Synthesis and characterization of mixed-morphology CePO4 nanoparticles

Article abstract of DOI:10.1016/j.jssc.2006.12.009

Cerium phosphate nanoparticles with diameters of 10-180 nm were synthesized by a variety of solution techniques. X-ray diffraction (XRD) determined the crystalline phase(s) present in each sample. Population, shift, and spin-lattice relaxation ³¹P solid-state nuclear magnetic resonance (NMR) measurements accounted for all the ³¹P nuclei expected in each sample, and were able to distinguish between phosphorous nuclei in different environments and phases. Transmission electron microscopy (TEM) characterized the morphology and crystallinity of the powder samples as well as of the sintered compacts of the powders. In conjunction with TEM, energy-dispersive spectroscopy (EDS) provided a measure of the composition of the bulk intergranular regions within each CePO₄ sample. The presence of an amorphous, phosphate-rich intergranular phase was found in those samples prepared by dissolution of ceria in H₃PO₄ under various conditions.

Full text of DOI:10.1016/j.jssc.2006.12.009

ARTICLE IN PRESS
Journal of Solid State Chemistry 180 (2007) 840–846
www.elsevier.com/locate/jssc
Synthesis and characterization of mixed-morphology
CePO4 nanoparticles
Ã
L. Karpowicha, S. Wilckeb, Rong Yuc, G. Harleya, J.A. Reimerb, L.C. De Jonghea,c,
aDepartment of Materials Science and Engineering, University of California at Berkeley, Berkeley, CA 94720, USA
bDepartment of Chemical Engineering, University of California at Berkeley, Berkeley, CA 94720, USA
cMaterials Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA 94720, USA
Received 1 September 2006; received in revised form 29 November 2006; accepted 2 December 2006
Available online 15 December 2006
Abstract
Cerium phosphate nanoparticles with diameters of 10–180 nm were synthesized by a variety of solution techniques. X-ray diffraction
(XRD) determined the crystalline phase(s) present in each sample. Population, shift, and spin-lattice relaxation 31P solid-state nuclear
magnetic resonance (NMR) measurements accounted for all the 31P nuclei expected in each sample, and were able to distinguish between
phosphorous nuclei in different environments and phases. Transmission electron microscopy (TEM) characterized the morphology and
crystallinity of the powder samples as well as of the sintered compacts of the powders. In conjunction with TEM, energy-dispersive
spectroscopy (EDS) provided a measure of the composition of the bulk intergranular regions within each CePO4 sample. The presence of
an amorphous, phosphate-rich intergranular phase was found in those samples prepared by dissolution of ceria in H3PO4 under various
conditions.
r 2006 Elsevier Inc. All rights reserved.
Keywords: CePO4; Cerium phosphate; Monazite; Nanoparticle; 31P solid-state NMR
1. Introduction
synthesized by different methods and examined by X-ray
Diffraction (XRD), transmission electron microscopy
(TEM), energy-dispersive spectroscopy (EDS), and 31P
solid-state nuclear magnetic resonance (NMR).
Cerium orthophosphate is a synthetic analog of the
naturally occurring mineral monazite, which often contains
trace amounts of La, Th, and other rare-earth elements
[1,2] and possesses the most stable chemical properties of
naturally occurring minerals [3]. Monazite cerium phos-
phate is also a heat-stable phosphate with a melting
temperature close to 2000 1C [2], making CePO4 attractive
for use in high-temperature, ceramic applications. Cerium
phosphate has been studied as a potential storage material
for nuclear waste [4], a poison to automobile catalyst [5,9]
and, more recently, as a potential proton conduction
membrane for use in hydrogen fuel cells [10]. Studies have
shown that the chemical nature and particle size of
synthesized phosphates is highly dependent on the proces-
sing technique [11–15]. To clarify the relationship between
the synthetic route and the chemical nature, CePO4 was
2. Experimental methods
Four different samples of CePO4, both commercially
purchased and in-house synthesized, were studied. Sample 1
was 99% pure cerium (III) phosphate powder in the
rhabdophane phase, as purchased from Alfa Aesar
(AA39215-22). Sample 2 was prepared by heating the as-
received powder at 800 1C in air for 8 h, resulting in a
transformation to the monazite phase. Samples 3 and 4
were synthesized via reaction between CeO2 powder
(Nanostructured & Amorphous Materials, Inc., 99.95%)
and liquid H3PO4 (J.T. Baker Analyzed ACS Reagent,
85%). Cerium oxide and phosphoric acid were combined in
acetone with a ratio of P:Ce ¼ 1:1, and the suspensions
were attritor milled for 2 h at 400 RPM. After drying
overnight, the resulting powders were ground in an alumina
Ã
Corresponding author. Fax: +510 486 4881.
E-mail address: dejonghe@lbl.gov (L.C. De Jonghe).
0022-4596/$ - see front matter r 2006 Elsevier Inc. All rights reserved.
doi:10.1016/j.jssc.2006.12.009

ARTICLE IN PRESS
L. Karpowich et al. / Journal of Solid State Chemistry 180 (2007) 840–846
841
Fig. 1. X-ray diffraction (XRD) measurements of rhabdophane CePO4 (Sample 1), monazite CePO4 (Sample 2), and mixtures of CePO4 and CeO2
(Samples 3 and 4).
Fig. 2. Transmission electron microscopy (TEM) images showing the morphology of Samples 1–4. Sample 1: TEM image of shows a needle-shaped
rhabdophane crystallite supported on an amorphous carbon film. Sample 2: After 8 h at 800 1C, the as-received powder has transformed into equiaxed
monazite powder. Sample 3: the larger particles are monazite CePO4, while the smaller particles are unreacted CeO2. Amorphous material is also present
between the grains (marked as a) Sample 4: powder composite particle after reaction of a ceria powder with phosphoric acid at 100 1C. The darker core
particle (ꢀ50 nm) is unreacted or partly reacted CeO2, while the lighter, outer particles (ꢀ10–20 nm) are monazite (CePO4) that accumulated around the
ceria particle either as a result of the reaction topology or as a result of hetero-agglomeration.

ARTICLE IN PRESS
L. Karpowich et al. / Journal of Solid State Chemistry 180 (2007) 840–846
842
this microscope is 0.23 nm and the information limit is
0.16 nm. Chemical composition analysis was done by EDS.
31P NMR spectra were acquired at 1.54 T in a Doty
probe tuned to a frequency of 26.56 MHz using a home-
built spectrometer with a Tecmag console. Shift and
population measurements were referenced to aqueous
85% H3PO4 (Fisher-Scientific). A 1.7 ms 901 single pulses
was used for the shift, peak width, and population
measurements of each sample. An inversion recovery
experiment determined the spin-lattice relaxation time,
T1, while a Hahn echo with varying t delays measured the
spin–spin relaxation time, T2. All shift, peak width, T1 and
T2 measurements were made both static and while spinning
at 11 kHz. While spinning at 11 kHz, t delays were
rotosynchronized for the inversion recovery and Hahn
echo experiments. A recycle delay exceeding 5ÃT1 was set
for every pulse sequence. All experiments were repeated
with an empty rotor to verify that the signal detected was
solely from the CePO4 sample being studied.
3. Results and discussion
Fig. 3. Transmission electron microscopy (TEM) lattice images of CePO4
particles in Samples 1–4. Sample 1: crystalline rhabdophane needle on
amorphous support film. Sample 2: monazite particle; no amorphous
phosphate phase is present. Sample 3: crystalline monazite; in addition,
TEM also found amorphous phosphate in between grains (see also Fig. 2-
Sample 3). Sample 3: the (1 1 0) CePO4 plane is indicated in the high
resolution TEM image of a crystalline monazite. Sample 4: TEM close-up
of the small, crystalline monazite particles surrounding the central ceria
particle shown in Fig. 2-Sample 4.
3.1. X-ray diffraction
The XRD results in Fig. 1 show Sample 1 to be phase-
pure rhabdophane CePO4. Rhabdophane (PDF 35-0614) is
a hydrated, hexagonal form of cerium phosphate, typically
expressed as CePO4ÃxH2O with x ¼ 0.3ꢁ0.5 [15]. After
heating the same powder at 800 1C for 8 h, an irreversible
transformation to the monazite phase of CePO4 occurs
(Sample 2), as expected based on previous work [13,14].
Monazite (PDF 32-0199) is the monoclinic structure of
unhydrated pure CePO4. No residual rhabdophane peaks
are observed in this sample, confirming that the transfor-
mation to monazite has gone to completion. Monazite
CePO4 and residual CeO2 were detected by XRD for both
Samples 3 and 4 that were prepared by the aqueous acid
synthesis procedure. The amount of residual CeO2 in
Sample 3 is lower relative to Sample 4. This is likely due to
the significantly lower temperatures used for the synthesis
of Sample 4.
mortar and pestle and sieved with US Standard #325
stainless steel sieve. Sample 3 powder was then uniaxially
die-pressed at 3000 psi followed by a cold isostatic press at
180 kpsi. The resulting pellets were calcined and free-
sintered at 800 1C for 20 h. Sample 4 was similarly
synthesized by placing CeO2 and H3PO4 (P:Ce ¼ 1:1) in
acetone, attritor milling, then heating to 100 1C for 2 h with
magnetic stirring, and drying overnight under ambient
pressure and temperature. The powder was then ground,
and sieved as described above.
The four resulting powders, as well as Sample 3 pellets,
were characterized with XRD to identify the crystalline
phases present. A Siemens D500 diffractometer and
Phillips Analytical X’Pert PRO MP diffractometer were
used with CuKa radiation, and step size of 0.0512y from
101 to 701. Patterns were matched using the international
center for diffraction data (ICDD)–powder diffraction files
(PDF) database.
Powder samples 1, 2, and 4 were prepared for TEM
analysis by dropping dilute specimen suspensions on
copper grids covered with amorphous carbon films. TEM
foils were prepared from Sample 3, first grinding slices
down to 80 mm and mechanically dimpling these to 10 mm,
followed by ion milling to electron transparency. The TEM
experiments were carried out using a Philips CM200
microscope, operating at 200 kV. The point resolution of
3.2. Transmission electron microscopy (TEM)
TEM probed the morphology, composition, and crystal-
linity of the CePO4 powders. The TEM results in Figs. 2
and 3 show that Sample 1 consists of 10–20 nm by
30–300 nm rod-shaped CePO4 rhabdophane crystals with
no detectable amorphous material. Sample 2 consists of
30–80 nm approximately equiaxed monazite crystallites,
and also shows no evidence of amorphous phases. The
sintered pellet of Sample 3 powder consists of 150–200 nm
equiaxed CePO4 grains as well as unreacted 20–60 nm
CeO2 particles. In addition, the TEM shows the presence of
amorphous intergranular material. The amount of residual
CeO2 in Sample 3 is low; the XRD (Fig. 1) indicates
approximately 10% of CeO2 by volume and agrees with the

ARTICLE IN PRESS
L. Karpowich et al. / Journal of Solid State Chemistry 180 (2007) 840–846
843
Fig. 4. (a), (b): TEM images of amorphous triple junctions (marked a), and grain boundaries in Sample 3 pellet. (c): Typical EDS analysis of grain
boundary regions (including triple boundary junctions). The white arrow in (d) indicates where analysis was done. Note the presence of impurities evident
in the EDS spectrum, as well as the excess of phosphorus relative to CePO4 in the amorphous regions.
Table 1
Summary of physical properties of samples 1–4 as determined by XRD, TEM, and EDS
Sample
Composition
Crystal structure
Morphology
Particle size
Ce:P ratio
1
2
3
Rhabdophane CePO4ÃxH20
Monazite CePO4
Monazite CePO4/residual CeO2
Hexagonal (P6222)
Monoclinic (P21/n)
Monoclinic (P21/n)
Rod-shaped particles
Spherical particles
Spherical particles
10-20 nm diameter ꢂ 30–300 nm length
30–80 nm diameter
53:47
53:47
52:48
100–200 nm CePO4 grains
20–60 nm CeO2 grains
4
Monazite CePO4/CeO2 composite
Monoclinic (P21/n)
CePO4 shell
CeO2 core
10–40 nm thick CePO4 shell
30–40 nm diameter CeO2 core
Shell-51:49
Core-94:6
amount of ceria visible in Fig. 2 (3). Fig. 4 includes an EDS
analysis of the intergranular amorphous material in
Sample 3. The intergranular material is generally phos-
phate-rich, and its composition is not uniform throughout
the sample. Additionally, sodium and aluminum impurities
were detected, presumably introduced from the Pyrexs
sample container or the attritor milling process during
synthesis. Sample 4 contains 10–40 nm thick polycrystalline
CePO4 shells around a core of 30–40 nm diameter CeO2
particles. It is not clear if this represents an actual reaction
morphology, or if it is the result of hetero-agglomeration
during TEM specimen preparation. The EDS analysis
results show that the CePO4 phase in all of the samples can
be viewed as stoichiometric CePO4 within typical error of
EDS (about 2–3%).
The lowest formation temperature of monazite CePO4
reported to date, using a different synthetic route starting
with Ce(CH3COO)3Ã2H2O and (NH4)H2PO4 in diethylene
glycol, was at 180 1C [12]. Crystalline CePO4 in the
monazite phase is not typically observed to form from
the solution of CeO2 and H3PO4 until a transition
temperature of about 660 1C is reached [13], which is much
higher than the synthesis temperature of 100 1C for Sample
4. A plausible reason for the formation monazite at this
low temperature is that the high surface energy of the CeO2
nanoparticles promoted their partial dissolution in the

ARTICLE IN PRESS
L. Karpowich et al. / Journal of Solid State Chemistry 180 (2007) 840–846
844
phosphoric acid, forming CePO4 by a solution/precipita-
tion process (Table 1).
data 31P NMR signals quantitatively accounted for all
anticipated phosphorus atoms. For example, pure, non-
hydrated monazite CePO4 should contain 2.56Ã1021 31P
nuclei/g. Sample 2, confirmed as monazite CePO4 through
XRD and TEM, was observed to contain the same number
of nuclei/g within experimental uncertainty (Table 2).
3.3. Nuclear magnetic resonance
31P NMR peaks were observed for all four samples
(Fig. 5). The shift, relaxation times T1 and T2, peak widths,
and quantitative population measurements are summarized
for all of the observed 31P NMR peaks in Tables 2 and 3. Xu
et al. reported they were unable to obtain a 31P NMR signal
from non-hydrated monazite CePO4 [5], citing that the
paramagnetic interaction between the 31P nucleus and the
unpaired electron on Ce3+ atoms broadened the peak past
their spectral range. We found the static FWHH peak for the
monazite CePO4 to be ꢀ500 ppm broad, which is difficult to
detect with standard high-resolution NMR methods. In all
To assess further the role of paramagnetism in our NMR
data, we conducted variable temperature studies, and
found that the 31P NMR peak of monazite CePO4 shifts
with temperature (Fig. 6). Such temperature-dependent
shifts are consistent with Curie–Weiss paramagnetic
7
behaviour, as seen in Li and 17O studies of oxides [6,7].
Furthermore, the magnitude of the spin-lattice relaxation
times for all 31P lines (Table 2) are also consistent with the
presence of paramagnetic ions. The source of the para-
magnetism is likely Ce3+ ions.
Assignments of the 31P peaks to specific sites within the
phosphates proceeds as follows. The peaks at ꢁ87 to
ꢁ94 ppm may be assigned to the P nuclei experiencing
paramagnetic (Fermi-contact) shifts in the monazite CePO4
phase, as all samples containing this crystal structure
exhibited this signal in their spectra. An estimate of the
diamagnetic shift of 31P in monazite can be obtained by
taking the infinite temperature limit (1/T-0) of the data
shown in Fig. 6. The result is a shift of 1473 ppm for
Sample 3. This diamagnetic shift is comparable to the
chemical shifts of the Li, Na, and K orthophosphate [8].
Additionally, all samples containing the hydrated rhabdo-
phane phase of CePO4 showed peaks near ꢁ55 ppm, which
can be assigned to Fermi-contact shifted 31P nuclei in
hexagonal phase of cerium phosphate.
The 31P peaks at 9 and 1 ppm appear only in the samples
synthesized via aqueous reaction of CeO2 with phosphoric
acid. Given that the intensities of the NMR peaks match well
the overall composition of CePO4, we suggest that the peaks
at 1 and 9 ppm may be associated with the phosphate-rich
amorphous phases observed under TEM. For Sample 3, the
Table 3
Summary of 31P static NMR measurements of the four CePO4 samples
Fig. 5. 31P NMR spectra of the four CePO4 samples obtained in a 1.54 T
magnet while spinning the sample at the magic angle at 11 kHz. 0 ppm is
referenced to 31P in 85% H3PO4. The peak of Sample 3 near 0 ppm and the
peak of Sample 4 near ꢁ90 ppm were multiplied by 5 relative to the rest of
the spectra to make their shapes more apparent. Multiple spin-lattice (T1)
constants were measured for the peaks near 0 ppm, suggesting that the 31P
nuclei corresponding to this shift are in multiple chemical environments.
Sample
d (ppm)
T1 (ms)
T2 (ms)
FWHH (ppm)
1
2
3
4
ꢁ33
ꢁ72
ꢁ82
ꢁ5
8
14
0.18
0.19
0.2
456
548
510
156
14
13 (50%), 56 (50%)
0.2
Table 2
Summary of 31P 11 kHz MAS NMR measurements of the four CePO4 samples
Sample d (ppm) Peak assignment
T1 (ms)
T2 (ms) FWHH (ppm) Population ( ꢂ 1021 nuclei/g)
1
2
3
3
4
4
ꢁ55
ꢁ94
ꢁ91
9
Rhabdophane CePO4 bulk
Monazite CePO4 bulk
Monazite CePO4 bulk
10
15
14
1.7
4.1
4.2
62
25
29
35
70
18
2.6
2.5
1.9 (89% of total)
0.2 (11% of total)
0.8 (33% of total)
1.6 (67% of total)
Amorphous intergranular phosphate-rich phase 73 (39%), 890 (61%) 4.0
Monazite CePO4 bulk
Amorphous phosphate- rich phase
ꢁ87
1
16 2.0
11 (56%), 156 (44%) 0.5

ARTICLE IN PRESS
L. Karpowich et al. / Journal of Solid State Chemistry 180 (2007) 840–846
845
intensity ratio of the second peak is approximately 11% of
the total 31P signal, which correlates well with the amount of
unreacted ceria present (estimated at 10 vol%). As the P:Ce
ratio for the synthesis was 1:1, any unreacted oxide results in
a corresponding amount of free H3PO4. This acid is surmised
to form the intergranular phosphate-rich amorphous phase.
The intensity ratio of the peak near 0 ppm to the bulk
monazite peak at ꢁ87ppm for Sample 4 may be similarly
analysed. Approximately 33% of the P intensity is from the
peak at ꢁ87 ppm, which is associated with the bulk signal
from CePO4. The remaining 67% of the population with
shift at 9 ppm is inferred to be P nuclei in the amorphous
interface or surface regions in Sample 4.
amorphous phase are suggestive of paramagnetic relaxation,
yet the temperature dependence of these shifts (not shown)
was found to be very weak. Furthermore, the 31P NMR
peaks at 9 and 1 ppm (in Samples 3 and 4, respectively)
exhibit bi-exponential spin-lattice relaxation behaviour
(Fig. 7), suggesting at least two different relaxation
environments for the nuclei corresponding to each peak.
Approximately half of the intensity of the NMR peaks
attributed to the amorphous phase is characterized by a
longer spin-lattice relaxation time (40.1 s), suggestive of a
region containing fewer Ce3+ paramagnetic centres. Thus
the amorphous phase NMR peaks and relaxation rates are
consistent with a core-shell model [9] where a spherical
diamagnetic core of material is surrounded by a paramag-
netically relaxed shell, with the intensity ratios used for an
overall mass balance. Our data can be fit to such a model
that yields results consistent with the bulk compostion and
length scales observed with electron microscopy; the
microscopy data, however, are clearly not consistent with
a spherical geometry for the amorphous phase.
A qualitative understanding of the amorphous phase
region may be further developed via analysis of the spin-
lattice relaxation times. The short spin-lattice relaxation
times associated with the peaks emanating from the
4. Conclusion
Powder and sintered samples of nanoparticulate cerium
phosphate were prepared through various syntheses.
Transmission electron microscopy and X-ray diffraction
indicated that in the samples synthesized by reacting ceria
with phosphoric acid, an intergranular phosphate-rich
amorphous phase persisted even after heat treatments of
up to 20 h at 800 1C. The use of 31P-NMR has proved to be
a reliable method of distinguishing between phosphorus
nuclei in the monazite and rhabdophane phases of CePO4.
Additional 31P signals were detected near 0 ppm, their
intensity correlating with the amount of unreacted ceria
and therefore the amount of phosphate-rich amorphous
phases. The complex T1 and T2 relaxation times for these
samples supports the presence of the amorphous, phos-
phate-rich material.
Fig. 6. Variable temperature 31P NMR spectra of the bulk CePO4 peak in
Sample 3 obtained in a 1.54 T magnet while spinning the sample at the
magic angle at 11 kHz. 0 ppm is referenced to 31P in 85% H3PO4.
Fig. 7. Signal intensities of the peaks near 0 ppm in Samples 3 and 4 grow biexponentially with increasing t delays, corresponding to two separate spin-
lattice T1 constants for each peak.

ARTICLE IN PRESS
L. Karpowich et al. / Journal of Solid State Chemistry 180 (2007) 840–846
846
While the exact nature of the amorphous regions has not
References
been characterized, it is probable that these phosphate-rich
intergranular phases will play a role in ionic diffusion and
conductivity. Proton conduction in rare-earth phosphates
has been attributed to the presence of (PO4)Hꢁ2 defects at
PO4 anion sites that are induced by pyrophosphate phases
such as P2O7 [7,16]. A nano-grained membrane, synthe-
sized as shown above, would result in a higher ratio of
intergranular to crystalline CePO4. At the nanoscale, the
volume fraction of amorphous interfacial material is no
longer negligible, and charge transport through this region
may contribute to proton conduction [17]. Exploiting this
characteristic of nano-grained CePO4 with phosphorus-
rich intergranular phases may well lead to enhanced proton
conductivity through the grain boundaries.
[1] G.W. Beall, L.A. Boatner, J. Inorg. Nucl. Chem. 43 (1980) 101.
[2] Y. Hikichi, T. Nomura, Y. Tanimura, S. Suzuki, J. Am. Ceram. Soc.
73 (12) (1990) 3594.
[3] O.H. Leonardos, Econ. Geol. 69 (1974) 1126–1128.
[4] M. Rappaz, M.M. Abraham, J.O. Ramey, L.A. Boatner, Phys. Rev.
B 23 (3) (1981) 1012.
[5] L. Xu, G. Guo, D. Uy, A.E. O’Neill, W.H. Weber, M.J. Rokosz,
R.W. McCabe, Appl. Catal. B-Environ. 50 (2004) 113.
[6] B. Gee, C.R. Horne, E.J. Cairns, J.A. Reimer, J. Phys. Chem. B. 102
(1998) 10142.
[7] S.B. Adler, J.A. Reimer, Solid State Ion. 91 (1996) 175.
[8] M. Duncan, Chemical Shift Tensors, second ed., Farragut Press,
Madison, WI, 1997.
[9] M. Lopez Granados, F. Cabello Galisteo, P.S. Lambrou,
R. Mariscal, J. Sanz, I. Sobrados, J.L.G. Fierro, A.M. Efstathiou,
J. Catal. 239 (2006) 410.
[10] N. Kitamura, K. Amezawa, Y. Tomii T. Hanada, N. Yamamoto,
T. Omata, Y. Otsuka-Yao-Matsuo, J. Electrochem. Soc. 152 (4)
(2004) A658.
Acknowledgments
[11] H. Onoda, H. Nariai, A. Moriwaki, H. Maki, I. Motooka, J. Mater.
Chem. 12 (2002) 1754.
This work was supported by the Director, Office of
Science, Office of Basic Energy Sciences, Materials Sciences
and Engineering Division, of the US Department of
Energy under Contract No. DE-AC02-05CH11231, and
by a National Science Foundation Graduate Student
Fellowship (SW). Dr. Shashi Vyas is thanked for NMR
discussions. The authors acknowledge support of the
National Center for Electron Microscopy, Lawrence
Berkeley Lab, which is supported by the US Department
of Energy under Contract # DE-AC02-05CH11231.
[12] C. Feldmann, H.O. Jungk, J. Mater. Sci. 37 (2002) 3251.
[13] T. Masui, H. Hirai, N. Imanaka, G. Adachi, Phys. Status Solidi A
198 (2) (2003) 364.
[14] S. Lucas, E. Champion, D. Bernache-Assollant, G.J. Leroy, J. Solid
State Chem. 177 (2004) 1312.
[15] S. Lucas, E. Champion, D. Bregiroux, D. Bernache-Assollant,
F. Audubert, J. Solid State Chem. 177 (2004) 1302.
[16] K. Amezawa, H. Maekawa, Y. Tomii, N. Yamamato, Solid State
Ion. 145 (2001) 233–240.
[17] P. Heitjans, S. Indris, J. Phys: Condens. Matter 15 (2003) R1257.