Towards autonomous high-throughput multiscale modelling of battery interfaces

Abstract

To date, battery research largely follows an "Edisonian"approach based on experimental trial-and-error in contrast to a systematic strategy of design-of-experiments. Battery interfaces are arguably the most important yet the least understood components of energy storage devices. To transform the way we perform battery research, theory and computations can be used simultaneously to understand and guide the design of meaningful and targeted experiments. However, it is well known that modelling of battery interfaces is computationally prohibitive in terms of both resources and time due to the large size of systems to provide realistic and descriptive models. Recently, automated and intelligent in silico tools have been developed to accelerate the description of materials, such as workflows designed to generate, handle and analyse hundreds of thousands of materials data and at different scales. Here, we assess the latest computational strategies, outline unresolved questions, and propose future directions that will guide and drive future developments of interfaces in energy storage devices. The future directions include the development of complementary experimental techniques, such as high-throughput automated materials synthesis, operando characterization, cell assembly and integrated platforms for device testing.