The role of Pd-Ga bimetallic particles in the bifunctional mechanism of selective methanol synthesis via CO2 hydrogenation on a Pd/Ga 2O3 catalyst

Article abstract of DOI:10.1016/j.jcat.2012.05.005

The effect of palladium-gallia interaction in Pd(1 wt.%)/β-Ga ₂O₃ during selective methanol synthesis by CO₂ hydrogenation was studied. A detailed quasi-in situ transmission electron microscopy analysis of the as-prepared H₂-reduced catalyst, without exposing it to air, showed that Ga-Pd bimetallic (nano)particles were formed under a reductive atmosphere at or above 523 K. However, these particles were unstable; upon air exposure, a dramatic and extensive encapsulation of the metallic crystallites by Ga₂O₃ occurred. In addition, the function of the bimetallic particles in the mechanism of methanol synthesis was investigated by in situ infrared spectroscopy at 0.7 MPa. The results confirmed those of previous studies in which the stepwise hydrogenation of (bi)carbonate to formate and then to methoxy groups on the Ga₂O₃ surface took place via a bifunctional pathway. In this pathway, the role of the Ga-Pd bimetallic crystallites was to provide atomic hydrogen, via spillover, to the oxidic surface and to hamper both CH₃OH decomposition and CO production.

Full text of DOI:10.1016/j.jcat.2012.05.005

Journal of Catalysis 292 (2012) 90–98
Contents lists available at SciVerse ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
The role of Pd–Ga bimetallic particles in the bifunctional mechanism
of selective methanol synthesis via CO2 hydrogenation on a Pd/Ga2O3 catalyst
b
b
b
b
Sebastián E. Collins a, , Juan J. Delgado , César Mira , José J. Calvino , Serafín Bernal ,
⇑
Dante L. Chiavassa a, Miguel A. Baltanás a, Adrian L. Bonivardi a
a Instituto de Desarrollo Tecnológico para la Industria Química, Universidad Nacional del Litoral and CONICET, Güemes 3450, 3000 Santa Fe, Argentina
b Dpto. de Ciencias de los Materiales, Ingeniería Metalúrgica y Quimica Inorgánica, Facultad de Ciencias, Universidad de Cádiz, Puerto Real E11510, Spain
a r t i c l e i n f o
a b s t r a c t
Article history:
The effect of palladium–gallia interaction in Pd(1 wt.%)/b-Ga2O3 during selective methanol synthesis by
CO2 hydrogenation was studied. A detailed quasi-in situ transmission electron microscopy analysis of the
as-prepared H2-reduced catalyst, without exposing it to air, showed that Ga–Pd bimetallic (nano)parti-
cles were formed under a reductive atmosphere at or above 523 K. However, these particles were unsta-
ble; upon air exposure, a dramatic and extensive encapsulation of the metallic crystallites by Ga2O3
occurred. In addition, the function of the bimetallic particles in the mechanism of methanol synthesis
was investigated by in situ infrared spectroscopy at 0.7 MPa. The results confirmed those of previous
studies in which the stepwise hydrogenation of (bi)carbonate to formate and then to methoxy groups
on the Ga2O3 surface took place via a bifunctional pathway. In this pathway, the role of the Ga–Pd bime-
tallic crystallites was to provide atomic hydrogen, via spillover, to the oxidic surface and to hamper both
CH3OH decomposition and CO production.
Received 24 April 2012
Accepted 2 May 2012
Available online 15 June 2012
Keywords:
Palladium
Gallium
Bimetallic
Methanol synthesis
HRTEM
Infrared spectroscopy
Ó 2012 Elsevier Inc. All rights reserved.
1. Introduction
hydrogenation of carbon dioxide at conventional process condi-
tions up to 500-fold (523 K, 3 MPa, H2/CO2 = 3) compared to Pd/
Methanol synthesis from CO2/H2 is a viable alternative for
reducing greenhouse gas net emissions. In this context, supported
precious metals, such as palladium, have been of particular interest
over the past decade [1], as it is well known that both the support
and the (eventual use of a) promoter can appreciably modify the
activity and/or selectivity of this metal in methanol synthesis [2–4].
The development of an active Pd/Ga2O3 catalyst for selective
methanol synthesis from CO2 hydrogenation able to compete with
the classical Cu/ZnO formulation was first reported by Fujitani et al.
[4]. Using a stoichiometric H2/CO2 mixture (523 K, 5 MPa), their Pd/
Ga2O3 preparations were 120-fold more active than Pd on SiO2 (an
inert support) and sevenfold more active than Cu/ZnO/Al2O3 [4,5]. It
was suggested that the other reaction product, CO, was obtained
(with selectivity close to 49%) via the reverse water–gas shift
reaction (RWGS) over the metallic particles.
SiO2. At the same time, methanol selectivity was enhanced up to
70%, compared to a meager 10% in the unpromoted catalyst.
Later, detailed studies on the chemical structure and reaction
mechanisms were undertaken. The interaction of H2 with Ga2O3/
SiO2, Pd/SiO2, Ga2O3–Pd/SiO2, and pure Ga2O3 polymorphs with
different crystallographic phases (a,b,c) was investigated by
in situ X-ray photoelectron and Fourier transform infrared spec-
troscopy (XPS and FTIR, respectively) [7,8]. The adsorption of CO2
over this set of gallium oxide polymorphs was also examined via
infrared spectroscopy. Two types of bicarbonate species (mono-
and bidentate), carboxylate, bridge carbonate, bidentate carbonate,
and polydentate carbonate species were identified [9]. Interest-
ingly, most of these carbonates were reversibly adsorbed on the
gallia surface.
A detailed study of CO2 hydrogenation to methanol, as well as of
methanol decomposition, on model Pd/gallia catalysts was like-
wise conducted, mainly by means of in situ FTIR [10,11]. It was
shown that the reaction intermediates, chemisorbed onto gallia,
were successively hydrogenated from (bi)carbonates to formate
(mono- and bidentate and bridged), methylenebisoxy, and meth-
oxy to give methanol, while hydrogen supply was accomplished
via spillover from the palladium metal particles. The same interme-
diates were identified during methanol decomposition (i.e., the re-
verse pathway) on pure Ga2O3 polymorphs and Pd/Ga2O3 catalysts
A sustained research effort has been conducted by some of us
during the past decade to understand the origin of the remarkable
performance of the palladium–gallia system for the selective
hydrogenation of CO2 to methanol. First, Bonivardi et al. [6]
showed that gallium oxide addition (as a promoter) to a Pd
(2 wt.%)/SiO2 catalyst increased the CH3OH production rate in the
⇑
Corresponding author. Fax: +54 342 4550944.
E-mail address: scollins@santafe-conicet.gov.ar (S.E. Collins).
0021-9517/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.jcat.2012.05.005

S.E. Collins et al. / Journal of Catalysis 292 (2012) 90–98
91
[11,12]. Further work using physical mixtures of Ga2O3/SiO2 and
Pd/SiO2 showed that atomic hydrogen, HS, was generated on the
silica-supported Pd particles and moved (‘‘spilled over’’) to the sup-
ported gallia particles, where the carbonaceous species were then
hydrogenated [13]. These reaction results indicated that (as was
also evidenced by in situ FTIR) the Ga2O3–Pd/SiO2 catalyst worked
as a true bifunctional system, while Pd/SiO2 did not.
The collected spectroscopic information was successfully used
to develop a comprehensive microkinetic model that was able to
account for the reaction performance of the gallia-promoted Pd
catalyst for the selective synthesis of methanol from CO2/H2 (to-
gether with the undesirable RWGS reaction) under a wide range
of operating conditions [14].
the reactor by flowing H2 (heating rate 3 K minÀ1, P = 0.1 MPa)
from RT to 523 K and maintaining the hydrogen flow for 1 h. Gas
chromatography was used to analyze the gas composition in two
Shimadzu 9A units arranged in parallel that employed Porapak
QS and Carbosieve SII-filled (1/8 in. ID, 3 m long) stainless steel col-
umns with TCD and FID, respectively.
After the reaction test, the reactor was depressurized, the feed
of reactants was replaced by pure H2 (60 cm3 minÀ1, at 523 K),
and the unit was allowed to cool to room temperature under this
flow. Then the reactor was swept with pure N2 (99.999%,
60 cm3 minÀ1), and oxygen pulses were gradually introduced into
the gas stream to passivate the catalyst before its exposure to air
and final storage in a desiccator. This sample is referred to hereaf-
ter as ‘‘postreaction Pd/Ga2O3.’’
Notwithstanding, some authors have postulated that Pd–Ga al-
loys may play a relevant role in the catalytic activity of the palla-
dium–gallia systems. Iwasa and coworkers [15,16], for example,
were the first to suggest that the remarkable activity and selectiv-
ity achieved by Pd/Ga2O3 catalysts, either in methanol synthesis or
in the steam reformation of methanol, was a consequence of the
new metallic catalytic function performed by a Pd–Ga alloy. Later,
a series of pure Pd–Ga intermetallic compounds showing high
activity and selectivity in the hydrogenation of acetylene in an eth-
ylene-rich stream were prepared by Schlögl and coworkers [17–
19]. Very recently, Pd/Ga2O3 catalysts have been investigated in
the methanol–steam reformation reaction using different oxidative
and reductive pretreatments [20–22]. The authors reported the
formation of Pd–Ga bimetallic particles during reductive pretreat-
ment above 523 K. In agreement with the work of Iwasa and
coworkers [15,16], the Pd–Ga bimetallic particles were claimed
to be the sites responsible for the increased selectivity to CO2 over
methanol decomposition to CO and H2. Likewise, the possibility of
the formation of a Pd–Ga alloy on Pd/Ga2O3 catalysts under meth-
anol synthesis conditions was suggested a few years ago [23], but
its relevance in the catalytic performance of this system in the CO2
hydrogenation remained unclear. Therefore, we present here an
investigation of a Pd/b-Ga2O3 model catalyst to gain further insight
into the role(s) of any possible alloy formation and its impact on
the catalytic performance of this material in H2/CO2 mixtures.
Reaction performance, in situ characterization, and high-pressure
IR spectroscopy were combined to compare the possible (or even-
tual) influence of the bimetallic Pd–Ga in the bifunctional mecha-
nism of methanol synthesis from CO2 hydrogenation.
2.3. High-resolution electron microscopy
High-resolution transmission (HRTEM) and high-angle annular
dark-field scanning–transmission (HAADF-STEM) micrographs
were obtained using a JEOL-2010F microscope equipped with a
field emission gun and operated at an acceleration voltage of
200 kV. The structural resolution of the equipment in the HRTEM
mode was 0.19 nm at the Scherzer defocus conditions, while the
diameter of the probe used in STEM was 0.5 nm. The intensity of
the HAADF-STEM images is proportional (sensitive) to the square
of the atomic number. Due to this proportionality, we can unam-
biguously discriminate between Pd (Z = 46) and the cation of the
support, Ga (Z = 31), for accurate determination of the average size
distribution of the metal particles.
Taking advantage of the STEM mode, elemental analysis studies
were performed at the nanometric scale by X-ray energy-dispersive
spectroscopy (X-EDS), using an Oxford INCA Energy 2000 system
coupled to the equipment. The chemical composition of the sup-
ported nanoparticles was studied using both spot and line spectrum
measurements. The latter approach was useful for determining the
distribution of Pd and Ga. The sample was mounted on a copper grid
by a dry method to prevent the sample from coming into contact
with any solvent.
2.4. Quasi-in situ transmission electron microscopy
A homemade quasi-in situ TEM system was used to evaluate the
formation of bimetallic nanoparticles in the catalyst without expo-
sure to the atmosphere. This setup featured a VTST4006 (GATAN
Inc.) high-resolution vacuum transfer holder. The tip of this holder
can be placed on a special quartz reactor to conduct thermochem-
ical treatments, after which the sample is introduced into a shuttle
chamber, under high vacuum, to place it inside the microscope
under anaerobic conditions. This approach allows observation of
samples without any further modification or degradation. The
samples were deposited onto a lacey carbon-coated molybdenum
grid, which was previously pretreated at 523 K under a H2 atmo-
sphere to prevent any contamination during the experiments.
2. Experimental
2.1. Catalyst
Pd(1 wt.%)/b-Ga2O3 was prepared by incipient wetness impreg-
nation using Pd(OAc)2 (Sigma) in acetone, with further drying/cal-
cination in air (673 K, 2 h), H2 reduction (2% H2/Ar, 723 K, 2 h), and
passivation with oxygen pulses at room temperature. This material
is named ‘‘as-prepared’’ or ‘‘prereaction’’ Pd/Ga2O3. The monoclinic
b-Ga2O3 was obtained by calcining gallia (Strem, 99.998%) at
1073 K for 5 h. CO pulses were used to determine the palladium
dispersion (DCO = 31%) on the as-prepared catalyst, which was re-
reduced at 523 K (pure H2, 1 h) using a conventional technique
[24]. The BET (N2) areas of the support and the catalyst were both
12 m2 gÀ1. Pd (1 wt.%)/SiO2 was prepared and characterized as de-
scribed previously [6].
2.5. Infrared studies
In situ transmission infrared spectroscopy was performed using
ꢀ23 mg cmÀ2 samples pressed into self-supported wafers at
5 ton cmÀ2. The wafers were located in a heated Pyrex cell with
water-cooled NaCl windows, which was attached to a conventional
high-vacuum system (base pressure = 1.33 Â 10À4 Pa) and a mani-
fold for gas flow. Infrared transmission spectra were acquired with
a Nicolet Magna 550 Series II FTIR spectrometer using an MCT-A
detector (resolution 4 cmÀ1, 100 scans). Further processing of the
spectra was carried out with Microcal Origin 4.1 software.
2.2. Catalytic activity tests
The catalytic performance was evaluated in a differential plug-
flow reactor [H2/CO2 = 3, 523 K, 0.7 MPa (SV = 7700 hÀ1) and 3 MPa
(SV = 80,000 hÀ1)] over 24 h. Each catalyst was re-reduced inside

92
S.E. Collins et al. / Journal of Catalysis 292 (2012) 90–98
Background correction of the spectra was achieved by subtracting
the spectra of the clean wafers after the activation pretreatment.
In situ reaction experiments, at high and low pressure, were
performed in a homemade stainless steel transmission cell fitted
with CaF2 windows. The temperature was controlled by two ther-
mocouples, one in the heating block and the other inserted inside
the cell space and in contact with the catalyst wafer. High-pressure
experiments were carried out by submitting the catalyst wafer to
in situ reduction in pure H2 from RT to 523 K (0.1 MPa, 5 K minÀ1
,
100 cm3 minÀ1). Then the flow was changed to a stoichiometric H2/
CO2 = 1/3 mixture, and the cell was pressurized to 0.7 MPa using a
backpressure regulator. Transmission spectra were acquired and
processed as described before.
3. Results
3.1. Characterization and catalytic activity
Table 1 summarizes the results of the morphological character-
ization, activity, and selectivity at steady state of the Pd(1 wt.%)/
b-Ga2O3 catalyst activated under pure H2 at 523 K for 60 min
(hereafter as-prepared + R523/1 h). The results of the hydrogena-
tion of CO2 (H2/CO2 = 3, 523 K) over 24 h on stream, both at 3 MPa
and at 0.7 MPa, are presented to compare the latter with in situ FTIR
experiments. For comparison purposes, a Pd(2 wt.%)/SiO2 reference
catalyst subjected to the same pretreatment protocol is also in-
cluded. Pure gallia was unreactive under these reaction conditions.
The specific rate of CO2 hydrogenation to methanol was approx-
imately 20-fold higher on the Pd/b-Ga2O3 catalyst than on the ref-
erence Pd/SiO2 catalyst, or approximately 40-fold higher per metal
site, based on their respective initial Pd dispersion. In addition, the
former was fourfold more selective to methanol. The temporal evo-
lution of the production rates of methanol and CO during 24 h of
reaction on both catalysts is shown in Fig. 1. As observed, the activ-
ity at the beginning of the reaction over Pd/Ga2O3 was much higher
toward the production of CO, but then decreased significantly.
Meanwhile, the reverse was observed for the activity toward meth-
anol. After approximately 3 h, the system reached a pseudo-steady
state, with similar activity (and selectivity) for the production of
methanol and CO. On the reference Pd/SiO2 catalyst, only a moder-
ate decrease in the rates to methanol and CO during the first hour
of reaction was observed. On this catalyst, the selectivity to CO was
also consistently and noticeably higher.
Fig. 1. Time evolution of the production rates of methanol and CO over the
Pd(1 wt.%)/b-Ga2O3 catalyst as-prepared + R523/1 h and the reference Pd(2 wt.%)/
SiO2 catalyst (H2/CO2 = 3, P = 3 MPa, T = 523 K).
3.2. Nanostructural and nanoanalytical analyses by transmission
electron microscopy
Fig. 2a and b shows a representative ex situ HRTEM image of the
as-prepared Pd/Ga2O3 catalyst and its Pd particle size histogram,
respectively. The structural parameters measured on the support
indicated that the interplanar spaces and angles correspond to the
monoclinic structure of b-Ga2O3. The Pd crystallites were uniformly
distributed on the surface of the sample. As shown in Fig. 2b, the
catalyst has a rather wide size distribution for the metal particles,
with most between 1 and 3.5 nm (dp = 2.1 nm, DTEM = 41%). It was
also noteworthy that the Pd crystallites appeared, in general, to
be free of decoration, although some nanoparticles looked as if they
were covered by a thin layer of an amorphous nature. Although this
decoration was not statistically measurable, it might well be an
indication of some alloy formation during the activation protocol,
as will be discussed later.
The nanostructural morphology and chemical composition of
the supported particles of the activated catalysis prior to the reac-
tion (i.e., the as-prepared catalyst re-reduced at 523 K) were inves-
tigated by quasi-in situ TEM analysis. An aliquot of the as-prepared
Pd/Ga2O3 stock was placed in a Mo grid and submitted to a reduc-
tion pretreatment in pure H2 (60 cm3 minÀ1, 523 K, 6 h) inside a
quartz microreactor (hereafter, as-prepared + R523). After reduc-
tion, the microreactor was evacuated (P < 10À6 mbar) for 60 min
These progressive (and opposite) changes in activity toward
methanol and CO on the Pd–Ga2O3 catalyst during the first 3 h of
reaction prompted us to investigate this phenomenon further.
Therefore, we set out to characterize the catalyst both before
(pre) and after (post) the reaction, using TEM and FTIR in situ.
Table 1
Catalytic performance of the catalysts for CO2 hydrogenation under pseudo-steady-state conditions.
Catalysts and supporta DTEMb(%) DCOc(%) S (BET) (m2 gÀ1
)
Reaction pressure (MPa) Activityd (mol gÀ1 sÀ1) Â 108 TOFe (sÀ1) Â 103 Selectivity (%)
CH3OH
CO
DMEf
CH3OH
CO
CH3OH CO DMEd
Pd(1 wt.%)/b-Ga2O3
41
31
12
3
0.7
3
220
102
0
201
220
0
0
0
75
35
–
69
75
–
52
32
0
48
68
0
0
0
–
–
b-Ga2O3
Pd(2 wt.%)/SiO2
–
58
–
56
12
265
Traces
0
3
11
85
2
9.5
12
88
a
b
c
Activation pretreatment of the as-prepared catalysts in flowing H2 (heating rate, 3 K minÀ1; P = 0.1 MPa) from RT to 523 K, maintaining the hydrogen flow for 1 h.
Metal dispersion of the as-prepared catalysts measured by TEM.
Metal dispersion of the as-prepared catalysts reduced at 523 K with pure H2 (1 h) measured by CO pulses at RT [24].
d
After 20 h of reaction in a differential plug-flow microreactor (pseudo-steady-state conditions): H2/CO2 = 3, T = 523 K, SV = 80,000 hÀ1 (3 MPa), SV = 7700 (0.7 MPa), CO2
conversion <1%.
e
TOF: Turnover frequency, calculated as follows: TOFi (sÀ1) = ith product flow rate (mol gÀ1 sÀ1)/quantity of metal surface sites (mol gÀ1); [i: CH3O or CO]. Quantity of metal
surface sites (mol gÀ1) = DCO Â metal loading (mol gÀ1).
f
DME: Dimethyl ether.

S.E. Collins et al. / Journal of Catalysis 292 (2012) 90–98
93
Fig. 2. HRTEM image of the prereaction Pd/b-Ga2O3 catalyst (a) and metal particle
size distribution (b).
and cooled under vacuum to RT. Next, the grid was transferred to
the microscope without exposure of the sample to air.
Fig. 3a shows an illustrative HAADF-STEM image obtained after
the reduction pretreatment. This HAADF-STEM image gave results
similar to those for the as-prepared prereaction catalyst: Pd nano-
particles were homogeneously distributed on the support. A slight
increase in particle size distribution was detected, though without
a pronounced change in the overall particle shape. The metal crys-
tallites appeared to be free of decoration.
Fig. 3. (a) STEM-HAADF image of the Pd/Ga2O3 catalyst reduced quasi-in situ in
pure H2 at 523 K. (b and c) X-EDS line spectra through the A–B and C–D trajectories
over selected metal particles (thin line, Ga; thick line, Pd; spatial resolution,
0.5 nm).
Metal particles located on convex surfaces of the support were
selected for chemical analysis by X-EDS. In particular, line-
spectrum measurements were carried out through these particles
along the lines traced in the figure. The intensity profiles of Pd
and Ga in two of these particles are shown in Fig. 3b and c, respec-
tively. The results clearly showed that gallium had been incorpo-
rated into the initial Pd nanoparticles during the reducing
pretreatment. The X-EDS analysis was carried out across 20 parti-
cles, and reproducible results were obtained.
A representative high-resolution image of a metallic particle ob-
served on the Pd/Ga2O3 catalyst after its exposure to pure H2 at
523 K is included in Fig. 4. To identify the possible formation of
Pd–Ga bimetallic particles, a digital diffraction pattern analysis of
the HRTEM images was performed. The plane distances and angles
observed were in good agreement with Pd2Ga5 in the [011] axis
zone [25]. However, the small differences between this phase
and the pure metallic Pd phase made it impossible to make unam-
biguous phase identification. Penner et al. previously reported that
the Pd5Ga2, Pd5Ga3, and Pd2Ga bimetallic phases exhibit
orthorhombic structures with similar lattice parameters (Pd5Ga3:
a = 5.50 Å, b = 10.53 Å, c = 4.40 Å; Pd5Ga2: a = 18.37 Å, b = 5.48 Å,
c = 4.08 Å; Pd2Ga: a = 5.46 Å, b = 4.03 Å, c = 7.81 Å), giving rise to
very similar diffraction patterns [20]. Thus, the close structural
relationship between these Pd–Ga phases precluded a definitive
assignment of a given bimetallic crystalline phase, as did the small
crystal size of the particles in the sample and the intrinsic experi-
mental error of the measurements. Nevertheless, the joint TEM and
CO adsorption results (shown in the next section) of the Pd/Ga2O3
catalyst reduced with hydrogen at 523 K allowed us to conclude
that bimetallic particles were formed on the surface of the gallia
support upon hydrogen reduction.
To assess the stability of these Pd–Ga bimetallic particles, a com-
plementary experiment was performed, in which the reduced cata-
lyst sample was exposed to air at RT for 24 h and again placed in the
microscope. The HRTEM and STEM-HAADF images confirmed that
several particles became covered by a thin layer of Ga2O3 of an
amorphous nature. Additionally, after 24 h on stream and air pas-
sivation at room temperature, the Pd/Ga2O3 catalyst was examined

94
S.E. Collins et al. / Journal of Catalysis 292 (2012) 90–98
covered by a thin layer that grew on top of the metal crystal planes.
A phase simulation of b-Ga2O3 along the zone [101] axis has been
inserted in the upper left corner of Fig. 5a. It fits perfectly to the
phase observed in both the oxide support and the thin layer deco-
rating the Pd particles. Therefore, we could conclude that the gallia
phase decorating the Pd particles was crystalline and monoclinic.
It was further noted that the thin layers of the support phase
grew epitaxially to the faceted surface planes of the metal parti-
cles. In some cases, these metal particles were settled on a stand,
or ‘‘pedestal,’’ of the support. It is likely that these stands grew un-
der the original contact surface between the Pd crystallites and the
gallia, following the same mechanism that produced the decora-
tion, as analyzed below.
In a recent work [26], HRTEM was used to show that a Ce–Au
alloy was formed on Au/CeO2 under reductive conditions but that,
upon reoxidation, ceria segregated onto the metal, encapsulating it
with a film of CeO2. A similar mechanism was also suggested for
other supported noble metals, such as platinum [27,28].
The lattice spacing of the encapsulated particles corresponded
to metallic Pd. That is, no relaxation (indicative of a bimetallic
phase or Pd–Ga alloy) was observed. However, as indicated before,
DDP analysis does not allow clear identification of Pd–Ga phases,
due to the interference of the oxide layer. Fig. 5b shows a moderate
increase in the metal particle sizes to 2.8 nm and a decrease in the
dispersion, DTEM (down to 33%), in comparison to the as-prepared
(i.e., reduced and passivated) sample. X-ray diffraction patterns
of the as-prepared, re-reduced (as-prepared + R523/6 h + air-
exposed), and postreaction Pd/Ga2O3 catalysts were also taken
(spectra not shown), but only the characteristic monoclinic struc-
ture of the b-Ga2O3 support was registered (i.e., no Pd, Pd-Ga alloy,
or PdO diffraction peaks). This result was most likely due to the low
metal loading and small particle size of the palladium crystallites.
Fig. 4. DDP analysis of a representative Pd–Ga particle by quasi-in situ HRTEM
microscopy.
by ex situ TEM. A representative HRTEM image and the particle size
distribution of the postreaction Pd/Ga2O3 catalyst are shown in
Fig. 5a and b, respectively. Dramatic nanostructural changes ap-
peared in the material: most of the palladium crystallites became
3.3. In situ Fourier transform infrared characterization
3.3.1. CO adsorption by infrared spectroscopy
The as-prepared Pd/b-Ga2O3 catalyst was exposed to a series of
reduction steps in pure H2 (760 Torr, 100 m3 minÀ1) at increasing
temperatures of 423, 523, and 723 K (60 min each time), evacuated
at the same temperature for 60 min, and cooled under vacuum to
RT. After each reduction treatment, the catalyst was exposed to
10 Torr CO at room temperature for 10 min. The collected spectra
are shown in Fig. 6. After the Pd/b-Ga2O3 catalyst that was reduced
at 423 K was exposed to CO, several absorption bands correspond-
ing to the linear (L = 2092 cmÀ1), bridged (B = 1975 cmÀ1), and
threefold bridged or hollow (H = 1915 cmÀ1) forms were observed
[29–31]. This spectrum was very similar to the one obtained after
the adsorption of CO (10 Torr, RT) on Pd/SiO2 [32]. After reduction
at higher temperatures (523 and 723 K), the spectra clearly showed
a progressive loss of intensity of the bridged CO bands (B and H)
and a moderate increase in the linear CO signal.
Kovnir et al. [19] reported the characterization of Pd–Ga bime-
tallic particles using CO chemisorption followed by FTIR. No
adsorption of CO in its bridged forms appeared on this Pd–Ga sys-
tem, and only one sharp band with a maximum at 2047 cmÀ1, as-
signed to CO adsorbed on Pd in the on-top position, was observed.
This finding was interpreted as being the result of the isolation of
Pd atoms on the surface of PdGa, as expected from the bulk crystal
structure of the intermetallic compound. The breakdown of the
arrangement among adjacent Pd atoms, or the increase in the dis-
tance between adjacent Pd atoms, prevented the adsorption of CO
in the bridged form. The significant red shift of the CO(L) band was
ascribed to a strong electronic modification of Pd in the intermetal-
lic compound, as detected by XPS measurements [19]. Identical
results were reported very recently by Haghofer et al. using
gallia-supported palladium [22].
Fig. 5. (a) HRTEM image of the postreaction Pd/b-Ga2O3 catalyst (inset: [101] zone
axis simulation image of a b-Ga2O3 crystal) and (b) metal particle size distribution.

S.E. Collins et al. / Journal of Catalysis 292 (2012) 90–98
95
Fig. 6. FTIR spectra after CO adsorption (10 Torr, 300 K, 10 min) over the
Pd/b-Ga2O3 catalyst prereduced in situ in pure H2 (60 min, 100 cm3 minÀ1) at
423, 523, and 723 K.
Fig. 7. Hydrogenation of surface (bi)carbonate groups bound to the gallia surface to
formate species over Pd/b-Ga2O3 prereduced at 423 K after switching from a pure
CO2 stream (100 cm3/min) to a H2/CO2 = 3 mixture (100 cm3/min). Total pressur-
e = 0.1 MPa. Solid squares: bicarbonates; solid circles: bidentate carbonates; open
squares: polydentate/monodentate carbonates.
We also performed a CO adsorption experiment (pulses), at RT,
over the activated catalyst (as-prepared + R523/1 h). The obtained
metal dispersion, DCO (calculated using a CO/Pd = 1/1 stoichiome-
try), was 31%, a value somewhat smaller than the one deduced
from TEM size distribution data (41%) (on the reference Pd/SiO2
catalyst, DTEM and DCO were equal; Table 1). These results also indi-
cated a loss of Pd metal sites at the surface of the particles due to
bimetallic particle formation, since CO chemisorption on this
material as measured by FTIR—spectra not shown—is irreversible.
surface carbonates on the postreaction catalyst (spectra not
shown). This result was expected due to the inhibition of hydrogen
dissociation at RT over the gallia-decorated metal particles.
3.4. High-pressure in situ reaction of H2 and CO2 followed by Fourier
transform infrared
An FTIR study of CO2 hydrogenation on the activated Pd/Ga2O3
catalyst (as-prepared + R523/1 h) was carried out at 0.7 MPa.
Before the reaction test was performed, the catalyst was reducÀe1d
3.3.2. CO2 hydrogenation at RT
The hydrogenation ability of the Pd–Ga bimetallic particles was
tested by in situ FTIR spectroscopy after activation of samples of
the as-prepared Pd/b-Ga2O3 catalyst. As reported above, palladium
catalysts supported on (or promoted with) Ga2O3 are able to hydro-
genate the (bi)carbonate species bonded to the gallia surface that
appear upon adsorption of CO2, producing formate even at ambient
temperature [10]. This test reaction was used here to measure the
ability of the metal particles to chemisorb and spillover hydrogen
to the support. Briefly, after reduction in pure hydrogen at 523 K
for 1 h, the sample wafers were evacuated at the same temperature
for 30 min and cooled to RT under vacuum. Then pure CO2 was flo-
wed (100 cm3 minÀ1) at room temperature for 15 min, producing
(bi)carbonate species on the gallium oxide surface [9]. Next, the
in the reactor cell at 523 K under H2 flow (100 cm3 min
0.1 MPa). Then the cell was purged with He (100 cm3 minÀ1
,
,
0.1 MPa) and reference spectra were collected. Next, a H2/CO2 mix-
ture (H2/CO2 = 3/1) was flowed through the cell at 523 K and
0.7 MPa. Infrared spectra were simultaneously collected every
5 min. Fig. 8a shows the time-resolved infrared spectra collected
after switching from pure H2 to H2/CO2 at 0.7 MPa. Almost imme-
diately, hydrogenated surface species bonded to the Ga2O3 surface
appeared (m-HCOO, br-HCOO, and CH3O) [10,12], as did linear
[CO(L)] and a very weak signal for bridged [CO(B)] bonded carbon
monoxide on the Pd sites [32]. The evolution of these surface
species is shown in Fig. 8b. The surface concentration of all these
species increased during the first 10 min and then reached a steady
state after approximately 40 min.
After the gas flow was switched back to pure H2, with the total
pressure kept constant (0.7 MPa), a sharp decrease in m-HCOO and
an increase in the CH3O concentration (Fig. 9a and b) occurred.
Meanwhile, the signal intensity of br-HCOO decayed steadily, but
no more CH3O groups were produced after 20 min (the CO2 partial
pressure inside the cell decreased to zero in <5 min—dotted line).
flow was changed from pure CO2 to a H2/CO2 = 3 (100 cm3 minÀ1
)
mixture. As shown in Fig. 7, complete hydrogenation of the (bi)car-
bonate surface species to formate (HCOO) was observed in the acti-
vated prereaction Pd/Ga2O3 catalyst after approximately 5 min on
stream at RT (m-HCOO [
s(OCO) = 1272 cmÀ1], br-HCOO
1382 cmÀ1 s(OCO) = 1362 cmÀ1] and b-HCOO [
cmÀ1]) [10].
m ,
as(OCO) = 1660 cmÀ1, d(CH) = 1309 cmÀ1
m
[
m
as(OCO) = 1580 cmÀ1
as(OCO) = 1600
, d(CH) =
,
m
m
The postreaction catalyst, which predominantly contained fully
encapsulated metal particles, was also tested for the hydrogena-
tion of carbonates at RT. A wafer of this material was reduced just
at 343 K, to avoid as much as possible the rupture (re-reduction) of
the gallia-layer-decorated Pd crystallites. As has already been
established, H2 treatment at such a temperature is sufficient to
reduce supported palladium particles to the metallic state. The
experimental results showed almost no hydrogenation of these
4. Discussion
4.1. Catalyst composition and reactivity
The time evolution of the catalytic activity of the activated
Pd/b-Ga2O3 catalyst (as-prepared + R523/1 h) upon its exposure
to the stoichiometric H2/CO2 mixture (Fig. 1) shows a remarkable

96
S.E. Collins et al. / Journal of Catalysis 292 (2012) 90–98
Fig. 8. (a) Time-resolved IR spectra over Pd/b-Ga2O3 at P = 0.7 MPa and T = 523 K
after switching from pure H2 (100 cm3 minÀ1
to H2/CO2 (molar ratio = 3/1,
140 cm3 minÀ1). (b) Time evolution of the integrated intensity of the IR bands of
the surface species after the switch: br-HCOO [
as(CO2) = 1580 cmÀ1], m-HCOO
as(CO2) = 1660 cmÀ1], CH3O [ (CO) = 1060 cmÀ1], and CO(L) [ (CO) = 2066 cmÀ1].
)
m
Fig. 9. (a) Time-resolved IR spectra over Pd/b-Ga2O3 at P = 0.7 MPa and T = 523 K
after switching from H2/CO2 (molar ratio = 3/1, 140 cm3 minÀ1
to pure H2
(100 cm3 minÀ1). (b) Time evolution of the integrated intensity of the IR bands of
the surface species after the switch: br-HCOO [
as(CO2) = 1580 cmÀ1], m-HCOO
as(CO2) = 1660 cmÀ1], CH3O [ (CO) = 1060 cmÀ1], and CO(L) [ (CO) = 2066 cmÀ1].
[m
m
m
)
m
change in the selectivity to methanol and CO during the first 3 h,
which was entirely absent in the reference Pd/SiO2 catalyst. Nano-
structural and nanochemical analyses by TEM of the catalyst in the
as-prepared hydrogen-reduced state at 523 K and the spent cata-
lyst taken from the reactor were also performed. As shown, the
as-prepared catalyst possessed predominantly metallic palladium
particles on the support, as well as some metal crystallites partially
covered by an amorphous layer of gallia. When the catalyst was re-
duced at 523 K (pure H2; 6 h) and then studied by TEM without
exposure to air, no decorated metal particles were observed. How-
ever, the presence of bimetallic Pd–Ga particles was revealed. As
mentioned above, the exact composition of these bimetallic
Pd–Ga nanoparticles could not be identified by means of its diffrac-
tion pattern due to the very close resemblance between pure pal-
ladium and bimetallic phases such as Pd5Ga2, Pd5Ga3, and Pd2Ga
[20]. Conversely, in the case of the air-exposed postreaction cata-
lyst, a remarkable encapsulation of the metal particles by a crystal-
line layer of b-Ga2O3 was observed.
[m
m
m
chemisorbs in bridged forms, but does so only in the linear or
on-top position (Fig. 7). This feature, showing the progressive
isolation of the Pd sites, was ascribed to the gradual formation of
Pd–Ga bimetallic particles. Similar results were recently reported
by Klötzer and coworkers. Using XRD and HRTEM, they showed
that Pd–Ga bimetallic particles were produced after reduction pre-
treatments at temperatures above 523 K. Furthermore, after the
sample was reoxidized, a gallium oxide layer that covered the me-
tal, causing its encapsulation, was also present [20,21].
Contemporary with our work, but in regard to the methanol
steam reforming reaction, Haghofer et al. [22] studied in depth a
Pd (5 wt.%)/Ga2O3 catalyst. By in situ XRD, they found that bulk
Pd2Ga and Pd–Ga bimetallics were formed after the material was
reduced at 673 and 773 K, respectively, and they appeared to be
stable in air at room temperature. However, they also reported
(as revealed via FTIR by CO chemisorption) that the surface compo-
sition of these bimetallic particles significantly changed either
after exposure of the catalyst to synthetic air at 303 K due to the
At the surface level, FTIR spectroscopy of adsorbed CO revealed
that upon submission of the catalyst to successive reductions with
hydrogen at increasing temperature, carbon monoxide no longer

S.E. Collins et al. / Journal of Catalysis 292 (2012) 90–98
97
formation of Ga2O3 patches or by contact with CO due to restruc-
turing. Nevertheless, under methanol steam reforming conditions
(using in operando FTIR), part of the particles consisted of metallic
Pd. This result contrasted with what had been observed after
hydrogen reduction and CO adsorption, thus indicating that partial
decomposition of the bimetallic phase had occurred. Therefore, we
could conclude that an analogous process occurred with our
Pd/b-Ga2O3 catalyst: (some phase of) Pd–Ga bimetallic particles
as formed after the activation of the catalyst at 723 K in 2% H2,
which then (partially, or superficially,) decomposed during the pas-
sivation treatment prior to shelving the as-prepared stock material.
Inside the plug-flow microreactor, the re-reduction of the as-
prepared catalyst at 523 K for 60 min began to form a Pd–Ga alloy,
but certainly the remarkable initial selectivity to CO, instead of to
methanol, was indicative of Pd–Pd domains on the surface of the
metal crystallites. It is likely that during the CO2 hydrogenation
reaction, intermetallic Pd–Ga had already formed and continued
incorporating Ga0, thus enriching the Pd–Ga alloy. It is also possi-
ble that the presence of the produced CO restructured the surface
of the nanoparticles during the reaction to bring about both the ob-
servable changes in catalyst selectivity during the first 3 h on
stream and the steady-state selectivity.
Our research indicated that bimetallic Pd–Ga particles were
responsible for the hydrogenation ability of our catalyst as well.
The experimental results indicated that bimetallic surfaces were
indeed present from the beginning of the hydrogenation of CO2
to methanol on this Pd/Ga2O3 catalyst. Furthermore, the presence
of bimetallic surfaces could be indirectly observed through FTIR
study of CO2 hydrogenation at 0.7 MPa and 523 K. As shown in
Fig. 9 and 10, extremely weak signals (almost indistinguishable
from the background) from bridged CO were registered, but
strongly bound linear CO was clearly observable. It is worth noting
that CO(B) species exhibit higher adsorption energy than linear
species on palladium; therefore, on supported Pd catalysts (e.g.,
Pd/SiO2), they remain adsorbed at temperatures much higher than
523 K [26]. Hence, the presence of these Pd–Ga bimetallic particles
during the H2/CO2 reaction could be inferred from in situ FTIR at
0.7 MPa (Fig. 8). At this pressure, the activated catalyst was able
to hydrogenate CO2 to methanol (Table 1).
Furthermore, these in situ results suggest that methoxy/metha-
nol formation from H2/CO2 took place as a consequence of
m-HCOO hydrogenation, while br-HCOO behaved more like a spec-
tator, or laggard reaction intermediate. The key role of monoden-
tate formate bound to Ga2O3 in methanol synthesis has already
been demonstrated on the basis of theoretical and experimental
results [33].
The Pd–Ga bimetallic phase was decomposed upon exposure of
the material to oxidizing conditions. Seemingly, gallium can segre-
gate to the surface of the particles and further oxidize after the pas-
sivation and removal of the catalyst from the reactor, thereby
producing the gallia-encapsulated Pd crystallites shown in the
HRTEM micrographs of Figs. 3 and 4.
The new experimental evidence reported here allowed reinter-
pretation of the dramatic change in selectivity of the Pd/Ga2O3 cat-
alyst at the beginning of the H2/CO2 reaction within the framework
of our postulated reaction mechanism. Thus, if the selectivity to-
ward methanol increased as a result of the progressive formation
of a Pd–Ga alloy under the (reductive) reaction conditions, said in-
crease in CH3OH production rate in detriment of CO production
could certainly be a consequence of the lower activity of the (new-
ly formed) Pd–Ga bimetallic particles to decompose methanol to
4.2. Role of Pd–Ga bimetallic particles in the reaction mechanism
As mentioned earlier, the reaction pathway for the hydrogena-
tion of CO2 to methanol over this same Pd/Ga2O3 catalyst was pre-
viously explored [10]. Based on detailed and systematic
spectroscopic identification of the surface intermediates, the fol-
lowing reaction pathway was proposed: (bi)carbonates, generated
by CO2 adsorption on surface sites of the gallium oxide, are hydro-
genated successively to formate(s), methylenebisoxy and methoxy
species, and then to methanol. Atomic hydrogen is provided by the
supported metal Pd crystallites through H2 dissociation on the me-
tal, generating active HS species that spill over onto the support,
thereby hydrogenating the carbonaceous intermediates bonded
to the gallia surface.
The results presented here also support the bifunctional model,
because the hydrogenation of (bi)carbonate species to formate in
the as-prepared + R523/1 h catalyst readily proceeds at room tem-
perature, through the spillover of HS from the bimetallic particles
to the support (Figs. 7 and 8).
In addition to carbon oxide(s) hydrogenation and methanol
steam reforming [15,16,20–22], Pd–Ga intermetallic catalysts have
also been investigated for the selective (or partial) hydrogenation
of acetylene in the recent years [17]. These catalysts, particularly
bulk Pd–Ga and Pd3Ga7, exhibited remarkable selectivity for the
partial hydrogenation of C2H2 in the presence of a large excess of
ethylene. The high selectivity was assigned to the isolation of the
palladium active sites due to the crystallographic structure of said
Pd–Ga intermetallic compounds, which resulted in a geometric
H2 and CO retaining the hydrogen dissociation (metallic) function.
This assertion is supported by recent work in the methanol steam
reforming reaction [15,16,20–22]). Therefore, methanol is pro-
duced via a bifunctional mechanism whereby H2 can still be disso-
ciated over the Pd–Ga bimetallic particles and spillover to the
gallia sites where CO2 is hydrogenated stepwise.
Therefore, because the ‘‘dynamic equilibrium between bimetal-
lic Pd–Ga formation by hydrogen and decomposition of the alloy
(at the superficial level) by CO occurs during the reaction’’ [22],
the high partial pressure of H2 (ca. 2 MPa) during the CO2/H2 reac-
tion could keep (or even increase the surface concentration of Ga0
on the metal particles). This behavior would counteract the decom-
position of the alloy by the produced CO. This in turn should de-
crease methanol decomposition, which is a property of pure Pd
arrays.
It must be recalled that the decrease in CO production with time
on the Pd/b-Ga2O3 catalyst was almost threefold larger than the
increase in methanol production (Fig. 1). At this point, we can
only dwell on along the following line of thought: carbon
monoxide production via the reverse water–gas shift reaction
(CO2 + H2 M CO + H2O) proceeds efficiently on Pd particles, as ob-
served in the case of Pd/SiO2 (Fig. 1). Yet, the progressive on stream
reformation of the Pd–Ga bimetallic in the (as-prepared + 523/1 h)
Pd/b-Ga2O3 catalyst effectively inhibited the RWGS reaction on the
metallic phase, and the material still remained efficiently active for
the provision of atomic hydrogen to the carbonaceous species
chemisorbed onto the support.
effect and led to weakly
p-bonded acetylene molecules on top of
the isolated Pd atoms. In addition, an electronic effect resulting
in the modification of the adsorption and desorption properties
of the acetylene and a kinetic effect due to the decreased availabil-
ity of hydrogen because of the absence of Pd hydrides were pro-
posed as well [17]. Nevertheless, the Pd–Ga intermetallic
compounds were still able to activate and dissociate H2 to selec-
tively hydrogenate the weakly adsorbed acetylene species.
5. Conclusions
A progressive change in selectivity during the CO2 reaction with
H2, increasing the yield of methanol at the expense of the undesired
CO, was observed under typical process conditions. The formation

98
S.E. Collins et al. / Journal of Catalysis 292 (2012) 90–98
[11] S.E. Collins, M.A. Baltanás, A.L. Bonivardi, Appl. Catal. A 295 (2005) 126.
[12] S.E. Collins, L.E. Briand, L.A. Gambaro, M.A. Baltanás, A.L. Bonivardi, J. Phys.
Chem. C 112 (2008) 14988.
[13] S.E. Collins, D.L. Chiavassa, A.L. Bonivardi, M.A. Baltanás, Catal. Lett. 103 (2005)
83.
[14] D.L. Chiavassa, S.E. Collins, A.L. Bonivardi, M.A. Baltanás, Chem. Eng. J. 150
(2009) 204.
[15] N. Iwasa, T. Mayanagi, N. Ogawa, K. Sakata, N. Takezawa, Catal. Lett. 54 (1998)
119.
[16] N. Iwasa, T. Mayanagi, W. Nomura, M. Arai, N. Takezawa, Appl. Catal. A 248
(2003) 153.
[17] J. Osswald, K. Kovnir, M. Armbruster, R. Giedigkeit, R.E. Jentoft, U. Wilda, Y.
Grinb, R. Schlögl, J. Catal. 258 (2008) 219.
[18] M. Armbrüster, K. Kovnir, M. Behrens, D. Teschner, Y. Grin, R. Schlögl, J. Am.
Chem. Soc. 132 (2010) 14745.
[19] K. Kovnir, M. Armbruster, D. Teschner, T.V. Venkov, L. Szentmiklosi, F.C. Jentoft,
A. Knop-Gericke, Yu. Grin, R. Schlögl, Surf. Sci. 603 (2009) 1784.
[20] S. Penner, H. Lorenz, W. Jochum, M. Stöger-Pollach, D. Wang, C. Rameshan, B.
Klötzer, Appl. Catal. A 358 (2009) 193.
[21] H. Lorenz, S. Penner, W. Jochum, C. Rameshan, B. Klötzer, Appl. Catal. A 358
(2009) 203.
[22] A. Haghofer, K. Föttinger, F. Girdsdies, D. Teschner, A. Knop-Gerike, R. Schlögl,
G. Rupprechter, J. Catal. 286 (2011) 13.
[23] S.E. Collins, M.A. Baltanás, A.L. Bonivardi, C. Mira, J.J. Calvino, S. Bernal, 14th
International Congress in Catalysis, June 15, 2008, Seoul, Korea.
[24] C.S. Fung, C.A. Querini, J. Catal. 138 (1992) 240.
[25] K. Khalaff, K. Schubert, J. Less-Common Met. 37 (1974) 129.
[26] T. Akita, M. Okumura, K. Tanaka, M. Kohyama, M. Haruta, Catal. Today 117
(2006) 62.
of Pd–Ga bimetallic particles was identified by means of a combina-
tion of quasi-in situ transmission electron microscopy techniques
and CO adsorption experiments. The role of these Ga–Pd bimetallic
particles in the reaction mechanism of methanol synthesis was the
same as that of pure Pd: to dissociate and spill over H to the gallia
sites, where CO2 was stepwise hydrogenated to methanol. Addi-
tionally, the decrease in CO production, by inhibition of the metha-
nol decomposition and/or reverse water–gas shift reaction, on the
bimetallic phase could be inferred. Gallia encapsulation of the Pd
particles constituted an ex situ artifact.
Acknowledgments
Financial support from the Agencia Nacional para la Promoción
de la Ciencia y Tecnología of Argentina (ANPCyT) and the Agencia
Española de Cooperación Internacional of Spain (AECI) is gratefully
acknowledged. S.E.C., M.A.B., and A.L.B. thank the Consejo Nacional
de Investigaciones Científicas y Técnicas (CONICET) and Universi-
dad Nacional del Litoral (UNL) for their continued support.
References
[1] L. Fan, K. Fujimoto, J. Catal. 150 (1994) 217.
[27] S. Bernal, F.J. Botana, J.J. Calvino, G.A. Cifredo, J.A. Pérez-Omil, J.M. Pintado,
Catal. Today 23 (1995) 219.
[28] S. Bernal, J.J. Calvino, M.A. Cauqui, J.M. Gatica, C. Larese, J.A. Pérez Omil, J.M.
Pintado, Catal. Today 50 (1999) 175.
[2] W.X. Pan, R. Cao, D.L. Roberts, G.L. Griffin, J. Catal. 114 (1988) 440.
[3] G.C. Chinchen, K.C. Waugh, D.A. Wham, Appl. Catal. 25 (1986) 101.
[4] T. Fujitani, M. Saito, Y. Kanai, Y. Watanabe, J. Nakamura, T. Uchijima, Appl.
Catal. A 125 (1995) L199.
[29] E. Ozensoy, D.W. Goodman, Phys. Chem. Chem. Phys. 6 (2004) 3765.
[30] T. Lear, R. Marshall, J.A. López-Sanchez, S.D. Jackson, T.M. Klapötke, M.
Bräumer, G. Rupprechter, H.J. Freund, D. Lennon, J. Chem. Phys. 123 (2005)
174706.
[5] T. Fujitani, I. Nakamura, Bull. Chem. Soc. Jpn. 75 (2002) 1393.
[6] A.L. Bonivardi, D.L. Chiavassa, C.A. Querini, M.A. Baltanás, Stud. Surf. Sci. Catal.
130D (2000) 3747.
[7] S.E. Collins, M.A. Baltanás, J.L. García-Fierro, A.L. Bonivardi, J. Catal. 211 (2002)
252.
[31] F.M. Hoffmann, Surf. Sci. Rep. 3 (1983) 1.
[32] S.E. Collins, M.A. Baltanás, A.L. Bonivardi, J. Mol. Catal. A 281 (2008) 73.
[33] M. Calatayud, S.E. Collins, M.A. Baltanás, A.L. Bonivardi, Phys. Chem. Chem.
Phys. 11 (2009) 1397.
[8] S.E. Collins, M.A. Baltanás, A.L. Bonivardi, Langmuir 21 (2005) 962.
[9] S.E. Collins, M.A. Baltanás, A.L. Bonivardi, J. Phys. Chem. B 110 (2006) 5498.
[10] S.E. Collins, M.A. Baltanás, A.L. Bonivardi, J. Catal. 226 (2004) 410.