CHARACTERIZATION OF NANOFILTRATION MEMBRANES FOR PREDICTIVE PURPOSES - USE OF SALTS, UNCHARGED SOLUTES AND ATOMIC-FORCE MICROSCOPY

An asymmetric nanofiltration membrane (Hoechst, PES5) has been charact erised by three different techniques: modelling of the rejection of si mple salts, modelling of the rejection of uncharged solutes and atomic force microscopy. Interpretation of experimental data for the rejecti on of three salts having common co-ion (LiCl, NaCl, KCl) with model ca lculations allows a characterisation of the membrane in terms of three parameters: an effective pore radius (r(p)), the ratio of effective t hickness over porosity (Delta x/A(k)) and an effective charge density (X). Interpretation of experimental data with model calculations for u ncharged solutes (Vitamin B-12, raffinose, sucrose, glucose, glycerin) allows a characterisation in terms of r(p) and Delta x/A(k). Atomic f orce microscopy (AFM) allows direct determination of surface pore radi us r(p)(s) and surface porosity A(k)(5). The AFM images provide direct confirmation of the presence of discrete surface pores in such membra nes. Further comparison of the characterisation obtained with salts an d that obtained with uncharged solutes shows that it is better to desc ribe the transport through such membranes as occurring through discret e pores rather than using an homogenous description of the membrane st ructure. It is also shown that the complexity of a ''space-charge'' de scription of the electric field distribution in the nanometre dimensio n pores of such membranes is not warranted. Direct experimental eviden ce of the charging mechanism of the membranes is provided. Overall cha racterisation parameters suitable for predictive purposes are suggeste d.