DISSERTATION DEFENSE
Nicholas ‘Cal’ Craven, Interdisciplinary Materials Science
*under the direction of Clare McCabe
“Reproducibility in Molecular Simulation: Approaches for the Screening Studies of Patchy Nanoparticles and Switchable Monolayer Films”
07.24.24 | 10:00AM CDT | 044 Engineering Science Building | Zoom
Molecular simulation is a growing branch of scientific study in the fields surrounding molecular discovery. It is an approximate method to gather atomistic and molecular level information from classical equations of motion. Growth in computational resources and algorithmic improvements has continue to expand its popularity and utility for understanding and prediction of molecular properties. As with any developing field, there is an ongoing need to develop and evolve community standards for highly reproducible scientific work. It is necessary to define the degree of rigor, robustness, and broad applicability of these standards. We have contributed to promoting such practices via the Molecular Simulation Design Framework, or MoSDeF, in order to perform TRUE simulations. TRUE stands for Transparent, Reproducible, Usable by others, and Extensible, and generalizes practices for conducting and reporting simulations in the best interests of the community. Using MoSDeF enabled methods, basic simulations were performed across five research groups and six simulation engines to compare the degree of accuracy one might expect when following such guidelines using varying methods and simulation complexity. From these results, expectations are discussed and recommended for what should be considered reasonable best-practices. Additionally, two simulation studies were performed following these protocols using MoSDeF for molecular screening. The initial study focuses on patchy nanoparticles, which are a useful tool for micro-nano structure control via self-assembly. Specifically, anisotropic coated nanoparticles as building blocks have been shown to form complex structures, including square-lattices, hexagonal-lattices, strings, rings, and polyhedral clusters. Silica nanoparticles of varying coating pattern, density, and size were screened for different phase behavior. A second study was performed on simulations of dynamic monolayer films which are designed to respond to the characteristic intermolecular forces of different liquid environments. These simulations identified key metrics to quantify and screen over the degree of responsiveness of a vast number of molecules, and methods were developed to perform an active learning search of this molecular space.