time (catalyst
activity and productivity) more than a third compared to the nominal life
time [19]. This pseudo-continuous ligand addition mainly compensates the
formation of phosphine oxides formed by traces of oxygen brought into the
system by the feedstock and the syn gas, which is produced by partial
oxidation of heavy oil or natural gas.
Besides the TPPTS-system a number of other sulfonated phosphine
ligand systems were investigated and tested on a pilot plant scale. Among
them are systems which are derived from biphenyl (BISBIS = sulfonated
bis(diphenylphosphinomethyl)biphenyl, varying grades of sulfonation of
BINAS
Figure 2. Formula of BINAS
BISBI, see Chapter 4) or binaphthyl structures (BINAS = sulfonated

NAPHOS, sulfonation grade between six and eight).
It turned out that rhodium-BINAS is the most active and selective water-
soluble hydroformylation catalyst [20]. Its reactivity is up to 10 times higher
than TPPTS, even low Rh/P-ratios (P/Rh = 7/1) give 1:b ratios of 98/2
(TPPTS: 94/6 with P/Rh = 80/1). Concerning the outstanding performance
data, BINAS would be clearly the ligand of choice. Compared to TPPTS,
these more complicated systems carry higher manufacturing costs. This
disadvantage could be compensated for by the lower amount of ligand
needed More importantly, BINAS shows higher decomposition rates than
TPPTS, which was up to now the overriding reason not to change the
established system. As a consequence up to now TPPTS is the best choice
for technical application in water based two-phase catalyst systems.
7.2.3 Kinetics and catalyst pre-formation
Kinetics.
When two-phase technology is applied, phase transfer phenomena can
play a dominant role in determining reaction rates and kinetics. The
hydroformylation of alkenes in a two-phase water/organic solvent system is
defined by the reaction of two gases, carbon monoxide and hydrogen, with
gaseous or liquid alkenes like ethene, propene or higher homologues.
Dependent on mass transport and reaction conditions (temperature, partial
pressures of the gases, stirring) and reactions partners (type of alkene), the
reaction of the alkene with CO and H2 takes place at the organic-aqueous
interface or in the aqueous bulk phase. No published data are available for
propene or butene.
One study has been carried out using octene-1, which is easy to handle,
thus yielding reliable results for the derivation of kinetic data, although
solubilities change with the composition of the system. Using the Rh/TPPTS
catalyst in an aqueous solution a rate law was determined for the two-phase
194 Chapter 7
hydroformylation of 1-octene to nonanal. The reaction rate was found to be

first order with respect to catalyst concentration. Also the dependency on the
alkene concentration is first order. At pressures below 5 bar a positive order
in the partial pressure of hydrogen pressure was found, but at pressures
above 10 bar this levels off to much lower values. Likewise, at pressures
below 5 bar a positive order in partial CO pressure was found, but at
pressures above 5 bar the usual retarding effect of increasing pressure was
found (Chapter 4). The rate law for the hydroformylation of 1-octene in an
aqueous two-phase system was described with the following equation, based
on the accepted mechanism of hydroformylation [4]:
k K1K2K3 [Rh ][ octene ][H2 ][CO ]

R=
1+[CO]
with k = reaction rate constant, K1, K2, K3, K4 = constants. At higher

pressures of dihydrogen the graphs seem to indicate a leveling off of the rate
increase with hydrogen pressure. If so, the rate-determining step would be
similar to the TPP system (Chapter 4).
For lower hydrogen partial pressures the following rate expression was
developed (hydroformylation of n-octene- 1 with ethanol as co-solvent to
increase the octene solubility [5]):
k[alkene]o[Rh][H2][CO]
Ro=
(1+KH2[H2])(1+Kco[CO]2)
The mixing of the reaction phases is also important for the reaction rate
so as to provide maximum liquid-liquid interfacial area in cases where the
reaction occurs mostly at the liquid-liquid interface and not in the bulk
aqueous phase. This can be achieved by high agitation speeds, optimized
stirrer type and geometry. In aqueous two-phase systems also the hold-up of
the aqueous phase plays an important role on the reaction rate. With too high
hold-up volumes the reaction rate starts to decrease (phase inversion yielding
the dispersed phase as reaction place which would be, in this case, the
organic phase). Thus it is recommended to keep the aqueous catalyst
solution at an optimized level to provide high productivity with regard to the
aldehyde formation. When using co-solvents like lower alcohols,
polyethylene glycol or acetonitrile the biphasic character of the system is
retained. The rate of the hydroformylation can be enhanced by several times,
albeit at the cost of a lower selectivity as a result of e.g. formation of acetals
or ethers.
7. Aqueous biphasic hydroformyla tion 195
Naturally, applying aqueous two-phase systems, the pH-value plays a

crucial role for catalyst activity. High reaction rates are observed at high pH
values (pH = 10, basic conditions). Also here the influence of catalyst
concentration on the reaction rate is highly beneficial (up to five times rate
increase with only tripling the catalyst concentration with hydroformylation
of 1-octene as an example) [6]. Increase of CO pressure under these
conditions is positive on the rate, but at too high CO partial pressures the rate
drops.
At low pH values (7 and lower, acidic conditions) the catalyst activity is
lower. Here an increase of catalyst concentration or CO partial pressure can
help only little. Thus hydroformylation is faster at higher pH values, but at
high pH-values severe side reactions of the reaction product can occur. For
example aldol condensation of butyric aldehydes can occur to form different
types of aldols and higher condensed systems that lead to an inferior quality
of the finished products like C4- aldehydes, alcohols, or 2-EH (2-ethyl-1-
hexanol).
In summary, the kinetics of hydroformylation in an aqueous/organic two
phase system applying water soluble Rh-complexes as catalysts are
dependent on the type of the Rh/ligand system, the pH-value, as well as on
the usage of additives and co-solvents. In particular the kinetics with respect
to the influence of the CO partial pressure varies between the different
systems. Of course the solubility of the alkene in the aqueous phase carrying
the catalyst plays a major role. Therefore, to achieve economically relevant
space time yields, higher alkenes require (i) co-solvents, (ii) catalyst binding
ligands like TPP, which facilitates the interfacial catalytic reaction by
increasing the concentration of the Rh-species at the interfacial area, or (iii)
the application of supported aqueous phase catalysis (Chapter 10).
Catalyst preparation / preformation

Of outstanding importance for the Rh/TPPTS system is the catalyst
preparation and pretreatment prior to the continuous use, in order to
guarantee a high catalyst lifetime and sufficient performance over the whole
runtime. Thus a carefully controlled TPPTS ligand synthesis, especially the
work up of the sulfonation mixtures, is a prerequisite to provide optimal
conditions during the hydroformy lation reaction.
Prior to the conversion of the alkene, the Rh/TPPTS catalyst system has
to be pre-formed (formation of the Rh-hydrido-carbonyl complex) applying
a specific chemical procedure is necessary to form the active Rh(1) species,
mostly starting from Rh(III) precursors. In general this is achieved under
hydroformylation conditions. Thus, treatment of the rhodium precursor with
syn gas in the absence of alkene for a couple of hours will transform it into a
Rh-hydrido-carbonyl complex.
196 Chapter 7
To achieve high catalyst lifetimes it is beneficial to keep the reaction

temperature below a certain ceiling value like 130 C to minimize ligand
degradation. To stabilize the active catalyst species it turned out to be
efficient if ligand is added by a saw-tooth-type dosing during the whole
reaction run. This keeps the Rh/P(III) ratio at a level which minimizes the
catalyst degradation (see also chapter 9). At the end of a catalyst life -
marked by significant and constant drop of productivity - the complete
aqueous catalyst solution is taken out of the production system and the
reactor is refilled with fresh catalyst solution. The old catalyst will be
worked up with full Rh-recovery (>90%) and even recovery of the portion of
the not degraded or oxidized TPPTS.
7.2.4 Process description
The Ruhrchemie Rhne-Poulene-Process applies a catalyst - TPPTS/Rh -

which is capable to convert both propene and raffinate-II feedstocks to
butyraldehyde and valeraldehyde, respectively. The potential of this process
is its flexibility to convert propylene or butenes in the same reactor system.
The design of the Ruhrehemie/Rhne-Poulene (RCH/RP) reaction system
and catalyst recovery method is completely different from that of
homogeneous hydroformylation processes. In principle, the two-phase
RCH/RP process applies a stirred tank reactor, followed by a phase separator
and a strip column. The reactor containing the aqueous catalyst solution
(Rh/TPPTS) is fed with propylene and syn gas. The oxo crude product is
passed on to the phase separator where gaseous components are removed
and the liquids separated into two phases, the aqueous catalyst solution and
the organic aldehyde phase. The catalyst solution is recycled to the reactor
passing a heat exchanger system, where steam is generated. The organic
phase containing the crude aldehydes is stripped with syn gas to remove
dissolved propylene that is recycled to the reactor. The crude aldehyde
mixture is split into iso- and n-aldehyde in an aldehyde distillation unit.
In the past at the Ruhrchemie site, the portion of the not reacted propene,
present in the waste gas from the two phase oxo reactors, was converted to
butyric aldehyde in a vent reactor, which was based on high pressure cobalt
hydroformylation technology. In 1998 the old cobalt technology was
eliminated by the implementation of a high-pressure rhodium process giving
a more selective process yielding a cleaner reaction product. The rhodium
system allows now the possibility to steer the overall 1:b ratio within a
comfortable range to serve the market needs. The capacity of this high-
pressure rhodium-unit is 70,000 tons per year.
The TPPTS ligand easily carries all of the used Rh into the water phase
by forming Rh/TPPTS complexes analogous to the complexes known from
the simple Rh/TPP system. Due to the sufficient strong coordination of the
TPPTS-ligand to the Rh metal-center almost no Rh leaching into the organic
phase is observed - a performance prerequisite with respect to the desired
economy of the process. Even with the low solubility of propene or butene in
water, the high reactivity of the used catalyst system is high enough to
provide a very efficient hydroformylation rate. To increase the solubility of
propylene in the water phase, plant trials were performed with polyethylene
glycol with various chain lengths, acting as an auxiliary-agent for better
propylene solubility. Good reaction rates (rate increase of more than 10 %)
were achieved, demonstrating a simple tool to increase productivity or to
switch to milder conditions. In summary, the major advantages of the
TPPTS-system are the simple, automatically performed mild catalyst
recycle, the comparable low Rh-inventory (rhodium savings), the high 1:b
ratio of ~20/1 and the beneficial low energy consumption of the whole
process as well as the improved utilization of propylene and synthesis gas
(CO/H2).
The water-based process is normally operated at a pH-value between 5
and 6 to provide reaction conditions, which depress unwanted side reactions
of the formed aldehydes. It turned out that a careful pH-control is beneficial
for various operation parameters of the protic reaction media: catalyst
reactivity and selectivity are directly influenced by and thus dependent on
the H3O+ concentration present in the aqueous phase.
7.2.5 Status of the operated plants
With the RCH/RP process it is possible to hydroformylate alkenes from

propene to pentene with satisfying space-time yields. Currently the process
is solely used for the conversion of propene and butene-1. In the early 1990s
developmental work was started to apply the RCH/RP process also to the
conversion of alkenes higher than C3. The overall economics looked
attractive and the market was prepared. Consequently, a 12,000 tons unit
went on-stream at the Oberhausen site in 1995, to convert raffinate-II (a
mixture of butene- 1 and butene-2) into C5-aldehydes that are predominantly
converted to the corresponding alcohols and acids in further downstream
processes at the site.
Compared to propene, butenes show a lower solubility in water. Due to
this the observed hydroformylation reaction rate of butene mixtures like
raffinate-II (in general 40 % butene-1, 40 % butene-2, 20 % butane isomers)
is significantly lower than that found for propene. However, a slight increase
in the catalyst concentration is sufficient to achieve satisfying space-time
yields for the pro-duction of valeric aldehydes. As a very selective catalyst
the Rh/TPPTS-system does not catalyze the hydroformylation of butene-2
198 Chapter 7
providing a simple method for the single conversion of the terminal alkene.
The catalyst isomerizes a small fraction of butene-1 to butene-2. The latter is
not converted under standard reaction conditions. Thus the butanes and the
unconverted butene-2 are separated form the reaction products in a stripping
column. Advantageously the RCH/RP technology can be applied to the
hydroformylation of C3- as well as C4-alkenes in the same production unit;
only minor changes (e.g. reaction temperature, partial pressures) are
necessary to switch between the different feedstock streams [2 13. This is not
the case for alkenes with five or more carbon atoms. Nevertheless the latter
feedstocks correspond to approximately 25 % of the worldwide oxo-
business.
Today the Ruhrchetme/Rhne-Poulene Process (RCH/RP) is successfully
operated at two locations in the world. In the early 1980s a 100,000 ton scale
process was developed and commercialized at the Oberhausen site of
Ruhrchemie AG (now Celanese AG). The first unit was started up in 1984.
In 1988 and in 1999 two additional lanes went on stream. The actual
combined capacity of all three units at Oberhausen (combined with the
conversion of not reacted propylene in a high-pressure hydroformylation set
up) amounts to more than 500,000 tons of C4-products. The 1:b-C4-
aldehydes produced are converted further to the corresponding C4-alcohols
or after aldol condensation and hydrogenation to 2-ethyl- 1 -hexanol (2-EH).
With an annual volume of 360,000 tons of 2-EH the Oberhausen facility
became the worlds largest 2-EH operating unit.
In 1996 the RCH/RP-process was licensed to Hanwha Corp. in South
Korea. With a total capacity of 120,000 tons the unit started up successfully
in 1997. The aldehydes are dedicated mainly for the manufacturing of
butanols.
All units are running well without any major problems being observed.
The process can be operated for long periods of time at steady and stable
conditions. In Table 1 typical process conditions and the composition of the
oxo crude product are shown as fifteen years averages.
7.2.6 Economics
Three major factors attribute to the reduction in the overall

manufacturing costs of the RCH/RP two-phase hydroformylation process:
A highly efficient energy recovery system
Minimum capital requirements for an RCH/RP oxo unit.
A high feedstock usage in combination with a classical reactor that
converts not reacted off gas.
These are results of the simplified process design, which was only
possible by applying a water-soluble catalyst. The catalyst/product
separation is achieved by a simple phase separation. Therefore, the

exothermal reaction heat of the hydroformylation reaction can easily be
recovered by a heat exchanger inside the reactor (high energy efficiency)
and, secondly, no further distillation steps and columns are required to
separate the catalyst from product (less capital).
Table 1. Reaction conditions and composition of crude product of the RCH/RP process
(15 years average)
Product [Unit] Typical Value Variation

n-Butanal[%] 94.5 95 - 91
i-Butanal [%] 4.5 4 -8
Heavy Ends [%] 0.4 0.2 - 0.8
1:b selectivity 19 13.3 - 32.3
Selectivity towards C4- 99 99
aldehyde [%]
Process conditions
Temperature [C] 120 1 10 - 130
Total pressure [bar] 50 40 - 60
Vol.(water)/Vol.(organics) 6 4-9
Heat recovery*) [%] >99 >99
Conversion [%] 95 85 - 99
*) Minus losses by radiation
Both factors together with the reduced fixed costs and the usage of an
own technology (no license fees in case of Celanese plants) makes the
RCH/RP process ca. 10 % cheaper in manufacturing costs (costs for ligand
synthesis already included) compared to classical rhodium process applying
a homogeneous phosphine-modified catalyst.
7.3 Reaction of various alkenes
7.3.1 Ethylene to propanal: why not applied?
The major advantage of applying two-phase technology is the easy

separation of catalyst phase and product phase. Due to the low solubility of
butanal, pentanal, etc. in water and vice versa the separation of the reaction
and product layer causes no problem in general, phase separation occurs fast
and complete. Propanal, however, derived from the hydroformylation of
ethylene, bears a significant miscibility with water. This causes two
problems:
a) the formed propanal is "wet", the water in the aldehyde has to be
removed be distillation, which is not trivial due to the formation of
azeotropic mixtures,
200 Chapter 7
b) together with the water dissolved in the propanal a significant

amount of catalyst is transported out of the reactor, a recovery of this
portion of the catalyst is not very efficient.
Nevertheless, the hydroformylation of ethylene itself in the water-based
Rh/TPPTS system is fast, but the advantage of ethylene, viz. its higher
solubility in the water phase, is counterbalanced by the unfavorably high
solubility of water in propanal. For this reason the hydroformylation of
ethylene is performed in homogeneous Rh/TPP systems to avoid the
aforementioned problems.
7.3.2 Long-chain alkenes
Concepts for the conversion of long-chain alkenes

Long chain alkenes cannot be hydroformylated economically by the
classical Ruhrehemic/ Rhne Poulenc process, because of low solubility of
alkenes higher than butene in water [7]. Especially for higher alkenes it
would be highly desirable to convert them with a two-phase system because
of the higher boiling points of the resulting aldehydes. In a homogeneous
system these aldehydes have to be stripped from the catalyst solution at a
high temperature which leads to formation of more heavy ends and faster
catalyst deactivation. Numerous attempts have been undertaken to
hydroformylate higher alkenes with a two-phase system. In principle the
difference in polarity between the organic and the catalyst phase has to be
decreased compared to the classical RCH/RP system. The following
principal approaches have been investigated and will be discussed below:
1. increase the lipophilic character of the catalyst phase,
2. increase the solubility of the catalyst in the product phase,
3. immobilize the catalyst on a support.
Cosolvents of the classical RCH/RP system

Increasing the lipophilic character will increase the solubility of the
alkene in the catalyst phase and thus it will increase the rate of
hydroformylation. The limits are the solubility of the cosolvents in the
lipophilic phase, and leaching of ligand and rhodium into the organic phase.
This means, that choosing the right system for the alkene to be converted is
crucial.
To increase the lipophilic character, cosolvents can be added to the
classical RCH/RP system. Solvents like methanol, ethanol, polyethylene
glycol strongly increase the hydroformylation rate [8]. The potential
disadvantages of these cosolvents are leaching into the organic phase and
their reactivity towards acetals. Leaching of catalyst and ligand seems to be
no problem, provided that certain concentrations of cosolvents are not
exceeded. Another disadvantage is a drop in 1:b selectivity with increasing

amount of cosolvent. This implies that the solvent inters with the catalyst.
Other solvents than water for the two-phase hydroformylation

Because of the increasing lipophilic character of the long chain
aldehydes, solvents other than water can be applied in a two-phase system.
TPPTS as ligand is almost exclusively soluble in water, but TPPMS (mono-
sulfonated TPP) is soluble in polar organic solvents. Abatjoglou developed a
two-phase system with TPPMS as ligand and N-methylpyrrolidone (NMP)
as solvent [9]. After the reaction, water is added to obtain phase separation.
Prior to recycling of the catalyst phase, the added water is removed by
distillation (see Chapter 8).
Ionic liquids have been used for the hydroformylation of higher alkenes.
Here the lipophilic character of the catalyst phase can be adjusted by proper
choice of the ionic liquid. In the recent past good results in hydroformylation
of 1-octene have been reported by P. Wasserscheidt and coworkers.
Linear:branched ratios up to 16 with special ionic ligands have been
achieved [ 10]. Bahrmann discovered that certain amines form ionic liquids
with TPPTS. In this instance the ligand itself is the ionic liquid [11].
Detergents in the classical RCH/RP system

Anionic detergents like soaps, AOS and alkylsulfonic acids show no
increase in hydroformylation rate compared to the classical system. Cationic
detergents, on the other hand, can increase the hydroformylation rate up to a
factor of 4. This is best explained by the fact that sodium of TPPTS is
exchanged by the detergent cation, which increases the solubility of the
catalyst in the organic phase. Technical application is hampered by loss of
rhodium into the organic phase and foaming [12].
Immobilization of the catalyst

Numerous attempts have been made to immobilize rhodium or its
complexes on polymeric or on inorganic supports. Immobilization without
ligands leads to systems that mainly hydrogenate the alkene [ 13]. Recently
van Leeuwen and coworkers described the immobilization of Rh - Xantphos
complexes in silicagel [14]. The system can be recycled 8 times without
significant loss of Rh or activity, but the reaction rate is very low.
Another possibility is the linkage of phosphines to linear polystyrene [15]
or other polymers like PEG and polyacrylic acid. These polymers can be
separated by membrane filtration or in some cases by thermoregulated phase
transfer [16]. In the latter case, the catalyst is soluble at lower temperature
and insoluble at higher temperatures. So the catalyst can be separated from
the products by phase separation at higher temperatures.
202 Chapter 7
References
1 Chauvin, Y.; Mumann, L.; Olivier, H. Angew. Chem. 1995, 107, 2941.
2 Horvth, T.; Rbai, J. Science, 1994, 266, 72.
3 a)Kuntz, E. D. CHEMTECH, 1987, 17 (9), 570. b) Kuntz, E. G., Fr. Pat. 2,314,910,
1977 (to Rhone-Poulene); Chem. Abstr. 1977, 87, 101944.
4 Bhanage, B.M.; Studies in hydroformylation of olefins using transition metal cataysts,
Ph. D. Thesis, University of Pune, 1995: b) Deshpande, R. M.; Chaudhari, R. V. Ind.
Eng. Chem. Res. 1988. 27, 1996.
5 Frohning, C. D.; Kohlpaintner C. W. in Aqueous-Phase Organometallic Catalysis (Eds.:
B. Cornils, W. A. Herrmnann), VCH, Weinheim, 1998, p. 295.
6 Chaudhari, R. V.; Bhanage B. M. in Aqueous-Phase Organometallic Catalysis (Eds.:
Cornils, B.; Herrmann, W. A.; VCH, Weinheim, 1998, p. 290; b) (Cornils, B.; Wiebus,
E., EP 0,158,246, 1985 (to Ruhrchemie AG).
7 Horvth, I.; Kastrup, R. V.; Oswald, A. A.; Mozeleski, E. J. Catal. Letters 1989, 2, 85.
8 Cornils, B.; Herrmann, W. A.; Aqueous Phase Organometallic Catalysis, Wiley-VCH
Weinheim, 1998, p. 306 ff.
9 Abatjoglou, A. G.; Bryant, D. R. US Patent, 4,731,486, 1988 (to Union Carbide
Cooperation). Chem. Abstr. 1988, 109, 57042.
10 Wasserscheid, P.; Salzer, A. XXXIII. Jahrestreffen Deutscher Katalytiker, Tagungsband
p. 66ff.
11 Bahrmann, H.; Schulte, M. DE 19756945 (to Celanese). Chem. Abstr. 1999, 131, 75261.
12 Cornils, B.; Herrmann, W. A. Aqueous Phase Organometallic Catalysis, Wiley-VCH
Weinheim, 1998, p. 306 ff.
13 Greiner, R.; Burger, B. XXXIII. Jahrestreffen Deutscher Katalytiker, Tagungsband p.
119 ff.
14 Sandee, A. J.; van der Veen, L. A.; Reek J. N. H.; Kamer, P. C. J.; Lutz, M.; Spek, A.
L.; van Leeuwen, P. W. N. M. Angew. Chem. Int. Ed. Engl. 1999, 38, 3231.
15 Bayer, E.; Schurig, V. Chemtech 1976, 212.
16 Jin, Z.; Zheng, X.; Fell, B. J. Mol. Catal. A.; Chem. 1997, 116, 55.
17 Grtner, R.; Cornils, B.; Springer, H.; Lappe, P., DE 3.235.030, to Ruhrchemie AG
(1982).
18 Kulpe, J. A. Dissertation, 1989, Technische Universitt Munchen; b) Kohlpaintner, C.
W. Dissertation, 1992, Techniche Universitt Munchen.
19 Frohning, C. D.; Kohlpaintner, C. W. in Aqueous-Phase Organometallic Catalysis (Eds.:
Cormils, B.; Herrmann, W. A. VCH, Weinheim, 1998, p. 304.
20 Herrmann, W. A.; Kohlpaintner, C. W.; Bahrmann, H.; Konkol, W. J. Mol. Catal. 1992,
73, 191; b) Herrmann, W. A.; Kohlpaintner, C. W.; Manetsberger, R. B.; Bahrmann, H. ;
Kottmann, H. J. Mol. Catal. 1995, 97, 65; c) Bahrmann, H.; Bach, H.; Frohning, C. D.;
Kleiner, H. J.; Lappe, P.; Peters, D.; Regnat, D.; Herrmann, W. A. J. Mol. Catal. 1997,
116,49; d) Herrmann, W. A.; Kohlpaintner, C. W. Angew. Chem. Int. Ed. Engl. 1993,
32, 1524.
21 a) Frohning, C. D.; Kohlpaintner, C. W. in Aqueous-Phase Organometallic Catalysis
(Eds.: Cornils, B.; Herrmann, W. A.; VCH, Weinheim, 1998, p. 302; b) Bahrmann, H.;
Frohning, C. D.; Heymanns, P.; Kalbfell, H.; Lappe, P.; Peters, D.; Wiebus, E. J. Mol.
Catal. 1997, 116, 35.
Chapter 8
Process aspects of rhodium-catalyzed

hydroformylation
Peter Arnoldy
Shell International Chemicals, Shell Research and Technology Centre Amsterdam, Badhuisweg
3, 1031 CM Amsterdam, The Netherlands.
8.1 Introduction
This chapter treats industrial applications of rhodium-catalyzed

hydroformylation technology. The field of hydroformylation, based on
homogeneous catalysts in all instances, is now dominated by technologies
based on rhodium and cobalt. The development of processes with modified
rhodium catalysts was initially driven by the desire to replace of some of the
older Co-based technologies by simpler processes with higher selectivities.
Later, rhodium processes have also been developed for new applications in
which Co-based technologies play no role. This chapter is centered around
the major problem of Rh-catalyzed hydroformylation, which is slowing
down the speed of commercial implementation, i.e., the high catalyst cost
related largely to the price of rhodium. Process development is directed to
full Rh containment, thus avoiding Rh loss.
Some economic aspects, including rhodium catalyst cost, are treated in
section 8.2. Catalyst performance aspects are treated in sections 8.3 (activity,
selectivity) and 8.4 (stability, loss routes for Rh and ligand). In 8.5 and 8.6,
several commercial processes are described. Four generic, industrially used
process types are described in 8.5, viz. processes using a stripping reactor, a
liquid recycle, a two-phase reaction, and an extraction after a one-phase
reaction. In 8.6, interesting, current developments in a few petrochemical
product areas are shortly discussed.
203
204 Chapter 8
8.2 Economics
Apprehension of processes and their commercial viability starts with a

good understanding of their economics. An important concept is the
manufacturing costs sheet including all cost elements, resulting in the overall
manufacturing costs of a chemical product. For a typical (hydroformylation)
process on petrochemical scale (10-500 kt/a), all cost elements are relevant:
variable costs (alkene, syn gas, utilities, catalyst), depreciation of capital
investment, and capital-related fixed costs (labor, maintenance). Catalyst
performance has a critical impact on the overall manufacturing costs,
influencing many of the cost elements via its selectivity and activity, and
determining catalyst cost directly via its stability and the extent of Rh and
ligand losses.
Direct catalyst cost seems to be only a small cost element. Catalyst costs
of about 1-3% of total costs are typical, especially in areas with mature
products and optimized processes, with some technology competition
present. Note however, that firstly other variable cost elements (alkene and
syn gas feed) can constitute higher overall costs, but they can be influenced
to a lesser extent because they are at least consumed stoichiometrically.
Secondly, an increase of catalyst consumption to about 10% of total costs by
lack of control easily results in a totally unacceptable cost structure. For
homogeneous processes, catalyst consumption/cost is generally a major
issue, because of both limited stability/lifetime and the occurrence of
physical losses, related to imperfect separation of catalyst from products (see
8.4). Catalyst cost determines most of the development effort in the area of
Rh-catalyzed hydroformylation. Figure 1 gives the development of the Rh
price since 1972 [1,2]. In 1999, circa 16,100 kg Rh has been produced
globally, of which 94% of Rh has been consumed for automobile exhaust
catalysts [3]. Rh prices sank in the nineties because of recession and reduced
Rh demand for exhaust catalysts (due to the introduction of Pd-only catalysts
in the US). Recently, Rh prices have started to rise again, e.g. because of the
return to Rh-containing exhaust catalysts which give a better NOx reduction
at high speed.
In the overall catalyst cost, also ligand consumption can play a role,
especially if (i) ligand structure/synthesis becomes more complicated, (ii)
ligands are not chemically stable, and (iii) high ligand/Rh molar ratios have
to be applied. A typical petrochemical ligand cost is ca. $10-100/kg, but
more complicated syntheses can lead to prices around $1000/kg. Ligand
costs rapidly escalate using expensive precursors or multi-step synthesis
routes. Also the small manufacturing scale and production of large waste
volumes pushes up the costs.
8. Process aspects of rhodium-catalyzed hydroformylation 205
Going from large-scale, petrochemical processes to smaller-scale fine-

chemical processes, the structure of the manufacturing cost sheet will
change. Feedstocks will be more expensive and minimization of capex
(capital expenditure), labor, and catalyst cost is somewhat less important.
The latter, in combination with the generally smaller total financial turnover
per product and the often shorter product lifecycle, will lead to less process
development. Processes can be carried out in batch rather than continuously,
generally using generic types of separation of catalyst and product (like
distillation), in spite of some incurring Rh losses. Even a Rh throw away
process (once-through in Rh) has been proposed for small-scale production
of expensive chemicals [4] !
Figure 1. Development of the rhodium price over the years
The following calculation shows how product value determines the need
for effective Rh recycle:
Using a Rh price of $30,000/kg and a Rh Concentration in the reactors of 300 ppmw, the
Rh value in the reactors is 9 $/kg reactor content. Assuming 50%w product in the reactors, a
typical petrochemical product value of $1/kg and the need to limit Rh cost to 1% of
manufacturing costs, leads to the need to recycle Rh not less than 1800 times before it can be
wasted. Using the same assumptions but taking a fine chemical product value of %100/kg
results in need for only 18 recycles. Going buck to the petrochemical example, I % Rh cost is
equivalent to a physical Rh loss with crude product of only 0.33 ppmw. assuming that there
are no other Rh loss pathways! Or in chemical terms, at a product mol weight of 72 (butanal),
a turnover is required of 4.3 million mol product/mol Rh!
206 Chapter 8
8.3 Catalyst selectivity and activity
8.3.1 Catalyst selectivity
While catalyst stability affects directly the catalyst cost, other catalyst
performance elements (activity and selectivity) also have their impact.
Catalyst selectivity (alkene loss to aldehyde byproduct, paraffin, alcohol,
heavy ends) determines alkene variable costs, but also capital costs, via
process simplifications and increased production of desired product in the
same facilities. Chemoselectivity to aldehydes is high for all Rh catalysts.
By-products can include aldehyde isomers, low-reactive alkene isomers,
alcohols, alkanes, and heavy ends. Some aldehyde isomers (generally
branched) have a significant value (e.g., isobutanal, some branched detergent
alcohol constituents).
Of all by-products, the formation of heavy ends constitutes the biggest
problem, generally not so much from a product point of view, but rather
from a process point of view. Heavy-ends accumulation can pose serious
process problems, since it can lead to a forced Rh-containing bleed. Figure 2
gives a survey of heavy ends formation reactions, as they can take place in
reactors and other hot places like distillation bottoms.
Figure 2. Heavy ends formation

Central in heavy-ends formation is the high reactivity of aldehyde

products, in aldol condensation, Tischenko reaction [5], acetalization and
oxidation reactions. Aldol condensation is base-catalyzed, acetal formation
is acid-catalyzed, so it is important to keep the process medium more or less
neutral. Acetals formation can only take place if alcohol by-product is
produced (or if a hydroxy group is part of the feedstock/product).
Activity for double-bond isomerization is a mixed blessing. For most Rh
catalysts, isomerization is slow with respect to hydroformylation, resulting in
some loss of reactive 1-alkene to inert 2-alkene. For some Rh catalysts (few
diphosphine ones and especially diphosphite-based ones), however, the rate
of isomerization is enhanced to such an extent that internal alkenes can be
converted to terminal aldehydes (see Chapters 3 and 4).
8.3.2. Catalyst activity
Catalyst activity improvements can be used only up to a certain extent for

reduction of capex for high-pressure reactors, but are generally translated
into reduction of the Rh concentration (limiting catalyst cost) or temperature
(improving selectivity). Extreme reduction of reactor size does not always
pay out. The highly exothemic hydroformylation reaction (28-35 kcal/mol
[6]) requires sufficient cooling area. While normally sparged/bubble
columns and CSTRs are used with enough volume available for internal
cooling, a small internal volume would necessitate external cooling, which
would add new capex elements. Also the risk of a thermal runaway and the
need to avoid mass transfer limitations (especially for CO) favors moderate
reaction rates.
Typically, reaction rates of around 0.5-2 mol.l-1.h-1 are applied
commercially, and translate at 300 ppmw Rh and a density of 800 kg/m3 into
-1
a turnover frequency (TOF) of circa 215-860 mol.mol-1.h . The following
may serve as an example: assuming 300 ppmw Rh in the reactors, a
-1
productivity of 2 mol.l-1.h , a product molecular weight of 72, 50%w
product in the reactors, a plant capacity of 100 kt/a, a stream factor of 8000
h/a, and a gas hold-up of 20%, an impressive reactor volume of 208 m3 is
required. The latter is probably split over more reactors, in parallel, or -
preferentially - in series in order to benefit from reduced back-mixing. With
such volumes, also the Rh inventory cost is quite high: at 30,000 $/kg Rh, it
is $1.25 million, which is a significant working capital and risk element.
Several other chapters indicate how activity can be tuned via variation of
type and concentration of ligand, as well as by other process parameters.
Kinetic equations can be specific for each catalyst system, since they are
dependent on the rate-determining step (see, eg., Chapter 4).
208 Chapter 8
8.4 Catalyst stability; degradation routes, losses and

recovery
As has become clear from the above, catalyst cost (Rh and ligand) should
be controlled very carefully. Especially Rh containment is the dominant
theme in development of Rh-based processes. Besides direct catalyst losses,
also catalyst deactivation can take place, which leads to a forced bleed of
deactivated catalyst. Rh can be recovered from such bleeds as well as from
crude product. Below, a survey is given of potential degradation/
deactivation, loss and recovery routes [7]. The chemistry of the degradation
reactions of the ligands will be discussed in Chapter 9, while in this chapter
we will concentrate on the process aspects.
8.4.1 Rhodium loss routes
Three different loss routes can be distinguished:

Chemical Rh loss via Rh plating. Rh, as one of the noble metals, has a
strong tendency to plate out as zero-valent metal, which can be present as
aggregates of colloid particles or as a film on wall surfaces. Mechanism
of such plating most likely goes via gradual growth of Rh-metal
clusters, till sufficient size is reached for adhesion to other particles or the
wall. Rh-Rh interactions are strong, as can be seen from the presence of
Rh dimers or Rh6(CO) 16 under normal hydroformylation process
conditions [8, 9]. The unmodified Rh carbonyl catalyst is stabilized
against plating by CO only as the ligand, leading to a need for high CO
pressures (total pressures of ca. 200 bar). In the presence of phosphorus
ligands, these take over the stabilization of Rh, and pressures can be
reduced to 1-50 bar. Rh plating can be suppressed by low Rh
concentration, high ligand/Rh ratios, use of chelating ligands
(diphosphines, diphosphites), and low temperature (high activity). Plating
can take place inside, but also outside the reactors, especially under harsh
distillation conditions (significant temperature, no CO pressure at all, air
ingress if vacuum distillation would be applied oxidizing protecting P-
ligands).
Physical Rh loss, with crude product (Rh leaching). Depending on the
process, this can be due to entrainment of catalyst in a gaseous or liquid
product phase, or some solubility in the liquid product phase In view of
the small Rh concentrations, this Rh will be more difficult to recover and
only at additional cost, e.g., using efficient adsorption beds for Rh
concentration.
Physical Rh loss, via a bleed (from reactor/catalyst recycle loop). A bleed
stream is generally required because of (i) accumulation of heavy ends in
the reactor system (when sufficient separation of catalyst and heavy ends
is not possible) or (ii) because of catalyst deactivation by interaction of
Rh with poisons, like ligand degradation products (see 8.4.2) or external
poisons such as sulfur species and dienes.
8.4.2 Ligand loss routes
Obviously, ligand will be lost if coordinated to Rh, via the physical Rh

loss routes mentioned above. Also free ligand can be lost, e.g., if the
volatility is similar to that of heavy ends (trimers/tetramers), in the event
they are removed by evaporation. But ligand degradation is often the root
cause for the occurrence of catalyst/ligand losses. Below a survey is given of
degradation pathways of the industrially used phosphorus ligands. Since
phosphines and phosphites show very different degradation pathways, they
are treated separately in the following.
8.4.2.1 Phosphine degradation

Oxidation can occur with molecular oxygen and hydroperoxides
(thermally or Rh-catalyzed), or with oxygenates like water, carbon
dioxide and allyl alcohol (always Rh-catalyzed). Oxygen can be present
in syn gas feed, but will mainly enter the process via air ingress during
vacuum distillation. Peroxides can easily form by contact of alkenes with
air.
Sulfidation can occur with simple sulfur components as can be present in
syngas or propene (H2S, COS), but also as a result of breakdown and
reduction of the sulfonate groups of sulfonated aryl-ligands).
P-C bond cleavage [7, 10- 16] results in formation of Rh-coordinated aryl-
and diarylphosphide fragments. P-C splitting has several consequences,
a.o. (i) formation of inactive rhodium compounds, (ii) formation of side
products derived from the aryl group stemming from the ligand, (iii)
formation of diarylalkylphosphines causing a lower activity, and (iv)
scrambling of aryl groups at phosphorus if different types of aryl groups
are present, resulting in formation of different phosphines with
potentially undesirable properties.
Reaction of nucleophilic phosphines with electrophiles forming
quaternary phosphonium salts, either thermally (with ,-unsaturated
aldehydes present in heavy ends) or Rh-catalyzed (alkenes).
8.4.2.2 Phosphite degradation

Phosphites probably undergo oxidation and sulfidation reactions like
phosphines, albeit at much lower rate due to their lower nucleophilicity.
Decomposition of phosphites can be much faster than that of phosphines due
210 Chapter 8
to their hydrolytic properties. Below some specific phosphite degradation

chemistry is summarized.
P-O bond splitting via hydrolysis with water (at least available in trace
amounts due to heavy ends formation) can be very fast, lion-catalyzed or
acid-catalyzed, depending on pH [ 17]. Hydrolysis produces
hydroxyphosphites; further hydrolysis, with/without additional oxidation
reactions, can result in formation of phosphorus acids, like H3PO3, H3PO4
and -hydroxyalkyl phosphonic acids. Since acids catalyze hydrolysis
and can be formed as a product of hydrolysis, hydrolysis kinetics show an
autocatalytic behavior, making efficient acid removal a process
requirement. The development of cyclic mono-phosphites [ 18, 19] and
especially rigid diphosphites [20] has led to a strong suppression of
hydrolysis activity.
Reaction with aldehyde to phosphonium salt adducts containing 2 or 3
mol aldehydes/mol phosphite. Reaction with water gives -hydroxyalkyl-
phosphinate esters, acetal phosphonate, and eventually phosphinic acids
[19, 21] (see Chapter 9).
P-O-bond splitting via alcoholysis. Alcoholysis can occur via Rh-
coordinated alkoxy species (ex aldehyde/Rh-hydride, or alcohol).
Alcoholysis of diphosphites leads to a mono-phosphite, which poisons
catalyst activity [ 17, 22].
Hydrogenolysis of P-O bonds [ 19], probably Rh-catalyzed.
H-C bond splitting can take place via oxidative addition to Rh. This so-
called orthometallation has been found for aryl phosphites [23]. In the
case of diphosphites this reaction is catalyzed by Rh clusters and results
in catalyst deactivation [9].
Alkyl phosphites cannot be used as ligands, since they rearrange
producing an alkylphosphonate esters via the Arbusov reaction [7].
8.4.3 Catalyst recovery processes
In spite of all efforts to contain Rh catalyst in the reactor and a primary

catalyst recycle loop and to maintain its catalytic performance, it is generally
required to take a bleed from the system. Such a bleed can be taken
continuously (in combination with continuous make-up of fresh catalyst).
More preferably (making maximum use of active Rh in reactors), bleeding is
delayed as long as possible by taking measures to compensate for
performance loss. These involve a gradual change to more severe process
conditions (temperature, Rh concentration), or the addition of fresh ligand
[24] or scavengers of ligand degradation products (such as maleic anhydride
[25]). At one point in time, however, a catalyst change is required, thus
marking the end of a catalyst life cycle. Catalyst (Rh and ligand) should be
recovered as completely as possible, from a point of view of both economy

and sustainability. There seems to be a preference for on-site catalyst
recovery, using methods that can be process-specific. In such cases,
significant additional capex is involved. Alternatively, use can be made of
the services of a noble metal reclaimer. Standard treatments of Rh-
containing bleed streams are:
distillation, in order to concentrate Rh by removing heavy ends and
poisons (preference for short-contact time distillation);
oxidation of ligand and ligand degradation products, enabling bare Rh
extraction into an acidic aqueous layer, from which active Rh catalyst can
be re-formed via addition of fresh ligand and reduction; and
combustion of Rh-containing waste.
8.5 Process concepts
The selection of the optimum process for commercialization is very

feedstock- and product-specific. The volatility (molecular weight) and
polarity of alkene feedstocks and products play an important role.
Fortunately, Rh-catalyzed hydroformylation can be carried using a wide
variety of ligands, allowing for extensive ligand variation and optimization.
Such ligand design should be done with several objectives in mind; besides
the normal wish list of activity, selectivity, and stability (of Rh catalyst and
free ligand), it is important that the ligand fits to the process concept, which
sets requirements on the ligand volatility and polarity. Also ligand cost and
availability play an important role. Attention should be paid to efficient
ligand manufacture as well as to measurement of physical and toxicological
properties (for notification, needed in the case of introduction of new
commercial chemicals).
The current industrial processes for rhodium hydroformylation can be
divided in four categories/types. Each type will be described below, using
industrial applications as example, in the chronological order of commercial
implementation. The general theme in all processes is good Rh containment.
The major principles for separation of catalyst from products are evaporation
(process types I and 11) and extraction (process types III and IVA/B). The
schemes that we present are not complete; facilities like alkene and syn gas
feed purification with adsorption beds, catalyst feed and recovery systems,
and upgrading of crude product (generally via distillation and further
reactions like hydrogenation and aldol condensation) are neglected. More
separation principles than Types I-IV are possible (using, e.g., solid
supports, membranes, crystallization), but will not described in here (see
Chapter 10).
212 Chapter 8
8.5.1 Type I: Stripping reactor process/Rh containment in reactor
The development of processes with modified Rh catalysts started mainly

in order to achieve advantages with respect to existing Co-based technology
(simpler process; lower selectivities to paraffin/alcohol/heavy ends; less
diethylketone in ethene hydroformylation; higher butanal linearity in
propene hydroformylation) [26, 27]. In view of the considered risk of Rh
loss [28], a logical start was to start with the conversion of the lightest 1-
alkenes available (ethene and propene) thus enabling a process in which the
Rh catalyst was contained as much as possible, i.e., by keeping Rh within the
reactor. For these light feeds, the volatile aldehyde products can be stripped
with excess gas from the liquid phase in a so-called stripping or sparged
reactor [29]. Such a process was independently developed by several
companies (Celanese (commercialized in 1974), Union Carbide/Davy
Powergas/Johnson Matthey (commercialized in 1975 for ethene, in 1976 for
propene) and BASF, and was called the low-pressure oxo process (LPO)
[6, 27, 30-34]. The catalyst used is always Rh/triphenylphosphine (TPP)
[26], solvents are the heavy ends formed (trimers and tetramers, which have
been claimed to stabilize the Rh catalyst [5]) as well as the ligand excess. A
limitation of the process is the coupling of reaction and separation
conditions; stripping requires a combination of high temperature (lower
selectivity: lower linearity, more paraffin), low pressure (lower activity) and
high gas rate (capex, utility costs). From this perspective, the concept seems
to be most suited for ethene, just acceptable for propene, and unacceptable
for butene and higher alkenes. The concept is essentially limited by the
volatility of especially heavy ends (trimers, tetramers). Their removal rate
via the gas phase should at least equal their rate of formation, and this
becomes rapidly more demanding with increasing carbon number.
Insufficient heavy ends volatility leads to either (i) heavy ends accumulation
and need for a bleed of heavy ends containing active catalyst (see 8.3.1), or
(ii) application of lower pressures or higher temperatures, with unacceptable
consequences for catalyst performance.
Figure 3 shows the stripping reactor scheme for propene
hydroformylation in Union Carbides gas recycle process [27, 30-32, 34,
35]. Reaction conditions are about 100 oC, 15-20 bar, 300 ppmw Rh, a molar
TPP/Rh ratio of 100-200. Mixing in the reactor is ensured by gas sparging,
but can be enhanced by mechanical stirring. A large syn gas compressor is
required to generate the gas flows required for stripping, via syn gas
recycling (syn gas recycle/feed ratios of ca. 10). Propene conversion per pass
is low (ca. 30%); unconverted propene is recycled with the syn gas and
overall propene conversions are in the range of 85-90%, probably dependent
on propene quality.
Figure 3. Process type I: stripping reactor (gas recycle process Union Carbide)
There is no need for staging via use of more reactors in series, because of
the presence of propene recycle. Butanal linearity is ca. 92%, circa 2%
paraffin is produced, but no butanols. A syn gas bleed is required to get rid
of inert gases (methane, nitrogen, propane), and it is inevitable that some
propene is lost via this bleed. Due to syn gas recycling, H2/CO ratios above
10 are achieved. A large reactor freeboard and a de-mister pad are required
to ensure that no liquid (containing Rh!) is entrained with the gas. Crude
product is condensed at high pressure and subsequently degassed and
stabilized by removal of residual propene. Level control is maintained via
gas flow and crude product recycling. BASF has developed a similar
process, using a slightly higher temperature (110 oC) and lower TPP
concentration, resulting in a smaller gas recycle, but a somewhat lower
linearity (86%) [32].
8.5.2 Type II: Liquid recycle process/use of distillative separation
This process has been developed for propene hydroformy lation by

Mitsubishi Chemical and Union Carbide/Davy Process Technology [32, 33,
36], as improvement of the stripping reactor concept. Product is now
removed from the reactors via the liquid phase and carefully evaporated
from the catalyst. The uncoupling of reactor and product/catalyst separation
214 Chapter 8
conditions leads to more optimum reactor conditions: the absence of a need

for huge syn gas recycling and better tuning of heavy ends removal in the
separate evaporation step. On the other hand, in the absence of recycling,
there is also the need for high propene conversion per pass, which is
probably achieved by using more reactors in series.
Figure 4. Process type 11: liquid recycle, distillative separation (Union Carbide process)
Figure 4 gives a typical process scheme for the Union Carbide process.
Reactor conditions are similar to those in the stripping reactor case,
temperature (90 oC) and Rh concentration (250 ppmw) are slightly lower.
The reactor is stirred for good gas/liquid mass transfer. Liquid product is
depressurized and transferred hot to an evaporation section, where Rh
catalyst, dissolved in mainly heavy ends, is separated and recycled. Probably
short-contact time distillation equipment (wiped- or falling-film evaporators)
is applied to protect the Rh catalyst. There is in principle a choice in
distillation conditions: atmospheric distillation leads to relatively high
temperature (thermal strain) with potential for Rh plating and aldehyde
oligomerization to heavy ends. Vacuum distillation would lead to lower
temperature, but oxygen ingress results in some oxidation of phosphorus
ligands and aldehyde; therefore vacuum distillation seems to be not
preferred. For lower aldehydes like butanal, atmospheric distillation is
applied. With increasing molecular weight of products (and heavy ends),
vacuum distillation becomes a requirement. Process for higher aldehydes are
less attractive, because of such harsh vacuum distillation conditions (with
limited stability of aldehydes and catalyst), but also due to the unavoidable
accumulation of heavy ends and the related need for a (catalyst-containing)
heavy ends bleed plus Rh recovery section.
8.5.3 Type III: Two-phase reaction/extraction process
This process has been developed as alternative to the previous mentioned

propene hydro formy lation processes, by Rhone Poulene and Ruhrchemie
and has been commercialized by Ruhrchemie in 1984 [16, 32, 37-39], and is
described in more detail in Chapter 7. The basic principle is the use of two
liquid phases in both reactor and separator, one phase being the crude
product and the other phase containing Rh catalyst and excess ligand, thus
allowing efficient catalyst/product separation. The second phase is generally
aqueous (polar), and water-soluble ligands have been designed. The most
successful ligand (used in the Ruhrchemie/ Rhone Poulene process) is
triphenylphosphine tri-meta-sulfonate (TPPTS). TPPTS synthesis via TPP
sulfonation [16] is not straightforward, but has been optimized using
purification via extraction with amines [40, 4 1] and minimizing phosphine
oxidation using water-free H2SO4/H3BO3 [42]. In the two-phase
hydroformylation reactor, apolar propene diffuses into the water phase and
apolar aldehydes and by-products leave the water phase again; so in fact
extraction takes place already in the reactor. Note that, while heavy ends
accumulation constitutes a major problem in distillative Type I/II processes,
this process solves this problem elegantly via solubility of the apolar heavy
ends in the product phase. In principle, this extraction system is an open
system, i.e., would allow Rh loss by simple solubility in the organic phase.
But practice shows that, in the propene hydroformylation case, solubility of
the Rh catalyst in crude product is extremely low (ca. 1 ppbw).
Figure 5. Process type III: two-phase reaction/separation

(Ruhrehemie/Rhone Poulene process)
216 Chapter 8
Figure 5 gives a description of this process. The reactor is mechanically

stirred, to minimize mass transfer limitations for both gad/liquid- and
liquid/liquid phase transfer. In spite of this, the reaction still seems to be
mass-transfer limited and restricted to the Iiquid/liquid interface [43], which
may be caused by high catalyst activity and temperature combined with the
low solubility of propene. The reactor cooling is integrated with the butanals
distillation reboiler. The temperature is ca. 120 oC, the pressure is ca. 50 bar
(H2/CO ratio of 1.0). The water phase contains ca. 300 ppmw Rh at a
TPPTS/Rh molar ratio of 50-100. The water/organic phase ratio in the
reactors is high (ca. 6) making water the continuous phase and giving an
overall high Rh concentration in the reactors.
The two-phase liquid is taken from the reactor (without cooling) to a
decanter vessel, in which excess syn gas is separated and the two liquids
separate by simple settling. The catalyst-containing aqueous phase is
recycled back to the reactor and is cooled against condensate to absorb some
of the reaction heat and produce low-pressure steam. The product phase
(containing all heavy ends) is stripped at high pressure with syn gas feed to
recover unconverted propene. The relatively high temperature of 120 oC
seems to be required because of the limited solubility of propene in water,
the desire for energy integration, and the need for high propene conversion
per pass (ca. 95% in one reactor). This higher temperature might have an
impact on catalyst stability, although a high ligand/Rh ratio is applied; no
data are available on Rh loss via plating or catalyst deactivation. High CO
pressure seems required, maybe for catalyst stability reasons, maybe because
of low CO solubility in water [44]. Heavy ends are minimized by pH control
(pH 5.5-6.2); the presence of some CO2 (1-3%) in the syn gas seems to be
beneficial to minimize heavy ends [45].
The major advantages of this process are: (i) the high level of heat
integration (making this process a steam exporter rather than a steam
importer), (ii) simple separation of catalyst and product/heavy ends (without
thermal stress of catalyst via distillation), (iii) better selectivities (99%
aldehydes, 95% linearity; no paraffin) and (iv) lower sensitivity to some
poisons (the preference of some poisons for the organic product layer). The
higher product linearity seems to be hardly an advantage in view of the value
and market size for isobutanal derivatives. The higher reactor pressure and
the heat integration might increase the capex somewhat.
8.5.4 Type IV Extraction after one-phase reaction
In this concept, catalysis is carried out in a homogeneous, one-phase

reaction system, while a two-phase catalyst extraction is done afterwards. A
one-phase reaction seems to be preferred, when the type III two-phase
reaction leads to (i) too low reaction rates because of low substrate solubility
or mass transfer limitations, or (ii) inefficient phase separation, because of
emulsification. The two separate phases form after reaction just by cooling
or, more typically, by addition of a solvent. In order to have efficient
extraction after the reaction, the distribution coefficients over two liquid
phases of catalyst and products should be sufficiently different. Generally
one phase is aqueous, and the other one is apolar. Given the product
(a)polarity, ligand design will result in the opposing ligand (a)polarity. Two
type IV concepts will be described: use of apolar catalyst for polar product
(type IVA), and use of polar catalyst for apolar product (type IVB).
8.5.4.1 Type IVA: apolar Rh catalyst, polar product

The Type IV concept has been commercialized for the first time in the
Rh-catalyzed hydroformylation of polar feed to polar product: allyl alcohol
(AA) conversion to 4-hydroxybutanal (for production of 1,4-butanediol
(BDO)) [46]. This process was developed by Kuraray and Daicel Chemical
Industries [47, 48] and commercialized in 1990 by Arco (Lyondell since
1998) [49], and is described in Figure 6.
Figure 6. Process type IV: one-phase reaction, extractive separation; type IVA: apolar
catalyst, polar product (Kurarays 1 ,4-butanediol process)
The catalyst used is Rh/TPP (apolar), to which 1,4-

bis(diphenylphosphino)butane (dppb) is added in small amounts (molar ratio
Rh/TPP/DPPB ca. 1 : 150:0.2). The presence of dppb elegantly avoids rapid
deactivation of the Rh catalyst by (i) poisoning by acyl intermediates or
methacrolein (formed by dehydration of branched aldehyde by-product), and
218 Chapter 8
(ii) ligand oxidation [50-53]. Very mild process conditions (ca. 60-65 C, 2-
2.5 bar) can be used, because of the high reactivity of AA compared with
unfunctionalized alkenes. These mild conditions will help to suppress heavy
ends (acetals) formation. Reaction takes place in one phase using an apolar
catalyst in an aromatic solvent like toluene. Multi-stage extraction with
water (ca. 30 C, water/organic phase volume ratio of ca. 1:1) leads to
recovery of the hydrophilic product in the aqueous phase, while essentially
all catalyst is left in the organic phase and can be recycled. Rh losses via the
aqueous phase are negligible (10 ppbw). Note that heavy ends as well as
TPPO (formed by oxidative addition of AAs C-O bond to Rh [52]) are less
polar and will accumulate in the apolar catalyst recycle, leading to need for
some bleed from the catalyst recycle. The reaction is very sensitive to CO
pressure. A too low a CO pressure results in low selectivities (production of
more propanal and propanol via AA isomerization and hydrogenation,
respectively) and catalyst deactivation. A too high a CO pressure leads to
lower linearity (< 70% rather than ca. 87%) as well as Rh catalyst loss with
the aqueous product phase. In order to achieve the optimum CO pressure,
good control of reaction rate and avoidance of mass transfer limitation (good
mixing) are important. The H2/CO ratio of syngas feed is ca. 4; CO
starvation is probably prevented by high syngas recycling rates, but addition
of CO in a second stage is an alternative option. AA concentrations should
be kept low: high AA conversions are preferred and back-mixing in the
reactors will help. Typical AA conversions are ca. 98%, propanal and
propanal selectivities are ca. 7 and 3 mol%, respectively. Recently Arco has
been able to achieve a linearity improvement (up to ca. 95%) by use of a
group 8 metal (Fe or Ru) as an additive [54]. There still seems to be scope in
Rh-catalyzed hydroformylation for BDO production: Lyondell has
announced to build 126 kt/a new capacity in the Netherlands in 2001, in
addition to their current 55 kt/a capacity in the US, which would be the
largest single stream BDO unit in the world [55].
8.5.4.2 Type IVB: polar Rh catalyst, apolar product

The Type IV concept can also be used in the field of higher alkenes, but
in this instance the product phase is the apolar one and a polar catalyst has to
be applied. In this field, technology developed by Union Carbide seems most
advanced (see below). Recently, Sasol announced the commercialization in
200 1 of a Rh-catalyzed hydroformylation (low-pressure oxo) technology,
licensed from Kvaerner (previously Davy McKee), for conversion of Sasols
Fischer-Tropsch 1-alkenes to detergent alcohols, on 120 kt/a scale [56]. On
the basis of the advanced status of Union Carbides higher alkene
technology, it seems probable that Sasol will apply the latter in South Africa.
Sasols Fischer-Tropsch streams consist of alkenes diluted in a soup of
paraffins, aromatics and oxygenates (e.g., alcohols) [57], and

hydroformylation of the full stream is feasible since product alcohols have a
higher boiling point and can be readily separated from the other components
by distillation. The alkenes are > 90% 1-alkenes, so are well suited for
hydroformylation by Rh/phosphine catalysts; a significant part of the 1-
alkenes is mono-methyl-branched, but these branches are well distributed
over the alkene chain [57], so probably, they hardly affect the
hydroformylation rate.
For use in their higher alkene process, Union Carbide use mono-
sulfonated ligands, most probably -diphenylalkane-a-sulfonates (alkane =
butane (DPBS) or propane (DPPS; DPBS seems preferred) [58]. These
ligands cannot give undesired aryl exchange reactions, and are probably not
expensive, since they can be conveniently produced from diphenylphosphide
and 1,4-butane- and 1 ,3-propane-sultone (carcinogenic), respectively [59].
Several production routes for the sultone precursors have been described
[60]. The presence of an alkyl group on the phosphorus leads to lower
activity than found for the Rh/TPP system, resulting in need for higher
temperature (ca. 110 C) and Rh concentration (ca. 300-400 ppmw), but a
lower ligand/Rh molar ratio (ca. 15). Pressure is low (ca. 7 bar) and H2/CO
ratios are high (ca. 4). Figure 7 gives a process scheme [33, 58, 61]. A series
of well-mixed reactors is used to obtain high alkene conversion (ca. 95%).
Selectivity to aldehydes is ca. 95%, with 2-alkenes being the major by-
product.
Figure 7. Process type IV: one-phase reaction, extractive separation; type IVB: polar catalyst,
apolar product (Union Carbide's higher alkene hydroformylation process)
220 Chapter 8
Aldehyde linearity is high (ca. 90%). Sufficient N-methyl-pyrrolidone

(NMP; ca. 40%w) and some water (1 -2%w) are applied to achieve a one-
phase system in the reactors. After reaction, water is added in a mixer (phase
ratio 1:1 v/v), followed by efficient phase separation in a settler, with
virtually all catalyst in the NMP/water layer. The crude product layer is
subjected to a multi-stage water extraction to remove residual NMP and
catalyst, and a final treatment over a silica-bed to reduce Rh leach levels
from 0.2 ppmw to 0.02 ppmw. The recycle catalyst layer (in NMP/water) is
dried in two steps, to evaporate water and achieve the low water
concentrations required for one-phase reaction, and then recycled to the
reactors. Water is recycled, from evaporators, via water extraction, to the
mixer. The flexibility of this process with respect to alkene carbon number
seems excellent: good performance has been found for C6-C14 alkenes [61].
8.6 Survey of commercialized processes and new

developments
Table 1 gives a survey of industrially applied Rh-catalyzed

hydroformylation processes, with indications of type of catalyst and process.
The list is probably not complete; especially smaller-scale applications might
have been missed, because of lack of open literature data. Below, some
interesting, new developments in a few product areas are shortly discussed,
when they have special process/catalyst features and/or involve new
petrochemical applications.
8.6.1 Hydroformylation of butenes
Conversion of linear butenes is mainly of interest for production of the

plasticizer alcohol 2-propyl- 1 -heptanol via pentanal. Linear butenes are
generally available as a Raffinate- 1 stream (after butadiene extraction of
cracker C4s) or a Raffinate-2 stream (after additional isobutene removal,
generally via MTBE production). The latter contains 50-65% 1 -butene, the
remainder being 2-butenes plus some butanes [24]. The Ruhrchemie-Rhone
Poulene process has been developed for butene conversion by Hoechst (now
Celanese) and is very similar to the process for propene (see Chapter 7 and
8.5.3) [24]. The same catalyst, Rh/TPPTS, is used. To compensate for lower
alkene solubility, 5-10 C higher temperature and slightly higher Rh
concentrations are applied. Occurrence of double-bond isomerization to 2-
butene results in a lower selectivity to aldehydes. The fact that only 1 -butene
is hydroformylated, leaving all 2-butene, makes this process less elegant
than the propene process.
Table 1. Survey of commercial applications of Rh-catalyzed hydroformylation processes
alkene products D# year process ligand kt/a Reference
a
C6-C14 higheralcohols M 1970 II none 23 Ch. 8.6.3
1 -alkenes
g
ethene propanol C,U 1974 I TPP 400 Ch. 8.5.1
propene butanol, B,C,U 1974 I TPP 4000 Ch. 8.5.1

b
isobutanol,
2-ethyl- I-hexanol,
neopentyl glycol
b
do do M,U 1978 II TPP Ch. 8.5.2
do do R 1984 III TPPTS 600 Ch. 7, 8.5.3

c d
1,2 -diacetoxy - vit-A B 70's II none 3 [32,34]
3-butene
d
1 ,4 -diacetoxy - do H 70's II TPP [32,34]
2-butene
1 -hexene, carboxylic C 1980 II TPP 18 Ch. 8.6.3
1-octene acids
branched isononanol M 1987 II TPPO 30 Ch. 8.6.2
internal octenes
f
3-methyl-3- 3-methyl-1,5- K 1988 II L* 3 [32,49,94,95]
butene- 1-ol pentanediol'
Allyl alcohol 1,4 b- utanediol K 1990 IVA TPP+dppb 180 Ch. 8.5.4.1
1-octenal 1,9 -nonanedioI K 1993 IVB TPPMS 2-3 [32,49,96,97]

f
do do K > 1993 II L* do [32,98,99]
f
1 -butene 2-propyl-1- Ho 1995 III TPPTS 40 Ch. 7, d 8.6.1
h
heptanol
1-butene/2- 2-propyl-1- U 1996 II diphosphite 80 Ch. 8.6.1,
butenes heptanol [100]
higher detergent Kv 2001 IVB DPBS 120 Ch. 8.5.4.2
1-alkenes alcohols (U?)
#
Developed by: M=Mitsubishi, C=Celanese, U=Union Carbide, B=BASF,
R=Ruhrchemie-RhonePoulene, H=Hoffman-LaRoche, K=Kuraray, Kv=Kvaemer,
a b c
Ho=Hoechst. Probably obsolete. Sum for Type I and Type II. 4-acetoxy-2-methyl-2-
d e
butenal; vitamin-A precursor. 3 kt/a sum for B and H. Together with -methyl--
f g h
valerolactone. L*=bulky monophosphite. Plus propanoic acid, methyl methacrylate. Plus
pentanoate esters.
222 Chapter 8
Union Carbide has focused on the development of diphosphite ligands

(see also Chapter 3), for conversion of butenes including 2-butene to linear
pentanal [20, 62, 63]. Compared to cyclic mono-phosphites [18, 19], they
have the advantages of substantial double-bond isomerization activity,
higher ligand stability and higher regioselectivity to linear aldehydes, albeit
at somewhat slower (but still significant) hydroformylation rate [20,64]. The
ligand applied commercially could well be 6,6-[3,3-bis(tert.butyl)-5,5-
dimethoxy- 1,1-iphenyl-2,2-diyl]-bis(oxy)bis-
dibenzo[d,f][ 1,3,2]dioxaphosphephin, or a related structure with additional
tert.butyl groups instead of the methoxy groups [20, 34, 64, 65]. The activity
for 2-butene is lower than for 1-butene, resulting (at ca. 250 ppmw Rh) in
temperatures of 85-90 C rather than the 70-75 C required for pure 1-
butene. Pentanal linearity is so high (ca. 97%) that no aldehyde purification
is required before aldol condensation. Related to the mild conditions, heavy
ends make is likely to be low and therefore their accumulation can probably
be avoided in this type II liquid recycle process. Catalyst/(by)product
separation occurs via atmospheric distillation (ca. 110 C). Reactor pressures
are low (circa 10 bar). The H2/CO ratio in the feed is probably slightly below
1, in order to have a high CO pressure enabling lower paraffin make and
suppression of inhibition by poisoning mono-phosphite. This high CO
pressure, however, leads to CO inhibition and a negative order in CO
pressure, resulting in potentially unstable reactor operation (cycling,
runaway). This problem is tackled by (i) reactor cooling with high cooling
flow and small temperature differential, and (ii) running at high conversion,
especially of H2 [66, 67]. Ligand/Rh molar ratios are 1.1 to 4, some ligand
excess being required for protection against Rh plating, the more so because
some ligand degradation takes place (formation of poisoning mono-
phosphites and strong phosphorus acids which can catalyze hydrolytic ligand
degradation). Some water is present for destruction of the poisoning mono-
phosphite, by hydrolysis to phosphorus acids (the latter hydrolysis being
much faster than that of the diphosphites themselves) [22]. The phosphorus
acids (like a-hydroxypentylphosphonic acid) have to be efficiently removed,
by addition of epoxides [17, 68], but preferentially by water extraction [69,
70]. A phosphate buffer can be applied to increase the acid uptake capacity
of the water phase and stabilize the pH at 4.5-7.5 [71]. Heterocyclic nitrogen
components (like benzoimidazole or benzotriazole) stabilize Rh clusters
formed during distillation [72], but also assist in trapping acids [66,69,70]. In
water extraction, the heterocyclic nitrogen components stay in the organic
phase. A group VIII metal (Ru preferred, 1-5 mol/mol Rh) can be added to
suppress Rh cluster formation and its catalysis of orthometallation [9].
8.6.2 Branched higher alkenes to mainly plasticizer alcohols
Plasticizers (mainly for PVC) are dialkylphtalates formed by reaction of

phtalic anhydride with plasticizer alcohols (C7-C11). Market size (1995) for
plasticizer alcohols is ca. 3900 kt/a including 2400 kt/a 2-ethyl- 1-hexanol
production ex n-butanal [73]. In the plasticizer alcohols market, branched
alcohols dominate. Most of the plasticizer alcohols are produced from
relatively cheap propene/butenes feedstocks. Besides the direct
hydroformylation of propene and butenes (giving 2-ethyl- 1 -hexanol and 2-
propyl-1-heptanol), propene and butenes are also oligomerized to C9 and C8
branched alkenes, respectively, which are then converted producing C9-C10
alcohols (isononanol, isodecanol), generally using Co-based
hydroformylation technology (with/without phosphine modification). Rh-
based technology has difficulty to enter this market, because (i) most Rh
catalysts are not suited for conversion of branched alkenes, (ii) the products
are cheap, (iii) no high selectivity to linear alcohols is required, and (iv)
heavy ends are more difficult to remove.
Mitsubishi Chemical has commercialized a Rh-based process on 30 kt/a
scale in 1987, for conversion of branched octenes into isononanol (INA)
using distillative separation of catalyst (type 11) [36, 74, 75]. The catalyst is
Rh/triphenylphosphine-oxide (TPPO; Rh/TPPO molar ratio ca. 20). TPPO is
claimed to be not just inert, but to be an activity promoter. Hydroformylation
of branched octenes takes place at ca. 130 C and 200 bar, with low Rh
concentrations (10-100 ppmw). Aldehyde yields are high (95-98%; the rest
being mainly useful alcohol). TPP (1-10 mol/mol Rh) is added after reaction
and before distillation, in order to prevent Rh plating during vacuum
distillation (at ca. 130 C bottoms temperature). Part of the TPP is oxidized
during distillation via air ingress, part is oxidized on-purpose after
distillation with air or hydroperoxide (formed by air oxidation of the alkene
feed). Heavy ends will be formed in significant amounts in this process due
to high temperatures in reaction and distillation; their accumulation results in
a forced catalyst-containing bleed, from which Rh has to be recovered in
separate facilities. Although the INA yield is high, this process seems to be
not very attractive from a Rh containment point of view (catalyst cost,
capital to recover catalyst).
8.6.3 Linear higher alkenes to mainly detergent alcohols
In the area of hydroformylation of linear higher alkenes, applications as

detergent alcohol dominate (mainly C12-C16) [76]. The market size for
detergent alcohols was ca. 1200 kt/a in 1995 [73]. Linearity (generally in the
range 40- 100%) relates to the need for detergency (micelles) and
224 Chapter 8
biodegradability. Figure 8 gives a survey of routes to detergent alcohols and

detergents [76, 77]. Most detergent alcohols are produced from ethene and
higher alkenes, but part is produced from natural oils and fats, by
transesterification with methanol followed by ester hydrogenation. Higher
alkenes can be produced in various ways. 1-Alkenes are produced via ethene
oligomerization, or from syn gas via the Fischer-Tropsch process, but
presently are produced in insufficient amounts to satisfy the need for
surfactant alcohols. Internal alkenes can be produced by an isomerization-
metathesis sequence (Shell process) or by dehydrogenation of alkanes (e.g.,
UOPs Pacol-Olex process). Commercially virtually all hydroformylation of
higher alkenes takes place using Co catalysts (Co-oxo and Shell process
using phosphine-modified Co catalyst), which convert both terminal and
internal alkenes: they are able to rapidly isomerize the double bond, leading
to formation of aldehydes/alcohols, with linearities up to 80%.
Figure 8. Routes to linear higher alkenes, detergent alcohols and detergents
For Rh-catalyzed hydroformylation of higher alkenes, preferably

phosphine-based catalysts are used for the conversion of 1 -alkenes, enabling
linearities of 80-90%. The low-volatile nature of the higher aldehydes and
their heavy ends clearly disadvantage any process with distillative
catalyst/product separation (type 11). Mitsubishi Chemical has reported the
start-up of such a unit in 1970 (23 kt/a), using the unmodified Rh catalyst
and C6-C14 1-alkenes [36], but that unit is probably obsolete. A better type II
process is run by Celanese since 1980; they convert C6-C8 1-alkenes to hear
aldehydes on 18 kt/a scale (for production of carboxylic acids), using
Rh/TPP and distillative separation under vacuum [78, 79]. Probably, C8 is
the maximum alkene carbon number to be used in a type II process.
The use of a process of the extractive type (type III or IVB) seems most
logical for these higher alkenes, since this enables efficient separation under
mild conditions of aldehydes and heavy ends from polar catalyst. For type III
processes, the major problem is the low solubility of higher alkenes in water
(the presence of salts in the aqueous layer making it even worse) [73,80],
thus limiting the reaction essentially to the interface. Combinations of the
following basic approaches can be used: (i) increase the alkene solubility in
the polar phase, (ii) increase the interfacial area, (iii) use a phase-transfer
catalyst, and (iv) enhance the reaction rate at the interface. To achieve these
objectives, either additives are used (generally selecting the Rh/TPPTS
catalyst) or the phosphine structure is modified. Most approaches, however,
seem to fail in the two-phase system. Once the problems around catalyst
activity are solved, significant problems are still present around loss of
catalyst and additives. Type IVB processes seem to be most convenient. In
8.5.4.2, the process developed by Union Carbide has been described. Many
groups are exploring similar approaches [73,81].
8.6.4 1,4-Bu tanediol
1,4-Butanediol (BDO) is a key intermediate in the petrochemical industry

(ca. 600 kt/a) [54, 82]. It is used mainly as monomer for the production of
the polyester polybutyleneterephtalate (PBT). But significant amounts are
also converted into tetrahydrofuran (THF) via dehydration (ca. 250 kt/a)
which finds applications as monomer for polytetramethyleneglycol
(PTMEG) and as solvent. BDO can also be dehydrogenated to -
butyrolactone (GBL) (ca. 100 kt/a), the main outlet of which is the
production of N-methyl-2-pyrrolidone (NMP) and N-vinylpyrrolidone
(monomer for PVP). THF and GBL can also be produced independently
from BDO, by hydrogenation of maleic anhydride (MALA). Figure 9 gives a
survey of routes to BDO and THF. The major commercial route to BDO is
the Reppe process, converting acetylene with formaldehyde to 2-butyne-1,4-
diol, followed by hydrogenation. Smaller amounts are produced via
acetoxylation of butadiene to 1,4-diacetoxy-2-butene and conversion of 1,4-
dichloro-2-butene. A last commercial route is Rh-catalyzed
hydroformylation of allyl alcohol (see 8.5.4. 1), obtained via isomerization of
propene oxide. The old acetylene route is still going strong, due to the fact
that a lot of acetylene extraction capacity is just available. But there is also a
lot of promise in routes using cheaper feedstocks like propene or butane.
Butane can be converted, via selective oxidation to MALA as intermediate,
into BDO, THF and GBL, or combinations thereof. New propene-based Rh-
hydroformylation routes to BDO are being explored, using allyl acetate or
acrolein. In view of the high cost of propene oxide, an alternative for allyl
226 Chapter 8
Figure 9. Routes to 1,4-butanediol, tetrahydrofuran and g-butyrolactone
alcohol production could be the hydrolysis of allyl acetate. Direct

hydroformylation of allyl acetate via Rh-catalyzed hydroformylation suffers
from poisoning by acetic acid. It might therefore be preferable to first
hydrolyze allyl acetate to crude allyl alcohol, and hydroformylate that crude
mixture. This still requires control of acetic acid traces, via extraction with
slightly alkaline water [83]. Alternatively, acrolein can be used. This cannot
be hydroformylated directly, but is converted to a cyclic acetal (using well-
available 1,3-diols like 2-methyl-propane- 1,3-diol (by-product) or 2,2,4-
trimethyl-pentane-2,4-diol). This acetal now contains an isolated terminal
double bond and can be hydroformylated. Subsequently, BDO is formed by
combined hydrolysis/hydrogenation. The best results can be obtained with
diphosphines with C4 bridge: at only 8 ppm Rh, full conversion can be
obtained with selectivity to linear aldehyde of > 99% [84].
8.6.5 Nylon monomers
Significant work is being carried out to develop new routes to nylon

starting from butadiene, including (generally Rh-catalyzed) hydroformy-
lation steps. Nylon-6 and nylon-6,6 are formed from the monomers -
caprolactam and adipic acid/hexamethyleen- l ,6-diamine, respectively [85].
Figure 10 gives a survey of routes to these monomers. -Caprolactam is
predominantly produced from cyclohexanone oxim, via multi-step routes
starting with benzene or toluene. Adipic acid is mainly produced by
cyclohexanol/cyclohexanone oxidation. Hexamethylene- 1,6-diamine is
produced mainly from adiponitrile. Adiponitrile, on its turn, is produced
from butadiene mainly via hydrocyanide addition or by electrochemical
coupling of acrylonitrile. There is potential for new routes to especially E-
caprolactam and adipic acid, using cheap butadiene.
Figure 10. Routes to nylon monomers
One route uses partial hydrogenation of adiponitrile to 6-

aminocapronitrile, followed by hydrolysis and ring-closure to -caprolactam
(BASF). Another uses butadiene double carbonylation to adipic acid
(Rhodia). And a third category consists of exploratory routes in which
hydroformylation (generally Rh-catalyzed) plays a role: converting (i)
butadiene directly to adipaldehyde, (ii) 3-pentenenitrile to 5-
formylvaleronitrile, or (iii) 3-pentenoate esters to 5-formylvalerates. The
direct hydroformylation of butadiene is probably the most difficult case,
with, a.o., low reactivity and selectivity [86, 87]. Rh-catalyzed
hydroformylation of 3-pentenenitrile and 3-pentenoate esters is explored
using generally diphosphite ligands since these enable the double-bond
isomerization to the terminal position. Hydroformylation of methyl 3-
pentenoates so far seems most promising. Typical numbers are an aldehyde
selectivity of ca. 90% (losses to paraffin and 2-alkene isomer) and an
aldehyde linearity of ca. 90%, at conversions of ca. 90%, but reactions are
generally very slow (10-20 h) [88-90].
Two process variants have been described. A type II process [91]
resembles the Union Carbide process for butenes hydroformylation (see
8.6.1). Rapid oxidation of the diphosphite ligands by air ingress in vacuum
distillation columns is mitigated by addition of an excess of sacrificial
ligand (tri-orthotolylphosphine). In a type IVA process, very much like the
Kuraray process (see 8.6.4.1), apolar and polar solvents are applied to effect
the desired combination of one-phase reaction followed by phase separation,
with most Rh/diphosphite catalyst in the apolar layer [92,93].
228 Chapter 8
References
1 London Fix, historical rhodium prices, http://www.kitco.com/egi-bin/yearly-graphs,egi,

2 New York spot prices, http://www.specialtymetals.com/prices.html.
3 Hillard, H. E., http://minerals.usgs.gov/minerals/pubs/commodity/platinum.
4 Solodar, A. J.; Sall, E. D.; Stout, L. E., Jr.,. in Catalysis of Organic Reactions (Malz,
R. E., Ed.), Chem. Ind. 68, Dekker, New York, 1996, 68, 119.
5 Pruett, R. L.; Smith, J. A. U.S. Pat. 4,148,830, 1979 (to Union Carbide); Chem. Abstr.
1971, 75, 109844.
6 Smith, F. J. Plat. Met. Rev. 1975, 19, 93.
7 van Leeuwen, P. W. N. M., Appl. Catal. 2000,
8 Moser, W. R.; Papile, C. J.; Brannon, D. J.; Duwell, R. J.; Weininger, S. J., J. Mol.
Catal. 1987, 41, 27 1.
9 Abatjoglou, A. G.; Bryant, D. R.; Maher, J. M., U.S. Pat. 5,756,855, 1998; U.S. Pat.
5,929,289, 1999 (to Union Carbide); Eur. Pat. Appl. 697391, 1996 (to Union Carbide);
Chem. Abstr. 1996, 124, 316213.
10 Billig, E.; Jamerson, J. D.; Pruett, R. L., J. Organornet. Chem. 1980, 192, C49.
11 Jamerson, J. D.; Pruett, R. L.; Billig, E.; Fiato, F. A., J. Organornet. Chem. 1980, 193,
C43.
12 Abatjoglou, A. G.; Billig, E.; Bryant, D. R., Organometallics 1984, 3, 923.
13 Abatjoglou, A. G.; Bryant, D. R., Organometallics 1984, 3, 932.
14 Moser, W. R.; Papile, C. J.; Weininger, S. J., J. Mol. Catal. 1987, 41,293.
15 Abatjoglou, A. G.; Billig, E.; Bryant, D. R.; Nelson, J. R., ACS Symp. Series 1992, 486,
Phosphorous Chemistry, pp. 229-239.
16 Herrmann, W. A.; Kohlpainter, C. W., Angew. Chem. 1993, 105, 1588.
17 Babin, J. E.; Maher, J. M.; Billig, E., U.S. Pat. 5,364,950, 1994 (to Union Carbide);
Chem. Abstr. 1994, 121, 208026.
18 Dennis, A.; Harrison, G. E.; Wyber, J.P., U.S. Pat. 4,482,749, 1984 (to Davy McKee);
Chem. Abstr. 1984, 100, 174267; U.S. Pat. 4,496,768, 1985 (to Davy McKee); Chem.
Abstr. 1984, 100, 191375.
19 Billig, E.; Abatjoglou, A. G.; Bryant, D. R.; Murray, R. E.; Maher, J.M., U.S. Pat.
4,599,206, 1986; U.S. Pat. 4,717,775, 1988 (to Union Carbide); Chem. Abstr. 1988,
109, 233177.
20 Billig, E.; Abatjoglou, A. G.; Bryant, D. R., U.S. Pat. 4,668,651 and 4,769,498, 1988 (to
Union Carbide); Chem. Abstr. 1989, 111, 117287.
21 Ramirez, F.; Patwardhan, A. V.; Heller, S. R., J. Am. Chem. Soc. 1964, 86, 514.
22 Maher, J. M.; Babin, J. E.; Billig, E.; Bryant, D. R.; Leung, T. W, U.S. Pat. 5,288,918,
1994 (to Union Carbide); Chem. Abstr. 1994, 121, 56996.
23 Trzeciak, A. M.; Ziolkowski, J. J., Z. anorg. allg. Chem. 1989, 577, 255.
24 Frohning, C. D.; Kohlpainter, C. W., in Aqueous-Phase Organometallic Catalysis
(Cornils, B; Herrmann, W. A., Eds.), Wiley-VCH, Weinheim 1998, pp. 294-306.
25 Konkol, W.; Bahrmann, H.; Herrmann, W. A.; Kohlpainter, C. W., Ger. Pat. 4135050,
1993 (to Hoechst); Chem. Abstr. 1994, 120, 79891.
26 Pruett, R. L.; Smith, J. A., U.S. Pat. 3,527,809, 1970 (to Union Carbide); Chem. Abstr.
1976, 84, 30433.
27 Brewester, E. A. V.; Pruett, R. L., U.S. Pat. 4,247,486 1981 (to Union Carbide); Chem.
Abstr. 1981, 95, 24271.
28 Cornils, B.; Payer, R.; Traenekner, Hydrocurbon Proc. 1975, 54, June 1975, 83.
29 Hershman, A.; Robinson, K. K.; Craddock, J. H.; Roth, J. F., Ind. Eng. Chem., Prod.
Res. Dev. 1969 ,8, 372.
30 Brewester, E. A. V., Chem. Eng. 1976, 83 (24), 90.

31 Fowler, R.; Connor. H.; Baehl, R. A., Hydrocarbon Proc. 1976, 55, Sept. 1976, 247;
Chemtech 1976, 772.
32 Frohning, C. D.; Kohlpainter, C. W., in Applied Homogeneous Catalysis with
Organometallic Compounds (Cornils, B; Herrmann, W. A., Eds.), VCH Weinheim
1996, pp. 29-104.
33 Billig, E.; Bryant, D. R., in Kirk-Othmer Encyclopedia of Chemical Technology, 4th
ed., Vol. 17, 1996, pp. 902-9 19.
34 Kieboom, A. P.; Moulijn, J. A.; Sheldon, R. A.; van Leeuwen, P. W. N. M., in
Catalysis: an Integrated Approach (Stud. Surf; Sci. Catal. 123; van Santen, R.A.; van
Leeuwen, P. W. N. M.; Moulijn, J. A.; Averill, B. A., Eds.), 2nd ed. 1999, pp. 29-80.
35 Bryant, D. R.; Billig, E., US. Pat. 4,277,627, 1981 (to Union Carbide); Chem. Abstr.
1978, 89, 163066.
36 Onoda, T., Chemtech 1993, 34.
37 Kuntz, E. G., Fr. Pat. 2314910, 1977 (to Rhone-Poulene); Chem. Abstr. 1977, 87,
101944.
38 Cornils, B.; Hibbel, J.; Konkol, W.; Lieder, B.; Much, J.; Schmidt, V.; Wiebus, E., Ger.
Pat. 3234701, 1984 (to Ruhrchemie); Chem Abstr. 1984, 100, 194022.
39 Kuntz, E. G., Chemtech 1987, 570.
40 Gaertner, R.; Cornils, B.; Springer, H.; Lappe, P., Ger. Pat. 3235030, 1984 (to
Ruhrchemie); Chem. Abstr. 1984, 101, 5533 1.
41 Bexten, L.; Cornils, B.; Kupies, D., Ger. Pat. 3431643, 1986 (to Ruhrchemie); Chem.
Abstr. 1986, 105, 117009.
42 Albanese, G.; Manetsberger, R.; Herrmann, W. A.; Schwer, C., Eur. Pat. Appl. 704451,
1996 (to Hoechst); Chem. Abstr. 1996, 125, 11 135.
43 Wachsen, 0.; Himmler, K.; Cornils, B., Catal. Today 1998, 42, 373.
44 Chaudhari, R. V.; Bhanage, B. B., in Aqueous-Phase Organometallic Catalysis
Cornils, B; Herrmann, W. A., Eds.), Wiley-VCH, Weinheim 1998, pp. 283-294.
45 Cornils, B.; Konkol, W.; Bach, H.; Daembkes, G.; Gick, W.; Wiebus, E.; Bahrmann, H.,
Ger. Pat. 3415968, 1985 (to Ruhrchemie); Chem. Abstr. 1986, 104, 209147.
46 Pittman, Jr., C. U.; Honnick, W. D., J. Org. Chem. 1980, 45, 2132.
47 Harano, Y., U.S. Pat. 4,465,873, 1984 (to Kuraray and Daicel Chemical Industries);
Chem. Abstr. 1984, 101,9088.
48 Matsumoto, M.; Miura, S.; Kikuchi, K.; Tamura, M.; Kojima, H.; Koga, K.; Yamashita,
S., U.S. Pat. 4,567,305, 1986; Eur. Pat. Appl. 129802, 1987 (to Kuraray and Daicel
Chemical Industries); Chem. Abstr. 1985, 102, 184723.
49 Nojori, N.; Misono, M., Appl. Catal. A General 1993. 93, 103.
50 Matsumoto, M.; Tamura, M., U.S. Pat. 4,215,077, 1980 (to Kuraray); Chem. Abstr.
1979 ,91, 174826.
51 Matsumoto, M.; Tamura, M., J. Mol. Catal. 1982, 16, 195.
52 Abatjoglou, A. G.; Bryant, D. R., Arab. J. Sci. Eng. 1985, 10, 429.
53 Yoshimura, N.; Tamura, M., in Successful Design of Catalyst (Inui, T., Ed.) 1988,
Elsevier, pp. 307-3 14.
54 Pitchai, R.; Gaffney, A. M.; Nandi, M. K.; Han, Y.-Z., U.S. Pat. 5,276,210, 1994 (to
Arco Chemical); Chem. Abstr. 1993, 119, 225573.
55 Eur. Chem News, 22-28 Nov. 1999, p. 20; 17-23 Jan. 2000, p. 33.
56 Chem. Eng. News, 1999, 18 Oct., p. 20.
57 Betts, M. J.; Dry, M. E.; Geertsema, A.; Rall, G. J. H., Int. PCT Pat. Appl. WO
97101521 1997 (to Sastech, Sasol Chemicals Europe); Chem. Abstr. 1997, 126, 173362.
230 Chapter 8
58 Abatjoglou, A. G.; Bryant, D. R.; Peterson, R. R., Eur. Pat. Appl. 350922, 1989; U.S.
Pat. 5,180,854, 1993 (to Union Carbide); Chem. Abstr. 1990, 113, 80944.
59 Paetzold, E.; Kinting, A.; Oehme, G., J. Prakt. Chem. 1987, 329, 725.
60 Breslow, D. S.; Skolnik, H., Multi-Sulfur and Sulfur and Oxygen Five- and Six-
Membered Heterocycles, Interscience, New York, 1966; Part 1, pp. 76-95; Part 2, pp.
775-789.
61 Abatjoglou, A. G.; Peterson, R. R.; Bryant, D. R., in Catalysis of Organic Reactions
(Malz, R. E., Ed.), Chem. Ind. 68, Dekker, New York, 1996, pp 133-139.
62 Billig, E.; Abatjoglou, A. G.; Bryant, D. R., U.S. Pat. 4,748,261, 1988; U.S. Pat.
4,885,401, 1989 (to Union Carbide); Chem. Abstr. 1987, 107, 7392.
63 Bryant, D. R.; Nicholson, J. C.; Briggs, J. R.; Packett, D. L.; Maher, J. M., U.S. Pat.
5,886,235, 1999 (to Union Carbide); Chem. Abstr. 1999, 130, 254055.
64 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.
65 Cuny, G. D.; Buchwald, S. L., J. Am. Chem. Soc. 1993, 115, 2066.
66 Nicholson, J. C.; Bryant, D. R.; Nelson, J. R, U.S. Pat. 5,763,679, 1998 (to Union
Carbide); Chem. Abstr. 1998, 129, 55721,
67 Nicholson, J. C.; Bryant, D. R.; Nelson, J. R.; Briggs, J. R.; Packett, D. L.; Maher, J. M.,
U.S. Pat. 5,874,639, 1999 (to Union Carbide); Chem. Abstr. 1999, 130, 184071.
68 Billig, E.; Bryant, D. R., U.S. Pat. 5.763,670, 1998 (to Union Carbide); Chem. Abstr.
1998, 129, 55719; U.S. Pat. 5,767,321, 1998 (to Union Carbide); Chem. Abstr. 1998,
129, 82975.
69 Bryant, D. R.; Nicholson, J.C., U.S. Pat. 5,763,671, 1998 (to Union Carbide); Chem.
Abstr. 1998, 129, 69096; U.S. Pat. 5,789,625, 1998 (to Union Carbide); Chem. Abstr.
1998, 129, 166637.
70 Bryant, D. R.; Nicholson, J. C.; Briggs, J. R.; Packett, D. L.; Maher, J. M., US. Pat.
5,917,095, 1999 (to Union Carbide); Chem. Ahstr. 1999, 131, 60316.
71 Bryant, D. R.; Leung, T. W.; Billig, E.; Eisenschmid, T. C.; Nicholson, J. C.; Briggs, J.
R.; Packett, D. L.; Maher, J. M., U.S. Pat. 5,874,640, 1999 (to Union Carbide); Chem.
Abstr. 1999, 130, 184072.
72 Leung, T. W.; Bryant, D. R.; Shaw, B. L., U.S. Pat. 5,731,472, 1998 (to Union Carbide);
Chem. Abstr. 1998, 128, 245454.
73 Bahrmann, H., in Aqueous-Phase Organometallic Catalysis (Cornils, B; Herrmann,
W. A., Eds.), Wiley-VCH, Weinheim 1998, pp. 306-321.
74 Tano, K.; Sato, K.; Okoshi, T, Ger. Pat. 3338340, 1984 (to Mitsubishi Chemical); Chem.
Abstr. 1984, 101, 170705.
75 Miyazawa, C.; Hiroshi, M.; Tsuboi, A.; Hamano, K., Eur. Pat. Appl. 272608, 1988 (to
Mitsubishi Chemical); Chem. Abstr. 1988, 109, 15 1931.
76 Wilne, C., in Alpha Olefins Applications Handbook (Lappin, G. R.; Sauer, J.D., Eds.),
Marcel Dekker, New York, 1989, pp. 139-199.
77 Wagner, J. D.; Lappin, G. R.; Zietz, J. R., in Kirk-Othmer Encyclopedia of Chemical
Technology, 4th ed., Vol. 1, 1991, pp. 893-913.
78 Celanese, Chem. Eng. 1981, 88 (22), 68.
79 Unruh, J. D.; Strong, J. R.; Koski, C. L., in Alpha Olefins Applications Handbook
(Lappin, G. R.; Sauer, J. D., Eds.), Marcel Dekker, New York, 1989, pp. 311-327.
80 Hanson, B. E., in Aqueous-Phase Organometallic Catalysis (Cornils, B; Herrmann,
W. A., Eds.), Wiley-VCH, Weinheim 1998, pp. 181-188.
81 Fell, B., Tenside Surf Det. 1998, 35, 326.
82 Hort, E. V.; Taylor, P., in Kirk-Othmer Encyclopedia of Chemical Technology, 4th ed.,
Vol.l, 1991, pp. 209-216.
83 Chen, S. C.; Chu, C. C., Lin, F. S.; Chou, J. Y., U.S. Pat. 5,426,250, 1995 (to Darien
Chemical).
84 Ichikawa, S.; Ohgomori, Y.; Sumitani, N.; Hayashi, H.; Imanari, M., Ind. Eng. Chem.
Res. 1995, 34, 971.
85 Weissermel, K., Arpe, H.-J., Industrial Organic Chemistry, 2nd ed., VCH, Weinheim,
1993, pp. 235-262.
86 Ohgomori, Y.; Suzuki, N.; Sumitani, N., J. Mol. Catal. 1998, 133, 289.
87 Fell, B.; Hermanns, P.; Bahrmann, H., J. prakt. Chem. 1998, 340, 459.
88 Hansen, C.B.; Teunissen, A.J.J.M., Eur. Pat. Appl. 712828 1995 (to DSM, du Pont);
Chem. Abstr. 1996, 125, 114182.
89 Wissing, E.; Teunissen, A. J.; Hansen, C. B.; van Leeuwen, P. W. N. M.; van Rooy, A.;
Burgers, D., Int.PCT Pat. Appl. WO 96/16923 1996 (to DSM, Du Pont); Chem. Abstr.
1996, 125, 114185.
90 Burke, P. M.; Gamer, J. M.; Tam, W.; Kreutzer, K. A., Teunissen, A. J. J. M.; Snijder,
C. S.; Hansen, C.B., Int. PCT Pat. Appl. WO 97/33854 1997 (to DSM, Du Pont); Chem.
Abstr. 1997, 127, 294939; U.S. Pat. 5,874,641 1999 (to DSM, Du Pont); Chem. Abstr.
1999, 130, 198123.
91 Borman, P. C.; Gelling, O. J., Eur. Pat. Appl. 839787 1998 (to DSM); Chem. Abstr.
1998, 128, 323141.
92 Bunel, E. E., Int. PCT Pat. Appl. WO 99106345 1999 (to du Pont, DSM); Chem. Abstr.
1999, 130, 167872.
93 Argyropoulos, J. N.; Bryant, D. R.,; Morrison, D. L.; Stockman, K.E., Int. PCT Pat.
Appl. WO 99138832, 1999 (to Union Carbide); Chem. Abstr. 1999, 131, 131509.
94 Tokitoh, Y.; Yoshimura, N., U.S. Pat. 4,663,468, 1987 (to Kuraray); Chem. Abstr. 1986,
104, 148739.
95 Tokitoh, Y.; Yoshimura, N., U.S. Pat. 4,808,737, 1989; Eur. Pat. Appl. 223103, 1989 (to
Kuraray); Chem. Abstr. 1987, 107, 154896.
96 Matsumoto, M.; Yoshimura, N.; Tamura, M., U.S. Pat. 4,510,332, 1985 (to Kuraray);
Chem. Abstr. 1984, 100, 22329.
97 Tokitoh, Y.; Yoshimura, N., U.S. Pat. 4,808,756, 1989 (to Kuraray); Chem. Abstr. 1987,
107, 96332.
98 Omatsu, T.; Tokitoh, Y.; Yoshimura, N., Eur. Pat. Appl. 303060, 1989 (to Kuraray);
Chem. Abstr. 1989, 111, 38870.
99 Tokuda, Y.; Ruda, K.; Suzuki, S., Jap. Pat. 11071318 1999 (to Kuraray); Chem. Abstr.
1999, 130, 222989.
100 Chem. Week 1994, Jul. 27, p. 26; Eur. Chem. News 1994, Oct. 3, p. 28.
Chapter 9
Catalyst preparation and decomposition

Institute of Molecular Chemistry, University of Amsterdam. Nieuwe Achtergracht 166, 1018
WV, Amsterdam, the Netherlands
9.1 Introduction
Catalyst preparation or setting up the catalytic system is closely related to

the decomposition of the catalysts. In this chapter we will deal with both
aspects. Rhodium hydroformylation catalysts can decompose in a variety of
ways: metal deposition, ligand decomposition, reactions with impurities,
dimer formation, and reaction of the metal center with the ligand. Sometimes
these side-reactions lead to temporary deactivation only and the catalyst
activity can be restored. Without attempting to be complete, we have
collected several examples of catalyst decomposition. In the area of
homogeneous catalysis there has always been less attention, certainly in
chemistry journals, for the stability, decomposition, and regeneration of the
catalyst as there is in the area of heterogeneous catalysis. This does not mean
that stability of the catalyst is not an issue, on the contrary. For the many
industrial applications that have come on stream in the last three decades,
catalyst stability has been a key issue, next to selectivity and activity. In
industry a considerable effort has been devoted to the study of catalyst
stability even though few accounts were published in patents or journals.
Many examples that we will mention stem from the work of the Bryant
group at Union Carbide Corporation.
9.2 Catalyst preparation
The "catalyst" for hydroformylation is a rhodium(I) hydride species that is

clearly distinct from the species that are active for hydrogenation. The
233
234 Chapter 9
hydrogenation catalysts are cationic Rh(I)+ or neutral Rh(I)Cl species.

Carbonylation of alcohols also requires an anionic Rh(I) species, e.g.
[Rh(CO)2I2]-. In modem catalysis we want to know the exact nature of our
catalysts and we would like to control its formation and stability.
Often rhodium(I) salts are used as the precursor for hydroformylation
catalysts. Under the reaction conditions (H2, CO, ligands, temperature > 25C)
these salts are converted to a rhodium hydride complex, although there are
several papers that seem to invoke cationic rhodium species as the catalysts. A
catalyst that does contain another rhodium species is the one reported by
Stanley, containing bimetallic rhodium species [Chapter 10]. Anions that have
been used include halides, conjugate bases of weak acids, thiolates, alkoxides,
etc. Chlorides have a particular deleterious effect on the activity (i.e. they are
not converted into hydrides under mild conditions). Chloride may be present in
phosphites or phosphines as a result of the synthesis; especially when the
ligands are used in a large excess this may be the cause for low activities. It has
been reported that addition of bases such as amines has a strong "promoting"
effect on such systems:
L3RhCl + H2 + CO + R3N L3Rh(CO)H + R3NH+Cl-
Rhodium salts of weaker acids are smoothly converted into rhodium

hydride without the addition of base:
L3RhA + H2 + CO L&h(CO)H + HA
A = acetate, 2,4-pentanedionate
When a large excess of carboxylic acid is present, rhodium carboxylate is

not converted into hydride [ 1]. A preferred dionate is 2,2,6,6-tetramethyl-3,5-
heptanedionate (dipivaloylmethane) especially because its solutions are stable
upon storage [2]. The use of stock solutions is recommended because very
small amounts of rhodium need to be applied when active catalysts are being
studied. In batch reactions the formation of rhodium hydride species may be
slow compared to catalysis. If so, one should pre-heat the catalyst system to
allow the formation of the catalyst before adding the alkene.
As will be discussed in section 9.3 purification of the reagents is
important as impurities may lead to decomposition of the catalytic precursor.
9. Catalyst preparation and decomposition 235
9.3 Catalyst decomposition
9.3.1 Metal plating or cluster formation
Formation of a metallic precipitate is the most common mechanism for

decomposition of a homogeneous catalyst. This is not surprising, since often
reducing agents such as dihydrogen and carbon monoxide are the reagents
used. Certainly this is true for hydroformylation. Precipitation of the metal or
metal cluster may be preceded by decomposition of the ligand, leaving
carbon monoxide as the only stabilizing ligand.
A typical example is the loss of carbon monoxide and dihydrogen from a
rhodium hydride tetracarbonyl, the hydroformylation catalyst discussed in
Chapter 2 (Figure 1).
4 HRh(CO)4 2 Rh2(CO)8 + H2 Rh4(CO)12 + 4CO
Figure 1. Decomposition of rhodium hydride carbonyl
High temperatures and low pressures accelerate the reaction. The catalyst
HRh(CO)4 must lose one molecule of CO before coordination of the alkene
substrate can take place (this is neither the catalysts resting state nor the
rate-determining step; for the sake of simplicity it is represented this way).
Thus, loss of CO is an intricate part of the catalytic cycle, which includes the
danger of a complete loss of ligands giving precipitation of the metal or one
of the carbonyl clusters. Precipitation of cobalt metal is quite common in
cobalt catalyzed hydroformylation, but because of the high cost of rhodium
(a thousand times more expensive than cobalt) precipitation of rhodium has
to be avoided under all circumstances. Addition of a phosphine or phosphite
ligand stabilizes the rhodium carbonyl species forming HRh(CO)4-n(PR3)n
complexes as discussed in Chapters 3-8. In most instances when ligands are
present, other complexes are formed upon decomposition rather than
rhodium carbonyl clusters or metal (vide infra).
Phosphorus free catalysts are used successfully both in industry and the
laboratory [3-5] and care should be taken to maintain conditions at which no
rhodium cluster or metal is formed. The precipitate observed first is that of
Rh4(CO)12.
9.3.2 Oxidation of phosphorus ligands
Phosphorus ligands are very prone to oxidation and thermodynamics

show that even water and carbon dioxide may oxidize alkyl phosphines to
the corresponding oxides. These reactions may be catalyzed by the transition
236 Chapter 9
metal in the system. Therefore, oxygen and hydroperoxides have to be

thoroughly removed from the reagents and solvents before starting our
catalysis. In spite of this common knowledge oxidation of phosphine ligands
has frequently obscured the catalytic results. Purification of the alkene feed
is often neglected. Since the alkene may be present in a thousand-fold excess
of the ligand, careful removal of hydroperoxides in the alkene is an absolute
must. Hydroperoxides are the ideal reagents for oxidizing phosphines.
Percolation over neutral alumina is usually sufficient for a hydroformylation
reaction. Treatment over sodium or sodium-potassium on a support will also
remove alkynes and dienes that may influence the catalyst performance.
Distillation from sodium may give isomerization of alkenes as an
undesirable side-reaction.
Practical recommendations and safety aspects. During the set-up of the
experiments air has to be excluded, although the reaction of oxygen with
electron-poor phosphorus compounds is much slower than that with
components of Ziegler like catalysts for instance. A simple calculation may
be useful. Suppose one uses an autoclave having a volume of 100 mL
containing 10 mL of solvent and a catalyst concentration of 1 mM, and a
tenfold excess of phosphine ligand (10 mM). The autoclave contains 0.1
mmole of phosphine. Thus, assuming a stoichiometry of one dioxygen for
two phosphine molecules we need as little as 6 mL of air to oxidize the
phosphine inventory! Flushing the tubing, pressurization and
depressurization of the autoclave with syn-gas, or evacuation, and cannula
and syringe techniques suffice to exclude oxygen in the system.
One can circumvent the problems caused by ligand oxidation by adding
an excess of the ligand, unless one is interested in the kinetics in relation to
the type of complexes present in the system. This and the addition of
triethylamine lead to a fast and practical way of doing hydroformylation
reactions. With these precautions the reaction is a highly accessible one for
people less experienced in catalysis. High-pressure equipment remains a
prerequisite (1 0 bar); we recommend the use of autoclaves, regularly tested
by the workshops, even though glassware for carrying out reactions at 5 bar
might be available.
The safety issues for working with high pressures, phosphorus
compounds, and syn-gas are well recognized. For high pressures and syn-gas
there will usually be government regulations concerning the construction,
maintenance, testing, and use of the equipment. For syn-gas one finds a
detection system in many laboratories and a regular rehearsal on how to act
when the alarm rings. Depressurization of the autoclaves (< 250 mL, < 20
bar) is done in the fume cupboards using a tube ending in the exhaust
channel. The rate of depressurization should be such that in the exhaust gas
the MAC value of CO is not exceeded.
The most commonly used phosphines such as tpp, tppms, and tppts have
been well tested for their toxicity owing to their use on industrial scale. No
particular dangers have been reported. The same holds for many phosphites,
which are also marketed as anti-oxidants. Apart from these compounds the
phosphorus compounds and their intermediates should be handled with the
same care as the toxic analogs that are known. Toxicity levels may be
comparable to common agrochemicals. Below we show one example of such
a group of phosphite ligands that turned out to be toxic (Table 1, Figure 2).
Most of the ligands we are using in the laboratory have not been tested.
Table 1. Toxicity data for a few phosphites [6]
Structure LD 50 (mg/kg, for mice,

Intraperitoneal)
R= CH 2 OH (Figure 2) > 500
R= Et (Figure 2) 1.1
R= Pr (Figure 2) 0.39
R= i-Pr (Figure 2) 0.22
Parathion (as P=S) 5.9
DFP (as P=O) 6.0
R = C2H5, C3H7, i-C 3H7, CH2OH
Figure 2. Structure of some toxic phosphites
9.3.3 Phosphorus-carbon bond breaking in phosphines
Oxidation of free phosphines was mentioned above as a reaction leading

to phosphine loss. Here we will discuss three further ways of phosphine
decomposition: oxidative addition of phosphines to low-valent metal
complexes, nucleophilic attack on coordinated phosphines, and aryl
exchange via phosphonium species. Interestingly in all cases the metal
serves as the catalyst for the decomposition reaction!
In his review [7] and feature article [8] (entitled I wonder where the
ligand went!) Garrou emphasizes the first mechanism, oxidative addition of
the phosphorus-carbon bond to low-valent metal complexes. More recently
experimental support for the other two mechanisms has been reported. In
Figure 3 the three mechanism are briefly outlined.
238 Chapter 9
Phosphonium ion formation
Figure 3. Phosphorus-carbon bond breaking
The latter mechanism has only been observed for palladium compounds
as yet, although it would also be feasible for rhodium(I) compounds giving
an anionic rhodium species and a phosphonium fragment. The actual
phosphorus carbon bond breaking occurs upon the reverse reaction, when the
phosphonium ion adds oxidatively to a low-valent metal.
Oxidative addition of P-C bond to a low-valent metal. Oxidative addition
of C-Br or C-CI bonds is an important reaction in cross-coupling type
catalysis, and while the reaction of a P-C bond is very similar, the breaking
of carbon-to-phosphorus bonds is not a useful reaction in homogeneous
catalysis. The reverse reaction, making of a carbon-to-phosphorus bond via
palladium or nickel catalysis, is becoming increasingly more important for
the synthesis of new phosphines [9]. P-C bond breaking is an undesirable
side-reaction that occurs in systems containing transition metals and
phosphine ligands and it leads to deactivation of the catalysts. The oxidative
addition of a phosphine to a low-valent transition-metal can be most easily
understood by comparing the Ph2P- fragment with a Cl- or Br- substituent at
the phenyl ring; electronically they are very akin, c.f. Hammett parameters
and the like. The phosphido anion formed during this reaction will usually
lead to bridged structures, which are extremely stable. Decomposition of
ligands during hydroformylation has been reported both for rhodium and
cobalt catalysts [10-12].
Thermal decomposition of RhH(CO)(PPh3)3, in the absence of H2 and
CO, leads to a stable cluster shown in Figure 4 containing 2-PPh2 fragments
[ 13] as was studied by Bryants group at Union Carbide. Presumably clusters
of this type also form in hydroformylation plants on the long run. Recovery
of rhodium from these inert clusters is a tedious operation. Reaction of the
cluster mixture with reactive organic halides such as allyl chloride has been
described to give allyldiphenylphosphine and rhodium chloride, which can
be easily extracted into a water layer. [ 14].
Figure 4. Rhodium cluster formed from decomposition of RhH(PPh 3)3CO
Under hydroformylation conditions also other products are found such as

benzaldehyde, benzene, and diphenylpropylphosphine. The mechanism for
their formation is outlined in Figure 5.
Figure 5. Formation of side-products during ligand decomposition
The phenyl group formed ends up as its hydrogenation or carbonylation

product. Instead of cluster formation the phosphido group may give a
reductive elimination starting from a rhodium propyl species, which gives
diphenylpropylphosphine as the product. The disadvantage of this phosphine
is that it is a stronger electron-donor than tpp and it leads to a less active
rhodium catalyst. Thus at some stage it has to be removed, which can be
done by utilizing its higher basicity by selective phosphonium salt formation.
Benzene and benzaldehyde byproducts were also observed in the
Ruhrchemie-Rhne Poulenc process using the trisulfonated analog of
triphenylphosphine [15], but the decomposition was reported to be much
slower for tppts as compared to tpp. This is an accidental favorable aspect of
the RC-RP process; it is unexpected because water might oxidize the
phosphine, or it might carry out a nucleophilic attack, thus initiating the
second mode of decomposition (vide infra).
Cluster or bimetallic reactions have also been proposed in addition to
monometallic oxidative addition reactions. The reactions do not basically
change. Several authors have proposed a mechanism involving
orthometallation as a first step in the degradation of phosphine ligands,
240 Chapter 9
especially in the older literature. orthometallation does take place as can be

inferred from deuteration studies, but it remains uncertain whether this is a
reaction necessarily preceding the oxidative addition (Figure 6):
Figure 6. Orthometallation
Subsequently the phosphorus-to-carbon bond is broken and the benzyne

intermediate inserts into the metal hydride bond. Although this mechanism
has been popular with many chemists there are many experiments that
contradict this mechanism. A simple para-substitution of the phenyl group
would answer the question whether orthometallation was involved as is
shown in Figure 7:
Figure 7. Disproof of orthometallation as the mechanism for P-C cleavage
Decomposition products of tolylphosphines should give 3-methyl

substituents if the orthometallation mechanism is operative. For palladium
catalyzed decomposition of triarylphosphines this is not the case [ 16].
Likewise, Rh, Co, and Ru hydroformylation catalysts give aryl derivatives
not involving C-H activation [ 17, 1 8]. Several rhodium complexes catalyze
the exchange of aryl substituents at triarylphosphines [ 18]:
R'3P + R3P (Rh) > R'R2P+R'2RP
These authors propose as the mechanism for this reaction a reversible

oxidative addition of the aryl-phosphido fragments to a low valent rhodium
species. A facile aryl exchange has been described for complexes
Pd(PPh3)2(C6H4CH3)I. Again the authors [19] suggest a pathway involving
oxidative additions and reductive eliminations. The mechanism outlined
below, however, can also explain the results of these two studies.
Phosphido formation has been observed for many transition metal
phosphine complexes [7, 8]. Upon prolonged heating and under an
atmosphere of CO and/or H2 palladium and platinum also tend to give stable

phosphido bridged dimers or clusters [20,21].
Nucleophilic attack. Current literature underestimates the importance of
nucleophilic attack as a mechanism for the catalytic decomposition of
phosphines, especially with nucleophiles such as acetate, methoxy, hydroxy
and hydride. For examples of nucleophilic attack at coordinated phosphorus
see references [20-25]. A very facile decomposition of alkylphosphines and
triphenylphosphine (using palladium acetate, one bar of hydrogen and room
temperature) has been reported [20] using acetate or hydride as the
nucleophile.
A detailed reaction proving the nucleophilic attack was shown for
platinum complexes [25]. The alkoxide coordinated to platinum attacks
phosphorus while the carbon atom coordinated to platinum migrates to
phosphorus. Thermodynamically the result seems more favorable, but
mechanistically this shuffle remains mysterious (see Figure 8).
Coordination to platinum makes the phosphorus atom more susceptible to
nucleophilic attack, and the harder atoms (P and O) and softer ones (C and
Pt) recombine as one might expect. The same mechanism was proposed by
Matsuda [22] for the decomposition of triphenylphoshine by palladium(II)
acetate. In this study the aryl phosphines are used as a source for aryl groups
that are converted into stilbenes via a Heck reaction. Even alkyl phosphines
underwent P-C bond cleavage via palladium acetate.
Figure 8. Intramolecular nucleophilic attack
A catalytic decomposition of triphenylphosphine has been reported [26]

in a reaction involving rhodium carbonyls, formaldehyde, water, and carbon
monoxide. Several hundreds of moles of phosphine can be decomposed this
way per mole of rhodium per hour! The reactions that may be involved are
shown in Figure 9.
Related to this chemistry is the hydroformylation of formaldehyde to give
glycolaldehyde, which would be an attractive route from syn-gas to ethylene
glycol. The reaction can indeed be accomplished and is catalyzed by
rhodium arylphosphine complexes [27], but clearly phosphine
decomposition is one of the major problems to be solved before
formaldehyde hydrofomylation can be applied commercially.
242 Chapter 9
Figure 9. Catalytic decomposition of tpp by formaldehyde and hydrogen
The interest in cross-coupling reactions in the last decade has lead to a

large number of reports dealing with the involvement of the ligands in these
reactions [e.g. 28-30]. The mechanism has not been elucidated for all cases.
Norton [28] proposes an intramolecular nucleophilic attack of a palladium
bonded methyl group to a coordinated aryl ligand and explicitly excludes the
intermediacy of phosphonium species, as deuterated phosphonium salts
present in solution and having the same composition did not participate in
the reaction. The mechanism involving phosphonium species will be
discussed in the next section.
Aryl exchange via phosphonium intermediates. More recently a variation
of the above mechanisms was reported by Novak [30]. Formally the
mechanism also involves nucleophilic attack at coordinated phosphines, but
after the nucleophilic attack the phosphorus moiety reductively eliminates
from the metal as a phosphonium salt. To obtain a catalytic cycle the
phosphonium salt re-adds to the zerovalent palladium complex (Figure 10).
Figure 10. Aryl exchange via phosphonium formation
Scavenging of free phosphines by electrophiles such as protons, other

metals, conjugated enones, etc. presents a potential route to phosphine loss
in catalytic systems. As yet, the participation of phosphonium intermediates
has not been reported for rhodium hydroformylation catalysts, but they could
be easily conceived, especially when dienes or enones are also present.
9.3.4 Decomposition of phosphites
Phosphites are easier to synthesize and less prone to oxidation than

phosphines. They are much cheaper than most phosphines and a wide variety
can be obtained commercially as they are used as anti-oxidants.
Disadvantages of the use of phosphites as ligands include several side
reactions: hydrolysis, alcoholysis, trans-esterification, Arbusov
rearrangement, O-C bond cleavage, P-O bond cleavage. Figure 11 gives an
overview of these reactions. In hydroformylation systems at least two more
reactions may occur, namely nucleophilic attack to aldehydes, and oxidative
cyclizations with aldehydes.
Figure 11. Decomposition pathways of phosphites
Phosphites have been extensively studied for their use as ligands in

rhodium-catalyzed hydroformylation (see Chapter 3). The first publication
on the use of phosphites is from Pruett and Smith, from Union Carbide [31].
The first exploitation of bulky monophosphites was reported by van
Leeuwen and Roobeek [32]. They found that very high rates can be obtained
for internal and terminal alkenes, but selectivities were low for linear
alkenes. The bulky phosphites not only gave higher rates than less bulky
phosphites, but they are also more resistant to hydrolysis. Bryant and
coworkers [33] introduced even more stable, bulky phosphites by the
244 Chapter 9
Figure 12. Bulky phosphite ligands
introduction of bisphenols instead monophenols for the phosphite synthesis

(see Figure 12). High selectivities are only obtained when diphosphites are
used.
Diphosphites came into focus after the discovery of Bryant and
coworkers at Union Carbide Corporation that certain bulky diphosphites lead
to high selectivities in the rhodium catalyzed hydroformylation of terminal
and internal alkenes [34] (see Figure 13). A plethora of diphosphites has
been tested and recorded in many patents authored by coworkers of several
companies [35-37], indicating their importance for the near future in this
field. The patents included in the references of this chapter serve only as
examples.
Figure 13. Examples of Union Carbide's diphosphites
The advantages of the diphosphite system for propene hydroformylation

as compared to the commercial triphenylphosphine system are that less
ligand and less rhodium are required and that higher rates can be obtained.
Also high selectivities for conversion of internal alkenes to linear products
have been reported.
It should be noted that all phosphites reported are aryl phosphites
(sometimes the backbones may be aliphatic) and that the favored ones often
contain bulky substituents. One of the reasons that aliphatic phosphites are
used only sparingly is that they are susceptible to the Arbusov rearrangement
while the aryl phosphites are not. Acids, carbenium ions, and metals catalyze
the Arbusov rearrangement. Many examples of metal catalyzed
decomposition reactions have been reported [38, 39] (see Figure 14).
Figure 14. Metal catalyzed Arbuzov reaction
Thorough exclusion of moisture can easily prevent hydrolysis of

phosphites in the laboratory reactor. In a continuous operation under severe
conditions traces of water may form via aldol condensation of the aldehyde
product. Weak and strong acids and strong bases catalyze the reaction. The
reactivity for individual phosphites spans many orders of magnitude. When
purifying phosphites over silica columns in the laboratory one usually adds
some triethylamine to avoid hydrolysis on the column.
Bryant and coworkers have extensively studied decomposition of
phosphites [40]. Stability involves thermal stability, hydrolysis, alcoholysis,
and stability toward aldehydes. The precise structure has an enormous
influence on the stability. Surprisingly it is the reactivity toward aldehydes
that received most attention. Older literature [4 1] mentions several reactions
between phosphites and aldehydes of which we show only two in Figure 15.
The addition of a phosphite to an aldehyde giving a phosphonate is the
most important reaction [40]. The reaction is catalyzed by acid and since the
product is acidic, the reaction is autocatalytic. Furthermore, acids catalyze
hydrolysis and alcoholysis and therefore the remedy proposed is continuous
removal of the phosphonate over a basic resin (Amberlyst A-21). The
examples in the patents illustrate that very stable systems can be obtained
when the acidic decomposition products are removed continuously.
The thermal decomposition of phosphites with aldehydes is illustrated in
Figure 16 [40]. In this experiment the ligands are heated for 23 hours at 160
C in the presence of pentanal. The figure shows the percentages of ligand
that have decomposed as measured by NMR spectroscopy. This was done in
the absence of a rhodium catalyst and so the numbers present a lower limit
for the decomposition. One can see that minor effects in the ligand structure
can have a tremendous influence on the rate of decomposition or rather the
rate of reaction with aldehydes. While simple hydrolysis or alcoholysis
might have been expected as the major source for ligand decomposition we
see that a study to what is actually taking place is worthwhile as an attempt
to stabilize the catalyst system.
246 Chapter 9
Figure 15. Reaction of phosphites with the aldehyde product
Figure 16. Stability of phosphites at 160 C after 23 hours (see text for explanation; numbers
give percentage of decomposition)
9.3.5 Formation of dormant sites
Dienes and alkynes are poisons for many alkene processes. In polyolefin
manufacture they must be carefully removed as they deactivate the catalyst
completely. Insertion of conjugated dienes is even much more rapid than
insertion of ethene and propene. The resulting -allyl species are inactive as
catalysts.
In rhodium catalyzed hydroformylation the effect is less drastic and often
remains unobserved, but surely diene impurities obscure the kinetics of
alkene hydroformylation [42]. Because the effect is often only temporary we
summarize it here under dormant sites. Hydroformylation of conjugated
alkadienes is much slower than that of alkenes, but also here alkadienes are
more reactive than alkenes toward rhodium hydrides [43, 44]. Stable -allyl
complexes are formed that undergo very slowly insertion of carbon
monoxide (Figure 17). The resting state of the catalyst will be a -allyl
species and less rhodium hydride is available for alkene hydroformylation.
Thus, alkadienes must be thoroughly removed as described by Garland [45],
especially in kinetic studies. It seems likely that 1,3- and 1,2-diene
impurities in 1 -alkenes will slow down, if not inhibit, the hydroformylation
of alkenes.
Figure 17. Reactions of diene to give -allyl complexes
Dimer formation. Active, monomeric catalyst species may be involved

the formation of inactive dimers. When this equilibrium is reversible it only
leads to a reduction of the amount of catalyst available and it does not bring
the catalysis to a full stop.
A well-known example is the formation of the so-called orange dimer
from HRh(PPh3)3CO, already reported by Wilkinson [46]. Since then it has
been reported by several authors [47 and references therein].
Figure 18. Dimer formation for hydrofomylation catalysts

248 Chapter 9
Since the hydroformylation reaction for most substrates shows a first

order dependence in 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 [47] (see Figure 18).
Work-up of hydroformylation solutions often leads to formation of
dimers. It seems likely that in a liquid recycle of a continuous reactor
rhodium will occur as such a dimeric species. Since the reaction with
hydrogen is reversible it presents a means to recycle rhodium. Dimer
formation is not restricted to phosphines as phosphites behave similarly [48].
When dimer formation becomes important one might attempt to
destabilize the dimer relative to the monomer. For instance making the
ligand very bulky might prevent dimer formation. Models show that the
ligand size must be substantially increased to arrive at the desired effect.
Another approach is so-called site isolation as was described by Grubbs
[49]. His well-known example concerns a titanocene catalyst that is used as a
hydrogenation catalyst. The intermediate titanium hydride is converted
almost completely to a dimer rendering the catalyst with a low activity.
Immobilization of the catalyst on a resin support prevents dimerization and
an active catalyst is obtained.
Ligand metallation. In early transition metal polymerization catalysis
often metalation of the ligand occurs leading to inactive catalysts. In late
transition metal chemistry the same reactions occur, but now the complexes
formed represent a dormant site and catalyst activity can often be restored.
Work-up of rhodium-phosphite catalyst solutions after hydroformylation
often shows partial formation of metallated species, especially when bulky
phosphites are used [50]. Dihydrogen elimination or alkane elimination may
lead to the metallated complex. The reaction is reversible for rhodium and
thus the metallated species could function as a stabilized form of rhodium
during a catalyst recycle. Many metallated phosphite complexes have been
reported, but we mention only two, one for triphenyl phosphite and rhodium
[51, 52] (see Figure 19) and one for a bulky phosphite and iridium [53].
Figure 19. Metallation in rhodium phosphite complexes

9.4 Concluding remarks
Decomposition of organometallic catalysts under reaction conditions of

catalytic processes lead to interesting organometallic chemistry. The study of
this chemistry is often worthwhile either to find ways of circumventing the
decomposition reactions or to find ways to work up the solutions such that
the metal can be regenerated for further use and separated from the
decomposed ligand fragments. From a stability viewpoint ligands having
phosphorus as the donor atom are not very attractive, but for none of the
metals active in hydroformylation (cobalt, platinum, palladium, rhodium)
there seems to be an alternative other than carbon monoxide. Arsines lead to
active catalysts, but it is not expected that they will form more stable
complexes than phosphines. For several metal catalyzed reactions recently
nitrogen ligands (Fe, Co, Pd) have been introduced which are often much
more robust than phosphorus ligands. Rhodium hydride carbonyls do not
seem to combine well with nitrogen ligands.
References
1 (a) Mieezynska, E.; Trzeciak, A. M.; Zilkowski, J. J. J. Mol. Catal. 1993, 80, 189. (b)
Buhling, A.; Kamer, P. C. J.; van Leeuwen, P. W. N. M. J. Mol. Catal. A: Chemical,
1995, 98, 69.
2 Coolen, H. K. A. C.; van Leeuwen, P. W. N. M.; Nolte, R. J. M. J. Org. Chem. 1996, 61,
4739.
3 Onada, T. ChemTech, September 1993 , 34.
4 Lazzaroni, R.; Pertici, P.; Bertozzi, S.; Fabrizi, G. J. Mol. Cat. 1990, 58, 75.
5 Vidal, J. L.; Walker, W. E. Inorg. Chem. 1981, 20, 249.
6 Bellet, E. M.; Casida, J. E. Science, December 1973, 182, 1135.
7 Garrou, P. E. Chem. Rev. 1985, 85, 171.
8 Garrou, P. E.; Dubois, R. A.; Jung, C. W. ChemTech, February 1985, 123.
9 Cai, D.; Payack, J. F.; Bender, D. R.; Hughes, D. 1.; Verhoeven, T. R.; Reider, P. J. J.
Org. Chem. 1994, 51 ,629.
10 Chini, P.; Martinengo, S.; Garlaschelli, G. J. Chem. Soc. Chem. Commun. 1972, 709.
11 Dubois, R. A.; Garrou, P. E. Organometallics, 1986, 5, 466.
12 Harley, A. D.; Guskey, G. J.; Geoffroy, G. L. Organometallics, 1983, 2, 53.
13 Billig, E.; Jamerson, J. D.; Pruett, R. L. J. Organomet. Chem. 1980 192, C49.
14 Miller, D. J.; Bryant, D. R.; Billig, E.; Shaw, B. L. U.S. Pat 4,929,767 (to Union Carbide
Chemicals and Plastics Co.) 1990; Chem. Abstr. 1991, 113, 85496.
15 Herrmann W. A.; Kohlpaintner, C. W. Angew. Chem. Int. Ed. Engl. 1993, 32, 1524.
16 Goel, A. B. Inorg. Chim. Acta, 1984, 84, L25.
17 Sakakura, T. J. Organometal. Chem. 1984, 267, 17 1.
18 Abatjoglou A. G.; Bryant, D. R. Organometallics 1984, 3, 932.
19 Kong, K-C.; Cheng, C-H. J. Amer. Chem. SOC. 1991, 113, 6313.
20 Sisak, A.; Ungvry, F.; Kiss, G. J. Mol. Catal. 1983, 18, 223.
250 Chapter 9
21 van Leeuwen, P. W. N. M.; Roobeek, C. F.; Frijns, J. H. G.; Orpen, A. G.

22 Kikukawa, K.; Takagi, M.: Matsuda, T. Bull. Chem. Soc. Japan, 1979, 52, 1493.
23 Bouaoud, S-E.; Braunstein, P.; Grandjean, D. ; Matt, D. Inorg. Chem. 1986, 25, 3765.
24 Alcock, N. W.; Bergamini, P.; Kemp, T. J.; Pringle, P. G. J. Chem. Soc. Chem. Commun.
1987, 235.
25 van Leeuwen, P. W. N. M.; Roobeek, C. F.; Orpen, A. G. Organometallics, 1990, 9,
2179.
26 Kaneda, K.; Sano, K.; Teranishi, S.; Chem Lett. 1979, 82.
27 Chan. A. S. C.; Caroll, W. E.; Willis, D. E. J. Mol. Catal. 1983, 19, 377.
28 Morita, D. K.; Stille, J. K.; Norton, J. R. J. Am. Chem. Soc. 1995, 117, 8576.
29 Segelstein, B. E.; Butler, T. W.; Chenard, B. L. J. Org. Chem. 1995, 60, 12.
30 Goodson, F. E.; Wallow, T. I.; Novak, B. M. J. Am. Chem. Soc. 1997, 119, 12441.
31 Pruett, R. L.; Smith, J. A. J. Org. Chem. 1969, 34, 327.
32 van Leeuwen P. W. N. M.; Roobeek, C. F. J. Organometal. Chem. 1983, 258, 343. Brit.
Pat. 2,068,377, USPat. 4,467,116 (to Shell Oil); Chem. Abstr. 1984, 101, 191 142.
33 Billig, E.: Abatjoglou, A. G.; Bryant, D. R.; Murray, R. E.; Maher, J. M. U.S. Pat.
4,599,206 (to Union Carbide Corp.) 1986; Chem. Abstr. 1989, 109, 233 177.
34 Billig, E.; Abatjoglou, A. G.; Bryant, D. R. (to Union Carbide Corporation) U.S. Pat.
4,769,498, U.S. Pat. 4,668,651; U.S. Pat. 4,748,261, 1987; Chem. Abstr. 1987, 107,
7392.
35 Burke, P. M.; Garner, J. M.; Tam, W.; Kreutzer, K. A.; Teunissen, A. J. J. MWO
97/33854, 1997, (to DSM/Du Pont); Chem. Abstr. 1997, 127, 294939.
36 Sato, K.; Kawaragi, Y.; Takai, M.; Ookoshi, T. US Pat. 5,235,113, EP 518241 (to
Mitsubishi); Chem. Abstr. 1993, 118, 191 183.
37 Rper, M.; Lorz, P. M.; Koeffer, D. Ger. Offen. DE 4,204,808 (to BASF); Chem. Abstr.
1994, 120, 133862.
38 Brill, T. B.; Landon, S. J. Chem. Rev. 1984, 84, 577.
39 Werener, H.; Feser, R. Z. Anorg. Allg. Chem. 1979, 458, 301.
40 Billig, E.; Abatjoglou, A. G.; Bryant, D. R.; Murray, R. E.; Maher, J. M. (to Union
Carbide Corporation) U.S. Pat. 4,717,775, 1988; Chem. Abstr. 1989, 109, 233177.
41 Ramirez, F.; Bhatia, S. B.; Smith, C. P. Tetrahedron, 1967, 23, 2067.
42 van den Beuken, E.; de Lange, W. G. J.; van Leeuwen, P. W. N. M.; Veldman, N.; Spek,
A. L.; Feringa, B. L. J. Chem. Soc. Dalton Trans. 1996, 3561.
43 van Leeuwen, P. W. N. M.; Roobeek, C. F. J. Mol. Catal. 1985, 31, 345.
44 van Rooy, A.; de Bruijn, J. N. H.; Roobeek, C. F.; Kamer, P. C. J.; van Leeuwen, P. W.
N. M. J. Organomet. Chem., 1996, 507, 69.
45 Fyhr, C.; Garland, M. Organometallics, 1993, 12, 1753.
46 Evans, D.; Yagupsky, G.; Wilkinson, G. J. Chem. Soc. (A) 1968, 2660.
47 Castellanos-Pez, A.; Castilln, S.; Claver, C.; van Leeuwen, P. W. N. M.; de Lange, W.
48 Buisman, Thesis, Amsterdam, 1995.
49 Bonds, W. D.; Brubaker, C. H.; Chandrasekaran, E. S.; C. Gibsons, Grubbs, R. H.; Kroll,
L. C. J. Am. Chem. Soc. 1975, 97, 2128.
50 van Rooy, A.; Buisman, G. J. H.; Roobeek, C. F.; Sablong, R.; van Leeuwen, P. W. N.
M. unpublished results.
51 Parshall, G. W.; Knoth, W. H.; Schunn, R. A. J. Am. Chem. SOC. 1969, 91, 4990.
52 Coolen, H. K. A. C.; van Leeuwen, P. W. N. M.; Nolte, R. J. M. J. Organomet. Chem.
1995, 496, 159.
53 Fernandez, E.; Ruiz, A.; Claver, C.; Castillon, S.; Chaloner, P. A.; Hitchcock, P. B.
Inorg. Chem. Comm. 1999 ,2, 21.
Chapter 10
Novel developments in hydroformylation
Joost N. H. Reek, Paul C. J. Kamer, and Piet W. N. M. van Leeuwen

Institute of Molecular Chemistry, University of Amsterdam, Nieuwe Achtergracht 166, 1018 WV
Amsterdam, The Netherlands.
10.1 Introduction
In this chapter we will describe several novel developments in catalytic

hydroformylation that might be of importance for future applications. This
involves novel catalysts that are highly active and selective, but more
importantly new systems that are sustainable. These processes should be
selective and efficient with a very small waste stream. A very good example
of an environmentally benign process is the aqueous biphasic system
developed by Ruhrchemie/Rhne-Poulene [ 1]. The water-soluble ligand
used (TPPTS: triphenylphosphine sulfonate) affords, in combination with a
rhodium precursor, an active catalyst for the hydroformylation of propene
that can be separated very easily from the organic phase and recycled
(Chapter 7). This aqueous biphasic system can be applied only to substrates
that are slightly water-soluble. Therefore, several alternatives for higher
alkenes have been investigated in the past decade. In this chapter we first
will discuss an interesting novel bimetallic active and selective catalyst
system that does not operate via the common mononuclear Rh(I) hydride
precursor as described in chapter 1. We will proceed with the description of
systems that show novel concepts in catalyst/product separation and end
with recent developments in supramolecular catalysis.
10.2 New bimetallic catalysts
It has been shown that a number of active sites of enzymes contain two
metal ions that give high activity via a cooperative bimetallic mechanism
253
254 Chapter 10
[2]. Synthetic binuclear complexes as catalysts, aiming at high activity, have

been subjects of intensive research [3]. However, the number of novel
bimetallic catalysts that are active and actually operating via a bimetallic
mechanism is extremely scarce. Stanley and co-workers reported a
bimetallic rhodium complex that is an active and selective hydroformylation
catalyst that was proposed to operate via a cooperative mechanism.
It has been shown that the tetraphosphine ligands 1 can simultaneously
complex two metal centers [4]. From an X-ray structure determination of
Rh2(nbd)2 (rac-l), it is clear that the two rhodium atoms are complexed in a
square planar geometry forming 5-membered rings with the ligand and a
metal-metal distance of 5.5 . The catalytic performance was compared with
mononuclear complexes of ligands 2-5. The activities of the rhodium
norbornadiene (nbd) complexes of ligands 1-7 have been tested in the
hydroformylation of 1-hexene [4] (Table 1). The results clearly show that
Rh2(nbd)2 (rac-1) forms a catalyst superior to all the other systems. Both the
activity and the selectivity for the linear products are much higher. The large
difference in activity between the bimetallic Rh2(nbd)2 (rac-1) and
monometallic Rh2(nbd)2 (2-5) strongly suggests that the former complex acts
via a different mechanism.
Stanley proposed a mechanism for the bimetallic hydroformylation that

slightly deviates from the mechanism suggested by Heck and Breslow for
the cobalt catalyzed reaction [5]. Spectroscopic evidence was found for the
existence of unusual dicationic bimetallic rhodium(II) complexes, supporting
the proposed mechanism [6] (see figure 1). The bimetallic cooperativity was
thought to originate from the intramolecular hydride transfer from the
rhodium hydride to the rhodium acyl species (in intermediate f), enhancing
the elimination of the aldehyde from the acyl intermediate.
10. Novel developments in hydroformylation 255
a
Table 1. Hydroformylation of 1-hexene using ligands 1-7
Ligand Initial TOF Aldehydes 1:b Isomerization Hydrogenation

-1
(mol.mol .h-1) (%) ratio (%) (%)
rac-1 636 85 27.5 8 3
meso-1 54 16 14 4 2
2 1 0.9 2 16 5
3 1.5 1.7 3 50 15
4 1.2 1.1 3 42 14
5 2.4 1.2 2.6 51 14
6 5.8 1.7 2.4 59 16
7 0 0 60 17
a
Catalyst: Rh2(nbd)2(BF4)2 and ligand. Reaction conditions: 6.2 bar CO/H 2 (1 : 1); ImM
catalyst in acetone, 1M hexene, reaction performed at 90 C for three hours.
Figure I. The proposed mechanism for the bimetallic hydroformylation using ligand 1
The bridge between the two metal centers is too large in ligands 6 and 7,
which prevents the cooperation between the metal centers. Indeed activities
and selectivities were found that were similar to the monometallic species.
Another important indication supporting this bimetallic cooperativity was an
256 Chapter 10
observed alkene inhibition of the catalytic reaction. At higher concentration

of the alkene a bis-acyl species can be formed, preventing the fast
intramolecular hydride transfer.
10.3 Novel developments in catalyst separation
The great potential of homogeneous catalysis in the selective preparation

of chemicals has been widely recognized and the number of industrial
applications is growing enormously. If a general solution would be available
for the often cumbersome separation of homogeneous catalysts from the
product, the number of commercial applications would even further increase.
A general solution allows the development of "general purpose operating
units", which is of special interest since it can lower the costs of chemical
processes. With this in mind a lot of investigators focussed on novel methods
of separating and re-using homogeneous catalysts. As yet, there is not one
general solution for this problem, but several interesting concepts have been
reported. In this section of the most representative concepts will be
discussed. As we will see, various attempts are based on extending the two-
phase Ruhrchemie/Rhne Poulenc process in such a way that it may become
applicable to larger organic substrates.
10.3.1 Micellar catalysis
One simple way to increase the solubility of organic substrates in

aqueous solution is the addition of co-solvents to an aqueous two-phase
system. This increases the reaction rate of the Rh/TPPTS catalyst, but at the
same time the selectivity drops [7]. Furthermore, it can result in troublesome
phase separation and leaching of the co-solvent into the product phase [7].
Instead of the addition of co-solvents triphenylphosphine can be added to the
two-phase Rh/TPPTS system. The local concentration of the active rhodium
complex at the interface will be increased, resulting in a rate enhancement in
the hydroformylation of 1-octene by a factor of 10-50 [8]. The active species
at the interface contains both TPPTS and triphenylphosphine ligands. Test
reactions performed in methanol confirmed that this increase in reaction rate
was due to promotion of interfacial catalysis. However, after recycling the
catalyst Rh/TPPTS was found in the aqueous phase whereas the PPh3 was in
the organic phase and in the subsequent reaction with freshly added organic
phase the activity of the catalyst is similar to that of the biphasic reaction in
the absence of PPh3.
Another way to improve the solubility of organic substrates in the
aqueous phase is the addition of amphiphiles (also called surfactants or
tensides) that form aggregates [9]. Above the critical aggregation
concentration (CAC) amphiphiles form organized structures in aqueous

solution in such way that the polar groups are pointing to the water phase
and the organic parts of the molecules stick together (Figure 2). Dependent
on the relative sizes of the head group and the organic part of the molecule
one observes the formation of micelles, vesicles or bilayers. Organic
substrates will dissolve in the organic part of the aggregates. Thus the
addition of amphiphiles will lead to an increased solubility of the substrate,
but a water-soluble catalyst (e.g. Rh/TPPTS) will be in a different phase. The
combination of organic substrates with a non water-soluble catalyst induces
high local concentrations of both components in the aggregate, which can
result in impressive rate enhancements [10]. This type of concentration or
medium effects introduced by micelles has been extensively studied [9]. The
problem of catalyst-product separation still subsists since both the catalyst
and the substrate will be extracted into the organic phase. Only recently
amphiphilic ligands have been used in catalysis.
Figure 2. Schematic presentation of the aggregation of amphiphiles into micelles, bilayers and
vesicles
Ligands that have an amphiphilic character and form the solubilizing

aggregate are themselves very interesting, since they can combine the
advantage of two-phase catalysis, i.e. easy catalyst separation, with a
reasonable solubilizing effect on organic substrates. Fell [ 11] and Hanson
[ 12, 13] reported the first compounds that combined these properties. Fell
prepared a series of zwitterionic phosphines with different organic chain
lengths (8) (see Figure 3) and he found an optimal catalytic performance
using the octane derivative (n=7) at 75 bar pressure (CO/H2=1: 1). Hanson
prepared the sulfonated ligands 9, and showed with dynamic light
258 Chapter 10
Table 2. Two-phase hydrofornylation of 1-octene comparing 9c, 10b and TPPTS

Ligand L/Rh TOF Yield nonanals 1:b
(mol.mol-1.h-1) (%)
9c 2 190a 38 2.4
9c 5 400 a 80 4.5
9c 9 435 a 87 8.1
TPPTS 2 70 a 14 2.2
a
TPPTS 5 145 29 3.1
TPPTS 9 220 a 44 3.5
10b 2 63 b 63 1.8
10b 5 76 b 76 3.7
b
10b 9 83 83 10.1
TPPTS 2 36 b 36 2.1
b
TPPTS 5 57 57 3.0
TPPTS 9 76 b 76 3.1
Reaction conditions: T=12OoC, initial pressure (at 25C) 14.3 bar (CO/H2=1 : 1),
[Rh]=0.0049 M, Rh/1-octene=500, MeOH/H2O=1:1. aTOF after 1 h. bTOF after 5 h.
scattering that at room temperature these amphiphiles indeed formed

micellar aggregates in aqueous solution.
The catalytic performance of the catalysts based on these amphiphilic
ligands was compared with that of TPPTS (see table 2). The reactions were
performed in aqueous methanol. High activity was observed for 9c compared
to TPPTS. In all cases the ligand concentration has a large effect on the
selectivity and activity of the catalyst. The highest selectivity towards the
linear product was found for the amphiphilic BISBI analogue 10b (1:b=10).
An effect of the co-solvent methanol on the selectivity of the reaction was
observed for TPPTS, which has been attributed to an increased intra-
complex ionic repulsion between the sulfonate groups. Using ligand 9c these
groups are further apart and as a consequence this repulsion is smaller. This
results in much better selectivities towards the linear product compared to
TPPTS. Under these conditions the BISBI analogue 10b only performed
slightly better than TPPTS and selectivity toward the linear product is lower
than that observed for the homogeneous reaction. Catalysis in aqueous
solution using these systems was not reported.
Amphiphilic diphosphines based on the Xanthene backbone have been
synthesized (11) [14]. The large natural bite angle of ligands based on the
Xanthene backbone has a beneficial influence on the regioselectivity of the
homogeneous hydroformylation catalyst. Furthermore, the chelate effect
results in a strong metal-ligand bond, which is of importance in recycling
experiments. Compounds 11b and 11c form vesicles in aqueous solution.
According to electron microscopy and light scattering experiments the size
of these vesicles was around 100 nm. Furthermore, the aggregates were
stable at temperatures up to 90 C and they were able to dissolve organic
Figure 3. Some of the amphilic phosphine ligans that have been reported
Figure 4. Schematic representation of catalysis using amphiphilic ligands that form vesicles
substrates in the organic phase of the aggregate.

These are obviously important properties for the application of these
ligands in the hydroformylation (at elevated temperatures) of organic
substrates such as 1 -octene in aqueous solution (see Figure 4). Importantly,
the addition of methanol resulted in the disruption of the vesicles.
Hydroformylation reactions were therefore performed in aqueous solution, at
70 C and 80 C using 15 bar CO/H 2 (1:1). The rate enhancement observed
for the aggregating systems was a factor of 14 and the high selectivity for the
linear product (1:b = 50) usually observed for these type of catalysts was
retained. Test reactions performed in aqueous methanol solution showed that
the enhancement was only a factor of 3. Recycling experiments showed that
the catalyst/product separation was good; the catalyst performance was
260 Chapter 10
retained completely in 4 consecutive runs and no rhodium leaching into the

organic phase was detected. Surprisingly, no emulsions were formed on
using 11b and 11c, allowing easy and fast separations.
10.3.2 Supported aqueous phase catalysis (SAPC)
Davis and Hanson developed a new concept of immobilizing

homogeneous catalysts denoted as supported aqueous phase catalysts
(SAPC) [15]. They reasoned that in aqueous biphasic catalysis the reaction
mainly takes place at the interface. In order to increase this interface they
used a high-surface-area hydrophilic support (figure 5). These materials have
a thin film of water adhered to the surface, in which the water-soluble
catalyst is dissolved. The reaction, performed in an organic solvent such as
toluene, occurs at the water-organic interface. The supported catalyst has a
Figure 5. Schematic representation of the concept of supported aqueous phase catalysis
very large interfacial area, which results in very efficient catalysis for
organic substrates. Furthermore, the catalyst stays completely on the support.
The important issues are the generality of the concept, the robustness of the
catalyst system and the influence of the thickness of the water layer [16].
This water layer has an enormous impact on the catalytic activity (figure 6).
If this layer is too thin the activity of the catalyst is much lower due to a
decrease of the catalyst mobility. The catalyst is bound to the silica resulting
in a heterogeneous system.
This was verified by 31P NMR spin relaxation measurements, which
show that the spin relaxation declines with increasing water content [ 17]. It
was observed in the hydroformylation of 1-heptene using a
HRh(CO)(TPPTS)3-SAPC that the TOF increases with a factor as high as
100 when going from 2.9 to 9 wt. % water (table 3).
Table 3. Hydroformylation performance of SAPC compared with homogeneous and biphasic

systems.
Catalyst system Substrate T (C) TOF (mol.mol-1.h-1)

SAPC (2.9 wt.% H 2 O) a 1-heptene 75 0.75
SAPC (9 wt.% H 2 O)a 1-heptene 75 72
Homogeneous b 1-heptene 75 288 g
c e f
SAPC Mixtureh 100 432 /432 /396 g
e f
Homogeneous d Mixture h 100 1656 /1800 /1800
c e
Biphasic Mixture h 125 17 / 5 f / l g
a b
HRhCO(TPPTS) 3, P=7 bar (H 2/CO=1:1) [16d]. HRhCO(PPh3)3 in toluene, P=7 bar
c d
(H2/CO=1:1) [16b]. HRhCO(TPPTS)3, P=51 bar (H2/CO=1:1) [18]. HRhCO(PPh 3) 3 in
h
hexane, P= 51 bar (H2/CO=1:1) [18]. The substrate was a 1:1:1 mixture of 1-hexene, 1-
e f g
octene and 1-decene. Heptanals. Nonanals. Undecanals.
At higher water contents of the support the water layer becomes too thick
and the substrate has to diffuse into the water layer, or the catalyst has to
diffuse to the interface. The result is a decrease in catalyst-product contact
time leading to lower activities. This sensitivity towards water is a drawback
of this otherwise attractive concept. Horvth performed experiments using
substrates with different solubilities in water and showed that, under optimal
conditions, this solubility did not influence the activity [ 18]. Furthermore, he
performed a hydroformylation reaction in a continuous system and even
under reaction conditions no leaching of rhodium complex was detected. The
water obviously leaches if the SAPC is used in a continuous flow system,
which in a practical application should be compensated for by using water-
saturated organic solvents.
Figure 6. The influence of the water contents of the hydrophilic support on the relative
catalytic activity
A water-soluble chelating diphosphine ligand (12) with a wide natural

bite angle has been prepared [19a]. These ligands are known to form very
selective catalysts for the hydroformylation of 1 -alkenes, yielding mainly
linear aldehyde as the product. These ligands were also studied as supported
aqueous phase catalysts and it was shown that these ligands performed well
262 Chapter 10
as SAPC; they were much more selective than the SAPC known so far [ 19b].
Recycling experiments showed that these catalysts retained their activity and
selectivity in at least ten consecutive runs, whereas under similar conditions
the TPPTS based catalyst showed a reduced performance in the fourth run.
Mortreux compared the activity of the SAPC catalysts with that of the
homogeneous analogue in the hydroformylation of methyl acrylate [20].
They observed an activity for the SAPC that was strongly dependent on the
amount of water present in the system. More remarkably, the optimized
activity of the SAPC was higher than that of the homogeneous systems. This
effect was ascribed to the polar interactions between the substrate and the
silica support. This effect was not observed for nonpolar substrates such as
propene, which supported the hypothesis.
In order to show the versatility of the method Davis extended the concept
to other hydrophilic liquids such as ethylene glycol and glycerol [21]. The
reactions then take place at the hydrophilic-hydrophobic liquid interface. In
this specific example the supported-phase concept was used for the
asymmetric reduction using a ruthenium catalyst. The obtained
enantioselectivity was higher than that of the SAPC, which was ascribed to
the decrease in hydrolysis of chloro-ligand from the ruthenium complex.
Naughton et al. used this supported homogeneous film catalysis concept for
the hydroformylation reaction using TPPTS as the ligand [22]. The low
molecular weight polyethylene glycol resulted in the formation of an
effective film.
10.3.3 Hydroformylation in supercritical fluids
Supercritical fluids, especially supercritical carbon dioxide (scCO2; T c =

3 1C, pc= 73.75 bar, dc = 0.468 g/mL), are receiving considerable interest as
alternative reaction media [23]. scCO2 is a non-toxic, cheap solvent, but
more importantly it has "gas like" properties that may provide advantages
compared to convential solvents. It is highly miscible with reactant gases,
there is no liquid/gas phase boundary, it has a high compressibility, very low
viscosity, and thus high diffusitivity. Several homogeneously catalyzed
reactions have been performed in supercritical fluids. The low solubility of
the catalyst can complicate this approach. Furthermore, the high investment
and operating costs caused by the relatively high pressures required for
supercritical fluids is a serious disadvantage when it comes to commercial

application for low-cost aldehydes.
Rathke was the first to perform a hydroformylation reaction in scCO2 [24].
They studied the hydroformylation of propene using CO2(CO)8 as the
catalyst. They found that the selectivity toward the linear aldehyde is slightly
higher (88%) compared to that found in benzene (83%). Leitner prepared
perfluoralkyl-substituted aryl-phosphines (13-15) in order to obtain scCO2
soluble ligands (figure 7) [25]. Rhodium complexes of these ligands
prepared with [(cod)Rh(hfacac)] were shown to have a much higher
solubility in scCO2 (a factor 150 compared to non-fluorinated complexes).
At higher densities ( = 0.75 mL-1) a complex concentration of 4.4 mM
could be reached. This high solubility allowed the hydroformylation of 1-
octene using rhodium phosphine complexes.
13 14 15
Figure 7. The perfluoroalkyl-substituted aryl phosphines/phosphite prepared for the

hydroformylation in scCO2
Table 4. Hydroformylation of 1-octene in scCO2

c
Ligand [Rh] L/Rh Psyn Ptot TOF Yield 1:b
(mM) (bar) (bar) (mol.mol-1.h-1) (%)
- 0.1 1 - 20 200 122 92 1.6
13 0.13 10 20 200 97 99 4.5
14 0.13 10 20 200 108 93 5.6
b
15 0.13 10 20 200 23 95 8.5
14d 0.66 4 45 210 14 97 3.2
e
14 1.6 6 60 235 11 90 4.6
14d 0.64 13 45 210 2 13 5.5
a
Conditions: [ 1-octene]=280 mmol, Vreactor=225mL, d=0.62 g mL-1, precursor
b
[cod]Rh(hfaca)], T=65 C. The reaction time was 88 h compared to approximately 20 for all
c d e
other cases. The total conversion to the aldehyde. [ 1-octene]=192 mM, Vreactor =25 mL. [ 1-
octene]=384 mM, V reactor = 25 mL.
Initially they studied the activity of solely the catalyst precursor,

[(cod)Rh(hfacac)], towards the hydroformylation of 1 -octene, and found that
it was remarkably active in scCO2 [28]. The activity increased 5-fold
compared to the reaction conducted under similar conditions in toluene.
Moreover, the presence of branched aldehydes and virtually no internal
alkenes indicate that in scCO2 the activity towards internal bonds is high.
264 Chapter 10
Below the critical temperature of CO2, thus in the liquid state, almost no
activity was observed, presumably due to the low solubility of the precursor
in liquid.
On using phosphine ligands the selectivity of the reaction increased, up
to 99% aldehyde was formed. The selectivity for the linear product increased
upon changing the P/Rh ratio from 4 to 10; the linear to branched (1:b) ratio
increased from 3.2 to 5.6. A further increase in P/Rh ratio did not improve
the selectivity further, but a dramatic drop in activity was observed. Similar
effects are observed for reactions performed in organic solvents, indicating
that the reactive species is indeed the phosphine-rhodium complex.
Interestingly, the phosphite ligand 15 also gives a high chemo- and regio-
selectivity, whereas in organic solvents under similar conditions this type of
catalysts gives a lot of isomerized products. -H elimination of the rhodium
alkyl complex results in the internal alkenes, whereas the CO insertion leads
to the aldehyde (see figure 6, chapter 4). The high CO concentration in
scCO2 favors the CO insertion, which results in higher selectivities
compared to the analogous reactions performed in organic media.
The catalyst-product separation was also studied. By changing the
pressure and temperature after the reaction they created a phase separation
from which the aldehydes could be collected as a colorless liquid (containing
less than 1 ppm rhodium). The catalysts of ligands 13-15 were used in 5
consecutive runs without significant changes in the catalytic performance.
The concept can be extended to asymmetric hydroformylation of styrene
in scCO2 using BINAPHOS as the chiral ligand. The initial experiments
using this ligand clearly show that the active catalyst in scCO2 does not
involve the chiral ligand [27]. The low solubility of the ligand was the major
problem, since the catalyst precursor in absence of ligand also gave a high
activity, but of course no enantioselectivity. Therefore a BINAPHOS
analogue was synthesized that contained perfluoroalkyl groups (16) [28].

Using this ligand very high enantioselectivities and regioselectivities were
obtained in the hydroformylation of styrene. These results were independent
of the phase behavior of the reaction mixture; the good performance was
similar using temperatures above and below the Tc of CO2. A slightly higher
regioselectivity was also observed on using 16 compared to BINAPHOS in
the homogeneous phase, which was ascribed partly to the ligand and partly
to the solvent.
Erkey studied the hydroformylation of 1 -octene in scCO2 using several
different fluoralkyl and fluoralkoxy functionalized aryl phosphines [29].
They found an increasing activity of the rhodium catalyst with decreasing
basicity of the phosphine ([3,5-(CF3)2C 6H3] 3P > [4-CF3C6H4]3P and [3-CF3
C6H4]3P > [4-CF3OC6H4]3P). The basicity was measured by the carbonyl
stretching frequencies of the complexes.
The low solubility of the catalyst in scCO2 is not necessarily a problem.
Cole-Hamilton used rhodium trialkylphosphine complexes for the
hydroformylation of 1-hexene in scCO2 [30]. The rhodium concentration
was about 6.5 mM with a P:Rh ratio of 6. The catalyst performed similarly
in scCO2 compared to experiments performed in toluene. The reaction rate
decreased with increasing PEt3, whereas higher Pco or PH2 resulted in an
enhanced reaction rate. The selectivity for the linear product slightly
increased on using scCO2 compared to toluene (1:b = 2.4 and 2.1
respectively).
Although the scCO2 approach is very elegant, one should keep in mind
that it is a relatively expensive method, especially for the bulk chemical
industry. For each recycling step the system has be depressurized and
pressurized again with the newly added substrate. This process requires more
powerful pumps and more capital expensive reactors than for example
biphasic aqueous phase process. Furthermore, the low solubility of potential
interesting substrates might hamper the commercialization of scCO2 in the
fine chemical industry.
10.3.4 Fluorous Biphase catalysis
The fluorous biphasic catalysis (FBC) is another elegant concept of

immobilizing catalysts in a different phase, which shows similarities with the
scCO2 approach. In the early nineties some experiments were already
described in the thesis M. Vogt [31a]. The concept was independently
investigated and reported by Horvth in 1994 [31b]. The system is based on
the immiscibility of fluorinated compounds with organic solvents, which is
due to the polar C-F bond and low polarizability of fluor making this a "non-
interacting" solvent. Ligands functionalized with long fluorinated alkyl
groups, the so-called "pony tails", are soluble in the fluorous phase, whereas
the substrates generally dissolve in the organic phase. Perfluoroaryl groups
are less suitable for this purpose, since they offer dipole-dipole interactions
and thus dissolve much better in organic solvents. The beauty of the concept
266 Chapter 10
stems from the temperature dependency of the miscibility of the solvents

(figure 8). At room temperature the two phases are well separated, but at
higher temperatures the fluorous biphasic system can become a
homogeneous one-phase system. This allows one to combine a
homogeneous reaction at higher temperature with catalyst-product
separations at lower temperatures. The exact solvent behavior obviously
depends on the solvents used [32]. Below the critical temperature phase
separation takes place, but still significant amounts of perfluorosolvent are
dissolved in the organic phase. This requires additional extraction or
distillation. Other major points of concern are catalyst leaching, solvent
costs, and contamination of the product with fluoro compounds.
Figure 8. The general concept of fluorous biphasic catalysis
After the initial report on the concept Horvth performed a detailed

kinetic study of rhodium catalyzed hydroformylation of l-decene in which
the fluorous-soluble ligand P[CH2CH2(CF2)5CF3]3 was compared with the
non- fluorous-soluble ligands P[(CH2)7CH3] and PPh3. Between the fluorous
pony tails and the phosphine donor atom a bridge of two CH 2 groups
insulates the two groups making the former two ligands electronically
identical.
The reactions were performed in 50/50 vol % toluene/C6F11CF3 at 100 C
and 11 bar CO/H2 pressure, thus under these conditions in a homogeneous
phase. For all three catalyst-systems a first order reaction rate in rhodium
concentration was observed. A first order in l-decene was observed for the
systems based on the alkyl phosphines, but for the PPh3 system the kinetics
appeared to be more complicated. The effect of the P/Rh ratio and phosphine
concentration on the selectivity and activity of the catalyst was studied (see
table 5). As observed for homogeneous systems (see chapter 3 and 4), the
P/Rh ratio does not affect the catalytic performance, whereas the absolute
phosphine concentration largely influenced the selectivity and activity of the
catalyst.
Table 5. Effect of the phosphine concentration on the selectivity and activity of the
hydroformylation of 1 -decene a
b
[L] mM P/Rh TOF 1:b
-1 -1
(mol.mol .h )
22.1 27 2160 3.3
22.5 39 1764 3.3
41.3 76 1440 4.4
82.1 79 756 5.0
152.2 103 288 6.3
304.0 102 144 7.8
23 Not given 1908 3.2
c
23 Not given 22572 3.2
d
23 Not given 792 2.3
a
Conditions: 50/50 vol. % toluene/C6F11CF3, T=100 C, pressure=11 bar CO/H2 (1:1),
b c d
initial [I-decene]= 1 M. P[CH2CH2(CF2)5CF3]3 PPh3 P[(CH2)7CH3]3.
At higher concentrations of the ligand a lower activity and higher 1:b ratio
was observed, which is in line with results previously obtained in organic
solvents. The activity of the fluorous pony-tail phosphine ligand was similar
to the alkyl phosphine, but remarkably the selectivity was identical to the
PPh3 system. No explanation was found for this effect. A minor effect of the
fluorous phase on the catalytic performance of the Rh/PPh3 was observed.
The activity was approximately 30 % lower in 50/50 vol % toluene/C6F11CF3
compared to toluene, and the 1:b ratio slightly increased. This effect was
attributed to differences in gas solubilities.
The fluorous biphase catalyst recovery concept was tested by performing
nine consecutive reactions in a batch reactor. A total loss of 4.2% of rhodium
was detected, and the decreasing 1:b ratio suggested some ligand leaching.
The total turnover number reached with the system was 35,000 mole of
aIdehyde/mole rhodium, with a rhodium loss of 1.18 ppm per mole of
aldehyde. Further optimization of this system, i.e. the use of heavier fluorous
solvents and longer pony tails, should decrease the amount of catalyst and
fluorous solvent leaching.
10.3.5 Dendrimer supported catalysts
A recent development in catalysis is the functionalization of dendrimers

with ligands that can give transition metal complexes suitable for catalysis
[33]. It is anticipated that dendritic catalysts can have true advantages over
their low molecular weight parent species [34]. Many times it has been
stated that these types of catalysts are in between homogeneous and
heterogeneous catalysts, combining the benefits of both systems. This should
result in catalysts that are well-defined (precise distribution of catalytic
sites), efficient (fast kinetics due to good accessibility), tunable by ligand
268 Chapter 10
design, and easily separated from the reaction mixture by nanofiltration

(batch wise or continuously). The recent developments of novel dendritic
catalysts suggest that this indeed is a very promising approach towards
efficient recyclable catalysts, but more detailed information about these
novel systems is required.
Figure 9. Phosphine functionalized dendrimers based on the polyamino dendrimers from

DSM (left) and carbosilane dendrimers (right)
Reetz [35] and van Leeuwen [36] reported similar phosphine

functionalized dendrimers, based on different types of dendrimers (figure 9).
Initial experiments in continuous membrane systems using palladium
complexes of these dendrimers as catalysts in the allylic substitution reaction
[35, 36] and hydrovinylation [37] show that the systems based on the higher
dendrimer generations are large enough for efficient separation by
nanofiltration. Both phosphine dendrimers are active in the rhodium-
catalyzed hydro formylation. The preliminary results suggest no significant
differences in catalytic performance compared to the smaller parent
compounds. Alper reported a similar diphosphine functionalized
polyamidoamine (PAMAM) dendrimer that was anchored onto a silica
support [38], a concept that was previously used for phosphite modified
polymers grafted on silica spheres of 12 nM [39]. The dendrimer was
actually synthesized on the support complicating detailed analysis of the
system. The Rh-PPh2-PAMAM-SiO2 catalyst was shown to be an active
catalyst for the hydroformylation reaction and, as expected, more selective
towards the branched product. The difference in chain-length (n = 0, 2, 4)
(see figure 10) had mainly an impact on the recyclability of the catalyst and
it was suggested that this is due to steric congestion of the more compact
dendrimers. Additional Soxhlet extraction experiments showed that rhodium
leached under catalytic conditions; both CO and substrate promote metal
leaching from the support. The phosphite-modified polymers grafted on
silica did not show any rhodium leaching when the hydroformylation was
performed under the proper conditions 139]. A constant conversion was
observed in a continuous flow experiment that was performed for 10 days in
benzene (residence time 3.3h).
Functionalized dendrimers offer new solutions to the catalyst/product
separation problem of homogeneous catalysts and might give catalysts with
unique novel properties. However, if these systems will ever be attractive
alternatives for commercial processes remains to be seen. They have to
compete with other systems consisting of cheaper polymer materials. Metal
leaching and catalyst decomposition are serious problems that should be
addressed. Especially the influence of the local very high concentration of
metal sites in the case of periphery functionalized dendrimers can promote
both these effects. The general problem of metal leaching can be solved by
the proper choice of ligands, as will be clear from an example given in the
next section.
Figure 10. The phosphine functionalized PAMAM dendrimers anchored on a silica support
10.3.6 Novel developments in polymer supported catalyst.
Polymer supported catalysts have attracted a lot of attention already since

the early sixties [40]. In principle, immobilized homogenous catalysts can
combine high activity and selectivity with facile catalyst-product separation,
but to date there are no commercial processes based on this concept. So far,
most systems suffered from either metal leaching, low selectivity and/or low
activity. However, there are some interesting developments in this field that
are very promising, which will be discussed in this section. Also in the light
of the current developments in combinatorial chemistry and rapid screening
techniques these developments are of interest. An effective support for
homogeneous catalysts can be very valuable for rapid synthesis and
270 Chapter 10
screening of novel catalysts. Moreover, the supported catalyst can directly be

used as an immobilized homogeneous catalyst, which is easily separated
from the reaction medium after the reaction by filtration.
The most widely studied polymers for catalyst-immobilization are
organic polymers such as cross-linked polystyrenes [41]. The major
advantages of these type of supports are the relative ease of functionalization
and the wide range of physical properties fine-tuned by the degree of cross-
linking. These polymers, however, suffer from major drawbacks as poor heat
transfer abilities, poor mechanical properties (they are pulverized in the
reactor) and solvent dependent swelling behavior. Inorganic supports such as
silica, alumina, glasses, clays and zeolites do not suffer from these
drawbacks.
An illustrative example of a hydroformylation catalyst immobilized on a
highly cross-linked polystyrene support was reported by Nozaki and Takaya
[42]. Several BINAPHOS ligands carrying vinyl groups have been prepared
and cross-polymerized with ethyl-styrene and divinylbenzene (Figure 1 1).
By using different ligands (17b-17d) the degree of freedom of the ligand in
the polymer matrix was varied. The influence of the degree of cross-linking
and the polymer/catalyst preparation method on the catalyst performance
was investigated (table 6). Generally, the results obtained with the polymer-
supported catalysts are similar to the homogeneous system (run 1).
Cross linked polymer

Initiator
Toluene/ 80C
17a R1=H, R2 =H
1
17b R =vinyl, R2=H
1
17c R =H,R2=vinyl
17d R1=vinyl, R2=vinyl
Figure 1I. Chiral phosphine-phosphite ligands on a highly cross-linked polystyrene polymer
Two methods of catalyst-preparation were tested; the co-polymerization

of the ligand and subsequent formation of the rhodium acetylacetonate
complex (run 2-7) and co-polymerization of the rhodium acac complex of
the ligand (run 8-11). This generally gave the same results, except for ligand
17d (cf. run 7 and 10). This ligand is frozen in the polymer matrix in a
different conformation, since it is co-polymerized at three positions of the
ligand. Some of these conformations are not able to form the rhodium
complex that gives the enantioselective hydroformylation catalyst. In
contrast, the other ligands have more freedom and can therefore form the
proper catalyst after the polymerization. The degree of cross-linking has no
significant influence on the selectivity of the reaction (runs 2, 4, 5, and 11).
The reaction can be performed in hexane, but the ee was slightly lower. The
recycling experiments clearly show the drawback of these types of polymers.
The polymers were partly crushed by the stirring bar and approximately 50%
of the initially charged rhodium was removed after the first run.
Table 6. Asymmetric hydroformylation of styrene using polystyrene supported rhodium

a
catalysts based on (R,S)-BINAPHOS)
Run Catalyst 1:b ee (%)

1 Rh(acac)( 17a) 0.12 92
2 (PS- 1 7b)-Rh(acac) 0.19 89
3b (PS-17d)-Rh(acac) 0.25 81
4c (PS-17d)-Rh(acac) 0.20 89
5d (PS- 17b)-Rh( acac) 0.23 84
6 (PS-17c)-Rh(acac) 0.12 89
7 (PS-17d)-Rh(acac) 0.14 68
8 (PS)-[Rh(acac)( 17b)] 0.18 90
9 (PS)-[Rh(acac)(17c)] 0.11 97
10 (PS)-[ Rh(acac)( 17d)] 0.15 85
11d (PS)-[Rh(acac)(17b)] 0.18 90
a
Conditions: 6.20 mmol styrene in benzene, 0.0031 mmol catalyst, 20 bar CO/H2 (1:1),
b c
60 C, 12 hours, conversions to aldehydes >99%. Hexane as solvent. Lower cross-linking
d
degree. Higher cross-linking degree.
The covalent anchoring of hydroformylation catalysts onto inorganic

silica surfaces was already studied in the late seventies [43]. Chloromethyl
silicone polymers were converted to the iodo-analog and subsequently
reacted to the diphenylphosphinemethyl polysiloxanes. After treatment with
RhCl(CO)(PPh3)2 these materials were used as catalyst in the
hydroformylation of 1-hexene (68 bar CO/H2 (1:1), 100 C, benzene). Under
these conditions, however, the metal dissociated from the support. The
resulting rhodium carbonyl species was responsible for the activity leading
to low selectivities.
Pantser pioneered the field of organo-functionalized silanes and
polysiloxanes. 3-Chloropropyltrialkoxysilanes were converted into
functionalized propyltrialkoxysilanes such as diphenylphosphine
propyltrialkoxysilane [44] . These compounds can easily be used to prepare
surface modified inorganic materials. Upon polycondensation or co-
272 Chapter 10
condensation of these compounds novel functional polymeric materials were

obtained that allowed transition metal complexation. This co-condensation,
later also referred to as sol-gel process [45], appeared to be a very mild
technique to immobilize catalysts. Two different routes towards these
functional polymers can be envisioned (Figure 12). One can first prepare the
metal complex and then do the co-condensation reaction (route I), or one can
prepare the metal complex after the polymerization step (route II). The
former route results in far better catalysts due to the higher structural
homogeneity. Also more stable catalysts are obtained showing less metal
leaching. Lindner performed a detailed study on the sol-gel processed ether-
phosphines and their ruthenium complexes using solid state NMR
spectroscopy [46]. The obtained different materials strongly influenced the
catalytic performance in the ruthenium catalyzed hydrogenation.
Figure 12. Two different routes to prepare sol-gel immobilized transition metal complexes
Pantser studied the hydroformylation of 1 -octene using a polysiloxane-

based rhodium phosphine-amine based catalyst in a trickle-bed reactor [47].
The selectivity for the linear product was low (1:b =1.5), but the catalytic
performance was retained over a period of 1000 h, and almost no rhodium
leaching was detected. Contrary to the homogenous phase reaction, the
usage of an excess of phosphine ligands did not result in larger amounts of
the linear product, which was attributed to the rigid structure of the
polymeric material.
Diphosphine ligands with wide natural bite angles based on the Xanthene
backbone have been functionalized with a propyltrialkoxysilane group (18)
[48]. Silica immobilized rhodium catalysts for the hydroformylation of 1-
octene were prepared via the sol-gel method. The catalyst system obtained is
very selectively producing the linear product and on recycling no
deterioration of the catalyst performance is observed for at least 8 cycles.
Furthermore, the reaction could be performed in pure substrate giving
reaction rates that are comparable with the homogenous system (80 C, 50
bar CO/H2 (1 : 1), rhodium:ligand=1 :10).
These silica immobilized systems prepared via the sol-gel method are
very promising organic-inorganic hybrid materials. The chemistry takes
place at the interface of the materials suppressing the detrimental influence
of the support [46c]. A series of network modifiers for the sol-gel process
are available, which can be used to optimize the support. This tool to further
improve the catalyst performance already proved to be valuable for several
reactions [46c]. For the hydroformylation reaction this still needs to be
explored.
18
Bergbreiter introduced an attractive novel concept by immobilizing

catalysts on "smart polymers"; these polymers show temperature dependent
solubility properties [49]. Initially these novel systems were based on
polyethylene polymers. At room temperature these polymers are completely
insoluble in all solvents but they are soluble in several solvents at elevated
temperature. Functionalized oligomers (M n = 2000) also exhibited this type
of behavior. Systems have been obtained that were catalytically active at
higher temperatures with similar activity as the low molecular weight parent
compounds and completely inactive and recyclable at room temperature. The
concept was extended using block copolymers (PEO-PPO-PEO) and poly(N-
isopropyl acrylamide) (PNIPAM) polymers. These materials phase separated
at elevated temperatures and formed a homogeneous solution at lower
temperatures. The concept remains the same, but now the catalysis takes
place at room temperature and the recycling can be achieved at higher
temperatures. Moreover, these materials are potentially useful in controlling
exothermic reactions.
Similar to the fluorous biphasic concept, a system was developed that
formed a homogenous phase at elevated temperature, but phase-separated at
room temperature [49c]. It was found that a mixture of functionalized
PNIPAM polymer, ethanol, heptane and water exhibited these properties.
The hydrogenation of 1 -octadecene and 1 -dodecene using a phosphine
functionalized PNIPAM with a rhodium precursor were taken as test
reactions and the high activity was found was similar to that of RhCl(PPh3)3.
At room temperature the mixture phase separated and the catalysis stopped
since the catalyst is completely insoluble in heptane. The substrate is
dissolved in the heptane allowing a facile catalyst/product separation without
the loss of activity. The concept is obviously limited to substrates that show
274 Chapter 10
a strong preference for the heptane layer. So far, none of the smart polymer
concepts has been tested for the hydroformylation reaction. However, the
current rapid developments in polymer technology and the already described
interesting systems makes the usage of smart functionalized polymers a
promising development.
10.4 Supramolecular catalysis
Supramolecular chemistry has been a very popular research topic for

three decades now. Most applications are foreseen in sensors and opto-
electronical devices. Supramolecular catalysis often refers to the
combination of a catalyst with a synthetic receptor molecule that pre-
organizes the substrate-catalyst complex and has also been proposed as an
important possible application. The concept, which has proven to be
powerful in enzymes, has mainly been demonstrated by chemists that
investigated hydrolysis reactions. Zinc and copper in combination with
cyclodextrins as the receptor dramatically enhance the rate of hydrolysis. So
far, the ample research devoted to transition metal catalysis has not been
extended to supramolecular transition metal catalysis. A rare example of
such a supramolecular transition metal catalyst was the results of the joined
efforts of the groups of Nolte and Van Leeuwen [SO]. They reported a
basket-shaped molecule functionalized with a catalytically active rhodium
complex that catalyzed hydrogenation reactions according to the principles
of enzymes. The system showed substrate selectivity, Michaelis Menten
kinetics and rate enhancement by cooperative binding of substrate
molecules. The hydroformylation of allyl catachol substrates resulted in a
complex mixture of products.
Reetz et al. developed water-soluble supramolecular transition metal
catalysts that are based on functionalized -cyclodextrins [5 1]. The b-
cyclodextrin is a frequently used building block that is soluble in the
aqueous phase and contains a hydrophobic cavity. Binding of organic
substrates takes place in this cavity and is driven by hydrophobic
interactions. Initial studies were focussed on selective hydrogenation and
optimization of the spacer length between the catalyst and the cavity. The
optimal supramolecular system (Figure 13) has been used to study
hydroformylation reactions. Several striking differences in catalytic
performance were observed for this supramolecular catalyst compared to the
parent compound. The two-phase aqueous hydroformylation of 1 -octene was
very efficient; at 80 C and 100 bar syn-gas pressure complete conversion
was achieved in less than 18 h (TON=3172). Moreover, a complete
chemoselectivity toward the aldehyde (>99%) was observed (no
isomerization). The activity of this system was higher than the aqueous
catalyst system based on TPPTS, even in the presence of phase transfer

catalysts. Also internal alkenes such as 3-octene, cyclic alkenes and
conjugated systems as styrene and 4-methyl- 1,3-pentadiene could be
hydroformylated.
Figure 13. The functionalized cyclodextrin building block as supramolecular catalyst for two-
phase catalysis. The actual catalysis is proposed to take place at the phase boundary
Interestingly, the selectivity for the linear product increased on using the
supramolecular system; the 1:b ratio was 3.2 compared to 1.5 for the parent
catalyst. A better test for the selectivity of this supramolecular catalyst is
allylbenzene, a substrate that easily isomerizes to methyl styrene. Again the
hydroformylation reaction was very selective and a higher 1:b ration was
observed on using the supramolecular system. The presence of an excess of
toluene, a guest molecule that competes for the cavity with the substrate,
resulted in a lower regioselectivity. This indicates that the formation of the
supramolecular complex is responsible for the higher selectivity.
19
Initial experiments aiming at recycling of the catalysts showed that the
supramolecular system was mainly in the water layer after separation from
the organic phase. Reuse of the aqueous phase revealed that 50% of the
catalytic activity was retained. Prior the work of Reetz several groups
studied the effect of cyclodextrins as inverse phase transfer materials on the
276 Chapter 10
hydroformylation reaction [52]. The general idea is similar to that described

above, but now the catalyst and the host molecule are not covalently linked
to each other. The cyclodextrin transfers the 1-decene from the organic to
aqueous phase where the hydroformylation takes place. Several
cyclodextrins (19) have been used (see Table 7).
Table 7. Hydroformylation of dec-1 ene in the presence of chemically modified cyclodextrins

a
[52b].
Type (R) a b TOF 1:b Aldehyde
(%)
- 6 2.7 60
(-) 0 18 6 3.2 85
(-) 0 24 6 2.5 66
(-) 0 21 12 2.1 78
(Me) 13 8 47 1.8 91
b
(Me) 13 8 83 1.9 95
(Me) 21 0 19 2.5 57
(COMe) 21 0 4 2.6 66
(COMe) 14 7 29 2.6 57
(CH2CHOHCH3) 6 15 20 2.0 84
(SO3Na) 9 12 4 2.8 69
a
Reaction conditions: Rh(aca)(CO)2 0.16 mmol, TPPTS 0.8 mmol, cyclodextrin 1.12
mmol, H2O 45 mL, dec-1-ene 80 mmol, undecane 4mmol, T=80C, pressure=50 bar
b
(CO/H2=1:1), t =8h. a is the number of R groups and b is number of free OH groups. 2.24
mmol cyclodextrin was added, full conversion after 6h.
The effect of the addition of unmodified cyclodextrins to a reaction

mixture on the aqueous hydroformylation of 1-decene is rather small. The
conversion was enhanced by a factor 2 upon the addition of -cyclodextrin.
The modified -cyclodextrins improved the activity to a greater extent; in
the optimized situation the activity increased with a factor of 14. These
modified cyclodextrins are more efficient since they are soluble in both the
organic phase and the aqueous phase thereby improving the efficiency of the
substrate transfer process. Similar to the systems described above, the
isomerization was suppressed under these optimized conditions. However,
internal alkenes could not be hydroformylated using this system, Another
striking difference is the selectivity of the reaction. The addition of the
modified cyclodextrins resulted in a decrease of the 1:b ratio (2.7 to 1.9),
whereas Reetz reported an increase for his supramolecular system! So far no
solid explanation has been found for this difference.
These types of supramolecular systems are expected to be too expensive
for commercial applications as yet. The prices of cyclodextrins, however, are
in the range of organic solvents and therefore this is not a limitation for
commercialization. Furthermore, cyclodextrins are lion-toxic, biodegradable
and available in large quantities, which makes them suitable for commercial
applications.
10.5 Conclusions
The above examples have shown that there is considerable activity in

searching for other ways to achieve separation of product and homogeneous
catalysts. The improvements needed are the same for all reaction types, also
reactions other than hydroformylation. Most of the new techniques have in
common that not only the product, but also the starting material and
byproducts are separated from the catalyst. This mixture of organic products
needs further separation, but as we have seen in Chapter 8 separation of
catalyst and byproducts such as heavy ends is an important issue in
hydroformylation of simple alkenes. There may be a future for immobilized
drop in catalysts as described above in the hydroformylation of fine
chemicals.
References
1 Kuntz, E. D. CHEMTECH, 1987, 17(9), 570.

2 (a) Reedijk, J., Ed.; Bioinorganic Catalysis, Dekker, New York, 1993. (b) Bertini, I.;
Gray, H. B.; Lippard, S. J.; Selverstone Valentine, J., Eds; Bioinorganic Chemistry,
University science books: Mill Valley, 1994.
3 van der Beuken, E.; Feringa, B. F. Tetrahedron, 1998, 54, 12985.
4 Brousard, M. E.; Juma, B.; Train, S. G.; Peng, W, -J.; Laneman, S. A.; Stanley, G. G.
Science, 1993, 260, 1784.
5 (a) Breslow, D. S.; Heck, R. F. Chem. Ind. 1960, 467. (b) Breslow, D. S.; Heck, R. F. J.
Am. Chem. Soc. 1961, 83, 4024.
6 (a) Matthews, R. C.; Howell, D. K.; Peng, W.-P.; Train, S.G.; Dale Treleaven, W.;
Stanley, G. G. Angew. Chem. Int. Ed. Engl. 1996, 35, 2253. (b) Stanley, G. G. in
Catalysis by di- and polynuclear metal cluster complexes, Chapter 10, Adams, R. D.;
Cotton, F. A. Eds. Wiley-VCH, 1998, New York.
7 (a) Hablot, I.; Jenck, J.; Casamatta, G.; Delmas, H. Chem. Eng. Sci. 1992, 47, 2689. (b)
Monteil, F.; Queau, R.; Kalck, P. J. Organomet. Chem. 1994, 480, 177.
8 Chaudhari, R. V.; Bhanage, B. M.; Desphpande, R. M.; Delmas, H. Nature. 1995, 373,
501.
9 (a) Tascioglu, S. Tetrahedron, 1996, 52, 11 113. (b) Christian, S. D.; Scamchorn, J. F.,
Eds. Solubilization in Surfactant aggregates, Marcel Dekker, New York, 1995. (c)
Fendler, J. H.; Fendler, E. J. Catalysis in Micellar and Macromolecular Systems,
Academic press, New York, 1975.
10 For example: Otto, S.; Engberts, J. B. F. N.; Kwak, J. C. T. J. Am. Chem. Soc. 1998, 120,
9517.
11 (a) Fell, B.; Papadogianakis, G. J. Mol. Catal. 1991, 66, 143. (b) Fell, B.;
Papadogianakis, G. J. Mol. Catal. A: Chem. 1995, 101, 179.
278 Chapter 10
12 (a) Ding, H.; Hanson, B. E.; Bartik, T.; Bartik, B. Organometallics, 1994, 13, 3761. (b)
Ding, H.; Hanson, B. E. J. Chem. SOC. Chem. Commun. 1994, 13, 2747. (e) Bartik, T.;
Ding, H.; Bartik, B.; Hanson, B. E. J. Mol. Catal. A; Chem. 1995, 98, 117.
13 (a) Ding, H.; Hanson, B. E.; Bakos, J. Angew. Chem. 1995, 107, 1728. (b) Hanson, B. E.;
Ding, H.; Kohlpainter, C. W. Catal. Today 1998, 42, 421.
14 Schreuder Goedheijt M.; Hanson, B. E., Reek, J. N. H.; Kamer, P. C. J.; van Leeuwen, P.
W. N. M. J. Am. Chem. Soc. 2000, 122, 1650.
15 Arhancet, J. P.; Davis, D. E., Merola, J. S.; Hanson, B. E. Nature, 1989, 339, 454.
16 (a) Davis, M. E. CHEMTECH, 1992, August, 498. (b) Archancet, J. P.; Davis, M. E.;
hanson, B. E. J .Catal. 1991, 129, 110. (c) Guo, I.; Hanson, B. E.; Tth, I.; Davis, M. E.
J. Organomet. Chem. 1991, 403, 221. (d) Arhancet, J. P.; Davis, M. E.; Hanson, B. E.
J, Catal. 1991, 129, 94. (e) Yuan, Y.; Xu, J.; Zhang, H.; Tsai, K. Catal. Lett. 1994, 29,
387. (f) Fremy, G.; Monfier, E.; Carpentier, J.-F.; Castanet, Y.; Mortreux, A. J. Catal.
1996, 162, 339.
17 Bunn, B. B.; Bartik, T.; Bartik, B.; Bebout, W. R.; Glass, T. E.; Hanson, B. E. J. Mol.
Catal. 1994, 94, 157.
18 Horvth, I. T. Catal. Lett. 1990, 6, 43.
19 (a) Schreuder Goedheijt, M.; Kamer, P. C. J.; van Leeuwen, P. W. N. M. J. Mol. Cat.
1998, 134, 243. (b) Sandee, A. J.; Slagt, V. F.; Reek, J. N. H.; Kamer, P. C. J.; van
Leeuwen, P. W. N. M. Chem. Commun. 1999, 1633.
20 Fremy, G.; Monflier, E.; Carpentier, J.-F.; Castanet, Y.; Mortreux, A.; Angew. Chem. Int.
Ed. Engl. 1995, 34, 1474.
21 (a) Wan, K. T.; Davis, D. E. Nature, 1994, 370, 449. (b) Wan, K. T.; Davis, D. E. J.
Catal. 1994, 148, 1.
22 Naughton, M. J.; Drago, R. S. J. Catal. 1995, 155, 383.
23 Jessop, P. G.; Ikariya, T.; Noyori, R. Chem. Rev. 1999, 99, 475.
24 Rathke, J. W.; Klingler, R. J.; Krause, T. R. Organometallics, 1991, 10 1350.
25 Kainz, S.; Koch, D.; Baumann, W.; Leitner, W. Angew. Chem. Int. Ed. Engl. 1997, 36,
1628.
26 Koch, D.; Leitner, W. J. Am. Chem. SOC. 1998, 120, 13398.
27 Kainz, S.; Leitner, W. Catal. Let. 1998, 55, 223.
28 Francio, G.; Leitner, W. Chem. Commun., 1999, 1663.
29 (a) Palo, D. R.; Erkey, C. Ind. Eng. Chem. Res. 1999, 38, 3786. (b) Palo, D. R.; Erkey, C.
30 Bach, I.; Cole-Hamilton, D. J. Chem. Commun. 1998, 1463.
31 (a) Vogt, M. PhD. Thesis, 1991, Rheinisch-Westfalischen Technischen Hochschule,
Aachen, Germany. (b) Horvth, I. T.; Rbai, J. Science, 1994, 266, 72.
32 (a) Grosse, A. V.; Cady, G. H. Ind. Eng. Chem. 1947, 39, 367. (b) Scott, R. L.; J. Am.
Chem. Soc. 1948, 70, 4090. (c) Hildebrand, J. H.; Cochran, D. R. F. J. Am. Chem. Soc.
1949, 71, 22. (d) Dorset, D. L. Macromolecules 1990, 23, 894. (d) Scott, R. J. J. Phys.
Chem. 1958, 62, 136.
33 For reviews on dendrimers see: (a) Tomalia, D. A.; Durst, H. D. Top. Curr. Chem., 1993,
165, 193; (b) Issberner J.; Moors, R.; Vgtle, F. Angew. Chem. 1994, 106, 2507; (c)
Frchet, J. M. J. Science, 1994, 263, 1710; (d) Newkome, G. R.; Moorefield, C. N.;
Vgtle, F., Dendritic Molecules-Concepts, Synthesis, Perspectives, WCH, Weinheim
1996. (e) Majoral, J. P.; Caminade, A. Chem. Rev., 1999, 99, 845. (f) Bosman, A. W.;
Janssen, H. M.; Meijer, E. W. Chem. Rev. 1999, 99, 1665.
34 (a) Knapen, J. W. J.; van der Made, A. W.; de Wilde, J. C.; van Leeuwen, P. W. N. M.;
Wijkens, P.; Grove, D. M.; van Koten, G. Nature, 1994, 372, 659. (b) Tomalia, D. A.;
Dvomic, P. R. Nature, 1994, 372, 617.
35 (a) Reetz, M. T.; Lohmer, G.; Schwickardi, R. Angew. Chem. Int. Ed. Engl. 1997, 36,
1526. (b) Brinkman, N.; Giebel, D.; Lohmer, G.; Reetz, M. T.; Kragl, U. J. Catal. 1999,
183, 163.
36 de Groot, D.; Eggeling, E. B.; de Wilde, J. C.; Kooijman, H.; van Haaren, R. J.; van der
Made, A. W.; Spek, A. L.; Vogt, D.; Reek, J. N. H.; Kamer, P. C. J.; van Leeuwen, P. W.
N. M. Chem. Commun. 1999, 1623.
37 Hovestad, N. J.; Eggeling, E. B.; Heidbchel, H. J.; Jastrzebski, J. T. B. H.; Kragl, U.;
Keim, W.; Vogt, D.; van Koten, G. Angew. Chem. Int. Ed. Engl. 1999, 38, 1655.
38 (a) Bourque, S. C.; Maltais, F.; Xiao, W. -J.; Tardif, O.; Alper, H.; Arya, P.; Manzer, L.
E. J .Am. Chem. SOC. 1999, 121, 3035. (b) Bourque, S. C.; Alper, H.; J. Am. Chem. Soc.
2000, 122, 956. (c) Arya, P.; Rao, N. V.; Singkhonrat, J.; Alper, H.; Bourque, S. C.;
Manzer, L. E. J. Org. Chem. 2000, 65, 1881.
39 Jongsma, T.; van Aert, H.; Fossen, M.; Challa, G.; van Leeuwen, P. W. N. M. J. Mol.
Catal. 1993, 83, 37.
40 Hartley, F. R.. Supported metal complexes, a new generation catalysts, Reidel,
Dordrecht, 1985.
41 (a) Pomogailo, A. D.; Whrle, D. in Macromolecule-Metal Complexes, Ciardelli, F.;
Tsuchida, E.; Whrle, D. Eds., Springer, Berlin, 1996. (b) Keim, W.; Driessen-Hlscher,
B. in Handbook of Heterogenous Catalysis, vol 1, Ertl, G.; Knzinger, H.; Weitkamp, J.
Eds. WILEY-VCH, Weinheim, 1997.
42 (a) Nozaki, K.; Itoi, Y.; Shibahara, F.; Shirakawa, E.; Ohta, T.; Takaya, H.; Hiyama, T.
J. Am. Chem. Soc. 1998, 120, 4051. (b) Nozaki, K.; Shibahara, F.; Itoi, Y.; Shirakawa,
E.; Ohta, T.; Takaya, H.; Hiyama, T. Bull. Chem. Soc. Jpn. 1999, 72, 191 1.
43 Farrell, M. O.; van Dyke, C. H.; Boucher, L. J.; Metlin, S. J. J. Organomet. Chem. 1979,
172, 367.
44 Deschler, U.; Kleinschmit, P.; Panster, P.; Angew. Chem. Int. Ed. Engl. 1986, 25, 236.
45 (a) Monaco, S. J.; Ko, E. I. CHEMTECH, 1998, 23. (b) Ingersoll, C. M.: Bright, F. V.
CHEMTECH, 1997, 26.
46 (a) Lindner, E.; Kemmler, M.; Mayer, H. A.; Wegner, P. J. Am. Chem. Soc. 1994, 116,
348. (b) Lindner, E.; Schneller, T.; Auer, F.; Wegner, P.; Mayer, H. A. Chem. Eur. J.
1997, 3, 1833. (c) Lindner, E.; Schneller, Auer, F.; Mayer, H. A. Angew. Chem. Int. Ed.
Engl. 1999, 38, 2155.
47 Wieland, S.; Panster, P. Catal. Org. React. 1994, 62, 383.
48 Sandee, A. J.; van der Veen, L. A.; Reek. J. N. H.; Kamer, P. C. J.; Lutz, M.; Spek, A.
L.; van Leeuwen, P. W. N. M. Angew. Chem. lnt. Ed. Engl. 1999, 38, 3231.
49 (a) Bergbreiter, D. E.; Catal. Today, 1998, 42, 389. (b) Bergbreiter, D. E.; Zhang, L.;
Mariagnanam, V. M. J. Am. Chem. SOC. 1993, 115, 9295. (c) Bergbreiter, D. E.; Liu, Y.-
S.; Osburn, P. L. J. Am. Chem. Soc. 1998, 120, 4250.
50 (a) Coolen, H. K. A. C.; van Leeuwen, P. W. N. M.; Nolte, R. J. M. Angew. Chem. Int.
Ed. Engl. 1992 ,31, 906. (b) Coolen, H. K. A. C.; Meeuwis, J. A. M.; van Leeuwen, P.
W.N. M.;Nolte,R. J. M. J. Am. Chem. Soc. 1995, 117, 11906.
51 Reetz, M. T.; Waldvogel, S. R. Angew. Chem. Int. Ed. Engl. 1997, 36, 865. (b) Reetz, M.
T. J. Heterocyclic Chem. 1998, 35, 1056. (c) Reetz, M. T. Catal. Today, 1998, 42, 399.
52 For hydroformylation reactions with cyclodextrins as additives: (a) Anderson, J. R.;
Campi, E. A.; Jackson, W. R. Catal. Lett. 1991, 9, 55. (b) Monflier, E.; Fremy, G.;
Castanet, Y.; Mortreux, A. Angew. Chem. Int. Ed. Engl. 1995, 34, 2269. (c) Monflier, E.;
Tilloy, S.; Fremy, G.; Castanet, Y.; Mortreux, A. Tetrahedron Lett. 1995, 36, 9481. (d)
Tilloy, S.; Bertoux, F.; Mortreux, A.; Monflier, E. Catal. Today, 1999, 48, 245.
281
Index
Index Terms Links
Acetalization 165 166

acrolein acetals 149
acrylonitrile 153
activity 207
aldol condensation 207 212 223
aldol reaction 157 177
alkylphosphines 67
alkynes 178
allyl alcohol 217
amination 168
amphiphilic ligands 256 259
aqueous two-phase catalysis 256
Arbuzov rearrangement 44 245
asymmetric hydroformylation 107-144
atropisomers 109
BDPP 134
Berry mechanism 114
bimetallic catalysts 253
BINAP 3 4 134
BINAPHOS 4 7 108 124 264 270
binaphthyl 125
BINAS 193
biphasic hydroformylation 135 136
BIPHEPHOS 147 170 178
BIPHOLOPHOS 136
BISBI ligands 7 82 83 84 85 86
87 136 147 258
BISBIS 192
bisphenol bridges 45
bite angle 5 8 10 82-99 135
BPPM 132
bridge length 45
catalyst decomposition 235 236 237 238 239 240
241 242 243 244 245 246
247 248 249
bulky diphosphite catalysts 45 108
bulky phosphite 6 7 37 38 39 40
41 42 43 44 151 158
1,4-butanediol 217 225
2-butene 46
Calix[4]arene based phosphites 59

catalyst cost 204
This page has been reformatted by Knovel to provide easier navigation.

282
Index Terms Links
catalyst preparation 17 195 233 234

catalyst recovery 210 211 212 213 214 215
216
catalyst separation 256
chelation control 134
chiral cooperativity 116
chiral diphosphines 131
chiral diphosphites 109
CHIRAPHOS 134
CO dissociation 74 93 100 101 102
cone angle 8 9 40
continuous flow system 261 269
cooperative effect 131
cosolvents 200
cyclodextrins 275
DEGUPHOS 134
dendrimer supported catalysts 268
detergent alcohols 223
detergents 201
deuterioformylation 16 24 25 26 27 30
31 33 65 84
dihydrofurans 150
dihydropyrans 151
dimer formation rhodium complexes 54 64 72 247
dimethoxyacrolein 153
dimethylbut-1-ene 18 25
dinuclear species 137
DIOCOL 133
DIOP 2 78 82 84 133
dioxaphosphepine 120
DIPAMP 2 4
diphosphines as ligands 76-102 131 132 133 134 135
136 137 138 139 140
diphosphites as ligands 44-59 108-120
dirhodium species 54 65 69 137
dissociation of CO 74 93 100 101 102
distillative separation 214
1,1-disubstituted alkenes 149 164
dormant sites 247
double-bond isomerization 207
DPBS 219
DPPB, dppb 217
DuPhos 3 4
Economics 198 204

electron withdrawing substituents 38 85
electronic effects 67 79 85 86 87 89

283
Index Terms Links
electronic parameter 8
electron-withdrawing phosphines 79
enantiofacial selection 128
energetics 100 101 102
ethyl acrylate 139 154
evaporation 211 212 213 214
extraction 211 215 216 217 218 219
220 225
extraction after one-phase reaction 216
Ferrocene diphosphine ligands 79

fluorous biphase catalysis 190 265
fluxional behavior 114
Gas recycle process 213

glucal 158 167
glucopyranose 123
Heavy -ends 207

Heck mechanism 64
heteroatom-directed
hydroformylation 160
-hydride elimination 41 80 101 102
hydroaminomethylation 172 177
hydrocyanation 1 121
hydrolysis phosphites 44
Immobilization 129 201

immobilized aqueous phase 260
in situ IR reflectance spectroscopy 68
in situ IR transmission spectroscopy 28 68 69 90 138
in situ NMR spectroscopy 49 50 51 52 68 69
80 91
indicator ligand 39
internal alkenes 55 56 57 58 59 95
263 275
IR spectroscopic studies 43
Josiphos 4
Kinetic studies 43 47 69 70 71 72
97 98 99 100 101 102
112
kinetics two-phase catalysis 193 194 195
kinetics, fluorous phase 266

284
Index Terms Links
Ligand effects 1 2 3 4 5 6
7 8 9 10 11 12
13 14 38 66 67 68
76-99
ligand loss 209 235 236 237 238 239
240 241 242 243 244 245
246
liquid recycle process 213
long chain alkenes 200
Manufacturing cost 204

mechanistic studies 22 40 41 42 43 44
48 49 50 51 52 53
54 55 68 69 70 71
72 96 97 98 99 113
114 115 116 117 118 119
120 124 138
metal plating 235
metalation 240 248
methyl methacrylate 5
methyl oleate 55 56 57
3-methylpentane-1,5-diol 39
micellar catalysis 256
monophosphines as ligands 63 64 65 66 67 68
69 70 71 72 73 74
75
monophosphites as ligands 37 38 39 40 41 42
43 44 121
muscone 149
Nanofiltration 268
NAPHOS 137 193
natural bite angle 82 88 96 97 98 99
NMP 220
NMR spectroscopy 49 50 51 52 68 69
80 91
nylon monomers 226
Organic synthesis, hydroformylation

in 145
orthometallation 54 55 210 240
oxidation of phosphorus ligands 209 235 236 237

285
Index Terms Links
P-C bond cleavage 209 237 238 239 240 241

3-pentenoate esters 58 227
pH aqueous phase catalysis 195
phase transfer 275
phase-transfer catalyst 225
phosphine ligands 63-100
phosphine-phosphite ligands 129 130 131
phosphites 35-61 109 -124

phospholes 67 82
phosphonium intermediates 242
phosphonium salt 210
phosphorus ligand effects 1
phyllantocin 162
platinum systems 107 132
P-O bond splitting 210
polyketone 4
polymer bound bulky phosphite 40
polymer supported catalysts 269 270 271 272 273 274
process aspects 196 197 198 203- 227

propanal 200
pyrrole 174
Quaternary phosphonium salts 210
Raffinate-II 197 220

regioselectivity 72 73 74 75 96 97
98 99
Reppe chemistry 1
resting state 40 41 42 43 69 70
71 72 74 81 96 113
138
rhodium containment 208
rhodium leaching 208
rhodium plating 208
rhodium price 204
ribose 122
Ruhrchemie Rhne-Poulenc-
Process 191 196
Silica immobilized systems 273

silylenol ethers 176
silylformylation 178 181
smart polymers 273

286
Index Terms Links
sol-gel process 271

solution structures 49 50 51 52 68 69
80 91 111 113 137
standard conditions 70
steric effects 38 40 46 68 113 114
115 116 117 118 119 120
stripping reactor 212
styrene 18 20 21 26 30 107- 142

substituted alkenes 55 56 57 58 59
substrate-directed stereoselectivity 155
sugar derivatives 121
sulfonated ligands 256 257 258 259 260 261
262
supercritical fluids 262 263
supported aqueous phase catalysis 260
supramolecular catalysis 274
Terpenes 167
Tischenko reaction 207
tpp 6 63 64 65 66 67
68 69 70 71 72 73
74 75 212 213 221
tppms 5 7 221
tppts 5 6 7 191 215 221
TPPTS, see tppts
transition state geometry 98
trifluoropropene 18 153
triphenyl phosphite 38
triphenylphosphine, see tpp
tris(ortho tert-butylphenyl)phosphite 37 38 39 40 41 42
43 44 221
tris(2,2,2-trifluoroethyl) phosphite 38
tum-stile mechanism 53 114 138
two-phase catalysis 190
two-phase reaction 215
Unmodified rhodium catalysts 15 16 17 18 19 20

21 22 23 24 25 26
27 28 29 30 31 32
33
Vinyl acetate 18 20
vinyl arenes 152
vinyl aromatics 107
vinyl esters 150
vinylpyridine 18 153

287
Index Terms Links
vinylpyrrole 153
Wilkinson 64 72
Wittig reaction 177
Xantphos ligands 87 88 89 90 91 92
93 94 95 147 201 259
262 273
xylofuranose 122

Series Editors:
R. Ugo, University of Milan, Milan, Italy
B. R. James, University of British Columbia, Vancouver, Canada
1*. F. J. McQuillin: Homogeneous Hydrogenation in Organic Chemistry. 1976

ISBN 90-277-0646-8
2. P. M. Henry: Palladium Catalyzed Oxidation ofHydrocarbons. 1980
ISBN 90-277-0986-6
3. R. A. Sheldon: Chemicals from Synthesis Gas. Catalytic Reactions of CO and
H2. 1983 ISBN 90-277-1489-4
4. W. Keim (ed.): Catalysis in C1 Chemistry. 1983 ISBN 90-277-1527-0
5. A. E. Shilov: Activation of Saturated Hydrocarbons by Transition Metal Com-
plexes. 1984 ISBN 90-277-1628-5
6. F. R. Hartley: Supported Metal Complexes. A New Generation of Catalysts. 1985
ISBN 90-277-1855-5
7. Y. Iwasawa (ed.): Tailored Metal Catalysts. 1986 ISBN 90-277-1866-0
8. R. S. Dickson: Homogeneous Catalysis with Compounds of Rhodium and lridium.
1985 ISBN 90-277-1880-6
9. G. Strukul (ed.): Catalytic Oxidations with Hydrogen Peroxide as Oxidant. 1993
ISBN 0-7923-1771-8
10. A. Mortreux and F. Petit (eds.): Industrial Applications of Homogeneous Cata-
lysis. 1988 ISBN 90-277-2520-9
11. N. Farrell: Transition Metal Complexes as Drugs and Chemotherapeutic Agents.
1989 ISBN 90-277-2828-3
12. A. F. Noels, M. Graziani and A. J. Hubert (eds.): Metal Promoted Selectivity in
Organic Synthesis. 1991 ISBN 0-7923-1184-1
13. L. I. Simndi: Catalytic Activation ofDioxygen by Metal Complexes. 1992
ISBN 0-7923-1896-X
14. K. Kalyanasundaram and M. Grtzel (eds.), Photosensitization and Photocata-
lysis Using Inorganic and Organometallic Compounds. 1993
ISBN 0-7923-2261-4
15. P. A. Chaloner, M. A. Esteruelas, F. Jo and L. A. Oro: Homogeneous Hydrogen-
ation. 1994 ISBN 0-7923-2474-9
16. G. Braca (ed.): Oxygenates by Homologation or CO Hydrogenation with Metal
Complexes. 1994 ISBN 0-7923-2628-8
17. F. Montanari and L. Casella (eds.): Metalloporphyrins Catalyzed Oxidations.
1994 ISBN 0-7923-2657-1
18. P.W.N.M. van Leeuwen, K. Morokuma and J.H. van Lenthe (eds.): Theoretical
Aspects of Homogeneous Catalysis. Applications of Ab Initio Molecular Orbital
Theory. 1995 ISBN 0-7923-3107-9
19. T. Funabiki (ed.): Oxygenases and Model Systems. 1997 ISBN 0-7923-4240-2
20. S. Cenini and F. Ragaini: Catalytic Reductive Carbonylation of Organic Nitro
Compounds. 1997 ISBN 0-7923-4307-7
21. A.E. Shilov and G.P. Shulpin: Activation and Catalytic Reactions of Saturated
Hydrocarbons in the Presence of Metal Complexes. 2000
ISBN 0-7923-6101-6
22. P.W.N.M. van Leeuwen and C. Claver (eds.): Rhodium Catalyzed Hydroformyla-
tion. 2000 ISBN 0-7923-6551-8
KLUWER ACADEMIC PUBLISHERS DORDRECHT / BOSTON / LONDON
*Volume I is previously published under the Series Title:

Homogeneous Catalysis in Organic and Inorganic Chemistry.

Mechanisms in Homogeneous Catalysis A Spectroscopi PDF

Uploaded by

Copyright:

Available Formats

Mechanisms in Homogeneous Catalysis A Spectroscopi PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Mechanisms in Homogeneous Catalysis A Spectroscopi PDF

Uploaded by

Copyright:

Available Formats

Catalysis by Metal Complexes

PIET W.N.M. VAN LEEUWEN

KLUWER ACADEMIC PUBLISHERS

2002 Kluwer Academic Publishers

All rights reserved

Created in the United States of America

Visit Kluwer Online at: http://www.kluweronline.com

This book covers the developments in rhodium catalyzed

2 Hydroformylation with unmodified rhodium catalysts 15

3 Rhodium phosphite catalysts 35

5 Asymmetric hydroformylation 107

5.2.5 Chiral cooperativity and effect of substituents in

6 Hydroformylation in organic synthesis 145

7 Aqueous biphasic hydroformylation 189

8 Process aspects of rhodium-catalyzed hydroformylation 203

8.6.3 Linear higher alkenes to mainly detergent

9 Catalyst preparation and decomposition 233

10 Novel developments in hydroformylation 253

Piet W. N. M. van Leeuwen

1.1 History of phosphorus ligand effects

In this chapter we will briefly review phosphorus ligand effects in

of ligand effects in the field of homogeneous catalysis using

Figure 1. Structures of dppe, SHOP ligand, DIOP, and DIPAMP

Bidentate ligands have played an important role in the development of

Figure 2. Structures of Phenyl--GLUP, BINAP, and DuPhos

A beneficial use of bidentate diphosphines was discovered in 1971 by

Figure 3. Asymmetric phosphine ligands BINAPHOS and Josiphos

In carbonylation chemistry using phosphine or phosphite complexes of

Shells process-related patents often use the propane bridged 1,3-bis-(di-2-

Figure 4 Ligands for bulk chemical processes

Another impressive ligand effect reported by Drent and coworkers [37]

The first generation of hydroformylation catalysts was based on cobalt

tppms "bulky" phosphite UCC ligand

formation. Since the mid-seventies the rhodium catalysts started to replace

In recent years the interest for hydroformylating higher alkenes with

1.3 Ligand parameters

ligands. An expansion of Tolmans list of -values was published by Bartik

For monodentate phosphorus ligands the cone angle is defined as the

Several studies have been conducted aiming ai the separation of steric

plotting oxidation activity versus position in the periodic table. In addition,

33 Metal-catalyzed Cross-coupling Reactions, eds. F. Diederich and P.J. Stang, Wiley-

Hydroformylation with unmodified rhodium catalysts

Raffaello Lazzaroni, Roberta Settambolo* and Aldo Caiazzo

The first investigations on rhodium-catalyzed hydroformylation were

persistent topicality is the fact that the mechanism of hydrofomylation is far

2.2 Regioselectivity in the rhodium-catalyzed

Figure 1 Regioisomeric aldehydes obtained from the rhodium-catalyzed hydroformylation of

The influence of substrate structure as well as of reaction parameters on

2.2.1 Catalyst precursors

The following unmodified catalyst precursors are those most frequently

2.2.2 Influence of the alkene structure on the regioselectivity

The rhodium-catalyzed hydroformylation of many simple and

In order to examine the influence of the alkene structure on the

These substrates cannot isomerize into internal alkenes, due to the

It is well known that alkenes with a hydrogens (linear 1-alkenes,

(R, R' = Ar, alkl )

The hydroformylation rate in the case of vinylidenic alkenes is very low

Table I. Isomeric composition of aldehydic products arising from styrene hydroformylation

[Rh(COD)(Cl)]2 80 40 1.5 95/5 13