(730g) A New Generation Reactive Force Field Based on Valence Bond Concepts with Polarized Charge Distributions, Fundamental Concepts and Application to Hydrogen Systems

Critical to the development and manufacturing of new generations of functional materials is the design of the nonequilibrium dynamical processing to achieve desired microstructures and transport properties. This requires methods for in silico designing and simulating the non-equilibrium dynamical processes required to manufacture complex functional materials to achieve novel functional and transport properties. Such non-equilibrium phenomena determine the behavior of materials for practical applications, and understanding them is critical to optimize material synthesizability, characterization, and design trade-offs.

We need methods that provide the accuracy of Quantum Mechanics (QM) for Reactive non-equilibrium dynamics simulations of extremely large spatial scales (1 billion atoms to describe a system 100 nm on a side) and time scales (milliseconds to seconds) to characterize the non-equilibrium dynamical processes required to synthesize complex chemical and material systems. Then we must match to the continuum simulations required for developing manufacturing processes. This work is the first step toward this goal.

A great deal of progress toward this goal has been achieved with the development of the ReaxFF reactive force by our group at Caltech. ReaxFF has proved capable of accurate descriptions of complex reaction barriers. It has proved that it is possible to construct a generic force field capable of describing complex reaction dynamics at nearly the accuracy of QM on systems that are many orders of magnitude too large for QM. However, optimizing the parameters of ReaxFF to fit QM is complex so that it can be difficult to rapidly achieve the desired accuracy for all aspects of a system.

We propose to build a new formulation of reactive force fields building on these successes but with the goal of matching the highest level QM while making it more systematic so that the transferable reactive force fields will be attainable that will enable non-equilibrium, adiabatic and non-adiabatic, reactive dynamics of large-scale functional material systems.

In this work we outline the new strategy which we illustrate for the case of hydrogen systems. Here we want to describe exactly the H2 dissociation, the H3 reaction surface, and the hydrogen equation of state from the best available QM computations and validate with experiment. Our new approach is modeled after valence bond concepts to understand bonding, showing that the chemical bond arises from the decrease in kinetic energy (contragradience) from overlapping orbitals.