100-year-old battery mystery cracked by AI

AI SCIENCE

100-year-old battery mystery cracked by AI

Scientists in the US have used AI to crack a 100-year-old mystery, which could help build better, longer-lasting batteries.

The problem involved Nanocrystals.

These tiny, disordered materials are too small and chaotic for traditional X-ray methods to reveal their atomic structure.

Unlike large pure crystals, nanocrystals light up in a way that’s hard to interpret, leaving researchers stuck.

Now, a team at Columbia Engineering has trained an AI model, PXRDnet, on tens of thousands of known materials to solve the puzzle.

Much like how ChatGPT learns the flow of language, this AI learned how atoms arrange themselves in nature… and it worked.

It can now identify structures as small as 10 angstroms wide (that’s a width thousands of times thinner than a human hair, not a Swedish basketball player).

Here’s what you should know:

• PXRDnet helps identify atomic structures in tiny, messy materials.

• Old-school X-ray tools couldn’t handle nanocrystal powders; this AI can.

• It could unlock new discoveries in batteries, electronics, and beyond.

10 angstroms of chaos

The breakthrough means scientists can finally study nanomaterials that were previously out of reach.

It’s also a clear sign of how much AI has progressed, solving problems once thought impossible.

“AI learned to solve something that’s puzzled researchers for a century, and did it with barely any physics knowledge,” said one of the team leads.

PXRDnet sounds like a password I’d forget in 2 seconds.