Formation of monometallic Au and Pd and bimetallic Au-Pd nanoparticles confined in mesopores via Ar glow-discharge plasma reduction and their catalytic applications in aerobic oxidation of benzyl alcohol

Article abstract of DOI:10.1016/j.jcat.2012.01.020

Successfully prepared via Ar glow-discharge plasma reduction, Au-Pd bimetallic nanoparticles were highly active in the selective oxidation of benzyl alcohol, showing a rate constant of 0.50 h^-1, which was 12.5 and 2× that of Au and Pd monometallic catalysts, respectively. Characterization analyses attributed the enhancement in both activity and selectivity to a Pd-rich shell/Au-rich core structure with abundant surface-coordination-unsaturated Pd atoms of those effectively confined and well-dispersed Au-Pd nanoparticles. As a green, efficient, and safe protocol, plasma reduction outperformed conventional H₂ thermal reduction due to the different particle nucleation and growth mechanism, which afforded modified morphology and surface chemistry of metal nanoparticles. Further oxidation and re-reduction of plasma-reduced Au-Pd catalyst resulted in the atomic rearrangement of nanoparticles, leading to inferior catalytic performance.

Full text of DOI:10.1016/j.jcat.2012.01.020

Journal of Catalysis 289 (2012) 105–117
Contents lists available at SciVerse ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Formation of monometallic Au and Pd and bimetallic Au–Pd nanoparticles
confined in mesopores via Ar glow-discharge plasma reduction and their
catalytic applications in aerobic oxidation of benzyl alcohol
a,b
a
a,
⇑
c
c
b
Yuanting Chen , Houpeng Wang , Chang-Jun Liu , Zhiyuan Zeng , Hua Zhang , Chunmei Zhou ,
b
b,
⇑
Xinli Jia , Yanhui Yang
a
Advanced Nanotechnology Center, School of Chemical Engineering and Technology, Tianjin University, Tianjin 300072, People’s Republic of China
School of Chemical and Biomedical Engineering, Nanyang Technological University, Singapore 637459, Singapore
School of Material Science and Engineering, Nanyang Technological University, Singapore 637879, Singapore
b
c
a r t i c l e i n f o
a b s t r a c t
Article history:
Successfully prepared via Ar glow-discharge plasma reduction, Au–Pd bimetallic nanoparticles were
highly active in the selective oxidation of benzyl alcohol, showing a rate constant of 0.50 h , which
was 12.5 and 2ꢁ that of Au and Pd monometallic catalysts, respectively. Characterization analyses attrib-
uted the enhancement in both activity and selectivity to a Pd-rich shell/Au-rich core structure with abun-
dant surface-coordination-unsaturated Pd atoms of those effectively confined and well-dispersed Au–Pd
Received 15 December 2011
Revised 30 January 2012
Accepted 30 January 2012
Available online 20 March 2012
ꢀ1
nanoparticles. As a green, efficient, and safe protocol, plasma reduction outperformed conventional H
2
Keywords:
Plasma
Gold
Palladium
Bimetallic nanoparticles
Benzyl alcohol oxidation
Mesoporous molecular sieve
thermal reduction due to the different particle nucleation and growth mechanism, which afforded mod-
ified morphology and surface chemistry of metal nanoparticles. Further oxidation and re-reduction of
plasma-reduced Au–Pd catalyst resulted in the atomic rearrangement of nanoparticles, leading to inferior
catalytic performance.
Ó 2012 Elsevier Inc. All rights reserved.
1
. Introduction
Tremendous efforts have been devoted to developing heteroge-
bimetallic catalysts is attributed not only to electronic interaction
between the two metals but also to the complex structure of bime-
tallic nanoparticles. Several studies tested different bimetallic
nanoparticle structures, such as core–shell structure [8], cluster-
in-cluster structure [9], and single-alloy structure [10]. Our group
reported on Au–Pd bimetallic nanoparticles with a Pd-cluster-on-
Au-cluster structure confined in a supercage of SBA-16 mesoporous
materials, which exhibited excellent catalytic activity for benzyl
alcohol oxidation [3].
Different preparation procedures may result in variation in the
particle size, morphology, structure, and electronic properties of
bimetallic nanoparticle catalysts, which significantly affect the
metal–support interaction and consequently the catalytic perfor-
mance. The pretreatment step, for example, reduction, is essential
to obtain uniform and well-dispersed metallic active species. Con-
neous noble metal catalysts from the viewpoints of atomic effi-
ciency and environmentally benign processes. Supported gold
(
Au) and palladium (Pd) nanoparticle catalysts have been investi-
gated intensively [1–4] since the first successful CO oxidation over
Au and the application of Pd in catalyzing aerobic oxidation of alco-
hols reported by Haruta et al. [5] and Blackburn and Schwartz [6],
respectively. Bimetallic nanoparticles are of greater interest than
monometallic ones for improving the catalytic properties due to
synergetic effects between two elements. Recently, several research
groups have examined supported Au–Pd bimetallic catalysts, which
have remarkably enhanced catalytic activity and product selectivity
in various alcohol oxidation reactions. Enache et al. reported an
ꢀ1
extraordinarily high TOF (269,000 h ) of Au–Pd/TiO
ethanol selective oxidation, showing that the combination of gold
and palladium in an alloyed nanoparticle configuration led to a
2
in 1-phenyl-
ventionally, the catalysts are reduced by flowing hydrogen (H
elevated temperatures or by chemical reactants, such as formalde-
hyde and hydrazine. Nonetheless, H requires special care in pro-
2
) at
2
2
5-fold improvement in catalytic activity [7]. The advantage of
duction, transportation, storage, and use, while most chemical
reducing agents currently employed in catalyst preparation are
hazardous to both the human body and the environment. Our group
previously reported a facile microwave-assisted polyol reduction
⇑
Corresponding authors. Fax: +86 22 27406490 (C.-J. Liu), +65 67947553 (Y.
Yang).
E-mail addresses: ughg_cjl@yahoo.com (C.-J. Liu), yhyang@ntu.edu.sg (Y. Yang).
(
MAPR) method, in which ethylene glycol served as both solvent
0
021-9517/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2012.01.020

1
06
Y. Chen et al. / Journal of Catalysis 289 (2012) 105–117
and reducing agent [4]; nonetheless, high temperatures and micro-
wave radiation are required. Recently, a green, rapid, facile, and
energy-efficient glow-discharge plasma reduction route for
reducing supported metal catalysts using inert gas (Ar) as the
plasma-forming gas has been reported [11–13]. This novel plasma
reduction is carried out at ambient temperature, which signifi-
cantly reduces the energy consumption and cost as well as emission
of the toxic and hazardous materials.
2.2. Catalyst characterizations
Powder X-ray diffraction (XRD) patterns were recorded on a
Rigaku D/Man-2500 V/Pc diffractometer under ambient conditions
using a filtered CuKa radiation source (k = 1.54056 Å) operated at
40 kV and 20 mA. Diffraction data were collected from 20° to 80°
(2h) at a scanning speed of 4°/min. Prior to the test, samples were
dried overnight at 100 °C.
Here, Ar glow-discharge plasma reduction was employed to
synthesize monometallic Au and Pd and bimetallic Au–Pd nanopar-
ticles confined in mesochannels of SBA-15 silicas. In view of the
importance of alcohol dehydrogenation to ketone and aldehyde,
which are valuable intermediates/products for the pharmaceuticals,
agrochemicals, and perfumery industries [14–16], the solvent-free
aerobic oxidation of benzyl alcohol to benzaldehyde is chosen as a
model reaction to test the catalytic performance of plasma-reduced
noble metal nanoparticle catalysts. The effects of plasma treatment
on the structure/configuration and catalytic activity of nanoparticle
catalysts were carefully examined and benchmarked against con-
Transmission electron microscopy (TEM) measurements were
taken on a Philips Tecnai G2 F20 system operated at 200 kV. The
samples were suspended in ethanol and dispersed on a holey car-
bon-coated Cu grid. The mean particle diameter was calculated
from the mean frequency distribution by counting ca. 200 particles
P
n
i
ꢃd
i
P
P
according to d ¼
sample, and n
, where
i
is the sum of i over the entire
n
i
is the number of particles with particle diameter
of d
i
in a certain range. The corresponding root-mean-square error
r
was
determined
using
. Scanning transmission electron microscopy-en-
n
i
ꢀ1
the
statistical
expression
r Pffi ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2
n
i
ðdꢀd
i
Þ
P
rðdÞ ¼
2
ventional H thermal reduction. Post-treatments (oxidation and
re-reduction) were carried out to further reveal the stability of
plasma-reduced bimetallic nanoparticles.
ergy-dispersive X-ray spectroscopy (STEM-EDX) was also con-
ducted using the same equipment combined with a Sirion200
field emission scanning electron microscope. High-resolution
TEM (HRTEM) was performed on a JEOL JEM-2010 operated at
200 kV, following the same sample preparation procedure as de-
scribed above.
2
. Experimental
2.1. Catalyst synthesis
X-ray photoelectron spectroscopy (XPS) was performed on a VG
Escalab 250 spectrometer equipped with an Al anode (Al
SBA-15 mesoporous silica was purchased from the laboratory of
Ka = 1486.6 eV). The background pressure in the analysis chamber
ꢀ7
Professor D. Y. Zhao (Department of Chemistry, Fudan University,
was lower than 1 ꢁ 10 Pa. Measurements were taken using
20-eV pass energy, 0.1-eV step and 0.15-min dwelling time. The cor-
rection of the binding energy (BE) employed the C1s peak of adven-
titious C at 284.6 eV. The background contribution caused by an
inelastic process was obtained by the Shirley method and sub-
tracted. The curve-fitting was done with a Gaussian–Lorentzian
function to deconvolute the overlapped peaks. The atomic ratios of
the different elements were estimated using the ratio of the normal-
ized peak intensities corrected with the atomic sensitivity factors
(5.240 for Au4f, 4.462 for Pd3d, 0.711 for O1s and 0.283 for Si2p).
Diffuse reflectance Fourier transform infrared (DRIFT) spectra of
adsorbed CO were obtained on a Tensor 27 spectrometer (Bruker)
equipped with a liquid-nitrogen-cooled mercury–cadmium–tellu-
rium (MCT) detector, a diffuse reflectance accessory, and a high-
temperature reaction chamber (Praying Mantis, Harrick). The
powder (7 mg) was loaded into the reflectance cell and purged with
2
People’s Republic of China, SBET ꢂ 600 m /g, d100 ꢂ 10 nm) [17] and
calcined in air at 500 °C for 3 h to remove the adsorbed water prior
to use. The monometallic Au and Pd and bimetallic Au–Pd catalysts
confined in SBA-15 were prepared via conventional incipient-wet-
ness impregnation with aqueous solutions of tetrachloroauratetri-
hydrate (HAuCl
4
ꢃ3H
2
O) and/or palladium nitrate dihydrate
(
Pd(NO O). The nominal metal loadings are 2 wt.% for mono-
3
)
2
ꢃ2H
2
metallic Au or Pd catalyst and 2 wt.% Au–2 wt.% Pd for Au–Pd bime-
tallic catalyst. The plasma was applied to reduce the metal
precursors, following the glow-discharge plasma setup and plasma
reduction protocol previously reported [18,19]. Typically, the sam-
ple (0.4 g) was loaded onto a quartz boat and placed in a quartz tube
(
i.d. 35 mm) with two stainless steel electrodes (o.d. 30 mm). When
the discharge tube was evacuated to 100 Pa, the glow-discharge
plasma was generated by applying 1000 V to the electrodes using
a high-voltage amplifier (Trek, 20/20B). The signal input for the
high-voltage amplifier was supplied by a function/arbitrary wave-
form generator (Hewlett–Packard, Model 33120A) with a 100-Hz
square wave. Ultrahigh-pure-grade argon (>99.999%) was used as
the plasma-forming gas. The plasma reduction was operated for
ꢀ1
flowing helium (20 mL min ) at 300 °C for 1 h to remove gas phase
and weakly bound water. CO adsorption was carried out in
ꢀ1
1.11 vol.% CO/He (20 mL min ) for 30 min at 25 °C, followed by
ꢀ1
purging with He (20 mL min ) for another 30 min at the same tem-
perature. The in situ DRIFT spectra were recorded for 64 scans at a
ꢀ
1
1
0 min per cycle (eight cycles in total), with manual mixing of the
4 cm resolution and illustrated using Kubelka–Munk (K–M) units.
sample at intervals between the treatments to ensure even exposure
to the plasma. The bulk temperature of the plasma (measured by
infrared imaging, Ircon modes 100PHT) was close to room tempera-
ture, indicating the negligible heating effect of the glow discharge.
The obtained catalysts were denoted as Au/S15-p, Pd/S15-p, and
AuPd/S15-p. For comparison, the impregnated AuPd/S15 catalyst
2.3. Catalytic activity
The solvent-free aerobic oxidation of benzyl alcohol using
molecular O was carried out in a batch-type reactor operated under
2
atmospheric conditions. Experiments were conducted using a three-
necked glass flask (capacity 25 mL) precharged with benzyl alcohol
and catalyst (benzyl alcohol/metal = 250/1 mmol/mg). The mixture
was stirred using a magnetic stirrer and heated in a silicon oil bath.
The system was equipped with a thermocouple to control the tem-
perature and a reflux condenser. In each reaction run, the mixture
was heated to 160 °C under vigorous stirring (stirring rate
1000 rpm). Oxygen was bubbled into the mixture at a constant flow
ꢀ
1
was reduced in a H
for 2 h and designated as AuPd/S15-H. To study the effect of post-
treatment, AuPd/S15-p was subjected to reoxidation in an O flow
of 30 mL min at 500 °C for 3 h, followed by re-reduction in a H
2 2
flow (purified H ) of 20 mL min at 400 °C
2
ꢀ1
2
ꢀ1
2
flow (5 vol.% H /He) of 60 mL min at 500 °C for 4 h, and designated
as AuPd/S15-pO and AuPd/S15-pR, respectively. The same post-
treatment procedures were also applied to AuPd/S15-H, which is de-
noted as AuPd/S15-HR. A diagram of the preparation procedure and
sample denotation is presented in Scheme 1.
ꢀ1
rate of 20 mL min
to initiate the reaction. After the allowed
reaction time, the catalyst powder was filtered off and the liquid

Y. Chen et al. / Journal of Catalysis 289 (2012) 105–117
107
Scheme 1. Sketch diagram of preparation procedures for Au, Pd, and Au–Pd catalysts with their corresponding structures.
organic products were analyzed using an Agilent 6890 gas
Au(111)
Au(200)
chromatograph equipped with an HP-5 capillary column (30 m long
and 0.32 mm in diameter, packed with silica-based SUPELCOSIL).
Dodecane was used as an internal standard to calculate benzyl alco-
hol conversion and benzaldehyde selectivity. The conversion, selec-
tivity, and quasi-turnover frequency (qTOF) are defined as follows:
Au/S15-p
Pd/S15-p
Au(200)
Au(311)
Pd(111)
AuPd/S15-p
AuPd/S15-H
[
Comp: Set the following three equations entirely roman]
PdO(101)
moles of reactant converted
moles of reactant in feed
moles of product formed
PdO(112)PdO(103) PdO(211)
AuPd/S15-pO
conversionð%Þ ¼
ꢁ 100%
ꢁ 100%
AuPd/S15-pR
AuPd/S15-HR
selectivityð%Þ ¼
qTOFðhꢀ1Þ ¼
moles of reactant converted
moles of reactant converted
moles of total active sites ꢁ reaction time
20
30
40
2
50
60
70
80
theta,degree
3
. Results
Fig. 1. XRD patterns of representative Au, Pd, and Au–Pd catalysts.
3
.1. Monometallic Au/S15-p and Pd/S15-p and bimetallic AuPd/S15-p
be taken into consideration. The diffraction peaks of metal oxides
are undetectable, suggesting that the Ar glow-discharge plasma
can effectively reduce the metal cation to the metallic phase at
room temperature [11].
The TEM micrographs of Au/S15-p, Pd/S15-p, and AuPd/S15-p
and their corresponding particle size distributions are depicted in
Fig. 2. The highly ordered mesoporous structure of SBA-15 is well
preserved after precursor impregnation and glow-discharge plasma
reduction. For Au/S15-p (Fig. 2a), Au nanorods are formed along the
channels of SBA-15, which are uniform in shape with a diameter of
ca. 5.9 nm and an aspect ratio of around 4. Spherical Pd nanoparti-
cles are highly dispersed and confined in the SBA-15 channels
for Pd/S15-p. The average particle size is determined to be
catalysts
During the plasma reduction, distinct color changes were ob-
served for the samples: from pastel yellow to brown purple for
Au/S15-p and from earth yellow to light gray for both Pd/S15-p
and AuPd/S15-p, implying the formation of metallic nanostruc-
tures over SBA-15 support.
The wide-angle XRD patterns of these samples are shown in
Fig. 1. A broad diffraction peak at ca. 22.4° is observed for all the
samples and ascribed to the amorphous silica framework of SBA-
1
6
5. Au/S15-p exhibits four diffraction peaks at 38.16°, 44.30°,
4.58°, and 77.54°, indexed to (111), (200), (220), and (311) reflec-
tions, respectively, for the face-centered cubic (FCC) lattice struc-
ture of Au (PDF# 65–2870). The peaks are sharp and intense,
indicating good crystallinity of metallic Au. Pd/S15-p shows only
one diffraction peak at 39.72°, assigned to Pd(111) reflection
5
.3 ± 0.6 nm, dramatically larger than the average size of single Pd
crystalline particles calculated by XRD (1.6 nm), which can be as-
cribed to the fact that powder XRD is sensitive to the size of coher-
ent domains, which can differ significantly from the particle size in
the case of multiple crystals composed of smaller primary nanopar-
ticles [20]. The different morphologies of nanostructures (nanorods
for Au versus nanoparticles for Pd) obtained after reduction in
glow-discharge plasma have been reported by Wang et al. and
attributed to the intrinsic properties of different metals, such as
surface energy and redox potential [12]. Bimetallic AuPd/S15-p also
shows a spherical morphology with a narrow particle size distribu-
tion centered at 5.7 nm.
(
PDF# 46–1043) that is weak and broad, indicating poor crystallin-
ity with structural defects. The average particle size based on the
111) reflection using the Scherrer formula is 5.8 and 1.6 nm for
(
Au/S15-P and Pd/S15-p, respectively. For bimetallic AuPd/S15-p,
reflections assigned to both metallic Au and Pd are clearly seen
individually, which rules out the possibility of a single-phase
homogeneous Au–Pd alloy. Besides the predominant presence of
Au–Pd bimetallic nanoparticles, the probability of sporadically
dispersed Au and Pd monometallic nanocrystallites should also

1
08
Y. Chen et al. / Journal of Catalysis 289 (2012) 105–117
Fig. 2. TEM results for mono- and bimetallic catalysts: (a) Au/S15-p; (b) Pd/S15-p; (c) AuPd/S15-p; (d) AuPd/S15-H. 1, Low-resolution TEM image; 2, HRTEM image; 3,
corresponding size distribution.
The HRTEM images are also depicted in Fig. 2. Au nanorods
consist of multiple crystals, showing clear lattice fringes with
d = 0.236 and 0.205 nm assigned to the (111) and (200) planes
of Au, respectively (Fig. 2a-2), which is in good agreement with
XRD results. Pd/S15-p exhibits a fringe distance of 0.227 nm
maximum at the center for Au and a uniform distribution through-
out the scanning track for Pd. The intensity of each element is as-
sumed to be proportional to the atomic densities. A similar EDX
result was obtained in our previously reported Au–Pd bimetallic
catalysts and denoted as a structure of Pd clusters on an Au-rich
core [3].
(Fig. 2b-2), corresponding to the Pd(111) plane. Fig. 2c-2 shows
that the bimetallic nanoparticles in AuPd/S15-p retain the FCC lat-
tice structure without the structural incoherence assignable to sep-
arate monometallic phases. A subtle phase contrast at the fringe is
observed, and the calculated lattice fringes of d = 0.235 and
In addition to the bulk morphology and structure, additional
information about the surface components of Au–Pd bimetallic
nanoparticles was further derived by XPS measurements. The
XPS spectra of Au4f, Pd3d, O1s, and Si2p core levels are shown in
Fig. 4, and the surface compositions are summarized in Table 1.
For AuPd/S15-p, BEs at 87.3 (4f5/2) and 83.5 (4f7/2) eV are character-
istic of metallic Au, while two symmetric peaks at 340.6 (3d3/2) and
335.5 (3d5/2) eV are assigned to metallic Pd, suggesting the com-
plete reduction of Au and Pd by Ar glow-discharge plasma. The ex-
pected Au/Pd molar ratio in bulk is 0.54. Nevertheless, AuPd/S15-p
shows a significantly decreased ratio of 0.21, implying that the
plasma-reduced Au–Pd bimetallic nanoparticles are surface-en-
riched in Pd. Hutchings and co-workers have attributed this Pd
surface enrichment to the formation of a Pd(shell)–Au(core)
0
.203 nm correspond to Au(111) and Au(200) facets, respectively.
Surprisingly, no lattice facets assigned to Pd can be determined.
The line-scan EDX analysis was carried out in the STEM model to
further confirm the local composition and element distribution of
the bimetallic nanoparticles. The EDX profiles of Au L
a and Pd La
X-rays are plotted in Fig. 3a along the line for a representative indi-
vidual nanoparticle (inset of Fig. 3a). The particle is ca. 6 nm in
diameter, and both Au and Pd are detected, confirming the success-
ful preparation of Au–Pd bimetallic nanoparticles. Different inten-
sity profiles are shown: a volcano-like distribution with the

Y. Chen et al. / Journal of Catalysis 289 (2012) 105–117
109
Three distinct bands are clearly observed for Pd/S15-p after
(a)
ꢀ1
exposure to CO: (1) a band at 2089 cm
attributed to linear
carbonyls on the (111) facet of Pd particles (designated as L); (2)
ꢀ1
a band at 1990 cm assigned to twofold bridged carbonyls on
the (100) facet of Pd particles (designated as B
band at 1948 cm corresponding to twofold bridged carbonyls
1
); and (3) a broad
AuPd/S15-p
ꢀ1
2
on the (111) facet of Pd particles (designated as B ). The intensities
Au
of these absorbencies significantly increase at higher CO coverage.
After 20 min of CO exposure, the intensities of the three bands no
longer change, indicating the saturation of CO adsorption on Pd par-
ticles. Upon purging under He, the characteristic peaks assigned to
gaseous CO completely disappear after 5 min. The intensities of
other bands also decrease and red-shift to lower frequencies due
to the decreased CO coverage. After purging with He for 30 min,
Pd
0
2
4
6
8
position,nm
only the features of L and B
2
remain, suggesting preferential and
more stable carbonyl adsorption on Pd(111) facets than on
(b)
2
+
Pd(100) facets. Furthermore, no band for Pd (between 2160 and
ꢀ1
+
ꢀ1
2
116 cm ) or Pd (between 2114 and 2140 cm ) can be seen, con-
firming the exclusive presence of surface metallic Pd species in Pd/
S15 [27].
AuPd/S15-pR
AuPd/S15-p exhibits CO-DRIFT spectra characteristic of Pd,
implying that Au may be covered (at least partially) by Pd in the
bimetallic nanoparticles, which further suggests the possibility of
a Pd-rich shell/Au-rich core structure. After purging under He for
Pd
Au
3
1
1
0 min, L and B CO bands red-shift from 2090 to 2085 and from
944 to 1939 cm , respectively, compared to monometallic Pd/
ꢀ1
S15-p, which can be attributed to either a geometric or an elec-
tronic effect [28]. First, Pd crystallites form small islands and rough
surfaces in the presence of Au cores, leading to weaker lateral
interaction between adsorbed CO molecules (i.e., decreased
dipole–dipole coupling in the adlayer). Second, due to the higher
electronegativity of Au than of Pd, Pd donates electrons to Au
0
2
4
6
8
10
12
14
position,nm
Fig. 3. The line-scan EDX intensity profiles of a single bimetallic particle of (a)
AuPd/S15-p and (b) AuPd/S15-pR along the line. The inset is the corresponding
STEM image.
and becomes electron-deficient, which is compensated for via in-
⁄
creased back donation in the 2
p
antibonding orbitals of adsorbed
structure [8]. The characteristic peaks for O1s and Si2p exhibit BEs
consistent with SiO . The surface O/Si atomic ratio is 2.69, which is
2
higher than the bulk stoichiometric ratio of 2.0 due to the coverage
of surface hydroxyl groups [21].
CO molecules. Therefore, not only has the Pd ensemble size been
affected, but also the electronic properties of Pd have been modi-
fied in the presence of Au. The synergistic effect between Au and
Pd on the local structure and active site distribution of bimetallic
nanoparticles can also be semiquantitatively reflected by the
intensity of bridge CO relative to linear CO ((B + B )/L) [19]. To ex-
To summarize these XRD, TEM, HRTEM, EDX, and XPS analyses, a
Pd-rich shell/Au-rich core structure can be proposed for AuPd/S15-
p catalysts. The so-called geared step-cycled reduction mechanism
due to the different redox potentials of Au and Pd may explain the
formation of such structures [3,22]: during the plasma reduction of
Pd and Au cations, the preferential reduction of Au is favored via an
electron transfer from Pd to Au because of the higher electronega-
1
2
clude the interference of gaseous CO, the CO-DRIFT spectra of CO-
desorption after 30 min of He purging were employed to analyze
the peaks. Outgassing at room temperature completely depletes
the B band, and only the area ratio of B /L is calculated, showing
1
2
1.65 and 1.43 for Pd/S15-p and AuPd/S15-p, respectively. CO is
preferentially bound linearly (L) on low-coordinated Pd atoms
and the B/L ratio decreases for small particles due to the larger
fraction of low-coordinated surface atoms [29]. In this study,
although it shows a slightly larger particle size than Pd/S15-p
(5.3 and 5.7 nm for Pd/S15-p and AuPd/S15-p, respectively),
AuPd/S5-p possesses more exposed Pd atoms of low coordination
(edge/corner sites), suggesting that the Pd nanocrystallites under-
go a major restructuring, leading to large exposure of coordination-
unsaturated Pd atoms.
2
+
tivity of Au. Pd precursor acts as an electron-mediate and acceler-
ates the reduction of Au cations. After the coalescence and growth
of Au clusters, Pd cations are adsorbed onto Au surface and further
reduced, creating a Pd shell/Au core structure. In addition, a reason-
able explanation for the expansion of Pd facets is the pseudomor-
phically epitaxial growth of Pd on Au-rich cores, which has been
suggested by Hori and co-workers [23].
Fig. 5 shows CO-adsorbed infrared spectra at room temperature
on various catalysts. For Au/S15-p,
a
doublet at 2172 and
ꢀ1
2
119 cm due to gas-phase CO is observed upon introducing CO.
The selective oxidation of benzyl alcohol with molecular O was
2
CO linearly bonded to low-coordinated metallic Au sites (step sites
over the Au surfaces and step sites at the interface of Au and the
employed as a chemical probe reaction to examine catalytic perfor-
mance of the plasma-reduced mono- and bimetallic catalysts. In the
absence of catalysts, only 7.5% of benzyl alcohol converts due to non-
catalytic oxidation. The SBA-15 support does not show any catalytic
activity (8.1% conversion). The evolution of the catalytic performance
with reaction time was studied and the results are illustrated in
Fig. 6, and the yield of benzaldehyde is shown in Fig. S2. Monometal-
lic Au catalyst possesses poor activity and quickly deactivates after
4 h, while both Pd/S15-p and AuPd/S15-p catalysts show monotoni-
cally increased benzyl alcohol conversions. After 8 h, only 25.7% con-
version is achieved over Au/S15-p, and Pd/S15-p can convert 83.6% of
ꢀ
1
support) should show a characteristic peak at ca. 2100 cm [24];
ꢀ
and the residual Cl after impregnation may change the electronic
properties of Au and shift this adsorption peak to a higher
ꢀ1
frequency [25]. Here, the band at 2119 cm will be ascribed to a
combination of gaseous CO and CO linearly adsorbed onto step sites
0
of metallic Au [24]. After gas-phase CO is removed, this absorbance
ꢀ1
is red-shifted to 2117 cm , with a substantial asymmetric tailing
in the high-frequency region, suggesting a wide distribution of CO
sites with different adsorption strengths [26].

1
10
Y. Chen et al. / Journal of Catalysis 289 (2012) 105–117
(
b) Pd 3d
Pd 3d5/2
335.5
(
a) Au4f
Au 4f 7/2
Pd 3d3/2
40.6
3
Au 4f 5/2
83.5
8
7.3
AuPd/S15-p
AuPd/S15-p
8
3.5
8
7.1
AuPd/S15-H
AuPd/S15-H
8
3.3
8
7.0
0
2+
Pd
Pd
0
Pd
2+
Pd
AuPd/S15-pR
AuPd/S15-pR
89
88 87 86 85 84 83 82 81 80
BE,eV
348 346 344 342 340 338 336 334 332
BE,eV
(c) O 1s
533.1
(d) Si 2p
103.3
AuPd/S15-p
AuPd/S15-p
AuPd/S15-H
AuPd/S15-H
AuPd/S15-pR
AauPd/S15-pR
538
536
534
BE,eV
532
530
528
108 107 106 105 104 103 102 101 100 99
BE,eV
Fig. 4. XPS spectra of AuPd/S15-p, AuPd/S15-H, and AuPd/S15-pR: (a) Au4f; (b) Pd3d; (c) O1s; (d) Si2p.
Table 1
Surface compositions calculated from XPS results of representative samples.
Sample
Surface atoms percentage (at.%)a
Surface Au/Pd ratio (at.%)
Surface O/Si ratio (at.%)
Relative abundance of Pd0 and Pd2+ (at.%)
Au
Pd
O
Si
Pd0
Pd2+
AuPd/S15-p
AuPd/S15-H
AuPd/S15-pR
0.060 (0.56)
0.10 (1.01)
0.10 (1.05)
0.27 (1.47)
0.19 (1.03)
0.29 (1.56)
72.67
73.18
73.01
27.00
26.53
26.59
0.21
0.53
0.36
2.69
2.76
2.75
100.00
100.00
59.29
0
0
40.71
a
Values in parentheses are wt.%.

Y. Chen et al. / Journal of Catalysis 289 (2012) 105–117
111
Au/S15-p
Pd/S15-p
AuPd/S15-p
AuPd/S15-H
2
089
L
B/L=1.65
B/L=1.43
B/L=0.80
COgas 2085
L
2089
L
COgas
0
COgas Au -CO
1990
B1
2001 1939
1
948
B2
COgas
1992
B
1946
B
B1
B2
a
b
a
b
a
b
a
b
1
2
c
c
c
c
d
e
d
d
d
e
f
e
e
g
f
f
g
h
i
j
f
h
i
g
g
j
h
i
j
h
i
j
2300
2200
2100
2000
1900
1800 2300
2200
2100
2000
1900
1800 2300
2200
2100
2000
1900
1800 2300
2200
2100
2000
1900
1800
-1
-1
-1
-1
wavenumber,cm
wavenumber,cm
wavenumber,cm
wavenumber,cm
Fig. 5. DRIFT spectra of CO adsorbed at 25 °C on Au/S15-p, Pd/S15-p, AuPd/S15-p, and AuPd/S15-H. Sample exposed to 1.1 kPa CO for (a) 1 min, (b) 5 min, (c) 10 min, (d)
0 min, and (e) 30 min and purged with He for (f) 1 min, (g) 5 min, (h) 10 min, (i) 20 min, and (j) 30 min.
2
benzyl alcohol. AuPd/S15-p exhibits excellent catalytic activity,
showing 100% conversion of benzyl alcohol. The plots in Fig. 6a are
fitted with first-order reaction kinetics, and the rate constants are
reaction, which is consistent with its low catalytic activity. In
contrast, the E for Pd/S15-P is extremely low (16.7 ± 0.7 kJ/mol),
suggesting an easily initiated reaction. Nonetheless, mass-trans-
port limitation and reactant diffusion resistance may be concerned
when the surface reaction occurs much faster than reactant
adsorption and/or product desorption [3]. The calculated E for
a
AuPd/S15-p (45.1 ± 4.2 kJ/mol) is moderate to readily initiate the
reaction as well as to obtain an optimal reaction–diffusion balance.
a
listed in Table 2. AuPd/S15-p shows
a rate constant k of
ꢀ1
0
2
.50 ± 0.05 h , which is 12.5 ꢁ that of Au/S15-p (0.04 ± 0.01) and
ꢁ that of Pd/S15-p (0.28 ± 0.02). The catalytic activity is also repre-
sented using qTOF (Fig. 6c). After the first 0.5 h of reaction, AuPd/
ꢀ1
S15-p shows a qTOF of ca. 27,000 h , which is more than twice that
of monometallic Pd and Au catalysts. Therefore, besides excellent
long-term stability, AuPd/S15-p also shows high catalytic activity.
The selectivity toward benzaldehyde differs remarkably among
these catalysts (Fig. 6b). Benzaldehyde is produced dominantly
a
Additionally, this E is remarkably higher than that of SBA-16-
supported Au–Pd catalyst (16.7 kJ/mol) in our previous report [3],
indicating that the ordered arrangement of mesochannels of SBA-
15 exhibits less resistance for electric charge and ion diffusion than
that of SBA-16.
(
selectivity > 98%) over Au/S15-p during the first 3 h, followed by
slightly decreased selectivity due to the overoxidation of benzalde-
hyde to benzoic acid, which occurs independent of the intervention
of catalyst [30]. Pd/S15-p shows a low benzaldehyde selectivity of
To clarify the origin of toluene and the reaction mechanism, we
studied the catalytic performance in 1 h of reaction time over these
three catalysts in a N atmosphere. The results are depicted in
2
5
9.0% after the first 0.5 h of reaction, with tremendous amounts of
Fig. 6d–f. When benzyl alcohol was exposed to the anaerobic con-
ditions, benzaldehyde and toluene were both produced, implying a
dehydrogenation mechanism in which benzaldehyde formation
occurs in the absence of oxidant. Au catalyst shows the highest
selectivity (>65%). For Pd/S15-p, 43.4% benzaldehyde and 56.6%
toluene are produced after first 10 min. Benzaldehyde is generated
as a subordinate component until 50 min of reaction, accompanied
by a large amount of toluene. Mixing Au with Pd can remarkably
suppress the formation of toluene over Pd catalyst, implying that
toluene is generated via either a hydrogen transfer or an oxygen
transfer mediated by surface-enriched Pd over AuPd/S15-p with
a Pd-rich shell/Au-rich core structure [30]. Fig. 6f shows that tolu-
ene and benzaldehyde are generated in almost equimolar amounts
in the initial stage of reaction over AuPd/S15-p, implying that the
disproportionation of benzyl alcohol occurs even at very low con-
versions. The initial rates, which are determined from the slopes of
toluene as by-product. The selectivity increases with the reaction
time, reaches its maximum at 3 h, and further deteriorates due to
formation of benzoic acid. After 8 h of reaction, only 53.2% benzal-
dehyde selectivity remains, accompanied by the large quantity of
benzoic acid due to overoxidation. AuPd/S15-p shows a good selec-
tivity profile in between monometallic Au and Pd catalysts, which
can be well maintained at high conversion up to 8 h.
The effect of reaction temperature from 120 to 160 °C was also
examined for these catalysts. In general, an elevated temperature
leads to an increased reaction rate. The Arrhenius plot is shown
5
in Fig. S1. The determined preexponential factors are 1.44 ꢁ 10 ,
7
1
5
.95 ꢁ 10 , and 2.87 ꢁ 10 for Au/S15-p, Pd/S15-p, and AuPd/S15-
p, respectively. The frequency factor A reflects the total number
of collisions per unit time, while the rate constant k is the number
of collisions that result in a reaction. Therefore, the ratio of k to A
(
k/A) is approximately the number of effective active sites for reac-
the conversion curves at zero conversions under aerobic (O
anaerobic (N ) conditions, are summarized in Table 2. It is note-
worthy that for these three catalysts, the initial rates are one order
of magnitude higher under O than under N , indicating the essen-
tial role of O in benzyl alcohol oxidation. Pd itself shows r com-
parable to that of Au–Pd bimetallic catalyst under N , further
suggesting that Pd is dominantly responsible for the reaction over
AuPd/S15-p. After O is introduced, Au–Pd bimetallic nanoparticles
2
) and
tions. As shown in Table 2, A is large while k is rather small for Au/
S15-p, implying that most collisions cannot initiate the reaction
due to insufficient effective active sites. In contrast, Pd/S15-p
shows the highest k/A ratio due to the abundance of effective active
sites. AuPd/S15-p possesses a moderate number of active sites. The
information derived from the k/A ratio complements the true acti-
2
2
2
2
0
2
vation energy E
listed in Table 2). Au/S15-p exhibits a remarkably high E
5.7 ± 2.9 kJ/mol, demanding a high-energy input to initiate the
a
measured according to the rate constant k (as
2
a
of
remarkably enhance the initial rate, showing an increasing rate of
benzyl alcohol disappearance and benzaldehyde appearance.
7

1
12
Y. Chen et al. / Journal of Catalysis 289 (2012) 105–117
reaction under O
2
reaction under N
2
1
00
(d) 100
3.0
(
(
(
a)
Au/S15-p
Pd/S15-p
AuPd/S15-p
Au/S15-p
r =0.25E-6 mol/s
0
2.5
2.0
1.5
1.0
8
6
4
2
0
0
0
0
0
80
60
40
20
0
benzaldehyde
-kt
x =1-e
kAu = 0.04 ± 0.01
kPd = 0.28 ± 0.02
kAuPd= 0.50 ± 0.05
toluene
benzyl alcohol
coversion
0.5
0.0
0
0
0
2
4
6
6
6
8
8
8
0.0
0.2
0.4
0.6
time,h
0.8
1.0
1.0
1.0
time,h
100
(e) 100
3.0
2.5
2.0
b)
Pd/S15-p
r =0.0027 mol/h
0
benzyl alcohol
coversion
8
6
4
2
0
0
0
0
0
80
60
40
20
0
toluene
1.5
1.0
0.5
benzaldehyde
Au/S15-p
Pd/S15-p
AuPd/S15-p
0.0
2
4
0.0
0.2
0.4
0.6
0.8
time,h
time,h
c) 30000
(f) 100
80
3.0
2.5
2.0
1.5
1.0
0.5
0.0
benzyl alcohol
coversion
AuPd/S15-p
Au/S15-p
Pd/S15-p
r =0.0042 mol/h
0
2
2
1
1
5000
0000
5000
0000
AuPd/S15-p
60
40
20
0
benzaldehyde
q
5000
toluene
0.8
0
2
4
0.0
0.2
0.4
0.6
time,h
time,h
Fig. 6. The time–activity profile for selective benzyl alcohol oxidation under aerobic conditions for (a) benzyl alcohol conversion; (b) benzaldehyde selectivity; (c) and qTOF
ꢀ
1
2 2
and under anaerobic conditions for (d) Pd/S15-p; (e) Au/S15-p; and (f) AuPd/S15-p. Reaction conditions: benzyl alcohol/metal = 250/1 mmol/mg; O (N ), 20 mL min ;
temperature, 160 °C.
Table 2
a
The average particle size and kinetic parameters of benzyl alcohol oxidation under anaerobic and aerobic conditions.
Particle size (nm)b
E
(kJ/mol)c
A (h )c
ꢀ1
Rate constant k (h )d
ꢀ1
Catalyst
a
0
Initial rate r (mol/h)
Anaerobic
Aerobic
Au/S15-p
Pd/S15-p
AuPd/S15-p
AuPd/S15-H
AuPd/S15-pO
AuPd/S15-pR
5.9(D)
5.3
5.7
5.8
13.9
6.7
75.7
16.7
45.1
—
—
—
5.95E7
2.87E1
1.44E5
0.04
0.28
0.50
0.36
0.28
0.33
0.0009
0.0027
0.0042
—
0.0024
0.0021
0.010
0.022
0.039
0.025
0.028
0.033
e
a
b
c
ꢀ1
Reaction conditions: benzyl alcohol/metal = 250/1 mmol/mg; O
Determined from TEM results.
2
(N
2
), 20 mL min ; temperature, 160 °C.
E
a
Determined from Arrhenius plots ln k ¼ ꢀ þ ln A.
RT
d
e
k is determined for reaction under aerobic conditions.
Not determined.

Y. Chen et al. / Journal of Catalysis 289 (2012) 105–117
113
3
.2. Plasma reduction vs. H
2
thermal reduction
a) 100
(
AuPd/S15-p
AuPd/S15-H
Au–Pd bimetallic catalyst was also prepared by conventional
8
0
thermal reduction under hydrogen and was denoted as AuPd/
S15-H. The XRD pattern of AuPd/S15-H catalyst, as depicted in
Fig. 1, is similar to that of AuPd/S15-p. The (111), (200), (220),
and (311) reflections of Au and the (111) reflection of Pd are
clearly identified, suggesting the FCC lattice structure of AuPd/
S15-H. However, the peak intensity of AuPd/S15-H is much lower
than that of the plasma-reduced sample, implying unsatisfactory
6
0
-kt
x
=1-e
40
kAuPd-p= 0.50
±
0.05
kAuPd-H=0.36 ± ±0.04
2
0
0
crystallinity after H
can be observed.
2
reduction. No characteristic peaks of PdO
The morphology of AuPd/S15-H bimetallic nanoparticles is
more directly shown in the TEM micrograph in Fig. 2d-1. Spherical
nanoparticles are highly dispersed on SBA-15, and the average par-
ticle diameter of AuPd/S15-H (ca. 5.8 nm) is similar to that of
AuPd/S15-p. Nonetheless, the particle size distribution is remark-
ably broad. A noticeable fraction of particles are smaller than
0
2
4
6
8
time,h
b) 100
(
8
0
2
nm, and many particles are larger than 12 nm (Fig. 2d-3). HRTEM
results show that the particle exhibits a decahedral multiply
twinned structure for AuPd/S15-H, and each twinned part main-
tains the FCC structure. Fig. 2d-2 presents two twinned boundaries
of a multiply twinned particle, which is distinctly different from
the Pd-rich shell/Au-rich core structure of AuPd/S15-p. The lattice
fringe assigned to the Au(111) facet is determined as 0.236 nm,
which is consistent with the value derived from the XRD pattern.
The surface compositions of AuPd/S15-H were also examined by
XPS measurements. As depicted in Fig. 4, AuPd/S15-H exhibits XPS
spectra of Au4f, Pd3d, O1s, and Si2p core levels similar to those of
6
0
40
2
0
0
AuPd/S15-p
AuPd/S15-H
0
2
4
6
8
time,h
2
AuPd/S15-p. The bimetallic nanoparticles after H thermal reduc-
0
0
tion consist exclusively of metallic Au and Pd species. The surface
Au/Pd ratio of AuPd/S15-H is determined as 0.53, which is almost
identical to the nominal bulk ratio (0.54), suggesting the absence
of preferential enrichment of Au or Pd on the surface. We speculate
that there is a possibility of aggregated bimetallic nanoparticles
with irregular Au–Pd distribution. The surface O/Si atomic ratio
is 2.76, which is also contributed to by the surface coverage of hy-
droxyl groups [21].
Fig. 7. The activity profile for selective benzyl alcohol oxidation using molecular O
over AuPd/S15-p and AuPd/S15-H: (a) benzyl alcohol conversion; (b) benzaldehyde
2
selectivity. Reaction conditions: benzyl alcohol/metal = 250/1 mmol/mg; O ,
2
ꢀ
1
2
0 mL min ; temperature, 160 °C.
that the selectivity toward benzaldehyde is also poor over AuPd/
S15-H and the best selectivity obtained at 3 h (82.4%) is ca. 15%
lower than over AuPd/S15-p. A large amount of benzoic acid is
produced and only 16% selectivity toward benzaldehyde remains
after 8 h. It was proposed that only selected sites allow the desorp-
tion of benzaldehyde, implying that the benzaldehyde selectivity is
dependent on the particle size as well as the active site distribution
CO-adsorbed DRIFT results are illustrated in Fig. 5. Despite sim-
ilarity to the CO-DRIFT spectra of AuPd/S15-p, the IR bands of
AuPd/S15-H are sharper, implying stronger adsorption of CO,
which can be attributed to either smaller particle size or weaker
metal–support interaction [13]. Upon introducing CO, both
[
32]. AuPd/S15-H has a large number of particles small in diameter,
linear-(L) and bridge-bound (B
0-min He purging, B disappears, indicating unstable carbonyl
adsorption on the Pd (100) facet. The L band is centered at
1 2
and B ) CO bands appear. After
which shows high oxygen heat of adsorption, that is, strong affinity
for oxygen, and is prone to oxygen poisoning. According to the oxi-
dative dehydrogenation reaction mechanism, the metal surfaces
will be covered with dehydrogenation products, such as organic
substrates/intermediates, oxygen, and hydrogen. As the reaction
proceeds, the concentration of benzyl alcohol decreases and excess
amounts of oxygen occupy the metal surfaces, leading to the ove-
roxidation of benzyl alcohol to benzoic acid and thereby the shar-
ply decreased selectivity.
3
1
ꢀ
1
2
B
089 cm , which is in accordance with AuPd/S15-p. Nonetheless,
ꢀ1
2
shows a higher frequency (1946 cm ) than AuPd/S15-p, which
is attributed to a geometric or electronic effect. The intensity of
bridge-bound CO relative to linear-bound CO ((B + B )/L) of
1
2
AuPd/S15-H is determined as 0.8, suggesting the high abundance
of low-coordinated Pd atoms due to the noticeable fraction of small
particles that contribute dominantly to the linearly binding CO.
The catalytic behavior of AuPd/S15-H was examined for the
selective oxidation of benzyl alcohol under aerobic conditions, as
shown in Fig. 7. Although AuPd/S15-H possesses a similar average
particle size, the catalytic activity is poorer than that of AuPd/S15-
p. The rate constant and the initial rate are determined as
3.3. Effect of post-treatment of AuPd/S15-p
These results have verified the superiority of Ar glow-discharge
plasma treatment to conventional H thermal reduction. To further
2
understand the nature of bimetallic nanoparticles formed under
plasma reduction, we introduced the post-treatment of the as-
synthesized AuPd/S15-p, that is, O
(AuPd/S15-pO) followed by H thermal reduction at 500 °C for 4 h
(AuPd/S15-pR). The effect of post-treatment was examined by the
following characterizations and catalytic tests.
The wide-angle XRD patterns of AuPd/S15-pO and AuPd/S15-pR
are depicted in Fig. 1. In both samples, Au exists as a metallic phase
ꢀ1
0
.36 ± 0.04 h and 0.025 mol/h, respectively (Table 2). Wang and
co-workers have proposed that benzyl alcohol molecule is
adsorbed via such a conformation that the benzene ring is parallel
to the terrace sites, while the b-H is attached to the edge/corner
sites; that is, the adsorption needs terrace atoms, while the b-H
cleavage requires the edge/corner sites. Thus, a balance of terrace
to edge/corner atoms is critical, which favors both adsorption of
benzyl alcohol and subsequent b-H cleavage [31]. It is noteworthy
2
calcination of AuPd/S15-p
2

1
14
Y. Chen et al. / Journal of Catalysis 289 (2012) 105–117
with a slight shift toward high angle for the characteristic peaks,
implying a shrunken lattice constant and possible lattice contrac-
tion. For AuPd/S15-pO, the peaks assigned to metallic Pd com-
pletely disappear. An intense peak at 33.76° and four weak
diffraction peaks at 41.92°, 54.58°, 60.16°, and 71.48° appear,
which are indexed to (101), (110), (112), (103), and (211) reflec-
tions of the tetragonal PdO lattice (PDF # 41-1107). Therefore, PdO
2
and PdO(101) planes, respectively. After H re-reduction, AuPd/
S15-pR shows an improved particle size distribution. Lattice facets
corresponding to Au(111) (d = 0.233 nm) and Pd(111) (d = 0.225
nm) can be observed in Fig. 8c-2. Fig. 3b shows the EDX profiles
0
of Au La and Pd La X-rays plotted along the line for a representative
individual nanoparticle over AuPd/S15-pR. In contrast to AuPd/S15-
p, Pd in AuPd/S15-pR displays a profile similar to that of Au, and the
intensity along the entire line scans is strong. Therefore, Au and Pd
should coexist in bimetallic nanoparticles, with Pd/PdO clusters
covering on the small Au core.
2
is the predominant phase after reoxidation under O . The diffrac-
tion peaks are sharp and intense, suggesting the aggregation of
PdO particles to a certain extent during the oxidative thermal
treatment. For AuPd/S15-pR, a shoulder peak at 39.5° assigned to
the Pd(111) plane appears. Surprisingly, the diffraction peaks of
PdO only become weak and broad rather than disappearing, imply-
ing the incomplete reduction of PdO. This is in contrast to the
exclusive presence of the metallic Pd0 phase for monometallic
Pd/S15 synthesized via the same preparation procedure in the pre-
vious report [11] and can be attributed to the synergistic interac-
tion between Pd and Au. Introducing Au may effectively hinder
the decomposition of PdO to Pd at high temperatures [33]. Doping
AuPd/S15-pR shows XPS spectra of Au4f, O1s, and Si2p core lev-
els similar to those of AuPd/S15-p, as depicted in Fig. 4. The Pd3d
spectrum consists of two asymmetric peaks attributed to Pd3d3/2
and Pd3d5/2 core levels (Fig. 4b). In addition to the characteristic
0
peaks assigned to metallic Pd , two small peaks around 341.3
2
+
and 336.4 eV corresponding to Pd species appear after deconvo-
2
+
lution. The calculated abundance of surface Pd is 40.71%. The
Au4f spectrum shows that Au remains in the metallic phase. The
0
0
peaks of both metallic Au and Pd shift to lower BEs due to the in-
creased number of Au–Pd bonds as the Au core becomes smaller,
2
Au with Pd is also suggested to ‘‘protect’’ TiO support under
reducing atmosphere [34]. From the XRD pattern, we suggest that
the structure of the nanoparticles in AuPd/S15-pR (Au, Pd, and
PdO) is ill defined; nonetheless, the presence of a homogeneous
phase of Au–Pd alloy can be ruled out due to the individual peaks
for Au and Pd.
The nanoparticle morphology and structure of AuPd/S15-pO and
AuPd/S15-pR, as well as their corresponding size distributions, are
directly shown in Fig. 8. Only large particles exist for AuPd/S15-pO
2
which is consistent with the EDX results. After H re-reduction,
the residual oxide species on the topmost Pd layer may induce
the inner Pd atoms at the Au–Pd interface to become partially pos-
itively charged, which significantly affects the Au–Pd bonding
strength and thereby imposes a strong impact on the bimetallic
synergistic interaction. The surface Au/Pd atomic ratio of AuPd/
S15-pR (0.36) is higher than that of AuPd/S15-p (0.21), which
corresponds to a Pd surface enrichment with larger abundance of
0
(
13.9 ± 3.3 nm), suggesting the substantial aggregation of Au and/or
surface Au species.
PdO nanoclusters. The HRTEM image in Fig. 8b-2 shows the lattice
constants d = 0.235 and 0.266 nm, corresponding to the Au(111)
CO-adsorbed DRIFT spectra of AuPd/S15-pO and AuPd/S15-pR
are illustrated in Fig. 9. For AuPd/S15-pO, besides the gaseous CO
Fig. 8. TEM results for different bimetallic Au–Pd catalysts: (a) AuPd/S15-p (for comparison); (b) AuPd/S15-pO; (c) AuPd/S15-pR. 1, Low-resolution TEM image; 2, HRTEM
image; 3, corresponding size distribution.

Y. Chen et al. / Journal of Catalysis 289 (2012) 105–117
115
band at 2172 cmꢀ1, only two adsorption bands at 2113 and
is suggested to be more active than PdO and predominantly
contributes to the high activity [36]. Our previous study also dem-
onstrated that the initial reaction activity is intimately related to
ꢀ1
1
956 cm appear. The higher-wavenumber band can be assigned
2+
to CO adsorbed onto Pd sites and the lower-frequency band is
due to Pd -CO-Pd species [35]. Both bands show a remarkable
red-shift to 2168 and 1948 cm , respectively, when purged under
+
+
0
the surface Pd concentration [4]. After reoxidation, AuPd/S15-pO
ꢀ
1
consists exclusively of Au and PdO species. Consequently, an induc-
tion period is required to transform PdO into metallic Pd nanopar-
ticles, which accounts for the low initial activity. The slightly higher
selectivity toward benzaldehyde over AuPd/S15-pO is in accor-
2
He for 5 min, which can be assigned to L and B bands of CO
adsorption on metallic Pd0 sites formed by the slow reduction of
PdO during CO adsorption and He outgasses (the peaks assigned
to gaseous CO
2
appear and are not shown here). Moreover, both
2
dance with the poor activity. Subsequent H re-reduction can re-
bands undergo a slight blue-shift compared to AuPd/S15-p, which
is contributed by the electronic effect associated with the weak-
ened surface CO bond; that is, the compression of the CO adlayer
by surface oxygen species (PdO) leads to higher frequencies. The
trieve an initial activity comparable to that of AuPd/S15-p,
implying the existence of surface Pd species. However, the conver-
0
sion reaches a plateau after 2 h, suggesting a rapid deactivation and
declining stability, which is contributed by lower hydrogen uptake
due to the higher abundance of surface Au on AuPd/S15-pR than on
AuPd/S15-p. The b-H elimination producing carbonyl and surface
hydride is known to be the rate-limiting step in the dehydrogena-
tion of alcohols. Therefore, the weakened capability to adsorb sur-
face metal hydride components may be responsible for the poor
long-term stability. Nonetheless, the lower abundance of surface
hydride species contributes to the high benzaldehyde selectivity
by effectively suppressing the hydrogenolysis of benzyl alcohol to
toluene, as shown in Fig. 10b.
band
Pd(100) facets, which is consistent with the exclusive presence
of PdO after O reoxidation. The calculated intensity of bridge-
1
B is undetectable, implying the absence of exposed
2
bound relative to linear-bound CO (B/L = 5.55) is higher than for
AuPd/S15-p (1.43), indicating that the bimetallic nanoclusters in
AuPd/S15-pO predominantly consist of close-packed planes with
few low-coordinated atoms due to the large particles and poor dis-
persion (13.9 ± 3.3 nm). After re-reduction, AuPd/S15-pR shows
CO-DRIFT spectra similar to those for AuPd/S15-p. Notably, the
B
1
band appears when CO gas is introduced, which suggests the
2
The catalytic performances in N over AuPd/S15-pO and AuPd/
exposure of Pd(100) facets and confirms the atomic rearrange-
ment after re-reduction. The B/L value of AuPd/S15-pR is 1.81,
which is slightly higher than that of AuPd/S15-p due to the slight
increase in particle size (6.7 ± 0.9 nm).
S15-pR are recorded within 1-h reaction time, as depicted in
Fig. 10c and d. A dehydrogenation mechanism is verified due to
the formation of benzaldehyde in the absence of oxidant. As the so-
lely detectable by-product over both catalysts, toluene is initially
produced in a moderate amount rather than equimolar to benzalde-
hyde, suggesting that after post-treatments (reoxidation and/or
re-reduction), the formation of toluene probably undergoes a
hydrogenolysis step rather than disproportionation of benzyl alco-
hol, which is suppressed by smaller amounts of surface metal hy-
The catalytic activities of AuPd/S15-pO and AuPd/S15-pR are
compared with that of AuPd/S15-p in Fig. 10. The conversion plots
are fitted with first-order reaction kinetics, showing the rate con-
ꢀ1
stants k as 0.28 ± 0.04 and 0.34 ± 0.08 h for AuPd/S15-pO and
AuPd/S15-pR, respectively. The initial rates r are 0.028 and 0.033
0
for these two samples, as listed in Table 2. Reoxidation and subse-
quent re-reduction show catalytic performances inferior to those of
plasma-reduced AuPd/S15-p catalyst due to the modified morphol-
0
dride. As listed in Table 1, the initial rates r over two catalysts
under anaerobic conditions are similar (AuPd/S15-pO, 0.0024 mol/
h; AuPd/S15-pR, 0.0021 mol/h) and are only half of that of AuPd/
S15-p. The decreased exposure of Pd may account for the low initial
activity, since Pd has been evidenced to be dominantly responsible
in Au–Pd bimetallic catalyst.
ogy and surface chemistry of catalyst after post-treatments. O
2
reoxidation of AuPd/S15-p leads to an apparent deterioration of
activity, particularly a depressed initial reaction rate. Metallic Pd0
AuPd/S15-p
B/L=1.43
AuPd/S15-pO
AuPd/S15-pR
B/L=5.55
B/L=1.81
COgas
2113
Pd
2088
L
2085
L
COgas
2
+
COgas
1956
2001 1939
1992 1946
B
+
+
B
B
Pd -CO-Pd
B
a
b
1
2
a
b
a
1
2
b
c
c
c
d
e
d
d
e
e
L 2068
B 1948
2
f
f
f
g
h
i
g
g
h
i
j
h
i
j
j
2300
2200
2100
2000
1900
1800 2300
2200
2100
2000
1900
1800
2300
2200
2100
2000
1900
1800
-1
-1
-1
wavenumber,cm
wavenumber,cm
wavenumber,cm
Fig. 9. DRIFT spectra of CO adsorbed at 25 °C on AuPd/S15-pO and AuPd/S15-pR, with the results of AuPd/S15-p included for comparison. Sample exposed to 1.1 kPa CO for (a)
min, (b) 5 min, (c) 10 min, (d) 20 min, and (e) 30 min and purged with He for (f) 1 min, (g) 5 min, (h) 10 min, (i) 20 min, and (j) 30 min.
1

1
16
Y. Chen et al. / Journal of Catalysis 289 (2012) 105–117
reaction under O
2
reaction under N
2
a) 100
(c)100
3.0
2.5
(
AuPd/S15-p
AuPd/S15-pO
AuPd/S15-pR
AuPd/S15-pO
r =0.0024 mol/h
0
80
60
40
20
0
80
60
40
20
0
benzaldehyde
2
.0
.5
benzyl alcohol
conversion
1
-kt
1.0
0.5
0.0
x =1-e
toluene
0.8
kAuPd-p=0.50 ± ±0.05
kAuPd-pO=0.28 ± ±0.04
kAuPd-pR=0.33 ± ±0.07
0
.0
0.5
1.0
1.5
2.0
2.5
3.0
0.0
0.2
0.4
0.6
1.0
time,h
time,h
100
3.0
(
b) 100
(d)
AuPd/S15-pR
r0=0.0021 mol/h
2.5
80
60
40
20
0
80
60
40
20
0
benzyl alcohol
conversion
2.0
benzaldehyde
1.5
1.0
0.5
toluene
0.8
AuPd/S15-p
AuPd/S15-pO
AuPd/S15-pR
0.0
1.0
0
1
2
3
0.0
0.2
0.4
0.6
time,h
time,h
Fig. 10. The time–activity profile for selective benzyl alcohol oxidation under aerobic conditions for (a) benzyl alcohol conversion; (b) benzaldehyde selectivity and under
ꢀ
1
2 2
anaerobic conditions for (c) AuPd/S15-pO; (d) AuPd/S15-pR. Reaction conditions: benzyl alcohol/metal = 250/1 mmol/mg; O (N ), 20 mL min ; temperature, 160 °C.
4
. Discussion
generated by the plasma, such as electrons and ions, directly reduce
the metal ions through a recombination process; (ii) an indirect
process involving the excitation (dissociation and ionization) of
The structures of Au, Pd, and Au–Pd catalysts proposed previ-
ously are summarized in Scheme 1. AuPd/S15-p shows Pd-rich
shell/Au-rich core structure with surface enrichment of coordina-
tion-unsaturated Pd atoms. The superior catalytic performance of
AuPd/S15-p to monometallic Au and Pd catalysts can be addressed
from the following two aspects. (i) The changed interatomic
distance of Pd on the surface may have both geometric and
electronic effects. Au can disperse/isolate Pd sites, preventing oxy-
gen poisoning of Pd in the liquid-phase oxidation [10]. Au, with
higher electronegativity, also withdraws electrons from Pd atoms,
leading to an enhanced interaction of Pd atoms with the substrate
and thereby improved activity and selectivity. Recently, it has also
been reported that the presence of nanocontacts triggers electron
transfer and localized charge redistribution in the contact region,
which are highly attractive to charged organic molecules, resulting
in the improvement of organic transformations [37]. (ii) As sug-
gested by Mori et al., low coordination Pd surface sites (edges
and/or corners) are more active than the highly coordinated ter-
race sites for b-H elimination, which is the rate-limiting step in
the oxidative dehydrogenation of alcohols. In this study, AuPd/
S15-p with more surface-coordination-unsaturated Pd atoms, as
evidenced by CO-DRIFT measurements, shows a strong positive ef-
fect toward b-H cleavage, thus leading to high activity.
the chemisorbed water molecular to generate active species (e.g.,
ꢀ
eaq) with strong reducibility, which is normally attributed to
nonhydrogen glow-discharge treatment [38]. In both processes,
metallic nanoparticles can be obtained with regular and uniform
distributions.
Au–Pd bimetallic catalysts are speculated to undergo different
atomic arrangements during Ar glow-discharge plasma treatment
2
at room temperature and H thermal reduction at elevated temper-
ature. Meanwhile, we admit that the catalyst after plasma reduc-
tion may be unstable and undergo a subtle change once it is
exposed to air. Due to the limitations of experimental conditions,
we cannot follow the in situ plasma reduction process at this junc-
ture and obtain real-time information on the catalyst. Therefore, we
2
introduced post-treatment processes (further O reoxidation and
H
2
re-reduction) on both AuPd/S15-p and AuPd/S15-H, and carried
out control experiments for the corresponding Au-Pd catalysts. The
structural information is illustrated in Fig. 1. AuPd/S15-H and AuPd/
S15-HR present similar XRD patterns, indicating the ‘‘stable’’ con-
2
figuration of bimetallic nanoparticles formed by H thermal reduc-
tion at high temperature. In contrast, the bimetallic Au–Pd
nanoparticles in AuPd/S15-p experience a structural transforma-
tion from a Pd-rich shell/Au-rich core structure to a cluster-in-clus-
ter structure. Their catalytic performances are also compared, as
depicted in Fig. 11 and Fig. S3. As shown in Fig. 11a, the plots of con-
version and selectivity almost coincide for AuPd/S15-H and AuPd/
S15-HR. Surprisingly, the plots in Fig. 11b exhibit the dramatic dif-
ference between AuPd/S15-p and AuPd/S15-pH. After post-treat-
ment, the selectivity increases from 67.7% to 81.5% during the
Compared to AuPd/S15-p, both inferior structure and catalytic
activity were observed over AuPd/S15-H prepared by conventional
hydrogen thermal reduction (Scheme 1 and Fig. 7). The advantage
of Ar glow-discharge plasma reduction over H thermal reduction
2
may be attributed to its plausible mechanism, involving two
processes: (i) a direct process, in which the high-energy species

Y. Chen et al. / Journal of Catalysis 289 (2012) 105–117
117
from H2 thermal reduction, affording modified morphology and
surface chemistry of metal nanoparticles. Further oxidation and
re-reduction of plasma-reduced Au–Pd catalyst resulted in the
atomic rearrangement of nanoparticles, leading to inferior catalytic
performance.
a) 100
100
80
60
40
20
0
(
benzaldehyde
selectivity
80
60
40
20
0
Acknowledgments
conversion
Y. Yang acknowledges funding from the Singapore Agency for
Science, Technology and Research (A⁄STAR), SERC Grant 102 101
0020 for financial support. Support from the National Natural Sci-
ence Foundation of China (under Contract 20990223) and the 111
Project of the Ministry of Education of China is also appreciated.
AuPd/S15-H
AuPd/S15-HR
0
.0
0.5
1.0
1.5
2.0
2.5
3.0
time,h
References
(
b) 100
100
80
60
40
20
0
[
[
[
1] G.J. Hutchings, Chem. Commun. (2008) 1148.
2] G.C. Bond, D.T. Thompson, Catal. Rev. Sci. Eng. 41 (1999) 319.
3] Y.T. Chen, H.M. Lim, Q.H. Tang, Y.T. Gao, T. Sun, Q.Y. Yan, Y.H. Yang, Appl. Catal.
A: Gen. 380 (2010) 55.
80
60
40
20
0
benzaldehyde
selectivity
[
4] Y.T. Chen, H.J. Zheng, Z. Guo, C.M. Zhou, C. Wang, A. Borgna, Y.H. Yang, J. Catal.
283 (2011) 34.
[5] M. Haruta, T. Kobayashi, H. Sano, N. Yamada, Chem. Lett. (1987) 405.
[
[
6] T.F. Blackburn, J. Schwartz, J. Chem. Soc. Chem. Commun. (1977) 157.
7] D.I. Enache, J.K. Edwards, P. Landon, B. Solsona-Espriu, A.F. Carley, A.A. Herzing,
M. Watanabe, C.J. Kiely, D.W. Knight, G.J. Hutchings, Science 311 (2006) 362.
conversion
[8] J.K. Edwards, B.E. Solsona, P. Landon, A.F. Carley, A. Herzing, C.J. Kiely, G.J.
Hutchings, J. Catal. 236 (2005) 69.
[
9] K. Asakura, Y. Yamazaki, H. Kuroda, M. Harada, N. Toshima, A Cluster-in-
Cluster Structure of the SiO -Supported PtPd Clusters, Kobe, Japan, 1992. 448.
2
AuPd/S15-p
AuPd/S15-pR
[
10] N. Dimitratos, A. Villa, D. Wang, F. Porta, D.S. Su, L. Prati, J. Catal. 244 (2006)
113.
[
[
[
[
[
11] H.P. Wang, C.J. Liu, Appl. Catal. B: Environ. 106 (2011) 672.
12] Z.J. Wang, Y.B. Xie, C.J. Liu, J. Phys. Chem. C 112 (2008) 19818.
13] Y. Li, R.T. Yang, C.-j. Liu, Z. Wang, Ind. Eng. Chem. Res. 46 (2007) 8277.
14] R.A. Sheldon, I. Arends, A. Dijksman, Catal. Today 57 (2000) 157.
15] R.A. Sheldon, I.W.C.E. Arends, G.-J.t. Brink, A. Dijksman, Acc. Chem. Res. 35
0.0
0.5
1.0
1.5
2.0
2.5 3.0
time,h
Fig. 11. Catalytic performance of Au–Pd catalysts: (a) AuPd/S15-H and AuPd/S15-
HR; (b) AuPd/S15-p and AuPd/S15-pR (solid, conversion; hollow, selectivity).
(
2002) 774.
[
[
16] R.A. Sheldon, J.K. Kochi, Metal-Catalyzed Oxidation of Organic Compounds,
Academic Press, New York, 1981.
17] D.Y. Zhao, J.L. Feng, Q.S. Huo, N. Melosh, G.H. Fredrickson, B.F. Chmelka, G.D.
Stucky, Science 279 (1998) 548.
first 0.5 h, whereas the conversion decreases from 69.0% to 51.0%
after 8 h. This evidence indirectly implies that during the Ar
glow-discharge plasma reduction, the bimetallic nanoparticles un-
dergo a different atomic arrangement of surface atoms, reaching a
highly active but ‘‘unstable’’ structure that can be affected readily
by post-treatment.
[18] Z.J. Wang, Y. Liu, P. Shi, C.J. Liu, Appl. Cataly. B: Environ. 90 (2009) 570.
[
19] C.J. Liu, K.L. Yu, Y.P. Zhang, X.L. Zhu, F. He, B. Eliasson, Appl. Catal. B: Environ.
7 (2004) 95.
20] X. Zhong, Y. Feng, I. Lieberwirth, W. Knoll, Chem. Mater. 18 (2006) 2468.
4
[
[21] T.H. Fleisch, R.F. Hicks, A.T. Bell, J. Catal. 87 (1984) 398.
[
[
22] N. Toshima, T. Yonezawa, New J. Chem. 22 (1998) 1179.
23] T. Akita, T. Hiroki, S. Tanaka, T. Kojima, M. Kohyama, A. Iwase, F. Hori, Catal.
Today 131 (2008) 90.
24] F. Boccuzzi, A. Chiorino, M. Manzoli, P. Lu, T. Akita, S. Ichikawa, M. Haruta, J.
Catal. 202 (2001) 256.
5
. Conclusions
[
[
[
25] A. Hugon, N. El Kolli, C. Louis, J. Catal. 274 (2010) 239.
26] N.F.P. Ribeiro, F.M.T. Mendes, C.A.C. Perez, M. Souza, M. Schmal, Appl. Catal. A:
Gen. 347 (2008) 62.
Bimetallic Au–Pd nanoparticles supported on SBA-15 mesopor-
ous materials were prepared by Ar glow-discharge plasma reduc-
tion. The size of metal nanostructure (diameter of metal
nanoparticles/nanorods) can be effectively controlled and tailored
by the highly ordered mesopores of SBA-15 silica. Adding Au to
Pd to form bimetallic Au–Pd nanocomposites well retains the mor-
phology of spherical nanoparticles with a similar particle size dis-
tribution. As a green, efficient, and safe reduction method, plasma
[27] V. Sanchez-Escribano, L. Arrighi, P. Riani, R. Marazza, G. Busca, Langmuir 22
2006) 9214.
[
[
[30] D.I. Enache, D. Barker, J.K. Edwards, S.H. Taylor, D.W. Knight, A.F. Carley, G.J.
Hutchings, Solvent-Free Oxidation of Benzyl Alcohol Using Titanic-Supported
Gold-Palladium Catalysts: Effect of Au–Pd Ratio on Catalytic Performance,
Limerick, Ireland, 2006. 407.
[31] Q.H. Zhang, W.P. Deng, Y. Wang, Chem. Commun. 47 (2011) 9275.
[32] C. Keresszegi, D. Ferri, T. Mallat, A. Baiker, J. Phys. Chem. B 109 (2005) 958.
(
28] C. Mihut, C. Descorme, D. Duprez, M.D. Amiridis, J. Catal. 212 (2002) 125.
29] G.C. Cabilla, A.L. Bonivardi, M.A. Baltanas, Catal. Lett. 55 (1998) 147.
reduction outperformed conventional H
2
thermal reduction, show-
ing uniform particle dispersions and size distributions. Character-
ization results suggested a Pd-rich shell/Au-rich core structure
with abundant surface-coordination-unsaturated Pd atoms for the
effectively reduced Au–Pd bimetallic nanoparticles, which ac-
counted for the good catalytic activity in the selective oxidation
of benzyl alcohol compared to Au and Pd monometallic catalysts.
At 1 h of reaction, AuPd/S15-p shows a conversion of 47.4% and
[
33] X. Guo, P. Brault, G. Zhi, A.l. Caillard, J. Guoqiang, C. Coutanceau, S. Baranton, X.
Guo, J. Phys. Chem. C 115 (2011) 11240.
[
34] M. Hosseini, S. Siffert, H.L. Tidahy, R. Cousin, J.F. Lamonier, A. Aboukais, A.
Vantomme, M. Roussel, B.L. Su, Catal. Today 122 (2007) 391.
35] I. Eswaramoorthi, A.K. Dalai, Int. J. Hydrogen Energy 34 (2009) 2580.
36] J.D. Grunwaldt, M. Caravati, A. Baiker, J. Phys. Chem. B 110 (2006) 25586.
37] C.L. Jiang, S. Ranjit, Z.Y. Duan, Y.L. Zhong, K.P. Loh, C. Zhang, X.G. Liu, Nanoscale
1 (2009) 391.
[
[
[
ꢀ1
[38] X. Liang, C.J. Liu, P. Kuai, Green Chem. 10 (2008) 1318.
TOF of 16,375 h . We believe that the particle nucleation and
growth mechanism in the plasma reduction process is different