(53d) Computational Design of Formulations and Biomaterials through Multiscale Modeling from All-Atom to Field-Theory | AIChE

(53d) Computational Design of Formulations and Biomaterials through Multiscale Modeling from All-Atom to Field-Theory


Nguyen, M., University of California, Santa Barbara
Delaney, K., University of California, Santa Barbara
Shell, M. S., UC Santa Barbara
Fredrickson, G. H., University of California, Santa Barbara
Sherck, N., University of California, Santa Barbar
Kohler, S., BASF
Solution formulations and biological materials typically involve complex multicomponent mixtures of multiple small molecules, surfactants, and macromolecules – and correspondingly vast design spaces associated with concentrations, chemistries, and sequences. To navigate the design landscape, computational design of such materials must address their self-assembly and phase behavior, which necessitate large scale modeling and advanced algorithms that can be prohibitively expensive to study by traditional particle-based molecular dynamics or even coarse-grained simulations. We address this challenge by taking advantage of statistical field theoretic simulations that efficiently probe large length scale behaviors, provide convenient access to thermodynamic quantities like the free energy, and hence readily enable phase behavior calculations. Historically, the predictive power of field-theoretic simulations has been limited by their dependence on emergent and oftentimes state-point dependent (e.g. chi) parameters with non-obvious connections to the underlying chemistries. Using information theoretic and statistical analyses, we developed new coarse graining techniques that take all-atom simulations as input and systematically derive chemically-specific coarse-grained field theory parameters that faithfully preserve thermodynamic behavior. This powerful new multiscale modeling workflow thus systematically incorporates high-resolution chemical detail into the large-scale simulation abilities of field theoretic simulations, in turn enabling de novo in-silico prediction and high-throughput screening of formulation behavior. We show how this approach enables detailed studies of the solution behavior of multi-component surfactant-polyelectrolyte formulations and of biomaterials comprising intrinsically disordered proteins.