Webinar: Computer-Accelerated Materials Design

The theory and simulation of molecules and materials has become increasingly accurate and predictive over the past few decades. The process of computing the chemical and physical properties of a known compound is now well established. The next challenge is to explore the vast space of unknown compounds, and to identify materials with the properties required to support the next-generation of technologies. This is being supported by rapid developments in both hardware (classical supercomputers and the first commercial quantum computers) and software (new algorithms and statistical approaches). Transfer of knowledge from the artificial intelligence community has the potential to supercharge chemical discovery by accessing a large phase space of potential compounds that is inaccessible by high throughput experiments or traditional calculations alone [1,2]. After providing a snapshot of the current status and future direction in this field, I will illustrate developments from our recent progress in the description of materials used in solar energy technologies [3,4]. Current bottlenecks in the field, including the absence of reliable and comprehensive structure-property databases, will be discussed. 

[1] “Computational screening of all stoichiometric inorganic materials” Chem 1, 617 (2016) [link]

[2] “Machine learning for molecular and materials science” Nature 559, 547 (2018) [link]

[3] “Performance-limiting nanoscale trap clusters at grain junctions in halide perovskites” Nature 580, 360 (2020) [link]

[4] “Upper limit to the photovoltaic efficiency of imperfect crystals” Energy and Environmental Science 13, 1481 (2020) [link]