Kinetic modeling of CL-20 decomposition by a chemical reaction neural network

CL-20 is a high energy density material (HEDM) with superior energetic properties. The study of its decomposition mechanism is of great significance to its application in the defense and aerospace industries. Multiple kinetic models of CL-20 decomposition under four heating rates are derived from thermogravimetric (TG) experimental data. The derivation is conducted using the chemical reaction neural network (CRNN). The derived kinetic model can accurately predict the mass change during CL-20 decomposition by inferring the main reaction pathway and kinetic parameters. Two representative kinetic models, including a one-step and a multiple-step model with five substances, are presented to reveal their species evolution along with decomposition. In a further analysis, a potential reaction mechanism of CL-20 decomposition is constructed by combining the multiple-step kinetic model with five substances together with the constraints from experiments and previous works. The reaction mechanism includes three reaction classes: initial decomposition, autocatalytic acceleration, and secondary reactions among products. This work demonstrates that the kinetic models from the CRNN framework can capture the thermal decomposition of CL-20 well. It is expected that the CRNN framework will contribute to the kinetic modeling of other solid-phase energetic materials in the future.

Automated summary

In this study, kinetic models of CL-20 decomposition are derived using the CRNN framework based on thermogravimetric data. The derived models accurately predict the mass change during decomposition and reveal the main reaction pathway. A potential reaction mechanism of CL-20 decomposition is constructed by combining the derived kinetic models with experimental constraints. The results demonstrate the effectiveness of the CRNN framework in capturing the thermal decomposition of CL-20 and its importance in future kinetic modeling of solid-phase energetic materials.