Supported palladium nanoparticles on hybrid mesoporous silica: Structure/activity-relationship in the aerobic alcohol oxidation using supercritical carbon dioxide

Article abstract of DOI:10.1016/j.jcat.2008.07.002

The preparation, characterization, and catalytic properties of Pd nanoparticles supported on mesoporous organic-inorganic hybrid materials are described for continuous-flow aerobic oxidation of alcohols using supercritical carbon dioxide (scCO₂) as a mobile phase. The nanoparticles were generated "bottom-up" from molecular precursors that were precoordinated to the support through suitable anchor units. The most active material allows high single-pass conversions in scCO₂ at temperatures as low as 60 °C. This high activity may be associated with the presence of small primary crystallites (approx. 2 nm) that conglomerate to ensembles about 25 nm in size, leading to a larger number of high-indexed planes in small volume units. These findings may provide useful guidelines for further catalyst design on the nanoscale for green oxidation methods.

Full text of DOI:10.1016/j.jcat.2008.07.002

Journal of Catalysis 258 (2008) 315–323
Contents lists available at ScienceDirect
Journal of Catalysis
www.elsevier.com/locate/jcat
Supported palladium nanoparticles on hybrid mesoporous silica:
Structure/activity-relationship in the aerobic alcohol oxidation using supercritical
carbon dioxide
Zhenshan Hou a, Nils Theyssen a, Axel Brinkmann a, Konstantin V. Klementiev b,c, Wolfgang Grünert b,
Michael Bühl a, Wolfgang Schmidt a, Bernd Spliethoff a, Bernd Tesche a, Claudia Weidenthaler a, Walter Leitner a,d,∗
a
Max-Planck-Institut für Kohlenforschung, Kaiser-Wilhelm-Platz 1, D-45470 Mülheim an der Ruhr, Germany
Lehrstuhl für Technische Chemie, Ruhr-Universität Bochum, Universitätsstr. 150, 44801 Bochum, Germany
Hasylab at DESY, D-22607 Hamburg, Germany
Institut für Technische und Makromolekulare Chemie, Lehrstuhl für Technische Chemie und Petrolchemie, RWTH Aachen, Worringerweg 1, D-52064 Aachen, Germany
b
c
d
a r t i c l e i n f o
a b s t r a c t
Article history:
The preparation, characterization, and catalytic properties of Pd nanoparticles supported on mesoporous
organic–inorganic hybrid materials are described for continuous-flow aerobic oxidation of alcohols using
supercritical carbon dioxide (scCO2) as a mobile phase. The nanoparticles were generated “bottom-up”
from molecular precursors that were precoordinated to the support through suitable anchor units. The
Received 8 April 2008
Revised 3 July 2008
Accepted 3 July 2008
Available online 25 July 2008
◦
most active material allows high single-pass conversions in scCO2 at temperatures as low as 60 C. This
high activity may be associated with the presence of small primary crystallites (approx. 2 nm) that
conglomerate to ensembles about 25 nm in size, leading to a larger number of high-indexed planes
in small volume units. These findings may provide useful guidelines for further catalyst design on the
nanoscale for green oxidation methods.
Keywords:
Palladium
Nanoparticles
Supercritical carbon dioxide
Alcohol oxidation
Mesoporous support
Hybrid materials
© 2008 Elsevier Inc. All rights reserved.
1. Introduction
the same time, scCO2 provides a safe reaction environment with
excellent mass and heat transport properties for aerobic oxida-
tions [40].
The selective oxidation of alcohols to the corresponding alde-
hydes or ketones is an important transformation for organic syn-
thesis on laboratory and industrial scales [1–3]. Transition metal-
catalyzed oxidations using molecular oxygen have been investi-
gated intensively as green alternatives to replace the more tra-
ditional stoichiometric procedures, which generate large amounts
of waste and require inefficient purification procedures [4–7]. In
particular, palladium-based systems in the form of supported het-
erogeneous catalysts, nanoparticles, and molecular palladium com-
plexes have been developed for aerobic alcohol oxidation and
found to convert various alcohols effectively and selectively to the
corresponding aldehydes and ketones [8–25]. In principle, the use
Recently, we have developed an efficient system for continuous-
flow aerobic alcohol oxidation using palladium nanoclusters that
were stabilized and immobilized in a liquid poly(ethylenglycol)
(PEG) matrix in combination with supercritical carbon dioxide
(scCO2) as the mobile phase [30]. This approach has the potential
to solve some of the problems typically associated with homoge-
neously dispersed nanoparticles, such as deactivation by formation
of Pd-black and difficulties in catalyst separation and recycling.
However, a stable well-defined Pd-nanoscale catalyst based on a
solid and mainly inorganic matrix seems to be even more de-
sirable, because it would allow the use of a simple and highly
efficient fixed-bed technology and thereby preclude any risk of de-
pletion or oxidative degradation of the matrix [32,41,42].
◦
of supercritical carbon dioxide (scCO2, Tc = 31.0 C, pc = 7.4 MPa)
[26–29] allows the implementation of such oxidations in a quasi-
solvent-free continuous-flow mode at temperatures far below the
boiling points of the organic substrates and products [30–39]. At
Herein we report that organic–inorganic hybrid materials, based
ꢀ
on mesoporous silica as a support and 2,2 -dipyridylamine as a
linker, allow the efficient immobilization and stabilization of pal-
ladium nanoparticles of defined size and aggregation for catalytic
aerobic alcohol oxidation using scCO2 as a reaction medium un-
der batchwise and continuous-flow, fixed-bed conditions (Fig. 1).
Recently, similar materials have been developed for applications in
conventional solvents [42–44]. In addition to the development of
Corresponding author at: Max-Planck-Institut für Kohlenforschung, Kaiser-
Wilhelm-Platz 1, D-45470 Mülheim an der Ruhr, Germany. Fax: +49 (0) 241 80 221
*
77.
E-mail address: leitner@itmc.rwth-aachen.de (W. Leitner).
0021-9517/$ – see front matter © 2008 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2008.07.002

316
Z. Hou et al. / Journal of Catalysis 258 (2008) 315–323
ethanol for 20 h. Before use, the extract was dried under vacuum
◦
at 80 C for 14 h.
2.1.3. Synthesis of 3
To a Schlenk-tube filled with 2.5 g of 2, 0.65 g of Pd(OAc)2
and 100 mL of dry acetone were added and stirred for 24 h. The
resulting suspension was filtered and washed with acetone and
◦
dichloromethane, then dried at 80 C for 14 h. The resulting prod-
uct was a yellow powder.
2.1.4. Synthesis of 4
First, 2 g of 3 was conditioned for 17 h by refluxing it in a 50-
mL mixture of toluene and benzyl alcohol (80:20, V/V). The result-
ing solid was then washed completely with toluene and acetone,
and then dried at 80 C for 12 h to obtain a dark-green powder.
Elemental analysis: 2.22% N and 7.02% Pd.
Fig. 1. Schematic drawing of the general reaction concept.
a quasi-solvent-free continuous-flow process, the present study in-
cludes a detailed analysis of the supported nanoscale particles. The
combined results indicate that the chosen reduction method of the
Pd(II) precursor (benzyl alcohol vs hydrogen) exerts a controlling
effect on the texture and oxidation state distribution of the ob-
tained palladium species, resulting in very significant differences
in the catalytic activity.
◦
2.1.5. Synthesis of 5
First, 0.7 g of 3 was loaded into an autoclave, then reduced for
◦
4 h at 40 bar H2 and 110 C. After reduction, the hydrogen was
released slowly from the autoclave, resulting in a brown powder.
2. Experimental
2.2. Characterization of the synthesized materials by XRD, BET area,
TEM, 29Si MAS NMR, XPS, and XAFS
2.1. Synthesis of materials
All organic solvents used in this work were dried by standard
procedures. All of the syntheses were performed under argon. 1H
NMR and 13C NMR spectra were recorded on a Bruker DPX 300
instrument. Pd(OAc)2, 5% Pd/C, and Pd%/Al2O3 were obtained from
Aldrich and used as received.
The powder X-ray diffraction (XRD) patterns for qualitative
phase analysis were collected on a Stoe STADI P transmission
diffractometer in transmission geometry with a primary monochro-
mator curved germanium (111) and a linear position sensitive
detector, with CuKα1: 1.54060 Å as a radiation source. The data
were collected in the range of 0–10 2θ with a step width of 0.01
2θ. For the measurements, the sample was prepared between two
polyacetate foils.
Nitrogen sorption isotherms were measured at 77 K with a
Quantachrome Instruments Nova 3000e sorption analyzer. Before
the measurements, the samples were evacuated at 393 K for 8 h.
Transmission electron microscopy (TEM) and energy-dispersive X-
ray analysis (EDX) were used to investigate structural features of
the catalysts with a Hitachi HF-2000 instrument. 29Si MAS NMR
spectroscopy was recorded on a Bruker Avance 500WB instrument
using a 4-mm MAS probe at a spinning rate of 10 kHz, 30 s recycle
delay, 2800 scans, and 2.2 μs π/4 pulse.
XPS measurements were performed with a Kratos HSi spec-
trometer with a hemispherical analyzer and a monochromatized
AlKα X-ray source (E = 1486.6 eV), operated at 15 kV and 15 mA.
For the narrow scans, an analyzer pass energy of 40 eV was ap-
2.1.1. Synthesis of 1
◦
◦
ꢀ
Commercially available 2,2 -dipyridylamine (1.0 g, 5.84 mmol)
was dissolved in 15 mL of dry THF. This solution was added to a
suspension of NaH (0.14 g, 5.85 mmol) in 50 mL of dry THF. The
◦
mixture was stirred under argon at 0 C for 1 h and then added
to a solution of 3-iodopropyltriethoxysilane (synthesized from 3-
chloropropyltriethoxysilane; 1.95 g, 5.92 mmol) in 10 mL of dry
THF. The resulting reaction mixture was stirred at room tempera-
ture under argon for 16 h. The solvent was evaporated under high
◦
vacuum at 35 C, and the resulting residue was taken up again in
toluene (100 mL). After filtration of the salts and washing with
toluene (3 × 10 mL), the filtrate was concentrated under vacuum,
resulting in a yellow oil (1.71 g, 4.52 mmol, 76.5%). 1H NMR (δ,
300 MHz, CDCl3): 0.48–0.53 ppm (m, 2H, CH2Si), 1.63–1.85 ppm
(m, 2H, CH2), 3.68 ppm (q, 6H, OCH2), 1.06 ppm (t, 9H, CH2CH3),
3.96 ppm (t, 2H, NCH2), 6.51–8.24 ppm (m, 8H, Ar H). 13C NMR
(δ, 75 MHz, CDCl3): 6.34 ppm (CH2Si), 23.9 ppm (CH2), 49.84 ppm
(CH2N), 57.78 ppm (OCH2), 18.18 ppm (CH2CH3), 113.0 (Ar), 117.14
(Ar), 138.60 (Ar), 148.76 (Ar), 156.01 (Ar). MS (EI): m/z = 376
plied. The hybrid mode was used as lens mode. The base pressure
−10
in the analysis chamber was 6 × 10
Torr. To account for charg-
+
+
ing effects, all spectra are referred to Si 2p at 103.45 eV.
[M + H ], 163 [Si(OC2H5)3 ]. Elemental analysis: Calculation for
XAFS measurements were carried out at the X1 station of HA-
SYLAB (DESY, Hamburg, Germany) using a double-crystal Si(311)
monochromator. The energy resolution was estimated from the an-
gular beam size as 7.1 eV (standard deviation) at the Pd K-edge.
Samples were pressed into discs of suitable thickness in a glove
box, and the discs were sealed air-tight in Kapton tape before be-
ing inserted into the XAFS experiment. The spectra were recorded
in transmission mode at low temperature (T = 80 K) to decrease
the Debye–Waller factors, and a Pd foil was measured simulta-
neously (between the second and third ionization chambers) for
energy calibration purposes. Details of the analysis procedure are
given in the ESI section.
C
19 H29N3O3Si: C: 60.77, H: 7.78, N: 11.19, O: 12.78, Si: 7.48; Found:
C: 60.44; H: 7.65, N: 10.88, Si: 7.36.
2.1.2. Synthesis of 2
The hybrid material was synthesized by co-hydrolysis and poly-
condensation of mixtures 1 and tetraethylorthosilicate (TEOS) in
the presence of n-hexadecylamine as a template. The molar com-
position of the mixture was 0.88 TEOS:0.12 1:0.28 n-hexadecyl-
amine:29.6 H2O:9.1 ethanol. The xerogel was prepared as de-
scribed previously [45]. The reaction mixture was stirred for 24 h.
The resulting solid was filtered and washed with ethanol. Surfac-
tant was removed from the product by Soxhlet extraction with

Z. Hou et al. / Journal of Catalysis 258 (2008) 315–323
317
Table 1
Key structural and analytical parameters for 2–5, together with catalytic data
2
3
4
5
Reduction
Color
–
None
Yellow
699
n.d.
Benzyl alcohol
Greenish
688
2.6
Hydrogen
Brown
677
n.d.
7.06
Colorless
−1
a
BET (m2
g
)
990
3.5
–
Pore size (nm)a
Pd-content (wt%)
7.12
7.01
N-content (wt%)
2.43
–
–
–
–
2.32
Pd(II) (ca. 100%)
Pd(II) (ca. 100%)
3%
60
98.7%
2.22
2.18
Oxidation state of bulk Pd (XAFS)
Oxidation state of surface Pd (XPS)
Single pass conversion of 6ab
TON (turnover number)
Selectivity for 7ab
Pd(0)/Pd(II), (33%:67%)
Pd(0)/Pd(II), (55%:45%)
41%
800
Pd(0)/Pd(II), (29%:71%)
Pd(0)/Pd(II) (84%:16%)
9%
180
98.6%
–
98.2%
a
For details of the sorption measurement see Supporting information; n.d. = not determined.
b
Average values over 28 h on stream, cf. Fig. 6.
2.3. Typical procedure for the alcohol oxidation with functionalized
hybrid catalysts under supercritical carbon dioxide
the preparation.) The hydrolysis/condensation of 1 with tetraethy-
lorthosilicate (TEOS) in the presence of n-hexadecylamine led to
the functionalized mesoporous material 2. After the template was
removed by extraction with ethanol, the product was treated with
an acetone solution of palladium acetate to give the silica bound
coordination compound 3 (Pd loading, 7.1%). Finally, the material
was reduced either with benzylic alcohol under reflux conditions
or with molecular hydrogen to yield the Pd-containing mesoporous
silica materials 4 and 5, respectively (Scheme 1). The prepara-
tion methods led to distinct materials with high reproducibility as
demonstrated by the excellent concordance of the XAFS spectra of
independent synthesized and measured samples (see Supporting
information).
Safety warning: The use of compressed gases and especially O2
in the presence of organic substrates requires appropriate safety
precautions and must be carried out using suitable equipment.
Batchwise catalytic oxidation of alcohol in scCO2: 75 mg of cata-
lyst 4 and 2.50 mmol alcohol (n(substrate)/n(Pd) = 50) was placed
in a 36 mL stainless-steel high-pressure reactor. The reactor was
pressured with a defined CO2/O2 mixture (19.5 g, c(O2) = 8 vol%)
and heated under vigorous stirring to the desired temperature for
a specified time (see Table 1). The total pressure was 18 MPa. Af-
◦
ter reaction, CO2 was released slowly through two −35 C serial
cold traps. The products were collected by washing the cold traps
and the autoclave with diethyl ether. The combined fractions were
then analyzed by 1H NMR and GC.
Table 1 presents information on the texture, composition, oxi-
dation state, and catalytic properties of the synthesized materials.
The XRD pattern of the colorless hybrid material 2 shows a sin-
Continuous-flow, fixed-bed catalytic oxidation of alcohol in scCO2:
The reactions were performed in a fixed-bed stainless steel tubu-
lar reactor with an inner diameter of 7.5 mm and 20 mL volume
(see Fig. 1). The temperature of the catalyst bed was controlled by
gle diffraction peak corresponding to a d-spacing of 4.1 nm. The
−1
N2 isotherms indicate a BET area of 990 m2 g
and confirm the
mesoporous structure with pore diameters around 3.5 nm [54–63].
These measurements, together with HRTEM micrographs of the
material, clearly indicate a wormhole pore structure. From the 29Si
MAS NMR spectra, the resonances near −100 and −110 ppm rep-
resent the Q3 HOSi(OSi)3 and Q4 Si(SiO)4 environments of the SiO4,
whereas the signals around −66 ppm arise from T3 connectivities
of the organic-functionalized silicon centers (see Supporting infor-
mation for details).
The structure of the mesoporous material 2 remained largely
unperturbed on incorporation of the Pd centers, but BET area
and pore sizes became somewhat smaller compared with the or-
ganic functionalized hybrid materials, as expected. The XRD pat-
terns showed no visible reflections at higher angles, indicating the
absence of long-range ordering. Differences in the Pd-containing
materials 3–5 were macroscopically indicated by their different
colors. The silica-bound coordination compound 3 was bright yel-
low, whereas 5 was a brown powder. Material 4 appeared greenish,
somewhere between the two other solids.
◦
a pumped thermal oil fluid, which kept the reactor at 60 C. Pure
quartz (80 mesh, 3.5 g) was used to dilute the catalyst bed (0.2 g
catalyst). A glass bead was placed above the catalyst bed to ensure
an optimal distribution of reactants at the beginning of the cata-
◦
lyst bed. A prewarmed (50 C) continuous stream of CO2/O2 was
passed through the tubular reactor at a flow rate of approximately
−1
7.5 l h
(exit flow at ambient conditions; total pressure, 15 MPa).
The substrate was fed into the CO2 stream before it entered the
reactor with a HPLC pump at a rate of 1.0 mL h−1. The reaction
◦
mixture was isolated in two sequential cold traps at −35 C from
the exit flow on depressurization of the supercritical solvent. The
cold traps was replaced periodically after about 2 h, and their
combined content was analyzed by 1H NMR and GC. Details on
the experimental setup are given in the Supporting information
(Fig. 5S).
3. Results and discussion
XPS analysis demonstrated that 3 contained exclusively Pd(II)
centers, as expected; however, two distinct signals indicated
the presence of two surface-bound species (Fig. 2a). Adsorbed
[Pd(OAc)2]3, or PdO, can be excluded as sources for these signals,
because no corresponding intensities were found in the EXAFS
spectra (Fig. 3b). The signal at lower binding energy (336.9 eV,
37 at%) can be assigned to a Pd environment with two oxygen
and two nitrogen ligands. The other signal (338.4 eV, 63 at%)
most likely resulted from a pure oxidic environment. These data
are most consistent with the formation of a trinuclear cluster
[(AcO)2Pd(μ-AcO)2Pd(μ-AcO)2Pd(N–N)–silica] as the main species
(theoretical: Pd[O]4: 66%, Pd[O]2[N]2: 33%) on the surface, likely
together with small amounts of a mononuclear coordination com-
pound of type [(AcO)2Pd(N–N)–silica]. These assignments are in
3.1. Catalyst preparation and characterization
In previous studies, Pd nanoparticles have been generated on
inorganic supports by impregnation (on, e.g., TiO2 [25], hydroxya-
patite [16], or active carbon [46] and by chemical vapor deposition
[47], or encapsulation/decomposition [48,49]) on MCM-41. We fol-
lowed a different technique, starting from molecular precursors
that were incorporated through organofunctional groups in the in-
organic framework during preparation of the mesoporous material
(one-pot method; Scheme 1) [42,50–53]. The trialkoxylsilyl deriva-
ꢀ
tive of 2,2 -dipyridylamine 1 was used as a coordinative linker
for the Pd-precursor. (See Supporting information for details of

318
Z. Hou et al. / Journal of Catalysis 258 (2008) 315–323
Scheme 1. Synthesis of functionalized organic–inorganic hybrid materials 4 and 5 containing Pd-nanoparticles on mesoporous silica.
Fig. 2. XPS analysis (Pd 3d spectra) for the surface-bound Pd-centers in materials 3 (a), 4 (b), and 5 (c). Species I and II are assigned to Pd(II) centers, whereas species III
results from Pd(0). See text for details.
line with calculated charges on Pd for appropriate model com-
pounds [64]. Based on the ratio of linker:Pd from elemental anal-
ysis, it may be concluded that ca. 55% of the linker remained
noncoordinated.
The XPS spectra of the material 5 indicate that the Pd(II) cen-
ters were reduced to a great extent by hydrogen treatment, result-
ing in 84% of Pd(0)-centers at the surface (Fig. 2c). In contrast, ma-
terial 4 contained significant amounts of unreduced palladium sites
corresponding to 45% Pd(II) (337.8 eV) and 55% Pd(0) (335.4 eV)
at the surface (Fig. 2b). Thus, hydrogen treatment appeared to be
more efficient for the reduction of surface-bound Pd-centers than
benzyl alcohol reduction. To obtain information on the metal parti-
cles not only at the surface, but also throughout the bulk material,
we performed a XAFS study to assess the overall oxidation state of

Z. Hou et al. / Journal of Catalysis 258 (2008) 315–323
319
Fig. 3. XAFS spectra of material 3–5 and reference compounds Pd foil, PdO and Pd acetate. (a) XANES, (b) EXAFS (Fourier transformation of k2-weighted spectra).
Pd and to gain insight into the local environment of the particles
inside the mesopores of the materials 3–5.
a similar manner as for material 4 (see Supporting information)
resulted in Pd compositions of 70.7% in unconverted precursor 3
and 29.3% in Pd nanoparticles, significantly smaller than those re-
quired for the best fit with material 4 (nanoparticles in material 5:
coordination number—6.6 0.6, diameter—0.7–1.1 nm). It should
be noted that the attempt to reproduce the EXAFS spectrum as a
linear combination of the component spectra resulted in a less sat-
isfactory agreement in material 5, indicating that the sample may
contain another, highly dispersed Pd(0) phase.
These XAFS measurement results indicate incomplete reduction
for the bulk material in both cases, but also suggest the presence
of significantly larger particles with partially metallic character
in material 4. A combination of TEM and EDX analysis provided
further insight into the nature of the two materials in terms of
nanoparticle fraction. In material 4, large nanoparticulate units
were observed, consisting of primary crystallites with an aver-
age size of 2.4 nm (in fair agreement with the 1.5 nm obtained
from XAFS) that aggregate into conglomerates with typical diam-
eters of about 25 nm (Fig. 5A, a–c). Due to the restricted inner
width of the mesoporous channels (about 4 nm), it is obvious that
these ensembles are present exclusively at the external surfaces.
No such large secondary particles were found in material 5, ob-
tained by hydrogen reduction. Most of the particles at the external
surface appeared to be primary particles, with an average diameter
of 6.5 nm (Fig. 5B, a–c). This value is almost an order of magni-
tude larger than that derived from XAFS data, further supporting
the conclusion that 5 contained an additional Pd(0) phase of much
higher dispersion that was not visible on TEM.
Fig. 3 shows XANES spectra together with the Fourier-trans-
formed EXAFS for materials 3–5 together with Pd-containing refer-
ence materials. In full accordance with the XPS data, the features
of material 3 are clearly distinct from PdO. A precise analysis of
the spectra from [Pd(OAc)2]3 also revealed explicit differences, in-
dicating that the presence of significant amounts of [Pd(OAc)2]3
was highly unlikely. Overall, the data are fully consistent with the
trinuclear coordination mode [(AcO)2Pd(μ-AcO)2Pd(μ-AcO)2Pd(N–
N)–silica] derived from the XPS data.
After reduction with benzyl alcohol, a signal for the first Pd–
Pd coordination sphere in metallic palladium was clearly visible in
the EXAFS spectrum (albeit with very small amplitude), along with
a signal for the residual coordination of Pd with a light element
(ca. 1.6 Å, uncorrected) in material 4. Target transformation analy-
sis of the XANES region revealed the presence of two components,
identified as precursor 3 and Pd nanoparticles, the data for which
were obtained from an independent study on Pd/C catalysts [65–
67]. These (carbon-supported) Pd nanoparticles exhibited a Pd–Pd
coordination number of 8.0 0.6, corresponding to a particle di-
ameter of 1.0–1.5 nm as derived from a geometrical correlation
between coordination number and (spherical) particle size [68].
The linear fit of the XANES region (Fig. 4) yielded a composition
corresponding to 32.6% of Pd in nanoparticles and 67.4% in un-
converted precursor 3, which directly corresponds to the degree
of reduction of the bulk material. Using this same ratio, we could
closely reproduce the EXAFS spectra by superimposing the spectra
measured with 3 and with the Pd nanoparticles [66,67] (Fig. 4),
suggesting that these were the only significant Pd phases present
in material 4.
3.2. Catalytic performance
Comparing the data from XPS and the analysis of the XAFS
measurements reveals that the palladium in the bulk of material
4 was reduced significantly less than the surface-exposed centers
(cf. Table 2). This difference was even more pronounced in mate-
rial 5, obtained from hydrogen reduction, for which XPS indicated
a reduction degree of 84%. The EXAFS spectrum of 5 showed con-
siderable resemblance to that of precursor 3, but with significant
differences in the XANES region (Fig. 3). Analyzing the data in
To assess the catalytic performance of the new hybrid mate-
rials 3–5, we evaluated then in the aerobic alcohol oxidation of
benzyl alcohol (6a) to give benzaldehyde (7a) using supercritical
carbon dioxide (scCO2) as the mobile phase under continuous-flow
conditions similar to those developed for PEG-stabilized Pd561 clus-
ters (Scheme 2; Fig. 1; see Supporting information for details) [30,
32]. Due to scCO2’s unique combination of gas-like mixing and
liquid-like solvent properties, a homogeneous feed of CO2, O2, and

320
Z. Hou et al. / Journal of Catalysis 258 (2008) 315–323
Fig. 4. XANES fit for material 4 and the resulting EXAFS.
Table 2
a
Aerobic oxidation of alcohols 6a–e catalyzed by 4 in scCO2
Entry
Substrate
Product
T
( C)
t
(h)
Conversion
(%)
Selectivity
(%)
◦
1
60
12
5
92.5
99.5
97.8
98.5
98.4
89.8
96.4
99.6
2
3
4
80
80
60
7
8
5
6
80
80
80
4
18
8
99.2
92.6
11.5b
92.6
98.7
93.0c
7
a
General reaction conditions: m(catalyst 4) = 75 mg, n(substrate) = 1.65 mmol, d(CO2/O2) = 0.52 g mL−1, c(O2) = 8 vol%.
Yield 8 = 5.5%, yield butyl butyrate = 5.2%, yield butanal = 0.8%.
b
c
Selectivity towards 8 and its butyl ester.

Z. Hou et al. / Journal of Catalysis 258 (2008) 315–323
321
(A)
(B)
Fig. 5. (A) TEM micrographs of material 4 before using it as a catalyst in the alcohol oxidation reaction. The magnification increases from (a) to (c). (B) TEM micrographs of
material 5 before using it as a catalyst in the alcohol oxidation reaction. The magnification increases from (a) to (c).
of products, avoiding the formation of overoxidized side products.
In addition, a low operation temperature would be expected to in-
crease the lifetime of the catalyst material, avoiding deactivation
processes. However, all of these beneficial properties can be ex-
ploited only with the proper choice of catalyst material.
Fixed-bed columns were prepared from the catalyst materials
mixed homogeneously with quartz for better mechanical stability
(catalyst:quartz ratio = 17.5:1). Short paths of glass were used to
Scheme 2. Aerobic alcohol oxidation with Pd-based mesoporous catalysts 3–5 under
supercritical conditions. For definition of 6a–e/7a–e and other products see Table 2.
seal the beds on either side of the column. The rest of the setup
was identical to that described previously [30,32]. Such parame-
ters as O2/substrate/CO2 ratio, system pressure, and flow rate were
6a–e/7a–e can be transported in a gas-like manner through the
fixed at standard conditions to benchmark the performance of the
catalyst bed at temperatures far below the boiling point of the sub-
strate and products [26–30,32–37]. Furthermore, scCO2 has been
found to enhance the rate of mass transfer inside the channel of
porous materials [69,70]. This behavior can facilitate rapid removal
◦
various catalyst materials. The temperature was set at 60 C, be-
cause preliminary screening indicated significant activity for mate-
rial 4 already at that value. This temperature is considerably lower

322
Z. Hou et al. / Journal of Catalysis 258 (2008) 315–323
increase in single-pass conversion to 65%, but palladium was de-
tected at around 5 ppm in the slightly yellow product, and the
selectivity was reduced to 90%.
To assess the possible scope of the continuous alcohol oxida-
tion in scCO2, catalyst 4 was evaluated in batchwise experiments
using structurally different alcohols 6a–e (Table 2). In the temper-
◦
ature range of 60–80 C, benzylic (6a), allylic (6b–c), and cyclic
aliphatic (6d) alcohols were converted in high yields into the
corresponding carbonyl compounds within similar reaction times
(entries 1–6), indicating that comparable space-time yields could
◦
be achieved under continuous-flow conditions. Operating at 60 C
generally gave somewhat higher selectivity, especially in the pres-
ence of other potentially oxidizable groups (entries 1/2 and 4/5). In
general, the reactions could be carried out at lower temperatures
or reduced reaction times compared with earlier systems based
on PEG-supported Pd561 clusters [30,32]. At the same time, the
overoxidation of primary alkanols, such as 1-butanol (6e), which
led to the corresponding acid (8) in free form and in form of the
corresponding butyl ester, was largely suppressed with catalyst 4.
Fig. 6. Continuous-flow fixed-bed aerobic oxidation of benzyl alcohol (6a) in scCO2
◦
with different catalysts; m(catalyst) = 0.20 g, T = 60 C, p(CO2/O2: 92/8) = 15 MPa,
−1
flow rate = 1 mL h−1, exit flow (normal pressure) = 7.5 L h
.
4. Conclusion
Here we have shown that the immobilization of palladium nan-
oclusters on organic/inorganic hybrid mesoporous silicates led to
highly active catalysts for aerobic alcohol oxidation in scCO2 under
batchwise and continuous-flow, fixed-bed conditions. The combi-
nation of a highly active solid catalyst material with scCO2 as a
mobile phase allowed a reaction engineering setup resembling a
typical gas-phase process at temperatures far below the boiling
point of the substrates and products. Together with the nature of
the support, this is believed to contribute to the catalysts’ high sta-
bility by preventing particle growth and metal leaching.
Application of this strategy requires a catalyst material of suffi-
ciently high activity. In this respect, the performance of material 4
seems very promising. Most notably, 4 showed significantly higher
activity than material 5, which is identical in terms of porous
structure, Pd loading, and bulk oxidation state. Detailed analysis of
the materials revealed two major differences that may be related
to the different activities. First, material 4 contained more unre-
duced Pd(II) centers at the surface. However, no evidence for an
interaction between the Pd(0) and Pd(II) centers was obtained in
this study, and the isolated Pd(II) centers were found to be largely
inactive in material 3 and in physical mixtures of 3 and 5 under
the present conditions [33,37,71,72]. Second, there was a signifi-
cant difference in the nature of the Pd(0) phases present in the
two materials [9]. The hydrogen reduction used for the preparation
of material 5 yielded medium-sized (approx. 6 nm) primary parti-
cles of very regular shape together with a second, highly dispersed
Pd(0) phase. Although this provided many accessible surface Pd
centers, it apparently was not the ideal situation for high activity
[73]. In contrast, the alcoholic reduction method for sample 4 re-
sulted in a texture containing small primary crystallites (ca. 2 nm)
that conglomerated to larger units of about 25 nm. Because of the
many different crystalline orientations in such ensembles, many
high-indexed planes are present in small volume units. The re-
sulting high number of accessible edges and corners provide many
potential sites for catalytically active Pd(0)-centers, giving a plau-
sible explanation for the high activity of material 4 [74,75].
Fig. 7. TEM micrograph of material 4 after using it as a catalyst in the oxidation of
6a (after 24 h on stream).
than that used with previous catalysts, which generally require
◦
◦
T ꢀ 80 C, in most cases ꢀ100 C.
As Fig. 6 shows, the commercial Pd/Al2O3 gave only negligible
single-pass conversion for benzyl alcohol 6a as a prototypical sub-
strate at 60 C under scCO2 continuous-flow conditions. Pd/C gave
◦
an initially significant conversion of around 12%, which rapidly
dropped to <5%. A similar marginal conversion was observed with
the heterogenized Pd(II) complex in material 3. Material 5, ob-
tained by hydrogen reduction, exhibited significant and stable con-
version around 9%. Finally, catalyst 4 showed a more than four-fold
greater activity and excellent stability in scCO2 under mild condi-
tions. On average, single-pass conversions of 41% and selectivities
◦
of >98% for 7a were achieved at temperatures as low as 60 C
and remained constant over more than 28 h on stream. Under
the nonoptimized continuous-flow conditions used in the present
study, these data correspond to a space-time yield of >100 g 7a
per liter of reactor volume per hour. Based on the actual catalyst
material, the productivity corresponds to 2.1 g of 7a per g of 4
−1
per hour (TON after 26 h = 800; TOF = 31 h
based on the total
amount of Pd).
TEM micrographs obtained for samples of material 4 recovered
from the reactor after 24 h on stream showed no indication of
significant further agglomeration of the Pd particles during the cat-
alytic process (compare Fig. 5A and Fig. 7). Furthermore, the level
of palladium in the colorless product was <1 ppm. These data sug-
gest negligible particle growth through Ostwald ripening and metal
leaching under these conditions. Clearly, the low-temperature ac-
tivity of the catalyst system, which allows operation of the contin-
The findings of the present study emphasize once again that the
application of unconventional reaction media, such as supercriti-
cal CO2, must be integrated with the development and molecular
understanding of suitable catalysts to exploit the potential ben-
efits offered by these advanced fluids. In the present study, the
reduction method used in nanoparticle preparation has a critical
influence on the size and agglomeration of the primary crystal-
lites, which in turn define the catalytic performance of the ma-
◦
uous process at 60 C, was beneficial in this respect. For example,
◦
increasing the reaction temperature with 7a to 80 C led to an

Z. Hou et al. / Journal of Catalysis 258 (2008) 315–323
323
terials. Interestingly, the data suggest that there is an optimum
balance between high surface area and surface defects, and thus
that nanoscale agglomerates can exhibit significantly higher ac-
tivities compared with isolated small primary particles in highly
dispersed phases. This may be a useful guideline for further cata-
lyst design on the nanoscale for green oxidation methods.
[33] J.-D. Grunwaldt, M. Caravati, A. Baiker, J. Phys. Chem. B 110 (2006) 9916.
[34] M. Caravati, D.M. Meier, J.-D. Grunwaldt, A. Baiker, J. Catal. 240 (2006) 126.
[35] M. Caravati, J.-D. Grunwaldt, A. Baiker, Phys. Chem. Chem. Phys.
278.
7 (2005)
[36] M. Caravati, J.-D. Grunwaldt, A. Baiker, Catal. Today 91–92 (2004) 1.
[37] J.-D. Grunwaldt, M. Caratati, M. Ramin, A. Baiker, Catal. Lett. 90 (2003) 221.
[38] S. Campestrini, M. Carraro, R. Ciriminna, M. Pagliaro, U. Tonellato, Adv. Synth.
Catal. 347 (2005) 825.
[39] R. Ciriminna, S. Campestrini, M. Pagliaro, Org. Biomol. Chem. 4 (2006) 2637.
[40] N. Theyssen, Z. Hou, W. Leitner, Chem. Eur. J. 12 (2006) 3401, and references
therein.
[41] B. Karimi, A. Zamani, J.H. Clark, Organometallics 24 (2005) 4695.
[42] B. Karimi, S. Abedi, J.H. Clark, V. Budarin, Angew. Chem. Int. Ed. 45 (2006) 4776.
[43] C. Li, Q. Zhang, Y. Wang, H. Wan, Catal. Lett. 120 (2008) 126.
[44] S.F.J. Hackett, R.M. Brydson, M.H. Gass, I. Harvey, A.D. Newman, K. Wilson, A.F.
Lee, Angew. Chem. Int. Ed. 46 (2007) 8593.
Acknowledgments
This work was supported by the Max-Planck-Gesellschaft and
the Fonds der Chemischen Industrie. The authors thank Dr. B. Zi-
browius (MPI für Kohlenforschung) for recording the 29Si MAS
NMR spectra.
[45] P.T. Tanev, T.J. Pinnavaia, Chem. Mater. 8 (1996) 2068.
[46] H. Bönnemann, G. Braun, W. Brijoux, R. Brinkmann, S.A. Tilling, K. Seevogel, K.
Siepen, J. Organomet. Chem. 520 (1996) 143.
[47] P. Mehnert, J.Y. Ying, Chem. Commun. (1997) 2215.
[48] M.D. Jones, M.J. Duer, S. Hermans, Y.Z. Khimyak, B.F.G. Johnson, J.M. Thomas,
Angew. Chem. Int. Ed. 41 (2002) 4726.
Supporting information
Further analytical characterizations using XRD, N2 sorption ex-
periments, 29Si MAS NMR spectroscopy and XAFS, and a schematic
presentation of the continuous-flow setup. This material is avail-
able free of charge at http://www.sciencedirect.com.
[49] B.F.G. Johnson, Top. Catal. 24 (2003) 147.
[50] M.H. Valkenberg, W.F. Hölderich, Catal. Rev. 44 (2002) 321.
[51] A.P. Wight, M.E. Davis, Chem. Rev. 102 (2002) 3589.
Please visit DOI: 10.1016/j.jcat.2008.07.002.
[52] D.E. De Vos, M. Dams, B.F. Sels, P.A. Jacobs, Chem. Rev. 102 (2002) 3615.
[53] G. Kickelbick, Angew. Chem. Int. Ed. 43 (2004) 3102.
References
[54] Y. Mori, T.J. Pinnavaia, Chem. Mater. 13 (2001) 2173.
[1] R.A. Sheldon, I.W.C.E. Arends, G.-J.T. Brink, A. Dijksman, Acc. Chem. Res. 35
(2002) 774.
[2] R.A. Sheldon, J.K. Kochi, Metal-Catalyzed Oxidation of Organic Compounds, Aca-
demic Press, New York, 1981, p. 350.
[3] B.M. Trost, I. Fleming, S.V. Ley, Comprehensive Organic Synthesis, Pergamon,
Oxford, 1991.
[55] W. Zhang, T.R. Pauly, T.J. Pinnavaia, Chem. Mater. 9 (1997) 2491.
[56] P.T. Tanev, T.J. Pinnavaia, Chem. Mater. 8 (1996) 2068.
[57] T.R. Pauly, T.J. Pinnavaia, Chem. Mater. 13 (2001) 987.
[58] C.C. Chen, E.J. McKimmy, T.J. Pinnavaia, K.F. Hayes, Environ. Sci. Technol. 38
(2004) 4758.
[59] Y. Liu, T.J. Pinnavaia, J. Mater. Chem. 14 (2004) 1099.
[60] S.S. Kim, T.R. Pauly, T.J. Pinnavaia, Chem. Commun. (2000) 835.
[61] R.J.P. Corriu, E. Lancelle-Beltran, A. Mehdi, C. Reye, S. Brandes, R. Guilard, J.
Mater. Chem. 12 (2002) 1355.
[62] T.J. Pinnavaia, D.B. Jackson, J.E.G. Mdoe, J.H. Clark, New J. Chem. 23 (1999) 539.
[63] R.J.P. Corriu, A. Mehdi, C. Reye, C. Thieuleux, Chem. Mater. 16 (2004) 159.
[64] Density functional calculations (RI-BP86/6-31G* level with Stuttgart–Dresden
pseudopotential on Pd) for the charge of the palladium atom with Natu-
[4] M.J. Schultz, M.S. Sigman, Tetrahedron 62 (2006) 8227.
[5] P.T. Anastas, J.C. Warner, Green Chemistry: Theory and Practice, Oxford Univer-
sity Press, Oxford, 1998.
[6] P.T. Anastas, M. Kirchhoff, Acc. Chem. Res. 35 (2002) 686.
[7] P.T. Anastas, R.L. Lankey, Green Chem. 2 (2000) 289.
[8] J. Muzart, Tetrahedron 59 (2003) 5789.
[9] B.-Z. Zhan, A. Thompson, Tetrahedron 60 (2004) 2917.
[10] B.A. Steinhoff, I.A. Guzei, S.S. Stahl, J. Am. Chem. Soc. 126 (2004) 11268.
[11] S. Shimazu, T. Uehara, A. Asami, T. Hara, N. Ichikuni, J. Mol. Catal. A: Chem. 282
(2008) 28.
[12] H.L. Wu, Q.H. Zhang, Y. Wang, Adv. Synth. Catal. 347 (2005) 1356.
[13] A. Corma, H. Garcia, A. Leyva, J. Mol. Catal. A: Chem. 230 (2005) 97.
[14] M.S. Kwon, N. Kim, C.M. Park, J.S. Lee, K.Y. Kang, J. Park, Org. Lett. 7 (2005)
1077.
[15] T. Mallat, A. Baiker, Chem. Rev. 104 (2004) 3037.
[16] K. Mori, T. Hara, T. Mizugaki, K. Ebitani, K. Kaneda, J. Am. Chem. Soc. 126 (2004)
10657.
[17] T. Iwasawa, M. Tokunaga, Y. Obora, Y. Tsuji, J. Am. Chem. Soc. 126 (2004) 6554.
[18] S. Paavola, K. Zetterberg, T. Privalov, I. Csöregh, C. Moberg, Adv. Synth.
Catal. 346 (2004) 237.
ral Population Analysis (NPA) give the following results: [Pd(OAc)
2]3: +0.80
[(AcO)2-Pd(η2-dma)] (mononuclear model compound for 3, dma = dipyridyl-
methylamin): +0.75 [(AcO)2Pd2(μ2-OAc)2Pd1(η2-dma)] (binuclear model
compound for 3): Pd2 = +0.79, Pd1 = +0.75 [(AcO)2Pd3(μ2-AcO)2Pd2(μ2-
OAc)2Pd1(η2-dma)] (open trinuclear model compound for 3): Pd3 and
Pd2
=
+0.81,
Pd1
=
+0.75
[Pd1(μ2-OAc)(AcO)Pd3(μ2-AcO)2Pd2(μ2-
OAc)2Pd1(η1-dma)] (closed trinuclear model compound for 3): Pd3 and
√
Pd2 = +0.81, Pd1 = +0.78.
[65] It is known that the XANES of small particles differs in amplitude from that of
a bulk material. Hence, we performed the linear XANES fitting using not the
spectrum of bulk Pd but spectra of Pd nanoparticles from an earlier study of
Pd/C catalysts, in which Pd(0) was the only palladium state detected [66]. At-
tempts to fit the XANES region of material 4 by a sum of the basis spectra of
Pd foil, PdO and [Pd(OAc)2]3 resulted in significantly less agreement between
measurement and models.
[19] T. Nishimura, S. Uemura, Synlett (2004) 201.
[20] U.R. Pillai, E. Sahle-Demessie, Green Chem. 6 (2004) 161.
[21] Y. Uozumi, R. Nakao, Angew. Chem. Int. Ed. 42 (2003) 194.
[22] G.-J.T.I. Brink, W.C.E. Arends, M. Hoogenraad, G. Vespui, R.A. Sheldon, Adv.
Synth. Catal. 345 (2003) 497.
[23] D.R. Jensen, M.J. Schultz, J.A. Mueller, M.S. Sigman, Angew. Chem. Int. Ed. 42
(2003) 3810.
[24] G. Kovtun, T. Kameneva, S.L. Hladyi, M. Starchevsky, Y. Pazdersky, I. Stoarov, M.
Vargaftik, I. Moiseev, Adv. Synth. Catal. 344 (2002) 957.
[25] K. Ebitani, Y. Fujie, K. Keneda, Langmuir 15 (1999) 3557.
[26] P.G. Jessop, W. Leitner (Eds.), Chemical Synthesis Using Supercritical Fluids,
Wiley–VCH, Weinheim, 1999.
[27] A. Baiker, Chem. Rev. 99 (1999) 453.
[28] P. Licence, J. Ke, M. Sokolova, S.K. Ross, M. Poliakoff, Green Chem. 5 (2003) 99.
[29] W. Leitner, Acc. Chem. Res. 35 (2002) 746.
[30] M. Caravati, J.-D. Grunwaldt, A. Baiker, Catal. Today 126 (2007) 27.
[31] Z. Hou, N. Theyssen, A. Brinkmann, W. Leitner, Angew. Chem. Int. Ed. 44 (2005)
1346.
[66] K.V. Klementiev, A.Yu. Stakheev, O.P. Tkachenko, N.N. Tolkachev, W. Grünert,
Hasylab Annual Report 2002.
[67] A.Yu. Stakheev, O.P. Tkachenko, G.I. Kapustin, N.S. Telegina, G.N. Baeva, T.R.
Brueva, K.V. Klementiev, W. Grünert, L.M. Kustov, Russ. Chem. Bull. 53 (2004)
528.
[68] M. Borovski, J. Phys. IV 7 (1997) 259.
[69] A.I. Cooper, Adv. Mater. 15 (2003) 1049.
[70] P.L. Dhepe, A. Fukuoka, M. Ichikawa, Phys. Chem. Chem. Phys. 5 (2003) 5565.
[71] J.-D. Grunwaldt, M. Caravati, A. Baiker, J. Phys. Chem. B 110 (2006) 25587.
[72] A.F. Lee, S.F.J. Hackett, J.S.J. Hargreaves, K. Wilson, Green Chem. 8 (2006) 549.
[73] F. Li, Q. Zhang, Y. Wang, Appl. Catal. A 334 (2008) 217.
[74] B. Ioan, A. Miyazaki, K.-I. Aika, Appl. Catal. B 59 (2005) 71.
[75] F. Maillard, S. Schreier, M. Hanzlik, E.R. Savinova, S. Weinkauf, U. Stimming,
Phys. Chem. Chem. Phys. 7 (2005) 385.
[32] Z. Hou, N. Theyssen, W. Leitner, Green Chem. 9 (2007) 127.