Influence of defect distribution on the thermoelectric properties of FeNbSb based materials

Abstract

Doping and alloying are important methodologies to improve the thermoelectric performance of FeNbSb based materials. To fully understand the influence of point defects on the thermoelectric properties, we have used density functional calculations in combination with the cluster expansion and Monte Carlo methods to examine the defect distribution behaviors in the mesoscopic FeNb_1−xV_xSb and FeNb_1−xTi_xSb systems. We find that V and Ti exhibit different distribution behaviors in FeNbSb at low temperature: forming the FeNbSb–FeVSb phase separations in the FeNb_1−xV_xSb system but two thermodynamically stable phases in FeNb_1−xTi_xSb. Based on the calculated effective mass and band degeneracy, it seems the doping concentration of V or Ti in FeNbSb has little effect on the electrical properties, except for one of the theoretically predicted stable Ti phases (Fe₆Nb₅Ti₁Sb₆). Thus, an essential methodology to improve the thermoelectric performance of FeNbSb should rely on phonon scattering to decrease the thermal conductivity. According to the theoretically determined phase diagrams of Fe(Nb,V)Sb and Fe(Nb,Ti)Sb, we propose the (composition, temperature) conditions for the experimental synthesis to improve the thermoelectric performance of FeNbSb based materials: lowering the experimental preparation temperature to around the phase boundary to form a mixture of the solid solution and phase separation. The point defects in the solid solution effectively scatter the short-wavelength phonons and the (coherent or incoherent) interfaces introduced by the phase separation can additionally scatter the middle-wavelength phonons to further decrease the thermal conductivity. Moreover, the induced interfaces could enhance the Seebeck coefficient as well, through the energy filtering effect. Our results give insight into the understanding of the impact of the defect distribution on the thermoelectric performance of materials and strengthen the connection between theoretical predictions and experimental measurements.