Microalgae have recently played a critical role in biofuel sources due to its rapid growth and low pollution compared to traditional fossil fuels [1,2]. In the microalgae production, harvesting cost can account for 20%−30% of the total cost of microalgae processing, which severely limits its commercial development [3,4]. Generally, microalgae harvesting methods include centrifugation, flocculation, flotation, membrane filtration or a combination of various techniques [5–7]. Recently, due to their low energy input, high efficiency and outstanding stability, membrane filtration processes has drawn considerable research attentions [8,9]. However, the membrane fouling problem is almost inevitable. Membrane fouling is generally divided into reversible fouling and irreversible fouling [10]. During the filtrating, microalgae cells and EOM firstly deposit on the membranes, which initially forms reversible fouling that can be physically removed [11,12]. When the pollutants are tightly attached to the membrane surface or deeply blocked in the pores, an irreversible gel layer formed [13,14]. The irreversible fouling usually requires chemical methods to be removed, which increases the cost of harvesting and limits its further applications [15–17]. Thus, understanding the fouling mechanism and determining compatible strategies to alleviate membrane fouling is essential to improve the harvesting efficiency and stability.

Previous studies have demonstrated that increasing the hydrophilicity of the membranes and introducing surface charge can effectively retard fouling [18–22]. The hydration of the membrane surface forms a hydration layer, which effectively prevents the foulants from directly contacting the membrane surface [19,23]. As a result, less fouling adheres to the membrane surface. Moreover, the earlier reversible fouling is formed mainly from microalgae cells and EOM, which are normally negatively charged [24,25]. Hence, hydrophilic surface and electrostatic repulsion can prevent the negative foulants from attaching to membrane surface, leading to less fouling. Polycarbonate (PC) membranes by incorporating polystyrene sulfonate (PSS) were prepared for harvesting microalgae and demonstrated excellent antifouling, better stability, and hydrophilicity [23]. It is not enough to alleviate the formation of moderately reversible fouling, irreversible fouling still needs to be removed by chemical means. Advanced oxidation processes (AOPs) have been developed to remove irreversible fouling [26–28]. For example, Fenton oxidation effectively generates ^•OH to degrade pollutants into carbon dioxide and water [29–31]. The electro-Fenton membrane, which blended porous carbon (PC) with carbon nanotubes (CNT) and Fe²⁺ had presented prominent water permeance and anti-fouling abilities [32,33]. Nevertheless, for the manufacture of membranes, it is necessary to select materials with favorable electrical conductivity and physio-chemical stability as the substrate, which limits the wide application of conductive films. In addition, the introduction of electrical energy undoubtedly increases energy consumption and the residual metal ions may cause secondary contamination. Low-energy photo-Fenton membranes may offer a promising solution. Photo-Fenton does not require additional electric energy consumption, and it exhibits excellent flux recovery because of its efficient photocatalytic reaction. However, most of the existing studies have investigated the application of photo-Fenton self-cleaning membranes in oil-water separation [34,35], dye degradation [36,37], and wastewater treatment [38,39]. At present, research on self-cleaning membranes that application in microalgae water separation is rare.

Therefore, in this study, superhydrophilic photo-Fenton membranes were prepared and applied to harvest microalgae. TA and APTES formed a hydrophilic nanocoating on the substrate by co-deposition method, giving the membrane outstanding hydrophilicity. TA provides abundant phenolic hydroxyl groups, which can be cross-linked with Fe³⁺ ions. Then in situ mineralized into β-FeOOH nanorods with heterogeneous photo-Fenton catalytic properties, which generate ^•OH under visible light irradiation to degrade irreversible fouling that block membrane pores. We evaluated the hydrophilicity, permeation flux of membranes with different times of coating deposition. In addition, the antifouling performance and flux recovery were further investigated.

Fig. 1a illustrates the deposition and mineralization process of photo-Fenton membrane. The morphologies of the modified and unmodified substrates were shown in Fig. 1b. The original substrate surface was smooth, and the surface after TA-APTES coating had abundant nanoparticles. These nanoparticles were formed by the hydrolysis of APTES, oxidation of TA to quinone, followed by Michael addition and Schiff base reaction [40]. After further mineralization, the substrate surface is evenly distributed β-FeOOH nanorods, there is no obvious difference in the general morphology of these substrates before and after modification, β-FeOOH nanorods uniformly wrap the substrate, which can prevent the substrate from being degraded during the photo-Fenton reaction.

The EDS patterns in Figs. 2a-f reveal that C, F, O, Fe and Si elements exist in the PVDF/TA-APTES@FeOOH-24 membranes. The functional groups on the membrane surface were analyzed by ATR-FTIR, as shown in Fig. 2g, compared to PVDF membrane, PVDF/TA-APTES membrane has a peak at 1720 cm⁻¹, which was attributed to the C=O vibration. In addition, 1608 cm⁻¹ also has a weaker absorption peak, which corresponds to the C=C stretching vibration of the aromatic ring. For the PVDF/TA-APTES@FeOOH-24 membrane, the new peak at 680 cm⁻¹ was attributed to the stretching vibration of Fe-O [41]. Moreover, the broad peak in the range of 3000–3600 cm⁻¹ was attributed to the vibrational stretching vibration of -OH. The peak broadens compared to the PVDF/TA-APTES membrane, which was attributed to the hydroxyl group vibration on the surface of β-FeOOH. The increasing of hydrophilic groups is essential to improve the hydrophilicity of the membrane. The chemical structure and crystal phase of the membranes were characterized by XRD, and it can be seen from Fig. 2h that the PVDF and PVDF/TA-APTES membranes exhibited similar XRD patterns, indicating that the TA-APTES network is an amorphous structure. For the PVDF/TA-APTES@FeOOH-24 membrane, new diffraction peaks appeared at 12° (110), 35.4° (211), 39.4° (301) and 56.2° (521) (Fig. 2b), demonstrating that β-FeOOH (JCPDS No. 34–1266) particles were successfully loaded on the membrane surface [42]. XPS measurements were used to further investigate the surface chemistry and electron transfer processes. Particularly, compared to the pristine PVDF membrane, new N 1s and O 1s peaks appeared for the PVDF/TA-APTES membrane, confirming the successful coating of the TA-APTES (Fig. 2i). The core-level XPS Fe 2p spectra is shown in Fig. 2j, where two satellite peaks at 719.08 eV and 732.92 eV are characteristic of mixed-valence Fe materials. The fitted peaks at 710.55 and 724.22 eV belong to Fe²⁺ while those at 712.56 and 726.17 eV belong to Fe³⁺ [43,44]. These results are consistent with the characterization by XRD and FT-IR, both indicating that the mixed valence β-FeOOH was uniformly deposited on the membrane surface.

To further study the visible light-induced photocatalytic degradation performance, the light absorption characteristics of PVDF membrane and PVDF/TA-APTES@FeOOH-24 membrane were detected by UV–vis diffuse reflectance spectrometer (Fig. 3a). The PVDF membrane exhibited very narrow absorption in the UV region, while the PVDF/TA-APTES@FeOOH-24 membrane exhibits strong and broad absorption in the visible light region of 400~600 nm. This indicates that more visible light can be absorbed after the introduction of β-FeOOH, which will enhance its photocatalytic activity under visible light. Furthermore, the band gap energy of the composite membrane can be obtained according to the following equation:

where α and hν represent the absorption coefficient and discrete photon energy. Respectively, E_g is the band gap, A is a constant, and the value of n is determined by the transition characteristics in the semiconductor (n = 1 means direct transition, n = 4 means indirect transition). The indirect transition value of β-FeOOH is 4, and the direct band gap of the PVDF/TA-APTES@FeOOH-24 membrane is calculated from the tangent intercept of the abscissa (αhν)² and the ordinate hν (2.11 eV, Fig. 3b), which is basically consistent with the reported 2.12 eV [45]. Here, methylene blue (MB) was used as the organic pollutant to verify the photo-Fenton performance of PVDF/TA-APTES@FeOOH-24 membrane. According to the dynamic curve of photo-Fenton catalyzed degradation in Fig. 3c, MB was almost completely degraded within 40 min. Meanwhile, the digital photos of each time interval in Fig. 3d recorded the process of MB solution color changing with prolonged irradiation time. Specifically, the color of the original blue MB solution continued to fade until it became colorless and transparent, with no visible transition in the appearance of the membrane before and after degradation. As shown in Fig. S1 (Supporting information), the concentration of MB decreased by 33.5% after 60 min without visible light. In contrast, under visible light irradiation, led to MB degradation efficiency of 97.4%. Although H₂O₂ can decompose organics, if used alone, as a consumable, H₂O₂ cannot continuously and rapidly generate ^•OH, which will result in low efficiency and high cost. The photocatalysis and Fenton catalysis in the PVDF/TA-APTES@FeOOH system can promote each other, accelerate the generation of ^•OH and promote the degradation of pollutants. These results demonstrate that the PVDF/TA-APTES@FeOOH-24 membranes decorated with β-FeOOH nanoparticles have excellent photo-Fenton catalytic activity.

The wettability of the membranes before and after modification was investigated by the dynamic water contact angle. As shown in Fig. 4a, the WCA of the pristine PVDF membrane did not change much with time and remained approximately 115.8°. Instead, the WCA of the PVDF/TA-APTES@FeOOH-24 membrane decreased to 0° in 5 s. In addition, the hydrophilic properties of the PVDF/TA-APTES@FeOOH-24 membrane decreased significantly with the increase of the coating time, and the WCA of the PVDF/TA-APTES@FeOOH-24 membrane decreased to 8° within 1 s and to 0° within 3 s (Fig. 4c). It shows that the coating can successfully modify the highly hydrophobic PVDF membrane into a superhydrophilic membrane. Generally, for membranes with higher hydrophilicity, the fouling tendency is lower because most fouling in water is hydrophobic [46]. Fig. 4b shows the pure water flux of various membranes. The pure water flux of the original PVDF membrane is about 6501 L/(m²·h·bar), and the TA-APTES coating improves the pure water flux of the membrane. The pure water flux of the PVDF/TA-APTES@FeOOH-0.5 membrane after only 0.5 h of coating reached 14,191 L/(m²·h·bar), an increase of 2.3 times. It is worth noting that with the increase of coating time, the pure water flux first increased and then decreased, and the pure water flux of the PVDF/TA-APTES@FeOOH-2 membrane deposited for 2 h was the highest (14,724 L/(m²·h·bar)). This may be because the long-term deposition of the coating blocks the original pores of the membrane (Fig. S2 in Supporting information), and the excessively dense structure will lead to a decrease in the permeation flux of the membrane. However, although the porosity of the membrane decreased (Fig. S3 and Table S1 in Supporting information), the hydrophilicity of its surface was quite excellent, and the water flux of the membrane increased significantly.

To consider the membrane permeation performance and energy saving purpose, PVDF/TA-APTES@FeOOH-2 membrane was selected to evaluate the membrane filtration harvesting microalgae and antifouling performance with a laboratory-scale staggered flow filtration system (Fig. S4 in Supporting information). Fig. 5a shows the pure water flux of the two membranes before filtration, after filtration without cleaning and after cleaning. The pure water flux of the filtered membranes was significantly reduced, and after cleaning, the pure water flux of the original PVDF membrane was 3928 L/(m²·h·bar), which was much smaller than that of the PVDF/TA-APTES@FeOOH-2 membrane (14,724 L/(m²·h·bar)). This is because the extracellular organic matter deposited on the membrane surface is degraded by strong oxidizing molecules such as reactive oxygen species after the PVDF/TA-APTES@FeOOH-2 membrane is treated with photo-Fenton. Normally, the fouling that clogs the membrane pores consists of reversible fouling and irreversible fouling. The microalgae cells attached to the membrane surface as reversible fouling can be removed by hydraulic cleaning, while as the cells secrete EOM, they will form irreversible fouling on the membrane surface through covalent and van der Waals forces. Fig. S5 (Supporting information) shows the optical photographs and SEM images of the original PVDF membrane and PVDF/TA-APTES@FeOOH-2 membrane after 4 cycles of filtering microalgae. Microalgae cells and EOM attached on the surface completely blocked the original membrane pores, and there was still a lot of fouling after cleaning. Although the surface of the modified membrane was also blocked, there were still some open pores. Surprisingly, the blocked fouling disappeared after being cleaned by photo-Fenton, and the surface structure remained intact. Additionally, after photo-Fenton recovered for 60 min, there was nearly no residual EOM on the PVDF/TA-APTES@FeOOH-2 membrane. The photo-Fenton process had degraded 97.9% polysaccharides and 97.8% proteins (Fig. S6 in Supporting information). To quantify the antifouling properties of the membrane, Fig. 5b illustrates the quantitative characterization of the relative fouling of the two membranes. There is a clear trend that the relative total fouling (73.42%) and irreversible fouling (3.26%) of the PVDF/TA-APTES@FeOOH-2 membrane are less than that of the original PVDF membrane (81.67%, 39.57%), and the lower the percentage of irreversible fouling, the better the antifouling performance of the membrane. Fig. 5c shows the changes of relative permeate flux with time during microalgae filtration by the original PVDF membrane and PVDF/TA-APTES@FeOOH-2 membrane. During filtration, microalgae cells and EOM were deposited on the membrane surface, resulting in a decrease in permeate flux. In the first cycle, there was no significant difference in the decreasing rate of flux between the two membranes. Interestingly, the initial permeate flux of the PVDF/TA-APTES@FeOOH-2 membrane was higher. We attribute the higher permeate flux to its more hydrophilic surface. The membrane with a hydrophilic layer has a high surface tension and can form a hydrated layer with the surrounding water molecules, which prevents the adhesion of hydrophobic fouling on the membrane surface. It is noteworthy that after one cycle, the modified membrane achieved 98.2% flux recovery after 1 h of photo-Fenton treatment, while the flux recovery of the original PVDF membrane was only 63.2%. As for the retention rate of microalgae, it can be seen from Fig. S7 (Supporting information) that the harvesting efficiency of the membranes were greater than 99%. Besides, the ratio of the four signal peaks in the ESR spectrum is 1:2:2:1 (Fig. 5d), which are characteristic peaks of ^•OH [47], indicating that PVDF/TA-APTES@FeOOH-2 membranes can produce ^•OH in situ. In order to characterize the stability of the membrane, we tested the pore size and the amount of iron ions leaching from the membrane before and after the photo-Fenton process. As shown in Fig. S8 (Supporting information), after photo-Fenton filtration, the pore size of the membrane had no obvious change, and the iron dissolution was 0.68 ± 0.13 mg/m² (Fig. S9 in Supporting information). The catalytic membrane prepared in this work is relatively stable and does not dissolve large amounts of iron. A conceptual diagram of the self-cleaning mechanism of the photo-Fenton membrane is given in Fig. 5e. Specifically, β-FeOOH nanorods play a dominant role in the self-cleaning process. Under the irradiation of visible light, β-FeOOH nanorods are excited and generate photogenerated electrons (e⁻) and hole pairs (h⁺), which catalyze H₂O₂ to produce a large amount of ^•OH. The strongly oxidizing ^•OH can decompose the irreversible fouling inside the membrane pore into CO₂, H₂O and small molecules. The excellent photo-Fenton self-cleaning performance of PVDF/TA-APTES@FeOOH-2 membrane has a broad application prospect in the field of long-term filtration for harvesting microalgae.

Moreover, the pure water flux, self-cleaning performance, hydrophilicity, FRR for microalgae harvesting of the PVDF/TA-APTES@FeOOH-2 membranes were compared with other membranes. As shown in Table 1 [5,8,18,23,46,48–50], it can be clearly seen that the PVDF/TA-APTES@FeOOH-2 membrane have higher FRR (98.2%) and higher pure water flux. It displays an excellent comprehensive antifouling performance for microalgae harvesting.

In summary, we successfully prepared photo-Fenton membranes for harvesting microalgae. The abundant hydrophilic groups as well as hydroxyl groups on the surface of β-FeOOH endowed the materials with superhydrophilic properties. We also investigated the pure water flux of the membranes with the coatings reacting for different times. Overall, these results show that the thicker the coating deposited and the denser the membrane, the lower the water flux. However, the water flux of the photo-Fenton membranes is still higher than that of the unmodified membrane. The pure water flux of PVDF/TA-APTES@FeOOH-2 membrane is 2.3 times that of pristine PVDF membrane, and the relative irreversible fouling (3.26%) is much smaller than that of pristine PVDF membrane (39.57%), showing excellent antifouling performance. In addition, the modified membrane has a flux recovery rate of 98.2%, which can effectively remove irreversible fouling. This work paves the way for the development of photo-Fenton membrane filtration harvesting microalgae techniques.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

This work was supported by the Fujian Provincial Science and Technology Cooperation Project (No. 20210002), National Natural Science Foundation of China (No. 31870994).

Supplementary material associated with this article can be found, in the online version, at https://doi.org/10.1016/j.cclet.2022.108020.