Obtaining of Fe3O4@C Core-Shell Nanoparticles as an Adsorbent of Tetracycline in Aqueous Solutions

Tetracycline is used to treat various diseases in animals and humans, so its high use causes this substance to be frequently found in effluents and in potable water reservoirs, causing serious problems for human health and the environment. Many processes are applied and evaluated in an attempt to removal tetracycline found in environment, such as adsorption. The application of iron-carbon nanoparticles with core-shell structure (Fe3O4@C) as adsorbent, can be an alternative to removal of these contaminants. In the present study Fe3O4@C nanoparticles were synthesized and used to adsorb tetracycline in aqueous solutions. The core-shell characterization was performed using X-ray diffraction techniques, infrared spectroscopy, surface area, chemical analysis and morphology, and adsorption capacity through isotherms. The results indicated that Fe3O4@C nanoparticles presented good tetracycline removal capability, 73.3%, when applied an initial concentration of antibiotic of the 30 mg L-1 and 0.5 g of adsorbent.


Introduction
Antibiotics are widely used in the treatment of diseases in animals and humans, as well as applied in food rations in order to increase animal growth rates 1,2 . Among the most commonly used antibiotics are tetracyclines 3 , which has high antimicrobial activity associated it's the aromatic structure that containing the naphthol chemical group that is antibacterial 4

.
However, approximately 50-90% of the doses of tetracyclines administered in health treatment procedures are not absorbed by organisms and are eliminated by humans and animals in sewage systems [5][6][7] . Tetracycline molecules have a complex chemical structure and the natural environmental and sewage conditions are not sufficient to decompose their chemical structure. Thus, the amount of tetracycline accumulated in sewage can be a serious environmental problem 8 .
Physical-pharmacological and biological methods have been used to remove the drugs from aqueous mediums. But recently new techniques have been evaluated for this propose, such as adsorption, membrane filtration, photocatalysis, ozonization and electroxidation [9][10][11] . However, these processes generally generate toxic secondary compounds and have low removal efficiency in short time periods and are costly 4,12 .
Consequently, the presence of these drugs and their secondary compounds in water or sewage systems 13,14 contributes to the natural selection of bacteria that are resistant to antibiotics [15][16][17] .
Among the methods of removal of tetracycline, the adsorption is a process fast, economical and low energy cost process, and does not form secondary compounds [18][19][20][21] . Adsorption is widely applied in environmental treatment applications around the world. Liquid-solid adsorption systems are based on the ability of certain solids to preferentially concentrate specific substances from solutions onto their surfaces 22 .
In this sense, the magnetic nanoparticles, composed of a core of magnetite and by a nanolayer of carbon have shown unique characteristics for application in adsorption processes 23 . Nanoparticles with magnetic properties can be efficiently separated from wastewater by using magnetic separators, avoiding the filtration process, which represents a difficulty in using nanosized materials at the treatment of great volumes of wastewaters in environmental remediation processes 24,25 . This core-shell nanostructure associates the high adsorption capability of the carbon nanoparticles with the high number of active sites and high magnetic response, which provides a fast and efficient separation of nanoparticles 26,27 . In addition, protects the core of carbon acids and alkaline environments, avoiding your deterioration 28,29 .
In this work, core-shell nanoparticles, magnetite covered by the carbon (Fe 3 O 4 @C), were obtained through hydrothermal process. These nanoparticles were physical and chemical characterized and their absorption capacity of tetracycline molecules was evaluated.
The new solution was transferred to a stainless steel (AISI 3014) reactor coated by Teflon with a volume of 200 mL. The reactor was closed and heated in a drying oven (Nova Ética, 402/3N model) at 180°C during 18 h. The solution was then removed from the reactor and cooled until room temperature. After this process the solution was washed three times with distilled water and washed two times with ethanol (99% purity; C 2 H 5 OH; Vetec). Finally, the nanoparticles were oven dried by 2 h at 70 °C.
X-ray diffraction analysis was made using a x-ray powder diffraction diffractometer (Bruker, model D8) with CuKα radiation target (λ = 1.5418 Ǻ), electrical tension of 40 kV, electrical current of 40 mA, wavelength with steps of 0.02, exposure time per step of 1 second and 2Theta angle range from 2° to 72°. The Rietveld refinement method was used for the identification of phases and the chemical composition, determination of the type of crystalline structure. The size of the crystallites was determined using the Scherrer equation with k = 0.94 31 .
FTIR analysis was performed with a Fourier transform infrared spectrophotometer (Jasco, FTIR 4200 model) with wavenumber range between 400 to 4000 cm -1 . The samples were previously prepared in a disk format mixing 5 mg of nanoparticles and 300 mg of potassium bromide (KBr).
Specific surface area of the core-shell nanoparticles was determined by Brunauer-Emmett-Teller -BET methods 32 (ASTM D 6556-17 33 ), using a static nitrogen adsorption instrument (Quantachrome, model Autosorb 1C). The diameter and volume of pores were calculated using the Barret, Joyner e Halenda (BJH) methods.
The samples were pretreated at 100 ºC under vacuum conditions (10 -7 Torr) by 3 h. For determine the density and the volume of the core-shell nanoparticles the gas pycnometer equipment (Quantachrome, Ultrapycnometer 1200e model) was used.
The core-shell morphology and the its surface chemical composition were analyzed using a scanning electron microscope (SEM-FEG, Tescan, Czech Republic, Mira 3 -20 kV) and an energy dispersive spectrometer (EDS), respectively.

Determination of the zero-charge point (pH PCZ )
Zero-charge point (pH PCZ ) of nanoparticles Fe 3 O 4 @C was determined according method described for Babic et al. 34

Adsorption kinetics
Adsorption kinetic tests were performed in Erlenmeyers with 100 mL of a solution of tetracycline with concentration of 30 mg L -1 (Fragon, purity 98.6%) and with 0.5 g of core-shell nanoparticles. The adsorption experiments were conducted at 25 ° C, pH 5.0 and on a shaker at 100 rpm.
Aliquots from the solutions were centrifuged at 3400 rpm by 10 min and then filtered using a membrane with diameter pore of 0.2 μm with the aid of a vacuum pump and then the passant solution were analyzed in the UV-Vis spectrophotometer (SpectraMax, Plus 384) in the wavelengths range between 250 and 550 nm. The characteristic absorber band for the tetracycline aqueous solution was identified at wavelength of 357 nm. All solutions were analyzed using quartz cuvettes and performed in triplicate.
A calibration curve to determine the tetracycline concentrations in the solutions were obtained with the UV-Visible spectrophotometer preparing solutions with different concentrations of tetracycline in distilled water, varying from 3 to 30 mg L -1 . For each concentration, the respective absorbance value in 357 nm was obtained and associated with the respectively concentration. A plot associating the absorbance values with the tetracycline concentration was obtained and defined as the calibration curve.
Experimental data were compared to the kinetic models of pseudo-first order and pseudo-second order to determine the kinetic parameters, according to Equations (01) and (02) respectively. The intraparticle diffusion model also available which presented in Equation (03). .
Where q t (mg g -1 ) is the amount of tetracycline in the solid phase of time; t (min) is the time of adsorption; q e (mg g -1 ) é a is the amount of tetracycline at the equilibrium; k 1 (min -1 ) is the first-order rate constant; k 2 (mg g -1 min -1 ) is the second-order rate constant; k in (mg g -1 min -1/2 ) is the intraparticle diffusion rate constant; C (mg g -1 ) is a constant related to the diffusion resistance.
The amount of tetracycline adsorbed at equilibrium was calculated by Equation (04).
Where: q e (mg g -1 ) is the adsorption capacity at the equilibrium; V (L) is the volume of the tetracycline solution, C 0 (mg L -1 ) is the initial concentration of tetracycline in the solution, C (mg L -1 ) the concentration of tetracycline at equilibrium and m (g) is the mass of the nanoparticles.
The kinetic parameters of the models evaluated are determined using Statistica ® 7.0 (StatSoft ® , USA). The Quasi-Newton method was applied to minimize the square sum of errors (SSE), which is calculated using Equation (5). q exp and q cal are the experimental and calculated adsorption capacity, respectively.

Adsorption isotherms
For the adsorption isotherm studies the flasks samples containing the solutions were agitated in the shaker at 100 rpm during 4 h (defined in preliminary tests). The solutions were prepared with 100 mL of a tetracycline solution with 30 mg L -1 and containing different mass of core-shell nanoparticles, between 0.1 and 0.9 g. The pH values of all solutions were adjusted at 5.0 using a solution of HCl 0.1 M (Vetec -purity de 99%). Subsequently, the flask samples were centrifuged at 3400 rpm (FANEM®, Excelsa II 206BL) by 10 min and filtered using the membrane with pore diameter of 0.2 μm and aided of a vacuum pump and then analyzed in the UV-Visible spectrophotometer using the characteristic wavelength in 357 nm.
The respective concentration values of the tetracycline in all solutions were determined comparing the absorbance values with the calibration curve. The isotherms were performed at three temperatures, being 25 °C, 35 °C and 45 °C and all experiments were conducted in triplicates.
In order to describe the interactive behavior between solute and adsorbent, the Langmuir, Freundlich e Langmuir-Freundlich models were applied to analyze the equilibrium adsorption characteristics, according with the Equation (06), (07) and (08), respectively.
Where q e (mg g -1 ) is the amount of tetracycline adsorbed at equilibrium on the core-shell nanoparticles; C e (mg L -1 ) is the concentration of tetracycline in the liquid phase at equilibrium; b L (L mg -1 ) is the Langmuir adsorption equilibrium that represents surface affinity; q max (mg g -1 ) is the amount of tetracycline adsorbed at saturation; k f and n f are the parameters of Freundlich models, where n f provides an indication whether the isotherm is favorable or unfavorable. Values of n f in the range between 1 to 10 represent favorable adsorption conditions; m LF is Langmuir-Freundlich constant; q mLF (mg g -1 ) is the maximum coverage capacity of the monolayer and b LF (L mg -1 ) is the affinity parameter Langmuir-Freundlich.
The isotherms parameters of the proposed models were found using the Statistica® 7.0 (StatSoft®, USA). The Quasi-Newton method was applied to minimize the square sum of errors (SSE), which is calculated using Equation (5).
The thermodynamic parameters were calculated by constructing the lnK plot as a function of T -1 in K. The equation of the better adjusted line was generated and the thermodynamic parameters calculated, in which the enthalpy and entropy variation were obtained from the slope and the intersection of the line. The equations used to calculate the constant K (Herny constant) and to calculate the parameters, Van't Hoff equation, are expressed in Equations (09), (10) and (11), respectively.
. Figure 1 shows the morphology of the Fe 3 O 4 @C structures. It is possible to observe that the nanoparticles formed have a spherical geometry, with diameters ranging from 30 to 70 nm, as well as tend to form agglomerates with dimensions of approximately 200 nm to 400 nm. The formation of these aggregate structures is an indicator that the surface energy of the nanoparticles is relatively strong. The results obtained resemble those described by Zhang et al. 36 and Fini, Niasari and Ghanbari 37 . The magnetite is a magnetic iron oxide phase originated from chemical reaction between the iron specimens and the glucose molecule in the reactor and is covered by the carbon of the core-shell nanoparticle. The peak inherent to carbon predicted at approximately 26º 45 was not clearly defined in the diffractogram possibly to the low concentration in relation to the nucleus of Fe 3 O 4 (magnetite).    Table 1 shows the Rietveld refinement for the x-ray diffractogram results for the core-shell nanoparticles. The results demonstrated the presence of carbon and magnetite in greater quantity, as expected. The lattice parameters determined for the unit cell of the magnetite and carbon phases was similar to those observed in literature, unit cell parameters: a=b = 2.47 Å and c = 6.79 Å for the crystalline carbon phase and a=b=c=8.38 Å for the magnetite phase [46][47][48] . The Scherrer method was applied considering the spherical crystallite form for the core-shell nanoparticles (k = 0.94) and the average values determined for the magnetite crystallite size were 69.33 nm, similar to the size values observed by scanning electron microscopes.

Characterization of nanoparticles
Information about the Functional groups on the surface of the core-shell nanoparticles Fe 3 O 4 @C was obtained with FT-IR, according Figure 4. The FT-IR analysis performed shows the presence of an absorption band at 561 cm -1 , which identify the Fe-O vibrational bond 49 , confirming the presence of iron oxide in the structure. In addition, the presence of carbon can be verified by the absorption bands in 1260 cm -1 , 1590 cm -1 and 2920 cm -1 , associated with the chemical groups C-O 50 , C-C 51 and C-H 52 , respectively.
The bands located at 3250 cm -1 can be attributed to the hydroxyl group or water hydrogen 53 , from the glucose or propylene glycol used in the synthesis of the nanoparticles or from the water molecules adsorbed on the nanoparticles of Fe 3 O 4 @C.
Surface area value of the core-shell nanoparticles was measured using the BET method and the value determined was 63.37 m 2 g -1 . The average values of the pore volume (0.19 cm 3 g -1 ) and pore size (8.99 nm) were determined using the mathematical model BJH with the results of BET. The surface porosity of the core-shell nanoparticles is representative of a mesoporous structure and is possibly related with the spaces formed between the agglomerates of the nanoparticles, since their value is high in relation to the dimensions of the nanoparticles. These values and characteristics are similar to those found by Zhong et al. 54 and Sayahi et al. 55 in studies with graphene@magnetite@ carbon nanosheets and magnetite/graphite nanocomposites, respectively. At pH = 5, used in kinetic and isothermal tests, tetracycline behaves as a zwitterion (dipole ion), possessing positive and negative charges, the latter being those that led to an adsorption on the surface of Fe 3 O 4 @C nanoparticles 58 . Figure 6 shows the adsorption kinetics results for the adsorptions of the tetracycline with a concentration of 30 mg L -1 in aqueous medium by a mass of 0.5 g of Fe 3 O 4 @C nanoparticles. It's possible to observe a significant decrease of the concentrations of tetracycline concentration in the aqueous medium during the first 50 min, removing approximately 58%. The high adsorption rate of the tetracycline molecules is possibly due to the higher tetracycline concentration gradient between the liquid phase (aqueous medium) and the solid phase (adsorbent) in this first time period of the process and also because that the adsorption is controlled by diffusion mechanisms on the surface of the adsorbent. The first diffusion mechanism is fast, because the diffusion processes occur on the boundary layer, followed by the diffusion processes of the adsorbate (tetracycline) into the pores and capillaries of the adsorbent structure (core-shell nanoparticles), thus, it corresponds to the intraparticle adsorption where the velocity is lower [59][60][61] .

Adsorption kinetics
It can be observed in the Figure 6, the concentration remaining in the liquid phase, after the balance to the adsorption process, was approximately 8 mg L -1 , in a time of 140 minutes. Therefore, for conditions of pH 5.0 and temperature of 25 °C the tetracycline concentration of 30 mg L -1 and adsorbent mass of 0.5 g, the removal was approximately 73%.   This high adsorption efficiency can be ascribed to abundant unoccupied active sites on the surface of the core-shell adsorbent and the high tetracycline concentration gradient in the initial stage, as well as the high specific surface area of core-shell nanoparticles 62 . In a study developed by Jafari et al. 63 , in which the authors used activated carbon-coated Fe 3 O 4 nanoparticles, AC@Fe 3 O 4 , as adsorbent to adsorb tetracycline, reached equilibrium in approximately 60 minutes, using a concentration of 20 mg L -1 tetracycline. The removal efficiency determined by the author was of approximately 50%, below the efficiency found in the present work.
To identify the mechanisms involved in the adsorption process, kinetic experimental results were fitted with kinetic models of pseudo-first order, pseudo-second order and intraparticle diffusion. The results of the adjustments, as well as the experimental results, are presented in Figure 7 (a) and (b).
The Table 2 shows the values for adsorption kinetic constants, determined from the application of the pseudofirst order, pseudo-second order and intraparticle diffusion model, for the tetracycline adsorption processes by the core-shell nanoparticles. In this table also presented the correlation coefficient of determination (R 2 ) for each model. The correlation coefficient is a good indicator of the good adjust of a theorical kinetic models and the highest value was established as the best fit. In addition to the values of R 2 , were also compared the values of the concentration of tetracycline adsorbed at equilibrium determined by experiments (q e(exp) ) and the correspondent concentrations determined from the adjustments of the theoretical kinetic models, Table 3. The value of the theoretical must be the closest to the value of q e(exp) . Thus, the adsorption of the tetracycline molecules by the core-shell nanoparticles occurs according to the pseudofirst order or pseudo-second order model.
According the R 2 values (Table 2), the model that best represents the tetracycline adsorption process by Fe 3 O 4 @C nanoparticles is the kinetic model of pseudo-second order. Analyzing the results presented in the Table 4, it's possible to observe smallest deviation between the values of q e(exp) and q e(calc) , for the pseudo-second order model.   Table 2. Kinetic parameters of the kinetic model of pseudo-first order, pseudo-second order and intraparticle diffusion for tetracycline.

Parameters
Pseudo-first order Pseudo-second order This means that the rate of adsorption is directly proportional to the square of the tetracycline concentration. For different adsorbents, several authors showed the success of the application of the pseudo-second order model for the representation of kinetic data for tetracycline adsorption 64,65,62 .
The kinetic model of pseudo-second order, predicts the behavior during the entire adsorption time, and indicates that the adsorption process can be occurs from a possible formation of complexes on the surface of the nanoparticles, probably occurring due to the ionic pairing mechanism. The study by Jafari et al. 63 evaluated the adsorption mechanism of tetracycline using Fe 3 O 4 nanoparticles coated with activated carbon, and obtained that the dominant mechanism was also pseudo-second order. The authors related that the adsorption process involves the exchange or sharing of electrons between tetracycline and nanoparticles. The study by Liu et al. 65 found a range for the pseudo-second order model of 0.0139 g mg -1 min -1 using MCM-41 impregnated with zeolite to adsorb tetracycline at 50 mg L -1 concentration. The value obtained by the authors is close to that found in this study.

Adsorption isotherms
Adsorption isotherms for the tetracycline adsorption processes were performed to evaluate the equilibrium relationship between the Fe 3 O 4 @C nanoparticles concentration and the tetracycline concentration in the aqueous medium phase at constant temperature. Figures 8, 9 and 10 show the adsorption isotherms obtained from the experimental data at different temperatures: 25 °C, 35 °C and 45 °C, respectively, as well the respectively curves obtained for the adjustment of the theorical models of Langmuir, Freundlich and Langmuir-Freundlich models.     (Table 4), as well the coefficient values of determination (R 2 ), indicate that the Freundlich model present the best fit for the three temperatures studied. This result indicates that the adsorption process occurs on heterogeneous, non-specific and energetically non-uniform surfaces, and the formation of mesopores will facilitate the adsorption of larger molecules. The parameter nf of the Freundlich model indicates how the adsorption sites are distributed in terms of their energy and the nf values in the range of 1 to 10 represent a favorable adsorption conditions, which is observed in this work, for the three temperatures studied.
Analyzing the results showed in Table 4 it is verified that the Langmuir-Freundlich model also presented a good fit, with little difference in the values coefficient of determination (R 2 ), for the three temperatures studied. It is possible to estimate the maximum adsorption capacity of tetracycline by nanoparticles, which was 14.734 mg g -1 , 5.603 mg g -1 and 3.391 mg g -1 for the 25 ºC, 35 ºC and 45 ºC, respectively. It is evident the reduction in the maximum capacity of adsorption with the increase of the temperature of the system, indicating that the tetracycline has greater affinity for the adsorbent when compared to the aqueous phase. This behavior was already expected, since the adsorption process is an exothermic process, being impaired by the increase in temperature 66 .
The Langmuir-Freundlich model has been applied to treat adsorption studies on heterogeneous surfaces, showing an exponential relation between free adsorbate and adsorbed adsorption. Therefore, good tuning of the Langmuir-Freundlich isotherm indicates that Fe 3 O 4 @C nanoparticles are heterogeneous with respect to the adsorption sites, which may be due to the contribution of the adsorption to the adsorption process. The model, when in low concentrations, follows the behavior of the Freundlich model, considering the adsorption in multilayers, whereas in high concentrations it follows the Langmuir model, considering the existence of a saturation point 67 .
Jafari et al. 63 evaluated the adsorption process of tetracycline using Fe 3 O 4 nanoparticles coated with activated carbon, with concentrations of tetracycline in the range of 100 to 300 mg L -1 . The Langmuir and Freundlich models were applied to describe the process, and the Langmuir model for all concentrations was the best to represents the experimental data. The maximum adsorption capacity obtained was 142.8 mg g -1 for 300 mg L -1 tetracycline. This value is higher than the one found in this work, possibly due to the fact that the initial concentration is higher and the adsorbent surface is different from the adsorbent used in this work. In the work of Zhang et al. 68 , in which the authors evaluated the adsorption of tetracycline by graphene-coated Fe 3 O 4 nanoparticle, the Langmuir model presented a better fit, demonstrating that the adsorption process belongs to the monolayer adsorption and the adsorption sites are evenly distributed.
Thermodynamic parameters related to the tetracycline adsorption process were determined from Equations 10 and 11. For the calculation of the thermodynamic parameters, the equilibrium constants found using the Langmuir-Freundlich isotherm were used. Table 5 shows the values of the thermodynamic parameters, entropy (ΔS), enthalpy (ΔH) and Gibbs free energy (ΔG).
The positive value of the standard enthalpy indicates the endothermic nature of the tetracycline adsorption process on nanoparticles. The negative value of Gibbs free energy, for the three temperatures studied, confirms the viability and spontaneous nature of the tetracycline adsorption process in nanoparticles. The positive value of the entropy indicates that there is affinity between the tetracycline and the surface of the adsorbent, that is, the positive value of ΔS indicates increasing randomness at the solid/liquid interface during the adsorption of tetracycline with time; there is a trend towards greater orderliness of the tetracycline molecules on the adsorbent surface during the formation of stable complexes between tetracycline and the Fe 3 O 4 @C nanoparticles.

Conclusion
Hydrothermal co-precipitation methodology proved to be an efficient and viable process for obtaining coreshell nanoparticles of magnetite and carbon. The results obtained by the characterizations indicated the formation of spherical core-shell nanoparticles of Fe 3 O 4 @C with dimensions between 20 to 70 nm, which forms agglomerates with dimensions between 200 to 400 nm.
Fe 3 O 4 @C nanoparticles have good capacity for adsorption of the tetracycline molecules in aqueous medium. The adsorption capacity determined was of 73.3% when used a concentration of 5 mg mL -1 of core-shell and 30 mg L -1 of tetracycline in aqueous medium.
Adsorption kinetics characteristics obtained for the adsorption by the Fe 3 O 4 @C nanoparticles were well described by the pseudo-second order model and the adsorption mechanisms by the Freundlich and Langmuir-Freundlich model. The adjustment of the experimental date by these theorical model establishes as the maximum adsorption capacity of 14.734 mg of tetracycline per gram of Fe 3 O 4 @C.
Adsorption process of tetracycline molecules by the Fe 3 O 4 @C nanoparticles is a spontaneous and endothermic dominated by physical adsorption mechanisms. Hence, Fe 3 O 4 @C can be used as an effective sorbent to remove tetracycline molecules in aqueous medium.