Influence of the preparation method on the morphological and composition properties of Pd-Au/ZrO2 catalysts and their effect on the direct synthesis of hydrogen peroxide from hydrogen and oxygen

Article abstract of DOI:10.1016/j.jcat.2009.09.010

Bimetallic Pd-Au samples supported on zirconia were prepared by different methods and tested for the direct synthesis of hydrogen peroxide under very mild conditions (room temperature and atmospheric pressure), outside the explosion range and without halides addition. Further catalytic tests were performed at higher pressure using solvents expanded with CO₂. Samples were characterized by N₂ physisorption, metal content analysis, XRD, HRTEM combined with X-ray EDS, TPR, and FTIR. The effect of the addition of gold to Pd in enhancing the yield of H₂O₂ is sensitive to the preparation method: the best catalytic results were obtained by depositing gold by deposition-precipitation (DP) and by introducing in a second step Pd by incipient wetness impregnation. The origin of the differences between samples is discussed. The role of Au in the catalytic reaction seems to be a complex one, changing the chemical composition of the metallic particles, their morphology, and charge of the exposed Pd sites.

Full text of DOI:10.1016/j.jcat.2009.09.010

Journal of Catalysis 268 (2009) 122–130
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Influence of the preparation method on the morphological and composition
properties of Pd–Au/ZrO2 catalysts and their effect on the direct synthesis
of hydrogen peroxide from hydrogen and oxygen
Federica Menegazzo a, Michela Signoretto a, Maela Manzoli b, Flora Boccuzzi b, Giuseppe Cruciani c,
Francesco Pinna a, Giorgio Strukul a,
*
a Department of Chemistry, Cà Foscari University and Consortium INSTM UdR Ve, Dorsoduro 2137, 30123 Venezia, Italy
b Department of Inorganic, Physical and Materials Chemistry, University of Torino, and NIS Centre of Excellence, via P. Giuria 7, 10125 Torino, Italy
c Department of Earth Sciences, University of Ferrara, Via Saragat 1, 44100 Ferrara, Italy
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 13 July 2009
Revised 11 September 2009
Accepted 11 September 2009
Available online 9 October 2009
Bimetallic Pd–Au samples supported on zirconia were prepared by different methods and tested for the
direct synthesis of hydrogen peroxide under very mild conditions (room temperature and atmospheric
pressure), outside the explosion range and without halides addition. Further catalytic tests were per-
formed at higher pressure using solvents expanded with CO2.
Samples were characterized by N2 physisorption, metal content analysis, XRD, HRTEM combined with
X-ray EDS, TPR, and FTIR. The effect of the addition of gold to Pd in enhancing the yield of H2O2 is sen-
sitive to the preparation method: the best catalytic results were obtained by depositing gold by deposi-
tion–precipitation (DP) and by introducing in a second step Pd by incipient wetness impregnation. The
origin of the differences between samples is discussed. The role of Au in the catalytic reaction seems
to be a complex one, changing the chemical composition of the metallic particles, their morphology,
and charge of the exposed Pd sites.
Keywords:
Hydrogen peroxide direct synthesis
Palladium
Gold
Pd–Au
Zirconia
Bimetallic catalyst
Reactivity
FTIR
Ó 2009 Elsevier Inc. All rights reserved.
Mechanism
1. Introduction
which it is generally agreed that palladium is the most effective
metal [8,9,14–16], it has been reported that the combination of
The direct reaction of H2 + O2 ? H2O2 is clearly the most atom-
efficient method to form hydrogen peroxide, but none of the
presently available processes has solved the productivity vs. safety
dilemma. In fact, the major problem of the direct route to hydrogen
peroxide is the poor selectivity to H2O2 vs. H2O that can be
achieved with known catalysts. As shown in Fig. 1, the process also
involves the thermodynamically highly favoured but undesirable
parallel and consecutive water-forming reactions. Another serious
problem that has limited the implementation for this process is the
significant risk of handling the explosive hydrogen/oxygen gas
mixture over an active catalyst. Therefore, despite several patents
[1–7] and recent literature [8–12], the direct synthesis of hydrogen
peroxide has not yet found the way to commercialization.
Pd with Au [17–19], Ir [20], Ag [21], and Pt [17,21–23] improves
the catalytic performance with respect to monometallic palladium
samples. As far as gold is concerned, since the pioneering work of
Haruta et al. [24], gold nanoparticles supported on metal oxides are
known to be active in some important industrial reactions [25–28],
the present reaction being not included. The reasons for the activ-
ity of small gold particles are still a matter of debate. It has been
shown that the catalytic activity of gold critically depends on the
preparation method, on the support type, and on the pre-treatment
procedure. The most widely accepted explanation for the variabil-
ity of gold catalytic properties focuses on the size of gold particles
and on the amount of low coordination sites of gold. In addition,
other factors have been considered in the literature: metal-support
interface, and charge transfer from the support or metal cationic
sites [29]. Several methods have been tested for making suitably
small gold particles: the deposition–precipitation (DP) method
has been qualified as the best so far. In this method [30] the pH
of a solution of HAuCl4 is raised by the addition of a base to the
point where the adsorption of the species in solution can react with
or be deposited on the support.
For a long time it has been known that alloying or combination
of two metals can lead to materials with special chemical proper-
ties due to an interplay of ‘‘ensemble” and ‘‘electronic” effects [13].
In particular for the direct synthesis H2O2, from its elements, for
* Corresponding author. Fax: +39 041 234 8517.
E-mail address: strukul@unive.it (G. Strukul).
0021-9517/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2009.09.010

F. Menegazzo et al. / Journal of Catalysis 268 (2009) 122–130
123
tion at pH 8.6 (pH fixed by adding a 0.5 M sodium hydroxide
solution). The sample obtained was then dried and calcined
under the same conditions reported above.
H2, k3
H2O2
H2O
- 136 kJ/mol
- 211 kJ/mol
(iii) The third catalyst (1Au2Pd) was prepared as the second cat-
alyst but by inverting the deposition order: first gold by DP,
then palladium by an IW method. The sample was finally
calcined at 773 K in flowing air (30 ml/min) for 3 h.
k1
k2
k4
H2 + O2
- 106 kJ/mol
- 242 kJ/mol
H2O + 1/2 O2
For the purpose of comparison a monometallic Pd catalyst (Pd)
was prepared by IW impregnation of H2PdCl4, and two monometal-
lic Au catalysts were synthesized both by IW impregnation and by
DP. After drying, the samples were calcined under the same condi-
tions reported for the bimetallic catalysts.
Fig. 1. Reactions involved in the direct production of H2O2.
We have already shown [31] that zirconia is a good support for
Pd-based catalysts for the direct synthesis of hydrogen peroxide
and that it is possible to prepare highly dispersed gold on zirconia
by DP [32]. We have also recently demonstrated [33,34] for both
plain and sulfated zirconia and ceria supports that while the mono-
metallic gold catalysts are inactive under mild experimental condi-
tions, the addition of a 1:1 amount of gold to a monometallic Pd
sample improves the productivity and especially the selectivity
of the process. In these samples gold must be in close contact with
Pd, as its presence profoundly changes both Pd dispersion and its
morphology and charge. However, no evidence for small Au parti-
cles was found, probably because of the preparation conditions. In
the present work, we wish to report the preparation by different
methods of a series of Pd–Au samples supported on zirconia in or-
der to fully understand the nature of palladium–gold interactions.
These catalysts were also tested for the direct synthesis of hydro-
gen peroxide both under very mild conditions (1 bar and 20 °C
and outside the explosion range) and at higher pressure (10 bar)
using solvents expanded with CO2 in order to increase productivity
further.
2.3. Methods
Surface areas and pore size distributions were obtained from N2
adsorption/desorption isotherms at 77 K (using a Micromeritics
ASAP 2000 analyzer). Calcined samples (300 mg) were pre-treated
at 573 K for 2 h under vacuum. Surface areas were calculated from
the N2 adsorption isotherm by the BET equation, and pore size dis-
tributions were determined by the BJH method [37]. Total pore
volumes were taken at p/p0 = 0.99.
The actual metal loadings were determined by atomic absorp-
tion spectroscopy after microwave disgregation of the samples
(100 mg).
X-ray powder diffraction (XRD) patterns were measured by a
Bruker D8 Advance diffractometer equipped with a Si(Li) solid
state detector (SOL-X) and a sealed tube providing Cu Ka radiation.
Measuring conditions were 40 kV ꢀ 40 mA. Apertures of diver-
gence, receiving, and detector slits were 1°, 1°, and 0.3°, respec-
tively. Data scans were performed in the 2h range 20–80° with
0.02° stepsize and counting times of 10 s/step. The normalized ref-
erence intensity ratio (RIR) method and the Rietveld refinement
method, respectively, implemented in the Bruker EVA and TOPAS
programs, were used to obtain the quantitative phase analysis
and the crystal size of zirconia polymorphs and metal phases in
the samples.
2. Experimental
2.1. Materials
ZrOCl2 (Fluka) was used as received for sample synthesis. All ki-
netic tests were performed in anhydrous methanol (SeccoSolv,
Merck, [H2O] < 0.005%). Commercial standard solutions of Na2S2O3
(Fixanal [0.01], Hydranal-solvent E, and Hydranal-titrant 2E, all
from Riedel-de Haen) were used for iodometric and Karl–Fischer
titrations.
A careful analysis of the catalysts morphology, structure, and
composition was performed by using a side entry stage high-reso-
lution transmission electron microscopy (HRTEM) JEOL JEM 3010
UHR (300 kV) equipped with a LaB6 filament and fitted with X-
ray EDS analysis by a Link ISIS 200 detector. The powdered samples
were ultrasonically dispersed in isopropanol and a few droplets of
the slurry were then deposited on a copper grid, coated with a por-
ous carbon film.
2.2. Catalyst preparation
TPR experiments were carried out in a home-made equipment:
samples (100 mg) were heated with a 10 K/min ramp from 298 K
to 1500 K in a 5% H2/Ar reducing mixture (40 ml/min STP).
FTIR spectra were taken on a Perkin–Elmer 1760 spectrometer
(equipped with a MCT detector) with the samples in self-support-
ing pellets introduced in a cell allowing thermal treatments under
controlled atmosphere. All pre-treatments on the as-prepared
samples were performed at room temperature (r.t.) before the
spectroscopic experiments. The pre-treatments were (i) prolonged
outgassing (1 h); (ii) 1-h outgassing and 1-h treatment under
50 mbar of H2; and (iii) 1-h outgassing at r.t., 1-h treatment under
H2 (50 mbar), 30-min outgassing followed by 1-h treatment under
50 mbar of O2 and 30-min outgassing. After each pre-treatment,
7 mbar of CO was admitted at 90 K, in order to avoid possible Pd
reduction by CO itself. Moreover, we chose to collect the spectra
at 180 K to eliminate the physisorbed CO contribution. The spec-
trum of the sample before the CO inlet was subtracted from each
spectrum. Finally, all spectra were normalized to the same palla-
dium content. Band integration was carried out by ‘‘Curvefit”, in
Spectra Calc (Galactic Industries Co.). The curvefits were performed
Zirconia was prepared by precipitation from ZrOCl2 at pH 10,
aged under reflux conditions for 20 h [35,36], washed free from
chloride (AgNO3 test), and dried at 383 K overnight. Zr(OH)4 was
then calcined in flowing air (30 ml/min) at 923 K for 4 h. Calcined
zirconia was used as a support for preparing three bimetallic Pd–
Au samples with the same Pd (1.3 wt%) and Au (1.2 wt%) loading,
but prepared with three different procedures:
(i) A catalyst (PdAu) was prepared by incipient wetness (IW)
co-impregnation of H2PdCl4 and HAuCl4 aqueous solutions,
followed by calcination at 773 K in flowing air (30 ml/min)
for 3 h.
(ii) Another sample (1Pd2Au) was prepared by depositing the
two metals separately by different techniques: palladium
by IW and gold by DP. Pd was deposited first on the support
from a H2PdCl4 aqueous solution, then the material was
dried at 383 K overnight and calcined at 773 K in flowing
air (30 ml/min) for 3 h. In a second step, Au was deposited
by DP by suspending the catalyst in a HAuCl4 aqueous solu-

124
F. Menegazzo et al. / Journal of Catalysis 268 (2009) 122–130
by 5–8 mixed Lorentzian–Gaussian curves (depending on the sam-
ple and on the pre-treatment), without any fixed parameter (max-
imum position, band half-width, etc.). The obtained integrated
areas were normalized to the Pd content of each sample.
250
200
150
100
50
0.6
0.5
0.4
0.3
0.2
0.1
0.0
2.4. Catalyst testing
2.4.1. H2O2 direct synthesis at atmospheric pressure
1
10
100
1000
Part of catalytic tests were carried out at atmospheric pressure
in a thermostatted glass reactor (293 K) according to a previously
described procedure [31,33]. Mixing was carried out with a
TeflonÒ-made rotor operating at 1000 rpm. Oxygen and hydrogen
were bubbled by a gas diffuser directly into the liquid phase with
a total flow of 50 ml/min. A H2:O2 4:96 (nonexplosive and lower
limit for non-flammable mixture [38]) gas mixture was used. The
0.03 M H2SO4 methanolic solution reaction medium (100 ml) was
pre-saturated with the gas mixture before the introduction of a
catalyst (135 mg). Samples were preactivated in situ first by H2
(15 min–30 ml/min) and then by O2 (15 min–30 ml/min) flow to
induce a Pd particle surface oxidation. During catalytic tests small
aliquots of the liquid phase were sampled through a septum and
used for water and hydrogen peroxide determination. H2O2 con-
centration was measured by iodometric titration, whereas water
was determined by volumetric Karl–Fischer method. The water
content in the reaction medium before catalyst addition was deter-
mined prior to each catalytic experiment. H2O2 selectivity at time t
was determined as follows:
Pore size (nm)
0
0.0
0.2
0.4
0.6
0.8
1.0
P/P
0
Fig. 2. N2 physisorption isotherms of calcined support and (inset) its BJH pore size
distribution.
inset). A type IV isotherm with hysteresis loop typical of mesopor-
ous materials is observed. A relatively low surface area (<60 m2/g)
and a mean pore size around 20 nm can be calculated, in agree-
ment with the previous works for zirconia prepared by this method
[31,33].
In Table 1 palladium and gold amounts on the final samples are
reported. As can be observed, in the 1Pd2Au catalyst the found pal-
ladium amount is significantly lower than the theoretical one. Part
of the metal is probably lost during gold DP, in spite of the interme-
diate calcination. On the other samples, the palladium amounts
correspond to those loaded, confirming that the IW impregnation
is a reliable preparation technique. On the contrary, the gold
amount found in the samples after calcination is lower than ex-
pected. This may be due to incomplete precipitation on the
support.
½H2O2ꢁ
SH
¼
2O2
½H2O2ꢁ þ ½H2Oꢁ
Kinetic data fitting was performed according to a procedure de-
scribed elsewhere to calculate the apparent kinetic constants that
apply to the complex reaction network shown in Fig. 1 [33]. It
should be emphasized that H2O2 disproportionation was not con-
sidered as it was found negligible under the reaction conditions
adopted [39].
3.2. Catalytic data
2.4.2. H2O2 direct synthesis using an autoclave
Some catalytic tests were performed at 293 K using an auto-
clave with a nominal volume of 250 ml. A 150 ml methanolic solu-
tion added with 250 ll of H2SO4 was used as the reaction medium.
3.2.1. H2O2 direct synthesis at atmospheric pressure
All catalytic tests were carried out in methanol, which is the
best solvent for this reaction according to the previous works
[31]. In fact it helps H2 solubility while avoiding the formation of
peroxides, at variance with higher alcohols [27]. Besides, the use
of methanol as a solvent allows water titration and therefore the
determination of the process selectivity. Since most oxidation reac-
tions involving H2O2 are carried out in organic solvents, often in
methanol, the synthesis in this solvent could be an advantage of
avoiding separation and concentration costs. Catalytic tests were
carried out in an acidic (H2SO4) methanol medium at room temper-
ature, atmospheric pressure, without halides addition, after an
activation process giving rise to a Pd particle surface oxidation
[33]. The concentration of the produced hydrogen peroxide is
shown in Fig. 3, while the corresponding productivities after 5 h
of reaction are reported in Table 1. We have already reported
The water content in the reaction medium was determined prior to
each catalytic experiment before the introduction of a catalyst
(50 mg). Typically, the autoclave was charged, purged three times
with CO2 (10 atm), and then filled with the reactants to give a total
pressure of 10 atm. A gas mixture with the following composition
was used: H2:O2:CO2 = 3.6:7.2:89.2 (nonexplosive and non-flam-
mable mixture [38]). Mixing was carried out with a TeflonÒ-made
rotor operating at 1200 rpm. Experiments were carried out for
30 min unless otherwise stated. Water and hydrogen peroxide
determination were measured after each catalytic test as reported
previously.
3. Results and discussion
3.1. Catalyst morphology and analytical properties
Table 1
Metal loading on final catalysts, chemisorption values and productivity after 5 h of
reaction at atmospheric pressure.
N2 physisorption analysis has been carried out in order to
determine the surface area and pore size distribution of the zirco-
nia prepared as a support. In this investigation, the choice of a mes-
oporous material as a support is very important, since the presence
of micropores could cause mass transfer problems, while a low sur-
face area would not allow a good dispersion of the Pd and Au active
phases. The N2 physisorption isotherm for the calcined zirconia
sample is shown in Fig. 2, with its BJH pore size distribution (see
Sample
% Pd
(wt%)
% Pd found
(wt%)
% Au
(wt%)
% Au found
(wt%)
Productivity
ðmmolH =gPd hÞ
2O2
Pd
PdAu
1Pd2Au
1Au2Pd
2.5
1.3
1.3
1.3
2.44
1.21
0.89
1.17
//
//
499
1.2
1.2
1.2
0.95
0.74
0.82
1027
1040
1429

F. Menegazzo et al. / Journal of Catalysis 268 (2009) 122–130
125
Table 2
150
100
50
Apparent kinetic constants for the bimetallic PdAu catalysts.
k01 ꢀ 104
k02 ꢀ 104
k03 ꢀ 104
Sample
1Au2Pd
PdAu
(mol Lꢃ1 minꢃ1
)
(mol Lꢃ1 minꢃ1
)
(minꢃ1
)
PdAu
1Au2Pd 4.1
1Pd2Au 4.7
3.5
2.0
2.2
2.4
1.1
0.88
6.4
T = 298 K, p = 1 bar, H2:O2 4:96 100 ml of a 0.03 M H2SO4 methanolic solution,
135 mg catalyst.
1Pd2Au
Data are consistent with the following observations: (i) with
1Au2Pd and PdAu the consecutive reactions in Fig. 1 are practically
negligible with respect to the parallel formation of hydrogen per-
oxide and water accounting for the constant formation of hydrogen
peroxide and the corresponding selectivity; (ii) the better perfor-
mance of 1Au2Pd with respect to the co-impregnated sample is
mainly due to a more efficient hydrogen peroxide formation, this
explains the higher selectivity; (iii) with 1Pd2Au the consecutive
hydrogenation of hydrogen peroxide is significantly present and
this accounts for the declining formation of hydrogen peroxide
and the corresponding selectivity with time.
Working in an acidic medium, metal leaching cannot, in princi-
ple, be excluded; therefore, the heterogeneity of the catalytic reac-
tion may be questionable [40]. To shed light on this point two
complementary experiments were performed with 1Au2Pd after
the first catalytic run: (i) the solid catalyst was filtered off and
the catalyst-free solution allowed to further react: no activity
was observed; (ii) the solid catalyst was recycled three times
showing no loss of activity and selectivity.
0
70
1Au2Pd
PdAu
60
50
40
30
20
10
0
1Pd2Au
0
60
120
180
240
300
360
Time (min)
Fig. 3. H2O2 production and selectivity at atmospheric pressure with PdAu (q),
1Au2Pd (w), 1Pd2Au (s) catalysts.
3.2.2. H2O2 direct synthesis using an autoclave
With the aim of further improving the productivity of H2O2, cat-
alytic tests at higher pressures were performed, still working out-
side the explosive region, at r.t. and without halides addition.
Methanol was chosen as the solvent due to the unquestionable
reasons mentioned above. CO2 was used as a diluent inert gas be-
cause the explosive region for H2/O2 mixtures is narrower than
that for other inert gases [27] and its acidity could help improve
hydrogen peroxide stability. Besides, carbon dioxide is generally
considered to be an environmentally benign solvent, as it is natu-
rally abundant, relatively non-toxic, and non-flammable [27,41].
As reported in Fig. 4, productivities at 10 bar are higher than
those at 1 bar for all examined catalysts by a factor even higher
than 10 times. This is not unexpected, but the unusual result is
the increase in selectivity. In fact, except for the worst sample
(1Pd2Au), the selectivity is higher when performing the reaction
at 10 bar instead of 1 bar, both for the monometallic Pd catalyst
and for the bimetallic PdAu and 1Au2Pd samples. Although mono-
metallic gold samples supported on zirconia do not show signifi-
cant H2O2 formation at 10 bar, the addition of gold to Pd results
in higher productivity and selectivity. As an example, the catalyst
[33] that gold alone, even supported on zirconia, is not active in the
H2O2 direct synthesis under the mild experimental conditions
used. From Table 1 it is evident that the addition of gold to palla-
dium results in samples with higher productivity than a monome-
tallic Pd catalyst. In fact H2O2 productivity increases from almost
500 for the Pd sample to over 1000 mmolH =gPd h for the three
2O2
bimetallic catalysts. However, as expected, the effect of the addi-
tion of gold to Pd in enhancing the yield of H2O2 is sensitive to
the preparation method: the best catalytic results were obtained
using the 1Au2Pd sample, as shown in Fig. 3. Such a catalyst allows
to achieve, already at atmospheric pressure, a H2O2 concentration
over 100 mM. The difference with respect to the 1Pd2Au sample
is really remarkable, since the latter achieves a H2O2 concentration
around 60 mM under the same conditions. Moreover Fig. 3 indi-
cates that hydrogen peroxide formation increases linearly with
1Au2Pd, while with the 1Pd2Au and PdAu samples productivity
declines after about 3 h, even if with the latter catalyst this behav-
ior is less evident. This means that, beyond this point, H2O2 decom-
position starts prevailing over H2O2 formation. This is confirmed by
the selectivity reported in Fig. 3. The sample with the highest pro-
ductivity (1Au2Pd) is also the most selective (52%) and stable over
5 h of time on stream. An almost similar behavior is observed for
the co-impregnated sample (PdAu) albeit with a lower selectivity
(40%). On the contrary, the 1Pd2Au sample is initially the most
selective (60%), but its selectivity decreases rapidly after 2 h and
after 5 h it is approximately 30%. All these experimental
observations can be easily accounted for on the basis of the appar-
ent kinetic constants k10 , k20 , and k30 (k04 was found negligible, see
Section 2) as can be seen in Fig. 1, incorporating hydrogen and oxy-
gen concentrations which under our experimental conditions are
constant. Results are presented in Table 2.
1Au2Pd has a productivity of almost 18; 000 mmolH =gPd h and
2O2
a
selectivity of 59%, higher than the Pd sample (ꢂ 9500
mmolH =gPd h and 49% selectivity). The most surprising result is
2O2
the totally different catalytic trend of 1Pd2Au catalyst with respect
to 1Au2Pd, the two catalysts differing only in the order in which
the metals are introduced on the support.
3.3. Catalyst characterization
The results reported above on the reactivity of the different cat-
alysts clearly indicate that the catalyst performance is strongly
dependent on the preparation method. This can induce different
metal particle morphologies and/or Pd–Au interactions. For this

126
F. Menegazzo et al. / Journal of Catalysis 268 (2009) 122–130
20000
15000
10000
5000
0
80
60
40
20
0
Fig. 4. Comparison between H2O2 productivity and selectivity for H2O2 direct synthesis performed at 1 bar and 10 bar.
purpose the samples were analyzed with different complementary
techniques with the aim of rationalizing the catalytic behavior and
hopefully identifying the criteria that may improve activity and
selectivity.
1.9%, to 1.4%, to 0.6% for AuPd, 1Au2Pd, and 1Pd2Au samples,
respectively. The interaction between gold and palladium is further
confirmed by the refined unit cell parameters, a = 4.058 Å
(1Au2Pd); 4.052 Å (AuPd); and 4.021 Å (1Pd2Au), that indicate
the formation of a gold-enriched alloy phase, in particular in the
first two cases (the reported refined unit cell parameter for pure
gold is in the range 4.06–4.07 Å).
3.3.1. XRD diffraction
Fig. 5 reports the XRD patterns of bimetallic samples, the 2h
positions corresponding to tetragonal and monoclinic ZrO2 being
indicated. According to the quantitative phase analysis, the PdAu
sample is richer in monoclinic with respect to tetragonal phase
by a factor of ꢂ2.5, while the monoclinic/tetragonal ZrO2 ratio is
close to 1 in both the 1Pd2Au and 1Au2Pd catalysts. This fact can
be ascribed to the different calcination treatments of the samples:
the co-impregnated PdAu has been calcined once, while the other
catalysts have been calcined twice after both Au DP and Pd IW
impregnation. In the same figure, the 2h positions corresponding
to the characteristic lines for pure Pd and Au are indicated. Despite
the characteristic peaks of pure Pd, pure Au, and possible AuPd al-
loys being largely overlapped to the diffraction peaks of monoclinic
ZrO2, a careful comparison of XRD patterns suggested the occur-
rence of a AuPd alloy with different contents in the three different
samples (Fig. 5, inset) This was confirmed by the Rietveld fits
which resulted in weight fractions of the PdAu alloy varying from
3.3.2. HRTEM and X-ray EDS analysis
HRTEM measurements were performed on both the bimetallic
samples and the Pd monometallic catalyst. The Pd sample showed
small roundish particles, uniform in shape and size, as indicated by
the narrow particle size distribution (not shown for the sake of
brevity). The average diameter is 1.5 nm. On the contrary, a high
heterogeneity of the metallic phase was observed in the presence
of gold. In particular, different morphologies and compositions of
the metallic phase were observed depending on the preparation
method.
The 1Au2Pd catalyst contains mainly large roundish bimetallic
particles with size between 30 and 100 nm, with an average diam-
eter of 55 nm. The X-ray EDS analysis showed that the composition
of these large particles is bimetallic, with a Au/Pd ratio ranging
from 0.3 up to 1.5, as summarized in Fig. 6b. In Fig. 6a, an image
Pd
Au
Au
Pd
°
37
38
39
40
*
2
θ
*
*
*
**
1Au2Pd
*
*
*
*
1Pd2Au
PdAu
10
20
30
40
50
60
70
80
2θ
Fig. 5. XRD patterns of Pd–Au bimetallic samples (s ZrO2 tetragonal; w ZrO2 monoclinic). Inset: 2h-region showing the most intense peak of the AuPd phase.

F. Menegazzo et al. / Journal of Catalysis 268 (2009) 122–130
127
Fig. 6. HRTEM image of a typical bimetallic particle (section a) and EDS Au/Pd ratio vs. particle size (section b). Image taken at an original magnification of 300,000ꢀ.
of a typical bimetallic particle is reported, while in section b the
Au/Pd ratio vs. particle size of the same sample is shown. An
approximately linear relationship between the composition and
the dimension of the bimetallic particles is evident: the gold con-
tent increases with the particle size. Independently from the em-
ployed preparation method (DP) that usually produces small gold
particles, after Pd impregnation and subsequent calcination the
presence of gold has been detected only in the large bimetallic par-
ticles. These findings are in agreement with those of Herzing et al.
[42], where the authors observed by STEM-XEDS that the amounts
of Au and Pd in the particles vary markedly with their size; the
smallest particles contain almost exclusively Pd and the concentra-
tion of Au increases as the particle size increases. This feature indi-
cates that a gold sintering process occurs.
the EDS Au/Pd ratio is larger than that obtained for 1Au2Pd and
it varies from about 3 up to 7, indicating that the bimetallic parti-
cles are richer in Au. The presence of very large particles containing
mainly Au can be related to the sample preparation procedure,
where gold has been deposited by co-impregnation, which is not
suitable for obtaining highly dispersed Au nanoparticles. This view
is confirmed by the HRTEM results on the monometallic gold sam-
ple prepared according to the same procedure (not shown for the
sake of brevity).
3.3.3. TPR analyses of fresh and used catalysts
After metal deposition and calcination, a TPR analysis was per-
formed to investigate the metal oxidation state and to identify pos-
sible interactions between palladium and gold. Fig. 7 (section a)
shows the TPR analyses for the fresh bimetallic samples. The TPR
profile of zirconia support (not shown) is completely flat and does
not show any peak, while the TPR profile of the monometallic pal-
ladium catalyst (not shown) shows only a negative peak at about
340 K, which corresponds to hydrogen evolution (mass spectrom-
etry evidence) and can be assigned to Pd b-hydride decomposition
[43,44]. This means that the as-prepared monometallic Pd sample
contains only room temperature reducible PdO species. On the
contrary, the TPR profile of bimetallic PdAu catalyst is completely
flat without Pd hydride decomposition evidence. This feature is a
good evidence [45,46] of Pd alloying, when Pd dispersion is not
high. So, any change of the Pd hydride peak could be related either
to alloy formation or to higher Pd dispersion in the bimetallic
sample.
Finally, some small metal particles with an average size of 2 nm
have also been observed on the 1Au2Pd sample (not shown for the
sake of brevity). However, EDS analysis revealed that they contain
only Pd.
A totally different morphology of the metallic phase was ob-
served in the case of the 1Pd2Au catalyst in which the order of
deposition of the two metals has been inverted. Only small parti-
cles have been detected on this sample. The size distribution is nar-
row and the particles have an average diameter of 3 nm (not
shown for the sake of brevity). The analysis of the diffraction
fringes of the particles evidenced the presence of Au, Pd, and
PdO, respectively. Particles of 15 nm were rarely observed and
their composition is bimetallic, as confirmed by the EDS analysis.
Finally, the PdAu sample exhibits small Pd particles of 2 nm
mean size as well as a vast amount of very large bimetallic parti-
cles with size in the 100–300 nm interval. However, in this case,
The TPR profile of fresh 1Au2Pd shows the presence of a Pd b-
hydride-like phase shifted up to about 360 K. Neutron diffraction
a
1Au2Pd
b
used 1Au2Pd
1Pd2Au
PdAu
used 1Pd2Au
used PdAu
300
400
500
600
700
300
400
500
600
700
Temperature, K
Temperature, K
Fig. 7. TPR patterns of fresh (a) and used (b) Pd–Au bimetallic samples.

128
F. Menegazzo et al. / Journal of Catalysis 268 (2009) 122–130
and thermal desorption spectroscopy studies [47] on different Pd–
Au alloys in the presence of deuterium showed that the hydrogen
or deuterium desorption characteristics change with the gold con-
tent. A different distribution of deuterium over octahedral and tet-
rahedral interstitial sites has been determined. It has been shown
that the tetrahedral occupancy increases with increasing Au con-
tent and this could be related to the improved activity in hydrogen
peroxide production. Moreover, an additional peak centered at
400 K related to the reduction of a AuPdO oxide-like phase is
observed.
a
.1 A
2110
1945
1898
2158
2001
Finally, also the TPR profile related to 1Pd2Au is featureless,
indicating a high dispersion of the metallic particles in agreement
with the HRTEM results that evidenced the presence of Pd and Au
nanoparticles.
b
c
.1 A
The TPR analyses of the bimetallic samples after reaction at
atmospheric pressure are also shown in Fig. 7 (section b), and
can help to understand some of the catalytic results reported
above. It should be noted that because of the catalytic reaction pro-
cedure these used catalysts are indeed the actual catalysts operat-
ing in the system. This point is supported by the observation that
catalysts can be recycled up to three times without loss of activity
and selectivity. Interestingly, all used samples show reduction
peaks above 650 K, associated with sulfate decomposition. As al-
ready observed for zirconia-supported catalysts [31,33], sulfate
ions, coming from sulfuric acid added to acidify the solution in cat-
alytic tests, can adsorb on the un-promoted support surface chang-
ing it into an actually sulfated sample. The TPR profile of the used
1Au2Pd catalyst, i.e. the sample with the highest activity and selec-
tivity, does not change after the catalytic reaction. It shows both
the Pd hydride decomposition peak, meaning that there is a palla-
dium–gold phase that is able to absorb hydrogen at ambient tem-
perature, and a reduction peak at 400 K, indicating the presence of
a AuPdO phase.
2105
2079
1938
2159
1889
1993
2087
.1 A
1926
1958
1895
2169
d
.1 A
2111
2134
1927
1895
1950
3.4. Nature of metallic surface sites determined by FTIR spectra
analysis of adsorbed CO
2161
2003
2000
Wavenumber [cm-1 ]
In order to study the nature of the exposed sites, we performed
CO adsorption at low temperature on the samples as-prepared or
pre-treated at r.t. under different atmospheres, i.e. only hydrogen
or hydrogen and then oxygen. These pre-treatments were accom-
plished in order to mimic the experimental conditions under which
the catalytic tests have been performed.
2200
2100
1900
1800
Fig. 8. FTIR spectra of CO adsorbed at 180 K on Pd (section a), 1Au2Pd (section b),
1Pd2Au (section c), and PdAu (section d) after pre-treatment in H2 at r.t. (fine
curves) and after pre-treatment in H2 and O2 at r.t. (bold curves).
CO adsorption at 180 K on the as-prepared samples (i.e. out-
gassed for 1 h at r.t.) revealed the presence of bands at about
2160 cmꢃ1 and 2135 cmꢃ1 related to CO adsorbed on Pd2+ ions sta-
bilized by neighboring chloride ions [48] due to the nature of the
Pd precursor and on Pdd+ at the surface of oxidized Pd0 particles
[49] (data not shown). The FTIR spectra collected after CO adsorp-
tion at 180 K on Pd (section a), 1Au2Pd (section b), 1Pd2Au (section
c), and PdAu (section d) pre-treated in H2 at r.t. (fine curves) and in
H2, then in O2 at r.t (bold curves) are shown in Fig. 8. All spectra
reported in Fig. 8 show similar features, in particular the position
of the bands, typically in the range of CO adsorbed on top, twofold
bridged or threefold bridged on Pd0 sites. In the case of 1Pd2Au
(section c) a band at 2169 cmꢃ1, due to CO adsorbed on the Zr4+
cations of the support, is also observed. The overall intensities of
the spectra are very different, the lowest intensity being observed
in the case of 1Au2Pd, while the intensities related to both 1Pd2Au
and PdAu are higher than the corresponding ones on the monome-
tallic sample. The lowest intensity of the spectrum of 1Au2Pd is re-
lated to the presence of large particles, as evidenced by HRTEM
analysis reported in the previous section.
estimated the % intensity decrease for every carbonylic species
and found that it is less evident in the case of the monometallic
catalyst (50–70%, section a), while it is larger on the bimetallic
samples, the highest one being observed in the 1Au2Pd sample
(95%, see section b). This phenomenon could be related to an
electronic and chemical effect, induced by gold on the palladium
exposed sites. In order to get a deeper understanding of the ob-
served features we carefully subjected the experimental spectra
of the hydrogen and hydrogen–oxygen pre-treated samples to a
curve fitting procedure (the spectra collected after hydrogen–oxy-
gen pre-treatment and related deconvolutions are shown in Fig. 9).
Oxygen adsorption sites without dissociation in close contact
with sites able to dissociate hydrogen are needed in order to
produce hydrogen peroxide with a high selectivity. The marked
decrease of the carbonylic bands related to the bimetallic samples
after H2, then O2 interaction indicates that a large fraction
of the surface of these catalysts is covered by an oxide layer
where oxygen can chemisorb molecularly according to a recent
paper [50] based on temperature-programed desorption (TPD)
measurements.
Generally, the overall intensity of all spectra is decreased after
pre-treatment in H2 and O2 (bold curves in Fig. 8), suggesting the
formation of a surface oxide layer, unable to adsorb CO [49]. We

F. Menegazzo et al. / Journal of Catalysis 268 (2009) 122–130
129
ready illustrated in the HRTEM-X-ray EDS section, mostly large
bimetallic particles are present in this sample. The higher oxidiz-
ability of palladium sites in this sample can be related to an elec-
a
.05 A
2158
tronic effect of gold atoms on the palladium ones, as
a
2108
consequence of the higher electronegativity of gold (2.54) with re-
spect to palladium (2.2).
1880
1941
A theoretical work [54] concerning atomic configurations of Pd
atoms in PdAu(1 1 1) bimetallic surfaces shows that the Pd–Au
bond is found to be slightly ionic, and it is stronger than Au–Au
and even Pd–Pd bonds. Therefore, gold atoms close to the surface
palladium atoms may cause the occurrence of a partial positive
charge on the latter and, consequently, their easier oxidation. Addi-
tionally, the bands at 2104 and 1985 cmꢃ1 are shifted with respect
to the frequency of the same CO species on the monometallic Pd
sample. The observed shifts can be due to differences in the lateral
interactions between the adsorbed CO. In particular, the band at
1985 cmꢃ1 is significantly red shifted, suggesting that the Pd0 sites
are more isolated, as a consequence of the alloying of either of the
two metals in this sample and not as a consequence of the smaller
size of the particles. In fact, this hypothesis can be excluded on the
basis of the HRTEM data already shown. Therefore, we assign the
band at 1985 cmꢃ1 to CO bridge bonded on Pd couples exposed
at the edges of the metal particles, modified by surrounding gold
atoms. These couples of Pd edge sites may be relevant to hydrogen
dissociation. These sites are less affected by the H2–O2 interaction
as indicated by the calculated decrease % (25%). On the contrary,
the band at 1941 cmꢃ1, assigned to bridged CO species on Pd0 ex-
posed at the surface of (1 1 1) or (1 0 0) terraces, does not show any
significant frequency shift and can be due to CO adsorbed on the
small pure Pd particles, detected by HRTEM and by the EDS analy-
sis previously discussed.
1809
1999
b
.05 A
2160
2134
2104
1941
1879
1985
c
.05 A
2131
2101
2169
1926
1966
2085
1881
d
.05 A
2161
2110
The spectra related to 1Pd2Au show quite different profiles (see
Figs. 8 and 9, sections c). First of all, the band at 2160 cmꢃ1 (Pd2+
sites) is totally missing, due to the removal of chlorine atoms dur-
ing the basification necessary for the Au deposition. Moreover, it is
evident from Fig. 9 that an additional band is required in order to
1944
1878
2001
2000
Wavenumber [cm-1]
fit the spectral region 2150–2000 cmꢃ1
. The weak band at
2101 cmꢃ1 may be due to CO adsorbed on the Au0 sites of the small
particles observed by HRTEM. This component was not observed
when the sample was pre-treated in H2, because of the very high
intensity of the bands related to CO on Pd0. Moreover, the total lack
of the band related to CO bridge bonded to Pd couples exposed on
edges is also evident, possibly as a consequence of a decoration ef-
fect of gold.
2200
2100
1900
1800
Fig. 9. FTIR spectra of CO adsorbed at 180 K on Pd (section a), 1Au2Pd (section b),
1Pd2Au (section c), and PdAu (section d) after pre-treatment in H2 and O2 at r.t.
(bold curves) and relative deconvolutions (fine curves).
The PdAu catalyst (Figs. 8 and 9, section d) showed almost the
same spectroscopic features observed in the case of the 1Au2Pd
sample (sections b), as far as the bands position and their behavior
upon the different pre-treatments are concerned, except in the
case of the component at 2003 cmꢃ1, due to bridged CO on the
Pd0 edges, which is not shifted at all if compared to the position
of the same band related to the monometallic sample (sections
a). HRTEM measurements already evidenced the presence of small
Pd particles.
FTIR characterization by adsorbed CO evidenced firstly that the
nature and the relative abundance of the metallic sites strongly de-
pend on the preparation method. A different behavior of Pd species
toward the H2, then O2 pre-treatment at r.t. indicates that an oxidic
layer is more easily produced on the bimetallic catalysts, due to the
Pd modification by the presence of Au. This modification may be a
consequence of the higher electronegativity of gold, inducing on
the near Pd atoms a partial positive charge. The highest effect is ob-
served on 1Au2Pd, which is also the most active catalyst. More-
over, Pd0 couples mainly localized on edge sites and strongly
modified due to the presence of a AuPd bimetallic phase, present
only on this sample, may explain the enhanced reactivity and
selectivity of the 1Au2Pd bimetallic catalyst.
Fig. 9, where the spectra of CO interaction on all samples after
H2–O2 pre-treatment are shown, allows some interesting consider-
ations. In the case of the monometallic catalyst, bands at 2108,
1999, 1941, 1880, and 1809 cmꢃ1 produced by CO interaction at
180 K are observed that can be assigned on the basis of their
behavior upon increasing temperature (not reported here for the
sake of brevity) and of literature data [51]. The band at
2108 cmꢃ1 is due to linear CO species on Pd0 sites exposed at the
surface of (1 1 1) facets, the one at 1999 cmꢃ1 can be assigned to
bridged CO on the Pd0 edges [52,53], while the component at
1941 cmꢃ1 can be related to bridged CO species on Pd0 exposed
at the surface of either (1 0 0) or (1 1 1) facets. Finally, the bands
at 1880 and 1809 cmꢃ1 are due to CO adsorbed on different three-
fold hollow Pd sites.
The most active 1Au2Pd catalyst, pre-treated in hydrogen–oxy-
gen, exhibits two components at 2160 and 2134 cmꢃ1, already as-
signed to CO adsorbed on Pd2+ and Pdd+ sites and weak components
at 2104, 1985, 1941, and 1879 cmꢃ1, assigned to CO on different Pd
sites (Fig. 8, section b). The weakness of these bands indicates that
palladium sites in this sample are the most easy to oxidize. As al-

130
F. Menegazzo et al. / Journal of Catalysis 268 (2009) 122–130
[14] T.A. Pospelova, N.I. Kobozev, Russ. J. Phys. Chem. 35 (1961) 262.
4. Conclusions
[15] J.H. Lunsford, J. Catal. 216 (2003) 445.
[16] J.M. Campos-Martin, G. Blanco-Brieva, J.L.G. Fierro, Angew. Chem. 118 (2006)
7116.
[17] G. Li, J.K. Edwards, A.F. Carley, G.J. Hutchings, Catal. Commun. 8 (2007) 247.
[18] P. Landon, P.J. Collier, A.J. Papworth, C.J. Kiely, G.J. Hutchings, Chem. Commun.
(2002) 2058.
[19] J. Edwards, B. Solsona, E. Ntainjua, A.F. Carley, A. Herzing, C. Kiely, G.
Hutchings, Science 323 (2009) 1037.
[20] G. Luft, U. Luckhoff, Chem. Eng. Technol. 66 (1994) 187.
[21] S. Abate, S. Melada, G. Centi, S. Perathoner, F. Pinna, G. Strukul, Catal. Today
117 (2006) 193.
[22] V.R. Choudhary, C. Samanta, T.V. Choudhary, Appl. Catal. A – Gen. 308 (2006)
128.
[23] Q. Liu, J.C. Bauer, R.E. Schaak, J.H. Lunsford, Appl. Catal. A – Gen. 339 (2008)
130.
[24] M. Haruta, T. Kobayashi, H. Sano, N. Yamada, Chem. Lett. (1987) 405.
[25] M. Haruta, N. Yamada, T. Kobayachi, S. Iijima, J. Catal. 115 (1989) 301.
[26] D. Andreeva, V. Idakiev, T. Tabakova, A. Andreev, J. Catal. 158 (1996) 354.
[27] J.K. Edwards, B. Solsona, P. Landon, A.F. Carley, A. Herzing, C. Kiely, G.
Hutchings, J. Catal. 236 (2005) 69.
[28] N.S. Patil, B.S. Uphade, D.G. McCulloh, S.K. Bhargava, V.R. Choudhary, Catal.
Commun. 5 (2004) 681.
[29] D.W. Goodman, Catal. Lett. 99 (2005) 1.
[30] F. Moreau, G. Bond, Catal. Today 122 (2007) 260.
[31] S. Melada, R. Rioda, F. Menegazzo, F. Pinna, G. Strukul, J. Catal. 239 (2006) 422.
[32] F. Menegazzo, F. Pinna, M. Signoretto, V. Trevisan, F. Boccuzzi, A. Chiorino, M.
Manzoli, ChemSusChem. 1 (2008) 320.
[33] F. Menegazzo, P. Burti, M. Signoretto, M. Manzoli, S. Vankova, F. Boccuzzi, F.
Pinna, G. Strukul, J. Catal. 257 (2008) 369.
[34] G. Bernardotto, F. Menegazzo, F. Pinna, M. Signoretto, G. Cruciani, G. Strukul,
Appl. Catal. A – Gen. 358 (2009) 129.
[35] S. Melada, M. Signoretto, F. Somma, F. Pinna, G. Cerrato, G. Meligrana, C.
Morterra, Catal. Lett. 94 (2004) 193.
[36] M. Signoretto, S. Melada, F. Pinna, S. Polizzi, G. Cerrato, C. Morterra, Micrpor.
Mesopor. Mater. 81 (2005) 19.
[37] S.J. Gregg, K.S.W. Sing, Adsorption, Surface Area and Porosity, second ed.,
Academic Press, 1982. p. 111.
In this work the influence of the preparation method of bimetal-
lic Pd–Au on zirconia catalysts for the direct synthesis of hydrogen
peroxide has been evidenced. It is quite clear from the reactivity
data that the order and the method by which the two metals are
deposited on the support are critical for obtaining a high activity
and a high selectivity in H2O2 formation. It has been proved that
depositing firstly Au by DP and subsequently Pd by IW is a method
that allows to obtain catalysts with a productivity at pressure as
high as 18; 000 mmolH =gPd h and a stable selectivity of 59%, sig-
2O2
nificantly higher than the one observed at atmospheric pressure.
The latter behavior is unusual (the opposite is normally observed)
and very promising.
The preparation method strongly influences the morphology
and composition of the metallic phase as was demonstrated by
HRTEM and X-ray EDS analysis. The 1Au2Pd most active catalyst
contains mainly large roundish bimetallic particles with a Au/Pd
ratio ranging from 0.3 up to 1.5. TPR related to 1Au2Pd showed,
among other features, the occurrence of a positive reduction peak
of a AuPdO oxide-like phase indicating an enhanced stability of an
oxidic phase for this catalyst.
FTIR results evidenced an enhanced oxidizability of the palla-
dium sites of the bimetallic catalysts, after the pre-treatment in
H2, then in O2 at r.t. On 1Au2Pd the Pd0 couples, mainly present
on the edge sites, are strongly modified as a consequence of the
presence of a AuPd phase. Moreover, they are oxidized to a lower
extent.
All together these results seem to suggest that the oxidic layer
acts as the activating phase of oxygen in a molecular form, while
hydrogen dissociates on clean Pd0 couples on edges, thereby
explaining the selectivity in hydrogen peroxide formation.
[38] B. Lewis, G. von Elbe, Combustion, Flames and Explosion of Gases, Academic
Press, New York and London, 1961.
[39] S. Melada, F. Pinna, G. Strukul, S. Perathoner, G. Centi, J. Catal. 237 (2006) 213.
[40] D.P. Dissanayake, J. Lunsford, J. Catal. 206 (2002) 173.
[41] E.J. Beckman, Green Chem. 5 (2003) 332.
[42] A.A. Herzing, M. Watanabe, J.K. Edwards, M. Conte, Z.-R. Tang, G.J. Hutchings,
C.J. Kiely, J. Chem. Soc., Faraday Discuss. 138 (2008) 337.
[43] W. Palczewska, in: Z. Paal, P.G. Menon (Eds.), Hydrogen Effects in Catalysis,
Dekker, New York, 1988, p. 373.
[44] Z. Karpinski, Adv. Catal. 37 (1990) 45.
[45] G. Fagherazzi, A. Benedetti, S. Polizzi, A. Di Mario, F. Pinna, M. Signoretto, N.
Pernicone, Catal. Lett. 32 (1995) 293.
Acknowledgment
We thank MIUR (Rome) for financial support.
References
[46] P. Canton, F. Menegazzo, M. Signoretto, F. Pinna, P. Riello, A. Benedetti, N.
Pernicone, Stud. Surf. Sci. Catal. 143 (2002) 1011.
[1] H.J. Riedl, G. Pfleiderer, US 2215883, 1940.
[47] D.E. Nanu, W.J. Legerstee, S.W.H. Eijt, W.G. Haije, J.F. Vente, M.G. Tucker, A.J.
Bottger, Acta Mater. 56 (2008) 6132.
[2] L.W. Gosser, US 4681751, 1987.
[3] W. Gosser, J.T. Schwartz, US 4772485, 1988.
[48] E.A. Sales, J. Jove, M. de Jesus Mendes, F. Bozon-Verduraz, J. Catal. 195 (2000)
88.
[49] T. Schalow, B. Brandt, M. Laurin, S. Schauermann, S. Guimond, H. Kuhlenbeck, J.
Libuda, H.J. Freund, Surf. Sci. 600 (2006) 2528.
[50] J.A. Hinojosa, H.H. Kan, J.F. Weaver, J. Phys. Chem. C 112 (2008) 8324.
[51] E. Ozensoy, D.W. Goodman, Phys. Chem. Chem. Phys. 6 (2004) 3765.
[52] K. Wolter, O. Seiferth, J. Libuda, H. Kuhlenbeck, M. Bäumer, H.J. Freund, Surf.
Sci. 402–404 (1998) 428.
[4] J. Van Weynbergh, J.-P. Schoebrechts, US 5447706, 1995.
[5] B. Bertsch-Frank, I. Hemme, S. Katusic, J. Rollmann, US 6387364, 2002.
[6] G. Paparatto, R. D’Aloisio, G. De Alberti, R. Buzzoni, US 6630118, 2003.
[7] K.M. Vanden Bussche, S.F. Abdo, A.R. Oroskar, US 6713036, 2004.
[8] R. Burch, P.R. Ellis, Appl. Catal. B – Environ. 42 (2003) 203.
[9] C. Samanta, V.R. Choudhary, Catal. Commun. 8 (2007) 2222.
[10] J.K. Edwards, A. Thomas, B. Solsona, P. Landon, A.F. Carley, G. Hutchings, Catal.
Today 122 (2007) 397.
[53] J.B. Giorgi, T. Schroeder, M. Bäumer, H.J. Freund, Surf. Sci. 498 (2002) L71.
[54] D. Yuan, X. Gong, R. Wu, Phys. Rev. B 78 (2008) 035441.
[11] D. Hancu, E.J. Beckman, Green Chem. 3 (2001) 80.
[12] Q. Liu, J.H. Lunsford, J. Catal. 239 (2006) 237.
[13] J.A. Rodriguez, Prog. Surf. Sci. 81 (2006) 141.