Abstract
Conventional research strategies for battery materials development have largely relied on an edisonian (trial-and-error) approach, in which each step in the discovery (value) chain depends sequentially on the success of the previous step(s). In recent years, a number of advances have emerged, such as the tight integration of virtual computations (usually at atomic scale) with materials design and operandi characterization techniques in a circular design loop, which can help accelerate the discovery cycle for next-generation battery technologies. This is illustrated by the progress made in the field of high-capacity positive electrodes for Li-ion technology, as well as in the field of materials for secondary metal-air batteries. However, this effort is not enough. Ideally, such a materials development process should engender (almost) simultaneous integration of experimental and theoretical research in an integrated development platform, enabling near-instantaneous cross-talk of results from complementary techniques.
Major efforts have been made to set up standardized infrastructures enabling users to store, maintain, track and share data in a well-defined and structured format, accessible from different platforms and for different purposes. Although these databases contain many millions of items of information, there is a need for tools and infrastructure to link these different databases, i.e. between experimental and theoretical data, but also between different lengths and time scales. The long-term vision is to develop a versatile, chemistry-neutral framework capable of increasing the rate of discovery of new battery materials and interfaces by a factor of 10. The backbone of this vision will eventually enable the reverse design of ultra battery materials, and even allow cross-cutting aspects such as manufacturability and recycling to be integrated directly into the discovery process. This seminar will attempt to illustrate this new research paradigm, in contrast to the Edisonian approach, via artificial intelligence.
