Rechargeable batteries power everything from electric vehicles to laptop computers. They are in-demand, but far from perfect. Improving them means finding the ideal mix of elements from the periodic table, each with unique properties alone and in combination.
Like drug discovery, the search can be overwhelming: thousands of possible materials, only a few with the right traits.
In a battery, the electrodes at each end and the electrolyte in the middle drive the electrochemical reactions that store and release energy.
The challenge is to identify materials that boost energy density and electron flow so batteries last longer and charge faster.
“If you had to choose materials armed with only the periodic table, it would probably take you 100 years,” said Professor Junjie Niu, industrial & manufacturing engineering, whose lab researches improvements for energy storage.
“AI is good for the selection of new materials to use in these parts of the batteries,” Niu said. “It helps us narrow the field to the top five or ten possibilities to meet the performance requirements and then my students focus on only the most promising.”
Different applications – say, a car battery versus a phone battery – require different qualities. So AI also scans and compiles data from past studies, accelerating the screening process and pointing to new directions.
Success lies in asking informed questions, he said.
“We’re not asking AI for the answers. We’re using what we know to ask for specific clues within certain parameters. Then we validate through our experiments.”