Investigation of methanol oxidation over Au/catalysts using operando IR spectroscopy: Determination of the active sites, intermediate/spectator species, and reaction mechanism

Article abstract of DOI:10.1021/ja1028809

FTIR spectroscopy coupled with mass spectrometry has been used to study the mechanism of methanol oxidation at low temperatures on nanostructured Au/CeO₂ and Au/TiO₂ catalysts. Activity and selectivity toward CO₂ have been

Full text of DOI:10.1021/ja1028809

Published on Web 07/19/2010
Investigation of Methanol Oxidation over Au/Catalysts Using
Operando IR Spectroscopy: Determination of the Active Sites,
Intermediate/Spectator Species, and Reaction Mechanism
Se´verine Rousseau,†,§ Olivier Marie,*,† Philippe Bazin,† Marco Daturi,†
Ste´phane Verdier,‡ and Virginie Harle´‡
Laboratoire Catalyse et Spectrochimie, ENSICAEN, UniVersite´ de Caen, CNRS,
6 Bd Mare´chal Juin, 14050 Caen, France, and Rhodia Recherches, 52 Rue de la Haie Coq,
93308 AuberVilliers Cedex, France
Received April 6, 2010; E-mail: marieo@ensicaen.fr
Abstract: FTIR spectroscopy coupled with mass spectrometry has been used to study the mechanism of
methanol oxidation at low temperatures on nanostructured Au/CeO2 and Au/TiO2 catalysts. Activity and
selectivity toward CO2 have been investigated through simultaneous analysis of adsorbed surface species
and gaseous species, and some key steps in the oxidation pathway, active sites, and intermediate species
are proposed. Among the detected species, some kinds of methoxy species formed on the support were
identified as intermediates, which further transform into formates whose oxidation was found to be the
rate-determining step for the reaction. The role of the support and the noble metal in the mechanism are
revealed using operando spectroscopy.
14,15
TiO2,2,7,12,13 or even CexTi(1-x)O2
as metal oxide supports
1. Introduction
greatly enhances the efficiency of “classical” Al2O3 supported
noble metal catalysts for VOC total oxidation. Not only do
supported noble metals such as Pd, Pt, Ag12,16-20 provide active
sites for oxidation reaction, but also transition metals (such as
Cu) supported onto the previously selected oxides3,21-23 are
reported to be efficient, especially for the CO preferential
oxidation. Indeed, the CO preferential oxidation reaction (known
as the PROX reaction) has been the subject of many efforts in
both the academic and the industrial world due to the high
In the frame of air depollution, several targets (such as
automobile cockpits, working or public frequented indoor halls,
restricted areas, refrigerated cells, domestic environment, etc.)
require more efficient compounds than activated carbon absorb-
ers. The aim being not only to remove the atmospheric pollutants
but also to transform them into inert species, an inexpensive
catalytic process efficient at room temperature (and even below)
under atmospheric conditions was investigated as a challenge.
Here, the complete oxidation of methanol (CH3OH) chosen as
model volatile organic compound (VOC) and its mechanism in
heterogeneous catalysis are reported. The oxidizing agent being
the dioxygen naturally present in the ambient air, the total
methanol oxidation stoichiometry is described by the following
equation:
(7) Solsona, B. E.; Garcia, T.; Jones, C.; Taylor, S. H.; Carley, A. F.;
Hutchings, G. J. Appl. Catal., A 2006, 312, 67–76.
(8) Andreeva, D.; Petrova, P.; Ilieva, L.; Sobczak, J. W.; Abrashev, M. V.
Appl. Catal., B 2008, 77, 364–372.
(9) Centeno, M. A.; Paulis, M.; Montes, M.; Odriozola, J. A. Appl. Catal.,
A 2002, 234, 65–78.
(10) Liotta, L. F.; Di Carlo, G.; Longo, A.; Pantaleo, G.; Venezia, A. M.
Catal. Today 2008, 139, 174–179.
(11) Milone, C.; Fazio, M.; Pistone, A.; Galvagno, S. Appl. Catal., B 2006,
68, 28–37.
(12) Rojluechai, S.; Chavadej, S.; Schwank, J. W.; Meeyoo, V. Catal.
Commun. 2007, 8, 57–64.
3
2
CH3OH + O2 f CO2 + 2H2O
(1)
Regarding the nature of potential candidates for active
catalysts, a brief survey of the literature data relative to air
pollution abatement indicates that the presence of CeO2,1-12
(13) Damyanova, S.; Cubeiro, M. L.; Fierro, J. L. G. J. Mol. Catal. A:
Chem. 1999, 142, 85–100.
(14) Gennequin, C.; Lamallem, M.; Cousin, R.; Siffert, S.; A¨ıssi, F.;
Abouka¨ıs, A. Catal. Today 2007, 122, 301–306.
† Universite´ de Caen.
‡ Rhodia Recherches.
(15) Lamallem, M.; Ayadi, H. E.; Gennequin, C.; Cousin, R.; Siffert, S.;
A¨ıssi, F.; Abouka¨ıs, A. Catal. Today 2008, 137, 367–372.
(16) Choudhary, V. R.; Samanta, C.; Choudhary, T. V. Appl. Catal., A
2006, 308, 128–133.
§ Current address: PSA Peugeot Citroe¨n - Centre Technique de Ve´lizy,
Route de Gisy, 78943 Ve´lizy Villacoublay, France.
(1) Gluhoi, A. C.; Lin, S. D.; Nieuwenhuys, B. E. Catal. Today 2004,
90, 175–181.
(17) Menegazzo, F.; Burti, P.; Signoretto, M.; Manzoli, M.; Vankova, S.;
Boccuzzi, F.; Pinna, F.; Strukul, G. J. Catal. 2008, 257, 369–381.
(18) Monyanon, S.; Pongstabodee, S.; Luengnaruemitchai, A. J. Power
Sources 2006, 163, 547–554.
(2) Liu, S. Y.; Yang, S. M. Appl. Catal., A 2008, 334, 92–99.
(3) Liu, W. China Particuol. 2005, 3, 383–394.
(4) Scire`, S.; Minico`, S.; Crisafulli, C.; Satriano, C.; Pistone, A. Appl.
Catal., B 2003, 40, 43–49.
(19) Monyanon, S.; Pongstabodee, S.; Luengnaruemitchai, A. J. Chin. Inst.
Chem. Eng. 2007, 38, 435–441.
(5) Shen, Y.; Yang, X.; Wang, Y.; Zhang, Y.; Zhu, H.; Gao, L.; Jia, M.
Appl. Catal., B 2008, 79, 142–148.
(20) Tabakova, T.; Boccuzzi, F.; Manzoli, M.; Chiorino, A.; Andreeva,
D.; Aldo Gamba, C. C.; Salvatore, C. Studies in Surface Science
Catalysis; Elsevier: New York, 2005; Vol. 155, pp 493-500.
(6) Sheng, P. Y.; Bowmaker, G. A.; Idriss, H. Appl. Catal., A 2004, 261,
171–181.
9
10832 J. AM. CHEM. SOC. 2010, 132, 10832–10841
10.1021/ja1028809  2010 American Chemical Society

Investigation of Methanol Oxidation
A R T I C L E S
Table 1. Some Physicochemical Properties of the Studied
Samples
interest to develop a process able to yield an ultra pure hydrogen
source for fuel cell applications. In this frame, intense investiga-
tions have been conducted during recent years, and especially
gold-containing materials were reported as very promising
catalysts to selectively convert CO into CO2 in the presence of
an excess of H2.1,18,19,21,22,24-29 In fact, gold has been for a
long time disregarded for any catalytic applications because this
noble metal is one of the most chemically inert element. The
very new life of gold in heterogeneous catalysis is due to
Haruta’s pioneer work30 who evidenced both a preparation
method to obtain Au supported nanoparticles (mean diameter
<5 nm) and the requirement of these nanosized gold particles
to generate highly active catalytic sites. More precisely, the new
coprecipitation method, which allows one to prepare gold
nanoparticles with an optimal size around 3 nm,31 even opened
the door to the low temperature catalytic oxidation, because it
appeared that CO could be converted into CO2 at temperature
as low as 200 K. From this discovery, the research on catalytic
properties for gold has increased exponentially. Nevertheless,
only a relative limited fraction of the reported studies deals with
mechanistic investigations, and the majority of them concern
CO oxidation at low temperature. These are the major reasons
why this study focuses on Au/CeO2 and Au/TiO2 nanosized
catalysts, aiming at understanding how they really work at the
molecular scale to provide information useful for extending their
efficiency to the oxidation of molecules more complex than CO,
that is, VOC compounds, represented by methanol in the present
case.
gold content/
wt %
specific area
(BET)/m2 g-1
gold particles
size/nm
sample name
TiO2
85
180
85
180
60
CeO2HSA
Au/TiO2
Au/CeO2HSA
Au/CeO2LSA
0.6
1
1
<3
<3
<3
the sample modifications and the methanol conversion, which was
also measured by online mass spectrometry.
Details on the operando system and the applied methodology
are reported in the Supporting Information. For this specific study
requiring efficient thermal regulation at temperatures close to room
temperature, a new reactor cell has been designed in our laboratory
(Figure 1). It is made of a stainless steel cylinder that carries a
toroidal sample holder in its center, where the catalyst is placed in
the form of a self-supported wafer of 10 mg cm-2. The thermo-
regulating system allows one to hold the sample at a fixed
temperature in a range from 263 to 373 K. The reaction chamber
hosting the sample is obviously tight with respect to the external
atmosphere, and its dead volume is extremely small allowing one
to obtain (i) time-resolved analysis, (ii) surface spectra without
superposition of the gas-phase signal, and (iii) fluid dynamics very
similar to that of a honeycomb system.
The reacting gas mixture composition was fixed to 700 ppm
CH3OH, 20 vol % O2 in argon, and the total flow was adjusted to
work with a constant gas hourly space velocity (GHSV) equal to
60 000 h-1. Before any experiment, each sample was simply treated
with argon and oxygen for a night at room temperature to stabilize
the surface state, without any thermal activation, to prevent metal
particle growth.
2. Experimental Section
Materials. The catalysts were prepared by depositing Au (via
trihydrated chlorauric acid) onto CeO2 and TiO2 commercial powder
materials with granulometry in the micrometer range. Two ceria
supports with pure fluorine phase having a high and a low level of
initial specific surface area and a pure anatase titania support were
chosen for this study. Details on the sample preparation are reported
in the Supporting Information.
In both cases, Au is finely dispersed onto the oxide supports
with a mean particle size below 3 nm as estimated by the X-ray
diffraction line broadening of Au. Furthermore, the specific surface
area of the oxides was unaffected by the Au deposition, as described
in Table 1.
3. Results and Discussion
3.1. Effect of the Support CeO2 or TiO2. The first aim of
this study was to investigate a potential activity of the support
oxides themselves. The evolution of the CH3OH conversion
(calculated from IR gas-phase analysis) with time on stream
(TOS) is given for CeO2-based samples in Figure 2. Focusing
on the CeO2HSA sample, a conversion loss to reach a plateau
without any catalytic activity was observed. Two hypotheses
are possible to explain this phenomenon of conversion drop:
(i) the catalysts deactivate with time on stream or (ii) the
disappearance of CH3OH is related to adsorption only and stops
when the surface is saturated.
Operando Tests. Operando IR studies were carried out using
the transmission technique to obtain quantitative data. Sample and
gas-phase IR spectra were alternatively collected to follow both
The upper part of Figure 2 indicates that during the first 30
min, the exposed surface cerium cations are saturated by
methoxy species coming from the dissociative methanol adsorp-
tion. According to abundant literature data,32-37 the band
evolving at 1102 cm-1 is associated with the formation of on
top methoxy species (type I) on Ce4+ cations, thus giving
evidence that after one night pretreatment under oxygen at room
temperature the main part of surface cations is oxidized. During
(21) Avgouropoulos, G.; Ioannides, T.; Papadopoulou, C.; Batista, J.;
Hocevar, S.; Matralis, H. K. Catal. Today 2002, 75, 157–167.
(22) Avgouropoulos, G.; Papavasiliou, J.; Tabakova, T.; Idakiev, V.;
Ioannides, T. Chem. Eng. J. 2006, 124, 41–45.
(23) Tabakova, T.; Boccuzzi, F.; Manzoli, M.; Sobczak, J. W.; Idakiev,
V.; Andreeva, D. Appl. Catal., A 2006, 298, 127–143.
(24) Avgouropoulos, G.; Manzoli, M.; Boccuzzi, F.; Tabakova, T.; Pa-
pavasiliou, J.; Ioannides, T.; Idakiev, V. J. Catal. 2008, 256, 237–
247.
(25) Ko, E.-Y.; Park, E. D.; Seo, K. W.; Lee, H. C.; Lee, D.; Kim, S.
Catal. Today 2006, 116, 377–383.
(32) Colo´n, G.; Pijolat, M.; Valdivieso, F.; Vidal, H.; Kasˇpar, J.; Finocchio,
E.; Daturi, M.; Binet, C.; Lavalley, J. C. J. Chem. Soc., Faraday Trans.
1998, 94, 3717.
(26) Luengnaruemitchai, A.; Osuwan, S.; Gulari, E. Int. J. Hydrogen Energy
2004, 29, 429–435.
(27) Panzera, G.; Modafferi, V.; Candamano, S.; Donato, A.; Frusteri, F.;
Antonucci, P. L. J. Power Sources 2004, 135, 177–183.
(28) Manzoli, M.; Avgouropoulos, G.; Tabakova, T.; Papavasiliou, J.;
Ioannides, T.; Boccuzzi, F. Catal. Today 2008, 138, 239–243.
(29) Wang, H.; Zhu, H.; Qin, Z.; Wang, G.; Liang, F.; Wang, J. Catal.
Commun. 2008, 9, 1487–1492.
(33) Finocchio, E.; Daturi, M.; Binet, C.; Lavalley, J. C.; Blanchard, G.
Catal. Today 1999, 52, 53.
(34) Daturi, M.; Binet, C.; Lavalley, J. C.; Galtayries, A.; Sporken, R. Phys.
Chem. Chem. Phys. 1999, 5717.
(35) Daturi, M.; Finocchio, E.; Binet, C.; Lavalley, J. C.; Fally, F.;
Perrichon, V.; Vidal, H.; Hickey, N.; Kasˇpar, J. J. Phys. Chem. B
2000, 104, 9186.
(30) Haruta, M.; Kobayashi, T.; Sano, H.; Yamada, N. Chem. Lett. 1987,
16, 405.
(36) Binet, C.; Daturi, M. Catal. Today 2001, 70, 155. 155.
(37) Binet, C.; Daturi, M.; Lavalley, J. C. Catal. Today 1999, 50, 207.
(31) Haruta, M. CATTECH 2002, 6, 102.
9
J. AM. CHEM. SOC. VOL. 132, NO. 31, 2010 10833

A R T I C L E S
Rousseau et al.
Figure 1. Detailed view of the CIRCOPE IR operando cell: (1) thermocouple, (2) oil tank, (3) oil circulation port connected to the cryostat, (4) wafer holder
(stainless steel), (5) sample (wafer), (6) adjusting nut for air-tightness, (7) gas flow port, (8) adjusting nut for KBr windows, (9) stainless steel ring, (10)
Teflon ring, (11) stainless steel cell body, (12) PEEK ring, (13) KBr window, and (14) Kalrez O-ring.
assigned to the methoxys doubly bridging Ce4+ cations having
an oxygen vacancy in the neighborhood (type II′ according to
Badri et al.38). For higher times on stream (>10 min), the amount
of adsorbed methanol keeps on increasing, and new bands in
the doubly bridging methoxy region (1053 and 1028 cm-1
)
emerge due to the coverage level effect. The first band is due
again to type II methoxys, that is, bridged species on Ce4+
cations, whereas the second band would characterize nondis-
sociative methanol physisorption.39 A shoulder at 1010 cm-1
typical for traces of triply bridging methoxy species (type III)37
is detected for the highest coverage level. Both the position of
the doubly bridging methoxy bands (1053 instead of 1065 cm-1,
as expected for an oxidized ceria) and the relative intensity of
the type III methoxy species slightly differ from spectra
previously reported for methanol adsorption on pure CeO2.38
These discrepancies may be due to the presence of remaining
adsorbed water and carbonate species (which may perturb other
adducts), because no thermal pretreatment was performed.
Finally, considering the band intensities, it is noticeable that
when the methanol adsorption equilibrium is reached, the peak
of type I methoxy band at 1102 cm-1 is around 1.5 times higher
than that relative to type II. Comparing the evolution of the
methoxy bands (lower part of Figure 2) with the profile of the
mass spectra reported in Figure 3A, it is evident that once
the surface is fully covered (30-40 min), the methanol level in
the exhaust gas is the same as that at the inlet, while there is no
CO2 production, thus indicating that there is no catalytic
conversion. Therefore, we can conclude that on the ceria support
the disappearance of CH3OH is related to adsorption only.
However, the m/z ) 18 signal (typical for H2O) is increasing
during the very first time on stream to go through a maximum
and further decrease to the zero level when the surface
equilibrium is reached. This phenomenon arises from the surface
dehydroxylation upon replacement of OH groups by methoxys.
A quantitative confirmation for pure adsorption phenomena
occurring with CeO2HSA was obtained from an estimation of
Figure 2. Bottom: Evolution of the methanol conversion (s) and the 1102
cm-1 band intensity (- - - -) with time on stream when the reaction is
conduced over CeO2HSA at (b) 323 K or Au/CeO2HSA at (0) 298 K and (9)
323 K. GHSV ) 60 000 h-1, 700 ppm CH3OH/20% O2 in Ar. Top: FTIR
subtraction spectra (the spectrum of the surface before reaction was
subtracted for clarity seek) of adsorbed species on CeO2HSA upon increasing
time on stream at 323 K: (a) 5 min, (b) 10 min, (c) 20 min, (d) 30 min, (e)
60 min, (f) 90 min, (g) 120 min, (h) 150 min, and (i) 180 min.
(38) Badri, A.; Binet, C.; Lavalley, J. C. J. Chem. Soc., Faraday Trans.
1997, 93, 1159.
the first 10 min under stream, another band centered at around
1040 cm-1 grows conjointly with the 1102 cm-1 one. That is
(39) Meyer, R.; Gunthard, H. S. J. Mol. Spectrosc. 1974, 52, 94.
9
10834 J. AM. CHEM. SOC. VOL. 132, NO. 31, 2010

Investigation of Methanol Oxidation
A R T I C L E S
Figure 3. Evolution of mass spectra signal (m/z ) 31 for methanol, 18 for water, 44 for CO2) intensities with time on stream [GHSV ) 60 000 h-1; flow:
700 ppm CH3OH + 20% O2 (when present) in Ar, T ) 323 K] when the reaction is performed over (A) CeO2, (B) Au/CeO2HSA, (B′) Au/CeO2LSA, and (C)
Au/CeO2HSA without O2.
the amount of adsorbed methanol under reaction flow. Indeed,
it was found that when the equilibrium is reached (plateau
regime), the coverage level almost corresponds to a monolayer
in which methoxy species are in close contact with their
neighbors. The details for the calculation are provided in the
Supporting Information.
The same trend for both the gas-phase data and the surface
species evolution was observed for TiO2 and Au/TiO2 catalysts,
confirming that for CeO2HSA, TiO2, and Au/TiO2 samples, the
initial apparent conversion is only due to adsorption phenomena.
For brevity’s sake, all of the catalytic data concerning TiO2-
based samples are not reported here but are available in the
Supporting Information.
conversion (evaluated after 1 h of TOS when the activity is
stabilized) at 323 K (50% calculated from IR online gas-phase
analysis) than at 298 K (15%). The synergetic effect between
the CeO2 support and the gold nanoparticles will be discussed
next.
In more detail, Figure 3B represents the MS data obtained
for CH3OH oxidation at 323 K over Au/CeO2HSA: coherently
with the IR analysis, the high initial CH3OH consumption
progressively decreases to reach a steady state after around 40
min on stream. The m/z ) 18 water signal is detected starting
from around 5-10 min, as it was the case for the Au-free CeO2
(Figure 3A), and during the first 20 min on stream, both CeO2HSA
and Au/CeO2HSA present the same MS signal evolution. In
particular, the m/z ) 44 associated with CO2 remains close to
zero, indicating in both cases that only methanol dissociative
adsorption first takes place leading to water release. On the
contrary, starting from 20 min on stream, an obvious gold effect
appears, because CO2 starts to be produced while the methanol
3.2. Effect of Gold Nanoparticles on CeO2. Contrarily to the
pristine support, when gold nanoparticles are deposited on ceria,
the activity stabilizes to a nonzero conversion (Figure 2). This
result indicates an efficient catalytic activity for Au/CeO2HSA
sample in methanol oxidation, which is confirmed by a higher
9
J. AM. CHEM. SOC. VOL. 132, NO. 31, 2010 10835

A R T I C L E S
Rousseau et al.
Figure 4. FTIR subtraction spectra in the δ(CH) and ν(CO) vibration range (the spectrum of the surface before reaction was subtracted for clarity seek) of
adsorbed species with time on stream [GHSV ) 60 000 h-1; flow: 700 ppm CH3OH + 20% O2 (when present) in Ar, T ) 323 K] when the reaction is
performed over (A) CeO2, (B) Au/CeO2HSA, and (C) Au/CeO2HSA without O2.
signal tends toward a stabilized value around 50% of the
bypassed level. This clearly indicates that the CO2 formation is
delayed when compared to the H2O formation and would
suggest that the methanol conversion proceeds in consecutive
steps. Operando technique was then applied to observe the
catalyst surface while working; the evolution of the adsorbed
species with time on stream should provide further information
about these consecutive steps.
3.3. CeO2 and Au/CeO2: Evolution of Adsorbed Species
with Time On Stream. The evolution with time on stream at
323 K of the spectra typical of adsorbed species over CeO2HSA
is reported in Figure 4A.
reduction into Ce3+. The cerium cations’ reduction appears to
be a very local consequence of the formate formation because
no evidence of methoxy species coordinated with Ce3+ could
be detected. The corresponding symmetric mode νs(OCO) is
much less sensitive to the cerium cation oxidation degree so
that only one band is detected at 1358 cm-1 for the formates.
To complete the formate species characterization, the 1371 cm-1
band can be assigned to their δ(CH) mode.33 It is worth
underlining that all of the band assignments reported above are
fully consistent with the vibration modes detected in the ν(CH)
stretching region, which are exhaustively described in the
Supporting Information.
Concerning the methoxy species, their ν(CO) modes were
already fully described in Figure 2: briefly, the type II are
Considering now the ν(CO) vibration modes for methoxy
species, the wavenumbers observed during reaction are rather
the same as those observed upon the first instants of adsorption
on the bare CeO2 support (when there is no shift due to coverage
effect), that is, at 1102 and 1042 cm-1 for type I and type II′
methoxys, respectively. Moreover, when the reaction takes place
over Au/CeO2HSA, the peak height for the type I methoxy band
is always lower than that relative to the type II′ methoxy,
whereas it was almost double in the case of pure adsorption on
the bare CeO2 (Figures 2 and 4A). One may wonder if this is
due to the fact that Ce4+ sites are less accessible in the presence
of supported Au. Knowing the gold weight loading, the
deposited gold particles mean size (Sm < 3 nm; see Table 1),
and the support specific area, it is deduced that less than 0.5%
of the CeO2 surface is thus occupied by Au nanoparticles
(calculation details are provided in the Supporting Information).
Thus, the much lower amount of type I methoxy observed during
reaction over Au/CeO2HSA cannot be explained by a much lower
number of Ce4+ adsorption sites due to Au nanoparticles
deposition. This last result indicates that, under reaction flow,
the surface never reaches methoxy saturation and suggests that
associated with the broad massif centered around 1053 cm-1
,
while the type I are responsible for the highest intensity band
at 1102 cm-1. The bands around 1450 cm-1 are associated with
the bending δ(CH) modes, while the “waves” effects observed
between 1600 and 1200 cm-1 are due to perturbations with the
already present adsorbed impurities (carbonates, water, etc., as
detected on the surface spectra before sending the reactant flow).
In a similar way, the evolution of the spectra with time on
stream at 323 K is reported for Au/CeO2HSA in Figure 4B: new
intense bands are observed at 1595, 1575, and 1553 cm-1. These
three bands correspond to the νa(OCO) vibrations of various
adsorbed formate species. The 1595 and the 1575 cm-1
components are characteristic of HCOO- species on Ce3+
cations, whereas the lower wavenumber component at 1553
33,40
cm-1 is due to bidentate formate species on oxidized Ce4+
.
The 1595 and 1575 cm-1 bands being the most intense, the
formate species’ appearance is thus accompanied by Ce4+
(40) Binet, C.; Jadi, A.; Lavalley, J. C. J. Chim. Phys. Phys.-Chim. Biol.
1992, 89, 1441.
9
10836 J. AM. CHEM. SOC. VOL. 132, NO. 31, 2010

Investigation of Methanol Oxidation
A R T I C L E S
“adlineation sites”) should be present at the interface between
gold and the support.41 Working on the hypothesis of a catalytic
reaction through a Mars van Krevelen mechanism, the oxygen
atoms consumed for methanol oxidation are replaced in the
oxide support surface vacancies by oxygen molecules from the
gas phase. More precisely, we propose that, at room temperature
and at the adlineation sites, the Au role consists of enabling
the Ce3+ formate formation from ceria surface oxygen species,
which further generates surface vacancies acting as oxygen
pumps (reverse spillover effect). It would also be tempting to
propose that the atomic active oxygen species necessary to
transform methoxys into formates are provided by a direct
oxygen molecule dissociation onto gold nanoparticles. This
possibility was in fact proposed by Schubert et al. in their elegant
study on the role of inert and active support materials for the
oxygen supply: according to them for Au supported on inert
materials, the oxygen supply would most likely proceed via
direct dissociative adsorption on the Au particles, while for Au
catalysts supported on active materials (such as CeO2) the
dominant reaction pathway would involve the adsorption of a
mobile, molecular oxygen species on the support with a further
dissociation at the interface with gold nanoparticles.42 The last
conclusion should be relevant to this work for which the ceria
support is known to possess an active redox surface acting as
an oxygen buffer. Moreover, the hypothesis of O2 direct
dissociation onto gold was here disregarded, because, to our
knowledge, no proof of such a dissociative adsorption has been
clearly provided.
Effect of Anaerobic Conditions. To illustrate the role of the
ceria mobile oxygen species, a similar experiment was con-
ducted with an anaerobic flow. In these conditions, the methanol
dissociative adsorption also takes place, as indicated by the water
yield (Figure 3C, m/z ) 18), but then the conversion evolution
is similar to that observed for pure CeO2 experiment (Figure
3A), that is, without any further oxidation product detected in
the gas phase (m/z ) 44). Focusing on the surface (Figure 4C),
some formates are however detected. Scrutinizing in depth the
spectra, a similar decline (although softer) of type I methoxy
was detected. This decrease corresponds to the formate appear-
ance (associated with the ceria buffered oxygen species con-
sumption), but both type II and type II′ reach a plateau without
any further decrease; their oxidation into formates involving a
neighboring surface or subsurface oxygen is prevented due to
a limited oxygen availability/mobility related to the absence of
O atoms (from the gas phase) able to re-equilibrate the solid
anionic content.
Figure 5. Evolution of selected IR bands typical of adsorbed species on
Au/CeO2HSA catalyst with time on stream at 323 K (700 ppm CH3OH +
20% O2 in Ar). The curve labeling refers to the corresponding band
wavenumber in cm-1
.
type I methoxy are continuously transformed, playing probably
a key role in the catalytic methanol oxidation.
To further investigate this possible methoxy key role, the
evolution of the intensity of characteristic bands upon TOS has
been reported for the Au/CeO2HSA (Figure 5). The 1040 cm-1
ν(CO) band, typical of type II′ methoxy, presents the same
feature for Au/CeO2 and for CeO2, characterized by a plateau
(see Figures 4 and 5).
On the contrary, the 1102 cm-1 component typical of type I
methoxy now goes through a maximum for the Au/CeO2
catalyst. This trend is typical for an intermediate species whose
transformation starts when a critical concentration (here surface
coverage) is reached. The decrease of the amount of type I
methoxy on the surface corresponds to the significant appearance
of the new bands typical for formate species at 1553, 1575,
and 1595 cm-1, which were not observed on the bare supports
nor on Au/TiO2 (not reported here). The fact that during reaction
over Au/CeO2HSA the type I methoxy band intensity never
reaches the same level as that observed upon adsorption
equilibrium over the bare CeO2HSA is thus due to the oxidation
of type I methoxy into formate species. Moreover, the 1053
cm-1 mode typical of type II methoxys is hardly observed over
Au/CeO2 under both CH3OH and O2 flow (Figure 4B), which
also suggests their transformation into formates in these
conditions. Indeed, the 1053 and 1040 cm-1 bands are overlap-
ping, and the intensity of the former is low so that its evolution
is better described in Figure 5. Actually, it then appears that
both type I and type II methoxy species have parallel evolution
under methanol and oxygen flow, thus indicating similar
reactivity in these conditions. On the contrary, the type II′
methoxy band intensity at 1040 cm-1 seems to remain stable
when formate species are formed (compare Figure 4A and B
and refer to Figure 5). This can be justified by the fact that the
environment of type II′ species is oxygen deficient, being
therefore unable to oxidize methoxy to formate species, whereas
type I and type II are coordinated to cations bearing a complete
oxygen 8-fold coordination. This oxygen needs the presence of
a metal (i.e., Au in our case) to be activated at low temperature,
leading to an increased surface and subsurface mobility via a
reverse spillover effect as reported by Leppelt et al., who also
concluded that a catalytic active perimeter (the so-called
This methanol oxidation mechanism going through formate
species was already observed on CeO2-based samples, but at
higher temperatures of 423-473 K.33 As aforementioned, the
gold nanoparticles would presumably increase the ceria oxygen
mobility, likely via a reverse spillover effect, thus decreasing
the methanol oxidation energy barrier and making it possible
at a lower temperature than on the bare support.
3.4. Methanol to CO2 Mechanism over Au/CeO2. The aim
for any operando experiment is to make the link between gas
phase and surface analysis data. For this purpose, selected time
on stream ranges will be considered, dealing first with the
experiment over Au/CeO2 at 323 K in the presence of both
CH3OH and O2.
(41) Leppelt, R.; Schumacher, B.; Plzak, V.; Kinne, M.; Behm, R. J. J.
Catal. 2006, 244, 137–152.
(42) Schubert, M. M.; Hackenberg, S.; van Veen, A. C.; Muhler, M.; Plzak,
V.; Behm, R. J. J. Catal. 2001, 197, 113–122.
9
J. AM. CHEM. SOC. VOL. 132, NO. 31, 2010 10837

A R T I C L E S
Rousseau et al.
(i) From zero to around 10 min on stream, the methanol
conversion is stable around 100%, but no reaction products are
observed in the exhaust gas stream (see Figures 2 and 3B).
Analysis of the surface data evidenced accumulation of meth-
oxys on the catalyst (types I, II, and II′ as described on Figures
4B and 5), indicating that methanol removal from the gas phase
is mostly due to dissociative adsorption. Indeed, only traces of
formates are detected on the surface during this period.
methanol oxidation reaches the steady state. The surface data
indicate that starting from around 45 min, the surface equilib-
rium between methoxy and formate species is almost achieved,
and the formate accumulation rate slows as indicated in Figure
5 by the slope change. At this step, we can hardly distinguish
whether formates are bare spectator species or reaction inter-
mediates. Some points are however in favor of the second
hypothesis; because the formate species are accumulating on
the surface and because the lowering of their accumulating rate
corresponds to the presence of CO2 as a reaction product, one
can propose that the complete methanol oxidation-determining
step consists of the formate transformation. In the case of the
low temperature water gas shift reaction over similar catalysts,
it was indeed reported that formates adsorbed on the ceria
support act as reaction intermediates in the dominant reaction
pathway41 yielding CO2 and H2 (in oxygen-free atmosphere).
Moreover, in the past, we already demonstrated on ceria-based
compounds that methanol can be adsorbed and then oxidized
through formate intermediates.33 The different possibilities for
formate transformation into CO2 and H2O were thus investigated.
The first possibility is the decomposition of formates into
hydroxyls and CO as described by eq 3. As already described,
step c would only occur when O2 is present in the flow.
Moreover, step d is consistent with the best known property of
gold nanoparticles, which is to be very active for CO oxidation,
as recently reviewed by C. Louis.43 The second possibility is
(ii) From 10 min to around 45 min TOS, the conversion drops
from 100% to stabilize around 50%. This phenomenon is
accompanied by water and CO2 production (Figure 3B),
indicating that catalytic reactions start to proceed during this
step. However, a delay between water and CO2 production is
observed attributed to first H2O release upon methanol dissocia-
tive adsorption and then to methoxy oxidation. The real catalytic
reaction, evidenced by CO2 release, takes place after 20 min
TOS, as confirmed by surface data showing methoxy species
decrease concomitant with formate accumulation on the catalyst,
which become the principal species in the infrared spectra
(Figures 4B and 5). At this step of the study, one can wonder
from Figures 3 and 4 why the monolayer (made of both
methoxys and formates) should be practically established to
observe a consistent catalytic activity (CO2 production). Con-
sidering the calculation reported in the Supporting Information,
the gold nanoparticles only account for 0.5% of the catalyst
area. Because it has been shown that the presence of gold
nanoparticles is necessary for catalytic activity, the active sites
must thus be on the metal itself or in its close neighborhoods.
Therefore, the reactants must first reach those sites before
undergoing any further reaction than single dissociative adsorp-
tion. Being at relatively low temperature, the species diffusivity
at the surface is limited, and statistically only a fully covered
surface allows one to obtain the required proximity to initiate
the reaction. Aiming at checking this point, the Au/CeO2 LSA
sample having the same gold loading but a support surface area
lowered to one-third was tested. As expected, the sample
oxidation activity evidenced by CO2 formation started at a time
roughly 3 times shorter (compare Figure 3B with B′); the surface
full coverage (characterized here by a total amount equal to
one-third of those on Au/CeO2HSA) is therefore a prerequisite
for initiating the oxidation reaction. This fact is again in perfect
agreement with the existence of an “active perimeter” at the
interface between Au nanoparticles and the oxide support,41
which must be reached to allow species catalytic conversion.
As a summary for phenomenon occurring until 45 min TOS,
eq 2 indicates that the transformation of type I methoxy is
accompanied by both H2O production and Ce4+ reduction into
Ce3+ (with a concomitant H2 production, as previously evi-
denced33), the whole being consistent with our data.
the direct formate oxidation (over Ce4+ or Ce3+) requiring in
one single step both surface dehydroxylation and neighboring
cerium cation reduction, thus implying a global order of reaction
higher than one according to eq 4.
(e)
HCOO- + OH- + 2Ce4+ 98 CO2 + H2O + 2Ce3+
(4)
One way to determine which of the two proposed formate
transformations is valid consists of tracking of the CO molecule,
which is only involved in the decomposition mechanism (eq
3). As expected according to the very high CO oxidation reaction
kinetics, the analysis of the gas-phase data never allowed one
to give any evidence of carbon monoxide at the reactor outlet;
however, focusing on the catalyst surface during reaction, the
presence of CO adsorbed on gold nanoparticles was detected.
Figure 6 illustrates that the main bands that are observed in
the 2300-2000 cm-1 region on both CeO2HSA and Au/CeO2HSA
are located around 2280, 2200, and 2050 cm-1 (with slight shifts
depending on the sample) and should be assigned to methoxy
ν(CO) overtones and combination modes. This assignment is
consistent with higher band intensities for the CeO2HSA sample,
which was shown to reach surface methoxy saturation, whereas
the Au/CeO2HSA sample transforms methoxys into formates.
For type II methoxy species, the mechanism is practically
the same, but will not be detailed here for the sake of brevity.
Note that in our case where the working atmosphere is oxygen
containing, H2 is consumed in the catalytic cycle and is not
detected in the gas exhaust stream.
(iii) Finally, for time on stream above 45 min, the conversion
remains stable around 50% (Figure 3); the complete catalytic
(43) Louis, C. In Nanoparticles and Catalysis; Astruc, D., Ed.; Wiley-
VCH: Weinheim, 2008; pp 475-503.
9
10838 J. AM. CHEM. SOC. VOL. 132, NO. 31, 2010

Investigation of Methanol Oxidation
A R T I C L E S
CO and leave an OH group on the surface, which comes back
to its initial state. CO is further oxidized to CO2 on gold
nanoparticles, while the Ce3+ cations are reoxidized to Ce4+
thus closing the catalytic loop (Figure 7, middle).
,
(iii) Type I species undergo the same transformation in the
presence of oxygen (Figure 7, top right). They also react in
anaerobic conditions; however, their oxidation stops after
formates formation on Ce3+ (Figure 7, top left) when all of the
available mobile oxygen atoms provided by the support have
been consumed.
3
For the sake of clarity, we should underline that the /2H2,
1
1/2H2, or /2O2 species mentioned in the mechanism might be
only entities necessary to equilibrate the reaction steps. In reality,
these entities are probably provided by surface OH groups or
capping oxygen atoms, but it would be too complicated to fully
detail the mechanism.
Finally, emphasis should be made on the fact that for both
type I and type II methoxy oxidation into formates, the proposed
mechanism involves the participation of active atomic oxygen
species, which are likely to form at the interface between gold
nanoparticles and the ceria support.42 This remark opens the
door to propose an explanation for the Au/TiO2 lack of activity
in methanol oxidation at room temperature without contradicting
its high efficiency in CO oxidation. Indeed, literature data
relative to complementary oxygen isotope exchange and ESR
experiments indicate that an Au/TiO2 sample that is highly
efficient in CO oxidation does not lead to any O2 dissociation
to atomic oxygen, but that on the contrary O2 adsorbs onto TiO2
Figure 6. FTIR subtraction spectra (the spectrum of the surface before
reaction was subtracted for clarity seek) of adsorbed species on (s) Au/
CeO2HSA catalyst and on (- - -) CeO2HSA upon increasing time on stream:
(a) 5 min, (b) 10 min, (c) 20 min, (d) 30 min, (e) 60 min, (f) 90 min, (g)
120 min, (h) 150 min, and (i) 180 min. The right part is an enlargement for
both samples after 120 min. GHSV ) 60 000 h-1, 700 ppm CH3OH +
20% O2 in Ar, T ) 323 K.
oxygen vacancies to form active superoxide O2 species.45
-
Considering the enlargement on the right part of Figure 6, the
presence of a minor band at 2091 cm-1 is detected, which is
only present for the gold-containing sample. According to the
literature data,44 this band should be assigned to CO adsorbed
on gold nanoparticles possessing a metallic character. The very
low intensity of this 2091 cm-1 band is consistent with a very
low surface concentration of highly active intermediate CO-
adsorbed species. Because adsorbed CO species were observed,
the formate decomposition mechanism is proven to occur in
our reaction conditions, and even if it is impossible to reject
the direct oxidation pathway, we propose that the latter is minor.
Furthermore, the direct oxidation pathway (eq 4) should give
rise to CO2 production even in the absence of oxygen, which
was not the case, as observed in Figure 3C.
At this step of the study, enough information were obtained
to propose a full description of the whole catalytic loop for the
Au/CeO2 catalyst. As depicted in Figure 7, the surface of the
catalyst is initially covered by water and carbonate impurities;
upon methanol dissociative chemisorption, water is desorbed
and methoxys cover the surface. Three types of species are
mainly observed: I, II, and II′. These species do not react
anymore on the bare support at room temperature, whereas they
undergo different transformations in the presence of gold
nanoparticles.
3.5. Kinetics Considerations. Preliminary kinetics consider-
ations are provided in the Supporting Information so that only
the main results are reported here. Figure 8 indicates that an
increase of the reaction temperature from 298 to 323 K at
constant space velocity leads to an increase (around 50%) of
the amount of adsorbed formate species. This seems a possible
confirmation that formates are active intermediate species whose
transformation is the rate-determining step. In fact, the steady-
state concentration of formates on the surface is the result of
their formation (methoxy conversion into formate) and their
consumption (formate oxidation into CO2). Both kinetic con-
stants increase with higher temperature (Arrhenius law); there-
fore, if the formate concentration is enhanced, their formation
increases faster than their oxidation, thus confirming the HCOO-
oxidation as the rate-determining step.
The effect of the reaction temperature (at constant space
velocity) on the conversion over Au/CeO2HSA was described
from Figure 2: at the steady state, the conversion increases from
15% at 298 K to 50% at 323 K. The activity was thus multiplied
by more than a factor 3, while the active intermediate formate
concentration was only multiplied by a factor 1.5. This result
was expected due to the variation of the kinetic constant with
the temperature, and an estimation of the apparent activation
energy Ea of methanol oxidation over Au/CeO2HSA sample was
calculated according to eq 5:
(i) Type II′ methoxy species do not participate in any
oxidation step, but they simply accumulate on the surface. On
this basis, they can be considered as spectator species (bottom
of Figure 7).
(ii) Type II species only react in the presence of oxygen: they
give rise to formates, first on Ce4+ and then on Ce3+ cations,
which constitute the redox pair necessary for the methoxy to
formate oxidation. Successively formates decompose to desorb
as(T1) [HCOO-]T2
T1T2
Ea ) -ln
.
R
(5)
-
as(T2)
(
)
(T2 - T1)
[HCOO ]T1
where as(Ti) stands for the specific activity at the temperature
Ti, [HCOO-]Ti stands for the formate surface concentration at
(44) Henao, J. D.; Caputo, T.; Yang, J. H.; Kung, M. C.; Kung, H. H. J.
Phys. Chem. B 2006, 110, 8689–8700.
(45) Liu, H.; Kozlov, A. I.; Kozlova, A. P.; Shido, T.; Asakura, K.;
Iwasawa, Y. J. Catal. 1999, 185, 252–264.
9
J. AM. CHEM. SOC. VOL. 132, NO. 31, 2010 10839

A R T I C L E S
Rousseau et al.
Figure 7. A sketch of the catalytic cycle for methanol oxidation over Au/CeO2 sample. Reaction mechanism for the methoxy formation (type I, upper part;
type II, middle part; type II′, bottom part), and then oxidation in the absence (left) or in the presence (right) of oxygen.
the temperature Ti, and R ) 8.314 J K-1 mol-1. The obtained
presents an apparent activation energy of 138 or 54 kJ mol-1
depending on the metal particle size.47 In both cases, the reported
Ea values are much higher than the one obtained in the present
study for a slower step. However, our Ea value (29 kJ mol-1) is
of the same order of magnitude (even slightly lower) that the
one reported (40 kJ mol-1) for the same rate-determining step
over similar Au/CeO2 catalysts.41 These comparisons with
apparent activation energy for the process is Ea ) 29 kJ mol-1
,
which is a relative small value when compared to some literature
data. For example, iron formate oxidation with O2 is reported
46
to require an activation energy of 238 kJ mol-1
.
Elsewhere,
Gates et al. reported that CO oxidation on gold-based catalysts
(46) Nobukawa, T.; Yoshida, M.; Kameoka, S.; Ito, S.; Tomishige, K.;
Kunimori, K. J. Phys. Chem. B 2004, 108, 4071.
(47) Aguilar-Guerrero, V.; Gates, B. C. J. Catal. 2008, 260, 351–357.
9
10840 J. AM. CHEM. SOC. VOL. 132, NO. 31, 2010

Investigation of Methanol Oxidation
A R T I C L E S
period required to observe real catalytic CO2 production and
the complementary experiments conducted in anaerobic condi-
tions proved the relevance of the gold-support interface to
initiate the low temperature catalytic reaction and the peculiarity
of ceria as a redox support. In fact, no methanol oxidation was
observed for gold over TiO2 in such conditions, likely due to
the absence of mobile atomic oxygen species on the latter
compound surface.
We have also highlighted that the specific area of the ceria
support is an important parameter to reach the steady-state
conditions for the reaction: the methoxy surface coverage until
around the monolayer appears necessary to initiate the CO2
production, thus indicating that only a minor part of them are
in the close neighborhoods of active gold-support interface
sites. We must stress that it is really thanks to the spectroscopic
survey of the catalysts under duty that we have been able to
overcome the challenging experimental conditions, which would
have impeached a traditional mechanism study from bare
gaseous data. Moreover, not only the nature of adsorbates
could be identified but also the study of their evolution versus
time allowed one to distinguish whether they behaved as
spectators or intermediates. However, intermediates’ evolu-
tion may be different depending on their position in the
pathway so that only the kinetic treatment of the spectroscopic
data allowed one to point out formate decomposition as the
rate-determining step.
Figure 8. Effect of the reaction temperature (A) T ) 298 K and (B) T )
323 K on the amount of adsorbed formate species on Au/CeO2HSA during
reaction. The IR spectra are given for the following time on stream: (a) 5
min, (b) 10 min, (c) 20 min, (d) 30 min, (e) 60 min, (f) 90 min, (g) 120
min, (h) 150 min, and (i) 180 min. GHSV ) 60 000 h-1, 700 ppm CH3OH
+ 20% O2 in Ar.
literature data thus clearly confirm that our Au/CeO2HSA catalyst
is very active for formate transformation and thus for methanol
oxidation, when present as traces in ambient air even at room
temperature.
4. Conclusions
Finally, a comparison with some literature data highlighted
the outstanding performances of the investigated Au/CeO2
materials as catalysts for VOC removal at ambient conditions
and permits us to propose them as reliable devices for confined
atmosphere purification.
In the present article, FTIR spectroscopy coupled with mass
spectrometry was used to study the methanol oxidation mech-
anism at low temperatures on nanostructured Au/CeO2 and Au/
TiO2 catalysts. The unique approach consisting of time-resolved
simultaneous analysis of adsorbed and gaseous species allowed
us to evaluate the activity and selectivity toward CO2 and to
identify consecutive steps in the oxidation pathway. Among all
of the detected adsorbed species, type I and type II methoxys
were evidenced as intermediates, which further transform into
formates, whose oxidation was found to be the rate-determining
step for the reaction. On the contrary, type II′ methoxys
(possessing an oxygen vacancy in the neighborhoods) were
proved to act as bare spectator species. Regarding the nature of
the active sites, experiment on the pristine support proved
without any doubt the necessity of gold to observe catalytic
oxidation at room temperature. Furthermore, both the induction
Supporting Information Available: Experimental section
including sample preparation procedure and conditions for
catalytic operando tests. Results and discussion section contain-
ing TiO2-based samples catalytic activity and estimation of the
methoxy monolayer over CeO2HSA; description of the ν(CH)
vibration range relative to Figure 4 and estimation of the gold
surface concentration; and hypotheses used for kinetic calcula-
tions. This material is available free of charge via the Internet
at http://pubs.acs.org.
JA1028809
9
J. AM. CHEM. SOC. VOL. 132, NO. 31, 2010 10841