Defluoridation efficiency of animal bone char and coconut shell charcoal: a comparative adsorption isotherm and kinetic study Open Access

Life on Earth depends on water, and having access to clean drinking water is critical to human health and wellbeing. However, high fluoride contamination of groundwater supplies poses serious health dangers to the public in many places (Srivastava & Flora 2020). Extended exposure to high fluoride concentrations can cause skeletal and dental fluorosis, which is characterized by bone abnormalities, intense pain, and mottling and discoloration of tooth enamel (Onipe et al. 2020). It is estimated that 884 million people worldwide lack access to safe drinking water, with developing countries bearing a disproportionate share of this burden. Groundwater is the main source of drinking water for approximately 300 million people in Africa (Onipe et al. 2020). However, the quality of these resources is frequently harmed by a variety of pollutants, arising from natural sources that has led to an elevated level of fluoride, arsenic, other chemicals (Onipe et al. 2020; Ligate et al. 2022).

Groundwater contamination, especially with fluoride has become a major concern in many areas, including parts of West Africa and the East African Rift Valley countries. In addition to posing health dangers, these pollutants fuel the ongoing fight toward achieving the Sustainable Development Goal of the UN, which calls for universal access to clean water and sanitary facilities (Ang & Mohammad 2020). Tanzania is one of the countries facing the problem of elevated groundwater where high fluoride levels of about 30% of its water sources make them unfit for domestic use (Ligate et al. 2021). A maximum fluoride concentration of 1.5 mg/L is recommended by the World Health Organization (WHO) for drinking water, whereas Tanzania allows a higher limit of 4 mg/L in areas with no alternative water sources (Ligate et al. 2021). This higher threshold raises concerns about potential long-term health effects on affected communities.

Numerous techniques for defluoridation have been investigated, including solvent extraction, membrane filtration, chemical precipitation, reverse osmosis, and electrocoagulation (Nyangi et al. 2020; Masindi & Foteinis 2021). These methods, however, frequently prove to be ineffective, costly, or ecologically unfriendly, producing significant volumes of hazardous chemical sludge that present disposal difficulties. For example, the highest fluoride removal effectiveness of chemical precipitation methods utilizing lime and alum is about 60–70%, while reverse osmosis systems, while very effective (removing 90–95% of the fluoride), are expensive and require frequent membrane replacement (Li et al. 2018; Bahrodin et al. 2021; El Messaoudi et al. 2024). The practical applicability of solvent extraction and electroplating processes are limited by their high operating costs and potential environmental dangers. Additionally, membrane filtration techniques such as electrodialysis and nanofiltration can remove fluoride with up to 98% effectiveness, but they produce concentrated brine streams that need additional treatment and are prone to fouling problems (Moran Ayala et al. 2018). There is a pressing need for cost-effective, sustainable, and locally adaptable solutions to address fluoride contamination in drinking water sources.

The adsorption process has proven to be an effective method for fluoride removal from groundwater, relying on the binding of fluoride ions to the surface of solid materials. Recent studies have highlighted the efficiency of various adsorbents for this purpose. Activated alumina is one of the most extensively used materials, achieving fluoride removal efficiencies of 90–95%, with adsorption capacities ranging from 1.1 to 2.2 mg/g depending on pH and initial fluoride concentration (Alhassan et al. 2021). Other bio-based adsorbents, such as tea waste and neem leaves, have also demonstrated removal efficiencies of 80–85% under optimized conditions. Among inorganic adsorbents, calcium-doped hydroxyapatite has shown an adsorption capacity of up to 20 mg/g, making it highly effective for fluoride remediation (Wang et al. 2024). Additionally, metal oxide nanoparticles, such as iron oxide, have exhibited exceptional fluoride removal efficiencies of up to 97% due to their high surface area and strong reactivity with fluoride ions (Alhassan et al. 2021). These findings highlight the versatility and effectiveness of adsorption in addressing fluoride contamination, offering environmentally sustainable and cost-effective solutions.

For the removal of fluoride, using natural adsorbents such coconut shells (CSs) and animal bones offer a viable substitute (Rego et al. 2021). The bone char prepared from cattle bones (CBs), has shown promise in water defluoridation by means of ion exchange adsorption between fluoride and the apatite found in bones (Shahid et al. 2020). Similarly, the adsorptive qualities of coconut charcoal which is made from CSs have been investigated for use in water treatment applications (Bhamare et al. 2022). The CBs are abundant materials in Tanzania as it was estimated that 479,071 tons are discarded daily, whereas CSs are locally available in many areas along the coast of Indian Ocean with an estimated 1,200 tons discarded daily to the environment per day (Kibona et al. 2022).

The disposal of CSs presents a significant environmental concern, particularly due to the large volumes generated in coconut-producing regions, where they contribute to landfills, increase organic waste, and can lead to methane emissions during decomposition, a potent greenhouse gas (Araga et al. 2019). Additionally, improper disposal can result in air pollution when shells are burned as waste. Recycling CSs and CBs into valuable products, such as activated carbon or biochar for water treatment, not only reduces these environmental impacts but also minimizes deforestation pressures by providing a sustainable alternative to wood-based materials (Nunes et al. 2020). Moreover, utilizing CSs for such purposes promotes local economic development by creating new revenue streams for farmers and small-scale industries through innovative, eco-friendly solutions. This approach supports the circular economy and reduces environmental degradation.

This study addresses a critical gap in the existing literature by providing a comparative analysis of animal bone char and CS charcoal as natural adsorbents for fluoride removal. While previous studies have explored these materials separately (Dar & Kurella 2023; El Messaoudi et al. 2024), this research uniquely evaluates their effectiveness under identical conditions, highlighting their potential as sustainable, cost-effective solutions for rural communities. The findings aim to inform local defluoridation strategies, contributing to global efforts to ensure safe drinking water access. Closing this gap may result in the creation of locally appropriate and reasonably priced defluoridation methods, especially in rural areas with limited access to cutting edge water treatment technologies. Therefore, this study aimed at

• i To evaluate the effectiveness of activated carbon derived from CBs and CSs in removing fluoride from water.

• ii To optimize the adsorption conditions (dosage, contact time, pH, and initial fluoride concentration) using Response Surface Methodology (RSM).

• iii To evaluate the adsorption capacities of animal bone char and CS charcoal adsorbents under identical experimental conditions for defluoridation.

Three grab samples were collected from three different boreholes in Nduruma Ward, Arumeru District, Arusha Region, Tanzania (coordinates: 3.491111°S, 36.8812222°E). Pre-cleaned 2-L polypropylene bottles were used for sample collection due to their chemical inertness, which prevents leaching or contamination of the groundwater samples. Before collection, the boreholes were flushed for approximately 5 min to remove stagnant water and obtain representative samples of fresh groundwater. The samples were then immediately preserved at 4 °C in sampling kits to minimize microbial activity and chemical changes, ensuring the integrity of the water's physicochemical properties. The samples were then immediately transported to the Water Institute laboratory in Dar es Salaam for further analysis.

The water samples were analyzed for pH, electrical conductivity, turbidity, fluoride, carbonates, and sodium following APHA Standard Methods (APHA 2005). The pH and fluoride concentrations were measured using a pH/ISE meter (H1 98402) with a fluoride ion-selective electrode (Method 4500-FC), ensuring precise detection of hydrogen ion activity and fluoride levels. Turbidity was determined using a calibrated Turbidimeter (Method 2130 B), which measures water cloudiness by detecting light scattering. Electrical conductivity was measured using a conductivity meter (Method 2510 B), which assesses the ability of water to conduct electric current. Carbonate concentrations were determined through acid–base titration (Method 2320 B), while sodium levels were quantified using flame photometry (Method 3500-Na B) or ion-selective titration, depending on the specific analytical requirements. All procedures adhered to the APHA standard methods for water and wastewater analysis to ensure the accuracy and reliability of the results (Federation and Association 2005).

The CSs were collected from local fruit markets in Dar es Salaam, Tanzania, while the CBs were collected from various butchers in Dar es Salaam. The solutions, reagents of sodium hydroxide and sulfuric acid were analytical grade, and apparatus used during sample analysis and calibration were analytical grade and acquired from Tecno Net Scientific Ltd, a chemical supplier in Dar es Salaam. The simulated water sample containing the desired levels of F^– was prepared by adding 2.21 g of sodium fluoride to 1 L of distilled water in a flask. Furthermore, the solutions of sodium hydroxide (0.1 M) and sulfuric acid (0.1 M) used in adjusting pH were also prepared using distilled water. All solutions were prepared with double distilled water produced by the Water Institute's laboratory.

The chemical composition of the adsorbents (CB and CS) was determined using X-ray fluorescence spectroscopy (XRF) (model: Bruker AXS S8). XRF analysis was performed to quantify the elemental composition of the adsorbents, focusing on key elements such as CaO and P₂O₅ in CB and CaO and K₂O in CS. The elemental changes before and after fluoride adsorption were analyzed to identify the mechanisms involved in fluoride binding. The crystallographic structure of the adsorbents was examined using X-ray diffraction (XRD) (model: Bruker AXS D2) with Cu Kα radiation (λ = 1.5406 Å). XRD patterns were recorded over a 2θ range of 10°–80° to identify crystalline phases in CB and the amorphous nature of CS. For CB and CS characteristic hydroxyapatite peaks were evaluated before adsorption, and post-adsorption changes were analyzed to confirm the adsorption reaction.

To validate the quadratic model, the ANOVA and regression were used as revealed by indicators, such as the F-value, P-value R² and adjusted R², to obtain the least square fit and interaction between variables and response for all the data reported in Table 2. The 2D plots on the effects of individual factors were plotted using Origin Pro 9. The three-dimensional (3D) plots that were used for explaining the combined effect of parameters on the reduction of TB and COD were generated using Design Expert 11. The statistical significance of the model and the factors involved were examined using F-value and P-value at 95% confidence level.

The proximate analysis data for activated carbon derived from CB and CS at varying temperatures (300, 500, and 700 °C) reveals expected trends based on the thermal decomposition and carbonization processes (Table 2). As the temperature increases, the MC and VM decreased, due to the progressive drying and devolatilization of the precursor materials. Conversely, the AC and FC percentages increased, with the former attributed to the concentration of inorganic minerals and the latter resulting from the enrichment of the carbonaceous residue. For CB, the maximum FC of 70.8% was observed at 500 °C, while CS showed its highest FC of 75% at 700 °C, indicating optimal carbonization at this temperature. Higher activation temperatures for CS (700 °C) promote more extensive carbonization and graphitization, leading to a higher carbon content (88.2%) compared to CB (82.5% at 500 °C). Conversely, CB displayed higher AC (8.4%) than CS (6.3%), likely due to the mineral components in bones. The decrease in MC (from 8.2 to 3.7% for CB, and 11.4 to 3.6% for CS) and VM (from 22.6 to 10.2% for CB, and 28.7 to 9.4% for CS) with increasing temperature are consistent with the devolatilization and thermal decomposition of the precursor materials (Ebelegi et al. 2022).

The XRF analysis of bone char (CB) and CS reveals notable differences in their elemental composition, reflecting their respective structural and chemical characteristics (the data are provided in Supplementary material). CB demonstrates a high content of calcium oxide (CaO, 43.0%) and phosphorus pentoxide (P₂O₅, 51.6%), which are key indicators of the presence of hydroxyapatite (Singh et al. 2020). This composition is well-suited for fluoride removal through ion exchange and precipitation mechanisms. In contrast, CS contains lower CaO (28.7%) but a significant proportion of potassium oxide (K₂O, 31.5%), which indicates the potential for surface complexation and electrostatic interactions during fluoride adsorption (Araga et al. 2019).

Post-adsorption XRF analysis highlights critical compositional shifts in both adsorbents. In CB, CaO increases to 46.8%, while P₂O₅ decreases to 48.8%, signifying the formation of fluorapatite (Ca₁₀(PO₄)₆F₂). This transformation confirms that fluoride ions replace hydroxyl ions in the hydroxyapatite lattice. For CS, a slight increase in CaO (to 29.8%) and K₂O (to 31.8%) indicates surface adsorption of fluoride ions, likely mediated by metal oxides (Shahid et al. 2020). These findings align with previous studies that demonstrated the critical role of CaO and K₂O in enhancing fluoride adsorption capacity (Araga et al. 2019).

Significant structural changes are observed post-adsorption for both adsorbents. In CB, the intensities of hydroxyapatite peaks increase (from ∼120 to ∼160 units at 2θ ≈ 32°–34°), reflecting its transformation to fluorapatite. This result validates the ion exchange mechanism, where fluoride ions replace hydroxyl ions in the crystal lattice. For CS, the broad amorphous carbon peak at 2θ ≈ 22° decreases in intensity (from ∼170 to ∼105 units) and broadens slightly, indicating fluoride incorporation into its carbonaceous matrix. These transformations are consistent with findings by Shahid et al. (2020) and Araga et al. (2019) who highlighted similar structural modifications during fluoride adsorption.

The data demonstrates that activated carbon derived from CB effectively adsorbs fluoride from aqueous solutions (Table 3). Fluoride removal efficiency ranged from 48 to 98%, with the highest efficiency achieved at a dose of 5.25 g, contact time of 65 min, pH 6.5, and initial fluoride concentration of 6 mg/L. Generally, higher doses improved adsorption performance, as evidenced by the comparison between 0.5 and 10 g doses. Longer contact times also enhance efficiency, demonstrated by an increase from 50 to 54% removal when extending time from 10 to 120 min at 10 g dose. The enhanced removal efficiency is attributable to the increased availability of active adsorption sites and sufficient time for adsorption to reach equilibrium (Pourhakkak et al. 2021). Additionally, the data points at pH 3 (S/N 3, 14, 20, and 26) exhibit lower removal efficiencies, which aligns with the expected trend of reduced adsorption at highly acidic conditions due to changes in surface charges and fluoride speciation. Higher initial fluoride concentrations (10 mg/L) in S/N 10, 15) generally result in lower removal efficiencies due to the saturation of available active sites (El Messaoudi et al. 2024).

The results showed that activated carbon derived from CS effectively removes fluoride, with removal rates varying between 42 and 95%. The highest removal (95%) occurs in experiment 11, using 5.25 g of dose, 120 min of contact time, pH 6.5, and 2 mg/L of initial concentration. The lowest removal (42%) is observed in experiment 18, with 10 g dose, 65 min, pH 6.5, and 10 mg/L concentration. This suggests that CS performs better at lower fluoride concentrations. Dose and contact time also influence performance, as seen in experiments 6 and 1, where increasing dose from 0.5 to 10 g at 120 min improves removal from 52 to 60%, similar trend as in CB. Notably, CB consistently exhibited slightly higher removal efficiencies compared to CS under similar experimental conditions, as evident from data points like S/N 1 (CB: 54%, CS: 60%), S/N 4 (CB: 98%, CS: 92%), and S/N 8 (CB: 52%, CS: 84%). This may be due to higher density of active adsorption sites of CB or more favorable surface properties for fluoride adsorption. The agreement between predicted and actual removal efficiencies for both adsorbents indicate the reliability of the predictive models.

The Analysis of Variance (ANOVA) presents the results of a factorial experimental design to investigate the effects of various factors on the removal of fluoride from water using CS) and CB as adsorbents (Table 4). For the CS model, the highly significant Model F-value of 27.36 (p-value < 0.0001) indicates that the model is statistically significant and can adequately explain the variation in the CS response. The factor D-Conc, exhibits the highest F-value of 166.59 (p-value < 0.0001), suggesting a substantial influence on the CS response. Additionally, the interactive effect AC and the quadratic term C² are also significant contributors to the model (F-value > 1, p-value < 0.05). The non-significant Lack of Fit (p-value = 0.065) confirms that the model fits the data adequately. For the CB model, the Model F-value of 129.53 and p-value < 0.0001 and the non-significant Lack of Fit (p-value = 0.3465) confirms the model's adequate fit to the data. Several factors exhibit significant effects, including dose, pH, concentration, and the interactive effect BC (F-value > 1, p-value < 0.05). The high R-squared values of 0.92 and 0.98 for the CS and CB models, respectively, indicate a good fit of the models to the experimental data, suggesting that the models can effectively predict the removal of fluoride from water using CB and CS as adsorbents (Chelladurai et al. 2021).

Figure 3(b) shows the effect of contact time on fluoride removal efficiency. Both adsorbents experienced a rapid initial increase in removal efficiency within the first 60–80 min, which is attributed to the abundance of vacant active sites on fresh adsorbents, promoting rapid adsorption kinetics (Scheufele et al. 2020). After this period, the removal rate slowed significantly as the process approached equilibrium. This could be due to saturation of active sites, formation of a boundary layer around the adsorbent particles, and increased difficulty for fluoride ions to diffuse to the remaining active sites (Wang & Guo 2023). CS reached equilibrium faster than CB, suggesting more favorable adsorption kinetics for CS compared to CB.

The influence of initial pH on fluoride removal efficiency is indicated by Figure 4(c). Both CB and CS exhibited a bell-shaped curve, with maximum removal occurring around pH 5–7. This behavior can be explained by the pH-dependent surface charge of the adsorbents and the speciation of fluoride ions in solution (Dotto & McKay 2020). At low pH < 5, the adsorbent surfaces may become positively charged, repelling the negatively charged fluoride ions and leading to lower removal efficiency. At high (pH > 8), the presence of hydroxide ions can lead to the formation of weakly adsorbing fluoride complexes, reducing the availability of free fluoride ions for adsorption (Dehghani et al. 2018). The slightly higher maximum removal efficiency of CB compared to CS within the optimal pH range further revealed the higher proportion of active sites or functional groups such as indicated by the chemical composition of the materials (Jemal et al. 2019).

Figure 4(d) illustrates the effect of initial fluoride concentration on removal efficiency. CB demonstrated superior performance, with its efficiency rising from ∼75% at 2 mg/L to a peak of ∼95% at 6–8 mg/L, followed by a slight decline at higher concentrations. This suggests CB has optimal adsorption capacity in the mid-range concentrations. In contrast, the removal efficiency of CS decreased with increasing initial fluoride concentration, at lower concentrations (<4 mg/L) there is highest removal efficiency of 98% that decreased to <60% at 10 mg/L. This indicates CS may become saturated more quickly or have limited adsorption sites compared to CB. The higher efficiency of CB can be attributed to its higher calcium content, which enhances fluoride adsorption through ion exchange mechanisms (Amit et al. 2011). The CB sustained high efficiency across a broader range makes it potentially more suitable for fluoride removal applications, especially at moderate to high fluoride concentrations, while CS may be more effective for low-concentration scenarios.

Figure 5(c) and 5(d) show the interaction between pH and initial fluoride concentration. At low pH (3–4) and high fluoride concentrations (8–10 mg/L), CB achieved optimal removal (∼98%), with decreasing performance at higher pH values. Conversely, CS performed optimally in the mid-range pH (5–7) and higher concentrations, though its efficiency dropped sharply at extreme pH levels. CB performed better in acidic environments, while CS exhibited greater versatility across typical natural water pH ranges.

Figure 5(e) and 5(f) presents the effect of pH and treatment time. CB shows a maximum removal efficiency of about 100% at pH 6–8 and contact times of 65–120 min. Its effectiveness decreased rapidly at lower pH levels and shorter contact times. CS demonstrated a maximum removal efficiency of approximately 90% under similar optimal conditions (pH 6–8, 65–120 min contact time). This is attributed to the increased availability of active adsorption sites on the CS surface at acidic conditions, as well as the prolonged contact time, allowing for more effective diffusion and adsorption of fluoride ions. CS exhibited a more gradual decline in efficiency as conditions deviate from the optimum, particularly maintaining better performance at lower pH levels compared to CB. Both adsorbents exhibited better performance at longer contact times, but CB showed greater sensitivity to pH changes, whereas CS maintained more consistent performance across varying conditions.

The optimization of operating factors for maximum fluoride removal was determined using numerical optimization based on BBD–RSM. The primary objective was to maximize fluoride removal efficiency at predetermined conditions. The initial fluoride concentration, solution pH, and adsorbent dose were kept within the experimental range, while fluoride removal was set to maximum. The fluoride removal of 90% was predicted at dose 3.9 g, 57 min treatment time, pH 7 and initial concentration of 3.2 mg/L for CS adsorbent. Meanwhile, the fluoride removal of 98% was predicted at dose 5 g, 67 min treatment time, pH 6.8 and initial concentration of 8.00 mg/L for CB adsorbent. At a predetermined conditions the CB attained 95% fluoride removal while CS removed fluoride by 88% showing good agreement of the models (Table 5).

The optimized conditions were further validated using real groundwater samples collected from Arusha, Tanzania. The three groundwater samples collected and analyzed showed varying fluoride concentrations 7 mg/L (pH 7.8), 4 mg/L (pH 6.8), and 9 mg/L (pH 7.5) for groundwater samples 1, 2, and 3. Additional physicochemical parameters were measured but are not included in this study. When applied to these samples, CB achieved a 96, 95, and 91% fluoride reduction, while CS achieved 90, 93, and 90% demonstrating the practical applicability of both adsorbents in treating real-world water samples.

The Freundlich model revealed a higher adsorption capacity constant (K_F) for CS (18.55) than CB (2.45), indicating that CS is better suited for heterogeneous adsorption. This heterogeneity is supported by the amorphous carbon structure observed in CS's XRD pattern, which provides diverse binding sites for fluoride ions. Additionally, the Freundlich n values for both adsorbents were within the favorable range (1 < n < 10), with CS showing a higher value (n = 1.785), suggesting more favorable adsorption conditions for CS compared to CB (n = 0.826) (Al-Ghouti & Da'ana 2020). The results suggest that BC, with its higher Langmuir adsorption capacity, is ideal for treating high-fluoride waters, while CS is more effective in environments with lower fluoride concentrations due to its stronger adsorbent–adsorbate interactions. These findings highlight the distinct strengths of each adsorbent, allowing for their targeted application based on water fluoride levels.

K₁, K₂, and K_in are the PFO kinetic rate constant (h⁻¹), PSO rate constant (gmg⁻¹min⁻¹), and IDF kinetic rate constant (gmg⁻¹ h^−0.5), respectively, and C is the boundary layer thickness.

The PFO model also showed high correlation (R² = 0.988 for BC, 0.996 for CS), with CS exhibiting a slightly higher rate constant (K₁ = 0.061 min⁻¹) than CB (K₁ = 0.058 min⁻¹). This suggests faster initial adsorption kinetics for CS, possibly due to its highly reactive surface functional groups. However, the PSO model's superior fit indicates that the overall adsorption process is controlled by chemisorption rather than physical adsorption. The IDM analysis revealed significant contributions of intraparticle diffusion to the overall adsorption process. The rate constants for intraparticle diffusion (K_int) were slightly higher for CS (0.05 g/mg·min⁰.⁵) than BC (0.04 g/mg·min⁰.⁵). Additionally, the diffusion layer thickness (Y) was greater for CB (0.69) compared to CS (0.64), suggesting a higher boundary layer effect in CB. This is likely due to CB's denser structure and larger particle size, which may hinder fluoride diffusion compared to the more porous and amorphous CS.

The PSO model's dominance confirms chemisorption as the rate-limiting step in fluoride removal for both adsorbents. CB is better suited for longer-term applications in high-fluoride environments, while CS's faster initial adsorption makes it effective for shorter treatment times or lower fluoride concentrations.

This study investigated and compared the effectiveness of activated carbon derived from CB and CSs for fluoride removal from water in a batch reactor. The BBD was used to optimize operating factors.

• CBs exhibited higher fluoride adsorption capacities (9.09 mg/g) compared to CS (4.55 mg/g) under optimized conditions.

• The optimal conditions for fluoride removal by CB were determined to be 5 g adsorbent dose, 67 min of treatment time, pH 6.8, and 8 mg/L initial fluoride concentration, achieving 96% fluoride removal.

• CS performed better at lower fluoride concentrations, with optimal conditions being 3.9 g dose, 57 min of treatment time, pH 7, and 3.2 mg/L initial fluoride concentration, achieving 90% fluoride removal.

• XRF and XRD analyses confirm that CB primarily removes fluoride through ion exchange, forming fluorapatite, while CS employs surface complexation facilitated by metal oxides and an amorphous carbon matrix.

• Both adsorbents followed pseudo-second-order kinetics, indicating chemisorption as the primary mechanism.

• The findings of this study support the use of locally available adsorbents, such as CB and CS, for cost-effective defluoridation, especially in rural areas.

• The study recommends using these materials for fluoride removal based on the degree of contamination. However, further research should investigate the regeneration and reusability of these adsorbents to improve their economic viability, evaluating their performance in continuous systems.

The authors acknowledge the Water Quality Laboratory of the Water Institute, (Tanzania) for providing space and facilities for this research work.

No funding was received for this research.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.