Study of the origin of the deactivation of a Pt/CeO2 catalyst during reverse water gas shift (RWGS) reaction

Article abstract of DOI:10.1016/j.jcat.2004.06.011

The effect of carbon deposition in the deactivation of a 2% Pt/CeO₂ under RWGS conditions was studied via an accelerated aging procedure consisting of measuring the catalyst activity (under 1% CO₂, 4% H₂, 300°C) before and after exposure to CO, CO₂, H₂, or CH₄ at 400°C. Exposure to CO led to severe deactivation while exposure to CO₂, CH₄, or H₂ led to moderate or no deactivation. Experiments consisting of measuring catalytic activity before and after exposure of the catalyst to CO followed by subsequent TPO experiments showed that exposing the catalyst to increasing amounts of CO led to increasing deposits of carbon and increasing loss of activity for the RWGS reaction. A clear correlation was found between the percentage of deactivation and the amount of carbon deposited. During the course of the RWGS reaction on this catalyst, carbon deposit took place and was thus one of the major reasons for the slow deactivation that was observed for 2% Pt/CeO₂.

Full text of DOI:10.1016/j.jcat.2004.06.011

Journal of Catalysis 226 (2004) 382–392
www.elsevier.com/locate/jcat
Study of the origin of the deactivation of a Pt/CeO catalyst
2
during reverse water gas shift (RWGS) reaction
a,∗
a
a
a
b
c
A. Goguet , F. Meunier , J.P. Breen , R. Burch , M.I. Petch , A. Faur Ghenciu
a
School of Chemistry, Queen’s University of Belfast, David Keir Building, Stranmillis Road, Belfast BT9 5AG, Northern Ireland, UK
b
Johnson Matthey Fuel Cells, Blounts Court, Sonning Common, Reading RG4 9NH, UK
c
Johnson Matthey Fuel Cells, 1397 King Road, West Chester, PA 19380, USA
Received 8 March 2004; revised 4 June 2004; accepted 8 June 2004
Available online 10 July 2004
Abstract
The effect of carbon deposition in the deactivation of a 2% Pt/CeO under RWGS conditions was investigated via an accelerated ageing
2
◦
procedure consisting of measuring the catalyst activity (under 1% CO , 4% H , 300 C) before and after exposure to CO, CO , H , or CH at
2
2
2
2
4
◦
4
00 C. Exposure to CO led to severe deactivation while exposure to CO , CH , or H led to moderate or no deactivation. Additional activity
2
4
2
measurements performed before and after exposure to CO followed by subsequent temperature-programmed oxidation (TPO) showed that
exposure to increasing amounts of CO led to increased deposits of carbon and deactivation. No evidence for metal sintering was found.
Additional TPO following catalyst exposure to RWGS conditions evidenced the deposition of carbonaceous deposits under the RWGS
reactant mix.

2004 Elsevier Inc. All rights reserved.
Keywords: Deactivation; Ceria; Reverse water gas shift; Pt; Temperature-programmed oxidation; Carbon deposition; Active site
1
. Introduction
to be removed as it is both a pollutant and a poison for the
platinum electrodes of the fuel cell.
To meet this goal, a water gas shift (WGS) converter, in-
cluding a high- and a low-temperature stage, is added after
the reformer to convert carbon monoxide with water to hy-
drogen and CO2. The WGS converter is generally followed
by an additional preferential oxidation stage to reduce the
CO level to about 10 ppm [5].
This new interest in the application of the WGS reaction
to fuel-cell applications has led to extensive research on im-
proved catalysts for the kinetically limited low-temperature
part of the WGS reaction. Among all the areas of research
investigated, one solution came from earlier work on three-
way catalysts and the observation of the effect of the addi-
tion of ceria to these catalysts. As a matter of fact, when
used with platinum group metals, it was observed that ceria
greatly promotes the WGS reaction [6–12].
In the past few years, much research has been devoted
to the development of fuel-cell power generation for trans-
portation applications. A main focus of the research has been
on providing the necessary hydrogen feedstock for the fuel
cell. Technical solutions range from liquid H2 tanks, high-
pressure H2 storage, medium-pressure H2 storage on hy-
drides or nanostructured carbons to on-board H2 production.
Among the technical solutions offered, the on-board reform-
ing of hydrogen-rich fuels such as natural gas, gasoline, or
methanol has the most potential [1–4].
Nevertheless, this solution also presents a challenging
drawback coming from the fact that the reformate gas al-
ways contains a fraction of carbon monoxide. This CO needs
*
Consequently, several metal/ceria catalysts were investi-
gated and some of them (e.g., Pt/CeO2) showed activity ex-
Corresponding author. Fax: +44 (0) 2890382117.
E-mail address: a.goguet@qub.ac.uk (A. Goguet).
0021-9517/$ – see front matter  2004 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2004.06.011

A. Goguet et al. / Journal of Catalysis 226 (2004) 382–392
383
ceeding that of conventional Cu/ZnO shift catalysts [5–17].
However, these investigations also revealed that the activity
and stability of these catalysts were closely linked to their
preparation conditions [12,18,19]. These observations un-
derlined the fact that a clear understanding of the reaction
mechanism taking place on these catalysts was compulsory
in order to develop improved ceria-based catalysts.
Surprisingly, only a few studies have been aimed at ad-
dressing this crucial question for ceria-supported catalysts.
The extent of this issue led Zalc and co-workers [5] to claim
that ceria-supported metal catalysts may not be practical for
the WGS reaction because of their high rate of deactivation.
Based on a study performed with Pt-based catalysts, Zalc
and co-workers proposed that the deactivation is due to irre-
To achieve this, several mechanistic studies were under-
taken and two possible mechanisms were proposed. Gorte
and co-workers [12,13,18,19] and Li and co-workers [20]
proposed a redox mechanism mainly based on kinetic stud-
ies.
versible overreduction of ceria by H . Gorte and co-workers
[17] performed an accelerated aging study of Pd/ceria and
2
Pt/ceria and stated that deactivation was due to the growth
of the metal particles under a CO-containing atmosphere
and discarded deactivation by overreduction. Gorte and co-
workers also proposed that the recovery of part of the activity
after reoxidation could be the consequence of the redisper-
sion of the metal under oxygen thermal treatment. In the
present study of the reverse water gas shift reaction, the
contribution of carbon deposition to the deactivation of a
In another approach, Shido and Iwasawa [7] performed
infrared and TPD studies on ceria-supported Rh. These au-
thors observed the formation of several surface species under
reaction conditions (formates, carbonyls, and carbonates)
and proposed that terminal hydroxyl groups on reduced ce-
ria react with CO to produce adsorbed bidentate formate
species. In the presence of water, these formates decompose
to produce, in a first step, H2 and unidentate carbonates. In a
second step, unidentate carbonates decompose to CO2 and a
hydroxyl group is regenerated. The rate-determining step is
proposed to be the decomposition of this formate to H2 and
a unidentate carbonate. In this mechanism it seems that the
function of the metal is to facilitate the reduction of the ceria
and the production of surface hydroxyl groups. This mech-
anism was supported by recent work performed by Jacobs
and co-workers [21,22].
More recently, Fu and co-workers [23] published an inter-
esting study that brought new insight into the effective role
played by the metal in these ceria-supported catalysts. These
authors removed most of the metal from ceria-supported
gold and platinum catalysts (to the point of removing all
the TEM observable particles) and showed that this treat-
ment did not affect the activity of the catalyst compared to
the untreated ones. Fu and co-workers proposed that part of
the metal, in the form of ions, diffused into the ceria and
anchored to surface defects. These ions were proposed to be
responsible for the increase in the reducibility of the support.
The remaining part of the metal in the form of nanoparticles
was considered to be unimportant in the WGS reaction.
Fu and co-workers advocated a mechanism similar to the
Pt/CeO catalyst has been investigated. Carbon deposition
2
through the decomposition of CO (the Boudouard reaction)
is well known with platinum group metals and we will
present results that indicate that this could be an important
factor under our experimental conditions:
0
−1
.
2
CO → C + CO2, ꢀH = 86.0 kJ mol
The results obtained are discussed in terms of both the mech-
anism responsible for the catalyst deactivation and its influ-
ence on the RWGS mechanism. The effect of carbonaceous
deposits on the activity of the Pt/ceria catalyst for the WGS
reaction will be reported in a subsequent paper.
2
. Experimental
The catalyst used in this study was a 2% Pt/CeO2 pro-
vided by Johnson Matthey p.l.c. The specific surface area
was measured by BET (Micromeretics ASAP 2010) and was
2
−1
found to be 180 m g . The Pt dispersion was measured
by H2 chemisorption (Micromeretics Autochem 2910) and
was found to be 17%. Note that the dispersion measure-
◦
ments were performed at −80 C to minimize the potential
spillover of H2 onto the support [26]. For all the experiments
described below, the gases (H2, He, CO2, CO, and CH4)
were >99.9% purity and were supplied by BOC.
“
normal-support activation” effect summarized by Golun-
ski and co-workers [24]. In this effect, the insertion of a
metal into a reducible support leads to a modification of the
reducibility of this support through the so-called “junction
effect”, that was first proposed by Frost [25] for methanol
synthesis catalysts. The most striking effect observed with
the NSA is a reversal of the normal roles of the metal oxide
and the precious metal through an encouragement of the for-
mation of highly active sites on the phase that normally acts
as the support.
In addition to the reaction mechanism insight, a better un-
derstanding of the mechanisms responsible for the catalyst
deactivation is also compulsory for further catalyst improve-
ment.
2
.1. Study of the ageing under CO, CO2, CH4, and H2
The deactivation of the 2% Pt/CeO2 catalyst was a slow
process under our typical reaction conditions (see Table 1).
Table 1
Conversion of CO as a function of time for 40 mg of 2% Pt/CeO under
RWGS reaction conditions: 290 C with 1% CO and 4% H , balance He
at 200 cm min
2
2
◦
2
2
3
−1
Time (h)
0
20
40
CO conversion (%)
21.7
20.6
19.1
2

384
A. Goguet et al. / Journal of Catalysis 226 (2004) 382–392
Fig. 1. Typical experimental cycle for ageing investigations of 2% Pt/CeO for the RWGS reaction.
2
Therefore, we decided to apply an accelerated ageing
procedure inspired by the one developed by Gorte and co-
workers [14].
The principle of our procedure consisted of measuring the
catalyst activity before and after exposure to pure CO, CO2,
H2, or CH4. After the second activity test the catalyst was
reoxidized and a third activity test was performed. This third
test was performed to investigate possible recovery of the
activity.
The experimental setup consisted of an in situ high-
temperature diffuse–reflectance IR cell (from Spectra-Tech)
fitted with ZnSe windows. The cell was connected to H2, Ar,
and CO2 gas cylinders through low-volume stainless-steel
lines. The gas flows were controlled by Aera mass-flow con-
trollers, which were regularly calibrated. The DRIFTS cell
was installed in a Bruker Equinox 55 spectrometer, operat-
−1
ing at a resolution of 4 cm
.
The amount of 2% Pt/CeO2 catalyst used was about
30 mg (particle diameter <150 µm). Typically, the cata-
lyst was placed in a ceramic crucible and reduced in situ
Typically, 20 mg of catalyst (2% Pt/CeO2) was placed
in a 1 -inch tubular reactor and reduced at 300 C under H2
◦
4
◦
3
−1
at 300 C in pure H . Following this step, the sample was
(
200 cm min ) for 1 h. The catalyst was then exposed to
2
submitted to the reaction mixture (1% CO2 + 4% H2,
the RWGS reaction mixture (1% CO2 + 4% H2, balance He,
3
−1
◦
3
−1
◦
50 cm min ) for 2 h at 300 C. After 2 h, the reactor
was cooled down to room temperature, the flowing gas was
switched from the reaction mixture to a flow of 20% O2/Ar,
2
00 cm min ) for 2 h at 300 C. Following this step, the
◦
temperature was increased to 400 C and the catalyst was ex-
posed to either pure CO, H2, CH4, or CO2 for 1 h. After the
exposure, the temperature was brought back to 300 C and
◦
and the temperature was increased to 600 C at a rate of
◦
◦
−1
5
C min . During the temperature ramp, an IR spectrum
a new catalytic test was performed under the RWGS con-
ditions. Following this test, the catalyst was reoxidized at
was recorded every 3 min.
◦
◦
4
00 C under 20% O2/Ar and then tested again at 300 C
2
.2.2. TPO after exposure to increasing amounts of CO
under RWGS conditions (see Fig. 1 for details).
The concentrations of the reactants and products were de-
termined using an on-line gas chromatograph (Perkin-Elmer
700), equipped with a Hayesep Q column and a FID, fitted
with a methanator to allow the detection of CO and CO2.
These experiments had a structure similar to the ones de-
scribed in Fig. 1. The difference came from the fact that a
TPO step experiment was added between the zones B and C
instead of a simple reoxidation.
8
Typically, the catalyst activity was measured before and
after exposure to various amounts of CO (controlled by vary-
ing the percentage of CO in the gas mix (1, 10, or 100%) as
well as the time of exposure (5 to 1440 min)). Then, TPO
experiments were performed by ramping from room tem-
2
.2. Temperature-programmed oxidation (TPO)
investigations
◦
◦
−1
2
.2.1. TPO with DRIFT
perature up to 400 C (5 C min ) under 1% O /He. The
2
A first TPO of the 2% Pt/CeO2 previously exposed to the
concentrations of CO and CO formed were followed by gas
2
RWGS reaction mixture (1% CO2 + 4% H2) was performed
in a DRIFT apparatus. This experiment allowed us to deter-
mine the temperature range over which IR-sensitive surface
species were desorbed/decomposed under the TPO condi-
tions.
chromatographywith a point recorded every 2 min. The TPO
experiments were followed by a third activity test in order to
investigate possible activity recovery.
Two additional experiments were performed with a 2%
Pt/SiO2 catalyst and a pure CeO2 as references. They al-

A. Goguet et al. / Journal of Catalysis 226 (2004) 382–392
385
lowed us to investigate the decoupled behavior of the metal-
lic phase (silica is an inert support) and the ceria support.
These catalysts were submitted to the same experimental
cycle as that described for the ceria-supported platinum cat-
alyst. The exposure to CO was performed using 10% CO/He
for 120 min. These conditions were chosen because they cor-
responded to the conditions under which the 2% Pt/CeO2
displayed the maximum loss of activity.
of deactivation were calculated with the following formula:
ꢀ
ꢁ
Conv.zone B or C
Conv.zone A
% Deactivation = 100 − 100 ×
.
Note that the reoxidation step was performed under 20%
O2/Ar for all experiments except for the experiment involv-
ing the exposure to CH4 where it was carried out under 1%
O2/Ar.
It can be seen that an exposure to pure CO led to a se-
vere deactivation (86%) and that a subsequent reoxidation
allowed only a partial recovery of the activity. An exposure
to either pure CO2 or pure CH4 led to a moderate loss of ac-
tivity and subsequent reoxidation led to a total recovery of
the initial activity. An exposure to pure H2 had almost no
impact on the catalyst activity. Neither had the subsequent
reoxidation.
2
.2.3. TPO after exposure to increasing time in the RWGS
mixture
TPO experiments were performed after various times of
exposure of the 2% Pt/CeO2 to the RWGS reactant mixture
1% CO2, 4% H2). These experiments were performed in or-
(
der to investigate the buildup of carbonaceous deposits under
true RWGS reaction conditions.
Typically, 40 mg of catalyst was placed in a 5 mm (i.d.)
tubular reactor, reduced at 300 C for 1 h under H2, and
◦
3.2. Temperature-programmed oxidation investigations
then exposed to the RWGS reaction mixture for 5, 24, 72, or
◦
9
6 h at 300 C. Then, the reactor was cooled down to room
3.2.1. TPO with DRIFT
temperature, the reaction mixture was switched to a flow of
Several IR bands were readily observed following the
introduction of the (1% CO2 + 4% H2) reaction mixture
over the 2% Pt/CeO2 sample. The bands formed were char-
◦
1
5
% O2/He, and the temperature was increased to 500 C at
C min . The CO and CO2 concentrations were followed
◦
−1
−1
at the outlet of the reactor with a quadrupole mass spectrom-
eter (VG Gaslab300).
acteristic of surface formates (e.g., 2950 and 2830 cm ),
−
1
carbonates (e.g., 866 and 851 cm ), and carbonyls (e.g.,
−1
It is worth noting that the deactivation under these condi-
tions was very slow. Therefore, the aim of these experiments
was only to investigate the formation of carbonaceous de-
posits under reaction conditions. No attempt was made to
evaluate the extent of deactivation as it was too small to be
accurately measured.
2065 cm ) as already described in detail by Shido and
co-workers [7], Li and co-workers [27,28], Binet and co-
workers [29], and Bourane and co-workers [30]. The inten-
sity of all the IR bands reached a steady-state value within
1
0–20 min on stream.
Fig. 2a reports the spectra recorded during the tempera-
ture ramp of the TPO experiment. Note that the first and the
second spectra were recorded at room temperature before
and after the admission of the 20% O2/Ar onto the catalyst,
respectively. Fig. 2b reports a plot of the area of bands as-
3
. Results
−1
sociated with surface formates (2830 cm ) and carbonates
3
.1. Study of the ageing under CO, CO2, CH4, and H2
−
1
(
866 and 851 cm ) as a function of the reactor temperature.
The band representative of the carbonyl species (2065
For all the experiments, the initial CO2 conversion of the
cm 1) disappeared as soon as the 20% O2/Ar was admit-
−
fresh catalyst (zone A) was always close to 19%. Table 2
reports the percentage of deactivation between the fresh cat-
alyst and the catalyst exposed to CO, CO2, H2, or CH4
ted to the reactor at room temperature. It is worth noting that
◦
a temperature increase of 50 C was observed in the catalyst
bed when switching to 20% O2/Ar even though the reactor
temperature was set at room temperature. This is likely to be
due to the exothermic oxidation reaction of the CO adsorbed
on Pt. In Figs. 2a and 2b, it can be clearly seen that formate
(
zone A/zone B) as well as between the fresh catalyst and
the reoxidized catalyst (zone A/zone C). These percentages
Table 2
−1 −1
(
2830 cm ) and carbonate band (866 and 851 cm ) inten-
Percentage deactivation of the 2% Pt/CeO catalyst for the RWGS reaction
2
3
−1
◦
sities decrease rapidly with increasing temperature; 95% of
the bands attributed to the carbonate and formate species has
(
1% CO + 4% H , balance He, 200 cm min ) at 300 C, before and
2
2
◦
after exposure to pure CO, H , CH , or CO for 1 h at 400 C and after
2
4
2
◦
◦
reoxidation at 400 C under 20% O /Ar
disappeared before 250 C. Note that small negative bands
2
were observed in the carbonate region at higher tempera-
tures. These bands probably correspond to carbonates on
strong basic sites.
Species used
for ageing
before zone B
% Deactivation
between zone A
and zone B
% Deactivation
between zone A
and zone C
CO
86.0
11.6
2.4
27.9
3.0
4.7
CO
2
3.2.2. TPO after exposure to increasing amounts of CO
Fig. 3 reports the activities obtained during a typical ex-
perimental cycle (see Fig. 1). In this example, the catalyst
H
2
CH
19.8
−11.5
4

386
A. Goguet et al. / Journal of Catalysis 226 (2004) 382–392
(a)
(b)
Fig. 2. (a) 3D plot of the spectra recorded during the temperature ramp of the TPO experiment. (b) Plot of the area of the bands representative of surface (×)
−1
−1
formates (2830 cm ) and (ꢀ) carbonates (866 and 851 cm ) as a function of the reactor temperature during the TPO performed in the DRIFT apparatus.
was exposed to 10% CO/He for 45 min between zones A
and B. Table 3 summarizes the results of all the experiments
performed. This table reports the percentage of deactivation
between the fresh catalyst and the catalyst exposed to differ-
ent amounts of CO (zone A/zone B) as well as the percentage
of deactivation between the fresh catalyst and the reoxidized
catalyst (zone A/zone C).
It was seen that an exposure to CO always led to a notice-
able deactivation of the catalyst. Increasing the amount of
CO introduced to the catalyst led to increased deactivation.
The subsequent reoxidation under 1% O2/He always led to
an almost total recovery of the initial activity. Note that the
extent of the recovery of the activities was different from the
one reported in Table 2 where only partial recovery was ob-
◦
Fig. 3. Conversion of CO for 2% Pt/CeO at 300 C for 1% CO , 4% H .
2
2
2
2
served. This difference was apparently due to the different
reoxidation conditions used in each experiment. This differ-
ence will be discussed later.
In this example, the catalyst was exposed to 10% CO/He during 45 min
between zones A and B.

A. Goguet et al. / Journal of Catalysis 226 (2004) 382–392
387
Fig. 4. TPO profiles for the 2% Pt/CeO exposed to 10% CO for (a) 126 min, (b) 45 min, (c) 15 min, and (d) blank experiment (no catalyst and flowing of
2
10% CO for 60 min between zones A and B).
Table 3
with the temperature range of the peak obtained with the
TPO.
Percentage of deactivation of the 2% Pt/CeO catalyst for the RWGS reac-
tion as a function of the amount of CO introduced to the catalyst
2
◦
The second peak at 260–290 C was attributed to the oxi-
Time
of exposure
% of CO
of the gas
mixture
% Deactivation
between zone A
and zone B
% Deactivation
between zone A
and zone C
dation of carbon deposits because: (1) the temperature of the
maximum of this peak was above the temperature of desorp-
tion/decomposition of the carbonyl, formates, and carbon-
ates (see Figs. 2a and 2b), (2) the intensity of this peak varied
substantially with the time of exposure to CO, indicating an
accumulation of an oxidizable form of carbon.
(min)
3
8
00
40
1
1
1
13
10
16
24
42
54
51
78
77
70
74
2.5
1.6
3.0
0.1
0.1
3.3
2.6
5.0
3.9
1440
10
15
20
45
10
10
10
10
10
100
100
100
This attribution was further confirmed by the result of a
◦
TPO following exposure of the catalyst at 400 C to CH4
instead of CO. This TPO showed a large peak of CO2 at
126
◦
2
50–280 C (Fig. 5) that is located at the exact same position
5
as the second peak observed during the TPO of the cata-
lyst exposed to CO. These results corresponded well with
the fact that an exposure of PGM catalysts to CO or CH4
leads to carbon deposition through CO disproportionation or
1
3
5
0
7.1
−10.2
Fig. 4 reports an example of the profiles of the CO2 con-
CH cracking.
4
centration recorded during the TPO step of the experiment
cycles. Note that no CO was detected in any of the TPO per-
formed. The example given in Fig. 4 corresponds to results
obtained with the catalyst exposed to 10% CO for 15, 45,
and 126 min.
Fig. 6 reports the plot of the percentage of deactivation
(reported in Table 3) against the area of the peak at 260–
◦
290 C obtained during the TPO step of each experiment
cycle.
It can be seen that the percentage of deactivation in-
creased almost linearly with the amount of carbon deposited
during the first 60–70% of deactivation. Above this value,
a slope break was observed and the extent of deactivation
decreased more slowly with the amount of carbon. This re-
sult indicates a correlation between deactivation and carbon
deposition.
The profiles of all the TPO experiments were similar to
the ones reported in Fig. 4. They consisted of two peaks at
◦
1
00–110 and 260–290 C, respectively.
◦
The peak at 100–110 C was identical in shape and in-
tensity for all the TPO experiments. Therefore, this peak
corresponds to species that are not affected by prolonged ex-
posure to CO and associated with the deactivation. These
species are probably formates and carbonates. This attribu-
tion was further confirmed by the TPO performed with the
DRIFT apparatus. Indeed, the range of temperature during
Regarding the results obtained with 2% Pt/SiO and the
2
CeO support, the catalytic tests showed that these catalysts
2
had poor activity for the RWGS reaction under the condi-
tions explored. The conversions of CO obtained in zone A
2
◦
which formates and the carbonates desorbed/decomposed
(fresh catalyst reduced at 300 C) were ꢁ1% for the ceria
◦
(
between room temperature and 250 C) corresponded well
and ꢁ2% for the 2% Pt/SiO2.

388
A. Goguet et al. / Journal of Catalysis 226 (2004) 382–392
◦
Fig. 5. TPO experiments obtained after exposure of the 2% Pt/CeO to (a) pure CH for 60 min or (b) 10% CO/He for 10 min at 400 C between zones A
2
4
and B.
Fig. 7 reports the TPO profiles obtained for 2% Pt/SiO2
and CeO2 after exposure to 10% CO for 126 min. The profile
obtained with the 2% Pt/CeO2 is included for comparison.
This profile was truncated for clarity purpose (for full scale
see Fig. 4). The profile obtained with the ceria presented two
◦
peaks: one at 100–110 C, similar to the one observed for
2
% Pt/CeO2 and a second peak, very small and very broad
◦
in the range 260–290 C. The profile obtained with the 2%
Pt/SiO2 showed a small and poorly defined peak that seemed
to extend between 50 and 300 C. We estimated that the C/Pt
◦
ratio is about 0.5 which is what would be expected for carbon
deposition on Pt.
3
.2.3. TPO after exposure to increasing time in the RWGS
mixture
Fig. 8 reports a typical result obtained during the TPO
Fig. 6. Percentage of deactivation between zones A and B against the
amount of carbon deposited on the 2% Pt/CeO surface (area of the peak at
2
experiments after the exposure of the catalyst to the RWGS
reactant mixture (72 h exposure at 300 C). Note that only
◦
260–290 C).
◦
Fig. 7. TPO experiments obtained after exposure to 10% CO for 126 min for (a) 2% Pt/CeO , (b) CeO , and (c) 2% Pt/SiO .
2
2
2

A. Goguet et al. / Journal of Catalysis 226 (2004) 382–392
389
◦
Fig. 9. Amount of CO released during the peak at 270 C against the length
2
◦
of exposure of the 2% Pt/CeO to the RWGS reactant mixture at 300 C
Fig. 8. TPO experiment following 72 h exposure of the 2% Pt/CeO to the
2
2
◦
(1% CO , 4% H ).
RWGS reactant mixture at 300 C (1% CO , 4% H ).
2
2
2
2
CO led to increasing deposits of carbon and increasing de-
activation of the 2% Pt/CeO2.
the mass 44 representative of CO2 is reported. No contribu-
tion for CO was observed.
The plot of the percentage of deactivation against the
area of the peak characteristic of the carbon deposits clearly
showed that there is a correlation between the amount of
carbon deposited on the surface of the catalyst and its deac-
tivation. This correlation was almost linear for deactivation
up to 60–70%. Above this value, a change of slope could be
observed.
All the TPO profiles (entailing 5, 24, 72, or 96 h exposure
at 300 C to the RWGS reaction mixture) displayed similar
◦
features: a first very sharp peak at room temperature, a sec-
◦
ond large and broad peak centered around 100 C, and a third
◦
peak around 270 C. It was possible to attribute the different
peaks observed by comparing these results to the results pro-
vided by the DRIFT experiment and the results of the TPO
following CO exposure.
The results of the experiments performed with unpro-
moted ceria and 2% Pt/SiO2 seem to indicate that the car-
bon deposition is located on the support (ceria) and not
only the platinum. As a matter of fact the TPO performed
with the pure ceria exhibited a small but significant peak at
The first very sharp peak corresponds to the oxidation of
the carbonyl (note that, in Fig. 8, this peak was truncated
◦
for clarity). The second large and broad peak (at 100 C) is
attributed to the decomposition of the formate and carbon-
◦
◦
about 260–290 C characteristic of carbon deposits. Under
ate species. The third peak around 270 C, that appeared to
the same conditions, the TPO of the 2% Pt/SiO2 showed a
small and broad peak. The amount of carbon deposited was
estimated from the dispersion of the Pt (30%) to correspond
to a C/Pt ratio of about 0.5, assuming that no carbon is de-
posited on the silica.
Additional evidence for the fact that the carbon is de-
posited on the surface of the ceria was given by the amount
of carbon deposited on the 2% Pt/CeO2 during exposure to
10% of CO for 126 min. The carbon to surface platinum ratio
was calculated knowing the platinum dispersion of the cata-
lyst (17%) and was found to be ca. C/Pts = 60. This value,
far above unity, indicates that the carbon could not be lo-
cated only on the platinum and therefore was mainly located
on the ceria surface.
increase in intensity with time of exposure to the RWGS re-
action mixture, was attributed to carbonaceous deposits.
It is worth noting that an additional broad peak located at
◦
about 410 C was also observed. This peak did not increase
in intensity with time on stream. It is possible that this peak
corresponds to carbonates on strong basic sites. This feature
could therefore correspond to the small negative bands that
◦
were observed above 350 C during the DRIFT experiment.
Fig. 9 reports the amount of CO2 released during the peak
◦
at 270 C against the length of exposure of the catalyst to the
RWGS reactant mixture. This result clearly shows that the
“
intensity” of this peak increases with the time of exposure
of the catalyst to the RWGS reactant mixture. Therefore it
can be concluded that during the course of the RWGS on the
2
% Pt/CeO2, deposition of carbonaceous species occurs.
It is interesting to note that the amount of carbon de-
posited on the 2% Pt/CeO2 was larger than the quantity de-
posited on the unpromoted ceria under the same conditions.
This result tends to indicate that the platinum is involved in
the formation of the carbon deposited on the ceria support.
A possible explanation of this observation is that the Pt in-
creases the reducibility of the ceria. TPR results showed that
the addition of a noble metal to the ceria led to an easier re-
duction of the ceria [31,32]. Moreover, some related work
showed that CO disproportionation took place on partially
4
. Discussion
The results of these experiments highlight the potential
role played by carbon deposition in the deactivation of the
% Pt/CeO2 catalyst under RWGS reaction conditions. As
2
a matter of fact, the catalytic tests and the TPO experiments
showed that exposing the catalyst to increasing amounts of

390
A. Goguet et al. / Journal of Catalysis 226 (2004) 382–392
reduced ceria [33,34]. Therefore, when platinum is added
to the ceria, a larger fraction of the support would be in a
reduced state as compared to the pure ceria, leading to in-
creased carbon deposition. Another possible explanation is
that the Pt increases the CO disproportionation rate and that
the carbon is easily transferred from the metal to the support.
Either or both of these effects are likely to be responsible
for the increased amount of carbon deposition on the 2%
Pt/CeO2 compared to the unpromoted ceria.
was clearly demonstrated that the rate of deposition was slow
under our standard conditions which would explain the slow
deactivation rate that was observed for the 2% Pt/CeO2 un-
der these normal reaction conditions. Furthermore, the fact
that the TPO peak attributed to the oxidation of carbona-
ceous species deposited under RWGS reaction conditions is
◦
located in a similar temperature range (260–290 C) as the
peak due to the oxidation of the carbon deposited during the
accelerated ageing in CO suggests that the two carbon de-
posits are of a similar type.
In the present study of the RWGS reaction, carbon depo-
sition could be solely responsible for the deactivation. No
metal sintering seemed to occur for our Pt/ceria catalyst.
This absence of sintering was evidenced by the fact that a
total recovery of the initial activity was obtained after re-
oxidation of the catalyst exposed to CO (see Table 3). Note
that the total recovery of the activity was only obtained when
the reoxidation was performed under “soft” conditions, that
is, 1% O2/Ar with a slow increase of the temperature from
Another interesting piece of information could be ex-
tracted from the quantification of the amount of carbon de-
posited on the catalyst. As a matter of fact, after an exposure
to 10% of CO for 126 min, the amount of carbon deposited
−5
on the catalyst was 2.1 × 10 mol, which led to 78% deac-
tivation of the catalyst (see Fig. 6). Assuming that the carbon
atomic radius was 77 pm and that the carbon was deposited
as a monolayer, this amount of carbon deposit would only
cover 6.5% of the total surface of the CeO2 support (the BET
◦
room temperature to 400 C.
2
−1
When the reoxidation was performed under “hard” con-
surface area of the 2% Pt/CeO2 was 180 m g ). Therefore,
a coverage of only 6.5% of the total surface area of the sup-
port led to almost 80% of deactivation of the catalyst. This
result indicates that the fraction of the support involved in
the mechanism of the RWGS reaction is very small.
◦
ditions with admission of 20% O2/Ar directly at 400 C, the
recovery of the activity was only partial (see Table 2). This
result indicates that the conditions under which the reoxi-
dation was performed were crucial for the recovery of the
activity of our Pt/ceria catalyst. It is proposed that, when the
Moreover, although the critical size of the platinum par-
ticles involved in the RWGS reaction mechanism is not
clear [23], the average radius of the platinum particles in
our sample was calculated from the platinum dispersion (i.e.,
17%), and found to be ca. 3.0 nm. From this average radius
and assuming that the carbon was depositing around the plat-
inum particles, it is possible to calculate the radius of the
“ring” of carbon deposit surrounding each metal particle (see
Fig. 10, R1 for Pt particle radius, R2 for radius of the carbon
ring; the Pt particles were considered to be hemispherical).
The calculation gave a value of R2 equal to 12 nm, which
corresponds to only 4 times the radius of the platinum parti-
cle (note that in the hypothesis of particles equally spaced on
the surface of the support, the distance between 2 particles
would be ∼150 nm). This result would mean that the “active
part” of the support would correspond only to the part that is
very close to the platinum particles.
This hypothesis would therefore explain why the activity
of the unpromoted ceria is so low for the RWGS reaction.
As was noted before, TPR results showed that the addition
of a noble metal to the ceria led to an easier reduction of the
latter. This effect was explained by the fact that the oxygen
was diffusing from the support to the metallic surface, eas-
ing the reaction with the reductant (H2 or CO) adsorbed on
the metallic surface. It is therefore possible to envisage that
the portion of the support close to the metal particle could
be in a highly reduced state (a gradient is created between
the vicinity of the metal particles and rest of the surface of
the support) and would correspond to the active part of the
support (e.g., reactive toward the reduction of CO2).
◦
reoxidation was performed under hard conditions at 400 C,
the instantaneous oxidation of the adsorbed CO and car-
bon deposits induced a strong local temperature increase
that probably led to sintering of the platinum and therefore
to irreversible deactivation. When the reoxidation was per-
formed under soft conditions, the oxidation of the adsorbed
CO and the carbon deposits took place at a lower temper-
ature and the sintering did not take place. The fact that a
total recovery of the activity was obtained under our reoxida-
tion conditions and that many studies stressed that exposing
supported platinum to oxidizing conditions leads to sinter-
ing and not to redispersion [35,36] seems to indicate that the
activity recovery of our Pt/ceria was due to removal of car-
bon deposits from the surface and not to redispersion of a
sintered metallic phase.
Concerning the deactivation by overreduction proposed
by Zalc and co-workers [5], the result of the artificial ageing
under H2 (see Table 2) seems to indicate that this phenom-
enon is not taking place under our conditions. Nevertheless,
our ageing experiments were performed on a short time
scale (1 h) and it is possible that the conditions used were
not severe enough to fully create overreduction. Therefore,
overreduction cannot be completely discarded and a more
extensive study is necessary to fully address this question.
The role of carbon deposition in the deactivation of the
catalyst was further supported by the TPO experiments fol-
lowing exposure of the catalyst to our “standard” RWGS
reaction conditions. These results clearly showed that car-
bon deposition was indeed taking place under our normal
RWGS reaction conditions, as it did to a larger extent during
the harsher accelerated ageing conditions reported above. It
In this hypothesis, the slope change observed on Fig. 6
(plot of the percentage of deactivation against the amount

A. Goguet et al. / Journal of Catalysis 226 (2004) 382–392
391
5
. Conclusion
Carbon deposition plays a major role in the deactivation
of the 2% Pt/CeO2 catalyst under RWGS reaction condi-
tions. Experiments consisting of measuring catalytic activity
before and after exposure of the catalyst to CO followed
by subsequent TPO experiments showed that exposing the
catalyst to increasing amounts of CO clearly led to increas-
ing deposits of carbon and increasing loss of activity for the
RWGS reaction.
A clear correlation was found between the percentage of
deactivation and the amount of carbon deposited. Moreover,
the results obtained seemed to indicate that the carbon de-
position is located primarily on the support (ceria) and not
on the platinum. This result indicates that the support was
“
active” in the reaction and that the deactivation took place
through gradual coverage of the support. Further informa-
tion about the fraction of active support was also revealed.
It was found that a very small coverage of the total surface
area of the support led to strong deactivation of the catalyst.
This result indicated that the fraction of the support involved
in the reaction is small, probably located next to the metal
and corresponding, at most, to a ring of four times the ra-
dius of the metallic particles. It is proposed that, during the
reaction, the portion of the support close to the metal parti-
cles could be in a highly reduced state (a gradient is created
between the vicinity of the metal particles and outer edge of
this support ring) and would correspond to the active part of
the support. This portion of active support is thought to be
involved in the reduction of the CO2 to CO.
Fig. 10. Schematic representation of the carbon ring surrounding the plat-
inum particles.
of carbon deposited) could find an explanation. This slope
change would correspond to the point at which the whole
active part of the support is covered by carbon deposits. The
remaining activity, beyond the point of slope change, would
be the activity of the platinum only. Note that this interpre-
tation was reinforced by the results obtained with the 2%
Pt/SiO2. For this catalyst the support is obviously inactive
and the conversion observed (∼2%) was relatively close to
the 4% conversion observed after 80% loss of activity of the
Moreover, these results are consistent with results ob-
tained with the RWGS reaction mixture. TPO experiments
following the exposure of the 2% Pt/CeO2 catalyst to the
RWGS reaction mixture led to the occurrence of a peak
2
% Pt/CeO2.
Moreover, on the basis of this hypothesis, it is likely that
the carbon was deposited directly on the support and not on
the metal first. That would mean that the progressive cover-
age by the carbon would start from the support followed by
the metal when the active part of the support was covered
and not the reverse.
Note that this hypothesis is also compatible with the
mechanism proposed by Fu and co-workers [23] more re-
cently. These authors proposed that the metal phase is not
directly involved in the mechanism of the reaction in the case
of the ceria-supported metal catalysts. In this approach, the
real role of the metal would be to increase the reducibility of
the support rendering the support as the active phase. If such
a mechanism is indeed taking place, only a direct coverage
of the support by the carbon deposits could be responsible
for the deactivation of the catalyst.
Our study clearly demonstrates that carbon deposition
is at least one of the main components of the deactivation
of Pt/ceria catalysts for the RWGS reaction. It also clearly
shows the crucial role of the support in being at least part of
the active phase of the reaction. The exact role of the metal-
lic phase could not be totally clarified and will be addressed
in more detail in a future publication.
◦
at 260–290 C, characteristic of carbon deposits. This peak
grew in intensity with the time of exposure of the catalyst
to the RWGS reactant mixture. These results confirmed that,
during the course of the RWGS reaction on this catalyst, car-
bon deposit took place and was thus one of the major reasons
for the slow deactivation that is observed for 2% Pt/CeO2.
References
[1] Y. Jamal, M.L. Wyszinski, Int. J. Hydrogen Energy 19 (1994) 557.
[2] M.A. Pena, J.P. Gomez, J.L.G. Fierro, Appl. Catal. A 144 (1996) 7.
[3] G. Maggio, S. Freni, S. Cavallaro, J. Power Sources 74 (1998) 17.
[4] A.N. Fatsikostas, D.I. Kondarides, X.E. Verykios, Chem. Commun.
(2001) 851.
[
[
5] J.M. Zalc, V. Sokolvskii, D.G. Löffler, J. Catal. 206 (2002) 169.
6] B.I. Whittington, C.J. Jiang, D.L. Trimm, Catal. Today 26 (1995) 41.
[7] T. Shido, Y. Iwasawa, J. Catal. 141 (1993) 71.
[8] J.T. Kummer, J. Phys. Chem. 90 (1986) 4747.
[9] E.C. Su, W.G. Rothschild, J. Catal. 99 (1986) 506.
[
[
10] R.K. Hertz, J.A. Sell, J. Catal. 94 (1985) 166.
11] K.C. Taylor, Automobile Catalytic Converters, in: J.R. Anderson, M.
Boudard (Eds.), Catalysis: Science and Technology, vol. 5, Springer,
Berlin, 1984.

392
A. Goguet et al. / Journal of Catalysis 226 (2004) 382–392
[
12] T. Bunluesin, R.J. Gorte, G.W. Graham, Appl. Catal. B 15 (1998) 107.
13] S. Hilaire, X. Wang, T. Luo, R.J. Gorte, J. Wagner, Appl. Catal. A 215
[24] S. Golunski, R. Rajaram, N. Hodge, G.J. Hutchings, C.J. Kiely, Catal.
Today 72 (2002) 107.
[
(2001) 271.
[25] J.C. Frost, Nature 334 (1988) 577.
[
[
[
26] S. Bernal, F.J. Botana, J.J. Calvino, M.A. Cauqui, G.A. Jobacho, J.M.
Pintado, J.M. Rodriguez-Izquierdo, J. Phys. Chem. 97 (1993) 4118.
27] C. Li, Y. Sakata, T. Arai, K. Domen, K. Maruya, T. Onishi, J. Chem.
Soc., Faraday Trans. I 85 (6) (1989) 1451.
28] C. Li, Y. Sakata, T. Arai, K. Domen, K. Maruya, T. Onishi, J. Chem.
Soc., Faraday Trans. I 85 (1989) 929.
[
14] X. Wang, R.J. Gorte, Catal. Lett. 73 (2001) 15.
[15] Q. Fu, A. Weber, M. Flytzani-Stephanopoulos, Catal. Lett. 77 (2001)
87.
[
[
[
[
16] T. Luo, R.J. Gorte, Catal. Lett. 85 (2003) 139.
17] X. Wang, R.J. Gorte, J.P. Wagner, J. Catal. 212 (2002) 225.
18] T. Bunluesin, R.J. Gorte, G.W. Graham, Appl. Catal. B 14 (1997) 105.
19] H. Cordatos, T. Bunluesin, J.S. Stubenrauch, J.M. Vohs, R.J. Gorte,
J. Phys. Chem. 100 (1996) 785.
[29] C. Binet, M. Daturi, J.C. Lavalley, Catal. Today 50 (1999) 207.
[30] A. Bourane, O. Dulaurent, D. Bianchi, J. Catal. 196 (2000) 115.
[31] V. Pitchon, J.F. Zins, L. Hilaire, G. Maire, React. Kinet. Catal. Lett. 59
[
[
[
[
20] Y. Li, Q. Fu, M. Flytzani-Stephanopoulos, Appl. Catal. B 27 (2000)
(1996) 203.
179.
[32] E. Rogemond, R. Frety, V. Perrichon, M. Primet, S. Salasc, M. Chev-
21] G. Jacobs, L. Williams, U. Graham, D. Sparks, B.H. Davis, J. Phys.
Chem. B 107 (2003) 10398.
rier, C. Gauthier, F. Mathis, J. Catal. 169 (1997) 120.
[33] C. Li, Y. Sakata, T. Arai, K. Domen, K. Maruya, T. Onishi, J. Chem.
Soc., Chem. Commun. (1991) 410.
[34] A. Holmgren, B. Andersson, J. Catal. 178 (1998) 14.
[35] A. Goguet, D. Schweich, J.P. Candy, J. Catal. 220 (2003) 280.
[36] H.H. Lee, E. Ruckenstein, Catal. Rev.-Sci. Eng. 25 (1983) 475.
22] G. Jacobs, L. Williams, U. Graham, G.A. Thomas, D. Sparks, B.H.
Davis, Appl. Catal. A 252 (2003) 107.
23] Q. Fu, H. Saltsburg, M. Flytzani-Stephanopoulos, Science 301 (2003)
935.