Synthesis of magnetic nanocomposite Fe3O4@ZIF-8@ZIF-67 and removal of tetracycline in water

We prepared a double-layer magnetic nanocomposite Fe3O4@ZIF-8@ZIF-67 by layer-by-layer self-assembly. Fe3O4@ZIF-8@ZIF-67 was used to remove tetracycline from an aqueous solution via a combination of adsorption and Fenton-like oxidation. Depending on the outstanding porous structure of the Fe3O4@ZIF-8@ZIF-67, a high adsorption capacity for tetracycline was 356.25 mg g−1, with > 95.47% removal efficiency within 100 min based on Fenton-like oxidation. To better understand the mechanisms involved in integrated adsorption and Fenton-like oxidation, various advanced characterization techniques were used to monitor the changes in morphology and composition of Fe3O4@ZIF-8@ZIF-67 before and after removal of tetracycline. Scanning electron microscopy/energy-dispersive X-ray spectroscopy (SEM/EDS), Fourier transform infrared spectroscopy (FTIR), and X-ray diffraction (XRD) all supported adsorption and Fenton oxidation of tetracycline. This study extends the application of Fe3O4@ZIF-8@ZIF-67 for environmental remediation.


Introduction
Antibiotic consumption has increased globally from usage in human and animal disease treatment, growth promotion, and prophylaxis (Kovalakova et al. 2020). Most ingested antibiotics are released into the aquatic environment instead of being metabolized by organisms (Kraemer et al. 2019). Tetracycline (TC), which ranks second in terms of global production and usage, was discovered in the 1940s (Jeong et al. 2010). Recent results suggest that the concentration of TC in the surface water is ~ 0.15 ug/L (Guo et al. 2017). Environmental TC residues may destroy ecosystems and result in the development of antibiotic-resistant bacteria (ARB) . Thus, it is critical and urgent to remove TC from the aquatic environment.
Successful TC removal has been achieved by adsorption (Bao et al. 2018), photocatalytic oxidation , electrochemistry (Bao et al. 2018), ozonation (Wang et al. 2011), Fenton and Fenton-like oxidation (Kong et al. 2020), and biodegradation (Shao et al. 2019). However, the development of these methods has been confined because of their disadvantages, such as a poor adsorption performance, low visible-light response, high energy consumption, complicated equipment, and harsh reaction conditions (Wang et al. 2011Bao et al. 2018;Shao et al. 2019;Kong et al. 2020;Zhao et al. 2020). A combined process may overcome these limitations by integrating two or more strategies, such as the combination of adsorption and microbial metabolism , adsorption and photocatalytic oxidation (Wang et al. 2019), electrochemistry and photocatalytic oxidation (Liu et al. 2009), and photocatalytic and Fenton oxidation (Han et al. 2020). In contrast, the combination of adsorption and Fenton-like oxidation is environmentally friendly, efficient, and operationally straightforward (Weng et al. 2020). Therefore, it is critical to source a material that can achieve adsorption and Fenton-like oxidation.
Metal-organic frameworks (MOFs), which are a type of porous coordination polymer, are comprised of metal ions or clusters that are coordinated to organic linkers (Cook et al. 2013;Tibbetts and Kostakis 2020). The past three decades have witnessed enormous growth in MOF research from synthesis to uses in adsorption, storage, separation, and catalysis (Stock and Biswas 2012). Numerous previous studies have shown that MOFs have an excellent effect on TC removal from wastewater with their large surface areas and high porosity (Li et al. 2020a;Xiao et al. 2020). To date, most studies have focused on the use of MOFs as an adsorbent, and little information is available on the performance and mechanism of TC removal from wastewater via synergetic adsorption and Fenton-like oxidation using magnetic MOFs.
In this study, a double-layer magnetic MOF (Fe 3 O 4 @ ZIF-8@ZIF-67) was prepared by a solvothermal method in layer-by-layer self-assembly. The structure of the double layer could expand the porosity and specific surface area for adsorption and provide more active sites for Fenton-like oxidation. Good magnetic properties could contribute to recycle from the reaction system. The main aim of this study was to discuss TC removal by using Fe 3 O 4 @ZIF-8@ZIF-67 by synergetic adsorption and Fenton-like oxidation.

Synthesis of Fe 3 O 4 NPs
Fe 3 O 4 nanoparticles (NPs) were prepared by a modified coprecipitation method based on a previous study (Li et al. 2020b). They were synthesized as follows. FeCl 3 ·6H 2 O (1.35 g) and FeSO 4 ·7H 2 O (0.695 g) were dissolved in distilled water (100 mL). Nitrogen gas was injected. The mixture was heated to 70℃ under vigorous stirring until a turbid solution formed. Heating was continued for 25 min. NH 3 ·H 2 O (5 mL) was added to the mixture, and the solution was stirred at 80℃ for 30 min. The mixture was cooled to room temperature. The solid was separated from the solution by magnetic separation and washed several times with deionized water.

Synthesis of Fe 3 O 4 @ZIF-8@ZIF-67
Fe 3 O 4 @ZIF-8@ZIF-67 was prepared by using a facile solvothermal process. Zn(NO 3 ) 2 ·6H 2 O (2.98 g) was dissolved in CH 3 OH (35 mL) to form a solution. Fe 3 O 4 and the solution were mixed under ultrasound for 15 min, followed by the addition of a methanol solution (20 mL) of 2-methylimidazole (6.57 g) under ultrasound for 30 min. A methanol solution (15 mL) of Co(NO 3 ) 2 ·6H 2 O (2.91 g) was added to the abovementioned solution, and the ultrasound was maintained for 30 min. The product was collected with an external magnet, washed twice with methanol, and dried at 70℃ for 6 h.

Characterization
Microstructures of the nanocomposite were examined by scanning electron microscopy (SEM, Zeiss Sigma 300, Germany) and X-ray energy-dispersive spectrometry (EDS, BRUKER Quantax EDS with XFlash6 detector, Germany). The nanocomposite crystal structure was detected by X-ray diffraction (XRD, BRUKER D8 Advance, Germany). The infrared spectrum was measured by Fourier transform infrared spectroscopy (FTIR, Nicolet iS5, America). The magnetic properties of the prepared nanoparticles were measured on a vibrating samples magnetometer (VSM, LakeShore 7410, USA). The N 2 adsorption-desorption isotherms were recorded on a fully automatic gas adsorption analyzer (BET, Micromeritics, ASAP 2460 3.00, USA) with a degassing temperature of 120℃ for 8 h, and the adsorption parameters of samples were obtained by Brunauer-Emmett-Teller method.

Batch experiments
Batch removal experiments were conducted in 150-mL conical flasks containing 50 mL of an aqueous sample of TC with a temperature of 25℃ (the influence of temperature was disregarded) and performed on a 150-rpm shaker for 100 min. If necessary, the pH was adjusted with HCl or NaOH (0.1 M).
When the experiment was completed, the adsorbent was separated from the solution by an external magnet, and the solution was filtered through a 0.22-μm microporous membrane to obtain the supernatant. The concentration of TC was determined by ultraviolet-visible spectrophotometry at 355 nm. The TC concentration was calculated from a standard curve. The removal efficiency Re (%) was calculated from (1): The adsorption capacity (q t ) of TC at any time was calculated using (2): where C 0 (mg/L) is the initial concentration of TC, C t (mg/L) is the concentration of TC at time t (min), V (mL) is the volume of the TC solution, and M (mg) is the adsorbent dosage.

Results and discussion
Characterization SEM was carried out to observe the morphological of Fe 3 O 4 , Fe 3 O 4 @ZIF-8, ZIF-8@ZIF-67, and Fe 3 O 4 @ ZIF-8@ZIF-67, and the changes after adsorption and Fenton-like oxidation of Fe 3 O 4 @ZIF-8@ZIF-67. SEM micrographs (Fig. 1b) show that Fe 3 O 4 @ZIF-8 exhibited uniform rhombic dodecahedral shape crystals and was not undermined by Fe 3 O 4 NP loading (Sajjadi et al. 2018). Figure 1c and d shows that the material nanostructure of ZIF-8@ZIF-67 and Fe 3 O 4 @ZIF-8@ZIF-67 changed from a granular to a rod-shaped morphology with the introduction of ZIF-67. Fe 3 O 4 NPs were distributed on the surface and inside the ZIF-8@ZIF-67. After adsorption, the morphology of Fe 3 O 4 @ZIF-8@ZIF-67 was similar to that of the as-prepared one (Fig. 1e), but the particle surface became a little rougher, which is attributed to TC (2) Fenton-like oxidation, the morphology of Fe 3 O 4 @ZIF-8@ ZIF-67 was not changed obviously ( Fig. 1f) but showed several pinholes and folds. This appearance most likely resulted because the generated OH acted on the structure. EDS element mapping was used to investigate the elemental distribution of Fe 3 O 4 @ZIF-8@ZIF-67. Figure 2 shows that C, N, Zn, and Co elements are uniformly distributed in the shell layer, while Fe and O elements are mainly dispersed in the core section of the composite. The mapping of C, N, Zn, and Co elements can be attributed to ZIFs, and the presence of Fe and O can be ascribed to Fe 3 O 4 . The fact that C, N, O, Fe, Zn, and Co elements are all detected confirmed the successful preparation of Fe 3 O 4 @ZIF-8@ZIF-67 nanocomposites.
To determine the chemical structure, Fe 3 O 4 , Fe 3 O 4 @ ZIF-8, ZIF-8@ZIF-67, and Fe 3 O 4 @ZIF-8@ZIF-67 were analyzed by XRD. The XRD pattern for Fe 3 O 4 @ZIF-8@ ZIF-67 and ZIF-8@ZIF-67 ( Fig. 3a and (Jiang et al. 2016). The introduction of Fe 3 O 4 NPs did not destroy the crystallinity of ZIF-8 or ZIF-67, which confirms that Fe 3 O 4 @ZIF-8@ZIF-67 was synthesized. To verify the crystalline structure stability, the XRD patterns of Fe 3 O 4 @ ZIF-8@ZIF-67 after adsorption and Fenton-like oxidation ( Fig. 3b) were analyzed. In Fig. 3b, the peak position and intensity appointed to Fe 3 O 4 remained consistent before and after adsorption and Fenton-like oxidation, which shows that the Fe 3 O 4 structure was stable. After batch reactions and Fenton-like oxidation, the magnetic property of Fe 3 O 4 remained unchanged and met the practical application of repeated and strong separation. However, after adsorption and Fenton-like oxidation, the diffraction peak intensities indexed to ZIF-8@ZIF-67 decreased compared with that observed prior to reaction, which may be attributed to TC adsorption, which resulted in clogged pores of Fe 3 O 4 @ ZIF-8@ZIF-67 and changes in the nanoparticle surface (Li et al. 2019a). The crystal form and structure of the Fe 3 O 4 @ ZIF-8@ZIF-67 did not show changes because of the existence of characteristic peaks. These results agree with the SEM observations (Fig. 1).
FTIR analysis was carried out to prove the synthesis of Fe 3 O 4 @ZIF-8@ZIF-67 and to obtain an improved understanding of the possible reaction mechanisms (Fig. 4). As shown in Fig. 4b, the appearance of a peak at 581 cm −1 could be related to Fe-O-Fe vibration, which suggests that Fe 3 O 4 NPs was loaded successfully. The peaks at 1136-1306 cm −1 are ascribed to imidazole ring vibration, whereas those at 2917-3107 cm −1 were assigned to the stretching vibration of the saturated hydrocarbon C-H (CH 3 ) and unsaturated hydrocarbon C-H of the 2-methylimidazolium. Therefore, the 2-methylimidazolium served as an organic ligand in the nanocomposite. The peak at 425 cm −1 was caused by the Zn-N and Co-N stretching vibration, and the peaks at 994-1106 cm −1 resulted from C = N in ZIF, which confirms the presence of a ZIF-8 and ZIF-67 structure. These characteristic peaks confirm the synthesis of nanocomposite Fe 3 O 4 @ZIF-8@ZIF-67 in accordance with the XRD results (Fig. 3). In Fig. 5, after adsorption and Fenton-like oxidation, the 1618 cm −1 peak that was associated with the benzene ring broadened indicates that TC was adsorbed on the Fe 3 O 4 @ZIF-8@ZIF-67. Compared with the original Fe 3 O 4 @ZIF-8@ZIF-67, the main characteristic peaks exhibited no obvious alterations after adsorption or Fenton-like oxidation; thus, it was inferred that the magnetic nanocomposite structure had not been destroyed.
The hysteresis loop from VSM was used to obtain the magnetic properties of Fe 3 O 4 , Fe 3 O 4 @ZIF-8, and Fe 3 O 4 @ ZIF-8@ZIF-67. According to the VSM results in Fig. 5, the saturation magnetization (Ms) values of Fe 3 O 4 , Fe 3 O 4 @ ZIF-8, and Fe 3 O 4 @ZIF-8@ZIF-67 were 70.21 emu/g, 37.16 emu/g, and 18.38 emu/g, respectively. The phenomenon that the magnetic intensity decreased shows that ZIF-8 and ZIF-67 were coated successively. The as-prepared nanocomposite still exhibited sufficient magnetic responsibility and could be separated with an external magnetic field to facilitate recycling and reuse. Figure 6 shows that Fe 3 O 4 @ ZIF-8@ZIF-67 started to decompose strongly at 500℃, which indicates its good thermal stability.
The surface area and porosity of Fe 3 O 4 @ZIF-8@ZIF-67 and Fe 3 O 4 @ZIF-8 were shown in Supplementary Fig. S1. The BET surface area of Fe 3 O 4 @ZIF-8 was 346.7 m 2 /g, and that of Fe 3 O 4 @ZIF-8@ZIF-67 was slightly lower (328.7 m 2 /g). The isotherms are a combination of type I isotherms, which suggested a typical microporous structure. The detailed porosity results of the two adsorbents were listed in Supplementary Table S1.

Adsorption performance of Fe 3 O 4 @ZIF-8@ZIF-67
Different factors affect the adsorption process, and the Fe 3 O 4 @ZIF-8@ZIF-67 dosage, temperature, pH, initial concentration of TC, and HA were investigated by a variablecontrolling strategy. To determine the optimum adsorbent dosage, we explored the adsorbent effect on the TC removal efficiency. As shown in Fig. 7a, with an increase of adsorbent from 5 to 20 mg, the TC removal efficiency improved and reached a maximum removal efficiency of more than 88%. This simply results from sufficient adsorption sites for adsorption. However, the adsorption capacity dropped gradually from 352.88 to 181.00 mg/L when the amount of the adsorbent further increased from 20 to 40 mg. This is because the unsaturated adsorption is caused by the excessive adsorbent, resulting in the decrease in adsorption per unit mass of Fe 3 O 4 @ZIF-8@ ZIF-67, which leads to a decrease in adsorption capacity.
Temperature determines the velocity of the molecular motion and the energy of the molecular surface, which affects the mass transfer rate. Therefore, it is important to study the effect of temperature on the adsorption process. Figure 7b shows that, with an increase in temperature, the adsorption capacity of Fe 3 O 4 @ZIF-8@ZIF-67 on TC increases slightly, which indicates that the adsorption is an endothermic process. This phenomenon may be attributed to the improved dispersion rate of TC molecules as the temperature increases, and as a result, TC molecules can pass through the external boundary faster.
As the initial concentration of TC increases, the adsorption capacity of TC improves, which shows that the initial concentration of contaminants may affect the mass transfer rate. A higher concentration increases the effective collision probability between the adsorbate and the adsorbent, which causes the adsorption to move in a positive direction. When the initial concentration reaches 120 mg/L, the removal efficiency exceeds 90%. However, there is no further significant change in removal efficiency when the initial concentration continues to increase. This behavior may be related to the saturation of active sites in Fe 3 O 4 @ZIF-8@ZIF-67, which cannot adsorb total TC at higher concentrations.
We investigated the adsorption quantity of double-layer MOF on TC at different pH. As indicated in Fig. 7c, the removal efficiency of TC exceeded 88% and remained stable for a pH of 5-7. Under strongly acidic (pH 3) or basic (pH 11) conditions, respectively, the adsorption efficiency decreases. This may be due to certain collapse of MOFs structure under acidic (pH 3) conditions (Han et al. 2018). At high ambient pH (pH 11), the amount of deprotonation of phenolic hydroxyl groups of TC increases, which may weaken the adsorption interaction.
It is meaningful to explore the effect of HA on the adsorption of TC because HA exists extensively in natural water and wastewater. Figure 7d shows the variation tendency in adsorption capacity of Fe 3 O 4 @ZIF-8@ZIF-67 with the concentration of HA range from 0 to 8 mg/L. The adsorption capability of Fe 3 O 4 @ZIF-8@ZIF-67 hardly changes as the HA concentration increases. This phenomenon shows that Fe 3 O 4 @ZIF-8@ZIF-67 is an efficient adsorbent in HAenriched water.

Catalytic performance of Fe 3 O 4 @ZIF-8@ZIF-67
The removal efficiencies of TC were evaluated using Fe 3 O 4 @ZIF-8@ZIF-67 and H 2 O 2 (Fig. 8). The sole H 2 O 2 system induced only 1.31% removal efficiency of TC within 180 min, indicating insignificant self-activated oxidation of H 2 O 2 . With the introduction of Fe 3 O 4 @ZIF-8@ ZIF-67, 95.76% of TC was removed in Fe 3 O 4 @ZIF-8@ ZIF-67/H 2 O 2 system, which was higher than the combined removal efficiency with H 2 O 2 (1.31%) and Fe 3 O 4 @ ZIF-8@ZIF-67 (89.03%) alone. These results show that Fe 3 O 4 @ZIF-8@ZIF-67 could be a favorable catalyst for H 2 O 2 activation. The addition of H 2 O 2 compensated for the deficiency that Fe 3 O 4 @ZIF-8@ZIF-67 alone had a weak adsorption efficiency for low concentration of TC, and the TC removal from aqueous solution rose to a new high. Figure 8 (inset) shows that the TC removal efficiency of the Fe 3 O 4 @ZIF-8@ZIF-67/H 2 O 2 system minus the Fe 3 O 4 @ZIF-8@ZIF-67 system decreased progressively, which indicates that Fenton-like oxidation acted mainly on the TC molecules that were adsorbed on the Fe 3 O 4 @ ZIF-8@ZIF-67 instead of the free TC molecules in an aqueous solution. Therefore, with synergetic adsorption and Fenton-like oxidation, TC molecules were adsorbed preferentially on the Fe 3 O 4 @ZIF-8@ZIF-67 and, afterwards, some were oxidized by radicals that were generated by H 2 O 2 . This is similar to the literature report by Hou et al. that an efficient "capture" is crucial for the following "destroy" (Hou et al. 2020).

Fig. 6 Thermogravimetric analysis diagram of Fe 3 O 4 @ZIF-8@ZIF-67
In Fenton-like reactions, the concentration of H 2 O 2 , the primary pH of the solution, and the adsorbent dosage may influence the oxidation performance, and thus, the effect of the abovementioned conditions on the catalytic oxidation of TC was investigated as follows.
As shown in Fig. 9, TC removal efficiency boosts with increasing Fe 3 O 4 @ZIF-8@ZIF-67 dosage from 5 to 20 mg, attributable to the increase of availability of active sites and the generation of radicals (Ahmed et al. 2016). The maximum removal efficiency (95.75%) was shown at the dosage of 20 mg. When the amount of the adsorbent further increased, the removal efficiency barely improved. This may be due to the limiting of H 2 O 2 concentration and the scavenging effect of excess adsorbent to hydroxyl radicals, which were wildly reported in previous works (Ahmed et al. 2016;Shi et al. 2018).
H 2 O 2 can oxidize a variety of organic compounds, such as carboxylic acids, alcohols, and esters, into inorganic states, and thus, it is important in Fenton-like reactions. Therefore, it is necessary to explore the influence of H 2 O 2 concentration. As shown in Fig. 10, with increasing H 2 O 2 concentration from 0 to 35 mM, the removal efficiency of TC improves from the initial 90.40% to a maximum of 98.97%. During this process, the increased hydroxyl radicals improved the oxidation efficiency. With a further increase of H 2 O 2 concentration, the removal efficiency (3) pH is an important factor to control the generation of ions and free radicals in Fenton-like reactions. As shown in Fig. 11, when the pH ranges from 3 to 9, the removal efficiency was sustained over 90%, which shows that Fe 3 O 4 @ ZIF-8@ZIF-67/H 2 O 2 system has a good TC removal under acid and neutral conditions. When pH was exceeded 9, the removal efficiency drops to 78.69%. This is possibly due to the auto self-decomposition of H 2 O 2 to H 2 O and O 2 , and the rapid conversion of ·OH to its less active conjugate base, ·O ‾ , in alkaline conditions (Babuponnusami and Muthukumar 2012). Compared with TC absorption by independent Fe 3 O 4 @ZIF-8@ZIF-67, the removal efficiency by Fe 3 O 4 @ZIF-8@ZIF-67/H 2 O 2 system increased the most (12.24%) when pH was 3. This is because H 2 O 2 was more likely to produce ·OH in an acidic environment to promote oxidation reaction, and the oxidation potential of ·OH is fairly strong in acidic conditions (Burbano et al. 2005). Overall, with the synergistic effect of absorption and oxidation, the Fe 3 O 4 @ZIF-8@ZIF-67 heterogeneous Fenton system can be used in a wider pH range (pH 3-9) than traditional Fenton-like system and has better environmental adaptability.

Kinetics analysis
TC removal consists of two sections, i.e., adsorption and Fenton-like oxidation. The following explains the two sections:

Adsorption kinetics
Adsorption kinetics were described by the pseudo-first-order (Eq. (6)) and pseudo-second-order (Eq. (7)) kinetic models: where q e and q t (mg/g) are the amounts of TC adsorption at equilibrium and time t (min), respectively, and k 1 (min −1 ) and k 2 (g/mg/min) are the rate constants for the pseudo-firstorder and pseudo-second-order kinetic models, respectively. The best-fit kinetic parameters of TC adsorption are presented in Table 1 at 25℃. The resulting best linear correlation coefficients for the pseudo-second-order kinetic model (R 2 = 0.99788) were greater than those for the pseudo-firstorder kinetic model (R 2 = 0.97004) (Table 1), which shows that the adsorption process followed the pseudo-secondorder kinetic model at 25℃ and chemisorption was dominant in the speed limit.

Oxidation kinetics
To explore the TC removal process, data of the Fenton-like oxidation fitted the pseudo-first-order (Eq. (8)) and pseudosecond-order (Eq. (9)) kinetic models: where C 0 is the initial concentration of TC and C t (mg/L) is the concentration of TC at t min, and k obs (min −1 ) and k (g/ mg/min) are the rate constants for the pseudo-first-order and pseudo-second-order kinetic models, respectively. As shown in Table 2, the correlation coefficients for the pseudo-firstorder kinetic model (R 2 = 0.94738) were greater than those for the pseudo-second-order kinetic model (R 2 = 0.93004), which means that the pseudo-first-order kinetic model was more appropriate for oxidation.

Comparison of various materials
The double-layer magnetic MOF (Fe 3 O 4 @ZIF-8@ZIF-67) was compared with the single-layer magnetic MOF (Fe 3 O 4 @ ZIF-8) and other common adsorbents (AC and nZVI) to reflect the adsorption performance. Figure 12 shows that the adsorption efficiency of Fe 3 O 4 @ZIF-8@ZIF-67 (90.01%) was two or more times that of Fe 3 O 4 @ZIF-8 (38.47%), which is attributed primarily to the high porosity and larger specific surface area of the former, which results from the double-layer structure. In contrast with AC (68.53%) and nZVI (18.01%), Fe 3 O 4 @ZIF-8@ZIF-67 showed an exceedingly good adsorption efficiency. The above results indicate that Fe 3 O 4 @ZIF-8@ZIF-67 was an outstanding TC adsorbent. A comparison of the synergetic adsorption and Fenton-like oxidation of TC by various materials is shown in Fig. 13. For the two classical Fenton reagents, Fe (nZVI) and Fe 2+ (FeSO 4 ·7H 2 O), as high as 82.70% and 84.50% TC could be removed in 100 min, respectively, which depended on ·OH radical generation. By comparison, Fe 3 O 4 @ZIF-8@ ZIF-67 (95.47%) exhibited a superior removal performance compared with the conventional Fenton reagents. After the   (Guo et al. 2020).
In order to investigate the synergetic removal mechanism of Fe 3 O 4 @ZIF-8@ZIF-67 composite, Fe 3 O 4 , Fe 3 O 4 @ZIF-8, and ZIF-8@ZIF-67 were applied to activate H 2 O 2 under the same condition. With the existence of H 2 O 2 , the TC removal efficiency in the Fe 3 O 4 @ZIF-8 system increased by 7.99%, whereas the ZIF-8@ZIF-67 system (93.81%) showed an increase of 34.85% (Figs. 12 and 13). Fe 3 O 4 @ZIF-8 and ZIF-8@ZIF-67 all have a similar structure of Zn-ZIFs, which indicates that Co-ZIFs had high catalysis and was the primary active substance. Wu et al. reported a similar finding by using Fe 3 O 4 @Zn/Co-ZIFs composite to activate PMS for carbamazepine degradation, and they found that Co-ZIFs was the primary active substance (Wu et al. 2020). Zn-ZIFs could not catalyze H 2 O 2 , but the adsorption was dominated because of the stable valence states of Zn (Wu et al. 2020). The promotion of removal efficiency in the Fe 3 O 4 @ZIF-8 system resulted mainly from the slight catalysis of Fe 3 O 4 NPs that were attached to the ZIF-8 surface (Fig. 13). Overall, Fe 3 O 4 @ZIF-8@ZIF-67 exhibited a significantly higher removal than other materials in this work.

Removal mechanism
To determine the type of free radicals that were produced during the reaction, free radicals quenching tests were carried out. Tert-butanol (TBA) and ethanol are two commonly used scavengers of free radicals. Ethanol is regarded as scavengers of both SO 4 · ‾ and ·OH (Li et al. 2019b), while TBA is considered as the scavenger of only ·OH (Hu et al. 2018). The Fenton-like oxidation reaction mainly catalyzes the H 2 O 2 to produce potent oxidative hydroxyl radicals (·OH) to degrade organic pollutants. Therefore, tertbutanol was introduced as a trapping agent to get ·OH. As shown in Fig. 14, with the increase in TBA concentrations (0-60 mM), the TC removal efficiency decreased from 94.48 to 84.99%, which indicates that ·OH played an important role in the degradation of TC. The removal of TC at a high TBA concentration is mainly attributed to the adsorption ability of Fe 3 O 4 @ZIF-8@ZIF-67. This was similar to the report by Hou et al., where they compared TC removal efficiency by nZVI/MIL-101(Cr) and nZVI/MIL-101(Cr) + H 2 O 2 + TBA systems and found both had almost the same TC removal efficiency (Hou et al. 2020).
Based on these results, a proposed pathway and mechanism for Fenton-like oxidation of TC using Fe 3 O 4 @ZIF-8@ ZIF-67 was proposed. This possible mechanism was supported by the obtained characterization (SEM, EDS, XRD, FTIR, VSM, and TG), and kinetics (adsorption and oxidation) data suggested that Fe 3 O 4 @ZIF-8@ZIF-67 functioned as an adsorbent and a catalyst in the Fenton-like oxidation system. TC molecules were adsorbed rapidly onto the Fe 3 O 4 @ZIF-8@ZIF-67. The H 2 O 2 was activated by Co 2+ in Fe 3 O 4 @ZIF-8@ZIF-67 to generate numerous ·OH radicals. TC molecules were oxidized by powerful oxidizing ·OH, degraded into smaller molecules, and even mineralized to H 2 O and CO 2 (Hou et al. 2020).

Fe 3 O 4 @ZIF-8@ZIF-67 reusability and stability
For the recycle tests, the Fe 3 O 4 @ZIF-8@ZIF-67 were separated by magnetism and vortexed with ethanol during adsorption experiments or washed with deionized water during Fenton-like oxidation experiments. After drying at 60℃, the same amount of Fe 3 O 4 @ZIF-8@ZIF-67 was used for the next recycle with experimental conditions unchanged. Figure 15 shows the reusability of the adsorption and Fentonlike oxidation experiments of Fe 3 O 4 @ZIF-8@ZIF-67 over five cycles. The removal efficiencies of TC in the adsorption and oxidation experiments decreased to different extents; the removal efficiency in the oxidation experiment was always better than that in the adsorption. The removal efficiencies in the adsorption experiment decreased from 88.96 to 79.27%, and that in the oxidation experiment decreased from 93.38 to 82.94%, which was sufficiently high for practical application. The decrease in removal performance after repeated use may be resulted from the changes in a porous structure that is caused by clearing and washing between the repeated experiments or the absorption of pollutants or byproducts on the active sites of Fe 3 O 4 @ZIF-8@ZIF-67 (Wu et al. 2020;Zhou et al. 2020).
Moreover, the Co ion leaching in Fe 3 O 4 @ZIF-8@ZIF-67/ H 2 O 2 system was also monitored after each cyclic experiment ( Supplementary Fig. S2). In the first two cycles, the leached concentration of Co ion was about 0.6 mg/L. In the following three cycles, the leached Co ion increased to 0.8 mg/L with the increase in the number of cycles. The leaching Co ion concentration was below the permissible limit (1 mg/L) according to the Chinese National Standard (GB 25,467-2010). The SEM and FTIR spectroscopy was conducted to further investigate the stability of recovered Fe 3 O 4 @ZIF-8@ZIF-67 after five runs of Fenton-like oxidation experiments. SEM image ( Supplementary Fig. S3) shows that the Fe 3 O 4 @ZIF-8@ZIF-67 after one cycle was similar to the fresh one. However, after five cycles, the rodshaped morphology of Fe 3 O 4 @ZIF-8@ZIF-67 changed and the surface was eroded. After the cycle experiments, the intensity of FTIR peaks assigned to ZIF decrease with respect to Fe 3 O 4 ( Supplementary Fig. S4). This may be due to the leaching of Co ion, resulting in the reduction of the ZIF ratio in the nanocomposite. In addition, Fe 3 O 4 @ZIF-8@ ZIF-67 was immersed in HCl (pH 3) and NaOH (pH 12) solution to test the acidic and basic stability ( Supplementary  Fig. S4). The spectra obtained from pH 3 changed obviously in intensity, while those obtained from pH 12 changed a little, which was consistent with previous literature that ZIF has exceptional stability in alkaline water (Park et al. 2006).

Conclusions
The as-prepared magnetic nanocomposite Fe 3 O 4 @ZIF-8@ ZIF-67 is a potential material for TC removal from water in combination with adsorption and Fenton-like oxidation. The adsorption and Fenton-like oxidation experiments indicated that removal efficiency of up to 95.47% was achieved at an initial TC concentration of 160 mg/L, a Fe 3 O 4 @ZIF-8@ZIF-67 dose of 0.4 g/L, and an H 2 O 2 concentration of 30 mM within 100 min at 25℃. Initial kinetic adsorption data best fit a pseudo-second-order kinetic model (R 2 ≥ 0.997), whereas the oxidation process best fit a pseudo-first-order model (R 2 ≥ 0.947). The SEM, EDS, XRD, FTIR, VSM, and TG results confirmed that Fe 3 O 4 @ZIF-8@ZIF-67 was synthesized and had excellent structural stability and magnetic property. The mechanism is that the high conductivity of Fe 3 O 4 NPs promoted Co 2+ and Co 3+ cycling. ·OH radicals were generated by H 2 O 2 and oxidized the adsorbed TC molecules on the Fe 3 O 4 @ZIF-8@ZIF-67, which mineralized to H 2 O and CO 2 . The findings of our study provide theoretical guidance and technical support to treat antibiotics in water by the combined process, and inspiration and a new perspective for morphological design and performance optimization of the new-generation MOFs.

Supplementary Information
The online version contains supplementary material available at https:// doi. org/ 10. 1007/ s11356-021-18042-9. Data availability All data generated or analyzed during this study are included in this published article.

Declarations
Ethics approval and consent to participate Not applicable.

Competing interests
The authors declare no competing interests.