Enhanced catalytic degradation of amoxicillin with TiO2–Fe3O4 composites via a submerged magnetic separation membrane photocatalytic reactor (SMSMPR)

A novel photo-Fenton catalytic system for the removal of organic pollutants was presented, including the use of photo-Fenton process and a submerged magnetic separation membrane photocatalytic reactor (SMSMPR). We synthesized TiO2–Fe3O4 composites as the photocatalyst and made full use of the magnetism of the photocatalyst to realize the recollection of the catalyst from the medium, which is critical to the commercialization of photocatalytic technology for wastewater treatment. The photo-Fenton performance of TiO2–Fe3O4 is evaluated with amoxicillin trihydrate (AMX) as a target pollutant. The results indicate that the TiO2–Fe3O4/H2O2 oxidation system shows efficient degradation of AMX. Fe3O4 could not only enhance the heterogeneous Fenton degradation of organic compounds but also allow the photocatalyst to be magnetically separated from treated water. After four reaction cycles, the TiO2–Fe3O4 composites still exhibit 85.2% removal efficiency of AMX and show excellent recovery properties. Accordingly, the SMSMPR with the TiO2–Fe3O4 composite is a promising way for removing organic pollutants.


Introduction
The penetration of pharmaceuticals into the environment has drawn considerable attention in recent years. [1][2][3] Several effective alternative water treatment technologies have been considered in recent studies, including adsorption, 4,5 membrane ltration, 6-8 coagulation 9 and advanced oxidation technologies. [10][11][12] Photocatalytic environmental remediation has always been the research focus since the breakthrough of water splitting was covered by Fujishima and Honda in 1972. 13,14 Various photocatalysts have been used in the photocatalysis process, among which titanium dioxide (TiO 2 ) is the most widely studied and applied owing to its advantages of strong oxidizing ability, chemical stability, nontoxicity, low cost and high hydrophilicity. 13,15 The photo-Fenton process is a typical combination of two kinds of advanced oxidation process (AOP) and shows high oxidative removal efficiency of organics because of the highly enhanced generation of reactive hydroxyl radicals ($OH). [16][17][18] However, one of the major drawbacks, separation and recycling of catalyst particles from large quantities of water, needs further cost and restrains the practical application of photocatalytic process. Accordingly, many studies were trying to explore an economic way to get the catalysts back. The ternary magnetic composite of Fe 3 O 4 @TiO 2 /SiO 2 was prepared and used to remove Rhodamine B from wastewater. Results indicated that the Fe 3 O 4 @TiO 2 /SiO 2 showed high photocatalytic activity and most importantly, it was recyclable. 19 Fan et al. coated Fe 3 O 4 / SiO 2 magnetic core with titania using hydrothermal synthetic method. With external magnetic eld, the Fe 3 O 4 /SiO 2 /TiO 2 magnetic nanocomposites could be separated from the suspension successfully. TiO 2 decorated with Fe 3 O 4 could generate greater photocatalytic activity as Fe 3+ can occupy both of the electron capture position and hole capture position, resulting in the decrease of electron-hole pair recombination of TiO 2 . However, aggregation of TiO 2 particles can inuence the optical properties and photoactivity of catalyst, 20 meaning the necessary of preventing the particles from aggregation during the photocatalysis process.
Herein, we developed a special combination of TiO 2 -Fe 3 O 4 catalysts, photo-Fenton process and SMSMPR. The incorporation of Fe 3 O 4 can improve the catalytic activity of TiO 2 (based on the energy level theory), and endowed TiO 2 with excellent reusability with the help of external magnetic eld. 19 The two associated oxidation process, photocatalysis and Fenton oxidation can ensure the degradation of organics in an efficient way. The well-designed submerged magnetic separation membrane photocatalytic reactor can realize the waste water treatment and the separation of catalyst. The aeration in the reactor during the photocatalysis process will alleviate the agglomeration of TiO 2 particles, which is essential to guarantee the specic surface area of photocatalyst. Also, the built-in backwashing treatment inside the SMSMPR could enhance the self-purication ability of membranes. We also evaluated the main factors inuencing the removal of organics, tried to nd out the optimum reaction condition and explored the possibility of cyclic utilization of the reaction system.

Synthesis of TiO 2 -Fe 3 O 4
TiO 2 -Fe 3 O 4 particles were synthesized by a facile hydrothermal synthetic method. 21 Firstly, 200 mL mixture of ethanol and water (v:v ¼ 1 : 1) was prepared. And different amounts of TiO 2 and 5 mM Fe(NO 3 ) 3 $9H 2 O were distributed in above solution by ultrasonic treatment for 3 h. Then it was dried at 60 C, ground to powder and placed into a beaker, which was put in a 500 mL Teon-lined autoclave. 60 mL ammonia solution was added into the autoclave beforehand and then sealed aer the beaker was introduced. A drying oven was used to maintain a constant temperature for alkaline treatment in autoclave (180 C for 12 h). Aerwards, the obtained samples were annealed at 200 C for 5 h and the obtained powder was distributed in 100 mL EG by ultrasonic treatment for 3 h, aer which the mixture was transferred to a 500 mL Teon-lined autoclave and maintained at 180 C for another 12 h. Finally, the obtained precipitates were washed three times with ethanol and distilled water and dried at 60 C in vacuum. The nal as-prepared products contained 10, 15, 20 and 25 wt% Fe 3 O 4 , respectively.

Characterization
Scanning electronic microscope (SEM, JSM 7800F), transmission electron microscope (TEM) and high resolution transmission electron microscope (HRTEM, JEM-2100F) images were obtained to characterize the morphological features of samples. X-ray diffraction (XRD) was carried out using a Bruker D8 Advance X-ray diffractometer with Cu-Ka radiation over a scan rate of 10 min À1 in the 2q range from 5 to 80 . A Tensor 27 Fourier-transform infrared spectroscopy (FTIR) spectrometer (Nicolet 6700) was used to record the FTIR spectra of catalyst. Xray photoelectron spectroscopy (XPS) analysis (PerkinElmer PHI 5000C, AlKa) of TiO 2 -Fe 3 O 4 nanocomposites was also performed.

Photo-Fenton performance using SMSMPR
The photo-Fenton process was performed with the SMSMPR and a schematic illustration of the SMSMPR for organic contaminant removal was depicted in Fig. 1. In this study, AMX (30 mg L À1 ) was chosen as the model organic contaminant. The concentration of AMX was measured by the total organic carbon (TOC) using a total organic carbon analyzer (multi 3100, Analytik Jena, Germany). Firstly, a peristaltic pump was utilized to pump AMX from the feed tank into an stainless steel container (L Â W Â H ¼ 200 mm Â 200 mm Â 500 mm) at a ow rate of 100 mL min À1 , which accommodated low-pressure mercury lamps (TUV, Philips), perforated aeration pipes, the hollow ceramic membranes (150 mm in length, 120 mm in width, and 4 mm in thickness), a powerful magnet connected to an iron rod and the mixture of organic solution and catalyst. Low-pressure mercury lamps (LPML; Philips) were placed around the ceramic membrane for photocatalysis and the light intensity quantitatively of a 100 W LPML was 1200 mW cm À2 . The perforated aeration pipes, connected to an air compressor, was placed at the bottom of the reactor to produce dissolved oxygen-rich micro-porous bubbles, which can keep the catalyst from gathering and settling down to the bottom. Aer the wastewater was successfully pumped into the reactor, the catalyst was dosed in and the air compressor was opened to uidize the catalysts. The low-pressure mercury lamps were switched on and the photocatalysis began aer 45 min, during which time the pharmaceutical solution and catalyst can be well-mixed and reach the equilibrium. As for the ceramic membranes, they were used for the separation of effluents from the catalyst slurry. Aer the photo-Fenton process nished, another peristaltic pump was used to pump the effluent out. Furthermore, the ceramic membranes were connected to the air compressor and can be used for backwashing, making the cyclic utilization of the system possible. Finally, an external magnetic eld, provided by a powerful magnet connected to an iron rod was also employed to realize the recollection of catalysts.

Results and discussion
Microimages of chemicals can provide its visual and intuitive morphology. The morphologies of photocatalysts were evaluated by SEM and TEM, and the corresponding images are presented in Fig. 2. Based on Fig. 2a and c, pure TiO 2 had irregularly spherical shape with average diameter of 40 nm and the TiO 2 particles were uniformly distributed in the plane. Aer combined with Fe 3 O 4 ( Fig. 2b and d), the nanoparticles showed obvious aggregation effect and smaller dimension, representing the presence of Fe 3 O 4 in the composites and interaction between TiO 2 and Fe 3 O 4 . Actually, during the formation of TiO 2 / Fe 3 O 4 composites, the positively charged Fe 3+ and negatively charged TiO 2 can be attracted by each other and connected in the beginning. As the temperature becoming higher, the evaporated ammonia reacted with Fe 3+ to produce Fe(OH) 3 on the surface of TiO 2 and the Fe(OH) 3 were in situ formed on TiO 2 . Aer the annealed treatment at 200 C for 5 h, the obtained Fe(OH) 3  stretching vibration of OH and surface absorbed water. 19,24,25 Compared with FTIR spectra of pure TiO 2 , the peaks displayed in the FTIR spectra of TiO 2 /15 wt% Fe 3 O 4 composites arose from not only OH group but also Fe-O vibration at 1087 cm À1 . 26 As displayed in Fig. 4 XPS measurements were carried out to analyse the compositional and chemical states of samples and the results are shown in Fig. 5. Fig. 5a presents XPS spectrum of pure TiO 2 and TiO 2 /15 wt% Fe 3 O 4 composites. The full-scale XPS spectra indicated that there were elements O, Ti and C in TiO 2 , while TiO 2 /15 wt% Fe 3 O 4 composites showed additional peaks of Fe 2p. The peak positions of Fe 2p 3/2 and Fe 2p 1/2 were 709.7 eV and 723.5 eV, respectively, which coincided with the previous literature. 28 Actually, Fe 3 O 4 should be described as FeO$Fe 2 O 3 , so the peaks are the comprehensive performance of Fe 2+ and Fe 3+ . As for O 1s (Fig. 5c), the peak of pure TiO 2 shied from 529.2 eV to 528.7 eV for TiO 2 /15 wt% Fe 3 O 4 , implying the interaction between TiO 2 and Fe 3 O 4 .
The removal efficiency of AMX under different conditions was investigated and the results were depicted in Fig. 6. As can be seen from the results, only 3.9% of the organic matter was removed even aer 100 min in the absence the activation of Fe 3 O 4 , while the removal efficiency of AMX can increase to 85.0% with 15 wt% Fe 3 O 4 dosage (Fig. 6a). Then it slightly decreases to 80.0 and 60.5% as the Fe 3 O 4 content increases to 20% and 25%, respectively. According to Table 1, the BET surface areas of catalysts decrease from 59.5 to 24.8 m 2 g À1 as the loading amount of Fe 3 O 4 increases from 0 to 25%, which is almost independent of the removal efficiency of AMX.  29 Apparently, almost no obvious decrease of AMX was observed if there was no H 2 O 2 , which is essential in the oxidation of organics in photo-Fenton process (Fig. 6b). Within the rst 30 minutes, more H 2 O 2 corresponded to higher degradation rates. Aer 30 minutes, it showed the same trend when the concentration of H 2 O 2 was below 24 mM. Undoubtedly, the increase of H 2 O 2 resulted in the increase of $OH, and thus the improvement in Fenton oxidation process. However, further increasing in H 2 O 2 dosage from 24 to 30 mM, resulted in slight decrease in AMX removal indeed. This observation could be ascribed to the mechanism that excessive H 2 O 2 could act as a self-scavenger for OH, following eqn (1) and (2), which leads to the greatly decrease of $OH generation. 30 The initial pH of solution plays a vital role in chemical reaction by affecting the charge and other physicochemical property of substance in the mixture. The effect of pH on the photo-Fenton degradation of AMX was evaluated at pH values in the range of 1.01-7.45 (Fig. 6c). It is generally recognized that homogeneous Fenton process happens in acidic conditions and it came to the same conclusion in this photo-Fenton system. 30,31 The AMX removal increased from 19.1% to 66.6% when the initial pH value decreased from 7.45 to 4.79. With further lower pH, more AMX was degraded and reached the highest at pH 2.84. This can be explained by the scramble of Fe 3+ between OH À and H 2 O 2 . In other words, more Fe 3+ will interact with OH À rather than H 2 O 2 when pH is higher, leading to lower oxidation efficiency. Additionally, the LPML as the UV radiation source was used in the photoreactor. The light intensity in the  Paper SMSMPR was investigated to evaluate the kinetics of the photo-Fenton reaction via the electron hole formation, separation, and recombination rates (Fig. 6d). It was obvious that the concentration of AMX showed only slight decrease in the dark (light intensity ¼ 0 W), indicating the limited adsorption capacity of TiO 2 /15 wt% Fe 3 O 4 to AMX. At a low light intensity in  L À1 (for b, c and d), and pH ¼ 2.84 (for a, b and d). the reactor (<200 W), the reaction rate increased remarkably with increasing light intensity, as the generation of electrons and holes was the predominant process. With further increase of light intensity (>200 W), the reaction rate increased slightly, owing to the high electron hole recombination. Meanwhile, electrons may have easily transferred from the catalyst to oxygen under the higher irradiation intensity, resulting in the generation of $O 2 À , which is the rate limiting step for larger TiO 2 particles. Therefore, light intensity of 200 W is the suitable operating condition for the SMSMPR.
In order to reveal the charge transfer process between TiO 2 and Fe 3 O 4 , AMX degradation mechanisms by TiO 2 /15 wt% Fe 3 O 4 composite were further explored. First, the trapping experiments were conducted in the minied system. Specically, 0.02 g photocatalysts were added into 50 mL AMX solution with a concentration 30 mg L À1 in the 100 mL photocatalysis reactor. EDTA-2Na (10 mM) and tert-butanol (TBA, 10 mM) were used as a hole scavenger and a hydroxyl radical scavenger ($OH), respectively (Fig. 7). The addition of a scavenger of holes (EDTA-2Na) caused a change in the photodegradation of AMX (50.9%). The degradation of AMX was signicantly inhibited in the presence of TBA (15.2%). Thus, it is believed that $OH and h + should be the main active species in the photo-Fenton degradation of AMX process.
The mechanism of high AMX degradation in this photo-Fenton system was revealed in Fig. 8. In the TiO 2 /15 wt% Fe 3 O 4 composite, excited electrons in the TiO 2 rapidly transfer to the Fe 3 O 4 . The quick separation of photogenerated electron-hole pairs can effectively reduce the h + /e À pairs recombination, which contributes to AMX degradation by the h + leaving in the valence band (route 1, Fig. 8). Meanwhile, introduction of Fe 3 O 4 provided an additional $OH generation pathway for AMX degradation (route 2, Fig. 8). The redox potential of Fe(II)/Fe(III) in Fe 3 O 4 cycle is effective to activate H 2 O 2 for the generation of $OH. More importantly, the photogenerated electron trapped by Fe 3 O 4 could facilitate the reduction process of Fe(III) to Fe(II) and then enhance the Fe(III)/ Fe(II) cycle during the Fenton process. Therefore, the easier Fe(III)/ Fe(II) cycle in TiO 2 /15 wt% Fe 3 O 4 composite contributed to the higher activity and stability for AMX degradation.
To examine the stability and repeatability of catalyst and SMSMPR, cycle experiments were carried out. Aer each cycle, an external magnetic eld was imposed to the reactor and the photocatalyst was reclaimed, washed and used in next cycle. Fig. 9a presented the AMX degradation rates of four cycles, with the reaction time 100 min for each cycle. Except the rst run (maybe because some loosely-combined catalyst leached), there was not signicantly declined degradation ability of catalytic composites with more runs and the removal efficiency of AMX was as high as 85.3% even aer 4 cycles. We also used FTIR, XRD and HRTEM to evaluate the difference between pristine and used TiO 2 /15 wt% Fe 3 O 4 . There was no obvious difference and the characteristic peaks of OH and Fe-O can be seen in both of the FTIR spectra (Fig. 9b). Fig. 9c illustrated that TiO 2 / 15 wt% Fe 3 O 4 sample aer 4 reaction cycles maintained morphology, lattice parameters and crystallinity similar to those of fresh samples. There is no obvious morphology change from HRTEM images (Fig. 9d). These results reveal the high material stability of TiO 2 /15 wt% Fe 3 O 4 and can be used as an environmentally friendly catalyst for photo-Fenton reaction.
Subsequently, SEM was utilized to investigate the surface change of membrane in the SMSMPR and the results were delineated in Fig. 10. As can be seen, the raw ceramic membrane had rough surface and many micron-scale pores on the surface. Aer one cycle experiment, the TiO 2 -Fe 3 O 4 particles were thickly deposited on the surface and even inside the pores of the membrane. However, most of the nanoparticles went away and the membrane recovered clean aer backwashing treatment and exposure to the external magnetic eld.
It is reported that the deposition of solid and pollutant can increase the trans-membrane pressure, lower the water ux and raise the cost of membrane technology. 32 We also examined the trans-membrane pressure change during the cycle experiments   and the result is given in Table 2. The initial trans-membrane pressure was measured aer the cycle experiment started for 5 min and the nal trans-membrane pressure was measured aer the cycle experiment started for 100 min. As can be observed, the trans-membrane pressure was lied much higher aer one single cycle run. For example, the initial transmembrane pressure was only 0.47 kPa, but the nal was as high as 0.99 kPa aer the second cycle started for 100 min.
Although backwashing treatment can recover the transmembrane pressure to some extent, the recovery effect was quite limited. By comparison, the use of external magnetic eld maintained the trans-membrane pressure within a relatively low level even aer 4 cycles. Recently, researchers have used different kinds of magnetic photocatalysts to degrade organic pollutants. Chang et al.   phenol. 33 However, they utilized photocatalysis only instead of photo-Fenton process and then the degradation efficiency of organics reached maximum with the catalyst amount as much as 3 g L À1 . As described above, photo-Fenton system has relatively higher degradation rate of pollutants under acidic conditions in most of the literature and we got the same result in the paper. Actually, researchers have been trying to overcome this drawback by adding chelating agents, such as ethylenediamine-N,N 0 -disuccinic acid (EDDS), ethylenediaminetetraacetic acid (EDTA) and nitrilotriacetic acid (NTA) in recent years, 34 which could be adopted by the system in the paper. Another study also carried out recycling experiments to measure the repetitive use of the catalyst. With even better performance in reusability, the asprepared Fe 3 O 4 @TiO 2 /SiO 2 photocatalyst showed just a slight drop (from 94.5% to 90.1%) in removal efficiency aer six cycles. 19 This means that the repeatability of the SMSMPR system may be improved if stronger interaction between Fe 3 O 4 and TiO 2 was formed and further study is needed.

Conclusions
In summary, a novel photocatalysis reactor was successfully built and employed in the photo-Fenton process. Fe 3 O 4 grown on a TiO 2 -Fe 3 O 4 composite not only enhances the heterogeneous Fenton degradation of refractory organic compounds but also provides magnetism of the photocatalyst for magnetic separation from treated water. We combined an SMSMPR with the magnetic TiO 2 -Fe 3 O 4 catalyst. The prepared TiO 2 -Fe 3 O 4 composites showed high photo-Fenton catalytic activity for degradation of AMX. Cycle experiments demonstrate that the combination of backwashing treatment with magnetic separation could enhance the stability and reusability of the SMSMPR, promoting its practical application for removal of organic pollutants in aqueous solution.

Conflicts of interest
There are no conicts to declare.