Computational approaches to finding earth-abundant thermoelectric materials*

Good thermoelectric materials should possess a combination of seemingly incompatible properties, such as high electronic mobility and low lattice thermal conductivity. Therefore, search for crystalline materials with glass-like thermal conductivity has been an active field of research. Several cubic I-V-VI$_2$ semiconductors, the paradigm for which is AgSbTe$_2$, have been shown to exhibit minimal values of lattice thermal conductivity at ambient temperatures when the phonon mean free path equals the interatomic distance. These modes are due to the existence of highly polarizable lone $s^2$ electron pairs on the group V cations. Electrostatic repulsion between the lone $s^2$ pairs and the valence charge on group VI anions tends to favor locally distorted bonding configurations and may lead to unstable phonons. We present the results of first-principles density functional theory (DFT) calculations of phonon dispersion and electron-phonon interactions in cubic I-V-VI$_2$ semiconductors, where the group I elements are Cu, Ag, Au or alkali metals, the group V elements are P, As, or Bi, and the group VI elements are S, Se, or Te. Compounds that have only marginally stable phonons have extremely large Gr\"uneisen parameters that result in a thermal conductivity limited by Umklapp processes to values at the amorphous limit above 200 K. Following the {\it ab initio\/} calculation, we synthesized AgSbTe$_2$, AgSbSe$_2$, AgBiTe$_2$, NaSbTe$_2$, NaSbSe$_2$, and NaBiTe$_2$ and report their thermal conductivity and specific heat: in all cases, the experiments confirm the theory.