Carbon-supported gold nanocatalysts: Shape effect in the selective glycerol oxidation

Article abstract of DOI:10.1002/cctc.201200535

Gold nanoparticles were supported on two types of carbon nanofibres with different degree of graphitisation. The investigation of these materials with an aberration-corrected transmission electron microscope showed that the degree of the surface graphitis

Full text of DOI:10.1002/cctc.201200535

DOI: 10.1002/cctc.201200535
Carbon-Supported Gold Nanocatalysts: Shape Effect in the
Selective Glycerol Oxidation
Di Wang,*[a, b] Alberto Villa,[a, c] Dangsheng Su,[a] Laura Prati,[c] and Robert Schlçgl[a]
Gold nanoparticles were supported on two types of carbon
nanofibres with different degree of graphitisation. The investi-
gation of these materials with an aberration-corrected trans-
mission electron microscope showed that the degree of the
surface graphitisation strongly influences the structures of the
supported gold particles. The more ordered graphitic layers of
the carbon nanofibre surface led to Au particles more pre-
ferred to immobilise on their {111} plane, exhibiting more
facet area. In contrast, disordered carbon nanofibre surfaces
led to random orientation of supported particles. The different
shape of similarly sized Au nanoparticles allowed determining
the effect of support surface structures on the selectivity of
the catalyst in the liquid-phase oxidation of glycerol, highlight-
ing the higher C3 product selectivity on the {111} surface.
Based on these results, we could also gain new insight in the
effect of Au nanoparticle size on the selectivity in the liquid-
phase oxidation of glycerol, that is, larger particles were more
selective toward C3 products than the smaller ones.
Introduction
Since the discovery of Haruta that supported gold nanoparti-
cles (AuNPs) are highly active catalysts for CO oxidation,[1]
a series of studies have revealed that gold is active in several
other catalytic reactions as well, such as hydrochlorination of
alkynes, production of hydrogen peroxide, and oxidation of al-
cohols and polyols in the liquid phase,[2] if it is highly dispersed
on different supports. Gold catalysts were proved to work also
in homogeneous oxidation with various oxidants including
air.[3] In recent years, gold and gold-based catalysts have at-
tracted more and more interest in their application for catalytic
transformation of biomass-derived chemicals; they have poten-
tial in “green chemistry” through aerobic oxygen as the oxi-
dant instead of other chemical oxidants that produce waste.[4]
The origin of activity is still under debate both in gas- and
liquid-phase reactions; however, activity is generally attributed
to the electronic and geometric structures of nanometre-sized
Au particles, which differ from bulk Au.[5] In contrast to the lim-
ited understanding of the catalytic mechanisms,[6] numerous
experiments have shown that the catalytic activity is highly de-
pendent on the nanostructures of the particles and the sup-
port. Particularly, in the oxidation of CO by Au catalysts hemi-
spherical particles were shown to be more active than spheri-
cal ones, which supports the importance of the perimeter in-
terfaces between AuNPs and the support.[7,2a] The importance
of the interface in the O2 activation has also been revealed by
DFT calculation.[8] Moreover, dynamic environmental TEM stud-
ies have recently shown that AuNPs can wet the support sur-
face differently, depending on the nature of the material, and
the wetting can be also influenced by adsorption of O2.[9] De-
pending on the O2 partial pressure and the size of the AuNPs,
O2 adsorption has been shown to modify the structure of the
particles by rounding the nanocrystals at elevated pressure,
whereas close-packed (111) facets were more stable at low
partial pressure.[10] The support also plays an essential role in
tuning the fine-structures of Au particles. It was also suggested
that highly disordered supports lead to high activity and the
perimeter interfaces could contain oxidic Au species (AuOx)[11]
À
derived from partially charged Au ions binding with OH or O2
species derived from O2 or the support itself. Therefore, it is ex-
pected that several interlocked factors (structure and size, sup-
port, and experimental conditions) determine the overall activi-
ty of supported metal-based catalysts.
Most of the studies dealing with shape effects concern with
gas-phase reactions and, therefore, do not take into consider-
ation the presence of a solvent that could lead to significant
variations in terms of both activity and selectivity. Here, we
have focused on the effect of the support on the selectivity of
glycerol oxidation through modification of the shape of sup-
ported AuNPs. This reaction has been studied by many re-
search groups as an important application of gold-based cata-
lysts.[12] It has been shown that both reaction rate and path are
sensitive to the particle size and the support.[13,12c] The mecha-
[a] Dr. D. Wang, Dr. A. Villa, Dr. D. Su, Prof. R. Schlçgl
Department of Inorganic Chemistry
Fritz Haber Institute of the Max Planck Society
Faradayweg 4-6, 14195 Berlin (Germany)
[b] Dr. D. Wang
Institute of Nanotechnology
Karlsruhe Institute of Technology
Hermann-von-Helmholtz-Platz 1
76344 Eggenstein-Leopoldshafen (Germany)
Fax: (+49)721-608-28976
E-mail: di.wang@kit.edu
[c] Dr. A. Villa, Prof. L. Prati
Dipartimento CIMA L.Malatesta
Universitꢀ di Milano
via Venezian 21, I-20133 Milano (Italy)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cctc.201200535.
ChemCatChem 0000, 00, 1 – 8
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
1
&
ÞÞ
These are not the final page numbers!

D. Wang et al.
nism of the reaction has been studied for Au–activated carbon
(AC) catalysts from theoretical and kinetic viewpoints,[14] high-
lighting the possible active role of the support and of AuNP
step and edge sites in O2 activation. From the activity point of
view, Corma et al. have shown that the reaction rate directly
depends on the surface atom, thus classifying this reaction as
not structure-sensitive.[15] The selectivity of the reaction has
also been studied with respect to the CÀC scission in native
H2O2 production, degradation, or both,[16] as well as by com-
paring glyceric acid production to hydroxyacetone produc-
tion,[17] even with some doubts on the analyses emerging
recently.[18]
Structure of CNFs
In the HRTEM image of the PR24-PS nanofibre in Figure 1a, de-
posited layers of carbon approximately 2 nm thick are shown
on the surface, which consisted of highly defective and discon-
Carbon-based materials, such as AC, carbon nanotubes
(CNTs), and carbon nanofibres (CNFs), have been widely used
as catalyst support because it is possible to manipulate them
by specific chemical and thermal treatments aimed to tune the
surface chemistry. Theoretical studies on the interaction be-
tween Au clusters and graphite shows that the stability and
the structure of Au are highly dependent on the Au cluster
size and the defects on graphitic layers.[19] We have shown that
the exact type of carbon support exerts a strong influence on
the catalytic properties (activity and selectivity) even when the
Au particles were preformed by sol immobilisation methods.[20]
Therefore, to study these findings in more detail and the possi-
ble role of carbon support in controlling the structures of
nanoparticles, supported systems synthesised by a two-step
methodology (generation of nanoparticles followed by their
immobilisation on the support) with two different types of
commercial CNFs were prepared. The correlation between Au
particle structure and the selectivity of glycerol oxidation was
then analysed by using aberration-corrected high-resolution
(HR)TEM.
Figure 1. HRTEM images of a) PR24-PS and b) PR24-LHT CNFs.
tinuous graphitic layers. In contrast to PR24-PS, PR24-LHT ex-
hibited a more graphitised surface of more ordered layers, as
shown in Figure 1b. The layers of deposited carbon were only
approximately 0.5 nm thick. At a distance of approximately 20–
30 nm from the carbon layer surface, a well-ordered graphite-
like structure was seen for both CNFs. This structure had no
direct contact with the Au particle subsequently deposited on
it and the distance to the surface was relatively large; it is thus
expected to have no strong impact on the structure of Au par-
ticles on the CNF surface. In a previous characterisation of
these two carbon supports by Raman spectroscopy, the ratio
of D band (at ꢀ1350 cmÀ1) and G band (at ꢀ1580 cmÀ1) inten-
sities were measured to indicate the graphitisation degree. A
lower D/G band intensity ratio and the appearance of much
more prominent 2D band intensity for PR24-LHT than for
PR24-PS indicated an increasing degree of graphitisation for
PR24-LHT.[23] Therefore, our TEM observation is consistent with
the previous Raman spectroscopic results.
Results and Discussion
Both CNFs used as support in this work have an average diam-
eter of approximately 100 nm. The CNFs denoted as PR24-PS
were produced by pyrolytically stripping the as-produced fibre
to remove polyaromatic hydrocarbons from the fibre surface.
The PR24-LHT were produced by heat-treating the fibre at
15008C. This treatment converts any chemically vapour-depos-
ited carbon present on the surface of the fibre into a more or-
dered graphitic structure.
Structures of gold particles supported on CNFs
Pregenerated polyvinyl alcohol (PVA) protected Au nanoparti-
cles were immobilised on CNFs by impregnation and TEM anal-
yses revealed that the distribution and the size of the Au parti-
cles were similar on PR24-PS and PR24-LHT CNF supports. Al-
though some CNFs were found with few or no Au particles,
most CNFs carried evenly distributed Au particles in a narrow
particle size distribution. The respective overviews of the Au/
PR24-PS and Au/PR24-LHT are shown in Figure 2. The mea-
sured median particle sizes for these two catalysts were (3.7Æ
0.2) nm and (3.5Æ0.1) nm, respectively.
HRTEM and high-angle annular dark field (HAADF) scanning
transmission electron microscopy (STEM) are very important
techniques to study the structure of nanoparticles at atomic
level and have been extensively applied to determine the con-
figuration of supported metal catalysts.[21] With aberration-cor-
rected HRTEM and HAADF STEM, the atomic configurations of
nanoparticles and clusters have been successfully resolved.[22]
We used a FEI Titan 80–300 microscope operating at 300 kV
equipped with a CEOS image Cs corrector to study the struc-
tures of two different CNF supports, the termination of Au cat-
alysts surface, and the dependence of Au nanoparticles mor-
phology on the support structure. The revealed different struc-
tures of supported Au particles were then directly correlated
to the selectivity observed in the aerobic oxidation of glycerol.
We studied the structure of the two catalysts, Au (1 wt%) on
PR24-PS and on PR24-LHT CNFs, by aberration-corrected
HRTEM and found that the final shape of the Au particles was
related to the surface structures of the support, although the
Au particles were preformed in the sol. To obtain a good statis-
&2
&
www.chemcatchem.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 8
ÝÝ
These are not the final page numbers!

Carbon-Supported Gold Nanocatalysts
Figure 2. TEM overview images of a) Au/PR24-PS and b) Au/PR24-LHT.
tic evaluation on the particle morphology, 167 particles with
metal-support interface parallel to the electron beam were
imaged in the case of Au/PR24-PS. The median size of these
particles was (3.8Æ0.2) nm, which is in good agreement with
the evaluation from the overview images. Therefore, the statis-
tic evaluation was representative. The Au particles were at-
tached to the PR24-PS nanofibres without distinguishable ori-
entation relation between Au low-index crystallographic
planes and the CNF surface. Among the examined 167 parti-
cles, only 36 particles (21.6%) were found to have one set of
{111} planes parallel to the CNF surface and the rest were ran-
domly oriented (Figure 3). Most of the particles exhibited
a polyhedral shape, for example, cuboctahedron, decahedron,
icosahedron, and truncated alternatives. In Figure 4a and 4b,
two Au particles of approximately 3 nm size are seen with cu-
boctahedron and icosahedron shape, respectively. The Au par-
ticle in cuboctahedron shape has clean surfaces. However,
a closer inspection of the intersection of the two {111} surfaces
and two {100} surfaces revealed one missing atom (marked by
an arrow). Furthermore, this particle is situated on the CNF sur-
face not with its (100) plane or (111) plane parallel to the CNF
surface but it is inclined. Consequently, the atoms around the
intersection ridge of the (100) and (111) surfaces in contact
with CNF are missing so that the interface between the parti-
cle and CNF is smoothed without any sharp edge. Such config-
Figure 4. Aberration-corrected HRTEM images of Au particles supported on
PR24-PS with a) modified cuboctahedral configuration, b) icosahedral config-
uration showing interaction between particle and CNF surface, and c) sur-
face carbon decoration with uncovered windows (each indicated by arrows).
uration reduces the contacting area. The interatomic distances
between the atoms in the bottom layer and the ones in the
neighbouring sublayer were measured and compared with the
distances between the atoms at inner planes. No systematic
contraction or expansion was observed for the carbon-contact-
ing atomic layer, suggesting that no C diffusion into the sub-
surface layer of Au particles has occurred. In Figure 4b, the ico-
sahedron nanoparticle was imaged on neither its six-fold nor
its five-fold axes, therefore it exhibits a complicated projected
contrast. Again, the modification of the interface between the
particle and the CNF support can be seen. The contacting
plane on the particle is more flattened than the side exposed
to the vacuum and some carbon decoration is observed at the
perimeter of the interface (indicated by the arrow in Fig-
ure 4b), which is attributed to either strong metal-carbon in-
teraction or the residual protection agent, PVA. These two ex-
amples offer direct evidence that both metal particles and CNF
support undergo considerable reconstruction, which modifies
the exposed surfaces and interfaces.
It was also frequently observed that the particle was well-
covered by a layer of carbonaceous species, but usually not
completely. In Figure 4c such an example of a particle decorat-
ed with the layer is seen. In the places indicated by arrows,
windows without covering can be distinguished, which owes
to the residual PVA applied during the formation of the metal-
lic Au nanoparticles. Besides the particle size control and parti-
cle stabilisation, the protecting agent may exert additional ef-
fects on the properties of metal catalysts through offering
binding between particle and support and acting as “cement”.
It is therefore apparent that the contact between the metal Au
and the support surface is mediated by the nanoparticle pro-
Figure 3. Statistical evaluation of the orientation of 167 particles from Au/
&
PR24-PS and 218 particles from Au/PR24-LHT imaged by HRTEM. : particles
&
with (111) planes parallel to CNF surface; : particles with other orientation.
The Au/PR24-LHT shows a much higher percentage of particles with (111)
planes parallel to the CNF surface than the Au/PR24-PS.
ChemCatChem 0000, 00, 1 – 8
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemcatchem.org
&
3
&
ÞÞ
These are not the final page numbers!

D. Wang et al.
tector (PVA). This is in agreement with the observation by
Hutchings et al. reporting on the different degree of wettability
between carbon and TiO2.[24] From Figure 4c, we can see that
the surface decoration is bound directly to the CNF support at
the perimeter of the interface. The holes in the covering are
essential to the accessibility of the reactant to the particle.
Moreover, it should be noticed that such carbonaceous surface
decoration also makes the particle surface rougher so that it
has less flat facets and more uncoordinated sites.
The most noticeable difference between Au/PR24-LHT and
Au/PR24-PS was that statistically the {111} planes of Au parti-
cles on PR24-LHT are more frequently observed to be parallel
to the CNF surface than in the case of PR24-PS. In the case of
Au/PR24-LHT, 218 particles were imaged in HRTEM images. The
median size of these particles was (3.6Æ0.2) nm, which is in
agreement with the evaluation from overview images. Among
these particles, 102 particles (46.8%) were found to have one
set of {111} planes parallel to the CNF surface (Figure 3). This
percentage was much higher than in the case of Au/PR24-PS
(21.6%). The correlation was extraordinarily prominent for Au
particles with sizes smaller than 3 nm. Below this size, the per-
centage of the oriented particles reached 72.1%, in contrast to
44.8% for the particles with size between 3 nm and 4 nm, and
only 21.2% for the particles larger than 4 nm. The histogram of
size distribution for oriented particles and all particles is shown
in Figure 5. In Figure 6a a representative particle with one set
of {111} lattice planes on the CNF surface is shown. The parti-
cle has a configuration similar to the cuboctahedron shape
with {111} and {100} surfaces exposed. A structural model was
constructed and is shown in Figure 6b. The particle has a rela-
tively large (111) base plane contacting with the supporting
CNF. However, the contact angle is larger than 908 and there is
no distinguishable carbonaceous adlayer binding the Au metal
particle to the CNF support. Moreover, a stacking fault is ob-
served at the second stacking plane on the CNF. Our observa-
tions strongly suggest that the more graphitised CNF surface
stabilises the first few layers of Au (111) stacks and the whole
particle keeps more or less the regular shape and less surface
decorated carbon is found on such particle. This tendency is
more prominent for smaller particles. Other examples of such
kind of regularly shaped Au particles exposing the low index
Figure 6. Aberration-corrected HRTEM images of Au particles supported on
PR24-LHT. a) A representative 2–4 nm particle with (111) surface epitaxially
parallel to the graphitic layer of CNF. b) Structure model derived from the
image in part a. c) A larger particle with spherical shape showing carbon
binding to the CNF surface.
surfaces are reported in Figure S1 in the Supporting Informa-
tion. Their {111} planes are more or less parallel to the CNF
surfaces, although no strict epitaxial relation with the undulat-
ing graphitic layers could be addressed. To further confirm that
such tendency is related with highly graphitic carbon surface,
Au particles deposited on thin amorphous C film were also in-
vestigated and there was no orientation preference between
Au particles and amorphous
Figure S2).
C film (HRTEM image in
Similarly to the case of Au/PR24-PS, the lattice spacing in Au
particles was constant from top to bottom within the accuracy
of the measurement, which means that no carbon was dis-
solved in the Au particle. Statistically, particles larger than
4 nm were more inclined to form spherical shape than the one
shown in Figure 6c. Among the 218 imaged particles, 52 parti-
cles (25%) were larger than 4 nm, 39 of them (75%) were
found in spherical shape. Owing to the relatively low fraction
of interface area, carbon binding is essential for the stabilisa-
tion of these particles on the support surface against leaching
and coarsening. Previous experimental data showed that parti-
cle dispersion increased with the number of acidic groups,
suggesting that such particles are positioned on top of stable
oxygen functional groups owing to the “oxophilicity” of
gold.[25]
Liquid-phase oxidation of glycerol
The two Au/CNFs prepared were very similar in particle size
and distribution but showed a prevalent different shape. This
finding allowed us to investigate more detailed the well-
Figure 5. Histogram of size distributions of the particles of the Au/PR24-LHT
&
&
system. : epitaxially oriented particles; : all particles.
&4
&
www.chemcatchem.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 8
ÝÝ
These are not the final page numbers!

Carbon-Supported Gold Nanocatalysts
known particle size effect observed in the liquid-phase oxida-
tion of glycerol.[12] This reaction has been studied by several
groups owing to its importance in the valorisation of biomass-
based processes.[12d] Glycerol can be smoothly oxidised under
basic or neutral conditions.[12a–c,16c] On the basis of experimen-
tal evidences, it is now accepted that the larger the AuNP size
is, the more the activity decreases.[12a–c,16a,26] The selectivity of
the reaction toward C3 products (glyceric, tartronic acids) ap-
pears to be related to the particle size as well as the activity
but in the opposite sense.[12a–c,16a,26] We tested the Au/CNFs
(PR24-PS and PR24-LHT) under the usual conditions reported
observed that Au/PR24-PS presents a similar selectivity toward
glyceric acid to that observed for similarly sized AuNPs sup-
ported on AC (58%). Formate and glycolate, by probing the
CÀC bond cleavage, were formed with selectivities of 19 and
22%, respectively. Conversely, Au/P24-LHT produced these
compounds with considerably higher selectivity (40 and 35%,
respectively). Considering, as highlighted before, that the
AuNPs were prepared by the same methodology with similar
particle sizes and distribution, we concluded that the shape
and the configuration of the Au particles are important factors
that determine selectivity. In the reaction described herein, the
specific particle shapes and configurations were induced by
the supports onto the preformed AuNPs. In particular, in glyc-
erol oxidation, the direct contact between PR24-LHT and {111}
surface of Au(PVA) nanoparticles led to the exposure of low-
index Au surfaces, suggesting that the active sites
in the literature (glycerol 0.3m, NaOH/glycerol=4 molmolÀ1
,
304 kPa O2, T=508C) to determine the shape effect on the ac-
tivity and selectivity of the reaction. The results, reported in
Table 1, confirmed that the catalyst activities mainly depended
on these surfaces could promote the C1 and C2
Table 1. Catalytic results of Au/CNFs in the liquid-phase oxidation of glycerol.[a]
products, that is, the CÀC bond cleavage.
To find a possible explanation of this result, we
Catalyst
Au particle TOF
Selectivity [%]
Glyceric acid Formic acid Glycolic acid Oxalic acid
also investigated the behaviour of AuNPs on amor-
phous AC. In this case, AuNPs deposited on amor-
phous carbon as shown in Figure S2 did not show
a preferential orientation. In addition, their activity
and selectivity (conversion 91%, selectivity to glycer-
ate 58%; see Table 1) were very similar to those of
Au/PR24-PS. Therefore, we concluded that a funda-
mental factor in ruling the overall selectivity is the
preferentially exposure facets of the AuNPs. Tenta-
tively we could correlate the high selectivity toward
the CÀC cleavage shown by low-indexed surface to
size
[hÀ1
]
[nm]
Au/CNFs-PS 3.7
Au/CNFs-LHT 3.5
948 56
982 22
950 58
19
40
20
22
35
21
2
2
1
Au/AC
3.6
[a] Glycerol solution (0.3m with a ratio NaOH/glycerol=4 molmolÀ1), and the catalyst
(ratio substrate/metal=1000 molmolÀ1) were mixed in distilled water (total volume
10 mL). The reactor was pressurised at 304 kPa of O2 and thermostatted at 508C.
on particle size and distribution with similar activities of the
two catalysts. The activities in both cases were high and full
conversion was obtained within 1 h reaction time. However, at
closer inspection, we observed strong differences between the
two catalysts, which can be mainly addressed to the different
shape of the AuNPs, whereas the size and the distribution
were very similar. We cannot completely exclude a support
effect on the reactivity, but the similarity of the CNF employed
made this effect reasonably small. From the data of Table 1, we
the H2O2 production. Native H2O2 has been recognised as one
of the main factors affecting selectivity, which decreases as
H2O2 production and stability increase.[16] It has been shown re-
cently that Au (111) surfaces are the most suitable for H2O2
direct synthesis.[27] According to the widely accepted mecha-
nism shown in Scheme 1, H2O2 is formed through O2 reduction
by metal hydride. In our case H2O2 could form faster or decom-
pose to a lesser extent on Au (111) sites than on other sites.
Therefore, the local higher presence of H2O2, if there are Au
Scheme 1. Mechanism of H2O2 formation on Au through O2 reduction.
ChemCatChem 0000, 00, 1 – 8
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemcatchem.org
&
5
&
ÞÞ
These are not the final page numbers!

D. Wang et al.
Glycerol oxidation: glycerol (0.3m), and the catalyst (substrate/
total metal=1000 molmolÀ1) were mixed in distilled water (total
volume 10 mL) with NaOH (NaOH/glycerol=4 molmolÀ1). The reac-
tor was pressurised at 300 kPa of O2 and set to 508C. Once this
temperature was reached, the gas supply was switched to O2 and
the monitoring of the reaction started. The reaction was initiated
by stirring. Samples were removed periodically and analysed by
HPLC with a column (Alltech OA- 10308, 300ꢁ7.8 mm) with UV
and refractive index (RI) detection to analyse the sample mixture.
Aqueous H3PO4 solution (0.1 wt%) was used as the eluent. Prod-
ucts were identified by comparison with the original samples.
(111) facets preferentially exposed as it was the case in Au/
PR24-LHT, could promote the CÀC bond cleavage and enhance
the selectivity to glycolate and formate. In all cases no leaching
of Au was observed by inductively coupled plasma measure-
ments on the collected solution after reaction.
Conclusions
In a carefully designed synthesis with preformed Au nanoparti-
cles (AuNPs) from the same batch (i.e., with the same narrow
particle size distribution) immobilised onto different carbon
supports, we demonstrated that the shapes of the AuNPs de-
pended on the different interaction with the support. The
methodology enables the exclusive study of a single factor, for
example, particle size, distribution, or support, which are often
convoluted in complex catalysts. By aberration-corrected high-
resolution TEM, the structures of Au particle surfaces, including
added or missing atoms, the deviation from equilibrium lattice
sites and uncoordinated sites, and the decoration at the inter-
face were imaged directly.
CNFs were purchased from Apply Science Company. They have an
average diameter of approximately 100 nm and a minimal CVD
layer of carbon on the surface of the fibre over a graphitic tubular
core. The CNFs denoted as PR24-PS were produced by pyrolytically
stripping the as-produced fibre to remove polyaromatic hydrocar-
bons from the fibre surface. The PR24-LHT CNFs were produced by
heat-treating the fibre at 15008C. This converts any CVD carbon
present on the surface of the fibre to a short-range-ordered
structure.
HRTEM images were acquired by using a FEI Titan 80–300 transmis-
sion electron microscope with CEOS image Cs corrector. It was op-
erated at accelerating voltage of 300 kV. The TEM samples of both
Au/CNF catalysts were prepared by directly putting the powder on
Cu grids coated with holey carbon film, without using any ultra-
sonic dispersion in any solvent to avoid any possible surface deco-
ration on Au particles.
The catalytic performance in the glycerol oxidation reaction
could be related directly to the structural difference observed.
Particularly, we focused on the activity of the catalysts deter-
mined principally by the particle size and distribution because
similar activities were observed for the three catalysts Au/
PR24-PS, Au/PR24-LHT, and Au particles on amorphous activat-
ed carbon. However, the selectivities in the reaction differed,
which was attributed to the preferential orientation of AuNPs
and the particle shapes. We concluded that Au/PR24-LHT was
more selective toward CÀC bond cleavage because more or-
dered graphitic layers of the CNF surface led to Au particles
bonded through their {111} planes thus exhibiting more facet
area. These findings demonstrate the importance of particle
shape in determining the selectivity of a catalytic reaction and
provide new insight into the support in heterogeneous catalyt-
ic reactions.
Acknowledgements
The authors acknowledge DAAD and CRUI for financial support
to personnel exchange program. The authors also acknowledge
Euminafab for part of the TEM characterisation carried out in KIT,
Campus North.
Keywords: carbon · glycerol · gold · oxidation · transmission
electron microscopy
[1] M. Haruta, T. Kobayashi, H. Sano, N. Yamada, Chem. Lett. 1987, 16, 405–
408.
[2] a) G. C. Bond, D. T. Thompson, Catal. Rev. Sci. Eng. 1999, 41, 319–388;
b) M. Haruta, Gold Bull. 2004, 37, 27–36; c) G. J. Hutchings, Gold Bull.
2004, 37, 3–11; d) A. S. K. Hashmi, G. J. Hutchings, Angew. Chem. 2006,
118, 8064–8105; Angew. Chem. Int. Ed. 2006, 45, 7896–7936.
[3] a) A. S. K. Hashmi, C. Lothschuetz, M. Ackermann, R. Doepp, S. Anan-
tharaman, B. Marchetti, H. Bertagnolli, F. Rominger, Chem. Eur. J. 2010,
16, 8012–8019; b) A. S. K. Hashmi, M. C. B. Jaimes, A. M. Schuster, F. Ro-
minger, J. Org. Chem. 2012, 77, 6394–6408.
[4] a) A. Behr, J. Eilting, K. Irawadi, J. Leschinski, F. Lindner, Green Chem.
2008, 10, 13–30; b) R. A. Sheldon, I. W. C. E. Arends, A. Dijksman, Catal.
Today 2000, 57, 157–166; c) R. A. Sheldon, I. W. C. E. Arends, G. J. ten
Brink, A. Dijksman, Acc. Chem. Res. 2002, 35, 774–781.
[5] a) M. Valden, X. Lai, D. W. Goodman, Science 1998, 281, 1647–1650;
b) H. Tsunoyama, H. Sakurai, Y. Negishi, T. Tsukuda, J. Am. Chem. Soc.
2005, 127, 9374–9375.
[6] A. S. K. Hashmi, Angew. Chem. 2010, 122, 5360–5369; Angew. Chem. Int.
Ed. 2010, 49, 5232–5241.
Experimental Section
The metallic Au nanoparticles was first prepared by the sol
method.[28] Solid NaAuCl4·2H2O (0.043 mmol) and PVA solution
[2 wt%, 103 mL, PVA/Au (w/w)=1] were added to H2O (130 mL).
After 3 min, NaBH4 solution (0.1m, 1.3 mL) was added to the
yellow solution under vigorous magnetic stirring. The ruby-red Au0
sol was immediately formed.
Within a few minutes of sol generation, the Au nanoparticles (acidi-
fied until pH 2 by sulfuric acid) were immobilised by adding the
CNF support under magnetic stirring. After additional 18 h, the
slurry was filtered and the catalyst washed thoroughly with dis-
tilled water. ICP analyses were performed on the filtrate by a Jobin
Yvon JV24 to verify the metal loading on carbon. The total metal
loading was 1 wt%.
[7] M. Haruta, J. New Mater. Electrochem. Syst. 2004, 7, 163–172.
[8] M. Boronat, A. Corma, Dalton Trans. 2010, 39, 8538–8546.
[9] S. Giorgio, M. Cabiꢂ, C. R. Henry, Gold Bull. 2008, 41, 167–173.
[10] F. Mittendorfer, N. Seriani, O. Dubay, G. Kresse, Phys. Rev. B 2007, 76,
233413–233416.
Reactions were carried out in a 30 mL glass reactor equipped with
a thermostat and an electronically controlled magnetic stirrer con-
nected to a 5000 mL reservoir charged with oxygen (300 kPa). The
oxygen uptake was monitored by using a mass flow controller con-
nected to a PC through an A/D board to plot a flow time diagram.
[11] G. C. Bond, D. T. Thompson, Gold Bull. . 2000, 33, 41–50.
&6
&
www.chemcatchem.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 8
ÝÝ
These are not the final page numbers!

Carbon-Supported Gold Nanocatalysts
[12] a) S. Carrettin, P. McMorn, P. Johnston, K. Griffin, C. J. Kiely, G. J. Hutch-
ings, Phys. Chem. Chem. Phys. 2003, 5, 1329–1336; b) F. Porta, L. Prati, J.
Catal. 2004, 224, 397–403; c) S. Demirel-Gꢃlen, M. Lucas, P. Claus, Catal.
[21] a) A. K. Datye, D. J. Smith, Catal. Rev. Sci. Eng. 1992, 34, 129–178; b) S. J.
Mejꢆa-Rosales, C. Fernꢇndez-Navarro, E. Pꢂrez-Tijerina, D. A. Blom, L. F.
Allard, M. Josꢂ-Yacamꢇn, J. Phys. Chem. C 2007, 111, 1256–1260; c) J. M.
Thomas, P. L. Gai, Adv. Catal. 2004, 48, 171–227; d) D. Wang, A. Villa, F.
Porta, L. Prati, D. S. Su, J. Phys. Chem. C 2008, 112, 8617–8622.
[22] a) Z. Y. Li, N. P. Young, M. Di Vece, S. Palomba, R. E. Palmer, A. L. Bleloch,
B. C. Curley, R. L. Johnston, J. Jiang, J. Yuan, Nature 2008, 451, 46–49;
b) T. Ling, L. Xie, J. Zhu, H. Yu, H. Ye, R. Yu, Z. Cheng, L. Liu, L. Liu, G.
Yang, Z. Cheng, Y. Wang, X. Ma, Nano Lett. 2009, 9, 1572–1576; c) S.
Chen, W. Sheng, N. Yabuuchi, P. J. Ferreira, L. F. Allard, Y. Shao-Horn, J.
Phys. Chem. C 2009, 113, 1109–1125; d) D. S. Su, T. Jacob, T. W. Hansen,
D. Wang, R. Schlçgl, B. Freitag, S. Kujawa, Angew. Chem. 2008, 120,
5083–5086; Angew. Chem. Int. Ed. 2008, 47, 5005–5008.
´
Today 2005, 102–103, 166–172; d) B. Katryniok, H. Kimura, E. Skrzynska,
J.-S. Girardon, P. Fongarland, M. Capron, R. Ducoulombier, N. Mimura, S.
Paula, F. Dumeignil, Green Chem. 2011, 13, 1960–1979.
[13] a) C. L. Bianchi, S. Biella, A. Gervasini, L. Prati, M. Rossi, Catal. Lett. 2003,
85, 91–96; b) A. Villa, A. Gaiassi, I. Rossetti, C. L. Bianchi, K. van Ben-
them, G. M. Veith, L. Prati, J. Catal. 2010, 275, 108–116.
[14] a) B. N. Zope, D. D. Hibbits, M. Neurock, R. J. Davis, Science 2010, 330,
74–78; b) C. Shang, Z.-P. Liu, J. Am. Chem. Soc. 2011, 133, 9938–9947.
[15] A. Abad, A. Corma, H. Garcia, Chem. Eur. J. 2008, 14, 212–222.
[16] a) W. C. Ketchie, Y. Fang, M. S. Wong, M. Murayama, R. J. Davis, J. Catal.
2007, 250, 94–101; b) L. Prati, P. Spontoni, A. Gaiassi, Top. Catal. 2009,
52, 288–296; c) A. Villa, G. M. Veith, L. Prati, Angew. Chem. 2010, 122,
4601–4604; Angew. Chem. Int. Ed. 2010, 49, 4499–4502.
[23] J.-P. Tessonnier, D. Rosenthal, T. W. Hansen, C. Hess, M. E. Schuster, R.
Blume, F. Girgsdies, N. Pfꢅnder, O. Timpe, D. S. Su, R. Schlçgl, Carbon
2009, 47, 1779–1798.
[17] a) S. Demirel, K. Lehnert, M. Lucas, P. Claus, Appl. Catal. B 2007, 70,
637–643; b) E. G. Rodrigues, M. F. R. Pereira, J. J. Delgado, X. Chen,
J. J. M. ꢄrf¼o, Catal. Commun. 2011, 16, 64–69; c) E. G. Rodrigues,
S. A. C. Carabineiro, J. J. Delgado, X. Chen, M. F. R. Pereira, J. J. M. ꢄrf¼o,
J. Catal. 2012, 285, 83–91.
[24] L. Kesavan, R. Tiruvalam, M. Hasbi Ab Rahim, M. Izham bin Saiman, D. I.
Enache, R. L. Jenkins, N. Dimitratos, J. A. Lopez-Sanchez, S. H. Taylor,
D. W. Knight, C. J. Kiely, G. J. Hutchings, Science 2011, 331, 195–199.
[25] L. Prati, A. Villa, C. E. Chan-Thaw, R. Arrigo, D. Wang, D. S. Su, Faraday
Discuss. 2011, 152, 353–365.
[18] a) E. G. Rodrigues, M. F. R. Pereira, J. J. M. ꢄrf¼o, Catal. Commun. 2012,
25, 110; b) E. G. Rodrigues, S. A. C. Carabineiro, J. J. Delgado, X. Chen,
M. F. R. Pereira, J. J. M. ꢄrf¼o, J. Catal. 2012, 292, 246–247.
[19] a) K. P. McKenna, A. L. Shluger, J. Phys. Chem. C 2007, 111, 18848–
18852; b) J. Akola, H. Hꢅkkinen, Phys. Rev. B 2006, 74, 165404–165412;
c) K. Jiang, A. Eitan, L. S. Schadler, P. M. Ajayan, R. W. Siegel, N. Grobert,
M. Mayne, M. Reyes-Reyes, H. Terrones, M. Terrones, Nano Lett. 2003, 3,
275–277.
[26] N. Dimitratos, J. A. Lopez-Sanchez, D. Lennon, F. Porta, L. Prati, A. Villa,
Catal. Lett. 2006, 108, 147–153.
[27] Y.-G. Wang, J.-G. Wang, Catal. Lett. 2012, 142, 601–607.
[28] a) D. Wang, A. Villa, F. Porta, D. S. Su, L. Prati, Chem. Commun. 2006,
1956–1958; b) N. Dimitratos, A. Villa, D. Wang, F. Porta, D. S. Su, L. Prati,
J. Catal. 2006, 244, 113–121.
[20] a) F. Porta, L. Prati, M. Rossi, G. Scari, J. Catal. 2002, 211, 464–469; b) S.
Gil, L. Munoz, L. Sanchez-Silva, A. Romero, J. L. Valverde, Chem. Eng. J.
2011, 172, 418–429; c) E. G. Rodrigues, M. F. R. Pereira, X. Chen, J. J. Del-
gado, J. J. M. ꢄrf¼o, J. Catal. 2011, 281, 119–127.
Received: August 6, 2012
Published online on && &&, 0000
ChemCatChem 0000, 00, 1 – 8
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemcatchem.org
&
7
&
ÞÞ
These are not the final page numbers!

FULL PAPERS
D. Wang,* A. Villa, D. Su, L. Prati,
R. Schlçgl
Order, order! More ordered graphitic
layers on the supporting carbon nano-
fibers surface led to Au particles
bonded through their {111} plane, ex-
hibiting more facet area. This catalyst
presented higher selectivity toward CÀC
bond cleavage in the liquid-phase oxi-
dation of glycerol.
&& – &&
Carbon-Supported Gold
Nanocatalysts: Shape Effect in the
Selective Glycerol Oxidation
&8
&
www.chemcatchem.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 8
ÝÝ
These are not the final page numbers!