To support our application areas, we develop new techniques and software tools in computational chemistry and physics.
Lattice dynamics and temperature
working closely with Atsushi Togo (Kyoto University) on the development and testing of Phonopy and Phono3py. Our current focus is theoretical vibrational spectroscopy (IR and Raman), anharmonic phonon-phonon interactions (lifetime effects), and thermal conductivity. See recent work on soft modes in the record thermoelectric material SnSe.
High-throughput screening and machine learning
we are developing a low-cost screening package based on exploring new materials using simple chemical descriptors. Our implementation is in the open-source Python package SMACT. See our recent perspective in Nature on the potential for machine learning tools in the chemical sciences.
Thermodynamics and synthesizability
in addition to electronic properties, first-principles simulation techniques also provide accurate total energies, which can be used to construct phase diagrams and reaction fields as a function of temperature and pressure. These can be used to identify processing windows for entirely new phases and properties. One application has been Cu2ZnSnS4.
Imperfect crystals and disorder
all crystals are defective owing to configurational entropy. Materials modelling can probe both the probability of defect formation as well as their structural, vibrational and optical properties. This includes the effect of disorder and defects on electron-hole recombination in photovoltaic devices, and approaches to achieve “defect tolerence”. We are involved in the application the multi-scale ChemShell package for embedded crystals, inspired by the Mott-Littleton procedure for the calculation of point defects in crystals.