Data-driven design of electronic band structure for materials
Eric B. Isaacs and Chris Wolverton
Department of Materials Science and Engineering, Northwestern University,
Evanston, Illinois, USA
Although electronic band structure is routinely computed, designing a material with a desired band structure remains an outstanding challenge. In this talk, I will describe an approach using materials database screening with materials attributes based on the constituent elements, nominal electron count, atomic coordination environment, and thermodynamics. Using the over half a million real and hypothetical inorganic crystals of the Open Quantum Materials Database, this approach is applied to two disparate band structure design problems. In the first, we seek a “pudding-mold” band structure containing both flat and dispersive components, which leads to large thermoelectric power factor. One of the identified compounds, BaPdS2, exhibits ultralow lattice thermal conductivity in addition to the pudding-mold band structure, leading to remarkable thermoelectric figure-of-merit approaching 3. In the second, we search for materials with a single correlated d band at low energy, an important yet rare property of the cuprate superconductors, which may serve as benchmarks for the one-band Hubbard model. Several Cu and Fe compounds, including bromide, oxide, selenate, and pyrophosphate chemistries, achieve the desired electronic structure and exhibit novel correlation properties such as Mott insulating behavior and antiferromagnetism.