Activated carbon adsorption enhanced removal of organic contaminants using a catalytic ceramic membrane with peroxymonosulfate activation

Abstract

Granular activated carbon (GAC) was respectively loaded into catalytic ceramic membrane channels (MG₁) and connected at the rear end of the catalytic ceramic membrane (MG₂) to improve the removal of emerging contaminants via peroxymonosulfate (PMS) activation. Compared to the catalytic ceramic membrane system (M_Co–Mn/PMS) without GAC coupling, the MG₁/PMS and MG₂/PMS systems demonstrated complete bisphenol A (BPA) removal within 5 min, with total organic carbon (TOC) removal rates increasing to 66% and 73%, respectively, and exhibited exceptional stability. Quenching experiments revealed that the MG₁/PMS and MG₂/PMS systems involved hybrid radical pathways involving HVMO and SO₄˙⁻. Liquid chromatography and mass spectrometry (LC-MS) analysis identified 6 and 5 transformation intermediates through two distinct pathways, respectively, indicating that the overall toxicity of the degradation intermediates was reduced, as confirmed by toxicity analysis. Fitting of the composite pollution model showed that the M_Co–Mn/PMS system exhibited significant antifouling ability, and the membrane fouling mechanism was mainly based on intermediate standard blockage.

---

Songxue Wang

Dr Songxue Wang earned his Ph.D. in Municipal Engineering from the Harbin Institute of Technology, China, in 2021. He is currently an Associate Professor in the School of Environmental and Municipal Engineering at Qingdao University of Technology, China. His primary research focuses on functional membrane design and advanced oxidation processes (AOPs) for water treatment applications.

1. Introduction

Membrane separation technologies, particularly ultrafiltration (UF) membranes, have become indispensable in potable water production and wastewater reclamation.^1,2 Ceramic membranes, as inorganic ultrafiltration membranes, have a long service life and exhibit superior thermal stability, chemical resistance, and mechanical strength, making them ideal for organic wastewater purification.^3–5 Over the past few decades, endocrine-disrupting compounds (EDCs) have emerged as a critical environmental concern.⁶ These persistent contaminants exhibit exceptional bioaccumulation potential and trophic transfer efficiency, posing significant risks to aquatic organisms and human health.^7,8 Therefore, removing EDCs has become a defining challenge for 21st-century environmental engineering.

However, ceramic membranes exhibit inadequate rejection efficiencies for small molecular weight organic compounds and progressive membrane fouling, leading to flux decline due to irreversible pore blockage and cake-layer formation. Conventional membrane separation processes further exhibit limited efficacy against recalcitrant emerging pollutants.^9–11 The integration of advanced oxidation processes (AOPs) with ceramic membrane filtration has consequently gained significant research traction. This technology generates multiple reactive oxygen species (˙OH, SO₄˙⁻, and ¹O₂) through heterogeneous activation pathways, achieving remarkable degradation rates over a wide pH range.^12–14 Yu et al. prepared an optimized spinel-incorporated catalytic poly (vinylidene fluoride) membrane (spinel-PVDF), which effectively demonstrated the degradation of a wide range of pollutants. The membrane demonstrated high stability, maintaining over 95% degradation efficiency for 120 h at a flux of 100 L m⁻² h⁻¹ (LMH).¹⁵ Zong et al. synthesized an FeOCl-modified carbon-based catalytic membrane (CM-FeOCl), which could effectively activate 80% of H₂O₂ under neutral conditions, and as a result, a high tetracycline (TC) removal efficiency exceeding 90% could be attained. The authors observed a 45% reduction in the total organic carbon (TOC) removal rate.¹⁶ Wu et al. developed a new catalytic system based on exfoliated MXene-modified BiFeO₃ (MX-BFO) for efficient degradation of antibiotics. Within 30 min, a 50 mg L⁻¹ TC solution could be completely degraded in the system with a TOC removal rate of 54.5%.¹⁷ In addition, other excellent studies have discussed coupling methods between ceramic membranes and peroxymonosulfate (PMS)-based AOPs and their treatment of pollutants.^18–21 Researchers generally believe that the coupling of ceramic membrane filtration and PMS will become a transformative technology for next-generation water treatment systems to address emerging pollutants.

Although the combination of AOPs and membrane filtration technology provides significant advantages, including strong mechanical strength, excellent chemical stability, and resistance to organic pollution under long-term oxidation conditions, achieving complete mineralization of contaminants remains a daunting challenge. Current research predominantly focuses on single catalytic membrane AOP systems for the removal of emerging pollutants, frequently neglecting comprehensive organic matter elimination, as evidenced by persistent TOC levels in treated effluents.^22–24 Conventional strategies for efficient TOC reduction, including nanofiltration (NF), reverse osmosis (RO), electrochemical processes and AOPs, typically require substantial energy input and involve intricate operational protocols.^25,26 These limitations highlight the need for more sustainable and efficient mineralization strategies. Based on our previous work, the single catalytic membrane AOP system (M_Co–Mn/PMS) achieves efficient degradation of the targeted pollutant bisphenol A (BPA). The degradation pathway of BPA and the low removal rate of TOC indicated that the M_Co–Mn/PMS system oxidized BPA into small molecules, achieving a limited degree of mineralization.²⁷ Granular activated carbon (GAC), with its microporous structure and high specific surface area, provides an effective solution for adsorbing low-molecular-weight degradation byproducts and is widely used in the field of water treatment.²⁸ To achieve deep removal of organic contaminants, a hybrid system was developed through the strategic integration of the catalytic ceramic membrane with GAC for stable effluent quality.

In this study, catalytic ceramic membranes were prepared by a facile impregnation–sintering method, and two distinct membrane modules were engineered by varying the positions of the GAC. A comparative investigation was conducted to evaluate the catalytic efficacy of the M_Co–Mn/PMS, MG₁/PMS and MG₂/PMS systems. The investigation systematically evaluated the TOC removal performance and PMS activation efficiency of these three systems. Simultaneously, the stability and reusability of the two GAC-integrated systems were also examined. The main active species involved in BPA degradation in the MG₁/PMS and MG₂/PMS systems were identified, and the degradation pathways were elucidated and toxicity analyses performed. Finally, the antifouling performance of M_Co–Mn was evaluated, and the main causes of membrane fouling were analyzed.

2. Materials and methods

2.1. Chemicals and reagents

Cobaltous nitrate hexahydrate (Co(NO₃)₂·6H₂O), methanol (MeOH), tert-butanol (TBA), and dimethyl sulfoxide (DMSO) were purchased from Sinopharm Chemical Reagent Co., Ltd (Shanghai, China). Bisphenol A (BPA) was purchased from Aladdin Biochemical Co., Ltd (Shanghai, China). L-Histidine, p-benzoquinone (BQ), and peroxymonosulfate (PMS, KHSO₅·0.5KHSO₄·0.5K₂SO₄) were purchased from Shanghai Macklin Biochemical Co., Ltd (Shanghai, China). Unless otherwise stated, all chemicals employed in this study were of analytical grade and were not subjected to further purification. Deionized (DI) water was prepared using ultrapure water acquired from a Millipore Milli-Q water system. Raw water from the Dingjiahe Reservoir in Qingdao (China) was used to investigate membrane fouling and was subjected to preliminary treatment using a 0.45 µm membrane. The raw water quality parameters of Dingjiahe reservoir are shown in Table S1. The initial pH of the reaction solution was adjusted to 7 ± 0.2, and the temperature was maintained at 25 ± 0.5 °C. Granular activated carbon (GAC) was purchased from Xinzhiyuan Activated Carbon Co., Ltd (China). The specific surface area and pore size distribution parameters of the GAC material are provided in Fig. S1 and Table S2. The flat-sheet ceramic membrane prepared in our laboratory was employed in this study, and its characteristic parameters are provided in Table S3.

2.2. Fabrication of the catalytic ceramic membrane

The catalytic ceramic membrane was prepared by a dip-coating and calcination approach. The detailed preparation procedure has been described in our previous work.²⁷ Briefly, 4.5 g of Co(NO₃)₂·6H₂O and 3.0 g of MnCl₂·4H₂O were dissolved into 500 mL of DI water to prepare the impregnating solution. The pristine membrane (M₀) was immersed in the impregnating solution for 24 h, and the membrane was subsequently dried in an oven at 70 °C for 6 h. After drying, the membrane precursor was calcined in a muffle furnace at 600 °C for 3 h under static air, with a heating rate of 5 °C min⁻¹. After cooling to room temperature, the catalytic ceramic membrane loaded with the Co–Mn catalyst was obtained and denoted as M_Co–Mn.

Two integrated catalytic ceramic membrane-activated carbon systems were constructed in this study. In one system, 5 g of GAC was loaded into the M_Co–Mn membrane channels by encapsulation with epoxy resin, and the effective catalytic membrane area was ∼90 cm², which was then assembled with an immersion ceramic membrane filtration device and designated as the MG₁/PMS system. In another system, an activated carbon pipe was prepared using a 1 cm diameter polypropylene hose filled with 5 g of GAC, secured with a 1 cm³ screen at both ends of the pipe and connected to the M_Co–Mn membrane by a quick coupling. This integrated system was named the MG₂/PMS system. The specific physical devices of the two systems are shown in Fig. 1.

Fig. 1 Schematic diagram of the two integrated catalytic ceramic membrane-activated carbon systems.

2.3. Experimental procedures

First, 200 mg L⁻¹ of BPA solution was prepared by mixing 0.2 g of BPA powder in an appropriate amount of deionized water in a 1000 mL volumetric flask and diluting to a constant volume of 1000 mL.

As shown in Fig. 1, the submerged ceramic membrane filtration experiment was carried out in a 1 L feed tank with 10 mg L⁻¹ of BPA solution. The initial solution pH was adjusted with 0.1 M HCl and 0.1 M NaOH. After adding PMS to the feed tank, the peristaltic pump was started at 40 rpm to maintain a membrane flux of 60 L m⁻² h⁻¹ (LMH). The start time was recorded when permeate first exited the outlet. The samples were collected, filtered through a 0.22 μm filter membrane and immediately quenched with 0.2 mL of MeOH solution for subsequent analysis. After each reaction, DI water was used for backwashing, and the membrane module was dried for subsequent experiments.

2.4. Analytical methods

A scanning electron microscope (SEM) (SIGMA300, Zeiss, Germany) and energy dispersive spectroscopy (EDS) were used to analyze the morphology and surface elemental distribution of the catalytic membrane samples. The BPA concentration was measured using high-performance liquid chromatography (HPLC) (LC-20AT, Shimadzu, Japan) equipped with a 4.6 × 250 mm C18 column. The detection wavelength was set at 230 nm, and the mobile phase consisted of ultrapure water/MeOH (35 : 65, v/v). The concentration of PMS in the reaction system was dynamically monitored using an ultraviolet spectrophotometer (UV-vis) (UV-2600, Shimadzu, Japan). The absorbance value was measured at 735 nm in a 10 mm quartz cuvette based on the colorimetric reaction of 2,2′-azinobis (3-ethylbenzothiazoline-6-sulfonic acid) (ABTS), and the PMS concentration was determined from the standard curve (R² > 0.999). The total organic carbon (TOC) removal performance in the two systems was analyzed by a total organic carbon analyzer (TOC-L, Shimadzu, Japan). The degradation products of BPA were analyzed and identified using an ultra-performance liquid chromatograph–mass spectrometer (LC/MS) (Ultimate 3000, Thermo, America). In order to evaluate the toxicity of BPA degradation intermediates, the quantitative structure–activity relationship (QSAR) method was used in this study. The acute toxicity, developmental toxicity, mutagenicity and bioaccumulation of BPA and its degradation intermediates were evaluated using the toxicity assessment software (T.E.S.T.).

2.5. Membrane fouling analysis

The resistance in-series model was used to quantitatively analyze the types of membrane fouling in this study. The intrinsic membrane resistance was calculated based on the transmembrane pressure (TMP) when filtering deionized water. After hydraulic backwashing, membrane fouling can be divided into reversible membrane resistance (R_r) and irreversible membrane resistance (R_ir), as shown in eqn (1) and (2):

R_f = R_ir + R_r (2)

where R_t is the total membrane fouling resistance (m⁻¹); TMP is the trans-membrane pressure (Pa); µ is the dynamic viscosity of the feed water (Pa s); J is the permeate flux (LMH); R_m is the intrinsic membrane resistance (m⁻¹); R_f is the total fouling resistance (m⁻¹); R_ir is the hydraulic irreversible fouling resistance (m⁻¹); and R_r is the hydraulic reversible fouling resistance (m⁻¹).

During the experiment, the flux was maintained at 60 LMH. First, the TMP was recorded while filtering 100 mL of DI water using a pressure gauge, and the average value was calculated as TMP₀. The R_m value of the membrane was calculated according to eqn (3). Subsequently, 1200 mL of the water sample was added and operated under the same conditions. When the processed volume reached 500 mL, filtration was stopped and set as the endpoint of a single filtration test. The transmembrane pressure difference was recorded as TMP₁. The R_f value was evaluated using eqn (4). After hydraulic backwashing, the average TMP was again recorded during filtration of 100 mL of DI water, which was designated as TMP₂. R_r and R_ir could be calculated according to eqn (5) and (6):

3. Results and discussion

3.1. Process performance of the integrated systems

3.1.1. BPA removal of the integrated systems. M₀ and M_Co–Mn were assembled to establish baseline and catalytic membrane systems, respectively. The BPA removal performance was evaluated under the same conditions, and the results are illustrated in Fig. 2a. The comparative experiments revealed that M₀ and M_Co–Mn exhibited minimal BPA adsorption. Furthermore, PMS alone and the M₀/PMS system demonstrated very low BPA removal. When PMS was added, the BPA removal rate of M_Co–Mn/PMS could reach 85%, indicating that the catalytic membrane could effectively degrade BPA. After coupling GAC, MG₁ and MG₂ could adsorb more than 90% of BPA, so by combining advanced oxidation with GAC adsorption, the MG₁/PMS and MG₂/PMS systems could remove 100% of BPA in 5 min.

Fig. 2 (a) BPA removal in different systems and (b) removal efficiency of BPA and TOC in different systems ([BPA]₀ = 10 mg L⁻¹, [PMS]₀ = 0.05 mM, [Flux] = 60 LMH).

The MG₁/PMS and MG₂/PMS systems could degrade BPA in a relatively short time, which might be because GAC itself has the function of adsorption and catalysis.²⁹ As shown in Fig. S2, the activation rate of PMS after reaction for 5 min in the three systems indicated that the utilization rate of PMS was not significantly improved after coupling with GAC. Therefore, GAC mainly adsorbed BPA in the MG₁/PMS and MG₂/PMS systems. As shown in Fig. 2b, the TOC removal rates of the three systems were compared after 60 min of reaction. The results showed that the TOC removal rate of the M_Co–Mn/PMS system was only 50%, while those of the MG₁/PMS and MG₂/PMS systems could reach 66% and 73%, suggesting that the TOC removal rate could be improved by coupling GAC. In the MG₁/PMS system, BPA molecules were oxidized and degraded into small molecules by reactive oxygen species (ROS) on the membrane surface and were then adsorbed and removed by GAC in the channel. Notably, the MG₂/PMS system demonstrated superior TOC removal efficiency, potentially attributable to the elongated external GAC tube configuration, which could prolong the contact time between GAC and small molecular substances. Therefore, the coupling of activated carbon with a catalytic ceramic membrane can significantly improve the removal rates of BPA and TOC.

3.1.2. Impact of operational parameters. A systematic investigation was conducted to evaluate key operational parameters (membrane flux, initial BPA concentration, PMS concentration, pH, and operational temperature) in the M_Co–Mn/PMS, MG₁/PMS and MG₂/PMS systems. Fig. 3a shows the effect of membrane flux on BPA removal in the integrated systems. It was found that the removal rate of BPA decreased slowly with increasing membrane flux. Increasing the membrane flux shortened the contact time between PMS and the catalyst, resulting in fewer active species being produced and a lower removal efficiency of BPA.³⁰Fig. 3b shows the effect of the BPA initial concentration on BPA degradation in the three systems. It was observed that with the increase of BPA initial concentration from 2 mg L⁻¹ to 20 mg L⁻¹, the BPA removal rates of the M_Co–Mn/PMS, MG₁/PMS and MG₂/PMS systems decreased by 20%, 15% and 13%, respectively. This might be due to a decrease in the number of active free radicals obtained by each BPA molecule at higher concentrations of BPA, resulting in a decrease in the efficiency of BPA removal.³¹ The decreases in BPA removal rates in the MG₁/PMS and MG₂/PMS systems were relatively small, primarily attributed to the synergistic adsorption capacity of activated carbon that enabled non-degraded BPA molecules to be removed by activated carbon adsorption.³² The effects of PMS concentration on BPA degradation in the M_Co–Mn/PMS, MG₁/PMS and MG₂/PMS systems are shown in Fig. 3c. The BPA removal rates of the M_Co–Mn/PMS, MG₁/PMS and MG₂/PMS systems increased by 43%, 29% and 26%, respectively, upon increasing the PMS concentration from 0.01 mM to 0.1 mM. In addition, the effects of initial solution pH on BPA removal by the M_Co–Mn/PMS, MG₁/PMS and MG₂/PMS systems were investigated. As shown in Fig. 3d, pH variation exerted a relatively weak influence on the M_Co–Mn/PMS, MG₁/PMS and MG₂/PMS systems. Remarkably, the MG₁/PMS and MG₂/PMS systems maintained a BPA removal rate of over 90% even under extreme acidic (pH 3) or alkaline (pH 11) conditions. The temperature-dependent catalytic performance was systematically evaluated, as illustrated in Fig. 3e. When the temperature increased from 15 °C to 35 °C, the BPA removal rates of the M_Co–Mn/PMS, MG₁/PMS and MG₂/PMS systems increased by 8%, 12% and 10%, respectively. Systematic evaluation revealed that temperature changes had little effect on the BPA removal efficiency in these systems. The effects of water quality factors, represented by chloride ions (Cl⁻), bicarbonate (HCO₃⁻), sulfate (SO₄²⁻) and nitrate (NO₃⁻), on BPA removal in the MG₁/PMS and MG₂/PMS systems were investigated, and the results are shown in Fig. 3f. The addition of inorganic anions showed no significant effect on BPA removal rates in any of the three systems. Notably, the M_Co–Mn/PMS membrane demonstrated intrinsic resistance to multi-ion interference, maintaining its degradation performance independent of activated carbon coupling. Comprehensive evaluation revealed that both MG₁/PMS and MG₂/PMS systems maintained exceptional BPA removal efficiencies of 90% across diverse environmental conditions, including variations in initial pH, operational temperature, and the presence of competing inorganic anions. While system performance demonstrated sensitivity to membrane flux, initial BPA concentration, and PMS concentration, the degradation efficiency consistently exceeded 70%, underscoring the robust nature of these hybrid processes. These results further confirmed the stability and reliability of the two combined processes in the treatment of complex water quality, as well as robustness to interference from environmental interference factors.

Fig. 3 Effects of (a) permeate flux, (b) BPA concentration, (c) PMS concentration, (d) pH, (e) operating temperature, and (f) coexisting anions on BPA removal in M_Co–Mn/PMS, MG₁/PMS and MG₂/PMS systems ([BPA]₀ = 10 mg L⁻¹, [PMS]₀ = 0.1 mM, [Flux] = 60 LMH, pH = 7).

3.2. Analysis of dominant reactive species and stability

In this study, the dominant reactive species responsible for BPA degradation in the MG₁/PMS and MG₂/PMS systems were investigated by a series of free radical quenching experiments. MeOH was used as the capture agent for ˙OH and SO₄˙⁻, TBA as the capture agent for ˙OH, and DMSO as the probe for high-valence metal-oxo species (HVMO).³³ As shown in Fig. 4a, after adding 500 mM MeOH to the MG₁/PMS and MG₂/PMS systems, the degradation rates of BPA decreased by 7.8% and 11.8%, respectively. When 500 mM TBA was added to the two systems, the degradation rates decreased by 2.8% and 4.1%, respectively. The results indicated that ˙OH and SO₄˙⁻ were both generated in the MG₁/PMS and MG₂/PMS systems, but they were not the dominant species responsible for BPA degradation in the two systems. Furthermore, when 10 mM DMSO, as the probe for high valence metal-oxo species, was added to the two systems, the degradation rates significantly decreased by 35.2% and 30.5%, respectively, in the MG₁/PMS and MG₂/PMS systems. The results showed that HVMO played a more significant role in the two systems. Notably, although HVMO played a more important role than ˙OH and SO₄˙⁻, the presence of activated carbon in the systems ensured that the BPA removal rate remained above 60% in both systems even after the quenching reaction. The high-resolution XPS spectra of Co 2p and Mn 2p before and after the catalytic reaction of M_Co–Mn are shown in Fig. S3. In the reaction system, Mn⁴⁺ accepted PMS electrons to form a metastable complex, followed by O–O bond cleavage to produce Mn⁵⁺, and Co²⁺ accepted PMS electrons to form Co⁴⁺. The Mn⁵⁺ and Co⁴⁺ directly oxidized BPA and were subsequently reduced to Mn⁴⁺/Mn³⁺ and Co²⁺, respectively. Concurrently, the Mn⁴⁺/Mn³⁺ cycle produced ˙OH and SO₄˙⁻, while Co²⁺ hydrolyzed to CoOH⁺ to further activate PMS for SO₄˙⁻, synergistically enhancing activation efficiency through the Co²⁺/Co³⁺ cycle. Therefore, in addition to adsorption for BPA removal, HVMO (Co⁴⁺/Mn⁵⁺) and SO₄˙⁻ jointly participate in the process of BPA degradation.

Fig. 4 Effects of (a) coexisting anions and (b) long cycle operation on the removal of BPA in different systems ([BPA]₀ = 10 mg L⁻¹, [PMS]₀ = 0.1 mM, [Flux]=60 LMH, pH = 7).

The degradation efficiency of BPA by the M_Co–Mn/PMS, MG₁/PMS, and MG₂/PMS systems during long-term operation was compared to evaluate the stability. As shown in Fig. 4b, the single MG₁ and MG₂ systems initially achieved BPA removal through activated carbon adsorption, but the BPA removal efficiency significantly decreased as the reaction progressed due to adsorption saturation. The M_Co–Mn/PMS system demonstrated a marked decline in catalytic efficiency, exhibiting an initial BPA removal efficiency of 83% at 30 min that progressively deteriorated to 40.6% after 420 min of continuous operation, suggesting relatively poor stability of BPA removal in the single M_Co–Mn/PMS system. In comparison, the MG₁/PMS and MG₂/PMS systems maintained superior operational stability, with initial removal efficiencies of 94.4% and 93.3% at 30 min, respectively, which were only attenuated to 61.2% and 69.1% after 420 min of operation. This long-term stable removal efficiency was attributed to the sustained adsorption of activated carbon and the synergistic effect of ceramic membrane catalytic oxidation, effectively avoiding the problems of adsorption saturation and reduction in catalytic efficiency.^34,35

3.3. Analysis of degradation pathway and toxicity

3.3.1. Degradation pathway. The degradation intermediate by-products of BPA in the MG₁/PMS and MG₂/PMS systems were elucidated through LC-MS analysis. The results obtained are shown in Fig. 5. As shown in Fig. 5a, there were two potential pathways for the degradation of BPA in the MG₁/PMS system, resulting in 6 possible degradation intermediates. In pathway 1, active species generated by the reaction resulted in the breaking of the C–C bond of BPA to form the intermediate product P1 (m/z = 150) connected to a methyl radical. P1 subsequently underwent an oxidation reaction and combined with ˙OH to generate P2 (m/z = 153). In pathway 2, BPA underwent a hydroxylation reaction to generate P3 (m/z = 243), which was a common pathway for activating PMS to degrade BPA.³⁶ Under the action of active species, the benzene ring underwent cleavage and produced P4 (m/z = 149),³⁷ which was then oxidized and degraded into small molecule substances P5 (m/z = 60) and P6 (m/z = 73), ultimately mineralizing into CO₂ and H₂O. As illustrated in Fig. 5b, there were also two possible degradation pathways in the MG₂/PMS system, generating a total of 5 possible intermediate products. Pathway 1 in the MG₂/PMS system was consistent with Pathway 1 in the MG₁/PMS system. Furthermore, some intermediate products underwent intermolecular coupling via H⁻ extraction (Pathway 2), resulting in the formation of polymer P3 (m/z = 362).³⁸ These coupling products were further cleaved by active species into smaller fragments P4 (m/z = 210) and ultimately degraded into P5 (m/z = 60) or mineralized to CO₂ and H₂O.

Fig. 5 BPA degradation pathways of (a) the MG₁/PMS system and (b) the MG₂/PMS system.

The above results show that, in the MG₁/PMS system, the activated carbon was in close contact with the catalytic ceramic membrane, and GAC preferentially adsorbed BPA and its macromolecular intermediate products (such as P1 and P3), accelerating the attack of free radicals on pollutants through a local enrichment effect and inhibiting coupling side reactions (the formation of P3 in pathway 2).³⁹ Compared with the 11 intermediate products produced by the M_Co–Mn/PMS system previously studied, the number of intermediate products in the MG₁/PMS system was decreased to 6.²⁷ The degradation pathway of the MG₁/PMS system was significantly shortened, and the TOC removal rate was increased to 66%. In the MG₂/PMS system, as the GAC was located at the back end of the oxidation reaction, oxidation and adsorption proceeded step by step. Oxidative intermediate products (such as P3, m/z = 362) underwent intermolecular coupling reactions (pathway 2) to form macromolecular substances. The GAC tube extended the hydraulic retention time, enhanced the contact between GAC and small molecules, effectively adsorbed BPA and its transformation intermediate products, and increased the TOC removal rate to 73% (Fig. 2b).

3.3.2. Evaluation of potential toxicity. This investigation systematically assessed the toxicity of intermediate products generated in the MG₁/PMS and MG₂/PMS systems. The toxicity evaluation was performed using the median lethal concentration (LC₅₀) for fathead minnows, the oral LD₅₀ for rats, the LC₅₀ for Daphnia magna Straus, and the bioaccumulation factor (BF) with quantitative structure–activity relationship (QSAR) modeling predictions.^40–42 As shown in Fig. 6, the toxicity assessment revealed that the fathead minnow LC₅₀ and the Daphnia magna Straus LC₅₀ of intermediate products produced in both systems were significantly higher than those of BPA. The overall acute toxicity levels of the MG₁/PMS and MG₂/PMS systems were comparable. While the LD₅₀ of P3 in the MG₂/PMS system was 780 mg kg⁻¹, which was 77 mg kg⁻¹ lower than that in the MG₁/PMS system, the GAC tube adsorbed most P3, effectively mitigating its bioavailability. Bioaccumulation potential analysis demonstrated that the BF values of all intermediate products in the MG₁/PMS and MG₂/PMS systems were lower than those of BPA. The BF values of the polar small molecule products P5 and P6 were less than 10, indicating particularly low accumulation potential, meaning that they are not easily enriched in organisms. Although the BF value of P3 in MG₂/PMS was 45, its poor water solubility and low environmental persistence reduced the actual enrichment risk. Therefore, the biological enrichment risks of these two systems were controllable, as BPA was transformed into products with significantly reduced toxicity and lower bioaccumulation potential.

Fig. 6 Toxicity variations of (a and e) fathead minnow LC₅₀, (b and f) Daphnia magna LC₅₀, (c and g) oral rat LD₅₀ and (d and h) bioconcentration factor in the degradation pathways of BPA in the MG₁/PMS and MG₂/PMS systems.

3.4. Membrane fouling analysis

A systematic evaluation of membrane fouling behavior was conducted using Dingjiahe Reservoir water as the background raw water, with a particular focus on the relative transmembrane pressure (P/P₀) changes and the reversibility of fouling in the catalytic ceramic membrane system. As shown in Fig. 7a, the M_Co–Mn/PMS system exhibited excellent antifouling performance. After three cycles, its final P/P₀ remained at 5.33, while P/P₀ values of the M₀, M_Co–Mn, and M₀/PMS systems were 9.5, 8.5, and 7.55, respectively. Fig. 7b shows that the R_f values of M₀, M_Co–Mn, and M₀/PMS were 1.369 × 10¹² m⁻¹, 1.305 × 10¹² m⁻¹ and 1.253 × 10¹² m⁻¹, respectively. However, the M_Co–Mn/PMS system exhibited a substantially reduced R_f of 4.23 × 10¹¹ m⁻¹, representing a reduction of over 65% compared to the other three systems. In the M_Co–Mn/PMS system, the Co–Mn catalyst on the ceramic membrane effectively activated PMS to generate reactive species for oxidizing and decomposing large organic molecules, thereby reducing the formation of filter cake layers. In addition, the active species alleviated pore blockage by degrading small-molecule pollutants adsorbed in the membrane pores.

Fig. 7 (a and b) Membrane fouling generated, R_r and R_ir by different systems filtering actual water. (c and d) Effects of different PMS concentrations on membrane fouling generated, R_r and R_ir under the M_Co–Mn/PMS system filtering actual water. (e and f) Effects of different fluxes on membrane fouling generated, R_r and R_ir under the M_Co–Mn/PMS system filtering actual water ([Flux] = 60 LMH, [PMS] = 0.2 mM).

A systematic investigation of the changes in P/P₀ and the reversible effects of membrane fouling in the M_Co–Mn/PMS system under different PMS concentrations was conducted to demonstrate the regulatory effect of PMS concentration on membrane fouling,⁴³ as shown in Fig. 7c and d. As the PMS concentration increased from 0 mM to 0.3 mM, after three cycles, the P/P₀ value of the M_Co–Mn/PMS system decreased from 8.5 to 1.96, accompanied by parallel mitigation of both R_r and R_ir, which decreased accordingly to 2.56 × 10¹¹ m⁻¹ and 1.42 × 10¹¹ m⁻¹. Notably, no significant differences between the conditions of 0.2 mM and 0.3 mM PMS were observed in both P/P₀ and R_f values, suggesting that the antifouling performance reached saturation at 0.3 mM PMS. Therefore, at an appropriate PMS concentration, the M_Co–Mn/PMS system could significantly improve the alleviation efficiency of membrane fouling, reduce irreversible fouling resistance, and enhance the long-term stable operational capability of the membrane. Furthermore, Fig. S4 and Table S4 show the fitting analysis of five composite pollution models for the M₀ and M_Co–Mn/PMS systems. M₀ exhibited the highest fitting degree with the model of complete blockage of the filter cake layer, R² = 0.9992. In the M_Co–Mn/PMS system, the intermediate standard blockage model became predominant at 0.2 mM PMS, R² = 0.9925, indicating partial pore constriction without complete occlusion. The addition of PMS significantly enhanced the decomposition of pollutants on the membrane surface, reduced the accumulation of pollutants at the membrane pore orifices, and facilitated deeper foulant penetration into the membrane matrix. Although increasing the PMS concentration to 0.3 mM slightly reduced the fitting quality, R² = 0.9923, the persistent dominance of the intermediate standard blockage model confirmed the system's stability. The minor fitting degrees of other models remained stable across different PMS concentrations. The M_Co–Mn/PMS system optimized the antifouling ability and reduced membrane fouling through precise PMS concentration control.

The effects of different membrane fluxes on membrane fouling in the M_Co–Mn/PMS system were investigated and are shown in Fig. 7e and f. At a flux of 40 LMH, R_r and R_ir were 0.68 × 10¹¹ m⁻¹ and 1.58 × 10¹¹ m⁻¹, respectively, with a P/P₀ of 2.12. When the flux was 60 LMH, P/P₀ was 2.23 and R_f was 2.5 × 10¹¹ m⁻¹. The M_Co–Mn/PMS system exhibited excellent antifouling performance with minimal parameter fluctuations. In contrast, when the flux increased to 80 LMH, after three operating cycles, the P/P₀ value reached 6.73, accompanied by severe membrane fouling. The R_f and R_ir values were 4.81 × 10¹¹ m⁻¹ and 3.35 × 10¹¹ m⁻¹, respectively. Under relatively low flux conditions, the contact time between pollutants and active species was prolonged, achieving complete oxidative degradation. Conversely, excessive flux accelerated the migration of pollutants, leading to severe membrane fouling.

The surface characterization through SEM elucidated the formation of actual water membrane surface fouling in membrane filtration of different systems, as shown in Fig. 8. Membrane samples were backwashed with 0.2 mM PMS solution for 30 min and left to stand undisturbed for 24 h after a complete filtration cycle. M₀ developed a dense cake layer with irregular aggregates, demonstrating classical complete pore-blocking behavior. The slight reduction in pollutants on the surface of M_Co–Mn could be attributed to improvements in membrane surface morphology via the Co–Mn catalyst. M_Co–Mn had a certain adsorption effect on pollutants and could reduce membrane fouling to a certain extent. Notably, 0.2 mM PMS incorporation achieved almost complete membrane surface cleaning. The M_Co–Mn/PMS system exhibited exceptional antifouling performance. The elemental distribution on the membrane surface of the M_Co–Mn/PMS system after multiple filtration cycles was analyzed using EDS, as shown in Fig. S5 and Table S5. Even after multiple filtration cycles, Co and Mn elements were still uniformly distributed on the membrane surface. Notably, the mass fraction of the C element was 26.42%, indicating that a large amount of organic matter adhered to the membrane surface. Combining SEM and EDS analysis results, it can be seen that the M_Co–Mn/PMS system can not only effectively decompose pollutants and reduce membrane fouling, but also maintain the long-term stability and activity of the catalyst.

Fig. 8 SEM images of (a and b) M₀, (c and d) M_Co–Mn and (e and f) M_Co–Mn/PMS after filtering actual water.

4. Conclusions

In this study, granular activated carbon (GAC) was loaded into the catalytic ceramic membrane channels (MG₁) and connected at the rear end of the catalytic ceramic membrane (MG₂) for improving the removal of emerging contaminants via PMS activation. Compared with the catalytic ceramic membrane system without GAC coupling (M_Co–Mn/PMS), the MG₁/PMS and MG₂/PMS systems demonstrated complete removal of BPA within 5 min, with TOC removal rates increasing to 66% and 73%, respectively. The quenching experiments showed that the MG₁/PMS and MG₂/PMS systems utilized hybrid oxidation pathways involving HVMO (Co⁴⁺/Mn⁵⁺) and SO₄˙⁻. LC-MS analysis indicated that degradation occurred through two pathways, and compared with the 11 intermediate products produced by M_Co–Mn/PMS, the intermediate products in the MG₁/PMS and MG₂/PMS systems were reduced to 6 and 5, respectively. The comprehensive characterization through LC-MS and toxicity analysis confirmed the effective treatment of low-concentration wastewater and an overall reduction in toxicity. The fitting of the composite pollution model showed that the M_Co–Mn/PMS system exhibited significant antifouling ability, and the membrane fouling mechanism was mainly based on intermediate standard blockage. After three consecutive operational cycles, the M_Co–Mn/PMS system significantly exhibited a remarkable 69.1% reduction in R_f compared to the M₀ alone, indicating its significant advantage in controlling membrane fouling. Therefore, the construction of the MG₁/PMS and MG₂/PMS systems provides a research direction for the efficient removal of emerging contaminants and the deep purification of water.

Author contributions

All authors have approved the final version of the manuscript. CRediT: Songxue Wang: investigation and writing – review & editing; Yiding Wang: data curation and writing – original draft; Zeng Xue: investigation; Zixuan Gao: visualization; Yazun Sun: formal analysis; Jingjing Xia: methodology; Baoxiu Zhao: investigation; Wenxiang Xia: resources; Xiuzhen Wei: conceptualization and writing – review & editing; and Lu Zeng: validation.

Conflicts of interest

There are no conflicts of interest to declare.

Data availability

The data supporting the findings of this study are available within the paper and its supplementary information (SI). Supplementary information is available. See DOI: https://doi.org/10.1039/d5nr03774c.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (No. 52100010) and the Natural Science Foundation of Shandong Province (No. ZR2021QE185).

References

1. S. Van Geluwe, L. Braeken and B. Van der Bruggen, Water Res., 2011, 45, 3551–3570.
2. B. Li, B. Qi, Z. Guo, D. Wang and T. Jiao, Chemosphere, 2023, 327, 138528.
3. S. Lee, M. Dilaver, P. Park and J. Kim, J. Membr. Sci., 2013, 432, 97–105.
4. D. Wang, X. Duan, X. He and D. D. Dionysiou, Environ. Sci. Pollut. Res., 2016, 23, 23772–23780.
5. C. Ursino, R. Castro-Muñoz, E. Drioli, L. Gzara, M. Albeirutty and A. Figoli, Membranes, 2018, 8, 18.
6. T. Guo, K. Pan, Y. Chen, Y. Tian, J. Deng and J. Li, Sci. Total Environ., 2024, 907, 167792.
7. M. Furhacker, S. Scharf and H. Weber, Chemosphere, 2000, 41, 751–756.
8. F. S. vom Saal, S. C. Nagel, B. L. Coe, B. M. Angle and J. A. Taylor, Mol. Cell. Endocrinol., 2012, 354, 74–84.
9. X. Cheng, H. Liang, F. Qu, A. Ding, H. Chang, B. Liu, X. Tang, D. Wu and G. Li, Chem. Eng. J., 2017, 308, 1010–1020.
10. E. O. Ezugbe and S. Rathilal, Membranes, 2020, 10, 89.
11. R. Yuan, S. N. Ramjaun, Z. Wang and J. Liu, J. Hazard. Mater., 2011, 196, 173–179.
12. C. Zhang, H. Chu, Q. Ma, Y. Chen and J. Fan, Appl. Sci., 2023, 13, 3182.
13. A. Medici, G. Luongo, S. Pedatella, L. Previtera, G. Di Fabio and A. Zarrelli, Molecules, 2024, 29, 2237.
14. Y. Wang, C. Li, Y. Meng, Z. Guo, S. Cui, X. Fu, H. Liu, W. Xia and W. Li, Water Res., 2024, 250, 121055.
15. L. Yu, M. Zha, Y. Li, J. Zhang and Y. Wang, Water Res., 2025, 284, 123991.
16. Z. Pan, X. Zhou, X. Liu, C. Song, W. Qian, H. Yu, R. Xu, L. Li, S. Zhao and T. Wang, J. Water Process Eng., 2025, 74, 107770.
17. M. Wu, C. Wang, R. Wang, P. Sun, B. Zhou, Y. Xiong and X. Dong, Sep. Purif. Technol., 2025, 376, 134067.
18. A. Hossain, Y. Nishimura, U. Salma, M. Tokumura, T. Nishino, M. Raknuzzaman, K. Noro, K. Watanabe, T. Amagai and M. Makino, Results Eng., 2024, 24, 103183.
19. Y. Wang, W. Wu, X. Yu, Z. Sun, H. Wang and J. Guo, J. Membr. Sci., 2025, 732, 124250.
20. H. Zhao, Z. Zhang, H. Yu, D. Wu, H. Wang and X. Zhang, Environ. Res., 2025, 282, 122097.
21. X. Zhang, W. Zhang, J. Li, T. Wang, H. Miao, Q. Fan, H. Zhu, Z. Yang and C. Kong, Sep. Purif. Technol., 2024, 328, 125135.
22. K. Nelson, A. C. Mecha and A. Kumar, Results Mater., 2025, 27, 100750.
23. Q. Yin, Y. Wang, W. Li, C. Wei, Y. Chen, C. Liu, X. Ling and Z. Chen, J. Environ. Chem. Eng., 2024, 12, 113103.
24. P. Santiago-Espiñeira, P. García-Muñoz, P. Campayo-Navarro and J. Rodríguez-Chueca, J. Environ. Manage., 2025, 388, 125983.
25. F. E. Titchou, H. Zazou, H. Afanga, J. El Gaayda, R. Ait Akbour, P. V. Nidheesh and M. Hamdani, Chem. Eng. Process., 2021, 169, 108631.
26. D. A. Nguyen and A. Jang, Water Res., 2024, 266, 122537.
27. S. Wang, J. Xia, Y. Zha, J. Zhu, Y. Wang, R. Huang, S. Shan, B. Yan, B. Ma, W. Xia and J. Tian, Chem. Eng. Sci., 2025, 301, 120698.
28. J. Shin and B. An, Chemosphere, 2024, 357, 142025.
29. W. Chen, X. Zhang, M. Mamadiev and Z. Wang, RSC Adv., 2017, 7, 927–938.
30. T. Lim, P. Yap, M. Srinivasan and A. G. Fane, Crit. Rev. Environ. Sci. Technol., 2011, 41, 1173–1230.
31. S. Liu, Z. Zhang, F. Huang, Y. Liu, L. Feng, J. Jiang, L. Zhang, F. Qi and C. Liu, Appl. Catal., B, 2021, 286, 119921.
32. M. Wei, L. Gao, J. Li, J. Fang, W. Cai, X. Li and A. Xu, J. Hazard. Mater., 2016, 316, 60–68.
33. Y. Lei, Y. Yu, X. Lei, X. Liang, S. Cheng, G. Ouyang and X. Yang, Environ. Sci. Technol., 2023, 57, 5433–5444.
34. X. Zhu, Y. Zhang, S. Zhang, S. Yang and Q. Xu, J. Alloys Compd., 2022, 924, 166500.
35. S. Zhu, Y. Xu, Z. Zhu, Z. Liu and W. Wang, Chem. Eng. J., 2020, 384, 123298.
36. Y. Huang and Y. Huang, J. Hazard. Mater., 2009, 167, 418–426.
37. J. Du, J. Bao, Y. Liu, H. Ling, H. Zheng, S. H. Kim and D. D. Dionysiou, J. Hazard. Mater., 2016, 320, 150–159.
38. X. He, X. Yang, C. Zhang, Y. Xiao and Y. Tang, J. Cleaner Prod., 2023, 414, 137571.
39. Z. Gao, L. Liu, J. Liu, W. Wang, N. Yang, L. Tao, Z. Li, Q. Gu and X. Liu, Nat. Synth., 2025, 4, 84–96.
40. X. Yu, S. Wang, Y. Zhang, X. Yu, X. Zhao, H. Gao, H. Yang, L. Fang and A. Syed, Surf. Interfaces, 2024, 54, 105222.
41. J. P. Meador, Environ. Pollut., 2021, 291, 118201.
42. X. Yu, S. Wang, Y. Zhang, H. Gao, X. Zhou, D. Li, H. Yang, L. Fang, H. Zhang and A. Syed, Ceram. Int., 2024, 50, 29528–29546.
43. M. Zhou, J. Chen, S. Yu, B. Chen, C. Chen, L. Shen, B. Li and H. Lin, Chem. Eng. J., 2023, 451, 139009.

---