On the structural stability and oxygen permeation behavior of inorganic SrCo_0.8Fe_0.2O_3−δ membranes

Abstract

The high oxygen permeability combined with reasonable structural stability of perovskite-type ABO_3−δ compounds is vital for their potential applications in gas separation, solid oxide fuel cells, sensors, etc. Hence, an attempt is made to develop SrCo_0.8Fe_0.2O_3−δ-based dense membranes with sol-gel-derived oxalates and study their phase stability and oxygen permeation. While X-ray diffraction confirms the presence of a perovskite-type cubic phase above 800 °C, X-ray photoelectron spectroscopy reveals the presence of cobalt and iron in 3+ and 4+ oxidation states with O₂ ²⁻, O₂ ⁻ and O⁻ species. The electrical conductivity increases up to a characteristic temperature and decreases slowly thereafter via pronounced carrier scattering. A 1.5-mm-thick membrane displays reasonable oxygen permeability of 1.05 × 10⁻⁶ mol cm⁻² s⁻¹ at 900 °C but has inadequate stability. Partial substitution of iron with zirconium is shown to improve permeability and stability significantly. Thus, SrCo_0.8Fe_0.15Zr_0.05O_3−δ membrane shows promise for oxygen permeation purposes.

Introduction

Perovskite-type cubic oxides with composition ABO_3−δ (A = Sr, La; B = Co, Fe; δ = oxygen deficiency) have ability to accommodate wide range of “δ” with mixed oxidation states of species at B− site [1]. As a consequence, they display significant oxygen permeability and have potential applications in gas separation as inorganic membranes, oxygen pumps, solid oxide fuel cells, oxidative coupling of methane as catalyst, oxygen sensors, etc. [2–7]. Their structural, electrical and other physico-chemical properties can be further controlled and/or improved by varying composition, preparation method, and treatment processes [8]. Numerous studies undertaken generally involve synthesis by solid state reaction of metal oxides and/or salts at elevated temperatures for extended hours [9, 10]. However, in-homogeneity and phase instability of products continue to cause uncertainty in results and property correlations [11]. Sol-Gel is an alternative route for synthesis and offers certain advantages, viz., homogeneous mixing of precursors at atomic level, better stoichiometry control, high purity, and low cost [12]. In an ideal perovskite-type cubic cell with composition (ABO₃), A and B ions are divalent and tetravalent, respectively. While A− ions assume the corner with oxygen ions at the face-centered positions and form together a cubic close-packed arrangement, B− ion occupies the octahedral void at the cube centre. The total positive charge of A− and B− site cations taken together is +6. Whenever the total positive charge is reduced due to mixed valance elements and/or their partial substitutions at A−/B− sites, system neutrality is ensured by additional oxygen vacancies. The system then becomes non-stoichiometric with chemical formula ABO_3−δ, where δ stands for the oxygen deficiency. Owing to excellent thermal, chemical, and structural stability with mixed ionic and electronic conductivity, some of them have found applications in catalytic reactor, energy storage, electrodes, etc.

An important member of the ABO_3−δ family is the strontium iron cobalt oxide (SrCo_0.8Fe_0.2O_3−δ) which exhibits good oxygen permeability but inadequate long-term stability for practical applications at high temperatures [13–16]. The various synthesis routes and conditions reported for SrCo_0.8Fe_0.2O_3−δ membranes with a few important characteristics are given in Table 1.

Table 1 Synthesis route and conditions with a few important characteristics of SrCo_0.8Fe_0.2O_3−δ membranes

The present investigation addresses this issue with sol-gel synthesis of SrCo_0.8Fe_0.2O_3−δ-based membranes and their characterization in terms of phase formation, stability at elevated temperatures, electrical conductivity, and oxygen permeation behavior. The uniqueness of work lies in the (a) preparation of SrCo_0.8Fe_0.2O_3−δ by a simple oxalate-based sol-gel route, (b) in situ phase evaluation at temperatures requiring membrane operation, and (c) correlation of structural changes with conductivity. Besides, evidence is advanced for the existence of (i) a cobalt-deficient brownmillerite-type orthorhombic phase together with Co₃O₄ in the temperature range of 200–700 °C and (ii) perovskite–type cubic phase above 800 °C. X-ray photoelectron spectroscopy (XPS) studies of SrCo_0.8Fe_0.2O_3−δ samples are undertaken for the first time to gather information about the multiple oxidation states of iron and cobalt species. The paper further illustrates high oxygen permeability flux of Jo₂ ∼ 1.65 ml cm⁻² min⁻¹ (or 1.23 × 10⁻⁶ mol cm⁻² s⁻¹) at 900 °C for 1-mm-thick SrCo_0.8Fe_0.2O_3−δ membranes but with poor operational stability. The significant improvement in the operational stability realized with partial substitution of iron with zirconium is explained and possible reasons are outlined.

Experimental

SrCo_0.8Fe_0.2O_3−δ powder was synthesized by a sol-gel process [12] using appropriate amounts of anhydrous strontium nitrate {Sr(NO₃)₂}, cobalt nitrate hexahydrate {Co(NO₃)₂.6H₂O}, ferric nitrate nanohydrate {Fe(NO₃)₃.9H₂O}, and oxalic acid dihydrate {C₂H₂O₄.2H₂O} precursors with ethanol {C₂H₅OH} as solvent. For the preparation of SrCo_0.8Fe_0.2-y Zr _y O_3−δ (y = 0.05) powder, an appropriate amount of zirconyl nitrate hydrate {ZrO(NO₃)₂.H₂O} was also taken in the initial precursor mixture. Firstly, metal nitrates and oxalic acid were dissolved separately in ethanol under constant stirring for 30 min to get respective transparent solutions. Due to poor solubility of strontium nitrate in ethanol, distilled water was added dropwise to achieve its complete dissolution. Secondly, the nitrate salt solutions were poured in a tray and mixed thoroughly before adding oxalic acid solution gradually and stirred continuously to form a thick gel. Thirdly, the product was digested (i.e., left as such for reactions to occur) for 12 h and dried subsequently at 80 °C for 24 h in an oven. The resulting material was found to contain numerous sticks of nearly same diameter (0.2–0.3 μm) but of various lengths (0.5–3 μm), arranged parallel to each other, and clustering in different regions with random orientation (Fig. 1). At some places, the sticks overlap and give appearance of a thatched roof, mat, or nest of twigs. The dried sol-gel product was ground to get fine powder—presumably of an oxalate as evident from the Fourier transform infrared (FTIR) analysis. It was subsequently decomposed by raising the temperature at a rate of 4 °C/min up to 950–1150 °C and cooled slowly to room temperature after maintaining there for 5 h to obtain the oxide powder. The exact decomposition condition for each composition was ascertained and optimized by thermo-gravimetric analysis combined with X-ray diffraction studies. The resulting black powder was finally ground and sieved through a 240 mesh for characterization in terms of phase(s), microstructure, chemical states of surface species, etc. For electrical conductivity and oxygen permeation measurements, synthesized powder was first pressed in a die (size ∼10 mm diameter) at a hydraulic pressure of 50 kN and then heated at 1300 °C for 5 h. The pellets/discs/membranes of diameter ∼10 mm and thickness (1–2 mm) thus produced displayed relative density of better than 95 % of the bulk value.

Fig. 1

Scanning electron micrographs (SE mode) of sol-gel-derived oxalate used for obtaining the SrCo_0.8Fe_0.2O_3 ‐ δ powder

An X-ray diffractometer (PANalytical X’ Pert PRO) with CuKα₁ radiation (wavelength λ = 1.540598 Ǻ) was engaged for identification of phase(s) present. The stability and phase changes were analyzed using the X-ray diffraction patterns recorded in situ in the temperature range of 27–1000 °C. A field emission scanning electron microscope (Carl Zeiss NTS GmbH-SUPRA 40VP) was used for observing the morphology of dried sol-gel products, oxide powders, and membranes. The oxidation states of iron and cobalt species lying on the surface were determined by XPS using a PHI 5000 Versa Probe (ULVAC) spectrometer with AlK_α (1486.6 eV) radiation. The electrical conductivity was measured by placing the sintered membrane disc in a sample holder, having silver electrodes and a chromel–alumel thermocouple, up to 700 °C with an LCZ meter (HP 4194 Å) at 1 kHz. The oxygen permeation characteristics of membranes were monitored in the temperature range of 700–1000 °C using a reactor developed in the laboratory with air as a feeding gas (400 ml min⁻¹), helium as a carrier (45 ml min⁻¹), and a gas chromatograph (NUCON, model 5765) equipped with a molecular sieve–5A capillary column. The inconel washer coated with fine alumina (size ∼50 nm) paste served as an efficient sealant for SrCo_0.8Fe_0.2O_3−δ (SCFO) membranes. Moreover, in order to prevent radial leakage of oxygen in the reactor, alumina paste was applied as well over the side-curved surface of the disc membrane. To make correction for the leakage of air through pore and/or around the membrane, the nitrogen content of the collected gas was also measured simultaneously. The sample was maintained at each temperature for about 30–45 min to attain equilibrium before noting down the permeation reading. If C _O and C _N represent the concentration of oxygen and nitrogen, respectively on the sweep side, F (ml min⁻¹) is the helium flow rate, and A (cm²) is the effective membrane cross-sectional area, the oxygen permeation flux Jo₂ (ml cm⁻² min⁻¹) is given by [21]

$$ J{o}_2=\left[Co-{C}_N\times \alpha \right]\ \left(F/A\right) $$

with α = (0.21/0.79) [Molecular weight of nitrogen / Molecular weight of oxygen]^1/2.

where C _N × α represents the oxygen content through leakage and (0.21/0.79) stands for the oxygen to nitrogen fractional ratio in air. The value of “A” for the membrane disc was ∼0.608 cm². The leakage was usually less than 1 % in these experiments. The oxygen permeation flux can as well be expressed in units of mol cm⁻² s⁻¹, the conversion relation being 1 ml cm⁻² min⁻¹ = 7.44 × 10⁻⁷ mol cm⁻² s⁻¹.

Results and discussion

Formation of SrCo_0.8Fe_0.2O_3−δ

Figure 2 shows the weight (W%) and dW(%)/dT versus temperature plots obtained by heating the sol-gel-derived oxalate powder in a thermogravimetric set-up at a rate of 4 °C min⁻¹ up to 900 °C in air. Three main stages of the weight loss are observed. In the first, a weight loss of ∼17.1 % occurs up to 315 °C and is associated with removal of water. The peaks around 170 and 268 °C appearing in the dW(%)/dT versus T curve correspond to absorbed and crystallizing water, respectively. In the second stage, a weight loss of ∼20.3 % results in the temperature range 315–340 °C and can be attributed to release of remaining crystallizing water and carbon monoxide (notice a sharp peak at 328 °C in the dW(%)/dT versus T plot). In the third stage, a weight loss of ∼16.6 % takes place in the temperature range 340–900 °C. This weight loss amounts to removal of remaining CO and CO₂. The CO reacts with the atmospheric oxygen and forms CO₂ before escaping. The decrease in magnitude of slope in the temperature ranges 460–644 °C and 644–877 °C is perhaps arising due to oxygen absorption and corresponding increase in weight. The decomposition is nearly complete at ∼900 °C with a total loss of about 54 %.

Fig. 2

Weight percent (W%) and d(W%)/dT versus temperature (T) plots of dried sol-gel product used for the synthesis of SrCo_0.8Fe_0.2O_3−δ

In accordance with thermogravimetry data, the dried sol-gel oxalate product was subjected to decomposition at 950 °C or above for 5 h in air to obtain a stable oxide. The X-ray diffraction (XRD) patterns of resulting products after decomposition of oxalate powders at 950, 1100, and 1150 °C are shown in Fig. 3. Notice that the diffraction pattern depicted in Fig. 3c corresponds to a single phase and can be indexed with a perovskite-type cubic structure of SrCo_0.8Fe_0.2O_3−δ with the lattice parameter a = 3.8650 ± 0.0002 Ǻ. Other patterns (Fig 3a, b) contain extra peaks marked by asterisk (*), indicating the presence of secondary phase(s). The 2θs, relative intensities, and hkl values of various diffraction peaks of SrCo_0.8Fe_0.2O_3−δ (obtained by decomposition of oxalate at 1150 °C for 5 h in air) are given in Table 2 together with the known data of Sr(Co_0.81Fe_0.19)O_2.78, which exhibits a perovskite-type cubic structure with lattice parameter a ∼ 3.849 Ǻ, Z = 1, space group $Pm\overset{-}{3}m$, and density (ρ) = 5.545 g cm⁻³ (JCPDS File no. 82-2445). Rietveld analysis undertaken by placing strontium at the corner (A site), cobalt/iron at the centre (“B” site), and oxygen ions at the face-centered positions of the perovskite-type ABO₃ cubic unit cell with space group $Pm\overset{-}{3}m$ gives a diffraction profile matching well with the observed pattern (Fig. 3d). The fitting parameters of profile being weighted R-factor (R _wp) = 8.1 %, R-factor (R _p) = 6.5 % and goodness χ² = 2.7 for the case δ = 0.2. The average crystallite size (D) as deduced from the Scherrer formula [D = 0.9 λ / (β cos θ), where λ is the wavelength of X-rays, β is the corrected full width at half maximum (FWHM) and θ is the Bragg angle], taking silicon as standard and the 110 diffraction peak data, is ∼100 ± 2 nm. The above results suggest the formation of a perovskite-type cubic phase of SrCo_0.8Fe_0.2O_3−δ by decomposition of the sol-gel derived oxalate at 1150 °C for 5 h in air. Its morphology shows clusters/agglomerates of small crystallites in varying size and shape (Fig. 4).

Fig. 3

X-ray diffraction patterns of SrCo_0.8Fe_0.2O_3−δ powder produced by decomposition of sol-gel-derived oxalate at a 950, b 1100, and c 1150 °C for 5 h in air, and d diffraction profile obtained by Rietveld analysis on the basis of a perovskite-type cubic structure with space group Pm3m. Asterisk (*) represent peaks of secondary phase(s)

Table 2 The 2θs, relative intensities, and hkl values of various diffraction peaks for SrCo_0.8Fe_0.2O_3-δ and Sr (Co_0.81 Fe_0.19)O_2.78 systems

Fig. 4

Scanning electron micrographs (SE mode) of SrCo_0.8Fe_0.2O_3−δ powder by decomposition of sol-gel-derived oxalate at 1150 °C for 5 h in air showing clusters/agglomerates of small crystallites at a low and b high magnifications

The oxygen deficiency parameter δ in SrCo_0.8Fe_0.2O_3−δ has been estimated by invoking the principle of charge balance, assuming the oxidation states of iron and cobalt as 3+ and/or 4+, and varying their amounts using the relation δ = (1–x–y)/2, where x = (0–0.8) and y = (0–0.2) denote Co⁴⁺ and Fe⁴⁺ content, respectively in structural formula Sr²⁺Co³⁺ _0.8–x Co⁴⁺ _x Fe³⁺ _0.2–y Fe⁴⁺ _y O²⁻ _3−δ. In ideal case, δ is zero where A and B sites in perovskite cubic structural formula have species of oxidation states 2+ and 4+, respectively. However, if B species display both 3+ and 4+ valences, δ becomes non-zero, indicating the presence of oxygen vacancies. The value of δ in that case lies in between zero and 0.50.

Phase changes with temperature

A few typical XRD patterns of SrCo_0.8Fe_0.2O_3−δ, synthesized by decomposition of the sol-gel-derived oxalate at 1150 °C for 5 h in air, recorded in the temperature range 27–1000 °C are shown in Fig. 5. Notice that the pattern at 27 °C can be indexed on the basis of perovskite-type cubic unit cell with lattice parameter a = 3.8650 Å (as discussed above). With increase in temperature, extra diffraction lines appear and splitting occurs as well of some peaks. Above 200 °C, lines marked with asterisk (*) indicate emergence of Co₃O₄ (JCPDF no. 80-1544) as a secondary phase. An evidence for the presence of Co₃O₄ in SrCoO_3−δ as secondary phase has been reported earlier, too [22–24]. The system follows a reversible reaction [18 SrCoO_3−δ ↔3Sr₆Co₅O₁₅ + Co₃O₄ ;δ = 5/18] with a cobalt-deficient Sr₆Co₅O₁₅ phase exhibiting a rhomboderal structure with Z = 3 and space group R32; lattice parameters on hexagonal axes being a = 9.5035(2) Å, c = 12.3966(4) Å. In the present case, the composition is different (SrCo_0.8Fe_0.2O_3−δ) but with Co₃O₄ precipitation, the primary cobalt deficient phase now corresponds to an orthorhombic structure (signature being splitting of cubic 110 peak); lattice parameters being a _o ∼ a√2, b _o ∼ 4a and c _o ∼ a√2, where “a” is the size of the SrCo_0.8Fe_0.2O_3−δ basic cubic unit cell [15, 25, 26]. This phase matches well with the brownmillerite (Ca₂AlFeO₅)-type orthorhombic structure having lattice parameters a _o = 5.5981 (3) Å, b _o = 15.7832(8) Å, c _o = 5.4692(3) Å, Z = 4, space group Ibmm (JCPDF no. 82-2446) and supposedly displays anionic vacancy ordering. Accordingly, A²⁺B³⁺O_2.5 (A = Sr²⁺, B = Co³⁺, Fe³⁺) contains BO₆ octahedra and BO₄ tetrahedra in alternating layers along the b-axis [22]. The split in the cubic 110 peak is indicative of orthorhombic phase formation with oxygen vacancy ordering, as reported earlier in SrCo_0.4Fe_0.6O_3−δ produced via solid state reaction method [27]. XRD patterns in the intermediate temperature range (300–500 °C) are quite complex with several phases co-existing but included here for completeness. Nevertheless, XRD pattern (Fig. 5) at 500 °C matches well with reported results of Prado et al. [28] for the brownmillerite phase. Efforts are underway to produce a single cobalt-deficient brownmillerite orthorhombic phase for detail and meaningful structure analysis. Needless to say, the product usefulness is only at elevated temperatures as oxygen permeable membrane. Above 800 °C, the orthorhombic phase and Co₃O₄ recombine gradually to yield back a stable perovskite-type cubic phase with a = 3.9937 ± 0.0002 Å at 1000 °C [15, 20, 22, 27]. The increase in lattice parameter can be attributed to apparent thermal expansion of the cubic phase observed at 27 °C. These observations are consistent with those in Sr_0.9La_0.1Co_0.8Fe_0.2O_3−δ system where the brownmillerite (Ca₂FeAlO₅-type) orthorhombic phase transforms to a perovskite-type cubic at elevated temperatures [28]. Both the cubic and orthorhombic phases are usually present at the intermediate temperatures.

Fig. 5

XRD patterns of SrCo_0.8Fe_0.2O_3-δ (SCFO) recorded at different temperatures in the range of 27° – 1000°C

With oxygen desorption as molecules above 800 °C, the associated electrons release facilitates in part B⁴⁺ → B³⁺ conversion of Co and/or Fe species which, in turn, causes stretching of B–O bond. This fact is perhaps responsible for both the orthorhombic–cubic transformation and disordering of oxygen vacancies. The splitting of diffraction peak at 2θ ≈ 66.48° at 1000 °C represents the resolved CuKα components (α1 and α₂) of 220 reflection. The diffraction patterns reveal shift, splitting, and merger of peaks besides emergence of Co₃O₄ at different temperatures. These observations amply demonstrate the undergoing changes and can be summarized as

$$ \begin{array}{lll}\begin{array}{l}{\mathrm{SrCo}}_{0.8}{\mathrm{Fe}}_{0.2}{\mathrm{O}}_{3-\delta}\\ {}\mathrm{Perovskite}\ \hbox{-}\ \mathrm{type}\ \mathrm{cubic}\ \mathrm{phase}\end{array}\hfill & \overset{200-800{}^{\circ}\mathrm{C}}{\to}\hfill & \begin{array}{l}{\mathrm{SrCo}}_{0.8\hbox{--} 3\xi }{\mathrm{Fe}}_{0.2}{\mathrm{O}}_{3\hbox{--} \delta \hbox{--} 4\xi } + \xi\ {\mathrm{Co}}_3{\mathrm{O}}_4\\ {}\mathrm{Orthorhombic}\ \mathrm{phase}\ \mathrm{with}\ \mathrm{ordered}\kern0.5em \mathrm{oxygen}\ \mathrm{vacancies}\end{array}\hfill \\ {}\begin{array}{l}{\mathrm{SrCo}}_{0.8\hbox{--} 3\xi }{\mathrm{Fe}}_{0.2}{\mathrm{O}}_{3\hbox{--} \delta \hbox{--} 4\xi } + \xi\ {\mathrm{Co}}_3{\mathrm{O}}_4\\ {}\mathrm{Orthorhombic}\ \mathrm{phase}\ \mathrm{with}\kern0.5em \mathrm{ordered}\ \mathrm{oxygen}\ \mathrm{vacancies}\end{array}\hfill & \overset{ > 800\hbox{--} 1000{}^{\circ}\mathrm{C}}{\to}\hfill & \begin{array}{l}{\mathrm{SrCo}}_{0.8}{\mathrm{Fe}}_{0.2}{\mathrm{O}}_{3-\delta}\\ {}\mathrm{Perovskite}\hbox{-} \mathrm{type}\ \mathrm{cubic}\ \mathrm{phase}\kern0.5em \mathrm{disordered}\ \mathrm{oxygen}\ \mathrm{vacancies}\end{array}\hfill \end{array} $$

Since the brownmillerite orthorhombic structure contains ordered distribution of anion vacancies, movement of oxygen ions becomes somewhat restricted and leads to poor permeability [26]. The structural change above 800 °C involves fusion of orthorhombic phase with Co₃O₄ and disordering of oxygen vacancies. Such an arrangement combined with lattice expansion is expected to facilitate oxygen permeation in SrCo_0.8Fe_0.2O_3−δ cubic phase emerging above 800 °C. The difference in the peak intensity observed in XRD patterns of SrCo_0.8Fe_0.2O_3−δ sample at 1000 °C and room temperature is quite significant. The tentative reason could be the random distribution of oxygen vacancies at 1000 °C. Nevertheless, the difference is not understood as yet and is being further investigated currently.

X-ray photoelectron spectroscopy studies

Figure 6a shows the XPS survey spectrum of SrCo_0.8Fe_0.2O_3−δ, obtained with a PHI 5000 Versa Probe photoelectron spectrometer (model ULVAC) using AlKα radiation (1486.6 eV) under vacuum of better than 6.7 × 10⁻⁸ Pa at low resolution (step 1 eV) in the energy range 0–900 eV. The peaks identified belong to Sr 3d, C 1s + Sr 3p _1/2, Co 2p, Fe 2p and O 1s levels. The carbon 1s peak at 284.6 eV is treated as the standard for calibration. The corresponding fine spectra observed at high resolution (step 125 meV) in five different energy regimes for strontium and carbon are presented in Fig. 6b, c (Table 3). The de-convoluted Gaussian peaks fitting carried out with Origin 8.5 software for cobalt, iron and oxygen species are shown in Fig. 7. The values of binding energy deduced from Figs. 6 and 7 for different elements together with those reported in literature are summarized in Table 3. The nature of various species can now be described:

Fig. 6

a XPS survey spectra with portion corresponding to b Sr 3d and c C 1s + Sr 3p in SrCo_0.8Fe_0.2O_3−δ at high resolution

Table 3 Binding energies of various elements as deduced from XPS data and reported in literature for SrCo_0.8Fe_0.2O_3-δ system

Fig. 7

XPS high-resolution spectra showing Gaussian fitting of peaks for a Co 2p, b Fe 2p, and c O 1s in SrCo_0.8Fe_0.2O_3−δ

Sr 3d and Sr 3p

For analyzing strontium, its 3d and 3p peaks (Fig. 6b, c) recorded in the energy range 129–154 eV and 279–297 eV, respectively, were used. The binding energy of Sr 3d doublet corresponds to 133.2 and 134.8 eV (Fig. 6b) whereas that of Sr 3p _1/2 is 279.3 eV (Fig. 6c) in SrCo_0.8Fe_0.2O_3−δ sample. In addition, two peaks clearly visible at 284.6 and 289.4 eV belongs to C 1s. These results are consistent with the reported binding energy data of Sr 3d, Sr 3p, and C 1s (Table 3).

Co 2p

A high-resolution XPS scan of SrCo_0.8Fe_0.2O_3−δ in the energy range of 771–816 eV shows two peaks of Co 2p _3/2 and Co 2p _1/2 (Fig. 7a). Dupin et al. [33] have discussed the XPS spectra of cobalt in Li_0.66CoO₂ and LiCoO₂ systems in detail. Accordingly, Co 2p peaks appearing in the energy range 775–800 eV are actually built with signals from Co (3+) and (4+) states, i.e., a doublet with energy 779.8 and 795 eV corresponds to Co³⁺ ions and the other at 781.5 and 796.8 eV belongs to Co⁴⁺ ions. In the present case, XPS spectrum shown in Fig. 7a is further analyzed and peaks deconvoluted to obtain relative amounts of Co³⁺ and Co⁴⁺ species present. This exercise reveals two doublets with binding energies (779.7 and 795.1 eV) for Co³⁺ ions and (781.4 and 796.8 eV) for Co⁴⁺ ions. These energy values are in good agreement with those reported by Dupin et al. [33]. The peak position, FWHM, % peak area, oxidation state, and the energy difference between the Co 2p _3/2 peaks are summarized in Table 4. The relative amounts of Co³⁺ and Co⁴⁺ ions have been estimated in SrCo_0.8Fe_0.2O_3−δ system by considering the respective peak areas as 55.9 and 44.1 %, respectively.

Table 4 XPS data and oxidation states of elements in SrCo_0.8Fe_0.2O_3-δ

Fe 2p

High-resolution XPS scan of SrCo_0.8Fe_0.2O_3−δ depicted in Fig. 7b in the energy range 701–741 eV reveals information about the oxidation states of iron. The peak broadness suggests the presence of all Fe³⁺, Fe²⁺, and Fe⁴⁺ ions but arises due to electrostatic interactions, spin-orbit coupling between the 2p core hole and unpaired 3d electrons of the photo-ionized cation, and crystal field effects [34, 35]. Yunfei et al. [32] have classified peaks for Fe 2p _3/2 in the energy range of 709.6–709.9 eV, 710.9–711.9 eV, and 712–714 eV as belonging to Fe²⁺, Fe³⁺, and Fe⁴⁺ species, respectively. The Gaussian fitting of Fe 2p signal corresponds to two peaks with binding energy 710.9 and 714.2 eV and so can be attributed to Fe³⁺ and Fe⁴⁺ ions, respectively (Fig. 7b and Table 3). The peak position, FWHM, % peak area, oxidation state, and the energy difference between the Fe 2p _3/2 peaks are summarized in Table 4. Note that the Fe³⁺ ions present in SrCo_0.8Fe_0.2O_3−δ system amounts to just 8.5 % while the Fe⁴⁺ species share the remaining 91.5 %.

O 1s

The high-resolution XPS spectrum in the energy range 525–543 eV depicted in Fig. 7c gives the signatures of oxygen ions. The Gaussian fitting shows peak split with binding energies 529.5 and 531.3 eV (Fig 7c). Machocki et al. [31] identified two peaks of oxygen in the XPS spectrum of manganese-lanthanum oxides at energy in the range (i) 529.4–529.8 eV for O²⁻ ions and (ii) 531.2–531.8 eV for surface adsorbed O₂ ²⁻, O₂ ⁻, or O⁻ species. Thus, peaks of O 1s observed at 529.5 and 531.3 eV can be assigned to oxygen ion (O²⁻) and surface adsorbed oxygen, respectively. The peak position, FWHM, % peak area, oxidation state, and the energy difference between the peaks for O 1s in SrCo_0.8Fe_0.2O_3−δ are summarized in Table 4. These indicate that the surface-adsorbed O₂ ²⁻, O₂ ⁻, or O⁻ species amount to 91.4 % whereas the bulk O²⁻ ion is just 8.6 %.

Considering the relative amounts of Co³⁺, Co⁴⁺, Fe³⁺, and Fe⁴⁺ present and invoking the concept of charge neutrality, oxygen deficiency parameter (δ) is deduced as 0.23 for SrCo_0.8Fe_0.2O_3−δ system at room temperature. Although the figure corresponds to sample surface, yet, it is useful for understanding the phenomenon occurring in bulk at high temperatures (“Electrical conductivity measurements” and “Oxygen permeation characteristics” sections).

Electrical conductivity measurements

The electrical conductivity (σ) of various disc membranes (size ∼10 mm diameter, thickness ∼1.5/2.0 mm) measured with an LCZ meter (HP 4194A) in the temperature range 27–700 °C at 1 kHz are shown in Fig. 8. The general trend is that σ increases initially up to a characteristic temperature (T _c) and then decreases gradually; the maximum value being ∼2.50 S cm⁻¹ at T _c = 280 °C. The σ is expressed as σ = neμ, where n is electron concentration, e is electronic charge and μ is carrier mobility. The electrical conductivity can therefore be improved by increasing “n” or “μ” or both. The initial increase of σ up to T _c may be attributed to gradual rise in the carrier mobility (μ) with thermal energy uptake and/or structural changes. With rise in temperature, desorption of oxygen molecules is accompanied by release of electrons and, in turn, increases n. Some of these electrons cause Co⁴⁺ → Co³⁺ and Fe⁴⁺ → Fe³⁺ conversion and/or form color centers at anion vacancy sites while others remain free. As a consequence, the electron interaction among themselves becomes progressively stronger and carrier scattering by defects significant–leading to decrease in mobility. At a certain temperature above T _c, the mobility (μ) reduction occurs such that the gain caused in electron density is overcome, leading to net decrease in (σ). Thus, the loss of oxygen and the annihilation of electron holes are major causes for the decrease in the electrical conductivity at high temperatures. The phase changes, if occur with temperature, can also contribute to variation in σ appropriately.

Fig. 8

Electrical conductivity (σ) versus temperature plot of SrCo_0.80Fe_0.20O_3−δ and corresponding ln (σT) versus 1000/T plot in the inset with straight lines drawn in four different temperature intervals

The electrical conductivity data reported in the literature for SCFO system vary significantly and lie in the range ∼0.5–500 S cm⁻¹ depending upon the system, composition, preparation method, temperature, oxygen partial pressure, etc. [36–42]. The reason for such a vast difference in σ is not understood as yet. The σ values observed here are on the lower end with the maximum 2.50 S cm⁻¹ being within the above range. The measurements were made earlier by a four-point probe with surface contacts and hence data correspond to lateral conduction of sample disc. On the other hand, σ values presented here refer to conduction through membrane thickness (i.e., relevant to oxygen permeation process) and hence two-probe method was the opted for measurements. The sintered membrane disc (10 mm diameter, 1.5 mm thick) was placed in a sample holder (comprised of silver electrodes and a chromel–alumel thermocouple) and (σ) data collected up to 700 °C using an LCZ meter (HP4194A) at 1 kHz. With the electrical resistivity of silver (Ag) being 1.63 × 10⁻⁶ (Ω-cm) the electrode resistance is of the order of 10⁻⁵ Ω as against large resistance (0.4 Ω) of SCFO disc. Hence, the electrode resistance in no way causes interference in measurements. The resistance of disc if deduced with the maximum value of σ (= 500 S cm⁻¹, being two orders of magnitude higher than observed in the present case) comes out to be 4 × 10⁻⁴ Ω. This value is still 40 times larger in magnitude than the resistance of the electrodes. Therefore, the σ data reported in the paper belong essentially to the SCFO disc. Moreover, the variation trends in electrical conductivity described above are consistent with the reported observations in (a) SrCo_1−x Fe _x O_3−δ (x = 0.2–0.6), LaSr₃Fe_3−x Co _x O_10−δ (x = 0–1.5), SrFe_1.5–x Co _x O _y (x = 0.15–0.5), and La_0.3Sr_0.7Fe_1−x Ga _x O_3−δ (x = 0–0.4), (b) La_1−x Sr _x Co_0.2Fe_0.8 O_3−δ (x = 0–0.4), (c) La_1−x M _x Co_1−y Fe _y O_3−δ(M = Sr, Ba, Ca; 0.4 ≤ x ≤ 0.8; 0.2 ≤ y ≤ 0.8) and (d) Ba_0.5Sr_0.5Zn_0.2Fe_0.8O_3−δ perovskite-type cubic compounds obtained by solid state reaction, liquid mix process, glycine-nitrate combustion, and sol-gel route, respectively [36–42].

The electrical conductivity (σ) arising due to polaron hopping in oxides is given by [19]

$$ \sigma =\left(A/T\right) \exp \left(-{E}_{\mathrm{a}}/{k}_{\mathrm{B}}T\right) $$

where A is the material constant, E _a is the activation energy of hopping, k _B (= 8.6 × 10⁻⁵ eV K⁻¹) is the Boltzmann constant and T is the absolute temperature. Accordingly, ln (σT) versus 1/T plot should be a straight line with slope giving the activation energy (E _a). The ln (σT) versus 1/T plot shown in inset of Fig. 8 for SrCo_0.8Fe_0.2O_3−δ sample is nonlinear (or curved) but can be fitted with four different straight lines whose slopes correspond to activation energies of hopping in between 0.03 and 0.07 eV in the temperature range 27–700 °C. This finding suggests gradual slowing down of the hopping process with rise in temperature. Incidentally, structural changes discussed in “Phase changes with temperature” section are well correlated with the above trends as the four regimes correspond to perovskite-type cubic structure (27–110 °C), Co₃O₄, and cobalt-deficient cubic phase (115–250 °C), non-stoichiometric orthorhombic phase with Co₃O₄ (255–480 °C), and an orthorhombic phase of different composition with less Co₃O₄ content (485–700 °C), respectively. Kharton et al. [43] have found the activation energy for conduction in SrCo_0.8Fe_0.2O_3−δ (synthesized by solid state reaction) somewhat higher (i.e., 0.110 ± 0.002 eV in temperature range 27–247 °C) possibly due to presence of inherent compositional inhomogeneity and micro-structural differences in the membrane. Stevenson et al. [42] have reported activation energy in two ranges 0.03–0.11 eV and 0.16–0.29 eV in La_1−x M _x Co_1−y Fe _y O_3−δ (M = Sr, Ba, Ca; 0.4 ≤ x ≤ 0.8; 0.2 ≤ y ≤ 0.8) compounds, produced by glycine-nitrate combustion technique. These values may be attributed to compositional differences and/or secondary phases present. Their ln (σT) versus 1/T plots also show nonlinearity as described above possibly due to variation in crystal structure with temperature. This aspect is perhaps missed out because of non-availability of the crystal data at various temperatures. On the contrary, XRD analysis carried out in the present study clearly relates (i) origin of nonlinearity of ln (σT) versus 1/T plot and (ii) variation of conductivity to crystalline phases in SrCo_0.8Fe_0.2O_3−δ at different temperatures within 27–700 °C. The hopping of charge carriers follows Zener double exchange and involves the passage built-up by the overlap of transition metal cation and oxygen orbitals, represented by [44, 45]:

$$ {\mathrm{B}}^{n+}{\textstyle \hbox{-} }{\mathrm{O}}^{2-}{\textstyle \hbox{-} }{\mathrm{B}}^{\left(n+1\right)+}\kern0.5em \overrightarrow{{\mathrm{O}}^{2-}\to {\mathrm{O}}^{1-}+e}\kern0.5em {\mathrm{B}}^{n+}{\textstyle \hbox{-} }{\mathrm{O}}^{1-}{\textstyle \hbox{-} }{\mathrm{B}}^{n+}\kern0.5em \overrightarrow{{\mathrm{O}}^{1-}+e\to {\mathrm{O}}^{2-}}\kern0.5em {\mathrm{B}}^{\left(n+1\right)+}{\textstyle \hbox{-} }{\mathrm{O}}^{2-}{\textstyle \hbox{-} }{\mathrm{B}}^{n+} $$

In this transfer chain, the B–O bond length, oxygen vacancy and the B⁽ⁿ⁺¹⁾⁺/Bⁿ⁺ couple influence the carrier conduction. In the transition metal-based perovskite-type oxides having B⁴⁺ and B³⁺ ions at B-site, the fraction [B⁴⁺]/([B³⁺] + [B⁴⁺]) determines the available hopping sites for polaron [40]. XPS analysis showed the fractions of total 3+ ions (i.e., Co³⁺ and Fe³⁺) and 4+ ions (i.e., Co⁴⁺ and Fe⁴⁺) on the SrCo_0.8Fe_0.2O_3−δ surface at room temperature as 0.464 and 0.536, respectively. The higher fraction of B⁴⁺ ions may lead to formation of B³⁺ ions by capturing electrons released during oxygen desorption at elevated temperatures. As a consequence, B⁴⁺ ions and hopping sites both decrease and cause lowering in electrical conductivity [19]. Further, significant overlap of the transition metal 3d and oxygen 2p orbitals in linear disposition of B–O–B (B=Co, Fe) bond gives rise to desired improvement in the electronic transport as well [46].

Oxygen permeation characteristics

As described above, the oxygen permeation flux (J _O2) through the SrCo_0.8Fe_0.2O_3−δ membrane disc (diameter ∼ 10 mm and thickness ∼ 1.0–2.0 mm) was determined with a set-up developed in the laboratory in the temperature range 700–1000 °C using air as a feeder gas on the front side and helium as a carrier on the exit (or sweep) side. Figure 9a depicts the variation of oxygen permeation flux J _O2 with temperature for membranes of three thicknesses (1.0, 1.5, and 2.0 mm) at input air flow rate of 400 ml min⁻¹ and sweeping helium gas flow rate of 45 ml min⁻¹. It may be noticed that J _O2 invariably increases with temperature and decreases with increase in membrane thickness. The oxygen permeation flux J _O2 reaches a maximum value of ∼2.44 ml cm⁻²min⁻¹ (or 1.8 × 10⁻⁶ mol cm⁻² s⁻¹) at 1000 °C for 1.0-mm-thick disc. The oxygen permeation flux data of various membranes obtained at 700, 800, 900, and 1000 °C are presented in Table 5. The increase in permeation observed with rise in temperature can be attributed to (i) creation of additional anion vacancies, (ii) enhancement in surface reactions, and (iii) higher oxygen mobility. The ln (J _O2) vs. 1000/T plots shown in Fig. 9b are nearly linear in nature and follow the Arrhenius equation J _O2 = J _O exp. (− E _a/k _B T), where J _O is the pre-exponential factor, E _a is the activation energy of oxygen permeation and k _B is the Boltzmann constant. The activation energy values deduced from the slopes of straight lines in the temperature range 800–1000 °C are 50.0, 52.4, and 56.2 kJ mol⁻¹ (or 0.52, 0.54, and 0.58 eV/ion) for membranes of thickness 1.0, 1.5, and 2.0 mm, respectively. The increase in activation energy with the membrane thickness is in correspondence with the observed decrease in the oxygen permeation flux (Table 5).

Fig. 9

a Oxygen permeation flux (Jo₂) versus temperature and b ln (J _O2) vs. 1000/T plots of SrCo_0.80Fe_0.20O_3−δ membranes (diameter ∼ 10 mm) having thickness 1.0, 1.5, and 2.0 mm

Table 5 Oxygen permeation flux (J _O2) at different temperatures of SrCo_0.8Fe_0.2O_3-δ membranes having thickness 1.0, 1.5, and 2.0 mm (air flow rate = 400 ml min⁻¹, He flow rate = 45 ml min⁻¹)

The processes occurring at front and exit surfaces and within the membrane disc are illustrated schematically in Fig. 10. The driving force for oxygen permeation is the partial pressure gradient prevailing along the membrane thickness. The three main steps involved in oxygen permeation are the following:

• (a) Selective adsorption of oxygen molecules on the front surface (at high pressure) and their dissociation into ions by capturing electrons following the reaction O₂ + 4e⁻ → 2O²⁻

• (b) Migration/diffusion of oxygen ions through the bulk membrane via anion vacancies to the exit (sweeping) side, and

• (c) Conversion of oxygen ions into molecules by leaving behind the electrons before desorbing at the exit surface; the reaction being 2O²⁻ → O₂ + 4e⁻

• (d) Movement of electrons released at the exit surface back to the front side due to concentration gradient alone and/or via B⁴⁺ → B³⁺ conversion before hopping following the scheme

Fig. 10

Schematic diagram illustrating processes occurring at front and exit surfaces, oxygen transport, and electron movement in a typical SrCo_0.8Fe_0.2O_3−δ membrane

$$ \begin{array}{cc}{\mathrm{B}}^{3+}{\textstyle \hbox{-} }{\mathrm{O}}^{2-}{\textstyle \hbox{-} }{\mathrm{B}}^{4+}\to {\mathrm{B}}^{4+}{\textstyle \hbox{-} }{\mathrm{O}}^{2-}{\textstyle \hbox{-} }{\mathrm{B}}^{3+}\kern1em & \kern1em \left(\mathrm{B}=Co,Fe\right)\kern1em \end{array} $$

With increase in temperature, Jo₂ increases due to cumulative effects of (i) formation of additional anion vacancies, (ii) improvement in dissociation of oxygen molecules, and (iii) easy migration of O²⁻ ions and electrons in opposite directions within the membrane.

Earlier, SrCo_0.8Fe_0.2O_3−δ membranes were realized with products derived via liquid citrate and solid state reaction methods. The oxygen permeability flux (Jo₂) of 1-mm-thick membrane at 850 °C was found to be 1.8 × 10⁻⁷ or 6.3 × 10⁻⁷ mol cm⁻² s⁻¹ [18, 20]. However, Teraoka et al. [17] reported (Jo₂) an order higher (i.e., 2.31 × 10⁻⁶ mol cm⁻² s⁻¹) in case of SrCo_0.8Fe_0.2O_3−δ compound synthesized by solid state reaction under similar experimental conditions. The reason for such a huge discrepancy is not apparent and remains unanswered so far. But significant variation in Jo₂ is also observed even in identical membranes [47]. The Jo₂ in SrCo_0.8Fe_0.2O_3−δ membrane (thickness = 1.0 mm), produced with sol-gel-derived powder in the present study, is slightly higher (i.e., 9.9 × 10⁻⁷ mol cm⁻² s⁻¹ or 1.33 ml cm⁻² min⁻¹) at 850 °C (with input air flow rate of 400 ml/min and sweeping helium flow rate of 45 ml min⁻¹) than those reported by Kruidhof et al. [20] and Qiu et al. [18]. The improvement in Jo₂ can be associated with the sol-gel route employed for material preparation which ensures homogeneous mixing of precursor species at atomic/molecular level and provides better stoichiometry control leading to product of high purity. Further, it gives nanosize crystallites and, in turn, compact membrane with high mechanical stability.

Membrane stability

In order to ascertain the structural stability of SrCo_0.8Fe_0.2O_3 ‐ δ membrane, XRD patterns of both the front and exit surfaces were recorded and analyzed after completion of 3 cycles of oxygen permeation measurements at 700, 800, 900, and 1000 °C in succession. Figure 11 depicts the XRD patterns along with that of the original membrane (thickness 1.0 mm) for the sake of comparison. Note that the perovskite-type cubic phase continues to prevail on both sides even after three oxygen permeation cycles covering a total period of about 24 h in the temperature range of 700–1000 °C. Obviously, the SrCo_0.8Fe_0.2O_3 ‐ δ disc remains structurally stable throughout the oxygen permeation process. The scanning electron micrographs of the membrane surface depicted in Fig. 12 however indicate visible microstructural changes (in terms of grain growth on both the front and exit surfaces) after the permeation measurements. Also, the front surface is affected more than the exit one.

Fig. 11

X-ray diffraction patterns of SrCo_0.8Fe_0.2O_3−δ membrane a before and b, c after oxygen permeation process

Fig. 12

Scanning electron micrographs (SE mode) of SrCo_0.8Fe_0.2O_3−δ membrane: a as such, b front, and c exit surfaces after the oxygen permeation process

The membrane is further expected to display high oxygen permeability with long-term stability in order to become suitable for practical applications. To check the stability of SrCo_0.8Fe_0.2O_3−δ membrane, oxygen permeation flux (J _O2) was also monitored at 900 °C for 120 h continuously by maintaining flow rates of input air at front surface and carrier helium gas at exit side as 400 ml min⁻¹ and 45 ml min⁻¹, respectively. Figure 13 depicts the oxygen permeation flux (J _O2) for membrane of thickness 1.5 mm as ∼1.41 ml cm⁻² min⁻¹ for about ∼60 h only. Jo₂ decreases thereafter slowly by ∼7 % in next 60 h. Nevertheless, Jo₂ value is greater than 1.0 ml cm⁻² min⁻¹ set for economic and industrial viability consideration [34]. The oxygen permeability of SrCo_0.8Fe_0.2O_3−δ membrane no doubt lies within the acceptable limit but its sustainability is somewhat poor and requires improvement. Figure 14 depicts the XRD pattern of SrCo_0.8Fe_0.2−y Zr _y O_3−δ (y = 0.05) sample obtained by decomposition of the dried sol-gel at 1150 °C for 5 h in air. This corresponds to a perovskite-type cubic phase with a = 3.8690 ± 0.0002 Å. It may be emphasized that the membrane having slightly different composition SrCo_0.8Fe_0.2−y Zr _y O_3−δ (y = 0.05) prepared in the same manner exhibits better stability and higher oxygen permeability as well (Fig. 13). Previous studies also reveal improvement in phase stability with addition of fixed oxidation state elements in the perovskite structure but at the cost of oxygen permeability [19, 27, 48, 49]. This happens with zirconium incorporation at B-site in SrCo_0.8Fe_0.2–y Zr _y O_3−δ system as well for y = 0.075 and 0.1 because of SrZrO₃ presence perhaps as a secondary phase. But the membrane of a particular composition (y = 0.05) somehow shows improvement both in the cubic phase stability and oxygen permeability. The increase in Jo₂ as mentioned above is observed experimentally for the composition (y = 0.05) and taken as an example from the systematic study undertaken on SrCo_0.8Fe_0.2–y Zr _y O_3−δ (y = 0–0.100) membranes [50]. The findings has been attributed to (i) improved stabilization of perovskite-type cubic structure, (ii) increase in lattice parameter because of higher ionic radius of Zr⁴⁺ ion vis-à-vis Fe/Co species, and (iii) enhancement of surface reactions due to availability of the additional electrons there. It may be mentioned that the stability of perovskite-type ABO_3−δ cubic structure is realized earlier too by partial replacement of species at B-site with a variety of elements in different systems, e.g., La_0.2Sr_0.8Fe_0.8B_0.2O_3−δ (B = Al, Ga, Cr, Ti, Ta, Nb), SrNb_0.1Co_0.9–x B _x O_3−δ (x = 0–0.5, B = Fe), BaCo_0.7Fe_0.2B_0.1 O_3−δ (B = Sn) and SrCo_0.6Fe_0.4–x B _x O_3−δ (x = 0–0.15, B = Sn) [51–54].

Fig. 13

Oxygen permeation flux (J _O2) as a function of time for a SrCo_0.80Fe_0.20O_3−δ and b SrCo_0.80Fe_0.20−y Zr_yO_3−δ (y = 0.05) membranes (thickness = 1.5 mm) at 900 °C (1 ml cm⁻² min⁻¹ = 7.44 × 10⁻⁷ mol cm⁻² s⁻¹)

Fig. 14

X-ray diffraction patterns of SrCo_0.8Fe_0.2−y Zr _y O_3−δ (y = 0.050) obtained from dried sol-gel oxalate product after decomposition at 1150 °C for 5 h in air

Conclusions

From the present study, the following conclusions can be drawn:

• 1) The perovskite-type cubic phase of SrCo_0.8Fe_0.2O_3−δ (a = 3.8650 ± 0.0002 Å, Z = 1), produced by decomposition of sol-gel derived oxalate precursor at 1150 °C for 5 h in air, remains stable at elevated temperatures (800–1000 °C) and exhibits high oxygen permeability (e.g., 1.41 ml cm⁻² min⁻¹ at 900 °C for membrane of thickness 1.5 mm) but lacks long-term stability.

• 2) Cobalt and iron exist in both 3+ and 4+ oxidation states and determine not only the oxygen content but also the number of electrons available for the reaction O₂ + 4e⁻ → 2O²⁻ to occur at the front surface of the membrane.

• 3) The perovskite-type cubic phase with abundance of anion vacancies is crucial for high oxygen permeability and extended operational stability of SCFO membranes. This can be achieved by partial substitution of iron with zirconium. Optimum composition SrCo_0.8Fe_0.2–y Zr _y O_3−δ (y = 0.05) with J _O2 ∼ 2.0 ml cm⁻² min⁻¹ at 900 °C (membrane thickness 1.5 mm) for long duration continuously shows promise for oxygen permeable membrane application.