Abstract:

Glass-forming, disordered materials are ubiquitous in our world. Systems lacking long-range order like a liquid but with many of the mechanical properties of a solid have countless applications and encompass many classes of materials, including polymers, which in many ways define the modern age. The use of glassy polymers to make membranes for separations processes has been of interest for several decades now, though scale-up to replace distillation or other thermally-driven conventional methods has been slow. The development of novel polymer materials for use in low-energy membrane separations processes presents a significant opportunity for reduction in the global emissions and fossil energy usage, and is an area of active research. To this end, efforts in this field have led to new classes of highly tunable monomeric building blocks to facilitate these processes, including ladder-like polymers with intrinsic microporosity (PIMs). The discovery of new materials in such a vast design space may be aided by machine learning methods, which have been used to predict new polymers for separations membranes. In an efficient pipeline for materials discovery, predicted structures could be investigated using molecular simulation to gain in silico understanding of their behavior before the time and effort of experimental synthesis is expended. This requires simulation methods which are straightforward, repeatable, and can effectively model the real phenomena that occur during gas permeation.

This dissertation outlines simulation methods for the in-depth study of glassy polymer membranes and presents several computational investigations into the behavior of small, gas-like molecules within the interstices of these systems. Two levels of detail are considered to probe the phenomena of interest: a coarse-grained system in the spirit of the classic Kremer-Grest model, and an all-atom simulation using a polymer-consistent force field. In the first system, we investigate the temperature-dependent dynamics of penetrant molecules in a linear polymer melt with molecular dynamics through measurement of diffusion coefficients and the structural relaxation time. These results are analyzed through the lenses of both a self-consistent cooperative hopping theory (SCCHT) and a machine-learning framework, softness, which has been previously applied to the dynamics of glass-forming systems. In the second system, we model the sorption of industrially relevant small gases into the polymer PIM-1 using a hybrid Monte Carlo and molecular dynamics method to replicate experimental sorption isotherms. We then model this pure-gas sorption in several other glassy polymer systems, including two recently developed catalytic arene-norbornene annulated (CANAL) polymers. Finally, we probe competitive sorption behavior in these polymers under mixed-gas CO2/CH4 feed conditions to model the natural gas upgrading process. These explicitly modeled mixed-gas simulations are compared with predictions based on pure-gas results from computation and experiments. The methods described in this work may be extended to different polymer systems or gas mixtures and can be used to complement and inform the experimental design and development of polymer membranes for gas separations.

Zoom Information:

Meeting ID: 953 0272 3835
Passcode: 053806053806
Link: https://upenn.zoom.us/j/95302723835?pwd=MEUr2JJUbxg9NqUebJ42kErKNXKbOW.1