Influence of the nature of the noble metal on the lean C3H6-assisted decomposition of NO on Ce0.68Zr0.32O2-supported catalysts

Article abstract of DOI:10.1016/j.molcata.2006.01.001

The Selective Catalytic Reduction (SCR) of NO_x assisted by C₃H₆ is investigated on Ce_0.68Zr_0.32O₂ (CZ)-supported Platinum Group Metals (PGMs: Pd, Rh, Pt) catalysts. This study shows that th

Full text of DOI:10.1016/j.molcata.2006.01.001

Journal of Molecular Catalysis A: Chemical 249 (2006) 71–79
Influence of the nature of the noble metal on the lean C3H6-assisted
decomposition of NO on Ce0.68Zr0.32O2-supported catalysts
Cyril Thomasa,∗, Olivier Gorcea,1, Franc¸oise Villainb,c, Gerald Djega-Mariadassou
a
´
´
a Laboratoire de Re´activite´ de Surface, UMR CNRS 7609, Universite´ Pierre et Marie Curie, 4 Place Jussieu, Case 178, 75252 Paris Cedex 05, France
b Laboratoire de Chimie Inorganique et Mate´riaux Mole´culaires, UMR CNRS 7071, Universite´ Pierre et Marie Curie,
4 Place Jussieu, 75252 Paris Cedex 05, France
c Laboratoire pour l’Utilisation du Rayonnement Electromagne´tique (LURE), Centre Universitaire Paris-Sud BP34, 91898 Orsay Cedex, France
Received 15 November 2005; accepted 5 January 2006
Available online 8 February 2006
Abstract
The Selective Catalytic Reduction (SCR) of NOx assisted by C3H6 is investigated on Ce0.68Zr0.32O2 (CZ)-supported Platinum Group Metals
(PGMs: Pd, Rh, Pt) catalysts.
This study shows that the order of reactivity of the PGMs supported on CZ is: Pd ꢀ Pt ≥ Rh. The performances of the most active CZ-supported
catalysts are comparable to those reported for SiO2- or Al2O3-supported PGMs catalysts in terms of N2 production. In contrast, the CZ-supported
catalysts are much more selective to N2 than those supported on SiO2 or Al2O3, which makes them of particular interest. This difference of
selectivity to N2 of the CZ-supported catalysts suggests that the mechanism of the SCR of NOx assisted by C3H6 proposed for Pd/CZ catalysts
[C. Thomas, O. Gorce, C. Fontaine, J.-M. Krafft, F. Villain, G. Dje´ga-Mariadassou, Appl. Catal. B: Environ. 63 (2006) 201] is also valid for Pt or
Rh/CZ catalysts. In this mechanism, ad-NO2 species react with C3H6 to form ad-RNOx compounds that further decompose to NO and oxygenates
(CxHyOz). The role of these oxygenates is to reduce the support to provide the sites responsible for NO decomposition to N2 (Ce3+ surface cations).
The much higher activity of the Pd-based catalysts in the SCR of NOx assisted by C3H6 might be attributed to the greater ability of the PdOx
phase to adsorb NO2. This property might be assigned to the lower interaction of the PdOx phase with CZ compared with that of the PtOx or RhOx
phases.
© 2006 Elsevier B.V. All rights reserved.
Keywords: SCR of NOx; Platinum Group Metals; C3H6; Ceria–zirconia; Catalysts
1. Introduction
als (PGMs) have received much interest in the SCR of NOx [2].
However, the rather poor stability of metal ion-exchanged zeo-
lites and the deactivation of base metal oxides in the presence of
water in the feed constitute the main drawbacks of these cata-
lysts to be used under real automotive exhausts [2]. On the other
hand, supported PGMs exhibit rather low selectivity to N2, N2O
being formed in substantial amounts [3]. Although not regu-
lated, it must be emphasised that this pollutant is a well-known
powerful greenhouse gas [4,5].
PGMs have mostly been supported on SiO2 or Al2O3 [3]
and very little attention has been paid to other supports such
as ceria–zirconia. To our knowledge, only three studies have
been reported for such catalysts for the SCR of NOx by
propene by: (i) Orlik et al. on Rh/CeZrO2 [6], (ii) Centi et
al. on Pt/Ce0.50Zr0.50O2/Al2O3 [7] and (iii) Liotta et al. on
Pt/Ce0.50Zr0.50O2 [8]. Recently, we reported on the perfor-
mances of Pd/Ce0.68Zr0.32O2 (Pd/CZ) catalysts in the SCR of
Because of environmental concerns, the reduction of NOx
from lean exhausts sustains strong interest in the academic and
industrial communities. Despite extensive investigations per-
formed on this challenging issue, the catalyst able to comply
with NOx emission regulations on lean–burn internal combus-
tion engines has still to be discovered [1]. A large number of
catalysts have been evaluated in the Selective Catalytic Reduc-
tion (SCR) of NOx by hydrocarbons [2,3]. Metal ion-exchanged
zeolites, base metal oxides and supported Platinum Group Met-
∗
Corresponding author. Tel.: +33 1 44 27 36 30; fax: +33 1 44 27 60 33.
E-mail address: cthomas@ccr.jussieu.fr (C. Thomas).
Present address: Renault sas, Centre Technique de Lardy, 1 alle´e Cornuel,
1
91510 Lardy, France.
1381-1169/$ – see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.molcata.2006.01.001

72
C. Thomas et al. / Journal of Molecular Catalysis A: Chemical 249 (2006) 71–79
NOx assistedbyC3H6 [9]. Thesecatalystsexhibitedmuchhigher
selectivity to N2 than Pd0/SiO2. This difference was assigned to
the occurrence of a deNOx mechanism on Pd/CZ different from
that of the dissociation of NO on zero-valent PGMs atoms [10].
This work reports on the influence of the nature of the PGMs
(Pd, Rh, Pt) supported on CZ in the SCR of NOx assisted by
C3H6. The validity of the proposed mechanism for Pd/CZ cata-
lysts [9] to other PGMs is discussed.
Pd) was prepared by subsequent incipient wetness impregna-
tions of the support by aqueous solutions of RhCl3·3H2O and
PdCl2·3H2O (Johnson Matthey). After impregnation, the cata-
lyst was aged for 2 h and dried in air at 120 ◦C for 3 h.
2.1.6. Rh-Pt/CZ
The preparation of the Rh(0.44)Pt(0.91)/CZ was similar to
that of Rh(0.44)Pd(0.54)/CZ (Section 2.1.4), except that an
aqueous solution of H2PtCl6·6H2O (Johnson Matthey) was used
instead of that of PdCl2 to carry out the impregnation of a Rh/CZ
catalyst.
2. Experimental
2.1. Catalyst synthesis
2.2. Catalyst characterisation
2.1.1. Ceria–zirconia (Ce0.68Zr0.32O2, CZ)
Metal and chlorine contents were determined by chemical
analyses (CNRS—Vernaison). Chlorine contents were about
1 wt% for all CZ-supported catalysts.
The CZ solid solution was provided by Rhodia. The CZ was
added to a chloridric acid solution (pH 1.9) prepared by adding
HCl to distilled water. After ageing for 2 h under vigorous stir-
ring, CZ was filtered and washed with distilled water before
being dried in air at 120 ◦C for 3 h. The samples were then
reduced under pure H2 (Air Liquide) for 2 h at 500 ◦C with a
flow rate of 40 mLNTP min−1 of H2 per gram of catalyst and
finally gently exposed to air at RT. Such a preparation was per-
formed for the sake of comparison with noble metal catalyst
synthesis.
The specific surface areas were determined by physisorption
of N2 at 77 K using a Quantasorb Jr. dynamic system equipped
with a thermal conductivity detector (TCD). The specific surface
areas were calculated using the BET method. The specific sur-
face area of the CZ-supported catalysts was about 100 m2 g−1
before and after testing, whereas that of Rh(0.40)Pd(0.54)/SiO2
was about 50 m2 g−1
.
2.2.1. Determination of the percentage of exposed
2.1.2. Rh/CZ
zero-valent metal atom (PEM0)
The CZ-supported rhodium catalyst (0.40 wt% Rh) was
prepared by anionic exchange from an acidic solution of
RhCl3·3H2O with the support [11]. The acidic solution of
RhCl3·3H2OwasaddedtoanaqueoussuspensionofCZ(pH1.9)
prepared as described in Section 2.1.1. After being exchanged,
Rh(0.40)/CZ was filtered and washed with distilled water before
being dried in air at 120 ◦C for 3 h. This sample was then reduced
under pure H2 (Air Liquide) for 2 h at 500 ◦C with a flow rate of
40 mLNTP min−1 of H2 per gram of catalyst and finally gently
exposed to air at RT.
With the well-known reducibility of the CZ support [12],
the determination of the percentage of exposed zero-valent
metal atom was done by means of hydrogenation reactions
[13,14]. Benzene hydrogenation was used to titrate exposed
zero-valent rhodium or platinum atoms, whereas propene hydro-
genationwasusedtotitrateexposedzero-valentpalladiumatoms
[14].
Briefly, prior to benzene hydrogenation, the catalyst sample
(50 mg deposited on sintered glass of a pyrex reactor) was heated
in flowing H2 (100 mLNTP min−1) at atmospheric pressure with
a heating rate of 3 ◦C min−1 up to 500 ◦C and was kept at this
temperature for 2 h. After cooling to 50 ◦C under H2, the reaction
was started. The partial pressure of benzene was 51.8 Torr, and
the total flow rate was 107 mLNTP min−1 with H2 as balance.
The procedure used for propene hydrogenation was similar
to that of benzene hydrogenation except that the sample weight
was 30–60 mg, the reduction temperature was 300 ◦C and the
reaction temperature was −78 ◦C.
2.1.3. Pd/CZ
As already described [9], the two ceria–zirconia-supported
Pd catalysts (0.54 and 0.89 wt% Pd) were prepared by incipi-
ent wetness impregnation of the chlorinated CZ by an aqueous
solution of PdCl2·3H2O (Johnson Matthey). In these particular
cases, the impregnations were carried out after exposure of the
reduced CZ to air at RT (Section 2.1.1). After impregnation, the
catalysts were aged for 2 h and dried in air at 120 ◦C for 3 h.
The composition of the effluents was analysed by means
of on-line gas chromatographs (Hewlett-Packard 5890, FID)
equipped with a paraffins-olefins-naphtenes-aromatics (PONA,
HP, 50 m long, 0.20 mm inner diameter, 0.5 ␮m film thickness)
or a CP-Al2O3/KCl (Chrompack, 50 m long, 0.32 mm inner
diameter, 5 ␮m film thickness) capillary columns for benzene
or propene hydrogenation, respectively. The only detected prod-
ucts were cyclohexane and propane.
2.1.4. Rh-Pd/CZ
After exposure of a reduced Rh/CZ catalyst to air at RT, the
Rh(0.44)Pd(0.54)/CZ catalyst was prepared by an impregnation
procedure of Pd identical to that used for the preparation of the
Pd/CZ catalysts (Section 2.1.3). After impregnation, the catalyst
was aged for 2 h and dried in air at 120 ◦C for 3 h.
From the initial reaction rates, the numbers of exposed zero-
valent metal atoms were calculated according to turnover rates
of 0.22 and 0.40 s−1 for benzene and propene hydrogenation
reactions carried out at 50 and −78 ◦C, respectively [14].
2.1.5. Rh-Pd/SiO2
For comparison with Rh-Pd/CZ, a silica (Degussa, Aerosil
50)-supported Rh-Pd catalyst (0.40 wt% Rh and 0.44 wt%

C. Thomas et al. / Journal of Molecular Catalysis A: Chemical 249 (2006) 71–79
73
2.2.2. XANES
Catalytic experiments were carried out with C3H6 as reduc-
tant. Typically, the composition of the C3H6–NO–O2 reacting
mixture was 1900 ppm C3H6, 340 ppm NO and 8% O2 in N2.
These reactants were fed from independent gas cylinders (Air
Liquide) of N2 diluted gas mixtures.
The reactor outflow was continuously analysed using the
combination of four different detectors. An Eco Physics CLD
700 AL chemiluminescence NOx analyser (for NO and total
NOx (i.e. NO + NO2)) allowed the simultaneous detection of
both NO and NOx. Two Ultramat 6 IR analysers were used to
monitor N2O, CO and CO2. A FID detector (Fidamat 5A) was
used to follow the concentration of hydrocarbons. We checked
that under our experimental conditions, CO and CO2 had a neg-
ligible response on the N2O IR analyser, whereas that of C3H6
was significant. C3H6 contribution to the N2O signal was taken
into account to calculate the “true” N2O concentration ([N2O])
as follows (Eq. (1)):
X-ray absorption measurements were carried out using syn-
chrotron radiation of the XAS13 station at LURE (Orsay,
France) on-line D42 (May 3–4, 2001). Fluorescence yield spec-
tra of Rh(0.44)Pd(0.54)/CZ were recorded at RT using a Ge
solid detector (Eurisys, seven elements) combined with a mul-
tichannel analyser to select the Rh or Pd K␣ fluorescences,
whereas those of the reference compounds (Rh and Pd foils,
25 and 20 ␮m thick, respectively, Rh2O3 (>99.99% pure, Lan-
caster) and PdO (>99% pure, Lancaster)) were collected in
the transmission mode. The incident beam was monochroma-
tised using a Ge(4 0 0) double monochromator. XANES spectra
were collected at the Rh and Pd K edges with a sampling step
of 2.0 eV/point from 23,150 to 23,370 eV and from 24,300 to
24,500 eV, respectively, and an integration time of 10 and 2 s in
fluorescence and transmission modes, respectively. The energy
was calibrated using the Pd foil. XANES spectra were obtained
from Rh(0.44)Pd(0.54)/CZ either after oxidation or reduction
treatments under flowing air or H2 (100 mLNTP min−1) for 2 h
at 500 ◦C. To avoid the exposure of the reduced sample to O2,
the catalyst was then transferred into a UV cell (Suprasil quartz,
5 cmlong, 1 cmwideand1 cmhigh)thatwasfinallysealedunder
vacuum (5 × 10−5 Torr). In a previous study [9], we checked
that the high purity silica glass did not interfere with incident
and fluorescent beams. The cell was set to 45◦ with respect to
the incident RX beam and the fluorescence detector. The extrac-
tion of the data was done with Michalowicz’s software package
[15,16].
[N2O]0 × [C3H6]meas.
[N2O] = [N2O]meas.
−
(1)
[C3H6]inlet
where [N2O]0, [C3H6]inlet, [C3H6]meas. and [N2O]meas. are the
concentrations of N2O due to the contribution of the inlet con-
centration of C3H6, the inlet concentration of C3H6, C3H6 and
N2O concentrations measured in the course of the reaction,
respectively. As N2 was used as balance, the conversion of NOx
to N2 was calculated assuming 100% nitrogen mass balance (Eq.
(2)):
Conversion of NOx to N2 (%)
2.3. Catalytic runs
[NOx]inlet − ([NOx]outlet + 2[N2O])
=
(2)
[NOx]inlet
Prior to catalytic runs, the samples were calcined in situ
in dry air at 500 ◦C (4 ◦C min−1) for 2 h with a flow rate of
500 mLNTP min−1 g−1 catalyst. Steady-state and temperature-
programmed desorption (TPD) experiments were carried out.
After being contacted with the appropriate gas mixture at RT,
TPD experiments were carried out from RT to 500 ◦C with a
heating rate of 10 ◦C min−1. Before TPD experiments, the cata-
lyst samples were flushed in N2 to remove physisorbed species
from the adsorption mixture.
where [NOx]inlet and [NOx]outlet are the concentrations of NOx
at the inlet and at the outlet of the reactor, respectively.
3. Results
3.1. Catalyst characterisation
3.1.1. Determination of the percentage of exposed
TheseexperimentswerecarriedoutinaU-typequartzreactor.
The sample (0.2 g) was held between plugs of quartz wool and
the temperature was controlled through a WEST 4000 tempera-
turecontroller using a Ktypethermocouple. Reactant gaseswere
fed from independent mass flow controllers (Brooks 5850E).
The total flow was 250 mLNTP min−1 to which corresponds a
zero-valent metal atoms (PEM0)
Table 1 lists the percentage of exposed zero-valent metal
atoms of CZ-supported Rh and Pd catalysts. As reported in
previous studies from our group [11,17], CZ-supported Rh cat-
alysts scarcely exhibit exposed zero-valent rhodium atoms. On
the other hand, the existence of exposed zero-valent palladium
atoms, under the present experimental conditions of propene
hour space velocity (HSV) of 112,500 h−1
.
Table 1
Determination of the percentage of exposed zero-valent metal atoms (PEM0) over Ce0.68Zr0.32O2 (CZ) supported catalysts by means of benzene and propene
hydrogenation reactions [14]
Catalysts
Rh(0.40)/CZ
Pd(0.54)/CZa
Pd(0.89)/CZa
Rh(0.44)Pd(0.54)/CZ
Rh(0.44)Pt (0.91)/CZ
Rh0 (%)
Pd0 (%)
Pt0 (%)
2
–
–
–
17
–
–
16
–
0
18
–
0
–
3
a
Data reported from Ref. [9] for comparison purpose.

74
C. Thomas et al. / Journal of Molecular Catalysis A: Chemical 249 (2006) 71–79
Fig. 2. Steady-state catalytic conversion of NO to NO2 in the course of the
NO–O2 reaction (340 ppm–8%, balance N2): (♦) Ce0.68Zr0.32O2 (CZ), (ꢁ)
Rh(0.40)/CZ and (ꢀ) Rh(0.44)Pd(0.54)/CZ.
turbations induced on the metal by the reduced support, as
also reported by Bernal et al. in the case of Pt/CeO2 cata-
lysts [19]. On the other hand, the comparison of the shape of
the XANES spectra of the bimetallic sample at the Pd K edge
either after calcination or after reduction correspond fairly well
to those of the PdO and the Pd metal references (Fig. 1b),
respectively. As for Pd(0.54)/CZ [9], this suggests that Pd is
present as PdO clusters in the CZ-supported Pd catalyst. Inter-
estingly, as identical comments stand for the bimetallic catalyst
and the monometallic ones, it might be concluded that the metals
do not seem to interact in the Rh(0.44)Pd(0.54)/CZ bimetallic
catalyst.
3.1.3. NO oxidation
Fig. 1. XANES spectra (a) at the Rh K edge for Rh2O3 (᭹), the Rh foil (ꢁ) and
Rh(0.44)Pd(0.54)/CZ (calcined in air at 500 ◦C (- - -) or reduced in H2 at 500 ◦C
(—)); (b) at the Pd K edge for PdO (ꢀ), the Pd foil (ꢂ) and Rh(0.40)Pd(0.54)/CZ
(calcined in air at 500 ◦C (- - -) or reduced in H2 at 500 ◦C (—)).
The NO oxidation reaction, which has been assumed of key
importance in lean deNOx catalysis [9,20–23], was investigated
over different catalysts under steady-state conditions. Fig. 2
shows that NO oxidation is the greatest at 400 ◦C and decreases
at higher temperatures because of thermodynamic limitations. It
must be emphasised that the addition of PGMs does not enhance
NO oxidation, as was also the case for Pd(0.89)/CZ [9], and this
reaction is only catalysed by CZ.
hydrogenation, is obvious. The percentage of exposed zero-
valent palladium atoms is close to 17% for all catalysts. As
already suggested [9], it is therefore very likely that after the
calcination step, Pd was present as PdO.
The fairly low benzene hydrogenation activity of
Rh(0.44)Pt(0.91)/CZ is assigned to Pt only (Table 1).
The synthesis of this catalyst was carried out starting with the
same Rh/CZ catalyst precursor as that used for the preparation
of Rh(0.44)Pd(0.54)/CZ, the rhodium contribution of which to
benzene hydrogenation is null (Table 1).
3.1.4. Adsorption of NO–O2–TPD in N2 or in C3H6–O2
The TPD profiles in N2 or C3H6–O2 obtained after exposure
of CZ and Pd(0.89)/CZ to NO–O2 (340 ppm–8% O2 in N2) have
been described in details in previous articles [9,23]. The TPD
traces found on Rh(0.40)/CZ and Rh(0.44)Pd(0.54)/CZ (not
shown) are similar to those of CZ and Pd(0.89)/CZ [9], respec-
tively. For all catalysts, NO2 is the major NOx desorbed species
of the TPD in N2, whereas NO is the chief product of the TPD
in C3H6–O2 (Table 2). The amounts of adsorbed NOx species
correspond well to those desorbed for the TPD in N2, whereas
the nitrogen balance of the TPD in C3H6–O2 indicates that part
of the stored NOx transform to N2 in the presence of C3H6 in
the gas feed (Table 2). The interaction of pre-adsorbed NOx with
C3H6 was revealed on Rh(0.40)/CZ and Rh(0.44)Pd(0.54)/CZ
(not shown), as was the case with CZ and Pd(0.89)/CZ [9].
However, this interaction occurred up to higher temperatures
for Rh(0.44)Pd(0.54)/CZ than for Rh(0.40)/CZ (not shown), as
3.1.2. XANES measurements
XANES spectra of Rh(0.44)Pd(0.54)/CZ after calcination or
reduction are shown in Fig. 1. The shape of the XANES spec-
trum at the Rh K edge after calcination of the sample closely
resembles that of the Rh2O3 oxide reference (Fig. 1a). This
suggests similar oxidation state and local geometry of Rh to
those of Rh2O3. Surprisingly, the shape of the XANES spec-
trum of the bimetallic catalyst after reduction differs markedly
from that of the Rh foil (Fig. 1a). Such a behaviour, which is
also observed on Rh(0.40)/CZ [18], is attributed to the forma-
tion of electro-deficient Rh clusters owing to electronic per-

C. Thomas et al. / Journal of Molecular Catalysis A: Chemical 249 (2006) 71–79
75
Table 2
Amounts of adsorbed or desorbed NOx (NO + NO2) and desorbed N2O in the course of the TPD experiments after exposure of the catalysts to NO–O2 (340 ppm–8%)
in N2
Catalysts
Experiment
Amounts of adsorbed (ads.) or desorbed (des.) species (10−5 mol g−1
)
a
b
TNO
NOx ads.
NOdes.
NO2 des.
NOx des.
N2Odes.
N balancec
2
CZd
TPD in N2
TPD in C3H6–O2
115
123
136
135
54.8
48.2
8.1
19.0
46.5
7.8
54.6
26.8
0.0
5.8
0.1
4.9
Rh(0.40)/Z
Pd(0.89)/CZd
TPD in N2
TPD in C3H6–O2
53.6
63.8
9.9
26.1
41.0
9.2
50.8
35.2
0.0
8.5
1.4
5.8
TPD in N2
TPD in C3H6–O2
47.6
51.8
6.2
20.3
36.4
5.1
42.7
25.4
0.0
5.4
2.5
7.8
Rh(0.44)Pd(0.54)/CZ
TPD in N2
TPD in C3H6–O2
45.5
57.8
7.5
23.9
37.4
8.2
44.9
32.1
0.0
6.4
0.3
6.5
a
Temperature (◦C) of the maximum of the first NO2 desorption peak from the TPD in N2 after adsorption of NO–O2.
The moderate fluctuation of the amount of adsorbed NOx is related to slightly different times of exposure and not to differences in the starting conditions.
N balance calculated with respect to N2 equivalent.
b
c
d
Data reported from Ref. [9] for comparison purpose.
was found for Pd(0.89)/CZ compared with CZ [9]. Similar to
Pd(0.89)/CZ, the first NO2 desorption peak of the TPD in N2
of Rh(0.44)Pd(0.54)/CZ is shifted to higher temperatures com-
pared with those of CZ and Rh(0.40)/CZ (Table 2).
selectivity to N2 of Rh(0.40)Pd(0.54)/SiO2 is much lower and
is of about 50%. Supported Pd catalysts are the most efficient
in lean deNOx, whereas Rh(0.40)/CZ is the less active of the
CZ-supported PGMs catalysts.
It is worthwhile to report that the sum of the SCR of NOx
to N2 of Rh(0.40)/CZ and Pd(0.54)/CZ corresponds well to that
of Rh(0.44)Pd(0.54)/CZ (Fig. 4a). Regarding the bimetallic cat-
alysts, the deNOx activity of Rh(0.44)Pd(0.54)/CZ (Fig. 3e) is
higher than that of Rh(0.44)Pt(0.91)/CZ (Fig. 3f), which clearly
shows that Pt is less active than Pd in the SCR of NOx when
supported on CZ. The picture is less obvious for the comparison
of Rh and Pt as the deNOx activity of a monometallic Pt catalyst
was not investigated in the present work. Nevertheless, the com-
parison of the deNOx activity of Rh(0.40)/CZ and the difference
of that of Rh(0.44)Pt(0.91)/CZ with that of Rh(0.40)/CZ may
be informative on that particular point. Fig. 4b shows that the
activity of Pt, deduced from the above-mentioned difference, is
close to that of Rh but takes place at lower temperature. For CZ-
supported PGMs catalysts, the catalytic activities for the SCR
of NOx follow the order: Pd ꢀ Pt ≥ Rh.
3.1.5. Steady-state C3H6–NO–O2 reaction
Fig. 3 shows the performances of the catalysts for the SCR
of NOx under steady-state conditions with stepwise increase
of the temperature. The temperature of maximum of N2 for-
mation and the corresponding NOx and C3H6 conversions are
listed in Table 3. The main effect of the introduction of PGMs
is to enhance C3H6 activation as the T50 C3H6 (tempera-
ture at 50% conversion of C3H6) of the CZ-supported PGMs
catalysts decrease significantly compared with CZ (Table 3).
It must also be emphasised that C3H6 oxidation is steeper
on Rh(0.40)Pd(0.54)/SiO2 than on the CZ-supported catalysts
(Fig. 3).
Except Rh(0.40)/CZ, which shows low deNOx activity com-
parable to that reported by Orlik et al. on a similar catalyst
[6], CZ-supported PGMs catalysts exhibit much higher deNOx
activity than CZ. The selectivity to N2 is, however, very compa-
rable for all CZ-based materials and remains higher than 80%
at maximum NOx to N2 conversion (Table 3). Conversely, the
Finally, the introduction of 1.7 wt% water in the gas feed (not
shown) at maximum deNOx activity results in a moderate (10%)
and reversible deactivation [9].
Table 3
Temperature of maximum of N2 formation, percentage of NOx converted to N2 at this temperature, selectivity to N2 at Tmax N2, C3H6 conversion at Tmax N2
and temperature at 50% conversion of C3H6 (T50 C3H6) over Ce0.68Zr0.32O2 (CZ)- and SiO2-supported catalysts for the steady-state NO–O2–C3H6 reaction
(340 ppm–8%–1900 ppm, N2 balance)
Catalysts
Tmax N2 (◦C)
NOx to N2 at
Tmax N2 (%)
Selectivity to N2
at Tmax N2 (%)a
C3H6 conversion
at Tmax N2 (%)
T50 C3H6 (◦C)
CZ
361
318
263
243
288
289
309
8
9
81
83
83
87
81
85
50
34
30
38
50
68
45
77
380
355
303
243
263
294
290
Rh(0.40)/CZ
Pd(0.54)/CZ
Pd(0.89)/CZ
Rh(0.44)Pd(0.54)/CZ
Rh(0.44)Pt(0.91)/CZ
Rh(0.40)Pd(0.44)/SiO2
21
29
28
16
20
a
Selectivity to N2 referred to as (N2/(N2 + N2O) × 100).

76
C. Thomas et al. / Journal of Molecular Catalysis A: Chemical 249 (2006) 71–79
Fig. 3. Steady-state catalytic conversions of C3H6 and NOx in the course of the C3H6–NO–O2 reaction (1900 ppm–340 ppm–8%, balance N2): (a) Ce0.68Zr0.32O2
(CZ), (b) Rh(0.40)/CZ, (c) Pd(0.54)/CZ, (d) Pd(0.89)/CZ, (e) Rh(0.44)Pd(0.54)/CZ, (f) Rh(0.44)Pt(0.91)/CZ and (g) Rh(0.40)Pd(0.44)/SiO2.
4. Discussion
metallic phases deposited on CZ. To overcome these difficul-
ties, the use of model hydrogenation reactions was considered
to characterise the metallic phases, and hence the metal oxide
phases indirectly. From these experiments (Table 1), it is obvi-
ous that the percentage of metal exposed as zero-valent metal
atoms is much greater for Pd catalysts than for Rh or Pt ones.
These results agree well with the work of Summers and Ausen
in which these authors concluded to the greater interaction of Pt
compared to that of Pd with CeO2 [25]. Accordingly, the com-
parison of the XANES spectra of Pd and Rh catalysts (Refs.
[9,18] and Fig. 1) shows that the reduction of Pd is easier than
that of Rh for CZ-supported catalysts. These XANES spectra
The characterisation of the metallic phases, and hence of
metal oxide phases, supported on ceria-based catalysts is a very
challenging problem [24]. The characterisation of metal parti-
cles by means of H2 or CO chemisorptions was not performed
because it is well known that CZ-supported catalysts adsorb
large amounts of these probe molecules and that the analysis
of CO chemisorption faces stoichiometric problems [24]. In
addition, the low contents of metal involved in the synthesised
catalysts along with the elevated surface area of the CZ support
prevented the use of TEM to obtain further information on the

C. Thomas et al. / Journal of Molecular Catalysis A: Chemical 249 (2006) 71–79
77
reported [9], this discrepancy also suggests that the lean deNOx
mechanism occurring on the CZ-supported catalysts is different
from that elucidated on supported zero-valent PGMs atoms by
Burchetal. [10]. ForPd/CZcatalysts, amechanismhasbeenpro-
posed [9]. Given that the selectivity to N2 of the CZ-supported
Rh and Pt catalysts closely resembles that of Pd/CZ and differs
markedly from that of the corresponding supported zero-valent
atoms [22,26–28], such a mechanism might be generalised to
the CZ-supported PGMs catalysts studied in the present work
(Fig. 5). In this mechanism, the SCR of NOx proceeds through
the interaction of NO2 (cycle 1), adsorbed on the NMOx cat-
alytic function, with C3H6 to form RNOx species that further
decompose to oxygenates (CxHyOz) and NO (cycle 2). These
oxygenates act as reductants of CZ to create the catalytic sites
responsible for the decomposition of NO to N2 (cycle 3). This
mechanism is supported by the adsorption–desorption exper-
iments (Section 3.1.4) which show that: (i) NO is the major
desorbed NOx species of the TPD in C3H6–O2, whereas NO2
is the major desorbed species of the TPD in N2. Such a result
closely corresponds to gas phase experiments for which NO2 is
stoichiometrically transformed to NO in the presence of C3H6
along with the simultaneous formation of oxygenates [23]. (ii)
The interaction of C3H6 with pre-adsorbed NO2 species is most
pronounced for CZ-supported Pd catalysts (Section 3.1.4) that
exhibit the greatest deNOx activities (Table 3).
NO decomposition on Ce3+ sites, resulting from the reduction
of CZ (cycle 3) by the formed oxygenates, has originally been
demonstrated by Niwa et al. [29] and confirmed later by other
authors [30–32]. It is worthwhile reporting that this decomposi-
tion pathway has also been considered in three-way catalysis in
which CO is used as reductant [33].
Fig. 4. NOx to N2 conversion of (a) Rh(0.44)Pd(0.54)/CZ (᭹) and the sum of
those of Rh(0.40)/CZ and Pd(0.54)/CZ (ꢂ), and (b) Rh(0.40)/CZ (ꢁ) and the
difference of those of Rh(0.44)Pt(0.91)/CZ and Rh(0.40)/CZ (ꢂ) in the course
of the C3H6–NO–O2 reaction (1900 ppm–340 ppm–8%, balance N2).
The occurrence of a deNOx mechanism on CZ-supported
PGMs catalysts different from that depicted on reduced PGMs
atoms [10] is also supported by the order of reactivity observed
for the CZ-supported PGMs materials: Pd ꢀ Pt ≥ Rh. This reac-
tivity order is indeed different from that found on SiO2- or
Al2O3-supported catalysts [34,35] among which supported Pt
catalysts exhibit the greatest deNOx activity and supported Rh
ones the lowest activity. The deNOx reactivity order reported
for SiO2-supported PGMs catalysts [34] is in good agreement
with the specific activities found for C3H6 oxidation by Cant
and Hall [36]. This similarity corroborates the fact that on zero-
valent PGMs atoms, the SCR of NOx proceeds through the
dissociation of NO and the recombination of adsorbed N atoms
to form N2, the oxygen left from the NO dissociation process
being cleaned by C3H6 oxidation [10]. In contrast, the reactiv-
ity order observed on CZ-supported PGMs catalysts suggests
that PdOx catalyses mild oxidation of C3H6 more readily than
PtOx or RhOx do. Adsorption–desorption experiments (Section
3.1.4) indicate that this peculiarity might be attributed to the abil-
ity of the CZ-supported PdOx phase to adsorb NO2 compared
with those of PtOx or RhOx and, thus, to form RNOx species
that further decompose to oxygenates being considered as the
real reductants [37,38] of the SCR process (Fig. 5). This NO2
adsorption discrepancy between the NMOx phases might be cor-
related with the lower interaction of the PdOx phase with CZ
compared with those of RhOx and PtOx. The indirect evidence
also reveal that Rh and Pd do not seem to interact with each
other for the bimetallic catalyst (Section 3.1.2).
Qualitatively, the hydrogenation results suggest that the Pd
metal particles, hence the size of the Pd metal oxide particles as
only minor sintering effects are expected for reduction temper-
atures equal or lower than 500 ◦C [24], must be comparable or
bigger than the Rh or Pt metal particles. It is, indeed, difficult to
consider that the greater interaction of Pt or Rh compared to Pd
with CeO2 would lead to bigger metal particles in the case of
the Pt or Rh catalysts, the latter catalysts yielding to the lower
percentages of metal exposed as zero-valent metal atoms after
reduction (Table 1).
The introduction of PGMs to CZ clearly promotes the acti-
vation of C3H6 and the deNOx activity (Table 3). In addition,
the selectivity to N2 is much higher on the CZ-based materi-
als compared with that found on SiO2-supported zero-valent
PGMs catalysts (Table 3 and Ref. [9]). The different N2 selec-
tivity obtained for CZ- and SiO2-supported catalysts supports
the assumption that the PGMs are stabilised as PGMs oxide
phases (NMOx) by CZ and do not undergo to complete reduc-
tion under the reacting mixture in agreement with the work of
Liotta et al. on a Pt(1.0)/Ce0.60Zr0.40O2 catalyst [8]. As recently

78
C. Thomas et al. / Journal of Molecular Catalysis A: Chemical 249 (2006) 71–79
Fig. 5. (a) Schematic representation of the catalytic functions, NMOx representing PGMs oxide phases and (b) mechanism of the lean C3H6-assisted decomposition
of NO over PGMs/Ce0.68Zr0.32O2 (CZ) catalysts, , RNOx, CxHyOz representing an oxygen from the surface of the CZ support, organic nitrogen-containing and
oxygenates compounds of undefined composition, respectively [9].
of which is provided by the XANES measurements (Fig. 1 and
Refs. [9,18]) showing that the PdOx phase is more easily reduced
than the RhOx one. The higher interaction of Pt compared to
that of Pd with CeO2 has been revealed by Summers and Ausen
through CO-FTIR investigations [25]. By means of CO-FTIR
and HRTEM measurements, Bernal et al. also concluded to
strong interaction between Pt and CeO2 [19].
At first sight, the SCR of NOx activities found for the Pd-
based catalysts of this study do not favourably compare with
those reported on SiO2- or Al2O3-supported PGMs catalysts.
Nevertheless, it must be pointed out that the SCR of NOx
activities of the present study correspond to the selective trans-
formation of NOx to N2, whereas literature data mainly report
on global activities including the production of both N2 and
N2O [28,34,39–43]. For some of the aforementioned studies,
N2 selectivity has not been reported [28,39–42]. As it is well
established that the selectivity to N2 hardly exceeds 50% in the
temperature range of interest on supported zero-valent PGMs
catalysts [22,26–28], the SCR of NOx to N2 activities found in
the present study (Table 3) are very comparable to those of sup-
ported zero-valent PGMs catalysts. Therefore, this makes the
Pd/CZ catalytic systems of greater interest than the SiO2- or
Al2O3-supported PGMs ones because of their much lower pro-
duction of N2O, the formation of which is to be avoided from
a global warming point of view [4,5]. The SCR of NOx to N2
activities obtained in the present work are slightly lower than
those reported by Centi et al. [7] or Liotta et al. [8] for CZ-
based Pt catalysts. In contrast, these catalysts were slightly less
selective to N2 than ours. These studies were, however, realised
with lower C3H6 and O2 concentrations, with lower HSV and
with higher NO concentration than ours. These differences in
reaction conditions make the comparison difficult as we noticed
that increases of C3H6 and NO concentrations and decreases of
the O2 concentration and the HSV resulted in an increase of the
SCR activity (not shown).
5. Conclusion
This study shows that the order of reactivity of the PGMs
supported on CZ is: Pd ꢀ Pt ≥ Rh. The performances of the
most active CZ-supported catalysts are comparable to those
reported for SiO2- or Al2O3-supported PGMs catalysts in terms
of N2 production. In contrast, the CZ-supported catalysts are
much more selective to N2 than those supported on SiO2 or
Al2O3, which makes them of particular interest. This difference
of selectivity to N2 of the CZ-supported catalysts suggests that
the mechanism of the SCR of NOx assisted by C3H6 proposed
for Pd/CZ catalysts [9] is also valid for Pt or Rh/CZ catalysts.

C. Thomas et al. / Journal of Molecular Catalysis A: Chemical 249 (2006) 71–79
79
In this mechanism, ad-NO2 species react with C3H6 to form
ad-RNOx compounds that further decompose to NO and oxy-
genates (CxHyOz). The role of these oxygenates is to reduce the
support to provide the sites responsible for NO decomposition
to N2 (Ce3+ surface cations).
[9] C. Thomas, O. Gorce, C. Fontaine, J.-M. Krafft, F. Villain, G. Dje´ga-
Mariadassou, Appl. Catal. B: Environ. 63 (2006) 201.
[10] R. Burch, P.J. Millington, A.P. Walker, Appl. Catal. B: Environ. 4 (1994)
65.
[11] F. Fajardie, J.-F. Tempe`re, J.-M. Manoli, O. Touret, G. Blanchard, G.
Dje´ga-Mariadassou, J. Catal. 179 (1998) 469.
The much higher activity of the Pd-based catalysts in the
SCR of NOx assisted by C3H6 might be attributed to the greater
ability of the PdOx phase to adsorb NO2. This property might
be assigned to the lower interaction of the PdOx phase with CZ
compared with that of the PtOx or RhOx phases.
[12] S.H. Overbury, D.R. Mullins, in: A. Trovarelli (Ed.), Ceria Surfaces and
Films for Model Catalytic Studies Using Surface Analysis Techniques,
Imperial College Press, London, 2002, p. 328.
[13] F. Fajardie, J.-F. Tempe`re, G. Dje´ga-Mariadassou, G. Blanchard, J. Catal.
163 (1996) 77.
[14] L. Salin, C. Potvin, J.-F. Tempe`re, M. Boudart, G. Dje´ga-Mariadassou,
J.-M. Bart, Ind. Eng. Chem. Res. 37 (1998) 4531.
Although appreciable SCR of NOx to N2 is achieved on the
CZ-supported catalysts, enhancement of the activities is still
needed to comply with most recent emission regulations. The
proposeddeNOx mechanism(Fig. 5), whoseobviouscomplexity
lies in the fact that the three catalytic cycles have to turn over
simultaneously, suggests ways in which deNOx activity could
be improved. Superior activity in the NO oxidation reaction is
to be obtained (Fig. 5, cycle 1), as the evaluated catalysts do not
allow for maximum production of NO2 that remains far from
thermodynamic equilibrium at maximum N2 formation (Fig. 2).
Improvement of the selective mild oxidation of C3H6 remains a
very challenging task (Fig. 5, cycle 2), as this moderate oxidation
process always competes with total oxidation [44]. The quality
of the catalytic sites responsible for the decomposition of NO
(Fig. 5, cycle 3) might also be considered. These different issues
are currently being investigated in our group.
[15] A. Michalowicz, Logiciels pour la chimie, Socie´te´ Franc¸aise de Chimie,
Paris, 1991, p. 102.
[16] A. Michalowicz, J. Phys. IV France 7 (1997) 235.
[17] I. Manuel, J. Chaubet, C. Thomas, H. Colas, N. Matthess, G. Dje´ga-
Mariadassou, J. Catal. 224 (2004) 269.
[18] C. Fontaine, C. Thomas, J.-M. Krafft, O. Gorce, F. Villain, G. Dje´ga-
Mariadassou, in preparation.
[19] S. Bernal, G. Blanco, J.M. Gatica, C. Larese, H. Vidal, J. Catal. 200
(2001) 411.
[20] C. Yokoyama, M. Misono, J. Catal. 150 (1994) 9.
[21] T. Okuhara, Y. Hasada, M. Misono, Catal. Today 35 (1997) 83.
[22] K.O. Haj, S. Ziyade, M. Ziyad, F. Garin, Appl. Catal. B: Environ. 37
(2002) 49.
[23] O. Gorce, F. Baudin, C. Thomas, P. Da Costa, G. Dje´ga-Mariadassou,
Appl. Catal. B: Environ. 54 (2004) 69.
[24] S. Bernal, J.J. Calvino, J.M. Gatica, C. Lo´pez Cartes, J.M. Pintado, in: A.
Trovarelli (Ed.), Chemical and Nanostructural Aspects of the Preparation
and Characterisation of Ceria and Ceria-Based Mixed Oxide-Supported
Metal Catalysts, Imperial College Press, London, 2002, p. 85.
[25] J.C. Summers, S.A. Ausen, J. Catal. 58 (1979) 131.
[26] T. Tanaka, T. Okuhara, M. Misono, Appl. Catal. B: Environ. 4 (1994)
L1.
Acknowledgments
[27] R. Burch, T.C. Watling, Catal Lett. 37 (1996) 51.
[28] R. Burch, T.C. Watling, Appl. Catal. B: Environ. 11 (1997) 207.
[29] M. Niwa, Y. Furukawa, Y. Murakami, J. Colloid Interface Sci. 86 (1982)
260.
[30] A. Martinez-Arias, J. Soria, J.C. Conesa, X.L. Soane, A. Arcoya, R.
Cataluna, J. Chem. Soc. Faraday Trans. 91 (1995) 1679.
[31] S.H. Overbury, D.R. Mullins, D.R. Huntley, Lj. Kundakovic, J. Catal.
186 (1999) 296.
Rhodia contributed to part of the financial support for this
`
´
work; the Ministere de l’Enseignement Superieur et de la
Recherche organisation supported the work of Dr. O. Gorce
(Grant 98-4-10713). We also thank G. Blanchard for his interest
in this work and P. Lavaud for his invaluable help in technical
support.
[32] M. Daturi, N. Bion, J. Saussey, J.-C. Lavalley, C. Hedouin, T. Seguelon,
G. Blanchard, Phys. Chem. Chem. Phys. 3 (2001) 252.
[33] G. Dje´ga-Mariadassou, F. Fajardie, J.-F. Tempe`re, J.-M. Manoli, O.
Touret, G. Blanchard, J. Mol. Catal. A: Chem. 161 (2000) 179.
[34] R. Burch, P.J. Millington, Catal. Today 29 (1996) 37.
[35] G.R. Bamwenda, A. Ogata, A. Obuchi, J. Oi, K. Mizuno, J. Skrzypek,
Appl. Catal. B: Environ. 6 (1995) 311.
References
[1] R.J. Farrauto, R.M. Heck, Catal. Today 55 (2000) 179.
[2] V.I. Paˆrvulescu, P. Grange, B. Delmon, Catal. Today 46 (1998) 233.
[3] R. Burch, J.P. Breen, F.C. Meunier, Appl. Catal. B: Environ. 39 (2002)
283.
[4] Y. Traa, B. Burger, J. Weitkamp, J. Chem. Soc. Chem. Commun. 21
(1999) 2187.
[5] J. Perez-Ramirez, F. Kapteijn, K. Schoffel, J.A. Moulijn, Appl. Catal.
B: Environ. 44 (2003) 117.
[6] S.N. Orlik, V.L. Struzhko, T.V. Mironyuk, G.M. Tel’biz, Theor. Exp.
Chem. 37 (2001) 311.
[36] N.W. Cant, W.K. Hall, J. Catal. 16 (1970) 220.
[37] G. Dje´ga-Mariadassou, Catal. Today 90 (2004) 27.
[38] G. Dje´ga-Mariadassou, M. Boudart, J. Catal. 216 (2003) 89.
[39] J. Li, J. Hao, L. Fu, T. Zhu, Topics Catal. 30–31 (2004) 81.
[40] H. Shinjoh, T. Tanabe, K. Yokota, M. Sugiura, Topics Catal. 30–31
(2004) 319.
[7] G. Centi, P. Fornasiero, M. Graziani, J. Kasˇpar, F. Vazzana, Topics Catal.
16–17 (2001) 157.
[8] L.F. Liotta, A. Longo, A. Macaluso, A. Martorana, G. Pantaleo, A.M.
Venezia, G. Deganello, Appl. Catal. B: Environ. 48 (2004) 133.
[41] R. Burch, T.C. Watling, Catal Lett. 43 (1997) 19.
[42] T. Okuhara, Y. Hasada, M. Misono, Catal. Today 35 (1997) 83.
[43] R. Burch, J.A. Sullivan, T.C. Watling, Catal. Today 42 (1998) 13.
[44] J. Haber, W. Turek, J. Catal. 190 (2000) 320.