Reversible immobilization of a molecular catalyst and challenges of catalyst characterization

Article abstract of DOI:10.1002/chem.200802389

The immobilization of molecular catalysts on support materials enables the combination of easy separation of solid catalysts with the high activity and selectivity of molecular catalysts. We have been able to apply a cross-metathesis reaction between term

Full text of DOI:10.1002/chem.200802389

FULL PAPER
DOI: 10.1002/chem.200802389
Reversible Immobilization of a Molecular Catalyst and Challenges of
Catalyst Characterization
R. Palkovits,[a] D. Arlt,[b] H. Stepowska,[b] and F. Schꢀth*[a]
Abstract: The immobilization of mo-
lecular catalysts on support materials
enables the combination of easy sepa-
and complexes on a support material.
Diphosphine ligands and corresponding
complexes can be covalently immobi-
lized and then split again from the sup-
port material after use in catalytic reac-
tions, and subsequently be reused by
repeated immobilization. The method
is widely applicable and offers, there-
fore, an interesting approach to a cova-
lent, but reversible immobilization of
various molecular catalysts.
ration of solid ca
activity and selectivity of molecular ca-
talysts. We have been able to apply a
ACHTUNGTRENNUNGtalACHTUGNTRENyNUGN sts with the high
A
ACHTUNGTRENNUNG
Keywords: biphep · heterogeneous
catalysis · hydrogenation · immobi-
lization · metathesis
cross-metathesis reaction between ter-
minal olefins, as versatile tool with
which to reversibly immobilize ligands
Introduction
lysts. Consequently, numerous studies on the immobilization
of molecular catalysts exist. In addition to immobilization,
adsorption effects, ionic interactions and entrapment meth-
ods have been applied to support molecular catalysts. Over-
views about these methods can be found elsewhere.[3–8] Re-
cently, many studies have focused on the covalent immobili-
zation of molecular catalysts on ordered mesoporous sup-
port materials, such as MCM-41 and SBA-15.[9,10] The appli-
cation of these highly ordered support materials is owed to
the idea of a preferred orientation between substrate and
pore system, in addition, the regular pore system facilitates
the analysis of the system, which makes ordered mesoporous
materials ideal model systems. The confinement in the pore
system is thought to induce additional geometric restrictions
and influence selectivity and enantioselectivity.[11] Other
than the mentioned effect on selectivity, such ordered mate-
rials exhibit a uniform pore system with narrow pore size
distribution. They can therefore provide uniform local envi-
ronments for the immobilized species concerning surface
curvature, surrounding surface sides and pore geometry. Es-
pecially in the field of enantioselective hydrogenation reac-
tions several studies report on the covalent immobilization
of molecular catalysts on ordered mesoporous materials.[12,13]
The aim of most investigations on covalent immobilizations
is a permanent anchoring of the molecular catalyst on the
support material. Nevertheless, most systems deactivate and
require regeneration after a certain time in use. In such cat-
alyst systems, the chiral ligand is by far the most expensive
component, and this is thus the most important part of the
system to be recovered.
Synthesis methods for sustainable chemistry are of growing
interest in chemical industry. Especially, homogeneously cat-
alyzed reactions that allow direct access to enantiopure
products[1] with high selectivities and, for some reactions,
close to 100% atom efficiency. Despite various research ac-
tivities in the field of homogeneously catalyzed reactions,
only a few systems are used on an industrial scale, mainly
owing to difficulties in separation and recycling of costly
chiral ligands. Asymmetric hydrogenation of different ke-
tones is of special industrial interest, as it allows direct
access to important precursors for fine and specialty chemi-
cals for pharmaceutical and agrochemical applications .[1]
Consequently, various research activities focus on new and
further developments of molecular catalysts.[2] The immobili-
zation of complexes on support materials presents an inter-
esting approach combining the easy separation of solid cata-
lysts with the high activity and selectivity of molecular cata-
[a] Dr. R. Palkovits, Prof. Dr. F. Schꢀth
Max-Planck-Institut fꢀr Kohlenforschung
Kaiser-Wilhelm-Platz 1, 45470 Mꢀlheim an der Ruhr (Germany)
Fax : (+49)208-306-2995
E-mail: schueth@mpi-muelheim.mpg.de
[b] Dr. D. Arlt, Dr. H. Stepowska
Ligand Chemie GmbH
Papenhauser Strasse 10, 32657 Lemgo (Germany)
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.200802389.
Chem. Eur. J. 2009, 15, 9183 – 9190
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
9183

We therefore studied a concept to covalently immobilize
molecular catalysts through a reversible anchoring reaction.
If successful, the molecular catalysts could be removed from
the support after use, re-purified and subsequently be re-im-
mobilized. In this approach, we investigated the cross-meta-
thesis reaction between alkene-chloroACTHNUTRGENNGUsilACHUTGTNRENNaGU nes and molecular
catalysts bearing a terminal olefin group as a tool with
which to reversibly immobilize molecular catalysts. In this
reaction, an alkene-modified ligand and an alkenyl-chloro-
silane are reacted through a cross-metathesis reaction and
the reaction product is immobilized on a silica support.
Since cross-metathesis reactions are reversible, a back-meta-
thesis reaction becomes possible to cut-off the molecular
catalyst from the support after use. Accordingly, the molecu-
lar catalyst could be recycled and immobilized on fresh sup-
port material. The only prerequisite for this approach is the
presence of olefin groups. Hence, it can widely be used to
reversibly immobilize various types of ligand molecules or
complexes.
In our studies, we have been able to apply this method to
immobilize a hexenoxy-modified phosphine ligand through
a cross-metathesis reaction with suitable chlorosilane mole-
cules serving as anchoring groups. The phosphine ligands
have been immobilized in the oxidized form and could suc-
cessfully be reduced without any leaching. The correspond-
ing molecular catalysts were formed by adding metal precur-
sor and further ligands. The immobilized molecular catalysts
reached high catalytic activity in the hydrogenation reaction
of acetophenone to phenylethanol compared to the molecu-
lar catalyst in the homogeneous form and compared to cata-
lytic activities reported in literature.[14] Additionally, the mo-
lecular catalysts could be cut off the support material after
use by means of another cross-metathesis reaction. Subse-
quently, the cut-off catalyst was immobilized on fresh sup-
port material.
Figure 1. Nitrogen sorption isotherm of SBA-15 and the corresponding
DFT pore distribution analysis.
uptake at low relative pressure, which can be assigned to
filling of micropores in the material. At higher relative pres-
sure, around 0.8 p/p0, the nitrogen uptake increases, corre-
sponding to filling of the mesopores. Desorption appears at
lower relative pressure resulting in a hysteresis. The steep
course of the desorption branch indicates a narrow pore size
distribution of the mesopores. Assuming cylindrical pores,
an average mesopore diameter of 9.4 nm can be determined
from the position of the desorption branch. The applied
SBA-15 material exhibits
a specific surface area of
743 m2 gÀ1 with a total pore volume of 1.5 cm3 gÀ1 of which
0.2 cm3 gÀ1 are located within the micropores.
The covalent immobilization requires anchoring of func-
tional groups on the surface of the support material. In the
case of silica materials, the anchoring is accomplished by
surface silanol groups. Solid-state 29Si NMR measurements
allow determination of the amount of silanol groups in the
material. The method however, is a bulk analysis method;
therefore 29Si NMR yields the total amount of silanol groups
in the material including not only the silanol groups accessi-
ble on the surface, but also those hidden inside the walls of
the material. To distinguish between surface and in-wall sila-
nol groups, the silica surface has been reacted with trime-
thylchlorosilane (TMCS). In the course of this reaction most
surface silanol groups available on the surface of the materi-
al are consumed. Analysis of 29Si NMR spectra recorded
before and after reaction with TMCS yields type and con-
centration of silanol groups in the material. Further on,
comparison of both spectra enables us to conclude on the
amount of surface silanol groups versus those hidden inside
pore walls, which are consequently not accessible as anchor-
ing points for immobilization reactions.[15] Some investiga-
tions have shown that other silylation agents are more effi-
cient than TMCS and reach a higher coverage of the surface
silanol groups.[16] Nevertheless, TMCS resembles the func-
tional silane that was used later in the immobilization step,
and thus allows a realistic assessment of the number of sur-
face silanol groups accessible for immobilization.
Results and Discussion
The ordered mesoporous silica material SBA-15 has been
chosen as a support material. It offers a two-dimensional
hexagonal arrangement of mesopores, each surrounded by a
corona of micropores. The choice of SBA-15 was governed
by two considerations: 1) The material provides high surface
area with a high concentration of reactive silanol groups,
and 2) the regular structure and porosity substantially facili-
tates analysis by sorption, diffraction methods and electron
microscopy. Additionally, a great number of NMR studies
have been performed with SBA-15, which provides a good
benchmark. This point of facile characterization is signifi-
cant, since the comprehensive characterization of the mate-
rial is important considering the complex task of a reversible
but covalent immobilization.
The nitrogen sorption isotherm of the starting material
and the results of the corresponding DFT analysis for the
determination of pore volume and diameter are presented
in Figure 1. The nitrogen sorption isotherm exhibits an
The 29Si NMR spectra of the starting material and the ma-
terial after reaction with TMCS is presented in Figure 2. For
9184
www.chemeurj.org
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 9183 – 9190

Reversible Immobilization of a Molecular Catalyst
FULL PAPER
system. One has to stress at this point that, although direct
immobilization without previous isolation of the ligand from
the product mixture is quite convenient, also remaining un-
reacted hexenyldimethylchlorosilanes will be immobilized.
Additionally, contamination of the material with Grubbs
catalyst and therefore with Ru metal can not be excluded
when the material is not washed thoroughly. XPS investiga-
tion of the materials did not indicate the presence of Ru
after the immobilization step (Figure S2 in the Supporting
Information). Nevertheless, elemental analysis reveals that
ruthenium levels around 0.5 wt.%, with appreciable scatter,
were observed.
In fact, different analytical techniques are necessary to
verify that the reaction proceeds in the described manner.
Solid-state 31P NMR was measured to confirm the presence
of the ligand system on the support material after the immo-
bilization reaction. The 31P NMR spectrum (Figure 3) dis-
Figure 2. 29Si MAS NMR spectra of blank (top) and trimethylsilylated
(bottom) SBA-15 (number of scans: 3200; spinning rate: 12000).
the starting material, three distinct resonances at À88, À98
and À109 ppm are shown, corresponding to Q2 (SiO(OH)2),
Q3 (SiO3/2OH) and Q4 (SiO2) type silicon atoms, respective-
ly. The 29Si NMR spectrum of the material after reaction
with the TMCS confirms the consumption of surface silanol
groups and reveals a decreased intensity of Q2 and Q3 group
signals and an increase of the Q4 signal intensity. The addi-
tional resonance at 18 ppm can be assigned to newly intro-
duced silicon atoms in the trimethylsilyl groups. Quantita-
tive analysis of the 29Si NMR spectra based on the differ-
ence between Q2 and Q3 group concentration before and
after surface modification shows that 24% of the silicon
atoms are present as surface silanol groups. Hence, these si-
lanol groups are available as anchoring groups for immobili-
zation reactions.
Figure 3. Solid-state 31P MAS NMR spectrum of the phosphine oxide
ligand immobilized on SBA-15 (number of scans: 4000, spinning rate
12000).
To accomplish the immobilization of a molecular catalyst,
5-hexenyldimethylchlorosilane (HMCS) was reacted with a
5,5’-dichloro,6,6’-dihexenoxybiphenyl-2,2’-diyl)bis(diphenyl-
phosphine oxide) (biphepO) type ligand modified with two
hexenoxy groups instead of the two methoxy groups. Ligand
structure and reaction are shown in Scheme 1. The ligand
and HMCS were reacted in a cross-metathesis reaction uti-
lizing a 2nd generation Grubbs catalyst.[17] Subsequently, the
product mixture was reacted with surface silanol groups of
the support material without previous isolation, resulting in
a covalent bond between support material and ligand
plays one distinct signal at 30 ppm with spinning side bands
at À30 and 90 ppm, in agreement with the chemical shift
measured in the liquid-phase 31P NMR prior to immobiliza-
tion. Accordingly, the ligand is present on the support mate-
rial after the immobilization reaction.
Furthermore, the successful cross-metathesis reaction be-
tween HMCS and ligand system had to be proven. For this
purpose, solid-state 13C NMR spectra have been recorded
(Figure 4). Cross-metathesis reactions of terminal olefins
result in the formation of in-chain double bonds with a
chemical shift of around
135 ppm.
Unfortunately,
carbon atoms from the aromat-
ic rings of the ligand have
chemical shifts in the same
range and their contribution to
the pronounced signal at
135 ppm is difficult to estimate.
Nevertheless, the low intensity
of the resonances at 115 and
140 ppm, corresponding to ter-
minal double-bond carbons,
points towards
a successful
cross-metathesis reaction be-
tween the ligand and chloro-
AHCTUNGTRENGNsUN ilane.
Scheme 1. The ligand immobilization procedure by means of a cross-metathesis reaction between the ligand
and hexenylchlorosilane followed by reaction of chlorosilane with the surface silanol groups of the support
material.
Chem. Eur. J. 2009, 15, 9183 – 9190
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
9185

F. Schꢀth et al.
both NMR techniques enables verification of the results, as
a higher concentration of phosphorus species on the support
than covalently attached silanes would clearly point towards
adsorption of ligands instead of anchoring through the pro-
posed reaction pathway. From quantitative analysis of the
29Si NMR spectrum (Figure 5) follows that 11% immobi-
lized silicon species are present corresponding to about 1ꢂ
10À1 mmolgÀ1. Quantification of the 31P NMR spectrum re-
veals that the material contains 5.7 mass% immobilized
phosphine
ligand,
corresponding
to
about
6.4ꢂ
10À2 mmolgÀ1. From this it can be calculated that 7% of the
immobilized silicon species bear a ligand. Considering the
bulkiness of the ligand molecules, the reached loading ap-
pears to be realistic for a covalent anchoring.
Figure 4. Solid-state 13C CP/MAS NMR spectrum of the phosphine oxide
ligand immobilized on SBA-15 with assignment of the resonances to the
corresponding carbon atoms (number of scans: 36000, spinning rate:
10000).
Immobilizing phosphine ligands on metal oxide supports
often results in oxidation of the ligands during immobiliza-
tion. We therefore directly immobilized the ligand in the
oxidized form. Consequently, transfer of a suitable reduction
method from the homogenous system to the immobilized
version is necessary, so we adapted methods known for the
reduction of phosphine oxide ligands in the liquid-phase.[18]
To avoid side reactions between the phosphine ligand and
the silica support, as described by Blꢀmel et al.,[19,20] the re-
maining surface silanol groups have been trimethylsilylated
prior to the reduction step by using trimethylsilylimidazole
(TMSI). The 31P NMR spectrum after reduction of the
ligand is presented in Figure 6. The resonance at 30 ppm
To finally conclude on the covalent nature of the immobi-
lization, 29Si NMR measurements have been performed. The
29Si NMR spectrum of the silica material with the immobi-
lized ligand (Figure 5) exhibits a clear resonance at 16 ppm.
Figure 5. Solid-state 29Si MAS NMR spectrum of the phosphine oxide
ligand immobilized on SBA-15 (number of scans: 2400, spinning rate:
10000).
This signal can be assigned to silicon atoms from anchored
hexenyldimethylsilyl groups previously reacted with the
ligand and confirms the covalent nature of the immobiliza-
tion. Signals at À88, À98 and À109 ppm are related to Q2,
Q3 and Q4 type silicon atoms. Corresponding to the con-
sumption of Q2 and Q3 silicon atoms in the immobilization
reaction, their signal intensity is reduced compared to the
blank material (Figure 2).
To close the line of argument, we performed a cross-
check. As stated before, solid-state 31P NMR as well as
29Si NMR can be used for quantitative analysis. Quantifica-
tion of the immobilized species has been carried out apply-
ing both techniques. 31P NMR allows quantification of the
concentration of phosphorus species on the support materi-
al, whereas 29Si NMR allows for the determination of the
concentration of immobilized silane species. The combina-
tion of both results in a conclusion on the percentage of
silane species bearing a ligand, which are the ones that have
been reacted successfully in the cross-metathesis reaction
prior to immobilization. Additionally, the combination of
Figure 6. Solid-state 31P MAS NMR spectra of the immobilized phos-
phine ligands before (top) and after (bottom) the reduction (number of
scans: 4000, spinning rate: 12000).
almost vanished together with the corresponding spinning
side bands. At the same time, a new resonance at À18 ppm
appeared, which is in good agreement with the chemical
shifts reported in literature for this type of ligand.[18] Quan-
tification shows that more than 95% of the phosphorus
atoms were reduced. The high reduction yield proves, on
one hand, the accessibility of the ligands to substrates and at
the same time emphasizes again the truly covalent nature of
the immobilization, as no change in the overall phosphorus
concentration could be observed (determined by 31P NMR).
The nitrogen sorption isotherm taken after the reduction
step excludes pore blocking effects and emphasizes that the
pore system is still accessible (Figure 7). Slight decreases in
total pore volume and average mesopore diameter com-
9186
www.chemeurj.org
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 9183 – 9190

Reversible Immobilization of a Molecular Catalyst
FULL PAPER
Figure 7. Nitrogen sorption isotherm of the SBA-15 after immobilization
and reduction of the phosphine ligand.
pared to the blank SBA-15 support can be assigned either to
the required space for the immobilized molecules within the
material or to changes in the adsorption properties of the
surface modified material.
Figure 9. Solid-state 31P MAS NMR spectrum of the immobilized catalyst
after reaction with the ruthenium precursor and Dpen (number of scans:
4000, spinning rate 14000).
13C NMR spectra were recorded to control the integrity of
the ligand structure after the reduction treatment. A com-
parison of 13C NMR spectra of the material just after immo-
bilization of the ligand and after the reduction is shown in
Figure 8. Both spectra are consistent with each other and
confirm the integrity of the immobilized system even after
appears at 48 ppm, which can be attributed to phosphorus
species in the complex. The signal at 26 ppm however, con-
tains two different species. On one hand phosphorus atoms
are oxidized during reaction with the metal precursor and
Dpen, on the other hand phosphorus atoms react with the
support to create P-O-Si type bonds with the surface silanol
groups, as described by Bluemel et al..[19,20] Small resonances
at 115, 95 and À45 ppm can be assigned to spinning side
bands of the main resonances. Quantification reveals that
about 30% of the phosphorus species could be reacted with
metal precursor and Dpen to form the target complex corre-
sponding to a concentration of 1.9ꢂ10À2 mmolgÀ1 molecular
catalyst on the support material. This concentration corre-
sponds to about 1.9 mggÀ1 ruthenium in the material. Owing
to the background contamination with ruthenium from the
metathesis step and the different intermediate steps, which
could change the contamination level, the results of elemen-
tal analysis carry a substantial error and are semi quantita-
tive at best. The measured concentrations are always in-
creased by values in the sub-percentage range. However,
owing to these uncertainties, the quantification of the active
species by 31P NMR appears to be much more reliable than
elemental analysis. A further increase in complex concentra-
tion should be possible by adapting the method for the reac-
tion with the metal precursor and Dpen to the requirements
of supported ligands, probably going to lower temperatures
and less polar solvents.
Figure 8. 13C CP/MAS NMR spectra of the immobilized ligand before
(top) and after (bottom) the reduction procedure (number of scans:
36000, spinning rate: 10000).
several steps at enhanced temperatures and in different sol-
vents. The resonance at 0 ppm can be assigned to carbon
atoms in trimethylsilyl groups. Consequently, the slight shift
and increase in signal intensity for this resonance after re-
duction are a result of the trimethylsilylation of remaining
surface silanol groups with TMSI.
To synthesize the final immobilized molecular catalyst
from the immobilized phosphine ligand, a reaction with the
ruthenium metal precursor and coordination of diphenyl
ACHTUNGTRENNUNGyldiamine (Dpen) is required. The intended immobilized
eth-
The immobilized molecular catalyst was applied in the hy-
drogenation of acetophenone with hydrogen at atmospheric
pressure. For comparison, two molecular catalysts have been
tested under identical reaction conditions in homogeneous
phase. For this, a ruthenium based complex with 2,2’-bis(di-
paratolylphenylphosphino)-1,1’-binaphthyl (tol-Binap) and
Dpen, was prepared. The liquid-state 31P NMR spectra of
the complex displayed a single resonance at 45 ppm, corre-
molecular catalyst and the 31P NMR spectrum measured
after reaction with ruthenium precursor and Dpen are pre-
sented in Figure 9. The resonance at À18 ppm can be as-
signed to phosphorus atoms from the immobilized phos-
phine ligands that did not react with the metal precursor to
form a complex, but are still in the reduced form. Phospho-
rus atoms in the desired coordination should exhibit a chem-
ical shift around 48 ppm.[14] In fact a rather broad resonance
sponding to the target complex.[14] Furthermore, the hex
oxy functionalized biphepO ligand has been employed to
ACHTUNGTRENNUNGen-
AHCTUNGTRENNUNG
Chem. Eur. J. 2009, 15, 9183 – 9190
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
9187

F. Schꢀth et al.
prepare a homogeneous analogue of the immobilized molec-
ular catalysts. The ligand was thus reduced and reacted with
ruthenium precursor and Dpen to form the corresponding
molecular catalyst in the homogeneous phase. The liquid-
phase 31P NMR spectrum of the complex exhibits three reso-
nances at À14, 28 and 48 ppm, corresponding to reduced,
but not complexed species, oxidized phosphorus atoms and
successfully complexed—and therefore catalytically active—
species, respectively. The catalytic activity and enantiomeric
excess (% ee) of the two homogeneous systems and the im-
mobilized molecular catalyst in the hydrogenation of aceto-
phenone to phenylethanol are summarized in Table 1. Inter-
estingly, both homogeneous molecular catalysts show com-
parable catalytic activity and an enantiomeric excess of the
R enantiomer, whereas the immobilized catalyst exhibits
slightly reduced catalytic activity, but a superior enantiomer-
ic excess of the R enantiomer.
Figure 10. Hydrogen uptake with reaction time for the hydrogenation of
acetophenone applying the immobilized molecular catalyst (A) and hy-
drogen uptake of the liquid phase after filtration (after 3 h) (B).
metathesis reaction should allow us to cleave the immobi-
lized complex or ligand again from the support and thus
achieve re-immobilization on a fresh support. Herein, we
evaluated this approach and applied another cross-metathe-
sis reaction utilizing again HMCS to cleave the complex
from the support material and to allow its immobilization
on fresh support material. The envisaged reaction is shown
in Scheme 2. HMCS and the supported molecular complex
are reacted in another cross-metathesis reaction, aiming at a
reaction of the in-chain double bond to connect the complex
with the support material. The resulting product bears again
a chlorosilane functionality allowing subsequent re-immobi-
lization on fresh support material. Aside from those indicat-
ed in Scheme 2, further products from back-metathesis are
possible, including molecules resulting from ring-closure
metathesis. Nevertheless, a significant fraction of the cleaved
molecules will be suitable to react in another cross-metathe-
sis with HMCS to achieve re-immobilization. Indeed, solid-
state 31P NMR measurements before and after the back-
metathesis reaction (Figure 11) illustrate a clear decrease of
all resonance intensities. This decrease can be correlated
with the reduced number of phosphorus components on the
Table 1. Catalytic activity of homogeneous and immobilized molecular
catalysts with different ligands in the hydrogenation of acetophenone to
phenylethanol.
Entry
Ligand
TON
TOF [hÀ1
]
ee [%]
1
2
3
tol-Binap
hexO-biphep
SBA15-(hexO-biphep)
8200
10000
3000
342
417
125
61
62
75
A control experiment has been conducted to exclude the
presence of non-supported complexes in the case of the im-
mobilized molecular catalyst. The mixture was filtered after
3 h reaction time and the hydrogenation reaction was con-
tinued utilizing the liquid phase only (Figure 10). The reac-
tion mixture did not show any further hydrogen uptake in
the following 7 h reaction time except for fluctuations owing
to small temperature changes. Consequently, the presence of
catalytically active species in
the liquid phase can be exclud-
ed. Overall, the immobilized
molecular catalyst shows some-
what reduced TON and TOF.
Nevertheless, the enantiomeric
excess is improved compared
to the homogeneous system.
Thus, a significant contribution
of ruthenium contamination
from the cross-metathesis reac-
tion on the catalytic activity
seems to be unlikely. Hence,
the results on the catalytic ac-
tivity are very promising and
allow for further optimization
of the system, mostly concern-
ing the full complexation of
immobilized species.
Cross-metathesis is a reversi-
ble reaction, therefore, the ap-
plication of another cross-
Scheme 2. A cross-metathesis reaction applying hexenyl-dimethylchlorosilane to remove the complex from the
support material and allow immediate re-immobilization on fresh support material.
9188
www.chemeurj.org
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 9183 – 9190

Reversible Immobilization of a Molecular Catalyst
FULL PAPER
of cross-metathesis, the hexenoxy-functionalized ligand has been reacted
with hexenyldimethylchlorosilane in threefold excess, together with 2nd
generation Grubbs catalyst. The reaction mixture was then added to
SBA-15. After immobilization of the ligand, the material was filtered and
dried. Trimethylsilylation of the remaining surface silanol was accom-
plished in toluene with trimethylsilylimidazole as silylation agent. For re-
duction of the immobilized ligands, a standard procedure applying tri-
chlorosilane and tributylamine in m-xylol has been applied. Complex for-
mation has been accomplished according to a method described by
Noyori et al.[14] for which the reaction time was reduced slightly to sup-
press side reactions.[19, 20] The catalytic activity was tested in the asymmet-
ric hydrogenation of acetophenone to phenylethanol at atmospheric pres-
sure, as described elsewhere.[14] To recycle the ligand, supported catalyst
was reacted with HMCS in the presence of 2nd generation Grubbs cata-
lyst. Consequently, the reaction mixture was filtered and the liquid phase
was transferred onto fresh support material. All materials have been
characterized using X-ray diffraction and nitrogen sorption. Solid-state
NMR spectra (13C, 29Si, and 31P) were measured by using a Bruker
Avance 500WB spectrometer using a 4 mm MAS probe at a spinning
rate of at least 10 kHz. Cross-polarization (CP) was applied for 13C NMR
under the following experimental conditions: 2 s recycle delay, 36,000
scans, 1 ms contact time, and 4.4 ms 1H p/2 pulse. The experimental condi-
tions for 29Si NMR were 30 s recycle delay, 2,400–3,600 scans, and 2.2 ms
p/4 pulse, and those for 31P NMR 15 s recycle delay, 4,000 scans, and
3.8 ms p/2 pulse. High-power proton decoupling was applied for all meas-
urements. The program DMFIT[22] was used for the deconvolution of 29Si
NMR spectra. The reactions mixtures of the catalytic experiments were
analyzed using GC to determine conversion and enantiomeric excess. Ex-
periments were conducted under argon atmosphere using dried solvents.
Figure 11. Solid-state 31P MAS NMR spectra were measured before (top)
and after (bottom) the back-metathesis reaction (number of scans: 4000,
spinning rate: 14000).
support— those that could successfully be cut-off from the
support and re-immobilized on fresh support material
(around 40%, see Figure S1 in the Supporting Information).
Although no full back-metathesis could be reached, adapta-
tion of reaction parameters—especially concerning the com-
plex to silane ratio—are well known to influence reachable
yields in cross-metathesis reactions to a large extent. Conse-
quently, adaptation of the reaction conditions for immobili-
zation of every individual type of ligand or complex will be
required, but will allow reversible immobilization of various
types of ligands or complexes, with the only prerequisite,
that they carry a double-bond functionality.
Acknowledgement
We would like to thank Bodo Zibrowius for NMR measurements.
Conclusion
[1] R. Noyori, Angew. Chem. 2002, 114, 2108–2123; Angew. Chem. Int.
Ed. 2002, 41, 2008–2022.
[2] Q.-H. Fan, Y.-M. Li, A. S. C. Chan, Chem. Rev. 2002, 102, 3385–
3466.
Cross-metathesis reactions could be applied for a reversible,
but covalent immobilization of molecular catalysts. Hex-
ACHTUNGTRENNUNGenoxy functionalized biphepO type ligands and hexenyldi-
[3] D. E. de Vos, I. F. J. Vankelecom, P. A. Jacobs, Chiral Catalyst Im-
mobilization and Recycling, Wiley-VCH, Weinheim, 2000.
[4] C. Copꢃret, M. Chabanas, R. P. Saint-Arroman, J.-M. Basset,
Angew. Chem. 2003, 115, 164–191; Angew. Chem. Int. Ed. 2003, 42,
156–181.
methylchlorosilane were reacted in a cross-metathesis reac-
tion and the reaction product was subsequently anchored on
a silica support material. The immobilized phosphine oxide
ligACHTUNGTRENNUNGands could be reduced and the desired ruthenium-based
[5] C. Saluzzo, M. Lemaire, Adv. Synth. Catal. 2002, 344, 915–928.
[6] J. Horn, F. Michalek, C. C. Tzschuke, W. Bannwarth, Top. Curr.
Chem. 2004, 242, 43–75.
[7] B. Cornils, W. A. Hermann, J. Catal. 2003, 216, 23–31.
[8] E. Lindner, T. Schneller, F. Auer, H. A. Mayer, Angew. Chem. 1999,
111, 2288–2309; Angew. Chem. Int. Ed. 1999, 38, 2154–2174.
[9] T. Joseph, S. S. Deshpande, S. B. Halligudi, A. Vinu, S. Ernst, M.
Hartmann, J. Mol. Catal. A 2003, 206, 13–21.
[10] P. N. Liu, P. M. Gu, F. Wang, Y. Q. Tu, Org. Lett. 2004, 6, 169–172.
[11] J. Thomas, T. Maschmeyer, B. F. G. Johnson, D. S. Shephard, J. Mol.
Catal. A 1999, 141, 139–144.
[12] A. Ghosh, R. Kumar, J. Catal. 2004, 228, 386–396.
[13] B. Kesanli, W. Lin, Chem. Commun. 2004, 20, 2284–2285.
[14] H. Doucet, T. Ohkuma, K. Murata, T. Yokozawa, M. Kozawa, E.
Katayama, A. F. England, T. Ikariya, R. Noyori, Angew. Chem.
1998, 110, 1792–1796; Angew. Chem. Int. Ed. 1998, 37, 1703–1707.
[15] R. Palkovits, C.-M. Yang, S. Olejnik, F. Schꢀth, J. Catal. 2006, 243,
93–98.
hydrogenation catalysts could be formed. The immobilized
molecular catalyst was tested in the hydrogenation of aceto-
phenone to phenylethanol. The immobilized molecular cata-
lyst showed slightly reduced activity compared to the homo-
geneous system, the enantiomeric excess, however, was im-
proved. Additionally, the envisaged back-metathesis to
allow cleavage of the complex from the support followed by
re-immobilization on a fresh support material could be ach-
ieved. Although further optimizations of the single reaction
steps are necessary, we believe that the method is widely ap-
plicable and offers an interesting approach to covalently, but
reversibly, immobilize various molecular catalysts.
Experimental Section
[16] K. Murtrey, J. Liquid Chromatography 1988, 11, 3375–3384.
[17] T. M. Trnka, R. H. Grubbs, Acc. Chem. Res. 2001, 34, 18–29.
[18] H.-C. Wu, J.-Q. Yu, J. B. Spencer, Org. Lett. 2004, 6, 4675–4678.
[19] J. Blꢀmel, Inorg. Chem. 1994, 33, 5050–5056.
Ordered mesoporous silica (SBA-15) was synthesized following a proce-
dure described by Ryoo et al.[21] For the ligand immobilization by means
Chem. Eur. J. 2009, 15, 9183 – 9190
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
9189

F. Schꢀth et al.
[20] J. Sommer, Y. Yang, D. Rambow, J. Blꢀmel, Inorg. Chem. 2004, 43,
[22] D. Massiot, F. Fayon, M. Capron, I. King, S. Le Calvꢃ, B. Alonso, J.-
O. Durand, B. Bujoli, Z. Gan, G. Hoatson, Magn. Reson. Chem.
2002, 40, 70–76.
7561–7563.
[21] M. Choi, W. Heo, F. Kleitz, R. Ryoo, Chem. Commun. 2003, 1340–
Received: November 17, 2008
1341.
Revised: March 26, 2009
Published online: August 4, 2009
9190
www.chemeurj.org
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 9183 – 9190