Activity of Rhx+ Species in CO Oxidation and NO Reduction in a CO/NO/O2 Stoichiometric Mixture over a Rh/CeO2-ZrO2 Catalyst

Article abstract of DOI:10.1006/jcat.1998.2222

The catalytic activity of a 0.30 wt% Rh/CeO₂-ZrO₂ catalyst (25% ZrO₂ in weight) 188 m²g^-1 was studied in the oxidation of CO and reduction of NO in a CO/NO/O₂ stoichiometric mixture, in the temperature range of 298-773 K. The hydrogenation of benzene, a structure insensitive reaction was used for counting surface Rh(0) atoms after either thermal treatment of the catalyst in hydrogen at 773 K, or reduction by the stoichiometric CO/O₂/NO reaction mixture. The bibliography shows that the turnover rate over Rh^x+ is lower than that over Rh(0) by two orders of magnitude. Rhodium supported on CeO₂-ZrO₂ submitted to the above reducing treatments was found to be totally inactive in benzene hydrogenation but active in CO/NO/O₂ reaction. As a conclusion the present catalyst consisted exclusively of Rh^x+ surface species. The catalytic behaviour of this rhodium-supported catalyst was interpreted in terms of multiple active sites containing one Rh^x+ ion grafted in the support and able to chemisorb CO and/or NO as for a rhodium homogeneous complex. This Rh^x+ is associated with two adjacent oxygen vacancies (and corresponding Ce³⁺ cations) linked to the reduction of the support and permitting oxygen and NO chemisorption. A method was defined for counting surface Rh^x+ containing multiple sites by measuring the number of NO molecules adsorbed at low temperature and desorbing at the temperature of light-off of CO. A (NO/Rh^x+) ratio of 2 was found, in agreement with a dinitrosyl species reaction intermediate.

Full text of DOI:10.1006/jcat.1998.2222

JOURNAL OF CATALYSIS 179, 469–476 (1998)
ARTICLE NO. CA982222
x+
Activity of Rh Species in CO Oxidation and NO Reduction in a CO/NO/O2
Stoichiometric Mixture over a Rh/CeO -ZrO Catalyst
2
2
�
Franck Fajardie, Jean-Fran c¸ ois Temp e` re,† Jean-Marie Manoli,† Olivier Touret,‡
�
, 1
Gilbert Blanchard, and G e´ rald Dj e´ ga-Mariadassou†
�
Rh oˆ ne Poulenc, Centre de Recherches d’Aubervilliers, 52 rue de la Haie Coq, 93308 Aubervilliers Cedex, France; †Universit e´ P&M Curie,
Laboratoire de R e´ activit e´ de Surface, CNRS UMR 7609, Case 178, T 54, 4 Place Jussieu 75252 Paris Cedex 05, France;
‡Rh oˆ ne Poulenc, Chimie Secteur I.O.M. Usine de La Rochelle B.P. 2049, 17010, La Rochelle Cedex, France
Received March 11, 1998; revised June 12, 1998; accepted July 7, 1998
(
5) and that the rate-determining step corresponds to the
The catalytic activity of a 0.30 wt% Rh/CeO2-ZrO2 catalyst (25% reaction of adsorbed CO molecule with adsorbed oxygen
2
� 1
ZrO2 in weight) 188 m g was studied in the oxidation of CO atoms to produce CO (6). When carried out in transient
2
and reduction of NO in a CO/NO/O2 stoichiometric mixture, in the
temperature range of 298–773 K. The hydrogenation of benzene, a
structure insensitive reaction was used for counting surface Rh(0)
atoms after either thermal treatment of the catalyst in hydrogen at
conditions, the metallic surface can be considered as fully
covered by CO at low temperatures, which prevents the
O2 dissociation (7–9). Increasing reaction temperature can
make O2 able to compete with CO for metallic adsorp-
7
73 K, or reduction by the stoichiometric CO/O2/NO reaction mix-
tion sites. Once dissociatively adsorbed, oxygen oxidizes
CO to CO2, whose desorption generates new adsorption
sites. When the metallic phase is dispersed on a reducible
support (CeO2), CO can oxidize at lower temperatures by
x+
ture. The bibliography shows that the turnover rate over Rh is
lower than that over Rh(0) by two orders of magnitude. Rhodium
supported on CeO2-ZrO2 submitted to the above reducing treat-
ments was found to be totally inactive in benzene hydrogenation
but active in CO/NO/O2 reaction. As a conclusion the present cata- reacting with oxygen ions at the metal–support interface
x+
lyst consisted exclusively of Rh surface species. The catalytic be- (3, 7, 10, 11). Zafiris et al. (9, 12) for Rh/CeO2 catalysts
haviour of this rhodium-supported catalyst was interpreted in terms have assumed that oxygen from reduced ceria migrate to
x+
of multiple active sites containing one Rh ion grafted in the sup-
the metallic rhodium surface where it can oxidize CO.
The CO/NO reaction occurs more easily on rhodium
metal than on Pt(0) or Pd(0) (13, 14), owing to its greater
port and able to chemisorb CO and/or NO as for a rhodium homo-
geneous complex. This Rh is associated with two adjacent oxygen
vacancies (and corresponding Ce cations) linked to the reduction
of the support and permitting oxygen and NO chemisorption. A
method was defined for counting surface Rh containing multiple
sites by measuring the number of NO molecules adsorbed at low
temperature and desorbing at the temperature of light-off of CO.
A (NO/Rh ) ratio of 2 was found, in agreement with a dinitrosyl
x+
3
+
ability to dissociate NO with regards to other metals, due to
the position of the Fermi level in rhodium metal, localized
x+
�
above the � orbital of NO (15, 16). The reverse situation
occurs for Pt(0) and Pd(0). When NO adsorbs on Rh(0),
x+
an enhanced charge transfer takes place from Rh(0) to the
�
species reaction intermediate.
�c 1998 Academic Press
�
orbital, increasing the back donation effect and, con-
NO
sequently, the weakening of the N-O bond.
Contrary to the CO/O2 reaction, the CO/NO reaction is
structure sensitive (1, 17, 18), due to the fact that NO dis-
sociates differently as a function of the considered metal-
lic planes (18). For alumina supported rhodium, Kaspar
et al. (17) have shown that at room temperature more than
I. INTRODUCTION
During the last years, the various reactions occurring be-
tween CO, NO, and O2 over three-way metallic catalysts
were the subject of numerous studies (1–4). As yet, active
sites for these reactions are not well defined. Authors gen-
erally believe that zero-valent metal is responsible for the
catalytic activity (1, 2) as is the case for CO/O2 and CO/NO
reactions over rhodium catalysts. It is known that on
rhodium metal, the CO/O2 reaction is structure insensitive
6
0% of NO molecules dissociate on the Rh(0) particles and
that the NO dissociation increases with the particle diam-
eter; so, above 15 A, more than 90% of the adsorbed NO
dissociates over Rh(0) particles, the remaining 10% des-
orbing as N2O and NO at 563 K. According to Cho et al.
˚
(
19, 20), the N2O desorption at low temperature is corre-
lated with the beginning of the CO/NO reaction which can
only start with the accessibility of the active sites initially
occupied by N2O. This N2O formation clearly indicates that
1
To whom correspondence should be addressed. E-mail: djega@ccr.
jussieu.fr.
469
0021-9517/98 $25.00
Copyright �c 1998 by Academic Press
All rights of reproduction in any form reserved.

4
70
FAJARDIE ET AL.
nitrogen atoms can bind, prior to NO dissociation. Thus, the ciation centers, or they can accomodate oxygen species
formation of N2O does not necessarily involve the inter- originating from the dissociation of O2 on the metallic
action at the metal surface of an adsorbed nitrogen atom surface,
with an adsorbed NO, as usually admitted, but can also
result from the breaking of the NO bond of hyponitrite
�
�
CO + (OV) → CO2gas + (V) + ( )
species (N2O2)2 originating from the dimerization of two
adjacent adsorbed NO (21–23): (NO)2ads.→N2Ogas + Oads..
N2O formation was also observed over Pt/Al2O3 and Pt-
CeO2/Al2O3 catalysts, especially in the case of highly dis-
persed metallic phase (22). Let us also note, that rhodium
sites generating (NO)2 species do not exclusively consist
of zerovalent rhodium atoms, since dinitrosyl complexes
can be evidenced on exchangeable rhodium ions in zeolites
�
�
�
NO + (V) → N + (OV)
�
�
(V) + O → ( ) + (OV),
where � is the adsorption site on the metal, (V) is the oxygen
vacancy on the support, (OV) is the oxygen ion located in
an oxygen vacancy.
From the above bibliography data, it appears that the
CO/NO/O2 reactions are complex and result from the com-
petition between the CO/O2 and CO/NO reactions. The ac-
tive sites are not yet well defined, but it is generally admitted
that the reaction takes place both on the reduced metal and
at the metal support interface.
High thermal stability and enhanced reducibility for the
ceria-containing support are essential requirements. It has
been found that incorporation of ZrO2 into a solid solution
with CeO2 strongly favours the reduction of the support
(
24).
As for CO/O2 reaction, reduced ceria can significantly
improve the catalytic activity of the system in the CO/NO
reaction (25, 26), by the presence of oxygen vacancies at
the metal support interface favouring the NO reduction;
oxygen atoms originating from the dissociation of NO at
the interfacial sites could locate in the oxygen vacancies,
while retaining a labile character which makes possible an
additional contribution to the CO oxidation.
In the presence of the three reagents (CO, NO, O2), the
two preceding reactions are competing, due to the fact that
NO and O2 compete for the active sites of the samples for
the CO/NO/O2 reactions. Leclercq et al. (3) have observed
for Pt/Al2O3 low NO conversion as long as O2 is present in
the reagent mixture and have concluded that the NO ad-
sorption sites could be blocked by adsorbed oxygen. Con-
versely, for Rh/Al2O3, Oh et al. (27) have shown that the
presence of NO in the reagents could inhibit the CO/O2 re-
action, at low CO/NO conversion. At low temperature of
reaction, these authors have found that undissociated NO
molecules are present at the surface and could block the
adsorption sites for the CO/O2 reaction. When the reaction
temperature increases, these NO molecules dissociate on
the metal, with a simultaneous formation of Nads. and Oads.
adsorbed species which can easily desorb from the metal-
lic surface (Nads. + Nads. → N2 and CO + Oads. → CO2), re-
generating active sites for the CO/O2 reaction. Even if
conclusions are sometimes controversial, as it is difficult
to determine which of NO and O2 inhibits the active
(
29, 30). These observations makes the catalytic properties
of the CeO2-ZrO2 catalysts of strong interest. We have se-
lected a CeO2-ZrO2 solid solution of high specific surface
area loaded with rhodium. This last metal being chosen for
its ability to enhance the NO conversion (1, 2).
The present work was devoted to the characterization
identification and counting) of active sites for the CO/
(
NO/O2 reactions present on a high surface area Ce/Zr
oxides supported rhodium catalyst. Catalyst preparation,
metal content, reactant mixture composition were carefully
chosen to study the Rh/CeO2-ZrO2 sample using experi-
mental conditions as close as possible to those used for real
postcombustion reactions.
EXPERIMENTAL
Catalyst Preparation
The CeO2-ZrO2 support (25% ZrO2 in weight) was a
2
� 1
sites at low temperature, on the whole it appears that the high specific surface area (188 m g ) solid solution sup-
CO/NO/O2 reactions can be considered as the succession of plied by Rh oˆ ne Poulenc, after calcination at 623 K, they
the CO/O2 and CO/NO reactions, with closely dependent were nitrates containing support even after this calcination
light-off temperatures (temperature for 50% conversion of step. The Rh(Cl)/CeO2-ZrO2 samples were prepared by
reactant).
anionic exchange from an acidic solution of RhCl3 � 3H2O
Ceric oxide additive, here again, confers to the catalysts (Johnson Matthey) adjusted with HCl at pH = 1.9. The
some promoting effect on their activity in the CO/NO/O2 amount of RhCl3 � 3H2O introduced into the solution was
reactions, by generating at the metal–support interface ad- calculated to obtain a metal loading of 0.30 wt% . It was veri-
ditional bifunctional sites (3, 28); under an oxygen-deficient fied that the presence of residual nitrate ions did not perturb
atmosphere, CO can migrate at the metal support interface the anionic exchange process. After exchange the samples
to be oxidized by a ceria lattice oxygen with concomitant were filtered, washed with distilled water, and dried in air
oxygen vacancy formation. The so-created oxygen vacan- for 3 h. The anionic exchange with rhodium complexes was
cies at the metal–support interface can act as NO disso- reproducible.

x+
ACTIVITY OF Rh SPECIES IN CO OXIDATION
471
Catalyst Characterization
(101 kPa.) in a differential dynamic microreactor. The total
�
1
flow rate was set to 100 ml min . The partial pressure of
3
Specific surface area measurements were determined by benzene was 51.8 Torr (6.9 � 10 Pa) so that a zero-order
physisorption of nitrogen at 77 K using a Quantasorb Jr. dy- relative to benzene was obtained. In a typical run the cata-
namic system, fitted out with a catharometer detector. The lyst prior to run was pretreated at 773 K in flowing H2 or in
specific areas were calculated using the BET method. Sam- the postcombustion mixtures (CO/NO/O2 or CO/O2). The
ples studied by high resolution electron microscopy were catalyst was then cooled to temperature of reaction under
ground and dispersed in pure ethanol; the suspension was helium and the C6H6/H2 mixture was passed over the cata-
stirred in an ultrasonic bath and one drop was deposited lysts. Cyclohexane was the only product of the reaction.
on a carbon-coated grid. HRTEM images were performed
with a JEOL JEM 100 CXII apparatus equipped with a top in the CO/NO/O2 reaction in temperature-programmed ex-
entry device and operating at 100 kV. periments from 298 to 773 K (light-off conditions) at a
Infrared spectra of the various samples were recorded constant heating rate of 453 Kh with a space velocity
The catalytic activities of the samples were characterized
�
1
�
1
using a Br u¨ cker IFS 66-V interferometer. After thermal from 90 to 125,000 h . Catalytic tests were performed at
treatment at a given temperature, samples were cooled to atmospheric pressure in a flow reactor. The feed stream
�
1
room temperature and flushed with argon. Spectra were (15 Lh ) consisted of a stoichiometric gas mixture con-
recorded in situ at 298 K and accumulated during 32 scans. taining 1.5 vol% CO, 0.20 vol% NO, and 0.65 vol% O2 in
�
1
The standard resolution was 4cm
.
helium, the composition of which being monitored by mass
Temperature-programmed reductions (TPR) were per- flowmeters. The catalyst sample (200 mg) was placed into a
formed on 50 mg of CeO2-ZrO2 or Rh/CeO2-ZrO2 sam- quartz reactor in an oven whose heating rate was monitored
ples placed in a conventional quartz microreactor using a by a West 2050 programmator. First a CO/NO/O2 reaction
gas mixture of 5 vol% H2 diluted in argon. The total flow up to 773 K was made before any catalytic run in order
rate was controlled by mass flow meters and was set to to get reproducible data and, more particularly, to desorb
3
� 1
2
5cm min . Experimentswere carried out from room tem- any nitrate preadsorbed ions due to the support synthesis.
�
1
perature to 1173 K at a heating rate of 5 K min . TPR cal- Continuous analysis of the gas mixture at the inlet and at
ibration was carried out by simulating the complete hydro- the outlet of the reactor was performed by a Maihak Finor
gen consumption of the hydrogen content of the reactant infrared detector for CO, a Thermoelectron chemilumines-
mixture (5 vol% H2/Ar). For that, a 100% Ar flow was sent cent NO/NOx analyzer for NO, and by gas chromatography
in the reactor, while the reactant mixture was flowing in the to control the eventual presence of N2O in the effluents.
reference branch. Such a calibration allows the correct eval- Prior to each catalytic run, fresh samples were heated at
uation of the surface area of the TPR peaks. Samples were 773 K in the reactant mixture (453 K h ), then cooled to
previously submitted to an oxidative treatment at 673 K, room temperature in helium. Experimental data were given
in order to eliminate NO3 ions originating from the sup- as conversion versus temperature plots allowing the deter-
�
1
�
port synthesis and make reproducible TPR plots. Note that mination of the temperatures of light-off (50% conversion)
such a treatment does not significantly modify the surface of reactants and the temperature at which 98% of the total
area of the samples, no more than their rhodium or chlo- conversion has been reached(T98% ).
ride contents. It was assumed that the rhodium phase con-
sisted in well dispersed Rh2O3 cristallites after the oxidizing
treatment.
RESULTS AND DISCUSSION
X-ray-induced photoelectron spectra of the reduced
The catalytic activity of CeO2-ZrO2 or rhodium sup-
CeO2-ZrO2 supports were recorded on a spectrometer
ported over CeO2-ZrO2 samples in the postcombustion re-
equipped with a non-monochromatized X-ray source (Mg-
actions was characterized after a standard thermal pretreat-
K�1, 2) and an electrostatic hemispherical analysis device
ment in the stoichiometric feed.
working in a fixed analysis transmission (FAT) mode. Bind-
ing energies were corrected from charge effects by ref-
erence to the C1s peak of carbon contamination fixed at
1
. Reducing Character of the CO/NO/O2 Reactant Mixture
E(C1) = 285 eV. The pressure in the spectrometer cham-
It was observed after pretreatment of catalysts at 773 K,
�
9
� 7
ber was approximately 5 � 10 Torr (6.6 � 10 Pa) dur- either under the stoichiometric CO/NO/O2 mixture, or un-
ing data acquisitions. Prereduced samples were transferred der pure hydrogen, that the catalytic behaviour displayed
under vacuum from the reactor into glass tubes and then by these materials in the postcombustion reactions are sim-
placed into the spectrometer chamber under argon atmo- ilar, as shown by the temperatures of light-off of CO and
sphere using a glove box.
NO and 98% conversion temperatures (Table 1).
Benzene hydrogenation measurements were carried out
It was easily demonstrated that under hydrogen at 773 K,
in a conventional system described elsewhere (32). Re- Ce4 ions of ceria after reduction reduced to Ce , as
+
3+
actions were performed at a total pressure of 760 Torr shown first by the increase of intensity of the XP signals

4
72
FAJARDIE ET AL.
TABLE 1
Catalytic Characteristics of the Reduced Samples in the CO/
NO/O2 Reactions (1) Prereduced at 773 K in the (CO/NO/O2) Re-
actant Mixture, (2) Prereduced at 773 K in H2
Rh
TloCO
TloNO
T98%
Catalysts
CeO2-ZrO2
(wt% )
�CO�
(K)
(K)
(K)
0.00
83
613
736
at 773 K
�
CO = 73.8
�NO = 96.5
(
(
1) Rh/CeO2-ZrO2
2) Rh/CeO2-ZrO2
0.30
0.30
82
80
477
456
500
480
562
553
FIG. 2. Thermoprogrammed reduction in a 5 vol% H2/Ar (heating
Note. � = conversion % , Tlo = temperature of light-off, T98% = tem-
�
1
3
� 1
rate 5 K � min. , total flow 25 cm min. ) of the fresh Rh/CeO2-ZrO2
sample (solid line) and of the corresponding unloaded support (dashed
line).
perature at which the total conversion reaches 98% , �CO� = CO conver-
sion when NO just begins to be converted.
0
0
u (EB = 904.7 eV) and v (EB = 886.2 eV) characteristics
of Ce3 in the X-ray-induced photoelectron spectra of the tion can suggest the occurrence of an hydrogen spillover
+
Ce(3d) level (33) (Fig. 1).
effect implying the formation of rhodium metal. In agree-
Then by the presence of hydrogen consumption peaks ment with already published data (29, 31) the rhodium pre-
in the TPR profiles (Fig. 2) which can be ascribed to sur- cursor is reduced in Rh(0) atoms at low temperatures (in
face and bulk ceria reduction occurring respectively at 673 the 323–523 K temperature range, probably below 373 K).
and 785 K and at lower temperatures in the presence of At this point of the TPR process, where Rh(0) is available,
rhodium (below 473 K and at 548 and 733 K). The shift of a spillover effect of H2 in the presence of rhodium metal
the TPR peaks towards the lower temperatures of reduc- occurs and favours the surface reduction of the support.
Furthermore, as soon as formed, Rh(0) atoms migrate over
the surface, locate into cationic vacancies of the support,
x+
4+
and are reoxidized as Rh with surrounding Ce cations;
4+
Zr is considered to enhance the stabilization of rhodium
at the oxidized state.
Finally by the appearance of the 2124 cm 1 infrared band
�
(
above 623 K for the bare support, and as low as 423 K
for the rhodium-containing catalyst), characteristic of the
2
2
3+
F5/2 → F7/2 transition of the Ce ion formed in both the
ceria surface and the bulk of the oxide support according
to Binet et al. (34) (Fig. 3).
The use of the reactant mixture (CO/NO/O2) instead of
H2 for the pretreatment of Rh/CeO2-ZrO2 does not mod-
ify the catalytic behaviour of this sample. This implies a
similar reducing effect of the stoichiometric mixture and of
H2 with regards to the solid. It was checked that a fresh
2
� 1
0
.3 wt% Rh/CeO2 (144 m g ) did not hydrogenate ben-
zene, whereas after treatment under H2 or under the sto-
ichiometric CO/NO/O2 reactants mixture at 723 K, the
C6H6/H2 reaction occurred. The percentages of zero-valent
metal exposed reduced, using the structure-insensitive re-
action concept (32), were quite similar whatever the nature
of the pretreating atmosphere (20.0% under H2 and 19.0%
under the reactant mixture); this unambiguouly affords ev-
idence of the reducing character of the CO/NO/O2 feed.
The catalyst reduction under the stoichiometric mixture is
necessarily due to the presence of CO which can reduce
ceria (35), without modifying the oxidation state of Zr4
as in the presence of H2). Moreover, we can state that the
Rh/CeO2-ZrO2 reduction takes place below 523 K, before
+
(
FIG. 1. XP spectra of the Ce3d signal in the CeO2-ZrO2 support be-
fore (1) and after (2) reduction in hydrogen at 773 K.

x+
ACTIVITY OF Rh SPECIES IN CO OXIDATION
473
HRTEM or XPS, due to the lack of contrast for the electron
microscopy plates and, in both cases, due to the very low
rhodium content of the samples. In contrast, it was possible
to show unambiguously the presence (or absence) of Rh(0)
on the surface of the samples before and after pretreat-
ment in the reactant mixture at 773 K by measuring their
hydrogenous activity in hydrogenation of benzene (32).
In fact, it was verified that the considered Rh/CeO2-ZrO2
(
25 wt% ZrO2) samples never display any activity in the
C6H6/H2 reaction. One can wonder about the origin of the
lack of hydrogenous activity of these samples. It cannot
be ascribed to some lack of accessibility of Rh(0) for
three main reasons: first, there is no support sintering, as
CeO2-ZrO2 reduces at 773 K in H2 with only 20% loss of its
initial surface area (36) and there is no metal encapsulation
by the reduced support at 773 K, since for lower reduction
temperatures (573 K) for which the encapsulation effect
cannot occur, the reduced sample is already inactive
in benzene hydrogenation. Furthermore as it will be
seen, hereafter, rhodium species are quite accessible for
CO/NO/O2 reactions. So, assuming that everything remain
the same when the reactant mixture is used as reducing
agent, it appears that rhodium in Rh/CeO2-ZrO2 treated
at 773 K in H2, or in the CO/NO/O2 mixture, is stabilized
FIG. 3. Infrared spectra in the 2000–2300 cm� 1 frequency interval of
the CeO2-ZrO2 support as a function of the temperature of reduction in a
vol% H2/Ar.
x+
at the catalyst surface as non hydrogenant “oxidized Rh
species,” due to a strong “rhodium-support” interactions.
5
the beginning of the CO/O2 reaction since the conversion 2. Catalytic Activity of the Reduced CeO -ZrO Support,
2
2
versus Tplots of the not pretreated sample does not differ
2
in the CO/NO/O Reactions (Fig. 4a)
from that of the sample pretreated in the reactant mixture
Let us note that the reactant mixture composition is such
that the totality of O2 in this feed cannot oxidize more than
(
Fig. 4, Table 1). In the presence of CO, the reduction of
3+
the metallic precursor (Rh ) to Rh(0) can also be pre-
dicted, concomitantly with the reduction of the support.
Unfortunately, for the Rh(0)/CeO2-ZrO2 system, the even-
tual presence of zero-valent rhodium cannot be shown by
8
7% CO. At 523 K NO desorption occurs, suggesting that
NO molecules were preadsorbed on the active sites created
during the reduction step of the sample. Simultaneously,
CO oxidation starts, involving exclusively oxygen of the
reactant mixture and in no case oxygen originating from
NO dissociation, as NO is leaving the surface. CO and NO
conversions are not complete at 673 K but can reach 100%
at higher temperatures. In the present case active sites must
be Ce3 ions and their associated oxygen vacancies.
So, concerning the reduced CeO2-ZrO2 support, the fol-
lowing reaction steps can be proposed:
+
(
a) NO from the reactant mixture adsorbs at 298 K in the
oxygen vacancies of the reduced support.
b) in the 523–573 K temperature range adsorbed NO
(
desorbs from the support surface, liberating the oxygen va-
cancies required for the dissociative adsorption of O2 which
reacts (at the dissociated state) with CO adsorbed on the
support.
(
c) at higher temperature (653 K), when the totality of the
FIG. 4. CO (dashed line) and NO (solid line) conversion curves in the
reactant oxygen is consumed in the oxidation of the quasi-
totality of CO, a new NO adsorption can occur, leading to
NO reduction and CO oxidation.
2
98–773 K temperature range of the CeO2-ZrO2 support (a) and of the
Rh/CeO2-ZrO2 sample (b), both pretreated in the CO/NO/O2 mixture at
73 K.
7

4
74
FAJARDIE ET AL.
3
. Catalytic Activity of the Reduced Rh/CeO2-ZrO2
Catalyst in the CO/NO/O2 Reactions (Fig. 4b)
a temperature of reaction lower than that on the bare sup-
port. The promoting effect of rhodium could be ascribed to
x+
an electronic tranfer from Rh to the reduced ceria sup-
In the presence of rhodium (Fig. 4b), the reaction steps
port leading to the lowering of the Ce-O bond (8) which
confers some labile character to the oxygen in ceria and
allows the CO/O and NO → N2O or N2 reactions to take
place at low temperature.
of the postcombustion reaction are quite similar to those
observed in the study of the activity of the unloaded sup-
port (Fig. 4a). However, the catalytic activity is strongly en-
hanced in the presence of metal, and appears at 393 K; i.e.
1
40 K less than that of the bare support (533 K). The CO/O2
x+
4
. Titration Method of Rh Species Active in CO/NO/O2
Reactions Present at the Surface of the
Rh/CeO2-ZrO2 Catalyst
reaction is still associated with a concomitant molecular NO
desorption starting at a temperature as low as 373 K, prior
to the beginning of the CO/NO reactions. Furthermore,
when the NO reduction occurs, low amounts of N2O are
(a) Principle. Asfound above, the surface catalyticsitesof
shown by chromatographic analysis of the effluents. This Rh/CeO2-ZrO2 in CO/NO/O2 reactions can be considered
x+
suggests that NO molecules adsorbed in the oxygen vacan- as a mixed site consisting of one Rh ion and two oxy-
cies can form either dinitrosyl groups or dimers involved in gen vacancies associated to Ce3 ions. The NO desorption
+
the mechanism of reduction of NO to N2.
peak appearing concomitantly with the beginning of the
The strong activity enhancement of rhodium-supported CO/O2 reaction is characteristic of these mixed sites. In fact,
samples confirms that the oxidized rhodium species are it seems obvious that NO molecules desorbing during this
quite accessible at the support surface and are part of the peak, desorbed from the mixed sites (one NO per oxygen
x+
active site. The occurrence of molecular NO desorption in vacancy, or Rh ). Counting the number of NO molecules
the presence of rhodium clearly indicates that rhodium is desorbing at the light-off temperature of CO brings infor-
oxidized since Rh(0) (17, 37) as well as Pt(0) (4) would mation on the number of rhodium ions implied in the mixed
have already dissociated the NO molecule at 298 K. The sites. The rhodium content in the weight of the catalyst used
existence of the molecular NO desorption steps occurring for one run being known from chemical analysis and as-
x+
simultaneously with the appearence of the CO/O2 activity suming that every Rh is accessible to the reaction, the
still suggests that NO was previously adsorbed in the oxy- stoichiometry of NO/Rh can be easily deduced from the
gen vacancies belonging to the active site, so impeding the NO/Rh ratio.
dioxygen dissociation in these vacancies and, consequently,
Desorption of NO preadsorbed on the catalyst surface
the CO oxidation. So, it appears that the active site could in experimental conditions as close as possible as those se-
present a mixed character and consists of oxidized rhodium lected in the postcombustion reaction must lead to a quan-
species and oxygen vacancies associated with Ce3 cations. titative titration of the active Rh species.
+
x+
A reaction pathway close to that proposed for the nude
support can be considered as follows:
(b) Adsorption–desorption of NO on activated Rh/
CeO2-ZrO2 catalysts
1
. Catalyst pretreatment; formation of active sites.
(
a) Desorption at 373 K of the preadsorbed molecular
Mixed sites first have to be created by contacting the glob-
ally reductive reactant mixture. The composition of the feed
x+
NO in oxygen vacancies; CO adsorption on Rh ions, and
reaction with oxygen atoms originating from the dissocia-
tion of the dioxygen of the feed in free oxygen vacancies
pertaining to the mixed sites accessible after the desorp-
tion at 373 K of the preadsorbed molecular NO in oxygen
vacancies.
�
1
for catalyst activation (total flow = 15 Lh ) was 1.5 vol%
CO/0.65 vol% O2 in helium.
Let us note, that the temperatures and intensities of the
NO desorption peaks observed at the light-off temperature
of the CO/O2 reaction for two samples pretreated either
in the CO/NO/O2 or in the CO/O2 (without NO) mixtures
are quite similar (Table 1, Fig. 4). This indicates that what-
ever the mixture used (CO/NO/O2 or CO/O2), after ther-
mal treatment at 773 K and cooling at room temperature at
(
b) As soon as dioxygen of the feed is totally consumed
by CO, NO adsorbs again in free oxygen vacancies of the
mixed sites. To explain the low concentrations of N2O in the
gas phase when NO begins to react, it can be assumed to be
produced by the decomposition of a surface dinitrosyl or
dimer (NO)2 species adsorbed on the mixed sites. NO can
be considered as a ligand for Rh , leading to a Rh (NO)2
species which decomposes immediately to N2O and then to
N2, releasing oxygen atoms in the vacancies. CO is another
2
98 K in helium, that the support surface is reduced, leading
to the formation of oxygen vacancies partly engaged in the
mixed sites. Nevertheless, in order to avoid any NO adsorp-
tion at low temperatures during the pretreatment stage of
the titration, the CO/O2 mixture was used as the support
reducing agent.
x+
x+
x+
potential ligand for Rh which can then be oxidized by
consumption of these oxygen atoms.
2
. Adsorption of NO on the activated material, at room
x+
� 1
The presence of Rh leads consequently to a very impor- temperature by flowing 0.2 vol% NO in helium (15 Lh ).
tant catalytic effect, since it allows the reaction to occur at NO adsorption was proceeded by flowing 0.2 Vol% NO/He

x+
ACTIVITY OF Rh SPECIES IN CO OXIDATION
475
TABLE 2
NO/Rh Evaluation as a Function of the Rh Content
in Rh/CeO2-ZrO2 catalysts
Rh (wt% )
NO/Rh
0.30
1.97
0.24
1.82
0.18
1.92
0.16
2.20
0.11
1.96
�
�
Note. NO determined from NO adsorption–desorption measure-
ments, and taking only into consideration the number of NO molecules
desorbing at the light-off temperature of the CO/O2 reaction. Rh = total
rhodium content in the sample.
surrounded by two oxygen vacancies of the reduced sup-
port and associated with Ce3 cations.
+
Nevertheless, note that the preceding results do not give
FIG. 5. Thermodesorption in the reducing (CO/O2) mixture and the the oxidation state of rhodium in the mixed site. After
2
98–773 K temperature range of NO preadsorbed at 298 K on the surface Howitt et al. (38) Rh3 can be excluded, as it is inactive
+
of the Rh/CeO2-ZrO2 sample previously reduced in the (CO/O2) mixture.
in the postcombustion reactions. Elsewhere, Soria et al.
(
39) by EPR spectroscopy of rhodium ions in Rh/CeO2
to the stabilization of the initial NO concentration at the
outlet of the reactor.
2+
systems reduced at 773 K showed the presence of Rh
ions.
3
. Thermodesorption of NO adsorbed on the Rh/
In order to get further information about the value of
the rhodium oxidation state in the mixed sites, we stud-
ied by infrared spectroscopy the adsorption of CO on the
Rh/CeO2-ZrO2 sample reduced at 773 K in H2 (Fig. 6).
Despite the poor resolution of the IR bands of CO ad-
sorbed on rhodium, two couples of bands were shown at
�
1
CeO2-ZrO2 surface by flowing (15 Lh ) 1.5 vol% CO/
0
.65 vol% O2/He.
After adsorption of NO on Rh/CeO2-ZrO2 at 298 K and
flushing of the reactor by helium, the sample was submitted
to a thermodesorption from 298 to 773 K in flowing CO/O2
mixture with the same composition as for the CO/NO/O2
reactions (Fig. 5). Two NO desorption peaks can be seen.
It is possible to discriminate the contribution of NO, either
adsorbed on the support or located in the mixed sites. The
first one concerns NO molecules desorbing between 298
and 423 K, before the beginning of the CO/O2 reaction;
consequently, these molecules do not interact with the ac-
tive sites, but are weakly adsorbed on the support surface.
The second one corresponds to NO desorbing at the light-
off temperature of the CO/O2 reaction. The measurement
of the surface area of this peak leads to the accurate num-
ber of NO molecules retained by the actives sites once the
first peak has been completely eluted.
�
1
� 1
2
063–2008 cm and 2046–1975 cm . These bands could
be identified as those ascribed to the vibrations of C-O at
2
� � 1
1
040–2030 cm and 2110–2075 cm in the gem carbonyl
1+
Rh (CO)2 species, more or less shifted, depending on the
degree of interaction of the oxygen atoms of the CO lig-
ands with the reduced support. So, it is likely that the ox-
idation state of rhodium in the mixed sites could be Rh1
state.
+
Counting the number of NO molecules desorbing be-
tween 423 and 523 K (second desorption peak) (Fig. 5) leads
to a NO/Rh ratio close to 2. In order to check this stoichiom-
etry, the NO/Rh ratio was determined for Rh/CeO2-ZrO2
samples with different total rhodium contents (Table 2). It
can be seen that the NO/Rh ratio remains fairly constant
with a mean value of 2. This result not only confirms the ra-
tio NO/Rh, but also confirms that all the rhodium is atomi-
cally dispersed at the surface of the support in an oxidized
state accessible for the CO/NO/O2 reactions. Furthermore,
it shows that the intermediate reaction must be either a
dinitrosyl or a dimer (NO)2.
�
1
FIG. 6. Infrared vibration bands (in the 2300–1700 cm frequency
interval) of CO adsorbed on the Rh/CeO2-ZrO2 sample reduced at 773 K
in H2 during 2 h as a function of the partial pressure of CO: (1) 0.4 kPa,
So, it can be concluded that the active site in the
(
CO/NO/O2) reactions at the surface of the activated
x+
Rh/CeO2-ZrO2 catalysts consists of dispersed Rh ions (2) 0.8 kPa, (3) 1.6 kPa, (4) 4.6 kPa, (5) 10.2 kPa, (6) 13.5 kPa.

4
76
FAJARDIE ET AL.
4. Burch, R., and Watling, T. C., in “First International Congress on
CONCLUSION
Catalysis and Automotive Pollution Control, Brussels, April 1997,”
Vol. 1, p. 69.
. Oh, S. H., and Eickel, C. C., J. Catal. 128, 526 (1991).
. Oh, S. H., Fisher, G. B., Carpenter, J. E., and Goodman, D. W., J. Catal.
The catalytic activity of Rh/CeO2-ZrO2 samples with re-
5
6
gard to the reducing stoichiometric CO/NO/O2 reaction
mixture was studied. It was shown that the catalytic ac-
tivity of these samples was due to the presence of mixed
100, 360 (1986).
7
. Serre, C., Garin, F., Belot, G., and Maire, G., J. Catal. 141, 9 (1993).
8. Serre, C., Garin, F., Belot, G., and Maire, G., J. Catal. 141, 1 (1993).
. Zafiris, G. S., and Gorte, R. J., J. Catal. 143, 86 (1993).
0. Mannila, P., Salmi, T., Haario, H., Luoma, M., H a¨ rk o¨ nen, M., and
Sohlo, Appl. Catal. B 7, 179 (1996).
x+
1+
sites consisting of oxidized Rh species (Rh ) stabilized
3+
9
with two oxygen vacancies (and associated to Ce cations)
of the reduced CeO2-ZrO2 support.
1
A titration method of these active mixed sites was devel-
oped using NO adsorption–desorption measurements on
1
1
1. Oh, S. H., and Eickel, C. C., J. Catal. 112, 543 (1988).
2. Zafiris, G. S., and Gorte, R. J., J. Catal. 139, 561 (1993).
the activated surface in experimental conditions as close as 13. Gorte, R. J., Schmidt, L. D., and Gland, J. L., Surf. Sci. 109, 367 (1981).
1
4. Wickham, D. T., Banse, B. A., and Koel, B. E., Surf. Sci. 243, 83 (1991).
possible to the catalytic ones; it was deduced that the stoi-
x+
15. Nieuwenhuys, B. E., Surf. Sci. 126, 307 (1983).
1
1
chiometry of the mixed site consisted of one Rh and two
6. Bowker, M., Guo, Q., and Joyner, W., Surf. Sci. 257, 33 (1991).
7. Kaspar, J., de Leitenburg, C., Fornasiero, P., Trovarelli, A., and
Graziani, M., J. Catal. 146, 136 (1994).
3+
oxygen vacancies associated to Ce ions. The promoting
x+
effect of Rh species on the catalytic activity in the post-
combustion reactions of these samples was clearly shown 18. Oh, S. H., and Eickel, C. C., J. Catal. 128, 526 (1990).
1
2
2
2
2
9. Cho, B. K., J. Catal. 148, 697 (1994).
by comparison with the reactivity of the unloaded CeO2-
ZrO2 support. This one only displayed catalytic activity for
temperatures 150 K higher than that of the sample contain-
ing rhodium. Rh/CeO2-ZrO2 was found to be still active in
three-way catalysis, even after drastic ageing. The ageing
resistant character of these samples working alternatively
under oxidative and reducing conditions could be now in-
terpreted in terms of active sites consisting of both Rh(0)
0. Cho, B. K., Shanks, B. H., and Bailey, J. E., J. Catal. 115, 486 (1989).
1. Iizuka, T., and Lunsford, J. H., J. Mol. Catal. 8, 391 (1980).
2. Liang, J., Wang, H. P., and Spicer, L. D., J. Phys. Chem. 89, 5840 (1985).
3. Martinez-Arias, A., Soria, J., Conesa, J. C., Seoane, X. L., Arcoya, A.,
and Cataluna, R., J. Chem. Soc. Faraday Trans. 91, 1679 (1995).
4. Kung, M. C., and Kung, H., Catal. Rev. 27, 425 (1985).
5. Oh, S. H., J. Catal. 124, 477 (1990).
6. Cho, B. K., J. Catal. 131, 74 (1991).
7. Oh, S. H., and Carpenter, J. E., J. Catal. 101, 114 (1986).
8. Harrison, B., Diwell, A. F., and Hallet, C., Plat. Metals Rev. 32, 73
2
2
2
2
2
x+
and Rh species (40).
(
1988).
2
3
9. Fornasiero, P., Di Monte, R., Rao, G. R., Kaspar, J., Meriani, S.,
Trovarelli, A., and Graziani, M., J. Catal. 151, 168 (1995).
0. Balducci, G., Kaspar, J., Fornasiero, P., Graziani, M., Saiful Islam, M.,
and Gale, J. D., J. Phys. Chem. B 101, 1750 (1997).
ACKNOWLEDGMENTS
The authors acknowledge CNRS and Rh oˆ ne Poulenc for their financial
grant. The technical assistance of M. Lavergne in the HRTEM measure- 31. Fornasiero, P., Kaspar, J., and Graziani, M., J. Catal. 167, 576 (1997).
ments and of G. Dufour in the XPS study are also gratefully acknowledged.
32. Fajardie, F., Temp e` re, J-F., Dj e´ ga-Mariadassou, G., and Blanchard, G.,
J. Catal. 163, 77 (1996).
3
3. Ramaroson, E., Temp e` re, J-F., and Guilleux, M-F., Vergand, F., Roulet,
H., and Dufour, G., J. Chem. Soc. Faraday Trans. 88, 1211 (1992).
4. Binet, C., Badri, A., and Lavalley, J. C., J. Phys. Chem. 98, 6392 (1994).
REFERENCES
3
1
2
3
. Taylor, K. C., in “Automotive Catalytic Converters, in Catalysis: Sci- 35. Laachir, A., Perrichon, V., Bernal, S., Calvino, J. J., and Cifredo, G. A.,
ence and Technology” (J. R. Anderson and M. Boudart, Eds.), Vol. 5.
Springer-Verlag, Berlin, 1984.
. Cooper, B. J., Evans, W. D. J., and Harrisson, B., in “Catalysis and
Automotive Pollution control” (A. Crucq and A. Frenet, Eds.), Vol. 1,
p. 117. Elsevier, Amsterdam, 1987.
J. Mol. Catal. 89, 391 (1994).
36. Fajardie, F., thesis, Universit e´ P.&M. Curie, Paris, 1996.
37. Peden, C. H. F., Belton, D. N., and Schmieg, S. J., J. Catal. 155, 204
(1995).
38. Howitt, C., Pitchon, V., and Maire, G., J. Catal. 154, 47 (1995).
. Leclercq, G., Dathy, C., Mabilon, G., and Leclercq, L., “Stud. Surf. 39. Soria, J., Martinez-Arias, A., Coronado, J. M., and Conesa, J. C., Appl.
Sciences and Catalysis, Capoc II” (A. Crucq, Ed.), Vol. 71, p. 181.
Elsevier, Amsterdam, 1991.
Surf. Sci. 70, 245 (1993).
40. Salin, L., thesis, Universit e´ P.&M. Curie, Paris, 1997.