Experimental and Modeling Study of the Nanofiltration of Alcohol-Based Molecules and Amino Acids by Commercial Membranes

The nanofiltration performance of three commercial membranes was analyzed by the Steric Pore Model (SPM) and the extended Nernst-Planck diffusion equation inside membrane pores. The model was completed with the equation to predict the concentration polarization, and the mass transfer coefficient was determined by considering the presence of a feed spacer. The model parameters that characterized the performance of the membrane were the hydrodynamic coefficient, which accounts for the possible variations in solute size and membrane pore radius, the effective membrane thickness, and the water permeability coefficient. All experiments were conducted at fixed feed pH of 6. The rejections of uncharged solutes (glucose for membranes with a high molecular weight cut-off (MWCO) and glycerol and ethylene glycol for membranes with a low MWCO) allowed the model parameters to be determined. We found that glycerol and ethylene glycol overestimate the membrane pore radius due to their ability to interact with the membrane matrix. Therefore, the rejection of glycine as a small amino acid was explored to characterize the membranes with low MWCO since these molecules do not interact with the membrane matrix and have an almost zero charge at pH values between 4.5 and 6.5. Based on the experimental rejections, it was stated that glucose and glycine could be separated by these membranes operating in continuous diafiltration mode.