Membrane adsorption of endocrine disrupting compounds and pharmaceutically active compounds

Membrane adsorption of endocrine disrupting compounds and pharmaceutically active compounds

Journal of Membrane Science 303 (2007) 267–277 Membrane adsorption of endocrine disrupting compounds and pharmaceutically active compounds Anna M. Co...

517KB Sizes 2 Downloads 38 Views

Journal of Membrane Science 303 (2007) 267–277

Membrane adsorption of endocrine disrupting compounds and pharmaceutically active compounds Anna M. Comerton a,∗ , Robert C. Andrews a , David M. Bagley b , Paul Yang c a

b

Department of Civil Engineering, University of Toronto, 35 St. George Street, Toronto, Ontario, Canada M5S 1A4 Department of Civil & Architectural Engineering, University of Wyoming, 1000 E. University Avenue, Laramie, WY 82071, United States c Applied Chromatography Section, Laboratory Services Branch, Ontario Ministry of the Environment, 125 Resources Rd. Etobicoke, Ontario, Canada M9P 3V6 Received 15 May 2007; received in revised form 13 July 2007; accepted 17 July 2007 Available online 22 July 2007

Abstract Adsorption is one of the main mechanisms contributing to compound removal by membrane filtration, in addition to size exclusion and charge repulsion. In this study, the adsorption of 22 endocrine disrupting compounds and pharmaceutically active compounds by ultrafiltration (UF), nanofiltration (NF) and reverse osmosis (RO) membranes was investigated using 24-h bottle tests at 21 and 4 ◦ C. Two natural waters (Lake Ontario and effluent from a membrane bioreactor (MBR)) and one laboratory-grade water were examined. Adsorption was strongly correlated with compound log Kow and membrane pure water permeability, and moderately correlated with compound water solubility. Adsorption was observed to be highest by the UF membrane followed by the NF and RO membranes. The influence of temperature on adsorption in the range examined was found to be insignificant. Three compounds for which deuterium-labelled surrogates were available (acetaminophen, carbamazepine, gemfibrozil) were examined to determine the influence of water matrix on adsorption. Adsorption of gemfibrozil may have been hindered due to competition for adsorption sites from the organic matter present in the lake water and MBR effluent. © 2007 Elsevier B.V. All rights reserved. Keywords: Endocrine disrupting compounds (EDCs); Pharmaceutically active compounds (PhACs); Membrane filtration; Adsorption; Organic matter

1. Introduction Recent improvements in analytical techniques have allowed the detection of trace levels (ng/L) of endocrine disrupting compounds (EDCs) and pharmaceutically active compounds (PhACs) in wastewater effluents [1–3], source waters [4,5] as well as treated drinking waters [6,7]. This has led to increasing concern over the potential adverse ecological and human health impacts resulting from the presence of EDCs and PhACs in the environment. In general, the removal of these compounds by conventional wastewater and drinking water treatment processes is not effective or well understood. It is therefore of interest to determine the ability of advanced treatment processes, such as membrane filtration, to remove these organic micropollutants. EDC and PhAC removal by membranes can vary significantly and is influenced by the physical–chemical properties of



Corresponding author. Tel.: +1 416 978 3220; fax: +1 416 978 3674. E-mail address: [email protected] (A.M. Comerton).

0376-7388/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.memsci.2007.07.025

the compounds, the membrane properties and the feed water characteristics. A recent study by Snyder et al. [8] included several long-term investigations of the removal of EDCs and PhACs from natural waters by membranes in pilot- and fullscale treatment plants. However, to date, most other studies investigating the removal of EDCs and PhACs by membranes have been conducted using laboratory-grade water spiked with individual compounds in the absence of organic matter and at concentrations higher than those found in the environment (␮g/L versus ng/L). In addition, many studies have been conducted over short periods of time (<24 h) and using small volumes of water (<10 L). These factors may lead to overestimation of membrane removal efficiency due to the high initial rate of compound adsorption to the membrane [9–11]. Initial removal via adsorption eventually stabilizes when equilibrium is achieved at which point other mechanisms contribute to compound removal (e.g. size exclusion, charge repulsion). In fact, after equilibrium has been reached, adsorption may adversely impact retention because it has been shown that adsorbed compounds can dissolve into the membrane active layers, then diffuse through the

268

A.M. Comerton et al. / Journal of Membrane Science 303 (2007) 267–277

polymer, and finally desorb at the permeate side of the membrane [12]. In addition, desorption of adsorbed compounds to the membrane’s permeate side may occur when the compound concentration in the feed water is lower than the equilibrium concentration [13]. Therefore, the investigation of adsorption is important to improve the understanding of membrane removal mechanisms. Adsorption is expected to be driven by hydrophobic and hydrogen bonding [10]. Kiso et al. [14] found that the rejection of pesticides by nanofiltration and reverse osmosis membranes was correlated with log Kow , a measure of compound hydrophobicity in its neutral state. Yoon et al. [9] examined the retention of 52 EDCs and PhACs by ultrafiltration and nanofiltration during filtration under non-equilibrium conditions and observed that removal was highest for hydrophobic compounds. Molecular polarity can also significantly influence compound adsorption to membranes. The dipole of polar compounds can be directed towards the membrane’s oppositely charged surface allowing the compound to enter more easily into the membrane structure [15,16]. Compound adsorption is also related to membrane pore size. Membranes with larger pores allow compounds to access adsorption sites in the membrane’s skin layer, support layer and pores in addition to its surface, whereas access to these internal sites may be limited with tighter membranes [17,18]. Membrane surface roughness has been attributed to an increase in colloid attachment on the membrane surface and in turn an increase in fouling. The greater the surface roughness, the larger the membrane surface area, leading to more opportunities for particle contact [19–21]. It is expected that organic micropollutants will exhibit similar behaviour and adsorb more readily to membranes with rough surfaces. Very little research has focussed on the influence of organic matter in natural waters on EDC and PhAC removal by membranes. Yoon et al. [17] reported that the presence of natural organic matter (NOM) did not significantly change the adsorption of 17␤-estradiol (E2) and fluoranthene to UF and NF membranes in batch adsorption tests. However, a more recent study reported a decrease in EDC and PhAC adsorption to the membrane and postulated that this was due to competition from the NOM for membrane adsorption sites [9]. Therefore, there is a need to investigate the influence of organic matter on the adsorption of organic micropollutants. The objective of this study was to examine the adsorption of EDCs and PhACs at concentrations typically found in the environment (ng/L range) as a function of compound characteristics and membrane properties. The study also investigated the influence of the presence of organic matter in natural waters as well as water temperature on the adsorption of EDCs and PhACs. 2. Experimental 2.1. Compound selection and characterization Twenty-two endocrine disrupting compounds and pharmaceutically active compounds were selected to be representative of various classes (e.g. pesticides, antibiotics, hormones) of organic micropollutants found in wastewater and in drinking

water sources. These compounds also represent a range of properties (i.e. solubility, hydrophobicity/hydrophilicity, polarity) that are expected to influence compound adsorption by membranes. A summary of the EDCs and PhACs examined and their properties is presented in Table 1. 2.2. Membrane selection and characterization Membranes examined in this experiment included the UE10 (TriSep, UE10, Goleta, CA) polysulfone ultrafiltration (UF) membrane, the NF270 (Dow Chemical Co., Filmtec NF270400, Midland, MI) and TS80 (TriSep, 4040-TS80-TSF, Goleta, CA) polyamide nanofiltration (NF) membranes, and the X20 (TriSep, 4040-X201-TSF, Goleta, CA) polyamide reverse osmosis (RO) membrane. The characteristics of these membranes are presented in Table 2. Pure water permeability (PWP) and molecular weight cut-off (MWCO) associated with each membrane were determined using a commercially available stainless steel crossflow membrane filtration unit (Sepa CF II med/high foulant system, Sterlitech Corp., Kent, WA) equipped with 17 mil diamond spacers. PWP provides an indication of the maximum flux that can be achieved by the membrane and was determined by taking the slope of the average flux of Milli-Q® water through the membrane as measured over a range of feed pressures. The MWCO of a membrane represents the molecular weight of a molecule that is rejected at 90%. The MWCO of each membrane was estimated using the solute transport technique described by Singh et al. [22] via rejection tests using solutions containing polyethylene glycol (PEG) molecules (87976, Sigma–Aldrich, Oakville, ON) of varying molecular weights (200, 400, 600, 1000, 2000 Da) but at a constant solute concentration of 10 ppm by weight. Membrane zeta potential provides a measure of the electrical charge of the membrane surface and was measured using a SurPASS electrokinetic analyzer (Anton Paar, Graz, Austria) following streaming current methodology described by Childress and Elimelech [23]. Membrane contact angle is an index of the hydrophilicity/hydrophobicity of a membrane surface. A contact angle of less than 90◦ indicates that the membrane surface is hydrophilic, whereas a contact angle above 90◦ means that the membrane surface is hydrophobic. Therefore, the higher the contact angle, the more hydrophobic the membrane surface. The static contact angle of dry membrane samples was measured in triplicate via the sessile drop technique, described by Chen and Wada [24], with Milli-Q® water (drop volume of approx. 10 ␮L) using a First Ten Angstroms dynamic contact angle analyzer (FTA200, Folio Instruments Inc., Kitchener, ON). Surface characterization of the membranes was performed using a Bioscope Atomic Force Microscope (Digital Instruments VEECO, Santa Barbara, CA) in tapping mode AFM [25]. A silicon cantilever (Tap300, NanoDevices Inc., Santa Barbara, CA) with a spring constant of 40 N/m and a nominal tip apex radius of 10 nm was used to obtain images at a 300 kHz resonant frequency. Membrane samples were air dried and attached to a microscope slide using double-sided tape. Root mean square (RMS) roughness and mean roughness (Ra ) were determined in triplicate over a scan size of 3 ␮m × 3 ␮m using the data

A.M. Comerton et al. / Journal of Membrane Science 303 (2007) 267–277

269

Table 1 Investigated EDCs and PhACs and their properties Compound

Classification

Molecular weight (g/mol)

Water solubilitya (mg/L)

log Kow a

Dipole momentb (μ) (D)

Spiked concentration (C0 ) (ng/L)

Acetaminophen Alachlor (Lasso) Atraton Bisphenol A (BPA) Caffeine Carbadox Carbamazepine DEET Diethylstilbesterol Equilin 17␣-Estradiol 17␤-Estradiol (E2) Estriol (E3) Estrone (E1) 17␣-Ethynyl estradiol (EE2) Gemfibrozil Metolachlor Oxybenzone Sulfachloropyridazine Sulfamerazine Sulfamethizole Sulfamethoxazole

PhAC (antipyretic) EDC (pesticide) EDC (pesticide) EDC (plasticizer) PhAC (stimulant) PhAC (antibiotic) PhAC (antiepileptic/antidepressant) EDC (insect repellent) EDC (estrogen replacement) EDC (estrogen replacement) EDC (natural hormone) EDC (natural hormone) EDC (natural hormone) EDC (metabolite of E2) EDC (ovulation inhibitor)

151 270 211 228 194 262 236 191 268 268 272 272 288 270 296

14000 240 1800 120 21600 15000 17.7 449 12 1.41 3.9 3.6 441 30 11.3

0.46 3.52 2.69 3.32 −0.07 −1.37 2.45 2.38 5.07 3.35 3.94 4.01 2.45 3.13 3.67

1.38 3.64 1.71 1.67 1.04 3.31 1.67 1.93 1.62 2.33 2.75 2.75 3.22 2.04 2.64

560 500 500 1035 500 560 680 500 530 490 525 505 470 545 550

PhAC (lipid regulator) EDC (pesticide) EDC (sunscreen) PhAC (antibiotic) PhAC (antibiotic) PhAC (antibiotic) PhAC (antibiotic)

250 284 228 285 278 270 253

10.9 530 68.6 7000 1500 1050 610

4.77 3.13 3.79 0.31 0.89 0.54 0.89

3.57 2.75 3.27 1.73 1.28 0.93 2.12

685 500 500 435 725 810 715

a b

Obtained from the Syracuse Research Corporation (SRC) PhysProp database (http://www.syrres.com/esc/physdemo.htm). Estimated using the chemical modelling software, HyperChem (7.5 student edition, Hypercube Inc., Gainsville, FL).

analysis software provided with the Digital Instruments system (Nanoscope 5.30r3). RMS is defined as the standard deviation of the measured membrane surface height values (Z-values) within a given area. Ra is the arithmetic average of the surface height deviations from the centre plane, for which the volume enclosed by the image above and below this plane is equal [22]. Measured roughness values are presented in Table 2. 2.3. Water selection and characterization Two natural waters (Lake Ontario and MBR effluent) and one laboratory-grade water (Milli-Q® ) were investigated in this study. Lake Ontario is a source of drinking water for millions of people living in Southern Ontario and New York State. Although EDC and PhAC levels in Lake Ontario have not been studied specifically, detectable levels of various EDCs and PhACs have been reported in many surface waters around the world [4–7] and are expected to be typical of most surface waters in populated areas. Lake Ontario water was collected from influent

to the Ajax Water Supply Plant (Ajax, Ontario, Canada), which serves a population of approximately 175,000. Effluent from the membrane bioreactor (MBR) of a municipal wastewater treatment plant (WWTP) in Port McNicoll, Ontario, Canada was also obtained. The MBR system incorporates ZeeWeed® 500 ultrafiltration membranes (Oakville, Ontario, Canada) and produces approximately 200 m3 of permeate per day while operating at a mixed liquor suspended solids concentration of between 12 and 15 g/L. EDCs and PhACs are expected to be present in municipal wastewater; many will not be completely removed by the MBR process, particularly if they are not easily degradable [26,27]. Finally, laboratory-grade (Milli-Q® ) water was used as a control at neutral pH (adjusted with NaOH or HCl), buffered with sodium bicarbonate (1 mM) and containing 3 mM of NaCl to provide background electrolytes [15]. The pH (A-32908-06, Labcor Technical Sales Inc., Concord, ON), conductivity and total dissolved solids (TDS) (23226-523, VWR, Mississauga, ON), total and dissolved organic carbon (TOC/DOC), and specific UVA254 (SUVA) were measured

Table 2 Membrane characteristics Type

Membrane

PWP (×10−5 L/m2 h Pa (L/m2 h bar))

MWCO (Da)

Zeta potentiala (mV)

Contact angle (◦ )

UF NF NF RO

UE10 NF270 TS80 X20

32c 15.5 ± 1.5 4.5 ± 0.3 2.5 ± 0.5

10000a 400 <200 <200

−92 −87 −15 −87

49.3 29.8 56.6 55.0

a b c

Streaming current measurement at pH 6 in 10−3 M KCl solution. Based on a membrane scan size of 3 ␮m × 3 ␮m. Based on the manufacturer’s specifications.

± ± ± ±

3.5 0.3 2.0 1.8

RMSb (nm) 10.0 6.8 89.1 81.0

± ± ± ±

2.5 1.6 20.1 15.7

Ra b (nm) 7.8 5.3 69.4 63.4

± ± ± ±

1.9 1.3 15.2 11.3

270

A.M. Comerton et al. / Journal of Membrane Science 303 (2007) 267–277

Table 3 Water characteristics Parameter

pH Conductivity (␮S/cm) TDS (ppm) TOC (mg/L) DOC (mg/L) SUVA (L/mg m) %Hydrophobic %Hydrophilic %Transphilic a b

Water Lake Ontario

MBR effluent

Milli-Q®

7.99 298 194 2.5 2.5 1.40 48.8 ± 5.6 34.8 ± 2.5 16.4 ± 3.2

7.88 614 382 3.7 3.7 0.77 57.9 ± 5.2 25.5 ± 3.7 16.6 ± 2.0

7.0 ± 0.1 480 ± 5 320 ± 3 0.3 0.3 NDa NMb NM NM

ND: not detected. NM: not measured.

for each type of water. The humic content or aromaticity of organic matter present in a water can be determined by SUVA (=UVA254 /DOC) [28]. DOC is the fraction of TOC in a sample that passes through a filter with a pore size of 0.45 ␮m (Gelman Supor, Gelman Sciences, Ann Arbor, MI). UVA254 gives an indication of the presence of UV absorbing compounds, such as aromatics or compounds that have conjugated double bonds, which are characteristic of humic substances. UVA254 was analysed using a diode array spectrophotometer (Hewlett Packard 8452A, Mississauga, ON) using a wavelength of 254 nm and a quartz crystal cuvette with a 1 cm path length. In addition, the hydrophobic, hydrophilic and transphilic organic matter fractions of each water were determined following the method developed by the U.S. Geological Survey [29] and described by Standard Method 5510 C [30]. Specific characteristics of the waters are presented in Table 3. 2.4. 24-H bottle tests Bottle tests were performed in triplicate to estimate the extent of individual compound adsorption for each type of membrane (UE10, NF270, TS80, X20) and water type (Lake Ontario, MBR effluent, Milli-Q® ). Membrane coupons (19 cm × 14 cm), cut into small pieces (approx. 4 cm × 4 cm), were placed in 1-L amber bottles filled with a specific water and spiked with the 22 EDCs/PhACs at a dosage of approximately 500 ng/L (see Table 1). Controls for each water type consisting of either Lake Ontario water, MBR effluent or Milli-Q® water only (no membrane) and spiked with the EDCs/PhACs were performed in triplicate and included to correct for any influence from the water matrix itself as well as any potential compound losses due to adsorption on the bottle walls. The bottles were mixed for 24 h using a bottle inverter (∼0.833 Hz (∼5 rpm)) at two temperature conditions to be representative of extremes experienced in Canada: 21 and 4 ◦ C. After 24 h, the membranes were removed from the bottles and approximately 2 mL of a 25% sodium thiosulphate solution were added to preserve the samples until analysis. A study by Chang et al. [13], investigating the adsorption of estrone to hollow microfiltration membranes in 24 h batch adsorption tests, determined that equilibrium was reached within 5 h. Similarly, a 72-h batch

Fig. 1. 72-H batch adsorption test results for the NF270 membrane in Milli-Q water (not all data shown).

adsorption test incorporating all investigated EDCs and PhACs with the NF270 membrane in Milli-Q® water confirmed that equilibrium was reached within 24 h (see Fig. 1). 2.5. Analytical methods PhACs (acetaminophen, carbadox, carbamazepine, gemfibrozil, sulfachloropyridazine, sulfamerazine, sulfamethizole, sulfamethoxazole) and EDCs (bisphenol A, diethylstilbesterol, equilin, 17␣-estradiol, 17␤-estradiol, estriol, estrone, 17␣-ethynyl estradiol) were analysed following a liquid chromatography/tandem mass spectrometry (LC/MS/MS) method [31] developed by the Ontario Ministry of the Environment (MOE) (Toronto, Ontario, Canada). All analyses were conducted at the MOE laboratory using an Applied Biosystems LC/MS/MS system (MDS SCIEX 4000QTRAP) with an Agilent LC (HP1100, Mississauga, ON), equipped with a Hypersil Gold guard column (2.1 mm i.d. × 5 mm) and analytical column (2.1 mm i.d. × 100 mm). Method detection limits (MDLs) for these compounds ranged from 2 to 64 ng/L. Triazine pesticides (alachlor, atraton, metolachlor), caffeine, DEET, and oxybenzone analyses were conducted following a gas chromatography/time of flight/mass spectrometry (GC/TOF/MS) method [32] developed by the MOE. All of these analyses were conducted using a Hewlett Packard 6890 Gas Chromatograph (Mississauga, ON), equipped with on-column injector and electronic pressure control (EPC) options, a Hewlett Packard 7673 autosampler, a LECO Pegasus II TOF Mass Spectrometer, and a DB-5MS capillary column (40 m, 0.18 mm i.d., 0.18 ␮m (df)). MDLs for these compounds ranged from 25 to 60 ng/L. Total organic carbon (TOC) was analysed using an O-I Corporation Model 1010 Analytical TOC Analyzer (College Station, Texas), following the wet oxidation method described in Standard Method 5310 D [30]. The MDL for TOC was 0.2 mg/L. 3. Results and discussion Measured EDC and PhAC concentrations (C) following the 24-h bottle tests were compared with the initial concentrations

A.M. Comerton et al. / Journal of Membrane Science 303 (2007) 267–277

271

Table 4 Adsorption of EDCs and PhACs (C/C0 ) from Milli-Q® water at 21 ◦ Ca Compound

Acetaminophen Alachlor (Lasso) Atraton Bisphenol A Caffeine Carbadox Carbamazepine DEET Diethylstilbesterol Equilin 17␣-Estradiol 17␤-Estradiol Estriol Estrone 17␣-Ethynyl estradiol Gemfibrozil Metolachlor Oxybenzone Sulfachloropyridazine Sulfamerazine Sulfamethizole Sulfamethoxazole a b

Membrane X20

TS80

NF270

UE10

None

0.81 ± 0.19 0.78 ± 0.06b 0.95 ± 0.13 0.91 ± 0.02 1.15 ± 0.19 1.04 ± 0.12 0.82 ± 0.13 1.03 ± 0.15 0.79 ± 0.03 0.79 ± 0.10 0.84 ± 0.04 0.94 ± 0.20 0.68 ± 0.04 0.75 ± 0.13 0.91 ± 0.07 0.81 ± 0.10 0.81 ± 0.11 0 0.88 ± 0.20 0.92 ± 0.13 1.31 ± 0.09 1.13 ± 0.01

0.80 ± 0.04 0.57 ± 0.03 0.84 ± 0.06 0.84 ± 0.04 0.98 ± 0.11 1.14 ± 0.09 0.87 ± 0.03 0.90 ± 0.07 0.40 ± 0.05 0.43 ± 0.04 0.69 ± 0.04 0.68 ± 0.02 0.44 ± 0.03 0.39 ± 0.03 0.59 ± 0.06 0.89 ± 0.01 0.63 ± 0.03 0 0.94 ± 0.03 0.84 ± 0.02 1.14 ± 0.09 1.17 ± 0.02

0.74 ± 0.08 0.51 ± 0.03 0.82 ± 0.12 0.84 ± 0.05 0.94 ± 0.36 1.02 ± 0.09 0.68 ± 0.00 0.91 ± 0.12 0.29 ± 0.04 0.32 ± 0.05 0.49 ± 0.06 0.44 ± 0.06 0.57 ± 0.06 0.31 ± 0.06 0.39 ± 0.11 0.72 ± 0.06 0.54 ± 0.08 0 1.05 ± 0.05 0.90 ± 0.03 1.61 ± 0.03 1.27 ± 0.02

0.93 ± 0.24 0.11 ± 0.01 0.50 ± 0.03 0 0.92 ± 0.02 0.96 ± 0.14 0.91 ± 0.05 0.37 ± 0.03 0 0 0.01 ± 0.01 0 0.42 ± 0.08 0 0 0.17 ± 0.02 0.08 ± 0.07 0 0.70 ± 0.04 0.88 ± 0.05 1.41 ± 0.15 0.89 ± 0.03

1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00

± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

0.12 0.06 0.10 0.02 0.08 0.11 0.00 0.10 0.20 0.03 0.07 0.06 0.01 0.04 0.07 0.04 0.08 0.04 0.20 0.09 0.14 0.11

Normalized C/C0 values are reported. See Table 1 for C0 values. C/C0 values in bold are significantly different from control with no membrane (95% confidence).

(C0 ) spiked into each bottle. Values for a given water type were normalized by dividing each C/C0 value by the C/C0 value for that water’s corresponding control (membrane type ‘None’) to account for compound loss due to adsorption onto the bottle or any contribution from the water itself. Therefore, a normalized C/C0 value of 1.0 indicated no adsorption by the membrane whereas a value of 0 indicated complete adsorption by the membrane. Adsorption within a water type was compared using the Tukey method for pairwise comparisons at a confidence level of 95% [33]. The Tukey method is best suited for situations where confidence intervals are required only for pairwise difference contrasts. For each water type examined in the bottle tests, Tukey’s simultaneous confidence intervals for all pairwise comparisons τ i − τ j , i = j, with an overall confidence level of at least 100(1 − α)% is given by:     1 qv,n−v,α 1 √ (1) MSerror + (¯yi − y¯ j ) ± ri rj 2 where n (observations) = 15, v (treatments (membrane conditions)) = 5, r (replicates) = 3, α (0.05) = 95% confidence level, qv,n–v,a (q5,10,0.05 ) = 4.65. v r ¯ i. )2 i=1 j=1 (yij − y (2) MSerror = v(r − 1) The null hypothesis, Ho :

v  i=1

ci τi = 0

(3)

is rejected, and compared C/C0 values are deemed significantly different, if the contrast ci τ i is larger than the minimum significance difference (i.e. the confidence interval excludes zero). The normalized C/C0 values for each membrane from MilliQ® water, Lake Ontario water and MBR effluent at 21 ◦ C are presented in Tables 4–6, respectively. Normalized C/C0 values that are significantly different from the control with no membrane (95% confidence) are highlighted in bold. 3.1. Influence of compound characteristics on adsorption Correlations between normalized C/C0 values at 21 ◦ C and four compound characteristics (log Kow , water solubility, dipole moment and molecular weight), determined via linear regression analysis using a commercial statistical software package (Design-Expert 7.1.2, Stat-Ease, Inc., Minneapolis, MN), are shown in Table 7. Correlations that were determined to be statistically significant at the 95% confidence level (p < 0.05) are highlighted in bold. The results from the correlation analysis show that log Kow values have the most impact on adsorption. In particular, a strong correlation (r > 0.80) between log Kow and adsorption by the UE10 membrane exists for all water types (see Fig. 2). It should be noted that the adsorption of some compounds, gemfibrozil (log Kow = 4.77, pKa = 4.7) and carbamazepine (log Kow = 2.45, pKa = 3.6) in particular, was observed to be lower than expected based on their hydrophobicity (log Kow ) alone. This may be a result of charge repulsion caused by deprotonation due a higher water pH than the compound acid dissociation constant (pKa ) value [15].

272

A.M. Comerton et al. / Journal of Membrane Science 303 (2007) 267–277

Table 5 Adsorption of EDCs and PhACs (C/C0 ) from Lake Ontario water at 21 ◦ Ca Compound

Acetaminophen Alachlor (Lasso) Atraton Bisphenol A Caffeine Carbadox Carbamazepine DEET Diethylstilbesterol Equilin 17␣-Estradiol 17␤-Estradiol Estriol Estrone 17␣-Ethynyl estradiol Gemfibrozil Metolachlor Oxybenzone Sulfachloropyridazine Sulfamerazine Sulfamethizole Sulfamethoxazole a b

Membrane X20

TS80

NF270

UE10

None

0.96 0.76 0.88 0.73 0.69 0.97 0.82 0.86 0.33 0.65 0.72 0.73 0.82 0.65 0.65 0.90 0.82 0.15 0.71 0.82 0.90 0.88

0.81 ± 0.05b

0.63 ± 0.05 0.47 ± 0.09 0.95 ± 0.14 0.59 ± 0.10 0.83 ± 0.20 0.80 ± 0.03 0.35 ± 0.03 0.79 ± 0.11 0.46 ± 0.05 0.23 ± 0.06 0.37 ± 0.09 0.31 ± 0.07 1.03 ± 0.05 0.23 ± 0.07 0.30 ± 0.11 0.76 ± 0.08 0.63 ± 0.14 0 0.58 ± 0.03 0.53 ± 0.04 0.39 ± 0.03 0.57 ± 0.03

1.03 ± 0.09 0.07 ± 0.06 0.42 ± 0.07 0 0.69 ± 0.16 0.97 ± 0.14 0.66 ± 0.04 0.31 ± 0.06 0 0 0 0 0.38 ± 0.02 0 0 0.52 ± 0.04 0.11 ± 0.01 0.02 ± 0.03 0.97 ± 0.02 0.77 ± 0.03 0.85 ± 0.06 0.77 ± 0.03

1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00

± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

0.01 0.07 0.08 0.07 0.16 0.01 0.03 0.06 0.04 0.01 0.03 0.03 0.18 0.06 0.05 0.06 0.04 0.03 0.04 0.03 0.02 0.02

0.58 ± 0.03 0.89 ± 0.04 0.64 ± 0.07 0.84 ± 0.11 0.88 ± 0.08 0.41 ± 0.02 0.80 ± 0.03 0.66 ± 0.05 0.37 ± 0.04 0.55 ± 0.04 0.51 ± 0.06 0.86 ± 0.05 0.31 ± 0.03 0.47 ± 0.06 0.89 ± 0.07 0.72 ± 0.02 0 0.55 ± 0.03 0.49 ± 0.03 0.46 ± 0.08 0.55 ± 0.03

± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

0.01 0.06 0.24 0.03 0.29 0.15 0.06 0.14 0.06 0.02 0.11 0.08 0.07 0.04 0.10 0.01 0.03 0.04 0.09 0.06 0.13 0.06

Normalized C/C0 values are reported. See Table 1 for C0 values. C/C0 values in bold are significantly different from control with no membrane (95% confidence).

Although the correlations between water solubility and adsorption by the UE10 membrane are not strong, they are statistically significant (95% confidence). The influence of water solubility is exemplified by the adsorption of equilin (water

solubility = 1.41 mg/L) by all membranes for all waters when compared to the lack of significant adsorption of caffeine (water solubility = 21,600 mg/L) under any condition (see Tables 4–6). Solubility is moderately correlated with log Kow (r = −0.68,

Table 6 Adsorption of EDCs and PhACs (C/C0 ) from MBR effluent at 21 ◦ Ca Compound

Membrane X20

Acetaminophen Alachlor (Lasso) Atraton Bisphenol A Caffeine Carbadox Carbamazepine DEET Diethylstilbesterol Equilin 17␣-Estradiol 17␤-Estradiol Estriol Estrone 17␣-Ethynyl estradiol Gemfibrozil Metolachlor Oxybenzone Sulfachloropyridazine Sulfamerazine Sulfamethizole Sulfamethoxazole a b

0.93 0.94 0.95 0.86 0.95 0.65 0.90 1.01 0.35 0.69 0.77 0.62 1.08 0.81 0.81 0.79 0.86 0.67 0.88 0.81 0.83 0.88

± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

0.03 0.25 0.03 0.05 0.01 0.03 0.05 0.06 0.18 0.07 0.06 0.14 0.31 0.06 0.05 0.02 0.15 0.59 0.08 0.06 0.04 0.05

TS80

NF270

UE10

None

0.80 ± 0.20 0.64 ± 0.02b 0.98 ± 0.08 0.74 ± 0.09 1.06 ± 0.03 0.83 ± 0.19 0.77 ± 0.00 0.98 ± 0.10 0.24 ± 0.03 0.39 ± 0.00 0.61 ± 0.04 0.55 ± 0.02 0.84 ± 0.23 0.48 ± 0.10 0.46 ± 0.10 0.87 ± 0.05 0.69 ± 0.03 0 0.75 ± 0.00 0.73 ± 0.02 0.74 ± 0.03 0.76 ± 0.02

0.98 ± 0.03 0.61 ± 0.05 0.92 ± 0.09 0.87 ± 0.05 1.02 ± 0.11 0.67 ± 0.07 0.79 ± 0.04 0.98 ± 0.11 0.09 ± 0.02 0.29 ± 0.04 0.47 ± 0.05 0.40 ± 0.04 0 0.34 ± 0.06 0.30 ± 0.02 0.81 ± 0.05 0.57 ± 0.05 0.02 ± 0.03 0.75 ± 0.06 0.72 ± 0.03 0.75 ± 0.05 0.77 ± 0.03

0.80 ± 0.24 0.11 ± 0.00 0.57 ± 0.06 0.04 ± 0.05 0.96 ± 0.09 0.78 ± 0.20 0.81 ± 0.12 0.40 ± 0.02 0.01 ± 0.01 0.03 ± 0.05 0.04 ± 0.04 0.01 ± 0.01 0.68 ± 0.24 0.02 ± 0.04 0.05 ± 0.08 0.40 ± 0.11 0.14 ± 0.03 0 0.95 ± 0.06 0.75 ± 0.05 0.89 ± 0.11 0.77 ± 0.05

1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00

Normalized C/C0 values are reported. See Table 1 for C0 values. C/C0 values in bold are significantly different from control with no membrane (95% confidence).

± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

0.12 0.05 0.09 0.01 0.12 0.28 0.06 0.05 0.01 0.02 0.01 0.16 0.54 0.04 0.15 0.04 0.05 0.15 0.29 0.03 0.03 0.20

A.M. Comerton et al. / Journal of Membrane Science 303 (2007) 267–277

273

Table 7 Correlations between normalized C/C0 and compound properties (21 ◦ C) Compound property

Water matrix Milli-Q®

Lake Ontario

MBR effluent

ra

pb

r

p

r

p

X20 membrane log Kow Water solubility (mg/L) Dipole moment (D) Molecular weight (g/mol)

−0.55c 0.41 −0.49 0.08

0.008 0.136 0.020 0.738

−0.50 0.24 −0.11 −0.14

0.018 0.290 0.636 0.524

−0.34 0.14 −0.10 −0.30

0.117 0.545 0.659 0.177

TS80 membrane log Kow Water solubility (mg/L) Dipole moment (D) Molecular weight (g/mol)

−0.67 0.41 −0.39 −0.54

0.001 0.060 0.076 0.009

−0.25 0.45 −0.01 −0.30

0.254 0.035 0.958 0.174

−0.53 0.43 −0.27 −0.36

0.011 0.044 0.231 0.103

NF270 membrane log Kow Water solubility (mg/L) Dipole moment (D) Molecular weight (g/mol)

−0.72 0.33 −0.46 −0.05

0.030 0.129 0.030 0.816

−0.37 0.40 −0.05 −0.27

0.088 0.067 0.819 0.225

−0.51 0.44 −0.47 −0.59

0.015 0.038 0.028 0.004

UE10 membrane log Kow Water solubility (mg/L) Dipole moment (D) Molecular weight (g/mol)

−0.87 0.54 −0.53 −0.22

<0.001 0.010 0.011 0.333

−0.85 0.61 −0.39 −0.29

<0.001 0.003 0.076 0.184

−0.84 0.58 −0.45 −0.28

<0.001 0.005 0.035 0.215

a b c

r: Pearson’s product-moment coefficient. p: Statistical significance of correlation. Correlations in bold are statistically significant (p < 0.05) (95% confidence).

p = 0.002) so it is expected that, in general, adsorption will increase with decreasing water solubility values and increasing compound hydrophobicity, as represented by log Kow . Finally, a weak (r ∼ = −0.5) but statistically significant correlation between dipole moment and adsorption is observed for the Milli-Q® water. Several studies have shown that the dipole of polar compounds can be directed towards the membrane’s oppositely charged surface allowing the compound to enter more easily into the membrane structure [15,16]. This may explain, for example, why although alachlor (μ = 3.64) and bisphenol A (μ = 1.67) have similar water solubility (240 mg/L versus 120 mg/L) and

Fig. 2. EDC/PhAC adsorption to the UE10 membrane as a function of log Kow .

log Kow (3.52 versus 3.32) values, adsorption of alachlor is greater than that of bisphenol A. 3.2. Influence of membrane properties on adsorption In general, EDC/PhAC adsorption by the membranes was observed in the following order: UE10 > NF270 ≈ TS80 > X20. Adsorption of the estrogen hormones equilin, 17␣-ethynyl estradiol, 17␣-estradiol, and 17␤-estradiol, as shown in Fig. 3, presents a good example of the differences in adsorption between membranes. Pure water permeability provides an indication of the maximum flux that can be achieved by the membrane. As expected, PWP is proportional to membrane pore size with the X20 RO membrane having the lowest flux and the UE10 UF membrane offering the highest flux. This is also supported by the molecular weight cut-off of the membranes which show an increase in flux with increasing MWCO values (see Table 2). There is a strong correlation between the normalized C/C0 values of the four estrogen hormones shown in Fig. 3 and membrane PWP (r = −0.95, p < 0.001). Sorption studies have shown that solute adsorption is not restricted to the membrane surface but can also occur in the membrane skin layer, the membrane support layer, and the membrane pores. Distinguishing between the adsorption occurring in the various parts of the membrane structure is difficult [17,34]. Yoon et al. [17] and Combe et al. [18] observed that adsorption is related to pore radius. Membranes with larger pore sizes allow solute to access the membrane’s internal adsorp-

274

A.M. Comerton et al. / Journal of Membrane Science 303 (2007) 267–277

Fig. 3. Membrane adsorption of estrogen hormones at 21 ◦ C (error bars represent standard deviation of three replicates).

tion sites, whereas access to these internal sites may be limited with tighter membranes. Therefore, the more porous UE10 UF membrane (PWP = 32 × 10−5 L/m2 h Pa (32 L/m2 h bar)) may allow more compound adsorption within its structure in addition to its surface when compared to the X20 RO membrane, which is essentially poreless (PWP = 2.5 × 10−5 L/m2 h Pa (2.5 L/m2 h bar)). An increase in pore size is expected to result in more access to the membrane’s internal adsorption sites and, in turn, result in an increase in compound adsorption. No statistically significant correlations were observed between adsorption and the other membrane properties measured. 3.3. Influence of water matrix on adsorption Three deuterium-labelled surrogate compounds (d4acetaminophen, d10-carbamazepine, d6-gemfibrozil) were available and added to each sample before EDC/PhAC analysis. Surrogate compounds are expected to behave similarly to their parent compound (i.e. acetaminophen, carbamazepine,

gemfibrozil) during the analytical process and are well detected by the analytical instrumentation. Using the analytical recovery of the surrogate compound in each sample, a correction can be made for the parent compound which takes into account differences in recovery attributable to analytical error in an individual sample or due to water matrix effects. For compounds without surrogates, analytical recovery of all the samples within a sample batch (10 samples) is assumed to be the same as that of laboratory-grade water spiked with a known amount of the EDCs and PhACs analysed and extracted with the sample batch. Therefore, results for compounds corrected with the recovery of their surrogates are expected to be more accurate because they take into account any analytical error and any water matrix effects attributed to that specific sample. Adsorption results for acetaminophen, carbamazepine and gemfibrozil corrected with their surrogates at 21 and 4 ◦ C are presented in Tables 8 and 9, respectively, and can be compared with the results obtained without surrogate correction (at 21 ◦ C, see Tables 4–6). C/C0 values that are significantly different from

Table 8 Adsorption of compounds (C/C0 ) at 21 ◦ C corrected with their surrogatesa Compound

Membrane X20

TS80

NF270

UE10

None

Milli-Q water Acetaminophen Carbamazepine Gemfibrozil

0.93 ± 0.12 0.87 ± 0.11 0.82 ± 0.10

0.85 ± 0.04 0.83 ± 0.03b 0.84 ± 0.04

0.85 ± 0.03 0.68 ± 0.02 0.67 ± 0.06

0.92 ± 0.07 0.70 ± 0.05 0.16 ± 0.01

1.00 ± 0.03 1.00 ± 0.02 1.00 ± 0.04

Lake Ontario water Acetaminophen Carbamazepine Gemfibrozil

0.93 ± 0.04 0.93 ± 0.15 0.88 ± 0.02

0.95 ± 0.02 0.77 ± 0.01 0.97 ± 0.05

0.92 ± 0.03 0.64 ± 0.04 0.78 ± 0.05

0.99 ± 0.04 0.66 ± 0.02 0.54 ± 0.02

1.00 ± 0.03 1.00 ± 0.03 1.00 ± 0.04

MBR effluent Acetaminophen Carbamazepine Gemfibrozil

0.97 ± 0.03 0.76 ± 0.02 0.99 ± 0.04

0.85 ± 0.17 0.81 ± 0.19 0.92 ± 0.04

0.96 ± 0.01 0.64 ± 0.01 0.94 ± 0.03

0.93 ± 0.29 0.64 ± 0.16 0.40 ± 0.09

1.00 ± 0.09 1.00 ± 0.15 1.00 ± 0.05

a b

Normalized C/C0 values are reported. See Table 1 for C0 values. C/C0 values in bold are significantly different from control with no membrane (95% confidence).

A.M. Comerton et al. / Journal of Membrane Science 303 (2007) 267–277

275

Table 9 Adsorption of compounds (C/C0 ) at 4 ◦ C corrected with their surrogatesa Compound

Membrane X20

TS80

NF270

UE10

None

Milli-Q water Acetaminophen Carbamazepine Gemfibrozil

1.18 ± 0.07 1.10 ± 0.03 1.37 ± 0.03

1.05 ± 0.21 1.92 ± 0.85 1.23 ± 0.16

0.98 ± 0.11 0.80 ± 0.07 1.03 ± 0.06

0.97 ± 0.03 0.93 ± 0.03 0.09 ± 0.06b

1.00 ± 0.03 1.00 ± 0.04 1.00 ± 0.06

Lake Ontario water Acetaminophen Carbamazepine Gemfibrozil

0.69 ± 0.18 0.87 ± 0.04 0.80 ± 0.11

0.86 ± 0.26 0.85 ± 0.07 0.87 ± 0.13

0.53 ± 0.00 0.70 ± 0.01 0.71 ± 0.04

0.70 ± 0.21 0.32 ± 0.02 0.25 ± 0.02

1.00 ± 0.03 1.00 ± 0.10 1.00 ± 0.04

MBR effluent Acetaminophen Carbamazepine Gemfibrozil

1.04 ± 0.03 0.85 ± 0.02 0.93 ± 0.01

0.68 ± 0.05 0.98 ± 0.06 1.06 ± 0.05

0.82 ± 0.18 0.72 ± 0.13 0.85 ± 0.05

0.90 ± 0.02 0.58 ± 0.07 0.49 ± 0.06

1.00 ± 0.08 1.00 ± 0.08 1.00 ± 0.04

a b

Normalized C/C0 values are reported. See Table 1 for C0 values. C/C0 values in bold are significantly different from control with no membrane (95% confidence).

the control with no membrane (95% confidence) are highlighted in bold in the tables. At 21 ◦ C, adsorption of these compounds in Milli-Q® water does not differ significantly from the results obtained with the compounds before surrogate correction with the exception of carbamazepine adsorption by the UE10 membrane. However, some differences in adsorption were observed for the Lake Ontario water (gemfibrozil and carbamazepine by X20, acetaminophen by TS80 and NF270) and for the MBR effluent (gemfibrozil and carbamazepine by X20, gemfibrozil by NF270). Similar observations were made for the adsorption results obtained at 4 ◦ C. This indicates that water matrix effects may influence compound analytical recoveries and, in turn, membrane adsorption results. Therefore, in order to avoid masking the influence of water matrix on adsorption due to differences in individual analytical sample recoveries, the influence of water matrix on adsorption was examined for compounds with surrogates only. The three compounds presented in Fig. 4 are representative of a non-polar (μ = 1.38 D) and hydrophilic (log Kow = 0.46)

compound (acetaminophen), a non-polar (μ = 1.67 D) and moderately hydrophobic (log Kow = 2.45) compound (carbamazepine), and a polar (μ = 3.57 D) and hydrophobic (log Kow = 4.77) compound (gemfibrozil). The Tukey method for pairwise comparisons (confidence level of 95%) [33] was used to compare compound membrane adsorption in the three water matrices. No adsorption of hydrophilic acetaminophen occurred and so no differences between water matrices were expected or observed. Carbamazepine was adsorbed under most of the experimental conditions and no significant differences were observed. Finally, adsorption of gemfibrozil also occurred under most of the experimental conditions due to its hydrophobicity. However, higher adsorption in the Milli-Q water, with the UE10 membrane in particular, indicates that there may be competition for membrane adsorption sites from the organic matter present in the Lake Ontario water and MBR effluent. Unfortunately, the analysis of the influence of water matrix on membrane adsorption is limited because only three compounds could be corrected with surrogate compounds. Therefore, further investigation of the influence of

Fig. 4. Influence of water matrix on adsorption of compounds corrected with surrogate at 21 ◦ C (error bars represent standard deviation of three replicates).

276

A.M. Comerton et al. / Journal of Membrane Science 303 (2007) 267–277

water matrix on adsorption with the use of a larger number of surrogate compounds is suggested. 3.4. Influence of temperature on adsorption Adsorption of EDCs and PhACs at 21 and 4 ◦ C was compared using the Tukey method for pairwise comparisons at a confidence level of 95% [33]. Eighty two percent (217 of 264) of the measurements showed no statistically significant differences between adsorption at these two temperatures. 4. Conclusions The adsorption of 22 endocrine disrupting compounds and pharmaceutically active compounds (PhACs) by ultrafiltration, nanofiltration and reverse osmosis membranes in Lake Ontario water, MBR effluent and Milli-Q® water was examined in this study. Adsorption was strongly correlated with compound log Kow and membrane pure water permeability, and moderately correlated with compound water solubility. Therefore, in general, adsorption is expected to increase with decreasing compound water solubility and increasing compound hydrophobicity, as represented by log Kow . In addition, the most significant adsorption was observed with the UF membrane followed by the NF membranes and the RO membrane. In order to avoid masking the influence of water matrix on adsorption due to differences in analytical sample recoveries, the use of compounds with surrogates is recommended for the investigation of the influence of water matrix on adsorption. Three compounds with surrogates (acetaminophen, carbamazepine, gemfibrozil) were examined in this study and it was observed that adsorption of gemfibrozil may have been hindered due to competition for adsorption sites from the organic matter in the Lake Ontario water and MBR effluent. Finally, the influence of temperature on membrane adsorption was found to be insignificant in the range studied (21 and 4 ◦ C). The extent of adsorption observed in this batch study demonstrates the importance of evaluating EDC and PhAC removal in laboratory experiments under equilibrium conditions in order to accurately estimate membrane removal efficiencies. An understanding of EDC and PhAC adsorption is also important to improve the understanding of membrane removal mechanisms. Acknowledgements This work was funded in part by the Canadian Water Network and the Ontario Ministry of the Environment. We would like to acknowledge Stephanie Lemanik, Xiaoming Zhao and Chunyan Hao of the applied chromatography laboratory at the Ontario Ministry of the Environment in Toronto for their assistance in analysing the EDC and PhAC samples. We would also like to thank Dr. Thomas Luxbacher of Anton Paar for generously taking the time to perform zeta potential measurements and analysis of the membranes. Finally, we would like to thank the Ajax Water Supply Plant and the Port McNicoll Wastewater Treatment Plant for providing water samples.

References [1] T. Reemtsma, S. Weiss, J. Mueller, M. Petrovic, S. Gonz´alez, D. Barcelo, F. Ventura, T.P. Knepper, Polar pollutants entry into the water cycle by municipal wastewater: a European perspective, Environ. Sci. Technol. 40 (2006) 5451. [2] N. Lindqvist, T. Tuhkanen, L. Kronberg, Occurrence of acidic pharmaceuticals in raw and treated sewages and in receiving waters, Water Res. 39 (2005) 2219. [3] C. Baronti, R. Curini, G. D’Ascenzo, A. Di Corcia, A. Gentili, R. Samperi, Monitoring natural and synthetic estrogens at activated sludge sewage treatment plants and in a receiving river water, Environ. Sci. Technol. 34 (24) (2000) 5059. [4] P.E. Stackelberg, E.T. Furlong, M.T. Meyer, S.D. Zaugg, A.K. Henderson, D.B. Reissman, Persistance of pharmaceutical compounds and other organic wastewater contaminants in a conventional drinking-watertreatment plant, Sci. Total Environ. 329 (2004) 99. [5] D.W. Kolpin, E.T. Furlon, M.T. Meyer, E.M. Thurman, S.D. Zaugg, L.B. Barber, H.T. Buxton, Pharmaceuticals, hormones, and other organic wastewater contaminants in U.S. streams, 1999–2000: a national reconnaissance, Environ. Sci. Technol. 36 (6) (2002) 1202. [6] G.A. Loraine, M.E. Pettigrove, Seasonal variations in concentrations of pharmaceuticals and personal care products in drinking water and reclaimed wastewater in Southern California, Environ. Sci. Technol. 40 (2006) 687. [7] H.M. Kuch, K. Ballschmiter, Determination of endocrine-disrupting phenolic compounds and estrogens in surface and drinking water by HRGC–(NCI)-MS in the picogram per liter range, Environ. Sci. Technol. 35 (2001) 3201. [8] S.A. Snyder, S. Adham, A.M. Redding, F.S. Cannon, J. DeCarolis, J. Oppenheimer, E.C. Wert, Y. Yoon, Role of membranes and activated carbon in the removal of endocrine disruptors and pharamaceuticals, Desalination 202 (2007) 156. [9] Y. Yoon, P. Westerhoff, S.A. Snyder, E.C. Wert, Nanofiltration and ultrafiltration of endocrine disrupting compounds, pharmaceuticals and personal care products, J. Membr. Sci. 270 (2006) 88. [10] L.D. Nghiem, A.I. Sch¨afer, M. Elimelech, Removal of natural hormones by nanofiltration membranes: measurement, modeling, and mechanisms, Environ. Sci. Technol. 38 (2004) 1888. [11] K. Kimura, G. Amy, J. Drewes, Y. Watanabe, Adsorption of hydrophobic compounds onto NF/RO membranes: an artifact leading to overestimation of rejection, J. Membr. Sci. 221 (2003) 89. [12] G.B. Van den Berg, C.A. Smolders, Diffusional phenomena in membrane separation processes, J. Membr. Sci. 73 (1992) 103. [13] S. Chang, T.D. Waite, A.I. Sch¨afer, A.G. Fane, Adsorption of endocrineactive compound estrone on microfiltration hollow fiber membranes, Environ. Sci. Technol. 37 (2003) 3158. [14] Y. Kiso, Y. Nishimura, T. Kitao, K. Nishimura, Rejection properties of non-phenylic pesticides with nanofiltration membranes, J. Membr. Sci. 171 (2000) 229. [15] L.D. Nghiem, A.I. Sch¨afer, M. Elimelech, Pharmaceutical retention mechanisms by nanofiltration membranes, Environ. Sci. Technol. 39 (2005) 7698. [16] B. Van der Bruggen, D. Schaep, D. Wilms, C. Vandecasteele, Influence of molecular size, polarity and charge on the retention of organic molecules by nanofiltration, J. Membr. Sci. 156 (1999) 29. [17] Y. Yoon, P. Westerhoff, J. Yoon, S.A. Snyder, Removal of 17␤-estradiol and fluoranthene by nanofiltration and ultrafiltration, J. Environ. Eng. ASCE 130 (12) (2004) 1460. [18] C. Combe, E. Molis, P. Lucas, R. Riley, M.M. Clark, The effect of CA membrane properties on adsorptive fouling by humic acid, J. Membr. Sci. 154 (1999) 73. [19] P. Xu, J.E. Drewes, T.-U. Kim, Effect of membrane fouling on transport of organic contaminants in NF/RO membrane applications, J. Membr. Sci. 279 (2006) 165. [20] X. Zhu, M. Elimelech, Colloidal fouling of reverse osmosis membranes: measurements and fouling mechanisms, Environ. Sci. Technol. 31 (1997) 3654.

A.M. Comerton et al. / Journal of Membrane Science 303 (2007) 267–277 [21] M. Elimelech, X. Zhu, A.E. Childress, S. Hong, Role of membrane surface morphology in colloidal fouling of cellulose acetate and composite aromatic polyamide reverse osmosis membranes, J. Membr. Sci. 127 (1997) 101. [22] S. Singh, K.C. Khulbe, T. Matsuura, P. Ramamurthy, Membrane characterization by solute transport and atomic force microscopy, J. Membr. Sci. 142 (1998) 111. [23] A.E. Childress, M. Elimelech, Effect of solution chemistry on the surface charge of polymeric reverse osmosis and nanofiltration membranes, J. Membr. Sci. 119 (1996) 253. [24] J.D. Chen, N. Wada, Edge profiles and dynamic contact angles of a spreading drop, J. Colloid Interface Sci. 148 (1992) 207. [25] K.C. Khulbe, T. Matsuura, Characterization of synthetic membranes by raman spectroscopy, electron spin resonance, and atomic force microscopy: a review, Polymer 41 (2000) 1917. [26] M. Clara, B. Strenn, O. Gans, E. Martinez, N. Kreuzinger, H. Kroiss, Removal of selected pharmaceuticals, fragrances and endocrine disrupting compounds in a membrane bioreactor and conventional wastewater treatment plants, Water Res. 39 (2005) 4797. [27] K. Kimura, H. Hara, Y. Watanabe, Removal of pharmaceutical compounds by submerged membrane bioreactors (MBRs), Desalination 178 (2005) 135.

277

[28] J. Cho, G. Amy, J. Pellegrino, Membrane filtration of natural organic matter: factors and mechanisms affecting rejection and flux decline with charged ultrafiltration (UF) membrane, J. Membr. Sci. 164 (2000) 89. [29] G.R. Aiken, D.M. McKnight, K.A. Thorn, E.M. Thurman, Isolation of hydrophilic organic acids from water using nonionic macroporous resins, Org. Geochem. 18 (4) (1992) 567. [30] American Public Health Association (APHA), in: A.D. Eaton, L.S. Clesceri, E.W. Rice, A.E. Greenberg (Eds.), Standard Methods for the Examination of Water and Wastewater, 20th ed., Washington, USA, 2005. [31] Ontario Ministry of the Environment (MOE), The determination of emerging organic pollutants in environmental matrices by LC-MS-MS analysis, Laboratory Services Branch Method E3454, Etobicoke, ON, 2007. [32] Ontario Ministry of the Environment (MOE), The determination of polycyclic aromatic hydrocarbons and triazine pesticides in water matrices by gas chromatography-time of flight-mass spectrometry, Laboratory Services Branch method E3435, Etobicoke, ON, 2007. [33] A. Dean, D. Voss, Design and Analysis of Experiments, Springer-Verlag New York, Inc., New York, 1999. [34] Y. Kiso, A. Mizuno, R.A. Adawiah binti Othman, Y.-J. Jung, A. Kumano, A. Ariji, Rejection properties of pesticides with a hollow fiber NF membrane (HNF-1), Desalination 143 (2002) 147.