Redox interplay at copper oxide- (Ce,Zr)Ox interfaces: Influence of the presence of NO on the catalytic activity for CO oxidation over CuO/CeZrO4

Article abstract of DOI:10.1016/S0021-9517(02)00084-2

A catalyst composed of copper oxide supported on cerium-zirconium mixed oxide (CuO_x/CeZrO₄) has been studied with respect to its activity for CO oxidation under stoichiometric conditions employing either oxygen or oxygen with a small amount of NO as oxidant. The nature of copper oxide entities, as well as the redox properties of the catalyst following interactions with CO and O₂-NO, has been studied by XPS, EPR, and static NO adsorption infrared spectroscopy while in situ DRIFTS has been employed to follow processes occurring at the catalyst surface under reaction conditions. Characterization of the copper oxide species in both fully oxidized and partially reduced states reveals that they are significantly affected by interactions with the underlying support. On the basis of catalytic activity results and in combination with analysis of the evolution of particular Cu ⁺ carbonyls, comparing the CO oxidation reaction in the presence and absence of NO, it is proposed that the two basic factors affecting the catalytic performance of this type of system are the facility for achieving a partially reduced state for the copper oxide phase at the interfacial zone and the redox properties of the CuO_x/CeZrO₄ interface.

Full text of DOI:10.1016/S0021-9517(02)00084-2

Journal of Catalysis 214 (2003) 261–272
www.elsevier.com/locate/jcat
Redox interplay at copper oxide-(Ce, Zr)O interfaces:
x
influence of the presence of NO on the catalytic activity
for CO oxidation over CuO/CeZrO4
a,∗
a
a
a
a,1
A. Martínez-Arias, M. Fernández-García, A.B. Hungría, A. Iglesias-Juez, O. Gálvez,
b
a
a
c
J.A. Anderson, J.C. Conesa, J. Soria, and G. Munuera
a
Instituto de Catálisis y Petroleoquímica, CSIC, C/ Marie Curie, Campus Cantoblanco, 28049 Madrid, Spain
b
Division of Physical and Inorganic Chemistry, University of Dundee, Dundee, DD14HN Scotland, UK
c
Departamento de Química Inorgánica, Universidad de Sevilla, 41092 Sevilla, Spain
Received 6 August 2002; revised 21 October 2002; accepted 24 October 2002
Abstract
A catalyst composed of copper oxide supported on cerium–zirconium mixed oxide (CuOx/CeZrO ) has been studied with respect to its
4
activity for CO oxidation under stoichiometric conditions employing either oxygen or oxygen with a small amount of NO as oxidant. The
nature of copper oxide entities, as well as the redox properties of the catalyst following interactions with CO and O –NO, has been studied
2
by XPS, EPR, and static NO adsorption infrared spectroscopy while in situ DRIFTS has been employed to follow processes occurring at
the catalyst surface under reaction conditions. Characterization of the copper oxide species in both fully oxidized and partially reduced
states reveals that they are significantly affected by interactions with the underlying support. On the basis of catalytic activity results and in
+
combination with analysis of the evolution of particular Cu carbonyls, comparing the CO oxidation reaction in the presence and absence of
NO, it is proposed that the two basic factors affecting the catalytic performance of this type of system are the facility for achieving a partially
reduced state for the copper oxide phase at the interfacial zone and the redox properties of the CuOx/CeZrO interface.
4

2003 Elsevier Science (USA). All rights reserved.
2+
Keywords: Copper oxide; Ce–Zr mixed oxide; Ceria-based catalysts; CO oxidation; Nitric oxide; XPS; Cu EPR; NO adsorption infrared; In situ DRIFTS
1
. Introduction
cluding structural, redox states, and bifunctional promotions
have been proposed to account for the synergetic effects ob-
served [1].
Catalysts based on ceria or cerium–zirconium mixed ox-
One of the reactions in which the promoting effect is most
significant is carbon monoxide oxidation. For this reaction,
it is interesting that ceria-related oxides promote the activity
of both noble metals (for instance, Rh, Pt, and/or Pd, typi-
cally present in TWC) and nonnoble metals or metal oxides
[1,2,5–16]. Among the latter, outstanding activities, compa-
rable to those exhibited by noble metal catalysts, are shown
by copper catalysts [5–17]. Different hypotheses have been
made to account for the synergetic effects on CO oxidation
observed upon establishment of interactions between copper
oxide and ceria–related oxides [5–17]. Concerning the na-
ture of the active copper species, it is generally accepted that
these are related to well-dispersed states of copper (having
a small nuclearity and probably in the form of copper oxide
clusters [16]) in contact with the ceria-based support [5–7,
ides present a wide range of applications, such as three-way
catalysts for automobile exhaust gas emission control, re-
moval of SOx–NOx from fluid catalytic cracking flue gases,
electrocatalysts over fuel cell electrodes, and catalysts for
various oxidation and hydrogenation reactions [1–4]. For
most of these applications, ceria–related compounds are
thought to operate mainly as promoters of the active metal
(
or metal oxides) with which they are in contact [1]. De-
pending on the type of reaction and the experimental condi-
tions involved in each case, different promoting effects in-
*
Corresponding author.
E-mail address: amartinez@icp.csic.es (A. Martínez-Arias).
Present address: Departamento de Química-Física, Facultad de
1
Química, Universidad Complutense, 28040 Madrid, Spain.
0021-9517/03/$ – see front matter  2003 Elsevier Science (USA). All rights reserved.
doi:10.1016/S0021-9517(02)00084-2

262
A. Martínez-Arias et al. / Journal of Catalysis 214 (2003) 261–272
9
,11,14–17]. Correlation between the reducibility of highly
phases prepared by dissolving similar molar amounts of
cerium (III) nitrate hexahydrate and zirconyl nitrate for the
first and tetramethylammonium hydroxide pentahydrate for
the second, dispersed in an organic solvent (n-heptane),
using Triton X-100 (Aldrich) as surfactant and hexanol
as co-surfactant. Following centrifugation, decanting, and
rinsing of the resulting solid with methanol, it was dried
at 383 K for 24 h and finally calcined in air at 773 K for
2 h. Further details on the preparation and characteristics
of this support can be found elsewhere [27]. Thorough
precipitation of both cerium and zirconium components was
achieved for this support (on the basis of ICP-AES analysis
of the final product); it consisted of ca. 5-nm crystallites
(based on TEM results; XRD estimation of the particle
size showed good agreement with the TEM observation)
presenting the fluorite-related pseudocubic phase t (on the
basis of XRD and Raman analyses) and showing reasonably
good compositional homogeneity at the atomic scale [27].
A sample of copper catalyst supported on the CeZrO4
mixed oxide (referred to hereafter as CuZC) was prepared
by incipient wetness impregnation of the above support
using an aqueous solution of Cu(NO3)2 · 3H2O (to give a
final copper loading of 1 wt%, representing ca. 157 µmol
of Cu per gram of catalyst). The resulting material was
dried overnight at 383 K and subsequently calcined in air
at 773 K for 2 h. XRD and Raman analyses of the CuZC
sample showed only features similar to those observed for
the copper-free support.
dispersed copper oxide species and their catalytic activity
appears to be well established [9,11,14–20]. Recent results
have shown that these sites display a facile redox interplay
with the reactants, suggesting that the reaction follows a re-
dox mechanism [16]. Furthermore, the redox states of both
catalyst components (copper and ceria-related oxides) ap-
pear to undergo concomitant changes, strongly suggesting
that the active sites are located at the interfacial region [16],
in agreement with previous postulations [5,7,14,17,19]. Dif-
ferences between the activities of copper oxide catalysts sup-
ported on different ceria-based oxides have been related to
the different redox activities of the O2 /V (where V de-
−
Ö
Ö
notes a doubly ionized oxygen vacancy, using the Kröger–
Vink notation) interface sites in each case [16]. A somewhat
different mechanistic proposal has been recently put forward
by Wang et al. [21]. In their approach, the reaction is pro-
posed to take place in two general sets of processes covering
an induction and a light-off period. Although a role for inter-
face vacancies at the support is recognized, this role appears
to be mainly restricted to facilitating oxygen activation dur-
ing the induction steps prior to light-off, while sites respon-
sible for light-off behavior are proposed to be related to oxy-
gen vacancies at the copper oxide [21]. On the other hand,
an important role of the surface structure of the ceria-related
oxide in the copper oxide activity for the reaction has been
proposed by Skårman et al. in a report showing significant
differences between the activities of copper oxide thin films
supported on CeO2{111} and CeO2{001} surfaces [22].
The present work aims to provide further insight into
processes involved in the mechanism of CO oxidation in
this kind of system by extending previous results obtained
ꢀꢀ
All the gases employed were of commercial purity (CO:
99.9%; O2: 99.99%; NO: 98.5%, with 1% NO2 and 0.5%
N2O as the main impurities) and, for experiments using the
vacuum line (static IR and EPR experiments), were further
purified by vacuum distillation before storage.
for the same reaction over a CuO/CeZrO catalyst [16]. In
4
this case, a small amount of NO (within a stoichiometric
CO–O2–NO mixture) has been included in the reactant
stream in order to introduce a certain distortion into the
CO oxidation processes. It should be noted that NO is
able to compete with CO for oxidized copper sites and
with oxygen for interfacial vacancies [23–25] and could
therefore potentially affect the redox properties of both
catalyst components and, as a consequence, the overall
catalytic activity of the system. Preliminary catalytic activity
results, for bimetallic Pd–Cu catalysts, have shown that
NO has little influence on CO oxidation activity over the
2.2. Techniques
Light-off catalytic activity tests were carried out using
a glass flow reactor system loaded with ca. 3 g of sample.
Catalyst particles in the range 0.125–0.250 mm were se-
lected after pelleting, grinding, and sieving in order to min-
imize the pressure drop and internal diffusion effects, while
the reactor geometry was optimized to avoid significant ex-
ternal diffusion effects [28]. Analysis of the feed and out-
let gas streams was performed using a Perkin–Elmer FTIR
spectrometer model 1725X, coupled to a multiple reflection
transmission cell (Infrared Analysis Inc. long path gas mini-
CuO/CeZrO catalyst [26]. A more detailed study of the
4
reasons underlying this behavior is presented here.
3
cell, path length 2.4 m, internal volume ca. 130 cm ), the
O2 concentration in the gas stream was determined with a
paramagnetic analyzer (Servomex 540 A). Prior to catalytic
testing, the catalysts were subjected to a standard calcina-
tion pretreatment using 3% O2/N2 flow at 773 K for 1 h,
cooled to room temperature (RT) in the same flow, and fi-
nally briefly purged in N2. The catalytic tests were per-
formed under stoichiometric conditions in 1% CO and 0.5%
O2 or 1% CO, 0.45% O2, and 0.1% NO, using N2 as car-
rier gas at atmospheric pressure, a fixed space velocity of
2
. Experimental
2
.1. Materials
2
−1
A high-surface-area (SBET = 96 m g ) Zr–Ce (atomic
ratio 1 : 1) mixed oxide was used as the support. It was
prepared by a microemulsion–coprecipitation method by
mixing two reverse microemulsions containing aqueous

A. Martínez-Arias et al. / Journal of Catalysis 214 (2003) 261–272
263
4
−1
−3
3
× 10 h (roughly corresponding to 0.18 gscm ), and
and introduced into the pretreatment chamber where it was
−1
temperature ramps of 5 Kmin ; it must be noted that analy-
sis of the bypass flow shows that around 45 ppm of NO2 is
present at RT in the latter gas mixture under these exper-
imental conditions, due to partial homogeneous NO/NO2
equilibration. In all cases, the runs were commenced after
an initial equilibration period of ca. 5 min in the reactant
flow at RT.
Diffuse reflectance infrared Fourier transform spectra
DRIFTS) under reaction conditions were recorded using
a Perkin–Elmer 1750 FTIR spectrometer equipped with an
MCT detector. The spectra were taken at resolution 4 cm
and 10 scans were accumulated for every spectrum. The
DRIFTS cell (Harrick) was fitted with CaF2 windows and
a heating cartridge which allowed samples to be heated to
73 K. Samples of ca. 80 mg were calcined in situ (in a
manner similar to that of the catalytic activity tests) and
then cooled to RT before introducing the reaction mixture
outgassed at 473 K for 2–3 h, until a pressure of less than 2×
−8
10 Torr was achieved, the sample then being reoxidized
under 1 Torr of O at the same temperature. Subsequently, it
2
was reduced under 1 Torr CO at 373 K and then contacted
−4
−2
with increasing doses of 10 , 10 , and 1 Torr of either
O2 or NO. Following each treatment, the sample was moved
into the ion-pumped analysis chamber, where it was further
−9
outgassed until a pressure less than 2 × 10
Torr was
(
attained (2–3 h). This low pressure was maintained during
all the data acquisition by ion pumping of the chamber. After
each treatment, XP spectra in the relevant energy windows
were collected for 20 to 90 min, depending on the peak
−1
intensities, at a pass energy of 44 eV (1 eV = 1.602 ×
−19
J), which is typical of high-resolution conditions.
1
0
7
The intensities were estimated by calculating the integral
of each peak after subtraction of an S-shaped Shirley-
type background with the help of UNIFIT for Windows
(
passing through the catalyst bed) consisting of 1% CO
and 0.5% O2 or 1% CO, 0.45% O2, and 0.1% NO in N2,
(
Version 3.2) software [29]; atomic ratios were then derived
using the appropriate experimental sensitivity factors. All
binding energies (BE) were referenced to the adventitious
C 1s line at 284.6 eV. This reference gave BE values with an
3
−1
and using a total flow of ca. 80 cm min and ramps of
−
1
5
Kmin . A PC-controlled gas blender was used to monitor
the composition of the inlet gases.
Transmission infrared spectra under static conditions
were recorded at RT with a Nicolet 5ZDX FTIR spectrom-
eter with a resolution of 4 cm , 128 scans per spectrum,
subtracting the gas phase contribution in all cases. Thin
self-supporting disks (ca. 20 mg cm ) were prepared by
pressing the powders at 2500 Ncm and were handled in
standard greaseless cells, where they could be subjected to
thermal or adsorption treatments. The reduction treatments
under CO were carried out under conditions similar to those
indicated below for the EPR experiments.
ꢀ
ꢀꢀ
4+
accuracy of ± 0.1 eV; the peak u characteristic of Ce was
thus obtained at 917.0± 0.1 eV. In the case of Ce 3d spectra,
3
+
4+
−1
factor analysis (FA) was used to calculate the Ce /Ce
ratios in each set of spectra recorded, using the methodology
developed in a previous work [30].
−
2
−
2
3
. Results
Electron paramagnetic resonance (EPR) spectra were
recorded at 77 K with a Bruker ER 200 D spectrometer
operating in the X-band and calibrated with a DPPH stan-
dard (g = 2.0036). Portions of sample of about 30 mg were
placed inside a quartz probe cell with greaseless stopcocks
using a conventional high-vacuum line (capable of maintain-
3.1. Catalytic activity tests
The results obtained for the CO–O2 and CO–O2–NO
reactions are shown in Fig. 1. Mass balance analysis of
the evolution of the different gas reactants and products
present during the CO–O2–NO reaction indicates that the
NO conversion profile is affected by adsorption–desorption
effects for T ꢀ 450 K; therefore, only conversion points
above that temperature can be considered as representative
of NO reduction with CO. Analysis of the various reactivity
profiles shows that decomposition of adsorbed NOx species
−3
−2
ing a dynamic vacuum of ca. 6 ×10 Nm ) for the differ-
ent treatments. In all cases, the sample was pretreated in 300
−2
Torr (1 Torr = 133 Nm ) of pure oxygen at 773 K for 2 h.
CO reduction treatments at a specific reduction temperature
(
Tr) were made under static conditions using 100 Torr of CO,
heating for 1 h at the corresponding Tr and subsequently out-
gassing at the same temperature for 0.5 h. Quantitative eval-
uation of the amount of species present in the spectra was
performed by double integration of the corresponding EPR
spectra and comparison with a copper sulfate standard.
X-ray photoelectron spectra (XPS) were recorded with
a Leybold–Heraeus spectrometer equipped with an EA-200
hemispherical electron multichannel analyzer (from Specs)
and a 120-W, 30-mA Mg-Kα X-ray source. A PC was
used to control the instrument and recording of the spectra.
(
below ca. 430 K) produces mainly NO and concomitantly
a very small amount of N2O, while NO reduction with CO
taking place at T ꢁ 450 K) is essentially 100% selective
(
to N2. Interestingly, CO oxidation with O2 (responsible for
CO conversion at T ꢀ 450 K) is very slightly affected by the
presence of NO in the reactant mixture. This contrasts with
the behavior observed for noble metal catalysts promoted by
ceria-related oxides for which, under conditions similar to
those employed here, the activity of CO oxidation with O2
is decreased by the presence of NO in the reactant mixture
[26,31].
The CuZC sample (0.2 g) was lightly pressed into a small
2
(
4 × 4 mm ) pellet and then mounted on a sample rod

264
A. Martínez-Arias et al. / Journal of Catalysis 214 (2003) 261–272
3
.2. Infrared results
DRIFT spectra recorded at increasing temperatures under
CO–O2–NO reaction conditions are shown in Fig. 2. A rea-
sonable correlation is apparent between the evolution of gas
phase CO and CO2 (appearing as relatively broad bands at
−1
2
200–2050 and 2400–2300 cm , respectively) and the cat-
alytic activity results (Fig. 1). Unfortunately, the smaller ex-
tinction coefficient and concentration of NO(g) prevent its
detection in the spectra under the conditions employed. Only
−1
one band in the range 2111–2105 cm appears under the
reaction conditions in the carbonyl-stretching region of the
spectra. This band is characteristic of carbonyl species ad-
sorbed on copper sites in contact with ceria-related oxides
and, although its frequency would appear somewhat low for
+
a cationic copper site, it has been attributed to Cu carbonyls
on the basis of its relatively high thermal stability during out-
gassing [16,18,19]. The evolution of its intensity as a func-
tion of the reaction temperature in the presence and absence
of NO in the reactant mixture is displayed in Fig. 3. Notably,
significant differences are observed between the two exper-
iments in the evolution of the band intensity, in spite of the
close similarities observed for the CO–O2 activity profiles
(
Fig. 1), and as confirmed by the simultaneous monitoring
of the CO2 gas phase bands during the in situ DRIFTS ex-
periments (Fig. 3). A better degree of correlation is observed
+
when catalytic activity and the νCO of the Cu carbonyl
are compared as functions of increasing reaction tempera-
ture (Fig. 3, bottom). In this respect, it is significant that the
profiles in the presence of NO show a rough degree of cor-
relation between the onset of CO oxidation and a small in-
crease in the Cu carbonyl intensity along with a sharper
wavenumber decrease for that band. It is also clear that no
Fig. 1. Top: Conversion profiles obtained during the CO–O –NO and
2
+
CO–O reactions over CuZC. Bottom: Outlet N O and NO concentrations
2
2
2
observed during the course of the CO–O –NO reaction.
2
Fig. 2. In situ DRIFTS spectra observed during the course of the CO–O –NO reaction at the indicated temperatures. The label INIT refers to the spectrum
2
recorded prior to introduction of the reaction mixture.

A. Martínez-Arias et al. / Journal of Catalysis 214 (2003) 261–272
265
Fig. 4. Static infrared results observed for the CuZC sample reduced under
CO at Tr = 373 K (the spectrum obtained after this treatment is taken as
the reference for the difference spectra shown in the figure): (a) following
exposure to 10 Torr of NO at RT; (b) subsequent exposure to 100 Torr
of NO at RT; (c) subsequent outgassing at RT; (d) exposure to 10 Torr of
NO + 45 Torr of O at RT.
2
for T ꢂ 393 K. Other changes in the spectral range 1800–
−1
1
000 cm which occurred during the course of the reac-
tion are shown in Fig. 2, although complete resolution of the
bands involved in these changes is difficult due to overlap-
ping with bands already present for the initial sample, most
likely due to bulk carbonate-type species, difficult to elim-
inate due to the relatively strong basic character of ceria-
related materials [1]. It appears in any case that a band at
+
Fig. 3. Top: Intensities observed by DRIFTS for the Cu carbonyl at
2
−
111–2105 cm (solid lines) and the CO (g) bands (dotted lines) during
2
1
the course of the CO–O (circles) and CO–O –NO (squares) reactions.
Bottom: Evolution of νCO for the Cu carbonyl during CO–O (circles)
2
2
+
2
and CO–O –NO (squares) reactions.
2
−1
1
461 cm evolves from 393 K while species giving rise to
a set of bands and unresolved features at 1575, 1545, 1522,
−1
correlation exists between CO coverage (assuming it is pro-
portional to band intensity) and the Cu carbonyl stretch-
and 1266 cm are formed at intermediate temperatures (be-
tween 393 and 453 K).
+
ing frequency (Fig. 3). This is further confirmed by exam-
ining the postreaction catalyst. For this purpose, the catalyst
was cooled under N2 at the end of the CO–O2–NO reac-
tion, and 1% CO was flowed over the catalyst at RT before
flushing the cell with N2 at RT and recording a spectrum
under this inert atmosphere. Under these conditions (result
Transmission infrared spectra following NO or NO + O
adsorption at RT on CuZC reduced under CO at 373 K are
shown in Fig. 4. Exposure to 10 Torr of NO mainly generates
2
−1
species giving bands at 1870, 1788, and 1104 cm , while
weaker bands or shoulders are also apparent at 1613, 1569,
−1
1510, 1471, 1429, 1286, 1104, and 826 cm (Fig. 4a).
A subsequent increase in the NO pressure (up to 100 Torr)
+
not shown), the Cu carbonyl band displayed an integrated
−1
intensity roughly four times larger than for the experiment
conducted at 303 K under reaction conditions (Fig. 2) and
resulted in the disappearance of the band at 1788 cm and
an increase of the rest of the bands with the maximum at
−1
−1
−1
shows a maximum at 2104 cm
.
1870 cm being shifted to 1883 cm (Fig. 4b). The main
modification produced upon subsequent outgassing at RT
Comparison with similar experiments performed in the
absence of NO [16] enables one to determine the formation
of bands derived from NOx adsorption or reaction. A band
−1
was the disappearance of the band at 1883 cm (Fig. 4c).
On the other hand, strong bands and other unresolved
−1
at 1870 cm appeared instantly upon contact with the CO–
O2–NO reactant mixture at 303 K and strongly decreased
with increasing reaction temperature and was practically un-
detectable for T > 363 K. This band can be attributed to
features appeared upon exposure to NO + O , forming sets
2
of overlapping bands in the ranges 1625–1400 and 1290–
−
1
1240 cm with two weaker but well-defined bands at 1010
−1
and 801 cm
.
2
+
−1
nitrosyl species adsorbed at exposed Cu cations [32]. Ad-
Bands in the range 1900–1700 cm appearing upon NO
adsorption are typical of adsorbed nitrosyl species, the dif-
ference between them being mainly related to the changes
induced in the N–O bond resulting from changing the oxi-
−1
ditionally, a band at 1184 cm also appeared upon contact
with the reactant mixture at 303 K, reaching maximum in-
tensity in spectra with the sample at 333 K and disappearing

266
A. Martínez-Arias et al. / Journal of Catalysis 214 (2003) 261–272
dation state of the adsorption site [32]. Thus, bands at 1870
−1
(
or 1883) cm , which are similar to those observed under
CO–O2–NO reaction conditions (Fig. 2), are assigned to ni-
2
+
trosyl species adsorbed on Cu cations, while the band at
−
1
+
1
788 cm is assigned to nitrosyl species adsorbed on Cu
−1
sites [32]. The band at 1184 cm , which also appeared un-
der reaction conditions (Fig. 2), is typical of chelating nitrite
species coordinated to cerium cations [25]; its assignment
to a species chemisorbed at sites on the support rather than
at exposed copper oxide sites is supported by its observa-
tion on a Pd/CeZrO sample treated under similar condi-
4
−1
tions [33]. The absorption at 1184 cm can be assigned to
the νasym(NO2) mode of the complex and the weaker band at
−1
8
26 cm to the δ(NO2) mode [25,32]; the νsym(NO2) mode
of the adsorbed complex, which is expected with a very
2
+
Fig. 5. Intensities of EPR signals due to Cu
species (solid lines) and
−
Ce –O species (dashed lines) as a function of the amount of O
2
−1
weak intensity at ca. 1270 cm [25], is probably obscured
by features corresponding to other species. The shoulder at
4+
2
(
circles) or NO (squares) adsorbed at RT on CuZC reduced under CO at
−1
1
104 cm is assigned to trans-hyponitrite species coordi-
Tr = 373 K.
nated to cerium cations [25]. The series of bands at 1625–
−1
1
400, 1290–1240, 1010, and 801 cm can be assigned to
CuZC sample was prepared by reducing the initial calcined
sample under CO at 373 K (followed by outgassing at the
same temperature). Then, increasing amounts of either O2 or
NO were chemisorbed over that sample at RT, followed by
thorough outgassing at the same temperature. The evolution
different adsorbed nitrate or nitro species [32,34–36].
3
.3. EPR results
of the Cu2 signals (identified mainly as Cu
+
2+
species
A previous report demonstrated in some detail the nature
of Cu2 signals in CuZC in its initial calcined state as well as
the evolution of the species as a function of treatments under
CO at different reduction temperatures [16]. Approximately
+
within small copper oxide clusters) was then monitored
during the reoxidation treatments. Following each of these
treatments, the samples were exposed to O2 (a dose of
0 µmol per g of catalyst) at 77 K (followed by outgassing
at 77 K) and the evolution of O2 –Ce
3
8% of the entire copper content was detected for the initial
7
−
4+
calcined sample, the spectrum being mainly formed by a fea-
signals was
tureless broad signal showing extremes at g = 2.21 and g =
monitored. The results obtained are shown in Fig. 5 and
provide evidence that O2 chemisorption induces a higher
degree of copper oxidation than NO chemisorption and, in
addition, the former is more effective in the elimination of
2
.04, attributed to Cu2 cations in small copper oxide clus-
+
ters, and a minor signal at g|| = 2.233 and g⊥ = 2.036 show-
ing four-line hyperfine splittings in each of its components
−
3
−1
−3
−1
−
4+
with A|| = 16.0 × 10 cm and A⊥ = 1.8 × 10 cm
,
centers which might form O2 –Ce species at 77 K.
2
+
attributed to isolated Cu cations in a (generally speaking)
tetragonally distorted octahedral symmetry. The portion of
the copper which remained undetectable by EPR for the ini-
tial sample was proposed, on the basis of XPS results, to be
3.4. XPS results
No significant changes in the Cu/(Ce+Zr) and Ce/(Zr+
Ce) atomic ratios were observed (Table 1) during the
course of the treatments following a protocol very similar
to that used for EPR experiments (see above). Values for
those ratios were similar to those observed for the initial
calcined sample or for the copper-free support, where a
certain surface Ce enrichment was apparent [16,37]. These
similarities indicate that the incorporation of the active
due to magnetically coupled Cu2 species forming part of
+
less dispersed copper oxide particles. A significant decrease
in the Cu2 intensity, affecting both signals present in the
spectrum, was produced upon reduction of the sample under
CO at RT and 373 K. Thus, ca. 24 and 8% of the entire cop-
+
2
+
per loading was detected as Cu after those experiments.
Exposure of the reduced sample to oxygen at 77 K led to the
appearance of superoxide species chemisorbed on cerium
copper phase does not modify the surface of the CeZrO and
4
−
4+
cations (formally O2 –Ce species) giving rise to signals
at gz = 2.031–2.029, gx = 2.016, and gy = 2.011 [16]. For-
mation of those species provided the principle evidence to
indicate that support reduction had occurred, considering
that no significant changes in copper dispersion occur during
the series of redox treatments.
Concerning changes in the redox state of cerium, the
CuZC sample reduced in CO at 373 K contains ca. 15%
that paramagnetic Ce3 species are not easily detectable by
EPR due to their short spin–lattice relaxation time [2].
On the basis of those results, a specific experimental
protocol focused on determining the redox properties of
CuZC upon interaction with O2 or NO. For that purpose, a
+
Ce , which is almost fully reoxidized (< 5% Ce remain-
3+
3+
−4
ing) immediately upon contact with 10 Torr of either O2
or NO at RT. This provides further evidence for the pres-
3
+
ence of Ce , as detected by EPR (through the formation
of Ce4 –O2 species), after the reduction step. The pres-
+
−

A. Martínez-Arias et al. / Journal of Catalysis 214 (2003) 261–272
267
ence of a small amount of Ce3 after the reoxidation ex-
periments probably results from the effect of the ultra-high-
vacuum environment removing the most labile chemisorbed
oxygen species [16,25]. Certain changes were also observed
in the Zr 3d signals during the course of the treatments.
Thus, the relatively broad doublet observed for the sam-
ple reduced with CO at 373 K was narrowed upon contact
with small amounts of NO, yielding a situation closer to that
presented by the initial sample calcined at 773 K (3d5/2 at
+
to the ꢀα = 2ꢀR relationship. This suggests a strong
contribution of the CeZrO4 interface to the relaxation
process, as previously reported by Mejias et al. for ultrathin
TiO2 films on different types of supports [40]. On the other
hand, according to its position (Fig. 6), highly dispersed
ea
+
Cu ions in Cu-exchanged Y-zeolites would experience a
much lower core-hole extraatomic relaxation that can be
related to the smaller polarizability of the oxide ions within
the zeolitic framework than of oxide ions in bulk Cu2O
[39,41]. Thus, we may conclude, in agreement with the
EPR data above, that after reduction with CO at 373 K
the active phase in CuZC is mostly reduced to a Cu2O-
like state, probably forming thin two-dimensional patches
or small oxide clusters tightly interacting with the surface of
the CeZrO4support, in such a way that the less polarizable
(in comparison with bulk compounds) oxide ions at this
1
82.2 ± 0.1 eV and 3d3/2 180.8 ± 0.1 eV). However, no sig-
nificant modification of these signals was produced upon in-
teraction with O2. The broadening of the Zr 3d peaks can be
tentatively interpreted as due to the high relative heterogene-
ity in the Zr4 coordination environments for the sample re-
duced by CO at 373 K rather than to changes in its oxidation
state under such mild reduction conditions. The broader and
slightly asymmetric Zr 3d doublet probably corresponds to
a deconvolution of spectra and can be reasonably fitted us-
ing the same single set of parameters employed for the initial
calcined sample, although larger FWHM values must be em-
ployed.
+
+
interface actively participate in the Cu core-hole relaxation
process.
Table 1 shows data for EB and EK obtained following
reoxidation of the CO-reduced sample by either O2 or NO,
following a similar protocol to that employed for the EPR
experiments. As observed in Fig. 6, increasing doses of
O2 at RT produce displacements in the Wagner chemical
state diagram which, after a treatment with 1 Torr of O2
at 473 K, reach a position that fits on the line of slope −3
To examine in detail changes in the active Cu phase after
each redox treatment, the energies of the Cu 2p3/2 and
Cu(L3M45M45, G) peaks in Table 1 were plotted (Fig. 6),
1
ꢀ
together with the modified Auger parameter (α = EB +
+
EK), as Wagner-type chemical state diagrams. According
corresponding to CuO. This suggests that Cu cations in the
2
+
to Wagner [38], the shifts in modified Auger parameter
Cu2O-like phase are being progressively oxidized to Cu
,
ꢀ
(
α = EB + EK), which can be accurately measured in the
a process that is limited to ca. 40–45% after 1 Torr O2
exposure at RT but reaches 100% at 473 K. This is supported
by analysis of the EB and EK values for the intermediate
oxidation steps, which correspond to deconvoluted spectra
presence of static sample charging and shown as lines of
slope −1 in the diagrams, can be related to differences in the
ꢀ
ea
extraatomic relaxation energy (ꢀα = 2ꢀR ) for the core
+
2+
photohole according to a nonlocal screening mechanism
that include Cu /Cu
components in the appropriate
(
final state effects). On the other hand, within Moretti’s
ratios. Two points are worthy of note here; first, a further
reduction with CO at 373 K completely restores the original
EB and EK values corresponding to Cu2O-like patches with
the same contribution of the support to the core photohole
relaxation process; and second, according to the EB and
EK values obtained for the sample reoxidized in 1 Torr of
O2 at 473 K, a quite similar effect of the interaction with
the support on the relaxation process seems to occur in the
fully oxidized CuO-like state. This suggests that the copper
dispersion is roughly maintained throughout the series of
treatments, since copper agglomeration and growth to give
electrostatic model [39], which assumes a nonlocal core-
hole screening relaxation mechanism, the lines of slope
−
3 in these diagrams correspond to species with the same
chemical state (i.e., similar initial state effects).
From the Wagner’s plot in Fig. 6, it may be concluded that
Cu in the initially CO-reduced CuZC catalyst is chemically
similar to Cu in bulk Cu2O. However, the shift toward
+
ꢀ
a lower value of α within the dashed line of slope −3
crossing through the point for bulk Cu2O indicates a much
smaller extraatomic relaxation of the photohole, according
Table 1
XPS data of CuZC reduced under CO at 373 K and subsequently reoxidized with O or NO at RT
2
ꢀ
Treatment
Cu/(Ce + Zr)
Ce/(Ce + Zr)
O/(Ce + Zr)
Cu 2p EB
eV)
Cu (AES) EK
α (Cu)
(
(eV)
(eV)
CO/1 Torr/373 K
0.076
0.064
0.077
0.082
0.59
0.59
0.59
0.59
2.00
1.98
2.07
2.06
932.8
933.1
933.1
933.3
915.3
915.7
915.6
916.0
1848.1
1848.6
1848.7
1849.3
−4
O /10 Torr/RT
2
−
2
O /10 Torr/RT
2
O /1 Torr/RT
2
CO/1 Torr/373 K
0.079
0.070
0.072
0.081
0.59
0.58
0.59
0.59
2.00
2.08
2.02
2.17
933.0
932.9
933.0
933.2
915.2
915.3
915.6
915.6
1848.2
1848.2
1848.6
1848.8
−4
NO/10 Torr/RT
−
2
NO/10 Torr/RT
NO/1 Torr/RT

268
A. Martínez-Arias et al. / Journal of Catalysis 214 (2003) 261–272
duced surface of the CeZrO4 at RT, as shown by the ob-
served evolution of the Ce 3d XP spectra during the course
of the treatments.
4
. Discussion
Our previous report dealing with the CO oxidation reac-
tion (employing O2 alone as oxidant) over the CuZC catalyst
proposed that the reaction follows a redox mechanism (of the
Mars–van Krevelen type) in which the catalyst oscillates be-
tween an initial fully oxidized state (that may be represented
as Cu2 –O –Ce , taking into account the interface char-
+
2−
4+
acter of the active sites and considering formal charges for
+
the sake of simplicity) and a partially reduced state (Cu –
3
+
V –Ce ) [16]. This proposal was mainly based on the fol-
Ö
lowing experimental evidence [16].
(
i) Reduction or oxidation of the catalyst is readily
achieved upon interaction at low temperature with CO or O2,
respectively. In fact, these redox processes take place in
CuZC even at RT, as indicated by in situ DRIFTS, EPR,
CO-TPD results, and XPS. This contrasts with the greater
difficulties encountered in trying to carry out similar redox
processes over either pure CuOx or the Cu-free ceria-related
oxide support [9,21,42], thus indicating the involvement
in these redox processes of relatively strong synergetic
interactions between both catalyst components. Involvement
Fig. 6. Wagner diagram showing the evolution of Cu (2p and AES) XPS
parameters during reoxidation of CuZC reduced under CO at Tr = 373 K
(
open circles) with increasing amounts of either O (open squares) or NO
2
2+
+
of the Cu –Cu redox pair instead of other alternatives
(open triangles). Parameters obtained for the sample reoxidized under 1 Torr
0
0
(
e.g., Cu2 –Cu or Cu –Cu [21]) was mainly based on
+ +
of O at 473 K are also shown (as a crossed circle point). Dashed lines of
2
slope −3 are shown as a reference guide. Values for the reference samples
the absence of spectroscopic evidence of the formation of
(
full circles) can be found elsewhere [39,41].
the Cu0 state under reaction conditions, along with the
2
+
relative ease with which the initial Cu state was restored
3
D particles, either in a reduced or oxidized state, would
upon interaction of the reduced catalyst with O , which
2
produce a loss of the interaction with the CeZrO4 surface
and an increase in the contribution of oxide ions from bulk-
like phases that would lead to displacements toward the
Cu2O or CuO reference values shown in the diagram.
is confirmed here by the XPS data analysis. Evidence for
the presence of Cu2 states for the initial calcined CuZC
catalyst was obtained by XPS and agrees with other XPS
or EXAFS experiments performed on samples similar to the
CuZC [15,42].
+
+
2+
A gradual reoxidation from Cu to Cu is also pro-
duced following interaction of the CO-reduced sample with
increasing amounts of NO at RT. However, a lesser degree
(ii) Regarding the nature of active sites, and in accor-
dance with proposals by other authors [5,9,17], the inter-
face sites between the copper oxide and the support were
invoked in the reaction mechanism. This follows from the
of oxidation (ca. 20–25% of Cu2 ), is achieved, compared
with reoxidation under O2, upon contact with 1 Torr of NO.
Furthermore, the degree of oxidation only increases up to
ca. 50% under heating at 473 K under 1 Torr of NO. Addi-
tionally, the concomitant growth of a small, very broad N 1s
signal, centered at 403.5 eV, is observed during the course
of these experiments. This indicates the formation of tightly
chemisorbed NOx species (not necessarily only on the Cu2O
phase), in agreement with infrared results. In summary, and
as previously found by EPR (Fig. 5), XPS data confirm that
O2 is a stronger oxidant than NO, either at RT or 473 K,
for the reduced active Cu-phase in our CuZC catalyst. In
addition, the strong interaction between the active copper
component and the support is retained under the mild ex-
perimental conditions (T < 473 K). On the other hand, the
small doses of both O2 and NO are able to oxidize the re-
+
+
fact that the formation of Cu carbonyl species (as charac-
−1
terized by DRIFTS) giving a band at ca. 2110 cm was ob-
served under reaction conditions. As mentioned above, as-
+
signment of this band to Cu carbonyl species, although at
0
a position typical of Cu carbonyls [43], was based on the
relatively high stability of those carbonyl species during out-
gassing [16], in contrast to the typical behavior expected for
0
Cu carbonyls [19,44]. Indeed, the relatively low wavenum-
ber observed for such particular carbonyls, in comparison
+
with other Cu –CO species, was attributed to an interaction
with the basic support [16,18,19,45], thus indicating that the
species were formed on copper sites affected by an interac-
tion with the ceria-related support [12,18,19]. Both the oxi-
dation state of the copper adsorption centers responsible for

A. Martínez-Arias et al. / Journal of Catalysis 214 (2003) 261–272
269
these carbonyls and its ascription to centers interacting with
the support are supported by the XPS results here, as indi-
cated by the analysis performed above. EPR results, on the
work by Harrison et al. [15]. It is interesting to note that
in that study an important improvement in the parameter
refinement was reported to be produced when a cerium atom
at a distance of ca. 4.2 Å was included in the EXAFS
analysis for that phase [15]. This suggests that the most
active copper species can be related either to some kind
of copper–cerium mixed oxide structure at a local level
(as proposed for other systems of this kind [15,50,51]) or
to thin two-dimensional raft-like copper oxide layers (2D
−
4+
other hand, showed the formation of O2 –Ce complexes
upon oxygen chemisorption at 77 K on a mildly reduced
CuZC sample (Tr = RT) [16], in contrast to results observed
for the support alone [19,46]. Such complexes were almost
completely absent in CuZC after warming to RT. This was
demonstrated to be due to the particular ease by which oxy-
gen reduction occurred over the reduced surface (leading to
diamagnetic species such as peroxide or oxide anions, whose
formation requires the transfer of two or more electrons),
which can be related to the high density of electrons (in the
form of reduced cerium centers of the support) that become
concentrated at the interface zone following reduction.
patches), in which the properties of the Cu2 cations would
differ from those in pure CuO as a consequence of their
interaction with the underlying ceria-related support. This
is supported by XPS analysis (Fig. 6), which indicates that
most of the CuO in the CuZC catalyst is present as a
highly dispersed thin phase in close interaction with the
CeZrO4 surface, so that the local electronic properties of the
+
2
+
Furthermore, among the various kinds of Cu entities
present in the catalysts (in the form of isolated cations,
small CuO-type patches or clusters, or less dispersed copper
oxide particles), it was proposed, mainly on the basis of
their higher redox activity, that the active sites for the
reaction were those involving contact between small CuO-
type clusters and the ceria-related support [16]. This agrees
with other proposals in the literature, which suggest that the
copper-related active centers are associated with the finely
dispersed states of copper oxide, which are able to maximize
the number of CuOx-support interface sites [5,11,17]. In
2
+
Cu ions (including the core-hole relaxation process) are
greatly modified when compared with those of bulk CuO (or
Cu2O). A similar scenario has been envisaged for Cu/TiO2
catalysts on the basis of IR spectroscopic evidence [52].
This modification induced by the support interaction can
facilitate the Cu2 /Cu redox exchange conditions and
+
+
+
the stabilization of adsorbed CO on the Cu ions, which
are readily formed even under stoichiometric CO + O2 or
CO + O2 + NO mixtures.
+
Analysis of the evolution of the Cu carbonyls giving the
this respect, it may be noted that isolated Cu2 species
also show significant redox activity toward CO–O2 [16,19],
although they were disregarded as being the main source
of catalytic activity in CuZC or similar catalysts, as they
usually represent only a very small portion of the whole
copper content of the catalyst [16].
+
band at 2111–2105 cm under reaction conditions reveals
−1
interesting differences depending on the presence/absence
of NO in the reaction mixture (Fig. 3). First, it can be
considered that the intensity of this band is, in principle,
a function of three different factors: the availability of
+
Cu sites at the sample surface (in contact with the ceria-
To the best of our knowledge, little is known about
the structural/morphological aspects of the active CuO-type
clusters. The above description of the active phase of copper
in terms of such copper oxide aggregates is mainly based
on the fact that, in contrast with other supports such as
alumina [47], no specific compounds appear to be formed
between ceria-type and copper oxides, although some degree
of solubility of oxidized copper in ceria seems possible in
view of the changes in the fluorite lattice constant observed
for this type of system [48]. Indeed, previous XPS results
have revealed that the stabilization of most of the copper in
this kind of catalyst is in the form of CuO-type phases [19].
related support, as discussed above); the concentration of
CO in the reactant gas phase, which is a function of the
extent of reaction; and the reaction temperature, which
affects the adsorption/desorption equilibrium. Concerning
the latter, the significant resistance of those carbonyls to
desorption should be noted, since for CuZC significant
intensity is still observed after RT CO adsorption on a (CO–
O2) post-reaction sample followed by extensive treatment
from RT to 373 K under an inert gas flow [16]. Thus,
taking the above factors into account and assuming that
copper appears in a Cu2 state before interaction with
the reaction mixture (as supported by XPS results), it
appears reasonable that the initial low-temperature increase
in band intensity under CO–O2 is due to copper reduction
under the stoichiometric reaction atmosphere. Taking into
account that the evolution of the CO concentration must
be similar for both reactions (considering the similar CO
oxidation activities), the suppressed formation of the copper
site giving this band in the presence of NO must, in
principle, be related to either one of the other two factors or
both, i.e., an NO-induced suppression of copper reduction
+
From the EPR prospective, Cu2 cations within those CuO-
type clusters, in addition to displaying significant redox
activity (Fig. 5 and [16]), are characterized by a relatively
broad signal without a clearly defined hyperfine structure
in any of its components [16]. This is attributed to the
+
presence of magnetic interactions between the Cu2 cations
within the clusters, of a magnitude not sufficiently large
to form antiferromagnetic couplings, which dominate in
the case of well-formed CuO crystals [49]. Those spectral
characteristics are also compatible with the presence of
amorphous copper oxide layers with a disordered structure,
which are considered to be the most active phase in the
+
+
to Cu or a blocking effect of surface copper cations
by NOx-derived adsorbed species. Analysis of the NO-
derived bands for the in situ DRIFTS spectra in Fig. 2

270
A. Martínez-Arias et al. / Journal of Catalysis 214 (2003) 261–272
does not allow a definitive conclusion to be drawn in this
respect. Only two bands of this kind are clearly observed
at reaction temperatures that might be related to species
that might affect the CO oxidation activity or the NO-
degree of oxidation at the copper oxide phase (at least
at low temperatures, before or immediately after CO–O2
reaction onset) is achieved under stoichiometric CO–O –NO
2
than under stoichiometric CO–O . This is apparently in
2
+
induced suppression of Cu -carbonyl formation. In the first
contradiction with EPR and XPS results (Figs. 5 and 6).
place, the Cu2 nitrosyl giving the band at 1870 cm does
not appear to be responsible for the mentioned effects, on
the basis of the mismatch between the evolution of the
intensities of the respective bands (Fig. 2), since both the
nitrosyl and the carbonyl show maximum intensity at 303 K
and decrease at higher reaction temperatures. Second, the
+
−1
While both NO and O show a large capability to oxidize
2
the reduced copper oxide phase (as also inferred from the
evolution of the copper-related nitrosyls in the static infrared
experiments of Fig. 4), the results clearly show the higher
oxidizing power of O2 than NO under the static equilibrium
conditions employed. Experiments are under way to try to
elucidate this phenomenon and two hypotheses are currently
being considered: first, the effects of synergetic NO–O2
interactions; second, the existence of a prevailing kinetic
control governing the interactions of the different reactants
under the dynamic conditions employed for the in situ
DRIFTS or the catalytic tests.
−1
chelating nitrite species give the band at 1184 cm . This,
however, corresponds to a species adsorbed on the support
and could only affect copper sites indirectly. Nevertheless,
as shown in more detail in a previous study [25], those
species are proposed to be formed by the interaction of
NO with Ce4 sites most likely formed during support
dehydroxylation (observed by in situ DRIFTS). Therefore,
their formation would not in principle be affected by
CuOx → CeZrO4 electron transfer processes that could
eventually lead to a higher degree of oxidation of the
copper oxide phase. Those electron transfer processes would
+
+
The decrease in wavenumber of the Cu carbonyl ob-
served just prior to the onset of the CO–O2 reaction (Fig. 3)
suggests that the reaction is triggered as soon as a certain
degree of reduction is achieved at the copper oxide phase.
This resembles somewhat the processes which are proposed
during the so-called “induction period,” which precede cat-
alyst light-off, in the reaction model proposed by Wang et
al. [21]. However, in contrast to that model, the different in-
generate Ce3 –VÖ sites, and these are not involved in
the formation of the chelating nitrite [25], thus making
it unlikely that formation of these species is involved to
any extent in the NO-induced suppression of the copper
oxide reduction process. In any case, it may be noted
that the evolution of both NO-derived species (Fig. 2), but
mainly that of the chelating nitrite species, accounts for
the NOx adsorption/desorption phenomena observed at low
reaction temperatures (ꢀ 430 K) in the activity profiles
+
+
tensities of the Cu carbonyl observed in the presence or
absence of NO just before reaction onset suggests that the
location of the active sites for CO oxidation is largely re-
stricted to a limited region of the catalyst, i.e., to copper
oxide–CeZrO4 interface positions, and that the reaction is
not significantly extended onto copper oxide sites far beyond
that region. A plausible model to account for the results ob-
tained is depicted schematically in Fig. 7. Active sites are
basically restricted to positions at the interface between the
thin CuOx 2D-patches or more or less flat copper oxide lay-
ers (in order to account for the existence of relatively strong
interactions between them and the underlying (Ce, Zr)Ox
support, according to XPS, infrared, and EPR results). Con-
cerning processes occurring on the copper oxide phase, the
(
Fig. 1). Regarding the effects of other NO-derived species
+
on the Cu -carbonyl formation, this is more difficult to
resolve, due to the overlapping of several different bands
in the corresponding spectral region. Therefore, it is not
clear whether a blocking of CuOx sites by nitrate or
nitro species is taking place, although bands corresponding
to these species appear at relatively high temperatures
and would not therefore affect CO oxidation activity or
+
NO-induced suppression of Cu -carbonyl formation to
any great extent. It would appear that the lower Cu -
+
presence of NO would mainly affect sites on the CuO layer
x
at positions relatively far from the active interface bound-
carbonyl intensity achieved in the presence of NO must
be mainly related to the maintenance of a higher degree
of oxidation at the copper oxide surface. This is supported
by the comparatively higher wavenumber observed for
+
aries, restricting formation of Cu carbonyls at those sites
either through a blocking effect (by chemisorbed species) or
by an oxidative interaction which favors the establishment
of a higher degree of oxidation at these sites. On the basis
of the absence of NO effects on the CO oxidation catalytic
activity, it appears that NO does not strongly interact with
the active interface sites and that the interactions with CO
and oxygen prevail in that zone. It can be assumed that both
the copper oxide and the support components of the inter-
face are involved to some extent in the reaction mechanism
by providing sites for activation of CO and O2. In this way,
the absence of NO competition with CO at the copper oxide
component of the interface can be related with the attainment
at a local level of a relatively high degree of reduction in that
+
the Cu carbonyl when NO is present in the reaction
mixture (Fig. 3) since, according to infrared studies of
reduced copper oxide samples [18,44,53], the vibrational
+
frequency of Cu carbonyls decreases by increasing the
+
degree of reduction around the Cu adsorption center.
This hypothesis is supported by the fact that the lower
−1
wavenumber for that band (2104 cm ) is observed in the
post-reaction experiment where a maximum overall degree
of reduction has been achieved on the basis of observation
of the greatest intensity for the band under that condition.
A question arises as to the reason that a higher overall

A. Martínez-Arias et al. / Journal of Catalysis 214 (2003) 261–272
271
Fig. 7. Schematic model of the proposed active surface for the CuZC catalyst during the course of the CO–O and CO–O –NO reactions.
2
2
zone (in agreement with analysis of the results of Fig. 3), if
it is taken into account that CO predominantly interacts with
reduced copper sites (and NO with oxidized sites) during
CO–NO competition [23]. On the other hand, the prevail-
ing interaction of oxygen, with respect to NO, with the ac-
5. Conclusions
The presence of copper oxide entities, most likely form-
ing 2D patches or thin raft-like layers and strongly interact-
ing with the CeZrO4 support, is inferred from analysis of the
Wagner diagram of the XPS results, both for the fully oxi-
tive Ce3 –V interface sites at the support is demonstrated
+
Ö
by the EPR results (Fig. 5). Thus, the largest hindrance to
2+
+
dized (Cu ) and partially reduced (Cu , achievable upon
−
4+
O2 –Ce formation produced when oxygen, in compari-
son with NO, is preadsorbed at RT reveals that oxygen is a
more powerful oxidant than NO for the reduced active sup-
port sites.
reduction under CO at 373 K) states of the catalyst, in agree-
ment with analysis of Cu signals by EPR and of Cu car-
bonyl infrared bands.
2+
+
The behavior of the CuOx/CeZrO4 catalyst reduced
under CO at 373 K toward reoxidation by either O2 or NO
has been examined by XPS and EPR. Both molecules readily
oxidize both catalyst components, although oxygen displays
the greater oxidizing power for both sites at the CuOx or (to
a greater extent) CeZrO4 parts of the catalyst.
No significant modifications were observed in the cat-
alytic activity for CO oxidation (with oxygen) when NO is
introduced (in a relatively small amount) into the stoichio-
metric reactant mixture. Analysis of this observation with
the most important differences observed by in situ DRIFTS
Concerning other mechanistic aspects, the fact that the
CO–O2 catalytic activity appears to be restricted to the in-
terface positions of the catalyst along with observation that a
certain reductive preactivation of the copper oxide phase ap-
pears to be produced prior to the onset of reaction suggests
that processes involved in the reaction mechanism for this
kind of system must essentially depend on the properties of
the interface between the partially reduced copper oxide en-
tities and the ceria-based support. Two factors stand out as
being most important in accounting for the catalytic proper-
ties for carbon monoxide oxidation in this kind of system.
Both essentially depend on the characteristics of the copper
oxide–ceria-based support synergetic interactions, although
some specific properties of the individual components may
also play a certain role. First, the facility for achieving par-
tial reduction of the copper oxide entities at the interface
zone, which, according to previous studies [6,9,21], is fa-
vored by an increase in dispersion of this component may
be most significant in accounting for the different activities
as a function of the level of copper oxide dispersion [6,9,
+
with respect to the intensity of Cu carbonyls (at sites in-
teracting with the CeZrO4 support) in the presence and ab-
sence of NO suggest that the active zones of the catalyst for
CO oxidation are largely restricted to the interface region. In
+
turn, the evolution observed in the wavenumber of the Cu
carbonyls under reaction conditions suggests that the copper
oxide component is reductively preactivated prior to the on-
set of CO oxidation. In general terms, the results obtained
suggest that the two main factors governing the catalytic ac-
tivity of this kind of system are the facility for partial pre-
reduction of the copper oxide component at the interface
zone (favored upon decreasing copper oxide particle size,
although the role of the support as a redox promoter must
also be considered) and the redox properties of the interface
between that partially reduced copper oxide component and
the underlying ceria-based support (specifically related to
2
1]. In the second place, the redox properties of the inter-
face sites between partially reduced copper oxide and the
ceria-based support, which probably depends to some extent
on the individual redox properties of the ceria-based sup-
port (for instance, the degree of unsaturation of the oxygen
vacancy defects that can be generated, its surface acid/base
properties, its electronic characteristics, etc.), may be the
most important factor in explaining differences in the cat-
alytic activities when the nature of the ceria-based support is
varied [16].
the characteristics of the CuO –support contacts and there-
fore strongly dependent on the support nature, i.e., the pres-
ence/absence of zirconium or other dopants in its composi-
tion).
x

272
A. Martínez-Arias et al. / Journal of Catalysis 214 (2003) 261–272
Acknowledgments
[26] A.B. Hungría, A. Iglesias-Juez, A. Martínez-Arias, M. Fernández-
García, J.A. Anderson, J.C. Conesa, J. Soria, J. Catal. 206 (2002)
81.
2
A.M.-A., A.B.H., and A.I.-J. thank the Comunidad de
Madrid for grants received, under which this work was
conducted, and for financial assistance (to A.M.-A.) under
the program “Ayudas para estancias breves en centros
de investigación extranjeros.” Financial help by CICYT
[27] A. Martínez-Arias, M. Fernández-García, V. Ballesteros, L.N. Sala-
manca, J.C. Conesa, C. Otero, J. Soria, Langmuir 15 (1999)
4796.
[28] A. Martínez-Arias, M. Fernández-García, A. Iglesias-Juez, A.B. Hun-
gría, J.A. Anderson, J.C. Conesa, J. Soria, Appl. Catal. B 38 (2002)
1
51.
29] R. Hesse, T. Chassé, R. Szargan, Fresenius J. Anal. Chem. 365 (1999)
8.
30] J.P. Holgado, R. Alvarez, G. Munuera, Appl. Surf. Sci. 161 (2000)
01.
(
Project MAT2000-1467) is also acknowledged.
[
[
[
4
References
3
31] A. Martínez-Arias, M. Fernández-García, A. Iglesias-Juez, A.B. Hun-
[
1] A. Trovarelli, Catal. Rev. Sci. Eng. 38 (1996) 439.
gría, J.A. Anderson, J.C. Conesa, J. Soria, Appl. Catal. B 31 (2001)
51.
[32] A.A. Davydov, in: C.H. Rochester (Ed.), Infrared Spectroscopy of
Adsorbed Species on the Surface of Transition Metal Oxides, Wiley,
New York, 1990, p. 75.
[
2] A. Trovarelli (Ed.), Catalysis by Ceria and Related Materials, in:
Catalytic Science Series, Vol. 2, Imperial College Press, London, 2002.
3] E.P. Murray, T. Tsai, S.A. Barnett, Nature 400 (1999) 649.
4] S. Park, J.M. Vohs, R.J. Gorte, Nature 404 (2000) 265.
[
[
[
5] W. Liu, A.F. Sarofim, M. Flytzani-Stephanopoulos, Chem. Eng. Sci.
[33] A. Martínez-Arias, unpublished results.
4
9 (1994) 4871.
[34] G. Ghiotti, F. Prinetto, Res. Chem. Intermed. 25 (1999) 131.
[35] P. Fornasiero, J. Kašpar, Collect. Czech. Chem. Commun. 66 (2001)
1287.
[36] K. Hadjiivanov, V. Avreyska, D. Kissurski, T. Marinova, Langmuir 18
(2002) 1619.
[
[
[
6] W. Liu, M. Flytzani-Stephanopoulos, J. Catal. 153 (1995) 304.
7] W. Liu, M. Flytzani-Stephanopoulos, J. Catal. 153 (1995) 317.
8] A. Tschöpe, W. Liu, M. Flytzani-Stephanopoulos, J.Y. Ying, J. Ca-
tal. 157 (1995) 42.
[
9] M.-F. Luo, Y.-J. Zhong, X.-X. Yuan, X.-M. Zheng, Appl. Catal. A 162
[37] A. Martínez-Arias, M. Fernández-García, A.B. Hungría, J.C. Conesa,
G. Munuera, in preparation.
(1997) 121.
[
[
[
10] P.W. Park, J.S. Ledford, Catal. Lett. 50 (1998) 41.
[38] C.D. Wagner, Faraday Discuss. Chem. Soc. 60 (1975) 291.
[39] G. Moretti, J. Electron. Spectrosc. Relat. Phenom. 95 (1998) 95.
[40] J.A. Mejias, V. Jiménez, G. Lassaletta, A. Fernández, J.P. Espinós,
A.R. González-Elipe, J. Phys. Chem. 100 (1996) 16255.
[41] G. Moretti, Zeolites 14 (1994) 469.
[42] Y. Hu, L. Dong, M. Shen, D. Liu, J. Wang, W. Ding, Y. Chen, Appl.
Catal. B 31 (2001) 61.
[43] P. Hollins, Surf. Sci. Rep. 16 (1992) 51.
11] M.-F. Luo, X.-M. Zheng, Acta Chem. Scand. 52 (1998) 1183.
12] A. Martínez-Arias, J. Soria, R. Cataluña, J.C. Conesa, V. Cortés
Corberán, Stud. Surf. Sci. Catal. 116 (1998) 591.
[13] P. Bera, S.T. Aruna, K.C. Patil, M.S. Hegde, J. Catal. 186 (1999) 36.
[14] J.B. Wang, W.-H. Shih, T.-J. Huang, Appl. Catal. A 203 (2000) 191.
[15] P.G. Harrison, I.K. Ball, W. Azelee, W. Daniell, D. Goldfarb, Chem.
Mater. 12 (2000) 3715.
[
16] A. Martínez-Arias, M. Fernández-García, O. Gálvez, J.M. Coronado,
[44] M.B. Padley, C.H. Rochester, G.J. Hutchings, F.J. King, J. Catal. 148
(1994) 438.
J.A. Anderson, J.C. Conesa, J. Soria, G. Munuera, J. Catal. 195 (2000)
2
07.
[45] A.A. Davydov, Kinet. Katal. 26 (1985) 157.
[46] A. Martínez-Arias, M. Fernández-García, C. Belver, J.C. Conesa,
J. Soria, Catal. Lett. 65 (2000) 197.
[47] M. Fernández-García, I. Rodríguez-Ramos, P. Ferreira-Aparicio, A.
Guerrero-Ruiz, J. Catal. 178 (1998) 253.
[48] S. Hoçevar, J. Batista, J. Levec, J. Catal. 184 (1999) 39.
[49] F. Mehran, S.E. Barnes, G.V. Chandrashekhar, T.R. McGuire, M.W.
Shafer, Solid State Commun. 67 (1988) 1187.
[
[
17] D.-H. Tsai, T.-J. Huang, Appl. Catal. A 223 (2002) 1.
18] A. Martínez-Arias, R. Cataluña, J.C. Conesa, J. Soria, J. Phys. Chem.
B 102 (1998) 809.
19] A. Martínez-Arias, M. Fernández-García, J. Soria, J.C. Conesa, J. Ca-
tal. 182 (1999) 367.
[
[
20] P. Bera, S. Mitra, S. Sampath, M.S. Hegde, Chem. Commun. 927
(2001).
[
[
21] J.B. Wang, D.-H. Tsai, T.-J. Huang, J. Catal. 208 (2002) 370.
22] B. Skårman, L.R. Wallenberg, P.-O. Larsson, A. Andersson, J.-O.
Bovin, S.N. Jacobsen, U. Helmersson, J. Catal. 181 (1999) 6.
23] Y. Fu, Y. Tian, P. Lin, J. Catal. 132 (1991) 85.
[50] W. Daniell, N.C. Lloyd, C. Bailey, P.G. Harrison, J. Phys. IV France 7
(1997) C2–963.
[51] M. Fernández-García, E. Gómez Rebollo, A. Guerrero Ruiz, J.C.
Conesa, J. Soria, J. Catal. 172 (1997) 146.
[52] F. Coloma, B. Bachiller-Baeza, C.H. Rochester, J.A. Anderson, Phys.
Chem. Chem. Phys. 3 (2001) 4817.
[
[
24] M. Fernández-García, A. Martínez-Arias, J.A. Anderson, J.C. Conesa,
J. Soria, Stud. Surf. Sci. Catal. 130 (2000) 1325.
[
25] A. Martínez-Arias, J. Soria, J.C. Conesa, X.L. Seoane, A. Arcoya,
[53] Y.A. Lokhov, V.A. Sadykov, S.F. Tikhov, V.V. Popovskii, Kinet.
Katal. 26 (1985) 177.
R. Cataluña, J. Chem. Soc. Faraday Trans. 91 (1995) 1679.