A Generalized Method for the Synthesis of Carbon-Encapsulated Fe3O4 Composites and Its Application in Water Treatment

In this paper, a simple and environmentally friendly method was developed for the preparation of highly stable C@Fe3O4 composites with controllable morphologies using sodium alginate as the carbon source and the easily obtained α-Fe2O3 as the precursors. The morphologies of the as-prepared C@Fe3O4 composites, inherited from their corresponding precursors of α-Fe2O3, survived from the annealing treatments, were characterized by the field-emission scanning electron microscopy (FESEM), transmission electron microscopy (TEM), X-ray diffraction (XRD) and inductively coupled plasma-atomic emission spectroscopy (ICP-AES). The C@Fe3O4 composites resisted to oxidation, acidification and aggregation, exhibiting porous structures and ferromagnetic properties at room temperature. Moreover, the adsorption performance of the C@Fe3O4 composites was evaluated by absorbing MB (methylene blue) in liquid environment. Experiments indicated that the C@Fe3O4 composites exhibited highly enhanced adsorption capacities and efficiencies as compared with their corresponding precursors of α-Fe2O3. This generalized method for the synthesis of C@Fe3O4 composites provides promising applications for the highly efficient removal of MB from industrial effluents.


Introduction
Among the various kinds of contaminants causing water pollution, dye pollution is becoming more serious in our environment and has become the focal point of research regarding the future of wastewater treatment, since organic dyes have been ubiquitously utilized in various fields, including plastics, leather, textiles, and cosmetics, etc. [1,2]. However, the unrestricted applications of these organic compounds result in serious environmental pollution because of their resistance to decomposition [3,4]. Particularly, as a representative organic dye, the removal of methylene blue (MB) from wastewater is an environmental issue and a lot of research studies have been carried out in this field. MB is a heterocyclic aromatic compound with extensive use in the textile industry and it is also present in the effluents of wastewater coming from other industries [5]. The acute exposure to MB may cause some harmful effects such as increased heart rate, shock, vomiting, jaundice, and tissue necrosis in humans [6,7]. Hence, it is essential to remove MB from industrial wastewater. Among various physical and chemical approaches, the adsorption technique is the most widely used method because of its simplicity, flexibility, cost-effectiveness and high efficiency for the dye removal [8][9][10]. Recently, magnetic adsorbent materials have gained special attention in water purification, based on their numerous advantages such as high separation efficiency, simple manipulation process, and mild operation conditions [11][12][13][14][15]. As one of the spinel ferrites, Co 0.5 Zn 0.5 Fe 2 O 4 nanoparticles are significant magnetic nano-adsorbents because of their excellent surface properties and tunable size. Hence, Lv and his colleagues synthesized the magnetic Co 0.5 Zn 0.5 Fe 2 O 4 nanoparticles via the facile nitrate-alcohol solution combustion process, which was further applied to remove methyl blue (MB) from an aqueous medium with its maximum adsorption capacity of 98.5 mg·g −1 [11]. Moreover, because of the excellent magnetism, electrical property, and adsorption capacity, Liu and his co-workers also prepared the magnetic Ni 0.3 Co 0.4 Zn 0.3 Fe 2 O 4 nanoparticles through a rapid-combustion process to adsorb MB from aqueous solution, and its maximum adsorption capacity was up to 101.1 mg·g −1 [12]. In addition, to achieve the rapid recovery of the adsorbent by an external magnetic field, Zhou et al. have reported the adsorption of MB using synthesized lignin-based magnetic Fe 3 O 4 nanoparticles and the maximum adsorption capacity of MB dye was 234.27 mg·g −1 [13]. Similarly, to take advantage of the simple magnetic separation, Iram et al. also synthesized the Fe 3 O 4 hollow nano-spheres via a one-pot simple template-free hydrothermal method to remove neutral red dye contaminants from the aqueous solution [14]. Additionally, Sharma and his co-workers have also applied the magnetic Fe 3 O 4 nanoparticles to remove the new coccine from aqueous solutions to achieve the rapid recovery of the adsorbent by an external magnetic field [15].
Although these ferrite materials based on Fe 3 O 4 nanoparticles have drawn considerable attention in wastewater treatment, they turned out not to be of practical application because of some drawbacks. It is well known that the Fe 3 O 4 is highly susceptible to oxidative and acid conditions and can easily agglomerate into bigger clusters, which result in losing the specific properties of the magnetic nanoparticles and decreasing adsorption capacities [16]. Hence, covering Fe 3 O 4 nanoparticles with a protective layer was beneficial to improving its stability. One of the most effective and feasible approaches is to cover the Fe 3 O 4 nanoparticle's surface with a carbon layer, which could protect it from being oxidized and reducing the magnetic interactions of magnetic nanoparticles.
To date, a series of carbon sources have been utilized to prepare carbon-encapsulated magnetic nanoparticles. For instance, Wang et al. synthesized the Fe 3 O 4 @C nanoparticles through an in situ solid-phase method to decolorize high concentrations of MB [17]. Kong et al. successfully synthesized Fe 3 O 4 @C nanoparticles using glucose as the carbon resource to fabricate an adsorbent for the removal of MB and cresol red (CR) [18]. Ai and his co-workers have prepared the graphene nanosheet (GNS)/magnetite (Fe 3 O 4 ) composite by in situ converting FeCl 3 to Fe 3 O 4 and reducing graphene oxide (GO) into graphene simultaneously in an ethylene glycol solution for the removal of MB [19]. Banerjee and his colleagues also prepared a porous adsorbent of a carbon-Fe 3 O 4 nanocomposite by using benzene dicarboxylic acid (BDC) as a carbon resource for the removal of MB using [20]. Although these carbon-coated ferromagnetic nanoparticles have been well developed, their practical applications are still limited for all kinds of reasons, such as high pollution, high-energy consumption, tedious synthetic steps, and low yields during the fabrication process. Thus, to meet the growing need of C@Fe 3 O 4 composites, searching for suitable and "green" carbon sources is still significant to assure safe reactants and to avoid environmental pollution in the preparative process.
Currently, as far as we know, investigations on the preparation of C@Fe 3 O 4 compound materials were rarely carried out using sodium alginate as the carbon source which is an inexpensive, non-toxic and renewable bio-resource belonging to the concept of "green chemistry". Moreover, α-Fe 2 O 3 , which takes the role of the most stable ferrite oxides in an environmental condition, has been extensively applied in many fields [21], such as adsorption [22,23], catalysis [24,25], sensors [26], nanofluids [27], and lithium-ion batteries [28,29] because of its low cost, safety, and it is environmentally benign. Therefore, to protect Fe 3 O 4 from being oxidized and applying suitable and "green" carbon sources, a simple and environmentally friendly method has been developed for the preparation of highly stable dispersions of C@Fe 3 O 4 composites with controlled morphologies, including nanorods, nanodisks, and nanoparticles, using sodium alginate as the carbon source and the easily obtained α-Fe 2 O 3 as the precursors. The α-Fe 2 O 3 precursors were successfully prepared through a simple hydrothermal method, followed by an ionic polymerization to produce α-Fe 2 O 3 /carbon shell beads. The dried up α-Fe 2 O 3 /carbon shell beads obtained above were calcined at 400 • C for 2 h in a tube furnace and acid treatment, obtaining the high yields of C@Fe 3 O 4 composites. These C@Fe 3 O 4 composites exhibit the similar morphologies as their precursors as well as a ferromagnetic characteristic at room temperature with a higher coercivity (Hc) value. A high-enhanced adsorption capacity and efficiency for the removal of MB were also observed in comparison with the precursors of α-Fe 2 O 3 . In particular, the commercially available α-Fe 2 O 3 materials were also applied to serve as the precursor mainly based on the following considerations. Firstly, it is used to verify whether this method can be generally applied to prepare similar C@Fe 3 O 4 composites with different kinds of precursors. Secondly, it is used to compare the adsorption performance of C@Fe 3 O 4 composites for MB removal by using different kinds of precursors and further to confirm the morphology effect on the adsorption performance.

Properties of the Synthesized Samples
XRD data were obtained to establish the composition and phase purity of the asprepared α-Fe 2 O 3 precursors, as shown in Figure 1a. The XRD pattern of α-FeOOH confirms that the initial yellow products after the hydrothermal synthesis are pure orthorhombic phase goethite iron oxide (α-FeOOH, JCPDS No. 81-0463). At the same time, the cell parameters of α-FeOOH, obtained through the least square fitting method, are a = 4.596 Å, b = 9.837 Å, and c = 3.197 Å, which further corroborate a similar structure as α-FeOOH (JCPDS No. 81-0463). The position and relative intensity of the reflection peaks of the α-Fe 2 O 3 nanorod and nanodisk confirm the structure of the hematite, which agree well with the XRD pattern of α-  Figure 1b,c shows the SEM images of iron-based samples. It can be clearly seen that all these samples have uniform and smooth surfaces. The diameter of the α-FeOOH nanorod is measured to be 120 nm, while 110 nm for the α-Fe 2 O 3 nanorod, which can be attributed to the annealing treatment, making the structure of α-Fe 2 O 3 nanorod much tighter. The disk-like structure of α-Fe 2 O 3 is observed with a diameter of around 175 nm, as depicted in Figure 1d. The size and morphology of the iron-based samples were further examined using transmission electron microscopy (TEM). Figure 2a also clearly displays a rod-like morphology with its porous structure of the α-FeOOH nanorod, of which the lattice fringes are measured by HRTEM (Figure 2d) with d-spacings of 0.261 and 0.262 nm, corresponding to the spacings of the (0 −2 1) and (0 2 1) planes of α-FeOOH, respectively.  The size and morphology of the iron-based samples were further examined using transmission electron microscopy (TEM). Figure 2a also clearly displays a rod-like morphology with its porous structure of the α-FeOOH nanorod, of which the lattice fringes are measured by HRTEM (Figure 2d) with d-spacings of 0.261 and 0.262 nm, corresponding to the spacings of the (0 −2 1) and (0 2 1) planes of α-FeOOH, respectively. Similarly, the porous structure of the α-Fe 2 O 3 nanorod is clearly shown in Figure 2b, as well as the Brunauer-Emmett-Teller (BET) measurements in Supplementary Materials ( Figure S1). The corresponding disk-like structure of the α-Fe 2 O 3 nanodisk is also depicted in Figure 2c. The representative HRTEM images taken from the α-Fe 2 O 3 nanorod and nanodisk sample are shown in Figure 2e Figure 3a. Moreover, there are no other diffraction peaks that have originated from other phases, suggesting that the as-synthesized samples have higher purity. However, due to the similar patterns between γ-Fe2O3 and Fe3O4, the XRD analyses cannot provide enough evidence to confirm the transformation [30]. Thus, the Raman measurements are provided for further structural confirmation, which is also a frequently used and non-ruinous method to characterize carbonaceous material, especially applied to certify the degree of graphitization [31]. The corresponding Raman spectra are shown in Figure 3b-d. The weak peaks at 666, 667, and 655 cm −1 of three C@Fe3O4 composites are attributed to the A1g mode of magnetite, according to Phase et al. [32,33], suggesting that the main component of C@Fe3O4 composites is Fe3O4, although no visible peak is recorded at 540 cm −1 . The peaks at 1372, 1385, and 1400 cm −1   Figure 3a. Moreover, there are no other diffraction peaks that have originated from other phases, suggesting that the as-synthesized samples have higher purity. However, due to the similar patterns between γ-Fe 2 O 3 and Fe 3 O 4 , the XRD analyses cannot provide enough evidence to confirm the transformation [30]. Thus, the Raman measurements are provided for further structural confirmation, which is also a frequently used and non-ruinous method to characterize carbonaceous material, especially applied to certify the degree of graphitization [31]. The corresponding Raman spectra are shown in  [32,33], suggesting that the main component of C@Fe 3 O 4 composites is Fe 3 O 4 , although no visible peak is recorded at 540 cm −1 . The peaks at 1372, 1385, and 1400 cm −1 are associated with the vibrational modes of carbon atoms with dangling bonds for the in-plane-terminated and disordered graphite, labeled as the D band. The corresponding G band peaks are found at 1594, 1594, and 1595 cm −1 , corresponding to the E 2g mode for the vibration of all sp 2 -bonded carbon atoms in a two-dimensional (2D) structure [31]. The degree of graphitization can be ascertained from the I D /I G value (relative intensity ratio of the D-G band) [16]. Furthermore, the I D /I G values of the three samples are all approximately 0.8, suggesting that these carbon materials have a high degree of disordered graphitization, which is coincident with the XRD analysis results.  Similarly, the structures of the C@Fe3O4 composites were also characterized by SEM and TEM observations. As shown in Figure 4a-c, it can be clearly observed that the morphologies of the precursors were not damaged during the process of calcination. However, the structures of the C@Fe3O4 composites became much tighter after annealing treatment in comparison to that of the precursors, which can be attributed to the presence of the carbon layer on the surface of the C@Fe3O4 composites. Meanwhile, the surfaces of the C@Fe3O4 composites are getting rougher in comparison with pure α-Fe2O3. Magnified TEM images further demonstrate that the Fe3O4 particles are surrounded by a thick carbon layer (Figure 4d-f). Thus, the advent of the broad XRD peaks in Figure 3a is ascribed to the presence of the nanosized Fe3O4 modified by a carbon layer. A similar phenomenon was also found in previous research [34]. Similarly, the structures of the C@Fe 3 O 4 composites were also characterized by SEM and TEM observations. As shown in Figure 4a-c, it can be clearly observed that the morphologies of the precursors were not damaged during the process of calcination. However, the structures of the C@Fe 3 O 4 composites became much tighter after annealing treatment in comparison to that of the precursors, which can be attributed to the presence of the carbon layer on the surface of the C@Fe 3 O 4 composites. Meanwhile, the surfaces of the C@Fe 3 O 4 composites are getting rougher in comparison with pure α-Fe 2 O 3 . Magnified TEM images further demonstrate that the Fe 3 O 4 particles are surrounded by a thick carbon layer (Figure 4d-f). Thus, the advent of the broad XRD peaks in Figure 3a is ascribed to the presence of the nanosized Fe 3 O 4 modified by a carbon layer. A similar phenomenon was also found in previous research [34]. treatment in comparison to that of the precursors, which can be attributed to the presence of the carbon layer on the surface of the C@Fe3O4 composites. Meanwhile, the surfaces of the C@Fe3O4 composites are getting rougher in comparison with pure α-Fe2O3. Magnified TEM images further demonstrate that the Fe3O4 particles are surrounded by a thick carbon layer (Figure 4d-f). Thus, the advent of the broad XRD peaks in Figure 3a is ascribed to the presence of the nanosized Fe3O4 modified by a carbon layer. A similar phenomenon was also found in previous research [34].   Figure 5b,d,f, which can be attributed to the (1 1 1) plane of the carbon layer. The existence of carbon was further corroborated by the energy dispersive X-ray spectroscopy (EDS) analysis ( Figure S2). The HRTEM images of C@Fe3O4 composites were also presented in Figure 5 to compare with that of their corresponding precursors. For the C@Fe3O4 nanorod and nanodisk, the lattice fringes with d-spacings of 0.297 and 0.266 nm are clearly observed, which correspond to the spacings of the (2 2 0) and (0 0 3) planes of Fe3O4, respectively (Figure 5a,c). Both lattice fringe angles are observed to be 90°, which is consistent with the theoretical calculation of the angle between the (2 2 0) and (0 0 3) planes. For C@Fe3O4 nanoparticles (Figure 5c), the typical lattice spacings are observed as 0.297 and 0.251 nm, corresponding to the lattice distance of (2 2 0) and (1 1 3) planes, respectively. Moreover, the lattice spacing of the carbon layer (approximately 0.207 nm) is also presented in Figure 5b,d,f, which can be attributed to the (1 1 1) plane of the carbon layer. The existence of carbon was further corroborated by the energy dispersive X-ray spectroscopy (EDS) analysis ( Figure S2). BET gas sorptometry measurements were conducted to examine the porous nature of these C@Fe3O4 composites. Figure 6 shows the N2 adsorption/desorption isotherm and the pore-size distribution of the C@Fe3O4 composites (a) nanorod, (b) nanodisk, and (c) nanoparticle, respectively. The isotherms of the three C@Fe3O4 composites clearly indicate their characteristic mesoporous structure, identified as type IV. The specific BET surface areas of nanorod, nanodisk, and nanoparticles are 31.407 m 2 ·g −1 , 26.682 m 2 ·g −1 BET gas sorptometry measurements were conducted to examine the porous nature of these C@Fe 3 O 4 composites. Figure 6 shows the N 2 adsorption/desorption isotherm  Figure S1). The pore-size distribution is obtained from BJH analysis, which shows numerous pores with less than 10 nm in these C@Fe 3 O 4 composites, thus, strongly supporting the mesoporous structures of these C@Fe 3 O 4 composites. The pores in both C@Fe 3 O 4 nanorod and C@Fe 3 O 4 nanodisk originated from the porous nature of the materials themselves, while the pores in C@Fe 3 O 4 nanoparticles originated from aggregates among a lot of small nanoparticles obtained from the observation of TEM images of the three C@Fe 3 O 4 composites. Moreover, as observed from the BJH pore-size distribution curve ( Figures 6 and S1), the sharp-pointed hole size distribution is approximately 4 nm for C@Fe 3 O 4 composites, while the pore sizes are approximately 20-30 nm in the precursors, which can be explained by the presence of the carbon layer generating smaller pore diameters and relatively larger surface areas [34], thus, enhancing the adsorption abilities of C@Fe 3 O 4 composites for the removal of MB in solution, eventually. Above all, the TEM and BET characterizations clearly demonstrate that a hexagonal close-packed oxide ion array of α-Fe 2 O 3 was successfully transitioned to the cubic close-packed array of Fe 3 O 4 after the calcination process. Additionally, the porous structure of the sample was not destroyed even with the significant structural changes. composites for the removal of MB in solution, eventually. Above all, the TEM and BET characterizations clearly demonstrate that a hexagonal close-packed oxide ion array of α-Fe2O3 was successfully transitioned to the cubic close-packed array of Fe3O4 after the calcination process. Additionally, the porous structure of the sample was not destroyed even with the significant structural changes. The magnetic properties of C@Fe3O4 composites were studied under ambient temperature (300 K) with an external magnetic field of −20,000 Oe to 20,000 Oe, as illustrated in Figure 7. The hysteresis loops of magnetization of three C@Fe3O4 composites show ferromagnetic behaviors. The saturation magnetization (Ms), remanent magnetization (Mr), and coercivity (Hc) values were shown in Table 1. It could be clearly seen that the saturation magnetizations (Ms) of C@Fe3O4 composites were lower in comparison to that of raw Fe3O4 (92 emu·g −1 ) [35]. However, their Ms values were higher than other formerly researched carbon-encapsulated Fe3O4 composites, for instance, Fe3O4-C microrods (0.91 emu·g −1 ), Fe3O4@C composites (13.0 emu·g −1 ) [36], and Fe3O4-C nanowires (5.11 emu·g −1 ) [37]. The decrease could be ascribed to the widespread presence of carbon layer on the Fe3O4 nanoparticles' surfaces, leading to their relatively ordered interpenetration in the carbon layer, which would hinder the optional movement of Fe3O4 nanoparticles and weaken the magnetic interactions between these magnetic nanoparticles. In addition, the Hc values were higher in comparison to that of raw Fe3O4 nanoparticles. Nevertheless, these Hc values were lower that of the formerly researched Fe3O4-C coaxial The magnetic properties of C@Fe 3 O 4 composites were studied under ambient temperature (300 K) with an external magnetic field of −20,000 Oe to 20,000 Oe, as illustrated in Figure 7. The hysteresis loops of magnetization of three C@Fe 3 O 4 composites show ferromagnetic behaviors. The saturation magnetization (Ms), remanent magnetization (Mr), and coercivity (Hc) values were shown in Table 1. It could be clearly seen that the saturation magnetizations (Ms) of C@Fe 3 O 4 composites were lower in comparison to that of raw Fe 3 O 4 (92 emu·g −1 ) [35]. However, their Ms values were higher than other formerly researched carbon-encapsulated Fe 3 O 4 composites, for instance, Fe 3 O 4 -C microrods (0.91 emu·g −1 ), Fe 3 O 4 @C composites (13.0 emu·g −1 ) [36], and Fe 3 O 4 -C nanowires (5.11 emu·g −1 ) [37]. The decrease could be ascribed to the widespread presence of carbon layer on the Fe 3 O 4 nanoparticles' surfaces, leading to their relatively ordered interpenetration in the carbon layer, which would hinder the optional movement of Fe 3 O 4 nanoparticles and weaken the magnetic interactions between these magnetic nanoparticles. In addition, the Hc values were higher in comparison to that of raw Fe 3 O 4 nanoparticles. Nevertheless, these Hc values were lower that of the formerly researched Fe 3 O 4 -C coaxial nanofibers (324.5 Oe) [38]. This phenomenon can be explained based on the higher shape anisotropy and decreased particle size of the C@Fe 3 O 4 composites. The higher shape anisotropy of the C@Fe 3 O 4 composites will induce them to magnetize in the direction of the easy axis of magnetization. Since C@Fe 3 O 4 composites were relatively randomly oriented in comparison to that of the raw Fe 3 O 4 nanoparticles, the projections of the magnetization vector in the direction of the magnetic field will be lower in comparison to that of these raw Fe 3 O 4 nanoparticles accompanied by higher shape anisotropy. Furthermore, the C@Fe 3 O 4 composites investigated in this study have lower shape magnetic anisotropy in comparison to that of the Fe 3 O 4 -C coaxial nanofibers possessing single crystalline Fe 3 O 4 cores, which, ultimately, brought about the decrease in Hc values. Hence, it can be concluded that the magnetic properties of the C@Fe 3 O 4 composites were closely related to the features of the materials, such as particle size, morphology, and the characteristics of micro-pores [39]. to the features of the materials, such as particle size, morphology, and the characteristics of micro-pores [39].  Annotation: The saturation magnetization (Ms) represents the maximum magnetization that a magnetic material can achieve when magnetized in an external magnetic field. Remanent magnetization (Mr) donates that the magnetic material still maintains a certain magnetization in the direction of the original external magnetic field; this occurs after removing the external magnetic field when the magnetic material is magnetized to saturation. The coercivity (Hc) represents that the magnetic field intensity is required to reduce its magnetization to zero; this occurs after the magnetic material has been magnetized to magnetic saturation.

Adsorption Kinetics of the C@Fe3O4 Composites
In this study, three C@Fe3O4 composites and their precursors were used to absorb MB from the aqueous solution. Figure 8    Annotation: The saturation magnetization (Ms) represents the maximum magnetization that a magnetic material can achieve when magnetized in an external magnetic field. Remanent magnetization (Mr) donates that the magnetic material still maintains a certain magnetization in the direction of the original external magnetic field; this occurs after removing the external magnetic field when the magnetic material is magnetized to saturation. The coercivity (Hc) represents that the magnetic field intensity is required to reduce its magnetization to zero; this occurs after the magnetic material has been magnetized to magnetic saturation.

Adsorption Kinetics of the C@Fe 3 O 4 Composites
In this study, three C@Fe 3 O 4 composites and their precursors were used to absorb MB from the aqueous solution. Figure 8  To investigate the adsorption kinetic process and mechanism, four kinetic models were applied to fit the kinetic adsorption data. Firstly, the linear regressions of the pseudo-first-order kinetic and pseudo-second-order kinetic models were used to fit the adsorption kinetics data for MB adsorption onto the obtained samples [40,41]. The above two kinetic models were illustrated as Equations (1) and (2) for pseudo-first-order kinetic and pseudo-second-order kinetic models, respectively: where qe and qt (mg⋅g −1 ) are the amounts of MB adsorbed at equilibrium and at time t (min), respectively. k1 (min −1 ) is the pseudo-first-order kinetic rate constant, which represents the adsorption rate when the adsorption process conforms to the pseudo-first-order kinetics equation. Additionally, k2 (g⋅mg −1 ⋅min −1 ) is the pseudo-second-order kinetic rate constant. Additionally, the fitting results were shown in Figure 9a,b, in company with These results clearly indicate that the MB adsorption onto the C@Fe3O4 composites can be well described by the pseudo-second-order kinetic model and that chemical adsorption might be involved in the process. Moreover, in this study, the k1 values followed the order of 0.1531 (min −1 ) > 0.0477 (min −1 ) > 0.0182 (min −1 ) for C@Fe3O4 composites nanorod, nanoparticle, and nanodisk, respectively, which was consistent with the comparisons of adsorption performance for the above C@Fe3O4 composites. Although the MB adsorption onto the C@Fe3O4 composites could not be well described by the pseudo-first-order kinetics model due to the poor R 2 values, it can still be inferred that the larger BET surface area, suitable morphology structures, and pore-size distributions would be beneficial for increasing the adsorption rate. Additionally, the k2 values followed the order of 0.0205 g·mg·min −1 ) > 0.0120 (g·mg·min −1 ) > 0.00085 (g·mg·min −1 ) for C@Fe3O4 nanorod, nanoparticle, and nanodisk composites, re- To investigate the adsorption kinetic process and mechanism, four kinetic models were applied to fit the kinetic adsorption data. Firstly, the linear regressions of the pseudofirst-order kinetic and pseudo-second-order kinetic models were used to fit the adsorption kinetics data for MB adsorption onto the obtained samples [40,41]. The above two kinetic models were illustrated as Equations (1) and (2) for pseudo-first-order kinetic and pseudosecond-order kinetic models, respectively: where q e and q t (mg·g −1 ) are the amounts of MB adsorbed at equilibrium and at time t (min), respectively. k 1 (min −1 ) is the pseudo-first-order kinetic rate constant, which represents the adsorption rate when the adsorption process conforms to the pseudo-first-order kinetics equation. Additionally, k 2 (g·mg −1 ·min −1 ) is the pseudo-second-order kinetic rate constant. Additionally, the fitting results were shown in Figure 9a,b, in company with   Secondly, the intra-particle diffusion model was utilized to obtain a more particular knowledge of adsorption data, and the equation was expressed as Equation (3) [42]: where ki (mg•g −1 •min −1 ) is the rate constant of the intra-particle diffusion model and C (mg•g −1 ) is a constant, which is proportional to the extent of boundary layer thickness. As shown in Figure 9c and Table S1, the adsorption of MB onto the C@Fe3O4 composites can be divided into three typical processes, including external, interfacial, and eventual equilibrium processes. The first stage with the sharp increase in adsorption amount was mainly controlled by the outside surface adsorption. Additionally, in this stage, a great quantity of MB molecules was immobilized onto the adsorption site of the adsorbents. Subsequently, the MB molecules diffused into the pores of the adsorbents, which were adsorbed by the interior surface of pores, in which the intra-particle diffusion became the primary rate-determining step. The third step demonstrates the final equilibrium  Secondly, the intra-particle diffusion model was utilized to obtain a more particular knowledge of adsorption data, and the equation was expressed as Equation (3) [42]: where k i (mg·g −1 ·min −1 ) is the rate constant of the intra-particle diffusion model and C (mg·g −1 ) is a constant, which is proportional to the extent of boundary layer thickness. As shown in Figure 9c and Table S1, the adsorption of MB onto the C@Fe 3 O 4 composites can be divided into three typical processes, including external, interfacial, and eventual equilibrium processes. The first stage with the sharp increase in adsorption amount was mainly controlled by the outside surface adsorption. Additionally, in this stage, a great quantity of MB molecules was immobilized onto the adsorption site of the adsorbents. Subsequently, the MB molecules diffused into the pores of the adsorbents, which were adsorbed by the interior surface of pores, in which the intra-particle diffusion became the primary rate-determining step. The third step demonstrates the final equilibrium state. Moreover, none of the three fitting lines crossed the origin, suggesting that intra-particle diffusion is not the only rate-controlling step. Furthermore, another kinetic fitting equation of the Boyd kinetic model was also utilized to confirm the practical rate-determining step regarding the MB immobilization onto the C@Fe 3 O 4 composites. Additionally, the equation was illustrated as Equations (4)-(6) [43]: where F is donated as the fraction of solute, while parameter B is the mathematic function of F. As illustrated in Figure 9d and Table S2, none of the three curves of the Boyd model plot crossed the origin, suggesting that liquid film diffusion was the rate-limiting step for MB immobilization onto the three C@Fe 3 O 4 composites. Additionally, the fitting results achieved a good agreement with that of the intra-particle diffusion model. Furthermore, the kinetic fitting parameters of the immobilization process on the three α-Fe 2 O 3 precursors were also evaluated by using these four kinetic models, as mentioned above ( Figure S3, Tables S2 and S3).

Adsorption Kinetics of the C@Fe 3 O 4 Composites
To investigate the maximum adsorption capacity of C@Fe 3 O 4 composites for MB removal, the adsorption experiments on different initial concentrations of MB were carried out at room temperature and the results of the experiments were shown in Figure 10. It could be seen that the adsorption capacities of C@Fe 3 O 4 composites were increased with the increase in the MB solution concentration. Particularly, the adsorption capacities of C@Fe 3 O 4 composites increased rapidly in the situation of low initial concentrations, indicating that the C@Fe 3 O 4 composites had a large number of active sites. Moreover, the adsorption capacities of the C@Fe 3 O 4 composites became flat as the initial MB concentration went up, suggesting that the limited active sites were fully occupied by an excess of MB dye. Moreover, two isotherm models of Langmuir and Freundlich were applied to understand the type of interaction of adsorbents (C@Fe 3 O 4 composites) with adsorbate (MB). The Langmuir isotherm was often applied to express the monolayer adsorption on homogeneous and energetically uniform surfaces, while the Freundlich isotherm model was extensively applied to determine the multilayer adsorption course [44,45]. Additionally, the two models were expressed as Equations (7) and (8) for the Langmuir and Freundlich isotherm model, respectively: q e = Q max bC e 1 + bC e (7) where q e (mg·g −1 ) represents the adsorption capacity of MB onto the C@Fe 3 O 4 composites under the balance situation, C e (mg·L −1 ) represents the MB concentration under the balance situation, Q max (mg·g −1 ) represents the maximum adsorption capacity MB onto the C@Fe 3 O 4 composites, b (L·mg −1 ) is related to the adsorption energy constant, K f represents the indicator of adsorption quantity, and 1/n represents the adsorption intensity.  The fitting results of the two above models were illustrated in Figure 10 and Table  3, which indicated that the Langmuir model yields a better fit for the experimental equilibrium adsorption data than the Freundlich model based on the comparison of the correlation coefficients (R 2 ) of the two equations. This fact indicates that the binding energy is uniform on the whole surface of these C@Fe3O4 composites, suggesting a mono-layer coverage of MB on the surface of the C@Fe3O4 composites. Furthermore, another data parameter of RL (RL = 1/(1 + bC0)) was also applied to understand the adsorption process [46]. It could be also seen that all the RL values obtained from the above function of the MB adsorption onto the three C@Fe3O4 composites are all within the range from 0 to 1, suggesting a favorable immobilization procedure. Moreover, the Freundlich constant n is greater than 1, which reveals the favorable adsorption of MB onto the C@Fe3O4 composites. This is consistent with the conclusion of the Langmuir model. The maximum adsorption capacities (Qmax) calculated from the above Langmuir model shown in Table  S4 are 34.54 mg·g −1 , 32.94 mg·g −1 , and 33.64 mg·g −1 for the C@Fe3O4 nanorod, nanaodisk and nanoparticle composites, respectively. Generally speaking, suitable morphology structures, suitable pore-size distribution, and large specific surface area will be beneficial to promoting their adsorption properties in the process of dye adsorption [47,48]. The C@Fe3O4 nanorod owned the largest adsorption capacity, while the C@Fe3O4 nanodisk owned the lowest adsorption capacity, which can be explained as follows: Comparing with the morphology structures of the three C@Fe3O4 composites, as shown in Figure 4, the C@Fe3O4 nanodisks are stuck much tighter and a thicker carbon layer is wrapped outside of the cubic Fe3O4, making it difficult for the MB molecules to combine with the active sites, thus, decreasing the adsorption capacity of the C@Fe3O4 nanodisk. The C@Fe3O4 nanorod sample owns a relatively thin carbon layer wrapped outside of the cubic Fe3O4 (Figure 4d), making it easier for the MB molecules to combine with the active sites. Moreover, the C@Fe3O4 nanorod exhibited a larger surface area than the C@Fe3O4 nanodisk, and nanoparticle, which is further beneficial to its adsorption process toward MB adsorption. The difference in adsorption capacity is the result of the morphology structures and BET surface area, but has little connection with pore size, since the three C@Fe3O4 nanorod, nanodisk, and nanoparticle composites have similar pore sizes. The fitting results of the two above models were illustrated in Figure 10 and Table 3, which indicated that the Langmuir model yields a better fit for the experimental equilibrium adsorption data than the Freundlich model based on the comparison of the correlation coefficients (R 2 ) of the two equations. This fact indicates that the binding energy is uniform on the whole surface of these C@Fe 3 O 4 composites, suggesting a mono-layer coverage of MB on the surface of the C@Fe 3 O 4 composites. Furthermore, another data parameter of R L (R L = 1/(1 + bC 0 )) was also applied to understand the adsorption process [46]. It could be also seen that all the R L values obtained from the above function of the MB adsorption onto the three C@Fe 3 O 4 composites are all within the range from 0 to 1, suggesting a favorable immobilization procedure. Moreover, the Freundlich constant n is greater than 1, which reveals the favorable adsorption of MB onto the C@Fe 3 O 4 composites. This is consistent with the conclusion of the Langmuir model. The maximum adsorption capacities (Q max ) calculated from the above Langmuir model shown in Table S4 are 34.54 mg·g −1 , 32.94 mg·g −1 , and 33.64 mg·g −1 for the C@Fe 3 O 4 nanorod, nanaodisk and nanoparticle composites, respectively. Generally speaking, suitable morphology structures, suitable pore-size distribution, and large specific surface area will be beneficial to promoting their adsorption properties in the process of dye adsorption [47,48]. The C@Fe 3 O 4 nanorod owned the largest adsorption capacity, while the C@Fe 3 O 4 nanodisk owned the lowest adsorption capacity, which can be explained as follows: Comparing with the morphology structures of the three C@Fe 3 O 4 composites, as shown in Figure 4, the C@Fe 3 O 4 nanodisks are stuck much tighter and a thicker carbon layer is wrapped outside of the cubic Fe 3 O 4 , making it difficult for the MB molecules to combine with the active sites, thus, decreasing the adsorption capacity of the C@Fe 3 O 4 nanodisk. The C@Fe 3 O 4 nanorod sample owns a relatively thin carbon layer wrapped outside of the cubic Fe 3 O 4 (Figure 4d), making it easier for the MB molecules to combine with the active sites. Moreover, the C@Fe 3 O 4 nanorod exhibited a larger surface area than the C@Fe 3 O 4 nanodisk, and nanoparticle, which is further beneficial to its adsorption process toward MB adsorption. The difference in adsorption capacity is the result of the morphology structures and BET surface area, but has little connection with pore size, since the three C@Fe 3 O 4 nanorod, nanodisk, and nanoparticle composites have similar pore sizes.

Synthesis of C@Fe 3 O 4 Nanorods
C@Fe 3 O 4 nanorods were fabricated through a three-step process. (1) α-Fe 2 O 3 nanorods were first obtained by a simple method. Specifically, 1.0 mmol of FeCl 3 and 2 mmol of citric acid were added into 40 mL of water under vigorous stirring. A total of 25 mmol of NaOH(s) was added into the above solution while stirring for approximately 10 min. After that, the whole mixture was transferred into a Teflon-lined stainless-steel autoclave (LC-KH-50, Lichen Instrument Technology Co., Ltd., Shaoxing, China) for hydrothermal treatment at 140 • C for 12 h. As the autoclave cooled to room temperature naturally, the precipitates (α-FeOOH) were separated by centrifugation (KL02A, Hunan Kaida Scientific Instrument Co., Ltd., Changsha, China), washed with distilled water and absolute ethanol, and dried in vacuum at 60 • C for 12 h. Finally, to obtain the α-Fe 2 O 3 nanorods, the as-prepared α-FeOOH was subjected to calcination in a muffle furnace (SX4-2-12, Sanfu Hezhong Technology Development Co., Ltd., Beijing, China) at 400 • C for 2 h with a temperature increasing rate of 1 • C·min −1 . (2) The α-Fe 2 O 3 nanorods/alginate composite beads were obtained through an ionic polymerization method. Particularly, sodium alginate powder with 1.5 g was stirred continuously in distilled water with 100 mL at 353.15 K for 12 h, gaining a viscous, transparent, and homogeneous sample solution. Subsequently, α-Fe 2 O 3 nanorods with 1.5 g was mixed into the above-mentioned solution and continually stirred for another 30 min. The above mixed solution containing sodium alginate and α-Fe 2 O 3 nanorods was added dropwise to the 0.18 mol·L −1 CaCl 2 solution with a syringe to obtain magnetic beads, which were then washed a few times by using distilled water and collected by drying in vacuum at 60 • C for 12 h, ultimately. A schematic illustration is shown in Scheme 1 for the fabrication of α-Fe 2 O 3 /alginate beads. (3) The as-prepared α-Fe 2 O 3 /alginate beads were further heated at 400 • C for 2 h in a tube furnace with a heating rate of 1 • C·min −1 under Ar atmosphere. The sample was treated with acid solutions, washed a few times using distilled water and collected by drying in vacuum at 60 • C for 12 h, ultimately. After the above treatment, the C@Fe 3 O 4 nanorods were obtained. The C@Fe 3 O 4 nanodisk and C@Fe 3 O 4 nanoparticle were also prepared using α-Fe 2 O 3 nanodisk and commercially available α-Fe 2 O 3 as the precursors through a similar procedure as mentioned above. The detailed synthetic procedures are available in the Supplementary Materials.

Characterization
X-ray diffraction (XRD) patterns were obtained using a Philips X'Pert Pro Super diffractometer (Amsterdam, The Netherlands) with Cu Kα radiation (λ = 1.54178 Å). The Raman spectra were recorded at an ambient temperature on a SPEX 1403 spectrometer (Metuchen, NJ, USA) with an argon-ion laser at an excitation wavelength of 514.5 nm. The scanning electron microscope (SEM) was carried out using a Hitachi S-4800 (Tokyo, Japan), which was equipped with an X-ray energy dispersive spectrometer of GENE-SIS4000. The high-resolution transmission electron microscopy (HRTEM) was performed on JEOL 2010 (Tokyo, Japan) with an accelerating voltage of 20 kV. N2 adsorption was measured using the Brunauer-Emmett-Teller method (ASAP-2020, USA) at 77 K. A superconducting quantum interference device (SQUID, San Diego, USA) was applied to investigate the magnetic properties of the samples.

Adsorption Experiment
Groups of experiments were designed to investigate the kinetics and isotherms of MB adsorption onto the magnetic materials. The stock solution of MB with 0.01 mol·L −1 was firstly prepared and then paired to certain concentrations in the following experiments.
The isothermal adsorption experiments were carried out under different initial MB concentrations (8. where C0 (mg·L −1 ) and Ce (mg·L −1 ) are the initial and equilibrium concentration of MB in the solution, V (L) is the total volume of the suspension, and m (g) is the mass of adsorbent. To determine the adsorption efficiency, the kinetic experiments were performed by using different flasks containing 0.8 g·L −1 of C@Fe3O4 composites. At the desired time intervals, the magnetic C@Fe3O4 composites were separated by using a hand-held permanent magnet, and the supernatant was collected for concentration measurements. The absorbance of the MB dye was measured using a UV-visible spectrophotometer (Shimadzu UV-1800 spectrophotometer, Kyoto, Japan). The maximum wavelength of ab-

Characterization
X-ray diffraction (XRD) patterns were obtained using a Philips X'Pert Pro Super diffractometer (Amsterdam, The Netherlands) with Cu Kα radiation (λ = 1.54178 Å). The Raman spectra were recorded at an ambient temperature on a SPEX 1403 spectrometer (Metuchen, NJ, USA) with an argon-ion laser at an excitation wavelength of 514.5 nm. The scanning electron microscope (SEM) was carried out using a Hitachi S-4800 (Tokyo, Japan), which was equipped with an X-ray energy dispersive spectrometer of GENESIS4000. The high-resolution transmission electron microscopy (HRTEM) was performed on JEOL 2010 (Tokyo, Japan) with an accelerating voltage of 20 kV. N 2 adsorption was measured using the Brunauer-Emmett-Teller method (ASAP-2020, USA) at 77 K. A superconducting quantum interference device (SQUID, San Diego, USA) was applied to investigate the magnetic properties of the samples.

Adsorption Experiment
Groups of experiments were designed to investigate the kinetics and isotherms of MB adsorption onto the magnetic materials. The stock solution of MB with 0.01 mol·L −1 was firstly prepared and then paired to certain concentrations in the following experiments.
To determine the adsorption efficiency, the kinetic experiments were performed by using different flasks containing 0.8 g·L −1 of C@Fe 3 O 4 composites. At the desired time intervals, the magnetic C@Fe 3 O 4 composites were separated by using a hand-held permanent magnet, and the supernatant was collected for concentration measurements. The absorbance of the MB dye was measured using a UV-visible spectrophotometer (Shimadzu UV-1800 spectrophotometer, Kyoto, Japan). The maximum wavelength of absorbance for the MB dye is λ max = 664 nm. All the adsorption experiments were carried out under solution pH values of 6.0 ± 0.1 and a temperature of 298.15 K. Moreover, the adsorption isotherm experiments were carried out for 48 h on an oscillator to achieve an adsorption equilibrium. Furthermore, all adsorption experiments were carried out in triplicate in order to ensure the accuracy of the experimental data.

Conclusions
Herein, a series of C@Fe 3 O 4 composites with three morphologies of nanorod, nanodisk, and nanoparticles were successfully fabricated via a three-step process, using the easily obtained α-Fe 2 O 3 as the precursor. The α-Fe 2 O 3 precursors were first fabricated, followed by the generation of α-Fe 2 O 3 /carbon shell beads using sodium alginate as a carbon source.
The magnetic hysteresis loop measurement shows that the C@Fe 3 O 4 composites display ferromagnetic properties at room temperature. The saturation magnetization values of the C@Fe 3 O 4 composites are lower, while the coercivity (Hc) values of the C@Fe 3 O 4 composites are higher in comparison to that of the raw Fe 3 O 4 nanoparticles, which could be ascribed to the widespread presence of carbon layer, the shape anisotropy, and the decreased particle size of C@Fe 3 O 4 composites. These C@Fe 3 Table  S1: Parameters of intra-particle diffusion and Boyd kinetic models for the adsorption of MB onto C@Fe 3 O 4 composites. Table S2: Parameters of pseudo-first-order and pseudo-second-order for the adsorption of MB onto α-Fe 2 O 3 samples. Table S3: Parameters of intra-particle diffusion and Boyd kinetic models for the adsorption of MB onto α-Fe 2 O 3 samples. Table S4:

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