Influence of the properties of layer-by-layer active layers on forward osmosis performance

Journal of Membrane Science 423-424 (2012) 536–542

Contents lists available at SciVerse ScienceDirect

Journal of Membrane Science journal homepage: www.elsevier.com/locate/memsci

Influence of the properties of layer-by-layer active layers on forward osmosis performance Saren Qi a,b, Weiyi Li a,b, Yang Zhao a,b, Ning Ma a,b, Jing Wei a,b, Ting Wei Chin a,b, Chuyang Y. Tang a,b,n a b

School of Civil and Environmental Engineering, Nanyang Technological University, Singapore 639798, Singapore Singapore Membrane Technology Centre, Nanyang Technological University, Singapore 639798, Singapore

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 Accepted 1 September 2012 Available online 10 September 2012

The advancement of forward osmosis (FO) technology requires separation membranes with appropriate transport characteristics. The layer-by-layer (LbL) method exhibits great flexibility for fabricating the active layer of FO membrane with controllable separation properties. The current work focused on investigating the effect of LbL active layer properties on the FO performance. A series of FO membranes were prepared with varied number of polyelectrolyte bilayers which were composed of positively charged poly(allylamine hydrochloride) (PAH) and negatively charged poly(sodium 4-styrene-sulfonate) (PSS), with either PAH or PSS as the terminating layer. The active layers were characterized in terms of contact angle, surface roughness, and zeta potential, which were exploited to explain the variations of the intrinsic transport properties (the hydraulic permeability and the solute permeability) of the polyelectrolyte multilayer films. FO filtration experiments were carried out to assess the performance of the same series of FO membranes. Both the filtration flux and the FO efficiency were demonstrated as a strong function of the LbL active layers. This dependency was rationalized by analyzing the relative importance of the different transport mechanisms during the FO processes, which were inherently correlated to the intrinsic transport properties of the multilayer films. The current investigation not only justifies the feasibility of improving the FO performance by properly controlling the number of the polyelectrolyte bilayers and the surface charge, but also makes the underlying mechanisms comprehensible. & 2012 Elsevier B.V. All rights reserved.

Keywords: Forward osmosis membrane Layer-by-layer adsorption Membrane characterization FO performance optimization

1. Introduction Forward osmosis (FO) is an emergent membrane separation technology with potential applications in desalination [1–3], wastewater treatment [4], food processing [5], membrane bioreactor [6,7], and so on. One of the key factors limiting the development of FO technology is the fabrication of high performance FO membranes. Similar to conventional membranes for pressure-driven membrane processes (e.g., reverse osmosis and ultrafiltration), FO membranes are composed of a thin selective layer (active layer) and a porous support layer. However, the osmotically driven FO suffers from the loss of driving force caused by internal concentration polarization (ICP), i.e., the severe change of the solute concentration within the support layer. Previous studies [8–10] indicate that the extent of the ICP during FO processes strongly depends on the transport properties of FO

n Corresponding author at: School of Civil and Environmental Engineering, Nanyang Technological University, 50 Nanyang Avenue N1-1b-35, Singapore 639798, Singapore. Tel.: þ 65 6790 5267; fax: þ65 6791 0676. E-mail address: [email protected] (C.Y. Tang).

0376-7388/$ - see front matter & 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.memsci.2012.09.009

membranes, which could be tuned by changing the microstructures of both the active layer and support layer. As one of the electrostatic self-assembly (ESA) techniques [11,12], the layer-by-layer (LbL) method has been successfully applied to the fabrication of FO membranes [13–15]. In this method, the active layer is formed by depositing oppositely charged polyelectrolytes in an alternate sequence onto a porous substrate. The LbL method has the flexibility in controlling the permeation properties of the active layer with more degrees of freedom compared to the conventional phase inversion method [13]. In previous studies, attention was mainly given to the physicochemical properties of the LbL films, e.g., the surface water wettability [16], the zeta potential [17], and the solute diffusivity [18]. However, these studies are unable to provide direct information for optimizing the LbL active layer for enhancing the FO performance owing to the highly nonlinear transport phenomena in the support layer. The objective of the current study was to correlate the properties of the LbL FO membranes to the filtration performance during FO processes. Lab-scale FO membrane specimens were prepared by LbL method. A series of characterization experiments were carried out to determine the properties of the active layers as a function

S. Qi et al. / Journal of Membrane Science 423-424 (2012) 536–542

of the number of the polyelectrolyte bilayers. Based on these characterization results, the effects of the LbL layer active layers on the FO performance will be analyzed in conjunction with the FO filtration experimental data, and implications on optimizing the LbL method for FO membrane will be discussed. 2. Experimental 2.1. Membrane fabrication The polyacrylonitrile (PAN) substrate was fabricated using the phase inversion method as introduced in our previous work [13–15]. Specifically, the polymer solution was prepared by dissolving PAN (weight averaged molecular weight Mw 150,000, SigmaAldrich) and Lithium chloride (LiCl, anhydrous, MP Biomed) into N,N-dimethylformamide (DMF, Z99.8%, Sigma-Aldrich) with a mass ratio of 18:2:80. The degassed polymer solution was cast onto a flat glass plate by a stainless steel casting knife (Elcometer Pte Ltd, Asia) with a gate height of 150 mm. Then, the cast polymer film was immediately immersed into tap water at room temperature  20 1C, and the porous polymer structure was formed via the phase separation. After the immersion precipitation, the PAN polymer film was washed by fresh tap water followed by a rinse of DI water. In order to impart the negative charge and enhance the hydrophilicity, the prepared PAN substrate was soaked in the alkali solution of 1.5 M NaOH (anhydrous, pellets Z98%, Sigma-Aldrich) at 45 1C for 1.5 h. Specially, the PAN substrate with NaOH treatment was designated as PAN-OH substrate [13,14]. Besides, the hydrophilicity of the PAN-OH substrate [19] and its finger-like porous structure (see Fig. A1 in Appendix and [13–15]) also help to improve the mass transfer inside the substrate. The polycation and polyanion solutions were prepared by dissolving poly(allylamine hydrochloride) (PAH, average Mw  112,000 to 200,000, Polyscience) and poly(sodium 4-styrene-sulfonate) (PSS, average Mw  70,000 30 wt% in water, Sigma-Aldrich), respectively, into NaCl (99%, Merck) solution of 0.5 M. Then, the polyelectrolyte bilayers were formed by alternately exposing the surface of negatively charged PAN-OH substrate to the PAH and PSS solutions. The duration of each exposure in the polyelectrolyte solution was 30 min, and each adsorption was followed by a rinse of DI water for one minute. As indicated by our previous studies [13,14], the electrolyte rejection of the LbL skin layer might be substantially reduced in high ionic strength surroundings owing to the Donnan exclusion effect. To reduce such adverse effect, the chemical crosslinking of LbL active layer with glutaraldehyde (GA, 25% in water, Sigma-Aldrich) was performed following our previous study [14]. In particular, we denoted the crosslinked LbL membranes by xLbLn, in which n represents the number of the electrolyte bilayers. According to this notation, for example, xLbL3.0 has 3 PAH/PSS bilayers and the outmost layer (terminating layer) is PSS; xLbL3.5 has one more PAH coating (i.e., half of a bilayer) in addition to the 3 PAH/PSS bilayers, with PAH being the terminating layer. In the current study, we prepared three PSS-terminated (T-PSS) active layers (xLbL1.0, xLbL2.0, and xLbL3.0) and four PAH-terminated (T-PAH) active layers (xLbL0.5, xLbL1.5, xLbL2.5, and xLbL3.5). 2.2. Membrane characterization Three different measurements were implemented for determining the surface characteristics of the prepared xLbL active layers. First, an OCA contact angle system (DataPhysics Instruments GmbH, Germany) was employed to measure the contact angle on the surface of the outmost polyelectrolyte films. Specifically, the sessile drop method [13,20] was adopted for the contact angle measurement, and the measurement was repeated

537

14 times for individual membrane specimens. The surface roughness was detected using an atomic force microscope (AFM, Park system, Korea) with the noncontact mode. The scan area was 25 mm2, and the scan rate was varied from 0.5 to 1.0 Hz. The measured roughness for each membrane sample was the average value from 4 different samples. Finally, the zeta potential was measured with an eletrokinetic analyzer (EKA, SurPASS, Anton Paar GmbH, Austria) using a 10 mM NaCl background electrolyte. The zeta potential values were given by the best fittings with the Helmoltz–Smoluchowski equation [13], and the measurement for each membrane sample was repeated 4 times. The intrinsic transport properties of the xLbL active layers were measured by performing the reverse osmosis (RO) filtrations. The bench-scale filtration setup was equipped with the membrane cell of effective area 42 cm2. Prior to each measurement, the membrane specimens were compacted with an applied pressure of 5 bars for at least 3 h. In the measurements for the hydraulic permeability A, pure water was used as the feed to determine the bulk flux JV as a function of the transmembrane hydraulic pressures DP. Then, the value of A was obtained by the linear fitting of JV versus DP. As revealed in the literature [1,8], the effect of the ICP on RO processes could be neglected since RO processes are mainly driven by the hydraulic pressure difference. Therefore, the values of the solute permeability B for a variety electrolyte solutes were obtained by fitting the RO filtration experimental results with the formula B¼A(DP Dp)(1/R 1) [13], where Dp is the osmotic pressure difference across the membrane, and R is the apparent solute rejection based on conductivity measurement (Ultrameter II, Myron L Company, Carlsbad, CA). Four electrolyte solutions were employed as the feed solution: 5 mM MgCl2, 7.5 mM MgSO4, 5 mM Na2SO4, and 7.5 mM NaCl (all were purchased from Merck). All the filtrations were run with an applied pressure of 5 bars, and a cross flow rate of  20 cm/s for the feed stream was adopted to minimize the external concentration polarization (ECP). 2.3. Evaluation of FO performance The FO filtration was implemented by applying the draw solution (DS, the phase of high osmotic pressure) and the feed solution (FS, the phase of low osmotic pressure) to the different sides of the FO membrane, and the bulk flow (water) was thus driven from the FS side to the DS side. In the current filtration experiments, the feed solution was DI water while the draw solution was a series of solitary electrolyte solutions, including MgCl2, MgSO4, and Na2SO4. With the aid of the thermophysical modeling software (OLI Stream Analyzer 3.1, Morris Plains, NJ), the concentration of the each electrolyte was varied so that the osmotic pressure holds the value of 38 bars. The FO membranes were accommodated in an FO membrane cell with an effective area of 42 cm2, and both DS and FS streams were pumped into the cell with a counter-current mode. The cross flow velocities for both streams were set at  18.75 cm/s so as to minimize the effect of ECP [13]. The change of the FS weight was recorded as a function of filtration time, which was then used to determine the bulk flux. In particular, two membrane orientations were adopted, i.e., the orientation with the active layer exposed to the draw solution (AL-DS) and the orientation with the active layer facing the feed solution (AL-FS). All these FO filtration experiments were repeated 3 times for the same operating conditions.

3. Results and discussion 3.1. Surface characteristics of xLbL active layers The surface characteristics of the xLbL active layers are of particular value to envisage the transport processes occurring at

538

S. Qi et al. / Journal of Membrane Science 423-424 (2012) 536–542

70 60 Contact angle (°)

relatively smooth surface compared to the T-PAH films with a similar thickness. The results of the zeta potential measurement are shown in Fig. 1(c). The value of the zeta potential is a measure of the electric potential difference between the mobile diffusion layer on the charged surface and the bulk solution, thereby reflecting the surface electric properties. According to Fig. 1(c), the sign of the surface charge was reversed, and a positive value of  37 mV was yielded when the first polycationic PAH layer was deposited on the negatively charged PAN substrate (i.e., membrane xLbL0.5). After depositing the next negatively charged PSS, the zeta potential of xLbL1.0 was reduced to  5 mV. As depositing more layers of the polyelectrolyte film, oscillatory changes of the surface potential around the value of 10 mV were displayed with amplitudes about 10 mV.

T-PAH T-PSS Substrate

50 40 30 20 70

Rougness (nm)

60 50 40

3.2. Intrinsic transport properties of xLbL active layers

30 20 10 0

Zeta potential (mV)

40 30 20 10 0 -30 0.0

0.5

1.0

1.5 2.0 Bilayer number

2.5

3.0

3.5

The intrinsic transport properties of the xLbL active layers are relevant to study the global osmotic behaviors of the FO membrane. In line with the classical solution–diffusion model [23], the transport nature of the polyelectrolyte multilayer films is characterized by the hydraulic permeability A and the solute permeability B. Both the values of A and B were measured by the RO filtration experiments, and plotted in Fig. 2 with the bilayer number on the abscissa. Similarly, the results of the PAH-terminated and PSSterminated films are distinguished by the solid and void symbols, respectively. The values of A for the various active layers were obtained from the filtration experiments with pure water and displayed in Fig. 2(a). It illustrates that the values of A were reduced as increasing the bilayer number for the active layers with the same type of terminating layer. This is consistent with the fact that the 6

Fig. 1. Surface characterization of the xLbL active layers with different bilayer number: (a) the contact angle, (b) the surface roughness and (c) the zeta potential.

A(L/m2 h bar)

5 4 3 2 1 0 100

T-PAH MgCl2 MgSO4 Na2SO4 NaCl

80

B (10-7 m/s)

the interface between the electrolyte solutions and the polyelectrolyte films. Our current work of surface characterization was focused on detecting the variation of the surface hydrophilicity (contact angle), surface roughness, and the surface electric property (zeta potential) upon increasing the number of the polyelectrolyte bilayers as well as the type of terminating layer (i.e., either PAH or PSS). All the characterization results were plotted in Fig. 1 as a function of the number of the bilayers, which was varied from 0.5 to 3.5. The results for the T-PAH active layers are designated by the solid squares whereas the void squares represent the data for the T-PSS active layers. Specially, the experimental results for the virgin substrate are also displayed by the void circles as a reference for comparative study. Referring to Fig. 1(a), the contact angles of T-PSS membranes were significantly lower compared to those of T-PAH membranes. This observation is consistent with the previous studies [16,21] revealing that negatively charged sulfo group usually possess higher hydrophilicity due to the high tendency to form hydrogen bonding. As increasing the number of the bilayers, the contact angle on the positively charged surface (T-PAH) is appreciably increased from  351 to  601. In contrast, the change in contact angle values for the PSS-terminated films was less than 101. Membrane surface roughness increased with the increased number of polyelectrolyte layers (Fig. 1b), which could be attributed to the random adsorption of the polyelectrolytes [22]. As increasing bilayer number, the roughness had an increasing tendency for both T-PAH and T-PSS. It is also noted that T-PSS gave rise to a

T-PAH T-PSS

60

T-PSS

40 20 0

0.5

1.0

1.5 2.0 2.5 Bilayer number

3.0

3.5

Fig. 2. Intrinsic transport properties of the xLbL active layers with different bilayer number: (a) the hydraulic permeability A and (b) the solute permeability B.

S. Qi et al. / Journal of Membrane Science 423-424 (2012) 536–542

539

thickness of the active layer is increased as depositing more polymer films, thereby engendering higher overall hydraulic resistance. When comparing the data from the active layers with different terminating layers, we note that the T-PSS films yielded higher values of A than the T-PAH films with similar number of bilayer. This periodic dependence of the A value on the terminal layer type is directly supported by the experimental results of the contact angle measurement as shown in Fig. 1(a), which indicate substantially higher degree of water wettability for the T-PSS membranes. The preferential partition of water into the active layer facilitates the reduction of the transfer resistance at the solution–polymer interface, thus enhancing the global conduction. The plot in Fig. 2(a) also indicates that further growth of the multilayer films might eventually suppress this periodic variation, since the frictional resistance likely plays a more important role for the thicker rejection layers (2.5–3.5 bilayers) compared to surface wettability. Fig. 2(b) demonstrates the experimental results of the solute permeability B, which describes the diffusion rate of the solute across the active layer. Four different electrolyte solutions were employed as the feed stream in the RO filtrations for determining the values of B. As revealed in the literature [18,24], the transfer of ions in charged membranes is governed by both the steric and Donnan exclusions, which respectively reflect the frictional and electric interactions between the ions and the membrane structure. Therefore, the electrolytes were deliberately chosen so that two cations (Na þ and Mg2 þ ) and two anions (Cl  and SO24  ) with different valences and hydration radii would be covered in this investigation. The expected intricacies concerning the different ion transport mechanisms were reflected by the experimental results in Fig. 2(b). According to the values of B for the initial deposition of PAH (xLbL0.5), the rejection of the salts was primarily dictated by the constituent cations in the solution. Specifically, the salts containing magnesium ions had relatively low permeabilities in contrast to those having sodium ions. As discovered by the experimental results of the zeta potential (Fig. 1c), coating the first layer of PAH onto the virgin PAN substrate resulted in a positive surface potential of the active layer. It is perhaps for this reason that the divalent magnesium ions were more severely repelled by the active layer compared to the monovalent sodium ions. A close inspection also reveals the impact of the geometrical characteristics of the anions on the diffusion of the salts sharing the same cations. It shows that the solute transfer rates were slightly retarded by the sulfate ions having a larger hydration shell (278 pm [25]) than the chloride ions (224 pm [25]). When the PAH top layers were covered by the negatively charged PSS, the salts with divalent anion SO24  were consistently better rejected compared to the salts with monovalent Cl  . It is also noted that the contributions of the constituent cations to the salt rejection were varied with the different anions. In the case of the salts with SO24  , Donnan exclusion was the key factor controlling the solute transfer. Therefore, MgSO4 exhibited a higher value of B in comparison with that of NaSO4 since Mg2 þ was more effective to screen the electrostatic interaction between the anion and the negatively charged PSS. In contrast, Donnan exclusion played a less important role for the salts with Cl  , whereas the steric effect was significant. This explains the observation that MgCl2 was better rejected than NaCl owing to the relatively large hydrated ion size of Mg2 þ (299 pm [25]).

driving force during FO processes. As indicated in previous studies [10,23], the ICP process inside the support layer of an FO is affected by the support layer structure (i.e. structural parameter S). Nevertheless, the rejection layer can also affect ICP directly by a phenomenon known as reverse diffusion induced ICP [13,26–28]. In the AL-DS orientation, draw solutes can diffuse through the active layer and accumulate inside the support layer to cause a severe ICP (even if the feed solution contains no dissolved solutes). In the AL-FS orientation, the leakage of draw solutes through the active layer causes reduced the solute concentration inside the support layer and thus enhances the degree of ICP. Therefore, one of the major tasks of optimizing the FO performance was to investigate the dependence of the ICP during the FO processes on the variations of the xLbL active layer. The ICP process in the support layer is heavily affected by the relative importance of the water transport (i.e., the bulk phase drift in the context of dilute solutions) and the solute transport. For osmotically driven processes, this relative importance is dictated by the transport nature of the active layer, and can be quantified by the Peclet number for the active layer Pea, which was defined in our previous study [29] as ADp/B. When the value of Pea is much greater than unity, it implies that the water transport might be the dominant mechanism in the porous substrate. The experimentally determined values of Pea for the various active layers were plotted in Fig. 3(a). In spite of the fluctuation arising from the variation of the terminating layers, it appears that the values of Pea for all the electrolyte solutes were getting greater as increasing the bilayer number. The values of Pea directly reflected the ratio of A to B since the transmembrane osmotic pressure was fixed at the same value for all the FO filtration experiments (Dp ¼38 bars). Although the similar decreasing tendency was yielded for both A and

3.3. Effect of xLbL active layers on FO performance

Fig. 3. Variations of the relative importance between different transport mechanisms during the FO process as a function of the polyelectrolyte bilayer number: (a) the Peclet number Pea indicating the relative importance of the water transport and the solute transport; and (b) the dimensionless solute permeability Ks indicating the relative solute transport resistance between the active layer and the support layer.

The concentration profile formed in the vicinity of the interior surface of the active layer could seriously reduce the concentration difference across the active layer, thereby decreasing the

600 500

Pea

400

T-PAH T-PSS MgCl2 MgCl 2 MgSO4 MgSO 4 Na2SO4 Na 2 SO 4 Pea = AΔπ/B

300 200 100 0 80 70

Ks= Ds/SB

60 Ks

50 40 30 20 10 0 0.5

1.0

1.5 2.0 2.5 Bilayer number

3.0

3.5

S. Qi et al. / Journal of Membrane Science 423-424 (2012) 536–542

70

AL-DS

Jv (L/m2 h)

60 50 40 T-PAH

30

T-PSS

MgCl2 MgSO4 Na2SO4

20 10 0 1.00 ηFO= JV/AΔπ

ηFO

0.75

0.50

0.25

0.00

0.5

1.0

1.5 2.0 2.5 Bilayer number

3.0

3.5

Fig. 4. FO performance of the xLbL active layers with varied bilayer number in the filtrations with the orientation of AL-DS: (a) the filtration bulk flux and (b) the FO efficiency (the fraction of the effective driving force).

70 60 50 Jv (L/m2 h)

B as shown in Fig. 2, the rate of change of B was greater than that of A, thereby giving rise to the increasing values of Pea. Another important factor affecting the ICP process is the relative solute resistance between the active layer and the support layer, which could be quantitatively described by the dimensionless solute permeability Ks ¼Ds/SB [29]. The quantity Ds is the solute diffusion coefficient in aqueous solutions, and S is the structural factor referring to the equivalent membrane thickness for the support layer [10]. In view of this criterion, the effect of the support layer on the ICP could be significant when the value of Ks is around unity, and this is especially true for the case in which the solute transport is dominant (Pea 51) [29]. Similarly, the values of Ks were calculated for the FO membranes with a variety of xLbL active layers as demonstrated in Fig. 3(b) (the values of Ds were 1.61  10  10 m2/s for MgSO4, 5.57  10  10 m2/s for MgCl2 and 6.19  10  10 m2/s Na2SO4 from the OLI Stream Analyzer software, and S¼109 mm was measured by the method in our previous studies [13]). As the same substrate was used for all the FO membranes, the oscillatory increase of the value of Ks straight forwardly reflected the response of B to the alternate adsorption of the polyelectrolyte layers. It is remarkable that the initial deposition of positively charged PAH (xLbL0.5) gave rise to the Ks less than 10 for all electrolyte solutes. This indicates that the diffusive resistance of the single PAH active layer was comparable to that of the PAN support layer rendering a regime in which the effect of the ICP could be significant. The values of Ks were significantly increased as increasing the bilayer number. All the prepared FO membranes with xLbL active layers were employed in the FO filtration experiments with two membrane orientations (AL-DS and AL-FS). Particularly, in addition to the bulk flux, our attention was given to the FO efficiency ZFO, which refers to the fraction of the effective driving force during FO processes. Numerically, ZFO is evaluated as the ratio of the measured bulk flux JV to the maximum bulk flux JVmax given by osmotic pressure difference between the draw and feed solution (ADp) [29]. First, the filtrations were performed with the active layers exposed to the draw solution, and both the measured bulk flux and the calculated FO efficiency were respectively presented in Fig. 4 as a function of the bilayer number. It is interesting to note that the bulk flux was progressively enhanced, and asymptotically attained a value of  60 L/m2h as the polyelectrolyte film became thicker. The growth of the bulk flow was primarily attributed to the influence of the ICP, which was identified by the variations of the FO efficiency in Fig. 4(b). It is evident that a very low FO efficiency of about 0.25 was yielded for the FO membranes with xLbL0.5 active layer indicating the severe loss of driving force caused by the ICP. This observation is consistent with the analysis of Pea and Ks. Although relatively small values of Pea were yielded in this initial regime, the ICP was aggravated due to the comparable solute permeabilities of the active layer and the support layer (i.e., the value of Ks is of order one) as illustrated in Fig. 3. As the number of the polyelectrolyte layer was substantially increased, the transfer resistance of the solutes in the support layer was getting less important, and the net growth of the driving force tended to increase the filtration flux. On the other hand, the increase of the bulk flux turned the water transport into the dominant mechanism in the support layer (i.e., the value of Pea was increased), which had negative impact on the ICP. As a result, the enhancement of the FO efficiency was retarded, thereby giving rise to a plateau regime as shown in Fig. 4(b). The experimental results for the AL-FS orientation were plotted in the same way as shown in Fig. 5. As revealed in the literature [10,13,29], when the porous support layer is exposed to the high-concentration draw solution, more severe polarization phenomena might occur in the support structure as a greater

T-PAH T-PSS MgCl2 MgSO4 Na2SO4

AL-FS

40 30 20 10 0 1.00 ηFO= JV/A∆π 0.75

ηFO

540

0.50

0.25

0.00

0.5

1.0

1.5 2.0 2.5 Bilayer number

3.0

3.5

Fig. 5. FO performance of the xLbL active layers with varied bilayer number in the filtrations with the orientation of AL-FS: (a) the filtration bulk flux and (b) the FO efficiency (the fraction of the effective driving force).

S. Qi et al. / Journal of Membrane Science 423-424 (2012) 536–542

xLbL2.0

541

xLbL2.5

Cross section of xLbL2.0 Fig. A1. Typical field emission scanning electron microscope (FE-SEM, JSM-7600 F) images of xLbL membranes: (a) top surface of xLbL2.0; (b) top surface of xLbL2.5; (c) cross section of xLbL2.0 and (d) magnified cross section of xLbL2.0.

concentration gradient is needed to balance the bulk transport with higher solute concentrations. This is in agreement with the observation in Fig. 5 that the FO filtrations with AL-FS engendered relatively low bulk fluxes (Fig. 5a). For membranes coated with 0.5  2 LbL bilayers, higher water flux was observed for T-PSS than T-PAH, which is explained by the more hydrophilic surface of T-PSS (Fig. 1a). Nevertheless, when further increasing the bilayer number (i.e., 2.5–3.5 bilayers), no significant difference was observed between T-PSS and T-PAH membranes, suggesting that the FO performance in this region was largely controlled by the bulk properties of the LbL layers (Fig. 2) while the surface properties was less important. Correspondingly, FO efficiencies (Fig. 5b) were also remained at a comparably low level of about 0.25 for membranes with 2.5–3.5 bilayers. Although the similar increase of the FO efficiency was observed, the bulk flux exhibited a minimum for the FO membrane with xLbL1.5 as indicated by Fig. 5(a). This observation could be rationalized by the fact that although the FO efficiency can be viewed as the normalized bulk flux, the bulk flux is not in a linear relationship with the FO efficiency when the value of A is varied. Hence, the enhancement of the bulk flux caused by the alleviation of the ICP might not be able to compensate the reduction of the bulk flux caused by the increased hydraulic resistance, i.e., smaller values of A. In light of this reasoning, it is also expected that the bulk flux might be decreased in the regime corresponding to a very thick polyelectrolyte film, even though a high FO efficiency could be held. While the current study mainly focused on the effect of LbL structure (especially the terminating layer) on FO performance, it is also important to note that other factors, including solution physicochemical properties (e.g., diffusivity, ion size, and viscosity), structure of substrates, and surface properties of substrates, can affect ICP thereby influencing FO performance [26,30,31]. Further membrane performance optimization can be achieved by combined rejection layer improvement, substrate optimization, and operational condition (e.g., draw solution) selection.

4. Summary In this work, a series of experiments were performed to investigate the dependence of the FO performance to the modification of the active layer formed by the xLbL method. The FO membranes were prepared by depositing various number of PAH/PSS bilyers onto a PAN substrate. In the surface characterization measurements, periodic variations of the surface hydrophilicity, surface roughness, and zeta potential were observed as varying the bilayer number and the terminating-layer charge. The evolution of the intrinsic transport properties of the polyelectrolyte active layers (the hydraulic permeability A and the solute permeability B) was then interpreted based on the surface characteristics revealing the complex interplay between the electrolytic solutions and the charged multilayer films. With the aid of two dimensionless quantities (the Peclet number Pea and the dimensionless solute permeability Ks), the effects of the xLbL modification on the ICP were discussed based on the relative importance between different transport mechanisms during the FO processes. This quantification analysis successfully justifies the correlations between the FO performance and the varied properties of the xLbL active layer: (i) the adverse impact of the ICP on the membrane throughput could be mitigated by properly controlling the number of the polyelectrolyte bilayers; (ii) there was a tradeoff between the FO efficiency and the hydraulic loss for the optimization of the active layer. All the presented results provide better knowledge of optimizing the xLbL technique for fabricating the FO membranes with enhanced performance.

Acknowledgments This project (Reference no. MEWR C651/06/173) is supported by the Environment and Water Industry Program Office of Singapore (under the funding of National Research Foundation).

542

S. Qi et al. / Journal of Membrane Science 423-424 (2012) 536–542

Appendix A SEM images of membrane surface and cross-sections are shown in Fig. A1. References [1] J.R. McCutcheon, R.L. McGinnis, M. Elimelech, A novel ammonia–carbon dioxide forward (direct) osmosis desalination process, Desalination 174 (1) (2005) 1–11. [2] J.R. McCutcheon, R.L. McGinnis, M. Elimelech, Desalination by ammonia– carbon dioxide forward osmosis: influence of draw and feed solution concentrations on process performance, J. Memb. Sci. 278 (1–2) (2006) 114–123. [3] S. Zhao, L. Zou, C.Y. Tang, D. Mulcahy, Recent developments in forward osmosis: opportunities and challenges, J. Memb. Sci. 396 (0) (2012) 1–21. [4] E.R. Cornelissen, D. Harmsen, K.F. de Korte, C.J. Ruiken, J.J. Qin, H. Oo, L.P. Wessels, Membrane fouling and process performance of forward osmosis membranes on activated sludge, J. Memb. Sci. 319 (1–2) (2008) 158–168. [5] M.I. Dova, K.B. Petrotos, H.N. Lazarides, On the direct osmotic concentration of liquid foods. Part I: Impact of process parameters on process performance, J. Food Eng. 78 (2) (2007) 422–430. [6] A. Achilli, T.Y. Cath, E.A. Marchand, A.E. Childress, The forward osmosis membrane bioreactor: a low fouling alternative to MBR processes, Desalination 239 (1–3) (2009) 10–21. [7] W.C.L. Lay, Q. Zhang, J. Zhang, D. McDougald, C. Tang, R. Wang, Y. Liu, A.G. Fane, Study of integration of forward osmosis and biological process: membrane performance under elevated salt environment, Desalination 283 (0) (2011) 123–130. [8] T.Y. Cath, A.E. Childress, M. Elimelech, Forward osmosis: principles, applications, and recent developments, J. Memb. Sci. 281 (1–2) (2006) 70–87. [9] G.T. Gray, J.R. McCutcheon, M. Elimelech, Internal concentration polarization in forward osmosis: role of membrane orientation, Desalination 197 (1–3) (2006) 1–8. [10] C.Y. Tang, Q. She, W.C.L. Lay, R. Wang, A.G. Fane, Coupled effects of internal concentration polarization and fouling on flux behavior of forward osmosis membranes during humic acid filtration, J. Memb. Sci. 354 (1–2) (2010) 123–133. [11] G. Decher, Fuzzy nanoassemblies: toward layered polymeric multicomposites, Science 277 (5330) (1997) 1232–1237. [12] P. Bertrand, A. Jonas, A. Laschewsky, R. Legras, Ultrathin polymer coatings by complexation of polyelectrolytes at interfaces: suitable materials, structure and properties, Macromol. Rapid Commun. 21 (7) (2000) 319–348. [13] Q. Saren, C.Q. Qiu, C.Y. Tang, Synthesis and characterization of novel forward osmosis membranes based on layer-by-layer assembly, Environ. Sci. Technol. 45 (12) (2011) 5201–5208. [14] C. Qiu, S. Qi, C.Y. Tang, Synthesis of high flux forward osmosis membranes by chemically crosslinked layer-by-layer polyelectrolytes, J. Memb. Sci. 381 (1-2) (2011) 74–80. [15] S. Qi, C. Qiu, Y. Zhao, C.Y. Tang, Double-skinned forward osmosis membranes based on layer-by-layer assembly – FO performance and fouling behavior, J. Memb. Sci. 405–406 (2012) 20–29.

[16] M. Kolasin´ska, R. Krastev, P. Warszyn´ski, Characteristics of polyelectrolyte multilayers: effect of PEI anchoring layer and posttreatment after deposition, J. Colloid Interface Sci. 305 (1) (2007) 46–56. [17] A.N.D. Egueh, B. Lakard, P. Fievet, S. Lakard, C. Buron, Charge properties of membranes modified by multilayer polyelectrolyte adsorption, J. Colloid Interface Sci. 344 (1) (2010) 221–227. [18] B.W. Stanton, J.J. Harris, M.D. Miller, M.L. Bruening, Ultrathin, multilayered polyelectrolyte films as nanofiltration membranes, Langmuir 19 (17) (2003) 7038–7042. [19] J.R. McCutcheon, M. Elimelech, Influence of membrane support layer hydrophobicity on water flux in osmotically driven membrane processes, J. Memb. Sci. 318 (1–2) (2008) 458–466. [20] C.Y.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 II. Membrane physiochemical properties and their dependence on polyamide and coating layers, Desalination 242 (1-3) (2009) 168–182. [21] M. Kolasin´ska, P. Warszyn´ski, The effect of nature of polyions and treatment after deposition on wetting characteristics of polyelectrolyte multilayers, Appl. Surf. Sci. 252 (3) (2005) 759–765. [22] T. Ogawa, B. Ding, Y. Sone, S. Shiratori, Super-hydrophobic surfaces of layerby-layer structured film-coated electrospun nanofibrous membranes, Nanotechnology 18 (2007) 16. [23] S. Loeb, L. Titelman, E. Korngold, J. Freiman, Effect of porous support fabric on osmosis through a Loeb–Sourirajan type asymmetric membrane, J. Memb. Sci. 129 (2) (1997) 243–249. [24] J.M.M. Peeters, J.P. Boom, M.H.V. Mulder, H. Strathmann, Retention measurements of nanofiltration membranes with electrolyte solutions, J. Memb. Sci. 145 (2) (1998) 199–209. [25] Y. Marcus, A. Rashin, A simple empirical model describing the thermodynamics of hydration of ions of widely varying charges, sizes, and shapes, Biophys. Chem. 51 (2–3) (1994) 111–127. [26] J. Wei, C. Qiu, C.Y. Tang, R. Wang, A.G. Fane, Synthesis and characterization of flat-sheet thin film composite forward osmosis membranes, J. Memb. Sci. 372 (1–2) (2011) 292–302. [27] C. Boo, S. Lee, M. Elimelech, Z. Meng, S. Hong, Colloidal fouling in forward osmosis: role of reverse salt diffusion, J. Memb. Sci. 390–391 (2012) 277–284. [28] Lee, S., Boo, C., Elimelech, M., Hong, S.. Comparison of fouling behavior in forward osmosis (FO) and reverse osmosis (RO). J. Memb. Sci. 365 (1–2) (2010) 34–39. [29] W. Li, Y. Gao, C.Y. Tang, Network modeling for studying the effect of support structure on internal concentration polarization during forward osmosis: model development and theoretical analysis with FEM, J. Memb. Sci. 379 (1–2) (2011) 307–321. [30] S. Zhao, L. Zou, Relating solution physicochemical properties to internal concentration polarization in forward osmosis, J. Memb. Sci. 379 (1–2) (2011) 459–467. [31] N.Y. Yip, A. Tiraferri, W.A. Phillip, J.D. Schiffman, L.A. Hoover, Y.C. Kim, M. Elimelech, Thin-film composite pressure retarded osmosis membranes for sustainable power generation from salinity gradients, Environ. Sci. Technol. 45 (10) (2011) 4360–4369.