Membrane characterization by dynamic hysteresis
Journal of Membrane Science 366 (2011) 17–24
Contents lists available at ScienceDirect
Journal of Membrane Science journal homepage: www.elsevier.com/locate/memsci
Membrane characterization by dynamic hysteresis: Measurements, mechanisms, and implications for membrane fouling Sangyoup Lee a , Eunsu Lee a , Menachem Elimelech b , Seungkwan Hong a,∗ a b
School of Civil, Environmental & Architectural Engineering, Korea University, 1, 5-ka, Anam-Dong, Sungbuk-Gu, Seoul 136-713, Republic of Korea Department of Chemical Engineering, Environmental Engineering Program, Yale University, New Haven, CT 06520-8286, USA
a r t i c l e
i n f o
Article history: Received 17 March 2010 Received in revised form 15 September 2010 Accepted 17 September 2010 Available online 25 September 2010 Keywords: Dynamic hysteresis RO membrane Surface heterogeneity Charge distribution Membrane fouling
a b s t r a c t The surface characteristics of reverse osmosis membranes and their relation to membrane fouling are systematically investigated by measuring membrane dynamic hysteresis based on the Wilhelmy plate method. Dynamic hysteresis represents the difference between the forces applied to a membrane surface when it is advanced into and withdrawn from a liquid or solution. Our results demonstrate that the chemical surface heterogeneity of various RO membranes could be quantified by measuring their dynamic hysteresis. The chemical heterogeneity was mostly related to the distribution of surface charge rather than average zeta potential. There was a remarkable correlation between the chemical surface heterogeneity and membrane dynamic hysteresis. It was clearly shown that dynamic hysteresis varied substantially with respect to the solution chemistry of test solutions. The dynamic hysteresis of RO membranes measured in the presence of organic foulants was further related to the flux-decline rate determined from bench-scale fouling experiments. It was found that higher flux-decline rate was obtained for RO membranes with larger dynamic hysteresis. Based on the results in this study, it is demonstrated that dynamic hysteresis measurements can be a promising tool for characterizing membrane surfaces as well as assessing membrane fouling. © 2010 Elsevier B.V. All rights reserved.
1. Introduction Reverse osmosis (RO) membranes are currently being used in a wide range of applications, including brackish/seawater desalination, drinking water treatment, and wastewater reuse. Efficient operation of RO membranes in these applications, however, is often hampered by membrane fouling. Since RO membrane fouling is a surface phenomenon, surface characteristics of RO membranes play a major role in membrane fouling. For this reason, there have been numerous studies investigating the surface characteristics of RO membranes and relating these characteristics to the rate and extent of membrane fouling [1–7]. Consequently, there have been significant advances in membrane fabrication technology, especially in surface modification, where membrane surface properties have been tailored toward reducing membrane fouling as well as enhancing membrane permeability [8–13]. Surface characteristics affecting membrane fouling can be divided into chemical and physical characteristics. The former mainly includes surface charge and hydrophobicity, where zeta potential and contact angle measurements are the major relevant characterization tools, respectively. Physical characteristics
∗ Corresponding author. Tel.: +82 2 3290 3322; fax: +82 2 928 7656. E-mail address: [email protected] (S. Hong). 0376-7388/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.memsci.2010.09.024
often refer to surface roughness, which is commonly determined by atomic force microscopy (AFM). Morphological surface heterogeneity (i.e., variations in the distribution of peak and valley structure) also belongs to physical surface characteristics [14]. Based on these surface characteristics and their relation to membrane fouling, it has been found that the propensity for fouling increases for membranes that are less negatively charged, more hydrophobic, and rougher [1,3,4,7,15–17]. However, these characterization methods are time-consuming and require relatively expensive instruments. Furthermore, it is difficult to isolate the effect of these surface characteristics. For example, changes in surface charge upon variation of feed water solution chemistry could result in the alteration of surface hydrophobicity and perhaps morphological surface roughness [14]. Therefore, a simple analytic technique to characterize various surface characteristics and the interplay between them could be a great asset to the study of membrane surface characterization. Recently, it has been demonstrated that dynamic hysteresis could be a useful parameter to characterize the physical surface properties of RO membranes [18]. Dynamic hysteresis values of various RO membranes were related to both the average surface roughness and the roughness distribution (i.e., morphological surface heterogeneity). It was shown that dynamic hysteresis is more closely related to the distribution of peak and valley structure of roughness than the average roughness. Furthermore, it has
18
S. Lee et al. / Journal of Membrane Science 366 (2011) 17–24
been shown that dynamic hysteresis measurements can be performed with various test liquids to reflect solution chemistry. By using a non-polar liquid (i.e., diiodomethane), the interference of chemical surface characteristics with physical surface characteristics could be excluded and, thus, a more accurate correlation between dynamic hysteresis and physical surface heterogeneity was obtained. In that study, however, the chemical aspects of dynamic hysteresis were not investigated. The main objective of this study is to determine the chemical surface characteristics of RO membranes by dynamic hysteresis and relate these characteristics to membrane fouling propensity. The chemical surface characteristics of RO membranes, including surface charge and its distribution, were investigated first. The distribution of surface charge was characterized by varying solution pH and anionic surfactant (sodium dodecyl sulfate) concentration during dynamic hysteresis measurements. Next, the membrane dynamic hysteresis was determined in the presence of organic foulants and the results were related to flux-decline data obtained from lab-scale RO fouling experiments. This study delineates, for the first time, the use of dynamic hysteresis for characterizing the chemical surface characteristics of RO membranes and the direct relationship to membrane fouling.
2. Background 2.1. Definition and relevance Dynamic hysteresis (or dynamic force hysteresis) represents the difference between the advancing and receding forces of a surface in a given solution chemistry. This parameter indicates the physicochemical surface properties of a substrate, such as chemical composition, roughness, swelling, chemical heterogeneity, adsorption, desorption, energy level of surface electrons, and surface configuration changes [19,20]. For RO membranes, the difference between the advancing and receding forces reflects the imperfection of surface properties (e.g., chemical heterogeneity, roughness, or mobility of functional groups or adsorbed molecules) or a capacity of the RO membrane surface to be reoriented or swell when it comes into contact with various test solutions [20,21]. This implies that if dynamic hysteresis approaches a value of zero, the substrate is chemically and physically uniform. Increase in dynamic hysteresis could be caused by the alteration of numerous surface properties, such as surface roughness, chemical composition, zeta potential, and the degree of hydration. In addition, the low surface energy of the RO membrane can increase dynamic hysteresis [20,21]. In this study, the term ‘dynamic hysteresis’ is slightly different from the conventional term of ‘dynamic contact angle hysteresis’, which is usually determined by the Wilhelmy plate method [22–28]. Dynamic contact angle hysteresis is also directly related to the difference in forces acting on the liquid–solid interfaces during the advancing and receding of a solid sample in a test liquid. During the receding phase, however, a thin residual liquid film remains on the solid surface and adds some weight to the force during dynamic contact angle measurements, thereby causing an error in the estimated receding contact angle. Since the liquid film remains on the surface during the receding phase (i.e., representing complete wetting), the surface tension of a test liquid is often overestimated and, thus, the receding contact angle goes to 0◦ . This means that dynamic contact angle hysteresis often tends to account for only the advancing force acting on the solid–liquid interface. To circumvent this problem, Lee et al. [18] and Keller et al. [22] proposed that the use of ‘dynamic (force) hysteresis’ instead of ‘dynamic contact angle hysteresis’ could be more appropriate when the effect of the liquid residue during the receding procedure is taken into
account. More details on this account are discussed in our previous study [18]. 2.2. Governing equations Dynamic hysteresis is calculated by extrapolating the corresponding immersion and emersion force per unit length (F/L) lines to zero immersion depth. The tensiometric method for measuring contact angle determines the force that is present when an RO membrane sample is brought into contact with a liquid. If the forces of interaction, geometry of the substrate, and surface tension of the liquid are known, the dynamic hysteresis can be calculated. First, the surface tension of the liquid using the du Nouy ring method [22–24] has to be measured. To test the RO membrane sample, the membrane is first hung on a balance. The liquid is then raised until it contacts the substrate (membrane sample). When this contact occurs, the change in forces is detected and a microbalance registers the elevation as zero depth of immersion. As the substrate is pushed into the liquid, the forces on the balance are recorded. The forces on the balance are: Ftotal = wetting force + weight of probe − buoyancy
(1)
The weight of the probe is tared and the effect of the buoyancy force is removed by extrapolating the graph back to zero depth of immersion. The wetting force is composed of the advancing and receding forces, which are defined as [18,20]: FA = W + P cos A
(2)
FR = W + P cos R − B
(3)
where W is the weight of the membrane sample, P the perimeter of the membrane sample, the surface tension of the test liquid, A and R the advancing and receding contact angles between test liquid and membrane sample, respectively, and B is the buoyancy force. Since the effect of the buoyancy force is eliminated (discussed above), the difference between the receding and advancing forces can be defined as dynamic hysteresis [18,19,21–23]. Dynamic hysteresis = FA /P − FR /P
(4)
Here, the measured F is divided by membrane sample perimeter, P, to express the dynamic hysteresis as force per unit length (F/L, mN/m) [18,21]. In a graphical analysis, the dynamic hysteresis is the vertical width between the parallel advancing and receding force lines (expressed as F/L) versus the moving distance, as illustrated in Fig. 1. 3. Materials and methods 3.1. RO membranes Four commercial seawater RO membranes were used. The RO membranes were RE-8040 (Woongjin Chemical), SW-30HR (Dow Chemical), TM-820 (Toray), and SWC-5 (Hydranautics). All membranes were thin film composite (TFC) polyamide membranes with an average salt rejection over 99.5%. General properties of the membranes in terms of zeta potential, contact angle, and surface roughness were determined and are listed in Table 1. For storage, the membranes were immersed in deionized (DI) water at 4 ◦ C and water was replaced regularly. The membranes were cut into strips measuring 15 mm in width and 20 mm in length, and had a thickness of 0.5 mm. Prior to each analysis, the membranes were put into 20–25 ◦ C DI water for 2 h as all analyses in this study were carried out at room temperature.
S. Lee et al. / Journal of Membrane Science 366 (2011) 17–24
19
Fig. 1. Measurement of dynamic hysteresis: (a) Wilhelmy plate method and (b) example of experimental data determined during the advancing and receding of the probe to illustrate the concept of dynamic hysteresis.
3.2. Solution chemistry The ionic strength of the solutions was controlled by NaCl (Fisher Scientific). NaCl was added to DI water (D7429-33, Easy Pure RO system, Lab Science, Korea) and was mixed for 1 h at room temperature (20–25 ◦ C). When needed, solution pH was adjusted with 0.1 M of NaOH or HCl. Stock solutions of EDTA (ethylenediaminetetraacetic acid) and divalent cations (i.e., CaCl2 ) were prepared at concentrations of 0.1 mM. Suwannee River natural organic matter (NOM) (IHSS) and alginate (alginic acid sodium salt (Sigma–Aldrich)) were used as model organic foulants at concentrations of 20 and 500 mg/L, respectively. All test solutions were freshly prepared 1 h before each experiment and disposed of after the experiments. 3.3. Dynamic hysteresis As described in Section 2, dynamic force hysteresis was determined based on the Wilhelmy plate method [22–28]. During the measurement, a membrane sample was held by an automated microbalance, then pushed into or pulled out of a test liquid. The measurements were carried out using a Sigma 701 microbalance (KSV Instruments, Ltd., Finland) interfaced with a PC for automatic control and data acquisition. A liquid cell containing a test liquid moved up and down at a constant speed rate repeatedly during the measurements. All parameters employed during the force hysteresis measurements are listed in our previous study [18]. The surface tension of each test liquid was measured by the du Nouy ring method at test liquid temperatures of 20–22 ◦ C and atmosphere temperatures of 20–24 ◦ C, with a humidity of 19–37% [18,21]. The ring was rinsed with ethyl alcohol prior to each measurement. More details on the method and instrument are given in our previous study [18]. 3.4. Static contact angle Contact angle measurements were performed with a goniometer (DM 500, Kyowa Interface Science, Japan). Equilibrium contact angle measurements, as described by Marmur [29], were adopted. Table 1 Characteristics of RO membranes (measured at pH 6.5–7.0 and a temperature of 20–25 ◦ C). Membrane
Contact anglea (◦ )
Zeta potentialb (mV)
Roughnessc (nm)
TM-820 RE-8040 SWC-5 SW-30HR
79.0 (±5.2) 73.6 (±7.6) 72.4 (±4.9) 24.0 (±3.4)
−18.3 (±2.2) −24.0 (±1.3) −21.2 (±1.6) −30.1 (±3.1)
91.5 (±9.2) 86.1 (±13.7) 97.0 (±11.8) 103.4 (±15.5)
a b c
Determined by a contact angle goniometer. Determined in a background electrolyte of 10 mM KCl and pH 7. Determined by AFM imaging.
The equilibrium contact angle was the average of the left and right contact angles. Ten measurements were conducted for each membrane and the reported values are the average of the 10 equilibrium contact angles. 3.5. Zeta potential Membrane zeta potential was analyzed by a streaming current electrokinetic analyzer (SurPASS, Anton Paar GmbH, Graz, Austria) following the procedure described by Luxbacher [30]. Zeta potential values were obtained based on the Fairbrother and Mastin equation [31]. For streaming potential measurement, 0.01 M KCl was used as a background electrolyte solution and solution pH was varied from 2 to 10. The operating pressure ranged from 0 to 500 mbar and the temperature was about 25 ◦ C. 3.6. Surface roughness and heterogeneity Membrane surface roughness was analyzed by AFM imaging (PUCOStation AFM, Surface Imaging Systems, Herzogenrath, Germany). Liquid phase AFM imaging was conducted in contact mode with silicon probes, the backsides of which had a 30-nm-thick aluminum reflex coating layer for better resolution and stability in liquid phase applications (AppNano, Applied NanoStructures, Inc., Santa Clara, CA). The probe had a spring constant of 0.1 N/m (±0.08 N/m), resonance frequency of 28 kHz (±10 kHz), tip radius of 5–6 nm, tip height of 14 m (±2 m), and cantilever length of 225 m (±10 m). The RO membranes were immersed in a liquid cell containing pre-adjusted test solution in terms of pH and ionic strength. All membranes were scanned three times at randomly selected scan positions. The surface roughness of each membrane was quantified as the root mean square (RMS) roughness, which is the RMS deviation of the peaks and valleys from the mean plane. Approaching force ranged from 4.0 to 6.0 N/m with a scan speed of 0.7 line/s and a scan area of 10 m × 10 m. Scanned images were analyzed using SPIP software (Surface Imaging Systems, Herzogenrath, Germany). Each image was flattened by a baseline prior to roughness analyses. In addition, surface morphological heterogeneity was determined from the scanned images by calculating the density of summit that reflects the heterogeneity in terms of the distribution of peak and valley structure on the membrane surface [18,32]. 3.7. Lab-scale RO fouling test A laboratory-scale cross-flow RO membrane test unit, similar to that described in our earlier publications [6,14,33], was used for the fouling experiments. Flux-decline test conditions were 10 mM NaCl, pH 7, temperature of 20 ◦ C, total operating time of 300 min, initial flux of 12.75 m/s, and pressure of 20 bar.
20
S. Lee et al. / Journal of Membrane Science 366 (2011) 17–24
Fig. 2. Membrane zeta potential as a function of solution pH. Experiments were carried out with 10 mM KCl as a background electrolyte and a temperature of 25 ◦ C.
4. Results and discussion 4.1. Dynamic hysteresis and chemical surface heterogeneity 4.1.1. Determination of chemical surface heterogeneity by dynamic hysteresis The zeta potential of the four membranes was determined for eight different solution pH conditions (from pH 2 to 10) and displayed in Fig. 2. It can be seen that the SW-30HR membrane exhibits
Fig. 4. Conceptual description of the changes in charge distribution of a homogeneous surface with respect to solution pH. This description corresponds to the SW-30HR membrane, which exhibits no isoelectric point as shown in Fig. 2.
a negative charge throughout the whole pH range, while the other three membranes have an isoelectric point between pH 3.5 and 4.5. These zeta potential values are correlated with the corresponding dynamic hysteresis values in Fig. 3. It is shown that dynamic hysteresis changes with variations in solution pH. This trend can be explained by the heterogeneity of the functional groups on the membrane surface.
Fig. 3. Variation of the dynamic hysteresis of RO membranes with respect to solution pH: (a) SW-30HR, (b) RE-8040, (c) SWC-5, and (d) TM-820.
S. Lee et al. / Journal of Membrane Science 366 (2011) 17–24
21
Fig. 5. Conceptual description of the changes in charge distribution of a heterogeneous surface with respect to solution pH. This description corresponds to the RE-8040, SWC-5, and TM-820 membranes, which exhibit isoelectric points as shown in Fig. 2.
In the case of SW-30HR, the surface is mainly composed of acidic functional groups, making the membrane negatively charged regardless of the pH conditions. Since the charge is distributed evenly across the surface, the dynamic hysteresis values of the SW30HR are comparatively low and uniform. It can also be seen that the dynamic hysteresis values decrease with increasing solution pH. As the solution pH (and zeta potential) increases, the number of negatively charged functional groups across the membrane surface also increases, resulting in more uniform charge distribution and, consequently, lower dynamic hysteresis values. This behavior is illustrated schematically in Fig. 4. At low pH values, the membrane surface is partially uncharged, and, as the pH increases, the
uncharged acidic functional groups deprotonate and become negatively charged making the surface much more uniform. Unlike the SW-30HR, the other membranes displayed different behavior, with the dynamic hysteresis values fluctuating near the isoelectric point. These fluctuations can be related to the chemical properties of the membranes. When the isoelectric point of the membrane is approached, positively charged functional groups deprotonate, imparting to the membrane surface a net neutral charge. The chemical homogeneity of the membrane surface then increases, thereby decreasing the dynamic hysteresis as illustrated in Fig. 5. Above the isoelectric point, the surface functional groups continue to deprotonate, making the surface charge partially neg-
Fig. 6. Influence of anionic surfactant (SDS) on the changes in dynamic hysteresis of RO membranes: (a) SW-30HR, (b) TM-820, (c) SWC-5, and (d) RE-8040.
22
S. Lee et al. / Journal of Membrane Science 366 (2011) 17–24
ative. As the pH further increases, the surface functional groups continue this process until the membrane surface becomes fully deprotonated at high pH, making it chemically homogeneous with all sites being negatively charged. These features are also observed with the dynamic hysteresis values in Fig. 3. In the case of the RE-8040 membrane (Fig. 3(b)), dynamic hysteresis decreases at the isoelectric point of the membrane (pH 3.5–4.5). As the pH increases beyond the isoelectric point, the dynamic hysteresis starts to increase because of the chemical heterogeneity, while at high pH values, when the membrane surface is fully deprotonated, the dynamic hysteresis decreases again. Similar trends can also be seen with the other membranes (Figs. 3(c) and (d)), implying that chemical heterogeneity, such as charge distribution, can be evaluated by a simple measurement of dynamic hysteresis.
4.2.1. Influence of membrane–NOM interaction on dynamic hysteresis To evaluate the degree of foulant adsorption and the resulting effects on the values of the dynamic hysteresis, two membranes (SW-30HR, TM-820) were tested with solutions containing NOM, divalent cations, and EDTA. Membranes were first tested in DI water for reference (blank), and then in DI water containing 20 mg/L of NOM to examine the effect of NOM adsorption on dynamic hysteresis. The effects of divalent cations (i.e., CaCl2 ) and EDTA were also tested to further evaluate the role of solution chemistry. The speed parameter was fixed at 0.36 mm/min to prevent the speed of the plate from interfering with the interaction between the foulants and membranes [18,34]. Results showed that dynamic hysteresis decreased with NOM addition (Fig. 8). When the membranes were advanced into the liquid, the NOM adsorbed onto the surface, which screened the chemical heterogeneity and reduced the dynamic hysteresis values. When divalent cations were added, the NOM adsorption was enhanced, further lowering the dynamic hysteresis. This is because the divalent cations interact with humic carboxyl groups and reduce the charge of the NOM. Calcium ions also initiate bridging between NOM and the membrane surface, which enhances NOM adsorption [33]. On the contrary, when a chelating agent, such as EDTA (ethylenediaminetetraacetic acid), is added to the solution, it preferentially reacts with divalent cations, restoring dynamic hysteresis to the condition where only NOM is presented. In conclusion, these results show that adsorption of foulants to the membranes
impacts dynamic hysteresis, which may shed light on the degree of membrane fouling. 4.2.2. Relating dynamic hysteresis to membrane fouling Bench-scale fouling tests with alginate as a model foulant were performed to investigate whether dynamic hysteresis can be used to predict the fouling propensity of RO membranes. The concentration of alginate was set to 500 mg/L to accelerate fouling and make it easier to distinguish the initial fouling behavior. In practical RO operations, chemical cleaning is performed when water flux declines by 15–20% of the initial flux. Hence, this study focuses on the initial stage of fouling, where the interaction between the foulant and the membrane surface predominates [6].
Dynamic hysteresis (mN/m)
4.2. Dynamic hysteresis and membrane fouling
Fig. 7. Reduction in the dynamic hysteresis of RO membranes due to adsorption of SDS molecules and subsequent masking of charge heterogeneity.
40
SW-30HR
35 30 25 20 15
Fouling
10 Blank
Dynamic hysteresis (mN/m)
4.1.2. Influence of SDS on the chemical surface heterogeneity and dynamic hysteresis By using surface-active molecules, such as surfactants, the membrane chemical surface heterogeneity can be altered, which impacts dynamic hysteresis. The dynamic hysteresis of the four RO membranes was measured in solutions containing different concentrations of sodium dodecyl sulfate (SDS). SDS is an anionic surfactant with a hydrophobic tail and hydrophilic head group that can be adsorbed on the membrane surface. Figs. 6 and 7 demonstrate that the dynamic hysteresis values for the membranes decrease with increasing SDS concentration. This implies that adsorbed SDS molecules can decrease the membrane chemical surface heterogeneity. As the concentration of the SDS solution increased, SDS molecules were abundantly adsorbed on the membrane surface and, consequently, dynamic hysteresis decreased. This behavior is attributed to the masking of surface charge heterogeneity of the RO membranes by the adsorbed SDS molecules. Therefore, it is confirmed that the membrane surface charge behavior, as described in the previous subsection, plays an important role in controlling the chemical surface heterogeneity of RO membranes.
NOM
Cleaning
(a)
NOM+ NOM+Ca2+ Ca2+ +EDTA
40
TM-820
35 30
Fouling
25
Cleaning
20 15
(b)
10 Blank
NOM
NOM+ NOM+Ca2+ Ca2+ +EDTA
Fig. 8. Influence of NOM adsorption and desorption on the variation of dynamic hysteresis: (a) SW-30HR and (b) TM-820. Fouling and cleaning processes with respect to calcium and EDTA addition, respectively, are described.
S. Lee et al. / Journal of Membrane Science 366 (2011) 17–24
23
Fig. 9. Flux reduction of RO membranes determined from bench-scale NOM fouling runs. The percent flux reduction was measured 30 min after the initiation of the fouling run with 500 mg/L alginate. Other experimental conditions were: initial flux of 15 m/s, cross-flow velocity of 8.5 cm/s, background electrolyte of 10 mM NaCl, and temperature of 20 ◦ C.
Fig. 9 presents the flux reduction for the tested membranes taken 30 min after the initiation of the fouling run. The SWC-5 membrane showed the highest flux reduction followed by the TM820, RE-8040, and SW-30HR membranes. The results indicate that the initial flux decline is closely related to the dynamic hysteresis. Fig. 10 shows the relation between the flux reduction and dynamic hysteresis, as well as the relation between the flux reduction and zeta potential. It can be seen that the dynamic hysteresis has a better correlation with the flux decline than the membrane zeta potential. Since the zeta potential only shows the average of the membrane surface charge, not the distribution, a membrane with high zeta potential but heterogeneous charge will be more easily
Fig. 11. Conceptual description of the foulant–membrane interaction with respect to heterogeneity in the charge distribution on the membrane surface.
fouled compared to a membrane with a uniform charge distribution. An illustration of this phenomenon is shown in Fig. 11. Accordingly, this observation suggests that charge distribution is more important than simple zeta potential measurements when analyzing RO membrane fouling.
5. Conclusion The use of dynamic hysteresis as a novel tool for characterizing RO membranes and evaluating their fouling propensity was investigated. For characterizing the chemical surface characteristics of RO membranes, dynamic hysteresis can be used as a measure of the distribution of surface charge on the membrane. It was shown that dynamic hysteresis varied noticeably for the tested RO membranes, even for membranes with similar average zeta potential. In addition, the applicability of using dynamic hysteresis to assess the fouling propensity of RO membranes was investigated. The results showed that NOM adsorption was favorable to surfaces with heterogeneous charge distribution. This was further confirmed by alginate fouling experiments where RO membranes with greater flux reduction exhibited higher dynamic hysteresis values. A remarkable correlation between flux reduction and dynamic hysteresis was obtained. Based on this study, it is clearly demonstrated that dynamic hysteresis can be a simple and reliable tool for characterizing membrane surfaces as well as assessing membrane fouling propensity.
Acknowledgements
Fig. 10. Relating flux reduction to (a) dynamic hysteresis and (b) zeta potential of RO membranes. Regression coefficient (R2 ) is included.
This research was supported by the Seawater Engineering & Architecture of High Efficiency Reverse Osmosis (SEAHERO) program supported by the Ministry of Land, Transport and Maritime Affairs (MLTM) and partly by the World Class University (WCU) program (Case III) through the National Research Foundation of Korea, which is funded by the Ministry of Education, Science and Technology (R33-10046).
24
S. Lee et al. / Journal of Membrane Science 366 (2011) 17–24
References [1] C. Jucker, M.M. Clark, Adsorption of aquatic humic substances on hydrophobic ultrafiltration membranes, J. Membr. Sci. 97 (1994) 37–52. [2] 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) 253–268. [3] C.-C. Ho, A.L. Zydney, Effect of membrane morphology on the initial rate of protein fouling during microfiltration, J. Membr. Sci. 155 (1999) 261– 275. [4] Y. Shim, H.J. Lee, S. Lee, S.H. Moon, J. Cho, Effects of natural organic matter and ionic species on membrane surface charge, Environ. Sci. Technol. 36 (2002) 3864–3871. [5] S. Al-Jeshi, A. Neville, An investigation into the relationship between flux and roughness on RO membranes using scanning probe microscopy, Desalination 189 (2006) 221–228. [6] S. Lee, M. Elimelech, Relating organic fouling of reverse osmosis membranes to intermolecular adhesion forces, Environ. Sci. Technol. 40 (2006) 980–987. [7] K. Boussu, A. Belpaire, A. Volodin, C. Van Haesendonck, P. Van der Meeren, C. Vandecasteele, B. Van der Bruggen, Influence of membrane and colloid characteristics on fouling of nanofiltration membranes, J. Membr. Sci. 280 (2007) 220–230. [8] J. Pieracci, J.V. Crivello, G. Belfort, Photochemical modification of 10 kDa polyethersulfone ultrafiltration membranes for reduction of biofouling, J. Membr. Sci. 156 (1999) 223–240. [9] V. Freger, J. Gilron, S. Belfer, TFC polyamide membranes modified by grafting of hydrophilic polymers: an FT-IR/AFM/TEM study, J. Membr. Sci. 209 (2002) 283–292. [10] S. Belfer, R. Fainshtain, Y. Purinson, J. Gilron, M. Nyström, M. Mänttäri, Modification of NF membrane properties by in situ redox initiated graft polymerization with hydrophilic monomers, J. Membr. Sci. 239 (2004) 55–64. [11] L.-F. Liu, S.-C. Yu, Y. Zhou, C.-J. Gao, Study on a novel polyamide–urea reverse osmosis composite membrane (ICIC–MPD). I. Preparation and characterization of ICIC–MPD membrane, J. Membr. Sci. 281 (2006) 88–94. [12] B.J.A. Tarboush, D. Rana, T. Matsuura, H.A. Arafat, R.M. Narbaitz, Preparation of thin-film-composite polyamide membranes for desalination using novel hydrophilic surface modifying macromolecules, J. Membr. Sci. 325 (2008) 166–175. [13] J.-H. Li, Y.-Y. Xu, L.-P. Zhu, J.-H. Wang, C.-H. Du, Fabrication and characterization of a novel TiO2 nanoparticle self-assembly membrane with improved fouling resistance, J. Membr. Sci. 326 (2009) 659–666. [14] J. Yang, S. Lee, E. Lee, J. Lee, S.K. Hong, Effect of solution chemistry on the surface property of reverse osmosis membranes under seawater conditions, Desalination 247 (2009) 148–161. [15] A.E. Childress, S. Shivaji, Deshmukh, Effect of humic substances and anionic surfactants on the surface charge and performance of reverse osmosis membranes, Desalination 118 (1998) 167–174. [16] E.M.V. Hoek, S. Bhattacharjee, M. Elimelech, Effect of membrane surface roughness on colloid–membrane DLVO interactions, Langmuir 19 (2003) 4836–4847.
[17] J.A. Brant, K.M. Johnson, A.E. Childress, Examining the electrochemical properties of a nanofiltration membrane with atomic force microscopy, J. Membr. Sci. 276 (2006) 286–294. [18] E. Lee, S. Lee, S.K. Hong, A new approach to the characterization of reverse osmosis membrane by dynamic hysteresis, Desalinat. Water Treat. 18 (2010) 257–263. [19] F. Rupp, L. Scheideler, J. Geis-Gerstorfer, Effect of heterogenic surfaces on contact angle hysteresis: dynamic contact angle analysis in material sciences, Chem. Eng. Technol. 25 (2002) 9–15. [20] T. Kasemura, S. Takahashi, N. Nakane, T. Maegawa, Surface dynamics for poly (vinyl alkylates) via dynamic contact angle and adhesion tension relaxation, Polymer 37 (1996) 3659–3664. [21] C.M. Weikart, M. Miyama, H.K. Yasuda, Surface modification of conventional polymers by depositing plasma polymers of trimethylsilane and of trimethylsilane + O2 , J. Colloid Interface Sci. 211 (1999) 28–38. [22] A.A. Keller, V. Broje, K. Setty, Effect of advancing velocity and fluid viscosity on the dynamic contact angle of petroleum hydrocarbons, J. Petrol. Sci. Eng. 58 (2007) 201–206. [23] X. Xie, N.R. Morrow, J.S. Buckley, Contact angle hysteresis and the stability of wetting changes induced by adsorption from crude oil, J. Petrol. Sci. Eng. 33 (2002) 147–159. [24] A.M. Gaudin, T.G. Decker, Contact angles and adsorption in the system quartz water–dodecane modified by dodecyl ammonium chloride, J. Colloid Interface Sci. 24 (1967) 151–158. [25] A. Nabe, E. Staude, G. Belfort, Surface modification of polysulfone ultrafiltration membranes and fouling by BSA solutions, J. Membr. Sci. 133 (1997) 57–72. [26] H. Kamusewitz, W. Possart, D. Paul, The relation between Young’s equilibrium contact angle and the hysteresis on rough paraffin wax surfaces, Colloids Surf. A: Physicochem. Eng. Aspects 156 (1999) 271–279. [27] S.-H. Lee, C.-S. Lee, D.-S. Shin, B.-G. Kim, Y.-S. Lee, Y.-K. Kim, Micro protein patterning using a lift-off process with fluorocarbon thin film, Sens. Actuators B 99 (2004) 623–632. [28] C. Brunot, B. Grosgogeat, C. Picart, C. Lagneau, N. Jaffrezic-Renault, L. Ponsonnet, Response of fibroblast activity and polyelectrolyte multilayer films coating titanium, Dent. Mater. 24 (2008) 1025–1035. [29] A. Marmur, Equilibrium contact angles: theory and measurement, Colloid Surf. A 116 (1996) 55–61. [30] T. Luxbacher, Electrokinetic characterization of flat sheet membranes by streaming current measurement, Desalination 199 (2006) 376–377. [31] F. Fairbrother, H. Mastin, Studies in electro-endosmosis. Part I, J. Chem. Soc. Trans. 125 (1924) 2319–2330. [32] K.J. Stout, P.J. Sullivan, W.P. Dong, E. Mainsah, N. Luo, T. Mathia, H. Zahouani, The development of methods for the characterization of roughness on three dimensions. Publication no. EUR 15178 EN of the Commission of the European Communities, Luxembourg, 1994. [33] Y. Yu, S. Lee, S.K. Hong, Effect of solution chemistry on organic fouling of reverse osmosis membranes in seawater desalination, J. Membr. Sci. 351 (2010) 205–213. [34] M. Morra, E. Occhiello, F. Garbassi, On the wettability of poly (2hydroxyrthylmethacrylate), J. Colloid Interface Sci. 149 (1992) 1.
Contents lists available at ScienceDirect
Journal of Membrane Science journal homepage: www.elsevier.com/locate/memsci
Membrane characterization by dynamic hysteresis: Measurements, mechanisms, and implications for membrane fouling Sangyoup Lee a , Eunsu Lee a , Menachem Elimelech b , Seungkwan Hong a,∗ a b
School of Civil, Environmental & Architectural Engineering, Korea University, 1, 5-ka, Anam-Dong, Sungbuk-Gu, Seoul 136-713, Republic of Korea Department of Chemical Engineering, Environmental Engineering Program, Yale University, New Haven, CT 06520-8286, USA
a r t i c l e
i n f o
Article history: Received 17 March 2010 Received in revised form 15 September 2010 Accepted 17 September 2010 Available online 25 September 2010 Keywords: Dynamic hysteresis RO membrane Surface heterogeneity Charge distribution Membrane fouling
a b s t r a c t The surface characteristics of reverse osmosis membranes and their relation to membrane fouling are systematically investigated by measuring membrane dynamic hysteresis based on the Wilhelmy plate method. Dynamic hysteresis represents the difference between the forces applied to a membrane surface when it is advanced into and withdrawn from a liquid or solution. Our results demonstrate that the chemical surface heterogeneity of various RO membranes could be quantified by measuring their dynamic hysteresis. The chemical heterogeneity was mostly related to the distribution of surface charge rather than average zeta potential. There was a remarkable correlation between the chemical surface heterogeneity and membrane dynamic hysteresis. It was clearly shown that dynamic hysteresis varied substantially with respect to the solution chemistry of test solutions. The dynamic hysteresis of RO membranes measured in the presence of organic foulants was further related to the flux-decline rate determined from bench-scale fouling experiments. It was found that higher flux-decline rate was obtained for RO membranes with larger dynamic hysteresis. Based on the results in this study, it is demonstrated that dynamic hysteresis measurements can be a promising tool for characterizing membrane surfaces as well as assessing membrane fouling. © 2010 Elsevier B.V. All rights reserved.
1. Introduction Reverse osmosis (RO) membranes are currently being used in a wide range of applications, including brackish/seawater desalination, drinking water treatment, and wastewater reuse. Efficient operation of RO membranes in these applications, however, is often hampered by membrane fouling. Since RO membrane fouling is a surface phenomenon, surface characteristics of RO membranes play a major role in membrane fouling. For this reason, there have been numerous studies investigating the surface characteristics of RO membranes and relating these characteristics to the rate and extent of membrane fouling [1–7]. Consequently, there have been significant advances in membrane fabrication technology, especially in surface modification, where membrane surface properties have been tailored toward reducing membrane fouling as well as enhancing membrane permeability [8–13]. Surface characteristics affecting membrane fouling can be divided into chemical and physical characteristics. The former mainly includes surface charge and hydrophobicity, where zeta potential and contact angle measurements are the major relevant characterization tools, respectively. Physical characteristics
∗ Corresponding author. Tel.: +82 2 3290 3322; fax: +82 2 928 7656. E-mail address: [email protected] (S. Hong). 0376-7388/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.memsci.2010.09.024
often refer to surface roughness, which is commonly determined by atomic force microscopy (AFM). Morphological surface heterogeneity (i.e., variations in the distribution of peak and valley structure) also belongs to physical surface characteristics [14]. Based on these surface characteristics and their relation to membrane fouling, it has been found that the propensity for fouling increases for membranes that are less negatively charged, more hydrophobic, and rougher [1,3,4,7,15–17]. However, these characterization methods are time-consuming and require relatively expensive instruments. Furthermore, it is difficult to isolate the effect of these surface characteristics. For example, changes in surface charge upon variation of feed water solution chemistry could result in the alteration of surface hydrophobicity and perhaps morphological surface roughness [14]. Therefore, a simple analytic technique to characterize various surface characteristics and the interplay between them could be a great asset to the study of membrane surface characterization. Recently, it has been demonstrated that dynamic hysteresis could be a useful parameter to characterize the physical surface properties of RO membranes [18]. Dynamic hysteresis values of various RO membranes were related to both the average surface roughness and the roughness distribution (i.e., morphological surface heterogeneity). It was shown that dynamic hysteresis is more closely related to the distribution of peak and valley structure of roughness than the average roughness. Furthermore, it has
18
S. Lee et al. / Journal of Membrane Science 366 (2011) 17–24
been shown that dynamic hysteresis measurements can be performed with various test liquids to reflect solution chemistry. By using a non-polar liquid (i.e., diiodomethane), the interference of chemical surface characteristics with physical surface characteristics could be excluded and, thus, a more accurate correlation between dynamic hysteresis and physical surface heterogeneity was obtained. In that study, however, the chemical aspects of dynamic hysteresis were not investigated. The main objective of this study is to determine the chemical surface characteristics of RO membranes by dynamic hysteresis and relate these characteristics to membrane fouling propensity. The chemical surface characteristics of RO membranes, including surface charge and its distribution, were investigated first. The distribution of surface charge was characterized by varying solution pH and anionic surfactant (sodium dodecyl sulfate) concentration during dynamic hysteresis measurements. Next, the membrane dynamic hysteresis was determined in the presence of organic foulants and the results were related to flux-decline data obtained from lab-scale RO fouling experiments. This study delineates, for the first time, the use of dynamic hysteresis for characterizing the chemical surface characteristics of RO membranes and the direct relationship to membrane fouling.
2. Background 2.1. Definition and relevance Dynamic hysteresis (or dynamic force hysteresis) represents the difference between the advancing and receding forces of a surface in a given solution chemistry. This parameter indicates the physicochemical surface properties of a substrate, such as chemical composition, roughness, swelling, chemical heterogeneity, adsorption, desorption, energy level of surface electrons, and surface configuration changes [19,20]. For RO membranes, the difference between the advancing and receding forces reflects the imperfection of surface properties (e.g., chemical heterogeneity, roughness, or mobility of functional groups or adsorbed molecules) or a capacity of the RO membrane surface to be reoriented or swell when it comes into contact with various test solutions [20,21]. This implies that if dynamic hysteresis approaches a value of zero, the substrate is chemically and physically uniform. Increase in dynamic hysteresis could be caused by the alteration of numerous surface properties, such as surface roughness, chemical composition, zeta potential, and the degree of hydration. In addition, the low surface energy of the RO membrane can increase dynamic hysteresis [20,21]. In this study, the term ‘dynamic hysteresis’ is slightly different from the conventional term of ‘dynamic contact angle hysteresis’, which is usually determined by the Wilhelmy plate method [22–28]. Dynamic contact angle hysteresis is also directly related to the difference in forces acting on the liquid–solid interfaces during the advancing and receding of a solid sample in a test liquid. During the receding phase, however, a thin residual liquid film remains on the solid surface and adds some weight to the force during dynamic contact angle measurements, thereby causing an error in the estimated receding contact angle. Since the liquid film remains on the surface during the receding phase (i.e., representing complete wetting), the surface tension of a test liquid is often overestimated and, thus, the receding contact angle goes to 0◦ . This means that dynamic contact angle hysteresis often tends to account for only the advancing force acting on the solid–liquid interface. To circumvent this problem, Lee et al. [18] and Keller et al. [22] proposed that the use of ‘dynamic (force) hysteresis’ instead of ‘dynamic contact angle hysteresis’ could be more appropriate when the effect of the liquid residue during the receding procedure is taken into
account. More details on this account are discussed in our previous study [18]. 2.2. Governing equations Dynamic hysteresis is calculated by extrapolating the corresponding immersion and emersion force per unit length (F/L) lines to zero immersion depth. The tensiometric method for measuring contact angle determines the force that is present when an RO membrane sample is brought into contact with a liquid. If the forces of interaction, geometry of the substrate, and surface tension of the liquid are known, the dynamic hysteresis can be calculated. First, the surface tension of the liquid using the du Nouy ring method [22–24] has to be measured. To test the RO membrane sample, the membrane is first hung on a balance. The liquid is then raised until it contacts the substrate (membrane sample). When this contact occurs, the change in forces is detected and a microbalance registers the elevation as zero depth of immersion. As the substrate is pushed into the liquid, the forces on the balance are recorded. The forces on the balance are: Ftotal = wetting force + weight of probe − buoyancy
(1)
The weight of the probe is tared and the effect of the buoyancy force is removed by extrapolating the graph back to zero depth of immersion. The wetting force is composed of the advancing and receding forces, which are defined as [18,20]: FA = W + P cos A
(2)
FR = W + P cos R − B
(3)
where W is the weight of the membrane sample, P the perimeter of the membrane sample, the surface tension of the test liquid, A and R the advancing and receding contact angles between test liquid and membrane sample, respectively, and B is the buoyancy force. Since the effect of the buoyancy force is eliminated (discussed above), the difference between the receding and advancing forces can be defined as dynamic hysteresis [18,19,21–23]. Dynamic hysteresis = FA /P − FR /P
(4)
Here, the measured F is divided by membrane sample perimeter, P, to express the dynamic hysteresis as force per unit length (F/L, mN/m) [18,21]. In a graphical analysis, the dynamic hysteresis is the vertical width between the parallel advancing and receding force lines (expressed as F/L) versus the moving distance, as illustrated in Fig. 1. 3. Materials and methods 3.1. RO membranes Four commercial seawater RO membranes were used. The RO membranes were RE-8040 (Woongjin Chemical), SW-30HR (Dow Chemical), TM-820 (Toray), and SWC-5 (Hydranautics). All membranes were thin film composite (TFC) polyamide membranes with an average salt rejection over 99.5%. General properties of the membranes in terms of zeta potential, contact angle, and surface roughness were determined and are listed in Table 1. For storage, the membranes were immersed in deionized (DI) water at 4 ◦ C and water was replaced regularly. The membranes were cut into strips measuring 15 mm in width and 20 mm in length, and had a thickness of 0.5 mm. Prior to each analysis, the membranes were put into 20–25 ◦ C DI water for 2 h as all analyses in this study were carried out at room temperature.
S. Lee et al. / Journal of Membrane Science 366 (2011) 17–24
19
Fig. 1. Measurement of dynamic hysteresis: (a) Wilhelmy plate method and (b) example of experimental data determined during the advancing and receding of the probe to illustrate the concept of dynamic hysteresis.
3.2. Solution chemistry The ionic strength of the solutions was controlled by NaCl (Fisher Scientific). NaCl was added to DI water (D7429-33, Easy Pure RO system, Lab Science, Korea) and was mixed for 1 h at room temperature (20–25 ◦ C). When needed, solution pH was adjusted with 0.1 M of NaOH or HCl. Stock solutions of EDTA (ethylenediaminetetraacetic acid) and divalent cations (i.e., CaCl2 ) were prepared at concentrations of 0.1 mM. Suwannee River natural organic matter (NOM) (IHSS) and alginate (alginic acid sodium salt (Sigma–Aldrich)) were used as model organic foulants at concentrations of 20 and 500 mg/L, respectively. All test solutions were freshly prepared 1 h before each experiment and disposed of after the experiments. 3.3. Dynamic hysteresis As described in Section 2, dynamic force hysteresis was determined based on the Wilhelmy plate method [22–28]. During the measurement, a membrane sample was held by an automated microbalance, then pushed into or pulled out of a test liquid. The measurements were carried out using a Sigma 701 microbalance (KSV Instruments, Ltd., Finland) interfaced with a PC for automatic control and data acquisition. A liquid cell containing a test liquid moved up and down at a constant speed rate repeatedly during the measurements. All parameters employed during the force hysteresis measurements are listed in our previous study [18]. The surface tension of each test liquid was measured by the du Nouy ring method at test liquid temperatures of 20–22 ◦ C and atmosphere temperatures of 20–24 ◦ C, with a humidity of 19–37% [18,21]. The ring was rinsed with ethyl alcohol prior to each measurement. More details on the method and instrument are given in our previous study [18]. 3.4. Static contact angle Contact angle measurements were performed with a goniometer (DM 500, Kyowa Interface Science, Japan). Equilibrium contact angle measurements, as described by Marmur [29], were adopted. Table 1 Characteristics of RO membranes (measured at pH 6.5–7.0 and a temperature of 20–25 ◦ C). Membrane
Contact anglea (◦ )
Zeta potentialb (mV)
Roughnessc (nm)
TM-820 RE-8040 SWC-5 SW-30HR
79.0 (±5.2) 73.6 (±7.6) 72.4 (±4.9) 24.0 (±3.4)
−18.3 (±2.2) −24.0 (±1.3) −21.2 (±1.6) −30.1 (±3.1)
91.5 (±9.2) 86.1 (±13.7) 97.0 (±11.8) 103.4 (±15.5)
a b c
Determined by a contact angle goniometer. Determined in a background electrolyte of 10 mM KCl and pH 7. Determined by AFM imaging.
The equilibrium contact angle was the average of the left and right contact angles. Ten measurements were conducted for each membrane and the reported values are the average of the 10 equilibrium contact angles. 3.5. Zeta potential Membrane zeta potential was analyzed by a streaming current electrokinetic analyzer (SurPASS, Anton Paar GmbH, Graz, Austria) following the procedure described by Luxbacher [30]. Zeta potential values were obtained based on the Fairbrother and Mastin equation [31]. For streaming potential measurement, 0.01 M KCl was used as a background electrolyte solution and solution pH was varied from 2 to 10. The operating pressure ranged from 0 to 500 mbar and the temperature was about 25 ◦ C. 3.6. Surface roughness and heterogeneity Membrane surface roughness was analyzed by AFM imaging (PUCOStation AFM, Surface Imaging Systems, Herzogenrath, Germany). Liquid phase AFM imaging was conducted in contact mode with silicon probes, the backsides of which had a 30-nm-thick aluminum reflex coating layer for better resolution and stability in liquid phase applications (AppNano, Applied NanoStructures, Inc., Santa Clara, CA). The probe had a spring constant of 0.1 N/m (±0.08 N/m), resonance frequency of 28 kHz (±10 kHz), tip radius of 5–6 nm, tip height of 14 m (±2 m), and cantilever length of 225 m (±10 m). The RO membranes were immersed in a liquid cell containing pre-adjusted test solution in terms of pH and ionic strength. All membranes were scanned three times at randomly selected scan positions. The surface roughness of each membrane was quantified as the root mean square (RMS) roughness, which is the RMS deviation of the peaks and valleys from the mean plane. Approaching force ranged from 4.0 to 6.0 N/m with a scan speed of 0.7 line/s and a scan area of 10 m × 10 m. Scanned images were analyzed using SPIP software (Surface Imaging Systems, Herzogenrath, Germany). Each image was flattened by a baseline prior to roughness analyses. In addition, surface morphological heterogeneity was determined from the scanned images by calculating the density of summit that reflects the heterogeneity in terms of the distribution of peak and valley structure on the membrane surface [18,32]. 3.7. Lab-scale RO fouling test A laboratory-scale cross-flow RO membrane test unit, similar to that described in our earlier publications [6,14,33], was used for the fouling experiments. Flux-decline test conditions were 10 mM NaCl, pH 7, temperature of 20 ◦ C, total operating time of 300 min, initial flux of 12.75 m/s, and pressure of 20 bar.
20
S. Lee et al. / Journal of Membrane Science 366 (2011) 17–24
Fig. 2. Membrane zeta potential as a function of solution pH. Experiments were carried out with 10 mM KCl as a background electrolyte and a temperature of 25 ◦ C.
4. Results and discussion 4.1. Dynamic hysteresis and chemical surface heterogeneity 4.1.1. Determination of chemical surface heterogeneity by dynamic hysteresis The zeta potential of the four membranes was determined for eight different solution pH conditions (from pH 2 to 10) and displayed in Fig. 2. It can be seen that the SW-30HR membrane exhibits
Fig. 4. Conceptual description of the changes in charge distribution of a homogeneous surface with respect to solution pH. This description corresponds to the SW-30HR membrane, which exhibits no isoelectric point as shown in Fig. 2.
a negative charge throughout the whole pH range, while the other three membranes have an isoelectric point between pH 3.5 and 4.5. These zeta potential values are correlated with the corresponding dynamic hysteresis values in Fig. 3. It is shown that dynamic hysteresis changes with variations in solution pH. This trend can be explained by the heterogeneity of the functional groups on the membrane surface.
Fig. 3. Variation of the dynamic hysteresis of RO membranes with respect to solution pH: (a) SW-30HR, (b) RE-8040, (c) SWC-5, and (d) TM-820.
S. Lee et al. / Journal of Membrane Science 366 (2011) 17–24
21
Fig. 5. Conceptual description of the changes in charge distribution of a heterogeneous surface with respect to solution pH. This description corresponds to the RE-8040, SWC-5, and TM-820 membranes, which exhibit isoelectric points as shown in Fig. 2.
In the case of SW-30HR, the surface is mainly composed of acidic functional groups, making the membrane negatively charged regardless of the pH conditions. Since the charge is distributed evenly across the surface, the dynamic hysteresis values of the SW30HR are comparatively low and uniform. It can also be seen that the dynamic hysteresis values decrease with increasing solution pH. As the solution pH (and zeta potential) increases, the number of negatively charged functional groups across the membrane surface also increases, resulting in more uniform charge distribution and, consequently, lower dynamic hysteresis values. This behavior is illustrated schematically in Fig. 4. At low pH values, the membrane surface is partially uncharged, and, as the pH increases, the
uncharged acidic functional groups deprotonate and become negatively charged making the surface much more uniform. Unlike the SW-30HR, the other membranes displayed different behavior, with the dynamic hysteresis values fluctuating near the isoelectric point. These fluctuations can be related to the chemical properties of the membranes. When the isoelectric point of the membrane is approached, positively charged functional groups deprotonate, imparting to the membrane surface a net neutral charge. The chemical homogeneity of the membrane surface then increases, thereby decreasing the dynamic hysteresis as illustrated in Fig. 5. Above the isoelectric point, the surface functional groups continue to deprotonate, making the surface charge partially neg-
Fig. 6. Influence of anionic surfactant (SDS) on the changes in dynamic hysteresis of RO membranes: (a) SW-30HR, (b) TM-820, (c) SWC-5, and (d) RE-8040.
22
S. Lee et al. / Journal of Membrane Science 366 (2011) 17–24
ative. As the pH further increases, the surface functional groups continue this process until the membrane surface becomes fully deprotonated at high pH, making it chemically homogeneous with all sites being negatively charged. These features are also observed with the dynamic hysteresis values in Fig. 3. In the case of the RE-8040 membrane (Fig. 3(b)), dynamic hysteresis decreases at the isoelectric point of the membrane (pH 3.5–4.5). As the pH increases beyond the isoelectric point, the dynamic hysteresis starts to increase because of the chemical heterogeneity, while at high pH values, when the membrane surface is fully deprotonated, the dynamic hysteresis decreases again. Similar trends can also be seen with the other membranes (Figs. 3(c) and (d)), implying that chemical heterogeneity, such as charge distribution, can be evaluated by a simple measurement of dynamic hysteresis.
4.2.1. Influence of membrane–NOM interaction on dynamic hysteresis To evaluate the degree of foulant adsorption and the resulting effects on the values of the dynamic hysteresis, two membranes (SW-30HR, TM-820) were tested with solutions containing NOM, divalent cations, and EDTA. Membranes were first tested in DI water for reference (blank), and then in DI water containing 20 mg/L of NOM to examine the effect of NOM adsorption on dynamic hysteresis. The effects of divalent cations (i.e., CaCl2 ) and EDTA were also tested to further evaluate the role of solution chemistry. The speed parameter was fixed at 0.36 mm/min to prevent the speed of the plate from interfering with the interaction between the foulants and membranes [18,34]. Results showed that dynamic hysteresis decreased with NOM addition (Fig. 8). When the membranes were advanced into the liquid, the NOM adsorbed onto the surface, which screened the chemical heterogeneity and reduced the dynamic hysteresis values. When divalent cations were added, the NOM adsorption was enhanced, further lowering the dynamic hysteresis. This is because the divalent cations interact with humic carboxyl groups and reduce the charge of the NOM. Calcium ions also initiate bridging between NOM and the membrane surface, which enhances NOM adsorption [33]. On the contrary, when a chelating agent, such as EDTA (ethylenediaminetetraacetic acid), is added to the solution, it preferentially reacts with divalent cations, restoring dynamic hysteresis to the condition where only NOM is presented. In conclusion, these results show that adsorption of foulants to the membranes
impacts dynamic hysteresis, which may shed light on the degree of membrane fouling. 4.2.2. Relating dynamic hysteresis to membrane fouling Bench-scale fouling tests with alginate as a model foulant were performed to investigate whether dynamic hysteresis can be used to predict the fouling propensity of RO membranes. The concentration of alginate was set to 500 mg/L to accelerate fouling and make it easier to distinguish the initial fouling behavior. In practical RO operations, chemical cleaning is performed when water flux declines by 15–20% of the initial flux. Hence, this study focuses on the initial stage of fouling, where the interaction between the foulant and the membrane surface predominates [6].
Dynamic hysteresis (mN/m)
4.2. Dynamic hysteresis and membrane fouling
Fig. 7. Reduction in the dynamic hysteresis of RO membranes due to adsorption of SDS molecules and subsequent masking of charge heterogeneity.
40
SW-30HR
35 30 25 20 15
Fouling
10 Blank
Dynamic hysteresis (mN/m)
4.1.2. Influence of SDS on the chemical surface heterogeneity and dynamic hysteresis By using surface-active molecules, such as surfactants, the membrane chemical surface heterogeneity can be altered, which impacts dynamic hysteresis. The dynamic hysteresis of the four RO membranes was measured in solutions containing different concentrations of sodium dodecyl sulfate (SDS). SDS is an anionic surfactant with a hydrophobic tail and hydrophilic head group that can be adsorbed on the membrane surface. Figs. 6 and 7 demonstrate that the dynamic hysteresis values for the membranes decrease with increasing SDS concentration. This implies that adsorbed SDS molecules can decrease the membrane chemical surface heterogeneity. As the concentration of the SDS solution increased, SDS molecules were abundantly adsorbed on the membrane surface and, consequently, dynamic hysteresis decreased. This behavior is attributed to the masking of surface charge heterogeneity of the RO membranes by the adsorbed SDS molecules. Therefore, it is confirmed that the membrane surface charge behavior, as described in the previous subsection, plays an important role in controlling the chemical surface heterogeneity of RO membranes.
NOM
Cleaning
(a)
NOM+ NOM+Ca2+ Ca2+ +EDTA
40
TM-820
35 30
Fouling
25
Cleaning
20 15
(b)
10 Blank
NOM
NOM+ NOM+Ca2+ Ca2+ +EDTA
Fig. 8. Influence of NOM adsorption and desorption on the variation of dynamic hysteresis: (a) SW-30HR and (b) TM-820. Fouling and cleaning processes with respect to calcium and EDTA addition, respectively, are described.
S. Lee et al. / Journal of Membrane Science 366 (2011) 17–24
23
Fig. 9. Flux reduction of RO membranes determined from bench-scale NOM fouling runs. The percent flux reduction was measured 30 min after the initiation of the fouling run with 500 mg/L alginate. Other experimental conditions were: initial flux of 15 m/s, cross-flow velocity of 8.5 cm/s, background electrolyte of 10 mM NaCl, and temperature of 20 ◦ C.
Fig. 9 presents the flux reduction for the tested membranes taken 30 min after the initiation of the fouling run. The SWC-5 membrane showed the highest flux reduction followed by the TM820, RE-8040, and SW-30HR membranes. The results indicate that the initial flux decline is closely related to the dynamic hysteresis. Fig. 10 shows the relation between the flux reduction and dynamic hysteresis, as well as the relation between the flux reduction and zeta potential. It can be seen that the dynamic hysteresis has a better correlation with the flux decline than the membrane zeta potential. Since the zeta potential only shows the average of the membrane surface charge, not the distribution, a membrane with high zeta potential but heterogeneous charge will be more easily
Fig. 11. Conceptual description of the foulant–membrane interaction with respect to heterogeneity in the charge distribution on the membrane surface.
fouled compared to a membrane with a uniform charge distribution. An illustration of this phenomenon is shown in Fig. 11. Accordingly, this observation suggests that charge distribution is more important than simple zeta potential measurements when analyzing RO membrane fouling.
5. Conclusion The use of dynamic hysteresis as a novel tool for characterizing RO membranes and evaluating their fouling propensity was investigated. For characterizing the chemical surface characteristics of RO membranes, dynamic hysteresis can be used as a measure of the distribution of surface charge on the membrane. It was shown that dynamic hysteresis varied noticeably for the tested RO membranes, even for membranes with similar average zeta potential. In addition, the applicability of using dynamic hysteresis to assess the fouling propensity of RO membranes was investigated. The results showed that NOM adsorption was favorable to surfaces with heterogeneous charge distribution. This was further confirmed by alginate fouling experiments where RO membranes with greater flux reduction exhibited higher dynamic hysteresis values. A remarkable correlation between flux reduction and dynamic hysteresis was obtained. Based on this study, it is clearly demonstrated that dynamic hysteresis can be a simple and reliable tool for characterizing membrane surfaces as well as assessing membrane fouling propensity.
Acknowledgements
Fig. 10. Relating flux reduction to (a) dynamic hysteresis and (b) zeta potential of RO membranes. Regression coefficient (R2 ) is included.
This research was supported by the Seawater Engineering & Architecture of High Efficiency Reverse Osmosis (SEAHERO) program supported by the Ministry of Land, Transport and Maritime Affairs (MLTM) and partly by the World Class University (WCU) program (Case III) through the National Research Foundation of Korea, which is funded by the Ministry of Education, Science and Technology (R33-10046).
24
S. Lee et al. / Journal of Membrane Science 366 (2011) 17–24
References [1] C. Jucker, M.M. Clark, Adsorption of aquatic humic substances on hydrophobic ultrafiltration membranes, J. Membr. Sci. 97 (1994) 37–52. [2] 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) 253–268. [3] C.-C. Ho, A.L. Zydney, Effect of membrane morphology on the initial rate of protein fouling during microfiltration, J. Membr. Sci. 155 (1999) 261– 275. [4] Y. Shim, H.J. Lee, S. Lee, S.H. Moon, J. Cho, Effects of natural organic matter and ionic species on membrane surface charge, Environ. Sci. Technol. 36 (2002) 3864–3871. [5] S. Al-Jeshi, A. Neville, An investigation into the relationship between flux and roughness on RO membranes using scanning probe microscopy, Desalination 189 (2006) 221–228. [6] S. Lee, M. Elimelech, Relating organic fouling of reverse osmosis membranes to intermolecular adhesion forces, Environ. Sci. Technol. 40 (2006) 980–987. [7] K. Boussu, A. Belpaire, A. Volodin, C. Van Haesendonck, P. Van der Meeren, C. Vandecasteele, B. Van der Bruggen, Influence of membrane and colloid characteristics on fouling of nanofiltration membranes, J. Membr. Sci. 280 (2007) 220–230. [8] J. Pieracci, J.V. Crivello, G. Belfort, Photochemical modification of 10 kDa polyethersulfone ultrafiltration membranes for reduction of biofouling, J. Membr. Sci. 156 (1999) 223–240. [9] V. Freger, J. Gilron, S. Belfer, TFC polyamide membranes modified by grafting of hydrophilic polymers: an FT-IR/AFM/TEM study, J. Membr. Sci. 209 (2002) 283–292. [10] S. Belfer, R. Fainshtain, Y. Purinson, J. Gilron, M. Nyström, M. Mänttäri, Modification of NF membrane properties by in situ redox initiated graft polymerization with hydrophilic monomers, J. Membr. Sci. 239 (2004) 55–64. [11] L.-F. Liu, S.-C. Yu, Y. Zhou, C.-J. Gao, Study on a novel polyamide–urea reverse osmosis composite membrane (ICIC–MPD). I. Preparation and characterization of ICIC–MPD membrane, J. Membr. Sci. 281 (2006) 88–94. [12] B.J.A. Tarboush, D. Rana, T. Matsuura, H.A. Arafat, R.M. Narbaitz, Preparation of thin-film-composite polyamide membranes for desalination using novel hydrophilic surface modifying macromolecules, J. Membr. Sci. 325 (2008) 166–175. [13] J.-H. Li, Y.-Y. Xu, L.-P. Zhu, J.-H. Wang, C.-H. Du, Fabrication and characterization of a novel TiO2 nanoparticle self-assembly membrane with improved fouling resistance, J. Membr. Sci. 326 (2009) 659–666. [14] J. Yang, S. Lee, E. Lee, J. Lee, S.K. Hong, Effect of solution chemistry on the surface property of reverse osmosis membranes under seawater conditions, Desalination 247 (2009) 148–161. [15] A.E. Childress, S. Shivaji, Deshmukh, Effect of humic substances and anionic surfactants on the surface charge and performance of reverse osmosis membranes, Desalination 118 (1998) 167–174. [16] E.M.V. Hoek, S. Bhattacharjee, M. Elimelech, Effect of membrane surface roughness on colloid–membrane DLVO interactions, Langmuir 19 (2003) 4836–4847.
[17] J.A. Brant, K.M. Johnson, A.E. Childress, Examining the electrochemical properties of a nanofiltration membrane with atomic force microscopy, J. Membr. Sci. 276 (2006) 286–294. [18] E. Lee, S. Lee, S.K. Hong, A new approach to the characterization of reverse osmosis membrane by dynamic hysteresis, Desalinat. Water Treat. 18 (2010) 257–263. [19] F. Rupp, L. Scheideler, J. Geis-Gerstorfer, Effect of heterogenic surfaces on contact angle hysteresis: dynamic contact angle analysis in material sciences, Chem. Eng. Technol. 25 (2002) 9–15. [20] T. Kasemura, S. Takahashi, N. Nakane, T. Maegawa, Surface dynamics for poly (vinyl alkylates) via dynamic contact angle and adhesion tension relaxation, Polymer 37 (1996) 3659–3664. [21] C.M. Weikart, M. Miyama, H.K. Yasuda, Surface modification of conventional polymers by depositing plasma polymers of trimethylsilane and of trimethylsilane + O2 , J. Colloid Interface Sci. 211 (1999) 28–38. [22] A.A. Keller, V. Broje, K. Setty, Effect of advancing velocity and fluid viscosity on the dynamic contact angle of petroleum hydrocarbons, J. Petrol. Sci. Eng. 58 (2007) 201–206. [23] X. Xie, N.R. Morrow, J.S. Buckley, Contact angle hysteresis and the stability of wetting changes induced by adsorption from crude oil, J. Petrol. Sci. Eng. 33 (2002) 147–159. [24] A.M. Gaudin, T.G. Decker, Contact angles and adsorption in the system quartz water–dodecane modified by dodecyl ammonium chloride, J. Colloid Interface Sci. 24 (1967) 151–158. [25] A. Nabe, E. Staude, G. Belfort, Surface modification of polysulfone ultrafiltration membranes and fouling by BSA solutions, J. Membr. Sci. 133 (1997) 57–72. [26] H. Kamusewitz, W. Possart, D. Paul, The relation between Young’s equilibrium contact angle and the hysteresis on rough paraffin wax surfaces, Colloids Surf. A: Physicochem. Eng. Aspects 156 (1999) 271–279. [27] S.-H. Lee, C.-S. Lee, D.-S. Shin, B.-G. Kim, Y.-S. Lee, Y.-K. Kim, Micro protein patterning using a lift-off process with fluorocarbon thin film, Sens. Actuators B 99 (2004) 623–632. [28] C. Brunot, B. Grosgogeat, C. Picart, C. Lagneau, N. Jaffrezic-Renault, L. Ponsonnet, Response of fibroblast activity and polyelectrolyte multilayer films coating titanium, Dent. Mater. 24 (2008) 1025–1035. [29] A. Marmur, Equilibrium contact angles: theory and measurement, Colloid Surf. A 116 (1996) 55–61. [30] T. Luxbacher, Electrokinetic characterization of flat sheet membranes by streaming current measurement, Desalination 199 (2006) 376–377. [31] F. Fairbrother, H. Mastin, Studies in electro-endosmosis. Part I, J. Chem. Soc. Trans. 125 (1924) 2319–2330. [32] K.J. Stout, P.J. Sullivan, W.P. Dong, E. Mainsah, N. Luo, T. Mathia, H. Zahouani, The development of methods for the characterization of roughness on three dimensions. Publication no. EUR 15178 EN of the Commission of the European Communities, Luxembourg, 1994. [33] Y. Yu, S. Lee, S.K. Hong, Effect of solution chemistry on organic fouling of reverse osmosis membranes in seawater desalination, J. Membr. Sci. 351 (2010) 205–213. [34] M. Morra, E. Occhiello, F. Garbassi, On the wettability of poly (2hydroxyrthylmethacrylate), J. Colloid Interface Sci. 149 (1992) 1.
