Efficient removal of perfluorinated compounds with the polyamide nanofiltration membrane and membrane fouling resistance analysis

Perfluorinated compounds (PFCs), as synthetic substances, have been widely applied in various industries such as chemicals, textiles, leather, and fire extinguishing foams, due to their unique properties of high chemical stability, surface activity, and thermal resistance (Banks et al. 2020; Zhang et al. 2021; Jl et al. 2022). However, in recent years, PFCs have raised increasing concerns regarding their impact on the global environment and human health, primarily ascribed to their toxicity, adverse effects, and resistance to biodegradation in the environment (Pasecnaja et al. 2022). As a class of persistent organic pollutants, the presence of aqueous solutions containing hazardous PFCs in our main water sources has become a matter of great concern, as it can potentially lead to significant adverse effects as well as pose risks of diseases and disorders to both abiotic and biotic ecosystems (Zhang et al. 2013; Jz et al. 2021; Pang et al. 2022). Among the various perfluorinated organic pollutants, perfluorooctanoic acid (PFOA) and perfluorooctane sulfonate (PFOS) are two of the most prevalent and representative compounds. Due to their strong polarity and thermal stability, they are highly resistant to degradation by natural factors such as ultraviolet irradiation, chemical reactions, and microbial metabolism (Boo et al. 2018; Yang et al. 2020; Yuan et al. 2020; Tian et al. 2021). Consequently, there is an urgent need to develop efficient and reliable techniques for the removal of PFOA and PFOS from aquatic environments.

To date, various technologies have been employed for the removal of PFCs from water, including advanced oxidation, photocatalysis, adsorption, and membrane separation (Tang et al. 2007; Meragawi et al. 2020; Ding et al. 2023). Among these technologies, nanofiltration (NF) membrane technology stands out as a promising option for the efficient removal of PFCs due to its high selectivity and reliability (Steinle-Darling & Reinhard 2008; Park et al. 2017). Commercial polymeric NF membranes have shown the ability to remove nearly 90% of PFOA or PFOS. However, the trade-off between permeability and selectivity remains a significant challenge for the widespread application of NF technology (Meng et al. 2021; Kwon et al. 2022).

Two-dimensional transition metal carbides, nitrides, or carbonitrides (MXene), as a novel type of two-dimensional material, possess abundant surface hydrophilic functional groups, such as hydroxyl and epoxy groups, which bring it excellent hydrophilicity. When used as the interlayer during the preparation of polyamide (PA) NF membranes, these properties contribute to reducing pore blockage in membranes and promoting the uniform distribution of monomers in the aqueous phase during interfacial polymerization, thereby preventing penetration into the membrane pores and resulting in a thinner PA layer (Naguib et al. 2011; Jun et al. 2019; Peng et al. 2019; Wu et al. 2020; Xu et al. 2020). Considering the rough surface and hydrophilicity of in situ produced MXene-TiO₂ via PMS oxidation, it is desirable to utilize MXene-TiO₂ as an intermediate layer to enhance the performance of the PA layer, achieving a smaller thickness, larger specific surface area, higher flux, and excellent salt cutoff rate. The tunable interlayer channels hold the potential to further improve the trade-off between permeability and selectivity for PFCs removal (Li et al. 2017a; Ding et al. 2020; Gao et al. 2020a, 2020b; Long et al. 2021).

Membrane fouling is a process in which pollutants accumulate within the membrane pores or on the membrane surface during the separation process, primarily due to the interaction between the pollutants and the membrane (Liu et al. 2023). This phenomenon often leads to pore blockage and the formation of a cake layer, resulting in increased resistance to membrane permeability (Chang et al. 2002; Pei et al. 2006; Wang et al. 2009). In natural water environments, the composition of pollutants is complex, involving the coexistence of organic substances and ions, which poses a significant challenge to the stable purification of NF membranes (Liu et al. 2022; Ma et al. 2022). During the process of PFCs removal, membrane fouling is also observed (Saravia et al. 2006). Therefore, when employing MXT-NFM for PFCs removal, it is crucial to consider the occurrence of membrane fouling in the presence of PFCs and other substances. Humic acid (HA), as one of the natural organic matter (NOM) components in water, is a major contributor to membrane fouling and can affect the removal of PFCs due to enhancement of scaling effect by its hydrophilic functional groups (Guan et al. 2018; Li et al. 2020). Additionally, cations can also influence membrane fouling. For example, calcium ion (Ca²⁺), a common cation, can interact with other pollutants, thereby affecting the structure and deposition rate of the filter cake layer and exacerbating membrane fouling (Sutzkover-Gutman et al. 2010; Fei & Shih 2011; Miao et al. 2017). Therefore, in this study, HA, as a representative of common organic compounds, and Ca²⁺, as typical ions, will be investigated to explore their role in PFCs removal and to verify whether MXT-NFM exhibits excellent membrane fouling resistance in complex systems (Wang et al. 2020). Furthermore, the membrane fouling mechanism of PFCs in coexistence systems has not been clearly defined in current research. Therefore, the extended Derjaguin-Landau-Verwey-Overbeek (XDLVO) theory, a cutting-edge analytical method, will be employed to evaluate the development trend and dominant factors of membrane fouling. The objective is to gain a scientific understanding of the mechanism underlying membrane fouling (Ojaniemi et al. 2012; Gao et al. 2019).

In this study, we fabricated an NF membrane, MXT-NFM, with MXene-TiO₂ as the intermediate layer structure for the purification of PFCs in aqueous solutions. The presence of the MXene-TiO₂ intermediate layer alleviated membrane pore plugging and resulted in the formation of a thinner and more uniform PA layer. These structural modifications contributed to enhanced water transport and selectivity (Hitsov et al. 2015; Gao et al. 2020a, 2020b; Wang et al. 2023). Experimental results demonstrate that MXT-NFM exhibited high removal efficiency and flux for PFCs. Moreover, the presence of Ca²⁺ and HA enhances the rejection of PFCs, and a higher concentration of Ca²⁺ reduced membrane fouling. The XDLVO theory was employed to evaluate the interaction between pollutants and the membrane, revealing that the particle size of pollutant colloids and the acid–base interaction play crucial roles during the membrane fouling development process. These experimental findings confirmed the excellent performance of MXT-NFM in removing PFCs and mitigating the adverse effects associated with membrane fouling.

Ti₃AlC₂ powder, lithium fluoride (LiF, >99.0%), sodium chloride (NaCl, >99.0%), sodium sulfate (Na₂SO₄, >99.0%), hydrochloric acid (HCl, 32%), anhydrous piperazine (PIP, >99.0%), trimesoyl chloride (TMC, >99.0%), perfluorooctanoic acid (PFOA, >99.0%), perfluorooctane sulfonate (PFOS, >99.0%), humic acid (HA, >99.0%), potassium monopersulfate (PMS, >99.0%), and N-Hexane (C₆H₁₄, >99.0%) were purchased from Shanghai Aladdin Biochemical Technology Co, Ltd. A polyethersulfone (PES) microfiltration membrane with a nominal pore size of 0.22 μm was used as the substrate. All reagents used in this study are of analytical grade and used without further purification. Milli-Q deionized water with a resistivity of 18.1 MΩ cm at 25 °C was used for the preparation of all solutions.

Ti₃C₂T_x nanosheets were synthesized in situ using a mild HF method. In a typical procedure, 1 g of LiF was added to 15 mL of HCl (6 M), and the mixture was continuously stirred for 5 min. Then, 1 g of Ti₃AlC₂ powder was gradually added to the etchant and the reaction was carried out at 35 °C for 24 h. Subsequently, the resulting mixture was washed by deionized water with centrifugation for 5 cycles to achieve neutral pH. The black Ti₃C₂T_x slurry was collected and redispersed in 100 mL of deionized water. The mixture was then sonicated under flowing nitrogen for 1 h to achieve delamination and followed by centrifugation at 3,500 rpm for 30 min.

For the preparation of MXene-TiO₂, a PMS solution was added to the MXene solution in a mass ratio of 25:1. The mixture was then mixed with a small amount of deionized water and transferred to a centrifuge tube. Ultrasonication was carried out for 5 min until the solution became clear and colorless, resulting in the desired MXene-TiO₂ solution. To prevent any interference from oxidation, the solution should be used immediately after preparation.

To prepare the PIP-based PA layer, a PIP aqueous solution with a mass fraction of 0.75 wt% was first prepared. Subsequently, a TMC oil phase solution with a mass fraction of 0.038 wt% was prepared by dissolving TMC into N-hexane.

The PES substrate, which had been pre-soaked in deionized water, was then naturally dried and placed into a self-made mold. The aqueous solution was poured into the mold and kept for 5 min. Excess solution was then poured out, and any residual solution on the membrane surface was wiped clean.

Next, the oil phase solution was poured onto the membrane and left to react for 2 min. Afterward, the excess solution was removed, and the membrane was taken out and soaked in deionized water for 2 min to remove any residual reagent. The pure PA NF membrane was obtained by placing the cleaned base membrane into an 80 °C vacuum oven for 3 min and denoted as pure membrane. The membrane was then stored in deionized water for further use.

For the preparation of MXT-NFM, 0.5 mL of the MXene-TiO₂ solution was sprayed onto the base membrane before pouring the aqueous phase. The membrane was allowed to stand for 10 min. Subsequently, it was soaked in deionized water for 30 min and followed by drying in air. The obtained membrane was used as the substrate for the preparation of MXT-NFM, and the remaining steps followed the same process as the preparation of the pure membrane.

X-ray diffraction (XRD) analysis was conducted over the 2θrange of 5-70° with the scanning rate of 0.2 s step⁻¹. Morphologies of MXene nanosheets and MXene-TiO₂ were characterized by high-resolution transmission electron microscopy (HRTEM; FEI Tecnai G2 F20, Japan). The micrographs of membranes were visualized by the field emission scanning electron microscope (FESEM; Hitachi S4800, Japan). Fourier transform infrared spectroscopy (FTIR; TENSOR 27, Bruker, Germany) was applied to investigate the near-surface functional groups of samples within a scanning range of 100–4,000 cm⁻¹. Atomic force microscopy (AFM; NSG10, NT-MDT, ca. 330 kHz) was further employed to assess the membrane surface roughness. Water contact angles were tested on a contact angle measurement system (CAM200) to evaluate the hydrophilcity of membrane surface. The Anton Paar SurPASS electrokinetic analyzer was used to detect the zeta potential of the as-prepared NF membranes by using KCl aqueous solution (1 mM, pH = 3.0–10.0).

To assess the membrane purification performance toward PFCs, PFOA solution (100 μg/L) and PFOS solution (100 μg/L) were used as the feed solutions. The process described above was repeated, and the purification performance of the membrane was evaluated using Equations (1) and (2). The discharged solutions were collected in injection bottles for precise composition analysis.

SEM images of the surface and cross-section of the pure membrane and MXT-NFM in Figure 2(d) and 2(g) demonstrate that the addition of the intermediate layer contributed to the formation of a wrinkled structure in the PA layer, compared to the relatively smooth surface of the pure membrane. The wrinkled structure is primarily attributed to two factors: Firstly, the intermediate layer reduced the pore size of the substrate, hindering water penetration into the pores, and promoting interfacial polymerization on the surface of the PES substrate membrane. Secondly, the hydrophilic functional groups on the intermediate layer reduced the diffusion rate of the PIP monomer in the aqueous phase, resulting in a lower reaction rate with TMC and the formation of the wrinkled structure. Moreover, the addition of the intermediate layer significantly reduced the thickness of the PA layer due to the enhanced interaction between the amine monomers and the hydrophilic MXene-TiO₂, which impeded the diffusion of PIP monomer and its reaction with the organic phase (Figure 2(e) and 2(f)).

AFM images in Figure 2(f) and 2(i) reveal the surface roughness of the NF membranes. The MXT-NFM membrane exhibited a higher surface roughness with an Ra value of 18.31 nm, approximately four times greater than that of the pure membrane. The rougher surface can potentially enhance the permeable area of the PA layer, facilitating more efficient water molecule transport and resulting in water flux improvement.

Figure 3(b) displays the water contact angles of the pure membrane and MXT-NFM, measuring 44.9°and 22.4°, respectively. Test results indicate that the incorporation of MXene-TiO₂ effectively enhances the hydrophilicity of the NF membrane. This improvement can be attributed to two possible factors: (1) the presence of hydrophilic groups on the surface of MXene-TiO₂ and (2) the intrinsic structure of MXene-TiO₂ provided additional water pathways within the NF membrane, reducing steric hindrance and facilitating water molecule migration and diffusion.

The charged properties of the membrane surface have a significant impact on membrane retention performance. Zeta potential analysis was employed in this study to investigate the difference in surface charge between the pure membrane and MXT-NFM. As illustrated in Figure 3(c), the pure membrane exhibited a negative charge until the pH exceeded 3.2, whereas MXT-NFM displayed a negative charge until the pH exceeded 3.8. The enhanced negative surface charge of MXT-NFM would promote anion repulsion during the process of filtration.

Permeance and rejection toward PFCs of the NF membranes were further investigated. Figure 3(d) demonstrates that the permeance of MXT-NFM was significantly increased, reaching approximately three times that of the pure membrane. This improvement can be attributed to the enhanced hydrophilicity and surface roughness of MXT-NFM. Simultaneously, the rejection of MXT-NFM for PFOA and PFOS was 84.76 and 99.81%, respectively, slightly higher than that of the pure membrane. The higher rejection can be attributed to the more negatively charged surface of MXT-NFM, which strengthened the electrostatic repulsion. Notably, the rejection of PFOS was higher than that of PFOA, primarily due to the larger size of PFOS molecules, which increases steric hindrance while passing through the NF membrane.

The surface charge of the NF membrane plays a crucial role in the removal of charged compounds. It is widely recognized that the membrane charge is closely associated with the pH value of the water sample. Additionally, the pH value of the water sample also affects the chemical forms of organic compounds in water, thereby influencing the interaction between the NF membrane and organic compounds. Accordingly, it is important to investigate the impact of initial pH on the removal of PFCs by the MXT-NFM membrane. As shown in Figure 4(b), the rejections of PFOS and PFOA by the MXT-NFM membrane increased from 94.32 to 99.73% and 82.79 to 95.48%, respectively, as the pH value of the water sample increased from 3 to 10. This indicates that the pH value has a significant influence on the removal efficiency of PFOS and PFOA by the MXT-NFM membrane, primarily due to the electrostatic effect. Based on the trend observed in the zeta potential diagram (Figure 3(c)), when the pH value of the MXT-NFM membrane was lower than the isoelectric point, H⁺ ions in the water sample neutralized the negative charge on the membrane surface, resulting in a positively charged membrane surface. Conversely, when the pH value was higher than the isoelectric point, the concentration of OH⁻ ions in the water sample increased, enriching the membrane surface with negative charges. The interaction between the MXT-NFM membrane and PFOS/PFOA is predominantly governed by electrostatic absorption. Under low pH conditions, the enhanced positive charge on the MXT-NFM membrane surface promoted electrostatic absorption, leading to ineffective interception of PFOS and PFOA. Similarly, with the increment of pH value, the charge density of anions increased, contributing to enhanced electrostatic impedance and rejections of these two types of PFCs. In summary, the alkaline environment facilitated the improved removal performance of MXT-NFM for PFOA and PFOS in practical applications.

To evaluate the influence of temperature removal, the initial concentrations of PFOS/PFOA were kept at 100 μg/L, and the water temperature was adjusted within the range of 5–35 °C. Figure 4(c) demonstrates that the change in temperature exhibited relatively negligible effect on the removal of PFCs. Although the diffusion degree of PFCs and water molecules increased with temperature, the impact on the physicochemical properties remained limited.

Under the condition of 100 μg/L PFOS/PFOA concentrations, the effect of operation pressure on the removal of PFCs by the MXT-NFM was investigated, as shown in Figure 4(d). The changes in the rejections of both PFOS and PFOA were negligible as the pressure increased from 3 to 6 bar. This result indicates that the intermediate layer could act as a robust membrane skeleton, providing excellent rigid support and effectively preventing excessive deformation to the MXT-NFM structure.

In natural water samples, there is a coexistence of organic compounds and inorganic salt ions. The complexation reactions between those compositions and the occurrence of membrane fouling can impact the removal efficiency of NF membranes for target pollutants. Therefore, it is essential to consider the influence of the coexistence of inorganic salt ions and organic substances when studying the removal efficiency of MXT-NFM for PFCs. In this study, HA and Ca²⁺ were selected as representative substances to investigate the effect of membrane fouling on the removal of PFCs under coexistence conditions.

To investigate the impact of organic compound coexistence on membrane fouling, a coexistence condition of 5 mg/L HA and 100 μg/L PFOS/PFOA was employed. The water flux variation is depicted in Figure 5(c). When HA and PFCs coexisted, the flux of MXT-NFM experienced a significant decrease and eventually reached a stable state after a certain period of time. With 2 h filtration process, the flux values for the PFOS/HA and PFOA/HA systems were recorded as 0.928 and 0.945, respectively. In this coexistence scenario, the water flux decay rate of MXT-NFM was consistently higher than that observed for PFCs alone. Several factors might contribute to this phenomenon. Firstly, the central portion of HA molecule consists of hydrophobic aromatic and fatty groups. Driven by hydrophobic interactions and pressure, HA was prone to adsorb onto the membrane surface, forming a dense organic fouling layer. This layer increased the mass transfer resistance and consequently led to a rapid decline in membrane flux. Secondly, the negative charge density of pollutants in the coexistence of HA and PFOS was lower than that in the coexistence of HA and PFOA, resulting in stronger electrostatic repulsion between the latter and MXT-NFM. This stronger repulsion led to a lower membrane flux decay rate in the presence of HA and PFOA. After hydraulic cleaning, the water flux of MXT-NFM was restored to a high level, accompanied by lower irreversible fouling rates (R_ir), as shown in Figure 5(d). These test results indicate that the overall anti-fouling performance of MXT-NFM remained at an ideal level, despite the significant initial impact of HA on permeance (Li et al. 2017b; Gao et al. 2018).

To investigate the impact of coexistence of inorganic salts and organic compounds on water flux attenuation, CaCl₂ was selected to simulate the inorganic salt environment in water. The experimental conditions were set as follows: the initial concentrations of PFOS/PFOA, HA, and Ca²⁺ were 100 μg/L, 5 mg/L, and 2 mM. The operation pressure was 4 bar.

As shown in Figure 5(e), the specific flux of MXT-NFM in the PFCs/HA/Ca²⁺ system exhibited a significant decrease compared with that of the HA/PFCs coexistence system. After 2 h of filtration, the specific flux values were measured as 0.906 for the PFOS/HA/Ca²⁺ coexistence system and 0.917 for the PFOA/HA/Ca²⁺ coexistence system. The reduction was because the addition of Ca²⁺ weakened the negatively charged properties of HA, PFCs, and membrane surface. Consequently, the electrostatic repulsion between MXT-NFM and HA/PFCs decreased, leading to greater aggregation of the two pollutants on the membrane surface. Additionally, the electrostatic repulsive force between the pollutants decreased, resulting in a denser fouling layer on the MXT-NFM surface and a drastic decrease in membrane flux.

As the deposition continued to increase, the adsorption and desorption of pollutants reached an equilibrium state, while the water flux of MXT-NFM approached a stable level. After hydraulic cleaning, the flux recovery remained satisfactory, demonstrating the high anti-fouling performance of MXT-NFM under the coexistence of HA, Ca²⁺, and PFCs. This can be attributed to the excellent hydrophilicity of MXT-NFM, which facilitated a tight combination between the membrane surface and water molecules, forming a compact hydration layer and reducing the interaction between pollutants and membrane surface (Shao et al. 2011; Chang et al. 2012).

In general, the membrane fouling process can be divided into two stages: the early stage which is primarily controlled by the interaction between free pollutants and the clean membrane surface, and the later stage which is mainly controlled by the interaction between free pollutants and the pollutants deposited on the membrane surface. To understand more details about the fouling mechanism, the XDLVO theory was applied to evaluate three energy components governing the interaction: van der Waals interaction energy (LW), electrostatic double-layer interaction energy (EL), and acid–base interaction energy (AB). By analyzing these interaction relationships between contaminants and membrane, as well as between contaminants themselves, we can gain insights into the fouling mechanism.

Supplementary Table S1 presents the zeta potential, colloid particle size, and contact angles of MXT-NFM and six different pollutant colloids, along with three probe liquids. When PFOS or PFOA coexisted with HA, the contact angle of the colloid decreased, indicating increased hydrophilicity. At the same time, the increased negative zeta potential can be attributed to the presence of carboxyl and hydroxyl groups in HA molecules, which enhanced the hydrophilicity and negative zeta potential of PFOS and PFOA. On the other hand, when Ca²⁺ was added, the contact angle of the colloid and its hydrophobicity increased, while the negative zeta potential decreased. This can be explained by the reaction of Ca²⁺ with PFCs and HA, weakening the negative charge on the carboxyl and hydroxyl groups in HA which resulted in a reduction in the overall hydrophilicity and negative zeta potential of the mixed pollutants. Furthermore, the coexistence of these components resulted in the formation of complex organic compounds with larger particle sizes.

Supplementary Table S2 presents the surface tension parameter and condensation free energy of MXT-NFM under different colloid environments. The surface tension data indicates that both the membrane and contaminants exhibited high electron donor composition (γ⁻) and relatively low electron receptor components (γ⁺). This implies that both the membrane and colloid possessed a strong electron-donating capacity and negative charged surface. The condensation free energy (ΔG_sws) refers to the free energy per unit area when two surfaces of the same material are immersed in a solvent (water) and come into contact with each other. ΔG_sws serves as a quantitative index for evaluating the hydrophilicity of membrane and pollutant. The positive value of ΔG_sws indicates a hydrophilic surface, while the negative value indicates a hydrophobic surface. Furthermore, a higher absolute value signifies a stronger hydrophilicity or hydrophobicity (Li et al. 2019).

Based on the data in Supplementary Table S2, the PFOS/HA and PFOA/HA systems exhibited greater hydrophilicity compared to the single PFOS and PFOA systems. However, the hydrophilicity of PFOS/HA/Ca²⁺ and PFOA/HA/Ca²⁺ systems were lower than that of PFOS/HA and PFOA/HA systems. It is important to note that the condensation free energy alone may not be suitable for quantitatively evaluating the interaction energy between the membrane and the pollutant colloids. Therefore, the interaction energy should be further evaluated to understand the actual relationship between the membrane and the pollutant colloids.

According to the XDLVO theory, interaction between the membrane and the pollutant is repulsive if the interaction energy between them is positive, indicating that the membrane repels the colloidal pollutants. Conversely, if the interaction energy is negative, it signifies that the membrane is attractive to the colloidal pollutants. Additionally, a higher positive (or negative) value indicates a smaller (or larger) degree of membrane fouling (Lin et al. 2014; Gao 2019).

Supplementary Table S3 presents the interaction energy between MXT-NFM and six types of pollutant colloids. When the interaction distance between the pollutant and the membrane surface was 0.158 nm, the acid–base interaction energy dominated the total interaction energy, which determined the development trend of membrane fouling. Notably, in the presence of PFOA, the electrostatic double-layer interaction energy between membrane and colloid (U_mlc^EL) value between MXT-NFM and the pollutant was positive, indicating a repulsive electrostatic interaction, which suggests that membrane fouling would be relatively slower. Conversely, when the contaminant contained PFOS, the U_mlc^EL value between MXT-NFM and the pollutant was negative, possibly due to significant differences in surface charge density. This negative value indicated a certain attraction between MXT-NFM and the colloids, which was consistent with the trend of membrane flux decay. The particle size of mixed pollutants was significantly larger than that of the single PFOS and PFOA systems, and particle size greatly influenced the magnitude of interface free energy between membrane and colloid (U_mlc^XDLVO). A larger particle size corresponded to a more negative U_mlc^XDLVO value and a stronger attractive force between the membrane and the colloid, facilitating colloid particle adsorption on the membrane surface and resulting in more severe membrane fouling.

In the later stage of membrane fouling, the MXT-NFM surface is gradually covered by a fouling layer, wherein the interaction between the pollutants near the membrane surface and the deposited pollutants becomes crucial for membrane fouling. Supplementary Table S3 shows that the U_mlc^XDLVO between pollutants was negative in the case of single PFCs systems, primarily due to the strong hydrophobicity of PFOA or PFOS, which contributed to the mutual attraction between the pollutants. However, upon adding HA, the U_mlc^XDLVO of PFOA/HA and PFOS/HA pollutants became positive due to the strong hydrophilicity of HA, leading to mutual exclusion between the pollutants. With the further addition of Ca²⁺, the U_mlc^XDLVO maintained a positive value, albeit slightly lower than that of the PFOA/HA and PFOS/HA systems. This decrease was mainly attributed to the reduced hydrophilicity of the pollutants caused by the addition of Ca²⁺, thereby decreasing the mutual exclusion between the pollutants. Similar to the early stage of membrane fouling, the absolute value of U_mlc^XDLVO in the coexisting system was significantly higher than that in the single PFCs system, which was mainly attributed to the larger particle size of pollutants in the coexisting system, emphasizing the significant influence of colloid particle size on the interaction energy. Furthermore, as shown in Supplementary Table S4, the acid–base interaction energy still dominated the total interaction energy in the later stage when the interaction distance was 0.158 nm, indicating that the hydrophobic interaction between pollutant colloids determined the development trend of membrane fouling.

In summary, during the entire process of membrane fouling, the particle size of the pollutant colloids and the acid–base interaction play crucial roles in the development of membrane fouling.

In this study, MXene-TiO₂ was chosen as the intermediate layer of a PA membrane to investigate the practical performance of MXT-NFM in the treatment of PFCs. Experimental findings demonstrate that MXene-TiO₂ interlayer endowed the MXT-NFM with excellent hydrophilicity reduced thickness of the PA layer, thereby overcoming the trade-off effect for PFCs removal. During membrane fouling tests, MXT-NFM demonstrated significant resistance to membrane fouling with high removal efficiency of PFCs in the complex pollutant. The application of the XDLVO theory reveals that colloidal particle size and acid–base interactions were critical factors influencing the development of membrane fouling.

This work was supported by the Key-Area Research and Development Program of Guangdong Province (2020B0101130001) and the National Natural Science Foundation of China (No. 51978239).

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.