Improved Rhodium Hydrogenation Catalysts Immobilized on Oxidic Supports

Article abstract of DOI:10.1002/adsc.200202197

Wilkinson-type rhodium hydrogenation catalysts immobilized on oxidic supports via mono-and bidentate phosphine linkers have been studied by ³¹P solid-state NMR, and their recycling stability and lifetime with respect to hydrogenation of 1-dodecene, 2-cyclohexen-1-one, and 4-bromostyrene have been improved substantially.

Full text of DOI:10.1002/adsc.200202197

COMMUNICATIONS
Improved Rhodium Hydrogenation Catalysts Immobilized on
Oxidic Supports
C. Merckle, J. Blümel*
Department of Organic Chemistry, University of Heidelberg, Im Neuenheimer Feld 270, 69120 Heidelberg, Germany
Phone: . ‡ 49)-6221-54-8470, Fax: . ‡ 49)-6221-54-4904, e-mail: J.Bluemel@urz.uni-heidelberg.de
Received: November 27, 2002; Accepted: January 19, 2003
In this contribution we will demonstrate that the
Abstract: Wilkinson-type rhodium hydrogenation
catalysts immobilized on oxidic supports via mono-
and bidentate phosphine linkers have been studied
by 31P solid-state NMR, and their recycling stability
and lifetime with respect to hydrogenation of 1-
dodecene, 2-cyclohexen-1-one, and 4-bromostyrene
have been improved substantially.
efficiency, stability and lifetimes of surface-bound
Wilkinson-type rhodium catalysts for olefin hydrogena-
tion, one of the most important and best developed
areas of catalysis,[15] can be improved substantially by
simple measures, namely the proper choice of pore
diameter and surface coverage of the support, as well as
using chelating ligands.
For the easy comparison of the hydrogenation char-
acteristics with our previous results,[6] and those of other
groups,[16] we used the standard set of substrates for the
catalytic runs, 1-dodecene .2), 2-cyclohexen-1-one .3),
and 4-bromostyrene .4). In the work presented here, all
the substrates 2 ± 4 were selectively hydrogenated, and
dodecane, cyclohexanone, and 4-bromoethylbenzene
were the only products .Scheme 1).
Keywords: alkene reduction; catalytic hydrogena-
tion; immobilization; P ligands; rhodium; silica;
solid-state NMR
The immobilization of homogeneous catalysts[1] via
The rate of conversion was determined by NMR and
linkers on supports is of growing importance because GC analysis. The complex ClRh[PPh2.CH2)3Si.OEt)3]3
it allows an easy separation of the catalyst from the .5) was synthesized starting from the precursor
reaction mixture as well as its recycling.[2,3] For some [.COD)RhCl]2, characterized and immobilized to give
time our group, like others, has pursued this idea by 5i as described previously.[6] The chelate complexes 6 ± 9
immobilizing homogeneous rhodium and nickel cata- .Scheme 2) were obtained by reaction of
lysts on inorganic oxidic supports via bifunctional Cl.PPh3)Rh.COD) with the corresponding chelating
phosphine linkers such as Ph2P.CH2)3Si.OEt)3 .1).[4±7]
The main disadvantage of immobilized catalysts has
always been leaching, the uncontrolled detachment of
the metal fragment from the support, with or without the
linker. Over time, we could identify and eliminate some
sources of leaching. First of all, one has to prevent the
oxidation of the phosphine by the combined action of
the silica surface and the ethoxysilane groups during the
immobilization step.[4] Then, the proper support materi-
al has to be chosen. In order to get strong covalent and
irreversible bonding of the ethoxysilane group, well-
dried silica[8,9] is best suited, while titania or magnesia
lead to persistent linker leaching.[10] Since monodentate
phosphines do not necessarily bind the metals to the
support in a chelating manner,[11] chelating phosphines
have been designed,[12] which should also provide a
stronger metal-ligand bonding.[7] These ligands[12] addi-
tionally allow one to study the dependence of catalytic
activity on their bite angles and spacer lengths. All the
amorphous surface-modified materials are best studied
with our improved analytical methods of optimized
Scheme 1. Hydrogenation of 2 ± 4 with homogeneous .6 ± 9)
and immobilized .6i ± 9i) catalysts.
classical solid-state NMR[13] and suspension NMR.[10,14]
584
ꢀ 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/adsc.200202197
Adv. Synth. Catal. 2003, 345, 584 ± 588

Rh Hydrogenation Catalysts Immobilized on Oxidic Supports
COMMUNICATIONS
Figure 2. Hydrogenation of 4-bromostyrene .4) with catalysts
6 ± 9 in toluene at 608C, with a catalyst to substrate ratio of
1:100.
phosphines.[12] Every unsymmetrical chelating ligand
leads to two inseparable isomers of the square-planar
complexes in a statistical ratio. However, as will be
elaborated in a future paper,[17] both isomers are equally
active hydrogenation catalysts, so that we can safely
apply the mixtures here. Immobilization of 6 ± 9 on silica
.see Experimental Section), using the standard proce-
dure[6] after their full characterization[18] gave the sur-
face-bound versions 6i ± 9i .Scheme 2).
Scheme 2. Immobilization of rhodium catalysts by chelating
linkers.
The success of all the immobilizations can be demon-
strated by 31P CP/MAS NMR[13] of the resulting
materials. The top spectrum in Figure 1 shows the
signals of 7i. Due to the more inhomogeneous surround-
ings of the phosphorus nuclei on the silica surface the
lines are broader than the ones in solution .7, middle
spectrum, Figure 1) or in the spectrum of the analogous
polycrystalline complex[19] Cl.PPh3)Rh.dppe) .bottom
spectrum). The signal intensities of the isotropic lines of
7i and of the polycrystalline complex are different from
the solution NMR signals of 7 in THF, because each
signal has its own chemical shift anisotropy[20] .CSA).
The larger the CSA, the more signal intensity is found in
the rotational sidebands at the expense of the isotropic
line. Nevertheless, it is obvious that no uncomplexed
phosphines[12] or oxidic species[4] are present.
As we could show previously, the efficiency of the
immobilized catalyst 5i is comparable to that of 5.[6] In
order to demonstrate that this is also true for all
substrates and for the rhodium catalysts with chelating
ligands, we compared the catalytic activities of the
catalysts 6 ± 9 with those of 5i and 6i ± 9i on silica. As
expected, the catalysts with chelating ligands need
slightly higher temperatures .608C) for the hydrogena-
tions than 5 or 5i .258C). Interestingly, as shown for
bromostyrene .4) and cyclohexenone .3) as the sub-
strates in Figures 2 and 3, the bite angles of the
phosphines or the spacer lengths do not make a crucial
difference with respect to the efficiency of the homoge-
neous hydrogenation, irrespective of the substrate. The
Figure 1. 162.0 MHz 31P CP/MAS spectra of amorphous 7i
.top, rotational frequency 13 kHz) and polycrystalline
Cl.PPh3)Rh.dppe) .bottom, spinning speed 15 kHz); spin-
ning sidebands are denoted by asterisks, other measurement
parameters, see ref.[13] Middle spectrum: 121.5 MHz
31P{1H} NMR spectrum of 7 in THF.
Adv. Synth. Catal. 2003, 345, 584 ± 588
asc.wiley-vch.de
ꢀ 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
585

C. Merckle, J. Blümel
COMMUNICATIONS
allowed the batch-wise recycling of the rhodium hydro-
genation catalysts for seven times,[6] with the catalysts
6i ± 9i 100% conversion is reached for all substrates in
every one of the 13 cycles. This is for example shown
with 6i on silica and 2 as the substrate in Figure 5.
Therefore, we conclude that our chelating ligands
generally, even for very different metal moieties,[7]
lead to diminished leaching and thus to improved
stabilities and lifetimes of immobilized catalysts.
Since it is known for rhodium catalysts that dimeriza-
tion leads to unreactive species,[21] we sought to improve
our catalysts further by diluting them on the silica
surface, and thus keep the reactive metal centers apart
from each other± one of the additional options that
immobilization offers in contrast to homogeneous
catalysis. Indeed, as can be seen in Figure 6 for 5i on
silica and 2, when the surface coverage is decreased
stepwise from the maximal coverage m to half the
surface coverage .m/2), then to m/4, and finally to m/10,
the turnover frequency .TOF) in a single run increases
substantially. The improvement is greatest on going
from m to m/2 and m/4, when the time requirement for
100% conversion decreases from 17 h to about 9 h .see
also Table 1). Diluting 5i further to m/10, however, only
saves roughly another half hour. Therefore, we conclude
that as soon as the active metal centers are far enough
apart from each other, a condition reached here at m/4,
no dimerization is possible any more and the maximal
lifetime of the catalyst with respect to this dimerization
deactivation mechanism is reached. Further dilution to
m/10 is unnecessary and only increases the bulk material
of the support, leading to ªdead reactor volumeº in
potential industrial applications. The same observation
is made for immobilized nickel catalysts,[7] and all our
catalysts with chelating ligands.[17]
Figure 3. Hydrogenation of cyclohexenone .3) with the
catalysts 6 ± 9 in toluene at 608C, with a catalyst to substrate
ratio of 1:100. Details of the experimental setting see ref.[6]
Figure 4. Hydrogenation of cyclohexenone .3) with the silica-
bound catalysts 6i ± 9i in toluene at 608C, with a catalyst to
substrate ratio of 1:100. Details of the experimental setting
see ref.[6]
The fact that the change of the surface coverage of the
complexes also affects the catalytic activity furthermore
corroborates the assumption that during the immobili-
difference in activity of the catalysts with various ligands zation step the metal centers are distributed evenly on
is even smaller in the case of 6i ± 9i .Figure 4). On going the surface, and do not settle down in patches. This is
from 6 ± 9 to the immobilized versions 6i ± 9i the reaction another proof that with rigorously dried silica .see
times increase to about two to three times the original above), which contains no more adsorbed water and
values, because now the substrates have to diffuse into only few isolated Si-OH groups,[22] no cross-linking of
the pores of the solid support .see also below). However, the ethoxysilane groups can take place,[8,9] and therefore
this disadvantage is made up for by the possibility of no ªclusteringº of the metal centers. Cross-checking this
recycling. All catalysts 6 ± 9 and 6i ± 9i give 100% result with silica, which was only dried at 258C and
conversion and optimal selectivity for all substrates 2 ± should therefore be wet enough to allow cross-linking,
4. Although every one of the catalysts 6i ± 9i is useful and leads to a loss of catalytic activity .Figure 6).
each has its merit for one special catalytic setting, overall
As long as the inorganic oxidic supports are dried
however, the catalysts 9 and 9i show generally very good rigorously, by dilution ofthe metal centers on thesurface
performance for all olefins, and might be tried first for one can also improve those catalysts immobilized via
different reactions by other groups.
phosphine linkers, for example, on alumina .Table 1).
All the chelating ligands improve the lifetimes of the Going from maximal surface coverage m of 5i to m/10
immobilized catalysts substantially, which was to be reduces the time required for hydrogenating 2 from 48 h
expected regarding the results of a parallel project with to about 10 h.
nickel catalysts.[7] While monodentate phosphine li-
The advantageous effect of diminished surface cover-
gands in the case of dodecene as the substrate only age is independent of the average pore diameter of the
586
ꢀ 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
Adv. Synth. Catal. 2003, 345, 584 ± 588

Rh Hydrogenation Catalysts Immobilized on Oxidic Supports
COMMUNICATIONS
Table 1. The effects of different supports and surface cov-
erages with 5i on the time required for quantitative conver-
sion of 2 to dodecane, and of 3 to cyclohexanone. Numbers in
brackets give the average pore diameters; #alumina with
neutral surface;[10] *silica dried at 258C under vacuum; m
denotes maximal surface coverage.
Support for 5i Surface
Time for 100% Conversion [h]
Coverage
1-Dodecene
2-Cyclohexen-
1-one
Al2O3 .90 Š)#
m
48.0
11.7
9.7
31.5
17.0
12.0
8.7
39.7
11.0
10.5
29.5
13.3
9.5
9.0
8.7
14.5
10.5
9.5
9.5
15.0
9.0
7.7
7.3
m/4
m/10
m
SiO2 .40 Š)*
SiO2 .40 Š)
m
Figure 5. Batch-wise recycling of the catalyst: hydrogenation
of dodecene .2) with 6i on silica at 608C in toluene with a
catalyst to substrate ratio of 1:100. Details of the exper-
imental setting see ref.[6]
m/2
m/4
m/10
m
m/2
m/4
m/10
m
8.3
SiO2 .60 Š)
SiO2 .100 Š)
17.3
11.7
7.7
7.3
18.0
13.0
5.0
m/2
m/4
m/10
4.3
shows convincingly that the effect of catalyst dilution on
the surface even overrides the advantages of chelate
bonding. With 5i quantitative conversion of 2 is obtained
already after 100 h in the 13th run, while 120 h are
needed with 6i immobilized with maximal density on the
support .Figure 5). Combining all the advantageous
effects of optimal support, surface coverage, and
chelating linkers will be the subject of further inves-
tigations.[17]
Figure 6. Hydrogenation of dodecene .2) b y5i on SiO2 .40
Š). m denotes maximal surface coverage .0.10 mmol 5 on 1 g
of SiO2); the asterisk denotes maximal surface coverage of 5i
on silica dried in vacuum at ambient temperature; catalyst to
substrate ratio 1:100; details of the experimental setting see
ref.[6]
support, as can be seen in Table 1 for values between 40
and 100 Š. Larger pores should allow better diffusion of
bulky substrates to the innermost catalytic centers.
Indeed, dodecene can be hydrogenated with 5i on SiO2
.100 Š) in only half the time required for 5i on SiO2 .40
Š) as the support. This result is also displayed for the
catalytic system 5i/2 in Figure 7. With more compact
substrates, such as 3 or 4, this effect is less pronounced.
For example, for 5i/3 the gained time difference is only
1.4 h .Table 1).
Since the linkers are bound covalently to the support,
and unlike merely physically adsorbed species[8,10] they
cannot migrate on the surface, the dilution of the metal
centers is persistent. Therefore, it is not surprising that
dilution also enhances the lifetime of 5i, when the
Figure 7. Hydrogenation of dodecene .2) b y5i on silica with
an average pore diameter of 100 Š in toluene. m denotes
maximal surface coverage .0.05 mmol 5 on 1 g of SiO2);
catalyst to substrate ratio 1:100; details of the experimental
catalyst is recycled in a batch-wise manner. Figure 8 setting see ref.[6]
Adv. Synth. Catal. 2003, 345, 584 ± 588
asc.wiley-vch.de
ꢀ 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
587

C. Merckle, J. Blümel
COMMUNICATIONS
Rambow and Mrs. A. Fischer for preparing countless samples
and supervising the catalytic runs.
References and Notes
[1] B. Cornils, W. A. Herrmann, Applied Homogeneous
Catalysis with Organometallic Compounds, VCH, New
York, 1996.
[2] a) F. R. Hartley, Supported Metal Complexes, D. Reidel
Publishing Co., Dordrecht, Holland, 1985, and literature
cited therein; b) J. H. Clark, Supported Reagents in
Organic Reactions, VCH Weinheim, 1994.
[3] D. E. DeVos, I. F. J. Vankelecom, P. A. Jacobs .Eds.),
Chiral Catalyst Immobilization and Recycling, Wiley,
VCH, Weinheim, 2000.
Figure 8. Batch-wise recycling of the catalyst. Hydrogenation
of dodecene .2) with 5i on SiO2 .40 Š) in toluene with half-
maximal surface coverage m/2 .0.05 mmol 5 on 1 g of SiO2)
and a catalyst to substrate ratio of 1:100; details of the
experimental setting see ref.[6]
[4] J. Blümel, Inorg. Chem. 1994, 33, 5050.
[5] K. D. Behringer, J. Blümel, Inorg. Chem. 1996, 35, 1814.
[6] C. Merckle, S. Haubrich, J. Blümel, J. Organomet. Chem.
2001, 627, 44.
[7] a) S. Reinhard, K. D. Behringer, J. Blümel, New J.
Ï
Chem., in press; b) S. Reinhard, P. Soba, F. Rominger, J.
Blümel, Adv. Synth. Catal. 2003, 345, 589.
[8] J. Blümel, J. Am. Chem. Soc. 1995, 117, 2112.
[9] K. D. Behringer, J. Blümel, J. Liquid Chromatogr. 1996,
19, 2753.
[10] C. Merckle, J. Blümel, Chem. Mater. 2001, 13, 3617.
[11] K. D. Behringer, J. Blümel, Chem. Commun. 1996, 653.
[12] G. Tsiavaliaris, S. Haubrich, C. Merckle, J. Blümel,
Synlett 2001, 391.
In conclusion, we have demonstrated that even simple
measures can improve the efficiency, stability, and
lifetime of immobilized catalysts. These results are of a
general nature with respect to different catalytic reac-
tions, metal centers, substrates, linkers, supports, surface
coverages and pore diameters. They clearly further
extend the utility of immobilized catalysts derived from
the covalent linkage of homogeneous metal complexes
to oxidic supports.
[13] S. Reinhard, J. Blümel, Magn. Reson. Chem., in press.
[14] K. D. Behringer, J. Blümel, Z. Naturforsch. 1995, 50b,
1723.
Â
[15] P. A. Chaloner, M. A. Esteruelas, F. Joo, L. A. Oro,
Experimental Section
Homogeneous Hydrogenation, Kluwer, Boston, 1994.
[16] D. Rutherford, J. J. J. Juliette, Ch. Rocaboy, I. T. Hor-
All catalytic reactions were performed under inert gas using
Schlenk techniques, prior to admitting hydrogen. If not stated
otherwise, silica with 40 Š average pore diameter, 750 m2/g
surface area, and a particle size of 0.063 to 0.2 mm was dried at
6008C under vacuum for 24 h prior to use and stored under N2.
All catalysts were immobilized according to the standard
procedure given in ref.[6] The details of the solid-state NMR
measurements are given in ref.[13]
Â
vath, J. A. Gladysz, Catalysis Today 1998, 1268, 1.
[17] C. Merckle, J. Blümel, in preparation.
1
[18] The compounds were fully characterized by H, 13C, and
31P NMR, HR-MS, and all the complexes gave satisfac-
tory elemental analyses; detailed data see refs.[6,17]
[19] Unfortunately all the rhodium complexes 6 ± 9 are oily
viscous liquids due to the alkyl spacers and ethoxysilane
groups and cannot be crystallized.
[20] T. M. Duncan, A Compilation of Chemical Shift Aniso-
tropies, Farragut Press, Chicago, 1990.
[21] U. Schubert, K. Rose, Transition Met. Chem. 1989, 14,
291.
[22] K. Iler, The Chemistry of Silica, John Wiley, New York,
1979.
Acknowledgements
We thank the Deutsche Forschungsgemeinschaft ꢀDFG,SFB
623) and the Fonds der Chemischen Industrie ꢀFCI) for
financial support. We are also grateful for the help of Mrs. D.
588
ꢀ 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
Adv. Synth. Catal. 2003, 345, 584 ± 588