Zhangbin
Guan‡
ab,
Bingyu
Wang‡
ab,
Yan
Wang
ab,
Jing
Chen
ab,
Chunyang
Bao
ab and
Qiang
Zhang
*ab
aKey Laboratory of New Membrane Materials, Ministry of Industry and Information Technology, School of Environmental and Biological Engineering, Nanjing University of Science and Technology, Nanjing 210094, P. R. China. E-mail: zhangqiang@njust.edu.cn
bInstitute of Polymer Ecomaterials, School of Environmental and Biological Engineering, Nanjing University of Science and Technology, Nanjing 210094, P. R. China
First published on 1st December 2021
Poly(ionic liquid)s (PILs) are widely used to improve the anti-fouling ability of membranes due to their high charge density and excellent hydrophilicity. However, it is difficult to eradicate the irreversible pollutants deposited inside the pores of membranes. In this study, an iron-containing PIL (Fe-PIL) membrane was constructed for the first time to enhance the anti-fouling property of a membrane via a heterogeneous Fenton reaction. Cu(0)-mediated reversible deactivation radical polymerization was used to synthesize polysulfone-based block copolymers with pyridine pendant groups, which were then quaternized for coordination with Fe(II) bromide to yield the targeted Fe-PILs. The presence of Fe-PILs enabled the heterogeneous Fenton reaction within the membrane pores and exhibited a wide range of pH tolerance, which could altogether remove contaminants within minutes. The membrane showed good anti-fouling performance and reusability. In addition, membrane filtration and the heterogeneous Fenton reaction exhibited a synergistic effect due to the confinement effect, improving the permeability and rejection of the membrane. This research developed a novel synthetic strategy and catalytic function of Fe-PILs for advanced membranes.
Poly(ionic liquid)s (PILs) are functional materials that combine the unique properties of ionic liquids with the functionalities of polymer materials, such as excellent stability, processability, and flexibility.7 Recently, PILs have been successfully used to prepare membranes, showing a wide range of applications in gas separation,8 fuel cells,9 seawater desalinization,10 acid recovery,11 antibacterial treatment,12 protein concentration,13 amino acid separation,14 and pigment wastewater treatment.15 PILs, as a kind of polyelectrolyte, had good hydrophilicity due to their charged structure, which made the polymer membrane less prone to fouling and easier to clean.16 It could reduce membrane fouling to a large extent; however, the irreversible pollutants that have been deposited on the surface of the polymer membrane are still difficult to remove.
A heterogeneous Fenton reaction overcame the shortcomings of the original Fenton reaction, such as being restricted by pH, producing a large amount of iron mud, etc.17–19 It has received increasing attention as a novel antifouling strategy in membrane science.20 However, when metal-based catalysts, such as Fe2O3,21 MnO2,22 TiO2,23etc., were directly loaded into the membrane, it would cause problems such as incompatibility with the polymer and higher roughness of the membrane surface. On the other hand, organometallic polymers with good solubility in the casting solution and high compatibility with other polymers could lower the surface roughness of the membrane, further reduce the degree of membrane fouling, and avoid the blockage of the membrane pores by nanoparticles.24 Wu et al. (2017) synthesized a series of polystyrene-b-PIL block copolymers to prepare a nano gold modified honeycomb membrane, which was successfully used for photocatalytic degradation of Congo red dye.25 Zheng et al. (2017) prepared an imidazolium-type iron-containing PIL (Fe-PIL) membrane to study its effect on the antibacterial activity of Staphylococcus aureus (S. aureus) and Escherichia coli (E. coli).26 In addition, iron-containing porous polyelectrolyte membranes (PPM) can be prepared based on the ionic complex between imidazolium-based PILs and 1,1-ferrocene dicarboxylic acid.27 However, as far as we know, there are few studies on the use of Fe-PILs as catalysts for the Fenton reaction. The application of Fe-PILs to prepare membranes with enhanced anti-fouling ability has not ever been reported.
Post-polymerization modification, such as quaternization and an anion exchange reaction, effectively prepared PILs with defined structures.28 At present, living/controlled free radical polymerizations such as reversible addition fragmentation chain transfer radical polymerization (RAFT) and atom transfer radical polymerization (ATRP) are commonly used for synthesizing well-defined PILs.29,30 Among them, RAFT is well known for its versatility to a wide range of monomers and solvents.31 The use of ATRP was relatively limited, mainly due to the competition of nitrogen-containing monomers with the ATRP catalyst system.32 For example, 4-vinylpyridine could be complexed with metal ions through σ bonds and π bonds, resulting in incomplete polymerization and wide PDI, making it difficult to control the molecular weight precisely.33 The newly developed Cu(0)-mediated reversible deactivation radical polymerization (Cu(0)-RDRP) could tolerate nitrogen-containing monomers, which reacted in the aqueous system at low temperatures.34–36 We hypothesize that well-defined Fe-PILs could be synthesized by Cu(0)-RDRP following post-polymerization modification and anion exchange from pyridine-containing functional polymers.
Herein, Fe-PILs have been successfully synthesized for the fabrication of heterogeneous Fenton membranes. Functional block copolymers with pyridine pendant groups were synthesized by Cu(0)-RDRP to prepare polysulfone blend membranes. A subsequent quaternization reaction and coordination with Fe(II) bromide successfully generated Fe-PILs on the surface of the membrane. The PIL membrane showed low surface roughness, excellent anti-fouling performance, and catalytic ability to mediate heterogeneous Fenton reactions. The catalytic and self-cleaning performances of the prepared Fe-PIL membranes were evaluated using methyl blue (MB). The synergistic reaction mechanism between membrane filtration and the Fenton reaction was also discussed.
Scheme 1 Synthesis of the poly(4-vinylpyridine)-b-polysulfone-b-poly(4-vinylpyridine) (PSF-b-P4VP) block copolymer via Cu(0)-RDRP. |
In this work, PSF-b-P4VP was obtained via direct Cu(0)-RDRP of 4-VP, which was conducted in a DMSO system at room temperature using chlorine terminated PSF (PSF-Cl) as the macroinitiator and Cu(0)/CuCl2/Me6TREN as the catalyst. The polymerization condition was [PSF-Cl]:[CuCl2]:[Me6TREN]:[4-VP] = [1]:[0.4]:[0.4]:[100] and the conversion could reach 93.5% after 24 hours according to the weight method. The results of FTIR, GPC, and 1H NMR were used to study the structure of PSF-based polymers.
The successful polymerization was directly verified using the FTIR spectra. As shown in Fig. 1A, a new peak appeared at 1760 cm−1 in the FTIR spectra of PSF-Cl compared with PSF-OH, attributed to the –COOR stretching vibration of PSF-Cl. After polymerization, new peaks at 1412 cm−1, 1557 cm−1, and 1591 cm−1 in the FTIR spectra of PSF-b-P4VP were due to the stretching vibration of the pyridine ring.37
Fig. 1 (A) Fourier transform infrared (FTIR) spectra and (B) gel permeation chromatography (GPC) spectra of PSF-based polymers. |
The successful synthesis of PSF-b-P4VP was also confirmed by GPC, which showed a relatively low polydispersity (Mw/Mn = 1.30, Fig. 1B) for the final polymer product PSF-b-P4VP. The Mn value of PSF-b-P4VP was significantly increased from 9500 Da (Mw/Mn = 1.22, PSF-Cl) to 27500 Da (Mw/Mn = 1.30), which indicated the success of Cu(0)-RDRP. However, the Mn value of PSF-Cl showed inconspicuous change from that of PSF-OH, which showed the slight increase in the molecular weight after the modification.
The chemical composition and structure of PSF, PSF-Cl, and PSF-b-P4VP could be determined using the 1H NMR spectra, as shown in Fig. 2. The aromatic protons of the PSF-OH main chain show chemical shifts at 6–8 ppm, and a new peak of PSF-Cl at 4.1 ppm was contributed by the methine proton of CPC, indicating the successful synthesis of the macroinitiator. Meanwhile, the 1H NMR spectra of PSF-b-P4VP further indicated the presence of the C–H group of the pyridine ring in 4-vinylpridine at 8.33 ppm and 6.39 ppm.37,38
Fig. 2 The molecular structure and 1H nuclear magnetic resonance (NMR) spectra of PSF-based polymers. |
According to Table S1,† the PSF-b-P4VP/PSF blend membranes (M1–M4) were prepared using various casting solutions constituted of PSF-b-P4VP and PSF in DI water via the NIPS process. Then, M4 was functionalized via quaternization with 3-bromo-1-propanol (M5) and a subsequent coordination reaction to generate Fe-PIL membranes (M6).
The top surface and cross-section SEM images of different membranes (M1–M6) are compared in Fig. 3. Fig. 3A1–F1 show that different membranes have a smooth surface without cracks, agglomeration, etc. A highly porous surface was observed in M5–M6 membranes (Fig. 3E1 and 3F1) compared with M1–M4 membranes (Fig. 3A1–D1). We believe that the appearance of surface pores was possibly due to the quaternization reaction in the bromo-1-propanol (10%) solution under thermal conditions. The previous study concluded that the combination of triblock copolymers and thermal and solvent annealing treatments could effectively increase the average pore size of the membrane,39 which supported this result of increasing the pore size of the M5 and M6 membranes (Table S2†). Furthermore, as shown in Fig. 3A2–F2, different cross-sectional membrane morphologies were due to various proportions casting solutions constitute. The cross-sectional image of the M1 membrane showed that the asymmetric structure of the M1 membrane was composed of a top dense skin layer, a middle finger-like structure, and a bottom cell-like structure.40 As the percentage of the polymer PSF-b-P4VP increased, the finger-like structure of the M2–M4 membranes became obvious. This can be explained as delayed mixing of the phase separation process resulting from the increasing PSF-b-P4VP content in the casting solution, leading to larger finger-shaped holes.41,42 With the introduction of PILs and Fe(II) to the membranes (M5 and M6), the larger finger-shaped structure was retained well, verifying that quaternizing the pyridine group and chelating Fe(II) on the membrane surface would not destroy the membrane structure.
The membrane surface roughness was further characterized by AFM as shown in Fig. 3A3–F3. The AFM scanning area is 5 μm × 5 μm, and the obtained roughness parameters (Ra) of all the membranes by AFM are shown in Table S2.† It can be found that as the PSF-b-P4VP content increased, the surface of M1–M4 became rougher (larger Ra value), which was attributed to the fact that self-separated hydrophilic P4VP segments on the top membrane promoted higher surface roughness.43 Interestingly, the roughness of the M5 membrane showed a downward trend from 17.01 nm to 10.02 nm after quaternization, which may be because the quaternized pyridyl group became positively charged, and the chains stretched to charge repulsion.44
To understand the roles of different surface elements in Fenton membranes, the surface elemental content and valence state in the M6 membrane were analyzed by XPS. The results in Table S3,†Fig. 4A, and Fig. S2† show that the M6 membrane was dominated by C (70.3%), O (23.5%), and Fe (2.9%), while M1 and M4 membranes were only dominated by C (78.5–79.3%) and O (15.4–16.7%), which confirmed that the successful preparation of Fe-PIL membrane (M6).
The C 1s, O 1s, and Fe 2p deconvolution of the composite M6 membrane is further shown in Fig. 4B–D. According to the C 1s spectra deconvolution (Fig. 4B), the binding energy peaks of 284.8 eV, 285.8 eV, and 288.0 eV were assigned to C–C/CC, C–O–C/C–OH, and CO in the M6 membrane, respectively. The O 1s spectra deconvolution of the M6 membrane in Fig. 4C showed three binding energies at 530.0 eV, 531.5 eV, and 533.0 eV, which shows the presence of lattice oxygen, C–O, and CO in the M6 membrane, respectively. The Fe 2p spectra deconvolution in Fig. 4D was at a binding energy of 710.0 eV and 712.5 eV for Fe 2p3/2, and 723.2 eV and 725.5 eV for Fe 2p1/2, which shows the presence of Fe(II) and Fe(III) in the M6 membrane.45,46 Hence, the rich hydroxyl groups content in the composite M6 membrane were attributed to the quaternization reaction. Moreover, the EDS images also showed that C, O, and Fe elements were homogeneously distributed in the surface of the M6 membrane (Fig. 4E–H), confirming that the Fe-PIL membrane had excellent dispersibility obtained by post-modification of the membrane. Besides, the excellent scalability resulting from the infinite combination of cations and anions in PILs also increased the compatibility and dispersity of C, O, and Fe elements.
In order to study the anti-fouling performance, the rejection ratio and flux of the membrane to BSA are shown in Fig. 5C. The rejection ratio of M1–M4 significantly increased (p < 0.05) from 97.8 ± 0.1% to 99.1 ± 0.1%, 98.9 ± 0.1%, and 98.7 ± 0.2%, respectively, attributed to pore size screening. Meanwhile, the BSA solution's flux to the membrane was dramatically promoted almost 22-fold from 7.2 ± 0.5 (M1) to 163.7 ± 6.0 L m−2 h−1 bar−1 (M4). It is worth noting that the M5 membrane rejection ratio (96.5 ± 0.4%) and flux (133.3 ± 10.1 L m−2 h−1 bar−1) to BSA were significantly lower (p < 0.05) than those of the M4 membrane after the quaternization reaction, which is primarily ascribed to the electrostatic attraction between the positively charged membrane and the negatively charged BSA in the PBS buffer solution (pH = 7.4), causing more severe membrane fouling.49 After coordinating with Fe(II), the electrostatic attraction between the M6 membrane and BSA was further increased, reducing the membrane flux to BSA (113.6 ± 3.1 L m−2 h−1 bar−1).
To further evaluate the antifouling ability of the membranes (M1–M6), the FRR, Rir, and Rr of the membranes are shown in Fig. 5D. Generally, 100% FRR indicates the robust antifouling ability of the membrane. The FRR of the M1 membrane was only 44.0%, and Rir was much higher than Rr, indicating that it belonged to irreversible pollutants. However, the M6 membrane had higher FRR (82.5%), and Rr was higher than Rir, which meant that compared with M1, M6 has more reversible pollutants. Fig. 5E further shows the relative flux change of different solutions (water and BSA solution) to the membrane over time. The relative flux is defined as the ratio of the real-time flux to the initial water flux. It was found that all membrane fluxes (Jw1) behave similarly with only a slight drop in 1 hour of pure water permeation. However, when pure water was converted to BSA solution, the membrane fluxes (Jp) showed a sharp drop due to a protein layer of the filter formed by deposition and adsorption to the surface of membranes. Different membranes were subjected to simple cleaning for the flux recovery test, and it was found that the M6 membrane flux recovered, which was observably higher than that of the M1 membrane. In summary, these results showed that with the addition of the PSF-b-P4VP block polymer and the quaternization/coordination reaction, the antifouling ability of the membrane was effectively improved.
Fig. 6A shows the degradation kinetics of MB by the M6 membrane at different reaction pH values. It can be seen that the MB degradation rate (k) and degradation efficiency (1 − C/C0) increased with decreasing pH, and pH 3 was the optimal pH, as shown in Fig. 6A and B. The results showed that acidic conditions favored MB removal in heterogeneous Fenton reaction membrane systems, which is consistent with the previous studies.50 However, this Fenton reaction system could also work under neutral or even alkaline conditions, which was a great advantage considering that it would avoid the necessity of tuning the pH.
The effect of the H2O2 dosage on the heterogeneous Fenton membrane reaction is shown in Fig. 6C. It was observed that the MB degradation rate (k) and degradation efficiency (1 − C/C0) were initially a noticeable increase at 0–2 mM H2O2 dosage and then a mild decrease at 2–5 mM, as shown in Fig. 6C and D. It was mainly because of ˙OH scavenging at a higher H2O2 dosage.51 Therefore, an H2O2 dosage of 2 mM was selected as the best H2O2 dosage.
Moreover, the reaction efficiency of the heterogeneous Fenton system was currently limited by insufficient Fe(III)/Fe(II) cycles. The reason for this phenomenon is that the reduction rate constant of Fe(III) to Fe(II) by H2O2 was about 4 orders of magnitude weaker than that of Fe(II) oxidation by H2O2, resulting in the reaction process consuming Fe(II) rapidly.52 Thus, the Fe(III)/Fe(II) cycles were promoted by adding the reducing agent AA to increase the reaction efficiency of the heterogeneous Fenton reaction. Thus, the effect of the AA dosage on the heterogeneous Fenton membrane reaction is shown in Fig. 6E. It was observed that the MB degradation rate (k) and degradation efficiency (1 − C/C0) were originally a surging rise at 0–2 mM AA dosage and then a slight drop at 2–5 mM, as shown in Fig. 6E and F. This might be because excessive AA will react with ˙OH, which would make the degradation effect worse.53 Therefore, an AA dosage of 2 mM was selected as the best AA dosage. In summary, the heterogeneous Fenton membrane could efficiently degrade MB in minutes in a broad pH range from 3.0 to 11.0. The addition of AA could significantly accelerate the Fenton reaction.
It is known that an ˙OH radical is the primary reactive species generated during Fenton catalytic degradation.50–52 To distinguish ˙OH in the Fenton membrane, the terephthalic acid (TA) scavenger was chosen in this study via fluorescent probe technology. TA (non-fluorescent) has been proposed to only react with ˙OH radicals and was converted into fluorescent 2-hydroxyterephthalic acid (HTA), which was measured using a fluorescence spectrophotometer with 315 nm excitation wavelength.54 As shown in Fig. 7B, it was found that the fluorescence intensity at 445 nm gradually increased with time, indicating that a large number of ˙OH radicals were generated during the reaction. This result proved that the generated ˙OH radicals were the primary reactive species during degradation.
In addition, the performance of the dynamic heterogeneous Fenton membrane reaction was evaluated in five consecutive cycles to evaluate the reusability of the membrane (Fig. 8B). It was found that the flux and rejection ratio only decreased by 4.9% and 3.1% after five cycles, respectively, indicating that the PSF-based Fe-PIL membrane has good stability in the actual membrane separation process. This result indicated that the PSF-based Fe-PIL membrane had excellent self-cleaning ability and separation stability. The synergy of membrane filtration and the Fenton process could increase permeability and rejection.
Footnotes |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/d1py01345a |
‡ These two authors contributed to this paper equally. |
This journal is © The Royal Society of Chemistry 2022 |