Nanophase fluorite-structured CeO2-ZrO2 catalysts prepared by high-energy mechanical milling: Analysis of low-temperature redox activity and oxygen storage capacity

Article abstract of DOI:10.1006/jcat.1997.1705

The utilization of mechanical milling for the preparation of catalysts based on ceria structurally modified with zirconia is presented. It is shown that room-temperature high-energy ball milling is an effective tool for the synthesis of nanophase CeO

Full text of DOI:10.1006/jcat.1997.1705

JOURNAL OF CATALYSIS 169, 490–502 (1997)
ARTICLE NO. CA971705
–
Nanophase Fluorite-Structured CeO2 ZrO2 Catalysts Prepared
by High-Energy Mechanical Milling
Analysis of Low-Temperature Redox Activity and Oxygen Storage Capacity
ꢀ
ꢀ
ꢀ
Alessandro Trovarelli, Francesca Zamar, Jordi Llorca,† Carla de Leitenburg,
ꢀ
Giuliano Dolcetti, and Janos T. Kiss‡
ꢀ
Dipartimento di Scienze e Tecnologie Chimiche, Universita` di Udine, via Cotonificio 108, 33100 Udine, Italy; †Departament de Quimica
Inorga`nica, Universitat de Barcelona, Diagonal 647, 08028 Barcelona, Spain; and ‡Department of Organic Chemistry,
Jozsef Attila University, Szeged, Dom ter 8, H-6720, Hungary
Received January 2, 1997; revised March 12, 1997; accepted March 12, 1997
tively. These redox properties are strongly enhanced if for-
eign cations such as Zr (3, 4), Gd (5), Pr (6), Tb (7), and
Pb (8) are introduced into the CeO2 lattice by forming solid
solutions. This is the result of enhanced oxygen ion mobil-
ity inside the modified fluorite lattice originating from the
formation of a defective, fluorite-structured solid solution
in which cations of different radius and/or charge have re-
placed cerium atoms. Much effort has been focused on the
preparation of these mixed oxides, which are potentially
useful as catalysts for sensors, for carrying out selective ox-
idations, and for providing low ignition temperatures for
catalyzed combustion reactions. The goal here is to lower
the temperature required for O2 diffusion through the cata-
lyst, which could lead to greater catalyst efficiency at lower
temperatures. However, further development of these ma-
terials requires detailed information on their structural fea-
tures and above all a precise knowledge of the role of their
various components and the influence of preparation pro-
cedure on redox behavior.
The utilization of mechanical milling for the preparation of cata-
lysts based on ceria structurally modified with zirconia is presented.
It is shown that room-temperature high-energy ball milling is an ef-
–
fective tool for the synthesis of nanophase CeO2 ZrO2 solid solution
in a wide composition range. The use of combined X-ray diffraction
analysis, Raman spectroscopy, and electron microscopy indicate
that the milling process induces the formation of true solid solutions
with a contraction of the cell parameter for cubic ceria following
the introduction of Zr into the lattice. This, in turn, remarkably en-
hances the oxygen storage/transport and redox capacity compared
to pure ceria and zirconia or to a simple mixture thereof. An un-
usual resistance to high-temperature cycling was also evidenced.
These features were analyzed by the study of the reduction profile
of doped ceria carried out by temperature-programmed reduction
at different milling times. The oxygen storage capacity (OSC) of the
catalysts was also evaluated; both the total and the kinetic acces-
sible OSC indicated that the best composition is CexZr1ꢁxO2 with
x > 0.5. This was correlated to the structural features and to the
presence of a high concentration of ions with redox character (i.e.,
c
Ce4+ ions) which favor oxygen mobility.
ꢂ 1997 Academic Press
Special attention has been focused in recent years on the
preparation of ceria structurally doped with ZrO2. These
materials show enhanced thermal (9), redox (3, 4, 7, 10),
and catalytic properties (11, 12) compared to pure undoped
ceria and their commercial development is being actively
INTRODUCTION
Materials containing ceria have been the subject of thor- pursued, especially for the auto-exhaust market (13–15).
ough investigation in recent years because of their useful- Several approaches have been used to prepare these ma-
ness in a number of applications in different areas of chem- terials, from high-temperature calcination of a mixture of
istry. One such area is catalysis, where ceria is currently the oxides (3), which gives materials with very low surface
being used as an active component in the so-called three- areas, to coprecipitation followed by supercritical drying of
way catalysts for automotive exhaust treatment as well as the hydrogels to produce high surface area aerogels (16).
in several other catalyst formulations (1, 2). One of the High-energy mechanical milling (MM) has also been shown
most important roles of CeO2 in these multicomponent to be a promising technique for the generation of nano-
systems is to provide surface active sites and to act as an structured metal and metal-oxide phases (17). MM con-
oxygen storage/transport medium by shifting between Ce3+ sists of a high energy ball milling process that can be used
and Ce4+ under reductive and oxidizing conditions, respec- to produce complex materials from a blend of elemental
490
0021-9517/97 $25.00
c
Copyright ꢂ 1997 by Academic Press
All rights of reproduction in any form reserved.

–
REDOX BEHAVIOR OF CeO2 ZrO2 PREPARED BY MECHANICAL MILLING
491
powders. Since its introduction in the early 1970s by structured material corresponding to {111}, {200}, {220},
Benjamin (18), there have been a number of mechano- {311}, {222}, and {400} diffraction planes.
chemical studies of the synthesis of nonequilibrium struc-
Samples for scanning electron microscopy analysis
tures and microstructures, including amorphous alloys, (SEM) were mounted on stubs and metalized with gold by a
solid solutions alloys, nanocrystalline materials, intermetal- sputtering system device (S150A Edwards). They were ob-
lic compounds, and metastable phases. Several treatments served by a Stereoscan 250 scanning electron microscope
employing mills have been applied at various stages of (Cambridge Instrument).
mixed oxides preparation to enhance catalytic activity or
Conventional transmission electron microscopy com-
selectivity (19–21), for example by changing sample mor- bined with dispersive X-ray microanalysis (EDX) was car-
phology or by inducing surface modifications. Mixed oxides ried out with a Hitachi H 800-MT electron microscope
have also been prepared entirely by high-energy mechan- working at 200 kV and equipped with a Kevex analyti-
ical milling (22, 23) and recently mixed oxides containing cal system. High-resolution electron microscopy (HREM)
CeO2 and dopants such as ZrO2, TbOx, and HfO2 have been combined with EDX was performed using a Philips CM-30
specifically prepared for catalytic applications (7, 24).
electron microscope working at 300 kV (LaB6) with 0.20-
The aim of this paper is to characterize the properties of nm point-to-point resolution and equipped with a Link an-
–
CeO2 ZrO2 mixed oxides prepared by room-temperature alytical system. Samples were supported on carbon-coated
high-energy mechanical milling of cubic CeO2 and mono- gold or copper grids by depositing a drop of the specimen
clinic ZrO2. Particular attention was paid to the kinetic of suspended in methanol. The sample was then transported
the milling process and to the structural/morphological fea- to the microscope stage and the equipment was pumped
tures of the final material. Redox behavior of the catalysts down for several hours before observation. Magnification
was investigated in depth, with special focus on oxygen of HREM images and electron diffraction patterns were
storage capacity, both the total and the kinetically limited. calibrated with pure CeO2 (a = 0.541 nm) under the same
These studies yielded unambiguous evidence for the forma- electron optical conditions.
tion of a mixed-oxide phase in a wide composition range by
The Raman spectra were taken on a Bio-Rad Digilab
milling the pure oxides at room temperature. Finally, and Division dedicated FT-Raman spectrometer between 3500
in agreement with recent findings, the redox and oxygen and 70cmꢁ1 at 4cmꢁ1 optical resolution employing500-mW
storage properties of ceria are strongly enhanced if Zr is in- energy of a Nd:YVO4 excitation laser at the sample posi-
troduced into the CeO2 lattice, especially when the dopant tion. Glass capillary sample holders were used. The number
level does not exceed 50 mol% .
of accumulated scans was 512. The spectra were processed
employing the GRAMS/386-based software supplied with
the FT-Raman spectrometer.
Textural properties were measured with a Carlo Erba
Sorptomatic 1900 instrument using the BET method.
EXPERIMENTAL
Catalyst Preparation
CeO2 (99.999% , surface area 5–18 m2/g) was purchased
from Aldrich and used as received. ZrO2 was prepared by
calcination of hydrous zirconia (MEL Chemicals) at 973 K
(surface area 27 m2/g). Preparation of the mixed oxides
was carried out with a high-energy vibratory ball-mill (Spex
8000) at an oscillation frequency of 20 Hz and an amplitude
of approximately 20 cm. Powders were loaded into a 50-ml
ZrO2 vial equipped with six Y-doped, high wear-resistant
Zirconia balls (Ø 10 mm, Tosoh Corporation). The ball-to-
powder ratio was 18/1 (i.e., 18 g of balls to 1 g of powder)
and milling time ranged from 0.5 to 12 h.
Oxygen Storage Capacity (OSC) and Redox Activity
Capacity to store and release oxygen was measured un-
der conditions where both total and kinetic-accessible OSC
can be detected. Total OSC was measured after reducing
the catalyst (50 mg) at the appropriate temperature (650 K)
under H2 flow (2 h, 35 ml/min STP). Before measuring O2
uptake, the catalyst after reduction was treated at 650 K un-
der inert flow for 30 min, and then pulses of a mixture con-
taining oxygen (1% O2 in He) were injected up to the break-
through point. Dynamic OSC was measured using pulses of
oxygen (250 ꢂl, 1% O2 in He) and hydrogen (250 ꢂl) alter-
nately injected (every 3 min) into the catalyst bed at 650 K.
Structural and Morphological Studies
Powder X-ray data were collected with a Siemens pow- OSC was evaluated from oxygen consumption in the first
˚
der diffractometer using CuKꢀ radiation (ꢁ = 1.5418 A) O2 pulse following an hydrogen pulse.
by placing samples in aluminum sample holders. High-
Redox activity was measured by temperature-prog-
resolution powder XRD data for structural analysis were rammed reduction (TPR) experiments which were carried
collected over a 14-h period from 22 = 20–75 with a 0.02ꢃ out bytreatingsamples(50–70mg) in a U-shaped quartzmi-
step size. Signals were fitted with Voigt functions and data croreactor inserted into a furnace (300–1473 K, 10 K/min)
were collected from the six main reflections of a fluorite- with a mixture of 5% H2 in Ar (35 ml/min STP) and by

492
TROVARELLI ET AL.
monitoring H2 consumption with a thermal conductivity de-
tector. Quantitative calculations of the extent of reduction
were performed by calibrating the apparatus for hydrogen
consumption with reduction of CuO to Cu, as previously
specified (25).
RESULTS AND DISCUSSION
Phase Composition of Mixed Oxides
Figure 1 reports the X-ray diffraction patterns of pow-
ders obtained by ball milling monoclinic ZrO2 (50% mol)
and cubic CeO2 (50% mol). Diffractograms were collected
after 0, 0.5, 2, 4, and milling for 9 h. The simple mixture
of CeO2 and ZrO2 (i.e., without milling) showed an XRD
pattern which is the sum of the diffractions of cubic CeO2
and monoclinic ZrO2. After milling for 2 h, a modification
in the X-ray pattern indicated that a solid-state reaction
had occurred. The main peaks relative to cubic CeO2 and
monoclinic ZrO2 had partially disappeared and were re-
placed by broad signals, which are indicative of a modifica-
tion of the crystal structure. After 9 h milling, the XRD pat-
tern indicated the formation of a single-phase Ce0.5Zr0.5O2
solid solution. Prolonged milling produced no further vari-
ations in the diffraction signals observed. This is clear from
Fig. 2, where the diffraction peaks corresponding to the
{220} plane are reported at a number of milling times. Pro-
file (a), with a peak at 47.50ꢃ, corresponds to the difraction
of the {220} plane of CeO2 in a simple mechanical mixture
FIG. 2. Diffraction of {220} plane of cubic fluorite-structured material
containing mixture of 50% mol CeO2 and 50% mol ZrO2 (a), the same
milled for 0.5 h (b), 2 h (c), 4 h (d), 7 h (e), and 9 h (f). The peaks were
fitted with Voigt functions (——) and deconvolution (- - -) was performed
with PeakFit software.
of CeO2 and ZrO2. After 0.5 and 2 h milling, the intensity
of this signal decreased and the growth of a new broad fea-
ture centered at higher values of 22 was observed. These
modifications indicate that solid-state reaction induces a
partial amorphization and/or a partial modification of crys-
tal size and lattice parameters as a result of the mechanical
impact of the powders with the balls. It has been reported
that mechanical milling of powders induces fracture and
deformations through high-energy collisions between balls
and particles (17). This in turn leads to a modification of
crystallite size also detected by transmission electron mi-
croscopy analysis (see below). After 9 h milling, the shape
and position of the signal stabilized at a value of 49.20ꢃ.
A complete X-ray analysis of the six main reflections indi-
cated the formation of a cubic fluorite-structured cell with
a = 0.5243(5) nm.
Solid solutions with a CeO2 molar content of 20% and
80% were obtained using the same methodology. The ki-
netic of the process, as revealed by X-ray diffraction, in-
dicates that the transformation of starting materials into
–
–
Ce Zr O solid solution is complete in about 9–11 h. Ma-
jor peaks are indexed as a cubic fluorite-structured material
with a = 0.5336(8) nm for Ce0.8Zr0.2O2 and a = 0.5183(3) nm
FIG. 1. X-ray diffraction patterns of mixture of 50% mol cubic CeO2
(᭢) and 50% mol monoclinic ZrO2 (ꢀ) milled for 0 h (a), 0.5 h (b), 2 h (c),
4 h (d), and 9 h (e).

–
REDOX BEHAVIOR OF CeO2 ZrO2 PREPARED BY MECHANICAL MILLING
493
–
FIG. 3. Cell volume of (Ce,Zr) O solid solutions obtained by milling
ceria and zirconia for 9 h. Experimental values (᭺) vs values calculated
using the empirical relationship developed in Ref. (26) (ꢄ ꢄ ꢄ).
FIG. 4. Peak corresponding to diffraction of plane {400} in
Ce0.5Zr0.5O2 (a), and Ce0.2Zr0.8O2 (b) obtained by ball-milling ceria and
zirconia for 9 h. Solid lines represent the experimental signal fitted with
Voigt functions (r2 = 0.994), while dotted lines show the deconvolution
profile of Ce0.2Zr0.8O2.
for Ce0.2Zr0.8O2. The XRD lines of the fluorite phases are
extremely broad and in consequence the lattice constant
cannot be accurately determined. However, the values ob-
tained are in good agreement with those calculated by the
weak intensity scattering between 70 and 700 cmꢁ1 char-
acterize the Raman spectrum of pure CeO2. Weak bands
around 260 and 607 cmꢁ1 and a shoulder near 405 cmꢁ1 are
also observed. The main peak at 465 cmꢁ1 is due to the F2g
Raman active mode typical of a fluorite-structured mate-
rial. The spectrum of Ce0.8Zr0.2O2 features a sharp band
at 465 cmꢁ1 plus a strong intensity scattering in the region
–
empirical relationship developed by Kim for CeO2 ZrO2
(26), as may be seen in Fig. 3.
–
–
For the binary system Ce Zr O, there are three possible
structures: monoclinic, tetragonal, and cubic; and several
stable or metastable phases having tetragonal symmetry
were found and separatelycharacterized (27). At high CeO2
concentrations (CeO2 > 80% mol), the cubic phase is fa-
vored, whereas for ZrO2-rich solid solutions the monoclinic
phases are observed, although it is well established that cu-
bic ZrO2 is stabilized by incorporating a small amount of an
oxide like Y2O3 or CaO. For intermediate composition both
cubic and tetragonal phases were detected, and it is very dif-
ficult on the basis of X-ray diffraction experiments to distin-
guish between the two, owing to the small difference in the
cell parameters. The structural analysis of Ce0.2Zr0.8O2 de-
serves some additional comment with regard to the possible
coexistence of the tetragonal and cubic phases in this sam-
ple. Peak analysis does not reveal any splitting of the {200},
{220}, and {311} signals but comparison of the diffraction of
the {400} plane for Ce0.2Zr0.8O2 and Ce0.5Zr0.5O2 indicates
the presence of an additional component at lower 2 values
for Ce0.2Zr0.8O2. Deconvolution of the XRD profile (Fig. 4)
is consistent with the splitting of the peaks corresponding to
the {004} and {400} planes at 72.08ꢃ and 73.24ꢃ, indicating a
partial tetragonalization of the structure. Calculation of the
cell parameters using these peaks shows that a = 0.5165 nm
and c = 0.5237nm with c/a = 1.014. Raman spectra collected
in the region 100–900 cmꢁ1 (Fig. 5) confirm these findings.
The spectrum of pure ZrO2 features several bands in this re-
gion which are attributable to monoclinic ZrO2 (28), while
FIG. 5. Raman spectra of ZrO2 (a), CeO2 (b), Ce0.8Zr0.2O2 (c), and
Ce0.2Zr0.8O2 (d). Samples c and d are obtained by milling an appropriate
one sharp single band (located at 465 cmꢁ1) and a very amount of CeO2 and ZrO2 for 9 h.

494
TROVARELLI ET AL.
100–700 cmꢁ1, which probably belongs to a cubic, fluorite- out to study the homogeneity of samples and the formation
structured phase. In the Ce0.2Zr0.8O2 sample, the intensityof of true solid solutions. A typical electron micrograph ob-
the 465 cmꢁ1 band fades markedly and additional bands are tained by HREM of mechanically alloyed Ce Zr O solid
–
–
observed at approximately 613, 316, 254, 142, and 117 cmꢁ1
,
solution is shown in Figs. 7a and 7b. The sample comprises
as indicated in Fig. 5. The new spectral features are presum- individual crystalline particles with sizes ranging from 4 to
ably due to a partial tetragonalization of the sample. It is 15 nm. A considerable degree of lattice imperfection can
–
–
known that Raman spectralfeaturesoftetragonalCe Zr O be observed (see, for example, the arrows in Fig. 7b). The
samples are characterized by six bands in this region (29). degree of imperfection can be explained by the mechanism
The further three bands, located at 379, 191, and 179 cmꢁ1
,
responsible for material transfer during the milling pro-
indicate that a small amount of monoclinic ZrO2 is still cess, which may include continuous bond rupture and for-
present. The low intensity and broadness of these peaks mation, local temperature rises, microdeformation of the
are probably caused by the small crystallites generated by surface and hydrostatic stresses (31). EDX microanalysis
mechanical milling. For very small crystallites (8–15 nm), performed on individual particles (probe size 5–8 nm) re-
the lack of long range order relaxes momentum conserva- vealed in all cases the presence of a mixed oxide phase for
–
–
tion selection rules in the Raman scattering process, which Ce Zr O. Chemical composition within individual parti-
results in a shift and broadness of the peaks (30). Unfor- cles was kept constant as revealed by different sets of EDX
–
–
tunately none of these data are known for Ce Zr O solid profiles performed on individual grains. Homogeneity and
solution, which makes comparison impossible.
composition from one grain to another was found to be
within less than 10% deviation. A representative EDX pro-
file corresponding to a single particle in the mechanically
alloyed Ce0.8Zr0.2O2 solid solution is shown in Fig. 8. No seg-
regation of separate CeO2 or ZrO2 phases was observed in
any case by EDX, which was consistent with selected-area
diffraction experiments. However, a comparison with solid
solutions prepared by conventional coprecipitation method
did indicate a slightly lower degree of homogeneity, which
can be partly eliminated by prolonging milling time. This
lower homogeneity may be due to preferential enrichment
of one of the two components in some areas of the material
as a result of an incomplete reaction and/or the leaching
Morphological Features
The formation of solid solutions and morphological fea-
tures were also studied by electron microscopy and EDX.
The morphology of doped ceria powders was similar for
the three samples. Figure 6 reports the SEM micrograph
of Ce0.8Zr0.2O2, which shows that the sample comprises an
aggregation of packed particles only a few micrometers
large. This is in agreement with previous observations of
calcia- and magnesia-doped zirconia prepared by mechani-
cal milling (19). Detailed HREM-EDX analysis was carried
FIG. 6. SEM micrograph of Ce0.8Zr0.2O2 obtained by MM.

–
REDOX BEHAVIOR OF CeO2 ZrO2 PREPARED BY MECHANICAL MILLING
495
FIG. 7. Low magnification HRTEM image of Ce0.8Zr0.2O2 solid solution obtained by mechanical milling for 9 h (a) showing lattice imperfection
indicated by arrows (b).
of ZrO2 from the apparatus (balls and vial). One of the cation in composition. Nevertheless, as can be verified by
major drawbacks of this methodology is the presence of weight gain or loss after reaction, leaching from milling ap-
impurities such as iron and tungsten from the milling me- paratus is limited to less than 1% (i.e., less than 10 mg for
dia (17). To limit leaching, we used apparatus made from 1 g preparation).
– –
zirconia, which has two advantages over traditional metal
Measurements of the Ce Zr O solid-solution lattice
devices. The loss of material from the balls is limited and spacing were also performed on several crystallites rep-
in any case introduces not impurities as such but a modifi- resentative of the Ce0.8Zr0.2O2 sample (Fig. 9). This was

496
TROVARELLI ET AL.
done by dividing the distance between the lateral faces of a
given particle by the whole number of spacings as visualized
by HREM. The average value for (111) lattice spacing was
found to be 0.308 nm (corresponding to a = 0.533 nm). Tak-
ing into account that (111) lattice spacing for pure ceria is
0.312 nm (a = 0.541 nm), our estimate indicated a shrinking
of the CeO2 fluorite-type lattice when Zr4+ was introduced.
This is consistent with XRD analysis, where a cubic lattice
with a = 0.5336(8) nm was found.
Hydrogen Uptake/Reduction Features
The introduction of ZrO2 into the CeO2 lattice has re-
cently been reported to strongly affect the reduction fea-
tures of ceria (3, 4). This occurs through structural modifi-
cations of the fluorite-type lattice of ceria as a consequence
FIG. 8. Energy-dispersive X-ray profile corresponding to single par-
4+
of the substitution of Ce4+ (ionic radius 0.97 A) with Zr
˚
ticle of Ce0.8Zr0.2O2 obtained by milling the reagents for 9 h. Copper sig-
nals are due to the grid used in the specimen preparation for electron
microscopy.
˚
(ionic radius 0.84 A). The effect of this substitution is to
decrease cell volume, lowering the activation energy for
oxygen-ion diffusion within the lattice and consequently
favoring reduction. The introduction of Zr also enhances
the formation of structural defects which are expected to
play an important role in determining reduction/oxidation
behavior. We therefore analyzed the reduction features of
–
–
mechanically alloyed Ce Zr O samples by studying their
TPR profiles. The reduction profile of pure CeO2 is charac-
terized by two peaks (Fig. 10, curve a). There is one peak
at temperature of ca. 800 K, which is strongly dependent
FIG. 10. TPR of CeO2 (a), Ce0.2Zr0.8O2 (b), Ce0.5Zr0.5O2 (c), and
Ce0.8Zr0.2O2 (d). Samples b, c, and d are obtained by milling the appropri-
ate amount of CeO2 and ZrO2 for 9 h. Dotted line indicates temperature at
which hydrogen consumed equals that necessary for reduction of surface
according to Ref. (35).
FIG. 9. High-magnification HRTEM image of Ce0.8Zr0.2O2 particle
showing {111} planes.

–
REDOX BEHAVIOR OF CeO2 ZrO2 PREPARED BY MECHANICAL MILLING
497
on surface area and which is generally attributed to the
TABLE 1
reduction of the uppermost layers of Ce4+ in CeO2, and
another peak at higher temperature originated by reduc-
tion of the material in the bulk (32). The overall degree
of reduction is 48% at 1450 K, corresponding to the for-
mation of CeOx with x = 1.76. This was evaluated by inte-
grating the TPR profile and by assuming that all adsorbed
hydrogen contributes to ceria reduction. The TPR of me-
chanically alloyed ceria–zirconia samples shows completely
different behavior (Figs. 10b–10d). The onset of H2 con-
sumption in solid solutions is observed at ca. 500 K, several
degrees lower than in pure undoped CeO2, and it is not
severely influenced by cerium content. The appearance of
this strong component of H2 adsorption at lower temper-
atures can be related to several factors including a higher
degree of surface Ce4+ reduction at lower temperatures
and/or an increase of H2 adsorption to produce hydride-
like or bronze-like species in which hydrogen is inserted in
the lattice (33, 34). High H2 adsorption has recently been
–
–
Surface Area and Reduction Features of Ce Zr O
Prepared by MM
Surface area,a H2 consumed,
x in
y in
Sample
CeO2
Ce0.8Zr0.2O2
Ce0.5Zr0.5O2
Ce0.2Zr0.8O2
m2/g
ꢂmol/gb
CeOx ZrO2 CeZrOy
c
d
–
5–18
27–33
16–20
9–13
14–18
20–25
27
1336(93)
1626(114)
1388(97)
669(47)
—
1.76
1.67
1.59
1.555
—
1.76
1.736
1.795
1.911
—
e
e
e
e
Ce0.65Zr0.35O2
Ce0.9Zr0.1O2
e
—
—
—
—
—
—
ZrO2
a The samples are made in 1–2 g batches. These data specify the range of
surface areas obtained within several repeated preparations. In the case of
CeO2 the different areas correspond to different lot numbers.
b
The H2 uptake was calculated by integration of the TPR profile; the
error is reported in parentheses.
c Extent of reduction of CeO2 in Ce Zr O solid solutions.
–
–
d
–
–
Overall reduction in Ce Zr O solid solutions.
e Milling time, 9 h.
–
–
reported for Cu Ce O solid solutions in which stored hy-
drogen is then used to reduce cerium at lower tempera-
tures (33). To estimate the amount of hydrogen required
for reduction of surface Ce4+ in ceria–zirconia samples, we
used the approach developed by Johnson and Mooi for pure
CeO2 (35). This method, which relies on the quantification
of H2 adsorbed by TPR, is not entirely beyond criticism as
there is a possibility of other phenomena affecting overall
H2 consumption in high surface area samples (36). How-
ever the assumption that H2 consumed in the first peak can
be related to surface area has been successfully applied in
some cases (37). It is evident that in our case the reduction
of the first Ce4+ layer requires a negligible amount of hy-
drogen in comparison to H2 adsorption responsible for the
low-temperature TPR peak of ceria–zirconia. The temper-
ature at which hydrogen consumption coincides with that
required for reduction of all surface cerium is indicated in
Fig. 10. Clearly no separation between surface and bulk re-
duction can be made. We therefore attributed increased H2
adsorption mainly to reduction of Ce4+ on both surface and
bulk as a result of an increase in the oxygen diffusion rate,
although we cannot exclude that some adsorbed hydrogen
could be stored first and only desorbed or consumed for
reduction subsequently. The lower oxygen mobility in the
bulk and the low surface area could account for the absence
of low temperature features in pure CeO2. The degree of
reduction of Ce4+ at 1450 K, as estimated by integration,
ranges from 66% in Ce0.8Zr0.2O2 to 89% in Ce0.2Zr0.8O2 and
it is better detailed in Table 1.
that of pure ceria, since ZrO2 does not show any H2 up-
take in this region. As the milling time increases there is a
lowering of the intensity of this peak simultaneously with
the enhancement of the low-temperature peak. This causes
strong promotion of hydrogen uptake in milled samples
particularly in the 800–1100 K region, where the effect of
milling is evident (see Fig. 12). The reason for this enhance-
ment is the gradual introduction of Zr into CeO2 lattice,
with a concurrent reduction of particle size down to a few
nanometers, a process which is complete in about 9 h, as is
demonstrated by X-ray analysis.
The formation of solid solution in the mill is very well
highlighted by monitoring the H2 consumption in the TPR
of a mixture of 50 mol% of CeO2 and 50 mol% of ZrO2 sub-
jected to different milling times. This is reported in Fig. 11
for samples milled from 0 to 8 h. The simple mixture of
FIG. 11. TPR of mixture of 50% mol CeO2 and 50% mol ZrO2 milled
CeO2 and ZrO2 shows a TPR profile which is similar to for 0 (a), 0.5 (b), 1 (c), 2 (d), 3 (e), 5 (f), and 9 h (g).

498
TROVARELLI ET AL.
28 m2/g), while only a small shift of the maximum of the re-
duction peak is observed over Ce0.5Zr0.5O2. This indicates
that the structural defects created by milling do not strongly
–
–
modify the redox behavior of Ce Zr O solid solutions. It is
likely that under the conditions (especially high tempera-
ture) typical of a TPR analysis, defects created by milling re-
arranges to more stable, defect-free structures. Therefore, it
is mainly the chemical/structural nature of the compounds,
more than the presence of defects induced by milling which
determines the redox properties of the material, at least for
this range of temperatures.
–
Another interesting feature of CeO2 ZrO2 solid solu-
tions is their unusual redox behavior after high-temp-
erature redox cycling (10, 12), that is the reduction of Ce4+
in solid solution is not affected (12) or even promoted (10)
if samples are thermally treated at high temperatures. This
feature was tested also in our samples prepared by mechan-
ical milling. Figure 14 shows the repeated TPR cycles of a
FIG. 12. Hydrogen consumption calculated from TPR of mixture of
50% mol CeO2 and 50% mol ZrO2 milled for 0, 0.5, 1, 2, 3, 5, and 9 h.
–
mixture of 50% CeO2 50% ZrO2 after 0.5, 2, 4, and 9 h
of milling. In this case the samples have been subjected to
the following procedure: (i) milling for the time indicated,
(ii) first TPR from 300 to 1450 K, (iii) oxidation at 773 K
under air, and (iv) second TPR. As can be seen the first
treatment under TPR conditions does not influence the hy-
drogen uptake at low temperature, although the surface ar-
eas are strongly influenced by the thermal treatment (they
drop to less than 2 m2/g for all samples). This is in contrast
to what happens with pure ceria, in which thermal treat-
ments strongly affect redox behavior and repeated TPR
cycles dramatically reduce hydrogen adsorption at lower
temperature (10, 12). In this case almost the opposite be-
havior is observed, the residual high-temperature feature
In order to understand how milling (and not the chemi-
cal nature) of the material could modify the redox behavior
we carried out a TPR analysis on a sample of CeO2 and
Ce0.5Zr0.5O2 prepared by conventional coprecipitation and
then subjected to ball milling for 9 h. The results are re-
ported in Fig. 13. The reduction features of CeO2 and of
Ce0.5Zr0.5O2 are not strongly modified; an increase of the
low-temperature feature is observed on CeO2 (attributable
to an increase of surface area induced by milling, 6 vs
–
of the CeO2 ZrO2 mixture milled for 9 h disappeared af-
ter the first thermal cycle. This can be explained by a slight
rearrangement of atoms from their original position driven
FIG. 13. Effect of milling on the TPR of ceria and ceria–zirconia pre-
pared by conventional route. (a) TPR of CeO2 (Aldrich, 6 m2/g), (b) TPR
of sample a after milling for 9 h; (c) TPR of Ce0.5Zr0.5O2 prepared by
coprecipitation (68 m2/g); (d) TPR of sample c milled for 9 h.
FIG. 14. TPR of mixture of 50% mol CeO2 and 50% mol ZrO2 milled
for 0, 0.5, 2, 4, and 9 h. Fresh samples (A) and recycled samples (B).

–
REDOX BEHAVIOR OF CeO2 ZrO2 PREPARED BY MECHANICAL MILLING
499
by temperature (10) or by a more complete degree of for-
mation of solid solution induced by treating at 1450 K. As
a matter of fact, samples milled for 0.5 and 2 h, under con-
ditions where solid solution cannot be completely formed,
strongly increase their reduction features at low temper-
ature after the first TPR. Overall this remarkable redox
behavior, which is unaffected by textural properties, and
thus not dependent on surface area, can be explained by
the increase of bulk anion mobility into a modified ceria.
Oxygen Storage/Release Properties
One of the main roles of ceria in auto-exhaust catalysts is
that of rapidly switching between CeO2 and CeO2ꢁx under
lean and rich conditions, respectively (2). In pure ceria, this
process is strongly dependent on surface redox phenomena
(38). A reduction of surface area therefore has a negative
impact on the capacity to rapidly store and release oxygen.
With the introduction of dopants into the CeO2 lattice there
are at least two beneficial effects: (i) an increase in ther-
mal stability which reduces surface-area loss upon aging;
and (ii) an increased diffusion rate of oxygen which posi-
tively influences the rapid exchange of oxygen, not only on
the surface but throughout (3, 5, 39). High oxygen storage
capacity should enlarge the operating A/F window with-
out negatively affecting TWC performances. We therefore
FIG. 15. Total oxygen storage capacity of CexZr1ꢁxO2 calculated on
per gram basis (᭺), and on per cerium atom basis (ꢀ). Samples obtained
after 9 h of milling.
the formation of solid solutions. Whereas in the first case a
maximum is observed for CexZr1ꢁxO2 with 0.6 ꢅ x ꢅ 0.8, the
composition giving the largest adsorption per mole of Ce is
in the range x = 0.2–0.5. These results are in agreement with
the data obtained by TPR analysis. The lower the cerium
content, the higher the degree of reduction of Ce4+ in solid
solution. The total oxygen consumption needed for reoxi-
dation calculated per cerium atom is therefore expected to
increase at low Ce concentrations. Conversely, the oxygen
consumed per gram of catalyst increases as the concentra-
tion of redox sites on the samples rises, although the degree
of reduction of Ce4+ is lower.
–
studied the oxygen storage/release behavior of CeO2 ZrO2
mixed oxides prepared by mechanical milling under con-
ditions in which OSC is either limited by thermodynamic
factors (i.e., number of oxygen vacancies) or is kinetically
controlled. The former gives an estimate of the total O2 that
can be exchanged following reduction and depends on the
total number of reduced accessible sites present on the cata-
lyst (for example oxygen vacancies or Ce3+ in CeO2), while
the latter involves measuring oxygen that is kinetically ac-
cessible under conditions which are typical of auto-exhaust
catalysts, i.e., rapid oscillations of the gas mixture from a
reducing to an oxidizing environment.
A rather different picture is observed if only the kinet-
ically accessible OSC is measured (Fig. 16). In this case,
regardless of the set of data considered, a maximum of
O2 adsorption is observed at higher Ce content, with a
Total OSC for CexZr1ꢁxO2 is reported in Fig. 15. As indi-
cated under Experimental, between reduction and oxida-
tion the sample wastreated under inert flowat temperatures
varying from 650 to 873 K. This was done in order to ascer-
tain whether H2 adsorption/desorption could affect OSC
measurements. The oxygen uptake values were unaffected
by thermal treatments, indicating that reaction of adsorbed
H2 with O2 is unlikely under these conditions. This is also
consistent with the relatively low surface area of our sam-
ples and with the observation that H2 adsorption on CeO2
is strongly dependent on textural properties (40). Two sets
of data are represented in Fig. 15 to show the oxygen ad-
sorbed per gram of catalyst and per cerium atom. Oxygen
adsorption is at a minimum with pure CeO2 (ZrO2 does not
show any adsorption) and it gradually increases with the in-
troduction of ZrO2. Oxygen consumption, either calculated
FIG. 16. Dynamic oxygen storage capacity of CexZr1ꢁxO2 calculated
on per gram basis (᭺) and on per cerium atom basis (ꢀ). Samples obtained
on a per gram or on a per mole of Ce basis, is influenced by after 9 h of milling.

500
TROVARELLI ET AL.
composition Ce0.9Zr0.1O2, even though there is no substan- obtained with Ce-rich samples? To answer these questions
tial difference in samples with x = 0.5–0.9. This means that fully, we need to consider several important aspects in the
in CeO2-rich ceria–zirconia solid solutions, oxygen is much development of these materials, especially the structural
more readily and rapidly available than in samples where features and the role of the various components in the
ZrO2 content exceeds 50% . These data are in qualitative mixed oxide.
– –
agreement with those recently reported by Cuif et al. (14)
The reoxidation of Ce in Ce Zr O mixed oxides depends
where OSC was found to have a maximum at the composi- on a number of factors including the participation at various
tion Ce0.6Zr0.4O2. In the above case, however, the amount stages of the surface and the bulk of the oxide. The first
of adsorbed oxygen differs considerably from our findings, step is the surface activation of oxygen which may occur
a result which may be attributed to different experimen- through several intermediates depending on the electron
tal conditions (temperature and reductant/oxidant concen- donor ability of the ceria surface and gives as a final species
tration) and the different methods employed for catalyst a double charged oxygen anion (2):
preparation. Similar systems have been recently analyzed
+4eꢁ
O2ads ꢄ ꢄ ꢄ
ꢄ ꢄ ꢄ 2O2ꢁ
also by Tanaka et al. (41). They found that fresh, Pt-loaded
Ce0.8Zr0.15Y0.05 has the best OSC behavior compared to
catalysts having lower cerium content. The effect of ther-
mal treatment in this case is to reduce oxygen adsorption
capacity, especially, of samples having high cerium loading.
An enhancement of the H2 uptake and a beneficial effect
This process is followed by the migration of oxygen ions
into the bulk which, for a defective fluorite-structured solid
solution, can be described by a vacancy mechanism. It is
therefore expected that the smaller the cell volume, the
lower the energy required to hop oxygen through the lat-
tice. This would imply that an increase in ZrO2 content will
always produce an observable increase in dynamic OSC,
which is in contradiction with experimental results. Conse-
quently, additional factors must be considered to explain
this behavior. Structural analysis has shown that for low
CeO2 content, it is likely that under milling conditions the
formation of tetragonal solid solution takes place. Oxygen
diffusion in a tetragonal structure should be hindered for
two reasons. The first is the anisotropy of diffusion, which
will limit the overall diffusion rate in a polycrystalline mate-
rial because of the different orientation of the grain bound-
aries (3). The second reason is the effective dimension of
the channel radius through which oxygen migrates, which
is favored for a cubic structure. This could explain the poor
OSC behavior observed in Ce0.2Zr0.8O2 but not the excel-
lent oxygen adsorption of cubic, CeO2-rich, samples. A pos-
sible explanation for this behavior may lie in the consider-
ation that the transfer of oxygen anions into the bulk must
be accompanied by a counterflow of electrons towards the
surface (42). This requires the presence of chains of ions
with redox properties. In the present case, only Ce can help
charge transfer by electron exchange in accordance with the
following equations, which are similar to those described by
Sinev et al. for Pr-based oxides (42):
–
of CeO2 ZrO2 solid solution on the performances of Pt,Rh
TWC catalysts has been also recently reported by Nunan
et al. (13) which found that on aged catalysts, the introduc-
tion of more than 18% of ZrO2 into CeO2 lattice produces
important advantageson the chemicalpromotionaleffect of
CeO2, particularly with CO/NO mixtures. Likewise, Ranga
Rao et al. (15) found a strong enhancement of NO decom-
position rate over Rh/Ce0.6Zr0.4O2 reduced at 673 K, com-
pared to Rh/CeO2; the driving force for this enhancement
was attributed to the formation of oxygen vacancies and to
their participation in the catalytic cycle. The choice of this
composition was done on the basis of the observed redox
behavior, which showed a maximum over cubic CexZr1ꢁxO2
with x = 0.4–0.5 (3). Moreover, previous results (12) indi-
cated that the introduction of 20 mol% ZrO2 into CeO2
strongly affects the kinetic of oxygen adsorption by lower-
ing the activation energy for O2 uptake.
Overall previous results seem to indicate that there is a
broad range of composition where the oxygen adsorption
and catalytic features are influenced by formation of solid
solution. These compositions are dependent on several pa-
rameters such as the preparation method, which should
allow formation of a mixed oxide phase, the aging tem-
perature, which affect textural and morphological features,
and/or the presence of noble metals, which enhances oxy-
gen adsorption by cooperating synergetically with the sup-
port. In the present case, where thermal treatments are not
employed and in the absence of noble metals, the number
of variables reduces and the structural factors should play
the most important role.
Ce4a+ + Ce3b+
Ce4+ + O2ꢁ
Ce3+ + Ce4b+
*
)
a
Ce3+ + Oꢁ
*
)
It is therefore expected that the presence of chains of
neighboring Ce atoms will accelerate the process. On de-
creasing Ce content, a decrease in the average distance be-
tween cerium atoms should hinder electron transfer in the
lattice with a negative effect on oxygen adsorption capabil-
ities and so explain the increase in OSC that accompanies
To explain the different results obtained under dynamic
conditions, we need answer to the following questions:
(i) why does Ce0.2Zr0.8O2 show extremely poor O2 adsorp-
tion under a kinetically controlled regime even when cal-
culated on a Ce basis? (ii) why are the best performances

–
REDOX BEHAVIOR OF CeO2 ZrO2 PREPARED BY MECHANICAL MILLING
501
an increase in cerium content. Another contribution to elec- if not all the collisions contribute to the formation of solid
tronic conductivity may come from the dimensions of the solutions, it will be necessary for a few to possess sufficient
crystallites prepared by mechanical milling. It has recently kinetic energy to produce the temperature rises necessary
been reported that the transport properties of nanocrys- for mechanical alloying to take place.
talline CeO2ꢁx are greatly enhanced by reduced resistance
in grain boundaries and higher electronic conductivity (43).
These properties, which are attributed to the formation of
interfacial defects, are also expected to play an important
role here, where materials with a high number of defects
are generated. The presence of defects and lattice mis-
match in the development of materials with a high oxy-
gen storage/release capacity has also been recognized by
Egami et al. (44), who found that the presence of nanoscale
segregated CeO2 phase in close contact with ZrO2 and/or
CONCLUSIONS
Summarizing, mechanical milling is an appropriate new
tool for preparing mixed-oxide catalysts with redox behav-
ior similar to that observed over solid solutions prepared
by conventional methods. The following factors need to be
highlighted:
ꢆ Dissolution of ZrO2 into CeO2 lattice was observed
in all the composition range with formation of a fluorite-
structured, cubic solid solution for ZrO2 content not ex-
ceeding 50 mol% and with a partial tetragonalization for
higher zirconium loading.
ꢆ The formation of solid solution was followed by X-
ray diffraction and temperature-programmed reduction. A
strong enhancement of the redox properties of CeO2 was
observed by introduction of ZrO2.
–
CeO2 ZrO2 enhances the OSC of ceria–zirconia mixtures.
The authors attributed the redox behavior of a nonhomo-
–
–
geneous Ce Zr O solid solution to an unusual nanoscale
phase segregation of CeO2 with exposure of high-energy
surfaces, where O2 removal is favored, therefore minimiz-
ing the importance of having an homogeneous single phase.
Mechanics of the Milling Process
ꢆ The capacity of reversibly adsorbing oxygen was also
measured. Under dynamic conditions CeO2-rich solid so-
lutions showed the best behavior, which can be correlated
to a high oxygen mobility and to the presence of a large
number of atoms with redox character.
Although the main purpose of this paper was to present
a novel method for preparing ceria–zirconia mixed-oxide
catalysts with a high oxygen storage capacity and enhanced
redox features, we would like to focus here on the mecha-
nisms that may be considered to explain the process.
In the mill the central event which determines micro-
scopic deformation, with continuous bond rupture and for-
mation, is the ball-powder-ball collision. These collisions
are, in turn, influenced by several macroscopic parameters
such as oscillation frequency, milling media, charge ratio
(total ball mass/powder mass) and time. Under our exper-
imental conditions the number of ball-to-ball impacts may
be estimated at around 800/sec (31). Therefore, in a 9-h
experiment, approximately 25 million collisions will take
place. If we assume that every collision is effective in in-
ducing breaking and welding of particles, this number is
not high enough to explain an homogeneous formation of
a mixed-oxide phase. Other factorsmust also be considered.
Since melting, mixing, and solidification is the most impor-
tant sequence that leads to the formation of solid solution,
one possible mechanism is the local surface melting that
occurs during impaction of the powder. Although this as-
sumption is reasonable, it must be excluded because of the
high temperatures that would be required for the process
and which are well above those estimated for a mill. Local
powder temperature rises during impact resulting from the
kinetic energy of the milling balls cannot be higher than
200–300 K and are not therefore sufficient to induce ox-
ide surface melting. However, this local temperature rise
can induce an enhancement of microdiffusion and plastic
deformation even in brittle materials (31) which can con-
tribute to the formation of a mixed-oxide phase. Thus even
Finally, the possibility of modifying lattice oxygen mobil-
ity and the kinetic of Ce(IV) reduction by introduction of
small quantities of dopants is a powerful additional tool in
the design of innovative catalytic materials with elevated
oxygen storage/transport capacity.
ACKNOWLEDGMENTS
The authors thank MURST and CNR for financial support. MEL chem-
icals (UK) is also acknowledged for providing samples of zirconia. The
FT-Raman spectrometers were purchased with the help of the PHARE-
ACCORD Program H-9112-0152.
REFERENCES
1. Taylor, K. C., “Catalysis, Science and Technology” (J. R. Anderson
and M. Boudart, Eds.), p. 119. Springer Verlag, Berlin, 1984.
2. Trovarelli, A., Catal. Rev.-Sci. Eng. 38, 439 (1996).
3. Fornasiero, P., Di Monte, R., Ranga Rao, G., Kaspar, J., Meriani, S.,
Trovarelli, A., and Graziani, M., J. Catal. 151, 168 (1995).
4. Murota, T., Hasegawa, T., Aozasa, S., Matsui, H., and Motoyama, M.,
J. Alloys Comp. 193, 298 (1993).
5. Cho, B. K., J. Catal. 131, 74 (1991).
6. Logan, A. D., and Shelef, M., J. Mater. Res. 9, 468 (1994).
7. Zamar, F., Trovarelli, A., de Leitenburg, C., and Dolcetti, G., Stud.
Surf. Sci. Catal. 101, 1283 (1996).
8. Zhang, Y., Andersson, A., and Muhammed, M., Appl. Catal. B: Env-
iron. 6, 325 (1995).
9. Pijolat, M., Prin, M., Soustelle, M., Touret, O., and Nortier, P., J. Chem.
Soc. Faraday Trans. 91, 3941 (1995).

502
TROVARELLI ET AL.
10. Fornasiero, P., Balducci, G., Di Monte, R., Kaspar, J., Sergo, V., 28. Keramidas, V. G., and White, W. B., J. Am. Ceram. Soc. 57, 22
Gubitosa, G., Ferrero, A., and Graziani, M., J. Catal. 164, 173 (1996). (1974).
11. Zamar, F., Trovarelli, A., de Leitenburg, C., and Dolcetti, G., J. Chem. 29. Kim, D.-J., Jung, H.-J., and Yang, I.-N., J. Am. Ceram. Soc. 76, 2106
Soc. Chem. Commun. 965 (1995).
(1993); Hirata, T., Asari, E., and Kitajima, M., J. Solid State Chem.
12. de Leitenburg, C., Trovarelli, A., Llorca, J., Bini, G., and Cavani, F.,
Appl. Catal. A: Gen. 139, 161 (1996).
110, 201 (1994).
30. Richter, H., Wang, Z. P., and Levy, L., Solid State Commun. 39, 625
13. Nunan, J. G., Williamson, W. B., and Robota, H. J., SAE Paper 960798,
1 (1996).
14. Cuif, J., Blanchard, G., Touret, O., Marczi, M., and Que´me´re´, E., SAE
Paper 961906, 73 (1996).
(1981).
31. Davis, R. M., McDermott, B., and Koch, C. C., Metall. Trans. A 19A,
2867 (1988).
32. Yao, H. C., and Yao, Y. F. Y., J. Catal. 86, 254 (1984).
15. Ranga Rao, G., Fornasiero, P., Di Monte, R., Kaspar, J., Vlaic, G., 33. Wrobel, G., Lamonier, C., Bennani, A., D’Huysser, A., and Aboukaı¨s,
Balducci, G., Meriani, S., Gubitosa, G., Cremona, A., and Graziani,
M., J. Catal. 162, 1 (1996).
16. Sun, Y., and Sermon, P. A., J. Mater. Chem. 6, 1025 (1996).
17. Koch, C. C., and Whittenberger, J. D., Intermetallics 4, 339 (1996).
18. Benjamin, J. S., Metall. Trans. 1, 2943 (1970).
A., J. Chem. Soc. Faraday Trans. 92, 2001 (1996).
34. Cunningham, J., O’Brien, S., Sanz, J., Rojo, J. M., Soria, J. A., and
Fierro, J. L. G., J. Mol. Catal. 57, 379 (1990).
35. Johnson, M. F. L., and Mooi, J., J. Catal. 103, 502 (1987); Johnson,
M. F. L., and Mooi, J., J. Catal. 140, 612 (1993).
19. Zazhigalov, V. A., Haber, J., Stoch, J., Bogutskaya, L. V., and
Bacherikova, I. V., Appl. Catal. A: Gen. 135, 155 (1996).
20. Mori, S., Xu, W.-C., Ishidzuki, T., Ogasawara, N., Imai, J., and
Kobayashi, K., Appl. Catal. A: Gen. 137, 255 (1996).
36. Zotin, F. M. Z., Tournayan, L., Varloud, J., Perrichon, V., and Frety,
R., Appl. Catal. A: Gen. 98, 99 (1993).
37. Perrichon, V., Laachir, A., Bergeret, G., Fre´ty, R., Tournayan, L., and
Touret, O., J. Chem. Soc. Faraday Trans. 90, 773 (1994).
21. Trovarelli, A., Matteazzi, P., Dolcetti, G., Lutman, A., and Miani, F., 38. Schmieg, S. J., and Belton, D. N., Appl. Catal. B: Environ. 6, 127
Appl. Catal. A: Gen. 95, L9 (1993). (1995).
22. Michel, D., Faudot, F., Gaffet, E., and Mazerolles, L., J. Am. Ceram. 39. Miki, T., Ogawa, T., Haneda, M., Kakuta, N., Ueno, A., Tateishi, S.,
Soc. 76, 2884 (1993). Matsura, S., and Sato, M., J. Phys. Chem. 94, 6464 (1990).
23. Chen, Y. L., Qi, M., and Yang, D. Z., and Wu, K. H., Mater. Sci. Eng. 40. Bernal, S., Calvino, J. J., Cifredo, G. A., Gatica, J. M., Perez Omil,
A 183, L9 (1994). J. A., and Pintado, J. M., J. Chem. Soc. Faraday Trans. 89, 3499 (1993).
24. de Leitenburg, C., Trovarelli, A., Zamar, F., Maschio, S., Dolcetti, G., 41. Tanaka, H., and Yamamoto, M., SAE Paper 960794, 231 (1996).
and Llorca, J., J. Chem. Soc. Chem. Commun. 2181 (1995).
25. Trovarelli, A., de Leitenburg, C., Dolcetti, G., and Llorca, J., J. Catal.
151, 111 (1995)
42. Sinev, M. Yu., Graham, G. W., Haach, L. P., and Shelef, M., J. Mater.
Res. 11, 1960 (1996).
43. Chiang, Y.-M., Lavik, E. B., Kosacki, I., Tuller, H. L., and Ying, J. Y.,
Appl. Phys. Lett. 69, 185 (1996).
26. Kim, D.-J., J. Am. Ceram. Soc. 72, 1415 (1989).
27. Yashima, M., Arashi, H., Kakihana, M., and Yoshimura, M., J. Am. 44. Egami, T., Dmowski, W., and Brezny, R., SAE Paper 970461, 000
Ceram. Soc. 77, 1067 (1994). (1997).