Machine Learning Enabled Quickly Predicting Of Detonation Properties Of N-Containing Molecules For Discovering New Energetic Materials

Energetic materials are widely used in the fields of military, civil engineering, and space exploration. The discovery of new energetic materials is essential to develop next-generation technologies of weapon, mining, construction, and rocket propelling. In this study, a machine-learning-assisted method is developed for accelerating the discovery of new energetic materials via efficient prediction and quick screening. Suitable neural networks are established for accurately predicting the detonation properties of various N-containing molecules based on their structures, including density (rho), detonation velocity (D), and detonation pressure (P). Then, the minimum database volume for high-precision extended prediction is determined. A proof-of-concept study for discovering new energetic compounds using machine learning is carried out, and 31 new N-containing molecules with outstanding detonation properties are discovered. It is expected that the development of next-generation energetic materials is greatly accelerated by the application of this strategy assisted by machine learning.