Inside a battery, like a layered cake, different solid layers sit side by side. The meeting point, the interface, behaves capriciously: stresses concentrate there, and over time the battery ages. To understand these processes, we need computer models that see every atom. But accurate calculations are too expensive, and fast ones are too crude.

The authors propose the FIRE method — a way to fine-tune a general-purpose model for a specific task. The model first learns on a variety of materials, then, like a chef tasting a new dish and adjusting the recipe, refines its predictions to match the accuracy of expensive quantum calculations. This requires ten times less data than usual.

Tests on six different interfaces in batteries showed: the energy error is less than one thousandth of an electronvolt per atom, and the force error is about twenty millielectronvolts per angstrom. This is ten times more accurate than previous models. The trained model correctly reproduced the mechanical and electrochemical properties of the materials, matching experiments.

The FIRE method opens the door to predictive materials modeling that was previously impossible due to computational costs. Such models will help design batteries with longer lifetimes and better safety.