CO multipulse TAP studies of 2% Pt/CeO2 catalyst: Influence of catalyst pretreatment and temperature on the number of active sites observed

Article abstract of DOI:10.1016/j.jcat.2007.10.028

CO multipulse temporal analysis of products (TAP) experiments were used to characterize a ceria-supported platinum catalyst after various oxidative and reductive pretreatments using O₂, H₂O, CO₂, and H₂. Based on the amount of CO consumed, using the final CO-saturated catalyst composition as the common state point, the oxidatively pretreated catalyst could be described using a general scale. From a kinetic analysis of the CO multipulse responses, two kinetic regimes corresponding to two types of active sites could be identified. As the temperature was raised, the number of the most active sites did not change while the amount of the less active site increased. Comparison of the number of active sites determined from the TAP data reported herein with that determined by a previous steady-state isotope transient kinetic analysis experiment showed excellent agreement. This correlation indicates that the (very fast response) TAP experiments can provide information regarding the number and type of active sites that are relevant to a catalyst under real reaction conditions.

Full text of DOI:10.1016/j.jcat.2007.10.028

Journal of Catalysis 253 (2008) 303–311
www.elsevier.com/locate/jcat
CO multipulse TAP studies of 2% Pt/CeO2 catalyst: Influence of catalyst
pretreatment and temperature on the number of active sites observed
S.O. Shekhtman, A. Goguet, R. Burch, C. Hardacre ∗, N. Maguire
CenTACat, School of Chemistry and Chemical Engineering, Queen’s University Belfast, Belfast BT9 5AG, Northern Ireland, UK
Received 5 July 2007; revised 8 October 2007; accepted 30 October 2007
Abstract
CO multipulse temporal analysis of products (TAP) experiments were used to characterize a ceria-supported platinum catalyst after various
oxidative and reductive pretreatments using O , H O, CO , and H . Based on the amount of CO consumed, using the final CO-saturated catalyst
2
2
2
2
composition as the common state point, the oxidatively pretreated catalyst could be described using a general scale. From a kinetic analysis of
the CO multipulse responses, two kinetic regimes corresponding to two types of active sites could be identified. As the temperature was raised,
the number of the most active sites did not change while the amount of the less active site increased. Comparison of the number of active sites
determined from the TAP data reported herein with that determined by a previous steady-state isotope transient kinetic analysis experiment showed
excellent agreement. This correlation indicates that the (very fast response) TAP experiments can provide information regarding the number and
type of active sites that are relevant to a catalyst under real reaction conditions.
© 2007 Elsevier Inc. All rights reserved.
Keywords: TAP; Catalyst; Characterization; Ceria; Platinum
1. Introduction
reaction mechanisms by which the WGS and the PrOx reactions
proceed [14–25]. These may be split into two broad themes,
a redox process whereby the reactants oxidize and reduce the
surface sequentially and an intermediate-derived process in-
volving the formation of a surface species such as carboxylate,
carbonate, or formate. Burch has recently proposed that both
mechanisms may operate over the same catalyst for the WGS
reaction with the reaction conditions determining the dominant
route [9].
Over the last two decades, the temporal analysis of prod-
ucts (TAP) technique, developed by Gleaves and co-workers
[26,27], has been successfully applied to non-steady-state ki-
netic characterization and investigation of reaction mechanisms
of model and multicomponent industrial catalysts [28]. The
TAP technique provides an opportunity to control the catalyst
composition such that it does not change significantly during
a single-pulse experiment because there are far fewer mole-
cules pulsed than surface active sites/species. Furthermore, the
catalyst state may be changed incrementally by exposing the
catalyst to multiple pulses of gas [29].
The development of hydrogen fuel cells is among the most
promising of all of the future energy solutions, providing ef-
ficient power with little environmental impact. However, pro-
viding hydrogen in a form that has sufficient purity not to de-
activate the fuel cell is a major challenge. The reforming of
hydrogen-rich fuels, including from renewable sources, is of
major importance [1–4]. Such a reforming process involves a
catalytic water–gas shift (WGS) step, which converts water into
hydrogen while also removing CO, a poison for the fuel cell,
forming CO2; further removal of CO is achieved by selective
oxidation (PrOx) without oxidation of the hydrogen [5]:
CO + H2O ꢀ CO2 + H2.
Recent studies have shown that noble metals (e.g., Pt, Au) sup-
ported on CeO2 are promising low-temperature catalysts for
these reactions [6–13]. However, there a number of conflicting
(1)
*
In traditional approaches to the characterization of ac-
tive sites, the sites are quantified using separate probe mole-
Corresponding author.
E-mail address: c.hardacre@qub.ac.uk (C. Hardacre).
0021-9517/$ – see front matter © 2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2007.10.028

304
S.O. Shekhtman et al. / Journal of Catalysis 253 (2008) 303–311
cule adsorption experiments, for example, in calculation of
the turnover frequency. In a multipulse TAP experiment, the
amount of reactant consumed (analogous to an adsorption ex-
periment) and the amount of product formed (characteristic
of active surface species) in each single-pulse experiment are
measured. These amounts are added from pulse-to-pulse to
characterize the catalyst composition before each pulse [29].
On the completion of the multipulse experiment (i.e., when no
further reaction is occurring), the total amount of the reactant
consumed and product formed are determined; this is analogous
to the uptake. An important advantage of the multipulse TAP
experiment is that it allows measurement of a gradual change
in the number of active sites/species as well as a change in the
overall activity associated with the remaining sites. Thus, in a
multipulse experiment it is possible to view the complete de-
pendence of catalyst activity as a function of the amount of
available active sites. A detailed analysis of this dependence
allows the sites to be distinguished kinetically with respect to
how rapidly the activity changes (e.g., decreases) as each site is
taken up.
This paper provides the first description of how this method-
ology may be applied to multipulse TAP data. Such a multipulse
experiment is particularly informative when combined with a
thin-zone TAP reactor (TZTR) [30], which ensures uniformity
in the catalyst zone [31]. The TZTR multipulse experiment al-
lows state-by-state transient screening of the catalyst to be used
to examine the active sites [32]. The catalyst state is contin-
uously changed by a long multipulse experiment and probed
kinetically by each single-pulse TAP experiment. A complete
characterization of the catalyst using this methodology involves
quantification/scaling of all catalyst states based on the amount
of consumed molecules (e.g., according to reduction or oxida-
tion), followed by understanding of the observed single-pulse
kinetics versus this scale of catalyst state.
The TAP pulse-response experiments were performed in a
TAP-I reactor (Autoclave Engineers) using a stainless steel
microreactor (41 mm long and 5.5 mm i.d.). In all of the
experiments, the TZTR method was used where the reactor
was packed with 2–3 mg of catalyst sandwiched between two
500 mg beds of silicon carbide. All particles were 250–450 µm
in size. The reactor temperature was measured by a thermo-
couple positioned in the center of the catalyst bed. Reactants
and products were recorded using a UTI 100C quadrupole mass
spectrometer with a total collection time of 10 s. Each pulse
comprised a 1:1 mix of 13CO and argon. 13CO was used to
eliminate the influence of the high background observed at an
AMU of 28 as part of the residual gas atmosphere. Masses
at AMUs of 29, 40, and 45 were followed in the TAP tran-
sient mode. A typical multipulse experiment comprised >6000
pulses. The characteristic single-pulse responses were obtained
by averaging a number of responses, varying from 2 responses
at the beginning of the experiment to 10 responses at the end of
the multipulse experiment. In the entire multipulse experiment
reported, pulse intensity was determined from the monitored
argon response intensity (which gradually changed during the
experiment) and the pressure drop in the pulse chamber.
The following procedure was used for all experiments:
1. Pretreatment of the catalyst by exposing it to different gases
at 1 atm pressure. Four different pretreatments were inves-
tigated:
I. 20% O2 flow at 300 ◦C for 1 h.
II. 20% O2 flow at 300 ◦C for 1 h, followed by flowing H2
at 300 ◦C for 1 h, followed by 10% H2O flow at 300 ◦C
for 1 h.
III. 20% O2 flow at 300 ◦C for 1 h, followed by flowing
H2 at 300 ◦C for 1 h, followed by 100% CO2 flow at
300 ◦C for 1 h.
IV. 20% O2 flow at 300 ◦C for 1 h, followed by flowing H2
at 300 ◦C for 1 h.
The present work used the TZTR multipulse experimental
methodology to characterize the active sites for CO conversion
on a Pt/CeO2 catalyst studied previously using TAP for the re-
verse WGS reaction [33]. CO was used as a probe molecule in
multipulse experiments performed as a function of temperature
and normal pressure catalyst pretreatment under O2, H2, CO2,
and H2O. The characterization shows the presence of two dis-
tinct sites, the numbers of which are strongly determined by the
catalyst state.
2. Evacuation of the reactor to 10−6 Torr.
3. Vacuum pulse-response TAP experiments using CO multi-
pulses at the desired temperature.
The exit flow rate time dependence for each reactant/product
and inert was determined using standard mass spectrometry
fragmentation patterns and sensitivity factors normalized to ar-
gon.
2. Experimental
3. Results and discussion
The 2% Pt/CeO2 catalyst, provided by Johnson Matthey, had
a BET surface area of 180 m2 g−1 and a platinum dispersion
of 17%. The metal dispersion was measured by H2 chemisorp-
tion; note that the dispersion was measured at −80 ◦C, to mini-
mize the potential spillover of H2 onto the support [34]. This
corresponds to a concentration of surface platinum atoms of
∼2 × 10−8 molmgc−a1t . All gases were >99.9% purity and were
supplied by BOC. Double-distilled deionized water (18.2 Mꢀ)
was used. Before each reaction, the catalyst was activated under
flowing 20% oxygen in nitrogen for 1 h.
3.1. Influence of pretreatment
This section reports the results of CO multipulse experi-
ments over various catalysts that underwent different pretreat-
ments. During each multipulse experiment, the intensity of the
CO responses increased (i.e., CO consumption decreased) with
increasing pulse number; ultimately reaching a point at which
CO molecules were no longer consumed by the catalyst. The
single-pulse CO conversion was determined as the ratio be-
tween the amount of CO consumed and amount of CO injected

S.O. Shekhtman et al. / Journal of Catalysis 253 (2008) 303–311
305
◦
Fig. 1. Single-pulse CO conversion variation with respect to the amount of consumed CO prior to each CO pulse at 300 C as a function of O
(
), H O ( ),
2
2
CO
(
), and H
(
2
) catalyst pretreatments.
2
into the reactor in a single-pulse. The change in catalyst com-
position in a multipulse experiment was characterized by the
amount of CO molecules consumed by the catalyst before each
pulse in the series (calculated from a single-pulse CO consump-
tion and pulse intensity monitored by the argon responses).
This scale is much more informative than the pulse number,
because both pulse intensity and CO consumption change dur-
ing a TAP multipulse experiment. Thus, the change in catalyst
composition varies significantly from pulse to pulse, and the
scale-based on the amount of CO consumed explicitly quanti-
fies this change.
Fig. 1 shows typical single-pulse CO conversion profiles as a
function of the amount of CO consumed before each CO pulse
over 2 mg of pretreated 2% Pt/CeO2 catalyst in a CO multipulse
TAP experiment at 300 ◦C. Four profiles are shown to illus-
trate the influence of the normal pressure pretreatments (I–IV).
The total amount of CO consumed in the full multipulse ex-
periment, the first pulse activity, and the profiles generally are
different for the different catalyst pretreatments. These results
indicate that, as expected, different normal pressure pretreat-
ment and evacuation regimes had different effects on catalyst
surface composition.
Despite these differences, a more informative graphical rep-
resentation of conversion profiles was found. In this represen-
tation, the pulse at which no more CO was consumed was used
as a common end point, corresponding to the point at which
the catalyst was saturated with CO in all four cases. Fig. 2
presents the same (as in Fig. 1) single-pulse CO conversion pro-
files plotted using this common end point. This figure clearly
shows that the single-pulse CO conversion observed in multi-
pulse experiments over differently preoxidized catalysts (i.e.,
using O2, H2O, and CO2) follows a single curve when plot-
ted against this new general scale. Within this general scale,
a different oxidizing pretreatment is reflected as a merely dif-
ferent starting point on the general curve according to total CO
consumption for each pretreatment, indicating that I (O2) >
II (H2O) > III (CO2). The maximum amount of CO consump-
tion was found after the oxygen treatment, corresponding to
1.28 × 10−6 molmgc−a1t , whereas after the H2O treatment, con-
sumption was lower, by ∼5 × 10−8 molmgc−a1t . This difference
is within the range of the number of surface platinum atoms,
possibly indicating that H2O oxidizes the same sites as O2
except those sites associated with platinum atoms. Compared
with the O2 pretreatment, CO2 pretreatment resulted in l−ow1 er
CO consumption. The difference of 2.3 × 10−7 molmgcat is
clearly greater than the number of platinum atoms. In addi-
tion, the first pulse activity dropped by ∼25% compared with
the oxygen treatment case. For all oxidation pretreatments,
a small amount of CO2 (10–15% of CO consumed) was ob-
served in the form of a broad feature indicating slow reaction/
desorption.
As discussed below, most of the CO is consumed during the
reaction to form a stable surface intermediate, such as surface
carbonate. Therefore, although CO2 pretreatment will reoxi-
dize the catalyst surface after reduction, it also will produce
surface carbonates, which may block the active sites and cause
lower first pulse activity and a decrease in total CO consump-
tion compared with either O2 or H2O pretreatment. In summary,
the oxygen pretreatment results in a completely oxidized cata-
lyst surface, both platinum and support; the water oxidizes only
the oxide support, and not the metal sites as follows from dif-
ference; and the CO2 pretreatment not only oxidizes the oxide
support, but also blocks some other active sites through the for-
mation of surface carbonates.
In contrast to the oxidative treatments, after H2 pretreatment
only, no detectable amount of CO2 produced was observed,
although a significant CO consumption activity for the initial
pulses was seen (Fig. 1). This is likely due to CO disproportion-
ation, with the CO2 formed during the reaction being trapped
through strong readsorption on the reduced ceria [35]. The con-
version profile for H2 pretreatment did not follow the general
curve as for the oxidized catalyst (Fig. 2). With an increasing

306
S.O. Shekhtman et al. / Journal of Catalysis 253 (2008) 303–311
◦
Fig. 2. Single-pulse CO conversion variation with respect to the amount of consumed CO prior to each CO pulse at 300 C as a function of O
(
), H O ( ),
2
2
CO
(
), and H
(
) catalyst pretreatments. Compared with the traces shown in Fig. 1, the curves have been shifted in order for all the profiles to have a common
2
2
ending point providing evidence of a general scale over a range pretreatments.
number of CO pulses, the activity decreased, but for the re-
duced catalyst, fewer pulses of CO were required before the
catalyst activity reached zero. This finding was expected, be-
cause the capacity to adsorb/react with CO after a reductive
treatment will decrease due to removal of labile oxygen from
the support. Interestingly, the amount of CO consumed far ex-
ceeded the number of exposed Pt atoms (CO/Pt ∼ 40) which is
consistent with previous observations made for this catalyst that
the carbon produced through CO disproportionation is unlikely
to be located only on the Pt [36]. In this case the disproportiona-
tion either occurs on the metal and part of the carbon spills over
the support, or at least part of the disproportionation reaction
occurs directly on the ceria support.
We analyzed single-pulse responses using moment formal-
ism as described in detail previously [29]. According to this
theory, each moment of observed response (M0, M1, M2) al-
lows the determination of only one independent kinetic para-
meter (R0, R1, R2), called the primary kinetic coefficient. For
the TZTR CO multipulse responses, the zeroth coefficient, R0,
is calculated from the zeroth moment, M0, as
The second coefficient, R2, is calculated from the second
moment, M2, as
ꢀ
ꢁ
M12
2τcatM02 τcatM03
M2
τin τinR0
τin
−R2 =
−
+
+ 2R1 +
,
(4)
6
5
τcat
where R2 has the dimension of time (s) and describes CO ap-
parent time delay on the catalyst surface.
LrLcat
τcat
=
2Din
is the residence time in the catalyst zone, and
εinL2r
τin =
2Din
is the residence time in inert zones. The explicit expres-
sions (2)–(4) have been recently detailed by Shekhtman et
al. [37].
The fact that R1 and R2 calculated for CO were >0 through-
out the entire multipulse experiment suggested a holdup of the
CO in the reactor bed (as proposed in [29]). The same con-
clusion follows from the comparison of typical argon and CO
responses observed at the end of a multipulse experiment shown
in Fig. 3 where it is clear that the CO response is broader. This
indicates that the CO molecule adsorbs reversibly during the
multipulse experiments. The fact that the CO consumption is
not balanced by the CO2 release indicates that the adsorbed CO
also transforms into a long-lived surface intermediate. These
conclusions lead to the following mechanism for CO adsorp-
tion/desorption and reaction:
1 − M0
R0 =
,
(2)
τcatM0
R0 has the dimension of reciprocal time (s) and represents an
apparent rate constant of CO consumption.
The first coefficient, R1, is calculated from the first moment,
M1, as
ꢀ
ꢁ
M1
τcatM02
1
τcat
R0
R1 =
− τin
+
.
(3)
k
a
3
−→
CO
COads,
←−
k
d
R1 is dimensionless and represents CO apparent “intermediate-
gas” constant, which relates gaseous and adsorbed CO.
k
trans
COads −→ Intermediate.
(5)

S.O. Shekhtman et al. / Journal of Catalysis 253 (2008) 303–311
307
Fig. 3. Typical normalized argon (grey thick) and CO (black) responses observed near the end of multipulse experiment.
◦
Fig. 4. Rate constant of surface CO transformation versus the amount of consumed CO prior to each CO pulse at 300 C as a function of O
(
), H O ( ), and
2
2
CO
( ) catalyst pretreatments. Dashed lines indicate two linear regimes.
2
The rate constants of these elementary steps (i.e., for CO
alysts also follows a single curve. This curve may be treated
as a combination of two linear relationships that are indicative
of two types of active sites, as indicated by the lines shown in
Fig. 4.
adsorption, desorption, and surface reaction) were determined
from the primary kinetic coefficients (R0, R1, and R2) us-
ing relationships reported in Table 1 (reversible adsorption +
irreversible reaction) of [29]. In particular,
3.2. Influence of reaction temperature
k
transR12
R0R2
1
ktrans R0 R1
R1 R2
−
=
and kd =
.
(6)
The variation in CO conversion dependence versus the
amount of CO consumed for the oxygen-pretreated catalyst
as a function of reaction temperature was measured to deter-
mine the influence of temperature on of the number and activity
of different active sites. Oxygen pretreatment was chosen be-
cause it provides the most extensive multipulse dependence, as
discussed in the previous section. In all the experiments, the
normal pressure oxidation and evacuation were conducted at
300 ◦C to produce a similar catalyst surface. Thereafter, the
As reported previously in [33], the desorption constant re-
mained around 1.5 s−1 during all multipulse experiments while
the transformation constant varied. As found in Fig. 2, the
single-pulse CO conversion observed in the multipulse exper-
iments over differently preoxidized catalysts followed a sin-
gle curve when plotted against the general scale. Fig. 4 shows
the surface transformation constant plotted against the general
scale. Clearly, ktrans observed over differently preoxidized cat-

308
S.O. Shekhtman et al. / Journal of Catalysis 253 (2008) 303–311
Fig. 5. The single-pulse CO conversion versus the amount of CO consumed prior to each CO pulse observed for oxygen pretreated catalyst at 250 ( ), 300 ( ),
◦
and 350 C ( ).
◦
Fig. 6. CO apparent rate constant versus the amount of CO consumed per mg of catalyst for the oxygen pretreated catalyst at 250 ( ), 300 ( ), and 350 C ( ).
temperature was adjusted to the desired value under vacuum
before the corresponding CO multipulse.
perature, each curve is again characterized by two regions, with
the transition point between the regions denoted by a dashed
line in Fig. 5. The amount of CO consumed within the region
in which the conversion decreased most rapidly did not change
with the temperature, remaining at ∼1.5 × 10−7 molmgc−a1t . In
contrast, in the second region, the amount of CO consumed in-
creased with increasing temperature, leading to the observation
that the overall amount of CO consumed also increased. A sim-
ilar pattern occurred for the apparent rate constant (R0) shown
in Fig. 6. The apparent rate constant decreased rapidly for the
first 1.5 × 10−7 mol of CO consumed, then decreased much
more gradually to zero activity. Again, the two regions found in
Fig. 5 are clearly shown in Fig. 6.
Figs. 5 and 6 show the single-pulse CO conversion and CO
apparent rate constant [i.e., R0 calculated from Eq. (2)], as a
function of the amount of CO consumed before the correspond-
ing CO pulse observed at 250, 300, and 350 ◦C over 3 mg of
catalyst, respectively. Good consistency was found for the to-
tal amount of CO consumed in the multipulse experiment and
the kinetic characteristics at 300 ◦C between 2 and 3 mg of cat-
alyst, indicating that the TZTR remained well defined. From
Fig. 5, in general, the single-pulse conversion increased with
temperature throughout the entire multipulse experiment, as ex-
pected. Furthermore, the total amount of CO consumed in a
completed multipulse experiment also increased with temper-
ature, possibly indicating increasingly deeper reduction of the
support with increasing temperature. Irrespective of the tem-
According to a general concept of active sites, the rate con-
stant should be proportional to the amount of available sites
and should decrease linearly as these sites are taken up. The

S.O. Shekhtman et al. / Journal of Catalysis 253 (2008) 303–311
309
(a)
(b)
◦
Fig. 7. CO apparent rate constants observed in multipulse experiments at 250 ( ), 300 ( ), and 350 C ( ) and rate constants, k
, calculated using the
model
model described by Eq. (2) (straight lines) as a function of the CO consumed. The x-axis of plot (b) is expanded compared with plot (a) to illustrate the fit in the
fast and slow change regions, respectively.
decrease in the CO apparent rate constant may be fitted by a
combination of two linear functions at all three temperatures,
but not by a single linear or quasi-linear function. The linear
decrease of the rate constant with CO consumption is indica-
tive of domination of a single type of active site when each CO
molecule takes up an active site while being consumed. The dif-
ferent slopes of the linear dependency indicate different activity
for the sites. Consequently, these two regions can be associated
with different active sites. The overall activity can be modeled
as a weighted average of the rate constants for an individual site
and the number of sites available,
experiment, and the amount of CO consumed in each pulse is
measured. The amount of CO consumed that reacts with sites 1
and 2 is proportional to the activity of the sites [i.e., k1(1 − θ1)
or k2(1−θ2), respectively], as calculated for the previous pulse.
The pulse-to-pulse evolution of model kmodel [Eq. (7)] was then
calculated from the observed single-pulse CO consumption and
fitted to experimentally observed kinetic dependencies (Fig. 6)
using least squares regression. Fig. 7 shows this fit, indicat-
ing that the reaction data is consistent with a two-site model
(but not a single-site model) in which the sites of very different
activity work independently in a parallel manner. Significant
differences in activity clearly separate the contribution of each
kind of site into two quasi-linear regions. The values of the fit-
ting parameters [i.e., the numbers (N1, N2) and activity (k1, k2)
of the two kinds of active sites] are given in Table 1. Table 1
also compares the equivalent site density determined using a
steady-state isotope transient kinetic analysis (SSITKA) exper-
iment at 225 ◦C [20]. Similar values for the calculated amounts
of two distinct kinds of active sites were found for the TAP
kmodel = k1(1 − θ1) + k2(1 − θ2),
(7)
where θ1 = NCO,1/N1 is the coverage of CO over more active
sites, N1; θ2 = NCO,2/N2 is the coverage of CO over less active
sites, N2; and k1 and k2 represent the activity of corresponding
sites.
With CO consumption, both NCO,1 and NCO,2 increase from
zero to maximum values (N1, N2) over the entire multipulse

310
S.O. Shekhtman et al. / Journal of Catalysis 253 (2008) 303–311
Table 1
Fitting parameters of the two site model shown in Eq. (7) associated with the low and high activity sites
Temperature
( C)
More active sites
Less active sites
◦
−1
−1
)
cat
−1
−1
)
Number, N (mol mg
)
Kinetic constant, k (s
)
Number, N (mol mg
Kinetic constant, k (s
1
1
2
2
cat
−7
−6
250
300
350
1.30 × 10
450
900
1050
0.75 × 10
11
21
55
−7
−6
1.35 × 10
1.15 × 10
−7
−6
1.33 × 10
1.95 × 10
−1
−1
Activation energy: 23 kJ mol
Activation energy: 44 kJ mol
−1
−1
◦
−7
−7
SSITKA data at 225 C [20]
CO precursor: 2.2 × 10 molmg
CO precursor: 8.4 × 10 molmg
2
cat
cat
and that only the temperature at which the reaction is performed
governs which one predominates.
4. Conclusion
In this work, CO multipulse TAP experiments were per-
formed to characterize the number and type of active sites on a
Pt/CeO2 CO oxidation/WGS catalyst. The influence of normal
pressure pretreatment and reaction temperature were studied in
detail. The influence of different normal pressure pretreatments
was quantified based on the amount of CO consumed. The ki-
netic characteristics (e.g., conversion) observed over differently
preoxidized catalysts followed a single curve when plotted us-
ing the final CO-saturated catalyst composition as the common
state. Different pretreatments are represented as different start-
ing points on this curve (indicating different levels of oxida-
tion).
Fig. 8. Arrhenius plot for the rate constants k ( ) and k ×10 ( ) determined
1
2
from the fitting using Eq. (2) for the high and low activity sites, respectively.
Based on kinetic data obtained at different temperatures, the
catalyst was found to contain two distinct, independent sites
each with significantly different activity. These sites were con-
sidered to be associated with the metal/metal-modified support
and the support corresponding to high and low activity, respec-
tively. The number of low activity sites was found to increase
with temperature, whereas little temperature variation was seen
for the higher-activity sites. The higher activity of each type
of site followed an Arrhenius dependence, with higher activa-
tion energy for less-active sites. Good agreement was found
between the TAP and SSITKA experiments, demonstrating that
TAP experiments can provide information regarding the active
sites relevant to a catalyst under real reaction conditions.
vacuum multipulse experiments and the reverse WGS SSITKA
study [33]. From the SSITKA data, the amounts of gaseous
CO and CO2 released by the catalyst surface after the isoto−pe1
switch were determined to be NCO = 2.2 × 10−7 molmgcat
and NCO = 8.4 × 10−7 molmgc−a1t , respectively, at 225 ◦C. The
2
activity (k1, k2) of each site can be described by an Arrhe-
nius dependence as shown in Fig. 8 with activation energies
of 23 kJ mol−1 for the higher activity site and 44 kJ mol−1 for
the lower activity site.
The thermal variation may be investigated by assigning the
more active sites to those associated with the platinum parti-
cles (i.e., on both the metal and those in the close vicinity),
because the number of these sites exceeds the amount of plat-
inum atoms. The less active sites can be associated only with the
support. This is consistent with an earlier observation made for
this catalyst associating the most active sites with close proxim-
ity of the Pt particles [36]. These sites can be easily reduced and
thus likely to show little temperature variation in the number of
sites available. In contrast, sites further away from the metal are
more difficult to reduce, resulting in a significant temperature
variation. Interestingly, the fact that the two sites have different
apparent activation barriers demonstrates that the lower-activity
site will dominate at higher temperatures, resulting in a switch
to a different reaction mechanism as a function of temperature.
This is particularly relevant to the WGS reaction, for which a
debate is ongoing as to whether an associative or a redox mech-
anism predominates. It is possible that both mechanisms occur
Acknowledgments
The authors thank Johnson Matthey for supplying the cata-
lyst. Funding was provided by the EPSRC under the CARMAC
project. A studentship (for N.M.) was provided by DEL.
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] Q. Fu, A. Weber, M. Flytzani-Stephanopoulos, Catal. Lett. 77 (2001) 87.

S.O. Shekhtman et al. / Journal of Catalysis 253 (2008) 303–311
311
[7] S. Ricote, G. Jacobs, M. Milling, Y. Ji, P.M. Patterson, B.H. Davis, Appl.
Catal. A 303 (2006) 35.
[8] 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 (2005)
22553.
[9] R. Burch, Phys. Chem. Chem. Phys. 8 (2006) 5483.
[10] A.F. Ghenciu, Curr. Opin. Solid State Mater. Sci. 6 (2002) 389.
[11] H. Son, A.M. Lane, Catal. Lett. 76 (2001) 151.
[24] O. Pozdnyakova, D. Teschner, A. Wootsch, J. Krohnert, B. Steinhauer,
H. Sauer, L. Toth, F.C. Jentoft, A. Knop-Gericke, Z. Paal, R. Schlogl,
J. Catal. 237 (2006) 17.
[25] A. Wootsch, C. Descorme, D. Duprez, J. Catal. 225 (2004) 259.
[26] J.T. Gleaves, J.R. Ebner, T.C. Kuechler, Catal. Rev. Sci. Eng. 30 (1988)
49.
[27] J.T. Gleaves, G.S. Yablonskii, P. Phanawadee, Y. Schuurman, Appl. Catal.
A 160 (1997) 55.
[12] R. Farrauto, S. Hwang, L. Shore, W. Ruettinger, J. Lampert, T. Giroux,
Y. Liu, O. Ilinich, Ann. Rev. Mater. Res. 33 (2003) 1.
[13] D. Duprez, J. Chim. Phys. Phys. Chim. Biol. 92 (1995) 1952.
[14] T. Bunluesin, R.J. Gorte, G.W. Graham, Appl. Catal. B 15 (1998) 107.
[15] R.J. Gorte, S. Zhao, Catal. Today 104 (2005) 18.
[16] T. Shido, Y. Iwasawa, J. Catal. 136 (1992) 493;
[28] G.S. Yablonsky, M. Olea, G.B. Marin, J. Catal. 216 (2003) 120.
[29] S.O. Shekhtman, G.S. Yablonsky, J.T. Gleaves, R. Fushimi, Chem. Eng.
Sci. 58 (2003) 4843.
[30] S.O. Shekhtman, G.S. Yablonsky, J.T. Gleaves, S. Chen, Chem. Eng. Sci.
54 (1999) 4371.
[31] S.O. Shekhtman, G.S. Yablonsky, Ind. Eng. Chem. Res. 44 (2005) 6518.
[32] S.O. Shekhtman, G.S. Yablonsky, J.T. Gleaves, R. Fushimi, Chem. Eng.
Sci. 59 (2004) 5493.
[33] A. Goguet, S.O. Shekhtman, R. Burch, C. Hardacre, F.C. Meunier, G.S.
Yablonsky, J. Catal. 237 (2006) 102.
[34] S. Bernal, F.J. Botana, J.J. Calvino, M.A. Cauqui, G.A. Jobacho, J.M. Pin-
tado, J.M. Rodriguez-Izquierdo, J. Phys. Chem. 97 (1993) 4118.
[35] F.C. Meunier, D. Tibiletti, A. Goguet, D. Reid, R. Burch, Appl. Catal.
A 289 (2005) 104.
[36] A. Goguet, F. Meunier, J.P. Breen, R. Burch, M.I. Petch, A. Faur Ghenciu,
J. Catal. 226 (2004) 382.
T. Shido, Y. Iwasawa, J. Catal. 141 (1993) 71.
[17] G. Jacobs, B.H. Davis, Appl. Catal. A 284 (2005) 31.
[18] G. Jacobs, S. Ricote, B.H. Davis, Appl. Catal. A 302 (2006) 14.
[19] F.C. Meunier, D. Tibiletti, A. Goguet, S. Shekhtman, C. Hardacre, R.
Burch, Catal. Today 126 (2007) 143.
[20] A. Goguet, F.C. Meunier, D. Tibiletti, J.P. Breen, R. Burch, J. Phys. Chem.
B 108 (2004) 20240.
[21] A. Wootsch, C. Descorme, D. Duprez, J. Catal. 225 (2004) 259.
[22] A. Trovarelli, Catal. Rev. Sci. Eng. 38 (1996) 439.
[23] O. Pozdnyakova, D. Teschner, A. Wootsch, J. Krohnert, B. Steinhauer,
H. Sauer, L. Toth, F.C. Jentoft, A. Knop-Gericke, Z. Paal, R. Schlogl,
J. Catal. 237 (2006) 1.
[37] S.O. Shekhtman, N. Maguire, A. Goguet, R. Burch, C. Hardacre, Catal.
Today 121 (2007) 255.