Efficient Photocatalytic Degradation of Tetracycline on the MnFe2O4/BGA Composite under Visible Light

In this work, the MnFe2O4/BGA (boron-doped graphene aerogel) composite prepared via the solvothermal method is applied as a photocatalyst to the degradation of tetracycline in the presence of peroxymonosulfate. The composite’s phase composition, morphology, valence state of elements, defect and pore structure were analyzed by XRD, SEM/TEM, XPS, Raman scattering and N2 adsorption–desorption isotherms, respectively. Under the radiation of visible light, the experimental parameters, including the ratio of BGA to MnFe2O4, the dosages of MnFe2O4/BGA and PMS, and the initial pH and tetracycline concentration were optimized in line with the degradation of tetracycline. Under the optimized conditions, the degradation rate of tetracycline reached 92.15% within 60 min, whereas the degradation rate constant on MnFe2O4/BGA remained 4.1 × 10−2 min−1, which was 1.93 and 1.56 times of those on BGA and MnFe2O4, respectively. The largely enhanced photocatalytic activity of the MnFe2O4/BGA composite over MnFe2O4 and BGA could be ascribed to the formation of type I heterojunction on the interfaces of BGA and MnFe2O4, which leads to the efficient transfer and separation of photogenerated charge carriers. Transient photocurrent response and electrochemical impedance spectroscopy tests offered solid support to this assumption. In line with the active species trapping experiments, SO4•− and O2•− radicals are confirmed to play crucial roles in the rapid and efficient degradation of tetracycline, and accordingly, a photodegradation mechanism for the degradation of tetracycline on MnFe2O4/BGA is proposed.


Introduction
As a therapeutic drug for humans and livestock, tetracycline (TC) is largely utilized to prevent disease and promote the growth of livestock. With the widespread application of tetracycline, severe environmental pollution occurs when a large amount of tetracycline is discharged with the metabolism of human beings and animals and with industrial sewage into the environment [1,2]. Therefore, it is imperative to develop efficient techniques to eliminate TC pollution. Currently, TC sewage can be treated by biological, physical, and chemical technology [3]. However, since TC has a wide range of antibacterial effects, it is difficult to eliminate its biological toxicity through the traditional biological degradation method. Moreover, the biological method is limited by a long treatment cycle and large space for a number of bulky facilities. As for the physical treatment method, it often leads to the incomplete degradation of TC. In contrast, chemical oxidation is proven to be facile, practical, and efficient for the degradation of TC because of its easy operation, low cost, high efficiency, etc. [4].
In recent years, advanced oxidation process (AOP) as a standard chemical technology has widely been applied in the control of water pollution [5], in which hydroxyl radical (•OH) and sulfate radical (SO 4 •− ) play a role of strong oxidants to mineralize organic pollutants. In comparison with •OH (1.7-2.7 V), SO 4 •− has the advantages of high oxidation potential (2.5-3.1 V) [6], a wide range of pH applications, and a long half-life. Therefore, SO 4 •− is more conducive than •OH to the rapid and efficient degradation of organic pollutants. Normally, SO 4 •− generates when peroxymonosulfate (PMS) is activated by electricity [7], heat [8], ultraviolet [9], ultrasonic [10], carbon materials [11], transition metals, and/or their compounds [12,13], in which the peroxide bond -O-O-in PMS breaks to form SO 4 •− and •OH [14,15]. Because of their relatively narrow band gap (1.5-2.5 eV), spinel ferrites are proper candidates for photocatalysts applicable in advanced oxidation processes [16,17]. In particular, MnFe 2 O 4 has been evidenced to be an effective photocatalyst for the degradation of various organic pollutants by virtue of its facile synthesis, high chemical activity, excellent stability, magnetic separation, and environmental friendliness [4, 17,18]. On the other hand, large-scale application of MnFe 2 O 4 in the field of organic pollutants degradation has been limited for its low conductivity, easy aggregation, poor adsorption performance, and quick combination of photogenerated electron(e − )-hole(h + ) pairs [19,20]. To date, researchers have made considerable progress in breaking these limitations through the construction of the composites of MnFe 2 O 4 nanoparticles and low-dimensional nanomaterials. Zhou et al. synthesized the MnFe 2 O 4 /β-CD (Beta cyclodextrin) composite, which could effectively activate PMS, and displayed quite good photocatalytic degradation performance towards 2,4-dichlorophenol in a broad pH range [21]. The MnFe 2 O 4 /CeO 2 composite synthesized by Okla et al. showed a large decrease in the recombination rate of photogenerated electron(e − )-hole(h + ) pairs, leading methylene blue dye to be efficiently degraded [22]. He et al. applied the SrCoO 3 /MnFe 2 O 4 /MoS 2 composite to the photocatalytic degradation of levofloxacin and confirmed the synergistic effect established between PMS and visible light [23].
Aiming at accelerating the transfer and separation of photogenerated carriers, graphene is often composited with photocatalysts to reduce the recombination rate of electron-hole pairs. It was reported that the MnFe 2 O 4 /rGO composite was superior to MnFe 2 O 4 as a photocatalyst for the degradation of methylene blue under a UV lamp [4] and that PMS could be effectively activated by graphene for the degradation of methyl violet, methyl orange, and methylene blue [24]. In comparison with graphene nanosheets, 3D graphene aerogel (GA) has the advantages of strong adsorption capacity and high specific surface area, which makes it an ideal material for preparing some composite photocatalysts. In the composites of Ag 3 PO 4 /GA [25], TiO 2 /RGA [26], and TiO 2 /MoS 2 /GA [27], the semiconducting photocatalysts are uniformly dispersed among the pores of GA, improving not only their photocatalytic performance but also their stability and sustainability. Moreover, the electronic structure of GA could be easily adjusted by the covalent modification of some other non-metal elements, and a synergistic effect between the doped GA and the semiconducting photocatalyst often appears in the process of photocatalysis. For example, Chanez et al. [28] reported the enhancement of visible-light photocatalytic pollutant degradation and hydrogen evolution with a 3D nitrogen-doped GA (N-GA), and Ren et al. [29] evidenced the rapid degradation of tetracycline on the spinel CoFe 2 O 4 nanoparticles anchored on N-GA. On the other hand, the BGA displayed a superior efficiency to GA towards the degradation of acridine orange under visible light irradiation [30], and we reported previously that a strong synergetic effect established between MnFe 2 O 4 and BGA during the degradation of rhodamine B on the MnFe 2 O 4 /BGA [31].
In this context, the magnetic MnFe 2 O 4 /BGA composite photocatalyst was synthesized and applied to activate PMS for the efficient removal of TC from its aqueous medium under visible light. Experimental results confirmed that the as-prepared MnFe 2 O 4 /BGA composite showed significantly enhanced photocatalytic activity than that of both MnFe 2 O 4 nanoparticles and BGA sheets for the degradation of TC. The excellent photocatalytic performance of the MnFe 2 O 4 /BGA composite could be ascribed to the type I heterojunction created on the 0D-2D point contact interface between MnFe 2 O 4 nanoparticles and BGA sheets, which led to the broad and strong visible light absorption and the high-efficient photogenerated carrier separation.

Materials Characterizations
The morphology of BGA, MnFe 2 O 4 , and MnFe 2 O 4 /BGA was observed by SEM and TEM. It can be seen from Figure 1a that the irregular MnFe 2 O 4 nanoparticles aggregate together, which is a limit to their catalytic performance. On the other hand, porous BGA is clearly observed in Figure 1b, whose average pore size is about 2.5 µm. It is no doubt that the pores of BGA are large enough to host MnFe 2 O 4 nanoparticles. As observed in Figure 1c, the MnFe 2 O 4 nanoparticles are evenly dispersed among BGA sheets, which is beneficial to the decrease in the agglomeration state and the increase in the specific surface area of MnFe 2 O 4 nanoparticles. The TEM image shown in Figure 1d confirms that MnFe 2 O 4 nanoparticles are closely wrapped by graphene sheets, which makes the MnFe 2 O 4 nanoparticles well dispersed. On the other hand, the BGA sheets form a conductive framework in the MnFe 2 O 4 /BGA composite for the photogenerated carriers, which is also helpful for the efficient separation of the photogenerated electrons and holes. and BGA sheets, which led to the broad and strong visible light absorption and the highefficient photogenerated carrier separation.

Materials Characterizations
The morphology of BGA, MnFe2O4, and MnFe2O4/BGA was observed by SEM and TEM. It can be seen from Figure 1a that the irregular MnFe2O4 nanoparticles aggregate together, which is a limit to their catalytic performance. On the other hand, porous BGA is clearly observed in Figure 1b, whose average pore size is about 2.5 µm. It is no doubt that the pores of BGA are large enough to host MnFe2O4 nanoparticles. As observed in Figure 1c, the MnFe2O4 nanoparticles are evenly dispersed among BGA sheets, which is beneficial to the decrease in the agglomeration state and the increase in the specific surface area of MnFe2O4 nanoparticles. The TEM image shown in Figure 1d confirms that MnFe2O4 nanoparticles are closely wrapped by graphene sheets, which makes the MnFe2O4 nanoparticles well dispersed. On the other hand, the BGA sheets form a conductive framework in the MnFe2O4/BGA composite for the photogenerated carriers, which is also helpful for the efficient separation of the photogenerated electrons and holes.  Figure 2a are the XRD patterns of powdered MnFe2O4, BGA, and MnFe2O4/BGA. For MnFe2O4, the sharp peaks observed at 8.18°, 29.82°, 35.17°, 42.60°, 52.83°, 56.41°, and 61.76° could be assigned to the (111), (220), (311), (400), (422), (511), and (440) lattice planes of spinel MnFe2O4 (JCPDS 10-0319), respectively. Accordingly, highly pure and crystalline MnFe2O4 is prepared by the co-precipitation method. Whereas BGA demonstrates two broad peaks at 24.3° and 43°, which are ascribed separately to the (002) and (100) lattice planes of BGA. It is worth noting that both characteristic peaks of spinel MnFe2O4 and BGA are observed in the XRD pattern of MnFe2O4/BGA, indicating that spinel MnFe2O4 particles are successfully composited with BGA sheets.  Figure 2a   As displayed in Figure 2b, MnFe2O4 demonstrates a characteristic Raman scattering peak at 616 cm −1 [32], and the D and G bands of BGA are separately observed at 1358 and 1589 cm −1 . Of interest to note is that all of these Raman scattering features clearly appear in the MnFe2O4/BGA composite, offering another support to the feasible preparation of the MnFe2O4/BGA composite. It is well known that the D and G bands of carbon nanomaterials are associated with the disordered carbons defects in graphene lattice and the vibration of the sp 2 hybridized carbon atoms in graphite, respectively. Usually, the intensity ratio of D-band to G-band (ID/IG) is applied to estimate the defects of carbon nanomaterials. For example, it can be seen from Figure 2b that the ID/IG of MnFe2O4/BGA (0.95) is higher than that of BGA (0.86), indicating that the former has more carbon defects than the latter [24]. These extra defects might be brought about by the mutual interaction between BGA and MnFe2O4 nanoparticles, which not only prevent the magnetic MnFe2O4 nanoparticles from aggregation but also act as active sites for the photocatalytic degradation of the TC.

Shown in
XPS was applied to investigate the chemical states of the elements in the MnFe2O4/BGA composite. The survey XPS spectrum shown in Figure 3a confirms the elemental composition of the MnFe2O4/BGA composite, i.e., Fe, Mn, B, C, and O. In the deconvoluted Fe 2p spectrum (Figure 3b), the peaks located at 725.4 and 711.1 eV are ascribed to Fe 2p 1/2 and Fe 2p 3/2 , respectively, along with two satellite ones at 733.4 and 719.7 eV [6], evidencing the presence of Fe 3+ . In the deconvoluted Mn 2p spectrum (Figure 3c), the peaks located at 653.3 and 641.5 eV are assigned to Mn 2p 1/2 and Mn 2p 3/2 , respectively, along with satellite ones at 644.9 eV, proving that the chemical state of Mn is Mn 2+ [23]. The deconvoluted B 1s spectrum (Figure 3d) could be fitted by the two subpeaks centered at 192.4 and 193.7 eV, which are ascribed separately to BC3 and BC2O/BCO2 [33,34]. The deconvoluted C 1s spectrum (Figure 3e  As displayed in Figure 2b, MnFe 2 O 4 demonstrates a characteristic Raman scattering peak at 616 cm −1 [32], and the D and G bands of BGA are separately observed at 1358 and 1589 cm −1 . Of interest to note is that all of these Raman scattering features clearly appear in the MnFe 2 O 4 /BGA composite, offering another support to the feasible preparation of the MnFe 2 O 4 /BGA composite. It is well known that the D and G bands of carbon nanomaterials are associated with the disordered carbons defects in graphene lattice and the vibration of the sp 2 hybridized carbon atoms in graphite, respectively. Usually, the intensity ratio of D-band to G-band (I D /I G ) is applied to estimate the defects of carbon nanomaterials. For example, it can be seen from Figure 2b that the I D /I G of MnFe 2 O 4 /BGA (0.95) is higher than that of BGA (0.86), indicating that the former has more carbon defects than the latter [24]. These extra defects might be brought about by the mutual interaction between BGA and MnFe 2 O 4 nanoparticles, which not only prevent the magnetic MnFe 2 O 4 nanoparticles from aggregation but also act as active sites for the photocatalytic degradation of the TC.
XPS was applied to investigate the chemical states of the elements in the MnFe 2 O 4 /BGA composite. The survey XPS spectrum shown in Figure 3a confirms the elemental composition of the MnFe 2 O 4 /BGA composite, i.e., Fe, Mn, B, C, and O. In the deconvoluted Fe 2p spectrum (Figure 3b), the peaks located at 725.4 and 711.1 eV are ascribed to Fe 2p 1/2 and Fe 2p 3/2 , respectively, along with two satellite ones at 733.4 and 719.7 eV [6], evidencing the presence of Fe 3+ . In the deconvoluted Mn 2p spectrum (Figure 3c), the peaks located at 653.3 and 641.5 eV are assigned to Mn 2p 1/2 and Mn 2p 3/2 , respectively, along with satellite ones at 644.9 eV, proving that the chemical state of Mn is Mn 2+ [23]. The deconvoluted B 1s spectrum (Figure 3d  In their N2 adsorption-desorption isotherms shown in Figure 4a, BGA, MnFe2O4, and MnFe2O4/BGA demonstrate typical IV isotherms with a type H3 hysteresis loop, indicative of the random distribution of pores and the interconnection of pores. Moreover, the calculated specific surface area of MnFe2O4/BGA (142.425 m 2 g −1 ) is much larger than that of BGA (108.874 m 2 g −1 ) or pure MnFe2O4 (79.068 m 2 g −1 ). Barrett-Joyner-Halenda (BJH) analyses shown in Figure 4b reveal that the MnFe2O4/BGA composite is of mesoporous structure with pore size mainly distributed around 15.66 nm and total pore volume of 0.594 cm 3 g −1 . In comparison with that of BGA, the average pore size of MnFe2O4/BGA is decreased to some extent, which might be a result of the interactions between MnFe2O4 nanoparticles and BGA sheets. The large specific surface area of MnFe2O4/BGA might promote the adsorption of contaminants and offer more reactive sites to accelerate the generation of free radicals [6]. In their N 2 adsorption-desorption isotherms shown in Figure 4a, BGA, MnFe 2 O 4 , and MnFe 2 O 4 /BGA demonstrate typical IV isotherms with a type H3 hysteresis loop, indicative of the random distribution of pores and the interconnection of pores. Moreover, the calculated specific surface area of MnFe 2 O 4 /BGA (142.425 m 2 g −1 ) is much larger than that of BGA (108.874 m 2 g −1 ) or pure MnFe 2 O 4 (79.068 m 2 g −1 ). Barrett-Joyner-Halenda (BJH) analyses shown in Figure 4b reveal that the MnFe 2 O 4 /BGA composite is of mesoporous structure with pore size mainly distributed around 15.66 nm and total pore volume of 0.594 cm 3 g −1 . In comparison with that of BGA, the average pore size of MnFe 2 O 4 /BGA is decreased to some extent, which might be a result of the interactions between MnFe 2 O 4 nanoparticles and BGA sheets. The large specific surface area of MnFe 2 O 4 /BGA might promote the adsorption of contaminants and offer more reactive sites to accelerate the generation of free radicals [6]. The magnetic behavior of MnFe2O4 and MnFe2O4/BGA was analyzed by a magnetic test system (VSM) at room temperature. As shown in Figure 5a, the saturation magnetization of MnFe2O4 is 76.45 emu g −1 , and that of MnFe2O4/BGA is 8.96 emu g −1 . Even though the saturation magnetization of MnFe2O4/BGA is largely decreased in comparison with that of MnFe2O4, it is still large enough to guarantee the effective separation of the photocatalyst from the liquid reaction system by an applied magnetic field (see the inset of Figure 5a). Thus, the MnFe2O4/BGA composite photocatalyst is reusable [17]. From an economic point of view, the recyclability of the catalyst is an important consideration for its practical applications.  The magnetic behavior of MnFe 2 O 4 and MnFe 2 O 4 /BGA was analyzed by a magnetic test system (VSM) at room temperature. As shown in Figure 5a, the saturation magnetization of MnFe 2 O 4 is 76.45 emu g −1 , and that of MnFe 2 O 4 /BGA is 8.96 emu g −1 . Even though the saturation magnetization of MnFe 2 O 4 /BGA is largely decreased in comparison with that of MnFe 2 O 4 , it is still large enough to guarantee the effective separation of the photocatalyst from the liquid reaction system by an applied magnetic field (see the inset of Figure 5a). Thus, the MnFe 2 O 4 /BGA composite photocatalyst is reusable [17]. From an economic point of view, the recyclability of the catalyst is an important consideration for its practical applications.  The magnetic behavior of MnFe2O4 and MnFe2O4/BGA was analyzed by a magnetic test system (VSM) at room temperature. As shown in Figure 5a, the saturation magnetization of MnFe2O4 is 76.45 emu g −1 , and that of MnFe2O4/BGA is 8.96 emu g −1 . Even though the saturation magnetization of MnFe2O4/BGA is largely decreased in comparison with that of MnFe2O4, it is still large enough to guarantee the effective separation of the photocatalyst from the liquid reaction system by an applied magnetic field (see the inset of Figure 5a). Thus, the MnFe2O4/BGA composite photocatalyst is reusable [17]. From an economic point of view, the recyclability of the catalyst is an important consideration for its practical applications.  The thermal stability and quantitative composition of the MnFe 2 O 4 /BGA composite were analyzed by TGA. It can be seen from Figure 5b that the MnFe 2 O 4 /BGA composite demonstrates a weight loss of 4.60 wt% between 30 and 290 • C, which is due to the evapo-ration of physical and chemical adsorbed moisture. The major weight loss of 60.50 wt% happens between 290 and 500 • C and is attributed to the combustion of BGA. The residues after 500 • C are ascribed to MnFe 2 O 4 nanoparticles, which are stable till the highest temperature of TGA measurement (800 • C). On the one hand, the relative content of MnFe 2 O 4 nanoparticles to BGA is basically consistent with that before the solvothermal process. On the other hand, the MnFe 2 O 4 /BGA composite is thermally stable to the temperature of 290 • C, satisfying the needs for practical applications.
The optical absorption property of photocatalysts is a crucial factor in determining their light-harvesting capacity. Shown in Figure 5c are the UV-visible-diffuse reflectance spectra of MnFe 2 O 4 and MnFe 2 O 4 /BGA. It is obvious that the MnFe 2 O 4 /BGA composite has much higher absorbance than the MnFe 2 O 4 nanoparticles in the UV-vis-NIR range. According to the Kubelka-Munk model, the direct band gaps of semiconducting materials could be estimated by the plot of (αhν) 2 versus (hν), where α stands for extinction coefficient, and hν is photon energy. As shown in Fig 5d,

Catalytic Performance
The experimental parameters, including the doping effect of BGA, MnFe 2 O 4 /BGA dosage, PMS dosage, initial pH, and initial TC concentration, were optimized. Shown in Figure 6a is a calibration curve of TC standard solutions, which displays the mass concentration dependence of the absorbance of TC measured at the maximum absorption wavelength (λ max = 357 nm). Accordingly, the concentration of TC in the reaction system was determined by its optical absorbance.
Experiments about the doping effect of BGA on the catalytic performance of MnFe 2 O 4 /BGA were conducted with the initial TC concentration, PMS dosage, composite catalyst dosage, and pH value of 20 mg L −1 , 0.1 mmol L −1 , 0.20 g L −1 , and 6, respectively. As shown in Figure 6b, with the change of n from 0.1 to 1.0, the removal rate of TC is initially increased and then decreased within 60 min, and the highest removal rate reaches 67.58% when n = 0.5. With the increase in the load of MnFe 2 O 4 nanoparticles on BGA sheets, more photogenerated charge carriers occur in the catalytic system. However, if BGA sheets are covered with too many MnFe 2 O 4 nanoparticles, the light absorption of the composite catalyst would be hindered because of the shortage of efficient specific surface of BGA to light, which limits the generation of photogenerated carriers and then reduces the photocatalytic activity of the composite catalyst.
Experiments about the MnFe 2 O 4 /BGA-0.5 dosage on the catalytic performance of MnFe 2 O 4 /BGA were carried out with the initial TC concentration, PMS dosage and pH value of 20 mg L −1 , 0.1 mmol L −1 , and 6, respectively. It can be seen from Figure 6c that the removal rate of TC increases monotonously with the enlargement of MnFe 2 O 4 /BGA dosage. When the dosage of MnFe 2 O 4 /BGA increase gradually, more and more SO 4 •− could be generated, which would contribute to more and more active sites for the degradation of TC. Of interest to note is that when the MnFe 2 O 4 /BGA dosages are 0.2 and 0.3 g L −1 , the removal rates of TC reach 92.15% and 95.10% within 60 min, respectively. On the other hand, much visible light would be scattered or reflected if too many MnFe 2 O 4 /BGA composites are dispersed in the catalytic reaction system. Thus, a dosage of 0.2 g L −1 was adopted for the MnFe 2 O 4 /BGA composite catalyst in this work. The effect of PMS dosage on TC degradation by MnFe2O4/BGA was investigated with an initial TC concentration of 20 mg L −1 , MnFe2O4/BGA dosage of 0.20 g L −1 , and pH value of 6. As displayed in Figure 6d, the removal rate of TC increases greatly when the dosage of PMS changes from 0.01 to 0.1 mmol L −1 , which reaches 92.15% at the PMS dosage of 0.1 mmol L −1 within 60 min. This is because the steady-state concentration of SO4 •− in the reaction system would be enhanced with the increase in the dosage of PMS, offering more collision probability between SO4 •− radicals and TC molecules. However, the removal rate of TC only increases from 92.15% to 92.92% within 60 min when the dosage of PMS changes from 0.1 to 1.0 mmol L −1 , which might be owing to the self-quenching reaction of SO4 •− radicals [6], the mutual quenching reaction of •OH and SO4 •− radicals and the quenching reaction of HSO5 − to SO4 •− radicals [38] (see Equations (1) and (2)). Thus, for the best utilization of the available SO4 •− radicals, the dosage of PMS was optimized to be 0.1 mmol L −1 . The effect of PMS dosage on TC degradation by MnFe 2 O 4 /BGA was investigated with an initial TC concentration of 20 mg L −1 , MnFe 2 O 4 /BGA dosage of 0.20 g L −1 , and pH value of 6. As displayed in Figure 6d, the removal rate of TC increases greatly when the dosage of PMS changes from 0.01 to 0.1 mmol L −1 , which reaches 92.15% at the PMS dosage of 0.1 mmol L −1 within 60 min. This is because the steady-state concentration of SO 4 •− in the reaction system would be enhanced with the increase in the dosage of PMS, offering more collision probability between SO 4 •− radicals and TC molecules. However, the removal rate of TC only increases from 92.15% to 92.92% within 60 min when the dosage of PMS changes from 0.1 to 1.0 mmol L −1 , which might be owing to the self-quenching reaction of SO 4 •− radicals [6], the mutual quenching reaction of •OH and SO 4 •− radicals and the quenching reaction of HSO 5 − to SO 4 •− radicals [38] (see Equations (1) and (2)). Thus, for the best utilization of the available SO 4 •− radicals, the dosage of PMS was optimized to be 0.1 mmol L −1 .
The initial pH dependence of TC degradation by MnFe 2 O 4 /BGA was checked out with an initial TC concentration of 20 mg L −1 , PMS dosage of 0.1 mmol L −1 , and MnFe 2 O 4 /BGA dosage of 0.20 g L −1 within 60 min. As can be seen from Figure 6e, the removal rate of TC gradually increases when the pH value changes from 2 to 6, and then largely decreases when the pH value changes from 6 to 10, which is owing to the reactions between H + /OH − and reactive radicals (•OH and SO 4 •− ). Nonetheless, the removal rate of TC maintains more than 85.00% when the pH value increases from 2 to 8, indicating that the MnFe 2 O 4 /BGA composite is applicable to TC degradation over a wide pH range.
The influence of initial TC concentration on the TC degradation by MnFe 2 O 4 /BGA was estimated with a PMS dosage of 0.1 mmol L −1 , MnFe 2 O 4 / BGA dosage of 0.20 g L −1 , and pH value of 6. As shown in Figure 6f, the removal rate of TC decreases with the increase of the initial TC concentration. The higher the initial TC concentration, the more TC molecules would be adsorbed on the surface of the MnFe 2 O 4 /BGA composite, which leads to the cover of active sites and limits the generation of free radicals. When the initial TC concentration is 20 mg L −1 , the removal rate of TC reaches as high as 92.15% after 60 min. This indicates that the MnFe 2 O 4 /BGA composite catalyst is an effective activator of PMS for the degradation of high-concentration TC.
To highlight the outstanding degradation ability of the PMS+MnFe 2 O 4 /BGA system towards TC, seven comparative experiments were carried out under the optimized conditions, i.e., PMS dosage 0.1 mmol L −1 , catalyst dosage 0.20 g L −1 , pH = 6, and initial TC concentration 20 mg L −1 . It can be seen from Figure 7a that the TC removal ratios reach 5.24%,30.73%, 28.56%, 47.53%, 55.14%, 59.03%, and 92.15%, respectively, in the seven designed systems. It is obvious that PMS or MnFe 2 O 4 /BGA alone demonstrates very limited degradation capacity towards TC in water and that BGA or MnFe 2 O 4 could not degrade TC efficiently in the presence of PMS. In contrast, the MnFe 2 O 4 /BGA shows the most effective removal of TC in the presence of PMS. It should be noted that the catalytic performance of the MnFe 2 O 4 /BGA catalyst is significantly better than that of the MnFe 2 O 4 /GA catalyst, confirming the importance of BGA. As electron-deficient centers, the doped B atoms in BGA could trap photogenerated electrons, which is beneficial for the transfer and separation of photogenerated carriers. Moreover, the electrons trapped by the B atom, as well as the structural defects caused by B doping, are the reaction sites for TC degradation [39]. In addition, BGA also plays multiple roles in the adsorption of TC, extending visible light absorption and suppressing the agglomeration of MnFe 2 O 4 nanoparticles. As indicated in Figure 7b, among the seven comparative experiments, MnFe 2 O 4 /BGA also is of the highest rate constant (4.1 × 10 −2 min −1 ), which is larger than the sum of that of MnFe 2 O 4 (1.27 × 10 −2 min −1 ) and that of BGA (9.85 × 10 −3 min −1 ). Therefore, it is assumed that a synergistic effect is established between MnFe 2 O 4 and BGA, which is responsible for the greatly enhanced photocatalytic activity of the MnFe 2 O 4 /BGA composite catalyst.
A comparison table of the photocatalytic degradation of TC on various catalysts in PMS advanced oxidation processes is shown in Table 1. It can be seen from Table 1 that the MnFe 2 O 4 /BGA composite displayed a degradation rate much higher than most of those reported in the literature. Co 3 O 4 /g-C 3 N 4 , for example, showed a lower degradation rate towards TC under the same experimental conditions. Even with much less amount of catalyst and PMS dosages, the MnFe 2 O 4 /BGA composite still exhibited a degradation rate compared with those of ZnFe 2 O 4 /Bi 2 O 2 CO 3 /BiOBr, CuWO 4−x /Bi 12 O 17 C l2 , or Bi 2 WO 6 /natural hematite. Though MoS 2 /Ag/g-C 3 N 4 , g-C 3 N 4 /FeWO 4 , and MCN/NCDs demonstrated higher degradation rates than the MnFe 2 O 4 /BGA composite, the data were collected with the participation of expensive metal, excess of PMS, and excess of both PMS and catalyst, respectively. Thus, the MnFe 2 O 4 /BGA composite achieved in this work makes it possible to eliminate tetracycline from wastewater in an economical way. A comparison table of the photocatalytic degradation of TC on various catalysts in PMS advanced oxidation processes is shown in Table 1. It can be seen from Table 1 that the MnFe2O4/BGA composite displayed a degradation rate much higher than most of those reported in the literature. Co3O4/g-C3N4, for example, showed a lower degradation rate towards TC under the same experimental conditions. Even with much less amount of catalyst and PMS dosages, the MnFe2O4/BGA composite still exhibited a degradation rate compared with those of ZnFe2O4/Bi2O2CO3/BiOBr, CuWO4−x/Bi12O17Cl2, or Bi2WO6/natural hematite. Though MoS2/Ag/g-C3N4, g-C3N4/FeWO4, and MCN/NCDs demonstrated higher degradation rates than the MnFe2O4/BGA composite, the data were collected with the participation of expensive metal, excess of PMS, and excess of both PMS and catalyst, respectively. Thus, the MnFe2O4/BGA composite achieved in this work makes it possible to eliminate tetracycline from wastewater in an economical way.   The reusability of the MnFe 2 O 4 /BGA composite catalyst was investigated by four successive cycling experiments. After every degradation of TC, MnFe 2 O 4 /BGA was collected by magnetic separation and subjected to the following recycling test. As displayed in Figure 8a, the removal ratio of TC declines from 92.15% to 74% in the second cycling experiment, indicative of a large loss of the catalyst activity. Nonetheless, from the second via the third to the fourth cycling experiments, the removal ratio of TC remains about 70%, confirming the cycle-to-cycle stability of the MnFe 2 O 4 /BGA composite catalyst. The apparent reaction rate constants displayed in Figure 8b offer direct support to the aforementioned results. Therefore, it is assumed that the loss of the catalyst activity might be owing to the mass loss and the leaching of metal ions for the MnFe 2 O 4 /BGA composite catalyst. ment, indicative of a large loss of the catalyst activity. Nonetheless, from the second via the third to the fourth cycling experiments, the removal ratio of TC remains about 70%, confirming the cycle-to-cycle stability of the MnFe2O4/BGA composite catalyst. The apparent reaction rate constants displayed in Figure 8b offer direct support to the aforementioned results. Therefore, it is assumed that the loss of the catalyst activity might be owing to the mass loss and the leaching of metal ions for the MnFe2O4/BGA composite catalyst.

Activation Mechanism
The scavenging assay was performed to evaluate the reactive oxidation species (ROS) in the photocatalysis of TC on MnFe2O4/BGA, AgNO3, AO, TBA, phenol, and p-BQ were applied as scavengers for quenching e − , h + , •OH, SO4 •− and O2 •− , respectively. It can be seen from Figure 9a that the main active species in the catalytic degradation of TC are O2 • and SO4 •− . On the other hand, DMPO was used as a spin-trapping agent for the radicals of O2 •− , •OH and SO4 •− in the reaction system. As shown in Figure 9b, the electron spins of DMPO-O2 •− , DMPO-•OH, and DMPO-SO4 •− were clearly detectable, offering another support to the existence of O2 •− oxidation reaction in addition to the conventional ones of •OH and SO4 •− free radicals.

Activation Mechanism
The scavenging assay was performed to evaluate the reactive oxidation species ( confirming the cycle-to-cycle stability of the MnFe2O4/BGA composite catalyst. The apparent reaction rate constants displayed in Figure 8b offer direct support to the aforementioned results. Therefore, it is assumed that the loss of the catalyst activity might be owing to the mass loss and the leaching of metal ions for the MnFe2O4/BGA composite catalyst.

Activation Mechanism
The scavenging assay was performed to evaluate the reactive oxidation species (ROS) in the photocatalysis of TC on MnFe2O4/BGA, AgNO3, AO, TBA, phenol, and p-BQ were applied as scavengers for quenching e − , h + , •OH, SO4 •− and O2 •− , respectively. It can be seen from Figure 9a that the main active species in the catalytic degradation of TC are O2 • and SO4 •− . On the other hand, DMPO was used as a spin-trapping agent for the radicals of O2 •− , •OH and SO4 •− in the reaction system. As shown in Figure 9b, the electron spins of DMPO-O2 •− , DMPO-•OH, and DMPO-SO4 •− were clearly detectable, offering another support to the existence of O2 •− oxidation reaction in addition to the conventional ones of •OH and SO4 •− free radicals.   Figure 10b, it can be seen that the arc radius of the Nyquist curve for MnFe 2 O 4 /BGA is smaller than that for MnFe 2 O 4 or BGA, which indicates a more effective separation of photogenerated e − -h + pairs and a higher efficiency of charge immigration across the electrode/electrolyte interface. Simultaneously, the photocurrent responses of MnFe 2 O 4 , BGA and MnFe 2 O 4 /BGA were also investigated by transient photocurrent tests. It can be seen from Figure 10c that the photocurrent of MnFe 2 O 4 /BGA is much larger than the sum of those for MnFe 2 O 4 and BGA, confirming the aforementioned synergistic effect established between MnFe 2 O 4 and BGA. Displayed in Figure 10d are the photoluminescence spectra of MnFe 2 O 4 , BGA, and MnFe 2 O 4 /BGA, which were excited by a laser with a wavelength of 250 nm. The room temperature luminescence peaks are observed at 420, 532, and 750 nm, respectively, in all the samples. The symmetrical peak at 750 nm might be owing to the third-order diffraction of the optical grating inside the monochromator. In comparison, the peaks at 420 and 532 nm correspond to the transitions associated with MnFe 2 O 4 and/or BGA, which result from the recombination of photogenerated electron-hole pairs and that of photogenerated holes with singly ionized oxygen vacancies [57,58], respectively. It is obvious that the photoluminescence of the MnFe 2 O 4 /BGA composite material is greatly reduced relative to those of MnFe 2 O 4 and BGA. This reduction is most likely due to the quenching effect caused by the charge transfer from MnFe 2 O 4 nanoparticles to the BGA sheet. Such charge transferled fluorescence quenching is indicative of the effective promotion of charge transfer and efficient suppression of the recombination of photogenerated electron-hole pairs [59].
trode/electrolyte interface. Simultaneously, the photocurrent responses of MnFe2O4, BGA and MnFe2O4/BGA were also investigated by transient photocurrent tests. It can be seen from Figure 10c that the photocurrent of MnFe2O4/BGA is much larger than the sum of those for MnFe2O4 and BGA, confirming the aforementioned synergistic effect established between MnFe2O4 and BGA. Displayed in Figure 10d are the photoluminescence spectra of MnFe2O4, BGA, and MnFe2O4/BGA, which were excited by a laser with a wavelength of 250 nm. The room temperature luminescence peaks are observed at 420, 532, and 750 nm, respectively, in all the samples. The symmetrical peak at 750 nm might be owing to the third-order diffraction of the optical grating inside the monochromator. In comparison, the peaks at 420 and 532 nm correspond to the transitions associated with MnFe2O4 and/or BGA, which result from the recombination of photogenerated electron-hole pairs and that of photogenerated holes with singly ionized oxygen vacancies [57,58], respectively. It is obvious that the photoluminescence of the MnFe2O4/BGA composite material is greatly reduced relative to those of MnFe2O4 and BGA. This reduction is most likely due to the quenching effect caused by the charge transfer from MnFe2O4 nanoparticles to the BGA sheet. Such charge transfer-led fluorescence quenching is indicative of the effective promotion of charge transfer and efficient suppression of the recombination of photogenerated electron-hole pairs [59]. graphene and the oxygen on the surface of MnFe2O4 (B…O-Fe) as well as that between the unreduced oxygen in graphene and the Fe on the surface of MnFe2O4 (C-O…Fe-O) [31]. BGA and MnFe2O4 form heterojunctions, which create a built-in electric field at their interface to broaden light absorption spectral range, enhance charge separation, inhibit charge recombination and accelerate the kinetic process. The data shown in Figures 5, 7 and 9 offer strong support for the formation of type-I heterojunctions at the interface between BGA and MnFe2O4. When the MnFe2O4/BGA composite is irradiated by visible light, electron-hole pairs are generated from the semiconducting BGA and MnFe2O4, respectively (see Equations (4) and (5)). Because the conduction band (CB) energy level of MnFe2O4 is higher than that of BGA, the photogenerated electrons would transfer from MnFe2O4 to BGA owing to the contact electric field. On the other hand, since the valence band (VB) energy level of MnFe2O4 is lower than that of BGA, the photogenerated holes would migrate from MnFe2O4 to BGA, forming a straddled band alignment (see Equations (6) and (7)). Though both the photogenerated electrons and holes transfer from MnFe2O4 to BGA across type-I heterojunctions, the difference in the migration rates of electrons and holes is responsible for the efficient separation of the photogenerated charge carriers. Because of the unique configuration of BGA-loaded MnFe2O4 nanoparticles, the photogenerated charge carriers in BGA are exposed to the liquid phase and are effectively involved in the photocatalytic degradation of TC. Accordingly, the reactive radicals of O2 •− , SO4 •− , and •OH are formed via the processes illustrated in Equations (8)-(10), respectively. The photogenerated electrons react separately with PMS and dissolved oxygen in the water to produce SO4 •− and O2 •− , while the photogenerated holes react with H2O to yield •OH. As indicated in Equation (11), the excellent photocatalytic performance of MnFe2O4/BGA towards the When the MnFe 2 O 4 /BGA composite is irradiated by visible light, electron-hole pairs are generated from the semiconducting BGA and MnFe 2 O 4 , respectively (see Equations (4) and (5)). Because the conduction band (CB) energy level of MnFe 2 O 4 is higher than that of BGA, the photogenerated electrons would transfer from MnFe 2 O 4 to BGA owing to the contact electric field. On the other hand, since the valence band (VB) energy level of MnFe 2 O 4 is lower than that of BGA, the photogenerated holes would migrate from MnFe 2 O 4 to BGA, forming a straddled band alignment (see Equations (6) and (7)). Though both the photogenerated electrons and holes transfer from MnFe 2 O 4 to BGA across type-I heterojunctions, the difference in the migration rates of electrons and holes is responsible for the efficient separation of the photogenerated charge carriers. Because of the unique configuration of BGA-loaded MnFe 2 O 4 nanoparticles, the photogenerated charge carriers in BGA are exposed to the liquid phase and are effectively involved in the photocatalytic degradation of TC. Accordingly, the reactive radicals of O 2 •− , SO 4 •− , and •OH are formed via the processes illustrated in Equations (8)-(10), respectively. The photogenerated electrons react separately with PMS and dissolved oxygen in the water to produce SO 4

Materials
All of the following chemicals were of analytical grade and used without further purification:

Preparation of the MnFe 2 O 4 /BGA Composite
As illustrated in Figure 12, MnFe 2 O 4 was synthesized by the co-precipitation of MnCl 2 ·4H 2 O and FeCl 3 ·6H 2 O with a molar ratio of 1:2 at pH = 12 [4], whereas graphene oxide (GO) was prepared from graphite powder by modified Hummers method [60]. The MnFe 2 O 4 /BGA composite was achieved through a hydrothermal procedure. A mixture of 140 mg GO, a certain amount of MnFe 2 O 4 (the mass ratios of MnFe 2 O 4 /GO are 0.1, 0.3, 0.5, 0.7 and 1.0, respectively), 50 mL deionized water, 20 mL ethylene glycol, and 280 mg H 3 BO 3 were mechanically stirred for 30 min and then transferred to a Teflon liner stainless steel autoclave. After hydrothermal treatment at 180 • C for 12 h, the product was immersed in an aqueous solution of ethanol (10%) for 12 h, followed in water for another 12 h, and finally subjected to a freeze-dried treatment for 48 h. Such obtained samples were named named MnFe2O4/BGA-n (n = 0.1, 0.3, 0.5, 0.7, or 1.0) according to the mass ratio of MnFe2O4/GO. In this context, MnFe2O4/BGA refers to MnFe2O4/BGA-0.5 without a specific statement. The MnFe2O4/GA composite and BGA were synthesized by the same procedure without the addition of H3BO3 or MnFe2O4, respectively.

Degradation of TC on the MnFe2O4/BGA Composite
The catalytic activity of the MnFe2O4/BGA composite catalyst was evaluated by photocatalytic degradation of TC. First, a certain amount of catalyst was added to 100 mL of 20 mg L −1 tetracycline solution. Then, after adjustment of the initial pH with 0.1 mol L −1 NaOH or 0.1 mol L −1 H2SO4 solutions, the reaction mixture was mechanically stirred for 10 min to ensure that the adsorption equilibrium of TC on the composite catalyst was reached. When a certain amount of PMS solution was injected, the reaction mixture was irradiated by a 300 W xenon lamp coupled with a 400 nm cut-off filter. At every ten minutes, 3 mL of the reaction mixture was taken out and filtered with 0.22 µm membrane. The concentration of residual TC in the filtrate was determined by its optical absorbance at 357 nm, and the removal rate of TC was calculated by the following formula.
where C0 is the initial concentration of TC, and C is the residual concentration of TC at a specific sampling time.
The kinetic rate of TC degradation on the MnFe2O4/BGA composite catalyst was calculated in line with general pseudo-first-order kinetics.
−ln = kt (13) where C0 is the initial concentration of TC, C is the residual concentration of TC at reaction time (t), and k is the pseudo-first-order rate constant.

Characterization Methods
Investigation on the phase and element compositions was carried out on an X-ray diffractometer (XRD, D8 Advance, Bruker, Berlin, Germany) and an X-ray photoelectron spectroscopy (XPS, KRATOS, Stratford, UK), respectively. In comparison, morphology observation was conducted on a scanning electron microscopy (SEM, QUANTA 200S, FEI, Eindhoven, The Netherlands) and a transmission electron microscope (TEM, JEM2100, JEOL, Tokyo, Japan). Raman scattering data were collected by a Jobin Yvon (Palaiseau, France) HR 800 micro-Raman spectrometer with 458 nm excitation. Fourier transforms infrared spectra (FT-IR) were recorded by a Perkin Elmer (Waltham, MA, USA) Spectrum 100FT-IR spectrometer with background correction by referring to KBr pellets.

Degradation of TC on the MnFe 2 O 4 /BGA Composite
The catalytic activity of the MnFe 2 O 4 /BGA composite catalyst was evaluated by photocatalytic degradation of TC. First, a certain amount of catalyst was added to 100 mL of 20 mg L −1 tetracycline solution. Then, after adjustment of the initial pH with 0.1 mol L −1 NaOH or 0.1 mol L −1 H 2 SO 4 solutions, the reaction mixture was mechanically stirred for 10 min to ensure that the adsorption equilibrium of TC on the composite catalyst was reached. When a certain amount of PMS solution was injected, the reaction mixture was irradiated by a 300 W xenon lamp coupled with a 400 nm cut-off filter. At every ten minutes, 3 mL of the reaction mixture was taken out and filtered with 0.22 µm membrane. The concentration of residual TC in the filtrate was determined by its optical absorbance at 357 nm, and the removal rate of TC was calculated by the following formula.
where C 0 is the initial concentration of TC, and C is the residual concentration of TC at a specific sampling time.
The kinetic rate of TC degradation on the MnFe 2 O 4 /BGA composite catalyst was calculated in line with general pseudo-first-order kinetics.

−ln
where C 0 is the initial concentration of TC, C is the residual concentration of TC at reaction time (t), and k is the pseudo-first-order rate constant.

Characterization Methods
Investigation on the phase and element compositions was carried out on an X-ray diffractometer (XRD, D8 Advance, Bruker, Berlin, Germany) and an X-ray photoelectron spectroscopy (XPS, KRATOS, Stratford, UK), respectively. In comparison, morphology observation was conducted on a scanning electron microscopy (SEM, QUANTA 200S, FEI, Eindhoven, The Netherlands) and a transmission electron microscope (TEM, JEM2100, JEOL, Tokyo, Japan). Raman scattering data were collected by a Jobin Yvon (Palaiseau, France) HR 800 micro-Raman spectrometer with 458 nm excitation. Fourier transforms infrared spectra (FT-IR) were recorded by a Perkin Elmer (Waltham, MA, USA) Spectrum 100FT-IR spectrometer with background correction by referring to KBr pellets. Thermogravimetric analysis differential thermal analysis (TG-DTA) was performed at a heating rate of 10 • C min −1 in the air with an EXSTAR TG/DTA 7300 thermogravimetric analyzer (Hitachi, Tokyo, Japan). Brunauer-Emmett-Teller (BET) surface area collected by N 2 adsorption-desorption method at 77 K on an ASAP 2010M analyzer (Micromeritics, Norcross, GA, USA). The magnetic behavior of MnFe 2 O 4 /BGA was recorded with an MPMS-3 superconducting quantum interference device (SQUID) magnetometer (Quantum Design, San Diego, CA, USA) in the applied field range of ±20 kOe at room temperature. The UV-vis diffuse reflectance spectra (DRS) were recorded with a UV-2550 spectrophotometer (Shimadzu, Tokyo, Japan). The optical absorbance of the TC solution was measured by a UV-3600 spectrophotometer (Shimadzu, Tokyo, Japan). Electron spin resonance (ESR) spectra were recorded with a Bruker N500 ESR spectrometer (Bruker, Rheinstetten, Germany). Cyclic voltammetry (CV), electrochemical impedance spectroscopy (EIS) and the transient photocurrent-time curves (I-t) were conducted on an SP-300 electrochemical workstation (Bio-Logic, Seyssinet-Pariset, France) in a three-electrode system with a sheet of platinum, Ag/AgCl electrode, and 0.5 mol L −1 Na 2 SO 4 as a counter electrode, reference electrode and electrolyte, respectively. The working electrode is an FTO plate (1 cm 2 ) homogeneously pasted with 20 mg of the composite catalyst. Photoluminescence spectra were recorded by the Edinburgh FLS980 fluorescence and phosphorescence spectrometer.

Conclusions
In summary, the MnFe 2 O 4 /BGA composite has been prepared by hydrothermal method. Homogenous dispersion of MnFe 2 O 4 nanoparticles and effective exfoliation of BGA sheets are realized simultaneously, leading MnFe 2 O 4 nanoparticles anchored on BGA sheets to form intimate type-I heterojunctions owing to the chemical interactions between MnFe 2 O 4 nanoparticles and BGA sheets. In comparison with pure MnFe 2 O 4 nanoparticles or BGA, the as-prepared MnFe 2 O 4 /BGA composite displays significantly enhanced photocatalytic activity for the degradation of high-concentrated TC under visible light irradiation, which could be mainly attributed to the type-I heterojunction induced efficient separation of photogenerated charge carriers excited from both MnFe 2 O 4 and BGA as well as the extended visible light response. It is no doubt that this work will provide useful information for the development of a wide-spectral-responsive photocatalyst based on type I band alignment related to the narrow band gap of BGA.
Moreover, a series of AOS quenching experiments evidenced that the active radicals of