The continuous catalytic debenzylation of 1,4-dibenzyloxybenzene with H2 in THF expanded with high pressure CO2

Article abstract of DOI:10.1021/op100136g

The continuous debenzylation of 1,4-dibenzyloxybenzene (1a) with H ₂ and supported Pd catalysts has been studied in high-pressure CO₂ + tetrahydrofuran. A range of parameters and catalysts have been tested, with one of the most successful results giving 86% conversion and 86% selectivity to the monodebenzylated product, 1,4-(benzyloxy)phenol (2a). In the absence of CO₂, the reaction was unselective with no formation of 2a. When complete debenzylation was required, the best catalyst was 5% Pd supported on Deloxan APII, giving quantitative yields of hydroquinone (3), the completely debenzylated product, at temperatures as low as 50 °C. This Deloxan-supported catalyst was also substantially active at only 30 °C.

Full text of DOI:10.1021/op100136g

Organic Process Research & Development 2010, 14, 1202–1208
The Continuous Catalytic Debenzylation of 1,4-Dibenzyloxybenzene with H2 in THF
Expanded with High Pressure CO2
Geoffrey R. Akien,†,‡ Jean-Christophe Legeay,†,§ Andrew Wells,⊥ and Martyn Poliakoff*,†
School of Chemistry, The UniVersity of Nottingham, UniVersity Park, Nottingham, United Kingdom NG7 2RD, and
AstraZeneca, Chemical Sciences, Loughborough, United Kingdom LE11 5RH
Abstract:
preparative supercritical fluid chromatography (SFC) versus
preparative HPLC,4 the overall energy cost for SFC was larger
than for HPLC, precisely because of the high energy costs of
cooling and heating CO2 to recycle it. When larger quantities
of organic modifiers are used, the fluid may then be termed a
gas-expanded liquid5,6 (GXL), which has improved solvent
power compared to modified scCO2, but still reduces the volume
of organic solvent required.3
One barrier to the application of GXLs in complex organic
reactions is the paucity of research on pharmaceutically relevant
molecules.6 The most obvious examples in this area include
the complexation of indomethacin with copper acetate,7 the
hydrogenation of 2-(6′-methoxy-2′-naphthyl)propenoic acid to
form (S)-naproxen,8,9 the selective hydrogenation of rac-
sertraline imine,10 and the asymmetric hydrogenation of methyl-
(Z)-R-acetamidocinnamate.11 In an attempt to rectify this
situation, this paper describes work on the O-debenzylation of
a model compound, 1,4-dibenzyloxybenzene (1a). Benzyl
groups are frequently used as protecting groups for hydroxyl
and amine moieties.12-20 Like many compounds of interest as
pharmaceutical intermediates, 1a has a relatively low solubility
in common organic solvents, and it is also symmetric, making
selective removal of only one of the benzyl groups a challenge.
Simplistically, the overall reaction probably takes place in two
The continuous debenzylation of 1,4-dibenzyloxybenzene (1a) with
H2 and supported Pd catalysts has been studied in high-
pressure CO2 + tetrahydrofuran. A range of parameters
and catalysts have been tested, with one of the most
successful results giving 86% conversion and 86% selectivity
to the monodebenzylated product, 1,4-(benzyloxy)phenol
(2a). In the absence of CO2, the reaction was unselective
with no formation of 2a. When complete debenzylation was
required, the best catalyst was 5% Pd supported on Deloxan
APII, giving quantitative yields of hydroquinone (3), the
completely debenzylated product, at temperatures as low
as 50 °C. This Deloxan-supported catalyst was also sub-
stantially active at only 30 °C.
Introduction
Continually increasing consumer and regulatory pressures
on the pharmaceutical industry are strong drivers for reduced
use of organic solvents throughout the manufacturing chain.1
One potential strategy is to replace some of the organic solvents
with supercritical fluids (SCFs),2 typically supercritical CO2
(scCO2). CO2 is attractive because of its low cost, low reactivity,
and relatively accessible critical conditions (31.1 °C and 73.8
bar). Unfortunately, scCO2 has a low polarity so is unsuitable
for dissolving the majority of pharmaceutical molecules of
interest. Indeed, in some cases these molecules are barely
soluble in the majority of common organic solvents. Adding
small quantities of ‘modifiers’, organic solvent (typically <10
mol %), to the scCO2 improves the solvent power of the fluid
significantly, thereby rendering the scCO2 more useful for a
wider range of organic intermediates, while still retaining some
of the benefits of reduced organic solvent use.3 However, it is
important to acknowledge that there is a significant energy cost
in compressing the CO2. In a recent life-cycle analysis of
(4) Van der Vorst, G.; Van Langenhove, H.; De Paep, F.; Aelterman,
W.; Dingenen, J.; Dewulf, J. Green Chem. 2009, 11, 1007–1012.
(5) Jessop, P. G.; Subramaniam, B. Chem. ReV. 2007, 107, 2666–2694.
(6) Akien, G. R.; Poliakoff, M. Green Chem. 2009, 11, 1083–1100.
(7) Warwick, B.; Dehghani, F.; Foster, N. R.; Biffin, J. R.; Regtop, H. L.
Ind. Eng. Chem. Res. 2000, 39, 4571–4579.
(8) Combes, G. B.; Dehghani, F.; Lucien, F. P.; Dillow, A. K.; Foster,
N. R. Asymmetric Catalytic Hydrogenation in CO2 Expanded Metha-
nol: An Application of Gas Anti-Solvent Reactions (GASR). In
Reaction Engineering for Pollution PreVention; Abraham, M. A.,
Hesketh, R. P., Eds.; Elsevier: Amsterdam, 2000; pp 173-181.
(9) Combes, G.; Coen, E.; Dehghani, F.; Foster, N. J. Supercrit. Fluids
2005, 36, 127–136.
(10) Clark, P.; Poliakoff, M.; Wells, A. AdV. Synth. Catal. 2007, 349, 2655–
2659.
(11) Shariati, A.; Sheldon, R. A.; Witkamp, G. J.; Peters, C. J. Green Chem.
2008, 10, 350–354.
* Author for correspondence. E-mail: martyn.poliakoff@nottingham.ac.uk.
† The University of Nottingham.
(12) Baltzly, R.; Buck, J. S. J. Am. Chem. Soc. 1943, 65, 1984–1992.
(13) Baltzly, R.; Russell, P. B. J. Am. Chem. Soc. 1950, 72, 3410–3413.
(14) Baltzly, R.; Russell, P. B. J. Am. Chem. Soc. 1953, 75, 5598–5602.
(15) Baltzly, R.; Russell, P. B. J. Am. Chem. Soc. 1954, 76, 5776–5777.
(16) Sajiki, H.; Kuno, H.; Hirota, K. Tetrahedron Lett. 1997, 38, 399–
402.
‡ Present address: Department of Biology and Chemistry, City University of
Hong Kong, 83 Tat Chee Avenue, Kowloon, Hong Kong SAR.
§ Present address: Transformations Inte´gre´es de la Matie`re Renouvelable,
ESCOM-UTC, 1-Alle´e du Jean-Marie Buckmaster, 60200 Compie`gne, France.
⊥ AstraZeneca, Chemical Sciences, Loughborough, U.K. LE11 5RH.
(1) Constable, D. J. C.; Dunn, P. J.; Hayler, J. D.; Humphrey, G. R.;
Leazer, J. L.; Linderman, R. J.; Lorenz, K.; Manley, J.; Pearlman,
B. A.; Wells, A.; Zaks, A.; Zhang, T. Y. Green Chem. 2007, 9, 411–
420.
(17) Blaser, H. U.; Indolese, A.; Schnyder, A.; Steiner, H.; Studer, M. J.
Mol. Catal. A: Chem. 2001, 173, 3–18.
(18) Clayden, J.; Greeves, N.; Warren, S.; Wothers, S., Organic Chemistry,
7th ed.; Oxford University Press: Oxford, 2001.
(2) Handbook of Green Chemistry, Volume 4: Supercritical SolVents;
Leitner, W., Jessop, P. G., Eds.; Wiley-VCH: Weinheim, 2010.
(3) Wei, M.; Musie, G. T.; Busch, D. H.; Subramaniam, B. J. Am. Chem.
Soc. 2002, 124, 2513–2517.
(19) Dugger, R. W.; Ragan, J. A.; Ripin, D. H. B. Org. Process Res. DeV.
2005, 9, 253–258.
(20) Carey, J. S.; Laffan, D.; Thomson, C.; Williams, M. T. Org. Biomol.
Chem. 2006, 4, 2337–2347.
1202
•
Vol. 14, No. 5, 2010 / Organic Process Research & Development
10.1021/op100136g  2010 American Chemical Society
Published on Web 07/19/2010

Scheme 1. Reaction scheme for the debenzylation of a
dibenzyloxybenzene (1) with H2 using a Pd catalyst, to form
the corresponding (benzyloxy)phenol (2) and the
side-product toluene (4); 2 can then be subsequently
debenzylated to form the hydroquinone (3a) or its isomers,
catechol and resorcinola
entific), helium (He) (99.999%, BOC Gases), hydroquinone (2a,
USP grade, Sigma), m-(benzyloxy)phenol (2b, Wako), m-
dibenzyloxybenzene (1b, 98%, Alfa Aesar), methylcyclohexane
(>99.5%, Fluka), 1,4-benzoquinone (98+%. Alfa Aesar), 1,4-
(benzyloxy)phenol (2a, 99+%, Aldrich), 1,4-dibenzyloxyben-
zene (1a, 98%, Alfa Aesar), ethyl acetate (laboratory reagent
grade, Fisher Scientific), magnesium sulfate (anhydrous, Fisher
Scientific), 4-methoxybenzyl chloride (98%, Sigma-Aldrich),
nitrogen (g99.99%, BOC Gases), 2% Pd/C (charcoal, pellets,
Johnson Matthey, batch number: M02161), 2% Pd/Al2O3 (type
335 powder, Johnson Matthey, batch number: M05226), 2%
Pd/SiO2 (egg-shell type 0.8 mm spheres, Johnson Matthey,
batch number: 99046), 2% Pd/SiO2-Al2O3 (type 31, Johnson
Matthey, batch number: 113031054), 5% Pd/Al2O3 (type
5R335, Johnson Matthey, batch number: M05227), 5% Pd/
CaCO3 (type 21 powder, Johnson Matthey, batch number:
DLR0320), petroleum ether (40-60 °C, laboratory reagent
grade, Fisher Scientific), potassium carbonate (laboratory
reagent grade, Fisher Scientific), resorcinol (3b, ACS 99+%,
Alfa Aesar), THF (laboratory reagent grade 99.5%, Fisher
Scientific), toluene (3, laboratory reagent grade, Fisher Scien-
tific), 4-(trifluoromethoxy)benzyl bromide (>99%, Fisher Sci-
entific). Deionised water was supplied from an Elga PureLab
S water purification system.
a The suffix describes various different analogues: a) para-, R ) H; b) meta-;
R ) H, c) para-, R ) OMe; d) para-, R ) OCF3. The other by-products observed
in this work are those where the aromatic core has been fully or partially
hydrogenated.
steps, as shown in Scheme 1, with the sequential removal of
the benzyl groups.
The pellets of 2% Pd/C and 2% Pd/SiO2 catalysts were too
large for the reactor; thus, they were crushed, and only the
63-125 µm sieved particles were used. The rest of the catalysts
(with the exception of 5% Pd/CaCO3. which was much finer)
were mainly supplied in the 125-250 µm size range.
The hydrogenation of 1a to yield selectively the intermediate
1,4-(benzyloxy)phenol (2a) was investigated by Sajiki et al.
some years ago.16 Using 5% Pd/C and H2 at atmospheric
pressure, the terminal product, hydroquinone (3a), was obtained
in quantitative yields.16 When 0.5 equiv of pyridine were added,
the partially poisoned catalyst then only produced a 17:83 ratio
of 2a:3a (the yield was not specified). Others have used different
hydrogen sources: a 90% yield of 2a was possible with tetralin
and 10% Pd/C in ethanol,21 but only 46.5% using cyclohexene
and 5% Pd/C.22 The presence of both reacted and unreacted
transfer hydrogenation sources may complicate later separation
of the products from the reaction mixture, so we thought it
worthwhile to investigate the potentially greener direct hydro-
genation of 1a instead.
The details of the synthesis of 1c and 1d are included in
Supporting Information.
Reaction Analysis. Analysis of the reaction mixtures was
carried out with a Shimadzu GC-2010 equipped with an AOC-
20s autosampler, and fitted with a Restek RTX-5 column (10
m × 0.10 mm internal diameter, 0.1 µm film thickness) using
He as the carrier gas, with detection using a flame ionisation
detector (FID). The method used for the separation of the
reaction mixture from the debenzylation of 1 was as follows:
0.2 µL injection volume, 250 °C inlet temperature, 50:1 split
ratio, temperature program 45 °C for 5 min, increasing to 320
°C at a rate of 50 °C·min-1, held for a further 2 min, 350 °C
FID temperature, samples dissolved in THF. Products were
identified by comparison with authentic samples and by
GC-MS (Thermo Finnegan TraceGC 2000 fitted with a
PolarisQ MS detector). Mass balance experiments showed
quantitative recovery of the substrates, so quantitative analysis
was performed with the aid of external standards.
Reactions. Safety Note. Flow reactors haVe a comparatiVely
small Volume under pressure. NeVertheless, equipment with the
appropriate pressure and temperature rating should always be
used for high-pressure experiments. The continuous flow
equipment was essentially identical to that used previously for
the hydrogenation of sertraline imine,10 apart from the mixer
being replaced by another tube identical to the reactor, but filled
with 1-1.5 mm glass beads and heated to the same temperature
as the reactor. Fresh catalyst was used for each experiment.
1a is largely insoluble in most solvents, apart from THF,
dichloromethane, and chloroform. Earlier work has shown THF
Thus, this paper presents a study on the catalytic reaction
of 1 with H2 in a GXL, with the aim of establishing how much
the selectivity of the reaction can controlled while achieving
as high as possible a yield of 2 using commercially available
heterogeneous catalysts.
Experimental Section
Materials. The following materials were used without
further purification: acetone (laboratory reagent grade, Fisher
Scientific), air (99.9%, BOC Gases), cesium carbonate (99.5%,
Fisher Scientific), chloroform (laboratory reagent grade, Fisher
Scientific), chloroform-d (99.8% D, Sigma-Aldrich), CO2 (SFC
grade, 99.999%, BOC Gases), 1,4-cyclohexanediol (99%,
Lancaster), 1,4-cyclohexanedione (>99%, Fluka), dichlo-
romethane (laboratory reagent grade, Fisher Scientific), N,N-
dimethylformamide (anhydrous, 99.8%, Sigma-Aldrich), H2
(99.995%, BOC Gases), HCl (∼12.0 mol·dm-3, Fisher Sci-
(21) Li, G.; Wang, P.; Zhao, W.; Chen, J. Zhejiang Yike Daxue Xuebao
1987, 16, 154–156.
(22) Li, L.; Pang, Q. Huaxi Yauxue Zazhi 1996, 11, 234–236.
Vol. 14, No. 5, 2010 / Organic Process Research & Development
•
1203

to be a good solvent for N-debenzylations;23 thus, it was selected
for this investigation, using a 1 concentration of 0.05 mol ·dm-3
(1.93 wt %). THF is also capable of dissolving large quantities
of CO2,27 giving an excellent opportunity to tune the reaction.
The standard conditions for each experiment were, unless
otherwise stated, flow 0.5 mL ·min-1 THF solution, 0.5
mL ·min-1 CO2, 100 °C, 100 bar, 5.5 equiv of H2, with ∼300
mg of 2% Pd/SiO2-Al2O3 catalyst, prereduced under N2/H2 at
100 °C and 100 bar for 2 h. For experiments where parameters
were varied, each value was held for at least 30 min, and
samples were collected every 5 min. The first sample at each
new value was discarded, and the mean and standard deviation
(shown with error bars) were calculated from the remaining
samples. For the experiments where the conditions were fixed,
samples were collected every 10 min, and no averaging was
applied to the data. Experiments varying the residence time (not
shown) indicated that the reaction was kinetically controlled.
Results and Discussion
Figure 1. Debenzylation of 1a over 2% Pd/SiO2-Al2O3; the
effects of increasing the temperature from 80 to 120 °C in 5 °C
increments at 100 bar pressure. For other conditions, see the
Experimental Section. For simplicity, the plots in all figures in
this paper are labeled in the same way, namely, conversion of
1 (b), selectivity for 2 (0), selectivity for 3 (∆), selectivity for
byproducts (3).
Overall Strategy. As explained above, our aim has been to
achieve a yield of 2 as high as possible using H2 as the hydrogen
source with commercially available heterogeneous catalysts.
Originally, we intended to study a wide range of compounds.
However, there was little difference between the debenzylation
of the para (1a) and meta (1b) isomers of 1, although 1b
produced more byproducts in the first 45 min and the terminal
product, resorcinol (3b), was not detected at all. By comparison
∼5 mol % of 3a was detected from the debenzylation of the
para isomer. Overall however, the differences were still small,
and consistent with previous observations24 that, for debenzy-
lations, electronic effects are more important than steric ones.
By changing the substituents on the benzyl group, the
electronic properties can be changed, possibly changing the
selectivities. To this end, the 1,4-methoxybenzyl (1c) and 1,4-
trifluoromethoxybenzyl (1d) analogues of 1a were synthesised
(see Supporting Information). Unfortunately, 1c was found to
be too insoluble to be reacted in the equipment used here. On
the other hand, 1d was noticeably more soluble than 1a, but
did not react until 200 °C, producing only small quantities of
unidentified products with GC retention times approximately a
minute later than the starting material.
important parameters for controlling conversion and selectivity
in a continuous reaction in scCO2. Therefore, the effects of
temperature were investigated first.
Figure 1 shows that increasing the temperature increases the
conversion, but it also reduces the selectivity for 2a by
promoting further debenzylation of 2a to form the doubly
debenzylated product hydroquinone, 3a. Under these conditions,
a phase transition from a single liquid phase to a liquid-vapor
biphase is expected27 at approximately 100 °C. A phase
transition is likely to have an effect on the residence time in
the reactor with a concomitant change in the conversion and
selectivity. Thus, one might expect a discontinuity in the plots
at 100 °C. However, Figure 1 shows that, surprisingly, this
transition does not have any significant effect; there is no sudden
change in conversion and selectivity at this temperature.
Effect of Changing the Ratio H2:1a. In an attempt to
improve the selectivity for 2a, the amount of H2 available for
the reaction was varied by changing the ratio of H2:1a by a
factor of 10, while maintaining the temperature and pressure
constant.
Figure 2 shows that, at 110 °C, increasing the ratio of H2:
1a from 1.45 to 14.5 increases the conversion of 1a as might
be expected, but also decreases the selectivity for 2a. As with
temperature, there is a clear trade-off between high conversion
and high selectivity. At higher H2:substrate ratios, there is a
notable leveling off; further increases in the amount of H2 have
little effect, suggesting that mass-transport limitations were
overcome under these conditions.
Therefore, our study has focused on the reactions of the para
isomer (1a). The first stage of the investigation focused on using
a single catalyst, 2% Pd/SiO2-Al2O3, and on identifying the
effects of individual parameters, including temperature, pressure,
and CO2 flow rate, while the other parameters were held
constant. The study was then extended to a wider range of other
Pd catalysts to establish the extent to which the nature of the
catalyst affects the control of the reaction.
Effect of Temperature at Constant Pressure. Past
experience10,25,26 suggests that temperature is one of the most
(23) Clark, P. D. Synthesis of Drug Intermediates in Carbon Dioxide. Ph.D.
Thesis. University of Nottingham: Nottingham, U.K., 2006.
(24) Gaunt, M. J.; Yu, J. Q.; Spencer, J. B. J. Org. Chem. 1998, 63, 4172–
4173.
Effect of Pressure at Constant Temperature. Figure 1
indicated that there was little apparent effect on the reaction of
a phase transition from two phases to one, promoted by
increasing temperature. However, the phase state of a binary
mixture, CO2 + organic solvent, can also be manipulated
(25) Bourne, R. A.; Stevens, J. G.; Ke, J.; Poliakoff, M. Chem. Commun.
2007, 4632–4634.
(26) Chapman, A. O.; Akien, G. R.; Arrowsmith, N. J.; Poliakoff, M. Green
Chem. 2010, 12, 310–315.
(27) Yuan, H. M.; Olesik, S. V. J. Chromatogr., A 1997, 785, 35–48.
1204
•
Vol. 14, No. 5, 2010 / Organic Process Research & Development

two distinct parts. The increase in conversion is rapid up to the
dashed line at 100 bar and slower thereafter, probably reflecting
a phase transition from two phases to a single phase at
approximately 100 bar (for the CO2 + THF system).27 Other
experiments using the pressure drop method30 showed that the
influence of the other components (H2 and 1a) is small enough
to be ignored. The effect of pressure on the fluid is different
for different numbers of phases. When there are two phases,
the system is in effect a trickle-bed reactor; increasing the
pressure will increase the quantity of CO2 and H2 dissolved in
the liquid phase, improving both the mass transport and the
liquid-phase H2 concentration. On the other hand, when there
is a single phase (>100 bar), increasing the pressure increases
the density of the reaction mixture, therefore increasing the
residence time. The change in overall density for a biphasic
system is more difficult to quantify, and the increase in dissolved
CO2 will also affect the solubility of the substrate in the liquid
phase.
It should be noted that the phase behavior in an operating
reactor will be slightly different from that of the binary mixtures
which are usually studied because the actual composition of
the mixture changes along the length of the reactor as H2 is
consumed and 1a is converted to 2.
Figure 2. Debenzylation of 1a, varying the H2:1a ratio at 110
°C; for labelling of plots, see Figure 1.
A comparison of Figures 1 and 3 hints at an apparent
contradiction. In Figure 1, the phase transition at 100 °C has
no apparent effect on the reaction, while in Figure 3 the phase
change at 100 bar does have an effect. The likely reason for
this is that the phase transitions are not the same. In Figure 1,
the transition is from two phases to a less dense single phase,
whereas as the pressure increases in Figure 3, the system is
going from two phases to a relatiVely denser single phase.
Increase in CO2 Flow Rate. Figure 4 shows the effect of
increasing the CO2 flow rate is increased conversion of 1a with,
initially at least, only a modest drop in selectivity towards 2a;
this is surprising in view of the probable decrease in residence
time caused by the dilution of the starting material. However,
increasing the proportion of CO2 causes a considerable drop in
the solvent power of the THF solvent.5,6 Therefore, one should
be cautious in drawing conclusions about the effect of high flow
rates of CO2 on the hydrogenation because the points in Figure
4 do not necessarily represent results at a steady state. For
example, at the highest CO2 flow rate it was only possible to
collect a single sample, before precipitation of 1a by the
antisolvent power of CO2 blocked the equipment and ended
the experiment prematurely.
Figure 3. Debenzylation of 1a, increasing the pressure at 100
°C without changing the ratio of CO2 to organic solvent. 2a
was the sole product (i.e the selectivity was ∼100%). So, in this
Figure, we merely show the conversion of 1a. The dashed
vertical line indicates the pressure at which the system switches
from two phases to one.
between single and multiphase by changing the applied pres-
sure.28 Of course, in a continuous reactor, increasing the pressure
will also change the residence time. However, previous work
on the desymmetrisation of a diol in scCO2,29 showed that it
was possible to change the outcome of the reaction completely
by inducing a phase change by changing the pressure. Figure
3 shows the effect of pressure on the debenzylation of 1a at
100 °C.
There is likely to be a change from one phase to two phases
when the CO2 composition (flow rate) is increased beyond 70
mol % CO2. When there are two phases, an increased CO2 flow
rate means an increased superficial gas velocity, which leads
to increased catalyst wetting and a reduced liquid film thick-
ness,31 and hence to better usage of the catalyst and higher
conversions. Such an effect has been reported before by Baiker
Generally speaking, increasing the pressure increases the
conversion, but in Figure 3, the increase in conversion shows
(28) McHugh, M. A.; Krukonis, V. J. Supercritical Fluid Extraction:
Principles and Practice, 2nd ed.; Butterworth-Heinemann: Boston,
1994.
(29) Licence, P.; Gray, W. K.; Sokolova, M.; Gray, W. K.; Poliakoff, M.
J. Am. Chem. Soc. 2005, 127, 293–298.
(30) Akien, G. R.; Skilton, R. A.; Poliakoff, M. Ind. Eng. Chem. Res. 2010,
49, 4974–4980.
(31) Al-Dahhan, M. H.; Larachi, F.; Dudukovic, M. P.; Laurent, A. Ind.
Eng. Chem. Res. 1997, 36, 3292–3314.
Vol. 14, No. 5, 2010 / Organic Process Research & Development
•
1205

Figure 5. Structures of the unexpected byproducts 5 (4-
(cyclohexylmethoxy)phenol) and 6 (4-(cyclohexylmethoxy)cy-
clohexanol).
much higher than with CO2; this may be a secondary effect
due to the much longer residence time in the reactor, because
smaller volumes of gas correspond to a necessarily higher
overall density. (iii) Unexpectedly, significant amounts (5-10
mol %) of monodebenzylated cyclohexyl products, 5 and 6
(Figure 5), were observed when N2 and/or H2 was used; we
can find no precedent for this in the literature. (iv) Most
importantly, the reaction is almost completely unselective when
CO2 is absent; therefore, we have the key result that the presence
of CO2 is introducing selectivity towards 2a.
Variation of Pretreatment of the Catalyst. We encoun-
tered some reproducibility issues in the course of this work, at
least some of which appeared to have been the result of
inadvertent variations in the pretreatments of the catalyst in situ,
before the reaction mixture was flowed through the reactor. The
manufacturers of the 2% Pd/SiO2-Al2O3 catalyst suggest that
it may be preferable not to pretreat heterogeneous Pd catalysts,
since the activity of untreated catalysts is higher for debenzy-
lations.33 Since the conversion is strongly linked to the selectiv-
ity, it was hoped that pretreating the catalyst would not only
affect the activity but might also influence the selectivity. To
this best of our knowledge, there is no information in the
primary literature on the effects of different pretreatments for
selective debenzylation with heterogeneous catalysts.
Figure 4. Debenzylation of 1a; the effect of changing the CO2
flow rate, hence the mol % of CO2 in the reaction mixture at
100 °C and 100 bar, and therefore the overall composition of
the fluid. The data correspond to a change in flow rate from
0.125 to 1.25 mL ·min-1. For labeling of the plots, see Figure 1.
Table 1. Variation of mobile phase for the debenzylation
of 1aa
selectivity (%)
gas
conversion of 1a (%)
2a
3a
by-products
CO2
44.0
100
100
88.7 11.2
0.4 59.4
0.1
40.3
34.7
N2
H2 only
0
65.3
a Standard conditions were used: 0.5 mL ·min-1 CO2, 0.5 mL ·min-1 THF
solution (0.05 mol ·dm-3 1a), 100 °C, 100 bar, 5.5 equiv of H2, with ∼300 mg of
2% Pd/SiO2-Al2O3 catalyst, prereduced under N2/H2 at 100 °C and 100 bar for
2 h. Where CO2 was replaced with N2, it was at the same mass flow rate.
Our results are summarized in Figure 6, and they confirm
that the most active catalyst was the one which had undergone
no prereduction at all. There was a considerable difference
between the activity of the catalyst after the different pretreat-
ments. There was also variability between the experiments, with
some differences between repeats under nominally identical
conditions. This is indicated in Figure 6 by the gradually
increasing conversion in the long experiment (∼8 h), where a
true steady state never seemed to be reached. It seems doubtful
that metal sintering is the cause of this instability, since this
usually occurs only higher temperatures (>500 °C).34-36 The
selectivity towards 2a is not shown explicitly in Figure 6
because it was broadly the same for all pretreatments, although
selectivity was generally slightly lower for the untreated catalyst
than in other experiments. Thus, despite the variability in the
results, it seems reasonable to conclude that pretreatment is
unlikely to be a useful tool for tuning the selectivity of the
catalyst in this reaction.
and co-workers in the Pd-catalysed dehydrogenation of 1-phe-
nylethanol in high pressure CO2.32
Effect of CO2 versus Other Gases. Figure 4 indicated that
increasing the quantity of CO2 in the solvent mixture increases
the conversion of 1, so the presence of CO2 appears to be
important for the selectivity of the reaction. To gain further
insight into the possible role of CO2, three separate experiments
were conducted under conditions which would allow direct
comparisons between different gases: the first experiment was
with CO2, the second with N2 replacing CO2 as the “mobile
phase”, and the third with no additional gas at all, i.e. only with
H2, but still at high pressure. The results are summarized in
Table 1.
Table 1 shows that the results of the three experiments were
strikingly different. The effects of changing the gas are large,
namely (i) The conversion of 1a is much higher with N2 than
with CO2; presumably, this is because CO2 has a higher
solubility in the liquid phase, thereby diluting the liquid phase
more and, hence, reducing the rate of reaction and the
conversion. (ii) When only H2 is present, the conversion is again
(33) A New Series of Catalysts for Deprotection Reactions; Johnson
Matthey: London; www.jmcatalysts.com/pharma/pdfs-uploaded/O%
20Debenz%20Flyer_1.pdf (accessed Jan 5, 2009).
(34) Chen, J. J.; Ruckenstein, E. J. Catal. 1981, 69, 254–273.
(35) Maxted, E. B.; Ali, S. I. J. Chem. Soc. 1964, 1127–1132.
(36) Liu, R. J.; Crozier, P. A.; Smith, C. M.; Hucul, D. A.; Blackson, J.;
Salaita, G. Appl. Catal., A 2005, 282, 111–121.
(32) Burgener, M.; Mallat, T.; Baiker, A. J. Mol. Catal. A: Chem. 2005,
225, 21–25.
1206
•
Vol. 14, No. 5, 2010 / Organic Process Research & Development

Figure 7. Debenzylation of 1a over 5% Pd/Deloxan APII,
variation of temperature; for labelling, see Figure 1.
Figure 6. Observed conversion during debenzylation of 1a over
2% Pd/Al2O3 under constant conditions (100 bar, 100 °C, 5.5
equiv of H2, 0.5 mL ·min-1 both for CO2 and THF solution),
varying pretreatment methods for the catalyst. The selectivities
were largely the same for a given conversion level and so are
not shown. Pretreatment conditions: (A) no pretreatment, (B)
brief prereduction under CO2 at 100 °C for 30 min with 0.71
mol % H2; (C) extensive prereduction under CO2 at 150 °C
for 2 h with 7.56 mol % H2; (D) extensive prereduction under
N2 at 150 °C with 500 mLN ·min-1 N2 (equivalent to the same
mass flow rate as 0.636 mL ·min-1 liquid CO2) for 2 h with
3.94 mol % H2, (E) high temperature prereduction under N2
at 250 °C for 2 h with 7.56 mol % H2; and (F) heating under
pure N2 at 150 °C with 500 mLN ·min-1 for 2 h.
Debenzylation Using Other Catalysts. 5% Pd/Deloxan
APII: Variation of Temperature. The investigation was now
widened to other catalysts in a search for greater selectivity for
the debenzylation of 1a in CO2-expanded THF. The results with
the Degussa catalyst 5% Pd/Deloxan APII were particularly
striking. Figure 7 shows that the catalyst was active at
surprisingly low temperatures. This catalyst gave levels of
conversion and selectivity at only 30 °C, similar to those
observed with 2% Pd/SiO2-Al2O3 at much higher temperatures.
Under these conditions, the residence time in the reactor is
only ∼30 s; so this is a remarkable reaction. The catalyst was
also capable of cleanly producing hydroquinone in almost
quantitative yields at only 50 °C. Given this excellent perfor-
mance, it is disappointing that the Deloxan catalysts are no
longer available commercially.
ConVersion Vs SelectiVity. The results described above
suggest a consistent trade-off between conversion of 1a and
selectivity towards 2a, as would be expected from a reaction
where the rates of the two steps, 1a f 2a and 2a f 3a, are
comparable. Thus, as the final stage of our study, we investi-
gated the correlation of the conversion of 1a versus selectivity
for 2a for a whole series of catalysts and reaction conditions.
The results are summarized in Figure 8.
Figure 8. Conversion versus selectivity for 2a for all 36
experimental runs of the debenzylation of 1. Two temperature
ramps are highlighted with lines joining the points: 60-100 °C
with 5% Pd/CaCO3 (O), and 30-60 °C with 5% Pd/Deloxan
APII (0). The best combination of conversion and selectivity
(both ∼85%) was obtained with 5% Pd/CaCO3.
rather worse. However, there is one catalyst that really stands
out. 5% Pd/CaCO3 gave 86% conversion and 86% selectivity
for 2a at only 65 °C (the circled point in Figure 8).
Catalysts in continuous reactors can easily give rise to local
hotspots during exothermic reactions such as these.37 This makes
CaCO3 particularly interesting as a support for reactions in
scCO2 because the CO2 is likely to suppress degradation to CaO
if the catalyst were to overheat locally.
Figure 8 shows that the majority of data points lies
approximately on the arc of a circle. This implies that the
relationship between conversion and selectivity is similar for
most of the catalysts and conditions with a few performing
(37) Hyde, J. R.; Walsh, B.; Poliakoff, M. Angew. Chem., Int. Ed. 2005,
44, 7588–7591.
Vol. 14, No. 5, 2010 / Organic Process Research & Development
•
1207

Conclusions
In this paper we have described the first systematic study of
catalyst was 5% Pd/CaCO3, which gave 86% conversion with
86% selectivity. These results strongly suggest that it will be
rewarding to study further the reactions in CO2-expanded
solvents of large molecules of potential pharmaceutical interest.
a continuous O-debenzylation in CO2-expanded solvents.
Although experimentally challenging, the continuous deben-
zylation of 1a with H2 and supported Pd catalysts in high-
pressure CO + THF has been quite successful. We have
established the effects on the reaction of varying temperature,
H2:substrate ratio, CO2 flow rate, pressure, prereduction, and
catalyst. The overall conclusion is that those reaction parameters
which increase the conversion all have energy or materials costs
associated with them. Therefore, a more detailed economic
analysis is still needed to identify strategies for increasing the
yield at the lowest possible cost.
Nevertheless, our study has produced three key results: (i)
the reaction can give good selectivity towards the monodeben-
zylated product 2asthat is towards the desymmetrisation of
the symmetrical substrate, 1a; (ii) this selectivity was only
observed in the presence of CO2, a rare example of a major
effect directly attributable to CO2; and (iii) the most effective
Supporting Information Available
Details of the synthesis of 1c and 1d. This material is
available free of charge via the Internet at http://pubs.acs.org.
Acknowledgment
We thank the EPSRC and AstraZeneca for funding and
Johnson Matthey and Degussa for the donation of catalysts.
We are also very grateful to R. A. Stockman for his advice,
and to M. Guyler, P. Fields, and R. Wilson for their technical
assistance.
Received for review May 17, 2010.
OP100136G
1208
•
Vol. 14, No. 5, 2010 / Organic Process Research & Development