In this paper, a novel positively charged membrane was prepared through interfacial polymerization technique between polyethyleneimine in aqueous phase and trimesoyl chloride in organic phase. Next, cross-linking of polyamide (PA) layer using ρ-xylylene dichloride (XDC) and glutaraldehyde (GA) was studied. The influences of cross-linking concentrations on the separation and permeation performance of membrane were also investigated. Membranes were characterized in terms of their chemical structure, the cross-sectional and surface morphologies, contact angles, molecular weight cut-off (MWCO) and effect of pH feed solution. The salt rejection sequence of CaCl2 >NaCl > Na2SO4 showed a positive charge at the membrane surface after cross-linking reaction. The MWCO of primary PA membrane decreased from 1,135 to 775 and 885 Da for XDC and GA, respectively. XDC membrane shows highest CaCl2 divalent cationic rejection (95.5%) and lowest water flux (21.1 L/m2.h). This study illustrates a promising method for fabrication of positively charged membrane in cation separation.

INTRODUCTION

Nanofiltration (NF) is a novel and powerful separation technology that displays separation characteristics in the intermediate range between ultrafiltration (UF) and reverse osmosis (RO) (Zhanga et al. 2011). Pore size of NF membranes changes between UF and RO membranes and the nominal molecular weight cut-off (MWCO) of a NF membrane ranges from 200 to 1,000 Da (Cheng et al. 2012). Transfer mechanism of NF membranes involves both pore sieving as in UF and diffusion as in RO. Separation mechanisms involve steric hindrance and Donnan exclusion effects (electrostatic interaction of charges between solutes and membrane surface). These characteristics lead to a relatively high permeate flux at low operating pressure and high selectivity. These advantages have led to a growing list of NF membrane applications, such as water softening (Mulder 1996), dyes substances separation and purification (Koyuncu 2002), wastewater treatment (Van der Bruggen et al. 2003), natural organic matter separation (Schafer et al. 1998), recovery of valuable pharmaceutical products (Tessier et al. 2005), and removal of heavy metal ions (e.g. Ni2+, Cd2+, Cu2+, Zn2+, Co2+, Pb2+) from industrial wastewater before discharging (Bougen et al. 2001; Mohamm et al. 2004; Bouranene et al. 2008).

Heavy metals are harmful to public health, plants and animals because of their higher toxicity, non-biodegradable and persistent nature (Gonzalez-Munoz et al. 2006). Separation of heavy metals from wastewater has received considerable attention during recent years caused by the stringent environmental laws. Heavy metals are found naturally in the earth but become concentrated with human industrial activities, such as electroplating, mining, battery manufacturing, metal finishing, paints and treated woods, textile industries, and automotive industries (Li et al. 2008). Several methods have been developed for the treatment of heavy metals in wastewater, such as chemical precipitation, coagulation/coprecipitation, adsorption, ion exchange, and floatation and end-up standard discharges (Ku et al. 2001; Kurniawan et al. 2006; Fu & Wang 2011). However, low selectivity, high costs, long time of operation, and the large sludge containing toxic compounds produced during these processes are the main disadvantage of these methods. Different types of membrane separation operations show the greatest promise in the commercial treatment of heavy metal wastewater. Among membrane processes RO operation was limited because of high energy consumption and low water permeability. Microfiltration and UF processes have a low rejection for heavy metal treatment. But NF technology with acceptable rejection, low energy consumption (moderate pressure), and high water permeability is a promising technology for removal of heavy metal from wastewater and other applications (Muthukrishnan & Guha 2006). However, according to the NF separation mechanism (Donnan exclusion effects) positively charged NF membranes are more efficient for the removal or recovery of heavy metal and multivalent cations. Several methods, such as coating (Ba et al. 2010; Wang et al. 2013), chemical cross-linking (Huang et al. 2006; Sun et al. 2011), UV photo-grafting (Deng et al. 2011; Liu et al. 2015), and plasma grafting (Buonomenna et al. 2009) were developed for prepared positively charge NF membranes.

So, positively charged NF membranes have gained significant attention in recent years. An et al. (2011) prepared positively charged copolymers of poly (2-methacryloyloxy ethyl trimethylammonium chloride-co-2-hydroxyethyl acrylate) (PDMCHEA) through free radical copolymerization. The water flux and salt rejection for MgCl2 and NaCl were 19.1 and 20.6 L/m2.h and 94.3 and 60.7% at 25 °C under 0.6 MPa, respectively. Li et al. (2011) prepared hydrophilic and positively charged NF onto polysulfone UF substrate by UV-induced graft polymerization of N,N-dimethylaminoethyl methacrylate and followed by quaternization by chemical cross-linking. The rejections of the composite membranes to different salts followed the order Na2SO4 < NaCl< MgSO4< MgCl2< CaCl2. Zhang et al. (2014) prepared NF membranes using the phase inversion method using cardopoly(arylene ether sulfone) with pendant tertiary amine groups (PES-TA). Zeta potential measurements indicated that the PES-TA membranes were positively charged below pH 10.7. Zhang et al. (2014) synthesized novel positively charged composite nanofiltration membranes were facilely prepared by polydopamine (PDA) deposition followed by poly(ethyleneimine) (PEI) grafting on polyethersulfone (PES) substrates. The rejection of salts was increased but the pure water flux was decreased with the increase of PDA deposition time, PEI concentration, PEI reaction temperature and time. Yu et al. (2015) prepared high flux positively charged nanofiltration membrane via ‘blending-phase inversion’ method using silica spheres modified by poly (ionic liquid) brushes. The results revealed that the hybrid membranes have enhanced surface hydrophilicity, water permeability, thermal stability, and mechanical properties.

Most of the commercial NF membranes have a negatively charge and are often made from polyamide using interfacial polymerization into the microporous substrate (An et al. 2011). In the interfacial polymerization technique, two reactive monomers usually amine and acyl chloride are dissolved in two insoluble phases, such as water and n-hexane, and the polymerization reaction occurs onto the interface of phases on the surface of the microporous support (Akbari & Mojallali Rostami 2014). As a result, a dense ultra-thin active layer formed separately onto a microporous support and it has a major role in determination of membrane characters, such as water flux, selectivity, morphology, roughness, hydrophilicity and type of membrane charge.

The main aim of the current study is preparation of a novel positively charged membrane based on polyamide membrane using interfacial polymerization method. Positively charged PA NF membrane was prepared by interfacial polymerization of PEI and triethylamine (TEA) in aqueous phase and trimesoyl chloride (TMC) in organic phase. The following PA active top layer was cross-linked using -xylylene dichloride (XDC) and glutaraldehyde (GA) agents. Performance of PA NF membranes was determined by filtration of aqueous feed solutions containing 1,000 ppm of NaCl, Na2SO4, and CaCl2 salts. The morphological, topological and chemical characteristics of the resulting membranes were studied by scanning electron microscopy (SEM), atomic force microscopy (AFM), and FT-IR, respectively. MWCO of PEI/TMC, XDC and GA membranes were calculated using filtration different molecular weight of polyethylene glycol (PEG). The relationship between salt rejection, water flux, and cross-linking agent concentrations was discussed.

EXPERIMENTAL

Materials

Polyacrylonitrile (PAN) fibers were purchased from textile wastes of SepehrManufacturing Company and were used as polymer for support membranes preparation. PEI (Mw ∼ 2,000) and GA purchased from Sigma-Aldrich Company. N,N-dimethylformamide (DMF), TMC, TEA, n-hexane, sodium hydroxide (NaOH) and hydrochloride acid (HCl) were obtained from Merck Chemicals (Germany). Also, PEG with different molecular weight were purchased from Merck Company. p-XDC purchased from Across. All chemical agents were used without further purification.

Preparation of microporous polyacrylonitrile support membrane

Microporous PAN membranes were used as a supporting substrate for the PA membrane, which were prepared by traditional phase-inversion method (Schafer et al. 2005). Homogeneous casting solution was prepared by dissolving of PAN (16% wt) in DMF (84% wt) at 70 °C for 12 hours. Polymer solutions were cast on flat glass plates with a uniform thickness of 250 μm by a casting knife. Immediately after casting, the glass plates were immersed in a coagulation bath containing distilled water at 25 °C acting as the non-solvent. Microporous PAN substrate membrane stored in deionized water at ambient temperature for after modification and experiments.

Fabrication of PA membrane

The selective PA active layer of NF membrane was prepared by conventional interfacial polymerization method onto the microporous PAN membrane. First, PAN supporting membrane was dipped into the PEI/TEA aqueous solution for 60 min and then the excess solution was removed using pieces of filter paper until the solution remained on the membrane surface. Then, the support membrane with aqueous phase was installed in a frame and the top surface of membrane was exposed to the TMC organic solution for 60 seconds. The effect of PEI (0.5, 1, 1.5, 2 and 2.5% w/w) and TMC (0.1, 0.2, 0.3 and 0.4% w/v) concentrations was investigated. Finally the membrane was thoroughly rinsed with de-ionized water and stored in deionized water at 4 °C before carrying out evaluation studies.

Cross-linking PA membrane by XDC and GA

As follows, quaterized PA membrane was reached by chemical cross-linking with XDC and GA. Chemical cross-linking shows at least two advantages: (1) PA active layer will be more compact and highly selective; (2) PA membrane surface will be more positively charged after cross-linking reaction. The PA membranes were immersed into cross-linker XDC in ethanol (0.2, 0.4, 0.6, 0.8 and 1% w/w) or GA in water (0.2, 0.6, 1, 1.4 and 1.8% w/w) solutions. The cross-linking reaction between PEI and XDC/GA was carried out at room temperature for 5 h and 12 h respectively, and is schematically described in Figure 1. After cross-linking reaction, the modified membranes were washed three times with water/ethanol by vibration to remove unreacted XDC/GA.
Figure 1

Schematic diagram for preparing positively charged PA membrane using chemical cross-linking with XDC and GA.

Figure 1

Schematic diagram for preparing positively charged PA membrane using chemical cross-linking with XDC and GA.

Separation and permeation properties

The performance of all the membranes were characterized by measurement of water flux and salt rejection of NaCl, Na2SO4 and CaCl2 salts with total concentration of 1,000 ppm at ambient temperature. Performances of prepared membranes were analyzed using a batch cross-flow system with 21 cm2 effective membrane surface area (Figure 2). The water flux (J, L/m2.h) and salt rejection (R, %) of the membrane was determined by using Equation (1) and Equation (2), respectively. 
formula
1
 
formula
2
where V (liter) is the volume of permeated water, A (m2) is the effective membrane area and t (hour) is the filtration time. Also Cp and Cf (mg/L) are permeated and feed solution concentration, respectively. All membrane samples were tested for at least three measurements and results have been averaged.
Figure 2

Cross-flow filtration for determining water permeation and salt rejection.

Figure 2

Cross-flow filtration for determining water permeation and salt rejection.

Characterization of membranes

Fourier transform infrared spectroscopy (FT-IR) (Nicolet Magna-IR 550) was employed to investigate the surface chemical composition changes of PAN membrane before and after modification and cross-linking in the region of 400–4,000 cm−1. The membrane surface hydrophilicity was characterized by water contact angle measurement using a contact angle goniometer (DSA 100 KRUSS GMBH, Germany) equipped at room temperature. A total of 4 μL of ultrapure water was lowed onto a dry membrane surface with a microsyringe and after 3 seconds, the value of water contact angle was recorded. At least five measurements in different locations of the membrane surface sample were carried out and then the curves of contact angle as a function of the drop age were plotted to compare the hydrophilicity of different membranes.

Membranes surface and cross-section morphologies were obtained by SEM (SEM KYKY-EM 3,200, China). The membrane samples were fractured in liquid nitrogen to avoid destroying the cross-section structure and then sputtered with gold before observation. AFM (model Park Scientific Instrument, CP Auto Probe) was used to image the membranes. The images were obtained in the same way from different places of each membrane surface sample at room temperature in air. Dry membranes were cut and glued on glass substrate and then membrane surfaces were imaged in a scan size of 5 μm × 5 μm.

The MWCO value is referred to the molecular weight of PEG that was 90% rejected by membrane and pore sizes were calculated from the relationship between the MWCOs of PEG and their Stokes radii as follows (Causserand et al. 2004): 
formula
3
 
formula
4

The MWCO of the membranes was evaluated with Filtration of different molecular weight PEG (200, 400, 1,000, and 1,500 g/mol). PEG concentration was measured using a total organic carbon analyzer (TOC-VCPHSHIMADZU).

RESULTS

Optimization of PEI and TMC concentration

A key advantage of the PA-TFC membrane is that the materials for microporous support and PA active top layer can be chosen and optimized separately to reach the best separation performance and membrane stability (Petersen 1993). The microporous support membrane was prepared for suitable strength against operating conditions and PA active layer being responsible for membrane water flux and selectivity. The selective PA active layer of TFC membrane was prepared by conventional interfacial polymerization method on the microporous PAN membrane. Concentrations of PEI in aqueous phase and TMC in organic phase were separately investigated to gain the best performance of PA-TFC membrane.

First, in interfacial polymerization process different amounts of PEI concentration was applied (0.5, 1, 1.5, 2 and 2.5% w/w) at 0.3% w/v TMC monomer and water flux and CaCl2 rejection illustrated in Figure 3. The water flux decreased and CaCl2 salt rejection increased with increase in PEI concentration. With increasing PEI concentration from 0.5 to 1.5% w/w, water flux rapidly decreased from 68.3 to 31.9 L/m2.h while rejection enhanced from 5.6 to 81.8%. At the higher PEI concentration both water flux and salt rejection roughly become constant because formed PA layer was prevented from more amine diffusion in the organic phase (Freger et al. 2002).
Figure 3

Effect of TMC and PEI concentration on the water flux and NaCl , CaCl2, and Na2SO4 salt rejections.

Figure 3

Effect of TMC and PEI concentration on the water flux and NaCl , CaCl2, and Na2SO4 salt rejections.

The variation of PA-TFC membrane performance with the concentration of TMC in organic phase showed the same trend and is shown in Figure 3. At first, water flux declined rapidly as the concentration of TMC in organic phase increased between 0.1 (394.8 L/m2.h) and 0.2 (43.2 L/m2.h), because at the 0.1% concentration a fault-free PA active top layer on the PAN support surface was not formed. At the higher concentration water flux declined with a slight slope from 43.2 at 0.2% to 24.8 L/m2.h at 0.4%. More TMC concentration namely more TMC monomers at interface of the two phases, so the PA active top layer became thicker, leading to more resistant to cross-flow water molecules transfer and resulting lower flux (Berezkin & Khokhlov 2006). Salt rejections show that the rejection increased as the concentration of TMC increased from 0.1 to 0.3% and then rejection approximately become constant. As a result, the modified PA-TFC membrane with 0.3% of TMC concentration seems to be the best result, because at the higher concentration of TMC water flux declined from 32 to 24.8 L/m2.h without considerably change in CaCl2 salt rejection.

The effect of XDC and GA concentration on the PA-TFC membrane performance

Figure 4 shows the effect of XDC and GA cross-linking concentration on the PA-TFC membrane water flux, when the PEI and TMC concentration are kept at 2% w/w and 0.3% w/v, respectively. In general, the change tendencies of the water flux of both cross-linking membranes are the same. The water flux of two membranes first decreases and then increases with increasing XDC and GA cross-linking concentration but XDC membrane shows intensive variation. This is probably assigned to higher XDC activity compared with GA agent and its ability to quaternization reaction (Li et al. 2011). The cross-linking reaction between PEI in PA polymer backbones and XDC or GA agent is enhanced with increasing the XDC or GA concentration (Musale & Kumar 2000; Dong et al. 2007). This leads to a more compact PA selective layer and consequently water flux indicated reduction at 0.6% XDC and 1% GA concentrations. This is explained by noting that the degree of chemical cross-linking reaction between PEI and XDC or GA is improved with increasing cross-linking agent concentration. At higher cross-linking concentration, the XDC or GA is exceeded and changed membrane size when its concentration is higher (Akbari & Mojallali Rostami 2014). Also, these results are probably attributed to the fact that more XDC or GA only use up to amines quaternization and not to cross-linking.
Figure 4

Effect of XDC and GA concentration on the water flux and NaCl , CaCl2, and Na2SO4 salt rejections.

Figure 4

Effect of XDC and GA concentration on the water flux and NaCl , CaCl2, and Na2SO4 salt rejections.

In general, the change tendencies of the salt rejection and the water flux of membrane are opposite. Also, the rejection behaviors of NaCl, Na2SO4 and CaCl2 of the cross-linked membranes are depicted in Figure 4. The salt rejection order follows: CaCl2 > NaCl > Na2SO4, which was in agreement with the Schaep &Vandecasteele (2001) and Huang et al. (2008) sequences. It was found that CaCl2 and NaCl rejection of the cross-linked membrane increased with raising the XDC (from 0.2 to 0.6% w/w) or GA (from 0.2 to 1% w/w) concentration as water flux declined. With increasing cross-linking concentration the degree of chemical cross-linking reaction between PEI and XDC or GA agents is improved, leading to a denser barrier PA active top layer with many positive charges on the membrane surface. As a result divalent cationic CaCl2 salt rejection increased from 85.2 to 95.5% for XDC and 83 to 88.7% for GA cross-linked membranes. Also, divalent anionic Na2SO4 salt rejection decreased due to aggregation of positive charges on the membrane surface. In both membranes, the salt rejection for NaCl first increased and then decreased because it contained equal monovalent anions and cations. However, the CaCl2 rejection decreased and the pure water flux increased when the XDC or GA concentration was higher than 0.6 and 1% w/w, respectively. Also, comparison of two diagrams indicated that more variation of water flux and salt rejection occurred when XDC applied as a cross-linking agent. This phenomenon probably contributed to higher XDC reactivity rather than GA cross-linking agent and quarternized amine groups with independent positively charge.

Effect of pH feed solution on the membrane performance

The effect of pH on rejection salt ions was investigated by addition of NaOH (pH = 11) or HCl (pH = 3) to salt feed solution and varying its pH accordingly. In general, NF separation mechanisms frequently involve both distribution of membrane pore size and Donnan exclusion effects (Yu et al. 2010). Ions rejection by NF membranes is dependent on the electrical interaction between ions and membrane surfaces charge. So, pH of feed solution can be dependent on ion rejection because membrane surface charges varied with the pH. The functional groups on the PA active top layer (carboxylic acid, amine, and amide) and pH feed solution are responsible for charge density value of the membrane surface (Teixeira et al. 2005). Therefore, charged properties of cross-linking membranes were evaluated; pH adjusted at 3, 7, and 11 using HCl and soda (NaOH).

The effect of pH feed solution on the PA-FTC membrane cross-linked with XDC and GA are shown in Figure 5. Both membranes showed a same variation in neutral and acidic pH ranges while in basic condition, a different behavior can be seen for salt rejection. In basic pH, carboxylate anions formed on the PA layer surface which enhanced negatively charge density on the membrane surface. Rejection of Na2SO4 divalent anionic salt reached a maximum value of 37.4%, and CaCl2 and NaCl rejection decreased to 78.9 and 41.5%, respectively, for XDC membrane. The rejection sequence of the salts did not change with pH variation (CaCl2 > NaCl > Na2SO4) because quaternized amines existed in the PA layer. However, in GA membrane, surface charge and rejection sequence of the salts in basic pH varied extremely. Rejection of Na2SO4 divalent anionic salt was highest (67.7%) while CaCl2 divalent cationic and NaCl monovalent salts declined to 55.2% and 50.1%, respectively. In acidic condition, the amine groups attracted the proton converted to type IV amines, so charge density of the membrane surface became more positive. This change leads to higher rejection for CaCl2 (98.3% for XDC and 95.6% for GA) and remarkable decrease in Na2SO4 rejection (18.6% for XDC and 20.5% for GA). Also, the NaCl salt rejection increased considerably because it contained equal monovalent ion of both cation (Na+) and anion (Cl). However, higher rejection of divalent cationic salt (CaCl2) and lower rejection of divalent anionic salt (Na2SO4) in all pH ranges (except GA membrane in basic pH) proved that membrane surface had a positive charge even at neutral condition (Chiang et al. 2009).
Figure 5

pH effect of feed solution pH on the NaCl , CaCl2, and Na2SO4 salt rejections of XDC and GA membrane.

Figure 5

pH effect of feed solution pH on the NaCl , CaCl2, and Na2SO4 salt rejections of XDC and GA membrane.

Molecular weight cut-off

Figure 6 shows the retention curves of the PA and XDC and GA cross-linking membranes for PEG of different molecular weights (200, 400, 1,000 and 1,500 Da). The MWCO is the molecular weight of the smallest macromolecule which is rejected more than 90% by membrane and illustrated in Table 1. PEG is used to measure the MWCO value because of a wide range of molecular weight and ineffectiveness in membrane properties. The membranes pore sizes were calculated from the relationship between the MWCOs of PEG and their Stokes radii using Equations (3) and (4). Thus, the PA, XDC, and GA membranes had estimated MWCOs of 1,135, 775, and 885 Da, corresponding to the pore radii of 1.33 nm, 1.09 nm, and 1.17 nm, respectively. Results indicated that XDC and GA membranes have a smaller pore size compared to the PA membrane. These results proved that XDC and GA cross-linking agent reaction with PA layer amines and changed pore radii of membrane surface. Also, XDC membrane exhibits the lowest pore radii which can be attributed to higher activity of XDC agent. However, the results are compatible with the membrane water flux values and salt rejection that was presented earlier.
Table 1

MWCO, pore size, and contact angles of PAN substrate and PA, XDC, and GA membranes

  PAN substrate PA XDC GA 
MWCO (Da, ±10) – 1138 775 885 
Pore size (nm) – 1.33 1.09 1.17 
Contact angle (°, ±1 °) 71.4 62.4 43.5 55.7 
  PAN substrate PA XDC GA 
MWCO (Da, ±10) – 1138 775 885 
Pore size (nm) – 1.33 1.09 1.17 
Contact angle (°, ±1 °) 71.4 62.4 43.5 55.7 
Figure 6

PEG rejections of polyamide , XDC and GA membranes.

Figure 6

PEG rejections of polyamide , XDC and GA membranes.

Water contact angle of membranes

The contact angle of water is usually used to investigate the changes in the molecular structure and hydrophilicity of the membrane surfaces. The water contact angles of PAN support substrate, polyamide and cross-linked with XDC and GA membranes are shown in Table 1. As expected, all the polyamide membranes have a lower water contact angle than those of the PAN support membrane. Also, the contact angle decreases with the cross-linking of polyamide top layer. The water contact angle of polyamide, XDC and GA membranes were 62.4 °, 43.5 ° and 55.7 °, respectively. As a result, the enhancement of the hydrophilicity of the cross-linking membranes was due to the quaternization reaction of polyamide and formation independent cation on the membrane surface. XDC membrane has a smallest water contact angle because XDC cross-linked has a higher activity compared with GA agent.

FT-IR spectrum

The chemical structure of PAN support membrane and polyamide active layers was prepared by interfacial polymerization and was identified by FT-IR. As shown in Figure 7(a), peaks around 1,731 and 3,628 cm−1 were assigned to the stretching vibration of carbonyl (C=O) and hydroxyl (—OH) of carboxylic acid groups in the PAN chains, respectively (Zhang et al. 2009). Strong and sharp peak at 2,243 cm−1 corresponded to stretch vibration cyanide groups of PAN polymer chains. Peaks at 2,936 and 1,450 cm−1 were due to stretching vibration of C—H and bending vibration of methylene (CH2), respectively.
Figure 7

FT-IR spectrums of the PAN substrate (a), polyamide (b), XDC (c) and GA (d) membranes.

Figure 7

FT-IR spectrums of the PAN substrate (a), polyamide (b), XDC (c) and GA (d) membranes.

Figure 7(b) shows spectrum of polyamide TFC NF membrane synthesis using PEI monomer. The traces of PAN support membrane and polyamide active layer can be found in the spectra of TFC membrane. Two peaks at 1,627 and 1,729 cm−1 are related to stretch vibration of amide and acid carbonyl groups in polyamide chain structures, respectively (Zoua et al. 2010). Two peaks at 3,429 cm−1 and 1,370 correspond to stretch vibration N—H and C—N groups of PEI polymer chains. The FT-IR spectra confirm that the interfacial polymerization of polyamide active layer was successfully done on the surface of PAN support membrane.

Figures 7(c) and 7(d) show FT-IR spectra of polyamide membrane after cross-linking reaction using XDC and GA, respectively. The strong peak at 677 cm−1 is attributed to stretching vibration of C—Cl bond of XDC (Li et al. 2011). Also, the bending vibration of CH2–Cl of XDC compound is shown at 1,261 cm−1 that proves cross-linking reaction of PEI chains at polyamide active layer. In Figure 7(d) GA peaks such as vibration aldehyde carbonyl and vibration aldehyde C—H overlapping with polyamide peaks are seen. In this case we have reverted to other tests such as SEM, AFM, MWCO and contact angle analysis and performance tests.

Scanning electron microscopy

To examine the morphological changes of PAN support membrane after surface modification, the surface and cross-section SEM observations were conducted with a different magnification and images are shown in Figure 8 and Figure 9, respectively. The surfaces image of PAN support membrane (Figures 8(a) and 8(b)) was uniform and no pore structure could be found on the active layer. Figure 9 shows that the PAN support membrane has an asymmetric structure, with a thin and dense top layer and a much thicker highly porous substrate. However, the surface morphology dramatically changed when interfacial polymerization reaction occurred on the PAN support membrane (Figures 8(c) and 8(d)).
Figure 8

Surface SEM images of the PAN substrate (a, b), PA (c, d), XDC (e, f), and GA (g, h) membranes.

Figure 8

Surface SEM images of the PAN substrate (a, b), PA (c, d), XDC (e, f), and GA (g, h) membranes.

Figure 9

Cross-section SEM images of the PAN substrate (a, b), PA (c, d), XDC (e, f), and GA (g, h) membranes.

Figure 9

Cross-section SEM images of the PAN substrate (a, b), PA (c, d), XDC (e, f), and GA (g, h) membranes.

The PAN support membrane became rougher and had a higher roughness and membrane surface shows many small fine grain-like and belt-like structures. Surface morphology variation demonstrates that thin and dense polyamide active layer was successfully formed on the PAN support membrane. After cross-linking reaction, the grain-like structure gradually decreased and the belt-like structures increased and became larger. This phenomenon is probably attributed to the quaternization cross-linking reaction between XDC and GA and active amine groups in PA backbone. As seen in Figures 8(d), 8(f) and 8(h), XDC membrane has fewer grains compared with PA and GA membrane, respectively. This may be due to higher reactivity of XDC agent.

Atomic force microscopy

The AFM images of PAN substrate, PA, XDC and GA membrane is illustrated in Figure 10. Also, the AFM parameters of roughness (Sa), root mean square roughness (Sq), and average ten point height (Sz) are presented in Table 2. It can be seen that PAN substrate show a unique surface structure with 4.3 nm roughness. After polymerization onto the microporous support surface homogeneous and fine structures can be seen. PA membrane roughness parameters increased to 19.7, 26.5 and 114.9 nm for Sa, Sq and Sz, respectively, which confirms the formation of rougher surface after polymerization reaction. Cross-linking reaction increased PA membrane roughness. Generally, roughness sequence followed the order of XDC > GA > PA > PAN substrate. XDC membrane roughness is maximum (66.9 nm) because it has higher chemical activity and quaternization ability. Increasing the roughness leads to an increase in membrane fouling, which is in competition with increasing hydrophilic membrane surface (Johnson et al. 2015). Also, it can be seen that XDC and GA membrane morphology completely varied compared with primary PA membrane and belt-like structure with a little fine grain covered membrane surface uniformly. The results from AFM morphology and parameters are consistent with the morphologies observed by SEM images.
Table 2

Surface roughness parameters of the PAN substrate, PA, XDC, and GA membranes

  Roughness (±3) (SaRoot mean square (SqPeak maximum (Sz
PAN substrate 4.3 5.8 31.6 
PA 19.7 26.5 114.9 
XDC 66.9 93 382.7 
GA 64.8 84.4 379.9 
  Roughness (±3) (SaRoot mean square (SqPeak maximum (Sz
PAN substrate 4.3 5.8 31.6 
PA 19.7 26.5 114.9 
XDC 66.9 93 382.7 
GA 64.8 84.4 379.9 
Figure 10

AFM images of the PAN substrate (a, b), PA (c, d), XDC (e, f), and GA (g, h) membranes.

Figure 10

AFM images of the PAN substrate (a, b), PA (c, d), XDC (e, f), and GA (g, h) membranes.

CONCLUSION

In the present study novel positively charged PA-TFC membranes were successfully prepared by cross-linking PA layer with XDC and GA agents. It was observed that the quaternization of PA layer affected nanofiltration properties and led to higher rejection of divalent cationic salt (CaCl2). This demonstrates that the new positively nanofiltration membrane is appropriate for separation of cationic material solutions, which is essential for some industries. For both types of cross-linking agents, it was observed that the rejection of cationic salt increased with an increase in agent concentration. The XDC membrane indicated positive charge in the pH range from 3 to 11 due to the quaternization of the tertiary amine groups in the PA layer.

REFERENCES

REFERENCES
An
Q.
Ji
Y.
Zhao
Q.
Chen
H.
Gao
C.
2011
Preparation of novel positively charged copolymer membranes for nanofiltration
.
Journal of Membrane Science
376
,
254
265
.
Berezkin
A. V.
Khokhlov
A. R.
2006
Mathematical modeling of interfacial polycondensation
.
Journal of Polymer Science Part B: Polymer Physics
44
,
2698
2724
.
Bougen
A.
Rabiller-Baudry
M.
Chaufer
B.
Michel
F.
2001
Retention of heavy metal ions with nanofiltration inorganic membranes by grafting chelating groups
.
Separation and Purification Technology
25
,
219
227
.
Bouranene
S.
Fievet
P.
Szymczyk
A.
Samar
M. E. H.
Vidonne
A.
2008
Influence of operating conditions on the rejection of cobalt and lead ions in aqueous solutions by a nanofiltration polyamide membrane
.
Journal of Membrane Science
325
,
150
157
.
Buonomenna
M. G.
Lopez
L. C.
Davoli
M.
Favia
P.
Agostino
R. D.
Drioli
E.
2009
Polymeric membranes modified via plasma for nanofiltration of aqueous solution containing organic compounds
.
Microporous and Mesoporous Materials
120
,
147
153
.
Causserand
C.
Rouaix
S.
Akbari
A.
Aimar
P.
2004
Improvement of a method for characterization of ultrafiltration by measurement of tracers retention
.
Journal of Membrane Science
238
,
177
190
.
Chiang
Y. C.
Hsub
Y. Z.
Ruaan
R. C.
Chuang
C. J.
Tung
K. L.
2009
Nanofiltration membranes synthesized from hyperbranched polyethyleneimine
.
Journal of Membrane Science
326
,
19
26
.
Fu
F. L.
Wang
Q.
2011
Removal of heavy metal ions from wastewaters: a Review
.
Journal of Environmental Management
92
,
407
418
.
Gonzalez-Munoz
M. J.
Rodriguez
M. A.
Luque
S.
Alvarez
J. R.
2006
Recovery of heavy metals from metal industry wastewaters by chemical precipitation and nanofiltration
.
Desalination
200
,
742
744
.
Huang
R. H.
Chen
G. H.
Yang
B. C.
Gao
C. J.
2008
Positively charged composite nanofiltration membrane from quaternized chitosan by toluene diisocyanate cross-linking
.
Separation and Purification Technology
61
,
424
429
.
Johnson
D.
Galiano
F.
Deowan
S. A.
Hoinkis
J.
Figoli
A.
Hilal
N.
2015
Adhesion forces between humic acid functionalized colloidal probes and polymer membranes to assess fouling potential
.
Journal of Membrane Science
484
,
35
46
.
Kurniawan
T. A.
Chan
G. Y. S.
Lo
W. H.
Babel
S.
2006
Physico-chemical treatment techniques for wastewater laden with heavy metals
.
Chemical Engineering Journal
118
,
83
98
.
Mulder
M.
1996
Basic Principles of Membrane Technology
.
Kluwer Academic
,
Dordrecht
.
Petersen
R. J.
1993
Composite reverse osmosis and nanofiltration membranes
.
Journal of Membrane Science
83
,
181
150
.
Schaep
J.
Vandecasteele
C.
2001
Evaluating the charge of nanofiltration membranes
.
Journal of Membrane Science
188
,
129
136
.
Schafer
A. I.
Fane
A. G.
Waite
T. D.
2005
Nanofiltration: Principles and Applications
,
Elsevier Ltd
,
New York
.
Teixeira
M. R.
Rosa
M. J.
Nystrom
M.
2005
The role of membrane charge on nanofiltration performance
.
Journal of Membrane Science
265
,
160
166
.
Van der Bruggen
B.
Vandecasteele
C.
Gestel
T. V.
Doyenb
W.
Leysen
R.
2003
A review of pressure-driven membrane processes in wastewater treatment and drinking water production
.
Environmental Progress
22
,
46
56
.