AI helps find a replacement for lithium used in electric vehicle batteries

The new material, “unknown to us and not found in nature, could potentially reduce lithium use in batteries by 70%” and “have enormous environmental, safety and economic benefits,” Microsoft researchers say in cloudblogs.microsoft.com/quantum.

Earlier this year, the new material, for which no further data was provided, was successfully tested in a prototype battery capable of powering a light bulb. Microsoft researchers initially started with 32 million submissions and found more than 500,000 stable candidates. However, identifying candidates was only the first step. Now it was necessary to find materials with suitable properties for this task, in this case for a new solid-state battery electrolyte. To do this, they turned to artificial intelligence and supercomputing, which allowed them to reduce the number of candidate materials to just 18 and in just 80 hours. According to statements to Reuters by a company representative, in this way they were able to significantly shorten a process that traditionally would have taken decades.

The researchers say the discovery is “the first real-world example of many to be achieved in a new era of scientific discovery driven by artificial intelligence.”
The new material could help solve problems associated with lithium, such as environmental issues associated with its mining and its shortage in the market. Lithium is now a key component in smartphones and electric vehicle batteries.

“Lithium and other strategic elements used in batteries are limited resources with limited and geographically concentrated supplies,” says Vijay Murugesan, PNNL group leader, in a blog post. “One of the main focuses of our work at PNNL has been to find new materials for the growing energy storage needs of the future; sustainable materials that conserve and protect the Earth’s limited resources,” he adds.