Dormant states of rhodium hydroformylation catalysts: Carboalkoxyrhodium complex formed from enones in the alkene feed

Article abstract of DOI:10.1002/anie.200351884

Catalysts get food poisoning: Conversion of the active 1-alkene hydroformylation catalyst 1 with reactive impurities in the feed, such as dienes and enones, into unreactive rhodium intermediates leads to deactivation of 1. For example, 2 is much more reac

Full text of DOI:10.1002/anie.200351884

Angewandte
Chemie
Catalyst Deactivation
Dormant States of Rhodium Hydroformylation
Catalysts: Carboalkoxyrhodium Complex Formed
from Enones in the Alkene Feed**
Scheme 1. Compound 1 as a mixture of two trigonal-bipyramidal iso-
mers.
Edyta B. Walczuk, Paul C. J. Kamer, and
Piet W. N. M. van Leeuwen*
the two isomers.[7] HP-IR studies of the hydroformylation of
1-octene showed that, while 1-octene was present in the
reaction mixture, the main species observed was 1. Minor
decomposition of 1 was due to the formation of an inactive
The stability, deactivation, and regeneration of the catalyst
are, along with activity and selectivity, important issues in
homogeneous catalysis.[1] The turnover number (TON) of a
catalyst, a reflection of the catalyst stability, is one of the
pivotal parameters for implementation of a process. Deacti-
vation can occur in many ways, but sometimes deactivation is
only temporary and the catalyst activity is restored. For long-
term catalyst performance in the hydroformylation reaction,
avoiding deactivation of the catalyst caused by reactive
impurities in alkene feeds is of great importance. Such
impurities may trap active rhodium catalysts either in a
temporary or a permanent inactive state. Dienes and alkynes
are poisons for many catalytic processes involving alkenes.[1]
Since hydroformylation of dienes is much slower than that of
alkenes,[2] diene impurities might thus slow down 1-alkene
hydroformylation, if the resting state of the catalytic cycle of
the diene is appropriate.[3] Examples of such deactivation are
reported in patent literature.[4]
We have studied the deactivation of hydroformylation
catalysts by controlled reaction with the most likely impurities
in alkene feeds, such as dienes, alkynes, and enones. The
hydrido rhodium complex, the common resting state in
hydroformylation, at which the catalytic cycle starts, was
prepared in a high-pressure (HP)-IR autoclave under syn gas
(CO/H2), from the catalyst precursor, [Rh(acac)(CO)2]
(acac = 2,4-pentadione) and excess of PPh3 at 808C. After
dimer.[8] New bands, characteristic of the dimer, appeared in
À1
~
the IR spectrum at n = 2023, 1972, and 1957 cm .
Several linear dienes, 2,4-hexadiene, 1,5-hexadiene, and
1,3-pentadiene were tested as inhibitors of the catalyst. The
addition of 2,4-hexadiene to 1 formed in situ resulted in the
slow growth of two new bands in the IR spectrum (Table 1) at
=
Table 1: Influence of poisons/inhibitors on the Rh-C O bands of 1 under
syn gas.
À1
=
1 + poison/inhibitor
Rh-C O bands [cm
]
–1953, 1986, 1997, 2043
2,4-hexadiene
1967, 2016[a]
1,3-pentadiene
1-octyne
trans-3-nonen-2-one
3-buten-2-one (MVK, 2)
1949 (w), 1957 (s), 2016 (s)
1946, 1988, 2020, 2042, 2073, 2120[a]
1946 (w), 1984 (vs)
1946 (w), 1984 (vs)
[a] Weak bands from 1 also present.
the cost of the four absorption bands characteristic of 1, but
these four initial bands did not disappear completely. 1,5-
Hexadiene underwent fast hydroformylation and its addition
to the reaction mixture did not change the nature of the
resting state of the rhodium complex and it did not influence
the rate of 1-octene hydroformylation either.
Addition of the conjugated 1,3-pentadiene to 1, however,
resulted in the immediate formation of new bands in the IR
spectrum (Table 1). A 60-fold excess of diene gave complete
conversion of 1 and new bands were observed immediately
after addition of the diene (Figure 1). When the ratio of
diene:Rh is 20:1, the signals of 1 did not disappear completely.
After 10 min the concentration of 1 was about 50% of its
initial value. In both cases 1,3-pentadiene underwent hydro-
formylation to a mixture of aldehydes.
Addition of 1,3-cyclohexadiene or cyclopentadiene to 1
did not affect the IR spectrum. Both the cyclic dienes
underwent fast hydroformylation and new bands character-
istic of the corresponding unsaturated aldehydes appeared in
the IR spectra.
Internal and terminal alkynes proved to have completely
different effects on 1. Addition of 3-hexyne did not affect the
nature of 1, whereas the addition of 1-octyne immediately
caused irreversible formation of new rhodium species, as
indicated by several new bands (Table 1), which remained
present throughout the experiment (Figure 2). Furthermore,
complete
conversion
of
[Rh(acac)(CO)2]
into
[RhH(CO)2(PPh3)2] (1), a substrate or a mixture of substrates
was added to the reaction mixture from a separately
pressurized reservoir. We have used in situ high-pressure IR
spectroscopy in conjunction with (high-pressure) NMR
spectroscopy to detect and characterize the catalyst under
actual catalytic conditions.[5] Employing NMR spectroscopy
studies Brown and Kent have shown that 1 exists as a mixture
of two rapidly equilibrating trigonal-bipyramidal isomers in a
diequatorial (ee) to equatorial-apical (ea) isomer ratio of
85:15 (Scheme 1).[6]
~
The HP-IR spectrum of 1 has four carbonyl bands (at n =
2043, 1997, 1986, and 1953 cmÀ1) that have been assigned to
[*] Prof. Dr. P. W. N. M. van Leeuwen, Dr. E. B. Walczuk,
Dr. P. C. J. Kamer
Institute of Molecular Chemistry
University of Amsterdam
Nieuwe Achtergracht 166, 1018 WV Amsterdam (The Netherlands)
Fax: (+31)20-525-6456
E-mail: pwnm@science.uva.nl
[**] We are indebted to Sasol Technology Ltd. for financial support and
to Dr. M. J. Green and co-workers for stimulating discussions.
Angew. Chem. Int. Ed. 2003, 42, 4665 –4669
DOI: 10.1002/anie.200351884
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4665

Communications
Figure 1. In situ HP-IR study of hydroformylation of 1,3-pentadiene catalyzed by 1 (1,3-pentadiene:1=60:1).
the linear a,b unsaturated ketones. The branched unsaturated
ketone, mesityl oxide (4-methyl-3-penten-2-one), did not lead
to a decrease of the concentration of 1.
The large impact of 2 on the concentration of 1 prompted
us to perform a detailed study of the temporary deactivation.
A high initial concentration of 2 in the reaction mixture gave a
very slow initial hydroformylation of 1-octene as measured by
syn-gas consumption. At low concentration of 2, only minor
inhibition occurred and the normal rates of the hydroform-
ylation of 1-octene were observed (Figure 4).
Figure 2. In situ HP-IR study of hydroformylation of 1-octyne catalyzed
by 1; c spectrum of 1, a spectrum after the addition of 1-octyne.
1-octyne underwent slow hydroformylation to a mixture of
products.
Addition of unsaturated ketones, such as trans-3-nonen-2-
one and 3-buten-2-one (MVK, 2), to a solution of 1
immediately gave rise two new bands in the IR spectrum
(Table 1) and the disappearance of the four bands character-
istic of 1 (Figure 3).
Note that when most of 2 had reacted, 1 was regenerated
and became again the predominant species. The addition of 2-
cyclohexen-1-one to 1 had the same effect as the addition of
Figure 4. Comparison of the rates of hydroformylation of 1-octene and
mixtures of 1-octene and 2 catalyzed by 1.
Figure 3. In situ HP-IR study of the reaction of 2 with 1.
4666 ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2003, 42, 4665 –4669

Angewandte
Chemie
Similar competition experiments were carried out with
the other substrates 1-octyne, 1,3-pentadiene, cyclopenta-
diene, and 2-cyclohexen-1-one. The substrates 1-octyne, 2-
cyclohexen-1-one, and 2 changed the nature of the resting
state of the rhodium species. When these new rhodium species
were present in the reaction mixture, hydroformylation of 1-
octene stopped or was retarded. When most of the inhibitor
was converted and its concentration reached a certain lower
level, the hydrido complex 1 again became the predominant
species and hydroformylation of 1-octene resumed. Com-
pounds not changing the resting state (cyclopentadiene, 2-
cycloocten-1-one (not shown)) or that are converted very
quickly (1,3-pentadiene) did not influence the rate of the 1-
octene hydroformylation (Figure 5).
extremely unstable nature attempts to isolate the intermedi-
ates were not successful, but they could be identified in situ.
Their structure was unambiguously determined by NMR
spectroscopy.[9] Two h1-oxygen bound rhodium enolate com-
plexes (4; Scheme 2) were identified and characterized by IR
À1
~
spectroscopy they give a carbonyl band at n = 1968 cm . The
À1
~
two carbonyl bands at n = 1690 and 1719 cm characteristic
of 2 disappear.
When the reaction of 3 with 2 was complete and a mixture
of 4a and 4b was formed, CO/H2 or CO was bubbled through
the reaction mixture for 6 min at À608C. The reaction of 4
with CO was studied by NMR (1H, 31P{1H}, 13C{1H}, COSY,
31P–31P correlation, 31P–13C correlation and 31P–103Rh correla-
tion) and HP-IR spectroscopy.
1H and 31P{1H} spectra taken after the reaction showed
complete conversion of 4 and formation of new species (5;
Scheme 3) which we characterized using HP-IR and several
NMR spectroscopy techniques.
Scheme 3. Reaction of 4 with CO.
The 31P{1H} spectrum changed totally when CO was
present in the reaction mixture. The two doublets character-
istic of 4 disappeared and two new doublets of doublets at d =
26.15 and 28.53 ppm were observed. The data were confirmed
by simulation of the spectrum of 5 (Figure 6).[10]
Figure 5. Comparison of the rates of hydroformylation of 1-octene and
mixtures of 1-octene and other substrates catalyzed by 1.
HP 1H and 31P NMR spectroscopy was employed to study
the intermediate products. However, the reaction between 1
and 2 at room temperature was very fast giving a mixture
containing many new compounds, which we did not attempt
According to the 31P–31P correlation spectra these two
doublets of doublets belong to two inequivalent phosphorus
atoms bonded to the same rhodium(i) species. This result was
confirmed by 31P–103Rh correlation spectroscopy.
to identify. As
a
model catalyst we then used
1H NMR spectroscopy showed formation of two E/Z
isomers that differed in the vinyl part in a ratio of 30:70. The
E isomer (5a) gave three resonance signals at d = 0.8, 1.35,
and 2.6 ppm. Another set of three signals at d = 1.28, 1.50, and
4.15 ppm was assigned to the Z isomer (5b). The phenyl
groups of both isomers gave overlapping signals at d = 6.8–
7.8 ppm. The two isomers could not be distinguished by
31P NMR spectroscopy.
A more detailed examination of the structure of 5 was
carried out employing 13CO instead of 12CO, which was
bubbled through the solution of the rhodium enolate com-
plexes 4. The 13C{1H} spectrum taken at À408C showed one
[RhH(CO)(PPh3)3] (3) in the reaction with 2 in a ratio of
1:1.2. The reaction was initially monitored at À808C and then
the reaction mixture was gradually heated to À208C while 1H
and 31P NMR spectra were recorded. Subsequently, the NMR
tube was cooled to À608C and 1H, COSY, 31P{1H}, and 13C{1H}
NMR spectra were recorded. 31P NMR showed complete
conversion of 3 and two new doublets at d = 26.74 and
29.99 ppm and a signal assigned to one equivalent of free PPh3
at d = À6.26 ppm were observed. This result indicates for-
mation of two isomeric insertion products, both containing
two equivalent triphenylphosphane units. Because of their
Scheme 2. Reaction of 3 with 2 under argon.
Angew. Chem. Int. Ed. 2003, 42, 4665 –4669
www.angewandte.org
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4667

Communications
upon carbonylation of four-coordinate iridium alkoxides,
have been reported.[12]
The formation of 5 is the cause for slow or no hydro-
formylation of 1-octene when 2 is present in the reaction
mixture. The unsaturated ketone is much more reactive
towards 1 than the alkene and forms carboalkoxy complexes
with rhodium thus blocking its activity for the 1-octene
hydroformylation. Catalyst 1 for the hydroformylation of 1-
octene is only restored after the enone has been converted
into 2-butanone. The formation of formate esters from the
enone substrate was not detected by GC-MS and HP-NMR
spectroscopy. The analysis of the reaction mixture confirmed
that the only organic product present after the reaction was 2-
butanone. The rhodium enolate complexes react very fast
with CO leading to the formation of 5, which does not react
with H2 as formate formation is apparently very slow. Instead,
a de-insertion has to occur to allow for hydrogenation of 4,
Figure 6. Top : 31P{1H} spectrum obtained after bubbling CO/H2
through the solution containing the rhodium enolate complexes 4,
=
O PPh ; bottom: simulation of this spectrum.
*
3
resonance for an acyl group at
235.5 ppm and one resonance for
rhodium-bound carbonyl group at
198 ppm. Coupling of the acyl
carbon with the rhodium (J(Rh,C) =
22 Hz) and phosphorus nuclei
(J(P,C) = 80 Hz) was observed. The
terminal carbonyl groups are coupled
to the rhodium nucleus (J(Rh,C) =
76 Hz). The 31P–13C correlation spec-
trum showed strong cross-peaks for
the two double doublets occurring in
the 31P{1H} NMR spectrum and for
the signal of a carbonyl ligand at
198 ppm and that of a rhodium-acyl
resonance at 235.5 ppm. Therefore,
we conclude that the two signals
present in the 13C{1H} NMR spec-
trum and two double doublets in the
31P{1H} NMR spectrum belong to the
same carboalkoxyrhodium complex,
which exists as a mixture of E/Z
isomers.
Scheme 4. Deactivation mechanism of the rhodium hydride complex by 2.
The reaction of 4 with CO was also performed in the HP-
IR autoclave. When 2 was added to a cyclohexane solution of
which results in butanone formation. In equilibrium, during a
catalytic reaction, the concentration of 4 is too low to detect
by HP-IR spectroscopy. When most of 2 has been converted
into butanone, the hydroformylation of 1-octene can proceed.
(Scheme 4).
3 the bands in the IR spectrum characteristic of the rhodium
À1
~
hydride complex (n = 2011 and 1938 cm ) disappeared and a
À1
~
new band characteristic of 4 appeared at n = 1968 cm . Then
CO/H2 or CO was added to the reaction mixture and the latter
In conclusion unsaturated impurities in the alkene feed,
such as enones, dienes, and alkynes, deactivate the hydro-
formylation catalyst. Enones transform the “active” 1 tem-
porarily into an inactive dormant state by formation of
carboalkoxyrhodium complex. The hydroformylation catalyst
is reformed by hydrogenation, which explains why the
inhibiting effect can be reduced by high H2 pressure.[3]
À1
~
band disappeared and new bands at n = 1984 and 1946 cm
were observed, indicative of coordination of two carbonyl
ligands. The positions of these bands in the IR spectrum were
exactly the same as those observed after addition of 2 to the
solution of 1 under CO/H2 (Figure 3).
The NMR and IR spectroscopy studies have shown that
addition of CO to the solution of 4 results in coordination of a
CO molecule to the rhodium center and insertion of a second
one to the rhodium–oxygen bond forming five coordinate
rhodium(i) complexes (5; Scheme 4).[6,11] Structurally analo-
gous carboalkoxy complexes of iridium, which are formed
Experimental Section
High-pressure IR experiments were performed in an SS-316 50-mL
autoclave equipped with IRTRAN windows (ZnS, transparent above
4668
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2003, 42, 4665 –4669

Angewandte
Chemie
700 cmÀ1, 1 = 10 mm, optical path length = 0.4 mm), a mechanical
stirrer, a temperature controller, and a pressure device. In a typical
experiment, [Rh(CO)2(acac)] (0.008 g, 3.0 ” 10À5 mol) and PPh3 (6.0 ”
10À4 mol) of were dissolved in cyclohexane (15 mL) under argon. The
solution was brought into the autoclave and after flushing and
pressurizing with CO/H2, the HP-IR cell was placed into a Nicolet 510
FTIR spectrometer. Then the reaction mixture was heated to 808C. A
substrate or a mixture of substrates were added to the reaction
mixture from a separately pressurized reservoir by means of over-
pressure once the active catalyst, [RhH(CO)2(PPh3)2] (1), was
formed. The IR spectra were recorded while the samples were stirred.
4: A solution of [RhH(CO)(PPh3)3] (3; 0.05 g, 5.4 ” 10À5 mol;
Strem) in [D8]toluene was prepared in a 5-mm NMR tube under
argon. The tube was placed in a NMR machine and cooled to À608C.
The NMR tube was then ejected and 3-butene-2-one (2, 5.5 mL, 6.5 ”
10À5 mol) was added and the tube placed back in the precooled NMR
[7] a) 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, C. Bo, J. Am. Chem. Soc. 1998, 120, 11616; b) A.
Castellanos-Pꢂez, S. Castillꢁn, C. Claver, P. W. N. M. van Leeu-
wen, W. G. J. de Lange, Organometallics 1998, 17, 2543.
[8] a) D. Evans, G. Yagupsky, G. Wilkinson, J. Chem. Soc. A 1968,
2660; b) A. S. C. Chan, H.-S. Shieh, J. R. Hill, J. Chem. Soc.
Chem. Commun. 1983, 688.
[9] a) G. A. Slough, R. G. Bergman, C. H. Heathcock, J. Am. Chem.
Soc. 1989, 111, 938; b) G. A. Slough, R. Hayashi, J. R. Ashbaugh,
S. L. Shamblin, A. M. Aukamp, Organometallics 1994, 13, 890;
c) G. A. Slough, J. R. Ashbaugh, L. A. Zannoni, Organometal-
lics 1994, 13, 3587; d) S. Sato, I. Matsuda, Y. Izumi, Chem. Lett.
1985, 1875.
[10] The spectra were simulated using gNMR4.1; P. H. M. Budzelaar,
Ivory Soft, Cherwell Scientific Publishers, Oxford, 1999.
[11] a) J. Zhang, M. Poliakoff, M. W. George, Organometallics 2003,
22, 1612; b) K. Fagnou, M. Lautens, Chem. Rev. 2003, 103, 169;
c) S. C. van der Slot, P. C. J. Kamer, P. W. N. M. van Leeuwen,
J. A. Iggo, B. T. Heaton, Organometallics 2001, 20, 430.
[12] a) W. M. Ress, J. D. Atwood, Organometallics 1985, 4, 402;
b) W. M. Ress, M. R. Churchil, Y.-J. Li, J. D. Atwood, Organo-
metallics 1985, 4, 1162; c) W. M. Ress, M. R. Churchil, J. C.
Fettinger, J. D. Atwood, Organometallics 1985, 4, 2179; d) M. R.
Churchil, J. C. Fettinger, W. M. Ress, J. D. Atwood, J. Organo-
met. Chem. 1986, 304, 227.
machine and the NMR spectra were recorded. E isomer (4a):
1
=
H NMR (300 MHz, C7D8, À608C): d = 1.49 (s, C(CH3), 3H), 2.15
=
=
(d, J(H,H) = 6 Hz, 3H, CH(CH3)), 3.97 (q, J(H,H) = 6 Hz, 1H,
CH(CH3)), 6.8–7.5 ppm (m, 30H, PPh3); 31P{1H} NMR (121.5 MHz,
C7D8, À608C): d = 26.74 ppm (d, J(Rh,P) = 145.00 Hz); 13C{1H} NMR
(75.4 MHz, C7D8, À608C): d = 14.95 (s, CH3), 29.21 (s, CH3), 92.15 (s,
=
1
CH), 159.03 (s, Rh-O-C), 191.46 ppm (dt, Rh-CO); Z isomer (4b):
H NMR (300 MHz, C7D8, À608C): d = 1.36 (s, 3H, C(CH3)), 1.53
=
=
=
(d, J(H,H) = 6 Hz, 3H, CH(CH3)), 4.63 (q, J(H,H) = 6 Hz, 1H,
CH(CH3)), 6.8–7.5 ppm (m, 30H, PPh3); 31P{1H} NMR (121.5 MHz,
C7D8, À608C): d = 29.99 ppm (d, J(Rh,P) = 146.25 Hz); 13C{1H} NMR
(75.4 MHz, C7D8, À608C): d = 14.29 (s, CH3), 28.95 (s, CH3), 88.17 (s,
=
CH), 160.36 (s, Rh-O-C), 191.46 ppm (dt, Rh-CO); 4a and 4b: IR
(C5H12): 1968 cmÀ1 (s).
5: First a mixture of 4 was prepared in a 5-mm NMR tube
according to the description given above. Then CO or CO/H2 was
bubbled through the solution for 6 min at À608C and the NMR
spectra were recorded. E isomer (5a): 1H NMR (300 MHz, C7D8,
=
=
À608C): d = 0.8 (d, J(H,H) = 7.2 Hz, 3H, CH(CH3)), 1.35 (s, 3H,
=
C(CH3)), 2.6 (q, J(H,H) = 7.2 Hz, 1H, CH(CH3)), 6.8–7.8 ppm (m,
1
30H, PPh3); Z isomer (5b): H NMR (300 MHz, C7D8, À608C): d =
=
=
1.28 (d, J(H,H) = 6.9 Hz, 3H, CH(CH3)), 1.50 (s, 3H, C(CH3)),
=
4.15 (q, J(H,H) = 6.9 Hz, 1H, CH(CH3)), 6.8–7.8 ppm (m, 30H,
PPh3); 5a and 5b: 31P{1H} NMR (121.5 MHz, C7D8, À608C): d = 26.15
(dd, J(Rh,P) = 132 Hz, J(P,P) = 30 Hz), 28.53 ppm (dd, J(Rh,P) =
72 Hz, J(P,P) = 30 Hz); 13C{1H} NMR (75.4 MHz, C7D8, À408C):
d = 198 (d, J(Rh,C) = 76 Hz, Rh-CO), d = 235.5 ppm (dd, J(Rh,C) =
22 Hz, J(P,C) = 80 Hz, Rh-COO); IR (C5H12): 1984 (vs), 1946 (w),
1652 cmÀ1 (w).
Received: May 13, 2003 [Z51884]
Keywords: high-pressure spectroscopy · homogeneous
.
catalysis · hydroformylation · reaction mechanisms · rhodium
[1] P. W. N. M. van Leeuwen, Appl. Catal. A 2001, 212, 61.
[2] a) P. W. N. M. van Leeuwen, C. F. Roobeek, J. Mol. Catal. 1985,
31, 345; b) A. van Rooy, J. N. H. de Bruijn, C. F. Roobeek,
P. C. J. Kamer, P. W. N. M. van Leeuwen, J. Organomet. Chem.
1996, 507, 69.
[3] K. F. Muilwijk, P. C. J. Kamer, P. W. N. M. van Leeuwen, J. Am.
Oil Chem. Soc. 1997, 74, 223.
[4] E. Billig, A. G. Abatjoglou, D. R. Bryant, R. E. Murray, J. M.
Maher (Union Carbide Corporation), US Pat. 4,717,775, 1988;
[Chem. Abstr. 1989, 109, 233177].
[5] M. Diꢀguez, C. Claver, A. M. Masdeu-Bultꢁ, A. Ruiz,
P. W. N. M. van Leeuwen, G. C. Schoemaker, Organometallics
1999, 18, 2107.
[6] J.M. Brown, A. G. Kent, J. Chem. Soc. Perkin Trans. 2 1987,
1597.
Angew. Chem. Int. Ed. 2003, 42, 4665 –4669
www.angewandte.org
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 4669