Journal of Membrane Science 433 (2013) 72–79 Contents lists available at SciVerse ScienceDirect Journal of Membrane Science journal homepage: www.elsevier.com/locate/memsci Correlating chlorine-induced changes in mechanical properties to performance in polyamide-based thin film composite membranes Jung-Hyun Lee a,b, Jun Young Chung a, Edwin P. Chan a, Christopher M. Stafford a,n a b Polymers Division, National Institute of Standards and Technology, 100 Bureau Drive, Gaithersburg, Maryland 20899, United States Center for Materials Architecturing, Korea Institute of Science and Technology, 39-1 Hawolgok-dong, Seongbuk-gu, Seoul 136-791, Republic of Korea a r t i c l e i n f o a b s t r a c t Article history: Received 24 August 2012 Received in revised form 14 January 2013 Accepted 18 January 2013 Available online 26 January 2013 In this study, chlorine-induced structural changes of fully-aromatic and semi-aromatic polyamide (PA) active layers used in reverse osmosis (RO) and nanofiltration (NF) membranes, respectively, were investigated by a combination of mechanical property measurements and performance studies. Our results indicated that chlorination causes quite different changes in mechanical properties of the active layer depending on the chemical nature of the PA, as identified by increased brittleness for a fullyaromatic PA and improved ductility for a semi-aromatic PA. Moreover, the results revealed that the mechanical responses of the PA active layers after chlorine exposure correlate to the overall membrane performance. A significant increase in water flux and a large decrease in salt rejection were observed for the RO membrane after chlorination, which can be ascribed to the increased fragility and resultant defects of the oxidized fully-aromatic PA network. In sharp contrast, the chlorination of the NF membrane resulted in a slightly reduced water flux accompanied with improved salt rejection, suggestive of structural compaction and densification of the semi-aromatic PA network induced by enhanced chain flexibility. We contend that our thin film measurement methodology provides key mechanical property measurements of the PA active layer and begins to bridge the gap between compositional chemical analyses and membrane performance measurements. Published by Elsevier B.V. Keywords: Thin film composite membranes Mechanical properties Chlorination Reverse osmosis Nanofiltration 1. Introduction Thin film composite (TFC) membranes have attracted intense interest as leading materials for desalination and sustainable water purification [1–3]. Such membranes are predominantly composed of a thin cross-linked polyamide (PA) layer supported by a porous polysulfone (PSF), which in turn is reinforced with a nonwoven polyester fabric [3]. The overall membrane performance is strongly affected by the structure of the top PA active layer, which is produced by interfacial polymerization of a triacid chloride and a diamine. Commercial RO membranes typically have a fully-aromatic PA layer derived from trimesoyl chloride (TMC) and m-phenylenediamine (MPD), whereas NF membranes have a semi-aromatic PA layer derived from TMC and piperazine. Although PA active layers in TFC membranes can meet the desired performance criteria (e.g., permselectivity, permeability), they typically suffer from structural deterioration by chlorine, commonly used as a disinfecting agent for biofouling control [3–5]. A fundamental understanding of the mechanism underlying the degradation caused by chlorine is n Corresponding author. Tel.: þ1 301 975 4368; fax: þ1 301 975 4924. E-mail address: chris.stafford@nist.gov (C.M. Stafford). 0376-7388/$ - see front matter Published by Elsevier B.V. http://dx.doi.org/10.1016/j.memsci.2013.01.026 therefore of paramount importance in predicting the reliability and long-term durability of these types of membranes, as well as in developing chlorine-resistant membranes [6]. Degradation of membranes by chlorine is a well-known phenomenon and has been studied for more than two decades. Early studies have shown that the chlorination of PA-based TFC membranes involves chlorine attack on the amide nitrogen (N-chlorination) and aromatic ring bonded to amide nitrogen (ring-chlorination), which consequently causes specific structural changes of the PA network [4,5]. In recent years, several mechanisms have been developed to account for the membrane performance changes upon chlorination, including conformational deformation [7,8], amide bond cleavage [9], chain tightening [10–12], change in hydrophobicity [13,14], and the physical separation of the PA layer from the support [15]. However, the origin of performance failure is still not fully understood due to the complex nature of membrane degradation, involving changes in both chemical and physical properties. It is well established that the transport properties (water and salt passage) of PA active layers are controlled by the fractional free volume (often described by a pore size and pore size distribution) available in the PA active layer, which is a function of the cross-linking density, molecular density and network 73 J.-H. Lee et al. / Journal of Membrane Science 433 (2013) 72–79 stiffness [3,16–18]. Similarly, the mechanical properties of these materials are dictated by the network structure, as defined by the cross-linking density, molecular density, and chain stiffness of the PA active layer. Thus, measurements of the mechanical behavior of PA active layers could shed light on performance changes upon chlorine exposure and potentially enable projections of membrane durability over time. Due to the ultrathin and insoluble nature of the cross-linked PA, physical property measurements of the active layer remain a challenge. In response to this challenge, researchers have designed inventive ways to tackle this problem, the most recent being the use of AFM-based nano-thermal analysis (nanoTA) to measure changes in the apparent glass transition temperature of PA layers upon chlorination [19] and the use of a pendant drop mechanical analysis (PMDA) to measure changes in stiffness and ductility of in-situ interfacially polymerized PA shells [15]. In the present study, we employed a method based on wrinkling–cracking phenomena to quantify the chlorine-induced changes in the mechanical properties of PA active layers. Specifically, our approach provides quantification of the stiffness, strength, and ductility of the extremely thin and delicate films and coatings [20,21]. Using this approach, we systematically investigated the chlorine-induced structural changes of chemically different, thin PA active layers used in commercial RO and NF membranes, having MPD-based fully-aromatic and piperazine-based semi-aromatic PAs, respectively. We then evaluated the water flux and salt rejection of the chlorine-treated membranes and correlated performance with the observed mechanical behaviors of the PA active layers in an attempt to elucidate the failure mechanism from a mechanical aspect. Additionally, we carried out chemical analysis by XPS and FT-IR to obtain complementary information on the mechanism of chlorine-induced degradation process. 2. Experimental 2.1. General Equipment and instruments or materials are identified in the manuscript in order to adequately specify the experimental details. Such identification does not imply recommendation by the National Institute of Standards and Technology, nor does it imply the materials are necessarily the best available for the purpose. The error bars presented throughout this manuscript represent one standard deviation of the data, which is taken as the experimental uncertainty of the measurement. 2.2. Materials and sample preparation Two types of PA-based TFC membranes were investigated: (1) a commercial RO membrane (SWC4þ , Hydranautics/Nitto Denko, Oceanside, CA) and (2) a commercial NF membrane (NF270, Dow Filmtec, Minneapolis, MN). Both membranes have a three-layered structure that consists of a top thin PA active layer (made by interfacial polymerization of a triacid chloride and a diamine), a microporous PSF interlayer, and a nonwoven polyester fabric support layer. Although structurally similar, the two membranes are functionally distinct, which is attributed to the different monomers used to generate the PA network structures. SWC4þ has a fully-aromatic PA layer formed from TMC and MPD (aromatic primary amine), whereas NF270 has a semi-aromatic PA layer made from TMC and piperazine (aliphatic secondary amine). The main chemical and physical characteristics of the membranes used are presented in Table 1. As-received membranes were rinsed thoroughly with deionized (DI) water, and then vacuum dried for 24 h prior to experiments. A sheet of poly(dimethylsiloxane) (PDMS; Sylgard 184, Dow Chemical Co.) was used as the substrate for mechanical analysis. The PDMS substrate was prepared by hand-mixing base monomer and curing agent with a 10:1 ratio by mass, and then post-curing the mixture at 75 1C for 2 h. After cooling, the cross-linked PDMS with thickness of 2.5 mm was cut into 75 mm  25 mm specimens. The Young’s modulus of the PDMS was estimated to be 1.8 MPa7 0.1 MPa by a Texture Analyzer (TA.XT2i, Texture Technologies) and the Poisson’s ratio of the PDMS was assumed to be 0.5. Thin PA layers supported on the PDMS substrate were prepared by transferring the active layers from commercial RO and NF membranes onto the PDMS surface using a procedure described Table 1 Chemical structure and selected structural properties of PA active layers of SWC4þ and NF270 membranes. Membrane Structure O O H H O O H H C C N N C C N N SWC4 þ (RO) C O n COOH O/N ratioa Thicknessb (nm) rmsb (nm) 1.17 70.02 2037 21 1307 35 1.42 70.05 17.7 7 0.5 5.77 1.0 1-n N H Fully-aromatic (TMC-based) O O C C N N O O C C N N NF270 (NF) C O N n Semi-aromatic (piperazine-based) a b Measured by XPS. Measured by AFM. COOH 1-n 74 J.-H. Lee et al. / Journal of Membrane Science 433 (2013) 72–79 previously [20], which followed a similar procedure described by Freger and coworkers [22,23]. Briefly, a membrane sheet was cut into pieces of 15 mm  15 mm. Next, a drop of isopropyl alcohol (IPA) was placed on the PA side of the three-layered membrane (PA/PSf/fabric), and the membrane was gently pressed onto the PDMS surface with the PA side facing down. After IPA drying, the sample (fabric/PSf/PA/ PDMS) was immersed into dimethyl formamide (DMF) to peel off the nonwoven fabric of the membrane. The isolated PA/PDMS bilayer was obtained by gentle and complete dissolution of the residual PSf layer of the membrane with DMF and dichloromethane. It has been previously deduced that such solvent rinsing has a minimal effect on the chemical and structural properties of the PA active layer [22,23]. 2.3. Chlorination Chlorination was performed by soaking samples (both membrane sheets and PA/PDMS specimens) in an aqueous sodium hypochlorite solution for different periods of exposure time (0rtexp r168 h). It is well known that the oxidation of PAs is highly sensitive to the chlorine concentration and pH. However, a moderately high chlorine concentration (1000 mg/L) under acidic conditions (pH¼4) was chosen as a case study to demonstrate the ability of our measurement approach to capture chlorine-induced changes in the mechanical and performance properties of the PA active layer. The contents of the solution bath were gently mixed with a magnetic stirrer. Samples were taken at various time intervals, and then rinsed thoroughly with DI water prior to measurements. 2.4. Morphological analysis The surface morphology of the fully-aromatic and semiaromatic PA layers was characterized by atomic force microscopy (AFM, MFP-3D, Asylum Research) and scanning electron microscopy (SEM, LEO 1530). The root-mean-square (rms) roughness of the PA layers was estimated from the topography image of a 10 mm  10 mm area using AFM operated in tapping mode. The reported roughness value is the average of three random area scans. For analysis with SEM, a thin layer of gold was sputtered on the surface of pristine and chlorinated membranes. SEM micrographs of the membrane surfaces were then obtained at an accelerating voltage of 5 kV. 2.5. Mechanical analysis Changes in the three key mechanical properties (elastic modulus, fracture strength, and onset fracture strain [indicator of ductility]) of the PA layers of the RO and NF membranes upon chlorine exposure were examined by a combined wrinkling– cracking technique. Detailed experimental procedure and data analysis of this technique were reported previously [20,21]. In brief, a PA/PDMS bilayer (untreated or after exposure to chlorine) was mounted onto a custom-built strain stage equipped with a strain controller, and then a uniaxial tensile strain was applied in a stepwise manner with a strain interval of E3%. Surface images at each strain increment were recorded with either an optical microscope (Labophot-2, Nikon) or an optical profilometer (NewView 7300, Zygo). Uniaxial stretching of the PA/PDMS induces periodic surface wrinkling patterns parallel to the applied strain (e) due to lateral Poisson contraction, as presented in Fig. 1. The plane-strain modulus of the PA layer was determined from the measured wrinkle wavelength (l) according to Enf ¼3Ens (l/2ph)3, where En ¼E/(1  n2) is the plane-strain modulus, E is the Young’s modulus, n is the Poisson’s ratio (the subscript f and s denote the PA film and PDMS substrate, respectively), and h is the thickness of the PA layer [24]. Tapping-mode AFM was used to measure the layer thickness following the protocol described previously [22]. As illustrated in Fig. 2, the PA layer undergoes fracture with regularly-spaced cracks perpendicular to the strain direction, whose crack density increases with the increased level of applied strain. This ‘‘thin film cracking’’ behavior is well studied in the regime where the average crack spacing (d) is shorter than the critical dimension (dc ¼4hEf/Es). In this regime, the average crack density (1/d) is proportional with the applied strain as given by the following equation: 1/d ¼Es(e  en)/2hsn, where sn and en are the fracture strength and onset fracture strain of the layer, respectively. Based on this equation, sn and en were determined from the slope and the intercept to zero crack density, respectively, of the linear fit to the experimentally obtained 1/dvs. e plot. Data on mechanical properties for the fully-aromatic PA layer (the active layer of SWC4þ membrane) were reproduced from our previous report [20], while those for the semi-aromatic PA layer (the active layer of NF270 membrane) were obtained for this study. 2.6. Chemical analysis XPS and FT-IR were employed to probe the chemical changes of both the fully-aromatic and semi-aromatic PA layers at various chlorine exposure times. Samples were dried under vacuum for 24 h after thorough cleaning of chlorinated membranes by DI water. XPS was performed on a Kratos AXIS Ultra DLD spectrometer with monochromated Al Ka radiation operating at 1486 eV and scanning over a binding energy range of 0 to 1200 eV with a dwell time of 100 ms. The analyzer pass energy was 160 eV for survey spectra. The chemical composition corresponding to each Fig. 1. Surface images of the strain-induced wrinkle patterns on (a) the PA layer of SWC4þ RO membrane (optical microscopy image; scale bar ¼ 20 mm) and (b) the PA layer of NF270 NF membranes (optical profilometer image; scale bar ¼20 mm). The arrows indicate the direction of the applied strain (e). 75 J.-H. Lee et al. / Journal of Membrane Science 433 (2013) 72–79 Fig. 2. Optical microscopy images illustrating the progressive crack growth with increasing level of applied strain (e): (a) the PA layer of SWC4 þ RO membrane and (b) the PA layer of NF270 NF membrane. The scale bars represent 200 mm. chlorination time was averaged from three survey scans. FT-IR spectra were recorded using Nicolet Nexus 670 spectrometer equipped with an attenuated total reflection (ATR) element in the range of wavenumber (400 to 4000) cm  1. All data were obtained at room temperature. 2.7. Performance analysis Changes in membrane performance before and after chlorine exposure (24 h) were assessed with 1500 mg/L NaCl aqueous solution at pH 5.8 in a dead-end filtration unit (HP4750 Stirred Cell, Sterlitech). Water flux and salt (NaCl) rejection were determined under various operating pressures (400 psi to 800 psi) at room temperature. The membranes, both pristine and exposed to chlorine, were cut into circular coupons with a diameter of 49 mm, washed thoroughly with DI water, and then placed in the filtration unit. Performance data were collected after steady-state flow conditions had been reached (from 30 min to a few hours, depending on the membrane type and operating conditions). Water flux was determined from the amount of the collected permeate (V) over the effective membrane area (A¼14.6 cm2) for a fixed time interval (t) as given by V/At. Salt rejection was calculated from the NaCl concentrations in the feed (Cf) and the permeate (Cp), which were determined by measuring the electrical conductivity with a conductance meter (Orion 3 Star Plus, Thermo Scientific) using the following equation: salt rejection (%)¼100  (1 Cp/Cf). The performance data reported were the averages of measurements of at least three samples. For the sake of comparison, the data for chlorinetreated membranes were normalized by those for their pristine (untreated) counterparts. 3. Results and discussion 3.1. Pristine RO and NF membranes Table 1 shows the chemical structure and selected structural properties of PA active layers in SWC4þ (RO) and NF270 (NF) membranes. SWC4 þ has a MPD-based, fully-aromatic PA active layer, whereas NF270 has a piperazine-based, semi-aromatic PA Table 2 Mechanical properties of the isolated PA active layers of SWC4þ and NF270 membranes. Membrane SWC4 þ (RO) NF270 (NF) E*f (Modulus, GPa) 1.4 7 0.5 1.3 7 0.1 sn (Fracture strength, en (Onset fracture MPa) strain, %) 66.9 73.3 353 720 14.07 4.1 44.5 7 6.4 layer. The choice of amine monomer (e.g., MPD vs. piperazine) for interfacial polymerization with TMC is known to play a major role in affecting the structural properties of the resultant PA layer, such as cross-linking density, layer thickness, and surface roughness [25,26]. In this study, the atomic ratio of oxygen to nitrogen (O/N) determined by XPS analysis was used to estimate the degree of cross-linking. Note the theoretical O/N ratio is 1.0 for fully cross-linked PA and 2.0 for linear PA in these systems [9]. From XPS results, it was confirmed that the O/N ratio in the fully-aromatic PA layer (1.17) was lower, and thus had a higher cross-linking density, than that in the semi-aromatic PA layer (1.42) [27]. In addition, AFM image analysis showed that the fully-aromatic PA layer was thick (203 nm) and rough (rms roughness of 130 nm), while the semi-aromatic PA layer was thin (17.7 nm) and fairly smooth (rms roughness of 5.7 nm). These results were in good agreement with previous reports [22,25,26]. To gain further information on the structural characteristics of the PA layers in RO and NF membranes, the mechanical properties, including modulus, fracture strength, and onset fracture strain, of the isolated PA active layers were measured using a combined wrinkling–cracking technique. This technique is detailed in the Experimental Section, but involves the uniaxial stretching of a PA/ PDMS bilayer, which induces both surface wrinkling and film cracking. The modulus of the PA layer was calculated from the measured wrinkling wavelength (Fig. 1). The fracture strength and onset fracture strain of the PA layer were determined by monitoring the progressive crack growth with increasing applied strain (Fig. 2). Table 2 summarizes the mechanical properties of both the fullyaromatic (RO) and semi-aromatic (NF) PA layers. The plane-strain modulus of the RO PA layer (1.4 GPa) was found to be similar to that of the NF PA layer (1.3 GPa), whereas both fracture strength and onset 76 J.-H. Lee et al. / Journal of Membrane Science 433 (2013) 72–79 fracture strain were significantly higher for the NF PA layer (353 MPa, 44.5%) than for the RO PA layer (66.9 MPa, 14.0%). In an effort to rationalize the mechanical properties of both RO and NF PA active layers, we must look more closely at the structure of these films upon interfacial polymerization. The RO PA layer is comprised of a highly cross-linked network with rigid repeating units dominated by aromatic moieties, while the NF PA layer has a relatively less cross-linked network and less rigid aliphatic moieties. The flexible nature of the NF PA network could justify its higher onset fracture strain compared to the RO PA network. However, one might expect that the RO PA layer would have a higher fracture strength due to its higher crosslink density and rigid aromatic monomers. This notion is reinforced by literature results that show fully-aromatic PA had a higher rupture strength (SR) than that of semi-aromatic PA [15,28]. However, a key difference between our approach and the pendant drop method (PDMA) is that the thickness of the membrane is a direct input into the analytical solution for wrinkling–cracking. It has been shown that interfacially polymerized aromatic PAs are highly heterogeneous through the thickness of the film, and it has been suggested that these active layers could be described as having a three-layer structure comprised of a dense core surrounded by more loosely crosslinked outer layers [29,30]. The repercussions of such heterogeneity would be that the entire film may not contribute to the measured fracture strength, i.e. the wrinkling–cracking results correspond to a volume-average across the entire heterogeneous film but may only represent the fracture strength of just the core layer (having an unknown thickness). Thus, by using the thickness of the entire film in the calculations, the apparent fracture strength appears to be lower than expected for such a rigid network. In contrast, the piperazine-based, semi-aromatic (NF) PA layer was characterized to be fairly dense and nearly depth-homogeneous [27,29]. Therefore, the entire film contributes equally to the fracture strength of the material. The heterogeneous structure of RO PA active layers could also explain why their modulus is not higher than that of NF PA layers, as the thickness of the entire layer is used to calculate the modulus. an increase in the O/N ratio for the RO PA layer with chlorine treatment indicates that the loss of cross-linking density occurs presumably by chain cleavage of secondary amides, which has often been proposed to explain the structural changes of the fullyaromatic PA by chlorine attack under harsh chlorinating Fig. 3. XPS measurements of the averaged atomic ratio of (a) chlorine to nitrogen (Cl/N) and (b) oxygen to nitrogen (O/N) of SWC4 þ RO (closed symbols) and NF270 NF (open symbols) membranes as a function of chlorine exposure time (texp; Cl concentration¼ 1000 mg/L, pH¼ 4). 3.2. Chlorinated RO and NF membranes XPS and FT-IR were used to probe the chemical changes of PA active layers in RO and NF membranes upon chlorine exposure. Fig. 3 shows the atomic ratio of chlorine to nitrogen (Cl/N, Fig. 3a) and oxygen to nitrogen (O/N, Fig. 3b) obtained from XPS spectra on the RO and NF membranes as a function of chlorine exposure time (texp). The Cl/N ratio for the fully-aromatic (RO) PA layer (closed symbols in Fig. 3a) increased up to 1.6, suggesting that the chlorine is likely bound to the secondary amide nitrogen via N-chlorination and to the aromatic group through ringchlorination. In contrast, the Cl/N ratio for the semi-aromatic (NF) PA layer (open symbols in Fig. 3a) did not increase significantly and was much less than 1. Both results were in agreement with the previous report [9] and suggest that the level of chlorine resistance of the PA layers depends strongly on their chemical structure. In particular, tertiary amide nitrogen and aliphatic group bonded to amide nitrogen were found to be chemically inert to chlorine, which could account for the much lower chlorine uptake observed for the piperazine-based NF PA layer [5,9,31,32]. Nevertheless, it should be noted that a small amount of chlorine could be incorporated by chlorine attack on the dangling, un-crosslinked nitrogens in the NF PA layer [9]. The relative content of oxygen to nitrogen (O/N ratio) was used as a relative measure of the degree of cross-linking. A drastic increase in the O/N ratio was observed for the RO PA layer (closed symbols in Fig. 3b), consistent with the previous report [9]. Such Fig. 4. FT-IR spectra of (a) SWC4þ RO and (b) NF270 NF membranes at different chlorination times (Cl concentration ¼1000 mg/L, pH ¼4). J.-H. Lee et al. / Journal of Membrane Science 433 (2013) 72–79 Fig. 5. Changes in the three key mechanical properties of the PA active layers of SWC4þ RO (closed symbols) and NF270 NF (open symbols) membranes as a function of chlorine exposure time (texp; Cl concentration ¼ 1000 mg/L, pH ¼4): (a) normalized plane-strain modulus, (b) normalized fracture strength, and (c) normalized onset fracture strain. The subscript 0 refers to the value for the untreated (pristine) PA layer. The dotted horizontal lines in (a)–(c) indicate the value of ‘‘1’’, which represents no difference between the values obtained for the pristine and chlorinated PA layers. conditions [3,5,9,33]. On the other hand, a minor decrease in the O/N ratio was seen for the NF PA layer (open symbols in Fig. 3b), suggesting that chain scission is not a plausible mechanism underlying structural changes for the piperazine-based PA. This result, together with the observed low chlorine uptake, indicates that the semi-aromatic PA layer has better chlorine tolerance compared to the fully-aromatic PA layer. FT-IR was used to study the near-bulk chemical changes occurring in the PA layers upon chlorine exposure because it has a much deeper penetration depth than XPS (penetration depth of less than 10 nm) [7–9]. Fig. 4 illustrates the FT-IR spectra changes of the RO and NF membranes at different chlorination times. Three characteristic peaks at 1661 cm  1 (amide I, C¼O stretching), 1541 cm  1 (amide II, N–H in-plane bending), and 1610 cm  1 (hydrogen-bonded CQO) were observed for the pristine RO membrane (Fig. 4a) [10,34,35]. The intensities of peaks at 1610 cm  1 and 1541 cm  1 decreased rapidly and completely disappeared within a few hours of chlorination. In addition, it was found that the amide I band at 1661 cm  1 shifted progressively to higher frequency with increasing chlorine exposure. Previous studies have attributed these spectral changes to the destruction of intermolecular hydrogen bonds between CQO and N–H groups resulting from the substitution of chlorine for hydrogen in amide nitrogen via N-chlorination [7,8,10,36]. In the case of the NF membrane (Fig. 4b), the peaks at 1610 cm  1 and 1541 cm  1 (typical for the fully-aromatic RO membrane) were absent, confirming the 77 piperazine-based PA chemistry of the NF membrane [34]. In contrast with the RO membrane, no discernible changes in the FT-IR spectra upon chlorination were observed for the NF membrane, further supporting that the chemical feature of tertiary amide and aliphatic ring of the piperazine-based PA is tolerant to chlorine attack [9], in excellent agreement with the XPS results obtained in this study. To gain a deeper insight into the overall changes in the structural and physical properties of the PA layers in response to chlorine exposure, the combined wrinkling-cracking technique was again employed to characterize the mechanical properties of the isolated PA active layers. Fig. 5 details the changes in the modulus, fracture strength, and onset fracture strain of the fullyaromatic (RO) and semi-aromatic (NF) PA layers as a function of chlorine exposure time (texp). In the case of the RO PA layer, the modulus increased slowly with increased chlorination time (closed symbols in Fig. 5a), whereas both fracture strength and onset fracture strain decreased rapidly at the early stage of chlorine exposure ( E24 h) and then decreased gradually at long exposure times (closed symbols in Fig. 5b and c). These results illustrate the ‘‘embrittlement’’ behavior of the fully-aromatic PA layer, similar with observations by other authors [4,15]. The N-chlorination and concomitant ring-chlorination in the fully-aromatic PA networks have been shown to cause conformational changes by disrupting the intermolecular hydrogen bonding and ultimately destroying PA linkages by chain cleavage, which will impact the mechanical behavior as well. Although there appears to be a decrease in crosslink density as measured via XPS and FT-IR, ‘‘embrittlement’’ cannot be explained by chain scission as noted previously [15]. One hypothesis is that upon oxidative chain scission, other bonds are formed (e.g., radical recombination) that result in a net change in the structure of the PA membrane that is not captured by the traditional metric of O/N ratio. Further chemical analysis would need to be conducted to test this hypothesis. Interestingly, the semi-aromatic (NF) PA layer exhibited a mechanical behavior significantly different from the fully-aromatic (RO) PA layer. The most noticeable differences were observed in the modulus (Fig. 5a) and onset fracture strain (Fig. 5c). While chlorination induced an ultimate increase in the modulus along with a substantial reduction in the fracture strain for the RO PA layer, a slight decrease in the modulus with a remarkable increase in the fracture strain was observed for the NF PA layer, suggesting an enhanced ductility of the NF PA layer in response to chlorine exposure. Additionally, the fracture strength of the NF PA layer decreased with increasing chlorination time, similar to the trend of the RO PA layer; however, its decrement was much smaller for the NF case. The observed mechanical behavior of the NF PA layer upon chlorination suggests that the incorporated chlorine disrupts local hydrogen bonding, resulting in an increase in the chain flexibility of a polymer, but not causing chain cleavage as postulated for the RO PA layer. Therefore, these structural changes in the NF PA layer are expected to enhance the mobility and extensibility of polymer chain, manifesting in improved ductility together with minor decreases in both strength and modulus. 3.3. Membrane performance before and after chlorination Correlating the mechanical changes of the PA active layers upon chlorination with their membrane performance is of great interest to elucidate the membrane failure mechanism, because the performance is linked with the mechanical properties and integrity of the ‘‘discriminating’’ PA active layer. Besides permselectivity, high mechanical strength and integrity ensure the ability of the PA active layer to withstand high hydraulic pressures during operation; otherwise, catastrophic performance failure could result [5,37]. The water flux and salt (NaCl) rejection 78 J.-H. Lee et al. / Journal of Membrane Science 433 (2013) 72–79 normalized by those of the pristine counterparts of the chlorinated (for 24 h) RO and NF membranes are presented in Fig. 6a and b, respectively. In the case of the RO membrane, chlorination drastically increased water flux and decreased salt rejection for all operating pressure conditions examined (filled bars in Fig. 6), which is consistent with the typical performance behavior of RO membranes treated with acidic and/or relatively concentrated hypochlorite solutions [9,10,19]. Since membrane performance is correlated directly with the structure of the active layer, the observed performance failure of the RO membrane by chlorination can be interpreted in terms of the changes in the mechanical properties (volume-averaged structural properties) of the fullyaromatic PA layer. Upon chlorination, the RO PA layer becomes increasingly fragile with a significant loss of its strength and ductility (i.e., embrittlement) due to the permanent structural changes of the PA networks. Eventually, such an increased fragility of the RO PA layer could lead to the formation of the permanent physical damage such as cracks and ruptures under the mechanical stresses induced by high operating hydraulic pressures. These arguments are supported strongly by the representative images from SEM (Fig. 7). The fracture-like features are clearly seen on the PA surface of the chlorinated RO membrane after permeation tests (Fig. 7b). These mechanical defects observed on the PA active layer are likely to result in the observed high permeation of both water and salt through the RO membrane, as discussed by others [15,28]. In sharp contrast, a noticeable reduction in water flux and an increase in salt rejection upon chlorination are observed for the NF membrane, irrespective of the applied pressures examined in this study (open bars in Fig. 6). Similar to the chlorine-induced mechanical behavior of the PA layer discussed above, the observed opposite effect of chlorination on the performances of the RO and NF membranes implicitly proposes that the transport properties are correlated to the changes in the mechanical properties of the PA barrier layers reflecting the specific structural alteration of the PA networks due to the presence of chlorine. In the present study, mechanical characterization demonstrated that chlorination of the NF PA layer remarkably enhanced its ductility due to promoted chain flexibility resulting from the weakening of intermolecular interactions upon the incorporation of chlorine. The resultant more ductile and flexible NF PA active layer could lead to structural compaction and densification of the PA chain network at high operating hydraulic pressures [7,10,12,32,35], unlike the case of the RO PA layer where physical defects were created. In fact, no mechanical damage was present on the PA surface of the chlorine-treated NF membrane, as illustrated in Fig. 7d. The possible structural compaction of the NF PA active layer could reduce the available free volume for transport [10,32] and thus suppress the passage of both water and salt through the membrane, which could account for the decreased permeate flux and increased salt rejection of the chlorinated NF membrane. Fig. 6. (a) Normalized water flux and (b) normalized NaCl rejection of chlorinated SWC4þ RO (filled bars) and NF270 NF (open bars) membranes at different operating pressures (chlorination condition: Cl concentration ¼1000 mg/L, pH¼ 4, and texp ¼24 h). Note that the normalized values were obtained by dividing the measured values of the chlorinated membranes by those of the pristine membrane. The dotted horizontal lines in (a) and (b) indicate the value of ‘‘1’’, which represents no difference between the values obtained for the pristine and chlorinated membranes. The numbers labeled in (a) indicate the values for the NF270 membrane. 4. Conclusions We employed a mechanical analysis based on wrinkling– cracking to capture the chlorine-induced changes in mechanical properties of commercial polyamide active layers used in reverse osmosis and nanofiltration membranes. The MPD-based, fullyaromatic PA representative of RO membranes became more brittle and fragile, while the piperazine-based, semi-aromatic PA representative of NF membranes became more ductile and flexible. The embrittlement of the RO PA active layer upon chlorination was attributed to the structural changes associated with chain cleavage and structural rearrangement, as evidenced by XPS and FT-IR analysis. Conversely, the NF PA active layer Fig. 7. SEM surface images of (a and b) SWC4þ RO and (c and d) NF270 NF membranes after performance test, (a) and (c) are pristine membranes, while (b) and (d) are chlorinated ones (chlorination condition: Cl concentration ¼ 1000 mg/L, pH ¼4, and texp ¼24 h). The scale bars of the main images indicate 30 mm, and the one for the inset image in (b) represents 5 mm. exhibited a less pronounced change in crosslink density and enhanced chain flexibility, resulting in increased ductility of the active layer. The observed performance changes in the RO and NF membranes upon chlorine exposure were successfully correlated to the J.-H. Lee et al. / Journal of Membrane Science 433 (2013) 72–79 mechanical responses of the PA barrier layers. A remarkable increase in water flux and a decrease in salt rejection of the chlorinated RO membranes was observed, and we hypothesized that mechanical defects (fractures) caused by the increased brittleness of the active layer were responsible for the decrease in performance. In contrast, the chlorinated NF membrane resulted in a decrease in water flux together with an increase in salt rejection, which was postulated to be a result of structural compaction of the PA layer due to enhanced chain flexibility and mobility upon chlorination. The findings of this study suggest that mechanical analysis of the active layer, along with traditional chemical analysis, can provide a more complete picture of failure modes in PA active layers upon exposure to chlorine. Acknowledgments The authors thank Hydranautics/Nitto Denko and Dow Filmtec for providing membrane materials. This work is an official contribution of the National Institute of Standards and Technology; not subject to copyright in the United States. This work was financially supported in part by a grant from the Center for Materials Architecturing of KIST. References [1] M. Elimelech, W.A. Phillip, The future of seawater desalination: energy, technology, and the environment, Science 333 (2011) 712–717. [2] M.A. Shannon, P.W. Bohn, M. Elimelech, J.G. Georgiadis, B.J. Marinas, A.M. Mayes, Science and technology for water purification in the coming decades, Nature 452 (2008) 301–310. [3] G.M. Geise, H.S. Lee, D.J. Miller, B.D. Freeman, J.E. McGrath, D.R. Paul, Water purification by membranes: the role of polymer science, J. Polym. Sci. Part B: Polym. Phys. 48 (2010) 1685–1718. [4] S. Avlonitis, W.T. Hanbury, T. Hodgkiess, Chlorine degradation of aromatic polyamides, Desalination 85 (1992) 321–334. [5] J. Glater, S.K. Hong, M. Elimelech, The search for a chlorine-resistant reverseosmosis membrane, Desalination 95 (1994) 325–345. [6] H.B. Park, B.D. Freeman, Z.B. Zhang, M. Sankir, J.E. McGrath, Highly chlorinetolerant polymers for desalination, Angew. Chem. Int. Ed. 47 (2008) 6019–6024. [7] Y.N. Kwon, J.O. Leckie, Hypochlorite degradation of crosslinked polyamide membranes: II. Changes in hydrogen bonding behavior and performance, J. Membr. Sci. 282 (2006) 456–464. [8] Y.N. Kwon, J.O. Leckie, Hypochlorite degradation of crosslinked polyamide membranes: I. Changes in chemical/morphological properties, J. Membr. Sci. 283 (2006) 21–26. [9] V.T. Do, C.Y. Tang, M. Reinhard, J.O. Leckie, Degradation of polyamide nanofiltration and reverse osmosis membranes by hypochlorite, Environ. Sci. Technol. 46 (2012) 852–859. [10] G.D. Kang, C.J. Gao, W.D. Chen, X.M. Jie, Y.M. Cao, Q. Yuan, Study on hypochlorite degradation of aromatic polyamide reverse osmosis membrane, J. Membr. Sci. 300 (2007) 165–171. [11] H.D. Raval, J.J. Trivedi, S.V. Joshi, C.V. Demurari, Flux enhancement of thin film composite RO membrane by controlled chlorine treatment, Desalination 250 (2010) 945–949. [12] S.C. Yu, M.H. Liu, Z.H. Lu, Y. Zhou, C.J. Gao, Aromatic-cycloaliphatic polyamide thin-film composite membrane with improved chlorine resistance prepared from m-phenylenediamine-4-methyl and cyclohexane-1,3,5-tricarbonyl chloride, J. Membr. Sci. 344 (2009) 155–164. [13] X.F. Zhai, J.Q. Meng, R. Li, L. Ni, Y.F. Zhang, Hypochlorite treatment on thin film composite RO membrane to improve boron removal performance, Desalination 274 (2011) 136–143. View publication stats 79 [14] A. Simon, L.D. Nghiem, P. Le-Clech, S.J. Khan, J.E. Drewes, Effects of membrane degradation on the removal of pharmaceutically active compounds (PhACs) by NF/RO filtration processes, J. Membr. Sci. 340 (2009) 16–25. [15] N.P. Soice, A.R. Greenberg, W.B. Krantz, A.D. Norman, Studies of oxidative degradation in polyamide RO membrane barrier layers using pendant drop mechanical analysis, J. Membr. Sci. 243 (2004) 345–355. [16] J.G. Wijmans, R.W. Baker, The solution-diffusion model — a review, J. Membr. Sci. 107 (1995) 1–21. [17] D.R. Paul, Reformulation of the solution-diffusion theory of reverse osmosis, J. Membr. Sci. 241 (2004) 371–386. [18] S.H. Kim, S.-Y. Kwak, T. Suzuki, Positron annihilation spectroscopic evidence to demonstrate the flux-enhancement mechanism in morpohlogy-controlled thin-film-composite (TFC) membrane, Environ. Sci. Technol. 39 (2005) 1764–1770. [19] S.H. Maruf, D.U. Ahn, J. Pellegrino, J.P. Killgore, A.R. Greenberg, Y. Ding, Correlation between barrier layer Tg and a thin-film composite polyamide membrane’s performance: effect of chlorine treatment, J. Membr. Sci. 405–406 (2012) 167–175. [20] J.Y. Chung, J.H. Lee, K.L. Beers, C.M. Stafford, Stiffness, strength, and ductility of nanoscale thin films and membranes: a combined wrinkling–cracking methodology, Nano Lett. 11 (2011) 3361–3365. [21] J.H. Lee, J.Y. Chung, C.M. Stafford, Effect of confinement on stiffness and fracture of thin amorphous polymer films, ACS Macro Lett. 1 (2012) 122–126. [22] V. Freger, Swelling and morphology of the skin layer of polyamide composite membranes: an atomic force microscopy study, Environ. Sci. Technol. 38 (2004) 3168–3175. [23] A. Ben-David, S. Bason, J. Jopp, Y. Oren, V. Freger, Partitioning of organic solutes between water and polyamide layer of RO and NF membranes: correlation to rejection, J. Membr. Sci. 281 (2006) 480–490. [24] J.Y. Chung, A.J. Nolte, C.M. Stafford, Surface wrinkling: a versatile platform for measuring thin-film properties, Adv. Mater. 23 (2011) 349–368. [25] V. Freger, Nanoscale heterogeneity of polyamide membranes formed by interfacial polymerization, Langmuir 19 (2003) 4791–4797. [26] F.A. Pacheco, I. Pinnau, M. Reinhard, J.O. Leckie, Characterization of isolated polyamide thin films of RO and NF membranes using novel TEM techniques, J. Membr. Sci. 358 (2010) 51–59. [27] C.Y. Tang, Y.N. Kwon, J.O. Leckie, Effect of membrane chemistry and coating layer on physiochemical properties of thin film composite polyamide RO and NF membranes: I. FTIR and XPS characterization of polyamide and coating layer chemistry, Desalination 242 (2009) 149–167. [28] I.J. Roh, Influence of rupture strength of interfacially polymerized thin-film structure on the performance of polyamide composite membranes, J. Membr. Sci. 198 (2002) 63–74. [29] D.G. Cahill, V. Freger, S.Y. Kwak, Microscopy and microanalysis of reverseosmosis and nanofiltration membranes, MRS Bull. 33 (2008) 27–32. [30] O. Coronell, B.J. Marinas, D.G. Cahill, Depth heterogeneity of fully aromatic polyamide active layers in reverse osmosis and nanofiltration membranes, Environ. Sci. Technol. 45 (2011) 4513–4520. [31] T. Kawaguchi, H. Tamura, Chlorine-resistant membrane for reverse-osmosis: 1. Correlation between chemical structures and chlorine resistance of polyamides, J. Appl. Polym. Sci. 29 (1984) 3359–3367. [32] P.R. Buch, D.J. Mohan, A.V.R. Reddy, Preparation, characterization and chlorine stability of aromatic-cycloaliphatic polyamide thin film composite membranes, J. Membr. Sci. 309 (2008) 36–44. [33] R. Singh, Polyamide polymer-solution behavior under chlorination conditions, J. Membr. Sci. 88 (1994) 285–287. [34] S.D. Arthur, Structure property relationship in a thin-film composite reverseosmosis membrane, J. Membr. Sci. 46 (1989) 243–260. [35] Y.N. Kwon, C.Y. Tang, J.O. Leckie, Change of chemical composition and hydrogen bonding behavior due to chlorination of crosslinked polyamide membranes, J. Appl. Polym. Sci. 108 (2008) 2061–2066. [36] A. Ettori, E. Gaudichet-Maurin, J.C. Schrotter, P. Aimar, C. Causserand, Permeability and chemical analysis of aromatic polyamide based membranes exposed to sodium hypochlorite, J. Membr. Sci. 375 (2011) 220–230. [37] L.A. Hoover, J.D. Schiffman, M. Elimelech, Nanofibers in thin-film composite membrane support layers: enabling expanded application of forward and pressure retarded osmosis, Desalination 308 (2013) 73–81.