Catalytic supercritical fluid extraction: Selective hydroformylation of olefin mixtures using scCO2 solubility for differentiation

Article abstract of DOI:10.1039/c0gc00299b

A new reaction concept that allows one to control the substrate selectivity of a catalytic reaction by supercritical fluid extraction is demonstrated for the hydroformylation of long-chain olefins as a prototypical example.

Full text of DOI:10.1039/c0gc00299b

View Article Online / Journal Homepage / Table of Contents for this issue
COMMUNICATION
www.rsc.org/greenchem | Green Chemistry
Catalytic supercritical fluid extraction: selective hydroformylation of olefin
mixtures using scCO2 solubility for differentiation†
Timo J. Koch,a Simon L. Desseta and Walter Leitner*a,b
Received 7th July 2010, Accepted 30th July 2010
DOI: 10.1039/c0gc00299b
A new reaction concept that allows one to control the
substrate selectivity of a catalytic reaction by supercritical
fluid extraction is demonstrated for the hydroformylation
of long-chain olefins as a prototypical example.
Catalysts can usually discriminate between reactive substrates
based on the functionality of the compound, or the shape
and/or size of the compound. However, for substrates bearing
identical reactive functions and being of similar shapes, the
catalyst alone is sometimes not able to achieve acceptable
levels of differentiation. The integration of mass separation
processes and catalytic reactions provides a possible approach
to address this challenge and examples based on reactive
extraction and reactive distillation are currently being explored.1
In reactive extraction, the discrimination is based on differences
in solvation (typically polarity), whereas reactive distillation
exploits differences in vapour pressures at sufficiently high
temperature. In the present communication, it is demonstrated
that supercritical fluid extraction, SFE, provides an alternative
approach that can be efficiently coupled with transition metal
catalysed transformations, allowing the selective conversion of
substrates with almost identical polarities at temperatures far
below their boiling points.
The intrinsic solubility of a compound in compressed carbon
dioxide, i.e. its “CO2-philicity”, depends largely on a combi-
nation of its volatility and polarity.2 Moreover, the extent of
solubility can be tuned by adjusting the density of the carbon
dioxide.3,4 Therefore, by contacting a mixture of substrates with
compressed CO2 and immobilising a catalyst in a separate
compartment in the CO2 phase, only those constituents of the
mixture displaying significant solubility in the CO2 phase at a
given density would be converted (Fig. 1).
Fig. 1 Schematic representation of the concept for selective reaction
of a substrate mixture in a CO2 biphasic system (S = substrate; P =
product) – S1 has a higher “CO2-philicity” then S2 allowing its selective
transportation to the catalyst, where it is converted to P1.
typically obtained in equimolar proportions indicating that there
is no differentiation from the Rh/PPh3 catalyst between the
substrates under conventional homogeneous conditions.6,7
Rhodium phosphine catalysed hydroformylation of long-
chain olefins in supercritical CO2 (scCO2) has been widely
studied under homogeneous8 and multiphasic conditions.9 In
particular, compressed carbon dioxide has been used as the
mobile phase for continuous hydroformylation, transporting the
dissolved olefin to the reactor containing a catalyst immobilised
on solid supports.10 However, for all these systems, only pure
alkenes, especially 1-octene, have been used as substrates. Given
the high selectivity of SFE we speculated that it should be
possible to identify suitable conditions where CO2 could act as
a selective mass separation vector for olefins of different chain
length.
To identify suitable conditions for the differentiation, the
continuous extraction of an equimolar mixture of 1-octene
The hydroformylation of long-chain olefins was chosen as
a prototypical example to validate this concept. It is well
established that molecular rhodium phosphine catalysts exhibit
high chemo- and regioselectivity, but show no significant
differentiation in rate for different terminal olefins above
C5.5 For example, when a mixture of linear terminal olefins,
namely 1-hexene, 1-octene and 1-decene (1 : 1 : 1 mixture) is
hydroformylated in hexane, the corresponding aldehydes are
◦
and 1-octadecene at 80 C and different pressures of CO2 was
investigated (Table 1). The extractions were carried out using a
standard high pressure reactor connected to a constant pressure
CO2 feed line.‡ The flow was controlled by adjusting the opening
of a thermostatted decompression needle valve at the outlet
of the reactor. The outlet flow passed through two successive
traps, one at ambient temperature and the second at -73 ◦C,
and finally through a rotameter before being vented to the
atmosphere. The extractions were carried out for 8 h and the
weights and compositions (GC analysis) of the extracts were
determined every hour. The material balance of the extractions
was determined by gravimetry and found to be ≥95%.
aInstitut fu¨r Technische und Makromolekulare Chemie, RWTH Aachen
University, Worringerweg 1, 52074, Aachen, Germany.
E-mail: leitner@itmc.rwth-aachen.de; Fax: +49 (0)241 80 22177;
Tel: +49 (0)241 26480
When the extraction was carried out at low pressure, i.e. pCO
2
bMax-Planck-Institut fu¨r Kohlenforschung, Kaiser-Wilhelm-Platz 1,
45470, Mu¨lheim an der Ruhr, Germany
£ 90 bar, very high selectivity toward the shorter olefin was
observed (1-octene selectivity > 97%). By raising the pressure at
constant temperature, i.e. raising the CO2 density, the extraction
† Electronic supplementary information (ESI) available: Material and
methods, catalyst testing. See DOI: 10.1039/c0gc00299b
This journal is
The Royal Society of Chemistry 2010
Green Chem., 2010, 12, 1719–1721 | 1719
©

View Article Online
Table 1 Continuous extraction of a 1-octene/1-octadecene mixture
charge (15 ppm). Moreover, when testing the crude product
phases for blind activity, no further conversion was observed.
Owing to the stability exhibited by the catalysts supported
on ion exchange resins, the use of complex 1 immobilised on
Amberlyst 26 OH for the hydroformylation of a mixture of long-
chain olefins under selective SFE conditions was investigated.
The catalyst was held in a basket placed in the top part of
the reactor ensuring there was no direct contact between the
liquid substrate phase and the catalyst in absence of CO2.
Using this reaction set-up, the hydroformylation of a mixture
of 1-octene and 1-octadecene (1 : 1) under 20 bar of syngas at
different partial pressures of CO2 was conducted (Fig. 2). As
expected, no conversion of any of the substrates was observed
when the reaction was carried out in the absence of CO2. In
the presence of compressed CO2 at low density, only 1-octene
was converted to the corresponding aldehydes (24% conversion,
l/b = 2.7) leaving the 1-octadecene unreacted. Further increase
of the CO2 pressure to 110 bar led to higher conversions (octene
conversion = 32, 48 and 52% at 72, 92 and 107 bar, respectively),
presumably due to more efficient transport of the substrate to
the catalyst, but again almost exclusively 1-octene was reacted
with the same levels of regio-selectivity (2.7 < l/b < 2.9). Only
for CO2 pressures above 110 bar, the conversion of the C18 olefin
(1 : 1) with compressed CO2 at 80 ◦Ca
pCO (bar)
C8 Selectivity (%)b
2
50
70
90
110
130
99.9
99.6
97.4
91.7
87.2
◦
a 80 C, 8 h, 50 cm3 min-1 (STP); b (mole 1-octene)/(mole 1-octene +
mole 1-octadecene), determined by GC, average over 8 h extraction.
selectivity was found to decrease. This was anticipated since the
solvent power of compressed CO2 is directly linked to its density.
Thus at high density, CO2 becomes a “better” solvent hence
decreasing its discrimination ability toward the two different
olefins. The very high selectivities observed for the extraction at
pCO £ 90 bar indicated that under these conditions CO2 could,
2
a priori, be used as a selective transport vector to bring one of
the components of the mixture to an immobilised catalyst that is
located in a separated compartment above the liquid substrates.
An additional prerequisite for the successful combination of
SFE and catalytic reaction was therefore the efficient compart-
mentalisation of the catalyst from the liquid substrate phase. In
other words, the catalyst must be efficiently immobilised onto
a support so that no catalyst is transported to the liquid phase
where it would catalyse the non-selective reaction and override
the selectivity gained trough SFE. Ionic compounds usually ex-
hibit very little solubility in CO2. Therefore the well known ionic
water-soluble hydroformylation complex [HRh(CO)(TPPTS)3],
1, (TPPTS = triphenylsphosphine trisulfonic acid trisodium salt)
was immobilised on various supports and the obtained catalysts
were tested for the hydroformylation of olefins in homogeneous
supercritical phase. Two different types of supports were used:
controlled pore glasses following the concept of supported
aqueous phase catalysts (SAPC),11 and anion exchange resins
where the catalyst is maintained on the support by ionic
interactions between the cationic groups from the surface and
the negatively charged ligand.12 The catalysts were prepared by
wet impregnation of the supports with an aqueous solution of
complex 1 under a syngas atmosphere. Filtration and drying
afforded the supported catalysts as free flowing yellow coloured
solids.
started to become significant (22% conversion at pCO = 135 bar)
2
and at 160 bar the same levels of conversion were observed for
both substrates (38% conversion).
For controlled glass supports, the catalysts were active for
the hydroformylation reaction but deactivated rapidly upon
recycling. Moreover, very low linear-to-branched ratio, l/b, were
observed (0.5 < l/b < 1.5). Deactivation might occur either by
leaching of the catalyst or by change of the hydration level of
the support which was shown to have a dramatic effect on the
catalyst performances.13 In any case, the poor robustness of the
system under supercritical conditions was not suitable for the
present application.
When using the anion exchange resins Dowex Marathon A2
and Amberlyst 26 OH as supports, active catalysts were obtained
and the activity remained stable over 4 consecutive recyclings.
Linear-to-branched ratios of 2.7, typical for rhodium triph-
enylphosphine type catalysts, were observed. The maximum
rhodium leaching, measured by ICP, after reaction over a wide
range of CO2 pressure never exceeded 1.5% of the initial rhodium
Fig. 2 Hydroformylation of a 1-octene/1-octadecene mixture (1 : 1) at
different pressures of CO2.
To ensure that the reaction was taking place in the CO2
phase and not catalysed in the liquid substrate phase by leached
catalyst, all crude reaction mixtures were analysed by ICP-
AES for rhodium and phosphorus. Rhodium concentrations
were below the detection limit (< 1 ppm) and the highest
concentration of phosphorus detected was 10 ppm (1% of the
initial charge) which is fully in line with the batch reactions. Very
interestingly, the octene selectivities observed for the reaction
and for the extraction at different pressures followed very similar
trends (Fig. 3).
This observation, together with the very low levels of rhodium
leaching supports the conclusion that the reaction is genuinely
1720 | Green Chem., 2010, 12, 1719–1721
This journal is
The Royal Society of Chemistry 2010
©

View Article Online
connected to an HPLC pump ensuring there was no contact between
the catalyst and the substrate. The magnetic stirrer was switched on and
the autoclave pressurized with CO2 to the desired pressure. After 20 h,
the autoclave was cooled down and the bottom valve connected to a
cold trap (acetone/dry ice). The content of the reactor was emptied by
gently opening the valve to the cold trap. The content of the cold trap
was analysed by GC for organics and by ICP-AES for rhodium and
phosphorus.
1 (a) Reactive Separation Processes, S. Kulprathipanja, ed., 2002,
Taylor & Francis, New York; (b) for recent examples see: S. H. Shuit,
K. T. Lee, A. H. Kamaruddin and S. Yusup, Fuel, 2010, 89, 527;
(c) G. J. Harmsen, Chem. Eng. Process., 2007, 46, 774.
2 (a) W. Leitner, Nature, 2000, 405, 129–130; (b) K. Zosel, Angew.
Chem., Int. Ed. Engl., 1978, 17, 702–709.
3 N. R. Foster, F. P. Lucien and R. Mammucari, Basic physical
properties, phase behavior and solubility, in Handbook of Green
Chemistry, Volume 4: Supercritical Solvents, W. Leitner and P. Jessop,
ed., 2010, Wiley-VCH, Weinheim, p 77.
Fig. 3 1-Octene (triangle) and 1-octadecene (circle) selectivity during
extraction (black) and hydroformylation (grey).
4 For recent application of this principle to separate catalyst and
product in continuous flow mode see: T. Harwardt, G. Francio` and
W. leitner, Chem. Commun, 2010,, 46, 6669.
5 (a) Rhodium Catalyzed Hydroformylation, P. W. N. M. van Leeuwen
and C. Claver, ed., 2000, Kluwer, Dordrecht; (b) C. D. Frohning,
C. W. Kohlpaintner and H. W. Bohnen, Hydroformylation (Oxo
Synthesis, Roelen reaction), in Applied Homogeneous Catalysis with
Organometallic Compounds 2nd Ed, B. Cornils and W. A. Herrmann,
ed., 2002, Wiley-VCH, Weinheim, p 31.
taking place in the CO2 phase, in which density acts as the
differentiating agent between the reactive substrates.
In summary, we have demonstrated that the selective catalytic
transformation of one compound from a mixture of substrates
could be achieved using the solubility in CO2 as a discrimination
tool. This constitutes a new approach in combining catalysed
transformations with mass separation processes in reactive
multicomponent systems. Extension of this concept to other
relevant mixtures and transformations appears promising for
integrated catalytic processes.
We would like to thank the Agence Nationale de la Recherche
and the Deutsche Forschungsgemeinschaft (ANR-DFG) for
their joint financial support. Additional financial support by
The Cluster of Excellence “Tailor-Made Fuels from Biomass”
(TMFB) is gratefully acknowledged.
6 I. T. Horva´th, Catal. Lett., 1990, 6, 43.
7 In aqueous biphasic systems, olefins with chain lengths > C5 are
typically hydroformylated at lower rates owing to the lower water
solubility as compared to shorter-chain substrate: (a) E. Wiebus and
B. Cornils, Biphasic systems: water-organic, in Catalyst Separation,
Recovery and Recycling, D. J. Cole-Hamilton, and R. P. Tooze, ed.,
2006, Springer, Dordrecht, p 105; (b) H. Bahrmann, S. Bogdanovic
and P. W. N. M. van Leeuwen, Higher alkenes in Aqueous-Phase
Organometallic Catalysis 2nd Ed., B. Crornils, and W. A. Herrmann,
ed., 2004, Wiley-VCH, Weinheim, p 391.
8 (a) D. Koch and W. Leitner, J. Am. Chem. Soc., 1998, 120, 13398–
13404; (b) W. Leitner, Acc. Chem. Res., 2002, 35, 746; (c) C. M.
Rayner, P. M. Rose and D. C. Barnes, Synthetic Organic Chemistry
in Supercritical Fluids, in Handbook of Green Chemistry, Volume 4:
Supercritical Solvents, W. Leitner and P. Jessop Eds, 2010, Wiley-
VCH, Weinheim, p 189 and references therein.
9 (a) P. B. Webb, M. F. Sellin, T. E. Kunene, S. Williamson, A. M. Z.
Slawin and D. J. Cole-Hamilton, J. Am. Chem. Soc., 2003, 125, 15577;
(b) D. J. Cole-Hamilton, Science, 2003, 299, 1702 and references
therein.
10 (a) U. Hintermair, Z. Gong, A. Serbanovic, M. J. Muldoon, C. C.
Santini and D. J. Cole-Hamilton, Dalton Trans., 2010, 38, 8501; (b) U.
Hintermair, G. Zhao, C. C. Santini, M. J. Muldoon and D. J. Cole-
Hamilton, Chem. Commun., 2007, 1462; (c) R. P. J. Bronger, J. P.
Bermon, J. N. H. Reek, P. C. J. Kamer, P. W. N. M. van Leeuwen, P.
Licence and M. Poliakoff, J. Mol. Catal. A: Chem., 2004, 224, 145;
(d) F. Shibahara, K. Nozaki and T. Hiyama, J. Am. Chem. Soc., 2003,
125, 8555.
Notes and references
‡ Extraction: in an autoclave (100 cm3) equipped with a CO2 inlet, a CO2
outlet connected to two successive traps and a flowmeter, a mechanical
stirrer, a heating jacket, a pressure gauge and a thermocouple, the
olefin mixture (1-octene/1-octadecene (1 : 1), 7.7 g) was introduced.
The autoclave was heated to 80 ◦C, stirred (500 rpm) and the CO2
was introduced at the desired pressure. The extraction was started by
opening the needle valve to the CO2 outlet and adjusting the flow to
50 cm3 min-1. The traps were changed every hour, the content were
weighted and analyzed by GC. The extraction was stopped after 8 h,
the reactor was cooled and vented. The liquid phase remaining in the
autoclave was weighed and analysed by GC. Catalyst preparation: the
dried support was impregnated by suspension in an aqueous solution
of [HRh(CO)(TPPTS)3], 1, (prepared according to ref. 14) for 12
h with gentle bubbling of syngas. After filtration and drying under
vacuum, the supported catalyst was obtained as a slightly yellow, free
flowing solid. Hydroformylation: under inert atmosphere, the basket,
loaded with catalyst (300 mg, 7.34 wt% of [HRh(CO)(TPPTS)3]), 1,
was introduced into the autoclave (20 cm3) and mounted in the upper
part. The autoclave was sealed, heated to 80 ◦C and pressurized with
syngas (20 bar). The olefin mixture (1-octene/1-octadecene (1 : 1), 1.5
cm3), was introduced at the bottom of the autoclave through a valve
11 J. P. Arhancet, M. E. Davis, J. S. Merola and B. E. Hanson, Nature,
1989, 339, 454.
12 P. Barbaro and F. Liguori, Chem. Rev., 2009, 109, 515.
13 (a) J. P. Arhancet, M. E. Davis, J. S. Merola and B. E. Hanson,
J. Catal., 1990, 121, 327; (b) T. I. Horvath, Cat. Lett., 1990, 6,
43.
14 I. Toth, B. E. Hanson, I. Guo and M. E. Davis, Catal. Lett., 1991, 8,
209.
This journal is
The Royal Society of Chemistry 2010
Green Chem., 2010, 12, 1719–1721 | 1721
©