Triphenylphosphane-functionalised amphiphilic copolymers: Tailor-made support materials for the efficient and selective aqueous two-phase hydroformylation of 1-octene

Article abstract of DOI:10.1002/chem.200600718

Amphiphilic copolymers (random P1 and block P2) based on 2-oxazolines were synthesised with triphenylphosphane ligands covalently linked to the polymers by means of a metal-free synthesis route. The resulting macroligands were used in the aqueous two-phase hydroformylation of 1-octene. The influence of the polymer architecture (random and block copolymers) on activity and selectivity of the hydroformylation reaction was investigated and compared with that of nonfunctionalised copolymers (random P3 and block P4) and Rh^I/ triphenylphosphane trisulfonate as a water-soluble catalyst. The highest activities were observed for the random copolymer P1 (p = 50 bar, T = 100°C, c = 8 × 10^-4 mol L^-1) with a turnover frequency (TOF) of 3700 h^-1, whereas the corresponding block copolymer P2 reached TOF numbers of 1630 h^-1. Additionally, both macroligands indicated efficient suppression of isomerisation and led to almost constant n/iso selectivities of about 3 after complete substrate conversion. Copolymers P3 and P4 showed, under identical reaction conditions, strong isomerisation after 40-60% conversion (n/iso≈0.7) and maximum activities of 1560 h ^-1 (P3) and 1330 h^-1 (P4) at a concentration of 5 × 10^-3 mol L^-1.

Full text of DOI:10.1002/chem.200600718

DOI: 10.1002/chem.200600718
Triphenylphosphane-Functionalised Amphiphilic Copolymers: Tailor-Made
Support Materials for the Efficient and Selective Aqueous Two-Phase
Hydroformylation of 1-Octene
Martin Bortenschlager,[a] Nicole Schçllhorn,[a] Anton Wittmann,[a] and
Ralf Weberskirch*[a, b]
Abstract: Amphiphilic copolymers
(random P1 and block P2) based on 2-
oxazolines were synthesised with tri-
phenylphosphane ligands covalently
linked to the polymers by means of a
metal-free synthesis route. The result-
ing macroligands were used in the
aqueous two-phase hydroformylation
of 1-octene. The influence of the poly-
mer architecture (random and block
copolymers) on activity and selectivity
of the hydroformylation reaction was
investigated and compared with that of
nonfunctionalised copolymers (random
P3 and block P4) and RhI/triphenyl-
phosphane trisulfonate as a water-solu-
ble catalyst. The highest activities were
observed for the random copolymer P1
sponding block copolymer P2 reached
TOF numbers of 1630 hÀ1. Additional-
ly, both macroligands indicated effi-
cient suppression of isomerisation and
led to almost constant n/iso selectivities
of about 3 after complete substrate
conversion. Copolymers P3 and P4
showed, under identical reaction condi-
tions, strong isomerisation after 40–
60% conversion (n/isoꢀ0.7) and maxi-
mum activities of 1560 hÀ1 (P3) and
1330 hÀ1 (P4) at a concentration of 5”
10À3 molLÀ1.
(p=50 bar,
T=1008C,
c=8”
10À4 molLÀ1) with a turnover frequency
(TOF) of 3700 hÀ1, whereas the corre-
Keywords: amphiphiles
formylation immobilization
polymers · supported catalysts
·
hydro-
A
·
·
Introduction
transfer processes by increasing either the mutual solubility
of the components or the mobility across the aqueous–or-
ganic phase boundaries. The employment of cosolvents, such
as methanol, ethanol, or propanol;[3] the use of supported
aqueous-phase catalysts (SAPC)[4] or additives such as sur-
factants;[5] modified cyclodextrin derivatives;[6] amphiphilic
phosphane ligands with surfactant structures[7] and various
polymer-supported catalysts[8] have been reported with this
in mind. Other strategies to overcome the mutual immisci-
bility of the catalyst and the substrate include the use of so-
called “smart” ligands, which display inverse temperature
behaviour in water and transport the catalyst at higher tem-
perature in the organic phase and move it back to the aque-
ous phase at lower temperatures,[9] fluorous/organic biphasic
mixtures[10] or the application of supercritical CO2 as sol-
vent[11] in the hydroformylation.
Hydroformylation of terminal olefins is one of the most im-
portant homogeneously catalysed reactions in industry for
the large-scale production of aldehydes. Aqueous two-phase
catalysis of lower olefins by using the Ruhrchemie/Rhône–
Poulenc process has revolutionised this reaction by using an
environmentally benign reaction medium with the option of
catalyst recycling.[1,2] In the case of higher olefins, however,
the hydroformylation reactions are limited by the solubility
of the substrate in the aqueous phase.
Over the past decades many approaches have been ex-
ploited to overcome this limitation and accelerate mass-
[a] Dr. M. Bortenschlager, Dipl. Ing. N. Schçllhorn, Dipl.
Ing. A. Wittmann, Dr. R. Weberskirch
The reasons for the still somewhat lower activities com-
pared with homogeneous hydroformylation are twofold. In
the case of amphiphilic additives such as surfactants, the
Department Chemie, Lehrstuhl für Makromolekulare Stoffe
TU München, Lichtenbergstrasse 4, 85747 Garching (Germany)
Fax : (+49)892-891-3562
E-mail: Ralf.Weberskirch@mytum.de
water-soluble [Rh(CO)H(TPPTS)3] catalysts (TPPTS=tri-
A
[b] Dr. R. Weberskirch
phenylphosphane trisulfonate) as well as the substrate are,
at first glance, both in the aqueous phase. On a microscopic
level, however, the hydrophobic substrate presumably stays
Current Address:
Bayer MaterialScience AG, BMS-NB-New Technologies
Building D 167, 51368 Leverkusen (Germany)
520
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2007, 13, 520 – 528

FULL PAPER
in the hydrophobic domains of the micellar aggregates while
the catalyst is homogeneously dissolved in the aqueous
phase and thus phase-separated from the substrate. On the
other hand, the use of smart ligands, organic/fluorous sol-
vent mixtures or supercritical CO2, provide nearly homoge-
neous reaction conditions during hydroformylation. Howev-
er, the high surface area and the resulting micellar effect,
which account for the good activities when using amphi-
philes, are gone. The question remains how to combine the
advantages of both approaches: forcing the catalyst and the
substrate into the same hydrophobic domain on the one
hand, while taking advantage of the micellar effect, that is,
compartmentalisation and concentrating reactants, thereby
altering the chemical rate of the reaction, on the other
hand. In recent years our work has focussed on the synthesis
of various amphiphilic block copolymers functionalised with
metal catalysts. Beyond simple catalyst immobilisation on a
polymer, these soluble macroligands form micellar aggre-
gates in water and consequently act as nanoreactors for the
efficient conversion of hydrophobic substrates in aqueous
media.[12] In the case of aqueous two-phase catalysis, this ap-
proach becomes particularly attractive. Here the micellar
environment forces the substrate and catalyst into the same
micellar core phase, and due to the insolubility of the poly-
mer in the 1-octene substrate, the polymer-supported cata-
lyst should remain in the aqueous phase under hydroformy-
lation conditions (see Scheme 1). Here we describe the
gated as carriers for phosphane ligands and tested in the hy-
droformylation of longer-chain alkenes. However, little is
known about the effect of different copolymer compositions
on activity and selectivity in the hydroformylation reaction.
The cationic ring-opening polymerisation of 2-oxazolines
provides an excellent methodology for the preparation of
different polymer compositions, including random copoly-
mers and block copolymers.[13]
Recently, Persigehl et al. reported on the synthesis of
TPP-functionalised amphiphilic block copolymers for cataly-
sis application.[14] The TPP unit was introduced into the po-
lymer by means of the polymer-analogous coupling of diphe-
nylphosphane to the iodoaryl-containing polymer precursor
in the presence of a Pd catalyst. Although the coupling reac-
tion proceeded quantitatively, it turned out that the residual
Pd catalyst was extremely difficult to remove and effected
hydroformylation activities dramatically due to the very low
amounts of rhodium employed therein. Consequently, we
decided to develop a metal-free route to TPP-functionalised
copolymers. The introduction of the triphenylphosphane
moieties to the amphiphilic copolymers was achieved by
using a polymer-analogous amide coupling of 4-diphenyl-
phosphanobenzoic acid and a poly(2-oxazoline) with pend-
ent primary amine groups (Scheme 2). The tert-butoxycar-
bonyl (Boc)-protected amine-functionalised monomer 2-[5-
(amino-tert-butoxycarbonyl)pentyl]-2-oxazoline
(BocOx)
was synthesised according to a literature procedure.[15] In
the next step, a random copolymer (PP1) and a block copo-
lymer (PP2) consisting of 2-methyl-2-oxazoline (MeOx), 2-
nonyl-2-oxazoline (NonOx) and BocOx were synthesised by
using living cationic polymerisation. The random copolymer
PP1 was synthesised by using simple copolymerisation of
the three monomers. In the case of PP2, sequential polymer-
isation of 2-methyl-2-oxazoline was used to form the hydro-
philic block, followed by the polymerisation of a mixture of
NonOx and BocOx to give the hydrophobic block (see also
Scheme 2). Methyl triflate (CH3OTf) was used as the initia-
tor and piperidine as the highly efficient termination
agent.[16]
Scheme 1. Left: Current situation of water-soluble catalysts phase-sepa-
rated from the hydrophobic substrate dissolved in the micellar phase.
Right: Our approach of amphiphilic macroligands that form micelles
with the catalyst, and the substrate during the reaction, in the hydropho-
bic core.
The NonOx monomer is necessary to increase the hydro-
phobicity of copolymers P1 and P2. The chosen ratio of
MeOx/NonOx/BocOx was 30:4:4. The copolymers show
narrow polydispersity indexes (PDIs) of 1.15 (PP1) and 1.09
(PP2), indicative of a living polymerisation mechanism.
After the polymerisation of the precursor polymers PP1 and
PP2 was complete, free primary amine groups were ob-
tained after deprotection in methanolic hydrochloric acid.
The success of the deprotection reaction to give PP1–NH2
preparation of two amphiphilic copolymers (random and
AB block copolymers) with triphenylphosphane (TPP) units
covalently attached to the polymer backbone. The two co-
polymers were studied for the aqueous two-phase hydrofor-
mylation of 1-octene and the results were compared with
the results from two nonfunctionalised copolymers, which
were used as polymeric amphiphiles in the presence of the
well-known RhI/TPPTS system as a water-soluble catalyst.
and PP2–NH2 can be followed by observing the disappear-
1
À
ance of the H NMR signals of the CH3 Boc group at d=
1.38 and 1.36 ppm. Triphenylphosphane groups were linked
to the polymer by amide bonds because of their high stabili-
ty, especially towards hydrolysis, which is important for any
catalytic reactions carried out in water. Therefore, an excess
of 4-diphenylphosphanobenzoic acid (1.5 equiv per amine
function) was reacted with the amino-deprotected PP1–NH2
and PP2–NH2 by using dicyclohexylcarbodiimide (DCC) as
Results and Discussion
Synthesis of amphiphilic, TPP-functionalised copolymers:
Several water-soluble polymers have already been investi-
Chem. Eur. J. 2007, 13, 520 – 528
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
521

R. Weberskirch et al.
Scheme 2. Synthesis of the precursor polymer PP2 with Boc-protected amine functionalities and its subsequent conversion to the triphenylphosphane-
functionalised amphiphilic block copolymer P2 (stat.=statistical/random).
the coupling reagent (Scheme 2). The success of the reaction
was monitored through changes in the 1H (Figure 1) and
31P NMR spectra.
The random P3 and block P4 copolymers based on 2-
methyl-2-oxazoline and 2-nonyl-2-oxazoline were prepared
as references. The ratio chosen for the hydrophilic (MeOx)
to the hydrophobic (NonOx) monomers was 30:6 to provide
good water solubility of the polymers while still retaining an
amphiphilic character. The resulting copolymers show very
narrow PDIs of 1.03 for both polymers P3 and P4. The
molar masses and copolymer compositions were calculated
1
from H NMR spectra and are summarised in Table 1
Table 1. Analytical data of the copolymers PP1–P4.
[a]
[b]
Polymer Composition (1H NMR)
Mn [gmolÀ1
]
Mw/Mn
PP1
PP2
P1
P2
P3
MeOx34.9NonOx4.6BocOx4.4
MeOx33.6NonOx3.5BocOx3.8
MeOx34.9NonOx4.6TPP-Ox4.4
MeoO33.6NonOx3.5AmOx0.7TPP-Ox3.1
MeOx29.0NonOx7.1
5100
4620
5540
5123
3970
3470
1.15
1.09
1.08
1.17
1.03
1.03
P4
MeOx27.3NonOx5.3
[a] From NMR data. [b] Solvent: DMAc; RI detector; poly(methyl meth-
acrylate) calibration.
Figure 1. 1H NMR spectrum of P2 (CD2Cl2, 300.13 MHz, T=208C).
Hydroformylation of 1-octene: In the first set of experi-
ments we investigated the nonfunctionalised copolymers P3
(random copolymer) and P4 (block copolymer) in the aque-
ous two-phase hydroformylation of 1-octene at different co-
polymer concentrations. As catalyst precursor we used [Rh-
The amount of triphenylphosphane moieties that were
successfully linked to the polymer was calculated by using
1H NMR spectroscopy to be 100% for P1 and 81% for P2.
Successful coupling of the TPP ligand to the copolymers was
confirmed by using 31P NMR spectroscopy showing one
signal at d=4.81 (P1) and 4.77 ppm (P2), which can be as-
cribed to the polymer-bound 4-diphenylphosphanobenzoic
acid. Unreacted 4-diphenylphosphanobenzoic acid at d=
À4.60 ppm, as well as the corresponding phosphane oxide as
possible byproduct at d=22.7 ppm, were not observed and
supported our 1H NMR observations.
A
phosphane ligand triphenylphosphane trisulfonate (TPPTS)
to immobilise the catalyst in the water phase. The reaction
parameters were chosen to meet industrial conditions (T=
1008C, p(CO/H2)=50 bar). The ratio of ligand to rhodium
A
was 5. The results of 1-octene hydroformylation at different
copolymer concentrations are summarised in Table 2.
522
www.chemeurj.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2007, 13, 520 – 528

Tailor-Made Supports
FULL PAPER
type of copolymer studied, the
results correspond well with
the findings of Schulz who re-
ported on random poly(2-oxa-
zoline) copolymers that do not
show a critical micelle concen-
tration.[16] Instead, they seem
to be able to form intramolec-
ular aggregates even at low po-
lymer concentrations, whereas
at higher copolymer concentra-
tions intermolecular aggrega-
tion becomes dominant. This
could be one reason to explain
the high hydroformylation ac-
tivity as well as the low surface
tension at low concentrations.
Regarding the selectivity of
the hydroformylation of 1-
octene, high isomerisation ac-
tivity was observed with both
P3 and P4 as surfactant (Fig-
ures 2 and 3, respectively).
Table 2. Catalytic activities of the two-phase hydroformylation of 1-
octene using nonfunctionalised poly(2-oxazoline)s P3 and P4 as the am-
phiphiles.
[a]
[a]
cpolymer [molLÀ1
]
TOF [hÀ1
]
TOF [hÀ1
]
P3
P4
5”10À3
4”10À4
1”10À5
1”10À6
1”10À7
1560
1200
1120
840
1330
640
490
370
210
800
[a] Determined by the initial formation rate of aldehydes, olefin isomeri-
sation not taken into account; reaction conditions: T=1008C; p(CO/
H2)=50 bar; c([Rh
(acac)(CO)2])=2”10À4 molLÀ1; rhodium/TPPTS=5;
substrate/1-octene, substrate/rhodium=10000.
AHCTREUNG
ACHTREUNG
Figure 2. Product distribution as a function of conversion in the hydrofor-
mylation of 1-octene using P3 as the amphiphile (cpolymer =5”
10À3 molLÀ1); black=1-octene, grey=internal octenes, white=nonanal,
hashed area=isoaldehydes.
In general, the random copolymer P3 shows higher activi-
ties than the block copolymer P4 at the same copolymer
concentration. The activity of P3 increases steadily with in-
creasing polymer concentration with TOF numbers of
800 hÀ1 (cpolymer =10À7 molLÀ1) up to 1560 hÀ1 (cpolymer =5”
10À3 molLÀ1). The catalyst activity is already rather high
(TOF=800 hÀ1 at c=10À7 molLÀ1), even at very low P3 con-
centrations, whereas the block copolymer P4 has almost no
effect on the activity at the same concentration. However,
the hydroformylation activity drastically increased from 640
to 1330 hÀ1 by increasing the concentration of P4 from
c
polymer =4”10À4 to cpolymer =5”10À3 molLÀ1.This effect can
be ascribed to the formation of micellar aggregates in this
concentration range and the increased solubility of 1-octene
in water, which is also observed when using surfactants.[5]
The random copolymer P3 on the other hand does not
show a sudden increase of the hydroformylation activity. It
is well known that the aggregation behaviour of random co-
polymers in water is much more complex. Depending on the
Figure 3. Product distribution as a function of conversion in the hydrofor-
mylation of 1-octene using P4 as the amphiphile (cpolymer =5”
10À3 molLÀ1); black=1-octene, grey=internal octenes, white=nonanal,
hashed area=isoaldehydes.
Chem. Eur. J. 2007, 13, 520 – 528
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
523

R. Weberskirch et al.
After approximately 20% octene conversion (based on the
formation of aldehydes), 60% of 1-octene is already trans-
formed into internal octenes in both cases. This is reflected
by the n/iso ratios (see Figure 6, below), which drop from 3
at the beginning of the reaction to below 1 after 40% con-
version. The reason for the moderate selectivities is the low
phosphane/rhodium ratio of 5:1 we applied to the hydrofor-
mylation reaction. Olefin isomerisation, which needs to be
suppressed to achieve high n/iso ratios, can only be avoided
by using either bidentate phosphanes[17] or an excess of
monodentate phosphanes of up to 100 equivalents (phos-
phane/rhodium) to generate a sterically demanding environ-
ment around the rhodium centre as a prerequisite for high
selectivities.
ratios remain almost constant at 3.0 throughout the whole
catalytic process. Obviously, olefin isomerisation is efficient-
ly suppressed within the micellar aggregates even at high
substrate conversion (see Figures 4 and 5). Covalent linking
Although the activities in the presence of both copoly-
mers P3 and P4 are still rather high, n/iso selectivities de-
crease very fast with substrate conversion to values below 1.
These low values indicate that the small excess of ligands
versus rhodium with a ratio of 5 is not sufficient by far to
generate a sterically demanding environment around the
metal centre.
Figure 4. Product distribution as a function of conversion in the hydrofor-
mylation of 1-octene using P3 as the amphiphilic macroligand (cpolymer
=
8”10À4 molLÀ1); black=1-octene, grey=internal octenes, white=nona-
nal, hashed area=isoaldehydes.
In the next set of experiments we tested the amphiphilic
copolymers P1 (random) and P2 (block copolymer). As al-
ready mentioned before, covalent linking of the triphenyl-
phosphane moieties to the hydrophobic parts results in am-
phiphilic copolymers that are able to form aggregates in
water. Rhodium, as the catalytically active metal, is then
forced into the hydrophobic domains as a result of complex-
ation with the hydrophobic phosphane ligands. As a conse-
quence, the active catalyst should therefore be in direct con-
tact with the hydrophobic substrate, which is also solubilised
inside the aggregates. Reaction conditions were the same as
Figure 5. Product distribution as a function of conversion in the hydrofor-
described above (T=1008C, p(CO/H2)=50 bar). Both co-
A
mylation of 1-octene using P4 as the amphiphilic macroligand (cpolymer
=
8”10À4 molLÀ1); black=1-octene, grey=internal octenes, white=nona-
nal, hashed area=isoaldehydes.
polymers show high activities, which are summarised in
Table 3 for two different concentrations each. In particular,
of the TPP ligand to a polymer backbone is crucial for main-
taining high selectivities. Amphiphilic ligands such as phos-
phanoethylsulfonatoalkyl thioethers still show high isomeri-
sation ratios under similar reaction conditions and Rh/
ligand ratios of 5, as can be seen in a very interesting report
recently published by Oheme and co-workers.[5c]
The results of n/iso selectivities for all four copolymer sys-
tems P1–P4 are again summarised in Figure 6. According to
these results, the micellar core already provides a sterically
demanding environment for the rhodium catalyst at very
low ligand to metal ratios of 5. These conditions lead to se-
lectivities that can only be achieved under typical biphasic
conditions with a RhI/TPPTS catalyst system when using a
large excess of ligand versus metal.
To get a better appreciation for the excellent selectivities
obtained in the presence of the functionalised macroligands,
we carried out a homogeneous hydroformylation of 1-
octene in toluene using triphenylphosphane as the ligand at
different ligand/Rh ratios. The results are shown in Figure 7.
At TPP/Rh ratios of 2 and 5 no effect of the conversion on
the n/iso selectivity of the reaction products can be detected,
Table 3. Catalytic activities of the two-phase hydroformylation of 1-
octene using triphenylphosphane-functionalised poly(2-oxazolines) P1
and P2 as the amphiphiles.
Polymer
cpolymer [molLÀ1
]
TOF[a] [hÀ1
]
P1
P1
P2
P2
4”10À4
8”10À4
4”10À4
8”10À4
2200
3700
1100
1630
[a] Determined by initial formation rate of aldehydes, olefin isomerisa-
tion not taken into account; reaction conditions: T=1008C, p(CO/H2)=
AHCTREUNG
50 bar, ligand/rhodium=5, substrate: 1-octene, substrate/rhodium=
10000.
the TOF numbers of the functionalised random copolymer
P1 is significantly higher than those of the nonfunctionalised
copolymers P3 at similar amphiphile concentrations, sug-
gesting excellent substrate/catalyst contact during hydrofor-
mylation.
More surprising, however, are the n/iso selectivities ob-
served in these hydroformylation experiments. In contrast to
the nonfunctionalised copolymers P3 and P4, the n/iso
524
www.chemeurj.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2007, 13, 520 – 528

Tailor-Made Supports
FULL PAPER
Figure 6. n/iso ratios as a function of conversion of the hydroformylation
*
Figure 8. Results of the surface-tension measurements: =nonfunction-
*
of 1-octene: =nonfunctionalised random copolymer P3 (cpolymer =5”
10À3 molLÀ1),
=functionalised random copolymer P1 (cpolymer =8”
*
alised random copolymer P3, =functionalised random copolymer P1,
=nonfunctionalised block copolymer P4, & =functionalised block co-
*
10À4 molLÀ1), =nonfunctionalised block copolymer P4 (cpolymer =5”
10À3 molLÀ1) and & =functionalised block copolymer P2 (cpolymer =8”
10À4 molLÀ1).
^
^
polymer P2.
ered the surface tension to about 60 mNmÀ1; the block co-
polymers had little influence on the surface tension. When
the functionalised polymers (P1, P2) were compared with
their nonfunctionalised counterparts (P3, P4), it was evident
that the surface activity seems to be strongly dependent on
the polymer architecture (random or block copolymer).
However, the different copolymer compositions for both the
random copolymers and the block copolymers seemed to
have had very little effect on surface tension.
Recycling experiments: In the last set of experiments we
wanted to study the possibility of separating the polymer-
bound catalyst and reusing it in a new hydroformylation
cycle. Both functionalised copolymers P1 and P2 were stud-
ied at a concentration of 4”10À4 molLÀ1. After the reaction
was completed, both copolymers P1 and P2 showed distinct
differences after phase separation (see Figure 9). In the
presence of random copolymer P1, only poor phase separa-
tion occurred and the macroligand accumulated preferably
at the interface and partly even in the organic phase. In con-
trast, phase separation occurred immediately in the presence
of block copolymer P2 after cooling down. The organic
phase was nearly colourless while the macroligand seemed
to be homogeneously dissolved in the aqueous phase, again
reflecting the differences in interfacial behaviour of the
Figure 7. Homogeneous hydroformylation of 1-octene in toluene at dif-
ferent triphenylphosphane/rhodium ratios.
and only a slight increase of the selectivity (to 1.2 and 1.7)
can be detected at ligand/metal ratios of 10 and 25, respec-
tively, at 80% conversion. Only when using a 500-fold
excess of phosphane versus rhodium were n/iso values of 2.9
observed at 90% conversion.
In the last set of experiments we wanted to get a better
understanding of why the random copolymers P1 and P3
showed higher activities in all experiments when compared
with the block copolymers P2 and P4. Aside from the al-
ready well-known differences in aggregate formation in
aqueous media between block copolymers and random co-
polymers, the interfacial behaviour of these two types of co-
polymers at the water–hydrophobic interface will be crucial
for the transport of 1-octene across the water–octene inter-
face inside the micellar domains. Therefore, surface-tension
measurements were conducted for all four copolymers at
the air–water interface at different concentrations by using a
ring tensiometer (see Figure 8).
The results of the surface-tension measurements show a
clear difference between the random copolymers (P3, P1)
and the block copolymers (P4, P2). The random copolymers
reduce the surface tension much more than the correspond-
ing block copolymers at any given concentration. The
random copolymers revealed high surface activities even at
very low polymer concentrations (c=10À6 molLÀ1) and low-
Figure 9. Phase separation of the aqueous biphasic 1-octene hydroformy-
lation mixture in the presence of block copolymer P2 (left) and the
random copolymer P1 (right).
Chem. Eur. J. 2007, 13, 520 – 528
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
525

R. Weberskirch et al.
random copolymers versus the block copolymer. The block
copolymer system P2 was further considered for recycling
experiments (four cycles, each 240 min of reaction time at
T=1008C and with a synthesis gas pressure of 50 bar). The
results are summarised in Table 4.
phenylphosphane ligands covalently attached to the polymer
backbone results in aggregates in the aqueous phase, with
the rhodium catalyst located in the hydrophobic domains.
The substrate is also predominately solubilised in these do-
mains during hydroformylation. Our method was able to
provide the product(s) with excellent activities, with TOF
numbers of up to 3700 hÀ1, and constant n/iso ratios of 3.0 at
remarkably low ligand/metal ratios of 5. In addition, sur-
face-tension measurements of the random versus block co-
polymer indicated strong differences in their interfacial be-
haviour, in agreement with the catalytic activities. The
simple design of the macroligands provides a novel and effi-
cient way for the hydroformylation of long-chain alkenes
under aqueous biphasic conditions, and might open new pos-
sibilities for aqueous biphasic reactions of hydrophobic sub-
strates in general.
Table 4. Catalytic activities of the two-phase hydroformylation of 1-
octene using P2 as the amphiphile in four consecutive cycles.
Cycle
Conversion [%]
TOF [hÀ1
]
Are[a] [%]
1
2
3
4
29
18
7
760
450
170
100
100
59
23
4
13
[a] Are =activity compared with the first cycle.
The first cycle was set to 100% activity, and the activity
drops dramatically thereafter. In the fourth cycle only 13%
activity is retained compared with the first cycle, which is in
excellent agreement with the recent results of Oehme and
co-workers. Two reasons might account for this observation.
The first possibility is that the catalyst is oxidised during re-
cycling of the catalyst system; the second reason for the ob-
servation is related to the stability of the rhodium hydride
complex. To address this problem we conducted two experi-
ments under homogeneous conditions: in the first experi-
ment an aqueous solution and CTAB/TPPTS/[Rh-
Experimental Section
Acronyms: 2-(5-Aminopentyl)-2-oxazoline (AmOx), N,N-dimethylaceta-
mide (dmac), piperidine (pip), 2-{5-[(3-diphenylphosphano)benzamidyl]-
pentyl}-2-oxazoline (tppox).
Measurements: 1H (300.13 MHz) and 31P NMR (121.50 MHz) spectra
were recorded on a Bruker ARX 300 spectrometer. Gel permeation
chromatography (GPC) was carried out on a Waters GPC 510 equipped
with UV and refractive index (RI) detectors using polystyrene calibration
standards for the poly(2-oxazoline) samples in dmac solvent. Gas chro-
matography analyses were performed on a Varian CP-3380 equipped
with a flame ionisation detector FID/1177, capillary column CP-Sil 8 CB
(length 25 m), with helium as the mobile phase. Surface tensions were
measured with a Lauda ring tensiometer TE1C.
ACHTREUNG
AHCTREUNG
hydrophobic catalyst. The reactor was cooled down after a
4 h reaction with 1-octene, depressurised, and more 1-octene
was then added. Whereas the activity of the TPP/toluene
mixture remained constant in the second cycle, the activity
of the aqueous mixture was only 25% compared with the
first run. These results led us to the conclusion that water
plays a key role in deactivating the rhodium hydride species
under aqueous biphasic conditions. Obviously, the catalyst is
only active under hydroformylation conditions, suggesting
that catalyst recycling and reuse in the Ruhrchemie/Rhône–
Poulenc process is more complex and requires more process
know-how.
As an important note, it should be emphasised that the
problems of catalyst recycling, when using rhodium/phos-
phane systems under aqueous two-phase conditions, can be
overcome by using, for example, an N-heterocyclic carbene/
rhodium complex as previously published. Simple catalyst
recycling was achieved under normal laboratory conditions,
and no catalyst deactivation was observed even after four
cycles of aqueous two-phase hydroformylation of 1-octe-
ne.[12b]
Materials: All chemicals were purchased from Aldrich and Fluka and
used, unless otherwise noted, without further purification. Water-free sol-
vents were purchased from Fluka (dichloromethane) or dried by using
standard procedures (diethyl ether, KOH molecular sieve 4 Š, BTS cata-
lyst; acetonitrile and chlorobenzene, CaH2). Synthesis gas (V(CO)=
50%, rest H2) was purchased from Linde AG. Liquid chemicals and sol-
vents used in the hydroformylation reactions were degassed and saturat-
ed with nitrogen.
Synthesis of polymers: A 100 mL polymerisation vessel was filled with
acetonitrile (40 mL), chlorobenzene (10 mL) and methyltriflate (1 equiv)
resulting in a 20–40 mmolar solution. To this solution, 2-methyl-2-oxazo-
line (30 equiv), 2-nonyl-2-oxazoline (P3, P4: 6 equiv; PP1, PP2: 4 equiv)
and 2-[5-(amino-tert-butoxycarbonyl)pentyl]-2-oxazoline (PP1, PP2:
4 equiv) were added. The reaction mixture was heated to 908C and stir-
red for 12 h.
In the case of the block copolymers (PP2, P4), 2-methyl-2-oxazoline was
added first; the monomers forming the second block were added after
12 h of stirring at 908C. The mixture was then stirred for another 12 h at
908C.
After cooling the solution to 408C, piperidine (2.5 equiv) was added as
the terminating agent and the mixture was stirred for 4 h. The solvent
was separated by means of rotary evaporation and the solid residue was
diluted in chloroform (20 mL). Potassium carbonate (50 equiv) was
added and the solution was stirred for 4 h at room temperature. The po-
lymer was precipitated in diethyl ether, filtered and dried under reduced
pressure.
Conclusion
Copolymer PP1: Yield 3.86 g (79%); composition (determined by
1H NMR end-group analysis): MeOx34.9NonOx4.6BocOx4.4; GPC (DMAc,
A new approach to a polymer-supported catalyst for aque-
ous-two phase hydroformylation of 1-octene was developed.
The use of an amphiphilic, water-soluble copolymer with tri-
RI, poly(methyl methacrylate) calibration): Mn =6454 gmolÀ1
, Mw =
7422 gmolÀ1, PDI=1.15; 1H NMR (CDCl3, 300.13 MHz, 208C): d=0.83
(t; NonOx-CH3), 1.21 (brs; CH2), 1.38 (s; Boc-CH3), 1.56 (brs;
526
www.chemeurj.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2007, 13, 520 – 528

Tailor-Made Supports
FULL PAPER
COCH2CH2, pip-C3H2, pip-C4H2), 2.03–2.40 (m; COCH3, COCH2, pip-
ic acid group); 31P{1H} NMR (CDCl3, 121.50 MHz, 208C): À4.81 ppm (s;
2
À
À
C H2, Boc-NH2 CH2 CH2), 2.91 (s; CH3 end group, ZCH3,C=O), 3.05 (brs,
P-phosphane).
À
Boc-NH CH2), 3.14 (CH3 end group, ECH3,C=O), 3.42 (brs, CH2 back-
Copolymer P2: Yield 0.210 g (67%); composition (determined by
1H NMR end-group analysis): MeOx33.6NonOx3.5AmOx0.7TPPOx3.1; GPC
bone), 4.87 ppm (brs; Boc-NH).
(DMAc, RI, poly(methyl methacrylate) calibration): Mn =6470 gmolÀ1
,
Copolymer PP2: Yield 2.58 g (56%); composition (determined by
PDI=1.17; 1H NMR (CD2Cl2, 300.13 MHz, 208C): d=0.88 (t; NonOx-
1H NMR end-group analysis): MeOx33.6NonOx3.5BocOx3.8; GPC (DMAc,
CH3), 1.24 (brs; CH2), 1.55 (brs; COCH2CH2, pip-C3H2, pip-C4H2), 2.08,
RI, poly(methyl methacrylate) calibration): Mn =6138 gmolÀ1
, Mw =
2
6690 gmolÀ1, PDI=1.09; 1H NMR (CDCl3, 300.13 MHz, 208C): d=0.81
À
2.11, 2.30 (3”s; COCH3, COCH2, pip-C H2), 2.59 (brs; PhC(O)NH
À
CH2), 2.87 (brs; NH2 CH2), 3.02 (CH3 end group, ECH3,C=O), 3.42 (brs;
(t; NonOx-CH3), 1.19 (brs; CH2), 1.36 (s; Boc-CH3), 1.53 (brs;
COCH2CH2, pip-C3H2, pip-C4H2), 2.01–2.27 (m; COCH3, COCH2, pip-
backbone CH2 ), 6.95 (t; m-CH of benzoic acid group), 7.29 (s; phenyl
CH), 7.71 ppm (d; o-CH of benzoic acid group); 31P{1H} NMR (CDCl3,
121.50 MHz, 208C): À4.77 ppm (s; P-phosphane).
2
À
À
C H2), 2.61 (s; Boc-NH2 CH2 CH2), 2.88 (s; CH3 end group, ZCH3,C=O),
À
3.00 (brs; Boc-NH CH2), 3.11 (CH3 end group, ECH3,C=O), 3.39 (brs;
backbone CH2), 4.86 ppm (brs; Boc-NH).
Hydroformylation experiments: All experiments were carried out in a
300 mL Parr high-pressure reactor.
Copolymer P3: Yield 3.05 g (72%); composition (determined by
1H NMR end-group analysis): MeOx29.0NonOx7.1; GPC (dmac, RI, poly(-
Aqueous two-phase experiments: The reactor was evacuated, flushed with
nitrogen and filled with polymer, [Rh(acac)(CO)2] (3–10 mg, depending
methyl methacrylate) calibration): Mn =4586 gmolÀ1, Mw =4733 gmolÀ1
,
A
PDI=1.03; 1H NMR (CDCl3, 300.13 MHz, 208C): d=0.90 (t; NonOx-
CH3), 1.29 (brs; CH2), 1.63 (brs; COCH2CH2, pip-C3H2, pip-C4H2), 2.11,
2.13, 2.17 (3”s; COCH3, COCH2), 2.39 (brs; pip-C2H2), 3.00 (s; CH3 end
group, ZCH3,C=O), 3.08 (s; CH3 end group, ECH3,C=O), 3.50 ppm (brs; back-
bone CH2).
on polymer concentration, leading to a phosphane/rhodium ratio of 5:1),
water (50 mL) and 1-octene leading to a substrate/catalyst ratio of 10000.
For the case of nonfunctionalised polymers (P3 and P4), triphenylphos-
phanotrisulfonate sodium salt was added (phosphane/rhodium ratio=
5:1). Undecane was added as the internal standard. The mixture was
pressurised twice to 30 bar synthesis gas (CO/H2 =1:1) to clean all sup-
plies before the pressure was adjusted to 50 bar with a back-pressure reg-
ulator. The autoclave was then heated to 1008C and kept at this tempera-
ture. Samples were taken every 15 to 30 min and the products were quan-
tified by using gas chromatography.
Copolymer P4: Yield 4.54 g (91%); composition (determined by
1H NMR end-group analysis): MeOx27.3NonOx5.3
;
GPC (DMAc, RI,
poly(methyl methacrylate) calibration): Mn =4433 gmolÀ1
,
Mw =
4551 gmolÀ1, PDI=1.03; 1H NMR (CDCl3, 300.13 MHz, 208C): d=0.89
(t; NonOx-CH3), 1.27 (brs; CH2), 1.60 (brs; COCH2CH2, pip-C3H2, pip-
C4H2), 2.09, 2.12, 2.16 (3”s; COCH3, COCH2), 2.32 (brs; pip-C2H2), 2.97
(s; CH3 end group, ZCH3,C=O), 3.07 (s; CH3 end group, ECH3,C=O),
3.47 ppm (brs; backbone CH2).
Homogeneous hydroformylation: The reactor was evacuated, flushed
with nitrogen and filled with degassed toluene (100 mL), [Rh-
(acac)(CO)2] (5 mg, 4”10À4 molLÀ1; 1 equiv), triphenylphosphane (1–
500 equiv) and 1-octene leading to a substrate/catalyst ratio of 10000.
Undecane (2 mL) was added as the internal standard. The mixture was
pressurised twice to 30 bar synthesis gas (CO/H2 =1:1) to clean all sup-
plies before the pressure was adjusted to 50 bar with a back-pressure reg-
ulator. The autoclave was then heated to 1008C and kept at this tempera-
ture. Samples were taken every 15 to 30 min and the products were quan-
tified by using gas chromatographic analysis before the reaction was ter-
minated after 5 h.
Polymer-analogous removal of Boc-groups (PP1-NH2, PP2–NH2): Boc-
functionalised polymer PP1 or PP2 (1 g) was dissolved in a mixture of
methanol (20 mL) and hydrochloric acid (2 mL, 37%). The solution was
stirred for 2 h at 608C. After cooling the solution to room temperature,
potassium carbonate (1 g) was added and the suspension was stirred for
another 4 h. After separating the polymer solution from excessive potas-
sium carbonate by means of filtration, the polymer was precipitated in di-
ethyl ether, filtered and dried under reduced pressure.
Copolymer PP1-NH2: Yield 2.70 g (82%); GPC (DMAc, RI, poly(meth-
yl methacrylate) calibration): Mn =6214 gmolÀ1 PDI=1.19; 1H NMR
,
(CDCl3, 300.13 MHz, 208C): d=0.85 (t; NonOx-CH3), 1.23 (brs; CH2),
1.58 (brs; COCH2CH2, pip-C3H2, pip-C4H2), 2.06–2.33 (m; COCH3,
[1] B. Cornils, W. A. Herrmann, R. Schlçgl, C.-H. Wong, Catalysis from
A to Z, Wiley, Weinheim, 2000.
[2] H. Chen, Y. Li, J. Chen, P. Cheng, Y. He, X. Li, J. Mol. Catal. A
1999, 149, 1–6.
[3] a) F. Monteil, R. Queau, P. Kalck, J. Organomet. Chem. 1994, 480,
177–184; b) P. Purwanto, H. Delmas, Catal. Today 1995, 24, 135–
140.
[4] a) J. P. Arhancet, M. E. Davis, J. S. Merola, B. E. Hanson, Nature
1989, 339, 454–455; b) J. P. Arhancet, M. E. Davis, J. S. Merola,
B. E. Hanson, J. Catal. 1990, 121, 327–339.
2
À
COCH2, pip-C H2), 2.83 (brs; NH2 CH2, CH3 end group, ZCH3,C=O), 3.01
(CH3 end group, ECH3,C=O), 3.44 ppm (brs; CH2 backbone).
Copolymer PP2-NH2: Yield 4.55 g (93%); GPC (DMAc, RI, poly(meth-
yl methacrylate) calibration): Mn =6104 gmolÀ1 PDI=1.18; 1H NMR
,
(CDCl3, 300.13 MHz, 208C): d=0.85 (t; nonOx-CH3), 1.24 (brs; CH2),
1.57 (brs; COCH2CH2, pip-C3H2, pip-C4H2), 2.06–2.38 (m; COCH3,
2
À
COCH2, pip-C H2), 2.68 (brs; NH2 CH2), 3.02 (CH3 end group, ECH3,C=
O), 3.44 ppm (brs; backbone CH2).
Synthesis of phosphane-functionalised macroligand (P1, P2): Amine-
functionalised polymer PP1-NH2, PP2-NH2 (0.4 mmol, 4 equiv of amine
functions) was diluted in degassed dichloromethane (60 mL). Dicyclohex-
ylcarbodiimide (2.9 mmol, 7.2 equiv) and 4-diphenylphosphanobenzoic
acid (2.4 mmol, 6 equiv) were added and the reaction mixture was stirred
at room temperature for 48 h. After addition of potassium carbonate
(2 g) and stirring for another 8 h, the white solid was separated by filtra-
tion. The clear polymer solution was poured into degassed diethyl ether.
The precipitate was filtered under argon atmosphere and dried under re-
duced pressure.
[5] a) H. Chen, Y. Li, J. Chen, P. Cheng, Y. He, X. Li, J. Mol. Catal. A
1999, 149, 1–6; b) M. Li, X. Hu, H. Chen, H. J. Zheng, X. Y. Huang,
Y. Z. Li, X. J. Li, Chin. Chem. Lett. 2004, 15, 1022–1023; c) E. Pet-
zold, G. Oehme, C. Fischer, M. Frank, J. Mol. Catal. A 2003, 200,
95–103; d) C. C. Miyagawa, J. Kupka, A. Schumpe, J. Mol. Catal. A
2005, 234, 9–17.
[6] L. Leclerq, M. Sauthier, Y. Castanet, A. Mortreux, H. Bricout, E.
Monflier, Adv. Synth. Catal. 2005, 347, 55–59.
[7] a) H. Ding, B. E. Hanson, T. Bartik, Organometallics 1994, 13,
3761–3763; b) B. E. Hanson, Coord. Chem. Rev. 1999, 185/186, 795–
807; c) S. Bischoff, M. Kant, Ind. Eng. Chem. Res. 2000, 39, 4908–
4913; d) M. Schreuder Goedheijt, B. E. Hanson, J. Reek, P. C. J.
Kamer, P. W. N. M. van Leeuwen, J. Am. Chem. Soc. 2000, 122,
1650–1657.
[8] a) J. Chen, H. Alper, J. Am. Chem. Soc. 1997, 119, 893–895;
b) A. N. Ajjou, H. Alper, J. Am. Chem. Soc. 1998, 120, 1466–1468.
[9] a) X. Zheng, J. Jiang, X. Liu, Z. Jin, Catal. Today 1998, 44, 175–182;
b) Z Jin, X. Zheng, B. Fell, J. Mol. Catal. A 1997, 116, 55–58.
P1: Yield 2.70 g (46%) GPC (DMAc, RI, poly(methyl methacrylate) cal-
ibration): Mn =5970 gmolÀ1
,
PDI=1.08; composition (determined by
1H NMR end-group analysis): MeOx34.9NonOx4.6TPPOx4.4 1H NMR
.
(CD2Cl2, 300.13 MHz, 208C): d=0.87 (t; Nonox-CH3), 1.26 (brs; CH2),
1.54 (brs; COCH2CH2, pip-C3H2, pip-C4H2), 2.04, 2.29 (s, brs; COCH3,
2
À
À
COCH2, pip-C H2), 2.60 (brs; NH2 CH2), 2.73 (brs; PhC(O)NH CH2),
2.93 (CH3 end group, ECH3,C=O), 3.40 (brs; CH2 backbone), 7.00 (t; m-CH
of benzoic acid group), 7.22 (s; phenyl CH), 7.75 ppm (d; o-CH of benzo-
Chem. Eur. J. 2007, 13, 520 – 528
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
527

R. Weberskirch et al.
[10] I. T. Horvath, G. Kiss, R. A. Cook, J. E. Bond, P. A. Stevens, J.
Rabaj, E. J. Mozeleski, J. Am. Chem. Soc. 1998, 120, 3133–3143.
[11] D. Koch, W. Leitner, J. Am. Chem. Soc. 1998, 120, 13398–13404.
[12] a) M. T. Zarka, O. Nuyken, R. Weberskirch, Chem. Eur. J. 2003, 9,
3228–3234; b) M. T. Zarka, M. Bortenschlager, K. Wurst, O.
Nuyken, R. Weberskirch, Organometallics 2004, 23, 4817–4820;
c) M. T. Zarka, O. Nuyken, R. Weberskirch, Macromol. Rapid
Commun. 2004, 25, 858–862; d) T. Kotre, O. Nuyken, R. Weber-
skirch, Macromol. Chem. Phys. 2004, 205, 1187–1196; e) D. Schçn-
felder, K. Fischer, M. Schmidt, O. Nuyken, R. Weberskirch, Macro-
molecules 2005, 38, 254–262.
[14] P. Persigehl, R. Jordan, O. Nuyken, Macromolecules 2000, 33, 6977–
6981.
[15] S. Cesana, J. Auernheimer, R. Jordan, H. Kessler, O. Nuyken, Mac-
romol. Chem. Phys. 2006, 207, 183–192.
[16] R. C. Schulz, Makromol. Chem. Macromol. Symp. 1993, 73, 103–
115.
[17] a) Z. Freixa, P. W. N. M. van Leeuwen, Dalton Trans. 2003, 10,
1890–1901; b) J. I. vam der Vlugt, R. vam Duren, G. D. Datema, R.
den Heeten, A. Meetma, J. Fraanje, K. Goubitz, P. C. J. Kamer,
P. W. N. M. van Leeuwen, D. Vogt, Organometallics 2005, 24, 5377–
5382; c) H. Chen, Y. Z. Li, R. X. Li, P. M. Cheng, X. J. Li, J. Mol.
Catal. A 2003, 198, 1–7.
[13] a) S. Kobayashi, Prog. Polym. Sci. 1990, 15, 751–824; b) K. Aoi, M.
Okada, Prog. Polym. Sci. 1996, 21, 151–208; c) S. Kobayashi, H.
Uyuma, J. Polym. Sci. A 2002, 40, 192–209.
Received: May 23, 2006
Published online: October 16, 2006
528
www.chemeurj.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2007, 13, 520 – 528