Separation processes are widely used throughout the chemical and pharmaceutical industries. Polymer membranes have the potential to significantly improve both energy usage and the costs of separation processes by reducing reliance on fractional distillation. For this to occur methods to control the porosity of the polymer membranes must be identified. The P84 molecule is a relatively complex co-polymer with numerous strongly interacting rigid groups, with a persistence length of over 1.1 nm, and the region in which filtration pores form in the membrane is typically 50–80nm thick, whilst the pores of interest within the membrane are typically less than 0.5 nm in size. P84 membranes are used commercially to separate molecules from organic solvents, in a process called organic solvent nanofiltration. Recent experiments with membranes produced from the P84 polyimide molecule found that altering the solvent used in the initial stage of manufacture radically altered the size of the sub-nanometre pores in the filtration region of the membrane. This effect was not expected, and could not be explained by the available models for polymer membrane formation.
I present here a model as well as key results developed during my investigation of the formation of P84 polymer membranes. The model uses a mixture of fully atomistic molecular dynamics simulations of a single P84 molecule in solvent and coarse grained Monte Carlo simulations containing hundreds of complete polymer molecules. It demonstrates that the experimentally observed changes in pore sizes in P84 membranes can be explained by the differing interaction energies between the solvents and the polymers. I further present a new method
for coarse graining aromatic polymers in molecular dynamics simulations which has been shown to permit the time step to be increased from 1 fs to 5 fs whilst maintaining all-atom accuracy.