Abstract: Batteries power our world, but atomic-scale 'potholes' inside the material can cause them to fail without warning. These defects are often too small or faint for traditional tools to see. To solve this, I developed an interpretable multimodal AI that acts as a detective: combining different data streams with the laws of physics to reveal what is truly happening inside the material. While I demonstrate this on batteries, the real contribution is a universal framework. This method provides trustworthy, mechanistic answers for any material system, helping us design better tech, from longer-lasting phones to next-generation microelectronics.

Bio: Haili Jia is a postdoctoral appointee in Argonne’s Nanoscience and Technology Division. Her work integrates AI with quantum-mechanical simulations to extract structure–function insights from complex materials data and accelerate nanoscale materials discovery. She earned her Ph.D. in chemical engineering from The Johns Hopkins University.