Abstract

Classical molecular simulation is a fundamental tool for interpreting and predicting the chemistry of an incredible host of systems, ranging from simple liquids to complex materials and biomolecules. In order to simulate chemically-relevant timescales and system sizes, most simulations are currently performed with the aid of force fields: computationally-inexpensive, parameterized mathematical expressions that approximate the exact potential energy surface of a system. Consequently, the accuracy and predictive capabilities of molecular simulation are directly tied to the accuracy of the underlying force field, thus making accurate force field development a central challenge in theoretical chemistry.

The results presented in this talk will focus on two important advances we have made in developing inter-molecular atom-atom force fields. In the first part of the talk, I will focus on how iterated stockholder atoms (ISA), an atom-in-molecule electron density partitioning scheme, can result in an improved, inexpensive model for the short-range effects that arise as a function of overlapping atom-in-molecule densities. This new methodology offers substantial improvements over previous Lennard-Jones or Born—Mayer approaches, yielding optimal isotropic atom-atom force fields. In the latter half of the talk, I will discuss how, with minimal additional parameterization, we can also selectively incorporate the effects of anisotropic atom-in-molecule density overlaps into all components of the model. These cost-effective, ‘atomically-anisotropic’ force fields demonstrate sub kJ/mol errors over a wide variety of systems, and I will conclude with several examples of how our newly-derived methodology can serve as a basis for increasingly accurate and predictive molecular simulation.