(621a) Molecular Modeling, Simulations, and Machine Learning for Understanding Structure and Properties in Polymer Solutions | AIChE

(621a) Molecular Modeling, Simulations, and Machine Learning for Understanding Structure and Properties in Polymer Solutions

Authors 

Jayaraman, A. - Presenter, University of Delaware, Newark
Wu, Z., UNIVERSITY OF DELAWARE
Collins, A., University of Delaware
Methylcellulose (MC), a cellulose derivative polymer, exhibits unique phase behavior wherein it is soluble in aqueous solutions at room temperature and at elevated temperatures self-assembles to form fibrils that can in turn form gels and networks. For targeted design of MC based materials for applications (e.g., food additive, drug delivery, tissue engineering, etc.), there is a need to understand this phase behavior, especially the hierarchical structure - assembled networks of fibrils, multiple MC chains packing into a fibril, and individual MC chain conformation - as a function of extent of methylation in the MC chains. In this talk, we present a comprehensive study of the hierarchical structure in MC solutions using a combination of machine learning and genetic algorithm-based analyses of experimental scattering measurements and physics-driven multiscale modeling and simulations. We first use machine learning (ML) enhanced Computational Reverse-Engineering Analysis of Scattering Experiments (CREASE) method to interpret published small-angle scattering (SAS) profiles of commercially available MC chains forming fibrils in aqueous solutions. This ML-CREASE approach provides a quantitative characterization of MC fibril dimensions (i.e., fibril diameter distribution and fibril stiffness). We complement the ML-CREASE interpretation with structural information about chain packing within the fibrils and behavior of water around MC chains obtained through coarse-grained and atomistic molecular dynamics simulations. Together, this top down (ML-CREASE) and bottom up (multiscale modeling and simulation) computational approach provides explanation as to why commercial MC chains assemble into fibrils with consistent diameters regardless of concentration and molecular weight, and how and why the MC chains pack in parallel configuration in the fibril. We also go beyond commercial MC chains and describe how and why the extent of methylation in the MC chains affect their packing and assembly into fibrils.