Combined removal of diesel soot particulates and NOx over CeO2-ZrO2 mixed oxides

Article abstract of DOI:10.1016/j.jcat.2008.07.016

CeO₂ and Ce-Zr mixed oxides with different Ce:Zr ratios were prepared; characterised by Raman spectroscopy, XRD, TEM, N₂ adsorption at -196 °C, and H₂-TPR; and tested for soot oxidation under NO_x/O₂. Among the different mixed oxides, Ce_0.76Zr_0.24O₂ provided the best results. Ce_0.76Zr_0.24O₂ presented greater activity than pure CeO₂ for soot oxidation by NO_x/O₂ when both catalysts were calcined at 500 °C (soot oxidation rates at 500 °C are 14.9 and 11.4 μg_soot/s, respectively), and the catalytic activity of CeO₂ decayed significantly with calcination temperature (from 500 to 1000 °C), whereas Ce_0.76Zr_0.24O₂ presented enhanced thermal stability at temperatures as high as 1000 °C. In addition, Ce_0.76Zr_0.24O₂ catalysed the reduction of NO_x by soot at around 500 °C more efficiently than CeO₂, thereby contributing to the decreased NO_x emission level. The catalytic activity of CeO₂ and Ce_0.76Zr_0.24O₂ for soot oxidation by NO_x/O₂ depended on the textural properties (BET area; crystallite size), but other properties of the oxides, such as redox behaviour and/or enhanced lattice oxygen mobility, also played a significant role.

Full text of DOI:10.1016/j.jcat.2008.07.016

Journal of Catalysis 259 (2008) 123–132
Contents lists available at ScienceDirect
Journal of Catalysis
www.elsevier.com/locate/jcat
Combined removal of diesel soot particulates and NO over CeO –ZrO mixed
x
2
2
oxides
∗
I. Atribak, A. Bueno-López , A. García-García
Department of Inorganic Chemistry, University of Alicante, Ap. 99 E-03080, Alicante, Spain
a r t i c l e i n f o
a b s t r a c t
Article history:
CeO2 and Ce–Zr mixed oxides with different Ce:Zr ratios were prepared; characterised by Raman
Received 17 April 2008
Revised 15 June 2008
Accepted 31 July 2008
Available online 30 August 2008
◦
spectroscopy, XRD, TEM, N2 adsorption at −196 C, and H2-TPR; and tested for soot oxidation under
NOx/O2. Among the different mixed oxides, Ce0.76Zr0.24O2 provided the best results. Ce0.76Zr0.24O2
presented greater activity than pure CeO2 for soot oxidation by NOx/O2 when both catalysts were calcined
◦
◦
at 500 C (soot oxidation rates at 500 C are 14.9 and 11.4 μgsoot/s, respectively), and the catalytic activity
◦
of CeO2 decayed significantly with calcination temperature (from 500 to 1000 C), whereas Ce0.76Zr0.24O2
presented enhanced thermal stability at temperatures as high as 1000 C. In addition, Ce0.76Zr0.24O2
catalysed the reduction of NOx by soot at around 500 C more efficiently than CeO2, thereby contributing
Keywords:
Catalysed soot oxidation
Ceria
◦
◦
Ce–Zr mixed oxide
Solid solution
Soot
to the decreased NOx emission level. The catalytic activity of CeO2 and Ce0.76Zr0.24O2 for soot oxidation
by NO /O2 depended on the textural properties (BET area; crystallite size), but other properties of the
x
oxides, such as redox behaviour and/or enhanced lattice oxygen mobility, also played a significant role.
NOx
©
2008 Elsevier Inc. All rights reserved.
Diesel pollution control
1
. Introduction
tem [5–8]. The energetics of the Ce4 /Ce reduction step and the
corresponding formation of oxygen vacancies are likely involved
[9]. Indeed, fundamental solid-state properties, such as the precise
role of structural defects and dopants, the nature of redox reac-
tions to create electronic species (both within the bulk and at the
surfaces), and the mechanism of oxygen migration clearly are cru-
cial to gaining insight into these important materials.
+
3+
Ceria is an important material in the framework of the three-
way catalysts (TWCs) used in gasoline automobile catalytic con-
verters for the treatment of exhaust gases [1,2]. One of the key
functions of this material is the ability of cerium to switch be-
tween the Ce4 and Ce
+
3+
oxidation states and to incorporate more
or less oxygen into their crystal structure depending on various
parameters, such as the gaseous atmosphere with which they are
in contact, temperature, and pressure [3]. Therefore, early on, ce-
ria played a main role as an oxygen storage component to extend
the three-way window on the lean side of stoichiometry by act-
ing as a sink for gas-phase oxygen during rich-to-lean transients
in air-to-fuel ratio [1–3]. On the other hand, the oxygen storage
component also could promote oxidation of reductants, like CO,
during lean-to-rich transients [3]. TWC formulation has undergone
significant advances during the past 20 years, mainly using ceria in
solid solution with other metal oxides, most notably zirconia [1,4].
The advanced TWC formulations are capable of much higher tem-
perature operation than their predecessors and have dramatically
improved long-term emissions performance [2,4].
The unique features of oxygen storage capacity have made Ce–
Zr binary oxides important in numerous catalytic processes besides
TWCs, including CO oxidation [10], light hydrocarbon combustion
[
11,12] and VOC oxidation [13]. The key steps of these reactions
are the supply of oxygen by the readily reducible mixed oxide
and its reoxidation by oxygen. The Ce–Zr mixed oxides are bet-
ter catalysts than bare CeO2, because the partial substitution of
4+
4+
Ce
(ionic radii = 0.97 Å) with Zr
(ionic radii = 0.84 Å) leads to
deformation of the lattice, improving its oxygen storage capacity,
redox properties, and thermal resistance [5–8,14,15].
The important role of the active oxygen generated by CeO2
in the catalysed oxidation of soot by O2 also has been reported
recently [16]. This oxide is of interest as a catalyst for the regen-
eration of soot traps fitted in the exhaust of diesel engines [17,18].
Doping of ceria with Zr [19], La [20,21], or Pr [21] has been found
to result in more active catalysts for soot oxidation by O2, but little
information has been reported regarding soot combustion under
NOx/O2 mixtures catalysed by Ce–Zr mixed oxides.
Adding zirconium to ceria to form a mixed-oxide solution is
4
+
known to greatly enhance the reducibility of Ce
in the catalyst
material. This has generated considerable interest in the Ce–Zr sys-
The catalytic behaviour of CeO2 for soot oxidation in O2 is not
the same than that in the presence of NOx. A previous study com-
pared the catalytic activity of pure TiO2, ZrO2, and CeO2 for soot
Corresponding author. Fax: +34 965 90 34 54.
E-mail addresses: atribak@ua.es (I. Atribak), agus@ua.es (A. Bueno-López),
a.garcia@ua.es (A. García-García).
*
0
021-9517/$ – see front matter © 2008 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2008.07.016

124
I. Atribak et al. / Journal of Catalysis 259 (2008) 123–132
oxidation under NOx +O2 mixtures [22] and concluded that CeO2’s
best activity is related to its capacity to accelerate the NO conver-
sion to NO2 (TiO2 and ZrO2 do not catalyse this reaction), NO2
being much more oxidizing than NO and O2 [23]. Rare earth (Sm,
Y, La, or Pr)-doped ceria catalysts also have been studied for soot
oxidation by NOx/O2; La and Pr doping enhanced ceria activity and
stability, whereas Sm and Y did not [24,25].
CeO2 has been reported to catalyse the reduction of NOx by
soot, removing both diesel pollutants (soot and NOx) simultane-
ously [22]. Pisarello et al. [26] studied the simultaneous removal of
NOx and soot with catalysts containing Co, K, and/or Ba supported
on MgO, La2O3, and CeO2 and concluded that CeO2 supplied the
necessary oxygen for the redox mechanism that occurs during the
reaction.
Blank experiments were performed under the aforementioned ex-
perimental conditions but without soot, that is, using only catalyst.
2.3. Characterisation techniques
The textural properties were determined by multipoint N ad-
2
◦
sorption at −196 C using an automatic Quantachrome Autosorb-
6B instrument. Data were treated in accordance with the BET
method. The samples had been previously degassed for 4 h at
◦
250 C under vacuum.
A JEOL JEM-2010 microscope was used to obtain TEM images of
the catalysts. Few droplets of an ultrasonically dispersed suspen-
sion of each catalyst in ethanol were placed in a grid and dried at
ambient conditions for TEM characterisation.
The aim of the present work was to study the catalytic per-
formance of Ce–Zr mixed oxides for the removal of soot under
simulated diesel exhaust conditions in a gas flow containing O2
and NOx. Special attention was given to the simultaneous removal
of NOx. The catalytic activity of the Ce–Zr mixed oxides was com-
pared with that of pure CeO2, and the thermal stability of the pure
and mixed oxides was studied in depth, because thermal stability
is a key requirement for the application of these materials in an
actual soot trap.
Raman spectra of the catalysts were recorded in a Bruker
RFS 100/S Fourier transform Raman spectrometer with a variable
power Nd:YAG laser source (1064 nm). A total of 64 scans at
85 mW laser power (70 mW on the sample) were recorded, and
no heating of the sample was observed under these conditions.
X-ray difractograms of the catalysts were recorded in a Bruker
D8 advance diffractometer, using CuK radiation. Spectra were reg-
α
◦
◦
◦
istered between 10 and 80 (2θ) with a step of 0.02 and a time
per step of 3 s. The average crystal size (D) of the catalysts was
determined using the equations of Scherrer and Williamson–Hall.
Sherrer’s equation is
2. Experimental
K · λ
2
.1. Catalyst preparation
D =
,
β · cos θ
Ce–Zr mixed oxides with different metal ratios were pre-
where λ is the wavelength of the radiation used (λ = 0.15418 nm
for CuKα); β is the full width at half maximum of the diffraction
peak considered (111); K is a shape factor, taken to be 0.9 (with 1
being a perfect sphere); and θ is the diffraction angle at which the
peak appears.
Estimating the crystal size of doped oxides presents some prob-
lems, because the introduction of foreign cations within the lattice
deforms the structure and affects the β values. The Williamson-
Hall equation separates the effects of size and strain in the crystals
and is more convenient for estimating crystal size of mixed oxides,
pared through a co-precipitation route. The required amounts of
ZrO(NO3)2·6H2O and/or Ce(NO3)3·6H2O (supplied by Aldrich) were
dissolved in water, and the hydroxides were precipitated by drop-
ping an ammonia solution to maintain the pH at about 9. The
◦
precipitates were dried at 90 C in air overnight and calcined in
◦
air for 3 h at different temperatures in the range of 500–1000 C.
Pure CeO2 and ZrO2 also were prepared following the same proce-
dure. The CeO2 and ZrO2 pure oxides are designated CeO2–T and
ZrO2–T, respectively, and the Ce–Zr mixed oxides are designated
CexZr1−xO2–T (formal composition, 0 < x < 1), where x is the mo-
lar fraction of CeO2 in the mixed oxides and T is the calcination
temperature.
0
.9λ
4(ꢀd)sin θ
d cos θ
β
Total
= βSize + βStrain
=
+
,
D cos θ
where βTotal is the full width half maximum of the XRD peak and
d is the difference of the d spacing corresponding to a typical
2.2. Catalytic tests
ꢀ
peak. A plot of βTotal cos θ against 4 sin θ yields the average crys-
tal size from the intercept value. Despite the improvement intro-
duced by the Williamson–Hall equation over Scherrer’s equation,
the crystal sizes estimated for the Ce–Zr mixed oxides prepared
in this study must be considered semiquantitative values, because
some of the data obtained are not fully consistent with the BET
areas estimated by N2 adsorption.
The catalytic tests were performed in a tubular quartz reac-
tor coupled to specific NDIR-UV gas analysers for CO, CO2, NO,
NO2, and O2 monitoring. First, 20 mg of soot and 80 mg of
the selected catalyst were mixed under so-called “loose contact”
conditions [27] and diluted with SiC to avoid pressure drop and
favour heat transfer. The gas mixture comprised 500 ppm NOx, 5%
O2, and balance N2, and the gas flow was fixed at 500 ml/min
The redox behaviour of the catalysts was examined by H2-TPR
in a Micromeritics Pulse ChemiSorb 2705 device consisting of a
tubular quartz reactor coupled to a TCD detector to monitor H2
consumption. The reducing gas used was 5% H2 in He. The temper-
−
1
(
GHSV = 30,000 h ).
Two types of experiments were carried out. In temperature-
programmed reactions, the gas mixture was fed to the reactor, and
then the temperature was increased from room temperature up to
◦
ature range explored was from room temperature to 900 C, and
the heating rate was 10 C/min.
◦
◦
7
00 C at a ramp rate of 10 C/min. In isothermal reactions, the
◦
◦
temperature was raised from room temperature up to 500 C, with
the soot-catalyst mixture maintained in N2 flow, and then the N2
flow was replaced by the reaction gas mixture. These experiments
were conducted until total conversion of soot occurred.
3
. Results and discussion
3
.1. Temperature-programmed reactions: Soot oxidation
The model soot used was a carbon black from Degussa (Prin-
tex U), with <0.1% ash content, 5% volatile matter, 92.2% C, 0.6% H,
.2% N, and 0.4% S. Soot conversion profiles were determined from
CO and CO2 evolved, and the selectivity of the different catalysts
toward CO emission was determined by
Fig. 1 compares the catalytic activity of two series of Ce–Zr
0
◦
mixed oxides, calcined at 500 and 1000 C, respectively. The tem-
perature required to convert 50% of the soot (T50%) in each ex-
periment is plotted as a function of the Zr molar fraction of the
catalysts used. The T50% temperature for the uncatalysed reaction
CO/COx (%) = 100CO/(CO + CO2).

I. Atribak et al. / Journal of Catalysis 259 (2008) 123–132
125
Fig. 1. Effect of the Ce:Zr ratio on soot combustion.
Fig. 2. Effect of the Ce:Zr ratio on CO emission.
◦
was 606 C under the experimental conditions used here, and most
of the samples decreased this temperature. Among the catalysts
◦
calcined at 500 C, CeO2-500 and Ce0.76Zr0.24O2-500 exhibited the
best performance, and increasing the Zr loading above this mo-
lar fraction had a negative effect on the activity because of the
lower amount of Ce available. This is consistent with the fact that
the couple Ce3
+/4+
is the responsible of the catalytic activity for
soot oxidation of CeO2-based catalysts. The same activity trend
has been reported previously [19] for CexZr1−xO2-catalysed soot
combustion in air (0 ꢀ x ꢀ 1.0 composition; catalysts calcined at
◦
5
00 C; soot and catalyst in tight contact), where the highest ac-
tivity was obtained with cerium-rich mixtures (with x = 1.0 and
x = 0.75). The arguments provided [19] to explain these results
were that cerium-rich samples have greater availability of surface
4
+
Ce
sites and that these materials (solid solutions) have a better
ability to donate oxygen for soot oxidation. The highest catalytic
activity of CexZr1−xO2 mixed oxides with x ∼ 0.75, compared with
higher zirconium loading, also has been reported for the selective
catalytic reduction of NOx by hydrocarbons [28].
Fig. 3. Effect of the calcination temperature of catalyst on soot combustion.
the amount of CO and CO2 yielded during the catalysed soot oxi-
dation reactions.
The activity of the catalysts decreased when they were calcined
◦
at 1000 C (Fig. 1), and the degree of thermal deactivation de-
Fig. 2 shows the percentage of CO evolved during the experi-
pended on their composition. Pure CeO2 and pure ZrO2 became
ments performed with the series of catalysts calcined at 500 and
◦
inactive when calcined at 1000 C. Ce0.76Zr0.24O2 exhibited a par-
◦
1
000 C. The uncatalysed reaction yielded 65% CO; all of the cata-
◦
tial decrease in activity after calcination at 1000 C. Ce–Zr mixtures
with a Zr molar fraction >0.24 calcined at 1000 C maintained
the poor activity of the counterpart catalysts calcined at 500 C.
lysts except CeO2-1000 decreased this value. The best results were
obtained with CeO2-500 and with Ce–Zr mixtures with a Zr molar
fraction ꢀ0.46. The calcination temperature of the Ce–Zr mixtures
had only a minor effect on CO selectivity.
◦
◦
The thermal stability of CeO2 is a critical aspect of its application
in catalysis, as was seen in TWCs, where the low stability of the
pure CeO2 used in early formulations required the development of
advanced CeO2-based materials with enhanced thermal resistance
To summarise the findings shown in Figs. 1 and 2, the best for-
mulation obtained was Ce0.76Zr0.24O2. This material exhibited the
same activity and selectivity for CO2 formation as pure CeO2 when
[2]. Pijolat et al. [29] investigated the effect of foreign cation dop-
◦
both catalysts were calcined at 500 C but showed enhanced ther-
ing on the thermal stability of CeO2 and concluded that among the
mal stability, maintaining part of its activity for soot oxidation and
selectivity for CO2 formation after calcination at temperatures as
4
+
4+
4+
3+
3+
3+
different cations investigated (Th , Zr , Si , La , Y , Sc ,
3
+
2+
2+
Al , Ca , and Mg ), those with ionic radii smaller than that
◦
high as 1000 C. This conclusion is in agreement with that reached
4
+
of Ce
effectively stabilised the CeO2 against sintering. This ob-
by Aneggi et al. [19] for CeO2- and Ce0.75Zr0.25O2-catalysed soot
◦
servation is consistent with the improved thermal resistance of
Ce0.76Zr0.24O2 compared with pure CeO2.
combustion in air (with catalysts calcined at 500 and 800 C),
These authors concluded that the main difference between pure
ceria and cerium–zirconium solid solution is related to the stabil-
ity after calcination.
Another feature of the soot oxidation catalysts that must be
analysed together with the decrease in soot oxidation temperature
is the production of CO2 and/or CO as a soot gasification product,
with CO2 being the desired gas product because of CO’s high toxic-
ity. A general feature of carbon combustion reactions is that the CO
production increases and CO2 production decreases with increasing
temperature. Therefore, on one hand, when a catalyst decreases
the soot combustion temperature, CO2 formation is favoured and,
on the other hand, CeO2-based catalysts are effective for catalysing
the oxidation of CO to CO2 [2]. Both catalyst properties determine
We studied the effect of calcination temperature on the cat-
alytic activity of CeO , ZrO , and Ce0.76Zr0.24O2 in more detail at
2
2
◦
500–1000 C. Figs. 3 and 4 show the results obtained for activ-
ity and selectivity toward CO2 formation, respectively. In all series
of catalysts, the catalytic activity for soot oxidation (Fig. 3) de-
creased with calcination temperature. As expected, ZrO2 had very
poor activity regardless of calcination temperature [22]. CeO2 and
Ce0.76Zr0.24O2 showed similar activity when catalysts were calcined

126
I. Atribak et al. / Journal of Catalysis 259 (2008) 123–132
(
a)
Fig. 4. Effect of the calcination temperature of catalyst on CO emission.
◦
at 500 C, but not at higher calcination temperatures. In general,
the Ce0.76Zr0.24O2 catalysts were more active than their counter-
part CeO2 catalysts. The activity of CeO2 decreased quite mono-
◦
tonically from 500 to 1000 C, with CeO2-1000 not active at all.
In contrast, the activity of Ce0.76Zr0.24O2 for soot oxidation clearly
◦
decreased between 500 and 600 C (with T50% increasing from
◦
◦
5
21 C to 543 C, respectively) but was not modified significantly
further until 900 C, indicating very high thermal stability within
the range of 600–900 C. The activity of Ce0.76Zr0.24O2 decreased
further only between 900 and 1000 C, but some activity remained.
◦
◦
◦
The effect of calcination temperature on the selectivity toward
CO2 formation was found to depend significantly on the cata-
lyst formulation (Fig. 4). Regardless of the calcination temperature,
Ce0.76Zr0.24O2 yielded a low CO percentage (13 ± 6%) and ZrO2
yielded a high CO percentage (53 ± 12%). CO formation during
CeO2-catalysed soot oxidation increased slightly with catalysts cal-
(
b)
◦
Fig. 5. Effect of the Ce:Zr ratio on: (a) NOx elimination and (b) O2 elimination (cat-
cined between 500 and 600 C but increased dramatically between
◦
alysts calcined at 500 C).
◦
6
00 and 800 C, almost reaching the CO percentage yielded during
the uncatalysed reaction. These results confirm that Ce0.76Zr0.24O2
was the best among the different formulations studied, based on
its activity for soot oxidation, high thermal stability, and high se-
lectivity toward CO2 formation as a soot oxidation product.
whereas the NOx chemisorption in both catalysts observed in
blank experiments (Fig. 6a) were the same. This suggests a
certain contribution of another NOx elimination pathway (NOx
reduction by soot), in addition to NOx chemisorption.
◦
•
•
NOx elimination at around 500 C. The elimination of NOx at
around this temperature can be attributed in part to the catal-
ysed soot–NOx reaction, with NOx consumed due to reduction
by soot. This is the most interesting NOx elimination pathway,
because it allows the simultaneous abatement of NOx and soot
by the reaction of both pollutants with each other. The forma-
tion of N2O as a reaction product was not detected in other
3
.2. Temperature-programmed reactions: NOx elimination
The elimination of NOx during the TPRs also depended on
the catalyst used, as shown in Fig. 5a for catalysts calcined at
◦
5
00 C. This type of profile was previously studied in detail for
pure CeO2 catalysts [22]. Considering as an example the NOx elim-
ination profile obtained with Ce0.76Zr0.24O2 (see Fig. 5a), three dif-
◦
◦
experiments followed by gas chromatography [22]. O also was
2
ferent processes can be distinguished between 225 C and 700 C:
◦
◦
converted at around 500 C due to soot combustion, as shown
The maximum NOx elimination level was reached at 400 C, and
in Fig. 5b. As mentioned in the previous section, the catalytic
activity for soot oxidation of the series of catalysts calcined
two shoulders appeared at higher temperatures, around 500 and
◦
6
00 C. This profile suggests three different NOx elimination path-
◦
at 500 C decreased with increasing Zr molar fraction (Fig. 1),
ways. The shape of the NOx elimination curves depends on the
catalyst used; in other words, the relative importance of each NOx
elimination pathway differs for each catalyst. The three NOx elimi-
nation profiles are as follows:
with Ce0.76Zr0.24O2-500 being the most active Ce–Zr mixture.
◦
NOx elimination at around 600 C. The elimination of NOx at
around this temperature can be attributed mainly to the un-
catalysed soot–NOx reaction, occurring along with the soot–O2
reaction (see Fig. 5b). As shown in Fig. 5a, this is the only NOx
elimination pathway for the least active catalysts (ZrO2-500,
Ce0.16Zr0.84O2-500, and Ce0.36Zr0.64O2-500).
◦
•
NOx elimination around 400 C and lower temperatures. The
elimination of NOx is mainly attributed to NOx chemisorption
on the catalysts, as supported by the blank experiments (with-
out soot) shown in Fig. 6a. NOx chemisorption on Ce–Zr mix-
◦
tures (Fig. 5a) at around 400 C decreased by increasing the Zr
A key factor influencing the catalysed soot oxidation reactions
molar fraction, because NOx chemisorption occurred on Ce but
not on Zr [22]. The NOx elimination levels reached at 400 C
by Ce0.76Zr0.24O2-500 and CeO2-500 (Fig. 5a) differed slightly,
in the presence of NOx is NO2 production by the different cata-
lysts. Fig. 6b plots the NO2 percentage (on the basis of NO + NO2)
◦
◦
as a function of temperature for blank experiments. Above 275 C,

I. Atribak et al. / Journal of Catalysis 259 (2008) 123–132
127
(
a)
(
a)
(b)
(b)
Fig. 7. Effect of the calcination temperature of catalyst on NOx elimination: (a) CeO2-
series and (b) Ce0.76Zr0.24O2-series.
Fig. 6. Blank experiments (without soot): (a) NOx elimination and (b) NO2 forma-
tion.
Ce0.76Zr0.24O2-500 and CeO2-500 catalysed the oxidation of NO to
NO2, reaching a maximum NO2 level corresponding to the ther-
modynamic equilibrium at 450 C and decreasing at higher tem-
Table 1
◦
Data estimated from isothermal reactions at 500
C
◦
Catalyst
CO/COx
Time for 75% soot
conversion (min)
Average soot oxidation
rate (μgsoot/s)
(
%)
peratures following thermodynamics. In contrast, the catalysts cal-
◦
cined at 1000 C were not effective for converting NO into NO2,
CeO2-500
CeO2-1000
Ce0.76Zr0.24O2-500
Ce0.76Zr0.24O2-1000
2.1
11.3
3.5
24
172
19
11.4
1.2
14.9
4.1
consistent with the lower activity of the Ce0.76Zr0.24O2-1000 and
CeO2-1000 samples for soot oxidation compared with the counter-
36.8
62
◦
part samples calcined at 500 C. It is well established [30] that the
catalytic activity for soot oxidation of ceria-based catalysts under
NOx mixtures is related to these catalysts’ ability to accelerate NO2
production, with NO2 being more oxidizing than NO and O2.
We also studied the effect of the calcination temperature of
CeO2 and Ce0.76Zr0.24O2 on these catalysts’ NOx elimination capac-
ity; the curves obtained during the corresponding TPRs are shown
in Figs. 7a and 7b, respectively. As shown in Fig. 7a, the amount of
different catalysts tested and showed enhanced thermal stability
with regard to bare CeO2.
◦
3.3. Isothermal reactions at 500 C
As described earlier, the TPR results indicate that some of the
◦
NOx removed below 500 C was decreased with calcination tem-
perature, with a dramatic decrease seen from CeO2-600 to CeO2-
catalysts tested are able to promote the soot–NO reaction at
x
◦
around 500 C, thereby allowing the simultaneous removal of both
7
00. Removal of NOx through the uncatalysed reaction (at around
NOx and soot. Considering the best performance of Ce0.76Zr0.24O2,
its catalytic activity was tested under isothermal conditions at
◦
600 C) was the main NOx elimination pathway for CeO2 calcined
◦
◦
at 800 C and higher.
500 C and was compared with that of bare CeO . In both cases,
2
◦
The behaviour of the Ce0.76Zr0.24O2-series (Fig. 7b) differed
from that of the CeO2 series (Fig. 7a). Ce0.76Zr0.24O2 sustained a
gradual decrease in NOx removal capacity, and only the Ce0.76Zr0.24-
O2-1000 sample exhibited a NOx profile ascribed solely to the
uncatalysed NOx–soot reaction. The NOx elimination profiles in-
dicate that Ce0.76Zr0.24O2 is the best formulation among those
studied, because it demonstrated the best performance among the
catalysts calcined at 500 and 1000 C were tested. Fig. 8a shows
the soot oxidation rates, expressed per gram of soot remaining
in the reactor, and Fig. 8b shows the corresponding NOx elimi-
nation profiles. Table 1 compiles the average soot oxidation rates
◦
estimated from isothermal experiments at 500 C, as well as the
percentage of CO evolved during the experiments and the time
required for 75% soot conversion to CO + CO2. Of note, the un-

128
I. Atribak et al. / Journal of Catalysis 259 (2008) 123–132
(
a)
Fig. 9. BET surface areas in terms of calcination temperature of catalysts.
1000 compared with Ce0.76Zr0.24O2-500. Under our experimental
conditions, 75% of the soot used in the experiments was burnt
off after 19 min with Ce0.76Zr0.24O2-500 and after 62 min with
Ce0.76Zr0.24O2-1000. The practical implication of this result is that
regeneration of a DPF loaded with Ce0.76Zr0.24O2-1000 would take
about three times longer than regeneration of the same filter
loaded with Ce0.76Zr0.24O2-500. The activity of CeO2-500 also de-
◦
creased after calcination at 1000 C, but in this case the calcination
temperature had a dramatic effect. It took 172 min for CeO2-1000
to convert 75% of the soot used in the experiment, compared
with 24 min for CeO2-500. Because thermal ageing of the cata-
lyst can occur in a real filter, thermally stable active phases are
required, and Ce0.76Zr0.24O2 is preferable to pure CeO2. In addi-
tion, the activity of CeO2-500 for soot oxidation and NOx reduc-
◦
tion under isothermal conditions at 500 C was lower than that
(b)
of Ce0.76Zr0.24O2-500 (Fig. 8). These differences were not detected
in the transient conditions of the TPRs (Figs. 1 and 3), indicating
the importance of performing experiments under isothermal con-
ditions, which also provide a more realistic picture of the catalytic
behaviour under real conditions.
◦
Fig. 8. Isothermal reactions at 500 C: (a) soot oxidation rate and (b) NOx elimi-
nation. (The units of soot oxidation rate are mg of soot converted per second and
gram of soot remaining in the reactor.)
◦
catalysed soot oxidation at 500 C did not occur under these ex-
perimental conditions.
3
.4. Study of the catalytic activity decay due to thermal ageing
Ce0.76Zr0.24O2-500 was the most active catalyst for soot oxida-
tion (Fig. 8a) and NOx reduction (Fig. 8b). The soot oxidation rate
during the Ce0.76Zr0.24O2-500-catalysed experiment rose with soot
conversion until about 25% of conversion, which can be attributed
to an increased number of active sites for oxygen chemisorption
on the soot surface as a result of its progressive oxidation [31,32].
After a certain conversion, the steady state was reached, and a con-
stant rate was maintained. As Table 1 shows, the main reaction
The thermal deactivation of CeO2 and Ce0.76Zr0.24O2 was stud-
ied, and their BET surface areas are shown as a function of the
calcination temperature in Fig. 9. The BET surface areas of the sam-
◦
2
ples calcined at 500 C were approximately similar (∼65 m /g),
and the surface areas of all of the oxides decreased significantly
with calcination temperature. Ce0.76Zr0.24O2 exhibited much bet-
ter thermal resistance than pure CeO , because of the stabilising
2
◦
product was CO2, mainly for catalysts calcined at 500 C, with very
role of Zr. In this type of mixed oxide, the BET value is related
low CO emissions (according to TPR results). The reduction of NOx
to the external surface area of the particles, with high surface
area meaning small particle size and vice versa. The TEM pictures
shown in Fig. 10 are in agreement with this argument as well as
with the crystal sizes determined from XRD by the Scherrer and
Williamson-Hall equations (Table 2). In the TEM pictures, CeO2-
(Fig. 8b) decreased progressively with soot conversion, as expected
because less reductant was available, and NOx elimination was null
after about 75% soot conversion. But a small amount of NOx was
further eliminated once the soot was almost completely consumed,
and the NOx elimination profiles rose slightly at the end of the
experiment. This phenomenon was attributed to the chemisorp-
tion of NOx on the catalyst [22]. If soot were available (in Fig. 8,
500 and Ce0.76Zr0.24O -500 appear as quite homogeneous agglom-
2
erations of small particles with particle sizes of ca. 10–20 nm, con-
sistent with the crystal sizes determined from the Williamson-Hall
◦
<
75% soot conversion for Ce0.76Zr0.24O2-500), then the reaction of
equation (22 and 21 nm, respectively). After calcination at 1000 C,
NOx (mainly NO2) with soot would be preferable to the storage of
NOx on the catalyst; but once soot were not available (>75% soot
conversion), NOx chemisorption on the catalyst would take place,
being the only possible NOx elimination pathway.
The calcination temperature of the catalyst diminished the cat-
alyst activity, and the soot oxidation rate (Fig. 8a) in steady-
state conditions was about threefold lower with Ce0.76Zr0.24O2-
the CeO2 particles grew due to thermal sintering, forming particles
larger than 100 nm (Fig. 10b), whereas Ce0.76Zr0.24O -1000 main-
2
tained smaller particles of around 20–30 nm (Fig. 10d). Along this
line, the Williamson-Hall equation (Table 2) estimated crystal sizes
of 107 nm for CeO2-1000 and 65 nm for Ce0.76Zr0.24O2-1000.
The structural characterisation of CeO2 and Ce0.76Zr0.24O2 sam-
ples was carried out by XRD and Raman spectroscopy (Figs. 11

I. Atribak et al. / Journal of Catalysis 259 (2008) 123–132
129
Fig. 10. TEM images: (a) CeO2-500, (b) CeO2-1000, (c) Ce0.76Zr0.24O2-500, (d) Ce0.76Zr0.24O2-1000.
(
a)
(a)
(
b)
(b)
Fig. 12. Raman characterisation of catalysts: (a) CeO -series and (b) Ce0.76Zr0.24O -
Fig. 11. XRD characterisation of catalysts: (a) CeO2-series and (b) Ce0.76Zr0.24O2-
2
2
series.
series.

130
I. Atribak et al. / Journal of Catalysis 259 (2008) 123–132
Table 2
Crystal sizes determined from XRD
Crystal size
with Williamson–
Crystal size with
Scherrer’s
Hall’s equation (nm)
equation (nm)
CeO2-500
CeO2-600
CeO2-700
CeO2-800
CeO2-900
CeO2-1000
22
22
46
63
116
107
14
18
32
39
58
55
Ce0.76Zr0.24O2-500
Ce0.76Zr0.24O2-600
Ce0.76Zr0.24O2-700
Ce0.76Zr0.24O2-800
Ce0.76Zr0.24O2-900
Ce0.76Zr0.24O2-1000
10
21
20
17
19
65
11
13
9
12
13
22
and 12, respectively). All of the diffractograms in Figs. 11a and 11b
show the main reflections typical of a fluorite-structured material,
with fcc unit cells at 28.5, 33.1, 47.6, and 56.5 , corresponding to
Fig. 13. H2 consumption profiles during H2-TPR for selected catalysts.
4+ 3+ 4+
to Ce , because Zr
◦
Ce
is a nonreducible cation. It is gener-
the (111), (200), (220), and (311) planes [33]. XRD revealed no ev-
idence of phase segregation for the Ce–Zr mixed oxide (Fig. 11b),
with no characteristic ZrO2 peaks. Some degree of inhomogeneous
distribution of cerium and zirconium cannot be ruled out, however,
as we discuss next in the context of H2-TPR characterisation.
Incorporation of Zr into the fluorite structure of CeO2 caused
lattice deformation. At the same calcination temperature, the in-
tensity of the CeO2 peaks was greater than that of the counterpart
Ce0.76Zr0.24O2 mixed oxide. This is a consequence of the better
arrangement of the atoms into the framework of pure CeO2 and
the decreased number of lattice defects. The effect of calcination
temperature is clearly shown in Fig. 11a. With increasing calcina-
tion temperature, the peaks of CeO2 narrowed. This is related to
an increase in crystal size, which is consistent with BET and TEM
characterisation, and also with the negative effect of temperature
on the catalytic activity of CeO2. The same effect of temperature
can be seen in the patterns in Fig. 11b but to a lesser extent, in
agreement with the greater thermal stability of the Ce–Zr mixed
oxide.
ally accepted [39] that two peaks characterise the reduction profile
of pure CeO2. The first peak, centred at around 500 C in the pro-
◦
file of CeO2-500, is attributed to the reduction of the uppermost
4+
◦
layers of Ce ; the second peak, centred at 800 C, originated
from the reduction of the bulk. The H2 consumption profile of
Ce0.76Zr0.24O2-500 also had this shape, but the peak intensity ratio
(surface/bulk reduction) was higher for the mixed oxide. This sug-
gests enhanced oxygen mobility within the Ce0.76Zr0.24O2 lattice
compared with the bare CeO2. On the other hand, a certain de-
gree of inhomogeneous distribution of cerium and zirconium also
can be inferred. A broad H2 consumption band would be expected
for a true Ce–Zr solid solution, where the surface and bulk re-
duction would occur concurrently, but a clear distinction between
surface and bulk reduction would not be expected. Wu et al. [40]
reported a bimodal H2-TPR profile for a Ce0.5Zr0.5O2 sample, which
they ascribed to a possible “shell/core structure” concordant with
a heterogeneous surface elemental distribution. These authors re-
ported a surface Ce/Zr ratio of 1.3, higher than the nominal ratio
(1.0) for this sample. The results of XPS analysis performed with
Raman characterisation supports the XRD findings. All of the
CeO2 and Ce0.76Zr0.24O2 samples (Figs. 12a and 12b) exhibited the
our Ce0.76Zr0.24O2 mixed oxide also showed a Ce/Zr ratio of 5.0,
well above the value of 3.2 expected for this nominal composition.
Along this line, Nagai et al. [41,42] also assessed (by XAFS) the Ce-
and Zr-rich domains in a Ce0.5Zr0.5O2 solid solution prepared by
−1
typical structure of CeO2, with the main band at 460 cm
at-
tributed to the only allowed Raman mode (F2g) of a fluorite-type
structure [34,35]. The Raman spectra of these fluorite-type oxide
structures are dominated by oxygen lattice vibrations and are sen-
◦
co-precipitation and subsequent calcination at 500 C.
sitive to the crystalline symmetry [36]. The presence of Zr4+ in
The profile of CeO2-1000 contains only the bulk reduction peak,
because of the very low BET surface area of this sample. The
drastic effect of the calcination temperature on the surface re-
dox properties of CeO2 is not so obvious in Ce0.76Zr0.24O2. The
H2 consumption profile of Ce0.76Zr0.24O2-1000 shows a broad band
instead of two well-defined peaks. The onset temperature of this
band is consistent with a surface reduction process. The generally
accepted explanation for this type of profile is that surface and
bulk reduction occur concurrently [39]; that is, there is no clear
distinction between the surface and bulk peaks, because oxygen
located within the bulk comes to the surface as surface oxygen is
consumed. This type of profiles is characteristic of materials with
good oxygen mobility.
the CeO2 lattice deformed the structure, and the intensity of the
characteristic fluorite peak decreased significantly, as can be seen
by comparing Figs. 12a and 12b. This deformation has been re-
ported to favour oxygen mobility, affecting the redox behaviour
of the material [37]. The calcination temperature also affected the
arrangement of atoms in the CeO2 lattice (Fig. 12a), and the in-
tensity of the Raman peak increased with calcination temperature
due to the better arrangement of atoms. Raman characterisation
also provided evidence of the improved thermal stability of the
Ce–Zr mixed oxide. As shown in Fig. 12b, the intensities of the
◦
signals of Ce0.76Zr0.24O2 calcined between 500 and 900 C were
about the same, and they increased slightly after calcination at
◦
1
000 C. The shift of the Raman signal toward lower energies that
To summarise, the characterisation of CeO2 and Ce0.76Zr0.24O2
suggests that both materials have a fluorite structure, and that
◦
occurred in Ce0.76Zr0.24O2 calcined between 500 and 900 C with
Ce–Zr incorporates Zr4 cations into the CeO2 framework. This
incorporation of foreign cations enhances the thermal resistance
of CeO2, diminishing thermal sintering. As a consequence of this
improved thermal resistance, Ce0.76Zr0.24O2 calcined at high tem-
perature has a greater BET surface area and better redox properties
compared with pure CeO2.
+
regard to Ce0.76Zr0.24O2-1000 is a sign of the improved oxygen
mobility, which could be related to the presence of oxygen vacan-
cies [38].
We investigated the redox properties of selected samples by
H2-TPR; the H2 consumption profiles obtained are plotted in
Fig. 13. H2 consumption must be attributed to the reduction of

I. Atribak et al. / Journal of Catalysis 259 (2008) 123–132
131
the same selectivity for CO2 formation as a soot oxidation product.
But Ce0.76Zr0.24O2 showed enhanced thermal stability compared
with pure CeO2 (as deduced from XRD, Raman, TEM, N2 adsorp-
tion and H2-TPR characterisation), maintaining part of its activity
for soot oxidation and selectivity toward CO2 formation after calci-
◦
nation at temperatures as high as 1000 C.
In addition, Ce0.76Zr0.24O2 catalysed the reduction of NOx by
◦
soot at around 500 C more efficiently than CeO2 did, thereby con-
tributing to decreased NOx emissions. The catalytic activity of CeO2
and Ce0.76Zr0.24O2 for soot oxidation by NOx/O2 depends on their
textural properties (BET area; crystallite size), but other properties
of the oxides, like redox behaviour and/or enhanced lattice oxygen
mobility, play significant roles as well.
Acknowledgments
Fig. 14. T50% parameter versus BET surface area of catalysts (CeO2-series and
Ce0.76Zr0.24O2-series).
Financial support was provided by the Spanish Ministry of Edu-
cation and Science (Project CTQ2005-01358) and the ABL contract
funded by the Ramon y Cajal Program and the Generalidat Valen-
ciana.
It is not easy to distinguish between the relative contribution of
structural properties (e.g., BET area; particle size) and other prop-
erties, such as redox behaviour or lattice oxygen mobility, in the
catalytic activity toward soot combustion of CeO2-based pure and
mixed oxides. To gain insight into this distinction, Fig. 14 plots the
temperature as a function of BET surface area for 50% soot conver-
sion obtained from TPRs of the different CeO2 and Ce0.76Zr0.24O2
samples. A relationship between the parameters can be clearly
seen, suggesting that the catalytic activity of pure CeO2 and of
the mixed oxide Ce0.76Zr0.24O2 depends on surface area. A relation-
ship between surface area and catalytic activity for soot oxidation
in temperature-programmed oxidation by O2, with soot and cata-
lyst in tight contact, also has been also reported for the catalyst
Ce0.95Fe0.05O1.975 [19]. This catalyst was calcined at different tem-
References
[
[
[
1] M.V. Twigg, Appl. Catal. B 70 (2007) 2.
2] J. Kašpar, P. Fornasiero, M. Graziani, Catal. Today 50 (1999) 285.
3] A. Trovarelli, Catalysis by Ceria and Related Materials, Catalytic Science Series,
vol. 2, Imperial College Press, 2002, p. 281.
[4] H.S. Gandhi, G.W. Graham, R.W. McCabe, J. Catal. 216 (2003) 433.
[5] P. Fornasiero, G. Balducci, J. Kašpar, S. Meriani, R. di Monte, M. Graziani, Catal.
Today 29 (1996) 47.
[6] C.E. Hori, H. Permana, K.Y.S. Ng, A. Brenner, K. More, K.M. Rahmoellerd, D. Bel-
ton, Appl. Catal. B 16 (1998) 105.
[7] M. Daturi, E. Finocchio, C. Binet, J.C. Lavalley, F. Fally, V. Perrichon, H. Vidal, N.
Hickey, J. Kašpar, J. Phys. Chem. B 104 (2000) 9186.
◦
peratures in the range of 500–750 C, achieving surface areas of
2
9
2–10 m /g. A linear correlation between surface area and activ-
[8] N. Hickey, P. Fornasiero, J. Kašpar, J.M. Gatica, S. Bernal, J. Catal. 200 (2001)
181.
2
ity was obtained for surface areas <40 m /g, whereas the vari-
ation in activity was not so relevant for larger surface areas. In
the present study, this threshold was not observed, probably be-
cause the soot and catalyst were in loose, not tight contact. For
experiments in loose contact, we also would expect a threshold
in surface area values above which the catalytic activity no longer
improved; however, this limit seems to be above the maximum
[9] G. Balducci, J. Kašpar, P. Fornasiero, M. Graziani, M. Saiful Islam, J. Phys. Chem.
B 102 (1998) 557.
[
10] B.M. Reddy, P. Lakshmanan, P. Bharali, P. Saikia, G. Thrimurthulu, M. Muhler,
W. Grulnert, J. Phys. Chem. C 111 (2007) 10478.
[11] S. Eriksson, S. Rojas, M. Boutonnet, J.L.G. Fierro, Appl. Catal. A 326 (2007) 8.
[
12] D. Terribile, A. Trovarelli, C. de Leitenburg, A. Primavera, G. Dolcetti, Catal. To-
day 47 (1999) 133.
2
surface area reached in the current study (67 m /g).
[13] J.I. Gutiérrez-Ortiz, B. de Rivas, R. López-Fonseca, J.R. González-Velasco, Appl.
Catal. B 65 (2006) 191.
Despite the relationship between catalyst surface area and the
T50% values obtained in the TPRs (Fig. 14), the isothermal reactions
[14] F. Fally, V. Perrichon, H. Vidal, J. Kašpar, G. Blanco, J.M. Pintado, S. Bernal, G.
◦
Colon, M. Daturi, J.C. Lavalley, Catal. Today 59 (2000) 373.
performed at 500 C indicate that Ce0.76Zr0.24O2-500 had better ac-
[
15] I. Atribak, A. Bueno-López, A. García-García, Catal. Commun. 9 (2008) 250.
16] A. Bueno-López, K. Krishna, M. Makkee, J.A. Moulijn, Catal. Lett. 99 (2005) 203.
tivity than CeO2-500, and both samples had the same BET surface
area. These differences can be attributed to the improved redox
properties, deduced from H2-TPR (Fig. 13), and/or enhanced lattice
oxygen mobility, deduced from Raman spectroscopy (Fig. 12), of
the mixed oxide. In conclusion, the structural properties of CeO2-
based pure and mixed oxides play important roles in the catalytic
activity for soot oxidation, and other properties, such as redox
behaviour and enhanced lattice oxygen mobility, also affect per-
formance.
[
[17] B.A.A.L. van Setten, M. Makkee, J.A. Moulijn, Catal. Rev. 43 (2001) 489.
[
[
18] J.P.A. Neeft, M. Makkee, J.A. Moulijn, Fuel Proc. Technol. 47 (1996) 1.
19] E. Aneggi, C. de Leitenburg, G. Dolcetti, A. Trovarelli, Catal. Today 114 (2006)
40.
[20] A. Bueno-López, K. Krishna, M. Makkee, J.A. Moulijn, J. Catal. 230 (2005) 237.
21] K. Krishna, A. Bueno-López, M. Makkee, J.A. Moulijn, Appl. Catal. B 75 (2007)
89.
22] I. Atribak, I. Such-Basáñez, A. Bueno-López, A. García García, J. Catal. 250 (2007)
5.
23] A. Setiabudi, J. Chen, G. Mul, M. Makkee, J.A. Moulijn, Appl. Catal. B 51 (2004)
[
1
[
[
[
7
4
. Conclusion
9.
24] K. Krishna, A. Bueno-López, M. Makkee, J.A. Moulijn, Appl. Catal. B 75 (2007)
201.
Based on our findings in the current study, we can conclude
that the catalytic activity of CeO2 for diesel-exhaust purification
[25] K. Krishna, A. Bueno-López, M. Makkee, J.A. Moulijn, Appl. Catal. B 75 (2007)
210.
could be significantly improved by doping CeO2 with Zr4 , with
Ce0.76Zr0.24O2 being the best formulation among those prepared
and tested. This Ce–Zr mixed oxide had a slightly higher activity
than bare CeO2 for soot oxidation by NOx/O2 when both cata-
+
[
26] M.L. Pisarello, V. Milt, M.A. Peralta, C.A. Querini, E.E. Miró, Catal. Today 75
2002) 465.
27] B.A.A.L. van Setten, J.M. Schouten, M. Makkee, J.A. Moulijn, Appl. Catal. B 28
2000) 253.
(
[
(
◦
lysts were calcined at 500 C (with soot oxidation rates of 14.9 and
[28] M. Adamowska, S. Muller, P. Da Costa, A. Krzton, P. Burg, Appl. Catal. B 74
◦
1
1.4 μgsoot/s, respectively, at 500 C), and both catalysts exhibited
(2007) 278.

132
I. Atribak et al. / Journal of Catalysis 259 (2008) 123–132
[
29] M. Pijolat, M. Prin, M. Soustelle, O. Touret, P. Nortier, J. Chem. Soc. Faraday
[36] M. Fernandez-García, A. Martínez-Arias, A. Iglesias-Juez, C. Belver, A.B. Hungría,
J.C. Conesa, J. Soria, J. Catal. 194 (2000) 385.
Trans. 91 (1995) 3941.
[
[
30] A. Setiabudi, M. Makkee, J.A. Moulijn, Appl. Catal. B 50 (2004) 185.
31] A. Bueno-López, A. García-García, J.A. Caballero-Suárez, Environ. Sci. Tech-
nol. 36 (2002) 5447.
[37] P. Fornasiero, J. Kašpar, M. Grazini, J. Catal. 167 (1997) 576.
[38] S. Rossignol, C. Descorme, C. Kappenstein, D. Duprez, J. Mater. Chem. 11 (2001)
2587.
[
[
[
[
32] A. Bueno-López, A. García-García, A. Linares-Solano, Fuel Proc. Technol. 77–78
2002) 301.
33] D. Terribile, A. Trovarelli, J. Llorca, C. de Leitenburg, G. Dolcetti, Catal. Today 43
1998) 79.
34] A. Nineshige, T. Taji, Y. Muroi, M. Kobune, S. Fujii, N. Nishi, M. Inaba, Z. Ogumi,
Solid State Ionics 135 (2000) 481.
[39] G.L. Markaryan, L.N. Ikryannikova, G.P. Muravieva, A.O. Turakulova, B.G.
Kostyuk, E.V. Lunina, V.V. Lunin, E. Zhilinskaya, A. Aboukais, Colloids Surf. A 151
(1999) 435.
[40] W. Xiadong, L. Qing, W. Xiaodi, W. Duan, J. Rare Earths 25 (2007) 416.
[41] Y. Nagai, T. Yamamoto, T. Tanaka, S. Yoshida, T. Nonaka, T. Okamoto, A. Suda,
M. Sugiura, J. Synchrotron Rad. 8 (2001) 616.
(
(
35] L.N. Ikryannikova, A.A. Aksenov, G.L. Markayan, G.P. Muravieva, B.G. Kostyuk,
A.N. Kharlanov, E.V. Linina, Appl. Catal. A 210 (2001) 225.
[42] Y. Nagai, T. Yamamoto, T. Tanaka, S. Yoshida, T. Nonaka, T. Okamoto, A. Suda,
M. Sugiura, Catal. Today 74 (2002) 225.