INTRODUCTION 
Daily release of large amounts of dyes through wastewater from various industries such as textiles, leather, paper and plastics can be very dangerous to the environment. Due to their complex structure, dyes are often stable and resistant to biodegradation, which are often toxic, carcinogenic and mutagenic [1, 2]. Production of toxic amines through reduction and breaking of azo bonds is another problem of depletion of dyes into water sources, which leads to severe effects on vital human organs such as the brain, liver, kidneys, reproductive system and central nervous system [3]. Very small amounts of dyes are clearly visible in the water and may be toxic to organisms in the water. Therefore, the removal of dyes from the process or wastewater will have an impact on the environment. 
The methods used to remove dyes mainly include physical, chemical and biological methods [4]. Most of these methods are costly and lead to the production of sludge and even dangerous by-products. Among these methods, the adsorption process by inexpensive adsorbents including clay, zeolites, silicates, agricultural wastes, bio-sorbents in terms of initial cost, wastewater reuse, simplicity and flexibility in design, easy operation and non-toxicity is a superior method. Production of high-quality effluent and no formation of hazardous substances such as ozone and free radicals are other advantages of this method [5]. 
Many studies have been conducted to find suitable adsorbents to reduce the concentration of colorants in water, including activated carbon [6, 7], peat [8, 9], chitin [10-12], silica [13], and others [14-16]. 
In recent years, many adsorbents based on agricultural wastes (also activated carbon and by-products such as almond shell [5], glossogyne tenuifolia leaves  [17], lignocellulosic waste [18] have been reported and applied for the removal of ionic and nonionic dye molecules from water.
Activated carbon (AC) is considered as a non-toxic adsorbent and universal adsorbent for the removal of pollutants such as dyes and heavy metals from the wastewater due to its high surface area, porous structure, high adsorption capacity, fast adsorption kinetics and general material as support for loading nanomaterials. AC contain various reactive sites such OH, COOH and C=O groups which together with the nanoparticle properties synergically improve the efficiency of adsorption procedure [19].
Unfortunately, two important disadvantages of using activated carbon as an adsorbent include: 1) activated carbon is difficult to separate from the solution and, after use in water and wastewater treatment, is discarded with process sludge, leading to secondary contamination [20].  2) The cost of AC is relatively high and they are difficult to reproduce, which limits the use of activated carbon in paint wastewater treatment.
To overcome these problems, many researchers and scientists are developing magnetic adsorbents from low-cost materials available as alternatives to activated carbon materials. But these alternative materials must have sufficient adsorption capacity as well as good reproductive capacity. 
On the other hand, because of their ease of control and rapid separation, magnetized sorbent have recently been used in water treatment [21, 22] and used to remove dyes and other water molecules [23-25]. The higher adsorption capacity of magnetic adsorbents for cationic basic dyes has been attributed to the interactions of hydroxyl groups with dye molecules [26]. 
It is also important to combine a low-cost material (such as seashell) with activated carbon that reduces its consumption in the production of composite adsorbents.
The main objective of this work is to introduce new magnetic adsorbent from sea shell (SS) and activated carbon (WAC) that produced from walnut (SS/WAC/Fe3O4 nanocomposite) and study the adsorption potential of SS/WAC/Fe3O4 nanocomposite for the removal of methylene blue (MB) dye from aqueous solution) Fig. 1). The effects of several operating parameters such as pH of solution, adsorbent dosage, initial MB dye concentration and adsorption time on the MB dye adsorption were investigated and optimized by Box-Behnken design (BBD) combined with response surface methodology (RSM).  
The adsorption mechanism of MB dye onto SS/WAC/Fe3O4 nanocomposite is also investigated in this report. However, there is no information available in literature regarding the optimization of MB dye adsorption on SS/WAC/Fe3O4 nanocomposites. 

MATERIALS AND METHODS
Instruments and reagents
Iron(III) chloride hexahydrate (FeCl3.6H2O), Iron(II) chloride tetrahydrate (FeCl2.4H2O) Nitric acid, Hydrochloric acid, sodium hydroxide and Methylene blue with the highest purity available were purchased from Merck chemical company. By dissolve 100 mg of solid dye in 500 ml of double-distilled water was prepared the MB stock solution (200 mg L1), and dilute it appropriately to obtain the daily working concentration. The high-purity chemicals come from the chemical companies Merck and Aldrich. A Varian model 640 spectrophotometer was used to record FT-IR spectra using pressed KBr pellets. A double beam Perkin-Elmer Lambda 25 spectrophotometer was used to record UV-Vis spectra to ensure the formation of nanoparticles. A Philips model X’PertPro diffractometer (Cu Ka = 1.5406 Å) was used to carry out X-ray diffraction (XRD) measurements at a scanning rate of 2º/min in the 2θ range (10-80). Scanning electron microscopy (SEM) was performed on a TESCAN model VEGA3-XMU equipped with energy dispersive spectroscopy (EDS) analysis. FESEM images were recorded using TESCAN MIRA3 FESEM operating at 15 kV. The chemical compositions of the prepared nanostructures were determined by EDS (energy dispersive X-ray spectroscopy) performed in FESEM. The BET specific surface area (SBET), pore volume (micropore volume, Vmi; and total pore volume, VT), and pore size distribution of the prepared magnetic adsorbent were determined by N2-physisorption using the surface area analyzer model Autosorb-1C (Quantachrome, USA). The multipoint Brunauer, Emmett, and Teller (BET) surface area was determined from the nitrogen adsorption/desorption isotherm. The BET isotherm equation is typically applied on the adsorption isotherm in P/P0 range of 0.02-0.99. Prior to the measurement, the sample was degassed at 200 °C for 6 h in an out-gassing station to remove any adsorbed water or entrapped gases in the sample.

Synthesis of Fe3O4 nanoparticles
The magnetization of alternative ferrite nanoparticles synthesized by co-precipitation method depends more on parameters such as reaction temperature, suspension pH, initial molar concentration , etc. [27]. 
In a typical synthesis, FeCl3. 6H2O (1.1 g) and FeCl2.4H2O (0.54 g) were dissolved in 100 ml of distilled water under constant stirring and blowing nitrogen for 60 min. The resulting solution is then placed in an oil bath adjusted at 85 °C. Afterwards, NaOH (25 mL and 5 M) as the pH controller agent was subsequently added dropwise to the solution under stirring and blowing nitrogen to reach pH value 10. After that, the solution is then kept in an oil bath at 85 °C under a nitrogen atmosphere for 30 minutes. Finally, the brown precipitates were washed several times with ethanol and distilled water and dried at 60 °C for 24 h.

Preparation of activated carbons from walnut shell (WAC)
In the usual synthesis, the waste walnut shells are first washed with distilled water and dried at 90°C for 6 hours in an oven, and mechanically pretreated by crushing in a mill, therefore, 5 g of walnut shell was washed several times with distilled water. Afterwards, the walnut shell was heated at 450 °C for 20 min.  The resulting product was then mixed with NaOH (3:1, NaOH: walnut shell) and heated at 400 °C for 25 min.  After that, the black product was washed several times with distilled water. Finally, the final product obtained was dried at 120 °C for 4 h [28].

Preparation of magnetized sea shell and Walnut shell based activated carbon (SS/WAC/Fe3O4)
The synthesis method was referring to some reported work [27, 29]. In brief, as-prepared Fe3O4 (200 mg) nanoparticles were dispersed in a 0.1 mol L-1 HNO3 solution for 10 min, then followed by washing with deionized water several times. Subsequently, the treated magnetic nanoparticles were dispersed in 25 mL solution contain 0.1 g seashell and 0.1 g activated carbon that obtained from walnut shell, under vigorous stirring for 20 min. After stirring for another 20 min in ultrasonic bath, the solution was transferred to a Teflon-sealed autoclave and kept at 180 °C for 12 h then cooled naturally. The resulting product was separated with a magnet and washed several times with deionized water and ethanol was dried at 60 °C for 12 h. 
In order to achieve the role of seashell as well as activated carbon in adsorption efficiency, in addition to SS/WAC/Fe3O4 magnetic nanocomposites, WAC/Fe3O4 and SS/Fe3O4 samples were also prepared as mentioned above.

Experimental design
Box-Behnken experimental design (BBD) was used to investigate the adsorption efficiency (or MB removal percentage) of SS/C/Fe3O4 for MB (response). Minitab software (version 14) was used to design of experiments and analysis of the experimental results. As shown in Table 1, a three-level BBD was employed to evaluate the significance of the effects of parameters including pH of solution, adsorbent dosage (g L-1), initial MB dye concentration (mgL-1) and adsorption time (min) and a total of 27 runs were carried out (Table 2). The order of the experiments was randomized to minimize the effects of uncontrolled factors.
A second-order polynomial equation was utilized to fit the MB removal percentage as presented in the following Eq:

Where Ypredicted is the predicted response MB removal percentage. The parameter β0 is the constant coefficient and βi is the linear coefficient of the input parameter, βii, are the quadratic coefficients,  βij, are the different interaction coefficients between input parameters, while Xi and Xj represent the independent variables and ε is the error of the model [30, 31].

Adsorption isotherms
The adsorption experiment was carried out in a batch mode as follows: 0.232 g L-1 dosage of  SS/WAC/Fe3O4 adsorbent, initial concentration of dye was 1–10 ppm and the pH of the solution was retained 6.5 using 0.1 M NaOH and 0.1 M HCl solution. The different concentration solutions were positioned in the water bath shaker at 100 rpm for 40 min. The aliquots for MB were examined in UV–Vis Spectrometer at the wavelength of 667 nm.

Measurements of dye uptake
The dye concentrations were determined according to calibration curve obtained at maximum wavelength (667 nm) over working concentration range. The efficiency of MB removal was determined at different experimental conditions optimized according to the BBD method. The experiments were also performed in the initial MB concentration range of 1-10 mg L-1 to obtain adsorption isotherms. The percentage of MB removal is calculated using the following formula:

Where C0 (mg L-1) and Ct (mg L-1) are the target concentrations at the beginning and subsequent time t respectively. The adsorbed MB amount (qe (mg g1)) was calculated by the following mass balance relationship:

Where C0 and Ce (mg / L) are the initial concentration and equilibrium concentration of the dye in the aqueous solution, V
where C0 and Ce (mg L-1) are the initial concentration and equilibrium concentration of the dye in the aqueous solution, respectively. Also, V (L) and W (g) are the volume of the solution and the mass of the adsorbent, respectively. 

RESULTS AND DISCUSSION
Characterization of SS/WACC/Fe3O4
The FT-IR spectra of Seashell, activated carbon, Fe3O4 and SS/WAC/ Fe3O4 are shown in Fig. 2. The strong band at 1014 cm-1 in the FTIR of WAC spectrum can be attributed to C-C coupling. 400 to 600 cm-1 refers to the metal-oxygen (Fe-O) stretching mode in the Fe3O4 structure [32]. The FT-IR spectrum data obtained for seashell shows the absorption bands of at 1463 cm-1, 1083 cm-1, 858 cm-1, and 714 cm-1, which are described as characteristics of carbonate ions in calcium carbonate and it is the main vibration mode of this particular compound [33]. The band at 3450 cm-1 is attributed to (O-H) vibrations in hydroxyl groups. The band appearing at 1770.71 cm-1 is attributed to carbonyl (C=O) groups. The appearance of band at 1143 cm-1 can be assigned to (C-O) stretching vibrations. The peak at around 889.21 and 669.32 cm-1 is ascribed to the aromatic C-H out of-plane vibrations [34].  The FT-IR analysis of SS/WAC/Fe3O4 confirm that there was no change in functional groups.  
Fig. 3 shows the X-ray diffraction patterns of shells, activated carbon, Fe3O4 and SS/WAC/Fe3O4 nanocomposites. Analysis of the diffraction pattern of Fe3O4 shows a cubic spinel structure and corresponding to the ICDD card no 01-1111, which is characteristic of this phase due to the strong reflection on the (311) plane. The peaks designated as (220), (311), (400), (422), (511), and (440) planes correspond to the characteristics of the cubic unit cell of the cubic spinel structure [35]. After calculating the grain size with the Scherer formula, we get the Fe3O4 particle size as 20.22 nm. 
In the XRD pattern of seashell (SS) corresponding to the ICDD card no 05-0453, the peak at 2θ = 29.5 is the characteristic peak of aragonite (CaCO3)  [36]. The XRD diffraction pattern of Fe3O4, SS and WAC is compared with the XRD pattern of SS/WAC/Fe3O4 in Fig. 3, which only confirms the presence of SS, WAC and Fe3O4 compounds in SS/WAC/Fe3O4.
Field emission scanning electron microscopy (FE-SEM) is used to observe the surface physical surface morphology of Seashell, Fe3O4 and SS/WAC/Fe3O4 (Fig. 4). Fe3O4 has a spherical structure with a size in the range 40-60 nm, and walnut activated carbon (WCA) has a microporous structure. SEM image of the WAC is shown in Fig. 4. Many pores are clearly visible on the surface of the WAC. The morphology of seashell indicates that the seashell has a sheet like structure.  FE-SEM of the SS/WAC/Fe3O4, clearly shows the presence of SS, WAC and Fe3O4 is clearly visible.
The elemental composition of the SS/WAC/Fe3O4 nanocomposite was also analyzed by EDS spectroscopy (Fig. 5), which also showed C (12.73%), Ca (22.73%), O (54, 33%) and Fe (10.65 %) in the SS/WAC/Fe3O4 nanocomposite; in addition, the elemental mapping image confirmed the presence of C, Ca and Fe (Fig. 6).
The hysteresis curve of the SS/WAC/ Fe3O4 nanocomposite measured at a magnetic field of 8 to +8 kOe and room temperature (298 K), is shown in Fig. 7, showing superparamagnetic properties and high saturation magnetization. From 0 to 8 kOe, the magnetization (M) increases sharply; M is almost saturated at around 2 kOe. For SS/WAC/ Fe3O4 nanocomposites, the saturation magnetization (Ms) is about 23.97 emu/g, and the reported specified Ms value is 84 emug1 for bulk Fe3O4 particles and 65 emug1 for Fe3O4 nanoparticles [37, 38]. The measured magnetization of nanocomposite was found to be considerably lower than the values measured from bulk magnetite [39].The decrease in saturation magnetization of SS/WAC/Fe3O4 nanocomposite can be attributed to the decrease in Fe3O4 content in the composite and the diamagnetic contribution from SS and WAC, but it was still large enough to be used as magnetically separable adsorbent.
The determination of the iodine value is usually an additional test of the N2/77 K adsorption isotherm, which is used to measure the surface of the microspores in the pores of the material. The specific surface area (SSA) is a measure of the adsorbent surface area per unit mass and generally depends on the surrounding phase that can modify the surface [40]. 
The pore size distribution, surface area and N2 adsorption/desorption isotherm of SS/WAC/Fe3O4 nanocomposite are analyzed by BET analyzer as shown in Fig. 8. The surface area of the SS/WAC/Fe3O4 nanocomposite is found to be 101.75 m2 g-1 which is slightly greater than the surface area of a-Fe2O3 (14.6 m2 g-1), Fe3O4-C spindles (35.1 m2 g-1) [41] and limestone (12 m2 g-1) [42] as reported in the literature. The BJH pore size distribution is used to determine the pore volume of SS/WAC/Fe3O4 nanocomposite. The total pore volume and average pore diameter are 0.2358 cm3/g and 1.29 nm, respectively. The loop of the adsorption/desorption curve indicates type-IV adsorption isotherm. The particles are highly mesoporous and have uniform pore size [40].

Determine the isoelectric point of SS/WAC/Fe3O4 nanocomposite
Prepare 0.1 M NaCl solutions with initial pH (pHi) adjusted from 2 to 10 with HCl and NaOH. Take 10 flasks with a capacity of 50 mL for each 0.05 g SS/WAC/Fe3O4 bottle, add 25 mL NaCl solution was prepared above, for 24 h, then filter the solution and redefine the pH (pHf) of the solutions. The results from Fig. 9 determine the isoelectric point of SS/WAC/Fe3O4 is equal to 7.2. This shows that when pH < pHpzc, the surface of SS/WAC/Fe3O4 nanocomposite is positively charged, when pH > pHpzc, the surface of SS/WAC/Fe3O4 nanocomposite is negatively charged.

Optimization of MB removal parameters by SS/WAC/Fe3O4 nanocomposite 
In the BBD step, the adsorption experiments were carried out according to the design matrix (Table 3) and the main, interaction and quadratic effects were evaluated. Four independent variables at three levels were selected, and the responses of 27 experiments were reconstructed and presented in Table 3. Analysis of variance (ANOVA) was calculated using MINITAB 14 (Table 3). Degrees of freedom (DF), sum of squares (Seq SS), modified sum of squares (Adj SS), modified mean square (MS Adj), F statistics and P values of the model expression are listed in Table 3. Data analysis gave a semi-empirical expression of MB removal with following equation (4):
     

 Based on the analysis of Eq. (4) depicted that the variables x2 (adsorbent dosage) and x4 (adsorption time) exhibited a positive relationship in the MB removal by SS/WAC/Fe3O4 nanocomposite. 
The results of ANOVA in Table 3 show that the F-value of the model is 120.13, which indicates that the probability that the “model F value” is high noise is only 0.01%, and most of the changes in the responses can be attributed to regression and that equations and models are meaningful [30].
The Predicted R2 is 0.97911 was in reasonable agreement with the Adjusted R2 (0.9200).  As shown in Fig. 10, plot of experiment response (YExperimental) versus the predicted response indicated a good fit and it can be seen that the data points are well spaced around the straight line (R2 = 0.9664) and confirming the statistical validation of the model  [30].
The main effect plot was used to evaluate the parameters, which give high removal efficiency (Fig. 11). The main effect plot was used to evaluate para meters, which give high removal efficiency of MB (Fig. 11). Fig.11 shows sorbent dosage, in the range of 0.05-0.3 gL-1. Increase the adsorbent dose to 0.25 gL-1 increased the removal efficiency and the efficiency decreases become constant afterwards. Therefore, 0.25 gL-1 was chosen as the optimal absorbent dosage.  Addition of MB concentration upper 4.0 ppm, decreased MB removal efficiency, as shown by the results in Fig. 11. Therefore, 4.0 ppm was chosen as the optimal MB concentration.  The MB removal increased up to pH=5.5., after which that no increase was observed in MB removal efficiency until pH=7.5. Therefore, pH=6.5 was chosen as the optimal pH. This can be explained as follows: when the pH value < pHpzc surface of SS/WAC/Fe3O4 is positively charged due to H+ ion adsorption. Also methylene blue is also a cationic dye. Therefore, there is a repulsive force between the cationic dye and the adsorbent surface. In addition, at lower pH concentrations of large H+ occur competitive adsorption with positively charged cationic dyes at adsorption centers. Therefore, the adsorption efficiency is low at low pH values. At pH values > pHpzc surface of SS/WAC/Fe3O4, negatively charged by OH- adsorption and appearing electrostatic suction force between negatively charged SS/WAC/Fe3O4 and positive charged MB dye cause to high adsorption efficiency at large pH values [43].
Removal time which was studied in the range of 10-60 min (Fig. 11).  Increasing extraction time up to 40 min increased the removal efficiency for the MB. The efficiency decreases afterwards. Thus, the optimal removal time was 40 min. 
As a result, the proposed adsorbent (0.23 g L-1) is applicable for successful removal of MB (10 mg L-1) in short time (40.5 min) with high adsorption capacity (37.17 mg g-1).  The equation of adsorption capacity is presented in Eq. 6. 
Under optimal conditions, for comparison, the adsorption of methylene blue for different adsorbents was investigated and is shown in Table 4. As can be seen, the percentage of removal of MB in SS/WAC/Fe3O4 is higher than that of WAC/Fe3O4, SS/Fe3O4, WAC and SS. Certainly WAC together with the Fe3O4 nanoparticle properties and SS synergically improve the efficiency of adsorption procedure.

Adsorption isotherm
The adsorption isotherm provides information about the physicochemical properties of the adsorption process as a single process and are tools for describing and predicting type and intensity of the adsorbent and adsorbate interaction. In order to find a suitable model that can be utilized for design process, equilibrium data should appropriately fit into Langmuir, Freundlich and Redlich–Peterson models [44].
The accuracy of the model for predicting experimental data was evaluated with correlation coefficient (R2) and the results are given in the table 5.
The Langmuir isotherm simulates the monolayer adsorption of the adsorbate onto homogenous adsorbent surface without migration of adsorbate molecules in the surface plane. The Langmuir isotherm equation is expressed as follows:
                                                    
     Where b and a are Langmuir constants related to adsorption capacity (mg/g) and adsorption equilibrium constant (l/mg), respectively. qe (adsorption capacity) is the amount of dye adsorbed at equilibrium per gram of adsorbent and it is defined as follows:
                  
                                     
                                                                                                          
Ci and Ce are the initial and equilibrium concentration of dye solutions respectively (mgL-1), V is the volume of the solution (l) and m is the mass (g) of the dry adsorbent used. The linear plots of 1/qe versus 1/Ce suggest the applicability of the Langmuir isotherms. The values of b and a were determined from slope and intercepts of the plots. According to the Langmuir model results (Fig. 12 and Table 5), the maximum adsorption capacity (b) and adsorption equilibrium constant (a) was 37.17 mg/g and 8.67 l/mg respectively. The larger constant values ​​a and b indicate better absorption. The results ​​obtained in this study, is remarkable compared to previous studies.
The Freundlich isotherm model (Fig. 13) is an exponential equation, so it is assumed that there is an interaction between adsorbed molecules on a heterogeneous surface, and it is not limited to forming a single layer. This model is given as:
                                                                                                   
                                                                             

Where k and n are Freundlich constants related to adsorption capacity and adsorption intensity respectively. Their values are obtained from the intercept and slope respectively (Table 5) from the linear plot of logqe against logCe.
The results showed that MB adsorption on the surface of SS/WAC/Fe3O4 nanocomposite follows the Langmuir model and has the highest correlation coefficient. 
The maximum capacity (qe) obtained by Langmuir isotherm for MB sorption was 37.17 mg/g for SS/WAC/Fe3O4 nanocomposite sorbent, which indicates the applicability of SS/WAC/Fe3O4 nanocomposite sorbent for treatment of real wastewater containing a high amount of understudy MB and also indicates the superiority of SS/WAC/Fe3O4 nanocomposite sorbent in comparison to previously reported material (Table 6) and this sorbent is an effective and promising sorbent to sorption of dyes from aqueous solutions. activated carbon obtained from walnut shell contains several reaction centers, such as OH, COOH and C=O groups, which, together with the properties of nanoparticles, synergistically improve the efficiency of the adsorption process [21].

SS/WAC/Fe3O4 nanocomposite recyclability
The recovering and reusability test of the SS/WAC/Fe3O4 nanocomposite was also investigated for the removal of MB. After the completion of the reaction, the composite was separated from the MB solution by magnet and dispersed in 10 mL weak acetone solution and sonicated for 10 min and then washed with distilled water several times and then was explored for MB removal in the succeeding cycles. We repeat the above process up to five cycles. The concentration of MB in the solution is about 10 mg/L. Removal efficiency of MB was reduced from 93% in the first cycle to 85% in final cycle (Fig. 14). 

CONCLUSION
In this study, the possibility of modifying the surface of seashell with activated carbon obtained from walnut shell and Fe3O4 and also its efficiency to remove methylene blue dye from aqueous solution was investigated. FT-IR results showed that seashell surface modification with activated carbon and Fe3O4 was well done. Also, the FE-SEM image of SS/WAC/Fe3O4 showed that these Fe3O4 particles are in nanoscale and their size distribution is not wide, which increases the dye removal efficiency. In this study, effective parameters optimization was performed by optimizing the response surface methodology. Optimal adsorption conditions including pH=6.5, dye concentration of 4.0 mg L-1, adsorbent amount of 0.232 g L-1 and contact time of 40 min were obtained. The small amount of proposed adsorbent is applicable for successful removal of MB in short time with high adsorption capacity equal to 37.2 mg g-1.
The study of Langmuir and Freundlich equilibrium isotherms showed that the MB dye adsorption process on the SS/ C /Fe3O4 adsorbent is good and with a correlation coefficient of 0.9893 and 0.9779, respectively. The Langmuir correlation coefficient is higher than the Freundlich model, so the Langmuir isotherm is selected and introduced to predict the adsorption behavior of methylene blue dye from the aqueous medium by the adsorbent. The SS/WAC/Fe3O4 could be reused and recovered several times with no significant loss of adsorption capacity.

CONFLICT OF INTEREST
The authors declare that there is no conflict of interests regarding the publication of this manuscript.