Performance of solvent-resistant membranes for non-aqueous systems: Solvent permeation results and modeling

Membrane processes like reverse osmosis (RO) and nanofiltration (NF) can be low energy consuming operations as compared to the traditional chemical engineering unit operations and have been widely used for aqueous systems. Since such membrane processes are low energy consuming operations, their use in non-aqueous systems would offer considerable energy savings. Thus, the study is directed towards development and experimental verification of membrane materials and transport models to explain permeation properties of non-aqueous solvent systems. The understanding of polymer-solvent interactions is critical towards the development of suitable materials and also the prediction of the transport mechanisms. Pure solvent permeation studies were conducted to understand the mechanism of solvent transport through polymeric membranes. Different membrane materials (hydrophilic and hydrophobic) as well as different solvents (polar and non-polar) were used for the study. Pure solvent fluxes for hydrophilic membranes used showed that polar solvents (methanol, ethanol, iso-propanol) had a significantly higher flux (8-10 times) than that of the non-polar solvents (pentane, hexane, octane). On the contrary, the non-polar solvent flux was two to four times that of the polar solvents for hydrophobic membranes. For example, hexane flux at ∼13 bar through a hydrophobic silicone based NF membrane was ∼0.6 × 10^-4 cm³/cm² s. And that through a hydrophilic aromatic polyamide based NF membrane was ∼6 × 10^-4 cm³/cm² s. A simple model based on a solution-diffusion approach is proposed for predicting the pure solvent permeation through hydrophobic polymeric membranes. The model uses molar volume and viscosity of the solvent as parameters for predicting the pure solvent permeability. The model reasonably predicts the pure solvent permeation (R² = 0.89, S.E. ∼ 4%) for hydrophobic membranes. The model has also been experimentally verified using high solution temperatures and also literature experimental data. To extend the predictions to different membranes (hydrophilic and hydrophobic), surface energy and sorption values have been used as a parameter along with the solvent physical properties.