pubs.acs.

org/Langmuir
2010 American Chemical Society

Direct Synthesis of Hydrogen Peroxide and Benzyl Alcohol Oxidation Using

Au-Pd Catalysts Prepared by Sol Immobilization
James Pritchard, Lokesh Kesavan, Marco Piccinini, Qian He, Ramchandra Tiruvalam,
Nikolaos Dimitratos, Jose A. Lopez-Sanchez, Albert F. Carley, Jennifer K. Edwards,
Christopher J. Kiely, and Graham J. Hutchings*,

School of Chemistry, Cardiff University, Main Building, Park Place, Cardiff CF10 3AT, United Kingdom, and

Center for Advanced Materials and Nanotechnology, Lehigh University, 5 East Packer Avenue, Bethlehem,
Pennsylvania 18015-3195

Received April 21, 2010. Revised Manuscript Received April 28, 2010

We report the preparation of Au-Pd nanocrystalline catalysts supported on activated carbon prepared via a sol-
immobilization technique and explore their use for the direct synthesis of hydrogen peroxide and the oxidation of benzyl
alcohol. In particular, we examine the synthesis of a systematic set of Au-Pd colloidal nanoparticles having a range of
Au/Pd ratios. The catalysts have been structurally characterized using a combination of UV-visible spectroscopy,
transmission electron microscopy, STEM HAADF/XEDS, and X-ray photoelectron spectroscopy. The Au-Pd
nanoparticles are found in the majority of cases to be homogeneous alloys, although some variation is observed in
the AuPd composition at high Pd/Au ratios. The optimum performance for the synthesis of hydrogen peroxide is
observed for a catalyst having a Au/Pd 1:2 molar ratio. However, the competing hydrogenation reaction of hydrogen
peroxide increases with increasing Pd content, although Pd alone is less effective than when Au is also present.
Investigation of the oxidation of benzyl alcohol using these materials also shows that the optimum selective oxidation to
the aldehyde occurs for the Au/Pd 1:2 molar ratio catalyst. These measured activity trends are discussed in terms of the
structure and composition of the supported Au-Pd nanoparticles.

Introduction particularly effective for the direct synthesis of hydrogen peroxide

Supported metal nanoparticles are receiving significant atten-
tion as catalysts for a broad range of selective chemical transfor-
mations. In particular, Au nanoparticles have been shown to be
particularly effective for a range of redox reactions.1-7 These
include the low temperature oxidation of CO,7 especially when
used as part of the processes for purifying a fuel cell feedstock,8
the synthesis of vinyl chloride by the hydrochlorination of ethyne,9
the selective oxidation of alkenes to epoxides10,11 and alcohols to
aldehydes,12 as well as selective hydrogenation.13 Recently, we have
shown that the alloying of Pd with Au can enhance the activity of
the nanoparticles, and such materials have been found to be
Part of the Molecular Surface Chemistry and Its Applications special
issue.
*To whom correspondence should be addressed. Telephone: 44 29 2087
4059. Fax: 44 29 2087 4059. E-mail: hutch@cardiff.ac.uk.
(1) Bond, G. C.; Thompson, D. T. Catal. Rev.-Sci. Eng. 1999, 41, 319388.
(2) Bond, G. C.; Thompson, D. T. Gold Bull. 2000, 33, 4151.
(3) Haruta, M. Gold Bull. 2004, 37, 2736.
(4) Hashmi, A. S. K. Gold Bull. 2004, 37, 5165.
(5) Meyer, R.; Lemaire, C.; Shaikutdinov, Sh. K.; Freund, H.-J. Gold Bull. 2004,
37, 72124.
(6) Hashmi, A. S. K.; Hutchings, G. J. Angew. Chem., Int. Ed. 2006, 45, 7896
7936.
(7) Haruta, M.; Kobayashi, T.; Sano, H.; Yamada, N. Chem. Lett. 1987, 16,
405408.
(8) Landon, P.; Ferguson, J.; Solsona, B. E.; Garcia, T.; Carley, A. F.; Herzing,
A. A.; Kiely, C. J.; Golunski, S. E.; Hutchings, G. J. Chem. Commun. 2005, 3385
3387.
(9) Hutchings, G. J. J. Catal. 1985, 96, 292295.
(10) Sinha, A. K.; Seelan, S.; Tsubota, S.; Haruta, M. Angew. Chem., Int. Ed.
2004, 43, 15461548.
(11) Hughes, M. D.; Xu, Y.-J.; Jenkins, P.; McMorn, P.; Landon, P.; Enache,
D. I.; Carley, A. F.; Attard, G. A.; Hutchings, G. J.; King, F.; Stitt, E. H.;
Johnston, P.; Griffin, K.; Kiely, C. J. Nature 2005, 437, 11321135.
(12) Abad, A.; Conception, P.; Corma, A.; Garcia, H. Angew. Chem., Int. Ed.
2005, 44, 40664069.
(13) Corma, A.; Serna, P. Science 2006, 313, 332334.
from its elements14,15 and the oxidation of alcohols.16 In this
paper, we examine supported Au-Pd alloy nanoparticles prepared
by sol-immobilization for two important classes of redox reac-
tion; the direct synthesis of hydrogen peroxide and the oxidation
of benzyl alcohol are chosen as two representative reactions.
Hydrogen peroxide (H2O2) is a major commodity chemical15
and has significant uses in cleaning and bleaching. At present,
H2O2 is produced using an indirect process which involves the
sequential hydrogenation and oxidation of an alkyl anthraqui-
none, thereby avoiding the potential for explosive contact be-
tween hydrogen and oxygen to occur.17 The direct synthesis of
hydrogen peroxide is not a facile process, since the formation of
water is preferred under all conditions, and the quest to find a
suitable catalyst for the direct reaction between H2 and O2 has
remained a challenge for almost a century, and as such it has
attracted considerable attention. Supported Pd catalysts have
been the focus of much of the research.18-26 We have found14,15
(14) Edwards, J. K.; Solsona, B.; Ntainjua, E.; Carley, A. F.; Herzing, A. A.;
Kiely, C. J.; Hutchings, G. J. Science 2009, 323, 10371041.
(15) Edwards, J. K.; Hutchings, G. J. Angew. Chem., Int. Ed. 2008, 47, 91929198.
(16) Enache, D. I.; Edwards, J. K.; Landon, P.; Solsona-Espriu, B.; Carley,
A. F.; Herzing, A. A.; Watanabe, M.; Kiely, C. J.; Knight, D. W.; Hutchings, G. J.
Science 2006, 311, 362365.
(17) Hess, H. T. In Kirk-Othmer Encyclopaedia of Chemical Engineering;
Kroschwitz, I., Howe-Grant, M., Eds.; Wiley: New York, 1995; Vol. 13, p 961.
(18) Henkel, H.; Weber, W. US Patent No. 1108752, 8, 25, 1914.
(19) Van Weynbergh, J.; Schoebrechts, J. P.; Colery, J. C. US Patent No.
5447706, 9, 5, 1995.
(20) Paparatto, G.; DAloisio, R.; De Alberti, G.; Furlan, P.; Arca, V.; Buzzoni,
R.; Meda, L. Eur. Patent No. 0978316A1, 7, 29, 1999.
(21) Zhou, B.; Lee, L.-K. US Patent No. 6168775, 1, 2, 2001.
(22) Nystrom, M.; Wangard, J.; Herrmann, W. US Patent No. 6210651, 4, 3,
2001.
(23) Lunsford, J. H. J. Catal. 2003, 216, 455460.
(24) Dissanayake, D. P.; Lunsford, J. H. J. Catal. 2002, 206, 173176.

16568 DOI: 10.1021/la101597q Published on Web 05/12/2010 Langmuir 2010, 26(21), 1656816577
Pritchard et al. Article

that a combination of Pd with Au promotes both the activity and

selectivity for the direct H2O2 synthesis reaction. Furthermore,
these catalysts are also very effective for the oxidation of alcohols,
and once again the addition of Pd to Au increases both selectivity
and activity.16
In our early studies,14,15 we used a relatively straightforward
wet coimpregnation method to prepare gold-palladium nano-
particles and concluded that either homogeneous alloys or core-
shell structures could be formed. We have found that relatively
small particles14 may be associated with the reactivity for the H2O2
reaction when supported on carbon. However, relatively low
number densities of these ultrasmall alloy particles are formed
using the coimpregnation method.14,15 Preforming colloidal nano-
particles could overcome this problem if the size distribution can
be retained when the nanoparticles are immobilized on the
support. In this paper, we explore the use of a sol-immobilization
method, and in particular examine the effect of the Au-Pd ratio,
for carbon-supported nanoparticles, on their activity for the direct
synthesis of hydrogen peroxide. In addition, since peroxy species
are known to be effective for the selective oxidation of alcohols,
we also investigate the selective oxidation of benzyl alcohol to
determine structure/activity relationships for these materials.

Experimental Section
Preparation of Au, Pd, and Au-Pd Nanoparticles Using a
Sol-Immobilization Method. For the preparation of support
Au-Pd colloidal materials, an aqueous solution of PdCl2 (Johnson
Matthey) and HAuCl4 3 3H2O (Johnson Matthey) of the desired
concentration were prepared. Poly(vinyl alcohol) (PVA) (1 wt %
aqueous solution, Aldrich, MW = 10 000, 80% hydrolyzed) and
an aqueous solution of NaBH4 (0.1 M) were also prepared. For
example, a catalyst comprising Au-Pd nanoparticles with 1 wt %
total metal loading on a carbon support (Aldrich G60) was
prepared as follows: To an aqueous PdCl2 and HAuCl4 solution
of the desired concentration, the required amount of a PVA
solution (1 wt %) was added (PVA/(Au Pd) (w/w) = 1.2); a
freshly prepared solution of NaBH4 (0.1 M, NaBH4/(Au Pd)
(mol/mol) = 5) was then added to form a dark-brown sol. After
30 min of sol generation, the colloid was immobilized by adding
activated carbon (acidified at pH 1 by sulfuric acid) under vigor-
ous stirring conditions. The amount of support material required
was calculated so as to have a total final metal loading of 1 wt %,
and the molar ratio of Au/Pd was varied by adjusting the relative
concentrations of the metals in solution. After 2 h, the slurry was
filtered and the catalyst was washed thoroughly with distilled
water (neutral mother liquors) and dried at 120 C for 16 h.
Catalytic Reactions. Hydrogen Peroxide Synthesis. Cata-
lyst testing was performed using a stainless steel autoclave (Parr
Instruments) with a nominal volume of 100 mL and a maximum Figure 1. Representative bright field TEM micrographs showing
working pressure of 14 MPa. The autoclave was equipped with an Au-Pd particles immobilized on the activated carbon support:
overhead stirrer (0-2000 rpm) and had provision for measure- (a) Au only; (b) Au/Pd = 7:1; (c) Au/Pd = 2:1; (d) Au/Pd = 1:1;
ment of temperature and pressure. For the standard reaction (e) Au/Pd = 1:1.85; (f) Au/Pd = 1:2; (g) Au/Pd = 1:7; (h) Pd only.
conditions we have employed previously, the autoclave was
charged with the catalyst (0.01 g) and solvent (5.6 g MeOH and Hydrogen Peroxide Hydrogenation. Hydrogen peroxide hydro-
2.9 g H2O), and purged three times with 5%H2/CO2 (3 MPa) and genation was evaluated using a Parr Instruments stainless steel
then filled with 5% H2/CO2 and 25% O2/CO2 to give a hydrogen/ autoclave with a nominal volume of 100 mL and a maximum
oxygen ratio of 1:2 at a total pressure of 3.7 MPa. Stirring working pressure of 14 MPa. To test each catalyst for H2O2 hydro-
(1200 rpm) was commenced on reaching the desired temperature genation, the autoclave was charged with catalyst (0.01 g) and a
(2 C), and experiments were carried out for 30 min. H2O2 yield solution containing 4 wt % H2O2 (5.6 g MeOH, 2.22 g H2O, and
was determined by titration of aliquots of the final filtered 0.68 g H2O2 (50%)). The charged autoclave was then purged three
solution with acidified Ce(SO4)2 (710-3 mol/L). Ce(SO4)2 solu- times with 5% H2/CO2 (0.7 MPa) before filling with 5% H2/CO2
tions were standardized against (NH4)2Fe(SO4)2 3 6H2O using to a pressure of 2.9 MPa at 20 C. The temperature was allowed to
ferroin as indicator. decrease to 2 C followed by stirring (at 1200 rpm) of the reaction
mixture for 30 min. The amount of residual H2O2 after reaction
was determined using the method described previously.
(25) Dissanayake, D. P.; Lunsford, J. H. J. Catal. 2003, 214, 113120.
(26) Choudhary, V. R.; Samanta, C.; Gaikwad, A. G. Chem. Commun. 2004, Benzyl Alcohol Oxidation. Benzyl alcohol oxidation was
20542055. carried out in a stirred reactor (100 mL, Parr Instruments). The

Langmuir 2010, 26(21), 1656816577 DOI: 10.1021/la101597q 16569

Article Pritchard et al.

Figure 2. Particle size distributions of the Au-Pd particles immobilized on the activated carbon support: (a) Au only; (b) Au/Pd = 7:1;
(c) Au/Pd = 2:1; (d) Au/Pd = 1:1; (e) Au/Pd = 1:1.85; (f) Au/Pd = 1:2; (g) Au/Pd = 1:7; (h) Pd only.

16570 DOI: 10.1021/la101597q Langmuir 2010, 26(21), 1656816577

Pritchard et al.
Figure 3. Representative STEM-HAADF images showing Au-Pd homogeneous alloy particles: (a,b) Au/Pd=2:1; (c,d) Au/Pd=1:1; (e,f)
Au/Pd = 1:2.
vessel was charged with benzyl alcohol (40 mL) and catalyst
(0.05 g). The autoclave was then purged 5 times with oxygen leav-
ing the vessel at 10 bar gauge. The stirrer was set at 1500 rpm and
the reaction mixture was raised to the required temperature and
the reaction time was started as soon as the required reaction
temperature was reached. Samples from the reactor were taken
periodically, via a sampling system. For the analysis of the pro-
ducts a GC-MS and GC (a Varian star 3400 cx with a 30 m CP-Wax
52 CB column) were employed, and the products were identified
by comparison with known standards. For the quantification of
the amounts of reactants consumed and products generated, an
external calibration method was used.
Nanoparticle Characterization. UV-visible spectroscopy
of the colloidal sols was used to monitor the intensity and position
of the plasmon resonance band of gold. Samples for examination
by transmission electron microscopy (TEM) were prepared by
dispersing the dry catalyst powder onto a holey carbon film sup-
ported by a 300 mesh copper TEM grid. Samples were then sub-
jected to bright field diffraction contrast imaging and X-ray
energy-dispersive spectrometry (XEDS) experiments in order to
determine particle size distributions and compositions. The instru-
ment used for this analysis was a JEOL 2000FX TEM operating
at 200 kV. A subset of samples were examined in an aberration
corrected JEOL 2200FS TEM instrument operated in STEM-
HAADF imaging and STEM-XEDS modes at 200 kV in order to
study the structural and compositional details of individual metal
nanoparticles.
X-ray photoelectron spectra were recorded on a Kratos Axis
Ultra DLD spectrometer employing a monochromatic Al KR
X-ray source (75-150 W) and analyzer pass energy of 160 eV (for
survey scans) or 40 eV (for detailed scans). Samples were mounted
using double-sided adhesive tape, and binding energies were
referenced to the C(1s) binding energy of adventitious carbon
contamination which was taken to be 284.7 eV.
Results and Discussion
Preparation and Characterization of Au, Pd, and Au-Pd
Nanoparticles. We prepared a range of Au, Pd, and Au-Pd
colloidal sols (Au/Pd = 1:0, 7:1, 2:1, 1:1, 1:1.85, 1:2, 1:7, 0:1) and
examined these by UV-visible spectroscopy. For the pure Au sol,
(27) Bianchi, C. L.; Canton, P.; Dimitratos, N.; Porta, F.; Prati, L. Catal. Today
2005, 102-103, 203212.
Langmuir 2010, 26(21), 1656816577
Article
Table 1. Mean and Median Particle Sizes of the Various Immobilized
Au-Pd Sols Investigated in This Study
Au/Pd molar ratio Au only 7:1 2:1 1:1 1:1.85 1:2 1:7 Pd only
mean size (nm) 3.7 3.9 3.5 4.9 3.7 3.2 4.3 5.7
median size (nm) 3.5 3.5 3.2 4.6 3.3 2.9 4.0 5.4
the appearance of the plasmon resonance was observed at 505 nm.
This resonance peak is characteristic for gold nanoparticles with
particle sizes below 10 nm.27,28 In the case of the Pd monometallic
sol, there is no surface plasmon band.27 The spectra resulting from
the AuxPdy bimetallic sols show the disappearance of the gold sur-
face plasmon band on addition of Pd, as previously reported for
this system,27,29 which has been ascribed to changes in the band
structure of the Au particles due to alloying with Pd. These results
suggest the formation of bimetallic random-alloy nanoparticles,
which is in agreement with previously reported data.30-32
Representative bright field TEM micrographs (Figure 1) were
acquired in order to measure the particle size distributions of the
colloidal metal nanoparticles after immobilization and drying on
the activated carbon support (Figure 2). The particle size data for
all the colloid compositions used in this study are summarized in
Table 1. The median particles sizes for the monometallic Au and
Pd colloids were found to be 3.5 and 5.4 nm, respectively. The
median particle sizes for all the AuxPdy alloys were in the 2.9-
4.6 nm range, and did not show any discernible systematic size
trend with colloid composition. Figure 3 shows representative
STEM-HAADF images of the alloy particles in the 2:1, 1:1, and
1:2 Au-Pd samples. In each case, the metal particles were found
to be random homogeneous Au-Pd alloys with a face-centered
cubic (fcc) structure. If core-shell morphologies, or any other gross
form of Au-Pd segregation, were present in these nanoparticles,
these would have been apparent by atomic mass (z) contrast in
(28) Link, S.; El-Sayed, M. A. J. Phys. Chem. B 1999, 103, 42124217.
(29) Lopez-Sanchez, J. A.; Dimitratos, N.; Miedziak, P.; Ntainjua, E.; Edwards,
J. K.; Morgan, D.; Carley, A. F.; Tiruvalam, R.; Kiely, C. J.; Hutchings, G. J.
Phys. Chem. Chem. Phys. 2008, 10, 19211930.
(30) Deki, S.; Akamatsu, K.; Hatakenaka, Y.; Mizuhata, M.; Kajinami, A.
Nanostruct. Mater. 1999, 11, 5965.
(31) Scott, R. W. J.; Wilson, O. M.; Oh, S.-K.; Kenik, E. A.; Crooks, R. M.
J. Am. Chem. Soc. 2004, 126, 1558315591.
(32) Dash, P.; Bond, T.; Fowler, C.; Hou, W.; Coombs, N.; Scott, R. W. J.
J. Phys. Chem. C 2009, 113, 1271912730.
DOI: 10.1021/la101597q 16571
Article
Figure 4. Set of HAADF images and corresponding XEDS point spectra from particles at the low (a,b), median (c,d), and high (e,f) ends of
the particle size distribution for the Au/Pd (1:7) sample. They demonstrate a systematic particle size/composition variation, where the smaller
particles progressively become more Au-rich and Pd-deficient.
these HAADF images. It is interesting to note that the Au-Pd
particles remained pseudospherical (i.e., did not strongly wet the
activated C support material).
Energy dispersive X-ray spectra were acquired from areas con-
taining 100 metal particles in order to qualitatively check alloy
compositions. The relative Au to Pd peak areas from such aver-
aged spectra agreed with the nominal x and y values for each of the
AuxPdy alloy compositions. Furthermore, XEDS spectra obtained
from individual particles showed characteristic Au M lines and Pd L
lines confirming that intimately mixed AuPd alloys had been formed
rather than physical mixtures of pure Au and Pd particles. However,
it was noticeable that, within specific AuxPdy samples, the XEDS
spectra obtained from individual particles did show considerable
systematic composition variations as a function of particle size. For
example, Figure 4 shows a set of HAADF images and correspond-
ing XEDS point spectra from particles at the low, median, and high
ends of the particle size distribution for the Au/Pd (1:7) sample.
While all the particles were homogeneous AuPd alloys, the smallest
particles were Au-rich, whereas the largest ones tended to be Pd-rich.
It is interesting to note that this systematic composition/size trend is
diametrically opposite to the situation found for Au-Pd nanopar-
ticles prepared by impregnation methods where the larger particles
tended to Au-rich and the smaller Pd-rich.33
(33) Herzing, A. A.; Watanabe, M.; Kiely, C. J.; Edwards, J.; Conte, M.; Tang,
Z. R.; Hutchings, G. J. Faraday Discuss. 2008, 138, 337351.
16572 DOI: 10.1021/la101597q
Pritchard et al.
The Pd(3d) and Au(4f) X-ray photoelectron spectra for the
series of catalysts are shown in Figures 5 and 6, respectively. A dis-
tinct shift in the Au(4f7/2) binding energy is observed in the
composition range Au/Pd = 1:0 (binding energy = 84.2 eV) to
Au/Pd = 1:1 (84.0 eV), being constant thereafter; this may reflect
alloying with the Pd. Analysis of the Pd(3d) spectra is complicated
by the severe overlap between the Pd(3d) doublet and the Au(4d5/2)
component. However, when we calculate the experimental surface
Pd/Au molar ratios based on the Pd(3d5/2) Pd(3d3/2) Au(4d5/2)
combined integrated intensities and the Au(4f) intensity together
with known relative sensitivity factors, we are able to correct the
molar ratio by simply subtracting 0.5 from the raw experimental
values. The corrected experimental ratios are plotted against
nominal ratios (based on the preparation recipes) in Figure 7
and show a very clear linear correlation for Pd/Au ratios up to Pd/
Au=1:1, and the X-ray photoelectron spectroscopy (XPS)-derived
Pd/Au molar ratio matches the ratios used in the synthesis. How-
ever, as we increase the relative ratio of Pd/Au in the synthesis
above unity, XPS-derived surface ratios are consistently approxi-
mately a factor of 2 lower than expected. This could in principle
arise from (i) a systematic difference in alloy composition with
particle size, (ii) the presence of Pd(core)-Au(shell) nanoparti-
cles, (iii) differential deposition of Au and Pd relative to the
intended composition, or (iv) preferential adsorption of smaller
Pd-rich nanoparticles deeper into the carbon structure. The TEM
characaterization we have described supports explanation (i) as
Langmuir 2010, 26(21), 1656816577
Pritchard et al.
Figure 5. Pd(3d) spectra for the series of Au-Pd/carbon catalysts; nominal Au/Pd ratios are as indicated.
the most likely explanation. The STEM HAADF/XEDS analysis
rules out explanation (ii), and TEM-XEDS analysis and visual
inspection during the immobilization procedure rule out any sig-
nificant contribution of (iii). An explanation based in (iv) would
be in agreement with similar suggestions in the literature34 but still
is purely speculative and lacks any substantial evidence at this
time. Importantly, HAADF with XEDS point spectra analysis
demonstrate that, for the catalysts with a high Pd/Au ratio, a
systematic particle size/composition variation occurs. The larger
particles progressively become more Pd-rich and Au-deficient,
and, considering the high surface sensitivity of XPS, this effect
alone could provide a satisfactory explanation for the systematic
deviation observed between Pd/Au ratios used in the synthesis
and calculated from XPS data. The Pd nanoparticles comprise
predominantly Pd0 species (Pd(3d5/2) binding energy ca. 335 eV)
for Pd/Au ratios<1, but the presence of Pd2 species is indicated
for higher Pd contents (Figure 5), with the highest concentration
corresponding to the Pd-only catalyst.
(34) Bianchi, C.; Porta, F.; Prati, L.; Rossi, M. Top. Catal. 2000, 13, 231236.
Langmuir 2010, 26(21), 1656816577
Article
Hydrogen Peroxide Synthesis. The supported Au-Pd cata-
lysts were investigated for the direct synthesis of H2O2, and the
results are shown in Table 2. The catalysts were found to be parti-
cularly effective for this reaction, and the synthesis activity exhibited
a maximum at Au/Pd = 1:2 molar ratio. In our previous studies
using catalysts prepared using the impregnation method,14,15 we
used a Au/Pd mol ratio of 1:1.85, which is very close to this opti-
mum ratio. However, the catalysts prepared by sol-immobilization
are considerably more active than those prepared using the impreg-
nation method.14,15 Our earlier catalysts used 5 wt % total metal
loading and gave a productivity of H2O2 synthesis of 120 mol H2O2/
kgcat/h, but the comparable molar ratio prepared using sol-immo-
bilization has a productivity of 158 mol H2O2/kgcat/h using only
20% of the previous metal loading. This demonstrates that using the
sol-immobilization method we are able to utilize the metal more
effectively, as in the impregnation method very large particles are
generated and these have very limited catalytic activity.
In our earlier study, we observed that in the catalysts prepared
using the impregnation method the very small particles comprised
mainly Pd and that 3-5 nm Au-Pd nanoparticles comprised
DOI: 10.1021/la101597q 16573
Article Pritchard et al.

Figure 6. Au(4f) spectra for the series of Au-Pd/carbon catalysts; nominal Au/Pd ratios are as indicated.

typically 98% Pd 2% Au. As the activity was considered to be synthesis reaction, it could be expected that a material which has a
associated with these small particles, this led us to conclude that high activity for hydrogenation would, in principle, make an
for optimum activity we needed to find a method to fabricate such active catalyst. This is, however, not the case, since catalysts that
particles. Initially the results of the present study might suggest are effective in the initial hydrogenation of oxygen and which
that the above is not necessarily the case, since the Pd rich Au-Pd selectively generate hydrogen peroxide are also equally effective at
colloidal nanoparticles (i.e., Pd/Aug 3) display lower productiv- the hydrogenation of hydrogen peroxide to water in a subsequent
ity than those with a Au/Pd = 1:2 molar ratio. However, only hydrogenation reaction. Hence, the overall reaction can be viewed
detailed microscopy can reveal whether we have increased the as a set of competing reactions as shown below:
population of 3-5 nm Au-Pd nanoparticles of the preferred
ratio. It must also be taken into account that this effect could also
be explained on the basis that the hydrogenation activity of the
catalysts is expected to increase with Pd content and this cancels
the positive effect of adding Pd to the Au nanoparticles for the
direct synthesis, as described in the next section.
Hydrogen Peroxide Hydrogenation. The direct reaction
between H2 and O2 to form hydrogen peroxide involves the
hydrogenation of molecular oxygen, since it is the H-H bond
that is broken and the O-O bond is retained. Hence, catalysts
that are effective for the direct synthesis reaction are also effective
hydrogenation catalysts, as their primary role is the activation of The successful design of a catalyst for the direct synthesis of
H2 at low temperatures. In designing catalysts for the direct H2O2 has to balance the opposing requirements of sufficient

16574 DOI: 10.1021/la101597q Langmuir 2010, 26(21), 1656816577

Pritchard et al. Article

Table 3. Hydrogenation of H2O2 Using Carbon-Supported Au-Pd

Catalysts Prepared by Sol-Immobilizationa
Au/Pd molar ratio hydrogenation activity (mol H2O2/kg-cat/h)

1:0 27
7:1 64
3:1 214
2:1 238
1:1 522
1:1.85b 546
1:2 617
1:3 650
1:7 683
0:1 360
a
Reaction conditions: same as those for synthesis but in the absence
of 160 psi 25%O2/CO2 and with 4 wt % H2O2 in solvent. b 1:1 by weight;
this composition has been used in previous studies prepared using
coimpregnation.16-20

Figure 7. XPS-derived corrected Pd/Au molar ratios plotted as a

function of nominal (expected) Au/Pd ratios. The dotted line
shows the expected ratio.

Table 2. Direct Synthesis of H2O2 Using Carbon-Supported Au-Pd

Catalysts Prepared by Sol-Immobilizationa
productivity (mol H2O2/kg-cat/h)
Au/Pd molar ratio 30 min synthesis 2 min synthesis

1:0 4 57
7:1 20 52
3:1 38 124
2:1 73 290
1:1 156 462
b
1:1.85 158 728
1:2 188 613
1:3 162 636
1:7 162 588
0:1 120 386
a
Reaction conditions: 10 mg catalyst (10 mg) in 5.6 g of methanol and
2.9 g of water solvent, 420 psi 5%H2/CO2 160 psi 25%O2/CO2, pro-
ductivity calculated after 2 and 30 min reaction with stirring 1200 rpm at
2 C. b 1:1 by weight; this composition has been used in previous studies
using catalysts prepared by coimpregnation.16-20

activity to activate molecular H2, but simultaneously ensure that

there is insufficient hydrogenation activity to react the H2O2. This
is the crucial feature, as the selectivity based on H2 utilization is
the most important parameter. Recently, we have shown14 that
the subsequent hydrogenation of H2O2 can be switched off by a
simple acid pretreatment of the carbon support for catalysts
prepared by impregnation.
We have therefore examined the H2O2 hydrogenation activity
of the carbon supported sol-immobilized catalysts, and the results
are shown in Table 3. It is apparent that as the Pd content of the
Au-Pd nanoparticles increases, so does the hydrogenation activ-
ity. This is to be expected, since Pd has a very high propensity for
the dissolution of H2 and hydrogenation reactions. Also, it must
be noticed that all the catalysts discussed in this paper contain a
fixed percentage of 1 wt % total metal content independently of
the Pd/Au molar ratio, and this results in an overall increase in
metal atomic concentration on the catalyst. Hence, increasing the
Pd/Au ratio leads to an overall increase in the metal available
to catalyze the reaction. Interestingly, the addition of Au to Pd
enhances the rate of H2O2 hydrogenation. This is unexpected but
once again shows the synergy in reactivity that is observed with
AuPd bimetallic catalysts when compared with their monome-
tallic counterparts.
Figure 8. Specific activity comparison based on moles of metal
present for benzyl alcohol oxidation, hydrogen peroxide synthesis,
and hydrogen peroxide hydrogenation with 1 wt % Au-Pd/C
prepared by sol-immobilization with a Pd/Au molar ratio of 1.85.
Key: (a) (solid black tilted square) H2O2 productivity calculated
after 2 min reaction (left axis) and (red empty square) TOF of ben-
zyl alcohol calculated after 30 min reaction (right axis). (b) (solid
black tilted square) H2O2 productivity calculated after 30 min (left
axis) and (empty red square) rate of H2O2 hydrogenation calcu-
lated after 30 min.
The promotion effect of gold on palladium for the hydrogena-
tion of H2O2 is shown in Figure 8, where activities are normalized
per mole of metal in the catalyst. Figure 8b indicates that a very
small amount of gold (Pd/Au ratio = 7) more than doubles the
activity of pure Pd, whereas subsequent increases in the amount of
gold have very little effect. A tentative explanation comes from

Langmuir 2010, 26(21), 1656816577 DOI: 10.1021/la101597q 16575

Article Pritchard et al.

Table 4. Effect of the Molar Au/Pd Ratio on Benzyl Alcohol Oxidation Results with 1 wt % (Aux-Pdy)/C Catalysts Prepared by the Colloidal
Methoda
selectivity (%)
entry catalyst conv (%) toluene benzaldehyde benzoic acid benzyl benzoate

1 1% Au/CSI 6.6 23.9 63.9 3.1 9.0

2 1% (7Au-1Pd)/CSI 11.6 1.9 66.3 7.2 24.5
3 1% (3Au-1Pd)/CSI 23.3 6.0 72.5 8.6 12.9
4 1% (2Au-1Pd)/CSI 52.0 10.9 77.5 3.5 8.1
5 1% (1Au-1Pd)/CSI 71.1 4.0 69.8 19.9 6.3
6 1% (1Au-1.85Pd)SI 80.7 3.4 67 23.1 6.4
7 1% (1Au-2Pd)/CSI 90.8 4.5 67.4 22.9 5.1
8 1% (1Au-3Pd)/CSI 94.7 4.2 67 23.5 5.3
9 1% (1Au-7Pd)/CSI 88.7 1.7 68.4 24.1 5.8
10 1% Pd/CSI 59.3 6.8 74.7 10.4 8.1
a
Reaction conditions: benzyl alcohol, 0.05 g of catalyst, T = 120 C, pO2 = 150 psi, stirring rate 1500 rpm, 6 h reaction.

Scheme 1. General Reaction Pathway for Benzyl Alcohol Oxidation Table 5. Comparison of Benzyl Alcohol Oxidation at Iso-Conversion
of 50% with 1 wt % (Aux-Pdy)/C Catalysts Prepared by the
Colloidal Methoda
selectivity (%)
benzoic benzyl
catalyst toluene benzaldehyde acid benzoate

1% (2Au-1Pd)/CSI 11.0 78.6 3.5 6.9

1% (1Au-1Pd)/CSI 5.4 79.7 7.4 7.5
1% (1Au-1.85Pd)CSI 4.3 81.9 6.6 7.2
1% (1Au-2Pd)/CSI 7.5 83.6 3.5 5.4
1% (1Au-3Pd)/CSI 6.6 85.4 2.4 5.5
1% (1Au-7Pd)/CSI 1.9 85.8 5.5 6.8
1% Pd/CSI 6.4 76.0 8.9 8.7
a
Reaction conditions: benzyl alcohol, 0.05 g of catalyst, T = 120 C,
pO2 = 150 psi, stirring rate 1500 rpm. Catalysts containing higher amounts
of Au could not be compared at iso-conversion, as the conversion with
these catalysts was typically <50% for reasonable reaction times.

the monometallic 1 wt % Au/C was very poor, but on increasing

close inspection of XPS data in Figure 4b of sample Pd/Au=7. It
appears that, in this particular sample, the Pd is more reduced. We
have previously found some indication from XPS that Pd alloyed
with gold can display higher concentration on the surface than the
analogous monometallic Pd catalyst35 and this effect could ex-
plain the increase of hydrogenation activity when doping palla-
dium with small amounts of gold.
Overall, these results show that the small Au-Pd nanoparti-
cles have a very high H2O2 hydrogenation activity, and cer-
tainly much higher than that for Au-Pd catalysts prepared by
impregnation. Hence, the rate of H2O2 synthesis will be much
higher than that gleaned from the overall reaction productivity
since the H2O2 is actually being consumed rapidly. For this rea-
son, we investigated the initial rate of H2O2 synthesis (Table 2)
and clearly this is the case, and hydrogen peroxide productivity
values are around 4 times higher for the most active catalysts
when measurements are taken at short times, when little hydro-
gen peroxide is yet present and so its subsequent hydrogenation
is minimized.
Benzyl Alcohol Oxidation. The catalytic oxidation of benzyl
alcohol under free solvent conditions has been chosen as an app-
ropriate model reaction for exploring the catalytic performance of
the bimetallic Aux-Pdy catalysts (Scheme 1). The results of the
catalytic tests at 120 C are presented in Table 4. The activity of
the Pd content we observed a progressive increase in catalytic
activity. This increase of activity reached a broad maximum with
a Au/Pd molar ratio between 1:1 and 1:3, as values above 95%
conversion are achieved. A further increase in the Pd content res-
ulted in a progressive decrease in catalytic activity. It is important
to note that even in the presence of a minor amount of gold or
palladium (i.e., in the case of 1:7 and 7:1 Au/Pd molar ratios) a
significant increase in the catalytic activity was observed; speci-
fically, we observed an increase in activity by a factor of 2.3-2.6
with respect to the monometallic Au and Pd catalysts. In order to
determine the effect of the Au/Pd molar ratio on selectivity, the
catalytic data were compared at iso-conversion, and Table 5
shows the catalytic data at 50% conversion. These data show that
the catalytic performance in terms of reaction specificity is not
particularly significant and all catalysts give high selectivities to
benzaldehyde which is the primary product (Scheme 1). Hence,
the main effect of alloying Au with Pd is an effect of enhancing
activity. Although at high conversions with Pd-rich catalysts
the selectivity to benzoic acid is enhanced. In particular, the for-
mation of toluene, which is generated from benzyl alcohol by a
hydrogen transfer reaction, does not increase at iso-conversion as
the Pd content increases. We have shown from the H2O2 hydro-
genation studies that hydrogenation increases with Pd concen-
tration in the Au-Pd alloy, and so we might have expected to
observe higher toluene selectivities as the Pd content is increased.
To the contrary, selectivity values at iso-conversion indicate that
the selectivity toward toluene is highest for the gold-rich catalysts
(Pd/Au = 0.5).
Comments on the Structure Activity Relationships of
Au-Pd Nanoparticles. In this study, we have shown that the

16576 DOI: 10.1021/la101597q Langmuir 2010, 26(21), 1656816577

Pritchard et al.
carbon supported Au-Pd nanoparticles are very effective for the
three reactions, namely, the direct synthesis of H2O2, its subseq-
uent hydrogenation, and the selective oxidation of benzyl alcohol.
For all these reactions, we show that the optimum ratio of Au/Pd
is ca. 1:1.85. This is shown in Figure 8 where the catalytic activity
per mole of metal of the Pd/Au series of catalysts for hydrogen
peroxide synthesis, hydrogen peroxide hydrogenation, and benzyl
alcohol oxidation is plotted. This is an unexpected result, as there
is no clear evidence that the same active sites are responsible for
the same activity for these two very different reactions. TOF val-
ues for benzyl alcohol oxidation reach a maximum>60 000 h-1,
whereas productivity values for hydrogen peroxide synthesis
reach up to 10 000 mol H2O2 produced 3 mol of metal-1 3 h-1.
These two values are difficult to compare because we are report-
ing the moles of H2O2 obtained after a catalytic test is completed,
but not the actual number of moles synthesized as the consecutive
hydrogenation reaction is taking place. Using the particle size
distribution obtained via TEM, we have been able to calculate the
distribution of atoms on the catalyst surface for the composition
of Pd/Au = 1.85 which coincides with the optimized activity
catalyst for the two reactions. Taking these values into account,
we can recalculate the activities to using only those atoms exposed
on the catalyst surface. For benzyl alcohol oxidation, a TOF of
196 000 mol benzyl alcohol reacted per surface atom of metal per
hour is observed. For hydrogen peroxide production, a value of
31 000 molecules of H2O2 produced per surface atom of metal per
hour is observed. This is further evidence of the very high activity
that the Au-Pd alloy system can produce when the preparation
method and composition are properly optimized.
A realistic comparison of turnover frequencies for these two
reactions is very difficult, and in Figure 8a we used initial activity
rates for hydrogen peroxide to reduce to a minimum the effect of
the consecutive hydrogenation reaction and/or decomposition
Langmuir 2010, 26(21), 1656816577
Article
of hydrogen peroxide. This allowed us to evaluate the highest
possible productivity values under these reaction conditions
and also to discern that we have the same activity profile as in
benzyl alcohol oxidation. Figure 8 contrasts the direct synthesis
of hydrogen peroxide versus its hydrogenation, with both
reactions being carried out for a period of 30 min. Remarkably,
both reactions showed a very similar dependence on the Pd/Au
ratio, and the data suggest a similar volcano plot as shown in
Figure 8a. At these longer reaction times, it is evident that
productivity values are much lower, but the hydrogenation
activity indicates that the ability of the catalyst to react hydro-
gen peroxide is also greater under the same optimum values of
Pd/Au. Unfortunately, this indicates that improved catalytic
yields for hydrogen peroxide production are still accompanied
by a substantial loss in hydrogen selectivity. Although the
reaction conditions for the two reactions are different, it is
important to note that activity values per surface of metal atom
calculated after 30 min online for the two reactions are 6700
molecules of H2O2 produced per surface atom of metal surface
per hour and 23 000 molecules of H2O2 hydrogenated per
surface atom of metal per hour. This demonstrates that, for
these catalysts prepared by sol-immobilizion, the hydrogenation
of H2O2 is far more facile. As we have previously done for
catalysts prepared by impregnation, we now need to modify the
catalyst formulation further at this point so that we can
preferentially enhance the direct synthesis, but not the subse-
quent hydrogenation reaction.
Acknowledgment. We thank the EPSRC and SOLVAY for
financial support.
(35) Dimitratos, N.; Lopez-Sanchez, J. A.; Anthonykutty, J. M.; Brett, G.;
Carley, A. F.; Tiruvalam, R. C.; Herzing, A. A.; Kiely, C. J.; Knight, D. W.;
Hutchings, G. J. Phys. Chem. Chem. Phys. 2009, 11, 49524961.
DOI: 10.1021/la101597q 16577