492
B.J. Mitchell et al. / Materials Research Bulletin 35 (2000) 491–501
1. Introduction
Recently, the mixed-conducting oxide SrFeCo0.5Oy (SFC2) has been identified as a
potential dense ceramic membrane that can be used to separate gas at elevated temperatures.
Indeed, Balachandran et al. [1] demonstrated that extruded membrane tubes of SFC2 can be
used for partial oxidation of methane to produce syngas (CO ϩ H2) in a methane conversion
reactor operating at Ϸ850°C. The oxygen flux obtained from the separation of air in this
reactor is commercially feasible, and the use of this technology can significantly reduce the
cost of oxygen separation [2].
Typically, ABO3Ϫ␦ perovskites which, as synthesized are single-phase materials, have
been extensively studied for oxygen diffusion properties or membrane applications [3–6].
SFC2 differs from the ABO3Ϫ␦ perovskites in that it is not a single-phase material under
standard synthesis conditions and times. Four known phases, Sr2Fe3ϪaCoaO6.5Ϫ␦ (236),
SrFe1ϪbCobO3Ϫ␦ (perovskite/brownmillerite), Co3ϪcFecO3 (rocksalt), and Co3ϪdFedO4
(spinel), are certainly identifiable; their presence depends upon synthesis conditions [7,8].
Guggilla and Manthiram [9] and Ma et al. [10] have shown that substitution of Co for Fe in
SrFe1.5ϪxCoxOy produces single-phase materials over a limited Co range (with limiting
composition ϷSrFe1.2Co0.3Oy). This finding suggests that Co3ϩ is not particularly stable
within the 236 structure. The enhanced performance of SFC2 when compared with that of the
individual phases is not understood and is the focus of continuing research.
The commonly accepted crystallographic representation for the layered structure
Sr2(Fe,Co)3O6.5Ϫ␦ is shown in Fig. 1. This structure is formed from perovskite and double-
layer intergrowths. The Fe/Co atoms are in a ϩ3 oxidation state and occupy three distinct
environments: octahedral (sixfold), trigonal bipyramidal (distorted fivefold), and square
pyramidal (distorted fivefold). The orthorhombic structure of 236 (Iba2 space group) is
related to the basic perovskite cell by the relationships a ϭ 2͌2ap, b Ϸ 5ap, c ϭ ͌2ap.
¯
Fig. 2 shows the crystal structure of the Sr(Fe,Co)O3Ϫ␦ perovskite (Pm3m space group)
and the defect ordered Sr2(Fe,Co)2O5 (Ibm2 space group shown) brownmillerite. In the
perovskite structure, six oxygen atoms that form regular octahedra surround the Fe/Co atoms.
The structure can readily adopt substoichiometry in the oxygen sub lattice and oxygen
vacancies are usually formed randomly. Brownmillerite has a supercell ordering of vacancies
in the perovskite structure a ϭ ͌2ap, b ϭ 4ap, c ϭ ͌2ap, where layers of corner sharing
(Fe,Co)O6 octahedra alternate along b with (Fe,Co)O4 distorted tetrahedra. The ordered
vacancies can be disordered at high temperature, where the material adopts the disordered
perovskite structure.
Because any operational membrane is subject to a wide range of pO2’s with air at one side
(pO2 Ϸ 10Ϫ0.7 atm) and at the methane side (pO2 Ϸ 10Ϫ18 atm), there must be a gradient
across the tube. How the pO2 and oxygen diffusion actually varies from point to point across
an operating membrane is outside the scope of this paper; Bouwmeester et al. [11] and Kim
et al. [12] address this matter in more detail. Instead, we will focus on the phase compositions
that are present at some pO2 values. The stabilities of the aforementioned phases in SFC2
under the probable pO2’s, which have until now been unknown, can be measured using
in-situ diffraction techniques. Neutrons offer advantages over X-rays for studying
Sr(Fe,Co)1.5Oy membrane materials since the coherent scattering lengths are not related to