Liverpool team uses AI for battery breakthrough

Researchers at Liverpool University have used AI tools to aid the discovery of a new solid-state material that rapidly conducts lithium ions.

University of Liverpool

Described in the journal Science, the material was designed to act as a solid electrolyte in lithium-ion (Li-ion) batteries. According to the Liverpool team, it is comprised of non-toxic Earth-abundant elements, and delivers high enough conductivity to replace the liquid electrolytes in current Li-ion battery technology.

The solid-state electrolyte is based on a Li7Si2S7I chemistry with an interconnected set of 15 crystallographically separate lithium sites of varying geometries. It’s claimed this provides a diverse path of conduction pathways for lithium ions, and therefore high conductivity.

“This research demonstrates the design and discovery of a material that is both new and functional,” said Professor Matt Rosseinsky, from Liverpool University’s Department of Chemistry. “The structure of this material changes previous understanding of what a high-performance solid-state electrolyte looks like.

“Specifically, solids with many different environments for the mobile ions can perform very well, not just the small number of solids where there is a very narrow range of ionic environments. This dramatically opens up the chemical space available for further discoveries.”

The discovery of the new material was driven by calculations and decisions made by team members of Liverpool’s Department of Chemistry, assisted by AI. According to Prof Rosseinsky, using AI in this guided, collaborative way - rather than applying it to a task completely independently – is more likely to lead to meaningful materials discoveries.  

“Recent reports and media coverage herald the use of AI tools to find potentially new materials,” he said in a statement. “In these cases, the AI tools are working independently and thus are likely to recreate what they were trained on in various ways, generating materials that may be very similar to known ones.

“In contrast, this discovery research paper shows that AI and computers marshalled by experts can tackle the complex problem of real-world materials discovery, where we seek meaningful differences in composition and structure whose impact on properties is assessed based on understanding.

“Our disruptive design approach offers a new route to discovery of these and other high-performance materials that rely on the fast motion of ions in solids.”