Light-off behaviour of PdO/γ-Al2O3 catalysts for stoichiometric CO-O2 and CO-O2-NO reactions: A combined catalytic activity-in situ DRIFTS study

Article abstract of DOI:10.1016/S0021-9517(03)00277-X

Three PdO/γ-Al₂O₃ catalysts differing in Pd loading (between 0.05 and 1 Pd wt%) have been examined with regard to their light-off catalytic activity for CO oxidation and NO reduction reactions under stoichiometric conditions. Catalytic activity results are explained on the basis of DRIFTS analysis of the adsorbed species present under reaction conditions. It is shown that differences between the catalysts (apart from the expected increasing activity with the Pd loading for both reactions) are considerably greater for NO reduction than for CO oxidation reactions. This is explained by structural differences between the active metallic Pd particles formed during the particle nucleation/growth process that takes place during the course of the light-off run upon interaction with the reactant mixture. On the basis of differences in the natures and relative intensities of adsorbed CO and NO species present during competition for atop and bridging sites over the Pd particles, the correlation (within the Pd loading studied) between NO reduction capability and Pd loading is attributed to the increasing NO dissociation efficiency as the relative size of the particles formed during the course of the reaction is increased.

Full text of DOI:10.1016/S0021-9517(03)00277-X

Journal of Catalysis 221 (2004) 85–92
www.elsevier.com/locate/jcat
Light-off behaviour of PdO/γ -Al2O3 catalysts for stoichiometric
CO–O2 and CO–O2–NO reactions: a combined catalytic
activity–in situ DRIFTS study
A. Martínez-Arias,a,∗ A.B. Hungría,a M. Fernández-García,a A. Iglesias-Juez,a
J.A. Anderson,b and J.C. Conesa a
a
Instituto de Catálisis y Petroleoquímica, CSIC, Campus Cantoblanco, 28049 Madrid, Spain
b
Division of Physical and Inorganic Chemistry, University of Dundee, Dundee DD14HN, Scotland, United Kingdom
Received 28 April 2003; revised 12 June 2003; accepted 13 June 2003
Abstract
Three PdO/γ -Al O catalysts differing in Pd loading (between 0.05 and 1 Pd wt%) have been examined with regard to their light-off
2
3
catalytic activity for CO oxidation and NO reduction reactions under stoichiometric conditions. Catalytic activity results are explained on the
basis of DRIFTS analysis of the adsorbed species present under reaction conditions. It is shown that differences between the catalysts (apart
from the expected increasing activity with the Pd loading for both reactions) are considerably greater for NO reduction than for CO oxidation
reactions. This is explained by structural differences between the active metallic Pd particles formed during the particle nucleation/growth
process that takes place during the course of the light-off run upon interaction with the reactant mixture. On the basis of differences in the
natures and relative intensities of adsorbed CO and NO species present during competition for atop and bridging sites over the Pd particles,
the correlation (within the Pd loading studied) between NO reduction capability and Pd loading is attributed to the increasing NO dissociation
efficiency as the relative size of the particles formed during the course of the reaction is increased.
2003 Elsevier Inc. All rights reserved.
Keywords: Pd/γ -Al O catalysts; CO oxidation; NO reduction; Light-off catalytic activity; In situ DRIFTS
2
3
1. Introduction
lyst configuration in terms of which metal–oxide interactions
(Pd–alumina, Pd–promoter or both) are most favourable in
promoting NO reduction reactions [4,7,8], in particular dur-
ing the light-off period during which the greatest proportion
of the whole toxic emission is produced over a driving cy-
cle [1]. Thus, some reports show that introduction of the
Ce-containing promoter to the Pd-only catalyst produces a
decrease in NO conversion at relatively low temperatures
(around 600 K) although it is enhanced at higher temper-
atures (around 770 K) [4], suggesting that Pd–promoter
interactions may be detrimental for light-off performance
of the system. However, other authors propose that oxy-
gen vacancies created upon reduction of the promoter in
the proximity of palladium can act as promoting sites for
NO reduction, leading to an enhanced NO conversion dur-
ing light-off [7,8]. It is possible that the reasons behind
these discrepancies lie mainly in differences in the particular
configuration of active components present in each specific
case. Such differences may significantly affect NO reduction
Three-way catalysts (TWC) have been widely used to
reduce pollutant emissions from gasoline engine-powered
vehicles [1]. Basic components of these systems usually in-
clude Rh, Pt, and/or Pd as active metals, zirconia-ceria as
promoter, and alumina as a high surface thermally stable
support [1,2]. More recently, considerable attention has been
paid to the use of Pd as the single active metal component
in TWC on the basis of economical aspects (the high cost
and limited supply of Rh) and the availability of cleaner
fuels and considering also its remarkable activity for ox-
idation reactions [3,4]. However, some limitations are ap-
parent for Pd-only TWC with respect to their performance
in NO reduction reactions [4–6]. This is partly linked to
the controversy existing with respect to the optimum cata-
*
Corresponding author.
E-mail address: amartinez@icp.csic.es (A. Martínez-Arias).
0021-9517/$ – see front matter
2003 Elsevier Inc. All rights reserved.
doi:10.1016/S0021-9517(03)00277-X

86
A. Martínez-Arias et al. / Journal of Catalysis 221 (2004) 85–92
processes given the structural dependency of NO activation
processes over Pd catalysts and the dynamic operating mode
of the TWCs, which may induce important redox changes
in the active components as a function of the variable at-
mosphere [1,9]. These aspects have motivated the present
study, in which preoxidised Pd/Al2O3 catalysts differing in
the Pd loading present in each case were examined with
the aim of determining the factors affecting their light-off
performance for CO oxidation–NO reduction processes un-
der different conditions of Pd dispersion/interaction with the
alumina support. For this purpose, the catalytic activity of
the systems under stoichiometric CO–O2 and CO–O2–NO
light-off conditions has been examined in combination with
in situ DRIFTS results in order to correlate catalytic proper-
ties with the state of the catalyst surface and the processes
taking place on it under reaction conditions.
recording one spectrum (average of 25 scans at 4 cm−1 res-
olution) generally every 15–20 K. The gas mixture (using
the same concentrations as employed for the catalytic tests)
was prepared using a computer-controlled gas blender with
ca. 75 cm3 min−1 passing through the catalyst bed at at-
mospheric pressure, which roughly corresponds to the space
velocity condition employed for the reaction tests. Online
NOx analysis at the outlet of the IR chamber was performed
by chemiluminescence (Thermo Environmental Instruments
42C).
3. Results and discussion
The main results of the light-off catalytic activity tests for
the CO–O2 and CO–O2–NO reactions are shown in Fig. 1.
Results obtained for the CO oxidation with oxygen alone
show a gradual shift of CO conversion values to higher tem-
perature with decreasing Pd loading. The form of the curves
shows a kind of “staircase” shape for the 1% Pd catalyst and,
to a lesser extent, for the 0.5% Pd one; this may also occur
for the 0.05% Pd catalyst although it shows a smoother pro-
file in this respect. Different factors could account for these
irregularities, such as the presence of deactivation phenom-
ena resulting from self-poisoning effects or catalytic inter-
ferences among reactants or intermediates [5,9,10]. Another
possibility, considering the essentially transient character of
these experiments, is related to the occurrence of changes in
the number or nature of active sites during the course of the
reaction as a consequence of the interactions of the catalyst
with the reactant atmosphere. This point will be addressed
later when examining in situ DRIFTS results.
The presence of NO in the reactant mixture produced
a shift to higher temperature for the light-off of CO with
O2. The magnitude of this shift is generally larger for the
1% Pd catalyst than for the other two catalysts, the inhibit-
ing effect of NO on CO oxidation being particularly strong
for the 1% Pd catalyst at the lower conversion temperatures
(ꢀT20 = 82, 23, and 46 K, while ꢀT50 = 34, 22, and 25 K,
for the 1, 0.5, and 0.05% Pd catalysts, respectively; Tx refers
to the x% isoconversion temperature). Important differences
between the catalysts were observed comparing the ability
of the two oxidants (O2 and NO) for oxidizing CO in each
case. Analysis of the CO and NO conversion values dur-
ing runs under CO–O2–NO shows that while CO oxidation
with NO is strongly favoured with respect to the CO oxi-
dation with O2 for the 1% Pd catalyst, the opposite occurs
for 0.05% Pd while the 0.5% Pd catalyst represents an inter-
mediate case in which CO oxidation with both NO and O2
takes place almost simultaneously. As a consequence, the
range of activities amongst the catalysts are much greater
for NO reduction than for CO oxidation with O2. In fact,
analysis of the N2O yield profile indicates that the 1% Pd
catalyst is active for NO reduction even from 303 K, in con-
trast with the other two catalysts. For this analysis, it must
be noted that the first two to four NO conversion points
2. Experimental methods
Three Pd/Al2O3 catalysts (with 0.05, 0.5, and 1 wt%
metal loadings) were prepared by incipient wetness impreg-
nation of a γ -Al2O3 support (supplied by Condea, SBET
=
180 m2 g−1) with aqueous solutions of Pd(NO3)2 ·xH2O,
followed by drying overnight at 383 K and calcination under
air at 773 K for 2 h.
Catalytic tests using stoichiometric mixtures of 1% CO +
0.5% O2 or 1% CO + 0.45% O2 + 0.1% NO (N2 balance)
at 3 × 104 h−1 were performed in a Pyrex glass flow reactor
system. Gases were regulated with mass flow controllers and
analysed online using a Perkin–Elmer 1725X FTIR spec-
trometer coupled with a multiple reflection transmission cell
(Infrared Analysis, Inc.). Oxygen concentrations were de-
termined using a paramagnetic analyser (Servomex 540A).
The experimental error in conversion values obtained un-
der these conditions was estimated as 7%. Prior to cat-
alytic testing, in situ calcination under synthetic air at 773 K
was performed, followed by cooling in synthetic air and a
N2 purge at room temperature. A characteristic test con-
sisted of increasing the temperature from 298 to 823 K at
5 K min−1. Catalyst particles in the 0.125–0.250 mm range
were selected after pelleting, grinding and sieving in order
to minimise the pressure drop and internal diffusion effects
during the catalytic tests. The absence of significant external
diffusion effects was also evidenced by the similar activity
profiles obtained during tests at 30,000 h−1 under different
flow conditions.
DRIFTS analysis of adsorbed species on the catalyst un-
der reaction conditions was carried out using a Perkin–Elmer
1750 FTIR spectrometer fitted with an MCT detector. The
DRIFTS cell (Harrick) was fitted with CaF2 windows and a
heating cartridge that allowed samples to be heated to 773 K.
Samples of ca. 80 mg were calcined in situ (in a way sim-
ilar to what was employed for the catalytic tests) and then
cooled to 298 K in synthetic air before the reaction mix-
ture was introduced and heated at 5 K min−1 up to 673 K,

A. Martínez-Arias et al. / Journal of Catalysis 221 (2004) 85–92
87
served for the catalysts. Pd can be assumed to be present
initially as an oxidised phase in all the catalysts, as a re-
sult of the oxidising preconditioning that has been applied
to the samples. This has been verified for samples similar to
those studied here by in situ XANES studies, which showed
the evolution from an inactive PdO-type state to the active
metallic Pd state during the course of an experiment per-
formed under stoichiometric CO–O2–NO under conditions
similar to those employed here [11]. In this respect, for com-
parative purposes, it must be taken into account that the Pd
reduction process taking place during the course of the reac-
tion may depend to some extent on the degree of interaction
between the oxidised Pd species and the alumina support,
i.e., a greater degree of difficulty for reduction of oxidised Pd
is expected as the degree of interaction with alumina (which,
in principle, can be enhanced by decreasing the size of the
oxidised Pd entities) is increased [12]. On the other hand, it
is necessary to consider the important sensitivity of NO ac-
tivation/reduction processes on the structural details of the
metallic Pd particles present under reaction conditions, in
terms of not only particle size but also particle shape [9,13].
Another aspect to consider is the temperature and structure
dependences of adsorption/activation competition processes
between the different reactants [9,14–17]. In order to clarify
these aspects, two of the samples (the 0.5 and 1% Pd sam-
ples) have been examined by in situ DRIFTS (Figs. 2–6).
Unfortunately, in the 0.05% Pd sample the low Pd loading
did not permit a reliable analysis by this technique since the
very weak signals obtained for the species chemisorbed on
Pd were masked by a much more intense CO(g) signal.
Certain differences were noted comparing CO2(g) and
NO(g) evolutions between experiments in Figs. 1 and 3.
These must be mainly attributed to differences in the geom-
etry and/or reactant flow conditions of the catalytic reactors
between both types of experiment. Nevertheless, qualita-
tive correlation between experiments was apparent. The
results obtained during the CO–O2 reaction over both cata-
lysts show the formation of basically two types of carbonyl
species (Figs. 2A and 2B). In the first place is a band
at 2099–2066 cm−1, which shifts to lower wavenumber
and shows a decrease in intensity with increasing reac-
tion temperature; this corresponds to atop carbonyl species
chemisorbed on metallic palladium particles [8,12,14–18].
Second, a band appears at 1980–1968 cm−1, which grows
appreciably with reaction temperature (but slightly de-
creases at 463 K with concomitant downward shift for the
1% Pd sample) and whose frequency shows a maximum at
intermediate reaction temperatures. This can be attributed to
carbonyl species chemisorbed on bridging sites of metal-
lic palladium particles [8,11,14–18]. Roughly in parallel
with the latter, a third carbonyl species appeares as a low-
frequency shoulder extending to 1850–1800 cm−1. Its posi-
tion is not well defined due to its relatively large linewidth
and its overlap with the 1980–1968 cm−1 bridged carbonyl
band. This shoulder can be attributed to carbonyl species
chemisorbed on threefold hollow sites of the metallic pal-
Fig. 1. Main results obtained during light-off tests under stoichiometric
CO–O (top) and CO–O –NO (middle and bottom) for the 1% Pd/Al O
2 3
2
2
(squares), 0.5% Pd/Al O (rhombi), and 0.05% Pd/Al O (circles) cata-
2
3
2 3
lysts. In the middle, full and open symbols correspond to NO and CO
conversion, respectively.
(below ca. 330 K) present a certain contribution correspond-
ing to NO adsorption processes and not to NO reduction
processes, as inferred from mass balance analysis. Concern-
ing N2 selectivity during the NO reduction process, analysis
of the N2O yield shows that N2O formation increases up to
a maximum (lower for the 1% Pd catalyst), roughly follow-
ing the initial NO conversion profiles, and then decreases,
the catalysts becoming almost fully selective to N2 at high
temperatures. This indicates that the N2O selectivity grad-
ually decreases with the reaction temperature with the cat-
alysts showing almost full N2O selectivity for the reduction
process during the first low-temperature NO reduction stage.
Different factors can be considered as playing a role in
the different CO oxidation and NO reduction behaviours ob-

88
A. Martínez-Arias et al. / Journal of Catalysis 221 (2004) 85–92
Fig. 2. DRIFTS spectra taken at the indicated temperatures under stoichiometric CO–O (A, B) and CO–O –NO (C, D) for the 0.5% Pd/Al O (A, C) and 1%
2
2
2 3
Pd/Al O (B, D) samples. A reference spectrum containing exclusively the CO(g) bands has been employed to subtract the corresponding contribution from
2
3
each spectrum in order to isolate contributions from chemisorbed species.

A. Martínez-Arias et al. / Journal of Catalysis 221 (2004) 85–92
89
Fig. 3. Results of online analysis of DRIFTS experiments showing the evolution of integrated contribution of CO (g) bands observed by DRIFTS and the NO
2
x
signal of the chemiluminescence detector obtained over the course of the indicated experiments.
ladium particles [18,19]. Both catalysts show an increase
in bridged/atop carbonyl intensity ratio with the reaction
temperature. This evolution can be attributed, in accordance
with previous studies [20,21], to the fact that the relative
size of the metallic palladium particles increases with the
reaction temperature during the course of the PdO → Pd re-
duction process that occurs (in agreement with the general
increasing intensity of metallic palladium carbonyls and as
confirmed by in situ XANES experiments performed over
catalysts similar to the ones studied here, as found else-
where [11]) upon interaction with the reactant mixture. This
hypothesis obviously assumes that low-temperature O2 ad-
sorption, which most likely would preferentially involve
covering of bridging sites [22], must be strongly inhibited
under the CO-rich conditions employed [23]. The main dif-
ference between both catalysts in the process of genesis
of the metallic Pd particles concerns a somewhat greater
difficulty in generation of bridging sites in the 0.5% Pd sam-
ple (compare Figs. 2A and 2B). This can be related to the
lower metallic Pd particle size achieved for the sample with
lower Pd loading during the process of nucleation/growth of
metallic particles. This becomes more apparent at the lower
reaction temperatures for which only incipient germs of the
metallic Pd particles have apparently developed.
Fig. 4. Wavenumber shift of the atop carbonyl adsorbed on metallic Pd parti-
cles and CO (g) evolution as a function of the reaction temperature during
2
the light-off runs under CO–O in the DRIFTS cell for 0.5% Pd/Al O
2
2 3
(rhombi) and 1% Pd/Al O (squares).
2
3
to coupling [24]), which increases with coverage. The ob-
served evolution indicates, as widely demonstrated [19,25],
that the reaction is limited by thermal desorption of CO
from the metallic Pd particles within the typical Langmuir–
Hinshelwood kinetic scheme. This also agrees with the rela-
tively sharp termination (at values approaching 100% con-
version) of the light-off profiles for both catalysts (1 and
0.5%, Fig. 1), which is typical of negative order kinetics
for the relevant reactant [26] (CO in this case), as expected
when the catalyst operates under the aforementioned limit-
A fair degree of correlation was observed (Fig. 4) be-
tween the frequency decrease of the atop carbonyl band
and the CO oxidation rate taking into account that the fre-
quency position of the palladium carbonyls is mainly deter-
mined (for each type of carbonyl) by dipolar coupling effects
(those correspondingto atop species being the most sensitive

90
A. Martínez-Arias et al. / Journal of Catalysis 221 (2004) 85–92
ing step [19]. It somewhat contrasts with the approach to a
sigmoidal-type termination of the profile for the 0.05% Pd
catalyst (Fig. 1), which suggests [26] that the catalyst may
be operating under a different kinetic regime in which the
reaction is essentially limited by the rate of CO adsorption
on the Pd particles (i.e., it becomes positive order with re-
spect to CO), as expected at the relatively high conversion
temperatures at which it is active [19]. Some limitation for
Pd reduction may also appear in this sample as a conse-
quence of PdO-support interaction [12], which can also limit
to some extent the onset of CO oxidation. On the other hand,
the similarities in the shapes of the CO conversion or CO2
production profiles between the 0.5 and the 1% catalysts
(Figs. 1 and 3) suggest that the main differences between
them are essentially related to the different numbers of ac-
tive centres present in each case. In this respect, it must be
considered that the nucleation/growth process of formation
of active metallic particles is practically the same for both
catalysts (according to DRIFTS results in Figs. 2A and 2B)
and that the small structural differences between the parti-
cles formed for each catalyst (as discussed above) are not
reflected in strong kinetic changes for the CO–O2 reaction,
according to the essentially structurally insensitive character
of this reaction over Pd [19]. Nevertheless, the increase in
the number of active sites occurring as a consequence of the
progressive growth of metallic particles during the course of
the reaction for any of the catalysts may well be responsible
for the irregularities (giving rise to the staircase-type shape
mentioned) observed in the activity profiles at lower temper-
ature.
Important differences are observed comparing DRIFTS
spectra in the absence and in the presence of a small amount
of NO in the reactant mixture (Fig. 2). Such differences are
more acute for the 1% Pd sample, which is in agreement
with the greater effect of NO on the CO oxidation catalytic
properties of this system. Thus, an important suppression on
the formation of metallic Pd carbonyls is observed for the
latter catalyst, most particularly at lower reaction tempera-
tures (compare Figs. 2B and 2D), while new features at 2143
and ca. 2125 cm−1 (sh) are observed at T <∼ 423 K and
show decreasing intensity with increasing reaction tempera-
ture. These bands can be attributed to atop carbonyl species
chemisorbed at oxidised Pd sites [8,19]. Bands due to metal-
lic Pd carbonyl species (atop carbonyls at 2098–2044 cm−1
and bridged carbonyls at 1970–1920 cm−1) begin to ap-
pear only at T ꢀ 333 K for this sample (Fig. 2D). Although
both bands generally show lower intensity in the presence
of NO, it is worth noting that bridged carbonyls are com-
paratively more affected by this adsorbate. The evolution
of the frequency shifts observed for these carbonyls by in-
creasing the reaction temperature correlates well with the
catalytic activity for CO oxidation, in a manner similar to
that shown above. Additionally, two bands at 2232 and ca.
2250 cm−1, corresponding to NCO species chemisorbed on
alumina [8,17], were observed at T ꢀ 393 K and showed
maximum intensity in spectra of the sample at 498 K. These
isocyanate species are proposed to be formed following NO
dissociation on the metallic palladium particles and subse-
quent N–CO reaction and spillover onto the alumina support
where they accumulate [8,17,27,28]. It can be noted that the
onset temperature for their formation correlates well with
the increase in the NO reduction catalytic activity, as also
noted in previous studies [8,11]. According to a recent re-
port [28], the isocyanate species formed in this kind of cat-
alyst are relatively little reactive toward NO + O2 mixtures
and can thus be considered spectators under the conditions
employed here; nevertheless, they can present an interesting
practical role as intermediates in the formation of ammonia
upon interaction with components of the automobile exhaust
gases [28].
As mentioned above, significant differences are observed
between the 0.5 and the 1% Pd samples with respect to the
formation/evolution of the different Pd carbonyl species in
the presence of NO (Figs. 2C and 2D). Thus, the extent to
which NO hinders the formation of metallic Pd carbonyls at
low reaction temperatures is considerably less for the 0.5%
Pd sample. Common to both samples, the presence of NO
influences the formation of bridging species to a greater
extent than the atop carbonyls. However, for the 0.5% Pd
sample, atop species adsorbed on metallic Pd particles are
clearly formed even after the initial contact at low temper-
ature with the reactant mixture, while the presence of atop
species adsorbed on oxidised Pd entities appears negligible.
It is interesting to note that the onset of NO reduction for
both samples (at ca. 393 and 423 K for the 1 and 0.5% sam-
ples, respectively), in accordance with the initial detection
of chemisorbed isocyanate species (Figs. 2C and 2D) and in
reasonable agreement with catalytic activity results (Fig. 1)
and online gas analysis (Fig. 3), coincides with a sharp in-
crease in the population of bridged carbonyl species.
The increase in NO reduction activity with Pd loading is
in line with literature results showing that this activity in-
creases with the relative size of the metallic Pd particles, at
least up to a certain limiting value [9,13]. In our particular
case, the reasons behind this behaviour can be related, on the
basis of DRIFTS results, to the particular CO–NO competi-
tion for atop–bridging sites on the active metallic palladium
particles and to the NO dissociation capability as a function
of the size/structural details of the particles formed during
the course of the light-off run. The main difference between
the catalysts is related to the presence of oxidised Pd car-
bonyls (bands at 2143–2125 cm−1) for the 1% Pd sample
at low reaction temperatures. The formation of these car-
bonyls may result from dissociative NO adsorption on the
Pd particles, which leads to the formation of the oxidised Pd
(Pdn+) chemisorption sites. This is consistent with the cat-
alytic activity results (Fig. 1), which show that the 1% Pd
sample is able to dissociate NO even at 303 K. Formation
of the Pdn+ sites can be associated with the presence of NO
in the reactant stream on the basis of comparison between
DRIFTS experiments conducted in the presence and absence
of this oxidant (Fig. 2). These oxidised Pd species can be

A. Martínez-Arias et al. / Journal of Catalysis 221 (2004) 85–92
91
Fig. 5. Postreaction DRIFTS experiments at 303 K for (a, b) 0.5% Pd/Al O
2
3
and (c, d) 1% Pd/Al O . Following the run under CO–O up to 483 K, the
2
3
2
sample was cooled under N and then 1% CO was flowed through the cell
2
(a, c) (dotted line) and subsequently flushed with N . (b, d) Same as the
2
former, but after the CO–O –NO reaction up to 573 K.
2
Fig. 6. DRIFTS spectra under CO–O –NO at the indicated temperatures in
2
the region relevant to adsorbed nitrosyl species for 0.5% Pd/Al O (top)
2
3
formed during the whole run, even if only in small amounts,
as demonstrated by postreaction experiments (Fig. 5). These
data again support the proposal that oxidised Pd sites are
linked with the presence of NO in the reactant stream. The
stabilization of these Pdn+-bonded carbonyls in significant
quantities only at lower reaction temperatures can be re-
lated to their lower thermal stability with respect to carbonyl
species adsorbed on metallic particles, as evidenced by the
respective behaviour toward flushing with N2 at room tem-
perature (Fig. 5). The combined results suggest a greater NO
dissociation efficiency on the particles formed for the 1% Pd
sample. This can be used to explain the greater activity of
such a catalyst, taking into account that the NO dissocia-
tion process has been proposed as rate limiting, at least for
relatively low temperatures and within a range of relatively
small particle sizes [13,29].
and 1% Pd/Al O (bottom).
2
3
the relatively low PNO/PCO employed for the catalytic ex-
periments as well as the presence of overlapping bands due
to species chemisorbed on the support. Thus, the spectra of
both samples (Fig. 6) show bands due to bicarbonate species
adsorbed on the alumina (at 1654 and 1439 cm−1) [30] and
of species giving a band at 1543 cm−1 that grew with the
reaction temperature and can be assigned to nitrate species
chemisorbed on alumina [30]. A more careful analysis of
this region reveals the presence of bands at 1648–1640
(showing a slight high-temperature red shift) and 1612 cm−1
that may be attributed to nitrosyl species chemisorbed on
bridging sites of the metallic Pd particles [14–16]. No hint
of formation of atop nitrosyl species (expected to give a
band at 1750–1730 cm−1 [15–17]) was observed in these
spectra, which suggests that when both CO and NO are
present under the employed conditions, the latter molecule
interacts preferentially with bridging sites while CO forms
mainly atop species through either exchange or NO-induced
site transfer (bridging → atop [14,17]) processes. Compar-
By considering the influence of NO on the relative
amounts of atop–bridging adsorption sites, indicated by the
respective carbonyl intensities, one would expect a relatively
stronger interaction of NO with bridging sites and with the
1% Pd catalyst. The analysis of the region relevant to ad-
sorbed nitrosyl species is complicated as a consequence of

92
A. Martínez-Arias et al. / Journal of Catalysis 221 (2004) 85–92
ison of bridging nitrosyl species for both catalysts reveals
that, surprisingly, the 0.5% Pd catalyst appears to sustain a
greater amount of such species at low temperature (consider-
ing the grea−te1r extinction coefficient of the bicarbonate band
at 1439 cm with respect to that of the overlapping band
at 1654 cm−1 for this analysis [30]). NO reverse spillover
effects, which would be favoured when a higher amount of
support (i.e., for lower metal loading) is exposed, may also
play a role in this observation [13]. Nevertheless, the ob-
served results are compatible with the higher NO reduction
activity of the 1% Pd catalyst and apparently reflect the more
efficient NO dissociation over this catalyst within the bal-
ance of NO adsorption/dissociation processes taking place
on the surface of the catalysts.
(A.B.H.) for financial help under the “Ayudas para estancias
breves en centros de investigación extranjeros” program.
Thanks are due to Dr. R. Cataluña for the preparation of
some of the catalysts. Financial help by project CICyT (Ref.
MAT 2000-1467) is also acknowledged.
References
[1] E.S.J. Lox, B.H. Engler, in: G. Ertl, H. Knözinger, J. Weitkamp (Eds.),
Environmental Catalysis, Wiley–VCH, New York, 1999, p. 1.
[2] R.M. Heck, R.J. Farrauto, Appl. Catal. A 221 (2001) 443.
[3] A. Martínez-Arias, M. Fernández-García, A.B. Hungría, A. Iglesias-
Juez, K. Duncan, R. Smith, J.A. Anderson, J.C. Conesa, J. Soria,
J. Catal. 204 (2001) 238.
[4] Z. Hu, C.Z. Wan, Y.K. Lui, J. Dettling, J.J. Steger, Catal. Today 30
(1996) 83.
4. Conclusions
[5] J.R. González-Velasco, J.A. Botas, R. Ferret, M.P. González-Marcos,
J.-L. Marc, M.A. Gutiérrez-Ortiz, Catal. Today 59 (2000) 395.
[6] S. Tagliaferri, R.A. Köppel, A. Baiker, Appl. Catal. B 15 (1998) 159.
[7] J.H. Holles, M.A. Switzer, R.J. Davis, J. Catal. 190 (2000) 247.
[8] A. Martínez-Arias, M. Fernández-García, A. Iglesias-Juez, A.B. Hun-
gría, J.A. Anderson, J.C. Conesa, J. Soria, Appl. Catal. B 31 (2001)
51.
A combined catalytic activity in situ DRIFTS study
of stoichiometric CO–O2 and CO–O2–NO reactions over
PdO/γ -Al2O3 catalysts with different Pd loadings has been
carried out. Results for the CO–O2 reaction are in line with
many studies in the literature showing that the catalysts fol-
low typical Langmuir–Hinshelwood kinetics in which the
reaction is initially rate limited by CO desorption from the
metallic particles except for a sample with a very small Pd
loading for which the reaction onset is shifted to signifi-
cantly higher temperatures, in which case CO adsorption
may become the rate-determining step for the reaction. The
light-off profiles show certain irregularities at low reaction
temperatures which are attributed to the fact that the num-
ber of active sites undergoes a change during the course of
the reaction as a consequence of the process of formation
of metallic Pd particles. A comparison between the catalyst
activities for CO oxidation and NO reduction under stoi-
chiometric CO–O2–NO reveals significant differences for
NO reduction processes. These are explained on the basis of
different structural characteristics of the Pd particles formed
in each case following interaction with the reactant mix-
ture. In this respect, it is shown that NO dissociation (which
may be the rate-determining step under the employed con-
ditions) is most favoured over the larger Pd particles formed
for the highest loaded sample, as deduced from catalytic ac-
tivity results and DRIFTS analysis of the nature (redox state
and coordination number of the adsorption site) and the rela-
tive intensities of atop–bridging carbonyl or nitrosyl species
formed for each catalyst.
[9] D.R. Rainer, S.M. Vesecky, M. Koranne, W.S. Oh, D.W. Goodman,
J. Catal. 167 (1997) 234.
[10] N. Matthess, D. Schweich, B. Martin, F. Castagna, Top. Catal. 16/17
(2001) 119.
[11] M. Fernández-García, A. Martínez-Arias, A. Iglesias-Juez, A.B. Hun-
gría, J.A. Anderson, J.C. Conesa, J. Soria, J. Catal. 214 (2003) 220.
[12] K. Otto, L.P. Haack, J.E. deVries, Appl. Catal. B 1 (1992) 1.
[13] L. Piccolo, C.R. Henry, J. Mol. Catal. A 167 (2001) 181.
[14] R. Raval, G. Blyholder, S. Haq, D.A. King, J. Phys. Condens. Matter 1
(1989) SB165.
[15] X. Xu, P. Chen, D.W. Goodman, J. Phys. Chem. 98 (1994) 9242.
[16] S.M. Vesecky, P. Chen, X. Xu, D.W. Goodman, J. Vac. Sci. Technol.
A 13 (1995) 1539.
[17] K. Almusaiteer, S.C. Chuang, J. Catal. 184 (1999) 189.
[18] I.V. Yudanov, R. Sahnoun, K.M. Neyman, N. Rösch, J. Hoffmann, S.
Schauermann, V. Johánek, H. Unterhalt, G. Rupprechter, J. Libuda,
H.-J. Freund, J. Phys. Chem. B 107 (2003) 255, and references therein.
[19] X. Xu, D.W. Goodman, J. Phys. Chem. 97 (1993) 7711.
[20] K. Wolter, O. Seiferth, H. Kuhlenbeck, M. Bäumer, H.-J. Freund, Surf.
Sci. 399 (1998) 190.
[21] M. Frank, M. Bäumer, Phys. Chem. Chem. Phys. 2 (2000) 3723.
[22] T. Engel, G. Ertl, Adv. Catal. 28 (1979) 2.
[23] I. Meusel, J. Hoffmann, J. Hartmann, J. Libuda, H.-J. Freund, J. Phys.
Chem. B 105 (2001) 3567.
[24] P. Hollins, Surf. Sci. Rep. 16 (1992) 51.
[25] J. Libuda, I. Meusel, J. Hoffmann, J. Hartmann, L. Piccolo, C.R.
Henry, H.-J. Freund, J. Chem. Phys. 114 (2001) 4669.
[26] F. Duprat, Chem. Eng. Sci. 57 (2002) 901.
[27] F. Solymosi, J. Rasko, J. Catal. 63 (1980) 217.
Acknowledgments
[28] N. Macleod, R.M. Lambert, Chem. Commun. (2003) 1300.
[29] B. Hammer, J. Catal. 199 (2001) 171.
A.B.H. and A.I.-J. thank the Comunidad de Madrid for
grants under which this work has been carried out and
[30] A. Martínez-Arias, M. Fernández-García, A. Iglesias-Juez, J.A. An-
derson, J.C. Conesa, J. Soria, Appl. Catal. B 28 (2000) 29.