Nanofiltration-coupled catalysis to combine the advantages of homogeneous and heterogeneous catalysis

Article abstract of DOI:10.1039/b009898l

In a hybrid process that combines nanofiltration with homogeneous catalysis, the best possible reaction rates, chemoselectivities and enantioselectivities are obtained in a continuous operation mode while recycling the catalyst.

Full text of DOI:10.1039/b009898l

View Article Online / Journal Homepage / Table of Contents for this issue
Nanofiltration-coupled catalysis to combine the advantages of homogeneous
and heterogeneous catalysis
Koen De Smet, Sven Aerts, Erik Ceulemans, Ivo F. J. Vankelecom* and Pierre A. Jacobs
Centre for Surface Chemistry and Catalysis, Faculty of Agricultural and Applied Biological Sciences, Katholieke
Universiteit Leuven, Kardinaal Mercierlaan 92, 3001 Leuven, Belgium. E-mail: ivo.vankelecom@agr.kuleuven.ac.be
Received (in Cambridge, UK) 12th December 2000, Accepted 21st February 2001
First published as an Advance Article on the web 8th March 2001
In a hybrid process that combines nanofiltration with
homogeneous catalysis, the best possible reaction rates,
chemoselectivities and enantioselectivities are obtained in a
continuous operation mode while recycling the catalyst.
while a good rejection of the the catalyst and reasonable fluxes
are preserved in the NF.
The continuous enantioselective hydrogenation of dimethyl
itaconate (DMI) with Ru–BINAP (MW 929 Da) and of methyl
2-acetamidoacrylate (MAA) with Rh–EtDUPHOS (MW
723 Da) (Fig. 2) were selected because of the excellent
performance5 of these catalysts and their industrial relevance.6
Several ways to heterogenise these complexes have been
reported already7–9 but most did not equal the performance of
the homogeneous catalyst, either in ee, activity or in the range
of possible substrates.
In industrial processes, heterogeneous catalysts are generally
preferred as they facilitate removal of the catalyst after reaction1
and allow a continuous operation mode. On the other hand,
preparing a heterogeneous catalyst can be tedious and might
demand a high preparative effort. In certain cases, mass or heat
transfer limitations in the solid state catalyst may lead to
decreased activities, and homogeneous reactions generally
show higher chemo- and enantio-selectivities.2
In the reported hybrid process (Fig. 1), a reaction takes place
in a continuously stirred tank reactor, thus reaching activities
and selectivities as in homogeneous reactions. The liquid is
contacted with a nanofiltration (NF) membrane that allows
products to permeate but rejects the dissolved catalyst. This set-
up is made possible by the recent development of solvent
resistant NF membranes.3 They have a molecular weight cut-off
(MWCO) in the range of 200–700 Da and working conditions
below 40 °C and 35 bar.
In related work by Giffels et al.,4 the same membranes are
used but they behave only like ultrafiltration membranes under
the reaction conditions applied. This implies that an enlarge-
ment of their oxazaborolidines is still necessary to have them
rejected by the membrane. Derivatisation of the catalyst in order
to enlarge it—e.g. by linking it to polymers or by forming
dendrimers—is avoided in our experiments by operating the
membrane filtration under true nanofiltration conditions. The
catalysts are thus retained by the membrane without the need to
derivatise them first, and they can be used off the shelf in the
form in which they are readily available.
In particular, chiral catalysts are among the preferred systems
for this hybrid membrane/catalysis process due to their
extremely high cost and their sensitivity towards traditional
heterogenisation methods. Furthermore, most of these catalysts
contain transition metal complexes with a molecular weight
above 500 Da and high activities and selectivities under
moderate reaction conditions. In the reported set-up, the
hydrogen pressure needed for the hydrogenation of the
substrates, forms—without any additional cost or equipment—
the driving force for the membrane permeation. The whole
hybrid process is operated in such a way that a sufficient amount
of product with high purity is yielded in the catalytic process,
Fig. 2 Reactions with (a) Rh–EtDUPHOS and (b) Ru–BINAP.
Because the membrane fluxes in the NF part of the process
are coupled via the hydrogen pressure to the reaction rates
during catalysis, the reaction rate and enantiomer excess were
determined first in a batch-wise reference reaction at pressures
that fall within the range for nanofiltration (Table 1).
The homogeneous reactions were carried out in magnetically
stirred 10 ml autoclaves. 0.35 g (2.5 mM) MAA was dissolved
in 9 ml MeOH and flushed with N2 before adding 1.7 mg Rh–
EtDUPHOS (2.35 mM). For the homogeneous reaction with
Ru–BINAP, 0.6 g (4 mM) DMI was dissolved in 9 ml MeOH,
flushed with N2 and subsequently mixed with 4.3 mg (4.6 mM)
catalyst. The results10 are shown in Fig. 1. For the hydro-
genation of DMI with Ru–BINAP, no literature data could be
found. The enantiomer excess and the activity for reactions with
Rh–EtDUPHOS were lower than reported in the literature.11
Table 1 Hydrogenations of MAA and DMI with, respectively, Rh–
EtDUPHOS and Ru–BINAP
Substrate/ Pressure/
catalyst bar
TOF/
h21
Ee (%)
T/°C
Rh–EtDUPHOSa
Rh–EtDUPHOS
Ru–BINAP
2000
1050
850
2
4
10
22
30
37
> 2000
703
330
99.4
95
93
a Burk et al.11
Fig. 1 Reactor set-up for NF-coupled catalysis.
DOI: 10.1039/b009898l
Chem. Commun., 2001, 597–598
This journal is © The Royal Society of Chemistry 2001
597

View Article Online
Table 2 Methanol fluxes at different temperatures and pressures for MPF-
60 membranes (KOCH)
Pressure/bar
10
T/°C
Flux/kg m22 h21
30
40
50
60
30
1.2
1.6
2.6
3.2
1.7
15
This is believed to be due to a less thorough pre-treatment of
solvents and reagents as compared with literature.
A NF membrane generally does not discriminate between
reactants and hydrogenated products, given their negligible
difference in MW, shape or polarity. This means that the
conditions that determine the membrane flux—such as mem-
brane type, membrane area, applied pressure gradient, tem-
perature and type of solvent12—can be adjusted to the catalytic
conditions that determine the conversion of the reactant.
Methanol fluxes through the NF membrane (KOCH, MPF-60),
are given in Table 2 for different temperatures and pressures: as
expected, higher temperature and pressure are tools used to
increase the flux through the membrane and thus realise a
shorter residence time (t) for a given reactor volume.
The continuous reactions were carried out in a stirred 100 ml
autoclave containing an MPF-60 membrane at the bottom. The
permeate was collected in a cooled flask (278 °C). Both feed
and permeate were analysed by GC and AAS to determine
retention (retained concentration/feed concentration) of re-
actants, products and catalyst. For the hydrogenation of DMI
Fig. 4 Conversion and enantiomer excess as a function of time for the
continuous NF-coupled hydrogenation with Rh–EtDUPHOS.
tion of the catalyst—possibly due to oxidation of the phosphine
ligand—is assumed but still needs further investigation. The
total TONs for the hydrogenation with Ru–BINAP and Rh–
EtDUPHOS are, respectively, 1950 and 930.
These two continuous reactions demonstrate the general
concept of this hybrid process to perform homogeneous
reactions in a continuous mode whenever the membrane is able
to retain the catalyst and does not retain the products. Even
though the system is limited by working conditions—like
solvent, temperature and pressure—it is believed that the
concept can be applied in many different types of reaction and
for a wide range of catalysts and substrates, especially in the
field of fine chemical synthesis.
This work was supported by the Belgian Federal Government
in the frame of an IAP-PAI grant on Supramolecular Catalysis.
K. D. S. acknowledges ‘het Vlaams Instituut voor de bevorder-
ing van het wetenschappelijk-technologisch onderzoek in de
industrie’ (IWT) for a grant as doctoral research fellow.
I. F. J. V. acknowledges a fellowship as Post-doctoral
Researcher from the Fund for Scientific Research (FWO).
with Ru–BINAP at 37 °C and 10 bar, the feed solution (C0
=
0.4 mM) was pumped at a rate of 3.6 ml h21 to the reaction
mixture (V = 14 ml, C0 = 0.4 mM and 33.7 mg Ru–BINAP).
The hydrogenation of MAA was performed at 35 °C and 10 bar
(V = 16 ml, C0 = 0.13 mM and 8.5 mg Rh–EtDUPHOS) with
the feed solution (C0 = 0.13 mM) added at a rate of 3.5
ml h21
.
To fully prove the concept, activities should remain un-
changed after several refreshments of the reactor volume and
the complex should be retained sufficiently. The hydrogenation
of DMI shows a constant enantiomer excess as a function of
time (Fig. 3). The very small decrease in conversion, becoming
apparent after several hours, can be ascribed to the incomplete
rejection of Ru–BINAP ( > 98%). Nevertheless, this nano-
filtration-coupled catalysis allowed the continuous hydro-
genation of ten reactor volumes—as indicated by the vertical
lines on the graph—with an enantiomer excess of 93%, which
equals those reached under homogeneous conditions. For the
hydrogenation of MAA with Rh–EtDUPHOS (Fig. 4), the
decrease in enantiomer excess and conversion in the long term
is slightly more significant. Since the 97% retention of this
complex cannot alone account for this effect, a slow deactiva-
Notes and references
1 E. Lindner, T. Schneller, F. Auer and H. A. Mayer, Angew. Chem., Int.
Ed., 1999, 38, 2154.
2 I. F. J. Vankelecom and P. A. Jacobs, in Immobilisation of Chiral
Catalysts, ed. D. De Vos, I. F. J. Vankelecom and P. A. Jacobs, VCH
Weinhein, 2000, ch. 2, pp. 19–42.
3 KOCH International B.V., Membrane Systems Division.
4 G. Giffels, J. Beliczey, M. Felder and U. Kragl, Tetahedron:
Asymmetry, 1998, 9, 691: S. Rissom, J. Beliczey, G. Giffels, U. Kragl
and C. Wandrey, Tetrahedron: Asymmetry, 1999, 10, 923.
5 S. Akutagawa, Appl. Catal., 1995, 128, 171
6 S. C. Stinson, C&EN, September 1998, pp. 83–104.
7 D. J. Bayston, J. L. Fraser, M. R. Ashton, A. D. Baxter, M. E. C.
Polywka and E. M. Moses, J. Org. Chem., 1998, 63, 3137.
8 R. ter Halle, B. Colasson, E. Schulz, M. Spagno and M. Lemaire,
Tetrahedron Lett., 2000, 41, 643.
9 K. T. Wan and M. E. Davis, Nature, 1994, 370, 449; I. F. J. Vankelecom,
D. Tas, R. F. Parton, V. Van de Vyver and P. A. Jacobs, Angew. Chem.,
1996, 108, 1445; I. F. J. Vankelecom, A. Wolfson, S. Geresh, M.
Landau, M. Gottlieb and M. Hershkovitz, Chem. Commun., 1999, 23,
2407.
10 The GC analysis was carried out on a Chiraldex G-TA (Chrompack)
with N2 as carrier, whereas MAA reactions were analysed on a Chirasil-
DEX CD (Chrompack) column with H2 as carrier. The amount of
catalyst in the permeate was determined by measuring the Ru
concentration by atomic absorption spectroscopy (Varion Techtron
AA6) at 349.9 nm and the Rh concentration at 343.5 nm.
11 M. J. Burk, J. E. Feaster, W. A. Nugent and R. L. Harlow, J. Am. Chem.
Soc., 1993, 115, 10 125.
12 D. R. Machado, D. Hasson and R. Semiat, J. Membr. Sci., 1999, 163, 93;
D. R. Machado, D. Hasson and R. Semiat, J. Membr. Sc., 1999, 166, 63;
J. A. Whu, B. C. Baltzis and K. K. Sirkar, J. Membr. Sci., 2000, 170,
159.
Fig. 3 Conversion and enantiomer excess as a function of time for the
continuous NF-coupled hydrogenation with Ru–BINAP.
598
Chem. Commun., 2001, 597–598