Progress and Future Directions: The ChIMES Machine Learned Interatomic Model Framework for Chemistry in Energetic Materials

Many of the safety and performance-related properties of energetic materials are governed by complex condensed phase chemistry under extreme conditions precluding direct experimental investigation. Atomistic simulations can provide valuable insights into such systems while circumventing the challenges of physical experiments, but they rely critically on the availability interatomic potentials that are simultaneously exhibit the quantum accuracy necessary to describe condensed phase chemistry under extreme conditions and the computational efficiency necessary to do so on the largeand associated spatiotemporal scales (e.g., approaching 1 μm and μs). Machine-learned interatomic potentials (ML-IAP), which aim to serve as a computationally efficient surrogate for the quantum mechanical potential energy surface, have provided a promising path forward. However, developing robust ML-IAP for condensed phase chemistry in molecular materials presents a significant challenge due to the expansive associated physicochemical space and disparate governing energy scales (i.e., governing bonded, non-bonded, and reactive interactions). In this presentation, we discuss challenges, progress, and future directions in ground-up ML-IAP development for EM within the context of our ChIMES framework.



This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National

Laboratory under Contract DE-AC52-07NA27344.