New techniques are urgently demanded to remove organic micropollutants (OMPs) such as endocrine disrupting compounds, pharmaceuticals and pesticides from our drinking water sources. With increasing salinity of surface waters in coastal regions, salt removal has also become an issue in the production of drinking water. When advanced oxidation processes are combined they are capable to remove OMPs to a large extent, but without the removal of salts. Dense membranes like reverse osmosis (RO) or flat-sheet nanofiltration (NF) membranes – which require an extensive pretreatment – can sufficiently remove OMPs and salts, but only at low permeabilities and by producing a saline waste stream. Hollow fiber NF membranes seem very promising in this respect as they have higher permeabilities, lower energy consumptions and can be applied directly. In this study, the removal of salts and OMPs from synthetic surface water and real lake water from the Valkenburgse Meer (VM) by the polyelectrolyte multilayer (PEM) coated dNF40 membrane was studied in order to make it suitable for dune infiltration water. This type of membrane is fabricated by depositing polycations and polyanions alternately on a porous support medium, also known as the layer by layer (LbL) technique, which enables control of the membrane’s surface charge and permeability. Hollow fiber NF membranes have the ability to separate ions or solutes with smaller sizes and hydraulic radii than the pore size of the membrane due to electrostatic repulsion. The membrane is negatively charged at neutral pH.

The dNF40 membrane showed ion retentions of up to 27% for Cl-, 98% for SO42-, 87% for Mg2+, 76% for Ca2+ and 17% for Na+ in once-through configuration with a hydraulic permeability of 5.8 Lm-2h-1 and molecular weight cut-off (MWCO) of ~200 Da. However, lower ion retention values were observed for filtration of real lake water, except for SO42-. Retentions were unaffected by cross-flow velocity, but increased with increasing permeate flux. Moreover, the ion retentions reduced considerably when the system recovery increased to 75%. In addition to the salt retentions, the retention of a mix of 20 distinctively different pharmaceuticals, both positively charged, negatively charged and neutral OMPs with molecular weights between 119 and 748 gmol−1, was investigated. An average retention as high as 79% and 89% was reached for SW (no natural organic matter (NOM)) and VM water (11.6 mgL-1 of NOM), respectively. As expected, the negatively charged pharmaceuticals were retained best as a result of electrostatic interactions with the negatively charged membrane, followed by the positively charged compounds, of which repulsion is probably promoted by underlying polycation layers in the membrane structure. Neutral compounds were retained less.

The dNF40 membrane showed to be resistant to fouling, while no extensive pretreatment was applied, making it suitable for direct treatment of surface water. Furthermore, OMPs were largely removed, hardness was partially reduced and NOM was almost completely removed, but insufficient NaCl was retained to meet dune infiltration water standards. Therefore, an additional step becomes necessary in the pretreatment of surface water. Four suggestions were provided.