E ﬀ ective Adsorption of Methylene Blue dye onto Magnetic Nanocomposites. Modeling and Reuse Studies

Featured Application: Treatment of polluted water with dyes in textile and paper industries. Abstract: In the present study, new adsorbent beads of alginate (A) / maghemite nanoparticles ( γ -Fe 2 O 3 ) / functionalized multiwalled carbon nanotubes (f-CNT) were prepared and characterized by several techniques, e.g., N 2 adsorption-desorption isotherms, Fourier transformed infrared spectroscopy (FTIR), X-ray di ﬀ raction (XRD), thermogravimetric analysis (TGA / DTG), scanning electron microscopy (SEM), and vibrating sample magnetometry (VSM) and further tested for the adsorption of the dye methylene blue (MB) from water. The beads (A / γ -Fe 2 O 3 / f-CNT) presented a relatively low BET speciﬁc surface area value of 59 m 2 g − 1 . The magnetization saturation values of A / γ -Fe 2 O 3 / f-CNT beads determined at 295 K was equal to 27.16 emu g − 1 , indicating a magnetic character. The time needed to attain the equilibrium of MB adsorption onto the beads was estimated within 48 h. Thus, several kinetic and isotherm equation models were used to ﬁt the kinetic and equilibrium experimental results. The number of adsorbed MB molecules per active site, the anchorage number, the receptor sites density, the adsorbed quantity at saturation, the concentration at half saturation and the molar adsorption energy were quantiﬁed using the monolayer model. The calculated negative ∆ G 0 and positive ∆ H 0 values suggested the spontaneous and endothermic nature of the adsorption process. In addition, A / γ -Fe 2 O 3 / f-CNT composites can be used at least for six times maintaining their signiﬁcant adsorptive performance and could be easily separated by using a magnet from water after treatment. The magnetization saturation of A / γ -Fe 2 O 3 / f-CNT beads was slightly lower than that of A / γ -Fe 2 O 3 beads, but it stills enough to be e ﬃ ciently separated from the aqueous medium. From these magnetization measurements, the mass fractions of γ -Fe 2 O 3 nanoparticles in A / γ -Fe 2 O 3 and A / γ -Fe 2 O 3 / f-CNT beads were estimated to be 65 and 60%, respectively. These results revealed that the synthesized magnetic beads are excellent materials for wastewater treatment in large-scale applications because they can be easily separated from water using a magnetic ﬁeld. by controlling both the surface chemistry of the and the of the The experiments out in a pH of 3.0–10.0 and the results It could be observed that MB uptake increased from 37.1 to 82.7 mg g − 1 with increasing the pH solution from 3 to 10. These results are directly related to the chemical surface properties of the adsorbent and MB molecules.


Introduction
Dyes are widely used in many industrial sectors such as textile and paper companies [1]. These industries dump large amounts of wastewater into streams without any prior treatment. Their discharges present a source of pollution that can be very harmful to human and environmental health. The protection of the environment and the improvement of water quality by means of the treatment of these industrial effluents have become an issue of concern in recent years. Methylene blue (MB) is a synthetic dye considered a suspected carcinogenic agent. It can cause breathing difficulties, nausea and huge quantities of waste. Likewise, these powdered materials cannot be used in a fixed-bed column because of the regeneration costs and the clogging of the reactor [34]. In general, the encapsulation of the adsorbent in alginate beads is an excellent solution to overcome this problem. In this context, the purpose of this study is to prepare an adsorbent material by encapsulating functionalized maghemite γ-Fe 2 O 3 and carbon nanotube (f-CNT) in alginate beads (A) and used them for methylene blue uptake.

Preparation of Magnetic γ-Fe 2 O 3 Nanoparticles
γ-Fe 2 O 3 nanoparticles synthesis was carried out following a modified procedure described by Bee et al. [28]. Thus, magnetite particles were prepared by co-precipitation after adding ammonium hydroxide solution to the mixture of FeCl 2 4H 2 O and FeCl 3 6H 2 O solutions. Then, magnetite particles were oxidized to maghemite in nitric acid solution at 90 • C using a boiling Fe(NO 3 ) 3 solution. Maghemite nanoparticles were then isolated and rinsed using acetone and ether and dispersed in water. In order to create a negative charge and to stabilize the surface of the obtained nanoparticles, maghemite was coated by adding 5 g of sodium citrate to 50 mL of magnetic solution at 90 • C for 90 min. Citrate functionalized maghemite nanoparticles were dispersed in water, affording a brown ferrofluid with a pH value of 7.7.

Preparation of (A/γ-Fe 2 O 3 /f-CNT) Beads
The composite beads (A/γ-Fe 2 O 3 /f-CNT) were synthesized as follows: 2 g of sodium alginate was dissolved in 100 mL of dispersion containing 16 mL of citrate functionalized ferrofluid. 0.4 g of f-CNT was then added to the solution at continuous stirring. This magnetic solution was dropped in a calcium chloride solution (4%) in order to form the beads. The beads were gently stirred for 24 h in calcium solution to obtain the total gelation of the alginate. After 24 h, A/γ-Fe 2 O 3 /f-CNT beads were separated, washed and stored in distilled water. Magnetic beads without f-CNT were also prepared by the same method and named as A/γ-Fe 2 O 3 .

Characterization Techniques
The textural properties of MWCNTs, γ-Fe 2 O 3 and A/γ-Fe 2 O 3 /f-CNT were explored by adsorption-desorption of N 2 at 77 K in an ASAP 2020 system (Micromeritics, Norcross, GA, USA). Before the analysis, the samples were outgassed at 250 • C for 3 h. The specific surface area (S BET , m 2 g −1 ) of the materials was evaluated by Brunauer, Emmett and Teller (B.E.T.) equation in a pressure range of P/P 0 = 0.15-0. 35. In addition, the external surface area (S ext, m 2 g −1 ) was determined according to t-plot method and the total pore volume value (V Total , cm 3 g −1 ) considered was at a relative pressure of P/P 0 = 0.99. FTIR spectra of γ-Fe 2 O 3 , f-CNT, A, A/γ-Fe 2 O 3 and A/γ-Fe 2 O 3 /f-CNT before and after MB adsorption were obtained in a FTIR 8400S spectrometer (Shimadzu, Kyoto, Japan) in the range of 400-4000 cm −1 , using KBr pellet technique. The magnetic properties of γ-Fe 2 O 3 , A/γ-Fe 2 O 3 and A/γ-Fe 2 O 3 /f-CNT were determined using a 7404-S vibrating sample magnetometer (Lake Shore, Princeton, Wi, USA) at room temperature. The crystalline phases in γ-Fe 2 O 3 , f-CNT and A/γ-Fe 2 O 3 /f-CNT materials were analysed by X-ray diffraction (XRD) using an X´Pert PRO diffractometer (Malvern Panalytical Inc., Westborough, MA, USA) with Cu Kα radiation (45kV, 40 mA) in the 2θ range between 10 • and 80 • at a scanning rate of 2 • min −1 . The spectra were analyzed by the Rietveld Refinement method using the Material Analysis Using Diffraction (MAUD) program and the average crystal size was calculated using the Debye-Scherrer equation. Thermogravimetric analysis (TGA/DTG) was conducted in an EXTAR 6000 thermal analyzer (Seiko, Mahwa, NJ, USA) under nitrogen atmosphere at a heating rate of 10 • C min −1 , from 30 to 900 • C. Thus, TGA/DTG analysis in air atmosphere (from 30 to 900 • C, 10 • C min −1 ) of γ-Fe 2 O 3 particles, f-CNT, alginate/γ-Fe 2 O 3 beads and A/γ-Fe 2 O 3 /f-CNT composite were accomplished. Finally, the surface morphology of the materials was studied by scanning electron microscopy (SEM) using a JSM 6400 microscope (JEOL, Peabody, MA, USA) equipped with a thermoionic cathode and a 25 kV detector. EDX analysis of A/γ-Fe 2 O 3 /f-CNT was accomplished. Additionally, the morphology of functionalized carbon nanotubes (f-CNT) was explored by transmission electronic microscopy (TEM) in a JEOL JEM 2100 microscope (200 kV, 0.25 nm of resolution).

Adsorption Experiments
The effect of pH on MB adsorption onto A/γ-Fe 2 O 3 /f-CNT was studied varying the initial pH values from 3 to 10. The volume used was of 50 mL, operating at room temperature, ±25 • C. Kinetic batch adsorption tests were accomplished using a MB initial concentration of 230 mg L −1 . For this, 50 mg of A/γ-Fe 2 O 3 /f-CNT was dispersed in MB solution (50 mL) at natural pH (5.2) and room temperature. At regular time intervals, a sample was magnetic separated. The equilibrium adsorption experiments were accomplished at different initial MB concentrations, ranging from 5 to 270 mg L −1 , mixed with 50 mg of γ-Fe 2 O 3 /f-CNT. The dispersions were maintained at a constant temperature of 25 • C, at a natural pH of 5.2. All the adsorption experiments were carried out without stirring. When the equilibrium was reached, the magnetic adsorbent could be separated from the aqueous medium by using a magnet. The effect of temperature on the adsorption was studied at different temperatures, i.e., 283, 293, 303 and 313 K using a thermostatic chamber.
MB concentration was determined on a Shimadzu UV-1700 spectrophotometer at a wavelength of 664 nm. The amount of adsorbed dye at equilibrium time (q e , mg g −1 ) was determined using Equation (1): where, C 0 and Ce (mg L −1 ) are MB concentration at initial and equilibrium time, respectively, m (g) is the weight of adsorbent and V (L) is the solution volume.

Desorption of MB Dye and Reusability Studies
A/γ-Fe 2 O 3 /f-CNT composite beads were used for six consecutive adsorption-desorption cycles in order to investigate their regeneration efficiency and examine their cost effectiveness. After adsorption, the desorption of MB from the beads was carried out using 10 mL of 0.1M HNO 3 solution , using 10 mg of adsorbent. The desorption efficiency (DE) was calculated by using Equation (2): where, q (mg g −1 ) is the adsorbed amount of MB, C (mg L −1 ) is the MB concentration in the desorption solution, V (L) is the desorption volume and m (g) is the mass of adsorbent used in the desorption tests.

Analysis of the Experimental Data
In order to find the best correlation of the experimental data, the regression coefficient (R 2 ) and residual root-mean squared error (RMSE) values (Equations (3) and (4), respectively) were calculated using Origin 8.5 software (OriginLab Corporation, Northampton, MA, USA).
(q e,t. exp .n − q e,t.cal.n ) 2 n n−1 (q e,t. exp .n − q e,t. exp .n ) 2 where, n denotes the number of experimental data, q e,t. exp and q e,t.cal are the experimental and calculated adsorption capacities at equilibrium or at any time, respectively.

Characterization of the Tested Adsorbents
N 2 adsorption-desorption isotherms obtained at 77 K of f-CNT, γ-Fe 2 O 3 particles and A/γ-Fe 2 O 3 /f-CNT beads are depicted in Figure 1. The textural parameters, specific surface area (S BET ), external surface area (S ext ), pore volume (V total ) and pore diameter of the tested materials are collected in Table 1. According to IUPAC classification [35], carbon nanotubes (f-CNT) are materials that show a mainly mesoporous structure; with a contribution of microporosity (~40 cm 3 g −1 at low P/P 0 values) and an important quantity of meso-macropores (type IVa isotherm, according IUPAC classification). These adsorbents show that both brands of N 2 adsorption isotherm are parallel and near to the verticality, characteristic of materials with cylindrical pores, where capillary condensation occurs (Table 1). Nevertheless, γ-Fe 2 O 3 and A/γ-Fe 2 O 3 /f-CNT beads showed N 2 adsorption isotherms where, at low relative pressure values, the amount of N 2 adsorbed is very low (lower than 10 cm 3 g −1 STP ). These are isotherms representative of essentially mesoporous material, without poor contribution of microporosity on their structure. Thus, f-CNT material showed a low-moderate specific surface area (265 m 2 g −1 ), since relatively low BET area values were found for γ-Fe 2 O 3 nanoparticles and A/γ-Fe 2 O 3 /f-CNT composites, 35 and 59 m 2 g −1 , respectively.
In order to find the best correlation of the experimental data, the regression coefficient (R 2 ) and residual root-mean squared error (RMSE) values (Equations (3) and (4), respectively) were calculated using Origin 8.5 software (OriginLab Corporation, Northampton, MA, USA).

Characterization of the Tested Adsorbents
N2 adsorption-desorption isotherms obtained at 77 K of f-CNT, γ-Fe2O3 particles and A/γ-Fe2O3/f-CNT beads are depicted in Figure 1. The textural parameters, specific surface area (SBET), external surface area (Sext), pore volume (Vtotal) and pore diameter of the tested materials are collected in Table 1. According to IUPAC classification [35], carbon nanotubes (f-CNT) are materials that show a mainly mesoporous structure; with a contribution of microporosity (~40 cm 3 g −1 at low P/P 0 values) and an important quantity of meso-macropores (type IVa isotherm, according IUPAC classification). These adsorbents show that both brands of N2 adsorption isotherm are parallel and near to the verticality, characteristic of materials with cylindrical pores, where capillary condensation occurs (Table 1). Nevertheless, γ-Fe2O3 and A/γ-Fe2O3/f-CNT beads showed N2 adsorption isotherms where, at low relative pressure values, the amount of N2 adsorbed is very low (lower than 10 cm 3 g −1 STP). These are isotherms representative of essentially mesoporous material, without poor contribution of microporosity on their structure. Thus, f-CNT material showed a low-moderate specific surface area (265 m 2 g −1 ), since relatively low BET area values were found for γ-Fe2O3 nanoparticles and A/γ-Fe2O3/f-CNT composites, 35 and 59 m 2 g −1 , respectively.  The FTIR spectra of (A), f-CNT, γ-Fe 2 O 3 nanoparticles, A/γ-Fe 2 O 3 and A/γ-Fe 2 O 3 /f-CNT beads are shown in Figure 2a. A wide band at 3400-3441 cm −1 was observed in all spectra and attributed to the stretching vibration of -OH groups coming from the physically adsorbed water. The peak observed at 2923 cm −1 for A, A/γ- The thermal stability of the materials was evaluated by thermogravimetric analysis. Figure 3a-e show the TGA/TDG curves of the different samples accomplished in a temperature range of 30 to 900° C. The alginate curve showed an initial weight loss of 12.77 % between 30 and 200 °C corresponding to the loss of water. At temperature between 200 and 500 °C, the glycosidic bonds were destroyed which corresponds to a weight loss of 47.76 % [43]. The third loss of mass (16.71%) could be attributed to the formation of CaCO3 [44]. Finally, the loss of mass of 8% corresponded to the decomposition of CaCO3 which begins at 600 °C. Beyond 700 °C, only quicklime remained as the residue of the decomposition of CaCO3, representing 14.75% of the initial mass of alginate. Thus, the total weight loss for alginate is 85.24%. One endothermic peak at 70°C and two exothermic peaks at 483°C and 533 °C were observed. The three peaks can be attributed to the dehydration of alginate, degradation of alginate and formation of CaCO3, respectively. For γ-Fe2O3, the TGA/TDA (Figure 3b) curve suggested a weight loss of 8.16% at temperatures ranging from 30 to 240 °C, and 12.53% at temperatures from 240 to 475 °C. These mass losses could be ascribed to two endothermic peaks (at The corresponding interplanar distance at 2θ = 43.83 • is of d (400) = 0.2085 nm, that is very close to that reported in the literature [39,40]. The diffractogram fitted very well to the cubic symmetry of γ-Fe 2 O 3 with the P4 1 32 space group. The Rietveld analysis by the MAUD software confirmed the good crystallinity of γ-Fe 2 O 3 with a cubic elementary mesh (a = 0.8341 nm). This value is in agreement with that reported in the literature for the same nanoparticles [41]. The average value of γ-Fe 2 O 3 crystals obtained by the refinement was of 10.59 nm, whereas the size calculated by Scherrer equation was of 10.24 nm. This value is in agreement with that obtained by Idris et al. [42], where resulted in 10 nm.
By comparison between the f-CNT diffractogram used in this study and that of the non-functionalized CNT reported in the literature [38], it could be concluded that the functionalization of CNTs with acid does not alter the crystallographic character of the material; only two peaks at 2θ = 25. The characteristic peaks of f-CNT were not identified in the patterns of composite material (A/γ-Fe 2 O 3 /f-CNT) due to the low initial f-CNT weight ratio used in the preparation of alginate beads (0.4%). In addition, the presence of the peak at 2θ = 43.83 o (400) in maghemite XRD pattern caused the disappearance of the characteristic peak of f-CNT at 2θ = 42.81 • (100). These findings may be also attributed by the presence of alginate in the material, which reduces the purity of f-CNT in the final composite. It should be noted that A/γ-Fe 2 O 3 /f-CNT have the same characteristic peaks that γ-Fe 2 O 3 have, indicating that the presence of alginate and f-CNTs in the composite did not alter the crystallographic character of γ-Fe 2 O 3 .
The thermal stability of the materials was evaluated by thermogravimetric analysis. Figure 3a-e show the TGA/TDG curves of the different samples accomplished in a temperature range of 30 to 900 • C. The alginate curve showed an initial weight loss of 12.77 % between 30 and 200 • C corresponding to the loss of water. At temperature between 200 and 500 • C, the glycosidic bonds were destroyed which corresponds to a weight loss of 47.76 % [43]. The third loss of mass (16.71%) could be attributed to the formation of CaCO 3 [44]. Finally, the loss of mass of 8% corresponded to the decomposition of CaCO 3 which begins at 600 • C. Beyond 700 • C, only quicklime remained as the residue of the decomposition of CaCO 3 , representing 14.75% of the initial mass of alginate. Thus, the total weight loss for alginate is 85.24%. One endothermic peak at 70 • C and two exothermic peaks at 483 • C and 533 • C were observed. The three peaks can be attributed to the dehydration of alginate, degradation of alginate and formation of CaCO 3 , respectively. For γ-Fe 2 O 3 , the TGA/TDA (Figure 3b) curve suggested a weight loss of 8.16% at temperatures ranging from 30 to 240 • C, and 12.53% at temperatures from 240 to 475 • C. These mass losses could be ascribed to two endothermic peaks (at 89.27 and 305.4 • C) related to the evaporation of the water adsorbed on the surface of γ-Fe 2 O 3 and to the degradation of organic matter resulting from the presence of citrate ions, respectively. The third loss in weight (16.08%) at 593.67 • C was attributed to the transition of γ-Fe 2 O 3 to hematite, α-Fe 2 O 3 [45]. The total loss in mass for γ-Fe 2 O 3 was of 36.77%. Above 845 • C, a γ-Fe 2 O 3 mass weight of 63.22% remained. Likewise, the total weight loss of the f-CNT (Figure 3c) was only of 8.53%, with an endothermic peak at 85 • C. In the case of A/γ-Fe 2 O 3 (Figure 3d), three losses in mass (9, 27.77 and 14.76%) were observed at 30-758 • C. These losses of mass are due to the loss of water, the thermal degradation of the alginate and citrates, the formation of CaCO 3 and the transition of γ-Fe 2 O 3 to α-Fe 2 O 3 , respectively. The presence of alginate in A/γ-Fe 2 O 3 decreased the thermal stability of γ-Fe 2 O 3 , leading to a total weight loss of 51.53%. The residue corresponded to metal oxides and quicklime. The results showed that A/γ-Fe 2 O 3 is stable at temperatures above 758 • C and that the material contains 48 ± 1 wt% of inorganic matter and 51 ± 1 wt% of organic matter, indicating that the mass ratio between inorganic and organic matter in A/γ-Fe 2 O 3 is approximately of 1.0. In the case of A/γ-Fe 2 O 3 /f-CNT (Figure 3e), the weight loss is higher than that observed in the other materials. Four losses in mass were observed; the first (10.13%) corresponded to the dehydration of the beads; the second (11.29%) and the third (66.52%) attributed to the degradation of the organic matter and the fourth (3.35%) was due to the degradation of CaCO 3 and the transition of γ-Fe 2 O 3 towards α-Fe 2 O 3 . The organic matter content was of 91.29%, while the inorganic matter was of 8.43%. These mass losses corresponded to one exothermic peak at 202 ºC and two endothermic peaks at 89 • C and 725 • C. fourth (3.35%) was due to the degradation of CaCO3 and the transition of γ-Fe2O3 towards α-Fe2O3. The organic matter content was of 91.29%, while the inorganic matter was of 8.43%. These mass losses corresponded to one exothermic peak at 202 ºC and two endothermic peaks at 89 °C and 725 °C.  Figure S2). γ-Fe2O3 nanoparticles are of spherical shape, showing many aggregations due to the drying before SEM analysis (Figure 4c). The elemental composition of γ-Fe2O3 obtained by EDX analysis was as follows: %C of 19.96, %O of 25.15, %Na of 7.92 and %Fe of 46.97%, confirming the coating of the particles with sodium citrates. The nanoparticles showed a larger size than that determined by XRD; this is due to the aggregation of nanoparticles because of van der Waals forces and magnetic repulsions [46]. For A/γ-Fe2O3 beads (Figure 4d), their external surface appeared very rough with a uniform dispersion of γ-Fe2O3 nanoparticles. Finally, in the case of A/γ-Fe2O3/f-CNT (Figure 4e), the surface is very rough and heterogeneous, observing areas with larger aggregations than those observed in A/γ-Fe2O3 beads. The coupling of f-CNTs with γ-Fe2O3 nanoparticles in the alginate matrix lead to improve the mechanical properties of the composite and to have aggregations in the beads [47].
The magnetization curves of γ-Fe2O3, A/γ-Fe2O3 and A/γ-Fe2O3/f-CNTs materials are shown in Figure 5. The superposition of the curves indicated that γ-Fe2O3 nanoparticles maintained their magnetic properties when they were encapsulated in the beads. According to the shape of the obtained curves, the materials have a magnetic memory with superparamagnetic behavior. The   The magnetization saturation of A/γ-Fe2O3/f-CNT beads was slightly lower than that of A/γ-Fe2O3 beads, but it stills enough to be efficiently separated from the aqueous medium. From these magnetization measurements, the mass fractions of γ-Fe2O3 nanoparticles in A/γ-Fe2O3 and A/γ-Fe2O3/f-CNT beads were estimated to be 65 and 60%, respectively. These results revealed that the synthesized magnetic beads are excellent materials for wastewater treatment in large-scale applications because they can be easily separated from water using a magnetic field.    The magnetization saturation of A/γ-Fe2O3/f-CNT beads was slightly lower than that of A/γ-Fe2O3 beads, but it stills enough to be efficiently separated from the aqueous medium. From these magnetization measurements, the mass fractions of γ-Fe2O3 nanoparticles in A/γ-Fe2O3 and A/γ-Fe2O3/f-CNT beads were estimated to be 65 and 60%, respectively. These results revealed that the The magnetization saturation of A/γ-Fe 2 O 3 /f-CNT beads was slightly lower than that of A/γ-Fe 2 O 3 beads, but it stills enough to be efficiently separated from the aqueous medium. From these magnetization measurements, the mass fractions of γ-Fe 2 O 3 nanoparticles in A/γ-Fe 2 O 3 and A/γ-Fe 2 O 3 /f-CNT beads were estimated to be 65 and 60%, respectively. These results revealed that the synthesized magnetic beads are excellent materials for wastewater treatment in large-scale applications because they can be easily separated from water using a magnetic field.

Effect of Initial PH in Adsorption Tests
The pH of the solution is an important factor that could affect the adsorption process by controlling both the surface chemistry of the adsorbent and the speciation of the adsorbate. The experiments were carried out in a pH range of 3.0-10.0 and the results are reported in Figure 6. It could be observed that MB uptake increased from 37.1 to 82.7 mg g −1 with increasing the pH solution from 3 to 10. These results are directly related to the chemical surface properties of the adsorbent and MB molecules. synthesized magnetic beads are excellent materials for wastewater treatment in large-scale applications because they can be easily separated from water using a magnetic field.

Effect of Initial PH in Adsorption Tests.
The pH of the solution is an important factor that could affect the adsorption process by controlling both the surface chemistry of the adsorbent and the speciation of the adsorbate. The experiments were carried out in a pH range of 3.0-10.0 and the results are reported in Figure 6. It could be observed that MB uptake increased from 37.1 to 82.7 mg g −1 with increasing the pH solution from 3 to 10. These results are directly related to the chemical surface properties of the adsorbent and MB molecules. Thus, the point of zero charge (pHPZC) of the beads was found to be 7.2 in MB solution [48]. The results showed that a remarkable decreasing of the adsorption capacity was observed at pH values < pHPZC, related to the pKa values of alginate (3.38 and 3.65) [46]. At pH values higher than 3.65, more negative charges are generated on the surface of the beads, while MB molecules showed positive charge, thus promoting the adsorbate-adsorbent electrostatic interactions between the negatively charged groups (-COOH and -OH) in the beads surface and the positively charged molecules of MB. In addition, at pH < pKa values of alginate, other interactions such as hydrogen bonding between sulfur and/or nitrogen atoms in MB and protonated carboxylate groups might be dominant in the adsorption process [34]. Same results were found in the literature for MB adsorption onto other composite beads [5][6][7][8][9][10][11][12][13][14][15][16]49]. Figure 7 shows the kinetic adsorption curve of MB adsorption onto A/γ-Fe2O3/f-CNT composite beads. Adsorption increased with increasing contact time due to the abundant availability of active sites on the adsorbent structure [50]. After, these sites were sequentially occupied and the adsorption capacity assume less valuable. In this case, the equilibrium time was achieved within 48 hours. The experimental kinetic data were fitted to the non-linear form of pseudo-first order (PFO) and pseudosecond order (PSO) models (Equations (5) and (6) Thus, the point of zero charge (Ph PZC ) of the beads was found to be 7.2 in MB solution [48]. The results showed that a remarkable decreasing of the adsorption capacity was observed at Ph values < Ph PZC , related to the pKa values of alginate (3.38 and 3.65) [46]. At Ph values higher than 3.65, more negative charges are generated on the surface of the beads, while MB molecules showed positive charge, thus promoting the adsorbate-adsorbent electrostatic interactions between the negatively charged groups (-COOH and -OH) in the beads surface and the positively charged molecules of MB. In addition, at Ph < pKa values of alginate, other interactions such as hydrogen bonding between sulfur and/or nitrogen atoms in MB and protonated carboxylate groups might be dominant in the adsorption process [34]. Same results were found in the literature for MB adsorption onto other composite beads [5][6][7][8][9][10][11][12][13][14][15][16]49]. Figure 7 shows the kinetic adsorption curve of MB adsorption onto A/γ-Fe 2 O 3 /f-CNT composite beads. Adsorption increased with increasing contact time due to the abundant availability of active sites on the adsorbent structure [50]. After, these sites were sequentially occupied and the adsorption capacity assume less valuable. In this case, the equilibrium time was achieved within 48 h. The experimental kinetic data were fitted to the non-linear form of pseudo-first order (PFO) and pseudo-second order (PSO) models (Equations (5) and (6), respectively) [51,52]:

Kinetic Studies
q t = k 2 q 2 e t 1 + k 2 q e t (6) where, q e and q t (mg g −1 ) are the adsorption capacities at equilibrium and at any time, respectively; and k 1 (min −1 ) and k 2 (g mg −1 min −1 ) are the adsorption rate constants of the PFO and PSO models, respectively. The fitting of the experimental data to both models is illustrated in Figure 7. According to the calculated values of R 2 and RMSE, it could be concluded that the adsorption process was well-fitted by the PSO model, obtaining a rate constant value of 0.030 g mg −1 min −1 .
Appl. Sci. 2019, 9,x FOR 11 of 19 where, qe and qt (mg g −1 ) are the adsorption capacities at equilibrium and at any time, respectively; and k1 (min −1 ) and k2 (g mg −1 min −1 ) are the adsorption rate constants of the PFO and PSO models, respectively. The fitting of the experimental data to both models is illustrated in Figure 7. According to the calculated values of R 2 and RMSE, it could be concluded that the adsorption process was wellfitted by the PSO model, obtaining a rate constant value of 0.030 g mg −1 min −1 .

Adsorption Isotherm Studies.
The MB adsorption isotherm data were evaluated by three adsorption isotherm equations, e.g., Langmuir, Freundlich and Sips models [53][54][55]. Langmuir model is related to a homogeneous monolayer adsorption process, while Freundlich model is widely used for heterogeneous multilayer adsorption. Overall, Sips model is a combination of both Langmuir and Freundlich equations. Langmuir, Freundlich and Sips models can be expressed as follows (Equations (7)-(9), respectively):

?
m L e e L e q K C q K C = + 1/n e F e q K C = where, qe (mg g −1 ) is the equilibrium adsorption capacity and Ce (mg L −1 ) is the equilibrium MB concentration in solution; qm (mg g −1 ) is the maximum monolayer capacity and KL (L mg −1 ) is the Langmuir constant, indicative of the affinity of the adsorbent towards the pollutant. Likewise, KF (L g −1 ) and 1/n are parameters of Freundlich equation; the value of 1/n is representative of the adsorption intensity. Finally, bs (L mg −1 ) is a constant of Sips equation, which depends on the temperature, and 1/s is another parameter related to the adsorption energy. Experimental and theoretical MB adsorption data are shown in Figure 8 and the calculated adsorption equilibrium parameters are collected in Table 2. According to the obtained low RMSE and high R 2 values, it could be affirmed that Sips model described very satisfactorily the experimental MB equilibrium adsorption data. Thus, the maximum adsorption capacity calculated from the model

Adsorption Isotherm Studies
The MB adsorption isotherm data were evaluated by three adsorption isotherm equations, e.g., Langmuir, Freundlich and Sips models [53][54][55]. Langmuir model is related to a homogeneous monolayer adsorption process, while Freundlich model is widely used for heterogeneous multilayer adsorption. Overall, Sips model is a combination of both Langmuir and Freundlich equations. Langmuir, Freundlich and Sips models can be expressed as follows (Equations (7)-(9), respectively): q e = q m K L C e 1 + K L C e (7) q e = K F C 1/n e (8) where, q e (mg g −1 ) is the equilibrium adsorption capacity and C e (mg L −1 ) is the equilibrium MB concentration in solution; q m (mg g −1 ) is the maximum monolayer capacity and K L (L mg −1 ) is the Langmuir constant, indicative of the affinity of the adsorbent towards the pollutant. Likewise, K F (L g −1 ) and 1/n are parameters of Freundlich equation; the value of 1/n is representative of the adsorption intensity. Finally, b s (L mg −1 ) is a constant of Sips equation, which depends on the temperature, and 1/s is another parameter related to the adsorption energy. Experimental and theoretical MB adsorption data are shown in Figure 8 and the calculated adsorption equilibrium parameters are collected in Table 2. According to the obtained low RMSE and high R 2 values, it could be affirmed that Sips model described very satisfactorily the experimental MB equilibrium adsorption data. Thus, the maximum adsorption capacity calculated from the model (320.5 mg g −1 ) was found very close to the experimental value (307.0 mg g −1 ), suggesting the homogenous surface of the adsorbent and the formation of a monolayer of MB molecules on its surface in the studied range of MB concentration (5-270 mg L −1 ). In this study, 1/n value resulted in 0.47, lower than 1.0, indicative of a favorable adsorption process.  The maximum MB adsorption capacity values on several adsorbents reported in the literature are collected in Table 3. The significant adsorption capacity of A/γ-Fe2O3/f-CNT revealed that this composite is an excellent adsorbent for the removal of MB contaminant from wastewater.   The maximum MB adsorption capacity values on several adsorbents reported in the literature are collected in Table 3. The significant adsorption capacity of A/γ-Fe 2 O 3 /f-CNT revealed that this composite is an excellent adsorbent for the removal of MB contaminant from wastewater.

Statistical Physics Analysis (Monolayer Model)
The monolayer model [60] can be considered as the general case of Langmuir model. It is applied to describe an entire set of experimental data. The monolayer model with single energy assumes that a site can accept "n" number of molecules. The equation of this model is: where, n is the number of molecules fixed per adsorbent surface, C 1/2 (mg L −1 ) is the concentration at half saturation, N M (mg g −1 ) is the receptor site density effectively occupied, q e (mg g −1 ) and C e (mg L −1 ) are the MB adsorption capacity and concentration, respectively, at equilibrium time.
According to the study reported by Meili et al. [62], if the number of dye molecules is lower than 1.0, it indicates the multi-anchorage adsorption process, which reflects a parallel position of the dye molecule adsorbed on the surface. This means that the dye molecules can simultaneously interact with several functional groups present on the adsorbent surface. On the contrary, if n value is higher than 1.0, adsorption can be considered as a multi-molecular process in which the molecules can interact in vertical position with the functional groups of the beads surface.
In this study, n parameter took a value of 1.62. Hence, the adsorption process can be considered as a multi-molecular process, assuming a vertical geometry of the dye molecule in the adsorption process. Based on this model, it is possible to characterize the interactions between the MB molecules and the beads surface by the adsorption energy determination. The energetic parameter can be calculated from: where, C s (mg L −1 ) is the solubility of MB in water and C 1/2 (mg L −1 ) is the half saturation concentration determined by the fitting method. The value of the energy of adsorption was found of 19.1 kJ mol −1 , suggesting that the adsorption is of endothermic and physical nature. The same result was obtained by Meili et al. [62] for the adsorption of MB onto an agricultural Algerian olive cake waste.

Thermodynamic Adsorption
The effect of the temperature on MB adsorption onto A/γ-Fe 2 O 3 /f-CNT composite beads was studied at four different values, i.e., 283, 293, 303 and 313 K. The results showed a positive correlation between MB adsorption and temperature, suggesting the endothermic nature of the process, which is in agreement with the results obtained from the monolayer model. The thermodynamic parameters, e.g., the enthalpy or heat of adsorption (∆H • ), the change in the standard Gibbs' energy (∆G • ) and entropy (∆S • ) were calculated using the following equations [27]: log where, q e (mg g −1 ) is the equilibrium MB adsorption capacity, C e (mg L −1 ) is the equilibrium MB concentration, ∆H • (kJ mol −1 ) is the enthalpy variation, ∆S • (J mol −1 K −1 ) is the entropy, ∆G • (kJ mol −1 ) is the change in the standard Gibbs' energy, K c is the equilibrium constant, T (K) is the temperature and R is the universal gas constant (8.314 J mol −1 K −1 ). The enthalpy (∆H • ) and entropy (∆S • ) parameters were evaluated by plotting log(q e /C e ) versus 1/T, resulting in values of 36.9 kJ mol −1 and 195.1 J mol −1 K −1 , respectively. The obtained positive value of the enthalpy is indicative of an endothermic process. Thus, the enthalpy value resulted smaller than 40 kJ mol −1 , indicating that MB adsorption onto A/γ-Fe 2 O 3 /f-CNT composite beads can be considered as a physisorption process, governed by electrostatic interactions, mainly van der Waals forces. ∆G • parameters at 283, 293, 303 and 313 K were of −18.3, −20.2, −22.2 and −24.1 kJ mol −1 , respectively. The negative ∆G • values indicated the spontaneous and the physical nature of MB adsorption process onto the synthesized composites. The change in the adsorption entropy (∆S • ) was positive, suggesting that the MB molecules increased their randomness at the solid/liquid interface during the adsorption.

Proposed Adsorption Mechanism of MB onto A/γ-Fe 2 O 3 /f-CNT Composite Beads
The adsorption mechanism was investigated by FTIR analysis. The FTIR spectra of MB and A/γ-Fe 2 O 3 /f-CNT composite before and after adsorption are depicted in Figure 9. The adsorption mechanism of methylene blue has been extensively reported in the literature. Thus, Wang et al. [63] have previously found that MB molecules could be adsorbed on alginate and acidified MWCNTs beads, on one hand, by π-π dispersion interactions with the aromatic regions of MWCNTs, and, on the other hand, by electrostatic attractions with the carboxylate groups present in alginate and MWCNTs. So, different adsorption mechanisms can be proposed when the adsorbent is a composite formed by different materials. The interactions between MB molecule and A/γ-Fe 2 O 3 /f-CNTs beads were verified by FTIR analysis (Figure 9).

Adsorbent Recycling.
As it was elucidated in the pH effect adsorption studies, A/γ-Fe2O3/f-CNT composite beads seem to be ineffective for MB removal at pH values lower than 3, indicating a high possibility that MB adsorbed ions be desorbed in acidic medium (0.1M HNO3 solution). Desorption efficiencies of the accomplished consecutive adsorption-desorption cycles are depicted in Figure 10.
At low pH values, the functional groups present on the beads surface are protonated, which leads to reduce the electrostatic attractions between MB molecules and the adsorbent surface. This conduct to the diffusion of MB molecules to the acid solution. As it could be observed in the Figure,  6 consecutive adsorption-desorption tests using a nitric acid solution as elution reagent were assessed. Thus, A/γ-Fe2O3/f-CNT composite beads could be reused for 6 cycles maintaining its high adsorptive ability, observing similar MB adsorption/desorption ratio, ranging from 82 to 87%. The results confirmed that A/γ-Fe2O3/f-CNT beads are efficient adsorbents of methylene blue, since they can be rapidly and effectively regenerated. Same findings have been described in the literature [66]. After adsorption, the FTIR spectrum of A/γ-Fe 2 O 3 /f-CNT clearly showed the MB signature, indicating its adsorption onto the composite beads. The bands appearing at 1394 and 1339 cm −1 are referred to the aromatic rings of MB. The characteristic C=C band of the aromatic rings detected at 1179 cm −1 in the MB spectrum appeared at 1175 cm −1 in the FTIR of A/γ-Fe 2 O 3 /f-CNT after adsorption [64]. In addition, the C-H bond vibrations of MB at 885, 1220 and 1249 cm −1 in A/γ-Fe 2 O 3 /f-CNT/MB were observed [65]. A shift in frequency values from 1622 to 1629 cm −1 for the carboxylate and carbonyl groups of A/γ-Fe 2 O 3 /f-CNT beads after adsorption was determined. Variation in intensity, shift and occurrence of new vibrations in A/γ-Fe 2 O 3 /f-CNTs-MB suggested that the mechanism of MB uptake by A/γ-Fe 2 O 3 /f-CNTs beads may be controlled by electrostatic interactions involving the functional groups present in the beads. These results are in agreement with the studies of pH and temperature effects on MB adsorption.

Adsorbent Recycling
As it was elucidated in the pH effect adsorption studies, A/γ-Fe 2 O 3 /f-CNT composite beads seem to be ineffective for MB removal at pH values lower than 3, indicating a high possibility that MB adsorbed ions be desorbed in acidic medium (0.1M HNO 3 solution). Desorption efficiencies of the accomplished consecutive adsorption-desorption cycles are depicted in Figure 10. Figure 9. FTIR spectra of MB and A/γ-Fe2O3/f-CNT beads before and after MB adsorption.

Adsorbent Recycling.
As it was elucidated in the pH effect adsorption studies, A/γ-Fe2O3/f-CNT composite beads seem to be ineffective for MB removal at pH values lower than 3, indicating a high possibility that MB adsorbed ions be desorbed in acidic medium (0.1M HNO3 solution). Desorption efficiencies of the accomplished consecutive adsorption-desorption cycles are depicted in Figure 10.
At low pH values, the functional groups present on the beads surface are protonated, which leads to reduce the electrostatic attractions between MB molecules and the adsorbent surface. This conduct to the diffusion of MB molecules to the acid solution. As it could be observed in the Figure,  6 consecutive adsorption-desorption tests using a nitric acid solution as elution reagent were assessed. Thus, A/γ-Fe2O3/f-CNT composite beads could be reused for 6 cycles maintaining its high adsorptive ability, observing similar MB adsorption/desorption ratio, ranging from 82 to 87%. The results confirmed that A/γ-Fe2O3/f-CNT beads are efficient adsorbents of methylene blue, since they can be rapidly and effectively regenerated. Same findings have been described in the literature [66]. At low pH values, the functional groups present on the beads surface are protonated, which leads to reduce the electrostatic attractions between MB molecules and the adsorbent surface. This conduct to the diffusion of MB molecules to the acid solution. As it could be observed in the Figure, 6 consecutive adsorption-desorption tests using a nitric acid solution as elution reagent were assessed. Thus, A/γ-Fe 2 O 3 /f-CNT composite beads could be reused for 6 cycles maintaining its high adsorptive ability, observing similar MB adsorption/desorption ratio, ranging from 82% to 87%. The results confirmed that A/γ-Fe 2 O 3 /f-CNT beads are efficient adsorbents of methylene blue, since they can be rapidly and effectively regenerated. Same findings have been described in the literature [66].

Conclusions
Magnetic beads of alginate citrate-stabilized maghemite nanoparticles and functionalized multiwalled carbon nanotubes (A/γ-Fe 2 O 3 /f-CNT) were prepared, characterized and used as adsorbent for methylene blue dye. The kinetic and equilibrium experimental data were well-explained by the pseudo-second order kinetic model and Sips isotherm equation, respectively. The maximum adsorption capacity determined by Sips model was of 320.5 mg g −1 , very close to the experimental value (307.0 mg g −1 ), suggesting a homogenous adsorption process in the studied range of MB concentrations. The studies of pH and temperature effect on the adsorption process revealed that the beads exhibited an important adsorptive stability over a large range of pH values (4-10) and the endothermic and spontaneous nature of the process. The statistical study using the monolayer model demonstrated the multi-molecular nature of the adsorption process, with a vertical adsorption geometry of the dye. The magnetic beads could be used at least for six adsorption/desorption successive cycles, maintaining its adsorptive capacity. According to the results of this study, the synthesized composite beads can be considered as high effective adsorbents of the model compound, revealing as excellent materials for dyes removal in flow-through adsorption WWTP applications. Funding: This research has not received any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.