CuFeS₂/MXene-Modified Polyvinylidene Fluoride Membrane for Antibiotics Removal through Peroxymonosulfate Activation

Abstract

In this research, the CuFeS₂/MXene-modified polyvinylidene fluoride (PVDF) membrane was prepared to activate peroxymonosulfate (PMS) to remove moxifloxacin (MOX) and its morphology; surface functional groups and hydrophilicity were also studied. The parameters of the catalytic membrane/PMS system were optimized, with an optimal loading of 4 mg/cm² and a PMS dosage of 0.20 mM. High filtration pressure, alkaline conditions, and impurities in water could inhibit MOX removal. After continuous filtration, the removal efficiency of MOX using the catalytic membrane/PMS system and PVDF membrane was 68.2% and 9.9%, respectively. Batch filtration could remove 87.8% MOX by the extra 10 min contact time between the catalytic membrane and solution. During the filtration process, CuFeS₂/MXene on the surface of the catalytic membrane activated PMS to produce SO₄^•−, HO^•, and ¹O₂, and MOX was removed through adsorption and degradation. Taking humic acid (HA) as the model foulant, reversible fouling resistance in the catalytic membrane/PMS system was 22.8% of the PVDF membrane. The catalytic membrane/PMS system weakened the formation of the cake layer by oxidizing HA into smaller pollutants and followed the intermediate blocking cake filtration model. The novelty of this research was to develop a CuFeS₂/MXene–PVDF membrane-activated PMS system and explore its application in antibiotics removal.

1. Introduction

The polyvinylidene fluoride (PVDF) membrane had excellent mechanical strength, chemical stability, and thermal stability. Its ultrafiltration process was widely used in drinking water production, wastewater treatment, reverse osmosis pretreatment, and so on [1]. However, membrane separation technology was only a physical process [2], and membrane fouling easily occurred due to the hydrophobicity and low surface energy of the PVDF membrane [3]. Typically, ultrafiltration membranes were not effective in removing antibiotics. PVDF ultrafiltration membranes removed 13.3% of sulfamethoxazole and 11.9% of tetracycline by filtering artificial water containing different antibiotics (50 μg/L), respectively [4].

Catalytic membranes were prepared by modifying the membranes with catalysts, which could couple with advanced oxidation technology. Combining oxidative degradation and membrane separation, catalytic membranes received attention for the removal of antibiotics. Alpatova et al. [5] immobilized Fe₂O₃ on a PVDF membrane in order to catalyze H₂O₂ to remove organic contaminants, which was beneficial for the reuse of heterogeneous catalysis. Li et al. [6] prepared the ZnO@Ti₃C₂T_x membrane to generate HO^• and ^•O₂⁻ under visible light, showing a good separation performance for six cycles with RhB rejection > 88.48%. Zhou et al. [7] prepared the D-UiO-66/graphite/PVDF membrane, producing HO^• and ^•O₂⁻ at a low current density (0.01 mA/cm²), with the removal efficiencies of four antibiotics more than 96.6%. Yue et al. [8] activated PMS with the Co²⁺/MXene membrane to effectively remove tetracycline (>97%) within 5 min and concluded that filtration enhanced the contact between PMS and Co²⁺. Heterogeneous catalysts activated persulfate to produce SO₄^•− with excellent oxidation properties, and the reaction system did not consume energy. The choice of catalytic materials was crucial for catalytic membranes to activate persulfate.

Metal oxides, zero-valent metals and the metal–organic framework were commonly used to activate persulfate, accompanied by changes in the valence state of metal elements. Polymetallic catalysts were effective at activating persulfate because the coupling reactions between different metals could increase the production of SO₄^•− [9]. Metal sulfides exhibited high activity in catalytic reactions, and S^2- could promote the redox circulation of multi-valence metals in persulfate activation reactions [10]. Chalcopyrite (CuFeS₂) was a kind of bimetallic mineral with abundant reserves and a wide distribution, which had potential applications in catalytic reactions. Cu⁺ and Fe²⁺ played a bimetallic synergy in PMS activation, and the reduction in S²⁻ promoted the regeneration of the lower valence metal. Due to the efficient electron transfer properties, CuFeS₂ was used to thermally activate H₂O₂ to degrade dye Rhodamine B [11]. Based on experimental data and theoretical calculations, Huang et al. [12] concluded that the Cu(I)/Fe(III) redox couple in CuFeS₂ was crucial to activating peroxydisulfate. Nie et al. [13] reported that CuFeS₂ could be used to activate PMS to degrade bisphenol A, with a degradation efficiency of 99.7% and a mineralization efficiency of 75% within 20 min. Zhou et al. [14] activated peroxymonosulfate (PMS) with 1 g/L of CuFeS₂ and 95.7% sulfisoxazole, which was degraded within 5 min. CuFeS₂/PMS could still degrade 83.1% bisphenol S in the five-cycle experiment [15], indicating its excellent reusability. These studies indicate that CuFeS₂ is an efficient catalyst for PMS activation.

However, heterogeneous metal catalysts still had particle aggregation and metal ion leaching, which could be alleviated by combining with carbon-based materials. C₃N₄–Fe–rGO prevented the acid leaching and agglomeration of Fe, which is conducive to maintaining the activity of the reaction site [16]. Biochar could improve the stability and dispersion of catalysts [17]. Carbon nanotubes could promote adsorption and electron transfer among the catalysts, PMS, and pollutants [18]. MXene could immobilize metal ions due to the surface with rich functional groups. Xia et al. [19] found that CoOOH was well dispersed on MXene and concluded that MXene, as a supporting material, could enhance the catalytic activity. For Fe₃O₄@MXene, Ti on the MXene surface was helpful for the acceleration of the Fe(II)/Fe(III) cycle, maintaining the activation of sodium percarbonate [20]. Xie et al. [21] prepared FeS₂@MXene for water electrolysis and concluded that MXene could reduce the agglomeration phenomenon of FeS₂ and regulate the electron density. MXene could be used as the carrier of the transition metal catalyst and play a synergistic role in the activation reaction of persulfate. Therefore, it was hypothesized that the combination of CuFeS₂ and MXene could effectively activate PMS. CuFeS₂/MXene catalytic membranes could potentially be utilized for the removal of antibiotics in water treatment applications.

In addition, some research showed that catalytic membranes could play a role in mitigating the formation of membrane fouling. The g-C₃N₄ composite membrane prepared by Li et al. [22] exhibited a self-cleaning performance under visible light, and polyvinyl alcohol (PVA) coated on the membrane surface could enhance hydrophilicity. Zhao et al. [23] prepare ad FeOCl-functionalized ceramic membrane to activate peroxyacetic acid to generate HO^• and ¹O₂ as the main species, reducing the extent of pore blocking and cake layer formation. Bai et al. [24] loaded layered double oxides (MnAl-LMO, CuAl-LMO, and CoAl-LMO) onto ceramic membranes to activate persulfate, resulting in the production of SO₄^•−, HO^•, and ¹O₂, which could reduce membrane fouling and improve water quality.

The objective of this research was to synthesize the CuFeS₂/MXene–PVDF membrane and investigate the antibiotic removal performance, its influences, and its mechanism using the catalytic membrane/PMS system. Due to the frequent detection in aquatic environments and potential risk to the ecosystem, moxifloxacin (MOX) was selected as the target pollutant in this research. MOX removal efficiency and the water flux of the membrane were investigated. Parameters of the catalytic membrane/PMS system were optimized, and the effects of operating conditions and the water matrix were studied. The mechanisms of MOX removal by the catalytic membrane/PMS system and the anti-fouling property of membranes were studied. Compared with other studies, a novel catalytic membrane was prepared to remove MOX through the synergistic activation of PMS and membrane rejection performance. This research provided the theoretical reference for antibiotics removal using PVDF membrane-coupled advanced oxidation technology.

2. Materials and Methods

2.1. Materials

Moxifloxacin (MOX, 99%), PMS (KHSO₅·0.5KHSO₄·0.5K₂SO₄), CuCl, phosphoric acid (H₃PO₄), acetonitrile (CH₃CN), methyl alcohol (MeOH) and humic acid (HA) were purchased from Aladdin, Shanghai, China. LiF, FeCl₃·6H₂O, polyethylene glycol (PEG), and glutaraldehyde (GA) were obtained from Macklin, Shanghai, China. HCl, NaOH, NaHCO₃, Na₂S₂O₃·5H₂O, (NH₂)₂CS, anhydrous ethanol, tertiary butyl alcohol (TBA), and L-histidine (L-his) were acquired from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). Ti₃AlC₂ (MAX) powder (400 mesh) was provided by HWRK Chemical, Beijing, China. H₃PO₄ and CH₃CN were of the HPLC grade, while other chemicals were analytical grade except for MOX. Ultrapure water was used throughout the experiments. The PVDF membrane (80 mm, 0.1 µm) was provided by Longjin Membrane Technology Co., Ltd. (Nantong, China). In total, 50 mg/L of the MOX stock solution was prepared by dissolving 50 mg of MOX in 1 L of ultrapure water. The MOX stock solution was diluted to 2 mg/L as the initial concentration in the experiments.

2.2. Preparation of CuFeS₂/MXene–PVDF Membrane

In this research, CuFeS₂/MXene was synthesized by the hydrothermal method (Note S1). CuFeS₂/MXene was ultrasonically dispersed in 50 mL of ultrapure water and deposited onto a PVDF membrane by vacuum filtration. Then, 50 mL of PEG (20 wt%) and GA (20 wt%) solutions were successively poured onto the membrane and pressurized with 0.05 MPa N₂. By changing the amount of CuFeS₂/MXene, the catalytic membrane with different catalyst loadings (2~4 mg/cm² CuFeS₂/MXene) could be prepared.

2.3. Membrane Characterization Method

The morphologies and microstructures of CuFeS₂, the PVDF membrane and the CuFeS₂/MXene–PVDF membrane were observed using a scanning electron microscope (SEM, FEG650, FEI, Mitchel Field, Long Island, NY, USA). The crystal structures of CuFeS₂, CuFeS₂/MXene, and CuFeS₂/MXene–PVDF were characterized with an X-ray diffractometer (XRD-6000, Shimadzu, Kyoto, Japan). Elemental composition was detected by X-ray photoelectron spectroscopy (XPS, Escalab 250Xi, Thermo Fisher, Waltham, WA, USA). The functional groups on the surface of membranes were determined by Fourier transform infrared spectrometer (FTIR, Spectrum Two, Perkin-Elmer, Waltham, WA, USA). The contact angles of membranes were measured using a contact angle meter (DCA20, Dataphysics, Filderstadt, Germany).

2.4. Experimental Procedure

For the continuous filtration experiment, the catalytic membrane was fixed on the ultrafiltration cup to filter 2 L of MOX under N₂ pressure, taking 1 mL sample from the filtrate and adding Na₂S₂O₃ immediately to terminate the reaction. The effects on the MOX removal by the catalytic membrane/PMS system were studied, such as CuFeS₂/MXene loading, PMS dosage, and filtration pressure. The effects of the water matrix on MOX removal were investigated by adjusting the initial pH with HCl or NaOH and adding NaHCO₃ and HA as impurities in water. For batch filtration, the solution should be in contact with the catalytic membrane for 10 min before filtration. Quenching experiments and electron paramagnetic resonance techniques (EPR, EMXplus-9.5/12, Bruker, Karlsruhe, Germany) were conducted to speculate on the main reactive species. After the addition of the quencher (MeOH, TBA, or L-his), the catalytic membrane (loading = 4 mg/cm², area = 38.5 cm²) activated 0.20 mM PMS to remove 250 mL of MOX under static conditions.

The catalytic membrane was completely submerged in water, stirring mechanically at different speeds for 1 h before being taken out and dried. The mass loss of modified materials could be evaluated by the weighing method. The membrane was used to filter 300 mL of HA (10 mg/L) at 0.02 MPa pressure, continuously weighing the mass of the filtered liquid. Based on filtration volume (V), filtration time (T), and water flux (J), equations of the relevant models were fitted and calculated, which could then be used to analyze the formation mechanism of membrane fouling.

2.5. Analysis Methods

The concentration of MOX was analyzed by high-performance liquid chromatography (HPLC, Agilent 1200 Series, Agilent Technologies, Santa Clara, CA, USA) with an XDB C-18 (5 μm, 4.6 × 150 mm) column and UV detector. The detection wavelength was 290 nm, the mobile phase was acetonitrile/water (25/75, V/V), and the flow rate was 1 mL/min.

2.6. Computational Methods

The density functional theory (DFT) was used to study the degradation pathway of MOX. The molecular formula of MOX was imported into GaussView 5.0 software (Gaussian Inc., Wallingford, CT, USA), and the DFT calculation was performed in Gaussian 09 software (Gaussian Inc., USA). The parameters were set corresponding to the B3LYP/6-31 + G (d, p) energy level theory, Natural Bond Orbitals, and aqueous phase model. The three-dimensional coordinates and Hirschfeld charges of each atom of MOX after geometric optimization were obtained. Then, the condensed Fukui function (CFF) was calculated according to Equations (1) and (2) [25]. Atom K with high values of f⁰ and f⁻ was susceptible to free radical attack and electrophilic attack, respectively.

$$\mathrm{Free}\mathrm{radical}\mathrm{attack}{f}_k^0=(q_{N-1}^k-q_{N+1}^k)/2$$

(1)

$$\mathrm{Electrophilic}\mathrm{attack}{f}_k^{-}=(q_{N-1}^k-q_N^k)$$

(2)

where q^k is the atom charge of atom K at the corresponding state; N, N − 1, and N + 1 indicate that the MOX molecule has a charge of 0, +1, and −1, respectively.

After the calculation file (.fch) of MOX was output from Gaussian 09, the Mayer analysis menu in Multiwfn software 3.8 (Tian Lu, China) was used to obtain the Mayer bonds of MOX. The Mayer bonds represented the stability of the chemical bond.

The water flux J (L m⁻² h⁻¹) of membranes could be calculated according to Equation (3) as follows:

$$J=\frac{V}{AT}$$

(3)

where V (L) is the filter volume; A is the effective area membrane (3.85 × 10⁻³ m²); and T (h) represents the filtration time.

The catalytic performance of the membrane was evaluated by calculating the removal efficiency of MOX (Equation (4)).

$$R=1-\frac{C_p}{C_0}$$

(4)

where C_p and C₀ represent the MOX concentrations of the filtrate and influent, respectively.

HA (10 mg/L TOC) was selected as the model foulant, and the membrane resistance R₁, reversible fouling resistance R₂, and irreversible fouling resistance R₃ of the membranes were obtained from Equations (5)–(7) as follows:

$$R_1=\frac{\mathsf{\Delta }P}{{\mathsf{\mu }J}_0}$$

(5)

$$R_2=\frac{\mathsf{\Delta }P}{{\mathsf{\mu }J}_1}-\frac{\mathsf{\Delta }P}{{\mathsf{\mu }J}_2}$$

(6)

$$R_3=\frac{\mathsf{\Delta }P}{{\mathsf{\mu }J}_2}-\frac{\mathsf{\Delta }P}{{\mathsf{\mu }J}_0}$$

(7)

where μ represents dynamic viscosity (8.9 × 10⁻⁴ Pa·s); ∆P represents filtration pressure; J₀ (before filtering HA), J₁ (after filtering HA) and J₂ (after hydraulic cleaning) represents the water flux J (L m⁻² h⁻¹) of membranes, respectively.

3. Results and Discussion

3.1. Membrane Characterization

SEM could observe the surface morphology of CuFeS₂/MXene, the PVDF membrane, and the catalytic membrane. As a type of MXene, Ti₃C₂T_x with a typical layered structure was prepared by etching Ti₃AlC₂ (Figure 1a). In Figure 1b, a large amount of granular CuFeS₂ was loaded onto the surface of MXene. As shown in Figure 1c, the PVDF membrane was a loose porous network structure with uneven pore size. The size of CuFeS₂/MXene particles was larger than PVDF membrane pores, indicating that CuFeS₂/MXene could accumulate on the surface of the PVDF membrane. In Figure 1d, the presence of CuFeS₂/MXene and the formation of pores were observed on the membrane surface.

The phase composition and structure were inspected by XRD, as shown in Figure 2a. According to the standard XRD data for CuFeS₂ (JCPDS 83-0983), the diffraction signals at 2θ = 29.4°, 49.0°, 57.9°, and 79.5° corresponded to the (1 1 2), (2 0 4), (3 1 2), and (3 1 6) planes of CuFeS₂. The strong diffraction signals and sharp peak shapes indicated that CuFeS₂ was crystallographically complete, and the arrangement of the atoms inside was relatively regular. Xu et al. [26] used XRD to characterize CuFeS₂ and observed similar characteristic peaks. The characteristic peaks of the composite at 18.2°, 39.0°, and 60.1° were correlated with the (0 0 4), (1 0 4), and (1 1 0) planes of MXene [19], indicating the successful synthesis of CuFeS₂/MXene. For CuFeS₂/MXene, the peak corresponding to the (0 0 2) plane was shifted to a lower angle compared to the standard XRD data for Ti₃AlC₂ (JCPDS 52-0875), suggesting that some particles were loaded onto the layered structure of MXene [20]. In addition, the low intensity of some peaks belonging to MXene may be due to the formation of CuFeS₂ on the MXene surface. XPS was applied to analyze the elemental composition of CuFeS₂, CuFeS₂/MXene, and the CuFeS₂/MXene–PVDF membrane, and the results are presented in Figure 2b–d. Cu, Fe, and S were obtained from CuFeS₂, and Ti was derived from MXene. The results of SEM, XRD, and XPS indicate the successful synthesis of the CuFeS₂/MXene–PVDF membrane.

The functional groups of membranes were analyzed by FTIR (Figure 3). The characteristic peak at 839 cm⁻¹, which corresponded to the β phase of the PVDF membrane, was clear [27]. The absorbance of 2943 and 1402 cm⁻¹ was due to the stretching vibration of -CH₂, and the peaks at 1170, 1073, and 878 cm⁻¹ were related to -CF₂. Li et al. [28] obtained similar FTIR spectra when characterizing PVDF membranes. The FTIR spectrum of the catalytic membrane showed a broad band in the range of 3200~3500 cm⁻¹ and a characteristic peak at 1730 cm⁻¹, indicating that -OH and -CHO were introduced on the membrane surface. Strong signals of CuFeS₂ in FTIR spectra [29] included 1147, 1120, 1040, 993 and 960 cm⁻¹. Yu et al. [30] observed the FTIR characteristic peak of CuFeS₂ at 570 cm⁻¹. The catalytic membrane prepared in this research also showed strong absorbance at the corresponding wave number, indicating the presence of CuFeS₂ on the surface of the membrane. According to the results of previous observations of MXene [31], 1630 and 655 cm⁻¹ in Figure 3 correspond to -COOH and Ti-O, respectively.

The hydrophilicity of the surfaces of membranes could be quantified through the contact angles, and the results are shown in Figure 4. The PVDF membrane exhibited a contact angle of 131.9°, and it was presumed that its hydrophobic quality was attributed to the repulsion of water molecules by the C-F bond. The catalytic membrane had super hydrophilicity, with a contact angle of 3.2°. PEG and GA were added as adhesives during the preparation of the catalytic membrane, so hydrophilic groups such as -OH and -CHO were introduced. In addition, the zeta potentials of PVDF and the catalytic membrane were −2.31 mV and −0.72 mV at pH = 5.75 (initial pH of the filtered solution), respectively. When the pH value was 5.22, the zeta potential of the catalytic membrane was zero. Zhao et al. measured the isoelectric point of the PVDF membrane at pH = 3.4 and concluded that the detected zeta potential values of COF-LZU1, which modified the membranes slightly, increased with the enhancement of surface hydrophilicity [32].

3.2. Parameter Optimization of the Catalytic Membrane/PMS System

Catalytic membranes could degrade antibiotics through advanced oxidation techniques. Before the filtration experiment, the bonding strength between modified materials and the PVDF membrane was investigated by shear water flow (Table S1). Only when using vacuum filtration did CuFeS₂/MXene easily fall off the PVDF membrane. However, after coating PEG and GA on the membrane surface, the loss of modified materials was significantly reduced. It was speculated that PEG and GA could effectively adhere CuFeS₂/MXene to the surface of the PVDF membrane. At 0.02 MPa filtration pressure, the MOX removal efficiencies by different systems were investigated in continuous filtration (Figure 5a). The removal efficiency of MOX by the CuFeS₂/MXene–PVDF membrane that activated the PMS system was 31.1%. Compared with other systems (removing 9.9%~13.4% MOX), it could be concluded that the catalytic membrane/PMS system promoted the removal of MOX through oxidative degradation. During the filtration process, CuFeS₂/MXene on the surface of the catalytic membrane activated PMS and generated reactive species.

Using CuFeS₂/MXene–PVDF membranes with different catalyst loadings to activate PMS, the removal efficiencies of MOX were 21.1%, 31.1%, and 48.2% (Figure 5b), respectively. The increase in loadings could promote MOX removal in the catalytic membrane/PMS system because more CuFeS₂/MXene indicated that there were more reaction sites in the PMS activation. At 0.02 MPa filtration pressure, the water flux of the PVDF membrane was 630.7 L m⁻² h⁻¹. As the CuFeS₂/MXene loading on the membrane surface increased from 1 to 4 mg/cm², the water flux decreased from 496.8 to 344.8 L m⁻² h⁻¹, promoting PMS activation by increasing the contact time between PMS and the catalytic membrane. Zheng et al. [2] used a catalytic membrane to activate PMS to remove pollutants and concluded that the k_obs values of RhB degradation increased with the increasing Cu-MOF-74 loadings. Figure 5c showed the removal efficiencies of MOX from different catalytic membrane/PMS systems, indicating that the catalytic performance of CuFeS₂/MXene was better than that of CuFeS₂. It was presumed that CuFeS₂ and MXene played a synergistic role in activating PMS. The addition of PMS (0.20, 0.16, and 0.12 mM) resulted in MOX removal efficiencies of 68.2%, 54.1%, and 48.2% (Figure 5d), respectively. It showed that the increase in the PMS concentration was advantageous for MOX removal. With the progress of filtration, the reaction sites on the catalytic membrane surface may be covered by pollutants [33], which could hinder the reaction between CuFeS₂/MXene and PMS. After selecting an appropriate loading (4 mg/cm²) and PMS dosage (0.20 mM), the following experiments were carried out. In this research, the removal efficiency of 2 L of MOX by the catalytic membrane/PMS system was 68.2%, which was higher than that of the PVDF membrane (9.9%). The slight removal of antibiotics by the PVDF membrane was attributed to solute adsorption into the membrane’s porous structure [34]. Wang et al. [35] activated PMS with the Ti catalytic membrane and TiO₂/Ti catalytic membrane, and the removal efficiency of ciprofloxacin was 6.9% and 7.2%, respectively. It indicated that CuFeS₂/MXene activated PMS effectively and promoted MOX removal through oxidative degradation.

3.3. Removal Efficiency under Varying Conditions

The effects of the operating conditions and water matrix on MOX removal by the catalytic membrane/PMS system were studied. Figure 6a shows the MOX removal at different filtration pressures. When the filtration pressure increased from 0.02 MPa to 0.08 MPa, the removal efficiency of MOX decreased from 68.2% to 18.5%. Low filtration pressure was conducive to the MOX removal by the catalytic membrane/PMS system. At 0.08 MPa, it was speculated that MOX was mainly removed by adsorption because the catalytic membrane could not effectively activate PMS within a short contact time. Under the different filtration pressures (0.08, 0.05, and 0.02 MPa), the water flux of the catalytic membrane was 1150.3, 548.3, and 344.8 L m⁻² h⁻¹, respectively. Filtration pressure affected the oxidative degradation effect by affecting the contact time between PMS and the catalytic membrane. Yue et al. [8] obtained a similar view that the degradation efficiency of antibiotics gradually decreased with the increase in pressure due to the short residual time of PMS on the catalytic membrane. In Figure 6b, the MOX removal efficiency of batch filtration increased by 19.6% compared with the original filtration method. It illustrated that the batch filtration was beneficial to the activation and the oxidative degradation of MOX.

By changing the initial pH value in the solution, the effect of a strong acid or alkali on MOX removal efficiency was investigated (Figure 7a). The initial pH of 2 mg/L of the MOX solution was 5.75, which was close to the theoretical value calculated based on the dissociation coefficient [36]. Filtering 1.5 L of the solution when the initial pH values were 3.10, 5.75, and 10.89, the removal efficiency of MOX was 83.4%, 82.5%, and 74.9%, respectively. This shows that the catalytic membrane/PMS system is more suitable for removing MOX under acidic conditions. pH could affect PMS activation by changing the fractions of different PMS species [37] and the surface charge of the catalyst. When the initial pH was 3.10, PMS almost existed in the form of HSO₅⁻, and the surface of CuFeS₂/MXene was positively charged. It promoted the activation of PMS through the electrostatic interaction between CuFeS₂/MXene and PMS. When the initial pH was 10.89, the main component of PMS was HSO₅²⁻, which repelled CuFeS₂/MXene with a negative surface charge. In addition, SO₄^•− was easy to use to generate HO^• under alkaline conditions [38] (Equation (8)), and the oxidation performance and action time of SO₄^•− were greater than that of HO^• [39].

SO₄^•− + OH⁻ → HO^• + SO₄²⁻

(8)

Impurities in the water could also interfere with the degradation of MOX due to scavenging and adsorption competition. The effects of NOM and anions on the catalytic membrane/PMS system were investigated by selecting HA and HCO₃⁻, respectively, as shown in Figure 7b. When HA and HCO₃⁻ were used as impurities in water, the removal efficiencies of MOX were 80.7% and 77.3%, respectively. HCO₃⁻ may react with SO₄^•− and HO^• to form active substances with a lower oxidation capacity (Equations (9) and (10)) [15,40], which weakenes the degradation of MOX. Free radicals were more likely to attach to HA than pollutants and then react with functional groups in HA [41]. HA preempted the active sites on the catalyst surface through strong π-π stacking [42], so the interaction between CuFeS₂/MXene and PMS could be blocked.

SO₄^•− + HCO₃⁻ → HCO₃^• + SO₄²⁻

(9)

HO^• + HCO₃⁻ → CO₃^•− + H₂O

(10)

3.4. Mechanism of MOX Removal by Catalytic Membrane

When the solution was filtered by a catalytic membrane, MOX adsorption and PMS activation could occur. According to the FTIR spectral results, the catalytic membrane surface contained functional groups, such as -OH, -CHO, and -COOH, which could adsorb MOX. Zou et al. [43] concluded that the oxygen-containing functional groups in adsorbents were an important mechanism for adsorbing MOX due to the stronger H-bonding effect. CuFeS₂/MXene loaded on the membrane surface could activate PMS to produce reactive species, and the activation mechanism was studied. Cu⁺ and Fe²⁺ on the CuFeS₂ surface reacted with PMS to generate SO₄^•−, which was accompanied by the increase in the valence states of metals (Cu²⁺ and Fe³⁺). Then, SO₄^•− and HSO₅⁻ could further generate HO^• and ¹O₂ through their reaction with H₂O [44]. The low valence sulfur [13], Cu⁺ [45], and MXene [46] contained in CuFeS₂/MXene could promote the regeneration of Cu²⁺/Fe³⁺ into Cu⁺/Fe²⁺, respectively.

MeOH was used to eliminate HO^• and SO₄^•−; TBA was chosen as a quencher of HO^•, and L-His could quench ¹O₂. Using the catalytic membrane/PMS system to remove MOX under static conditions, Figure 8a shows the effects of quenchers on MOX removal. With the addition of TBA, L-his, or MeOH, the removal efficiency of MOX was 70.1%, 68.2% and 26.6%, respectively. It can be inferred that SO₄^•−, HO^•, and ¹O₂ are beneficial for MOX removal. Using DMPO and TEMP as the capture of radicals, the characteristic signal peaks of DMPO-HO^•, DMPO-SO₄^•−, and TEMP-¹O₂ are observed in the EPR spectrum (Figure 8b). The quenching experiments and EPR results show that SO₄^•−, HO^•, and ¹O₂ play an important role in MOX degradation.

CFF was used to illustrate the regioselectivity of radicals attacking (Table S2), where f⁰ and f⁻ corresponded to radical attack and electrophilic attack, respectively. The values of f⁰ and f⁻ of the reactive sites were usually larger than other regions, such as the atoms (Num. 1, 2, 5, 13, 19, 21, and 29). In Figure 9a, different colors showed the numerical ranges of the Mayer bonds of MOX. Mayer bonds of N1-C14 and C22-O27 were 1.26 and 1.60, respectively, so it was difficult to remove N and O. The Mayer bonds of N1-C2, N1-C5, C10-C21, N19-C24, and C13-F29 were 0.62, 0.49, 0.78, 0.91, and 0.93, respectively, which were easily broken under the attack of active species. Therefore, the degradation of MOX may involve the cleavage of the nitrogen-containing heterocycle (NCH) (C1-C2 and C1-C5), the removal of the cyclopropyl group (C24-N19), decarboxylation (C10-C21) and defluorination (C13-F29). Similar MOX degradation pathways have been reported in other studies, such as g-C₃N₄/photocatalysis [47] and FeMnCoO_x/PMS [48]. The mechanism of MOX removal by the catalytic membrane/PMS system included adsorption and degradation, as shown in Figure 9b. CuFeS₂/MXene on the membrane surface was in contact with PMS, generating reactive species (SO₄^•−, HO^• and ¹O₂) for the degradation of MOX on the membrane surface and in the solution. On the one hand, the PVDF membrane served as an effective carrier for CuFeS₂/MXene, alleviating the problems of catalyst loss and difficulties in recycling. On the other hand, CuFeS₂/MXene could enhance MOX removal by the PVDF membrane through advanced oxidation.

3.5. Anti-Fouling Property

HA could produce membrane fouling through adhesion and blockage, which was selected as a common foulant to assess the anti-fouling performance of membranes. The contamination of the membrane surface was reversible and could be removed by hydraulic cleaning, while irreversible membrane fouling caused by pore blockage and adsorption required chemical reagents to restore water flux. Based on the water flux of membranes and Equations (5)–(7), the resistances in different membrane treatment systems were calculated, as shown in Figure 10. For the PVDF membrane, the oxidation of PMS itself could reduce the total resistance and reversible fouling resistance by 7.5% and 21.0%, respectively. However, PMS also oxidized HA into smaller pollutants, which was easier to enter the PVDF membrane’s pores, and irreversible fouling resistance increased by 13.2%. For the catalytic membrane, HA caused accumulation on the membrane surface, and reversible fouling resistance was even more serious. However, due to the small pore size of the catalytic membrane, HA could not easily enter the membrane pore, so irreversible fouling resistance was lower than that of the PVDF membrane. The calculated total resistance and reversible fouling resistance of the catalytic membrane/PMS system were 69.9% and 22.8% of the PVDF membrane, respectively. This indicates that membrane fouling could be effectively alleviated by the catalytic membrane/PMS system. The catalytic membrane could activate PMS efficiently, reducing the degree of membrane fouling. On the one hand, in situ, PMS activation occurred when the feed water passed through the catalytic layer of the membrane, which could enhance the repulsive interactions between pollutants and membranes according to the interface free energy calculations [24]. On the other hand, the active species produced by activating PMS could promote the degradation of pollutants with high molecular weight into smaller fractions and oxidize the hydrophobic matter into hydrophilic matter [7,49]. Although the loading of CuFeS₂/MXene increased the inherent resistance of the catalytic membrane, the total resistance and reversible fouling resistance were reduced through PMS activation.

Hermia modeled and analyzed the filtering process (Equation (11)), and the different values of n corresponded to four single models [50]. The PVDF membrane and catalytic membrane were used to filter HA in order to study the formation mechanism of membrane fouling (Figure 11 and Table S3). The Cake filtration model (R² = 0.9944) might explain the formation of PVDF membrane fouling compared with other models (R² = 0.9276~0.9710). Thus, it could be assumed that HA formed a cake layer on the surface of the PVDF membrane.

$${\mathrm{d}}^2\mathrm{t}/\mathrm{d}{\mathrm{V}}^2=\mathrm{k}({\mathrm{dt}/\mathrm{dV})}^{\mathrm{n}}$$

(11)

For the catalytic membrane/PMS system, the fitting effects of the cake filtration model (R² = 0.9885) and intermediate blocking model (R² = 0.9755) were better than the other models. The fitting effects of single models were not satisfactory, but the intermediate blocking cake filtration model (Equation (12)) [51] resulted in a better fitting effect (R² = 0.9991). It was shown how the cake layer tended to transform into intermediate blocking, and Simon et al. [52] also found a similar conclusion using caustic cleaning to alleviate membrane fouling. It is speculated that the catalytic membrane could activate PMS to produce reactive species, and HA can be oxidized into smaller pollutants to enter the pores. The catalytic membrane/PMS system alleviated membrane fouling by oxidizing HA and weakening the formation of the cake layer, as follows:

$$\frac{1}{K_i}\ln (1+\frac{K_i(\sqrt{1+{{2K}_{c}J}_0^{2}t}-1)}{K_c{J}_0})=y$$

(12)

where J₀ is the initial water flux; y is the filtration volume of the membrane per unit area; and K_i (m⁻¹) and K_c (s/m²) are the constants of the intermediate blocking model and cake filtration model, respectively.

4. Conclusions

The CuFeS₂/MXene–PVDF membrane was prepared to active PMS and remove MOX, and it was characterized by the measurement of SEM, FTIR, and contact angles. CuFeS₂/MXene accumulated on the surface of the PVDF membrane, leading to the formation of new pores. The coating of PEG and GA enhanced the bonding strength between CuFeS₂/MXene and the PVDF membrane surface and also introduced hydrophilic functional groups, such as -OH and -CHO. The contact angles of the PVDF membrane and catalytic membrane were 131.9° and 3.2°, respectively, indicating superhydrophilicity for the latter. Optimal parameters for the catalytic membrane/PMS system included loading 4 mg/cm² of CuFeS₂/MXene onto the PVDF membrane and adding 0.20 mM of PMS. In the continuous filtration of a 2 L solution, the catalytic membrane/PMS system exhibited a 58.3% higher removal efficiency of MOX compared to the PVDF membrane alone. Low filtration pressure and batch filtration could promote PMS activation by enhancing the contact reaction between the catalytic membrane and solution. However, OH⁻, HCO₃⁻, and HA could inhibit MOX removal by affecting the generation and transformation of free radicals. Besides adsorption, the catalytic membrane also activated PMS to generate SO₄^•−, HO^•, and ¹O₂, which could degrade MOX through the removal of the NCH and cyclopropyl group, decarboxylation, and defluorination. Based on the fouling resistance calculation and the intermediate blocking cake filtration model (R² = 0.9991), the catalytic membrane/PMS system could reduce reversible membrane fouling and weaken the formation of the cake layer. These findings indicate the potential application of catalytic membrane-activated persulfate technology in the field of water treatment.