Early Warning of Battery Thermal Runaway Enabled by Ag-Decorated SnO2 and Machine Learning: Diethyl Carbonate Recognition.

The safe utilization of lithium-ion batteries depends on the management of thermal runaway to avoid electrolyte leakage. It is crucial to make early warning by developing sensors for rapid detection of electrolyte (e.g. diethyl carbonate, DEC). Herein, the superior gas sensing capabilities of SnO2 modified with Ag nanoparticles (SA NPs) have been demonstrated for DEC recognition and warning by machine learning. The research findings show that the introduced silver effectively modulates energy band structure of pristine SnO2, and contributes its catalytic function. As-optimized nanomaterial (SA-5) exhibits 12.6 times response than that of SnO2 toward 10 ppm DEC at 140°C, along with high selectivity, stability, and resistance to silicon poisoning. Kinetic and DFT calculations reveal that silver leads to more negative adsorption energy (-0.49 eV) and fast adsorption kinetic, verifying the synergistic effect of the Ag and SnO2 interface and catalytic function. The extracted transient features were combined with Support Vector Machine (SVM) algorithm to achieve intelligent identification of electrolyte with classification accuracy of 99.3% for carbonates and their mixtures. Notably, a portable device was employed to manifest its practical monitoring ability in simulated environments and real battery leakage, and demonstrates promising application for warning the safety of batteries.