Water decontamination in terms of Hg(II) over thiol immobilized magnesium ferrite: Gum Arabic biosorbent—response surface optimization, kinetic, isotherm and comparing study

Mercury thrown even at low concentrations in water resources causes a severe threat to the ecosystem hence the presented work describes a low-cost and environmentally friendly clean-up platform for decontamination of water in terms of mercury ions. For this purpose, MgFe2O4—Gum Arabic composite was prepared with a simple ultrasound-assisted precipitation route and further modified with l-Cysteine as a thiol resource to improve its adsorption characteristic. The structure of the biosorbent was characterized by XRD, FESEM, VSM, EDX, and BET techniques. Effective parameters on mercury adsorption were optimized with response surface methodology using Box–Behnken design. The maximum removal efficiency was obtained at pH of 5, contact time of 15 min and adsorbent dosage of 19 mg by magnetic GA. After thiol functionalization, optimum variables changed to pH of 2, time of 15 min and dosage of 5 mg. Isotherm study indicated that mercury biosorption onto the magnetic Gum Arabic and thiol immobilized sorbent followed Langmuir and Freundlich model with adsorption capacity of 96 mg g−1 and 250 mg g−1, respectively. Results of the kinetic study revealed that mercury adsorption followed a pseudo-second order model. To study the ability of biosorbent as a reusable compound a mixture of HCl (0.5 M) and thiourea (2%) was employed to release the adsorbed ions from the sorbent surface moreover, it showed 90% removal efficiency after three cycles of sorption and desorption which confirmed the presented composite is a reusable biosorbent.


Introduction
Contamination of surface and groundwater with heavy metals is a special concern in terms of environmental protection. Among heavy metals mercury is a priority toxic compound as the European Union (EU) legislation appointed the allowable level of 0.07 l μg/L in surface water [1,2]. Mercury is a highly reactive and volatile metal that is relatively soluble in living tissues and water resources hence it can cause damage to the kidney and nervous system as well as birth and chromosome defects [3,4]. Mercury thrown in the environment has increased dramatically in recent decades owing to extensive development in textile, fertilizer, leather and military industries [5][6][7]. The hazardous effects of mercury as well as its high tendency for accumulation onto living ecosystems and food chains persuaded researchers to find effective methods for remediation of mercury contaminants [8,9]. Up to now several technologies such as coagulation, precipitation, ion exchange, and adsorption have been developed for mercury removal [10]. Because of the economical aspects and environmental benign; adsorption is a more suitable process for water treatment compared with the mentioned techniques [11,12]. Numerous adsorbents such as carbon-based materials [13], chelating resins [14], chitosan [15,16], graphene oxide [17] and Moroccan stevensite [18] have been applied in the adsorption process. Despite the worthiness of the most mentioned adsorbents in the literature some of them show weaknesses like low capacity and high equilibrium time causes low desire to choose them for water treatment hence it is quite necessary to develop novel adsorption based plane for water treatment [19]. In recent decades, a growing attention was paid to the use of nanostructures for clean water supply. This is because of the unique properties of nanosorbents such as high surface area and great active sites which made the adsorption process fast, simple and more efficient [20][21][22][23].
Biosorption, an adsorption process that uses materials with natural resources, attracted great attention for water decontamination since it uses inexpensive and environmentally friendly materials. The most employed biosorbents for metal adsorption include rice husk, chitosan, wool, cellulose, Jujuba seeds and so on [24][25][26]. However, they suffer from some drawbacks such as water solubility or low dispersibility in water that limits their technological applications leading to unsatisfactory adsorption behavior [27].
Based on the above-described merits this work was developed to present an efficient green route for water decontamination in terms of mercury ions. To reach this goal a novel biosorption system has been selected and developed. Gum Arabic (GA), a natural negatively charged polysaccharide, was selected as the main fragment of the biosorption system owing to its safety, availability and biocompatibility [28]. In other words, GA is easily available for the preparation of designed materials; however, it suffers from solubility in water and low adsorption efficiency for metal ions. Chemically modification is an efficient strategy to overcome the mentioned drawback to enhance its adsorption properties toward metal ions [29][30][31]. Hence l-Cysteine was used as a low toxic chelating agent. The selection of cysteine was based on the fact that it has tremendous potential for metal adsorption due to the presence of amine, thiol and hydroxy groups. Based on hard-soft acid-based theory mercury and sulfur are classified as soft groups that effectively react with each other. [32,33]. Epichlorohydrin (ECH) was used as a bridge to attach the l-Cysteine onto the GA structure. The performance of the biosorption system can be improved by preparing a magnetic composite of them. The prepared composite combines outstanding traits of two components through increases in surface area and active sites. At the first step of the magnetic composite preparation, the employed metal ions (Mg and Fe) react with the GA functional groups hence the growth of the magnetic nanoparticles in GA matrices increases their stiffness and decreases its solubility in water [34]. The magnetic property of composite improves the adsorption properties of GA since it exhibited easy separation and large flexibility in further surface modification. One of the suitable candidates in magnetic GA preparation is MgFe 2 O 4 as a subset of spinel-ceramic ferrites with high corrosive stability, biocompatibility and eco-friendly due to the presence of Mg 2+ instead of heavy metals [35]. In brief, this research focuses on efficient mercury removal hence l-Cysteine anchored magnetic MgFe 2 O 4 -GA based biosorbent has been prepared and used. In other words, a combination of LC with magnetic GA generates an interesting advanced material with improved adsorption properties relative to the unmodified magnetic GA. Moreover, effective parameters on mercury adsorption i.e., time, pH and adsorbent dosage were optimized by response surface methodology (RSM). In comparing one factor at a time this technique is simpler and faster [36] since RSM varies several variables simultaneously with a limited number of experiments. Isotherm and kinetic models for mercury adsorption were also studied and the results were compared with the literature.

Materials and instruments
Distilled water was used to prepare all stock solutions and required aqueous media for the synthetic process. GA was supplied from a local market in Tehran, Iran. All chemical reagents were analytical grade with a purity of more than 98%. Ferric nitrate (Fe(NO 3 ) 3 ·9H 2 O), magnesium nitrate (Mg(NO 3 ) 2 ), epichlorohydrin (ECH), sodium hydroxide (NaOH) and l-Cysteine were supplied from Merck (Darmstadt, Germany) and used to prepare magnetic modified GA. To prepare a stock solution of Hg 2+ with a concentration of 1000 mg L −1 ; mercury chloride (Merck product) was dissolved in distilled water at a pH of 3 using hydrochloric acid prepared with a dilution of concentrated HCl (32%). Ethanol with a purity of 95% was supplied from Bidestan Company (Qazvin, Iran).
The morphology of the prepared magnetic modified GA was characterized at the accelerating voltage of 15 kV with Field Emission Scanning Electron Microscopy (FE-SEM) using Mira 3 Tescan; the Czech Republic. N 2 adsorption/desorption experiment was performed with the Brunauer, Emmett and Teller (BET) method. The magnetic behavior of the composite was studied by Vibrating Sample

3
Magnetometer (VSM, model MDKFD, Iran). The crystallinity of the composite was studied with Cu-Ka (λ = 1.540589 Ǻ) radiation in the 2θ range of 2°-100° using a powder X-ray diffraction analyzer (Phillips powder diffractometer, X' Pert MPD). The elemental composition of the prepared composite was determined with Energy dispersive X-ray spectrometry (EDX) using the Oxford ED-2000 instrument (England). Transmission electron microscopy (TEM) images were recorded with the Zeiss-EM10C instrument (Germany). The FT-IR spectra were recorded using Equinox 55, Bruker instrument in the wavenumber range of 400-4000 cm −1 using the attenuated total reflection method. The pH adjustments were performed with a digital pH meter (model 692, metrohm, Herisau, Switzerland). Magnetic separation was assisted with an external magnetic field using a neodymium-iron-boron (Nd 2 Fe 12 B) magnet.

Preparation of magnetic GA and modified composite
Magnetic GA was synthesized by a simple ultrasound-assisted coprecipitation method. For this purpose about 2.0 g of GA was dissolved in 100 mL distilled water under magnetic stirring at 50 °C. Then 0.64 g of Mg(NO 3 ) 2 and 2.0 g of Fe(NO 3 ) 3 ·9H 2 O was dissolved in 50 mL water have been dripped into the GA solution at room temperature. To assist in adsorption of the ions onto the GA structure, the mixture was stirred for 10 min then 25 mL of NaOH solution (2 M) was added to GA-metal ions solution and the mixture was sonicated at 60 °C for 1 h. The prepared magnetic GA was collected and washed with distilled water and ethanol then dried at 70 °C for 3 h.
Thiol immobilized magnetic GA was prepared using ECH as a linker through a refluxing route. To activation of hydroxyl functional groups on the magnetic GA surface, 2.0 g of magnetic GA was dispersed in 100 mL of NaOH solution (0.2 M) using an ultrasound wave within 1 h. After cooling the mixture to room temperature, 5 mL of ECH was added and the mixture was stirred at room temperature for 12 h. The prepared chloro magnetic GA was washed once with ethanol and redispersed in the 50 mL of ethanol-distilled water (1:1) solution containing 2.0 g of l-Cysteine. The mixture was refluxed at 50 °C under magnetic stirring for 24 h then collected, washed with distilled water twice and ethanol once then dried at 80 °C for 6 h.

Adsorption experiments
Optimization of effective parameters on the adsorption efficiency was performed by Response surface method (RSM) tests using Box-Behnken Design (BBD). A design expert 7.0 software was employed and a three-parameter including pH, contact time and adsorbent dosage was selected as effective parameters. Optimization experiments were performed at 17 designed runs, 5 center points and one block. Sample solutions were 10 mL with a mercury concentration of 20 mg L −1 . The values of the parameters are shown in Table 1.
Isotherm and kinetic data were extracted at the optimum value of parameters. Mercury concentration was 1-150 mg/L with volume of 10 mL. After equilibrium, Polymer Bulletin (2023) 80:8355-8375 the concentration of the mercury ions in the supernatant is determined by ICP-AES analysis. The removal percentage (%R) is calculated based on the initial (C 0 ) and remaining concentration of mercury after equilibrium (C e ) using the following equation.

Results and discussion
Characterization Elemental analysis of the magnetic GA and final composite was performed with the EDX analysis. According to the results in Fig. 1 magnetic GA is composed of C, O, Fe and Mg which is corresponded to the component of the employed raw material. After preparing LC immobilized magnetic GA two new peaks corresponding to the elements of S and N appeared which proves that the nanocomposite preparation was successful.
To study the crystalline structure of as-synthesized MgFe 2 O 4 -GA and thiol immobilized magnetic GA, the XRD analysis was employed. According to results in Fig   observed results are in good agreement with the reference number 5247-006-98 for magnesium ferrite. The GA shows the main peak at 2θ° equal to 22° which is corresponded to the (110) pseudo-orthorhombic reflections of the polysaccharide [37]. After functionalization with thiol, a sharp peak around 2θ° = 22° appeared which can be assigned to new crystallinity owed to LC. In other words, GA and LC structures contain several numbers of hydroxyl, carboxyl and amine functionality which can interact with each other through strong hydrogen bonding leading to the regularity of the composite structure that appears as of new crystalline structure. Characteristic peaks of the magnetic GA are also observable with high intensity which confirmed the formation of the target composite material. The magnetic behavior of the thiol functionalized magnetic GA in terms of magnetization versus applied field is shown in Fig. 3. The value of magnetic saturation (M s ) for the final composite is 1.8 emu/g which is lower than some reports in the literature [38]. This can be owed to the low contribution of the magnetic nanoparticles in total volume fraction as well as the disaffiliation of the GA as an inert coating layer on the total magnetization. It has been observed that the M r value of the magnetic composite was 0.06 emu/g. It indicated that the prepared magnetic material probably is classified as a superparamagnetic compound since in the absence of an external magnetic field the remanence value was near zero.
The FESEM image of the magnetic GA and thiol immobilized composite is shown in Fig. 4. The image indicated that both materials are in an aggregated form composed of a flake-like structure. The composite structure is composed of GA, LC and magnetic fragments hence the observed structure can be assigned to the intermolecular and intramolecular hydrogen bonds between them. In other words, GA acts as a stabilizer to prevent direct magnetic interaction between the particles composed of hydroxyl functional groups. They can react with each other and with hydroxyl groups of ferrite as well as with amine and carboxyl groups of LC through hydrogen bonding in various directions thus; the nanocomposite is regarded as a combined structure. The TEM image of the modified magnetic GA is illustrated in Fig. 5. It can be seen that the flake-like structures are composed of magnetic nanoparticles as the GA is covered by the nanoparticles and the nanoparticles have been dispersed in the GA matrices. GA acts as a stabilizer for the magnetic nanoparticles which prevents excess growth of them.
To study the pore size diameter and volume the N 2 adsorption-desorption experiment has been performed for magnetic GA and cysteine immobilized composite moreover, the corresponding Barrett, Joyner, and Halenda (BJH) pore size distribution curve is obtained. Results are shown in Fig. 6. As can be seen the curve for magnetic GA belong to a type II isotherm that possesses slit shape pores with parallel walls. The isotherm has a minor hysteresis loop owed to the filling and emptying of the pores by capillary condensation [39]. Table 2 shows the BET characteristics of the materials. As can be seen, the pore diameter of the material is in the range of 11-18 nm and the pore size increased from magnetic GA to final composite. Pore size distribution indicated that the materials belong to mesopores compounds [40]. Moreover, it can be seen that the total pore volume and surface area of the final composite increased concerning magnetic GA as a result of new crystalline formation on the composite structure. In other words, a large number of hydroxyl and amine groups on the GA and cysteine structure linked together through stronger intra and intermolecular hydrogen bonds led to the regularity of the composite structure that   increases the pore volume. Moreover, l-Cysteine immobilization increases the roughness of the surface of magnetic GA causing an increase in its surface area of it. FT-IR spectrums of magnetic GA and modified magnetic GA are shown in Fig. 7 The spectrum of magnetic GA showed stretching vibration at 2500-3500 cm −1 , 2880 cm −1 , 1050 cm −1 and 1374 cm −1 and 1627 cm −1 corresponding to COOH and O-H vibration, stretching of C-H and C-O, the OH bending vibration of primary alcoholic groups and C=O groups, respectively [41]. Octahedral and tetrahedral stretching vibration of Fe-O complexes appeared at 578 cm −1 . Relative to the magnetic GA in the spectrum of modified sorbent the intensity of C-H stretching at 2880 cm −1 increased owing to the increase in the aliphatic C-H density. Moreover, the intensity of the broad peak at 2500-3500 cm −1 is increased as a result of the high density of NH and carboxylic acid groups of the l-Cysteine.

Fitting of process models
A polynomial equation has been extracted using Box-Behnken design and used for the calculation of regression coefficients. The removal percentage (%R) has been linked to decoded variables. Results are depicted in Eqs. 2-3 corresponding to magnetic GA and thiol functionalized composite, respectively. The equations A, B, and C are independent variables including pH, contact time and sorbent dosage, respectively.
The significance of coefficients was verified from F and P values extracted from the analysis of variance (ANOVA). Results are depicted in Tables 3 and 4  probability value (P) were < 0.0035 and < 0.0008 which proves the regressions are statistically significant. Based on the results of p values, the independent variable of A (pH) is significant using thiol-modified magnetic GA. Besides, results showed that the variable of C and quadratic terms of A 2 significantly affect the mercury removal using both magnetic GA and thiol-modified sorbent.
The predicted response was compared to the actual values and the results are shown in Fig. 8a, b. It can be seen that the predicted response (%R) is in good  [42]. The normal distribution of residual extracted from the normal probability plot (NPP) was used to validate the ANOVA results. According to Fig. 9a, b data stay on straight lines with low violations from the assumptions hence emphasizing the normality of the data. Results of the perturbation plots are shown in Fig. 10a, b  . 9 The NNP for mercury removal using magnetic GA (a) and cysteine immobilized magnetic GA (b)

3
The plots imply comparative effects of the parameters on the removal. It can be seen that the curve of A and C has steep curvatures and confirmed that mercury removal is sensitive to pH and adsorbent dosage.
The 3D plots based on pH and dosage are shown in Fig. 11. Results showed that the optimum levels of the variables were pH of 5, time of 15 min, and adsorbent dosage of 19 mg using magnetic GA. The optimum value of parameters using the thiol functionalized sorbent is pH of 2, time of 15 min, and adsorbent dosage of 5 mg. The accuracy of the results was evaluated after performing three experiments under optimum conditions. It was found that there was a good agreement between the calculated responses (83%) and the mean of three experimental responses (85%).

Kinetics study
A kinetic study is useful to estimate the involved adsorption mechanisms by evaluating rate expressions. Moreover, they express sorption rates. In view of practical application, a large-scale adsorption system can be designed based on the results of the kinetic study [43]. Kinetic parameters were extracted by performing several experiments at the optimum condition and time range below 15 min. Mercury concentration was 20 mg/L and sample volumes were 20 mL. Two main kinetic models, i.e., pseudo-first-order and pseudo-second-order were used to evaluate the results. The mentioned kinetic models can be expressed as the following equations: In the above equations, the parameters of k 1 represent the pseudo-first-order model constant (min −1 ) and k 2 imply the pseudo-second-order rate constant (g mg −1 min −1 ). The Q e and Q t express the values of the amount adsorbed per unit mass at equilibrium and at any time t, respectively [44]. Table 5 and Fig. 12 show the results of the studied kinetic models and confirmed that the pseudo-second-order model has better linearity compared to the pseudofirst-order model moreover, obtained Q e based on the pseudo-second-order model show lower deviation from the experimental values; hence, this model can better describe the adsorption kinetic by magnetic GA and thiol modified sorbent.

Adsorption isotherms
Two main adsorption isotherm models, i.e., Langmuir and Freundlich isotherm models are employed to study adsorbate-adsorbent interaction. The models were described using the nonlinear equations with MATLAB R2013a software. Relative to the linear isotherm model in nonlinear equations the error distribution and isotherm parameters are fixed in the same axis [45]. The model was illustrated with the following equations.

3
In the above equations, C e (mg L −1 ) is the equilibrium concentration of the analyte in the liquid phase, Q e (mg g −1 ) is adsorption capacity, K f and n are coefficients of the Freundlich model and b and Q m are Langmuir coefficients [46][47][48][49].
The equilibrium parameter (R L ) was used to express the essential characteristics of the Langmuir equation. The parameter is a dimensionless constant which is calculated using the following equation.
In this equation, C 0 is the initial solute concentration. The value of R L may be equal to zero which indicates the isotherm is irreversible. 0 < R L < 1 indicate that the model is favorable, R L = 1 confirms the model is linear and R L > 1 shows the model is unfavorable [50]. Results are shown in Fig. 13 and Table 6. Based on the R L value it can be seen that mercury adsorption onto both adsorbents is favorable. fitted with the Langmuir model using magnetic GA as adsorbent. However, mercury adsorption onto the thiol functionalized magnetic GA fitted with the Freundlich model. Based on the results maximum adsorption capacity of magnetic GA and composite is 96 mg g −1 and 250 mg g −1 . Results illustrated that mercury adsorption onto both adsorbents is not a net physical or chemical process as mercury adsorption onto the final composite can be classified as a multilayer physical adsorption on the  heterogeneous surface. This type of sorption is include direct adsorption of the first layer on the surface of the sorbent along with precipitation or hydrogen bonding type attachment of other layers on the surface of the first layer [51][52][53].

Adsorption mechanism
Metal ions-solid surface interaction takes place through physical or chemical processes. Based on the results of the kinetics studies, the mercury adsorption onto both sorbents followed pseudo-second order model and hence can be classified as a chemisorption process. Isotherm study showed that adsorption onto magnetic GA followed the monolayer Langmuir model. Besides, the adsorption onto the thiol functionalized magnetic GA followed the multilayer Freundlich model. Therefore, it can be concluded that mercury sorption is along with a physicochemical process. In other words, mercury adsorption followed a complex mechanism as both the chemical and physical adsorptions contribute at the same time to the adsorption process. Mercury interaction with the functional groups of the sorbent can be classified as an electrostatic attraction and complexation reaction (Fig. 14) According to zeta potential measurement (Fig. 15) magnetic GA that contains a hydroxyl (-OH) and carboxyl (-COOH) functional groups has a negative charge at pH higher than 2. It means that the electrostatic attraction occurs between the mercury ion and deprotonated functional groups of magnetic GA as with increasing the pH negative charge of the sorbent is increased which caused higher adsorption efficiency. Mercury adsorption on the surface of thiol functionalized sorbent takes place in an acidic solution this means that the lone pairs of electrons of the S, N and O elements on the l-Cysteine could rapidly complex mercury ions [54]. In fact results of zeta potential for l-Cysteine immobilized magnetic GA showed zero charges around a pH of 3. This result confirmed that active groups of amino and thiol have the main role to capture target ions. In other words, electrostatic interaction has a small share in mercury adsorption onto thiol immobilized magnetic sorbent. It seems that at a pH higher than the pH of zero charges, Hg(II) adsorption should be feasible owing to electrostatic interaction as a result of an increase in the negative charge of the sorbent surface but the efficiency decreases. This is due to the increase in the amount of OH − groups that react with Hg(II) ions to form Hg(OH) 2 and Hg(OH) − 3 , which reduces the adsorption efficiency [55].

Desorption and reusability
Regeneration of the sorbent is a main factor that makes the sorption process more economical hence, different concentrations of HCl-thiourea solution with a concentration of 0.5 M of HCl and 0.5 %, 1 % and 2% thiourea are tested for release of the adsorbed mercury ions. The selection of thiourea was based on the hard-soft acidbase theory since its sulfur group can easily react with the mercury ions. It is found that a mixture of HCl (0.5 M) and 2% thiourea releases adsorbed mercury ions with an efficiency of 95-97%. Moreover, the reusability of the final sorbent was examined with several sorptions-desorption run and it is found that after 3 cycles the removal efficiencies reached from 97 and 95% to 90% which confirms its efficiency as a regenerable adsorbent.

Comparison of the adsorbent's performances
In this research, two adsorbents including magnetic GA and thiol immobilized adsorbent was used for the mercury removal from the aqueous solution. According to the results, the final composite showed better performance relative to magnetic GA. The maximum adsorption capacities are 96 mg g −1 and 250 mg g −1 for Fig. 15 Zeta potential measurement of magnetic GA and l-Cysteine modified magnetic GA the magnetic GA and final composite, respectively. The reason may be related to the fact that mercury is a soft acid species and as a result, they have lower tendencies toward oxy/hydroxy functional groups. Nevertheless, the final composite is composed of sulfur groups that are ready to capture mercury ions through selfassembly routes.
The mercury adsorption efficiency of the presented sorbents has been compared to some reports in the literature and the results are shown in Table 7. The performances of the prepared sorbents are appropriate with the aspect of adsorption time as follows a fast adsorption kinetic. Moreover, they demonstrate satisfactory sorption capacities which are higher than the values for most reported adsorbents. Besides, the employed sorbent is a green adsorbent as its components include environmentally friendly material which doesn't generate secondary byproducts.

Conclusions
In summary, a nanocomposite of ferrite-Gum Arabic and l-Cysteine immobilized composite was synthesized and employed for mercury removal from water solution. Effects of important parameters on removal efficiency were studied with RSM and results showed that solution pH and adsorbent dosage are effective parameters on the sorption efficiency. Adsorption onto the sorbents followed the pseudo-second-order kinetics model. Isotherm's study showed that the adsorption onto magnetic GA followed the Langmuir model; however, the adsorption onto the final composite was fitted with the Freundlich model. Results confirmed that mercury adsorption followed the physicochemical sorption mechanism. The magnetic GA-cysteine nanocomposite could be considered a green adsorbent with a good adsorption capacity of 250 mg g −1 which proved the efficiency of sorbent for decontamination of water in terms of mercury ions.