Continuous flow homogeneous catalysis using supercritical fluids

Article abstract of DOI:10.1039/b316311c

The continuous flow hydroformylation of 1-octene catalysed by Rh/[RMIM][Ph₂PC₆H₄SO₃] (R = 1-propyl, 1-pentyl or 1-octyl) dissolved only in the steady state reaction mixture and using scCO₂ as a transport vector for both substrates and products gives rates up to 160-240 catalyst turnovers h^-1 with low rhodium leaching over a 12 h period at a total pressure of 125-140 bar.

Full text of DOI:10.1039/b316311c

Continuous flow homogeneous catalysis using supercritical fluids
Paul B. Webb and David J. Cole-Hamilton*
School of Chemistry, University of St. Andrews, St. Andrews, Fife, Scotland, UK KY16 9ST.
E-mail: djc@st-and.ac.uk; Fax: +44-1334-463808; Tel: +44-1334-463805
Received (in Cambridge, UK) 16th December 2003, Accepted 9th January 2004
First published as an Advance Article on the web 9th February 2004
The continuous flow hydroformylation of 1-octene catalysed by
Rh/[RMIM][Ph PC SO ] (R = 1-propyl, 1-pentyl or 1-octyl)
dissolved only in the steady state reaction mixture and using
scCO as a transport vector for both substrates and products
gives rates up to 160–240 catalyst turnovers h
rhodium leaching over a 12 h period at a total pressure of
25–140 bar.
irritation.10 These solvents may also prove to be prohibitively
expensive for use in the manufacture of bulk or commodity
chemicals. In this paper, we report a method for overcoming both
of these problems by removing the ionic liquid altogether.
Imidazolium based ionic liquids with a long alkyl chain are
capable of dissolving both the alkene substrate and the aldehyde
product. This suggests that they are not very polar and we reasoned
that it might be possible to dissolve the ligand and rhodium
precursor in the steady state mixture of substrate and products that
would be generated if no ionic liquid were present. If this were the
case, it should be possible to use the steady state reaction mixture
as the solvent and to transport substrates into the reactor and
2
6
H
4
3
2
2
1
with low
1
The separation of the products of a reaction from the catalyst and
any solvent can be such a serious problem for reactions involving
homogeneous catalysts that their commercialisation is sometimes
inhibited. This is especially the case if the products are of low
1
volatility and/or the catalyst is of low thermal stability. A range of
products from the reactor dissolved in scCO
using the reactor described previously4 showed that
[PrMIM][Ph PC SO ] exhibits some solubility in nonanal at the
2
. Our initial studies
new approaches to this separation problem is being explored,2 but
amongst them those which involve continuous flow processing are
especially attractive, not only because the catalyst/product separa-
tion is built in as an intrinsic part of the process, but also because all
of the catalyst always remains in the reactor in its active state and
high throughputs can be achieved from relatively small reactors.
We recently reported that the rhodium catalysed hydroformylaton
of long chain alkenes, a reaction of considerable potential
significance for the production of plasticiser, soap and detergent
range alcohols, can be carried out as a continuous process using a
,3
2
6
H
4
3
reaction temperature so that the reaction can be carried out using
nonanal as the initial solvent. GC analysis of liquid fractions
collected downstream over a 12 h period showed that the reaction
2
proceeded. Initially nonanal was recovered from the scCO in the
decompression system, but the composition of the recovered
product altered with time to reach a steady state, as indicated by the
linear : branched (l : b) ratio of the aldehydes. This initial reaction
was carried out at 200 bar total pressure and it proved very difficult
to adjust the gas and liquid flow rates so as to maintain a constant
liquid volume in the reactor, such that at the end of the reaction, the
reactor was almost dry and the catalyst, together with excess ligand
was largely present in a solid form. In the absence of a second
solvent the rate of removal of reaction products must be balanced
by the rate at which products are formed within the system so as to
maintain a constant volume of reaction solution. Working at 140
bar (Entry 1, Table 1), the control was easier but the overall rate was
still low. In addition, substantial rhodium leaching occurred in the
early part of the reaction. This suggests that the ligand may not be
sufficiently soluble in the mixture for catalyst formation to take
place efficiently and rhodium complexes without bound phos-
4
,5
supercritical fluid — ionic liquid biphasic system. Reaction rates
similar to those obtained using rhodium catalysts for commercial
propene hydroformylation were obtained with very low rhodium
leaching (0.012 ppm in the recovered product), when using a
catalyst formed in situ from [PrMIM][Ph
1
2
2
PC
-propyl-3-methylimidazolium) and [Rh(acac)(CO)
,4-pentanedione) in [OctMIM][bistrifluoromethylsulfonyl amide]
(scCO ) as the
6
H
4
SO
3
] (PrMIM =
2
] (acacH =
(
oct = 1-octyl) and using supercritical CO
2
2
4
transport vector. Although the linear to branched ratio of the
products (l : b) was low (3–4 : 1), this could be improved to 40 : 1
albeit at a slightly lower rate using a ligand based on the xantphos
skeleton.6 A similar process has subsequently been used for a
variety of other reactions involving metal based or enzyme
2
phines are removed because of their solubility in scCO .
7
catalysts. The main problem with the SCF-IL biphasic system for
In order to increase the solubility of the ligand and catalyst and
maintain the substrate product level within the reactor, the reaction
was carried out at 140 bar using the more lipophilic ligand,
alkene hydroformylation is that the overall pressure of the system is
high at 200 bar. This high pressure is required, at least in part,
because the aldehyde has to be extracted from the ionic liquid in
which it is soluble.8 There is also a residual question over the long
term stability and toxicity of the ionic liquid solvent; those with
long alkyl chains having recently been shown to cause skin
[OctMIM][Ph
adjust the CO
2
PC
, and CO/H
6
H
4
SO
3
]. Under these conditions, it was easier to
2
flow rates to obtain a good mass balance
,9
2
2
1
and a reaction rate of 239 catalyst turnovers h was obtained
throughout the 12 h reaction (Fig. 1 and Table 1, Entry 5). Fig. 1
Table 1 Hydroformylation of 1-octene at 100 °C catalysed by rhodium complexes of different phosphines, using scCO
into and the products out of the reactor
2
to transport the substrate and gases
a
[
Rh]
CO flowc/
mmol min
Octene flow/
mmol min
TOF /
h
e
Conversionf
(%)
leaching /
ppm
g
[
Rh]/
[P]/
mol dm
2
CO flow/
nL min
23
23
21
21
21
Pd/bar
21
Entry
Ligand Rb
mol dm
1
2
3
4
5
a
Pr
0.012
0.011
0.0057
0.015
0.015
0.19
0.16
0.043
0.22
0.22
0.47
0.55
0.55
0.65
0.65
b
1.2h
2.89
2.78
1.59
2.34
0.042
0.32
0.32
1.27
1.27
c
140
125
125
140
140
22
80
162
208
239
25
76
77
50
57
35i
0.1–0.5
0.1–0.5
5–10
Pent
Pent
Oct
Oct
5–10
3
= 1 : 1; d Total pressure; e mol product (mol catalyst
Starting solvent is 1-nonanal (16 cm ), nL = normal litre; In [RMIM] [Ph
% aldehyde in recovered product; [Rh] in recovered product; h CO : H
; = 1 : 2; High leaching early in the reaction.
2
2
PC
6
H
4
SO
3
]; CO : H
2
h)2
1
f
g
i
6
12
C h e m . C o m m u n . , 2 0 0 4 , 6 1 2 – 6 1 3
T h i s j o u r n a l i s © T h e R o y a l S o c i e t y o f C h e m i s t r y 2 0 0 4

occurred leading to lower rates, a higher concentration of unreacted
substrate, further precipitation of catalyst and so on. In fact, to
support the catalyst concentrations initially used requires a reaction
solution with a composition that is at least 70 mol % aldehyde. This
is illustrated further by entries 2 and 3 of Table 1; despite a large
difference in the initial concentration of catalyst the same
conversion is observed suggesting that not all of the catalyst is
actually in solution at the higher loadings (although we note that the
increased phosphine concentration will also reduce the rate).
Precipitation of the catalyst was overcome by using a lower catalyst
concentration and a lower flow rate of octene so as to increase its
residence time within the catalyst solution thereby increasing the %
conversion to products.
In conclusion, we have demonstrated that it is possible to carry
out the hydroformylation of medium chain alkenes in a continuous
flow system using scCO
products, but with no other components apart from the substrates,
catalyst and products. Since the scCO can in principle be recycled,
2
as the transport vector for substrates and
Fig. 1 Continuous flow hydroformylation of 1-octene in nonanal (initial
solvent) with scCO as the flowing phase. [OctMIM][Ph PC SO ] (1.45
g, 2.69 mmol) and [Rh(acac)(CO) ] (45.9 mg, 0.18 mmol) in 1-nonanal (16
2
2
H
6 4
3
2
2
21
this represents a process which is potentially emissionless. This
new approach has advantages over our previously reported systems
involving ionic liquids because it can be carried out at much lower
pressures and the one component of the system over which there
was some environmental doubt has been removed. We anticipate
that this kind of process could be applicable to a wide range of
catalytic reactions. Removal of the ionic liquid offers further
advantages in terms of reduced capital costs. The pressure we
currently use (125 bar) is much closer to the pressure used in the
commercial production of medium — long chain aldehydes, which
is carried out using cobalt/phosphene catalysts at 80–100 bar,
however, the temperature we use (100 °C) is much lower than that
used in the commercial cobalt process (180 °C). A preliminary
patent application covering this work has been filed.11
2
mL). Flow rates: CO (0.65 nL min ), synthesis gas (1 : 1, 3.72 mmol
min ), 1-octene (0.2 mL min21, 1.27 mmol min ); 100 °C, 140 bar.
2
1
21
also shows the change in l : b ratio of the recovered aldehydes
during the reaction. Initially the only aldehyde present was nonanal
(it was the starting solvent), but the l : b ratio changed smoothly
towards the steady state value of 3.2 : 1 characteristic of the
products expected from the reaction using this type of ligand.4
Reducing the flow rate of synthesis gas, but retaining the overall
2
pressure by increasing the CO partial pressure, reduces the
reaction rate (compare Entries 5 and 4 of Table 1), as observed
4
when using an ionic liquid, because the octene partitions less into
the liquid phase, where the catalyst is located. Changing the alkyl
chain length on the imidazolium cation has a more dramatic effect.
We thank the EPSRC for a Fellowship (P. B. W.).
2 6 4 3
[PrMIM][Ph PC H SO ] is not sufficiently soluble in the organic
phase to give good catalysis. This has two effects, firstly the
reaction rate is low, but secondly the rhodium leaching is high early
in the reaction because rhodium complexes are formed which do
Notes and references
1 B. Cornils, in Applied Homogenous Catalysis with Organometallic
Compounds, ed. B. Cornils and W. A. Herrmann, Weinheim, 1996.
2 D. J. Cole-Hamilton, Science, 2003, 299, 1702.
not contain phosphine, and these are soluble in scCO
MIM][Ph PC SO ] is very soluble in the organic phase and
therefore gives good reaction rates but has some solubility in
scCO , hence the leaching. [PentMIM][Ph PC SO ] has inter-
mediate properties; it dissolves in the organic phase, but not at such
high levels as [OctMIM][Ph PC SO ], and shows reasonable
2
. [Oct-
2
6
H
4
3
2
2
6
H
4
3
3
C. C. Tzschucke, C. Markert, W. Bannwarth, S. Roller, A. Hebel and R.
Haag, Angew. Chem., Int. Ed. Engl., 2002, 41, 3964.
2
6
H
4
3
4
P. B. Webb, M. F. Sellin, T. E. Kunene, S. M. Williamson, A. M. Z.
Slawin and D. J. Cole-Hamilton, J. Am. Chem. Soc., 2003, 125,
catalytic activity but the leaching into the collected product is much
reduced.
Whereas [OctMIM][Ph PC H SO ] is highly soluble in nonanal
2 6 4 3
the [PentMIM] analogue only shows appreciable solubility at
elevated temperatures. Initial attempts to hydroformylate octene
1
5577.
M. F. Sellin, P. B. Webb and D. J. Cole-Hamilton, Chem. Commun.,
001, 781.
5
2
6 T. E. Kunene, P. B. Webb and D. J. Cole-Hamilton Chem. Commun.,
2003, manuscript in preparation.
7 S. V. Dzyuba and R. A. Bartsch, Angew. Chem., Int. Ed. Engl., 2003, 42,
continuously using [PentMIM][Ph
identical to those used for [OctMIM][Ph
2
PC
6
H
4
SO
PC
3
] under conditions
SO ], were un-
2
6
H
4
3
1
48.
L. A. Blanchard, D. Hancu, E. J. Beckman and J. F. Brennecke, Nature,
999, 399, 28.
L. A. Blanchard and J. F. Brennecke, Ind. Eng. Chem. Res., 2001, 40,
87.
successful. GC analysis of liquid fractions recovered downstream
revealed a gradual drop in the % aldehyde despite some initial
conversion of substrate as indicated by the presence of branched
product. In this case a steady state reaction mixture was never
achieved. This rapid drop in substrate conversion occurred because
although the phosphine is soluble in nonanal it is insoluble in
octene/aldehyde mixtures. As the concentration of alkene within
the reaction solution increased, precipitation of the catalyst
8
9
1
2
1
0 B. Jastorff, R. Stormann, J. Ranke, K. Molter, F. Stock, B. Oberheit-
mann, W. Hoffmann, J. Hoffmann, M. Nuchter, B. Ondruschka and J.
Filser, Green Chem., 2003, 5, 136.
11 P. B. Webb and D. J. Cole-Hamilton, GB Patent, 2003, 0300733.3.
C h e m . C o m m u n . , 2 0 0 4 , 6 1 2 – 6 1 3
613