The first-principles methods such as density functional theory (DFT) often produce quantitative predictions for physics and chemistry of materials with explicit descriptions of electron’s behavior. We were able to provide information of electronic structures with chemical doping and metal-insulator transition of rare-earth nickelates that cannot be easily accessible with experimental characterizations. Moreover, combining with mean-field microkinetic modeling, we utilized the DFT energetics to model water gas shift reactions catalyzed by Fe3O4at steady-state and determined favorable reaction mechanism. However, the high computational costs of DFT calculations make it impossible to investigate complex chemical processes with hundreds of elementary steps with more than thousands of atoms for realistic systems. The study of molecular high energy (HE) materials using the reactive force field (ReaxFF) has contributed to understand chemically induced detonation process with nanoscale defects as well as defect-free systems. However, the reduced accuracy of the force fields canalso lead to a different conclusion compared to DFT calculations and experimental results. Machine learning force field is a promising alternative to work with comparable simulation size and speed of ReaxFF while maintaining accuracy of DFT. In this respect, we developed a neural network reactive force field (NNRF) that was iteratively parameterized with DFT calculations to solve problems of ReaxFF. We built an efficient and accurate NNRF for complex decomposition reaction of HE materials such as high energy nitramine 1,3,5-Trinitroperhydro-1,3,5-triazine (RDX)and predicted consistent results for experimental findings. This work aims to demonstrate the approaches to clarify the reaction details of materials using the first-principles methods and machine learning force fields to guide quantitative predictions of complex chemical process.

<div>For decades, the response of high-energy (HE) density materials at extreme conditions of pressure and temperature from strong insults like burning or impact have been studied in depth by the shock community. Shock physicists aim to develop a fundamental understanding for coupled chemical and physical processes across orders of magnitude spatial and temporal regimes. In order to succeed, this requires extensive collaboration between experiments and simulations, ranging from the electronic to the engineering scales. The end goals would be to develop predictive multiscale models capable of explaining ignition and initiation of HE systems and composites. The collected works in this thesis detail my contributions to the field of HE materials, specifically addressing the chemical reactivity at the atomistic level using reactive molecular dynamics (MD) simulations.</div><div><div>Through this endeavor, we aim to develop a critical understanding for the decomposition processes of HE materials. We begin with a validation the reactive force field, ReaxFF, by addressing the very strong anisotropic shock sensitivity in 2,2-Bis[(nitrooxy)methyl]propane-1,3-diyl dinitrate (PETN) through direct comparison of time-evolved spectra between experiments and simulations. Such strong orientation dependence is thought to relate to the initial decomposition events. Therefore we compare spectra at three different shock pressures, where we observe similar timescales for the disappearance of the NO2 symmetric and antisymmetric stretch modes. A more detailed chemical species analysis indicates that the NO2 molecular species could be considered the primary intermediate which initiates the decomposition process. Furthermore, these results suggest that the combination of explicit MD simulations and ultrafast spectroscopy will be key to the development of a detailed understanding of chemistry at extreme conditions.</div></div><div><div>Following the validation study, we further our understanding of reactivity in HE systems by investigating the differences in kinetics between an ordered and disordered system. It has been shown that shocked material is often severely strained, causing a loss in crystalline order. This in turn results in the disordered materials, such as amorphous solids, having</div><div>faster reactivity due to their higher internal energy and/or lower thermal conductivity. Our results indicate that extra energy is required to break the long-range order in bulk crystalline systems, thus resulting in slower decomposition rates. Further analyses of thermal hotspots point towards slightly faster chemical propagation in the amorphous samples due to lower thermal conductivity. These results provide an understanding for how molecular disorder can be attributed to increased reactivity.</div></div><div><div>After developing an understanding for the initial decomposition processes of HE materials, we turn our attention to a growing interest in the community which is the developing reduced order chemistry models for use in multiscale efforts. Many schemes report mechanisms that are obtained from experiments, which can have large error bars depending on the apparatus and/or extraction technique, or from gas phase simulations, which may not be relevant at shock conditions. To circumvent these issues, we develop a coarse-grained chemical kinetics model from all-atom reactive MD simulations by taking advantage of an unsupervised dimensionality reduction machine learning technique called non-negative matrix factorization. Doing so allows us to represent the overall decomposition chemistry as latent concentrations akin to reactants, intermediates, and products, which we then use to extract kinetics parameters and heats of reaction. These values are implemented into a continuum model, where we could simulate the criticality of thermal hotspots at regimes beyond the reach of MD, as well as verify how uncertainties in the parameters vary as a function of hotspot sizes.</div></div><div><div>Finally, we close with significant progress made towards on-going and future work, where we address two of the most challenging ideas in the field of HE materials: 1) developing definitive chemistry models at extreme conditions, and 2) improving coarse-grained descriptions for multiscale modeling.</div></div>