Utilizing the Molecular Simulation Design Framework (MoSDef) to screen soft matter systems


Utilizing the Molecular Simulation Design Framework (MoSDef) to screen soft matter systems

Wed, 08/02/2023 - 14:00


Clare McCabe
Heriot Watt - IB3

Soft matter systems (those easily deformed at room temperature - e.g., liquids, polymers, foams, gels, colloids, and most biological materials) are ubiquitous in chemistry, but they pose particular computational challenges since the differences in potential energy between distant configurations are on the same order as the thermal motion, requiring time and/or ensemble-averaged data to be collected over long simulation trajectories for property evaluation. Furthermore, performing a molecular simulation of a soft matter system involves multiple steps, which have traditionally been performed by researchers in a “bespoke” fashion. The result is that many soft matter simulations published in the literature are not reproducible based on the information provided in the publication, and large-scale screening (as envisaged in the Materials Genome Initiative) of soft materials systems is a formidable challenge.

To address the issues of reproducibility and automation needed to enable computational screening, we have been developing the Molecular Simulation and Design Framework (MoSDeF, http://mosdef.org) software suite. MoSDeF includes the open­source mBuild (https://github.com/mosdef­hub/mbuild), Foyer (https://github.com/mosdef­hub/foyer) and GMSO (https://github.com/mosdef-hub/gmso) packages. We will introduce MoSDeF and its capabilities in this presentation. We will also show how, by combining MoSDeF with the Glotzer group’s Signac­flow workflow manager (https://bitbucket.org/glotzer/signac­flow), we can facilitate screening of soft matter systems over chemical/structural parameter spaces.

Specifically, results will be presented for the lubrication of nanoscale devices featuring surfaces functionalized by monolayers in sliding contact. Using MoSDeF, a combinatorial screening study was performed to explore tens of thousands of unique monolayer films. While this approach enables us to determine systems with favorable properties more rapidly than could be accomplished through experiment, the screening process still requires a significant amount of time and computing resources. Here, we illustrate the value of coupling MD simulations with machine learning (ML) in order to guide the screening process and reduce the simulations needed in order to optimize system designs. The ML-derived structure-property relations then enable screening of hundreds of thousands of candidate monolayer films, since the structure-property relation requires trivial computational resources.