Lawrence Berkeley National Laboratory

We propose a new method for accurately calculating electrical transport properties of a lightly-doped thermoelectric material from density functional theory (DFT) calculations, based on experimental data and density functional theory results for the corresponding undoped material. We employ this approach because hybrid DFT calculations are prohibitive for the large supercells required to model low dopant concentrations comparable to those achieved experimentally for high-performing thermoelectrics. Using zinc antimonide as our base material, we find that the electrical transport properties calculated with DFT and Boltzmann transport theory exhibit the same trends with changes in chemical potential as those computed with hybrid DFT, and propose a fitting algorithm that involves adjusting the computed Fermi energy so that the resulting Seebeck coefficient trends with temperature match experimental trends. We confirm the validity of this approach in reproducing the experimental trends in electrical conductivity and Seebeck coefficient versus temperature for Bi-doped $\beta-$Zn$_4$Sb$_3$. We then screen various transition metal cation dopants, including copper and nickel, and find that a Cu dopant concentration of 2.56% in Zn$_{39}$Sb$_{30}$ exhibited a 14% increase in the thermoelectric power factor for temperatures between 300-400 K. We thus propose that transition metal dopants may significantly improve the thermoelectric performance of the host material, compared to heavy and/or rare-earth dopants.