Understanding Complicated Chemistry of Organic Materials Using Machine Learning Interatomic Potential

Understanding the chemical reactivity of organic materials under extreme conditions is important in many fields, such as chemistry, materials science, pharmaceutical, astronomy<i>. </i>Molecular dynamics simulations have become a powerful method that can provide insights into the detailed chemistry at the atomistic level. However, its accuracy depends strongly on the interatomic potential. Here, we present our effort to develop a quantum accurate Chebyshev Interaction Model for Efficient Simulation (ChIMES) many-body reactive potential to study the complicated chemistry of 1,3,5-Triamino-2,4,6-trinitrobenzene (TATB) under a shockwave. We discuss the techniques to control model accuracy and transferability as well as minimal training data selection. Our ChIMES potential has capability of reproducing the structural properties and chemistry of TATB at high accurate quantum level for a wide range of thermodynamic conditions. The shock simulations of TATB using ChIMES show good agreements with available experimental data.