Progress Towards Quantum Accurate Atomistic Simulations of Shock Propagation and Release in DT

Large-scale atomistic simulations of inertial confinement fusion (ICF) experiments naturally include microscopic physics missing from traditional radiation-hydrodynamic codes. Thus, they can better model kinetic processes such as species separation in CH ablators and the subsequent hydrogen streaming and mixing into the deuterium-tritium fuel that occur during strong shocks in these experiments.

We will present progress towards quantum accurate atomistic simulations of ICF using machine learning interatomic potentials (ML-IAPs). First, we will discuss using a recently developed quantum-accurate potential for deuterium gas using the Chebyshev Interaction Model for Efficient Simulations (ChIMES) framework. We show that due to an improved description of the molecular-to-atomic transition, this model can better reproduce the ab initio equation of state, radial distribution functions, and principal Hugoniot than bond order potentials.

Second, we will show that even ML-IAPs struggle to be truly transferable across the entire range of thermodynamic conditions. We will discuss strategies for how to augment existing ML-IAP models with temperature dependent corrections to more accurately describe interatomic interactions including ionization in these simulations.