Rhodium-catalyzed hydrocarbonylation of acetic acid into higher acids

Article abstract of DOI:10.1016/S1381-1169(98)00065-X

The hydrocarbonylation of acetic acid into higher homologues catalyzed by rhodium/iodide systems has been investigated at 20 MPa and 220°C. In homogeneous catalysis the most convenient precursor proved to be [RhCI₂(CO)₂]^- prepared from [RhCl(CO)₂]₂ in the presence of LiI; mean turnover frequencies of 67 h^-1 and selectivities as high as 80% in propionic acid were obtained. In addition, in heterogeneous catalysis, rhodium supported upon activated carbon was observed to be an efficient system for the conversion of acetic acid into propionic acid (80% selectivity) in a fixed bed reactor. The reaction mechanism is thought to be an iodoacetyl rather than an ethanol pathway in the homogeneous system, while the two seem likely to be in operation in the heterogeneous system.

Full text of DOI:10.1016/S1381-1169(98)00065-X

w
i D
Ž
.
Journal of Molecular Catalysis A: Chemical 136 1998 269–278
Rhodium-catalyzed hydrocarbonylation of acetic acid into
higher acids
a
b
c
b,)
Philippe Serp , Laurence Chateau , Roselyne Feurer , Alain Kiennemann
,
a
Philippe Kalck
a
Laboratoire de Catalyse et Chimie Fine, Ecole Nationale Superieure de Chimie de Toulouse, 118, route de Narbonne, 31077 Toulouse
´
Cedex, France
´
b
Laboratoire d’Etudes de la ReactiÕite Catalytique des Surfaces et Interfaces, Ecole Europeenne Chimie Polymeres et Materiaux, 1, rue
´
´
`
´
Blaise Pascal, BP 296 67008 Strasbourg Cedex, France
´
c
Laboratoire de Cristallographie, ReactiÕite et Protection des Materiaux, Ecole Nationale Superieure de Chimie de Toulouse, 118, route de
´
´
´
´
Narbonne, 31077 Toulouse Cedex, France
´
Accepted 2 February 1998
Abstract
The hydrocarbonylation of acetic acid into higher homologues catalyzed by rhodiumriodide systems has been
w
Ž
. xy
investigated at 20 MPa and 2208C. In homogeneous catalysis the most convenient precursor proved to be RhI2 CO
2
1
prepared from RhCl CO
in the presence of LiI; mean turnover frequencies of 67 hy and selectivities as high as 80% in
w
Ž
. x
2
2
propionic acid were obtained. In addition, in heterogeneous catalysis, rhodium supported upon activated carbon was
Ž
.
observed to be an efficient system for the conversion of acetic acid into propionic acid 80% selectivity in a fixed bed
reactor. The reaction mechanism is thought to be an iodoacetyl rather than an ethanol pathway in the homogeneous system,
while the two seem likely to be in operation in the heterogeneous system. q 1998 Elsevier Science B.V. All rights reserved.
Keywords: Hydrocarbonylation; Homologation; Acetic acid; Propionic acid; Rhodium catalysis; Activated carbon; Iodide promoter
1. Introduction
sis of C1–C6 alcohols from synthesis gas, which
give mixtures of products, there are only a small
number of processes which meet these require-
ments.
From COrH2, heterogeneous rhodium cata-
lysts have been reported to selectively afford
The production of C2 oxygenates and higher
alcohols or acids starting from raw materials
such as synthesis gas or methanol, represents an
attractive challenge provided one single product
can be selectively produced. If we exclude the
hydroformylation of olefins, which requires the
previous production of an alkene, or the direct
oxidation of hydrocarbons and the direct synthe-
w
x
ethanol 1,2 . To a lesser extent rhodium cata-
lysts promoted with iridium, manganese oxide
and alkali metals were shown to directly pro-
w
x
duce acetic acid 3–7 . Carbonylation of
w
Ž
.
2 xy
methanol catalyzed by RhI2 CO
in the
)
presence of iodides as promoters was developed
Corresponding author. Fax: q33-3-8841-6863.
1381-1169r98r$ - see front matter q 1998 Elsevier Science B.V. All rights reserved.
Ž
.
PII: S1381-1169 98 00065-X

(
)
270
P. Serp et al.rJournal of Molecular Catalysis A: Chemical 136 1998 269–278
on an industrial scale by Monsanto and gives
rise to selectivities higher than 99% with regard
acetic acid either by homogeneous catalysis
starting from various rhodiumriodide precur-
sors or by heterogeneous catalysis using a 2%
w x
to methanol 8 . Recent improvements of this
Ž
.
process have been patented by British Petroleum
wrw Rhractivated carbon catalyst prepared
w x
w x
9 and Hoechst Celanese 10 . Iridium com-
pounds in particular have been claimed to be
active soluble catalysts to carbonylate methanol
by an Organometallic Chemical Vapor Deposi-
w
x
tion method 26 .
w
x
to acetic acid 9,11 . Many attempts have been
reported in the literature to heterogenize the
catalytic system either in batch or in fixed bed
2. Experimental part
w x
reactors 8 . Generally the catalysts comprise
2.1. Catalyst preparation
w
x
rhodium supported upon various polymers 12
for liquid phase reactions or on activated carbon
w
Ž
.
x w x w
xw
w
Ž
.
x w
x
RhCl CO
26 , AsPh4 RhI2 CO
27 ,
x w x
29
2 2
.
2
w
x
w
x
w
Ž
. Ž
x w
x
Ž
.
13,14 , various inorganic acids 15,16 , or zeo-
Rh CO acac
28 and Rh4 CO
2
12
w
x
lites 17,18 . The reaction is carried out at higher
temperatures in the vapour phase. In such condi-
tions an iodide promoter remains necessary
complexes were prepared using published meth-
ods.
w
Ž
.Ž
.x
The complex RhI3 CO dppp was prepared
as follows: 1.74 mmol 450 mg of
w
x
Ž .
while the proposed mechanism 19,20 is similar
w
x
w
Ž . Ž
Rh CO acac was dissolved, under a nitro-
2
gen atmosphere, in 70 ml of methanol at 208C.
.x
to that assumed for homogeneous catalysis 11 .
On these various supports, operating in condi-
tions approaching those of the homogeneous
process, numerous heavier acids were produced.
A significant increase in the pressure )20
MPa , the temperature 2508C , as well as the
Ž
.
One equivalent of 1,3-bis diphenylphosphino
propane was added to the stirred solution to-
gether with 60 ml of acetic acid and 5 ml of a
57% solution of hydriodic acid. Immediate pre-
cipitation of a dark orange powder occurred.
This precipitate was filtered, washed with dry
ether and dried under reduced pressure. Recrys-
Ž
.
Ž
.
introduction of synthesis gas instead of carbon
w
x
monoxide 15 resulted in large amounts of C3,
C4 and C5 carboxylic acids with a classical
Ž
.
Ž
Ž
.
Anderson–Schulz–Flory ASF distribution ca.
tallization dichloromethane–hexane at y258C
.
20% wrw propionic acid .
yielded the air stable complex as thin dark red
Ž . Ž .
The production of a single Cn carboxylic acid
needles yield 85% . Infrared KBr pellets : nCO
1
1
2084 cmy ; H NMR CD2Cl2, d: ppm : 7.1–
Ž .
by homologation of the Cn acid or its corre-
sponding ester has a limited reference within the
y1
Ž . Ž . Ž . Ž .
8.1 m, 20H , 3.6 2H , 2.02 2H , 2.12 2H .
31
1
w
x
Ä 4 Ž . Ž .
P H NMR CD2Cl2, d: ppm : y0.59 d ,
literature 21–25 since a few papers deal with
the high pressure homogeneous catalysis in the
presence of ruthenium, rhodium, nickel or palla-
dium complexes with an iodide promoter.
Among these catalytic systems, rutheniumr
iodide mixtures afforded the best results, al-
JRh – P s96 Hz.
Full details of the preparation of the sup-
ported rhodium catalyst by an organometallic
chemical vapour deposition method have been
w
Ž
x
reported elsewhere 26 . Typically, vapours of
w
. x
though the catalytic activity was limited to 8
the complex RhCl CO 2 2, diluted in a he-
lium–hydrogen mixture, were contacted with
activated carbon maintained in a fluidized bed.
The duration of the deposit was adjusted to
hy at 46 MPa, and 2208C 24 . Furthermore, to
the best of our knowledge, we present here the
first approach in heterogeneous catalysis of the
homologation reaction by hydrocarbonylation of
carboxylic acids or esters.
1
w
x
Ž
.
obtain 2% wrw Rh on activated carbon. This
one-step method permitted the production at
Ž
.
This paper reports on the selective prepara-
tion of propionic acid by hydrocarbonylation of
low temperature 1258C of well-dispersed
metallic rhodium aggregates. The activated car-

(
)
P. Serp et al.rJournal of Molecular Catalysis A: Chemical 136 1998 269–278
271
bon used in this study is commercially available
was regulated by a Tescom 70 MPA, the gas
flow measured by a Setaram 3100 Flowmeter
and regulated by an electronic controller cou-
pled with a Brooks 5835N electrovalve. At the
exit of the reactor the liquids were trapped at
108C under high pressure. The pressure of the
exit gas was reduced to 2 MPa and the gas
phase was analyzed by G.C. with both TCD and
FID detectors. The CH3OH–CH3I mixture or
acetic acid–alkyl iodide mixture was cooled to
08C to avoid evaporation and introduced into
the device by a model 303 Gilson metering
pump. Prior to the catalytic test, the catalyst
2
Ž
.
Aldrich , it possesses a BET area of 725 m
gy1, and was used without further activation or
modification of its surface properties. EDS anal-
ysis revealed the presence of traces of sulphur,
Ž
.
and of small amounts total ca. 1% of various
Ž
w
x
w
x
w
x
w x.
metals Fe f Cu ) Cr f Ni . The rhodium
catalyst prepared by this method contained 2%
Ž
. Ž
Rh wrw as determined by both atomic ab-
.
sorption and ICP and possessed a mean particle
size of 3 nm CO chemisorption measurements .
Ž
.
2.2. Homogeneous catalytic tests
y
1
Ž
.
Ž
0.5 g was reduced by pure hydrogen 2 l h
y
1
.
The homogeneous hydrocarbonylation of
acetic acid was performed in a Hastelloy-B2
100-ml autoclave, equipped with a 500-ml bal-
last vessel to maintain a constant reaction pres-
sure. The reactor was charged with 180 mmol
gcat. during heating from room temperature to
2308C 28C min
y
1
Ž
.
and then 15 h at 2308C. The
hydrogen was then released and the synthesis
gas was introduced progressively. Catalytic ac-
tivities were compared on the basis of methanol
Ž
Ž
.
acetic acid, 6.2 mmol of hydriodic acid 1.4 g of
or acetic acid molar % remaining in the final
mixture. Indeed, it has been shown that, under
the present experimental conditions, some acetic
acid can be formed by carbonylation of some
methanol produced by reaction of CO and H2
upon the rhodium catalyst.
y
4
.
a 57% solution in water and 1.48 10 g at of
rhodium. After purging with nitrogen, the pres-
sure was fixed at 10 MPa COrH2 s2 and,
when the temperature of 2208C was reached 15
min , the pressure was adjusted to 20 MPa. The
stirring 1500 rpm was maintained for the en-
tire run 6 h . Following a slow decompression
of the reactor, the liquid products were analyzed
on a Carlo Erba GC 6000 gas chromatograph
Ž
.
Ž
.
Ž
Ž
.
.
3. Results and discussion
Ž
equipped with a 50 m CPWAX 58CB i.d.: 0.32
3.1. Homogeneous catalytic study
.
Ž
mm column or by GCrMS Perkin-Elmer
QMASS 910 . The gaseous products were ana-
lyzed on a Perkin–Elmer Autosystem GC.
2.3. Heterogeneous catalytic tests
.
For the homogeneous hydrocarbonylation of
acetic acid the activity of various catalysts was
examined, and aqueous HI was selected as the
iodide-containing promoter. A total syngas pres-
sure of 20 MPa was maintained during all of the
catalytic run at a constant temperature of 2208C.
The best catalytic activity was observed for a
COrH2 molar ratio of 2. The different molar
The catalytic tests were performed in a stain-
Ž
.
less steel Z2CND17-12 reactor of 8 mm i.d.
which was lined by copper 6 mm inner diame-
ter. The length of the heated element was 0.43
m and the catalyst was maintained in the central
section of the reactor by glass wool plugs. The
COrH2 mixture was prepared in the laboratory
and controlled by G.C. TCD detector . The
syngas was compressed to 70 MPa in a 500-ml
high pressure vessel. The pressure of the device
w
x w x w
x w
x
ratios of CH3COOH r Rh , CH3COOH r HI ,
w
x w
x
and HI r Rh were respectively 1215, 29 and
42, in order to be near to the conditions required
for the low pressure methanol carbonylation
reaction, mainly for the concentrations of the
promoter. Table 1 presents the conversion of
acetic acid and the yields in C3–C5 acids for
Ž
.

(
)
272
P. Serp et al.rJournal of Molecular Catalysis A: Chemical 136 1998 269–278
Table 1
Acetic acid hydrocarbonylation
y1
a
Ž
.
Ž
.
Ž
.
Rhodium catalysts
TOF h
Conversion acetic acid %
Carboxylic acid yield mol %
C3
n-C4
iso-C4
C5
RhCl3 P3H2O
12
22
42
34
26
46
6
11
21
17
13
23
96.3
77.8
81
49.9
95.8
95.6
total 3.7
y
w
w
w
w
w
xw . x
AsPh4 RhI2 CO
2
Ž
9.7
8.9
21.1
12
9.8
25.2
0.5
0.3
3.5
y
Ž
. x
RhCl CO
2
Ž
. Ž 2 .x
Rh CO acac
2
Ž
.12 x
Rh4 CO
RhI3 CO dppp
total 4.2
total 4.4
Ž
.Ž .x
y
aCarboxylic acid yield with regard to acetic acid converted.
Reactives charge: rhodium 1.4 10y4 g at., HI 6.2 mmol, acetic acid 0.18 mol.
Typical operating conditions: 2208C, 2 MPa at the working temperature, COrH2 2r1 , 6 h.
Ž
.
the investigated catalysts, namely RhCl3 P3H2O,
ties of insoluble rhodium were observed. In a
separate experiment, it was verified that this
‘rhodium black’ precipitate did not induce any
homologation of acetic acid under the operating
conditions.
w
xw
Ž
.
x w
Ž
.
x
w
Ž
.
AsPh4 RhI2 CO , RhCl CO 2 2, Rh CO -
2
Ž
.
x w
Ž
.
2 x
w
Ž
.Ž
.x
acac , Rh4 CO
and RhI3 CO dppp .
12
From these results, it appears that the produc-
tion of higher acids starting from acetic acid is a
very selective process since almost no ester was
produced for all the investigated systems. In-
deed, the analysis of the liquid phase by GCrMS
allowed us to identify, besides small amounts of
w
Ž
. Ž
.x
The compound Rh CO acac appears to
2
be the precursor permitting the widest homolo-
gation since 50% propionic acid, 46% butyric
acid, and 4% pentanoic acid were produced.
This system is efficient as regards the homolo-
gation reaction itself, but the desired selectivity
in a single acid is not met. High selectivities in
propionic acid were obtained starting from
Ž
iodine containing compounds C2 H5I4CH3I
.
)C3H7I)C4H9I , only traces of ethanol and
of the corresponding ethyl esters. However, the
main competing reaction to rhodium-catalyzed
acetic acid hydrocarbonylation is the water–gas
w
Ž
. x
RhCl3 P 3H2O and Rh4 CO , this latter
12
Ž
.
shift WGS reaction as evidenced by a signifi-
cant quantity of carbon dioxide present in the
gaseous phase. Contrary to the ruthenium sys-
species encouraging twice the conversion of
acetic acid with respect to the rhodium salt. In
terms of selectivity, only 4% butyric acid was
w
x
w
tems described by Knifton 24 , which produced
significant amount of ethane and propane, the
analysis of the gaseous phase did not reveal
produced. Concerning the complex RhCl-
Ž
. x
CO 2 2, it is known that the addition of hydri-
x w
w
Ž
.
x
odic acid affords the species H RhI2 CO
11 .
2
Ž
hydrocarbon formation only traces of methane
were found . This could be due to the low
For this reason we prepared the homologous
.
w
x w
Ž
.
x
complex AsPh4 RhI2 CO . Surprisingly,
bearing in mind that the same molar quantity of
rhodium was introduced, RhCl CO led to
2
partial pressure of hydrogen in the present con-
ditions, but the main reason probably arises
from the low hydrogenating character of the
rhodium catalyst 11,19 compared to ruthe-
nium. Although the stoichiometric ratio required
for the homologation reaction is 0.5, the CO
insertion step is probably rate-limiting since the
highest conversions were obtained for a COrH2
ratio of 2. In addition, it should be mentioned
that, after each catalytic run, significant quanti-
w
Ž
.2 x2
twice the activity of the pre-formed anionic
complex, the various selectivities in C3–C5
acids being identical. In our opinion the differ-
ence in productivity stems from the nature of
w
x
the counter ion, either Hq or AsPh4 q. Due to
w
x
w
xq
its steric hindrance, the cation AsPh4 should
w
Ž
.
2 xy
present no interaction with the RhI2 CO
species, whereas Hq may give rise to an ion-

(
)
P. Serp et al.rJournal of Molecular Catalysis A: Chemical 136 1998 269–278
273
Table 2
Ž
.
selectivity in propionic acid ca. 96% was ob-
tained. As compared with the other cases, little
insoluble rhodium was found after the catalytic
run: therefore this system appears to be attrac-
tive.
FTIR analysis of the rhodium solution after catalysis
y1
Ž
.
Rhodium catalysts
nCO cm
Assignment
Ž
.
Ž
.
w
Ž
. xy
RhCl3P3H2O
2090 vs , 2035 vw
RhI4 CO
2
w
xw
Ž
. x
Ž
.
.
AsPh4 RhI2 CO
2083 vs
y
2
w
Ž
. x
Ž
Ž
.
w
Ž
Ž
Ž
Ž
Ž
Ž
. xy
RhCl CO
2094 vs , 2071 vs
RhI4 CO
,
2
2
2
w
w
w
w
w
.xy
After the catalytic runs, infrared spectra were
recorded. The results are displayed in Table 2.
Infrared measurements for the dppp complex
RhI4 CO
RhI4 CO
RhI4 CO
RhI4 CO
w
Ž
. Ž
12 x
.x
Ž
Ž
.
Ž
Ž
.
.
. xy
Rh CO acac
2090 vs , 2030 vw
2
.
w
Ž
.
.2 xy
Rh4 CO
2090 vs , 2070 vs
,
2
.xy
revealed that the single nCO at 2089 cmy
1
w
Ž
.Ž
.x
Ž
.
.Ž
.x
RhI3 CO dppp
2089 vs
RhI3 CO dppp
w
Ž
.Ž
.x
belongs to the RhI3 CO dppp species. Ex-
w
x w
Ž
.
x
cept for AsPh4 RhI2 CO which gave a nCO
2
1
pairing interaction with the anionic rhodium
sub-complex. The dramatic accelerating effect
of Lewis acids upon CO insertion, including
that of proton acids has previously been re-
band at 2083 cmy that remains unassigned, the
w
Ž
.
xy
other systems gave rise to RhI4 CO
and
which were identified by their
w
Ž
.
xy
RhI4 CO
2w
x
nCO bands 32 . In a separate experiment, addi-
tion of HI to the complex AsPh4 RhI2 CO
at room temperature in acetic acid afforded the
w
x
w
xw
Ž
.
2 x
ported 30 . It is known that significant amounts
of water are required in the Monsanto process to
w
Ž
.
2 xy
w xw Ž
.x
maintain the stability of the RhI2 CO
cata-
complex AsPh4 RhI4 CO . All attempts to
w
x
Ž
.
lyst 10 , this species being generated from vari-
ous rhodium salts or complexes. In the present
case, there is a competition between the rate of
formation of the active species starting from
isolate these two tetraiodorhodium III species
from a catalytic solution by addition of
w
x
w
x
AsPh4 Cl or AsPh4 I failed. The only dark red
crystalline product isolated was proved by its
w
Ž
.
x
w
Ž
. Ž
.
x
w
x
RhCl3 P3H2O, RhCl CO 2 2, Rh CO acac
X-ray crystal structure to be in fact AsPh4 I3.
2 x2
it was deter-
2
w
Ž
.
x
w
Ž
.
and Rh4 CO , and the rate of decomposition
Starting from RhCl CO
12
Ž
.
into insoluble materials due to the low water
concentrations. Interesting results were also ob-
tained with RhI3 CO dppp , where dppp is
the ligand bis diphenylphosphino propane. In a
recent paper, Moloy and Wegman 31 demon-
strated the stabilizing effect of a diphosphine
ligand upon a rhodium–ruthenium system for
the reductive carbonylation of methanol to
ethanol. Table 1 shows that a 23% conversion
of acetic acid was achieved in 6 h and a high
mined that around 85% wrw of the rhodium
introduced decomposed into insoluble rhodium.
The stabilization of the generated species was
particularly investigated. As in the carbonyla-
tion of methanol 10,11 or methyl acetate 33 ,
the addition of increasing amounts of LiI was
investigated. Adding up to 10–20 equivalents of
LiI per rhodium led to the observation of an
w
Ž
.Ž
.x
Ž
.
w
x
w
x
w x
Ž
.
increase in catalytic activity see Table 3 . The
highest value of the turnover frequency was
Table 3
Influence of the LiI concentration on the acetic acid conversion
y1
a
w
x w
LiI r Rh
x
Ž
.
Ž
.
TOF h
Acetic acid
conversion %
Carboxylic acid yield mole %
Ž
.
Propionic
iso-Butyric
n-Butyric
Pentanoic
no LiI
2
4
10
20
40
42
50
59
67
63
48
21
25
29
33
31
24
81
8.9
8.3
5.8
6.2
7.8
6.8
9.8
9.3
7.1
9.3
8.6
9.1
0.3
0.3
0.2
1
0.3
y
80.1
86.9
83.5
83.3
84.1
aCarboxylic acid yield basis acetic acid converted.

(
)
274
P. Serp et al.rJournal of Molecular Catalysis A: Chemical 136 1998 269–278
67 hy1, around eight times the value observed
Ž
allows to work at high temperature 2508C–
2858C , with a stable activity over long reaction
w
x
.
by Knifton 24 with a ruthenium catalyst. For
higher concentrations of LiI the activity was
Ž
. w x
periods 500 h 13 . The main role of this
support is to promote the alkyl iodide oxidative
addition to metal centres which have a low
affinity for halides. This promoting effect could
be explained by an electronic transfer between
the support and the metal, resulting in an accel-
eration of the limiting steps of the carbonylation
w
x w
x
reduced. While for LiI r Rh s10, around 5%
insoluble rhodium were found following a cat-
alytic experiment, almost no insoluble material
w
x w
x
was detected at LiI r Rh s40. For the car-
bonylation of methanol or methyl acetate, the
effect of LiI has been frequently discussed,
w
x
w x
particularly by Fulford et al. 34 . The role of
LiI is not only to stabilize the active species
reaction 36 .
In sharp contrast with supports such as La2O3
or SiO2, the leaching of rhodium from the
activated carbon supported catalyst during the
methanol carbonylation reaction was much re-
duced. In addition, although the temperature
was increased from 208C with regard to the
w
Ž
.
2 xy
RhI2 CO but also to exert a promoting
effect upon the oxidative addition of CH3I.
After a catalytic run and a slow depressurization
of the reactor, the solutions were analyzed by
infrared. Two nCO bands were detected at
2093 and 2073 cmy and were assigned to
conditions previously used for silica 20 , the
1
w
x
)
w
w
Ž
.
xw
Ž
.
2 x
w
Ž
.
n x
Ž
.
Li solvent RhI4 CO
and Li solvent
RhI4 CO respectively, a slight shift being due
to the changing in the cation species 32 .
2% wrw RhrC catalyst displayed a con-
stant activity during the 16–115 h investigated
period, whereas we have observed that for a 2%
n
Ž
.x
w
x
Ž
.
wrw RhrSiO2 catalyst the catalytic activity
3.2. Heterogeneous catalytic study
decreased dramatically after 60 h. The anchor-
ing of rhodium was greatly improved on acti-
vated carbon with regard to silica since after the
catalytic tests the loss of metal was determined
From previous work dealing with the
methanol carbonylation into acetic acid it ap-
pears that the activity of carbon supported metal
catalysts follows the order: Rh)Ir)Ni)Pd
)Co)Ru)Fe 35 . No significant synergic
effect was observed upon the addition of a
second metal to the active metallic phase. How-
ever, the use of activated carbon as a support
to be 35% from RhrC) 115 h run and 90%
Ž
.
Ž
.
from RhrSiO2 60 h run . Our results in a fixed
bed reactor give strong evidence that the hydro-
carbonylation is due to supported rhodium. Pre-
vious investigations in batch reactors had led
the authors to observe a poor catalytic activity
w
x
Table 4
Hydrocarbonylation of methanol to higher acids on 2% RhrC)
Selectivitya
Productivity
Ž .
Reaction time h
y1
y1
AcMe
AcEt
C2
C3
C4
C5
g g
hy1
g g hy1
cat.
Rh
16
24
41
49
65
73
89
97
115
2.2
0.2
0.7
y
3.2
1.6
1.2
1.7
1.5
1.5
1.4
1.3
1.4
79
81
81
80
79
79
78
78
78
14
15
15
16
17
17
17
17
18
1.6
2.0
2.2
2.2
2.6
2.3
2.6
2.6
2.6
0.2
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.2
0.49
0.58
0.50
0.48
0.45
0.57
0.57
0.53
0.61
24.5
29
25
24
y
22.5
28.5
28.5
26.5
30.5
0.4
0.4
0.3
0.5
a
Ž
.
Selectivity basis on the product of the hydrocarbonylation molar percentage, H2O not included .
Operating conditions: Ps21 MPa, Ts2508C, COrH2 s60r40, gas flows3 l hy1
g
, CH3OHrCH3Is7.7r1 M, CH3OHqCH3Is
y1
cat.
0.2 ml hy1
g
.
cat.
y1

(
)
P. Serp et al.rJournal of Molecular Catalysis A: Chemical 136 1998 269–278
275
for RhrC) catalysts and to conclude that it was
Table 6
Catalytic behavior of 2% RhrC) for hydrocarbonylation reac-
due to homogeneous catalysis resulting from the
tions
w
x
rhodium leaching 37 . Concerning the La2O3
support, we observed not only a loss of rhodium
but also a solubilisation of the support by acetic
acid to afford lanthanum acetate.
Reaction Remaining Selectivityb
Ž .
time h
substratea
AcMe AcEt C3
C4
C5
14
23
38
45
15
24
30
76.3
79.5
77.4
79.1
71.7
74.3
77.9
13.9
y
27.2
12.6
8.5
10.4
18.3
7.7
52.9
77.6
5.6 0.3
9.3 0.8
In order to attain a significant chain growth
the experiments were performed at 21 MPa and
2508C and for a duration of more than 100 h:
the activity remained constant and no methanol
was detected in the reaction mixture. The results
obtained with a 2%RhrC) are presented in
Table 4.
y
79.9 10.8 1.6
76.9 11.6 0.9
69.9 11.4 0.3
78.4 13.4 0.6
y
y
y
y
4.2
77
16.7 0.7
a Molar percentage; the four first rows correspond to CH3COOH
qC2 H5I, and the three last rows to CH3COOCH3 qCH3I.
bSelectivity calculated from the products of the hydrocarbonyla-
Ž
.
Ž
tion molar percentage, H2O not included .
A high selectivity in carboxylic acids )
.
Operating conditions: Ps21 MPa, T s2508C, COrH2 s60r40,
95% was observed with particularly few esters
gas flows3 l hy1
g
, substraterRIs7.7r1 M, substrateqRI
y1
cat.
being produced, and a significant chain growth
y1
s0.2 ml hy1
g
.
cat.
Ž
.
20% M . However, since large quantities of
acetic acid were obtained, the chain growth
remained too short and followed an ASF distri-
bution. These results were in opposition to the
anticipated selective production of propionic or
butyric acids. In order to reach this objective a
study of the direct hydrocarbonylation of acetic
acid was undertaken. The catalytic results using
the same 2% RhrC) catalyst are summarized
in Table 5. Under the same experimental condi-
tions as for methanol it proved possible to pre-
pare propionic acid with an excellent selectivity
most the 20% of products remaining. It is worth
mentioning that an induction of ca. 24 h was
required which corresponded to a stabilization
and an activation period for the catalyst. During
this initial period, significant amounts of esters
Ž
.
methyl- and ethyl acetate were formed. Be-
yond this stage, the selectivity in higher acids as
Ž
.
well as the activity with respect to acetic acid ,
was practically constant. The productivity of the
catalyst was around 0.1 g of higher acids per g
cat.y hy1. Methanol needed to methyl acetate
1
Ž
.
f80% , butyric and pentanoic acids being al-
was provided both by carbon monoxide hydro-
genation and methyl iodide hydrolysis. As
methanol produced under the reaction condi-
tions should afford very quickly methyl acetate,
ethanol was presumably produced by acetic acid
reduction instead of by direct homologation of
methanol.
Table 5
Catalytic behaviour of 2% RhrC) catalyst for acetic acid hydro-
carbonylation
Reaction Remaining Selectivityb
AcOHa
Ž .
time h
AcMe AcEt C3
C4
C5
Concerning the activity of the supported cata-
lyst it is well known that the presence of chlo-
rine atoms has a significant poisonous effect in
16
24
41
48
64
72
88
65
75
77
73
74
74
72
51.2
16.8
1.3
0.3
y
10.9
5.3
3.9
3
2.7
2.3
1.5
32.9
4.8 0.5
62.8 13.3 1.8
75.7 16.3 2.7
75.9 18.7 2.0
78.5 17.2 1.7
78.5 17.5 1.7
72.6 22.6 3.3
w
x
COrH2 reactions 38 . Such an effect was also
Ž
.
y
observed here. Indeed, with a 2% wrw
RhrC) catalyst prepared by conventional im-
pregnation of the support by RhCl3 P3H2O
y
a Molar percentage.
bSelectivity calculated from the product of the hydrocarbonylation
Ž
.
similar mean particle size of 4–5 nm , the
Ž
.
molar percentage, H2O not included .
productivity of carboxylic acids after an induc-
tion period of 24 h was significantly reduced:
85% remaining acetic acid, the product distribu-
Operating conditions: Ps21 MPa, T s2508C, COrH2 s60r40,
y1
gas flow s 3
l
hy1
g
,
CH3COOHrCH3I s 7.7r1 M,
cat.
CH3COOHqCH3Is0.2 ml hy1
g
.
cat.
y1

(
)
276
P. Serp et al.rJournal of Molecular Catalysis A: Chemical 136 1998 269–278
tion being similar to that shown in Table 5. This
classical catalyst contained around 20 at.% of
chlorine whereas those prepared by OMCVD
contained only 3–4 at.% remaining chlorine.
Under the same experimental conditions, it
has also been verified that ethanol hydrocar-
bonylation also leads to both propionic and
those formed during the reduction steps would
be consumed by the WGS reaction, thus shifting
Ž .
the equilibrium shown in Eq. 1 towards the
right.
CH3COOHqHImCH3COIqH2O
1
Ž .
In addition, at least in homogenous condi-
Ž .
tions, the formation of the acetato rhodium I
Ž
butyric–pentanoic acids 64% and 5% respec-
Ž
.
or III species appears rather unlikely. Further-
more, as almost no ethyl propionate was de-
tected in the reaction medium, the presence of
ethanol as a starting material for propionic acid
production seems equally unlikely. Finally, since
the direct use of ethanol rather than acetic acid
as reactant resulted in the formation of half as
much propionic acid under the same experimen-
tal conditions, the ethanol route for the hydro-
carbonylation of acetic acid may be discarded.
Thus, presumably the acetyl species, produced
by the oxidative addition of CH3COI to the
active rhodium species, is reduced by hydrogen
to generate an ethyl rhodium species which give
an acyl species in a CO insertion step. Recently,
.
tively , 14% of starting alcohol being unreacted.
For this latter reaction, the activity of the cata-
lyst was slightly lower than for methanol hydro-
carbonylation yet higher than for acetic acid
hydrocarbonylation. Substitution of methyl io-
dide by ethyl iodide led to the results presented
in Table 6. In this case, while there was a
decrease in the induction period the differences
between the two iodides with respect to cat-
alytic activity or selectivity into higher acids
were not significant. The results obtained start-
ing from either methyl acetatermethyl iodide or
acetic acidrethyl iodide were quite comparable
Ž
.
for both activity and selectivity Table 6 indi-
cating the same mechanistic pathway for the
two starting materials.
wŽ
.
Ž
.
xy
the EtCO Rh CO I3
complex has been
characterized during the carbonylation of ethy-
2
w
x
lene reaction 39 .
4. Mechanistic considerations
4.2. Heterogeneous catalysis
4.1. Homogeneous catalysis
From previous work carried out using a
w
x
Ž
w x
Knifton 24 reported the high pressure ,46
RhrPr6O11 catalyst 20 , as well as from the
literature results, it appears that the mechanisms
involved in the methanol carbonylation are simi-
lar in homogeneous and heterogeneous catalysis
.
MPa, 2208C hydrocarbonylation of acetic acid
with ruthenium homogeneous catalysts and pro-
posed a catalytic cycle in which the key step is
the formation of acetyl iodide. Oxidative addi-
tion, reduction of the acyl to an ethyl group, CO
insertion and reductive elimination of propionyl
iodide are the principal steps of this catalytic
sequence.
In the present work, it is worth noting that
significant amounts of CO2, and therefore hy-
drogen, were produced by the water–gas shift
reaction reducing the amounts of water in the
medium. The hypothesis that acetic acid in the
presence of hydriodic acid gives rise to acetyl
iodide appears attractive. The small volumes of
w
x
40,41 . Furthermore, it is known that the car-
bonylation of ethanol and higher alcohols is at
least ten fold times slower than that of methanol
w
x
11,8 . While in homogeneous catalysis, the hy-
drocarbonylation mechanism of acetic acid is
modelled largely on the alcohol carbonylation,
this might not be the case for heterogeneous
catalysis. Here the formation of acetate species
must to be taken into account. Two routes might
Ž
.
therefore be considered Scheme 1 .
Ž .
i Hydrolysis of acyl iodide in acetic acid
can occur which leads to its chemisorption on
Ž
.
water introduced as the HI solution 57% and
the catalytic surface and produces a carboxylate

(
)
P. Serp et al.rJournal of Molecular Catalysis A: Chemical 136 1998 269–278
277
Scheme 1. Mechanistic pathway for the homologation of acetic acid.
w
x
species. The reduction of this acetate species to
an alkoxy species has been widely studied for
cobalt, nickel or iron heterogeneous catalysts on
CH3CH2-Rh–I , which would insert rapidly
carbon monoxide in agreement with the low
quantities of ethyl iodide observed during the
run.
Ž
. w x
various supports SiO2, Al2O3, C . . . 42 , but
w
x
Orita et al. 43–45 have concluded in their
studies with COrH2 that the presence of the
carboxylate species is essential for oxygenate
The ethyl species could also be obtained via
a facile dehydrogenation of ethanol, yielding
ethylene and subsequently ethyl-Rh formation.
Ž
.
Ž .
EtOH formation. The reduction of acetate to
It is difficult to differentiate route i from
Ž .
alkoxy species could occur readily under the
present experimental conditions. Ethanol formed
through hydrogenolysis or hydrolysis of the
ethoxy species could lead to the esterification of
acetic acid into ethyl acetate or to propionic
acid via ethanol carbonylation. Equally it was
shown over 5% RhrLa2O3 operating condi-
tions of Table 5 that if ethanol leads to a
selectivity comparable to that obtained from
route ii as illustrated by IR results following
chemisorption over 5% RhrPr6O11 of C2 H5I,
CO and H2 route i or C2 H5OH, CO and H2
route ii and heating. In the two cases propi-
onate species were characterized propionate
Ž
.
Ž
.
Ž
Ž
.
Ž
.
bands at 2980, 2944 n CH3 ; 1548 naCO ;
Ž
Ž . Ž .
1470 daCH3 ; 1375, 1309, 1253 dsCH3 ; 1079,
1
1028, 873, 803 cmy CH2 . Moreover, in the
.
Ž
.
two cases the n CO band at 1679 cmy seems to
1
Ž
.
acetic acid 60–63% propionic acid , the activ-
characterize a propionyl species in agreement
1
ity would be higher productivity in higher acids:
with the band of an acyl group at 1624 cmy
Ž
1
1
y1
0.5 g cat.y hy compared with less than 0.1 g
20 and the band at 1679 cm appearing after
w
x
cat.y hy1 . The rate determining step could
a propionaldehyde chemisorption. The presence
of propionate and propionyl species was con-
firmed by chemical trapping experiments. In-
1
.
then be the reduction of carboxylates into alkoxy
species, explaining the weaker reactivity of
acetic acid compared to ethanol. This result is
contrary to that obtained in homogeneous catal-
ysis.
Ž
deed, over the same catalytic surface Rhr
.
Ž
Pr6O11 , after the adsorption of C2 H5I or
C2 H5OH , CO, H2 and the trapping by a great
.
Ž .
ii Alternatively, direct chemisorption of acyl
iodide on the catalytic surface could occur to
excess of CH3I, methyl propionate and methyl
ethyl ketone were characterized by gas chro-
matography. It is worth to mention that the
support alone did not promote the formation of
w
x
produce a CH3CO–Rh–I species. Subsequent
reduction would give rise to the ethyl species

(
)
278
P. Serp et al.rJournal of Molecular Catalysis A: Chemical 136 1998 269–278
w
w
x
11 P.M. Maitlis, A. Haynes, G.J. Sunley, M.J. Howard, J.
these propionate and propionyl species, whereas
the presence of acetate and formate species
occurred, showing clearly that the presence of
rhodium is necessary for chain growth.
All the results indicated the partial transfor-
mation of C2 H5I into ethanol upon the surface.
This was verified by IR, since after C2 H5I
chemisorption the ethoxy bands appeared at
Ž
.
Chem. Soc., Dalton Trans. 1996 2187.
x
Ž
.
12 J. Hjortkjaer, Y. Chen, B. Heinrich, Appl. Catal. 67 1991
269.
w
w
x
x
Ž
.
13 R.G. Schultz, P.D.D. Montgomerry, J. Catal. 13 1969 105.
14 K.K. Robinson, A. Hershman, J.H. Craddock, J.F. Roth, J.
Ž
.
Catal. 27 1972 389.
w
x
15 E. Tempesti, A. Kiennemann, S. Rapagna, C. Mazzocchia, L.
Ž
.
Guifre, Chem. Ind. 1991 548.
w
w
w
x
Ž
.
16 G. Ritter, G. Luft, Chem. -Ing. -Tech. 59 1987 485.
17 M.S. Scurell, R.F. Howe, J. Mol. Catal. 7 1980 535.
18 T. Yashima, Y. Orikasa, N. Takahashi, N. Hara, J. Catal. 59
x
Ž
.
Ž
.
Ž
.
2978, 2927, 2875 n CH3 ; 1475, 1442 dCH2 ;
x
Ž
.
Ž
Ž
.
1386 dCH3 ; 1148, 1097 na C–C ; 1056
Ž
.
1979 53.
w
w
x
x
Ž
.
19 D. Forster, J. Chem. Soc., Dalton Trans. 1979 1639.
20 L. Chateau, J.P. Hindermann, A. Kiennemann, E. Tempesti,
Ž
.
.
n CO ; 882 ns C–C ; 826, 793 and 718 and
ethyl bands at 1209, 946 and 850 cmy1. How-
ever, the CH2 and CH3 vibrations could not be
differentiated between ethoxy or ethyl species.
Due to the simultaneous presence of these two
Ž
.
J. Mol. Catal. 107 1996 367.
w
w
w
w
w
w
x
Ž
.
21 J.F. Knifton, Chem. Tech. 1981 609.
22 J.F. Knifton, Hydrocarbon Process. 60 1981 113.
23 J.F. Knifton, J. Chem. Soc., Chem. Comm. 1981 41.
24 J.F. Knifton, J. Mol. Catal. 11 1981 91.
25 J.F. Knifton, J. Mol. Catal. 16 1982 101.
26 Ph. Serp, R. Feurer, R. Morancho, Ph. Kalck, J. Catal. 157
x
Ž
.
Ž
x
.
x
Ž
Ž
.
.
Ž .
systems the discrimination between routes i
x
Ž .
x
and ii is unlikely. Presumably the two mecha-
Ž
.
1995 294.
nisms should prevail.
w
w
x
x
Ž
.
27 L.M. Vallarino, Inorg. Chem. 4 1965 161.
28 Y.S. Varshavskii, T.G. Cherkasova, Russ. J. Inorg. Chem. 12
Ž
.
Ž
.
1967 899, Engl. Transl. .
w
w
x
29 Ph. Serp, Ph. Kalck, R. Feurer, R. Morancho, Rhodium
Acknowledgements
Ž
.
Express 9 1995 5.
x
30 S.B. Butts, T.G. Richmond, D.F. Shriver, Inorg. Chem. 20
This work was supported by the European
Economic Community through a Brite-Euram
Ž
.
1981 278.
w
w
x
x
Ž
.
31 K.G. Moloy, R.W. Wegman, Organometallics 3 1989 2883.
32 B.T. Heaton, C. Jacob, S. Moffet, J. Organomet. Chem. 462
Ž
.
project BREU CT 91 0458 DOEO . L Chateau
and Ph. Serp thank the EEC for their financial
support. Comptoir Lyon-Alemand-Louyot SA is
gratefully acknowledged for a loan of rhodium
and particularly for a helpful collaboration with
Dr. J.-P. Guerlet, N. Petit and M. Malheiro.
Ž
.
1993 347.
w
w
w
x
33 J.R. Zoeller, V.H. Agreda, S.L. Cook, N.L. Lafferty, S.W.
Ž
.
Polichnowski, D.M. Pond, Catal. Today 13 1981 278.
x
34 A. Fulford, C.E. Hickey, P.M. Maitlis, J. Organomet. Chem.
Ž
.
398 1990 311.
x
35 K. Kujimoto, K. Omata, T. Shikada, H. Tominaga, in: D.R.
Ž
.
Fahey Ed. , Industrial Chemicals via C1 Processes, ACS
Ž
.
Symp. Ser., 328 1987 208.
w
w
x
x
Ž
.
36 H. Yagita, K. Fujimoto, J. Mol. Catal. 69 1991 191.
37 M.E. Halttunen, M.K. Niemela, A.O.I. Krause, A.I. Vuori, J.
¨
References
Ž
.
Mol. Catal. 109 1996 209.
w
w
x
38 R.B. Anderson, The Fischer–Tropsch Synthesis, Academic
w x
Ž
.
1
M. Ichikawa, Bull. Chem. Soc. Jpn. 51 1978 2268 and
2273.
Press, 1984.
x
39 D.C. Roe, R.E. Sherida, E.E. Bunel, J. Am. Chem. Soc. 116
w x
Ž
.
2
V. Ponec, Catal. Rev. Sci. Eng. 29 1987 183.
T. Nakajo, K. Sano, S. Matsuhira, H. Arakawa, J. Chem.
Ž
.
1994 1163.
w x
3
w
w
x
x
Ž
.
40 T.W. Decklewa, D. Forster, Adv. Catal. 34 1986 81.
41 P.R. Ellio, J.M. Pearson, A. Haynes, H. Adams, N.A. Bailey,
Ž
.
Soc., Chem. Comm. 1987 647.
w x
Ž .
4
R. Burch, M.I. Petch, Appl. Catal. A 88 1992 39.
Ž
.
P.M. Maitlis, Organometallics 13 1994 3215.
w x
Ž
.
5
R. Burch, M.I. Petch, Appl. Catal. A 88 1992 77.
H. Hamada, Y. Kuwahara, Y. Kintaichi, T. Ito, K. Wak-
w
w
x
42 J. Cressely, D. Farkhani, A. Deluzarche, A. Kiennemann,
w x
6
Ž
.
Mat. Chem. Phys. 11 1984 413.
Ž
.
abayashi, H. Iijima, K. Sano, Chem. Lett. 1984 1611.
x
43 H. Orita, S. Naito, K. Tamaru, J. Chem. Soc. Chem. Comm.
w x
Ž
.
7
Y. Sugi, K. Takeuchi, H. Arakawa, Chem. Lett. 1985 1315.
M.J. Howard, M.D. Jones, M.S. Roberts, S.A. Taylor, Catal.
Today 18 1993 325, and references quoted therein.
C.S. Garland, M.F. Giles, J.G. Sunley, Eur. Patent Appl. 0
643 034 A1 priority 10.09.1993 to BP Chemicals .
Ž
.
1984 150.
w x
8
w
w
x
x
Ž
.
44 H. Orita, S. Naito, K. Tamaru, J. Catal. 90 1984 183.
45 H. Orita, S. Naito, K. Tamaru, J. Phys. Chem. 89 1985
3066.
Ž
.
Ž
.
w x
9
Ž
.
w
x
10 B.L. Smith, G.P. Torrence, A. Aguilo, J.S. Alder, US Patent
Ž
.
5 001 259 priority 19.03.1991 to Hoechst Celanese .