Abstract
For thermoelectric applications, ab initio methods generally fail to predict the transport properties of the materials because of their inability to predict properly the carrier concentrations that control the electronic properties. In this work, a methodology to fill in this gap is applied on the NiTiSn half Heusler phase. For that, we show that the main defects act as donor of electrons and are responsible of the electronic properties of the material. Indeed, the presence of Nii interstitial defects explains the experimental valence band spectrum and its associated band gap reported in the literature. Moreover, combining the DOS of the solid solutions with the determination of the energy of formation of charged defects, we show that Nii defects are also responsible of the measured carrier concentration in experimentally supposed “pure” NiTiSn compounds. Subsequently the thermoelectric properties of NiTiSn can be calculated using a fully ab initio description and an overall correct agreement with experiments is obtained. This methodology can be extended to predict the result of extrinsic doping and thus to select the most efficient dopant for specific thermoelectric applications.