An investigation of possible mechanisms for the water-gas shift reaction over a ZrO2-supported Pt catalyst

Article abstract of DOI:10.1016/j.jcat.2006.09.004

The present work investigates the reactivity of the surface species observable by in situ DRIFTS formed over a Pt/ZrO₂ during the water-gas shift (WGS) reaction. A DRIFTS cell/mass spectrometer system was operated at the chemical steady state during isotopic transients to yield information about the true nature (i.e., main reaction intermediate or spectators) of adsorbates. Only carbonyl and formate species were observed by DRIFTS under reaction conditions; the surface coverage of carbonate species was negligible. Isotopic transient kinetic analyses revealed that formates exchanged uniformly according to a first-order law, suggesting that most formates observed by DRIFTS were of the same reactivity. In addition, the time scale of the exchange of the reaction product CO₂ was significantly shorter than that of the surface formates. Therefore, a formate route based on the formates as detected by DRIFTS can be ruled out as the main reaction pathway in the present case. The number of precursors of the reaction product CO₂ was smaller than the number of surface Pt atoms, suggesting that carbonyl species or some infrared-invisible complex (not excluding formates) at the Pt-zirconia interface was the main reaction intermediate. A simple redox mechanism could also explain the present results.

Full text of DOI:10.1016/j.jcat.2006.09.004

Journal of Catalysis 244 (2006) 183–191
www.elsevier.com/locate/jcat
An investigation of possible mechanisms for the water–gas shift reaction
over a ZrO -supported Pt catalyst
2
a
a,∗
a
a
a
b
b
D. Tibiletti , F.C. Meunier , A. Goguet , D. Reid , R. Burch , M. Boaro , M. Vicario ,
b
A. Trovarelli
a
CenTACat, School of Chemistry and Chemical Engineering, Queen’s University Belfast, Belfast, BT9 5AG, Northern Ireland
b
Dipartimento di Scienze e Tecnologie Chimiche, Università di Udine, 33100 Udine, Italy
Received 18 July 2006; revised 31 August 2006; accepted 5 September 2006
Available online 10 October 2006
Abstract
The present work investigates the reactivity of the surface species observable by in situ DRIFTS formed over a Pt/ZrO during the water–gas
2
shift (WGS) reaction. A DRIFTS cell/mass spectrometer system was operated at the chemical steady state during isotopic transients to yield
information about the true nature (i.e., main reaction intermediate or spectators) of adsorbates. Only carbonyl and formate species were observed
by DRIFTS under reaction conditions; the surface coverage of carbonate species was negligible. Isotopic transient kinetic analyses revealed that
formates exchanged uniformly according to a first-order law, suggesting that most formates observed by DRIFTS were of the same reactivity.
In addition, the time scale of the exchange of the reaction product CO was significantly shorter than that of the surface formates. Therefore,
2
a formate route based on the formates as detected by DRIFTS can be ruled out as the main reaction pathway in the present case. The number
of precursors of the reaction product CO was smaller than the number of surface Pt atoms, suggesting that carbonyl species or some “infrared-
2
invisible” complex (not excluding formates) at the Pt–zirconia interface was the main reaction intermediate. A simple redox mechanism could
also explain the present results.
©
2006 Elsevier Inc. All rights reserved.
Keywords: Formate; DRIFT; FTIR; Operando; In situ; Spectroscopy; Water–gas shift; Reaction intermediate; Spectator; SSITKA; Platinum; Zirconia
1
. Introduction
former atmosphere, and (iv) no degradation on exposure to
air or temperature cycles. Formulations based on noble met-
als (NM), such as Pt and Au on reducible supports, have been
recently proposed [3,6–8]. Pt-supported on zirconia [9,10] or
mixed-oxide ceria–zirconia [11,12] have been of particular in-
terest.
Although numerous mechanistic and kinetics studies have
been carried out in recent years, disagreement remains about
the nature of the reaction intermediates and the active sites
of NM supported on redox oxides. Two reaction mechanisms
are mainly proposed in the literature: the “adsorptive mecha-
nism” (involving in particular formate surface species) and the
The water–gas shift (WGS) reaction and the correspond-
ing low- and high-temperature catalysts (i.e., Cu and hematite-
based formulations, respectively) are well-established features
of large-scale industrial production of hydrogen [1,2]. Yet,
the interest in novel low-temperature WGS catalysts has been
growing steadily in relation to fuel cell technology [3–5]. The
need to develop fuel cell processors capable of more efficiently
converting fuels into high-purity hydrogen has led to intensive
research into developing new catalytic formulations able to sat-
isfy the necessary criteria for this application, that is, (i) high
activity in the temperature range 473–553 K, (ii) no need for
activation before use, (iii) resistance to poisoning under a re-
“
regenerative mechanism” (redox). In the regenerative mecha-
nism [13–20], water adsorbs and dissociates on a partly reduced
support, releasing H2 and reoxidising ceria. In parallel, CO(g)
adsorbs on metallic sites to form a NM-bound carbonyl species,
which then reduces the support and releases CO2. This model is
consistent with that proposed by Grenoble and Estadt [21], who
*
Corresponding author. Fax: +44 28 90 382117.
E-mail addresses: f.meunier@qub.ac.uk (F.C. Meunier),
marta.boaro@uniud.it (M. Boaro).
0021-9517/$ – see front matter © 2006 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2006.09.004

184
D. Tibiletti et al. / Journal of Catalysis 244 (2006) 183–191
studied the effect of various supports on the reaction rate and
concluded that the WGS reaction occurred through a bifunc-
tional mechanism; the metal activated the carbon monoxide,
and support sites were the principal sites for water activation.
Various authors have also reported a relationship between cata-
lyst activity and the redox properties of the support [22–24].
In the adsorptive (or associative) mechanism [25–30], CO
and H2O are proposed to adsorb on the catalyst and form a sur-
face intermediate, which subsequently decomposes as H2 and
CO2. Many research groups have tried to identify the nature of
the main intermediate species mainly by isotopic labelling ex-
periments and FT-IR spectroscopy. Formate species have been
proposed for the forward-WGS reaction [25,31–33] and car-
bonates for both the forward [34] and reverse-WGS reaction
tivity of ZrO2- and CeO2-based materials strongly depends on
their method of preparation. As an example, some of us have
recently shown that the activity in CO oxidation on similarly
prepared ceria powders is strongly dependent on type of sur-
face planes exposed [41]. These in turn are modified by ther-
mal treatments. Moreover, it must be stressed that ZrO2 alone
is present in different polymorphs (tetragonal and monoclinic)
that exhibit different structural and acid–base surface proper-
ties; therefore, the similarities between ZrO2 and CeO2 are not
always obvious [42].
In contrast to some other catalysts, for which complications
due to the influence of an easily reducible support (e.g., CeO2
or Fe2O3) can make it difficult to define a unique mechanism
for the WGS reaction, plain ZrO2 is less readily reduced (see
the TPR data in Ref. [9]); thus, it may be possible to better
differentiate between processes on the metal or at the metal–
support interface and those occurring across the support. In the
present work, the nature and reactivity of the surface species
formed under operando conditions during the WGS reaction
over Pt supported on ZrO2 is presented and the role of formate
species is discussed.
[
35]. In work published by Jacobs et al. [12], it has been pro-
posed that one of the metal’s main roles would be to favour the
formation of reactive surface hydroxyl groups (type-II) on the
oxide support, which would lead to increased formation of for-
mate species.
Our laboratory has recently shown that the reactivity of
formates and other surface species formed over Pt/CeO2 cat-
alyst under reverse was–gas shift (RWGS) feed conditions
was strongly dependent on the experimental procedure used,
stressing the need to use steady-state and operando methods
Based on a DRIFT + MS/SSITKA technique [39,43,44], the
data reported here unambiguously show that the route based
on the formate observed by DRIFTS is only a “minor” reaction
pathway under the present experimental conditions used for our
Pt/ZrO2. This stresses that another route, the details of which
are yet unclear but could possibly be related to a redox-type
mechanism, must be present and accounts for most of the sam-
ple activity under the present conditions. Although plain ZrO2
can be difficult to reduce even at high temperatures (e.g., not
[
36]. The formate species, monitored by operando DRIFTS
combined with the utilisation of isotopic tracer methods (i.e.,
SSITKA), appeared to undergo isotopic exchange slower than
the formation of the reaction product (i.e., 13CO) under steady-
state conditions, indicating that these species were not “main”
reaction intermediates. It was proposed that formates were
therefore “minor” reaction intermediates, meaning that these
species are associated with a reaction pathway that may lead
to the reaction product. But that would account for only a mi-
nor fraction of the actual product yield. Importantly, our work
indicated that two or more reaction pathways may occur simul-
taneously over Pt/CeO2.
◦
below 800 C [9]), various groups have reported temperature-
programmed reaction studies clearly showing that Pt-promoted
◦
zirconia was starting to reduce below 200 C [9,12], and thus a
redox mechanism is possible, at least at the metal–support in-
terface, under the conditions reported here.
Gong et al. [37] studied the WGS reaction using density
functional theory (DFT). In the redox mechanism, the highest
energetic barrier is ca. 0.9 eV, corresponding to the hydroxyl de-
composition OH → O + H. The highest energetic barrier in the
adsorptive mechanism is the formate decomposition, HCOO
2. Experimental
2.1. Sample preparation
The support used was a commercial ZrO2 supplied by Mag-
nesium Elektron Limited, calcined at 923 K. The platinum was
deposited by incipient wetness impregnation of H2PtCl6·6H2O
(Aldrich). The support was first dried and then impregnated
with an appropriate volume of the aqueous solution of Pt pre-
cursor, corresponding to the support pore volume. The solu-
tion was added dropwise, stirring the catalyst precursor until
pore saturation. To achieve the desired loading, the impregna-
tion was repeated three times. The sample was dried at 383 K
overnight, crushed, and calcined at 823 K in a tubular reactor
→
CO2 + H, at 1.02 eV. If there were an OH group on the
surface near the surface HCOO, then the hydrogen on the sur-
face formate could pass to the hydroxyl, leading to CO2 and H2
with an even lower barrier. However, the authors concluded that
both mechanisms were possible and that surface coverage could
play an important role in determining the reaction pathway. In
this regard, Farrauto et al. [38] and Meunier et al. [39] recently
proposed that surface carbonates are a potential main reaction
intermediate that normally readily decomposes to CO2(g), but
that overreduction of the CeO2 support could lead to a strength-
ening of the carbonate bonds, resulting in self-poisoning of the
catalyst by one of the reaction products [40].
3
−1
for 1 h in air (200 cm min ), using a temperature ramp of
−
1
10 K min
.
Noble metals solely supported on either ZrO2 or CeO2 are
typically less active than the corresponding mixed oxide-based
materials, but it is important and enlightening to investigate the
individual components. We also must point out that the reac-
2.2. Sample characterization
The specific surface area was measured by the BET method
on a Tristar 3000 (Micrometrics) gas adsorption analyser. X-ray

D. Tibiletti et al. / Journal of Catalysis 244 (2006) 183–191
185
powder diffraction patterns were collected using a Philips
PW3040/60 X’pert PRO instrument (equipped with an X’celer-
ator detector) operated at 40 kV and 40 mA, with an Ni-filtered
CuKα radiation source. The diffractograms were collected with
achieve ca. 12% conversion. The reaction mixture (i.e., 2% CO,
7% H O, 12% H in Ar) was introduced at a total flow rate of
2
2
3
−1
100 cm min . The steady-state conditions, as far as the na-
ture and concentration of surface species measured by DRIFTS
are concerned, was reached in less than 30 min. A series of
back and forth isotopic switches were carried out to assess the
reproducibility and precision of the experiment. No kinetic dif-
ferences could be observed between the forward and backward
switches, and symmetric exchange curves were obtained. It was
noticed that any set of 12C or C species could be used to dis-
cuss the results.
◦
a step size of 0.02 and a counting time of 40 s per angular ab-
◦
◦
scissa in the range of 20 –145 .
The Pt dispersion was estimated using H2 chemisorption
analysis, carried out on a Micromeritics ASAP 2000 analyser.
The H2 chemisorption experiments were conducted at 183 K on
13
0
.2 g of catalyst. The calcined sample was subjected to a stan-
dard cleaning procedure consisting of heating the sample under
O2 (5%)/Ar at 773 K for 1 h. The samples were then prere-
The IR data were reported as log 1/R, with R = I/I , where
0
−
1
duced in a flow of H2 (35 ml min ) at 473 K. After 2 h at this
temperature, the samples were evacuated at 673 K for 4 h and
cooled under vacuum to the adsorption temperature (183 K).
Typically, an equilibration time of 10 min was used. Adsorbed
volumes were determined by extrapolation to zero pressure of
the linear part of the adsorption isotherm (100–400 Torr) after
elimination of the so-called “reversible hydrogen adsorption.”
A stoichiometry of 1:1 between the surface Pt site and hydro-
gen atom was assumed.
R is the sample reflectance, I is the intensity measured on the
0
sample after reduction in hydrogen, and I is the intensity mea-
sured under reaction condition. The function log 1/R gives a
better linear representation of the band intensity against sample
surface coverage than that given by the Kubelka–Munk func-
tion for strongly absorbing media such as our catalyst [45,46].
2
.4. Analysis of the isotopic exchange rate of the surface
species by IR and rate of formation of the reaction products
by MS
2
.3. In situ DRIFTS studies and SSITKA–DRIFTS/MS
A least squares linear regression method was used to quan-
tify the extent of the isotopic exchange of the formate species,
using the initial (i.e., all 12C) and final (i.e., all C) DRIFTS
spectra as references. Two different regions of the spectrum
were used to integrate the signal of formate species (vide in-
fra). As far as the mass spectrometric analysis was concerned,
the masses 28, 44, 29, 45, and 4 (amu), were followed and the
data collection was started at least 10 min before the isotopic
switches to observe any possible MS baseline drift. The con-
centrations of the reactants and products were also determined
using an on-line gas chromatograph (Perkin–Elmer Clarus),
equipped with a HayeSep Q column and a flame ionization de-
tector fitted with a methanator to allow the accurate and precise
quantification of CO and CO2.
experiments
13
The in situ DRIFTS experiments were carried out as de-
scribed previously [40]. In brief, it consisted of an in situ
high-temperature diffuse reflectance IR cell (Spectra-Tech) fit-
ted with ZnSe windows. The crucible was modified to ensure
essentially plug-flow conditions through the catalyst bed. The
interface between the ceramic reactor and the metallic base
plate was sealed with PTFE tape to prevent any sample bypass
[
39]. The cell was connected to the feed gas cylinders through
low-volume stainless steel lines. The purity of the gases used
i.e., H2, CO and Ar, supplied by BOC) was >99.95%. The
(
1
3
CO was 99% pure (supplied by Cambridge Isotope Labo-
ratories). The gas flows were controlled by Aera mass flow
controllers, which were calibrated regularly. The water was sup-
plied by bubbling Ar through a metallic water saturator, the
temperature of which was controlled by an Eurotherm 2416
controller. A four-way valve was used to allow rapid switch-
ing between two reaction feeds, when appropriate. The DRIFTS
cell was located in a Bruker Equinox 55 spectrometer, oper-
ating at a resolution of 4 cm . The outlet of the DRIFTS
cell was connected to a quadrupolar mass spectrometer (Hiden
HPR20). The mass spectrometer was equipped with a capillary
inlet system with bypass allowing fast response (i.e., 100 ms)
for the sampling. Both feed lines were connected to an ultra-
low differential pressure transducer (Honeywell 395–257) and
a high-precision metering valve (Nupro) to equalise the corre-
sponding pressure drops. No variation in the concentration of
water was observed on switching between feeds.
3
. Results and discussion
3
.1. Catalyst structure
The sample characteristics are gathered in Table 1. Fig. 1
shows the XRD patterns of the support and the platinum-loaded
catalyst. In both cases, the ZrO2 was mainly monoclinic, al-
though a tetragonal phase could also be observed. The diffrac-
−
1
Table 1
Structural properties of the samples
Sample Pt
name loading
wt%)
Pt
Average
Specific
XRD
structure
Average
pore
size
c
disper- Pt particle surface
sion
a
b
(
size
area
2
−1
)
(nm)
(m g
(nm)
The catalyst loaded in the reactor had a particle diameter
150 µm. Before any measurement, the sample was reduced
ZrO
Pt/ZrO2
–
4
–
75%
–
1.5
38 (923 K) m and t
35 (523 K) m and t
11.8
10.2
2
<
in situ for 30 min at 453 K in a 5% H2/Ar mixture at a total
flow of 100 cm min . After the reduction step, the cell was
purged with Ar, and the temperature of the reactor was set to
a
b
c
Determined from the dispersion.
In parentheses the temperature of calcination.
m = monoclinic; t = tetragonal.
3
−1

186
D. Tibiletti et al. / Journal of Catalysis 244 (2006) 183–191
Fig. 1. XRD patterns of (a) ZrO , (b) Pt/ZrO . In both cases, the patterns corresponded to 90% of the monoclinic polymorph (JCPDS 861450) and 10% of the
2
2
tetragonal polymorph (JCPDS 79-1770).
1
2
Fig. 2. In situ DRIFT spectra of the Pt/ZrO at 473 K in (a) Ar after prereduction and (b) exposed to the WGS mixture (i.e., 2% CO, 7% H O, 12% H in Ar) and
2
2
2
13
(
c) exposed to the corresponding labelled feed (i.e., 2% CO, 7% H O, 12% H in Ar). The absorbance with respect to a mirror signal is reported.
2 2
tion patterns of the Pt-containing catalysts did not show any
peak related to the metal or metal oxide phases, indicating that
high dispersion was obtained and the Pt crystallite size was on
the order of nanometers. Furthermore, there was no evidence
of solid solution formation between Pt and the support; no shift
of the support peak was observed. After calcination at 923 K
feed) are shown in Fig. 2. The absorbance with respect to the
signal obtained using a mirror in place of the sample is reported.
−1
In all cases, the large band observed at around 825 cm was
associated with zirconia phonon modes, and the broad and ill-
−1
defined band observed between 3700 and 3000 cm was due
to H-bonded hydroxyl groups [47]. Note that the negative band
2
−1
−
1
for 2 h, the ZrO specific surface area was equal to 38 m g
,
2
at 2360 cm was due mostly to ambient CO signal variations
2
2
−1
whereas that of the Pt sample was 35 m g . The samples ex-
hibited type IV adsorption isotherms (data not shown), with an
average pore size of 10 nm.
and is of no significance here.
A series of bands around 2055 cm was observed in both
the sample under Ar (Fig. 2a) and that under the C-containing
−1
12
WGS feed (Fig. 2b). These bands are unambiguously assigned
to Pt-bound carbonyl species [48]. The fact that such species
were present over the sample not yet exposed to the WGS feed
can be simply reconciled by the fact that some C-containing
species, such as carbonates, were reduced during the pretreat-
ment yielding CO, which then adsorbed onto the Pt. In a similar
3
.2. In situ DRIFTS data
The DRIFTS spectra of the Pt catalyst at 473 K after prere-
duction under Ar and under the WGS mixtures (i.e., 2% CO,
7
13
% H O, 12% H in Ar, and the corresponding C-labelled
2 2

D. Tibiletti et al. / Journal of Catalysis 244 (2006) 183–191
187
1
2
Fig. 3. In situ DRIFT spectrum of the surface species observed over a 4% Pt/ZrO at 473 K at steady state under (a) 2% CO, 7% H O, 12% H in Ar, and then
2
2
2
13
under 2% CO, 7% H O, 12% H in Ar for (b) 0.75, (c) 2 and (d) 30 min. The signal of the reduced catalyst at the same temperature under Ar was used as
2
2
background.
way, ill-defined bands at around 2900 and 1500 cm 1 can be
observed over the prereduced sample and were mostly formate-
type species formed by CO adsorption over the support [49–
−
any band other than formates in the 1700–1200 cm region, in
which OCO stretching modes typically appear.
−1
The details of the spectra obtained under the 13C-containing
feed (Fig. 3b) were very similar to those observed under C,
except for the fact that most bands exhibited a red shift. Note
that the band at 797 cm (Fig. 3a) was associated with the
OCO bending mode of the C-formate, which was not ex-
pected to shift when exchanging C with C, as was con-
firmed here (Fig. 3b). Because no carbonate species were
present, either the CH or the OCO stretching region could be
used to quantify the relative exchange of the formate species
during the SSITKA experiments. These analyses were car-
12
5
4].
The corresponding absorbance spectra using the signal ob-
tained over the sample under Ar as reference background are
given in Fig. 3. Fig. 3a clearly shows that the surface con-
centration of C-containing formates was markedly increased
under the 12CO-containing WGS feed. Most of the clearly
observable bands (i.e., 797, 1360, 1382, 1575, 2737, 2870,
931, and 2966 cm ) were all associated with surface formate
species [47]. The band structure between 3000 and 2700 cm
is quite complex (Fig. 3a, left), and various assignments have
already been discussed by Busca and co-workers [47]. The
931 cm band is actually somewhat reminiscent of a band
associated with the C-containing species (Fig. 3b), but we
can confidently exclude any contamination due to the C-
containing feed and clearly confirm the earlier assignment by
Busca et al. of this particular band to a ( C-containing) for-
mate. Some minor modifications of the carbonyl region (i.e.,
100–1700 cm ) peak were observed under WGS feed and
those could possibly be related to modification of the shape of
Pt particles or competitive adsorption phenomena over the Pt
surface.
−
1
12
12
12
13
−1
2
−
1
−
1
ried out by using the 3100–2600 cm region and the 1667–
1411 cm region, based on a least squares linear regression
−
1
−1
12
13
2
method using the initial C and final C spectra as refer-
ences. It is interesting to note that the use of either region led
to essentially the same transient response (Fig. 4a). This obser-
vation emphasised the accuracy and reliability of this analytical
method. In addition, the semilogarithmic plot (Fig. 4b) of the
same data indicated that the exchange curve of the formates
was a single exponential curve, suggesting that the reactivity of
all of the formate species as observed by IR was uniform and
followed a first-order law.
13
13
12
−1
2
It is noteworthy the absence of any detectable bands in the
3.3. MS experiments
−
1
spectral region at around 860 cm , which would be typically
associated with the carbonate out-of-plane bending [36]. This
stresses that under reaction conditions, the amount of adsorbed
carbonate was negligible, as already indicated by the lack of
The isotopic exchange of the reactant 13CO and that of
the reaction product 13CO were recorded by MS experiments
2
(Fig. 5), simultaneous to the DRIFTS data presented in the pre-

188
D. Tibiletti et al. / Journal of Catalysis 244 (2006) 183–191
Table 2
Quantitative data derived from the MS-SSITKA curves
N
, precursors
N
, precursors
2(g)
NPt, surface
N , surface
CO
CO
O
2
of CO
of CO
Pt atoms
oxide ions
(g)
18
−1
18 −1
g
cat
18
−1
18 −1
10
g
10
10
g
10
g
cat
cat
cat
52 ± 5
17 ± 2
93 ± 7
520 ± 10
plot of the tracer and that of the 13CO is related (via CO con-
centration and flow rate) to NCO, that is, the surface precursors
of CO(g) [e.g., carbonyl or formate that desorbs as CO(g)] [44,
5
5–57]. In a similar manner, the area between the tracer curve
13
and that of the CO leads to the number of surface precur-
2
sors of CO2 [e.g., CO2(ads), carbonates, if any, or formates].
These values of precursor amounts, along with the number of
surface Pt and O sites, are given in Table 2. The latter set of val-
ues was calculated from the Pt dispersion as determined by H2
chemisorption (assuming H:Pt = 1:1), the surface area of the
catalyst, and the average surface density of oxide ions over the
(a)
−
2
zirconia monoclinic surface (ca. 15 O nm ).
These data highlight two important facts. First, the number
of precursors of the reaction product CO2 (i.e., NCO = 17 ×
2
1
8
−1
1
0
g
) was significantly lower than the number of surface
). The number of Pt atoms
located at the perimeter of the metal particles (assumed to be
cat
18
−1
Pt atoms (i.e., NPt = 93 × 10
g
cat
18
−1
hemispheres of diameter 1.5 nm) was 33 × 10
g
. Therefore
cat
it is possible that the catalytically active sites were related only
to Pt atoms, for example, all Pt sites or even only those located
at the metal–oxide interface.
(b)
12
Fig. 4. (a) Comparison of the variation of the DRIFT signal of C-formate
species during the isotopic exchange evaluated using (◦) the CH stretching re-
Second, the number of adsorbed species redesorbing as
−
1
gion around 2860 cm and (×) the OCO asymmetric stretch region around
18 −1
) was smaller than the num-
cat
CO(g) (i.e., NCO = 52 × 10
g
−
1
1550 cm
(see text for integration details). (b) Same as (a), plotted in a
13
ber of Pt sites minus the number of precursors of CO2 (i.e.,
semi-logarithmic mode. Feed: 2% CO, 7% H O, 12% H in Ar, following
2
2
18
−1
12
Pt
N − N
= 75 × 10
g
). This makes it possible that all
steady state under 2% CO, 7% H O, 12% H in Ar. T = 473 K.
CO2
cat
2
2
or the main part of the CO(g) precursors originated simply from
Pt-bound carbonyls. Some of the CO(g) could have also arisen
from the decomposition of the formates located on the ZrO2.
However, this latter route is likely to be minor because the time
scales of the CO exchange (ca. 0.2 min) and the formate ex-
change (5 min) were significantly different.
3
.4. Comparison of the DRIFTS and MS data during the
SSITKA experiments
The comparison of the DRIFTS 13C-formate and MS 13CO2
data during the isotopic exchange (Fig. 6) reveals in an unam-
biguous manner that these two species were exchanged on a
significantly different time scale and thus were unlikely to be
related to one another. The formates half-exchanged in about
1
3
13
2 min, whereas the CO half-exchanged in about 0.1 min. It
Fig. 5. MS traces of Kr, CO and CO for the isotopic switch experiment
2
2
13
over the Pt/ZrO catalyst. Feed: 2% CO, 7% H O, 12% H in Ar, following
steady state under 2% CO, 7% H O, 12% H + 1% Kr in Ar. T = 473 K.
can be seen that 90% of the CO produced was already ex-
2
2
2
2
12
12
2
2
changed after 0.3 min, whereas only 10% of the C-formates
had been replaced by the 13C counterpart at this point. This ob-
servation clearly indicates that the formates seen by DRIFTS
are not responsible for the formation of the reaction product
CO2. Because the exchange of formates appeared to be uniform
(Fig. 4b), it is unlikely that any significant fraction of the for-
mates observed by DRIFTS reacted faster than the rest of these
vious section. The relative intensity of the MS signal of Kr, used
as a tracer, reached steady state in less than 0.1 min (Fig. 5).
This value gives an indication of the time resolution of our
setup. The exchange of both the reactant and product were es-
sentially complete in less than 0.5 min. The area between the

D. Tibiletti et al. / Journal of Catalysis 244 (2006) 183–191
189
far away from the interface would not significantly contribute
to the turnover number.
In addition, the SSITKA data have shown that the number
of CO2 precursors was merely similar to the number of Pt–Zr
interface sites, and probably much lower than the number of
formates that could bind to oxide sites. (The latter number is
not known, but the number of surface oxide ions is 30-times
larger than the number of CO2 precursors; see Table 2.)
One referee suggested that by switching off the water feed
(
and replacing it by the same flowrate of inert gas), a large
increase of the formate concentration would be observed, sup-
porting the presence of highly reactive formate as main reaction
intermediates, which would not be observable by IR at steady-
state conditions. Although the observation described by the ref-
eree may be true, we do not believe that it would constitute
unambiguous evidence for the involvement of fast formates, but
we cannot reject this argument either. The increased formate
concentration could be due simply to modification of the sur-
face structure and/or oxidation state of the catalyst, due to the
removal of an oxidiser (i.e., water) while a reducing agent (i.e.,
CO) was kept in the feed. TPR [9,12] and EPR studies [62,63]
support the formation of some reduced centres (i.e., Zr ) al-
ready at the temperature relevant to the present work, these
reduced centres having been proposed as the active sites for CO
hydrogenation via formate species [64–66]. Bell et al. have also
proposed that formation of the anion vacancies may actually fa-
cilitate the reaction of CO with neighbouring surface hydroxyl
group and lead to surface formate species at temperatures sim-
ilar to that used here [67,68]. Caution must be observed when
using non-steady-state conditions to study the nature and reac-
tivity of surface species over this type of catalyst, even if ZrO2
is typically less easily reduced than CeO2, for which the effect
of feed composition on the surface reactivity has been clearly
demonstrated [40].
1
3
Fig. 6. Normalised MS trace of CO and intensity of the IR band of the
2
13
C-containing formate following the isotopic switch experiments over the
Pt/ZrO . Feed: 2% CO, 7% H O, 12% H in Ar, following steady state under
13
2
2
2
12
2%
CO, 7% H O, 12% H in Ar. T = 473 K.
2
2
3+
12
Fig. 7. Variation of the DRIFT signal of C-formate species during the isotopic
exchange reported as a function of the square root of time. Feed: 2% CO, 7%
H O, 12% H in Ar, following steady state under 2% CO, 7% H O, 12% H
13
12
2
2
2
2
Although the present work could not highlight the nature of
the main reaction intermediates (e.g., carbonyl or formates lo-
cated on the Pt or the Pt–ZrO2 interface), it can be concluded
that the “formates observed by IR” located on the ZrO2 sur-
face were not the main pool of surface compounds leading to
CO2, although those may lead to CO2 at a much slower rate.
These IR-visible formates were minor reaction intermediates
and not truly spectator species, because those were eventually
exchanged. However, under normal time scales they could not
make any significant contribution to the production of CO2. It
would be useful to quantify the exact proportion of CO2 formed
by the IR-observable formates; however, this would require de-
termining the absolute formate surface concentration and the
selectivity to which those decompose to either CO or CO2,
which are not straightforward measurements.
in Ar. T = 473 K.
species observed by DRIFTS to account for the formation of
CO2.
The slower exchange of formates compared with that of CO2
could have been related to the fact that formates reacted to yield
CO2 only at the interface with Pt, as proposed by Davis et al.
[
58], and thus slow surface diffusion could have limited their
rate of exchange. Diffusion on a surface is a two-dimensional
phenomenon, and, as in mono-dimensional diffusion, the up-
take or exchange of species following a traditional Fickian dif-
fusion model is expected to yield a linear variation of the con-
centration with the square root of time, up to ca. 50% exchange
[
59–61]. The plot relating to the formate isotope exchange as a
function of the square root of time was not linear (Fig. 7), and
thus the surface diffusion of formates to the metal sites was not
the limiting step in the process of formate exchange observed
here. The data reported in Fig. 4b. actually indicated that the
exchange followed a single exponential law, related to a first-
order process. In any case, and with involvement of formate
diffusion, the misfit between the formate DRIFTS exchange
As far as the main reaction pathway is concerned, the only
IR-observable species that seemed to have a kinetic response
comparable to the rate of formation of carbon dioxide is the car-
bonyl species. It is however impossible to accurately quantify
the carbonyl exchange; see the discussion section on carbonyls
in [46]. However, the number of CO precursors is lower than
2
the number of surface platinum atoms, assuming that one CO
plot and that of the CO (Fig. 6) still indicates that the formates
adsorbed per surface Pt atom. This leads us to suggest that on
2

190
D. Tibiletti et al. / Journal of Catalysis 244 (2006) 183–191
a support such as zirconia, which is yet difficult to reduce, the
WGS mechanism could involve a simple redox mechanism as
described by Flytzani–Stephanopoulos et al. and Gorte et al.
References
[1] C. Rhodes, G.J. Hutchings, A.M. Ward, Catal. Today 23 (1995) 43.
[2] M. Laniecki, M. Malecka-Grycz, F. Domka, Appl. Catal. A Gen. 196
[
13–19], possibly at the metal–support interface. This could in-
(
2000) 293.
3] A.N. Fatsikostas, D.I. Kondarides, X.E. Verykios, Catal. Today 75 (2002)
45.
volve activation of water at an oxygen vacancy on the surface
of the zirconia adjacent to a Pt particle, followed by transfer of
O(ads) to the Pt, where it could react with CO to release CO2.
We emphasise that these considerations of possible reaction
intermediates strictly apply only to this catalyst under these re-
action conditions. Elsewhere [36,39,69], we have frequently
emphasised the point that the dominant reaction mechanism
for the WGS reaction is dependent on experimental parame-
ters, such as the reaction temperature or the gas composition.
To this we can now add the choice of support, because it
seems abundantly clear that the mechanism on zirconia under
these experimental conditions is different from that on ceria-
or ceria/zirconia-supported catalysts. Finally, even for this cat-
alyst, it must be stressed that the use of different experimental
conditions (such as much higher water concentration, e.g., 52%,
as used by Davis et al. [30], instead of the 7% used here) or a
different metal or method of preparation of a zirconia-supported
catalyst could result in different conclusions. The feed used in
the present study also lacked the CO2 that normally would be
present in a true fuel processor. Some experiments with added
CO2 were carried out, but the analysis of the data was partic-
ularly complex because of the occurrence of the reverse reac-
tion. It should be stressed that the dominant mechanism could
change with increasing partial pressure of CO2.
[
1
[4] K.A. Adamson, Energy Policy 32 (2004) 1231.
[5] C.C. Elam, C.E.G. Padro, G. Sandrock, A. Luzzi, P. Lindblad, E.F. Hagen,
Int. J. Hydrogen Energy 28 (2003) 601.
[6] V. Idakiev, T. Tabakova, A. Naydenov, Z.-Y. Yuan, B.-L. Su, Appl. Catal.
B Environ. 63 (2006) 178.
[
[
7] W. Ruettinger, O. Ilinich, R.J. Farrauto, J. Power Sources 118 (2003) 61.
8] R. Farrauto, S. Hwang, L. Shore, W. Ruettinger, J. Lampert, T. Giroux,
Y. Liu, O. Ilinich, Ann. Rev. Mater. Res. 33 (2003) 1.
9] P.S. Querino, J.R.C. Bispo, M.D. Rangel, Catal. Today 107–108 (2005)
920.
[
[10] H. Iida, A. Igarashi, Appl. Catal. A Gen. 303 (2006) 48.
[11] D. Tibiletti, A. Amieiro-Fonseca, R. Burch, Y. Chen, J.M. Fisher, A. Go-
guet, C. Hardacre, P. Hu, A. Thompsett, J. Phys. Chem. B 109 (47) (2005)
22553.
[12] S. Ricote, G. Jacobs, M. Milling, Y. Ji, P.M. Patterson, B.H. Davis, Appl.
Catal. A Gen. 303 (2006) 35.
[13] W. Liu, M. Flytzani-Stephanopoulos, J. Catal. 153 (1995) 317.
[
[
14] Q. Fu, H. Saltsburg, M. Flytzani-Stephanopoulos, Science 301 (2003) 935.
15] Y. Li, Q. Fu, M. Flytzani-Stephanopoulos, Appl. Catal. B Environ. 27
(
2000) 179.
16] T. Bunluesin, R.J. Gorte, G.W. Graham, Appl. Catal. B Environ. 15 (1998)
07.
17] R.J. Gorte, S. Zhao, Catal. Today 104 (2005) 18.
[
1
[
[18] S. Hilaire, S. Sharma, R.J. Gorte, J.M. Vohs, H.-W. Jen, Catal. Lett. 70
2000) 131.
19] S. Hilaire, X. Wang, T. Luo, R.J. Gorte, J. Wagner, Appl. Catal. A Gen. 215
2001) 271.
(
[
[
[
(
4
. Conclusion
20] C.M.Y. Yeung, F. Meunier, R. Burch, D. Thompsett, S.C. Tsang, J. Phys.
Chem. B 110 (2006) 8540.
21] D.C. Grenoble, M.M. Estadt, D.F. Ollis, J. Catal. 67 (1981) 90.
Only carbonyl and formate species were observed by in
situ DRIFTS during the WGS reaction over our Pt/ZrO2. The
steady-state surface coverage of carbonate species was negligi-
ble. Isotopic transient kinetic analyses revealed that formates
exchanged uniformly according to a first-order law, suggest-
ing that most formates observed by DRIFTS were of the same
reactivity. In addition, the time scale of the exchange of the re-
action product CO2 was significantly shorter than that of the
surface formates. Therefore, the formate route, which is based
on the formates seen on DRIFTS, can be ruled out as being
the main reaction pathway in the present case. The number of
precursors of the reaction product, CO2, was lower than the
number of surface Pt atoms, suggesting that carbonyl species,
or some “infrared-invisible” complex (not excluding formates)
at the interface between Pt and zirconia was the main reaction
intermediate. A simple redox mechanism could also explain the
present results.
[22] A. Luengnaruemitchai, S. Osuwan, E. Gulari, Catal. Commun. 4 (2003)
15.
2
[23] A. Venugopal, J. Aluha, M.S. Scurrell, Catal. Lett. 90 (2003) 1.
[24] S.L. Swartz, M.M. Seabaugh, C.T. Holt, W. Dawson, Fuel Cells Bull. 30
(2000) 7.
[25] T. Shido, Y. Iwasawa, J. Catal. 141 (1993) 71.
[26] G. Jacobs, L. Williams, U. Graham, G.A. Thomas, D.E. Sparks, B.H.
Davis, Appl. Catal. A Gen. 252 (2003) 107.
[27] G. Jacobs, L. Williams, U. Graham, D. Sparks, B.H. Davis, J. Phys. Chem.
B 107 (2003) 10398.
[28] G. Jacobs, U.M. Graham, E. Chenu, P.M. Patterson, A. Dozier, B.H.
Davis, J. Catal. 229 (2005) 499.
[
[
29] H. Iida, A. Igarashi, Appl. Catal. A Gen. 303 (2006) 48.
30] E. Chenu, G. Jacobs, A.C. Crawford, R.A. Keogh, P.M. Patterson, D.E.
Sparks, B.H. Davis, Appl. Catal. B Environ. 59 (2005) 45.
[31] G. Jacobs, E. Chenu, P.M. Patterson, L. Williams, D. Sparks, G. Thomas,
B.H. Davis, Appl. Catal. A Gen. 258 (2004) 203.
[
[
[
32] G. Jacobs, P.M. Patterson, L. Williams, E. Chenu, D. Sparks, G. Thomas,
B.H. Davis, Appl. Catal. A Gen. 262 (2004) 177.
33] G. Jacobs, P.M. Patterson, L. Williams, D. Sparks, B.H. Davis, Catal.
Lett. 96 (2004) 97.
34] M. Weibel, F. Garin, P. Bernhardt, G. Maire, M. Prigent, in: Catalysis
and Automotive Pollution Control II, vol. 71, Elsevier, Amsterdam, 1991,
p. 195.
Acknowledgments
The authors thank MEL for providing the ZrO2 sample and
Dr. Tiziano Montini, University of Trieste for performing the
chemisorption measurements. Financial support was provided
by MIUR (Italy) under the PRIN projects (to A.T.), the Carmac
EPSRC project, and the CONCORDE Coordination Action on
Nanostructured Oxides (NMP2-CT-2004-505834).
[35] R.A. Koeppel, A. Baiker, C. Schild, A. Wokaun, J. Chem. Soc. Faraday
Trans. I 87 (1991) 2821.
[36] D. Tibiletti, A. Goguet, F.C. Meunier, J.P. Breen, R. Burch, Chem. Com-
mun. (2004) 1636.
[37] X.Q. Gong, P. Hu, R. Raval, J. Chem. Phys. 119 (2003) 6324.

D. Tibiletti et al. / Journal of Catalysis 244 (2006) 183–191
191
[
[
[
38] X.S. Liu, W. Ruettinger, X.M. Xu, R. Farrauto, Appl. Catal. B Environ. 56
[53] N.B. Colthup, L.H. Daly, S.E. Wiberley, Introduction to Infrared and Ra-
man Spectroscopy, Academic Press, Boston, 2000.
[54] K. Nakamoto, Infrared and Raman Spectra of Inorganic and Coordination
Compounds, Wiley–Interscience, New York, 1986.
[55] S.L. Shannon, J.G. Goodwin, Chem. Rev. 95 (1995) 677.
[56] C.N. Costa, A.M. Efstathiou, J. Phys. Chem. B 108 (2004) 2620.
[57] Y. Schuurman, C. Mirodatos, Appl. Catal. A Gen. 151 (1997) 305.
[58] G. Jacobs, S. Ricote, U.M. Graham, P.M. Patterson, B.H. Davis, Catal.
Today 106 (2005) 259.
(2005) 69.
39] A. Goguet, F.C. Meunier, D. Tibiletti, J.P. Breen, R. Burch, J. Phys. Chem.
B 108 (2004) 20240.
40] F.C. Meunier, D. Tibiletti, A. Goguet, D. Reid, R. Burch, Appl. Catal. A
Gen. 289 (2005) 104.
[
41] E. Aneggi, J. Llorca, M. Boaro, A. Trovarelli, J. Catal. 234 (2005) 88.
42] Z.-Y. Ma, C. Yang, W. Wei, W.-H. Li, Y.-H. Sun, J. Mol. Catal A
Chem. 227 (2005) 119.
[
[
43] D. Tibiletti, Ph.D. thesis, School of Chemistry, Queen’s University Belfast
[59] J. Crank, The Mathematics of Diffusion, second ed., Clarendon Press, Ox-
ford, 1975.
(2004).
[
[
[
44] M.W. Balakos, S.S.C. Chuang, G. Srivinas, J. Catal. 140 (1993) 281.
45] J.M. Olinger, P.R. Griffiths, Anal. Chem. 60 (1988) 2427.
46] D. Tibiletti, A. Goguet, D. Reid, F.C. Meunier, R. Burch, Catal. Today 113
[60] G. Muller, T. Narbeshuber, G. Mirth, J.A. Lercher, J. Phys. Chem. 98
(1994) 7436.
[61] F.C. Meunier, L. Domokos, K. Seshan, J.A. Lercher, J. Catal. 211 (2002)
366.
[62] E. Guglielminotti, Langmuir 6 (1990) 1455.
[63] C. Morterra, E. Giamello, L. Osio, M. Volante, J. Phys. Chem. 94 (1990)
3111.
[64] N.B. Jackson, J.G. Ekerdt, J. Catal. 101 (1986) 90.
[65] R.G. Silver, C.J. Hou, J.G. Ekerdt, J. Catal. 118 (1989) 400.
[66] J.C. Frost, Nature 334 (1988) 577.
(
2006) 94.
47] G. Busca, J. Lamotte, J.C. Lavalley, V. Lorenzelli, J. Am. Chem. Soc. 109
1987) 5197.
[
[
[
(
48] P. Bazin, O. Saur, J.C. Lavalley, M. Daturi, G. Blanchard, Phys. Chem.
Chem. Phys. 7 (2005) 187.
49] A. Holmgren, B. Andersson, D. Duprez, Appl. Catal. B Environ. 22 (1999)
215.
[50] A. Trovarelli, Catal. Rev. Sci. Eng. 38 (1996) 439.
[51] C. Binet, M. Daturi, J.C. Lavalley, Catal. Today 50 (1999) 207.
[52] C. Li, Y. Sakata, T. Arai, K. Domen, K. Maruya, T. Onishi, J. Chem. Soc.
Faraday Trans. I 85 (1989) 929.
[67] N.O. Gonzales, A.K. Chakraborty, A.T. Bell, J. Phys. Chem. B 101 (1997)
10058.
[68] M.D. Rhodes, A.T. Bell, J. Catal. 233 (2005) 198.
[69] R. Burch, Phys. Chem. Chem. Phys., invited article, in preparation.