A cartridge system for organometallic catalysis: Sequential catalysis and separation using supercritical carbon dioxide to switch phases

Article abstract of DOI:10.1002/anie.200461768

(Figure Presented) A judicious combination of an organometallic catalyst and workup with supercritical CO₂ resulted in a recyclable system for mediating a variety of transformations. Between reactions the catalyst is "switched off" and recovered as a precipitate from the reaction mixture when ScCO₂ is introduced. Simultaneously the product is isolated by extraction with ScCO₂. The key is the unique solubility properties of the Rh catalyst arising from the poly-(ethyleneglycol) chains in the phosphine ligands (see scheme).

Full text of DOI:10.1002/anie.200461768

Angewandte
Chemie
Green Chemistry
A Cartridge System for Organometallic Catalysis:
Sequential Catalysis and Separation Using
Supercritical Carbon Dioxide to Switch Phases**
Maurizio Solinas, Jingyang Jiang, Othmar Stelzer†, and
Walter Leitner*
There is continued interest in homogeneous organometallic
catalysis[1] as an important synthetic tool in the research and
production of fine chemicals and pharmaceuticals. As a
consequence, effective separation techniques are needed
increasingly for the clean isolation of products and catalyst
recycling.[2,3] A promising approach toward this goal is
provided by regulated systems in which a defined substitution
pattern in the periphery of the metalꢀs coordination sphere
affects the catalystꢀs solubility, making it possible to switch
between two states: the catalyst is homogeneously dissolved
during the reaction and precipitates quantitatively in the
separation stage.[4] The most commonly used parameters to
induce the precipitation are temperature,[5] solvent polarity,[6]
and pH of the solution.[7] It has been shown also that
compressed and in particular supercritical carbon dioxide[8]
(scCO2, Tc = 31.08C, pc = 73.75 bar) can be used to control the
solubility of transition-metal catalysts, such that one can
switch from the reaction mode to the separation stage in
certain cases.[9,10]
Ideally, in a regulated system all reaction components are
removed from the precipitated catalyst quantitatively under
mild conditions to ensure that the catalyst remains available
for repeated application in homogenous form after redissolv-
ing. In particular, it would be attractive to devise a “catalyst
cartridge” with a single catalyst batch for a range of
transition-metal-catalyzed conversions utilizing different sub-
strates or even different reaction types in the same apparatus
[*] Dr. M. Solinas, Dr. J. Jiang, Prof. Dr. W. Leitner
Institut fꢀr Technische und Makromolekulare Chemie
RWTH Aachen
Worringer Weg 1, 52074 Aachen (Germany)
Fax: (+49)241-8022177
E-mail: leitner@itmc.rwth-aachen.de
Prof. Dr. W. Leitner
Max-Planck-Institut fꢀr Kohlenforschung
Kaiser-Wilhelm-Platz 1, 45472 Mꢀlheim an der Ruhr (Germany)
Dr. J. Jiang, Prof. Dr. O. Stelzer†+
FB 9, Anorganische Chemie
Universitꢁt Wuppertal
Gaussstrasse 20, 42097 (Germany)
Dr. J. Jiang
Permanent address:
Dalian Institute of Technology, Dalian 116023 (P.R. China)
[+] Deceased on January 27, 2002.
[**] This work was supported by the German Ministry of Research and
Education (BMBF) under the ConNeCat program as part of the
project “Regulated Systems for Multiphase Catalysis—smart sol-
vents/smart ligands”, the Deutsche Forschungsgemeinschaft, and
the Fonds der Chemischen Industrie.
Angew. Chem. Int. Ed. 2005, 44, 2291 –2295
DOI: 10.1002/anie.200461768
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2291

Communications
with an integrated separation technique.[11]
Several limitations hamper the development
of efficient systems for cartridge catalysts
based on the more conventional regulated
systems:
1) Thermal and/or chemical stress can lead to
deactivation of the organometallic cata-
lyst.
2) Additives used to induce precipitation
may accumulate in the catalyst compart-
ment or require additional operations for
purification of the product stream.
Scheme 1. Hydroformylation as a benchmark reaction to validate the cartridge catalyst
system 1/[Rh(acac)(CO)2]/scCO2.
(Scheme 1). In a first set of experiments we found that the
3) Reaction and separation are difficult to integrate if
pressurized gases are used as reagents.
4) As the catalyst must be soluble in a broad range of
reaction mixtures of different polarities, it can be difficult
or even impossible to find a suitable solvent for separation
without cross-contamination and catalyst leaching.
hydroformylation of 2a catalyzed by a catalyst formed in situ
from 1 and [Rh(acac)(CO)2] (acac = acetylacetonate; P/Rh =
5:1) could be “switched off” completely by the introduction of
CO2 into the reactor. Thus, conversion achieved within two
hoursꢀ reaction time (T= 708C, p(H2/CO) = 50 bar) dropped
from 99% in the absence of CO2 to 66.4% at a density
d(CO2) = 0.35 gmLÀ1 to 0% at d(CO2) = 0.57 gmLÀ1. Visual
inspection revealed that this complete stop of the reaction was
accompanied by precipitation of a yellow-orange solid,
demonstrating the perfect separation of the catalyst and
substrate 2a.[14] When CO2 was introduced after the two-hour
reaction time, the same efficient separation was induced in
the product mixture (Figure 2). A mixture of 3a and 4a
We report here on the design and application of a system
approaching the cartridge-catalyst ideal based on the use of a
poly(ethyleneglycol) (PEG) modified phosphine ligand in a
CO2-regulated reaction/separation sequence (Figure 1).
In contrast to liquid solvent systems, separation with
scCO2 can be regulated through a combination of polarity and
volatility.[12] The design of ligand 1[13] reflects this principle. A
PEG chain of moderate length (Mw = 750, n = 16) was used
having a medium polarity compatible with a broad range of
chemical environments and an almost negligible vapor
pressure considerably below that of typical reaction products
in fine-chemical applications. This substitution pattern makes
catalysts based on ligand 1 soluble in a variety of organic
substrates without the need for additional solvents; at the
same time they are expected to be completely insoluble in
scCO2 even at high density.
Figure 2. CO2-induced quantitative precipitation of the rhodium cata-
lyst formed from 1 and [Rh(acac)(CO)2] during hydroformylation of 2a.
a) Reaction mixture without CO2, b) during addition of CO2, and
We chose the hydroformylation of 1-octene (2a) initially
as our benchmark reaction to validate these considerations
c) after introduction of CO2 at a density of approximately 0.5 gmLÀ1
.
Figure 1. The cartridge catalyst system should be flexible enough such that any of a variety of transition-metal-catalyzed conversions of substrates
Si with different reagents Rx can give the desired product Pix by means of one single catalyst batch with the same apparatus and separation tech-
niques. In the present case, a PEG-modified phosphine ligand defines the solubility properties of a rhodium catalyst, and scCO2 is used to switch
from the reaction stage (homogeneously dissolved catalyst) to the separation/extraction stage (precipitated catalyst).
2292
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2005, 44, 2291 –2295

Angewandte
Chemie
(n/iso = 2.5:1) was isolated quantitatively in solvent-free form
with low levels of metal contamination (ca. 5 ppm, vide infra)
by extraction with scCO2. The reaction/separation sequence
was repeated successfully six times with no significant changes
in conversion or selectivity (conversion: 99.3–99.7%, n/iso:
2.3:1–2.5:1, isomerization: 1.6–2.5%, recovery of organic
material: 91–103%).
Based on these findings, a catalyst cartridge based on
1/[Rh(acac)(CO)2] (P/Rh = 5:1) was successfully imple-
mented for the hydroformylation of a set of structurally
diverse olefins 2a–e using scCO2 to induce a phase switch and
mass separation (Table 1).[11] After a recycling experiment
impossible, to achieve a comparable efficient separation by
standard solvent extraction owing to the similarities in
polarity between catalyst and some of the substrates.
Encouraged by the promising performance of the car-
tridge system 1/[Rh(acac)(CO)2]/scCO2 for hydroformyl-
ation, we wanted to extend our approach to a series of
different reaction types. It is now necessary that the precipi-
tated catalyst is able to enter different catalytic cycles
independent of its previous history. The hydrogenation,
hydroformylation, and hydroboration of styrene 2e were
chosen as test reactions for our catalyst cartridge (Scheme 2).
All three reactions involve phosphine rhodium hydride
fragments as part of the active
intermediates, providing a reason-
Table 1: Hydroformylation of 2a–e using MeOPEG750-PPh2 (1)/[Rh(acac)(CO)2]/scCO2.
able chance of linking the different
catalytic cycles.
Cycle[a]
Substr.
Substr./Rh
Conv. [%]
Prod., sel. [%]
scCO2 extraction[b]
T [8C]/p [bar]/V [L]
Recov. [%]
We were very pleased to find
that the reaction sequence shown
in Scheme 2 can be indeed realized
with the cartridge system 1/[Rh
1
2
3
4
5
6
7
8
9
2a
2a
2c
2c
2d
2d
2e
2e
2a
984
1010
1018
1003
997
1010
1039
1015
1056
>99
>99
>99
>99
>99
>99
>99
>99
94.1
3a, 73.7
3a, 72.2
3c, 70.6
3c, 70.6
3d, 68.8
3d, 68.9
4e, 90.0
4e, 89.8
3a, 73.0
45–50/80–96/130
45–50/85–95/140
45–50/120–130/188
45–50/120–140/200
45–50/90–100/120
45–50/95–105/130
45–50/86–95/115
45–50/88–100/120
45–50/80–100/100
87.0
92.7
99.8
100
100
100
91.0
91.0
86.2
(acac)(CO)2]/scCO2.
Excellent
results were obtained over
extended periods of time inde-
pendent of the starting point and
the order of the subsequent reac-
tions. Figure 3 shows the results of
a representative series starting with
hydroboration and following the
[a] For details see the Experimental Section. [b] The exit flow was maintained at 0.5–0.7 LminÀ1; the total
volume of CO2 is given as a liter of gas at standard conditions.
with 2a as described above, the substrate was changed to 2b,
which was equally well converted and isolated in two
subsequent runs. The same was shown for styrene (2e) and
the functionalized olefin 2d. In cycle 9 substrate 2a was
employed again; the results demonstrated that the catalyst
had largely preserved its high activity and selectivity through-
out the procedure. This was further confirmed by an addi-
tional run in which the 2a/Rh ratio was increased to 3000:1;
the resulting turnover frequency was 852 hÀ1 and the selec-
tivity for 3a was 72.2%.
The conditions of the extraction step were not optimized
in detail, albeit slightly higher pressures and longer extraction
times were applied for the less volatile products. The products
were collected from the CO2 stream upon depressurization in
two sequential cold traps kept at À608C. Small amounts of the
more volatile products 3a and 4e were lost by evaporation in
this simple setup, but in general the recovery of organic
material was good to excellent. Most importantly, the
products of the individual reactions were obtained without
any significant cross-contamination of material from the
previous runs (< 0.5% according to GC analysis). A total
turnover number of almost 11000 was achieved over the nine
consecutive reactions. Rhodium and phosphorous contami-
nation in all samples was determined by ICP MS (inductively
coupled plasma mass spectrometry), and the total loss of the
original catalyst amounted to 1.2% for rhodium and 2.4% of
phosphorous over the entire sequence. It is important to note
that none of the products 3a–e/4a–e would be amenable to
isolation by distillation without thermal decomposition of the
catalyst. Furthermore, it would be very difficult, if not
Scheme 2. A series of different catalytic reactions carried out sequen-
tially with the cartridge catalyst system 1/[Rh(acac)(CO)2]/scCO2;
pinBH=pinacolborane.
reaction/separation sequence of Scheme 2 in a clockwise
manner
The series of reactions was carried out in a 24-mL stainless
steel reactor equipped with a window. The substrate-to-metal
ratio 2e/Rh was adjusted to approximately 1000:1 in all cases,
and no additional solvent was used. The catalyst was formed
in situ in the first run from 1 and [Rh(acac)(CO)2] (P/Rh =
5:1). As the initial reaction a hydroboration of 2e was carried
out with pinacol borane at room temperature for (3 h)
Angew. Chem. Int. Ed. 2005, 44, 2291 –2295
www.angewandte.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2293

Communications
Although the metal contamination (ca. 5 ppm) achieved
under the nonoptimized conditions reported here would be
too high for larger scale production in many cases, it may be
acceptable in synthetic laboratory-scale applications. Further
reduction of the catalyst leaching seems possible by improved
engineering and optimization of the extraction conditions.
Another limitation of the system at its current stage of
development is imposed by the range of catalytic trans-
formations that can be achieved with the ligand type 1.
However, the synthetically straightforward modification of
achiral or chiral ligands with PEG chains together with the
significant solubility of a large body of organic compounds in
scCO2 suggest that this strategy can be extended also to other
broadly applicable catalytic systems. Side chains other than
PEG may also be envisaged to provide similar solubility
properties, for example, ionic groups such as those reported
recently for continuous-flow hydroformylation of propene.[10e]
Figure 3. Representative example for a series of different catalytic reac-
tions linked sequentially with the cartridge system 1/[Rh(acac)(CO)2]/
scCO2 (see Scheme 2 and text for details).
Experimental Section
Safety warning: Experiments using compressed gases must be carried
out only with appropriate equipment and under rigorous safety
precautions.
followed by extraction of all organic material with scCO2
(458C/60–100 bar). The conversion of 2e was 87%, the two
regioisomeric borane adducts 5 and 6 were the only significant
products (7 was formed in 1% yield), and the selectivity for
the branched product 5 was 84%. A new batch of 2e was
added to the material remaining in the reactor, and a
hydroformylation was carried out and worked up as described
above (see Table 1, entry 8). Complete conversion with a
selectivity of 90% for 4e was achieved. In the third catalytic
process, hydrogenation of 2e (T= 508C, p(H2) = 50 bar, t =
2 h) and subsequent extraction with scCO2 (458C/80 bar)
gave ethylbenzene (7) in practically quantitative yield.
Finally, the sequence was completed with a hydroboration,
which gave results identical to those from the initial run (89%
conversion, 83% selectivity for 5). All samples were analyzed
for their rhodium and phosphorous content; the total loss
over all four reactions was only 0.16 and 1.3%, respectively.
In summary, we have demonstrated for a series of
rhodium-catalyzed reactions that the combination of a
PEG-modified phosphine ligand and scCO2 as phase switch
provides an efficient cartridge system for organometallic
catalysis. As all reactions examined here were carried out neat
and proceeded as 100% atom-efficient addition reactions, no
solvent waste or by-products were generated and the products
were isolated directly in high purity. The equipment required
for handling the supercritical fluid is largely identical to that
established in supercritical extraction or chromatography.
Thus it would be possible to retrofit a commercially available
technology platform to accommodate this reaction–separa-
tion system. The method is readily scalable to small- or
medium-size production, and can be automated and even
parallelized for the generation of libraries of structural
diversity on a laboratory scale. This seems particularly
intriguing for reactions involving pressurized gases, for
which the integration of reaction and separation in one
operation unit is notoriously difficult in the context of
conventional homogeneous catalysis.
Typical procedure for the catalysis/separation sequence, exem-
plified by the hydroformylation of 2a–e. Awindow-equipped stainless
steel reactor (V= 24 mL) was charged under argon atmosphere with
[Rh(acac)(CO)2] (0.01 mmol) and ligand 1[13] (0.05 mmol). Substrate
2a (10 mmol) was added under argon to give a clear orange solution,
and the reactor was heated to 708C. The reaction was started by
introduction of synthesis gas (H2/CO = 1:1, 50 bar) and allowed to
proceed for two hours during which the color of the reaction mixture
changed to dark red. At the end of the reaction, the autoclave was
cooled to 508C, and carbon dioxide was added to reach a density of
approximately 0.5–0.6 gmLÀ1. An orange solid precipitated and most
of the liquid product dissolved. The extraction was performed by
flushing CO2 through the reactor at a flow rate of approximately 0.5–
0.7 mLminÀ1 (STP); the reactor was maintained at a temperature
between 45 and 508C and a pressure between 80 and 100 bar. After
exiting the reactor, the CO2 stream was vented to ambient pressure
through a series of two cold traps (À608C). The extraction process
was monitored visually through the windows of the reactor and
continued for twice the time after which no liquid product was visible
in the reactor. The contents of the cold traps were combined and
collected for off-line analysis, and the reactor was charged again with
a new batch of substrate for the next cycle.
Received: August 23, 2004
Revised: November 16, 2004
Published online: March 10, 2005
Keywords: green chemistry · homogeneous catalysis ·
.
phosphine ligands · rhodium · supercritical fluids
[1] a) Transition Metals for Organic Synthesis (Eds.: M. Beller, C.
Bolm), Wiley-VCH, Weinheim, 1998; b) Applied Homogeneous
Catalysis with Organometallic Compounds (Eds.: B. Cornils,
W. A. Herrmann), 2nd ed., Wiley-VCH, Weinheim, 2000.
[2] a) Green Chemistry: Theory and Practice (Eds.: P. T. Anastas, J.
Warner), Oxford University Press, Oxford, 1998; b) R. A.
Sheldon, Pure Appl. Chem. 2000, 72, 1233 – 1246; c) P. T.
Anastas, M. M. Kirchhoff, Acc. Chem. Res. 2002, 35, 686 – 694.
[3] a) R. T. Baker, W. Tumas, Science 1999, 284, 1477 – 1479; b) D. J.
Cole-Hamilton, Science 2003, 299, 1702 – 1706; c) Aqueous-
2294
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2005, 44, 2291 –2295

Angewandte
Chemie
Phase Organometallic Catalysis (Eds.: B. Cornils, W. A. Herr-
mann), 2nd ed., Wiley-VCH, Weinheim, 2004.
[4] a) D. E. Bergbreiter, Chem. Rev. 2002, 102, 3345 – 3384; b) Z.
Jin, Y. Wang, X. Zheng in Aqueous-Phase Organometallic
Catalysis (Eds.: B. Cornils, W. A. Herrmann), 2nd ed., Wiley-
VCH, Weinheim, 2004, pp. 301 – 311.
[5] a) D. E. Bergbreiter, R. Chandran, J. Am. Chem. Soc. 1987, 109,
174 – 179; b) D. E. Bergbreiter, L. Zhang, V. M. Mariagnanam, J.
Am. Chem. Soc. 1993, 115, 9295 – 9296; c) E. A. Karakhanov,
E. A. Runova, G. V. Berezkin, E. B. Neimerovets, Macromol.
Symp. 1994, 80, 231 – 240.
[6] a) M. Wende, J. A. Gladysz, J. Am. Chem. Soc. 2003, 125, 5861 –
5872; b) V. K. Dioumaev, R. M. Bullock, Nature 2004, 434, 530 –
532.
[7] a) T. Malmstrꢁm, C. Andersson, J. Mol. Catal. A 1997, 116, 237 –
245; b) D. E. Bergbreiter, Y.-S. Liu, Tetrahedron Lett. 1997, 38,
3703 – 3706.
[8] a) Chemical Synthesis Using Supercritical Fluids (Eds.: P. G.
Jessop, W. Leitner), Wiley-VCH, Weinheim, 1999; b) W. Leitner,
Acc. Chem. Res. 2002, 35, 746 – 756.
[9] a) D. Koch, W. Leitner, J. Am. Chem. Soc. 1998, 120, 13398 –
13404; b) S. Kainz, A. Brinkmann, W. Leitner, A. Pfaltz, J. Am.
Chem. Soc. 1999, 121, 6421 – 6429; c) G. Franciꢂ, K. Wittmann,
W. Leitner, J. Organomet. Chem. 2001, 621, 130 – 142; d) W.
Leitner, A. M. Scurto in Aqueous-Phase Organometallic Catal-
ysis (Eds.: B. Cornils, W. A. Herrmann), 2nd ed., Wiley-VCH,
Weinheim, 2004, pp. 665 – 685.
[10] a) M. F. Sellin, D. J. Cole-Hamilton, J. Chem. Soc. Dalton Trans.
2000, 11, 1681 – 1683; b) C. D. Ablan, J. P. Hallett, K. N. West,
R. S. Jones, C. A. Eckert, C. L. Liotta, P. G. Jessop, Chem.
Commun. 2003, 2972 – 2973; c) M. Wei, G. T. Musie, D. H.
Busch, B. Subramaniam, J. Am. Chem. Soc. 2002, 124, 2513 –
2517; d) G. B. Combes, F. Dehghani, F. P. Lucien, A. K. Dillow,
N. R. Foster in Reaction Engineering for Pollution Prevention
(Eds.: M. A. Abraham, R. P. Hesketh), Elsevier Science, Dor-
drecht, 2000, pp. 173 – 181; e) P. B. Webb, D. J. Cole-Hamilton,
Chem. Commun. 2004, 612 – 613.
[11] For examples of sequential catalytic reactions of different olefins
on polymer-bound heterogenized rhodium catalysts, see: a) A. J.
Sandee, J. N. H. Reek, P. C. J. Kamer, P. van Leeuwen, J. Am.
Chem. Soc. 2001, 123, 8468 – 8476; b) F. Shibahara, K. Nozaki, T.
Hiyama, J. Am. Chem. Soc. 2003, 125, 8555 – 8560.
[12] For this reason in his pioneering work on natural product
extraction, Kurt Zosel referred to supercritical-fluid extraction
as “Destraktion”, an artificial word composed from Latin
“destillare” and “extrahere”. K. Zosel, Angew. Chem. 1978, 90,
748; Angew. Chem. Int. Ed. Engl. 1978, 17, 702.
[13] Ligand 1 was obtained as a colorless solid in 81% yield on a 30-g
scale from the reaction of MeOPEG750-OS(O)2Me with LiPPh2
in THF. 1H NMR (CD3OD): d = 2.40 (t, 2H), 3.38 (s, 3H), 3.54–
3.65 (m, 71H), 7.35–7.48 ppm (m, 10H); 13C{1H} NMR d = 30.0
(d, JCP = 13 Hz), 69.7 (d, JCP = 23 Hz), 59.2 (s), 71.3–71.7 (m),
129.7 (s), 129.9 (d, JCP = 12 Hz), 133.9 (d, JCP = 19 Hz), 139.9 ppm
(d, JCP = 13 Hz); 31P{1H} NMR d = À21.1 ppm.
[14] It is important to note that rhodium catalysts based on other very
poorly CO2-soluble ligands tend to give some background
reaction under similar conditions either through formation of
liquid phases[10a] or through small amounts of unmodified soluble
rhodium carbonyl complexes.[9c] These ligands cannot be recy-
cled with the high efficiency required for cartridge catalysts as
experimentally verified for the hydroformylation of 2a with the
system PPh3/[Rh(acac)(CO)2]/scCO2, where the selectivity and
activity were found to decrease significantly upon recycling.
Angew. Chem. Int. Ed. 2005, 44, 2291 –2295
www.angewandte.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2295