Robert McKinney1 2 Prashun Gorai1 2 Eric Toberer1 2 Vladan Stevanovic1 2

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

The emergence of inorganic quasi-2D crystals has generated an interest within the material science community due to a diverse range of properties and relative ease of implementation into complex structures. Quasi-2D materials garner considerable attention because of their potential applications in nanoelectronics, optoelectronics, energy storage, and thermoelectrics. Computational identification of quasi-2D materials typically only considers materials with Van der Waals bonding between the layers. Complex variants of Van der Waals materials often do not have a Van der Waals gap (such as CsBi4Te6), but are still sought out specifically for properties that stem from their layered behavior. The same techniques used for identifying quasi-2D Van der Waals materials can also be applied to search for layered materials with stronger bonding. In this work, we have identified a group of layered ternary materials which are characterized as having binary layers separated by an ionically-bonded spacer element between the layers. We conducted a high-throughput search through ternary systems within the inorganic crystal structure database (ICSD) and identified over 700 compounds which we classify as ionically-bonded layered materials. As confirmation that we correctly identify such materials, we find that the largest grouping of spacer elements comes from the group I and II elements, which fits with the general understanding that many of these materials have spacer elements such as Cs, Ba, Li, Ca, and La. To assess the elastic stability of these structures, we preformed DFT relaxation and elastic calculations within the plane-wave VASP code. From DFT, we found approximately 45% of the compounds within the set with nonzero bandgap. From the elastic tensor, we assessed the anisotropy of these materials using the universal anisotropy index (AU), which varied from completely isotropic (AU ≈ 0) to highly anisotropic (AU > 10). From the elastic tensor we can calculate the speed of sound for any lattice vector using Christoffel's equations. We use this in conjunction with isotropic material parameters to predict the lattice thermal conductivity using a semi-epirical model that we have previously developed. We find that the majority of these layered compounds are predicted to have low lattice thermal conductivity, with 80% having an average value lower than 10 W/mK and 30% with κl lower than 2 W/mK. This ratio holds for both zero-gap and finite bandgap. Within this set of ionically-bonded layered materials, we expect to find many suitable candidates for thermoelectric performance and other applications necessitating low thermal conductivity.