Hydroxyapatite-Modified Zeolite for Fluoride Removal from Drinking Water: Adsorption Mechanism Investigation and Column Study

Abstract

This study investigates the synthesis and application of hydroxyapatite (HAp)-modified zeolite materials for efficient fluoride removal from groundwater-based drinking water. Characterization confirmed the successful incorporation of HAp onto the zeolite surface and the formation of a stable composite. EDS analysis revealed the presence of Ca and P after modification, while FTIR and XRD confirmed the structural integrity of HAp during adsorption. ZH8 exhibited the highest F-removal efficiency of 92.23% at pH 3, 30 °C, [F⁻] = 6 ppm and dose = 10 g/L. Meanwhile, HAp-modified zeolite showed high F-selectivity, and the competing ions had limited interference. The Langmuir model best described the adsorption process, suggesting monolayer adsorption with a maximum capacity of 39.38 mg/g for ZH8. The process followed pseudo-first-order kinetics, with equilibrium achieved within 4 h. Regeneration studies demonstrated that ZH8 maintained over 85% efficiency for three cycles, highlighting its reusability. Column studies validated the material’s practical applicability, with breakthrough times of up to 23 h under optimal conditions (flow rate: 8 cm³ min⁻¹, bed depth: 30 cm, feed concentration: 7.5 ppm) and a maximum yield of 99% at [F⁻] = 5 ppm with V_b = 10.8 L. The Thomas model best described the column adsorption process, indicating chemical adsorption as the dominant mechanism. These findings demonstrate the potential of HAp-modified zeolite, particularly ZH8, as an effective adsorbent for fluoride removal in real-world applications.

1. Introduction

The presence of fluoride in drinking water is a growing concern due to its potential adverse effects on human health. Numerous studies have highlighted the detrimental effects of long-term consumption of drinking water with a high fluoride concentration [1,2,3,4,5]. Dental and skeletal fluorosis are among the most commonly observed health issues associated with excessive fluoride intake through drinking water. To ensure the safety and well-being of individuals, the removal of fluoride from potable water is paramount. Fluoride levels in the North–Central Province in Sri Lanka are significantly higher than the permissible limit, especially during the dry season, and more than 1/3 of the groundwater sources exceed the permissible limit (1.0 mg/L) [6]. Additionally, recent studies suggest a synergistic effect between fluoride, cadmium and water hardness in contributing to chronic kidney disease of unknown etiology (CKDu) in Sri Lanka [7]. Moreover, small-scale drinking water treatment plants in rural Sri Lanka, particularly those using reverse osmosis (RO) as the primary membrane process, face challenges in consistently delivering safe drinking water due to seasonal variations, especially in the North, North–Central and North–Western provinces. These variations affect water quality, making it difficult to maintain acceptable standards [8].

Various methods exist for fluoride removal from drinking water, each with limitations [9]. Coagulation and precipitation are costly, produce harmful sludge and depend on interfering ions [10]. Electrocoagulation (EC) is efficient but hindered by high initial and operational costs and sludge handling [11,12]. Ion exchange relies on costly resin preparation and unsustainable regeneration processes [9,12,13]. Dialysis, though energy-efficient, is slow and limited to industrial use [9]. Electro-dialysis (ED) faces challenges with high energy consumption and capital costs [14,15]. Membrane processes like nanofiltration (NF) and reverse osmosis (RO) are effective, with NF being more cost-effective than RO, but both have problems with hazardous concentrate disposal, high capital investment and operational expenses. Adsorption is a cost-effective and widely used method for fluoride removal in drinking water, with natural materials gaining attention due to their eco-friendliness, high surface area, ion exchange capacity and chemical stability. Natural minerals like zeolite, montmorillonite, kaolinite and bentonite, often modified with rare earth metals or high-affinity substances, exhibit superior fluoride removal capabilities. These materials, combining multiple components at the nanoscale, offer an enhanced performance compared to individual components, making them promising adsorbents for water treatment [16,17,18,19,20]. In particular, zeolite, an alumino-silicate material, is known for its excellent ion exchange and adsorption properties [18]. Surface modification of zeolites with organic and inorganic materials was performed to improve their anion adsorption capacity, addressing their inherent negative surface charge [21,22,23].

The recent studies show that these surface modifications of zeolite materials are capable of removing fluoride from water sources to mitigate the fluoride contamination. In one of the latest studies, Zhao et al. investigated the removal of phosphate and fluoride using lanthanum hydroxide-modified zeolite (LMZ) [24]. The results showed that LMZ effectively adsorbed both phosphate and fluoride, though fluoride adsorption capacity decreased by 57.9% in the presence of phosphate, which was unaffected. Another study used lanthanum-modified zeolite (LMZ), synthesized from coal fly ash, for efficient fluoride removal from water [25]. The maximum adsorption capacity was 141.5 mg/g at a F/La molar ratio of 4.21. Lanthanum-modified zeolite was also used in another study performed by Lai et al., which investigated the mechanism of fluoride removal from simulated zinc sulfate solutions [26]. The results demonstrated a maximum adsorption capacity of 20.83 mg/g at 303 K and 23.04 mg/g at 313 K. A composite adsorbent was synthesized using zirconium oxychloride, chitosan and artificial zeolite for the removal of fluoride (F⁻) from water in a study performed by Chen et al. [27]. The adsorption isotherms fit the Freundlich and Langmuir models, with a maximum adsorption capacity of 10.75 mg/g at room temperature, suggesting chemisorption as the primary mechanism.

Coal fly ash was used in another study to synthesize aluminum hydroxide-coated zeolite (AHZ), and the researchers found efficient fluoride removal from wastewater [28]. The study demonstrated that AHZ has a maximum fluoride adsorption capacity of 18.12 mg/g. Over 92% of fluoride was removed within 2 h. Ebsa studied fluoride removal using clinoptilolite zeolite modified with HTAB, achieving optimal efficiency at 10 mg/L fluoride and a 5 g/L HTAB dosage after 60 min. Unmodified zeolite was ineffective, while modified zeolite required precise control of pH, temperature and runtime [29]. In a study performed by Gao et al., a micron zirconia/zeolite molecular sieve (ZrO₂-Ze) composite was synthesized to study the effect of fluoride removal in drinking water [30]. Optimal adsorption conditions were determined at an adsorbent dose of 2 g/150 mL, contact time of 8 h, pH of 6 and temperature of 25 °C, achieving a fluoride removal rate of 94.89%. Natural zeolite was modified with magnesium (II), aluminum (III) and titanium (IV) ions, in a study performed by Ma et al., as adsorbents for fluoride removal from wastewater [31]. The competitive adsorption tests demonstrated that the modified zeolites exhibited high selectivity for fluoride ions.

But in most of these studies, rare earth minerals were used and energy intensive modification routes were followed, which are not cost-effective solutions. Among the high-affinity substances, hydroxyapatite, in particular, showed enhanced, efficient removal performances of fluoride from aqueous solutions in various studies performed in recent years [32,33,34,35]. Hence, a zeolite-based, hydroxyapatite-modified adsorbent has the potential to be introduced as an innovative solution for existing fluoride contamination of groundwater in Sri Lanka. Thus, the development of a new surface-modified zeolite-based material is necessary to initiate innovation. Hence, in this study, surface modification was achieved through the simple precipitation of hydroxyapatite (HAp) onto synthetic zeolite beads. It was then characterized by various techniques to ensure the successful incorporation of HAp onto the surface. Then, the performance was evaluated by investigating the effect of pH and competing ions on the removal efficiency of fluoride. After that, the adsorption mechanism was studied in isotherm and kinetic experiments. Finally, the practical applicability was assessed in a column study, with us varying main parameters like the flow rate, feed concentration, and bed height. The results of this work show the potential for F⁻ removal and offer ideas for ensuring a safe drinking water supply.

2. Materials and Methods

2.1. Materials

Raw zeolite beads with a particle size of 5 mm were purchased from MACKLIN (Shandong Keyuan Biochemical Co., Ltd., Jinan, Shandong, China). NaF was used to prepare the fluoride stock solution (purity > 98%, MACKLIN, Shandong Keyuan Biochemical Co., Ltd., Jinan, Shandong, China). DI water was used for all the experiments. The reagents used in this study including NaH₂PO₄, NaOH, CaCl₂, FeCl₃.6H₂O (purity > 99.5%) and HCl were all analytical-grade (Sinopharm Chemical Reagent Co. Ltd., Shanghai, China).

2.2. Surface Modification of Raw Zeolite

Zeolites were dried in an oven at 150 °C to open the pore channel on the surface as pre-preparation. Surface modification of zeolite with hydroxyapatite (HAp) was performed by the simple precipitation method, which was used in a previous study [32]. Firstly, 400 mL of 0.0072 mol HAp aqueous solution was prepared in a conical flask by adding NaH₂PO₄·2H₂O and NaOH followed by adding of CaCl₂. A white-colored HAp precipitate was formed in the solution. Before adding the zeolites, the initial pH was adjusted according to the pre-determined values. We then added 10 g of pre-prepared zeolites beads to the solution with constant stirring at 25 °C. Then, the solution was agitated at 120 rpm for 4 h at 35 °C followed by aging at 25 °C for 12 h. Next, the zeolite beads were filtered and washed with DI water until a neutral pH was achieved, and the beads were dried overnight at 75 °C. Finally, the prepared zeolite beads were kept in airtight containers for future use. The prepared materials were categorized according to the initial pH (6, 7, 8, 9, 10) of the HAp solution in which the zeolite beads were added. Thus, the materials were named ZH6, ZH7, ZH8, ZH9 and ZH10. The acronym ZH stands for “zeolite” and “HAp” and the number denotes the pH value. For instance, ZH6 means that the precipitation of HAp onto zeolites surface is performed at an initial pH of 6. Raw zeolite beads were named Raw-Z.

2.3. Characterization

X-ray diffraction spectra of the synthesized materials were obtained using a D8 advance powder X-ray diffractometer (Bruker AXS, Karlsruhe, Germany) to analyze the crystalline structure. Material identification was performed from XRD patterns using MDI Jade 6 software (MDI Materials Data Inc., Livermore, CA, USA). The elemental analysis was performed by energy-dispersive X-ray spectroscopy (Quattro C, Thermo Fisher Scientific Inc., Waltham, MA, USA). The materials were then further analyzed to investigate the bonding formation using Fourier-transform infrared spectroscopy (FTIR, Nicolet iZ10 FTIR spectrometer, Thermo Fisher Scientific Inc., USA). FTIR patterns were identified using OMNIC 8.2.0.387 software (Thermo Fisher Scientific Inc., USA).

2.4. Batch Study

Unless otherwise stated, all the batch experiments were carried out using a 6 ppm fluoride solution, with a 10 g/L adsorbent dose at 30 °C. As per the recent studies, the maximum fluoride level was 6.89 ppm, with an average of 1.23 ppm and a standard deviation (STD) of 0.99 [6] in groundwater in the north–central province in Sri Lanka. Hence, 6 ppm was chosen as the standard for batch studies, which was prepared using 100 ppm fluoride stock solution. The fluoride removal efficiency of each material was tested in pH 3, 4, 5, 6, 7, 8 and 9 using a 50 mL fluoride solution. The pH was adjusted by using 1 M NaOH and 1 M HCl solutions and measured by a laboratory pH meter (FP20, Mettler-Toledo GmbH, Greifensee, Switzerland). After adding the sorbent, the solution was kept in an incubator shaker for 6 h at 120 rpm. The shaker time period and the adsorbent dosage were determined by choosing the optimum results of the preliminary experiments. The final F⁻ concentration was then measured using the F⁻ ion-specific electrode (PXSJ-216F ion meter, Shanghai Yidian Scientific Instrument Co., Ltd., Shanghai, China) by adding an appropriate amount of total ionic strength adjustment buffer solution (TISAB). Experiments for each material were performed in triplicate and the removal efficiency was calculated according to Equation (1):

$$\mathrm{R}\%={\displaystyle \frac{{\mathrm{C}}_0-{ \mathrm{C}}_{\mathrm{e}}}{{\mathrm{C}}_0}}\times 100\%$$

(1)

where R% is the removal efficiency, C₀ is the initial concentration (mg/L) and C_e is the equilibrium concentration (mg/L).

The effect of competing ions was tested using different cations (Na⁺: 200 ppm, K⁺: 25 ppm) and anions (Cl⁻: 250 ppm, Br⁻: 220 ppm, NO₂⁻: 50 ppm, NO₃⁻: 20 ppm, SO₄²⁻: 280 ppm). The selection of these ions and their initial concentrations were based on the latest results obtained in a previous investigation of the groundwater quality in the north–central province of Sri Lanka [7]. The experiment was performed using a 400 mL volume of the fluoride solution at an initial pH of 6.5. The initial pH of 6.5 was selected as the optimum value after a study of the removal efficiency for the selected material, ZH8. After adding the sorbent material, the solution was constantly stirred at 120 rpm for 6 h. We then collected 10 mL samples from 0 to 240 min at 30 min intervals. Following that, the samples were analyzed for final concentrations of the ions using an inductively coupled plasma optical emission spectrophotometer (Optima 8300, Perkin Elmer, Waltham, MA, USA) and ion chromatography (Dionex ICS-1000, Temecula, CA, USA). Out of the five materials synthesized, ZH7 and ZH8 had the better performances for fluoride removal. Thus, those materials were selected for the isotherm and kinetic study.

2.5. Isotherm Study

The concentrations of F⁻ were varied from 0 to 400 ppm to study the effect of concentration on adsorption. In each 50 mL solution, the initial pH was 6.5 and the interaction time was 6 h. After the equilibrium phase was achieved, the final concentration in the solution was measured, and the amount of fluoride adsorbed (q_e, mg/g) was calculated by Equation (2):

$${\mathrm{q}}_{\mathrm{e}}={\displaystyle \frac{{\mathrm{C}}_0-{ \mathrm{C}}_{\mathrm{e}}}{{\mathrm{C}}_{\mathrm{s}}}}$$

(2)

where C₀ is the initial concentration (mg/L), C_e is the equilibrium concentration(mg/L) and C_s is the adsorbent dosage (mg/L). The variation in q_e with C_e was then non-linearly fitted using three adsorption isotherm models: Langmuir, Freundlich and Temkin. The relevant equations for these models are given in Section S1 (Supplementary Information).

2.6. Kinetic Study

The reaction kinetics were investigated by maintaining the same laboratory conditions during the kinetic study as in the isotherm study. Samples were drawn from the solution initially at 5, 10, 15, 20 and 30 min and in 30 min intervals afterwards for up to 6 h to measure the concentration. The experiments were duplicated during the study, and the amount of fluoride adsorbed at time t (q_t, mg/g) was calculated by Equation (3):

$${\mathrm{q}}_{\mathrm{t}}={\displaystyle \frac{{\mathrm{C}}_0-{ \mathrm{C}}_{\mathrm{t}}}{{\mathrm{C}}_{\mathrm{s}}}}$$

(3)

where C₀ is the initial concentration (mg/L), C_t is the concentration at time t (mg/L) and C_s is the adsorbent dosage (mg/L). The variation in q_t with C_e was then non-linearly fitted using three adsorption kinetic models: pseudo-first-order, pseudo-second-order and Elovich. The relevant equations for these models are given in Section S2 (Supplementary Information).

2.7. Regeneration and Re-Use

The selected adsorbent from the batch studies was used for a regeneration and re-use study for up to five consecutive cycles. The experimental conditions were the same as in the batch studies (Section 2.4) during the adsorption process of each cycle. The solution was held for 6 h at 120 rpm. The removal efficiency of fluoride was as per Equation (1). After each cycle, the adsorbent was regenerated with 1 M NaOH solution followed by thorough washing with DI water until a neutral pH was achieved. Then, the adsorbents were dried overnight at 75 °C and used for the next adsorption cycle. The desorbed fluoride ions were then diluted and disposed according to the material handling and safety guidelines of the laboratory. When regenerating the adsorbents in practical scenarios in water treatment plants, the sludge-handling guidelines of national and international regulations should be strictly followed.

2.8. Column Study

Practical application of the synthesized material as a pre-treatment unit was investigated in a continuous-flow fixed-bed column study. Three main experimental parameters—feed concentration, flow rate and bed depth—were selected, which were used in other similar studies [32,36,37]. The range of fluoride was varied as 5, 7.5 and 10 ppm, the flow rate as 8, 12 and 16 cm³ min⁻¹ and the bed depth as 20, 30 and 50 cm. All the column runs were performed at 30 °C and ambient pressure. For the complete utilization of the material bed, the contaminated water was fed from the bottom to the top using a peristaltic pump. Samples were collected at 1 h intervals, and the concentration was measured. The breakthrough point (t_b) was defined as the time at which the outlet concentration reached the maximum permissible limit for fluoride in drinking water according to the Sri Lankan standard (1 ppm). The experimental values were further analyzed using three adsorption models: Thomas, Yoon–Nelson and Bohart–Adams. The details of these models are presented in Section S3 (Supplementary Information). To investigate the removal efficiency from t = 0 to t_b, the breakthrough volume (V_b) and percentage yield (Y_b %) were calculated as per Equations (4)–(8):

$${\mathrm{m}}_{\mathrm{in},{\mathrm{t}}_{\mathrm{b}}}=\underset{0}{\overset{{\mathrm{t}}_{\mathrm{b}}}{{\displaystyle \int }}}{\mathrm{C}}_0 \mathrm{Q} \mathrm{dt}$$

(4)

$${\mathrm{m}}_{\mathrm{out},{\mathrm{t}}_{\mathrm{b}}}=\underset{0}{\overset{{\mathrm{t}}_{\mathrm{b}}}{{\displaystyle \int }}}{\mathrm{C}}_{\mathrm{t}} \mathrm{Q} \mathrm{dt}$$

(5)

$${\mathrm{m}}_{\mathrm{ads},{\mathrm{t}}_{\mathrm{b}}}={ \mathrm{m}}_{\mathrm{in},{\mathrm{t}}_{\mathrm{b}}}-{\mathrm{m}}_{\mathrm{out},{\mathrm{t}}_{\mathrm{b}}}$$

(6)

$${\mathrm{V}}_{\mathrm{b}}= \mathrm{Q} \times {\mathrm{t}}_{\mathrm{b}}$$

(7)

$${\mathrm{Y}}_{\mathrm{b}}(\%)={\displaystyle \frac{{\mathrm{m}}_{\mathrm{ads},{\mathrm{t}}_{\mathrm{b}}}}{{\mathrm{m}}_{\mathrm{in},{\mathrm{t}}_{\mathrm{b}}}}}\times 100\%$$

(8)

where t_b is the breakthrough time (h), C₀ is the feed concentration (mg dm⁻³), C_t is the concentration at time t (mg dm⁻³), Q is the flow rate (cm³ min⁻¹), m_in,tb is the total mass of fluoride fed up to t_b, m_out,tb is the total mass of fluoride output up to t_b, m_ads,tb is the total mass of fluoride adsorbed up to t_b, V_b is the breakthrough volume (L) and Y_b (%) is the percentage yield up to t_b.

3. Results and Discussion

3.1. Characterization of the Modified Zeolite

3.1.1. EDS

The EDS spectrum and elemental mapping in Figure 1a show the elemental analysis of the Raw-Z material, which was commercially purchased. According to the spectra, the high intensities at 1.4867, 1.7399 and 5.249 keV values indicate the elemental composition of Al, Si and O, respectively, which are the main elements of the zeolite material. Other than that, an additional peak is observed at 1.0409 keV, which corresponds to Na. It can be seen from the mapping that these elements are uniformly distributed along the surface of the material. Figure 1b shows the EDS spectrum for the material after modification. All five materials synthesized (ZH6, ZH7, ZH8, ZH9, ZH10) showed the same EDS spectra and elemental mapping. After surface modification, two corresponding peaks for the element Ca and one peak for P were found at 0.341, 3.69 and 2.01 keV values, which is a clear indication of the successful incorporation of those elements into the surface during the chemical precipitation process of HAp [Ca₁₀(PO₄)₆(OH)₂]. Also, the peak values and the distribution of those elements are consistent with the spectra. The corresponding peak at 676.8 keV in Figure 1c after adsorption confirms the favorable fluoride uptake by the active sites during the adsorption process.

3.1.2. FTIR

In order to study the functional groups in the synthesized materials, FTIR spectra were used, and the results are illustrated Figure 2 and Figure S1. A broad peak in the range of 3685–3028 cm⁻¹ was discovered, corresponding to stretching vibration of the O-H group in raw Zeolite-A, and another broad peak at 1651 cm⁻¹ was also discovered, corresponding to bending vibration of the same group [38,39]. The Si-O and Al-O asymmetric stretching modes were also detected at 1002 cm⁻¹. After modification, a sharp peak was obtained at 1035 cm⁻¹, corresponding to stretching vibration of phosphate (PO₄³⁻) groups, which is an indication of the effective precipitation of HAp on the surface of the material. A small peak was observed after adsorption at 756 cm⁻¹, which corresponds to the formation of hydroxyl bonding of fluorohydroxyapatite [32]. That may be an indication of the adsorption process, which might be described as a ligand-exchange reaction [39]. The adsorption process did not alter the bonding of the phosphate group, which can be attributed as demonstrating satisfactory formation and good stability with the raw material.

3.1.3. XRD

The X-ray diffraction (XRD) patterns for the synthetic zeolite and its surface-modified forms, ZH (zeolite modified with hydroxyapatite) and ZH-8 after adsorption (ZH-8 after ads.), are presented in Figure 3 and patterns for other materials are presented Figure S2. The patterns reveal the crystalline nature of zeolite, with its characteristic peaks prominently displayed at specific 2θ values. The peaks at 7.33°, 10.326°, 12.63°, 16.29°, 21.87°, 24.2°, 27.34°, 30.18° and 34.43° can be attributed to the crystalline structure of the zeolite framework, consistent with the previously reported literature for similar types of synthetic zeolites (ZeoliteA-syn, JCPDS 82-2400). These diffraction peaks are indicative of the well-ordered crystalline structure inherent to synthetic zeolite, corresponding to its microporous framework.

Following surface modification of zeolite with hydroxyapatite (HAp), the XRD pattern for ZH shows additional peaks at 31.9° and 49.8°, which are characteristic of HAp, indicating successful incorporation of the hydroxyapatite phase onto the zeolite surface (JCPDS 25-0166). The presence of these peaks suggests the formation of a composite material, where HAp is deposited on the zeolite surface, thereby enhancing the material’s functionality for potential applications such as adsorption. Interestingly, the ZH-modified samples showed a similar pattern to ZH, suggesting that substitution of HAp did not make any changes in the crystal structure of Zeolite-A. Since fluoride ions replace the -OH groups in HAp, the crystal structure after adsorption is unaltered; hence, the XRD patterns remained identical. This result signifies the good stability of the synthesized adsorbents after adsorption [32].

3.2. Adsorption Performance

3.2.1. Effect of pH on Removal Efficiency

Figure 4 illustrates the effect of pH on the removal efficiency (R%) of fluoride from water using the five different materials synthesized, which showed a general trend of decreasing efficiency with an increasing pH across all materials. The removal efficiency of the raw material was significantly enhanced by the incorporation of hydroxyapatite. The pH range was from 3 to 9, covering both acidic and basic conditions beyond the range of permissible pH for drinking water. At a low pH of 3, the removal efficiency can be observed to be the highest for all materials. This trend indicates that fluoride removal is more effective under acidic conditions, which can be ascribed to the increased protonation of active sites at a lower pH, leading to stronger attraction of the negatively charged fluoride ions [39,40]. ZH8 exhibited the highest efficiency of 92.23% while ZH6, ZH9 and ZH10 showed a lower efficiency at this pH. It seems the hydroxyapatite bond formation with the zeolite surface is slightly lower at pH 6, 9 and 10 whereas the optimum pH for modified hydroxyapatite is pH 7 or 8. The removal efficiency of ZH8 remains somewhat stable during the drinking water pH range around the value of 67%.

3.2.2. Effect of Competing Ions

When the adsorption process happens in natural water sources, the adsorption surface not only attracts the fluoride ions but also the co-existing anions and cations. Thus, there may be a competition with fluoride ions for sorption onto the surface by some of those ions. Figure 5 shows the variation in the concentrations of fluoride, chloride and bromide ions during the adsorption process. As can be seen, the final concentration of fluoride was higher than that in the previous experiments, and the concentration of chloride ions was also reduced with time. This can be correlated with the fact that chloride has the same electronegativity and valence as in fluoride ions and a relatively lower ionic radius compared to other ions, hence interfering with the adsorption process. The concentrations of other ions remained constant during the whole time, as can be seen from Figure S3. This result is an indication that the material ZH8 has a sorption selectivity for fluoride and that the co-existing ions cause limited interference with the adsorption process.

3.3. Adsorption Mechanism

3.3.1. Isotherm Study

Figure 6 shows the relationship between the equilibrium concentration of the adsorbate (C_e) and the amount adsorbed per unit mass of adsorbent (q_e). The Langmuir isotherm model, depicted by the red solid line, demonstrates a strong fit with the experimental data for both materials, implying that the adsorbent–adsorbate formed a monolayer and uniform active site with similar sorption energy [32,40]. The maximum adsorption capacity (Q_m) was determined to be 34.68 mg/g with a Langmuir constant (K_L) of 0.807 L/mg for ZH7 and 39.38 mg/g with a Langmuir constant (K_L) of 0.692 L/mg for ZH8. A comparison of the adsorption values with other similar adsorbents in previous studies is given in Table S4. Out of the 15 adsorbents listed, 11 adsorbents have a lower adsorption value than the ZH8 material. The adsorption intensity values (n) in the Freundlich model for both the materials are greater than 1, which implies a favorable adsorption process for fluoride removal [39]. In the Temkin isotherm model, the heats of sorption (B_T) are positive for both materials, at 6.909 and 6.805 J/mg for ZH7 and ZH8, respectively, indicating an exothermic reaction, and since the values are greater than 0.385 J/mg, adsorption occurs as chemisorption [41,42]. The relevant parameters for the models are provided in Table S1; although all models provide a good fit, the Langmuir model is superior, highlighting the homogeneity of the adsorption sites and that the monolayer adsorption process is the dominant mechanism in this system. Amongst the two materials, ZH8 shows the better adsorption performance.

3.3.2. Kinetic Study

The fluoride ion removal ability was studied in a batch kinetic study of two materials, and the results are depicted in Figure 7. The contact time was up to 6 h. During the first 15 min, the adsorption rate was rapid, due to the abundance of active sites on the surface, where more than 50% of the equilibrium adsorption capacity was achieved, which then gradually rose during the next 210 min. During this slow phase from 15 min to 210 min, the adsorption rate become sluggish due to the occupation of numerous energy sites, resulting in a reduction in vacant active sites [39]. The equilibrium phase was achieved after 4 h, where the concentration remained almost unchanged. The fitting results are depicted in Table S2 for both the materials, which best fit with the pseudo-first-order model, indicating the involvement of chemical adsorption [39,43]. Among the two materials, ZH8 shows a higher adsorption rate, implying it has more active sites and relatively larger potential surface area. Also, it has the higher correlation coefficient value, from which it can be concluded that it is the best material in terms of adsorption kinetics.

3.4. Practical Application Study

3.4.1. Regeneration and Re-Use

The results of our regeneration and re-use study are illustrated in Figure 8. During the first three cycles, the efficiency remained above 85%, suggesting the effective desorption of fluoride and replacement of the -OH functional groups in HAp [32]. In the fourth and fifth cycles, the efficiency is reduced below 80%. This can be attributed to the fact that there may be irrevocable active adsorption sites, which cannot replace the fluoride ions with -OH groups. However, the efficiency is still comparatively high for up to five cycles. This suggests that the synthesized material has potential practical applicability for use for commercial purposes. The adsorbents used should be disposed with at an authorized landfill site for strict compliance with the guidelines for solid waste management of national and international institutions. According to recent studies, the zeolite adsorbents used have the potential to be recycled and used for the fabrication of cementitious materials for construction purposes [44,45,46], or co-composted with winery waste for soil management [47,48]. Before their use for such purposes, stabilization of the contaminants through thermal treatment should be applied as a safety measure to reduce the risk of leaching into the environment.

3.4.2. Column Study

The breakthrough curve in Figure 9a displays the effects of varying flow rates on the contaminant breakthrough time. The breakthrough times for flow rates of 8, 12 and 16 cm³ min⁻¹ were 23, 10.5 and 4.5 h, respectively. The trends indicate that, at increased flow rates, breakthrough occurred earlier, with a steeper curve slope, suggesting a shorter contact time between the material and the contaminated water. Further, when the flow is accelerated, it reduces the residence time, leaving insufficient time to achieve the equilibrium, limiting the efficiency of the adsorption process [36]. Meanwhile, at lower flow rates, breakthrough times increase, producing a more gradual curve. That pattern can be attributed to the fact that lower flow extended residence times within the column, allowing more time for adsorption interactions, hence providing sufficient time to reach the equilibrium [32]. Thus, optimizing the flow rate would be essential to balance throughput with the contaminant removal efficiency. Slower flow rates may be beneficial for scenarios demanding high contaminant removal, while faster rates might suit cases where higher processing volumes are needed for water with lower contaminant levels.

The impact of varying bed depths on the breakthrough time is revealed in Figure 9b. The breakthrough times were observed to be 4.5, 10.5 and 16.5 h for bed depths of 20, 30 and 50 cm. Higher bed depths yield a more gradual breakthrough curve, indicating extended contaminant retention and a prolonged saturation time compared to shorter bed depths, where breakthrough is faster and the curve steeper. An increased bed depth enhanced adsorption efficiency by providing a greater volume of adsorbent, thus increasing the number of active adsorption sites available for interaction with contaminants [32]. Moreover, a greater bed depth provided a larger adsorption interface, delaying contaminant saturation and increasing the adsorption capacity. These results suggest that using deeper beds could reduce the frequency of adsorbent replacement or regeneration, ultimately improving the suitability for real field applications.

The breakthrough curves for the effect of the feed concentration are presented in Figure 9c. The breakthrough times were 15, 10.5 and 6.5 h for feed concentrations of 5, 7.5 and 10 ppm, respectively. The breakthrough curves for higher concentrations are steeper compared to those for lower concentrations, proving that saturation is reached more quickly when the adsorbent is exposed to higher contaminant levels. This finding highlights that the adsorption active sites on the column’s adsorbent material are occupied more quickly, leading to earlier breakthrough. This can be explained by the higher driving force produced with higher concentration levels, which reduces the breakthrough time. Also, when the concentration is high, more fluoride ions are introduced to the surface of the material per unit time inside the column, which accelerates the adsorption process, hence leading saturation to occur more quickly than at lower concentration levels [36].

The experimental parameter values for the three adsorption models are presented in

Table 1. Mathematical parameters of the Thomas, Yoon–Nelson and Bohart–Adams models.

Parameter	Thomas Model			Yoon–Nelson Model			Bohart–Adams Model
	${\mathit{k}}_{\mathit{T}\mathit{H}}$	${\mathit{q}}_0$	R2	${\mathit{k}}_{\mathit{Y}\mathit{N}}$	$\mathit{\tau }$	R2	${\mathit{k}}_{\mathit{B}\mathit{A}}$	${\mathit{N}}_0$	R2
Bed Depth (cm)
20	0.074	0.94	0.998	0.539	7.98	0.995	0.029	198.7	0.985
30	0.042	3.29	0.999	0.316	16.48	0.995	0.042	185.4	0.989
50	0.037	6.96	0.996	0.280	23.22	0.995	0.061	138.6	0.985
Flow Rate (cm3 min−1)
8	0.020	5.84	0.998	0.230	24.81	0.988	0.055	124.5	0.985
12	0.042	3.29	0.999	0.316	16.48	0.999	0.042	185.4	0.988
16	0.081	1.00	0.988	0.613	7.53	0.988	0.032	218.2	0.985
Feed Concentration (mg dm−3)
5	0.031	6.43	0.998	0.250	23.0	0.998	0.025	218.6	0.985
7.5	0.042	3.29	0.999	0.316	16.48	0.998	0.042	185.4	0.989
10	0.051	1.32	0.995	0.572	9.799	0.995	0.057	146.9	0.995

. The highest correlation coefficient (R²) is reported for the Thomas model, which is greater than 0.995. This indicates that the adsorption process obeys Langmuir adsorption kinetics, which matches the previously performed isotherm studies. Moreover, according to the other assumptions based on which the Thomas model was developed, plug flow occurs in the bed with no axial dispersion, and the rate of adsorption is determined by chemical effects rather than diffusion [37]. The relationship between k_TH and q₀ was inversely proportional, and the highest uptake capacity of 6.96 mg g⁻¹ was found when the bed depth was at its maximum. This value is higher than one from the previous literature, as presented in Table S5 [49]. The rate constant k_BA and N₀ values in the Bohart–Adams model was also inversely proportional, which indicates that the adsorption process is controlled by the external mass transfer over the initial phase of adsorption [36].

The results of the breakthrough volume (V_b) and percentage yield (Y_b%) are presented in

Table 2. Parameters related to the removal efficiency of fluoride up to the breakthrough time (t_b).

	Variation with Bed Depth (cm)			Variation with Flow Rate (cm3 min−1)			Variation with Feed Concentration (ppm)
	20	30	50	8	12	16	5	7.5	10
$t_b$	4.5	10.5	16.5	23	10.5	4.5	15	10.5	6.5
$m_{in,t_b}$	24.3	56.7	89.1	82.8	56.7	32.4	54	56.7	35.1
$m_{ads,t_b}$	23.02	54.37	86.4	64.99	54.37	30.81	53.87	54.37	23.65
$V_b$	3.24	7.56	11.88	11.04	7.56	4.32	10.80	7.56	4.68
$Y_b(\%)$	94.73	95.88	96.94	78.48	95.88	95.1	99.77	95.88	67.37

. Since the best-suited model was the Thomas model, its equation with relevant k_TH and q₀ parameter values for each experimental condition was used for the C_t/C₀ term to calculate m_ads,tb. The yield exceeded 94% in five out of seven experimental conditions. The yields were as high as 99% and 95% when the feed concentrations were 5 and 7 ppm, which shows the greater performance even at the maximum polluted groundwater levels in Sri Lanka. These results of the column study show the potential of the ZH8 material for fluoride removal from groundwater in practical applications.

4. Conclusions

In this study, hydroxyapatite (HAP) was successfully incorporated onto the surface of commercially available zeolite beads (5 mm). The FTIR and XRD spectra showed that the HAp structure on the surface of the material was not altered during the adsorption process, which indicated good stability with the raw material. The removal efficiency of ZH8 remained somewhat stable during the drinking water pH range around the value of 67%. The co-existing ions had limited interference during adsorption, which showed the sorption selectivity for fluoride of the material ZH8. The Langmuir model was best fitted (R² > 0.98), and the maximum adsorption capacity (Q_m) was determined to be 34.68 mg/g with a Langmuir constant (K_L) of 0.807 L/mg for ZH7, while it was 39.38 mg/g with a Langmuir constant (K_L) of 0.692 L/mg for ZH8. Both ZH7 and ZH8 materials fitted with the pseudo-first-order model in the kinetic study, which further indicated the involvement of chemical adsorption. Out of the five cycles run during re-use, the efficiency remained above 85% during the first three cycles, suggesting the effective desorption of fluoride and replacement of the -OH functional groups in HAp. Out of the three breakthrough models, the Thomas model was best fitted, which obeys Langmuir adsorption kinetics, and the greater R² value (>0.99) indicated that the rate of adsorption is determined by chemical effects rather than diffusion. The highest yield was observed to be 99% at a feed concentration of 5 ppm, and the breakthrough volume was 10.8 L, which indicates the potential for practical application even at high fluoride contamination levels. In conclusion, the surface-modified ZH8 material might be used as a solution for the existing fluoride contamination problems in the groundwater of rural areas in Sri Lanka, with it used in an adsorption pre-treatment column for the existing membrane treatment train.