The relative importance of heterogeneous and homogeneous methanol carbonylation using supported rhodium catalysts in the liquid phase

Article abstract of DOI:10.1016/S0022-328X(97)00438-5

Rhodium catalysts supported on ZrO₂, carbon, a cross-linked polystyrene with pendant Ph₂P groups (SDT) or polyvinylpyrrolidone have been tested as catalysts for the heterogeneous carbonylation of methanol in the bulk liquid phase. In all cases, leaching of the catalyst into solution occurs and the observed increase in the rate of production of methyl ethanoate with time is attributed to the rate of the homogeneous reaction being faster than that of the heterogeneous. A detailed kinetic analysis for the catalyst supported on polyvinylpyrrolidone suggests that the rate constant for the homogeneous reaction is ca. 2× that of the heterogeneous at 150°C.

Full text of DOI:10.1016/S0022-328X(97)00438-5

Ž
.
Journal of Organometallic Chemistry 551 1998 229–234
The relative importance of heterogeneous and homogeneous methanol
1
carbonylation using supported rhodium catalysts in the liquid phase
a
a
b
c
Nicola De Blasio , Margaret R. Wright , Enzo Tempesti , Carlo Mazzocchia ,
a,)
David J. Cole-Hamilton
a
School of Chemistry, UniÕersity of St. Andrews, St. Andrews, Fife, KY16 9ST , Scotland
Dipartimento di Chimica e Fisica per i Materiali, UniÕersita di Brescia, Via Valotti 9, 25123 Brescia, Italy
Dipartimento di Chimica Industriale e Ingegneria Chimica, Politecnico di Milano, Piazza Leonardo da Vinci 32, 20133 Milano, Italy
b
`
c
Received 9 June 1997; received in revised form 3 July 1997
Abstract
Ž
.
Rhodium catalysts supported on ZrO2, carbon, a cross-linked polystyrene with pendant Ph2P groups SDT or polyvinylpyrrolidone
have been tested as catalysts for the heterogeneous carbonylation of methanol in the bulk liquid phase. In all cases, leaching of the
catalyst into solution occurs and the observed increase in the rate of production of methyl ethanoate with time is attributed to the rate of
the homogeneous reaction being faster than that of the heterogeneous. A detailed kinetic analysis for the catalyst supported on
polyvinylpyrrolidone suggests that the rate constant for the homogeneous reaction is ca. 2= that of the heterogeneous at 1508C. q 1998
Elsevier Science S.A.
Keywords: Methanol; Rhodium catalyst; Kinetic analysis
1. Introduction
with methyl iodide, methanol and water is recharged
into the reactor. During the product separation phase,
the rhodium complex is not taking part in the catalytic
reaction so that the throughput of product per mole of
rhodium is reduced from its optimum value. There is,
therefore, scope for the development of a process which
uses a genuine flow system and ideally this will involve
a heterogeneous catalyst with the reactants and products
being transported in the liquid or gas phase.
The carbonylation of methanol accounts for the pro-
duction of 3.5 million tons of ethanoic acid per year.
The reaction is promoted by hydrogen iodide, which
reacts with methanol to form iodomethane. The C–C
bond forming reaction is the metal catalysed carbonyla-
tion of iodomethane to ethanoyl iodide. These organic
reactions are outlined in Scheme 1. The preferred pro-
cess is based on a homogeneous rhodium complex,
There have been many reports of the attempted de-
velopment of heterogeneous catalysts for methanol car-
w
Ž
.
xy w x
1 , although an iridium-based process
Rh CO I2
2
w x
has recently been commercialised 2 . A full understand-
ing of the chemistry and advances in the process engi-
neering involved mean that the problems usually associ-
ated with homogeneous catalysts, such as the separation
of the product from the catalyst and the solvent, have
been overcome and plants can be run continuously for
w
x
w
x
bonylation either in the liquid 3–5 or the gas 6,7
phase. In the liquid phase supports such as ion exchange
w
x
resins 3,4 and polyphosphines derived from styrene
w
x
divinyl benzene copolymers 5,8 have proved effective,
w x
several years without rhodium loss 1 . The process is
run as a continuous batch process so that some of the
catalytic mixture is removed from the reactor, the prod-
uct is separated by distillation and the catalyst together
)
Corresponding author.
Scheme 1. Organic reactions occurring during the carbonylation of
methanol.
1 Dedicated to Professor Peter Maitlis, a great scientist and friend.
0022-328Xr98r$19.00 q 1998 Elsevier Science S.A. All rights reserved.

(
)
230
N. De Blasio et al.rJournal of Organometallic Chemistry 551 1998 229–234
Ž
.
Ž
.
Ž
.
but the similarity in rates found for the homogeneous
and heterogeneous systems together with the observa-
Johnson Matthey , bis prop-2-yl ether Aldrich and
Ž
.
rhodium on carbon 1%, Engelhard were used as re-
x w x w x w
w x
w x
w
Ž
.
Ž
.
x
tion of rhodium leaching 5 has led to the proposal 1
that the rhodium species is dissolved from the catalyst
during the reaction and some of it is redeposited once
the CO pressure is released. The working catalyst would
then be identical to that used in the homogeneous
systems. Attaching the rhodium complex to a carbon
support via a diphosphine ligand increases the stability
compared with the diphenylphosphino derived
ceived. Rh2 OAc P2MeOH 12 , Rh2Cl2 CO 12
4
4
and rhodium complexes supported on zirconia were
w
x
prepared by literature methods 13 .
2.1. Preparation of supported catalysts
[
]
2.1.1. RhrSDT 14
Ž
.
w
Ž
.4 x
SDT 1 g was added to a solution of Rh2Cl2 CO
0.1 g in chloroform 125 cm and the mixture stirred
3
Ž
.
Ž
.
w x
polystyrene, but deactivation still occurs 8 . In some
Ž
.
until the solution was colourless 48 h . The polymer
cases, higher rates have been claimed for heterogeneous
catalysts supported on quaternised or oxidised
Ž
was collected and washed with chloroform 2=10
cm3 and diethyl ether 10 cm . The rhodium contain-
3
.
Ž
.
w x
polyvinylpyridines 9 , than normally observed for ho-
Ž
Ž
.
ing polymer 0.47 g was activated by stirring in
mogeneous systems, but direct comparisons under iden-
tical conditions were not carried out. High stability
towards rhodium leaching has been claimed for rhodium
3
.
iodomethane 4.5 cm in the dark for 40 min. The
methyl iodide was then removed in vacuo. Atomic
absorption analysis showed the rhodium content to be
2.21%. This catalyst was used in one experiment in
which the conversion after 4 h was 1.5% and 19% of
the rhodium had leached from the catalyst.
w
x
supported on polyvinylpyrrolidone 10 .
In this paper, we report studies aimed at evaluating
the effect of different supports upon the activity and
stability towards leaching of the catalyst in liquid phase
batch reactions, in an attempt to identify whether a
genuine heterogeneous reaction can occur under these
conditions. The most promising catalyst in terms of
stability and activity was then selected for a more
quantitative evaluation of the relative rates of the homo-
geneous and heterogeneous process. Studies of the use
of this catalyst in the gas-phase heterogeneous reaction
[
]
2.1.2. RhrpolyÕinylpyrrolidone 10
Ž
.
Cross-linked polyvinyl pyrrolidone 3.0 g , methanol
10 g , glacial ethanoic acid 18.8 g , iodomethane 7.5
Ž
.
Ž
.
.
Ž
.
Ž
.
w
Ž
x
Ž
.
g , water 2 g and Rh2 OAc P2MeOH 0.07 g were
4
placed in an autoclave under nitrogen and stirred at
1908C for 2 h. After cooling, the precipitate was col-
lected by filtration and dried in vacuo. Atomic absorp-
tion analysis showed the rhodium content to be 0.72%
compared with a theoretical value of 0.94% if all the
rhodium had been adsorbed.
w
x
will be reported separately 11 , but it showed excellent
stability with no apparent leaching of rhodium occurring
over several tens of hours of reaction.
2.2. Catalytic reactions
2. Experimental
Atomic absorption analyses were carried out by the
University of St. Andrews Microanalytical Service and
the Polytecnico di Milano on Pye Unicam PU9400X,
fitted with a graphite furnace or PU9000 flame spec-
trometers. GC analyses of liquid phase products were
carried out using a Philips PU4500 gas chromatograph
running with JCL 6000 software and fitted with a flame
ionisation detector. Nitrogen was the carrier gas and a
non polar capillary column SGE BP1 was employed.
Gas phase products were analysed using a Carbosphere
80r100 mesh column in a Pye Unicam Series 204 gas
chromatograph fitted with a thermal conductivity detec-
tor.
An autoclave, fitted with a glass liner was charged
Ž
with the catalyst ca. 5 mmol of Rh, weighed and
.
introduced in air . It was then fitted with a special head
that allowed it to be used like a Schlenk tube. The
autoclave was degassed with nitrogen and a mixture of
3
3
Ž
.
Ž
.
methanol 4.2 cm and iodomethane 0.8 cm was
injected in. The autoclave was pressurised to 40 bar
with carbon monoxide and sealed. It was then heated to
the reaction temperature using a heating band. The
reaction temperature stabilised within 10–15 min. It
was stirred with a magnetic stirrer generally at a rate of
350 rpm. It was shown that, although the reaction yield
after short times increased with stirring speed, that at
longer times decreased because catalyst was forced up
the sides of the glass liner and some stuck there. After
the reaction, the autoclave was cooled in a water bath
for at least 1 h and the pressure slowly released. It was
found that some solution had distilled into the space
between the glass liner and the autoclave body. This
had the same concentration of methyl ethanoate as the
bulk solution. The solution was removed in air and
filtered to collect the solid catalyst. Its volume was
Ž
.
Methanol was dried by distillation from magnesium
methoxide, formed in situ from magnesium turnings.
Ž
.
Iodomethane Aldrich was redistilled and protected
from light. Chloroform was distilled from P2O5 and
protected from the light. All operations were carried out
under dry oxygen free nitrogen using standard Schlenk
line and catheter tubing techniques.
Ž
Triphenylphosphine, polymer-supported SDT,
.
Ž
.
Aldrich , polyvinyl pyrrolidone Aldrich , RhCl3 P3H2O

(
)
N. De Blasio et al.rJournal of Organometallic Chemistry 551 1998 229–234
231
Ž
measured and it was analysed by GLC, using bis prop-
.
2-yl ether as an internal standard, and atomic absorption
Ž
.
for Rh . The final volume of the solution was always
less than 5 cm3 2–3.5 cm . Some of the losses were
3
Ž
.
3
Ž
.
mechanical because of handling 1–1.5 cm but some
volatiles were lost during the venting process. On one
occasion, the vented gases were passed through an
3
Ž .
Ž
.
ethanolrCO2 s trap. The recovered liquid 0.5 cm
had the same concentration of methyl ethanoate as the
bulk liquid from the autoclave so that analysis of the
autoclave solution is a reliable indicator of the concen-
tration of methyl ethanoate. It cannot, however, be
relied upon for quantitative analysis of the rhodium
leaching.
The recovered catalyst was analysed by atomic ab-
sorption having been dissolved completely in concen-
Ž
.
trated sulphuric acid at 3008C and nitric acid 65% .
In all cases, GC analysis of the products showed,
methanol, iodomethane and methyl ethanoate together
with traces of 1,1-dimethoxyethane.
Reactions were checked for reproducibility and the
methyl ethanoate yield was found to be reproducible
within "10%.
3. Results
Six rhodium based catalysts were prepared, three on
Ž
.
w x
zirconium IV oxide 6 with the rhodium derived from
w
Ž
.
12 x
Ž
. Ž
Rh4 CO
.
1.3%, A; 4%, B or RhCl3 P3H2O 4%,
w x
Ž
.
C ; one on carbon 7 from RhCl3 P3H2O 1%, D ; one
on SDT, a copolymer of styrene and divinylbenzene
w
x
with pendant Ph2 P units 5,8 , from rhodium trichloride
Ž
.
2.2%, E and the sixth on PVP, polyvinylpyrrolidone,
w
x
Ž .
Ž
.
10 from rhodium II ethanoate 0.72%, F . All of these
Ž .
Fig. 1.
a
Plots of yield of methyl ethanoate against time for the
catalysts were tested in bulk liquid phase reactions for
the carbonylation of methanol at 1508C and pCO s40
bar. The reactions were stopped after a variety of times
up to 10 h and the yield of methyl ethanoate was
measured by GLC. In addition, both the liquid phase
and the collected solids were analysed for their rhodium
content by atomic absorption spectroscopy. In order to
attempt to prevent rhodium redeposition after the reac-
tion, the autoclaves were cooled rapidly from the reac-
tion temperature and the reaction mixtures were filtered
immediately after venting the autoclaves. The major
form of precipitation of rhodium in other systems is as
Ž
.
Ž
.
carbonylation of methanol catalysed by A – – I – – , B – – B – – ,
y4
C
– ' – , D n , and F – v – Rh s7.66=10
mol dmy3
.
.
Ž
.
Ž
.
Ž
. w
x
Initial Rh s10
mol dmy3, pCO s40 bar, T s1508C, MeI 0.8
y3
w
x
Ž
.
Ž
3. Ž .
cm3 , MeOH 4.2 cm . b The same data replotted for catalyst F.
with time. This is shown more clearly for F in Fig. 1b.
The rather unusual upward curvature of these graphs
with time suggests that a more active catalyst is being
produced as the reaction proceeds. A comparison of
catalysts B and C which differ only in the rhodium
2
2 Alternative explanations might include that the medium changes
as the reaction proceeds, or that the reaction is autocatalytic i.e.,
w x
w
Ž
.
xy
RhI3 1 , formed from Rh CO I4 , which in turn is
2
.
Ž
w
Ž
xy
formed from the active Rh CO I2 at low pCO. The
2
.
catalysed by one of the products . However, all reactions were
Ž
.
rhodium III complexes are very dark in colour, whereas
carried out to low conversion so that changes in the medium should
be minor; and there is no literature evidence for autocatalysis in the
methanol carbonylation system. Furthermore, for catalyst B, all the
catalyst leaches after ca. 4 h and in the period 4–8 h, the graph of
yield of methyl ethanoate against time becomes linear. This supports
our suggestion that the generally observed upward curvature arises
because the leached homogeneous catalyst is more active than the
supported catalyst.
w
Ž
.
xy
Rh CO I2 is pale yellow. All the reaction solutions
2
were pale yellow when they were filtered.
Fig. 1a shows the evolution of methyl ethanoate
formation with time for five of the supported catalysts.
Various trends are apparent. Thus, in almost all cases
the rate of production of methyl ethanoate increases

(
)
232
N. De Blasio et al.rJournal of Organometallic Chemistry 551 1998 229–234
precursor used to load the catalyst shows that a more
active catalyst is obtained from loading with
w
Ž
. x
Rh4 CO than with RhCl3 P3H2O. There may, how-
12
ever, be different effects for different supports since
Ž .
RhCl3 P3H2O on carbon D shows rather high activity,
although the data in this case is rather scattered. A
comparison of A and B, which differ only in the
rhodium loading indicates that the loading has little
effect on the rate provided that the absolute amount of
rhodium present is the same. The overall activities of
the catalysts are broadly similar, except that C has
Ž
rather low activity whilst F has the highest activity in
w
x
.
this case Rh is lower than for the other catalysts .
The extent of rhodium leaching from the catalysts
determined from the concentration of rhodium in solu-
tion is presented in Fig. 2a. In all cases, extensive
leaching occurs and this would account for the increase
in the rate of methyl ethanoate production with time if
the solubilised rhodium species more active than the
supported form. Once again, the loading of the catalyst
appears to have little effect upon the stability of the
Ž
.
catalyst A and B and these two catalysts show very
substantial leaching with all the rhodium being lost after
Ž
;3 h. For catalysts C and D both loaded from RhCl3 P
.
3H2O , the leaching is less severe than for the catalysts
12 x
w
Ž
.
obtained from Rh4 CO
and the interaction with
Ž .
ZrO2 C appears to be stronger than that with carbon
Ž .
D . The most stable catalyst towards rhodium leaching
is F, with only ca. 2 ppm of rhodium being observed in
the solution after 2 h of reaction. This rises to 40 ppm
after 10 h.
4. Kinetic analysis
The unusual shape of the graph of methyl ethanoate
production against time, which we have attributed to the
soluble rhodium catalyst being more active than its
supported analogue, prompted us to attempt to analyse
the data in more detail to try to obtain an estimate of the
relative rate constants for the homogeneous and hetero-
geneous reactions.
Ž .
Ž .
Fig. 2. a Extent of rhodium leaching % determined by measuring
the solution concentration of rhodium, assuming a solution volume of
5 cm3 for catalysts A – – I – – , B – – B – – , C – ' – , D n ,
Ž
.
Ž
.
Ž
.
Ž .
Ž
.
Ž .
and F – v – . Conditions as for Fig. 1a. b Extent of leaching of
rhodium determined from analysis of the recovered solid catalyst for
F. This data was used in the kinetic analysis.
w x
system 1 , the overall rate of production of methyl
Ž .
The leaching process can be described by Eq. 1 :
ethanoate is then given by:
ki
Msupported | Msolv
1
Ž .
d MeOAc
n
m
p
ky
i
sk1 MeI Msupported CO
dt
The overall reaction is then made up of the two
processes, as exemplified in Eqs. 2 and 3 :
q
r
s
qk2 MeI Msolv CO
4
Ž .
Ž .
Ž .
w
x
We can assume that MeI does not change during
the reaction because any that is carbonylated is regener-
ated rapidly by Eqs. 5 and 6 :
MeCOIqMeOH™MeOAcqHI
HIqMeOH™MeIqH2O
k1
2MeOHqCO
™
MeOAcqH2O
2
Ž .
Msupported
Ž .
Ž .
k2
5
Ž .
2MeOHqCO ™ MeOAcqH2O
3
Ž .
Msolv
6
Ž .
w
x
w
x
Assuming that the reaction rate depends on MeI
In addition, we can assume that CO remains ap-
proximately constant because we are working at low
w
x
rather than MeOH , as observed in the homogeneous

(
)
N. De Blasio et al.rJournal of Organometallic Chemistry 551 1998 229–234
233
conversion, but it is also known that, at least the
w
x
homogeneous 1,15,16 and the gas-phase heteroge-
w
x
neous 17 reactions are zero order in CO for pCO )5
bar.
Ž .
These assumptions allow us to simplify Eq. 4 to
d MeOAc
n
r
skX1 Msupported qkX2 Msolv
7
Ž .
dt
X
X
m
q
w
x
w
x
where k1 sk1 MeI and k2 sk2 MeI ; m and q are
the orders of reaction with respect to MeI, while n and
r are the orders of reaction for Msupported and Msolv
,
respectively.
There is substantial evidence that the homogeneous
methanol carbonylation reaction is first order in rhodium
w
x
1,15,16 but we can be less certain about the heteroge-
neous. If all the rhodium atoms are isolated in the form
of supported complexes, we can expect first-order be-
haviour, although at high Rh loadings, two Rh atoms
may be involved in the rate-determining oxidative addi-
w
x
Fig. 3. Plot of rate of methyl ethanoate productionr Msolv against
w
x w
x
Msupported r Msolv for catalyst F. Conditions as for Fig. 1.
w x
tion of MeI 5 . If rhodium particles are present, the
reaction may have an order less than 1.
Ž .
We can, therefore, simplify Eq. 7 further:
under these conditions. Straight lines are also obtained
if fractional orders in rhodium are assumed for the
heterogeneous reaction, so the data are insufficient to
d MeOAc
n
skX1 Msupported qkX2 Msolv
8
Ž .
dt
w
x
.
determine the order with respect to Msupported
The homogeneous rate constant for the oxidative
Ž .
Rearranging Eq. 8 ,
w
Ž
.
xy
addition of iodomethane to Rh CO I2 , the rate de-
2
n
d MeOAc
Msupported
termining step of the catalytic reaction, has been mea-
Msolv skX1
qkX2
9
Ž .
Ž
. w x
sured both in Sheffield 0–358C 18 and in St. An-
dt
Msolv
Ž
. w x
drews 25–808C 19 . Extrapolation of the values to
1508C gives k2 as 0.015 and 0.11 dm3 moly1 sy1
respectively. We have also measured the rate of the
Ž
w
x. Ž
.
d MeOAc r dt can be determined by measuring
w
x
the slope of the tangents to the graph of MeOAc
Ž
.
w
x
w
Ž
.
xy
carbonylation of methanol catalysed by Rh CO I2
versus time from Fig. 1a,b . Although both Msolv and
Msupported were measured, problems associated with
possible evaporation of the solution during venting or
while awaiting analysis suggested that the more accu-
rate values should be those of Msupported . These were
2
w
x
at 1508C. The absolute rate is somewhat dependent
upon the reactor, but the values of k2 calculated from
these measurements are in the range 0.044–0.12 dm3
moly1 sy1. This compares with a value of 0.10 dm3
moly1 sy1 calculated from published activation parame-
w
x
w
x
used in the calculations with Msolv being obtained by
w
x
w
x
w x
subtraction of Msupported from the total Rh present
ters for the entire cycle 18 . In order to retain the
Ž
w
x
.
w
x
Ž
during the reaction. d MeOAc rdt r Msolv was then
heterogeneous catalyst within the solution see Section
w
x w
x
.
plotted against Msupported r Msolv . This analysis was
2.2 , we have had to use rather lower stirring speeds in
carried out for catalyst F since this catalyst showed the
this study of the supported heterogeneous catalysts than
in our normal studies and have shown that, at least at
low reaction times the rate of the reaction depends quite
markedly on the stirring rate. This perhaps accounts for
the lower value of k2 measured from the kinetic analy-
sis presented above. Indeed, since catalyst B is com-
pletely leached into solution after 4 h, the reaction in
the period 4–8 h using this catalyst should represent the
purely homogeneous reaction under the conditions that
we have used for our studies of catalyst F. The value of
k2 calculated for the reaction catalysed by B in the
period 7–8 h is 0.027 dm3 moly1 sy1, very close to the
value of 0.031 dm3 moly1 sy1 calculated from the
kinetic analysis of the reaction catalysed by F.
Ž
least leaching the rhodium leaching based on the analy-
ses of the recovered catalysts are shown in Fig. 2b of
.
all of the catalysts and the highest activity for methyl
ethanoate production.
Graphs were plotted for various n and that for ns1
is shown in Fig. 3. The values of kX1 and kX2 can be
calculated since kX1 is the slope of the graph and kX2 is
the intercept. Assuming that the reaction is first order in
Ž
methyl iodide and knowing its concentration 2.56 mol
dmy3 , it is possible to calculate the heterogeneous k1
.
.
Ž
Ž
.
and homogeneous k2 rate constants as 0.014 and
0.031 dm3 moly1 sy1, respectively. The homogeneous
rate constant is then 2.2= faster than the heterogeneous

(
)
234
N. De Blasio et al.rJournal of Organometallic Chemistry 551 1998 229–234
w x
3
R.S. Drago, E.D. Nyberg, A. El A’mma, A. Zombeck, Inorg.
5. Conclusions
Ž
.
Chem. 20 1981 641.
w x
Ž .
We conclude that all the supported rhodium com-
plexes tested show substantial leaching of the catalyst
into solution during the bulk liquid phase carbonylation
of methanol. Polymer supports are more effective at
preventing leaching than zirconium oxide or carbon. For
rhodium on polyvinylpyrrolidone, we have provided
evidence that both the heterogeneous and homogeneous
forms are active with the rate constant for the homoge-
neous reaction being ca. 2= that for the heterogeneous,
4
P. Panster, R. Gradl, US Patent 4,845,163 1989 .
M.S. Jarrell, B.C. Gates, J. Catal. 40 1975 255.
w x
Ž
.
5
w x
6
E. Tempesti, A. Kiennemann, S. Rapagna, C. Mazzocchia, L.
Ž
.
Giuffre, Chem. Ind. 1991 548.
w x
Ž .
7
R.G. Schultz, P.D. Montgomery, J. Catal. 27 1969 13.
Ž .
C.M. Bartish, L.J. Hayes, US Patent 4,325,834 1982 .
Ž .
C.R. Marston, G.L. Goe, Eur. Pat. Appl. 277 1988 824.
w x
8
w x
9
x
w
w
w
w
10 M.O. Scates, R.J. Warner, G.P. Torrence, US Patent 5,281,359
Ž
.
1994 .
x
11 N. de Blasio, E. Tempesti, C. Mazzocchia, A. Kaddouri, D.J.
Cole-Hamilton, 1997, manuscript in preparation.
w
x
assuming that both reactions are first order in Rh .
x
12 G.A. Rempel, P. Legzdins, H. Smith, G. Wilkinson, Inorg.
Ž
.
Synth. 13 1972 90.
x
13 A. Ceriotti, S. Martinengo, L. Zanderighi, C. Tonelli, A. Ianni-
Acknowledgements
Ž
.
bello, A. Girelli, J. Chem. Soc., Faraday Trans. 1 80 1984
1605.
We thank Joanne Rankin for permission to use her
results and for helpful discussions, the EC for funding
under the ERASMUS programme N. de B. and Prof.
w
x
14 E. Tempesti, A. Kiennemann, S. Rapagna, G. Jenner, L. Giuf-
fre, Italian Patent 1,230,855 1989 .
15 D. Forster, J. Am. Chem. Soc. 98 1976 846.
16 P.M. Maitlis, A. Haynes, G.J. Sunley, M.J. Howard, J. Chem.
Soc., Chem. Commun. 1996 2187.
17 K.K. Robinson, A. Hershman, J.H. Craddock, J.F. Roth, J.
Catal. 27 1972 389.
18 A. Haynes, B.E. Mann, G.E. Morris, P.M. Maitlis, J. Am.
Chem. Soc. 115 1993 4093.
19 J. Rankin, A.D. Poole, A.C. Benyei, D.J. Cole-Hamilton, J.
Ž
.
Ž
w
w
x
x
.
Ž
.
w
Ž
. x
Ph. Kalck for the generous gift of Rh4 CO .
12
Ž
.
w
w
w
x
Ž
.
References
x
w x
Ž
.
1
M.J. Howard, M.D. Jones, M.S. Roberts, S.A. Taylor, Catal.
Ž
.
x
Today 18 1993 325.
w x
2
Ž .
Ž
.
Chem. Ind. 1996 483.
Chem. Soc., Chem. Commun. 1997 , submitted.