Machine-Learning molecular dynamics simulations of Shock Response and Spallation Behavior in PPTA Crystals

Abstract

The shock response of poly(p-phenylene terephthalamide) (PPTA) crystals is investigated using molecular dynamics simulations combined with a machine learning potential. Considering the anisotropy of PPTA crystals, the directions dominated by hydrogen bonding and van der Waals forces are examined, respectively. First, a machine learning potential capable of simulating the shock behavior of PPTA is developed and validated. The potential is demonstrated to achieve excellent accuracy, showing high consistency with density functional theory results. Based on the established machine learning potential, multiscale shock techniques are employed to simulate shock compression at various particle velocities. The Hugoniot curves of PPTA crystals reveal three distinct stages of shock response: elastic, plastic, and cross-linking. With increasing particle velocity, the b axis of PPTA crystals is found to exhibit a greater tendency for plastic deformation. Plasticity along the a axis is characterized by the planarization of adjacent benzene rings within the chains, while along the b axis, it involves the breaking and reformation of hydrogen bonds. The spatiotemporal evolution of thermodynamic parameters and spallation during shock wave propagation is further uncovered through non-equilibrium molecular dynamics simulations. The shock response mechanisms of PPTA fibers are elucidated, providing a foundation for subsequent simulations and their application in impact protection structures.