Water Journal : Water Journal December 2011
contaminants of concern technical features 90 DECEMBER 2011 water by the surface charge of the membrane and the degree of ionisation of the acid (Bellona and Drewes, 2005). At pH levels below its pKa, ibuprofen was less than 50% charged and, being hydrophobic, is adsorbed on the membrane. At pH levels above the pKa, adsorption of ibuprofen was less with increasing pH because of the increase in the amount of the anionic form, which has a strong electrostatic repulsion influence. However, this effect does not appear to be as strong as hydrophobic interactions. Rejection was at a minimum at pH 5. It was found that ibuprofen was predominantly removed by adsorption and above pH 5 it was removed by electrostatic repulsion. A tighter membrane gave higher rejections. The radius of the ibuprofen molecule is 0.38 nm and the pore radii of NF membranes used in another study were 0.8--1.4 nm (Park and Choi, 2005). Hence, rejection of this pharmaceutical compound has to be by a mechanism other than one involving molecular size, confirming that the high negative charge on the molecule at pH 7 and the negative character of the membranes are responsible for repulsion of the micropollutant. Because conventional coagulation and adsorption on GAC are inefficient in the removal of estrogens, there are many examples that use further treatment involving membranes (Bodzek and Dudziak, 2006). NF has been applied in this way with polyamide and cellulosic commercial membranes. The membranes alone removed >63% of a mixture of six natural and synthetic estrogens, but after coagulation with poly(aluminium chloride) this was elevated to >81.5% removal. Depending on the type of estrogen, with the polyamide membrane the retention coefficients ranged from 81.5% for estriol to 100% for mestranol. Much higher values were obtained for synthetic estrogens, which have a lower solubility in water. Of course, the role of the coagulant is to simply insolubilise the micropollutants, so a better coagulant is needed for this purpose. A specifically designed inorganic coagulant should have some hydrophobic surfaces too. A recent example is an organic-modified polysilicato iron made for treating dye wastewater (Fu et al., 2011). A recent study has been published of the interactions between trace organics, NF membranes, fouling layers, and NOM macromolecules present in raw water (Hajibabania et al., 2011). The adsorption of organics onto the membrane surface and their association with NOM were shown to be dependent on the physicochemical properties of the organics and the molecular weight of the NOM. Model compounds were used to simulate the effect of the nature of the foulant on adsorption behaviour and fouling. The characteristics of the NOM influenced the amount of adsorption of trace organic compounds. A decrease in rejection was observed with alginate- and humic acid-fouled membranes. Some groups bound more with alginate rather than humic acid, and this enhanced adsorption could be closely correlated with the number of functional groups and the molecular size of the NOM. It was found that two counteractive mechanisms were commonly involved in the rejection of hydrophilic non-ionic solutes: adsorption onto the NOM resulting in an increased rejection, and the presence of cake-enhanced concentration polarisation leading to a decrease in rejection. The negative effect of fouling on the removal of trace organics was more dominant than the influence of adsorption onto the NOM. The effect of fouling on the rejection of hydrophobic trace organics was more limited. Their rejection was strongly influenced by the surface charge densities of the fouled membranes, since a decrease in rejection was observed with alginate- and humic acid-fouled membranes. For hydrophobic ionic compounds, two mechanisms were again involved: electrostatic repulsion of the compounds with the membrane surface, and to a lesser extent, adsorption onto NOM in the bulk and on the fouling layer. It was hoped that once the properties of the main trace organics were properly defined, along with the characteristics of the feed water, long-term rejection performances of NF membrane could be better predicted. In practical applications of NF/RO membranes to municipal wastewater treatment, the feed water always contains organic macromolecules at TOC levels of up to 10 mg/L (Kimura et al., 2009). These are mainly soluble microbial products produced during biological treatment processes such as an activated sludge process. A study of the influence of these organic macromolecules on the removal of six pharmaceuticals by NF/RO membranes was undertaken. Two types of biological treatment, conventional activated sludge followed by media filtration (tertiary treatment) and a membrane bioreactor (MBR), were examined as sources of feed waters to NF/RO membranes. Removal of the pharmaceuticals was higher from these feed waters than from de-ionised pure water spiked with the pharmaceuticals, with the increase being significant in the case of the NF membrane. Alteration of the membrane surface because of membrane fouling and association of the pharmaceuticals with organic macromolecules were postulated to account for this increase in removal. It was proposed that the organic macromolecules present in tertiary effluent enhanced removal of the pharmaceuticals by NF because of a modification of the membrane surface. However, the organic macromolecules present in the MBR effluent seemed to enhance removal of the pharmaceuticals by NF because of association with them. The different mechanisms highlight the different properties of the organic macromolecules present in the two types of effluent. The prediction of membrane rejections based on a detailed understanding of organic compound rejection levels as a function of the properties of the compounds and the membrane is a formidable task. An alternative way is to develop high-science models of quantitative structure-property relations (QSPR) that take into account the simultaneous correlation of organic compound rejection with multiple molecular parameters for the membrane. Artificial neural networks have this capability for building multi-parameter QSPRs with wide applicability (Libotean et al., 2008). Such models have been developed using the results of RO performance covering 50 different organic compounds and five different commercial RO membranes at the Orange Let's not waste it...
Water Journal April 2012
Water Journal November 2011