ELECTRONICALLY CONDUCTIVE PEROVSKITE TYPE MATERIALS Klaus Reichmann, Nikos Katsarakis, Angelika Reichmann Institute for Chemical Technology of Inorganic Materials Graz University of Technology INVITED PAPER MIDEM '98 CONFERENCE 23.09.98 - 25.09.98, Rogaška Slatina, Slovenia Keywords: perovskite type materials, electronically conductive materials, oxide materials, electronic conductance, ion conductance, proton conductance, polaronic conductance, band type conductance, poiarons, semiconductor technology, band formation, band structures, temperature dependence Abstract: Since oxide materials with exceptional electronic properties recently found their application in semiconductor technology, materials research has got important stimulation. New preparative techniques offer the possibility to integrate materials, whose utilisation for a long time was considered to be confined to the ceramics world. This contribution will give a review on a group of electronically conductive oxides with perovskite structure. A short introduction into the crystallography is followed by an overview of theoretical models of band type conduction and polaronic conduction. General considerations for the formation of bands are discussed and the conditions for itinerant or localised electrons are outlined. Examples are given to illustrate these concepts. Elektronsko prevodni perovskltnl materiali Ključne besede: materiali tipa perovskite, materiali prevodni elektronsko, materiali oksidni, prevodnost elektronska, prevodnost ionska, prevodnost protonska, prevodnost polaronska, prevodnost tipa pas, polaroni, tehnologija polprevodnikov, oblikovanje pasov, strukture pasovne, odvisnost temperaturna Povzetek: Uporaba oksidnih materialov z izjemnimi elektronskimi lastnostmi v polprevodniški tehnologiji je dodatno vzpodbudilo raziskovanje na področju materialov. Nove tehnike priprave so omogočile izdelavo komponent, v katerih so kombinirani oksidni materiali z različnimi karakteristikami. V prispevku obravnavamo skupino elektronsko prevodnih oksidov s perovskitno strukturo. Po kratkem uvodu, ki obravnava kristalografijo teh materialov, nadaljujemo s pregledom teoretičnih modelov pasovnega in polaronskega prevajanja. Obravnavamo splošne pogoje za tvorbo prevodnih pasov kakor tudi pogoje za pojav lokaliziranih elektronov. Z nekaj primeri tudi podpremo opisane koncepte. 1. INTRODUCTION Perovskites represent a very common type of ternary compounds with the general formula ABO3. They exhibit a wide range of interesting electrical and magnetic properties, which depend primarily on the character of the d-electrons of the metal cation at the B-site. Most ABO3 compounds are semiconductors or insulators. However a few of them show metal-like electronic conductivity, while others are good ionic conductors, Similar is the range of magnetic properties including the interesting effect of giant magnetoresistance. The source of electric conductivity in some cases is the electronic structure and the formation of bands like in SrRuOs or LaNiOs. Such compounds theoretically can be treated as "metals". Due to the temperature dependence of spin and valence states and coupling of the electronic orbitals non-metal to metal transitions or semiconductor to metal transitions can be observed. Another reason for conductivity may be the formation of poiarons as charge carriers. Poiarons are electrons, partially localised by the polarisation of the lattice. The transport of a polaron in an electric field needs the hopping over an energy barrier located between neighbouring cations of the same species with different valence state. The preparation and modification of such mixed valence compounds can be done by doping with heterovalent cations (examples will be given). In other cases the mixed valence is caused intrinsically by oxygen deficiency. The temperature characteristics of such polaron conductors is similar to semiconductors because of the activated charge transport. Just to complete this overview it has to be mentioned that also ionic conductivity occurs in perovskite type oxides as oxide ion conduction (e.g. LaAIOs and Ca-TiOs) or as proton conduction (e.g. doped SrZrOs). The effect depends strongly on the concentration and distribution of vacancies in the lattice. Ionic conductivity is combined in some cases with a certain electronic conductivity. For some applications e.g. electrode materials for solid oxide fuel cells (SOFC) such a mixed ionic and electronic conductivity is highly appreciated. Principal investigations and the theoretical treatment of these types of conductors have been done years and decades ago. The application of these materials however is just on the start. Nowadays conductive perovskites are under investigation as electrode material as well as solid electrolyte for solid oxide fuel cells (SOFC), to replace noble metal electrodes or as sensor material, As thin films these compounds have raised attention as buffer layers or even electrodes for ferroelectric /1/ or superconductive /2/ thin films. New preparative techniques such as pulsed laser deposition, magnetron sputtering or chemical solution deposition are vital for extending the field of application. 2. STRUCTURAL CONSIDERATIONS Particular for the crystallography of ABXs perovskites is the combination of cations of different size. The bigger A-cation, similar in size to the anion X, forms together with this anion a close packed cubic structure. There the A-cation is twelve-fold co-ordinated. The smaller B-cation occupies octahedrally co-ordinated interstices in that structure entirely formed by the anions. By this way, octahedra containing B-cations are linked at their corners to form a three-dimensional framework (fig. 1). Fig. 1: Unit cell of an ideal perovskite ABOs- The stability of the perovskite structure is primarily derived from the electrostatic energy achieved by the twelve-fold co-ordination of the A-cation. These sites are formed by the corner sharing octahedra containing the B-cation. Thus the first prerequisite of a stable perovskite is the existence of a rigid octahedral framework, which, in turn, requires the preference of the B-cation for an octahedral co-ordination. Moreover a high effective charge is favourable. Since any A-cation must occupy the relatively large interstices created by these corner sharing octahedra, a second prerequisite is the appropriate size of the A-cation. If the A-cation is too large, the B-X bond length cannot be optimised, thus hexagonal stacking with face sharing octahedra becomes competitive /3,4/. If the A-cation is too small, A-X bonding stabilises structures with a lower anionic co-ordination around the A-cation. It should be noted that the ionic radii strongly influence the bonding between the ions and by that way also the band structure. Goldschmidt /5/ defined a very useful relationship for the stability of perovskites containing a tolerance factor t. RA+Rx=tV2(RB+Rx) (1) Ra, Rb and Rx are the empirical ionic radii of the respective ions. The perovskite structure occurs only for values of 0.8 < t < 1.1. The ideal close packing with cubic structure corresponds to t = 1. In most cases however orthorhombic and rhombohedral distortions occur, but also tetragonal, triciinic and monoclinic structures are found. Small values for t (t < 0.8) correspond to a comparable size of A-and B-cations and lead to more close packed structures like ilmenite. For t > 1 the space available for the B-cation in its oxygen cage becomes so large that a displacement is possible. This is the origin of the ferroelectric effect of BaTiOs. The valences of the A- and B-cation can be chosen nearly arbitrarily as long as they sum up to six. Thus perovskites can be classified as l-V-perovskites (e.g. KNbOs), li-IV-perovskites (e.g. BaTiOs) and lll-lll-perovskites (e.g. LaCoOs). Even ReOs can be treated as perovskite with Re6+ as B-cation and a vacancy as "A-cation". Because of the different size of the cations, an inversion, i.e. an exchange between A- and B-cations like in spinels, is impossible. On the other hand the perovskite structure is very tolerant towards defects and so deviations from stoichiometry (oxygen excess or deficiency) can cause mixed-valence compounds. The defect distribution can be statistical or ordered, forming superstructures. Well known for such defect superstructures are the perovskite type high temperature superconductors. 3. PEROVSKITES WITH BAND STRUCTURE Several perovskite oxides exhibit metallic conductivity. Typical examples are ReOs, AxWOs, AIVI0O3 (A = Ca, Sr, Ba), SrVOs, LaTiOs and LaNiOs. An early but still very valuable approach to the band structure of transition metal compounds was derived by Goodenough /6, 7, 8/. With empirically formulated criteria for the overlap of cation-cation and cation-anion-cation orbitals, Goodenough rationalises the nature of the d-electrons in transition metal compounds and the conditions for localised and itinerant electrons. The concept is based on the transfer energy term by, which measures the strength of the interaction between two neighbouring atoms i and j: (2) In this equation H is the Hamilton operator for the electronic wave functions or orbitals ^Fi and of the neighbouring atoms i and j and Ey is the one-electron energy term. The expression (Ti is known as overlap integral. Although it is not possible to get good absolute estimates of by, one can predict its variation in a series of isostructural compounds. In oxides with significant cation-cation interaction, by is proportional to the reciprocal cation-cation separation. Where the cation-anion-cation interaction is important, by is related to the covalent mixing of the cation-anion orbitals. For small values of by, the outer d-electrons are localised, for large values of by they are itinerant in a band and behave like in a metal. In a series of isostructural compounds, there is a critical value of the transfer energy, separating the localised from the itinerant electron regime. This critical transfer energy be is expressed in terms of the position of the cation in the periodic table, the principal quantum number of the d-orbital, the formal charge and the total spin of the cation. In the case of the perovskite the B-cations are octahe-drally co-ordinated by the anions. That means that d-orbitals of the B-cation are no longer degenerated but split into eg-and tag-orbitals. This splitting has to be taken into account for estimating the overlap integrals. Figure 2 illustrates the position and the interesting electron Orbitals in the perovskite structure. The B-cations are placed in the corners of a cube with the anions inbetween on the edges. To simplify at one B-cation only the eg-orbitals and on another B-cation only the t2g -Orbitals are drawn. For the one anion the p-orbitals are drawn. The bonding between anion and B-cation thus can be a- or T:-type. In principle also a bonding between B-cations across the face of the cube has to be considered. Hence the following overlap integrals between neighbouring B-cations labelled 1, 2 and 3 (fig. 2) are possible: Aacc = H^ts) (3) Aacac = CFe1 ATicac = <^t1 (4) (5) As mentioned before, be is related to the atomic number, the formal charge, the principal quantum number of the d-electrons and the total spin of the B-cation. Applied to the LaBOs-series, with B from the first period of the transition metals (Ti3+, V3+, Cr3+, Mn3+, Fe3+, Co3+, Ni3+) it turns out that the total spin determines the electron behaviour /9/. Figure 3 contains the data for the electrical resistivity p and the activation energy Ea for the conductivity of these LaBOs compounds. In LaTiOs and LaNiOs (Ni in the "low-spin" configuration) the spin of the transition metal ions S is equal to 1/2 resulting in itinerant electrons and metallic behaviour (low resistivity and low activation energy). Compounds with S > 1 for the B-cations, such as V3+, Cr3+, Mn3+ and Fe3+ are insulators with localised electrons (high resistivity and high activation energy). A special case is found with the compound LaCoOa where the temperature dependent population of "low-spin"- and "high-spin"-states causes a transformation from an insulator (or better semiconductor) to a metallic conductor around 1200 K /10, 11 /. Aocc is the overlap integral between the t2g orbitals between the cations 2 and 3 (the label cc is for cation-cation overlap). Since the distance across the face of the cube is 5 - 6 A the contribution of Agcc is considered negligibly small. Aacac and Ancac are the corresponding overlap integrals along the edge of the cube involving the covalent bonding with the anion. These integrals determine the behaviour of the d-electrons and if they are large enough, it is appropriate to construct collective electron orbitals or bands. On the other hand, if these overlap integrals are small, the d-electrons are localised on discrete cationic sites, Therefore it is possible to define a critical overlap integral that is proportional to a critical transfer energy be and to distinguish between systems with localised d-electrons and such with itinerant or "band" electrons. Fig. 2: Unit ceil of a perovsi»>5