Discovering Early-Stage Gas Generation Kinetics Enables Thermal Runaway Early Warning in Lithium-Ion Batteries

Early warning of thermal runaway (TR) in lithium-ion batteries remains constrained, as conventional indicators emerge after irreversible failure. Here, we establish a novel framework to advance gas-based early warning by uncovering the earliest chemically driven processes preceding heat accumulation (HA). Operando gas measurements reveal a previously unrecognized chemical activation (CA) stage prior to HA, where coupled electrolyte-electrode reactions initiate at 60 ${}^{\circ }C$ , causing continuous trace gas accumulation without measurable temperature rise. To elucidate the underlying mechanisms, a physics-informed gas generation kinetics network (GGKNet) is developed by coupling chemical reaction kinetics with gas adsorption-desorption, electrolyte evaporation, and lithium-involved reactions. The framework quantitatively captures species-resolved gas generation across states of charge and reconstructs pathway-resolved kinetics during CA and HA stages. Validations under TR scenarios demonstrate that GGKNet achieves superior prediction of gas onset timing, temporal evolution, and spatial distribution, thereby translating kinetic-level insights into guidance for practical early warning systems.