Facile Fabrication of Novel NiFe2O4@Carbon Composites for Enhanced Adsorption of Emergent Antibiotics

Water purification is becoming one of the most pertinent environmental issues throughout the world. Among common types of water pollution involving heavy metals, pharmaceutical drugs, textile dyes, personal care products, and other persistent organic pollutants, the pollution of antibiotic drugs is increasingly emerging due to their adverse effects on microorganisms, aquatic animals, and human health. Therefore, the treatment of such contaminants is very necessary to reduce the concentration of antibiotic pollutants to permissible levels prior to discharge. Herein, we report the use of NiFe2O4@C composites from a bimetallic-based metal-organic framework Ni-MIL-88B(Fe) for removal of ciprofloxacin (CFX) and tetracycline (TCC). The effect of production temperatures (600–900 °C), solution pH (2–10), NiFe2O4@C dose (0.05–0.2 g/L), concentration of antibiotics (10–60 mg/L), and uptake time (0–480 min) was investigated systematically. Response surface methodology and central composite design were applied for quadratic models to discover optimum conditions of antibiotic adsorption. With high coefficients of determination (R2 = 0.9640–0.9713), the proposed models were significant statistically. Under proposed optimum conditions, the adsorption capacity for CFX and TCC were found at 256.244, and 105.38 mg/g, respectively. Recyclability study was employed and found that NiFe2O4@C-900 could be reused for up to three cycles, offering the potential of this composite as a good adsorbent for removal of emergent antibiotics.


Introduction
Global industrialization not only accompanies drastically increasing demands for clean water, but it can also trigger many potential pollutions related to water resources. Therefore, water purification is currently one of the most pertinent environmental issues throughout the world. There are common types of water pollution caused by heavy metals, pharmaceutical drugs, textile dyes, personal care products, or other persistent organic pollutants [1][2][3]. Among them, the pollution of pharmaceutical drugs, including antibiotics, is alarmingly emerging due to their adverse effects on microorganisms, aquatic animals, and human health through bioaccumulation in food chains [4]. Tetracycline (TCC, molecular formula: C 22 H 24 N 2 O 8 , molar weight: 444.435 g/mol) acts as a broad-spectrum antibacterial antibiotic [5]. It has been widely consumed in livestock industries and human therapies mainly due to low-cost production [6]. It was estimated that TCC accounts for one third of the total production, and is ranked second in the consumption of worldwide antibiotics [7]. Meanwhile, ciprofloxacin (CFX, molecular formula: C 17 H 18 FN 3 O 3 , molar weight: 331.346 g/mol) presents as one of the most widely used second-generation quinolones [8]. It was reported that millions of prescriptions were consumed for antibacterial purposes, eye irritations, and bone infections in the US [9]. The occurrence of TCC and CFX pollution can

Instrumentation
The D8 Advance Bruker powder diffractometer was used to record the X-ray powder diffraction (XRD, Hitachi Inc., Krefeld, Germany) profiles using Cu-Kα beams as excitation sources. The S4800 instrument (JEOL, Tokyo, Japan) was implemented to capture the scanning electron microscope (SEM) images with magnification of 7000× using an accelerating voltage source (15 kV). The N 2 adsorption/desorption isotherm measurements were recorded on the Micromeritics 2020 volumetric adsorption analyzer system (Micromeritics Inc., Norcross, GA, USA). The FT-IR spectra were recorded on the Nicolet 6700 spectrophotometer (Thermo Fischer Scientific Inc., Waltham, MA, USA). The UV-Vis spectrophotometer (Shimadzu, Kyoto, Japan) was used to determine the concentrations of TCC and CFX antibiotics at 273 and 272 nm in wavelength, respectively.

Synthesis of Ni-MIL-88B(Fe) Material
The Ni-MIL-88B(Fe) material could be fabricated by solvothermal method, referring to a previous study with some modifications of amount of salts and solvents [40]. In detail, the synthesis process is summarized in Figure 1. As a typical procedure, 0.3 g H 2 BDC, 0.4309 g Fe(NO 3 ) 3 ·9H 2 O, and 0.1268 g NiCl 2 ·6H 2 O were each dissolved completely in separate beakers containing DMF (40 mL) under stirring for 15 min. All solutions were then poured into a larger beaker, and CH 3 CN (40 mL) was carefully added under stirring for the next 30 min. The homogeneous mixture was divided and transferred into Teflon-lined autoclaves with appropriate volume. All autoclaves were loaded in a heating oven at 100 • C for 15 h. After cooling down the oven, the suspension was extracted, washed with DMF (three times) and methanol (three times) to make sure that all residual metals and impurities were removed. The solid was separated from the solution by centrifugation. Finally, all bold red yellow crystals were dried by vacuum system and stored for usage as template precursor for NiFe 2 O 4 @C synthesis.

Synthesis of NiFe2O4@C Materials from Ni-MIL-88B(Fe)
As-produced Ni-MIL-88B(Fe) was used as a template to synthesize NiFe2O4@C. As a typical procedure, 2.0 g Ni-MIL-88B(Fe) were heated (ramping rate of 3 °C/min) at various temperatures, 600, 700, 800, and 900 °C, from the room temperature for 4 h. To make sure that degasification of oxygen was complete, nitrogen (100 mL/min) was flowed continuously throughout the samples for 1 h before pyrolysis. After pyrolysis, the system temperature was allowed to reduce gradually to room temperature beneath nitrogen flow (100 mL/min). The pyrolysis products were collected and labeled as NiFe2O4@C-x, where x denotes the pyrolysis temperature (NiFe2O4@C-600, NiFe2O4@C-700, NiFe2O4@C-800,  As-produced Ni-MIL-88B(Fe) was used as a template to synthesize NiFe 2 O 4 @C. As a typical procedure, 2.0 g Ni-MIL-88B(Fe) were heated (ramping rate of 3 • C/min) at various temperatures, 600, 700, 800, and 900 • C, from the room temperature for 4 h. To make sure that degasification of oxygen was complete, nitrogen (100 mL/min) was flowed continuously throughout the samples for 1 h before pyrolysis. After pyrolysis, the system temperature was allowed to reduce gradually to room temperature beneath nitrogen flow (100 mL/min). The pyrolysis products were collected and labeled as NiFe 2 O 4 @C-x, where x denotes the pyrolysis temperature (NiFe 2 O 4 @C-600, NiFe 2 O 4 @C-700, NiFe 2 O 4 @C-800, and NiFe 2 O 4 @C-900).
where, C 0 (mg/L) and C f (mg/L) denote the initial and finial concentrations, respectively; W (g) and V (L) denote the mass of adsorbent and volume of the solution, respectively.

Experimental Design for Optimization
After analyzing the most influential effect of a range of factors, three main factors involving antibiotic concentration, sorbent mass, and pH were selected to discover the optimum adsorption of antibiotic ciprofloxacin (CFX) and antibiotic tetracycline (TCC). The response surface methodology was applied for statistical analysis, and proposed solutions to reach the highest adsorption capacity. In detail, based on the Box-Behnken design, Table 1 summaries the list of factors and value ranges for adsorption models of TCC and CFX on adsorbent.  [42,43]. Moreover, graphitic carbon was found due to the presence of a broad peak between 20 and 30 • . This crystallite phase may be formed by the destruction of ligand structure containing aromatic rings, converting them into carbon layers or matrix. A previous study reported the same phenomenon caused by carbonizing metal-organic framework under nitrogen to break down the coordination bonds, and form ionic bonds as of NiFe 2 O 4 [44]. The existence of the magnetic component aids the separation of the sorbent materials while the porous carbon component may enhance the adsorption of antibiotic drugs.

FT-IR Analysis
The FT-IR spectra presented in Figure 3 show the moderate difference between original Ni-MIL-88B(Fe) and NiFe2O4@C-x materials. Overall, there are several peaks that still remained, but others are absent from the original structure. In detail, a strong absorption band at the wavenumber 3500-3750 cm −1 was typical for the -OH functional groups in all samples [45]. Another absorption band at 1600-1631 cm −1 was maintained, and characteristic of the vibration of C=O bonds, regardless of a minor shift of wavenumber from 1600 in Ni-MIL-88B(Fe) to 1631 cm −1 in NiFe2O4@C-x [46]. The strong emerging absorption band at 1392 cm −1 characterized the symmetrical and asymmetrical oscillations of carboxylate groups (-COO) of ligand on Ni-MIL-88B(Fe), but it was not observed on NiFe2O4@C-x series [47]. This finding provided the information of breaking out the carboxylate ligands under pyrolysis. Clear evidence of the presence of C-H at 2926, 2864, and 750 cm −1 was presented in NiFe2O4@C-x materials [48]. These chemical bonds were typical on graphical carbon with high hydrophobicity. The peaks shifting from 558 to 568 cm −1 were observed since the temperature of pyrolysis increased and it was attributable to the existence of Fe-

FT-IR Analysis
The FT-IR spectra presented in Figure 3 show the moderate difference between original Ni-MIL-88B(Fe) and NiFe 2 O 4 @C-x materials. Overall, there are several peaks that still remained, but others are absent from the original structure. In detail, a strong absorption band at the wavenumber 3500-3750 cm −1 was typical for the -OH functional groups in all samples [45]. Another absorption band at 1600-1631 cm −1 was maintained, and characteristic of the vibration of C=O bonds, regardless of a minor shift of wavenumber from 1600 in Ni-MIL-88B(Fe) to 1631 cm −1 in NiFe 2 O 4 @C-x [46]. The strong emerging absorption band at 1392 cm −1 characterized the symmetrical and asymmetrical oscillations of carboxylate groups (-COO) of ligand on Ni-MIL-88B(Fe), but it was not observed on NiFe 2 O 4 @C-x series [47]. This finding provided the information of breaking out the carboxylate ligands under pyrolysis. Clear evidence of the presence of C-H at 2926, 2864, and 750 cm −1 was presented in NiFe 2 O 4 @C-x materials [48]. These chemical bonds were typical on graphical carbon with high hydrophobicity. The peaks shifting from 558 to 568 cm −1 were observed since the temperature of pyrolysis increased and it was attributable to the existence of Fe-O [49,50]. This proof consolidated the successful conversion of coordination bonds of Ni-MIL-88B(Fe) into ionic bonds of NiFe 2 O 4 @C-x. It is possible that carbon acts a strong reductant to transform Fe(III) into Fe(II) species. The previous study explained the same event occurred on MIL-53(Fe) or Fe(BDC) (BDC = 1,4-benzenedicarboxylic acid) material [51].

Morphological Analysis
Morphological characteristics of original Ni-MIL-88B(Fe) and NiFe2O4@C-x materia by scanning electron microscopy (SEM) analysis. Figure 4A shows that Ni-MIL-88B(F particles have a relatively uniform size distribution with mainly hexagonal shape. Th observation was well consistent with that of a previous study [40]. When carbonizing N MIL-88B(Fe) material at different temperatures, the morphology of the pyrolysis materia changes markedly as shown in Figure 4B-E. The results indicate that the morphology NiFe2O4@C-x became relatively amorphous, and their surface has more defects at high calcination temperature. This finding was explained due to the effect of thermal energ causing the destruction of hexagonal structure, and strong rearrangement of metal atom under magnetic field.

Morphological Analysis
Morphological characteristics of original Ni-MIL-88B(Fe) and NiFe 2 O 4 @C-x materials by scanning electron microscopy (SEM) analysis. Figure 4A shows that Ni-MIL-88B(Fe) particles have a relatively uniform size distribution with mainly hexagonal shape. This observation was well consistent with that of a previous study [40]. When carbonizing Ni-MIL-88B(Fe) material at different temperatures, the morphology of the pyrolysis materials changes markedly as shown in Figure 4B-E. The results indicate that the morphology of NiFe 2 O 4 @C-x became relatively amorphous, and their surface has more defects at higher calcination temperature. This finding was explained due to the effect of thermal energy, causing the destruction of hexagonal structure, and strong rearrangement of metal atoms under magnetic field.

Nitrogen Adsorption/Desorption Isotherm Analysis
Ni-MIL-88B(Fe) and NiFe2O4@C-900 were selected as representatives to characterize the nitrogen adsorption and desorption isotherms as shown in Figure 5. The results depict that the isotherms resemble between mixed Type II and Type IV curves with the presence of hysteresis, indicating that the structures of Ni-MIL-88B(Fe) and NiFe2O4@C-900 obtained mesoporous and microporous properties. In addition, BET surface area and pore volume of NiFe2O4@C-900 were measured, at 68.3 m 2 /g, and 0.1541 cm 3 /g, respectively, which were considerably lower than those of original Ni-MIL-88B(Fe), at 268 m 2 /g, and 0.14 cm 3 /g, respectively. This finding reconfirmed that the pyrolysis process causes the collapse of crystalline structure, and agglomeration of magnetic NiFe2O4 nanoparticles during pyrolysis at high temperature (900 °C). However, NiFe2O4@C-900 with a sufficiently large surface area and diverse surface groups can attain good adsorption efficiency results.

Nitrogen Adsorption/Desorption Isotherm Analysis
Ni-MIL-88B(Fe) and NiFe 2 O 4 @C-900 were selected as representatives to characterize the nitrogen adsorption and desorption isotherms as shown in Figure 5. The results depict that the isotherms resemble between mixed Type II and Type IV curves with the presence of hysteresis, indicating that the structures of Ni-MIL-88B(Fe) and NiFe 2 O 4 @C-900 obtained mesoporous and microporous properties. In addition, BET surface area and pore volume of NiFe 2 O 4 @C-900 were measured, at 68.3 m 2 /g, and 0.1541 cm 3 /g, respectively, which were considerably lower than those of original Ni-MIL-88B(Fe), at 268 m 2 /g, and 0.14 cm 3 /g, respectively. This finding reconfirmed that the pyrolysis process causes the collapse of crystalline structure, and agglomeration of magnetic NiFe 2 O 4 nanoparticles during pyrolysis at high temperature (900 • C). However, NiFe 2 O 4 @C-900 with a sufficiently large surface area and diverse surface groups can attain good adsorption efficiency results.

Effect of Pyrolysis Temperature
Calcination condition strongly affects the structure and adsorption properties of materials. Thus, the composite materials produced at various temperature were compared in adsorption capacity to CFX and TCC. Figure 6 shows that NiFe2O4@C-900 achieved the

Effect of Pyrolysis Temperature
Calcination condition strongly affects the structure and adsorption properties of materials. Thus, the composite materials produced at various temperature were compared in adsorption capacity to CFX and TCC. Figure 6 shows that NiFe 2 O 4 @C-900 achieved the highest adsorption capacity in both CFX and TCC antibiotics, 80.0 mg/g and 65.5 mg/g, respectively. This can be explained as follows. At higher pyrolysis temperatures, the structures of NiFe 2 O 4 @C-x are more defective, higher degree of graphitic carbons, and larger number of functional groups. Therefore, the adsorption capacity of NiFe 2 O 4 @C-x was higher at the higher produced temperatures.

Effect of Pyrolysis Temperature
Calcination condition strongly affects the structure and adsorption properties of materials. Thus, the composite materials produced at various temperature were compared in adsorption capacity to CFX and TCC. Figure 6 shows that NiFe2O4@C-900 achieved the highest adsorption capacity in both CFX and TCC antibiotics, 80.0 mg/g and 65.5 mg/g, respectively. This can be explained as follows. At higher pyrolysis temperatures, the structures of NiFe2O4@C-x are more defective, higher degree of graphitic carbons, and larger number of functional groups. Therefore, the adsorption capacity of NiFe2O4@C-x was higher at the higher produced temperatures.
Herein, Ni-MIL-88B(Fe) attained a relatively good adsorption capacity to TCC antibiotic (63.7 mg/g), which was very close to that of NiFe2O4@C-900. However, this material exhibited some disadvantages such as poor stability in aqueous solvent (particularly at low pH), difficult recovery after treatment, and low recyclability. As a result, Ni-MIL-88B(Fe) was not selected to perform further investigations.  Herein, Ni-MIL-88B(Fe) attained a relatively good adsorption capacity to TCC antibiotic (63.7 mg/g), which was very close to that of NiFe 2 O 4 @C-900. However, this material exhibited some disadvantages such as poor stability in aqueous solvent (particularly at low pH), difficult recovery after treatment, and low recyclability. As a result, Ni-MIL-88B(Fe) was not selected to perform further investigations.

Effect of Solution pH
Surface charge of magnetic composites and ionization of antibiotic molecules can be controlled by pH value. Thus, the influence of solution pH 2-10 on CFX and TCC adsorption capacity was investigated in this work. Figure 7 shows the highest adsorption capacity of CFX and TCC on NiFe 2 O 4 @C-x series at pH 4 and pH 3, respectively, except for the case of CFX adsorption by NiFe 2 O 4 @C-600, where optimum pH was found at 8.0. NiFe 2 O 4 @C-900 exhibited the highest CFX and TCC adsorption capacities, at 100.0 mg/g and 118.7 mg/g, respectively. These optimum pH values found herein were consistent with some previous studies [12,52]. Therefore, the CFX solution at pH 4, and TCC solution at pH 3 were selected to investigate for the effect of adsorbent dose.

Effect of Composite Dose
Adsorbent dosage is optimized to reach lower amounts of used materials, but higher level of antibiotic treatment. In this study, composite dosage was set between 0.05 and 0.2 g/L. The adsorption capacity and removal efficiency were measured to totally assess the effect of adsorbent dosage. Figure 8 indicates that composite dose has strong effects on adsorption of CFX and TCC. Overall, the dose increased from 0.05 to 0.2 g/L, the removal efficiency would gradually increase, but uptake capacity would gradually decrease in most cases. These findings are because higher dose of materials provided more adsorptive sites to improve the percentage of antibiotic removal. However, the use of higher amount of composite could lead to a decrease of uptake capacity based on Equation (2). The same phenomenon of adsorption of CFX and TCC on various composites was found in the previous studies [53,54]. Surface charge of magnetic composites and ionization of antibiotic molecules can be controlled by pH value. Thus, the influence of solution pH 2-10 on CFX and TCC adsorption capacity was investigated in this work. Figure 7 shows the highest adsorption capacity of CFX and TCC on NiFe2O4@C-x series at pH 4 and pH 3, respectively, except for the case of CFX adsorption by NiFe2O4@C-600, where optimum pH was found at 8.0. NiFe2O4@C-900 exhibited the highest CFX and TCC adsorption capacities, at 100.0 mg/g and 118.7 mg/g, respectively. These optimum pH values found herein were consistent with some previous studies [12,52]. Therefore, the CFX solution at pH 4, and TCC solution at pH 3 were selected to investigate for the effect of adsorbent dose.

Effect of Composite Dose
Adsorbent dosage is optimized to reach lower amounts of used materials, but higher level of antibiotic treatment. In this study, composite dosage was set between 0.05 and 0.2 g/L. The adsorption capacity and removal efficiency were measured to totally assess the effect of adsorbent dosage. Figure 8 indicates that composite dose has strong effects on adsorption of CFX and TCC. Overall, the dose increased from 0.05 to 0.2 g/L, the removal efficiency would gradually increase, but uptake capacity would gradually decrease in most cases. These findings are because higher dose of materials provided more adsorptive sites to improve the percentage of antibiotic removal. However, the use of higher amount of composite could lead to a decrease of uptake capacity based on Equation (2). The same phenomenon of adsorption of CFX and TCC on various composites was found in the previous studies [53,54].   In particular, the adsorbents produced at higher temperatures gave higher adsorption of both CFX and TCC. Indeed, take dose of 0.1 g/L as an example, the percentages of CFX and TCC removal by NiFe 2 O 4 @C-600 obtained at only 32.3% and 19%, respectively, but those by NiFe 2 O 4 @C-900 reached the highest values, at 72.7% and 68.4%, respectively. At the same trend, uptake capacity of CFX and TCC by NiFe 2 O 4 @C-600 obtained at only 80.7 mg/g and 42.7 mg/g, respectively, but those by NiFe 2 O 4 @C-900 reached the highest values, at 181.7 mg/g and 154 mg/g, respectively. This was due to pyrolysis conditions' influence on pore structure and surface area of composites. Finally, we considered several factors of material mass as well as adsorption efficiency to select a dose of 0.1 g/L for further investigations. Figure 9 shows the amount of CFX and TCC antibiotics adsorbed on NiFe 2 O 4 @C-x composite against contact time. In general, the adsorption occurred rapidly at the initial periods (0-30 min) due to the availability of adsorptive sites. Most of the kinetic adsorption curves reached a threshold of equilibrium at about 180 min. Adsorption CFX and TCC antibiotics by NiFe 2 O 4 @C-x was unconducive after this moment because desorption was initiated. This event was because the water molecules would enhance the occupation of pores of NiFe 2 O 4 @C-x, pushing the antibiotic molecules away from the adsorbents. Moreover, the immersion of composite adsorbent in aqueous solutions reduced their stability. Based on such observations, optimum time of 180 min was selected to explore the effect of further parameters such as antibiotic concentration. tion curves reached a threshold of equilibrium at about 180 min. Adsorption CFX and TCC antibiotics by NiFe2O4@C-x was unconducive after this moment because desorption was initiated. This event was because the water molecules would enhance the occupation of pores of NiFe2O4@C-x, pushing the antibiotic molecules away from the adsorbents. Moreover, the immersion of composite adsorbent in aqueous solutions reduced their stability. Based on such observations, optimum time of 180 min was selected to explore the effect of further parameters such as antibiotic concentration. According to Figure 9, NiFe2O4@C-900 gave the highest removal percentage and adsorption capacity. Therefore, this material was used to calculate the kinetic constants from pseudo first-order, pseudo second-order, Elovich, and Bangham. Table 2 compares the fitness of kinetic models by coefficient of determination (R 2 ). With R 2 values greater than 0.9, adsorption kinetics described the experimental data well. TCC adsorption was followed by both Elovich and Bangham, while CFX adsorption was followed by pseudo second-order equation. According to Figure 9, NiFe 2 O 4 @C-900 gave the highest removal percentage and adsorption capacity. Therefore, this material was used to calculate the kinetic constants from pseudo first-order, pseudo second-order, Elovich, and Bangham. Table 2 compares the fitness of kinetic models by coefficient of determination (R 2 ). With R 2 values greater than 0.9, adsorption kinetics described the experimental data well. TCC adsorption was followed by both Elovich and Bangham, while CFX adsorption was followed by pseudo second-order equation.   To gain insight into the adsorption equilibrium, the isotherm equations including Langmuir, Freundlich, and Temkin were fitted using their nonlinear forms. Table 3 lists the isotherm constants for adsorption of CFX and TCC on NiFe 2 O 4 @C-900 material. Based on R 2 values, the best fitting order of CFX adsorption isotherm models would be: Temkin > Langmuir > Freundlich, and that of TCC adsorption isotherm models would be: Langmuir > Langmuir > Temkin. As a result, adsorption of CFX and TCC on NiFe 2 O 4 @C-900 followed by Temkin and Langmuir, respectively. With a range of values R L and 1/n between zero and one, the adsorption of antibiotics in this study offered a favorable process. More importantly, the values of maximum adsorption capacity of CFX and TCC identified from Langmuir were very high, 737.42 mg/g and 827.34 mg/g, respectively. It is therefore proposed that NiFe 2 O 4 @C-900 can be a good adsorbent to adsorb CFX and TCC from water.

Model Design
From results obtained from investigating the effect of each factor, the optimization model was designed by selecting the powerful factors such as antibiotic concentration, adsorbent dosage, and pH. Based on the widely used central composite design, the range of experimental values are shown in Table 1. There are twenty experimental runs for each antibiotic adsorption model, giving twenty experimental results. The corresponding twenty predicted results were listed in Table 4. Overall, the adsorption capacity of CFX antibiotic reached the highest value at 285 mg/g (entry 10), and the lowest at 33.21 mg/g (entry 1). Meanwhile, TCC absorption capacity was the highest value of 105 mg/g (entry 16), and the lowest value of was 29 mg/g (entry 9).
The relationship between the antibiotics adsorption capacity on NiFe 2 O 4 @C-900 and independent variables x 1 , x 2 , and x 3 generated from Design-Expert 11 software expressed in the actual equation is as follows: Q TCC (mg/g) = −120.61 + 9.89 x 1 + 320.26 x 2 + 55.84 x 3 − 11.25 x 1 x 2 − 0.21 x 1 x 3 + 87.50 where x 1 is the antibiotic concentration (mg/L), x 2 is the adsorbent mass (g/L), and x 3 is the pH value. From Equations (3) and (4), all factors had positive, and great effects on adsorption capacity, which means that the value of response would increase with the increase of factors.
To assess the compatibility of each model, statistical analysis such as ANOVA (analysis of variance) discloses the significance based on the model terms such as sum of squares, degrees of freedom, mean of squares, F-and p-values, as well as fitness statistics. In detail, the statistically analytical results are summarized in Table 5. According to Table 5, CFX and TCC antibiotic treatment models exhibit F-values of 29.74 and 37.57, respectively, corresponding to p-values less than 0.0001. This means that both models presented a chance of 0.01% that F-values could occur due to noise. As a result, both design models were statistically significant at a reliability level of 95%. More importantly, all factors which obtained p-value less than 0.0001, were statistically significant at the same reliability level. As such, the models have been successfully designed with the statistical significance of all factors [55].
Some fitness statistics such as coefficient of determination (R 2 ) and adequate precision could be used to diagnose the compatibility of models. Obviously, R 2 values for both models (0.9640-0.9713) were very close to 1.0, suggesting the outstanding fitness properties. In addition, both models had adequate precision (AP) values between 17.6179 and 19.9834. A model with AP value greater than 4.0 offers a high degree of goodness and compatibility. Here, AP values were both greater than 4.0, indicating the high compatibility of proposed models, providing the appropriate signals for the quadratic models [56]. The coefficient of variation (CV%) is a statistical measure of data dispersion level in a collection of data relative to the mean. As observed, both indicators (6.05-11.81%) were lower than 15%, suggesting that the dispersion of experimental data of proposed models should be acceptable.
Diagnostic plots can be used to evaluate whether a model is compatible with the experimental results. One of the most common ones is predicted versus actual plots as shown in Figure 11A,B. It was obvious that model data acquired a good correlation since the data points were concentrated in straight lines. Moreover, plots in Figure 11C,D again supported the evidence of random distribution of data points without any trends or scatters.

Optimization of Process Parameters
The influence of influential factors on the adsorption capacity of TCC and CFX antibiotics by NiFe2O4@C-900 can be depicted by three-dimensional surface response plots as shown in Figure 12. Herein, two of three factors would be changed while another was kept at a central level or zero. Overall, the grid regions that described adsorption capacity were converged, indicating that the model could be optimized in the range of investigated values of factors. In detail, for response surfaces of the CFX adsorption model in Figure  12A,C,E, the adsorption capacity would be optimized at around pH 4. This finding was again compatible with surveyed results in the previous sections. The great improvement could be obtained if the CFX concentrations were set to high values (30-40 mg/L). NiFe2O4@C-900 adsorbent dose (0.05-0.15 g/L), however, showed a moderately influential effect, but an interactive correlation with other factors.

Optimization of Process Parameters
The influence of influential factors on the adsorption capacity of TCC and CFX antibiotics by NiFe 2 O 4 @C-900 can be depicted by three-dimensional surface response plots as shown in Figure 12. Herein, two of three factors would be changed while another was kept at a central level or zero. Overall, the grid regions that described adsorption capacity were converged, indicating that the model could be optimized in the range of investigated values of factors. In detail, for response surfaces of the CFX adsorption model in Figure 12A,C,E, the adsorption capacity would be optimized at around pH 4. This finding was again compatible with surveyed results in the previous sections. The great improvement could be obtained if the CFX concentrations were set to high values (30-40 mg/L). NiFe 2 O 4 @C-900 adsorbent dose (0.05-0.15 g/L), however, showed a moderately influential effect, but an interactive correlation with other factors.
Meanwhile, for response surfaces of the TCC adsorption model, the response regions could be highly converged, which are highlighted with red color in Figure 12B,D,F. Blue color regions showed lower values of adsorption capacity. Based on optimum surfaces, the TCC concentrations were suggested from 20 to 30 mg/L to obtain higher magnitude of adsorption capacity. The conditions at around pH 3 would facilitate the adsorption of TCC, which was fitted with surveyed results in the previous sections. As similar as the case of CFX adsorption model, the effect of NiFe 2 O 4 @C-900 adsorbent dose (0.05-0.15 g/L) was moderate. However, TCC adsorption capacities were higher at lower NiFe 2 O 4 @C-900 dose which is described by Equation (2). Finally, the optimum solutions for adsorption of CFX were proposed as follows, concentration of 40.0 mg/L, adsorbent weight of 0.148 g/L, and pH 3.97. Meanwhile, the optimum solutions for adsorption of TCC were proposed as follows, concentration of 23.93 mg/L, adsorbent weight of 0.115 g/L, and pH 3.6. The desirability values were found at 0.885 and 1.0000 to obtain the highest capacities for CFX and TCC at 256.244, and 105.38 mg/g, respectively.
Model verification was used to repeat the proposed conditions, and attain the statistical errors of experimental capacity values. As expected, the verified errors were lower than 5.0%, suggesting the good compatibility of the quadratic models proposed in this study.

Recyclability Study
The research on material recyclability are an important strategy to demonstrate the potential effect of sustainable adsorbents on environmental remediation. This experiment was carried out under the following conditions: NiFe2O4@C-900 material weight of 0.15 g/L, CFX antibiotic concentration of 40 mg/L, time of 180 min, and pH 4. HCl solvent (0.1 mol/L in ethanol) was used as an effective eluent by referring to the study reported by [57]. Indeed, HCl in aqueous solution disassociates a species of Cl − . As a result, it can complete the adsorption with CFX, enhancing the effectiveness of desorption of CFX from NiFe2O4@C-900 material. Accordingly, Figure 13 shows that NFO NiFe2O4@C-900 could

Recyclability Study
The research on material recyclability are an important strategy to demonstrate the potential effect of sustainable adsorbents on environmental remediation. This experiment was carried out under the following conditions: NiFe 2 O 4 @C-900 material weight of 0.15 g/L, CFX antibiotic concentration of 40 mg/L, time of 180 min, and pH 4. HCl solvent (0.1 mol/L in ethanol) was used as an effective eluent by referring to the study reported by [57]. Indeed, HCl in aqueous solution disassociates a species of Cl − . As a result, it can complete the adsorption with CFX, enhancing the effectiveness of desorption of CFX from NiFe 2 O 4 @C-900 material. Accordingly, Figure 13 shows that NFO NiFe 2 O 4 @C-900 could be reused at least three times with a moderate change (~25%) in adsorption capacity, from 201.75 (1st run) to 149.0 mg/g (3rd run). However, there was a considerable decrease in adsorption (98.5 mg/g) for 4th run with 51.2%. Thereby, NiFe 2 O 4 @C-900 may be a potential material for antibiotic adsorption.

Conclusions
In this work, a series of NiFe2O4@C-x composites were successfully pr bimetallic-based metal-organic framework Ni-MIL-88B(Fe). Production (600-900 °C) had a strong effect on the structure and adsorption properti TCC antibiotics. The characteristics indicated the existence of NiFe2O4 nano carbon matrix/layer. The NiFe2O4@C-x composites possessed a range of fun For adsorption results, the composite produced at 900 °C obtained higher pacities for both CFX and TCC. The optimum pH investigations were ach for CFX and pH 3 for TCC. The other optimum conditions of NiFe2O4@C-90 time, and concentration were found, at 0.1 g/L, 180 min, and 20-30 mg/L Response surface methodology was applied for quadratic models with ve cient of determination and adequate precision, suggesting high compatibil imental data. For CFX adsorption, the experimental conditions were identi tration of 40.0 mg/L, adsorbent weight of 0.148 g/L, and pH 3.97. Additio adsorption, the experimental conditions were identified at concentration adsorbent weight of 0.115 g/L, and pH 3.6. The desirability values were and 1.0000, respectively. Under such proposed optimum conditions, the ex sorption capacities for CFX and TCC were found at 256.244 and 105.38 mg/ Recyclability study indicated NiFe2O4@C-900 could be reused up to three c

Conclusions
In this work, a series of NiFe 2 O 4 @C-x composites were successfully produced from a bimetallic-based metal-organic framework Ni-MIL-88B(Fe). Production temperatures (600-900 • C) had a strong effect on the structure and adsorption properties of CFX and TCC antibiotics. The characteristics indicated the existence of NiFe 2 O 4 nanoparticles in the carbon matrix/layer. The NiFe 2 O 4 @C-x composites possessed a range of functional groups. For adsorption results, the composite produced at 900 • C obtained higher adsorption capacities for both CFX and TCC. The optimum pH investigations were achieved at pH 4 for CFX and pH 3 for TCC. The other optimum conditions of NiFe 2 O 4 @C-900 dose, contact time, and concentration were found, at 0.1 g/L, 180 min, and 20-30 mg/L, respectively. Response surface methodology was applied for quadratic models with very high coefficient of determination and adequate precision, suggesting high compatibility with experimental data. For CFX adsorption, the experimental conditions were identified at concentration of 40.0 mg/L, adsorbent weight of 0.148 g/L, and pH 3.97. Additionally, for TCC adsorption, the experimental conditions were identified at concentration of 23.93 mg/L, adsorbent weight of 0.115 g/L, and pH 3.6. The desirability values were found, at 0.885 and 1.0000, respectively. Under such proposed optimum conditions, the experimental adsorption capacities for CFX and TCC were found at 256.244 and 105.38 mg/g, respectively. Recyclability study indicated NiFe 2 O 4 @C-900 could be reused up to three cycles, offering the potential of this composite adsorbent for removal of emergent antibiotics.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.