Membrane filtration in water recycling: removal of natural hormones L.D. Nghiem,.I. Schäfer and T.D.Waite Centre for Water and Waste Technology, School of Civil and Environmental Engineering, The University of NSW, Sydney NSW 52, ustralia (E-mail: a.schaefer@unsw.edu.au) bstract Recent detections of endocrine-disrupting chemicals (EDCs) in effluent are of great concern to sections of the community associated with the issue of water recycling. In vitro and in vivo studies by many researchers have confirmed the impacts of EDCs on trout at the common concentration encountered in sewage effluent. mongst many types of EDCs the impacts of steroid estrogens such as estrone, estradiol (natural hormones) and ethinylestradiol (a synthetic hormone) are prominent as they have far higher endocrine-disrupting potency than other synthetic EDCs. Given the continuous developments in membrane technology, tertiary treatment using membrane processes has been identified as a promising technology to provide a safeguard to water recycling practice and to protect the environment. This paper investigates retention and adsorptive behavior of the natural hormones estrone and estradiol by two commercial lowpressure nanofiltration membranes TFC-SR2 and TFC-S, using dead end stirred cell systems. The removal phenomena of estradiol are similar to that of estrone. p has been found to significantly influence the adsorption of estrone and estradiol by the membranes, presumably due to hydrogen bonding. This adsorption is critical in the risk of possible release of such hormones to the product waters. Total adsorbed amounts were calculated for standard membrane elements and are indeed important. Keywords dsorption; endocrine disrupters; hormones; nanofiltration; water recycling; water and wastewater treatment Introduction Recent detections of endocrine-disrupting chemicals (EDCs) in effluents are of great concern to sections of the community associated with the issue of water recycling. In vitro and in vivo studies by many researchers have confirmed the impacts of EDCs on trout at the common concentration encountered in sewage effluent. mongst many types of EDCs the impacts of steroid estrogens such as estrone, estradiol (natural hormones) and ethinylestradiol (a synthetic hormone) are prominent as they have far higher endocrine-disrupting potency than other synthetic EDCs (Johnson and Stumpter, 1). Performance of conventional wastewater treatment of different plants on removal of these compounds varies greatly and, concentrations of some steroid estrogens in secondary effluent are still able enough to harm wildlife such as fish in particular (Johnson and Stumpter, 1). In spite of the magnitude of this problem, research on the removal of EDCs in water and wastewater treatment remains to date very limited due to their relatively low concentration and the associated analytical difficulties. Given the continuous developments in membrane technology, tertiary treatment using membrane processes has been identified as a promising technology to provide a safeguard to water recycling practice and to protect the environment. Several researchers have shown that nanofiltration is capable of removing trace organics including natural hormones and a wide range of pesticides (Kiso et al.,, 1; Schäfer et al., 3). In our previous work, removal of the trace contaminant estrone using eight different nanofiltration and reverse osmosis membranes, which cover a wide pore size range, has been studied. It was found that estrone could be adsorbed to the surface of some membranes. This adsorptive phenomenon is of concern as it may result in contaminant leakage or bulk release when Water Science and Technology: Water Supply Vol 3 No 3 pp 155 1 IW Publishing 3 155
desorption occurs. This paper investigates retention and adsorptive behaviour of the natural hormones estrone and estradiol on two nanofiltration membranes TFC-SR2 and TFC-S and determines how much of such compounds are likely to accumulate on the membranes. Materials and methods Membranes TFC-S and TFC-SR2 were selected for this study due to their excellent permeability at low pressure. They were supplied by Fluid System (San Diego, US). Membrane types and pure water flux at 5 bar are summarised in Table 1. TFC-S is expected to have a smaller pore size as compared to TFC-SR2 due to its higher salt retention (data is not shown) and pure water flux differences. Filtration system and protocol schematic of the filtration system is shown in Figure 1. Experiments were carried out in a 185 ml stainless steel stirred cell. The inner diameter was 56.6 mm resulting in a membrane surface area of 21.2 1 4 m 2. n micon magnetic stirrer was used and the stirrer speed was set at rpm. Instrument grade air was used to pressurize the stirred cell. new membrane was used for each experiment. Each experiment was conducted in three steps. The membrane was compacted for 1 hour using MilliQ water at 1 bar. Pure water flux was then determined at 5 bar. In the third step, the reservoir was emptied and the cell filled with the test solution. The solution was filtered at 5 bar and six permeate samples of ml each were collected from a feed volume of 185 ml. retentate sample was also collected for analysis. Parameters used to quantify the efficiency of a membrane were flux (J) and solute retention (R) where the flux is defined as dv J 1 dt and retention as cp R = ( 1 ) c (1) (2) Table 1 Membrane types and pure water flux Membrane Type verage pure water Membrane Membrane flux* [L m 2 h 1 ] resistance [m 1 ] material TFC-S 55. ± 7.3 3.3 1 13 Polyamide on Polysulfon support TFC-SR2 77. ± 25.2 2.3 1 13 * verage values are derived from all experiments and variations are averaged F G C D E F G Stirred cell Magnetic stirrer Membrane Stainless steel porous support 2 L reservoir Pressurised instrument air inlet Feed inlet, safety and release valves Permeate outlet D C 156 Figure 1 Membrane filtration stirred cell set-up
Solution chemistry and chemicals ll chemicals were of analytical grade. Radiolabelled estrone-2,4,6,7-3 (N) and estradiol- 2,4,6,7-3 (N) were purchased from Sigma ldrich (Saint Louis, Missouri, US). The background electrolyte consisted of 1 mmol l 1 NaCO 3, and mmol l 1 NaCl. p was adjusted using 1 mmol l 1 Cl or 1 mmol l 1 NaO. Natural hormone characteristics and analysis Molecular structures of estrone and estradiol are presented in Figure 2. oth compounds are hydrophobic and have a very low solubility in water (Merck, 1996). The acid dissociation constant, pka, of estrone is 1.4 (Schäfer et al., 3). Estradiol has a very similar molecular structure to estrone; thus, it is expected to have the same pka value. ydroxyl and carbonyl functional groups of estrone and estradiol make them capable of participating in hydrogen bonding, as a proton-donor or proton-acceptor species. Feed solutions were prepared by spiking estrone or estradiol into background electrolyte solution to make up ng/l of estrone or estradiol, respectively. This is a typical concentration of natural hormones often encountered in surface waters and wastewaters. Estrone was analysed using a Packard Instruments scintillation counter. dsorption of estrone or estradiol to the membrane was determined by cutting the membrane into small pieces at the conclusion of each experiment. The membrane is then placed into a scintillation vial to which 5 ml of acetone is added. The vial was shaken vigorously and left for 1 hour for all estrone to dissolve. 1 ml of solution was extracted into another vial for air drying. The residue was redissolved with 1 ml of MilliQ water and 9 ml of scintillation liquid added prior to analysis. Results and discussion Effect of p on adsorption of estrone s indicated previously, eight membranes were screened for estrone retention and from those results two membranes were selected for further study; the TFC-SR2 and TFC-S due to an expected difference in pore dimension based on pure water flux (see Table 1) and salt retention. Figure 3 shows that adsorption of estrone by both membranes drops drastically with the dissociation of estrone at p 1.5. It is not surprising that adsorption capacities of the two membranes are almost identical as they are both of polyamide on polysulfon support. The experiments do not allow differentiation between adsorption on the active layer and the support material. ydrogen bonding was suggested as the mechanism of adsorption of estrone by the membrane (Schäfer et al., 3). ydroxyl groups are the most likely interaction sites due to the resonance structures of the aromatic groups. When dissociated, estrone loses its proton and becomes unable to participate in hydrogen bonding with membrane functional groups, resulting in a reduction in adsorption and lower retention. C 3 O O C 3 O O Figure 2 Molecular structure of estrone () and estradiol () 157
% Estrone adsorbed 3 1 TFC-SR2 TFC-S 2 4 6 8 1 12 p(-) Figure 3 Estrone adsorption as a function of p on TFC-SR2 and TFC-S membranes ( ng/l estrone; 1 mmol l 1 NaCO 3 ; mmol l 1 NaCl) dsorption effect on estrone retention Figure 4 compares retention by TFC-SR2 and by TFC-S at different p. Retention by both membranes decreases as p exceeds a pka value of estrone (1.5) in parallel with the decreased adsorption of Figure 3. owever, while retention by TFC-SR2 drops drastically with the dissociation of estrone at p 1.5, reduction in retention by TFC-S is more gradual. In particular, the graph shows a variation of about 1% for TFC-S membrane compared to % reduction for the TFC-SR2 membrane. This result indicates that both adsorption and size exclusion can influence retention of TFC-S membrane due to its smaller pore size as compared to TFC-SR2 membrane. From this result, it can be speculated that estrone has similar dimensions as the pore size of TFC-S membrane and size exclusion may be the dominant retention mechanism for smaller pore size membranes. dsorption effect on estradiol retention Despite the different in functional groups (estrone has a carbonyl group while estradiol has a second hydroxyl group, see molecule structures in Figure 2) other characteristics of the two compounds are very similar. Not surprisingly, estradiol adsorption and subsequently its retention by TFC-SR2 resemble that of estrone. The dissociation of the only, and in the case of estradiol, the first, hydroxyl group at p 1.5 has the greatest impact on retention. This also implies that results reported here could be applied for other estrogenic compounds such as estriol and ethinylestradiol. Time dependence of adsorption It appears that adsorption of trace contaminants on membranes is a temporary effect that occurs in the initial stages of filtration. While this adsorption should not be relied upon for the removal of trace contaminants, adsorption is likely to continue until the material is saturated and lead to the accumulation of large amounts of contaminants. 158 9 2 4 6 8 1 12 14 p (-) Permeate Concentration (ngl -1 ) 9 2 4 6 8 1 12 14 p (-) Figure 4 Permeate concentration and estrone retention by TFC-SR2 () and TFC-S () membranes as function of p ( ng/l estrone; 1 mmol l 1 NaCO 3 and mmol l 1 NaCl) Permeate Concentration (ngl -1 )
To investigate the limits of this adsorption and subsequent retention of saturated membranes, experiments were conducted with a series of fresh feed solutions for one membrane. Results from these experiments are presented in Figure 6. fter 7 feed volumes, both retention and adsorption reach equilibrium and retention decreases with a reduction in adsorption (adsorption data is not shown). This indicates that the measured retention is dominated by adsorption phenomena for both membranes and that the retention of such polar compounds is not guaranteed with these membranes. ccumulation of estrogenic compounds dsorption of estrone on 21.2 cm 2 of TFC-SR2 and TFC-S membranes at saturation in this laboratory scale study was determined. The amount of estrone that can accumulate on commercial membrane modules when waters containing such a compound in water or wastewater treatment are treated can then be estimated for different commercial membrane modules. The results are tabulated in Table 2. Results reported here indicate that the membrane may adsorb and therefore accumulate a significant amount of estrone. Given the extreme endocrine-disrupting potency of natural hormones and the risk of accidental bulk release of these contaminants, adsorption of estrogenic compounds appears as a critical issue in water recycling practice using membrane technology. % Estrone adsorbed 3 1 24 6 8 1 12 p (-) 9 2 4 6 8 1 12 14 p (-) Permeate concentration (ngl -1 ) Figure 5 Estradiol adsorption, permeate concentration and retention by TFC-SR2 as a function of p ( ng/l estradiol; 1 mmol l 1 NaCO 3 and mmol l 1 NaCl) 9 TFC-S TFC-SR2 3 3 9 Permeate Volume (ml) Figure 6 Retention as a function of filtration volume ( ng/l estrone; 1 mmol l 1 NaCO 3, mmol l 1 NaCl and p 7.8) Table 2 ccumulation of estrone on different commercial membrane modules in µg (feed contained ng/l estrone, 1 mmol l 1 NaOCO 3 and mmol l 1 NaCl; p 7.8) Module size 2.5 inch 4 inch 8 inch Membrane area (approx.) 2 m 2 8 m 2 m 2 TFC-SR2 66.8 µg 267.2 µg 1,6 µg TFC-S 44.6 µg 178.4 µg 1,115 µg 159
Conclusions In this study, we investigated the adsorptive phenomena of natural hormones estrone and estradiol by two low-pressure nanofiltration membranes. The removal phenomena of estradiol was similar to that of estrone. p has been found to significantly influence the adsorption of estrone and estradiol by the membranes, presumably due to speciation and subsequent changes in hydrogen bonding ability of the molecules. While adsorption also occurs in the case of the TFC-S membrane, size exclusion appears to be a contributing factor to retention by this membrane. Further studies are planned to identify a membrane with a valid retention mechanism rather than adsorption, ideally in conjunction with a low adsorption capacity to reduce the risk of a bulk release of trace contaminants and a low pressure requirement. cknowledgements The Queensland Government and the ustralian Research Council are thanked for project funding. We acknowledge Koch Membrane Systems (San Diego, US) for providing membrane samples. Symbols : Membrane Surface [m 2 ] c : ulk Concentration [mg L 1 ] c P : Permeate Conc. [mg L 1 ] J: Flux [L m 2 h 1 ] R: Retention [%] t: Time [h] V: Permeate Volume [L] References Johnson,.C and Sumpter, J.P. (1). Removal of endocrine-disrupting chemicals in activated sludge treatment works. Environmental Science & Technology, 35, 4697 43. Kiso, Y., Kon, T., Kitao, T. and Nishimura, K. (1). Rejection properties of alkyl phthalates with nanofiltration membranes. Journal of Membrane Science, 182, 5 214. Kiso, Y., Nishimura, Y., Kitao, T. and Nishimura, K. (). Rejection properties of non-phenylic pesticides with nanofiltration membranes. Journal of Membrane Science, 171, 229 237. Merck,.S. (1996). Merck index. 12th Ed. Merck & Co., Inc, New Jersey. Schäfer,.I., Nghiem, D.L. and Waite, T.D. (3). Removal of natural hormone estrone from aqueous solutions using nanofiltration and reverse osmosis. Environmental Science & Technology, 37, 182 188. 1