Prashun Gorai1 2 Anuj Goyal1 2 Brenden Ortiz1 Eric Toberer1 2 Vladan Stevanovic1 2

1, Colorado School of Mines, Golden, Colorado, United States
2, National Renewable Energy Laboratory, Golden, Colorado, United States

Zintl compounds are attractive as thermoelectric materials owing to their favorable charge transport properties and low lattice thermal conductivities. We have computationally assessed the potential for thermoelectric performance of 145 Zintl compounds and predicted that many of these Zintls, if doped with electrons (n-type), can outperform the hole-doped (p-type) materials. However, almost all known Zintl thermoelectric materials are exclusively p-type, including Yb14MnSb11, Sr3GaSb3, and Ca5Al2Sb6. Only recently, we have reported relatively high thermoelectric performance (zT~1) in two n-type Zintls, KAlSb4 and KGaSb4. To facilitate the search for new n-type Zintl thermoelectric materials, we have identified a simple empirical rule that correlates the average oxidation state of the anion (Anox) and the dopability (p- vs. n-type) of the material. Within Zintl pnictides, compounds with Anox < -1 are p-type while compounds with Anox = -1 can be doped n-type. The dopability of a material is intimately related to its defect chemistry. In Zintl compounds with Anox < -1, as in the case of LiZnSb, Ca5Al2Sb6, and CdSb, n-type doping cannot be achieved due to the presence of large concentrations of electron-killing defects, most commonly the cation vacancies. We demonstrate that in Zintl compounds with Anox = -1, such as KGaSb4, KSb and CdAs2, the electron-killing defects are present in negligible concentrations such that extrinsic n-type doping can be achieved without charge compensation. Using Anox as a tool for predicting dopability, we have screened the Inorganic Crystal Structure Database (ICSD) and identified candidate Zintl compounds that can be potentially doped n-type. Our defect calculations for these candidate materials confirm that these compounds can indeed be doped n-type.