The immobilization of rhodium-4-(diphenylphosphino)-2- (diphenylphosphinomethyl)-pyrrolidine (Rh-PPM) complexes: A systematic study

Article abstract of DOI:10.1002/adsc.200606161

A modular toolbox for the immobilization of homogeneous catalysts to various supports is described. It consists of functionalized chiral diphosphines and three different linkers based on isocyanate chemistry and it is used to attach the 4-(diphenylphosphino)-2-(diphenylphosphinomethyl)-pyrrolidine (PPM) ligand to a large variety of soluble, swellable and non-swellable solid organic polymers and to silica gels. As model reaction the hydrogenation of acetamidocinnamic acid derivatives, catalyzed with high enantioselectivity was chosen. Besides information on the usefulness of a particular type of support for synthetic applications, the experiments were also designed to address the question how parameters such as solubility, swellability, cage or pore size and solvent affect the rate and enantioselectivity of an immobilized catalyst. Rhodium complexes of ligands attached to soluble polymers and inorganic supports achieved ees up to 95% and turnover frequencies between 700 and 1400 h ^-1, very close to the values of the homogeneous Rh catalyst (ee 95%, TOF 1320 h^-1). Insoluble or strongly cross-linked organic polymers led to catalysts with lower enantioselectivity and activity. PPM ligands attached to water soluble dendrimer fragments allowed hydrogenation in water solution with ees up to 94%, albeit with much lower activity compared to reactions in methanol with the homogeneous catalyst.

Full text of DOI:10.1002/adsc.200606161

FULL PAPERS
DOI: 10.1002/adsc.200606161
The Immobilization of Rhodium-4-(diphenylphosphino)-2-
(
diphenylphosphinomethyl)-pyrrolidine (Rh-PPM) Complexes:
A Systematic Study
a,
a
Benoît Pugin * and Hans-Ulrich Blaser
a
Solvias AG, P.O. Box, 4002Basel, Switzerland
Fax : (+41)-61-686-6311; e-mail: benoit.pugin@solvias.com
Received: April 4, 2006; Accepted: June 13, 2006
Supporting information for this article is available on the WWW under http://asc.wiley-vch.de/home/.
Abstract: A modular toolbox for the immobilization ligands attached to soluble polymers and inorganic
of homogeneous catalysts to various supports is de- supports achieved ees up to 95% and turnover fre-
À1
scribed. It consists of functionalized chiral diphos- quencies between 700 and 1400 h , very close to the
phines and three different linkers based on isocya- values of the homogeneous Rh catalyst (ee 95%,
À1
nate chemistry and it is used to attach the 4-(diphe- TOF 1320 h ). Insoluble or strongly cross-linked or-
nylphosphino)-2-(diphenylphosphinomethyl)-pyrroli-
ganic polymers led to catalysts with lower enantiose-
dine (PPM) ligand to a large variety of soluble, swel- lectivity and activity. PPM ligands attached to water
lable and non-swellable solid organic polymers and soluble dendrimer fragments allowed hydrogenation
to silica gels. As model reaction the hydrogenation in water solution with ees up to 94%, albeit with
of acetamidocinnamic acid derivatives, catalyzed much lower activity compared to reactions in metha-
with high enantioselectivity was chosen. Besides in- nol with the homogeneous catalyst.
formation on the usefulness of a particular type of
support for synthetic applications, the experiments
were also designed to address the question how pa- Keywords: a-acetamidocinnamic acid derivatives; 4-
rameters such as solubility, swellability, cage or pore (diphenylphosphino)-2-(diphenylphosphinomethyl)-
size and solvent affect the rate and enantioselectivity pyrrolidine (PPM); hydrogenation; immobilization;
of an immobilized catalyst. Rhodium complexes of rhodium; supported catalysts
Introduction
on immobilized catalysts[5–7] and a large number of
research groups both in academia and in industry is
First efforts to heterogenize enantioselective homoge- still developing new immobilization methods or
neous catalysts via attachment to insoluble carriers adapting known ones for specific reactions.
started very shortly after the discovery of effective
Despite many successful examples, it became clear
rhodium complexes for hydrogenation reactions. The very early that immobilization is no panacea for enan-
basic idea was to combine the best features of homo- tioselective catalysis. The heterogenized catalysts are
geneous and heterogeneous catalysis, i.e., the enantio- not only much more complex (and much more expen-
selectivity of the homogeneous catalysts with the sive) than the homogeneous analogues, but in many
good handling and separation properties of heteroge- cases their catalytic performance with respect to
neous catalysts. Early pioneers were Kagan, who co- enantioselectivity and/or activity was far below that of
valently attached his diop ligand to a Merrifield their homogeneous counterparts. Many attempts were
[
1a]
[1b]
[2a]
resin
and later to graphite,
Stille,
the first to reported in the literature to identify and understand
study the effect of additional chiral groups in the po- the positive as well as negative effect of various im-
[
2b]
lymer backbone,
Achiwa, who immobilized the mobilization methods on the catalytic performance
4
-(diphenylphosphino)-2-(diphenylphosphinomethyl)-
and a number of possible reasons can be postulated:
Interactions between immobilized complexes due
[
3]
pyrrolidine (PPM) ligand by copolymerization and
Hetfleijs, the first to use silica gel as support.[ Later, to high local concentration on the carrier. This can
various different immobilization strategies were de- result in both positive effects due to cooperation[
veloped and today, there are hundreds of publications and negative effects due to catalyst deactivation via
4]
8]
Adv. Synth. Catal. 2006, 348, 1743 – 1751
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1743

Benoît Pugin and Hans-Ulrich Blaser
FULL PAPERS
dimerization.[9] On the positive side, fixing a complex The main elements of our system are functionalized
on a rigid support with high dilution can lead to an chiral diphosphines, three different linkers based on
[
12–14]
enforced isolation of the active sites.
isocyanate chemistry and various carriers.
This
Interactions between the active center and the sup- approach allows a systematic and quick access to a va-
port material, e.g., silanol groups of silica gel or aro- riety of immobilized chiral catalysts with the possibili-
matic rings of an organic polymer may affect the per- ty to adapt their material and catalytic properties to
formance of the immobilized catalyst.
the specific needs.
Restriction of conformational freedom due to steric
In order to make such a toolbox useful for industri-
interaction with the support, especially inside small al applications, the following criteria have to be fulfil-
pores or the network of strongly cross-linked poly- led:
mers, i.e., the effect of small cages. Also here, positive
e.g., when the ligand is too flexible) as well as nega- depicted in Figure 3 as well as a variety of inorganic
tive effects (e.g., for an optimized catalyst) on enan- and organic support materials are commercially avail-
(1) General and efficient preparation: The linkers
(
tioselectivity and rate can be expected.
able and the isocyanate chemistry is well developed.
(2) Access to a variety of properly functionalized
Mass transport effects due to diffusion limitations
of one or several reactants. Here the dominant effect chiral diphosphines: This is a major drawback of the
is the rate of the reaction, but due to mechanistic rea- covalent attachment strategy since most commercially
sons, enantioselectivity can be affected as well. available diphosphines have to be modified, requiring
Since it is not (yet) possible to predict which ligand an often sizeable synthetic effort.
and support will be suitable for a given substrate and
(3) Reasonable “molecular weight” of the immobi-
process, the ligand selection occurs via a screening ap- lized catalyst: For practical reasons, this should not
proach. To do this successfully, a large variety of exceed 10 kD per mole of immobilized metal com-
chiral ligands and metal precursors has to be immedi- plex.
ately at hand since time constraints for process devel-
Here we report on a systematic study where differ-
opment can be rather severe.[ In view of the fact ent carriers were attached to the same ligand using a
that almost no immobilized ligands are commercially common linker strategy. The goal was to study the
available the screening has to be carried out with ho- effect on rate and enantioselectivity of a reaction with
mogeneous catalysts. The best ligand should then be a high intrinsic enantioselectivity. As model we chose
immobilized as fast and reliably as possible and with- the hydrogenation of acetamidocinnamic acid deriva-
10]
out any loss in performance.
tives, catalyzed by Rh-PPM complexes [15] covalently
Inspired by the seminal paper of Nagel[ describ- attached to various supports (see Figure 2). Besides
ing a very active and selective Rh-pyrphos catalyst co- information on the usefulness of a particular support
valently attached to silica gel we have developed the for synthetic application, the experiments were also
modular toolbox schematically depicted in Figure 1. designed to study the effect of various parameters
11]
Figure 1. Elements and parameters of the modular toolbox
1744
www.asc.wiley-vch.de
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2006, 348, 1743 – 1751

Immobilization of Rhodium-4-(diphenylphosphino)-2-(diphenylphosphinomethyl)-pyrrolidine
FULL PAPERS
Figure 2. Test reaction using various rhodium-4-(diphenylphosphino)-2-(diphenylphosphinomethyl)-pyrrolidine (Rh-PPM)
catalysts.
(
e.g., solubility, swellability, cage or pore size, solvent) are feasible but due to the different reactivity of the
on rate and enantioselectivity of the heterogenized two isocyanate groups the position of the methyl
catalyst.
group is different for the two routes (see Figure 4).
However, we did not find any systematic differences
in the catalytic behavior between catalysts prepared
via the two methods (results not described here). The
reactive support route is more flexible than the reac-
Results and Discussion
Preparation and Characteristics of Immobilized
Catalysts
tive ligand route. First, not all OH or NH groups of
the support have to be reacted with TDI and, second-
ly, the loading of the PPM ligand can also be control-
2
All immobilized ligands were assembled from PPM, led. In case of incomplete loading with PPM, the
an appropriate isocyanate linker and the desired sup- second isocyanate group is either reacted with an al-
port as schematically depicted in Figure 3 (polymer- cohol, allowing modification of the polarity of the
supported ligands) and Figure 5 (water-soluble li- polymeric ligand (usually EtOH) or it can be used for
gands). A schematic overview on important proper- cross-linking (see below). While we have not fully
ties of different supports types is given in Table 1. characterized the immobilized ligands, we have esti-
The specific polymeric and inorganic supports used in mated the apparent molecular weight via phosphorus
this investigations are listed in Table 2, where also in- elemental analysis (see Tables 3 and 4).
formation on some structural characteristics can be
found.
The structure of the water soluble carriers are de-
picted in Figure 5. Their synthesis is rather straightfor-
For the immobilization of a ligand on a polymeric ward, first reacting carbonyl diimidazole with the car-
support, two routes are possible: Either the linker is rier followed by reaction with the PPM fragment.
first reacted with the N- or O-function of the ligand While the polyacrylic acid has already been used by
[
16]
and then with the support (reactive ligand route) or Andersson
to immobilize selected ligands, the ap-
the linker is first attached to the support and then re- plications of the two dendrimer fragments is new. For
acted with the ligand (reactive support route). For in- the synthesis of AcidD, the condensation reaction has
organic supports, only the reactive ligand route is to be carried out with the corresponding triester fol-
[
14]
used
whereas for organic polymers with TDI (tolu- lowed by base catalyzed hydrolysis after attachment
ene-2,4-diisocyanate) as linker precursor, both routes of the ligand.
Figure 3. Structure of ligand, linker precursor and immobilized ligand assembly for the various carriers/supports.
Adv. Synth. Catal. 2006, 348, 1743 – 1751
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.asc.wiley-vch.de
1745

Benoît Pugin and Hans-Ulrich Blaser
FULL PAPERS
Figure 4. Preparation of 4-(diphenylphosphino)-2-(diphenylphosphinomethyl)-pyrrolidine (PPM) ligands immobilized on or-
ganic polymers (X=O, NH).
Table 1. Organic polymers and inorganic solids used for the immobilization of metal complex catalysts.
Linear polymer
Slightly cross-linked poly-
mer
Highly cross-linked polymer
Amorphous inorganic sup-
port
E.g., non-cross-linked polymers
such as HEMA, PKHH
E.g., polystyrene with 0.5-
3% divinylbenzene
E.g., polystyrene with>5%
divinylbenzene
E.g., silica gel, alumina
Soluble, solvent dependent perfor- Swellable, solvent depen-
mance dent performance
Insoluble, slightly solvent de- Insoluble, solvent inde-
pendent performance
pendent performance
1746
www.asc.wiley-vch.de
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2006, 348, 1743 – 1751

Immobilization of Rhodium-4-(diphenylphosphino)-2-(diphenylphosphinomethyl)-pyrrolidine
FULL PAPERS
Table 2. Abbreviations and selected characteristics of the polymer and inorganic supports used in this study.
[a]
[b] [c]
Support
MW, D,
P
Soluble in
Insoluble in
mmol
OH
[
g]
PKKH, commercially available polyphenoxy resin
MW 50,000 D
MW 300,000 D
CH Cl , THF,
MeOH/THF
hexane, alco-
hols
3.5
2
2
(
Union Carbide), non-cross-linked
HEMA (Aldrich 192066), poly(2-hydroxyethylmetha-
crylate), non-cross-linked
MeOH/THF, al- hexane, CH Cl , 7.7
cohols, DMF
2
2
swells in THF
PS-PEG 1% (Fluka 81185). Polyethylene glycol bound slightly cross-
swells in THF,
slightly in alco-
hol
all solvents
all solvents
0.35
1.49
to PS cross-linked with 1% DVB
linked
PS 25% (Rohm & Haas, XAD-2). Polystyrene, cross-
linked with 25% divinylbenzene, hydroxymethylated
highly cross-
linked, macropo-
rous
-
[d]
Silica gels
Grace 332D 35–70
mm, P 19 nm
all solvents
9-10
Merck 100
Merck 60
Merck 40
D 200–500 mm, P 14 nm
D 60–200 mm, P 10 nm
D 60–200 mm, P 4.4 nm
mmol/
m
2
[
[
[
[
a]
b]
c]
d]
Average molecular weight.
Particle size (mm)
Mean pore diameter (nm).
For more detailed information, see ref.
[14]
Figure 5. 4-(Diphenylphosphino)-2-(diphenylphosphinomethyl)-pyrrolidine (PPM) tethered to various water soluble carriers.
Hydrogenation Experiments
trogen atom of the pyrrolidine ring is ideally suited to
be attached to a variety of linkers using our isocya-
The supported ligands described in the preceding nate technology. The hydrogenation experiments with
paragraph were applied for the Rh-catalyzed hydro- MAC (see Tables 3 and 4) were carried out in metha-
genation of methyl a-acetamidocinnamate (MAC) nol, the preferred solvents for this reaction and
and a-acetamidocinnamic acid (ACA) (Figure 1), i.e., MeOH/THF mixtures in order to dissolve or swell
the most important standard test reaction for enantio- some of the immobilized catalysts. For the hydrogena-
selective hydrogenation. The PPM ligand was chosen tion of ACA with the water-soluble catalysts, water
for two reasons: First, it was well documented that and MeOH were used as solvents (see Table 5). The
homogeneous Rh-PPM catalysts have a very good catalysts were prepared in situ from the appropriate
performance for this reaction[ and, secondly, the ni- ligand and Rh
15]
A
H
R
U
G
2
4
Adv. Synth. Catal. 2006, 348, 1743 – 1751
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.asc.wiley-vch.de
1747

Benoît Pugin and Hans-Ulrich Blaser
FULL PAPERS
diene) and the reactions were usually run to high con- cline may be attributed to the attachment of the cata-
versions. lyst within too small pores, which affects mass trans-
As can be seen from the results listed in Table 3, port and restricts the conformational freedom of the
the performance of the homogeneous Rh-BPPM cata- catalytic species. The solvent effects are also much
lyst is very little solvent dependent: The ees vary be- less pronounced when inorganic supports are used
À1
tween 95 and 96%, TOFs between 1300 and 1380 h
(
(entry 3.7). Whereas the silica gel with the larges
À1
entry 3.1). Similar catalyst performances are ob- pores achieves a TOF up to 1400 h
(entry 3.6),
served for anchored catalysts on dissolved polymers smaller pores lead to a decrease in activity which is
entries 3.2and 3.3). With few exceptions, ligands at- pronounced only for the catalyst with very small
(
tached to polymers achieve significantly lower TOFs pores and high ligand loading (entry 3.11). Enantiose-
and somewhat lower ees as soon as the polymer is no lectivities vary less systematically but in general the
longer soluble. When a linear polymer is precipitated same trend is observed: Restrictions of any kind lead
À1
the TOF drops to 12h (entry 3.2). We also observed to a drop of ee between a few percent to 25%
that catalysis comes to a complete standstill when a (entry 3.2). From this series of experiments we tenta-
catalyst attached to a swellable support was used in tively conclude that making the Rh-PPM centers less
solvents where it does not swell (unpublished results). and less accessible leads first to a reduction in rate
Interestingly, while the TOF and ee of the highly and only later to a decrease in enantioselectivity. This
cross-linked Rh-PPM-PS 25% catalyst are low (TOF interpretation was tested with a series of PKKH sup-
~
80, ee ~87%, entry 3.5) its performance is almost ports cross-linked to various degrees with two differ-
unaffected by the solvent, indicating that the high ent agents and the results are summarized in Table 4.
degree of cross-linking results in a rigid support that It has to be stressed that the degree of cross-linking
has no more swelling properties. The performance de- was not determined quantitatively and therefore we
À1
Table 3. Enantioselectivity (ee,%) and activity (TOF at ~50% conversion, h ) for the hydrogenation of methyl a-acetami-
[a]
docinnamate (MAC) with various supported Rh-PPM catalysts.
[b]
[c]
Entry
Catalyst
MW
MeOH
ee [%]
MeOH/THF
ee [%]
Comments
À1
À1
TOF [h ]
TOF [h ]
3
3
3
3
3
3
3
3
3
3
3
.1
.2
.3
.4
.5
.6
.7
.8
Rh-BPPM
554
2650
1905
15500
4700
12900
17500
8200
95
70
91
87
86
92
93
92
93
90
89
1380
12
480
240
70
~1400
~1200
~800
~920
~800
~270
96
95
95
90
87
1320[
homogeneous catalyst
d])
[e]
Rh-PPM-PKKH
Rh-PPM-HEMA
Rh-PPM-PS-PEG 1%
Rh-PPM-PS 25%
1200
A , insoluble in MeOH
[e]
1320
420
90
B , soluble polymer
B, not swellable in MeOH
[
e]
Rh-PPM-Grace 332
B
[
f]
Rh-PPM-Merck 100[
92
~800
B, low PPM loading
B, high PPM loading
B
B, low PPM loading
B, high PPM loading
f]
Rh-PPM-Merck 100
[
f]
f]
f]
.9
.10
.11
Rh-PPM-Merck 60[
9500
11300
4700
Rh-PPM-Merck 40[
Rh-PPM-Merck 40
[
[
[
[
[
[
a]
Reaction conditions: Ligand/Rh (COD) BF =1.2, S/C 200, 1 bar, 258C.
Estimated molecular weight of immobilized ligand.
MeOH/THF 3.5/1 (vol/vol).
2
4
b]
c]
d]
e]
f]
At>99% conversion.
A: reactive polymer method, B: reactive ligand method.
[14]
Data based on ref.
Table 4. Effect of cross-linking for Rh-PPM supported on PKKH prepared via the reactive polymer route A. Enantioselec-
À1
[a]
tivity (ee,%) and activity (TOF at ~50% conversion, h ) for the hydrogenation of MAC in MeOH/THF 3.5/1.
[
b]
À1
Entry
Catalyst/Cross-linking agent
MW
554
ee [%]
TOR [h ]
Comments
4
4
4
4
4
4
.1
.2
.3
.4
.5
.6
Rh-BPPM
Rh-PPM-PKKH
Rh-PPM-PKKH/PEG200-1
Rh-PPM-PKKH/PEG200-2
Rh-PPM-PKKH/decanediol-1
Rh-PPM-PKKH/decan e6 d0 iol-232
96
95
93
88
87
1320
1200
560
520
160
homogeneous catalyst
not cross-linked
partially cross-linked
strongly cross-linked
partially cross-linked
strongly cross-linked
2650
3670
3450
3760
74
<5
[
[
a]
Reaction conditions: Ligand/Rh(COD) BF =1.2, S/C 200, 1 bar, 258C.
Estimated molecular weight of immobilized ligand.
A
H
R
U
G
2 4
b]
1748
www.asc.wiley-vch.de
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2006, 348, 1743 – 1751

Immobilization of Rhodium-4-(diphenylphosphino)-2-(diphenylphosphinomethyl)-pyrrolidine
FULL PAPERS
Table 5. Enantioselectivity (ee) and activity (TOF at high conversion) for the hydrogenation of a-acetamidocinnamic acid
[a]
(
ACA) with water soluble Rh-PPM catalysts.
Entry
Catalyst
MeOH
H O
2
Comments
5
5
5
5
.1
Rh-BPPM
96
19 22
95
97
2600
89
1
homogeneous catalyst
.2Rh-P0 P0 M0 -PAA2 -
.3
.4
-
À1
Rh-PPM-AcidD
Rh-PPM-AmineD
2400
500
94
92100
240
H O at 15 bar: ee 96%, TOF 2000 h
2
[
a]
Reaction conditions: Ligand/Rh (COD) BF =1.2, S/C 200, 1 bar, 258C.
2
4
are not able to directly compare results of the differ- crease in ee and TOF. On the other hand, we cannot
ent types of polymers but only those with the same exclude forced interactions with either linker and sup-
cross-linker. Nevertheless, the results agree very well port.
with our tentative conclusion: While any type of
There are two motivations to render a catalyst
cross-linking leads to lower activity, ees drop only water-soluble. One is the possibility to work in a two
when the cross-linking is higher and/or when the link- phase system where the substrate is soluble in the or-
ing agent leads to more restriction (compare PEG200 ganic phase while the catalyst is “immobilized” in an
vs. 1,10-decanediol). Under extreme conditions activi- aqueous phase. This strategy allows catalyst recovery
ty drops to almost zero (entry 4.6).
via phase separation but in many cases transport re-
[
7]
As discussed in the Introduction, various effects striction occurs between the two phases. In this
can be responsible for the observed decrease in cata- series of experiments we were interested in the
lyst performance. Even though we cannot rigorously second opportunity, namely the possibility to hydro-
exclude this, we think that it is rather unlikely that genate water-soluble substrates in water as the only
[
17]
the results summarized in Tables 3 and 4 are due to solvent.
The results summarized in Table 5 clearly
chemical differences of the linkers and/or support ma- demonstrate that the isocyanate linker technology is
terials. The reason for this conclusion is that the same basically suitable for this purpose.
trends are observed for the different systems when
While the BPPM-Rh complex is insoluble in water
the ligands are more and more constricted either by and as a consequence shows extremely low activity in
increased cross-linking or by smaller pores. If we water (entry 5.1), the catalysts bound to a water-solu-
accept this interpretation, there are still several inter- ble dendrimer fragment achieve moderate to good
pretations possible. The first and at the moment the TOFs (entries 5.3 and 5.4). Nevertheless, when we
most plausible is that enclosing the catalyst in a rela- compare the results in water with those in methanol
tively rigid polymer matrix or in a very small pore no there are obvious limitations. Somewhat surprising
longer allows the Rh-PPM catalyst to adopt the opti- was the very low TOF for the catalyst based on poly-
mal geometry during the catalytic cycle. This can lead acrylic acid (entry 5.2). But also the ligands attached
to a decrease in ee if the product determining inter- to the two dendrimer fragments achieved much lower
mediates are affected and/or to a decrease in rate if, TOFs and somewhat lower ees in water than in
e.g., access of the substrate is hindered. On the other MeOH. One reason could be the much lower solubili-
hand, the restriction could lead to closer interactions ty of hydrogen in water, leading to a lower hydrogen
of the Rh centers either with each other (dimer for- concentration. Indeed, when we increased the pres-
mation due to high local concentration) or with the sure to 15 bar, both TOF and ee improved significant-
support and/or linker. To test the effect of high local ly for the Rh-PPM-AcidD catalyst (entry 5.4). Fur-
Rh concentration we prepared a PPM dimer, linked thermore, at 15 bar it was also possible to use this par-
via a short chain diamine depicted in Figure 6 and ticular catalyst with S/C ratios up to 10,000 (ee 94%,
À1
tested the Rh complex for the hydrogenation of MAC TOF 5000 h ). This last results indicates that some of
under our standard conditions. Both in MeOH and our experiments might actually be at least partially
MeOH/THF the ee and TOF were the same as for the mass transport controlled and that with a more effec-
Rh-BPPM catalyst, demonstrating that Rh-Rh inter- tive gas-liquid mixing system even higher TOF could
actions are not a probable cause for the observed de- be reached.
Finally we briefly discuss the potential of our mod-
ular toolbox for synthetic and technical applications.
Our results clearly show that it is possible to prepare
immobilized catalysts with a synthetically useful per-
formance, in several cases close to that of the homo-
geneous analogue. Indeed, we have recently applied
this technology to the Ir-catalyzed hydrogenation of a
Figure 6. Dimeric PPM ligand.
herbicide intermediate with a diphosphine immobi-
Adv. Synth. Catal. 2006, 348, 1743 – 1751
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.asc.wiley-vch.de
1749

Benoît Pugin and Hans-Ulrich Blaser
FULL PAPERS
lized on silica where turnover numbers>100,000 have fragments allow hydrogenation in water solution with
[
18]
been achieved
as well as to the Rh-catalyzed hy- ees up to 94%. However, catalyst activity is much
drogenation of folic acid to l-tetrahydrofolic acid in lower compared to reactions in methanol, maybe due
water with a diphosphine attached to acid with mod- to mass transfer restrictions. Many of the immobilized
[
19]
erate selectivity but very good turnover numbers.
catalysts described in principle fulfill the criteria for a
While we have not systematically investigated the fil- technical application, but there are also obvious limi-
tration and leaching behavior of our immobilized cat- tations.
alysts, we have indications that we can recover>95%
of our Rh complexes via filtration and that in most
cases Rh and P content is<5 ppm.
Experimental Section
However, there are obvious limitations to our im-
mobilization strategy especially for technical applica- Reagents and solvents used in this study purchased from Al-
drich or Fluka and used as received unless otherwise stated.
tions:
All manipulations were carried out under an Argon atmos-
phere.
(
1) In all cases, the ligand which has been identified
to have a sufficient catalyst performance has to be
functionalized with an appropriate O or N function
for tethering. While this is possible (we have synthe-
sized about ten relevant ligands), it makes the catalyst
much more expensive.
Synthesis of Immobilized Ligands
The preparation of ligands immobilized on silica gels is de-
scribed in ref.[ that of the water-soluble ligands and their
use in catalytic hydrogenations are described in ref.[
14]
(
2) Many of the polymer-based catalysts are either
19]
not as active as the homogeneous catalyst or are very
sensitive to solvent effects. Furthermore, the separa-
tion of soluble polymeric catalyst which have the best
catalyst performance needs either ultrafiltration
equipment (which is expensive) or a change in solvent
to precipitate the catalyst (which is inconvenient).
Polymers: PKHH, phenoxy resin, Union Carbide.
HEMA, poly(2-hydroxyethyl methacrylate), non-cross-
linked, Aldrich 192066. PS-PEG600 1%, polyethylene
glycol bound to polystyrene cross-linked with 1% divinyl-
benzene, Fluka 81185. PS
A
H
R
U
G
cross-linked with 25% divinylbenzene. This support was
(
3) Highly cross-linked polymers have high OH prepared starting from macroporous polystyrene (XAD-2,
group density but unfortunately the pore structure of Rohm & Haas) which was first chloromethylated and then
the PS 25% polymer seems not be optimal. To get a hydrolyzed as described in the literature.[20,21]
useful polymer, optimization of the pore structure
would be required without affecting the functional
group density which might be difficult to achieve.
The more active silica gel-based catalysts have a
larger apparent molecular weight due to limited sur-
face area of the large pore material but are in princi-
Representative Example for the Preparation of
Polymer-Supported 4-(Diphenylphosphino)-
2
-(diphenylphosphinomethyl)-pyrrolidine Ligand by
the ꢀReactive Polymer Methodꢁ A
ple suitable for technical application, especially on a PPM-PKHH:
Toluene-2,4-diisocyanate
(2.84 mL,
small scale. An alternative to the amorphous silica 19.8 mmol) was added dropwise to a solution of 350 mg
gels are mesoporous crystalline materials such as (1.23 mmol OH groups) of PKHH phenoxy resin in 10 mL
[
6a]
of dichloromethane. After addition of 0.020 mL of triethyla-
mine (catalyst) the solution was stirred at 508C for 200 min.
After cooling to room temperature, the excess of toluene-
MCM-41
which we have not applied in this study.
2
,4-diisocyanate was removed by repeating the following
Conclusions
procedure 4 times: Precipitation of the polymer by addition
of 50 mL of hexane, decantation of the supernatant solution,
re-dissolve polymer in 10 mL of dichloromethane.
We have shown that the covalent attachment of di-
phosphine ligands to various organic and inorganic
After the last decantation, the polymer was dissolved in
supports as well as water-soluble carriers is a feasible 10 mL of dichloromethane, 140 mg (0.31 mmol) of 4-(di-
strategy to heterogenize homogeneous catalysts. phosphino)-2-(diphenylphosphinomethyl)-pyrrolidine were
While ligands attached to soluble polymers and inor- added and the reaction mixture was stirred at 258C for 20 h.
Then 5 mL of ethanol and 0.020 mL of triethylamine were
added and the mixture stirred at 508C for 4 h. Finally the
ganic supports have similar catalyst performance as
the homogeneous analogue, insoluble or strongly
product was washed by repeating the following procedure 4
cross-linked polymers lead to catalysts with lower
times: precipitation with 50 mL of hexane/diethyl ether 3:2,
enantioselectivity and activity. There are indications
decantation of the supernatant solution, re-dissolve or swell
that this negative effect is mainly due to a confine-
product with 10 mL of dichloromethane. After the last de-
ment of the catalytically active species inside the po-
cantation the product was dried under reduced pressure at
lymer matrix or very narrow pores, although interac-
room temperature. The product was obtained as a white
tions with the linker and/or support cannot be exclud- powder that is practically soluble in dichloromethane. The
ed. PPM ligands attached to water-soluble dendrimer P-content was determined by microanalysis.
1750
www.asc.wiley-vch.de
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2006, 348, 1743 – 1751

Immobilization of Rhodium-4-(diphenylphosphino)-2-(diphenylphosphinomethyl)-pyrrolidine
FULL PAPERS
For details on the preparation of all polymer supported li- Acknowledgements
gands by the ꢁreactive polymerꢂ method see Supporting In-
formation.
We thank Heidi Landert and Andrea Schwendemann for
their careful experimental work.
Preparation of Polymer-Supported
4
-(Diphenylphosphino)-2-(diphenylphosphinome-
thyl)-pyrrolidine Ligands by the ꢀReactive Ligand
Methodꢁ B
References
[
1] a) W. Dumont, J. C. Poulin, P. T. Dang, H. B. Kagan, J.
Am. Chem. Soc. 1973, 95, 8295; b) H. B. Kagan, T. Ya-
magishi, J. C. Motte, R. Setton , Israel J. Chem. 1979,
Preparation of the ꢁreactive ligandꢂ (TDI-PPM): A solution
of 180 mg (0.4 mmol) of 4-(diphenylphosphino)-2-(diphenyl-
phosphinomethyl)-pyrrolidine in 3 mL of dichloromethane
was added to a solution of 0.58 mol of toluene-2,4-diisocya-
nate in 4 mL of dichloromethane at À508C. The cooling
bath was removed after stirring for 1 h and the solvent was
evaporated under reduced pressure at room temperature.
Addition of 25 mL of hexane under vigorous stirring caused
the product separate as an oil which sticks to the glass wall.
The hexane was decanted and the oil is washed 5 times with
1
7, 274.
[
2] a) For a review, see J. K. Stille, Reactive Polymers 1989,
1
1
0, 165; b) T. Matsuda, J. K. Stille, J. Am. Chem. Soc.
978, 100, 268; c) G. L. Baker, S. J. Fritschel, J. K. Stille,
J. Org. Chem. 1981, 46, 2960; d) R. Deschenaux, J. K.
Stille, J. Org. Chem. 1985, 50, 2299.
[
[
3] K. Achiwa, Chem. Lett. 1978, 905.
4] I. Kolb, M. Cerny, J. Hetfleijs, React. Kinet. Catal. Lett.
30 mL hexane. Finally the TDI-PPM (reactive ligand) was
1
977, 7, 199.
dried under high vacuum.
PPM-HEMA: A solution of 100 mg of HEMA polymer
in 3 mL of dimethylformamide was added to a solution of
[
[
5] Chiral Catalyst Immobilization and Recycling, (Eds.:
D. E. De Vos, I. F. J. Vankelecom, P. A. Jacobs), Wiley-
VCH, Weinheim, 2000.
6] Some recent overviews: a) P. McMorn, G. J. Hutchings,
Chem. Soc., Rev. 2004, 33, 108; b) Q.-H. Fan, Y.-M. Li,
A. S. C. Chan, Chem. Rev. 2002, 102, 3385; c) C. E.
Song, S. Lee, Chem. Rev. 2002, 102, 3495.
7] For a recent compilation on biphasic catalysis, see:
J. A. Gladysz, Chem. Rev. 2002, 102, 3215, introducing
a topical issue of Chem. Rev.
8] For an example, see: D. A. Annis, E. N. Jacobsen, J.
Am. Chem. Soc. 1999, 121, 4147.
170 mg of TDI-PPM in 2mL of dimethylformamide. After
addition of 0.01 mL of triethylamine (catalyst) the reaction
mixture was stirred at 558C for 16 h. After cooling to room
temperature, the product was washed in several steps: (1)
addition of hexane and vigorous stirring led to the formation
of two phases. The upper phase was separated. (2) addition
of diethyl ether. Stirring caused the product to oil out/solidi-
fy and the supernatant solution was removed (3–6) the oil/
solid was stirred in 5 mL of methanol, then 50 mL of hexane
were added. The supernatant solution was decanted from
the precipitate. After the last washing the product was dried
under reduced pressure at room temperature.
PPM-PS-PEG1%: A solution of approx. 130 mg of TDI-
PPM in 3 mL of dichloromethane was added to a suspension
of 2g of the PS-PEG polymer in 12mL of dichloromethane.
After addition of 0.01 mL of triethylamine (catalyst) the re-
action mixture was stirred at reflux temperature for 20 h.
After cooling to room temperature, the product was washed
several times with dichloromethane and finally dried under
reduced pressure.
[
[
[
9] H. U. Blaser, B. Pugin, F. Spindler, A. Togni, C. R.
Chimie 2002, 5, 1.
[
[
[
10] H. U. Blaser, E. Schmidt, in: Large Scale Asymmetric
Catalysis, (Eds.: H. U. Blaser, E. Schmidt), Wiley-VCH,
Weinheim, 2003, p. 1.
11] U. Nagel, E. Kinzel, J. Chem. Soc., Chem. Commun.
1
8
986, 1098; U. Nagel, J. Leipold, Chem. Ber. 1996, 129,
15.
12] B. Pugin, F. Spindler, M. Müller, EP 496699 and
4
96700, 1991 (assigned to Ciba-Geigy AG).
[
[
13] B. Pugin, J. Mol. Catal. 1996, 107, 273.
14] B. Pugin, M. Müller, Stud. Surf. Sci. Catal. 1993, 78,
Hydrogenation Experiments
1
07.
The ligand (0.015 mmol) was stirred for 15 min in either tet-
rahydrofuran or a mixture of tetrahydrofuran and methanol
to allow the polymer supports either to dissolve or to swell.
[15] PPM=4-diphenylphosphino-2-diphenylphosphinome-
thylpyrrolidine, K. Achiwa, J. Am. Chem. Soc. 1976, 98,
8265.
Then a solution of 0.0125 mmol of [Rh
of methanol was added. The mixture was stirred for further
5 min. During this time, the insoluble beads became in-
A
H
R
U
G
[16] T. Malmstrçm, C. Andersson, Chem. Commun. 1996,
1135; T. Malmstrçm, C. Andersson, J. Mol. Cat. A:
Chemical 1999, 139, 259.
2
4
1
tensely orange yellow while the solution discolored. Then, a
solution of 2.5 mmol of methyl acetamidocinnamate in
[17] D. Sinou, Adv. Synth. Catal. 2002, 344, 221.
[18] B. Pugin, H. Landert, F. Spindler, H. U. Blaser, Adv.
Synth. Catal. 2002, 344, 974.
1
6 mL of the solvent(s) indicated in the Tables was added.
The argon atmosphere was exchanged against hydrogen
ambient pressure) and the hydrogenation started by vigo-
[19] B. Pugin, R. Moser, V. Groehn, H. U. Blaser, Tetrahe-
(
dron: Asymmetry 2006, 17, 544.
rous stirring. The course of the hydrogenation was followed
by monitoring the hydrogen consumption. Conversion and
ee were determined by GC (Chirasil-L-val, 50 m capillary
column, carrier gas: helium).
[20] H. J. Van den Berg, G. Challa, U. K. Pandit, J. Mol.
Catal. 1989, 51, 13.
[21] J. M. J. Frechet, M. D. De Smet, M. J. Farall, Polymer
1979, 20, 675.
Adv. Synth. Catal. 2006, 348, 1743 – 1751
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.asc.wiley-vch.de
1751