Synthesis and Application of Fe₃O₄–ZrO₂ Magnetic Nanoparticles for Fluoride Adsorption from Water

Abstract

This study presents the synthesis, characterization, and application of magnetic magnetite–zirconium dioxide (Fe₃O₄–ZrO₂) nanoparticles as an efficient nanoadsorbent for fluoride removal from water. The nanoparticles were synthesized using a wet chemical co-precipitation method with Fe/Zr molar ratios of 1:1, 1:2, and 1:4, and characterized using Fourier-transform infrared (FTIR) spectroscopy, scanning electron microscopy (SEM), X-ray diffraction (XRD), and energy-dispersive X-ray spectroscopy (EDS). FTIR analysis confirmed the presence of Fe₃O₄ and ZrO₂ functional groups, while XRD showed that increased Zr content led to a dominant amorphous phase. SEM and EDS analyses revealed quasi-spherical and elongated morphologies with uniform elemental distribution, maintaining the designed Fe/Zr ratios. Preliminary adsorption tests identified the Fe/Zr = 1:1 (M1) nanoadsorbent as the most effective due to its high surface homogeneity and optimal fluoride-binding characteristics. Adsorption experiments demonstrated that the material achieved a maximum fluoride adsorption capacity of 70.4 mg/g at pH 3, with the adsorption process best fitting the Temkin isotherm model (R² = 0.987), suggesting strong adsorbate–adsorbent interactions. pH-dependent studies confirmed that adsorption efficiency decreased at higher pH values due to electrostatic repulsion and competition with hydroxyl ions. Competitive ion experiments revealed that common anions such as nitrate, chloride, and sulfate had negligible effects on fluoride adsorption, whereas bicarbonate, carbonate, and phosphate reduced removal efficiency due to their strong interactions with active adsorption sites. The Fe₃O₄–ZrO₂ nanoadsorbent exhibited excellent magnetic properties, facilitating rapid and efficient separation using an external magnetic field, making it a promising candidate for practical water treatment applications.

1. Introduction

The contamination of drinking water with fluoride from both natural sources and human activities is a significant global concern [1]. According to the World Health Organization (WHO), the maximum permissible fluoride concentration in drinking water is 1.5 mg/L [2,3,4]. Excess fluoride intake can lead to dental and skeletal fluorosis, posing serious health risks [5,6]. Various methods have been employed to remove fluoride from water, including precipitation, ion exchange, and adsorption [7,8,9]. Among these, adsorption stands out as a highly effective approach, capable of reducing fluoride levels below 1 mg/L [10,11]. The use of nanosized adsorbents is particularly promising due to their high surface area, which enhances adsorption efficiency [12,13].

Metal oxides such as titanium dioxide (TiO₂), aluminum oxide (Al₂O₃), and zirconium dioxide (ZrO₂) have demonstrated excellent selectivity for anion removal in aqueous solutions [14,15,16]. Among them, zirconium-based materials have attracted significant attention due to their strong affinity for fluoride ions [17,18,19]. However, challenges remain in using nanomaterials for water treatment, particularly in conventional packed-bed reactors and stirred tank reactors, where their small size complicates solid–liquid separation [20].

Magnetic adsorbents offer a practical solution to these limitations, combining high adsorption efficiency with easy recovery using an external magnetic field [21]. Magnetite (Fe₃O₄) has been widely employed for removing contaminants such as heavy metals [22], nanoplastics [23], dyes [24], pharmaceuticals [25], and fluoride [26]. Among the various synthesis methods for magnetic iron oxides, co-precipitation is preferred for its simplicity, scalability, and ability to produce high-purity Fe₃O₄ nanoparticles. This method involves the simultaneous precipitation of Fe²⁺ and Fe³⁺ salts in an alkaline medium, typically under nitrogen or inert atmospheres to prevent oxidation [27]. Optimizing parameters such as pH, temperature, ionic strength, and reaction time significantly influences particle size, crystallinity, and magnetic properties [28]. Additionally, co-precipitation has been successfully used to synthesize Fe₃O₄ composites with other oxides (e.g., ZrO₂, TiO₂, SiO₂), enhancing their stability and adsorption performance [29].

Magnetic zirconium–iron oxide nanoparticles (Fe₃O₄@ZrO₂) synthesized via co-precipitation have demonstrated promising adsorption capabilities for various contaminants. Studies have shown that Fe/Zr molar ratios influence surface area and adsorption capacity, with lower Fe/Zr ratios improving phosphate removal [30]. Similarly, Alam et al. [31] synthesized Fe₃O₄@mZrO₂ nanoparticles doped with rare-earth elements, achieving enhanced arsenic adsorption. Recently, Riahi et al. [20] reported the modification of Fe₃O₄ nanoparticles with zirconium oxide for fluoride removal.

In this study, we present a simple and efficient method for fluoride removal using magnetic Fe₃O₄–ZrO₂ nanoparticles. First, we synthesized and characterized the nanoparticles using the co-precipitation method. Then, we systematically investigated key factors affecting fluoride adsorption, including the Fe/Zr ratio, solution pH, contact time, initial fluoride concentration, and the presence of competing anions. Our results demonstrate that Fe₃O₄–ZrO₂ nanoparticles exhibit a high fluoride adsorption capacity, fast kinetics, and strong selectivity in complex water matrices. Furthermore, the nanoparticles can be efficiently separated using a magnetic field, enabling their potential application in practical water treatment systems.

2. Results and Discussion

2.1. Characterization of the Magnetic Fe₃O₄–ZrO₂ Adsorbents

2.1.1. FTIR Spectra Analysis

The FTIR spectra of the synthesized magnetic adsorbents are presented in Figure 1. All adsorbents exhibited similar absorption bands, though with varying intensities depending on the Fe/Zr ratios. The signal at 3300 cm⁻¹ corresponds to the stretching vibration of hydroxyl (-OH) groups from surface-bound water [32], which appears with comparable intensity in all spectra. The band at 880 cm⁻¹, associated with the bending vibrational mode of iron hydroxide [33], is present in the spectra of M1 and M2 but absent in M3. This absence is attributed to the stabilizing effect between the metal oxides, which promotes the crystalline phase of Fe₃O₄ as Zr content increases, thereby preventing the formation of iron hydroxide byproducts [34]. Similarly, the characteristic Fe₃O₄ fingerprint at 580 cm⁻¹, corresponding to the stretching vibrational mode of Fe₃O₄ [35], intensifies with increasing Zr content. Peaks at 1340 cm⁻¹ and 1600 cm⁻¹ are associated with the stretching and bending vibrational modes of zirconium–oxygen and zirconium hydroxide (Zr–OH), respectively [36].

2.1.2. X-Ray Diffraction (XRD) Analysis

Figure 2 presents the XRD patterns of the magnetic adsorbents. The diffraction peaks observed for M1 at 2θ values of 30.2°, 33.3°, and 35.5° correspond to the (111), (101), and (200) planes of zirconium oxide, respectively (JCPDS 96-152-2144) [20,37,38]. Peaks at 30.2°, 35.5°, 43.2°, 53.4°, 57.2°, and 62.7° were assigned to the (220), (311), (400), (422), (511), and (440) planes of Fe₃O₄ (JCPDS 96-900-5839) [20,30,39,40]. A peak at 21.1° corresponds to iron hydroxide, with no impurity peaks detected, confirming the spinel structure of Fe₃O₄ [20].

The XRD patterns reveal a progressive decrease in Fe₃O₄ diffraction peak intensity as ZrO₂ content increases. With increasing zirconium oxide, the amorphous phase dominates in M2 and M3 due to the lack of calcination [41], with only the (200) and (111) peaks of ZrO₂ remaining detectable. This phenomenon is attributed to the predominant amorphous phase of ZrO₂, which suppresses the diffraction signals of the crystalline Fe₃O₄ phase. Additionally, increased dispersion of Fe₃O₄ within the ZrO₂ matrix results in a dilution effect, further diminishing peak intensity [31]. These observations suggest that higher ZrO₂ content induces structural distortions in Fe₃O₄, reducing its crystallinity and promoting a more homogeneous composite material [42].

2.1.3. Morphology and Elemental Composition (SEM and EDS)

The morphology and elemental composition of M1, M2, and M3 were examined using SEM and EDS (Figure 3). The magnetic adsorbents consisted of elongated and quasi-spherical nanoparticles, structures previously reported in the literature [43]. As the zirconium oxide content increased, the proportion of elongated and quasi-spherical nanoparticles increased, while rectangular structures almost disappeared in M3. The decrease in elongated structures with increasing Zr content suggests that these initial structures were associated with iron oxides. The surface of the composites appeared relatively rough, composed of numerous aggregated nanoparticles, likely due to oxidation during SEM preparation.

EDS analysis (Figure 3D, Supplementary Information Figures S1–S3) confirmed the presence of Fe, Zr, and O in the synthesized material. The empirical formulas obtained—FeZrO_4·5, Fe₂Zr₄O₁₁, and FeZr₄O₁₁ for M1, M2, and M3, respectively—demonstrate that the Fe/Zr ratios used during synthesis were maintained in the final products, validating the proposed compositions.

2.2. Fluoride Adsorption Studies

2.2.1. Effect of pH on Fluoride Removal

Preliminary adsorption tests, where the adsorbents interacted with water samples for 24 h, showed that all three materials (M1, M2, and M3) captured nearly the same amount of fluoride: 99.0, 99.4, and 99.5%, respectively. Since zirconium was the most expensive reagent, we excluded M2 and M3 from further analysis, as their higher zirconia content did not provide a significant advantage in fluoride adsorption. Additionally, SEM images indicated that M1 exhibited the most uniform size distribution among the three materials; therefore, subsequent tests were conducted exclusively with the M1 adsorbent.

Fluoride adsorption onto the synthesized material occurs at the adsorbent’s surface, where ion exchange and electrostatic interactions take place. The adsorption capacity for F⁻ ions is significantly influenced by pH, as it dictates the surface charge distribution of the material [44].

The relationship between pH and the surface zeta potential of Fe₃O₄–ZrO₂ composites was studied by Zhang et al. [30]. The point of zero charge (PZC) values of the composites were in the pH range of 8–9. Superparamagnetic zirconia material had a PZC around pH 5.8 [45]. The PZC of Fe₃O₄ nanoparticles was about pH 7.38 [46]. According to these results, it is assumed that the M1 adsorbent had a similar value of PZC. When the solution pH is below this point, adsorbents have a positive charge. The positive surface charge promotes the electrostatic attraction of fluoride anions.

As illustrated in Figure 4, maximum adsorption was observed at pH 3, corresponding to a positively charged surface that facilitates efficient interaction with negatively charged fluoride ions. At higher pH levels, surface deprotonation results in a negatively charged surface, leading to electrostatic repulsion between the adsorbent and fluoride ions, as well as competition from hydroxyl ions for available adsorption sites. However, fluoride adsorption remained over 60% across a range of pH values from 1 to 7, indicating that the material retains its adsorptive properties under different conditions. Additionally, pH adjustment to acidic conditions is a simple and cost-effective strategy widely employed in water treatment processes, as it can be achieved using common acidifying agents such as sulfuric or hydrochloric acid without significantly increasing operational complexity [47]. Moreover, the natural fluoride contamination and industrial processes like metal smelting contribute vast amounts of untreated fluoridated wastewater [48,49]. Metallurgical wastewater often exhibits acidic pH due to the use of acids in various metal processing steps. Based on these considerations, a pH of 3 was selected for subsequent experiments.

2.2.2. Kinetics Analysis

The adsorption kinetics of fluoride onto the M1 composite was evaluated using three kinetic models: pseudo-first-order, pseudo-second-order (both linear and nonlinear forms), and the Elovich model. Experiments were carried out using an initial fluoride concentration of 10 mg/L, with 50 mg of adsorbent dispersed in 50 mL of deionized water at pH 3 and ambient temperature. In these graphs, it can be observed that more than 90% of the fluoride adsorbed at equilibrium (q_e, mg/g) occurred within the first 30 min from the start of the reaction. The results indicate that the equilibrium time (t_e) for fluoride adsorption from a 10.0 mg/L solution was approximately 2 h (Figure 5).

The best fit to the experimental data was obtained using the nonlinear pseudo-second-order model [50], expressed as follows:

q_t = (k₂·q_e²·t)/(1 + k₂·q_e·t)

where q_t (mg/g) is the amount of fluoride adsorbed at time t (min), q_e (mg/g) is the adsorption capacity at equilibrium, and k₂ (g·mg⁻¹·min⁻¹) is the pseudo-second-order rate constant. This model produced the highest correlation coefficient (R² = 0.9724) and the lowest reduced chi-square value (0.00572) (

Table 1. Kinetic models, equations, and estimated parameters for fluoride adsorption onto composite M1.

Model	Equation	qe (mg/g)	Constant(s)	R2	Reduced χ2
Pseudo-first order	qt = qe · (1 − e^(−k1·t))	9.69 ± 0.08	k1 = 0.2135 ± 0.0186 (1/min)	0.8063	0.04012
Pseudo-second order	qt = (qe2·k2·t)/(1 + qe·k2·t)	9.86 ± 0.04	k2 = 0.0731 ± 0.0057 (g/mg·min)	0.9724	0.00572
Elovich	qt = (1/β)·ln(1 + α·β·t)	–	α = 7.04 × 1019 ± 1.18 × 1021 (mg/g·min)	0.6095	0.08089
			β = 5.46 ± 1.79 (g/mg)

). The estimated parameters were q_e = 9.86 mg/g and k₂ = 0.07307 g/(mg·min).

In contrast, the pseudo-first-order model [51] is defined by the equation:

q_t = q_e·(1 − e^(−k₁·t))

where q_t (mg/g) is the amount of fluoride adsorbed at time t (min), q_e (mg/g) is the adsorption capacity at equilibrium, and k₁ (min⁻¹) is the pseudo-first-order rate constant. This model exhibited a weaker correlation with the experimental data (R² = 0.8063) (Table 1), indicating that it does not adequately describe the fluoride adsorption mechanism on M1.

Finally, the Elovich model [52] is given by the following:

q_t = (1/β)·ln(1 + α·β·t)

where α (mg·g⁻¹·min⁻¹) is the initial adsorption rate, and β (g·mg⁻¹) is related to surface coverage and activation energy. The Elovich model showed the poorest fit (R² = 0.6095) (Table 1), confirming that the pseudo-second-order model best represents the kinetic behavior of fluoride adsorption on M1.

These results support a mechanism dominated by chemisorption, likely involving electron sharing or exchange between fluoride ions and active sites on the composite surface [53].

2.2.3. Adsorption Isotherm Analysis

Experimental adsorption data and the best-fit linear equation are shown in Figure 6. Experiments were conducted with a rotating mixer at 40 rpm for 2 h, followed by magnetic separation of the adsorbents and measurement of fluoride concentrations using an ion-selective electrode. Tests were performed in duplicate.

The Freundlich, Langmuir, and Temkin isotherm models were applied to analyze fluoride adsorption equilibrium, with results compared in

Table 2. Freundlich, Langmuir, and Temkin isotherms for the adsorption of fluoride ions onto the Fe₃O₄–ZrO₂ composites.

Isotherm	Equation	Parameters
Model				R2	χ2
Langmuir		${\mathit{Q}}_{\mathit{m}\mathit{a}\mathit{x}}^{\mathit{0}}$	KL
		(mg/g)	(L/g)
Nonlinear	$q_e={\displaystyle \frac{Q_{max}^0{K}_L{C}_e}{1+K_L{C}_e}}$	70.40	0.21	0.905	56.52
Linear	${\displaystyle \frac{C_e}{q_e}}=\left({\displaystyle \frac{1}{Q_{max}^0}}\right)C_e+{\displaystyle \frac{1}{Q_{max}^0{K}_L}}$	70.42	0.43	0.942	-
Freundlich		KF	n
		(mg/g)
Nonlinear	$q_e=K_F{C}_e^n$	20.80	0.24	0.948	31.30
Linear	$\mathit{log}{q}_e=n\mathit{log}{C}_e+\mathit{log}{K}_F$	14.62	0.33	0.88	-
Temkin		bT	AT
		(J/mol)	(L/g)
Nonlinear	$q_e={\displaystyle \frac{RT}{b_T}}\left(\mathit{ln}{A}_T{C}_e\right)$	274.4	10.42	0.987	7.91
Linear	$q_e=B\mathit{ln}{C}_e+B\mathit{ln}{A}_T$	251.1	8.60	0.975	-

. The nonlinear model fitting and the inset in Figure 6 indicate that the Temkin model provided the best fit, as evidenced by superior regression coefficients (R²) and chi-square (χ²) values. The maximum adsorption capacity of Fe₃O₄–ZrO₂ nanoparticles, determined from the Langmuir model, was 70.4 mg/g, closely matching the experimentally measured capacity of 69.1 mg/g.

Langmuir adsorption parameters were obtained using both linear and nonlinear equations, where K_L (L/g) represents the Langmuir constant and Q⁰_max (mg/g) signifies the maximum monolayer coverage capacity. The Freundlich model describes adsorption intensity (n) and the isotherm constant (K_F). For the Temkin model, the adsorption parameters A_T (8.60–10.42 L/g) and b_T (251.1–274.4 J/mol) correspond to equilibrium binding and adsorption energy constants, respectively [54]. The positive B value (9.87) suggests an exothermic adsorption process [55].

The Gibbs free energy can also be calculated from the thermodynamic equilibrium constant, K₀, which is defined as follows:

$$K_0={\displaystyle \frac{a_s}{a_e}}={\displaystyle \frac{v_s{q}_e}{v_e{C}_e}}$$

where a_s is the activity of the adsorbed fluoride ion, a_e is the activity of the fluoride ion in the equilibrium solution, ν_s is the activity coefficient of the adsorbed fluoride, and ν_e is the activity coefficient of fluoride in the equilibrium solution. The above equation can be rewritten as follows:

$$\underset{q_e\to 0}{\mathit{lim}}={\displaystyle \frac{a_s}{a_e}}={\displaystyle \frac{q_e}{C_e}}=K_0$$

K₀ can be obtained by plotting ln (q_e/C_e) vs. q_e (Figure 7) and extrapolating to zero [56]. Its intersection gives the values of K₀. In this analysis, the value of K₀ could be calculated from the data obtained from the adsorption isotherms at room temperature.

The standard Gibbs free energy changes of adsorption (ΔG⁰) can be calculated according to the following:

$$\mathsf{\Delta }{G}^0=-RT\mathit{ln}{K}_0$$

The obtained values are shown in

Table 3. Equilibrium constant K₀ and Gibbs free energy of fluoride ion adsorption on Fe₃O₄–ZrO₂ at room temperature and C₀ = 5−300 mg/L.

	Fe3O4–ZrO2
Intersection, (ln K0)	4.185
ΔG0(kJ/mol)	−10.37
R2	0.9461

. The standard Gibbs free energy change (ΔG⁰) in the F⁻–Fe₃O₄–ZrO₂ system is −10.37 kJ/mol, indicating that the adsorption process is spontaneous.

The Dubinin–Radushkevich (D-R) equation was developed to account for the effect of the porous structure of an adsorbent [57]. It is expressed as follows:

$$q_e=q_{DR}{e}^{-K_{DR}{\epsilon }^2}$$

The linear form of the Dubinin–Radushkevich equation is as follows:

$$\mathit{ln}{q}_e=-K_{DR}{\epsilon }^2+\mathit{ln}{q}_{DR}$$

$$\epsilon =RT\mathit{ln}\left(1+{\displaystyle \frac{1}{C_e}}\right)$$

where q_e (mol/g) is the amount of fluoride adsorbed at equilibrium, C_e (mol/L) is the equilibrium fluoride concentration, q_DR (mol/g) is the theoretical adsorption capacity, K_DR (mol²/kJ²) is a constant related to the sorption energy, ε is the Polanyi potential, R is the universal gas constant (8.314 × 10⁻³ kJ/mol·K), and T is the temperature in Kelvin.

The parameters q_DR and K_DR in the above equation can be obtained as follows: a plot of ln q_e versus R²T² ln² (1 + 1/C_e) has a slope of -K_DR and an intercept of lnq_DR (Figure 8).

The mean free energy of adsorption E_DR is important for distinguishing between physical and chemical adsorption processes. It is defined as the change in free energy when 1 mole of a solute is transferred to the surface of an adsorbent from infinity, i.e., from the solution [58], and it can be calculated using the following equation [59]:

$$E_{DR}={\displaystyle \frac{1}{\sqrt{2K_{DR}}}}$$

The obtained E_DR value was 12.11 kJ/mol (

Table 4. Parameter values for fluoride adsorption on Fe₃O₄–ZrO₂.

qDR (mg/g)	KDR (mol2/kJ2)	EDR (kJ/mol)	R2
119.12	3.41 × 10−3	12.11	0.932

). The magnitude of E_DR is used to estimate the type of adsorption mechanism:

(i)

Physisorption if E_DR < 8 kJ/mol;

(ii)

Ion exchange if E_DR = 8.0–16.0 kJ/mol;

(iii)

Chemisorption if E_DR > 16.0 kJ/mol [60].

2.2.4. Effects of Interfering Anions on Fluoride Adsorption

Adsorption selectivity is a crucial factor in fluoride removal efficiency, as competing ions in wastewater can affect adsorption performance. Interference tests were conducted using common anions found in drinking water: nitrate (NO₃⁻), chloride (Cl⁻), sulfate (SO₄²⁻), bicarbonate (HCO₃⁻), phosphate (PO₄³⁻), and carbonate (CO₃²⁻). The results (Figure 9) showed that nitrate, chloride, and sulfate had a minimal impact on fluoride adsorption, aligning with previous studies [17].

However, bicarbonate and carbonate ions significantly reduced fluoride uptake by more than 50% at concentrations above 1 mM. This is attributed to the formation of a stable pentacyclic complex between carbonate ions and Zr⁴⁺, occupying active sites and competing with fluoride ions [61]. Phosphate ions also inhibited fluoride adsorption, primarily due to hydroxyl (-OH) groups formed during Na₃PO₄ hydrolysis, which increased the pH and intensified competition for active sites [30]. FTIR spectroscopy was used to confirm carbonate and phosphate ions’ adsorption on the surface of the M1 adsorbent (Figure S4).

These findings underscore the importance of understanding competing ion interactions to optimize fluoride removal using Fe₃O₄–ZrO₂ composites.

2.2.5. Comparison of Fe₃O₄–ZrO₂ with Other Fluoride Adsorbents

To evaluate the efficiency and practicality of Fe₃O₄–ZrO₂ as a fluoride adsorbent, a comparative analysis with other materials reported in the literature was conducted.

Table 5. Comparison of fluoride adsorbents with different compositions and synthesis methods.

Reference	Material	Max Adsorption Capacity (mg/g)	pH Range/Optimal	Contact Time	Synthesis Method	Magnetic/ Calcination/ Scalable
[62]	Fe3O4@MgO	98.4	2–11/7	30 min	SE, C 700 °C	Yes/Yes/No
[63]	Fe3O4-SiO2	5.6	ND/6.8	10 min	CP, C 500 °C	Yes/Yes/No
[64]	Fe3O4-Activated Carbon	146.2	4–11/7	75 min	CP, CA: 650 °C	Yes/Yes/No
[65]	Fe3O4-CaCO3	17.95	2–12/6.8	60 min	WI, C 500 °C	Yes/Yes/No
[66]	Fe3O4@SiO2/Cdots	21.4	ND/5.5	60 min	HT, SG	Yes/No/No
[67]	Fe3O4@NiO	63.86	ND/4	120 min	ED	Yes/No/No
[68]	Fe3O4-ZnO	7.8	ND	80 min	HT	Yes/No/No
[69]	SO42−-ZrO2–CeO2	10.37	2–10/7	35 s	SG	No/No/No
[70]	ZrO2-Chitosan Composite	44.44	3–11/4–10	600 min	SG, CL	No/No/No
[71]	ZrO2-MOFs	102.4	ND/7	24 h	ST	No/No/No
[72]	ZrO2-ZnO	107.41	ND/7	24 h	HT	No/Yes/No
[73]	CeO2-ZrO2	175	ND/4	24 h	ST	No/No/No
[74]	Zr(IV)-Polypyrrole-Zr iodate	183.5	ND/6.8	Column	P	No/No/No
[75]	UiO-66-NH2 Zr metal-organic frameworks	97	ND/8	60 min	ST, ES	No/No/No
This work	Fe3O4–ZrO2	70.8	1–10/3	120 min	CP	Yes/No/Yes

SE: solvent evaporation, C: calcination, CP: co-precipitation, CA: carbonization, WI: wet impregnation, SG: sol–gel, ED: electrodeposition, CL: chemical crosslinking, ST: solvothermal, P: polymerization, ES: electrospun, ND: no data.

summarizes key properties such as adsorption capacity, synthesis method, operational pH, contact time, and scalability. This comparison highlights the advantages of Fe₃O₄–ZrO₂, particularly in terms of synthesis simplicity, scalability, and magnetic separation capabilities.

Several adsorbents in the literature exhibit higher fluoride adsorption capacities than our adsorbent. For instance, Zr(IV)-Polypirrole-Zr iodate (183.5 mg/g) [74] and CeO₂-ZrO₂ (175 mg/g) [73] surpass its performance. However, these materials require complex synthesis methods such as polymerization or solvothermal techniques, which involve long reaction times, high energy consumption, and expensive reagents. Additionally, these materials are not magnetic, making their separation from water difficult, requiring filtration or centrifugation.

Magnetic adsorbents like Fe₃O₄ offer the advantage of easy separation from water using an external magnetic field, enhancing their practicality. Among them, Fe₃O₄-activated carbon (146.2 mg/g) [64] and Fe₃O₄@MgO (98.4 mg/g) [62] have higher adsorption capacities, but their synthesis involves high-temperature treatments (650–700 °C), making them less energy-efficient and costly.

In contrast, our adsorbent demonstrates a high adsorption capacity (70.8 mg/g) over a broad pH range (optimum at 3), making it applicable to various water conditions. Furthermore, its simple and scalable co-precipitation synthesis eliminates the need for high temperatures or long reaction times, ensuring an energy-efficient and cost-effective alternative. Unlike other Fe₃O₄-based composites, our adsorbent requires no calcination, simplifying its fabrication while maintaining superior adsorption performance. Compared to Fe₃O₄-SiO₂ (5.6 mg/g) [63] and Fe₃O₄-CaCO₃ (17.95 mg/g) [65], Fe₃O₄–ZrO₂ exhibits significantly enhanced fluoride removal efficiency, confirming its effectiveness as a viable and practical adsorbent.

For real-world water treatment applications, an adsorbent must combine efficiency, cost-effectiveness, and ease of recovery. Our adsorbent successfully balances adsorption capacity, scalability, and magnetic separation, making it a highly promising candidate for fluoride removal.

The reusability of the M1 composite was evaluated over four consecutive fluoride adsorption–desorption cycles, as shown in the Supplementary Information. The adsorption capacity gradually decreased to approximately 25% (Table S1). This decline in performance may be attributed to irreversible fluoride binding at active sites or incomplete regeneration of adsorption sites. Despite the reduction, the material retained measurable adsorption capacity, indicating its potential for limited reuse in low-demand applications.

3. Materials and Methods

3.1. Chemicals and Materials

Iron(II) chloride tetrahydrate (FeCl₂·4H₂O, 99%), iron(III) chloride hexahydrate (FeCl₃·6H₂O, 98%), zirconium oxychloride octahydrate (ZrOCl₂·8H₂O, 99%), and sodium fluoride (NaF, 99%) were obtained from Sigma-Aldrich Chemical Co. (St. Louis, MO, USA). Sodium hydroxide (NaOH, 97%) and hydrochloric acid (HCl, 37%) were supplied by Merck S.A. de C.V. (Naucalpan de Juárez, Estado de México, Mexico). All reagents were of analytical grade and used without further purification. Deionized water was used throughout the experiments.

3.2. Synthesis of Fe₃O₄–ZrO₂ Nanoparticles

Fe₃O₄–ZrO₂ nanoparticles were synthesized using a modified wet chemical co-precipitation method based on the previous literature [30]. Unlike the original approach, where iron sulfates were used, this synthesis employed iron chlorides (FeCl₃ and FeCl₂) due to their higher solubility in aqueous media, allowing for more homogeneous nucleation and facilitating a better dispersion of the zirconia phase. Briefly, a 200 mL aqueous solution containing 0.5 M FeCl₃, 0.25 M FeCl₂, and ZrOCl₂ was prepared in stoichiometric proportions according to the Fe/Zr ratios of 1:1, 1:2, and 1:4, designated as M1, M2, and M3, respectively. A 6 M NaOH solution was slowly added under vigorous mechanical stirring at room temperature until the pH reached 11, ensuring the co-precipitation of Fe₃O₄ and hydrated zirconia. The resulting precipitates were aged at 90 °C for 18 h, a step that promotes phase interaction, enhances structural stability, and reduces lattice defects. The Fe₃O₄–ZrO₂ composites were then magnetically separated using a neodymium magnet, thoroughly washed with distilled water until the supernatant was clear, and dried at 90 °C for 8 h. The adsorbents in this work were manually pulverized using a mortar and pestle. The final products were stored in a desiccator until further use.

3.2.1. Characterization

X-ray diffraction (XRD) patterns were recorded using a Malvern Panalytical X-ray diffractometer (Malvern Panalytical Ltd., Model Empyrean, Worcestershire, UK) (resolution: 0.0525°) with monochromatized Cu Kα radiation (λ = 1.5406 Å). Morphological and elemental analyses were performed with a JEOL JSM-7800F (Tokyo, Japan) scanning electron microscope (SEM) equipped with an energy-dispersive X-ray spectroscopy (EDS) module (X-Max 80, Oxford Instruments, Abingdon, UK). Fourier-transform infrared (FTIR) spectroscopy was conducted using a Perkin Elmer Frontier spectrophotometer (Waltham, MA, USA) in the range of 4000–500 cm⁻¹.

3.2.2. Batch Adsorption Experiments

A standard fluoride stock solution (1000 mg/L F⁻) was prepared by dissolving sodium fluoride in deionized water and stored in the dark. Fluoride solutions with concentrations ranging from 5 to 300 mg/L were obtained by diluting the stock solution.

To determine the most effective adsorbent, 50 mg of each sample (M1, M2, and M3) was dispersed in 50 mL of a 50 mg/L fluoride solution and mixed at 40 rpm for 24 h at room temperature using an IKA Loopster Digital shaker rotator (IKA Works, Inc., Wilmington, NC, USA). Afterward, the adsorbents were separated using a magnet, and the remaining fluoride concentration in the supernatant was measured with an ion-selective electrode (HI 98402, Hanna Instruments, Padua, Italy).

The effect of pH on fluoride adsorption was evaluated following the same procedure, varying the pH from 1 to 10 while using only the M1 adsorbent.

To study the adsorption isotherm, 50 mg of the M1 adsorbent was added to 50 mL of fluoride solutions with initial concentrations ranging from 5 to 300 mg/L at pH 3. The suspensions were stirred for 2 h, after which the adsorption capacity at equilibrium (q_e) was calculated using the equation:

$$q_e={\displaystyle \frac{\left(C_o-C_e\right)V}{m}}$$

where C₀ and C_e are the initial and equilibrium fluoride concentrations (mg/L), respectively, V is the volume of the solution (mL), and m is the mass of the adsorbent (g).

To perform the kinetic analysis, 50 mg of the M1 adsorbent was dispersed in 50 mL of fluoride ion solution with a concentration of 10 mg/L, and the mixture was stirred for different periods, from 5 min to 24 h. The remaining fluoride concentration was determined as indicated above.

To assess the influence of interfering ions on fluoride adsorption, 50 mg of the M1 adsorbent was added to water samples containing 10 mg/L fluoride along with one of the following ions: ($S{O}_4^{2-}$, $C{O}_3^{2-}$, $HC{O}_3^{-}$, $P{O}_4^{3-}$, $C{l}^{-}, or N{O}_3^{-}$) at concentrations of 0.1, 1, and 10 mM. The mixtures were allowed to react for 2 h, after which the adsorbent was separated using a magnet, and the remaining fluoride concentration in the solution was measured.

4. Conclusions

The synthesized Fe₃O₄–ZrO₂ nanoparticles demonstrated high efficiency in removing fluoride from aqueous solutions, with the Fe/Zr = 1:1 composition (M1) showing the highest fluoride adsorption capacity, reaching 70.8 mg/g at pH 3. The adsorption process followed the Temkin isotherm model, indicating strong interactions between the adsorbent and fluoride ions. Additionally, the nanoparticles exhibited excellent magnetic properties, allowing for easy separation using an external magnetic field, which enhances their potential for practical applications in water treatment. The adsorption capacity was found to be pH-dependent, with the highest efficiency observed at pH 3 and a decrease in performance at higher pH values due to electrostatic repulsion and competition with hydroxyl ions. However, fluoride adsorption remained over 60% across a range of pH values from 1 to 7, indicating that the material retains its adsorptive properties under different conditions, making it adaptable for various water treatment scenarios. Competitive ion studies showed that common anions like nitrate, chloride, and sulfate had a minimal effect on fluoride removal, indicating the selective nature of the adsorbent. However, bicarbonate, carbonate, and phosphate ions reduced the adsorption capacity due to their strong interactions with the adsorption sites. A comparative analysis with other fluoride adsorbents reported in the literature (Table 4) highlights the advantages of our adsorbent over alternative materials. While certain adsorbents exhibit higher fluoride uptake, many require complex synthesis methods, including solvothermal treatments, polymerization, or high-temperature calcination (500–700 °C), which increase energy consumption and production costs. Additionally, non-magnetic materials require additional separation steps, such as filtration or centrifugation, limiting their applicability. In contrast, our adsorbent combines high adsorption capacity, simple synthesis, and magnetic separation, making it a promising candidate for scalable and cost-effective water treatment solutions. Overall, our adsorbent shows great promise for selective fluoride removal from water, offering a balance of adsorption efficiency, magnetic separation, and minimal interference from most common ions. Its simple and scalable synthesis further supports its potential for real-world applications in addressing fluoride contamination in drinking water.