"Hydrogen-catalyzed" dehydrogenation: A supercritical conundrum

Article abstract of DOI:10.1002/anie.200502049

Hot spots: The surprising, apparently catalytic effect of H₂ in the dehydrogenation of 1 in supercritical CO₂ is the result of thermal hot spots (see thermal image of reactor) generated in the catalyst bed by an initial exothermic hy

Full text of DOI:10.1002/anie.200502049

Communications
Heterogeneous Catalysis
feedstocks. In batch reactors,this appears to be an overall
disproportionation reaction,producing two mole of B for
each mole of C. So,a surprising aspect of the reported
DOI: 10.1002/anie.200502049
dehydrogenation of V in scCO was that C was not observed,
2
“
Hydrogen-Catalyzed” Dehydrogenation:
making a simple disproportionation unlikely. The results
presented herein provide an explanation of this conundrum
and include the identification of two previously unsuspected
intermediates, 1 and 2 in Scheme 1.
A Supercritical Conundrum**
Jason R. Hyde,* Ben Walsh, and Martyn Poliakoff*
Initially,we investigated the effect of passing V+ scCO2
through the reactor in the absence of H . Our automated
2
[
16]
In recent years there has been increasing interest in organic
reactor (Figure 1) was therefore programmed to ramp the
reactions in supercritical carbon dioxide (scCO ). These
2
1–4]
[
reactions have included Friedel–Crafts alkylations,
hydro-
and espe-
Apart from the environmental
[
5–9]
[10–15]
[16–18]
formylation,
cially hydrogenation.
oxidation,
etherification,
[
19–24]
acceptability,the attractions of scCO 2 for hydrogenation are
improved mass transport,ease of continuous operation,and
[
25]
potentially improved selectivity.
Much of this work has
involved continuous reactors,which simplify scale-up. How-
ever,we show herein,that these reactors can produce
apparently misleading results which,when correctly inter-
preted,can open up new routes to compounds which are
difficult to synthesize otherwise.
The starting point for this paper
[
26]
was a report
of the attempted
selective hydrogenation of 4-vinylcy-
clohex-1-ene (V; Scheme 1),in a
small-scale (5 mL) continuous reac-
tor. Unexpectedly,this approach
resulted in the formation of ethyl
benzene (B; which is a dehydrogen-
ation product) even in the presence of
Figure 1. Schematic diagram of the automated apparatus. The organic
substrates and CO are delivered by HPLC pumps (Jasco PU–980 and
2
CO –1580, respectively); system pressure in the reactor (R) is main-
2
tained by a Jasco Back Pressure Regulator (BPR) which also facilitates
product separation. Samples of product are collected in-line by a
sample loop (L; Valco/Vici) and injected into a gas chromatograph
(GC) and the results are downloaded automatically to the computer.
a large excess of H and at temper-
2
The high-pressure lines are designated by heavy filled arrowed lines,
whilst data lines and the direction of data are depicted by light unfilled
arrowed lines. T1–6 are optional thermocouples placed within the
reactor.
atures apparently as low as 408C. This
report was significant firstly because
the dehydrogenation of V is industri-
Scheme 1. Structures of
ethyl cyclohexane (C)
and ethyl benzene (B)
[
27–31]
formed during the dehy- ally important
drogenation of 1-vinyl
and secondly
because the reactor design is one
which is very widely used for super-
critical chemistry.
cyclohex-4-ene (V); 1
and 2 are the intermedi-
ates detected in the
course of this reaction
and were identified by
NMR spectroscopy (see
Supporting Informa-
tion).
temperature of the catalyst bed (Pd 2%) from 808C to 2008C
over a period of 8 h at a constant pressure of 150 bar,
analyzing the products by GLC every 5 minutes. Figure 2,
The dehydrogenation of cyclohex-
yl compounds to the corresponding
benzene derivatives is an attractive
route to aromatic compounds from
cheap,plentiful biologically derived
shows that 1) in the absence of H ,a reaction does occur; 2)
2
this reaction yields both B and C,unlike the reaction in the
presence of H where C is not observed; 3) the ratio of B to C
2
was found to be 2:1 close to the ratio expected for
[
27–31]
disproportionation;
1
4) there are two additional products,
[
32]
and 2; 5) the reaction displays “light-off” (i.e. there is an
[
*] Dr. J. R. Hyde, Dr. B. Walsh, Prof. M. Poliakoff
ignition over a small temperature interval of only 2–38C;
similar behavior was also observed at 100,200,and 250 bar)
and 6) the reaction switches-off equally suddenly.
Analysis of the product stream by off-line GC/MS showed
that 2 had a parent ion with m/z 108,identical to that of V. By
contrast 1,had m/z 110,consistent with a partially hydro-
genated product. Further experiments,detailed in the ESI,
provided sufficient NMR spectroscopic data to identify the
structures of 1 and 2 (see Scheme 1).
School of Chemistry
University of Nottingham
University Park, NG72RD (UK)
Fax: (+44)115-913-058
E-mail: jason.hyde@nottingham.ac.uk
martyn.poliakoff@nottingham.ac.uk
[
**] We thank the EPSRC for Grant GR/S87409 and for a CASE
studentship to B.W. and Thomas Swan & Co. Ltd for support. We
thank Dr. Pete Licence, Dr. Stephen Ross, and Messrs Mark Guyler,
Peter Fields, Rich Wilson, David Litchfield, and James Warren for
their help and support.
Estimating the residence times in these reactors is
[
33]
straightforward but slightly convoluted.
However,once
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
the relationship between flow rate and residence time has
7588
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2005, 44, 7588 –7591

Angewandte
Chemie
by analyzing the GC data with the new analytical technique,
[
1,34]
2
DCOR-GC,
correlation function to chromatographic data.
DCOR-GC maps shown in Figure 4,are calculated from
which is an extension of the generalized
[
34]
The
2
Figure 2. Plot of the product distribution observed whilst the catalyst
bed is heated (bold line) between 140 and 1808C. Note the arrowed
auto-thermal ignition point at around 1508C, and the presence of
Figure 4. 2D-correlation analysis of the complete GC data from the
experiment illustrated in Figure 3. This analysis is mathematically
À1
À1
intermediates 1 and 2. Flow rates: CO 1 mLmin , V 0.2 mLmin
.
[38]
2
similar to the 2DCOS methodology for FTIR. The principle of
The product composition was sampled every 15 minutes over the 10-
hour period.
analysis for GC and FTIR are the same. The sequence in which the
various compounds occur in response to a perturbation (in this case
residence time) is obtained by comparing the signs (white positive,
black negative) of the off-diagonal peaks in the synchronous and
[
38]
asynchronous plots and applying Noda’s Rules. a) Synchronous map
showing the simultaneous relationships between products; b) asyn-
chronous map showing the “out-of-phase” relationships. In this plot,
further products, A, X, and Y, have been highlighted. High-resolution
color copies are available in the Supporting Information.
been established,it is relatively simple to investigate the
variation of reaction products with residence time. Figure 3
shows that 1) reaction products are detected at residence
times longer than 2 s; 2) the yields of B and C grow steadily
with increasing residence time,confirming that B and C are
final products; 3) the yields of 1 and 2 reach maxima and then
decrease,suggesting that 1 and 2 are intermediates,and 4) the
initial rates of formation of 2 and B are similar,suggesting
that 2 is not an intermediate in the formation of B.
the same experiment that gave the data in Figure 3. The off-
diagonal peaks in Figure 4a show that the formations of all of
the products are correlated to V but not necessarily to each
other,reflecting the existence of two parallel reaction path-
ways,namely hydrogenation and dehydrogenation. Together,
the two maps can be used to deduce the sequence of
formation of B, C, 1,and 2,see Scheme 2. The 2DCOR-GC
maps provide evidence for two additional intermediates, X
and Y,which are difficult to discern in the 1D chromatograms
but are easily seen in the maps. X and Y,correlate with 2 but
are formed before it. Although X and Y were not identified
spectroscopically,a simple GC structural–elution simula-
However,additional evidence for the relative positions of
the different species in the reaction scheme can be obtained
[
35]
tion suggests that they may have the structures shown in
Scheme 2. An additional weak peak, A in Figure 4,was
Figure 3. Plot of the product distribution for the dehydrogenation of V
in the absence of H as the residence time is varied. The traces are
2
labeled 1, 2, V, C, B as in Scheme 1. In principle, such data can be
used to model the kinetics of the reaction and hence to extract rate
constants for the different stages. However, we are limited in this case
Scheme 2. Postulated mechanism for the dehydrogenation/hydrogena-
tion of V to form B and C. Compounds 1 and 2, have been identified
by NMR spectroscopy, and X and Y have been tentatively identified by
elution–structure relationships.
by the fact that H is a key component in the reaction but its
2
concentration is unknown and indeed, will vary at different points in
the reactor.
Angew. Chem. Int. Ed. 2005, 44, 7588 –7591ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
7589

Communications
identified as vinyl cyclohexane,by comparison with an
The apparent catalytic effect of H in the dehydrogenation
2
authentic sample. Compound A does not correlate synchro-
nously with any other species and is probably the product of a
side reaction.
V is the result of thermal hot spots generated in the catalyst
bed by an initial exothermic hydrogenation “light-off” forcing
the dehydrogenation reaction. The important point is that the
possible presence of such hot spots in small-scale reactors
should not be overlooked in other exothermic reactions in
fixed bed reactors. Clearly it will be worth investigating the
effects of different supercritical solvents on hot spots since
different fluids can have significantly different thermal
properties. In addition our study has resulted in the isolation
of the diene, 2 by a potentially much simpler route than that
We now return to the original conundrum,namely why H 2
apparently catalyzes the dehydrogenation of V at temper-
[
26]
atures close to ambient. Figure 5 shows thermal images of
[
37]
previously reported.
Many of our experiments would not have been possible on
manually operated equipment because it was necessary to
collect large amounts of data over a whole series of different
reaction conditions,a process which would be impractically
long without automation. However,handling such large
volumes of data in itself presents problems and herein we
have demonstrated again the value of 2DCOR-GC for
presenting these data in a more easily interpretable form.
Overall,one of the more important lessons from this work is
that even apparently simple catalytic reactions are in reality
quite complex,but that current techniques for supercritical
chemistry are now capable of unraveling these complexities.
Figure 5. Evidence for hot spots during the dehydrogenation of V in
the presence of H . a) photograph of the reactor; b) and c) thermal
2
images of the same reactor with b) the pre-heater set to 408C, and
CO and V both flowing through it. Note the maximum temperature is
2
3
18C and c) with H flowing and all other conditions unchanged. Note
2
that the reactor skin has now reached nearly 658C. A thermocouple
place at the center of the reactor reads over 2008C. The exit temper-
ature remained close to ambient. High-resolution color copies are
available in the Supporting Information.
Experimental Section
Our automated continuous supercritical apparatus has recently been
[
16]
described in detail elsewhere. Figure 1 shows a schematic of the
design. Computer controlled HPLC pumps (Jasco CO –1580 and
2
Jasco PU–980) deliver the CO and organic substrate. The preheated
2
CO /organic mixture is passed over the catalyst bed (2% Pd/SiO /
2
2
Al O supported: Johnson Matthey),which is heated by cartridge
2
3
the reactor,which reveal that H 2 causes a highly exothermic
reaction and the generation of a distinct hot spot. A separate
experiment with an internal thermocouple,deliberately
located in the region of the hot spot,indicated that the
temperature of the catalyst bed rose to approximately 2008C.
A further experiment with a series of six thermocouples
positioned at regular intervals down the length of an 8 cm
reactor showed that over 24 h the hot spot gradually moves
down the catalyst bed; presumably the catalyst gradually
heaters and aluminum blocks. The temperature of the catalyst bed is
controlled by a programmable Eurotherm. After the reaction has
occurred and left the catalyst bed,the mixture of products and gases
are passed through a HPLC 4-port sample loop (Valco/Vici),which
samples 0.06 mL periodically into a GLC (Shimadzu GC17–A v.4),
whereupon product separation and analysis is carried out. The system
pressure and product separation is facilitated by a programmable
Back Pressure Regulator (Jasco BPR BP-1580-81). Data analysis is
performed off-line after the experimental is completed.
[
36]
becomes deactivated through charring or coking. Thus,the
Received: June 14,2005
Revised: August 23,2005
Published online: October 31,2005
explanation of the “catalysis” by H is that hydrogenation of
2
V rapidly heats a thin layer at the top of the catalyst bed to a
temperature well above the light-off temperature for dehy-
drogenation of V. Since the dehydrogenation is itself exo-
thermic,the reaction becomes almost self-sustaining and
therefore can continue with almost 100% yield of B without
the formation of C.
Keywords: carbon dioxide · dehydrogenation · heterogeneous
.
catalysis · reaction mechanisms · supercritical fluids
It is possible to suppress the dehydrogenation completely
by deliberately cooling the reactor to 108C,although the
catalyst bed itself is presumably at a higher temperature. This
approach was used to generate sufficient quantities of 1 and 2
for their identification by NMR spectroscopy (see Supporting
Information). Note that this low-temperature hydrogenation
of V is a much simpler route to 2 than the original synthesis by
[
1] J. R. Hyde,M. Poliakoff,R. Amandi,S. K. Ross,T. J. Lotz, Green
Chem. 2005, 7,288.
[2] J. M. Simard,M. S. Erickson, Abstr. Pap. Am. Chem. Soc. 1999,
217,271.
[
3] M. G. Hitzler,F. R. Smail,S. K. Ross,M. Poliakoff,
Chem.
Commun. 1998,359.
[
4] J. E. Chateauneuf,K. Nie, Abstr. Pap. Am. Chem. Soc. 2000, 219,
46.
4
[
37]
Trost and co-workers,
which involved a Diels–Alder or
[
[
5] W. Leitner, Pure Appl. Chem. 2004, 76,635.
6] A. R. Tadd,A. Marteel,M. R. Mason,J. A. Davies,M. A.
Abraham, J. Supercrit. Fluids 2003, 25,183.
cycloaddition reaction,and thus opens up the possibility of
manufacturing this compound on a larger-scale if required.
7590 www.angewandte.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2005, 44, 7588 –7591

Angewandte
Chemie
[
[
[
7] A. Marteel,J. A. Davies,M. R. Mason,T. Tack,S. Bektesevic,
M. A. Abraham, Catal. Commun. 2003, 4,309.
8] A. E. Marteel,T. T. Tack,S. Bektesevic,J. A. Davies,M. R.
Mason,M. A. Abraham, Environ. Sci. Technol. 2003, 37,5424.
9] N. J. Meehan,A. J. Sandee,J. N. H. Reek,P. C. J. Kamer,P.
van Leeuwen,M. Poliakoff, Chem. Commun. 2000,1497.
[
10] M. Caravati,J. D. Grunwaldt,A. Baiker, Catal. Today, 2004, 91–
2,1.
11] T. D. Crowe,P. J. White, J. Am. Oil Chem. Soc. 2003, 80,569.
9
[
[
12] M. L. Corazza,L. Cardozo,O. A. C. Antunes,C. Dariva,
Eng. Chem. Res. 2003, 42,3150.
Ind.
[
[
[
13] J. Liu,R. Zhang,L. Gao,T. Jiang,Z. M. Liu,J. He,B. X. Han,
G. Y. Yang, Chin. Chem. Lett. 2002, 13,1162.
14] G. T. Musie,M. Wei,B. Subramaniam,D. H. Busch,
Inorg.
Chem. 2001, 40,3336.
15] G. Jenzer,M. S. Schneider,R. Wandeler,T. Mallat,A. Baiker, J.
Catal. 2001, 199,141.
[
[
16] J. R. Hyde,M. Poliakoff,B. Walsh, Green Chem. 2005, 7,357.
17] W. K. Gray,F. R. Smail,M. G. Hitzler,S. K. Ross,M. Poliakoff, J.
Am. Chem. Soc. 1999, 121,10711.
[
[
[
18] P. Licence,W. K. Gray,M. Sokolova, J. Am. Chem. Soc. 2005,
1
27,293.
19] S. Ichikawa,M. Tada,Y. Iwasawea,T. Ikariya, Chem. Commun.
005,924.
20] P. Stephenson,P. Licence,S. K. Ross,M. Poliakoff, Green Chem.
004, 6,521.
2
2
[
[
21] F. Zhao,Y. Ikushima,M. Arai, J. Catal. 2004, 224,479.
22] F. Y. Zhao,S. Fujita,J. M. Sun,Y. Ikushima,M. Arai,
Today 2004, 98,523.
Catal.
[
[
[
[
23] T. Sato,C. V. Rode,O. Sato,M. Shirai, Appl. Catal. B 2004, 49,
81.
24] N. Hiyoshi,C. V. Rode,O. Sato,M. Shirai, J. Jpn. Pet. Inst. 2004,
7,410.
25] J. R. Hyde,P. Licence,D. Carter,M. Poliakoff, Appl. Catal. A
001, 222,119.
1
4
2
26] S. K. Ross,F. R. Smail,M. Sellin,R. Amandi,M. Poliakoff,
Chem. Eng. Trans. 2002, 2,561.
[
[
[
[
27] S. Clark,US Patent 5321180, 1994.
28] R. W. Rieve,H. Shalit, US Patent 4029715, 1977.
29] G. Ruckelshauss,US Patent 4048243, 1997.
30] H. H. Vogel,H. M. Weitz,E. Lorez,R. Platz,US Patent
3
903185, 1975.
[
[
31] D. W. Hart,L. H. Slaugh, US Patent 4375571, 1983.
32] O. Levenspiel,1st ed. , Chemical Reaction Engineering, 2nd Ed.,
Wiley,New York, 1962,230.
[
33] The residence time was estimated by calculating the velocity of
the fluid passing through the reactor per unit volume. Fluid
À1
velocity (ms ) was calculated at experimental temperature and
pressure,and therefore requires specific data,which was
obtained from appropriate equations of state,in this case from
the NIST database (http://www.nist.gov).
[
[
[
[
[
34] J. R. Hyde,R. A. Bourne,I. Noda,P. Stephenson,M. Poliakoff,
Anal. Chem. 2004, 76,6197.
35] Advanced Chemistry Development,Inc.,6.08,Toronto ON,
Canada,http://www.acdlabs.com, 2002.
36] J. H. Kim,D. J. Suh,T. J. Park,K. L. Kim, Appl. Catal. A 2000,
1
97,191.
37] B. Trost,T. Runge,L. Jungheim, J. Am. Chem. Soc. 1980, 102,
841.
38] I. Noda, Appl. Spectrosc. 1993, 47,1329.
2
Angew. Chem. Int. Ed. 2005, 44, 7588 –7591ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
7591