Thermal stability of alkali and alkaline-earth substituted LAMOX oxide-ion conductors

Article abstract of DOI:10.1039/b915080c

The high temperature demixing/recombination phenomenon previously observed in Ca-substituted La₂Mo₂O₉ oxide ion conductors [A. Selmi et al., Solid State Ionics, 2006, 177, 3051; Eur. J. Inorg. Chem., 2008, 1813] has been visualised using scanning electron microscopy and EDX analysis. The demixed state appears as CaMoO₄ straight solid streams erupted from pores within LAMOX grains. The thermal stability study is extended to other alkali and alkaline-earth substituted LAMOX compounds, all of which are shown, in temperature-controlled X-ray diffractograms, to present similar demixing/recombination processes. The most spectacular effect is observed in La_1.88K_0.12Mo_0.6W_1.4O _8.88 where demixing takes the form of a total decomposition, before full recombination at a higher temperature. Such a phenomenon is interpreted as originating from temperature-dependent solid solution limits with higher substitutional ranges at higher temperatures. It results in the metastabilisation of pure phases by quenching (or rapid cooling), whereas the stable state is demixed, as shown on slowly cooled samples. The Royal Society of Chemistry 2010.

Full text of DOI:10.1039/b915080c

View Article Online / Journal Homepage / Table of Contents for this issue
PAPER
www.rsc.org/dalton | Dalton Transactions
Thermal stability of alkali and alkaline-earth substituted LAMOX
oxide-ion conductors†
A. Selmi, C. Galven, G. Corbel and P. Lacorre*
Received 27th July 2009, Accepted 24th September 2009
First published as an Advance Article on the web 10th November 2009
DOI: 10.1039/b915080c
The high temperature demixing/recombination phenomenon previously observed in Ca-substituted
La2Mo2O9 oxide ion conductors [A. Selmi et al., Solid State Ionics, 2006, 177, 3051; Eur. J. Inorg.
Chem., 2008, 1813] has been visualised using scanning electron microscopy and EDX analysis. The
demixed state appears as CaMoO4 straight solid streams erupted from pores within LAMOX grains.
The thermal stability study is extended to other alkali and alkaline-earth substituted LAMOX
compounds, all of which are shown, in temperature-controlled X-ray diffractograms, to present similar
demixing/recombination processes. The most spectacular effect is observed in La1.88K0.12Mo0.6W1.4O8.88
where demixing takes the form of a total decomposition, before full recombination at a higher
temperature. Such a phenomenon is interpreted as originating from temperature-dependent solid
solution limits with higher substitutional ranges at higher temperatures. It results in the
metastabilisation of pure phases by quenching (or rapid cooling), whereas the stable state is demixed, as
shown on slowly cooled samples.
Suspecting that the phenomenon first observed in Ca sub-
1. Introduction
stitutions is more general, we undertook a thorough study of
the thermal stability of alkali and alkaline-earth substituted
LAMOX compounds. It helped elucidate both irreproducibilities
in measurements and discrepancies found in the literature, and
allowed us to propose an explanation for the origin of such a
phenomenon. The main results of this study are presented here.
The long term stability of solid oxide fuel cells (SOFC) often
depends on such phenomena as cation diffusion in cell core
materials, which is usually non-negligible in oxides with a high
concentration of mobile oxygen vacancies. Although not as
numerous as studies devoted to the oxide-ion transport properties
of SOFC electrolytes, those concerned with cation diffusion draw
the attention, for instance, on the importance of the process in the
degradation of the conductivity and mechanical properties.1
In the last few years, a significant amount of work has been
published on the properties of the new LAMOX family of fast
oxide-ion conductors derived from La2Mo2O9,2 but very few
of those were centred on cation diffusion, even though it was
suspected3,4 and further circumvented5 in studies on reactivity
with electrode materials. We started addressing the issue while
studying the effects of aliovalent substitutions on the properties
of La2Mo2O9. Such substitutions by an element with a different
valence offer the possibility to change the oxide ion/vacancy
balance, and therefore to modify the conduction properties of the
materials. We previously explored the properties of LAMOX com-
pounds partially substituted by Ca. A very unusual behaviour was
observed, with a series of reversed vacancy ordering/disordering
phase transformations including partial demixing and recombina-
tion processes,6,7 showing the importance of the diffusion processes
of both anionic and cationic species in this series. The demixing
process is accompanied by a decrease of oxide-ion conductivity,7
whose thermal evolution therefore appears to be irreversible in a
definite temperature range.
2. Experimental
2.1 Samples preparation
Powder samples were prepared by conventional solid state reac-
tions.◦Starting reactant powders (La2O3 previously annealed at
1000 C in order to remove adsorbed water and carbon dioxide,
CaCO3, SrCO3, BaCO3 kept dry in an air oven at 100 ◦C, K2CO3,
1.5H2O, MoO3, WO3) were weighed in stoichiometric amounts,
ground and first heated at 500 ◦C for 12 h in an alumina crucible.
After several intermediate heatings and regrindings, pure samples
were obtained after a final firing at temperatures reported in
◦
Table 1, and cooling down to room temperature at 5 C min-1.
Doubly substituted compositions involving Mo-W substitutions
were studied because of the stabilising effect of tungsten in
reducing atmospheres.8
The Ca4-LM pellet sample used in the SEM study was prepared
from the previous powder batch with the same composition. The
900 mg powder sample was first ball-milled (FRITSCH planetary
micromill pulverizette apparatus, with agate vials together with six
12 mm diameter agate balls, in ethanol) for 4 alternations of 15 min
milling sequences at 700 rpm, with 15 min pauses in between.
After ethanol was evaporated in an air oven, a drop of polyvinyl
alcohol solution was added to the ground powder, which was then
shaped as a pellet (diameter 13 mm, thickness ~0.6 mm) using a
uniaxial press (~0.15 GPa). The pellet was annealed at 300 ◦C for
5 h, and finally at 1200 ◦C for 4 h. It was then polished, after
Laboratoire des Oxydes et Fluorures, UMR CNRS 6010, Universite´ du
Maine, Avenue Olivier Messiaen, 72085 Le Mans cedex 9, France. E-mail:
philippe.lacorre@univ-lemans.fr
† Electronic supplementary information (ESI) available: Room tempera-
ture X-ray diffraction patterns of 19 as-prepared pure powder samples. See
DOI: 10.1039/b915080c
This journal is
The Royal Society of Chemistry 2010
Dalton Trans., 2010, 39, 93–102 | 93
©

View Article Online
Table 1 Composition, synthesis temperature and crystal symmetry of the room temperature phase for the series of pure compounds studied in this paper
Composition
T
synth./◦C
Symmetry
Composition
T
synth./◦C
Symmetry
Composition
T
synth./◦C
Symmetry
Ca1-LM
Ca2-LM
Ca3-LM
Ca4-LM
Ca5-LM
1100
1100
1100
1150
1150
Mono
Mono
~Cubic
Cubic
Cubic
Ba5-LM
Ba7.5-LM
Ba10-LM
Ba15-LM
1000
1000
1050
1200
Cubic
Cubic
Cubic
Cubic
K2-LM
K3-LM
K4-LM
K5-LM
K6-LM
875
900
975
1075
1060
Cubic
Cubic
Cubic
Cubic
Cubic
Composition
T
synth./◦C
Symmetry
Composition
T
synth./◦C
Symmetry
Composition
T
synth./◦C
Symmetry
Ca4-LM-W12.5
Ca4-LM-W25
Ca4-LM-W37.5
1150
1175
1175
Cubic
Cubic
Cubic
Ba5-LM-W70
Ba7.5-LM-W70
Ba10-LM-W70
1075
1100
1200
Cubic
Cubic
Cubic
K2-LM-W12.5
K4-LM-W12.5
K2-LM-W70
K3-LM-W70
K4-LM-W70
K6-LM-W70
875
975
1050
1100
1100
1150
Cubic
Cubic
Cubic
Cubic
Cubic
Cubic
being glued with Canada balsam, on a TRIEFUS Surfex 83 201
apparatus, according to the process: firstly, a 2 min polishing with
FEPA P1200 sandpaper (15.3 mm particles), followed by successive
polishes with flannel disks impregnated with ESCIL LP S lubricant
containing diamond powder (10 min with 6 mm particles, 10 min
with 1 mm particles, 20 min with 0.25 mm particles). In order to
reveal grain boundaries, a 30 min annealing at 1150 ◦C was carried
out (with 5 ◦C min-1 heating/cooling rate) prior to further studies.
For the conductivity measurements, a 5 mm diameter die
was used, resulting in pellets (~4.3 mm diameter and ~2.6 mm
thickness) with relative densities 96% (Ba5-LM) and 95% (Ba10-
LM-W70) after◦sintering. The annealing conditions were 15 h at
1000 ◦C and 5 C min-1 cooling down for Ba5-LM, and 3 h at
1175 ◦C, then quenching down to 900 ◦C and finally cooling down
2.4 Structural and microstructural characterisation
Temperature-controlled X-ray powder diffraction (XRD) patterns
were recorded on a PANalytical q/q Bragg-Brentano X’pert MPD
PRO diffractometer (Cu Ka1+2 radiations) equipped with the
X’celerator detector and with a HTK 1200 Anton Paar cham-
ber. Thermodiffractograms were recorded at given temperatures
between 24 and 1060 ◦C during 197 min in the 9–130◦ scattering
angle range with◦a 0.0167◦ step after a 20 min thermal stabilisation
(heating rate 10 C min-1 between each calibrated temperature3).
The program FullProf9 was used for Le Bail’s fit of the diffraction
patterns recorded at RT for crystal cell determinations, and at
846 ◦C on the K6-LM-W70 decomposed phase.
In order to check the purity of the slowly cooled samples,
complementary room temperature X-ray diffraction patterns were
collected on the same diffractometer (without the HTK chamber)
in the 24–32◦ scattering angle range with a 0.0084◦ step during
24 min, or 127 min for those with lower substituent contents.
The SEM study (imaging and EDX analysis) was carried out
on a Hitachi S2300 microscope. Prior to the analysis, a platinum
or carbon layer had been deposited at the pellet surface by RF
sputtering under vacuum.
◦
at 10 C min-1 for Ba10-LM-W70. Electrodes were deposited as
thin platinum films by RF sputtering on both faces of the pellets.
2.2 Terminology
In order to simplify the composition terminology of the studied
phases, we will use in this paper the following nomenclature
which focuses on the mol% substitution rate either on the
La or the Mo sites. A LAMOX compound with the general
formula La2-2xA2xMo2-2yW2yO9-d will be named Ax-LM-Wy, with
x = 100x and y = 100y representing the mol% substitutional
fraction of La and Mo, respectively. For instance, composition
La1.92Ca0.08Mo2O8.96 (4 mol% Ca-LAMOX) will be named Ca4-
LM, and composition La1.88K0.12Mo0.6W1.4O8.88 (6 mol% K +
70 mol% W-LAMOX) will be named K6-LM-W70. The com-
positions of most of the phases studied in this paper are given in
the figure captions in the ESI.†
3. Instability in ionic transport
The main motivation for aliovalent substitution in oxide-ion
conductors is to be able, for application purposes, to tune anion
conductivity by controlling the oxide ion/vacancy balance. In
order to clearly present such effects and make use of them, an im-
portant prerequisite is to ensure that the substituted compositions
are pure and thermally stable in the considered thermal domain.
As stated in the introduction, a previous study showed that it is not
the case for certain Ca-substituted LAMOX compounds, resulting
in ionic conductivity instability and irreversibility (see Fig. 9 in
ref. 7). When extending our studies to LAMOX pellet samples
with La substituted by other alkaline-earths such as barium,
we found the same type of irreversible behaviour in conducting
properties (see Fig. 1), which suggests that such phenomena are
probably widespread in aliovalently-substituted LAMOX oxide-
ion conductors. In order to better identify and understand them,
we undertook a thorough study on the thermal stability of powder
samples of these compounds.
2.3 Conductivity measurements
A Schlumberger Solartron SI 1260 frequency response analyser
working with Dielectric Interface Solartron 1296 in the frequency
range of 0.05 Hz to 10 MHz was used to collect the impedance
spectra in flow-dried air. After 30 min of thermal stabilisation in
dried air, each spectrum was recorded at an AC voltage of 0.05 V
with 40 points/decade. In order to determine the resistance of the
ceramic pellets, impedance data were analysed with the help of
Z-view 2.7 software.
94 | Dalton Trans., 2010, 39, 93–102
This journal is
The Royal Society of Chemistry 2010
©

View Article Online
Fig. 1 Thermal evolution of the conductivity of Ba5-LM (a) and Ba10-LM-W70 (b) LAMOX pellets upon heating, then cooling down after a thermal
plateau at 725 ◦C for 15 h (a) and 54 h (b).
narrow thermal range of purity (1050 ◦C), and once again, in this
4. Solid solution domains
series the thermal history has a strong effect on the phase purity.
Finding adequate synthesis conditions and solid solution limits
The cell volume deduced from X-ray diffraction patterns show a
in alkali and alkaline-earth substituted LAMOX compounds
linear evolution with Ba content up to 15 mol% Ba.
appeared to be problematic, since, in most cases, purity appears to
be highly dependent on the thermal history. In order to allow for
4.1.2 Alkali (Na, K). Sodium was tested as a possible lan-
thanum substitute in part because the 23Na nucleus gives access
comparisons between compositions, a common reference process
to NMR studies on a cationic site. Tealdi et al.15 mention the
preparation of a Na5-LM sample by a complex polymerisation
method. We attempted to prepare Na4-LM and Na6-LM samples
by solid state reactions. It resulted in impure cubic phases, showing
that enough sodium entered the LAMOX phase to make it cubic,
but that the substitution limit is lower than 4 mol% Na. The
impurity is the scheelite NaLa(MoO4)2, whose main diffraction
peak is close to the LAMOX (210) main reflection, and therefore
difficult to detect at low amounts. For this reason, and also because
of the too small Na amount for an NMR study of the pure phase,
we did not extend our studies further on lower Na contents.
In the literature, K substitution for La has been tested up to
7.5 mol%.16 Marrero-Lopez et al. indicate a solid solution limit of
3.75 mol% K.11 We prepared K2-, K3-, K4-, K5-, K6- and K10-
LM samples by solid state reactions. Phases were pure cubic up
to 6 mol% K (Fig. S8–S10, ESI†). The K10-LM sample showed
impurity peaks belonging to an undetermined phase (whose main
peak is close to the LAMOX (210) main reflection), and to the b-
KLa(MoO4)2 phase depending on the annealing temperature. The
room temperature cell volumes determined from X-ray diffraction
patterns (see Fig. 2(a)) show a non-linear evolution with K content,
with a maximum value of around 5–6 mol% K. We also noted in the
K3- and K4-LM samples, irreproducibilities in the phase purity
and in the value of the cell parameters (Fig. 2(a)), respectively,
depending on the thermal history.
was settled for all powder preparations, with a 5 ◦C min-1 heating
and cooling speed before and after thermal plateaus at given
temperatures. We will first examine the case of single (either alkali
or alkaline-earth) substitutions for lanthanum, then that of double
substitutions (alkali or alkaline-earth for lanthanum, and tungsten
for molybdenum).
4.1 Single substitutions
4.1.1 Alkaline-earth (Ca, Sr, Ba). Samples with La substi-
tuted by 1, 2, 3, 4 and 5 mol% Ca were checked to be pure,
whereas the Ca10-LM sample showed impurity peaks of the
CaMoO4 scheelite phase.7 Note that Subasri et al. indicate that the
substitution limit is Ca4-LM,10 and Marrero-Lopez et al. that it
is Ca2.5-LM.11 The room temperature X-ray diffraction patterns
of Ca1- and Ca2-LM samples show a monoclinic cell (Fig. S1,
ESI†) with superstructures as in a-La2Mo2O9, whereas the Ca4-
and Ca5-LM samples are cubic (Fig. S2 and S3, ESI†).
Several attempts to prepare pure samples of 5 mol% Sr were
carried out, with limited success. An apparently pure preparation
could not be reproduced. The impurity was found to be the
SrMoO4 scheelite phase, whose main diffraction peak is very close
to the main (210) peak of La2Mo2O9. It is therefore very difficult
to ascertain the absence of any impurities. The impurity peak
appeared to be more or less intense, and the cell volume varied
depending on the annealing temperature without any coherent
evolution. Given the above difficulties, no further attempts were
made with other Sr amounts.
4.2 Double substitutions
Three series of double substitutions involving Ca, or Ba, or K,
with W substitute for Mo, were prepared.
According to the literature, the maximum substitution rate of
lanthanum by barium is either 7.5 mol% or 10 mol%.10–14 We
prepared samples with 5, 7.5, 10, 15 and 20 mol% Ba. Ba5-, Ba7.5-,
Ba10- and Ba15-LM samples were prepared pure (Fig. S4–S7,
ESI†) at increasing synthesis temperature, whereas the Ba20-LM
sample showed impurity peaks, identified as being characteristic
of the BaMoO4 scheelite phase. The Ba10-LM sample has a very
As representative of Ca substitution, the Ca4-LM composition
was chosen as a basis for double substitutions involving tungsten.
It appears that, contrary to single W substitution,8,17,18 the
maximal range for W substitution in Ca4-LM samples is lower
than 50%, since pure doubly substituted compounds could be
prepared only with 12.5 mol% W, 25 mol% W and 37.5 mol%
This journal is
The Royal Society of Chemistry 2010
Dalton Trans., 2010, 39, 93–102 | 95
©

View Article Online
Fig. 2 Crystal cell volume as a function of the amount of substituting element for single K (a) and double Ca-W (b) substitutions. In both cases, the
thermal history has an effect on cell volume: in (a), the two points representative of the K4-LM composition correspond to two samples prepared with
different final annealing times at 975 ◦C; in (b), the four points representative of the Ca4-LM-W37.5 composition correspond to successive annealings of
the same sample at 1150 ◦C and 1175 ◦C.
W (Fig. S11–S13, ESI†). X-Ray diffraction patterns of the Ca4-
LM-W50 sample showed the presence of a more or less intense
impurity peak (depending on the temperature) at the basis of the
LAMOX main peak. As in the singly-substituted W series, the
cell parameter evolution with tungsten content (deduced from
X-ray diffraction patterns) passes through a maximum, but at
lower W content: around 12.5 mol% W (Fig. 2(b)). In this doubly-
substituted series as in the singly-substituted previous ones, both
purity (uneasy to ascertain due to peaks overlapping) and cell
parameters seem to be dependent on the sample thermal history
(Fig. 2(b)).
Contrary to double Ca-W substitutions, the preparation of
doubly substituted Ba-W samples with higher W content is
possible since we were able to synthesise by solid state reactions,
samples with 5, 7.5, 10 mol% Ba and 70 mol% W as pure phases
(Fig. S14–S16, ESI†). The Ba15-LM-W70 composition was not
obtained pure, with the BaMo(W)O4 phase as an impurity. Ba10-
LM-W70 was obtained pure after a thermal process which is
different from the standard one: a final short annealing at the
limit of the melting point (1200 ◦C) was followed by quenching to
room temperature. All the doubly substituted Ba-W samples were
obtained in the LAMOX cubic form.
5.1 Single substitutions
The most complex thermal behaviour is observed in the Ca series,
and more particularly in the Ca4-LM sample whose features are
thoroughly detailed in references 6 and 7. In this compound, at
intermediate temperature (around 400–600 ◦C), a double reversed
cubic/monoclinic/cubic (b/a/b) phase transition is observed due
to the metastability of the cubic b phase at low temperature (when
cooled down more slowly, the stable form of the Ca4-LM sample is
the monoclinic a phase). At higher temperature, above ca. 700 ◦C,
an impurity phase of the CaMoO4 type appears (see Fig. 3(a)).
Upon further heating, the impurity peaks first increase in
intensity, then decrease above 950–1000 ◦C, and finally disappear
at the synthesis temperature of the compound (1150 ◦C). We have
visualised this process (see Fig. 4) on a dense pellet sample of
Ca4-LM using scanning electron microscopy (SEM). A pellet
sample of relative density◦97(1)% was first prepared at 1200 ◦C
for 4 h, then cooled at 5 C min-1. After polishing and thermal
treatment (see experimental), the pellet was annealed for 10 h at
850 ◦C (inside the demixing thermal range), and cooled to room
temperature at 5 ◦C min-1. Fig. 4(a) gives the amazing SEM image
of the previously polished face. A large number of craters, a few
micrometers in diameter, are visible inside ~50 mm size grains. In
addition, many straight solid streams, ~2–3 mm in diameter and
~10–50 mm in length, are visible at the surface of the grains, as
if they erupted from pores. EDX analysis was used to map such
solid streams on a grain surface. It shows (Fig. 4(b)) that the
streams are rich in calcium and poor in lanthanum relative to the
grain surface, thereby corroborating the CaMoO4 composition of
these streams. In-depth EDX analysis (see Fig. (5)) performed
on this pellet after further deep polishing (~0.35 mm) shows that
CaMoO4 steaming occurs not only at the surface of the grains
but also inside the grains in the whole pellet, with no preferred
orientation of the streams. After further annealing of the pellet
for 12 h at 1150 ◦C, i.e. above the recombination temperature,
and cooling to room temperature at 5 ◦C min-1, the pellet surface
exhibited the morphology shown in Fig. 4(c). The craters, although
partially filled, appear to be as numerous as previously observed,
In the K-W series, double substitution was made possible up
to K6-LM-W70 (Fig. S17–S19, ESI†). All compositions with at
least 12.5 mol% W were obtained as pure cubic phases. As for the
single K substitutions, in the double K-W substitutions the cell
parameter evolution with K content is non-linear. Contrary to the
single W substitutions, in the double K-W substitutions the cell
parameter evolution with W content appears to be rather linear,
decreasing with increasing W content.
5. Thermal stability
A large number of the above prepared samples were studied by
temperature-controlled X-ray diffraction in order to probe their
thermal stability. The most salient features, concerning single and
double substitutions involving Ca, Ba and K are reported here.
96 | Dalton Trans., 2010, 39, 93–102
This journal is
The Royal Society of Chemistry 2010
©

View Article Online
Fig. 3 X-Ray diffraction patterns of Ca4-LM (a), Ba10-LM (b), and K6-LM (c) powder samples. Bottom: after synthesis at 1150 ◦C, 1050 ◦C and
1060 ◦C, respectively, then cooling down to RT at 5 ◦C min-1; middle: heating of pure powders (details of T-controlled XRD patterns); top: at 952 ◦C,
showing the maximal size of the main impurity peaks (*) comparatively to that of LAMOX diffraction peaks.
Fig. 4 SEM image (a) and EDX cartographies (b) of the Ca4-LM pellet sample after 10 h annealing at 850 ◦C showing straight solid streams (arrow)
of CaMoO4 composition; (c) SEM image of the same pellet after further 12 h annealing at 1150 ◦C.
and the solid streams have disappeared. EDX analysis of the
surface did not show any trace of CaMoO4, thus confirming full
recombination. A vivid description of the observed process is to
consider Ca-LM somewhat as a “La2Mo2O9 sponge of CaMoO4”.
As stated earlier,7 this process occurs in powders as well as in dense
pellets.
In the following, we will concentrate on this demix-
ing/recombination phenomenon in order to check whether this
behaviour is specific to the Ca series, or whether it is more general
in the aliovalently substituted LAMOX compounds (and if so, to
what extent in terms of composition). For this study, temperature-
controlled X-ray diffraction will be used.
In the Ba series, the only composition to remain pure cubic
in the whole studied thermal range (RT–952 ◦C) is the Ba5-
LM sample. The two other compositions studied, Ba7.5-LM and
Ba10-LM, show impurity peaks above ~900 ◦C (undetermined
impurity) and ~700 ◦C (BaMoO4) respectively (see Fig. 3(b)).
Since they have been prepared as pure phases at 1000 ◦C and
1050 ◦C respectively, it means that the Ba series also undergoes the
demixing/recombination phenomenon, all the more when the Ba
content is higher. The absence of any impurities at any temperature
in the Ba5-LM sample is somewhat in contradiction with the ob-
served decrease in conductivity (Fig. 1(a)). One has to admit that
the discrepancy originates either from the effect of sample shaping
This journal is
The Royal Society of Chemistry 2010
Dalton Trans., 2010, 39, 93–102 | 97
©

View Article Online
Fig. 5 SEM images (left) and EDX cartographies after in-depth polishing (~-0.35 mm) of the Ca4-LM pellet sample annealed at 850 ◦C.
(powder vs. pellet), or from the additional effect of electric field,19
or from a lower sensitivity of XRD comparative to impedance
spectroscopy to detect small amounts of BaMoO4 impurity.
The K series shows the same type of behaviour as the Ba
series. Whereas the K2-LM sample remains pure cubic up to
about 900 ◦C, the K4-LM and K6-LM samples show the presence
of an impurity above ~700 ◦C and ~500 ◦C respectively (see
Fig. 3(c)). Here again, since these two last compositions have been
prepared as pure phases at 975 ◦C and 1060 ◦C respectively, a
demixing/recombination phenomenon also occurs, especially at
high K content. It confirms that such a behaviour is not specific to
the Ca series, but general in all alkali and alkaline-earth substituted
LAMOX compounds. In the case of K substitution, the impurity
is b-KLaMo2O8, crystallised in the scheelite type form as CaMoO4
and BaMoO4 in the previous series.
Fig. 6 Part of the thermodiffractograms recorded on K2-LM-W70
(La1.96K0.04Mo0.6W1.4O8.96) upon heating, showing a demixing phenomenon
between 740 and 952 ◦C.
5.2 Double substitutions
Similar temperature-controlled X-ray diffraction measurements
have been carried out on doubly substituted Ba-W and K-W series.
In the Ba-W series, three samples were studied with 5, 7.5 and
10 mol% Ba, and 70 mol% W for each of them. Similarly to what
happens in singly Ba substituted samples, the only compound
which remains pure at all temperatures is Ba5-LM-W70. In the
other two compositions, a BaMo(W)O4-type impurity appears
above ~800 ◦C, whose peaks increase in intensity wi◦th temperature
the point in the case of the K2-LM-W70 sample. The impurity
peaks correspond to the b-KLaMo2O8 phase (as in the singly
K substituted cases with higher K content), plus another phase
called p, which will be identified afterwards. Similar impurities are
detected in the K4-LM-W70 sample between 740 ◦C and 950 ◦C.
From all the studied samples, the K6-LM-W70 compound
shows the most spectacular demixing/recombination phe-
nomenon, since the LAMOX phase totally disappears between
740 ◦C and 850 ◦C, before reappearing at higher temperature,
◦
at least up to 1000 C (Ba7.5-LM-W70) and 950 C (Ba10-LM-
W70), the higher thermal limits of our temperature-controlled
XRD experiments. For the last compound, this is consistent with
the observed decrease in conductivity (Fig. 1(b)). Given that
the two compounds were prepared as pure phases at 1100 ◦C
and 1200 ◦C respectively, one has to admit that scheelite-type
impurities disappear progressively above 1000 ◦C. The behaviour
is close to that observed in the singly Ba substituted series.
The K-W series behaves differently from the previous ones in
several ways. First of all, all the studied compositions (2, 4 and
6 mol% K, with 70 mol% W) show a demixing phenomenon,
above ~700 ◦C, ~700 ◦C and ~600 ◦C, respectively. An interesting
point is that all of them also exhibit a recombination phenomenon
at the highest temperature, showing that recombination takes
place at a lower temperature in this series (1050 ◦C, 1000 ◦C,
1050 ◦C, respectively) than in the previous ones. Fig. 6 illustrates
◦
and remaining the only (pure LAMOX) phase at ~1060 C (see
Fig. 7). Between 740 ◦C and 850 ◦C, two phases are detected:
the b-KLaMo2O8 scheelite and the p phase already detected as
an impurity in the samples with lower K amount, the p phase
being now the main phase. A careful examination of the high
temperature diffraction patterns enabled us to identify it, with the
use of the Inorganic Crystal Structure Database. The p phase is of
the a-La2W2O9 structural type, which is triclinic in symmetry and
different from the a-La2Mo2O9 structural type.20 Table 2 compares
◦
the cell parameters of the p phase at 846 C (obtained from Le
Bail’s fit of the diffractogram with the Fullprof program9) and of a-
La2W2O9 at room temperature.20 The difference in cell parameters
results from the difference in temperature and also probably from
partial substitution of tungsten by molybdenum.21
98 | Dalton Trans., 2010, 39, 93–102
This journal is
The Royal Society of Chemistry 2010
©

View Article Online
Table 2 Cell parameters of the p phase in sample K6-LM-W70 at 846 ◦C, compared to those of a-La2W2O9 at room temperature20
¯
˚
a/A
˚
b/A
˚
c/A
(Space group P1)
p Phase at 846 ◦
a/◦
b/◦
g /◦
C
7.2804(1)
7.2489(1)
7.3530(1)
7.2878(1)
7.1258(1)
7.0435(1)
96.375(1)
96.367(1)
94.734(1)
94.715(1)
70.700(1)
70.286(1)
a-La2W2O9 at RT20
- 6 h annealing at temperature Ta within the demixing thermal
range, followed by
- a slow cooling (15 ◦C h-1) to room temperature.
This thermal schedule was also applied to some samples with
lower substitutional rates, for which no demixing had been
observed previously. The studied compositions were Ca1-, Ca2-,
Ca3-, Ca4-LM (Ta = 950 ◦C), K2-, K3-, K4-LM (Ta = 900 ◦C),
Ba4-, Ba7.5-LM (Ta = 1000 ◦C), K6-LM-W70 (Ta = 900 ◦C and
1000 ◦C), Ba5-LM-W70 and Ba7.5-LM-W70 (Ta = 1000 ◦C).
All the compositions presenting the demixing/recombination
phenomenon at high T showed, after annealing and slow cooling, a
demixed stable state at room temperature. In addition, some other
compositions (such as Ca3- or even Ca2-LM, or K3-LM) also
showed the same demixed stable state. Only the compositions with
the lowest substitutional content, such as Ca1-LM and K2-LM,
showed no demixing (note that in these cases, the substitution rate
may be too low for demixing to be detected, if any). Once again, the
most spectacular case is found for the K6-LM-W70 composition
which appeared, after annealing at Ta = 900 ◦C, to be totally
demixed/decomposed at room temperature. Note however that
the demixing/decomposition kinetics is very slow, since cooling
Fig. 7 Part of the thermodiffractograms recorded on K6-LM-W70
(La1.88K0.12Mo0.6W1.4O8.88) upon heating, showing a decomposition phe-
nomenon between 634 and 952 ◦C.
5.3 Metastability of pure phases
The results presented above show that the demixing/
recombination process is recursive in alkali and alkaline-earth
substituted LAMOX compounds, and not restricted to Ca-
substituted LAMOX. Although already encountered in another
oxide compound,22 such a decomposition/recombination phe-
nomenon has apparently not been commented upon yet. The
similarity with the low temperature double reversed transition
observed in the 4 mol% Ca sample (see Fig. 8), namely b/a/b vs.
b/demixed/b, is troubling. Suspecting that both phenomena could
have the same origin, namely the metastability of a pure b phase
(whereas the intermediate a and the demixed phases, respectively,
would be the stable states), we have carried out further experiments
in order to check for it.
◦
down from higher Ta = 1000 C through the demixing thermal
◦
◦
range at 0.25 C min-1 (15 C h-1) is not slow enough to detect
any impurity peaks in the room temperature X-ray diffractogram:
a few hours annealing at the right temperature is necessary to
achieve total decomposition before cooling down (see Fig. 9).
From the above study, it appears clearly that all compositions
which present a demixing(decomposition)/recombination process
at high temperature are in a metastable state when they appear as
pure phases at room temperature.
6. Discussion
A series of thermal treatments have been performed on sam-
ples presenting the high temperature demixing(decomposition)/
recombination phenomenon, according to the following schedule:
From the previous results, one has to assume that the solid
solution range of alkali or alkaline-earth in LAMOX phases,
Fig. 8 Oxygen vacancy order/disorder phase transformations in LAMOX compounds with low substitutional content: (a) regular a/b phase transition;
(b) metastabilisation of the disorder b phase by rapid cooling; (c) recovery of the ordered a form by heating, followed by regular a/b phase transition, as
observed in the Ca4-LM compound, for instance.
This journal is
The Royal Society of Chemistry 2010
Dalton Trans., 2010, 39, 93–102 | 99
©

View Article Online
Fig. 9 Room temperature X-ray diffraction patterns of K6-LM-W70 powder samples: (a) raw powder as prepared at 1150 ◦C, then cooled to RT at
5 ◦C min-1 (pure); (b) raw powder annealed for 6 h at 1000 ◦C, then cooled to RT at 0.25 ◦C min-1 (pure); (c) raw powder annealed for 6 h at 900 ◦C, then
cooled to RT at 0.25 ◦C min-1 (decomposed).
Fig. 10 Tentative phase diagram of La2Mo2O9-AMoO4 solid solutions (A = Ca, Ba, (K0.5La0.5),. . .) showing a temperature-dependent solid solution
limit (a) with metastabilisation of a pure phase by rapid cooling (b) and recovery of the stable diphasic system by heating up (demixing) followed by
reincorporation of the substitute (recombination) (c).
varies with temperature and is higher at high temperature than
at low temperature, which is similar to what happens in eutectic
or eutectoid phase diagrams with solid solution, or in exsolution
processes. Solid solution extends to a much narrower range at room
temperature than originally postulated (Fig. 10(a)). However,
if ◦samples are quenched or cooled rapidly enough (even at
5 C min-1), high temperature stable solid solutions can be made
metastable at room temperature (Fig. 10(b)). When the metastable
composition is heated at a certain temperature the supplied energy
is large enough to turn the system back to its stable, demixed
state (Fig. 10(c)); when temperature keeps on increasing, LAMOX
incorporates more and more substitutes and eventually transits
back to a pure high temperature phase. It shows that the stable
room temperature solubility range of alkali or alkaline-earths
with lanthanum in LAMOX compounds is smaller than originally
described in previous papers and does not exceed a few percent
(typically 2 to 3% for Ca or K, maybe even less). It also explains
the discrepancies between the solid solution limits published in the
literature (see section 4.1), since metastability is highly dependent
on the cooling rate and on grain sizes, which can vary significantly
from one study to another.
Fig. 11). The phase compositions are in quantitative agreement
with such an analysis including the scheelite type molybdate,
according to the following transformations:
The most spectacular illustration of this phenomenon concerns
the doubly substituted K6-LM-W70 (La1.88K0.12Mo0.6W1.4O8.88
)
compound. In this composition, the two-phases fully decomposed
◦
state observed at 740 C is the stable state at room temperature.
The absence of any LAMOX phase in the mix reflects molybdenum
depletion from the LAMOX phase to the benefit of tungsten, to
the point that another a-La2W2O9 structural type (ICSD #93721),
with a smaller amount of molybdenum, becomes more stable (see
Fig. 11 Phase localisation and compositional demixing in the
La2Mo2O9-La2W2O9-KLaMo2O8 ternary phase diagram (green
LAMOX type structure, blue = La2W2O9 type structure, red = demixing
=
path of the K6-LM-W70 composition).
100 | Dalton Trans., 2010, 39, 93–102
This journal is
The Royal Society of Chemistry 2010
©

View Article Online
Fig. 12 The complex phase transformations observed in the Ca4-LM compound (c) as explained by the metastabilisation of the pure b-LAMOX phase,
and superimposition of anionic vacancy ordering at intermediate T (b) and cationic demixing at higher T (a) triggered by thermal stabilisation (see text).
The above considerations shed some light on the very complex
behaviour of the Ca4-LM compound already described earlier,6,7
in which the following, very unusual, succession of phases have
been observed upon heating:
the high temperature structural study of Tealdi et al. on K5-
LM23 really reflects intrinsic behaviour of LAMOX compounds
or peculiarities linked to the growing instability of the metastable
phase when temperature increases. Metastability can affect both
oxygen vacancy ordering and cationic segregation inside different
thermal ranges related to the relative mobilities of oxide ions and
cations.
Such metastability problems are not specific to the LAMOX
family. For instance, in what is considered as the standard solid ox-
ide electrolyte, yttria stabilised zirconia (YSZ), substantial yttrium
segregation is observed simply by cooling down the sample.24,25
Besides, according to the yttria-zirconia phase diagram, cubic
YSZ should also be considered as a metastable phase at room
temperature.1 It is therefore likely that metastability and counter-
ion segregation are current phenomena in oxide-ion conductors.
Most probably, ageing problems encountered at the solid oxide fuel
cell core level, which are usually considered as resulting from the
reaction with electrode materials, are partly due to the slow kinetics
of ionic migration at an intermediate temperature, as evidenced in
this paper.
b → a → b → demixing → b → decomposition.
The key element is the metastability of the pure high temperature
b-LAMOX phase at room temperature. The thermal range of
metastability of the pure LAMOX phase (see Fig. 12(a)) is
larger than the thermal range of metastability of the b-LAMOX
phase (see Fig. 12(b)). This is due to the fact that, in an oxide
ion conductor, anions are the most mobile species compared
to cations, and therefore the energy necessary to move oxide
ions (i.e. order/disorder transition of oxygen vacancies) is lower
than that necessary to segregate cations (i.e. substituting cation
demixing). Consequently, the stable state relative to oxygen
vacancy distribution (a phase) is recovered at lower temperature
than the stable state relative to cationic distribution (demixed
state). It follows that the thermal phase diagram (Fig. 12(c)) is
a simple superimposition of the thermal phase diagram of the
oxygen vacancy distribution (Fig. 12(b)) upon that of the cationic
distribution (Fig. 12(a)).
Acknowledgements
The authors thank CNRS and Re´gion Pays de la Loire for A. Selmi
PhD thesis grant. G. Petot-Ervas, C. Petot and G. Dezanneau
at SPMS (Ecole Centrale Paris), as well as Y. Piffard at Institut
Jean Rouxel (IMN Nantes), are acknowledged for their helpful
discussions.
7. Conclusion
In this paper, we have given evidence of the extension to alkali
and alkaline-earth substitutes of the demixing/recombination
phenomenon already observed in Ca-substituted LAMOX
compounds,7 which gives rise, when this compound is annealed in
the demixing thermal range, to the eruption of CaMoO4 straight
solid streams from pores within LAMOX grains. The phenomenon
has been identified as originating from a temperature-dependent
solid-solution limit of alkali and alkaline-earths (for lanthanum)
in LAMOX compounds, with higher solubility at higher tem-
perature. It is responsible for discrepancies in the solid solution
limits published in the literature. The room-temperature solid
solution range is very narrow (lower than 2 or 3 mol% for Ca
and K). All previous studies dealing with higher substitutional
contents therefore concern metastable, not stable state of matter,
which demixes when the temperature is high enough to recover
the stable diphasic state. For instance, one can wonder whether
References
1 M. Kilo, in Defects and Diffusion in Ceramics: An Annual Retrospective
Vii, ed. D. J. Fisher, Trans Tech Publications Ltd, Zurich-Uetikon,
2005, vol. 242–244, pp. 185–253.
2 P. Lacorre, F. Goutenoire, O. Bohnke, R. Retoux and Y. Laligant,
Nature, 2000, 404, 856–858.
3 G. Corbel, S. Mestiri and P. Lacorre, Solid State Sci., 2005, 7, 1216–
1224.
4 D. Marrero-Lopez, J. Pena-Martinez, J. C. Ruiz-Morales, D. Perez-
Coll, M. C. Martin-Sedeno and P. Nunez, Solid State Ionics, 2007, 178,
1366–1378.
5 H. C. Chang, D. S. Tsai, W. H. Chung, Y. S. Huang and M. V. Le, Solid
State Ionics, 2009, 180, 412–417.
This journal is
The Royal Society of Chemistry 2010
Dalton Trans., 2010, 39, 93–102 | 101
©

View Article Online
6 A. Selmi, G. Corbel and P. Lacorre, Solid State Ionics, 2006, 177, 3051–
3055.
7 A. Selmi, G. Corbel, S. Kojikian, V. Voronkova, E. Kharitonova and P.
Lacorre, Eur. J. Inorg. Chem., 2008, 1813–1821.
8 S. Georges, F. Goutenoire, Y. Laligant and P. Lacorre, J. Mater. Chem.,
2003, 13, 2317–2321.
9 J. Rodriguez-Carvajal, Phys. B, 1993, 192, 55–69.
10 R. Subasri, D. Matusch, H. Nafe and F. Aldinger, J. Eur. Ceram. Soc.,
2004, 24, 129–137.
16 X. P. Wang, Z. J. Cheng and Q. F. Fang, Solid State Ionics, 2005, 176,
761–765.
17 S. Georges, F. Goutenoire, O. Bohnke, M. C. Steil, S. J. Skinner, H. D.
Wiemhofer and P. Lacorre, J. New Mat. Electrochem. Syst., 2004, 7,
51–57.
18 D. Marrero-Lopez, J. Canales-Vazquez, J. C. Ruiz-Morales, J. T. S.
Irvine and R. Nunez, Electrochim. Acta, 2005, 50, 4385–4395.
19 D. Monceau, M. Filal, M. Tebtoub, C. Petot and G. Petotervas, Solid
State Ionics, 1994, 73, 221–225.
11 D. Marrero-Lopez, D. Perez-Coll, J. C. Ruiz-Morales, J. Canales-
Vazquez, M. C. Martin-Sedeno and P. Nunez, Electrochim. Acta, 2007,
52, 5219–5231.
12 T. M. He, Y. L. Huang, Q. He, Y. Ji, L. Pei, J. Liu and Z. Lu, J. Alloys
Compd., 2005, 388, 145–152.
13 F. Goutenoire, O. Isnard, E. Suard, O. Bohnke, Y. Laligant,
R. Retoux and P. Lacorre, J. Mater. Chem., 2001, 11, 119–
124.
20 Y. Laligant, A. Le Bail and F. Goutenoire, J. Solid State Chem., 2001,
159, 223–227.
21 J. A. Collado, M. A. G. Aranda, A. Cabeza, P. Olivera-Pastor and S.
Bruque, J. Solid State Chem., 2002, 167, 80–85.
22 O. Labidi, J. P. Wignacourt, P. Roussel, M. Drache, P. Conflant and H.
Steinfink, Solid State Sci., 2004, 6, 783–790.
23 C. Tealdi, L. Malavasi, C. Ritter, G. Flor and G. Costa, J. Solid State
Chem., 2008, 181, 603–610.
14 S. Basu, P. S. Devi and H. S. Maiti, Appl. Phys. Lett., 2004, 85, 3486–
24 Y. Y. Lei, Y. Ito, N. D. Browning and T. J. Mazanec, J. Am. Ceram.
Soc., 2002, 85, 2359–2363.
3488.
15 C. Tealdi, G. Chiodelli, L. Malavasi and G. Flor, J. Mater. Chem., 2004,
14, 3553–3557.
25 A. Rizea, G. Petot-Ervas, C. Petot, M. Abrudeanu, M. J. Graham and
G. I. Sproule, Solid State Ionics, 2007, 177, 3417–3424.
102 | Dalton Trans., 2010, 39, 93–102
This journal is
The Royal Society of Chemistry 2010
©