Improving adsorption performance of l-ascorbic acid from aqueous solution using magnetic rice husk as an adsorbent: experimental and RSM modeling

In this research, rice husk (RH) was utilized to prepare a magnetic adsorbent for adsorption of ascorbic acid (AA). The magnetic agent is iron(III) chloride (FeCl3). The impact of acid concentration in the range of 400–800 ppm, adsorbent dosage in the range of 0.5–1 g, and contact time in the range of 10–130 min were studied. The Langmuir model had the highest R2 of 0.9982, 0.9996, and 0.9985 at the temperature of 15, 25, and 35 °C, respectively, and the qmax values in these temperatures have been calculated at 19.157, 31.34, and 38.75 mg/g, respectively. The pseudo-second-order kinetic model had the best agreement with the experimental results. In this kinetic model, the values of q have been measured at 36.496, 45.248, and 49.019 mg/g at the acid concentration of 418, 600, and 718 ppm, respectively. The values of ΔHo and ΔSo were measured 31.972 kJ/mol and 120.253 kJ/mol K, respectively, which proves the endothermic and irregularity nature of the adsorption of AA. Besides, the optimum conditions of the design-expert software have been obtained 486.929 ppm of acid concentration, 0.875 g of the adsorbent dosage, and 105.397 min of the contact time, and the adsorption efficiency in these conditions was determined at 92.94%. The surface area of the RH and modified RH was determined of 98.17 and 120.23 m2/g, respectively, which confirms the high surface area of these two adsorbents.

for this study. Several types of reagents have been exploited as the adsorbent for the removal of metallic ions from contaminated water such as activated carbon 9 , fruit wastes 10 , mineral substances [11][12][13][14] , microbes 15 , waste materials 16 , and polymers 17 . In this study, a kind of agricultural waste was used as a natural adsorbent. The annual production of food and crop waste is increasing extensively, thus, it is crucial to manage food waste 18 . For avoiding this challenge, food waste can be converted into beneficial materials. The reuse of agricultural waste is proposed as a convenient and economical approach. Due to the desired performances and the low price of agricultural waste such as banana peel, orange peel, rice husk (RH), tea pulp, walnut shell 19 , Montmorillonite clay 20 , chicken beak 21 , zeolite 22 , and etc., these materials have received much attention. Also, various reagents can be grafted to the structure of the adsorbents for improving their performance such as polymers, metal hydroxide, acids, iron, and other chemical materials like xanthate 23,24 . Some valuable works that demonstrate the superiority of the modified adsorbent over the raw ones are indicated as follows. Foroutan et al. used walnut shell (WSA) and WSA/Starch/Fe 3 O 4 for the removal of copper ions from the water. The uptake capacity of the copper ions was attained at 29 and 45.4 mg/g for WSA and WSA/Starch/Fe 3 O 4 , respectively 19 . Ahmadi et al. used Montmorillonite clay (MC) and MC/starch/CoFe 2 O 4 for the removal of methylene blue (MB) and methyl violet (MV) from the wastewater. The uptake capacity of MV using MC and MC/starch/CoFe 2 O 4 nanocomposite were 29.76 and 43.95 mg/g, respectively, and the uptake sorption capacity of MB using these sorbents were 31.96 and 47.51 mg/g, respectively 20 . Foroutan et al. studied the performance of a chicken beak containing hydroxyapatite (HApB) and modified one with the zeolitic imidazolate framework-8 (ZIF-8) for the removal of nickel ions from the water. The uptake capacity of nickel ions has been obtained at 24.27 and 63.49 mg/g using HApB and HApB/ ZIF-8, respectively 21 . Savari et al. investigated the removal of fluoride from the water using zeolite-zirconium under pulsed and continuous sonication. The uptake capacity of the fluoride has been calculated at 32.98 and 31.73 mg/g at the pulsed and continuous situations, respectively 22 . It was concluded that the modification with the magnetic particles has the most impact on the uptake capacity relative to the other modifiers. The magnetic materials especially magnetic biochars are in the category of carbon which can remove the pollutants by electrostatic attraction between the pollutants and oxygen-containing functional groups. Additionally, the generation of magnetic biochar at a minimal oxygen supply increases the crystalline architecture of this adsorbent owing to its graphitic domains that are much smaller in comparison with the nanocarbon materials. Magnetic materials have some benefits including: (1) They can be synthesized in one step only causes decreasing energy loss, (2) They have a flexible and exclusive property due to their surface area, high adsorption capacity, and high degree of surface reactivity, (3) They are mostly known as the economic and cost-effective adsorbents, and (4) They can reduce the concentration of the toxic gases, metallic ions, and other pollutants dramatically 25 .
The RH and modified RH were exploited as the adsorbents mostly for the removal of different dyes, and heavy metal ions. In addition, it was observed that the RH could be employed for the adsorption of some drugs such as paracetamol and tetracycline ( Table 1). The use of RH and modified RH for the adsorption of AA has not been observed until now. Homagai et al. used modified RH for the removal of the crystal violet (CV) dye. The modification agent was xanthate. The highest adsorption capacity has been determined at 90.02 mg/g at a pH of 10 and a contact time of 60 min 26 . Bansal et al. used raw RH for the elimination of Cd(II) ions. The adsorption capacity was obtained at 3.83 mg/g at a pH of 2, and a time of 180 min 27 . Chowdhury et al. adsorbed malachite green using NaOH-rice husk. In their work, the uptake capacity has been obtained at 10 mg/g 28 . El-Shafey used sulfuric acid-rice husk for the removal of Se(IV) ions with an uptake capacity of 12 mg/g 29 . Hsu et al. removed the methacrylic acid using carboxyl-rice husk with an uptake capacity of 317.70 mg/g after 45 min 30 . Chen et al. could adsorb tetracycline after 40 min with an uptake capacity of 8.37 mg/g 31 . The other studies were brought in Table 1.
The objective of the present work was to exploit magnetic RH for trapping vitamin C from an aqueous solution. One of the strengths of this manuscript is the adsorption of vitamin C from an aqueous solution for the first time. The novelty of this work is using agricultural waste as a cost-effective and environmentally friendly adsorbent for reducing the AA from the pharmaceutical and food industries. It will help the industries for adsorbing the AA from the effluents and afterward leaching the AA from this adsorbent and using AA again in the production cycle. The other novelty is detecting the optimum point of the effective factors for increasing the yield using design-expert software. For this aim, various effective parameters including AA concentration, the dosage of the adsorbent, and contact time have been evaluated. Besides, the optimum operation conditions are

Preparation and standardization of iodine solution.
To prepare the standard iodine solution, 0.15 g of potassium iodate (KIO 3 ) was dissolved in a certain amount of deionized water. Then, 3 g of potassium iodide (KI) has been added to the mixture and its volume has been increased to 100 mL. The vessel containing the materials has been mixed using a stirrer to dissolve completely. Afterward, 1 mL of H 2 SO 4 with a concentration of 9 M was gradually introduced into the solution for accelerating the reaction until the appearance of the mixture altered from milky to dark red. Regarding the following reaction, 0.021 M of iodine solution was produced which was made fresh for daily consumption in order to obtain the desired result of the titration process. The reaction of AA with the iodine ions was displayed in Fig. 2 Adsorption. The adsorption experiments have been accomplished in the beaker containing 50 mL of AA solution and modified RH. All experiments were performed at room temperature and stirred at 450 rpm, The adsorbent dosage, AA concentrations, temperatures, and contact time were selected in the domain of 0.5-1 g, 400-800 mg/L, 288-308 K, and 10-120 min for finding the best performance conditions of the resin, respectively. After the adsorption process, the resulting suspension was placed in a centrifuge at 3000 rpm for 10 min. The content of the adsorbed AA has been measured using titration with iodine solution in the presence of starch www.nature.com/scientificreports/ as an indicator. The adsorption process has been repeated in three epochs to reduce the error of the experiment and their mean values were reported. The uptake capacity at the saturation state (qe) and removal efficiency of AA were calculated by the following formulas: (2) q e = (C 0 − C e )V W  Y illustrates the percentage of AA adsorption (predicted response); X c0 , X t , and X D are autocratic coded factors in the experiment. Furthermore, A 0 , A i , A ii , and A ij refer to the coefficients of the linear, quadratic, and interaction between them, respectively, and are random inaccuracy. The significance of the expanded regression models was estimated through the analysis of variance (ANOVA). The design of the experiment for the present adsorption process has been examined using the central composite design (CCD) method under RSM to determine the optimal conditions. According to Table 3, we define three symbols in the design-expert software such as A, B, and C, which refer to the acid concentration, adsorbent dosage, and contact time, respectively. Also, the value of α was defined at a number of 2 in this software, thus, the central point of the acid concentration, adsorbent dosage, and contact time was calculated at 600, 0.75, and 70, respectively. As can be seen, the number of the levels for these three parameters is 5. The ranges of the experimental parameters were selected with respect to the Refs. [43][44][45] .

Results and discussion
Characterization analysis. Figure 3 displays the FTIR of the adsorbents for determining the wavelength of the functional moieties in the adsorbent before and after AA adsorption. The peak in the wavelength of 3427 cm −1 was ascribed to the water molecules and the tensile vibrations of the hydroxyl group (-OH) which was due to the intermolecular hydrogen bonding of some reagents like alcohols, phenols, and carboxylic acids in the lignin and cellulose structure. Hydroxyl moieties are on the adsorbent plane. The peak in the wavelength of 2929 cm −1 is attributed to the symmetric and asymmetric tensile vibrations (C-H) of aliphatic acids in cellulose (which is present in the RH structure), which indicates the presence of an alkane functional group (-CH 3 or -CH 2 ) 46 . The peaks in the wavelength of 11,646 cm −1 correspond to asymmetric mobility of (C=O) (carbonyl groups (ketones and aldehydes), carboxylic acid or ester, as well as to tensile vibrations (CN) and flexural vibrations (NH) related to proteins 47 . The peak is also noted at the wavelengths of 1075 cm −1 , and 1646 cm −1 , which are attributed to the Si-O-Si tensile vibrations and the asymmetric tensile vibrations of C=O, respectively. Fe-O . Comparison of Fourier transform infrared spectroscopy before and after adsorption of AA shows that the peaks in the wavelength range of 3427, 2924, and 1646 cm −1 are ascribed to the hydroxyl functional moiety, symmetric and asymmetric tensile vibrations of C-H, and asymmetric (C=O), respectively. The peak is weakened in the wavelength of 1075 cm −1 which is attributed to the Si-O-Si tensile vibrations, which indicates the involvement of these factor groups during the adsorption process. In addition, the peak of Fe-O is amplified after the adsorption process, which can be analyzed in such a way that magnetic iron nanoparticles have participated in the adsorption procedure 19,20,52 . SEM test has been performed for studying the morphology of the RH surface before and after AA uptake. The SEM micrograph of the RH shows that the surface has a smooth surface (Fig. 4a) and after the modification, many pores have been dispersed through the RH and its porosity has increased (Fig. 4b). Figure 4c and d also showed the structure of the RH and magnetic RH after the adsorption of AA, respectively. Because of the accumulation of AA molecules, the adsorbent surface became smoother and had fewer depressions, which generally indicates the successful adsorption of AA on the adsorbent surface. The EDX patterns for the detailed study of compounds present in the RH and magnetic RH are displayed in Fig. 4e and f, respectively. This analysis confirms the presence of the iron particles after the modification. The EDX test also implied that the RH contains very small amounts of minerals and carbohydrates such as cellulose and lignin in its structure. Thus, the main elements in the structure of the RH are oxygen, carbon, and a small amount of iron 53 . Also, Fig. 4f is related to magnetic RH, which, as can be seen from the composition of the percentage, after placing the iron nanoparticles on the RH, the percentage of iron's composition has increased significantly and their carbon and oxygen have diminished, because the nanoparticles are well deposited on the surface of carbon, which is the main constituent of bran 54 . The main constituents of RH are about 70% of the inorganic part of rice bran, and it is mainly composed of silica and small amounts of alkali metal oxide. RH is composed mainly of lignin (20-30%), holo-cellulose (55-65%), SiO 2 (15-20%) and extracts (2-5%), which can be regarded as a natural organic-inorganic composite ( Table 4).
The XPS (X-ray photoelectron spectroscopy) images are illustrated in Fig. 4. The model of the XPS apparatus is SPECS FlexMod. Figure 4g and h show the XPS image of the raw and magnetic RH. In addition, the XPS image after the adsorption is depicted in Fig. 4i. The peaks in Fig. 4g are 285, 533, 400, and 103.50 eV, which relate to the C1s, O1s, N1s, and Si2p, respectively. The peak at the binding energy of 750 eV in Fig. 4h appears which corresponds to the Fe2p because of the presence of the iron elements in the magnetic RH. Figure 4h implies that the magnetization has occurred successfully. According to Fig. 4i, the binding energy of peaks for C1s, O1s, Si2p, and Fe2p has declined which is owing to the chelation of AA to these atoms, and it reveals the successful adsorption of AA onto the magnetic RH. The binding energy of C, O, Si, and Fe after the adsorption of AA are 250, 500, 400, and 95 eV, respectively 52 .
The X-ray diffraction patterns (Card No.   www.nature.com/scientificreports/ One of the disadvantages of traditional adsorbents is their separation from the solution, which prevents the adsorbent from reusing. The addition of iron not only increases the adsorption capacity but it can also be conveniently isolated from the adsorbent via employing a magnetic field. Figure 6 shows the magnetic properties of raw and modified RH with a vibrating sample magnetometer (VSM) system. The magnetic field was adjusted in the domain of − 15,000 to + 15,000 Oe at room temperature. Saturation magnetization values for the raw and magnetic RH have been determined at 8 and 57 emu/g, respectively. Regarding Fig. 6, the magnetic property www.nature.com/scientificreports/ of the raw RH is less than that of the magnetic one, which means that the Fe 3 O 4 particles have occupied the structure of rice successfully. The BET analysis for the RH and modified RH was conducted. The surface area of the RH and modified RH was determined at 98.17 and 120.23 m 2 /g, respectively. Besides, the pore volumes of the RH and modified RH have been obtained at 2.10 and 6.20 mm 3 /g, respectively. It reveals that the modification of RH increases the surface area, which causes improving the adsorption of AA onto the RH due to the increment of pores on the RH. Besides, the TGA analysis of the modified rice husk was performed and it is illustrated in Fig. 7. Regarding this figure, the thermal stability of the modified rice husk was 450℃, which demonstrates the high resistance of the modified rice husk to the temperature elevation.
RSM results. Variance analysis for the adsorption efficiency of AA is conducted by considering the temperature, acid concentration, and dosage in Table 6. In fact, the existence of P-value, F-value, and lack of fit can help us to decide about the significance or insignificance of these parameters. The other case that was brought in this section is predicting an equation for the adsorption process. The equation in terms of actual factors can be used to make predictions about the response for given levels of each factor, thus, defining an appropriate model or Other things 0.28 Figure 5. The XRD diagram of RH and magnetic RH. www.nature.com/scientificreports/ equation for the process can help us to describe the behavior of the system better. According to Table 5, different polynomials such as linear, quadratic, and cubic were fitted versus the experimental data in the design-expert software for selecting the closest model to the empirical results. This table reveals that the quadratic polynomial was suggested, thus, the experimental parameters have the degree of two. Equation (6) was also predicted by the software, which explains that there is an optimum point at the center of the quadratic graph instead of the corner points. Besides, this equation should not be used to determine the relative impact of each factor because the coefficients are scaled to accommodate the units of each factor and the intercept is not at the center of the design space. Regarding Table 6, the Predicted R 2 of 0.9488 is in reasonable agreement with the Adjusted R 2 of 0.9860. Adeq Precision measures the signal-to-noise ratio. A ratio greater than 4 is desirable. The calculated Adeq Precision is 47.553 which indicates an adequate signal. This model can be used to navigate the design space. Regarding Table 7, the P-value of the model has attained lesser than 0.0001 and demonstrated that the models are significant. The F-value for the adsorption percentage is measured at 102.15, which proved the models are significant. There is only a 0.01% possibility that F-values these large could occur because of the noise 57 . Regarding Table 6, each model hasn't lack fit, which proved that the model is appropriate. Regarding Table 6, the P-values of acid concentration, adsorbent dosage, and time have been obtained lesser than 0.0001, which shows the effectiveness of these factors. Table 6 also proves that the interaction between the acid concentration with time, acid concentration with adsorbent dosage, and adsorbent dosage with time are effective on the adsorption efficiency because their P-values are attained at 0.0075, 0.0048, and 0.0064 which are lesser than 0.05 58 . The model F-value for response is determined 102.15, which indicated the model is significant. There is only 0.01% probability that F-value this large could happen own to noise. In addition, the degree two of the time is effective because its value is 0.0024. According to the RSM results, the optimum experimental conditions have been predicted at the acid concentration of 486.929 ppm, adsorbent dosage of 0.875 g, and time of 105.397 min. In addition, the adsorption efficiency has been obtained at 92.936% at these optimum conditions with the desirability value of 1.00 which is the best value. It is essential to refer to this notice that the values of the acid concentration, adsorbent dosage, and contact time were selected "in range" and the value of response or % R was selected "maximize" in the designexpert software, because our goal was maximizing the adsorption efficiency with respect to the range values of the experimental conditions (Table 8). In Eq. (6), the symbols A, B, and C are ascorbic acid concentration,   In this way, we selected the investigated parameters which include the initial concentration, adsorbent dose and contact time by selecting the in range option in the desired range. The purpose of this work is to show values of parameters where R has its maximum value. For this reason, we choose the highest option for R, which is the absorption percentage value, so that the software is able to protect the optimal R value. As shown in the table, in the initial concentration = 718.92 ppm, adsorbent dose = 0.898 mg and contact time = 105 min, we reach the maximum value of absorption percentage which is 92.33%. Also, in the initial concentration = 481 ppm, adsorbent dose = 0.601 mg and contact time = 34 min, we reach the lowest absorption percentage value, which is 25%.
The 3D diagrams induce important information about the impact of three factors on the adsorption efficiency (response) simultaneously by analyzing the effect of interaction between two variables by setting a factor at its central level, which is displayed in Fig. 8. The observations revealed that the interaction of acid concentration and adsorbent dosage on the response was weak (Fig. 8a). As shown in Fig. 8a, the adsorption efficiency dropped mildly with increasing acid concentration, because the number of the penetrated molecules into the active centers increases and can block these centers and there is not an opportunity for the other molecule for filing the vacant cavities. Increasing the adsorbent dosage has a positive impact on the adsorption efficiency, because the number of active sites has been developed and porosity is increased (Fig. 8b,c). Figure 8b and c reveal that the adsorption occurs rapidly at the initial times, because of the presence of more unsaturated active centers at the primary times, and these sites are occupied by the AA with passing time which enhances the mass transfer resistance and the diffusion of the AA molecules is prohibited 59,60 . The normal probability graph can be utilized to control and survey the normality of the empirical data. The proximity of the squares to the direct line demonstrates the normal dispersion of the error with a mean of zero and a fixed value. According to Fig. 9a, the results of the adsorption efficiency are in the vicinity of the direct line, which reveals the normality of the obtained data of the response. The squares with various colors and straight routes are the experimental and anticipated values, respectively. The more focused the squares are on the straight route proves the better the data distribution. Regarding Fig. 9b, the adsorption percentage of AA placed close to the route displayed that the actual results obeyed a specified function, the data distribution trend was desired and the confidence in the obtained data is high 61 . The perturbation diagram assists to contrast the influence of all factors at a determined place in the design space. The responses are drawn by varying only one parameter over its limited area while keeping the other parameters unchanged. By default, Design-Expert imagines the criteria placed at the midpoint (coded 0) of the entire parameters. Figure 10 indicated that the change in the adsorption efficiency with each of the effective factors is not a direct route, which proves that the adsorption efficiency of AA is susceptible to variations of    www.nature.com/scientificreports/ entire influential parameters in the adsorption process. Besides, each of the factors has a dissociated point where the response is enlarged. Also, it was derived from Fig. 10, increasing the adsorbent dosage and temperature and decreasing the acid concentrations from the optimized value has a suitable influence on the adsorption process and ultimately the economy of the process 61 . The Box-Cox diagram set the best lambda value for the adsorption process at 2.13, indicating that the data did not need to be modified for further improvement (Fig. 11b). The Pareto diagram is another tool to investigate the significance of independent variables on a response (Fig. 11a). The pareto graph shows how well each variable satisfied the criteria: values near one are good. It is observed that the response is one and demonstrating that the aim of maximizing the adsorption percentage is occurred appropriately 20,62-64 . Isotherm modeling. Isotherm study provides important information for scaling up and designing the adsorption process 65,66 . Adsorption equilibrium studies provide comprehensive information to describe the adsorption of solutes between the sorbent surface and solution, especially at the equilibrium state. Figure 12 reveals the diagram of uptake capacity against the acid concentration. Regarding this figure, the uptake capacity increases by acid concentration in temperatures of 15, 25, and 35 °C. Table 9 shows the factors of the isotherm models for the adsorption of AA on the modified KOH rice husk (M-KOH-RH) at temperatures of 15, 25, and 35℃. In the present work, the empirical results have been fitted with the four isotherm models such as Langmuir, Freundlich, Temkin, and Dubinin-Radushkevich (D-R) 67 . The equations of four isotherm models are presented below: The constants in the Langmuir, Freundlich, and Tamkin relations are K L , K F , and β, respectively. In these equations, q e (mg/g), q m (mg/g), and C 0 (mg/L) refer to the saturation uptake capability, maximum adsorbent capacity, and initial adsorbent content. Also, ε is the polarization potential of the D-R equation, the constant K refers to the average free adsorption energy, and the q D (mg/g) refers to the theoretical saturation absorption capability. n is also the heterogeneity coefficient of the Freundlich model. The n parameter for the AA adsorption process using modified rice husk is 10.384 which shows that the adsorption process by these adsorbents is physical and is done easily 65 . A T (L/min) is an equilibrium connection constant of the Temkin model. Table 9 shows the different parameters of the adsorption isotherm with the linear fitting of four models at temperatures   4)] indicates the desirability (0 < R L < 1) of the adsorption process of AA by the adsorbent. In Freundlich's isothermal model, K f increases with increasing temperature. Also, the amount of n higher than unity indicates that the M-KOH-RH adsorbent can be used for all AA concentration ranges. In the Temkin isotherm model, the values of the β parameter also increase with increasing temperature, which means more heat is absorbed. Also, the A T parameter has the highest value in the isothermal isotherm model for the adsorbent at 35 °C, which indicates the adsorbent-adsorption interaction for the adsorbent at this temperature is larger than at the other temperatures. Examining the values of parameter which indicates the average free adsorption energy and the adsorption mechanism in terms of being physical or chemical. If the value of mean free energy is between 8 and 16 kJ/mol, the adsorption process is ion change type. If its value is less than 8 kJ/mol, it shows in adsorption that the mechanism of the adsorption process is physical 65 . In the D-R model, it is obvious that the physical adsorption mechanism (E > 8) is the dominant mechanism for adsorption 68 .
Kinetic modeling. In order to imply the physical or chemical mechanism between the resin and solute, the kinetic study of the adsorption is a useful and efficient method. Figure 13 shows the affinity of the adsorption percentage and uptake capacity to the time. According to this figure, the adsorption percentage and the adsorption capacity increase by time enhancement in both raw rice husk and the modified one. The equilibrium time for the rice husk and the modified rice husk has been observed at 120 and 30 min, respectively, which is owing to the penetration of AA between the layers of the adsorbents and AA deposited on the adsorbent surface. In the adsorption process, the active sites of the adsorbent become gradually altered and resulting in reduced adsorption rates 69,70 . It was concluded the modified rice husk could adsorb the AA rapidly. Adsorption kinetic data were obtained at the AA concentration of 481 ppm, and adsorbent dose of 0.898 mg using well-known kinetic models such as pseudo-first-order, pseudo-second-order, Elovich, and diffusion Intra-particle 71 . The kinetic equations are written below: (11) The equation of pseudo -first -order model : ln q e − q t = ln q e − k 1 t  www.nature.com/scientificreports/ In the above equations, q e (mg/g) , and q t (mg/g) indicate the uptake capability at the saturation state, and at the time of t, respectively. The constants of the pseudo-first-order and pseudo-second-order equations were denoted by k 1 (1/min) and k 2 (g/mg min(. In Eqs. (13) and (14), α and β are the constants of the Elovich model, which are called the adsorption rate (mg/g.min) and the desorption constant (g/mg), respectively. K p (mg (g min 0.5 ) −1 ) and C(mg/g) also describe the intra-particle diffusion rate and the boundary layer constant, respectively. Table 10 shows the various parameters calculated by fitting four kinetic equations. According to Table 10, the pseudo-second-order equation has a more correlation relative to the other kinetic equations, which means that the chemical behavior is dominant in the separation procedure in which the electron sharing occurred between the composite and the solutes and also the intensity of separation of molecules on the adsorbent surface is linear and a function of the number of active sites. The intra-particle diffusion of adsorption took place in two stages, the first stage has a higher R 2 and K p than the second stage, which indicates that the dominant process is the mobility of AA from a liquid phase to the surface of the resin 72,73 . Adsorption thermodynamic. Thermodynamics of the separation process has been studied at various heat degrees to detect the impact of temperature on the separation procedure. The initial solution temperature provides the energy needed for the system and affects the system's adsorption capacity. In order to evaluate the behavior of each process in terms of feasibility and spontaneity, its energy functions and entropy should be considered 19,21,74 . Thermodynamic parameters like Gibbs free energy change (∆G°), enthalpy change (∆H°), and entropy change (∆S°) contribute to interpreting the adsorption performance under several heat degrees. For this aim, the quantities of ∆G°, ∆H°, and ∆S° at 293, 303, and 313 K were obtained as follows and their values have been calculated with respect to Fig. 14: Where K C is the saturation constant, R is the universal gas constant (8.314 kJ/mol K) and T is the absolute temperature (K). The thermodynamic factors of the adsorption have been listed in Table 11. This means that the process of adsorption of AA on the surface of the magnetic sorbent is endothermic, because the sign of ΔH° is positive. Positive values for ΔS o indicate an increase of irregularity in the boundary layer between the solid phase and the liquid phase. Also, a low enthalpy value indicates weak interactions between AA and functional groups (12) The equation of pseudo -second -order model : ln q e − q t = ln q e − k 1 t Adsorption mechanism. The mechanism for the adsorption of AA on the plane of the treated resin has been drawn in Fig. 15. With respect to this Scheme, electrostatic interactions and functional groups interfered with the adsorption mechanism of AA on the modified RH. The electrostatic interaction exists between the divalent iron ions and C=O in the aromatic ring of the AA and silicate ions of the adsorbent with the hydroxyl groups of the aromatic ring of the AA. In addition, the bond formation between the hydroxyl ions of the modified RH and the C-H of the AA backbone is the participation of the functional moieties in the separation mechanism.   www.nature.com/scientificreports/ Moreover, the intra-particle diffusion equation in the kinetic study proves that the mass transfer mechanism of the AA adsorption divides into two parts including (a) Transportation of the AA particles from the AA solution onto the plane of the modified RH and (b) Penetration of AA molecules inside the channels of the modified RH. This kinetic model also explains that the first step or the first case is rapid. In addition, the separation mechanism of AA particles onto the plane of the resin is associated by releasing heat.
Comparison. The comparison between the present work with the other works was collected in Table 12.
This Table demonstrates that rice husk and modified rice husk were employed for the adsorption of different toxic components such as dyes, phenol, and metallic ions. According to this Table, both rice husk and modified rice husk have been used more for the adsorption of different dyes, and it was observed no papers about the adsorption of ascorbic acid onto the modified rice husk. Among the examined dyes, methylene blue could be adsorbed onto the modified rice husk (alginate/rice husk) with a higher uptake capacity (274.90 mg/g) relative Figure 15. Schematic for the adsorption mechanism. www.nature.com/scientificreports/ to the other dyes. Rice husk could also adsorb the synthetic dye with a considerable adsorption percentage which was 99.50%. Besides its good uptake capacity in the adsorption of methylene blue, modified rice husk (KOH/rice husk) could likewise adsorb phenol with a high uptake capacity which was attained at 215.27 mg/g. The ability of adsorption for modified rice husk was relatively low for metallic ions like Mn(II) and Co(II) ions. Since the modified rice husk has exhibited good results in the elimination of dyes, phenol, and metallic ions, the authors decided to evaluate the potential of the modified rice husk (KOH/rice husk) for adsorption of ascorbic acid. Regarding this table, the modified rice husk has shown a dramatic adsorption percentage (92.936%) for ascorbic acid, therefore, KOH/rice husk can be a suitable natural adsorbent for adsorbing the ascorbic acid.
Reusability. In order to save adsorbents, it is essential to survey the reusability of modified RH during adsorption. The reusability of the sorbents is an important factor in industrial applications. In our research, 0.1 mol/L HCl has been selected for leaching the AA from the modified RH 68 . In fact, the modified RH with the adsorbed AA molecules was soaked in the vessel containing 0.1 mol/L HCl for 30 min, and then the adsorbent was rinsed with the deionized water 3 times. Subsequently, it was dried in the oven. After this mentioned process, the modified RH was used again for trapping the AA molecules. The results of the desorption prove that the modified RH is a suitable adsorbent for leaching the AA molecules because the adsorption efficiency at the equilibrium point in the initial and the terminal cycle has been detected at 92 mg/g and 89.5 mg/g, respectively (Fig. 16). In other words, the adsorption efficiency of the modified RH decreased % 2.71 only. Hence, the modified RH is advised to use more than five times without a dramatic reduction in the adsorption efficiency, implying that the modified RH is completely reused frequently.

Effect of cation and anion.
In this section, we want to know how the anion and cation alter the adsorption of AA. The results are exhibited in Fig. 17. The cation and anion in this research are Ca 2+ and Cl − With respect to Fig. 17, the presence of an anion has a slight impact on the adsorption capacity, and the presence of a cation has almost a considerable effect on the adsorption capacity. The adsorption capacity of AA on the magnetic RH without the presence of cation and anion has been obtained at 92 mg/g. The adsorption capacity of AA on the magnetic RH in the presence of cation and anion has been obtained at 84.3 and 90.5 mg/g, respectively. Since the sorbent has a negative charge and anions have also a negative charge. Therefore, the repulsion electrostatic force is created between the anions and the surface of the adsorbent which has a little impact on the adsorption of AA onto the adsorbent. In addition, when the cations are added to the AA solution, they compete with the AA par-

Conclusion
In this current research, the RH was modified by adding Fe 2 O 3 to its network for the adsorption of AA from the aqueous solution for the first time. The characterization tests, isotherm, kinetic, thermodynamic modeling, and reusability test have occurred for studying the adsorption potential of the modified RH. Besides the experimental results, the design-expert software version 11 was considered for detecting the optimum experimental conditions in which the adsorption efficiency would be maximized at those conditions. These results have been derived which were written below: 1. The optimum conditions of the design-expert software have been obtained at 486.929 ppm of acid concentration, 0.875 g of the adsorbent dosage, and 105.397 min of the contact time, and the adsorption efficiency in these conditions was determined at 92.936%. 2. The reusability test exhibits that the adsorption efficiency at the equilibrium point in the initial and the terminal cycle has been detected at 92 mg/g and 78 mg/g, respectively, which showed the modified RH can be used more than five times without dramatic reduction in the adsorption efficiency. 3. The surface area of the RH and modified RH was determined at 98.17 and 120.23 m 2 /g, respectively. 4. The thermal stability of the modified rice husk was 450 °C. 5. The Langmuir model had the highest R 2 of 0.9982, 0.9996, and 0.9985 at the temperature of 15, 25, and 35 °C, respectively, and also the q max values in these temperatures have been calculated at 19.157, 31.34, and 38.75 mg/g, respectively. 6. The pseudo-second-order kinetic model had the best agreement with the experimental results. In this kinetic model, the values of q have been measured at 36.496, 45.248, and 49.019 mg/g at the acid concentration of 418, 600, and 718 ppm, respectively. 7. The values of ΔH° and ΔS° were measured at 31.972 kJ/mol and 120.253 kJ/mol K, respectively, which proves the endothermic and irregular nature of the adsorption of AA. 8. Electrostatic interactions and functional groups interfered in the adsorption mechanism of AA on the modified RH.

Data availability
The datasets used and/or analyzed during the current study available from the corresponding author on reasonable request.