Compositional optimization of polyimide-based SEPPI membranes using a genetic algorithm and high-throughput techniques.

Abstract

Asymmetric, nanosized zeolite-filled solvent resistant nanofiltration (SRNF) membranes, prepared from emulsified polyimide (PI) solutions via the earlier reported solidification of emulsified polymer solutions via phase inversion (SEPPI) method, were optimized for their performance in the separation of rose bengal (RB) from 2-propanol (IPA). All membranes were prepared and tested in a parallellized, miniaturized, and automated manner using laboratory-developed high-throughput experimentation techniques. Nine different synthesis parameters related to the composition of the casting solutions were thus optimized. In a first, “conventional” approach, a preliminary systematic screening was carried out, in which only four constituents were used, that is, Matrimid PI, NMP as solvent, THF as volatile cosolvent, and an NMP-based zeolite precursor sol as emulsifying agent. A combinatorial strategy, based on a genetic algorithm and a self-adaptive evolutionary strategy, was then applied to optimize the SRNF performance of PI-based SEPPI membranes. This directed approach allowed the screening of an extended, 9-dimensional parameter space, comprising two extra solvents, the two corresponding nanosized zeolite suspensions, as well as another cosolvent. Coupling with high-throughput techniques allowed the preparation of three generations of casting solutions, 176 compositions in total, resulting in 125 testable membranes. With IPA permeances up to 3.3 L.m−2 h−1 bar−1 and RB rejections around 98%, the combinatorially optimized membranes scored significantly better with respect to fluxes and selectivities than the best membranes obtained in the systematic screening. The best SEPPI membranes also showed much higher IPA permeances than two commercial SRNF membranes at similar or slightly lower RB rejections.