Accuracy and Transferability of Ab Initio Electronic Band Structure Calculations for Doped BiFeO3

Abstract

BiFeO3 is a multiferroic material and, therefore, highly interesting with respect to future oxide electronics. In order to realize such devices, pn junctions need to be fabricated, which are currently impeded by the lack of successful p-type doping in this material. In order to guide the numerous research efforts in this field, we recently finished a comprehensive computational study, investigating the influence of many dopants onto the electronic structure of BiFeO3. In order to allow for this large scale ab initio study, the computational setup had to be accurate and efficient. Here we discuss the details of this assessment, showing that standard density-functional theory (DFT) yields good structural properties. The obtained electronic structure, however, suffers from well-known shortcomings. By comparing the conventional DFT results for alkali and alkaline-earth metal doping with more accurate hybrid-DFT calculations, we show that, in this case, the problems of standard DFT go beyond a simple systematic error. Conventional DFT shows bad transferability and the more reliable hybrid-DFT has to be chosen for a qualitatively correct prediction of doping induced changes in the electronic structure of BiFeO3.