Fabrication and characterization of phosphorylated chitosan nanofiltration membranes with tunable surface charges and improved selectivities

Chemical Engineering Journal 352 (2018) 163–172

Contents lists available at ScienceDirect

Chemical Engineering Journal journal homepage: www.elsevier.com/locate/cej

Fabrication and characterization of phosphorylated chitosan nanofiltration membranes with tunable surface charges and improved selectivities

T

⁎

Yuefei Song , Qihua Hu, Tiemei Li, Yueke Sun, Xinxin Chen, Jing Fan Key Laboratory of Yellow River and Huai River Water Environmental and Pollution Control, Ministry of Education, School of Environment, Henan Normal University, Xinxiang 453007, China

H I GH L IG H T S

nanofiltration membranes were fabricated with tunable surface charges. • Novel with modulated substituted degree of phosphate group were synthesized. • PCSs and EDX analysis verified that it was introduced into the NFMs. • FTIR on NFMs’ permeation and selectivity performance were studied. • Impacts • NFM4 exhibits good chemical stability in the long-term operation.

A R T I C LE I N FO

A B S T R A C T

Keywords: Nanofiltration membrane Phosphorylated chitosan Degree of substitution Tunable surface charge Improved selectivity

Phosphorylated chitosan (PCS) with tailored amount of phosphate groups was synthesized, and novel composite nanofiltration membranes (NFMs) with tunable surface charges were prepared by coating PCS onto polyacrylonitrile (PAN) supporting layer and subsequently cross-linked by glutaraldehyde (GA). Chemical structures and compositions of PCS and NFMs, along with morphologies, surface charges, and performance of NFMs were represented by using FTIR, TG, ATR-IR, SEM-EDX, AFM, streaming potential analyzer, and cross-flow flat permeation test, respectively. The effect of different phosphorous abundance of PCS precursors on the surface charge and permselectivity of NFMs was investigated systematically. It was illustrated that the incorporation of phosphate group with high-substitute degree did enhance the surface charge and selectivity of NFM, respectively, while retaining its permeability. The resultant membrane showed a zeta potential of −77.6 mV at 1 mol·m−3 KCl electrolyte solution and pH 7.0, significantly more superior than that of the commercial DL membrane. For the feedwater (Na2SO4 + MgCl2, 1.0 g·L−1), the mass ratio of SO42−/Cl− and Mg2+/Na+ decreased from initial 0.5:1 in the feed to 1.58 × 10−2 and 0.254 in the permeate after filtration by the optimal NF membrane. Anionic dyes removal tests also confirmed the existence of negative charge characteristics from the prepared membranes, and pivotal role of the Donnan exclusion in separating performance. In addition, NFM4 exhibits good chemical stability in the long-term operation.

1. Introduction Over the past few decades, great advances in nanofiltration (NF) membrane have encouraged the widespread use for many purposes, including wastewater reclamation, water softening, desalination, whey demineralization, dye purification and so forth [1–3]. NF membrane no longer hinges solely on steric hindrance exclusion, but also rather depends on electrostatic interactions and dielectric effects [4,5], which results that NF process is desirable because of its selective separation of one solute over another [6]. The unique separating performance of NF membrane mentioned above has been reported to closely relate to the ⁎

Corresponding author. E-mail address: [email protected] (Y. Song).

https://doi.org/10.1016/j.cej.2018.07.010 Received 1 May 2018; Received in revised form 29 June 2018; Accepted 1 July 2018 Available online 04 July 2018 1385-8947/ © 2018 Elsevier B.V. All rights reserved.

porous structure, charged groups and ionic selectivity for its ultra-thin active layer [7]. Among these characteristics, there is no doubt that the surface charge plays an integral role in applications of NF membrane separation technology. For instance, Daraei et al. prepared a negative charged NF membrane by surface grafting of poly(acrylic acid), with an acid blue (–) rejection more than 95% [8]. Therefore, it is of great interest to exploit new membrane materials or modify traditional used materials to develop novel NF membranes with tailored/tunable functionalized active layer assuring high surface charge density, high hydrophilicity, as well as enhanced perm-selectivity [9]. Currently, many studies have proven that the preparation of novel

Chemical Engineering Journal 352 (2018) 163–172

Y. Song et al.

Nomenclature A cf,i cp,i Sanion

Scation t V β ΔEs ΔP

the membrane area, m2 concentration of solute i in the feed solution, mol·m−3 concentration of solute i in the permeate, mol·m−3 ions selectivities of anion, –

ions selectivities of cation, – the permeation time, h−1 the permeate volume, L the slope of a ΔEs-ΔP curve, mV·kPa−1 the streaming potential drops, mV the flow pressure drops, kPa

22.9 and 58.4 L·m−2·h−1, respectively [19]. Gao et al. developed a new O-(carboxymethyl)-chitosan NF membrane through surface functionalization with graphene oxide (GO) nanosheets. This membrane performed salt rejections of 92.9% and 62.3% for a feed Na2SO4 and NaCl concentration of 1.0 g·L−1, respectively, and the relevant permeate fluxes of 15.4 and 17.7 L·m−2·h−1 at 1.0 MPa and ambient temperature, and displayed higher salt rejection including monovalent and divalent constituents than the pristine membrane and most of the commercial NF membranes [20]. These studies showed that the active layers of NF membranes synthesized from existing CS derivatives could yield high salt rejection rates and moderate permeate fluxes, while the relatively low selectivity for divalent ions over monovalent ions suggested potentially low efficiency of water softening or other water treatment. So far, researches on novel NF membranes have mostly focused on improving the water flux and rejection rather than on ion selectivity [21]. Improved NF membrane selectivity will have a critical role in lowing energy consumption, eliminating the need for additional separation stages and adapting precise applications [22]. It has been shown that grafting high charged ion component to the functional layer could strength NF membrane surface charge and at the same time greatly alter the membrane selectivity [23]. Thus, as a notable high charge density polyelectrolyte, phosphorylated chitosan (PCS) appears to be one of the best candidate available to improve NF membrane surface performance substantially. The aim of this paper was to prepare a new kind of highly and tunable charged composite NF membrane from PCS with the flexibility of improving the membrane selectivity performance. Firstly, a series of the PCSs with modulated substituted degree of phosphate group, composition composed of CS and phosphorus pentoxide were synthesized. Secondly, their membranes were fabricated via surface coating on the polyacrylonitrile (PAN) supporting layer and chemical cross-linking reaction with glutaraldehyde (GA). Chemical compositions of PCSs and surface morphologies, properties of the active layers were characterized by FTIR, TG, ATR-IR, AFM, SEM and streaming potential analyzer. Finally, the regulation for preparation of PCS/PAN membranes with tunable surface charge and enhanced selectivity of divalent ions over monovalent ions and charged anionic dyes in standard conditions was discussed systematically, by manipulating of phosphorus content or substituted degree of phosphate group in the synthetic PCS precursors.

NF membrane with high surface charge density could be made by adding charged ion component to the membrane casting solution [10,11]. For example, Lin et al. informed that the fabrication of polyelectrolyte NF membranes with a Ni(OH)2 nanosheet layer allowed tunable surface charges, which could be regulated easily by the number of assembled layers [10]. Liu et al. found that modifying carbon nanotube with different charged polymers endowed the modified NF membranes possessing surface zeta potentials in the range of −40.81 to 17.46 mV at the pH of 7.0, and the antifouling data indicated that adjusting the surface charge could tune the interactions between the membrane surface and foulants, which could effectively suppress the irreversible fouling, resulting in 100% flux recovery [12]. Using the synthesized nano-MgO as positively charged precursor, a novel ceramic membrane was fabricated and exhibited high electrical performance [13]. Park et al. prepared a positively charged NF membrane that exhibited enhanced anti-fouling property by introducing a high-density positive charge derived from branched polyethyleneimine onto the selected membrane skin layer, and the corresponding zeta potential values suggested that the modified membrane had 19.52 mV of zeta potential, compared with −8.47 mV for the neat membrane under neutral condition [14]. All these findings confirmed that membrane surface charge density could be significantly enhanced after modification by grafting charged functional groups onto the membrane skin layer. Nevertheless, until recently, very few investigations dedicated to further understand the influence of charged ion component with different substitution degrees or abundance in the precursors on tunable surface charge density of NF membrane. Chitosan (CS) is a cationic polysaccharide usually acquired from the deacetylation of chitin polymer and routinely applied in membrane preparation for its easy surface-functionalization and environmental benignancy properties [15]. In recent years, CS and its derivatives have been used more often as NF membrane materials to form thin active layers aimed at overcoming the permeability and selectivity trade-off [16,17]. Modification processes mainly focus on a series of substitution reactions of hydroxyls on C-3 and C-6 positions, amino group on C-2 position of CS monomers [18]. Miao et al. prepared amphoteric composite NF membranes by grafting sulphated group onto CS by hexamethylene diisocyanate or epichlorohydrin as cross-linker. At 18 °C and 0.45 MPa, the rejections of 7.0 mmol·L−1 K2SO4, KCl solutions for both modified membranes were 90.5% and 37.6%, with permeate fluxes of 10.8 and 11.7 L·m−2·h−1, 90.8% and 32.5%, with permeation fluxes of

Table 1 Main structure features and characteristics of the anionic dyes adopted in this study. Dye

Molecular structure N

Methylene blue H3C N CH3

Methyl orange

Molecular weight (g·mol−1)

Maximal absorption wave-length (λmax, nm)

319.9

664

327.3

464

287.2

374

CH3

S +

N

Cl CH3

O

H3C N

N

N

S

O

Na

H3C O O

Alizarine yellow R

C

O2N

N

N

ONa

OH

164

Chemical Engineering Journal 352 (2018) 163–172

Y. Song et al.

2. Experimental and methods

added into 1000 ml three-neck bottle. Then, after feeding nitrogen to the mixture, phosphorus pentoxide with different doses (5–20 g) was added slowly while stirring in an ice-water batch to keep the temperature under 5 °C. After 2 h, the ice-water bath was removed, and 400 ml mixed solvent of deionized water and ethanol with the volume ratio of 1:1 was added into the resulting mixture to rinse the product. Afterward, the obtained product was washed and centrifuged repeatedly in water-ethanol mixed solvent until the solution pH reached approximately 7.0. Finally, PCS was dried in a vacuum oven at 30 °C for 24 h and stored in a desiccator. Afterwards, the composite NF membranes (NFMs) were fabricated through surface coating and chemical cross-linking methods (as shown in Fig. 1b). PCS aqueous solutions with different degrees of substitution (DSs) of phosphate group and pH 7.6 adjusted with NaHCO3 saturated solution were prepared by dissolving a certain amount of PCS with different abundance of phosphorous into deionized water at the preoptimized weight ratio of 3.0 wt% and filtered with G3 sand filter. After degasification, PCS solution was coated onto the dry surface of PAN porous layer, followed by curing in a vacuum oven at 50 °C for 1 h. Subsequently, the cured membranes were immersed in an aqueous solution containing 2.0 vol% GA, crosslinked at 30 °C for 20 min, and then rinsed with acetone and deionized water extensively. Afterward, the resultant NF membrane was gained via drying at 50 °C for 1.5 h, and the membranes were immersed in 1.0 wt% sodium dithionite solution before the characterization and performance tests.

2.1. Materials Flat-sheet porous PAN ultrafiltration membrane with a pure water permeability of approximately 1100 L m−2·h−1·MPa−1 and an average molecular weight cut-off (MWCO) of 8 × 104 Da was purchased from Lanjing Membrane Technology & Engineering Co., Ltd. (Sahnghai, China) and used as the supporting membranes. CS with a deacetylation degree of 95% and phosphorus pentoxide, as well as methane sulfonic acid, provided by Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China), were applied to form the surface functional layer of the composite membrane. GA obtained from Deen Chemical Reagent Co., Ltd. (Tianjin, China) were used as the cross-linking agent. High-pure nitrogen (≥99.99%) was obtained from Yuxin Gas Manufacturing Co., Ltd. (Xinxiang, China). Inorganic salts KCl, MgCl2, and Na2SO4 of analytical grade were adopted as the model solutes to characterize the ion selectivity of the resultant composite membranes. Anionic dyes such as Methylene blue, Methyl orange and Alizarine yellow R were all purchased from Aladdin and also used as model solutes for studying the charged organic compounds’ removal performance by the fabricated membranes. Main structure features and characteristics, such as molecular weights, molecular structures and maximal absorption wavelengths (λmax) of the dyes adopted are demonstrated in Table 1. All other reagents including anhydrous ethanol, polyethylene glycol (PEG), sodium carbonate, ammonium molybdate, and potassium persulfate were also of analytical grade and employed without further purification.

2.3. Characterization Chemical structures of PCS samples and PCS/PAN composite NF membranes were characterized using a Fourier transform infrared spectroscopy (FTIR, NEXUS, America) and attenuated total reflection infrared spectroscopy (ATR-IR, AVATAR 370, America), respectively. Thermogravimetric analyse (TGA, STA409, Germany) was carried out under nitrogen atmosphere over a temperature range of 20–800 °C at a

2.2. Synthesis of PCS polymer and their composite membranes PCS was synthesized using a modified Nishi’s method as reported in the literature [24] and the reaction scheme was shown in Fig. 1a. In brief, 3.4 g chitosan powder and 28 ml methane sulfonic acid were

O

O (1)

OH

OH O

NHR

O

HO

O

O P

CH 3SO 3 H, 0-5

O

HO

P2 O5

NH 2

O P

O

ONa ONa O

O

HO

(2) Na 2CO 3

m

O

ONa ONa

a

NH 2

O

NHAc

n

m HO

P

n

ONa

O R=H or COCH 3

O

O O P

O

OR OR

O P

O

OR OR

O NH 2

O O O P O

O P

O

GA, 50 oC

CH

O O m RO P

NH 2

NH n

OR

O

OR OR

O NH 2

m

O

OR OR

HO

O O RO P

(CH 2)3

CH

O RO

O

NH 2

n

NH

O

O

O

RO

b

RO

P

O

O

GA: CHO(CH 2)3CHO

m

RO P O RO O R:Na, H

Fig. 1. Schematic illustrations for the synthesis of PCS (a) and the fabrication of their NF membranes (b). 165

P OR

O

O

OR

O

n

Chemical Engineering Journal 352 (2018) 163–172

Y. Song et al.

Table 2 The reaction conditions and compositions of PCSs. Sample

Molar ratio (P2O5:CS)

P (At. %)

C (At. %)

DS

PCS PCS PCS PCS

1 3 5 7

3.05 5.45 9.04 10.16

49.89 49.17 48.01 47.43

0.20 0.38 0.73 0.87

1 2 3 4

[PO4

0.11 0.20 0.33 0.37

3−

]

b

DS

Membrane

0.22 0.42 0.80 0.95

NFM-1 NFM-2 NFM-3 NFM-4

a,b DS was defined as the molar ratio of phosphate to chitosan monomer, and determined by EDX and GB11893-89, respectively.

heating rate of 5 °C·min−1 to assess the stability of PCS. The surface and cross-section morphologies of the NF membranes were characterized by field-emission scanning electron microscope (FESEM, JSM-6390LV, Japan). The pre-dried samples were coated with gold prior to these observations. The morphology and roughness of the NF membrane were also examined by atomic force microscope (AFM, SUPRA40, German). NF membrane samples were rinsed with deionized water and dried out under the vacuum at 30 °C for 12 h prior to each characterization. The DS of phosphate group in each PCS sample was ascertained from the elemental phosphorus content (%P) measured by energy dispersive Xray (EDX) spectroscopy attached to the SEM in the low-vacuum mode at 8 kV and calculated according to determination of total phosphorus method (GB 11893-89), and the calculation adapted from [25] provided by Eq. (1).

50 × (%P ) × (168 + 81 × DS ) × 95 mPCS

PCS2

PCS3

PCS4

4000

3500

1250

907 cm-1

1000

750

500

-1

Wavenumbers (cm ) Fig. 2. FTIR spectra of CS and different PCSs.

100

CS PCS3

PCS1 PCS4

PCS2

600

800

80

95 31

(1)

where mPCS is the mass of the synthetic PCS, the result derived from the right half of the formula expresses as phosphate group content (% PO43−).

60

40

20

0

2.4. Tangential steaming potential measurements

0

200

400 o

Temperature ( C)

Surface charge characteristics of PCS/PAN composite NF membranes were conducted using a self-made laboratory-scale tangential streaming potential analyzer with a symmetric clamping cell, which installed two identical membranes (16.09 cm2 section areas, 135 μm channel depth) and two Ag/AgCl reference eletrodes (RO305, China). Accordingly, streaming potential (Es) was measured at room temperature and a variety of pressures ranging from 20 to 200 kPa in 1 mol·m−3 KCl solution (pH = 7.0). The measurement procedure was the same as that explained in our previous work [26] (also provided in Supplementary information file as Fig. S1). Based on the fact that the convective flow of charge is closely related to the flow pressure drops (ΔP) through charged NF membrane, the slope of a ΔEs-ΔP curve (β) could reflect the membrane surface charge properties qualitatively and responsibly [15].

Fig. 3. Thermogravimetric curves of CS and PCS samples.

before cross-linking

Transmittion (%)

β = −(∂ΔEs / ∂ΔP )

1078 cm-1

1381 cm-1

TG (%)

DS =

3435 cm-1

PCS1

Transmittion (%)

a

CS

(2)

1627 cm-1

1228 cm-1

1111 cm-1

after cross-linking

1172 cm-1

2.5. Cross-flow permeation performance 2000

The whole permeation experiments were conducted at temperature of 25.0 ± 0.3 °C, pressure of 0.8 ± 0.03 MPa and pH of 6.5 ± 0.05 for electrolyte solutions and 6.8 ± 0.05 for anionic dye solutions in a closed-loop operation by employing a customized tangential flow NF membrane evaluating device (Fig. S2) with four parallel circular membrane vessels having an effective membrane area of 113.04 cm2. Each group of permeation experiment lasted 2 h in order to achieve the steady state. Water flux (J) and solute rejection (R) were calculated by the following equations.

1800

1600

1400

1200

1000

800

-1

Wavenumbers (cm ) Fig. 4. ATR-IR spectra of NFMs before and after cross-linking reaction.

J=

166

V A×t

(3)

Chemical Engineering Journal 352 (2018) 163–172

Y. Song et al.

c p, i ⎞ R (%) = ⎜⎛1− ⎟ × 100 ⎝ cf , i ⎠

FTIR analysis results of chitosan and PCSs with different contents of phosphate group are depicted in Fig. 2. Absorption peak at 3435 cm−1 reveals for chitosan and all PCSs, which may be attributed to H-bonded NH2 and OH stretching vibration. After phosphorylation, the new peak at 907 cm−1 attributed to the P-O-C stretching indicates that the phosphorylation mainly occurred on the hydroxyl group of C-3 or C-5 positions of chitosan monomer. PCSs are also characterized by strong peaks at 1380 cm−1 and 1078 cm−1, which correspond to P]O and PeO stretching, respectively. Additionally, the ratios of the heights of the above mentioned characteristic peaks from PCS1 to PCS4 also grow incrementally over phosphorus content during the reaction, which implies that the phosphate group is indeed introduced into PCS precursor and the next synthesised NFMs in a tunable manner. The phosphorylation was also characterized by TGA analysis (Fig. 3). It was found that TGA of both chitosan and PCSs exhibited three different stages of weight loss. The first stage starts at 20 °C and continues up to 226 °C for chitosan and around 191 for PCSs, respectively, which accompanied by 5% weight loss. This may correspond to the loss of adsorbed and bound water. Presumably, grafting phosphate group to PCSs’ structures might make them decompose more rapidly. Then, when temperature was raised from 226 to 410 °C for chitosan, and from 191 to 285 °C for PCSs, there were 64% and 51% weight loss due to the deoxygenation of oxygen-containing groups and the decomposition of partial backbone of these compounds. With a further augment in temperature up to 800 °C, the TGA curves of PCSs showed a little shift as compared to chitosan, implying an intensified thermal stability. The results indicated that all PCSs were thermally stable up to 191 °C and met the requirement of practical application.

(4)

where V is the permeate volume, t is the permeation time and A is the membrane area; cf,i and cp,i are the concentration of the rejected solute i in the feed water and in the permeate, respectively. Ions selectivities (S) of cation and anion describe the differential validity of divalent ion rejection and monovalent ion rejection, which were determined by Eqs. (5) and (6) [27].

Scation =

Sanion =

RMg2 + (5)

RNa+

RSO42 − (6)

RCl−

3. Results and discussion 3.1. Characterizations of PCS precursors The chemical structures and compositions of PCSs were determined by EDX and FTIR. The EDX analyses of these precursors (Table 2) confirmed that the phosphorus content of PCS hinges on the molar ratio of phosphoric anhydride to chitosan monomer, that is, phosphate group in PCS actually increases with increasing phosphoric anhydride agent. For example, the phosphorus content increased from 3.05% (nP2O5: nCS monomer = 1) to 9.04% (nP2O5: nCS monomer = 5), indicating that more phosphate group was introduced with the increase of the molar ratio of phosphoric anhydride agent to chitosan monomer during PCS synthesis. Besides, even though this adjustable parameter was prolonged continuously, the phosphoric element proportion and DS of phosphate group did not increase markedly. For the purpose of verification, the phosphorus content of PCS and DS of phosphate group were also analysed by GB11893-89 (national standard) and the relevant data were also demonstrated in Table 2. The results reaffirmed the variation regularity of both parameters, and the calculated DS of phosphate group was highly identical with the result determined by EDX spectroscopy.

3.2. Characterizations of NFMs NF membranes were prepared from a series of PCSs with varying DSs of phosphate group ranging from 0.22 to 0.95 by the preparation technique depicted in Section 2.2. The specific conditions of fabrication were presented as follows: casting solution concentration of 3.0 wt%, curing time of 1 h at 50 °C, GA concentration of 2.0 vol% and crosslinking time of 1.5 h at 50 °C. Fig. 4 presents the ATR-FTIR spectra of a pristine supporting layer

(a)

(b)

(c)

(d)

Fig. 5. SEM images of surface (a–c) and cross-section (d) of PCS composite NF membranes: (a) NFM-1; (b) NFM-3; (c, d) NFM-4. 167

Chemical Engineering Journal 352 (2018) 163–172

Y. Song et al.

RMS=10.28 nm

RMS=7.70 nm

(a)

(d) RMS=16.94 nm

RMS=12.55 nm

(c)

(d)

Fig. 6. AFM images of (a) NFM-1 surface, (b) NFM-2 surface, (b) NFM-3 surface and (d) NFM-4 surface.

Water contact angle (o)

80

content of 33 wt% is dense and smooth as well as non-porous, with no appreciable defections being visualized (Fig. 5b). But with continued increase of this value to 37 wt%, several anomalous agglomerated micro-particles evenly distributed on the surface of NFM4 (Fig. 5c). This evolutionary trend may be resulted from the excess PCS abundance effect, which occurred in the homogeneous cross-linking reaction with GA. Detailed cross-sectional examination (Fig. 5d) shows that the thickness of the potential optimal NF membrane (NFM-4) is 545 nm, which is a proper thickness of selective active layer of the usual composite NF membranes fabricated via the surface coating and chemical cross-linking method [15]. It could be seen from AFM images (Fig. 6) that, all the composite membranes display a typical peak-and-valley surface morphology. Nevertheless, physical irregularity and obvious nodular (hills and valleys) are more pronounced with the enlargement of DS of phosphate group. In the meantime, the root mean square roughness (RMS) values of the composite membranes NFM-1, NFM-2, NFM-3 and NFM4 are 7.70, 10.28, 12.55 and 16.94 nm, respectively, exhibiting a trend of raising surface roughness with the increasing phosphate group content of PCS in casting solution. The ascending of RMS values could be caused by the introduction of polymer PCS. Firstly, PCS with high DS of phosphate group (hydrophilic and polar functional group) solution commonly accompanies with lower viscosity, which favors instantaneous demixing, causing a rise in the RMS and a decline in uniformity. Additionally, polymer PCS with high DS of phosphate group has ever-growing negative charge density and decreasing percentage of positively charged group at the same time, which tends to increase the inter- and intra-chain electrostatic repulsion. It would rise the rotational flexibility and freedom of polymer chains [18], which might increase the surface roughness of NF membrane. The contact angle between the air-water interface and the membrane surface as a function of the amount of phosphate group in the skin-layer was demonstrated in Fig. 7. As listed in Fig. 7, the mean surface contact angle declines dramatically from about 63.2° of the neat PAN membrane to 57.3, 55.3, 52.2, and 50.9° of the composite membranes NFM1, NFM 2, NFM 3 and NFM 4, respectively. It could be

60

40

20

0

UF substrate NFM-1

NFM-2

NFM-3

NFM-4

Fig. 7. Surface contact angles of the PAN substrate and NFMs determined with de-ionized water at 25.0 °C.

with mere the PCS coating (PAN-PCS), and with the cross-linked PCS functional layer (PAN-PCS-GA). Compared with the spectrum of the PAN-PCS, there is a distinct band at 1111 cm−1 ascribed to CeC stretching vibration of the straight-chain alkanes. Besides, two sharper and stronger characteristic peaks located at 1627 (fatty secondary amine group) and 1172 cm−1 (ether group) proved that the crosslinking reaction happened between primary amine or hydroxyl attached to chitosan monomer and GA. After the treatment, the 1228 cm−1 peak corresponding to OeH group almost completely disappeared. ATR-IR results described above indicate that cross-linking reaction has surely occurred. Membrane morphological structure was checked through FESEM and AFM, respectively. Fig. 5 presents the membrane surface and crosssection FESEM micrographs of the NFMs. In contrast to the functional layer of NFM1 with initial phosphate content (11 wt%) that shows a porous morphology (Fig. 5a), the surface of NFM3 with phosphate 168

Chemical Engineering Journal 352 (2018) 163–172

Y. Song et al.

a

0

-60

with the increase of DS of phosphate group. This result mainly due to the enrichment of phosphate groups in PCS casting solution onto NF membrane surface. Besides, based on the AFM results that with the increase of phosphorus content, the composite membrane surface roughness gradually enlarged, which would also contribute to a smaller contact angle value [28].

-90

3.3. Surface charge performance and MWCO of the prepared NFMs

NFM-1 NFM-3

Streaming potential (mv)

-30

NFM-2 NFM-4

Surface charges of these candidate membranes were evaluated through measuring the streaming potentials and calculating the zeta potentials. As demonstrated in Fig. 8, the composite membrane surface incorporated with abundant phosphate group in PCS casting solution becomes more negative charged. With the DS of phosphate group in PCS samples increases from 0.22 to 0.95, the slopes of fitting straights between ΔEs and ΔP expressed as β value calculated according to Eq. (2) and the average zeta potential (determined using Eq. (S1)) using 1 mol·m−3 KCl aqueous solution at pH 7.0 decrease from −0.185 mV·kPa−1, −27.96 mV for NFM1 membrane to −0.318, −0.411 and −0.437 mV·kPa−1, −43.4, −70.9 and −77.6 mV for NFM2, NFM3 and NFM4 membranes, respectively. On the basis of carefully analysing published experimental results, NFM4 exhibits a more negative zeta potential than that of the commercial DK [25] and other membranes (Table S1). In addition, it should be noted that the composite NF membrane surface charge could be tuned through changing the DS of phosphate group in the PCS precursor. The increased surface negative charges of the composite NFM1, NFM2, NFM3 and NFM4 membranes could be attributed to a growing number of net charge between the anionic and cationic groups. Firstly, the deprotonated tendency of phosphate group as a strong electrolyte is stronger than the protonation of amino group, therefore the membrane behaves like a negatively charged membrane [29]. Secondly, the everincreasing phosphate group in PCS chains with relative higher DS value increased the negatively charged density. But additionally, both the shielding effect of the formation of hydrogen bonding between the phosphate group of PCS chains and unreacted hydroxyl group of PCS molecules and the covalently linked PCS molecules make contributions to weaken the surface negative charge increasing trend of the composite membranes [30]. MWCO of the developed NF composite membrane was detected through permeation experiments using model solute of PEG with different fractions. The rejection rates of the acquired membranes versus PEG molecular weight were depicted in Fig. 9. It can be found from Fig. 9 that the removal rate of the membrane attains 90% when the molecular weight of PEG is equal to 409, 615, 818 and 912 Da which is the MWCO of the NFM1, NFM2, NFM3 and NFM4 membrane, respectively. It indicated that incorporation of PCS especially with high phosphorus content greatly increases membrane MWCO, which is probably due to the decrease of the degree of crosslinking of the formed functional layer.

-120 -150 -180

0

40

80

120

160

200

Pressure drop (kPa)

Surface zeta potential (mV)

b

0

-20

-40

-60

-80

NFM-1

NFM-2

NFM-3

NFM-4

Fig. 8. Surface charge phenomena of the composite NFMs tested using 1 mol·m−3 KCl aqueous solution at pH 7.0 and 25.0 °C (a. streaming potential vs pressure drop, b. surface zeta potential at pressure drop 180 kPa).

100

Rejection (%)

80

60

NFM-1 NFM-2 NFM-3 NFM-4

40

3.4. NFMs’ permeation and selectivity performance with different contents of phosphate group

20 200

400

600

800

1000

By choosing NFM-4 and some other ones as the model membrane and comparative objects, effects of different DS values of phosphate group in PCSs on NFMs’ permeation and selectivity performance were studied in several cases, including electrolyte solution, anionic dyes solution and its long-term operation. NFMs with different phosphorous abundance were evaluated with 1.0 g·L−1 two-component ternary electrolyte solution (Na2SO4 + MgCl2) with the molar ratio of 1:1 (pH = 6.5) at 25 °C (Fig. 10). As shown in Fig. 10, with increasing the DS of phosphate group from NFM1 to NFM4, the permeate flux and all ions rejections increased remarkably under the fixed operating pressure, which could be mainly due to the ever-improving hydrophilicity. As such, NFM-4

Molecular weight of PEG (Da) Fig. 9. PEG rejection curve for the composite NFMs measured with 0.5 g·L−1 PEG aqueous solution at 0.8 MPa and 25.0 °C.

found that all composite NF membranes with the introduction of PCS polymer coating on PAN substrate exhibited a remarkably lower contact angle than that of virgin UF membrane. This is understandable because PCS polymer is much more hydrophobic than the PAN substrate. At the same time, the water contact angles decreased gradually 169

Chemical Engineering Journal 352 (2018) 163–172

Y. Song et al.

a

b 100 80

5

90

Rejection (%)

20

3

60

2

30 20

NFM-4 0 0.4

NFM-2

0.8

1.2

1.6

SO42-

10

NFM-1

Na

0 0.4

2.0

0.8

c 100

4

Sanion

1.2

1.6

2.0

0

5

60

2

30 2-

10 0 0.4

2+

SO4

Mg

Na+

Scation

0.8

1.2

Cl

1

-

70

3

60

2

30 20

SO42-

10

Sanion 1.6

Rejection (%)

3

4

80

Selectivity (-)

Rejection (%)

70

20

2.0

0.8

NFM-4, flux NFM-4, rejection

60 60 40

Rejection (%)

80

20

20

Methyl orange

Alizarine yellow R Methylene blue

Sanion 1.6

2.0

0

−1

aqueous solution at different operating pressure (b. NFM1, c.

effects strengthened gradually. Further investigation of surface charge effect on the NFM1 and NFM4 membranes separating performance for anionic dyes including methyl orange, alizarine yellow R and methylene blue was carried out. The cross-flow NF tests were carried out at a dye concentration of 0.1 g·L−1, pressure of 0.8 MPa, pH of 6.8 ± 0.05, and temperature of 25 ± 0.03 °C. As shown in Fig. 11, the membranes fluxes for all dyes are close to each other. Rejections acquired from NFM1 and NFM4 for methyl orange and methylene blue are 52. 6% and 64.2%, 49.5% and 57.2%, while alizarine yellow R are 71.9% and 87.5%, respectively, affirming the vital role of Donnan exclusion in the removal mechanism. As can be seen in Table 1, molecular weights of these adopted dyes are similar, but the polarities or negative charges of the indicative functional groups (eCOO− for alizarine yellow R, -SO3− for methyl orange and eC]S+ for methylene blue) show significant difference [31]. In comparison to NFM1 membrane, NFM4 membrane incorporated with a high phosphate group abundance becomes more negatively charged, which facilitates the electrostatic repulsion between the anionic dye molecules and the membrane surface and thus increases the rejection rates to dye molecules. Time-dependant flux, divalent ions rejection rates (Mg2+ and SO42−) and selectivity of the optimal NF membrane were researched to verify its chemical stability. The operating pressure was fixed at 0.8 MPa and other experiment conditions were set as mentioned in Fig. 10 and the tested results were demonstrated in Fig. 12. As depicted in Fig. 12, NFM4 showed good operating stability in terms of permeability and tunable selectivity, which was primarily ascribed to the stable chemical cross-linking structure of NFM4. It must be particularly pointed out that after 200 h of operation, Sanion and Scation for NFM4

100 80

40

Scation 1.2

1

Cl-

Operating pressure (MPa)

Fig. 10. Effects of the content of phosphate group on the NFMs performance tested with 1.0 g·L NFM2, d. NFM4).

NFM-1, flux NFM-1, rejection

Mg2+

Na+

0 0.4

0

Operating pressure (MPa)

Permeability (L·m-2·h-1·MPa-1)

Scation

1

90

80

0

Cl-

d 100

5

90

100

Mg2+

Operating pressure (MPa)

Operating pressure (MPa)

120

+

Selectivity (-)

Flux (L·m-2·h-1)

40

70

Selectivity (-)

4

80

60

0

Fig. 11. Dye removals and steady-state permeabilities of the tested NFM-1 and NFM-4 membranes.

was regarded as showing the optimum performance of electrolyte solution. Meanwhile, owing to the existing differential growth rates for both divalent ions (SO42− and Mg2+) and monovalent ions (Na+ and Cl−) with the enlargement of operating pressure, Sanion and Scation presented different changing trends. According to contrast test results, it showed beyond all doubt that membrane’ selectivity of bivalent ion over monovalent ion improved dramatically. For example, Sanion and Scation for NFM1 membrane were 3.25, 2.21 at transmembrane pressure of 1.2 MPa, while these values for NFM2 and NFM4 increased to 4.02 and 2.65, 4.18 and 3.21, respectively. This is primarily because of the increased membrane surface charge density, the Donnan and dielectric 170

Chemical Engineering Journal 352 (2018) 163–172

Y. Song et al.

References

5.0 4.5

90

3

80 RSO

2-

Sanion

RMg

2+

Flux

4

70

Scation 2

60

[1] J. Zhu, M. Tian, Y. Zhang, H. Zhang, J. Liu, Fabrication of a novel ‘‘loose’’ nanofiltration membrane by facileblending with Chitosan Montmorillonite nanosheets for dyes purification, Chem. Eng. J. 265 (2015) 184–193. [2] C. Feng, J. Xu, M. Li, Y. Tang, C. Gao, Studies on a novel nanofiltration membrane prepared by cross-linking of polyethyleneimine on polyacrylonitrile substrate, J. Membr. Sci. 451 (2014) 103–110. [3] L.H. Andrade, B.C. Ricci, L.B. Grossi, W.L. Pires, A.O. Aguiar, M.C.S. Amaral, Nanofiltration applied in gold mining effluent treatment: evaluation of chemical cleaning and membrane stability, Chem. Eng. J. 323 (2017) 545–556. [4] M. Bauman, A. Košak, A. Lobnik, I. Petrinic, T. Luxbacher, Nanofiltration membranes modified with alkoxysilanes: surface characterization using zeta-potential, Colloids Surf. A 422 (2013) 110–117. [5] L.L. Shao, Q.F. An, Y.L. Ji, Q. Zhao, X.S. Wang, B.K. Zhu, C.J. Gao, Preparation and characterization of sulfated carboxymethyl cellulose nanofiltration membranes with improved water permeability, Desalination 338 (2014) 74–83. [6] Y.C. Xu, Z.X. Wang, X.Q. Cheng, Y.C. Xiao, L. Shao, Positively charged nanofiltration membranes via economically mussel-substance-simulated co-deposition for textile wastewater treatment, Chem. Eng. J. 303 (2016) 555–564. [7] P. Mou, S.D. Jons, Chemistry and fabrication of polymeric nanofiltration membranes: a review, Polymer 103 (2016) 417–456. [8] P. Daraei, S.S. Madaeni, N. Ghaemi, M.A. Khadivi, L. Rajabi, A.A. Derakhshan, F. Seyedpour, PAA grafting onto new acrylate-alumoxane/PES mixed matrix nanoenhanced membrane: preparation, characterization and performance in dye removal, Chem. Eng. J. 221 (2013) 111–123. [9] A.W. Mohammad, Y.H. Teow, W.L. Ang, Y.T. Chung, D.L. Oatley-Radcliffe, N. Hilal, Nanofiltration membranes review: recent advances and future prospects, Desalination 356 (2015) 226–254. [10] Z. Lin, Q. Zhang, Y. Qu, M.M. Chen, F. Soyekwo, C.X. Lin, A.M. Zhu, Q.L. Liu, LBL assembled polyelectrolyte nanofiltration membranes with tunable surface charges and high permeation by employing a nanosheet sacrificial layer, J. Mater. Chem. A 5 (2017) 14819–14827. [11] X. Song, R.S. Zambare, S. Qi, B.N. Sowrirajalu, A.P.J. Selvaraj, C.Y. Tang, C. Gao, Charge gated ion transport through polyelectrolyte intercalated amine reduced graphene oxide membranes, ACS Appl. Mater. Inter. 9 (2017) 41482–41495. [12] Y. Liu, Y. Su, J. Cao, J. Guan, R. Zhang, M. He, L. Fan, Q. Zhang, Z. Jiang, Antifouling, high-flux oil/water separation carbon nanotube membranes by polymer-mediated surface charging and hydrophilization, J. Membr. Sci. 542 (2017) 254–263. [13] X. Meng, Z. Liu, C. Deng, M. Zhu, D. Wang, K. Li, Y. Deng, M. Jiang, Microporous nano-MgO/diatomite ceramic membrane with high positive surface charge for tetracycline removal, J. Hazard. Mater. 320 (2016) 495–503. [14] S.Y. Park, S.H. Choi, J.W. Chung, S.-Y. Kwak, Anti-scaling ultrafiltration/microfiltration (UF/MF) polyvinylidene fluoride (PVDF) membranes with positive surface charges for Ca2+/silica-rich wastewater treatment, J. Membr. Sci. 480 (2015) 122–128. [15] J. Miao, L.C. Zhang, H. Lin, A novel kind of thin film composite nanofiltration membrane with sulfated chitosan as the active layer material, Chem. Eng. Sci. 87 (2013) 152–159. [16] V.K. Thakur, S.I. Voicu, Recent advances in cellulose and chitosan based membranes for water purification: a concise review, Carbohydr. Polym. 146 (2016) 148–165. [17] H.B. Park, J. Kamcev, L.M. Robeson, M. Elimelech, B.D. Freeman, Maximizing the right stuff: the trade-off between membrane permeability and selectivity, Science 356 (2017) 1138–1148. [18] C. Zhou, X.-L. Gao, S.-S. Li, C.-J. Gao, Fabrication and characterization of novel composite nanofiltration membranes based on zwitterionic O-carboxymethyl chitosan, Desalination 317 (2013) 67–76. [19] J. Miao, H. Lin, W. Wang, L.C. Zhang, Amphoteric composite membranes for nanofiltration prepared from sulfated chitosan crosslinked with hexamethylene diisocyanate, Chem. Eng. J. 234 (2013) 132–139. [20] J. Wang, X. Gao, J. Wang, Y. Wei, Z. Li, C. Gao, O-(carboxymethyl)-chitosan nanofiltration membrane surface functionalized with graphene oxide nanosheets for enhanced desalting properties, ACS Appl. Mater. Inter. 7 (2015) 4381–4389. [21] J. Abraham, K.S. Vasu, C.D. Williams, K. Gopinadhan, Y. Su, C.T. Cherian, J. Dix, E. Prestat, S.J. Haigh, I.V. Grigorieva, P. Carbone, A.K. Geim, R.R. Nair, Tunable sieving of ions using graphene oxide membranes, Nat. Nanotechnol. 12 (2017) 546–550. [22] J.R. Werber, C.O. Osuji, M. Elimelech, Materials for next-generation desalination and water purification membranes, Nat. Rev. Mater. 1 (2016) 1–15. [23] L.Y. Ng, A.W. Mohammad, C.Y. Ng, A review on nanofiltration membrane fabrication and modification using polyelectrolytes: effective ways to develop membrane selective barriers and rejection capability, Adv. Colloids Interface 197 (2013) 85–107. [24] R. Jayakumar, N. Selvamurugan, S.V. Nair, S. Tokura, H. Tamura, Preparative methods of phosphorylated chitin and chitosan – an overview, Int. J. Biol. Macromol. 43 (2008) 221–225. [25] N.R. Pires, P.L.R. Cunha, J.S. Maciel, A.L. Angelim, V.M.M. Melo, R.C.M. de Paula, J.P.A. Feitosa, Sulfated chitosan as tear substitute with no antimicrobial activity, Carbohydr. Polym. 91 (2013) 92–99. [26] Y. Song, W. Qin, T. Li, Q. Hu, C. Gao, The role of nanofiltration membrane surface charge on the scale-prone ions concentration polarization for low or medium saline water softening, Desalination 432 (2018) 81–88. [27] Y. Song, J. Xu, Y. Xu, X. Gao, C. Gao, Performance of UF–NF integrated membrane

Selectivity (-)

Rejection (%) or Flux (L·m-2·h-1)

100

1

50 0

50

100

150

200

0

Operating time (h) Fig. 12. Time-dependant flux, divalent ions rejection rates (Mg2+ and SO42−) and selectivity of the candidate membrane NFM-4 tested with 1.0 g·L−1 electrolyte aqueous solution at 0.8 MPa, pH 6.5 and 25.0 °C.

membrane were kept at 4.49 and 3.15, respectively, which were still higher than those of many other NF membranes (presented in Table S2) ranged from 1.81–4.22 to 0.57–3.06 under the same testing conditions, respectively [32,33]. 4. Conclusions Thin-film composite PCS nanofiltration membranes with tunable surface charges and improved selectivities have been successfully fabricated through incorporation of PCS with tailored amount of phosphate group. FTIR and EDX analysis verified that the phosphate group was successfully introduced into the NFMs in a tunable manner, resulting in a beneficial increase in surface charge density, pore size, hydrophilic property and ionic selectivity. The effect of different phosphorous abundance of PCS precursors on the surface charge and permselectivity of NF membranes was investigated systematically. In detail, the acquired NFM4 showed a zeta potential of −77.6 mV at 1 mol·m−3 KCl electrolyte solution and pH 7.0, significantly higher than that of the commercial DK membrane. For the feed mixture of aqueous (Na2SO4 + MgCl2, 1.0 g·L−1), the mass ratio of SO42−/Cl− and Mg2+/ Na+ decreased from initial 0.5:1 in the feed to 1.58 × 10−2 and 0.254 in the permeate after filtration by the optimal NF membrane. Anionic dyes removal tests also confirmed the existence of negative charge characteristics from the prepared membranes, and pivotal role of the Donnan exclusion in separating performance. In addition, NFM4 exhibits good chemical stability in the long-term operation. In conclusion, the design of novel composite membrane by introducing different phosphorous abundance of PCS precursors provides an alternative strategy for fabricating NF membranes with high surface charges and improved selectivities. Acknowledgments This work was financially supported by Project funded by the National Natural Science Foundation of China (No. 21606074, 51604099), China Postdoctoral Science Foundation (No. 2016M592297, 2017T100533), Henan Province Postdoctoral Science Foundation (No. 2015073), Henan Normal University Fund for Excellent Young Scholars (2016YQ03). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at https://doi.org/10.1016/j.cej.2018.07.010. 171

Chemical Engineering Journal 352 (2018) 163–172

Y. Song et al.

[31] S. Yu, Z. Chen, Q. Cheng, Z. Lü, M. Liu, C. Gao, Application of thin-film composite hollow fiber membrane to submerged nanofiltration of anionic dye aqueous solutions, Sep. Purif. Technol. 88 (2012) 121–129. [32] M. Liu, C. Zhou, B. Dong, Z. Wu, L. Wang, S. Yu, C. Gao, Enhancing the permselectivity of thin-film composite poly(vinyl alcohol) (PVA) nanofiltration membrane by incorporating poly(sodium-p-styrene-sulfonate) (PSSNa), J. Membr. Sci. 463 (2014) 173–182. [33] L.Y. Ng, A.W. Mohammad, C.Y. Ng, C.P. Leo, R. Rohani, Development of nanofiltration membrane with high salt selectivity and performance stability using polyelectrolyte multilayers, Desalination 351 (2014) 19–26.

process for seawater softening, Desalination 276 (2011) 109–116. [28] H. Li, W. Shi, Y. Zhang, Q. Du, X. Qin, Y. Su, Improved performance of poly(piperazine amide) composite nanofiltration membranes by adding aluminum hydroxide nanospheres, Sep. Purif. Technol. 166 (2016) 240–251. [29] B. Li, L. Huang, X. Wang, J. Ma, F. Xie, L. Xia, Effect of micropores and citric acid on the bioactivity of phosphorylated chitosan/chitosan/hydroxyapatite composites, Ceram. Int. 39 (2013) 3423–3427. [30] E. Idil Mouhoumed, A. Szymczyk, A. Schäfer, L. Paugam, Y.H. La, Physico-chemical characterization of polyamide NF/RO membranes: insight from streaming current measurements, J. Membr. Sci. 461 (2014) 130–138.

172