To investigate the thermal decomposition of perfluorinatedketones, ReaxFF parameters were optimized by training against QM-derived calculations using Monte Carlo Annealing method. The optimized parameters showed proper descriptions in predicting the geometries, relative reaction energies and bond dissociation curves of perfluorinatedketones. For validating the reliability of ReaxFF parameters, large-scale atomistic molecular dynamics (MD) with 1 ns long simulations were implemented for thermal decomposition of perfluorinatedketones over the temperature range from 300 to 5000 K. Our simulation results suggested that CC bond cleavages were prior than other bonds in the initial stages of perfluorinatedketones decomposition. Furthermore, kinetic rate constants and product distributions were also obtained from the simulations, and the apparent activation energies (E_a) were predicted to be 40.10 ± 1.27 and 42.03 ± 0.81 kcal/mol for C5 and C6 respectively. With regards to gas mixture decomposition, the main products, including CF₄, C₂F₄, C₂F₆, C₃F₆, C₃F₈, C₄F₈, CO, COF₂, CO₂, were observed experimentally. Concerning the insulating gas decomposition under the moisture condition, an etchant gas HF was observed as a characteristic product, which supplies new insights for perfluorinatedketones thermal decomposition process.

为了研究全氟酮的热分解，通过使用蒙特卡洛退火法对QM派生的计算进行训练，从而优化了ReaxFF参数。优化的参数在预测全氟酮的几何形状，相对反应能和键解离曲线方面显示出正确的描述。为了验证ReaxFF参数的可靠性，在300到5000 K的温度范围内，对全氟酮进行热分解，进行了1 ns长的大规模原子分子动力学（MD）模拟。在全氟化酮分解的初始阶段，C键的裂解要比其他键的裂解先。此外，还从模拟中获得了动力学速率常数和产物分布，并且对于C5和C6 ，表观活化能（E _a）分别预测为40.10±1.27和42.03±0.81 kcal / mol。关于气体混合物分解，主要产物包括CF ₄，C ₂ F ₄，C ₂ F ₆，C ₃ F ₆，C ₃ F ₈，C ₄ F ₈，CO，COF ₂，CO ₂，通过实验观察。关于湿气条件下的绝缘气体分解，观察到蚀刻剂气体HF是一种特有的产物，这为全氟酮的热分解过程提供了新的见识。