Probing the thermal decomposition mechanism of CF3SO2F by deep learning molecular dynamics

The urgent need to phase out SF₆, an extremely potent greenhouse gas prevalent in electrical grids, drives the search for eco-friendly insulation alternatives. Trifluoromethanesulfonyl fluoride (CF₃SO₂F) emerges as a promising candidate due to its excellent properties. However, understanding its thermal decomposition pathways and products under operationally relevant conditions is critical for evaluating its environmental feasibility and mitigating potential risks upon accidental release or during fault events. This study investigates the thermal decomposition mechanisms of CF₃SO₂F using a deep learning potential that combines ab initio accuracy with empirical MD efficiency. By leveraging machine learning driven molecular dynamics, we systematically analyze the yields and components of decomposition products versus temperatures, gas mixing ratios, and buffer gas. The results reveal that the bond-breaking pathways are temperature-dependent, with both elevated temperatures and higher buffer gas mixing ratios promoting its decomposition. Elevated gas pressure enhances the decomposition process by increasing the collision frequency among reactant species. Additionally, N₂ exhibits an inhibitory effect on decomposition under high pressure compared to CO₂. Experimental validation via a thermal decomposition platform confirms characteristic decomposition products. These findings are pivotal for guiding the rational design and safe deployment of CF₃SO₂F to achieve substantial greenhouse gas mitigation in the power industry.