Multiscale Reactive Model for 1,3,5-Triamino-2,4,6-trinitrobenzene Inferred by Reactive MD Simulations and Unsupervised Learning

When high-energy-density materials are subjected to thermalormechanical insults at extreme conditions (shock loading), a coupledresponse between the thermo-mechanical and chemical behaviors is systematicallyinduced. We develop a reaction model for the fast chemistry of 1,3,5-triamino-2,4,6-trinitrobenzene(TATB) at the mesoscopic scale, where the chemical behavior is determinedby underlying microscopic reactive simulations. The slow carbon clusterformation is not discussed in the present work. All-atom reactivemolecular dynamics (MD) simulations are performed with the ReaxFFpotential, and a reduced-order chemical kinetics model for TATB isfitted to isothermal and adiabatic simulations of single crystal chemicaldecomposition. Unsupervised machine learning techniques based on non-negativematrix factorization are applied to MD trajectories to model the decompositionkinetics of TATB in terms of a four-component model. The associatedheats of reaction are fit to the temperature evolution from adiabaticdecomposition trajectories. Using a chemical species analysis, weshow that non-negative matrix factorization captures the main chemicaldecomposition steps of TATB and provides an accurate estimation oftheir evolution with temperature. The final analytical formulation,coupled to a diffusion term, is incorporated into a continuum formalism,and simulation results are compared one-to-one against MD simulationsof 1D reaction propagation along different crystallographic directionsand with different initial temperatures. A good agreement is foundfor both the temporal and spatial evolution of the temperature field.

Automated summary

A coupled response between thermomechanical and chemical behaviors is induced in high-energy-density materials under extreme conditions (shock loading). This study develops a reaction model for the fast chemistry of 1,3,5-triamino-2,4,6-trinitrobenzene (TATB) and applies unsupervised machine learning techniques to extract its main chemical decomposition steps.