The effect of reaction conditions on the stability of Au/CeZrO4 catalysts in the low-temperature water-gas shift reaction

Article abstract of DOI:10.1016/j.jcat.2010.05.021

Au/CeZrO₄ catalysts are highly active for the water-gas shift reaction but tend to be unstable and deactivate with time on stream. In this study, in situ DRIFTS-GC was used to investigate the nature of the Au under a range of reaction conditions as the deactivation rate is observed to vary with feed conditions (water concentration, presence of CO₂) and reaction temperature. An analysis of the Au-CO bands during the reaction showed that the rate of change in the Au⁰-CO bands correlated with the deactivation rate under all conditions. Although Au^δ+-CO species may be highly active they are very unstable under the feed and, from a comparison of the behaviour of the Au-CO species under varied feeds and reaction temperatures, it is proposed that metallic gold is the predominant active state in these catalysts for low-temperature WGS reaction. Through varying the feed conditions of the water-gas shift reaction, it was possible to enhance the stability of the catalysts significantly. Following a pre-treatment of the catalyst under the full water-gas shift feed (2% CO, 2% CO₂, 8.1% H₂ and 7.5% H₂O) and removal of CO₂ from this feed, the catalyst was found to be very stable. The extent of deactivation can be altered by varying the feed conditions during a reaction. The discovery of a way to enhance Au catalyst stability for the low-temperature water-gas shift reaction has significant implications for the development and use of these catalysts.

Full text of DOI:10.1016/j.jcat.2010.05.021

Journal of Catalysis 273 (2010) 257–265
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
The effect of reaction conditions on the stability of Au/CeZrO4 catalysts
in the low-temperature water–gas shift reaction
b
a
c
H. Daly a, A. Goguet a, C. Hardacre a, , F.C. Meunier , R. Pilasombat , D. Thompsett
*
a CenTACat and School of Chemistry and Chemical Engineering, Queen’s University Belfast, Belfast BT9 5AG, Northern Ireland, UK
b LCS, Univ. Caen, CNRS 6 Bd Maréchal Juin, F-14050 Caen, France
c Johnson Matthey Technology Centre, Blounts Court, Sonning Common, Reading RG4 9NH, UK
a r t i c l e i n f o
a b s t r a c t
Article history:
Au/CeZrO4 catalysts are highly active for the water–gas shift reaction but tend to be unstable and deac-
tivate with time on stream. In this study, in situ DRIFTS-GC was used to investigate the nature of the Au
under a range of reaction conditions as the deactivation rate is observed to vary with feed conditions
(water concentration, presence of CO2) and reaction temperature. An analysis of the Au–CO bands during
the reaction showed that the rate of change in the Au0–CO bands correlated with the deactivation rate
under all conditions. Although Aud+–CO species may be highly active they are very unstable under the
feed and, from a comparison of the behaviour of the Au–CO species under varied feeds and reaction tem-
peratures, it is proposed that metallic gold is the predominant active state in these catalysts for low-tem-
perature WGS reaction.
Through varying the feed conditions of the water–gas shift reaction, it was possible to enhance the sta-
bility of the catalysts significantly. Following a pre-treatment of the catalyst under the full water–gas
shift feed (2% CO, 2% CO2, 8.1% H2 and 7.5% H2O) and removal of CO2 from this feed, the catalyst was found
to be very stable. The extent of deactivation can be altered by varying the feed conditions during a reac-
tion. The discovery of a way to enhance Au catalyst stability for the low-temperature water–gas shift
reaction has significant implications for the development and use of these catalysts.
Received 9 March 2010
Revised 24 May 2010
Accepted 30 May 2010
Available online 26 June 2010
Keywords:
Water–gas shift
Gold
Deactivation
Low temperature
Feed
DRIFTS
Ó 2010 Elsevier Inc. All rights reserved.
1. Introduction
[9]. Therein, the deactivation of the catalyst was shown to be
dependent on the concentration of water in the gas phase and
Au/CeZrO4 catalysts have been reported to be highly active for
the low-temperature water–gas shift reaction which makes them
interesting candidates for a range of applications including feed
gas cleanup for low-temperature fuel cells [1]. However, the com-
mercial application of these catalysts has been hindered by their
tendency to deactivate rapidly under reaction conditions. The
long-term stability is an important issue when catalysts are to be
implemented in commercial WGS applications; therefore, an
understanding of the deactivation mechanisms is critical in design-
ing new catalysts/processes able to achieve clean hydrogen
production via the WGS reaction. The physical origin and mecha-
nism of the deactivation of supported Au catalysts are still contro-
versial with the formation of formate/carbonate species [2–6],
sintering of Au [7], loss of ceria supported surface area due to
weakening of the gold–ceria interaction [8] and loss of the contact
between the Au and the support [9] all being reported.
the reaction temperature with significant thermal deactivation
occurring at temperatures above 250 °C. Considering both experi-
mental results and theoretical calculations, sintering of the Au
and the build-up of carbon-containing material were not thought
to be relevant and a common deactivation mechanism was pro-
posed in which water and the reaction temperature caused dew-
etting of the Au nanoparticle from the support reducing the
metal support interaction which is critical for the high activity of
these catalysts.
In the present work, DRIFTS with on-line GC monitoring of
the CO conversion has been used to probe the deactivation of
Au/CeZrO4 catalysts under a range of feed conditions, pre-treat-
ments and at different reaction temperatures. DRIFT spectros-
copy has been used to assess the species adsorbed on the
catalyst surface under reaction conditions with particular focus
on the Au carbonyl region. Infrared spectroscopy has been
widely used to characterise the oxidation/coordination state of
Au through analysis of the carbonyl band position and the nat-
ure of the active Au species for the WGS reaction is still much
debated with both cationic Au [6,8,10] and metallic Au [11–13]
proposed as the active species. Goguet et al. [9] reported a
Previous work in our laboratory investigated the deactivation of
a highly active 2 wt.% Au/CeZrO4 catalyst under WGS conditions
* Corresponding author. Fax: +44 28 9097 4687.
E-mail address: c.hardacre@qub.ac.uk (C. Hardacre).
0021-9517/$ - see front matter Ó 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2010.05.021

258
H. Daly et al. / Journal of Catalysis 273 (2010) 257–265
change in the CO adsorption properties of the Au nanoparticle
with the Au carbonyl band intensity decreasing with time on
stream. The aim of this study was to examine the correlation be-
tween the deactivation rate and the Au–CO band intensity in or-
der to clarify the nature of the active Au species for the WGS
reaction.
where H2 and CO2, respectively, have been removed from the ini-
tial gas composition. It should be noted that during the reaction
H2 and CO2 will be formed and therefore the reactions are not com-
pletely in the absence of these components but conditions where
lower amounts of H2 and CO2 and H2 will be present relative to
the full WGS feed used in this study which contains 8.1% H2 and
2.5% CO2.
The effect of pre-treating the catalyst with different compo-
nents of the WGS feed was also tested. The catalyst was heated
at 2 °C minÀ1 to 100 °C under He before introducing the respective
pre-treatment gases. The temperature was further ramped to
200 °C under the pre-treatment gases where it was held for 1 h be-
fore switching to He and cooled to 100 °C (2 °C minÀ1). Once at
100 °C the reaction feed was introduced and the temperature
ramped at 1 °C minÀ1 to 150 °C. The pre-treatments tested include
the full WGS feed, 10% H2, 10% H2O, 10% CO2, 7.5% H2O + 5% CO2
and 7.5% H2O + 10% CO2. An ex situ pre-treatment with carbonic
acid was also performed to investigate whether the deactivation
could be due to the in situ reaction of water and carbon dioxide
to form carbonic acid. The 1 wt.% Au/CeZrO4 catalyst was treated
with synthesised carbonic acid at room temperature. 1g of catalyst
was soaked with 20 cm3 of deionised water to form a slurry into
which 40 cm3 minÀ1 of pure CO2 was bubbled for 30 min under
continuous stirring. The resulting catalyst was filtered and dried
under vacuum at room temperature for 6 h before undergoing
the same reaction as detailed above following the in situ pre-
treatments.
2. Experimental
2.1. Catalyst preparation
Using the deposition–precipitation method, 1 wt.% Au/CeZrO4
catalysts was prepared . The CeZrO4 (Johnson Matthey) support
was slurried in water and heated to 60 °C with the pH adjusted
to 8.0 by the addition of Na2CO3 (0.05 M). Then, 1.2 mM of HAuCl4
(49.2 wt.% Au) was prepared and pumped into the stirred slurry at
10 cm3 minÀ1. The pH was monitored throughout using a TitraLab
TIM856 pHstat (Radiometer analytical), and excess Na2CO3 was
added to maintain the pH at 8.0. After the addition of the gold salt,
stirring was continued for a further hour at 60 °C. The catalyst was
recovered by filtration, thoroughly washed to remove soluble chlo-
ride and vacuum dried at room temperature for 6 h. The average
particle size of the Au for catalysts prepared by this method was
determined to be between 0.75 and 1 nm from STEM-HAADF
images (see Supplementary Information).
2.2. Activity testing
2.3. DRIFTS measurements
The catalyst was tested in a plug flow tube microreactor held in
a temperature-controlled furnace. The catalyst bed consisted of
50 mg of catalyst between two plugs of quartz wool and a thermo-
couple measured the temperature in the middle of the catalyst bed.
Both reactants and products were analysed by GC (Perkin Elmer;
Auto System XL ARNEL) fitted with a thermal conductivity detec-
tor. Aera mass flow controllers were used to control the flow (total
flow 100 cm3 minÀ1) of CO, CO2, H2 and He to the reactor. The pur-
ity of the gases was >99.95% (supplied by BOC); dry ice/acetone
traps were used on CO gas lines remove any metal carbonyl con-
taminants that can be found in CO cylinders. A stable flow of water
vapour to the reactor was established by passing the combined
flow of CO, CO2 and H2 through a saturator/condenser system.
The saturator consisted of a 200 cm3 heated stainless steel vessel
filled with deionised water through which the permanent gases
were bubbled. The temperature of the saturator was maintained
at a slightly higher temperature than that required to give a water
vapour pressure of 7.5 kPa. The excess water vapour from the sat-
urator was removed in a glass condenser set at the correct temper-
ature to deliver a constant stream containing 7.5 kPa of water
vapour to the reactor. The combined saturator/condenser system
gave very stable water flows required for accurate analysis of the
WGS reaction. The catalyst was heated at 1 °C minÀ1 to 100 °C un-
der He and then from 100 °C to 150 °C under the WGS feed which
consisted of 2% CO, 2.5% CO2, 8.1% H2 and 7.5% H2O, balance He.
These feed conditions were also used when investigating the effect
of reaction temperature. The reaction temperature was cycled be-
tween 150 and 400 °C over a 20-h period while monitoring the CO
conversion to assess how the reaction temperature affected the
activity and stability of the catalyst.
DRIFTS with on-line GC was used to monitor the surface of the
catalyst (50 5 mg) and the activity during the WGS reaction at
150 °C. The activity profiles obtained in the plug flow reactor and
the DRIFTS-GC setup were comparable. All the activity data shown,
except for in Fig. 7, result from the operando DRIFTS-GC experi-
ments. The DRIFT spectra were recorded using a Bruker Equinox
55 spectrometer using an average of 128 scans and a resolution
of 4 cmÀ1. The DRIFTS setup consisted of an in situ high-tempera-
ture diffuse reflectance IR cell (Spectra-Tech) fitted with ZnSe win-
dows which was modified in house to behave as a plug flow
reactor, the details of which have been previously reported
[14,15]. DRIFTS-GC experiments were performed while varying
the reaction temperature and under varied feed conditions of a full
WGS feed, a low water feed, a WGS feed without CO2 and also fol-
lowing a pre-treatment under the full WGS feed.
The effect of the amount of water in the feed on catalyst stabil-
ity was investigated. The catalyst was heated at 1 °C minÀ1 under
argon to 100 °C before introduction of the WGS feed which con-
tained either 7.5% or 1.0% water; the temperature was then further
increased to 150 °C. The reaction was monitored for 20 h with
spectra recorded every 30 min.
The effect of temperature on catalyst stability was also tested
under the full WGS feed (7.5% H2O). The catalyst was heated at
1 °C minÀ1 to 100 °C under argon before introduction of the WGS
feed. The temperature was further increased under the WGS feed
to 150 °C and held at this temperature for 1 h. The temperature
was then increased to 400 °C and held at 400 °C for 1 h before
returning to 150 °C. This temperature cycle was repeated over a
24-h period with infrared spectra recorded every 15 min while
the temperature was held at 150 °C.
When varying the feed composition, reactions were performed
in a feed without H2 or CO2 and also with varying amounts of water
in the feed (1% H2O or 7.5% H2O). The total flow rate was main-
tained at 100 cm3 minÀ1 by balancing the He concentration of
the flow. These feed conditions are referred to as a ‘‘CO2-free feed”,
a ‘‘H2-free feed”, ‘‘low water feed” (1% H2O) and ‘‘full WGS feed”
(7.5% H2O) thereafter. The H2 and CO2-free feeds refer to feeds
2.4. Isosteric heat of adsorption
The isosteric heat of adsorption of CO was obtained using the
Clausius–Clapeyron equation from the infrared data obtained over
an active and deactivated Au/CeZrO4 catalyst. In these experiments,

H. Daly et al. / Journal of Catalysis 273 (2010) 257–265
259
2 wt.% Au loading was used to provide more intense Au–CO bands
which aided the analysis from increased signal to noise, in particu-
lar for the deactivated catalyst. A pellet of the 2 wt.% Au/CeZrO4 cat-
alyst (30 mg) was loaded into a variable temperature transmission
cell (Specac). The background spectrum of the fresh catalyst under
helium at 25 °C was recorded before exposing the catalyst to the full
WGS feed at 150 °C for 30 min to activate the catalyst. Following
purging of the cell with helium, adsorption of x% CO in He
(x = 0.25, 1, 4 and 8, total flow rate 200 cm3 minÀ1) was performed
at 40 °C, 60 °C and 80 °C. The same catalyst pellet was then exposed
again to the full feed at 150 °C for 12 h which resulted in deactiva-
tion of the catalyst. The 2 wt.% Au/CeZrO4 catalyst lost ꢀ30% activity
following 12 h of reaction in the microreactor. The cell was purged
again with helium and then cooled to 40 °C. Adsorption of the same
concentrations of CO was performed on the deactivated catalyst at
0 °C, 20 °C and 40 °C. CO adsorption was performed at lower tem-
peratures for the deactivated catalyst due to the weaker intensity
of the Au–CO compared with the active catalyst.
Fig. 1. Normalised CO conversion for reaction under the full WGS feed at 150 °C (2%
CO, 2% CO2, 8.1% H2 and 7.5% H2O) of 0.2 wt.%, 1 wt.% and 2 wt.% Au/CeZrO4
catalysts.
2.5. Analysis of IR spectra
All DRIFT spectra are displayed as log 1/R, where R is the relative
reflectance (R = reflectance recorded under WGS feed/reflectance
recorded under Ar). DRIFT spectra are reported as log 1/R as the rel-
ative reflectance of the Au/CeZrO4 catalyst under the WGS feed is
greater than 60% and under these conditions log 1/R is proportional
to surface concentration of adsorbates [16,17]. All spectra have
been corrected for contributions due to unreacted gas-phase CO
and the electronic transition from reduced ceria [18–20] prior to
analysis with the presence of the electronic transition confirmed
from 13CO experiments [21]. The experiments were also carried
out using a 13CO feed (2% 13CO, 2.5% CO2, 8.1% H2 and 7.5% H2O,
balance He). The shift of the Au–13CO bands to lower wavenumber
allowed assessment of the ceria electronic transition band at
ꢀ2125 cmÀ1 which along with the gas-phase CO bands were sub-
tracted from all spectra using Bruker OPUS software prior to
analysis.
Fig. 2. DRIFT spectra (2250–1950 cmÀ1 region) recorded of 2 wt.%, 1 wt.% and
0.5 wt.% Au/CeZrO4 catalysts under the full WGS feed (2% CO, 2% CO2, 8.1% H2 and
7.5% H2O) at 150 °C. All spectra have been corrected for unreacted gas-phase CO and
the electronic transition of reduced ceria.
catalysts for reaction at 150 °C were found to be 520, 270 and
180 cm3 gÀ1 hÀ1, respectively.
3.2. Infrared analysis: DRIFT spectra under WGS feed
Several overlapping features exist in the Au–CO band region
and, therefore, to analyse the rate of change of the different Au spe-
cies with time on stream, the spectra were deconvoluted by fitting
four Gaussian curves to the data and integrating the fitted Gaussi-
ans (see Supplementary Information). The same analysis was per-
formed on all spectra recorded under the series of reaction
conditions used in this study. Normalised rate of change of Au–
CO bands has been plotted to allow comparison with rate of
deactivation.
DRIFT spectroscopy has been used to probe the cause of
Au/CeZrO4 deactivation through analysis of surface species formed
on the catalyst under realistic WGS reaction conditions. Fig. 3
shows DRIFT spectra of a 1 wt.% Au/CeZrO4 upon introduction of
the WGS feed at 150 °C and after 10 h on stream. Under this feed,
bands due to bidentate formates are observed (OCO vibration: mas
at 1588 cmÀ1 and ms at 1376 cmÀ1 and C–H stretching vibration
at 2852 cmÀ1) [22–24]. In addition, a number of overlapping
vibrations due to carbonate species are also observed in the
1100–800 cmÀ1 region and ꢀ1500 cmÀ1 which are assigned to
monodentate, bidentate, bulk carbonate and carboxylate OCO
stretching vibrations [3,19,20,23,25,26].
3. Results and discussion
3.1. Full WGS feed conditions – activity
Kim et al. proposed that carbonates and/or formates block the
active sites during the water–gas shift reaction causing deactiva-
tion of Au/CeO2 catalysts at 240 °C using a feed containing 10%
CO, 22% H2O, 43% CO2 and 19% H2 [2]. Therein, sintering and
over-reduction were determined not to cause deactivation. The
role of carbonates and/or formates in the deactivation was as-
sessed through comparison of X-ray photoelectron spectra of the
C 1s region for a fresh and deactivated catalyst which showed an
increase in the peak at 288.8 eV assigned to increased carbonates.
The DRIFT spectra of the fresh catalyst exposed to CO/H2 mixture at
240 °C also showed increased intensity of bands due to formates
and carbonates. Fig. 3 showed no increase in carbonate/formate
bands during the reaction and no correlation between the rate of
formation/amount of carbonates/formates under the WGS feed
and the deactivation rate.
The stability of 2 wt.%, 1 wt.% and 0.2 wt.% Au/CeZrO4 catalysts
during reaction under the full WGS feed at 150 °C is shown in
Fig. 1. Under the full WGS feed, all catalysts exhibit similar long-
term deactivation profiles with deactivation rates of 0.23% hÀ1
.
Some differences were observed for the 2 wt.% Au catalyst in that
it exhibits a faster initial deactivation, over the first 100 min, be-
fore deactivating at a slower rate which is comparable with the
lower loaded catalysts. Fig. 2 shows the DRIFT spectra recorded un-
der the WGS feed at 150 °C as a function of Au loading. In each
case, the same Au–CO bands were observed which shows that
the loading does not alter the types of Au species formed during
the reaction but only their relative proportion. The activity (rate
of CO conversion) of the 2 wt.%, 1 wt.% and 0.5 wt.% Au/CeZrO4

260
H. Daly et al. / Journal of Catalysis 273 (2010) 257–265
Fig. 4. DRIFT spectra of 2250–1950 cmÀ1 region focussing on the Au–CO bands;
spectra shown are after 0 (bold spectrum), 1, 3, 5, 8, 10, 12 and 20 h under full WGS
feed at 150 °C. All spectra have been corrected for the electronic transition of ceria
reduced and unreacted gas-phase CO.
Fig. 3. DRIFT spectra recorded at 150 °C upon introduction of: (a) 1% H2O WGS feed
and (b) upon introduction of 7.5% H2O WGS feed (dark spectrum); light spectrum
after 10 h under 7.5% H2O WGS feed.
ported on CeO2–TiO2 [32] and on Au/ZrO2 [35] and have been de-
scribed as being due to Au+ atoms incorporated in a nanoparticle
with the charge spread over the whole particle [34].
Dentwitz et al. proposed that the formation of thermally stable
monodentate carbonates, under a range of conditions from idea-
lised to realistic reaction atmospheres at 180 °C, to be the cause
of deactivation of a Au/CeO2 catalyst [3]. Infrared bands due to
bidenatate and monodentate carbonates and bidentate formate
species were integrated and the loss/formation recorded as a func-
tion of the feed conditions. Monodentate carbonates were ob-
served to increase under all feeds with the increase observed to
vary as the deactivation rate changed under the different condi-
tions used. This is in contrast with the results obtained in this
study where no correlation between the formation of formates/car-
bonates and the deactivation rate of the catalyst was obtained. The
data reported, herein, are in agreement, however, with the previ-
ous study on Au/CeZrO4 catalysts where no increase in activity
was observed following a TPO (Temperature Programmed Oxida-
tion) to 230 °C which suggested that deactivation is not due to ad-
sorbed carbonaceous species [9]. This variation is possibly due to
the different supports used in the studies with ceria–zirconia re-
ported to be less sensitive to the build-up of carbonaceous species
compared with ceria [27].
The identification of the major Au species from DRIFT spectros-
copy as being metallic Au is consistent with previous in situ char-
acterisation and DFT results for this catalyst [11]. From the
previous results, a model of the active site was proposed in which
a hemispherical metallic gold nanoparticle ꢀ1 nm was anchored to
the support through a gold adatom occupying a cerium lattice va-
cancy in the support [36]. In this case, the gold atoms in the cerium
vacancy would have a partial positive charge (Aud+) and the high
activity of the catalyst was linked to the intimate contact of the
nanoparticle with the support.
With increasing time on stream and subsequent catalyst deacti-
vation, the DRIFT spectra showed that bands due to Au0–CO and
Aud+–CO species decreased in intensity (Fig. 4). The relative inten-
sities of these two bands were also found to change on deactivation
with the Aud+–CO band decreasing faster than Au0–CO species
(Fig. 5). However, if the Aud+–CO species were an active site for
the WGS reaction, it would be expected that the relative intensity
of the bands would change in line with the deactivation rate; this
was not observed. The observed Aud+–CO species could be CO ad-
sorbed on isolated Au atoms locked into the support in cerium lat-
tice vacancies which are able to adsorb CO and but are rapidly
reduced under the feed/lost in a fast initial deactivation process.
The region between 2200 and 1950 cmÀ1 in Fig. 3 shows strong
Au–CO bands which are observed to form immediately upon expo-
sure to the WGS feed. With CO being a reactant in the WGS reac-
tion it is an effective probe molecule for characterising the state
of the Au under reaction conditions. From Fig. 3, it is evident that
after 10 h on stream there has been a significant change in the
Au–CO bands in the 2200–1950 cmÀ1 region. Fig. 4 shows the spec-
tra in the Au–CO region and Fig. 5 shows the temporal evolution of
the Au–CO band areas associated with Au0, Aud+ and AudÀ obtained
as a function of time on stream after the introduction of the full
WGS feed at 150 °C. Upon introduction of the feed, the main fea-
ture is a band at 2098 cmÀ1 which has a shoulder at higher wave-
number at 2125 cmÀ1. The band at 2098 cmÀ1 is assigned to CO
adsorbed on metallic Au particles and is consistent with spectra re-
ported previously for Au supported on CeO2 [23,28,29]. A band at a
similar position (2110–2098 cmÀ1) has also been observed for Au
on other supports (TiO2, ZrO2, Al2O3 and Fe2O3) [30,31] and again
was assigned to CO adsorbed on metallic Au nanoparticles, specif-
ically at the step sites or the perimeter of Au nanoparticles [32,33].
The band at 2125 cmÀ1 is between the region where carbonyl
bands due to adsorption on Au+ (2200–2150 cmÀ1) and Au0 occur
for oxide supported Au (2110–2090 cmÀ1) [34]. This band is,
therefore, assigned to Au which is partially oxidised, denoted as
Aud+. Partially oxidised Au species have been observed on Au sup-
Fig. 5. Rate of change of Au–CO bands with time on stream under the full WGS feed
(2% CO, 2% CO2, 8.1% H2 and 7.5% H2O) at 150 °C. *Bands in 2070–1850 cmÀ1 region
may be due to Au reconstruction or negatively charged Au.

H. Daly et al. / Journal of Catalysis 273 (2010) 257–265
261
In contrast, the rate of change of the Au0–CO species did correlate
with the long-term deactivation rate which suggests that metallic
gold is the active species and that the perimeter sites of the nano-
particle are able to strongly adsorb CO as proposed by Goguet et al.
[9]. The identification of metallic Au as the active Au species is in
contrast with recent work of Deng et al. [37] who used in situ EX-
AFS and XANES to study low-loading Au/CeO2 catalysts (<1% Au)
under the WGS feed (5% CO + 3% H2O at 100 and 200 °C). They ob-
served reduction of Au under the WGS feed and gradual Au
agglomeration with the extent of Au reduction being related to
the reduction potential of the feed. The maximum activity was
found for highly dispersed gold with lower activity observed when
metallic Au was present.
correlation between the change in the strength of CO adsorption
and the loss of WGS activity. A 2 wt.% Au catalyst was deactivated
under the full water–gas shift feed at 150 °C for 12 h which led ca.
30% loss in activity (Fig. 1).
Clausius–Clapeyron plots of ln P(CO) as a function of 1/T for dif-
ferent surface coverages of CO allowed the heat of adsorption to be
obtained for an active and deactivated catalyst (data shown in the
Supplementary Information). The heat of adsorption increases with
decreasing CO coverage for the active catalyst and ranged from 30
to 105 kJ molÀ1 (when extrapolated to zero coverage) and for the
deactivated catalyst 10–90 kJ molÀ1. The heat of adsorption over
the deactivated catalysts was greatly reduced and suggests that
the loss of activity can be correlated with the ability of Au to ad-
sorb CO. The lower heat of adsorption following reaction is consis-
tent with the activity being linked with the CO chemisorption
ability of the Au.
The heat of adsorption of CO at low coverage on the most active
Au/CeZrO4 catalyst in this study is larger than previously reported
values which are typically between 45 and 70 kJ molÀ1, for exam-
ple values for Au(3 3 2) [33] and Au(1 0 0) [48] are 55 3 kJ molÀ1
and 58 3 kJ molÀ1, respectively. The heat of adsorption on a
reconstructed Au(1 1 0)–(1 Â 2) surface also has a comparable va-
lue of 59 kJ molÀ1 [49]. DFT calculations for energy of CO adsorp-
tion on different Au surfaces showed that CO chemisorption is
favoured on lower coordination sites with values of 32.8, 66.6,
75.3 and 105.2 kJ molÀ1 for terrace sites on Au(1 1 1), step sites
on Au(3 3 1), kink sites on Au(8 7 4) and Au adatoms on the
Au(1 1 1) surface, respectively [44]. In the case of supported Au
As Au0–CO and Aud+–CO bands decreased, a new broad Au–CO
band at a lower wavenumber of the metallic Au–CO band was ob-
served to increase in intensity (Fig. 4). This broad feature, between
2070 and 1950 cmÀ1, is located in the region where CO adsorbed
on negatively charged Au has been reported on reduced Au/CeO2
and Au/Fe2O3 catalysts [28–30,38,39], defect-rich Au/TiO2 catalysts
active for low-temperature CO oxidation [40] and Au on MgO sur-
faces [41]. The assignment of these features is still ambiguous. One
possible hypothesis is that the gold nanoparticles become mobile
on dewetting from the support surface during deactivation and ad-
sorb on oxygen vacancies [42]. In these vacancies, the nanoparti-
cles would become negatively charged through charge transfer
into the nanoparticle [43]. However,
a band in the 2070–
1950 cmÀ1 region has also been observed on a 1 wt.% Au/Al2O3 cat-
alyst following adsorption of CO at room temperature. In this case,
initial CO adsorption led to a band at 2098 cmÀ1 which was as-
signed to CO adsorbed on kink or defect sites. Following exposure
to elevated CO pressures/longer exposure times, the 2098 cmÀ1
band decreased in intensity with a new band at 2070 cmÀ1 forming
which was assigned to CO adsorbed on terrace or step sites. The
formation of the band at 2070 cmÀ1 was proposed to be due to a
change in the adsorption site on the gold nanoparticle from the
restructuring of the nanoparticle. Studies on single crystals also
illustrated a shift in the CO vibrational frequency with coordina-
tion number of the Au with adsorption of CO on step, corner or ter-
race sites exhibiting different band positions [44,45]. Restructuring
of the Au and formation of a band in the 1950–2070 cmÀ1 region
has also been observed on model Au particles supported on TiO2
where a flattening of the Au nanoparticles was observed; however,
in this case the bands were assigned to adsorption on negative Au
formed from electron transfer from defect-rich titania to the Au
particles [46].
catalysts, Roze et al. have reported heats of adsorption of 43–
À1
62 kJ mol
for 1 wt.% Au/Al2O3 for a band at 2098 cmÀ1 assigned
to CO adsorbed on defect/kink sites of Au [50] while Derrouiche et
al. reported values of 47–74 kJ molÀ1 for Au/TiO2 [51].
The higher heat of adsorption of CO for our Au catalysts com-
pared with the values reported previously for CeZrO4 supported
Au is consistent with the strong interaction of the Au nanoparticles
with the support which significantly modifies the chemisorption
properties of the metal [9]. Upon deactivation as the gold cluster
becomes detached, the adsorption characteristics of the gold will
be modified and changes occur both in terms of the type and in
terms of the number of adsorption sites on the gold particle at
positions remote from the interface. Previously, the model pro-
posed was that interfacial gold atoms (Audd+) are the sites able to
adsorb CO at elevated temperatures [9]. This interpretation was
supported by DFT calculations, which showed that CO chemisorp-
tion is enhanced on the Au site that is near the interface between
the gold and the support compared to CO chemisorption on the
Au cluster without support. An important feature of this model
was that the metal cluster is bound to the surface of the oxide
via gold atoms (Aud+) in cerium vacancies in the oxide lattice. It
is the removal of these anchor points that is responsible for the
dewetting of the gold cluster and the loss of the adsorptive proper-
ties and in fine the catalyst activity. The Audd+ sites proposed in the
previous study were assigned as the perimeter sites of the nano-
particle, which more correctly corresponds with the Au0–CO bands
observed by DRIFTS.
Clearly, in the case of the catalysts investigated in this study,
the formation of the bands between 1950 and 2070 cmÀ1 showed
a significant change in the adsorption sites of the Au under the
WGS reaction. The formation of these bands could be due to a
structural change in the nanoparticle as the catalyst deactivates
with the loss of the strong CO adsorption sites at the interface of
the nanoparticle and the support (Au0–CO bands). Irrespective of
the exact attribution of the origin of these bands, their increase
in intensity, as the catalyst deactivates, clearly indicated that this
adsorption site/state of the gold is inactive for the water–gas shift
reaction.
3.4. Effect of water concentration in feed
3.3. Infrared analysis: isosteric heat of adsorption
The DRIFT spectra under the 1% water feed exhibited the same
Au–CO bands as that observed under the feed containing the high-
er water concentration with bands due to Au0–CO at 2098 cmÀ1
and Aud+–CO at 2125 cmÀ1. Under the lower water concentration
feed, the Au–CO bands are more intense. This is consistent with
the reaction being positive order with respect to the water and
non-zero order with respect to CO under the investigated reaction
conditions explored. Such conditions lead to a higher equilibrium
The in situ DRIFT spectra shown in this study exhibit Au–CO
bands which are more intense than those observed in other studies
run at comparable temperatures [3]. Therefore, with the strength
of the CO adsorption on the Au nanoparticle being linked to the
activity of the catalyst for the water–gas shift reaction [11,47],
the isosteric heat of adsorption over an active and a deactivated
Au/CeZrO4 catalyst was probed in order to investigate the possible

262
H. Daly et al. / Journal of Catalysis 273 (2010) 257–265
concentration of CO adsorbed on the Au with decreasing water
concentration as observed.
During the reaction using the 1% water feed, the Au0–CO and
Aud+–CO species are again observed to decrease with time on
stream and the broad band between 2070 and 1950 cmÀ1 observed
to increase. Fig. 6 shows the temporal evolution of the Au–CO
bands area with time on stream for WGS reaction conditions using
1% and 7.5% water. Clearly, the rate of decrease was lower for the
conditions using 1% compared with 7.5% water corresponding with
the lower deactivation rate observed for reaction under the 1%
water feed of 0.01% hÀ1 (compared to 0.23% hÀ1 for the full WGS
feed). Under both water feeds, the Aud+–CO species decreased more
rapidly than found for the Au0–CO species. Again, the Aud+–CO spe-
cies are lost faster than the rate of deactivation while the Au0–CO
species correlates well with the loss of CO conversion.
Fig. 7. Effect on CO conversion of removing H2 and CO2 from the WGS feed at 140
and 200 °C.
3.5. Removal of H2 and CO2 from the feed – activity data
It was already reported that the deactivation rate is linearly re-
lated to the water concentration of the feed [9] but little investiga-
tion has been reported as to how the other components of the WGS
feed affected the stability [24]. Following the reaction under the
full WGS feed in the microreactor system, CO2 and then H2 were
removed from the feed to investigate whether the presence of
these components influenced the stability.
Fig. 7 shows the sequential exposure of the catalyst to the full
WGS feed, CO2-free feed and H2-free feed at 140 °C. After 20 h of
reaction under the full WGS feed, the CO2 was removed from the
feed and the helium balance increased to maintain the feed con-
centrations. The conversion was found to increase highlighting
an inhibition effect of CO2; however, in addition, under the CO2-
free feed it was observed that the catalyst showed little on-stream
deactivation. To determine whether the stability enhancement was
permanent, CO2 was re-introduced and the reaction monitored for
a further 10 h. The presence of CO2 in the feed clearly reduced the
stability of the catalyst with the deactivation rate returning to that
obtained initially under the full feed. The catalyst, upon re-intro-
duction of CO2 to the feed (after ꢀ45 h total reaction time under
a WGS feed), also exhibits higher activity than would be obtained
following 45 h of reaction under the full WGS feed. 30% CO conver-
sion was found following the CO2-free feed pre-treatment com-
pared with ꢀ20% CO conversion under the same total reaction
time under the full WGS feed (Fig. 7). The effect of H2 was then
investigated in an analogous method to that of CO2. Removal of
H2 from the feed led to an increase in the CO conversion evidencing
an inhibition effect of H2 but no enhancement in the stability of the
catalyst relative to that found under the full WGS feed was ob-
served. The same series of feed conditions was repeated at 200 °C
and again, significant enhancement of the catalyst stability was ob-
served upon removal of CO2.
With the enhanced stability observed following CO2 removal
from the feed (deactivation rate of 0.07% hÀ1), further testing was
carried out with a WGS feed without additional CO2 added (feed
containing CO + H2O + H2). Fig. 8 compares the activity of
a
1 wt.% Au/CeZrO4 under the CO2-free feed and the full WGS feed
at 150 °C. Under the CO2-free feed there was only a slight enhance-
ment in stability which could be due to the varied feed composi-
tion changing the position of the equilibrium resulting in higher
conversion of water to hydrogen and hence a lower surface water
concentration and slower hydrolysis of the Au–support interface
which has been postulated as leading to the catalyst deactivation
[9]. This small change in stability is in contrast with the significant
enhancement in stability obtained when CO2 was removed from
the feed after the catalyst had had prior exposure to the full WGS
feed. In order to probe this in more detail, an investigation into
pre-treatment of the catalyst was explored.
Fig. 6. Comparison of rate of change of Au–CO bands with time on stream under the
full WGS feed (2% CO, 2% CO2, 8.1% H2 and 7.5% H2O) and low water feed (2% CO, 2%
CO2, 8.1% H2 and 1.0% H2O) for a 1 wt.% Au/CeZrO4 catalyst.
Fig. 8. Normalised CO conversion for WGS reaction at 150 °C under CO2-free feed
(2% CO, 8.1% H2 and 7.5% H2O, balance He) and the full WGS feed (2% CO, 2% CO2,
8.1% H2 and 7.5% H2O).

H. Daly et al. / Journal of Catalysis 273 (2010) 257–265
263
A reaction was carried out where a 1 wt.% Au/CeZrO4 catalyst
was exposed to the full WGS feed at 200 °C for 1 h as a pre-treat-
ment. Following the pre-treatment, the temperature was lowered
to 100 °C under He before introduction of a CO2-free WGS feed
and the temperature increased to 150 °C. The reaction was moni-
tored with DRIFTS-GC during the pre-treatment and the reaction.
Fig. 9 shows a comparison of the activity and the Au0–CO inte-
grated band area under a CO2-free feed with and without pre-treat-
ment under the full WGS feed. It is evident that pre-treatment of
the catalyst and then reaction under a CO2-free feed does led to
an enhancement in the stability which is also reflected in the
slower rate of loss of Au0–CO bands.
The DRIFTS results showed that the rate of change of Au0–CO
species followed the trend in stability with feed conditions
(Fig. 9). No Aud+–CO bands were observed under the CO2-free feed
following the pre-treatment at 200 °C under the full WGS feed.
During the full WGS pre-treatment temperature ramp to 200 °C,
the band due to Aud+–CO species was depleted and was no longer
evident at 200 °C under the feed. In fact, at 200 °C, under all feeds,
only Au0–CO bands were observed. As demonstrated already, the
Aud+–CO species are significantly less stable than Au0–CO species
and while Aud+ species may have high activity for the WGS reaction
they are not observed after pre-treatment of the catalyst when the
catalyst is still very active and stable. Only Au0–CO bands were ob-
served under the CO2-free feed (after pre-treatment) and their rate
of change corresponded with the deactivation rate as observed un-
der other feed conditions tested.
Fig. 10. Reaction of a 1 wt.% Au/CeZrO4 catalyst under a CO2-free WGS feed (2% CO,
8.1% H2 and 7.5% H2O) at 200 °C following different pre-treatments at 200 °C. The
effect of pre-treating the catalyst with the full WGS feed (2% CO, 2% CO2, 8.1% H2
and 7.5% H2O), 10% CO2, 10% H2, 10% H2O, 5% CO2 + 7.5% H2O and carbonic acid.
between carbonates/formates surface concentration and the ob-
served deactivation rate. In addition, surprisingly, running the
reaction initially without CO2 in the feed did not lead to the same
enhanced stability. Therefore, it is possible that CO2 is required ini-
tially to form the active catalyst but that prolonged exposure to
elevated amounts of CO2 in the presence of water influences the
structure of the Au nanoparticle and enhances the dewetting of
the nanoparticle from the support.
Alternative pre-treatments were also explored to determine
whether further enhancement could be obtained. Fig. 10 shows
the activity data obtained for reaction at 150 °C under a CO2-free
feed following pre-treatment with water, hydrogen, CO2, water
plus CO2 and carbonic acid. Carbonic acid was examined to test
the postulation that the effect of CO2 in the feed could be related
to its interaction with water to form carbonic acid. Pre-treatment
at 200 °C with hydrogen, water or CO2 did not have any effect on
the stability with a deactivation rate comparable to that observed
in reaction under the full WGS feed. Pre-treatment with a simulta-
neous feed of water and CO2 caused faster deactivation than water
or carbon dioxide alone. The stability also varied with the propor-
tion of CO2 in the water mixture with faster deactivation being ob-
served when the catalyst was pre-treated with 10% CO2 plus 7.5%
water compared with a feed containing 5% CO2 plus 7.5% water.
The deactivation rate was faster again following carbonic acid
treatment.
3.6. Effect of reaction temperature
Previously it was reported that elevated reaction temperatures
(>250 °C) led to rapid and irreversible deactivation [9]. In order to
investigate the impact of thermal deactivation on the various Au–
CO species, the reaction temperature was cycled between 150 °C
and 400 °C while recording DRIFT spectra during the reaction cy-
cles at 150 °C. Fig. 11 shows the normalised CO conversion, Aud+
–
CO and Au0–CO band areas as a function of time following cycles
between 150 and 400 °C. Following the first temperature cycle to
400 °C, a significant loss in activity of ꢀ50% was observed. After
this initial rapid loss of activity, the Aud+–CO band was no longer
observed which was consistent with loss of Aud+–CO following
the pre-treatment at 200 °C, as discussed above. With the loss of
activity, there was also a corresponding decrease in the Au0–CO
band at 2098 cmÀ1. With subsequent cycles at 400 °C, further de-
creases in conversion were observed with concomitant decreases
in the intensity of the Au0–CO band. These results showed once
again that the loss of the metallic Au adsorption sites can be corre-
lated with the long-term deactivation of the catalyst which sug-
gests a common deactivation mechanism. Further evidence for
the rate of reaction correlating with the Au0–CO band area is
shown in the inset of Fig. 11 where a plot of rate of reaction/
Au0–CO band area gives a constant value (within experimental
error). This shows that as the reaction progresses and the catalyst
deactivates, the loss of band area of Au0–CO species correlates with
the deactivation rate and further suggests a change in the CO
chemisorption ability of the Au to be related to the catalyst
deactivation.
Although the variation in the CO2 concentration of the feed has
been previously reported to correlate with deactivation [3], in this
study, removal of CO2 from the feed enhances stability. Under
these varied feed conditions there was again no correlation
It is interesting to note that following the rapid thermal deacti-
vation, no bands in the 2070–1950 cmÀ1 region were formed. If
these species were due to adsorption of CO on negatively charged
Au formed from the nanoparticle adsorption on an oxygen vacancy,
increased temperature would be expected to increase the number
of oxygen vacancies. Consequently, more negatively charged Au
would be expected to form as the catalyst deactivates; this is obvi-
ously not the case. As reported previously [9], deactivation at low
temperatures is likely due to hydrolysis of the gold nanoparticle/
support interaction causing a change in shape/structure of the
Fig. 9. Normalised CO conversion and Au0–CO band area during reactions at 150 °C
under a CO2-free WGS feed (2% CO, 8.1% H2 and 7.5% H2O) with and without pre-
treatment with the full WGS feed (2% CO, 2% CO2, 8.1% H2 and 7.5% H2O).

264
H. Daly et al. / Journal of Catalysis 273 (2010) 257–265
0.0002
0.0001
0.0000
Fig. 11. Comparison of catalyst activity and rate of change of Au–CO species at 150 °C following hour intervals under the full WGS feed (2% CO, 2% CO2, 8.1% H2 and 7.5% H2O)
at 400 °C. The inset shows a plot of the rate of reaction (molCO minÀ1 gcatÀ1)/Au0–CO band area with reaction time.
nanoparticle with the gold nanoparticle reconstructing and pro-
gressively losing the intimate contact with the support [9]. This
is supported by the formation of the Au–CO bands in the 2070–
1950 cmÀ1 region and these bands being attributed to particle
reconstruction similar to that described in Roze et al. [50]. How-
ever, with the thermal deactivation, a more extreme change in
the structure of the nanoparticle could be responsible for the great-
er and more rapid loss of activity and the corresponding different
CO adsorption behaviour compared to reaction at lower
temperatures.
with the changing CO sorption ability of the Au under the different
reaction conditions as observed by DRIFTS.
The activity and stability of these Au/CeZrO4 catalysts are very
sensitive to the feed composition. An initial period of reaction un-
der the full WGS feed is required before removing the CO2 from the
feed in order to obtain enhanced stability of these catalysts. There-
fore, it is proposed that the active Au sites are formed under the
full WGS feed wherein the Au nanoparticles are locked into the
support. Once the nanoparticles have formed, the feed composition
and reaction temperature will determine the rate at which the
nanoparticles become detached from the surface, with contact
with water leading to hydrolysis of the nanoparticle/support inter-
action. During reaction under the CO2-free feed, the surface con-
centration of water will be reduced (equilibrium shifting towards
hydrogen formation) which is one factor which could account for
the slower deactivation when CO2 is removed from the feed. How-
ever, faster deactivation is observed when the catalyst is pre-trea-
ted with CO2 + H2O or carbonic acid rather than with water alone.
This enhanced deactivation from the combined effect of CO2 and
water may be the formation of species such as H3O+/HCO3À which
may promote the dewetting of the nanoparticle through acid/
base-mediated hydrolysis. Therefore, removal of the CO2 once the
active state of the catalyst has been achieved greatly improves
the stability.
4. Conclusions
The adsorption of CO as a probe of the state of the Au is a widely
used technique which was well suited for this study because of the
strong CO chemisorption ability of these Au catalysts. The high
heats of adsorption (90–105 kJ molÀ1) is consistent with the in-
tense Au–CO bands observed at the reaction temperature of
150 °C and which allowed investigation into the Au species under
relevant conditions and monitoring of the change of these species
with time on stream. This DRIFTS-GC study under varying feed
conditions/reaction temperatures has shown metallic Au to be
the important species for the WGS reaction. Aud+ species were
shown to be less stable under the feed and were no longer ob-
served at reaction temperatures above 200 °C when the catalyst
was still active. It is unlikely that a very low number (below detec-
tion limit of DRIFTS technique) of extremely active sites could be
responsible for the high WGS activity observed. This is because
the rate of change of positively charged species did not match
the long-term deactivation rate while the change of Au0–CO spe-
cies did correlate well under all feed/reaction conditions explored.
It is proposed that formation of the AudÀ species are in fact due
to a change in the adsorption site due to the restructuring of the Au
under the WGS feed during the progressive dewetting of the gold
nanoparticle which is the deactivation mechanism proposed for
these catalysts. In this work, negatively charged Au species were
not observed following the high-temperature WGS reaction when
more oxygen vacancies would be expected. This result suggests
that the bands in the 2070–1950 cmÀ1 region may be related to a
restructuring of the Au rather than CO adsorption on negatively
charge gold species.
The use of these catalysts with membrane reactors where CO2
could be removed from the feed to provide a highly active and sta-
ble catalyst provides great potential for the long-term use of Au
catalysts in the production of hydrogen.
Acknowledgment
We would like to thank the EPSRC for funding as part of the
CARMAC and CASTech projects.
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.jcat.2010.05.021.
References
A reversible effect on the stability was observed upon removal
and re-introduction of CO2 into the feed. The thermal deactivation
was irreversible [9,11] and significantly faster than the deactiva-
tion observed on 1 wt.% Au/CeZrO4 under varied feed conditions.
Hence, while there may be a common deactivation mechanism,
the rate at which the structural change occurs may be correlated
[1] A.A. Fonseca, J.M. Fisher, D. Ozkaya, M.D. Shannon, D. Thompsett, Top. Catal. 44
(2007) 223.
[2] C.H. Kim, L.T. Thompson, J. Catal. 230 (2005) 66.
[3] Y. Denkwitz, A. Karpenko, V. Plzak, R. Leppelt, B. Schumacher, R.J. Behm, J.
Catal. 246 (2007) 74.
[4] M.M. Schubert, V. Plzak, J. Garche, R.J. Behm, Catal. Lett. 76 (2001) 143.

H. Daly et al. / Journal of Catalysis 273 (2010) 257–265
265
[5] M.M. Schubert, A. Venugopal, M.J. Kahlich, V. Plzak, R.J. Behm, J. Catal. 222
(2004) 32.
[6] W. Deng, J. De Jesus, H. Saltsburg, M. Flytzani-Stephanopoulos, Appl. Catal. A
291 (2005) 126.
[7] A. Luengnaruemitchai, S. Osuwan, E. Gulari, Catal. Commun. 4 (2003) 215.
[8] Q. Fu, W. Deng, H. Saltsburg, M. Flytzani-Stephanopoulos, Appl. Catal. B 56
(2005) 57.
[28] T. Tabakova, F. Boccuzzi, M. Manzoli, D. Andreeva, Appl. Catal. A 252 (2003) 385.
[29] M. Manzoli, F. Boccuzzi, A. Chiorino, F. Vindigni, W. Deng, M. Flytzani-
Stephanopoulos, J. Catal. 245 (2007) 308.
[30] F. Menegazzo, M. Manzoli, A. Chiorino, F. Boccuzzi, T. Tabakova, M. Signoretto,
F. Pinna, N. Pernicone, J. Catal. 237 (2006) 431.
[31] Y.Z. Chen, B.J. Liaw, W.C. Chang, C.T. Huang, Int. J. Hydrogen Energy 32 (2007)
4550.
[9] A. Goguet, R. Burch, Y. Chen, C. Hardacre, P. Hu, R.W. Joyner, F.C. Meunier, B.S.
Mun, D. Thompsett, D. Tibiletti, J. Phys. Chem. C 111 (2007) 16927.
[10] Q. Fu, H. Saltsburg, M. Flytzani-Stephanopoulos, Science 301 (2003) 935.
[11] D. Tibiletti, A.A. Fonseca, R. Burch, Y. Chen, J.M. Fisher, A. Goguet, C. Hardacre,
P. Hu, D. Thompsett, J. Phys. Chem. B 109 (2005) 22553.
[12] C.H. Kim, L.T. Thompson, J. Catal. 244 (2006) 248.
[13] J. Rodriguez, X. Wang, P. Liu, W. Wen, J. Hanson, J. Hrbek, M. Pérez, J. Evans,
Top. Catal. 44 (2007) 73.
[14] F.C. Meunier, D. Reid, A. Goguet, S. Shekhtman, C. Hardacre, R. Burch, W. Deng,
M. Flytzani-Stephanopoulos, J. Catal. 247 (2007) 277.
[15] F.C. Meunier, A. Goguet, S. Shekhtman, D. Rooney, H. Daly, Appl. Catal. A 340
(2008) 196.
[16] J. Sirita, S. Phanichphant, F.C. Meunier, Anal. Chem. 79 (2007) 3912.
[17] J. Couble, P. Gravejat, F. Gaillard, D. Bianchi, Appl. Catal. A 371 (2009) 99.
[18] C. Binet, A. Badri, J.C. Lavalley, J. Phys. Chem. 98 (1994) 6392.
[19] C. Binet, M. Daturi, J.C. Lavalley, Catal. Today 50 (1999) 207.
[20] F. Bozon-Verduraz, A. Bensalem, J. Chem. Soc. Faraday Trans. 90 (1994) 653.
[21] H. Daly, J. Ni, D. Thompsett, F.C. Meunier, J. Catal. 254 (2008) 238.
[22] G. Busca, J. Lamotte, J.C. Lavalley, V. Lorenzelli, J. Am. Chem. Soc. 109 (1987)
5197.
[32] A. Chiorino, M. Manzoli, F. Menegazzo, M. Signoretto, F. Vindigni, F. Pinna, F.
Boccuzzi, J. Catal. 262 (2009) 169.
[33] C. Ruggiero, P. Hollins, Surf. Sci. 377–379 (1997) 583.
[34] M. Mihaylov, H. Knözinger, K. Hadjiivanov, B.C. Gates, Chem. Ing. Tech. 79
(2007) 795.
[35] F. Boccuzzi, G. Cerrato, F. Pinna, G. Strukul, J. Phys. Chem. B 102 (1998) 5733.
[36] Y. Chen, P. Hu, M.H. Lee, H. Wang, Surf. Sci. 602 (2008) 1736.
[37] W. Deng, A.I. Frenkel, R. Si, M. Flytzani-Stephanopoulos, J. Phys. Chem. C 112
(2008) 12834.
[38] F. Boccuzzi, A. Chiorino, M. Manzoli, D. Andreeva, T. Tabakova, J. Catal. 188
(1999) 176.
[39] M. Date, H. Imai, S. Tsubota, M. Haruta, Catal. Today 122 (2007) 222.
[40] T. Diemant, Z. Zhao, H. Rauscher, J. Bansmanna, R.J. Behm, Top. Catal. 44 (2007)
83.
[41] B. Yoon, H. Hakkinen, U. Landman, A.S. Worz, J.M. Antonietti, S. Abbet, K. Judai,
U. Heiz, Science 307 (2005) 403.
[42] H.L. Chen, W.T. Peng, J.J. Ho, H.M. Hsieh, Chem. Phys. 348 (2008) 161.
[43] M. Sterrer, T. Risse, H.J. Freund, Appl. Catal. A 307 (2006) 58.
[44] L. Piccolo, D. Loffreda, F.J. Cadete Santos Aires, C. Deranlot, Y. Jugnet, P. Sautet,
J.C. Bertolini, Surf. Sci. 566–568 (2004) 995.
[23] A. Abd El-Moemen, A. Karpenko, Y. Denkwitz, R.J. Behm, J. Power Sources 190
(2009) 64.
[24] A. Abd El-Moemen, G. Kucerová, R.J. Behm, Appl. Catal. B 95 (2010) 57.
[25] C. Li, Y. Sakata, T. Arai, K. Domen, K. Maruya, T. Onishi, J. Chem. Soc. Faraday
Trans. 85 (1989) 929.
[26] B. Aeijelts Averink Silberova, G. Mul, M. Makkee, J.A. Moulijn, J. Catal. 243
(2006) 171.
[27] C.H. Kim, L.T. Thompson at GOLD2006: New Industrial Applications for Gold,
Limerick, Ireland, 2006.
[45] P. Hollins, Surf. Sci. Rep. 16 (1992) 53.
[46] T. Diemant, H. Hartmann, J. Bansmann, R.J. Behm, J. Catal. 252 (2007) 171.
[47] D.C. Grenoble, M.M. Estadt, D.F. Ollis, J. Catal. 67 (1981) 90.
[48] G. Mcelhiney, J. Pritchard, Surf. Sci. 60 (1976) 397.
[49] J.M. Gottfried, K.J. Schmidt, S.L.M. Schroeder, K. Christmann, Surf. Sci. 536
(2003) 206.
[50] E. Roze, P. Gravejat, E. Quinet, J.L. Rousset, D. Bianchi, J. Phys. Chem. C 113
(2009) 1037.
[51] S. Derrouiche, P. Gravejat, D. Bianchi, J. Am. Chem. Soc. 126 (2004) 13010.