Selective catalytic hydrogenation of organic compounds in supercritical fluids as a continuous process

Article abstract of DOI:10.1021/op970056m

We report a new method for continuous hydrogenation in supercritical fluids (CO₂ or propane) using heterogeneous noble metal catalysts on Deloxan aminopolysiloxane supports. The method has considerable promise both for laboratory-scale hydrogenation and for the industrial production of fine chemicals. It can be applied to a wide range of organic compounds including alkenes, alkynes, aliphatic and aromatic ketones and aldehydes, epoxides, phenols, oximes, nitrobenzenes, Schiff bases, and nitriles. Conversion of starting materials, product selectivity, and space-time yields of the catalyst are all high, and the reactors themselves are very small (5- and 10-mL volume). Supercritical hydrogenation enables the reaction parameters to be controlled very precisely. Results are presented for a series of different reactions showing product distributions, which are dependent on temperature, pressure, H₂ concentration, and the loading and nature of the catalyst. The hydrogenation of cyclohexene has been studied in some detail, and our results are related to the phase diagrams of the ternary system cyclohexane + CO₂ + H₂, which we present in a novel way, more suited to continuous reactors. Finally, we report that the supercritical hydrogenation of isophorone has advantages over conventional methods.

Full text of DOI:10.1021/op970056m

Organic Process Research & Development 1998, 2, 137−146
Articles
Selective Catalytic Hydrogenation of Organic Compounds in Supercritical Fluids
as a Continuous Process
†
†
‡
,†
Martin G. Hitzler, Fiona R. Smail, Stephen K. Ross, and Martyn Poliakoff*
The Department of Chemistry, The UniVersity of Nottingham, Nottingham, UK NG7 2RD, and Thomas Swan & Co. Ltd.,
Crookhall, Consett, UK DH8 7ND
Abstract:
processes. Applications range from extraction and chroma-
tography in analytical chemistry to reaction chemistry and
We report a new method for continuous hydrogenation in
supercritical fluids (CO
1
-9
2
or propane) using heterogeneous noble
preparation of new materials.
These fluids are gases
) and pressurized
c
above their critical pressures (P ). Their densities are
metal catalysts on Deloxan aminopolysiloxane supports. The
method has considerable promise both for laboratory-scale
hydrogenation and for the industrial production of fine chemi-
cals. It can be applied to a wide range of organic compounds
including alkenes, alkynes, aliphatic and aromatic ketones and
aldehydes, epoxides, phenols, oximes, nitrobenzenes, Schiff
bases, and nitriles. Conversion of starting materials, product
selectivity, and space-time yields of the catalyst are all high,
and the reactors themselves are very small (5- and 10-mL
volume). Supercritical hydrogenation enables the reaction
parameters to be controlled very precisely. Results are pre-
sented for a series of different reactions showing product
distributions, which are dependent on temperature, pressure,
heated above their critical temperatures (T
c
1
0,11
comparable to those of organic liquids, and the gaslike nature
of the fluids renders them completely miscible with perma-
1
2
nent gases such as H
2
. By contrast, the solubility of
gaseous H in conventional organic solvents is relatively low.
2
Thus, supercritical fluids are potentially attractive solvents
for hydrogenation reactions with significant advantages over
conventional methods, particularly because gas-phase reac-
tions often generate significant amounts of by-products and
conversion can be poor in liquid-phase reactions.
The properties of supercritical fluids make them highly
suitable as solvents for continuous flow reactors. Compared
to liquid-phase reactions, reactions in supercritical fluids are
characterized by reduced viscosity and enhanced mass
transfer. In addition, the good thermal transport properties
2
H concentration, and the loading and nature of the catalyst.
The hydrogenation of cyclohexene has been studied in some
detail, and our results are related to the phase diagrams of the
ternary system cyclohexane + CO
2
2
+ H , which we present in
(
1) Savage, Ph. E.; Gopalan, S.; Mizan, T. I.; Martino, Ch. J.; Brock, E. E.
AIChE J. 1995, 41, 1723.
a novel way, more suited to continuous reactors. Finally, we
report that the supercritical hydrogenation of isophorone has
advantages over conventional methods.
(2) Mandel, F. S. Proc. Conf. Fluorine in Coatings II, Munich, Germany;
PRA: Teddington, UK, 1997; Paper 10.
(
(
(
(
3) Brennecke, J. F. Chem. Ind. (London) 1996, 831.
4) Phelps, C. L.; Smart, N. G.; Wai, C. M. J. Chem. Educ. 1996, 73, 1163.
5) Clifford, T.; Bartle, K. Chem. Ind. (London) 1996, 449.
6) Kaupp, G. Angew. Chem., Int. Ed. Engl. 1994, 33, 1452; Angew. Chem.
1
994, 106, 1519.
Introduction
(
7) Katritzky, A. R.; Allin, S. M. Acc. Chem. Res. 1996, 29, 399.
Supercritical fluids are becoming increasingly attractive
as solvents for environmentally more acceptable chemical
(8) Parsons, E. J. CHEMTECH 1996, 30.
9) Hatakeda, K.; Ikushima, Y.; Ito, S.; Saito, N.; Sato, O. Chem. Lett. 1997,
45.
(
2
(10) McHugh, M. A.; Krukonis, V. J. Supercritical Fluid Extraction: Principles
*
Martyn.Poliakoff@nottingham.ac.uk.
University of Nottingham.
Thomas Swan & Co. Ltd.; skross@macline.co.uk.
& Practice; Butterworth-Heinemann, Boston, 1994.
(11) Eckert, C. A.; Knutson, B. L.; Debendetti, P. G. Nature 1996, 383, 313.
(12) Tsang, C. Y.; Streett, W. B. J. Eng. Sci. 1981, 36, 993.
†
‡
S1083-6160(97)00056-X CCC: $15.00 © 1998 American Chemical Society and Royal Society of Chemistry
Published on Web 02/24/1998
Vol. 2, No. 3, 1998 / Organic Process Research & Development
•
137

of supercritical fluids are an advantage because hydrogena-
tion is usually a highly exothermic reaction. However,
relatively few articles have been published on hydrogenation
in supercritical fluids,13-23 and most of those have involved
hydrogenation as a batch process, carried out in sealed
autoclaves. Reports on supercritical hydrogenation as a
continuous process are even rarer.
Industry, in particular, favours continuous processes
because they are more cost efficient and the reactors can be
kept smaller in size.2
3,24
This reduction in size reduces both
costs and safety problems of the high-pressure equipment
needed for supercritical reactions. There have been two
investigations into the hydrogenation of fats and oils in
continuous flow reactors using near-critical or supercritical
Figure 1. Flow apparatus for catalytic hydrogenation in
supercritical fluids, based on high-pressure modules (NWA
GmbH, L o1 rrach, Germany); see text for details of operation.
The components are labeled as follows: DU, pneumatically
CO
2
(scCO
2
, T
c
) 31.1 °C, P
c
) 73.8 bar) and propane
1,22
(scPropane, T
c
) 96.8 °C, P
c
) 42.6 bar).2
In a very
25
operated dosage unit for adding H
valves to control the flow rate and to separate the product(s)
from the fluid and excess H ; M, mixer; onIR, on-line FTIR
2
; E1 and E2, expansion
recent communication, we described a much more general
method for the selective hydrogenation of organic substrates
in supercritical fluids using a laboratory-scale continuous
flow reactor.
2
monitoring (Nicolet Impact 410); P, pump for the organic
substrate (Gilson 305); PH, preheater; PP, pneumatic pump
for CO
sor for H
2
or propane (module PM 101); PU, pneumatic compres-
(module CU 105); R, reactor (plus heater block not
In this paper, we describe our approach to supercritical
hydrogenation in greater detail. We begin by summarizing
the essential features of our reactor. Then, we describe a
series of reactions which illustrate the breadth of chemistry
accessible via the reactor. Next, we report a more detailed
study on the hydrogenation of alkenes, aimed at investigating
the effects of carrying out such reactions on a larger scale,
and in the final section, we describe the supercritical
hydrogenation of isophorone, a reaction with significant
industrial and environmental benefits.
2
shown) which is fitted with three thermocouples as illustrated
in Figure 8; S, pressure regulator for controlling the system
pressure; V, vent with flow meter (not shown).
their particular apparatus meets the necessary safety re-
quirements. The indiVidual components, which we describe
below, work well, but they are not necessarily the only
equipment of this type aVailable nor the most suitable for
the purpose.
Figure 1 shows a schematic view of the apparatus. The
pump, PP (PM 101, NWA GmbH), compresses the super-
critical fluid to maintain a system pressure of 40-200 bar,
which is controlled by the regulator, S. The organic substrate
is pumped with a standard HPLC pump, P (0.3-20.0 mL/
Experimental Section
Safety Hazard. CAUTION! The experiments described
in this paper inVolVe the use of relatiVely high pressures
and require equipment with the appropriate pressure rating.
It is the responsibility of indiVidual researchers to Verify that
min, Gilson 305). H
CU 105, NWA GmbH), and is added pulsewise via a dosage
unit, DU (based on pneumatic Rheodyne 6-port injection
valve) at a pulse rate chosen to give the desired H :substrate
ratio. The fluid, substrate, and H mixture is stirred
2
is compressed to 200-280 bar, PC
(
(
13) Howdle, S. M.; Healy, M. A.; Poliakoff, M. J. Am. Chem. Soc. 1990, 112,
804.
14) Rathke, J. W.; Klingler, R. J.; Krause, T. R. Organometallics 1991, 10,
350.
2
4
(
2
1
mechanically and preheated, M+PH. The reaction mixture
then enters the reactor, R, which can be heated independently
by an aluminium block. We found that alkynes, alkenes,
and Schiff bases reacted readily at an initial reactor temper-
ature of 40 °C but, in most cases, additional heating of the
reactor was necessary.
(
15) Jessop, Ph. G.; Ikariya, T.; Noyori, R. Nature 1994, 368, 231.
(16) Jessop, Ph. G.; Hsiano, Y.; Ikariya, T.; Noyori, R. J. Am. Chem. Soc. 1996,
1
18, 344.
17) Burk, M. J.; Feng, S.; Gross, M. F.; Tumas, W. J. Am. Chem. Soc. 1995,
17, 8277.
18) Minder, B.; Mallat, T.; Pickel, K.-H.; Steiner, K.; Baiker, A. Catal. Lett.
995, 34, 1.
(
(
1
1
(
(
19) Kr o¨ cher, O.; K o¨ ppel, R. A.; Baiker, A. Chem. Commun. 1996, 1497.
20) Pickel, K.-H.; Steiner, K. Proc. Int. Symp. Supercrit. Fluids, 3rd 1994, 3,
In the experiments described here, we used two inter-
changeable reactors made of 316 stainless steel tubing
2
5.
(
(
21) Tacke, T.; Wieland, S.; Panster, P. Process Technol. Proc. 1996, 12, 17.
22) H a¨ rr o¨ d, M.; Møller, P. Process Technol. Proc. 1996, 12, 43. H a¨ rr o¨ d, M.;
Macher, M. B.; H o¨ gberg, J.; Møller, P. Proc. Ital. Conf. Supercrit. Fluids
Their Appl., 4th Italy, 1997, 319.
(9-mm i.d.): length 78 mm (5-mL volume) and length 152
mm (10-mL volume). A frit at the bottom of the reactor
keeps the catalyst in place. The system has three thermo-
couples situated (i) inside the catalyst bed (Tcat), (ii) against
the reactor wall (Twall), and (iii) in the product stream leaving
the reactor (Tout). All three are used for monitoring the
reaction, but in practice, we have found that Twall gives the
most reproducible control of the heater.
(23) Roche Mag. 1992, 41, 2.
(24) Tundo, P. Continuous Flow Methods in Organic Synthesis; Ellis Horwood:
Chichester, UK, 1991.
(
25) Hitzler, M. G.; Poliakoff, M. Chem. Commun. 1997, 1667. Preliminary oral
presentations of this work have been given at the 4th International
Symposium on Supercritical Fluids, Sendai, Japan, May 1997, the Final
Meeting of EU COST Action D6, Santorini, Greece, June 1997, and the
6
th European Symposium on Organic Reactivity, Louvain, La Neuve,
Belgium, July 1997.
138
•
Vol. 2, No. 3, 1998 / Organic Process Research & Development

After passing through the reactor, the product stream can
various parameters in a supercritical flow reactor can be
controlled more or less independently. For example, in a
flow reactor, unlike a batch system, the temperature can be
altered without causing any change in the pressure inside
the reactor. Thus, reaction conditions can be controlled with
considerable precision and with high reproducibility of the
results.
2
6
be optionally flowed through a high-pressure IR cell for
on-line monitoring (Nicolet Impact 410 FTIR interferometer)
before the pressure is dropped stepwise to separate the
2
product(s) from the fluid and excess H using an expansion
module (PE 103, NWA GmbH). In this module, both of
the expansion valves E1 and E2 control the flow rate of the
exhaust gases. Additionally, a flow meter (not illustrated)
is connected to the vent (V) to measure the flow rate of the
fluid, which is typically set to 0.5-1.5 L of gas at 1 atm
(a) Hydrogenation of m-Cresol, 1: Effects of Temper-
ature and of H concentration. Our 5-mL reactor is a
2
relatively short tube so that residence times are <5 min with
and 20 °C. This flow rate corresponds to 0.9-1.8 g of CO
2
/
the CO flow rates typically used in our experiments. This
2
min. Thus, in summary, the fluid pressure is controlled at
the start of the system, together with the flow rates of the
substrate and/or H . The total flow of the reaction mixture
2
means that high reaction temperatures are needed to achieve
full conversion of substrates which have high activation
energies for hydrogenation. The disadvantage, however, is
that product selectivity decreases rapidly with increasing
temperature. This problem can be overcome, when neces-
sary, by using a longer reactor; the 10-mL reactor doubles
the residence time of a substrate, and a lower temperature
can be used with a correspondingly improved selectivity.
Thus, we have investigated the hydrogenation of m-cresol,
is controlled at the end of the system. This arrangement
means that increasing the flow rate of reactants will
automatically reduce the flow of fluid unless the total flow
is also increased. Thus, the faster the reactor runs, the less
fluid is used. This has important and beneficial consequences
for the operation of the reactor, as discussed in the last
sections of this paper.
A number of catalysts were tried, including Pt/C and Ni
metal. Several were effective, apart from Pt/C, which tended
to block the system, but we found Deloxan precious metal
catalysts (Degussa AG, Frankfurt, Germany) to be the most
reliable and convenient catalysts for use in our continuous
flow reactor. Therefore, these catalysts, which are based on
aminopolysiloxane supports, have been used in all the
experiments described here.
The organic products are recovered free of any solvent.
Thus, they can immediately be analyzed by NMR (Bruker
DPX 300), GC (Philips PU 4500), or GC-MS (Shimadzu
QP-5000) without further work-up. All reactions were run
using commercially available substrates (Aldrich). The
reactor described in this paper gives a good degree of
reproducibility between runs.
1
, in scCO
Pd APII Deloxan catalyst (particle size: 0.3-0.8 mm)
Scheme 1). Both the total pressure (120 bar) and the flow
2
in a 10-mL reactor containing 9 mL of the 5%
(
rate of 1 (0.5 mL/min) were kept constant throughout the
investigation. Even with this 10-mL reactor, temperatures
of at least 250 °C and a H :1 ratio of 3.5:1 were needed to
2
obtain 100% reaction of 1, see Figure 2. Under these
conditions, 3-methylcyclohexanone, 2, and 3-methylcyclo-
hexanol, 3, were formed in yields of 42% and 55%,
respectively. Increasing the temperature favoured hydro-
genolysis to methylcyclohexane, 4, and small amounts of
toluene, 5. Figure 3 shows how the yield of 3 can be raised
2
at the expense of 2 when the ratio of H :1 is increased from
3
.5:1 to 5:1 at a constant temperature of 250 °C. On the
2
other hand, lowering the H :1 ratio to 2.5:1 increases the
yield of 2 to 61.5%. It should be noted that the 5% Pd
Deloxan catalyst survived for at least 16 h with no signs of
coking or degradation even at reactor temperatures of up to
Results and Discussion
Table 1 summarizes some of the reactions which we have
investigated. These reactions have been chosen to test
different features of the reactor. The hydrogenations have
involved a wide range of organic functional groups including
alkenes, alkynes, aliphatic and aromatic ketones and alde-
hydes, epoxides, phenols, oximes, nitrobenzenes, Schiff bases
and nitriles. The hydrogenolysis of aliphatic alcohols and
ethers has also been studied. For those hydrogenations where
several products were possible, we have investigated the
effect of the reactor conditions on the selectivity of the
reaction. We have found that the most important reaction
parameters are the temperature of the reactor and the
4
00 °C.
b) Hydrogenation of Aldehydes and Ketones: Com-
(
parison of Aromatic and Aliphatic Compounds. The
hydrogenation of the CO group of aromatic ketones or
aldehydes required temperatures lower than those used for
the hydrogenation of m-cresol. Thus, benzaldehyde, 6, could
be hydrogenated quantitatively at only 95 °C (Scheme 2).
Using the 5-mL reactor filled with 4 mL of the 5% Pd APII
Deloxan catalyst, benzyl alcohol, 7, was isolated in 92% yield
as the major product, see Figure 4. The yield of 7 decreased
rapidly as the temperature of the reactor was increased; at
1
40 °C, toluene, 5, was formed in 77% yield. Significant
hydrogenation of the aromatic ring was observed only above
00 °C, and a sample collected at 270 °C contained 49%
2
concentration of H . Surprisingly, changes in the absolute
pressurestypically values of 40-200 barsonly became
important when the pressure was close to or below the critical
pressure of the fluid.
Compared to conventional methods of hydrogenation, the
particular advantage of using supercritical fluids is that the
2
methylcyclohexane, 4, and equal proportions of toluene, 5,
and cyclohexane, 9 (22-25%).
In contrast to benzaldehyde, propionaldehyde, 10, could
2
not be hydrogenated in scCO over the 5% Pd catalyst even
at temperatures as high as 300 °C. Better results were
(26) Poliakoff, M.; Kazarian, S. G.; Howdle, S. M. Angew. Chem., Int. Ed. Engl.
1
995, 34, 1275.
obtained with a different catalyst, 5% Ru APII Deloxan (4
Vol. 2, No. 3, 1998 / Organic Process Research & Development
•
139

Table 1. Summary of the results of supercritical hydrogenation described in this paper
140
•
Vol. 2, No. 3, 1998 / Organic Process Research & Development

Table 1 (Continued)
Scheme 1. Hydrogenation of m-cresol, 1
Figure 3. Plot illustrating the dependence of the product
distribution on the ratio of H : m-cresol, 1 (0.5 mL/min), in
scCO (0.65 L/min) at constant pressure (120 bar total) and
temperature (Twall ) 250 °C); catalyst 5% Pd APII Deloxan
9 mL) and product analysis by GC. The compounds are
2
2
(
numbered as in Scheme 1.
Scheme 2. Hydrogenation of benzaldehyde, 6
Figure 2. Plot showing the dependence of the product
distribution on the temperature of the reactor wall, Twall, for
2
the hydrogenation of m-cresol, 1 (0.5 mL/min), in scCO (0.65
L/min) at 120-bar total pressure with 5% Pd APII Deloxan
catalyst (9 mL). The compounds are numbered as in Scheme
2
1. Note that the ratio of H :1 was increased from 2:1 to 6:1 at
higher temperatures. Product analysis was by GC. In this and
subsequent figures, the solid curves are drawn merely to aid
visualization of the data.
(e.g., 150 °C) the Ru catalyst converted a substantial amount
of 11 into propene and water.
mL, particle size: <0.2 mm). However, even with this
catalyst, the overall conversion of 10 was only 68% with a
product selectivity to 1-propanol, 11, of only 35%, see Table
We found analogous differences in behavior between the
hydrogenation of aromatic and aliphatic ketones. For
example, acetophenone, 12, could be hydrogenated selec-
1. It appears that even at relatively modest temperatures
2
tively in scCO over the 5% Pd APII Deloxan catalyst to
Vol. 2, No. 3, 1998 / Organic Process Research & Development
•
141

Scheme 4. Hydrogenation of cyclohexanone, 17, and
cyclohexanol, 18
Scheme 5. Formation of carbamic acids from amines and
CO
2
from approximately 30% to 60%. The most obvious
rationalization is that, at constant flow rate, increasing the
pressure increases the residence time of the substrate in the
reactor and hence the degree of hydrogenation.
By contrast, the aliphatic ketone 17 was not hydrogenated
over the 5% Pd catalyst. It was hydrogenated selectively,
however, to cyclohexanol, 18, at 300 °C and a pressure of
Figure 4. Plot showing the dependence of the product
distribution on Twall (5-mL reactor) for the hydrogenation of
benzaldehyde, 6 (0.5 mL/min), in scCO
total pressure with 5% Pd APII Deloxan catalyst (4 mL). The
ratio of H :6 was increased from 2:1 to 8:1 at higher temper-
2
(1.0 L/min) at 120-bar
2
1
atures; product analysis by H NMR (CDCl
3
). The compounds
are numbered as in Scheme 2.
1
20 bar by a 3% Ru APII Deloxan catalyst (particle size:
Scheme 3. Hydrogenation of acetophenone, 12
0.8-1.8 mm) (Scheme 4). Even so, the yields of 18 were
low (17%), and the reaction generated by-products when run
at higher temperatures and pressures.
Interestingly, the hydrogenolysis of cyclohexanol, 18,
itself was best promoted by the 5% Pd catalyst. Even at
temperatures as high as 400 °C, the selectivity for cyclo-
hexane, 9, was 60% compared to the mere 15% which could
be achieved with the 5% Ru catalyst.
(
c) Hydrogenation of Cyclic Ethers. Tetrahydrofuran
THF), 21, can be obtained in more than 95% selectivity by
hydrogenation of furan, 20, over a 5% Pd catalyst in scCO
at 120 bar. A very good conversion of 98% was achieved
at 300-350 °C with at least a 3-fold excess of H , see Table
. The only side reaction was hydrogenolysis of 21 to
-butanol, 22 (2%). We were unable to improve the
(
2
2
1
1
efficiency of the direct transformation of 21 to 22 beyond
0% even under harsher conditions (e.g., 400 °C, 160 bar)
or by increasing the residence times with a reduced flow of
CO (e.g., from 0.75 to 0.2 L/min).
d) Hydrogenation of Nitro Compounds: The Use of
scPropane. Propane was favoured as the supercritical fluid
for the hydrogenation of nitrogen-containing substrates, such
as oximes, Schiff bases, nitriles, and nitrobenzenes, to the
corresponding amines, to avoid the formation of insoluble
carbamic acid salts. Carbamic acids are formed by the
1
2
(
Figure 5. Plot of the pressure dependence of the product
distribution for the hydrogenation of acetophenone, 12 (0.5 mL/
min), in scCO
Deloxan catalyst (4 mL). The ratio of H
:1 throughout the experiment while the total pressure was
2
(1.0 L/min) at Twall ) 240 °C with 5% Pd APII
2
:12 was constant at
2
1
varied; product analysis by H NMR (CDCl
are numbered as in Scheme 3.
3
). The compounds
reaction of amine groups with CO
we have found that scCO can be used for all reactions other
than those producing amines. Of course, scPropane has the
2
, see Scheme 5. In general
products 13-16, Table 1 (Scheme 3). As we have already
2
25
reported, it is possible to maximize the yield of any of the
hydrogenation products by varying the temperature between
2
advantage of a lower critcal pressure than CO , but this is
2
90 and 300 °C, and increasing the H :12 ratio (from 2:1 to
offset by its flammability.
6
:1) at a constant pressure of 120 bar, see Table 1.
The total pressure had an interesting effect on the product
Thus, nitrobenzene, 23, was hydrogenated (0.5 mL/min)
with a 2-fold excess of H in scPropane (0.75 L/min) at 80
2
distribution, see Figure 5. At 240 °C, the yields of
ethylbenzene, 14, and its ring-hydrogenated species 16 were
substantially affected by pressure changes while the yields
of 13 (0%) and 15 (4%) remained unchanged. The most
dramatic changes to 14 and 16 happened near and just below
bar in the 5-mL reactor (150-200 °C) over a 1% Pd APII
Deloxan catalyst (particle size: 0.2-0.5 mm) (Scheme 6).
Under these conditions, conversion to aniline, 24, was
quantitative with no impurities detectable by either GC or
NMR.
the critical pressure of CO
2
(74 bar). Over the pressure range
The distribution of products, 24-26 (and 9), was strongly
dependent on the loading of the catalyst and the choice of
40-140 bar, the yield of ethylcyclohexane, 16, was doubled
142
•
Vol. 2, No. 3, 1998 / Organic Process Research & Development

Scheme 6. Hydrogenation of nitrobenzene, 23
Scheme 7. Hydrogenation of Schiff bases, 27, and
oximes, 29
Scheme 8. Full hydrogenation of acetylenes and olefins
Figure 7. On-line FTIR spectra showing the differences in the
absorptions of of 1-octyne, 33, 1-octene, 34, and octane, 35;
Figure 6. The effects of catalyst on the product distribution
from the hydrogenation of nitrobenzene, 23, in scPropane at
spectra were recorded during hydrogenation at 60 °C in CO
2
1
20 bar and at temperatures of 200-250 °C with a 3-fold excess
of H . Panel a shows how 5% Pd APII Deloxan catalyst
transformed 23 into aniline, 24 (9%), cyclohexylamine, 25
29%), dicyclohexylamine, 26 (58%), and traces of cyclohexane,
(1%). Panel b shows that, with a 5% Pt APII Deloxan catalyst,
at 120 bar. The bands are all ν(C-H) vibrations; the left-hand
2
arrow indicates the ν(C-H) band of the acetylenic proton in
3
3 while the right-hand arrow shows the ν(C-H) band of the
(
9
olefinic protons in 34.
the amount of cyclohexane, 9, increased to 28% at the expense
of 25 (13%) and 26 (4%) with aniline, 24, as the major product
(
f) Hydrogenation of C-C Multiple Bonds: On-Line
Monitoring and Scale-Up. As might be expected, the
hydrogenation of CtC triple bonds or CdC double bonds
(46%). By comparison, 1% Pd APII Deloxan catalyst gave
quantitative conversion of 23 to aniline, 24, at 200 °C; analysis
by GC.
2
to single bonds was easily performed in either scCO or
the noble metal itself, see Figure 6. The formation of 9 is
particularly striking because its formation requires the
scPropane (Scheme 8). Both the 5% Pd and 5% Pt APII
Deloxan were extremely effective catalysts for this reaction
addition of no less than 7 mol of H
2
/mole of 23. This
2
while Ru on Deloxan showed poor conversion. Using CO ,
underlines the high efficiency and process intensification
which are possible in supercritical flow reactors.
a reactor temperature of 40 °C and a pressure of 120-140
bar were quite sufficient to start the reaction. With propane,
the reactor was preheated to 100 °C to be above the critical
(e) Hydrogenation of Oximes and Schiff Bases. Oximes
and Schiff bases were hydrogenated in scPropane to the
corresponding amines in high yields and with high selectivity
using the 5-mL reactor. For example, benzylmethylamine,
temperature of the fluid (T
The reactions in scCO
on-line FTIR, which proved to be effective for this purpose
c
) 96.8 °C).
were monitored in real time by
2
26
2
8, was formed in excellent yield (99%) from N-benzyli-
because, for example, octane, 35, showed no absorptions in
-
1
denemethylamine, 27, at temperatures of 40-50 °C and a
pressure of 120 bar (Scheme 7). Furthermore, toluene, 5,
was obtained in quantitative yield by increasing the temper-
ature to 200 °C and the ratio of H
excess.
the region of 3000-3500 cm while the starting materials
1-octyne, 33, and 1-octene, 34, had characteristic peaks in
this region, see Figure 7. The advantage of on-line FTIR
over on-line gas chromatography is that FTIR is a much
faster method, allowing immediate remedial action to be
taken as soon as loss of catalyst activity or product selectivity
is observed. Thus, FTIR provides an extremely versatile tool
for exploring the performance parameters of laboratory flow
2
from a 2-fold to a 4-fold
2
-Butanone oxime, 29, reacted with H at 150-200 °C
2
and 80 bar over either the 5% Pd or the 5% Pt APII Deloxan
catalyst. In both cases, the conversion of 29 was >95%.
The selectivity for 2-butylamine, 30, was 80.0% with the
Pd and 85.5% with the Pt catalyst. A side reaction led to
the formation of 2-butanone, 31 (10.0-17.5%), and 2-bu-
tanol, 32 (0.5-1.5%).
2
7
reactors (e.g., residence time ). Furthermore, FTIR may
(
27) Buback, M.; Hinton, C. Chapter 4 in High-Pressure Techniques in Chemistry
and Physics; Holzapfel, W. B., Isaacs, N. S., Eds.; OUP: Oxford, UK, 1997;
Chapter 4, p 151.
Vol. 2, No. 3, 1998 / Organic Process Research & Development
•
143

well have considerable potential for process optimization and
control of supercritical hydrogenation on an industrial scale.
Most of the reactions described above were carried out
with a relatively low flow rate of substrate, 0.3-2.0 mL/
min. To explore the limitations of substrate throughput, the
hydrogenation of 1-octene, 34, was tested at a considerably
higher flow rate, 10.0 mL/min, using the 5-mL reactor filled
with 4 mL of the 5% Pd APII Deloxan catalyst, see Table
Figure 8. Schematic view of the 5-mL reactor showing the
positions of the three thermocouples, Tcat, Tout, and Twall. The
wall thermocouple could be moved up and down the 5-mL
reactor over a distance of 4 cm to locate the position of the
hottest zone. When the reactor was used with a heating block,
1. Even at this rate, octane, 35, was formed quantitatively
over periods of up to 4 h without significant degradation of
catalyst performance or leaching of Pd metal. Other catalysts
which we tried were less successful; powdered Pd/C (Ald-
Twall was located in a fixed position in the block itself and its
2 3
rich) blocked the reactor, and Pd leached from Pd/Al O
signal was used to maintain the reactor at constant temperature.
(Degussa AG) at higher temperatures.
Cyclohexene, 36, was tested at even higher flow rates,
20 mL/min, the limit of our current pump. Again, hydro-
genation at the highest flow rate remained quantitative over
a period of hours. Hydrogenation at this rate represents a
high degree of efficiency. Conventionally, the efficiency of
a reactor is judged by its liquid hourly space velocity, LHSV
(volume of substrate converted per hour by a unit volume
of catalyst) or its space-time yield (kg of product/h per cubic
meter of catalyst). With a flow rate of 36 of 20 mL/min
and 4 mL of 5% Pd on Deloxan, our reactor achieved a very
-
1
5
high LHSV (300 h ) and space-time yield (2.5 × 10 kg
-1
-3
h
m ). In general, space-time yields are highest for very
small reactors where problems of heat removal are much
less serious. Therefore it is gratifying to find that our space-
time yield is comparable to that calculated for the hydroge-
nation of fats22 in a reactor of much smaller volume (7 µL)
Figure 9. Plots showing the dependence of temperature as
measured by (9) Tcat, (b) Tout, and (2) Twall on substrate flow
rate. Data were recorded for the hydrogenation of cyclohexene,
than ours (5 mL).
36, in scCO
catalyst (4 mL). The flow rate of 36 was increased from 0.5 to
0.0 mL/min with a 2-fold H excess. The overall gaseous flow
2
at 120-bar total pressure with 5% Pd APII Deloxan
The initial impetus for our research, and that of others,22
1
2
was the ability of a supercritical fluid to bring the organic
substrate and H into a single homogeneous phase. Super-
2
rate was held constant at 1.0 L/min (measured at atmospheric
pressure). The overall conversion of 36 to 9 remained at 96-
critical hydrogenation is demonstrably effective, but it is still
important to show whether the hydrogenation really does
occur in a single phase. We have therefore investigated the
hydrogenation of cyclohexene, 36, to cyclohexane, 9, in some
detail.
1
3
98% as analysed by H NMR (CDCl ).
point moved downwards and then remained at the bottom
of the reactor up to the maximum flow rate.
These measurements show that, at the higher flow rates,
The complete reaction involves a quaternary mixture (36,
c
the temperature of the catalyst bed is higher than the T of
9
, H
2
, and CO
2
). Fortunately, the critical temperatures of
36. Under these conditions, all of the components are
supercritical and therefore the system will be single-phase.
However, this will not necessarily be true at lower temper-
atures, so we have modeled the system to identify the areas
36 (T
c
) 287.3 °C) and 9 (T
c
) 280.3 °C) are sufficiently
close for the system to be considered as a ternary mixture.
Even so, phase investigations of ternary mixtures can be quite
involved, but the fact that our reactor runs at constant
pressure somewhat restricts the volume of phase space which
needs to be investigated. Thus, the temperature (Tcat, Twall
T
of miscibility for the system CO
for the system propane + H + cyclohexane at various
temperatures and a constant pressure of 120 bar using the
2
+ H
2
+ cyclohexane and
2
,
2
8,29
Peng-Robinson equation of state.
The results of such
out) was monitored at various points on the reactor, see
calculations are conventionally represented as a so-called
phase prism. However, our results are presented in a novel
manner, Figure 10. This representation is particularly
appropriate for the flow reactor because the regions of
miscibility (dark areas) are much easier to visualize than in
a phase prism. These calculations show that (a) at any
temperature, the miscibility of the system always improves
Figure 8. Figure 9 shows how these temperatures varied as
the flow rate of 36 was increased while the total pressure
was held constant at 120 bar. At 0.5 mL/min, Tcat stabilized
at 160 °C while Twall and Tout remained at 50 °C. Tcat
increased rapidly to 340 °C at a flow rate of 2.0 mL/min
but then remained constant over a narrow range from 340
to 360 °C as the flow was increased to 10.0 mL/min. At
flow rates of 0.5-3.0 mL/min, the highest value of Twall was
measured at a point within 0.5 cm from the top of the reactor.
As the flow rate increased from 4.0 to 6.0 mL/min, the hottest
(28) Kordikowski, A.; Robertson, D. G.; Poliakoff, M.; DiNoia, T. D.; McHugh,
M. A.; Aguiar-Ricardo, A. J. Phys. Chem. B 1997, 101, 5853.
(29) Peng, D. Y.; Robinson, D. B. Ind. Eng. Chem. Fundam. 1976, 15, 59.
144
•
Vol. 2, No. 3, 1998 / Organic Process Research & Development

3,3,5-trimethylcyclohexanone (dihydroisophorone), 38, which
is used as solvent for vinyl resins, lacquers, varnishes, paints,
and other coatings.
Most of the hydrogenation processes described in the
patent literature either show low overall conversion but high
selectivity for 38 or give high conversion with poor selectiv-
3
1
ity. A process which combines high conversion and high
selectivity is very desirable because the boiling points of 37,
3
8, and the by-products 39 and 40 are all very close to each
other and so purification of 38 by distillation is difficult and
costly.
We have found that our continuous supercritical method
combines quantitative conversion with high selectivity for
3
8. Furthermore, this is achieved without use of conven-
tional solvents, thereby eliminating the need for any work-
up of the product. Thus, we were able to hydrogenate 37 in
our 5-mL reactor at flow rates up to 2.0 mL/min at 120 bar
2
in scCO (0.75 L/min) over the 5% Pd APII Deloxan catalyst.
For reactor temperatures <200 °C, the conversion of 37 to
1
3
8 was quantitative with no 39 or 40 detectable by H NMR
Figure 10. Pictorial representation of the temperature varia-
tion of phase equilibrium for the system cyclohexane + CO
2
2
+
or GC. At higher reactor temperatures, the selectivity
dropped and 3,3,5-trimethylcyclohexanol, 39, and 1,1,3-
trimethylcyclohexane, 40, were also formed. With the 10-
mL reactor, the selectivity of hydrogenation could be
maintained at flow rates of 37 of e7 mL/min. With this
flow rate in the reactor, we were able to hydrogenate 7.5 kg
of 37 using only 2 g of 5% Pd on Deloxan, and even at the
end of the run the catalyst was giving 89% conversion of
H
0
using so-called “Gibbs triangles” (in steps of 30 °C between
and 270 °C). The dark areas represent homogeneous phases
(totally miscible) whereas the areas of immiscibility are left
white. The figures show the percentage of each triangle occupied
by the homogeneous phase. The system was modeled for 120
bar using the Peng-Robinson equation of state. Broadly, similar
results were obtained for the system cyclohexane + H +
2
propane.
3
7 to 38. These are much higher rates and selectivities than
Scheme 9. Hydrogenation of isophorone, 37
are usually obtained in the conventional hydrogenation of
3
1
7. Higher flow rates of 37 cause the temperature of the
0-mL reactor to exceed 200 °C. Using a 100-mL reactor,32
however, we have already achieved continuous hydrogena-
tion of 37 at 30 mL/min. It is interesting that, following
the principles of Figure 10, we have shown that 37 can be
hydrogenated successfully in our equipment using a mixture
which contained 70 mol % of 37 and only a small excess of
H . Thus, the process uses a relatively small amount of CO ,
2 2
a point of considerable economic benefit for future scale-
up.
2
when the concentration of H is reduced; (b) above 60 °C,
the regions of miscibility increase as the temperature
increases; and (c) at higher temperatures, relatively small
2
amounts of CO are sufficient to ensure miscibility.
The implications are straightforward. If a mixture is
homogeneous before the start of the reaction, it will remain
so because hydrogenation is an exothermic reaction which
Conclusion
2
consumes H . This conclusion will clearly not be true for
We have described a new approach to selective hydro-
genation. The combination of supercritical fluids and
continuous flow reactors permits a step change in the scale
and range of hydrogenations which can be carried out in the
nonspecialist laboratory. Even with a 5-mL reactor, we have
reached a throughput which is larger than that needed by
most synthetic organic laboratories. Depending on the
conditions, the hydrogenation can be mild and selectiVe as
in the hydrogenation of the CdC bond in isophorone, 37,
or it can be extreme as in the full hydrogenation of
all reactions; for example, many of the reactions in Table 1
generate H O (T ) 374 °C) as one of the products.
2 c
Nevertheless, the main point is almost certainly valid for
most systems; one of the principal roles of the fluid is to
2
render the substrate and H homogeneous as they enter the
3
0
reactor, and this homogeneity can often be achieved with
relatively low concentrations of fluid.
(g) Advantages of Supercritical Reaction. The Hy-
drogenation of Isophorone, 37. We now apply these ideas
to the hydrogentation of isophorone, 37, a functionalized
cyclohexene derivative of commercial interest in the fine
chemicals industry (Scheme 9). An important process is the
selective hydrogenation of the ring double bond which yields
(
31) See, for example: Cotrupe, D. P.; Wellman, W. E.; Burton, P. E. Patent
No. US 3446850, 1969; Patent No. FR 1549722, 1968. Bueschken, W.;
Hummel, J. Patent No. DE 19524969, 1997.
(
32) The 100-mL reactor was used at Thomas Swan & Co. Ltd. (Consett), with
a flow system similar to that described in this paper. Although scaling up
from a 5-mL to 10-mL reactor presented no problems, scaling up to a 100-
mL reactor necessitated the use of internal baffling to prevent an excessively
hot flow through the centre of the catalyst bed.
2
(30) In reactions where H O is generated, the product stream from the apparatus
usually consists of a mixture of aqueous and organic phases rather than an
emulsion.
Vol. 2, No. 3, 1998 / Organic Process Research & Development
•
145

nitrobenzene, 23, to cyclohexane, 9, and ammonia. Both
hydrogenations can be carried out on the same piece of
equipment, occupying the space of only half a fume hood.
All of the reactions described in this paper have involved
substrates and products which are liquids at room temper-
ature. This restriction is not chemical but primarily due to
the engineering problems of handling solid materials in a
flow system. Nevertheless we have shown that solids can
be hydrogenated in our apparatus by dissolving them in an
organic solvent (e.g., MeOH/THF). This enables the substrate
to be pumped into the system, and the product is recovered
as a solution. Although this approach is not as environ-
We are now actively pursuing scale-up. Already, our
maximum substrate flow rate with a 10-mL reactor is close
to that needed for the production of fine chemicals, and only
minor scale-up would be necessary for certain chemical
products of commercial interest. Furthermore, our reactor
can be applied to other organic reactions. Thus, we have
just shown that, with only minor modification, our flow
reactor can be used with a solid acid catalyst for supercritical
25
3
3
Friedel-Crafts alkylation of aromatics.
Acknowledgment
We thank Dr. A. Kordikowski for his help in modeling
the phase behavior of cyclohexane/CO /H . We thank
2
2
2
menally “green” as using pure CO , it works quite satisfac-
Thomas Swan & Co. Ltd. for fully funding the work at
Nottingham and Mr. D. Campbell and Mr. J. C. Toler for
their assistance. We are grateful to Degussa AG for donating
the catalysts. We thank Dr. M. W. George, Mr. M. Guyler,
Dr. S. M. Howdle, Dr. K.-H. Pickel, Mr. K. Stanley, Dr. T.
Tacke, and Dr. S. Wieland for their help. M.P. thanks EPSRC
and the Royal Academy of Engineering for fellowship
support, and F.R.S. thanks EPSRC for a studentship.
torily at least on the laboratory scale.
One of our most important findings is that high flow rates
of substrate do not necessarily require high flow rates of
CO
processes have proved costly to operate because large
quantities of CO are required to obtain even modest amounts
of product. The reasons why small amounts of CO are
2
. In the past, many apparently attractive supercritical
2
2
needed for these hydrogenations are clear from our inves-
tigation of the phase behavior. An additional attraction of
this technique is the possibility of recycling the supercritical
fluid. Although recycling has been widely applied to
Received for review October 29, 1997.
OP970056M
10
supercritical processes, we have not yet attempted to apply
it to our hydrogenation reactor.
(33) Hitzler, M. G.; Smail, F. R.; Ross, S. K.; Poliakoff, M. Chem. Commun.
1998, 359.
146
•
Vol. 2, No. 3, 1998 / Organic Process Research & Development