Selective oxidation of CO over model gold-based catalysts in the presence of H2

Article abstract of DOI:10.1016/j.jcat.2005.01.011

The model catalysts Au/Al₂O₃, Au/ZrO₂, and Au/TiO₂ were produced by laser vaporization of a metallic gold rod followed by deposition of the formed clusters onto the support powders. This technique allows to obtain a narrow size distribution of highly dispersed gold particles on the support and, most importantly, similar sizes whatever the support. This makes it possible to accurately study of the influence of the support identity on the catalytic reaction. A detailed investigation of the preferential oxidation of CO in the presence of H₂ was undertaken. Catalytic performances in the PROX reaction were compared with those in the oxidation of CO in the temperature range of 25-420°C. A boost in the conversion of CO was observed in the presence of H₂ at low temperature; the extent of this boost is dependent on the support identity. Hence the reactivity order found for CO oxidation (Au/Al₂O ₃ ? Au/ZrO₂ < Au/TiO₂) was changed. In fact, in the presence of H₂, the reaction rates for the oxidation of CO become rather similar on all three systems.

Full text of DOI:10.1016/j.jcat.2005.01.011

Journal of Catalysis 230 (2005) 476–483
www.elsevier.com/locate/jcat
Selective oxidation of CO over model gold-based catalysts
in the presence of H2
Cécile Rossignol, Sandrine Arrii, Franck Morfin, Laurent Piccolo, Valérie Caps,
Jean-Luc Rousset ∗
Institut de Recherches sur la Catalyse (CNRS), 2 Avenue Albert Einstein, F-69626 Villeurbanne, France
Received 5 November 2004; revised 17 December 2004; accepted 7 January 2005
Abstract
The model catalysts Au/Al O , Au/ZrO , and Au/TiO were produced by laser vaporization of a metallic gold rod followed by deposition
2
3
2
2
of the formed clusters onto the support powders. This technique allows to obtain a narrow size distribution of highly dispersed gold particles
on the support and, most importantly, similar sizes whatever the support. This makes it possible to accurately study of the influence of the
support identity on the catalytic reaction. A detailed investigation of the preferential oxidation of CO in the presence of H was undertaken.
2
◦
Catalytic performances in the PROX reaction were compared with those in the oxidation of CO in the temperature range of 25–420 C.
A boost in the conversion of CO was observed in the presence of H at low temperature; the extent of this boost is dependent on the support
2
identity. Hence the reactivity order found for CO oxidation (Au/Al O ꢀ Au/ZrO < Au/TiO ) was changed. In fact, in the presence of H ,
2
3
2
2
2
the reaction rates for the oxidation of CO become rather similar on all three systems.
2005 Elsevier Inc. All rights reserved.
Keywords: Gold catalysis; Model catalyst; Laser vaporization; CO oxidation, PROX
1. Introduction
structure on a number of preparation parameters, such as
temperature, pH, type of incorporation, etc. The main issue
seems be the limited choice in the gold precursors [12]. The
chlorinated HAuCl4 precursor is mostly used; the method
(deposition-precipitation, co-precipitation, or ion exchange)
is then selected according to the nature of the support on
which gold is deposited. This leads, after heat treatment, to
gold particles with different sizes on different supports and
makes it difficult to investigate any support effects in the
catalytic reactions [13]. In addition, these chemical meth-
ods suffer not only from low reproducibility but also from
unwanted incorporation of chloride residuals and/or other
contaminants (Na, etc.) in the catalyst, even after extensive
washing and/or postsynthetic heat treatments.
The catalytic potentials of supported gold-based nanopar-
ticles have been considered for a number of useful applica-
tions; among these, air-cleaning by oxidation of CO [1–5],
hydrogen purification in fuel cell technology via the water–
gas shift reaction [6–8] and the selective oxidation of CO in
the presence of hydrogen (SELOX or PROX for preferential
oxidation) [1,9–11] have received particular attention. The
advantage of using gold in these reactions is indeed its cat-
alytic activity at low temperature as compared with the other
metals from the platinum group. However, despite exten-
sive efforts in the preparation, characterization, and testing
of gold catalysts, the reaction mechanisms still remain un-
clear. This is due partly to the structure sensitivity of the
gold-catalyzed reactions and the sharp dependency of gold
Recently we reported on the preparation of model cat-
alysts using of laser vaporization [14]. With this method,
we managed to get highly dispersed small gold particles
(2–3 nm) at the surface of the traditional metal oxide pow-
ders (Al2O3, ZrO2, and TiO2) and a narrow size distribution
*
Corresponding author.
E-mail address: rousset@catalyse.cnrs.fr (J.-L. Rousset).
0021-9517/$ – see front matter
doi:10.1016/j.jcat.2005.01.011
2005 Elsevier Inc. All rights reserved.

C. Rossignol et al. / Journal of Catalysis 230 (2005) 476–483
477
of these particles. Furthermore, it was shown that since the
Table 1
Metal loadings (ICP analysis) and mean diameters of the gold particles in
fresh and used catalysts (TEM). In parentheses are the standard deviations
metallic clusters were formed before deposition, their size
was predetermined and similar size distributions were ob-
tained on all the supports. With this technique, indeed, the
metal particle size did not depend on the interaction of a
metal precursor with the support (and hence on the nature of
the support), since the preformed gold particles were physi-
cally deposited on the support. This ligand-free method thus
turned out to be a unique way to produce clean model cat-
alysts, avoid heat treatments and make it possible to accu-
rately study the influence of the identity of the support on
the catalytic oxidation of CO. A clear support effect on the
CO oxidation rates was indeed evidenced: titania was better
than zirconia, which was much better than alumina.
In this paper we investigate support effects in the selec-
tive oxidation of CO in the presence of hydrogen by using
the same gold-based model systems. Results are compared
with those obtained from the oxidation of CO. The influence
of H2 on the CO conversion, and its extent depending on
the support, will be presented and discussed. The selectivity
of the catalysts will also be discussed through a comparison
of their behavior in the PROX reaction and in the oxidation
of H2. A possible mechanism for the PROX reaction will
be suggested on the basis of these observations. Some con-
clusions about the type of catalyst that could be suitable for
PROX will then be drawn.
Sample
Gold metal
content
(wt%)
Mean diameter (nm)
Fresh
catalysts
Used
catalysts
Au/γ -Al O
Au/ZrO
Au/TiO
Au/TiO (Ref. WGC) 1.47
0.080
0.054
2.6 (1.6)
2.9 (1.7)
2.9 (1.7)
3.7 (1.5)
3.1 (1.6)
3.8 (1.9)
3.2 (1.3)
N/A
2
3
2
2
0.023
a
2
a
In house measurements give Au 1.38 wt%, Na 0.055 wt%, Cl 47 ppm.
is well described in a previous paper [14]; the results are
summarized in Table 1. The gold clusters deposited on the
different supports have similar and narrow size distributions
centered around 3 nm. They are in the metallic state Au0.
2.2. Catalytic activity measurements
The materials were tested in three reactions—the oxida-
tion of CO, the selective oxidation of CO in the presence
of H2 and the oxidation of H2—according to the same pro-
tocol. Only the incoming gas mixture was changed from
one reaction to the other. These tests were carried out at at-
mospheric pressure in a continuous-flow fixed-bed reactor.
750 mg of the as-obtained gold-based powders was loaded
into the quartz reactor, located in a ceramic furnace, which
resulted in a length of ∼ 10 mm for the catalytic bed. No
pretreatment was applied (an important advantage of this
catalyst over catalysts prepared from salt precursors is the
negligible amount of impurities and residuals). A Au/TiO2
reference catalyst (Type A, Sample Number 53) was also
purchased from the World Gold Council [15] and tested.
To allow direct comparison of the catalytic results, 11.6 mg
of this catalyst was diluted in 740 mg γ -Al2O3 (which is
inactive for the three reactions studied under the present con-
ditions). This ensures a gold content (0.17 mg) similar to
that in the test with the model Au/TiO2 catalyst (prepared by
laser vaporization).
2. Experimental
2.1. Catalyst preparation
The three catalysts used in this study (Au/Al2O3, Au/
ZrO2, and Au/TiO2) were prepared by low-energy cluster
beam deposition, which has been described in detail else-
where [14]. In brief, the second harmonic of a Nd:YAG
pulsed laser is used to vaporize the metal from the rod,
which creates a plasma. Cluster nucleation and growth oc-
cur when a continuous flow of inert gas (a mixture of He
and Ar) is introduced into the vacuum chamber. Differential
pumping extraction of the clusters through a skimmer yields
a cluster beam that carries neutral and ionized species. After
removal of the ionized clusters by electrostatic deflection,
the neutral clusters are deposited on the substrates. To de-
posit the clusters on the supports, which in the present case
are alumina (Condea Puralox Sc-Fa-215, 221 m2 g−1), tita-
nia (Hombikat T100 20-S20 Type, anatase 60 m2 g−1), and
zirconia (MEL chemical, 79 m2 g−1) powders, a device was
developed that stirs the powder in front of the cluster beam.
This device favors homogeneous cluster deposition on most
of the powder grains. After deposition, the samples are air-
transferred, characterized, and used for catalytic reactions.
For these experiments, a gold rod of 99.99% purity from
Goodfellow was used. The metal content of each catalyst
measured by chemical analysis (ICP or ICP-MS) is reported
in Table 1. The catalyst characterization (TEM-EDX, XPS)
The reactant flow consisted of a mixture of 2% CO +
2% O2 + 96% He for the oxidation of CO, 2% CO +
2% O2 + 48% H2 + 48% He for the preferential oxidation
of CO (PROX) and 48% H2 + 2% O2 + 50% He for the ox-
idation of H2 (all percentages are vol%). All the high-purity
(> 99.995%) gases were purchased from Air Liquide. Mass-
flow controllers (Brooks) were calibrated for the above gases
with a soap-bubble flow meter. In this way, accurate and re-
producible CO/O2/He, H2/O2/He or CO/O2/H2/He mixtures
could be generated at a typical flow rate of 50 Nml min−1
.
The gas mixture could then be sent either through the reactor
or directly to the gas chromatograph for analysis and calibra-
tion of the GC. Calibration of the GC was done with a gas
mixture containing 3% CO, 3% CO2 and 6% O2 in N2. The
reactor was heated to 450 ◦C and then cooled to 25 ◦C at a
heating (and cooling) rate of 1 ◦C min−1. Several reaction
cycles, each consisting of a heating and a cooling period,

478
C. Rossignol et al. / Journal of Catalysis 230 (2005) 476–483
were recorded to monitor possible hysteresis, deactivation
or activation. Product analysis was carried out on-line with
a Varian-Micro GC (CP2003) equipped with a TCD detec-
tor. Two columns, operated at 50 ◦C, were used in parallel:
a Molsieve 5A column (Ar as carrier gas) to quantify O2
and CO, and a poraPLOT Q column (He as carrier gas) to
quantify CO2. To avoid polluting the Micro GC columns,
the steam was trapped at the exit of the reactor in a con-
denser held at 3 ◦C, using a Peltier gas cooler (M&C cooler
ECP 1000). We used the usual procedure to ensure that the
activity was measured in the kinetic regime: the flow rates
and the mass of catalyst were varied in the same proportions
to provide a constant contact time and consequently a con-
stant space velocity. The conversion did not vary as the mass
flow was changed, indicating that the measured rate was not
limited by external transfers (film diffusion).
the poorest conversion of CO. Here we compare our model
catalysts with the Au/TiO2 reference catalyst prepared by
deposition-precipitation according to a protocol developed
by AIST (Japan). As can be seen in Fig. 1d, this catalyst is
superior for CO conversion compared with our catalyst pre-
pared by laser vaporization (Fig. 1c). Given that the average
particle size of gold (3.7 nm) is even larger than in our ma-
terial (2.9 nm), we attribute this higher activity to some side
effects due to the presence of other elements (Na 0.034 wt%,
Cl 47 ppm) or, more probably, to a specific interaction of
the gold particle with the support engineered during the heat
treatment that is required in the chemical methods of prepa-
ration. In any case, this comparison shows that the activity
of our model catalyst is of the same order of magnitude as
that of the reference catalyst for the oxidation of CO. Ac-
tually, this is also the case for the oxidation of H2 and the
oxidation of CO in the presence of H2, but since the point of
the present study is to compare catalytic properties on model
systems, the following discussion will focus only on them.
3. Results
As mentioned above, several reaction cycles, each con-
sisting of a heating and a cooling period, were performed.
The CO conversion (under CO oxidation conditions) was
found to be always lower in the first heating stage than in
the following runs. Later on, no modification of the catalytic
activity was observed in further cycles, indicating a stabi-
lization of the catalysts. To better understand this activation
of the catalyst during the first run, an experiment was made
with a pre-cycle (heating and cooling) under He (reactant-
free) before two cycles under reactive mixture. After this
first cycle in helium, the two subsequent cycles under CO
oxidation conditions were then found to be similar to those
obtained after a first cycle under a CO + O2 mixture. This
indicates that activation of the catalyst is not related to the
formation of active species during contact with the reac-
tive mixture. Activation of the catalyst through engineering
of irreversible metal–support interactions during heating to
450 ◦C is also ruled out on the basis of an experiment show-
ing the same first-cycle activation on the same used catalyst
left for 45 days in air/moisture. Desorption of water from the
catalyst surface upon heating at 450 ◦C seems to be respon-
sible for the catalyst activation. Hence results corresponding
to the second cycle (and further cycles), which were already
found to be more representative of the intrinsic activity of
gold catalysts [14,16], are used for the following discus-
sion.
3.2. H2 oxidation in a CO-free mixture (48% H2, 2% O2 in
He)
Before proceeding to the PROX reaction, we performed
H2 oxidation with a large excess of hydrogen. Because of the
large error in the quantification of H2O and the small change
in the H2 concentration due to the large excess of hydro-
gen (for total O2 consumption, only 8% of the hydrogen can
be oxidized), we evaluated the activity of the catalyst in the
oxidation of H2 by quantifying the O2 conversion. As can
be seen from Figs. 1a and 1b, Au/Al2O3 and Au/ZrO2 are
more active for the oxidation of H2 than for the oxidation of
CO, whereas Au/TiO2 is more active for the oxidation of CO
(Fig. 1c). Temperatures of half-maximum conversions are
indeed 230 ◦C lower, 25 ◦C lower, and 25 ◦C higher in H2
oxidation than in CO oxidation for Au/Al2O3, Au/ZrO2, and
Au/TiO2, respectively. Maximum H2 conversion is readily
achieved at 170 ◦C on Au/Al2O3, at 225 ◦C on Au/ZrO2, and
at 250 ◦C on Au/TiO2. We note that Haruta et al. [17] found,
for a standard gas containing 1 vol% H2 or CO in air, rather
identical activities for the two reactions over a Au/ZrO2 cata-
lyst prepared by coprecipitation; the better activity found for
H2 oxidation as compared with CO oxidation over our model
Au/ZrO2 catalyst could come from the large partial pressure
of hydrogen we employed. It is also worth noting that, if we
take into account the catalyst loadings (Table 1), the rate of
H2 oxidation (Fig. 2) is similar for all our model catalysts.
This has already been observed, under Haruta’s conditions,
with silica-, alumina-, and titania-supported gold catalysts,
over which H2 oxidation is much less influenced by metal–
support interactions than is CO oxidation [18].
3.1. CO oxidation in a H2-free atmosphere (2% CO,
2% O2 in He)
It is widely accepted that titania is currently the best sup-
port for gold nanoparticles in the catalytic oxidation of CO.
This result was confirmed in a previous paper [14], where
gold nanoparticles were dispersed on alumina, zirconia, and
titania supports with the same size distribution and found
to convert more CO on TiO2 than on ZrO2; Al2O3 yielded
3.3. PROX reaction (2% CO, 2% O2, 48% H2 in He)
Because the ability of the catalyst to oxidize CO more
readily than H2 is generally thought to be a prerequisite for

C. Rossignol et al. / Journal of Catalysis 230 (2005) 476–483
479
Fig. 1. CO and H conversion and selectivity as a function of temperature over Au/Al O (a), Au/ZrO (b), Au/TiO (c) and Au/TiO -WGC (d). (!) CO
2
2
3
2
2
2
conversion in H -free mixture, (P) H conversion in CO-free mixture, (") CO conversion in PROX, (Q) H conversion in PROX and (×) selectivity.
2
2
2
Fig. 2. CO oxidation rate vs temperature in H -free mixture (open symbols) and H oxidation rate vs temperature in CO-free mixture (dark symbols) over
2
2
Au/Al O (E and F), Au/ZrO (P and Q), Au/TiO (! and ") and Au/TiO -WGC (1 and 2). Only the points corresponding to a conversion below 50%
2
3
2
2
2
are plotted.
selective CO oxidation in the presence of H2 [11,18–21], we
expected the support effect to be the same in the PROX re-
action as in the CO oxidation; that is, titania is better than
zirconia and much better than alumina.
However, it is clear for all systems (Fig. 1) that, in
the presence of H2, the conversion of CO reaches a maxi-
mum at a relatively low temperature (65% on Au/Al2O3 at
119 ◦C, 60% on Au/TiO2 at 196 ◦C, and 55% on Au/ZrO2

480
C. Rossignol et al. / Journal of Catalysis 230 (2005) 476–483
Fig. 3. CO oxidation rate vs temperature in H -free mixture (open symbols) and in PROX mixture (dark symbols) over Au/Al O (E and F), Au/ZrO
2
2
2
3
(P and Q), Au/TiO (! and ") and Au/TiO -WGC (1 and 2). Only the points corresponding to a conversion below the maximum conversion reached in the
2
2
PROX reaction are plotted.
Table 2
a
Temperature of half-conversion, reaction rates and activation energies for the oxidation of CO, the oxidation of H and the PROX reaction over our catalysts
2
Catalyst
CO oxidation (H free)
2
H
oxidation (CO free)
CO oxidation in PROX
2
◦
◦
◦
b
◦
(H ) ( C)
1/2 2
T
1/2
( C)
Rate
E
T
1/2
( C)
Rate
E
T
max
( C)
Rate
T
a
a
Au/γ -Al O
350
188
153
77
0.02
0.05
0.42
2.6
22
34
29
28
121
150
177
92
0.39
0.38
0.36
3.2
31
27
31
34
119
172
196
102
0.62
0.29
0.32
3.7
135
166
201
123
2
3
Au/ZrO
2
Au/TiO
2
Au/TiO (Ref. WGC)
2
a
−1
T
= temperature of half-conversion, i.e. temperature at which 50% of the reactant has been converted; E = activation energy (kJ mol ) derived from
a
1/2
−1
−1
◦
Fig. 2; Rate = reaction rates (mmol g
s
) calculated at 80 C from Fig. 1; T
= temperature corresponding to the maximum CO conversion reached in
max
Au
the PROX reaction.
b
Temperature of half-conversion of H .
2
at 172 ◦C) and then decreases rapidly, which is attributed
first to the competition of adsorption between H2 and CO
(favorable to hydrogen when temperature increases) and sec-
ond to the consumption of the oxidant in the oxidation of
hydrogen. Indeed, the temperatures of the maxima corre-
spond approximately to the temperatures of half-conversion
of H2 in the oxidation of H2 in the presence of CO (Ta-
ble 2). At this point about 75% O2 is converted (50%
for the oxidation of H2 and about 25% for the oxidation
of CO).
Nevertheless, below those temperatures, the efficiency or-
der of the catalysts in the PROX reaction seems to be re-
versed as compared with that in the CO oxidation reaction:
CO conversion is now higher on the alumina-supported cat-
alyst than on the zirconia- and titania-supported catalysts.
Indeed, CO oxidation is largely enhanced on the alumina-
supported catalyst, when H2 is present in the reactive mix-
ture, whereas it is only slightly enhanced on the zirconia-
supported catalyst and basically unaffected on the titania-
supported catalyst. If we take into account the loadings of
the catalysts (Table 1), the rates in the catalytic oxidation
of CO in the PROX reaction are rather similar for all three
systems, as seen in Fig. 3. In other words, in the presence
of hydrogen, gold on alumina will convert as much CO as
gold on titania or gold on zirconia. It seems that hydrogen
masks the effect of the support in the oxidation of CO or,
more precisely, that the effect of hydrogen makes our cata-
lysts equivalent for the PROX reaction.
Hence the actual selectivity is found to be much higher
than that expected from the results obtained for CO and
H2 oxidation when O2 is the only counterpart reactant gas
(Fig. 1). A high selectivity for CO2 is achieved at low tem-
peratures on all the catalysts, and it decreases rapidly with
increasing temperature. Actually, the activities of the cata-
lysts for the oxidation of H2 decrease a little in the presence
of CO; the alumina-supported catalyst is the least affected.
With maximum H2 conversions reached at 230 ◦C over
Au/Al2O3, 300 ◦C for Au/ZrO2, and 325 ◦C for Au/TiO2,
Au/Al2O3 remains superior for H2 oxidation compared with
H2-free CO oxidation (see Fig. 1, dark triangles and open
circles), Au/TiO2 remains better for CO oxidation, while the
activity of Au/ZrO2 is now similar in both reactions. In any
case, as in the CO-free H2 oxidation reaction (Section 3.2),
H2 oxidation rates are similar for all supports.

C. Rossignol et al. / Journal of Catalysis 230 (2005) 476–483
481
4. Discussion
idation activity in the whole pressure range investigated.
This reduction diminishes as the CO partial pressure in-
creases and vanishes when the CO partial pressure reaches
2 kPa.
4.1. H2-induced boost in the CO oxidation rate
Generally, the CO conversion over supported Au catalysts
[11,16,22] is reported to decrease when H2 is introduced into
the reaction mixture, even if some exceptions exist [9], like
Au/MgO/Al2O3 [16]. Only for Pt catalysts supported on alu-
mina was it reported in the literature that CO oxidation was
enhanced by the presence of hydrogen [23]. In this case, the
physical origin of the H2-induced increase in the CO oxi-
dation rate has not been fully explained so far, but possible
mechanisms include the effect of hydrogen on the heat of
adsorption of CO and on the interaction of the hydroxylated
Al2O3 support with CO adsorbed on Pt [23].
In the present work, in the presence of hydrogen, the rate
of CO oxidation is boosted and similar on three very differ-
ent supports, reducible and nonreducible oxides, while they
are so different in the pure oxidation of CO. This could be
related to the fact that all three catalysts have a similar ac-
tivity in the oxidation of H2. It is possible that if hydroxyl
groups at the metal–support interface are important for the
CO oxidation reaction, the reaction of hydrogen or water
with the support produces the same amount of these ac-
tive species on all the supports, as was recently discussed
[29]. It is also possible that, under our conditions, molecu-
lar oxygen is transformed, through reaction with hydrogen,
into an active species that is capable of oxidizing CO [30].
This phenomenon has already been invoked in the selective
oxidation of propylene to propylene oxide (PO). The direct
synthesis of PO, with the use of molecular oxygen, is one of
the most important chemical reactions to remain unsolved
by catalysis [31]. Nevertheless, recent studies have shown
that Au catalysts can effectively catalyze the direct vapor-
phase conversion of propylene to PO when hydrogen is used
in addition to molecular oxygen [31,32]. Many highly reac-
tive intermediates, such as H2O2, are known to be involved
during the reaction between H2 and O2 [33]. It has been pro-
posed that these hydrogen peroxy-like species formed on the
Au surface are oxidants for the epoxidation reaction. Neu-
tron experiments from C. Sivadinarayana et al. [34] provide
evidence for the formation of hydrogen peroxide H2O2 and
hydroperoxo radicals (HOO.) from the reaction between H2
and O2 over supported Au nanoclusters. In addition, a D2
kinetic isotope effect on the PO formation rate has been
observed, indicating that hydrogen is involved in the rate-
limiting step of this reaction [35].
On the other hand, for gold catalysts, beneficial effects
of H2 on the CO oxidation reaction are mostly related to
the prevention of deactivation and regeneration. Costello and
co-workers [24] have indeed demonstrated that while the ac-
tivity of Au/Al2O3 catalyst declined rapidly with time on
stream in CO oxidation, the presence of hydrogen during
selective CO oxidation successfully prevented deactivation.
They also showed that a deactivated Au/Al2O3 catalyst (after
reaction in CO oxidation) could be regenerated by exposure
to H2 or even water vapor at room temperature. These re-
sults suggest [24,25] that the active sites probably contain
hydroxyl groups, which can be removed by CO oxidation.
These hydroxyl groups would participate in the reaction,
possibly by reaction with CO, forming an active intermedi-
ate in the CO2 production pathway. Deactivation during CO
oxidation would then be due to the depletion of the hydrox-
ylated active sites when the reaction intermediate transforms
into a stable surface species. These surface species, possibly
surface carbonates, can be removed by reaction with hy-
drogen to regenerate the hydroxyl groups. However, in this
work, H2 only helps to prevent deactivation of the catalyst
and to reactivate it; no beneficial effect on the intrinsic ac-
tivity for CO oxidation has been mentioned. Choudhary and
co-workers [26] have also reported that Au/TiO2 catalysts,
which are known to generally undergo rapid deactivation
during the CO oxidation reaction, could be completely re-
generated after subsequent H2–O2 treatments. Moreover, it
has been shown very recently [27] that the addition of water
to the CO + O2 mixture accelerated the rate of oxidation, in
accordance with previous work [28].
Thus it seems obvious that, in the presence of H2, two
phenomena should occur simultaneously. One of these,
which has a negative effect on CO oxidation, would be a
competition of adsorption between H2 and CO [16,22]. The
other one, which has a beneficial effect on CO oxidation,
would be the appearance of additional reactive intermediates
produced in the presence of H2. The competition between
these two phenomena should be closely related to the partial
pressure of CO, O2, and H2, which would explain why no
general trend exists in the literature with respect to an am-
plification of or a decrease in the CO oxidation rate in the
presence of hydrogen.
Furthermore, a pronounced effect of the H2 co-reactant
was observed in kinetic measurements. At a temperature of
80 ◦C and in the pressure range (0.03 kPa < PCO < 2 kPa)
investigated by Schumacher et al. [22] during their prefer-
ential CO oxidation (PROX) study on Au/TiO2 catalysts, it
was shown that H2 affects the CO oxidation, most probably
by competing with the CO adsorption on the Au nanopar-
ticles and reacting with oxygen, which results in a signif-
icantly higher CO reaction order. Indeed, the reaction or-
der for CO increased from about 0.35 to about 0.9 in the
presence of H2. This is consistent with a reduced CO ox-
4.2. Possible mechanism for the PROX reaction
The CO interaction with gold model catalysts on various
supports has been shown to be similar [36], and therefore
support effects seen in CO oxidation must arise from the in-
teraction of oxygen rather than CO with these catalysts. In
the present work, support effects exist neither in the case of

482
C. Rossignol et al. / Journal of Catalysis 230 (2005) 476–483
Scheme 1. Possible mechanism involved in the oxidation of CO in the presence of H , including (a) adsorption of CO and H and dissociation of H on a gold
2
2
2
particle, (b) reaction of gas phase O with adsorbed H atom and (c) reaction of the resulting oxidizing species with adsorbed CO to give CO .
2
2
H2 oxidation nor in the case of CO oxidation in the presence
of H2. Hence, in the presence of H2, the O2 activation seems
to proceed via another pathway as well.
A possible mechanism for the oxidation of CO in the
presence of hydrogen must finally be able to account for the
following observations:
5. Conclusion
In summary, a strong effect of hydrogen on the CO oxida-
tion rate has been demonstrated, implying that the ability of
the catalyst to oxidize CO more readily than H2 is not a pre-
requisite for efficiency in the selective oxidation of CO in the
presence of H2. Under our PROX conditions, Au/Al2O3 and
Au/ZrO2 have indeed become as active as Au/TiO2 for the
conversion of CO. No significant effect of the support has
been detected for the oxidation of CO in the presence of hy-
drogen (or for the oxidation of H2), as opposed to what had
been reported for the oxidation of CO in CO + O2 mixtures.
Paradoxically, it seems that, to be a good candidate for the
PROX reaction, a catalyst must be active for hydrogen ox-
idation. We conclude that materials other than titania could
be interesting supports for gold in the selective oxidation of
CO in the presence of H2.
(1) the absence of a support effect on the H2 oxidation rate,
(2) the absence of a support effect on the CO oxidation rate
in the PROX reaction,
(3) a support-dependent hydrogen effect on the CO oxida-
tion rates.
Hence, although no direct evidence has been obtained,
we propose that the presence of H2 in the CO + O2 mix-
ture will not only prevent deactivation but also contribute,
via highly oxidizing intermediate species, to the oxidation
of CO, in the same way as in the propylene epoxidation re-
action. H2 oxidation, which controls the rate of formation of
these species, would then be an important step in the PROX
reaction. The fact that the gold-catalyzed H2 oxidation rate
is basically independent of the support indicates that the cat-
alytic activity for H2 oxidation is primarily related to the
exposed surface area of gold, as noted in Ref. [18]. Since it is
generally accepted that dissociation of O2 does not occur on
gold particles, an Eley–Rideal-type mechanism (Scheme 1)
could be involved in the oxidation of H2, that is, dissocia-
tion of H2 on the gold particle followed by reaction with an
O2 molecule from the gas phase. The PROX reaction would
then proceed via reaction, on the gold particle, between an
adsorbed CO molecule and an adsorbed H–O–O species (or
any HxOy species). This would explain why the CO oxida-
tion rates are similar on all supports and why no significant
support effect but rather a hydrogen effect is observed on the
CO oxidation rate for the PROX reaction under our condi-
tions.
Acknowledgment
C.R. is grateful to CNRS for a 2003 postdoctoral position.
References
[1] S. Carrettin, P. Concepción, A. Corma, J.M. López Nieto, V.F. Puntes,
Angew. Chem. Int. Ed. 43 (2004) 2538.
[2] M.M. Schubert, S. Hackenberg, A.C. Van Veen, M. Muhler, V. Plzak,
R.J. Behm, J. Catal. 197 (2001) 113.
[3] G.C. Bond, D.T. Thompson, Gold Bull. 33 (2000) 41.
[4] J.D. Grünwaldt, A. Baiker, J. Phys. Chem. B 103 (1999) 1002.
[5] M. Okumura, S. Nakamura, S. Tsubota, T. Nakamura, M. Azuma,
M. Haruta, Catal. Lett. 51 (1998) 53.
[6] H. Sakurai, A. Ueda, T. Kobayashi, M. Haruta, Chem. Commun.
(1997) 271.
[7] D. Andreeva, V. Idakiev, T. Tabakova, L. Ilieva, P. Falaras, A. Bourli-
nos, A. Travlos, Catal. Today 72 (2002) 51.
However, adsorbed H atoms could also have affected
neighbor gold atoms in such a way that they can now
chemisorb and possibly dissociate O2 molecules or directly
react at the perimeter of the Au nanoparticle with oxy-
gen bound to the support. This last mechanism has already
been invoked for CO interacting with O2 on Au/MgO and
Au/TiO2 [37,38].
[8] F. Boccuzzi, A. Chiorino, M. Manzoli, D. Andreeva, T. Tabakova,
J. Catal. 188 (1999) 176.
[9] M. Lomello-Tafin, A. Ait Chaou, F. Morfin, V. Caps, J.L. Rousset,
Chem. Comm. (2005) 388.
[10] M.M. Schubert, V. Plzak, J. Garche, R.J. Behm, Catal. Lett. 76 (2001)
143.
[11] R.J.H. Grisel, C.J. Weststrate, A. Goossens, M.W.J. Crajé, A.M. van
der Kraan, B.E. Nieuwenhuys, Catal. Today 72 (2002) 123.

C. Rossignol et al. / Journal of Catalysis 230 (2005) 476–483
483
[12] V. Arunajatesan, CatGoldNews 2 (6) (2004).
[13] G.C. Bond, D.T. Thompson, Catal. Rev. Sci. Eng. 41 (1999) 319.
[14] S. Arrii, F. Morfin, A.J. Renouprez, J.L. Rousset, J. Am. Chem.
Soc. 126 (2004) 1199.
[15] Gold reference catalysts, Gold Bull. 36 (2003) 1.
[16] R.J.H. Grisel, B.E. Nieuwenhuys, J. Catal. 199 (2001) 48.
[17] M. Haruta, S. Tsubota, T. Kobayashi, H. Kageyama, M.J. Genet,
B. Delmon, J. Catal. 144 (1993) 175.
[18] M. Okumura, S. Tsubota, M. Haruta, J. Mol. Catal. A 199 (2003) 73.
[19] R.H. Torres Sanchez, A. Ueda, K. Tanaka, M. Haruta, J. Catal. 168
(1997) 125.
[20] R. Grisel, K.-J. Weststrate, A. Gluhoi, B.E. Nieuwenhuys, Gold
Bull. 35 (2002) 39.
[26] T.V. Choudhary, C. Sivadinarayana, C. Chusuei, A.K. Datye, J.P. Fack-
ler Jr., D.W. Goodman, J. Catal. 207 (2002) 247.
[27] J.T. Calla, R.J. Davis, J. Phys. Chem. B (2005).
[28] M. Daté, M. Haruta, J. Catal. 201 (2001) 221.
[29] M. Daté, M. Okumura, S. Tsubota, M. Haruta, Angew. Chem. Int.
Ed. 43 (2004) 2129.
[30] M.A. Sanchez-Castillo, C. Couto, W.B. Kim, J.A. Dumesic, Angew.
Chem. Int. Ed. 43 (2004) 1140.
[31] T. Hayashi, K. Tanaka, M. Haruta, J. Catal. 178 (1998) 566, and refer-
ences cited therein.
[32] T.V. Choudhary, D.W. Goodman, Top. Catal. 21 (2002) 35.
[33] M. Okumura, Y. Kitagawa, K. Yamaguchi, T. Akita, S. Tsubota,
M. Haruta, Chem. Lett. 32 (2003) 822.
[34] C. Sivadinarayana, T.V. Choudhary, L.L. Daemen, J. Eckert, D.W.
Goodman, J. Am. Chem. Soc. 126 (2004) 38.
[21] B. Qiao, Y. Deng, Chem. Commun. (2003) 2192.
[22] B. Schumacher, Y. Denkwitz, V. Plzak, M. Kinne, R.J. Behm, J. Ca-
tal. 224 (2004) 449.
[35] E.E. Stangland, K.B. Stavens, R.P. Andres, W.N. Delgass, J. Catal. 191
(2000) 332.
[23] M.J. Kahlich, H.A. Gasteiger, R.J. Behm, J. Catal. 171 (1997) 93.
[24] C.K. Costello, M.C. Kung, H.-S. Oh, Y. Wang, H.H. Kung, Appl.
Catal. A 232 (2002) 159.
[25] D. Cameron, R. Holliday, D. Thompson, J. Power Sources 118 (2003)
298.
[36] K.Sh. Shaikhutdinov, R. Meyer, M. Naschitzki, M. Bäumer, H.-J. Fre-
und, Catal. Lett. 86 (2003) 211.
[37] L.M. Molina, B. Hammer, Phys. Rev. Lett. 90 (2003) 206102.
[38] L.M. Molina, M.D. Rasmussen, B. Hammer, J. Chem. Phys. 120
(2004) 7673.