3.1. Morphological and structural analysis
The SEM images of Fe3O4, MIL-101(Fe) and MF-50 are shown in Fig. 2. It can be clearly seen that Fe3O4 is a structure consisting of clusters of nanospheres and MIL-101(Fe) is an octahedral structure, which is similar to that reported previously (Zhang et. al, 2021; Zhao et. al, 2020). For the MF-50 composite, it can be seen that it is still roughly an octahedral-like structure with a few nanospheres on its surface, demonstrating that the Fe3O4 and MIL-101(Fe) components are combined and in good contact. EDS supports the study of the chemical composition of MF-50. According to Fig. 2d, C, O and Fe are dispersed uniformly on the surface of MF-50, which proves its successful preparation.
The application of XRD is necessary in order to study the crystal structure and phase composition of the materials. The Fig. 3a shows the XRD spectra of Fe3O4, MIL-101(Fe) and MF-50. The XRD plots of Fe3O4 show diffraction peaks at 18.269°, 30.095°, 35.422°, 37.052° and 43.052°, which correspond to the JCPDS 19–0629 standard card for Fe3O4, demonstrating that the preparation of Fe3O4 is satisfactory. The XRD plots of MF-50 show peaks of Fe3O4 as well as MIL-101, proving that the material has been successfully compounded.
The use of XPS assists in the analysis of the chemical properties of the materials. The survey of XPS spectrum (Fig. 3b) shows that the MF-50 material contains the elements C, O, Fe and Cl. The Fig. 3c shows the Fe 2p spectra, where the area ratio of the Fe 2p 1/2 peak of Fe3O4 to the Fe 2p 3/2 peak is 2:1, which is consistent with the Fe(III):Fe(II) = 2:1 feature of Fe3O4; as for MIL-101, the difference between the Fe 2p 3/2 (711.2 eV) peak and the Fe 2p 1/2 (724.9 eV) peak is 13.7 eV, which is similar to that previously reported (Zhao et. al, 2020)the spectrum of MF-50 is very similar to that of MIL-101, but its satellite peaks are shifted, demonstrating an interaction between Fe3O4 and MIL-101. As shown in the Fig. 2d, the O 1s spectra of both MIL-101 and MF-50 can be divided into three peaks corresponding to C-O, C = O, and Fe-O bonds (Guo et. al, 2020).The Fe-O peak of MIL-101 is lower than that of the C = O bond, while the contrary is observed for MF, indicating that Fe3O4 is successfully compounded with MIL-101.
3.2. Analysis of photoelectric properties and energy band structure
The employment of the UV-vis DRS allows exploring the light absorption characteristics of the material and deriving the bandgap energy of the material. As shown in the Fig. 4a, MIL-101 exhibits excellent light absorption in the UV-vis region and when compounded with Fe3O4, the absorption in the visible region is improved. This can be attributed to the coupling between Fe3O4 and MIL-101 enhancing the efficiency of visible light utilisation. The Kubelka-Munk function for counting the band gap energy of photocatalysts based on UV-vis DRS spectra is as follows: \({\alpha }\text{h}{\nu }=\text{A}{(\text{h}{\nu }-{\text{E}}_{\text{g}})}^{\text{n}/2}\) (Yang et. al, 2021b). Plotting the light energy (\({\alpha }\text{h}{\nu }\))2 versus energy (\(\text{h}{\nu })\), the band gap energy of MIL-101 and Fe3O4 are calculated to be 2.54 eV and 1.57 eV (Fig. 4b). The analysis of the XPS valence band spectrum can calculate the valence band potential of the materials. According to Fig. 4c, the valence band potentials of Fe3O4 and MIL-101 are 0.99 eV and 2.5 eV, respectively. The test of Mott Schottky can estimate the flat-band potentials. Based on Fig. 4d, the flat-band potentials of Fe3O4 and MIL-101 are − 0.45 eV and 0.06 eV, respectively, and both are n-type semiconductors. Typically, the flat-band potentials of n-type semiconductors are slightly higher (0.1–0.2 eV) than the conduction band potential (Yu et. al, 2016). Therefore, the conduction band potentials of Fe3O4 and MIL-101 are − 0.58 eV and − 0.04 eV, respectively. The energy band structure of both Fe3O4 and MIL-101 are consistent with previous reports (Li et. al, 2019; Wu et. al, 2020).
The determination of EIS allows the charge transfer capability of samples to be studied. The EIS surveys of samples were conducted in NaSO4 solution (0.20 M) with a frequency range of 10− 5 to 0.01 Hz. Typically, lower EIS curves mean lower resistance and higher charge transfer efficiency (Du et. al, 2021a). As shown in the Fig. 4e, the EIS of MIL-101 is lower than that of Fe3O4, and the EIS of the composites of the two is the lowest,, indicating that the heterojunction formed by the two contributes to the separation of photogenerated hole electron pairs.
3.3. Photodegradation performance
The Fig. 5a demonstrates the removal efficiency of OTC in the photo-Fenton system for MF with different Fe3O4 compound ratios. The removal efficiency is significantly improved after the Fe3O4 compounding because of the formation of heterojunctions and the increase of Fe sites. The highest removal efficiency (87.1%) occurred at MF-50/light/PS.
In order to investigate the role of the components of the reaction, a number of control experiments were carried out. As shown in Fig. 5b, there was no OTC degradation under visible light alone, indicating its structural stability. In the light/PS system, OTC only slightly degraded, demonstrating the ineffectiveness of PS activation by visible light. Also, the removal efficiency of OTC in photocatalytic systems was still relatively low. However, in the Fenton system (MF-50/PS), 61.1% of the OTC was removed, indicating that the sample had a superior activation effect on PS. Notably, in the photo-Fenton system, the synergistic effect of light, MF-50 and PS significantly enhanced the removal efficiency of OTC.
3.3.1. Effect of catalyst dosage
The dosage of catalyst is an important factor affecting the removal efficiency of OTC. We set up parallel experiments with different MF-50 doses (0.05, 0.1, 0.2, 0.3, 0.4 and 0.5 g/L respectively) and the results are shown in the Fig. 5c. The removal efficiency of OTC by the dark reaction increased with the amount of catalyst, because more catalyst provided more adsorption. While in the light reaction, the removal efficiency was well improved when the catalyst dosage was increased from 0.05 g/L to 0.2 g/L, but the removal efficiency was not significantly improved from 0.2 g/L to 0.5 g/L, which may be due to the scattering of light caused by the increase in the amount of catalyst.
3.3.2. Effect of pH
The initial pH value of the solution is an important factor influencing the removal of OTC. Hence, HCl (0.1 mol/L) or NaOH (0.1 mol/L) was adopted to tune the pH value of the OTC solution for investigating the impact of pH value on OTC removal. The Fig. 5d shows the removal efficiency of OTC in the MF/vis/PS system at different initial solution pH values. The dark reaction is less effective in an acidic environment. Combined with the results of zeta potential analysis (Fig. 5g), the zero potential point of the MF material is 9.48, while the OTC molecule exhibits electropositivity at pH < 5 (Sudhaik et. al, 2018)when the surface of the MF material also exhibits electropositivity and homogeneous repulsion. However, in strongly alkaline aqueous solutions, the activity of Fe activated persulfate is reduced, which decreases the removal efficiency of the material in the photoreactive phase. Meanwhile, excessive adsorption can reduce the utilization of light by the samples (Ye et. al, 2019).
3.3.3. Effect of OTC concentration
Investigating the degradation efficiency of MF-50 at different antibiotic concentrations is a major step in testing the suitability of the material. In this study, the removal efficiency of MF-50 coupled with PS was evaluated at OTC concentrations of 30, 40, 50, 60, 70 and 100 mg/L. According to Fig. 5e, MF-50/PS/light has good removal efficiency for OTC from 30 to 70 mg/L. When the concentration of OTC increases to 100 mg/L, the removal efficiency decreases slightly. This can be attributed to the fact that high concentrations of OTC are more susceptible to light scattering, which reduces the number of photons captured by MF-50. Also the increased competition of OTC molecules is a reason for the low removal efficiency of high OTC concentrations (Jiang et. al, 2018).
3.3.4. Effect of PS concentration
The addition of PS has a non-negligible effect on the removal of OTC, so it is necessary to investigate the optimum amount of persulfate to be added. As shown in the Fig. 5f, the removal rate of OTC increases when the amount of PS increases from 0.1 to 0.5 g/L. This improvement is caused by increased concentrations of active species associated with the conversion of more PS molecules into free radicals (Pan et. al, 2020). However, when the PS continued to increase from 0.5 g/L to 1.0 g/L, the removal efficiency of OTC did not improve, which may be because excessive oxidant would clean up the generated free radicals, resulting in a lower removal efficiency (Heidarpour et. al, 2020).
3.4. Recyclability and stability
Recyclability and stability are two important indicators of the materials' practicality. After each use, the material was recovered with magnets, washed and dried, and then removed the OTC, and so on four times, and the total iron in the water samples at the end of each experiment was measured. As shown in Fig. 6a, after four cycles, the removal efficiency was only reduced by 4.29% and none of the leached iron exceeded 0.035 mg/L. Furthermore, in combination with XPS analysis of the used material (Fig. 3b), the results showed that the structure of the material remained virtually unchanged, demonstrating the excellent stability of the material. The determination of the hysteresis lines helps to characterize the magnetic properties of the material. According to Fig. 6b, MF-50 is paramagnetic with a saturation magnetic strength of 3.52 emu/g.
3.5. Mechanistic analysis
To investigate the role of free radicals in the degradation of oxytetracycline, trapping experiments were carried out. In our work, MeOH (1 mol/L), IPA (1 mol/L), BQ (1 mmol/L), NaCl (1 mmol/L) were used as scavengers for •SO4− + •OH, •OH, •O2−, h+, respectively. As shown in the Fig. 7a, the addition of IPA significantly reduced the removal efficiency of OTC, indicating that •OH was involved in the removal process of OTC. After adding MeOH, the removal efficiency of OTC was further reduced, indicating that •SO4− also played a role in the removal process of OTC. While the addition of BQ and NaCl had no effect on the removal of OTC. In this way, •SO4− and •OH were the main active substances for OTC degradation. To further verify the reactive oxygen species in AUN-2 photocatalytic system, ESR spin-trap measurement involved 5,5-dimethyl-1-pyrroline N-oxide (DMPO) in aqueous solution. Experiments were conducted in darkness, irradiated with visible light for 5 min. As shown in Fig. 7b, the DMPO •OH adduct could observe a four-wire ESR signal under light, while the DMPO-•SO4− could observea a six-wire ESR signal, indicating that MF-50/light could effectively activate the PDS to produce •SO4− and •OH radicals, so as to obtain excellent OTC removal efficiency. In contrast, there was no obvious peak during the dark reaction, which also further illustrated the coupling of photocatalysis with persulfate.
Combined with the analysis of free radicals, the charge transfer paths can be deduced (Fig. 8). According to the analysis results of the material energy band structure, Fe3O4 and MIL-101 combine to form a heterojunction structure with interleaved energy levels, so both photocatalysts may form type II heterojunction and Z-type heterojunction structures. MF-50 can generate electron leaps upon photoexcitation for both Fe3O4 and MIL-101. Since the conduction band potential of Fe3O4 is lower than that of O2/•O2− (-0.33 eV vs. NHE), and •O2− does not participate in the degradation of OTC according to the trapping experiments, the electrons on Fe3O4 are subsequently transferred to the conduction band of MIL-101, which activates persulfate to generate persulfate. At the same time, h+ on the valence band of Fe3O4 should migrate to the valence band of MIL-101 and convert H2O to •OH. This is because the valence band potential of Fe3O4 is lower than •OH/H2O (2.38 eV vs. NHE), and h+ is not the main active substance in the active substance analysis. In addition, comparing Fe3+/Fe2+ (0.77eV vs. NHE), the Fe(III) and Fe(II) cycles present during the reaction. This is also evidenced by a new peak of 710.6 eV in the XPS Fe2p pattern of the used material, which corresponds to Fe(II).