Membrane separation technology is a rapidly growing field that has the potential to replace conventional thermal-energy-intensive separation techniques such as distillation and evaporation. These traditional methods consume large amounts of energy, resulting in high carbon emissions and a negative environmental impact. Polymer-based membranes are commonly used due to their simple operation, easy scaling-up and cheap costs. However, the application of this type of membrane is hindered by their solvent resistance in aggressive organic solvents. Exploring novel polymer-based membranes that can withstand harsh organic solvent conditions is a key step towards industrial applications of membrane processes as well as sustainable development. These membranes can be used in various applications, such as purifying liquids, gases, and dissolved substances.

The primary objective of the PhD research was to develop novel polymer-based membranes that exhibit improved chemical stability in harsh organic solvent conditions for organic solvent nanofiltration (OSN). Polybenzimidazole (PBI) has emerged as an ideal material for OSN membrane fabrication due to its excellent mechanical properties, thermal stability, and resistance in various organic solvents. However, pristine PBI membranes show unsatisfactory stability in polar aprotic solvents. Extensive research has been conducted on the covalent crosslinking of PBI membranes using difunctional organic acids. The obtained robust membranes were able to deal with polar aprotic solvents for one week.

Polyimide aerogel film was also investigated as a TFC membrane support, and selective layers of thin-film composites (TFCs) and covalent organic frameworks (COFs) were fabricated on it. The ultrapermeable aerogel support was stable in various organic solvents including harsh polar aprotic solvents. The support enabled the fabrication of both amorphous and crystalline selective layers at both room and higher temperatures.

Subsequently, the selective layer of TFC membranes obtained by formaldehyde and amines through interfacial polymerization was studied and optimized at room temperature. The resultant TFC membranes were stable in common solvents during solvent permeation testing. The thin layers exhibited excellent compatibility with the porous polyimide aerogel support. 

Finally, the establishment of predictive models for performance prediction of OSN membranes was achieved through artificial intelligence algorithms. Instead of developing fundamental mathematical equations, the researcher compiled a large dataset and built AI-based predictive models for both rejection and permeance, based on a collected dataset containing 38,430 datapoints with more than 18 dimensions.