Fluoride Removal from Aqueous Solutions by Using Super-Adsorbents of Chitosan/Orange Peels/Activated Carbon@MgO: Synthesis, Characterization, and Adsorption Evaluation

Abstract

Exposure to excessive concentrations of fluoride in potable water is harmful to human health; therefore, its limitation is deemed necessary. Among the commonly applied technologies, adsorption is selected, as it is a highly effective, simple, and economically efficient treatment. In the present study, several combinations of chitosan (CS), orange peels (OP), activated carbon (AC), and MgO were synthesized and tested as adsorbents in order to find the most effective derivative for fluoride extraction. The impact of the adsorbent dosage, pH level, contact time, and initial concentration was investigated to assess the feasibility of the chitosan/orange peels/activated carbon@MgO composite. According to the results, the modification of chitosan with AC, OP, and MgO in a unique adsorbent (CS/OP/AC@MgO), especially in acidic conditions (pH 3.0 ± 0.1) by using 1.0 g/L of the adsorbent, demonstrated the highest efficiency in F removal, up to 97%. The pseudo-second (PSO) order model and Langmuir isotherm model fit better to the experimental results, especially for CS/OP/AC@MgO, providing a Q_m = 26.92 mg/g. Thermodynamic analysis confirmed the spontaneous nature of the adsorption process. The structure and morphology of the modified OP/CS@AC-Mg were extensively characterized using BET, XRD, FTIR, and SEM techniques.

1. Introduction

Fluoride is among the essential elements for dental health, but if it exceeds 1.5 mg/L in potable water, it may lead to dental or skeletal fluorosis [1]. According to the World Health Organization (WHO), the allowable limit is 1.5 mg/L, but the recommended range is between 0.5 and 1 mg/L [2,3]. Groundwater with excess fluoride concentrations is found in many mid-latitude regions of the world, i.e., Africa (Uganda and Tanzania); China, India, and Sri Lanka in Asia, several states in the U.S.A; Mexico; and South American countries such as Chile and Argentina [1,4]; as well as in Europe, such as Hungary and Italy [5]. For example, the highest concentration reported was 2.8 g/L in Lake Nakuru in Kenya, while the mean fluoride concentration in Ethiopian Valley groundwater was 6.03 mg/L [6]. In Ghana, however, it may reach up to 4.37 mg/L, with a mean of 1–2.09 mg/L [7,8], while a high concentration larger than 3.5 mg/L was noticed in South Carolina and up to 6.27 mg/L in Texas groundwater [9]. Various industrial processes are responsible for increasing fluoride concentrations in wastewater, as they can be easily released from coal-fired power plants, electroplating and semiconductor production, phosphate fertilizer production, etc. [10].

Given the issues associated with fluoride’s presence in drinking water, it is considered essential to safely reduce its residual concentration. For this reason, according to the literature, various methods have been employed to extract fluoride from water, including (1) coagulation and precipitation [11,12,13], but both produce large amounts of residues; (2) electrodialysis [14], which is efficient and does not produce secondary pollutants, but the high power consumption increases the cost significantly [14,15]; (3) ion exchange, which is unfortunately an expensive process that requires careful maintenance and produces significant waste [16,17]; (4) adsorption [16,18,19,20]; and (5) membrane filtration, which leads to optimal levels of fluoride concentration [8]. Although membranes are generally considered one of the expensive technologies due to high operating and capital costs, especially when used in seawater desalination, in the context of ion adsorption using membrane filters, operational and capital costs are low due to a low energy footprint due to the nanofibrous nature of membrane filters (eventually furnishing very high water permeability) [21]. Adsorption has been considered a superior method since it offers simplicity, environmental safety, high efficiency, and cost-effectiveness [16,17,22], and thus was selected to be used in this study. Adsorbents can be produced from environmentally friendly materials and are highly selective for separation, making them effective in extracting heavy metal ions, anions, and organic dyes from wastewater [5,23,24].

Adsorbents of particular interest include those composed of natural polymers, especially polysaccharides like chitin and its derivative chitosan [5], as well as fruit peels [25]. Chitin, the second most abundant natural biopolymer after cellulose, is the most common aminopolysaccharide. It is sourced from crustacean shells, mollusk skeletons, and certain fungal cell walls [1]. Chitosan is an effective and sustainable adsorbent for removing fluoride anions due to its positively charged amino groups and linear polyamine structure, which allows for cross-linking and modification [1,10]. It is biodegradable, environmentally friendly, and relatively inexpensive, making it suitable for large-scale industrial applications. Thus, in terms of the flexibility of chitosan, it is useful that it can be modified to enhance its adsorption abilities [23]. As chitosan occurs unprocessed mainly in the form of flakes or powder, this is a significant problem in its utility, especially for column applications, as it can swell, have low mechanical strength, or crumble [1]. Therefore, to overcome these drawbacks, researchers have studied the combination of chitosan with the widely used activated carbon, as an adsorbent and substrate, to enhance it [26,27]. The combination of chitosan and activated carbon (AC) is expected to increase the effectiveness of removing fluoride ions and the economic utilization due to adsorbent recycling and regeneration. However, despite the extensive use of AC as an adsorbent, its economic viability is one of its drawbacks. As a result, researchers have investigated alternative adsorbents like agricultural waste, which provide a more affordable and sustainable option for fluoride extraction. Agricultural wastes, including fruit peels [25,28,29], have been successfully used as adsorbents. Orange peels are plentiful, low-cost, and readily accessible, making them an effective option for removing a range of pollutants from aqueous solutions, including dyes, heavy metals, and anions [11,12,30,31]. These peels contain cellulose, lignin, pectin, and chlorophyll pigments [32], which possess functional groups like amide, carboxyl, and hydroxyl groups that contribute to the effective removal of pollutants.

Recent studies have explored the use of magnesium (Mg) combined with various adsorbents or flocculants for fluoride removal. Pure magnesium salts are costly, making their use impractical. Magnesium oxide (MgO) is a more economical alternative with high defluoridation capacity [33], though its dust can impact filtration [34]. To enhance stability and leverage the benefits of each component, combining magnesium with materials like activated carbon or orange peels is suggested. Additionally, chitosan/MgO composites have been recognized as more effective for fluoride removal compared to pure MgO or chitosan alone [34].

Considering all the above, in this study, several combinations of chitosan and/or activated carbon, orange peels, and MgO were synthesized and tested as adsorbents for potential defluoridation. The proposed combinations have never been used before for F-removal, as a survey of the recent literature reveals. In order to find the most effective derivative, the adsorption efficiency of the different adsorbents and the impact of pH, mass, contact time, starting concentration, and temperature levels were investigated. In addition, the data were evaluated and modeled using adsorption kinetics and isotherm equations, as well as thermodynamic principles. Comprehensive characterization of the materials was performed both before and after fluoride ion adsorption on the optimal materials. This allows for the study and description of the structure of the synthesized composites and the mechanisms involved.

2. Materials and Methods

2.1. Materials

A sodium fluoride (NaF) solution from, Sigma-Aldrich, Merck KGaA, Darmstadt Germany was used to prepare simulated fluoridated water, with a 1 g/L stock solution made by dissolving 2.210 g of NaF in 1 L of deionized water. This stock solution was then diluted as needed for the experiments. Chitosan (310–375 kDa) and activated carbon were sourced from Sigma-Aldrich, Merck KGaA, Darmstadt, Germany, while magnesium oxide (MgO, PMS2 pure magnesia 96%, MW = 40.30 g/mol) was obtained from Alfa Aesar GmbH & Co KG, Karlsruhe, Germany. Oranges were purchased from a local market. pH adjustments were performed using diluted solutions of 37% HCl (Panreac, AppliChem, Barcelona, Spain) and ≥97.0% NaOH pellets (Sigma-Aldrich, Merck KGaA, Darmstadt, Germany), with NaOH, were also used in the synthesis of the materials.

2.2. Preparation of CS/OP/AC@MgO Adsorbent

The orange peels (OP) were initially collected before disposal and thoroughly washed with deionized water. Afterward, they were air-dried at room temperature for three days, sieved, and then oven-dried at 100 °C for 120 min. The dried peels were ground into a fine powder (75–125 μm) using a household mixer, sealed in a plastic bag, and stored for use in adsorption experiments and composite synthesis [31,35].

To synthesize the CS/OP/AC@MgO composite, activated carbon (AC) and magnesium oxide (MgO) were mixed with deionized water and sonicated for 1 h. Orange peels (OP) were added while stirring at 50 °C for another hour. Acetic acid and chitosan (CS) were then introduced and stirred for 3 h. Sodium hydroxide was added as a cross-linking agent, and the solution was stirred for 24 h. The mixture was filtered, washed, dried, and ground into powder. Similar methods are used to create other composite materials, as summarized in

Table 1. Composition of synthesized adsorbents.

A/A	Adsorbent	Constituents
1	CS/OP	Chitosan and orange peels (ratio of 1:2)
2	CS/AC	Chitosan and activated carbon
3	CS/MgO	Chitosan and magnesium oxide
4	AC/MgO	Activated carbon and magnesium oxide
5	CS/OP/AC	Chitosan, orange peels, and activated carbon
6	CS/OP@MgO	Chitosan, orange peels, and magnesium oxide
7	CS/AC@MgO	Chitosan, activated carbon, and magnesium oxide
8	CS/OP/AC@MgO	Chitosan, orange peels, activated carbon, and magnesium oxide

.

2.3. Analytical Determinations

In order to determine the residual F⁻ concentration, SPADNS photometric method [36] was used. According to this method, a non-colored complex (ZrF₆²⁻) is forrmed in a coloured solution when fluoride anions and zirconium react. Thus, while the concentration of fluoride increases, the color of the solution becomes lighter. The Zr–SPADNS complex used in the analysis was prepared by mixing two separately prepared solutions:

(1)

A total of 958 mg of SPADNS solution (Sigma-Aldrich, Merck KGaA, Darmstadt Germany) was added to 0.5 L of distilled water.

(2)

A total of 133 mg of 99.99% trace element zirconyl chloride (Sigma-Aldrich, Merck KGaA, Darmstadt Germany) was added to 25 mL of deionized water. Then, 350 mL of HCl was added to make a final 0.5 L diluted solution.

(3)

The Zr–SPADNS solution was created by mixing equal volumes of these two solutions and remains stable for 2 years when stored in the dark.

To measure residual fluoride concentration, 1 mL of the Zr–SPADNS reagent mixture was added to 5 mL of the sample (distilled water was used as the blank) and mixed thoroughly for two minutes. The relative absorbance at 570 nm was then measured using a UV-Vis spectrophotometer (UV/VIS uniSPEC 4, LLG-Labware, Meckenheim, Germany), with the absorbance values compared to a standard fluoride ion curve.

2.4. Adsorption Experiments

To evaluate the adsorptive ability of the materials for fluoride (F⁻) extraction, a series of experiments were conducted. Batch adsorption experiments involved adding adsorbents to 10 mL solutions of F⁻ at room temperature while continuously shaking. For comparison, the impact of contact time (0–24 h), starting F⁻ concentration (2–50 mg/L for isotherm tests and 5 mg/L for other experiments, based on literature [1,7,8,9]), and adsorbent dose (0.3–1.0 g/L) on the adsorption capacity were evaluated. Following adsorption, the water samples were filtered using a 0.45 µm nylon filter, and the filtrate was collected for further detailed analysis. The removal rate (R (%)) of F⁻ was assessed as follows (Equation (1)):

$$R\left(\%\right)=\left({\displaystyle \frac{C_0-C_f}{C_0}}\right)\times 100\mathrm{\%}$$

(1)

where C₀ is the starting F⁻ concentration (mg/L), and C_f is the F⁻ concentration after treatment (mg/L). The capacity (Q_e) (mg/g) was evaluated as follows (Equation (2)):

$$Q_e={\displaystyle \frac{(C_0-C_e)\times V}{m}}$$

(2)

where C_e indicates the F⁻ concentration (mg/L) at equilibrium, V (L) is the volume of solution, and m (g) is the mass of the adsorbent used.

2.4.1. Equilibrium Experiments

A specified quantity of adsorbent was added to 10 mL fluoride (F⁻) solutions with concentrations ranging from 2 to 50 mg/L. The results were analyzed using the Langmuir and Freundlich isotherm models [37,38]. The Langmuir model is described as follows (Equation (3)):

$$Q_e={\displaystyle \frac{Q_m{K}_L{C}_e}{1+K_L{C}_e}}$$

(3)

where Q_e represents the concentration of the adsorbate in the solid phase relative to the liquid phase at equilibrium (mg/g), Q_m indicates the maximum capacity (mg/g), and K_L denotes the adsorption energy for fluoride (F⁻) (L/mg).

The Langmuir model suggests that adsorption forms a single layer of adsorbate on the adsorbent’s surface without interactions between adsorbed molecules, with a maximum capacity (Q_m). The Freundlich equation describes the relationship between the equilibrium concentration of fluoride (F⁻) in mg/L and the adsorbent’s adsorption capacity (Q_e) in mg/g. The empirical relationship can be expressed as follows (Equation (4)):

$$Q_e=K_F{C}_e^{1/n}$$

(4)

where K_F is a constant related to the capacity, and 1/n is a constant associated with the adsorption intensity or the surface heterogeneity.

2.4.2. Kinetics Experiments

Kinetic experiments of F⁻ adsorption were performed by adding 0.5 g/L of each adsorbent to 10 mL of fluoride ion solution with an initial concentration of 5 mg/L in falcon tubes at ideal pH values and constant temperature. The samples were shaken and withdrawn at specified times between 1 and 1440 min to measure the residual concentration of F⁻.

The adsorption models can be divided in two major categories: empirical and phenomenological ones. Typical empirical models are the pseudo-first-order (PSO) and pseudo-second-order (PSO) one, the Elovich model and the diffusion model. Their main differences are (i) the empirical models contain parameters with no direct physical meaning; (ii) the empirical models cannot be used in experimental conditions different than those of which they have been derived; and iii) the empirical models do not consider the mass balance of the solute in the system (i.e., as the adsorbed solute increases, the bulk concentration decreases by the same quantity). It is noted that the empirical models are used in the majority of the literature and, in particular, the PSO model is almost always successful in fitting the experimental data. Recently it was shown that this is due to the pathology of the linearization of the model to fit the data [39]. This is why the non-linear fitting procedure is used in the present work.

As a result, the PFO and PSO kinetics were evaluated [24]. The obtained kinetic adsorption values were then investigated in order to determine the sorption rates and rate expressions of the mechanisms. The PFO and PSO models are presented below via Equations (5) and (6), respectively:

$$Q_t=Q_e(1-e^{-k_{1}t})$$

(5)

$$Q_t={\displaystyle \frac{k_2{Q}_e^2\mathrm{t}}{1+k_2{Q}_{e}t}}$$

(6)

where Q_t and Q_e denote the fluoride (F⁻) adsorbed (mg/g) at time t (min) and at equilibrium, respectively; k₁ and k₂ are the PFO and PSO rate constants (g/mg min); and t is the time (min).

On the contrary, the phenomenological models are based on physical principles, contain physically meaningful parameters, can be used for predictive purposes, and take into account the balance of the solute in the system. It is noted that the characterization of a model as phenomenological implies that all the above requirements are fulfilled. For example, the diffusion model is based on a physical mechanism, so it can be not characterized as phenomenological because it does not contain the solute mass balance. The phenomenological model used here is a one based on chemical engineering approach. It is assumed that the slow phase of the adsorption is the diffusion of the solute on the adsorbent matrix. In particular, as is typical for this type of adsorbent, the dominant mechanism of diffusion is surface diffusion. The full model consists of a partial differential equation combined with some algebraic relation [40]. To simplify the numerical solution, the partial differential equation is converted into an ordinary differential equation using a technique known as LDF (linear driving force). This method relies on a quadratic approximation of the solution profile within the assumed spherical adsorbent particle [41].

The governing equation for the changes in the adsorbed amount Q is described by Equation (7):

$${\displaystyle \frac{dQ}{dt}}=K\left(f\left(C\right)-Q\right)$$

(7)

where K is a coefficient, which can be evaluated as K = D/(15 r²); r is the adsorbent particle radius; C is the solute concentration; f is the solute isotherm; and D is the effective total diffusivity.

The bulk concentration changes over time can be evaluated as follows (Equation (8)):

$$C=C_0-Q\left({\displaystyle \frac{m}{V}}\right)$$

(8)

The isotherm fittings showed that the Langmuir isotherm explains the adsorption equilibrium data (see next section). According to this statement, the function f has the following form (Equation (9)):

$$f(C)={\displaystyle \frac{Q_m{K}_{L}C}{1+K_{L}C}}$$

(9)

where the parameters Q_m and K_L for each adsorbent are listed in Table 1 (Table 3). Specifically, the differential Equation (7) is discretized using a logarithmic grid and solved with the explicit Euler method on a fine grid.

2.5. Thermodynamics

Temperature influences the adsorption results, and its impact was assessed at various temperature levels, including 303, 313, and 323 K, using the optimal dose and pH value over a contact period of 90 min. The thermodynamic analysis of the adsorption process involves evaluating changes in Gibbs free energy (ΔG°, kJ/mol), enthalpy (ΔH°, kJ/mol), and entropy (ΔS°, kJ/mol·K). Four temperatures (303, 313, and 323 K) were used to calculate these parameters, with Equations (10)–(13) applied for the computations [42].

$${{\mathsf{\Delta }G}^0=\mathsf{\Delta }H}^0-T{\mathsf{\Delta }S}^0$$

(10)

$${\mathsf{\Delta }G}^0={-RT\mathrm{l}\mathrm{n}(K}_c)$$

(11)

$$K_c={\displaystyle \frac{C_s}{C_e}}$$

(12)

$$ln\left(K_c\right)=\left(-{\displaystyle \frac{{\Delta H}^0}{R}}\right)+{\displaystyle \frac{{\Delta S}^0}{R}}$$

(13)

where K_C is the thermodynamic constant, R is the universal gas constant (8.314 J/mol·K), and T is the temperature (K). C_s (mg/L) represents the quantity adsorbed at equilibrium, and C_e denotes the concentration (mg/L) at equilibrium. ΔG° was obtained from Equation (6), while ΔH° and ΔS° were determined from the slope and intercept of the plot of ln(K_C) versus 1/T.

2.6. Characterization Techniques

The surface of the CS/OP/AC@MgO composite was characterized using Scanning Electron Microscopy (SEM) (Jeol JSM-6390 LV, Japan); Fourier Transform Infrared Spectroscopy (FT-IR) (Perkin Elmer, New York, NY, USA); and Brunauer, Emmett, and Teller (BET) analysis software (Quantachrome NovaWin - Data Acquisition and Reduction for NOVA instruments ©1994-2012, Quantachrome Instruments version 11.02). Additionally, X-ray diffraction (XRD), using a Rigaku MiniFlex II (Rigaku Corporation, Tokyo, Japan), was performed to assess crystallinity over a 2θ range of 5–90° using a BRUKER D8 FOCUS diffractometer with CuKα radiation (λ = 0.154 nm).

3. Results and Discussion

3.1. Impact of Initial pH Solution—Comparison of the Materials

The pH influences the effectiveness of adsorbent materials. Batch experiments were conducted to examine how the initial pH and adsorbent composition affect fluoride (F⁻) removal. This study adjusted the pH between 3.0 and 9.0 ± 0.1 at 293 K using a constant adsorbent dose of 0.5 g/L for 24 h. The results, shown in Figure 1, indicated that the most effective materials were Cs/OP, CS/AC, CS/AC@MgO, and CS/OP/AC@MgO. These materials showed the best performance in an acidic environment, specifically at pH 3.0 ± 0.1, where the adsorption of fluoride ions was maximum. For example, CS/OP/AC@MgO, which is the most complex synthesized material in this study, showed the highest efficiency, with about 75% removal of F⁻ at 3.0 ± 0.1, following CS/AC@MgO (70% removal), indicating that the addition of OP in its structure enhanced the performance of the adsorbent. Therefore, Cs/OP, CS/AC, CS/AC@MgO, and CS/OP/AC@MgO were selected for further study in the following experiments. Acidic conditions for the extraction of fluoride ions from aqueous solutions were also found to be optimal in recent literature [43,44].

The point of zero charge (pH_pzc) of the optimal CS/OP/AC@MgO material was determined using the pH drift method [45] across a pH range of approximately 2 to 10. By plotting ΔpH against pH_initial using the pH shift method (Figure 2), the pHpzc for CS/OP/AC@MgO was calculated to be 8.95. Thus, at pH values below this (e.g., 3.0 < 8.95), the surface of CS/OP/AC@MgO is positively charged, which facilitates the electrostatic attraction of anions such as fluoride [45].

3.2. The Impact of the Adsorbent’s Dosage

The impact of the adsorbent’s dosage for the optimum adsorbents Cs/OP, CS/AC, CS/AC@MgO, and CS/OP/AC@MgO on F-removal was studied by using various doses ranging between 0.3 and 1.0 g/L, according to recent literature [5,46,47,48,49], while other parameters were kept constant. The results shown in Figure 3 indicate that the removal of fluoride ions (%) shows a significant increase at higher adsorbent doses. For the CS/OP/AC@MgO material, the fluoride removal efficiency increased from about 65% to 75% when the dose increased from 0.3 g/L to 0.2 g/L. The CS/AC@MgO material showed a similar trend, with the efficiency increasing from about 50% to 68% with the same dose variation. In the case of Cs/OP and CS/AC materials, a significant increase in adsorption efficiency was also observed, with efficiencies reaching about 53% and 63%, respectively, applying a dosage of 0.5 g/L. Further increasing the dosage to 0.8 g/L and 1 g/L resulted in improved removal for all materials, reaching 97% when 1.0 g/L of the optimum CS/OP/AC@MgO was used. As expected [5,49], increasing the amount of adsorbent increases the available surface area, i.e., the active sites for adsorption [49], resulting in a corresponding increase in fluoride removal efficiency.

3.3. The Impact of Contact Time

The kinetic behavior of adsorbent materials was investigated by analyzing fluoride (F⁻) adsorption over contact times ranging from 5 min to 24 h using Cs/OP, CS/AC, CS/AC@MgO, and CS/OP/AC@MgO. All other parameters were kept constant. The results, shown in Figure 4, indicated that the fluoride removal rate increased with contact time and plateaued after 180 min for Cs/OP (about 55%). For CS/AC, CS/AC@MgO, and CS/OP/AC@MgO, removal rates continued to increase up to 1400 min, with an additional 5% increase for each material. Therefore, 180 min is considered when most adsorption is completed. To optimize cost-effectiveness, a contact time of 90 min was chosen for further experiments. At this duration, the fluoride removal was 55% for CS/AC, 53% for CS/AC@MgO, and 61% for CS/OP/AC@MgO. Notably, CS/OP/AC@MgO achieved a residual fluoride concentration of 1.47 mg/L, below the WHO limit, and reached an ideal level of 1.2 mg/L after 24 h [3].

3.4. Adsorption Kinetics

3.4.1. Empirical Models

The kinetic study of F⁻ ions adsorption was carried out using pseudo-first (PFO)- and pseudo-second (PSO)-order models to understand better the mechanism and rate of adsorption on the various materials. The results are presented in Figure 5 and summarized in

Table 2. PFO and PSO kinetic model parameters for F⁻ adsorption on the optimal materials.

Adsorbent	Qe.exp (mg/g)	PFO Model			PSO Model
		K1 (L/mg∙min)	Qe.cal (mg/g)	R2	K2 (L/mg∙min)	Qe.cal (mg/g)	R2
CS/OP	5.73	0.02342	5.69	0.9763	0.00354	7.03	0.9777
CS/AC	6.49	0.02215	6.10	0.9697	0.00308	7.57	0.9744
CS/AC@MgO	6.96	0.00208	8.89	0.9605	0.01901	6.95	0.9613
CS/OP/AC@MgO	7.57	0.0355	6.89	0.9561	0.00604	7.93	0.9789

, highlighting the different behavior of each material in the adsorption of fluoride ions. The PFO model data showed that this model does not provide the best fit for most materials. In particular, the CS/OP/AC@MgO material provided a correlation coefficient of R² = 0.9561, the CS/AC material also showed a low fit with R² = 0.9697, and the CS/AC@MgO and Cs/OP materials showed similar low R² values (0.9605 and 0.9763, respectively), indicating that the adsorption of fluoride ions does not only depend on the availability of surface sites, but other factors are involved, and this model cannot describe sufficiently [50].

On the other hand, the pseudo-second-order model presented a much better fit to the experimental data, indicating that chemisorption, i.e., the chemical interaction of fluoride ions with the adsorbent materials, is the dominant mechanism of adsorption [50,51,52]. For example, the data showed that for CS/OP/AC@MgO, the fit was much better with R² = 0.9789, suggesting that the adsorption kinetics mainly depend on the chemical reactions on the material surface [51]. The Q_e.cal values for the PSO model were very close to the experimental Q_e,exp values, reinforcing the view that it accurately describes the adsorption. For example, for CS/OP/AC@MgO, the experimental value Q_e,exp was 7.57 mg/g, while the value calculated from the application of the equations was Q_e.cal 7.93 mg/g. In addition, CS/AC@MgO had, respectively, an experimental Q_e,exp value of 6.96 mg/g and a calculated Q_e.cal value of 6.95 mg/g, showing excellent agreement between the values.

3.4.2. Phenomenological Model

Moreover, it is noted that usually, in adsorption kinetic studies, the data show two-stage behavior: a fast stage in the beginning, followed by a slower one later. This behavior is modeled assuming that K is a function of Q. In extreme cases, a two-stage adsorption process is considered [53]. However, this double kinetics does not appear in the present data, so K can be assumed to take a constant value for each type of adsorption. The numerically solved model is fitted to the adsorption kinetics experimental data. The corresponding fitted curves are shown in Figure 6, together with experimental data. The only fitting parameter is the constant K. The optimum fitting values of K are 0.015 min⁻¹, 0.013 min⁻¹, 0.0105 min⁻¹, and 0.012 min⁻¹ for the adsorbents CS/OP, CS/AC, CS/AC@MgO, and CS/OP/AC@MgO, respectively. The range of the values of the parameter K is small, and given the scattering of the experimental data point, one could say that no significant variation of the kinetic among adsorbents is observed. An estimation of the effective diffusion coefficient D of F in the adsorbed particle can be found by employing the relation D = 15 Kr². Assuming an average value of K = 0.126 min⁻¹ and an estimation of the adsorbent particle radius equal to 50 μm, a characteristic value for the diffusivity D = 7.9×10⁻¹² m²/s is derived.

3.5. Adsorption Isotherms

Langmuir–Freundlich isotherm models, commonly used in recent literature for fluoride removal with chitosan-based adsorbents [5,46,54], were applied to evaluate the equilibrium adsorption data for fluoride ions on the optimal materials Cs/OP, CS/AC, CS/AC@MgO, and CS/OP/AC@MgO under optimal conditions (pH 3, dose of 0.5 g/L, 293 K, and contact time of 90 min). Experimental data were fitted to these models, and the non-linear plots are shown in Figure 7, with relevant equilibrium constants and coefficients summarized in

Table 3. Isotherm models constants for F⁻ adsorption on optimal materials.

Langmuir Isotherm Model
Adsorbent	Qm (mg/g)	KL (L/mg)	R2
CS/OP	25.3	0.1121	0.9952
CS/AC	25.4	0.1359	0.9969
CS/AC@MgO	26.92	0.1752	0.9969
CS/OP/AC@MgO	61.3	0.0663	0.9986
Freundlich Isotherm Model
Adsorbent	1/n	KF (mg/g)(L/mg)1/n	R2
CS/OP	0.4694	3.862	0.9567
CS/AC	0.4409	4.470	0.9591
CS/AC@MgO	0.4018	5.716	0.9513
CS/OP/AC@MgO	0.5931	5.623	0.9937

. The Langmuir model, which assumes monolayer adsorption on a homogeneous surface with fixed active sites, showed that CS/OP/AC@MgO had the highest maximum adsorption capacity (Q_m = 61.3 mg/g), and CS/AC@MgO also demonstrated significant capacity (Q_m = 26.92 mg/g). The high R² value for CS/OP/AC@MgO (0.9986) indicates an excellent fit to the Langmuir model, confirming monolayer [55] and homogeneous adsorption. In contrast, the Freundlich model [37] showed that CS/OP/AC@MgO had a high K_F value (5.623), indicating a good adsorption capacity at low fluoride concentrations. CS/AC@MgO also had a notable K_F value (5.716), suggesting a strong affinity for fluoride ions even in dilute solutions. However, the R² values for the Freundlich model were slightly lower than those for Langmuir, indicating that while the Freundlich model describes adsorption well, the Langmuir model is more suitable for these materials.

In conclusion, the data analysis shows that the Langmuir model is the most suitable for describing the adsorption of F⁻ in the examined materials, especially for CS/OP/AC@MgO, which showed the greatest adsorption capacity. Therefore, the high Q_m and R² values indicate that the adsorption on these materials is monolayer and homogeneous, which is characteristic of systems following the Langmuir model, with CS/OP/AC@MgO standing out as the most efficient according to isotherm data.

3.6. Effect of Temperature Thermodynamics

Figure 8 plots the adsorption capacity (% Removal) with temperature in Kelvin, following the thermodynamic study for the optimum materials Cs/OP, CS/AC, CS/AC@MgO, and CS/OP/AC@MgO at 293, 303, 313, and 323 K. The figure clearly shows that as the temperature rises, the capacity increases, leading to higher removal percentages. This enhanced adsorption efficiency at elevated temperatures suggests that the fluoride ion-adsorption process is endothermic. The increased removal rate at higher temperatures may be due to either stronger chemical interactions between fluoride ions and the adsorbent materials or the creation of new adsorption sites on the material surfaces [56]. For example, in the case of CS/OP/AC@MgO, there was an increase in the removal rate of F⁻ from 67% at 293 K to 83% at 323 K after 1.5 h, corresponding to a 1.24 and 0.83 mg/L residual fluoride concentration, respectively, values considered within the ideal desired established permissible limits [3].

The ΔH₀ and ΔS₀ values were calculated from the plot of ln(Kc) versus 1/T, with the optimal material, CS/OP/AC@MgO, showing a high R² correlation coefficient of 0.9998. The results, summarized in

Table 4. Thermodynamic parameters for F⁻ adsorption on optimal materials (C₀ of 5 mg/L; pH 3.0 ± 0.1.0; dose of 0.5 g/L; 303, 313, and 323 K; 1.5 h).

Adsorbent	T (K)	∆G0 (kJ/mol)	∆H0 (kJ/mol)	∆S0 (kJ/mol∙K)	R2
CS/OP	303	−0.521	15.933	0.0543	0.9811
	313	−1.064
	323	−1.607
CS/AC	303	−0.731	17.512	0.0602	0.9541
	313	−1.333
	323	−1.935
CS/AC@MgO	303	−1.374	23.593	0.0824	0.9834
	313	−2.198
	323	−3.022
CS/OP/AC@MgO	303	−2.854	19.481	0.0737	0.9998
	313	−3.591
	303	−4.328

, indicate negative ΔG₀ values for all materials, suggesting that fluoride ion removal via adsorption is spontaneous and does not require additional energy. The adsorption process was found to be endothermic, as evidenced by positive ΔH₀ values, and the positive ΔS₀ values reflect increased disorder between the solids and solutes [24].

3.7. Regeneration of CS/OP/AC@MgO

One of the important things that determines not only the efficiency but also the reusability of adsorbents is their potential for regeneration [57]. For this reason, the reusability of the most effective material for fluoride removal in this study, i.e., CS/OP/AC@MgO, was examined. Cycling experiments were conducted five times, using 5 mg/L as the initial concentration of F⁻ by adding 0.5 g/L of CS/OP/AC@MgO adsorbent at pH 3.0 ± 0.1 (adsorption pH). Subsequently to the first cycle, the residual solid CS/OP/AC@MgO was mixed with 0.1 M NaOH solution for 1.5 h (desorption pH = 10 ± 0.1). To eliminate alkalinity, the filtered material was washed with distilled water. This process was repetitive for an extra four cycles. As illustrated in Figure 9, the desorbed CS/OP/AC@MgO is possible to be reused for F removal for at least five cycles after regeneration. Quantifying the regeneration efficiency of the material, it follows that only about a 26% loss in the efficiency of the material is observed after five cycles of adsorption–desorption experiments, as in the first cycle, the removal was 67%, and after the fifth cycle, it was 41%. Therefore, CS/OP/AC@MgO material can be effectively reused several times for F removal.

3.8. Characterization of Optimum Adsorbents before and after Adsorption

Various techniques were used to characterize the surface of optimum Cs/OP, CS/AC, CS/AC@MgO, and CS/OP/AC@MgO adsorbents.

Table 5. Physical properties of optimal materials.

Physical Properties	CS/OP	CS/AC	CS/AC@MgO	CS/OP/AC@MgO
BET surface area, SBET (m2/g)	10.45	167	193	79.45
BJH pore size (Å)	12.8	8.1	6.8	6.4
Micropore volume, Vmicro (cm3/g)	0.00023	0.050	0.072	0.022
Total pore volume, VT (cm3/g)	0.013	0.107	0.127	0.057

shows the determined levels of BET surface area (S_BET), Barrett–Joyner–Halenda (BJH) average pore diameter, and total pore volume at P/P_o = 0.985. The micropore volume was also withdrawn in order to better determine the factor responsible for each material’s performance. Regarding S_BET, the order of the materials from the highest to the lowest values is CS/AC@MgO > CS/AC > CS/OP/AC@MgO > CS/OP, indicating that AC is the main contributor to the adsorption capacity of the material. The presence of MgO particles provides a subtle addition to the overall surface area, while OP substantially reduces this value; the surface area of CS/OP/AC@MgO is about 60% less than CS/AC@MgO.

Despite the fact that MgO is reported to reduce AC structural properties due to pore blocking by MgO particles [58], the polymeric matrix alters this interplay. According to Kim et al. [59], MgO addition to polymer/AC composites creates further roughness. In this study, the reduced pore size of CS/AC@MgO compared to CS/AC (6.8 Å and 8.1 Å, respectively) confirms the pore blocking, but both the inherited porosity of MgO particles and the additional surface roughness increases the surface area.

An interesting remark can be stated regarding fluoride removal, where CS/OP/AC@MgO shows the highest performance, although the measured structural properties do not suggest this. One possible explanation is that the formed crystals, under the presence of OP (Figure 7), possess active sites where fluoride ions can be attached or that the dissolution of these mineral salts creates a substantial concentration of Na⁺ in the solution, thus removing more F⁻. In any case, the difference between these two best materials regarding F-removal is relatively small.

Furthermore, as illustrated in Figure 10, according to SEM micrographs, the resulting surface of the prepared materials was investigated prior to and after the adsorption process. As observed in the different cases, the presence of orange peels results in a material with crystals formed on the organic matter (Figure 10a,d, before adsorption). The formation of crystals is attributed to the synthesis process, where the reaction of the acetic acid and NaOH can potentially grow crystals when in contact with the cellulosic fibers of the orange peel [60]. Interestingly, those crystals disappear after the use of the adsorbent in a fluoride-concentrated solution, exposing a relatively smooth surface, as is expected for orange peels [61]. Most probably, the crystals were removed by dissolving into the fluoride/water solution [62].

In the case of CS/AC, the morphology is different (Figure 10b before adsorption). Although the surface is still rough, no crystals are observed. As is characteristic of AC, the pores are more visible with a rather granular and uneven surface. After adsorption, the looser particles have been detached, leaving some areas smooth, where, most likely, the concentration of chitosan is higher. The addition of MgO does not seem to alter the surface characteristics of each two previous cases; however, it enhances the performance of the materials showcasing the best F removal percentage among the cases (Figure 1). The high affinity of F⁻ to MgO is well reported, and in this case, the MgO particles that are semi-embedded into the CS matrix provide a surface with substantial active sites to attract F⁻ ions [63].

SEM-EDS analysis of CS/AC@MgO and CS/OP/AC@MgO is presented in

Table 6. Elemental composition of CS/AC@MgO and CS/OP/AC@MgO adsorbents according to SEM/EDS analysis.

% (w/w)	CS/AC@MgO	CS/OP/AC@MgO
Au 1	65.89	47.61
C	4.98	13.13
Mg	5.00	6.65
N	2.64	4.86
Na	1.65	2.32
Cu	2.98	2.17
Re	4.56	2.57
Zn	1.04	1.31
Ca	2.86	0.98
Al	0.74	0.95
Tc	0.85	0.66
O	6.81	16.79

¹ Au is used for sample sputtering.

. According to the % (w/w) values of the EDS analysis, magnesium is detected on the surface of both composites, with Mg contributing 5.00% (w/w) to the structure of CS/AC@MgO and 6.65% (w/w) to the structure of CS/OP/AC@MgO. While EDS cannot distinguish between MgO and Mg(OH)₂, the significant presence of magnesium supports the possibility that both compounds are present. Previous studies have shown that MgO is poorly soluble in water [64,65], and the dissolution rate is highly dependent on factors such as temperature, water volume, and pH. Furthermore, the filtration step in this study takes place shortly after the introduction of MgO into the water, limiting the extent of conversion. Therefore, some MgO may remain unconverted, particularly in the solid phase. In addition, carbon, nitrogen, and oxygen are prominent elements in the structure of materials that exhibit high content rates. It is worth noting that in the case of CS/OP/AC@MgO, the percentage of carbon is higher, i.e., 13.13% versus 4.98% for CS/AC@MgO, owing to the possible presence of the cellulosic orange peel. Furthermore, Figure 11 illustrates the appearance of Mg on both CS/AC@MgO and CS/OP/AC@MgO.

Moreover, the distinct characteristic peaks in the FT-IR spectra (Figure 12 of the chitosan-based composites) indicate the distinct chemical structures of each material. The O-H and N-H stretching vibrations in CS produce prominent peaks around 3200–3500 cm⁻¹. The C-H stretching vibrations are around 2900 cm⁻¹, and the amide I and II bands are near 1650 cm⁻¹ and 1550 cm⁻¹, respectively. Orange peels add peaks centered around 3200–3500 cm⁻¹ from O-H stretching; however, a peak at 1730 cm⁻¹ from C=O stretching is expected due to polysaccharides such as cellulose and lignin being absent [66]. This can be attributed to the formation of the crystals. Additionally, C-N stretching contributes to an absorbance peak near 1250–1350 cm⁻¹, and the C-O-C stretching vibrations from glycosidic linkages are seen in the 1000–1100 cm⁻¹ range. Although all the samples are chitosan-based, spectra of the samples containing AC do not present all of the characteristic peaks, indicating that there are interactions of AC functional groups with those of the CS matrix (Figure 12b,d) [67,68]. In the case of MgO, its most prominent IR absorption bands are typically found at lower wavenumbers, particularly in the range of 400–600 cm⁻¹, which correspond to the Mg-O stretching vibrations. However, when MgO is part of a composite material, as in the present case, interactions between MgO and other components lead to the appearance of a peak around 1000 cm⁻¹ related to C-O-C stretching vibrations influenced by the presence of MgO [65].

Post-adsorption shifts in these peaks, particularly in the O-H, N-H, and amide regions, suggest that fluoride ions have interacted with these functional groups, indicating successful adsorption for all cases. More specifically, these changes are particularly evident in the O-H, N-H, and amide regions, indicating successful adsorption and bonding of fluoride ions with the composite’s active sites. The most persisting indication of F⁻ interaction with the adsorbents is the formation of strong hydrogen bonds. Thus, a shifting of C-H stretching to lower wavenumbers (3000 cm⁻¹) is observed in all the after-adsorption cases. Additionally, F⁻ adsorption creates small intensity peaks around 1000 cm⁻¹ due to C-O stretching, and especially in the case of CS/OP, it is more pronounced due to polysaccharides of the OP [69].

Furthermore, in Figure 13, the XRD patterns of Cs/OP, CS/AC, CS/AC@MgO, and CS/OP/AC@MgO are presented. In each one only the characteristic peak of chitosan exists at around 20°. The amorphous form of compounds often assists in the adsorption procedure of a substance [70].

3.9. Challenges and Future Perspectives

The effective removal of various pollutants is a challenge for researchers, and therefore, adsorption and membrane techniques are usually considered easier and more effective. In this study, a new proposed CS/OP/AC@MgO material was found to be suitable for fluoride extraction in batch/static experiments. As the batch tests used in this study are not representative of practical adsorption operations, although mainly used in relative literature, it is suggested, for future experiments, to use a continuous flow system (such as industrial operations), as large amounts of water can be treated in this way. For this reason, it is of increasing interest to combine adsorbents with membranes (i.e., adsorbent membranes) to better serve the needs of industrial operations [21]. Therefore, it is suggested for future experiments to place these solid adsorbents in a packed column or filtration column, as actually suggested for the adsorption of metal ions from aqueous sources such as brine, seawater, and polluted water [71,72,73,74]. Finally, it is suggested to carry out further studies on real groundwater (from different areas with different concentrations) in order to investigate more representative and realistic conditions. Thus, interfering ions present in real/simulated groundwater and the presence of contaminants that would seriously affect the sorption capacity will be examined.

4. Conclusions

This study concentrated on synthesizing, characterizing, and evaluating new chitosan-based adsorbent materials for fluoride ion removal from aqueous solutions. The findings indicated that modifying chitosan with materials like activated carbon (AC), magnesium oxide (MgO), and orange peels (OP) notably enhances its adsorption capacity. The optimal material, CS/OP/AC@MgO, showed the highest fluoride removal efficiency, with a removal rate of up to 97% under acidic conditions (pH 3.0 ± 0.1). The addition of MgO enhanced the positively charged surface of the materials, increasing the electrostatic attraction of the negatively charged fluoride ions, while the study of the kinetic models showed that the adsorption process mainly follows the pseudo-second-order model, indicating that the dominant mechanism is the chemisorption. In addition, a rather simple phenomenological kinetic model for the adsorption kinetics of fluoride ions on the surface of Cs/OP, CS/AC, CS/AC@MgO, and CS/OP/AC@MgO is presented in this study, which was applied comparatively to the numerical results and experimental data. As inferred, the application of a phenomenological model to fit the kinetic data leads to an estimation for the effective diffusion coefficient of F⁻ in the adsorbent equal to 7.9 × 10⁻¹² m²/s.

Furthermore, according to the best-fit Langmuir isotherm model, the CS/OP/AC@MgO material exhibited the highest maximum adsorption capacity (26.92 mg/g), confirming its efficiency in fluoride removal mainly through a monolayer-adsorption mechanism on homogeneous surfaces. Thermodynamic studies revealed that the adsorption is endothermic and spontaneous, while increasing the temperature further improved the efficiency of the process. According to the characterization results before and after adsorption, it was found that the addition of AC and MgO to chitosan notably enhances the physicochemical attributes of the adsorbent materials. Specifically, the materials incorporating activated carbon and magnesium oxide (CS/AC@MgO) showed the highest specific surface area of 193 m²/g. This indicates that activated carbon significantly contributes to the increase in active adsorption locations, while the addition of magnesium oxide further enhances surface roughness, providing additional active sites. However, it was observed that the presence of OP (CS/OP/AC@MgO) decreased the surface area, even though this adsorbent provided the highest fluoride removal efficiency. SEM micrographs confirmed that the addition of MgO does not significantly change the surface morphology, and FT-IR results demonstrated that fluoride ions interact with the functional groups of the materials, confirming successful adsorption through electrostatic interactions and hydrogen bonding. The XRD patterns showed the amorphous form of the materials. Overall, this research demonstrated that chitosan composites modified with environmentally friendly materials can effectively remove fluoride ions from water, presenting new opportunities for enhancing water quality and safeguarding public health.