We present a method for rapidly generating efficient k-point grids for Brillouin zone integration by using a database of pre-calculated grids. Benchmark results on 102 randomly-selected materials indicate that for well-converged calculations, the grids generated by our method have less than half as many irreducible k-points as Monkhorst-Pack grids generated using a more conventional method, significantly accelerating the calculation of properties of crystalline materials.

The performance of solid-state lithium ion batteries can be improved through the use of interfacial coating materials, but computationally identifying materials with sufficiently high lithiumion conductivity can be challenging. Methods such as ab initio molecular dynamics that work well for superionic conductors can be prohibitively expensive when used on materials that conduct lithium ions less well but are still suitable for use as interfacial coatings. We demonstrate a way to address this problem using machine-learned interatomic potentials models in the form of moment tensor potentials. To prevent the potentials from significantly deviating from density functional theory calculations, we use molecular dynamics simulations coupled with on-the-fly machine learning. This approach increases the efficiency of the calculations by 7 orders of magnitude compared to purely ab initio molecular dynamics, significantly reducing the uncertainty in calculated migration energies and improving agreement with experimentally determined activation energies. Using this approach, we identify two particularly promising materials for use as coatings in batteries as well as several others that are candidates for doping-enhanced ionic conduction.

Machine learning interatomic potentials powered by neural networks have been shown to readily model a gradient of compositions in metallic systems. However, their application to date on ionic systems tends to focus on specific compositions and oxidation states owing to their more heterogeneous chemical nature. Herein we show that a deep neural network potential (DNP) can model various properties of metal oxides with different oxidation states without additional charge information. We created and validated DNPs for Ag x O y , Cu x O y Mg x O y , Pt x O y , and Zn x O y , whereby each system was trained without any limitations on oxidation states. We illustrate how the database can be augmented to enhance the DNP transferability for a new polymorph, surface energies, and thermal expansion. In addition, we show that these potentials can correctly interpolate significant pressure and temperature ranges, exhibit stability over long molecular dynamics simulation time scales, and replicate nonharmonic thermal expansion, consistent with experimental results.

No abstract

A computational screen reveals that materials with the CrVO4 structure type may have very low migration barriers for proton conduction.