December 3
@
4:00 PM
–
5:00 PM
Michael Webb
Department of Chemical Engineering
Princeton University
Princeton, NJ
Data-driven Strategies to Navigate Sequence, Composition, and Architectural
Complexity in Polymer Design
Understanding and designing polymers with target structural and/or functional properties are grand challenges in materials science. The field of polymer physics provides invaluable scaffolding to elucidate general phenomena of polymer-based materials, but contributions fall short of proffering chemically specific insights or usefully guiding design. Meanwhile, artificial intelligence and machine learning (ML) have greatly enhanced design efforts for many materials classes, including polymers, but success has been mostly limited to chemically “simpler” systems, like linear homopolymers.
In this talk, I will describe our recent efforts to combine simulation, machine learning, and concepts from polymer physics to navigate complex polymer design spaces and accurately construct structure-function relationships for chemically and topologically diverse polymers. Examples will include the design of tunable biomolecular condensates as well as copolymers that modulate enzymatic activity. Furthermore, in considering the rheology of architecturally diverse polymers, I will explain how essential concepts from polymer physics can be combined with natively naive algorithms to enhance data efficiency and model performance. These results establish a pathway to formulate chemically specific predictions as perturbations from baseline theories to enhance future polymer design. These vignettes will highlight both methodological advancements as well as intriguing application areas.