A computational framework for evaluating molecular dynamics potential parameters employing quantum mechanics

A Lanjan and Z Moradi and S Srinivasan, MOLECULAR SYSTEMS DESIGN & ENGINEERING, 8, 632-646 (2023).

DOI: 10.1039/d3me00007a

Molecular dynamics (MD) and quantum mechanics (QM) calculations can be used to characterize novel materials and phenomena that experimental methods cannot capture. While QM provides accurate results, it has high computational costs and is applicable only to small system sizes. On the other hand, MD can work with larger systems and has better computational efficiency but is incapable of studying novel materials/phenomena due to its dependency on experimental data in the literature. Therefore, complex systems such as solid-electrolyte interface (SEI) layer formation cannot be comprehensively investigated by (I) experimental methods due to small time scales, (II) MD simulations because of the absence of experimental data, and (III) QM calculations due to the relatively large system. Herein, we report a suite of new nano-scale algorithms to facilitate studying complex material interphases and molecular systems with the accuracy and precision of QM calculations and at a speed and system size permissible using MD simulations. Our formulation addresses the most challenging aspect of performing an MD simulation, i.e., finding accurate potential (force field) parameters that are often derived from experimental methods. The computational framework presented in this work consists of seven main functions/algorithms that collectively help us account for the effects of nonbonded, bonded, angle, dihedral, and improper interactions in a system/molecule. It is now possible to use these simulations to design and study wholly new and novel materials and investigate phenomena at an atomic/molecular scale under different conditions without the need for prior experimental investigations. We have successfully validated our algorithms with respect to the experimental data of established materials such as H2O (a polar molecule), LiPF6 (an ionic compound), C2H5OH (ethanol), C8H18 (a long chain molecule), and ethylene carbonate (EC) (a complex molecular system). The obtained results have an accuracy of over 90%.

Return to Publications page