Berkeley Lab built a tool that tracks atom movement to forecast solid-state reactions, including the impurities that show up along the way
The Breakthrough
Researchers at Lawrence Berkeley National Laboratory built an AI model that predicts how reactions between solid materials play out over time. It is the first predictive model that accounts for how atoms move through materials during solid-state reactions, and its predictions point toward the best recipes for making advanced materials.
The Old Problem
Scientists can find a promising material on a computer. Making it in real life is harder. Producing solid materials often means mixing powders and heating them at high temperatures. This process frequently creates a mix of unexpected compounds instead of the target material. Getting the recipe right can take weeks. Sometimes it takes years. Older computational models tried to predict synthesis outcomes using thermodynamics alone. They could guess the end result but missed the path a reaction takes to get there, along with the impurities that form along the way.
What Changed
The new model tracks atom movement directly. It predicts the pathway a reaction follows, not just its final product. That includes the impurities most older models miss. Purity and yield decide whether a material works for real manufacturing, so this extra detail matters.
What the Lead Researcher Says
Kristin Persson is a senior scientist at Berkeley Lab and a professor of materials science and engineering at UC Berkeley. She said the model lets “materials scientists and industry stakeholders make promising new materials dramatically faster, and with higher purity and yield.” She said the tool can push solid materials toward manufacturing and commercialization faster, and close the gap between lab discovery and technology that reaches people.
Part of a Bigger Effort
Berkeley Lab has spent years building AI tools for materials science. The Materials Project database is one example, used by hundreds of thousands of researchers. Automated labs run by robots test candidate materials without waiting on manual trial and error. This new model fills a gap in that pipeline. It does not just predict which materials might work. It predicts how to make them.
Why It Matters
Batteries, fuel cells, and semiconductors all depend on new materials reaching production faster. A model that turns years of trial and error into minutes of computation can move materials from the lab bench to the market at a faster pace.
Source: Lawrence Berkeley National Laboratory News Center, “New AI Modeling Approach Accelerates the Development of Advanced Materials,” August 3, 2026.
