Continuous catalytic oxidation of solid alcohols in supercritical CO2: A parametric and spectroscopic study of the transformation of cinnamyl alcohol over Pd/Al2O3

Article abstract of DOI:10.1016/j.jcat.2006.03.009

Cinnamyl alcohol was oxidized to cinnamaldehyde in a continuous fixed-bed reactor with molecular oxygen over an alumina-supported palladium catalyst in supercritical carbon dioxide modified with toluene. A strong dependence of the reaction performance on pressure and oxygen concentration in the feed was found. Optimization of the reaction conditions resulted in a higher catalytic activity than in the liquid phase. At 120 bar, 80 °C, and double stoichiometric oxygen concentration, a turnover frequency of 400 h^-1 at a selectivity of 60% to cinnamaldehyde was achieved. Spectroscopic investigations and the knowledge of the selectivity pattern turned out to be crucial for a deeper understanding of the reaction allowing a rational optimization. Under almost all experimental conditions (even at high oxygen concentration) hydrogenated byproducts, stemming from internal hydrogen transfer reactions, were detected in the effluent. This indicated that alcohol dehydrogenation was the first reaction step; this finding was further confirmed by spectroscopic investigations. In situ XANES and EXAFS revealed that in the entire experimental range investigated, the palladium constituent was mainly in a reduced state, and its surface could be oxidized only in the absence of cinnamyl alcohol in the feed. Bulk-phase behavior studies and investigations at the catalyst-fluid interface, performed by visual inspection and combined transmission and ATR-IR spectroscopy, demonstrated that the reaction performed best in the biphasic region. Moreover, cinnamaldehyde and carbon dioxide, but hardly any toluene and cinnamyl alcohol, were detected inside the porous catalyst, indicating a significantly different product composition inside the porous catalyst compared with the bulk phase.

Full text of DOI:10.1016/j.jcat.2006.03.009

Journal of Catalysis 240 (2006) 126–136
www.elsevier.com/locate/jcat
Continuous catalytic oxidation of solid alcohols in supercritical CO2:
A parametric and spectroscopic study of the transformation
of cinnamyl alcohol over Pd/Al2O3
Matteo Caravati, Daniel M. Meier, Jan-Dierk Grunwaldt ∗, Alfons Baiker
Institute for Chemical and Bioengineering, Department of Chemistry and Applied Biosciences, ETH Zurich, Hönggerberg, HCI, 8093 Zurich, Switzerland
Received 9 January 2006; revised 14 March 2006; accepted 17 March 2006
Available online 24 April 2006
Abstract
Cinnamyl alcohol was oxidized to cinnamaldehyde in a continuous fixed-bed reactor with molecular oxygen over an alumina-supported pal-
ladium catalyst in supercritical carbon dioxide modified with toluene. A strong dependence of the reaction performance on pressure and oxygen
concentration in the feed was found. Optimization of the reaction conditions resulted in a higher catalytic activity than in the liquid phase. At
◦
−1
120 bar, 80 C, and double stoichiometric oxygen concentration, a turnover frequency of 400 h at a selectivity of 60% to cinnamaldehyde
was achieved. Spectroscopic investigations and the knowledge of the selectivity pattern turned out to be crucial for a deeper understanding of the
reaction allowing a rational optimization. Under almost all experimental conditions (even at high oxygen concentration) hydrogenated byproducts,
stemming from internal hydrogen transfer reactions, were detected in the effluent. This indicated that alcohol dehydrogenation was the first reac-
tion step; this finding was further confirmed by spectroscopic investigations. In situ XANES and EXAFS revealed that in the entire experimental
range investigated, the palladium constituent was mainly in a reduced state, and its surface could be oxidized only in the absence of cinnamyl
alcohol in the feed. Bulk-phase behavior studies and investigations at the catalyst–fluid interface, performed by visual inspection and combined
transmission and ATR-IR spectroscopy, demonstrated that the reaction performed best in the biphasic region. Moreover, cinnamaldehyde and
carbon dioxide, but hardly any toluene and cinnamyl alcohol, were detected inside the porous catalyst, indicating a significantly different product
composition inside the porous catalyst compared with the bulk phase.
© 2006 Elsevier Inc. All rights reserved.
Keywords: Aerobic oxidation; Cinnamyl alcohol; Supercritical CO ; Phase behavior; In situ ATR-IR; In situ EXAFS; Operando spectroscopy
2
1. Introduction
Among heterogeneously catalyzed oxidation reactions, the
selective oxidation of alcohols to the corresponding aldehyde
or ketone is important in fine chemistry [13–16]. These reac-
tions are typically performed under mild conditions over noble
metal-supported catalysts, using water as a solvent and mole-
cular oxygen or air as the sole oxidant. For weakly polar al-
cohols, poorly soluble in water, the use of supercritical CO2
has proven beneficial, precluding the need for detergents or or-
ganic solvents [17,18] and thus leading to a safer and more
environmentally friendly process. In several cases, higher rates
and selectivity than the ones obtained in conventional solvents,
as well as improved catalyst lifetime, were reported [17–21].
The reactions were performed either in a batch reactor or in a
continuous fixed-bed reactor in a wide temperature and pres-
sure range. Optimal reaction conditions often depend signifi-
Supercritical carbon dioxide is well suited as a solvent for
oxidation reactions because of its natural abundance, nontoxic-
ity, and inertness, which result in safer operation. Moreover, it
is fully miscible with molecular oxygen (air) and exhibits high
heat capacity [1–10]. Compared with most conventional sol-
vents, supercritical CO2 also results in improved internal and
external mass transport, often leading to an enhanced global
reaction rate. This is especially valid under single-phase condi-
tions [1,2,9,11,12].
*
Corresponding author.
E-mail address: grunwaldt@chem.ethz.ch (J.-D. Grunwaldt).
0021-9517/$ – see front matter © 2006 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2006.03.009

M. Caravati et al. / Journal of Catalysis 240 (2006) 126–136
127
[13,14,16,24,25]. Both oxidized palladium and metallic Pd sites
have been speculated to be the active species in the reaction.
Hence, complementary to the catalytic activity measure-
ments, in situ spectroscopic studies are required for a rational
process optimization. In the present study, we applied both in
situ infrared spectroscopy (in transmission and attenuated to-
tal reflection [ATR] mode) combined with video monitoring
and in situ X-ray absorption spectroscopy (XAS). Although
the IR/visual inspection studies could provide insight into the
phase behavior of the reaction mixture and the phenomena oc-
curring at the catalyst–fluid interface during the reaction, XAS
was used to gain insight into the catalyst structure under reac-
tion conditions. Additional insight into the reaction mechanism
was gained by the selectivity pattern observed during cinnamyl
alcohol oxidation, because the substrate can undergo different
side reactions (Scheme 2) [24,26].
Scheme 1. Reaction scheme of cinnamyl alcohol oxidation to cinnamaldehyde
with molecular oxygen in supercritical CO over an alumina-supported palla-
2
dium catalyst.
cantly on the catalyst, the substrate, and also the reactor type.
Deeper spectroscopic insight into such oxidation reactions has
been gained only recently [17,20,22,23]. Significantly different
features compared to liquid-phase oxidation have been demon-
strated. However, up to now, these kind of studies have been
limited.
In the fine chemicals industry, many of the commercially
interesting alcohols are solid. Only a few studies on the se-
lective oxidation of solid alcohols using supercritical CO2 as a
solvent have been reported [18,21], and in those studies discon-
tinuous reactors (batch reactors) were applied. Here we propose
a simple approach to continuously oxidize solid alcohols with
molecular oxygen in supercritical CO2. The method involves
dissolving the solid alcohol in a small amount of a liquid or-
ganic solvent and continuously feeding it as a reaction mixture
via a liquid pump. One advantage of this method is the flexi-
ble setup that allows continuous processing of both liquid and
solid alcohols without losing the technical and environmental
benefits of supercritical CO2 (i.e., nonflammability, process in-
tensification, improved mass and heat transfer). Moreover, the
alcohol can be easily dosed, and using a co-solvent can prevent
possible blocking of tubing and valves.
The present work focused on the oxidation of the solid and
technically relevant cinnamyl alcohol (Scheme 1). The influ-
ence of the main process variables (pressure, temperature, co-
solvent amount, oxygen concentration) on the reaction rate and
on the product/byproduct distribution was studied. In previous
studies, we showed that a profound knowledge of the phase be-
havior, the solid–fluid interface, and the catalyst structure is
vitally important for reactions in supercritical fluids [9,12,20,
23]. Moreover, even in liquid phase, controversy exists regard-
ing the nature of the active site and the reaction mechanism
2. Experimental
2.1. Catalytic measurements
The reactions were performed isothermally in a fixed-bed
stainless steel tubular reactor with an inner diameter of 13 mm
and a volume of 38 mL. The temperature of the catalyst bed
was monitored by an axially movable thermocouple placed in
the center of the reactor. A commercial shell-impregnated cat-
alyst consisting of 0.5 wt% Pd/Al2O3 (Engelhard 4586) was
crushed and used as a sieved fraction (ca. 0.5–1.4 mm). Pure
alumina (Engelhard 43299) of the same sieved fraction was
used to dilute the catalyst bed, to avoid possible hot spot for-
mation by the exothermic reaction. Then 2.5 g of catalyst and
1.5 g of alumina were loaded, corresponding to a bed length
of ca. 4 cm. Glass beads were placed above the catalyst to en-
sure optimal distribution of the reactants at the entrance of the
catalytic bed. The catalyst was prereduced with H2 before start-
ing the measurements, to improve its stability and guarantee
its complete initial reduction. Cinnamyl alcohol (Fluka, 96330,
purity ꢀ97.0%) and toluene (Fluka, 89681, purity >99.7) were
used as received. Typically, 10 g of cinnamyl alcohol (melt-
Scheme 2. Reaction network observed during the oxidation of cinnamyl alcohol with molecular oxygen in supercritical CO over a supported palladium catalyst.
2
Byproducts detected only in traces are not included.

128
M. Caravati et al. / Journal of Catalysis 240 (2006) 126–136
of 4 cm−1. To perform the experiments, first the alcohol with
toluene was loaded, then CO2 was fed into the cell at 40 ◦C, and
finally the desired pressure was adjusted by heating to 80 ◦C
and changing the volume. The pressure was increased from
120 to 160 bar and then back again to 120 bar while recording
both transmission IR and ATR-IR spectra after each pressure
change of 5 bar. Detailed descriptions of a similar experimental
setup and of the experimental procedure are available elsewhere
[23,27]. To coat the IRE with a thin catalytic layer, 5 mg of
5 wt% Pd/Al2O3 (Johnson–Matthey, 324) was suspended in
0.5 mL of water. The suspension was placed onto the crystal,
which was then dried overnight in air at room temperature.
2.3. XAS measurements
Fig. 1. Dependence of cinnamyl alcohol conversion and selectivity to cinnam-
aldehyde on the concentration of toluene in the feed. The data were obtained
◦
over a stabilized (equilibrated) catalyst; conditions: 150 bar, 80 C, 3.55 ×
In situ XAS experiments were performed in a specially con-
structed flow cell (details in Ref. [22]) at the Hamburger Syn-
chrotronlabor (HASYLAB at DESY, Germany) at beamline
X1 using a Si(311) double-crystal monochromator. The crys-
tals were slightly detuned to minimize higher harmonics, and
three ionization chambers filled with Ar were used to record
the intensity of the incident and transmitted X-ray beam. A Pd
reference foil was used for energy calibration. EXAFS spec-
tra were recorded around the Pd K-edge in the step-scanning
mode between 24000 and 25600 eV. The raw data were energy-
calibrated, smoothed, background-corrected, and normalized
using WINXAS 3.0 software [28]. The k3-weighted EXA−F1S
−4
−4
10 mol/min cinnamyl alcohol, 3.55 × 10 mol/min O , 0.233 mol/min
2
CO , 2.5 g of 0.5 wt% Pd/Al O .
2
2 3
ing point 30–33 ◦C, technical purity) were dissolved in 100 mL
of toluene (corresponding to a molar concentration of toluene
of 1.9%; Fig. 1), with the exception of experiments in which
the amount of toluene was varied to check its influence on
the reaction performance. The cinnamyl alcohol/toluene mix-
ture was fed to the reactor by a Gilson 305 HPLC pump with
a volumetric liquid flow of 0.482 mL/min, resulting in an alco-
hol molar flow rate of 3.55 × 10−4 mol/min. Oxygen (Pangas,
purity ꢀ99.999%) was supplied through a six-port switching
valve (Rheodyne, model 7000) dosing 0.05- or 1-mL pulses
at high pressure and constant frequency. The constant pressure
was maintained by a pressure regulator with CO2, and the total
gas flow was controlled at the vent. Product identification was
performed by GC/MS (HP-5973), and quantitative results were
obtained via GC analysis (HP-6890, FFAP column). Details on
the experimental setup can be found elsewhere [17].
data were Fourier-transformed in the interval k = 2.7–13 Å
.
The in situ flow cell provides optimal conditions for the simulta-
neous monitoring of both catalyst structure and catalytic activ-
ity [22]. The catalyst and the substrate/solvents used during the
operando XAS measurements were the same as those used dur-
ing the catalytic experiments in the conventional reactor. Cin-
namyl alcohol conversion and selectivity to cinnamaldehyde
were determined by GC analysis (HP-6890, FFAP column).
2.2. FTIR and phase behavior measurements
3. Results
Infrared spectroscopic investigations were carried out in a
custom-made stainless steel high-pressure view cell (SITEC
AG) that can withstand pressures up to 200 bar and temper-
atures up to 200 ◦C. A sapphire window covering the entire
diameter allows visual observation and digital imaging of the
bulk-phase behavior. The cell is also equipped with IR transmis-
sion windows (ZnSe, Komlass, internal diameter 8 mm, length
19.2 mm) that allow transmission IR measurements in the up-
per part of the reactor (path length of 0.4 mm). An internal
reflection element (IRE, Komlass, made of ZnSe, trapezoidal
shape, angle of incidence 60◦, length 27 mm, height 2 mm,
depth 10 mm) mounted at the bottom of the hollow cylinder
allows ATR-IR measurements. A motor-driven carriage moves
a set of mirrors that directs the light beam either to the IRE or
to the transmission windows, thus allowing quasi-simultaneous
transmission IR and ATR-IR measurements. Fourier transform
infrared (FTIR) spectra were recorded with a Bruker Optics
IFS-66/s spectrometer equipped with a liquid nitrogen-cooled
MCT detector, typically accumulating 200 scans at a resolution
3.1. Catalyst and co-solvent screening
The selective oxidation of cinnamyl alcohol to cinnamalde-
hyde (Scheme 1) was performed in a continuous-flow reactor
using supercritical CO2 as a solvent and molecular oxygen as
an oxidant. The solid cinnamyl alcohol was dissolved in a fixed
amount of organic co-solvent and then fed to the reactor as a
liquid solution in a continuous way. As a first preliminary step,
screening to identify the best catalyst and a suitable co-solvent
was performed, using the same activation procedure each time.
The catalysts were tested over one day; the results are reported
in Table 1. Palladium supported on alumina turned out to be the
only active catalyst. In addition, performing the reaction with
cinnamyl alcohol dissolved in toluene provided better perfor-
mance than that dissolved in acetone and THF, both in terms
of conversion and selectivity to aldehyde. Therefore, all further
investigations were performed over a 0.5 wt% Pd/Al2O3 using
toluene as a co-solvent.

M. Caravati et al. / Journal of Catalysis 240 (2006) 126–136
129
Table 1
the first working days, with the alcohol conversion in a stan-
dard experiment dropping from 60% on the first day to 14.9%
on the fifth day. In contrast, the selectivity to cinnamaldehyde
increased from 72 to 79%. An increase in selectivity with de-
creasing conversion has been observed in other sets of experi-
ments as well, and it can be considered a general feature of this
reaction under the reaction conditions typically applied in the
present study. Interestingly, after about 5 working days, no fur-
ther deactivation was observed, and the catalyst demonstrating
steady-state behavior was used for the parametric studies.
Conversion of cinnamyl alcohol and selectivity to cinnamaldehyde measured
during oxidation with molecular oxygen in supercritical carbon dioxide mod-
ified with different co-solvents and over different catalysts; in all cases the
catalyst was prereduced in the same manner at 100 C and the catalytic data
were recorded during the first day
◦
a
Catalyst
0.5 wt% Ru/Al O (Engelhard 4871)
0.5 wt% Pt/Al O (Engelhard 4751)
0.5 wt% Pd/Al O (Engelhard 4586)
Co-
solvent
Conver-
sion (%)
Selec-
tivity (%)
b
b
Toluene
Toluene
Toluene
Acetone
THF
–
–
60
44.3
33.3
–
2
3
b
b
–
2
3
71.5
62.8
64.3
2
3
0.5 wt% Pd/Al O (Engelhard 4586)
0.5 wt% Pd/Al O (Engelhard 4586)
2
3
2
3
3.4. Reaction network
a
◦
Conditions: 150 bar, 80 C, 0.15 mol% cinnamyl alcohol, 0.15 mol% O ,
2
1.9 mol% co-solvent in CO .
2
During the catalytic tests, different byproducts were detected
in the effluent and identified by GC/MS. Byproducts originat-
ing from hydrogenation, hydrogenolysis, decarbonylation, and
carbon–carbon double-bond cleavage were observed, but only
some of them were present in appreciable amounts and thus
taken into account in the reaction network shown in Scheme 2.
Note, in the following, we refer to the number appearing in
brackets after the product name for describing the compounds
formed.
b
Below detection limit of GC.
3.2. Reaction performance in toluene-modified dense CO2
In the next step, the effect of toluene concentration on the
reaction performance was investigated; the results are reported
in Fig. 1. A constant reaction rate was found on increasing the
amount of toluene from 1 to 2.9 mol%, with an alcohol con-
version of ca. 15% and enhanced selectivity with an increasing
amount of co-solvent. Increased amounts of toluene resulted in
lower activity (conversion ca. 5% at 15.4 mol% toluene), even
though the selectivity to the aldehyde kept increasing. On the
basis of these results, it was decided to perform the paramet-
ric study presented in this work at a toluene concentration of
1.9 mol%, which seems to represent a good compromise be-
tween the amount of toluene added, catalytic activity, and se-
lectivity to aldehyde. Note that at a lower toluene concentration
of 1 mol%, unstable setup conditions were observed, with the
CO2 flow rate oscillating slightly around the desired value. This
was probably due to the formation of solid substances in the ex-
pansion valves, which could be avoided by applying a higher
amount of co-solvent. Co-solvents [29–31] and expanded liq-
uids [32,33] have also been used in other studies to overcome
solubility limitations [34,35] and in general to tune the sol-
vent properties of SCFs without spoiling their advantageous
properties [36,37]. In some cases, higher reaction rates than
in neat CO2 were reported, indicating that the co-solvent may
play a crucial role in phase behavior, mass transfer properties,
and adsorption/desorption phenomena [29–31,38]. A similar
role of the co-solvent cannot be excluded in the present case,
and thus toluene probably should not be considered only as a
means of dissolving the solid alcohol. In principle, other so-
lutions to process the solid cinnamyl alcohol alone, without a
co-solvent, could be used; for example, injection of melted al-
cohol or the saturation method. However, these solutions would
have required a more complex experimental setup for dosing
the alcohol and for preventing possible solid deposit formation.
3.5. Systematic catalytic investigations
On the basis of the previous results, a parametric study was
performed over a stabilized 0.5 wt% Pd/Al2O3 catalyst using
toluene as a co-solvent. The influence of the total pressure on
conversion and selectivity to cinnamaldehyde is reported in
Fig. 2a. In the low-pressure region, a slight increase in con-
version with increasing pressure was observed. A maximum
was reached at 120 bar, with the turnover frequency (TOF)
increasing up to 400 h−1. At 130–140 bar, a sharp drop in
conversion occurred, with a TOF of 90 h−1 at 150 bar. This
behavior is in contrast to previous observations for benzyl alco-
hol [23], and we discuss it later in the light of phase behavior
and X-ray absorption spectroscopy results. In contrast, the se-
lectivity to cinnamaldehyde constantly increased with pressure,
showing a more marked jump between 130 and 140 bar, corre-
sponding to the drop in reaction rate. At 120 bar, where the
maximum reaction rate was observed, the selectivity to cin-
namaldehyde was 58.4%. Fig. 2b shows the product/byproduct
distribution, revealing that the main product cinnamaldehyde is
always present in the highest concentration. In the low-pressure
region (corresponding to the higher catalytic activity) the two
hydrogenated byproducts, 3-phenyl-1-propanol (2) and beta-
methylstyrene (5), are present in comparable amounts to cin-
namaldehyde (1).
Another important variable that has a significant effect on
reaction performance is the oxygen concentration in the feed.
Two sets of experiments at 120 and 150 bar were performed
with varying oxygen concentrations up to 10 mol% (Figs. 3
and 4). In a test performed without oxygen feed, similar cat-
alytic activity was measured at 120 bar (cinnamyl alcohol con-
version of 12.4%) and 150 bar (conversion of 13%), with dif-
fering selectivities to cinnamaldehyde (57.6% at 120 bar vs.
79% at 150 bar). The dehydrogenation via internal hydrogen
3.3. Catalyst deactivation
In contrast to previous studies on the selective oxidation of
benzyl alcohol in scCO2, in the case of cinnamyl alcohol, the
palladium catalyst demonstrated significant deactivation during

130
M. Caravati et al. / Journal of Catalysis 240 (2006) 126–136
◦
Fig. 2. Conversion of cinnamyl alcohol and selectivity to cinnamaldehyde (a), and product/byproduct distribution (b) as a function of pressure; conditions: T = 80 C,
0.15 mol% alcohol, 0.15 mol% O , 1.9 mol% toluene in CO (0.233 mol/min), 2.5 g of 0.5 wt% Pd/Al O .
2
2
2 3
Fig. 3. Conversion of cinnamyl alcohol and selectivity to cinnamaldehyde (a), and product/byproduct distribution (b) at 120 bar as a function of oxygen concentration
◦
in the feed; conditions: 120 bar, 80 C, 0.15 mol% alcohol, 1.9 mol% toluene in CO (0.233 mol/min), 2.5 g of 0.5 wt% Pd/Al O .
2
2 3
transfer seemed dominant, as also indicated by the byproduct
distribution. At 120 bar, the hydrogenated products 3-phenyl-
1-propanol (11.2%) and beta-methylstyrene (30.9%) were de-
tected, and at 150 bar, mainly 3-phenyl-1-propanol (19.2%) was
found. Adding oxygen (0.15 mol%, same concentration of al-
cohol) resulted in a markedly increased reaction rate at 120 bar
(Fig. 3a) with increasing alcohol conversion from 12.4 to 58.7%
and a virtually unchanged selectivity to aldehyde. The byprod-
uct distribution changed slightly, showing increased production
of 3-phenyl-1-propanol (2) and decreased production of beta-
methylstyrene (5). Increasing the oxygen concentration up to
10 mol% produced no clear trend, showing at first a slight in-
crease in conversion with oxygen concentration up to 1.2%,
followed by a slight deactivation. However, no strong deacti-
vation like that observed during benzyl alcohol oxidation [22]
was measured, and at 10 mol% O2 in the feed, cinnamyl alco-
hol conversion was still 51.4%. The selectivity to cinnamalde-
hyde increased with increasing amounts of oxygen (Fig. 3).
At 10 mol%, the product/byproduct distribution was 82.2%
cinnamaldehyde, 11.7% benzaldehyde, and 3.6% 3-phenyl-1-
propanol. Generally, with increasing oxygen partial pressure,
the hydrogenated byproducts 2 and 5 decreased and the oxy-
genated byproduct benzaldehyde (4) increased. This striking
result indicates the greater availability of surface oxygen on the
palladium active sites.
In contrast, adding 0.15 mol% oxygen at 150 bar (Fig. 4)
did not boost the reaction rate; compared with the oxygen-
free experiment, conversion and selectivity to cinnamaldehyde
remained more or less unchanged. As Fig. 4 shows, an in-
crease in oxygen concentration resulted in an initial decrease
(at 1.2 mol%), followed by a marked increase in catalytic ac-
tivity. The conversion jumped from 6.4% at 1.2 mol% O2 to
62% at 7.6 mol% O2. A further increase to 10 mol% O2 in
the feed led to only a slight deactivation, smaller than that for
benzyl alcohol [20,22]. At 10 mol%, cinnamyl alcohol conver-
sion was 53% and selectivity to cinnamaldehyde was 72.9%.
The byproduct distribution indicated the predominance of the
oxygenated products, but some internal hydrogen transfer was
still observed (20% of benzaldehyde vs. 5.2% of 3-phenyl-1-
propanol and traces of beta-methylstyrene).
3.6. Phase behavior studies
The phase behavior of the reaction mixture was studied by
video recording combined with transmission IR spectroscopy
of the bulk supercritical phase and ATR-IR spectroscopy of
the liquid phases. A similar approach was previously used for

M. Caravati et al. / Journal of Catalysis 240 (2006) 126–136
131
Fig. 4. Conversion of cinnamyl alcohol and selectivity to cinnamaldehyde (a), and product/byproduct distribution (b) at 150 bar as a function of oxygen concentration
◦
in the feed; conditions: 150 bar, 80 C, 0.15 mol% alcohol, 1.9 mol% toluene in CO (0.233 mol/min), 2.5 g of 0.5 wt% Pd/Al O .
2
2 3
◦
Fig. 5. Snapshots of the phase behavior of the reaction mixture in dense carbon dioxide at 80 C and containing 0.15 mol% cinnamyl alcohol, 0.15 mol% O ,
2
7.7 mol% toluene at: (a) 126.4 bar in the biphasic region; (b) 128.7 bar in the intermediate region; (c) 130.6 bar in the single phase region; (a movie of the phase
transition is provided in the electronic support information).
benzyl alcohol, demonstrating that information on phase com-
position as well as phase behavior can be obtained [23]. Varying
the amount of toluene had little effect on the transition from the
biphasic region to the monophasic region of the reaction mix-
ture. For four different compositions (1, 1.9, 7.7, and 15.4 mol%
toluene in 0.15 mol% alcohol/0.15 mol% O2/CO2), the transi-
tion to a single supercritical phase occurred between 120 and
130 bar. Below 120 bar, at which the greatest activity was mea-
sured, all the mixtures were present as biphasic systems. This
is apparent in the snapshots shown in Fig. 5, which were taken
while investigating the phase behavior of the mixture contain-
ing 7.7 mol% toluene. A pressure increase of ca. 4 bar resulted
in a sharp transition from a two-phase region at 126.4 bar
(Fig. 5, snapshot a) via an intermediate region with a higher
volume of the CO2-expanded toluene phase at 128.7 bar (Fig. 5,
snapshot b) to a single-phase region at 130.6 bar (Fig. 5, snap-
shot c). This phase transition can be best visualized by a movie,
provided as electronic supplementary information. The phase
behavior of the 1.9 mol% toluene-containing mixture (the mix-
ture used in the parametric study) was also studied by ATR-IR
spectroscopy because of the small amount of liquid present in
the cell. The results are reported in Fig. 6. The transition to a
single supercritical phase in the bulk started at 120 bar, as indi-
cated by the decreased intensity of the cinnamyl alcohol bands
Fig. 6. ATR-IR spectra recorded at the bottom of the view cell in the ν -band
2
region of CO at increasing pressure for the 1.9 mol% toluene containing re-
2
−1
action mixture; the cinnamyl alcohol bands at 730 and 692 cm indicate the
presence of an alcohol-rich liquid-like phase at 120 bar, a transition towards a
single phase is observed at high pressure by the shift of the ν -band of CO
2
2
−1
from 660 to 662 cm and by the decrease of the alcohol bands; (conditions:
◦
80 C, 0.15 mol% cinnamyl alcohol, 0.15 mol% O , 1.9 mol% toluene in CO ).
2
2
at 730 and 692 cm−1 and the shift to higher wavenumbers of
the ν2-band of CO2 (660 cm−1 at 115 bar in spectrum 1 and
662 cm−1 at 120 bar in spectrum 2). The shift of the ν2-band

132
M. Caravati et al. / Journal of Catalysis 240 (2006) 126–136
Fig. 7. Comparison of the ATR-IR spectra recorded at the bottom of the view
Fig. 8. On-line monitoring of the catalyst/fluid interface by ATR-IR spec-
troscopy during the selective oxidation of cinnamyl alcohol to cinnamaldehyde
in the view cell equipped with IR spectroscopy. On the right development of
cell of a 0.15 mol% cinnamyl alcohol, 0.15 mol% O , 1.9 mol% toluene in
2
◦
CO mixture at 120 bar and 80 C in the region of the aromatic deformation
2
−1
−1
−1
bands (730 and 692 cm ) and of the ν band of CO (662 cm ) recorded
the carbonyl stretching band of cinnamaldehyde (1683 cm ) during reaction,
2
2
−1
with and without a catalyst layer of 0.5 wt% Pd/Al O on the ZnSe crystal; in
on the left rise of byproducts bands (3215 and 3134 cm ) at the end of the
2
3
◦
the catalyst pores supercritical-like conditions are reached at a lower pressure
(see text).
reaction are shown; (conditions: 120 bar, 80 C, 0.15 mol% cinnamyl alcohol,
0.15 mol% O , 1.9 mol% toluene in CO , 5 mg of 5 wt% Pd/Al O deposited
2
2
2 3
on the crystal).
in the dense liquid-like phase is due to the interaction of CO2
and alcohol [23,39], but alkylcarbonic acids were not found in
significant amounts [40]. However, at 130 bar and after a long
equilibration time, the bands of alcohol were still visible (spec-
trum 3), and higher pressure was needed to completely dissolve
the substrate. At 150 bar, the bands of alcohol could hardly be
detected (spectrum 4), as a consequence of the very short path
length of the probing beam in the ATR-IR technique. However,
under single-phase conditions, the ATR-IR signals reflect only
the data that can also be extracted from the transmission IR re-
sults. These observations indicate a slower transformation than
the sudden transformation observed for the mixture containing
7.7 mol% toluene (Fig. 5), but in this case as well, at 120 bar
the mixture was still present as two phases.
may be due to the higher virtual pressure inside narrow pores,
which is known to lead to a shift in the critical temperature
[41–43]; therefore, we describe the behavior inside the porous
catalyst as “supercritical” even though it is well known that
the critical parameters are clearly defined only in the case of
pure substances [1,9,11]. In summary, the supercritical condi-
tions observed above 150 bar in the bulk phase were reached at
lower pressure and temperature in the catalyst pores.
Cinnamyl alcohol was then oxidized over the palladium cat-
alyst at 120 bar and 80 ◦C in the batch reactor infrared view
cell (0.15 mol% alcohol, 0.15 mol% O2, 1.9 mol% toluene,
97.8 mol% CO2). With increasing reaction time in the ATR-
IR spectra recorded at the catalyst–fluid interface there is
an increase of the cinnamaldehyde C–O stretching band at
1683 cm−1 (Fig. 8), whereas for the entire reaction time hardly
any band of alcohol was detectable. At the end of the reaction
(after ca. 280 min), the band of cinnamaldehyde was quite in-
tense, indicating that the aldehyde was not fully dissolved in
the bulk supercritical phase but remained in the catalyst pores as
well. Two intense bands at 3134 and 3215 cm−1 also appeared,
which may be assigned to byproducts containing carboxylic
groups. Performing the reaction at 150 bar resulted in a lower
reaction rate, in line with the observation during the catalytic
tests, and during the whole reaction time bands of aldehyde or
alcohol were hardly visible in the ATR-IR spectra. At the end of
the reaction (after ca. 680 min), two intense bands at 3128 and
3222 cm−1 appeared (not shown), which were very similar to
those observed at 120 bar and thus probably belong to similar
types of byproducts. The conversion of cinnamyl alcohol in the
batch reactor at 120 and 150 bar as a function of reaction time,
calculated from the area of the cinnamaldehyde C=O stretch-
ing band is shown in Fig. 9. These measurements were taken
while the phenomena were investigated at the catalyst–fluid in-
terface and are in good accordance with the catalytic results
3.7. Investigations at the catalyst–fluid interface by ATR-IR
spectroscopy
In situ ATR-IR studies at the solid–fluid interface of the Pd-
catalyst were performed by coating the ATR crystal with a layer
of active catalyst with the aim of gaining information on the
phase behavior inside the catalyst pores and on the interactions
among CO2, toluene, substrates, and the solid catalyst. The re-
action progress was simultaneously monitored by transmission
IR spectroscopy. The reaction conditions applied resembled as
closely as possible those encountered during the catalytic tests
in the continuous-flow reactor. In addition, experiments were
performed such that the spectroscopic data and catalytic activity
were reported at the same time, leading to real in situ investiga-
tions.
Interestingly, in the presence of a catalyst layer, hardly any
band of cinnamyl alcohol or toluene was detected in the ATR-
IR spectrum (Fig. 7) and the carbon dioxide ν2 band was not
shifted as it was in the liquid-like bulk phase. Hence the sit-
uation is similar to the single, supercritical bulk phase. This

M. Caravati et al. / Journal of Catalysis 240 (2006) 126–136
133
Table 2
Catalytic activity results measured under different reaction conditions during
cinnamyl alcohol oxidation over a palladium catalyst while in situ monitoring
a
the structure of the solid catalyst by X-ray absorption spectroscopy
Pressure Alcohol Oxygen Conver- Selectivity TOF
Spectrum
#
−1
b
(bar)
(mol%) (mol%) sion (%) (%)
(h
)
c
120
120
120
150
150
150
0.15
0.15
0.15
0.15
0.15
–
–
5.2
8.0
5.7
2.7
7.0
–
62.6
45.6
76.6
81.7
87.3
–
172
265
191
91
234
–
–
3
–
4
5
6
0.15
7.0
0.15
7.0
5.0
c
a
Conditions: 0.4689 g of 0.5 wt% Pd/Al O loaded in the EXAFS flow cell,
2
3
◦
T = 80 C, 1.9 mol% toluene, 0.233 mol/min CO .
2
b
EXAFS spectra shown in Fig. 10.
Not shown.
c
Fig. 9. Conversion as a function of time during selective oxidation of cin-
namyl alcohol to cinnamaldehyde at 120 and 150 bar in the spectroscopic batch
reactor view cell, extracted from the area of the carbonyl stretching band of cin-
experimental setup was developed to mimic as closely as possi-
ble the experiments at the synchrotron with those performed in
the home laboratory [22]. Therefore, the reaction conditions ap-
plied during the XAS investigations are exactly the same (with
respect to, e.g., pressure, temperature, CO2 flow, and reactant
concentration) as those typically applied in conventional experi-
ments. The spectra, focused on the XANES region, are reported
in Fig. 10a, and the corresponding Fourier-transformed EXAFS
spectra are shown in Fig. 10b. The catalyst was at first reduced
in 5% H2/He at 100 ◦C (spectrum 1), and then exposed to a
CO2 flow (0.233 mol/min) at 120 bar and 80 ◦C, resulting in
slight reoxidation of the palladium constituent, indicated by
the decreased intensity of the Pd–Pd backscattering at 2.75 Å
(Fig. 10b, spectrum 2). Cinnamyl alcohol (0.15 mol%), dis-
solved in toluene (1.9 mol%), was then added, with virtually
no change in the EXAFS spectrum observed (not shown) but
significant catalytic activity measured (Table 2). Also adding
oxygen (0.15 mol%) resulted in a further slight decrease in
Pd–Pd backscattering (Fig. 10b, spectrum 3). The reaction also
proceeded readily in the EXAFS flow cell (Table 2); a TOF of
265 h−1 was measured, in line with the results of experiments
in the conventional reactor (Fig. 3), emphasizing the proximity
◦
namaldehyde; conditions: T = 80 C, 0.15 mol% cinnamyl alcohol, 0.15 mol%
O , 1.9 mol% toluene in CO , 5 mg of 5 wt% Pd/Al O deposited on the crys-
2
2
2 3
tal.
described above. Experiments performed at 120 bar without the
addition of toluene (not shown) resulted in lower catalytic ac-
tivity, underlining the beneficial role of the co-solvent, which
thus cannot be considered solely as a means of dissolving the
solid alcohol. The phenomena at the solid–fluid interface did
not differ significantly from those observed in the presence of
toluene.
3.8. XAS experiments
Operando X-ray absorption experiments were performed to
shed more light on the oxidation state of the palladium catalyst
during cinnamyl alcohol oxidation and to investigate whether
the measured changes in activity could be traced back to mod-
ifications of the catalyst structure. The X-ray absorption spec-
troscopic data were measured under the reaction conditions in
which the catalytic activity exhibited the major changes. The
Fig. 10. In situ EXAFS investigations on a 0.5 wt% Pd/Al O catalyst at the Pd K-edge under different reaction conditions: (a) energy-calibrated and back-
2
3
ground-corrected XANES spectra; (b) Fourier transformed EXAFS spectra; (1) in 5% H /He; (2) in CO at 120 bar; (3) in 0.15% alcohol, 0.15% O , CO at
2
2
2
2
120 bar; (4) in 0.15% alcohol, 0.15% O , CO at 150 bar; (5) in 0.15% alcohol, 7% O , CO at 150 bar; (6) in 5% O , CO at 150 bar; (details in text).
2
2
2
2
2
2

134
M. Caravati et al. / Journal of Catalysis 240 (2006) 126–136
Fig. 11. Schematic representation of the phenomena taking place in the catalyst and at the Pd-particles; the presence of a dense liquid-like bulk phase, rich in
aromatics, favors the dissolution of cinnamaldehyde; inside the pores mainly scCO and cinnamaldehyde are present; the alcohol is not observed by ATR-IR
2
spectroscopy due to a rapid dehydrogenation reaction.
to real operando studies. Increasing the oxygen concentration
to 7 mol% resulted in lower catalytic activity with improved se-
lectivity to cinnamaldehyde (Table 2), but the EXAFS spectrum
remained virtually unchanged (not shown). Increasing the pres-
sure to 150 bar (at 0.15 mol% oxygen), reaching single-phase
conditions in the bulk supercritical phase, had no detectable ef-
fect on the catalyst structure, as shown by spectrum 4, which
exhibited features very similar to those of spectrum 3 in both the
XANES region (Fig. 10a) and the Fourier-transformed EXAFS
spectra (Fig. 10b). As expected, the catalytic activity dropped,
and a TOF of 91 h−1 was achieved under these conditions. In-
creasing the oxygen amount to 7 mol% at 150 bar resulted in
increased activity, with the TOF rising to 234 h−1, once again
in agreement with the measurements performed in the conven-
tional fixed-bed reactor (Fig. 4). Also at 7 mol% of oxygen in
the feed, no significant changes were observed in the oxidation
state of the palladium constituent, except for a further slight de-
crease in Pd–Pd backscattering (spectrum 5 in Fig. 10b). Stop-
ping the alcohol feed, thus leaving the catalyst in 5% O2 in CO2
at 150 bar and 80 ◦C, finally resulted in oxidation of the palla-
dium constituent. This is clearly shown by the decreased Pd–Pd
feature at 24.39 keV in the near-edge region of the X-ray ab-
sorption spectrum (Fig. 10a, spectrum 6) and is also confirmed
by the remarkably less intense Pd–Pd backscattering in the
Fourier-transformed EXAFS spectrum (Fig. 10b, spectrum 6).
Interestingly, under these conditions, Pd–O backscattering also
could be clearly observed at 1.5 Å. Fitting results showed that
the coordination number of Pd decreased from about 7 to 6
(R = 2.75 Å) and that of oxygen was about 2 (R = 1.98 Å).
ity pattern and spectroscopic investigations indicated different
reaction pathways in these two regions of high catalytic perfor-
mance than those observed in liquid-phase oxidation [45].
The effect of pressure on reaction rate can be traced back
to the characteristic phase behavior of the reaction mixture
(Fig. 5). At low pressure, two phases were present in the bulk,
with a CO2-rich gas-like phase (containing also part of the cin-
namyl alcohol and almost all of the toluene) in equilibrium with
an alcohol-rich liquid-like phase (Figs. 5 and 6). The higher
activity in the biphasic region of the bulk-phase behavior is op-
posite to that reported for benzyl alcohol [20,23] and clearly
indicates that the transition to a single bulk phase has a negative
influence on the reaction rate. Hence, gas–liquid mass trans-
fer is not rate-limiting. Whereas the bulk phase is present at
120 bar as a biphasic reaction mixture with a dense liquid-like
phase probably in contact with the catalyst particles (as sug-
gested by the ATR-IR investigations performed without coating
the catalyst on the crystal), ATR-IR spectroscopy, performed
by coating the crystal with a layer of an active catalyst, sug-
gests that the conditions inside the pores resemble the situation
found in the bulk phase under single-phase conditions (e.g., rich
in carbon dioxide). This situation is shown schematically in
Fig. 11. A liquid-like phase in the bulk is probably beneficial
because it leads to a higher alcohol concentration around the
particles and exhibits a higher solvation power, which is ben-
eficial for the extraction (i.e., desorption and transport) of the
aldehyde and the byproducts, including water. Cinnamaldehyde
was observed inside the porous catalyst (cf. Fig. 8), whereas this
was not the case for benzaldehyde, which can also be well dis-
solved by supercritical carbon dioxide [23,46].
The selectivity pattern under these conditions indicates a
high hydrogen concentration on the catalyst surface. In fact,
alcohol oxidation over noble metal catalyst proceeds via ox-
idative dehydrogenation, the first step being the adsorption and
dehydrogenation of the alcohol to the corresponding aldehyde
or ketone [16,47]. The hydrogen thus formed can be removed
either by oxidation with molecular oxygen [45] or, in the case
of cinnamyl alcohol, by internal hydrogen transfer [48], as indi-
cated in Fig. 11. The high catalytic activity at oxygen-free con-
ditions and 3-phenyl-1-propanol (2) as the main byproduct both
4. Discussion
In this study, cinnamyl alcohol was oxidized by molecular
oxygen for the first time in a continuous way over a palladium
catalyst in supercritical carbon dioxide. The reaction rate was
significantly higher than that in conventional organic solvents
[16,44]. A strong dependence of reaction rate on pressure and
oxygen concentration was found (Figs. 2–4), with high perfor-
mance achieved in the low-pressure region (90–120 bar) and at
high oxygen concentration (5–8 mol%). However, the selectiv-

M. Caravati et al. / Journal of Catalysis 240 (2006) 126–136
135
in the presence and in the absence of oxygen reveals that the re-
action readily proceeds by internal hydrogen transfer and that
hydrogen is available on the catalyst surface in high concen-
tration. This also favors hydrogenolysis reactions, as indicated
by the high concentration of beta-methylstyrene (5) in the ef-
fluent. Interestingly, despite the high concentration of surface
hydrogen, propylbenzene (from the hydrogenation of the C=C
double bond of beta-methylstyrene) was detected only in traces.
Higher amounts of beta-methylstyrene than in liquid phase ox-
idation were found [24,45] and this difference can probably be
traced back to its higher affinity for scCO2. As soon as it is
formed, it quickly desorbs and diffuses out of the catalyst pores
without further reaction.
was observed due to the high concentration of surface hydro-
gen available under all the reaction conditions investigated. In
situ XAS studies support this conclusion. Under almost all the
experimental conditions investigated the palladium constituent
was found to be in an almost fully reduced state during alcohol
oxidation. Because an increase in pressure from 120 to 150 bar
(at 0.15 mol% of oxygen in the feed) resulted in no changes
in the catalyst structure, we can exclude the possibility that
the liquid-like phase present in the bulk protected the catalyst
from overoxidation, by, for instance, decreasing the diffusion
rate of oxygen. In addition, also at high oxygen concentration
(7 mol%), no significant oxidation of the palladium constituent
was observed (Fig. 10). Interestingly, only when the alcohol
feed was stopped, thus leaving the catalyst in 5 mol% O2/CO2,
did an evident oxidation of the surface palladium occur. This
underscores the role of surface hydrogen in preventing oxida-
tion of the catalyst and indicates that the dehydrogenation step
actually proceeds. The results are also in good agreement with
the observed byproduct distribution, particularly of the prod-
ucts of hydrogenolysis. The internal hydrogen transfer is a very
sensitive probe for Pd–H species on the surface. Also at high
oxygen concentration, even though benzaldehyde was the major
byproduct, 3-phenyl-1-propanol (2) and 3-phenyl propionalde-
hyde (3), derived from internal hydrogen transfer (Scheme 2),
were still present in considerable amounts, demonstrating that
even under these conditions hydrogen is still available on the
palladium surface.
The increased reaction rate at 150 bar and high oxygen con-
centration (5–8 mol%; Fig. 4) may be attributed to a change
in bulk-phase behavior, because the reaction mixture becomes
biphasic at high O2 concentrations, as expected based on
the decrease in overall density [49]. But the phase composi-
tion/behavior inside the catalyst pores remained unchanged.
Again, two phases in the bulk phase and supercritical-like con-
ditions inside the catalyst pores led to a high catalytic activity.
The reaction pathway at high oxygen concentration is different,
as indicated by the byproduct distribution, which showed benz-
aldehyde as the main byproduct. Benzaldehyde can be formed
from cinnamic acid by oxidative cleavage of the carbon–carbon
double bond. This reaction is discussed in the literature, even
though it requires severe reaction conditions with strong oxi-
dants [50–52]. In the present case, the reaction may be favored
by rapid removal of benzaldehyde due to its high solubility
in supercritical carbon dioxide [23,46,53], which continuously
shifts the equilibrium of the reaction toward benzaldehyde.
Cinnamic acid was observed in only trace amounts during the
catalytic tests, whereas carboxylic species at the catalyst–fluid
interface were detected by infrared spectroscopy (Fig. 8). It may
be formed by acid-catalyzed hydration with the co-product wa-
ter of cinnamaldehyde, followed by oxidative dehydrogenation
of the hydrate [54]. Interestingly, in the experiment performed
at 120 bar, benzaldehyde was the main byproduct only above
5 mol% of oxygen in the feed, whereas at 150 bar it was the
main byproduct already at ca. 1 mol% of oxygen. This again
may be traced back to the bulk-phase behavior. At 150 bar,
supercritical conditions in the bulk imply improved external
mass transfer and thus higher oxygen concentration at the cat-
alyst surface. From previous results [31], we can exclude the
possibility that benzaldehyde is derived from the oxidation of
toluene. Typically, the reaction occurs at higher temperatures
[55,56].
5. Conclusion
In this paper we have presented a simple and effective
method for the continuous selective oxidation of solid cinnamyl
alcohol in supercritical carbon dioxide over a heterogeneous
palladium catalyst. Using toluene as a co-solvent allowed the
reaction to be performed efficiently and in a continuous man-
ner. During optimization, a high reaction rate with a TOF of
400 h−1 and selectivity to cinnamaldehyde of ca. 60% was
achieved. A marked dependence on pressure and oxygen con-
centration was found. Depending on the different reaction con-
ditions, hydrogenation, hydrogenolysis, and oxidative cleavage
of carbon–carbon double bonds were observed as side reac-
tions. Phase behavior studies, in situ transmission and ATR-
IR spectroscopy, as well as operando XANES/EXAFS studies,
were excellent tools for gaining deeper insight into the reaction
mechanism, the solid/fluid interface, and the state of the cata-
lyst. The oxidation reaction is significantly different to benzyl
alcohol oxidation in supercritical carbon dioxide. In contrast
to benzyl alcohol oxidation, a biphasic reaction mixture was
beneficial for the catalytic performance of the Pd-catalyzed ox-
idation of cinnamyl alcohol, and no deactivation due to oxygen
in high concentration was observed. Both the reaction network
and the in situ XANES and EXAFS studies under reaction con-
ditions revealed that Pd was mainly in the metallic state, even
in the presence of high amounts of oxygen. In addition, IR-
spectroscopic studies at the solid–fluid interface indicated that
the phase composition was significantly different in the catalyst
pores than in the bulk phase, particularly under reaction condi-
Oxygen has thus a strong influence on the phase behavior
and on the reaction pathway, but in contrast to the previous ob-
servation with benzyl alcohol [20,22] it did not play a direct
role in catalyst deactivation. Up to 10 mol% of oxygen in the
feed no strong deactivation could be observed (Figs. 3 and 4).
In principle, deactivation due to over-oxidation may take place
when the oxygen supply on the catalyst is higher than its con-
sumption by the reactions, thus leading to a surface oxidation of
the palladium constituent [14,16,24,57,58]. In the present case
no deactivation due to over-oxidation of the catalyst surface

136
M. Caravati et al. / Journal of Catalysis 240 (2006) 126–136
tions. This indicates that a good understanding of the fluid or
reaction mixture inside a porous catalyst is inevitable at near-
or supercritical conditions.
[16] T. Mallat, A. Baiker, Chem. Rev. 104 (2004) 3037.
[17] G. Jenzer, M.S. Schneider, R. Wandeler, T. Mallat, A. Baiker, J. Catal. 199
(2001) 141.
[18] A.M. Steele, J. Zhu, S.C. Tsang, Catal. Lett. 73 (2001) 9.
[19] R. Gläser, R. Jos, J. Williardt, Top. Catal. 22 (2003) 31.
[20] M. Caravati, J.-D. Grunwaldt, A. Baiker, Catal. Today 91–92 (2004) 1.
[21] S.C. Tsang, J. Zhu, A.M. Steele, P. Meric, J. Catal. 226 (2004) 435.
[22] J.-D. Grunwaldt, M. Caravati, M. Ramin, A. Baiker, Catal. Lett. 90 (2003)
221.
[23] M. Caravati, J.-D. Grunwaldt, A. Baiker, Phys. Chem. Chem. Phys. 7
(2005) 278.
[24] J.-D. Grunwaldt, C. Keresszegi, T. Mallat, A. Baiker, J. Catal. 213 (2003)
291.
Acknowledgments
The authors gratefully acknowledge the Swiss Federal Office
of Energy (BFE) for financial support and Engelhard Italiana
S.r.l. for providing the catalyst samples. They also thank the
Hamburger Synchrotron Labor (HASYLAB) at DESY (Ham-
burg, Germany) for beam time and Dr. E. Welter for the sup-
port during the measurements. This work at the synchrotron
radiation source was supported by the European Community
- Research Infrastructure Action under the FP6 Structuring the
European Research Area program (through the Integrated In-
frastructure Initiative [Integrating Activity on Synchrotron and
Free Electron Laser Science] contract RII3-CT-2004-506008).
Thanks are due to Drs. T. Mallat, D. Ferri, and C. Keresszegi
for help and discussion, and to R. Mäder and P. Trüssel (me-
chanical workshop, ETH Zürich) for their continuous technical
assistance. Dr. M. Burgener is acknowledged for the help during
the phase-behavior measurements, and SITEC Sieber Engineer-
ing AG is recognized for their steady interest and continuous
support.
[25] A.F. Lee, K. Wilson, Green Chem. 6 (2004) 37.
[26] C. Keresszegi, T. Mallat, J.-D. Grunwaldt, A. Baiker, J. Catal. 225 (2004)
138.
[27] M.S. Schneider, J.-D. Grunwaldt, T. Bürgi, A. Baiker, Rev. Sci. Instrum.
74 (2003) 4121.
[28] T. Ressler, J. Synchr. Rad. 5 (1998) 118.
[29] Z. Hou, B. Han, L. Gao, Z. Liu, G. Yang, Green Chem. 4 (2002) 426.
[30] Y. Chang, T. Jiang, B. Han, L. Gao, R. Zhang, Z. Liu, W. Wu, Ind. Eng.
Chem. Res. 42 (2003) 6384.
[31] M. Caravati, J.-D. Grunwaldt, A. Baiker, Appl. Catal. A 298 (2006) 50.
[32] M. Wei, G.T. Musie, D.H. Busch, B. Subramaniam, J. Am. Chem. Soc. 124
(2002) 2513.
[33] B. Kerler, R.E. Robinson, A.S. Borovik, B. Subramaniam, Appl. Catal.
B 49 (2004) 91.
[34] J.M. Dobbs, J.M. Wong, K.P. Johnston, J. Chem. Eng. Data 31 (1986) 303.
[35] J.M. Dobbs, J.M. Wong, R.J. Lahiere, K.P. Johnston, Ind. Eng. Chem.
Res. 26 (1987) 56.
Electronic support information available
[36] C.A. Eckert, K. Chandler, J. Supercrit. Fluids 13 (1998) 187.
[37] C.A. Eckert, D. Bush, J.S. Brown, C.L. Liotta, Ind. Eng. Chem. Res. 39
(2000) 4615.
[38] D.M. Ginosar, B. Subramaniam, J. Catal. 152 (1995) 31.
[39] S.G. Kazarian, M.F. Vincent, F.V. Bright, C.L. Liotta, C.A. Eckert, J. Am.
Chem. Soc. 118 (1996) 1729.
[40] K.N. West, C. Wheeler, J.P. McCarney, K.N. Griffith, D. Bush, C.L. Li-
otta, C.A. Eckert, J. Phys. Chem. A 105 (2001) 3947.
[41] L.D. Gelb, K.E. Gubbins, R. Radhakrishnan, M. Sliwinska-Bartkowiak,
Rep. Prog. Phys. 62 (1999) 1573.
A movie recorded while investigating the phase behavior of
a 0.15 mol% alcohol/0.15 mol% O2/7.7 mol% toluene in CO2
is available from the web pages of Journal of Catalysis.
Please visit DOI:10.1016/j.jcat.2006.03.009.
References
[1] P.G. Jessop, W. Leitner, Chemical Synthesis Using Supercritical Fluids,
Wiley–VCH, Weinheim, 1999.
[2] A. Baiker, Chem. Rev. 99 (1999) 453.
[3] P.G. Jessop, T. Ikariya, R. Noyori, Chem. Rev. 99 (1999) 475.
[4] A.D. Kasnevich, E.J. Beckman, Chem. Today June 2004 (2004) 38.
[5] E.J. Beckman, J. Supercrit. Fluids 28 (2004) 121.
[6] G. Musie, M. Wei, B. Subramaniam, D.H. Busch, Coord. Chem. Rev. 219
(2001) 789.
[42] K. Morishige, M. Ito, J. Chem. Phys. 117 (2002) 8036.
[43] M.S. Schneider, J.-D. Grunwaldt, A. Baiker, Langmuir 20 (2004) 2890.
[44] T. Mallat, Z. Bodnar, P. Hug, A. Baiker, J. Catal. 153 (1995) 131.
[45] C. Keresszegi, T. Bürgi, T. Mallat, A. Baiker, J. Catal. 211 (2002) 244.
[46] D. Walther, G. Maurer, Ber. Bunsen.-Ges. Phys. Chem. 96 (1992) 981.
[47] C. Keresszegi, D. Ferri, T. Mallat, A. Baiker, J. Catal. 234 (2005) 64.
[48] C. Keresszegi, T. Mallat, A. Baiker, New J. Chem. 25 (2001) 1163.
[49] J. Chrastill, J. Phys. Chem. 86 (1982) 3016.
[7] J.R. Hyde, P. Licence, D. Carter, M. Poliakoff, Appl. Catal. A 222 (2001)
119.
[8] H. Jin, B. Subramaniam, Chem. Eng. Sci. 58 (2003) 1897.
[9] J.-D. Grunwaldt, R. Wandeler, A. Baiker, Catal. Rev.-Sci. Eng. 45 (2003)
1.
[10] S. Campestrini, U. Tonnellato, Curr. Org. Chem. 9 (2005) 31.
[11] R. Wandeler, A. Baiker, CATTECH 4 (2000) 128.
[12] J.-D. Grunwaldt, A. Baiker, Phys. Chem. Chem. Phys. 7 (2005) 3526.
[13] M. Besson, P. Gallezot, Catal. Today 57 (2000) 127.
[14] J.H.J. Kluytmans, A.P. Markusse, B.F.M. Kuster, G.B. Marin, J.C.
Schouten, Catal. Today 57 (2000) 143.
[50] I.G. Aguilar, P.R. Rey, C. Baluja, Ozone Sci. Eng. 23 (2001) 177.
[51] D.G. Lee, T. Chen, Z. Wang, J. Org. Chem. 58 (1993) 2918.
[52] S. Lai, D.G. Lee, Synthesis 11 (2001) 1645.
[53] D. Walther, G. Maurer, J. Chem. Eng. Data 38 (1993) 247.
[54] P. Vinke, H.E. van Dam, H. van Bekkum, Stud. Surf. Sci. Catal. 55 (1990)
147.
[55] C.-C. Guo, Q. Liu, X.-T. Wang, H.-Y. Hu, Appl. Catal. A 282 (2005) 55.
[56] K.M. Dooley, F.C. Knopf, Ind. Eng. Chem. Res. 26 (1987) 1910.
[57] C. Keresszegi, J.-D. Grunwaldt, T. Mallat, A. Baiker, J. Catal. 222 (2004)
268.
[58] H.E. van Dam, A.P.G. Kieboom, H. van Bekkum, Appl. Catal. 33 (1987)
361.
[15] J. Muzart, Tetrahedron 59 (2003) 5789.