Studies on interactions between membranes (RO and NF) and pollutants (SiO2, NO3-, Mn and humic acid) in water

The comparison between two membrane separation processes, like reverse osmosis (RO) and nanofiltration (NF), and the interaction between membrane and pollutant in water cleaning-up were studied. The separation efficiency of pollutants like silica, nitrate, manganese and humic acids (HA) by means of RO and NF membranes (both of spiral wound type, in polyamide) was determined. The interaction between the membrane and other ions (e.g. Cu ~ casually present in water) also affected membrane cleaning. All pollutants were dissolved in tap water; in some tests they were singularly present, in others they were all together to simulate a polluted water. Particular attention to membrane washing was given testing various washing conditions for different fouling cases. In the separation of silica by RO, fouling problems were not observed in a three hours test at 600mg/l concentration. In RO tests with water containing all four pollutant species at initial concentrations of 126 mg/l silica, 263.8mg/1 NOr, 123.9mg/1 Mn ++, 129.4mg/1 humic acids at pH=8, rejections equal to 98%, 94%, 99% and 95.5%, respectively, were observed. These values of rejections were identical to that obtained when the pollutants were present separately, showing absence of interactions among them. The permeate flow rate was 5301/hxmodule with a recovery equal to 30% at an average transmembrane pressure of 30 bar. In the NF tests both water polluted with all four contaminants and torrent water (from torrent Emoli, Rende (CS)) were used. Obviously the mean rejections were lower than RO membrane, and equal to 35%, 6%, 80%, 35% respectively. Mn ÷+ rejection was the highest owing to the positive charge of the NF membrane. The permeate flow rate was equal to 5301/h×module at a mean pressure equal to 11 bar and T=25°C and comparable to that obtained using the RO module at 30 bar. The best washing for membrane cleaning was a NH3 aqueous solution 0.4% w/v. A simple washing with water was not able to remove Mn++ and Cu++ ions (owing to acid-base Lewis type interaction with polyamide membrane) and humic acids; the weak basic agent NH3 was able to avoid precipitation of insoluble hydroxides, like Mn(OH)2, and consequent plugging of membrane pores, with respect to the strong base NaOH. The obtained results show the importance to know the type of interaction membrane-pollutants and the chemical behaviour of pollutants to obtain the maximum benefit both in pollutant separation and in membrane cleaning. *Corresponding author. Presented at the European Conference on Desalination and the Environment: Water Shortage, Lemesos, Cyprus, 28-31 May 2001.

Reverse osmosis tests on the four chosen pollutants evidenced different membrane behaviour depending on type of pollutant and, consequently, also in membrane washing. Particular problems were not met in silica separation tests; mean rejections was 98% and silica permeate concentrations were lower than 10mg/l also when silica concentration in recirculation tank was high (600mg/l). On the contrary, in nitrate separation tests some problems were met during membrane washing. In fact, in the test with copper nitrate a basic washing with a weak base (NH3 0.4% w/v) was required to dissolve copper ion bounded on the membrane by Lewis interaction so avoiding its precipitation and transport blocking in membrane. Nitrate concentrations obtained in the permeate were about 18.5mg/1 and the mean rejections were equal to 97%. In nitrate separation tests, using sodium nitrate as pollutant, mean rejections were also high (94%) and average permeate concentrations were approximately 59mg/! of NO3- in presence of a high concentration (656.5 mg/1 o f NO3-) in the recirculation tank. In manganese separation tests rejections were about 99% and membrane washing (basic) restored initial flow-rates. In HA separation, some tests were carded out varying pH, obtaining highest permeate flow-rate (5501/h) at pH 5 and permeate concentrations practically zero that means 100% HA rejection. Basic washing with NH3 0.4% w/v confLrrned its effectiveness in removing adsorbed HA. The simultaneous separation tests using the reverse osmosis module showed good rejections that were practically equal to that ones obtained on a single pollutant, showing absence of interaction among pollutants. In nanofiltration tests the average rejections were obviously lower than reverse osmosis ones, and equal to 80% for Mn ~, 8% for NO3-, 9% for SiO2, 47% for HA. A basic washing with NH3 0.4% w/v was also required for restoring initial permeate flow-rates. In conclusion, when using membrane modules for carrying out tests in small scale it is important to use a significant feed volume for minimizing the variation o f concentration arising from adsorption on membrane. It is also important to minimize the washing cycles, both for economical reasons and for avoiding pollution from washing water. Other than low pollutants concentration and absence of colour and turbidity in permeate, obtained with RO and NF membranes, a lower space for membrane plants than traditional ones is required and pollutant removal is realized without by-products formation and without significant chemicals addition.