Interactions between point defects and internal interfaces in bismuth ferrite
MetadataVis full innførsel
BiFeO3 (BFO) is the most studied multiferroic material and a promising candidate for future information storage. One challenge for practical applications of this material is a thorough understanding and control of the defect structure. Despite the vast number of experimental and theoretical studies of the effect of epitaxial strain and point defects, a fundamental understanding of the interplay in between is still missing. Motivated by the strong strain dependence of both anion and cation vacancy formation in manganese oxides, this thesis investigate the effect of epitaxial strain on formation of Bi vacancies as well as Bi-O vacancy pairs by first principles DFT calculations. Two orientations of strained interface are considered, (111) and (001). For (001)-strain, the situation is further complicated by the strain-induced phase transition from rhombohedral ground state to a super-tetragonal so-called T-phase. Besides the epitaxial strain interfaces, ferroelectric domain walls also hold great potential as functional elements in nanoelectronics, the effects of intrinsic defects on the domain wall properties are also interesting, but not fully understood. Therefore, the third part of this PhD project is to investigate the effect of point defects on ferroelectric domain walls of bismuth ferrite. Firstly, the stability of Bi vacancies and Bi-O vacancy pairs as a function of epitaxial strain in the (111)-oriented BFO thin film has been studied by DFT calculations. The calculated formation enthalpy of point defects are strongly depends on epitaxial strain. Neutral Bi vacancies are energetically more favored by compressive strain than tensile strain when charge compensated by holes oxidizing Fe. We have also studied the stability of bismuth-oxygen vacancy pairs upon different strain states, the stability of Bi-O vacancy pairs aligned out-of-plane with respect to the (111) strain plane is enhanced both by compressive and tensile strain. Bi-O vacancy pairs aligned more parallel to the (111) plane are stabilized by compressive strain, but rather insensitive to the tensile strain. Vacancy pairs aligned out-of-plane, more parallel with the polarization direction, are favored over in-plane pairs for all investigated strain values. The stabilization of Bi vacancies, with or without adjacent O vacancies, is important for thin film growth as strain promotes Bi deficiency. Finally, the local chemical composition around ferroelectric domain walls is also expected to be Bi deficient due to the surrounding local strain fields. For the investigation on (001)-oriented substrates, we firstly reproduced the R to T phase transition upon compressive strain of more than 4.5%. Then by introducing Bi vacancies, we found that the Bi vacancy enthalpy of formation decreases with increasing compressive strain up to ~3.5%, then increases with further compressive strain, before dropping abruptly at the transition from rhombohedral R-phase to tetragonal T-phase. Bismuth-oxygen vacancy pairs are also stabilized under compressive strain. In-plane (IP) oriented vacancy pairs are strongly stabilized upon entering the T-phase, while out-of-plane oriented vacancy pairs are destabilized compared to the R-phase. The calculated spontaneous polarization is insensitive to single Bi vacancies, while IP vacancy pairs reduce the IP polarization components, yielding a net polarization direction close to . IP vacancy pairs under small strain levels cause local in-phase octahedral rotations in contrast with the anti-phase rotations of the stoichiometric bulk material. Bi deficiency is predicted to give p-type conductivity, in agreement with recent experimental work on thin films, domain walls and bulk ceramics, highlighting the importance of cation non-stoichiometry in BiFeO3. Our computational study of domain walls with oxygen vacancies in bismuth ferrite revealed a lower formation energy of oxygen vacancies at the domain wall than in the bulk for all three types of BFO domain walls. This implies that oxygen vacancies tend to migrate to, accumulate at, and pin, the domain walls. A reduction of the electronic bandgap was also found at the domain walls. It has long been a controversy whether experimentally observed enhanced conductivity of BFO domain walls should be attributed to either a structurally induced bandgap reduction or to the presence of oxygen vacancies. Our results indicate that both factors can contribute.