Predictive Multiscale Modeling of Polymer Formulations

A method is described for de novo prediction of the self-assembly and phase behavior of polymer melt or solution formulations without any experimental input. The technique employs all-atom molecular dynamics simulations of small systems containing oligomeric fragments of the polymers to obtain equilibrated reference data. Relative entropy methods are then used to best fit a coarse-grained (CG) bead-spring model of the oligomeric fluid with Gaussian soft-core, non-bonded interactions. The resulting CG model is analytically converted to a statistical field theory using Hubbard-Stratonovich transforms, enabling the application of field-theoretic simulation methods to efficiently assess structure and thermodynamics of the original polymeric formulation at the mesoscale. Two examples of executing this workflow are provided, one aqueous and one oil-based, both showing promising agreement between computational prediction and experiment.

This work has been performed in collaboration with M. Scott Shell.