Spin-state transition of iron in (${\mathrm{Ba}}_{0.5}{\mathrm{Sr}}_{0.5})({\mathrm{Fe}}_{0.8}{\mathrm{Zn}}_{0.2}){\mathrm{O}}_{3-\delta }$ perovskite

Abstract

The redox behavior of iron during heating of a high-performance perovskite for ceramic oxygen separation membranes was studied by combined electron energy-loss (EELS, esp. ELNES) and Mössbauer spectroscopical in situ methods. At room temperature, the iron in $({\mathrm{Ba}}_{0.5}{\mathrm{Sr}}_{0.5})({\mathrm{Fe}}_{0.8}{\mathrm{Zn}}_{0.2}){\mathrm{O}}_{3-\delta }$ (BSFZ) is in a mixed valence state of 75% ${\mathrm{Fe}}^{4+}$ in the high-spin state and 25% ${\mathrm{Fe}}^{3+}$ predominantly in the low-spin state. When heated to $900{}^{\circ }\mathrm{C}$, a slight reduction of iron is observed that increases the quantity of ${\mathrm{Fe}}^{3+}$ species. However, the dominant occurrence is a gradual transition in the spin-state of trivalent iron from a mixed low-spin/high-spin to a pure high-spin configuration. In addition, a remarkable amount of hybridization is found in the Fe–O bonds that are highly polar rather than purely ionic. The coupled valence/spin-state transition correlates with anomalies in thermogravimetry and thermal expansion behavior observed by X-ray diffraction and dilatometry, respectively. Since the effective cationic radii depend not only on the valence but also on the spin-state, both have to be considered when estimating under which conditions a cubic perovskite will tolerate specific cations. It is concluded that an excellent phase stability of perovskite-based membrane materials demands a tailoring, which enables pure high-spin states under operational conditions, even if mixed valence states are present. The low spin-state transition temperature of BSFZ provides that all iron species are in a pure high-spin configuration already above ca. $500{}^{\circ }\mathrm{C}$ making this ceramic highly attractive for intermediate temperature applications ($500–800{}^{\circ }\mathrm{C}$).

Introduction

Ceramic membranes can provide remarkable oxygen permeation fluxes at infinite selectivity without the need of external electrodes when they are based on heavily doped anion deficient cubic perovskite-type $A_{1-x}{A}_x^{\prime }{B}_{1-y}{B}_y^{\prime }{\mathrm{O}}_{3-\delta }$ oxides exhibiting mixed oxygen-ion and electron conductivities at elevated temperatures ($500–1000{}^{\circ }\mathrm{C}$) [1]. While the partial conductivity of electrons is distinctly higher than that of oxygen ions [1], [2], the oxygen deficit $\delta$ correlates directly with the obtainable permeation flux via the concentration of disordered oxygen vacancies. The perovskite lattice can tolerate a remarkable number of vacant oxygen sites if a reducible transition metal cation is located at the crystallographic B site in the center of the $B{\mathrm{O}}_6$ octahedron. Upon heating under constant pressure, entropy may free up even more oxygen from a previously equilibrated perovskite. The quantity of released oxygen may be quite high as the requirement of charge neutrality is no longer preserved solely by small concentrations of cation vacancies but by the possibly flexible redox behavior of the B site cation(s). However, one requirement for the oxygen vacancies to be mobile is to preserve a cubic perovskite structure [3], where the redox behavior of the B site cation(s) plays the key role as well. Hence, a better understanding of the redox behavior can be used to tailor improved membrane materials with excellent phase stability under strongly reducing conditions.

The pioneering work of Teraoka et al. in the 1980s [4], [5], [6], [7] exceeded the already high standards for obtainable oxygen fluxes. As a consequence, the search for high-flux materials (i.e., exhibiting oxygen permeation fluxes of one to two orders of magnitude higher than cubically stabilized zirconia equipped with shortcircuit external electrodes [4], [8]) focused until today mostly on complex perovskites hosting cobalt on their crystallographic B site. The current state-of-the-art material with respect to oxygen permeation and phase stability above $900{}^{\circ }\mathrm{C}$ is $({\mathrm{Ba}}_{0.5}{\mathrm{Sr}}_{0.5})({\mathrm{Co}}_{0.8}{\mathrm{Fe}}_{0.2}){\mathrm{O}}_{3-\delta }$ (BSCF) [3], [9], [10], [11]. Recently, in situ high-temperature electron energy-loss spectroscopy (HT-EELS) on BSCF has shown that cobalt is reduced from an average formal oxidation state of $2.6+$ to $2.2+$ and iron from $3.0+$ to $2.8+$, if BSCF is heated in the vacuum chamber of a transmission electron microscope (TEM) from room temperature to $950{}^{\circ }\mathrm{C}$ [12]. The average valence of the B site cations, $2.7+$ at room temperature and $2.3+$ at $950{}^{\circ }\mathrm{C}$, give oxygen contents $3-\delta$ of 2.3 and 2.2, respectively. That is in good agreement with oxygen stoichiometries estimated by thermogravimetric analysis (TGA) and neutron powder diffraction [13]. The experiments in [12] however, give direct proof of the easier reducibility of cobalt over iron in a highly doped perovskite-type oxide. Also, comparative TGA of different perovskite-type materials indicate that during heating the polyvalent B-site cobalt ions are reduced far more easily than iron, manganese or nickel [14].

The flexible redox behavior of cobalt provides on the one hand high oxygen fluxes at high temperatures. On the other hand it leads to two major problems hindering the reliable use of the BSCF material (and cobaltites in general) under important technical requirements. First, it causes a large coefficient of thermal expansion (CTE) that can lie in the range of $20–24\times {10}^6{\mathrm{K}}^{-1}$ over a wide temperature range [13], [15]. The resulting dilatation causes huge thermal stresses and thus cracks form easily in the membranes, especially if operated at steep oxygen potential gradients. Steep gradients can be obtained by making membranes very thin (i.e., $<200\mathrm{\mu }\mathrm{m}$) [15], [16], [17]. This is of interest to increase flux densities for the design of compact membrane units. Second, the valence instability of cobalt introduces inherent phase instability to the cobaltites at intermediate temperatures (ITs, ca. $500–800{}^{\circ }\mathrm{C}$) [9], [18], [19]. This is due to a coupled valence/spin-state transition of cobalt, and cobaltites containing ${\mathrm{CoO}}_6$ octahedra with ${\mathrm{Co}}^{3+}$ in low-spin configuration tend to prefer face sharing (contributions of hexagonal stacking) rather than corner sharing (cubic stacking) [19]. The breakdown of the cubic perovskite structure principally limits the long-time stability of the BSCF material under the conditions required for the operation of a membrane material in the IT range. The IT range, however, is of special interest for membrane-based dehydrogenation processes in the synthesis of basic chemicals like ethylene [20] and propylene [21] at high selectivity, and for the novel concept of solid oxide fuel cells (SOFCs) [22].

Recently, the search for alternative materials has led to the development of the cobalt-free perovskite-type oxide $({\mathrm{Ba}}_{0.5}{\mathrm{Sr}}_{0.5})({\mathrm{Fe}}_{0.8}{\mathrm{Zn}}_{0.2}){\mathrm{O}}_{3-\delta }$ (BSFZ) [23], [24]. Wang et al. [24] demonstrated its potential for use in a membrane reactor for the partial oxidation of methane (POM) at $900{}^{\circ }\mathrm{C}$. The reported oxygen permeation flux of $2.5\mathrm{ml}{\min }^{-1}{\mathrm{cm}}^{-2}$ on 1.25 mm-thick membranes at $900{}^{\circ }\mathrm{C}$ indicates a high conductivity as well as high surface exchange rates for oxygen. Wei et al. [25], [26] quantified the latter by measuring low polarization resistances of 0.22, 0.46, and $0.98\mathrm{\Omega }{\mathrm{cm}}^2$ at 700, 650, and $600{}^{\circ }\mathrm{C}$, respectively. They also used BSFZ as a cathode material in an IT-SOFC at $500–650{}^{\circ }\mathrm{C}$ at power densities up to $180\mathrm{mW}{\mathrm{cm}}^{-2}$. The room temperature structure of BSFZ has been refined in a cubic unit cell ($a=0.3990(0)\mathrm{nm}$) by the Rietveld method [27]. In situ X-ray diffraction (XRD) has shown that BSFZ remains in the cubic structure if heated to $900{}^{\circ }\mathrm{C}$ in air or in low oxygen partial pressures down to $10\times {10}^{-7}\mathrm{Pa}$ [24]. Good phase stability of cubic BSFZ in the IT range has been demonstrated recently for up to 100 h [28] and it is further implied by the observation that the perovskite can be synthesized by a sol–gel-based method in the pure phase at $750{}^{\circ }\mathrm{C}$ [27], [29], [30]. This is contrary to the cobaltite BSCF that requires higher temperatures in an analogous synthesis ($950{}^{\circ }\mathrm{C}$) [19], [31]. The present work focuses on a thorough atomic level understanding of electronic effects in the redox behavior of the BSFZ perovskite. Combined in situ electron energy-loss spectroscopy (EELS) and Mössbauer spectroscopy show that a coupled ${\mathrm{Fe}}^{4+}/{\mathrm{Fe}}^{3+}$ valence and ${\mathrm{Fe}}^{3+}$ low-spin to high-spin transition play a key role. This is reflected in anomalies found in integrative investigations of temperature-dependent weight-losses and lattice dilatations.