AWWA WQTC63981

The objective of this study was to assess the influence of both hydrophobic-hydrophobic and charge interactions on the rejection of selected trace organic pollutants. In order to experimentally generate the rejection values, bench-scale laboratory experiments were carried out. Trisep 4040 TS80-TSF spiral-wound nanofiltration membrane elements were used for all experiments. In order to investigate the influence of solute-hydrophobicity on the rejection, three different categories of uncharged solutes were selected, based on their log K_ow-value: hydrophilic molecules (log K_ow<1), hydrophobic molecules (log K_ow>3), and a category of solutes of intermediate hydrophobicity (1ow<2). Compounds of increasing molecular weight were chosen, to construct rejection curves in function of molecular weight. Some interesting trace organics (e.g. NDMA and MTBE) however, were also included in the testing. In order to investigate the influence of charge, several pharmaceuticals were selected, based on their molecular weight and charge. Positively and negatively charged pharmaceuticals were selected, as well as some neutral pharmaceuticals. All organic contaminants were analyzed by GC/MS (gas chromatography, combined with mass spectrometry) or LC/MS (liquid chromatography, combined with mass spectrometry). The feed water matrix for the filtration experiments with the neutral compounds was Milli-Q water. For the pharmaceutical experiments, surface water from the Weesperkarspel treatment plant in Amsterdam (DOC-content 5 mg/l) was used. Compounds were spiked into the feed waters in concentrations of 2 µg/l. A bench-scale, cross-flow membrane filtration test unit for 4040 membrane modules was used in this study. The unit utilizes a multi-impellor centrifugal pump (Grundfos CRN 3) capable of providing pressures of up to 25 bar at a flow rate of 3 m³/h. Temperature of the feed tank was kept constant at 20ºC using a cooling system (Tamson TLC 10B). Feed, permeate and concentrate flows were monitored by rotameters. The bench-scale NF experiments were carried out at a recovery of approximately 10%. All test unit parts in contact with the solution are made of stainless steel to minimize adsorption of the organic solutes used. Bench-scale filtration experiments were performed at a constant cross-flow rate (Qf = 1500 l/h, cross-flow velocity = 0.2 m/s (which correspond to a concentration polarization factor of 1.02)) and transmembrane pressure for 4 days at 20ºC and pH 7-8. Experiments lasted for 4 days to reach saturation of the membrane with the compound of interest, so no overestimation of the rejection was obtained. Samples of feed and permeate streams were taken for analysis of organic solutes concentration. Before use, membranes were rinsed for 2 hours with tap water (to remove the preservation liquids present in the membranes) and characterized for MgSO₄-rejection with a 500 ppm MgSO₄-solution in Milli-Q water. Includes 12 references, tables, figures.