Abstract

In this study, a positively charged nanofiltration (NF) membrane was prepared by interfacial polymerization for separation of divalent cations, whereby a nanomaterial (modified graphitic carbon nitride (g-C3N4) with poly(dopamine), PDA-C3N4) was incorporated into the active layer of the NF membrane. PDA-C3N4 sheets were synthesized from g-C3N4 sheets prepared by thermal oxidation of melamine, and the preparation conditions of NF membrane were also optimized. The results show that the roughness of PDA-C3N4 embedded NF membrane decreases, and the hydrophilicity and the permeation increase. The membrane also shows high rejection for divalent cations (Mg2+, Ca2+, Ba2+, Cu2+ and Zn2+) but low rejection (36.8%) for monovalent cation (Li+), as well as good fouling resistance performance. The fabricated membrane has the potential for treatment of industrial wastewater.

INTRODUCTION

Nanofiltration (NF) membranes can be used to separate multivalent ions and small organic compounds with a molecular weight of 200–1,000 Da (Van der Bruggen et al. 2001; Van der Bruggen & Vandecasteele 2003). However, most commercial NF membranes are negatively charged and capable of separating multivalent anions such as sulfates due to the Donnan exclusion principle. The Donnan exclusion effect on ions in the solution is that the co-ions are repelled and the counterions are attracted. The Donnan exclusion effect increases with the increasing of co-ion charge and decreases with the increasing of counterion charge (Pontalier et al. 1997; Wen et al. 2006), which plays a determining role in the efficiency of the separation of charged molecules. Therefore, positively charged NF membranes are more effective than negatively charged membranes in separating heavy metals (e.g., Cu2+, Zn2+, Ba2+) from the environment. Thus, considerable effort has been made to prepare positively charged NF membranes.

A number of methods have been developed for this purpose, such as chemical cross-linking (Ji et al. 2015), layer-by-layer assembly (Chen et al. 2015), interfacial polymerization (IP) (Wang et al. 2013) and ultraviolet-induced photografting polymerization (Deng et al. 2011). Among these methods, IP offers a simple and controllable approach to prepare composite NF membranes. An active layer can be formed at the interface via polymerization between two monomers of amine and acyl chloride (Cadotte 1977, 1981; Cadotte et al. 1980). 1,4-Bis(3-aminopropyl)piperazine (DAPP) has been widely used to fabricate positively charged NF membranes, but the permeability of these NF membranes is low (Li et al. 2015b). Therefore, many attempts have been made to improve the permeability of NF membranes by introducing nanomaterials into the active layer, such as carbon nanotubes (Zhang et al. 2017), graphene oxide (Sun & Wu 2018), and SiO2 (Lv et al. 2017).

Graphitic carbon nitride (g-C3N4) is a new nanoporous material with a lamellar structure like graphite layers. g-C3N4 has outstanding mechanical, thermal and chemical stability owing to its tri-s-triazine structure unit (Cao et al. 2015), and active groups such as NH and/or NH2 in g-C3N4 nanomaterials can strongly interact with polymer chains. More importantly, g-C3N4 can be synthesized using commonly available materials. g-C3N4 has been widely used to prepare pervaporation membranes (Cao et al. 2015; Wang et al. 2018) and gas separation membranes (Li et al. 2015a; Tian et al. 2016) due to the periodic hole defects of the g-C3N4 lattice, as well as photocatalytic membranes due to its excellent photocatalytic performance (Zhao et al. 2016; Li et al. 2017). However, g-C3N4 has rarely been used to fabricate NF or reverse osmosis (RO) membranes. Shahabi et al. (2019) synthesized a modified RO membrane using g-C3N to enhance the desalination performance, and the results showed that this membrane had better permeation, rejection for NaCl and antifouling performance than the unmodified one. Chen et al. (2016) prepared a negatively charged NF membrane via IP using g-C3N4 nanosheets, and it showed higher water flux and salt rejection of Na2SO4 (above 84.0%) compared with pristine membrane.

The main objective of this study is to enhance the permeability of positively charged NF membrane prepared by IP using g-C3N4 sheets. In order to improve the stability of g-C3N4 sheets in the active layer, dopamine (DA), a low-molecular-weight catecholamine able to self-polymerize to form a stable poly(dopamine) (PDA) layer on inorganic and organic materials (Azari & Zou 2012; Sun et al. 2012; Li et al. 2014), was introduced onto the surface of g-C3N4 sheets. In this study, a positively charged NF membrane was prepared by IP using polyether sulfone (PES) ultrafiltration membrane as the substrate, and DAPP and trimesoyl chloride (TMC) as the aqueous and organic monomer, respectively. The properties of g-C3N4 modified with PDA were characterized by scanning electron microscopy (SEM), energy dispersive spectroscopy (EDS), and Fourier transform infrared spectroscopy (FTIR). Then, the modified g-C3N4 was incorporated into the active layer, and the morphology, permeate flux, rejection and fouling resistance performance of the NF membrane were investigated.

EXPERIMENTAL SECTION

Materials

PES ultrafiltration membrane was purchased from Microdyn-Nadir Co., Ltd (Xiamen, China). TMC (99%, analytical reagent grade (AR)), DAPP (98%, AR), n-hexane (98%, AR), DA (98%, AR), tris(hydroxymethyl) aminomethane (Tris-HCl, AR), bovine serum albumin (BSA, AR), magnesium chloride hexahydrate (MgCl2·6H2O, AR), calcium chloride (CaCl2, AR), barium chloride (BaCl2, AR), cupric chloride dihydrate (CuCl2·2H2O, AR), zinc chloride (ZnCl2, AR) and lithium chloride (LiCl, AR) were purchased from Aladdin Co., Ltd (Shanghai, China). Polyethylene glycols (PEGs, AR) with a molecular weight of 400, 600 and 800 Da, glucose (AR), sucrose (AR), raffinose (AR), ethanol (99.5%, AR), sodium hydroxide (NaOH, AR) and hydrogen chloride (HCl, AR) were purchased from Titan Polytron Technologies Inc. (Shanghai, China). All chemicals were used without further purification. Aqueous solutions were prepared with pure water obtained using a RO membrane system (UPT-1-5T, China).

Membrane preparation

Preparation of PDA-C3N4

g-C3N4 was prepared by thermal decomposition of melamine prepared by ourselves and the detailed information was described in our previous work (Zhang et al. 2019). The typical step was as follows: 10 g of melamine was placed in a furnace (SX2-4-10, China) and heated to 550 °C at a rate of 5 °C·min−1 for 4 h. Then, the sample was cooled to room temperature, and the resulting powder was placed in a ball mill (BM4Pro, China) for 8 h to prepare small g-C3N4 particles.

g-C3N4 was modified via adhesion with PDA in Tris-HCl buffer solution (pH = 8.5, 50 mM). Typically, g-C3N4 and DA (1:1, w:w) were added to an Erlenmeyer flask with 200 mL of Tris-HCl buffer solution, and stirred at room temperature for 47 h. After that, the mixture was filtrated, washed with deionized water (DI-water), and then dried at 40 °C under vacuum for 48 h to obtain modified g-C3N4, which was denoted as PDA-C3N4.

Fabrication of PES composite nanofiltration membrane

PES ultrafiltration membranes were immersed into ethanol for 2 h to remove bubbles and impurities in and on the membranes. After that, these membranes were taken out and thoroughly washed with DI-water. A polyamide layer was formed on the PES support via IP using the aqueous solution of DAPP and the organic solution of TMC. The PES support was immersed in DAPP solution, and excess DAPP solution was removed using a filter. Then, the PES support was soaked in the TMC solution for some time. Excess TMC solution was removed from the PES support, and the membranes were cured at 70 °C for 8 min and then stored in DI-water before use. For g-C3N4 and PDA-C3N4 composite membranes, 0.005% of g-C3N4 or PDA-C3N4 was dispersed in DAPP solution using a probe sonicator (Biosafer 650-92, China) for 20 min. The substrate PES membrane and NF membranes prepared with DAPP/TMC, DAPP/TMC/g-C3N4 and DAPP/TMC/PDA-C3N4 were denoted as M0, MP, MPC and MPPC, respectively. The schematic of the NF membrane is shown in Figure S1 (Supplementary Material).

Characterizations

All samples were dried in a vacuum oven at 80 ± 0.2 °C for 24 h before characterization.

The cross-sectional and surface morphologies of membranes and inorganic materials were characterized by SEM (S4800, Japan). The surface roughness of membranes was analyzed by atomic force microcopy (AFM, Ntegra Prima, Russia), and then the root mean square (Rms) roughness and the average (Ra) roughness were calculated.

The functional groups on the surface of inorganic materials and membranes were characterized by FTIR (Thermo Fisher 6700, USA) with a wavenumber from 600 to 4,000 cm−1, EDS (S4800, Japan) and X-ray photoelectron spectroscopy (XPS, ESCALAB 250Xi, USA). The hydrophilic property was measured at five different positions for each sample by a contact angle goniometer (JC2000D2, China) at room temperature with 1.5 μL of DI-water as the droplet.

The zeta potential was characterized by streaming potential measurement with an electrokinetic analyzer (SurPASS, Anton Paar, Austria) using 1 mmol·L−1 potassium chloride (KCl) solution as the electrolyte solution at 20 ± 0.5 °C. Two pieces of membrane with an area of 10 × 20 mm were cut and pasted on the top of the cell, and then the streaming potential with pH was measured. The pH of the electrolyte solution was adjusted with 0.1 mol·L−1 NaOH or HCl.

The molecular weight cut-off (MWCO) of membranes was determined using several neutral solutes such as glucose, sucrose, raffinose and PEGs (0.3 g·L−1) under 0.6 MPa. The concentration of neutral solutes before and after filtration was measured by a total organic carbon analyzer (TOC-VCPN, Japan).

Separation performance of membranes

A laboratory scale cross-flow filtration module with an effective area of 35.3 cm2 was used to test the performance of membranes (Figure S2, Supplementary Material), including the volume permeate flux and the rejection ratio of saline. The feed tank jacket was circulated with cooling water to stabilize the temperature at 20 ± 0.5 °C. Before testing, each membrane was compacted with DI-water at 0.5 MPa and 20 ± 0.5 °C for 30 min to reach a stable state. Then, the saline solution (2.0 g·L−1) composed of MgCl2, CaCl2, BaCl2, CuCl2, ZnCl2 and LiCl was continuously fed for 30 min at 0.6 MPa and 20 ± 0.5 °C, and the steady flux and concentration of cations were measured. The volume permeate flux (Jv, L·m−2·h−1) was calculated by formula (1). 
formula
(1)
where V, A, and Δt are the volume of permeate solution (L), the active surface area of membrane (m2), and the permeation time (h), respectively.
The rejection ratio (R, %) of ions was calculated by formula (2). 
formula
(2)
where Cp and Cf are the concentration of permeate (g·L−1) and feed (g·L−1), respectively.

Fouling resistance performance of membranes

The fouling resistance property was investigated using 0.1 g·L−1 BSA solution in the cross-flow filtration module at 20 ± 0.5 °C. The membrane was pressurized with DI-water at 0.6 MPa and 20 ± 0.5 °C for 30 min to reach a stable state, and the volume permeate flux (Jv0) was measured. Then, the BSA solution was continuously fed for another 3.5 h, and the steady flux was measured at a certain time (1 h), which was denoted as Jv1. After that, the fouled membrane was thoroughly washed with DI-water for 10 min at 0.1 MPa and 20 ± 0.5 °C. Finally, the pure water volume permeate flux (Jv2) of the cleaned membrane was tested again at 0.6 MPa and 20 ± 0.5 °C. The above fouling resistance tests were stopped after two cycles.

The relative flux (RF), flux recovery ratio (FRR), total fouling ratio (FRt), reversible fouling ratio (FRr) and irreversible fouling ratio (FRir) were calculated by the following formulas: 
formula
(3)
 
formula
(4)
 
formula
(5)
 
formula
(6)
 
formula
(7)

RESULTS AND DISCUSSION

Characterization of modified g-C3N4

The surface morphologies and relative elemental contents of g-C3N4 and PDA-C3N4 were measured by SEM and EDS. As shown in Figure 1, g-C3N4 shows a characteristic lamellar structure with irregular wrinkles (Kang et al. 2018; Shahabi et al. 2019). After modification, the size of PDA-C3N4 becomes larger and more uniform and the surface is smoother, compared with g-C3N4. EDS was performed to determine the composition of g-C3N4 and PDA-C3N4 sheets. As shown in Table 1, g-C3N4 is composed of nitrogen (N), carbon (C) and a trace amount of oxygen (O) with a C:N ratio of about 0.76, which is very close to the theoretical value of 0.75. Compared with g-C3N4 sheets, the relative content of C and O elements is increased, while that of N element is decreased in the PDA-C3N4 sheet due to the high percentage of O and C elements in PDA, suggesting that PDA has been anchored onto the g-C3N4 sheets.

Table 1

Surface relative elemental content of g-C3N4 and PDA-C3N4 from EDS

SamplesRelative elemental content (at%)
CNO
g-C3N4 41.73 ± 1.61 54.79 ± 0.68 3.48 ± 0.64 
PDA-C3N4 48.60 ± 1.97 38.38 ± 1.77 13.03 ± 0.68 
SamplesRelative elemental content (at%)
CNO
g-C3N4 41.73 ± 1.61 54.79 ± 0.68 3.48 ± 0.64 
PDA-C3N4 48.60 ± 1.97 38.38 ± 1.77 13.03 ± 0.68 
Figure 1

SEM images of g-C3N4 (a), PDA-C3N4 (b) and FTIR spectra of g-C3N4 and PDA-C3N4 (c).

Figure 1

SEM images of g-C3N4 (a), PDA-C3N4 (b) and FTIR spectra of g-C3N4 and PDA-C3N4 (c).

FTIR was performed to characterize the chemical composition of g-C3N4 and PDA-C3N4, as shown in Figure 1(c). The strong peak at around 812 cm−1 is derived from the vibration of tri-s-triazine ring (Lotsch et al. 2007; Tian et al. 2016). Some absorption bands in the range of 1,245–1,640 cm−1 are also detected, which are assigned to the stretching vibration of C-NH-C and N-(C)3 units (Tian et al. 2016). Compared with g-C3N4, the absorption band at 3,404 cm−1 in PDA-C3N4 is much stronger, which corresponds to the O-H and N-H groups introduced from PDA via adhesion.

Effects of fabrication conditions on the membrane performance

The active layer plays a key role in the separation performance of NF membranes. The effects of DAPP concentration, DAPP immersion time, TMC concentration and reaction time on the membrane rejection and flux were investigated, as shown in Figure 2. In Figure 2(a), the rejection of MgCl2 and LiCl is decreased from 96.9% to 92.3% and from 50.1% to 40.3% with the increase of DAPP concentration, respectively, but the permeate fluxes of the composite membranes with MgCl2 and LiCl solutions first increase and then decrease. Similar results have also been reported by Chen et al. (2002) and Fang et al. (2013). This is probably because the polymerization rate is lower at lower monomer concentrations, leading to the formation of a thin and loose polyamide layer, but it increases with the increase of monomer concentration and thus the polyamide layer becomes thicker and more compact. However, as the DAPP concentration further increases, the TMC supply is deficient, which results in an increase in unreacted amine groups, but a decrease in the cross-linking degree and low salt rejection (Fang et al. 2013; Li et al. 2015b). As shown in Figure 2(b), the rejection of MgCl2 changes from 86.2% to 95.7% and then to 90.0% with the increase of DAPP immersion time, but the DAPP immersion time has little effect on the rejection and flux of LiCl solution and the flux of MgCl2 solution.

Figure 2

Effect of DAPP concentration (a), DAPP immersion time (b), TMC concentration (c) and reaction time (d) on membrane rejection and flux of MgCl2 and LiCl.

Figure 2

Effect of DAPP concentration (a), DAPP immersion time (b), TMC concentration (c) and reaction time (d) on membrane rejection and flux of MgCl2 and LiCl.

It can be seen from Figure 2(c) that the rejection of MgCl2 and LiCl changes little and the fluxes of both MgCl2 and LiCl solutions decrease as the TMC concentration increases from 0.05% to 0.20%, indicating that the concentration of TMC in organic solutions has a strong effect on the flux of composite membranes (Chen et al. 2002). This is because the polymerization rate increases with the increase of TMC concentration, so that a thick polyamide layer is formed. Figure 2(d) shows the relationship between membrane performance and reaction time. The salt rejection of MgCl2 and LiCl changes little as the reaction time increases, while the flux increases until the reaction time reaches 2 min, after which it decreases with further increase of the reaction time to 4 min. This is because more unreacted acyl chloride groups are introduced from TMC solution and then hydrolyzed to carboxyl groups that can improve the permeability of membrane (Chai & Krantz 1994; Fang et al. 2013). The polyamide layer on the membrane surface becomes thicker as the reaction proceeds, indicating that membranes with good performance can be obtained at the optimum reaction time.

Considering the rejection and flux of MgCl2 and LiCl, the optimum fabrication conditions are as follows: the DAPP concentration is 1.0%, the DAPP immersion time is 2 min, the TMC concentration is 0.1%, and the reaction time is 2 min. All NF membranes were prepared under the optimum condition thereafter.

Characterization and properties of membranes

FTIR and XPS spectra of membrane surfaces

The chemical structures of M0, MP, MPC and MPPC membrane surfaces were analyzed by FTIR and XPS. Figure 3(a) shows a new peak at 3,385 cm−1 for MP and MPC, which corresponds to O-H and N-H vibration. The peaks at 1,644/1,673 cm−1 (amide I, C=O band) and 1,576 cm−1 (amide II, C-N stretch) are arising from IP, while those at 1,072 and 1,011 cm−1 are attributed to the stretching vibration of -COO originating from the hydrolysis of –COCl in TMC (Zhang et al. 2015). The vibration strength of MP, MPC and MPPC membranes at 1,644/1,673 cm−1 (amide I, C=O band) and 1,576 cm−1 (amide II, C-N stretch) is higher than that of M0 membrane. These characteristic bands suggest that the IP reaction occurs on the surface of PES membranes.

Figure 3

FTIR (a) and XPS (b) spectra of M0, MP, MPC and MPPC membranes.

Figure 3

FTIR (a) and XPS (b) spectra of M0, MP, MPC and MPPC membranes.

The relative elemental contents and XPS spectra of the active layer are shown in Figure 3(b). Three major emission peaks are observed at 532, 400 and 285 eV, which are attributed to O1s, N1s and C1s, respectively. There are two small peaks at 232 and 168 eV, which are attributed to Cl2p and S2p, respectively. However, the content of Cl is very low. Compared with M0 membrane, the relative content of N and O elements increases, while that of C element decreases in MP membrane, which is attributed to the IP reaction between DAPP and TMC on the membrane surface. However, the content of N element is further increased from 12.13% to 16.01% in MPC and to 14.50% in MPPC due to the high percentage of N element in g-C3N4 and PDA-C3N4, suggesting that a dense polyamine layer is formed. The high-resolution C1s and N1s XPS spectra of M0, MP, MPC and MPPC membranes were deconvoluted and fitted to obtain more information about the membrane surface chemistry. Figure S3 (Supplementary Material) shows that there are four carbon species at 291.5, 286.3, 285.8 and 284.8 eV in the C1s XPS spectra of M0 membrane, corresponding to the C-S, C-O, C-N and C-C bond of PES, respectively. For MP and MPC membranes, the peaks at 287.7, 285.8 and 284.8 eV could be assigned to the O=C-N bond of amide, C-N bond of amine and C-C bond, respectively. The peak of O=C-N bond indicates that the IP reaction has occurred. For MPPC membrane, the peaks at 287.7, 286.3, 285.8 and 284.8 eV correspond to the O=C-N bond of amide, C-O band of PDA and phenolic hydroxyl, C-N bond of amine and C-C bond, respectively (Zhao et al. 2014). In the N1s XPS spectra, for M0 membrane, the peak at 399.4 eV could be attributed to the C-N bond of PES. For MP membrane, the peaks at 402.4, 399.7 and 389.9 eV correspond to the protonation of N, O=C-N bond of amide and C-N bond, respectively. For MPC membrane, three peaks are observed at 401.4, 399.7 and 389.9 eV, which correspond to the protonation of N and terminal amino functions (C-N-H), O=C-N bond of amide and C-N bond (Hai et al. 2016). Compared with MPC membrane, a new peak can be seen at 398.4 eV in MPPC membrane, which is attributed to sp2-hybridized nitrogen (C=N-C) of PDA and g-C3N4 (Wang et al. 2015). The above XPS spectra suggest that a new amide group between DAPP and TMC has been formed and PDA-C3N4 or g-C3N4 has been successfully anchored on the active layer.

Surface morphology and roughness of membranes

The surface and cross-sectional SEM images of M0, MP, MPC and MPPC membranes are shown in Figure 4. Many pores can be observed on the surface of M0 membrane with finger- and spongy-like structures, which is characteristic of membranes prepared by non-solvent induced phase inversion method (Li et al. 2010). However, the surface of MP, MPC and MPPC membranes is denser with no visible pores compared with M0 membrane. Notably, the surface of MP membrane is smooth with many grains on the surface which may be caused by the aggregation of g-C3N4 sheets. However, the MPPC membrane has a denser surface and more grains than the MPC membrane, which is attributed to the introduction of PDA-C3N4 sheets. The cross-sectional images show that MP, MPC and MPPC membranes are comprised of a skin layer, and their thickness increases with the introduction of g-C3N4 or PDA-C3N4 sheets due to the lamellar structure of g-C3N4 or PDA-C3N4 sheets.

Figure 4

Surface and cross-sectional morphology of M0 ((A) and (a)), MP ((B) and (b)), MPC ((C) and (c)), and MPPC ((D) and (d)) membranes.

Figure 4

Surface and cross-sectional morphology of M0 ((A) and (a)), MP ((B) and (b)), MPC ((C) and (c)), and MPPC ((D) and (d)) membranes.

Figure 5 shows the 3D AFM images and surface roughness of M0, MP, MPC and MPPC membranes. It is seen that the roughness of membranes follows the order of MP membrane > M0 membrane > MPC membrane > MPPC membrane. The surface of M0 membrane is smooth with a Ra of 4.385 nm and a Rms of 5.458 nm, while that of MP membrane is rough with a Ra of 6.792 nm and a Rms of 8.809 nm. However, the surface of MPC and MPPC membranes is smoother than that of M0 and MP membranes, which is partly due to the formation of hydrogen bands between the polyamide layer and g-C3N4 or PDA-C3N4 sheets (Shahabi et al. 2019), and partly to the fact that the PDA-C3N4 sheet has a smoother and more uniform structure than the g-C3N4 sheet.

Figure 5

AFM images of M0 (a), MP (b), MPC (c) and MPPC (d) membranes.

Figure 5

AFM images of M0 (a), MP (b), MPC (c) and MPPC (d) membranes.

Hydrophilicity of membranes

The surface hydrophilicity can be indicative of the permeate flux and anti-fouling performance of membranes. In order to evaluate the hydrophilicity of membranes more accurately, the dynamic contact angle was measured. Figure 6(a) shows that the contact angle of MP and MPC membranes is higher than that of M0 membrane. Although the contact angle of MPPC membrane exhibits the same trend as that of MP and MPC membranes in the first 60 s, it is lower than that of M0 membrane in the last 60 s, indicating an increase in hydrophilicity of MPPC membrane due to the introduction of PDA-C3N4 in the active layer.

Figure 6

Contact angles of M0, MP, MPC and MPPC membranes (a), surface zeta potential of M0, MP and MPPC membranes (b) and rejection curves of neutral solutes for MPPC membrane (c).

Figure 6

Contact angles of M0, MP, MPC and MPPC membranes (a), surface zeta potential of M0, MP and MPPC membranes (b) and rejection curves of neutral solutes for MPPC membrane (c).

Zeta potential of membranes

The zeta potential values of M0, MP and MPPC membranes are shown in Figure 6(b). Their isoelectric points are pH = 3.63, 6.00 and 6.69, respectively, indicating that MP and MPPC membranes are positively charged at pH < 6.00 and pH < 6.69, respectively. Therefore, the electrolyte solutes with higher cationic charged densities and/or lower anionic charged densities are rejected more effectively (Veríssimo et al. 2006). The surface zeta potential curve of MPPC membrane has the typical shape of amphoteric surfaces, indicating the presence of basic groups (-NH2 and =NH) and acidic groups (-COOH) on the surface. Generally, the NF membrane was used in pH ≈ 6.0 (Chiang et al. 2009; Lv et al. 2015), so the MPPC membrane is positively charged during the NF operation process.

Molecular weight cut-off of MPPC membrane

The MWCO can be used to evaluate the pore size of the MPPC membrane, which is obtained by the molecular weight as the rejection of neutral solutes reaches 90% (Liu et al. 2017). Figure 6(c) shows the rejection curves of MPPC membrane to glucose, sucrose, raffinose and PEGs. It shows that the MWCO of MPPC membrane is about 740 Da, demonstrating that the MPPC membrane is an NF membrane due to the formation of a dense and compact active layer on the surface of membrane.

Separation performance and fouling resistance of composite membranes

Permeation and separation performance of composite membranes

The permeation and separation performance of composite membranes for 2.0 g·L−1 MgCl2 and 2.0 g·L−1 LiCl solutions were measured, respectively. Figure 7(a) shows that the addition of g-C3N4 in DAPP solution results in a decrease in the permeate fluxes of composite membranes, while the opposite is true for the addition of PDA-C3N4. The permeate flux of the MPPC membrane with MgCl2 and LiCl solutions is 17.3 and 24.0 L·m−2·h−1, respectively, which is higher than that of MP membrane. The rejection for MgCl2 and LiCl is above 89.0% and below 45.0% for all membranes, respectively. It is noted that the MPPC membrane has the lowest rejection for LiCl and high rejection for MgCl2, and thus it can be used to separate monovalent and multivalent ions. In conclusion, the MPPC membrane outperforms other membranes in terms of permeation and rejection performance.

Figure 7

Permeate flux and salt rejection of the prepared membranes for 2.0 g·L−1 MgCl2 and LiCl solution (a) and divalent salts solution (b).

Figure 7

Permeate flux and salt rejection of the prepared membranes for 2.0 g·L−1 MgCl2 and LiCl solution (a) and divalent salts solution (b).

The rejection and flux of MP and MPPC membranes for divalent salts are shown in Figure 7(b). It is clear that the rejection of MP and MPPC membranes for ions follows the order of Mg2+ > Ba2+ ≈ Ca2+ > Cu2+ > Zn2+, which can be explained by the hydrated radius and diffusion coefficient of ions (Table 2). For positively charged membranes, the rejection of divalent cations with a larger hydrated radius or Stokes radius is higher than that with a smaller hydrated radius or Stokes radius. The diffusion coefficient plays an important role in rejection. It is seen that Ba2+ has the lowest diffusion coefficient, indicating that Ba2+ has the lowest diffusion rate. The opposite is true for Zn2+ with a relatively larger hydrated radius and a higher diffusion coefficient. As a consequence, the rejection is determined not only by the hydrated radius but also by the diffusion coefficient.

Table 2

Bulk diffusion coefficient and radius

SoluteDiffusion coefficient (Lide 2002) (10−5)/cm2·s−1Stokes radius (Nightingale 1959) (25 °C)/nmHydrated radius (Nightingale 1959) (25 °C)/nm
Li+ 1.029 0.238 0.382 
Mg2+ 0.706 0.347 0.428 
Ca2+ 0.792 0.310 0.412 
Ba2+ 0.541 0.290 0.404 
Cu2+ 0.714 0.325 0.419 
Zn2+ 0.703 0.349 0.430 
Cl 2.032 0.121 0.332 
SoluteDiffusion coefficient (Lide 2002) (10−5)/cm2·s−1Stokes radius (Nightingale 1959) (25 °C)/nmHydrated radius (Nightingale 1959) (25 °C)/nm
Li+ 1.029 0.238 0.382 
Mg2+ 0.706 0.347 0.428 
Ca2+ 0.792 0.310 0.412 
Ba2+ 0.541 0.290 0.404 
Cu2+ 0.714 0.325 0.419 
Zn2+ 0.703 0.349 0.430 
Cl 2.032 0.121 0.332 

Figure 7(b) also exhibits that the permeate flux of MPPC membrane increases dramatically compared with the MP membrane. For MPPC membrane, the salt permeate flux follows the order of Cu2+ > Zn2+ > Ca2+ > Ba2+. It is known that the hydrophilicity of membranes increases as the contact angle decreases. As discussed in Section ‘Hydrophilicity of membranes’, MPPC membrane has a lower contact angle than MP membrane, indicating higher hydrophilicity and permeate flux. The hydroxyl and amino groups originating from PDA-C3N4 sheets can increase the hydrophilicity of MPPC membrane (Dong et al. 2014; Shahabi et al. 2019), and the pore size and the lamellar structure of PDA-C3N4 sheets also can contribute to enhancing the permeation (Kim & Nair 2013).

Fouling resistance performance of MPPC membrane

The fouling resistance performance of M0 and MPPC membranes was assessed by testing the filtration performance for BSA solution. Clearly, the MPPC membrane performs well in resisting protein adsorption. The relative flux values and fouling resistance properties of M0 and MPPC membranes during the cyclic filtration of 0.1 g·L−1 BSA solution are illustrated in Figure 8. For each circle (Figure 8(a)), the stable permeate flux of MPPC membrane is lower than the pure water flux, indicating that the adsorption/deposition and back diffusion of BSA molecules reaches an equilibration (Bi et al. 2013). Figure 8(b) shows that the MPPC membrane has good fouling resistance properties (FRR, FRt, FRr and FRir) to protein. As the filtration cycle increases, the values of FRt and FRr increase, while the values of FRR and FRir change slightly. These results confirm that the MPPC membrane with PDA-C3N4 can efficiently reduce membrane fouling and exhibits good stability during the cyclic protein filtration.

Figure 8

Fouling behavior (a) and fouling resistance properties (FRR, FRt, FRr, FRir) (b) of M0 and MPPC membranes.

Figure 8

Fouling behavior (a) and fouling resistance properties (FRR, FRt, FRr, FRir) (b) of M0 and MPPC membranes.

CONCLUSIONS

g-C3N4 was modified by PDA to improve its stability and hydrophilicity. A positively charged NF membrane was prepared for separating divalent ions by IP using DAPP as aqueous monomer and TMC as organic monomer on PES support membrane, and PDA-C3N4 was added in the DAPP solution to improve the permeability of composite membranes. The presence of PDA-C3N4 in the active layer of NF membrane is favorable to improve the permeability, especially for divalent ions. For example, the permeate flux for CuCl2 is increased from 16.7 to 29.2 L·m2·h−1, while the salt rejection changes only slightly. The fabricated membrane has good fouling resistance performance for BSA. In conclusion, the membrane has great potential for separating divalent cations from wastewater, and g-C3N4 can be used in the membrane industry for water treatment.

ACKNOWLEDGEMENTS

This research is financially supported by the National Natural Science Foundation of China (21764011), the Foundation from Qinghai Science and Technology Department (2020-HZ-808) and Thousand Talents Program of Qinghai Province.

SUPPLEMENTARY MATERIAL

The Supplementary Material for this paper is available online at https://dx.doi.org/10.2166/wst.2020.092.

REFERENCES

REFERENCES
Cadotte
J. E.
1977
Reverse osmosis membrane
.
US Patent
4039
,
440
.
Cadotte
J. E.
1981
Interfacially synthesized reverse osmosis membrane
.
US Patent
4277
,
344
.
Cadotte
J. E.
Petersen
R. J.
Larson
R. E.
Erickson
E. E.
1980
A new thin-film composite seawater reverse osmosis membrane
.
Desalination
32
,
25
31
.
Cao
K. T.
Jiang
Z. Y.
Zhang
X. S.
Zhang
Y. M.
Zhao
J.
Xing
R. S.
Yang
S.
Gao
C. Y.
Pan
F. S.
2015
Highly water-selective hybrid membrane by incorporating g-C3N4 nanosheets into polymer matrix
.
J. Membr. Sci.
490
,
72
83
.
Chen
S. H.
Chang
D. J.
Liou
M. R.
Hsu
C. S.
Lin
S. S.
2002
Preparation and separation properties of polyamide nanofiltration membrane
.
J. Appl. Polym. Sci.
83
,
1112
1118
.
Chiang
Y. C.
Hsub
Y. Z.
Ruaan
R. C.
Chuang
C. J.
Tung
K. L.
2009
Nanofiltration membranes synthesized from hyperbranched polyethyleneimine
.
J. Membr. Sci.
326
(
1
),
19
26
.
Li
M. M.
Xu
J.
Chang
C. Y.
Feng
C. C.
Zhang
L. L.
Tang
Y. Y.
Gao
C. J.
2014
Bioinspired fabrication of composite nanofiltration membrane based on the formation of DA/PEI layer followed by cross-linking
.
J. Membr. Sci.
459
,
62
71
.
Lide
D. R.
2002
Handbook of Chemistry and Physics
.
CRC Press
,
Boca Raton, FL, USA
.
5-93-94
.
Pontalier
P. Y.
Ismail
A.
Ghoul
M.
1997
Mechanisms for the selective rejection of solutes in nanofiltration membranes
.
Sep. Purif. Technol.
12
,
175
181
.
Tian
Z. Z.
Wang
S. F.
Wang
Y. T.
Ma
X. R.
Cao
K. T.
Peng
D. D.
Wu
X. Y.
Wu
H.
Jiang
Z. Y.
2016
Enhanced gas separation performance of mixed matrix membranes from graphitic carbon nitride nanosheets and polymers of intrinsic microporosity
.
J. Membr. Sci.
514
,
15
24
.
Wen
X. M.
Ma
P. H.
Zhu
C. L.
He
Q.
Deng
X. C.
2006
Preliminary study on recovering lithium chloride from lithium-containing waters by nanofiltration
.
Desalination
49
,
230
236
.
Zhao
J. J.
Su
Y. L.
He
X.
Zhao
X. T.
Li
Y. F.
Zhang
R. N.
Jiang
Z. Y.
2014
Dopamine composite nanofiltration membranes prepared by self-polymerization and interfacial polymerization
.
J. Membr. Sci.
465
,
41
48
.

Supplementary data