Homogeneous reactions in supercritical carbon dioxide using a catalyst immobilized by a microporous silica membrane

Article abstract of DOI:10.1002/1521-3773(20011203)40:23<4473::AID-ANIE4473>3.0.CO;2-0

Membrane separation technology is successfully applied for the immobilization of a homogeneous catalyst (a (1H, 1H,2H,2H-perfluoroalkyl)-dimethylsilyl-substituted derivative of Wilkinson's catalyst) in a continuous process that uses supercritical carbon dioxide as solvent. The catalyst is separated from the products by a microporous silica membrane (see scheme).

Full text of DOI:10.1002/1521-3773(20011203)40:23<4473::AID-ANIE4473>3.0.CO;2-0

COMMUNICATIONS
applicationto CCDC, 12 UnionRoad, Cambridge CB21EZ, UK (fax:
(‡44)1223-336-033; e-mail: deposit@ccdc.cam.ac.uk).
[13] R. A. Leising, J. S. Ohman, K. J. Takeuchi, Inorg. Chem. 1988, 27,
3804 3809.
[14] a) R. E. Shepherd, H. Taube, Inorg. Chem. 1973, 12, 1392 1401;
b) S. S. Isied, H. Taube, Inorg. Chem. 1976, 15, 3070 3075.
[15] T. Matsubara, C. Creutz, Inorg. Chem. 1979, 18, 1956 1966.
[16] L. R. Allen, P. R. Craft, B. Durham, J. Walsh, Inorg. Chem. 1987, 26,
53 56.
[17] M. H. V. Huynh, J. M. Lasker, M. Wetzler, B. Mort, L. F. Szczepura,
L. M. Witham, J. M. Cintron, A. C. Marschilok, L. J. Ackerman, R. K.
Castellano, D. L. Jameson, M. R. Churchill, A. J. Jircitano, K. J.
Takeuchi, J. Am. Chem. Soc. 2001, 123, 8780 8784.
[18] C. A. Bessel, J. A. Margarucci, J. H. Acquaye, R. S. Rubino, J.
Crandall, A. J. Jircitano, K. J. Takeuchi, Inorg. Chem. 1993, 32,
5779 5784.
successful operationof homogeneous catalysis inscCO 2 is the
solubility of the catalyst, which canbe achieved by attaching
perfluoroalkyl groups to the ligands.[5]
Herein, we present a solution that overcomes these draw-
backs. We employ a continuous reactor for homogeneous
catalysis inscCO 2 that allows anintegrated catalyst separation
by immobilizationof a perfluoroalkylated catalyst onone side
of a microporous silica membrane. The low cohesive energy
density of perfluoroalkyl groups is used to realize the required
solubility of the catalyst inscCO 2, and simultaneously the
increased size ensures effective retention of the catalyst in the
membrane reactor. The reactants and products, which are also
dissolved inthe scCO 2, candiffuse through the membrane.
The principle is demonstrated by using a (1H,1H,2H,2H-
perfluoroalkyl)dimethylsilyl-substituted derivative of Wilkin-
son×s catalyst, 1[6] and a supported microporous silica
[19] a) Typically, rates of aqua-ligand exchange in ruthenium(ii) an d
ruthenium(iii) complexes range from 1 Â 10À6 mÀ1 sÀ1 to 1 Â
10À2 mÀ1 sÀ1; b) L. M˘nsted, O. M˘nsted, Coord. Chem. Rev. 1989,
94, 109 150.
Homogeneous Reactions in Supercritical
Carbon Dioxide Using a Catalyst Immobilized
by a Microporous Silica Membrane
membrane with an average pore size of 0.6 nm.[7] The
mechanical strength of ceramic silica membranes allows the
use of relatively harsh reaction conditions. As a model
Leo J. P. vandenBroeke,* Earl L. V. Goetheer,
ArjanW. Verkerk, Elwinde Wolf, Berth-JanDeelman,
Gerard van Koten, and Jos T. F. Keurentjes
reaction the hydrogenation of 1-butene was first carried out
[8]
inscCO ina batch reactor
at 200 bar and 353 K. The
2
catalyst 1 was prepared insitu from [RhCl(cod)] (cod ˆ
2
Although homogeneous catalysis offers many advantages
over heterogeneous catalysis in terms of catalytic selectivity
and activity, the difficult separation of the catalyst from the
products and the use of environmentally harmful organic
solvents are two major drawbacks.[1, 2] The problem of catalyst
separationhas beenaddressed by several groups, adn a
number of new concepts for immobilization of homogeneous
cis,cis-1,5-cyclooctadiene) and six equivalents of P(p-(Si-
Me2CH2CH2C8F17)C6H4)3. Inthe batch reactor the turnover
frequency (TOF) at 25% conversion was found to be 9400 hÀ1.
The high solubility of hydrogeninscCO 2, as compared to the
solubility in conventional solvents,[6] is responsible for the 10-
fold higher activity. For the sake of comparison, the solublity
of hydrogenintoluene is 2.7 m m (at 1 bar and 298 K) and in
fluorous solvents, like c-C6F11CF3, it is 6.1 mm (at 1 bar and
298 K).
[3]
catalysts have beendeveloped. With respect to the use of
more environmentally friendly solvents, the application of
high-density gases such as supercritical carbon dioxide
(scCO2; Tc ˆ 304.2 K and Pc ˆ 73.8 bar) has some clear
advantages.[2, 4] Supercritical solvents are completely miscible
with gaseous reagents, and, therefore, avoid possible diffusion
limitationingas liquid reactions. Animportant issue for
In the continuous membrane set-up (Figure 1), permeation
and reaction experiments were typically conducted with a
feed pressure of 200 bar and a transmembrane pressure
[9]
between0.5 and 10 bar.
Experiments have been performed
with different pressure differences across the membrane to
study the effect of the residence time on the conversion.
[‡]
[*] Dr. L. J. P. vandenBroeke, E. L. V. Goetheer, A. W. Verkerk,
Prof. Dr. J. T. F. Keurentjes
Process Development Group
Department of Chemical Engineering and Chemistry
Eindhoven University of Technology
PO Box 513, 5600 MB Eindhoven (The Netherlands)
Fax : (‡31)40-244-6104
E-mail: l.j.p.van.den.broeke@tue.nl
Figure 1. Continuous reaction and separation concept. The membrane
reactor is operated in a dead-end configuration.
‡
[ ] Dutch Polymer Institute
PO Box 513, 5600 MB Eindhoven (The Netherlands)
E. de Wolf, Dr. B.-J. Deelman,[‡‡] Prof. Dr. G. vanKoten
Debye Institute, Department of Metal-Mediated Synthesis
Utrecht University
Up to a feed side pressure of 200 bar high fluxes of carbon
dioxide were obtained through the microporous membrane.
Anoverview of the characteristics of the membrane reactor is
giveninTable 1.
Padualaan8, 3584 CH Utrecht (The Netherlands)
‡‡
[
] ATOFINA Vlissingen BV
PO Box 70, 4380 AB Vlissingen (The Netherlands)
Angew. Chem. Int. Ed. 2001, 40, No. 23
¹ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2001
1433-7851/01/4023-4473 $ 17.50+.50/0
4473

COMMUNICATIONS
Table 1. Characteristics of the membrane reactor.
Property
latter appears highly unlikely since the size of 1, as obtained
from a MMFF94 molecular mechanics structural optimization
(4 nm) and dynamic light scattering (2 nm in c-C6F11CF3), was
found to be much larger than the average pore size of the
membrane (0.6 nm). Oxidation of 1 due to residual traces of
oxygeninthe carbondioxide or the feed is a more likely
explanation for the observed decrease in activity.
thickness selective silica layer
length membrane
200 nm
0.30 m
outer diameter ceramic membrane
membrane reactor volume
CO2 permeance[a] (200 bar, 353 K)
0.014 m
35.0 mL
3.0 Â 10À3 molmÀ2 sÀ1 barÀ1
[a] Flux divided by the pressure difference across the membrane.
In conclusion, we have demonstrated the successful appli-
cationof membrane separationtechnology for the immobi-
lization of homogeneous catalysts in a continuous process
using a supercritical solvent. It is clear that the methodology
presented here could be highly relevant for the further
development of clean chemical processes based on homoge-
neous catalysis carried out in environmentally friendly high-
density gases.
In the continuous process[10] about two permeated reactor
volumes were needed to reach a constant conversion of about
40% (Figure 2). In the experiment a pressure difference of
Received: July 5, 2001 [Z17434]
[1] P. G. Jessop, T. Ikariya, R. Noyori, Chem. Rev. 1999, 99, 475.
[2] D. Adams, Nature 2000, 407, 938.
[3] M. G. L. Petrucci, A. K. Kakkar, Adv. Mater. 1996, 8, 251; J. W.
Knapen, A. W. van der Made, J. C. Wilde, P. W. N. M. van Leeuwen, P.
Wijkens, D. M. Grove, G. van Koten, Nature 1994, 372, 659; W. A.
Herrmann, C. W. Kohlpainter, Angew. Chem. 1993, 105, 1588; Angew.
Chem. Int. Ed. Engl. 1993, 32, 1524; E. de Wolf, G. vanKoten, B.-J.
Deelman, Chem. Soc. Rev. 1999, 28, 37.
[4] Chemical synthesis using supercritical fluids (Eds.: P. G. Jessop, W.
Leitner), Wiley-VCH, Weinheim, 1999; S. L. Wells, J. DeSimone,
Angew. Chem. 2001, 113, 534; Angew. Chem. Int. Ed. 2001, 40, 518; B.
Lin, A. Akgerman, Ind. Eng. Chem. Res. 2001, 40, 1113.
Figure 2. Conversion, y [%], of 1-butene to butane versus permeated
volume per reactor volume, for the continuous reaction/separation process.
[5] S. Kainz, D. Koch, W. Baumann, W. Leitner, Angew. Chem. 1997, 109,
1699; Angew. Chem. Int. Ed. Engl. 1997, 36, 1628.
[6] B. Richter, A. L. Spek, G. van Koten, B.-J. Deelman, J. Am. Chem.
Soc. 2000, 122, 3945; B. Richter, E. de Wolf, G. vanKoten, B.-J.
Deelman, J. Org. Chem. 2000, 65, 3885.
3 bar is applied, and this corresponds to a residence time of
62 min. The TOF, based on the initial amount of catalyst
added, is 4000 hÀ1, which is about 60% lower thaninthe batch
experiment (TOF ˆ 9400 hÀ1). Overnight the pressure of the
system was reduced to 60 bar, leaving no pressure difference
between the feed and the permeate side of the membrane.
During the depressurization of the membrane reactor the
catalyst precipitated and no further reaction occurred. The
precipitated catalyst was used for a new cycle by pressuriza-
tionof the membrane reactor. The shutdownprocedure was
carried out twice to check the stability of the catalyst under
those conditions. At the start of a new cycle, the system was
first pressurized to 200 bar for 2 h to dissolve the catalyst
before creating a transmembrane pressure difference. At the
end of the third run a conversion of about 33% was obtained
(TOF ˆ 3000 hÀ1). An overall turnover number of about 1.2 Â
105 was obtained (in 32 h), and a total of 32 reactor volumes
had permeated through the membrane. The overall operating
time, including start-ups, reaction cycles, and shutdowns, was
approximately 75 h.
[7] M. K. Koukou, N. Papayannakos, N. C. Markatos, M. Bracht, H. M.
Veen, A. Roskam, J. Membr. Sci. 1999, 155, 241.
[8] The batch reactor, with a volume of 26 mL, consists of two saffire
windows for UV/Vis monitoring. The volume of the reactor can be
controlled with a piston, which assures sampling without disturbing
the reaction conditions. The concentration of 1 was 1.0 Â 10À6 molLÀ1
,
and 1-butene and hydrogen were present in a concentration of 0.02
and 0.08 molLÀ1, respectively, at a total pressure of 200 bar.
[9] For the permeation experiments and the hydrogenation reaction in
the membrane reactor the same high-pressure set-up was used. The
catalyst was first synthesized insitu ina high-pressure reactor. The
catalyst was introduced into the system by gently flushing CO2
through the high-pressure reactor. The reactionwas started by the
addition of the 1-butene and hydrogen to the membrane module. The
membrane was pressurized at the feed and permeates side up to
200 bar with the reaction mixture, by using a LKB HPLC pump and
was monitored by a Meyvis 802-C pressure module. The permeate
needle valve was opened to create a transmembrane pressure varying
between0.5 to 10 bar. The compositionof the permeate was
monitored by GC and by GC-MS.
[10] During the continuous reaction experiments the concentration of 1
was 1.0 Â 10À6 molLÀ1, and 1-butene and hydrogen were present in a
concentration of 0.02 and 0.08 molLÀ1, respectively.
[11] A 2.2 mL high-pressure viewing cell was used in combination with a
spectrophotometer; measurements recorded at 280 and 410 nm.
InFigure 2 an18% decrease inconversionis observed
(from 40% to 33%). Analysis of the permeate stream by UV/
Vis spectroscopy[11] and by inductively coupled plasma atomic
absorptionspectrometry (ICP-AAS) showed that there was
no transport of the catalyst or of the free ligand through the
membrane; thus, it can be concluded that complete retention
of 1 occurred. Evenif the decrease inconversionwould have
been caused by leaching of the catalyst, the retention of the
catalyst per permeated reactor volume is still over 99%. The
4474
¹ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2001
1433-7851/01/4023-4474 $ 17.50+.50/0
Angew. Chem. Int. Ed. 2001, 40, No. 23