Science

MIT researchers report a peer-reviewed AI model that combines three-dimensional crystal structures with scientific language. In tests, the team says it identified candidate crystal materials for solid-state lithium-ion battery electrolytes better than a text-only baseline. That is a screening tool, not a finished battery design: the candidates still need to be made and tested in the lab. The study shows one way AI can use atomic arrangement, not just words in papers, to help researchers prioritize materials worth investigating.

A peer-reviewed laboratory study points to a possible target for a future mission to Saturn's moon Enceladus. Researchers froze salty water and used X-rays to watch natron crystals form. Some crystals trapped tiny pockets of fluid, which could preserve clues about the moon's hidden ocean chemistry. This is not evidence of life on Enceladus. Natron has not yet been confirmed on its surface, and the experiment has not shown that it can preserve biological material. But the result could help missions decide which fresh plume deposits are most useful to sample.