Preferential CO oxidation in hydrogen (PROX) on ceria-supported catalysts, part I: Oxidation state and surface species on Pt/CeO2 under reaction conditions

Article abstract of DOI:10.1016/j.jcat.2005.10.014

The CO content of hydrogen feed to proton-exchange membrane fuel cells (PEMFCs) must be kept under 1-100 ppm for their proper operation. This can be achieved by using catalysts able to selectively oxidize CO in the presence of excess hydrogen (PROX). The present study reports on the mechanism of the PROX reaction on Pt/CeO₂ catalyst, using catalytic tests, in situ DRIFTS, high-pressure XPS, HRTEM, and TDS techniques. Bulk metallic, pronounced adsorbate-induced surface Pt, and a small amount of oxidized Pt sites were detected by in situ high-pressure XPS under PROX conditions. The preoxidized ceria surface was strongly reduced in pure H₂ but significantly reoxidized under PROX conditions (i.e., O₂ + CO in excess hydrogen) at T=358 K. The remaining small amount of Ce³⁺ decreased with increasing temperature. HRTEM found well-crystallized CeO₂ particles (8-10 nm) in the case of activated (pre-oxidized) sample that transformed in a large extent to an oxygen deficient ceria supercell structure after PROX reaction. Metallic Pt particles (2-3 nm) and small (0.5-0.6 nm) Pt clusters were found by HRTEM. These findings were in accordance with the variations in relative intensity of the corresponding PtCO bands (DRIFTS). Different types of carbonate and formate species were detected (XPS and DRIFTS). Their possible role in the reaction mechanism is discussed. Resolved OH bands could not be found by DRIFTS in the PROX reaction mixture indicating significant amount of adsorbed water in a hydrogen-bonded structure. Its presence seems to suppress hydrogen oxidation while CO oxidation still takes place, as the metallic particles are covered by CO (DRIFTS). The direct contribution of surface water in a low-temperature water-gas shift-type reaction in the PROX mixture is proposed.

Full text of DOI:10.1016/j.jcat.2005.10.014

Journal of Catalysis 237 (2006) 1–16
www.elsevier.com/locate/jcat
Preferential CO oxidation in hydrogen (PROX) on ceria-supported catalysts,
part I: Oxidation state and surface species on Pt/CeO2
under reaction conditions
a
a,b,∗
a
b
b
b
c
O. Pozdnyakova , D. Teschner
, A. Wootsch , J. Kröhnert , B. Steinhauer , H. Sauer , L. Toth ,
b
b
a
b
F.C. Jentoft , A. Knop-Gericke , Z. Paál , R. Schlögl
a
Institute of Isotopes, CRC, HAS, POB 77, Budapest, H-1525, Hungary
b
Department of Inorganic Chemistry, Fritz-Haber-Institut der Max-Plank-Gesellschaft, Faradayweg 4-6, D-14195 Berlin, Germany
c
Research Institute for Technical Physics and Materials Science, HAS, Budapest, POB 49, H-1525, Hungary
Received 4 September 2005; revised 13 October 2005; accepted 17 October 2005
Available online 17 November 2005
Abstract
The CO content of hydrogen feed to proton-exchange membrane fuel cells (PEMFCs) must be kept under 1–100 ppm for their proper operation.
This can be achieved by using catalysts able to selectively oxidize CO in the presence of excess hydrogen (PROX). The present study reports on the
mechanism of the PROX reaction on Pt/CeO catalyst, using catalytic tests, in situ DRIFTS, high-pressure XPS, HRTEM, and TDS techniques.
2
Bulk metallic, pronounced adsorbate-induced surface Pt, and a small amount of oxidized Pt sites were detected by in situ high-pressure XPS
under PROX conditions. The preoxidized ceria surface was strongly reduced in pure H but significantly reoxidized under PROX conditions
2
3+
(
i.e., O + CO in excess hydrogen) at T = 358 K. The remaining small amount of Ce decreased with increasing temperature. HRTEM found
2
well-crystallized CeO particles (8–10 nm) in the case of activated (pre-oxidized) sample that transformed in a large extent to an oxygen deficient
2
ceria supercell structure after PROX reaction. Metallic Pt particles (2–3 nm) and small (0.5–0.6 nm) Pt clusters were found by HRTEM. These
findings were in accordance with the variations in relative intensity of the corresponding Pt–CO bands (DRIFTS). Different types of carbonate
and formate species were detected (XPS and DRIFTS). Their possible role in the reaction mechanism is discussed. Resolved OH bands could not
be found by DRIFTS in the PROX reaction mixture indicating significant amount of adsorbed water in a hydrogen-bonded structure. Its presence
seems to suppress hydrogen oxidation while CO oxidation still takes place, as the metallic particles are covered by CO (DRIFTS). The direct
contribution of surface water in a low-temperature water–gas shift-type reaction in the PROX mixture is proposed.

2005 Elsevier Inc. All rights reserved.
Keywords: Hydrogen purification; Fuel cell; Preferential CO oxidation; PROX; Platinum; Pt/CeO ; Ceria; High pressure XPS; In situ DRIFTS; TDS; HRTEM
2
1
. Introduction
(ca. 5–15%) of CO are formed together with H2, H2O, CO2, and
CH4. The concentration of carbon monoxide in the hydrogen
feed must be kept below 1–100 ppm for the proper operation
of PEMFC [1]. A subsequent water–gas shift (WGS) reaction
reduces the amount of CO to 0.5–1% [2]. This amount of CO
is still high, and it can be removed by preferential oxidation
Increased attention now turned to development of fuel cell-
powered systems, because of their expectedly low environ-
mental impact [1]. The proton-exchange membrane fuel cell
(
PEMFC) is fueled with clean, CO-free hydrogen, which can
be produced by steam reforming (STR), autothermal reform-
ing (ATR), or partial oxidation (POX) of natural gas, light-oil
fractions, and alcohols [2,3]. Unfortunately, significant amounts
(
PROX) and/or by methanation of CO using mainly Pt and Pt–
Ru bimetallic catalysts [4,5].
Several catalytic formulations have been tentatively tested in
the PROX reaction so far. Supported noble metals, such as Au
[6–9], Pt [6,9–13], Rh [10], Ru [10,14], and the bimetallic Pt–
Sn system [12,15], were found applicable for the reaction. In
*
Corresponding author. Fax: +49 30 8413 4676.
E-mail address: teschner@fhi-berlin.mpg.de (D. Teschner).
0021-9517/$ – see front matter  2005 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2005.10.014

2
O. Pozdnyakova et al. / Journal of Catalysis 237 (2006) 1–16
general, the suitable catalyst must adsorb CO and provide acti-
vated oxygen, while hydrogen adsorption must be suppressed.
Carbon monoxide is thought to adsorb on the metal phase (Au
or Pt). CO adsorption on gold is much weaker than on Pt, how-
ever, manifesting itself in low steady-state CO coverage at op-
erational conditions [7–9]. But oxygen does not adsorb on these
metals either, because the surface is fully covered by CO (on Pt
at T < 440 K [11,13]) or because of the very low sticking coef-
ficient (on Au [16]). Therefore, oxygen should be activated on
different sites, either on the reducible support [6–9,13] or—in
the case of modified Pt catalysts—on partially oxidized patches,
such as Ru(ox) [14], Sn(ox) [15], Ce(ox) [12], or Fe(ox) [17].
In this way, oxygen should diffuse to interface sites and subse-
quently react either at the perimeter or after a spillover process
of one of the components on the adjacent metal/oxide site.
The important role of a reducible support manifests itself in
Various direct and indirect experimental evidence has been
presented for the reaction mechanism in the PROX reaction
on ceria-supported Pt [13]; nevertheless, some controversy and
yet-unresolved questions remain. For example, what is the sur-
face state of the catalysts during the reaction? Is Pt reduced in
the presence of the PROX mixture despite the oxidizing pre-
conditioning? Is the reduction degree of reducible oxides cru-
cial to supply activated oxygen to the reaction? Can the strong
temperature dependence of the selectivity on Pt/CeOx also be
explained simply by changes in the CO coverage, as has been
suggested for alloy samples [14,15]? What are the possible sur-
face species formed during the PROX reaction? Are they the
same species as formed during CO oxidation (without hydro-
gen)?
To gain further insight into the reaction mechanism of the
PROX reaction on Pt/CeO2 catalysts, we examined changes
in the oxidation state of ceria and Pt occurring under various
reaction environments by high-pressure XPS, as well as the sur-
face species formed after adsorption of CO alone, CO + O2,
or CO + H2 + O2 (PROX mixture) onto oxidized catalysts by
in situ DRIFTS. In a companion paper (Part II), we report on
Pd/CeO2 catalysts using the same techniques; because the pal-
ladium samples are far less active in the PROX reaction, we
compare the two systems to obtain a better understanding of
the reaction mechanism.
the different behavior of Pt/Al O and Pt on ceria containing
2
3
supports in the PROX reaction [13]. Supporting Pt on Al2O3
results in increasing O2 conversion and a maximum CO conver-
sion as a function of temperature at around 440 K in the PROX
reaction [10,11,13], whereas oxygen conversion on Pt/CeO is
2
very high even at low temperature, with a sharp “light off”
activity at around 363 K [13]. The maximum selectivity as a
function of temperature was observed at much lower tempera-
ture on Pt/CeO than on Pt/Al O , and, accordingly, the maxi-
2
2
3
mum CO conversion appears at a temperature of around 370 K
on the former sample [13]. The different behavior of the sam-
ples is interpreted by different reaction mechanisms [13]. On
Pt/Al O , a competitive Langmuir–Hinshelwood kinetics can
2
. Experimental
2.1. Catalysts
2
3
be supposed, because Pt can adsorb all reactants involved. On
ceria-supported Pt, a noncompetitive Langmuir–Hinshelwood
mechanism has been described, involving CO activation on
Pt particles and their reaction with oxygen activated from the
support on the metal/oxide interface at low temperature (T =
Two catalysts with nominally 1 and 5% metal loadings were
prepared on ceria support (Rhodia Catalysts & Electronics,
2
−1
France; BET = 96 m g [41]) by wet impregnation with an
aqueous solution of Pt(NH ) (OH) [13,42]. The impregnated
3
4
2
3
63–423 K). Both competitive and noncompetitive reaction
samples were dried at 393 K overnight and calcined for 4 h at
773 K in flowing air (30 mL/min), then reduced at 673 K for
4 h in flowing H2 (30 mL/min). Dispersion was determined by
low-temperature (223 K) H2 adsorption after reduction [43,44].
The obtained values were D = 62% for 1% Pt/CeO2 and 18%
for 5% Pt/CeO2.
pathways, as well as reaction on the ceria support, have been
proposed for higher temperatures [13].
From a more general standpoint, cerium oxide is active
in transient oxygen storage, and as a catalyst support, it can
promote oxidation even under oxygen-poor conditions [18].
Ceria-supported platinum [19,20], rhodium [21,22], and pal-
ladium [23] catalysts are also remarkably active in the low-
temperature oxidation of CO, and they are able to oxidize car-
bon monoxide even in the absence of oxygen, in the so-called
2.2. Catalysis
Catalytic tests were carried out in an atmospheric continuous-
flow glass reactor system with stainless steel tubing and con-
nections. Analytical-grade hydrogen, oxygen, and CO were
used as inlet gases and controlled by Brooks mass-flow con-
trollers calibrated earlier. Product analysis was performed by
a gas chromatograph (TCD) equipped with a polar column
(Poropak Q) to separate CO and H O from the other effluent
“
oxygen storage capacity” (OSC) measurement [24–26].
A number of studies have been devoted to ceria-containing
catalysts, investigating chemisorptive [27,28], structural [29,
0], and catalytic properties [18,31]. Adsorption of CO on
3
ceria-supported catalysts leads to the accumulation of various
carboxylate, carbonate, and formate species on the ceria sur-
face, as shown on infrared spectroscopic studies [32–37]. The
oxidation state of ceria was also studied after different pretreat-
ments by XPS [38–40]; it was found that hydrogen exposure
2
2
gases, as well as by a hydrogen compensated flue-gas ana-
lyzer (MRU DELTA 65-3) for CO and O quantification. Only
2
CO and H O were detected as products; methane did not ap-
2
2
leads to a high concentration of Ce3 at surface and subsur-
face sites, and also that the surface layer of partly reduced ceria
easily reoxidizes after exposure to O2 at room temperature.
+
pear in our experimental conditions. The total gas inlet was
100 N mL/min, containing 1% CO, 0.4–1% O (oxygen ex-
2
cess, λ, 0.8–2), and the rest H2.

O. Pozdnyakova et al. / Journal of Catalysis 237 (2006) 1–16
3
Catalysts were activated in situ in flowing air (30 mL/min) at
73 K before catalytic testing. A charge of 82 mg 1% Pt/CeO2
cal grade and controlled by mass flow controllers. The total gas
inlet was 50 N mL/min, containing 1% CO in N2 (referred to
as CO alone), 1% CO and 1% O in N (CO + O ), or 1% CO
5
and 73 mg 5% Pt/CeO2 was used in the catalytic reactor and
reactivated between different measurement series by the afore-
mentioned treatment [13]. CO and O2 conversion, as well as the
selectivity, were calculated as reported previously [13].
The total conversion was defined as the oxygen consump-
tion,
2
2
2
and 1% O in H (PROX).
2
2
All measurement series were carried out on ca. 100 mg
of fresh sample previously pretreated in situ in flowing air
(30 mL/min) at 573 K. The catalyst was purged in N while
2
cooling to the desired reaction temperature of 383 or 523 K.
A spectrum of the activated sample before adsorption was col-
lected in N2 at the reaction temperature, and then the reaction
mixture—premixed in a bypass—was introduced to the cata-
lyst in one step. Spectra were collected as a function of time
on stream for 90 min in all cases. Only spectra taken under the
steady-state conditions are shown herein.
Gas composition was analyzed only in the case of PROX re-
action by a Pfeiffer Thermostar mass spectrometer. Selectivity
and activity values were calculated using the same formula as
presented earlier.
in
out
O2
n
− n
O2
O2
Xtotal = XO =
× 100 (%).
(1)
2
in
n
CO conversion, considering no side-reactions, can be defined
as
in
CO
out
CO
out
n
− n
n
CO2
XCO =
× 100 =
× 100 (%).
(2)
in
CO
in
n
n
CO
Selectivity is defined as the ratio of the oxygen transformed into
CO2 to the total oxygen consumed as
out
n
2.5. High-pressure X-ray photoelectron spectroscopy
CO2
S =
× 100 (%).
)
(3)
in
out
2
(n − nO2
O2
The in situ XPS experiments were performed at beamline
U49/2-PGM2 at BESSY in Berlin. Details about the setup were
published [45–47]. Briefly, the differentially pumped electro-
static lens system is the key feature of the system, allowing
investigation of the sample in the mbar pressure region.
The 5% Pt/CeO2 sample was chosen for these measure-
ments to ensure a reliable signal-to-noise ratio in the Pt 4f re-
gion. A pressed pellet containing about 100 mg of 5% Pt/CeO2
catalyst was placed on a temperature-controlled heater, and
was then activated in situ in oxygen (0.5 mbar, 573 K). Gas
flow (∼20 N mL/min) into the reaction cell was controlled
using calibrated mass flow controllers and leak valves. The
PROX mixture contained 0.48 mbar H2, 0.032 mbar CO, and
The relation between conversion and selectivity is defined as
SXO2
XCO =
λ (%),
(4)
1
00
where λ is the oxygen excess factor, by definition,
in
in
in
2n
2c
2
pO2
2[O2]
in
O2
O2
λ =
=
=
=
c
CO
.
(5)
in
CO
in
n
pCO
[CO]
2
.3. Temperature-programmed desorption
Temperature-programmed desorption (TDS) measurements
were performed in a standard UHV apparatus equipped with
an atmospheric chamber. The atmospheric part was used to
introduce the sample for pretreatments and adsorbing gases
at non-UHV pressures. The 5% Pt/CeO2 was investigated by
TDS. Different gases (gas mixtures) were adsorbed at room
0
.015 mbar O2. Gas-phase analysis was carried out using a
quadrupole Balzers mass spectrometer connected through a
leak valve to the experimental cell. Selectivity and activity val-
ues were calculated using the same formulas as presented ear-
lier.
−
2
temperature on the sample: CO alone (10 mbar for 30 min),
−2
−2
Ce 3d, O 1s, C 1s, and Pt 4f spectra were recorded with pho-
ton energies of hν = 1035, 650 (and 920), 670, and 460 eV,
respectively. The binding energies (BEs) were calibrated, if
possible, against the Fermi level of the sample or using internal
CO + O2 (10 mbar O2 + 2 × 10 mbar CO for 20 min),
−
1
−2
and H2 + CO + O2 (4.7 × 10 mbar H2 + 2 × 10 mbar
−2
CO + 10 mbar O2 for 20 min). After adsorption, the sample
was evacuated and transferred to the UHV chamber to follow
the desorption pattern. The heating rate was 1 K/s. A Balzers
mass spectrometer was used to follow the evolution of desorb-
ing gases.
ꢀ
ꢀꢀ
references, such as the Ce 3d V (882.4 eV) and U (916.7 eV)
hybridization states [48] or the Ce 4f state in the band gap. The
energy calibration was necessary because surface charging oc-
curred due to the emission of photoelectrons. In the presence
of hydrogen (i.e., in most cases) this was only 1–3 eV, in O2,
however, charging up to 20 eV was observed. Decomposition of
the Pt 4f and C 1s regions was performed using Gauss–Lorentz
curves. The metallic Pt 4f component at 71 eV was fitted using
Gauss–Lorentz profiles with an exponential tail.
2
.4. In situ diffuse reflectance infrared spectroscopy
A diffuse reflection attachment “Selector” from Graseby
Specac was placed in a Bruker FTIR spectrometer equipped
with a D315M MCT detector to collect diffuse reflectance in-
frared (DRIFT) spectra. The spectrometer was purged with pu-
rified air. Reactions were conducted in a gold cup (2.5 mm high,
2.6. High-resolution transmission electron microscopy
8
.5 mm o.d., 7.2 mm i.d.) placed in a Graseby Specac “envi-
ronmental chamber” with a ZnSe window. A spectrum of KBr
recorded in N2 served as background. Inlet gases were analyti-
High-resolution transmission electron microscopy (HRTEM)
investigations were performed in a Philips CM200 FEG elec-

4
O. Pozdnyakova et al. / Journal of Catalysis 237 (2006) 1–16
tron microscope operated at 200 keV. The microscope was
equipped with a Gatan imaging filter (GIF 100), Gatan Slow-
Scan Camera, and EDX system. The sample was prepared from
a piece of the 5% Pt/CeO2 pellet that was previously activated
in O2 in the in situ XPS cell as was done for the in situ XPS
experiments. Another sample of 5% Pt/CeO2 was also investi-
gated after PROX reaction at 358 K (from the XPS cell).
DRIFTS results are shown mainly on the 1% Pt/CeO2. Both
catalysts were tested in the catalytic reactor, and 1% Pt showed
slightly better performance in the PROX reaction; however, this
difference was negligible (Table 1).
Activity patterns measured in different setups showed sim-
ilar trends, albeit with different actual values (Table 1). In all
cases, increasing temperature decreased the selectivity toward
CO oxidation, and the lower O conversion was, as a rule,
2
3
. Results
accompanied by higher selectivity toward CO oxidation. The
differences in activity pattern can be attributed to the different
contact times due to the varying dead volume and volume flow
rates in the different experimental setups. In the flow reactor
system, the sample covers the whole cross-section of the reac-
tor, whereas in the DRIFTS experiments, the gas mixture passes
3
.1. Catalytic reaction
Fig. 1 compares the CO oxidation activity on 1% Pt/CeO2
in the absence and presence of hydrogen. A sharp “light off”
activity around 365 K was observed at high oxygen excess with-
out hydrogen, which became less pronounced with decreasing
oxygen excess (Fig. 1a), in accord with earlier observations
3
over the catalyst powder in a ca. 100-cm cell. In the case of
high-pressure XPS, the 10-mm-diameter pellet was placed in a
ca. 8-L chamber, and the total pressure was only 0.5 mbar, as
opposed to the atmospheric pressure used in other setups.
[
(
19–21]. In the presence of hydrogen, the oxygen conversion
not shown) was very high (ca. 98–100%), whereas CO conver-
3
.2. Temperature-programmed desorption (TDS)
sion decreased as a function of temperature [13]. The oxygen
was consumed for the production of water. Nevertheless, the
natural selectivity of Pt toward CO oxidation as opposed to H2
oxidation, manifested as lower λ values, resulted in higher se-
lectivity (Fig. 1c).
The catalytic data observed in different experimental setups
are compared in Table 1. Unfortunately, the 1% Pt/CeO2 sample
could not be measured by XPS because of the low Pt content. To
receive well-resolved Pt spectra in the high-pressure XPS appa-
ratus, the 5% Pt/CeO2 sample was prepared and measured here.
Fig. 2 shows the desorption curves of CO2 and water from
the 5% loading catalyst after adsorbing different gases or gas
mixtures. Formation of CO2 after adsorption of CO alone
clearly occurred, but its amount was an order of magnitude
lower than the amount desorbed after oxygen-containing mix-
tures. Interestingly, more CO2 was produced in the presence
of hydrogen, particularly around 358 K. Water desorption was
very strongly enhanced after adsorption of PROX mixture espe-
Fig. 1. CO oxidation activity in the absence (a) and presence of hydrogen, PROX (b) as well as selectivity towards CO oxidation in the PROX reaction (c) on 1%
Pt/CeO catalyst as a function of temperature at different oxygen excess λ. The oxygen conversion was almost total in the PROX reaction under all conditions.
2
Table 1
PROX reaction in different experimental setups and on different catalysts. Reaction mixtures contained 1% CO, 0.5 or 1% O (O excess λ = 1 or 2) in H . Total
2
2
2
pressure: p = 1 atm in catalytic reactor and DRIFTS cell and p = 0.5 mbar in high pressure XPS
Catalytic reactor
DRIFTS
XPS
1% Pt/CeO
5% Pt/CeO
1% Pt/CeO
5% Pt/CeO
2
2
2
2
O excess (λ)
2
1
1
2
2
1
1
2
2
2
2
1
1
T (K)
383
523
383
523
383
523
383
523
383
523
358
523
X
(%)
65
66
99
38
38
100
83
42
99
47
24
100
59
60
98
41
41
100
84
43
99
42
21
100
44
30
73
22
14
81
9
83
12
6
18
38
CO
S (%)
(%)
X
O2

O. Pozdnyakova et al. / Journal of Catalysis 237 (2006) 1–16
5
Fig. 2. Temperature-programmed desorption (TDS) curves of CO (a) and water (b) from the 5% loading catalyst after adsorbing different gas mixtures at room
2
−
2
−2
−2
−1
temperature: CO alone corresponds to 10 mbar CO for 30 min, CO + O to 10 mbar O +2×10 mbar CO for 20 min, and PROX mixture 4.7×10 mbar
2
2
−
2
−2
H + 2 × 10 mbar CO + 10 mbar O for 20 min. (Intensities are shown as measured, without calibration.)
2
2
Fig. 3. DRIFT spectra of 1 and 5% Pt/CeO at T = 383 K (a) after activation in N and (b) in the presence of PROX reaction mixture (1% CO, 1% O in H ).
2
2
2
2
cially up to 500 K. Its amount (after calibration) was an order
of magnitude higher than expected considering the selectivity
values given in Table 1.
shown in Fig. 4. The bands in this region are summarized in
Table 2. Adsorption of CO on Pt/CeO at 383 K in the ab-
2
sence of both oxygen and hydrogen produced one main band
−
1
at 2063 cm , with significant broadening at the low-frequency
3
.3. In situ DRIFTS
−1
side, and shoulders at about 2095 and 2083 cm (Fig. 4a). The
strong band at 2063 cm and the shoulder at about 2083 cm
−
1
−1
DRIFT spectra of 1 and 5% Pt/CeO2 samples recorded after
activation in oxygen as well as in different reaction environ-
could be attributed to linear CO species adsorbed on a step-
0
site and terrace Pt atoms, respectively (Table 2) [15,32,34,35].
ments (CO alone, CO+O , and PROX mixture) exhibited simi-
2
−1
The shoulder at ∼2095 cm is due to CO adsorbed on a partly
lar features irrespective of the metal loading (Fig. 3). Therefore,
oxidized Pt site in the vicinity of reduced Pt [32,34,35]. The
the discussion from now on is limited to the 1% Pt/CeO sam-
2
−
1
broad band at ca. 1970–1960 cm recalls the feature observed
ple. For clarity, the spectral features observed for four charac-
teristic vibrational regions are presented separately (Figs. 4–7),
and the characteristic bands corresponding to different surface
species are summarized in Tables 2–4.
−
1
at 2000–1960 cm on Pt/Al2O3 and Pt/CeO2–SiO2 catalysts
and is assigned to CO adsorption at the metal–support inter-
face [49–52]. Kappers et al. [53] attributed this downshift of the
ν(CO) band as an indicator of ion–dipole interaction between
O atom of the CO species with charged cations from the sup-
port. They showed that adsorbed water can preferentially shield
this interaction. A weak ν(CO) band characteristic of bridge-
−
1
3
.3.1. DRIFT spectra in the region 2200–1700 cm
The DRIFT spectra of CO adsorbed on the 1% Pt/CeO2 un-
−
1
der different conditions in the region of 2200–1700 cm are

6
O. Pozdnyakova et al. / Journal of Catalysis 237 (2006) 1–16
−
1
Fig. 4. DRIFT spectra in the 2200–1700 cm region (CO vibrations) of 1% Pt/CeO in the presence of (a) 1% CO alone in N , (b) in 1% CO and 1% O in N ,
2
2
2
2
and (c) in 1% CO and 1% O in H (PROX) at 383 and 523 K, respectively.
2
2
Table 2
IR bands on ceria supported Pt in the 2200–1800 cm region
−
1
−
1
Name
Wavenumbers (cm
)
In N
CO alone
CO + O2
PROX
2
Literature values
Our values 383 K 523 K 383 K 523 K 383 K 523 K 383 K 523 K
4
3
+
+
CO adsorbed on Ce
CO adsorbed on Ce
2177, 2156 [37],
2172–2154
−
−
+
+
+
+
low
+
2170–2127 [54]
2120–2127 [54]
2122–2131 [35]
2124
2121–2115
2095
−
−
−
−
−
−
−
−
−
−
−
−
−
−
Always present, their distinction is difficult
CO adsorbed on Pt(ox)
CO adsorbed on metallic Pt with Pt(ox) neighbor 2091–2096 [34,35]
CO linearly adsorbed on terraces
CO linearly adsorbed on step sites
CO on metal/support interface site
Bridged CO on Pt
+
+
+
low
+
+
+
+
low
+
low
+
low
+
2080–2085 [14,15,34,35] 2083–2075
2072–2065 [14,15,34,35] 2068–2041
1960–1970 [33,49–52]
1850 [14,33]
+
+
+
+
1960–1970
1877–1882
Always present, more at high T
Only small amount at low T
bonded CO species was resolved at 1880 cm 1. An additional
−
In the presence of oxygen in the gas mixture (Fig. 4b), only
moderate changes occurred compared with the CO alone case
(Fig. 4a); most of the ν(CO) bands were slightly blue-shifted
−
1
weak ν(CO) band observed at ca. 2124 cm and a doublet
−
1
encompassing broad bands at 2172 and 2157 cm have been
ascribed to CO linearly adsorbed on reduced and unreduced ce-
ria sites, respectively [37,54]. However, in our case these bands
overlap with ν(CO) of the gaseous carbon monoxide, and only
at ca. 2124 cm there is definitely a contribution from ad-
sorbed CO, because the P branch (of gas-phase CO) is weaker
−
1
−1
(∼5 cm ), with the intensity of the shoulder at 2095 cm
−1
relatively increased and that of the shoulder at 2083 cm de-
creased. These changes can be explained by slight oxidation
(probably hindered reduction) of the Pt particles.
−1
Under PROX conditions (Fig. 4c), a general shift to lower
wavenumbers was observed, with the main band shifted from
−
1
than the R branch (above 2143 cm ).
−
1
−1
.
2
063 to 2054 cm and the shoulder positioned at 2075 cm
Increasing temperature led to the following: (i) a decrease in
intensity of the main band and of the shoulder at 2083 cm and
a relative increase in intensity of the component at ca. 1970–
960 cm ; (ii) a shift of the main band to lower wavenumbers
from 2063 to 2051 cm ); (iii) a relative increase in the inten-
sity of the ν(CO) band on Ce sites compared to that one on
Ce sites; and (iv) a discernable feature at 2112 cm . The
broad band at 2112 cm is due to CO linearly adsorbed both
on Pt atoms interacting with oxygen (i.e., Pt atoms in a more un-
saturated coordination state (Pt ) [34,35]) and on Ce sites,
respectively.
−
1
Increasing temperature in the presence of hydrogen resulted in
−1
a further shift of these ν(CO) bands to 2041 and 2068 cm
,
−1
respectively, and a relative increase in intensity of the broad
1
(
−
1
−
1
ν(CO) band at 1965 cm compared with that at T = 383 K.
3+
4+
−1
−1
3.3.2. DRIFT spectra in the 1800–1000 cm region
−1
The DRIFT spectra of the 1% Pt/CeO sample under dif-
2
−
1
ferent conditions in the 1800–1000 cm region are shown in
Fig. 5. In this region different carbonate, carboxylate and for-
mate species gave several bands; their assignments are summa-
δ+
3+

O. Pozdnyakova et al. / Journal of Catalysis 237 (2006) 1–16
7
−
1
Fig. 5. DRIFT spectra in the 1800–1000 cm region (OCO vibrations) of 1% Pt/CeO at (a) T = 383 K and (b) T = 523 K after activation, in the presence of 1%
2
CO alone in N , in 1% CO and 1% O in N , and in 1% CO and 1% O in H (PROX), respectively.
2
2
2
2
2
Table 3
IR bands on ceria supported Pt for carbonate and carboxylate species. Bands in italic are OH vibrations
−
1
Name
Formula
Wavenumbers (cm
Literature values
)
In N
CO alone
CO + O2
PROX
2
Our values
383 K 523 K 383 K 523 K 383 K 523 K 383 K 523 K
Carbonate unidentate
Carbonate bidentate
1545, 1348, 1062 [36]
1562, 1286, 1028 [36]
1464, 1358, 1085
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
1565, 1298, 1014–1009
Carbonate polydentate Polymer
Bridged carbonate
1462, 1353, 1066 [33]
1463, 1353, 1050–1040
1740, 1393, 1232, 1145
+
+
+
−
+
+
+
−
+
+
+
−
+
−
+
−
1728, 1396, 1219, 1132 [36]
Bicarbonate
3617, 1613, 1391, 1218, 1045 [33] 3619, 1611, 1393, 1217, 1043 −
−
+
−
+
−
−
−
Carboxylate
1560, 1510, 1310 [36]
1514, 1316
+
−
+?
?
+
+
?
+
+
−
+
−
+
Carboxylic acid
1670–1695, 1338–1310 [35,36]
3590, 1700–1600, 1335
−
+
+
rized in Tables 3 and 4. The identification of carboxylic acids,
bicarbonates and formate species were also based on their other
characteristic O–H (Fig. 6) or C–H bands (Fig. 7), respectively.
in N2 (Fig. 5a). Increasing temperature resulted in a decrease of
the bicarbonate and slight increase of the bidentate carbonate
species on the surface of the activated sample (Fig. 5b).
Exposure to 10 mbar of CO alone or in the presence of
oxygen yielded bands that were grouped as follows: (i) strong
−
1
Unidentate carbonate (strong bands at 1464 and 1358 cm
−
1
and a weak band at 1080 cm ), bidentate carbonate (features
−
1
−1
at ca. 1565, 1300, and 1013 cm ), and traces of bicarbonate
medium bands at 1393 and 1232, a weak band at ca. 1050,
and a shoulder at ca. 1611 cm ) and carboxylate (feature at
bands at 1393 and 1217 cm and weak ones at 1611 and 1043
−
1
−1
(
cm ; (ii) a strong broad band at 1298 cm and weak bands at
−
1
−1
−1
1565 and ∼1010 cm ; and (iii) a medium band at 1545 cm
−
1
−1
ca. 1514 and a very weak band at 1316 cm ) species were
and weak bands at 1372 and 1360 cm . The first group of
bands was assigned to bicarbonate species, which correlates
present on the surface of the O2 pretreated sample at T = 383 K

8
O. Pozdnyakova et al. / Journal of Catalysis 237 (2006) 1–16
Table 4
IR bands on ceria supported Pt for formate and –OH species. Bands in italic are OH or CH vibrations
−
1
Name
Formula
Wavenumbers (cm
Literature values
)
In N
CO alone
CO + O2
PROX
2
Our values
383 K 523 K 383 K 523 K 383 K 523 K 383 K 523 K
2
2
945, 2852, 1558, 1369, 1329 [37] 2936, 2848, 1545,
Bidentate formate
−
−
+
+
−
+
+
+
845, 1547, 1358 [55]
1372, 1360
2
2
933, 2852, 1575, 1358 [55],
945, 1587, 1329 [37]
2950, 2844, 1582,
1372, 1330
Bridged formate
Type I OH
−
+
−
+
+
+
+
+
−
+
+
+
+
+
+
Ce–OH
3710 [33,59]
650 [59]
3680–3660 [33,
8]
3708–3700
?
3
Type II-A OH
3658–3667
+
+
+
+
+
+
?
?
+
5
Type II-B OH
3651 [33,58]
3619–3625
3517
+?
+
+?
+
−
+
+
Type III OH
3500 [59], 3600 [33,58]
+
+
+
+
+
+
+
+
−
1
Fig. 6. DRIFT spectra in the 3900–3300 cm region (OH vibrations) of 1% Pt/CeO at (a) T = 383 K and (b) T = 523 K after activation, in the presence of 1%
2
CO alone in N , in 1% CO and 1% O in N , and in 1% CO and 1% O in H (PROX), respectively.
2
2
2
2
2
−
1
well with the appearance of strong ν(OH) band at 3619 cm
on partially reduced ceria through reaction between CO and sur-
face hydroxyl groups [37].
and concomitant decrease of another OH band (Fig. 6) observed
on exposure of sample to CO or CO+O2 gas mixtures. The sec-
ond group of bands belongs to bidentate carbonate species. The
relative intensities of the bands of different carbonate species
were not affected by the presence of oxygen in the gas mix-
ture, except for bands of bicarbonate, which were enhanced in
CO + O2 atmosphere. The third group of bands appearing only
in the presence of CO alone, complemented by the appearance
of ν(CH) bands (Fig. 7), belongs to bidentate formates formed
Under PROX conditions, bicarbonate species observed in
H2-free gas mixtures did not form (Fig. 5). The ν(OCO) bands
of bidentate formate species became poorly resolved, with over-
all decreased intensities. At the same time, a strong broad fea-
−
1
ture developed at ca. 1582 cm , which may be attributed to the
νas(OCO) band of bridged formate species [55]. The significant
−
1
broadening above 1611 cm corresponds to surface water (see
later). The ν(CO3) bands of bidentate and unidentate carbon-

O. Pozdnyakova et al. / Journal of Catalysis 237 (2006) 1–16
9
−
1
Fig. 7. DRIFT spectra in the 3000–2650 cm region (CH vibrations) of 1% Pt/CeO at (a) T = 383 K and (b) T = 523 K after activation, in the presence of 1%
2
CO alone in N , in 1% CO and 1% O in N , and in 1% CO and 1% O in H (PROX), respectively.
2
2
2
2
2
ate species of similar intensities were present in the spectra in
H2-rich atmosphere.
of mono-coordinated (OH-I) and doubly bridging (OH-II) hy-
droxyl groups on the ceria surface was observed (Fig. 6b).
The presence of CO alone or O2 + CO at T = 383 K
(Fig. 6a) resulted in the appearance of ν(OH) bands at 3619
At T = 523 K, the presence of CO alone and CO + O2 gas
−
1
mixtures gave a strong band at 1582 cm and weak bands at
−
1
−1
and 3590 cm , along with a decreased intensity of the ν(OH-
1
372, 1540, and 1360 cm (Fig. 5b), accompanied by the si-
multaneous appearance of a strong ν(CH) band at 2850 cm
−
1
−1
I) band at 3704 cm . The development of a strong band at
−
1
−1
(Fig. 7). These features indicate that bridged formate and prob-
3619 cm and decaying of the band at 3704 cm were at-
−
ably traces of bidentate formate species are present on the sur-
tributed to the formation of HCO3 species through the re-
action of CO with terminal (OH-I) hydroxyl groups. (Accord-
ingly, bicarbonate vibrations were found in Fig. 5, as discussed
face. Under PROX conditions at T = 523 K, bands belonging
−
1
to bridged formate species (1582, 1372, and 1330 cm ) were
significantly higher than those at T = 383 K in the presence of
H2 or at T = 523 K in the absence of H2.
−1
earlier.) The other developed ν(OH) band at 3590 cm may
reflect the presence of carboxylic acid (–COOH) groups on the
surface of the ceria support, because it correlates with the weak
broad features observed under the same conditions at ca. 1690,
Ill-defined bands due to unidentate and/or polydentate [56]
and bidentate carbonate species were also observed at T =
−
1
1
510–1506, and 1330 cm (Fig. 5). Decreasing ν(OH) bands
5
1
23 K. Under PROX conditions, the poorly resolved bands at
335, 1690, and 1510–1506 cm (Fig. 5) reflect the presence
−
1
can be correlated with the development of formate species, as
described later.
of protonated carboxylate species, carboxylic acid (–COOH),
The presence of oxygen had little effect on the position and
relative intensities of the ν(OH) bands, although the intensity of
on the surface, as also indicated by the corresponding ν(OH)
−
1
3
590 cm (Fig. 6).
−
1
the band at 3704 cm was enhanced in the presence of oxygen
compared with that in a CO-alone gas mixture (Fig. 6a). After
CO adsorption at T = 523 K, no distinct ν(OH) band of bicar-
bonate species was resolved, whereas ν(OH) stretching of all
−
1
3
.3.3. DRIFT spectra in the 3900–3300 cm region
Four OH groups were reproduced after oxidative activation
−1
on the 1% Pt/CeO2 sample at T = 383 K (Fig. 6a). They are
denoted as type I, II, or III according to the number of cerium
cations constituting the adsorption site [57]. The sharp peak at
surface hydroxyl groups were downshifted (∼10 cm ) com-
pared with that at T = 383 K.
Under PROX conditions at T = 383 K, isolated bands be-
come convoluted into intense and broad absorption features,
among which those centered at ca. 3689, 3658 (OH-IIA), and
3590 (carboxylic acid) could be somewhat resolved. Decreased
resolution, a general broadening of stretching ν(OH), and a
strong increase in absorption toward lower wavenumbers in-
−
1
3
704 cm was assigned to mono-coordinated OH (type I); the
−1
−1
band at 3651 cm with the shoulder at ca. 3625 cm was as-
signed to two types of bridging OH (types II-A and II-B); and a
broad band centered at 3517 cm was assigned to triply bridg-
ing OH (III) species (Table 4) [58,59]. At T = 523 K, a slight
downshift of all ν(OH) bands and a decrease in the number
−
1
−1
dicate considerable H-bonding. A band at ca. 1640 cm and

10
O. Pozdnyakova et al. / Journal of Catalysis 237 (2006) 1–16
the observation of H2O as a product are further evidence of
the presence of significant amounts of water on the surface.
H bonding with the Ce–OH groups suggests that the water is
adsorbed predominantly on the support. Increasing tempera-
ture under PROX conditions decreased the intensity of features
of adsorbed water and increased the intensity of the bands at
assigned to the ν(CH) band of formyl species or may originate
from other CH-containing species.
3.4. High-pressure XPS
The 5% Pt/CeO2 sample was investigated in different
gaseous environment using our novel high-pressure XPS setup
−1
−1
3
625 cm (OH-IIB) and 3590 cm .
[
46,47]. The sample was first activated in 0.5 mbar O2 at 573
−
1
3
.3.4. DRIFT spectra in the 3000–2650 cm region
K, then cooled in O2. At ∼340 K, oxygen was evacuated and
replaced by hydrogen (0.48 mbar). As the ambient temperature
was reached, first CO (0.032 mbar) and then O2 (0.015 mbar)
were added to the hydrogen. The surface state of the sample
was investigated in these different environments. The catalytic
activity observed during high-pressure XPS measurements is
characterized in Table 1.
The main bands in this vibration region can be assigned
to bridged and bidentate formate species as follows (Fig. 7):
848 cm as ν(CH), 2936–2950 cm as a combination of
νas(OCO) and the C–H in-plane vibration, and 2719 cm as
overtone 2δ(C–H), which at around δ(C–H) = 1360 cm is
indeed hardly visible because of the strong carbonate vibra-
tions (Fig. 5). The C–H vibrations of the two different for-
mate species, bidentate and bridged, are quite similar; how-
−1 −1
2
−
1
−
1
Fig. 8 depicts a part of the Ce 3d region in oxygen at
573 K, in hydrogen at 300 K, and in the PROX mixture at
358 K. As expected, the surface of ceria was more or less
completely oxidized in oxygen, whereas hydrogen induced the
−
1
ever, the difference in νas(OCO)— ∼1545 cm for bidentate
−1
and ∼1582 cm for bridged (Table 4) [37,55]—makes their
distinction possible. Consequently, the combination mode of
νas(OCO) and the C–H in-plane vibration should also be higher
for the bridged species than for the bidentate species [60].
After oxidative treatment of the Pt/CeO2 sample, no bands
formation of Ce3 species. The peaks V at ∼880.7 eV and V
+
0
ꢀ
3+
at ∼885.5 eV are characteristic for Ce species, whereas V
ꢀ
ꢀ
(∼882.4 eV) and V (∼888.7 eV) correspond to oxidized ce-
ria. (The same notation is used as established by Burroughs et
al. [48] and popularly applied in other papers as well [38,39].)
The use of 1035 eV synchrotron radiation ensures—due to the
low KE electrons (∼140 eV in this part of the spectrum)—that
mainly the topmost few (two or three) Ce layers were probed.
The reduction degree of ceria in hydrogen was estimated to be
−
1
were observed in the 3000–2600 cm region, whereas in CO
−
1
stream, strong bands at 2936 and 2848 cm and weak features
−
1
at ca. 2910, 2719, and ca. 2699 cm appeared (Fig. 7a). The
main bands were assigned to surface formate species [37,61,
2], mainly to bidentate formate (see above). It was proposed
6
that surface formate species on ceria were formed through the
reaction of CO with terminal (OH-I) [55] and/or bridging (OH-
II) [37,63] hydroxyl groups; these OH bands decreased accord-
ingly, as discussed in the preceding section. The spectra under
conditions of CO + O2 exhibited no distinct bands in the 3000–
−
1
2
600 cm region. In the PROX gas mixture, at T = 383 K,
because of the strong “hydrogen-bonded OH–water structure,”
the background increased as a function of wavenumbers, and
thus the bands were poorly resolved. Nevertheless, a ν(CH)
−1
band at 2844 cm was observed, accompanied by weak broad
−
1
features at ca. 2716 and 2952 cm . These bands might re-
flect the presence of both bidentate and bridged formate species
on the surface [the latter probably being in excess, see corre-
sponding νas(OCO) (Fig. 6)] or imply that the broad band at
−
1
2
844 cm probably involves some contribution from ν(CH)
vibrations, which are not associated with the formate species
(
hydrocarbonaceous residue).
At T = 523 K, under conditions of CO alone and CO +
−
1
O2, the position of the main ν(CH) band at 2848 cm was
hardly affected. However, the ratio of bridged and bidentate
formate species seemed to change; more bridged formate and
less bidentate formate could be estimated than at low tempera-
ture, because of the relative increase in intensity of the band at
−
1
2
2
950 cm and significant decrease in intensity of the band at
−
1
936 cm , in accordance with the position of the correspond-
Fig. 8. Part of the Ce 3d region of the 5% Pt/CeO at different conditions: 1,
2
ing νas(OCO) bands (see earlier). The presence of hydrogen in
the gas mixture at elevated temperature resulted in a significant
increase in intensity of the bands belonging to bridged formate
species and the appearance of several weak bands that may be
in 0.5 mbar O at 573 K; 2, in 0.48 mbar H at RT; 3, in ∼0.5 mbar PROX
2
2
mixture at 358 K; 4, in ∼0.5 mbar PROX mixture at 523 K; 5, in a mixture of
0.032 mbar CO and 0.015 mbar O (as hydrogen was switched off after 4). The
2
measurements were carried out in sequence as indicated by the numbers.

O. Pozdnyakova et al. / Journal of Catalysis 237 (2006) 1–16
11
Fig. 9. (a) Pt 4f region of 5% Pt/CeO at different conditions: 1, in 0.48 mbar H at RT (after O activation at 573 K); 2, in ∼0.5 mbar PROX mixture at 358 K;
2
2
2
3
, in ∼0.5 mbar PROX mixture at 523 K; (b) 1, in 0.48 mbar H at 573 K; 2, in ∼0.5 mbar PROX mixture at 358 K; 3, in ∼0.5 mbar PROX mixture at 523 K. The
2
measurements were carried out in sequence as indicated by the numbers.
roughly 50%. Assuming that the reduction at low temperature
affects just the top layer, this layer should be fully reduced to
fraction of this Pt species is too high to be explainable solely by
surface Pt, because the spectra were taken with 460-eV photons,
which are insufficiently surface-sensitive to account for such an
intense component. Possible explanations could be that some
special CO-induced compound-like platinum atoms give rise to
this Pt 4f peak, or that the particles are extremely small and ad-
ditional surface reconstruction occurs. Bridged CO cannot be
resolved from our spectrum. On further heating of the sample to
523 K, the new peaks disappeared, and the spectrum closely re-
sembled that measured after activation. The Pt-oxidized1 state
decreased slightly in the reaction mixture.
In the second experiment, the sample after activation was
reduced in hydrogen at 573 K and tested in the PROX reac-
tion. The sample’s catalytic behavior was comparable to that of
nonreduced sample, showing similar activity at low T and se-
lectivity decay at 523 K. The corresponding spectra are shown
in Fig. 9b. During the 573 K hydrogen treatment, platinum was
totally reduced and could be fitted with the typical asymmet-
ric function. In the PROX reaction mixture at low T , all the
aforementioned components appeared, albeit at smaller relative
intensities. This indicates that the Pt-oxidized1 species can be
formed in the reaction mixture already at 358 K (or lower).
Considering the high excess of reducing gases (H2 and CO),
the identification of this component as “bulk” PtO is not likely.
It is more probable that, under less severe reducing conditions,
the interface atoms are polarized by the negatively charged an-
ions of the supporting ceria. If so, then the Pt particles should
be very small (a few nm or even smaller).
3+
Ce . In the PROX mixture, the surface was clearly reoxidized,
however. (Note that the gas mixture contained just 3% oxygen,
whereas the other 97% had a reducing character.) This reoxi-
dation was slightly more pronounced at 523 K, indicating that
oxygen has a very strong sticking probability on ceria surfaces
ꢀ
ꢀ
even at higher T . The ratio of the V to the V hybridization
states changed in the PROX reaction mixture, although the sur-
face was oxidized more or less similarly as in pure O2.
Fig. 9a shows the platinum 4f core level in hydrogen and
in the reaction mixture at two different temperatures. In oxy-
gen, charging of about 17 eV made the platinum peak (not
shown) too broad to allow evaluation of its form. But introduc-
ing hydrogen at close to room temperature may induce hardly
any changes in the valence state of platinum; thus we consider
this spectrum representative for the oxygen-activated sample.
The spectrum indicates two distinct components, one in the
zero-valent state and the other with a BE shift of +1.6 eV. Iden-
tification of this second component is controversial according to
the literature. Several studies found a Pt2 state (e.g., PtO) in
a +1–1.8 eV distance relative to metallic Pt [64,65]; however,
other studies [66,67] reported PtO in a +2.4–3 eV distance and
PtO2 at +3.6–4.5 eV [66,68]. Different surface–core level shift
+
(
SCLS) values were observed for adsorbed O2 on Pt single crys-
tals [67,69]. At this point, without arguing in favor of any of
these possibilities, we call this component “Pt-oxidized1.” At
3
58 K under PROX conditions, the spectrum changed appre-
ciably; this component was no longer resolved, and a shoulder
at 72.0 eV appeared. The 7/2-to-5/2 splitting is still not well
separated (intensity at ∼73.5 eV); therefore, the Pt-oxidized1
component should not disappear completely. Moreover, a tail
at around 77.2 eV is now clear, indicating a nonresolvable 7/2
component at ∼73.9 eV. We call this “Pt-oxidized2.” According
to the literature [69], SCLS of adsorbed CO gives BE shifts of
Carbon 1s spectra of the activated sample in the reaction
mixture at low and high temperatures are shown in Fig. 10.
Because of variations in the form of the spectra, we used sev-
eral components (actually, seven are shown) to adequately fit
the measured curves. But unambiguous assignment is hardly
possible, because many different species can have very simi-
lar BEs [70]. As a first approximation, we consider the lowest
BE peak (284.2 eV) as graphene or graphite, the peaks around
286 eV as adsorbed CO in bridged (285.6 eV) and in linear
(286.3 eV) coordination to Pt, and the other high-BE peaks
∼
+1.0 and +0.4 eV, identified as on-top and bridged bonded
CO, respectively. Therefore, we may identify the 72.0 eV com-
ponent as Pt with linearly bonded CO. On the other hand, the

12
O. Pozdnyakova et al. / Journal of Catalysis 237 (2006) 1–16
characterize the microstructure of the sample and to try to iden-
tify oxidized Pt (see above). The ceria particles (8–10 nm) were
strongly stacked together despite the careful sample prepara-
tion, allowing no phase identification in most of the area of
the specimen. Furthermore, depicting platinum particles was
widely hindered by the strong contrast of the ceria lattice.
Therefore, very few Pt particles could be found. A selected
image is shown in Fig. 11a, in which accidentally two Pt par-
ticles can be seen. The particle about 2 × 3 nm in size is [110]
oriented, as can be deduced from its power spectrum (see the
inset). [The power spectrum is the square of the modulus of the
Fourier transform of the two-dimensional image (or a chosen
part of the image) and represents the (computer-generated) dif-
fraction pattern of the image.] The contrast of the particle was
slightly enhanced by filtering the corresponding spots (inverse
lattice fringes) of the power spectrum, back-transforming, and
adding it to the original image. Fig. 11b shows another part of
the sample with unidentified dark spots on the surface of ce-
ria. These spots did not behave like the dark and light spots in
the carbon foil (flickering), and thus they do not belong to the
foil, but may be very small (5–6 Å) Pt clusters. If so, then they
should be partially oxidized through the relatively strong inter-
action with ceria. Oxidized “bulk” platinum could not be found
in any form on the images, however. Therefore, we identify the
Fig. 10. C 1s region of 5% Pt/CeO at different conditions: 1, in ∼0.5 mbar
2
PROX mixture at 358 K; 2, in ∼0.5 mbar PROX mixture at 523 K.
(
except that of 290.6 eV, which is gas-phase CO) as different
oxidized carbon on Pt, but mainly on the ceria surface. Be-
cause only a negligible amount of bridged CO (i.e., bonded
between two Pt atoms) was detected by DRIFTS, the species at
2
85.6 may belong to CO bonded to the Pt–ceria interface (CO
−1
band at ∼1960 cm ). But this interface CO can be considered
7
2.6 eV Pt 4f component as small Pt clusters (or interface Pt
atoms) strongly interacting with ceria. Most of the ceria parti-
cles were in the form of CeO (fluorite structure) with [110],
bridged adsorbed CO as well, considering that C is coordinat-
ing to Pt and O is interacting with a coordinatively unsaturated
2
n+
(
cus) Ce site.
[
111], and [001] orientation. In some cases a small lattice ex-
pansion and a stronger angular distortion could be calculated
mainly because of some defective structure. In addition, EDX
measurements were carried out to roughly characterize the bulk
Pt content of the sample. Using an area of 300 nm in diame-
ter, the calculated Pt content was ∼4 wt%. Using several small
3
.5. High-resolution transmission electron microscopy
High-resolution transmission electron microscopy (HRTEM)
measurements were performed on the activated 5% Pt/CeO2 to
Fig. 11. (a) High-resolution TEM image of 5% Pt/CeO after activation in oxygen at 573 K. The inset shows the power spectrum of the platinum particle situated in
2
the middle of the TEM image. (b) Other section of the sample showing very small dark spots (indicated by circles) that might correspond to Pt clusters in the size
of 5–6 Å.

O. Pozdnyakova et al. / Journal of Catalysis 237 (2006) 1–16
13
(
13 nm) regions, calculations give a minimum value of 4 wt%
The first observation indicates that the reaction mechanism
(surface species) must have been the same in the experiments
and a maximum value of 5.3 wt%; thus, platinum should be
distributed quite homogeneously in the sample.
with CO alone and with CO + O . The role of oxygen was lim-
2
HRTEM after PROX reaction found a markedly higher num-
ber of reduced ceria particles with a more pronounced reduc-
tion degree. Careful analysis of the ceria particles revealed an
oxygen-deficient structure (CeO1.695) in which the vacancies
ordered and formed a supercell with a lattice constant more than
twice that of CeO2. For further details, see Part II.
ited to reoxidation of support sites, because the Pt particles were
almost totally covered with CO, as postulated earlier [18,24–
26]. It correlates well with the observed strong affinity of partly
reduced ceria surface toward gas-phase oxygen; it can be reox-
idized even at low temperatures (T = 358 K) in the presence of
the PROX mixture, 3% O2 (see Fig. 8). Accordingly, the partic-
ipation of oxygen from the ceria support in CO oxidation was
proved earlier both from the observation of zero-order oxygen
pressure dependence and from oxygen-exchange measurements
4
. Discussion
16
18
between C O- and O-predosed catalysts [21].
Interaction of CO alone with the preoxidized catalyst can be
Disproportionation of carbon monoxide 2CO → C + CO2
is another reported reaction pathway for OSC measurement on
Pt/CeO2 at high temperatures [44]. The spillover and interac-
tion of this carbon species with OH groups can also result in
the formation of bands in the region of CH vibrations. Never-
theless, the reaction mechanism of CO disproportionation is not
clarified. In fact, this process can take place via dissociation of
CO and then reaction of the left-over oxygen with another ad-
sorbed CO molecule.
The suppression of C–H bond formation results from ex-
cess oxygen present in the gas phase reoxidizing these species
to carbonates or to CO2. However, at high temperature, con-
comitantly with the evolution of these bands, the formation of
seen in DRIFT spectra mainly as CO bonded to the metallic
particles (Fig. 4). This adsorbed CO can then react with spilled-
over oxygen to form gas-phase CO (observed at those times
2
in the gas-phase spectra during DRIFTS experiments, but not
shown here) [24–26,44]. The TDS measurement of adsorbed
CO (Fig. 2) confirms this model. CO adsorbed on ceria sites
and/or spilled over from the metal to the support was present
only in a low proportion (Fig. 4), whereas a pronounced in-
−
1
crease in the carboxyl region at 1800–1000 cm was observed
Fig. 5). Thus, CO on support sites must have oxidized rapidly
to form different types of carbonate species [44]. The pro-
nounced CO desorption on CO adsorption in the TDS spectra
(
2
(
Fig. 2) also supports this idea. At the same time, the interac-
−
1
gas-phase CO is suppressed (Fig. 12). The additional increase
tion of CO species with the support OH groups (3704 cm
)
2
in the C–H region is attributed not to formates, but rather to
methylidyne or partly hydrogenated graphitic species (Fig. 10)
formed after CO dissociation.
In the presence of the PROX reaction mixture, a completely
different picture is observed in the OH region of DRIFT spec-
tra (Fig. 6). Poorly resolved overlapping curves were observed
at T = 383 K, indicating a strong hydrogen-bonded system on
the surface. Accordingly, XPS of the O 1s region also indicated
a pronounced amount of adsorbed water on the ceria surface at
results in the formation of formate and bicarbonate species, as
clearly shown in Figs. 5–7.
In the case of CO oxidation with oxygen, mainly the same
changes were observed in the DRIFT spectra as in experiments
with CO alone. The only difference was that the ν(CO) bands
−1
at 2060–2100 cm were slightly blue-shifted and the bands
in the C–H region were less pronounced when oxygen was
present; they even disappeared at T = 383 K. Thus, the pres-
ence of oxygen hardly affected CO adsorption and suppressed
the formation of C–H bands.
low temperatures (see Part II, in comparison with Pd/CeO ). At
2
Fig. 12. Correlation of integrated intensities of surface C–H bands and gas-phase CO band. Both of them originate from in situ DRIFTS data and are expressed in
2
Kubelka–Munk (KM) units.

14
O. Pozdnyakova et al. / Journal of Catalysis 237 (2006) 1–16
the same time, ceria was reduced only slightly at low tempera-
tures (Fig. 8).
The interaction of CO at the metal–support interface with
the Ce cation gives rise to a strong enhancement of the band
−
1
Before the reaction mechanism for the PROX reaction is de-
scribed, it must be emphasized that Pt itself is active both in
hydrogen and CO oxidation, and an obvious reaction route is
that products of these reactions, CO2 and H2O, desorb from
Pt. The selectivity is determined only by the competitive ad-
sorption of CO, H2, and O2, as has been clearly shown for the
Pt/Al2O3 system [11,13]. In this case, hydrogen adsorption on
Pt (and hence the possibility toward nonselective oxidation), is
strongly suppressed by the high (but not total) CO coverage,
even in excess H2. Nevertheless, Pt/CeO2 behaves differently
than Pt/Al2O3 in the PROX reaction [13], as emphasized ear-
lier.
at ∼1965 cm as most of the water desorbs from ceria, in
good agreement with the findings of Kappers et al. [53]. This
“bridged-like” interface species correlates with the enhanced
intensity in the C 1s region identified as bridged CO. Its pro-
nounced appearance is accompanied by increased C–H vibra-
tion in DRIFT spectra and, accordingly, in the XPS C 1s region.
−
1
We suggest that this band at 1960–1970 cm is a possible pre-
cursor for CO dissociation on the boundary site, and thus the
precursor for methylidyne or hydrogenated graphitic species.
At the same time, formate species became more pronounced.
If CO does not dissociate at the interface but rather hops to
the ceria, then surface formates can be formed in the reaction
with OH groups. But in the absence of surface water, they ac-
cumulate near the Pt particles and represent a “dead end” for
the reaction. The formation of all of the aforementioned (C–H)
species shows an apparent negative correlation with CO2 pro-
duction (Fig. 12); thus the catalyst is deactivating.
Our results show that accumulation of both CO2 (in the form
of carbonates and formates) and H2O (in “hydrogen-bonded
OH–water structure”) is possible on the ceria support, espe-
cially at lower temperatures. These structures can be formed
either by spillover of reaction products (CO2 and H2O) or, more
likely, by spillover of hydrogen or CO to the support. The pos-
sibility of direct adsorption (e.g., H2 [71]) on the support also
cannot be neglected. Among all of these possibilities, hydro-
gen spillover is the most likely. The hydrogen on the support
reacts rapidly with oxidized ceria to form OH groups on the
To sum up, the most feasible mechanism for CO oxidation
in PROX mixture is as follows. At the beginning, a significant
amount of water accumulates on the ceria via spillover of ad-
sorbed hydrogen atoms from the platinum. This water reacts
in a LTWGS reaction with linearly bonded CO at the Pt–ceria
interface, forming CO2 and hydrogen. Hydrogen instead of des-
orption regenerates surface water by reacting with oxygen. At
high T , the hydrogen-bonded structure decomposes and water
desorbs. As a result, CO can bind in bridged-like manner at the
interface, giving rise to dissociation and/or formate formation.
Thereafter, the empty interface site can be quickly refilled by a
new CO molecule, continuing this side reaction. Although sur-
face formates as intermediates in PROX cannot be completely
excluded, they seem to represent a rather dead end for the reac-
tion. The key feature appears to be the linearly bonded CO at
the Pt–ceria interface in a close interaction with surface water.
surface, whereas ceria is reduced to Ce3 and adsorbed water
accumulates on the ceria surface. This water then reacts with
carbonate and carboxylate groups, resulting in hydroxylation
of the surface (see the relatively high amount of carboxylic acid
in the PROX reaction, as revealed by DRIFTS). The reduced
+
3+
Ce centers on the surface seem to be reoxidized; however,
HRTEM indicates that these reduced defect sites formally move
into the bulk of ceria, forming an oxygen-deficient supercell
during the reaction. The presence of adsorbed water seems to
suppress hydrogen oxidation while CO oxidation still occurs,
as the metallic particles are covered by CO, and most probably
because oxygen spillover to the metallic site is not hindered by
this hydrogen-bonded structure. The promoting effect of sur-
face water on CO oxidation over Pt catalysts was postulated
recently on microkinetic modeling [72]. The possibility of a
direct contribution of water in a low-temperature water–gas
shift (LTWGS)-type reaction has also been proposed [72]. In
this way, surface water (and/or OH groups) would react at the
perimeter of Pt particles with linearly bonded CO molecules
5. Conclusions and summary
The preferential oxidation of CO in the presence of excess
hydrogen (PROX) on Pt/CeO2 catalysts was examined by dif-
ferent spectroscopic techniques (DRIFTS and XPS) in combi-
nation with catalytic measurements in desired in situ appara-
tuses. Characterization of the sample by TDS and TEM gave
corresponding results. The first major conclusion of this paper
is that the results of these techniques, especially in situ DRIFTS
(surface species) and high-pressure XPS (oxidation states under
reaction conditions), complement one another perfectly. The
catalytic data of the PROX reaction measured in the two in situ
spectroscopic systems showed trends identical to those deter-
mined in the atmospheric catalytic reactor.
[
72] or with surface formate [63,73] or carboxyl species [72],
which are thought to be intermediates in a LTWGS reaction
producing CO2 and H2. Each of these possibilities would result
in relatively high selectivity toward CO oxidation in the PROX
reaction at low temperature.
Increasing the temperature in the PROX reaction leads to
the following: (i) decreased selectivity toward CO oxidation
(
(
Figs. 1 and 12); (ii) further oxidation of the ceria surface
Fig. 8); (iii) decreased linearly adsorbed CO species seen both
The surface species determined by DRIFTS in the presence
of CO alone and a CO + O2 mixture were very similar. Thus
CO oxidation would occur similarly in the presence and the ab-
sence of oxygen, supporting the mechanism of CO oxidation
proposed earlier for ceria-supported noble metal catalysts. The
role of gas-phase oxygen is limited to the regeneration of re-
duced ceria sites; however, in the presence of hydrogen, the
in XPS (Fig. 10) and DRIFTS (Fig. 4); (iv) increase of the band
−1
around 1965 cm (Fig. 4); (v) increased intensity of C–H vi-
brations in the DRIFT spectra (Fig. 7); (vi) increased amount of
graphite-like species in the C 1s XP spectra (Fig. 10); and (vii)
decreased surface OH–water H-bonded structure (Fig. 6).

O. Pozdnyakova et al. / Journal of Catalysis 237 (2006) 1–16
15
carbonate and formate species found by DRIFTS during the
PROX reaction strongly indicate that this reaction differs from
CO oxidation.
[17] X. Liu, O. Korotkikh, R. Farrauto, Appl. Catal. A 226 (2002) 293.
[
[
[
[
[
18] A. Trovarelli, Catal. Rev. Sci. Eng. 38 (1996) 439.
19] Y.Y. Yung-Fang, J. Catal. 87 (1984) 162.
20] S. Johansson, L. Österlund, B. Kasemo, J. Catal. 201 (2001) 275.
21] R. Taha, D. Martin, S. Kacimi, D. Duprez, Catal. Today 29 (1996) 89.
22] T. Bunluesin, H. Cordatos, R.J. Gorte, J. Catal. 157 (1995) 222.
DRIFTS found no resolved OH bands in the PROX reaction
mixture, indicating a significant amount of adsorbed water on
the ceria surface in a hydrogen-bonded structure. Its presence
seems to suppress hydrogen oxidation while CO oxidation still
occurs, as the metallic particles are covered by CO (DRIFTS).
On the other hand, no direct correlation between CO cover-
age and CO2 formation was found. The direct contribution of
surface water in a LTWGS reaction, causing apparently high
selectivity toward CO oxidation, was established. Increasing
temperature leads to slightly decreased CO coverage on metal
particles, desorption of surface water, and increase of formate
species. This latter two effects decrease the selectivity toward
CO oxidation as a function of temperature.
[23] E. Bekyarova, P. Fornasiero, J. Kaspar, M. Graziani, Catal. Today 45
1998) 179.
(
[
[
[
[
24] H.C. Yao, Y.F. Yu-Yao, J. Catal. 86 (1984) 254.
25] S. Bedrane, C. Descorme, D. Duprez, Catal. Today 73 (2002) 233.
26] S. Bedrane, C. Descorme, D. Duprez, Catal. Today 75 (2002) 401.
27] S. Bernal, J.J. Calvino, G.A. Cifredo, J.M. Rodriguez-Izquierdo, V. Perri-
chon, A. Laachir, J. Catal. 137 (1992) 1.
[28] S. Salasc, V. Perrichon, M. Primet, M. Chevrier, N. Mouaddib-Moral,
J. Catal. 189 (2000) 401.
[29] 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.
[30] W.J. Shen, Y. Ichihashi, H. Ando, Y. Matsumara, M. Okumura, M. Haruta,
Appl. Catal. A 217 (2001) 231.
Earlier investigations of the PROX reaction aimed at sup-
pressing hydrogen adsorption on the catalyst (usually metal)
surface while keeping CO adsorption and oxygen activation
[31] G. Rupprechter, J.J. Calvino, C. López-Cartez, M. Fuchs, J.M. Gatica, J.A.
Pérez-Omil, K. Hayek, S. Bernal, Stud. Surf. Sci. Catal. 130 (2000) 2021.
[32] A. Holmgren, B. Andersson, D. Duprez, Appl. Catal. B 22 (1999) 215.
[33] C. Binet, M. Daturi, J.-C. Lavalley, Catal. Today 50 (1999) 207.
[34] D.W. Daniel, J. Phys. Chem. 92 (1988) 3891.
(
usually on different sites) possible. But our results point to
the beneficial effect of surface water, which suppresses further
hydrogen oxidation and can directly participate in the water–
gas shift reaction. Another possible way to prepare a highly
selective PROX catalyst is to suppress water desorption (and
not water formation) and promote a LTWGS reaction. Further
investigation into more realistic (water- and CO2-containing)
reaction mixtures is planned.
[35] T. Jin, Y. Zhou, G.J. Mains, J.M. White, J. Phys. Chem. 91 (1987) 5931.
[36] C. Li, T. Arai, K. Domen, K. Maruya, T. Onishi, J. Chem. Soc., Faraday
Trans. 85 (1989) 929.
[37] C. Li, Y. Sakata, T. Arai, K. Domen, K. Maruya, T. Onishi, J. Chem. Soc.,
Faraday Trans. 85 (1989) 1451.
[
[
38] A. Pfau, K.D. Schierbaum, Surf. Sci. 321 (1994) 71.
39] A.Q. Wang, P. Punchaipetch, R.M. Wallace, T.D. Golden, J. Vac. Sci.
Technol. B 21 (2003) 1169.
[
[
[
40] M.A. Henderson, C.L. Perkins, M.H. Engelhard, S. Thevuthasan, C.H.F.
Peden, Surf. Sci. 526 (2003) 1.
41] Y. Madier, C. Descorme, A.M. Le Govic, D. Duprez, J. Phys. Chem. B 103
Acknowledgments
(1999) 10999.
Financial support was provided by the Hungarian National
Science Foundation (grant OTKA F046216), National R&D
Program (NKFP 3A058-04), the Athena Consortium, and the
Hungarian Academy of Sciences (a Bolyai grant to A.W.). The
authors thank the BESSY staff for providing technical assis-
tance during the in situ XPS measurements.
42] D. Teschner, A. Wootsch, K. Matusek, T. Röder, Z. Paál, Solid State Ion-
ics 141 (2001) 709.
[43] S. Kacimi, J. Barbier Jr., R. Taha, D. Duprez, Catal. Lett. 22 (1993) 343.
[
[
44] A. Holmgren, B. Andersson, J. Catal. 178 (1998) 14.
45] D.F. Ogletree, H. Bluhm, G. Lebedev, C. Fadley, Z. Hussain, M. Salmeron,
Rev. Sci. Instrum. 73 (2002) 3872.
[46] D. Teschner, A. Pestryakov, E. Kleimenov, M. Hävecker, H. Bluhm, H.
Sauer, A. Knop-Gericke, R. Schlögl, J. Catal. 230 (2005) 186.
References
[47] D. Teschner, A. Pestryakov, E. Kleimenov, M. Hävecker, H. Bluhm, H.
Sauer, A. Knop-Gericke, R. Schlögl, J. Catal. 230 (2005) 195.
[
48] P. Burroughs, A. Hamnett, A.F. Orchard, G. Thornton, J. Chem. Soc. Dal-
[
1] A.J. Appleby, F.R. Foulkes, Fuel Cell Handbook, Van Nostrand Reinhold,
New York, 1989.
ton Trans. 17 (1976) 1686.
[
[
49] D. Ferri, T. Bürgi, A. Baiker, Phys. Chem. Chem. Phys. 4 (2002) 2667.
50] R. Barth, R. Pitchai, R.L. Anderson, X.E. Verykios, J. Catal. 115 (1989)
[
[
[
2] J.N. Armor, Appl. Catal. 176 (1999) 159.
3] F. Aupretre, C. Descorme, D. Duprez, Catal. Commun. 3 (2002) 263.
4] C.D. Dudfield, R. Chen, P.L. Adock, Int. J. Hydrogen Energy 26 (2001)
61.
[51] J. Silvestre-Albero, A. Sepulveda-Escribano, F. Rodriguez-Reinoso, J.A.
763.
Anderson, Phys. Chem. Chem. Phys. 5 (2003) 208.
[
[
5] S.H. Lee, J. Han, K.-Y. Lee, J. Power Sources 109 (2002) 394.
6] G. Avgouropoulos, T. Ioannides, Ch. Papadopoulou, J. Batita, S. Hocevar,
H.K. Martalis, Catal. Today 75 (2002) 157.
[52] P. Hollins, Surf. Sci. Rep. 16 (1992) 51.
[53] M.J. Kappers, J.T. Miller, D.C. Koningsberger, J. Phys. Chem. 100 (1996)
3227.
[7] G.K. Bethke, H.H. Kung, Appl. Catal. 194–195 (2000) 43.
[8] M.J. Kahlich, H. Gasteiger, R.J. Behm, J. Catal. 182 (1999) 430.
[9] M.M. Schubert, M.J. Kahlich, H.A. Gasteiger, R.J. Behm, J. Power
Sources 84 (1999) 175.
[54] F. Bonzon-Verduraz, A. Bensalem, J. Chem. Soc., Faraday Trans. I 90
(1994) 653.
[55] T. Shido, Y. Iwasawa, J. Catal. 136 (1992) 493.
[
[
[
[
[
10] S.H. Oh, R.M. Sinkevitch, J. Catal. 142 (1993) 254.
11] M.J. Kahlich, H.A. Gasteiger, R.J. Behm, J. Catal. 171 (1997) 93.
12] S. Özkara, A.E. Aksoylu, Appl. Catal. A 251 (2003) 75.
13] A. Wootsch, C. Descorme, D. Duprez, J. Catal. 225 (2004) 259.
14] Y.-F. Han, M.J. Kahlich, M. Kinne, R.J. Behm, Phys. Chem. Chem.
Phys. 4 (2002) 389.
[56] C. Binet, M. Daturi, J.C. Lavalley, Catal. Today 50 (1999) 207.
[57] A. Tsyganenko, V. Filimonov, J. Mol. Struct. 19 (1973) 579.
[58] A. Badri, C. Binet, J.C. Lavalley, J. Chem. Soc., Faraday Trans. 92 (1996)
4669.
[59] A. Laachir, V. Perrichon, A. Badri, J. Lamotte, E. Chaterine, J.C. Lavalley,
J. El Fallah, L. Hilaire, F. Le Normand, E. Quemere, G.N. Sauvion, O.
Touret, J. Chem. Soc., Faraday Trans. 87 (1991) 1601.
[60] K. Nakamoto, IR and Raman Spectra of Inorganic and Coordination Com-
pounds, Wiley–Interscience, New York, 1987.
[
15] M.M. Schubert, M.J. Kahlich, G. Feldmeyer, M. Hüttner, S. Hackensberg,
H.A. Gasteiger, R.J. Behm, Phys. Chem. Chem. Phys. 3 (2001) 1123.
16] M. Haruta, Catal. Today 36 (1997) 153.
[

16
O. Pozdnyakova et al. / Journal of Catalysis 237 (2006) 1–16
[
61] G. Busca, J. Lamotte, J.C. Lavalley, V. Lorinzelli, J. Am. Chem. Soc. 109
[68] J.L.G. Fierro, J.M. Palacios, F. Tomas, Surf. Interface Anal. 13 (1988)
25.
(1987) 5197.
[
[
62] C. Li, K. Domen, K. Maruya, T. Onishi, J. Catal. 125 (1990) 445.
63] G. Jacobs, U.M. Graham, E. Chenu, P.M. Patterson, A. Dozier, B.H.
Davis, J. Catal. 229 (2005) 499.
[69] K. Dückers, K.C. Price, H.P. Bonzel, V. Chab, K. Horn, Phys. Rev. B 36
(1987) 6292.
[70] N.M. Rodriguez, P.E. Anderson, A. Wootsch, U. Wild, R. Schlögl, Z. Paál,
J. Catal. 197 (2001) 365.
[64] M.-C. Jung, H.-D. Kim, M. Han, W. Jo, D.C. Kim, Jpn. J. Appl. Phys. 38
(
1999) 4872.
65] J.S. Hammond, N. Winograd, J. Electrochem. Electroanal. Chem. 78
1977) 55.
[
[
[71] S. Bernal, J.J. Calvino, G.A. Cifredo, J.M. Gatica, J.A.P. Omil, J.M. Pin-
tado, J. Chem. Soc., Faraday Trans. 89 (1993) 3499.
[72] A.B. Mhadeswar, D.G. Vlachos, J. Phys Chem. B 108 (2004) 15246.
[73] G. Jacobs, P.A. Patterson, U.M. Graham, D.E. Sparks, B.H. Davis, Appl.
Catal. A 269 (2004) 63.
[
(
66] K.S. Kim, N. Winograd, R.E. Davis, J. Am. Chem. Soc. 93 (1971) 6296.
67] C.R. Parkinson, M. Walker, C.F. McConville, Surf. Sci. 545 (2003) 19.