Solid State Ionics 104 (1997) 57–65

Oxygen permeation, electrical conductivity and stability of the

perovskite oxide La 0.2 Sr 0.8 Cu 0.4 Co 0.6 O 32x
S. Kim, Y.L. Yang, R. Christoffersen, A.J. Jacobson*
Department of Chemistry, University of Houston, Houston, TX 77204 -5641, USA
Received 7 July 1997; accepted 22 July 1997

Abstract

The cubic perovskite La 0.2 Sr 0.8 Cu 0.4 Co 0.6 O 32x has high electronic and ionic conductivity at high temperatures. The
electronic conductivity initially increases with increasing temperature but then decreases due to loss of oxygen atoms from
the structure. The oxygen permeability measured on disc membranes is comparable to values reported for other
cobalt-containing perovskite oxides with high strontium content. The permeation flux decreases with time due to a
transformation of the disordered cubic perovskite to an ordered structure related to Brownmillerite. This ordered structure is
stable below 9758C in an oxygen partial pressure pO 2 5 0.01 atm. Membranes that have been maintained at high temperature
in an oxygen partial pressure gradient for . 400 h show decomposition to a multi-phase assemblage that includes
non-perovskite strontium cuprates. This decomposition may also contribute to the flux decline.

Keywords: Oxygen permeation; Perovskite oxide; Mixed conductivity; La 0.2 Sr 0.8 Cu 0.4 Co 0.6 O 32x

1. Introduction fluxes through perovskite membranes were first

studied by Teraoka et al. [1–3], who reported results
Mixed conducting perovskite oxides have received for a number of (La,Sr)(Co,Fe)O x compositions. A
considerable attention as possible candidate materials characteristic feature of these perovskite composi-
for application in solid state electrochemical devices tions that contributes to their suitable oxygen trans-
such as solid oxide fuel cells and sensors. These port properties is their high concentration of oxygen
materials also hold promise for use as ceramic vacancies. In order for the compounds to be of
membranes designed for oxygen separation from air. practical use as membranes, however, it is desirable
The driving force for oxygen permeation through that these large numbers of vacancies remain struc-
such membranes is the oxygen partial pressure turally disordered under the temperature and pO 2
gradient across the membrane. Since the flux of conditions that obtain on both sides of the mem-
oxygen ions is charge compensated internally by the brane. Otherwise, undesirable vacancy trapping ef-
transport of electronic charge carriers, these mem- fects [4] or volume changes [5] associated with the
branes can be used for gas separation without the development of a vacancy-ordered superstructure
need for electrodes or external circuitry. Oxygen may result. In particular, Adler et al. [4] have
suggested that specific types of local ordering re-
*Corresponding author. E-mail: ajjacob@uh.edu sulting in nano-scale ordered domain structures can

0167-2738 / 97 / $17.00  1997 Elsevier Science B.V. All rights reserved.

PII S0167-2738( 97 )00427-X
58 S. Kim et al. / Solid State Ionics 104 (1997) 57 – 65

be as detrimental to oxygen mobility as long-range The sample was cooled to room temperature by
ordering. shutting off the furnace, then reground and refired.
As well as being stable against undesirable poly- This treatment was repeated twice. The resulting
morphic transformations such as vacancy ordering, a powder sample was characterized by X-ray powder
membrane compound must resist decomposition to diffraction (XRD) and transmission electron micro-
other phases under the reducing conditions that exist scopy (TEM). X-ray diffraction measurements were
on the low pO 2 side of the membrane. Such de- made with a Scintag XDS 2000 automated diffrac-
composition is not necessarily limited to simple tometer in u –u geometry with a flat plate sample and
reduction reactions, since the pO 2 gradient across the Cu Ka ( l 5 1.54178 A) ˚ radiation. TEM was per-
membrane may cause cations to diffuse along with formed with a JEOL 2000FX microscope operating
the oxygen ions, causing the membrane to decom- at 200 kV.
pose by ‘kinetic demixing’ [6,7]. Irrespective of the The measurement of total oxygen content in
mechanism by which it occurs, decomposition typi- selected samples was performed by thermogravi-
cally results in a time-dependent decrease in the metric analysis as follows. Samples were preheated
oxygen flux through the membrane. in a thermobalance (DuPont 990) in 80% N 2 1 20%
Very recently, a new oxygen-deficient perovskite O 2 at 9508C for 2 h to remove traces of water,
composition, La 0.2 Sr 0.8 Cu 0.4 Co 0.6 O 32x , was synthes- cooled to ambient temperature, and then reduced by
ized by Genouel et al. [8]. This compound was heating in 95% N 2 1 5% H 2 (100 mL min 21 ) to
21
reported to have a cubic perovskite structure, sug- 9008C at 58C min . The sample was held at 9008C
gesting that both the cation sublattices and the for 3 h to ensure complete reduction.
oxygen vacancies are disordered. X-ray absorption Detailed measurement of oxygen contents as a
spectroscopy indicated that cobalt ions are present in function of temperature were made using a TA
the Co(III) state. The copper ions show mixed Instruments thermogravimetric analyzer. Measure-
Cu(II) / Cu(III) oxidation states. The system is un- ments were made up to 9008C at a heating rate of
usual in that it apparently contains a large con- 28C min 21 at two oxygen partial pressures, pO 2 5
centration of randomly distributed oxygen vacancies 0.2 and 0.01 atm, obtained using O 2 / N 2 mixtures.
at x 5 | 0.5. Phase transition temperatures were determined by
In this paper we evaluate the compound differential thermal analysis (SDT 2960 DTA-TGA,
La 0.2 Sr 0.8 Cu 0.4 Co 0.6 O 32x as a possible oxygen trans- TA Instruments). Measurements were made in argon
port membrane material and present data on its at an oxygen partial pressure of 0.008 atm (measured
fundamental transport properties, including electrical with an Ametek TM-1B oxygen analyzer) at a scan
conductivity and oxygen transport. Results pertinent rate of 208C min 21 .
to some of the additional membrane performance Structural changes occurring in La 0.2 Sr 0.8 Cu 0.4
criteria mentioned above, such as structural and Co 0.6 O 32x under conditions more reducing than the
phase stability, have also been obtained. original in-air synthesis were studied by XRD.
Samples ( | 5 g) in platinum boats were equilibrated
at 8478C and 9758C in N 2 / O 2 with pO 2 5 0.01 atm
2. Experimental for 20 h in a tube furnace. The samples were
quenched to room temperature under the same
The compound La 0.2 Sr 0.8 Cu 0.4 Co 0.6 O 32x was pre- atmosphere to preserve the high-temperature phase.
pared by a high temperature synthesis route using The as-quenched powder samples were immediately
La 2 O 3 (Aldrich, 99.99%), CuO (Aldrich, 99.99%), examined by XRD.
Co 3 O 4 (Aldrich, 99.995%), and SrCO 3 (Aldrich, For preparation of membranes, the as-synthesized
99.995%) as starting materials. The predried starting powder described above was ball-milled for 48 h in
materials (La 2 O 3 , 3h at 11008C, CuO, Co 3 O 4 , and ethanol. The powder was then packed into a tubular
SrCO 3 , 24 h at 1208C) were mixed in the stoichio- mold made of silicon resin and pressed at 40 000 psi
metric ratios and were heated at 28C min 21 in a in a cold isostatic press. Disc membranes with
platinum crucible in air to 10008C and held for 10 h. appropriate thicknesses were sliced from the re-
 S. Kim et al. / Solid State Ionics 104 (1997) 57 – 65 59

sulting cylinder with a diamond saw. The discs were and in a 1% O 2 / N 2 atmosphere at a flowrate 100 mL
heated at 28C min 21 in air to 10508C and held for 1 min 21 . The sample bars were prepared from a
h before cooling to room temperature also at 28C sintered disc. Measurements were made on both
min 21 . The resulting sintered discs had diameters of increasing and decreasing the temperature. Equilibra-
| 13 mm. The measured densities were typically tion times at each temperature were typically | 1 h.
. 90% of the calculated theoretical density assuming A Keithley 220 programmable current source and
the cubic perovskite unit cell dimensions. As a final Keithley 196 digital multimeter were used for DC
step, both sides of the discs were ground using 600 resistivity measurements.
grit SiC polishing paper.
The permeation measurement apparatus has been
described in detail in a previous publication [9]. The 3. Results and discussion
discs were sealed between two gold-rings and the
effective area for oxygen permeation was 0.5 cm 2 for The X-ray diffraction pattern of the as-synthesized
discs with 13 mm diameters. Air was fed through a La 0.2 Sr 0.8 Cu 0.4 Co 0.6 O 32x powder is shown in Fig. 1.
thin stainless tube into one side at typically 40 mL This sample has an oxygen stoichiometry of 3 2 x 5
min 21 , while helium gas was swept through the 2.57 as determined by thermogravimetric reduction
other side with a similar flowrate. Flowrates were in hydrogen. The X-ray pattern (Fig. 1) is consistent
controlled by MKS 247C mass flow controllers. An with formation of a cubic perovskite structure with
MTI M200 gas chromatograph connected to the exit a 5 3.876(7) A. ˚ These results are in agreement with
on the sweep side was used to measure both oxygen the previous work of Genouel et al. [8] although the
and nitrogen concentrations. The gas chromatograph oxygen stoichiometry is slightly higher. Our micro-
was frequently calibrated using a standard gas mix- probe characterization of membrane discs prepared
ture of O 2 and N 2 in helium. The disc temperature from the above powder before permeation show
was monitored by an inserted thermocouple. Any some local deviation in La / Sr ratio from the ideal
small oxygen leakage due to pin-holes or cracks on 20 / 80 value. This inhomogeneity may be a feature
the disc were monitored by measuring the con- of the original powder or may have developed during
centration of nitrogen in the helium sweep gas. The sintering of the membrane disc.
oxygen fluxes were then corrected for oxygen leak- In the X-ray powder pattern of the as-synthesized
age using the corresponding nitrogen concentrations. sample we observed no extra lines indicative of an
The total oxygen leakage from ambient air coming oxygen vacancy-ordered superstructure. This appar-
through the seals and / or from the pin-holes or cracks
in the membrane was typically , 1% of the total
oxygen flux. At each temperature, 24 h were allowed
for equilibration. Oxygen permeation measurements
were performed between 750 and 9508C.
After the permeation experiments the discs were
embedded in epoxy, then sectioned in half and
polished to facilitate cross-sectional examination by
scanning electron microscopy (SEM) and X-ray
microanalysis using a JEOL JXA-8600 electron
microprobe. Microprobe data and observations of the
post-permeation discs were compared to results
obtained on as-sintered discs that were similarly
sectioned and polished.
Electrical conductivities of sample bars (12 mm 3
3 mm 3 3 mm) were measured by a four-probe DC
method using platinum electrodes as a function of Fig. 1. X-ray powder diffraction patterns of the as-synthesized
temperature (1008C to 9158C) in 80% N 2 –20% O 2 (lower) and the reduced (upper) La 0.2 Sr 0.8 Cu 0.4 Co 0.6 O 32x samples.
60 S. Kim et al. / Solid State Ionics 104 (1997) 57 – 65

ent lack of vacancy ordering, at least as indicated by

X-ray diffraction, is the characteristic feature of this
compound, and is particularly unusual in light of the
fact that the sample in question was furnace cooled
as opposed to quenched. In other perovskites that
have high-vacancy contents and are structurally
similar to the present sample [4], high-resolution
TEM (HRTEM) and electronic diffraction studies
have shown that vacancy-ordering does occur, but
only in the form of local ordered domains | 5 nm in
size in which the unit cell is doubled [4]. Selected-
area electron diffraction patterns of our as-synthes-
ized sample exhibit a complex patterns of diffuse
scattering indicative of short-range atomic ordering
(Fig. 2), but the pattern of relatively continuous
diffuse intensity between the h011j reflections is not
Fig. 3. Variation of the oxygen stoichiometry of the as-synthesized
consistent with the doubled-cell microdomain struc- La 0.2 Sr 0.8 Cu 0.4 Co 0.6 O 32x versus temperature under pO 2 50.2 and
tures previously reported [4]. We conclude that the 0.01 atm.
as-synthesized sample can be regarded as largely
disordered in a long-range sense, with only a small
degree of short range order, which is not necessarily under pO 2 5 0.2, the composition remained constant
associated with oxygen vacancies. until | 3508C and then lost oxygen continuously up
The variation in the stoichiometry (3 2 x) of to a composition of 2.45 at 9008C. Reoxidation of
La 0.2 Sr 0.8 Cu 0.4 Co 0.6 O 32x at pO 2 5 0.2 and 0.01 atm the sample on cooling at the same rate was complete-
is shown in Fig. 3. The initial stoichiometry of the ly reversible above 3508C. Below 3508 a small
as-synthesized sample was 2.57. On heating to 9008C hysteresis was observed and the final composition
was slightly less oxidized than that of the initial
sample. At this stage the oxygen partial pressure was
lowered to 0.01 atm and the sample was reheated to
9008C. The weight loss corresponded to a com-
position 3 2 x 5 2.42 at 9008C. On cooling, the
sample reoxidized to a final composition of 2.545.
These data can be compared with the data reported
previously. Genouel et al. [8] reported a composition
of 2.42 for a sample quenched from 10008C in air.
Extrapolation of the TGA data in Fig. 3 to 10008C
would give 3 2 x 5 2.43, in good agreement.
Possible changes in the structure of
La 0.2 Sr 0.8 Cu 0.4 Co 0.6 O 32x at low oxygen partial
pressures (suggested by the flux decay observed in
the permeation experiments, see below) were investi-
gated by X-ray diffraction studies of quenched
samples and by DTA. The X-ray powder pattern of a
sample quenched from 9758C under pO 2 5 0.01 atm
showed that the sample had changed from cubic
Fig. 2. A [100] electron diffraction pattern of the
La 0.2 Sr 0.8 Cu 0.4 Co 0.6 O 32x powder sample, as-synthesized at perovskite to a phase apparently related to Brown-
10008C in air and cooled at 28C h 21 . Note the presence of diffuse millerite Ca 2 Fe 2 O 5 . The diffraction lines shown in
scattering intensity along the h011j reciprocal lattice rows. Fig. 1, although broad, could be indexed with an
 S. Kim et al. / Solid State Ionics 104 (1997) 57 – 65 61

aœ2 3 4a 3 aœ2 Brownmillerite unit cell with Samples that were furnace cooled in air in the
approximate lattice constants a 5 5.354 A,˚ b 5 16.05 present work have oxygen stoichiometry close to 2.5
˚A and c 5 5.421 A ˚ shown in Fig. 1. A sample and show diffraction lines characteristic of only the
quenched from 8478C gave a more complex powder perovskite structure identical to those that were
pattern with extra lines in addition to those from the annealed at 4508C for 24 h in oxygen [8]. The results
Brownmillerite phase, suggesting the presence of a suggest that the slow transformation to an ordered
two-phase mixture or a single phase with a more oxygen vacancy structure may be coupled to slow
complex vacancy ordering scheme. interdiffusion and rearrangement of the cations.
To investigate the temperature dependence of Some evidence for cation redistribution in the con-
transformation of the sample to the Brownmillerite- version of perovskite to Brownmillerite has previous-
like phase in more detail, a sample was prepared for ly been reported for SrCo 0.8 Fe 0.2 O 2.5 [10].
DTA analysis by annealing at 9508C in nitrogen for The results of electrical conductivity measure-
20 h and then cooling to ambient temperature. DTA ments on La 0.2 Sr 0.8 Cu 0.4 Co 0.6 O 32x at pO 2 5 0.20
scans were then obtained for this material by heating and 0.01 atm are shown in Fig. 5. The conductivity
and cooling in argon (pO 2 5 0.008) between 3258 linearly increases with increasing temperature at
and 10508C (Fig. 4). On heating, two endothermic temperatures below 580 K for pO 2 5 0.2 atm and
peaks were observed at 8628C and 9958C. The 690 K for pO 2 5 0.01 atm, respectively. Activation
transitions were reversible and two exothermic peaks energies of 4.36 and 7.57 kJ mol 21 can be derived
at 8258C and 9598C were observed on cooling. The according to the following Eq. (1). At higher tem-
endothermic peak at 8628C and the corresponding peratures, broad curvature can been seen in both sets
of conductivity data and they can be fitted according
exothermic peak 8258C correspond to very small
to Eq. (4) below. The general behavior is
heat changes and their origin is not yet known. The
similar to that reported recently for several
phase transition from the perovskite to Brownmiller-
La 12x Sr x Co 12y Fe y O 32x compositions [11]. The gen-
ite is most probably associated with the larger event
eral form of the conductivity is consistent with the
that occurs at 9958 and 9598C on heating and cooling
small polaron mechanism and can be expressed by
respectively. The phase transformation between
[12,13]:
perovskite and Brownmillerite structures only occurs
after long annealing under reducing conditions. s T 5 A exp(2Ea /kT ), (1)

Fig. 4. DTA data for the La 0.2 Sr 0.8 Cu 0.4 Co 0.6 O 32x sample an- Fig. 5. Electrical conductivity data versus 1000 /T under pO 2 50.2
nealed at 9508C in nitrogen for 20 h and slowly cooled to room and 0.01 atm. Solid lines are fitted data according to Eqs. (1) and
temperature. (4).
62 S. Kim et al. / Solid State Ionics 104 (1997) 57 – 65

where A is a constant containing a carrier con-

centration term, Ea is the activation energy, and k is
the Boltzmann constant. If the concentration of
charge carriers is constant, then plots of log s T
versus 1 /T are linear. When La is partially substi-
tuted by Sr, charge neutrality is maintained by the
oxidation of the B site cation and / or by the intro-
duction of oxygen vacancies according to the
¨
Kroger–Vink expression
[Sr 9La ] 5 [B B? ] 1 2[V O?? ]. (2)

At lower temperatures, the charge compensation is

predominantly by B cation oxidation [B ?B ] and the
increase in conductivity with increasing pO 2 (Fig. 5)
is consistent with the expected p-type electronic
conductivity. At temperatures greater than 3508C, the
thermogravimetric analysis data (Fig. 3) show that Fig. 6. Oxygen permeation flux versus 1000 /T. Open squares:
the oxygen content decreases with increasing tem- membrane thickness 0.8 mm; and solid squares: membrane
perature. The 32x dependence on temperature is thickness 2.0 mm.
approximately linear from 3508C to 9008C. The
oxygen loss is expected to decrease the electronic
hole concentration by reducing the number of oxi- An oxygen flux of 2.3310 27 mol cm 22 s 21 was
dized B cations according to the reaction obtained for the 0.8 mm thick membrane at 8638C
2B ?B 1 O xO → V ??O 1 B xB 1 ]12 O 2 (g). (3) with oxygen partial pressures of 0.2 and 4.37310 23
atm at the high and low pressure sides, respectively.
The form of the dependence of the conductivity on The oxygen permeation flux shows a strong depen-
temperature can be modified to include the reduction dence on the membrane thickness suggesting that the
in carrier concentration due to the linear decrease in overall oxygen permeation rate through this mem-
the oxygen content, as previously suggested by brane material is determined by bulk diffusion. The
Stevenson et al. [11]: flux at 8638C is comparable to the value of 4.63
10 27 mol cm 22 s 21 for La 0.2 Sr 0.8 CoO 32x extrapo-
s T 5 A(1 2 bT ) exp(2Ea /kT ). (4)
lated from results of van Doorn et al.
The data in Fig. 5 were fitted to this model above [14] obtained under similar conditions. A similar
600 K and to a model assuming constant carrier value (|5310 27 mol cm 22 s 21 ) was also obtained
concentration below 600 K. This treatment re- for La 0.2 Sr 0.8 Co 0.4 Fe 0.6 O 32x by Teraoka et al. [2].
produces general trends in the conductivity data very A useful membrane material, in addition to having
well (Fig. 5). The conductivity data are in the range a high permeation flux, must be stable for long
250 to 600 S cm 21 from ambient temperature to times in the appropriate oxygen partial pressure
9008C in air and are comparable to values reported gradient. We noted that for membranes of
for (La,Sr)(Fe,Co)O 32x phases. La 0.2 Sr 0.8 Co 0.4 Fe 0.6 O 32x a decay in the measured
Oxygen permeation fluxes through two disc mem- oxygen flux occurred when the membrane was kept
branes with thicknesses of 0.8 mm and 2.0 mm were for long times at high temperature. For example, the
measured as a function of temperature in the range oxygen fluxes through membranes with 0.8 and 2.0
7508C to 9608C and are shown in Fig. 6. The mm thicknesses after operation for 240 h at 7658C
apparent activation energy based on these data is were 5.7310 28 and 3.2310 28 mol cm 22 s 21 ,
109.965.0 kJ mol 21 . The data were not corrected in respectively. These values are 21.7% and 22.4%
these measurements for the effects of edge leakage. lower than the oxygen permeation fluxes shown in
 S. Kim et al. / Solid State Ionics 104 (1997) 57 – 65 63

Fig. 6 which were measured at the beginning of the clude perovksite phases with enriched La and Co
experiment. contents compared to the starting composition (e.g.,
We interpret the decline in the oxygen flux with La 0.36 Sr 0.64 Co 0.65 Cu 0.35 O 32x ), and grains of a non-
time to be the result of two different effects. One of perovskite Sr–Cu oxide with minor La and Co.
the main reasons this material was chosen initially as These decomposition products occur to greater or
a candidate for an oxygen membrane application was lesser extent in all areas of the sample, but were
the large concentration of disordered oxygen vac- found in their greater concentrations closer to the
ancies observed in the as-synthesized compound. ‘‘reduced’’ edge of the sample exposed to the helium
The X-ray and DTA data described above show, sweep (Fig. 7). The backscattered electron image in
however, that when the compound is annealed for Fig. 7 shows in detail that the brighter, La-rich
long times under reducing conditions, the disordered decomposition products appear to concentrate along
perovskite structure transforms to an ordered Brown- cracks or grain boundaries, consistent with decompo-
millerite structure. The transformation of the mem- sition beginning closer to the surface and then
brane to this ordered oxygen vacancy structure is proceeding into the membrane along high-diffusivity
expected on the low oxygen partial pressure side and pathways.
should reduce the oxygen permeation flux [9,15,16]. The decomposition assemblage suggests an overall
Unlike SrCo 0.8 Fe 0.2 O 32x , all of the permeation mea- reaction in which formation of a Sr-rich product
surements on La 0.2 Sr 0.8 Co 0.4 Fe 0.6 O 32x are obtained phases occurs in conjunction with loss of Sr and Cu
below the order–disorder transition temperature. from the starting perovskite composition. Such
A second effect that can lead to flux decline is ‘‘strontium segregation’’ has been reported previous-
decomposition of the membrane material into other ly in La 0.3 Sr 0.7 CoO 32x membranes where it was
phases driven by the oxygen partial pressure gra- attributed to the formation of SrO by kinetic demix-
dient. Decomposition of both the 0.8 and 2.0 mm ing [13,17–21]. In this latter case the SrO formation
membranes after the permeation experiments was was tied to an increase in the oxygen permeation flux
evident based on electron microprobe SEM imaging due to formation of a more porous surface layer. In
and microanalysis. The decomposition products in- the present case, there is no evidence of a porous

Fig. 7. Back-scattered electron image of the 3.0 mm-thick membrane showing the edge of the sample on the side of the He gas sweep. The
network of bright material is a La–Cu rich non-perovskite decomposition product.
64 S. Kim et al. / Solid State Ionics 104 (1997) 57 – 65

layer and segregation apparently leads to a flux the mechanism for the segregation involves kinetic
decline. Further experiments to separate the relative demixing due to different cation diffusivities [6]. The
effects of the order–disorder transition and phase origin of the decomposition we observe in
separation on the oxygen permeation are required. La 0.2 Sr 0.8 Cu 0.4 Co 0.6 O 32x is not yet clear but it may
also contribute to the observed flux decline. Further
study and analysis of the long term stability and the
4. Conclusions decomposition mechanisms of these complex perov-
skite compositions is clearly needed.
The perovskite oxide La 0.2 Sr 0.8 Cu 0.4 Co 0.6 O 32x
has high ionic conductivity due to the high con-
centration of oxygen vacancies present in the dis-
Acknowledgements
ordered structure and high p type electronic con-
ductivity. The combination of properties is reflected
We thank the Texas Center for Superconductivity
in the high initial permeation flux observed in dense
and the Robert A. Welch Foundation for financial
membranes of this composition. The permeation flux
support of this work. The work was supported in part
is observed to decline significantly with time due to
by the MRSEC program of the National Science
two problems that are of general importance in the
Foundation under Award Number DMR-9632667.
use of these systems in membrane reactors: vacancy
ordering and chemical decomposition. With regard to
vacancy ordering we find that air-synthesized
La 0.2 Sr 0.8 Cu 0.4 Co 0.6 O 32x is a disordered cubic References
perovskite, but upon annealing in 1% O 2 for 20 h,
the X-ray data indicate the formation of a phase [1] Y. Teraoka, T. Nobunaga, K. Okamoto, N. Miura, N.
related to Brownmillerite. The Brownmillerite phase, Yamazoe, Solid State Ionics 48 (1991) 207.
[2] Y. Teraoka, T. Nobunaga, N. Yamazoe, Chem. Lett. (1988)
once formed, shows a reversible phase transforma- 503.
tion at 9958C and 9598C on heating and cooling [3] Y. Teraoka, H. Zhang, S. Furukawa, N. Yamazoe, Chem.
respectively that is most probably associated with Lett. (1985) 1743.
disordering of the oxygen vacancies. The slow rate [4] S. Adler, S. Russek, J. Reimer, M. Fendorf, A. Stacy, Q.
of the initial formation of the Brownmillerite phase Huang, A. Santoro, J. Lynn, J. Baltisberger, U. Werner, Solid
State Ionics 68 (1994) 193.
suggests that the transformation is driven by re-
[5] U. Balachandran, J.T. Dusek, R.L. Mieville, R.B. Poeppel,
arrangements of the cations. The oxide ion diffusion M.S. Kleefisch, S. Pei, T.P. Tobylinski, C.A. Udovich, A.C.
rate is expected to be lower in the more ordered Bose, Appl. Catal. A 133 (1995) 19.
structure and consequently the flux decline that we [6] M. Martin, R. Schmackpfeffer, Solid State Ionics 72 (1994)
observe is due at least in part to this slow structural 67.
[7] H. Schmalzried, W. Laqua, Oxidation of Metals 13 (1981)
rearrangement. Similar studies on La 0.6 Sr 0.4 CoO 32x
339.
membranes below the order–disorder transition have [8] R. Genouel, C. Michel, N. Nguyen, F. Studer, M. Hervieu,
shown that the permeation flux decays slowly on a B. Raveau, J. Solid State Chem. 119 (1995) 260.
time scale of hundreds of hours [16]. [9] L. Qiu, T.H. Lee, Y.L. Yang, A.J. Jacobson, Solid State
Concerning the second problem of chemical de- Ionics 76 (1995) 321.
[10] W.T.A. Harrison, T.H. Lee, Y.L. Yang, D.P. Scarfe, L.M. Liu,
composition we observe formation of a non-perov-
A.J. Jacobson, Mat. Res. Bull. 30 (1995) 237.
skite strontium-rich phase, and associated La enrich- [11] J.W. Stevenson, T.R. Armstrong, R.D. Carneim, L.R. Peder-
ment of the initial starting composition, during son, W.J. Weber, J. Electrochem. Soc. 143 (1996) 2722.
membrane operation in an oxygen partial pressure [12] D. Emin, T. Holstein, Ann. Phys. 53 (1969) 439.
gradient. Strontium segregation has been reported [13] D.P. Karim, A.T. Aldred, Phys. Rev. B 20 (1979) 2255.
[14] R.H.E. van Doorn, H. Kruidhof, H.J.M. Bouwmeester and
previously in La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 32x membranes
A.J. Burggraaf, in: H.U. Anderson, A.C. Khandkar, M. Liu
used for methane conversion and also in (Eds.), Proc. of the First International Conference on
La 0.7 Sr 0.3 CoO 32x membranes in smaller oxygen Ceramic Membranes, Proc. Electrochem Soc. PV95-24,
partial pressure gradients. It has been suggested that Pennington, NJ, 1997, p. 138.
 S. Kim et al. / Solid State Ionics 104 (1997) 57 – 65 65

[15] L.M. Liu, T.H. Lee, L. Qiu, Y.L. Yang, A.J. Jacobson, Mater. [19] J.E. ten Elshof, H.J.M. Bouwmeester, H. Verweij, Appl. Cat.
Res. Bull. 31 (1996) 29. A General 130 (1995) 195.
[16] H. Kruidhof, H.J.M. Bouwmeester, R.H.E. van Doorn, A.J. [20] J.E. ten Elshof, B.A. van Hassel, H.J.M. Bouwmeester,
Burggraaf, Solid State Ionics 63–65 (1993) 816. Catal. Today 25 (1995) 397.
[17] R.H.E. van Doorn, I.C. Fullarton, R.A. de Souza, J.A. Kilner, [21] Y. Teraoka, T. Fukada, N. Miura, N. Yamazoe, J. Ceram.
H.J.M. Bouwmeester, A.J. Burggraaf, Solid State Ionics 96 Soc. Jpn. Inter. Ed. 97 (1989) 523.
(1997) 1.
[18] J.E. ten Elshof, H.J.M. Bouwmeester, H. Verweij, Solid State
Ionics 89 (1996) 81.