Removal of Mercury (II) by EDTA-Functionalized Magnetic CoFe2O4@SiO2 Nanomaterial with Core-Shell Structure

In order to reduce the difficulty and risk of operation, decrease the preparation time and improve the adsorption performance of magnetic nano-silicon adsorbent with core-shell structure, a carboxylated CoFe2O4@SiO2 was prepared by EDTA-functionalized method using a safe, mild and simple hydrothermal method. The results show that the prepared material of CoFe2O4@SiO2-EDTA has a maximum adsorption capacity of 103.3 mg/g for mercury ions (Hg(II)) at pH = 7. The adsorption process of Hg(II) is a chemical reaction involving chelation and single-layer adsorption, and follows the pseudo-second-order kinetic and Langmuir adsorption isotherm models. Moreover, the removal of Hg(II) is a spontaneous and exothermic reaction. The material characterization, before and after adsorption, shows that CoFe2O4@SiO2-EDTA has excellent recyclability, hydrothermal stability and fully biodegradable properties. To summarize, it is a potential adsorption material for removing heavy metals from aqueous solutions in practical applications.


Introduction
Currently, heavy metals, such as mercury, lead and copper, have extensively infiltrated natural water bodies that are becoming increasingly polluted [1]. Industrial wastewater generated in coal combustion and other industrial activities poses a threat to human health through bio-enrichment of aquatic organisms and sewage irrigation [2,3]. Heavy metals are easily accumulated in the food chain due to their non-degradable properties. Mercury pollution, in particular, is extremely dangerous because of its persistence, fluidity, high bioaccumulation and uncontrollability [4]. Moreover, it seriously jeopardizes the ecological environment, as well as human life and health [5]. Therefore, the World Health Organization (WHO) and the Environmental Protection Agency (EPA) have listed mercury as one of the most toxic elements of heavy-metal pollution [6]. The limits for mercury ions allowed in drinking water and surface water discharge standards are 2 and 10 µg/L, respectively. Mercury pollution is caused by industries, such as the chlor-alkali industry, plastics industry, dye chemicals industry and electronics industry, as well as activities like amalgamation of alfalfa [7]. Note that mercury exists in various chemical forms, including elemental mercury (Hg 0 ), methylmercury (CH 3 Hg + ) and inorganic mercury (Hg(II)) in environment. Therefore, the economic and efficient removal of mercury from water bodies in environmental field has become a top priority.

Preparation of CoFe 2 O 4 @SiO 2
CoFe 2 O 4 was prepared using hydrothermal synthesis. Briefly, 2.379 g of CoCl 2 ·6H 2 O, 7.0634 g of Fe(acac) 3 , 2.0 g of polyethylene glycol and 8.679 g of CH 3 COONa were mixed in 120 mL of ethylene glycol and continuously stirred in a water bath at 313 K for 30 min. Then, the reaction was kept at 473 K for 16 h. The material was washed with ethanol and pure water and dried at 333 K. Next, 1.0 g of the as-prepared CoFe 2 O 4 was poured into pure water and sonicated for 20 min. The magnetic solution was continued to be stirred for 30 min and heated to 353 K in a water bath. Subsequently, 2 mL of ammonia water and 2 mL of tetraethyl orthosilicate (TEOS) were added and then the reaction was executed at 353 K for 3 h. Finally, the solution was magnetically separated, washed with pure water, and dried at 333 K.

Preparation of CoFe 2 O 4 @SiO 2 -EDTA
A total of 0.6 g of the as-prepared CoFe 2 O 4 @SiO 2 was added into a 150 mL solution of ethanol and pure water (4:1) at 313 K for 0.5 h. Then 2 mL of ammonia water was added to adjust the pH to 9, and then 2 mL of TEOS was slowly added. After 1 h, 2 mL APTES was slowly added and the mixture was continued stirring for 8 h. The solution was then magnetically separated, washed with pure water and dried at 333 K. In this method, CoFe 2 O 4 @SiO 2 -NH 2 was successfully prepared. Then, the amino group bounded to the surface of the silica was reacted with 9 mmol of EDTA anhydride in 160 mL of ethanol and acetic acid (1:1) at 343 K for 16 h. Then the magnetic separation material was washed with pure water and dried it at 333 K to obtain CoFe 2 O 4 @SiO 2 -EDTA. Please see the synthesized schematic diagram of CoFe 2 O 4 @SiO 2 -EDTA in Figure 1.

Sample Characterization
The appearance, structure and size of the material were characterized using transmission electron microscopy (TEM, Tecnai G2 F30, Hillsboro, OR, USA) and scanning electron microscopy (SEM, ZEISS-SUPRA 55, Jeona, Germany). X-ray diffraction (XRD, D8 Advance, Bruker, Karlsruhe, Germany) was used to analyse the crystal structure of the material. The functional groups in the material were characterized using Fourier transform infrared spectroscopy (FT-IR, Nicolet 6700, Thermo Scientific, Waltham, MA, USA). The magnetic strength was measured using a vibrating sample magnetometer (VSM, PPMS-9, Quantum Design, San Diego, CA, USA). The specific surface area (BET) was obtained using an N 2 adsorption-desorption apparatus (Autosorb-IQ2-MP, Quantachrome, Boynton Beach, FL, USA). Finally, using X-ray photoelectron spectroscopy (XPS, Escalab 250xi, Thermo Scientific), the binding energy of the elements contained in the materials was obtained.

Batch Adsorption Experiments
Before each experiment, to prevent the hydrolysis of mercury ions, 0.1 mL of HCl and 10 mL of HNO 3 in 1000 mL pure water was used as a protective solution. A certain amount of HgCl 2 was added to the protective solution to obtain a 1000 mg/L mercury stock solution. The effect of CoFe 2 O 4 @SiO 2 -EDTA on the adsorption of the mercury solution was studied at different pH levels. 0.01 g of the adsorbent was added into an Erlenmeyer flask containing 100 mL Hg(II) solution (C 0 = 20 mg/L), and then the pH was adjusted to 2-7 using 0.1 mol/L HCl and 0.1 mol/L NaOH [34]. After shaking for 12 h, the mixture was filtered through a 0.45 µm filter and analyzed by a cold atomic absorption spectrophotometry. All experiments were performed in triplicate and then averaged. The kinetic adsorption experiment was carried out under the conditions of initial concentration C 0 = 20 mg/L, pH = 7, and temperature (T) of 298 K with an adsorbent amount of 0.01 g. The contact times were 1, 3,6,10,15,30,45,60,90,120,150,180,210,240,270, 300 and 360 min. Thermodynamic adsorption experiments were carried out at 298, 308 and 318 K with an adsorbent amount of 0.01 g, pH = 7 and C 0 of 10, 20, 30, 40 and 50 mg/g. Before each experiment, to prevent the hydrolysis of mercury ions, 0.1 mL of HCl and 10 mL of HNO3 in 1000 mL pure water was used as a protective solution. A certain amount of HgCl2 was added to the protective solution to obtain a 1000 mg/L mercury stock solution. The effect of CoFe2O4@SiO2-EDTA on the adsorption of the mercury solution was studied at different pH levels. 0.01 g of the adsorbent was added into an Erlenmeyer flask containing 100 mL Hg(II) solution (C0 = 20 mg/L), and then the pH was adjusted to 2-7 using 0.1 mol/L HCl and 0.1 mol/L NaOH [34]. After shaking for 12 h, the mixture was filtered through a 0.45 µm filter and analyzed by a cold atomic absorption spectrophotometry. All experiments were performed in triplicate and then averaged. The kinetic adsorption experiment was carried out under the conditions of initial concentration C0 = 20 mg/L, pH = 7, and temperature (T) of 298 K with an adsorbent amount of 0.01 g.

Characterizations
From the SEM and TEM images (Figures 2 and 3), it can be known that the adsorbents of CoFe2O4, CoFe2O4@SiO2 and CoFe2O4@SiO2-EDTA have the diameters of about 50-90, 90-130 and 110-200 nm. In Figure 2a, CoFe2O4 nanomaterial is irregular, and its diameter is smaller than that of CoFe2O4@SiO2 in Figure 2b. The SEM image clearly shows that the surface of CoFe2O4@SiO2 is smoother than that of CoFe2O4, which indicates that SiO2 is successfully supported on the surface of CoFe2O4.
After the material was modified by EDTA, its dispersibility was greatly improved with the material becoming noticeably fluffy, as shown in Figure 2c. In the TEM image of Figure 2d, Black magnetic is covered by outer silicon dioxide. Moreover, the transparent polymer coated with SiO2 and has a distinct core-shell structure. The energy dispersive spectrometer (EDS) mapping technology was used to analyse the elements contained in the material. Figure 3 shows the images of various elements in the material of CoFe2O4@SiO2-EDTA. From the image, not only the elements of Co and Fe are evenly distributed on the surface of the material, but also the other four elements of C, N, O and Si appear correspondingly and are evenly distributed, which confirms that the EDTA is successfully loaded onto the particles of the magnetic nanomaterial CoFe2O4@SiO2. Figure 4a shows the XRD patterns of the prepared materials CoFe2O4, CoFe2O4@SiO2 and CoFe2O4@SiO2-EDTA, revealing their microcrystalline structure. The peak characteristics in the XRD pattern of CoFe2O4 are consistent with the JCPDS file (22-1086) [35]. Moreover, the diffraction peaks

Characterizations
From the SEM and TEM images (Figures 2 and 3 Figure 2a, CoFe 2 O 4 nanomaterial is irregular, and its diameter is smaller than that of CoFe 2 O 4 @SiO 2 in Figure 2b. The SEM image clearly shows that the surface of CoFe 2 O 4 @SiO 2 is smoother than that of CoFe 2 O 4 , which indicates that SiO 2 is successfully supported on the surface of CoFe 2 O 4 . After the material was modified by EDTA, its dispersibility was greatly improved with the material becoming noticeably fluffy, as shown in Figure 2c. In the TEM image of Figure 2d, Black magnetic is covered by outer silicon dioxide. Moreover, the transparent polymer coated with SiO 2 and has a distinct core-shell structure. The energy dispersive spectrometer (EDS) mapping technology was used to analyse the elements contained in the material. Figure 3 shows the images of various elements in the material of CoFe 2 O 4 @SiO 2 -EDTA. From the image, not only the elements of Co and Fe are evenly distributed on the surface of the material, but also the other four elements of C, N, O and Si appear correspondingly and are evenly distributed, which confirms that the EDTA is successfully loaded onto the particles of the magnetic nanomaterial CoFe 2 O 4 @SiO 2 . Figure 4a shows the XRD patterns of the prepared materials CoFe 2 O 4 , CoFe 2 O 4 @SiO 2 and CoFe 2 O 4 @SiO 2 -EDTA, revealing their microcrystalline structure. The peak characteristics in the XRD pattern of CoFe 2 O 4 are consistent with the JCPDS file   [35]. Moreover, the diffraction peaks in the CoFe 2 O 4 @SiO 2 and CoFe 2 O 4 @SiO 2 -EDTA are similar to those of CoFe 2 O 4 and no other peaks are observed, indicating that the crystal structure in the material is either silicon-coated or carboxyl-functionalized. The functional groups and chemical bonds in the material were analyzed using FT-IR spectroscopy. According to Figure 4b, the broad peaks at 3472 and 1612 cm −1 correspond to the -OH stretching vibration peaks in the water molecules on the surface of the material [36]. in the CoFe2O4@SiO2 and CoFe2O4@SiO2-EDTA are similar to those of CoFe2O4 and no other peaks are observed, indicating that the crystal structure in the material is either silicon-coated or carboxylfunctionalized. The functional groups and chemical bonds in the material were analyzed using FT-IR spectroscopy. According to Figure 4b, the broad peaks at 3472 and 1612 cm −1 correspond to the -OH stretching vibration peaks in the water molecules on the surface of the material [36]. In the CoFe2O4@SiO2 and CoFe2O4@SiO2-EDTA spectra, a new broad absorption peak at 1092 cm −1 is the stretching vibration peak of Si-O-Si [4], indicating that silica has been successfully coated on CoFe2O4. In the CoFe2O4@SiO2-EDTA spectrum, there is no common carboxyl peak at 1732 cm −1 , but a new absorption peak appears at 1405 cm −1 [37]. This peak may be caused by EDTA modification of -COO − and symmetrical stretching vibration of C-O in the group [38]. The newly emerging absorption peak at 1623 cm −1 may be attributed to the presence of an amide bond [39]. The results indicate that EDTA has been successfully grafted onto silica surface. In the CoFe 2 O 4 @SiO 2 and CoFe 2 O 4 @SiO 2 -EDTA spectra, a new broad absorption peak at 1092 cm −1 is the stretching vibration peak of Si-O-Si [4], indicating that silica has been successfully coated on CoFe 2 O 4 . In the CoFe 2 O 4 @SiO 2 -EDTA spectrum, there is no common carboxyl peak at 1732 cm −1 , but a new absorption peak appears at 1405 cm −1 [37]. This peak may be caused by EDTA modification of -COO − and symmetrical stretching vibration of C-O in the group [38]. The newly emerging absorption peak at 1623 cm −1 may be attributed to the presence of an amide bond [39]. The results indicate that EDTA has been successfully grafted onto silica surface.   Figure 5 shows the hysteresis loops of the three materials. The saturation magnetizations of CoFe 2 O 4 , CoFe 2 O 4 @SiO 2 and CoFe 2 O 4 @SiO 2 -EDTA are 58.94, 11.94 and 7.65 emu/g, respectively. CoFe 2 O 4 shows typical super-paramagnetism. After coated by SiO 2 , the magnetic property of CoFe 2 O 4 @SiO 2 is obviously weakened due to the existence of silicon shell. After modified by EDTA, the thickness of the surface organic portion of CoFe 2 O 4 @SiO 2 -EDTA increases, forming a non-magnetic carboxylic acid functional layer. Although the magnetic properties are significantly reduced, the magnetic material can still be easily separated from aqueous solution. Note that the synthesis of the magnetic materials facilitates solid-liquid separation, making the entire adsorption process more convenient.   Figure 5 shows the hysteresis loops of the three materials. The saturation magnetizations of CoFe2O4, CoFe2O4@SiO2 and CoFe2O4@SiO2-EDTA are 58.94, 11.94 and 7.65 emu/g, respectively. CoFe2O4 shows typical super-paramagnetism. After coated by SiO2, the magnetic property of CoFe2O4@SiO2 is obviously weakened due to the existence of silicon shell. After modified by EDTA, the thickness of the surface organic portion of CoFe2O4@SiO2-EDTA increases, forming a nonmagnetic carboxylic acid functional layer. Although the magnetic properties are significantly reduced, the magnetic material can still be easily separated from aqueous solution. Note that the synthesis of the magnetic materials facilitates solid-liquid separation, making the entire adsorption process more convenient.   N2 desorption-desorption measurement was used to investigate the structural properties and specific surface area (BET). Figure 6 shows the adsorption-desorption isotherm curves for the three materials. The adsorption data are listed in Table 1. The BET value of CoFe2O4 decreases after the loading of the silicon shell, possibly due to the dense silica coating, resulted in a decrease of the BET value. After the modification of EDTA with the TEOS added as an active agent, the BET value slightly decreases, probably because the functionalized groups and some extra silica occupy the surface. Due to the extra fluffy silica on the surface, the pore size increases slightly (from 7.73 to 13.02 nm).   Figure 5 shows the hysteresis loops of the three materials. The saturation magnetizations of CoFe2O4, CoFe2O4@SiO2 and CoFe2O4@SiO2-EDTA are 58.94, 11.94 and 7.65 emu/g, respectively. CoFe2O4 shows typical super-paramagnetism. After coated by SiO2, the magnetic property of CoFe2O4@SiO2 is obviously weakened due to the existence of silicon shell. After modified by EDTA, the thickness of the surface organic portion of CoFe2O4@SiO2-EDTA increases, forming a nonmagnetic carboxylic acid functional layer. Although the magnetic properties are significantly reduced, the magnetic material can still be easily separated from aqueous solution. Note that the synthesis of the magnetic materials facilitates solid-liquid separation, making the entire adsorption process more convenient.  N2 desorption-desorption measurement was used to investigate the structural properties and specific surface area (BET). Figure 6 shows the adsorption-desorption isotherm curves for the three materials. The adsorption data are listed in Table 1. The BET value of CoFe2O4 decreases after the loading of the silicon shell, possibly due to the dense silica coating, resulted in a decrease of the BET value. After the modification of EDTA with the TEOS added as an active agent, the BET value slightly decreases, probably because the functionalized groups and some extra silica occupy the surface. Due to the extra fluffy silica on the surface, the pore size increases slightly (from 7.73 to 13.02 nm). N 2 desorption-desorption measurement was used to investigate the structural properties and specific surface area (BET). Figure 6 shows the adsorption-desorption isotherm curves for the three materials. The adsorption data are listed in Table 1. The BET value of CoFe 2 O 4 decreases after the loading of the silicon shell, possibly due to the dense silica coating, resulted in a decrease of the BET value. After the modification of EDTA with the TEOS added as an active agent, the BET value slightly decreases, probably because the functionalized groups and some extra silica occupy the surface. Due to the extra fluffy silica on the surface, the pore size increases slightly (from 7.73 to 13.02 nm).
From the XPS spectrum of Figure 7a, the peaks of Co 2p, Fe 2p and O 1s in CoFe 2 O 4 , can be observed, and the Si 2p peak appears in the modified materials of CoFe 2 O 4 @SiO 2 and CoFe 2 O 4 @SiO 2 . The photoelectrons of N 1s and C 1s can be attributed to the peaks that occur after EDTA-modified grafting. Moreover, high-resolution scanning of Co 2p, Fe 2p, Si 2p, N 1s, O 1s, and C 1s in the material is sequentially performed. The peaks of Co 2p 3/2 at 780.2 and 782.0 eV are shown in Figure 7b, along with the bond energy of Co 2p 1/2 at 796.5 eV with satellite characteristics at 786.5 and 803.4 eV, respectively. Nanomaterials 2019, 9, x FOR PEER REVIEW 8 of 23 From the XPS spectrum of Figure 7a, the peaks of Co 2p, Fe 2p and O 1s in CoFe2O4, can be observed, and the Si 2p peak appears in the modified materials of CoFe2O4@SiO2 and CoFe2O4@SiO2. The photoelectrons of N 1s and C 1s can be attributed to the peaks that occur after EDTA-modified grafting. Moreover, high-resolution scanning of Co 2p, Fe 2p, Si 2p, N 1s, O 1s, and C 1s in the material is sequentially performed. The peaks of Co 2p3/2 at 780. 2   Note that the peak at 796.5 eV is attributed to Co 3+ oxide [40]. The primary peak of Fe 2p in Figure  7c at 711.0 eV is attributed to Fe 3+ , whereas that at 724.3 eV belongs to Fe 2+ [41]. Figure 7d shows the change of Si before and after modification with EDTA. The peak of Si 2p appears at 103.2 eV before modification and at 102.3 eV after modification. The significant decrease of the Si 2p peak after modification can be explained by the successful grafting of EDTA. Figure 7e shows a scan of N 1s with binding energies of C-N and N-H in the N element at 400.9 and 398.9 eV, respectively.  [40,43]. In the C 1s scan of Figure 7g, the four different binding energy values observed at 283.5, 284.2, 285.2 and 287.5 eV correspond to C in C-Si, C-H, C-C/C-N and O-C=O, respectively. Thus, through XPS characterization, SiO2 is supported onto CoFe2O4 and EDTA is successfully grafted [44].
As shown in Figure 8, the equipotential (pHiep) of the adsorbent is ~3.6; therefore, the material's surface is negatively charged at pH ≥ 3.6. According to the experimental results, the maximum adsorption capacity of CoFe2O4@SiO2-EDTA at pH 7 is closely related to the electrostatic interaction between the surface negative charge and the Hg 2+ cation. When pH is lower than pHiep, the decrease in the adsorption capacity of CoFe2O4@SiO2-EDTA can be attributed to the positive charge on the surface of the material and the repulsive force between Hg 2+ and the adsorbent.  Note that the peak at 796.5 eV is attributed to Co 3+ oxide [40]. The primary peak of Fe 2p in Figure 7c at 711.0 eV is attributed to Fe 3+ , whereas that at 724.3 eV belongs to Fe 2+ [41]. Figure 7d shows the change of Si before and after modification with EDTA. The peak of Si 2p appears at 103.2 eV before modification and at 102.3 eV after modification. The significant decrease of the Si 2p peak after modification can be explained by the successful grafting of EDTA. Figure 7e shows a scan of N 1s with binding energies of C-N and N-H in the N element at 400.9 and 398.9 eV, respectively.  [40,43]. In the C 1s scan of Figure 7g, the four different binding energy values observed at 283.5, 284.2, 285.2 and 287.5 eV correspond to C in C-Si, C-H, C-C/C-N and O-C=O, respectively. Thus, through XPS characterization, SiO 2 is supported onto CoFe 2 O 4 and EDTA is successfully grafted [44].
As shown in Figure 8, the equipotential (pH iep ) of the adsorbent is~3.6; therefore, the material's surface is negatively charged at pH ≥ 3.6. According to the experimental results, the maximum adsorption capacity of CoFe 2 O 4 @SiO 2 -EDTA at pH 7 is closely related to the electrostatic interaction between the surface negative charge and the Hg 2+ cation. When pH is lower than pH iep , the decrease in the adsorption capacity of CoFe 2 O 4 @SiO 2 -EDTA can be attributed to the positive charge on the surface of the material and the repulsive force between Hg 2+ and the adsorbent. Binding energy (eV)

Effect of EDTA Addition Amount
Generally, the amounts of EDTA and the added intermediate crosslinking agent of APTES significantly affect the material properties ( Figure 9). The adsorption amount of CoFe2O4@SiO2-EDTA increases with increasing amounts of these additives. When the EDTA amount is 9 mmol, the adsorbent has a maximum adsorption capacity of qe of 103.3 mg/g. However, when the amount is less than 9 mmol, then the active site of the functional group remains on the surface of CoFe2O4@SiO2 so that the adsorption amount is reduced. Furthermore, when the amount added is greater than 9 mmol, the excess amount of EDTA will self-condense in the material, resulting in the coverage of the active groups, whereby the grafting ratio is lowered and the adsorption is decreased. Therefore, for subsequent experiments, EDTA amount of 9 mmol was selected for the subsequent study.

Effect of pH
As an important factor in adsorption, pH value directly affects the surface charge of the adsorbent surface, the existence morphology of heavy metal mercury ions and the stability of the interaction between the functional groups and the mercury ions. According to the data presented in Figure 10, with increasing pH value, the adsorption capacity for mercury ions with CoFe2O4@SiO2 is

Effect of EDTA Addition Amount
Generally, the amounts of EDTA and the added intermediate crosslinking agent of APTES significantly affect the material properties ( Figure 9). The adsorption amount of CoFe 2 O 4 @SiO 2 -EDTA increases with increasing amounts of these additives. When the EDTA amount is 9 mmol, the adsorbent has a maximum adsorption capacity of q e of 103.3 mg/g. However, when the amount is less than 9 mmol, then the active site of the functional group remains on the surface of CoFe 2 O 4 @SiO 2 so that the adsorption amount is reduced. Furthermore, when the amount added is greater than 9 mmol, the excess amount of EDTA will self-condense in the material, resulting in the coverage of the active groups, whereby the grafting ratio is lowered and the adsorption is decreased. Therefore, for subsequent experiments, EDTA amount of 9 mmol was selected for the subsequent study.

Effect of EDTA Addition Amount
Generally, the amounts of EDTA and the added intermediate crosslinking agent of APTES significantly affect the material properties ( Figure 9). The adsorption amount of CoFe2O4@SiO2-EDTA increases with increasing amounts of these additives. When the EDTA amount is 9 mmol, the adsorbent has a maximum adsorption capacity of qe of 103.3 mg/g. However, when the amount is less than 9 mmol, then the active site of the functional group remains on the surface of CoFe2O4@SiO2 so that the adsorption amount is reduced. Furthermore, when the amount added is greater than 9 mmol, the excess amount of EDTA will self-condense in the material, resulting in the coverage of the active groups, whereby the grafting ratio is lowered and the adsorption is decreased. Therefore, for subsequent experiments, EDTA amount of 9 mmol was selected for the subsequent study.

Effect of pH
As an important factor in adsorption, pH value directly affects the surface charge of the adsorbent surface, the existence morphology of heavy metal mercury ions and the stability of the interaction between the functional groups and the mercury ions. According to the data presented in Figure 10, with increasing pH value, the adsorption capacity for mercury ions with CoFe2O4@SiO2 is

Effect of pH
As an important factor in adsorption, pH value directly affects the surface charge of the adsorbent surface, the existence morphology of heavy metal mercury ions and the stability of the interaction between the functional groups and the mercury ions. According to the data presented in Figure 10, with increasing pH value, the adsorption capacity for mercury ions with CoFe 2 O 4 @SiO 2 is always low, but that of CoFe 2 O 4 @SiO 2 -EDTA quickly increases. At pH of 7, CoFe 2 O 4 @SiO 2 -EDTA has a maximum capacity of 103.3 mg/g for Hg(II). When the pH is alkaline, there is no need for detection because the carboxylic acid in EDTA is neutralized under alkaline conditions, and the adsorption capacity is considerably affected. Furthermore, a high pH value is not conducive for practical applications. At low pH, the solution contains a large amount of H + , the EDTA functional group is completely protonated and electrostatic repulsion occurs with the mercury ion, and EDTA exists in the form of -COOH while there is shortage of coordination sites, which weakens the complexation between functional groups and Hg 2+ , and leads to low adsorption capacity. always low, but that of CoFe2O4@SiO2-EDTA quickly increases. At pH of 7, CoFe2O4@SiO2-EDTA has a maximum capacity of 103.3 mg/g for Hg(II). When the pH is alkaline, there is no need for detection because the carboxylic acid in EDTA is neutralized under alkaline conditions, and the adsorption capacity is considerably affected. Furthermore, a high pH value is not conducive for practical applications. At low pH, the solution contains a large amount of H + , the EDTA functional group is completely protonated and electrostatic repulsion occurs with the mercury ion, and EDTA exists in the form of -COOH while there is shortage of coordination sites, which weakens the complexation between functional groups and Hg 2+ , and leads to low adsorption capacity.
As pH value is increased, the H + in the carboxylic acid gradually dissociates from the functional group, Until the EDTA functional group is completely deprotonated and exists in the form of -COO − , and the competitiveness of H + becomes weaker as its concentration decreases, thereby enhancing the affinity of adsorbent towards Hg 2+ . Thus, the adsorption of Hg(II) by CoFe2O4@SiO2-EDTA is mainly through ion exchange and chelation [45].

Effect of Dosage
As shown in Figure 11, the effect of adsorption capacity of mercury ions was researched with different dosage of CoFe2O4@SiO2-EDTA under the same conditions. As the dosage is increased, the adsorption capacity for Hg(II) is rapidly decreased, and the adsorption efficiency is rapidly increased. Thus, the increase of dosage provides more active sites for the combination of -COOH and Hg(II), resulting in a high removal rate. However, the effective utilization between the material and Hg(II) is reduced, thereby decreasing the adsorption capacity. As pH value is increased, the H + in the carboxylic acid gradually dissociates from the functional group, Until the EDTA functional group is completely deprotonated and exists in the form of -COO − , and the competitiveness of H + becomes weaker as its concentration decreases, thereby enhancing the affinity of adsorbent towards Hg 2+ . Thus, the adsorption of Hg(II) by CoFe 2 O 4 @SiO 2 -EDTA is mainly through ion exchange and chelation [45].

Effect of Dosage
As shown in Figure 11, the effect of adsorption capacity of mercury ions was researched with different dosage of CoFe 2 O 4 @SiO 2 -EDTA under the same conditions. As the dosage is increased, the adsorption capacity for Hg(II) is rapidly decreased, and the adsorption efficiency is rapidly increased. Thus, the increase of dosage provides more active sites for the combination of -COOH and Hg(II), resulting in a high removal rate. However, the effective utilization between the material and Hg(II) is reduced, thereby decreasing the adsorption capacity.  Figure 12a shows the adsorption curve of Hg(II) as a function of time. Rapid adsorption occurs in the first 30 min, and the adsorption capacity reaches 92.63 mg/g. The -COOH groups in the CoFe2O4@SiO2-EDTA provides rich active sites, thus leading to the complexation and rapid interaction of Hg(II) in the first 30 min [46]. Then adsorption begins to increase slowly and approaches maximum at 360 min with a capacity of 103.13 mg/g, owing to the presence of a large number of -COOH groups.   Figure 12a shows the adsorption curve of Hg(II) as a function of time. Rapid adsorption occurs in the first 30 min, and the adsorption capacity reaches 92.63 mg/g. The -COOH groups in the CoFe 2 O 4 @SiO 2 -EDTA provides rich active sites, thus leading to the complexation and rapid interaction of Hg(II) in the first 30 min [46]. Then adsorption begins to increase slowly and approaches maximum at 360 min with a capacity of 103.13 mg/g, owing to the presence of a large number of -COOH groups.  Figure 12a shows the adsorption curve of Hg(II) as a function of time. Rapid adsorption occurs in the first 30 min, and the adsorption capacity reaches 92.63 mg/g. The -COOH groups in the CoFe2O4@SiO2-EDTA provides rich active sites, thus leading to the complexation and rapid interaction of Hg(II) in the first 30 min [46]. Then adsorption begins to increase slowly and approaches maximum at 360 min with a capacity of 103.13 mg/g, owing to the presence of a large number of -COOH groups. To explore the adsorption behavior of CoFe 2 O 4 @SiO 2 -EDTA in detail, three types of kinetic models were used to fit the experimental data, including a pseudo-first-order kinetic model (Equation (1)), pseudo-second-order kinetic model (Equation (2)) and intra-particle diffusion model (Equation (3)):

Adsorption Kinetics
where q e (mg/g) is the equilibrium adsorption capacity; q t (mg/g) is the adsorption capacity at time t (min); k 1 and k 2 are the adsorption rate constants; k d is the intraparticle diffusion rate constant and C (mg/g) is the boundary layer thickness. Based on the fitting results of the above three kinetic models and Table 2, from the linear fitting in Figure 12b,c, the regression coefficient of the pseudo-first-order model (R 2 = 0.925) is less than that of the pseudo-second-order model (R 2 = 0.999). From the pseudo-first-order fitting diagram, it can be observed that the distributions of points are scattered and the fitting is poor. Kinetic adsorption is not accounted for in the pseudo-first-order model. However, in the pseudo-second-order model, the fitting is extremely high [47]. Moreover, the fitting value of q e,cal (103.62 mg/g) is closer to the experimental value of q e,exp (103.13 mg/g). Thus, it can be concluded that the adsorption of Hg(II) in an aqueous solution with CoFe 2 O 4 @SiO 2 -EDTA can be explained by the pseudo-second-order kinetic model, and the adsorption process is subjected to chemisorption [45]. This behavior refers to the complexation between the adsorbent and mercury ions and forms a chemical bond during the process of adsorption [48].  Figure 12d shows a fitting plot of the intra-particle diffusion model. The adsorption process is divided into two stages. The first-stage is large pore diffusion, corresponding to fast adsorption and high k d1 values. Finally, the second stage is equilibrium adsorption, corresponding to low adsorption rates and k d2 values. However, in the linear fitting, the straight line does not pass through the origin, indicating that intra-particle diffusion is not the only factor affecting the adsorption rate of mercury. Figure 13a shows the adsorption of mercury ions in an aqueous solution with the adsorbent CoFe 2 O 4 @SiO 2 -EDTA at different temperatures. According to the above data, the material has an optimal adsorption effect at 298 K and pH of 7. As the temperature increases, the adsorption curve gradually decreases, indicating that the material is more suitable for using under room temperature conditions. Note that the adsorption isotherm is important for optimizing the using of the adsorbent, as it can be used to evaluate the adsorption capacity of the adsorbent and describe how the adsorbent interacts with the adsorbates. The Langmuir isotherm model (Equation (4)), Freundlich isotherm model (Equation (5)), Temkin isotherm model (Equation (6)) and Dubinin-Radushkevich isotherm model (Equations (7)-(9)) can be obtained using the following equations:  The Langmuir isotherm model (Equation (4)), Freundlich isotherm model (Equation (5)), Temkin isotherm model (Equation (6)) and Dubinin-Radushkevich isotherm model (Equations (7)-(9)) can be obtained using the following equations:

Adsorption Isotherms
ln q e = ln K F = 1 n ln C e (5) ln q e = ln q m − βε 2 (7) where Q m (mg/g) is the saturated adsorption capacity; C e (mg/L) is the equilibrium concentration; K L (L/mg) is the Langmuir adsorption equilibrium constant; K F and K T are all constants. R is the gas constant (8.314 J/mol/K); T (K) is the temperature, ε is the Polanyi potential energy and β is a constant related to average adsorption energy E (kJ/mol). The separation constant R L can be used to describe the adsorption characteristics of the Langmuir adsorption isotherm model using Equation (10).
where R L is a separation constant, which is dimensionless, and C 0 (mg/L) is the initial concentration. The fitting maps for the Langmuir, Freundlich, Temkin and Dubinin-Radushkevich isothermal models are shown in Figure 13b-e, and the corresponding data are shown in Table 3. According to the fitting data, the three R 2 of the Langmuir model are higher than those of the Freundlich model under different temperatures, indicating that the adsorption process is a single molecule adsorption process [34]. Therefore, the adsorbent exhibits a chemisorption behavior [49]. This conclusion is consistent with the kinetic model. From Table 3, it can be observed that the maximum adsorption capacities of the materials fitted by the Langmuir isothermal model are 143.85, 138.12 and 111.23 mg/g, respectively, when the temperature is 298, 308 and 318 K. Therefore, as the temperature increases, the adsorption capacity gradually decreases. Moreover, the value of R L is between 0 and 1, which indicates that the divalent mercury ions are easily adsorbed by the material. In the Freundlich isotherm model, the 1/n values are also between 0 and 1, indicating that the adsorption process is proceeding in a favourable direction. The relatively high R 2 values and the K T values in the Temkin isotherm model at three temperatures indicate a strong interaction between CoFe 2 O 4 @SiO 2 -EDTA and mercury ions; while the increasing b T values with the rising temperature indicate that the adsorption capacity is gradually decreasing. The result confirms that the high temperature does not favour the reaction.
In the Dubinin-Radushkevich model, when the value of the mean free energy E < 8 kJ/mol, the reaction belongs to physical adsorption; however, when the values of E is between 8 and 16 kJ/mol, ion exchange takes place [50], and the chemisorption mechanism is functioning when the values of E is exceeds 16 kJ/mol. As the average free energy E of the three adsorbents is over 16 kJ/mol, chemisorption is involved in the adsorption process, which is consistent with the abovementioned conclusions.
As shown in Table 4, CoFe 2 O 4 @SiO 2 -EDTA had a smaller specific surface area and a higher adsorption capacity, which is better than most adsorbents.

Thermodynamics
The nature and mechanism of the adsorption process can be analyzed by thermodynamics. The standard Gibbs free energy (∆G 0 , kJ/mol), enthalpy change (∆H 0 , kJ/mol) and entropy change (∆S 0 , kJ/mol/K) are represented by Equations (11) and (12).
where K d is equilibrium constant. From thermodynamic fitting of Figure 14 and the thermodynamic fitting parameters of Table 5, the ∆H 0 values are negative at three different mercury ion concentrations, indicating that the process of adsorbing Hg 2+ by CoFe 2 O 4 @SiO 2 -EDTA is exothermic. All of the ∆S 0 are negative, indicating that the order of the solid-liquid interface increases during the process of adsorption [36], ∆G 0 are also negative, indicating a spontaneous adsorption. As temperature increases, ∆G 0 gradually decreases, indicating that high temperature is not conducive to the progress of the adsorption process. Based on the fitting results and data, the adsorption of mercury ions by the CoFe 2 O 4 @SiO 2 -EDTA is a spontaneous exothermic reaction.

Reusability
To examine the potential ability of CoFe2O4@SiO2-EDTA in practical applications, the regeneration of the adsorbent was evaluated by repeatedly treating CoFe2O4@SiO2-EDTA-Hg with 0.1 mol/L HCl as a regenerant [45]. The mixture is stirred for 4 h under acidic conditions. Figure 15 shows that the adsorption capacity of the material decreases after each regeneration cycle. After three cycles, >90% of the adsorbed Hg(II) still remains; the adsorption capacity decreases by ~14.5% after

Reusability
To examine the potential ability of CoFe 2 O 4 @SiO 2 -EDTA in practical applications, the regeneration of the adsorbent was evaluated by repeatedly treating CoFe 2 O 4 @SiO 2 -EDTA-Hg with 0.1 mol/L HCl as a regenerant [45]. The mixture is stirred for 4 h under acidic conditions. Figure 15 shows that the adsorption capacity of the material decreases after each regeneration cycle. After three cycles, >90% of the adsorbed Hg(II) still remains; the adsorption capacity decreases by~14.5% after five cycles, which was a favourable result. Thus, the regeneration studies show that the adsorbent of CoFe 2 O 4 @SiO 2 -EDTA has a good potential application in actual water treatment [34].

Reusability
To examine the potential ability of CoFe2O4@SiO2-EDTA in practical applications, the regeneration of the adsorbent was evaluated by repeatedly treating CoFe2O4@SiO2-EDTA-Hg with 0.1 mol/L HCl as a regenerant [45]. The mixture is stirred for 4 h under acidic conditions. Figure 15 shows that the adsorption capacity of the material decreases after each regeneration cycle. After three cycles, >90% of the adsorbed Hg(II) still remains; the adsorption capacity decreases by ~14.5% after five cycles, which was a favourable result. Thus, the regeneration studies show that the adsorbent of CoFe2O4@SiO2-EDTA has a good potential application in actual water treatment [34].

Mechanism Speculation
To study the adsorption mechanism of CoFe2O4@SiO2-EDTA towards Hg(II) and analysis was carried out by using XPS technology. It can be seen from Figure 16a, a new peak of Hg appears after adsorption. The peaks of Hg 4f5/2 at 104.2 eV and of Hg 4f7/2 at 100.2 eV (Figure 16b) are attributed to the adsorbed HgCl2 [57]. It clearly shows that Hg is successfully adsorbed by CoFe2O4@SiO2-EDTA. In Figure 16c, the O-C=O and C-N/C-C are transferred to 285.5 and 287.8 eV, respectively, compared to the spectrum before adsorption [44]. The reason is the formation of coordination bonds between the carboxylic acid and Hg(II) on the surface of the material. Therefore, XPS spectra indirectly confirmed that a stable complex was formed by chelation between CoFe2O4@SiO2-EDTA and mercury ions during the process of adsorption.

Mechanism Speculation
To study the adsorption mechanism of CoFe 2 O 4 @SiO 2 -EDTA towards Hg(II) and analysis was carried out by using XPS technology. It can be seen from Figure 16a, a new peak of Hg appears after adsorption. The peaks of Hg 4f 5/2 at 104.2 eV and of Hg 4f 7/2 at 100.2 eV (Figure 16b) are attributed to the adsorbed HgCl 2 [57]. It clearly shows that Hg is successfully adsorbed by CoFe 2 O 4 @SiO 2 -EDTA. In Figure 16c, the O-C=O and C-N/C-C are transferred to 285.5 and 287.8 eV, respectively, compared to the spectrum before adsorption [44]. The reason is the formation of coordination bonds between the carboxylic acid and Hg(II) on the surface of the material. Therefore, XPS spectra indirectly confirmed that a stable complex was formed by chelation between CoFe 2 O 4 @SiO 2 -EDTA and mercury ions during the process of adsorption.
Compared the morphology of the different types of Hg(II) in aqueous solution and the binding sites that can be utilized, it can be observed that the adsorption effect is not very different. When the solution contains Cl − , the form of Hg(II) in the solution, such as HgCl 2 , HgCl + and Hg(OH) 2 , changes with pH value [16]. Under strongly acidic conditions, the surface of the adsorbent is negatively charged, and the EDTA functional group on the surface of the adsorbent is protonated, which limits carboxyl binding to Hg 2+ .
However, the surface of the adsorbent is positively charged with the increase of pH, and the presence of EDTA exhibits a strong chelating ability to metal ions. Mercury ions in the solution gradually form hydroxy mercury (HgOH and Hg(OH) 2 ) [58] and interaction with the adsorbent occurs, there is no static repulsion, and the adsorption effect quickly reaches the maximum value.
sites that can be utilized, it can be observed that the adsorption effect is not very different. When the solution contains Cl − , the form of Hg(II) in the solution, such as HgCl2, HgCl + and Hg(OH)2, changes with pH value [16]. Under strongly acidic conditions, the surface of the adsorbent is negatively charged, and the EDTA functional group on the surface of the adsorbent is protonated, which limits carboxyl binding to Hg 2+ .
However, the surface of the adsorbent is positively charged with the increase of pH, and the presence of EDTA exhibits a strong chelating ability to metal ions. Mercury ions in the solution gradually form hydroxy mercury (HgOH and Hg(OH)2) [58] and interaction with the adsorbent occurs, there is no static repulsion, and the adsorption effect quickly reaches the maximum value. Binding energy (eV) Figure 16. XPS spectra of survey scan of (a) and high-resolution scan of Hg 4f (b) and C 1s (c).
The adsorption mechanism of mercury by CoFe2O4@SiO2-EDTA is shown in Figure 17. A complex with cyclic structure is formed from the central COO − ions and the N-based ligand under the chelation. During the process of adsorption, there are two situations in Figure 17. By EDTAfunctionalized, the adsorbent forms a symmetric ligand or contain a free carboxyl group. Under these conditions, a stable complex is often formed by H2O and another bond, but H2O is usually replaced by an imine group on CoFe2O4@SiO2-EDTA to form a more stable six-position complex [34,37]. Understanding the complex mechanisms in the adsorption process requires collecting data to determine rates, steps and related parameters. Here, the removal of Hg(II) by CoFe2O4@SiO2-EDTA is evaluated considering different adsorption kinetics and adsorption isotherm models. The adsorption process of metal ions can be divided into two steps [58].
The first process involves the adsorption of Hg ions to the active sites on the surface of the material through functional groups, during which physical or chemical adsorption occurs. The experimental data show excellent adaptation in the pseudo-second-order kinetic and the Langmuir Figure 16. XPS spectra of survey scan of (a) and high-resolution scan of Hg 4f (b) and C 1s (c).
The adsorption mechanism of mercury by CoFe 2 O 4 @SiO 2 -EDTA is shown in Figure 17. A complex with cyclic structure is formed from the central COO − ions and the N-based ligand under the chelation. During the process of adsorption, there are two situations in Figure 17. By EDTA-functionalized, the adsorbent forms a symmetric ligand or contain a free carboxyl group. Under these conditions, a stable complex is often formed by H 2 O and another bond, but H 2 O is usually replaced by an imine group on CoFe 2 O 4 @SiO 2 -EDTA to form a more stable six-position complex [34,37]. Understanding the complex mechanisms in the adsorption process requires collecting data to determine rates, steps and related parameters. Here, the removal of Hg(II) by CoFe 2 O 4 @SiO 2 -EDTA is evaluated considering different adsorption kinetics and adsorption isotherm models. The adsorption process of metal ions can be divided into two steps [58].
The first process involves the adsorption of Hg ions to the active sites on the surface of the material through functional groups, during which physical or chemical adsorption occurs. The experimental data show excellent adaptation in the pseudo-second-order kinetic and the Langmuir isotherm models [59]. The homogeneous active site on the adsorbent CoFe 2 O 4 @SiO 2 -EDTA is a single molecule adsorption process for mercury ions. Therefore, chemical adsorption is the driving force in the rate-control step of the adsorption process. Moreover, the value of the mean free energy E in the Dubinin-Radushkevich isothermal model confirms this. Furthermore, this illustrates the complexation between the metal ions and the active sites on the adsorbent during the process of adsorption. A stable complex is formed by chelation between a carboxylic acid and Hg 2+ . To further understand the adsorption mechanism, the FT-IR spectra after adsorption show that the tensile vibration at 1646 cm −1 is lost. The result indicates that the carbonyl group of EDTA interacts with Hg(II). It can be seen from Figure 18, with the disappearance of the vibrational peak of carboxylic acid at 1403 cm −1 after adsorption, the peaks at 1353 and 1384 cm −1 are attributed to the adsorption of the Hg carboxylic acid [38]. This group forms a new chemical bond with Hg, resulting in a shift in the peak. The change of the peak value in the FT-IR spectrum also indirectly confirms that the adsorption process of mercury ions in aqueous solution by CoFe 2 O 4 @SiO 2 -EDTA is chemical adsorption. Finally, Figures 15 and 16 show that the CoFe 2 O 4 @SiO 2 -EDTA material before and after adsorption does not significantly change, thereby indicating that the material has good hydrothermal stability. molecule adsorption process for mercury ions. Therefore, chemical adsorption is the driving force in the rate-control step of the adsorption process. Moreover, the value of the mean free energy E in the Dubinin-Radushkevich isothermal model confirms this. Furthermore, this illustrates the complexation between the metal ions and the active sites on the adsorbent during the process of adsorption. A stable complex is formed by chelation between a carboxylic acid and Hg 2+ . To further understand the adsorption mechanism, the FT-IR spectra after adsorption show that the tensile vibration at 1646 cm −1 is lost. The result indicates that the carbonyl group of EDTA interacts with Hg(II). It can be seen from Figure 18, with the disappearance of the vibrational peak of carboxylic acid at 1403 cm −1 after adsorption, the peaks at 1353 and 1384 cm −1 are attributed to the adsorption of the Hg carboxylic acid [38]. This group forms a new chemical bond with Hg, resulting in a shift in the peak. The change of the peak value in the FT-IR spectrum also indirectly confirms that the adsorption process of mercury ions in aqueous solution by CoFe2O4@SiO2-EDTA is chemical adsorption. Finally, Figures 15 and 16 show that the CoFe2O4@SiO2-EDTA material before and after adsorption does not significantly change, thereby indicating that the material has good hydrothermal stability.   the rate-control step of the adsorption process. Moreover, the value of the mean free energy E in the Dubinin-Radushkevich isothermal model confirms this. Furthermore, this illustrates the complexation between the metal ions and the active sites on the adsorbent during the process of adsorption. A stable complex is formed by chelation between a carboxylic acid and Hg 2+ . To further understand the adsorption mechanism, the FT-IR spectra after adsorption show that the tensile vibration at 1646 cm −1 is lost. The result indicates that the carbonyl group of EDTA interacts with Hg(II). It can be seen from Figure 18, with the disappearance of the vibrational peak of carboxylic acid at 1403 cm −1 after adsorption, the peaks at 1353 and 1384 cm −1 are attributed to the adsorption of the Hg carboxylic acid [38]. This group forms a new chemical bond with Hg, resulting in a shift in the peak. The change of the peak value in the FT-IR spectrum also indirectly confirms that the adsorption process of mercury ions in aqueous solution by CoFe2O4@SiO2-EDTA is chemical adsorption. Finally, Figures 15 and 16 show that the CoFe2O4@SiO2-EDTA material before and after adsorption does not significantly change, thereby indicating that the material has good hydrothermal stability.

Conclusions
In this study, a core-shell structure of CoFe 2 O 4 @SiO 2 is successfully functionalized with EDTA by a safe, mild and easy hydrothermal method. The as-prepared material has a saturation magnetic strength of 7.65 emu/g. Moreover, the material has a maximum adsorption capacity of 103.3 mg/g for metal Hg(II) at pH 7, and rapid separation of the adsorbent from the solution by magnetism. The adsorption process has excellent correlation with pseudo-second-order kinetics and the Langmuir isotherm model; moreover, it is a single-layer adsorption and a spontaneous exothermic reaction. A stable complex is formed between the EDTA functional group and the Hg(II) ion by chelation, and the chemical reaction is the key to the rate control step of the adsorption process. To summarize, EDTA-Functionalized Magnetic CoFe 2 O 4 @SiO 2 Nanomaterial has good hydrothermal stability, recyclability and fully biodegradable properties, indicating that it is a potential adsorbent for removing heavy metals from water in practical applications.